The NEURONS and NEURAL SYSTEM: a 21st CENTURY PARADIGM

This material is excerpted from the full β-version of the text. The final printed version will be more concise due to further editing and economical constraints. A Table of Contents and an index are located at the end of this paper. A few citations have yet to be defined and are indicated by “xxx.”

James T. Fulton Neural Concepts [email protected]

August 1, 2016 Copyright 2011 James T. Fulton

1 [xxx equatorial and axial are introduced in definitions on page 30.] [xxx consolidate on angstrom or on nm throughout ] [xxx the choice of C(arboxylic)-path is more definitive than A(cidic)-path in differentiating from the H-best condition. Alternately, the label A(cetate)-path is more indicative of the actual (predominant situation. This path includes a large number of esters ending in -ate as well as the organic acids ending in -ic See Section 8.5.xxx.] [xxx G-Path clearly avoids duplication between the old and new set of path labels ] [xxx use the A-, G-, N - and P- path labels from now on, exccept in quotations. ] THIS CHAPTER IS PRESENTED IN DRAFT FORM AT THE CURRENT TIME BECAUSE OF THE NEED FOR THE INFORMATION BY THE GUSTATORY COMMUNITY.

8 Stage 1 & 2, Signal Generating & Processing Neurons1

“Science is made up with facts as a house is made from stones. But a collection of facts is no more a science than a pile of stones is a house.” —Poincare' , Hypotheses in Physics (1952) “In order to understand any part of nature, one must have both experimental data and a theory for interpreting the data and predicting new data.” – Shepherd, Outline of a Theory of Olfaction, 2005 This Part provides an in depth discussion of stage 1 operation (sensing) of the Gustatory Modality

8.5 The gustatory modality

Excepting the group at the University of Wisconsin, there has not been a large amount of exploratory research into gustation with an academic focus since the 1980's. Most of the work continues to relate to product development in the food industry, generally under the rubric flavor rather than taste.

Quoting van der Heijden in 1993, “Humans can perceive four tastes, of which sweet and bitter have received the most attention from scientists.” This will become evident when the sparse information concerning the elicitation of the acidic and salty sensations are reviewed. The literature fails to note the acidic sensation is based primarily on the sensing of organic acids. The sensation of salty is elicited almost exclusively by the hydrated sodium ion. While based on very fragmentary data, Boudreau has provided the most complete conceptual

1August 1, 2016 Signal Generation & Processing 8- 3 description of the overall gustatory modality2. Inconveniently, he uses the term ganglion in place of the more conventional nerve when speaking of the major nerves serving the oral cavity. The following discussion will repeatedly encounter conflicts between two different modalities of the neural system, the gustatory and the nocent modalities. While the gustatory modality is reasonably well understood at the concept level, its differentiation from the less well understood nocent (pain reporting) modality suffers. Many inorganic materials discussed in empirical gustation investigations (such as HCl and CaCl2 ) are actually nocentaphores. The nocentaphores will be discussed in Section 8.7. 8.5.1 Background for and summary--the gustatory modality hypothesis 8.5.1.1 Background

8.5.1.1.1 Historical documentation

Cagan & Kare edited a comprehensive volume with a focus on olfactory transduction in 19813. The individual paper authors explored a broad range of potential transduction mechanisms. Some focused on the role of proteins on the surface of the sensory neuron membrane. Others focused on the potential for the lipids of the membrane to be involved in the primary mechanism. No conclusions were drawn. Kurihara, Miyake & Yoshii explored the work of Kamo et al4 in 1980. Kamo et al. attempted to mimic the sensory response of the gustatory neurons much as Hodgkin did for the visual modality sensory neurons (pages 249-286). Their equation 2 expresses the two-way operation of the proposed fundamental chemical reaction of the transduction process in abstract form. They then formulated a set of second order differential equations based on the Law of Mass Action (which requires a totally reversible reaction in solution). Unfortunately, there are a vast array of possible mechanisms that are satisfied by a second order differential equation. These include the quantum-mechanical mechanisms such as the production of nuclear isotopes.

In their analysis, they ignored the rapidly rising attack transient and concentrated on the decreasing transient and the steady state value (prior to cessation of stimulation). However, their boundary conditions necessarily included the rapidly rising attack transient in their ultimate solution. While their particular solution of the differential equations is complete and includes the pulse condition (as opposed to just the impulse condition), there are two problems with their interpretation of it. First, they apparently did not treat the off-response properly by applying new boundary conditions. The off response is a direct and independent measurement of the second exponential in equation 2 and in the E/D equation of this work. Second, Kamo et al. did not treat the singularity within their equations properly. Their particular solution does not apply at w1 = w2. Removal of the singularity results in a different mathematical form that this author has labeled the Hodgkin condition (Section xxx). They noted Kashiwagura et al contained additional material on their investigation5.

2Boudreau, J. (1989) Neurophysiology and stimulus chemistry of mammalian taste systems In Teranishi, R. Buttery, R. & Shahidi, F. eds. Flavor Chemistry: Trends and Developments. Washington, DC: American Chemical Society Chapter 10

3Cagan, R. & Kare, M. ed. (1981) Biochemistry of taste and olfaction. NY: Academic Press Parts II & III

4Kamo, N. Kashiwagura, T. Kurihara, K. & Kobatake, Y. (1980) A Theory of dynamic and steady responses in chemoreception J theor Biol vol 83, pp 111-130

5Kashiwagura, T. Kamo, N. Kurihara, K. & Kobatake, Y. (1980) Interpretation by theoretical model of dynamic and steady components in frog gustatory response Am J Physiol (gastrointest.) Vol. 238, pp G445-G452 4 Neurons & the Nervous System

Scott & Mark explored coding within the taste system in 19876. Their analysis relied upon the chemical theory of the neuron. Their abstract opens with; “Attempts to define the organization of the taste system in terms of the physical characteristics of stimuli have been largely unsuccessful.” They noted specifically, “Molecular weight and pH did not relate to the total organizaton of the system. . . “ Their work, employing multidimensional scaling (MDS)will be addressed below. Fisher & Scott reviewed the subject of food flavours in 19977. The work is entirely conceptual, with few substantive sketches, and is based on behavioral studies. It makes the conventional assumptions that taste involves either ion channels passing molecules through the sensory lemma or proteins on the surface of the sensory lemma. It does not demonstrate either of these concepts is correct, nor does it provide primary data relative to these concepts. Their table 3.1, from Kinnamon & Getchell, illustrates the continuing conceptual character of their thesis. Pages 85–87 provide a comprehensive list of conceptual gustatory features that suggest four basic taste sensory channels.

Although dated, the discussion of the gustatory modality in Noback remains useful8. The description of the taste buds as bowl shaped features located behind pores in the lingual epithelium and containing on the order of 25 or more distinct gustatory sensory neurons is clear. He also notes the seldom reported high turnover rate among the sensory neurons, as opposed to just the sensory hair of each cell. “Each mature sensory neuron is replaced every 200 to 300 hours.” This turnover rate is similar to that of just the outer disks of the visual sensory neurons. It is also similar to the turnover rate of the piezoelectric proteins within the auditory sensory neurons. Like the retina of the visual system, the taste buds contain “sustentacular cells” that develop and replace the older sensory neurons. This replacement is associated with a transfer of the synapse with the orthodromic neurons from the old to the new sensory neuron.

Noback also notes the neurons immediately orthodromic to the sensory neurons emerge from the taste bud and become myelinated immediately (therefore stage 3 neurons). As in the case of the auditory neurons, this suggests the encoding of the analog signals occurs at the first Node of Ranvier and not within the soma of these stage 3 neurons. This node of Ranvier occurs in what is conventionally described as the dendrite of the signal projection neuron. However, the myelinated neuron component directly after the encoding node of Ranvier is an axon segment.

The number of gustatory sensory neurons is small, estimated at about 10,000 in human babies and the sensitivity of the modality is low relative to the olfactory modality (as many as 20,000 times the number of molecules are required for gustation as olfaction). However, it is probably the method of delivery more than the sensitivity of the receptors that is limiting. The quantum- mechanical character of the response of the gustatory sensory neurons suggest they are highly efficient at sensing molecules delivered to their immediate vicinity.

The subject of taste modifiers (additional chemicals applied with the stimuli) and enhanced stimuli (stimuli with an additional ligand not found naturally) have not been introduced into a comprehensive theory of gustatory sensing. However, a few enhanced stimuli have been documented having perceived intensities as much as 30 to 10,000 times greater than the strongest natural materials. Most of the stimuli with an artificial gustaphore have been centered on the attempts to find artificial super-sweeteners, frequently with an auxiliary goal that they be non-caloric from a nutrition perspective. Glendinning et al. summarized the conventional wisdom concerning gustation in 2000. They

6Scott, T. & Mark, G. (1987) The taste system encodes stimulus toxicity Brain Res vol 414, pp 197-203

7Fisher, C. & Scott, T. (1997) Food Flavours: Biology and Chemistry. Cambridge, UK: The Royal Society of Chemistry.

8Noback, C. (1967) The Human Nervous System. NY: McGraw-Hill pp 118 &136-139 Signal Generation & Processing 8- 5 noted,

“Among the sensory systems, the taste system is unusual in its capacity to respond to a large number of stimuli that vary greatly in molecular size, lipophilicity, and pH (e.g., salts), amino acids, sugars, acids, vitamins, fatty acids, and many toxic compounds). To accommodate this structural diversity, taste cells utilize a diverse array of transduction mechanisms. However, only a fraction of these mechanism occurs within any given taste cell, different subsets of taste cells appear to express different transduction mechanisms, and hence different molecular receptive rages.” All of Glendinning et al’s. discussion is dependent on the chemical theory of the neuron, and the concept of pores through the dendrolemma of the sensory neurons or the presence of proteins inserted into and traversing the dendrolemma (the conventional G-protein hypothesis). Their figures are limited to simple conceptual sketches of a dendrolemma. No differentiation of dendrolemma into special classes is considered. The figures depend on the porosity of the hydraulic barrier (tight junction) between the sensory neurons for the sensing of simple positive ions (illustrated as Na+ & H+). They note, the hydraulic barrier is known to be impermeable to amiloride in vertebrates. In summary, the Glendinning et al. material is an excellent source of laboratory results but provides no detailed theory of gustatory neuron operation. They do not address the time duration of the sensory mechanisms or explain the form of the C/D waveform. The internal conflicts within their models are characterized by the discussion of G-protein versus cAMP involvement in “sweet” sensing on page 325.

There is virtually no detailed discussions of the precise role of proteins in the gustatory modality, other than the assertion that such a role must exist and subsequent Bayesian inferences derived from analyses of the genetic code. McManus et al. made the strongest presentation in a two page communication in 19819. They asserted the most likely site along a protein to act as a sensory receptor was where proline residues disrupted the normal helical structure and exposed two hydroxyl groups to provide an AH,B bonding site. Their subsequent paper focused on polyphenols, rather than sugars, interacting with proteins.

Smith and Davis made an important assertion in 2000 (page 362) that they refuted almost immediately,

“A universal characteristic of mammalian gustatory neurons is their responsiveness to stimuli representing more than one of the classic four taste qualities.”

This statement suggests the major perceptions of taste do not correspond to any prior categorization of individual stimuli into classes. It also suggests that, like in the case of yellow in vision, the perception of a prominent taste may not be related to an individual sensory channel. Such a perception may be the result of signal manipulation within stage 4. The many multidimensional analyses of taste perceptions appears to support this assertion. However, Smith and Davis follow immediately with the statement conflicting with their above view (page 363). Speaking of the moth, Bombyx mori, they say,

“The most sharply tuned fiber is depolarized exclusively by specific isomers of the sugar alcohol, inositol, the other three fibers are less sharply tuned . . .” This work would recommend the term sugar alcohol be replaced by glyco-alcohol (from the perspective of gustation) since the chemical is perceived as sweet but the glycophore stimulating the receptor channel is not attached to a heterocyclic saccharide. Pure saturated aliphatic alcohols are intrinsically tasteless. However, two points are significant. First, saturated aliphatic alcohols (and aldehydes) rapidly form azeotropes in water–based solutions. Second, alcohols containing significant low levels of impurities have been used to rate alcohols on a perception scale in many pedagogical experiments.

9McManus, J. Davis, K. Lilley, T. & Haslam, E. (xxx) xxx J Chem Soc Chem Commun vol. 7, pp 309-311 6 Neurons & the Nervous System

A similar recommendation can be made regarding the sugar acids, such as ascorbic acid, that do not contain any carboxylic acid group (indicative of an organic acid) or any CH2OH group (indicative of a saccharide). They are predominantly sweet alcohols based on gustatory perception.. The common protocol of recording the number of action potentials in a five second interval obliterates any information regarding the shape of the excitation/de-excitation function in gustation. It is known that most animals can discriminate or identify a gustaphore in less than one second. - - - - Doty edited a handbook of olfaction and gustation in 2003 that provides both academic and clinical material10. It offers considerably more detailed information than Finger et al11. It focuses on anatomy and the chemicals of gustation but does not address the neurophysiology of olfaction or gustation at a significant level. The authors writing in Doty review many of the earlier theories of olfaction and gustation and generally finds them wanting. On the other hand, they do not converge on one specific theory or model that is able to describe chemical sensing. The handbook is very useful for data mining but offers no significant theory of chemical sensing. See Section 8.4.

- - - -

Spector & Travers (page 173) note the low likelihood of a chemotopic organization in the gustatory modality based on considerable investigation by the community. Travers et al. have provided a paper showing how complex the gustatory signaling paths are and the difficulty of preparing a set of stimulants and a protocol for evaluating this modality unambiguously12. Their figure 3 (lower right) is difficult to interpret, however it may show cases of a nominal pulse rate being reduced in the presence of a second stimulant, a feature of a differencing channel (at the M neurons in the figure presented below). It is common to find examples of summing (at the P neurons of the following figure) among at least two channels in the neural paths of the chorda tympani. Pfaffmann said as an example, “Three types of fiber were found : those stimulated (1) by acids, (2) by acid and by sodium chloride, and (3) by acid and by quinine13.” He also noted, “Since some of the fibers respond to such a surprisingly diverse group of stimuli, one is reminded that the single fiber of these experiments is inferred from the appearance of the potential record.” 8.5.1.1.2 Major problems with the RSC Jmol & JSmol Libraries

Molecular modeling and x-ray crystallography play a critical role in understanding the critical role of transduction in the gustatory modality. As noted in Section 8.4.1.2.3, the field of molecular modeling has developed unevenly and with little sophistication.

This work has relied upon the 3D Jmol and JSmol files in the archive provided by the Royal Society of Chemistry. The reliance has been primarily for determining the distance between various atoms within a molecule in 3D space. The Jmol archive has been evolving for the last decade but exhibits some significant shortcomings, particularly the ability of anyone to submit

10Doty, R. ed. (2003) Handbook of Olfaction and Gustation, 2nd revised and expanded edition. NY: Marcel Dekker

11Finger, T. Silver, W. & Restrepo, D. eds. (2000) The Neurobiology of taste and smell, 2nd Ed.. NY: Wiley- Liss

12Travers, S. Pfaffmann, C. & Norgren, R. (1986) Convergence of Lingual and Palatal Gustatory Neural Activity in the Nucleus of the Solitary Tract Brain Research, vol 365, pp 305-320

13Pfaffmann, C. (1941) Gustatory Afferent Impulses J Cell Comp Physiol vol 17, pp243+ Signal Generation & Processing 8- 7 files that have not been curated by the RSC before deposition in the database. The JSmol database was implemented a short time ago to complement the Jmol database that exhibits a variety of additional shortcomings (See Sections 8.4.1.2.3 & 8.6.1.6.3). News flash: The Jmol files are no longer available in 3D based on the cancellation of their internet security certification. This in turn was based on the “Cessation of Activity” as of 15 October 2015. It appears these files are being supplanted by the JSmol files curated by the same RSC. However, the JSmol 3D database was taken off the internet for an unspecified period as of 19 Nov 2015 (as was the ability to contact the curator via the website). While the JSmol files examined frequently have more header information than the Jmol files, the information is frequently disguised with a dummy author’s name (Marvin) appearing on large numbers of JSmol files. This work will continue to rely upon the Jmol and JSmol files previously downloaded to the computer files of this investigator in spite of their shortcomings until the RSC resolves its internal problems. It is hoped that the actual d-values are at least proportional, if not precise, to those given in ChemSpider. 8.5.1.1.3 How have the taste sensations been defined?

The chemical-sensing oriented portion of the psychophysical community has long speculated on the functional organization of the gustatory and olfactory systems. A major problem has been defining what is sensed by the gustatory and olfactory modalities. Initially, only the perceived responses of humans, as computed in stage 5 and expressed using stage 6 of the neural system, could be documented. More recently, electronic techniques have allowed recording sensations, in stages 1 through 4 within the neural system. However, no rational description of the ordering of gustatory or olfactory responses has been defined. Nor has the number of “dimensions” associated with that ordering been established. Squire et al. (pages 642-645) have provided a discussion of this situation.

Pfaffmann et al. have discussed the delineation of the gustatory responses to different stimuli into five classes described as;

S group–sweet or sugary N group– salty, and specifically “salty based on the presence of Na+ ion.” Q group– bitter, letter derived from the relevance of quinine H group– acidic. U group–umami, a potential taste apparently unique to monosodium glutamate in some investigator’s eyes.

Johnson (page 753) has provided a discussion of these properties. Such properties are always discussed in narrative form and no two authors agree at the detail level.

Some investigators have attempted to define a unique channel, the umami channel, associated with mono-sodium glutamate. However, this compound appears to be a typical gustant involving a sodium-based member of the N group, a moiety affecting the H group and a moiety affecting the S group (See Section 8.5.1.6.2 & 3). An explicit unique definition of the perception of umami has not been found in the literature. In 2015, Running et al. have attempted to identify a sixth gustatory channel that they describe as unique to the “fats” as defined in the food science field. As shown in Section 8.5.4.10, they are in fact discussing the fatty acids that primarily stimulate the acidic channel, labeled G1 in this work, but if unsaturated can stimulate other channels as well.

Squire et al. note (2003, page 613), “Most taste stimuli are hydrophilic molecules, including Na+ salt (salty), divalent salts, and KCl (salty and bitter), acids (sour), sugars (sweet), amino acids (sweet, bitter, and umami), and proteins (sweet and bitter). Some taste stimuli are lipophilic, including the bitter-tasting alkaloids and many synthetic sweeteners.” 8 Neurons & the Nervous System

Smith & Davis have provided a different behavioral perspective14, “These qualities and the behavior associated with them provide the means by which an animal makes ingestive decisions. Through these qualities, taste help to ensure the animals’ energy supply (sweet), maintain the proper electrolyte balance (salt), and avoid the ingestion of toxic substances (sour, bitter)” A major area of study has been limited to studying the ionic portion of the chemical spectrum. Another major area has been focused on the study of organic chemicals, primarily but not exclusively the simple sugars. While Spector and Travers have conceptually recognized two broad classes of receptors, these two areas of study have yet to converge on a comprehensive functional model at a detailed level. They noted in 2005 (page 177), “For all of the behavioral and electrophysiological work that has been conducted to date, it is revealing that definitive evidence distinguishing various modes of neural coding of taste quality remains to be seen.” Many anecdotal remarks also appear in the literature but these have not generally been incorporated into a comprehensive functional model either. On the same page, they note, “An examination of the literature leads one to speculate that some of this complexity might be due to experimental factors that potentially obscure the discovery of orderly principles. Most investigations have not employed statistically adequate numbers of stimulants and receptors to support drawing precise conclusions from the data.

Hudspeth & Tanaka provided a brief review that only touched on the chemical senses, taste and olfaction15. They note the classical concept suggests the taste sensations of sweetness, sourness, saltyness and bitterness. They quickly note the sensitivity of the human system to fat as well. Many alternates to this brief listing have been proposed. Nearly a century ago, Ikeda suggested an additional axis to the taste map involving glutamate (or more specifically monosodium glutamate)16. He named the sensation associated with glutamate umami17. Sugimoto & Ninomiya have provided a review of more recent work related to umami. Axel & Buck suggested a very large number of independent olfactory sensor types. They inferred several relationships between the olfactory system and over 1000 genes. xxx add reference and brief comments re: sensing organic chemicals xxx

8.5.1.2 Anatomy of the peripheral portion of the gustatory modality

A problem related to taste is the variation in sensitivity to different stimuli with location on the tongue. Age is known to play a significant role in taste, with many senior citizens experiencing a loss in taste sensitivity. Children are believed to exhibit an enhanced taste sensitivity possibly related to sensory cells covering a larger areas of the oral cavity. Eyzaguirre & Fidone have described the sensitivity of the human tongue using Figure 8.5.1-1 18. The sour taste is located at both edges of the tongue, salt and sweet at the tip of the tongue and bitter at the base of

14Smith, D. & Davis, B. (2000) Neural representation of taste In Finger, T. Silver, W. & Restrepo, D. eds. (2000) The Neurobiology of taste and smell, 2nd Ed.. NY: Wiley-Liss Chapter 14

15Hudspeth, A. & Tanaka, K. (2000) Sensory systems: editorial overview Curr Opin Neurobiol vol xxx, pp 443- 446

16Ikeda, K. (2002) New Seasonings Chem Senses vol 27, pp 847-849 A reprint of original

17Sugimoto, K. & Ninomiya, Y. (2005) Introductory remarks on umami research: candidate receptors and signal transduction mechanisms on umami Chem Senses vol 30(suppl 1) pp i21-i22

18Eyzguirre, C. & Fidone, S. (1975) Physiology of the Nervous System, 2nd Ed. Chicago, Il: Yearbook Publishers page 146 Signal Generation & Processing 8- 9 the tongue. Squire et al. (2003, pages 632-635) provide additional detail concerning the location of the gustatory sensory neurons. Johnson has provided more details of the human tongue and its connection to the main nerves. Johnson also provides a comprehensive posterior view of the palo- and neo-cortex that places the various neural paths in the brain associated with the gustatory modality in perspective. Bartoshuk has questioned diagrams such as this based on its probable origination in 1901 and on their experiments19. He noted the presentation suggests an all-or-nothing response for different areas. He also notes that was not the intent of the original author. “If taste qualities were arranged chemotopically on the tongue such that sweet receptors were found on the front, then damage to the chorda tympani nerve would selectively impair one’s ability to taste sweet. Not only does this not occur, damage to the chorda tympani often produces virtually no change in the subjective taste world of the patient.” He provided an alternate figure describing the sensitivity of the tongue showing the high degree of overlap between the sensations throughout the surface of the tongue.

More recent terminology for the types of taste buds found on the human tongue include;

1. fungiform papillae (found on the body or Figure 8.5.1-1 Semi-schematic representation of anterior 2/3 of the tongue). the tongue that is archaic (see text). The location 2. filiform papillae. of different sensory modalities and the areas 3. foliate papillae (found on the base or innervated by different cranial nerves (V, VII & IX) posterior 1/3 of the tongue). are shown. The vallate papillae are located in the 4. circumvallate papillae (found at the region marked by the dashed lines. From base of the tongue arranged in a V-shpae). Eyzaguirre & Fidone, 1975. The terms do not describe the character or content of the individual taste bud. While specific data is unavailable, it appears each taste bud incorporates a small number of sensory receptor cells of multiple types (probably all types but in different proportions or exposures depending on location)

- - - - -

The rough character of the tongue is due to the multiple papillae covering the forward sector of the tongue. Each papilla contains about 250 individual taste buds. Each taste bud forms a cavity containing a number of individual gustatory sensory neurons. The sensory neurons in each bud may sense different stimulants.

- - -- Mistretta studied the effects of age on the morphology of gustation and the number of gustatory

19Bartoxhuk, L. (1993) Genetic and pathological taste variation: what can we learn from animal models and human disease? In Margolis, F. ed. The Molecular Basis of Smell and Taste Transduction. NY: John Wiley & Sons pp 251-267 10 Neurons & the Nervous System sensory neurons in humans and other mammals20. She concluded, “The recent data from quantitative studies of taste buds in old humans, rhesus monkeys, and rats complement neurophysiological data on taste responses from aged rats and lead to the general conclusion that the peripheral taste system is maintained structurally and functionally across the life span. The number of the taste buds varies immensely with species. Kare has compiled a simple table illustrating this variation21.

8.5.1.2.1 Morphology of the gustatory modality

Figure 8.5.1-2 is a schematic/block diagram of the proposed gustatory system. It shows the taste bud schematically followed by a preliminary block diagram of the modality through stage 4. The synapses within the taste bud are not shown in detail. However, the neurons leaving the taste bud are known to become myelinated immediately. Lacking any data showing analog signal differencing of the initial gustatory signals, no connections are shown to the potential midget ganglion (difference) encoding neuron. The exemplar signal path is shown proceeding to a parasol ganglion neuron wherein the first Node of Ranvier performs the encoding (modulation) function. The nucleus solitarius is the “1st relay” and the parabrachial nucleus is the “2nd relay” in the gustatory signaling chain. Signals from these elements divide and proceed to both the ventral posteromedial thalamus (PVM) and to the amygdala/hypothalamus of the limbic system. The signals from the PVM are known to proceed to the region of the operculum of the parietal lobe (BA 2), the adjacent insula and inner surface of the temporal lobe (potentially BA 27). Aggleton & Passingham have provided more specific information on these locations in the macaque22. Rolls & Baylis have provided some data on the nearly parallel olfactory and gustatory tracts in the primate orbiotfrontal cortex23. Signals from the amygdala/hypothalamus are known to connect to the orbitofrontal cortex area of what this work calls the prefrontal cortex.

20Mistretta, C. (1989) Anatomy and Neurophysiology of the Taste System in Aged Animals Annal NY Acad Sci vol 561, pp 277-290

21Kare, M. (1971) Comparative study of taste In Beidler, L. ed. Taste: Handbook of Sensory Physiology, Vol IV, Part 2 Chap 13

22Aggleton, J. & Passingham, R. (1981) Stereotaxic surgery under x-ray guidance in the rhesus monkey Exp Brain Res vol 44, pp271-276

23Rolls, E. & Baylis, L. (1994) gustatory, olfactory and visual convergence within the primate orbitofrontal cortex J Neurosci vol 14, pp 5437-5452 Signal Generation & Processing 8- 11

Figure 8.5.1-2 Proposed schematic of the gustatory system. The nucleus solitarius and parabrachial nucleus may be the same element as defined by different investigators. Complications arise in evaluating the gustatory and nociceptor modalities due to their sharing certain nerves directed to the stage 2 signal processing engines. These engines appear to contain multiple summation channels (P) and differencing channels (M) delivering signals to a variety of higher level engines (not shown in detail). See text.

Each sensory neuron exhibits a small number of microvilli emanating from the neuron and into the cup formed by the taste pore in the lingual epithelium. This is illustrated more clearly in [Figure xxx ]

The figure shows the gustatory neurons delivering signals to nerves V (trigeminal), VII (facial), IX (glassopharyngeal) and X (vagus) depending on their function and location within the oral cavity. The routing is not widely accepted and many neurons appear to be routed via the nerve V, the trigeminal nerve, which serves large areas of the mucosal membrane within the oral and nasal cavities as well as lining the eye sockets. These nerves all enter the brain stem at the interface between the pons and the medulla. A major problem in evaluating the gustatory modality, illustrated in the previous and following figures, is the mixture of sensory signals that arrive at the Nuclei of Solitarius (NoS). In general, electrophysiological laboratory investigations must insure the data is unadulterated and applies specifically to gustaphore receptor channels. Specifically, by the time the signals reach the NoS, they cannot easily be distinguished between their origin within the nocent or gustatory channels. Figures 788 & 912 in a recent reprinting of an undefined edition of Gray’s Anatomy (widely circulated in Wikipedia, etc.) clearly illustrates the problem. Figure 788 shows the chorda tympani as a branch of nerve IX 12 Neurons & the Nervous System

rather than nerve VII as shown in Mistretta (1990). Figure 912 uses the less definitive label “facial nerve” more in agreement with Mistretta. It is important to acquire gustatory signals from within the chorda tympani or other branches of the main nerves before they are adulterated.

The afferent neurons within the taste bud frequently connect to multiple sensory neurons. However, this may be to support the rapid turnover of the sensory neurons with the afferent neurons connecting to only one functional neuron at a time. The literature does not suggest any differencing of neural signals occurs within the taste bud. Therefore, the figure shows a neuron from the taste bud connecting to a parasol (summing) ganglion neuron, P. The potential for a midget (differencing) ganglion neuron, M, has been retained for discussion purposes only.. The neurons emanating from the taste bud become myelinated immediately and prior to reaching the soma of the neurons. This myelination suggests the sensory neurons of the gustatory modality are much like those of the auditory modality in that the first Node of Ranvier appears prior to the soma and acts as the encoding element within this stage 3 neuron.

The figure illustrates the difficulty of isolating the gustatory neural signals from the nocent neuron signals due to their sharing the nerves projecting both signal types to the nuclei solitarius.

The nucleus of solitarius is known to sum signals from multiple sensory neurons but whether on a spatial basis or a stimulus basis remains unclear. This nucleus does not show an organization similar to a glomeruli or any recognizable chemotopic organization. Figure 8.5.1-3 is an schematic of the neural paths of mammals based specifically on the anatomy of sheep. The figure offers considerable definition of the separate neural paths and way points. It also shows the lack of a common focus of the gustatory signals as they pass through the rostral brainstem. Signal Generation & Processing 8- 13

Figure 8.5.1-3 Schematic of mammalian neural paths in gustation. Based on the sheep but applicable to any mammal. Mistretta noted, the nasoincisor duct with taste buds, schematically indicated on the anterior hard palate and innervated by the greater palatine nerve, has not yet been reported in the sheep. From Mistretta, 1990.

The role of the PBN remains unclear. Some authors suggest it is missing, or inconsequential in humans24. Others sketch it in to conceptual drawings25. Whether the nucleus solitarius and parabrachial nucleus should be considered stage 2 signal processing elements or merely relay points is yet to be resolved. Their forward connections suggest they should be considered elements of stage 4, signal manipulation.

24Mistretta, C. (1990) Taste Development In Coleman, J. ed. Development of Sensory Systems in Mammals. NY: John Wiley & Sons Chapter 14

25Rolls, E. & Scott, T. (2003) Central taste anatomy and neurophysiology In Doty, R. ed. Handbook of Olfaction and Gustation, 2nd Ed. NY: Marcel Dekker page 680 14 Neurons & the Nervous System

Many signal paths emanate from the nucleus solatarius and the parbrachial nucleus, with the majority believed to connect to the ventral posteromedial thalamus (VPM) and the amygdala/hypothalamus. Beyond those elements, the neural pathway of the gustatory modality is poorly documented. The VPM appears to connect to an area of the cerebral cortex known as the insula with spillover onto the operculum of the parietal lobe (BA 2) and the upper medial surface of the temporal lobe (?BA 27). This region may constitute a portion of the saliency map devoted to gustatory and olfactory representations available for review by the cognitive functions of stage 5. The amygdala/hypothalamus appears to connect to the orbitofrontal portion of the prefrontal cortex.

Anatomical data on the tongue of many species appears in Bradley26. Unfortunately, many textbooks have adopted a simple map from Boring (1942) redrawn from some very early work of Hanig (1901). The best description of the sensor location on the human tongue is shown in Figure 8.5.1-4 from Yanagisawa et al27. based on the studies of Collings28 [xxx add words ]

Figure 8.5.1-4 Relative magnitude estimates for locations on the human tongue ADD for the historical gustaphores. From Collings, 1974.

26Bradley, R. (1971) Tongue topography In Beidler, L. ed. Taste: Handbook of Sensory Physiology, Vol IV, Part 2, Chap 1

27Yanagisawa, K Bartoshuk, L. Karrer, T. et al. (1992) Anesthesia of the chorda tympani nerve:insights into a source of sygeusia Chem Sneses vol. 17, p 724 (abstract).

28Collings, V. (1974) Human taste response as a function of locus of stimulation on the tongue and soft palate Percept Psychophys vol 16, pp 169-174 Signal Generation & Processing 8- 15 The unique labels in this figure are described by Mistretta in Figure 8.5.1-5

Figure 8.5.1-5 Diagram of types of lingual gustatory papillae, the fungiform, foliate, and circumvallate, and of gustatory epithelium representative of that on the soft palate and epiglottis. The black dots represent individual taste buds. From Mistretta, 1989.

- - - -

Pfaffman et al. provided some early gustatory information using squirrel monkeys29. They recorded action potentials from the chorda tympani, a branch of the VII facial nerve. They introduced the concept of labeled lines with little detail. They did note, “Two-thirds of our sample of taste units fall readily into one of the four classic taste categories with a peak at one basic taste stimulus. ‘Side bands’ around such peaks produce a certain degree of multiple sensitivity. One-third of the responsive fibers, however, cannot be classified by a single ‘best stimulus’ but appear to have broad multiple sensitivity. All of their reported waveforms suggest parasol type encoding channels.

Recently Mattes has introduced data showing the gustatory modality of humans is also sensitive to a series of free fatty acids30. [xxx add words ] Lim & Lawless explored the taste of a variety of metal salts31. [xxx add words ] Rogers et al. have offered a limited tree focused on bitter taste relationships32. [xxx add words ] Lundy & Norgren experiments suggesting feedback of signals from the amygdala/hypothalamus and gustatory cortex to the parabrachial nucleus. However, while they located their stimulation and recording sites adequately for anatomical purposes, they did not identify the precise location of their sites with respect to the inputs and outputs of various elements containing millions of individual neurons.

29Pfaffman, C. Frank, M. Bartoshuk, L. & Snell, T. (1976) Coding gustatory information in squirrel monkey chorda tympani In Sprague, J. & Epstein, A. eds. Progress in Psychobiology and Physiological Psychology. NY: Academic Press. vol 6, pp 1-27

30Mattes, R. (2009) Oral Detection of Short-, Medium-, and Long-Chain Free Fatty Acids in Humans Chem Senses vol 34(2), pp 145-150

31Lim, J. & Lawless, H. (2005) Qualitative Differences of Divalent Salts: Multidimensional Scaling and Cluster Analysis Chem Senses vol 30(9), pp 719-726

32Rodgers, S. Busch, J. Peters, H. & Christ-Hazelhof, E. (2005) Building a Tree of Knowledge: Analysis of Bitter Molecules Chem Senses vol 30, pp 547–557 16 Neurons & the Nervous System

Citing Gilbertson33, Hudspeth & Tanaka have asserted that “mammals respond strongly to the taste of fat34.” As discussed in this Part and in Part 2A supporting olfaction, fats and fatty acids exhibiting a pair of double covalent bonds at the appropriate spacing are entirely capable of stimulating various GR’s and OR’s. Alternately, fats and fatty acids have many opportunities to mediate the chemical sensing of other gustants and olfactants. - - - - The mechanism of gustatory transduction has long been an unsolved problem. The likelihood of four distinct perceived tastes, sweet, sour, salty, bitter & have been proposes since ancient times. Umami became a potential fifth flavor during the early 1990's. Many, generally unsuccessful, efforts have been made to explain these perceived tastes based on the chemical structure of the stimulants35. - - - -

8.5.1.2.2 The morphology of the taste bud & sensory neuron

Figure 8.5.1-6 shows a modified figure from Mistretta (page 288), who modified a figure from Murray based on his photomicrographs36. Mistretta eliminated the myelination of the stage 3 neurons and labeled the hydraulic seals as tight junctions (which could be misinterpreted). She also annotated the Murray figure. The figure has been further annotated to show the presence of a somatosensory neuron as added to the Murray figure by Finger & Simon37. The figure as reproduced here is at a magnification of about 2100x.

The typical taste bud contains 5-50 sensory neurons and a number of supporting cells. It is believed that some of these supporting cells are responsible for the phagocytosis of non- functional sensory neurons at the end of their life expectancy (nominally 200 hours)38.

Murray began the practice of numbering the sensory neurons based on their reflected appearance after uranyl acetate staining; some stained cells appeared darker than others, Mistretta chose not to adopt this notation.

33Gilbertson, T. (1998) Gustatory mechanisms for the detection of fat Curr Opin Neurobiol vol 8, pp 447-452

34Hudspeth, A. & Tanaka, K. (1998) Sensory systems: editorial overview curr Opin Neurobiol vol 8, pp 443- 446

35Shallenberger, R. & Acree. T. (1967) Molecular theory of sweet taste Nature vol 216, pp 480-482

36Murray, R. (1973) The ultrastructure of taste buds In Friedmann, I. ed. The Ultrastructure of Sensory Organs. NY: American Elsevier Chapter 1

37Finger, T. & Simon, S. (2000) Cell biology of taste epithelium In Finger, T. Silver, W. & Restrepo, D. eds. (2000) The Neurobiology of taste and smell. NY: Wiley-Liss Chapter 12, page 293

38Suzucki, Y. Takeda, M. Obara, N. & Nagai,Y.(1996) Phagocytic cells in the taste buds of rat circumvallate papillae after denervation Chem Senses vol 21, pp 467-476 Signal Generation & Processing 8- 17

Figure 8.5.1-6 Morphological features of the mammalian taste bud. The analog waveforms from the sensory neurons are converted to stage 3 projection waveforms at the first node of Ranvier where the myelination begins. A second node of Ranvier is shown schematically as labeled. Note the hydraulic seals that protect the neural system from potential contamination. Note also the proximity of a somatosensory neuron. See text. Modified from an original in Murray, 1973.

Murray has provided a variety of actual micrographs of mammalian sensory neurons. However, they are difficult to interpret without considerable experience. A Figure 8.5.1-7 from Finger & Simon39, modified from Singh40, shows the same cytology of a gustatory sensory neuron but from a fly Drosophila. The importance of this figure is that it shows internal lemma as expected to

39Finger, T. & Simon, S. (2000) Cell biology of taste epithelium In Finger, T. Silver, W. & restrepo, D. eds. The Neurobiology of Taste and Smell. NY: Wiley-Liss Chapter 12, page 290

40Singh, R. (1997) Neurobiology of the gustatory systems of Drosophila and some terrestrial insects Micro Res Tech vol 39(6), pp 547-563 18 Neurons & the Nervous System support the electrolytic circuit of the neuron proposed by the Electrolytic Theory of the Neuron. It also shows the interstitial cells supporting homeostasis, and probably phagocytosis of the no longer functional sensory neurons.

The “outer dendritic segment” is not shown in enough detail in the figure for the purpose of this discussion. Figure 8.5.1-8 provides a more detailed view of mature mammalian dendritic segments in perspective view from Murray (page 55). Each dendritic segment ends in one or more structures called a colax. Each colax contains a group of 9(2) + 2 microvilli that emanate from its terminal surface just as they do in photoreceptor sensory neurons. The two in parenthesis indicates an individual pair of clsoely spaced cylindrical structures. The nine represents the arrangements of these cylindrical pairs around the center pair. Murray shows actual electron micrographs of these structures at 40,000x (in plan view).

The size of the microvilli remains open to discussion because of the difficulty in measuring them precisely. Murray (page 8) suggests they are about 100-200 millimicrons (nanometers) in width and 1-2 microns long. He notes others have set a minimum diameter of 50 millimicrons. Molecular structures do not exhibit hard edges at these scales. Murray has estimated the total number of microvilli at 30-40 with another investigator suggesting 5-9 microvilli from each colax. Murray notes Sclazi (1967) reported the microvilli were coated with a material identified as polysacchride in nature. More recently, the coating has been described as present in “rafts” on the surface.

Beets (1973, pg 229) has given similar dimensions for the microvilli, 0.12 microns wide and 2 microns long projecting into the Figure 8.5.1-7 Tracing of a gustatory sensory pore. neuron showing internal lemma. Cell is from the fly, Drosophila. See text. From Finger & Simon, The axons of the sensory receptor cells are 2000. even more difficult to measure, as many have diameters at or below the resolution limit of light microscopy. Many originating at sensory cells have been measured with diameters in the 0.05 to 0.5 micron diameter range. An average diameter is about 0.2 microns. Signal Generation & Processing 8- 19

8.5.1.2.3 The structure of the “sweet” lemma of the microvilli

Glendinning et al. have discussed the sweeteners (page 322), “These taste stimuli include sugars, sugar alcohols, D-amino acids, and some proteins.” They also note, “There are no generally agreed upon molecular features that are essential for a compound to elicit sweet taste.” This work takes a different view based primarily on the detailed analyses of Shallenberger & Acree and of Kier. Beidler has provided a highly conceptual view of the surface of a microvilli and a proposed mechanism of transduction41. The microvilli, or “curly hairs” of the sensory neurons is approximately xxx in diameter and xxx microns long. Slade has recently provided real time images of the microvilli on what appear to be chemosensory neurons using atomic force microscopy (AFM) (see Section 8.4.1.2.1). She notes the microvilli are typically straight and perpendicular to the dendritic surface unless knocked over during the investigation process.

At the Angstrom level, the wall of the microvilli can be considered a flat surface. The microvilli wall is specialized and consists Figure 8.5.1-8 Two views of the colax and microvilli of a sandwich of the outer or of a gustatory sensory neuron of rabbit. The plasmalemma and the reticulolemma of individual microvilli emanate from individual colax the typical dendrite in close proximity. The structures in groups of up to nine. See text. From total wall thickness is on the order of 160 Murray, 1973. Angstrom. Figure 8.5.1-9 shows the cross section of the proposed microvilli lemma of the gustatory sensory neuron. Dowhan & Bogdanov reproduce a figure remarkably similar to the right half of this figure42.

The polar heads of the globosides are the presumed receptors of the S-best sensory neurons. They are known to concentrate in “rafts” on the surface of the outer lemma. A better analogy is a “bunch of grass sprouts” emanating from the space between the bilayers of the lemma. Each exposed blade of grass consists of the polar head of the globoside molecule of the outer bi-leaf of the plasmalemma. The globosides consist of a pair of lipid chains occupying a cylinder about 9.7 Angstrom in diameter. The polar heads are much larger due to the array of sugars they contain. The diameter of the heads are unknown until their detailed structural orientation is determined. The polar heads are shown piercing the nominal surface of the microvilli into the pore space and consisting of three galactose sugars. Each galactose sugar provides one AH,B coordination site (shown as a white disk). The precise stereographic arrangement of the polar heads of the globoside molecules relative to the lipid body is unknown at this time. However,

41Beidler, L. (1971) Taste receptor stimulation with salts and acids In Beidler, L. ed. Handbook of Sensory Physiology, Vol 4/1: Olfactory Sensing. Chap 11

42Dowhan, W. & Bogdanov, M. (2002) Functional roles of lipids in membranes In Vance, D. & Vance, J. eds. Biochemistry of Lipids, Lipoproteins and Membranes. 4th Ed. NY: Elsevier Chapter 1 20 Neurons & the Nervous System the polar heads are known to form “rafts” on the external surface of the outer bilayer of the lemma. The glycophores are negatively charged at pH 7.0 and attract positively charged ligands in solution. The definition of a raft used in this work differs from that in Vance & Vance (2002, page 29). The raft of interest here consists of only the phosphatidyl molecules forming the sensory receptors of the various sensory channels that are embedded in the outer bilayer of type 4 lemma associated with the microvilli of the sensory neurons. The globosides are also known to have surfactant properties (Merck Index), although this may not be relevant when the nonpolar portion of the molecule is embedded in a liquid crystalline leaf of the microvilli lemma. As noted, the globosides are a major constituent of the type 4 lemma area of sensory neurons. The type 2 lemma area shown is capped by the electrostenolytic area providing power to this portion of the sensory neuron. It is intimately associated with an Activa formed within the type 4 area. This Activa is in intimate electrical contact with the type 2 area via the liquid crystalline water shown shaded. The electrolytic aspects of these structures will be discussed in more detail below. The microvilli of the gustatory sensory neurons, like the microvilli and cilia of all sensory neurons do not contain any pores for the passage of positive alkali or alkali ions. Nor do they contain any GPCR’s involved in the transduction process. However, they do incorporate what the community has struggled to define as “membrane rafts” supporting specialized, generally lipid based sections of outer bilayer lemma43. The term membrane raft has not gained wide adoption within the community since that time. “The definition adopted by the group was as follows: ‘Membrane rafts are small (10–200 nm), heterogeneous, highly dynamic, sterol- and sphingolipid-enriched domains that compartmentalize cellular processes.’” The electrophysiological processes associated with these areas are more important than their biochemical structures suggest. Their role will be addressed again in Chapters 2 & 3. [xxx put citation into 2 or 3 ]

43Pike, L. (2006) Rafts defined: a report on the Keystone symposium on lipid rafts and cell function J Lipid Res vol 47, pp 1597-1598 Signal Generation & Processing 8- 21 Figure 8.5.1-9 shows the electrophysiology of the “sweet” lemma shown above in greater detail. The microvilli lemma leaf closest to the dendroplasm is proposed to be primarily insulating except for the region of type 2 lemma where it is acts as an electrolytic diode. The leaf contacting the mucosa of the lingual pore is insulating in the type 1 areas and acts as an electrolytic diode in the type 2 and 4 areas. The bilayer acts as an active electrolytic device, an Activa, in the type 2 region when biased by the electrostenolytic process shown at the top of the type 2 column. The current through this Activa is controlled by the electrical potential of the hydronium of the type 2 area via the horizontal shaded area from the type 4 lemma column. In this instance, n hydronium refers to the liquid crystalline form of hydrogen-bonded water (H20) , and not the + simpler ionic form of hydrated hydrogen (H3O) discussed in many introductory textbooks. The current that passes through the diode of the outer leaf is determined by the chemical change in the polar head of the globoside (or cerboside) forming the leaf in this region. In the absence of any attack by a stimulus, the surface of the globoside in this area is polarized by the electrical potential of the hydronium at -24 mV and the potential of the polar head of the globoside combined. This net potential may have a significant impact on the stimulus/globoside interaction. The high charge gain of this configuration will be developed in the following section.

Figure 8.5.1-9 The electrophysiology of the gustatory “sweet” microvilli MOD UPPER RIGHT. This is the same histological and electrophysiology diagram as used for the olfactory sensory neurons. The shaded area on the left shows the electrical circuit configuration of the shaded histological area on the right. The C/D impulse response of this structure is shown at lower left (without considering any other circuit elements. See text.

[xxx the diagram assumes capacitive coupling instead of a dipole potential as source of drive to Activa. ] 22 Neurons & the Nervous System

The histological arrangement in the above figure is reminiscent of the simpler triune receptor concept of Hucho44, which was presented to, but has not gained traction in the pharmacology community. It differs from his type–I receptors involving a single protein bridging the lemma and his type–II receptors which involve three proteins, an ‘R’ on the external surface of the lemma, an ‘E’ on the inner surface and a ‘T’ connecting the two and defined as a coupling or G-protein. In this case, his concept is redefined as no proteins are involved. An important fact has emerged in conjunction with the above two figures and Section 3.2.4. It appears that the same carboxylic receptor, phosphatidylserine is used to both sense carboxylic acid gustaphores within the oral cavity and to capture glutamate within the fluid milieu of the body in order to power the neurons. The first mechanism (illustrated on the right of both figures) is not selective with respect to the acidophores. However, the second mechanism (on the left) requires only properly oriented glutamate be captured. The mechanism illustrated on the left employs an enzyme of the mGluR family of proteins to insure this selection process. 8.5.1.3 Electrophysiology of the sensory neurons

Teeter et al. have provided an interspecies review of the sensations generated in the gustatory modality45. However, their interpretations are all based on an undefined linear circuit network for the sensory receptors. No circuits or equations were presented. They generally assumed a change in plasma membrane impedance resulted from stimulation. 8.5.1.3.1 The electrophysiology of the “sweet” gustatory sensory neuron

The physical similarity of all of the sensory neurons is impressive. The major morphological difference is whether they have extended length axons (like the visual and olfactory sensory neurons) or whether the pedicle of the sensory neurons is on the surface of the soma (like the auditory and gustatory sensory neurons). Based on these cytological similarities, Figure 8.5.1-10 shows the proposed electrophysiology of the gustatory sensory neuron. It is virtually identical to the auditory and olfaction sensory neurons and varies only in the basic transduction area from the visual sensory neurons.

44Hucho, F. (1993) Transmitter receptors–general principles and nomenclature In Hucho, F. ed. Neurotransmitter Receptors. NY: Elsevier page 7

45Teeter, J. Funakoshi, M. et al. (1987) Generation of the taste cell potential Chem Senses vol 12(2), pp 217-234 Signal Generation & Processing 8- 23

Figure 8.5.1-10 Candidate cytology & electrophysiology schematics of the gustatory sensory neuron DUMMY ADD. The similarity to the other sensory neurons should be clear.

- - - - 24 Neurons & the Nervous System

. Earlier sections, including Section 3.2.4 provides additional details related to this figure.

8.5.1.4 Initial block diagram of the modality

Schifferstein has discussed the steps in gustation from a food technology perspective46. His totallly psychophysical steps appear remote from the physiological steps identified in this work.

Figure 8.5.1-11 shows a provisional block diagram of the gustatory modality. Frame A shows a simple linear representation of the gustatory modality that parallels that of xxx. However, the functional descriptions at the gaps in the line segments have been changed. Frame B shows a first alternate of greater complexity. Frame C shows a potential signaling architecture analogous with that of the visual modality of Section 8.2. and/or Chapter 17, Section 17.1.4 in “Processes in Biological Vision.” This frame separates the qualities associated with the individual channels of gustation from the overall intensity sensation processed through a separate “summation channel.” This summation channel will be given the labe, the R-channel in analogy to that designation in the visual modality. Thus, the top level architecture of the gustatory modality will be described as consisting of the A-Path, the G-Path, the N-Path, the P-Path and the R-Path at the stage 1 and stage 2 level. Whether there are difference channels between the signals generated by the stage 1 receptors will be developed below.

46Schifferstein, H. (2003) Human perception of taste mixtures In Doty, R. ed. (2003) Handbook of Olfaction and Gustation, 2nd revised and expanded edition. NY: Marcel Dekker Chap 38 Signal Generation & Processing 8- 25

Figure 8.5.1-11 Provisional block diagram of the gustatory modality ADD. C; a generalized architecture for gustation analogous to that described within the vision modality. See text.

Frame B expands the conceptual character of Frame A to illustrate xxx. 26 Neurons & the Nervous System

8.5.1.4.1 A proposed top level architecture of the gustatory modality

Frame C of the previous figure shows the probable character of the gustatory modality based on its close analogy to the visual modality. The analogy has been proposed based on the detailed analyses of the gustatory modality developed in this chapter. From a global perspective, it appears very likely the architecture is analogous to that of the visual modality but with the signal processing in stage 2 adapted to the specific needs of gustation. A common question is whether the neural signals recorded at the output of the sensory neurons correlate well with the perceived tastes of the subject. Frank has provided a comparison of the sensed and perceived responses using factor analysis that shows this is a close correlation in hamsters47. Her imposing a spherical overlay on her data leads one to believe the distance from the axes of various stimulants is more limited as their appearance on the graph moves toward the perimeter. A more open three-dimensional taste sensation space is suggested in this work (Section 8.5.3 xxx). Frank’s sensory recordings were at the chorda tympani nerve (generally action potential pulses within stage 3). The conclusion can be drawn that the stage 2 through stage 4 activity is designed to project and optimize the presentation of sensory neuron output data to be analyzed by the prefrontal cortex of stage 5.

As noted in Section xxx, the stage 2 signal processing within gustation may be limited. It appears any stage 2 signal processing does not occur in the taste buds or the oral cavity. Some signal processing may occur at the immediate output of the stage 1 sensory neurons. It is more likely this signal processing occurs in the nucleus solitarius tract (NST) or possibly the parabrachial nucleus (Sections 8.5.2 & 8.5.8).

There are suggestions of a differencing function between the stage 1 output signals within stage 2 based on several discussions in the MDS literature of gustation. However, it is not well isolated from the potential stage 4 information extraction function employed in developing the saliency map of gustation. The proposed saliency map architecture is strongly dependent on the basis functions of the MDS analyses presented in the literature (Section 8.5.2.4).

A differencing function within stage 2 is also strongly suggested by the apparent broad tuning of the individual sensory receptors of stage 1. Efficient information extraction in stage 4 would suggest that some pattern of direct projection of the A–, G–, N – & P– signals accompanied by at least a partial set of differencing signals is utilized (Section 8.5.5). [xxx confirm this relationship ]

No data has been found in the literature describing the signals from the temporal lobe of cerebral cortex describing the form of the data presented to the saliency map. 8.5.1.5 The chemistry most important to gustation

47Frank, M. (2000) Neuron types, receptors, behavior, and taste quality Physiol Behav vol 69, pp 53–62 Signal Generation & Processing 8- 27

8.5.1.5.1 The chemical families of carbohydrates involved in gustation

The complexity of carbohydrate chemistry is truly mind-boggling. The smaller portion of this class related to gustation remains in this category. The name carbohydrate was assigned based on a major misunderstanding that they were hydrates of carbon. A modern definition of a carbohydrate is given by Robyt48; “a carbohydrate is a polyhydroxy aldehyde or ketone, or a compound that can be derived from them by any of several means including, 1. reduction to give sugar alcohols, 2. oxidation to give sugar acids, 3. substitution of one or more of the hydroxl groups by various chemical groups, for example hydrogen may be substituted to give deoxysugars and amino group [NH2 or acetyl-NH] may be substituted to give amino sugars, 4. derivatization of the hydroxyl groups by various moieties, for example, phosphoric acid to give phospho sugars, or sulfuric acid to give sulfo sugars, or reaction of the hydroxyl groups with alcohols to give saccharides, oligosaccharides and polysacchaired.” Lindhorst gives the clearest and most extensive description of the terminology used to describe this very large region of organic chemistry49. He stresses the conformational chemistry of the family. The conformation of the molecules is critically important to the role of the carbohydrates in gustation, to the extent that Haworth projections, and other projections obfuscate the required structural elements of the molecules.

8.5.1.5.2 Chemical families pertinent to gustatory receptor identification

The following discussions will focus on modifications to the outer bilayer of the cilia and villi of sensory neurons as leading to the likely receptors of gustation. The chemistry in this area can be extremely complex, and involve mechanisms not familiar to the majority of researchers.

Several authors of texts on carbohydrates make gratuitous statements about the role of protein-based receptors in gustation and olfaction (without justification or citations). It will become clear in this work that proteins are not required in the process of chemical sensing by the neural system.

The complexity in the literature of potential lemma chemistry is partially due to the different time periods in which various investigators have worked. It is also due to the difficulty in controlling and identifying and avoiding uncontrolled chemical reactions involving constituents present in very complex molecules.

When discussing complex organic chemistry, it is common in textbooks and journal articles to address the degree of saturation in the bond structure in a cavalier manner. A more refined approach is needed in this work. Figure 8.5.1-12(A) discusses the major possibilities. The case of a saturated hydrocarbon is a boundary condition. The molecule contains no double or triple bonds. However, an unsaturated hydrocarbon remains largely undefined. An unsaturated hydrocarbon with only one double bond among many double bonds appears to retain most of the properties of its fully saturated sibling. However, molecules with alternating single and double bond patterns can exhibit significantly different electronic properties. The brain and neural system is known to contain many molecules exhibiting a repeating isoprene structure where carbon-carbon double bonds are separated by two carbon-carbon single bonds, a structure associated with the squalenes Alternately, .visual sensory organs of the neural system are known to contain large numbers of molecules exhibiting a repeating isoprene structure where carbon-carbon double bonds are separated by one carbon-carbon single bond. Such

48Robyt, J. (1998) Essentials of Carbohydrate Chemistry. NY Springer page 2

49Lindehorst, T. (2003) Essentials of Carbohydrate Chemistry and Biochemistry, 2nd Ed. NY Wiley-VCH Chap 2 28 Neurons & the Nervous System molecules are known as terepenes and are referred to as conjugated hydrocarbons over a significant and defined portion of the entire molecules. They include the very important retinal derivatives, the rhodonines, which exhibit significant intra– and inter–molecular electronic conductivity when present in the liquid crystalline state. In the case of the rhodonines, the conjugation even extends to the O– and OH– ligands of the structure and explains the multiband spectral performance of the visual system. Figure 8.5.1-12(B) attempts to normalize the terminology used to describe the chemistry involved in sensory neuron receptor identification. The goal is to define traceable names to the chemical receptors providing transduction of the presence of stimulants into graded electrical signals. To this end, the figure attempts to rationalize the words and figures of Lehninger (1970) with those of Benjamins et al. (2006). It employs an indenting style used to save space when using long descriptors. The chemicals of interest are at the convergence of two very large general families known as the phospholipids (phosphoglycerides) and the sphingolipids (containing no glycerol). They both contain two long chain hydrocarbons and a polar head group. The polar head groups typically exhibit a dipole potential and dipole moment because of their electrical asymmetry at the atomic constituent level. As normally conceived, the hydrocarbon chains of both families are nonpolar and electrically insulating. However, a thesis of the Electrolytic Theory of the Neuron is that at least one of the chains in the above phospholipids and sphingolipids can be present in a conducting (or semiconducting, i.e. directionally conducting) form. In this form, a dipole potential can be exhibited between the two ends of the overall molecule (particularly when the molecules are present in the highly oriented liquid crystalline state found in lemma bilayers). Signal Generation & Processing 8- 29

Figure 8.5.1-12 A normalized etiology of chemicals relating to proposed gustatory sensory receptors.

Electrically conductive hydrocarbon chains in the liquid crystalline state are well known, and are widely used in this age of the liquid crystal display technology. The conductivity of hydrocarbon chains is primarily a function of the degree of saturation (unsaturation) of each chain. Fully saturated and minimally unsaturated chains are electrical insulators. However, unsaturated chains, particularly of the fully conjugated form, exhibit significant conductivity because of their sharing of excited electron π-bonds along the length of their chains. In an appropriate electrolytic circuit configuration, these conductive chains are able to transfer a potential at their interface with their polar head to the extreme end of their hydrocarbon chain. It is proposed that the value of this potential is used within the neural system to identify and quantify the stimulants used in gustation. 30 Neurons & the Nervous System

Textbooks tend to show all phospholipids and sphingolipids as consisting of fully saturated fatty acid chains as a matter of simplicity and convenience. Benjamins et al. in particular show a range of saturated and unsaturated fatty acids of neurochemical interest. Their tabulation focuses on products of the isoprene rule, the squalenes that exhibit conjugate carbon-carbon bonds at every third carbon. A similar product of the isoprene rule, the terpenes exhibit conjugate bonds at every other carbon-carbon bond. The terpenes are known to be conductive and their may be other configurations providing conductivity, possibly involving cross-linking between the two parallel long chains. Four conductive phosphoglycerides have been identified as likely receptors in gustation. They have been identified on the right as being the receptors in of the four primary sensory channels of gustation. Frame C: The literature contains discussions of cerebrosides, of both the glycophospholipid and glycosphingolipid (defined as containing no glycerol) types containing one sugar ligand. Cerebrosides containing oligosaccharide complexes have been named gangliosides and globosides (the former containing sialic acid while the latter does not. As one travels down this figure, it is not always clear how precisely and exclusively the terms have been defined. The term gangliosides in particular seems to be used to describe both galtanophospholipids and galtanosphingolipids in the journal literature. Benjamins et al. have noted “galactocerebroside, in which galactose is β-glycosidically linked to cerebroside, constitutes about 16% of total adult human brain lipids.” Such a high concentration suggests it is a structural component rather than a sensory or signaling component. They note the galactocerebroside appears primarily in the white matter, especially in myelin, a passive electrically insulating material.

The routinely reported presence of the cerebrosides and gangliosides at high concentrations within the brain does not necessarily indicate they are functional within the neural signaling system. There is considerable potential for these molecules to participate as structural components of the neuron walls. The reduced presence of oxygen in the case of the sphingolipids also suggests they are not as electrically important as the phospholipids.

Benjamins et al. have provided a set of shorthand labels for many of the chemicals that will be discussed in detail later in this chapter50. Most of the phosphoglycerides will be described using their more specific labels; the carboxyphosphoglyceride of interest is described as phosphatidyl serine or PtdSer.

The aliphatic alcohols are missing from the above table. It is generally believed that these molecules do not stimulate the gustatory sensory receptors of humans. However, they apparently do stimulate the olfactory sensory receptors. This situation suggests frequent experimental difficulties in isolating the sensory neurons of the two modalities in the laboratory. The difficulties frequently lead to confusion in the literature over the role of these molecules in gustation. 8.5.1.5.3 The special case of the saturated aliphatic alcohols and aldehydes

The literature is conflicted concerning the perceived taste of the saturated aliphatic alcohols and aldehydes. With respect to gustation, the question is how do these molecules stimulate a GR in the absence of any intrinsic pair of atoms/electrophobic elements capable of participating in a DACB. The answer is unusual, the lower alcohols and aldehydes are appreciably soluble in water. This solubility is accounted for by the fact these chemicals become hydrogen bonded to the water (including that of the saliva) immediately. The distance between the two oxygen atoms of the hydrogen bond is 2.703 Angstrom. These two oxygen atoms are capable of forming a DACB with GR 2. The result is a perceived sweet odor for the solution as most frequently reported.

50Benjamins, J. Hajra, A. & Agranoff, B. (2006) In Siegel, G. et al. eds. Basic Neurochemistry, 7th Ed. NY: Elsevier Chapter 3 Signal Generation & Processing 8- 31 It is critically important when discussing the alcohols, aldehydes and organic acids (all polar chemicals containing oxygen atoms with incomplete orbital shells) to recognize that when they are placed in solution the oxygen atoms will form hydrogen bonds with water. Thus the active species within the saliva is not the molecule typically shown in text books but the hydrated form exhibiting two oxygen atoms separated by 2.703 Angstrom and capable of forming a DACB with OR 2 of the gustatory modality. The form could be described as R–O- -HO or as an organic hydroxyl ether where the O- -H—O structure is a hydrogen (London) bond. There is also a problem of obtaining alcohols and aldehydes with an impurity level below a few parts per billion. Ho, Johnson & Leon documented this problem in 200651. They showed that reagents available from reputable suppliers can not be relied upon in gustatory research due to undocumented impurities. 8.5.1.5.4 The unique role of the hydrated sodium ion in gustation

As will be seen below, the sensing of sodium and its salts play a minimal if any role in fish gustation. However, as the vertebrates moved onto land, the search for salt became a major nutritional requirement. It is quite possible, the minor change in the G-path “sweet” receptor to make it sensitive to the hydrated sodium ion was the best and easiest evolutionary solution to this problem.

Sensing the inorganic sodium ion by the new N-path GR of the gustatory modality depends entirely on the high degree of hydration immediately assumed by the sodium ion when it is formed in solution. This hydrated sodium ion looks to the gustatory receptors like a typical organic molecule with a specific distance (the d-value defined below) between the oxygen orbitals of the water molecules enclosing the sodium ion. See Section 8.5.4.4. 8.5.1.5.5 Inorganic acids and astringents excite the nocent (pain) modality

Activity related to the gustatory modality began in antiquity. During the 19th and 20th Centuries, that activity has been primarily within the psychophysical and food science communities. No significant theory of gustation appeared. As a result, the current terminology has evolved from the earliest psychophysical observations. The concepts of sweet, sour, salty and bitter evolved into a formalism of S(sweet)-best, H(acid)-best, N(salty)-best and Q(bitter)-best sensory channels. Unfortunately, this formalism does not correspond to the operation of the neural system. Specifically, the gustatory modality of the neural system does not sense the hydrogen ion level present in the saliva ( a Lowry-Bronsted acid concept, 1923), it senses the presence of the carboxyl ligand of organic compounds according to the Lewis (1916) definition of an acid, The situation is more complex, the sensing involves two coordinate bonds, each involving the sharing of an electron pair, but no net charge transfer between the stimulant and the receptor. Thus, no chemical reaction is involved. The coordinate bonds are fragile and temporary in character. The actual operating mode of the gustatory modality suggests a redefinition of the four channels described in this paragraph (See Section 8.5.2.6 xxx).

The role of inorganic (Lowry-Bronsted) acids and a variety of astringents is to stimulate the pain sensing portion of the nocent (nociceptor) modality of the neural system. The nocent modality is addressed in Section 8.7 following the olfactory modality of Section 8.6. Some organic compounds also affect the nocent modality. Carbolic acid (phenol), although called an acid since ancient times, is actually a very strong organic base. It is so caustic that is is highly poisonous (in fact, destructive of organic tissue). 8.5.1.5.6 Definition of specific stereo–molecular structures

Understanding the gustatory modality involves considering polyfunctional compounds, molecules with multiple properties normally associated with separate individual molecular

51Ho, S. Johnson, B. & Leon, M. (2006) Long hydrocarbon chains serve as unique molecular features recognized by ventral glomeruli of the rat olfactory bulb J Comp Neurol vol 498, pp 16–30 32 Neurons & the Nervous System families in textbooks. A major difference in properties is associated with the cis– versus trans– characterization. This characterization is further defined when used in conjunction with cyclic compounds. [xxx check if following their convention ] Spanning a long time, the literature is inconsistent when naming “sweet” stimulants. This work will show 1,2 cis-glycol or a partially dehydrogenated derivative is the primary glycophore of gustation. When present in a sugar, the oxygen atoms of this gustaphore are generally present as the OH–3 and OH-4 groups with one of the hydrogens removed. Glycophore is the term used by Shallenberger and his associates for the “sweet” gustaphore. This work will follow their convention. This allows the 1,2 cis-glycol to be differentiated from the 1,2 trans-glycol; the latter is employed in a glycolipid susceptible to the hydrated sodium ion natrophore of this work. In the older carbohydrate literature, cis– and trans– are used to describe the direction relevant hydroxyl ligands are pointing, rather than the side of the molecule they are on relative to each other. The precision with regard to the direction they are pointing is not quantified. The precise direction they are pointing is a critical factor in gustation. Therefore, a more modern and precise description is required. In this work, a diol (including a hydrated carboxyl) incorporated into a ring structure can have its relevant hydroxyl groups pointing perpendicular to the plane of the ring (axially) and be described as either axial-trans- or axial-cis- relative to that plane. similarly, the diol incorporated into a ring structure can have its relevant hydroxyl groups pointing generally outward (equatorially) in the plane of the ring. In this case, they can be pointing marginally above or marginally below the plane of the ring by up to a nominal 19 degrees in some sugars, even though most Newman diagrams suggest +/–30 degrees. If the relevant hydroxyls are both pointing in the same direction relative to the plane of the ring, they are described as equatorial-cis-. If one is pointing above the ring and one below the ring, they are described as equatorial-trans- .

The actual total dihedral angle in 3D space between the adjacent equatorial-trans Oxygen atoms in the saccharides is given as 64.9 degrees in most Jmol renderings of this molecule.

Figure 8.5.1-13 summarizes the notation used in this work and required to interpret chemical sensing adequately. The dihedral angles associated with the Newman representation are discussed in greater detail in Glusker et al52. Glusker et al. are the authority with respect to stereographic representation of molecules. They specifically note, “There is considerable misuse in the literature of the term ‘dihedral angle’ (used where the term ‘torsion angle’ is intended). Their figure 12.6 demonstrates that the Newman Diagram is a projection of the location of the atoms and groups of atoms onto a 2D surface drawn perpendicular to a particular axis drawn between two anchor atoms. The defined torsion angle (θ)of this projection is typically smaller than the dihedral angle (φ) in 3D space relative to the same two anchor atoms. However, the definition of the two angles results in the equation; dihedral angle = 180 – torsion angle (figure 12.11.

As noted in Section 8.4.1, the Newman Representation is not adequate for describing the ligand conformations of interest in chemical sensing. Precise three-dimensional distances between the major orbitals must be determined. No label could be found in the pedagogical literature for the 120 degree rotation between the two methyl groups in n-butane or other simple molecular forms.

52Glusker, J. Lewis, M. & Rossi, M. (1994) Crystal Structure Analysis for Chemists and Biologists. NY: Wiley- VCH Page 462-480 Signal Generation & Processing 8- 33

Figure 8.5.1-13 Conformational notation applied to a Newman representation. As noted earlier, the Newman representations are 2D projections of 3D conformations. While the equat-cis conformation appears similar to the eclipsed conformation, the 3D distance between the methyl groups are different. Similarly, the 3D distances between the methyl groups in the equat-trans conformation are different than in the gauche conformations. Expanded from Morrison & Boyd, 1971.

While the term glucophore is frequently used in place of glycophore, one should not draw the conclusion that the gustaphore is based exclusively on glucose. 8.5.1.5.7 Equilibrium in the context of gustation–a brief review

The subjects of equilibrium, of equilibrium states, the release/absorption of free energy as a result of chemical activity and changes in the concentration or reactants plays an important but poorly delineated role in gustation. The concepts are difficult to comprehend by many investigators during their first encounter. In the case of gustation, the very low free energies involved may require the introduction of thermal noise related to the average energy state of any molecule(s) during the short term. The subject of equilibria in chemistry and changes in free energy associated therewith have not spawned a single amorphous community of investigators. Studying the literature is complicated by the fact that the symbology used when discussing changes in the free energy have used the labels, ΔF, ΔG and ΔH for the same quantity over the years. In addition, as noted by xxx, certain disciplines have used a negative value for each of these quantities to indicate the release of energy while others have omitted the minus sign. 34 Neurons & the Nervous System

Adamson has provided definitions for some of these terms and a summary (page 218) “E= 3/2CRT, the root mean squared (RMS) velocity of a gas molecule (page 55).” It is equal to the internal energy of a definite portion of matter (page 121). Adamson uses a capital and italic E when referring to a molar basis. “H, a molar quantity called the enthalpy, is equal to E + PV (Page 130).” ΔH describes the total work, both PV and chemical, involved in the system. “G = H– TS where ΔG is the maximum chemical work involved in H (page 214).” T is absolute temperature and S is the entropy of the system. G is also known as the Gibb’s free energy. G0 is generally defined as the molar free energy of a gas in its standard or reference state of unit pressure ordinarily taken to be one atmosphere (page 256).” Both E and H are described as state functions where P, V & T of the gas equation are also described as state functions (page 130). This label us used to describe a change that is only defined by the function values at two end points of some change that is unaffected by the path of integration used to obtain these values. The term just-noticeable difference (JND) when used in gustation and olfaction is frequently tied closely to the ratio of the change in stimulus intensity to the RMS (root-mean-squared) change in the noise level due to thermal agitation at the temperature of the experiments. The RMS concept will not be addressed here but it is widely used in engineering and physics. The major question becomes whether the noise limit of a subject in gustation is due to the quantity, E related to the vibration of the basic molecules involved or is due to excess noise contributed by the sensory neural system?

Introductory college chemistry texts typically stress the dynamics of molecules at temperatures above absolute zero and define several types of equilibrium,

1. Physical equilibrium between molecules at the border between different states of mater or in the undissolved and dissolved states.

2. Ionic equilibrium between the non-ionized and ionized forms of an ionic molecule (typically while in solution).

3. Chemical equilibrium between a chemical and its constituent ionizable constituents or between a mixture of two or more ionizable chemicals and the resulting mix of ionized components and unionized chemicals.

To satisfy the needs of biological chemical sensing, it is probably necessary to define additional equilibrium conditions related to coordinate chemistry and present in isolation from the associated equilibrium conditions involving some of the individual participating constituents such as,

4. Coordinate equilibrium between ions, hydrated ions, and non-electrolytes, and their coordinate chemical bonding to the receptor material of sensory neuron receptors. When speaking of coordinate equilibrium, the energy levels involved are so low that the quantum mechanical variations in the energies of the various components at a given moment may be important. The result is a variation in the instantaneous state of equilibrium on very short time scales compared to the time scales employed in most gustatory (and olfactory) investigation protocols. Equilibrium involves three major characteristics; C There are always two opposing tendencies. C Although the two opposing tendencies neutralize each other at equilibrium, both tendencies are still in full operation. C A slight change in the conditions produces, a corresponding small and reversible change in the state of the system. This is the chief characteristic of equilibrium. Signal Generation & Processing 8- 35 In the context of gustation, equilibrium may be separated into multiple stages; equilibrium between solid and liquid phases, equilibrium between solvated and hydrated molecules within a liquid environment, equilibrium between molecules in solution and those in a captured context relative to the sensory neuron receptors, and equilibrium between molecules in solution and the same molecules present in the adjacent vapor state. Equilibrium reactions may be self terminating and run to conclusion when one of the products formed is insoluble, when one of the products is volatile or when water is one of the products formed. For most practical situations, water can be considered un-ionizable (typically 2 parts in 109 and giving rise to the concept of pH). Chemical equilibrium is typically influenced by changes in temperature, changes in pressure or changes in the concentration. The Law of Mass Action (or Law of Molecular Concentration) may be stated as “The speed of a reaction is proportional to the product of the molecular concentrations of the reacting substances.” An important parameter for the equilibrium at a given temperature, pressure and concentration is the equilibrium constant, K, given as the ratio of the speed of the reaction proceeding to the right divided by the speed of the reaction proceeding to the left in an equilibrium reaction written with a double arrow , W , over or in place of the equality sign. K is typically defined as the equilibrium value after a sufficient time interval to establish equilibrium. Pressure is not an important factor for equilibria occurring within the fluid state of matter.

In the case of ionizable molecule, the term ionization constant is frequently encountered instead of equilibrium constant. There are significant variations in the value of the ionization constant for special chemical situations. Two important situations are the polyprotic acids, sulfuric acid and phosphoric acid. These acids ionize in steps with each step in their ionization exhibiting a different ionization constant.

A search of the literature for the description of equilibria in the context of coordinate chemistry has not provided any guidance.

The mathematical definition of K follows the pattern shown in Figure 8.5.1-14 for simple molecules in solution. The top equation shows the simplest case of a molecule ionizing. It employs a double arrow indicating the reaction is reversible but will attain an equilibrium for a given temperature and total concentration. The reaction proceeds to the left based on the product of the concentrations shown on the right and equal to KR. The reaction will proceed to the right based on the concentration of the concentration of the complete molecule and equal to KL. The equilibrium constant for the reaction is given by K = KR/KL Row II shows a reaction consisting of two reactants on the left and three on the right. The equilibrium constant reflects the concentrations of each of the reactants. The expression for K in the right column also indicates that; for the reaction to proceed to the right, all of the terms on the left, A and B, must colllide with each other. For the reaction to proceed to the left, all of the terms in the numerator on the right, C, D & E must simultaneously collide with each other. This condition holds true for all of the forms shown in the figure. Row III shows a more complex reaction involving more than single units of each reactant. Following the style of the second row, the equilibrium constant of this type reaction involves recognizing the amount of each reactant present. The product of two identical terms can be shown as the square of the individual term as shown at far right. The value of the equilibrium constant indicate how fast the reaction reaches equilibrium. A value of K =100, 1000 or more indicates the reaction progresses rapidly to the right. If K < 1, the reaction proceeds to the left. 36 Neurons & the Nervous System

Row IV shows a more complex situation involving a polyprotic acid. Phosphoric acid ionizes in multiple steps. As a result, each step exhibits its own equilibrium constant. The nominal equilibrium constants for the three reactions are; 7.5 x 10–3, 6.3 x 10–8 & 3.6 x 10–13. The equilibrium constants usually represent the concentrations of the reactants at the beginning of a reaction, time / zero. The characteristic of a Figure 8.5.1-14 The equilibrium constant in different contexts ADD system in chemical equilibrium is that the speeds of the forward and reverse reactions have become equal and K = 1.00 during this period.

The question of major interest here are; C How are the equilibrium constants of coordinate chemistry calculated? C How is the equilibrium constant for a DACB calculated? C Can it be assumed the two hydrogen bonds are formed simultaneously? The question arises because of form III and IV in the above figure. Form III calls for all of the constituents of a reaction proceeding in one direction to collide with each other simultaneously. Form IV shows that a reaction requiring the presence of a component of the ionization of a polyprotic acid (or similar structure like a disaccharide becoming hydrolyzed) may be delayed until the necessary reactant is available.

The calculation of equilibrium constants for the step-wise dissociations of weak acids and bases becomes very unwieldy and graphic methods are usually employed53. 8.5.1.5.8 The change in free energy associated with a DACB –a brief review

If it can be determined how the association and dissociation of a DACB progresses, it is possible to consider how that association or dissociation might be impacted by other sources of energy that might drive the reaction(s). The formation of Each hydrogen bond of a DACB would be expected to involve a change of energy of less than 5 kcal/mole. This is the value typically given for the hydrogen bonds associated with water. Adamson54 has given that value when associated with water. He has also given a higher value of about 7 kcal/mole(0.3 eV) for each of the two hydrogen bonds in the dimer of acetic acid and suggested a net change of 14 kcal/mole (0.61 eV) for dissociation of the dimer. These are remarkably low energies compared to the conventional chemical bond between hydrogen and oxygen of 119 kcal/mole ( 5.1electron volts @1eV = 23,060 kcal/mole). Such small energies are typically associated with changes in conformation within a molecule as illustrated in Section 8.5.1.5.6. A few molecule form or decompose in these energy ranges.

53Adamson, A. (1973) A Textbook of Physical Chemistry. NY Academic Press page 551

54Adamson, A. (1973) A Textbook of Physical Chemistry. NY Academic Press page 298 Signal Generation & Processing 8- 37 Hydrogen iodide has a standard enthalpy of formation, ΔH0 of only 6.2 kcal/mole at 298Kelvin. xxx gives the best value as ΔG = – xxx at xxxC. This is an exceedingly small value relative to the energy fluctuations of a molecule at room or endothermic animal body temperature. . 8.5.1.5.9 The primary question regarding the transduction mechanism of gustation

The primary question regarding the mechanism supporting excitation and de-excitation process of a chemical sensory neuron receptor in gustation and olfaction is; How is the DACB initially formed and how and why is the DACB broken?. The answer is most likely given in terms of the molecular quantum mechanics described in the previous two subsections as they apply to solutions. At the detailed level, this description may be quite complex. However, the first order mechanism may be tractable. This subject will be pursued in Section 8.5.3.3. 8.5.1.6 Summary of the gustatory modality hypothesis

The following material will develop the theory of the gustatory modality as an explanation of the empirical data in the literature based on the Electrolytic Theory of the Neuron and transduction via a coordinate chemistry mechanism for each of the sensory channels of taste. The theory is comprehensive. It encompasses the sensing of all four classical sensations of taste. It will be labeled the Electrolytic Coordinate Chemistry Theory of Gustation.

The theory of gustatory sensing recognizes a three dimensional taste space with the four classical sensations occupying four of the available six vertices. The space is expandable in either of two ways, if additional data appears supporting a fifth channel related to the label “umami.” This sensation could occupy an available vertex in the three dimensional space or it could require expanding the space to a fourth dimension (as is the case in the visual modality when the sensing of ultraviolet light is recognized– Section xxx).

Moving from a conceptual understanding of gustation to a more formalized theory has provided new insights. These insights can be grouped as follows.

[xxx subdivide the following into groups with rational names. ] Group 1 Gustatory Transduction, global aspects C Gustatory transduction involves a coordination chemical process of very low energy that is totally reversible. No chemical “reaction” is involved in the transduction process. C Transduction occurs totally external to the sensory neuron lemma. No stimulant components pass through the cell wall. C The transduction mechanism uses a set of phospholipids, widely recognized as being present in neural tissue but of previously unknown utility, as the gustatory sensory receptors (GR’s). C The transduction mechanism employs a coordinate chemical union between the sensory receptors and the stimulant that is stereo-chemically specific. Group 2–Gustatory Transduction, Details • The phospholipids forming a specific active region of a GR are present as a liquid crystal portion of the dendritic tree, known cytologically as the cilia. • There are four specific types of GR in gustation that selectively isolate the individual gustaphores found among a wide range of stimulants. • The individual gustaphore is associated with the appropriate GR through a dual anti-parallel coordinate bond (DACB) arrangement. 38 Neurons & the Nervous System

• The perception of a taste is the result of a 2-step transduction process involving initially the isolation of gustaphores affecting a given GR type and subsequently the measurement of the dipole potential of the individual stimulant when bonded to the GR. • The net dipole potential change due to a specific stimulant may be influenced by the specific electrostatic field presented to the GR, resulting in the perception of super-sweet and super- bitter qualities. Group 3 Additional details of transduction C The transduction mechanism results in a change in dipole potential of the phospholipids that are known to be polar. This change in dipole potential is sensed by the first Activa of the sensory neuron. • The net change in dipole potential of the individual phospholipid GR is highly sensitive to nearby electrostatic fields, particularly those of the associated gustaphore. C The histology and physiology of the sensory neurons following the transduction mechanism is the same as that of other sensory neurons. They lack an external axon like the auditory sensory neurons and share a microvilli structure similar to the visual sensory neurons. C The sensory channels of the gustatory system exhibit the same excitation/de-excitation function and the same adaptation functions as other sensory modalities. Only the time constants vary to accommodate anatomical and vascular conditions. C The gustatory modality exhibits the same neural architecture as all other sensory modalities but the stage 2 signal processing occurring outside the cranium appears to be minimal.

Group 4 The Perception of stimulants within the neural system C The gustatory modality uses a small group of sensory receptors (nominally four) to sense a wide variety of chemical structures (in the thousands). C The small number of sensory channels operate independently and can therefore be represented as statistically independent and therefore orthogonal, resulting in a three- dimensional taste space within the neural system. This space can be unfolded into a single dimensional space equivalent to the spectra of the visual and auditory modalities. C The unfolded representation of the taste space highlights the common feature wherein many stimuli are able to stimulate multiple sensory channels. C The “salty” sensory receptor is in fact only sensitive to the fully hydrated sodium ion; the chlorine ion plays no role in the transduction mechanism. C The “sour” sensory channel is primarily sensitive to the carboxyl group of organic (Lewis) acids.

C The response of the gustatory modality to inorganic acids and various astringents (alkaline- earth salts) appears to be primarily via nociceptors. C The “sweet” sensory channel is primarily sensitive to a group, consisting mostly of either oxygen or hydroxyl oxygen at positions 1,2 of a cis-glycol, generally in a cyclic structure. [See a more precise definition of the 1,2 equatorial-trans-glycol given below] C The “sweet” sensory channel exhibits an overlay mode of operation, involving a three-point stereochemistry, that is supersensitive to man-made sweeteners (super-sweet glycophores. C The “bitter” sensory channel is primarily sensitive to organics with two orbitals capable of sharing an electron-pair and separated by a distance of 4.746 Angstrom. C The “bitter” sensory channel also exhibits an overlay mode of operation, involving a three-point Signal Generation & Processing 8- 39 stereochemistry, that is supersensitive to primarily man-made stimulants (super-bitter picrophores). C The frequently proposed independent “umami” sensory channel cannot be confirmed based on the available empirical data and rigorous multidimensional analysis procedures. The dominant chemical associated with this designation, monosodium glutamate, is unusual in that it exhibits multiple gustaphores, hence its ability to stimulate the salty, sweet and acid channels of gustation simultaneously. Eliel et al. have discussed the different kinds of molecular models available in both physical and computerized form55. No effort will be made here to standardize on one type of diagram or physical representation. [xxx see final version of why four GR’s paper for broader discussion) Simple line diagrams of chemicals leading up to include Haworth models will be used, augmented by computer models of specific chemicals prepared using Molecules-3D Pro, v2.10 from Molecular Arts. The specific geometric relationships between stimulants and receptors may not be shown faithfully because of the author’s limited familiarity with the program. However, the critical dimensions will be specified as clearly as possible. Figure 8.5.1-15 describes the sensory receptors related to taste. They are all phosphatidyl lipids derived from the normal lipids forming the outer bilayer of the microvilli of the sensory neurons. They differ primarily in the spacing between certain atoms in the structures that allow coordinate bonding with a variety of stimuli in the AH,B or AH,B,X configuration (Section 8.5.3).

The figure is complex. However, by following the logic of Dowhan, it provides a variety of answers. It begins at upper left describing the chemical structure of a typical phospholipid and its ability to bond with a variety of terminal groups through esterification. The diagram has been annotated here to define one of the lipid molecules to be electrically conductive, most likely through complete conjugation of its fatty acid chain, like the isoprene chain of Vitamin A. The ligand in the box is labeled on the right with the trivial name choline, or formally phosphatidylcholine (PtdCho). This ligand plus the next one below it, ethanolamine, are the principle phospholipids of the lemma of all cells. The examples of ligands below these two are those commonly found in the lemma of neurons and particularly sensory neurons of the gustatory system. There are four of these ligands, each supporting a separate sensory channel of gustation leading to the sensation of sourness (presence of an organic acid or acidophore), sweetness (presence of a glycophore), saltiness (presence of a hydrated ion of sodium or natrophore) and bitterness (presence of a picrophore).

Each of the ligands exhibits a distinctive spacing between two or more orbitals susceptible to coordinate bonding with stimulants exhibiting a similar spacing between elements supportive of coordinate bonding. In the case of the sourness channel, the most important orbitals of phosphatidylserine (PtdSer) exhibit a spacing of 2.34 Angstrom when nonresonant (2.07 Angstrom when resonant). This is the same spacing as the two oxygen atoms of the carboxylic acid family. Thus, any carboxylic acid is amenable to coordinate bonding to PtdSer and eliciting a sensation of sourness. The proposed location of this bonding is shown by the small rectangular bracket next to the two oxygen atoms. PtdSer is therefore defined as the receptor of the sensory neuron initiating the sourness, or acid, channel of gustation.

Carboxylic acid is only mildly acidic, indicating it is not highly ionized in water. The dominant species is the nonresonant carboxylic acid form with a complete hydroxyl group. The spacing between the oxygen atoms in this configuration is 2.34 Angstrom. Phosphatidylgalactose (PtdGal), also known as galactoceramide (CerGal) exhibits a spacing of 2.86 Angstrom between OH–3 and OH–4. This spacing is compatible with a very large group of sweeteners containing a 1,2 cis–glycol group that can dual-coordinate bond with this receptor and elicit a sensation of sweetness.

[xxx may duplicate previous section 8.5.1.xxx ] Spanning a long time, the literature is inconsistent when naming “sweet” stimulants. This

55Eliel, E. Wilen, S. & Doyle, M. (2001) Basic Organic Stereochemistry. NY: Wiley Interscience pg 25 40 Neurons & the Nervous System

work will show 1,2 cis-glycol or a partially dehydrogenated derivative is the primary glycophore of gustation. When present in a sugar, the oxygen atoms of this gustaphore are generally present as the OH–3 and OH-4 groups with one of the hydrogens removed. Glycophore is the term used by Shallenberger and his associates for the “sweet” gustaphore. This work will follow their convention. This allows the 1,2 cis-glycol to be differentiated from the 1,2 trans-glycol; the latter is employed in a glycolipid susceptible to the hydrated sodium ion natrophore of this work. In the older carbohydrate literatue, cis– and trans– are used to describe the direction relevant hydroxyl ligands are pointing, rather than the side of the molecule they are on relative to each other. The precision with regard to the direction they are pointing is not quantified. The precise direction they are pointing is a critical factor in gustation. Therefore, a more modern and precise description is required. In this work, a diol (including a hydrated carboxyl) incorporated into a ring structure can have its relevant hydroxyl groups pointing perpendicular to the plane of the ring (axially) and be described as either axial-trans- or axial-cis- relative to that plane. similarly, the diol incorporated into a ring structure can have its relevant hydroxyl groups pointing generally outward (equatorially) in the plane of the ring. In this case, they can be pointing marginally above or marginally below the plane of the ring by up to a nominal 19 degrees in some sugars, even though most Newman diagrams suggest +/–30 degrees. If the relevant hydroxyls are both pointing in the same direction relative to the plane of the ring, they are described as equatorial-cis-. If one is pointing above the ring and one below the ring, they are described as equatorial-trans- .

PtdGal is defined as the receptor of the sensory neuron initiating the sweetness channel of gustation. The ligand is shown in its standard notation that places the atoms of interest at the lower left. A bracket is shown to represent the stimulant bonding with the specified oxygen atoms.

The PtdGal receptor is shown twice in a dashed box because of an additional phenomenon that has been documented. Besides the common two-point coordinate bonding arrangement illustrated with a spacing of 2.6 Angstrom in the top half of the box, it was found that an enhanced sensation of sweetness could be elicited by some stimulants if they employed a three-point rather than two-point bonding strategy. This super sweet stimulant situation is shown in the lower half of the box.

Phosphatidylinositol (PtdIns) exhibits a spacing of 3.3 Angstrom between the two rightmost hydroxyl groups in the figure. This spacing is uniquely matched to the spacing between the oxygen atoms of adjacent water molecules hydrating with the sodium ion. Thus, this phospholipid is specifically attractive to coordinate bonding to the hydrated sodium ion and eliciting the sensation of saltiness. Note that this natrophore relates only to the sodium ion. The Signal Generation & Processing 8- 41

Figure 8.5.1-15 Summary: sensory receptors of gustatory modality based on the Electrolytic Theory of the Neuron and a coordinate chemistry mechanism. The glycol receptor is able to sense a variety of man-made “super sugars” using a tripartite coordinate union when they exhibit an electron rich area a prescribed distance from the two oxygen atoms of their AH,B structure. The bitter (quinine) receptor may also operate in a tripartite mode to sense a variety of “super bitter” stimuli. See text. Built using the style of Dowhan, 2002. 42 Neurons & the Nervous System anion of the salt plays no role in eliciting the salty sensation. PtdIns is defined as the receptor of the sensory neuron initiating the salty channel of gustation. Inositol monophosphate_19951069 demonstrates the esterification of inositol to the intrinsic Ptd. Phosphatidyl 3'-O-aminoacyl glycerol (Ptd3'Og) is the fourth unique phospholipid of the gustatory modality. It exhibits a spacing of 4.2 Angstrom between the doubly bonded oxygen and the amine group of this amino acid based ligand. This spacing is compatible with a very large group of chemical that can coordinate bond with it and elicit a sensation of bitterness. Ptd3'Og is defined as the receptor of the sensory neuron initiating the bitter channel of gustation. Based on the similarity in the stereochemistry of the bitter receptor and the sweetness receptor, the bitter receptor has been reproduced twice within a dashed box. It appears likely that some of the extremely bitter stimulants that have been found may employ a three-point coordinate bonding arrangement like the super sweeteners do. This potentiality is shown in the lower half of the box using the dashed triangle to suggest the bonding arrangement.. The coordinate bonding between the various stimulants and receptors on the external lemma of the sensory neurons involves a very low energy (5kcal/mole) and does not constitute a chemical reaction in the conventional sense. No reaction products are formed and the original species reappear when the hydrogen bonds are disrupted. The stimulants do not pass through the lemma of the sensory neuron.

The major difference between the sensory receptor for sweetness and saltiness involves the steric crowding of the hydroxyl ligands of the polar molecular structure. The galactose molecule exhibits considerably more steric crowding than does the highly symmetric inositol molecule. This results in a smaller dimension, d, between positions 3 and 4 of the galactose molecule than in inositol.

The phosphatidyls of the gustatory sensory receptors are specialized forms of type 4 lemma. They can be tabulated under simpler names indicative of their specific character, Figure 8.5.1- 16. The new laboratory channel designations in the right hand column are developed in the following paragraphs.

Figure 8.5.1-16 Lemma sub-type designations for the sensory receptors of gustation ADD. The active groups are attached to a generic phosphatidyl molecule (Ptd) that provides the electrical conductivity through the outer bilayer of the neuron dendrite. The label C–Best is used to designate the organic acid sensitive channel, that excludes sensitivity to inorganic acids.

Moulton speculated on the location of the sensory receptors on the surface of the sensory Signal Generation & Processing 8- 43 olfactory neurons in 1970 but there has been little discussion of the subject since56. The discussion was exploratory and drew no conclusions. He did note the vast surface area of the cilia of the sensory neurons compared to the estimated 3 sq. cm. of the olfactory epithelium in humans. Figure 8.5.1-17 summarizes the performance of the gustatory modality and presents a rectilinear graph for positioning stimulants in sensation space for the first time. As noted in Section 8.5.xxx, the signals passed to the brain by the sensory channels are treated as independent, and therefore orthogonal, by the brain. This orthogonal feature supports the presentation of the sensory data in a three-dimensional format, as suggested by the “basis factors” of multi- dimensional scaling applied to the gustatory modality. A d-value of 4.746 Angstrom is shown for the bitter sensory receptor. Some papers have suggested a value of 4.36 Angstrom derived from features of several astringents (primarily alkali- earth salts).

Figure 8.5.1-17 UPDATE XXX’s Summary: performance of the gustatory modality based on the Electrolytic Theory of the Neuron and a coordination chemistry mechanism. Top; stereo- geometry of the transduction process. Middle; a tabulation of pertinent labels and parameters. Bottom; a rectilinear sensation vs stimulus space describing the first order operation of the gustatory modality. Mean values are calculated or reported. Distributions are conceptual. [xxx Dashed line shows potential alternate bitter channel at d = 4.36 Angstrom]. See text.

56Moulton, D. (1970) Detection and recognition of odor molecules In Ohloff, G. & Thomas, A. eds. Gustation and Olfaction. NY: Academic Press pp 1-27 44 Neurons & the Nervous System

The discovery of a large number of man-made sweeteners, and obscure natural sweeteners, with immensely more perceived sweetness than the natural reference, sucrose, has changed the character of the search to understand the S-best sensory channel of gustation. van der Wel has provided a comprehensive list of these sweeteners57. He also noted the propensity of these sweeteners to exhibit multiple AH,B sites and potentially multiple AH,B,X sites (possibly with multiple X dimensions).. Figure 8.5.1-18 shows the first-order form of the gustaphores of taste. The titles are arbitrary but designed to emphasize the character of the gustaphore. Acidophore suggests the Lewis acid character of the acidophore, as opposed to a simple hydrogen ion. The first order acidophore is invariably derived from a carboxyl group and could be labeled a carboxylophore or an acetateophore (See Section 8.5.1.6.2). The glycophore suggests a sweetness, but could just as readily be labeled an ethylophore to indicate the two carbon backbone. The term glycophore is infrequently used and does not describe the general case. In the extended world of sweetness, at least one oxygen orbital can be replaced by another electron-pair sharing element. In gustation, the term orbital will be used to designate oxygen, nitrogen or sulfur, and potentially phosphorus, atoms with their outer electrons in hybridized form and exhibiting shareable electron-pairs. [xxx add C=C double bond and resonant benzyl ring? ]

In the more general case, the oxygen can be replaced by nitrogen, sulfur or an electronegative feature such as an unsaturated carbon bond or the π-bonding system of the benzene ring. In the more limited case, the term glycol is also appropriate as it describes two hydroxyl groups separated by two carbons. The controlling feature is the dimension, d, between the two orbitals. Trans– glycol configurations, do not exhibit the required d-value of about 2.6 Angstrom. The natrophore describes the configuration of the hydrated sodium ion giving what is commonly called the salty sensation. The anion of the salt plays no role in this sensation. The natrophore exhibits a 90 degree angle between the two orbitals due to the octahedral form of the hydrated sodium ion. The gustaphore contributing to the bitter taste has been labeled a picrophore to suggest its bitter taste. It could be described as the propophore to suggest the three carbon backbone of the structure.

57Van der Wel, H. (The role of organic chemistry in taste perception In van der Starre, H. ed. Proceedings of 7th ISOT London: IRL Press pp 13-19 Signal Generation & Processing 8- 45

An electronegative feature also appears to be important in a special situation leading to super sweetness, and potentially super bitterness. An electronegative feature a specific distance from the AH,B group appears to bond to a site on the receptor. This tripartite AH,B,X arrangement results in a sweetness sensation orders of magnitude (reported as great as 9500:1) greater than that of the dipartite arrangement (Sections 8.5.3 & 8.5.10). After the presence of two electron-pair sharing orbitals and the hydrogen bond associated with one of them, it is the distance, d, between the orbitals supporting the AH,B, or AH,B,X coordinate union that is the important feature of the gustaphores. The distances shown in the figure are nominal as little statistically precise data is available. There has been confusion previously in the case of the glycophore as to whether the distance measured was between the orbitals or the greater diagonal distance between one orbital and a hydrogen. 8.5.1.6.1 Defining the gustatory perception space

Three facts have emerged in the above discussion;

• the d-values of the gustaphores in a stimulant are key to interpreting gustation • the d-value continuum represents the fundamental dimension of gustation • the four gustatory paths within the neural system are treated as orthogonal Figure 8.5.1-18 The first-order gustaphores of taste. • the four gustatory neural paths represent The spacing, d, between the oxygen atoms (or four “labeled-lines” and alternate orbitals) is the critical dimension in • Each of the labeled lines may consist of determining the effectiveness of the gustaphore. multiple individual neurons in that path. See text. Based on these individual corollaries to the hypothesis, the following assertion is important; • the labeled line represent orthogonal paths that can be considered nodes along the fundamental dimension. The question becomes how is the information delivered to the higher information extraction engines of stage 4. Furthermore, how can the information be represented to match the way the stage 5 cognitive engines perceive the information.

Figure 8.5.1-19 provides the first calibrated graph of the gustatory perception space. The vertical lines represent the nominal d-values of the sensory neuron receptors. The distributions about these vertical lines represent the probability that a given gustaphore can interact with the gustatory receptor (GR) associated with that d-value. Currently the widths of these distributions are not known. A subsequent paper will show this horizontal number line can be folded at each of the nominal d-values to form a three dimensional taste perception space. 46 Neurons & the Nervous System

Figure 8.5.1-19 The one-dimensional effectivity graph of gustatory performance based only on the steric properties of the receptors and gustaphores. The term ring in the graph refers to a cyclic compound. Centroid values of each distribution are at nominals of 2.276, 2.82, 3.243 & 4.746 Angstrom.

This graph can be considered analogous to a spectral plot of the visual modality based on a constant amplitude narrow band spectral source traversing the wavelength spectrum. In that analogy, the effectivity shown here is similar to the probability of excitation of an individual chromophore of vision as a function of wavelength. By folding the visual modality representation at the centroids of the chromophore peaks, the familiar 3D and 2D perceived color spaces of vision are obtained.

Each response function is illustrated here with an arbitrary tuning width of ±5% . This tuning width is associated primarily with the individual GR and will be more completely evaluated later. This use of the term tuning is different from that commonly used within the psychology community. See section 8.5.2.3.5. It is proposed that the same folding procedure used to create the visual color solid of Munsell, etc. can be applied to gustation. The result is a 3D volume with the centroids of the various gustatory receptors (GR) at the nodes of the volume and the distances between the nodes represented by the d-value scale. The term “folding” used here is totally different from the “unfolding” term used to differentiate the data points of one subject from a larger set of subjects in a MDS representation. The concept is discussed in Chapter 7 of Cox & Cox. It does not appear to be widely used.. 8.5.1.6.2 A 3D olfactory perception space with calibrated scales Signal Generation & Processing 8- 47 This paragraph will develop the MDS space with calibrated scales before introducing both a quantitiative MDS space for gustation and a quantitative MDS-like space incorporating intensity information. When folding a continuous 1D fundamental dimension into a more useful 3D space several options arise; • Whether to employ a linear or logarithmic scale. • Whether to adopt a left-hand or right-hand 3D space configuration. For the purposes of this discussion, the lengths of the individual vectors associated with either the linear or logarithmic scales will be ignored in favor of a set of equal length vectors (resulting in a cubic perception space for illustration). It is also important to recognize that the higher information extraction engines of the brain can only add and subtract the amplitudes of the vectorial signals they receive over the mathematically independent channels of the gustatory modality, always staying within the boundary formed by the nodes at the corners of the perceptual space. [xxx how to explain this in more precise, but not necessarily more difficult language ] •Each gustaphore channel projects an analog intensity value to the information extraction engines. While this intensity value may be propagated as a pulse signal within stage 3 circuits, it is recovered at the output of stage 3 and processed within stage 4 as an analog signal.

While gustaphores with d-values marginally less than 2.26 and marginally greater than 4.74 Angstrom can form a DACB with a GR, the GR channels only report the signal intensities associated with the labeled lines arriving at the information extraction engines of the brain. These labeled lines represent the cardinal d-values associated with the GR’s (2.26, 2.82, 3.24 & 4.74 Angstrom). The extent of the stimulus application space extends only marginally beyond the values of the extreme nodes. The axes of the 3D representation interconnecting these nodes and representing the gustatory perceptual space do not extend beyond those extreme nodes.

Like the visual modality, the spacing between the nodes of the 1D gustatory dimension appear to be independent of any underlying logarithmic relationship. Therefore, a linear dimension shall be employed here.

Like the visual modality, the peaks in the sensitivity profile of gustation can profitably be considered nodes in a 3D representation of the perceived gustatory space.

Figure 8.5.1-20 shows two arbitrary foldings of the linear parameter, the d-value, into three dimensional form using the peak sensitivities of the GR’s in the modality as nodes of the perceived space. Frame A repeats figure 1. Frame B shows the intensity values reported to the stage 4 information extraction engines. While frame A indicates an effectivity profile for each gustatory channel, frame B does not. The signals delivered to the stage 4 information extraction engines arrive over “labeled lines.” There is no equivocation about their source. Frame B also indicated the vectorial length of each segment of the fundamental dimension between the nodes of gustation. To maintain the integrity of the 3D representation, it is critically important that the head-to- tail character between the individual vectorial segments of the fundamental dimension be maintained. If not required in setting the parameters for the computer solution to the MDS, the program will select its own relationship by default. This may not maintain the desired continuity of the fundamental dimension in the resulting representation.

Frame C shows the nodes of the perceived space labeled per their d-values using a Left-hand- rule for folding. Frame D shows the same framework as frame C but using a Right-hand-rule for folding. In both cases, the heavy line demonstrates the folding of the fundamental dimension. Dimension 1 represents the value +Δx, dimension 2 represents the value +Δy and dimension 3 represents the value +Δz. 48 Neurons & the Nervous System

The signs of the scales assigned automatically by the MDS program are irrelevant. It is important that the various nodes of the gustatory modality paths be located as shown in framed D. It is critically important that the location of these nodes be shown in a “standard representation” so that the representations from various investigators can be compared. Such a comparison has been labeled a “Procrustes analysis” in some documents (Cox & Cox, chapter 5). It requires employing rotation, mirroring and scaling until congruent representations are obtained. Then, a calculation of similarity can be performed if desired. Otherwise, a visual comparison can be made. Dunn-Rankin et al. have developed a complete procedure for performing the above standardization procedures under the heading “Principal Component Analysis (PCA) in their application workbook for Windows based computers58. They note, “Usually the general factor solution that emerges (from MDS) before rotation is not as interpretable as a solution that can be obtained by rotating the axes in order to contrast the factor loadings more effectively.” They provide the matrix for rotating the axes in the clockwise direction. They also use a routine called Varimax to determine the optimum angle of rotation. Abdi has expanded on the Varimax procedure59.

As is immediately apparent, these two frames are not superimposable (using the chemists terminology), but the A–G–N and G–N–P planes are identical in both views except for a direction sign change. Thus, the two volumes are stackable, frame D on to frame C. There is an intensity value from frame B associated with each of the nodal positions in both cases. This intensity value constitutes a fourth dimension within the complete data set.

If only one gustaphore is present during an experiment, the gustaphore will be represented by an analog value associated with its node in either frame C or D. Figure 8.5.1-20 Alternate representations of a 1D parameter in a 3D perception space. A; the 1D This is an important situation. The MDS dimension, d-values. B; the intensity values technique is qualitative. The typical delivered to the stage 4 engines. C; the 3D MDS analyses should only report perception space using the Left-hand-rule. D; the individual gustaphores at nodal 3D space using the Right-hand-rule. E; the locations. If all of the gustaphores perceived location of mixed gustaphores. See text.

58Dunn-Rankin, P. Knezek, G. Wallace, S. & Zhang, S. (2004) Scaling Methods, 2nd Ed. Mahwah, NJ: Erlbaum Associates Chapter 12

59Abdi, H. (2003) Factor Rotations in Factor Analyses. In Lewis-Beck M., Bryman, A., Futing T. Eds. Encyclopedia of Social Sciences Research Methods. Thousand Oaks (CA): Sage. http://www.utdallas.edu/~herve/Abdi-rotations-pretty.pdf Signal Generation & Processing 8- 49 exciting a single GR are not shown congruent at the nodal position, the precision of the data is not statistically significant. The diameter of the cluster of values obtained by repeating the experiment is inversely proportional to the precision of the results. If two gustaphores are present during an experiment, the perceived stimulant will appear as a point value between the nodes of the gustaphores at a position determined by the relative intensities (available from the data for the fourth dimension) of the gustaphores sensed by the GR’s. • Location 1 represents an intensity in the G–channel twice that in the C– channel. • Location 2 represents an intensity in the G–channel twice that in the N–channel. • Location 3 represents an intensity in the P–channel equal to that in the N–channel. • Location 4 represents an intensity in the N–channel twice that in the C–channel. • Location 5 represents an intensity in the P–channel twice that in the C–channel. Locations 4 & 5 are obviously not along the fundamental dimension of the d-value parameter. They are similar to the magentas of color vision. They represent the simultaneous presence of two gustaphores that are stimulating non-adjacent GR’s along the d-value dimension. Monosodium glutamate is a stimulant exhibiting three gustaphores. If they each excited an equal intensity signal in the N–, G– & A– channels. The perceived sensation would be in the center of the A–G–N triangle. For different perceived intensities for the individual gustaphores, the perception would move about within the plane of the A–G–N triangle.

As a corollary to the single gustaphore situation above, in MDS analyses involving multiple gustaphores, the reported position in perceptual space is only accurate if the same experimental protocol reports the individual gustaphores at their correct nodal positions. When a series of stimulants containing the same glycophore are shown as a cluster in an MDS representation, the diameter of the cluster is inversely proportional to the precision (adequacy) of the experimental protocol. This matter will be addressed further below.

The choice of the left-hand rule or right-hand rule coordinate systems shown above is arbitrary. Figure 8.5.1-21 shows a three-dimensional taste space based on the above and additional considerations based on Section 8.5. It is rotated from the right-hand rule coordinate system above. The figure shows the four basic taste sensations at four of the eight vertices of a cubic space. The folded segments of the fundamental dimension are shown as dimension 1 (A–G), 2 (G–N) and 3 (N–P). 50 Neurons & the Nervous System

Figure 8.5.1-21 Potential taste sensation space for a mammalian species exhibiting four primary taste sensation. The lower frame shows the nominal capture efficiency for a gustaphore of different d-value than the related GR. See text.

[xxx check if this is useful ] Signal Generation & Processing 8- 51 In the orthogonal coordinate system shown, the G–Path node occurs at the null value, 0,0,0. However, this system leaves the C–P and G–P axes as non-orthogonal with respect to the other axes. An alternate representation would shift the 0,0,0 value to the center of the volume and consider the middle of dimensions 1, 2 & 3 as representing the zero value for that dimension.. - - - - [xxx duplicated using different names after next figure ] The potential for a fifth basic taste, umami, to be present at one of the remaining vertices remains present. However, the subsequent analyses will show the major ligands of mono sodium glutamate are; • a carboxylic acid channel (A-path) gustaphore (limited to Lewis acids and their esters) • a 1,2 equatorial-trans-glycol channel (G-path) gustaphore. and • a hydrated sodium ion channel (N-path) gustaphore interacting with a 1,2 axial-trans inositol GR. Thus, mono-sodium glutamate, and in fact all putative stimulants perceived as umami-like are found to consist of multiple gustaphores with their combined perception located in the C–S–N plane.

It should be noted that the amplitude of the stimulus passed to the brain is not represented in this figure. Such an amplitude parameter exists in a separate dimension from those shown as will be developed in Section 8.5.2.3 and later. 8.5.1.6.3 The proposed qualitative 3D gustatory perception space

Figure 8.5.1-22 shows an alternate MDS space using a more relevant labeling and a right hand rule widely used in engineering. It is recommended as a template for future research. Note the labeled nodes do not incorporate any perceived intensity information. To use this presentation format, there must be an attempt to use gustants of equal gustatory efficacy. The preferred protocol is to use a series of single gustaphore gustants (SGG) analogous to the SOO of olfaction (Section 8.6.xxx). If some of the stimulants are multiple gustaphore gustants (MGG), their location in the perceived taste MDS representation will be at locations remote from the nodes dependent on the character and relative efficacy of the individual gustaphores. The precise locations will be determined by vector-based calculations. The representation of such MGGs can be expected to appear in one of the planes annotated on the figure (or within the overall volume if gustaphores exciting all four channels are present) if sufficient statistical precision is available from the test protocols.

The template has the option of; C stretching the cube shown to a rectangular solid and making the scales uniform in accordance with the d-values at the nodes or, C demarcating each dimension with scales of uniform, but different, spacing from the adjacent dimensions.

If the data offers sufficient precision and indicates the nodes are misplaced, a generic rotation of axes associated with the data is indicated as discussed in Section 8.5.2.3.1. 52 Neurons & the Nervous System

Figure 8.5.1-22 Preferred MDS space based on a right-hand rule more commonly used in gustatory research. The heavy line in the upper frame is a folded version of the abscissa of the lower frame. Available MDS software programs can reverse individual scales to arrive at this right-hand rule space configuration. See text.

The potential for a fifth basic taste, umami, to be present at one of the remaining vertices remains present. However, the subsequent analyses will show the major ligands of mono sodium Signal Generation & Processing 8- 53 glutamate are; • a carboxylic acid channel(A-path) gustaphore (limited to Lewis acids and their acetates) • a 1,2 equatorial-trans-glycol channel (G-path) gustaphore. and • a hydrated sodium ion channel (N-path) gustaphore interacting with a 1,2 axial-trans inositol GR. Thus, mono-sodium glutamate, and in fact all putative stimulants perceived as umami-like are found to consist of multiple gustaphore gustants with their combined perception located in the A–G–N plane. The perception of umami has occasionally been likened to “meat-like,” an obvious mixture of individual perceptions. Other MGGs absent any organic acids or acetates may be represented in the G-N-P plane. For MGGs including derivatives of all four types of SGGs, their representation will appear within the volume of the MDS space (with the continued assumption that all gustaphores present are of equal efficacy). The net efficacy relative to a single node may be affected by the presence of multiple gustaphores with the same nominal d-value. It is stressed that the amplitude of the signals passed to the brain is not represented in this figure and no intensity of the resultant perceived taste is present in the MDS space. Such an amplitude parameter exists in a separate dimension from those shown as will be developed in Section 8.5.1.6.4 and later in Section 8.5.2.3 .

- - - - -

When employing the MDS analytical technique, it is not usual to describe the axes (or dimensions) specifically. Figure 8.5.1-23 presents a table that will be useful later in comparing the data from various investigators employing the MDS analytical technique. As it stands, the table is not useful. Every investigator used different dimension labels and mixed sub-sets. and generally inadequately populated sub-sets.

Figure 8.5.1-23 Citations & parameters of recent MDS investigations ADD DATA to Smith. The dimensional data is based on best estimates by this investigator. CT; chordata tympani. NG; glossopharyngeal nerve. 54 Neurons & the Nervous System

Most of the available data sets ( Hellekant et al., 1997) are based on stimulants and not single gustaphore gustants. As a result, the electrophysiological data is not of nominal amplitude when collected within the neural system. Many of the data sets available have not populated the perceived gustatory space adequately to define their axes explicitly (Smith et al., 1979: Giza et al., 1991; ). Smith & Travers have investigated the efficacy response function for a variety of functions based on a largely conceptual model of the gustatory modality60. They sought to use many of the probability equations of information theory, particularly relative to entropy) in a deterministic context. “The entropy measure, as applied here, varies continuously from 0.0 for a unit that responds exclusively to one stimulus (i.e., narrowly tuned) to 1.0 for a cell that responds equivalently to all four of the basic compounds (i.e., broadly tuned). Subtle variations in the neural response profile of a cell, such as those produced by changes in stimulus concentration, are reflected in this measure. Thus, the use of the equation for entropy to describe the responsiveness of gustatory neurons provides a quantitative measure of their breadth of tuning that can be meaningfully applied to the problems of gustatory quality coding.”

“Investigations of both peripheral and central gustatory neurons in a variety of mammalian species have clearly demonstrated that these cells are typically broadly responsive to stimuli representing the four basic taste qualities (with many archaic citations) . This lack of stimulus specificity first led Pfaffmann (1955, 1959) to propose that taste quality is coded by the pattern of activity across a population of broadly tuned afferents, a concept which has been further elaborated and given a quantitative basis by Erickson (1963, 1967, 1968, 1974; Erickson et al., 1965). This across-fiber pattern theory accounts for the ability of a discrete number of neurons to code the tastes of the many thousands of potential gustatory stimuli and has received support from a number of behavioral studies of taste discrimination.”

“Although this pattern theory of quality coding can account well for the broad sensitivity of gustatory neurons, it has recently been proposed that taste quality may be represented in the nervous system in a more specific fashion (Nowlis & Frank, 1977; Pfaffmann, 1974; Pfaffmann, Frank, Bartoshuk & Snell, 1976). For example, if a neuron responds better to sucrose than to NaCl, HC1 or quinine, it may be carrying information only about "sweetness", the lesser response to these other stimuli being background noise. This labelled-line view of taste quality coding suggests that there are four separate neural channels in the mammalian gustatory system that carry information about the four distinct experiences of salty, sour, sweet and bitter (Nowlis & Frank, 1977). Since responses of these differentially but broadly tuned neurons can be handled well by either theoretical approach, the nature of the neural code for taste quality is a topic of much current interest.”

These remarks are largely consistent with this work but for different underlying reasons, specifically related to the interpretation of multiple channel stimulation as a noise-related phenomenon. In 1979, Smith & Travers did not consider that their individual gustants )such as quinine HCl and sucrose) might contain multiple gustaphores or that HCl might be primarily a nocent and not a gustant. After discussing a variety of approaches to interpreting the results of others, they note, “As a consequence of these various approaches to describing response breadth, conclusions about the specificity of populations of taste-responsive neurons are sometimes contradictory (Pfaffmann et al., 1979). For example, most cortical cells in the rabbit have been reported to respond to three or four of the basic taste stimuli (Yamamoto & Kawamura, 1975), whereas the typical cortical neuron in the dog and rat responds to only one or two stimuli (Funakoshi et al., 1972). Either there are species differences in the

60Smith, D. & Travers, J. (1979 A metric for the breadth of tuning of gustatory neurons Chem Sense Flav vol 4(3), pp 215-229 Signal Generation & Processing 8- 55 specificity of these central gustatory neurons or the various definitions of responsiveness and breadth of tuning are creating an apparent discrepancy in the interpretation of these data.” They describe their approach to the tuning breadth problem, “What is required to resolve this kind of problem is a metric for the breadth of responsiveness of gustatory neurons that will be sensitive to subtle distinctions in neural response functions rather than simply dichotomize units as either broadly or narrowly tuned. In a recent analysis of the response properties of gustatory neurons-in the hamster medulla, Travers & Smith (1979) introduced such a measure of response breadth based on the equation for entropy from information theory (Shannon & Weaver, 1949). This approach takes into account the relative magnitudes of the responses to each of the four basic stimuli and provides a continuous scale for quantifying the breadth of a neuron's sensitivities. The purpose of the present paper is to further elaborate this concept in order to provide a more quantitative approach to the problem of the breadth of tuning of gustatory neurons.” They note, “The application of the entropy equation as a measure of response breadth in the present situation is distinctly different than its use in the context of information theory, in which estimates of p, are based on response probabilities (Shannon & Weaver, 1949).”

If the investigation was repeated using SGGs, the results would be distinctly different and more repeatable.

The 1979 Smith & Travers paper did not report data from an MDS analysis; however, the 1983 paper did61. “The present investigation is designed to examine the relationship between the taste neuron types in the hamster brain stem suggested by the preceding paper and these proposed neural codes for taste quality. Multidimensional scaling of the stimulus relationships within and across the neuron groups is employed in an attempt to understand the role of these various classes of neurons in the coding of taste quality.”

They use the –best labeling of their four coordinates (based on the stimulus used) rather that the –path designations more directly related to the sensory neurons. They do note that the so-called H– neuron group is dominated by the response to organic acids and not inorganic acids exhibiting an H+ ion. They also make the important observation, “The exclusion of any one class of neurons from the population results in a dispersal of its most effective stimuli within a multidimensional ‘stimulus space’.” After a few more introductory remarks, they also note, “Thus, the neural distinction between many behaviorally discriminable stimuli depends on the simultaneous activity in different classes of neurons.” Concluding their Summary & Conclusions, they note, “Particular sets of neurons are critical for coding the tastes of particular classes of stimuli, but activity in more than one group of neurons is necessary for the unambiguous coding of taste quality.” These assertions are all compatible with the hypothesis of this work.

Their introduction began with, “It is almost without exception that mammalian taste neurons respond to stimuli representing more than one of the classical four taste qualities (many citations). It was the multiple sensitivity of fibers in the chorda tympani nerve that first led Pfaffmann (25-27) to propose that taste quality is coded by the relative amount of activity across a number of taste afferents. Since the response of most taste fibers can be modulated by both stimulus quality and intensity, the response of any particular cell is typically ambiguous with respect to either parameter.” They then state their goal, “The purpose of the present paper is to examine the relationship between the gustatory neuron types in the hamster brain stem and the population response of these neurons to a variety of taste stimuli.” They study focused on neuron types and not stimulus types! They use a variety of inorganic and organic molecules to determine the sensitivity of four different classes of neurons within the hamster parabrachial nuclei (PbN) neurons. “These data were chosen because the numbers of cells in each of the neuron classes was nearly equal in the PbN data. This was important since one of the goals of

61Smith, D. Van Buskirk, R. Travers, J. & Bieber S. (1983) Coding of taste stimuli by hamster brain stem neurons J Neurophys vol 50(2), pp 541-558 56 Neurons & the Nervous System the present analysis was to examine the stimulus relationships within each of the neuron groups separately.” Under methods, “Identification and quantification of neural responses were the same as reported in the preceding paper (35). The 5-s response measures were put into the same multivariate matrix (m neurons x yt stimuli), from which the subsequent analyses proceeded. From this matrix of 3 1 neurons x 18 stimuli, an across-neuron correlation matrix was generated, which was used as the input matrix to a multidimensional scaling program (KYST, Bell Telephone Laboratories). This program yielded a “stimulus space,” depicting the similarities and differences in the across-neuron correlational relationships.” They went on, “The number of dimensions necessary to represent these relationships adequately was chosen on the basis of the “stress” in the solution and the interpretability of the space. Once the “stimulus space,” defined by the responses of all the neurons, was created, additional multidimensional scaling analyses were conducted on subsets of the neurons, defined by the cluster solution in the preceding paper which suggested three neuron groups (S-, H-, and N-neurons) on the basis of similarities among their response profiles..” They did not report on Q–Best (P–Path) neurons although they did show two P–Path molecules in their three dimensional MDS. Smith et al. report their results in human perceptual space although the data was collected by interrogating hamster stage 3 projection neurons. Their figure 8 clearly shows separate groupings for the stimuli focused on the A–Path, G–Path, N–Path and P-Path. Their use of four different symbols to relate the stimulants affecting a particular neural channel maximally in their MDS analysis is in excellent agreement with the hypothesis of this work (with spreads primarily due to the limited number of test sequences performed and a lack of accounting for the use of MGGs. Their node at 0,0,0 is clearly related to the A–Path, their node I is related to the G–Path, their node II is related to the N–Path Node III locates the P–Path. The location of the data groups suggests a rotation of the coordinate system by about 10 degrees is appropriate.

Their paper goes on to form a set of two-dimensional MDS representations based on collapsing their 3D data set (with the confabulation of the results cautioned about in Section 8.5.xxx. Recognizing this confabulation, it is clear from their figures that removing one set of related stimulants does provide a multi-peaked response function for each set of stimulants.

They conclude in their Discussion that, “The present analyses have demonstrated the relationship between the “across-fiber pattern” and labeled-line approaches to the analysis of neural responses in the gustatory system. Although these two approaches are describing the same phenomena from different perspectives, their implications for the coding of taste quality are quite distinct. The results of the present analyses can be interpreted within either theoretical framework and, taken alone, cannot argue for one of these alternatives over the other. To the extent that particular classes of neurons are critical for establishing particular across-neuron patterns, the same cells are important for coding quality in both theoretical approaches. However, the implication for the roles of these cells in coding taste quality is quite different in the labeled-line and the across-fiber pattern hypotheses.” They go on, “Whether one chooses to accept a labeled-line or an across-fiber pattern interpretation of neural coding depends on whether one believes that activity in a single neuron represents, in and of itself, a single taste quality. Unless this point is experimentally testable, the distinction between these two theoretical approaches is primarily a philosophical one.” Fortunately, a clearer understanding of the signal processing context available leads to a testable situation. The data and discussion of the Smith et al. paper is quite supportive of the hypothesis of this work when one modification to their concept is introduced. Their discussion addresses the differences between the above two theoretical approaches without incorporating the concept that the signal paths are statistically independent (although their stimulating gustants may contain statistically related gustaphores) and therefore orthogonal. Considering the vectorial addition of the responses in the orthogonal representation of the perceptual space leads to the stage 4 information extraction neurons providing a composite signal to a lookup table that defines the overall gustatory perception stored in the appropriate region f the saliency map and available to the stage 5 cognitive neurons.

The Smith and the Danilova papers have begun providing a measure of the quality of their MDS representations using the Kruskal stress parameter. The Danilova paper provides more visibility into this parameter as a function of the number of dimensions set as a parameter in their MDS calculations. This parameter will be addressed further in Section 8.5.2. Signal Generation & Processing 8- 57 - - - - Danilova et al. provided gustatory data in an MDS format as a result of their investigations with the marmoset62,63. “Callithrix jacchus jacchus is a small New World monkey belonging to the infraorder platyrrhina, which shared ancestry some 38 millions years ago with the catarrhina group to which humans, apes and Old World monkeys belong.” The investigator is cautioned to be aware of the findings in Section 8.5.1.6.8. For reasons not clearly stated, the MDS representation in the 2002 paper did not include data related to the sodium channel (N–Path). It did include HCl as a stimulus along with a group of Lewis acids. It did include a variety of very complex gustant molecules (Table 1, example: QHCl). They stated, “In our recent studies, we have strived to use a large number of compounds. The main reason for this is that we wanted to address the question of quality and coding in taste not the particular taste of a given compound.” This approach overlooks the fact that many of their gustants incorporated many diverse gustaphores. The paper is discussed and the representation is shown in Section 8.5.4.5. As an aside relative to the Danilova & colleagues papers, data from the chorda tympani (CT)should not be combined with data from the glossopharyngeal nerve (NG) other nerves, and other orthodromic locations within the neural system without justification. They began their analyses with a series of cluster analyses and noted,

“The hierarchical cluster analysis distinguished three major clusters in both CT and NG(separately): S, Q, and H. The SCT fibers, 38% of all CT fibers, responded only to sweeteners. The SCT fibers did not respond during stimulation with salts, acids, and bitter compounds but exhibited OFF responses after citric and ascorbic acids, quinine hydrochloride (QHCl), and salts (in 80% of SCT fibers). SNG fibers, 50% of all NG fibers, also responded to sweeteners but not to stimuli of other taste qualities (except for citric acid, which stimulated 70% of the SNGfibers). Some sweeteners, including natural (the sweet proteins brazzein, monellin) and artificial [cyclamate, neohesperidin dihydrochalcone (NHDHC), N-3,5-dichlorophenyl-N’-(S)-α-methylbenzylguanidineacetate (DMGA), N-4-cyanophenylcarbamoyl-(R,S)-3-amino-3-(3,4-methylenedioxyphenyl) propionic acid (CAMPA)] did not elicit responses in the S fibers.”

These findings appear to conflict with the virtually universal reports that individual sensory path receptors exhibit broad tuning that assures they show some response to stimuli aimed at other, or at points along the d-value line between both, sensory paths.

The above paragraph was followed by an apparently contradictory paragraph,

“In general, the response profiles of the SCT and SNG clusters were very similar, the correlation coefficient between the responses to sweeteners in these clusters was 0.94. Both the QCT and the QNG fibers (40 and 46% of all fibers) were predominantly responsive to bitter compounds, although their responses to the same set of bitter compounds were quite different. Sweeteners with sweet/bitter taste for humans also stimulated the Q clusters. The H clusters (22 and 3% of all fibers) were predominantly responsive to acids and did not respond to stimuli of other taste qualities. However, bitter stimuli, mainly QHCl, inhibited activity in 70% of HCT fibers. Among a total of 90 fibers from both nerves there was only 1 NaCl-best fiber in CT. We found, however, that 35% of the CT fibers reacted to salts with inhibition of activity during stimulation, followed by an OFF response. This OFF response was diminished or eliminated by amiloride.”

62Danilova, V. Danilov, Y. Roberts, T. Tinti, J-M. & Hellekant, G. (2002) Sense of taste in a new world monkey, the common marmoset: Recordings from the chorda tympani and glossopharyngeal nerves J Neurophysiol vol 88, pp 579–594

63Danilova, V. & Hellekant, G. (2003) Comparison of the responses of the chorda tympani and glossopharyngeal nerves to taste stimuli in C57BL/6J mice BMC Neurosci vol 4, http://www.biomedcentral.com/1471-2202/4/5 58 Neurons & the Nervous System

The data in the Danilova & Hellikant(2002) paper have reported negative pulse rate data from neurons of the chorda tympani, when Chapter 9 will show the pulse rates are necessarily zero or positive. The negative values may be entirely due to their mathematical manipulations (example, subtracting a quiescent pulse rate from a gross pulse rate during stimulation). Their comments regarding the suppression of responses suggests their electronic probes were frequently interrogating stage 3 signal projection neurons of the differential type rather than of the summation type (see Chapter 9). - - - - - All of the phospholipids defined above are known to be associated with the sensory neurons of gustation. They all exhibit a dipole moment that is a function of their state of coordination. When present with one conjugated lipid in the liquid crystalline form, as found in a bilayer, they exhibit a dipole potential that can be measured. This potential is also a function of the state of coordination of the molecule. - - - - - Similarly, the system does not sense the presence of a salt, or even the presence of sodium per + se. It senses the presence of the specifically hydrated form, Na:(H20)6 of the sodium ion only. It specifically does not sense the presence of the chlorine ion in solution. The operation of the modality explains a critical aspect related to mono-sodium glutamate. This compound interacts with the hydrated sodium receptor and potentially both the glycophore and organic acid receptors. It does not interact with an independent umami sensory receptor.

Figure 8.5.1-24 provides a necessary conversion process if one is to transition from a behavioral to a more fundamental understanding of the gustatory modality. No chemical reactions are involved in gustation; no residues are formed. Only temporary coordinate chemical bonds are involved. The “acid” sensory channel does not sense a proton; it operates only in the Lewis acid sense in order to sense a carboxyl ligand.

Figure 8.5.1-24 The transition from behavioral to fundamental perspectives in gustation. All the gustaphores exhibit an AH,B molecular configuration and suitable stereographic configuration capable of forming a dual hydrogen bond with a receptor. Note: “picric acid” is not a Lewis acid. It is the historical name for 2,4,6-trinitrophenol_6688.

To be clear, the hydrogen ion, H+, is not sensed by the gustatory modality. The gustatory modality senses an AH,B stereochemical configuration capable of forming a DACB with a d = 2.276 Angstrom dimension. This is the nominal “natural” configuration of the carboxyl group, i.e., the organic (Lewis) acids and the esters of such acids, the esters. Similarly, the “sweet” sensory channel does not sense the chemical groups typically associated with a sugar; it senses the 1,2 equatorial-trans-glycol overlay group associated with a wide variety of organic compounds. In research, the preferred SGG is the monosaccharide, glucose, rather than the disaccharide, sucrose. The “salty” sensory receptor does not sense a salt; it senses only the fully hydrated sodium ion. The “bitter” sensory receptor does not sense any Signal Generation & Processing 8- 59 feature or chemical group uniquely related to the quinine molecule; it senses the picrophore, a pair of oxygen atoms separated by a three-carbon chain with a specific structural relationship. The simplest example of a picrophore is (2R, 4R)-pentanediol_2005883. There is no “umami” sensory receptor, the typical umami stimulant, stimulates more than one of the above sensory receptor channels. The gustatory section concludes with: The realization that the gustatory receptors depend on specific steric distances between orbitals capable of sharing electron-pairs and not upon the functional groups within a molecule or the interconnection of those groups. The hydrated state of any cation is critically important to the gustatory transduction process. When monosodium glutamate is ionized and hydrated in solution, it is capable of stimulating three of the four sensory receptors, the acid, the sweet and the sodium receptors. There is no evidence for a separate and unique umami sensory channel. It has been reported by Beets that sodium benzoate can elicit sensations of sweet, bitter, sour, salty and/or tasteless64. However, the elicited sensations appear to vary significantly among individuals. The following discussion does not surface how sodium benzoate could elicit a bitter or tasteless response. See Section 8.5.4.5.1.

8.5.1.6.4 An equivalent quantitative 3D olfactory perception space EMPTY

The qualitative MDS representation proposed above can be used for more quantitative investigations. These investigations typically employing electrophysiological interrogation of one or more stage 3 signal projection neurons associated with the gustatory (and not nocent) signaling channels. There are several obstacles to the use of a 3D MDS format;

1. The major obstacle is that the 3D MDS format cannot represent four independent (orthogonal) sensory channels and, in addition, an amplitude parameter.

2. A second major obstacle is accommodating within the protocol adequate control of the state of adaptation of all of the sensory receptors stimulated by the gustants.

3. The pulse code used among the stage 3 channels must be understood and evaluated properly to avoid to preserve the timing of the first pulse and recognizing the variation in pulse- to-pulse interval intrinsic to this coding.

The first obstacle indicates the test protocol must be restricted to stimulating no more than three of the independent sensory channel GRs (bordering the A-G-N or G-N-P planes) in order to accommodate intensity information along the third axis.

Without detailed knowledge of the efficacy profile of each of the GRs, it is difficult to insure or compensate for the state of adaptation of the GRs of an individual sensory channel. The pulse code used in gustation appears to be the same code used throughout the peripheral neural system (See Chapter 9). Because of the relatively crude method of stimulation used in both gustatory and olfaction research, it is typically best to make measurements related to the stage 3 pulse trains after they have stabilized, and typically after several seconds from onset. 8.5.1.6.5 The gustatory response versus molarity of stimuli

The output as a function of input intensity has proven very important in understanding the operation of the visual modality. Zotterman has provided data on the gustatory response of humans based on recordings of taste

64Beets, M. (1978) Op. Cit. pg 175 60 Neurons & the Nervous System nerves65 with most of the data attributed to Borg et al. (1967). They confirmed their integrated responses from the chorda tympani relative to the molarity of citric acid exhibited the same slope on log-log paper as psychophysical responses, 0.5. They then showed the slope of N-best channels exhibited a slope of 1.0 while the slope of H-best channels showed a slope of nominally 0.67 psychophysically for 14 students. In parallel experiments, the G-Path channels showed a slope near 1.0 and higher than that of the H-best channel. This data is very similar to that provided by Sato for the rat66. Sato more carefully recorded the saturation affects encountered at molarities near and above 0.1 M. Several of the figures in Sato and in Borg et al. show significant saturation that cannot be interpreted using a straight line as an overlay. No clear threshold levels were documented. The size of the database suggests more experiments are needed and saturation must be accounted for in reducing the data. xxx in his Fermich Award Address has provided a mixed list of stimulant to the gustatory (and probably the nociceptor modalities (page 41) with references.

8.5.1.6.6 The structural constraints on tasting sweet

Shallenberger & Acree presented a brief theory of sweetness in 1967 based on the presence of a glycol group in the stimulus67. It stated the initial interaction “is neither a proton transfer nor an electrostatic interaction, but probably involves London dispersion, the principle element of hydrogen bonds.” However, it is important to note the in vitro experiment involved neutral chemicals dispersed in water and not attached to any neural system. Much more comprehensive papers appeared in 197068 and 197169. The 1970 paper by Shallenberger reviewed the various previous theories of compounds that were perceived as sweet. None of these theories has survived their time. He also addressed the question of why some molecules tasted sweet while their conformal partners did not, and proposed it was due to stereographic hindrances at the GR site. He also demonstrated that the previous idea that L–sugars were tasteless while the D–sugars exhibited taste was a myth.

The 1970 paper addressed the question of why lead acetate and beryllium chloride are reported to taste sweet (page 133). He noted that these materials do not disassociate in the usual manner but form hydrates (much like the saturated alcohols are perceived as sweet due to their forming hydrates (a.k.a. azeotropes).

The 1970 paper also addressed two medical patients being treated for hypoparathyroidism. Simultaneously, they reported an anomaly in their gustatory modality wherein they could not perceive the sensation of sweetness (aglycogeusia). Based on this work, the condition appears to be genetic. The paper uses the terms sour and bitter rather loosely and a more stringent protocol may be needed to evaluate this condition. It is possible these patients were not forming the GR 2 receptor (proposed here to be PTDGal).

Later, a paper by Kier, Shallenberger & Lindley discussed a number of secondary structural

65Zotterman, Y. (1971) The recording of the electrical response from human taste nerves In Beidler, L. ed. Handbook of Sensory Physiology, Vol 4/1: Olfactory Sensing. Chap 6

66Sato, M. (1971) Neural coding in taste as seen from recordings from peripheral receptors and nerves In Beidler, L. ed. Handbook of Sensory Physiology, Vol 4/1: Olfactory Sensing. Chap 7

67Shallenberger, R. & Acree, T. (1967) Molecular theory of sweet taste Nature vol 216, pp 480-482

68Shallenberger, R. (1970) Molecular structure and taste In Ohloff, G. & Thomas, A. eds. Gustation and Olfaction. NY: Academic Press pp 126-133

69Shallenberger, R. & Acree, T. (1971) Chemical structure of compounds and their sweet and bitter taste In Beidler, L. ed. Taste: Handbook of Sensory Physiology, Vol IV, Part 2, Chap 12 Signal Generation & Processing 8- 61 features that affect the intensity of the sensation of sweetness70. The paper considered many alternatives and is very well structured. It also provided some information on the other basic tastes. Shallenberger presented a very comprehensive discussion of the structural aspects of sweetness in 1982 that appears to support the electrostatic premise71. In 1996, he provided a synopsis of his work based on an electrostatic assumption, i.e., that no reaction between the stimulus and the receptor occurred72. In 2000, Eggers, Acree & Shallenberger provided a review of their work over a 30 year span73. The critical nature of the specific conformation of a odorant was highlighted by Shallenberger. The Anti-anisaldehyde oxime is very sweet, yet its fraternal twin, Syn-anisaldehyde oxime, is tasteless. 8.5.1.6.7 Extending the chemoreception concepts of Shallenberger, Kier and Beets

[xxx incorporate Shallenberger in Ohloff & Thomas, page 126 ] Evans provided some early material related to understanding the transduction of the sweet sensation74 that appears to have led to the important documentation of Shallenberger and Acree in 1967. Evans used the terminology of the day to suggest transduction “involved a physical rather than a chemical reaction, but of course a very specific one.” He also discussed the properties of an unspecified stereoisomer of inositol as a stimulant. After Shallenberger et al. and Kier explored the coordinate bond pair concept in the context of the elicitation of a sweet sensation, Beets (1978, pg 188) suggested this concept could be extended to include all of the organic taste stimulants. This work will go farther and show that the concept also explains the major inorganic stimulant class, those eliciting a “salty” sensation. It will show the salty sensation is actually the sensation elicited by the hydrated form of the sodium ion acting as a stimulant. The salty sensation is actually elicited by both bases and salts of the sodium ion in hydration. However the activity of most bases containing sodium are too active chemically to be used in gustation.

In section 8.6, it will be shown that the same concepts apply to the modality of olfaction, with similar constraints related to the chemical activity of the phenols. While the phenols can cause injury if employed in olfaction experiments, their derivatives play an important role in the field of perfumery.

In 1990, a major symposium was held reviewing the work of many investigators during the 1980's. The record of that meeting appeared in 199175. The use of computerized molecular modeling was in its infancy at that time. However, the importance of 3D modeling was noted (page xxx) as well as the fact that the response of the sensory neurons to sugars followed the E/D mechanism described in detail in this work (page 295). Figure 8.5.1-25 shows Hellekant et al.’s interpretation of the E/D mechanism (Section 8.7). It clearly defines a finite delay time (less than 2 sec), a rise time, which should not include the delay time, steep enough to obscure its exponential character (time constant less than 0.25 seconds). See Section 8.5.6.2 for a discussion of this figure.

70Shallenberger, R. & Lindley, M. (1977) A lipophilic-hydrophobic attribute and component in the stereochemistry of sweetness Food Chem vol 77(2), pp 145-153

71Shallenberger, R. (1982) Advanced Sugar Chemistry. Westport, CT: AVI Publishing Chapter 10

72Shallenberger, R. (1996) The AH,B glycophore and general taste chemistry Food Chem vol 56(3), pp 209-214

73Eggers, S. Acree, T. & Shallenberger, R. (2000) Sweetness chemoreception theory and sweetness transduction Food Chem vol 68(1), pp 45-49

74Evans, D. (1963) Chemical structure and stimulation by carbohydrates In Zotterman, Y. ed. Olfaction and Taste. NY: Pergamon Press page 165+

75ACS Symposium (1991) Sweeteners: Discovery, Molecular Design , and Chemoreception. Washington, DC: ACS volume 450 62 Neurons & the Nervous System

An extension of the Shallenberger & Acree and Kier concepts of multiple bonding was also extended, at least conceptually by Tinti & Nofre (page 206). They noted the expression AH,B,X used in this work had been standardized by that time as AH,B,G. They proposed an expanded set of points of importance; AH, B, G, D, Y, XH, E1 & E2 without describing each in detail. The additional points were based on their Table I that summarized the presence of chemical groups that played some role in various sweeteners (and the sweetness of the resulting compound). Figure 8.5.1-26 shows their more complex situation. They did not develop what that role was. They noted, “To elicit sweetness, the simultaneous binding of the eight sites assumed to be involved in the sweetener- receptor interaction is not a prerequisite.” They did provide distances between some selected orbitals present in the various chemical groups. In other cases, they gave Figure 8.5.1-25 A representative summated the dimensions relative to an arbitrary recording from the chorda tympani nerve of a arrangement of orbitals. Their work did not rhesus monkey during a 10 s e c stimulation with a provide or involve the d-value between the sweetener. See text. Modified from Hellekant et complete family of orbitals defined in this al., 1991. work and drawn from the earlier proposals of Shallenberger & Acree, Kier and Beets. The original Tinti & Nofre work was presented in German and has not been translated or widely adopted by the English speaking research community. Their D group is dominated by the cyanides. While they gave dimensions between the various ligands in their figure, they did not provide angles

They did not identify any sensory receptor that could be impacted by their multiple varieties of gustants (or gustaphores) such as AH,Y, AH,B, XH,Y, XH,E1 or XH,E2. The identification of one or more individual gustaphores based on an overlay group within a complex molecule provides a totally different and simpler solution than the model of Tinti & Nofre.

Culbertson & Walters provided a paper in the same volume focused on a 3D model of the sweet taste receptor. Their model was based on the molecular design techniques available in that era and relied upon known structure activity relationships (SAR) as they were known at that time. However, they did not identify any Figure 8.5.1-26 Caricature of extended multipoint receptor and all of their drawings involved model of sugar-receptor coupling. G is shown two dimensional representations. large primarily because it involves very large molecules relative to the other chemical groups. Rohse & Belitz (chapter 13) provided an No bond lengths between these groups were early report on computer modeling to described. From Tinti & Nofre, 1991. discover the underlying rules of step one gustation. Again, there was no default model and they used many very complex molecules (exhibiting many individual gustaphores based on this work) in their analyses .They generally built Signal Generation & Processing 8- 63 on the ideas of Shallenberger & Acree and of Kier. Their framework provided more precision by defining an electrophilic/nucleophilic (e/n) system. Their description of their recognized e/n systems appears internally contradictory compared to similar categorization in this work (see Section 8.5.8.2). They also provided their finding of the perceived taste of these chemicals. However, the allowed tastes appear to have been limited to the dichotomy of sweet versus bitter (Tables II & IV). As part of their explorations, they did define what are called d-values in this work (the distance between pairs of electrophilic and neurophilic sites within an individual molecule). Unfortunately, they did not define the multitude of these values present in their typical MGG molecule. They did define a threshold for sweetness in many cases. Simon wrote on the mechanisms of sweet taste transduction (page237). However, the paper primarily addressed the conceptual possibilities rather than documenting the actual situation. Lindley addressed the subtle differences between similar chemicals that act as either sweetness inhibitors or non-inhibitors (page 251) but without explaining their significance. This material can be reinterpreted using the hypotheses of this work. Most recently, Erickson has written on the core ideas of taste76 and has drawn many comments. The paper was largely philosophical, contained no diagrams and provided little new information on the physiology of the gustatory modality. 8.5.1.6.8 Proteins as stimulants and/or gustaphores

The biological community has long asserted a role for proteins in gustation, however there has been a lack of definitive evidence. The assertion has been primarily associated with the recognition that a very few identified proteins are perceived as sweet. Their sweetness ranges from one hundred to three thousand times that of sucrose on a weight/weight basis. As noted by Caldwell et al.77,“Only six proteins — brazzein, curculin, mabinlin, monellin, pentadin, and thaumatin — have been identified that elicit a sweet taste response in humans.” Pentadin appears to be a di-brazzein resulting from food preparation.

The 1991 ACS symposium addressed above contained only one paper focused on the role of proteins in the perception of sweetness (page 28) along with a paper related to simple peptides (page 41). Several authors noted the continued inability of the community to identify any protein based sensory receptor. As noted by Lancet on page 234, “A most awaited development is the future identification, isolation and characterization of the protein receptors themselves.” This is clearly a Bayesian approach to science, anticipating a desired solution to a problem.

Section 8.5.9.1 will develop the role of “sweet proteins” in greater detail and include a theoretical explanation for their performance 8.5.1.6.9 Tests of the Electrolytic hypothesis of gustation

The bulk of the tests of the hypothesis are embedded in Section 8.5 itself. The successful re- interpretations of much of the previous literature is proof of the hypothesis beyond a reasonable doubt. However, some unique and important tests can be developed. 1. An important test of the hypothesis is whether the simpler acetates, absent any other odorophores, are perceived by the human as acidic. If so, they are clearly stimulating the A- Path of the gustatory modality. signaling plan as predicted by the hypothesis.

8.5.1.7 A dichotomy: the labeled-line and across-neuron-pattern theories

76Erickson, R. (2008) A study of the science of taste: On the origins and influence of the core ideas Behav Brain Sci vol 31, pp 59–105

77Caldwell, J. Abildgaard, F. Džakula, Z. Ming, D. Hellekant, G. & Markley, J. (1998) Solution structure of the thermostable sweet-tasting protein brazzein Nat Struct Mol Biol vol 5, pp 427 - 431 64 Neurons & the Nervous System

How the different stimuli are initially sensed in the peripheral portion of the gustatory system has evolved along two separate paths, the labeled-line theory and the across-neuron-pattern theory. These two approaches are seldom precisely defined and may not be totally distinct. Several authors offer different interpretations of their meaning.

Christensen & White provided a comprehensive overview of this subject78 from its introduction by Dethier in 1976. They note, “In the ‘labeled-line’ model, the sensory axons carrying information to the CNS are ‘absolutely restricted’ with respect to selectivity, whereas in an ‘across-fiber’ code, ‘each stimulus would produce a different and characteristic total response profile’ across the entire population of sensory neurons. [xxx add and reference olfactory portion, Section 8.6.6 & 7 ] Data is now available to resolve the labeled-line versus across-neuron-pattern dichotomy with respect to gustation. Data such as Figure 8.5.1-27 from Pfaffmann et al., and similar individual neurons make it very difficult to support the labeled-line concept in the gustatory modality organization79. They also highlight the analog character of the initial waveforms at the receptor neurons and within that modality. These individual neurons are responsive to a wide variety of stimuli and may contribute to an across-neural-pattern theory. Alternately, they may contribute to a simpler difference channel approach as used in vision.

78Christensen, T. & White, J. (2000) Representation of olfactory information in the brain In Finger, T. Silver, W. & Restrepo, D. eds. The Neurobiology of Taste and Smell, 2nd Ed. NY: Wiley-Liss

79Pfaffmann, C. Frank, M. Bartoshuk, L. & Snell, T. (1976) Coding gustatory information in the squirrel monkey Chorda Tympani, In Sprague, J. & Epstein, A. eds. Progress in Psychobiology and Physiological Psychology. NY: Academic Press. vol 6, pp 1-27 Signal Generation & Processing 8- 65

Figure 8.5.1-27 Records of total chorda tympani responses to water and taste stimuli applied to the anterior tongue. The first two deflections of the marker at the lower edge signal the flow of distilled water, the third deflection signals duration of stimulus flow, the fourth indicates rinse with distilled water. I division = 5 seconds. From Pfaffmann et al., 1976 66 Neurons & the Nervous System

8.5.1.8 Initial identification of human genetic differences

Bartoshuk has discussed the apparently genetic differences between, non–tasters, tasters and super–tasters with reference to specific individual gustaphores80. These differences and the inherent difficulty of comparing the perceived sensations among individual makes for interesting conversations among gustatory researchers. The discussion following the 1993 paper is particularly interesting to an outside observer. More recently, Drewnowski has addresses the genetics of human taste perception briefly and from a narrow perspective81. He notes, “The study of human taste genetics is largely the study of bitter taste.” He does note the important distinction that bitter tastes have a threshold at micro-molar levels, apparently to prevent ingested toxicity while the sweet tastes have a threshold at milli-molar or higher levels. He lists a remarkably wide list of chemicals exhibiting many different functional groups that all exhibit a bitter taste. Based on this list, and no underlying discussion of the gustatory sensing mechanism, he asserts, “That would suggest that bitterness is perceived through a variety of receptors and multiple transduction mechanisms.” Such a conclusion appears unrealistic based on a deeper understanding of the mechanisms involved.

8.5.1.9 Renewal of the gustatory sensory neurons

Van der Heijden has noted, “The taste cells are renewed continuously; the renewal cycle occurs in 10 days to 2 weeks82.” This is the same renewal cycle as found in the visual and auditory sensory neurons. 8.5.2 Analysis of perceived gustatory sensations–MDS and other techniques

Three tools have emerged as the principle means of understanding the taste performance of the gustatory modality. As noted by many, taste is a perceived sensation that may not be due purely to gustatory sensory neurons. In many cases, the olfactory sensory neurons appear to contribute a major portion of the overall taste experience. The somatosensory neurons of the oral cavity appear to also play a significant role.

In common with other sensory modalities, ethical and cultural restrictions have limited the collection of data from the human species related to gustation, other than that from psychophysical experiments. As a result, the major data bases are based on behavioral data from the hamster, rat and occasional monkey.

Table 16.1 by Breslin in Finger, Silver & Restrepo provides a list of thresholds for a variety of gustants and individual gustaphores collected from a variety of sources. No tolerances are provided and the mixed sources make it difficult to correlate the data.

The first goal of any gustatory analysis is to determine the minimum number of distinct sensory channel types required for a species to achieve its desired performance. This requires the collection of data using a variety of stimulus types. To interpret the available data, dendrograms and both two and three dimensional histograms have proven the most useful.

80Bartoshuk, L. (1993) Genetic and pathological taste variation: what can we learn from animal models and human disease? In Margolis, F. ed. The Molecular Basis of Smell and Taste Transduction: Ciba Foundation Symposium 179. NY John Wiley pp 251–267

81Drewnowski, A. (2003) Genetics of human taste perception In Doty, R. ed. Handbook of Olfaction and Gustation. NY: Marcel Dekker Chapter 40

82Van der Heijden, A. (1993) Sweet and bitter tastes In Acree, T. & Teranishi, R. eds. Flavor Science. Washington, DC: American Chemical Society Chapter 3 Signal Generation & Processing 8- 67 The recent availability of SYSTAT, a suite of programs that can be run on a nominal Apple computer system (and after version 13 on a Windows system, has led several investigators to adopt it. In its current form, it is designed to use the modern graphical interfaces of these two computer operating systems. It is basically a display program for relatively simple problems in statistics. It does not specifically address cluster analysis, multidimensional scaling or the optimization procedures related to these types of analyses. The program shares its name with a variety of system level commands and macros within the Linux and Unix environments. At this time the cluster analysis program in the BMDP2M software package is the preferred tool. For MDS analysis, the techniques derived from the original KYST program is preferred. 8.5.2.1 Dendrographic representation

It is important to differentiate between dendrograms based on data acquired at the sensory neurons, later in the neural signaling chain, or ultimately based on psycho-physical perceptions. The goal of stage 2 and stage 4 signal processing is to extract specific focused information from a complex signal environment. The different dendrograms are indicative of the progress in this process. They can be significantly different. When recording stage 3 signals, it is common to report the average pulse rate for an interval on the order of five seconds. This type of report obscures any transient performance and adaptation associated with the gustatory modality. Chemical sensing neurons typically exhibit an initial time constant of less than one second.

[xxx edit following about adjacent figures consider moving part to section 8.5.8.1] Figure 8.5.2-1 shows a typical dendrogram for the hamster from Smith et al83,84. Their work was clearly exploratory. Their dendrograms do not reflect sensory receptor responses (stage 1), but later responses (probably at the output of stage 2) after undefined neural signal processing. The figure represents signals from an unspecified location within (on the surface of) the parabrachial nuclei (PbN) which they differentiated from the nucleus tract of solitarius (NTS). They record extracellular signals and then infer a signal came from a specific cell by the pulse heights of the action potential streams. Additionally, they make a variety of Bayesian assumptions. They assume there are only four gustatory sensory channels. They make the conventional assumption that HCl is a gustatory stimulant without examining the question of whether it is a nociceptor stimulant. They also restrict their stimulation to the anterior portion of the tongue of the hamster, thus significantly skewing their results away from the contribution by the P-channel neurons. They included several neurons exhibiting negligible response to any of the stimulants in their MDS analyses. They restricted their MDS analyses to 2D (which would be appropriate in the absence of P-channel stimulants and receptors. They only employ 30 or 31 neurons depending on the portion of the brainstem examined and only 18 stimulants (only one of which was associated with the P-Path channel of the gustatory modality. Their 18 stimulants were dominated by a large variety of inorganic molecules, numbering 6. With these numbers, little statistical value can be associated with the results. In fact, they noted a significant difference in their results between their best-stimulus classification and their cluster analyses (footnote to Table I, pg 536 and the asterisks in the figure reproduced here).

The figure is more complete than most and includes the designations on the right applicable to individual neurons in the set. However, it is less complete than some in only showing the clustering for three situations, S or sweet-best, N or NaCl-best and H or HCl-best. They reported only one neuron maximally sensitive to the Q-best channel (and then only minimally). Note the asterisks denoting the perceived best response of the individual neurons listed on the right does not always correspond to the clustering on the left. This is a problem in experimental protocol development and may also be confounded by the lack of a sufficiently broad range of stimuli to uncover the true best response of specific neurons. Unless a very large set of neurons is used, the statistical independence of the individual clusters is not assured as the right side of the figure

83Smith, D. van Buskirk, R. Travers, J. & Bieber, S. (1983) Gustatory neuron types in hamster brain stem J Neurophysiol vol 50(2), pp 522-540

84Smith, D. van Buskirk, R. Travers, J. & Bieber, S. (1983) Coding of taste stimuli by hamster brain stem neurons J Neurophysiol vol 50(2), pp 541-558 68 Neurons & the Nervous System is approached. In discussing this and similar figures (page xxx, Smith & Davis employ three best conditions in this figure, four best conditions in immediately adjacent figures (adding quinone or Q-best) and at least provide marginal support for a fifth best condition (adding U or umami- best). They also note, citing several references, “Individual gustatory neurons, both peripheral and central, typically respond to stimuli representing several different taste qualities.” The abscissa is scaled using values from a fundamentally abstract analytical routine, which makes it difficult to compare, and account for the significant differences in the work of different investigators. The cluster diagram does not provide any left terminus. The midpoint of the left- most vertical line can be assumed to represent the sum of the entire ensemble. A similar, but more recent, dendrogram for a primate from Scott & Plata-Salaman omits the labeling of the best performance of the individual neurons but provides a different orthogonal scale from Smith. The scale remains abstract with both positive and negative values. It includes, S-best, N-best, Q-best and H-best categories. They have also added heavy lines attempting to show the level at which the different clusters achieve statistical independence. Their analysis shows that ~73% of the primate neural responses are to sensations described as sweet or salty. Those tuned to quinine constituted ~22% and only ~5% were oriented toward the detection of acids.

Brining et al85. examined the sensitivity of hamsters and describe the distance between their clusters in more detail than Smith et al. did.

8.5.2.2 ROC analysis

Lemon & Smith have recently consolidated their database and applied the “receiver operating characteristic (ROC) analysis technique of Green & Swets (1966) to Figure 8.5.2-1 Cluster dendrogram of 31 hamster action potential recordings from 162 (stage PbN neurons based on similarities in their neural 3) neurons at undefined locations on/within response profiles across 18 gustatory stimuli. The the NST of rats86. Their results included neuron numbers and their best-stimulus measured and standard deviation values. designations (S, H, or N) are shown on the right. Their reported spike rates were from a few The asterisks indicate stimulants that are out of to less than 75 pps. They used an AC place based on the best-stimulus criteria used by coupled oscilloscope that degraded the the authors. The distance between neurons or form of their action potentials significantly. cluster of neurons joined at each step is shown They applied their stimuli to the apical along the abscissa. S, H, & N indicate the three tongue at concentrations chosen to major clusters. From Smith et al., 1983. provide half-maximum response in the chorda tympani nerve. In applying the ROC procedure, Lemon & Smith did not report any neurons where the action pulses exhibited

85Brining, S. Belecky, T. & Smith, D. (1991) Taste Reactivity in the Hamster Physiol Behav vol 49, pp. 1265-1272

86Lemon, C. & Smith, D. (2006) Influence of response variability on the coding performance of central gustatory neurons J Neurophysiol vol. 26(28), pp 7433-7443 Signal Generation & Processing 8- 69 a high spontaneous discharge rate or where the action pulse rate decreased with stimulation of the sensory neurons. These facts suggest the neurons probed were either at the entrance to the NST or were at the exit and did not display any differential encoding. Their discussion suggests they did encounter variation in sensory neuron sensitivity or application protocol inconsistency that led them to assert, “Spike rate was found to be an unreliable predictor of stimulus quality for each neuron tested.” They also concluded, “Results revealed poor classification performance in some cases attributable to wide variations in the sensitivities of neurons that compose a cell type.” These findings are not unexpected based on their protocols, the very great adaptation capability of typical sensory neurons and the poor repeatability of stimulation of a specific neuron within a taste bud. In their discussion, they considered a number of stage 4 computational strategies without presenting any schematic of the appropriate sections of the neural system (or even a semantic model). MacMillan & Creelman87 have recently published “Detection theory: A User’s Guide.” for those interested in exploring the ROC technique in detail.

8.5.2.3 Employing MDS techniques in understanding gustation

An important mathematical technique originally developed to address non-quantitative phenomena in the social sciences has been applied to the sensory system of biology in recent years. It is a very powerful program that generally must be employed in a reiterative manner in order to process the data optimally. During the initial employment of the program, it provides estimates of the adequacy of its performance for a given set of orthogonal axes in interpreting the data set.

Kruskal & Wish were early pioneers in the development of MDS88. Their 1978 book is an excellent primer on the subject. This section will excerpt and quote liberally from that book. It includes citations to about a dozen MDS programs in use at that time, generally requiring large scale computers of that era. There are at least a dozen current programs available to perform MDS analyses, many in desktop computer environments. Detailing the differences between these is well beyond the scope of this introduction. A Google search is suggested if a specific program has not already been adopted by your institution. Two recent texts on MDS are by Borg89 and by Cox & Cox90 8.5.2.3.1 Background relative to the MDS technique–map ,making

Typically, a Highway atlas provides a map of a region and an associated tabulation of the distances between the major cities. Assume no information is given on the size of the cities or their elevation above ground. But consider the problem of establishing the correct relative locations of the cities if one is only given the distances between them in a matrix such as in Figure 8.5.2-2. In this case, the matrix has a null diagonal. From this matrix, it is readily assumed that the cities are all located on a two-dimensional surface. It may also be assumed they are all the same size. However, these assumptions may not be correct. The MDS program is designed to enable a researcher to uncover the “hidden structure” from a set of proximities such as the data in the matrix. This hidden structure, or geometric configuration of points, is the 2D map we are all familiar with. The points in frame (b) describe the geometric configuration of points for the cities in the matrix. The relative position of the cities looks familiar but not quite correct. Notice the program establishes two orthogonal axes (dimensions) with only relative scales. The axes are determined by the distribution of the data points. If that distribution changes (by adding or subtracting data points from the data set), a different set of axes are liable to be chosen.

87MacMillan, N. & Creelman, C. (1991) Detection Theory: A User’s Guide. NY: Cambridge Univ Press

88Kruskal, J. & Wish, M. (1978) Multidimensional Scaling. Sage University Paper series on Quantitative Applications in the Social Sciences, 07-011Beverly Hills, CA & London: Sage Publications

89Borg, I. & Groenen, P. (2005) Modern Multidimensional Scaling: Theory and Applications. NY: Springer

90Cox, T. & Cox, M. (2001) Multidimensional Scaling, Second Edition. Boca Raton, Fl: Chapman & Hall/CRC 70 Neurons & the Nervous System

Figure 8.5.2-2 Creation of a 2D MDS representation based on a matrix tabulation. Top; airline distances between ten United States cities. Note the null diagonal. Bottom; the initial geometric configuration of the cities based on applying MDS to the airline distances and the specification that the points all fall within a 2D plane. Rhumb line at ~+38.5 degrees N. latitude. Modified from Kruskal & Wish, 1978.

If other information is available, such as the latitude and longitude of any two cities, a rhumb line can be drawn between them and (assuming a Mercator type projection) the rhumb line can be considered the hypotenuse of a triangle with the differences in longitude and latitude forming the absolute dimensions related to the data set. In this situation (San Francisco & Washington are on the same latitude, ± 0.5 degrees), the data set can be rotated clockwise by 44 degrees to establish the new intrinsic dimensions 1 and 2 as the actual axes of the data set as in Figure 8.5.2-3. Cox & Cox showed a similar situation for a group of English cities (page 2) but after rotation, the cities remained a mirror image of their correct locations. They introduced a cosine matrix that can be used to rotate the 2D MDS output file before putting the data into the graphics display program. Rotation is about the vertical axis perpendicular to the 2D data Signal Generation & Processing 8- 71 set. The 2D matrix can be easily expanded to cause a left to right reversal of the data set and resulting imagery. The matrix can also be easily modified to introduce absolute scales by adding a multiplier with or without a displacement factor to each dimension. The matrix can also be expanded to process multi-D data sets based on the same procedures. It is easiest if the rotations about the two principle planes of a 3D data set are treated separately. The result is a properly oriented graphic based on absolute positioning, such as the d-value fundamental dimension or in the example longitude and latitude .

Figure 8.5.2-3 Creation of a 2D MDS with absolute scales based on knowledge from an additional source. Same ten cities as in previous figure. Note the intrinsic longitude scale is reversed relative to the relational scale provided by the MDS program. The two latitude scales are not correlated with each other or the longitude scales. See text. Modified from Kruskal & Wish, 1978.

Once the rotation is completed to align the data with a recognized set of coordinates, the scales of the two dimensions can be adjusted (independently) based on the known latitude and longitude of at least two cities. Note the significantly different representations between these two figures if an attempt is made to collapse the data points onto either axis. The resulting 1D representation is essentially meaningless in the first case. After the rotation into the intrinsic dimension space, the 1D representation describes the latitude or longitude of the cities properly in relative (and absolute) dimension. The geographic representations of Kruskal & Wish can be extended into 3D space to indicate the height above sea level of the cities, or the height of the tallest structures in each city. The folding of such data into one dimensional representations must be done carefully to maintain 72 Neurons & the Nervous System legitimacy. If both the height above sea level and the height of the highest structure are introduced into the database, the data can no longer be represented using a 3D presentation space. One dimension must be selected for elimination from the representation (and note of that data deletion should be included in the appropriate text or caption). It is therefore important to remember; C the order of the representation space must be appropriate for the number of dimensions of the database to be represented. • the initial dimensions generated by the MDS program are always arbitrary and have no intrinsic meaning. • the solutions to ordinary MDS are always subject to rotation and mirroring (which is most easily achieved in 2D representations. • when rotation is performed based on other information, the intrinsic dimensions may reveal significant relationships. - - - - The extraction of true axes from initial MDS analyses has become a high art in recent times. Several software packages are available under the labels promax rotation, varimax rotation, etc. In the current situation only extraction techniques maintaining orthogonality is useful. An introductory discussion by J. D. Brown of the University of Hawai’i at Manoa appears particularly helpful91. A lecture by A. Ainsworth of California State University (Northridge) is very useful92. See also Section 8.6.6.2.2.

In the general case, a linear regression analysis is often required after running an MDS program in order to surface the underlying (intrinsic) axes. In many cases, such regression analyses may generate non-orthogonal axes. The analyses may also suggest more dimensions than originally assigned to the data. Additional techniques drawn from factor analysis can highlight both local groupings as well as large mean distances between groupings.

Note the increase in longitude with direction opposite to the positive going direction of the relative dimension suggested by the MDS program. The directionality of the dimensions indicated by the relative scales suggested by the MDS program when in tabular form are arbitrary. They do not suggest a connectivity between these orthogonal dimensions when assembled into a multidimensional representation space.

If additional data is available in matrix form describing the altitude above sea level is available for the ten cities, the MDS program can combine the two matrices and generate a three dimensional map (geometric distribution of points). If an additional matrix is available describing the population of each city, MDS can include this data and develop a 4D geometric distribution of points, ad infinitum. The challenge becomes how to represent this multidimensional data set in the most useful form.

- - - - Determining the dimensionality of the complete data set is an important task in MDS. It is the primary reason the program is run iteratively. The criteria is the goodness-of-fit of the results to the original data set. Sometimes the goodness-of-fit is described as the badness-of-fit on philosophical grounds. In either case, the measure is a stress defined as the square root of a normalized “residual sum of squares.” The calculation of stress is a critically important element of an MDS program. Because of the complexities involved, different MDS programs use different optimization programs, and frequently give different results. A reported “stress” is MDS program specific. A suspiciously high stress may come about because of a failure of the routine used in a specific MDS program to

91Brown, J. (2009) Choosing the Right Type of Rotation in PCA and EFA In Shiken: JALT Test Eval SIG Newsletter vol 13(3), pp 20-25 http://jalt.org/test/PDF/Brown31.pdf

92Ainsworth, A. (xxx) Factor Analysis www.csun.edu/.../Psy524%20lecture 21FA cont.ppt Signal Generation & Processing 8- 73 converge. Most programs limit the number of iterations in the stress calculation to avoid meaningless results and print out an appropriate termination statement. Whenever a stress below 0.01 is found, the possibility of a fully or partially degenerate solution should be investigated. The calculation of stress as a function of dimensionality is based on probability calculations so the results are not static between iterations and can vary dramatically in some cases. To alleviate this problem several different “levels” of stress analysis have been defined and implemented within MDS programs. Because of the variation in macro’s used to calculate the stress, it is common to define the stress level using a program particular label such as the “Kruskal stress value.” It has become common to present the stress level for a set of data by providing the calculated stress for a 1D, 2D, 3D etc. MDS analysis, such as xxx, or as a single value for the MDS analysis chosen for presentation and discussion such as 0.048 for a Kruskal stress value for figure 6 of Hellekant et al., 1997. In gustation investigations, the stress levels are relatively high, even after assuming a 3D data space. This if because the data set is inherently 4D or higher, with the fourth dimension being the intensity of the signal produced by the stage 1 sensory neurons (hopefully in the absence of adaptation). Uneven exposure to the various stimulants can introduce a fifth dimension into the data set. To eliminate the fourth dimension from the data set, some investigators have adjusted the molarity of their test stimulants to generate a constant analog signal amplitude at the output of the sensory neurons or a nominal action potential pulse count on stage 3 neurons orthodromic to the sensory neurons. Adjusting the molarity in this way can ameliorate any difference in the solubility, ionizability or hydration efficiency associated with a stimulant. Adjusting the molarity in this way also accounts for multiple gustaphores of the same type within one stimulant. For stimulants incorporating multiple gustaphores of different type, a more complex procedure is required.

Hellekant et al. (1997) used a set of stimuli varying in molarity over a factor of 10,000:1 to achieve a nominal action potential pulse rate. Oddly, all of their stimulants except quinine were dissolved in an artificial saliva containing significant amount of the sodium ion. The presence of this ion would effectively bias nearly all of their results.

Failing to eliminate the signal intensity variation among the stimulants or gustaphores, and any variation in the stimulant application process, may distort the MDS representation significantly. It may result in the third dimension incorporating the components of both the third and all higher dimensions. The investigator will greatly simplify his work if he eliminates any higher order dimensions in his data set if possible before running any MDS program.

It is not clear how the information present in the higher level dimensions is distributed when the number of dimensions is reduced below the optimum or necessary number. The information is probably distributed in a structured manner. This assumption implies it is not treated randomly and that it can not be compensated for by employing more samples and relying upon the Central Limit Theorem. The example above relied upon quantitative proximities in miles. This is not the general case in most applications of MDS. The proximities are usually arrived at qualitatively, by asking subjects to estimate the similarity/differences between different objects or perceived taste. Such estimates are necessarily noisy, the subject will give different answers if the same samples are presented a second time, in either the same or different order. Thus, it is important to collect the estimated relationships between data points in a data set multiple times in order to obtain optimal results. If the estimates associated with different times are labeled, the MDS will show the degree of clustering of similar responses. This clustering can also be illustrated using a scatter diagram (sometimes labeled a Shepard Diagram) as in Figure 8.5.2-4. The scatter diagram typically shows the percentage of “same” judgements versus the inter-point distances. As with any noisy process, the optimum relationship between the data points will begin to appear as more data sets are added in accordance with the Central Limit Theorem applied to statistical processes. As the number of data points are accumulated in the scatter diagram, it becomes possible to discern the underlying relationship and consider writing an equation passing through 74 Neurons & the Nervous System the centroid of each set of points.

Figure 8.5.2-4 Scatter diagrams and their application to gustation. Left; typical scatter diagram with an estimate of the final noise free result (thin line). Right; a noise free data set and human gustation showing variations due to the collection probability width of the GR’s and the ultimate degenerate condition with constant collection probabilities over a specific width (horizontal line segments). To maintain similarity, right frame indicates 20% of the potential gustaphores were not captured by the defined GR’s. Left frame from Kruskal & Wish, 1978.

A step-like scatter diagram is indicative of the level of degeneracy in the data set, i.e., its lack of continuity in the underlying functional relationship between the points in the data set.

Recognizing the fundamental dimension of gustation before folding, it is possible to present a single scatter diagram (right frame) representing gustation in a species or higher designation of animal(s). For a comprehensive data set with enough data points, discrete step heights will be found at the nominal d-values of each GR present.

Thus a continuous scatter diagram supports the earlier Buck-like assumption of a large number of gustaphores supporting a large number of stimulants on a one-to-one basis. A highly discontinuous (degenerate) scatter diagram supports the combinatorial hypothesis presented here. Buck has recently changed her position to a combinatorial approach.

Degeneracy points directly to the d–values of the underlying individual gustaphores in taste. The Kruskal stress value approaches 0.0 with degeneracy. - - - - The terms metric and non-metric are used in a unique context in MDS. The term metric is used to describe a data set that can be described by a specific equation where the dependent variable is scalable with respect to the independent variable. The term non-metric is used to describe the situation where the dependent variable rises or falls relative to the independent variable but not in a scalable manner. Both the scatter diagram and a multidimensional representation of the complete data set may show a clumping of data points into one or more groups. This is defined as degeneracy in MDS. It is most apparent as a step-like pattern in the scatter diagram. This degeneracy is critically important to finding new relationships about the underlying phenomena (more specifically about finding new combinations of gustaphores in the taste modality). The quality of an MDS analysis is highly dependent on the number of data points in the original set and their distribution over the underlying multidimensional space. A rough rule has evolved that there should be at least twice as many stimulus pairs as parameters to be estimated, to Signal Generation & Processing 8- 75 assure an adequate degree of statistical stability. This can be expressed as the number of stimuli minus one should exceed four times the number of dimensions involved. In gustation, this appears to give a bare minimum requirement unless the stimulants are distributed approximately equally between the four sensory paths, A–, G–, N – & P–. 8.5.2.3.2 Dimensionality–selecting the number of dimensions

Kruskal & Wish discussed the appropriateness of selecting a given number of dimensions to represent a data set (pages 48-58 and Appendix C). They used both real data and random number theory to describe their findings. In general, what has come to be known as the Kruskal stress index is developed. A series of values for this index are frequently given for an ascending number of dimensions in the solution such as 0.246, 0.069, 0.035, and 0.018 for R equal 1, 2, 3 and 4 dimensions. As they note, a stress level of 0.02 represents a very good fit between the MDS representation and the underlying data set. Stress levels above 0.100 for a given dimensionality generally indicate the need to rerun the program using more dimensions or to collect more data or refine the data set. A stress level below 0.005 is indicative of a degenerate data set. Figure 8.5.2-5 shows what has come to be known as the Kruskal stress index as used by Cox & Cox (page 70) in one set of experiments.

Figure 8.5.2-5 The Kruskal stress index as a percentage. To operate at a number of dimensions below the first inflection point, and typically below 5%, which ever occurs first is a nominal goal in statistics. Five dimensions are the suggested complexity to be used for the underlying dataset. Many investigators will develop MDS representations for both 4 & 5 dimensions to determine which provides the best representation. From Cox & Cox, 1994.

Dropping from a two dimensional representation to a one dimensional representation is equivalent to projecting the 2D data set onto one of the axes (after appropriate rotation if necessary). Dropping the third coordinate from a three dimension representation to a two- dimension representation is the same as projecting that dimension perpendicularly onto the plane formed by the remaining two dimension. The effect of reducing a five-dimension representation to a four dimension representation is straight forward from a mathematical perspective but difficult to comprehend otherwise. 76 Neurons & the Nervous System

8.5.2.3.3 Multidimensional scaling applied to gustation

An adequate multidimensional scaling analysis of the taste system can provide strong evidence for resolving the question of the number of independent sensory channels. A problem involves the extreme sensitivity of the scaling procedure to the scope of the stimuli set. In 1963, Gesteland et al. distinguished eight types of neuroreceptors based on one set of criteria93. Holley et al in 1974 seven types of very selective cells and nine “categories” of less selective ones94. The multidimensional scaling technique has matured in recent years. Revial et al. used the technique to describe their data set obtained from the frog95. This set included the camphors and the benzenes and at least one thiophene (a class of sulfur compounds). “Data analysis revealed five factors that described 71% of the total variance, with one factor markedly prevailing over the other four.” “In other words, the olfactory space is multi-dimensional but, in the case of this particular set of compounds, one dimension is dominating.”

[xxx this paragraph needs integration to previously numbered section ] More recently still, Smith et al. have provided a broad investigation into the gustatory modality in the hamster96. The more limited scope of the taste sensation was characterized using only three dimensions, except for one outlier representing quinine hydrochloride. By stretching the QHCl criteria, they placed it in the acid (H) response class. Using this criteria, multidimensional scaling suggested the taste sensations formed into three groups when observed in elements of the brainstem, sugar sensitive S-neurons, salt sensitive N-neurons and acid sensitive H-neurons that could be displayed in a two-dimensional space.

These results support the conclusion that the answers investigators obtain in chemoreceptor research generally depend on what sample set they used.

Figure 8.5.2-6 shows the result of such a multidimensional scaling analysis of glossopharyngeal nerve (NG) fiber of the rhesus monkey data by Hellekant, et al97. This analysis asserts there are three “dimensions” to the stimulus sensing mechanism in the mammalian gustatory system. Each of these dimensions constitutes a value continuum that is orthogonal to the other two dimensions. The challenge is to define these continua using sufficiently precise terminology. The figure also highlights the fact that these dimensions are continua in character and not binary. Similar recordings from many others98,99,100 have used a three-dimensional space to describe the perceived chemoreceptor space, at least for taste. However, both groups omitted several

93Gesteland, R. Lettvin, J. Pitts, W. & Rojas, A. (1963) Odour specificities of the frog’s olfactory receptors in Zotterman, ed. Olfaction and Taste I. London: Pergamon

94Holley, A. Duchamp, A. Revial, M. Juge, A. & Macleod, P. (1974) Qualitative and quantitative discrimination in the frog olfactory receptors An N. Y. Acad Sci vol 237, pp 102-114

95Revial, M. Duchamp, A. Holley, A. & Cacleod, P. (1978) Odour discrimination by from olfactory receptors; a second study Chem Sens Flav vol 3, pp 7-22 & 23-33

96Smith, D. & Davis, B. (2000) Neural representation of taste In Finger, T. Silver, W. & Restrepo, D. eds. (2000) The Neurobiology of taste and smell, 2nd Ed.. NY: Wiley-Liss Chapter 14

97Hellekant, G. Danilova, V. & Ninomiya, Y. (1997) Primate sense of taste: behavioral and single chorda tympani and glossopharyngeal nerve fiber recordings in the Rhesus Monkey, Macaca mulatta. J Neurophysiol vol 77, pp 978-993

98Squire, L. et al. (2003) Op. Cit. pg 644

99Smith, D. Van Buskirk, R. Travers, J. & Bieber, S. (1983a) Gustatory neuron types in hamster brain stem J Neurophysiol vol 50(2), pp 522-540

100Smith, D. Van Buskirk, R. Travers, J. & Bieber, S. (1983b) Coding of taste stimuli by hamster brain stem neurons J Neurophysiol vol 50(2), pp 541-558 Signal Generation & Processing 8- 77 significant groups of stimulants. Using these additional groups may require additional dimensions to provide an appropriate framework for describing this wider set of stimulants. Hellekant has written extensively on monkeys beginning in 1975.

Figure 8.5.2-6 Multidimensional scaling of gustatory stimulants in monkey ADD. From figure 11 of Hellekant et al., 1997.

An important feature of the multidimensional scaling procedure is to achieve optimal alignment of the coordinate axes of the presentation. The procedure provides great flexibility in this area. For a review of the technique and references, see Shepard, et. al101. A figure of merit is used in

101Shepard, R. Romney, A. & Nerlove, S. (1972) Multidimensional scaling; theory and applications in the behavioral sciences. NY: Seminar Press 78 Neurons & the Nervous System these analyses to estimate the broadness of sensitivity of a specific sensory neuron channel. The figure used is an entropy given by the equation, H = .K(n)AΣpiAlogpi where the coefficient K(n) depends on the number, n, of stimulants in the set. Smith & Scott have given a brief discussion of this figure of merit102. The value of K(n) is adjusted to insure a maximum value for this function of 1.0.

A problem not addressed in the presentations based on multidimensional scaling relates to the underlying functions generating the three axes. The three-dimensional character of the multidimensional scaling is strikingly similar to that of the visual system103. The optimized dimensions in the visual case relate to the luminance channel and the two major chrominance difference channels of human vision104. These neurophysical channels describe the perceived scene as processed by stages 1 & 2 within the retina. In the case of vision, the two chrominance difference channels result in two of the three dimensions exhibiting zero’s or nulls along their axes. As a result, it is clear the chrominance outputs of the stage 2 signal processing are electrically bipolar signals. Because of the commonality of the spectral channels participation in the differencing, it is even possible to specify the relative polarity of the regions of these signals. If the signal processing in stage 2 of the chemoreceptor channels mimic that found in most other sensory channels, the knobs atop the vertical stems would point both up and down in [Figure 8.5-1], the presentation would exhibit bipolar data sets. However the CNS does not provide bipolar perceptions. It assigns names, e. g., salty, sugary etc., to perceptions relating to the individual relative peaks in its perceptual space. Whether they were perceive based on bipolar data presented in the saliency map of the parietal lobe or not is irrelevant.

Hellekant et al. made only a brief attempt to define the value continua in this figure. They described dimension 1 as extending from a group of sweeteners in the background of the left side and a group of bitter compound in the foreground. They described dimension 2 as isolating a group consisting of citric and aspartic acid, NaCl, sodium cyclamate and MSG but described no counter to that group. No discussion of the third dimension was offered.

Smith & Scott have provided a similar three-dimensional figure (from Giza & Scott, 1991105) showing the effect of amiloride on taste perceptions or rat for a variety of stimulants106.

8.5.2.3.4 Strange representations due to dimension reduction

Since gustation in mammals involves four distinct signaling channels at stage 1, the qualitative performance of the modality can be represented effectively in a 3-dimensional space. Such qualitative evaluations have generally sought to employ stimulants of equal effectiveness. As a result, the intensity of the perceived taste have not been considered. If variable concentration stimulants are employed (described as quantitative data in this discussion), an additional dimension is introduced that must be accommodated in selecting the number of dimensions employed in the MDS analysis and in the choice of graphic representations. Similarly when discussing the dimensionality of olfaction, it must be recognized that olfaction employs a 9-dimension qualitative space and a 10-dimension quantitative space (Section 8.6.2.9.7 &

102Smith, D. & Scott, T. (2003) Gustatory neural coding In Doty, R. ed. (2003) Handbook of Olfaction and Gustation, 2nd revised and expanded edition. NY: Marcel Dekker Chapter 35

103Romney, K. & Indow. T. (2002) A model for the simultaneous analysis of reflectance spectra and basis factors of Munsell color samples under D65. . . Proc Natl Acad Sci USA vol 99, no 17, pp 11543-11546

104Fulton, J. (2005) Processes in Biological Vision. www.sightresearch.net Chapter 17 Part 1b

105Giza, B. & Scott, T. (1991) The effect of amiloride on taste-evoked activity in the nucleus tractus solitarius of the rat Brain Res vol 550, pp 247-256

106Smith, D. & Scott, T. (2003) Gustatory neural coding In Doty, R. ed. Handbook of Olfaction and Gustation. NY: Marcel Dekker, Chap 35 page 752 Signal Generation & Processing 8- 79 8.6.7.4). Representing gustation in less than three dimensions leads to unusual representations that are difficult to interpret. Scott and colleagues have demonstrated this difficulty most clearly in 1989107. Their figures 5 and 8B can be assembled into one representation even though some of the molecules in the 3-dimensional representation are not present in the 2 and 1-dimensional representations. 8.5.2.3.5 2D MDS representation applied to gustation

A two-dimensional presentation can be obtained based on multidimensional analysis. This form provides the clearest separation of the perceived responses into distinct groups. Figure 8.5.2-7 shows such a presentation from Brining. The precise definition of position on this presentation is complex. The presentation shows there are two distinct perpendicular “dimensions” to the data. However, there is no defined orientation of the axes in this mode of analysis. The axes shown are arranged arbitrarily. The range of the data is scaled by the mathematical tools with the circle shown with a unit radius. It is interesting to note the location of the chlorine radical appears to be independent of these two dimensions. The sugars form a tight group isolated from the other less tightly grouped stimuli. The hydrogen ion present in the quinine hydrochloride sample appears to play no functional role in that stimulus. Smith & Scott (Doty, page 733) have noted the stimuli would be shown on the unit circle if their characteristics were determined exclusively by the two “dimensions.” Clearly, the N-Path representations include an additional dimension while the other selected stimuli provide responses dominated by the two dimensions shown. This is a serious limitation of a two dimensional representation of data containing additional dimensions. This limitation clearly indicates the data needs to be replotted. 8.5.2.3.6 Initial considerations related to entropy in gustation

Several of the papers discussed below include a discussion of a parameter, “entropy,” based on the information content of a message as first developed by Shannon in the 1940's. In its true Figure 8.5.2-7 Two-dimensional histogram of mathematical form, an entropy of 1.0 is hamster taste preferences for 12 different stimuli. indicative of a totally stochastic process of The data needs to be replotted in three multiple dimensions. In the context used in dimensions. See text. From Brining et al., 1991. these papers related to gustation, it is based on the equal stimulation of a large number of poorly defined putative stimulants as developed by Smith, Travers & van Buskirk108. Several authors have suggested this interpretation of entropy is analogous to the concept of white light in a colloquial or informal sense. In fact, white light is not a good analogy to this situation. The recognition of the combinatorial character of the stimulant identification process and the presence of an adaptation mechanism within gustation requires a much more substantive formula for entropy in gustation. Hellekant et al. (1997) have addressed the entropy of individual gustatory pathways at locations

107Scott, T. & Yaxley, S. (1989) Interaction of taste and ingestion In Cagan, R. ed. Neural Mechanisms in Taste Boca Raton, FL: CRC Press Chap. 7

108Smith, D. Travers, J. & van Buskirk, R. (1979) Brainstem correlates of gustatory similarity in the hamster Brain Res vol 4 pp 359-372. 80 Neurons & the Nervous System along the neural system, at the chorda tympani and at the glossopharyngeal nuclei. Such an entropy, different on a pathway basis at these two locations) is arguably different from the entropy associated with the total gustatory perception within stage 5 cognition. An overall entropy might be defined for that related to cognitive perception as separated from .a local entropy related to an individual pathway. In their brief discussion, Hellekant et al. used the term citric acid-best fibers in their table 5, but HCl-best fibers in table 4 to allow comparison between their work and an earlier study. They used citric-acid best because HCl played a negligible role in their study. As noted here, the more appropriate designation is the carboxylic (Lewis) acid-best pathway. Spector & Travis use tuning as a synonym for an entropy value (ranging from 0 to 1.0, page 151) they calculate based on a standardized set of four stimulants (that apparently does not recognize the concentration of gustaphores may differ from the concentration of stimulants containing those gustaphores). They do note, “Very critically, response breadth is affected profoundly by the compounds and concentrations chosen (page 151). Smith & Travers also used an entropy value approach to calculate a breadth of tuning on the assumption that there was only one gustatory sensory channel sensitive to all four of the historical gustatory stimulants109. Alternately, it could be asserted they were evaluating the cumulative gustatory perception to stimulation by the four stimulants.

8.5.2.4 Changes in multidimensional presentation EMPTY

[xxx ad introductory material. Cite one of Kruskal or other authors in support. ] 8.5.2.4.1 Rotation and displacement of MDS axes

Scott and colleagues presented a series of paper utilizing MDS techniques in 1984 through 1991. They did not totally recognize the underlying framework and limitations of the technique that was introduced by Kruskal beginning in 1978. There are substantive conditions on how the axes resulting from MDS analysis can be interpreted and either rotated or displaced ( Section 8.5.2.6.1). These papers are reviewed and their problems discussed in Section 8.5.8 and its subsections. 8.5.2.4.2 Conclusions from analysis of the data

The collected papers of Schiffman provide the most taste data in one place, even though it has not all been collected and presented in a single multi-dimensional space. Some of the discussion accompanying this data may not be relevant in 2010. Figure 2 of her 2000 paper exhibits some strange artifacts of the graphic arts (example, the leading edge of the E/D response for NaCl leans to the left as it rises). The introduction of that paper also includes several descriptions that may be overly limiting. The individual data sets do not represent a portion of the overall taste sensation space. They are the result of individual MDS analyses on incomplete datasets. Lemon & Smith have provided data recorded at the NST of 45 adult male Sprague Dawley rats for a limited group of primarily bitter stimuli110. Their protocol did not employ quinine HCl as a stimulus and they grouped acid/bitter in their dendrogram and 3D histogram. Lacking a quinine HCl-like stimulus, their 3D representation is probably distorted based on applying MDS techniques

109Smith, D. & Travers, J. (1979) A metric for the breadth of tuning of gustatory neurons Chem Senses Flav vol 4(1), pp 215-229

110Lemon, C. & Smith, D. (2005) Neural representation of bitter taste in the nucleus of the solitary tract J Neurophysiol vol 94, pp 3719–3729, Signal Generation & Processing 8- 81 to an incomplete dataset. The conclusion of Giza & Scott appears useful but incomplete in the absence of a group of organic acid stimulants and a larger subject sample size. “The results imply that specific receptors are responsible for the recognition and transduction of sodium salts and that this specificity in the peripheral taste nerves to be manifested in the NTS. Rohse & Belitz introduced computer modeling, basically using comparison techniques, to the search for the underlying structure of molecules perceived as “sweet.” They did not pursue the search into other channels of the gustatory modality. See Section 8.5.8.2. The methods of multi- dimensional analysis are quite sophisticated and do not lend themselves to total computerization using a cook-book program. Beginning with Coombs in 1950, great progress has been made in ordering psychological test data using methods that uncover the underlying “dimensionality” of the data. The work of Shepard in the 1960's led to a technique labeled multidimensional analysis111. It could determine the number of relevant underlying dimensions and the character of the monotonic function relating the data points. A feature of Shepards 1962 paper was his analysis of when the technique would fail (p. 240). A more sophisticated technique labeled single value decomposition (SVD) was developing in parallel with Shepard’s version of multidimensional analysis. The SVD method was described in detail, with examples, in Weller & Romney112. A broader review of these and other techniques can be found in Shepard, Romney & Nerlove113. Shepard asserted in 1962 that these techniques were not practical before the emergence of the high capacity digital computer because of the intense mathematical manipulations required (he used an IBM 7090 which was new in his time period). He could not have known that the fundamental requirements underlying these techniques would be frequently overlooked when it only required pressing a single key to invoke these powerful statistical techniques on a desktop computer.

The histograms plotted using olfactory and gustatory data have used several, but not all of the basis functions resulting from computerized analyses of the above types. In the case of vision, a complete SVD analysis provides four basis functions; a brightness or R–function versus spectral wavelength, and three basis functions (O–, P– and Q–functions) describing three chromatic difference functions versus spectral wavelength. It has been common in the past to ignore the O–function (the ultraviolet component) in vision resulting in an imprecise Q–function. In gustation, it can be expected that omission of the acids as a group in a test protocol will distort the resultant basis functions used to create the above histograms.

111Shepard, R. (1962) The analysis of proximities: multidimensional scaling with an unknown distance function Psychometrika vol27(2), pp 125-139 & vol 27(3), pp 219-246

112Weller, S. & Romney, A. (1990) Metric Scaling: Correspondence Analysis. Series #07-075 Newbury Park, Ca: Sage Publications

113Shepard, R. Romney, A. & Nerlove, S. (1972) Multidimensional scaling; theory and applications in the behavioral sciences. NY: Seminar Press 82 Neurons & the Nervous System

Figure 8.5.2-8 shows the underlying theory of the complete three-dimensional Munsell Color Space114. In practice, the lens of the eye truncates the effective spectral range at 400 nm and the two-dimensional Munsell Color Space results. Note carefully that the white point is not along the spectral locus forming the axes of the figure. Similarly, yellow at a nominal wavelength of 572 nm is not a node of the color space; although yellow is frequently perceived as a dominant color at higher cortical levels by combining brightness and chrominance information. A similar taste space is sought from the experimental data base using SVD.

Without a complete statistically relevant SVD analysis of the taste space in hand, a complete taste space can only be suggested. Figure 8.5.2-9 shows this suggested taste space. This space shows a primary locus of sensory neurons for organic acids at the zero coordinate point. It shows a second primary locus of sensory neurons for sugars (the G-Path) along one axis of the three-dimensional space, another primary locus of sensory neurons for sodium ions (the N-Path) along a second orthogonal axis and a final locus of primary sensory neurons for bitter substances (the P-Path) along the third orthogonal axis. From a behavioral perspective, the acid sugar axis might be considered primary, indicating a likely food at its most positive extent and an undesirable acidic condition at its negative extension. HCL and the alkaline and alkaline earth salts are considered nocents Figure 8.5.2-8 The foundation for the chromaticity in this work and not properly presented in diagram of tetrachromatic vision. W marks the the gustatory perception space of the location of tetrachromatic “white” in the overall saliency map of the neural system, diagram. W’ shows the location of “white” for a even though the inorganic acid, HCl, has long wavelength trichromat. W” shows the been traditionally considered a gustant. location of “white” for a theoretical short The HCl in QHCl is employed to achieve wavelength trichromat. greater solubility for the overall molecule. The quinine moiety contains the P-Path gustaphore.

114Fulton, J. (2005) Processes in Biological Vision. http://neuronresearch.net/vision/pdf/17Performance1a.pdf Section 17.3.3.1 Signal Generation & Processing 8- 83

Figure 8.5.2-9 A taste sensation space based on an incomplete experimental database paths and overlaid by a rotated right-hand rle coordinate space from Section 8.5.1.6.2. With a larger sample set, the values shown near each “Path” node would converge on, and define more precisely, each node. The dashed lines define the total 3-D taste sensation space and a right tetrahedron satisfying Henning’s description. HCl and the alkaline and alkaline earth molecules are not proper elements of this 3D perception space. They belong in a fourth dimension associated with a different modality. See text. Data points from Smith et al., 1983.

Hydrochloric acid is not properly represented in this figure. In dilute solution, this molecule is always dissociated and each ion is hydrated. The hydrogen ion is actually present as either + + + H3O , H5O2 , - - - - . H3O is frequently given the label “hydronium ion.” See Section 8.5.4.3.5 for a broader discussion. The chlorine ion is similarly hydrated. When fully hydrated, its arrangement is similar to that of sodium ion (Section 8.5.4.4) and it is surrounded by six water molecules. The hydration of the hydrogen ion leads to a potential O–H- - O configuration, a hydrogen bond with a d-value of 2.708 Angstrom (Table in Section 8.6.1.8). A DACB with this d-value can form with the GR 2 receptor and be perceived as sweet. In the more general case the chlorine ion is a strong astringent and will act as such and affect the nociceptors of the nocent modality located 84 Neurons & the Nervous System in the mouth (Section 8.7.3.1). The other monovalent and divalent alkali salts are also considered nociceptors rather than gustants. For hydrogen chloride and the alkali salts to be properly shown, a 4-dimensional (4D) MDS analysis would be appropriate. Multidimensional analysis of a complete data set shows that Henning’s tetrahedron (1916) is inappropriate. Henning described the four tastes as located at corners of a equilateral tetrahedron, much as Young had done earlier for the perception of color. While conceptually simple, such a representation does not meet the requirements of orthogonality to avoid cross contamination. The use of selected corners of a cube does provide such independence in both areas of perception, taste and color vision. A right tetrahedron (all face angles at one vertex are 90 degrees) can represent the taste space more appropriately. Note a right tetrahedron drawn within the orthogonal space shown (with its 90 degree vertices at A-Path) occupies only one third of the available taste space proposed here. While sodium saccharide and sucrose fall outside of the tetrahedron shown, they would fall within the orthogonal volume. Other basic tastes could occupy other vertices of a three dimensional cube (See analysis of Schiffman’s paper below). The proposed coordinate system has been overlaid with the data points from Smith et al. (1983b, figure 8). The P-Path/N-Path plane has been rotated by 10 degrees from Smith et al. This is a normal procedure and totally permissible in multidimensional analysis. It is assumed that with a larger and more statistically defendable data set, the data points clustered around each axis would converge on a single value for each class of stimulant thereby defining precisely the locus of the “Best” point for each class of stimulants. The resultant best points would define the peak sensitivity of the underlying sensory neuron types.

[xxx careful with MDS versus SVD ] Smith et al.(1983a) used a very small stimulus data set in their multi-dimensional scaling (MDS) process, including only one stimulant clearly associated with the bitter channel. they used six inorganic compounds that are not relevant to this study. Their response data set only included 31 neurons divided into three groups. It is important to note the data in figure 1 of this paper. The data is from the parabrachial nucleus and not from the actual sensory neurons as might be acquired from nerves VII, IX or X. The distribution of sensitivity among their S–neurons and H–neurons suggest these neurons exhibit the result of a summation process and are not typical distributions of sensory neurons. The power of MDS developed clusters along the expected primary axes, figure 8 of Smith et al. (1983b), but the groupings may be less tight than would be expected of sensory neurons sensing only the organic compounds plus the sodium compounds. As a side note, it is of interest that sodium saccharine did not appear to ionize in solution and thereby influence the N-neurons.

The dimensionality of this figure is arbitrary. However, a convention is preferred. The axes shown are those commonly found in three-dimensional presentations following the right-hand rule. The scales of the individual basis functions have been linearly [xxx vectorially ] modified so that the mean of A-Path occurs at zero and all other values are positive in all of these functions. This approach surfaces a critical thesis concerning gustation. The precise location and the degree of tuning associated with each descriptor, N-Path, P-Path etc. are properties of the sensory neurons. In a statistically adequate group of stimuli. and target systems, these properties are independent of the stimuli employed. While the precise chemistry and the neurological circuitry associated with these scales is unknown, it can be assumed the neurological circuitry at the level of the NST forms differences between the signals from pairs of these nodes (probably limited to pairs along the individual axes). These pairs can be labeled much as they are in the neurological circuitry of color vision space. In the gustatory case, the axes can be associated with; [xxx rewrite text to support the labels below and insure agreement with earlier dimensional spaces. ] Dimension nodes axes in Smith preferred axes Signal Generation & Processing 8- 85 of this work

Dimension 1 = A – S (x-axis) A - G (x-axis) Dimension 2 = S – N (y-axis) G - N (y-axis) Dimension 3 = N –P (z-axis) N - P (x-axis)

In this figure, the mean of the A-path locus is taken as the common junction defined by the basis functions of a new SVD analysis based on a larger, statistically adequate, data set. As noted above, the salts of the alkali earth metals, Calcium, and Magnesium, generally appear along the axis between the A-Path and N-path loci, although they belong to a separate node in a 4- dimensional analysis. This analysis suggests, or supports the hypothesis that, there are four primary types of sensory neurons in gustation; G-path, A-path, N-path and P-path. All gustatory stimulants excite one or more of these types. It may be useful to note that quinine hydrochloride contains the complicated organic molecule, quinine, which is considered an alkaloid (alkali-like) in many pharmaceutical applications. It does not appear to act as an alkaloid in gustation. Urea is a much simpler carbamide H2NCONH2 and is weakly basic. The data base is not large enough to address the question of a distinct umami sensitive sensory neuron. To expand the theory to include a fifth primary sensory neuron type would require; C development of a statistically relevant data base C a database in which all of the potential sensory neuron types were stimulated and C a SVD analysis to develop all of the relevant basis functions. Every indication from this work and the literature is that the umami perception is a composite resulting from the stimulation of the A-path, G-path and N-path simultaneously. See Section 8.5.4.9. 8.5.2.4.3 The basis functions of gustation

The basis functions, included as a step in the preparation of a histogram resulting from a multidimensional analysis or more complex single value decomposition (SVD), can provide a numerical scale for describing the tuning of individual sensory neurons. The SVD method is described in detail, with examples, in Weller & Romney115. A broader review of these and other techniques can be found in Shepard, Romney & Nerlove116. The work of Shepard in the 1960's led to a technique labeled multidimensional analysis117. It could determine the number of relevant underlying dimensions and the character of the monotonic function relating the data points. A feature of Shepards 1962 paper was his analysis of when the technique would fail (p. 240).

While many three-dimensional histograms of gustation have appeared in the literature, the authors have not generally presented the basis functions used to define their orthogonal axes. In the case of vision, where it is easy to provide a constant amplitude stimulus across the visual spectrum, the basis functions are found to describe the mean location and tuning of the visual sensory neurons precisely (Section 17.3.5.2.4 of “Processes in Biological Vision”). Scott & Giza have collected their data and organized it to suggest sensory profiles in their figure 4. However, the relationship to the basis functions is tenuous. This paper also includes a broad discussion highlighting some of the problems in defining the neural coding of gustation.

115Weller, S. & Romney, A. (1990) Metric Scaling: Correspondence Analysis. Series #07-075 Newbury Park, Ca: Sage Publications

116Shepard, R. Romney, A. & Nerlove, S. (1972) Multidimensional scaling; theory and applications in the behavioral sciences. NY: Seminar Press

117Shepard, R. (1962) The analysis of proximites: multidimensional scaling with an unknown distance function Psychometrika vol27(2), pp 125-139 & vol 27(3), pp 219-246 86 Neurons & the Nervous System

MDS is a very powerful analytical tool in the hands of an expert118. It involves considerable routine calculation that is amenable to and essentially requires computer support. Unfortunately, the use of a canned computer program in the hands of the naive can lead to beautiful but erroneous conclusions. The figures on pages 347 & 349 of Getchell et al119. (reproduced from Scott & Mark, 1987) can be used as an example. The data is plotted in two dimensions, but as the captions note, 5 to 6% of the data is unaccounted for using only two dimensions. The phenomena being analyzed involves at least a three-dimensional space. By eliminating one of the degrees of freedom in the computer routine, suitable presentations are obtained but they force the data points into a two dimensional space. The rotation of MDS axes is an additional experimental tool in MDS. The axes selected by the computer routine may not be the primal axes of the data. The choice is determined by factors unrelated to the fundamental axes of the data set. This is particularly true if the statistical sample size and experimental error associated with individual odorophores is inadequate. [xxx edit to agree with MDS paper ] An argument can be made that the primal axes of the figure on page 347 (credited to Scott & Mark120) should pass through NaCl with one axis extending to Fructose/Sucrose and the other perpendicular to that axis as shown in Figure 8.5.2-10. Using the proposed axes, several important points can be made. C Sucrose is not a suitable sample for these experiments, since it is broken down into glucose and fructose by hydrolyzation in the process of salivation. C Both fructose and sucrose contain multiple odorophores based on their ring structure and their CH2OH groups. Thus, these materials are shown farther along the axis than appropriate due to their higher odorophore concentration than expected by the investigators. C The group of chemicals clustered along the axes sloping to the upper right can be separated into the organic acids in the plane of the paper and the picro-receptor stimulants such as quinine, brucine and strychnine out of the plane of the paper.. C The constituents shown on the lower one-dimensional scale can be replotted relative to the new axes and separated into in-plane and out-of-plane components. C The role of the inorganic molecules in gustation (except those containing sodium) have not been addressed in this work. An exception is the role of hydrogen sulfide in both gustation and olfaction. See Section 8.6.3.5.2 for hydrogen sulfide’s role in olfaction. This gas, when hydrated, also plays a role in gustation where it stimulates the sensory neurons of the picric path.

118Weller, S. & Romney, A. (1990) Metric Scaling: correspondence analysis. Newbury Park, California: Sage University Press

119Getchell, T. Doty, R. Bartoshuk, L. & Snow, J. (1991) Smell and Taste in Health and Disease. NY: Raven Press

120Scott, T. & Mark, P. (1987). The taste system encodes stimulus toxicity. Brain Res vol 414(1), pp 197-203. Signal Generation & Processing 8- 87

Schiffman has presented a recent paper that provides data on a wider range of stimulants than usual in an attempt to demonstrate more than four basic tastes. She also attempts to show the equilateral tetrahedron of Henning is inadequate. It is interesting that Schiffman attempts to draw small equilateral tetrahedrons within the boundaries of her orthogonal three- dimensional taste space resulting from a conventional MDS analysis. It is argued here the orthogonal three-dimensional taste space represents a more modern theoretical right tetrahedron concept than did Henning’s equilateral tetrahedron. Schiffman presents a series of three- dimensional spaces derived from MDS for various limited ranges of stimulants. These spaces are necessarily incomplete.

Schiffman & Dackis presented an earlier paper that clearly recognized the multidimensional character of MDS Figure 8.5.2-10 REWRITE A two-dimensional taste analyses121. It contains very useful space with alternate axes applied. The responses information on a range of stimuli. They used are described as taste qualities as determined by an early MDS analysis program by activity profiles across neurons. Dimension 1 Guttman122. It appeared in the same accounts for 91% of the data variance in rats. journal as the previous paper by Shephard Dimension 2 accounts for 4% of the variance. Five mentioned above. They chose to use a percent of the variance is unaccounted for by three-dimensional space after considering these two dimensions. Modified from Scott & up to five dimensions. In that paper Mark, 1987. examining the role of amino acids, vitamins and fatty acids in taste, they noted, “Since the four- and five-dimensional solutions do not appreciably decrease the error and since they do not reveal new relations among the stimuli, . . . a three-dimensional solution was considered appropriate for the data.” They gave the errors as a function of dimension number in the paper; 1D-36%, 2D-21%, 3D-14%, 4D-10% and 5D-7%. Stopping at 3D with a 14% error is considerably poorer than in vision where the 3D error was below 5% in the Indow & Romney dataset.

It appears to be Schiffman”s challenge to show that her additional stimulants fall outside of the three-dimensional space of the four basic tastes. If they do, it is important to incorporate all of the taste sensations, including her metallic tastes in a new multi-dimensional analysis. That analysis should show the existence of more than three significant orthogonal basis functions (beside that for intensity alone). The results would be expected to define a four dimensional histogram which may be difficult to draw on paper, but nevertheless would exist. Schiffman’s interpretation of Henning’s and probably all other tetrahedrons as hollow with all tastes represented on its surface is probably inappropriate. A more likely interpretation is that the tetrahedron is a right tetrahedron where all tastes fall within its volume. Alternately, the three dimensional taste space has several corners (nodes) not presently occupied by any “basic taste.” By abandoning the tetrahedron concept altogether and allow the taste space to expand to a cubic volume, additional possibilities appear. Until proven otherwise, it can be assumed her metallic taste occupies one of the corners (Xmax, Ymax, 0 or Xmax, 0, Zmax or Xmax, Ymax, Zmax) in the taste space proposed here and shown above. [ xxx check it is still above ]

121Schiffman, S. & Dackis, C. (1975) Taste of nutrients: amino acids, vitamins, and fatty acids Percept Psychophys vol 17(2), pp 140-146

122Guttman, L. (1968) A general nonmetric technique for finding the smallest coordinate space for a configuration of points Psychometricka, vol 33, pp 469-506 88 Neurons & the Nervous System

It is appropriate to consider the proposal of the earlier Henning in 1916 as archaic and the use of MDS in the 1970's as primitive, not well understood and investigational. 8.5.2.4.4 The proposed human sensory space of gustation

Comparison between the multidimensional space of gustation and vision suggest a functional analogy. This analogy can be exploited based on the Electrolytic Theory of the Neuron. The analogy to be presented here is based on a number of hypotheses concerning gustation; C MDS analysis of a sufficiently large set of stimuli and large set of species specific subjects. C To exhibit effective gustaphores, the stimulant must be soluble in saliva. C The sensory region of the microvilli lemma exhibits a negative potential due to both the polarization of the sensory receptor molecules and the type 4 lemma of the dendroplasm. C The negative field in the region of the type 4 lemma is attractive to positively polarized ligands of stimuli molecules. C Transduction involves a coordination chemistry process (a narrow type of stereochemical process). C Transduction involves the change in the base potential of the Activa formed within the type 4 lemma of the microvilli, and the resulting change in current flow into the dendroplasm of the sensory neuron. C Transduction probability is the product of a sensory receptor capture cross-section for the gustaphore of interest times the correlation process reactivity coefficient for the sensory receptor type of interest. C The capture cross-section of the sensory receptor is the product of the number of sensory molecules within the area of the type 4 lemma times the number of reaction sites per sensory molecule. C The reactivity coefficient of a stimulus toward a specific sensory receptor is the product of the molarity of the stimulus in solution times the number of reactivity sites per molecule of the stimulus.

Figure 8.5.2-11 shows the proposed taste sensation space for gustation involving four primary taste sensations. Like in vision, signals from the individual sensory channels of gustation are treated as independent within the CNS. Thus, the primary taste receptors can be shown as orthogonal and located at the corners of a three-dimensional space. They are shown here with a color scale drawn from The Electrolytic Theory of Color Vision for orientation and comparison). The scale intervals are all equal in this figure, and equal to 0.10 Angstrom.

The figure shows the location of the four recognized sensory channels of the gustatory modality; the A-path or acido-receptor at 2.276 Angstrom, the G-path or glyco-receptor at 2.82 Angstrom, the N-path or natro-receptor at 3.243 Angstrom and the P-path or picro-receptor at 4.746 Angstrom.

The taste modality sensory neuron receptors have a narrow stereochemical capture range of about ±0.10 Angstrom, narrow band receptors compared to the wide band receptors of vision. Thus, an individual gustaphore must exhibit a d-value very similar to that of the receptor to stimulate the system. Many stimulants consist of multiple gustaphores. As a result, an effective d-value for the stimulant may lie anywhere along the d-value axis based on the relative effectiveness of the individual gustaphores. An assertion by Smith & Davis (page362) can be restated in the context of this work. A universal characteristic of mammalian gustatory neurons is the simultaneous response of distinct sensory channels to a single stimuli (consisting of multiple gustaphores). Signal Generation & Processing 8- 89

The potential for a stimulus to be perceived and described as exhibiting all four primary tastes is finite. Under this proposal, a great many active stimuli can be described as exhibiting multiple primary tastes simultaneously. By its name, it should be obvious that mono-sodium glutamate in solution is capable of stimulating both the gluco-receptor and the natro-receptor channels. The nomenclature of the chemical demonstrates why a separate channel for a umami gustaphore is not needed. The adaptation mechanism associated with each sensory neuron can result in the perceived taste of any stimulus to vary as a function of time, both short term and long term. The importance of this is the complex procedure used in formal wine-tasting.

Figure 8.5.2-11 Proposed taste sensation space for 8.5.2.4.5 The family of Neural a mammalian species exhibiting four primary taste sensations. The sensations are generated by the Response Functions acido-receptors, the gluco-receptors, the natro- receptors and the picro-receptors. Less specific The concept of a neural response function identifiers are shown in parentheses. Based on the (NRF) has appeared a number of times in d-value parameter of the Electrolytic Theory of the literature. It is simple to define a neural Taste & Smell. See text. response function as the response of a specific region of the oral cavity to a range of selected stimuli (using only an arbitrary sequence of the stimuli rather than a calibrated horizontal axis). However, a more detailed definition of such a function, or functions, has been missing. Multidimensional analysis clearly shows the basic taste space is three dimensional and each “best” sensory channel exhibits a sombrero hat shaped neural response function in this space. These hats are the appropriate descriptions of the individual neural response functions.

Woolston & Erickson have developed their concept of an NRF from a totally philosophical and psychophysical perspective123. Their data set includes a variety of stimuli, including the stimuli normally associated with the four best channels. However, they chose to limit their analysis to a two-dimensional taste space. Their conceptual NRF’s are derived from their two-dimensional taste sensation space determined from the NST of 26 Sprague-Hawley rats. They used the Guttman-Lingoes MDS program and show the percent error as a function of the number of independent dimensions used using two different criteria. Frame A of their figure 8 is unlikely for several reasons. First, it does not represent a section through the taste sensation space as described above. Second, it appears the three absorption probabilities shown appear to have been individually normalized to a common height. The actual situation for a given subject and a real stimuli is not properly represented by their caricature. Based on the taste sensation space hypothesized here, a number of more specific NRF’s can be detailed. The goal is to have one or more two-dimensional graphical NRF’s with one axis representing a calibrated variable and the second axis representing the response function. The problem is the three-dimensional character of the taste sensation space. It is possible to define a variety of two-dimensional NRF’s within this space.

123Woolston, D. & Erickson, R. (1979) Concept of Neuron Types in Gustation in the Rat J Neurophysiol vol 42(5), pp 1390-1406 90 Neurons & the Nervous System

MDS provides an arbitrary calibration for each axis of the taste sensation space. These calibrations are stable as long as the stimuli set and subject set are broad enough to encompass the entire taste sensation space. While the calibration is arbitrary, the scale is linear and the distance between any pair of “best” loci is therefore linear. A percentage scale can be calculated using this distance between pairs of “best”loci as the denominator. Figure 8.5.2-12 shows a nominal NRF between A-path and N-path along the X-axis. If all of the scales of the taste sensation space have been adjusted to form spherical shells in the previous figure, the plane of this presentation is rotationally symmetrical about the X-axis.

Figure 8.5.2-12 Proposed 2-D neural response function of the molecule X, NRFX.. The horizontal scale is relative. It can be described in terms of the absolute d-value scale. The individual responses are not normalized to the same level. The peak amplitude of each depends on the precise definition of NRF used. The relative neural responses for stimulus B are equal at a relative scale other than 50% because of the unequal heights of the individual responses. In practice, the individual responses of the sensory receptors are unknown. The dashed lines represent individual responses presented on a logarithmic vertical scale. Many stimulants incorporate multiple gustaphores in order to excite separate sensory receptor channels. See text.

The individual “best” response functions can represent several different combinations of factors. They can represent the relative density of sensory receptor molecules on the surface of the dendrolemma for a given “best” locus. They can represent the product of the relative density and the number of potential binding sites per molecule. They can represent the unbound sites remaining after previous exposure of the sites to stimuli. Since it is also possible for the stimuli to reflect different reactivity to the different “best” sites, it is possible to plot the graph to show the unbound sites for a given “best” locus times the reactivity of a given stimuli to that locus. As shown below, the typical gustatory or olfactory receptor)shows a very narrow response function (dashed lines in the figure and there is little overlap between the individual response functions. This situation is compatible with the “quantized” character of the gustaphores and odorophores of chemical sensing. Because of the available molecular geometries, only a limited set of d-values are available to stimulate the chemical sensing system. On the other hand, many stimulants incorporate multiple gustaphores or odorophores targeting multiple sensory receptor channels. When fully annotated according to the above possibilities, the figure should prove more useful than the conceptual figure 8(A) of Woolston & Erickson and result in more useful scatter plots than the ones they conceptualize.. In the context of gustation, the molar concentration of a stimulus does not necessarily indicate its total reactivity. That parameter is indicated by the molarity of the stimulus times the number Signal Generation & Processing 8- 91 of reaction sites associated with each molecule. [xxx edit ] Since a majority of the stimuli do not appear along one of the axes of the taste sensation space, NRFX, NRFY and NRFZ are of limited utility. An alternate approach is to calculate the likelihood of an individual stimulus’ interaction with each of the “best” sensory channels based on the above choice of parameters. The stimulus can then be described in terms of its relative participation with each of the best channels as [NaGlutamate]x,y,z or NaGlutamate:x,y,z where x,y,z are relative values indicating the salty, bitter and sweet sensations relative to the sour sensation respectively. Using A-path as the 0,0,0 point of the taste sensation space suppresses the role of A-path in the analysis. Since the space is linear, the zero point can be moved to anywhere in the space without introducing distortion. By moving the zero point to correspond to the zero point defined by the underlying MDS dimensions, the x,y,z values become more symmetrical about zero and the four “best” loci are treated more equally. Experience will demonstrate the best method of describing these stimulus values in a practical manner.

8.5.2.5 Conclusions from analysis of the MDS technique & examples EMPTY

The MDS technique is a powerful method of organizing qualitative data (as typically found in the social sciences. Dendrographic techniques provide a similar means to organize primarily qualitative data. Both techniques have been extended to introduce quantitative values for relating the representations of inherently qualitative data.

All of the techniques related to MDS are statistically based and highly dependent on the number of samples in the database. The goal is to use enough independent samples related to each independent sensory channel to achieve a true mean of the total sample (The Central Limit Theorem of the Mean). In this process, all of the samples related to a specific independent channel will congregate ever closer to the actual mean of that channel, ie., the central value of the specific independent channel. In the case of gustation, these central values will be the peak sensitivity of each sensory receptor, the d-value of that channel. With sufficient sample size, the perceived taste will be given precisely, again by the Central Limit Theorem, by the Convolution integral of the individual probability distributions124. Failure to achieve an adequate sample size leads to misplaced elements in a dendrogram and significant spread in the locations of individual samples presented in an MDS representation.

In both dendrographic and MDS analyses, it is important to determine the number of independent variables present in the data set. Failing this determination, the resulting representations of the dataset can not be successfully defended. Kruskal & Wish have developed a stress parameter applicable to either of these techniques. As shown in an earlier figure, the degenerate condition is the ideal. In practice, the Kruskal stress value (or parameter or index) at less than 0.048 (~5%) for the number of “dimensions” chosen is a satisfactory criteria. Data sets not reaching this criteria cannot be relied upon as accurate for research purposes. 8.5.2.5.1 The representation of an MDS dataset in the preferred form

The orientation of the dimension axes developed by current software packages will always be orthogonal but their orientation is entirely arbitrary. It is up to the investigator to determine the absolute orientation and scales of the data as presented based on the availability of additional data. This is a trivial problem if the data set is degenerate; the dataset will be presented with clusters of data points corresponding to the independent channels of the gustatory modality. Under less than degenerate conditions, some estimates must be used. The data points related to one of the expected nodes can be used to rotate the entire data set so the data points relating to that node lie along (or parallel to) the selected axes (but not necessarily at the node). The data points can then be rotated as a group about the selected axes until the other clusters align with their expected axes. If the software did not make the appropriate selection of dimension scales, some mirroring of the data set may be required to arrive at a congruence with the preferred right-hand rule geometry in the final representation.

124Panter, P. (1965) Modulation, Noise and Spectral Analysis. NY:McGraw-Hill pp 124-141 or many statistics texts 92 Neurons & the Nervous System

Once the data set is aligned with the desired axes, the absolute scales appropriate to the individual “dimensions” can be matched to the relative scales provided by the software package. The preferred coordinate system would use the d-values developed in the above preferred representation (or more precise d-values for the individual channels if developed in the future).

8.5.2.5.2Failure to accommodate hidden variables in MDS representations

Historically, investigators in gustation and olfaction have relied upon the use of standardized concentrations of their stimulants in the development of their dendrographic and MDS representations. These solutions have generally been based on molar measures based on; 1) the chemical formula of the stimulants rather than the presence of multiple gustaphores or odorophores (that are the effective stimulants), and 2) without due attention to the solubility of the molecules in the saliva and/or mucosa of the modality. Shepherd established that the concentrations he used did not result in equal intensity stimulation of the olfactory modality (Section 8.6.3.10.1). Any college level textbook would show that the solubility of any homologous chemical family varies significantly (and frequently becomes negligible for members of the family that are solids at the test temperature).

The preparation of a series solutions of molecules for use in gustation or olfaction experiments must accommodate the solubility of the molecules in the saliva and/or mucosa if meaningful dendrographic or MDS representations are to be achieved (unless an added dimension relating to the intensity of the neural signals is included in the overall data analysis. 8.5.3 The 2-step hypothesis of gustatory transduction

The literature of taste based on the chemical theory of the neuron contains a large number of almost completely conceptual descriptions of gustatory transduction. One of the most recent is presented by Gilbertson & Margolskee in Doty125. They describe all of their gustatory transduction mechanisms as dependent on a rise in intracellular Ca++ followed by the release of a chemical neurotransmitter at the pedicle of the sensory neuron. While many labeled materials and presumed operations are shown in their block diagram, no schematic or explanation of the steps in their operations are provided. They show the sensing of amino acids as entirely independent from the sensing of the sugars. Both are shown as requiring undefined pores through the microvilli. Their concepts do not account for the quiescent or dynamic potential at the pedicle of the sensory neuron, the adaptation characteristic of the sensory response, or the neural response function for any of the sensory channels.

[xxx use a different term for dendroplasm and dendroplasm in the following sections, cilia or microvilli lemma and plasma? See page 48 of Cagan & Kare first. they use microvilli lemma ] [xxx may need pages 213 to 246 in Cagan & Kare ]

After reviewing the literature of gustation from the perspective of the Electrolytic Theory of the Neuron, what conclusions can be drawn about the transduction mechanism of gustation? C The rapid replacement of the complete sensory neuron (~200 hrs) suggests transduction involves a substantive change in the structure of each sensory neuron. C The similarity of the gustatory pulse response to that of the visual, auditory and olfaction modalities suggests a quantum-mechanical mechanism is involved, probably involving a chemical reaction involving the outer lemma of the neurons. C The only detailed hypothesis relating to gustatory transduction in the literature relies upon the Law of Mass Action to describe a process described using a second order differential equation.

125 Gilbertson, T. & Margolskee, R. (2003) Molecular physiology of gustatory transduction In Doty, R. ed. (2003) Handbook of Olfaction and Gustation, 2nd revised and expanded edition. NY: Marcel Dekker Chap 34 Signal Generation & Processing 8- 93 C All previously discussed transduction mechanisms in this and related works have involved a change in state of one of the reactants/participants, thus ruling out the Law of Mass Action as an underlying premise. C The great variety of lipids found in the lemma of neurons suggest a sufficient variation in lipid reaction with stimuli to define a set of independent variables in gustation and explain the groupings of stimulus effects in many multidimensional analyses of stimuli. Breslin’s work has suggested the phosphatidyl fatty acid of P-Path transduction. A globoside appears a candidate for the phosphoglyceride of G-Pathtransduction. C The complex grouping of stimuli in various multidimensional analyses based on the stimuli suggest the chemical structures of the stimuli are not an independent variable in gustation. C As demonstrated by Kashiwagura et al., preadaptation with NaCl affects the transduction of CaCl, a compound devoid of sodium. The following proposals will follow the fundamental chemistry found to be operative in the discussion of olfaction, coordination chemistry between very complex molecular ligands associated with specialized (type 4) regions of the lemma of microvilli emanating from individual dendrites.

The challenge is to determine the structure of these complex ligands and their electronic state prior to coordination with a variety of potential stimuli.

The stimuli themselves range from the nonpolar, non-ionizing sugars and sugar derivatives to the highly ionic H+ and Na+ which may be present in solution in more complex coordinate forms. - - - -

As a general introduction, the sensory neurons of the gustatory modality employ specialized (type 4) lemma on the microvilli emanating from their dendritic terminal. The specialized lemma consist of four different phospholipids known collectively as globosides, in the outer leaf of their bilayer structure. Mixed collections of these sensory neurons are grouped in taste buds embedded behind pores primarily in the epithelium of the tongue and oral cavity surfaces. In vivo, the globosides exhibit a negatively polarized molecular surface to the fluid environment of the oral cavity that offers a coordination chemistry association to molecules exhibiting a positive hydrogen ion or a positive sodium ion on their surface, and two more complex molecular arrangements, developed further below, commonly found in the sugars and in quinine hydrochloride. The possibility of additional sensory neuron types sensitive to the monosodium glutamate and various “metallic” salts cannot be dismissed at this time. Negative ions, such as the chlorine ion, are not attracted to the globosides. The coordination chemistry involved appears to generate a stable state that is long lasting. This change of chemical state may contribute to the need to replace the sensory neurons on a 200 hour cycle time.

A conducting form of PtdIns is also a candidate for inclusion in the globosides used as receptors of the gustatory modality. Inositol is considered a sugar derivative although it lacks an oxygen within the ring structure of this cyclohexane. Inositol occurs as nine isomers. The inositols, particularly the epi- and myo- isomers, are known to participate in a variety of coordinate relationships with the alkali and alkali-earth metal ions126. PtdIns is a known phospholipid of the cell lemma. The inositol group of the phospholipid could participate in the necessary coordinate chemistry and are prime candidates for the N-Path sensory neuron receptor. Additional background on the inositols is available127. Lehninger fortuitously identified the four major globosides of gustation in 1970 based on their

126Williams, R. & Atalla, R. (1981) Interactions of group II cations and borate ions with nonionic saccharides In Brant, D. ed. Solution Properties of Polysaccharides Washington, DC: American Chemical Soc. Chapter 22

127Cerny, M. Kocourek, J. & Pacak, J. (1963) The Monosaccharides. NY: Academic Press Chapter 35 94 Neurons & the Nervous System presence in neural tissue128. He also described their biological formation. - - - - Many multidimensional analyses and cluster analyses have shown multiple inconsistencies when interpreted in terms of a reversible chemical reaction leading to transduction. As in olfaction, the conflicts are sufficient to spread doubt that the underlying mechanism involves reversible chemistry. In all other excitation/de-excitation mechanisms studied, a quantum-mechanical mechanism was involved that was not reversible. Thus, it is useful to consider, non reversible mechanisms for gustation, particularly with the close association of gustation and olfaction. A leading candidate for explaining transduction in gustation is the proliferation of different lipids found in specialized areas of the bilayer lemma of the sensory neurons. These lipids have polar terminals that are sugars, alcohols, amines, cholines, and many other complex chemistries. It is possible that the major stimulants attach stereographically or attack these lipids, causing a reaction and generating a free electron that is transferred through the nonpolar lipid tail to the plasma inside the neuron. The reaction also changes the chemical constituency of the lipid involved. As a result of the reaction, the particular lipid molecule is no longer viable. With time, the viability of the entire neuron is degraded in this way and it must be replaced in an orderly process to maintain nominal gustatory sensitivity. Failure to perform this continual replacement, due to age or otherwise, leads to a loss of gustatory sensitivity.

Based on this hypothesis, the challenge is to identify the most likely lipid species able to react with the appropriate group of stimuli.

The many multidimensional analyses in the gustation literature invariable display a two dimensional plane with the stimulants grouped in not more than four major locations. The various cluster analyses also show a few major branches with the various minor branches splitting from these. See also Wright & Michels for the similar situation in olfaction.

It appears reasonable to hypothesize the gustatory sensory neurons operate in a mode very similar to the olfactory transducers and employ specialized regions of lemma encasing the protruding hair (villi) consisting of a few different types of lipid molecules. Under this hypothesis, the transduction function is primarily dependent on the type of lipid present and not on the precise nature of the stimulant. Thus, one type of lipid reacts to produce a signal that is interpreted as sweet, a second type as sour, a third type as xxx etc. irrespective of the chemical formula of the stimulant. A single stimulant could affect more than one type of receptor. The signals from these sensory neurons could be differenced higher up the neural pathway to provide finer bipolar signals describing the individual stimulants. Following the proposed hypothesis, the typical multidimensional analysis can be modified slightly in its support. Figure 8.5.3-1 shows a diagram from xxx with the added scales.

Figure 8.5.3-1 XXX modified multidimensional 8.5.3.1 Previous theories of gustatory analysis based on the polar head of lipids EMPTY.

128Lehninger, A. (1970) Biochemistry NY: Worth Publishing pp 196-200 & 522-526 Signal Generation & Processing 8- 95 transduction

All previous theories of gustatory transduction have been based on the chemical theory of the neuron and have been necessarily largely conceptual. Scott & Mark described the state of understanding of the transduction mechanism in the gustatory modality in 1987, “Attempts to define the organization of the taste system in terms of the physical characteristics of stimuli have been largely unsuccessful.” Interestingly, after annotating reviews of gustatory transduction up through 1991, Getchell et al. note on page 153, “To date, there have been no examples of 2nd messenger modulating ion channels directly in taste cells.” They also note that in the case of at least one bitter compound, denatonium, no membrane conductance change is thought to occur at all; . . .” They conclude, “Since the final pathway converges and most taste cells respond to more than one taste modality, it is not yet clear how taste qualities can be discriminated.” This work will propose an entirely different theory of transduction based on the Electrolytic Theory of the Neuron. The theory eliminates the ambiguities encountered by earlier theories and demonstrates that individual receptors respond to only one specific stereochemical structure and measure the dipole potential of the gustaphore. It is this measurement that is passed to the CNS along a conventional sensory neural circuit. 8.5.3.2 The AH,B & AH,B,X coordination chemistry of the gustatory channel

The conditions required in coordination chemistry are quite complex. Shallenberger has provided a glossary addressing these conditions and their definition (pages 297-301)

The globoside receptors of the desirable/”sweet” gustatory channel are subject to attack by a variety of chemicals. Shallenberger has explored this possibility extensively129. Jakinovich has also offered ideas in this area130. Figure 8.5.3-2 shows the basic coordination chemistry he proposes as the mechanism resulting in excitation of the G-Path sensory neuron. At the minimum, AH represents a hydroxyl proton and B represents a neighboring hydroxyl oxygen atom. While the configuration of these two units need not be identical on the two ligands, the situation strongly suggests a stereochemical relationship, at least in the local area of each ligand. The AH moiety may be OH, NH, NH2 or even CH in halogenated compounds. The B moiety may be O, N, an unsaturated center, or even the π-bonding cloud of the benzene ring.

A unique feature of the hydroxyl group was highlighted by DuBois et al131. The hydroxyl group can act as either the AH or the B element in the AH,B coordinate chemical bond. A single hydroxyl group can act as either a hydrogen bond donor (via the hydrogen atom) or acceptor (via the unpaired electrons of the oxygen atom). Thus two hydroxyl groups of a molecule can provide both the AH structure and the B structure, as they frequently do in the sugars.

Shallenberger & Acree noted a second stereochemical requirement can be used to describe the difference in sweetness between D-leucine and L-leucine. The latter is not sweet to the taste.

129Shallenberger, R. (1982) Op Cit. Chap 10, Sweetness: A stereochemical attribute.

130Jakinovich, W. (1981) Comparative study of sweet taste specificity In Cagan, R. & Kare eds Op. Cit. page 132

131Dubois, G. Walters, D. Schiffman, S. et al. (1991) Concentration–Response relationships of sweeteners In Walters, D. Orthoefer, F. & DuBois , G. eds. Sweeteners. Wash. DC: American Chemical Society Chapter 20, page 275 96 Neurons & the Nervous System

Figure 8.5.3-2 Proposed coordination chemistry of the G-Path sensory neurons clarifying the condition described by Shallenberger & Acree. In the simplest case, all A’s & B’s are hydroxyl oxygen and H’s are hydroxyl hydrogen. Their original text did not differentiate between the distance between AH and B. His later writings referred to the H,B distance as 3 Angstrom. Both numerics are ±7%. Modified from Shallenberger, 1971.

When originally investigating in 1967, Shallenberger & Acree asserted, the initial interaction “is neither a proton transfer nor an electrostatic interaction, but probably involves London dispersion, the principle element of hydrogen bonds.” This comment may be subject to modification where more is known of the environment involved. The biochemistry field has generally adopted the term Van der Waals forces as a synonym for the London forces of “London dispersion.” Wikipedia offers a good overview in this area. Culberson & Lee describe the multitude of potential gustatory sites based on Van der Waals forces (page 219) without producing a practical model of gustation132. They employed a large number of MGGs that did not allow reducing their data adequately. The use of multiple SGG’s (Section 8.5.1.6.3) would have led to more appropriate results. In 1970, Birch et al. initiated a study to determine if there was any reaction chemistry involved

132Culberson, J. & Walters, D.(1991) Three-dimensional model of the sweet taste receptor In Walters, D. Orthoefer, F. & DuBois , G. eds. Sweeteners. Wash. DC: American Chemical Society Chap. 16 Signal Generation & Processing 8- 97 in the AH,B relationship133,134. They concluded that, “there is a chemical reaction basis (stoichiometric basis) for the initial chemistry of the sweetness phenomenon.” While stoichiometry can be expected in coordinate bond arrangements, reaction chemistry exhibits a more demanding requirement, residues. No residues of such a reaction haveever been identified. Shallenberger & Acree (page 263) also established the distance between the AH and the B moiety had to be 3 Angstrom (~2.86–3.2 Angstrom or ±7%). This parameter allowed them to define the specific members of various sugars and other molecules that could participate in the AH,B relationship. In the case of glucose, they showed that only OH-4 and OH-3 were the logical choice for a primary AH,B relationship. Using a galactose, they were able to establish that OH-4 was AH and the only remaining possibility for B is O-3 (Shallenberger, 1982, pp 265-275). After discussing some very sophisticated principles of advanced sugar chemistry, Shallenberger continues on page 275-276 to show that the receptor site structure is disymmetric, but not asymmetric. “This planar structure is capable of resolving the chiral recognition problem presented by the fact that the enantiomeric amino acids possess different taste, but the enantiomeric sugars do not.” Based on the overall analysis, Shallenberger & Acree are able to account for virtually all of the variation in sweetness among the simple sugars, amino acids and other natural chemicals exhibiting the appropriate AH,B relationship. They have also described a mechanism for various hydrolyzed alkali and alkali earth ions to cause a perception of sweetness, particularly at low molar concentrations (page 269). After initially abandoning the tetrahedron concept of taste space in favor of an orthogonal space, Shiffman et al. have provided similar data on amino acids versus their acetylized relatives135. Glutamic acid and aspartic acid, being the only two negatively charged amino acids, would not be expected to taste sweet. Shallenberger & Acree confirmed this observation for glutamic acid (page 246). They also noted that γ-aminobutyric acid (GABA) is not sweet because of the distance between the amine and carboxyl group is too large. This is due to the loss of the closer COO– group, for material within the neural system, by electrostenolysis.

- - - -

Working in the same time period as Shallenberger, Kier has extended the idea of the AH,B coordination bond to include a tripartite or “three-point” coordination model136. This model supports additional discrimination between similar stimuli and also explains the “super sugar” qualities of many man-made sugar substitutes. Kier labeled the third point, X, while Shallenberger chose to use the label γ for “greasy.” Later, the third point became known as the dispersion point. Shallenberger & Lindley (1977) re-examined the work of these two groups as reported by Shallenberger (1982, page 269) and surfaced the fact they were measuring distances from different references, Figure 8.5.3-3. Quoting Shallenberger, “In the former (Kier), the A to B distance is estimated to be 2.6 Angstrom. In the latter (Shallenberger et al.), the AH proton to B orbital distance is estimated to be about 3 Angstrom.” Differences between these two geometries may be significant, but cannot be adequately resolved without numerics of three digit precision. In this work, the best available d-value between A and B is taken as 2.82 Angstrom for the sweet sensitive, G-path, GR.

133Birch, G. Cowell, N. & Eyton, D. (1970A) A quantitative investigation of Shallenberger’s sweetness hypothesis. J Food Tech vol 5, pp 277-280

134Birch, G. Lee, C. & Rolfe, E. (1970B) Organoleptic effect in sugar structures J Sci Food Agric vol 21, pp 650-653

135Schiffman, S. Moroch, K. & Dunbar, J. (1975) Taste of acetylated amino acids Chem Sens Flav vol 1, pp 387-401.

136Kier, L. (1972) A molecular theory of sweet taste J Pharm Sci vol 61(9), pp 1394-1397 98 Neurons & the Nervous System

Definitions: 1. As this discussion progresses, it will become appropriate to modify the definition of dispersion point. There are two locations to be discussed, one on the GR and one or more on the gustaphore(s). These locations are actually determined by the electrostatic fields of the specific moiety. While a centroid of an electrostatic field contour can be defined as the dispersion point, this point may vary with the contour chosen. The term dispersion centroid Figure 8.5.3-3 Comparison of AH,B,X geometries. (DC) may be more appropriate. Left, geometry reported by Kier. Right geometry reported by Shallenberger group. See text. 2. It is not clear yet that the dispersion centroid of a GR must be aligned precisely with a similar dispersion centroid on a gustaphore (like applying a finger axially to a door bell). It may be that the action of the dispersion point, or points, on a gustaphore is to change the location of the dispersion centroid of the GR resulting in a change in the dipole potential presented to the 1st amplifier of the sensory neuron. In this case, the location of the centroid of the electrostatic field of the gustaphore may be lateral to the centroid of the GR. This case also allows multiple locations for gustaphore centroids relative to the location of the GR centroid.

3. The lateral action of a gustaphore electrostatic field on the electrostatic field of the GR is compatible with multiple dispersion centroids located around the dispersion centroid of the GR.

4. In general, a point described as a dispersion point in the literature is more appropriately defined as the centroid of an electrostatic field subject to dislocation resulting in a change in the dipole potential presented to the 1st amplifier (signal detector) within the sensory neuron. One form of dislocation is dispersion.

With regard to the fourth point, Immel, writing in 1995, asserted that the dispersion point may not be a precisely located point (as quantified differently by Kier and the Shallenberger group). He noted the potential electrostatic fields associated with a wide group of super-sweet compounds were changing slowly over a significant portion of the surface of the molecules.

- - - -

The coordination chemistry relationships developed by Shallenberger et al. and by Kier, and the Electrolytic Theory of the Neuron provides a comprehensive description of the sweet sensory neuron that does not involve any GPCR proteins or any movement of sodium ions through the wall of the cell. The described performance of these sensory neurons is much more complete and detailed than that of any other hypothesis. - - - - The primary structure of a stimulant that is perceived as sweet can be defined as its glycophore. Shallenberger ((1982, page 265) defined the glycophore of the majority of sugars and many sugar derivatives as consisting of the hydroxyls at position 3 and 4. OH-4 acts as AH and O-3 acts as B. In other chemicals, the glycophore is represented differently but still exhibits the nominal 3 Angstrom spacing between AH and B. - - - - A more complete discussion of the AH,B,X relationship and its affect on perception appears in Section 8.5.10. Signal Generation & Processing 8- 99 - - - - Figure 8.5.3-4 shows several examples of chemicals considered to taste sweet but do not contain the structure of a sugar. Early investigators were unable to explain the reason for their perceived sweetness. Based on this hypothesis and the advances in computational chemistry, most of the gustaphores of these chemicals can be identified as shown by the brackets marked AH,B. The specifics of these chemical structures will be addressed in later sections. Preview: In the case of α-anisaldehyde oxime (now known as anisaldoxime, CAS 3717-22- 40), the chemical is quick to rearrange. As shown, it exhibits a hand-calculated d-value between the oxygen closest to the ring and the centroid of the ring equal to 2.82 Angstrom. This is the nominal d-value for a G-path glycophore. In the case of saccharin (planar except for the paired oxygen atoms), the hand-calculated d-value of 3.21 Angstrom between the sulfur and the centroid of the phenol ring is 14 % longer than ideal but potentially acceptable for a non-caloric glycophore. The d-value between either oxygen attached to the sulfur and the nitrogen is 2.5 Angstrom, 11% low compared to an ideal glycophore. In essence, this chemical exhibits three marginal glycophores. Each of the chlorine pairs in chloroform are separated by a d-value of 2.94 Angstrom as measured using the program Jmol (4% above the nominal 2.62 Angstrom). In this case, the chlorine atoms are the active orbital pairs in the AH,B relationship. 100 Neurons & the Nervous System

Figure 8.5.3-4 Sweet tasting non-sugars and the their AH,B relationships, with fructose as a reference. In the unsaturated alcohol, the B may consist of the dual bond between two carbons. Similarly, the B of α-anisaldehyde oxime appears to be the centroid of the nucleophilic phenol ring. See text. From Shallenberger & Acree, 1971.

Kier extended the stereochemical requirement on an effective sweetener based on a study of the amino acids and the growing number of man-made sweeteners (typically including amino acid ligands)137. [xxx edit ] Kier developed a three point union involving the AH,B coordination relationship (with a nominal spacing of 2.5 Angstrom) plus an additional bonding of undefined type about 3.5 Angstrom from the A entity and 5.5 Angstrom from the B entity. It is not clear if the difference between the nominal value of 2.5 Angstrom of Kier and the 3.0 Angstrom of Shallenberger & Acree is

137Kier, L. (1972) A molecular theory of sweet taste J Pharm Sci vol 61(9), pp 1394-1397 Signal Generation & Processing 8- 101 significant. The value of 2.5 Angstrom reported by Kier appears to involve more complex AH,B configurations than the simple carboxylic arrangement found in most Shallenberger & Acree formulas. This may suggest, the 2.5 Angstrom value is off-peak compared to the 3.0 Angstrom nominal. Kier conjectured the third union involved “an electron-rich position capable of undergoing electrophilic attack, engaging in localized charge transfer, or capable of participating in some type of bonding involving the electron component such as a dispersion interaction.” Chemicals able to satisfy this three point union exhibit sweetnesses 10's to xxx,000 times greater than fructose, suggesting a significantly larger change in the dipole potential of the receptor per mass unit of sweetener. Eggers, Acree & Shallenberger (2000) have discussed and developed the three-point hypothesis farther using the symbol γ to replace Kier’s X.

- - - - 102 Neurons & the Nervous System

This sensitivity to attack and the resultant transduction signal intensity can be portrayed more graphically as seen in Figure 8.5.3-5.

8.5.3.3 Proposed transduction mechanism in gustation

The section is presented as provisional. There remain no known texts with a focus on coordinate chemistry and most of the work reported in the journal literature is based on metal-based coordinate chemistry. The description of organic coordinate chemistry (lacking any metallic ions) is rudimentary. No material has been located describing equilibrium reactions associated with hydrogen (London) bonds, much less on the DACB type of such bonds.

The mechanism of transduction proposed in this work involves a two-step process. The same mechanism is employed in all four of the sensory channels and receptors. This situation is different than in the visual modality where a different mechanism is Figure 8.5.3-5 Proposed sensitivity of the used in the long-wavelength channel desirable/”sweet” sensory channel ADD using the receptor. multidimensional graphic of xxx in overlay as an example. The first step is a selection process wherein various gustaphores form a dual coordinate chemical bond with one of four types of sensory receptors. The second step involves a net change in the dipole electrostatic potential of the gustaphore/receptor complex compared to the previous dipole electrostatic potential of the sensory receptor alone that is presented to the first Activa of the sensory neuron. The first step provides a gross determination of the character of the gustaphore and the second provides a vernier measurement of the intensity of the gustaphore as a stimulus. A temporal aspect can be associated with the second step. Figure 1 of Jin et al illustrates this aspect. Immel has provided considerable information on the molecular electrostatic potential (MEP) of sugars (Section 8.5.5.1.3). The calculations would need to be extended to give the net dipole potential of a molecule when participating in a DACB. Immel did not discuss the specific location of the X feature on the receptors when participating in an AH,B,X (super-sweet) relationship. In fact, he did not consider the specific location of the X feature on the stimulants. He noted in his abstract to chapter 4, “Most informative in regard to the placement of the tripartite AH-B-X glucophore are the hydrophobicity distributions, which show the lipophilic X-part to be an entire, obviously quite flexible, region rather than a specific corner of the ‘sweetness triangle’.” - - - - Signal Generation & Processing 8- 103 The first two steps will initially be discussed jointly. Section xxx will concentrate on the vernier aspects of perception associated with the ΔDEP.

- - - - Hellekant et al. provided a conceptual drawing of the excitation/de-excitation process based on their data in 1991. It is shown in Section 8.5.1.6.7 and again in Section 8.5.6.6 where the modified time line at the bottom of the figure is discussed. It discusses the delay time due to the introduction of the stimulant and the beginning of the recovery time that includes the delay internal to the sensory neuron involved. The mechanisms underlying the drawing and associated uniquely with the Excitation-De-excitation Equations of Section xxx can now be discussed in greater detail based on the background provided in Section 8.5.1.6.7. C The total delay time consists of the time required by the protocol for the stimulant to reach the sensory neuron receptors plus the intrinsic electrolytic delay within the neuron before the emitter of the Activa achieves the threshol potential provided to minimize internal noise from being treated as authentic signal. C The response time includes the precise response waveform beginning with the end of the delay time measurable at the collector (or nearby axon pedicle) of the neuron. C The decay time represents the actual decay characteristic of the sensory neuron as the stimulating molecules are released from the binding site of the neuron. The waveform is a true exponential curve with a finite time constant characteristic of this part of the mechanism.

The premise adopted here is that the combining and releasing of the stimulant and phospholipid ester of the receptor can be described as an equilibrium reaction within the coordinate chemical sphere (rather than the conventional chemistry sphere). Taking the A-Path as an exemplar, the equilibrium equation for a DACB is shown, using form II of Section8.5.1.6.7 as,

- - - - -

PtdSer + odorophore º PtdSer:odorophore Where the odorophore presents with O and OH orbitals of a carboxylic group

As a result, the above equation can be written in an expanded form as,

PtdSer + O + OH º PtdSer (DACB, or :) odorophore The equilibrium constant for this situation can be written as,

K = KR/KL =[PtdSer:odorophore]/[PtdSer] C [OH] C [OH] - - - - - For a a molecule perceived as sweet, the equations change to reflect the proper receptor and stimulant structure.

PtdTyr + odorophore º PtdTyr:odorophore Where the odorophore presents with two independent OH groups in an equat.– trans–diol configuration. As a result, the above equation can be written in an expanded form as,

PtdTyr + OH + OH º PtdTyr (DACB, or :) odorophore The equilibrium constant for this situation can be written as,

K = KR/KL =[PtdTyr:odorophore]/[PtdTyr] C [OH] C [OH]

These equations assume the formation of the DACB occurs with the simultaneous presentation 104 Neurons & the Nervous System of the two orbitals of the stimulant to the two orbitals of the receptor. They also assume very little delay in the transit of the saliva by the stimulant compared to the measurable time delay of the test instrumentation.

As noted in Section 8.5..1.6.7, there remains a question of Signal Generation & Processing 8- 105 8.5.3.4 Proposed selection mechanism for “desireable/sweet” RENAME

It is important to have an understanding of the numbering system used in sugar chemistry and the possibility of sugars with the hydroxyl groups pointing in different directions. Figure 8.5.3-6 shows these relationships. Current convention calls for numbering to be clockwise beginning with the first carbon to the right of the oxygen in all hexose and pentose sugars. Note in the pentoses that the first carbon is taken as that in the methyl alcohol. Rotation of the axis of the moieties associated with a given carbon can affect the taste sensation significantly. For instance, α-D-mannose is sweet, but β-D-mannose, which is very similar in structure, is distinctly bitter according to Shallenberger & Acree (1971, pg 222) 106 Neurons & the Nervous System

[xxx rewrite the following to separate gangliosides from globosides and address third view of simpler receptor ] Using the description of the gustatory sensory neuron above, the question is what is the structure of the type xxx neurolemma covering the outer surface of the sensory hair(s). The exterior bilayer of the lemma of non-neural cells is known to be populated primarily by phosphatidyl choline (lecithin). However, in neural tissue, the exterior bilayer is known to consist of 20-25% of ganglioside, a very large family of phosphoglycerides138. Figure 8.5.3-7 shows the structure of two gangliosides. Only a few of the available configurations have been investigated in the laboratory139,140. The top frame shows a well studied ganglioside known as ganglioside GM2. Ferrier & Collins have provided an alternate

Figure 8.5.3-6 The numbering system of the simple sugars. Even the pentoses have six carbons. The axial elements associated with each carbon can invert structurally resulting in significantly different chemical properties.

138Christie, W. (2010) Gangliosides: structure, occurrence, biology and analysis http://lipidlibrary.aocs.org/Lipids/gang/index.htm

139Ledeen, R. ed. (1984) Ganglioside structure, function, and biomedical potential. NY: Plenum Press

140Porcellati, G. Ceccarelli, B. & Tettamanti, G. (1975) Ganglioside function: biochemical and pharmacological implications. NY: Plenum Press, c1976 Signal Generation & Processing 8- 107 Haworth Diagram of this ganglioside141. They describe one of the R moieties as ceramide, a saturated lipid. Similar forms of gangliosides are commonly found in the myelin sheaths of stage 3 neurons. The center of the frame shows a ganglioside with the complicated series of amino sugars replaced with a chain of D-galactose moieties bridged in the α(1–>3) arrangement. This arrangement provides three separate locations where a coordinate AH,B bond can occur between a stimulus and this molecule. Only the potential AH,B bond at O-4 and O-6 of galactose(I) is shown explicitly. One of the R moieties is shown fully saturated in order to provide conductivity along the length of the molecule. This modification may not be necessary as Hauser, and Sherer & Seelig have shown operation of the molecule as a molecular electrometer is possible without it (Section 8.xxx). This moiety is a polyethine, a member of the terpene family. Many variations are available within the polar head and the lipid tail of the gangliosides.

141Ferrier, R. & Collins, P. (1972) Monosaccharide Chemistry. NY: Penguin Books page 255 108 Neurons & the Nervous System

Figure 8.5.3-7 Two gangliosides associated with the neural system. Top; the ganglioside GM2 isolated from a Tay-Sach brain. Center; proposed ganglioside supporting the G-Path channel of gustation. The ganglioside contains three galactose ligands chained at the α(1–>3) sites and a polyethine (fully conjugated) lipid for one of the R moieties. This structure provides multiple glycophore binding sites using O-4 & O-6. An optional sialic acid moiety is shown as A at the lower left. Bottom; a simpler ganglioside utilizing only one galactose moiety if AH,B coordinate bonding is only possible using O-3 & O-4.

Lehninger has shown a ganglioside with two galactose moieties that could support two AH,B bonds142. The number of chained sugars can reach ten in some oligosaccharides. The first of three possible AH,B bonds is shown between the hydroxyl hydrogen at C-4 and a hydroxyl oxygen of the stimulus, and a hydroxyl oxygen at C-6 and a hydroxyl hydrogen of the stimulus separated by 2.6 Angstrom. The lower frame shows a simpler potential receptor with only one galactose moiety if the only acceptable AH,B coordinate bonding must use O-3 and O-4 as suggested by Shallenberger & Acree. Gennis has described a digalactosyl diglyceride using a α(1–>6) linkage arrangement that can be extended indefinitely while preserving the capability of O-3 & O-4 to support coordinate

142Lehninger, A. (1970) Biochemistry NY: Worth Publishing page 201 Signal Generation & Processing 8- 109 bond pairing143. The chain of galactose sugars (potentially labeled a galtan) can be extended indefinitely but the effect on the conductivity of the resulting structure is unknown. This family of phosphoglycerides is known to exhibit a negative polar head. In the case of the gustatory sensory neurons, this polarity is enhanced by the negative polarity of the interstitial liquid crystalline water of the type 4 microvilli bilayer. Thus, the in-vivo sensory ganglioside is highly attractive to positive ions introduced into the saliva in significant concentration. Shallenberger has described a series of n-galactose configurations employing different chaining arrangements (pages 224-230). Only two of the six arrangements provide the maximum opportunities for AH,B bonding.

- - - - Concatenation of the galactose moieties can cause an increased capture area for the sugars but potential interference with the L-configuraton sugars. The result would be an increased sensitivity of the channel to the D-configurations at the expense of the L-configurations. - - - -

As noted in Chapter xxx, a bilayer of sphingomyelin (of the same super family of glycosphingolipids as ganglioside) is known to form an electrical diode, i.e. the bilayer is able to conduct electrical charges asymmetrically from one polar head to the other polar head, after surface treatment of one of the electrolytes with 10– 7 g/ml of alamethicin by Mueller & Rudin144. Alamethicin is a cyclopeptide antibiotic containing a variety of amino-acids and at least one carboxyl group. The precise role of alamethicin remains unknown.

The presence of a semiconducting region of type 2 lemma employing a ganglioside in its outer layer introduces the immense possibilities of polysaccharide (sugar) chemistry to the gustatory transduction function. Each of the four saccharide units of the ganglioside exhibit slightly different arrangements and the sugar moiety is subject to attack in a variety of ways. Such polysaccharides are known to be subject to decomposition by dilute bases and dilute acids. They are heteropolysaccharides. They contain both α(1–>4) and α(1–>6) linkages. The presence of a simple saccharide as a stimulus may result in the chaining of the simple saccharide to the heteropolysaccharides of the type 4 membrane. Any addition, decomposition or rearrangement that causes a significant change in current flow into the plasma of the neuron, via the first Activa of the sensory neuron, will constitute the fundamental step in transduction.

The typical lemma does not have two bilayers of ganglioside. It is more likely to have an outer layer of ganglioside and an inner layer of phosphatidyl ethanolamine (PtdEtn). The important feature is that the bilayer form a electrolytic diode that can transport an electron into the dendroplasm of the sensory neuron when such an electron is released from the polar head of the ganglioside.

With the presence of a diode accounted for, any significant rearrangement, fragmentation or polymerization of the polar head could inject the negative charge associated with that head into the plasma. of the sensory neuron. In the laboratory, the sphingolipids as a family are known to be easily hydrolyzed (Lehninger, page 199).

[xxx edit ] The proposed transduction mechanism of the gustatory modality involves the disruption of the chemical structure of the ganglioside bilayer of the type xxx portion of the microvilli lemma and

143Gennis, R. (1989) Biomembranes. NY: Springer-Verlag pg 29

144Mueller, P. & Rudin, D. (1968) Action potentials induced in biomolecular lipid membranes Nature vol. 217, pp 713-719 110 Neurons & the Nervous System the injection of a negative charge (an electron) into the dendroplasm of the sensory neuron. The presence of the electron is sensed by the same electrolytic structure as employed in the olfactory modality. Most of the electrolytic structure is also shared with the visual and auditory modalities, however, the free electron is generated by different means. - - - - 8.4.4.6xxx Sugars as potential GR’s in gustation

[xxx move or renumber ] The significant presence of ganglioside or globoside attached to the surface of the dendrolemma of gustatory microvilli introduce the potential for these phosphoglyceride -sugars to provide the sensory substrate for gustation. Drefus et al. have defined the gangliosides as “a complex group of glycosphingolipids which contain one or more molecules of sialic acid145.” However, as Schauer et al. noted, “Sialic acids are derivatives of neuraminic acid with either an acetyl or glycolyl residue at the amino function and frequently one or more O-acetyl group(s) at C-4, C-7, C-8 and preferably at C-9146.” Thus the gangliosides are a very large family that has been only superficially explored in the literature. Merrill & Sandhoff, as an example, only discuss the saturated sphingolipids as a matter of expediency147. In the same volume, Cook & McMaster provide a set of nomenclature for the broader varieties of saturated and unsaturated fatty acids in phospholipids148. The gangliosides are widely associated with neuron tissue. The distribution of the gangliosides appears focused in the gray matter, the unmyelinated portions of neurons, but this remains controversial. Yamakawa writing in the same volume, noted the high concentration of at least a close relative, galactosyl ceramide, in the white matter of the brain149. Cook & McMaster discuss the wide spectrum of fatty acids needed within a mammalian species and note (page 187), “Unsaturated fatty acids also must be synthesized in cells, supplemented by essential fatty acids in the diet.” They also assert that trans-unsaturated fatty acids are not produced by mammalian enzymes. . .” Their statements about cis-unsaturated fatty acids may be too narrow to include the very small relative numbers of these acids needed in the body.

Galactose, which plays a major role in the following discussion was known as cerebrose in earlier times.

Benjamins et al. have discussed the cerebrosides in the context of their occurrence within the brain150. They distinguish between sphingolipids containing sialic acid (gangliosides) and those that do not (globosides). They hypothesize the creation of these chemicals within the chemical milieu of the brain. They also provide additional information about the long chain fatty acids of the sphingolipids. The material does not address the operation of the sensory neurons or their

145Dreyfus, H Urban, P. Harth, S. et al. (1976) Ritinal gangliosides: composition, evolution with age In Porcellati, G. ed. Ganglioside Function. NY: Plenum Press

146Schauer, R. Schroder, C. & Shukla, A. (1984) New techniques for the investigation of structure and metabolism of sialic acids In Ledeen, R. et al. eds. ganglioside structure, function, and biomedical potential NY: Plenum Press

147Merrill, A. & Sandhoff, K. (2002) Sphingolipids: metabolism and cell signaling In Vance, D. & Vance, J. eds. Biochemistry of Lipids, Lipoproteins and Membranes. 4th Ed. NY: Elsevier Chapter 14

148Cook, H. & McMaster, C. (2002) Fatty acid desaturation and chain elongation in eukaryotes In Vance, D. & Vance, J. eds. Biochemistry of Lipids, Lipoproteins and Membranes. 4th Ed. NY: Elsevier Chapter 7

149Yamakawa, T. (1984) Wonders in glycolipids-a historical review In Ledeen, R. et al. eds. ganglioside structure, function, and biomedical potential NY: Plenum Press

150Benjamins, J. Hajra, A. & Agranoff, B. (2006) In Siegel, G. et al. eds. Basic Neurochemistry, 7th Ed. NY: Elsevier Chapter 3 Signal Generation & Processing 8- 111 receptors. They do describe galactocerebroside (CerGal) as associated with their description of the long chain fatty acid, sphingosine. Crescenzi et al. have described the interaction of the Na+ ion with iso-carrageenan in water at 25°C151. Pass & Hales have explored the changes in enthalpy and entropy when the alkali metal ions interact with polysaccharides152. `Williams & Atalla have discussed the interaction of hexose sugars complexing with group II cations such as calcium and magnesium153. They noted that either 3 or 4 coordinate bonds are formed between the cation and one or two sugar residues. Going beyond the simplest examples is difficult because of the immense number of isomers associated with a specific formula polysaccharide. They note that inositol, the polar head of a candidate phosphoglyceride, forms nine isomers. A three ring polysaccharide, as found in ganglioside, can exhibit up to 56 isomers154. However, the stereo preferences of these sugars may limit the useful number of these configurations. Williams & Atalla noted the exceptional effectiveness of the calcium ion in coordinating with these sugar alcohols in aqueous solution. They speculate this may be due to the specific parameters of the orbital electrons in this species. Sodium is also known to complex with inositol.

8.5.4 The initial selection operation of the gustatory sensory receptors

The overall operation of the gustatory sensory neurons are analogous to the operation of all other sensory neurons studied. For more details of its operation, see Section xxx. Each sensory channel employs a different sensory receptor based on the phospholipid forming the type 4 region of its villi.

This section will develop the stimulant/receptor relationship used by each of the sensory channels of the gustatory modality. Based on the data in the previous section, there are only four distinct channels, a considerably simpler situation than in olfaction.

- - - [xxx edit this down ] Shallenberger (1996) has summarized his thesis on the operation of each of the chemoreceptor types. The terminology is that of the advanced chemist and will not be reported here. In essence, there are at least four types of gustatory sensory neurons capable of sensing each historical class of gustatory sensations. The sensory neurons all rely upon coordination chemistry concepts for transduction and no reaction chemistry result is involved in the transduction process. The features of molecular symmetry/dissymmetry and chirality/pseudo-chirality play a major role in distinguishing between the various stimuli. As a general rule, the transduction process involves the generation of an electrical signal as the result of a pseudo-acid/base titration phenomenon. Shallenberger’s thesis relegates the topology of the chemical structures to a secondary role in transduction. His abstract summarizes his position;

“When considered jointly, all tastes (sweet, salt, bitter, sour) are variations on a common electrostatic mechanism, and the primary distinction among them can be traced to the symmetrical nature of the interaction between the substance and the taste receptor. Sourness is a dissymmetric interaction between the hydronium ion (an acidophore) and

151Crescenzi, V. Dentini, M. & Rizzo, R. (1981) Polyelectrolytic behavior of ionic polysaccharides In Brant, D. ed. Solution Properties of Polysaccharides. Washington, DC: American Chemical Society Chapter 23

152Pass, G. & Hales, P. (1981) Interaction between metal cations and anionic polysaccharides In Brant, D. ed. Solution Properties of Polysaccharides. Washington, DC: American Chemical Society Chapter 24

153Williams, R. & Atalla, R. (1981) Interactions of Group II cations and borate anions with nonionic saccharides In Brant, D. ed. Solution Properties of Polysaccharides. Washington, DC: American Chemical Society Chapter 22

154Hough, L. (1977) Selective substitution of hydroxyl groups in sucrose In Hickson, J. ed. Sucrochemistry. Washington, DC: American Chemical Society 112 Neurons & the Nervous System

the taste receptor, whereas saltiness is a concerted symmetrical electrostatic interaction between the Na+ and Cl– ions (the halophore) and the receptor. Sweetness is elicited through a bilaterally symmetrical and concerted dipolar interaction between a glycophore and the receptor, while bitterness can be traced to either dissymmetric ionic or dipolar interactions between a picrophore and the receptor. As no products are ever formed, taste phenomena are collectively grouped as being due to electrostatic recognition interactions.” Shallenberger expresses each of these relationships using an equation based on the notation of Belitz et al. (1981). In all cases, transduction occurs via a nucleophilic/electrophilic (n/e) taste receptor on the surface of the microvilli emanating from the dendrolemma of the gustatory sensory neuron. Shallenberger concludes, “In taste chemistry, the substrate is not transformed, nor is there a direct physiological effect.” This statement is too simple. There is clearly a physiological effect in that the organism perceives a different state. More specifically, there is no physiological effect (that is physical transformation) recognized by the chemical theory of the neuron. There is a profound physiological effect associated with the electrophysiology of the sensory neuron that is reported to the CNS. The Electrolytic Theory of the Neuron recognizes this effect. A change in the dipole moment of the phospholipid receptor at bilayer interface results in a change in the potential at the base of the first Activa within the sensory neuron. This potential changes by a few hundreds of microvolts to a few millivolts, and results in a change in the electrical potential at the pedicle of the sensory neuron. See Section 8.5.10 xxx.

Shallenberger briefly mentions a “miracle fruit extract,” from the berry of the tree known as Synsepalum dulcificum. Mixed with a sour stimulus, this extract is known to turn the perception of many sour stimuli remarkably sweet. This material is now known as miraculin. The language of Shallenberger can be easily misinterpreted. In ChemSpider, miraculin is a synonym for triacontanol_62194 a totally saturated straight chain aliphatic. This material would contribute to a sensation of sweetness when it formed an azeotrope in solution. 8.5.4.1 Operation of the “sweet” gustatory sensory neuron

Birch has provided an overview of the sugar sensing situation based on the SAR approach155. The focus is on the degree of sweetness of several families of chemicals and the effect of substitutions within these chemicals. He does assert the removal of the anomeric hydroxyl group in sugars does not affect their degree of sweetness. He also reports on the removal of one oxygen at a time at each of the five possible positions in pyranose structures. His findings suggest some alternate alignments between a given sugar and a receptor may be possible or multiple sugar receptors are present that operate according to different rules

Van der Heijden el al. have tabulated the relative sweetness of a large variety of sweeteners156,157. They have also provided chemical diagrams for each of them showing multiple sites of AH,B interaction with the sensory receptors in many cases. Their discussion is the most recent advocating a third feature associated with each stimulant. They described the AH,B distance as ranging from 0.314 nm (3.14 Angstrom) to 0.333 nm (3.33 Angstrom) in these materials. The attack upon the globoside embedded in the microvilli lemma of the neuron causes a change in the potential of the base electrode of the Activa buried in the type 2 lemma of the microvilli. The globoside is no longer functional and it is replaced as part of the 200 hour renewal

155Birch, G. (1976) Structure-activity relationships in the taste of sugar molecules In Benz, G. ed. Structure- Activity Relationships in Chemoreception London: Information Retrieval Limited pp 111-118

156van der Heijden, A. van der Wei, H. & Peer, H. (1985) Structure-activity relationships in sweeteners. I. Nitroanilines, sulphamates, oximes, isocoumarins and dipeptides Chem senses vol 10(1), pp 57-72

157van der Heijden, A. van der Wei, H. & Peer, H. (1985) Structure-activity relationships in sweeteners. II. Saccharins, acesulfames, chlorosugars, tryptophans and ureasChem senses vol 10(1), pp 73-88 Signal Generation & Processing 8- 113 cycle of these sensory neurons. The injected electron is stored on the distributed capacitance of the plasma and results in a change in the base to emitter potential of the sensory Activa. This causes a modulation of the current through the collector-emitter circuit of the sensory Activa. capacitance. The sensory Activa and the distribution Activa are wired in a common emitter circuit. The change in the current through the first Activa causes a mirror change in current through the emitter-base circuit of the distribution Activa. This change results in a significant change in the collector potential (axon potential) of the distribution Activa of the neuron. The base-emitter circuit of the sensory Activa is normally biased just above cutoff and a quiescent current is typically present. This current is due to the presence of agents in the saliva that continually attack the globosides of the outer lemma, releasing a small continual electron current. As in the case of the visual modality, the rate of free electron generation within the base of the sensory Activa and the impedance of the collector current supply source are critical in the overall response of the sensory neuron. They control both the short term transient performance, the excitation/de-excitation (E/D) characteristic, and the long term adaptation properties of the neuron. Shallenberger & Acree (page 244) have reproduced a graph from Cameron, 1947, showing the sweetness as a function of concentration on a log-log graph. The slope is very nearly 1.0 for the simple sugars at concentrations up to 2.0%.

Shallenberger has addressed the Lemieux Effect as additional confirmation of the AH,B theory of G-Path channel sensory operation (page 267).

Van der Heijden (1993, page 108) addressed the potential for multiple sensory neuron receptors to account for the large range of chemical structures contributing to the sensation. Lindley (page 131), in the following article relating to sweetness antagonists reaches the opposite conclusion, “Currently available evidence is consistent with the conclusion that these sweetness inhibitors are competitive antagonists of sweet taste acting at a single receptor structure.” The limited set of stimulants used in prior multi-dimensional analyses cannot totally rule out this option. However, it appears unlikely after exploring the architecture of the neural system. Lindley (page 131), in an article following that of Van der Heijden and relating to sweetness antagonists reaches the opposite conclusion, “Currently available evidence is consistent with the conclusion that these sweetness inhibitors are competitive antagonists of sweet taste acting at a single receptor structure.”

The analysis of the models of taste cells available to Van der Heijden led him to draw much the same conclusions as in this work. [xxx tabulate ] 8.5.4.1.1 Review of the historical database BRIEF

The database of the sugars is extremely large and cannot be reviewed here in a reasonable number of pages. See xxx for an extensive review. A more limited tabulation is provided in the van der Heijden papers of 1985. Walters, Orthoefer & DuBois edited a large compendium of papers on Sweeteners in 1991158. In summary, that volume described a variety of attempts to come up with a practical theory of gustation related to both natural and artificial sweeteners. No satisfactory theory evolved from that work. Shallenberger (1993, section10.7.3) provided calculated values of the equilibrium constant for sucrose, galactose and an unspecified enzyme acting as the gustatory receptor based on the data of McBride and of Maes. He did note that minor changes in the amino acids of the assumed enzymes could be significant in his calculations. He offered no data supporting the presumed participation of enzymes or proteins as receptors in the gustatory process 8.5.4.1.2 Chemical identification of large classes of sugars (saccharides)

158Walters, D. Orthoefer, F. & DuBois , G. eds. (1991) Sweeteners: Discovery, molecular design and chemoreception Washington, DC: American Chemical Society ACS symposium #450 114 Neurons & the Nervous System

The sugars (saccharides) are invariably soluble and polarize electronically but do not ionize. Therefore, they are not described as electrolytes. To simplify the following initial discussion, only stimulants belonging to the polyhydroxyl aldehyde or ketone sugars will be considered. A broader discussion will follow this section. Any polysaccharides (particularly sucrose) applied to the chemical sensing modalities will be assumed to be hydrolyzed into monosaccharides by the mucosa. The resulting monosaccharides will be considered non-electrolytes. Jakinovich has noted a requirement that the three hydroxyl groups of the monosaccharide must be in the equatorial positions for the sugar to be sensed by flies159. His graphics also suggest the critical steric feature of the monosaccharides is the presence of an oxygen in the ring at position 6 adjacent to a carbon at position 5 that is supporting a CH2OH group. While on the right track, these assumptions are modified below. - - - - All monosaccharides are reducing sugars. They will reduce Fehling’s, Benedict’s, Nylander’s or Tollens’ reagents. Fehling’s reagent involves a Cupric ion complexed with tartrate ions. Benedicts’s reagent involves a Cupric ion complexed with a citrate ion. Nylander’s reagent involves bismuth complexed with a tartrate. Tollen’s reagent contains the diamminesilver(I) ion, [Ag(NH3)2]+. These reactions produce products that are frequently highly colored. This coloration may play a part in explaining the coloration of the mucus associated with the olfactory sensory neurons.

Wikipedia provides the following description of a reducing sugar.

A reducing sugar is any sugar that, in basic solution, forms some aldehyde or ketone. This allows the sugar to act as a reducing agent, for example in Benedict's reaction.

Reducing sugars include glucose, fructose, glyceraldehyde, lactose, arabinose and maltose. Significantly, sucrose and trehalose are not reducing sugars.

Benedict's reagent is used to determine if a reducing sugar is present. If it is a reducing sugar, the mixture will turn brick red. Fehling's solution can also be used for the same purpose, as both contain copper (II) ions, which are reduced to a brick red precipitate of copper (I) oxide when the solution is heated. Copper (II) forms a distorted tetrahedral coordination geometry in organic molecules160 that is not directly related to its conventional inorganic structures based on its valence of Cu2+. Cu(I) and Cu(II) both exhibit a conventional valence of 2+ even though they coordinate with four other atoms or ligands (Spiro, pg 114). Electron parametric resonance (EPR) measurements clearly identify three types of Copper in these molecules, Cu(I), Cu(II) and Cu(III) that are illustrated by Spiro (page 122). The complexity of the coordination of copper in a protein is illustrated on pages 35 and 310 of Spiro.

A reducing sugar occurs when its anomeric carbon is in the free form. Since sugars occur in a chain as well as a ring structure, it is possible to have an equilibrium between these two forms. When the hemi-acetal or ketal hydroxyl group is free, i.e. it is not locked, not linked to another (sugar)molecule, the aldehyde (or keto-) form (i.e. the chain-form) is available for reducing copper (II) ions. When a sugar is oxidized, its carbonyl group (i.e. aldehyde or ketone group) is converted to a carboxyl group. [xxx This last sentence can only be interpreted for sugars described using a straight chain chemical representation. All monosaccharides present in solution are properly represented using a ring representation (Lehninger, pg 220)] - - - -

159Jakinovich, W. (1981) Comparative study of sweet taste specificity In Cagan, R. & Kare, M. eds. Biochemistry of Taste and Olfaction. NY: Academic Press pg 119

160Spiro, T. ed. (1981) Copper proteins. NY: John Wiley & Sons pg 6 Signal Generation & Processing 8- 115 As noted above, there appears to be a major problem with consistency among texts, journal articles and Jmol files of the last quarter century attributed to a specific molecule perceived as sweet. This is particularly apparent among the sugars where multiple representations are available for a variety of the most studied sugars. The problem is exacerbated by the multiple configurations, conformations and anomers of the sugars and the continual revisions of the IUPAC naming rules during the 1980's. +, – describes the rotation of polarized light by the sugar of interest (page 74+ in Morrison & Boyd, 2nd Ed, 10th Printing, 1971). D, L– describe the configuration of the molecule about the lowest asymmetric carbon atom, the carbonyl atom being at the top of the Fischer Diagram (regardless of the direction they rotate polarized light (M & B, page 1000). α−, β– describes the configuration of the two enantiomers of a given sugar when in their cyclic form (when they are in solution) based on their degree of rotation of polarized light. The α version causes greater rotation than the β version. The difference is related to the phenomenon of mutarotation involving the orientation of the H and OH groups about the C1 carbon (pages 1002-04). The α– and β– forms of otherwise identical sugars are diastereomers differing only in their configuration relative to the C1 and are therefore defined as anomers (pages 1002-06). After a period of time in solution, a fixed temperature sensitive ratio between the α– and β– components is achieved.

Morrison & Boyd also highlight the problem in illustrating the cyclic sugars using either the Fischer Diagram or the Haworth Diagram when discussing the above conditions (page 1003). Only the conformational representations properly illustrate the position of the CH2OH moiety of the sugar properly. These moieties are almost universally in the equatorial position relative to the ring structure rather than perpendicular to it as suggested by the Haworth Diagram. However, while in the equatorial position, it is important to point out these moieties can either point up or down (by about 19 degrees) relative to the position of the OH groups associated with the ring of a particular sugar.

Variations in the above parameters are very important in determining the precise d-value of a specific sugar. 8.5.4.1.3 The unique “sugar alcohols” or glyco-alcohols

The name sugar alcohols, although of long-standing, appears inappropriate in the context of gustation. Many of these molecules lack the heterocyclic structure of the saccharides and frequently do not exhibit a ring structure at all (ex. mannitol_6015). Mannitol and many other molecules bearing this label do exhibit the glyco-alcohol structure required to be perceived as sweet (two hydroxyl groups separated by two carbons, C–C, that continue to assume an open but cyclic stereo structure..

Inositol is described above as an important “sugar alcohol involved in forming the sensory receptors. This sugar alcohol does exhibit a cyclohexane ring structure.

This work would recommend the term sugar alcohol be replaced by glyco-alcohol (from the perspective of gustation) since the chemical is perceived as sweet but the glycophore stimulating the receptor channel is not attached to a heterocyclic saccharide. 8.5.4.1.4 The unique “sugar acids” or glyco-acids

A class of chemicals perceived primarily as sweet are the sugar acids, named primarily by their method of fabrication rather than their chemical structure. They are similar to the sugar alcohols in this respect (Section 8.5.4.1.3). See also Section 8.5.8.3). There are three common groups of these molecules, the aldonic, aldaric and uronic acids. As in the case of gluconic acid_10240, they exhibit an even less complete ring structure while incorporating a carboxylic acid group. [xxx may want to eliminate d-values between pairs of C=C bonds. ] The JSmol file for gluconic acid, when viewed via the DS 4.1 representation, exhibits d-values of 2.075, 2.076, 2.704, 2.722, 2.803, 2.819, 2.934 Angstrom and many higher values. The hypothesis of this work indicates this molecule is very effective in exciting both the GR 1 and GR 2 channels 116 Neurons & the Nervous System of gustation via multiple DACBs and the potential for marginally exciting the GR 3 channels. However, the molecule is not a saccharide or a cyclohexane. 8.5.4.1.5 Sweetness antagonists (inhibitors)

Lindley provided a chapter in Acree & Teranishi (1993) on molecules that antagonized the perception of sweetness. Many of his sweeteners belonged to the class of “super sweet” molecules discussed in Section 8.5.4.2. His review of the state of the art at that time is particularly useful, although many of the open questions can be resolved based on this work (specifically the fact that the signals of the chorda tympani are orthodromic to stage 2 signal processing in gustation. These signals are frequently based on taking the difference between signals produced by stage 1 sensory receptors supporting different signal paths. Shallenberger (1993, section 10.8) discussed the selective and non-selective inhibition of the sensory receptors. It is also appropriate to discuss the inhibition achieved by interfering with the gustaphores, by either binding to them or by affecting their solubility in the mucosa. Shallenberger suggested the gymnemic acids may impact the receptors by indicating they had to be applied to the tongue before other stimulants. He also noted that several materials, including the gymnemic acids were “surface active” tast modifiers. Ziziphin_390256, is a very complex triterpene glycoside that is considered a taste-modifiers. It could form a DACB with a variety of gustatory receptors. After chewing leaves from the ziziphins, solutions sweetened with sucrose taste like water. The anti-sweet activity is reversible, but sweetness recovery on the tongue can take more than 10 minutes. However, it has no effect on the perception of the other tastes, bitterness, sourness and saltiness161. These characteristics suggest the ziziphins are selective gustatory receptor (GR 2) inhibitors. 8.5.4.2 Operation of the “super sweet” sensory neuron & AH,B,X

Beginning in the 1970's, several sweeteners became prominent that offered a sensation of sweetness far exceeding the previous natural sweeteners. Kier (1972), followed by Shallenberger & Lindley (1977) explored the potential chemical structures that could account for this circumstance. Both groups defined slightly different mechanisms that involved a third element of a tripartite arrangement based on the previous bipartite, AH,B, arrangement. The 1977 paper discusses distances with reference to the two orbitals without considering the location of the hydrogen atom. They described this feature as hydrophobic in nature and capable of participating in a “dispersion” bond (a form of van der Waal bond) with an appropriate hydrophobic bond on the sensory receptor. Figure 8.5.4-1 shows the arrangement of the glycophores of two common sugars based on this tripartite concept.

161Kurihara, Y. (1992) Characteristics of antisweet substances, sweet proteins, and sweetness-inducing proteins Crit. Rev. Food Sci. Nutr. 32 (3): 231–252. doi:10.1080/10408399209527598 Signal Generation & Processing 8- 117

Figure 8.5.4-1 The “super sweet” tripartite glycophores of two sugars. The glycophores of the stimulants are descibed by the shaded areas and the subscripts C. Left; the C-6 methyl carbon acts as the hydrophobic feature. Right; the C-6 methylene carbon acts as the hydrophobic feature. The term γ of Shallenberger & Lindley is interchangeable with the term X of Kier. From Shallenberger & Lindley, 1977

Shallenberger & Lindley also described the glycophores of β-D-glucose and β-L-glucose using the Haworth notation to show how they interacted with the same undefined sensory receptor in a tripartite bond. These sugars show the same degree of sweetness.

The Kier paper discusses several more exotic sweeteners, involving nitrogen and sulfur chemical elements.

Immel made an important assertion in the abstract to his chapter 4 regarding the MEP of the super sweeteners, “Most informative in regard to the placement of the tripartite AH-B-X glucophore are the hydrophobicity distributions, which show the lipophilic X-part to be an entire, obviously quite flexible, region rather than a specific corner of the ‘sweetness triangle’.” See Section 8.5.5.1.3. 118 Neurons & the Nervous System

[xxx need to reference or consolidate parallel molecule structure in Figure 8.5.-51 ] See Section 8.5.10.3 for additional information related to the operation of super sweeteners.

8.5.4.2.1 The unique non– saccharide sweeteners EMPTY

Van der Heijden and colleagues provided an extensive report in 1985 on the SAR properties of two groups of sweeteners that included the saccharins for completeness162,163. The nomenclature in the second paper, involving four parameters, differs from the three parameters used in this work. [xxx expand ] 8.5.4.3 Operation of the “acidic” gustatory sensory neuron

While an acid sensing channel has been associated with the gustatory system since historic times, reports on the related transduction process are virtually nonexistent in the scientific literature. It is easy to assume any acid transduction would involve a hydrogen ion, this may be an oversimplification. Referring to both recent history and his limited range of subject species, Boudreau has noted164, “The system was labeled ‘acid’ because the most stimulating compounds were Brknsted acids and the least stimulating were Brknsted bases. The most excitatory compounds were carboxylic acids for all species. Also stimulating, but at a variable rate, were phosphoric acids and a small number of nitrogen compounds functioning as Brknsted acids.”

This work will show that earlier studies, including that of Boudreau were erroneous. The acidic gustatory sensor neurons of animals are sensitive to and based on organic (Lewis) acids. They do not involve any activity related to an isolated hydrogen ion. It will also be shown that the phosphoric acids and nitrogen compounds function as normal Lewis acids and not as some unique form of Brknsted acids. 8.5.4.3.1 Background

Boudreau’s comments suggest the subtleties between the classic acids, Lewis acids and Brknsted acids need to be reviewed before attempting to understand the acidic sensory receptors.

A Brknsted acid is a proton donor and a base is a proton acceptor. The classic (or Brknsted) acid + 165 ionizes to form a simple proton that may or may not be hydrated, H[H20]n . Pitzer has discussed the hydration of the hydrogen ion in water at biological temperatures and pressures is typically + H[H20)1 . From a practical perspective, Lehninger has also noted, “the formalism introduced by J. N. Brknsted and T. M. Lowry is more useful than that of Lewis in describing acid-base reactions in dilute aqueous systems.” An acid-base reaction always involves a conjugate acid-base pair, made up of a proton donor and the correspond proton acceptor.” In transduction, no acid- base reaction occurs but dilute aqueous systems are generally involved. The more recent development of the Lewis Acid concept appears crucial to transduction.

162Van der Heijden, A. van der Wel, H. & Peer, H. (1985) Structure-activity relationships in sweeteners. I. Nitroanilines, sulphamates, oximes, isocoumarins and dipeptides Chem Senses vol 10(1) pp 57-72

163Van der Heijden, A. van der Wel, H. & Peer, H. (1985) Structure-activity relationships in sweeteners. II. Saccharins, acesulfames, chlorosugars, tryptophans and ureas Chem Senses vol 10(1) pp 73-88

164Boudreau, J. (1989) Neurophysiology and stimulus chemistry of mammalian taste systems In Teranishi, R. Buttery, R. & Shahidi, F. eds. Flavor Chemistry: Trends and Developments. Washington, DC: American Chemical Society Chapter 10

165Pitzer, K. (1982) Self-ionization of water at high temperature and the thermodynamic properties of the ions J Phys Chem vol 86, pp 4704-4708 Signal Generation & Processing 8- 119 Lehninger (page 46) has noted, “A Lewis acid is defined as a potential electron-pair acceptor, and a Lewis base is a potential electron-pair donor. The Lewis acid concept has led to the description of a broad range of covalent and coordinate chemistry bonds in organic chemistry. In acid channel transduction, the use of the Lewis concept appears to offer compatibility with the other coordinate chemistry concepts applicable to the other gustatory channels. This is particularly appropriate where oxygen is involved and it contributes a pair of electrons in a coordinate covalent bond (as in the AH,B coordinate arrangement). According to Morrison & Boyd (page 29), the Lewis acid concept is the most fundamental of the acid concepts. They also note, “To be acidic in the Lowry-Brknsted sense, a molecule must, of course, contain hydrogen.” This is not a requirement with regard to Lewis acids. However, there may be additional requirements on the stimulus, as noted above, besides presenting an available hydrogen ion. Ohloff has presented a theory of A-path sensory receptor operation involving a specific configuration of sensory receptor capable of hydrogen bonding166. It appears to represent a hybrid of the sensory mechanism proposed below. Basically, the sensory receptor is capable of closing a ring by employing a hydrogen bond contributed by the stimulus. There is no DACB relationship. He does not indicate how his proposal accommodates a Lewis acid. It is useful to note that in biochemistry, an apparent disassociation constant, K’, is used based on the analytical conditions of the investigator rather than the more precise disassociation constant, K, of thermodynamics.

It is useful here to consider the inorganic acids as largely outside the realm of natural foods and the sharing of pairs of covalent bonds as the typical situation in gustatory transduction involving + organic acids. This formalism allows the ammonium radical, NH4 to be treated as an organic acid. It is also compatible with treating the carboxyl group as the primary component of most of the organic acids.

A variety of phospholipids are known to be present on the outer surface lemma of the dendritic surfaces of gustatory sensory neurons. Specific molecules from these families, called plasmalogens (or phosphatidal derivatives) by some investigators, with one of the lipid chains unsaturated, are potential candidates for the sensory receptor role in the A-path sensory channel. A slightly more complex arrangement involves the presence of the same plasmalogens in a liquid crystalline surface structure exhibiting electrically conductive features. Such structures have become very significant in the display industry because of their remarkable electrical properties. 8.5.4.3.2 Proposed acidic channel sensory neuron receptor

Based on this discussion and those of the previous sections, the sensory receptor of the A-path (organic acid sensitive) sensory channel is likely to be a structure that can form a coordinate structure with the carboxylic acids. A likely candidate would be PtdSer, a phospholipid known to be present in the lemma of neurons and to contain a carboxylic acid group as shown in Figure 8.5.4-2.

166Ohloff, G. (1981) Bifunctional unit concept in flavour chemistry In Schreier, P. ed. Flavour ‘81. Berlin: Walter de Gruyter pp 757-770 120 Neurons & the Nervous System

Figure 8.5.4-2 Proposed phosphatidylserine shown in polar form This molecule is proposed as the active component of the GR 1 sensory receptor of gustation. It is commonly found on the surface lemma of the dendritic structure of gustatory sensory neurons.

In another depiction, the serine ligand can be described as in Figure 8.5.4-3. This ligand offers a variety of interpretations supporting an AH,B type coordinating situation. The two oxygen atoms shown on the right provide a natural receptor for any other carboxylic acid group seeking to form a DACB with a sensory receptor. The oxygen on the left is expected to readily form an ester with phosphatidic acid. The presence of the amino nitrogen makes it possible for this molecule to form DACBs with other than the carboxylic acids but is believed to be inhibited from this action by the stero-chemistry of the overall phosphatidyl serine and its similar molecules forming a liquid crystal on the surface of the outer lemma.

Discussions of carboxylic acid chemistry frequently begin with a dimer as shown in Figure 8.5.4-3 Serine as the potential sensory Figure 8.5.4-4. While typically shown with R receptor ligand. The oxygen at upper left forms and R’ identical in initial discussions, this is the ester with the phospholipid. The carboxyl not a criteria. The structure shown is planar. group at right forms the sensory receptor site of R and R’ may be quite different without the acid channel sensory neuron. causing major differences in the coordinate bond relationship. Where the differences introduce strains in the bond configuration, the Signal Generation & Processing 8- 121 transduction process may be less effective and the sensation of“acidity” reduced for the stimulus. Almenningen et al. have provided detailed dimensions for the gas phase monomer and the dimer of formic acid167. They provide statistical limits on their parameters after reviewing the earlier work of others. Their statistics may be influenced by the presence of the hydrate of formic acid, methanediol. Interestingly, the so-called hydrogen bond is measured from the centers of the two oxygen atoms. It is proposed that this formation of a dimer between the serine of PtdSer and a carboxylic acid stimulus constitutes the major transduction process in the (organic) acid-best channel of the gustatory modality. The transduction requirement is that the stimulus form a coordinate Figure 8.5.4-4 A gas-phase carboxylic acid dimer EDIT. The bond with the carboxyl dimensions are shown for the carboxyl group of the serine portion group of the serine, thereby of phosphatidyl serine acting as the (organic) acid receptor of changing the dipole the gustatory modality. potential at the base of the first Activa formed by the asymmetrical bilayer lemma of the sensory neuron microvilli. Any stimulus capable of forming a coordinate bond with the serine ligand will elicit a sensation of acidity within the neural system.

The hydrogen bond in each leg of such a dimer typically has an energy of 2-10 kCal/mole. This low energy illustrates the ease with which the bond can be broken following the transduction process.

As suggested by the difference in bond lengths relating to the left carbon, resonance does not appear to play a role in this structure. However, more analysis may be warranted. Valla has described nine different coordinate bonding arrangements for the carboxylic acid group, three that involve two bonding paths168. Because of the carboxyl group in their intrinsic structure, all amino acids can coordinate with the proposed acid sensory receptor and stimulate the acid sensory channel to a degree. Boudreau (page 130) reviewed the relative intensity of these sensations in several species.

- - - - The d-value for phosphatidyl serine present as a liquid crystal embedded in the outer lemma of the cilia of a sensory neuron is desired. At the present time, only a variety of values determined under different conditions are available, Figure 8.5.4-5. The top two values are shown for reference when considering whether the organic acid receptor might be sensitive to other stimulants. The bottom value for the resonant carboxylic group is also shown for reference. Eliminating these values, the remaining values range from 2.223 to 2.34 Angstrom. with the only measured value at 2.268 Angstrom. All other values are calculated from poorly documented models and bond lengths. [It must be noted that AminoAcidsGuide.com on the internet offers no references to its authenticity. Its source is hidden behind a privacy shield.} Based on the Jmol file from ChemSpider, the d-value for the carboxyl group of phosphatidyl serine in liquid crystalline form on the surface of the receptor neuron lemma will be taken as 2.276 Angstrom.

167Almenningen, A. Bastiansen, O. & Motzfeldt, T. (1969) A reinvestigation of the structure of monomer and dimer formic acid by gas electron diffraction technique Acta Chem Scand vol 23(8), pp 2848-2864

168Valla, V. & Bakola-Christianopoulou, M. (2007) Chemical aspects of organotin derivatives of beta-diketones, quinonoids, steroids and some currently used drugs: A review of the literature with emphasis on the medicinal potential of organotins Synth React Inorganic, Metal-Organic, and Nano-Metal Chem vol 37, pp 507–525 122 Neurons & the Nervous System

8.5.4.3.3 The polarized forms of carboxylic acids

Serine is typically described as an amino acid with an uncharged polar group (but with a polarized amino + – group NH2 CCOO . The form of the proposed organic acid receptor is based on the calculated d- value of the neutral, non- Figure 8.5.4-5 Potential d-values for phosphatidyl serine. The top resonant carboxylic acid two rows and the bottom row are shown for reference. The only form. The biochemistry measured d-value is 2.268 Angstrom for a dimer of Formic acid as literature frequently represents the carboxylic a gas by Almenningen et al. This value is also near the median for – the group of O, OH distances. See text. acid group as COO or alternately as a hydrated acid of the form COHOH with a net positive charge. Either of these configurations would suggest different bond lengths and angles resulting in a different d-values. It is necessary to determine what is the structure of the carboxyl group when it is part of the complete phosphatidyl serine molecule with the carboxyl group exposed to the solution of the saliva.

Looking at formic acid for a moment, it is a polar compound that is freely soluble in water in any proportion and are thus completely soluble in water169.

Akiya & Savage have addressed the form of formic acid in solution.170 They showed that formic acid decomposed in the presence of water but all of their results were for a temperature of 700 Kelvin and the necessary multiple atmosphere pressure. They appear irrelevant to the biological conditions. Under biological conditions, it appears the relevant influences are the potential polarization of the parent phosphatidic acid and the pH of the saliva.

Aloisio et al. addressed hydrogen bonding between formic acid and water at laboratory temperatures in the formation of aerosols171. Zhou et al. also addressed hydrogen bonding between water and formic acid in solution using some of the same graphics172. The graphics did not differentiate between single and double bonds to oxygen. Both studies involved computational chemistry. Many of the configurations studied would preclude the carboxyl group acting as a receptor for other than water in the absence of some agent to prevent this type of hydrogen bonding with water. It is possible that it is necessary to incorporate a more realistic saliva in the environment of the formic acid.

Zhou et al. did provide a discussion of dipole moments and bond lengths under different conditions. The dipole moments of the complex in solution and in the gas phase are 1.78 and 5.22 D, respectively. Such a large variation in the dipole moment produces a large change in the

169Xxx see but do not cite my copy of a discussion in ref files/Formic acid in water

170Akiya, N. & Savage, P. (1998) Role of Water in Formic Acid Decomposition AIChE J vol 44(2), pp 405-415

171Aloisio, S. Hintze, P. & Vaida, V. (2002) The Hydration of Formic Acid J Phys Chem A vol 106(2), pp 363-370

172Zhou, Z. Shi, Y. & Zhou, X. (2004) Theoretical Studies on the Hydrogen Bonding Interaction of Complexes of Formic Acid with Water J Phys Chem A vol 108(5), pp 813-822 Signal Generation & Processing 8- 123 solute-solvent interaction energy that favors the aligned configuration. For the R2, it is shortening by 0.042 Å in solution compared to the gas-phase value. The angle C–O - - -H varied from 103.9/ to 147.7/, while O- - -H–O changed from 147.9/ to 160.9/. Differences in other bond lengths between the gas-phase and solution-phase results are less than 0.020 Å, and differences in other angles are less than 4/. These changes in dipole moment and angle appear to be significant. Lehninger (page 196) shows PtdSer exhibits a complex polarization showing a negative value near the terminal carboxyl group, a positive value near the amine group and another negative value near the point of esterification. How, these values change with incorporation into the liquid crystal of the lemma and connection to the first Activa of the neural system is not clear at this time. Lehninger (pages 46-49) has discussed the effect of pH on the “apparent disassociation” constant, K’, of formic and acetic acid as a function of pH at 25o C. The apparent disassociation constant of the biochemist differs from the disassociation constant, K, of the physical chemist. At pH below 5.0, the dominant acetic acid species is the un-disassociated species. Formic acid is more easily disassociated with disassociation dominant above 4.0 pH. Both chemicals exhibit considerable buffering capacity over several whole number values of pH. Saliva pH is typically 6.35–6.85. The tendency of a polar molecule organic to form a shell of water molecules isolating it from the surrounding water would suggest a layer of water hydrogen bonded to the receptor that must be discouraged by some mechanism (possibly catalytic or enzymatic) to support efficient gustatory sensing.

Further discussion of this situation will require experimentation. For the present the best empirical estimate of the d-value of PtdSer appears to remain the value of 2.268. However, the best calculated value for PtdSer appears to be 2.276 from the Jmol file of ChemSpider (provided by the Royal Society of Chemistry). 8.5.4.3.4 The perception of carboxylic acid derivatives as acids

The range of carboxylic acid derivatives is quite large in organic chemistry. Many of these are formed through esterification of the hydroxyl group. The result is a very large number of molecules (the acetates) that still exhibit a d-value near 2.276 Angstrom due to the continued presence of two oxygen atoms capable of sharing their electron-pairs in a DACB. These acetates, if soluble, stimulate the GR 1 channel and are perceived as acidic. 124 Neurons & the Nervous System

8.5.4.3.5 The perception of inorganic acids as nocents–HCl

Only a few inorganic acids have played a historical role in gustation, primarily HCl which has frequently been employed as a standard absent any theoretical foundation. Most inorganic acids are strong relative to organic acids and considered toxic to the biological system. In the terms of chemical sensing, they are generally considered nocents at low concentrations and capable of serious structural damage (including destruction of tissue) at high concentrations. Most inorganic acids are totally ionized in solution and it becomes the hydrogen ion that is actually of interest. This ion does not exist in an isolated form except in elementary chemistry books. It is invariably in a coordinate chemical arrangement of the form H+:(H2O)n. Similarly, most of the negative ions of inorganic acids do not exist in their elementary form in solution. Quoting Powell et al.“ the number of water molecules in the first coordination shell of Cl– (the hydration number) is #6 and depends on both the ionic strength and the type of cation173.” Inorganic acids do not participate in coordinate bonding in their nominal formula state. Neither the proton or most of the residues other than the proton exist in their simple ionized configuration when in solution. Rather, they are typically hydrated. In the case of hydrochloric acid_307, the bond length of the gas is 1.327 Angstrom. In solution, the proton is generally hydrated in a + + +. structure consisting multiple water molecules, typically H5O2 , H7O3 or H9O4 . The simplest of these hydrates exhibits two O- -H bonds. Assuming these bonds are in line, the oxygen atoms could participate in a DACB with an appropriate receptor of the gustatory modality. This hydrate exhibits a d-value of 3.48 Angstrom and results in the stimulation of GR 3 and the perception of a salty material, not an acidic material.

Pauling & Marsh addressed the hydration of chlorine gas in 1952174. Their paper is very informative and suggests it is the hydrate of chloride that is important in gustation related to hydrochloric acid. They predict the chlorine is encased in a shell of water (6Cl2C46H2O) with the oxygen atoms forming much of the outer surface. The result is a chathrate, a compound in which molecules of one component are physically trapped within the crystal structure of another. Bieze et al. have more recently studied the hydration of the chloride ion175. Their conclusion was that Cl– was coordinately bound with six molecules of water resulting in a d-value between the oxygen atoms of 4.44 to 5.45 Angstrom. This range would be ineffective with respect to the primary gustatory receptors. However, it would be relevant to the olfactory modality (Section 8.6.xxx). Foresman & Brooks have reported on the hydration of the chloride ion in solution based on ab initio calculations176. Their analyses converge in general agreement with the conclusion of Bieze et al.

These levels of hydration suggest the inorganic acids will disturb gustation by seeking to acquire water molecules from the mucosa, saliva or cells of the body, rather than by forming a dual coordinate bond (DCB) with the receptors.

Picric acid is an interesting case and discussed in more detail in Section 8.5.4.5. Its name is ancient, suggesting its bitter taste was often confused with the sour taste of organic acids. It is an organic carbohydrate compound but does not contain a carboxyl group. It can become a Lewis acid through the loss of the hydrogen of the hydroxyl group or be considered a negative

173 Powell, D. Barnes, A. Enderby, J. et al. (1988) The hydration structure around chloride ions in aqueous solution Faraday Discuss Chem Soc vol 85, pp 137-146 DOI: 10.1039/DC9888500137

174Pauling, L. & Marsh, R. (1952) The Structure of Chlorine Hydrate PNAS vol 38, pp 112-118

175Bieze, T. Tromp, R. van der Maarel, J. et al (1994) Hydration of Chloride Ions in a Polyelectrolyte Solution Studied with Neutron Diffraction J Phys Chem vol 98 (16), pp 4454–4458

176Foresman, J. & Brooks, C. (1987) An ab initio study of hydrated chloride ion complexes: Evidence of polarization effects and nonadditivity Chem. Phys. vol 87(10), pp 5892-5895 Signal Generation & Processing 8- 125 ion by internal charge rearrangement. However, its major role in gustation is as a tastant containing three identical gustaphores with d-values of 4.746 Angstrom that stimulate the distinctly separate picric pathway significantly. 8.5.4.4 Operation of the “alkaline” gustatory sensory neuron

The alkali metals and the alkali earth metals are distinctly different and should not be lumped together when considering their theoretical taste performance. The alkali metals (Lithium, Sodium etc.) are singularly ionized while the alkali earth metals (Calcium, Magnesium etc.) are typically doubly ionized. Of greater importance is their different coordinate bonding potential. It was found earlier that the alkali metal ions were perceived as sweet at low concentrations and this is believed to be due to their hydration resulting in an AH,B coordination capability with the G-Path receptors.

Avenet & Lindemann (1989) have noted177, “Of the diversity of tastes induced by inorganic salts, that of NaC1 has found primary attention. For man, NaC1 is the only substance which has a pure salty taste, and the sodium ion appears to be more important for this taste than the anion.”

Quist & Marshall178,179 have described the hydration of NaCl and Na+ & Cl–. They note the nearly + constant hydration, Na[H2O]6 , regardless of temperature for concentrations above 1.2 0 moles/liter and NaCl[H2O]2 below 1.2 moles/liter. Zumdahl has provided the hydration energy of the sodium ion, presumably for the fully hydrated case, as –402 kJ/mol180. He also provided the first ionization energy of sodium as +495 kJ/mol.

Marshall provided the ionic radii for Li, Na & K as 0.6, 0.98 & 1.33 Angstrom respecively181. He also provided the atom radii for these materials as 1.56, 1.86 & 2.23 Angstrom.

Shallenberger (1996) noted a general consensus that only NaCl elicited a true salt taste. His words may be too casual. If the coordinate chemistry of the sodium ion is the principle contributor to the N-Path channel’s response, it should be clear that it is the sodium ion (possibly at a specific level of hydration) that elicits the maximum (or “true”) N-Path channel sensation. The presence of the chlorine ion is largely irrelevant.

Figure 8.5.4-6 illustrates the two most common states of hydration of the sodium ion. In dilute + solution, it is believed to form Na(H20)2 . The distance between the oxygen atoms of this hydrate + is 4.7 Angstrom. In more concentrated solutions, it is believed to form Na(H20)6 with the water molecules arranged at the vertices of an octahedron. The distance between the water molecules, that can act as AH,B coordinate structures, are nominally 3.3 Angstrom

177Avenet, P. & Lindemann, B. (1989) Perspectives of Taste Reception J Membrane Biol vol 112, pp 1-8

178Quist, A. & Marshall, W. (1968) The independence of isothermal equilibria in electrolyte solutions on changes in dielectric constant J Phys Chem vol 72(6), pp 1536-1544

179Quist, A. & Marshall, W. (1968) Electrical conductances of aqueous sodium chloride solutions from 0 to 800" and at pressures to 4000 bars J Phys Chem vol 72(2), pp 684-703

180Zumdahl, S. (1993) Chemistry, 3rd Ed. Lexington, MA: D. C. Heath page 329

181Marshall, S. (1956) Sodium: its manufacture, properties, and uses. NY: Reinhold 126 Neurons & the Nervous System

Figure 8.5.4-6 The sodium ion at hydration levels of 2 and 6. Only one of the three pairs of water molecules are shown on the right. See text.

The unique features of the sodium ion are two. First, the sodium ion is very small, nominally 0.95 Angstrom in diameter. Second, the ion is normally highly hydrated in aqueous solution. The hydration number is usually six (with the 2p shell fully involved), although it has been reported to Signal Generation & Processing 8- 127 be four, two or one in various special cases. With a hydration number of six, the ensemble exhibits a face centered cubic arrangement, with the water molecules located at the vertices of an octahedron. At a hydration number of six, the locations of the water molecules are at a spacing that is ideally arranged for coordination with a sensory receptor such as the oxygen rich PtdIns. The fully hydrated sodium ion exhibits three separate odorophores with a d-value of 3.3 Angstrom. This raises the probability that a given hydrate will be successful in coordinate bonding to a suitable sensory receptor. Figure 8.5.4-7 presents a potential hydrated sodium “dimer” similar to that shown for the carboxyl group. The dimensions are taken from Hille182 who credits them to Pauling. This figure shows the hydrogens of the water molecules are out of plane relative to the sodium and oxygen molecules. This suggests that one of the hydrogens can be moved into the plane of the molecules and form a hydrogen bond with a pair of electrons of the oxygen of the other hydrate.

Figure 8.5.4-7 A potential hydrated sodium “dimer.” Each of the units displayed consist of one sodium ion fully hydrated with six water (H2O)molecules, only two of which are shown. The short solid lines represent the O–H bonds. The heavy dashed lines represent the H- -O (hydrogen) bonds. The slight dog-leg between the O–H bonds and the H- -O bonds are typically ignored. The distance between the hydrogen bonds is taken to be 3.3 Angstrom.

As in the case of the carboxyl dimer, this figure suggests the spacing between the AH,B group of the sodium sensitive sensory receptor should optimally have a spacing near 3.3 Angstrom.

182Hille, B. (1971) The Hydration of sodium ions crossing the nerve membrane PNAS vol 68(2), pp 280-282 128 Neurons & the Nervous System

This small size is optimally matched to the coordination geometry of PtdIns. This makes the sensitivity of the N-Path receptor maximally sensitive to the sodium ensemble. - - - - [xxx merge with above or below ] No Protein Data Bank (pdb) file has been located, that have been curated by a prestigious source, describing the hydrated form of the sodium ion in solution. However, Megyes et al. have recently studied the hydration of the sodium ion in considerable detail and note several differences from the previously reported data183. The study focused on very caustic and viscous solutions incompatible with the gustatory modality. Their results do point toward further analysis but seem to favor a coordination number of six in cases of interest. They differentiate between the distance between the oxygen atoms of two water molecules in a high molar NaOH solution, using the notation OwC C COw. from two adjacent oxygen atoms associated with water molecules in the hydration sphere of sodium ions using the notation, OC C CO They note on page 5, “The OwC C COw distance was found to be between 2.80 and 2.85 Å. A drastic decrease of water-water coordination number can be observed with increase in concentration, showing that the hydrogen bonded structure of the bulk water is gradually destroyed and it completely disappears at the highest solute concentration. On the other hand, other OC C CO interactions appear with values in the range of 3.2–3.8 Å, arising from the interaction of the oxygen atoms in the hydration sphere of the sodium ions. Unfortunately it is not possible to describe quantitatively these interactions on the basis of x-ray diffraction measurements because the corresponding peak is rather broad and blurred.”

The best available bond lengths (C–O = 1.43 Angstrom and C–C = 1.53 Angstrom) and angles for the distance between pairs of adjacent opposed axial oxygen atoms for muco-inositol as 3.2435 Angstrom. This is a precise value of undefined accuracy as are other values below. No measured data has been located relative to these materials. 8.5.4.4.1 The details/confusion related toPtdIns

The chemistry of PtdIns is poorly documented because of the potential complexity of the family; the essentially unknown glycerides forming the sn-1 and sn-2 elements of triglyceride in a given situation, and the inositol can be drawn from a family of at least 9 isomers. Even the combining of these two moieties can occur under a variety of conditions involving the phosphate portion of the triglyceride. Text books generally show sn-1 and sn-2 as straight unsaturated stearic acid. However, most researchers show various other acids depending on the results of their research. The two glycerides are seldom straight. The conformation of these molecules also depends on the environment in which they are formed. The electrical conductivity of the glyceride chains has not been studied in any detail.

Parthasarathy & Eisenberg reviewed the subject in 1991, noting a major change in nomenclature in 1968. They also make confusing statements about the isomers of inositol. “Myo-inositol is the only stereoisomer to occur in phospholipids (pg 6)” and “This finding suggests that myo-inositol is not the only stereoisomer found in inositol phospholipids (pg 14).” Their figure 6 requires careful study because the hydroxyl groups are omitted from the chair conformation diagrams they present. A new set of rules was being prepared during preparation of their published article184. These rules (labeled recommendations) surfaced several critically important facts. To this day, • the recommendations only apply specifically to myo-inositol as indicated by their title, • the recommendations are provided with loopholes in order to serve different communities and • the rules applicable to a specific moiety change when it is associated with a second moiety. Several of these situations are explicitly stated in the recommendation.

183Megyes, T. Bálint, S. Grósz T. et al. (2008) The structure of aqueous sodium hydroxide solutions: a combined solution x-ray diffraction and simulation study J Chem Phys vol 128 044501

184NC-IUB (Moss, G. ed.) (1988) Numbering of atoms in myo-inositol http://www.chem.qmul.ac.uk/iupac/cyclitol/myo.html Signal Generation & Processing 8- 129 A notable fact is that the computational chemistry community has dispensed with using the numbering rules of formal chemistry as defined by the IUPAC/IUBMB when developing .mol and .pdb files. They have begun numbering the atoms sequentially according to their own arbitrary scheme. See Section 8.6.10.4. In the case of standalone muco-inositol, they have chosen an oxygen as atom number 1 and assigned #2 to its associated carbon and #3 to its associated hydrogen, then proceeded clockwise to the next carbon which becomes atom #4...

[xxx probably condense to one sentence in favor of the following material based on Molecular Arts ] [xxx edit to show the axial-trans– muco-inositol ] Generic inositol, or cyclohexane-1,2,3,4,5,6-hexol is a chemical compound with formula C6H12O6 or (-CHOH-)6, a sixfold alcohol (polyol) of cyclohexane with all of the hydroxyl groups in the equatorial plane of the hexane ring (± 19 degrees). Myo-inositol has been the most studied because of its occurrence in muscle tissue. Of most importance in this work is muco-inositol. The study of myo-inositol has resulted in a specific effort to standardize its terminology by the IUPAC/IUBMB, also known as the NC- IUB, cited above. As of the 1988 report, several options were still allowed to support different research purposes. The committee has generally avoided using Fischer Diagrams in their most recent work. It has stressed the need to use at least the Haworth Diagram and more satisfactorily, the chair or wire-frame presentation. With the more recent evolution of the computer-aided 3D representations based on .mol, .mol2 and .pdb files, they will undoubtedly be recommended by the committee in the future.

In its most stable conformational geometry, the myo-inositol isomer assumes the chair conformation, which puts the maximum number of hydroxyls to the equatorial position, where they are farthest apart from each other. In this conformation the natural myo isomer has a structure in which five of the six hydroxyls (the 1st, 3rd, 4th, 5th, and 6th are equatorial, whereas the 2nd hydroxyl group is axial. Figure 8.5.4-8 shows this configuration combined with a partial view of hydrated sodium. The chair diagram is described as muco-inositol 3-phosphate by Cosgrove in 1980185. However, the 1988 recommendations of the NC-IUBMB would assign the name muco-inositol phosphate based on their recommended renumbering after phosphorylation. The carbon associated with the phosphorous atom is assigned #1. Cosgrove spent less than two pages on muco-inositol in his book. The carbon- carbon bond lengths are 1.53 Angstrom. The carbon-oxygen bond lengths are 1.43 Angstrom. The net (out-of-plane) bond length between the OH-3 and OH-4 have been estimated to be about 3.3 Angstrom without considering the puckering of the ring. It is described as muco inositol_16736990 in the JSmol files of the RSC. However, the JSmol protocols are under review (see above & Section 8.5.1.1.2) No experimental value has been Figure 8.5.4-8 Muco-inositol phosphate and a fully located. The current best estimate based hydrated sodium. The muco-inositol is combined on the JSmol file and the DS 4.1 visualizer of with the phosphate portion of a phosphatidyl computational chemistry is 3.648 Angstrom. moiety and acting as the Na-channel receptor through dual coordinate bonds with a dimension, The figure is approximately to scale but d, of nominally 3.3 Angstrom. See text.

185Cosgrove, D. (1980) Inositol Phosphates: Their Chemistry, Biochemistry and Physiology. NY: Elsevier pg 82 130 Neurons & the Nervous System employs different display techniques to allow representation of the details related to the muco- inositol. In this representation, the two hydrogen bonds are antiparallel, one originating at the sodium ensemble and the other at the hydroxyl group. In 2012, the Molecular Arts software package prepared an alternate representation of inositol without identifying the specific conformation. 8.5.4.4.2 The details/confusion related to muco-inositol

The name inositol comes from the Greek word for “substance isolated from muscle.” Whereas myo-inositol has a long history in the food field, the role of the other stereo-isomers is largely unknown. The prefix myo is clearly redundant but found useful to help organize the nine stereo- isomers of inositol. The prefix muco- is suggestive of the presence of this variant in the mucus of the nose and potentially the saliva of the mouth. Muco-inositol has been recognized since at least the 1970's due to its role primarily relating to the external lemma of neurons.

The nomenclature of the very large number of isomers in the inositol family is confused, partly due to a major change in 1968. Cosgrove described the history of the inositols up to 1980 using a now archaic nomenclature based on the 1988 recommendations186. Citing Brownstein, he did point out the minimum energy required to convert the generic chemical between chair conformations via a boat conformation (pages 41 & 82). Parthasarathy & Eisenberg have described the subsequent history and nomenclature up to 1991 and describe it as “in statu nascendi”187. Even while they were writing their paper, a subsequent change (in which they participated) was being negotiated in 1988. As a result, any representation of inositol in the literature, such as that in the special issue of the Biochemical Journal of 1986, must be questioned and placed in historical context188.

Glusker et al. described six different conformations for derivatives of cyclohexane and more complex molecules in 1994 (top)along with a more definitive set of labels for the bonds extending away from the basic ring, based on conventional drawing practice) (bottom) in Figure 8.5.4-9 .

186D. (1980) Inositol Phosphates: Their Chemistry, Biochemistry and Physiology. NY: Elsevier Chap. 1 & 2

187Parthasarathy, R. & Eisenberg, F. (1991) Bio-, stereochemistry, and nomenclature In Reitz, A. ed. Inositol Phosphates and Derivatives. Washington, DC: American Chemical Society

188Special Issue (1986) The phosphates of phosphatidylinositol in blood Biochem J vol 238(2) Signal Generation & Processing 8- 131

Figure 8.5.4-9 Representations of a six-member ring from Glusker et al., 1994. Top; conformations of six-member rings. Bottom; dispositions of bonds in cyclohexene, a six-membered ring with one double bond. From Glusker et al., 1994.

Majumder & Biswas, writing in 2006, have addressed the variation in the rules over the years in detail189. Figure 8.5.4-10 reproduces their interpretation of the 1989 IUB rules applied to the conformations of the inositols.

Wikipedia has presented a pair of representations, a modified Fischer and a chair that do not appear to agree with each other even though the Merck Index is cited. Later on the page, a set of isomers are presented that do not appear to agree with Majumder & Biswas. The citation is to a 2006 thesis from the Imperial College London. A title but no author or more detailed source is provided. The thesis was neither published or archived. The stereochemistry associated with coordinate chemistry requires detailed knowledge of the conformation of the requisite chemicals. Discussions of gustation based on Haworth, Mills, zig-zag or Fischer projections190 can only be used where the underlying conformation has already been described. Cosgrove has provided both Haworth and chair diagrams (also known as Johnson/Tate diagrams) of the inositols that can be used to compare their utility (pages 4-6).

189Majumder, A. & Biswas, B. (2006) Biology of inositols and phosphoinositides. NY: Springer Chap. 1

190Lindehorst, T. (2003) Essentials of Carbohydrate Chemistry and Biochemistry. NY: Wiley pg 9 132 Neurons & the Nervous System

The Haworth diagrams do not indicate the equatorial character of the hydroxyls at all and the early chair diagrams indicate the difference between the adjacent equatorial hydroxyls poorly.

Figure 8.5.4-10 Stereo-isomers of inositol. The major interest of this work is in muco-inositol and the enantomiers, D-chiro- and L-chiro-inositol. The latter probably exist as a 50/50 mixture at body temperature. Muco-inositol potentially offers two receptor sites meeting the d-value criteria for a hydrated sodium sensory receptor. From Majumder & Biswas, 2006.

The muco-inositol of Majumder & Biswas exhibits a clear symmetry about the hydroxyl up group between the two down hydroxyl groups and the opposite side of the molecule. In Figure 8.5.4- 11, Shaughnessy has documented two conformations of muco-inositol in the answers to a final exam at the University of Alabama. His two forms are consistent with that of Majumder & Biswas. These presentations do not provide a numbering system for muco-inositol. To avoid introducing 4 additional numbers to this discussion, muco-inositol will be described using the 1C notation with the equatorial hydroxyl at position 1 being assumed to esterify with the phospholipid to form the gustatory sensory receptor or GR. The axial oxygen at O-4 can participate in a dual coordinate bond with a gustaphore with either the adjacent axial-trans- O-3 or O-5. Each of these pairs represents a natraphore with a d = 3.3 Angstrom. Whichever oxygen participates in the phosphorylation, the carbon associated with that oxygen becomes C1 based on the current IUPAC/IUBMB nomenclature cited above. Signal Generation & Processing 8- 133

Figure 8.5.4-11 Chair conformations of muco-inositol. Left; two axial down and one axial up providing two potential natrophores. The three equatorial hydroxyl groups provide two pairs of glycophores. Right; same molecule observed from a different position. From Shaughnessy, 2011.

Brownstein studied the confirmation of the inositols using low frequency nuclear magnetic resonance191. The data is limited. He noted the inter-convertibility of the two chair conformations with the axial groups becoming equatorial and vice versa.

4 Figure 8.5.4-12 shows the muco-inositol_16736990 clearly, except for the chair aspects of the 1C conformation. For discussion, the carbon at the bottom of the figure is defined as C-1. The upper axial oxygen, O-4, can participate in a dual coordinate bond with either or both of the axial-trans- hydroxyl groups at positions 3 or 5 (hidden by the ring carbons). The molecule exhibits a mirror symmetry about the vertical axis in this representation.

Dowd, French & Reilly have given energies for a variety of inositol conformations and noted in a scatter diagram the distance between ab initio calculations and

Figure 8.5.4-12 Muco-inositol conformation. The suggested numbering proceeds clockwise from the bottom carbon in this figure. Any of the three equatorial hydroxyl groups at the bottom of the figure could have esterified with the phospholipids of the sensory neuron lemma.

191Brownstein, S. (1959) Shifts in Nuclear Magnetic Resonance Absorption Due to Steric Effects. II. Polysubstituted Cyclohexanes J Am Chem Soc vol 81(7), pp 1606–1608 134 Neurons & the Nervous System measured values for inositol as of 1996192. Simperler et al193. have provided extensive data on the melting point and dimer formations of the isomers of inositol. The report includes a dimer of muco-inositol defined by a double H-bond of the antiparallel type between one equatorial and one axial hydroxyl group of each dimer. Muco-inositol only forms one dimer of this type. This dimer configuration is not used in the gustatory modality. Due to its symmetry, muco-inositol is optically inactive. Its melting point is 290 Celsius. Simperler et al. note the fact that muco-inositol can form ring crystalline motifs where each OH is involved in two distinct hydrogen bond links. This suggests each phospholipid sensory receptor of muco-inositol can form an axial-trans- dual coordinate bond with two gustaphores at once by having the middle axial-trans- hydroxyl participate into two pairings (See Section xxx). 8.5.4.4.3 The gustaphores of the inositol ion

The inositol ion (inosinate–) contains considerable oxygen. As a result, it exhibits two calculated glycophores with d-value near 2.82 Angstrom and two natrophores with d-value near 3.66 Angstrom. The d=3.66 natrophores are likely to form a dimer with the sodium channel receptor. The d=2.82 glycophores may be less optimum and thus less effective in stimulating the G–channel receptor.

The fact that some of the inosinates are perceived as salty, in the absence of any sodium or inorganic salts, provides strong support for the hypothesis of this work, that the GR 3 channel sensory receptor is PtdIns. PtdIns can easily form a dimer with these inosinates. Additional confirmation of the hypothesis are presented in Section 8.5.15. 8.5.4.4.4 The perception of sodium as sweet at low concentrations

[xxx see also comments re Kcl and LiCl in Dzendolet67.pdf ] Shallenberger & Acree have noted the perception of sweetness associated with salt in solution at low concentrations. The trait appears common among the low atomic weight alkali metal salts, Li, Na & K.

The literature related to the hydration of the alkali ions at low concentrations is extremely thin. The terminology is also contradictory due to age. An anhydrous compound, HNaO, is described by the CAS Register as sodium hydrate. In solution however, the sodium ion is known to exhibit a variety of states of hydration. It appears sodium can support a coordinate number of 1, 2, 4 or 6 depending on the circumstances, and particularly the concentration of the ion in water.

Shallenberger & Acree have suggested that Na[4], with a tetrahedral form, can create a coordinate bond with the G-Path sensory receptors because of the spacing between the pairs of hydrates, one acting as an electron acceptor and the other as an electron donor. Beginning in the same era, Marshall has indicated the ion is generally fully hydrated, Na[VI]+1, at low (gustatory) concentrations194. See Section 8.5.3XXX.4

The hydration of sodium at low concentrations explains the perception of a sweet taste to sodium salts. The hydration level would also suggest that the sodium or N-Path sensory channel is also a coordination chemistry based sensory channel that senses the sodium ion (without sensing the associated negative ion).

192Dowd, M. French, A. & Reilly, P. (1996) Analysis of Inositol Ring Puckering Austral J Chem vol 49(3), pp 327-335

193Simperler, A. Watt, S. Bonnet, P.. Jones, W. & Motherwell, W. (2006) Correlation of melting points of inositols with hydrogen bonding patterns Cryst Eng Comm vol 8, pp 589–600

194Marshall, W. (2008) Aqueous electrolyte ionization over extreme ranges as simple fundamental relation with density and believed universal; sodium chloride ionization from 0o to 1000oC and to 1000 mPa (10000 atm.) Nature Precedings : hdl:10101/npre.2008.2476.1 Signal Generation & Processing 8- 135 - - - -

8.5.4.5 Operation of the “bitter” gustatory sensory neuron

Bitter tastes are often associated with harmful substances like alkaloids and glucosides, but many other groups elicit a bitter sensation. The operation of the bitter sensory channel is not well documented. Belitz et al195. provided a theory of bitter taste related to the early work of Shallenberger. They suggest a topological requirement, involving a hydrophobic pocket, is involved in bitter or P-Path channel sensing. The concept appears to correlate well with experience involving a wide range of amides, amines and amino acids. However, Shallenberger & Acree noted in a larger context in 1971, “Only limited correlations between bitter taste and molecular structure can be found.” In 2000, Eggers, Acree & Shallenberger took a different position after reviewing the state of the art and noted, “These results suggest that both sweet and bitter reception share the same transduction components and that the non-sugar sweet receptor system is related to the bitter receptor system if it is not in fact the same.”

Kumazawa provided a discussion of the varied chemistry of bitter stimulants and potential bitter channel sensor operation but was unable to identify any receptor or provide any description of the pertinent sensory neuron196. [xxx merge with above and below ]

8.6.10.3 The detailed nomenclature of the picric channel stimulants

8.6.10.3.1 Review of the historical database

Data showing the relative perception of bitterness in humans is rare. Figure 8.5.4-13 shows a table from Van der Heijden from 1993. Note that nearly all of the references are to one source, the 1985 paper by Belitz & Wieser in the first issue of an obscure journal197. The other references are primarily in German.

195Belitz, H. Chen, W. Jugel, H. et al. (1981) Structural requirements for sweet and bitter taste In Schreier, P. ed. Flavour ‘81. Berlin: Walter de Gruyter pp 741-755

196Kumazawa, T. Nomura, T. & Kurihara, K. (1988) Liposomes as model for taste cells: receptor sites for bitter substances including n-c=s substances and mechanism of membrane potential changes Biochem vol 27, pp 1239

197Belitz, H. & Wieser, H. (1985) Bitter compounds: occurrence and structure-activity relationships Food Reviews Internat vol 1(1), pp271+ 136 Neurons & the Nervous System

There are a number of questions about this tabulation. As noted in this work, listing molecules by their functional group or similar chemical categorization is not meaningful in gustation. The presence of a specific ligand, as defined in this work, is the proper criteria. The presence of that ligand in a chemical does not make it a good representative of the chemical category listed.

Figure 8.5.4-13 Most potent of various classes of bitter compounds. The boxed values at upper right appear to be a drafting error. The most bitter Picric acid should be shown following the triterpenes (some being very bitter) at lower left. See text. Modified from Van der Heijden, 1993.

This fact is illustrated by the assertion by Van der Heijden page 81) relating to aspartame, “The LL optical isomer is sweet, and the DD, DL and LD isomers are not.” The title of the figure suggests that picric acid is one of the most potent of bitter stimulants. Yet the label potency is accompanied by a foot-note saying the column represents threshold values in nanomoles per liter unless otherwise noted. This statement would suggest picric acid has one of the highest thresholds in the list. The drastic difference in threshold between the ureas and above and the “heated protein extracts” below, or between the amides and the plant constituents below, Signal Generation & Processing 8- 137 would suggest the likelihood of a separation between the bitter and the super bitter compounds. Acree addressed what he described as the “flavor activity” in the same work as Van der Heijden. Acree noted,

“It is important not to confuse the definition of flavor activity with the detailed relationship between stimulus and response describe by psychophysical or dose-response functions for flavor. As is shown below, activity is usually defined at specific values of these functions. Typical flavor activity measured in flavor units is, flavor activity = (1/threshold) x concentration” His text is careful to delineate between flavor activity and the psychological response labeled flavor response. He goes on, “The concentration factor is the amount of compound detected in or added to a food. The inverse threshold factor is somewhat complicated conceptually and is sometimes difficult to measure. In simple terms, the threshold is the concentration of a compound in a particular food system below which the food system has no flavor. Thus compounds with high inverse threshold factors are flavorful even at low concentrations.”

The above formula can be rewritten as flavor activity = concentration/threshold. In this formulation, the threshold corresponds to the noise floor or other minimum level required for the system to respond, and the concentration describes the magnitude of the signal relative to that threshold. This is the context in which Acree discusses the function in his figure 2 of 1993 addressed below. In this context, the flavor activity is the intensity of the stimulus as measured at the input to the adaptation amplifier of the sensory neuron. The flavor response is the perceived stimulus level at stage 5, information extraction, of the neural system. The threshold level corresponds to the cutoff potential of the input amplifier (adaptation amplifier or 1st Activa) within the sensory neuron itself.

If the Acree formulation is appropriate, there are three problems with the above Van der Heijden figure; the column labeled potency is truly the threshold value for most of the values in the column and the values given for the phenols should be at the bottom of the list, as having a nominal threshold similar or below that of the triterpenes (and consequently requiring only a minimal concentration to elicit a psychological response. Finally, sucrose octaacetate should not be considered a typical sugar; it contains a powerful picrophore not found among the simple sugars. As a result, the most potent gustaphores (picric acid, the triterpenes, amarogentin and artabsin) are at the bottom of the list, in agreement with most of the literature.

In the case of the phenols, their threshold value is not near 1.0 nanomoles per liter. It appears Van der Heijden encountered an error in taking the reciprocal of a very small number. Acree addressed the dynamic range of the gustatory modality in 1993. His figure 2 suggest an instantaneous dynamic range with regard to a specific stimulant of about 100:1, which is similar to the more easily determined 200:1 in the visual and hearing modalities. His figure 1 suggests a overall dynamic range, resulting from adaptation, on the order of six orders of magnitude or more, also compatible with the measured ranges of vision and hearing.

8.6.10.3.2 Summarizing the picrophores of taste

Van der Heijden (page 95) noted the claim of Kubota & Kubo (1969) “that bitterness could be explained by using the AH-B concept, which had also been used for elucidating sweetness. However the AH-B distance for bitterness is 0.15 nm (1.5 Angstrom) [xxx check ]compared to 0.30 nm for sweetness.” He appears to be confusing acidity for bitterness in this case. The first property to note in this paragraph is that picric acid (CAS 88-89-1) is not an organic (carboxylic) acid. The name is the common one for a NO2 substituted phenol, 2,4,6-trinitrophenol of the formula, C6H2(NO2)3OH. It is frequently shown as an inorganic anion due to the loss of a 138 Neurons & the Nervous System hydrogen. Figure 8.5.4-14 illustrates the complete picric acid molecule. It is proposed that this picrophore has a d-value optimized for stimulating the picric channel sensory neuron that employs PtdAsn as its receptor. Taking picric acid as the premier member of the bitter tasting chemical family, there are two potential ligands of interest. They appear simultaneously and in multiple instances. The first ligand of interest is O=N-C-=C-=C- N=O and it is planar. Note the backbone includes two nitrogen atoms instead of an all carbon backbone. The notation C-=C is used to indicate a resonant linkage between the two carbons due to the parent ring structure. This gustaphore occurs three times in this (gustant) molecule In each case, the best available d-value is 4.746 Angstrom.

The second ligand of interest is the N-C-=C- =C-N sequence alone. The nitrogen atoms as configured in picric acid continue to have an unshared pair of electrons capable of supporting a dual antiparallel coordinate bond but would need to Figure 8.5.4-14 The premier gustant of the bitter or acquire a hydrogen through P-channel of gustation. Distances in Angstrom. rearrangement of from the solvent. The d- See text. From a PDB file on the Boston University value for these pairs remains 4.746 web site, 2012. Angstrom.

There is a potential third ligand in interest. It would originate at an nitrogen and terminate at an oxygen, or vice versa, O=N-C-=C-=C-N . The d-value of these ligands is only slightly higher at 4.948 Angstrom.

Each of these potential ligands occurs multiple times in each molecule providing a rich source of potential gustaphores and resulting in a very high effective concentration for this molecule..

The symmetry of the molecule suggests it has a dipole potential of near zero for all of the odorophores discussed. The literature shows a wide variety of ionic structures for picric acid. While of no concern with regard to coordinate bonding, they could affect the measurement of the dipole potential of the molecule. However, it is never shown as an organic acid. Figure 8.5.4-15 shows it in a triply ionized state198 but without the removal of the hydrogen from the hydroxyl group at top center. The noteworthy fact is the d-value of 4.746 Angstrom does not change for the noted oxygen pairs. It is likely that different dipole potentials will result from the various ionic forms.

198http://www.chemspider.com/Chemical-Structure.6688.html Signal Generation & Processing 8- 139 It is noteworthy that ChemSpider differentiates between the average mass of 229.103897 Da and the monoisotopic mass of: 228.997101 Da typically reported. Under a heading, “Names and Synonyms,” it also offers a validation citation, along with a variety of synonyms. The site makes a significant effort to provide traceability with its property assertions.

8.6.10.3.3 Amarogentin, artabsin and quinine

Amarogentin (CAS 21018-84-8) is a complex molecule of 586 Dalton that the Merck index suggests is one of the bitterest of compounds. It hydrolyzes into three parts in water, a sugar with a glucophore, O24–O26, with d-value of 2.754 Angstrom and two multiple ring structures. The one ring structure exhibits a picrophore, O28–O34, with a d-value of 4.648 Angstrom. Figure 8.5.4-15 Ionic forms of picric acid from the Thus amarogentin contains as a minimum literature. Neither the formula, or the coordinate two different types of gustaphore. Its bond capability, of the molecule changes due to psychophysical evaluation is difficult the ionization shown. See text. Example from because of this feature. ChemSpider, 2012.

Quinine is a complex molecule with a two phenol ring structure that supports a picrophore between two oxygen atoms with a d-value of 4.801 Angstrom. The molecule also exhibits two phenol rings that can also act as an electron-pair donor. These features offer additional sources of picrophores in conjunction with the other oxygen and nitrogen atoms present.

Artabsin (HMDB36641) has only three oxygen and zero nitrogen or other orbital atoms. The most relevant pairs in this molecule have d-values of 2.420 and 5.829 Angstrom. It does exhibit one penta hetrocyclic ring, one penta cyclic ring and one 7-point ring. Further analysis will be required to uncover the source of its picrophore(s). 140 Neurons & the Nervous System

Quinine provides a slightly more complex gustaphore relying on an out-of-plane condition. The ligand can be described as O-C-=C-=C-=C-O. The backbone is all carbon but the geometry is slightly more convoluted. The upper oxygen is in the plane of the two rings. However, the lower oxygen is out of the plane resulting in a longer distance between the oxygen atoms, a d-value of 4.801 Angstrom, as shown in Figure 8.5.4-16. The complex moiety on the right plays no significant role in the gustaphore selection process, although it may play a role in the dipole potential of the molecule. The slightly different d-value in the third significant digit may indicate the width of the sensory receptor acceptance range, or it may indicate a difference in optimization procedure used to optimize the molecule structure before defining this d-value. Artabsin (HMDB36641) has been described as a potent bitter stimulant. It has only three oxygen and zero nitrogen or other orbital atoms. The most relevant pairs in this molecule have d-values of 2.420 and 5.829 Angstrom. It does exhibit one penta hetrocyclic ring, one penta cyclic ring and one 7-point ring. Further analysis will be required to uncover the source of its picrophore(s). 8.6.10.3.4 The triterpenes Figure 8.5.4-16 Quinine as a stimulant of the Triterpenes are terpenes consisting of six P–channel of gustation. Distance is in Angstrom. isoprene units and have the molecular See text. A PDB file provided by David Woodcock formula C30H48. The simple terpene of consists of two isoprene units. Terpenes Okanaga Univ. Coll., 2012. typically exhibit five complete or partial ring structures. They are a large and varied class of hydrocarbons199, produced primarily by a wide variety of plants, particularly the gums and resins of the conifers, though also by some insects such as termites or swallowtail butterflies, which emit terpenes from their osmeterium. Triterpenes are precursors to steroids in both plants and animals. They generally consist of more than three ring structures. The simplest triterpenes exhibit little oxygen and no other orbitals involved in taste. Many triterpenes containing more oxygen are known and they offer many opportunities to exhibit picrophores.

199https://www.google.com/search?q=triterpenes&hl=en&tbo=d&rls=com.microsoft:en-us:IE-SearchBox&r lz=1I7GGLD_en&tbm=isch&tbs=simg:CAESEglApfbYCUQYkiE5VFo1_1Dz7jg&dur=4297&biw=995& bih=552&sei=WIqiUMyLOumY2wWP7ICIBA#q=triterpenes&hl=en&tbo=d&rls=com.microsoft:en-us:IE- SearchBox&rlz=1I7GGLD_en&tbs=simg:CAESEglApfbYCUQYkiE5VFo1_1Dz7jg&tbm=isch&bav=on.2 ,or.r_gc.r_pw.r_qf.&fp=d1a4c9382e039dc&bpcl=38093640&biw=995&bih=552 Signal Generation & Processing 8- 141

Figure 8.5.4-17 shows a typical triterpene associated with a sugar as well as one containing many more oxygen atoms and hydroxyl groups. The upper triterpene after hydrolization is tasteless while the lower triterpene is very likely to taste bitter due to its potential to form multiple gustaphores capable of forming a dual antiparallel coordinate bond with a d-value of near 4.746 Angstrom.

8.5.4.5.1 Review of diverse bitter gustants and gustaphores

The list of compounds historically presumed to be bitter tasting is quite long and involves many chemical structures as shown by Spielman et al200. in Figure 8.5.4-18. Van der Heijden has given a similar list of the major bitter stimulants that includes their relative potency. One of the goals of this work is to more clearly identify the role of these compounds. To this end, it is noted that the role of the astringents formed by the salts of the alkali earths are actually nocent chemicals and not bitter gustaphores in the context of the gustatory modality (Section 8.5.1.5.4). However, time will not allow the identification of all of these chemical groups. Many of the listed chemicals are now considered, and can be shown, to be stimulants containing multiple gustaphores exciting various neural pathways.

Van der Heijden listed picric acid as a reference gustant, but noted it was more than six orders of magnitude less bitter than the triterpenes and two plant constituents, amarogentin and artabsin. Van der Heijden has also provided chemical structures for each of these materials. Unfortunately, these structures are two- Figure 8.5.4-17 A typical triterpene linked to a dimensional Fischer (stick) diagrams. These sugar at lower left. The molecule exhibits little diagrams illustrate the historic difficulty of oxygen but the capability of many points of finding a common structural feature to oxygen addition. Below; a more highly account for the bitter sensation. oxygenated triterpene. See text. When noting large ranges in the perceived bitterness of materials, it becomes necessary to consider the possibility of a super-bitter category analogous to the super-sweet category in the glycophores of the sugars. This is done in Section 8.5.4.6.

A 3d representation of amarogentin, CAS #21018-84-8, clearly shows it exhibits a glycophore with d-value = 2.754 Angstrom (O24–C–C–O26) as well as a picrophore with d-value = 4.648 Angstrom (O28–C–C–O34). Its complexity suggests it is also capable of participating in a super-bitter perception such as that developed in the next major section. Caffeine is a simpler molecule than amarogentin, similar to the complexity of some of the man-

200Spielman, A. Huque, T. Whitney, G. & Brand, J. (1992) The diversity of bitter taste signal transduction mechanisms In Corey, D. & Roper, S. eds. Sensory Transduction. Woods Hole, MA: Marine Biological Laboratory Chapter 20 . 142 Neurons & the Nervous System made super-sweet molecules. It exhibits a d-value = 4.7 Angstrom (only one percent below the nominal 4.746 Angstrom of the proposed P-path GR). The AH,B path within caffeine is O4=C–N –C–N8. Further review and comparison of this and other presumed super-bitter molecules will be required to determine any presumed AH,B,X relationship that might affect the P-path GR significantly.

Figure 8.5.4-18 The diversity of historically bitter compounds EDIT or MERGE with van der Heijden, 1993 of figure 8.5.4-27 by adding threshold column a/o d-value column. Right column; proposed alternate designations. See text. Significantly modified from Spielman et al., 1991.

Spielman et al., writing in a Society of General Physiology symposium record, speculate on whether a single sensory neuron could possibly accommodate the detection of this broad array of chemical structures, or whether multiple sensory neurons are required. They offer only working hypotheses on the character of the sensory neurons based conceptually on the chemical theory of the neuron. Their assumptions are; 1. there is no one mechanism that would apply to all bitter stimuli, 2. there must be several signal transduction processes attuned to the broader chemical categories of bitter compounds, 3. one group of bitter compound can be detected by several different mechanisms, and 4. there are still bitter compounds to be discovered or synthesized. Van der Heijden reviewed the structural formulas for a variety of his bitter stimulants (page 95- 103) and concluded they did not meet the criteria for a suitable AH,B coordination bond because the structures did not meet his criteria for bitterness of 0.15 nm (1.5 Angstrom) [xxx chk ]. His model of a sweet-bitter has not drawn favor in the taste research community. As shown in this work, the AH,B spacing criteria to coordinate with the bitter sensory receptor is 0.42 nm (4.2 Angstrom) as developed below. [xxx 4.7xxx ] The ability of the AH,B coordination structure to explain the operation of the G-Path channel suggests a similar structure might be important in the P-Path channel. To date, only limited Signal Generation & Processing 8- 143 insight has been obtained into the operation of the P-Path receptor mechanism. Shallenberger & Acree have noted, “limited correlations between bitter taste and molecular structure exist because this taste sensation is not one that is enthusiastically evaluated.” More recently (2003), Drewnowski has noted201, “the biology of bitter taste perception is poorly understood.” and “The study of human taste genetics is largely the study of bitter taste.” The range of chemical structures stimulating the P-Path channel is wide. It includes many, but not all, alkaloids, tannins, methylxanthines, peptides, isoprenoids and many non-sodium salts (Section 8.xxx). Care must be taken to use the complete chemical name of a bitter stimulant, the common names used in pharmacology and the food technologies are frequently less than specific. - - - - A major problem for previous investigators has been their lack of any guidance on what to look for. With the exception of the comment about three electronegative oxygen atoms lying in a single plane by Shallenberger & Acree, the typical investigator has been looking at structures involving less distance between the specific atoms. The rectilinear sensation space of this work shows a quinine best receptor must involve A B spacings greater than 3.7 Angstrom to be independent of the other sensory channels.

Denatonium, Figure 8.5.4-19, usually available as denatonium benzoate and as denatonium saccharide, is the most bitter chemical compound known according to Wikipedia and Aversion Technologies, Inc... Threshold for the benzoate is approx. 0.05 ppm. Dilutions of as little as 10 ppm are unbearably bitter to most humans. 30 ppm is claimed to be undrinkable. Denatonium is a quaternary ammonium cation. It is a compound of a salt with an inert anion like benzoate or saccharide. The structure of denatonium benzoate, C28H34N2O3 and molecular weight of 446.58, is related to the local anesthetic lidocaine, differing only by the addition of a benzyl group to the amino nitrogen. Denatonium saccharide (meaning the salt of saccharin, a benzoic sulfilimine and not of a simple sugar) has the formula, C28H33N3O4S & molecular weight of 507.46..

Figure 8.5.4-19 Denatonium benzoate & saccharide, the most bitter compounds ADD & EDIT known. Labels don’t appear to match figure and difference between two upper forms is not justified. Top; benzoic acid. Bottom; denatonium. Et; ethyl ligand. Mt; methyl ligand. PH; phenyl ligand.

Denatonium is frequently used as an alcohol denaturant, along with brucine and quassin. These

201Drewnowski, A. (2003) Genetics of human taste perception In Doty, R. ed. (2003) Handbook of Olfaction and Gustation, 2nd revised and expanded edition. NY: Marcel Dekker Chap 40 144 Neurons & the Nervous System chemicals are frequently shown as in the figure. Even the Merck Index shows each of them as a pair of ions with the positive charge on the nitrogen of the denatonium. The applicable Jmol files also show two ions. No information on the solubility or ionization ability of these chemicals is given in the Index. Their actual role as picrophores can not be determined without showing their actual structure in three dimensions. They obviously contain a large number of potential orbitals and aromatic rings that could provide the appropriate d-values for stimulating the bitter channel. The complexity of the denatonium salts suggests they could stimulate almost any receptor of the gustatory modality depending on the degree of ionization of these salts. Figure 8.5.4-20 shows a 3D representation of denatonium benzoate. Based on the proposed hypothesis, the oxygen of the denatonium ion would not result in a taste perception when associated with either of the two rings (d- values of approx. 3.72 & 5.96 Angstrom. The upper nitrogen and ring exhibit a d- value of 2.785 Angstrom and could be perceived as sweet. The lower nitrogen and ring exhibit a d-value of approximately 3.69 Angstrom and could be perceived as salty. The benzoate ion is a simple organic acid and would be expected to contribute an acid taste to the solution (d-value of about 2.268 Angstrom). These values support the observation by others in Figure 8.5.4-20 Denatonium benzoate from Jmol. Section 8.5.1. The ability of the benzoate The view shows the fact the two rings of to also taste salty is in question because of denatonium are not in the same plane or parallel the accuracy of the oxygen to ring d- to the aliphatic chain. See text. values based on the Jmol of December 2012. It shows the distances between the two oxygen orbitals and the centroid of the ring as equal, which would require the two oxygen atoms to be in resonance.

Saroli has provided a major table of bitterness thresholds among a large group of denatonium derivatives associated with various halogens202. His brief discussion of the potential AH,B relationships for these molecules is confounded by the unusual dimensions he gives for the distances between the relevant orbitals. His 3.2 Angstrom between the electrophilic ammonium nitrogen atom and the nucleophilic oxygen of the carbonyl group differs markedly from the 1.43 Angstrom found using Discovery Studio 3.5 (DS3.5, 2012)from Accelyrs . However, hand calculations would suggest his value of 3.2 Angstrom is in error. Saroli also uses a different variant of the denatonium structure than CAS 1674-99-3. As noted earlier, the state of 3D computational chemistry remains primitive. Significantly different values are given using the same Jmol file found in the on-line Chemical Book (CAS 1674-99-3) when using the DS3.5 and the Argus Labs 4.0.1 (2004) visualizers due to the different degrees of planarity found. Both visualizers show four bonds to the carbonyl oxygen without any graphic identification or other explanation. It can be assumed the extra bonds, one to each of the nitrogen atoms are coordinate bonds rather than valence bonds. The Jmol vers 13.0.8 (2012) visualizer gives the same planar configuration as the Argus representation.

202Saroli, A. (1985) Interaction of denatonium chloride with the bitter taste receptors Z Lebensm Unters Forsch vol 180, pp 227-229 Signal Generation & Processing 8- 145

[xxx deprecate these paragraphs in favor of Sec 8.4.xxx in complete chap 8 printed version ] A problem at this time is the various visualization programs provide strikingly different representations. Compare the above to Figure 8.5.4-21. The upper 3D representation shows a totally planar structure for denatonium while the lower representation shows a non-planar structure more complex than that in the upper frame and previous figure. Note the four bonds to O13. This representation differs significantly from the Fischer diagram in the center of the figure. DS3.5 describes the bond between C7 and N12 as “aromatic.” It does not describe the bond between O13 and N1. The distances between the atoms of these three denatonium moiety representations vary drastically.

Until the actual structure of the various denatonium compounds can be determined when in solution, the hypothesis of this work can not be tested or extended to include this family of super-bitter picrophores.

Care must also be taken to identify the visualizer used in a specific case. They do not assign the same atom numbers to the same atoms within a compound.

No dependable (and defendable) conclusions can be drawn from these representations. Any attempt to extend the super bitter representation to the AH,B,X relationship is detered by the problems with these visualizers. The only potential picrophores of denatonium chloride identified to date are;

• the centroid of the benzyl ring farthest from the ammonium nitrogen and that nitrogen with a d-value using the Argus Lab’s visualizer of 4.72 Angstrom (6.794 using the DS3.5 visualizer)

•the centroid of the benzyl ring farthest from the ammonium nitrogen and the closer quadratic nitrogen with a d-value = Figure 8.5.4-21 Alternate structures for the 4.337 Angstrom using the DS3.5 visualizer denatonium family given by different visualizer (much shorter using the Argus visualizer). programs. The programs are unable to provide consistent representations of this family, including Only the 4.72 Angstrom d-value meets the the specific CAS 1674-99-3. See text. criteria of this hypothesis well. Denatonium is the first example of the nucleophilic character of a ring structure serving as an orbital in gustation discussed in this work.. - - - - Figure 8.5.4-22 shows the chemical structure of quinine. Quinine hydrochloride is frequently used in the laboratory as the quintessential bitter stimulant. Its complex structure makes it difficult to analyze without a definitive set of requirements. However, based on the theory developed here, the bipartite AH,B coordinate bond with the P-Path sensory receptor is clear as illustrated. The spacing between the two oxygen atoms is estimated at 4.2 Angstrom after accounting for the two-dimensional projection of this three-dimensional molecule. If the actual spacing is not close to this value, an alternate AH,B coordinate bond will be required based on the presence of Nitrogen at several points or the presence of the ring structures. 146 Neurons & the Nervous System

While nitrogen plays a major role in many bitter stimulants, it is not a necessary component of a P-Path stimulus. The tannins (tannic acids) are a complex family203 of (a) so-called condensed tannins and (b) hydrolyzable tannins. The latter are sugar esters with one or more trihydroxybenzenecarboxylic acids as shown in Figure 8.5.4-23 for corilagin (CAS 23094-69-1. The names based on their structure are largely irrelevant in gustation. The proliferation of hydroxyl radicals in these compounds make the potential for an AH,B coordinate bond of the required spacing quite likely. While shown in Fischer diagram form, and represented as planar using the Argus and Jmol visualizers, the molecule is in fact non-planar. To achieve planarity, both involve a forced closure of one outsized bond and very many foreshortened bond lengths. The DS3.5 Figure 8.5.4-22 Quinine, the preferred bitter taste visualizer shows a more rational 3D in the laboratory. The d-value shown was representation of the molecule. In that obtained with DS3.5 visualizer, as opposed to a representation, the molecule exhibits value of 4.2 Angstrom obtained with a Jmol hydroxyl pairs with a variety of d-values. visualizer and a Jmol file from kaist_ac_kr Several (at least six) are in the range (Woodcock). suggestive of stimulating the picric channel receptor, from 4.65 to 4.93 Angstrom.

The Argus visualizer provides a stunning electrostatic surface profile for corilagin. However, it cannot be relied upon due to the foreshortened dimensions also displayed. [xxx corilagin ESP 23094-69-1 ] [xxx corilagin DS3.5_23094-69-1.wpg ]

Lawless and Lee (Figure 6), following McManus et al204., described the potential for two adjacent oxygen atoms separated by two carbons in a resonant 6-member ring to form a dual antiparallel coordinate bond with an unspecified peptide chain. They did not discuss any requirement that the distance between the two oxygen atoms of the tannin match the distance between the oxygen and nitrogen of the peptide chain or any requirement that the lengths of the hydrogen bonds meet nominal requirements.

Lawless & Lee have used the terms astringent as well as bitter in their discussions of the tannins.

Figure 8.5.4-23 Corilagin, a tannic acid & a complex sugar ester, CAS 23094-69-1. The four rings, including the one on the right which is heterocyclic, are found in distinctly separate planes.

203Lawless, H. & Lee, C. (1993) Common chemical sense In Acree, T. & Teranishi, R. eds. Flavor Science. Washington, DC: American Chemical Society Chapter 2

204McManus, J. Davis, K. Lilley, T. & Haslam, E. (1981) The association of proteins with polyphenols J Chem Soc Chem Commun pp 309b-311 Signal Generation & Processing 8- 147

Figure 8.5.4-24 shows the chemical structure of caffeine, another bitter stimulus containing significant nitrogen but little oxygen, and its parent xanthine on the right. Analysis using the Jmol program, shows the d = 4.70 Angstrom between .N9 and O2 in both molecules. Such a value is compatible with binding of these molecules to the picric channel receptor with a nominal d = 4.827 Angstrom. The disparity of 2.7% is consistent with the weakly bitter perception of this chemical by many humans. [xxx check with DS3.5 visualizer ] [xxx alt use 4.746 and a new percentage ]

Quoting Glendinning et al., “Many different classes of compounds elicit bitter taste in humans and inhibit feeding in animals. These compounds include alkaloids, tannins, methyl- xanthines, peptides, isoprenoids, and many non-sodium salts. Because virtually all naturally occurring poisons taste bitter to humans, this taste quality is thought to have evolved as a mechanism for avoiding toxic substances.” Figure 8.5.4-24 Caffeine, 1,3,7-trimethylxanthine [xxx rewrite next two paragraphs consider and its parent xanthine on the right. Both exhibit moving all or part to the thiol section. ]] a d-value of 4.7 between the atoms shown in Breslin may have uncovered the key to the both the DS3.5 and Jmol visualizers, although the operation of the P-Path sensory neuron atom numbering schemes used differ. (Section 8.xxx). This may be an excellent clue to the operation of the P-Path sensory channel since phosphatidic acid is a potential transducer in the type 4 lemma of the P-Path sensory neuron microvilli. If the phosphatidic acid is reacting with the quinine hydrochloride with the precipitation of quinine, the phosphatidic acid may become chlorinated and become negatively charged. This action could mirror the nominal reaction between the phosphatidyl fatty acid of type 4 lemma and quinine hydrochloride, resulting in the precipitation of the quinine, permanent alteration of the phosphatidyl fatty acid and the injection of an electron into the plasma of the microvilli. A comparison of the state of polarization of phosphatidic acid and both PtdCho and PtdSer would appear to be useful. PtdSer is very similar structurally to phosphatidic acid and shows a positive polarization, just the opposite of the more common PtdCho forming a majority of the outer leaf of plasma lemma. Breslin (page 440) has noted the genetics-based work of Kalmus205. Kalmus found a N–C=S bond in a set of anti-thyroid drugs that may be significant. Individual humans exhibit considerable difference in sensitivity to these two compounds. As many as 25% of the population may be taste blind to the bitterness of these materials, suggesting a “spectral blindness” similar to protanopia in vision. Since these taste-blind subjects are not blind to other quinine related stimuli, it has been suggested there may be more than one P-Path receptor types. See Section 8.5.4.7.

205Kalmus, H. (1971) The genetics of taste In Beidler, L. ed. Handbook of Sensory Physiology: Vol. IV, Chemical Senses NY: Springer pp 165-179 148 Neurons & the Nervous System

Shallenberger & Acree have also reported the work of Kubota & Kubo206 suggesting an AH,B relationship with a 1.5 Angstrom spacing might be important in bitter channel transduction. However, they also suggest three electronegative oxygen atoms lying in a single plane may be a characteristic of quinine class stimuli, such as picric acid with three NO2 ligands. Figure 8.5.4-25 shows a pair of molecules that Shallenberger and Acree described as “most intriguing variations in sweet taste between two closely related compounds.” They noted the α– conformation was sweet tasting but the β– conformation of identical chemical structure was tasteless. Based on the double antiparallel coordinate bond (DACB) theory of this work, the difference is quite understandable; it is not the fact that the same groups are present in both cases, it is the fact that the distance between the electron-rich aromatic ring and the hydroxyl group varies so significantly. The sweet tasting α– or anti form has a d-value near 2.74 Angstrom while the β– or syn oxime has a much larger d-value of about 4.8 Angstrom because of the physical location of the hydroxyl group. Contrary to being tasteless, the β– form should stimulate the P-path and be perceived as bitter (possibly at a different concentration level). The relevant situation is shown in the lower half of the figure. It is critically important to discriminate between these two conformation when discussing taste. These two substances are given a wide variety of numbers on various databases. The closest source of consistent numbers given on one page appears to be from the NCBI of NLM of the NIH207. The α– conformation is given the CID of 5371924 (items 27, 28 & 29, with a minor variation in graphic for 28 & 29). The β– conformation is given the CID of 5371961 (items 1-26). It is the distance between the unshared pairs of orbital electrons that is the significant feature of the two conformations. In this case one of the unshared groups is provided by the aromatic ring. The only d-value of interest in this figure is that between the aromatic ring and the most remote oxygen. The ring and the remote oxygen can bind to the sensory receptor of the G-path. The sharing of a pair of electrons from an aromatic ring will be found to be a common and critical feature in the olfactory modality discussed in Section 8.6.

The geometry of the electronic cloud associated with the aromatic ring is poorly defined with relation to its participation in a coordinate bond situation. Initially, it will be assumed the centroid of the cloud is congruent with the centroid of the ring of atoms forming the ring. While the electronic cloud may extend above and below the plane of the ring, it will be assumed the cloud is dumbbell shaped and the two sections of the cloud are cylindrically symmetrical. Hence, the distance between an orbital atom and the geometric center of the ring will be taken as the pertinent distance in any DACB binding relationship. At a more sophisticated level, it may be shown that Figure 8.5.4-25 Two “most intriguing” examples of the pertinent centroid varies with auxiliary anisaldoxime Top row; representation from asymmetric moieties associated with the Shallenberger & Acree with dimensions added. ring just as the dipole moment of the overall Bottom row; 2D representations from a 3D molecule varies with the associated construction using Jmol with additional electron moieties. pairs annotated. See text. Modified and extended from Shallenberger & Acree, 1971.

206Kubota, T. & Kubo, I. (1969) Bitterness and chemical structure Nature vol 223, pp 97-99

207http://www.ncbi.nlm.nih.gov/sites/entrez Signal Generation & Processing 8- 149

Kubota & Kubo examined the diterpenes from a specific species of plants, Isodon, extensively. Using slightly different terminology than Shallenberger & Acree, possibly mis-translating their concept and relying on the Brknsted acid concept of acids and bases, they assert that a proton donor group (DH) and a proton acceptor group (A) must be within 1.5 Angstrom and form an intramolecular bond in order to form a “bitter unit.” They do not address the stimulus/receptor relationship. [xxx re-examine d-values The assertion of a distance between coordinate bonds of 1.5 Angstrom for bitter compounds has not appeared in the recent literature, except in its restatement by Kubo in his 1994 doctoral dissertation208. His determination of 1.5 Angstrom based on X-ray investigations may well refer to the length of the O–H- -O hydrogen bond (nominally 1.70 Angstrom) rather than the spacing between the two orbitals of the stimulant or GR. Kubo repeated the common phrase, “However, the rationalization of bitterness in terms of chemical structure has been a difficult and long standing problem.” His paper did not change that situation. He did provide a Fischer diagram for a wide range of bitter stimulants, 30, but did not specifically identify the picrophore on any of them. The picrophore associated with many of them is identified in this work. A paper by Yamada et al209. in 1999 obtained detailed information on the structure of several more diterpenes from Isodon and explored their bitterness relative to quinone. However, they did not explore the transduction process.

8.5.4.5.2 Potential picric channel receptors

The earlier discussion has suggested a family of receptors based on simple chemicals esterfied to phosphatidic acid. The same table in Leninger and in Yudkin & Offord of the 1980's suggest two possibilities for the bitter phospholipid receptor, the direct esterification of a simple molecule to the phosphatidic acid or the esterification of a simple molecule to phosphatidyl glycerol at the 3' OH of the glycerol. Figure 8.5.4-26 shows the simplest case of aspartic acid esterified to either of these potential moieties. The molecular structure is very complex in 3D space. However, the d-value between its various pairs of oxygen orbitals are interesting; 3.105, 3.273, 4.375 and 4.827 Angstrom. Some of these d-values are lost during the esterification process. Either of the oxygen orbitals shown closest together to the right of this two-dimensional representation and between the open boxes could be the location of esterification. The value of 4.827 is very close to the 4.746 value for the three odorophores of picric acid (with picric acid being one of the most bitter common chemicals and taken as the major bitter gustant). The other d-values do not suggest Figure 8.5.4-26 Potential phosphatidyl aspartic stimulation of any of the other channel acid receptor and picric acid as gustant. Open receptors. . boxes represent locations of antiparallel coordinate bonds. Note the three picrophores of In this proposed configuration, the precise each molecule of the gustant. See text.

208Kubo, I. (1994) Structural Basis for Bitterness Based on Rabdosia Diterpenes Physiol Behav vol 56(6), pp. 1203-1207,

209Yamada, Y. Sako, N. Ando, E. et al. (1999) New bitter diterpenes: Rabdosianone I & II: Biosci Biotechnol Biochem vol 63(3), pp 524-529 150 Neurons & the Nervous System formation of the one hydrogen bond, shown by the lower open box, appears clear. The formation of the second hydrogen bond, the upper open box, is less clear when using the standard form of aspartic acid generated by the Jmol and similar modeling programs. Interchanging the hydroxyl group at upper center of the picric acid with the carboxyl oxygen at upper left would support formation of the desired hydrogen bond within the upper open box. Taking phosphatidyl aspartic acid (PtdAsp) as the reference picric channel receptor, its d-value in the absence of any crowding becomes the reference value of d = 4.827 Angstrom. The ability of picric acid, with a d-value only 1.7% different, to bind in a DACB appears well within the acceptance range of the receptor. The configuration shown suggests many of the ring based bitter gustants; quinine (d = 4.801), caffeine (d = 4.700), xanthine (d = 4.700), xxx & xxx can bind to the proposed PtdAsp receptor without difficulty.

An alternate to PtdAsp would be its close relative, phosphatidyl asparagine (PtdAsn), with NH2 replacing the hydroxyl group farthest from the amine carbon. However, its promising bond lengths are slightly longer than for PtdAsp, 4.89 and 5.52 Angstrom.

8.5.4.5.3 Hydrated organic molecules as picric channel gustaphore

As noted in the discussion related to the hydrated sodium ion, many of the other bitter organic gustants may only be stimulants when in a hydrated state. 8.5.4.5.4 hydrated hydrogen sulfide as an inorganic picric channel gustaphore

Xxx add here It appears that hydrogen sulfide is also a bitter tasting gustant when in the hydrated state. Signal Generation & Processing 8- 151 8.5.4.5.6 OBSOLETE MATERIAL ON PICROPHORE/RECEPTOR MATCH

[xxx edit or drop. must redraw figure to show d-value from Jmol etc. ] The most probable phospholipid of the outer bilayer membrane of the microvilli capable of operating as the sensory receptor is a form of Ptd3'Og with one unsaturated fatty acid providing electrical conductivity along the length of the molecule. Figure 8.5.4-27 shows the structure of this candidate along with its predicted sensitivity in taste sensation space. As shown by the unspecified group, R, Ptd3'Og.is actually a large family. In their experiments, Silvius et al. found R to be mostly in the form of alanine210. In this figure, an acyl substitution has been introduced as an alternate to the alanine. Any number of R ligands could be introduced to provide an additional carbon and either a hydroxyl group or carboxyl oxygen. The active portion of the overall phosphatidyl molecule is shown at upper left beginning with the ester to phosphoric acid on the left. The pertinent feature is the distance of about 4.7 Angstrom between the oxygen atom, B, and the amine, AH. This broad spacing suggests this sensory receptor is positioned as shown in the lower frame of the figure, achieving a sensation space with minimum overlap with the other sensory channels. As noted earlier, the mean values of each distribution is calculable but the distributions are shown only conceptually.

XXX CAN PROBABLY ELIMINATE R’ AND THE Figure 8.5.4-27 Candidate sensory receptor DOUBLE BOND BY PLACING H2 ON THE performance for the “bitter” channel. Upper left, LOWER CARBON AS IN ASPARTIC ACID. active ligand of bitter sensory receptor. Upper right; quassin shown oriented so as to form an The upper right of the figure shows a very AH,B coordinate bond with the sensory receptor. complex molecule known for its bitterness, Bottom; rectilinear taste space showing the quassin (or Surinam quassia in commerce). quinine channel centered at an A B spacing of It has a bitterness threshold of 1:60,000. 4.7.46 Angstrom. See text. Note the symmetry of this molecule suggesting it could form the AH,B bond in multiple ways. A similar graphic can be prepared showing coordinate bonding with brucine, another important bitter stimulant.

210Silvius, J. Mak, N. & McElhaney, R. (1980) Lipid and protein composition and thermotropic lipid phase transitions in fatty acid-homogeneous membranes of Acholeplasma laidlawii b Biochim Biophysica vol 597(2), pp 199-215 152 Neurons & the Nervous System

Figure 8.5.4-28 shows the proposed sensory receptor coordinate bonding to the common laboratory stimulant quinine. The nominal spacing between the coordinate bonding orbitals of quinine is 4.801 Angstrom, approximately one percent greater than the nominal bonding orbital spacing of picric acid, 4.746 Angstrom. The receptor orbitals are oxygen and nitrogen while the stimulant orbitals are both oxygen. R is shown as carboxylic acid in the figure. The left-most oxygen forms the ester with the conductive phosphatidyl group. The potential for an enhanced bitter sensation resulting from a third site forming an AH,B,X bond like that of the sugars should not be dismissed. If the bitter sensory receptor is in fact 3'-O-aminoacyl glycerol, it exhibits a potential X site at either of the two oxygen atoms associated with the glycerol moiety and at either of the two oxygen atoms of the carboxyl group.

Figure 8.5.4-28 Candidate picric sensory receptor and quinine coordinate bonding RESCALE. SHOW left-most O as ester. 3D representation for quinine from David Woodcock, http://Kaist.ac.kr.edu. Signal Generation & Processing 8- 153 Figure 8.5.4-29 shows a key feature of the candidate P-path picrophores and GR’s. It is particularly interesting because of the range of chemicals that can satisfy the nominal picrophore requirement within an acceptable tolerance. The nominal situation based on picric acid is a d-value equal to 4.746 Angstrom. This value is determined by the five carbon chain immersed in a fusion of homocyclic six carbon rings. The rings may be incomplete, as long as the orbitals occupy the positions commensurate with complete rings. The minimalist condition can be labeled a (2R,4R)-pentanediol_2005883 with a d-value of 4.871. Note the bonds extending beyond the ends of the upper carbon chain. The pentanediol exhibits methyl groups on the two extreme carbons of the lower frame. A functional alternative is a 1,3 propanediol immersed in a structure of one or more rings if its d-value is close enough to nominal to support a P-path DACB. Generically, 1,3 propanediol is described as sweet in Merck Index. However, when in the restricted configuration required here, the resulting d-value prescribes a bitter perception. In the configurations shown, any of the oxygen orbitals can be replaced by a nitrogen or other orbital. In the bottom structure, exchanging one or more oxygen atoms for nitrogen has only a nominal effect on the net d-value.

The upper frame shows the nominal picrophore based on five carbons associated with multiple ring structures. In this configuration, the d-value of 4.746 Angstrom is independent of the particular orbital (shown as oxygen in this example). Exchanging either or both orbitals has no effect on the d-value of the structure. This configuration offers immense flexibility in orbital selection.

The lower frame shows potential alternate picrophores. Note the open bonds indicative of the presence of additional Figure 8.5.4-29 Potential P-path picrophores and structure meeting the above criteria. In this receptors. The candidates are symmetrical as configuration, the d-value of the immersed shown. However, each oxygen can be replaced ligand varies with the length of the with another type orbital. Top; the nominal particular orbital-carbon bond length. candidate where the choice of orbital does not For example, change the d-value. Bottom; an alternate where the d-value varies nominally with orbital. See text.

Terminal bond d-value Deviation type From 4.746 A

Both O–C 4.83 Angstrom +1.7% 1 O–C & 1 N–C 4.75 0 Both N–C 4.66 –1.8%

8.5.4.6 Operation of the “super-bitter” sensory neurons

Figure 8.5.4-30 from van der Heijden is presented to describe the potential families of super-bitter picrophores. It is likely that the chemicals with threshold sensitivities at concentrations at or below 0.001 nanomoles per liter are employing a different or augmented mechanism for exciting 154 Neurons & the Nervous System the P-path GR’s. It will be initially proposed that this mechanism is of the AH,B,X type so successfully applied to the super-sweet glycophores. Van der Heijden provided potential AH,B,X dimensions for his review of potential super-sweeteners but he did not extend the concept to these potential super-bitter compounds.

Figure 8.5.4-30 The bitter and super-bitter stimulants of van der Heijden EDIT. CHG sugars to sugar derivatives. As footnoted, potency refers to threshold sensitivity concentrations. Abbreviated from van der Heijden, 1993.

The table includes a triterpene labeled lucidenic acid D1 and cites Nishitoba, 1989 in S Agric Biol Chem vol 52 pp 1791+.. Figure 8.5.4-31 illustrates two stick figure representation of this chemical under the nomenclature, HMDB: 36856 (Lucidenic acid D1) and HMDB 38199 of the Canadian Human metabolome data base. While these representations suggest a complex non-planar structure for this chemical, the 3D Jmol representation confirms it is a nearly planar as shown in the next figure. It is very difficult to ascertain the picrophore in the upper representation. However, in the lower representation, the candidates are quite obvious and the actual picrophore involves the bracketed oxygen orbitals as will be confirmed using the next figure. Signal Generation & Processing 8- 155

Figure 8.5.4-31 2D representations of Lucidenic acid D1. Note the bracket at lower left indicating the pair of orbitals forming the picrophore in this chemical, d-value = 4.769 Angstrom.

Figure 8.5.4-32 displays a 3D representation of Lucidenic acid D1 (HMDB 28199) using Jmol. The hypothesis of this work calls for a DACB with an optimum d-value of 4.746 Angstrom. The measured value based on the Jmol model deviates from this prediction by only one part in one- thousand, a very good confirmation of the hypothesis of this work. The relative simplicity of this molecule makes it useful in searching for the dispersion centroid supporting the super-bitter character of this molecule. Two most likely dispersion centroids for an AH,B,X model of a super- bitter material are shown, X1 will be assumed to be the actual point until additional evidence from other chemicals is gathered. 156 Neurons & the Nervous System

Figure 8.5.4-32 Lucidenic acid D1 (HMDB 38199) from Jmol. The baseline picrophore measures 4.769 Angstrom. The candidate dispersion centroid, X1, is 8.013 Angstrom from one orbital and 5.749 Angstrom from the other. The alternate dispersion centroid, X2, is at 9.655 & 6.310 Angstrom with an acute angle of 120.62 degrees. The inset at upper right shows the molecule to be non- planar.

The fact that the d-value of lucidenic acid D1 was within one part in one thousand of the expected value based on the gustatory portion of the Electrolytic Theory of the Neuron is strong vindication of this theory. However, the application of the AH,B,X concept will require better understanding of the quantum-mechanical-(electrical) characteristics of the molecule.

Lucidenic acid is a general label for a very large family of triterpenes. Several lucidenic acids are shown in Figure 8.5.4-33 from Guoa et al211. These materials exhibit a striking resemblance to picric acid. One of these has been bracketed to indicate its picrophore. The rest of the molecules on this page have had their structure changed to the point they exhibit no picrophore and would not expect to be perceived as bitter.

211Xiao-Yu Guoa, X-Y. Hana, J. Yea, M. et al. (2112) Identification of major compounds in rat bile after oral administration of total triterpenoids of Ganoderma lucidum by high-performance liquid chromatography with electrospray ionization tandem mass spectrometry J Pharm Biomed Anal vol 63, pp 29–39 Signal Generation & Processing 8- 157

Figure 8.5.4-33 Variants of lucidenic acid ADD. Only the derivative at lower left still exhibits the pair of orbitals at the correct d-value to qualify as a picrophore. It can be expected to be bitter and probably super-bitter. From Guoa et al., 2112. 158 Neurons & the Nervous System

Figure 8.5.4-34 shows a 2D view of amarogentin (CAS #21018-84-8) and a similar view of artabsin. Unfortunately, the CAS # for artabsin is difficult to define and varies with databank. The Brookhaven PDB databank code of 1A31 appears more reliable. This version of artabsin includes a seven-carbon ring and two five sided rings (one homocyclic and one heterocyclic with the inclusion of an oxygen. The 3d representation of amarogentin, CAS #21018-84-8, clearly shows it exhibits a glycophore with d-value = 2.754 Angstrom (O24–C–C–O26) as well as a picrophore with d-value = 4.648 Angstrom (O28–C–C–O34). It would therefore qualify as a stimulant exhibiting both a A-path gustaphore and a P-path gustophore. Its complexity suggests it is also capable of participating in a super-bitter perception such as that developed in the next major section. A distinctly different variant of amarogentin labeled 692114 in the “Darwin” library of the GNU- Darwin project is very complex as displayed in 3D. Lacking additional support, this representation is impossible to analyze. The structure of artabsin is considerably simpler but exhibits no obvious picrophore involving only atoms for orbitals. Van der Heijden shows an artabsin with an eight-carbon ring and two five- sided rings (including an oxygen in one heterocyclic ring) citing Brieskorn, 1979, in German. However, other Jmol representations show a seven-carbon ring. Both arrangements strongly suggests the involvement of the ring structures in any functional AH,B picrophore and probably any AH,B,X picrophore. The only potential picrophore that has been developed using DS3.5 visualizer is that from the upper oxygen and the centroid of the seven-sided ring, a d-value of 4.953 Angstrom. The lower oxygen and the seven-sided ring qualifies as a A-Path acidophore with d = 2.860 Angstrom.

Figure 8.5.4-34 Potential AH,B,X geometries for amarogentin and artabsin. The artabsin is based on a seven-sided ring. Amarogentin exhibits at least one picrophore with a d-value of 4.648 Angstrom (2% below the nominal 4.746 Angstrom). Artabsin exhibits one picrophore between the upper oxygen orbital and the seven-sided ring with a d-value = 4.953 Angstrom (4% above the nominal 4.746 Angstrom). Both may also support a dispersion centroid qualifying them as super-bitter.

8.5.4.7 Summary of the proposed receptor d-values CONSOLIDATE Signal Generation & Processing 8- 159 Figure 8.5.4-35 shows a summary of the d-values for the proposed four sensory receptors of gustation and a provisional estimate of the width of each sensitivity. The sensitivity widths are taken as 5% of the center value of the path based on the family of bitter (P-path) picrophores discussed in earlier sections.

Figure 8.5.4-35 Proposed summary d-values for the gustatory receptors with an estimate of their sensitivity profiles.

Each of the gustatory receptors is functionally a diol esterified to a phosphatidyl lipid. The controlling feature of each GR is the d-value between the two oxygen orbitals of the diol. Immel has discussed a variety of sugars using the diol extension (Section 8.5.5.1.3). 8.5.4.8 Other gustaphores

There are a number of organic and inorganic gustaphores described in the literature that can be interpreted in the context of the hypothesis of this work.

Bryant & Mezine presented an extensive family of complex organic molecules in an investigation of nocent stimulation of the trigeminal nerve of the rat. Probably all of the molecules quality as gustants that are relevant to this work. However, the molecules contain large numbers of individual gustaphores and odorophores that make their value in sensory research limited. See Section 8.7.3.1.2.

8.5.4.8.1 CaCl2 & MgCl2 as gustaphores or nocents

The sensory significant of CaCl2 and MgCl2 are unclear from the literature. They are frequently reported as being perceived as bitter as well as or rather than salty. Alternately, they are described as astringents within the general class of nocents. The reasons for this will be developed below.

Both CaCl2 and MgCl2. are described as deliquescent, forming crystals including six molecules of water. In the biological context, they are so deliquescent that they are, like alum, astringent in the presence of biological tissue. The anhydrous calcium salt is deliquescent; it can 160 Neurons & the Nervous System accumulate enough water in its crystal lattice to form a solution. When calcium chloride is added to water (solubility–74.5 gr/100 ml), much heat is liberated. Magnesium chloride acts similarly. Its solubility at biological temperatures is 60 gr/100ml. While these materials completely ionize in water, it is the coordination chemistry of the hydrated cations that is significant in gustation. The anion plays no active role. Magnesium sulphate (milk of magnesia) is also perceived as both bitter and salty. When in solution, the cations of these chemicals form coordinate bonds with the water molecules. Calcium generally bonds to six or seven water molecules, occasionally eight or less than six. For six ligands, the structure is octahedral. For eight ligands, the structure is hexagonal bipyramidal. Other bonding levels typically employ one of these forms with one or more coordinate locations empty (although some sources report a pentagonal bipyramidal structure for this salt). At biological temperatures one calcium is predominantly bound to six water molecules. Calcium forms complexes with many molecules. The ligand distances vary between 2.1 and 2.6 Angstrom. When coordinate bonded to water, the ligand distance is normally 2.431 Angstrom. For a coordination number of eight, the hexagonal bipyramidal structure exhibits angles of 60 degrees between the six oxygen ions in the horizontal plane and 90 degrees between the oxygen ions in that plane and the two oxygen ions at the points of the pyramids. Figure 8.5.4- 36 shows the medial plane of the octahedral structure of the calcium ion fully coordinated with six water molecules after CaCl2 is dissolved in water. The ability of the oxygen atoms to participate in paired coordinate bonds with a sensory receptor is obvious. The nominal 212 +2 calcium to oxygen spacing is from Kim et al . As shown, the Ca(H2O)6 complex could +2 exhibit gustaphores with d-values of 3.437, 4.210 and 4.862 Angstroms. The smaller Mg(H2O)6 exhibits equivalent values of 2.9, 3.585 and 4.14 Angstrom Glusker et al., 1999). Bock et al. (1994) have noted that for MgCl2, the hydrogen atoms are at an angle of 120-127 degrees from the Mg–O bond. This angle would be 127.7 degrees if the pHOH was the nominal 104.5 degrees

Figure 8.5.4-36 Calcium cation fully coordinated with water when in solution. Coordinate bonds are shown dashed. The cation retains its conventional valence of +2. Dimensions in Angstrom. See text.

212Kim, S. Gregor, W. Peloquin, J. Brynda, M. & Britt, R. (2004) Investigation of the Calcium-Binding Site of the Oxygen Evolving Complex of Photosystem II Using 87Sr ESEEM Spectroscopy J Am Chem Soc vol 126, pp 7228-7237 Signal Generation & Processing 8- 161 Yang et al. discuss variations in the angles between the oxygen ions in stressed complexes213. Magnesium complexes in the same manner as calcium and with a similar variation in coordinate levels (with six being the predominant value 79% of the time214). The oxygen to oxygen spacing in the octahedron form is 2.9 Angstrom. The spacing between opposed oxygen ions is 4.14 Angstrom. [xxx duplicates above ] [xxx Is either form of calcium or magnesium good matches for the values of a picrophore, d=4.2 Angstrom, or Natrophore, d=3.3 ] The d-value of 4.14 Angstrom for the opposing oxygen ions of the hexagonal Magnesium complex makes this structure a potential picrophore if it is able to satisfy the stereo-chemical requirements and reach the appropriate sensory receptor site.

Figure 8.5.4-37 shows the relationship between the these “salts” and the receptive range of the gustatory sensory receptors. It clearly shows why these salts are perceived as bitter. Both have d-values matching the values of the bitter sensory channel. CaCl2 and MgCl2 are both picrophores. Both exhibit d-values that could qualify them as natrophores also, depending on the sensitivity range of the sensory receptors..

[xxx edit ] The d-values of these “salts” aid in determining the receptive range of the sensory receptors. In the case of the picrophores, they exhibit calculated-values +2 of 4.14 (Mg(H2O)6 ) to 4.22 (Quinine xxx). These values suggest a receptive range of at least 0.08 around a center value of 4.18. On the other hand, the gustaphores of these salts at 3.437 and 3.585 are less effective in stimulating the sodium sensitive channels and help define the sensitivity range of the sodium sensitive channel. The sodium sensitive channel can be considered to be centered at d = 3.30 with a range at most only marginally higher than 3.437. These values suggest a center value of 3.30 with a range on the order of 0.28 for the sodium sensitive channel.

This analysis suggests the picrophores of Magnesium salts are formed by the oxygen atoms separated by 120 degrees about the cation while the picrophores of calcium salts are formed by the oxygen atoms on opposite sides of the cation. In both cases, these hydrated ion structures exhibit multiple picrophores associated with each cation, raising the probability that these structures will form dual coordinated bonds with the Figure 8.5.4-37 The potential gustaphores of CaCl2 bitter sensory receptors. The nearest oxygen and MgCl2 ADD. Both salts exhibit significant atom pairs of these structures do not act as picrophores, at 4.14 and 4.21 Angstrom. The effective gustaphores. calcium salt may also exhibit a natrophore at 3.437 Angstrom, depending on the receptive 8.5.4.8.2 The thio moieties as range of the sensory neuron receptor. stimulants

213Yang, W. Lee, H-W Hellinga, H & Yang, J. (2002) Structural analysis, identification, and design of calcium- binding sites in proteins Proteins struct funct genet vol 47, pp 344-356

214Glusker, J. Katz, A. & Bock, C. (1999) Metal ions in biological systems Rigaku J pp 8-16 162 Neurons & the Nervous System

A variety of thiol compounds, R:SH (mercaptan) and R:SR’ (a thioether), appear as gustants in the literature. The lack of sufficient oxygen in some molecules has hampered their interpretation as gustaphores. However, the formula shows the Sulphur in these moieties exhibits unpaired electrons that can act as orbitals in a DACB arrangement. The amino acid, cysteine, is a member of this group. It is frequently perceived as “sulphurous” in both its D– and L– isomer forms. Sulphurous is a term outside the normal organic acid, sweet, sodium ion or picric channels of gustation. Engel has provided simple Fischer diagrams for an assortment of sulphur bearing compounds (including some homologs) providing a perception of sweetness215. Figure 8.5.4-38 shows the distances between the relevant orbitals in cysteine. It should be obvious that the sulphurous perception of cysteine is due to its ability to stimulate the acid receptor with its d-value of 2.223 Angstrom, the sodium channel receptor with its d-value of 3.472 Angstrom and the sweet receptor with its d-value of 2.785 Angstrom (and probably its d-value of 2.979 Angstrom). This situation is characteristic of many thiols. As will be shown when discussing the perception of gustaphores, the thiols do not appear at the nodes of a multidimensional space. Their perception results from stimulation of multiple sensory receptors simultaneously. Based on the above analysis, cysteine may not exhibit the identical set of gustaphores in its two isomeric forms. [xxx ]

Figure 8.5.4-38 The d-values associated with orbital pairs in the amino acid, cysteine. The multiple pairs of orbitals provide multiple d-values able to stimulate multiple gustatory receptors. See text.

215Engel, K-H. (1999) The importance of sulfur-containing compounds in fruit flavors In Teranishi, R. Wick, E. & Hornstein, I. eds Flavor Chemistry: Thirty Years of Progress pg 267 Signal Generation & Processing 8- 163

Shallenberger & Acree have provided some information concerning a possible structural arrangements important in understanding the role of thiols in gustation. Figure 8.5.4-39 shows this molecule. The complete structural configuration elicits a bitter sensation but the abbreviated form they presented in 1971 could not account for it. The thio-carbamide was originally found in phenylthiocarbamide (PTC) by Fox (1932). The Jmol representation of this complete molecule suggests it can be perceived as bitter based on the distance between the extreme nitrogen and the aromatic ring acting as an electron donor. The aliphatic portion is planar but slightly out of the plane of the ring. The distance from the extreme nitrogen to the center of the planar ring is 4.94 Angstrom. The effective location of the charge associated with the ring has not been determined.

Steudel & Steudel have presented some useful data on the bond lengths and angles associated with a variety of sulphur based compounds216. 8.5.4.8.3 The “water” gustatory sensory response

Several investigators have reported a “water” response for their sensory neurons. With an understanding of the concept of pre-adaptation, this response should not be surprising. A water flush after stimulation with any stimulant or even normal saliva can be expected to constitute a change in the taste bud environment. This previous exposure to a stimulant or saliva can be considered a pre-adaptation step with the water wash a subsequent stimulus test Figure 8.5.4-39 Phenythiocarbamide as presented interval. The sensory neurons can be in stick and 3D form. Stick form was shown for expected to report this change. Pages 262- discussion by Shallenberger & Acree in 1971. 3D 268 in Cagan & Kare discuss their form created using Jmol. The more complete experiments. structure suggests a perceived bitterness based on the phenyl ring acting as an electron donor. xxx has described the water taste in his Firmench Award Address217. “Concentrations of salt that are weaker than the levels in saliva may give rise to bitter tastes, which may in fact be water tastes. Water tastes occur when the mouth is completely or nearly-completely adapted to any suprathreshold concentration of a compound and then water is sampled.” 8.5.4.8.4 The “browned flavors” sensory response

Hodge, followed by Ohloff218, have studied the characteristics of a large range of molecules associated with the sensations resulting from the browning of food products during cooking. Hodge analyzed the chemical structure of a wide variety of molecules he associated with the browned flavor sensation219. He described these sensations as;

216Steudel, R. & Steudel, Y. (2009) Sulfur Dioxide and Water: Structures and Energies of the Hydrated Species SO2AnH2O, [HSO3]–AnH2O, [SO3H]–AnH2O, and H2SO3AnH2O (n = 0–8) Eur J Inorg Chem pp 1393–1405

217Xxx (2001) xxx In Spanier, A. et al. eds. Food Flavors and Chemistry. Royal Soc Chem pg 43

218Ohloff, G. (1981) Bifunctional unit concept in flavour chemistry In Schreier, P. ed. Flavour ‘81. Berlin: Walter de Gruyter pp 757-770

219Hodge, J. Mills, F. & Fisher, B. (1972) Compounds of browned flavor derived from sugar-amine reactions Wash, DC: Agricultural Research Service, U.S. Department of Agriculture. 164 Neurons & the Nervous System

“They are essential for the recognition and acceptance by taste of many processed foods, especially cereal-derived foods. Browned flavors include caramelized sugar aromas (1); food aromas that have been described variously as toasted, baked, nutty, or roasted (2); corny and amine-like aromas from cooked grains and meals; and both the desirable and undesirable burnt aromas and bitterish tastes of roasted malt, nuts, coffee, chicory, cocoa, meats, fruits, and vegetables (2). The objectionable burnt and bitter flavors of overheated or long-stored dehydrated foods also fall within this class.” Whereas Hodge noted the ability of many of these molecules to form an intramolecular hydrogen bond between a divalent oxygen and an adjacent hydroxyl moiety, this relationship did not suggest any arrangement between these molecules and a sensory receptor. Ohloff, in the same time period as Shallenberger & Acree were working on the sugars, proposed the “browned flavor” stimulants employed the same AH,B coordinate bonding to a sensory receptor as the sugars (with a spacing of less than 3 Angstrom) but with an additional relationship. He suggested this relationship involved the hydrophobic part of the molecule. He was unable to specify the precise relationship between this apolar portion of the molecule and the receptor as a substrate.

No multidimensional analyses involving these flavors could be located. It appears the “browned flavor” stimulants could support a dual channel sensation involving the G-Path channel and one or more other best channels. 8.5.4.8.5 The role of amines & amino acids in the taste sensation TIE 8.6.2.6.6

[ rewrite and broaden to cover both categories ] Belitz et al. have provided a simple table showing the critical role of the carboxyl group and a ligand containing some form of nitrogen in giving the amino acids a sweet taste220. More recently, Van der Heijden (page 104) has described the relative sweetness and bitterness of the amino acids based on their steric form. Of those amino acids addressed, the D– forms are invariably sweet and the L– forms invariably bitter. Van der Heijden provides a brief list of properties associated with amino acids and peptides and the sensation of bitterness. [xxx list ] Kier221, has provided a description of the sensations elicited by the amino acids that differs significantly from that of Van der Heijden.

Kato et al. have provided considerable background on the gustatory properties of the amino acids222. The discussion makes it very clear that it focuses on the requirements of the sensory receptors and not the obvious properties of the stimulants that determine the gustatory sensations.

Because the amino acids exhibit such a variation between sweetness, non-sweetness and bitterness, they appear to offer a chance to develop a better understanding of the structural relationships required to elicit sweet and bitter tastes in general. Unfortunately, the various authors provide different findings with regard to the perceived taste of each amino acid. Kato’s citations are primarily to investigators working in the 19th Century. 8.5.4.8.6 The phenols and aliphatic-aromatics

Phenol (C6H5OH) is the first member of a large family generally described as aromatic alcohols, ArOH, where Ar is phenyl, substituted phenyl or one of the other aryl groups. Hydrogen bonding plays an important role in both the intermolecular and intramolecular chemistry of the aromatics.

220Belitz, H. Chen, W. Jugel, H. et al. (1979) In Bourdreau, J. ed. Food Taste Chemistry. Washington, DC: American Chemical Society Chapter 4

221Kier, L. (1972) A molecular theory of sweet taste J Pharm Sci vol 61(9), pp 1394-1397

222Kato, H. Rhue, M. & Nishimura, T. (1989) Role of free amino acids and peptides in food taste In Teranishi, R. Buttery, R. & Shahidi, F. eds. Flavor Chemistry: Trends and Developments. Washington, DC: American Chemical Society Chapter 13 Signal Generation & Processing 8- 165 Phenol is generally known as carbolic acid and is both poisonous and caustic. However, its compounds play a major role in both gustation and, as the next major section, 8.6, will show in olfaction. While the name carbolic acid suggests an organic acid, is is actually a highly basic aromatic alcohol. The shared charges of the aromatic chemical structure is an electron donor. As such, it is able to perform as an “orbital” in the AH,B DACB mechanism. This capability provides many potential gustaphores when an aromatic ring is associated with one or more aliphatic side chains. When in an aliphatic-aromatic relationship, the electrostatic potential of the aromatic may become asymmetrical relative to the plane of the aromatic moiety. Jmol and the other 3D representations are then needed to calculate the distances between the center(s) of charge of the ring and the orbitals of the aliphatic structures. The halogens play a major role in determining the properties of a stimulant as a gustaphore. Such role is not apparent in the typical 2Drepresentations of molecules. The halogens can play an important role in the gustatory properties of the aromatics. They can effectively withdraw an electron from the charge cloud of the aromatic, disrupting its ability to act as a negative charge source in an AH,B bonding relationship. Many other ligands can also affect the ability of the aromatic to act as an effective orbital (Morrison & Boyd, 1971, pp 822- 840). 8.5.4.8.7 The non-hydroxyl guanidines

The role of the quanidine moiety frequently appears in the gustation literature. Quanidine (not to be confused with quinidine) is an excellent example supporting the hypotheses presented here. It is a simple molecule containing no oxygen or hydroxyl groups. Its orbitals are all NH or NH2 groups as shown in Figure 8.5.4-40. The material is typically described as a base based on its pH value although its structure shows it readily forms DACB’s with the A-Path GR’s leading to a perception of acidity.

Two of the orbital are connected to the third via a carbon atom with a d-value of 2.32 Angstrom. In addition, the two orbitals are connected to each other via a carbon with a d-value of 2.38 Angstrom. These values suggest this chemical should form a DACB with the A-Path GR (nominal d=2.276 Angstrom) quite effectively and probably act as a multi-acidophore stimulant. 166 Neurons & the Nervous System

Figure 8.5.4-40 Quanidine, a stimulant with three gustaphores. The molecule is planar. Each of the orbital pairs represents a gustaphore exciting the A-Path sensory receptor in accordance with the Electrolytic Theory of the Neuron and the hypothesis presented here.

When combined with a simple phenol, or other phenol variants, through dehydration, the resulting derivatives are noted for their perception of sweetness. The phenol guanidine combination at the nitrogen double bonded to the rest of the guanidine exhibits a d- value of 2.71 Angstrom which is well within the 5% tolerance on the DACB range of the G-path GR. [SC-45647 used by Hellekant et al. 1997 in art file from PubChem 3D viewer ]

When the guanidine moiety is combined with a carboxyl group, that group is capable of providing an additional C-path gustaphore to the overall structure. The combination is also capable of acting as a super sweetener of the AH,B,X variant as suggested in Figure 8.5.4-41. The structural requirements on an AH,B,X super sweetener are developed in Section 8.5.3.2. No Jmol 3D representation has been found that supported precise distance measurements for this molecule. There is a potential X site nominally 3-3.5 Angstrom from the AH site and nominally 3.5-4.0 Angstrom from the B Figure 8.5.4-41 The guanidine derivative, SC-45647 site. as a super-sweetener. The distance between the AH,B pair and the additional phenol ring results in an AH,B,X configuration. No Jmol model of this molecule has been located. See text. Signal Generation & Processing 8- 167 Hellekant et al. have studied several derivatives of the quanidine molecule (Section 8.5.8.3) in detail, especially when combined with a phenol by dehydration as well as a carboxyl group via a carbon. 8.5.4.8.8 Procaine and other local anesthetics

Procaine is a topical anesthetic frequently experienced by people during dental maintenance and generally perceived as bitter. Its structure is shown in Figure 8.5.4-42. The spacings relative to its gustatory performance is shown based on a Jmol representation of the Jmol file (CAS 59-46- 1 described as verified on Wikipedia). The value of 2.91 is shown as 3.729 using DS3.5. The value of 4.75 is shown as 4.876 using DS3.5. These differences are major and need to be resolved. The pair of larger values both support the perception of bitterness via the P-path. However, the pair of smaller values differ significantly.

8.5.4.8.9 Nutmeg and Mace

Nutmeg and mace (the spice and not the riot control gas) are two ingredients frequently used in cooking both for their color and delicate flavor. According to the Merck Index, XII Ed., 1996, the essential oil obtained by steam distillation of ground nutmeg is used widely in the perfumery and pharmaceutical industries. This volatile fraction typically contains 60-80% d-camphene by weight, as well as quantities of d-pinene, limonene, d-borneol, l-terpineol, geraniol, safrol, and myristicin. Figure 8.5.4-42 Structure of procaine with d-values as represented using the Jmol visualizer. The D-camphene (ChemSpider 83259) is a very values using DS3.5 are significantly different for the complex bicyclo-heptane hydrocarbon both values but particularly for the larger containing none of the orbitals normally distance.. associated with either olfaction or gustation except the C=C bond associated with a methylene group. A further extension of the hypothesis of this work will be required to address these chemicals as gustants. Some of them are volatile enough to be important in olfaction. Borneol is an alcohol with the hydroxyl group replacing the methylene group of camphene. Chrysantheonone is the aldehyde associated with borneol. The acetate is even more complex and contains two oxygen orbitals but no simple aromatic ring structures or covalent carbon bonds. See Section 8.6.2.7.4. 8.5.4.8.10 Heterocyclics–caramel and butterscotch

The perception of caramel/butterscotch is based on maltol_8066 (a.k.a., veltol), a heterocyclic of oxygen with both a carbonyl and hydroxyl attached directly to the ring. The resulting d-values to the hetercyclic oxygen are 3.629 and 3.974 Angstrom. If the ring is still able to support dislocated electrons, the d-values of 2.573 for the carbonyl oxygen and 2.597 for the hydroxyl oxygen should be considered. Caramel and butterscotch are categorized as gustants rather than olfactants due to their low vapor pressure. Quoting the Perfume Shrine; “You can create the scent of caramel with 3-hydroxy-4,5-dimethyl-2(5H)-furanone. If you take that molecule and add a small amount of ethyl butyrate, ethyl valerate, and phenethyl acetate, you get a nice fresh garden berry that would work great in an Escada launch. God forbid the public knew it.” Sotolon_56569, a.k.a., 3-Hydroxy-4,5-dimethyl-2(5H)-furanone is very similar to maltol, except it includes two additional methane groups attached to the main heterocyclic ring. These additional features probably contribute to the commercial utility of the material.

8.5.4.8.11 Heterocyclics–the pyridines 168 Neurons & the Nervous System

Ache & Carr have discussed a wide range of pyridines and their role in the gustation of crayfish223. They provide individual formulas and structures that can be analyzed with respect to their stimulation of various OR channels. They provide considerable material on the “tuning” of the crayfish receptors in terms of these compounds. Of specific interest is their figure 5 showing the linearity of the stage 3 pulse rate versus the stimulus concentration applied to the stage one GR’s when plotted using a linear ordinate and logarithmic abscissa. xxx, writing in Cagan, (page 112 have described the major pyridines using Figure 8.5.4-43 It illustrates a variety of pyridine molecules known to affect the chemoreceptors of crayfish224.

Figure 8.5.4-43 Effectiveness of 12 pyridines in stimulating the chemoreceptors of crayfish. The effectiveness of the substanced decreases from left to right in each row. See text. From Hatt & Schmiedel-Jacob, 1984.

8.5.4.9 The putative “umami” sensory response

The Japanese, beginning in 1908, have suggested a fifth fundamental taste sensation called umami. The term appears to have been derived from their word, umami (delicious). The designation appears to have a strong representation in their culture. Whether this sensation is the result of a mixture of the other four sensations, an enhancement of the overall taste sensation or a fifth sensation is yet to be resolved. Physiology and Behavior published a special issue devoted to Umami in 1991225. While useful data came out of that conference little information appeared supporting umami as a unique sensation of gustation. Yamaguch & Ninomiya have recently asserted, “Umami makes a variety of food palatable, although it is not palatable by itself226.” In the same article, they quote Ikeda as saying while “umami cannot be produced by any combination of taste qualities, many researchers believed that it could be duplicated by the four conventional tastes.” This work will show the latter is the case. 8.5.4.9.1 History of umami

223Ache, B. & Carr, W. (1989) Chemorecetion in aquatic invertebrates In Cagan, R. ed. Neural Mechanisms in Taste. Boca Raton, FL: CRC Press Chap. 5

224Hatt, H. & Schmiedel-Jacob, I. (1984) xxx J Comp Physiol vol 154, pp 855+

225- - - - (1991) Special Issue Physiol Behav vol 49(5)

226Yamaguchi, S. & Ninomiya, K. (1999) Umami and food palatability In Teranishi, R. Wick, E. & Hornstein, I. eds. Flavor Chemistry: Thirty Years of Progress. NY: Kluwer Academic/Plenum Chapter 36 Signal Generation & Processing 8- 169 Yamaguchi has been the leading investigator of umami. He has prepared the most extensive study of this putative sensation, including extensive lists of chemicals believed to involve the umami sensation227. More recently, he has presented some results of multi-dimensional analyses and asserted that umami is represented by a different dimension than the other four historical sensations228. However, the printed record of this poster presentation does not include any graphical material and speaks of dimensions and vertices of the taste sensation space that may suffer in translation. It does reference, but not cite in the paper, a larger paper of ten years earlier229. This paper did not include a full multi-dimensional analysis. It specifically did not include the basis factors demonstrating a four dimensional taste space. It leaves the question of whether umami is a taste enhancer or is sensed by a distinct sensory channel unanswered. It is yet to be resolved whether their interpretation introduces a fourth dimension in taste space or merely occupies another vertex of the three dimensional space. 8.5.4.9.2 Recent literature on umami

At this time, it is hypothesized that umami is represented in a 3D taste space as distinct from any of the four nodes but not orthogonal to them. A perception of umami is the result of simultaneous stimulation by a natrophore, an acidophore and a glycophore, the latter two due to the glutamic acid component of the glutamate ligand. As a result, the sensation of umami does not fall along any of the axes of the 3D taste space but within the volume of that space. The principal stimulant associated with umami is mono-sodium glutamate, an unusual chemical that includes the primary N-Path channel stimulant (hydrated sodium) and a well studied stimulant, glutamate. The glutamate ligand exhibits the ability to coordinate with the A-Path sensory receptor via its carboxyl group(s), and to potentially coordinate with the G-path sensory receptor via its spacing of an NH and O combination providing an AH,B union. The d-value of these atoms is 2.66 Angstrom versus the nominal 2.82 Angstrom for the G-path GR. The naturally occurring L-glutamate is reported to be tasteless while D-glutamate is sometimes reported to be sweet. The sensations elicited by mono-sodium glutamate are primarily those of salty, with a slightly acidic, and to some a slightly sweet taste. These conclusions are consistent with Belitz et al (1979).

The glutamate anion when present in the fluid milieu of the organism (not the oral cavity) also plays a critical role in powering the sensory neurons. In this role, it is frequently described as a neuro-facilitator. However, in excess, it is known to cause excessive neural performance. In this case, glutamate can increase the intensity of the taste sensation due to its increasing the static potential at the collector of the first Activa formed within the microvilli. The resulting increase in sensory neuron sensitivity could be reflected in the sweet channels and potentially in the acid and sodium channels as well.

Several sources have made comments to the affect that inosine and guanosine compounds may elicit a umami perception, but without data explaining how or why. These very complex compounds are likely to excite a variety of the GR’s other than the organic acid GR. Kurihara & Kashiwayanagi addressed umami in 1998 and described the principle chemicals associated with it as mono-sodium glutamate, disodium inosinate and disodium guanylate230. Disodium guanylate incorporates both a sodium ion and a glycophore. Disodium inosinate is unique in that the sodium excites the N-Path receptor, PtdIns and the inosinate is a derivative of the family forming

227Yamaguchi, S. (1979) The umami taste In Boudreau, J. ed Food Taste Chemistry. Washington, DC: American Chemical Soc Chapter 2

228Yamaguchi, S. & Komata, Y. (1987) Independence and primacy of umami as compared with the four basic tastes Annals NY Acad Sci vol 510, pp 725-726 A special issue of this journal

229Yamaguchi, S. (1979) The umami taste In Bourdreau, J. ed. Food Taste Chemistry. Washington, DC: American Chemical Society Chapter 2.

230Kurihara, K. & Kashiwayanagi, M. (1998) Introductory remarks on umami taste In Murphy, C. ed. Olfaction and Taste, XII. NY: New York Academy of Sciences pp 393-397 170 Neurons & the Nervous System the N-Path receptor, PtdIns. There is little evidence these chemicals excite an independent sensory channel. The inosinate forms a dual hydrogen bond dimer with the inositol of the PtdIns according to this theory. The dual hydrogen bond spacing is obviously the ideal 3.3 Angstrom. In 1991, Yamaguchi reported a large scale test comparing the sensations of Orientals and Caucasians of European origin231. No significant differences were found. Also in 1991, as part of an extensive study of the chimpanzee, Hellekant & Ninomiya specifically sought to understand the relevance of umami232. They asked the question and then responded as follows; “Is There an Umami Taste Quality? Unfortunately, we have not yet acquired enough data to answer this question. We have no fibers that responded exclusively to the umami compounds.” Continuing; “We can state that the present study has laid the foundation for an answer to this question, by creating an overview of the fiber types that exist in the chimpanzee, but does not answer the question.” Their subsequent study did not provide a clear answer to the above question233. Their results indicated MSG was only sensed by the N-path GR’s. They did note, “These results suggest that the bitter and sweet tastes are conveyed in specific and separate groups of nerve fibers in the chimpanzee.” 8.5.4.9.3 The underlying mechanism–the perception of umami

Figure 8.5.4-44 shows a selected list of the broad range of chemicals claimed to be associated with the umami sensation based on purely psychophysical tests234. Note the difficulty of determining any specific structural commonality between these chemicals using only two- dimensional “stick figure” diagrams (frequently with the positive ion omitted). However, note the commonality of the carboxyl group in a majority of the chemicals. The presence of this group suggests at least one of the sensations elicited by this family is the acidic sensation due to coordination with the serine sensory receptor (representing the H–best channel). The pair of adjacent OH groups in the pentane ring may be separated sufficiently to be able to coordinate with the AH,B receptor site of serine. The hydrated sodium ion of MSG is obviously able to elicit the salty sensation simultaneously.

231Yamaguchi, S. (1991) Basic properties of umami and effects on humans Physiol Behav vol 49, pp. 833-841

232Hellekant, G. & Ninomiya, Y. (1991) On the Taste of Umami in Chimpanzee Physiol Behav, vol 49, pp. 927-934

233Hellekant, G. and Ninomiya, Y. (1994) Bitter taste in single chorda tympani taste fibers from chimpanzee. Physiol Behav vol 56(6), pp 1185-1188,

234Schlichtherle-Cerny, Affolter, M. & Cerny, C. (2004) Taste-Active glycoconjugates of glutamate: new umami compounds In Hoffman, T. Ho, C-T & Pickenhagen, W. eds. Challenges in Taste Chemistry and Biology. Washington, DC: American Chemical Society Chapter 14 Signal Generation & Processing 8- 171

Figure 8.5.4-44 A selected group of stimulants represented as “umami” type in the literature. Note the presence of a carboxyl group in nearly every chemical (some shown more explicitly than others. See text. From Schlichtherle-Cerny et al., 2004

Thus, it is conceivable that mono-sodium glutamate represents a group of chemicals that stimulate more than one of the basic sensory channels. However, there are other chemicals in the group that do not contain either sodium or glutamate, including a long list of inosinates. The inosinates may be particularly liable to associate with the active portion of the sensory receptor of the N-Path channel which is believed to be an inositol ligand. Two organic acids are said to be 5-30 times more effective than monosodium L-glutamate in eliciting the umami response.

In the case of lactic acid, note the obvious presence of two gustaphores, the A-Path gustaphore (carboxylic acid) and the G-path gustaphore (HO–C–C–OH).

Figure 8.5.4-45 indicates the character of most stimulants described previously as perceived as umami. They are in fact gustants containing multiple gustaphores. The more complex gustants in the previous figure are not shown to avoid complexity. All of the gustants shown exhibit two or more gustaphores. The resulting perception is similar to magenta in vision; it requires the stimulation of multiple gustaphores of well known characteristics. All of the gustants shown exhibit an acidophore and all exhibit a glycophore. Only MSG and aspartic acid exhibit a natrophore. The hydrated sodium ion of MSG theoretically exhibits six natrophores. The nucleotides at upper right in the previous figure have not been analysed with respect to their gustaphores. Selected hand calculations suggest they are not important gustaphores. Most of them have d-values interdigitated with the GR’s of gustation. 172 Neurons & the Nervous System

Nucleotides are three component systems consisting of (1) a nitrogenous base, (2) a five carbon sugar, and (3) phosphoric acid. If the phosphoric acid is removed by partial hydrolysis, the remaining components are known as nucleosides. While ubiquitous in the discussions of this work, the phosphate group does not participate in gustation. The d-values between singly bonded oxygen orbitals is 2.63 Angstrom. The d-value between a doubly bonded oxygen and a singly bonded oxygen is 2.52 Angstrom. These values appear in the notch between the A-Path and G-path GR’s. Their limited perceived taste helps define the acceptability profile of these two GR’s. The d-values associated with the five- carbon sugar are also interdigitated between the d-values of the A-Path and G-path and the N-path and P-path.

Exploring the analogy with the visual system, two situations arise; the first where the gustant exhibits gustaphores affecting adjacent GR’s on the d-value line. This situation is analogous to the perception of yellow caused by the presence of a red and a green spectral component simultaneously. Lactic acid is a clear example of this situation. The second situation arises when the gustant exhibits two gustaphores that are not stimulating adjacent GR’s. The result is analogous to the perception of magenta Figure 8.5.4-45Stimulants associated with umami caused by the presence of a red and a blue due to their multiple gustaphores EDIT TOP LINE of spectral component simultaneously. No the four fundamental types. Frame A; the example of this situation appears in the acceptance range of the four fundamental GR’s previous two figures. of taste. Other frames as labeled. The sodium ion is highly hydrated in the example of frame B. The It is noteworthy that aspartic acid contains number next to some columns indicate the four distinct gustaphore types, one type potential number of gustaphores of that type stimulating each of the fundamental GR’s. present. All gustaphores are shown at unity It in fact has two copies of the acidophore, relative effectivity. carboxylic acid.. Danilova et al. (2001, page 984) identified a unique M-cluster that was most closely associated with MSG but included other stimulants containing at least two gustaphores associated with the N-path, two with the A-Path and one with the G-path. It was clearly not due to a single new and unique gustaphore. The M-cluster was analogous to white in color vision. It is perceived as off of the d-value line and it involved stimulation of multiple GR’s simultaneously. 8.5.4.10 The putative “non esterified fat” sensory response Signal Generation & Processing 8- 173 In 2015, Running et al. provided a paper suggesting that fats might be a separate class of gustants235. Their Abstract opens with, “Considerable mechanistic data indicate there may be a sixth basic taste: fat. However, evidence demonstrating that the sensation of nonesterified fatty acids (NEFA, the proposed stimuli for “fat taste”) differs qualitatively from other tastes is lacking. Using perceptual mapping, we demonstrate that medium and long-chain NEFA have a taste sensation that is distinct from other basic tastes (sweet, sour, salty, and bitter). Although some overlap was observed between these NEFA and umami taste, this overlap is likely due to unfamiliarity with umami sensations rather than true similarity.” Running et al. state in their introduction, “However, documentation that oral NEFA exposure elicits a perceptible and unique taste sensation, in addition to their olfactory and somatosensory sensations, is weak overall and absent in humans. Studies in rodent models indicate that taste aversions to nutritive oil and long chain fatty acids do not generalize to other taste sensations or to textural qualities (Pittman 2010), suggesting the sensation is unique in this species.” Their conclusion is even more damaging to their null hypothesis. “Notably, the taste sensation elicited by long chain fatty acids is not wholly consistent with the expectations of “fattiness.” Given the clear unpleasantness of the sensation in isolation, and the incongruity with the term “fatty,” which has strong textural context, we propose a new term to describe the taste of long chain NEFA.”

The nomenclature between the food community and the general organic chemistry community must be clarified. Their Table 1 lists a broad range of fatty acids, sugars and other test materials accompanied by the concentrations of each. The table does not list any “fats,” fatty acids esterified to glycerol 236. Their fatty acids are typically considered derivatives of the carboxylic acid family in the larger field of organic chemistry. The derivatives listed are generally unsaturated fatty acids.

Lenninger defines the major component of depot, or storage, fats in plant and animal cells (1972, page 192) as a neutral fat, or glyceride, i.e., a fatty acid ester of the alcohol glycerol. Running et al. define their “fats” as the term is used in the food science and food industry, specifically as non-esterified fatty acids. Thus, they are actually talking specifically about the fatty acids derived from the carboxylic acid family and not about fats per se. The individual fatty acids can be obtained from fats by hydrolysis. Whether hydrolysis occurs in a given situation depends very greatly on the solubility of the glyceride ester (fat) and the temperature (and duration of the temperature profile). Cooking has a major impact on the solubility and digestibility of fat. To the extent the fat is hydrolyzed, the investigator is no longer discussing a fat. He/She is discussing a fatty acid, a totally different chemical substance. When hydrolyzed, the resultant fatty acid is a NEFA by definition. The more important question is whether it is saturated or unsaturated.

The fatty acids are discussed in Section 8.5.4.3. Fully saturated carboxylic acid derivatives exhibit only one gustaphore, that of the carboxylic group with a nominal d-value of 2.268 Angstrom. They are perceived as acidic under the hypothesis of this work. If the derivative is not saturated, the molecule will exhibit additional gustaphores based on the distance between the C=C double bond(s) and the orbitals of the carboxylic group. As an example, oleic acid_393217 contains one double bond and exhibits d-values of 10.759, 11.735 and 2.079 Angstrom.. Linoleic acid_4444105 exhibits two double bonds (separated by three trans- carbon atoms) and its d-values are 10.739, 11.711, 2.078, 11.777, 10.892 and 3.257 Angstrom. As developed in Section 8.5.1.6, d-values greater than about 6 Angstrom are not effective in stimulating the gustatory modality but Section 8.6 will show they are significant in stimulating the olfactory and/or oskonatory modalities. Thus, both oleic and linoleic acids will produce a taste perception of acidic and linoleic will also produce a taste perception of salty. The relative intensity of the acidic and salty perceptions is not documented. An unsturated fatty acid with more than three carbons between any pair of

235Running, C. Craig, B. & Mattes, R. (2015) Oleogustus: The Unique Taste of Fat Chem Senses doi:10.1093/chemse/bjv036

236Gunstone, F. (1955) The component acids of chimpanzee fat Biochem J. vol 59, pp 454-457 174 Neurons & the Nervous System orbitals may cause an olfactory perception but will not cause a significant gustatory perception. Depending on the volatility of the material, it may be necessary to consider stimulation of the olfactory epithelium via retrograde nasal entry of the material.

There is no reason to believe Running et al. have defined or isolated a new gustatory channel or mechanism. 8.5.4.11 The mints as nocents instead of gustants

There has always been a debate concerning whether the mints constitute gustants or are in fact nocents. Mentha (also known as mint, from Greek míntha, is a genus of plants in the family Lamiaceae. The essential oils in the leaves of these plants are from 40 to 90% menthol with the remainder dominated by carvone, limonene and other biologically sensitive chemicals. The family consists of at least 20 members and at least 10 hybrids of commercial interest. Spearmint essential oil includes naturally occurring carvone and limonene. In the last lesson related to the R,S system in a discussion of chirality, the Khan Academy notes that ( R) carvone smells like spearmint while (S) carvone smells like caraway seed. (R)-(-)carvone_388655 exhibits d-values of 4.943, 4.461 & 2.951 Angstrom. (+)-(S) carvone exhibits d–values of 5.347, 4.278 & 2.950 Angstrom. The 3D representation of these isomers demonstrates they are distinctly different odorophores from the olfactory perspective.

Menthol is not currently addressed as an odorant in the theory of this work (see Section 8.6.3.3.1). It is grouped with camphor and eucalyptol as more complex molecular shapes not amenable with the DACB theory of odorants.

This investigator had a recent encounter with a commercial product known as Altoids and sold in the candy section of most food stores. Altoids mints are currently available in four flavors: peppermint, wintergreen, spearmint, and cinnamon. The variant of this product labeled peppermint contained sorbitol and natural flavors as their principle ingredients according to its label. To this investigator, this product was tasteless and odor-free but had a significant nocent properties when placed on the tongue. According to Wikipedia, sorbitol_5576, also known as glucitol, is a sugar alcohol with a sweet taste which the human body metabolizes slowly. It is frequently considered a non-caloric sweetener. Mannitol_6015 is a closely related sugar alcohol.

I infer that the natural flavors were dominated by menthol and that menthol was perceived as very cold but not perceptually sweet.

See Section 8.6.6.3 for a more global discussion of the role of the mints and menthol. Menthol exists in a variety of isomers (Section 8.6.13.1). 8.5.5 The vernier (intensity) operation of the gustatory modality

This section will only discuss the output of individual stage 1 sensory neurons as a result of gustatory stimulation. Under a variety of conditions, it is necessary to consider both the area and accessibility of the stage 0 stimulus presentation (effective capture area of each type of GR) and the stage 2 signal processing (potential forms of signal summation) before attempting to relate perceived tastes to real stimulants. Both of these aspects are discussed in the next chapter of this work. [xxx temporarily housed in Section 8.10 below the end mark of pt 2 for this chapter] It will also be shown in that later material that the long-standing debate relative to the labeled-line versus the across-neuron (nerve) pattern can be satisfactorily resolved based on additional clarity in the details being discussed. 8.5.5.1 Background

The head groups of the phospholipids forming the sensory receptors are known to be highly polarized, thereby presenting a large dipole electrostatic potential (DEP) to the input terminal of the first Activa within each sensory neuron relative to the potential of the gustatory cavity. The Signal Generation & Processing 8- 175 dual coordinate chemical bond formed between the individual gustaphore and the sensory receptor causes a significant change (and sometimes a drastic change in the case of the “super” stimulants) in this potential. After this introductory material, the abbreviation DEP will be shortened to DP. This area of chemistry has not been widely studied until recently. The current method of study is primarily one of mathematical analysis of the molecular structure of selected chemicals. The process is very computationally demanding even on modern computers. The computational cost of these calculations limit their pursuit. Venanzi & Venanzi and colleagues have been a principle group pursuing the calculations of the more general molecular electrostatic potentials (MEP) of selected tastants and gustaphores237,238,239,240,241. Their analyses related to amiloride contribute greatly to the hypotheses presented here both for the basic vernier change in the DEP generated by a typical tastant, and also an understanding of how the “super” tastants influence the sensory receptors. Amiloride has been implicated in both sodium transport through the cell wall in the chemical theory of the neuron, and as an antagonist of the N-Path sensory receptor. Only the role of amiloride as an antagonist in gustation is of interest here. The 1988a paper provides the geometry and potentials related to acesulfame, a relatively potent sweetner structurally related to saccharine and containing the same SO2 and N–C=O arrangements. In fact these structures are present in a six sided ring in acesulfame and a five sided ring in saccharin. . The 1988b paper explores a group of analogues of perillartine, many containing a cyclohexadiene ring. Many MEP maps are provided. Their discussion opens with;

“The electrostatic potential maps for the perillartine analogues, especially the most potent ones, yield information on the electronic features that determine the recognition pattern of these molecules. This recognition pattern can be used for understanding not only the activity of the analogues in this study but also the activity of other structurally similar analogues. In addition, from an analysis of all the maps at the two different distances, some conclusions can be drawn about the preliminary stages of the receptor-analogue interaction.”

From an analysis of the electrostatic potential contour maps of the perillartine analogues, the importance of two negative electrostatic potential regions emerges: (1) a broad region near the nitrogen and oxygen atoms of the oxime group, which reflects the effect of the lone pairs on each of these atoms and does not generally vary in its position or depth from analogue to analogue, and (2) a second region in the hydrocarbon section of the analogue; a region that not only varies in its depth, shape, and extension but also in its orientation with respect to the region identified with the oxime moiety. The depth and extent of this negative electrostatic region as well as its orientation with respect to the potential region around the oxime moiety appear to be crucial for receptor recognition and, thus, are significant indicators of taste potency. This result emphasizes the importance of the hydrocarbon domain: a point recently underscored by the structure-activity data and first made, in a different context, by Kier.

237Venanzi, T. & Venanzi, C. (1988a) A molecular electrostatic-potential study of acesulfame Analytica Chimica Acta vol 210 pp 213-218

238Venanzi, C. & Venanzi, T. (1988b) Ab initio molecular electrostatic potentials of perillartine analogues: Implications for sweet-taste receptor recognition J Med Chem vol 31(10) pp1879-1885

239Venanzi, C. & Venanzi, T. (1992) Molecular recognition of amiloride analogs: a molecular electrostatic potential analysis. 1. Pyrazine ring modifications J Med Chem vol 35 pp 1643-1649

240Venanzi, C. & Venanzi, T. (1993) MEP of side chains In Simon, S. & Roper, S. eds. Mechanism of Taste Transduction. Boca Raton, FL: CRC Press pp 427-462.Chap. 15

241Buono, R. Venanzi, T. Zauhar et al. (1994) Molecular dynamics and static solvation studies of amiloride J xxx pp xxx 176 Neurons & the Nervous System

They go on later; “Although the recognition model is used here as an indicator of biological activity, a complete description of the taste potency must involve other factors among which are conformational dynamics, polarization, steric, and hydrophobic effects.” The 1992 paper shows the dual coordinate chemical bond between several of their molecules. However, these bonds relate to an artificial formic acid with d-value= 1.8Angstrom and do not relate to the d-value of 3.3 required to bond to the N-Path receptor of this work. The paper contains extensive citations. The dual bond suggested by the artwork in this paper is a parallel rather than antiparallel bond. Closer 3D examination of the many electron pairs on nitrogen and oxygen atoms in amiloride, Figure 8.5.5-1, may uncover a potential coordinate bond of the appropriate spacing to block the N-Path receptors. While largely planar, amiloride is not planar in the important areas associated with blocking the N-Path receptor. Using the Jmol representation in ChemSpider, amiloride exhibits d-values of 2.196, 2.956, 3.586 and 3.601 Angstrom among others. The d-values when in solution have not been found in the literature. Their 1993 paper addressed the effect of side chains in the MEP of a tastant. Their focus on amiloride in the 1994 paper is not easily extended to other stimulants/depressants of interest with the needed precision. Like the 1992 paper, the figures provide only Fischer (planar) representations of very complex molecules. This paper is focused on the MEP of their molecules in solution. Overall, these papers lack a framework pointing to a specific null hypothesis underlying their exploratory research. Figure 8.5.5-1 The many lone pairs of electrons in amiloride available for dual coordinate bonding. Culbertson & Walters (1991) discussed a 3D The auxiliary lines indicate the distinct planar model of a sweet channel receptor that was structures in the figure. deduced from the effectiveness of a variety of artificial sweeteners. They described their model primarily from a stereographic perspective using a very conceptual caricature. 8.5.5.1.1 Dipole potential and related dipole moment

A critical aspect of intensity sensing in gustation involves measuring the dipole potential of a stimulant when in a DACB relationship with the appropriate GR. While the dipole potential of interest is related to the dipole moment of a stimulant in free space, the value of interest here is an integrated value when in solution and relative to the DACB relationship. It is the dipole potential presented to the GR at the DACB while the stimulant is in solution that is of fundamental interest. Narasimhan has presented some basic research on the dipole moments of a variety of saturated dicarboxylic (Lewis) acids of interest in gustation242. He focused considerable attention on the variation in dipole moments associated with various rotations around single carbon-coarbon bonds. He used dioxan as an aid to disassociation of these molecules in dilute solution.

242Narasimhan, P. (1957) Dipole moments of saturated dicarboxylic acids Signal Generation & Processing 8- 177

Venanzi, Buono et al. looked at additional aspects of the MEP of amiloride (a complex molecule “known to be a potent inhibitor of sodium transport in a variety of cellular and epithelial transport systems”) in 1995243. Figure 8.5.5-2 shows a rearrangement of a Venanzi & Buono et al. figure. The left frame shows amiloride in an uncombined configuration as exists in in the gas phase. The field lines are relatively conforming to the shape of the molecule. The conventional dipole moment, and dipole potential are derived from this configuration. The right frame shows the same molecule in a potential “encounter” with acetic acid. These two entities are not likely to form a DACB bond due to the difference in d-values of their relevant orbitals. Their paper did not address how amiloride might couple with sodium or a putative sodium receptor.

Figure 8.5.5-2 Computed molecular electrostatic potential maps of amiloride. Left; amiloride “in vacuum.” Right; amiloride in an encounter with acetic acid. The encounter involves the guanidine group of amiloride. Neither Venanzi & Buono or this work indicates an actual coordinate bond or bonds between the amiloride and acetic acid. Contours in units of kcal/mol. See text and original captions. Rearrangement of Venanzi, Buono et al., 1995.

A zero potential line is shown passing through the space occupied by the hydrogen bonds. Not shown is the presence of water surrounding these two structures in an overall 3D configuration. The presence of water would cause the zero potential line to curve around the amiloride in a closed 3D envelope. The parameter of interest here is the dipole potential (DP) between the midpoint of the DACB and the water environment surrounding the amiloride. It is proposed that this dipole potential is sensed by the sensory receptor in step 2 of the transduction mechanism of chemical sensing. In their amiloride forms 1, 18 & 19, similar to the case illustrated above, amiloride exhibits d-values of 3.586 Angstrom between their N1 and N12 and their N1 and N9. These gustaphores are more likely to couple to the sensory receptor of the N-Path and thereby block hydrated sodium from stimulating this receptor. Their chlorination of these forms may bring the d-values even closer to the 3.3 Angstrom value. Alternately, they could be coupling to a hydrated sodium entity and thereby blocking one of the potential DACB opportunities of the hydrated sodium to couple to the sensory receptor. Both of these possibilities are discussed in the Venanzi and colleagues papers. Since, hydrated sodium exhibits several sites of DACB coupling, it is suggested that the

243Venanzi, C. Buono, R. et al. (1995) From Maps to Models: A concerted computational approach to analysis of the structure-activity relationships of amiloride analogues In Reynolds, C., et al.; Computer-Aided Molecular Design. Washington, DC: American Chemical Society ACS Symposium Series, vol 589 178 Neurons & the Nervous System concentration of amiloride in a solution would play an important role in the ability of hydrated sodium to stimulate a sensory receptor. Venanzi, Buono et al. developed a “pharmacophore hypothesis” to identify a molecule that could form a stable blocking complex with an undefined sodium channel, based on certain assumptions concerning the structure of amiloride. Their focus was not on the DACB coupling but on substitutions in the the pyrazine ring. They note, “Although this approach has some features in common with the Active Analogue Approach of Maisha, it is, in contrast, not a traditional QSAR analysis.” Their pharmacophore does not relate to amiloride as a gustant with an appropriate gustaphore (the strongly basic guanidine group) as defined in this work. Venanzi, Buono et al. have noted two relevant forms of amiloride, “The Al free base conformer (OCCN=180/) was found to be more stable than the A4 (OCCN=0/) in solution. This sheds light on earlier NMR studies which were unable to distinguish between the two conformers in solution.” Venanzi, Buono et al. noted, “In order to investigate these issues further, in the absence of explicit knowledge of the molecular structure of the ion channel, we are using the technique of stereolithography to build plastic models of the global energy minimum conformers of amiloride analogues and of their complementary molecular shapes.” It will be proposed below that the effective dipole potential of the combination of the GR/gustaphore combination can present a very significantly different DP to the 1st amplifier of the sensory neuron as a result of the dispersion process associated with the AH,B,X relationship.

The dipole potential of the homologous members of a stimulant family may vary significantly being the dipole moment (measured in Debye and taken as the most easily measured parameter) is the product of the dipole potential (measured in electrostatic volts) and the size of the stimulant molecule (measured in an effective length between the charges required to account for the net dipole potential). It can also vary significantly among closely related derivatives. Methanol CH3OH, or CH4O, has a significant dipole moment of 1.71 Debye, Methanal (formaldehyde), CH2O has a value of 2.33. CH3NO2 has a large value of 3.5 Debye

8.5.5.1.2 Dipole potential calculation

The calculation of dipole potentials based on computerization is in a rapidly expanding, but still primitive state. The available calculations are primarily designed for stand-alone representations and do not normally calculate the dipole potential between one protected area on the surface of the molecule and the integrated surface of the remainder of the molecule immersed in a fluid (typically mucos, saliva or water).

The computational field appears to be centered on a program called PyMOL and an add-on called APBS244. PyMOL is a program based on the Python language and copyright 2004; “PyMOL is one lone scientist's answer to the frustration he encountered with existing visualization and modeling software as a practicing computational scientist.” APBS -- Adaptive Poisson-Boltzmann Solver. Quoting from the APBS User Guide/package-overview; “The APBS sub-directory examples contains several test systems which show how to use APBS for binding energy, solvation energy, and force calculations. The file examples/README.html contains descriptions of the test cases and links to anticipated results. Examples can be run and compared to expected results by running make test in each example directory. Additional examples are provided as part of the APBS tutorial (doc/html/tutorial/), described in more detail in the Documentation section.”

244http://www.poissonboltzmann.org/apbs/user-guide Signal Generation & Processing 8- 179

The APBS program is moving rapidly toward a Windows compatible graphic interface but its utility is still beyond this author’s capability at this time (11/2013). Much of its utility relies upon manipulating binary code packages and preparing large data files in specific formats. It can accept .PDB files from the Protein Data Bank. The PDB is focused on much larger molecules that of interest in biological chemical sensing. The PDB data file must be converted to a PQR file using the translator PDB2PQR, with several specified caveats. 8.5.5.1.3 Molecular electrostatic potential profiles

Immel245 has provided a dissertation on the molecular electrostatic potential (MEP) of a variety of sugars. The dissertation is extensive, as indicated in the citation, and can contribute to the hypothesis of this work. Only chapters 2, 3 & 4 out of 12 will be discussed here; the later chapters deal with ever more complex sugars. The material is important because it calculates and displays a 3D MEP that could be instrumental in calculating the dipole potential in this work. The dipole potential results from the integration of the MEP of the molecule with respect to the location of the DACB employed during coupling to the phospholipid of the sensory receptor. Chapter 2 is extensive and develops the detailed 3D structure of a variety of saccharides along with their MEPs. At one point (page 41), he notes, “Semiempirical calculations of other conformations of 2 [ß-D-fructose, ed.] are encumbered with the fact that the minimum energy geometries generated represent the state in vacuo, which may substantially be altered on solvation with water. This applies to the conformations emerging from very elaborate ab initio calculations and AM1-based semiempirical investigations, as well as to those emanating from the more simple PIMM88 force field methodology.” He concludes with, “In summary, much remains to be learned about the intricacies of the mechanism(s) involved in activation of sweet-sensitive cells, and direct solid evidence is urgently required.”

Chapter 3 of Immel opens with an important observation, “Sucrose and other sugars are over-functionalized with hydroxyl groups of almost identical properties, such that predictions of their relative chemical reactivities are highly speculative. Especially difficult is the assignment of relevant local electrostatic and hydrophobic molecular properties which determine their chemical and biological behavior. In view of the availability of modern molecular modeling techniques, the MOLCAD-program in particular, it deemed opportune to probe this methodology with a structure as complex as sucrose. In this chapter, the computational basics for generating molecular electrostatic potential profiles and lipophilicity patterns (MEP's and MLP's) are detailed and the far-reaching chemical implications and biological significance is discussed for sucrose with the aim to gain further insights into hydroxyl group reactivities and the mechanism by which sweetness is elicited.”

Within the context of gustation, the sugars are not over-functionalized because only specific conformational pairs of hydroxyl groups can participate in DACB with the GR 2 (G- Path) receptors.

His figure 3-1 stresses the importance of employing 3D representations of the sugars rather than the more convenient Haworth representations. It also shows the location of several hydrogen bonds in sucrose coupling the fructose and glucose moieties and preventing the participation of these hydroxyl groups in forming DACB with the sensory receptors. He has also analyzed the conformation of sucrose in other solvents than water. In this chapter, Immel makes the assumption that sucrose does not hydrolyze into fructose and glucose. Therefore, he focuses on how sucrose might bind to a receptor on strictly stereographic grounds using caricatures. Immel’s conclusions relative to sucrose sensing are modest and do not consider the electrostatic potentials of the combined phospholipid and monosaccharide structure resulting from the DACB.

245Immel, S. (1995) Computer Simulation of Chemical and Biological Properties of Saccharides: Sucrose, Fructose, Cyclodextrins, and Starch http://csi.chemie.tu-darmstadt.de/ak/immel/script/redirect.cgi?filename=http://csi.chemie.tu-darmstadt.de/ak /immel/publications/phdthesis/ Full text and index. Chapter 4; http://csi.chemie.tu-darmstadt.de/ak/immel/publications/phdthesis/chapter04.pdf 180 Neurons & the Nervous System

Chapter 4 is devoted to the calculation of molecular electrostatic potentials (MEP) of various sugars and assumes the AH-B-X concept. He notes, “The first rationalization of structure-sweetness relationships by Shallenberger and Kier presumes the existence of a common AH-B-X glucophore in all sweet substances, eliciting the sweet response via interaction with a complementary hydrogen bond donor and acceptor functionality and a hydrophobic site in the taste receptor. This very simple theory, also termed the "sweetness triangle",appears much too simple to explain all of the observations at the present state of knowledge, particularly when bearing in mind that sweet taste chemoreception is mediated by a cascade of complex biochemical processes that are little understood at the cellular and molecular level.” This work has expanded on the work of Shallenberger and Kier to overcome the “simplicity” objection, specifically as it relates to their 2D calculations and molecular representations. At the time of his dissertation, Immel noted, “It is unlikely that progress can be made on this rather speculative level by further reflections on the data presently available. If at all, it is clear that the three-dimensional molecular shape of the sweet substrates and their respective physico-chemical properties must be taken into consideration to mold a more precise overall picture of the substrate-receptor-interactions involved in sweetness. Starting from these realizations the results obtained from calculation of the molecular electrostatic potential profiles and hydrophobicity patterns of the ß-pyranoid form of D-fructose should be discussed in the sequel.” Immel provides little tabular results in Chapter 4. However, he does provide a series of Haworth diagrams with specific comments about the sweetness of various similar monosaccharide forms. The comments have not been fully correlated with this work to date. However, this can be done by calculating the d-values for the various conformations.

Immel also notes, “The most striking shortcomings of the proposals (i) and (ii) in Fig. 2-14 for the1,2-glucophore is seemingly the dramatic difference between fructose and sorbose, being initiated by simple inversion of the remote 5-OH group. The proposed hydrogen bond releasing effect of 5-OH, liberating the anomeric hydroxyl group 2-OH for interaction with the receptor by competitive intramolecular hydrogen bonding with the ring oxygen – being possible in fructose, but not in sorbose – was used to explain the taste difference of these two sugars. Due to the nearly unaltered sweetness of the 5-deoxyderivative 8 in relation to fructose, this explanation seems untenable.” The mere inversion of the 5-OH group is key to the sweetness of the monosaccharide in the context of this work! He continues, “The taste properties of sorbose are more readily rationalized, assuming the validity of Birch's assignment (iii, Fig. 2-14): a change in the steric and electronic characteristics in direct vicinity of the 3,4-diol system would change the sweetness quality substantially. [Fig. 2-14 describes conformations of fructose using Haworth diagrams, ed.] The same effect is observed in the case of sucrose – in which the 2,3-diol unit was assigned as the glucophoric moiety – when inverting the configuration at C-4 to "galacto-sucrose", sweetness is almost completely lost. Similarly, going from glucose (in which the 3- and 4-OH groups seem to represent the AH-B unit) to the 2-epimeric mannose, sweetness changes substantially.” Immel concludes his discussion with, “Although further evidence is required to settle this question unequivocally, as of now, major significance is attributed to the MLP's obtained for the two fructose conformers likely to prevail in solution: these(Fig. 4-3) clearly favor Birch's[101]proposal (iii in Fig. 2-14), which places the AH-B couple of the glucophore into the 3,4-diol grouping of fructose.” Note the use of the terms 1,2-glucophore and 3,4-diol in the above discussion. He notes in Chapter 2, “ the hydrophilic portions of these sweeteners are more compact, invariably located opposite to the hydrophobic region, and appear to contain the AH-B couple of the glucophore: the glucosyl-2- and 3-OH group in sucrose and sucralose, versus the 3,4-diol grouping in fructose.” It appears further study of the Immel dissertation is warranted if further confirmation of the hypothesis of this work is required. This would involve obtaining and familiarization with some of the software packages he used, and specifically the writing of routines for obtaining the dipole moment from spherical representations of the MEP shown in his dissertation. Immel has gone on to publish a long list of follow-up papers based on his dissertation that are listed on the dissertation website. 8.5.5.2 Analog intensity variation due to gustaphores

After selective dual coordinate binding of a gustaphore to the appropriate sensory receptor, there remains the task of perceiving the difference in efficacy between tastants involving the Signal Generation & Processing 8- 181 same gustaphore. This perception is supported by the electrostatic potential changes to the sensory receptor brought on by the presence of the bound gustaphore as described in Figure 8.5.5-3. More detailed circuit descriptions will be presented after some introductory material. 182 Neurons & the Nervous System

Figure 8.5.5-3 The mechanism(s) of intensity determination (including dispersion) in gustation. Top; circuit schematic of transduction elements. Bottom; caricatures expanded to show the proposed layout of the receptors and tastants from an electrostatic potential perspective. The single plate of a capacitor associated with each receptor is discussed in the text. Bottom left; quiescent state of sensory receptor. Bottom middle; state of circuit when stimulated by a nominal tastant. Bottom right; state of circuit when stimulated by a “super tastant,” a molecule exhibiting a single gustaphore and an exposed region of high electrostatic potential appropriately positioned to satisfy the AH,B,X criteria. See text. Signal Generation & Processing 8- 183 The mechanism relies upon the analog signal sensitivity of the first (adaptation) Activa of each sensory neuron. This mechanism has been described in detail in Chapter 2, Sections 2.2.3 & 2.2.4 and is also discussed in Section 8.5.1.3. The upper left panel shows the basic circuit of a gustaphore specific path beginning with the binding of a stimulant to a receptor ligand by way of the gustaphore/sensory receptor bond. This bond causes a change in the electrostatic potential at the base terminal (B) of the 1st (adaptation) Activa of the sensory neuron from its quiescent potential. This change is amplified by the 1st Activa and results in a current flowing through the emitter terminal (E, arrow on lead at lower left of upper frame). The lower three frames illustrate the individual states of the circuit as indicated. The phospholipid body shown in the upper panel is a very high impedance circuit resulting from a reverse biased diode formed by the tail of the phospholipid. Because of its impedance, very little current is required to flow through this element in order to generate a significant change in potential at the base terminal (B). The head group of the phospholipids identified as the sensory receptors in gustation are known to exhibit a very large dipole potential between their bonding terminal and their connection to the tail of the phospholipid. As a general rule, the change in the electrostatic potential presented at (B) is small relative to this large dipole potential. However, in the case of super sweeteners and super picric gustaphores, the change in the dipole potential of the receptor ligand is much larger (accounting for the high efficacy of these materials). See Section 8.5.10.

The lower left panel illustrates the quiescent state of the sensory receptor and 1st amplifier in the absence of any selected gustaphore. The receptor ligand exhibits its nominal large dipole potential between its exposed terminal in the lingual cavity and its termination at the phospholipid tail structure. This potential is passed to the other end of the tail structure and results in an electrostatic potential, labeled the quiescent voltage (Vq) measurable as shown with adequate instrumentation. This potential causes only a small current to flow through the 1st Activa and into the dendroplasm of the sensory neuron.

The lower middle panel illustrates the situation when a specific gustaphore of a tastant becomes bound to the sensory receptor. This tastant exhibits a “dipole potential” described as the potential of its AB, H structure relative to the integrated MEP for the rest of the tastant exposed to the saliva.

The distributed charge spread over the surface of the three-dimensional tastant molecule, its molecular electrostatic potential, MEP, makes it awkward to speak of its dipole potential. However, its effective dipole potential is the integrated potential of the entire surface exposed to the saliva relative to the potential of its region involved in coordinate bonding to the receptor.

The dipole potential of the tastant is usually small relative to the potential provided by the receptor ligand. It causes the potential at terminal B to rise, causing greater current flow through the 1st Activa into the dendroplasm. This increased current is amplified further in the 2nd Activa of the sensory receptor and generates a change in potential at the pedicle of the neuron axon. There are typically a large number of 1st Activa associated with an individual 2nd Activa. As a result, many small currents are summed to create the signal used to drive the 2nd Activa. This summation of currents causes a larger change in the pedicle potential of the 2nd Activa. It is this change in potential at the pedicle of an individual sensory neuron that is summed with the signal from other sensory receptors of the same class at higher levels within the neural system in order to create a perception of the intensity of a specific tastant. Calibrating the currents and potentials associated with a specific tastant and/or gustaphore is particularly difficult because of the difficulty of controlling the density per unit area of the tastant applied to a specific sensory neuron sensitive to that specific tastant. Certain tastants (particularly man-made tastants) are known to have very high efficacy relative to a specific gustaphore path (sometimes 10,000 times more effective than a simple gustaphore of the same class. The mechanism employed relies upon a secondary electrostatic potential on the surface of the tastant at a specific location unique to the receptor for that gustaphore path. This secondary electrostatic potential is able to cause a major change in the intrinsically high dipole potential of the receptor ligand. The result is a very large change in the base terminal (B) 184 Neurons & the Nervous System potential compared to that created by the direct binding of the gustaphore to the receptor. This mechanism is illustrated in the lower right panel. In sequence, the tastant must be selected due to its gustaphore forming a dual coordinate bond with the receptor, a nominal change in the base terminal potential is caused by the presence of the tastant in the signal path, and a much larger change in the intrinsic dipole potential of the receptor ligand is caused by the close proximity of the secondary electrostatic potential on the tastant to a similar (or complimentary) potential feature on the receptor. See Section 8.5.4.2 for a broader discussion. [xxx odorophore in this paragraph?? ] How to precisely model the actions involved with the “super tastants” is unclear. A model based strictly on electrostatics appears useful. In that case, the tastant can be modeled as a conventional single gustaphore of dipole potential, Vg, supported by an auxiliary potential, Vsg, presenting a charged area on the surface of the tastant as shown; and the sensory receptor can be modeled as two large capacity Leyden Jars connected in opposition and resulting in a small quiescent potential, Vq. The connection between the two electrostatic sources is supported by the plate of a capacitor as shown on the right. When the odorophore binds with the receptor, its small dipole potential causes the net potential, Vn, to become slightly more positive. However, the introduction of the additional charge associated with the super tastant in the vicinity of the capacitor plate of the receptor and at a small distance, X, introduces the possibility of changing the dipole potential of one half of the receptor significantly. The resulting small percentage change in the dipole potential of that half of the receptor can generate a very large percentage change in the net potential, Vs, because of the small change introduced into the difference between two large quantities. The result is a very large change, Vs – Vq, for the application of only a small amount of tastant, the hallmark of a super tastant.

8.5.5.2.1 Two distinct response–concentration characteristics for sweeteners

DuBois et al. have addressed the perceptual response of humans to variations in concentration of gustants consisting of both simple sugars and artificial sweeteners246. They find that these two classes exhibit different response vs concentration characteristics. The implication is that they employ different means of creating either the AH,B or AH,B,X relationship associated with their DACB. The difference appears to highlight the potential for aftertastes among the artificial sweeteners. It also appears likely, as they conclude, “Thus it is tentatively concluded that at least two routes to receptor cell activation ust exist.” Recognizing that some of their simple sweeteners are mono–, di– and poly–saccharides may be significant when analyzing this data. They discuss the Beidler equation in multiple forms. Changing from % concentration to ppm concentration complicates the comparisons.

The application of the Beidler equation involves a number of variables besides those explicitly included in the formula. 8.5.5.2.2 Structure of simple artificial and Anti-sweeteners

Piisola has provided an overview of non-sucrose based sweeteners for the student versed in organic chemistry structures247. Figure 8.5.5-4 shows the dimensions of xylitol, one of the simplest of the sugar-alcohols. [xxx no figure in aliphatic or ring form. In ring form it will probably resemble inositol. ]

246Dubois, G. Walters, D. Schiffman, S. et al. (1991) Concentration–Response relationships of sweeteners In Walters, D. Orthoefer, F. & DuBois , G. eds. Sweeteners. Wash. DC: American Chemical Society Chapter 20

247Pissola, A. (20014)Sugar Substitutes: The Chemistry of Synthetic Alternatives to Sucrose https://www.jyu.fi/kemia/tutkimus/orgaaninen/en/research/Pihko/gm/Piisola_Sugar_Substitutes_2014.pdf Signal Generation & Processing 8- 185 Sugar-alcohol is largely a historical term. Xylitol contains no saccharide or otherwise sugar- like structure. It is a highly hydroxylated, although not saturated, pentanol. In its aliphatic form it contains three glycophores with d = 2.87 Angstrom (0.287 nm) giving it a perceived sweetness but remaining non-caloric from a nutrition perspective. Xylitol is commonly thought to form a sugar like ring in solution. It is approximately as sweet as sucrose. This level of sweetness suggests it does not exhibit any super-sweet AH,B,X glycophores. See table in Section 8.5.9.6 for a comparison with other glycophores.

8.5.5.2.3 Structure of super sweeteners–acesulfame & saccharine

[xxx check use of odorophore in text below here. Some material copied from 8.6 ]

Mining the above papers and various textbooks provides additional information on the role of acesulfame and saccharine as artificial sweeteners and “super” sweeteners. It is generally reported that acesulfame, and Figure 8.5.5-4 The structural form of xylitol, a sweet presumably saccharine oxidize slowly in air. alcohol EMPTY. It is also reported that acesulfame combines with a hydroxyl group when solvated. It is not clear whether the reaction involves hydrogen bonding between the pyridine nitrogen and the oxygen or whether the bonding is directly between the nitrogen and the oxygen. Figure 8.5.5-5 suggests the bonding is direct.

This figure provides a totally different representation of these materials as sweeteners. The upper frame shows the materials as described based on dry crystallographic observation. The lower frame shows the same materials in their proposed solvated forms. In the solvated forms, the two chemicals exhibit a potential gustaphore, O=C–N–O very similar to the glycophores affecting the G– sensory path. The open question is the precise d-value for the two oxygen atoms s of these two forms. A d-value of 2.6 ± a yet to be determined tolerance would confirm their efficacy as sweeteners. Note the potentially different angles between the relevant atoms in these six member and five member ring molecules. The structure of the acesulfame is basically planar except for the ring oxygen which is out of plane by 22 degrees. Venanzi & Venanzi (1988a), Fig. 1, Figure 8.5.5-5 Acesulfame and saccharine in dry have provided detailed MEP maps for the and solvated forms. Acesulfames on the left, crystal form of acesulfame computed for saccharines on the right. Upper row; dry form planes parallel to but offset from the determined by crystallography. Lower row; specific structures in (1). The offset was a proposed solvated form. 186 Neurons & the Nervous System requirement for the computational tools available in their time. Unfortunately, their attention was centered on the oxygen atoms surrounding the sulfur atom rather than the glycophore proposed here. - - - - - [xxx discuss DEP in more detail. ] In the context of this work, the important DEP is that between the baseline of the AH,B structure and the portion of the molecule perpendicular to that baseline (for the simple sweetener) and for the DEP of the X atom relative to the AH,B baseline for the super sweetener. - - - - Signal Generation & Processing 8- 187

Figure 8.5.5-6 shows a potential configuration of acesulfame acting as a conventional sweetener and as an enhanced super sweetener. The representation on the left shows the molecule associated with a potassium ion, as usually found to avoid introduction of a natrophore if sodium were used. The chemistry of this molecule is quite complex. However, the hetrocyclic nitrogen can be considered a pyridine moiety with a free pair of electrons as shown. The pyridine nitrogen is considered a tertiary amine since it lacks any associated hydrogen. This structure is known to be sensitive to oxidation. It is commonly found to form pyridine N-oxide (M & B, pg 1088) or as shown on the right in this figure to combine with a hydroxyl group from the solute (M & B, pg 721). A mixed form of representation is used to stress the covalent form of the solvated acesulfame and its ability to support the anticipated dual coordinate bonding to a G–path sensory receptor. If the distance between the pyridine oxygen and the adjacent single covalently bonded oxygen is approximately 2.6 Angstrom, the material will qualify as a glycophore. The angles involving the pyridine nitrogen are remarkably similar to those for a carbon in the same position (M & B, pg 723). The potential of the solvated acesulfame to act as a super sweetener is clear. The potential for the molecule to exhibit a spacing between the AH,B baseline and one of the three oxygen atoms at upper left in the molecule, in accordance with the AH,B,X geometries of Section 8.5.3, is Figure 8.5.5-6 Acesulfame as a conventional and substantial. The required distances are 5.5 super sweetener. Left; crystalline form with Angstrom from the more distance oxygen potassium as a ligand to avoid sodium that would and 3.5 Angstrom from the closer oxygen. introduce a natraphore when hydrated. Right; the proposed solvated ligand See text. A similiar analysis can be applied to the saccharine molecule to explain its role as a sweetener and more specifically a super sweetener. In saccharine, the five member heterocyclic ring, a pyrrole, has similar properties to the pyridine ring and the distances to the two oxygen atoms bound to the sulfur atom are similar.

The artificial sweeteners–Saccharin and Aspartame

[xxx must edit the following paragraphs significantly and introduce aspartame 27_45 and or aspartame 32_88.wpg showing a 2.91 glycophore and possibly still a picrophore. ]

Figure 8.5.5-7 shows the structure of aspartame in its role as a super-sweetener. Aspartame is considered 150-200 times sweeter than sucrose. This is probably due to both its molecular structure and the change it causes in the dipole potential of the sweetness sensory receptor. The figure shows the AH,B,X group as formed of NH,O,O where the O at X is covalent oxygen An alternate AH,B,X group is formed of NH,O,O with the O at X is singly bonded to a carbon and a nitrogen at the lower portion of the figure. The configurations are compared in detail in Section 8.5.9.3 [xxx this call may change ]. 188 Neurons & the Nervous System

Figure 8.5.5-7 Aspartame as a super-sweetener with an angle of 27.45 degrees at X. It can also support an angle of 32.88 degrees. See text.

Aspartame has been found to play an important but controversial role in the sweetening of foods. Its perceived side effects have spawned a cottage industry of de-toxifying programs. The perceived problem is its breakdown into individual amino acids, and potentially methanol, following ingestion after performing its desired gustatory task. Figure 8.5.5-8 shows its principle “supersweet” glycophore is probably formed by the two nitrogens, able to form a pair of AH,B coordinate bond with a distance between them of 2.6 Angstrom and the oxygen shown acting as X. The structure is nearly identical to the O-3, O-4 and O-2 arrangement of galactose in the sweet channel receptor. Signal Generation & Processing 8- 189

Figure 8.5.5-9 shows the reason aspartame can taste bitter. It contains a picrophore that is nearly a perfect structural match to the sensory receptor, 3'-O’aminoacyl glycerol. The reason is both are based on the same ligand, aspartic amino acid. The chemical can also elicit a sour taste due to its carboxyl group. Figure 8.5.5-8 Aspartame showing the axis of its AH,B,X “supersweet” glycophore, the two nitrogen atoms, and the location of the oxygen at the X position. See text.

Aspartame incorporates three distinct gustophores.

- - - -

Figure 8.5.5-9 Aspartame bonding to the picric sensory receptor ADD HYDROGEN BONDS and spacing of 4.2 Angstrom..

The structure of saccharin is relatively simple. However, its role as a sweetener probably introduces known but new chemistry to this discussion. Figure 8.5.5-10 shows the structure of saccharin. Deprotonation of the nitrogen leads to the normal anion of saccharin. Sulfur exhibits 5 shareable electron pairs with four involved in oxygen bonding in this example. 190 Neurons & the Nervous System

The bond lengths and angles relating to saccharin are well documented248,249. Shang et al. also describe the distortions of the molecule relative to common descriptive drawing practice. Like other chemicals, the individual bond lengths can vary by up to ten percent from standard bond lengths in a specific moiety. The spacing between the two oxygen atoms associated with sulfur is within the 7% tolerance (2.42-2.78 Angstrom) of the spacing originally derived from Shallenberger & Acree for sweet channel stimulation. The third oxygen appears too far from the initial pair to represent the X of the AH,B,X relationship. However, the phenol ring appears well placed to satisfy this requirement. The shared electrons of the π bonds are known to provide the necessary field potential. However, the elevation view of the molecule shows the symmetrical arrangement of the AH & B oxygen atoms. As a result the triangle formed between these atoms and the X location is isosceles. The bitter after-taste associated with

Figure 8.5.5-10 The structure of saccharin using standard bond lengths (see text). The spacing between the two oxygen atoms is within the tolerance for a sweet odorophore based on standard bond distances. The triangle involving the phenol as X is also isosceles based on the elevation view of the molecule.

248Zhang, Y. Wang, Y & Yu, H. (1995) Structural properties of o-sulphobenzoimide in complex crystals Cryst Res Technol vol 30(6), pp 831-835

249Matos, M. Miranda, M. Morais, V. & Liebman, J. (2005) Saccharin: a combined experimental and computational thermochemical investigation of a sweetener and sulfonamide Molec Phys vol 103(2–3), 221–228 Signal Generation & Processing 8- 191 saccharin appears to be due to the odorophore formed between the isolated oxygen atom and the phenol with a d-value of about 4.2 Angstrom (Section 8.5.xxx).

8.5.5.2.4 Potential dispersion centroids of super sweeteners

- - - - Figure 8.5.5-11 shows a dimer situation with α-D-galactose representing both the sensory receptor and the stimulant. O-3 and O-4 of each molecule are forming the AH,B coordinate bonds and the four atoms define a plane (like that found in Ferrier).

Figure 8.5.5-11 ED A dimer situation with galactose as both sensory receptor and stimulant. The radii associated with the X element intersect at O-2 of the stimulant but do not intersect at any element of the receptor. Exchanging the centers of the rings results in an intersection near O-6 (or possibly C-6). For illustration only, the distances must be calculated in three-dimensional space. See text. 192 Neurons & the Nervous System

Circles are overlaid on the coordinate oxygen molecules of the stimulant to locate X with radii as estimated by Kier. Their intersection occurs at the nominal location of O-2 but do not intersect near any element associated with the receptor. The circles do not intersect at the methylene C-2 of the stimulant in this projection. By switching the centers of the circles, their intersection occurs near C-6 and O-6 of the stimulant. If O-2, or C-6 or O-6, are significant in the sensation of super sweetness, the coordination mechanism must bring one of these elements into some form of bond with the receptor, or an adjacent receptor molecule of the sensory receptor complex. Figure 8.5.5-12 conceptualizes an alternate coordinate bonding arrangement where the galactose receptor is bonding to an unspecified glycophore. In the case of the glycophore, the electron-pair orbitals O-2, O-3 and O-4 need not be oxygen-based. They can be any of the atoms or other species identified earlier in Section 8.5.3.1.

- - - - - 8.5.5.2.5 The super-bitter (picric) stimulants EMPTY

[xxx show or discuss similar graphics for these materials.

8.5.5.3 Defining the gustatory receptor (standing alone)

Significant information is now available concerning the steric geometry required of the sensory receptors of gustation. The challenge is to ascertain the most likely organic molecules used within the microvilli Figure 8.5.5-12 Concept: A stacked situation with of the sensory neurons that can provide a the sensory receptor galactose interfacing with voltage change required by the Electrolytic the glycophore of an undefined stimulant. A third Theory of the Neuron to elicit a gustatory (but strained) coordinate bond is shown between sensation. The most specific information the O-2 atoms, completing a tripartite union relates to the distance, d, between the two (triangle). electron-pair sharing atoms found to be present in all known gustatory stimulants.

The analyses performed here suggest the simplest steric forms of the gustatory receptor are molecules containing ligands that are dimers of the ligands of the gustaphores, except probably in the case of the sodium sensory receptor. True dimers tend to be planar structures. However, the tripartite form of some glycophores suggest the interaction between the stimulant and the receptor is probably not planar but involves a parallelism between the rings of the stimulant and the receptor to support the juxtaposition of the two X features in the AH,B,X coordinate bonding arrangement. This possibility introduces the question of the stimulant and the receptor being enantiomers, either totally or at least with respect to the elements related to the AH,B,X coordinate bonding. Because of the wide range of stimulants known to elicit a response via a single receptor, it appears likely that only a degree of enantiomerism is required to satisfy the coordinate bonding requirement. Ferrier has described the unit cell of β-D-glucose as consisting of four molecules in an orthorhombic configuration250. The molecules exhibited a complex orientation between each other with hydrogen bonding involving the ring oxygen & O-6 bonded to the adjacent O-3 & O-2

250Ferrier, W. (1960) The crystal structure of β-D-glucose Acta Cryst vol 13, pp678 Signal Generation & Processing 8- 193 respectively. The hydrogen atoms are not shown explicitly. This bonding was planar in character. The O-1 and O-4 oxygen play no role in coordinate bonding within the cell. The dimeric relationships assures an ability of the receptor to provide the necessary steric match to the major stimulants and provides a gradation in the sensation intensity as a function of the dimensional mismatch between the receptor and the less than ideal stimulant geometries. The phospholipids known to be associated with the sensory neurons, if not the actual receptor areas of those neurons, appear to provide this general capability. When present in the form containing one conductive fatty acid chain (probably conjugated), these phospholipids provide a generic receptor capability for each of the major taste sensations and an electrolytic path to the space between the bilayers of the respective lemmas. The literature of fatty acids originally assumed all fatty acid chains in lemma were fully saturated, then were present with a single conjugated link near the middle of the chain and more recently were present with a conjugated bond at every third carbon as in 4,7,10,13,16,19-docosahexenoic acid (Siegel, 1999, pg 49; 2006, pg 35). The structure on page 34 is alleged to be the isoprene subunit in natural polymerization. This material can be polymerized by two paths. One path employs addition polymerization that leads to the squalenes (double bonds at every third carbon). Cholesterol is the most common squalene. The second approach employs condensation polymerization and leads to the terpenes (double bonds at every other carbon) with the release of water. Vitamin A is the quintessential example of a fully conjugated terpene. Additional research is likely to uncover very small amounts of the fully conjugated fatty acids (a double bond at every other carbon according to the isoprene rule) in type 4 lemma dedicated to only the sensory receptor function. C Carboxylophore receptor– PtdSer provides a carboxyl ligand ideally located relative to the surface of the microvilli lemma. The two oxygen atoms of the nonresonant carboxyl group exhibit precisely the same dimension, d = 2.34 Angstrom, as do the acidophores associated with this sensory channel.

The ability of a single sensory receptor to coordinate with both organic acids and amino acids is illustrated in Figure 8.5.5-13.

The serine ligand of PtdSer is able to coordinate bond with other organic acids as well as with the amino acids.

C Natrophore receptor– PtdIns provides a ligand ideally located relative to the surface of the microvilli lemma and is known to associate with sodium in other processes. Its cyclic structure provides the opportunity to provide two electron-sharing atoms positioned to match the required dimension, d = 3.3 Angstrom, closely.

C Glycophore receptor– The variety of cyclic ligands capable of forming phospholipids and also coordinate bonding with the potential sweet and super sweet stimulants is very large. However, only a few of these ligands have been reported as present in sensory neuron tissue. [xxx edit this after extending above paragraph ] Phosphatidyl xxx provides a ring structure capable of coordinate bonding with the Figure 8.5.5-13 The serine ligand coordinate principle tripartite glycophores of the bonding with organic acids and amino acids. sweetness stimulants (and the bipartite glycophores of a much wider range of sweetness stimulants). 194 Neurons & the Nervous System

- - - - - Based on its presence in globosides, it is likely the sensory receptor ligand is a form of galactose. Such a ligand could coordinate bond with a galactose molecule acting as a stimulant in a planar dimer (like found in the natural crystal of glucose) or it could coordinate with a galactose molecule in a more complex three dimensional configuration. Figure 8.5.5-14 shows a dimer situation with α-D-galactose representing both the sensory receptor and the stimulant. O-3 and O-4 of each molecule are forming the AH,B coordinate bonds and the four atoms define a plane (like that found in Ferrier).

Figure 8.5.5-14 EDIT A dimer situation with galactose as both sensory receptor and stimulant. The radii associated with the X element intersect at O-2 of the stimulant but do not intersect at any element of the receptor. Exchanging the centers of the rings results in an intersection near O-6 (or possibly C-6). See text. Signal Generation & Processing 8- 195 C Picrophore receptor– Ptd3'Og provides a long ligand terminated by oxygen and the amine of nitrogen that is ideally located parallel to the surface of the microvilli lemma to coordinate with the picrophores. The dimension, d = 4.2 Angstrom. Van der Heijden addressed the chemical structures related to bitterness (pages 94– 108). However, the treatment is briefer and some confusion appears between sourness and bitterness when discussing the AH,B dimension, d. He asserts the dimension is 0.15 nm [xxx chk ] for bitterness (compared to 0.14 nm for the acidophores) instead of the value of 0.42 nm (4.2 Angstrom) [xxx chk ] for the picrophores developed above. His table VI describes picric acid as the most potent of the picrophores. He summarizes his findings in 1993 as, “Although some progress has been made upon more precise characterization of bitter-receptor sites, the specifications among the various classes of bitter principles are lacking. Even the existence of a bitterness receptor is rather speculative. The relationship between the bitter taste and the shape of molecules is more complicated.” The last sentence indicates the criteria for a bitter structure remained unknown at that time. 196 Neurons & the Nervous System

8.5.5.3.1 The nominal gustatory receptor EDIT

Figure 8.5.5-15 places the galactose-based sensory receptor in context with the sensory neuron. The sensory receptor is esterfied to the electrolytically conductive fatty acid portion of the phospholipid. The molecule normally exhibits a dipole potential. This dipole potential is a function of the overall configuration of the combination of the sensory receptor and any coordinate bonded stimulus. The resulting dipole potential at the bilayer membrane interface can change by δV based on changes in the bond state of the sensory receptor. The smallest change in δV is associated with the bipartite bonding via O-3 and O-4. This bonding is associated with a majority of the natural sweeteners. A super-sweet sensation (per molar concentration of sweetener) is associated with either tripartite bond, O-2, O-3 & O-4 or O-3, O-4 & O-6. Signal Generation & Processing 8- 197

The galactose-based sensory receptor offers the possibility of a “super-super-sweet” sensation resulting from the coordinate bonding to all four of its identified coordinate sites (O-3, O-4, O-2 & O-6) at once. Van der Heijden has supported this assertion (1993, page 79). Van der Heijden (pages 77–91) has attempted to move the understanding of enhanced sweeteners forward based on two papers published in 1985251, 252. He provided a series of figures describing potential sites he labels δ for enhanced sweetness relative to the nominal AH,B sites of sweeteners. He did note the importance of the distance A B rather than H B in describing the baseline feature associated with sweetness. His calculations, based on a modified computer program, STERIMOL, assume the broadest possible definition of the AH,B condition and all possible auxiliary sites within a given sweetener. As a result, many potential auxiliary sites are defined that may or may not contribute to perceived sweetness. He did not address the structural arrangements required to bond maximally with the potential sensory receptors. However, he did note the most effective sweeteners typically had an angle of 125 degrees at the A vertex.

Figure 8.5.5-15 The galactose-based receptor in 8.5.5.3.2 The electrolytic properties of context with the sensory microvilli. A bipartite the receptors coordinate bond (d = 2.6 Angstrom) is the minimal requirement for eliciting a sweet sensation. The phosphatidyl moieties defined above, Extension of the bonding to either tripartite when present in liquid crystalline form configuration elicits a super- sweet sensation. A constitute the receptor areas of the sensory super-super-sweet sensation is possible using a neuron villi. Excellent, although limited data quadripartite bond (O-2, O-3, O-4 & O-6). See is available on the electrolytic (particularly text. the surface potentials and the electron density profiles) characteristics of the materials of interest here. Nearly all of the available data relates to the two insulating and electrically inert, type 1, phospholipids, PtdCho and PtdEtn. However, much of the orientation data related to these moieties should be useful. Hauser253 and Hitchcock et al254. have provided some very early but invaluable information about these two

251Van der Heijden, A. van der Wei, H. & Peer, H. (1985) Structure-activity relationships in sweeteners. I. Chem Sens vol 10(1) pp 57-72

252Van der Heijden, A. van der Wei, H. & Peer, H. (1985) Structure-activity relationships in sweeteners. II. Chem Sens vol 10(1) pp.73-88

253Hauser, H. (1976) Phospholipid model membranes: Demonstration of a structure-activity relationship In Benz, G. ed. Structure-Activity Relationships in Chemoreception London: Information Retrieval Limited pp 13-24

254Hitchcock, P. Mason, R. Thomas, K. & Shipley, G. 1974) Structural chemistry of 1,2 dilauroyl-DL-phosphatidylethanolamine: Molecular conformation and intermolecular packing of phospholipids Proc. Nat. Acad. Sci. USA vol 71(8), pp 3036-3040 198 Neurons & the Nervous System moieties. The 1981 Hauser et al. paper added considerable information about the packing of the phospholipids when in both crystalline and liquid crystalline form255. Hitchcock et al. examined phosphatidyl xxx. Many have built on their initial findings. Weiner & White provided more information on a bilayer of phosphatidyl choline, PtdCho (which they label DOPC) in 1992256. Flewelling & Hubbell describe the high internal potential of a typical lemma bilayer and tentatively describe the source of the potential as the dipole moments of the phosphatidyl moieties forming the bilayer257. Davis has provided data, contradicting Hitchcock et al. to a degree. Davis found the individual phosphatidyl xxx molecules rotate about their long axis when in the liquid crystalline state at biological temperatures. Seddon & Marsh have provided X-Ray data showing additional detail regarding the organization of liquid crystalline films of two PtdEtn’s of different chain lengths258. Sherer & Seelig have explored the orientation of phosphatidyl choline polar heads and their dipole moment when subjected to the presence of strong electrolytes259. Significantly, they note, “In spite of intensive research, the specific functional roles of these headgroups are still, by and large, unknown. Also of interest, they used the term molecular electrometer (a scalar device) to describe the quite accurate response of the phospholipid to the electric charge at the membrane surface. It is the scalar dipole potential that is the proposed result of initial transduction in this theory. Sherer & Seelig attempt to rationalize the nomenclature of phospholipid chemistry and explore both the crystalline and liquid crystalline form of these materials. White & Wimley have explored the presence of simple peptides with phosphatidyl choline260. They address five fundamental questions. They also note the dipeptides remain associated with the polar head. Marsh has reported on the water content of the interior of bilayer membranes of PtdCho under various conditions261.

PtdCho and PtdEtn are particularly simple phospholipids that do not offer the requisite oxygen rich binding sites required of receptors. All of the candidate type 4, semiconducting, phospholipids exhibit significant oxygen atoms arranged to support coordinate bonding.

Figure 8.5.5-16 to show the operation of the microtubules (villi) at the molecular level. Lowe & Gold have shown that odorant sensitivity and the odorant-evoked inward transduction current are uniformly distributed along the cilia262. Based on this premise, Gold has asserted, “Thus, all components of the transduction mechanism must be present in the cilia263.” Lowe & Gold also showed that the latency of the transduction current is independent of the region of the cilia that are stimulated: This implies that current is generated at the site of odorant binding.

255Hauser, H. Pascher, I. Pearson, R. & Sundell, S. (1981) Preferred conformation and molecular packing of phosphatidylethanolamine and phosphatidylcholine Bioehim Biophys Acta, vol 650, pp 21-51

256Wiener, M. & White, S. (1992) Structure of a fluid dioleoylphosphatidylcholine bilayer determined by joint refinement of x-ray and neutron diffraction data: III. Complete structure Biophys J vol 61(2) pp 434-447

257Flewelling, R. & Hubbell, W. (1986) The membrane dipole potential in a total membrane potential model: Applications to hydrophobic ion interactions with membranes Biophys j vol 49, pp 541-552

258Seddon, J. Cevc, G. Kaye, R. & Marsh, D. (1984) X-ray diffraction study of the polymorphism of hydrated diacyl and dialkylphosphatidylethanolamines Biochem vol 23(12), pp 2634–2644

259Scherer, P. & Seelig, J. (1989) Electric charge effects on phospholipid headgroups. Phosphatidylcholine in mixtures with cationic and anionic amphiphiles Biochem vol 28(19), pp 7720–7728

260White, S. & Wimley, W. (1994) Peptides in lipid bilayers: structural and thermodynamic basis for partitioning and folding Cur Opin Struct Biol vol 4, pp 79-86

261Marsh, D. (2001) Polarity and permeation profiles in lipid membranes PNAS vol 98(14), pp 7777–7782

262Lowe, G, Gold, G. 1991. The spatial distributions of odorant sensitivity and odorant induced currents in salamander olfactory receptor cells. J Physiol vol 442, pp 147–68

263Gold, G. (1999) Controversial issues in vertebrate olfactory transduction Annu. Rev. Physiol vol 61, pp 857–71 Signal Generation & Processing 8- 199 The figure starts with the two bilayer membranes separated by only a thin layer of water crystallized into a hydronium liquid crystal by the confined quarters. The two bilayers are both present as liquid crystals. They are formed of insulating lipids except in the shaded areas. The insulating lipids are phosphoglycerides with fully saturated nonpolar tails (typically PtdEtn). In the shaded areas, at least one of the tails is not saturated and is conductive like the polar head. These phosphoglycerides are typically from the family of globosides. They are typically associated with neurons and exhibit more complex polar heads and unsaturated tails. It is proposed these globosides constitute semiconducting electrolytic materials. The complex heads provide the steric arrangement required to accommodate either the electrostenolytic process or the OR mechanism shown. The more detailed discussion of these materials appears in [Section 5.2.4.1 ] The shaded area on the left constitutes a first Activa. The semiconducting lipid of the dendrolemma constitutes a region of P-type electronic material. The semiconducting lipid of the reticulum also constitutes a region of P-type material. The region of hydronium between them is an N-type electronic material. The sandwich forms an active electrolytic PNP transistor device, an Activa. The polar head of the dendrolemma facing the mucosa provides a steric site suitable for accommodating the electrostenolytic power supply providing –150 mV to the Activa collector as shown (See Section xxx). The polar head of the reticulum facing the dendroplasm passes the electron current directly into the dendroplasm as shown by the dashed arrow. The magnitude of this current is controlled by the bulk potential of the hydronium N-type material.

Figure 8.5.5-16 The proposed molecular operation of the microtubules in GRN’s based on the Electrolytic Theory of the Neuron. A portion of the surface of a microtubule extending from the dendritic knob is shown. The portion is typically replicated many times. This proposal can be compared to the equivalent conceptual transduction mechanisms of the chemical theory of the neuron, such as Ache, 1994. See text.

The N-type material of the Activa is connected to an adjacent P-type region on the right, again formed of a semiconducting globoside. The junction of the P and N materials forms a PN diode junction and an electrical connection to the stimulus sensing complex, the GR. The diode is forward biased and acts like a “low” impedance electrical conductor. The polar head of the globoside forming the semiconducting dendrolemma is sterically selected to support a steric 200 Neurons & the Nervous System coupling with the GR in liquid crystalline form. The polar head also constitutes one plate of a capacitor, CT, with the GR as the dielectric. The other plate is formed by the interface between the GR and the mucosa. The GR may or may not exhibit a dipole moment in the absence of a odor stimulant (horizontal bar inside left half of GR). However, in the presence of a sterically appropriate odor stimulant, it is a zwitterion and will exhibit a dipole moment as shown by the inclined bar on the right. This second dipole moment is due to a quantum-mechanical change in the configuration of the GR molecule(s). This change in dipole moment between the absence and presence of a stimulant causes a change in the dielectric properties and the resultant voltage on the capacitor, CT. The change in voltage is passed through the forward biased PN junction to the base of the Activa. This voltage controls the amplitude of the current passing through the Activa. The change in the dipole moment of the GR acts to repel the stimulant from its steric coupling with the GR. Thus, the presence of the stimulant in a steric relationship is only temporary. The resultant voltage change at the Activa is also temporary. The stimulant molecule is free to go without encountering any chemical change in character. The waveform at lower left shows the current waveform entering the dendroplasm resulting from a group of stimulant molecules coupling and uncoupling from the GR within a time period short with respect to the time constant of the overall quantum-mechanical-electrolytic circuit. The mathematical form of this waveform, including its latency, δ, are developed in Section xxx.

Hauser has described the stereographic structure of PtdCho (he uses the common name lecithin) and PtdEtn and demonstrated how the dipole moment of these two materials is quite different. The dipole moment of PtdCho is aligned with the long axis of the molecule and perpendicular to the surface of the membrane. The dipole moment of PtdEtn is aligned nearly perpendicular to the long axis of the molecule. This would suggest the surface potential of these materials when formed into a liquid crystalline monolayer would be quite different. However, when so aligned, it is not the vector quantity, the dipole moment that is important by the scalar dipole potential.

Vogel et al. have demonstrated the conductivity of the typical phospholipid is sufficient to allow the measurement of a finite potential between its two ends when present as a monolayer film264. They have measured the potential across phosphatidyl choline and phosphatidylethanolamine when in the single layer liquid crystalline state. These potentials are found when the liquid crystalline film has reached an equivalent molecular cross-sectional area of 40 sq Angstrom for these materials. The potential is given as 555 mV for PtdEtn and 669 mV for PtdCho at 18-19C with the hydrophobic surface negative. These are large voltages relative to the input dynamic range of the first Activa of the sensory neuron (see the following sections). 8.5.5.3.3 Description of the operation of the sensory neuron in gustation

In the absence of adaptation phenomena (addressed below), the description of gustatory intensity mechanism appears to be straight forward. This operation is illustrated conceptually in Figure 8.5.5-17 using a lower range of dipole potentials than cited above. The dipole potentials are closer to those reported by electro-physiologists in Section 8.5.10 below. The label PtdGR is meant to represent a specific GR related to one of the sensory paths. The figure supports the earlier schematic of [Figure 8.5.5-2 ] In the left frame, the absolute difference between the putative dipole potential of PtdGR and st of PtdCho is shown as +20 mV resulting in a net potential, vq, at the base of the 1 amplifier of the GRN of –20 mV. This is at the bottom of the operating range of the 1st amplifier. In the center frame, the action of a nominal AH,B type of DACB between the gustaphore and the GR is described. A gustaphore is inserted between the saliva and the GR with a dipole potential of + 20 mV. This action shifts the net potential, vg, by +20 mV to zero mV. Zero mV is the beginning of the saturation condition of the amplifier. An analog intensity difference will be reported by the

264Vogel, V. & Mobius, D. (1988) Local surface potentials and electric dipole moments of lipid monolayers: contributions of the water/lipid and the lipid/air interfaces J Colloid Interface Sci vol 126(2), pp 408-420 Signal Generation & Processing 8- 201 GRN and perceived as representing the change in the dipole potential of any gustaphore with a dipole potential between zero and +20 mV. In the right frame, a gustaphore is introduced that causes a change in the dipole potential at the base of the 1st amplifier of only one mV due to the gustaphore. However, the gustaphore does cause a change in the dipole potential of the phospholipid/GR combination of 10 percent of its quiescent value. This change is represented by a 26 degree change in the dipole potential vector associated with the GR. If its quiescent value was 300 mV, this change alone causes the net potential at the base of the 1st amplifier to change by 30 mV, driving the amplifier 10 mV into saturation. To compensate for this change, the concentration of the gustaphore would need to be reduced by 30:1 to be perceived appropriately by the system. If the dipole change due to the gustaphore was only 0.1 mV, but the change in the GR remained at 30 mV, the concentration of the gustaphore would need to be reduced by a factor of 300:1 to be represented appropriately, etc. Achieving increases of perceived intensity on the order of several thousand to one are reasonably achieved by this approach.

Figure 8.5.5-17 Effective potential at the base of the 1st amplifier (Activa) of the GRN. Illustrative only. Down arrows are the effective dipole potential of the phospholipid/GR combination. Up arrows are the dipole potential of the PtdCho bilayer of the outer lemma of the sensory neuron. In the absence of any stimulant, the queiscent potential, vq, at the base of the first amplifier is –20 mV, i.e., cutoff. In the simple AH,B case, a stimulus with a dipole potential of +20 mV will raise the base potential, vn, to 0 mV, nominal saturation (maximum processable signal amplitude). In the AH,B,X case. a gustaphore introducing no change in dipole potential but causing a 10% change in the dipole potential of the phospholipid/GR combination will drive the base potential xxx mV beyond saturation. See text.

It is proposed that the net surface potential between the mucosa and the interior layer of the type 4 bilayer membrane consists of three scalar components, the intrinsic dipole potential of the phospholipid, the contribution to the total dipole potential due to the coordinate bond pairing with the odorophore (AH,B), and any additional change in dipole potential caused by the third- point association (X) between the odorophore and the receptor molecule. The intrinsic dipole can be considered the quiescent contribution of the phospholipid to the biasing of the first Activa of the sensory neuron with the sum of the contribution to the dipole potential from the AH,B pairing and the X association constituting the signal element in the total electrical potential. With the nominal gain of 200x for the Activa and a sensitivity of about 0.1 to 1.0 mV for the following 202 Neurons & the Nervous System circuitry, a change in the quiescent potential of the phosphatidyl moiety of the sensory receptor of less than five microvolts would constitute an adequate signal for eliciting a sensation within the CNS. Such a small change is not currently measurable in the laboratory, but the presence of such changes are indicated by the psychophysical data of Imamura et al. and of Randebrock for changes in the potential due to coordinate coupling of the receptors with alcohols and carboxylic acids of different chain lengths and internal bond arrangements (Section 8.6.2.3). While Hauser has also suggested that the molecules of a monolayer membrane of PtdEtn are interlocked by hydrogen bonds between the nitrogen and the adjacent oxygen of the phosphorous oxide group, the more recent and comprehensive data of Davis suggest the individual molecules are rotating rapidly about their long axis at biological temperatures. Such rotation would appear to aid the contact between the receptor molecules and the stimulant molecules in the mucosa.

Many empiricists have attempted to graphically conceptualize the transduction mechanism of olfaction. See Ache265 reproduced in Schild & Restrepo (page 455), as an example. The chemical neuron approach is much more complex, usually employs sweeping arrows and symbolism, many question marks and has not shown itself amenable to verification or deterministic calculations.

The above figure provides a closed form, verifiable and deterministic model of olfactory transduction. It employs no putative pores or channels passing complex ions under the control of undefined protein processes. 8.5.5.4 The perception versus stimulus intensity function

Green, Shaffer & Gilmore have presented data for sucrose and for ethanol asserting a semilogarithmic perception based on largely psychophysical experiments266. The perception of sucrose is labeled “sweet.” The perception of ethanol is labeled “irritation.” A function of perception versus temperature is also presented.

Ossebaard & Smith have provided good data showing the semi-logarithmic transfer function of the hydrated sodium-path of the gustatory modality for stimuli above a minimum concentration267. They specifically note the difference in the sensory mechanisms related to their presumed Na+ of NaCl and K+ of KCl. 8.5.6 Electrophysiology of gustation–the Excitation/De-excitation equation

[xxx consolidate adaptation material between 8.5.6 and end of 8.5.7 ] The performance of the gustatory modality can be described using two different test protocols; the first involving a short pulse lasting less than 1/3 of the duration of the shortest time constant in the stage 1 sensory neuron (an impulse) and the second a long square pulse lasting longer than the longest time constant in the stage 1 neuron (frequently a square wave). The first protocol surfaces the quantum-mechanical character of the intensity aspect of the transduction process. The second surfaces the adaptation aspects of the circuits within sensory neuron itself. The impulse response can only be effectively measured by probing the analog output of the stage

265Ache, B. (1994) Towards a common strategy for transducing olfactory information Semin Cell Biol vol 5, pp 55-63

266Green, B. Shaffer, G. & Gilmore, M. (1993) Derivation and evaluation of a semantic scale of oral sensation magnitude with apparent ratio properties Chem Senses vol 18(6), pp 683-702

267Ossebaard, C. & Smith, D. (1995) Effect of Amiloride on the Taste of NaCI, Na-gluconate and KCI in Humans: Implications for Na+ Receptor Mechanisms Chem Senses vol 20(1), pp 37-46 Signal Generation & Processing 8- 203 1 neuron. The square wave response is also best measured at the axoplasma of the stage 1 neuron but crude data can be acquired by probing the stage 3 action potential generating neurons. This method suffers from the incorporation of any stage 2 signal processing in the measured data unbeknownst to the investigator. [xxx Show or cite examples of the impulse response adjacent to the square pulse response ] [xxx use a variant of Fulton Squire03pg611.wpg found in fig 8.7.1-2, Figure 8.5.6-1 ]

8.5.6.1 The impulse response of the stage 1 neuron 8.5.6.1.1 Character of the DACB phenomenon

The literature has long struggled with the character of the “coupling” between the gustaphores and the GR’s of gustation. Conventional chemical reaction theory calls for the generation of reaction products that Figure 8.5.6-1 Framework for impulse response should be detectable. There are none. versus square pulse analyses ADD. Modified from Conventional chemical theory would also Squire, 2003. suggest a slow reversible process reaching some form of equilibrium. None has been documented.

The actual process involves the DACB coordinate bond relationship. The single coordinate bond known as the hydrogen bond is exceedingly weak, typically described as of 5 kCal or less energy per bond. Such a bond is near the thermal threshold for stability at biological temperatures. The requirement that pairs of these bonds be formed and maintained for a finite period is even more demanding. In general, such DACB relationship can only be maintained for a short interval before it is broken. The process of formation and destruction is repeatable until one of the constituents is removed from the area. However, this repetitive cycle changes the character of the transduction process. Each DACB becomes a quantum-mechanical process definable by one or more time constants.

With the realization that the transduction process is a quantum-mechanical process, it shows a very close analogy to that used in other stage 1 transduction neurons (particularly those of vision). In fact, the entire transduction–sensory signal amplification mechanisms appear virtually identical in all sensory neurons of the neural system (with the exception of the initial transduction step tailored to the specific modality).

8.5.6.1.2 Circuit description of the gustatory sensory neuron with GR REFOCUS EDIT

[ refocus on gustatory neuron ] With the exception of one simple two-terminal circuit applicable to chemoreception in general proposed by Kurihara et al., no circuit diagram of the olfactory sensory neuron could be found in the literature268. Their circuit diagram contains no capacitors or active elements and does not address the detailed operation of the olfactory sensory neuron. Figure 8.5.6-2 shows the proposed sensory neuron of olfaction based on the common functional circuit elements found in other sensory modalities. The circuit is amenable to direct quantum-mechanical stimulation

268Kurihara, K. Miyake, M. & Yoshii, K. (1981) Molecular mechanisms of transduction in chemoreception In Cagan, R. & Kare, M. eds. Biochemistry of taste and olfaction. NY: Academic Press Chap 13 204 Neurons & the Nervous System of the base region of the first Activa (as in photoreception) or to the application of an electrical potential to the base electrode (as in phonoreception). Because of the expected low energy of stimulation, the figure adopts the configuration used in hearing.

Figure 8.5.6-2 Candidate circuitry of the gustatory sensory neurons. All stage 1 and 2 elements operate in the analog domain. The steric feature of interest in gustation is the d-value of the DACB coupling. The Activa of the afferent synapse operates as an “active diode” with a bypass capacitor. The electrolytic power supply forming the battery associated with impedance (1) in the Activa collector circuit is shown in its hydraulic analog at the top of the figure. The dendrite of the stage 3 ganglion typically functions as the action potential encoding device.

In the case of phonoreception, the first Activa is biased into the operating range required by a conventional amplifier. The circuit is capable of accepting electrical signals in the tenths of a millivolt to ten millivolt range. Such a range appears compatible with the energy levels involved in the odor reception scheme suggested by Turin.

In hearing an AC electrical potential is created between the plates of a capacitor by piezo- electric action. In the case of olfaction, it is proposed a similar AC electrical potential is created between the plates of a capacitor by dielectric polarization due to an electronic rearrangement of the dielectric. the dielectric is formed by a stimulant sensing complex, GR. This rearrangement is temporary and caused by the temporary presence of a stimulant in steric coupling with the GR. The proposed configuration creates a quantum-mechanical shift in the electronic state of the GR every time an appropriate stimulant couples with the GR. This coupling at the molecular level would be a low probability event unless the GR was present in a liquid crystalline form on the surface of the input structure of the sensory neuron. It is proposed that the GR is present in such a form and is located on the surface of a specialized section of the dendroplasm of the sensory neuron. It is this specialized section that creates the Activa just as it does in other modalities. [xxx eliminate the capacitor part ] It is further proposed that the GR is present in a liquid crystalline form where part of the crystalline configuration forms the dielectric of a capacitor between the electrolyte of the mucosa and the base region of the Activa within the sensory neuron. Signal Generation & Processing 8- 205 Indicator dyes are characterized by their multiple electronic configurations. Each of these configurations exhibits a moment. The change in configuration can be described by a transition dipole moment. A non-zero transition dipole moment change in a material forming the dielectric of a capacitor will result in a change in the potential between the electrodes of that capacitor. The concept is discussed briefly in Adamson269. Maron & Lando go into the subject of dipole moments, and the various spectra related to molecular structure270. They provide a pedagogical energy diagram. For molecules of more that three atoms, the complexity of th spectra becomes quite high. However, the change in polarization associated with the vibration-rotation spectra are of interest here. The energies of interest are in the millivolt range and the associated wavelengths are in the near infra-red range. Petersen & Cone have provided an excellent paper discussing many aspects of the dipole moment of proteins, particularly rhodopsin, the disk protein associated with vision271. It forms a good tutorial as well as providing valuable protocols and experimental data. They provide a footnote; “the dipole moment, μ, is in Debyes (D), where 1 D = 10–18 statcoulomb-cm. Thus an electron and a proton separated by 1 Angstrom produce a dipole moment of 4.8 D. In other words, 1 charge-D = 4.8 D.” They show that most small proteins separate into two classes based on their dipole moment and their molecular weight, those with values of μ/W of 5 x 10–3 D/Dalton and a second group with μ/W clustered about 20 x 10–3 D/Dalton. Rhodopsin with a dipole moment of 720 Debye and a molecular weight of 35,000 ± 2000 Daltons, has a μ/W value of 20 x 10–3. Gerber et al. note that the proteins with the larger dipole moments all tend to aggregate and suggest that the dipole moments aid in the aggregation process. This ability of the protein to aggregate is crucially important in the formation of the outer disks of visual sensory neurons. Aggregation may play a similar role in forming the olfactory receptors (OR’s). The change in the dipole moment of rhodopsin upon saturating illumination is only 25 D (5 charge-D) or about 3%. Petersen & Cone give precise data on the dielectric constant of rhodopsin solutions and its change with applied frequency in a standard test cell. In the in-vivo situation, a value for the aggregated material is believed to be necessary.

[xxx only paragraph in this part mentioning organo-metallics ] Many complex organo-metallic compounds exhibit a dipole moment in the 2-5 Debye range272. More comprehensive general lists are available but they tend to be old273. Following further isolation of the chemistry of interest, a more detailed search may uncover the desired values for their dipole moments. The lower molecular weight of these organo-metallic materials (the molecular weight of copper tartrate is only 329 Daltons, give them much higher μ/W values than those of proteins. Nominal static μ/W values for the organo-metallics are in the 2–6 x 10-3 range. The transition dipole moment may be much larger relatively because of the small size of the molecules.

Takashima has provided excellent background on the dipole moment274. An on-line calculator for calculating the dipole moment of arbitrary proteins is available275. However, the structure of the protein must be known in detail.

269Adamson, A. (1973) A Textbook of Physical Chemistry. NY: Academic Press pp 911-913

270Maron, S. & Lando, J. (1974) Fundamentals of Physical Chemistry. NY: Macmillan Chapter 5

271Petersen, D. & Cone, R. (1975) The electric dipole moment of rhodopsin solubilized in Triton x-100 Biophys J vol 15, pp 1181-1200

272CRC (1975) Handbook of Chemistry and Physics. Cleveland, OH: CRC Press pg E-66

273Smyth, C. (1941) Dipole moment and bond character in organometallic compounds J. Org. Chem vol 06(3), pp 421–426

274Takashima, S. (1999) Computation of the dipole moment of protein molecules using protein databases Colloids and Surfaces A vol 148, pp 95-106

275Felder, C. Prilusky, J. Silman, I. & Sussman, J. (2007) A server and database for dipole moments of proteins Nucleic Acids Research vol 35, special Web Servers Issue. 206 Neurons & the Nervous System

Two technical challenges arise in attempting to define a steric mechanism of transduction. First, the capture cross-section of the transduction mechanism must be as large as possible. Second, the transition dipole moment must be as large as possible and it must be optimally sensed. Forming the GR’s into a liquid crystalline structure is one method of increasing capture cross- section and potentially increasing the efficiency of sensing the net transition dipole moment. - - - - Not shown explicitly in these diagrams are the large dipole potentials introduced by the inner and - - - - Figure 8.5.6-3 shows the circuit along with its cytological equivalent. Frame A shows the cytological layout of the proposed typical olfactory sensory neuron. All plasmalemma are electrically insulating bilayer membranes, except for specialized sections that are semiconducting. These specialized sections are active participants in forming active semiconductor devices, Activa, or in supporting the electrostenolytic process electrically powering the cell. All of the sensory neuron except for its axon is located within the olfactory epithelium and its mucosa. The body of the sensory neuron peripheral to the soma is arbitrarily defined as the dendrite portion. That portion between the soma and the olfactory bulb is arbitrarily described as the axon. Signal Generation & Processing 8- 207

The bulbous extreme end of the dendrite is labeled the knob. Its internal structure is not reported in the literature. Multiple individual microtubules emanate from the knob. These function as dendritic spines with multiple sensitive areas typically associated with synapses in non sensory

Figure 8.5.6-3 Proposed cytological and electrolytic description of the olfactory sensory neuron ADD & MODIFY. MODIFY frame A to show stimulation. DEFINE SBC. 208 Neurons & the Nervous System neurons. In the sensory neurons, these areas are modified to support transduction. Only the microtubules (cilia) lie in the mucosa. The number of microtubules emanating from the dendritic knob varies from species to species, from as few as four to as many as thirty. The typical length of the microtubules is about 50 microns in mammals276. The microtubules are splayed into a planar surface within the xxx thick mucosa. The three horizontal arrows in the dendroplasm indicate where the multilayer sandwich to their left actually extend up into the individual cilia of the neuron. The character of these layers changes from insulating to semiconducting at multiple points along each cilia just as they do in the microtubules of the visual sensory neurons. The two insets, A1 & A2, illustrate how the transduction mechanism may appear at one point along the length of the cilia. Inset A1 shows an Activa (black rectangle) supported by its electrostenolytic supply (E.S (1)) and the transduction element consisting of an GR forming the dielectric of a capacitor, CT, with its outer conducting surface formed by the mucosa. Inset A2 shows an alternate configuration where an auxiliary electrostenolytic source (E.S. (0)) appears on the outer surface of the capacitor. This source provides an electrical bias to the capacitor that may provide a higher sensitivity to the overall circuit as discussed in Section xxx.

The dashed arrows associated with each electrostenolytic supply show the direction of electron charge flowing into or out of the respective plasmas. The “conventional charge” defined erroneously by Benjamin Franklin flows in the opposite direction.

Charges from each of the Activa in each cilia are introduced into the dendroplasm and generate a potential between the dendroplasm and the podaplasm. This causes the Activa shown between the dendroplasm and the axoplasm to transfer approximately 200 times as much charge to the axoplasm. This additional charge passes through impedance associated with E.S. (4) and generates the output voltage at the synapses shown.

The sensory neuron delivers a tonic generator waveform at the remote glomeruli by diffusion. Because of the length of the axon required to pass through the cribriform plate and the low rate of signal propagation by diffusion, the bandwidth of the olfactory sensory neurons is necessarily low.

Frame B shows the same circuit configuration as in Frame A. [xxx Expand ]

Frame C shows the same circuit configuration as in Frame B but redrawn to illustrate the commonness of the circuit. It is described as an PNP type asymmetric differential pair in conventional electronic engineering terms. The left (or first) Activa is base driven and the right (or distribution) Activa is emitter driven. The left Activa provides a high impedance input and a gain of nominally 200:1. The right Activa provides a low impedance output at 1:1 gain suitable for driving multiple synapses. [xxx Expand ]

Frame A1 has been expanded further and discussed in detail in Section 8.5.5.1. Figure 8.5.6-4 reproduces a set of gustatory path waveforms from Danilova et al277. These are summated waveforms from two locations along the gustatory neural pathway. Their analog character is reconstructed from the action potential pulse streams recorded at these locations. They noted the recordings were from the whole nerve. They also noted, “All stimuli elicited a response except NaCl in the NG.” It is also noteworthy that they did not include HCl in their data set. HCl, an inorganic acid, elicited negligible response in their broad set of experiments involving these two nerves. Similar waveforms recorded at the chorda tympani were presented by Pfaffmann in 1976.

276Menco, B. (1983) The ultrastructure of olfactory and nasal respiratory epithelium surfaces In Reznick, G. & Stinson, S. eds. Nasal Tumors in Animals and Man . . . Vol 1. Boca Raton, Fl: CRC Press pp 45-102

277Danilova, V. Danilov,Y. Roberts, T. Tinti, J-M. Nofre, C. & Hellekant, G. (2002) Sense of Taste in a New World Monkey, the Common Marmoset: Recordings From the Chorda Tympani and Glossopharyngeal Nerves J Neurophysiol vol 88, pp 579–594 Signal Generation & Processing 8- 209 By comparing these responses with those obtained within the visual278 and auditory279 modalities, it is possible to recognize several phenomena. First, the analysis will be limited by the lack of a vertical scale for these waveforms. The paper did not say the waveforms were presented using the same vertical scale. Second, the concentration of the stimulant varied dramatically between these waveforms. [xxx include a simple version of formula and graph here if none in Section 8.2 or 8.3) The waveform for sucralose on the left (a low caloric artificial sweetener of the AH,B,X type with formula C12H19Cl3O8 ) has been overlaid with a nominal excitation/de-excitation response to aid this discussion. Several features are noteworthy; • The overall sucralose response does not show any sign of saturation or other distortion. • The delay between points A and B is due partly to the physical diffusion of the stimulant to the point of transduction at the GR on the surface of individual cilia of sensory neurons. A second part is due to a delay within the transduction mechanism of the sensory neuron that is stimulus intensity related. • The rise time constant, τrise, associated with the response starting at point B is a function of the stimulus intensity rather than a fixed value. • The decay time constant, τdecay, has a fixed value determined by the internal electrolytic components of the transduction mechanism. • The net response waveform following the beginning of stimulation is the difference between the exponential rising response and the exponential falling response (that both start at the same time, B. • The decay portion of the net response waveform is not related to and is not calculated from the peak of the net response. • The delay between the end of stimulation, C, and the discontinuity in the net response, D, is the same as the part of the delay between points A and B due to phenomenon within the transduction mechanism. • The precipitous drop at point D is a real part of the overall excitation/de-excitation function. • The delay following point D is a true exponential function equal precisely to the decay time constant, τdecay.

8.5.6.1.3 Analog waveforms generated by stage 1 neurons EMPTY REFOCUS

[ Refocus on gustatory modality ] The literature provides many analog generator waveforms from olfactory receptor neurons (Schild & Restrepo, page 443-444). These waveforms exhibit a latency as a function of stimulus intensity that is not compatible with experiments involving the putative photoexcitation of cAMP. This latency as a function of intensity is characteristic of the excitation/de-excitation mechanism proposed in this work. Farbman has provided a very clear patch-clamp recording from the olfactory cilium of a frog (page 110). The change in amplitude of up to 20 mV suggests the recording was from the axoplasm of the sensory neuron. The variation in the amplitude of the signal over time suggests some logarithmic compression due to the current to voltage conversion but no sign of hard saturation or of action potential generation.

278Fulton, J. (2004) Processes in Biological Vision. Section 7.2.4 http://neuronresearch.net/vision/pdf/7Dynamics.pdf

279Fulton, J. (2008) Processes in Biological Hearing. Section 5.4 http://neuronresearch.net/hearing/pdf/5Generation.pdf 210 Neurons & the Nervous System

Figure 8.5.6-4 Summated chorda tympani and glossopharyngeal nerves during taste stimulation of the tongue in a marmoset. The thich horizontal line at the bottom of each recording indicates the onset and end of stimulation using a computer controlled open flow system, the Taste- O_Matic. NO vertical scales were given. See text. From Danilova et al., 2002. Signal Generation & Processing 8- 211 Based on the overlay, it is clear that the summated sucralose waveform at the chorda tympani was the result of rapid diffusion of the stimulant to the point of transduction, the transduction process exhibited a very fast rise time indicating that the stimulus intensity was near the maximum acceptable by the relevant GR’s, the decay characteristic (although noisy) was precisely as expected and the decay time constant was nominally 2 seconds (subject to refining later). Looking only at the left panel of the figure, the waveform for sucrose shows a similar shape to that of sucralose except for the limited amplitude due to saturation in the transduction mechanism. The decay time constant was also apparent and had a value of approximately 2 seconds (compared to 12.5 msec in hearing). A similar degree of saturation appears to have occurred in the waveforms for QHCl and SC-45647. The waveforms for NaCl and citric acid are anachronistic during the washout interval following stimulation. The waveform for ethanol may reflect the presence of another gustaphore within the stimulant sample. Looking at the right panel, the lack of a response to NaCl was a common feature of their neural bundle accessed within the glossopharyngeal nerve. The citric acid response was similar to that from probing the CT (including the anachronistic response during the washout period. QHCl exhibited a nearly ideal response relative to the ideal excitation/de-excitation equation. The sucrose and SC-45647 responses show saturation within the transduction mechanism. The sucralose response was as expected. The ethanol response was closer to what would be expected by a typically tasteless aliphatic alcohol.

The minimum delay between points A and B is about 0.5 seconds for citric acid (as measured at the chorda tympani).

8.5.6.1.4 The generic Excitation/De-excitation equation applied to gustation EMPTY

The actual excitation/de-excitation equation is quite complex and can be found, including several specific ramifications in the citations provided above. The equation is associated with a quantum mechanical mechanism such as the forming and disbanding of DACB’s as hypothesized here. The relevance of the equation to and association with a gated channel phenomenon, such as proposed in the chemical theory of the neuron, has not been demonstrated.

8.5.6.1.5 Circuit parameters of gustatory transduction EDIT [ Refocus to gustatory modality transduction ]

Schild & Restrepo have included considerable parametric data (with citations) on the axonal portion of the olfactory sensory neurons. Statements like “whole cell capacitance” presume a two terminal structure for the neurons. Under the Electrolytic Theory of the Neuron, this label should be modified to the whole axoplasm capacitance as it was obtained by injecting charge into the axon and measuring the resulting voltage rise. Values in the 2-10 pF range are consistent with the sensory neurons of other modalities. They report a wide range of axon resting potentials, undoubtedly because of the lack of control of the dendrite environment while investigators made these measurements. A value of –70 mV would be expected and falls near the middle of the range quoted. They report an average time constant of the axoplasm of ~ 60 ms, giving a first order RC filter corner frequency of ~16 Hz. While lower than in other modalities, these values are consistent with the extended length of the axon required to pass through the cribriform bone. Measured time constants in some species have been as long as 100 ms. They did note an important point, the delay displayed in their generator potentials of figure 4 are proportional to the amplitude of the stimulus (page 433). Unfortunately, they have continued to repeat the reports based on conclusions improperly drawn the work of Gesteland in 1971280. That assertion was that sensory neurons exhibit action potentials (beginning on page 434). Schild & Restrepo

280Gesteland, R. (1971) Neural coding in olfactory receptor cells In Beidler, L. ed. Handbook of Sensory Physiology, Vol 4, Part 1: Chemical Senses. Berlin: Springer-Verlag pp 132-150 212 Neurons & the Nervous System note, “These reports were discussed controversially, see Getchell281.” Gesteland was very clear that his measurements were extracellularly, and in fact of the crudest kind. His measurements were made between the two sides of the olfactory epithelium. Because of his technique (his figure 1), he actually measured generator waveforms of opposite polarity to that at the axon of a cell “and with action potentials spikes superimposed” (page 144). He was measuring the complement of the generator waveform at the collector of the first Activa instead of at the collector of the second Activa (the normal axon potential (See Section xxx). It is now known his current path in figure 1 requires changing to reflect the documented role of the poditic terminal.(Section xxx). Gesteland does provide several parameters supporting the neural model developed here. [xxx review gesteland 1971 ] Additional data showing such action potentials has not appeared subsequently. It appears Gesteland relied upon data collected extracellulary (intercellularly). It exhibits low amplitude action potentials riding on top of generator waveforms. The combined waveform is easily explained as due to capacitive coupling under the Electrolytic Theory of the Neuron. Schild & Restrepo show generator waveforms on page 444 that are free of action potential features.

[xxx applies to an olfactory neuron ] Schild & Restrepo reproduce a figure from Kurahashi282 that may represent the diode characteristic of the axon load, Figure 8.5.6-5. If so, the load diode had a reverse cutoff current (I0) of –25 pA. Unfortunately as noted in the original caption, the current traces shown on the left were arbitrarily shifted. The nominal horizontal axis of each waveform does not correspond to zero current. The holding voltage would have been accompanied by a holding current that is not shown explicitly, but can be seen on the right.

281Getchell, T. (1986) Functional properties of vertebrate olfactory receptor neurons Physiol Rev vol 66, pp 772- 818

282Kurahashi, T. (1989) Activation by odorants of cation-selective conductance in the olfactory receptor cell isolated for the newt J Physiol (Lond.) Vol 419, pp 177-192 Signal Generation & Processing 8- 213

Figure 8.5.6-5 Odorant induced currents at various holding potentials from a newt, Cynops pyrrhogaster ADD. Current traces were arbitrarily shifted. See text. Modified from Kurahashi, 1989 214 Neurons & the Nervous System

8.5.6.2 The square pulse response of the stage 1 neuron

[xxx figure duplicates Figure -26 in the summary above ] Hellekant et al. have provided the clearest example of the E/D response of a gustatory neuron based on their empirical investigations283 in Figure 8.5.6-6. It is virtually identical to the theoretical response described here based on the Electrolytic Theory of the Neuron. A minor exception involves their labeling of the rise time as starting before the delay time has elapsed. The rise time as defined in this work involves the time from when the response deviates significantly from the baseline to the peak amplitude of the waveform. The deviation begins when the analog electrical signal intensity from the dendritic structure is first sensed at the emitter of the Activa within the neuron. Shallenberger (1993, section 10.7.1)was probably referring to this revised rise time when he noted casually “the rapid nature of taste transactions, which is usually about 50 microseconds.”

The interval labeled the “resume time” by Hellekant et al. is modified to extend the actual response time and more properly label the decay time in accordance with this work. The decay time is an exponential explicitly describing the decay time constant of the neuron. It constitutes the time following the cessation of stimulation as perceived at the emitter of the Activa and not at the end of the stimulation time. Extensive data is provided in the Hellekant et al. paper but many of the time constants and other values require re-interpretation in the light of the more highly defined terms of the E/D equation of sensory neurons.

8.5.6.2.1 Circuit parameters of gustatory adaptation

In the above figure, the decay time constant of the nominal gustatory neuron is approximately two seconds. The corrected rise time constant associated with the same neuron (difference between the beginning of the leading edge and the maximum amplitude) is on the order of 0.6 seconds. The delay time shown may include a diffusion time for the stimulant to reach the dendrite and may be dependent on the test protocol used. It is also dependent on the intensity of the stimulus. The rise time is also a function of the intensity of the stimulus. The variation in these values are illustrated in Table 2 of Hellekant et al.

Figure 8.5.6-6 The characteristics of the E/D response in gustation according to Hellekant et al. The lower scale has been expanded to delineate between the stimulation time and the response time. See text. Modified from Hellekant et al., 1991.

283Hellekant, G. Walters, D. Culberson, J. et al. (1991) Electrophysiological evaluation of sweeteners In Walters, D. Orthoefer, F. & DuBois , G. eds. Sweeteners. Wash. DC: American Chemical Society Chap. 22 Signal Generation & Processing 8- 215

8.5.6.3 Chemical kinetics at the receptor/gustaphore interface

Some of the papers cited below do not clearly indicate how they calculated their concentrations. Unless noted otherwise, this section will speak in terms of a gram molecular weight of solute dissolved in 100 ml of solvent. This is more important than might be expected for two reasons; first some of the materials are complex sugars that are hydrolyzed as they dissolve, second, many of the molecules considered are gustants containing multiple gustaphores. The effective concentration may not be the same as the concentration defined above and used by the authors. The E/D equation describes the physiological transduction process at the sensory receptors in precise and considerable detail. Because of its temperature term, it is even applicable to exothermic animals with poor physiological temperature control. The C/D equation also provides insight as to the chemical kinetics at the receptor/gustaphore interface on a global scale. It does not, however, describe the kinetics of the transduction process in detail. This section will address these kinetics.

The chemical kinetics of gustation have been studied since the 1960's. However, lacking an adequate model of the transduction mechanism, these studies have tended to fall back on the investigators concepts of the fundamentals of chemistry. Thus, the laws of chemical equilibrium related to solutions and the Law of Mass Action have played a significant role in these investigations. Much of the analyses have revolved around the perceptions of human subjects. These investigations have led to many approximations to overcome the difficulties with the models assumed and the mathematics involved. The work of Hill ( xxx, cited in DuBois), Beidler (1954 & 1961), Stone & Oliver (1966), Dzendolet (1967), Dubois et al. (1991), Shallenberger (1993) and a variety of others will be explicitly discussed here.

Most of the above investigators rearranged an equation first offered by Beidler in the 1950's in order to evaluate the constants applicable to the equation. They had little interest in using the equation in an operational context. The goal of this discussion is to gain insight into the chemical kinetics of the transduction process supporting the Excitation/De-excitation Equation of the previous section. This will further the overall understanding of the gustatory modality and probably other sensory modalities as well (such as olfaction, Section 8.6)

An initial global conclusion based on the composite data from the above authors is that it is possible to correlate the human perceived responses to gustatory stimulations as a function of a linear transfer function over a significant instantaneous range of stimulation. This perceived response also correlates well with the stage 3 pulse rates along various nerves serving the gustatory modality. This reinforces the assumption that psychophysical intensity profiles do linearly match similar electrophysiological intensity profiles.

The following discussions are based primarily on poorly defined conceptual models of gustatory transduction assuming a conventional chemical reaction in some state of equilibrium. Determining whether an equilibrium condition is ever achieved, or even applies to gustatory transduction is a goal of this discussion. Shallenberger reviewed much of the same material as reviewed below in 1993 in a text284. Little new material appeared in this volume relative to the present discussion. Several parameters were provided however.

8.5.6.3.1 The Beidler equation of chemical kinetics in transduction

284Shallenberger, R. (1993) Taste Chemistry. Glasgow, Scotland: Chapman & Hall 216 Neurons & the Nervous System

Beidler derived an equation for the concentration over perceived response (C/R) as a function of the stimulus concentration (See figure in following discussion of the Dzendolet “correction”).285 The derivation was based on his interpretation of the Mass Action Law. K was the conventional equilibrium constant for the reactants involved, the sensory receptor reaction sites and the molecules of odorant in solution in the saliva. Beidler’s analysis relied upon the following assumptions, “(I) The reactions involved in stimulation are in a time-independent state, very likely in thermodynamic equilibrium, since the response to 0.I M NaCI was shown to remain constant during I0 minutes of continued salt stimulation. The magnitude of this response is the same no matter whether immediately preceded by higher or lower concentrations of the salt. (2) Stimulation is very rapid as shown by the fact that a response may be recorded within 50 msec. after 0.2 K NaCl is applied to the surface of the tongue. (3) The responses are completely reversible. (4) Both the cations and anions enter the reaction although the magnitude of response is primarily determined by the presence of the cation. (5) As the strength of the stimulus is increased, a level of response is reached at which a further increase in stimulus does not result in an appreciable increase in response. (6) The receptors of the tongue respond to a large number of different substances and over a wide range of concentrations.”

The Beidler paper draws a variety of conclusions based on these assumptions. This work would question a variety of these assumptions.

(Item 1) His use of average pulse rates at the stage 3 chorda tympani implies steady-state conditions. However, the perception of saltiness is known to be stimulus intensity sensitive. His results would be quite different if he reported on the pulse rate during the first few seconds of the response to stimulation (as suggested in item 2..

(Item 3) The responses may not be revesable under conditions of adaptation typically associated with stimuli of high concentration over extended periods, or the presence of other molecular stimulations to the system. (Item 4) Beidler did not demonstrate that the anion played any role in his reported investigation. The role of the chlorine ion is quite independent of the role of the sodium ion with respect to the N-Path, the hydrated sodium ion sensitive channel of gustation) (Item 5) This item remains true and is discussed with respect to DuBois et al. below. (Item 6) This statement remains correct but can be made considerably more precise based on this work. Each type of receptor, GR 1, GR 2, GR 3 or GR 4 responds to a large number of stimulants belonging to one of the four stimulant classes, acidic, dulcal, hydrated sodium-like or picric.

Beidler self proclaimed his equation (shown in the following figure), “This is the fundamental equation relating the magnitude of response to the concentration of the applied chemical stimulus.” It is clearly not fundamental but derived and does not apply at R=0 where C/R is undefined. In his “Application to Electrophysiological Data” section, he notes, his fundamental equation “does not necessarily prove that the chosen theory of stimulation is a correct one.” Beidler also notes, “low values of the change in free energy(which he explored) should be expected for all species of mammals if the mechanism of taste stimulation as outlined in this paper is a general one.” The low values reported are noted, but the mechanism of state stimulation is disputed by this work. With regard to his analyses involving NACl stimulation, he notes, “clearly indicates that the ions of chemical stimulus are loosely bound to some substance of the taste receptor.” It is proposed here that this observation can be extended to a variety of non-ionic molecules capable of forming a DACB as well. Beidler concluded his 1954 paper with the provocative assertion, “The recently proposed (Baradi,

285Beidler, L. (1954) A theory of taste stimulation J Gen Physiol pp 133-139 Signal Generation & Processing 8- 217 A. F., and Bourne, G. H., Nature, 1951,168, 977. ) enzymatic reactions for chemoreceptors do not appear plausible for sodium salt stimulation of the taste receptors of the rat.” DuBois et al. offered a rearrangement of the Beidler equation in a more application oriented form where the output signal (response, R) is a function of the input stimulus (concentration, C) that is also shown in the following figure. This form describes the function, f( ) as linear in C for C below a critical point where it becomes asymptotic to a constant value, Rm. As noted below, diffeerent investigators have chosen to use a variety of forms of the Beidler equation that makes retrieval of the constants involved easier. - - - - Dzendolet expanded on what had become the baseline Beidler Equation in 1967. He showed that the assumptions made by Beidler resulted in his equation being limited to regions of high stimulus concentration applied to receptors exhibiting minimal quiescent stimulation levels. Such a situation is frequently described in engineering as the small signal condition (applicable to only a limited extent of the dynamic range available). Dzendolet sought to provide a broader (large signal condition) equation by removing some of the simplifications. Beidler defined Z as the number of sites which combined with the stimulus at the concentration C. Dzendolet indicated this expression was correct only for very high concentrations of the stimulus when Z is negligible. For the more general condition, he asserted that the mass action equation without Beidler’s assumption would replace C by C–Z. Based on this change, he defined the magnitude of the perceived response as R =aZ, where a is a constant and Z equals the number of filled receptor sites. Rm is defined as the perceived response when all available receptor sites are filled, aN. Dzendolet asserts his variant “has the advantage of being exact and applicable at all stimulus concentrations.” Figure 8.5.6-7 compares the Dzendolet equation compared to the equation of Beidler. Dzendolet chose to put all of the constants on the right hand side of the equation and leave the expression Rm - R as a term on the left. To express his equation in the form of Beidler, that term should be made a denominator under all of the right hand terms.

Dzendolet compared his equation to the earlier data of Beidler from fig. 5 of a 1953 paper. The data was collected from the stage 3 chorda tympani nerve of a rat to various concentrations of NaCl. He calculated Rm as equal to 29.0 and noted the sensitivity of the graphic to the value of Rm. For other experiments, Rm may be arbitrarily assigned the value of unity, and other responses treated as fractional values of Rm. Dzendolet spoke in terms of the Law of Mass Action and solution-based equilibria. They also assumed a reversible bond between the receptor and the gustaphore. This led him to speak of molecular associations and Figure 8.5.6-7 Comparison of the Beidler and dissociations of reaction chemistry rather Dzendolet equations describing the perceived than the DACB’s of coordinate chemistry. response reflecting a stimulus at a concentration C where a and K are constants and Rm is the Dzendolet provided a replot of the Beidler maximum perceived response based on data, Figure 8.5.6-8, that does not relate to calibration experiments. See text. the output as a function of input representation in gustation but is of the parametric type. It attempts to describe a more precise relationship between his stimulus concentration to response ratio, C(Rm–R)/R versus the perceived response (R) of the applicable chemical kinetics. The figure is provocative but must be discussed in its parametric context. As R approaches Rm, the abscissa must approach zero regardless of the value of C. Because of this feature, he related the curve to two distinct phases of transduction, an interval of chemical disassociation between the receptors and a gustaphore and a related interval of association between he same elements. He noted, the left-hand portion of the waveform is straight with a 218 Neurons & the Nervous System positive slope. However, the form of the graph cannot be linear on the right where it exhibits a lower negative slope. Dzendolet chose to sketch in a transition from the rising to the falling portion of the waveform and provide two linear dashed lines for reference. Signal Generation & Processing 8- 219

Based on his interpretation of this curve, the left portion describes the disassociation of the receptor/stimulus into a free receptor site and a free stimulus molecule (or ion of the molecule). Conversely, the right portion represents the association of receptor sites and stimulus molecules (or ions thereof). He concludes, “In other words, a response is initiated when a dissociation rather than an association reaction occurs at the receptor.” No physiological model supporting this mathematical interpretation was provided. It is difficult to accept the interpretation of Dzendolet when in the context of chemical equilibrium [xxx see Smith’s College Chem, page 233 ] since it implies the sensory receptors were initially associated with the stimulant before the stimulant was applied to the sensory organ itself. Beidler plotted C/R as a function of the independent variable, C, where C/R is clearly undefined at C = 0. It would be preferred if they focuses on either the response or the ratio of response to concentration as a function of concentration rather than the reciprocal Figure 8.5.6-8 Parametric graph of the kinetics of forms. A value of C/R approaching zero is gustation by Dzendolet ADD. Note, the ordinate equivalent to a value of R/C approaching does not relate to an independent variable of infinity (neither of which is a realistic excitation. The value of R is relative to Rmax = 1.0. situation). Dzendolet attributed different operating environments to the positive and negative slope regions (which appear inappropriate). See text. The dashed lines were his contributions to data from Beidler, 1953. From Dzendolet, 1967.

DuBois et al. cited the Beidler equation in a different form than as published in Beidler (1954). However, it is in a more conventional output as a function of input form. In this form, the equation can be graphed more conventionally. For values of C less than 1/K, the equation rises linearly with concentration. Significantly above C = 1/K, the equation is asymptotic to Rm. There is no reversal in the slope of R as a function of C. This form can be interpreted as; the number of occupied sensory receptor sites is increasing with the concentration of the applied stimulus, and as the number of available receptor sites becomes more limited, the effectivity of the stimulus is reduced as would be expected. This interpretation avoids the problem with the dissociation issue with the analysis of Dzendolet. It will also be shown to tie together the analyses of DuBois et al. addressed below.

By rearranging the equation of Dzendelot as DuBois did for the Beidler equation, a very similar output-input function to that of DuBois is obtained, R= f(C) is linear for (Rm–R) n a/K. (If Rm-R) o a/K, R becomes asymptotic to “a”. Dzendolet defined “a” as an arbitrary constant equal to Rm/N when all available receptor sites were filled. It appears the mass action equation of chemistry should be approached in differential form when used in calculations related to gustation and other transduction mechanisms. 8.5.6.3.2 Transduction kinetics of the amino acids

The amino acids play a major role in the transduction mechanism associated with gustation. Stone & Oliver employed two forms of the Beidler equation to investigate the human gustatory 220 Neurons & the Nervous System responses of a group of amino acids286. One employed linear scales to plot C/R= f( C). The other used logarithmic scales to illustrate a rearranged Beidler equation given by R/Rm = CK/(CK+1) and designed to show all of the data fitted a linear response proportional to the auxiliary function CK/(CK+1). The response, R, is clearly not linear with respect to the underlying concentration, C, by inspection. The authors employed the auxiliary function to avoid needing to account for how their data did not fit a single straight line over a significant range of concentrations. C represented molar concentrations in their stimuli. Figure 8.5.6-9 reproduces their figure 1. Note the scales are both linear and the ordinates are different for the two frames. most of the best estimate line for each amino acid is estimated from only a few data points. The slope of each line is an estimate of 1/Rm and the intercept with the vertical axis is an estimate of 1/KRm for that amino acid. Stone and Oliver asserted, “The data indicated that the evaluators rated the intensity of the stimuli in the expected manner; response rose rapidly and stabilized as concentration was maximized.” They also noted, “The evaluators also indicated their hedonic impressions and the taste qualities perceived.” However, this information was not included in this paper. A problem was also noted, “For some stimuli there was good agreement between evaluators (e.g., aspartic and glutamic acids, phenylalanine and arginine; for others there appeared to be an almost complete reversal, primarily at the lower concentrations.” Only five evaluators is a very small number for obtaining statistically significant data points. However, estimates were made. The rating system was crude, from 0 to 1.0; examples 0–no taste, 0.5– moderate intensity, 1.0– extremely intense following presentation of an identified standard with a rating of 0.50. A total of 20 responses was obtained from each evaluator for each concentration of each stimulus, but only the final 15 scores were retained for analysis. The amino acids were obtained from commercial sources (typically $98% purity) without any additional purification.

It is noteworthy that Shallenberger (1993, section 10.7.2) changed the modified form of R/Rm suggested by Stone & Oliver by omitting the term CK in the numerator on the right. He then proceeded to describe the resulting equation conceptually as a sigmoidal form. His analysis omitted the fact that the response, R/Rm is proportional to C below a transition point defined by KC = 1 and asymptotic to Rm for concentrations above this point. He also failed to describe the toe of his sigmoidal function as due to noise in the measurements. There is no real relationship between gustatory transduction and a sigmoidal function. The noise floor has been investigated by McBride287. His paper “demonstrates that the Beidler equation also provides a good description of the human taste response, as obtained by two psychophysical methods (JND cumulation, category rating).” It also addressed all four of the gustatory Paths defined in this work. The McBride paper relies upon a paper by Maes288, One of the major assertions of Maes after discussing the modifications introduced to simplify the plotting of the Beidler equation is that, “the linearizing plots should not be used at all for the quantitative evaluation of data. Direct, numerical iterative curve-fitting methods seem to give more reliable results.” He supports his position by noting, “even modest amounts of response variability may interfere badly with the evaluation of these plots.” He provides massive support for his arguements. No one has demonstrated the lower regions of the graphs of these modifications actually relate to the mechanisms of gustation. This work suggests these lower regions are noise limited regions just as they are in the visual and auditory modalities. See also Section 8.5.1.5.7 regarding just- noticeable-differences. Shallenberger (1993, section 10.7.3) also noted the danger of using so- called rating data when the data was plotted on an equal-interval(linear) scale with an arbitrary zero point.

286Stone, H. & Oliver, S. (1966) Beidler’s theory and human taste stimulation Percept Psychophysics vol 1, pp 358-360

287McBride, R. (1987) Taste psychophysics and the Beidler equation Chem Senses vol 12 (2), pp 323-332. doi: 10.1093/chemse/12.2.323

288Maes, F. (1985) Response noise affects the graphical evaluation of response versus concentration curves Chem Senses vol10 (1), pp 23-34. doi: 10.1093/chemse/10.1.23 Signal Generation & Processing 8- 221

The final estimates are shown in Figure 8.5.6- 10 along with estimates of the free energy change parameter for a presumed reaction where ΔF = –RTlnK. The values exhibit interesting groupings. The first six rows relate to on-polar amino acids. The last two rows relate to acidic amino acids believed to be critical to the electrostenolytic process providing power to the neural system (Section 3.2). Arginine is the only basic amino acid in the table. The last three are obviously distinct from the earlier amino acids in the table. The slopes of these last three amino acids are significantly different from the non-polar amino acids. ΔF for the non-polar amino acids is very low, suggesting a non-ionic bond with the sensory neuron receptor. Although not discussed in detail, Stone & Oliver suggest “the initial step is most likely adsorption, in agreement with Beidler’s conclusion (1954).” This work hypothesizes the first step involves a DACB between the stimulus and the neural receptor. The three substantially higher values for the bottom three rows of the table probably relate to a totally different mode of operation or the limited statistical precision of the raw data.

Figure 8.5.6-9 Beidler equation plotting data for amino acids, C/R versus C. Note the linear scales and different ordinate scales. The poins derived from the pooled data from five subjects. The method of least squares was employed to determine the equation for each line. From Stone & Oliver, 1966.

Figure 8.5.6-10 Amino acid slope values, maximum responses & equilibrium constants. The free energy is calculated from th equilibrium constant using the equation under the table. See text. The bottom pair of amino acids are acidic and involved in the electrostenolytic process powering the neural system. Arginine is the only basic amino acid in the table. The others are described as non-polar. From Stone & Oliver, 1966.

Shallenberger (1993, section 10.7.2) gave the parameter K a value of 9.8 (7–15) for NaCl stimulating rats when measured neurophysiologically, with the value depending on the protocol employed. Shallenberger (1993, section 10.7.2) gave the free energy change for NaCl binding 222 Neurons & the Nervous System to the N-Path receptor as –1.37 kJ/mol. and described this value as “reasonable value for an inonic interaction.” He did not indicate whether this value applied to NaCl before solvation or when in its fully hydrated ionic form. He did suggest the change could be attributed largely to a change in entropy. He cited the work of Simon that will be discussed in Section 8.5.6.3.5.

If the Stone & Oliver data was replotted to display R or R/Rm = f( C), the resulting curves exhibit a linear relationship below a breakpoint. For concentrations above this breakpoint, the response approaches Rm asymptotically. This is the typical saturation condition expected in the transduction process (Section xxx showing saturation in the reported P/D equation of vision ]. The Stone & Oliver representations do not support the analysis of Dzendolet, nor do the analyses of DuBois et al. discussed in the next subsection. However, Stone & Oliver did not discuss the equilibrium equation that they assume is the source of their data. The very low ΔF values suggest the DACB associations and dissociations involved occur at energy levels quite near the thermal noise level. A very sophisticated analysis of this situation may be required. 8.5.6.3.3 Comparing the human perceived response to simple sugars

Dubois et al. provided data in 1991 that has been redrawn in Figure 8.5.6-11 to show that using the molarity of molecules that hydrolyze upon going into solution is inappropriate. Fructose (mol. wt. = 180) is generally accepted to be the most sweet of the simple sugars. The ordinate scale is in percent concentration based on equal molarities of the dry sugars. However, the saccharides totally hydrolyze on going into solution. Thus the figure actually shows the responses of a variety of simple sugars on a scale of questionable concentration precision.

The best fit regression line for sucrose (a disaccharide subject to hydrolysis) shown in their figure 2 exhibits an intercept with the abscissa at 1.4% which is of doubtful theoretical importance and probably results from the difficulties of calibrating human perceptual responses. As a result of the intercept, he indicates a slope of the response to sucrose versus the stimulus to be 0.94. Except for this intercept, his regression line for sucrose overlays their fructose data very well. The slope of the regression line for fructose is given as 1.27. The slope for glucose is given as 0.60. D-Glucose is a monosaccHaride of the hexose family with a 6 carbon ring whereas D-sucrose (cane sugar) is a disaccharide containing one residue of D-glucose (blood sugar or corn sugar) and one of D- fructose (fruit sugar, also a hexose but with a 5 carbon ring). Fructose is generally found to exist in both furanose and pyranose forms (with about 20% furanose at 20C). DuBois et al defined their fructo-oligosaccharide sweetener as a footnote to their Table I. They gave the slope for the data shown as 0.27. DuBois et al., while discussing these slopes asserted that fructose exhibited a higher potency compared to that of sucrose. However, the slopes are nominally equal after considering the intercept problem in the sucrose data. There appears to be an uncontrolled variable present in this data. Each of the simple sugars of DuBois et al. is a gustant containing one or more gustaphores stimulating the G-Path via the GR 2 receptor of this work. To stimulate this sensory receptor, the requirement is that each gustant molecule exhibit one or more gustaphores of the two carbon equatorial-trans-diol form (See Section 8.5.1.6) Several of the gustants used occur in multiple forms in solution. α-D-glucopyranose (using the current systemic name) exhibits two gustaphores meeting this requirement. . β-D-glucopyranose (with the same molecular weight of 180) exhibits 3 and can be expected to be perceived as sweeter on a equal dry weight molar basis. Similarly, α-D-fructosefuranose exhibits two equat-trans-diol gustaphores while β-D- fructosefuranose exhibits only one equat-trans-diol gustaphore. Sucrose ( α-D-glucopyranosyl-β- D-fructosefuranoside) may hydrolyze in solution to exhibit two appropriate gustaphores due to the glucose residue and only one due to the fructose residue. Clearly these variables must be accounted for in any repetition of the DuBois et al. experiments involving simple sugars. The use of uncharacterized oligosaccharides are obviously not recommended as stimulants in such experiments. Single gustaophore gustants (SGG) or at least gustants with an equal numbe of gustaphores per equal molar quantity are preferred. Signal Generation & Processing 8- 223

Figure 8.5.6-11 The perceived response of humans to polysaccharides of nominal concentration based on their dry weight. The sucrose data from Dubois et al., figure 2 overlays the fructose curve but shows an intercept at a response of 1.4%. The alternate line with dog leg drawn through the glucose data is suggestive of a concentration near 25% was equal to the reciprocal value of the equilibrium constant (K) for this situation. See text. Composite from Dubois et al., 1991.

Taking the change in molar concentrations following hydrolyzation of the selected saccharide, the figure can be re-interpreted to indicate the average number of gustaphores present for each labeled sugar is important and the average number of gustaphores in the oligosaccharides (containng two to ten simple sugars) after hydrolyzation is typically lower than the mixture of fructoses and/or glucoses.

Based on the figure, the concentrations used by DuBois et al. apparently did not exceed the threshold where C = 1/K except there is an indication that the glucose data is approaching that threshold.

DuBois et al. also provided similar data for three very complex sugar alcohols (containing a heterocyclic structure of oxygen). While exhibiting an abundance of hydroxyl groups, they only exhibited a very few equat-trans-diol groups. Their perceived response versus concentration appears to reflect their very large molecular weights relative to their number of gustaphores. The concentrations had to be about twice as high to achieve a response comparable to that of the simple sugars. DuBois et al. considered a “Hill equation” as an alternative to the variant of the Beidler equation they presented. The applicability of the Hill equation of 1910 to various problems was reviewed by Weiss in 1997289. In general, the Hill equation is applicable to situations like the association of multiple atoms of oxygen to hemoglobin before the hemoglobin becomes active. He noted, “The Hill coefficient is best thought of as an interaction coefficient, reflecting the extent of positive cooperativity among multiple ligand binding sites.” He also makes the assertion, “Despite its appealing simplicity, the Hill equation is not a physically realistic reaction scheme, raising the question of whether it should be abandoned in favor of realistic schemes.” There is no known reason to consider the Hill equation with regard to individual receptor/stimulant binding or the binding of thousands, if not millions, of receptor sites on the dendritic structure of a given sensory neuron. It has been demonstrated on multiple occasions that the sensory receptors of the neural system are capable of reporting single events, be they individual visual photons, individual

289Weiss, J. (1997) The Hill equation revisited: uses and misuses FASEB J vol 11, pp 835-841 224 Neurons & the Nervous System audible phonons or minuscule stimulations applied to somatosensory receptors. In practice, such sensing is frequently obscured by noise either accompanying the stimulant (ex., photon noise) arising in the sensory receptor neuron. There does not appear to be any reason to consider the Hill equation in the context of gustation. The Hill equation reduces to the Beidler equation if no cooperation is required between multiple ligands of stimulant in order to achieve signal generation by the individual receptor neuron. The responses reported by DuBois et al. for both simple sugars and sugar alcohols appeared linear with respect to concentration up to the beginning of saturation in the number of empty and available receptor sites.

8.5.6.3.4 Comparing the human perceived response between sweeteners

DuBois et al. also provided human perceived responses to a list of artificial sweeteners of their time period (and including glycine for comparative purposes). There graphical data invariable exhibits a linear rise in the response as a function of concentration up to an inflection point where they became asymptotic to a constant response level. DuBois et al. used concentrations given in parts per million (ppm) rather than in molar units. The inflection points are given in ppm in Table II under the title, 1/K. These ppm values appear to track the increased effectiveness of these sweeteners (Section 8.5.xxx)

It can be argued that the Hill equation used to fit the data for three of the sweeteners can be replaced by the Beidler equation without seriously compromising the fit of the equation to the statistical error bars provided for the individual gustants.

While the response functions shown in figures 5 through 8 initially appear distinctly different from those for the simple sugars and sugar alcohols of figures 2 thorough 4, they are merely representations stressing different regions of the same underlying function. Note the significantly different scales for the ordinate axis among these figures. The different scales reflect two conditions; first, the significantly different sensitivity of the receptors to stimulants in an AH,B,X relationship to the less effective simple stimulants involving an AH,B relationship, and to the inflection point located at 1/K. The three functions shown in Figure 5 for alitame, aspartame and sucralose exhibit an intercept on the ordinate axis that does not go through zero concentration. The later figures do not show such an intercept. A major assertion of the DuBois et al. paper is, “it is tentatively concluded that at least two routes to receptor cell activation must exist.” The AH,B and the AH,B,X mechanisms appear to support this assertion. 8.5.6.3.5 Mechanisms of sweet taste transduction from Simon

Simon has provided a very comprehensive proposal related to the sweet transduction mechanism based primarily on his background and conceptualization290. It is sophisticated but does not provide a physiological model of the process other than the conventional sketches of neurons with ions and messengers traveling in diverse directions. He proposes to consider saccharides alone as well as with or without accompanying salts and with transduction only on the surface of the dendrites or also by passing through tight junctions in order to stimulate the basolateral membrane of sensory neurons. He then limits his investigation to only glucose and sucrose in the presence or absence of salts (apparently only NaCl). As of 1991, he also asserts, “Since receptors for naturally occurring saccharides have not yet been isolated rom lingual epithelia, there is no direct evidence for their existence.” He also asserts, “One reason for questioning the existence of receptors is that D– and L–glucose are equally sweet.” This is hardly a viable reason for doubting the existence of a type of receptor. It is a better reason for doubting his concepts of transduction.He does provide strong evidence that saccharides do not bind effectively to proteins, based on his presumption that proteins are an element involved in transduction.

290Simon, S. (1991) Mechanisms of sweet taste transduction In Walters, D. Orthoefer, F. & DuBois , G. eds. Sweeteners. Wash. DC: American Chemical Society Chapter 18 Signal Generation & Processing 8- 225 While his page 238 can be largely discounted due to limitation in his model(s), selected portions of the data related to thermodynamics is quite valueable in supporting the hypothesis of this work. His comments concerning the effects of amiloride on the transduction of saccharides will be addressed in Section 8.5.7.3. Simon proposed two distinct conceptual models; In the first model, the binding of saccharides to receptors opens an amiloride-inhibitable channel permitting small cations to enter and depolarize the taste cells. In the second model, the binding of ssaccharides to receptors activates a second messenger cascade that either closes potassium channels and /or opens chloride channels. In the second model, taste transduction can occur inthe absence of salts.”

Neither of these models employs the DACB relationship of Shallenberger and this work; neither the AH,B or AH,B,X configurations were considered. Neither of these models appears viable based on this work; however, much of the information in Simon is applicable in support of the hypothesis of this work. This assertion is based on his basic premise concerning the stimulation of sensory receptor neurons,

“It was demonstrated that the interaction of saccharides with taste cells depolarise them. The depolarization of taste cells implies that ions must flow across their cell membranes. Hence ions (salts) are involved in the response to saccharides (emphasis added).”

The implication is not the only option. Further more ions are not salts, they are ions that may have originated in dry salts before solvation. Simon clearly does not appreciate the options under the Electrolytic Theory of the Neuron. Holes (the absence of electrons) in a structured membrane can imitate positive ions but are identifiable by differences in speed of transport. Unhydrated Ions generally can not cell membranes. They must first be combined with other lipids or proteins. Hydrated sodium salts exibit unique d-values that allow them to form DACB bonds with the organic ligands of the sensory neuron receptors (Section 8.5.xxx). Simon then asserts that “the remainder of this manuscript will deal with the ion currents that are induced by the interaction of saccharides with taste cells.” This Bayesian assumption negates the value of this paper.

Simon illustrates six distinct conceptual models of how the sensory receptor neurons might operate based on the currently discussed chemical theory of the neuron. The figure was attributed to J. Verbrugge. He then discusses each one determining the difficulties associated with each variant. Multiple ions are entering or leaving each caricature at various undefined times and at various undefined rates. He also notes, “Taste cells also contain transport pathways not mentioned above.” As a result, the models are less than satisfying to a biophysicist. 8.5.6.3.6 Solubility of the natural sugars from Andersen et al.

Figure 8.5.6-12 reproduces the data of Andersen et al291. on the solubility of various sugars as a function of temperature. The data was collected from a variety of sources and more recent data may be available292. They showed significant differences exist at all temperatures of biological interest. In a separate table they showed the relative solubility of these natural sugars as measured by multiple investigators. The sequence of solubilities was routinely given as D- fructose>sucrose>D-glucose>maltose>D-galactose>lactose. The slopes of the solubilities are uniform with the exception of D-galactose and with some proclivity for D-glucose and maltose to show a higher slope at higher temperatures. A noted feature is the uniformity of both the monosaccharide and disaccharide members of this family.

291Andersen, H. Funakoshi, M. & Zotterman, Y. (1963) Electrophysiological responses to sugars and their depression by salt In Zotterman, Y. ed. Olfaction and Taste. NY:

292 Yalkowsky, S. He, Y. & Jain, P. eds. (2010) Handbook of aqueous solubility data Boca Raton, FL: CRC Press Also a 2003 edition. 226 Neurons & the Nervous System

Andersen et al. asserted, “These results indicate that the disaccharides are not split on the tongue or at the receptor sites when the stimulation of the proper receptors takes place.” This assertion appears to be incorrect and based on the observation that the molarity is changed when the disaccharides are hydrolyzed but the sugars as gustaphores are not. Based on this work, hydrolyzation leads to both a change in the molarity of the solution and an equal or similar creation of more gustaphores due to the creation of more equat-trans-diol gustaphore sites on the increased number of molecules (Section 8.5.6.3.3). The net effect is no significant change in activity level with hydrolyzation of most polysaccharides. The massive compilation of Yalkowsky et al. suggests that the solubility of the sugars is still not a settled science. Variations of up to 5:1 between values at a given temperature fora single sugar are not uncommon, even when measured by the same investigator during the same year (ex., fructose). Signal Generation & Processing 8- 227

Figure 8.5.6-12 Solubility of the natural saccharides versus temperature. A collage from multiple sources. From Andersen et al., 1963

8.5.7 Antagonists (blocking agents) & adaptation in the gustatory modality

Based on the previous analyses, it is apparent that there are a variety of means of interfering with the normal operation of the gustatory modality. These include, C chemically capturing a gustant prior to its effective stimulation of the sensory receptors. 228 Neurons & the Nervous System

C mechanically interfering with a gustant physically reaching the sensor receptors. C physically isolating the sensory receptors by causing a closure of the entrance to the taste buds. C chemically occupying the receptor sites on one or more types of receptors for an interval of time. C reducing the sensitivity of the sensory neurons by interfering with their electrostenolytic mechanisms electrically powering the neurons. In the temporal domain, it is also possible to depress the sensitivity of a given gustatory channel through the process of adaptation. High intensity stimulation of the channel for only a short period will cause a major reduction in its sensitivity for a prolonged period, based on the attack and decay time constants of the sensory receptor neurons. A variety of terms have been associated with the above mechanisms, mediation, amelioration, suppression, blocking, etc. One chemical playing a major role in this area is amiloride, most probably named by association with amelioration.

To avoid interfering with operation of the gustatory modality, the investigator should be careful not to employ nocents (such as inorganic acids, including HCl) or astringents (such as the alkali earth salts) in their protocols. See Section 8.5.4.3.5.

As in the process of powering the individual neurons, there are a series of antagonists that operate to impede the stimulation by various gustants and gustaphores. As dopamine is an antagonist to the neuro-facilitator glutamic acid, it has been asserted that amiloride is an antagonist to the sodium ions in exciting the sodium (N–) channel of gustation. There are other antagonists to the gustatory modality that will be addressed below.

Shallenberger (1993, section 10.8) has provided a valuable review of potential antagonists and their point(s) of action.

Recent computer generated representations of amiloride suggest it may not be an antagonist to natrophores but is in fact a strong picrophore that may distract the individual from perceiving the intensity of the natrophore. This will be discussed further in Section 8.5.7.2.

The mechanism is slightly different in gustation compared to neural fueling. But the result is largely the same. The antagonist is able to form a dual antiparallel coordinate bond with the sensory receptor site and prevent the appropriate gustaphore from reaching the receptor.

The ability of amiloride to interfere with the hydrated sodium ion is interesting because amiloride contains no sodium. It is a totally organic molecule that is able to bond to the totally organic sensory receptor site. Amiloride (DB00594) contains one oxygen and seven nitrogen orbitals. It offers an abundant number of orbital pairs that could provide a d-value near that of the hydrated sodium receptor at 3.24 Angstrom. Breslin, in his Firmenich Award Address has provided some data on the properties of sucrose and urea when mixed and when impacted by the presence of 0.3M sodium acetate293. When mixed, they tend to suppress the perception of each other. When sodium acetate was added, the perceived response of sucrose was nearly restored while the perception of urea was further suppressed.

8.5.7.1 Generic blocking agents

293Breslin, P. (2001) Human gustation and flavor In Spanier, A. et al. eds. Food Flavors and Chemistry. Royal Soc Chem pg 47 Also published in Flavor & Fragrance J. Signal Generation & Processing 8- 229 Andersen et al. reported on the suppression of the sweetness perception by salt294. They experimented on dogs under the assumption that all gustatory signals passed along the chorda tympani. Breslin has addressed the question of blockade of gustatory stimulants295. He also discusses enhancement, synergy, suppression and masking by multiple agents simultaneously. Although discussing, chlorhexidine, gymnema sylvestre, lactisole and a combination labeled PALG, the emphasis is on amiloride. It is important to determine what mechanism from those listed above is being considered by the following investigators. Are they blocking access to the receptors, blocking the formation of a DACB between gustant and receptor, or interfering with the X site in an AH,B,X relationship to overcome the capability of the super-sweeteners?

Breslin describes two known sweetness suppressants. He notes that gymnema applied to the oral cavity prior to application of a sweet tasting substance results in suppression of the perception of sweetness. He also notes lactisole is a fast-acting competitive antagonist to almost any compound tasting sweet. Lindley has written on the chemistry of anti-sweeteners in 1991296 and 1993297. His 1991 introduction discussed several molecules and their potential means of interfering with gustation, even addressing the subject of detergents applied to the sensory receptors. Unfortunately, the common names he used for his molecules are not in common use now and the gustants were diluted in 5% or 10%(w/v) sucrose. His procedure employed SAR techniques. His conclusions are well stated and suggest his inhibitors are competing for the same receptor locations as the sweeteners being inhibited. He did note that in the absence of an AH group within an AH,B structure, the abilitiy of his inhibitors was eliminated. Based on his conclusions,”then it is a logical extension to conclude there is a single receptor structure that ‘codes’ for sweetness.”

In 1993, he summarized as follows, “Currently available evidence is consistent with the conclusion that these sweetness inhibitors are competitive antagonists of sweet taste acting at a single receptor structure.”

Breslin asserts the sodium ion is the primary known blocker of the bitter taste sensation. He offers no description of the mechanism causing this action. Blocking in this manner could be related to a neural differencing in either the dimension 1 = Q – H or dimension 2 = N – H channels described in the multidimensional analyses above. The importance of the differencing might be less than that due to the summation of the various stimuli joining in forming the overall taste sensation.

He discusses PALG, a protein (especially lactoglobulin) bound to phosphatidic acid, as a good blocker of bitter tasting compounds, especially quinine which it precipitates. His terminology is that of a pharmacist or specialists in food product development. The designation phosphatidic acid describes a fatty acid lipid that is a precursor for many of the potential fatty acids in the sensory neuron lemma. Quinine in soluble form generally relates to quinine hydrochloride. This may be an excellent clue to the operation of the P-Path sensory channel since phosphatidic acid is a potential transducer in the type 4 lemma of the P-Path sensory neuron microvilli. If the phosphatidic acid is reacting with the quinine hydrochloride with the precipitation of quinine, the phosphatidic acid may become chlorinated and become negatively charged. This action could

294Andersen, H. Funakoshi, M. & Zotterman, Y. (1963) Electrophysiological responses to sugars and their depression by salt In Zotterman, Y. ed. Olfaction and Taste. NY:

295Breslin, P. (2000) Human gustation In Finger, T. Silver, W. & Restrepo, D. eds. The Neurobiology of taste and smell, 2nd Ed.. NY: Wiley-Liss Chapter 16

296Lindley, M. (1991) Phenoxyalkanoic acid sweentess inhibitors In Walters, D. Orthoefer, F. & DuBois , G. eds. Sweeteners. Wash. DC: American Chemical Society Chap 19

297Lindley, M. (1993) Sweetness antagonists In Acree, T. & Teranishi, R. eds. Flavor Science. Washington, DC: American Chemical Society Chapter 4 230 Neurons & the Nervous System mirror the nominal reaction between the phosphatidyl fatty acid of type 4 lemma and quinine hydrochloride, resulting in the precipitation of the quinine, permanent alteration of the phosphatidyl fatty acid and a significant change in the flow of current into the plasma of the microvilli. A comparison of the state of polarization of phosphatidic acid and both PtdCho and PtdSer would appear to be useful. As Breslin noted (page 437), “. . .a potent selective blocker of salty taste in humans remains to be discovered.” He also noted (page 438), “There are no known blocking agents for acid sourness or umami in humans at present (2000).” The theory of this work offers a variety of new approaches to discovering, or designing, such gustaphores xxx. In 2001, Breslin & Tharp presented a paper on the potential for blocking the N-Path and the P-Path by a complex nitrogen molecule, chlorhexidine298. They note, “Psychophysical studies using pharmacological blockers of specific taste qualities (e.g. bitterness or sweetness) hold the potential to provide insight into both the type and number of transduction mechanisms. For this technique to highlight a particular component of taste physiology, two prerequisites must be met. First, the agent must specifically block a taste quality or qualities and not taste in general. Second, the pharmacological agent must have a known biochemical action that could be effective on taste physiology.” xxx 8.5.7.2 Gymnemic acids as G-path blockers

Quoting Wikipedia, “Gymnemic acids are Glycosides isolated from the leaves of Gymnema sylvestre (Asclepiadaceae). Gymnemic acids like ziziphin and hodulcine are anti-sweet compounds, or sweetness inhibitors. After chewing the leaves, solutions sweetened with sucrose taste like water.

More than 20 homologues of gymnemic acid are found in the leaves. Gymnemic acid 1 has the highest anti-sweet properties. It suppresses the sweetness of most of the sweeteners including intense artificial sweeteners such as aspartame and natural sweeteners such as thaumatin, a sweet protein. The anti-sweet activity is reversible, but sweetness recovery on the tongue can take more than 10 minutes.”

The gymnemic acids appear to occupy the G-path GR sites much like Glycine and the catacholines (Sec 3.5.5.3 & 3.5.5.4) do in occupying the glutamate receptor sites in the electrostenolytic process fueling the neurons themselves. It is generally claimed that these chemicals as a group do not affect the other pathways of gustation. Shallenberger (1993, section 10.8) has reviewed the sketchy data available on whether gymnemic acid operates as a competitive or non-competitive inhibitor. Since it must be applied to the sensory receptors before the stimulant of interest, it appears it occupies the sensory neuron receptor sites prior to the stimulus attempting to occupy those same sites. The fact that washing out the impact of gymnemic acid on the sensory receptors may require time periods of up to ten minutes also suggests the acid has captured the receptor site.

The ability of lactisole, a dual gustaphore (triple if the sodium ion is present and hydrated) gymnetic acid is shown in Figure 8.5.7-1, to block normal sugar sensitivity at the GR is obvious from its configuration. It is claimed by some ( ? A dubious claim at best) that this chemical has no affect on the other gustatory pathways.

298Breslin, P. & Tharp, C. (2001) Reduction of Saltiness and Bitterness After a Chlorhexidine Rinse Chem Senses vol 26, pp 105–116 Signal Generation & Processing 8- 231 Hellekant et al. (1985) have described the ability of miraculin to overcome and/or alleviate the effects of the gymnemic acids299. Quoting Wikipedia, “Miraculin is a natural sugar substitute, a glycoprotein extracted from the fruit of Synsepalum dulcificum. The berry, which contains active polyphenols, was first documented by explorer Chevalier des Marchais, who searched for many different fruits during a 1725 excursion to his native West Africa. Figure 8.5.7-1 Lactisole, a sweetness blocker. The Miraculin itself is not sweet. However, after O-C-C=O structure is a glycophore that can form the taste buds are exposed to miraculin, a DACB with the GR but it does not readily release ordinarily sour foods, such as citrus, are from this bonding like other glycophores, thereby perceived as sweet. This effect lasts up to an preventing its replacement by other molecules. hour. Miraculin works by binding to the sweet receptors on the tongue. Miraculin's effect lasts as long as the protein is bound to the tongue, which can be up to an hour. It makes most acidic foods taste sweet, but does not improve the taste of bitter things.” 8.5.7.3 Affect of amiloride on the monkey & other species

Hellekant et al. (1988) addressed the affect of amiloride on the perception of sweetness to a variety of stimulants in the monkey, Macaca mulatto, at the chorda tympani of the monkey specifically300. As noted earlier, there may be a question as to whether the amiloride blocked the natrophore at the GR of stage 1 or actually overwhelmed the perception of saltiness due to the intensity of the amiloride acting as a picrophore at the output of a stage 2 neuron. Additional analysis of the protocol used by Hellekant et al. may be required to determine the character of the nerve they interrogated.

Avenet & Lindeman have provided additional information from the laboratory on the role of amiloride in taste related to several species301. Their temporal response measurements and current/voltage plots are highly supportive of the electrolytic theory of the neuron. and show the activity resulting from changes in the presence of amiloride clearly. The effect of amiloride on the voltage/current characteristic is quite clear.

[xxx see Simon, pg 239+ in Walters with regard to amiloride and the sugars and salts. ] Simon has noted several variants in the response to amiloride among the mammals (page 239) that make dependence on generalizations in the literature difficult. Later on page 248, in discussion of his Direct Activation Model concept and his figure 1a, Simon asserts, they “suggest there are two saccharide-stimulated chorda tympani (CT) pathways that can be distinguished by their amiloride sensitivity (emphasis in the original). Thus it appears that the amiloride-sensitive component is salt dependent and the amiloride-insensitive is salt (mucosal) independent.”

8.5.7.3.1 Current problem related to the available archives and visualizers

299Hellekant, G. Segerstad, H. Roberts, T. van der Wel, H et al. (1985) Effects of gymnemic acid on the chorda tympani proper nerve responses to sweet, sour, salty and bitter taste stimuli in the chimpanzee Acta Physiol Scand, vol 124(39), pp ~ 408+

300Hellekant, G. DuBois, G. Roberts, T. & van der Wei, H. (1988) On the gustatory effect of amiloride in the monkey (Macaca mulatto) Chem Senses Vol.13(l) pp.89-93

301Avenet, P. & Lindemann, B. (1988) Amiloride-Blockable Sodium Currents in Isolated Taste Receptor Cells J Membrane Biol vol 105, pp 245-255 232 Neurons & the Nervous System

Figure 8.5.7-2 provides two totally different representations of amiloride (DB00594 from the Canadian Data Bank), both showing the molecule as planar. The left frame using the Jmol visualizer suggests the lower two outer nitrogen atoms can act as a acidophore with d-value = 2.86 Angstrom but no pair capable of interfering with a natrophore was found. No atom pair offering a d-value suitable for interfering with a natrophore was identified. The right frame using the DS3.5 visualizer shows the same molecule, from the same Jmol file. In this case, the two lower outer nitrogen atoms have a d-value of 4.723 Angstrom that is better centered on the P-path channel of d-value = 4.746 Angstrom than its only potential natrophore of d-value = 3.599 Angstrom is centered on the N-path channel d=value of 3.3 Angstrom. Simple hand calculations using accepted bond lengths strongly suggests the Jmol visualizer should not be relied upon even though the underlying file may be correct, as represented by the DS3.5 visualizer. Using DS4.1 to visualize the similar Jmol file from Chemspider for amiloride_15403, gives a totally different conformation and significantly different d-values than either of the above sources. The Protein Data Bank (pdb) files related to muco-inositol, like other specific conformations of multi-conformational molecules, are also difficult to find in the literature, and many of those on prestigous sites, including the NIH, are corrupted (referring to a different stereoisomer than muco- inositol) as of November, 2012. An edited conversation with one of these curators said; ’‘In fact, practically all of the contents in our database is not generated by ourselves but aggregated from other databases. By the same token as above, errors/weaknesses in these databases therefore unavoidably find their way into our services until someone points them out to us.

If you search ( ) with "inositol", you'll find exactly the same situation: "Inositol - Compound Summary (CID 892) Also known as: myo-inositol, meso-Inositol, Scyllo-inositol, Muco-Inositol, Allo-inositol, i-Inositol, Myoinositol, epi-Inositol, mesoinositol"

Most likely, therefore, we got the data in ( ) from ( ). ( ( ) does distinguish between stereoisomers of inositol; though we have found many other errors in ( ) for other molecules - no one service/site is perfect in this regard.)

One little trick to test what we have before downloading an SD file (especially with 3D coordinates) is to first just request a (2D) drawing of a molecule. You'll see that all six stereocenters are undefined. What this means is that the 3D mol/SDF generator will simply *create* some default stereochemistry and geometry.”

In the majority of this work, Jmol files downloaded from ChemSpider will be relied upon. ChemSpider is the website of the Royal Society of Chemistry (RSC) in Great Britain. The files will be visualized using the Design Studio visualizer, DS3.5 (up until July 2015 and DS4.1 thereafter), as indicated earlier (Section 8.5.4). As of 1 December, 2015, the RSC has taken down their 3D representation of amiloride_15403 in their JSmol archive entirely along with probably all other JSmol representations of molecules.

See Section 8.5.1.1.2 for additional important information regarding the RSC files. Signal Generation & Processing 8- 233

Figure 8.5.7-2 Amiloride (DB00594) as represented by a Jmol file from the Canadian Data Bank using two different visualizers. Note the totally different distances between the same atoms. Left; using the Jmol visualizer. The lower two outer nitrogen atoms exhibit a d-value = 2.86 which is suggestive of a acidophore. Right; using the DS3.5 visualizer. The pair of nitrogen atoms with a d-value = 4.723 Angstrom is a nearly ideal picrophore. The other two pairs shown are relatively ineffective, but could achieve a DACB with the N-path GR. See text.

8.5.7.4 Adaptation and/or suppression by antagonists in gustation EMPTY xxx in his Firmenich Award Address (page 44) has addressed the definitions of adaptation and cross-adaptation in the context of gustation as used in the food industry.

[xxx significance of adaptation of individual sensory paths and the resultant after-taste, colors, and other terminology.

Evaluation by human perception of monosodium glutamate and other gustants containing three distinct gustaphores becomes extremely difficult. The requirement to control the state of adaptation of all of the affected channels before they are stimulated is a difficult one. The requirement to avoid desensitization of any of the sensory receptors associated with these channels by antagonists may be even more difficult.

8.5.8 Analysis of the literature based on the hypothesis

The empirical literature seeking an understanding of gustation is immense. However, much of that reviewed below emanated from a group centered at Pennsylvania State University (Pfaffmann and then Travers) and at the University of Wisconsin (Danilova and Hellekant) over a period of many years. Most papers totally ignore any theoretical framework for explaining the collected data. This section will review several groups of papers prepared by teams over a period of years. As time has gone by, it is easy to see the improvement in their test protocols, their data presentation and graphics. representations The framework developed in this work and confirmed by the recent empirical record shows that hydrochloric acid, HCl, does not play a role in the gustatory modality, except in its fully hydrated form (Section 8.5.4.3.5). The gustatory modality does not sense HCl and it is not reported in signals from the chorda tympani. HCl is an inorganic acid that attacks the neural system and attempts to digest biological tissue. Like capsaicin, it is classified as an 234 Neurons & the Nervous System

irritant. Both are sensed by the nociceptor modality. While HCl is also found in the digestive tract and is neutralized before reaching the large intestine, capsaicin is know to irritate the entire alimentary canal. On the other hand, the empirical investigations have lacked a theoretical framework. A particular problem has been the fact that the gustatory modality involves four orthogonal dimensions, the three dimensions supporting the selection mechanism and the one dimension describing the intensity of the total perceived sensation at the output of the saliency map. Assuming the modality only involved three dimensions has left all of the collected data in disarray, particularly with regard to MDS representations but also with regard to what stimuli were included in their stimulant sets. The MDS representations in the literature cannot be effectively compared because they have employed arbitrary “dimensions” selected by the computer programs and have distributed the values associated with the fourth dimensions among the values presented within the assumed three dimensional perception space. In future experimentation, it is necessary to allow for four dimensions in the MDS program resolving the collected data. The three dimensions containing the four nodes of the signal pathways can then be transferred to the presentation program of choice. At this point, any rotation, reflection and scaling can be performed to optimize the presentation. These procedures must be applied to the four dimensional data set. After these procedures have been completed, the values associated with the fourth dimension can then be displayed as a one-dimensional plot representing the perceived intensity for each of the stimulants. The one-dimensional plot as a function of the fundamental dimension (the d-value line) will be similar to figure 5 in in Hellekant et al. (1997) except the stimulants will be clumped around their appropriate node along the d- value line.

The added definition of the gustatory neural system and the definitized hypothesis based on the Electrolytic Theory of the Neuron presented above, requires a review, and reinterpretation, of the past literature. This is particularly true based on the recognition that;

• there are a minimal number of gustatory channels within the neural system (4) • that individual stimulants may exhibit multiple gustaphores of the same or mixed types and • that a combinatorial mechanism is employed to support a multitude of gustatory stimulants.

The demonstration of a long sought fundamental dimension ordering the effectiveness of the gustatory modality is also critically important in the reinterpretation of the literature.

An additional, critically important, factor is where the data was collected. The optimum data would be collected at the output of the stage 1 sensory neurons. However, this is extremely difficult while maintaining the integrity of the cells to normal stimulation. The next best data is obtained from the stage 3 neurons emanating from the taste buds and typically traveling along the chorda tympani before it is merged into the facial nerve (n. VII or IX depending on investigator). This data is not contaminated by any known stage 2 signal processing. It is in pulse coded format. Data may also be acquired from the glossopharyngeal nerve before it is delivered to the brain stem. However the opportunity for contamination of the signals is present because this nerve carries so many other signals (both afferent and efferent). Examining signals at the nuclei tract of solitarius (NTS) or the parabrachial xxx 8.5.8.1 Reinterpretation of Smith et al. of 1983 using hamsters

Smith, van Buskirk, Travers & Bieber issued two papers in 1983 that can be reinterpreted based on the hypothesis and supporting material presented here. A major problem with interpreting their data is the lack of statistical error bars associated with their raw data. Much of the data involves very small signals that cannot be interpreted with precision (parts of figures 3, 4, 6 & 7). As a result, the dendrograms and the 3D representation of their data does not always agree as indicated in the footnote to table 1 of the first paper. All of their data was acquired at either the nucleus tractus solitarius (NTS) or at the parabrachial nuclei (PbN), locations within the brainstem. No attempt was made to determine whether the collected signals were at the input to or at the output of these stage 2 signal processing engines. Nor was their any assertion that the signals were not corrupted at the input or only involved gustatory signals at the output of these engines. Signal Generation & Processing 8- 235 It is difficult to demonstrate the data is not corrupted from data supplied from nociceptors and other types of sensory neurons at these locations within the brainstem. Stimuli were only provided to the anterior portion of the tongue. They employ the historically recognized terms for H-best, Na-best, sucrose-best and quinine-best neurons. As they note, “Such an approach may well overlook other potentially important characteristics of a neuron.” In fact it also overlooks important characteristics of the stimulants as well. Their selection of stimulants shows a distinctly Bayesian approach to their experiments. The stimulants, and the molarity of each, were selected based on their expectations of what the results would look like. Their cluster analyses were performed using the BMDP2M software package and their MDS analyses employed the KYST package without any description of the default, and/or established, parameters for these programs. Their dendrograms included a number of anachronistic inclusions. For the NTS neurons, they only identified three major clusters, H–, S– and N –. The N – cluster was subdivided into three sub- clusters. Six neurons were included within these N – subclusters even though they were identified as either H-best neurons or S-best neurons. Their raw data in figure 4 shows distinctly that the four neurons labeled H-best were in fact not H-best. #23 was maximally sensitive to sucrose, the hydrated sodium ion and to HCl. #24 showed similar sensitivity. #26 & #27 exhibited very low signal intensities with peak sensitivities to sodium and HCl in the first case and HCl alone in the second case.

The sensitivity to HCl suggests these neurons are responding to nociceptors (see the comparable data of Hellekant et al. in the next section where HCl played a negligible role in data collected earlier in the signaling channels). Neuron #23 strongly suggests it is reproducing signals at the output of the NTS engine.

For the PbN dendrogram, a similar situation is found. Only three of the four historical clusters were found using their stimulants and software. One ostensibly Q-best neuron (#31) was placed in an H-cluster. According to figure 6, this neuron exhibited marginal sensitivity with some sensitivities registered as negative relative to the quiescent neuron output. A negative output is normally associated with the output of a differencing circuit within a stage 2 engine rather than the unidirectional signals associated with a stage 1 sensory neuron. #20 and #23 of the PbN neurons exhibited very minimal responses and are difficult to categorize authoritatively. Both #16 and #17 exhibit very broad sensitivities to sugars, salts, HCl and urea in the first case and to all but the sugars and quinine in the second case. #24 although described as H-best but found in the N- cluster, shows significant sensitivity to the hydrated sodium ion (nominally as great as to HCl). #30 shows negligible response. The response to HCl is only marginally higher than its response to hydrated sodium. These responses as a group suggest, the signals are recorded at the output of the NTS and PbN engines where the signals would be at maximum strength as they are encoded in stage 3 pulse form for projection to other engines farther up the brainstem.

The first paper does not include any 3D representations based on MDS analyses. It does present two 2D representations. However the distance between the clusters is of the same magnitude as the diameter of the clusters with the shape of some clusters significantly impacted by one neuron at a considerable distance from the other neurons in that cluster. While they only found one Q-best neuron, which they attribute to only stimulating the tip of the hamster tongue, the following Hellekant et al. paper found multiple Q-best (or P–channel) neurons in the tip of the monkey tongue. Smith et al. generally support the labeled-line presumption relative to gustation but do not make any strong assertion. In the second paper, the authors make a strong statement in their introduction that, “It is almost without exception that mammalian taste neurons respond to stimuli representing more than one of the classical four taste qualities.” They accompany this statement with a list of citations but no detailed discussion. This position is not supported by this work or by the following paper by Hellekant et al. They do make the important assertion that, “Since the response of most taste fibers can be modulated by both stimulus quality and intensity, the response of any particular cell is typically ambiguous with respect to either parameter.” This position is supported here and is the 236 Neurons & the Nervous System main reason investigators attempt the standardize the molarity of their stimulants relative to the pulse rate they encounter for specific neurons. Most of the data in their figure 1 & 2 reflect very broad sensitivity for individual neurons. They did provide one 3D representation of an MDS analysis reproduced here as Figure 8.5.8-1. Lacking a larger data set, the dashed circles can be used to indicate the mean location of the particular odorophores affecting each sensory receptor channel. The data set can be rotated to bring the vertical axis parallel to the vertical dashed line. Then the X,Y plane can be rotated to bring it into registration with the heavy dashed lines of the X and Y axes. Alternately, the heavy dashed lines can be considered the fundamental axes of the perceptual gustatory space for the data points shown. If HCl remains in this group, it must be recognized that its gustaphore results from its hydration (becoming an azeotrope in order to be sensed by the gustatory modality). Signal Generation & Processing 8- 237

Figure 8.5.8-1 Three-dimensional space showing the location of 18 stimuli. Note the location of the sugars compared to the organic acids. Note the separation between the ionizable salts of ionic sodium at lower right and Na-saccharin at upper center. Urea and QHCl occupy a distinctly separate area. The dashed circles suggest the mean location of the particular group of odorophores. The heavy dashed lines suggest the degree of rotation of the axes required to bring the data into alignment with the preferred coordinates of Section 8.5.1. Data and original coordinates From Smith et al. 1979.

8.5.8.2 Reinterpretation/expansion of Rohse & Belitz of 1991 238 Neurons & the Nervous System

The text of the paper by Rohse & Belitz302 was brief and only examined similarities between very complex molecules perceived as sweet or bitter without any discussion of underlying theoretical principles. They did note the importance of an “e/n system” relationship implying the importance of an electrophilic and nucleophilic components in the transduction process. They described a number of candidate e/n systems including, 1. OH/OH in sugars, + – 2. NH3 /COO in amino acids and peptides, 3. NH/SO2 in oxathiazinone, 4. OH/CO in β-hydroxyketones, – 5. CONH2/COO in carboxyalkyl benzamides, 6. CONH2/OR in alkoxyphenyl ureas and 7. NH2/CO in ureas. They did provide distances between these different group pairings. However, they did not indicate the points from which their distances were measured. Furthermore, they did not recognize the minimal properties required for sweetness within these molecules. Nor did they speak in terms of an overlay group structure associated with their pairings. Such an overlay structure can be critical because it can affect the net distance in 3D space between the actual orbitals. They also failed to address the participation of resonant structures in their computer modeling.

A revised (condensed) and expanded (to include the resonant structures) description of the overlay groups required for the perception of sweetness is suggested below,

[xxx edit is the H in NH groups really necessary? ] Electrophilic structure Nucleophilic structure Shorthand e/n notation

OH (ex., 1, 4 above) O OH/O (using AH,B concept) + NH3 O (of carbonal group) NH2 NH O (of either sulfxxx group) OH O (of carbonal group) OH/O

OH Any C=C double bond OH/C=C OH Any Huckel structure (resonant) OH/Ar ring

The OH group is susceptible of forming a hydrogen bond with a nucleophilic element of a separate molecule, resulting in a London bond. In a London bond, the hydrogen is held to one oxygen by a covalent bond and the other oxygen atom by purely electrostatic forces. The net bond strength is on the order of 5 kcal/mole compared to 50-100 kcal/mole for most covalent bonds (M&B, 1971, section 15.5). The distance between the two oxygen atoms, that are individually capable of participating in additional London bonds. is typically 2.703 Angstrom. As noted below, the ability of many saturated alcohols and aldehydes to form azeotropes in this manner accounts for their typically mild sweet taste. The chemistries of nitrogen and sulfur become very complex compared to that of oxygen in their roles in gustation. Their chemistries must be analyzed in greater detail than can be addressed here (See M&B, 1971, sections 11.15-18) The role of sulfur in gustation , and particularly the complex chemistry associated with sulfur bonded to oxygen within a sulfonyl group), has not been studied in this work. There are a variety of ways to satisfy the Huckel rule (4n + 2, n=0,1,2,3,etc.) within a molecule. The six sided benzyl ring is by far the most common.

302Rohse, H. Belitz, H. (1991) Shape of sweet receptors studied by computer modeling [1991] xxx Signal Generation & Processing 8- 239 The above listing does not include a pairing of two C=C double bonds in a DACB relationship and caution should be employed when determining the d-values of many compounds. The Rohse & Belitz paper is a transitional step between the work of Shallenberger & Acree and of Kier and this work. It only addresses a small group of large molecules perceived as sweet plus a few similar molecules perceived as bitter (without providing any definition of these terms. The paper is compatible with the more detailed hypothesis of this work. 8.5.8.3 Reinterpretation of Hellekant et al. of 1997 using M. mulatta

The paper of Hellekant et al. is very well constructed, provides statistical error bars on its basic data and interprets that data with great care. However, it lacked the fundamental dimension provided by the d-values of the DACB for their stimulants. It also overlooked the fact that one of their stimulants, sucrose is actually a disaccharide and its gustaphore molarity after hydrolysis in water is twice the molarity of sucrose. This fact is clearly reflected in their data in figures 5D and 10C. To compensate for the effect on molarity of hydrolysis, hydrolysis also has the potential for increasing the number of gustaphores in the solution by a factor of two. They collected data at both the chorda tympani (CT) supporting the anterior part of the tongue and the glossopharyngeal nerve (NG or xxx) supporting the posterior portion of the tongue. Hellekant et al. used several guanidine derivatives in parts of his investigation but does nor show them in his cluster analyses or MDS results. These derivatives were assigned numbers rather than names. Unfortunately, the NC-00174 and NC-00351 numbers are also used for other chemicals in the general field of chemistry. The chemical SC-45647 (C18H26N4O2, mol wt. 330.4) is shown as such in the literature but the term NC– in their figure 1 should probably be written as H2NC– as shown in its 3D representations to avoid confusion with cyanide groups. These chemicals are perceived as super sweet due to their AH,B,X arrangement with aromatic–N group acting as the AH,B structure with a d-value of 2.71 Angstrom.

Like Smith et al., Hellekant et al. employed both multidimensional scaling and dendrographic clustering techniques to organize their data. Unlike Smith et al., they also provided a large three dimensional table describing the performance of each of neuron interrogated.

They calculated entropy values for their two sets of data using the formula proposed by Smith & Travers, 1979. This analysis will show these calculations need to be repeated using different neuron groupings.

Hellekant et al. used an artificial saliva as their solvent rather than distilled water. To the extent this saliva mimics that of their subjects, it appears to remove another variable from their experiments. However, in some cases, it may introduce the hydrated sodium ion into stimulant samples where it is normally absent.

Their data from the chorda tympani is particularly clear. It forms four distinct clusters in their figure 4 using the historical labels, N –, H–, S– & Q–. The analysis to follow appears to justify the use of the new labels proposed in this work of N –, A–, S– and P–. This is particularly true because the hydrogen ion plays virtually no role in gustation and the compound quinine hydrochloride is not a good example of the bitter tasting chemicals generally related to, and displaying the complexity of picric acid. Only five cells appear misplaced based on their analyses. • RH95D13G is described as an H-neuron in the S-cluster based on its response to a variety of artificial sweeteners and the two organic acids. All of the artificial sweeteners include at least one carboxyl gustaphore. Therefore, • their S-cluster should be subdivided into those stimulant containing only a glycophore and those containing multiple gustaphores, typically a glycophore, a acidophore and frequently a natrophore. • this neuron is primarily sensitive to the organic acids and belongs in the A-Path of this work. 240 Neurons & the Nervous System

• RH92A23B is described as an H-neuron because of its sensitivity to the carboxyl gustaphore (C–path) but is placed in the S-cluster based on its strong response to xylitol. Xylitol is a sugar alcohol typically shown in aliphatic form with three glycophores with d-values of 2.87 Angstrom.. Typically sugar alcohols form an open ring-like structure when in solution. No purity of the xylitol was provided in the paper. This neuron also showed the highest sensitivity of any to HCl. Therefore, [xxx conclusion] See Section 8.5.4.1.4. • RH92U23P, RH92U24D & RH92A23H are described as H-neurons within the N-cluster based on their response to several glutamates and citric and aspartic acids. However, all four of the chemicals they responded to contain the acidophore (the carboxylic acid group). Therefore, • Their presence in the N-cluster is only partially justified based on their actual primary sensitivity to an acidophore • No data was provided concerning the purity of the citric and aspartic acids or any effort expended to avoid their contamination by sodium salts. • Neuron susceptible to one of the gustaphores in a complex stimulant cannot be classified based on their anecdotal (or undetailed) characteristics. It is concluded that the four H-neurons listed above within the N-cluster and one H-neuron listed within the S-cluster of the dendrogram of figure 4 are in fact sensitive to the acidophore present in all of the complex stimulants described as belonging to the N-cluster. These neurons show marginal to negligible sensitivity to NaCl in figure 3. MSG & MSG + GMP, are multiple gustaphore stimulants that should not be assigned to any exclusive N-channel of gustation.

It should be noted that only two neurons in the set from the chorda tympani showed significant sensitivity to HCl.

Figure 8.5.8-2 reproduces their figure 6 showing the data from their MDS analysis of the neurons within the chorda tympani without change. However, an overlay has been added describing the folded 3D representation of the fundamental dimension of gustation. Five facts are notable.

• First, their 3D space uses a set of perspective axes rather than truly orthogonal axes. • Second, the location of the G– node is only established approximately because the distance between the disaccharide sucrose and the monosaccharide fructose do not clearly establish the correct position of the glycophore node. • Third, as shown in their figure 5, the data set contains considerable variation in their average action potential counts. This variation constitutes a fourth dimension not accounted for in their analyses, except by the high Kruskal stress index of 0.048. The variation exceeds 10:1. • Fourth, some of their average counts in figure 5 are negative, suggesting some stage 2 signal processing prior to the creation of the action potential pulses in the stage 3 chorda tympani. • Fifth, their dimension 1 is essentially parallel to the G–N axis, their dimension 2 axis nominally parallel to the N–P axis and their dimension 3 axis is nominally parallel to the G–C axis of the proposed new 3D space based on the fundamental dimension. Only citric acid, aspartic acid are grossly misrepresented with respect to the new coordinate system. The one HCl data point is extraneous in this representation. Signal Generation & Processing 8- 241

Figure 8.5.8-2 Distribution of 23 stimuli in a 3D space based on data from 47 CT fibers. Kruskal stress level: 0.048. After accounting for the perspective coordinates used by Hellekant et al., their dimensions (axes) essentially parallel the folded segments of the fundamental dimension based on the d-values of the underlying gustaphores. See text. Modified from figure 6 of Hellekant et al., 1997.

RH92U27J is classified as a Q-neuron based on very marginal performance when stimulated by QHCl and no neuron responded significantly to denatonium benzoate at the molarity of the stimulants employed to excite the neurons within the chorda tympani. Their data from the glossopharyngeal nerve can be readily interpreted based on the above discussion. Only two neurons responded marginally to denatonium benzoate. Only one of their neurons responded significantly to HCl They did not show the denatonium in their graphic and the HCl location appears anachronistic compared to the location of citric acid and aspartic acid (both organic acids). Their representation of their 3D MDS analysis used the same perspective coordinate system as their figure 6 instead of a truly orthogonal system. The overlay does show the continuity of the d-value line (A–G–N–P) as developed in Section 8.5.1. With this continuity, the dimension scales could be augmented with the addition of explicit d-value scales. Their Kruskal stress value was marginally higher at 0.067. The folded fundamental dimension of this work best fits their data if rotated by 10 degrees clockwise in the plane of the paper relative to 242 Neurons & the Nervous System the arbitrary dimensions (axes) assigned by the MDS program. The location of the C–node is not established definitively based on the data available. All of the artificial sweeteners containing an acidophore as well as a glycophore are shown appropriately along the G–C dimension. The two stimulants exhibiting only a single acidophore, citric acid and aspartic acid, are shown anachronistically among the natrophores. No data was provided concerning the purity of these two acids or any effort expended to avoid their contamination by sodium salts. When examining the response of the neurons to a stimulant perceived as bitter by humans, sucrose octa-acetate (SOA) was added to the list of stimulants. This material has little chemical resemblance to sucrose. While it maintains a heterocyclic ring containing one oxygen, in other respects it resembles picric acid in the number of oxygen atoms exposed along its periphery. It was not included in most of their analyses. Caffeine is shown in a logical, if not demonstrated, position along the P–G axis. Sodium cyclamate is shown in a position dominated by its natrophore. NaCl + amiloride is shown near the P–node along the N–P dimension much as in their figure 6. In their supporting dendrogram (fig. 9), they define an M-cluster and no H- or N-cluster. The cluster is dominated by the stimulants containing multiple gustaphores, and in particular a natrophore and an acidophore. As discussed with respect to the CT neurons, the M-cluster designation should be abandoned and the neurons reassigned based on their dominant response to an N- cluster and a C-cluster. Specifically;

RH94E21D is strongly stimulated by NaCl, monosodium glutamate and both citric and aspartic acid.

Figure 8.5.8-3 reproduces their figure 11 showing the data from their MDS analysis of the neurons within the glossopharyngeal nerve without change. However, an overlay has been added describing the folded 3D representation of the fundamental dimension of gustation. [xxx change C– to A– ] Signal Generation & Processing 8- 243

Figure 8.5.8-3 Distribution of 18 stimuli in a 3D space based on data from 33 NG fibers. Kruskal stress level: 0.067. After accounting for the perspective coordinates used by Hellekant et al., their dimensions (axes) can be overlaid by a folded form of the fundamental dimension based on the d-value parameter developed in this work. The P-path node appears poorly defined by only two data points. The best fit appears to call for the fundamental axes to be rotated 10 degrees counterclockwise in the plane of the paper. See text. Modified from figure 11 of Hellekant et al., 1997.

As in the previous figure, several facts are notable. • First, their 3D space uses a set of perspective axes rather than truly orthogonal axes. • Second, as shown in their figure 10, the data set contains considerable variation in their average action potential counts. This variation constitutes a fourth dimension not accounted for in their analyses, except by the high Kruskal stress index of 0.067. The variation exceeds 10:1 • Third, the location of the G– node is only established approximately because the distance between the disaccharide sucrose and the monosaccharide fructose do not clearly establish the correct position of the glycophore node. .• Fourth, the location of the C–node cannot be adequately defined because of the anachronistic comingling of the citric acid and aspartic acid acidophores with the natrophores, probably due to the above variation. 244 Neurons & the Nervous System

• Fifth, some of their average counts in figure 10 are negative, suggesting some stage 2 signal processing prior to the creation of the action potential pulses they measured at the NG. • Sixth, their dimension 1 axis is nominally parallel to a G–P diagonal of the proposed new 3D space based on the fundamental dimension. Their dimension 2 is nominally parallel to the G–N axis. • Seventh, citric acid and aspartic acid are grossly misrepresented with respect to the new coordinate system. The one HCl data point is extraneous in this representation. • Eighth, the P-node is not well represented in this figure because of the presence of only two P- path picrophores in the figure and the significant distortion of the figure due to the comingling of the fourth dimension amplitude information into the other values in the data set.. The third picrophore, SOA was omitted from this figure for some reason. • Ninth, the combination of sodium chloride and amiloride should be shown next to NaCl in the figure but with its amplitude contribution, in the fourth dimension, greatly reduced. All of the artificial sweeteners, except aspartame, are properly located along the C–G axis associated with the dimension 3. The position of aspartane is appropriate if the stimulant contains a natrophore as well as its recognized glycophore and acidophore. They did introduce a series of stimulants known to be perceived as sweet as other artificial sweeteners by humans with only numbers for labels. The stimulants were all quanidine derivatives incorporating a phenol ring. Although outside the scope of this section, these can be identified as glycophores exhibiting a d-value of 2.71 Angstrom but absent a conventional glycol group. They form glycophores because of the d-value between the phenol ring and the nitrogen to which it is attached. Each of the quanidines also contained an acidophore. For unexplained reasons, these guanidines were reported in some of their experiments but not others. Their molarity was 10,000:1 lower than their reference sucrose, indicating they are acting as super- sweeteners. Their SC-45647 at 0.04mM was statistically sweeter than the other two numbered stimulants at 0.03mM.

While the code number SC-45647 is respected in the literature, The code numbers beginning with NC– are not unique in the literature and represent different chemicals in different catalogs within the chemical literature.

The Hellekant et al. paper includes many more pieces of information but they must all be reinterpreted within the hypothesis of gustation developed here before reiterating any conclusions drawn from them. They also introduced the fact that they have been publishing a set of similar papers related to the chimpanzee. These also require reinterpretation based on the current very successful hypothesis.

As an example their fig. 2 shows the variation in their “bar code” representations of the action potentials for different chemicals. These variations are due to the output of the stage 1 sensory neurons exhibiting considerable variation during this interval as shown by Dailova in the next section (her figs 1, 2, 7 & 8). The variation in pulse rate indicates clearly that counting the pulses during a five second interval and computing a mean rate is at best an exploratory technique.

Interestingly, they close with the description of ethanol as an “enigmatic taste” wherein here the alcohols are recognized as tasteless with regard to gustation but possibly irritants in higher concentrations affecting the nocicepters. 8.5.8.4 The investigations of Hellekant et al. of using chimpanzees

During an extended period beginning in 1984, the Hellekant team explored the gustatory modality of the chimpanzee on the assumption that this species was the closest to humans. Some of their results have been cited under appropriate sections above. They generally addressed the effect of a set of common examples of each type of gustatory stimulant in Signal Generation & Processing 8- 245 separate papers303,304. In the 1996 paper, they compared the responses at their “proper” chorda tympani among a variety of species to stimulants frequently used in human taste perception experiments305. In a 1997 paper they addressed the subject of the efficacy of ethanol to elicit a gustatory response in various species306. They quote Hallenberg (1914), the weakest concentrations give neither a sweet, sour, salty nor bitter sensation). Nevertheless, the opinion of Wilson et al. (1973), that low concentrations taste mainly sweet, seems to be the prevailing one. The closing sentence of their opening paragraph notes, “There is, however, less agreement on the taste of higher concentrations, especially if one disregards a burning sensation as a taste quality. In their discussion, they note, “However, as will be described in a following study, ethanol is a powerful stimulus of many fibers in the lingual proper nerve. Because this nerve lacks taste fibers, its activity must give rise to a different sensation. This may underlie and cause the tactile sensation of astringency often confused with bitterness according to Amerine and Roessler.” They concluded; “On the Question if Ethanol Has a Taste. It is evident that pure ethanol has a taste to the rhesus monkey. This conclusion rests on the fact that ethanol elicited a response in taste fibers conducting taste. The question then is does ethanol taste to humans? It may seem redundant to ask this question here, but many sources suggest that ethanol itself has no taste [e.g., (26)]. Recordings from the CT of chimpanzees yielded responses to stimulation with ethanol that were similar to those described here (unpublished data). Thus, ethanol elicits a taste not only in old-world monkeys, including the rhesus monkey, but also in a species that from a human standpoint is evolutionary closer. Further, the human CT recording published by Diamant et al. (5,6) shows a taste response to ethanol. Consequently, ethanol tastes to humans. It seems that the idea that ethanol in itself has no taste to humans should be put to rest.”

Their set of experiments required multi-molar concentrations of the alcohols to elicit significant responses within the chorda tympani. They did not demonstrate a unique fiber that responded to alcohol. In their MDS analyses using a small set of stimulants that provided a Kruskal stress level below 0.003, ethanol at very high molarities was grouped with the sugars. They also performed an MDS analysis on a set of mixtures of conventional stimulants and alcohol. It was accompanied by a discussion of the role of alcohol in wines. These results suggest the alcohol may be acting as an agonist to these other stimulants instead of a gustaphore. The rest of their paper fails to provide any description of the mechanism of taste related to gustation and this work will continue to consider the aliphatic alcohols as nociceptive channels rather than the gustatory channels.

Their conclusion that alcohol is a stimulant of the gustatory modality remains questionable in the context of this work.

They continued their study of the chimpanzee in 1997 in a different journal307. That study did not address the role of alcohol. It did however present a series of MDS analyses for the less controversial stimulants as well as further discussions of the role of amiloride and the potential for

303Hellekant, G. & Ninomiya, Y. (1994) Bitter Taste in Single Chorda Tympani Taste Fibers From Chimpanzee Physiol Behav vol 56(6), pp 1185-1188

304Hellekant, G. Ninomiya, Y. Dubois, G. Danilova, V. & Roberts, T. (1996) Taste in Chimpanzee: I. The Summated Response to Sweeteners and the Effect of Gymnemic Acid Physiol Behav vol 60(2), 469-479

305Hellekant, G. & Danilova, V. (1996) Species differences toward sweeteners Food Chem vol 56(3), pp. 323-328

306Hellekant, G. Danilova, V. Roberts, T. & Ninomiya, Y. (1997) The Taste of Ethanol in a Primate Model: I. Chorda Tympani Nerve Response in Macaca mulatta. Alcohol vol 14(5), pp 473-484

307Hellekant, G. Ninomiya, Y. & Danilova, V. (1997) Taste in Chimpanzees II: Single Chorda Tympani Fibers Physiol Behav vol 61(6), pp. 829–841 246 Neurons & the Nervous System a distinct umami perception. Their study did not identify a unique umami pathway. They did identify several sub-clusters under the N-path cluster in their dendrograms. Their third paper on the chimpanzee focused on the gymnemic acids and miraculin308. The paper also provided more support for the labeled-line interpretation of the gustatory modality. They did not resolve their issue completely, “In conclusion, the result of the discussion on labeled-line or across-fiber pattern coding depends on what level of taste these two coding mechanisms are considered; if a change of activity in all taste fibers is a prerequisite for the perception of each taste quality, then the results here clearly do not favor the across-fiber pattern for coding of taste in chimpanzees; on the other hand, the labeled-line theory does not explain the ability to distinguish compounds within the same taste quality and with the same temporal pattern.”

The third paper included a broader set of conclusions attempting to summarize the full set of papers related to the chimpanzee Their conclusion was the perceptions of taste in chimpanzee were indeed very similar to that of humans. However, many of their occasional putative discussions of the theoretical operation of the gustatory modality are not supported here. In 2003, the team presented another paper constituting an overview of their work of recent years focused on the evolution of the primates309. They presented a set of cladograms and dendrograms comparing several species among the primates. The diagrams did not exhibit any umami pathway. The presentation is interesting but involves considerable speculation about the motivation behind variances (if any) in the gustatory modalities of primates. 8.5.8.5 Reinterpretation of the 2002 paper of Danilova et al.

Danilova, working with the Hellekant team has provided a data filled paper on the Marmoset, a small New World monkey. However, lacking a theoretical model, the discussion involves many assumptions that are difficult to defend.

• The transient data of figure xxx shows the difficulty of using an average number of action potential pulses within a 5 sec. interval as a criteria. The actual pulse rate varies widely over a range of at least xxx to one within this interval. The recovered analog waveforms associated with stage 1 transduction exhibit significantly different rise and fall times that are also impacted by the solubility and transport velocity within the saliva of the individual stimulants. • The data of figure xxx shows the significantly different intensity of the signals recorded at both the chorda tympani and the glossopharyngeal nerve. Such a variation is not compatible with the use of MDS techniques limited to 3 dimensions (a situation that requires the intensity of the molarities of the stimulants (actually gustaphores) to be adjusted to generate similar intensity levels within the neural system. Otherwise, a four dimensional MDS is required to account for the variables present. • While presenting a 3D MDS representation, the stimulant set used was grossly deficient in natrophores and acidophores of appropriate molarity. The result is a data set that falls primarily along the glycophore-picrophore axis with virtually no representation at the acidophore node or the natrophore node. • The resulting 1D representation is distorted by the presence of the unrecognized variable related to the intensity variation among the signaling channels. In this case, it can be argued that the second and third dimensions are more related to the intensity parameter than to the P–N dimension and the C–G dimensions of the folded fundamental dimension.

308Hellekant, G. Ninomiya, Y. & Danilova, V. (1998) Taste in Chimpanzees. III: Labeled-line Coding in Sweet Taste Physiol Behav vol 65(2), pp 191–200

309Hladik, C-M. Pasquet, P. Danilova, V. & Hellekant, G. (2003) The evolution of taste perception: psychophysics and taste nerves tell the same story in human and non-human primates C. R. Palevol vol 2, pp 281–287 Signal Generation & Processing 8- 247 The team has provided an interesting 3D representation of the perceptions of the marmoset, a new world monkey. Their 3D representation is so compressed in dimension 3 as to suggest the data set was prepared as a 2D representation and was then plotted in a 3D coordinate system. However, this could be a scaling problem since no absolute scales were introduced. The stimulant set used is nearly devoid of simple acidophores and only included one hydrated sodium ion natrophore, NaCl. The result was a nearly 2D presentation exhibiting a high Kruskal stress index of 0.14, nearly three times the average value for other Hellekant team representations. Figure 8.5.8-4 reproduces her figure with an attempt to overlay the coordinate system with the folded fundamental dimension of this work.

Figure 8.5.8-4 A presentation based on a 3D MDS analysis with a limited gustaphore set shown in perspective rather than orthogonally. The red line (dog leg) can be considered to be in the G–N–P plane of the preferred gustatory perception space. Note the nocent, HCl in the gustatory MDS space. Replacing it with a Lewis acid would identify the node of the A(cidic)-Path more accurately. It is suggested that many of the gustants were not of equal stimulant intensity and a 4-dimensional MDS analysis may give a more interpretable perception space for this stimulant set. See text. Modified from Danilova et al., 2002.

They did not provide a significant analysis of the representation of their MDS program results. Many of the artificial sweeteners involved large and very complex molecular structures that are amenable to generating multiple picrophores as well as one or more of the wanted glycophores. They noted that many of the artificial sweeteners were indeed perceived as bitter by humans. 248 Neurons & the Nervous System

The Danilova team introduced a chemical, sucrose octa-acetate(SOA) into their experiments. The name is archaic and misleading. SOA is in fact a picrophore formed by replacing all of the hydroxyl groups (6) of sucrose with acetate groups (8) in a very complex geometry. While it has been described as a sugar acetate in earlier literature, it does not exhibit any chemical structure associated with a perception of sweetness (whether a glycol group or the quanidine derivative discussed above) and it is not perceived as sweet. Its perception is that of a bitter product due to the proliferation of hydroxyl groups on its periphery exhibiting d-values on the order of 4.74 Angstrom. [xxx reword this paragraph ] Her 3D data table shows a predominance of G–path and P– path stimulants to the virtual exclusion of N–path and C–path stimulants. Thus, the data set is more realistically a one dimensional data set along the G–P dimension with the data points projected onto a diagonal surface and then rotated into the dimension 1–dimension 2 plane to the essential elimination of data in dimension 3. The team made a throw-away reference to the H-best receptors in their study. However, the response of their inorganic acids was generally below threshold and not relevant. Of the 49 chorda tympani neurons interrogated by the Danilova team, not one showed significant sensitivity to HCl. The same can be said for the 41 glossopharyngeal neurons. As a result, that team used the terminology citrus acid-best rather than H-best in their discussion and figure captions. The only gustaphore of citrus acid is of course, its carboxylic acid group that leads to the designation of the A(cidic)–path in this work. 8.5.8.6 Reinterpretation of the Giza & Scott 1991 paper

Giza & Scott have extended the multidimensional gustatory presentation to account for differences in protocol for the same materials310. Figure 8.5.8-5 shows the effect of introducing amiloride into the protocol as a pre-wash before repeating the same stimulant protocol. Amiloride is a nitrogen and amine heavy cyclic hydrocarbon containing no alkali metal311. In its pharmaceutical form, it is combined with HCl and water, C6H8ClN7O•HCl•2H2O, to form a potassium-conserving antikaliuretic-diuretic agent, Whether its activity in gustation is also distinct with respect to potassium is unclear.

This figure was not dimensioned. The “dimensions” shown have been added for discussion. They approximate those in the above figure with dimension 1 being a Quinine-sugar axis and dimension 2 a sodium-sugar axis. The third dimension has been labeled the quinine-polycose axis for discussion. “Acid,” particularly organic acid, stimuli appear under represented in the figure. It is also clear that the correct MDS analysis would have employed four dimensions in order to account for the differences in intensity of the N-Path signals caused by the difference in concentration of the four NaCl solution stimuli.

Giza & Scott did not dwell on their term polycose except to indicate it was perceived as a unique taste they described only as starchy. This term is nebulous at best. It could describe a perception attributed more to an astringent, a nocent rather than a gustant. The popular literature describes starch as a polysaccharide, a long chain of glucose molecules linked at the α(1¸4) and branched at the α(1¸6) positions, and known as amylose when helically coiled. Amylose is not truly soluble in water but forms hydrated micelles It is considered largely tasteless but known to begin to taste sweet in the mouth due to the breakdown of the amylose into glucose and maltose due to enzymatic action attacking amylase. Prior to such breakdown, it may be perceived primarily by its texture via the somatosensory modality.

310Giza, B. & Scott, T. (1991) The effect of amiloride on taste-evoked activity in the nucleus tractus solitarius of the rat Brain Res vol 550, pp 247-256

311http://www.drugbank.ca/drugs/DB00594/structure?dim=3d A Jmol representation of amiloride Signal Generation & Processing 8- 249 The change of their polycose to glucose as a function of time and amylase concentration makes their results highly dependent on the timing of their protocols. Smith & Scott have reviewed this figure in Doty (page 751). They assert, “taste offers no known analogy to the pathology of color blindness.” Actually, it appears to offer a significant analogy to conventional color vision when interpreted as in this work (See Section 8.5.2.4.2). Color vision can be explained fully using a three-dimensional color space. More approximately, color vision can be explained using an incomplete two-dimensional color space known as the Munsell Color Space. This is similar to the situation in the previous figure. One Smith & Scott comment may need modification. They note, “Blocking the NaCl response of the N-best neurons with amiloride resulted in the inability of the remaining cells to discriminate between sodium and non-sodium salts, . . .” Giza & Smith were more specific, “Responses to all 7 stimuli that contained Na+ or Li+ were suppressed by amiloride.” Even this statement in their summary appears subject to modification as monosodium glutamate was unaffected by the amiloride treatment. In the original paper, Giza & Scott noted, “The most striking findings of this study of NTS taste cells are (1) the effects of amiloride on responsiveness to NaCl and LIC1 are either profound or minimal, depending on the cell, . . . .” A larger statistical sample is needed to resolve the effect on the potassium (another alkali metal) salt data point.

It is important to note the work of Ossebaard & Smith regarding amiloride in humans312. They suggest several subtleties indicating a lower effectiveness in humans than in some other species.

312Ossebaard, C. & Smith, D. (1995) Effect of amiloride on the taste of NaCl, Na-gluconate and KCl in humans: implications for Na+ receptor mechanisms Chem Senses vol 20, pp 37-46 250 Neurons & the Nervous System

Figure 8.5.8-5 Changes in three-dimensional taste spaces of one rat caused by amiloride as recorded at the NST. 15 stimuli were used in a before (A) and after (B) experiment where the tongue and palate were washed with amiloride. After amiloride application, all the Na+-bearing stimuli, except MSG, and the lithium salt migrate to the positions of the one organic acid, the non- sodium salts, and quinine. N1-N4, 0.01-0.3 M NaCl; Li, LiCl;.S; sucrose; Sa, Sodium saccharin; G, glucose; H, HCl; Ci, citric acid; Q, quinine hydrocholoride; K, KCl; Ca, CaCl2; M, MSG; P, polycose. From Giza & Scott, 1991.

- - - - Signal Generation & Processing 8- 251 Giza & Scott presented a 3D representation of the effect of amiloride on a variety of stimulants313. The before and after 3D plots did not label their axes, nor did they discuss the differences between the scales supporting the axes. Their block diagram of the gustatory modality generally follows an expanded version proposed in this work. Their discussion of entropy as applied to gustation deserves additional analysis. To be useful, the definition of entropy should be explored as it is found at the output of the saliency map for gustation. Dimensions 1, 2 & 3 were labeled arbitrarily in this work to aid discussion. Based on the hypothesis of this work, the stimulant set included a variety of non gustatory (primarily alkaline earth salts). Because of the limited range of the data set (no acidophores and only one picrophore) it is difficult to determine the character of the axes. It is reasonable to conclude the automatically attached scales of the before and after views are different due to the movement of the points within the data set that the MDS program manipulated. It can also be deduced; • G remained unchanged between the two views and S and SA only moved marginally. • Q moved significantly, suggesting the large movement of the N’s caused a re-scaling in dimension 1. • The large movement of all of the N’s in dimension 1 suggests dimension 1 is more concerned with the intensity dimension than the spatial dimensions associated with the d-value dimensions. • The components of each stimulant in dimension 3 remained largely unchanged for all of the stimulants suggesting dimension 3 was not associated with the perceived intensity of the stimulants. • The components of each stimulant in dimension 2 remained largely at the same locations except for the N’s.

It can be concluded that;

• At least the dimension 1 scales are different between the before and after plots • The few gustaphores employed do not allow for the delineation of the the complete perceived gustatory space • The consistency of the positions of the sugars and quinine suggest dimensions 2 and 3 are related to the G–P axis of gustation. • The large component changes associated with dimension 1 suggest it represents or is highly influenced by the intensity dimension of perception. This finding suggests amiloride is blocking the selective DACB of the N–stimulants with the N-path GR’s. • Amiloride had negligible impact on the other stimulants of the gustatory and nociceptive modalities of the neural system. 8.5.8.7 Reinterpretation of the review by Spector & Travers of 2005

Spector & Travers lamented the lack of a fundamental dimension in gustation in 2005. As a result of this work, their review can be considered largely archaic. Their observation that many experimental protocols lacked sufficient samples to be statistically relevant is confirmed by the variability observed in most of the reports of laboratory investigations reviewed above. Without providing error bars to quantify the utility of their data, the conclusions drawn by most investigators must be considered only exploratory in character. They do note their entropy varied substantially based on the stimulants employed in their experiments.

8.5.8.8 Reinterpretation of the nociceptor data from Kashiwagura et al. 1980

Placed in the context of this work, the investigations of Kamo et al. addressed in Section 8.5.1 applied to the chemoreception system as stated, while the investigations of Kashiwagura can be clearly related to the nociceptor modality of the neural system. Kashiwagura et al314. show the

313Giza, B. & Scott, T. (1991) The effect of amiloride on taste-evoked activity in the nucleus tractus solitarius of the rat Brain Res vol 550, pp 247-256

314Kashiwagura, T. Kamo, N. Kurihara, K. & Kobatake, Y. (1980) Interpretation by theoretical model of dynamic and steady components in frog gustatory response Am J Physiol (gastrointest.) Vol. 238, pp G445 252 Neurons & the Nervous System measured responses at the glossopharyngeal nerve of the bullfrog, Rana catesbeiana, for comparison with the mathematical analysis of Kamo et al. They varied both the flow rates of the stimulants and the temperature of the experiment. The experiments were carried out in-vivo at 18°C except when exploring the affect of temperature between 2.5°C and 20 oC, They noted the dynamic portions of their waveforms were sensitive to flow rate but the steady state values generally were not. [xxx condense next two paragraphs ] Kashiwagura et al. arbitrarily separated the pedestal (steady state) portion of their responses from the transient (or dynamic) portion without providing any physiological mechanism justifying such treatment. In fact, the pedestal portion represented the response of their test configuration under steady state conditions (after all transients had died away). Generally, their stimulus was removed far before this condition was realized. As a result, their pedestal height was generally taken as equal to the height of the response at termination of stimulation. Their page G447 states, “The theoretical definition of the steady state response is the response at t = 4, but it is not available experimentally. As an experimental approximation, the magnitude of the response at 4.5 min after the onset of stimulation is taken as the magnitude of the steady component.” This assumption assumes the stimulation is a step instead of a finite duration pulse as shown for all of their waveforms. In many cases their responses are not nearly horizontal after 4.5 min. While not clear, Kashiwagura et al. appear to use the expression dynamic component to refer to the overall waveform less a constant value present throughout the duration of the stimulus. They associate the fast component and the attack time constant of their discussion. The slow component is associated with a decay time constant. However, a more meaningful decay time constant for their waveforms is obtained by evaluating the decay transient after cessation of stimulation. This decay characteristic repeatedly suggests a true decay time constant of less than one second. If correct, the majority of their slow component is due to adaptation within the first amplifier of the sensory neuron.

Their focus on the application of alkali earth salts, acting generally as astringents, is a clear indication that their stage 3 signals from the glossopharyngeal nerve originated with the nociceptor neurons and not from the gustatory sensory neurons that are insensitive to the alkali earth salts.

- - - -

Gustatory experiments are traditionally carried out under conditions convenient to the investigator. Stimuli are applied by pouring or brushing the material onto areas of the tongue that are only coarsely delineated. Typical time intervals are measured in tens of seconds with initial application frequently reported using voice records from the investigator. The records are generally action potential streams taken at the stage 3 neurons orthodromic to the stage 1 sensory neurons, amplified and integrated. The integrated records faithfully represent the original analog waveforms of the sensory neurons prior to encoding by the stage 3 neurons. However, the rapidly rising attack portion of the waveform is frequently poorly reproduced. This makes determination of the precise shape of the leading edge of the responses difficult to determine.

- - - -

Figure 8.5.8-6(top) provides a block diagram supportive of the exploratory investigations of Kashiwagura et al. In the context of the Electrolytic Theory of the Neuron, Kashiwagura et al. explored the operation of the non-neural stage 0 of the nociceptor modality and both stage 1A, transduction and stage 1B, adaptation of stage 1, sensory signal development. They did not address any potential signal processing within stage 2. They did record their signals via a probing of the GP nerve at the point indicated following pulse signal encoding (action potential generation) in stage 3A. Unfortunately, the data is highly conflated. However, using the model of this work, the material can be teased apart with some success. The following discussion will examine portions of such an operation. Signal Generation & Processing 8- 253

Figure 8.5.8-6 Integral of stage 3 action potential recovered from the glossopharyngeal nerve of the bullfrog, Rana catesbeiana. Top; block diagram supporting the Kashiwagura et al. experiments. Bottom; waveforms from their paper with additional annotation. 1–2; pure delay interval. 2–3; attack interval. 3–4; adaptation interval. 4–5; actual decay interval after cessation of stimulation. See text. Lower frame modified from Kashiwagura, 1980.

Kashiwagura et al. introduced several equations based on their interpretation of the above block diagram as consisting of a single black box. This led to some confusion in their discussion. Omitting the details here, they observed on page G451, “As the flow rate becomes larger, the value of [their time constant becomes smaller]. An analytic solution of [the overall process from stage 0 to stage 3A], P, however, is not obtainable when the above equation is introduced into the kinetic equations set up for Equation 1. Use of an analog computer enables simulation for the time course of the response as shown below.” In fact, a suitable overall (analytical) equation is easily arrived at based on the block diagram proposed here.

They noted the character of their stimulation by defining the stimulation function of stage 0 as a square pulse with an exponential leading edge where the time constant of the exponential could be controlled. They then described a kinetic equation (typical of common chemistry) that did not consider the trailing portion of their waveforms following cessation of stimulation. They ascribed the leading edge of their waveform to one portion of the kinetic equation and the noisy decay portion of their waveforms to a second. The true decay characteristic of their kinetic equation is actually the decay characteristic after stimulation withdrawal. They provided a set of recorded waveforms illustrating a wide range of operating conditions (without vertical scale calibrations). Their figures have appeared in textbooks in several variants but unfortunately the captions are usually misleading as to the origin of the major decay in signal amplitude with time. This major decay is due to adaptation within the sensory neuron and not to the fundamental excitation/de-excitation mechanism of transduction. 254 Neurons & the Nervous System

Figure 8.5.8-7(bottom) provides a set of temperature related waveforms with waveform D further annotated. Based on the Electrolytic Theory of the Neuron presented here, these waveforms can be interpreted in considerably greater detail. They provided no vertical scale in this figure and it is quite likely that the individual waveforms exhibited considerably different amplitudes. Note the frames A & B were acquired at the same temperature. As noted by Kashiwagura et al., frame A shows the time constant related to the flow rate of the stimulant and not the leading edge of the excitation/de-excitation (E/D) mechanism of transduction. The performance in frame A is dominated by the time constant of stage 0. It is proposed here without further justification that the nociceptor transduction process utilizes the same excitation/de-excitation mechanism as all other sensory modalities of the sensory neurons (Section 8.5.6 xxx). This mechanism, and the associated adaptation mechanism employs a temperature parameter that is biologically limited and does not employ absolute zero Kelvin as its lower limit. The relationship with regard to temperature is quite complex but analytical. Their solution, P, for the overall response of the frog to stimulation did not include any temperature term and cannot be used to describe their measured waveforms where temperature is a variable.

Frames B through E show that the leading edge of the E/D mechanism is quite sharp, with total rise times (interval 2–3) of less than a few seconds following an absolute delay (interval 1–2) of a few seconds or less depending on temperature. The decay time constant of the E/D mechanism can be determined from the interval 4–5. However the scale is marginal for reading values off of this figure. The interval is about 12 seconds, suggesting a decay time constant of about 3 seconds or less at 12o C. The adaptation mechanism associated with the first amplifier (Activa) of the transduction process is quite prominent in interval 3–5. This parameter changes significantly with temperature and intensity of the stimulus. Since no vertical scale was provided with these waveforms, the effect of temperature and stimulus intensity cannot be de-convolved in these images. Based on frames D and E, it appears the adaptation time constant is about 100 seconds at 12–18o C. It may be considerably longer in frames B and C, as expected for an exothermic vertebrate.

If the waveforms of B through E are in-fact plotted to the same vertical scale, the change in the overall response with temperature is significant, although little if any change in the rate of rise and the decay characteristic after stimulation ends are apparent. This situation would suggest a significant change in the gain of the 1st amplifier and its adaptation time constant (both related to its chemical source of power) with temperature. 8.5.8.9 The conditioned taste aversion MDS data of Chang & Scott–1984

Chang & Scott have highlighted the effect on MDS results of varying the physiological state of rats in the collection of data (Section 8.5.8.9). In summary, based on the hypothesis of this work, their results are very difficult to interpret. First, their conditioning stimulus was not a SGG; it was a MGG containing at least two gustaphores closely related to many of their other stimulants. Second, they included the nocent, HCl, in their gustatory stimulant set. Third, they did not attempt to control the concentration of their gustants to achieve a near constant perceived sensation level. This mixing of stimulants from different modalities along with a mix of concentrations and the use of MGGs makes deconflating their data (specifically rotating the axes of their work into a standardized 3D space) very awkward. They did not describe the three dimensions of their data space. The data set they attempted to display in a 3D space requires at least 5 dimensions to properly analyze. Chang & Scott do not appear to have appreciated the state of the art of MDS analysis at the time of their work. They attempted to use a technique designed to address qualitative differences and include quantitative differences in their stimulant set.

- - - - Signal Generation & Processing 8- 255 Chang & Scott have highlighted the effect on MDS results of varying the physiological state of rats in the collection of data. Unfortunately, they did not employ a model of the gustatory modality before developing their protocol. They employed single neuron recording from a location within the Nucleus Tractus Solitarius. The selected neurons had to be sensitive to stimuli representing the four signaling channels the investigators were accustomed to, salty, inorganic acid, sweet and bitter. No organic acids were included in their basic stimulus set, but citric acid was added in a followup experiment. No other inorganic acids were included except HCl. They did include two different concentrations of sodium saccharin (0.25 and 0.0025 M), suggesting an additional dimension related to perceived intensity was present in their database. In summary, based on the hypothesis of this work, their results are very difficult to interpret. First, their conditioning stimulus was not a SGG; it was a MGG containing at least two gustaphores closely related to many of their other stimulants. Second, they included the nocent, HCl, in their stimulant set. Third, they did not attempt to control the concentration of their gustants to achieve a near constant perceived sensation level. This mixing of stimulants from different modalities along with a mix of concentrations and the use of MGGs makes deconflating their data (specifically rotating the axes of their work into a standardized 3D space) very awkward. They did not describe the three dimensions of their data space. The data set they attempted to display in a 3D sapce requires at least 5 dimensions to properly analyze. Figure 8.5.8-7 shows their results for a group of rats that had been conditioned to be wary of NaSaccharin. The paper is extensive and data-filled. They note,

(Previous) “Explorations of the neural substrates of conditioned taste aversions (CTAs) have focused principally on diencephalic and telencephalic structures. The nucleus tractus solitarius (NTS) is the initial gustatory relay in the rat’s hindbrain. It is worthy of investigation for its part in mediating CTAs in that it is sensitive to several physiological conditions which affect feeding while also being a site of anatomical convergence for vagal afferents from the viscera and centrifugal projections from areas (hypothalamus, amygdala) implicated in emotions and hedonics.”

“We compared single neuron responses from NTS to several taste stimuli in three groups of rats: (1) those receiving exposure to 0.0025 M sodium saccharin without physiological consequences; (2) those made ill through intraperitoneal injections of LiCl but having no obvious gustatory referent for their malaise (sensitization-pseudoconditioning controls); (3) those in which exposure to 0.0025 M sodium saccharin (the conditioned stimulus, CS) was paired with LiCl-induced poisoning (the unconditioned stimulus, US), creating a pronounced aversion to the saccharin.”

According to response profiles, NTS neurons in all three groups could be divided into subsets of about 30%, which showed a sweet-sensitive profile, and 70%, which were primarily sensitive to nonsweet qualities. The major effect of the conditioning procedure was to increase responsiveness to the saccharin CS only among the sweet-sensitive subset.”

“The axes of the spaces are unlabeled, for the attributes they represent-which should underlie taste just as frequency underlies pitch and wavelength, color-are not yet determined.” “The significance of the results is that: (1) CTAs affect sensory activity at a lower order level than had heretofore been demonstrated; (2) NTS shows sensitivity to yet another physiological condition, reinforcing the involvement of the hindbrain in hedonics and sophisticated taste-related processes; (3) there is a subset of taste neurons, rather distinct according to its sensitivity profile, which is also functionally unique in its response to conditioning by a sweet CS.” 256 Neurons & the Nervous System

Figure 8.5.8-7 Three dimensional MDS spaces representing a CTA group and a control group. NaSaccharine, CS, is the same coordinates in each space. See text. From Chang & Scott, 1984.

Note the presence of HCl and citric acid in their control space. The major problem with the paper and the above graphic is that they did not rely upon any theory of gustation and they subsequently did not rotate and rectify their MDS data to represent a three-dimensional perceived gustatory space. They did adjust their dimensionless representations so that NaSaccharie occurred at the same coordinates in each frame (2.5 units to the right and 2.0 units up in the checkerboard of both frames. However, this forced retention of one molecular location was not justified by any axiom of MDS analysis. The third dimension appears to be about 2,5 units above the plane of the checkerboard. However, it is not clear this is an important feature of their analysis. A more important shortcoming is the lack of any definition of the dimensions of each axis and any calibration of these axes. They described sodium saccharin as “a novel taste stimulus that served as the conditioned stimulus (CS).” The use of sodium saccharin (NaSac) as a reference is unfortunate since it is obviously a gustant containing at least two gustaphores.

8.5.9 Extending the hypothesis to include the sweet proteins

Only six proteins are known that are perceived as sweet by humans, and apparently old world monkeys. The success of the hypothesis presented here suggests it can probably be extended to account for this situation. The following paragraphs address this extension. While less well demonstrated by empirical investigations than the basic hypothesis presented above, there is adequate data to form such an extension. The extended hypothesis will include the following corrolary; Signal Generation & Processing 8- 257 • The six known sweet proteins (many of their derivatives and potentially other proteins) can form DACB’s with the G-path GR’s of this work by modifying specific amino acid residues located along gradual turns at the periphery of the space occupied by the proteins. The restriction to the gradual turns, such as the β-turn, eliminates those amino acids found within τηε α−helix and sheet portions of the protein. The requirement for a modified amino acid is consistent with our knowledge of a long list of rare amino acids as well as the recognition and good documentation of one such modified amino acid in brazzein. Brazzein is the smallest known (6.5 kDa) and one of the most well studied sweet proteins. The following discussion will limit the modifications to specific amino acids of the sweet proteins to the first order case; i. e., the introduction of an orbital (either oxygen or nitrogen associated with one or more hydrogens to support structural rules and/or support the requisite hydrogen bonding. The potential introduction of more uncommon orbitals, including the reliance on a phenol ring structure (such as in a modified tyrosine or phenylalanine) will be left to future study. 8.5.9.1 The study of protein mutations in human & primate gustation

[xxx edit text above and below the - - - marks into one text. Currently have two 8.5.9.1s ] This paragraph will provide a more specific framework leading to clarification of the role of the proteins as gustatory stimulants versus true gustaphores and the various situations in which proteins might exhibit a distinct taste.

Currently, proteins exist in three distinct structural forms; • native or those with a highly distinct (and generally unique) entwined ligand form, • man-made and either reconstructed or fabricated structures assumed to contain ligands in the same concatenation and be entwined (folded) identically to the native form and • denatured or proteins with an alternate (frequently simpler and unfolded) structural arrangement as a result of heating or specific chemical treatment.

Based on the hypothesis of this work, a protein can be considered a stimulant containing one or more individual gustaphores or be itself a gustaphore. A criteria for any gustaphore is that it be able to form a DACB with a GR. In the case of a typically structurally complex protein, it is necessary that any gustaphore be accessible at the proteins surface in order to form hydrogen bonds with a GR that is no more than 2.7 Angstrom distant. Thus, the configuration of the protein is critically important when exploring its gustatory properties.

While textbooks have frequently indicated that the individual amino acids forming a protein have no intrinsic biological or toxic effects, this statement needs qualification to separate gustation from other biological processes.

Jin and others have investigated the structure of proteins in the context of their gustatory properties. They note the relevance of the N – and C–terminals of the protein as well as a variety of surface ligands that could relate to individual gustaphores. Along the backbone of the protein, the side-chains, known as R-groups are particularly relevant. The C–terminal of a protein consists of a carboxylic acid group and is by definition a carboxylic acid gustaphore or acidophore under the hypothesis of this work if it satisfies the 2.7 Angstrom stereo-chemical criteria defined above. Frequently, this terminal is not readily available as an acidophore. The N –terminal of a protein consists of only an NH2 group and does not constitute a gustaphore according to this hypothesis. Many of the amino acids forming a protein exhibit side chains when linked known as R-groups. Glutamic acid and aspartic acid both contain carboxylic acid in their R-group that can potentially represent an acidophore along the backbone of the protein. They both exhibit ideal d-values of 2.7 Angstrom. Glutamine and asparagine have R-groups containing an O=C-NH2 structure that could act as an acidophore. They both exhibit a d-value of 2.22 Angstrom (18.5% low compared to the ideal 2.7 Angstrom for an acidophore). [xxx edit above values ] 258 Neurons & the Nervous System

Jin et al. have studied 25 derivatives (they describe them as mutations) of a simple wild protein, brazzein (6.5 kDa)315. They showed this protein was tasteless in its wild form but was perceived as sweet when an amino acid without an acidophore R-group was replaced by glutamic acid with its acidophore R-group (Section 8.5.9.1) The Jin et al. study did much to correlate the human perception of taste with the recordings from G-path neurons in the proper CT neurons (where care was taken to interrogate the chorda tympani before it merged with the lingual nerve. xxx316, writing in Cagan [xxx page 91], discusses the artificial sweeteners monellin and thaumatin (E957 in the sweetener index). He notes thaumatin includes 11 lysine residues. He also notes the stepwise acetylization of the lysine residues leads to a progressive loss in the sweeteners capability. Lysine exhibits a d-value of 2.605 Angstrom when incorporated on the outer surface of a protein and thus, each lysine residue is capable of forming a DACB with the GR 2 sensory receptor of gustation. See also Section 8.5.9.1. “Acetylization of more than four lysines residues renders it tasteless.” This assertion suggests the other lysine residues may be located internally in thaumatin. xxx notes, “The effect of thaumatin stimulation of the tongue has been recorded in several mammals.” They also assert it only occurs in catarrhine primates. “No neural response has been obtained in all other prosimian and platyrrhine primates tested.” This assertion follows the Thaumatin column in Table 1 of Hellekant et al317. in 1991. Monellin exhibits similar properties to thaumatin. With a few exceptions, its stimulation is also limited to the cattrhine animals. The exceptons suggest further investigation is needed with respect to both thaumatin and monellin.

It is generally recognized that solvent molecules do not penetrate the interior of the volume occupied by the proteins. For purposes of DACB bonding, the proposed gustaphores need to be located near the exterior surface of the protein volume. [xxx edit? ]

Figure 8.5.9-1 summarizes the characteristics of the known proteins perceived as sweet.

315Jin, Z. Danilova, V. Assadi-Porter, F. Markley, J. & Hellekant, G. (2003) Monkey Electrophysiological and Human Psychophysical Responses to Mutants of the Sweet Protein Brazzein: Delineating Brazzein Sweetness Chem Senses vol 28, pp 491–498

316Scott, T. & Yaxley, S. ( 1989) Interaction of taste and ingestion In Cagan, R. ed. Neural Mechanisms in Taste Boca Raton, FL: CRC Press Chap. 7

317Hellekant, G. Walters, D. Culberson, J. et al. (1991) Electrophysiological evaluation of sweeteners In Walters, D. Orthoefer, F. & DuBois , G. eds. Sweeteners. Wash. DC: American Chemical Society Chap. 22 Signal Generation & Processing 8- 259

Figure 8.5.9-1 Characteristics of proteins perceived as sweet by humans ADD Thautadin is more recently known as thaumatin (E957). See text.

- - - - -

There are currently six proteins known to be perceived as sweet by humans. There are indications this perception may be limited to humans, apes and old world monkeys318. These proteins generally satisfy the definition of a super sweet stimulant because of their high degree of sweetness compares to sucrose, typically in the 500:1 to 100,000:1 range on a molarity basis. The ratios are typically much lower on a weight/weight basis. These ratios and how they degrade on molecular substitution suggest the AH,B,X model of the Shellenberger team can be used to explain the gustatory operation of these proteins. The protein brazzein has been a popular subject because of its high temperature of denaturing, ~85 C. Quoting Jin et al, who cite Caldwell et al., “Fruit brazzein consist of two forms, one with pGlu at its N-terminus and one without (des pGlu) which is twice as sweet as the first form. The latter form is nominally 500 times sweeter than a 10% sucrose solution and up to 2000 times sweeter that a 2% sucrose solution on a weight basis.” Assadi-Porter et al give a different set of numbers without attribution, “Brazzein, as isolated from the fruit, is 500 times sweeter than sucrose on a weight basis (9500 times sweeter on a per molecule basis). The search for the mechanism by which “sweet proteins” interact with the sweet GR is a very active area. Most of the work remains conceptual and involving the structural docking of these proteins with a very complex putative T1R1 receptor derived from the genetic code. No mechanism of interaction between these elements has been offered.

318Assadi-Porter, F. Aceti, D. Cheng,H. & Markley, J. (2000a) Efficient Production of Recombinant Brazzein, a Small, Heat-Stable, Sweet-Tasting Protein of Plant Origin Arch Biochem Biophys vol. 376(2), pp 252-258 260 Neurons & the Nervous System

Some of the concepts used on searching for the fundamental gustatory mechanism involving the sweet proteins are quite colorful, such as involving a “venus fly trap model” (VFTM). Others have been seeking to understand the relationship between the sweet proteins and the GR using amino acid sequencing and structural analysis of the complete protein. Searching the amino acid sequence of these small proteins, 54 to 202 amino acid residues, appears to have been most successful. However, this path has yet to uncover the mechanism involved.

8.5.9.1.1 The folding of proteins

It has long been known that protein folding was dominated by hydrogen bonding between various amino acid residues frequently separated by multiple intermediate residues. Lehninger (1972, pg 112) provided an excellent illustration of this condition. The ultimate folding of a given protein is a direct result of these hydrogen bonds and an indirect result of the genetic code itself for that protein, since the code defines the sequence of the individual residues and thereby their initial spatial relationships.

Lattman & Rose made some important observations about protein folding under a provocative title in a significant early paper319; “The folding reactions of many small, globular proteins exhibit two-state kinetics, in which the folded and unfolded states interconvert readily without observable intermediates. Typically, the free energy difference, AG, between the native and denatured states of such a protein is quite small, lying in the range of approximately -5 to -15 kcal/mol. We point out that, under these circumstances, a population of native-like molecules will persist, even in the presence of mutations sufficiently destabilizing to change the sign of AG. Therefore, it is not energy per se that determines conformation. A corollary to this argument is that specificity-not stability-would be the more informative focus in future folding studies.”

They went on; “A protein molecule adopts its unique, three-dimensional equilibrium structure spontaneously under physiological conditions in many, if not all, cases. This native structure can be denatured readily by elevated temperatures or chemical perturbations, both of which induce chain disorder but leave covalent bonds intact. . . It is well established that the information necessary to drive this reversible disorder + order transition is encrypted solely within the linear amino acid sequence; hence, the structure of a protein is implicit in the gene that encodes it.”

Later they note, “There is a profound organizing aspect of two-state behavior that is often overlooked: proteins are either folded in a native-like manner or not folded at all.”

“What is the origin of conformational specificity? One attractive candidate has been internal packing. Globular proteins are known to have mean packing densities reminiscent of solids, a consequence of the exquisite complementarity between interior side chains, which fit together like pieces of a three-dimensional jigsaw puzzle (27). This experimental fact can be interpreted to mean that protein conformation is linked tightly to internal packing.” Finally, “In sum, we propose that the derivation of a reliable strategy to predict structure from sequence will depend critically upon elucidation of the stereochemical code that underlies conformational specificity.”

319Lattman, E. and Rose, G. (1993) Protein folding-what's the question? Proc Nati Acad Sci USA vol 90, pp 439-441 Signal Generation & Processing 8- 261 Thirteen years later, Rose et al. provided a major article on folding320. They provided a Part 5 as follows; Part 5. A Backbone-Based Theory of Folding This perspective has described 10 seemingly disparate aspects of protein folding. In particular: 1. The native fold is unique. The folding reaction is U º N, not U º N1 N2 . . . Ni. 2. Folding is reversible. 3. No covalent bonds are made or broken in the folding reaction, U º N. Only weak bonds are involved. 4. Folding conditions and unfolding conditions are similar, respectively, for most mesophilic proteins, regardless of sequence. 5. The U º N reaction is highly cooperative. Most single-domain proteins fold in an all-or-none manner. 6. The fold is built on a scaffold of hydrogen-bonded -helices and -strands. 7. The number of stable domains is limited to a few thousand. 8. Proteins typically avoid metastable kinetic traps under native folding conditions. 9. Protecting denaturing osmolytes fold unfold proteins by operating predominantly on the backbone in the unfolded state, dialing folding up/down but leaving the fold itself unaltered. 10. Stability and conformation are not synonymous. The native conformation can still be attained under grossly destabilizing conditions. Such conditions shift the U º N equilibrium toward either N or U, but not toward N* (i.e., an alternative folded state)..

In 2012, Porter & Rose provided additional perspective on folding with extremely colorful and complex figures321. Their approach was based on thermodynamic assumptions. Their principle contribution was a “Domain Identification Algorithm. Domains were identified in solved protein structures by using our structure-energy equivalence of domains (SEED) algorithm. The minimum size of a domain was fixed at 25 residues, approximating the size of a unit of supersecondary structure and the minimum chain length needed to attain a protein-like surface/volume ratio. No fixed limit was imposed on the maximum size. The algorithm identifies an optimal set of non overlapping units that maximizes both collective QRs and chain coverage.” They did not address sweet proteins that were near the shorter sequence end of their algorithm space. Their focus on long peptide chains is beyond the scope of that needed here. 8.5.9.1 Searching glycophore location based on complex protein theory de Vos et al. have provided the schematic of amino acid folding for thaumatin 1 at 3.1 Angstrom resolution322. Thaumatin consists of a single polypeptide chain of 207 amino acids, has a molecular weight of 22.2 kDa (Iyengar et al., 1979). The amino acid sequence (Iyengar et al., 1979) and 3-dimensional structure (De Vos et al., 1985) are known. The taste of thaumatin and that of another sweet protein, xxx Speaking of both thaumatin and monellin they noted, “The sweet taste of the proteins can be registered at a very low concentration (10' M), comparable to those in hormone-hormone receptor interaction. These proteins are about 100,000 times sweeter than sucrose on a molar basis and several thousand times sweeter on a weight basis. In fact, these two proteins are the two sweetest compounds known to man.” They also note, “five tripeptides in monellin have their homologous counterparts in thaumatin at residues 94-96, 100-102, 101-103, 118-120, and 128-130. All five regions are well exposed. It is likely that some of these regions are responsible for the immunological cross-reactivity between monellin and thaumatin, and their conformation may be important for sweet receptor binding.” They conclude without any discussion related to the theory of gustation, “Recently the gene for thaumatin II has been cloned and expressed (16). Thaumatin II is as intensely sweet as thaumatin I but is different from it at five

320Rose, G. Fleming, P. Banavar, J. & Maritan, A. (2006) A backbone-based theory of protein folding PNAS vol 103(45), pp 16623–16633

321Porter, L. & Rose, G. (2012) A thermodynamic definition of protein domains PNAS vol 109(24), pp 9420–9425

322De Vos, A. Hatadat, M. Van Der Wel, H. et al. (1985) Three-dimensional structure of thaumatin I, an intensely sweet protein Natl. Acad. Sci. USA vol 82, pp 1406-1409 262 Neurons & the Nervous System positions [residues 46, 63, 67, 76, and 113 (4)]. It is almost certain that thaumatin II has the same three-dimensional structure as thaumatin I described here.” Tancredi et al. provided a comparative study of three of the sweet proteins in 2004323. Schematics of each are shown in their figure 1. Their lack of commonality in structure, except for the various loops, is immediately obvious. As they note in the opening to their abstract, “The mechanism of interaction of sweet proteins with the [putative] T1R2-T1R3 sweet taste receptor has not yet been elucidated.” They synthesized a group of peptide sequences representing the “hairpin structures” thought to be representative of these structures. “However, none of the peptides has a sweet taste.” They noted in 2004, “In spite of several attempts, the glucophores of sweet proteins have not yet been identified with certainty, even for brazzein, monellin and thaumatin, the only sweet proteins of known structure.” They provide considerable data on their peptides. Their introduction ends with, “Here we present an extensive search for ‘sweet fingers’ on the surface of monellin, thaumatin and brazzein, the three sweet proteins best characterized from a structural point of view. As a result of the search three potential ‘sweet finger’ peptides were designed, synthesized and studied from a conformational point of view.”

Their conclusions end with, “In summary, we can conclude that the data presented in this paper do not support the idea of a localized ‘sweet finger’ site that could explain the properties of sweet proteins according to the rules established for small sweet molecules. Alternative explanations that could better explain the experimental observations, such as the ‘wedge model’ recently proposed by us [40], should therefore be explored.” In essence, they did not find any trace of the actual glycophore associated with these three sweet proteins. Only detailed examination of the 3D representations of their “sequences” could determine if they contained a glycophore (using the requirements based on the hypothesis presented here).

Temussi discussed his wedge model in 2002 under the heading, hypothesis324. After a discussion, he suggests a future plan of investigation.

Temussi provided an overview of the search for the gustatory sites of the sweet proteins in 2006, based primarily on the field of genetics and genetic code expression325. While citing the work of Shallenberger et al and Kier, he dismissed their work in terms of its applicability to the sweet proteins. He developed his wedge model introduced above by Tancredi et al.

8.5.9.2 Amino acid sequencing of the sweet proteins

In a 2003 study, the team of Jin - - - Hellekant examined the role of more than 25 derivatives of the wild protein brazzein (which they described as mutants) in human gustation326. Their conclusions do not appear surprising within the context of the hypothesis presented here,

“The effects of charge and side chain size were examined at two locations, namely positions 29 and 36. The findings indicate that charge is important for eliciting sweetness,

323Tancredi, T. Pastore, A. Salvadori, S. Esposito, V. & Temussi, P. (2004) Interaction of sweet proteins with their receptor A conformational study of peptides corresponding to loops of brazzein, monellin and thaumatin Eur. J. Biochem. v 271, pp 2231–2240

324Temussi, P.A. (2002) Why are sweet proteins sweet? Interaction of brazzein, monellin and thaumatin with the T1R2–T1R3 receptor. FEBSLett vol 526, pp 1–3

325Temussi, xxx (2006) xxx

326Jin, Z. Danilova, V. Assadi-Porter, F. Markley, J. & Hellekant, G. (2003) Monkey Electrophysiological and Human Psychophysical Responses to Mutants of the Sweet Protein Brazzein: Delineating Brazzein Sweetness Chem Senses vol 28, pp 491–498 Signal Generation & Processing 8- 263 whereas the length of the side-chain plays a lesser role. We also found that the N- and C-termini are important for the sweetness of brazzein. The close correlation (r = 0.78) between the results of the above two methods corroborates our hypothesis that S fibers convey sweet taste in primates.” They also draw other conclusions, “Fruit brazzein consists of two forms, one with pGlu at its N terminus and one without (des-pGlu) which is twice as sweet as the first form. The latter form is 500 times sweeter than a 10% sucrose solution and up to 2000 times sweeter than a 2% sucrose solution on a weight basis. It has the smallest molecular size among sweet proteins (6.5 kDa) and high solubility in aqueous solution.” des-brazzein has a weight of 6.353 kDa. This work would suggest that charge and side chain length play different roles. Side chain length is critical when attempting to meet the d-value criteria for a glycophore within a stimulant, whereas, the charge is largely related to the perceived intensity of a given stimulant. Jin et al. also provided a set of “bar code” recordings from the chorda tympani suggestive of the rate of response of the taste system following initial application of their stimulants. Caldwell et al. have indicated there are a total of six known simple proteins (as opposed to glycoproteins) perceived as sweet. Their figure 3 shows the ball-and-stick versions of brazzein, thaumatin and monellin obtained using MolScript. The general view is they exhibit no common structural features other than individual curved strands forming the external envelope of each molecule. Brazzein is much the simpler of the three.

Brazzein and the other five proteins qualify as a super-sweetener and an effort should be made to determine how it satisfies the proposed AH,B,X criteria for such sweeteners.

Figure 8.5.9-2 reproduces their protein and indicates locations of principle attention. 264 Neurons & the Nervous System

Figure 8.5.9-2 Diagram showing the position of mutations on the brazzein molecule and the corresponding changes in taste (red, increased sweetness, black, the same, light blue, decreased sweetness in comaprison with WT brazzein, dark blue, scored as water). Intramolecular disulfide bonds are shown as yellow lines. Three potential glycophores are shown by black dots at locations 14, 34 and 35. See text. From Jin et al., 2003.

Assadi-Porter et al. (2000, pg 259) has provided a figure equivalent to that of Jin et al. that is slightly easier to interpret. Their paper also focused on two different locations, pGlu1 and His31. Caldwell et al. provided an additional schematic of brazzein with some additional specifics. They asserted brazzein contains • a single α-helix (residues 21-29) • a three-stranded anti-parallel β-sheet (strand I, residues 5-7) • strand II (residues 44-50 • strand III (residues 34-39) • a poorly defined stretch of random coil between (residues 9-20) • a β-turn between strands II and III (residues 40-43) • a poorly defined N-terminus (residues 1-4) • a poorly defined C-terminus (residues 52-53. These descriptions were based on the overlay of a series of conformers showing the configuration of the molecule at individual times. They also provided additional detail regarding the disulfide bonds of the molecule. Their analyses were based primarily on examining conventional structural Signal Generation & Processing 8- 265 details. They did note a disulfide bridge between the N-terminus and the C-terminus. Their analysis also focuses attention on Arg43 and on His31 as important to sweetness. Their figure 2b shows some additional structural details concerning the residues along the brazzein chain. A comparison of the findings of Jin et al, Caldwell et al. and Assadi-Porter et al. will be made in Section 8.5.9.6. Brazzein has been found to exist in two very similar forms. The first is 54 residues long and has a “non-amino acid” group at its N-terminus called pGlu or pyroglutamic acid. The alternate variant is 53 residues long and is without this unique group (des-pGlu1). The numbering in both cases begins with the pGlu group even if absent. Assadi-Porter et al. (2000, pg 252) asserts the des- form is perceived as 2:1 sweeter than the pGlu form by humans. Figure 8.5.9-3 reproduces the amino acid sequence for brazzein from Caldwell et al.327. The figure includes three arrows pointing to residues of particular focus. These may also be modified and therefore “non amino acids.” The focus is partly due to the location of the residues #38-#45 near the exterior envelope of the molecule and on a curve that may contribute to a larger than typical d-value between some of the orbitals in the side chains of the residues. Pyroglutamic acid (pGlu, pE and a.k.a. pidolic acid) forming the N-terminal of the brazzein amino acid sequence has lost its carboxylic group during peptide bond formation, thereby losing its potential to act as an acidophore within brazzein. However Glu41 and Asp29 retain the ability to act as acidophores if adequately exposed to the GR’s for the G-path.

Figure 8.5.9-3 The amino acid sequence of brazzein with three positions marked. The potential for a modified serine residue at location 14 or 34 and a modified glycine at location 35 is worthy of further study. See text. From Caldwell et al., 1998.

- - - -

[xxx figure and text somewhat duplicates figure and text later re b-turn ] A fully functional 3D representation of brazzein is available from the PDB bank of Europe. That bank supports a viewer called Astex. Like other 3d viewers encountered in 2013, they are not fully implemented. Astex 3.0 is provided on an “as-is” basis under a GNU license with a latest copyright date of 2007 and at best a sketchy manual (click on user commands on left navigation panel). Figure 8.5.9-4 shows the complexity of brazzein in the vicinity of the β-turn using this viewer. Astex operates, and shares many commands, with Jmol and more specifically Discovery Studio vers. 3.5. Measuring the d-values of the various orbital pairs available in this area leads to the conclusion that the sharp curvature in this turn and the general location of Glu41 and Arg43 on the outside of the curve leads to spacings higher than those that can support a glycophore in this region. The curvature and location of the residues in the range of residues 30 to 33 may offer higher probability of locating the glycophore(s).

327Caldwell, J. Abildgaard, F. Dzakula, Z. et al. (1998) Solution structure of the thermostable sweet-tasting protein brazzein Nat Struct Biol vol 5(6), pp 427-431 266 Neurons & the Nervous System

Figure 8.5.9-4 A partial 3D ball and stick representation of the β-turn in brazzein EDIT. The area to the left of the broad yellow sheet is the β-turn of interest. Residue lys42 is at lower left, residue Glu41 is center left and residue Arg43 is at upper left. The interesting possibility that one of the carboxylic oxygen atoms of Glu41 and the oxygen associated with the peptide bond leading to Lys42 might form an AH,B group with a d-value near 2.8 Angstrom and the location of the guanidinium group of Arg43 might form the dispersion centroid, X, needs additional examination.

- - - - Assadi-Porter et al. (2000) designed a series of 15 synthetic genes that when placed in e-coli could lead to expression of a series of brazzein derivatives328. The experiments were quite successful. All but one of their derivatives appears to have folded properly into their native state and were perceived as super sweet by humans. Whether their synthetic genes were able to encode processes necessary to create non-amino-acid inserts into the amino acid chain is not clear. They did note the N-terminal and C-terminal of their derivatives were physically quite close together. They focused on residues near Arg43 on a portion of a flexible loop as particularly pertinent to sweetness. When unfolded, all of their derivatives became tasteless. They also noted that recreating their derivatives using only D-amino acids led to tasteless products. While possibly not adequately worded, they assert that Arg43 is necessary for sweetness in brazzein.

[xxx some duplication of words re glycine and serine ] Glycophores discussed in this work so far have all been found quite widely and involve well documented chemistry. For one of the above six proteins perceived as sweet to satisfy the requirements of containing at least one glycophore, they need to exhibit at least one structure easily accessible from outside of the protein periphery. This glycophore would consist of an HO-C- C-O structure with a d-value between the two oxygen atoms of nominally 2.82 Angstrom with a potential range of ±6% or more. This requirement suggests a trans- arrangement along a curved portion of the protein similar structurally to a simple phenol ring. Such a structure could be obtained from a serine absent its normal methylene group (CH2) or a hydroxylated glycine incorporated into a polypeptide chain. While formation of such an amino acid based on the genetic code is unlikely (no triplet code is recognized for them, they could be formed after insertion of the parent amino acid into the polypeptide chain (Lehninger, 1970, pp 72). Thus the glycine residue at locations 35 could be oxygenated (or involve a hydroxyl substitution for the terminal hydrogen) or the methylene group could be removed from the serine residue at either location 14 or 34. It is not obvious the current methods of protein decomposition and

328Assadi-Porter, xxx (2000) pp 259-xxx Signal Generation & Processing 8- 267 identification using the conventional amino acid analyzer329 would recognize such a “rare amino acid.” However, Caldwell et al. and others have recognized pGlu in a similar situation as discussed above. - - - - [xxx change to reflect above possibilities for modified amino acids ] A new characterization of some of the observations of Jin et al. regarding brazzein are of interest. “Two features are noteworthy. One is that addition of an N-terminal Ala or mutation of the side chain from Asp to Asn did not change the sweetness. The other is that mutations of residues close to one another in the protein sequence led to remarkable differences in sweetness. For example, mutation at position 29 (changing Asp29 to Ala, Lys or Asn) made the molecule much sweeter than WT (the wild type of Brazzein), while mutations at positions 30 or 33 (Lys30Asp or Arg33Ala) removed all sweetness. The same pattern occurred again at the $-turn region, where Glu41Lys gave the highest sweetness score among the mutants tested, whereas a mutation two residues distant (Arg43Ala) abolished the sweetness.” Their studies showed that it was not the proteins themselves that were gustaphores but that the amino acid residues at specific locations were gustaphores. “Specifically, our findings suggest that residues 29–33 and residues 39–43, plus residue 36 connecting these stretches, as well as the N- and C-termini determine the sweetness of brazzein. As a consequence of these studies, brazzein variants identified to have enhanced sweet qualities could become candidates for a new generation of low-caloric natural sweeteners.” Such residues all contain the glycophore found in any amino acid. They would be expected to be functional gustaphores based on the hypothesis presented here if they were physically exposed sufficiently to form a DACB with the G-path GR’s.

They noted the importance of changing the net charge on the protein by substitution of a residue. The role of charge in their discussion of the charge on the molecule differs significantly from the dipole potential of the molecule to be discussed in a following paper. That paper shows that the dipole potential of a molecule when bound to a GR is measured by the sensory neuron amplifier following the transduction mechanism at the surface of the GR.

- - - - [xxx edit paragraph and indent ] Their focus on locations 29 and 41 is interesting. Both of these locations were initially suitable as acidophore sites according to this hypothesis. The glutamic acid residue at location 41 appears to be the primary glycophore for brazzein at this time with changes at location 29 significantly impacting the total charge on the molecule when in the DACB relationship with the G-path GR..

Note, when discussing the effect of changing the residue at location 29, the protein is still acting as a glycophore during such experiments. This is a strong indication that the effective glycophore of brazzein is the glutamic acid residue at location 41. This residue appears to be on the exposed surface of the molecule near the β-turn nominally at location 43.

- - - -

In 2003, Assadi-Porter et al. reported on a study of the hydrogen bonds within the folded forms of brazzein in conjunction with their sequence-function investigations330. The study centered on the hydorgen bonds interconnecting residues within brazzein. They examined five single point mutants using isotope labeled atoms by high resolution NMR. This is basically a shotgun approach as no theory was offered as to how such hydrogen bond changes should affect the sweetness of these variants of their recombinant des-brazzein. They repeated their previous assertion that,

329Wade, L. (xxx) Organic Chemistry, 7th Ed. NY: Pearson Chapter 24, section 24.9B http://www.pearsonhighered.com/showtell/wade_032159231X/wade_032159231X.html

330Assadi-Porter, F. Abildgaard, F. Blad, H. & Markley, J. (2003) Correlation of the Sweetness of Variants of the Protein Brazzein with Patterns of Hydrogen Bonds Detected by NMR Spectroscopy J Biol Chem vol 278(33), pp 31331–31339 268 Neurons & the Nervous System

“The characteristics that make a compound taste sweet to humans are not well understood.” They did offer a table of sweetness threshold and sweetness intensity as a function of their mutations. They also offered an extensive table of apparent lengths of the hydrogen bonds they encountered based on an empirical equation from Cornilescu et al (1999). The values clustered around 2.91 Angstrom. 8.5.9.3 The high electrical charge of the peptides

Understanding the dipole potential of the sweet proteins when in their DACB state with a GR could be extremely valuable information. This potential plays a major role in the perception of the intensity of all gustaphores. As specifically noted by Assadi-Porter et al. (2000, pg 252), “Brazzein is highly charged and polar. . .Charged and polar residues also appear to be critical for the sweetness of thaumatin and monellin Their figure 1 has provided data on “Protein elution patterns on a preparative Vydac reversed phase HPLC column for brazzein isolated from fruit, recombinant wild-type (des-pGlu1-brazzein), and selected mutant brazzeins. “ They also provided considerable one- dimensional 1H-NMR data at 600 and 750 MHz. Their conclusions are noteworthy, particularly their assertion that the presence of Arg43 is essential for sweetness and that the area near Arg43 is critically involved in the perceived sweetness of these materials. They provide an explanation for the small extraneous peaks around 25 minutes in their HPLC results that may be subject to additional discussion.

Jin has studied the difference in gustatory performance between brazzein and a variety of its derivatives. These derivatives remained in the appropriate condition to act as glycophores during their evaluation. An important fact can be noted about locations 29 and 41. Both are occupied by the only two common acidic amino acids (negatively charged). Changing the ligand at these locations can be expected to, and did result in, major changes in the perceived sweetness of these molecules.

Xue has provided some charge data for monelllin331. Quoting from their abstract;

“A small number of proteins have the unusual property of tasting intensely sweet. Despite many studies aimed at identifying their sweet taste determinants, the molecular basis of protein sweetness is not fully understood. Recent mutational studies of monellin have implicated positively charged residues in sweetness. In the present work, the effect of overall net charge was investigated using the complementary approach of negative charge alterations. Multiple substitutions of Asp/Asn and Glu/Gln residues radically altered the surface charge of single-chain monellin by removing six negative charges or adding four negative charges. . . Despite different sizes and non-homologous primary, secondary, and tertiary structures (where known: brazzein [9]; monellin [10]; neoculin [11]; thaumatin [12]), these proteins share not only intense sweetness that develops more slowly than that of sugars, but an aftertaste that can persist for hours ([8,13] and data not shown).”

Xue et al. also note, “Despite different sizes and non-homologous primary, secondary, and tertiary structures, the sweet proteins share not only intense sweetness that develops more slowly than that of sugars, but an aftertaste that can persist for hours.” These characteristics are almost certainly related to the size of the molecules and their more limited Brownian motion when in the presence of predominantly smaller molecules. Their discussion opens with “Many years of study have eliminated numerous proposals about the molecular basis of sweet protein taste.” They did not offer any substantial change in this situation. They do dismiss the importance of local pockets on the surface of the GB’s and the so- called “sweet finger” of the protein wedging into such pockets. Xue et al. offer no theoretical framework or model to ground their primarily conceptual discussion. Rather than explore specific binding between stimulants and GR’s, they discuss the electrostatic properties of their proteins

331Xue, W.-F. Szczepankiewicz, O. Thulin, E. Linse, S. & Carey, J. (2009) Role of protein surface charge in monellin sweetness Biochim Biophys Acta vol 1794, pp 410–420 Signal Generation & Processing 8- 269 (native and derivative) in terms of their coulombic properties contributing to the selection and binding mechanisms. They also consider their coulombic properties as rate determining in diffusion rather than as properties measured by the stage 1 transduction process and reported as intensity information. Xue et al. do note the perceived sweetness of the sweet proteins does appear to increase as the charge on the stimulant becomes more positive, up to a saturation point. Such a polarity ad such a saturation are totally compatible with the electrophysiology of the stage 1 sensory neurons (Section 8.xxx)

- - - - Comparing the electrolytic properties of sucrose (which immediately hydrolyzes to glucose in solution) with those of the six sweet proteins (and especially brazzein) by chromatography and electrophoresis could provide important information about both their absolute and relative electrical properties. Performing similar comparisons with derivatives of brazzein prepared by Jin et al. would also provide excellent information. However, note the qualifications concerning chromatography in Section 8.5.9.5. 8.5.9.4 Studies related to monellin

Hellenkant & van der Wei have studied two of the best known of maybe a dozen “sweet” proteins, Monellin and thaumatin332. Monellin is a protein from Discoreophyllm cumminsii Diels and thaumatin is a protein from Thaumatococcus daniellii Benth. They note the extreme sweetness of these two proteins compared to sucrose, 60,000 to 100,000 times sweeter. Based on this work, they employ an AH,B,X coupling to the sensory receptors of OR 2. They provide several additional clues to how these proteins achieve the perception of sweetness and note their performance is largely limited to catarrhine primates. They elicit little perception of sweetness in non-catarrhine primates and all non-primates tested. In the future, this information might help determine the difference(s) between the OR 2 receptors of these distinct groups. The proteins are soluble in water at low concentrations and have a long lasting aftertaste suggesting their low volatility. The aftertaste is not easily affected by washout procedures.

Kohmura et al. have undertaken a decades long study of provided papers on monellin beginning in a journal no longer published. However the papers are available from the National Institute of Informatics of Japan333. The large number of papers are highly redundant with regard to the overall properties of monellin but provide useful highlights of their various experiments

The first abstract follows334,

“Monellin, a sweet protein, consists of two non covalently associated polypeptide chains, the A chain of 44 amino acid residues and the B chain of 50 residues. Two different primary structures have been reported for each of the A and B chains. The A and B chains corresponding to one of the reported monellin structures were synthesized by the stepwise solid-phase method using the Fmoc strategy in overall yields of 14.1% and 5.6%, respectively. The characterization of the synthetic peptides by HPLC, FAB-MS, amino acid analysis and sequencing fully supported the expected structures. The individual synthetic A and B chains were not sweet. Combination of the two chains, and subsequent HPLC purification gave monellin in a yield of 53.9%. The synthetic monellin had a distinct, lingering sweet taste (4000 times sweeter than sucrose) and was crystallized by a vapor diffusion method. The synthetic product was identical to natural monellin by HPLC, but not by tryptic mapping. These results indicate that the reported structure for monellin differs slightly from that of natural monellin.

332Hellenkant, G. & van der Wei, H. (1989) Taste modifiers and sweet proteins In Cagan, R. ed. Neural Mechanisms in Taste Boca Raton, FL: CRC Press Chap. 4

333http://ci.nii.ac.jp/info/en/articles/quickguide.html

334Kohmura, M. Nio, N. Ariyoshi, Y. (1990a) Solid-phase synthesis and crystallization of monellin, an intensely sweet protein Agri Biol Chem vol 54(6), pp 1521-1530 270 Neurons & the Nervous System

The fact that HPLC and tryptic mapping gave different results with regard to the natural and synthetic monellin may be of great importance. The second paper includes a complete amino acid sequence for monellin as shown in Figure 8.5.9-5335. Kohmura et al found that monellin consisted of two distinct peptide chains, A with 44 residues and B with 50 residues, that were not covalently bonded. They noted that two different primary structures had been reported for each of these chains with differences at several locations. After furhter investigation, the questionable locations were assigned the values of Asp22, Glu25 & Asp26 for the A chain and Glu49 & Asn 50 for the B chain. It is noteworthy that fusion of the two chains at the C-terminus of the B chain and the N-terminus of the A chain did produce a perception of sweetness for the result. Additional HPLC and tryptic data is found in the paper.

335Kohmura, M. Nio, N. Ariyoshi, Y. (1990b) Complete amino acid sequence of the sweet protein monellin Agri Biol Chem vol 54(9): pp 2219-2224 Signal Generation & Processing 8- 271

Figure 8.5.9-5 Complete amino acid chain for monellin. Top; A-chain of nomellin. Approximately 10% of the material carries an extra Phe at the N-terminus. Bottom, B-chain. Approximately 19% of which carries an extra Thr at the N-terminus and approximately 24% lacks the N-terminal Gly. From Kohmura et al., 1990b. 272 Neurons & the Nervous System

The third paper focused on derivatives of monellin336. They sought to determine the mode of interaction with the GR of the G-path by various sweet peptides but noted the difficulty of studying their peptides because of their ability to assume a variety of conformations in water. They then reverted to the study of the complete proteins. They continued to explore their materials without a clear theoretical framework. The paper did not provide any significant conclusions. The xxx paper was focused on the solid crystalline state337. It contains significant information that has not been reviewed in detail. Kohmura and team is continuing to study monellin. Search scholar for more recent articles.. In a 1994 paper338, they continued to seek to understand the selection and transduction of monellin by replacing various amino acids while maintaining an active perception of sweetness. As for other investigators, this negative approach does not lead to an explanation of transduction mechanism, only to the determination of what other residues increase or decrease perceived sweetness. No schematic could be found among their work showing the folding of monellin like the schematics of Jin et al. and Assadi-Porter et al. for brazzein. 8.5.9.5 Recent experiments to characterize rare amino acids

Chaudhuri & Yeates have recently described the fact that some proteins are produced from alternate readings of accepted stop codons of DNA and tRNA339. They note, “The difficulty arises from the distinction that, unlike other amino acids, rare amino acids are not coded for by dedicated codons. Instead, they are incorporated in special circumstances by the UGA (opal; selenocysteine) and the UAG (amber; pyrrolysine) codons, which are ordinarily interpreted as stop signals to terminate translation.” Their may be other situations yet to be discovered. This subject matter is very much on the forefront of biological research340.

Shen et al. have provided an excellent paper on identification and molecular cloning of the protein required for generating the rare protein selenocysteine341. They identified this specific protein by way of an unusual presence (a mobility shift) in an RNA electrophoretic assay. However, they did not identify, or discuss the identification of, the actual amino acid.

336Kohmura, M. Nio, N. Ariyoshi, Y. (1990c) Solid-phase synthesis and crystallization of [Asn22, Gln25, Asn26]-A-chain-[Asn49, Glu50]-B-chain-monellin, an analogue of the sweet protein monellin Agri Biol Chem vol 54(12), pp3157-62

337Kohmura, M. Nio, N. Ariyoshi, Y. (1991) Solid-phase synthesis of crystalline monellin, a sweet protein Agri Biol Chem vol 55(2), pp 539-45

338Kohmura, M. Nio, N. Ariyoshi, Y. (1994) Solid-phase synthesis and structure-taste relationships of analogs of the sweet protein monellin Biosci Biotech Biochem vol 58(8), pp 1522-1524

339Ghaudhuri, B. & Yeates, T. (2005) A computational method to predict genetically encoded rare amino acids in proteins Genome Biol vol 6, pg:R79

340Wang, Q. Parrish, A. & Wang L (March 2009). "Expanding the genetic code for biological studies". Chem. Biol vol 16(3), pp 323–36. doi:10.1016/j.chembiol.2009.03.001

341Shen, Q. Wu, R. Leonard, J. & Newburger, P. (1998) Identification and Molecular Cloning of a Human Selenocysteine Insertion Sequence-binding Protein J Biol Chem vol 273(10), pp. 5443–5446, Signal Generation & Processing 8- 273 Goodsell described the more advanced genetic coding used to identify and form selenocysteine342. “Quite surprisingly, cells modify their genetic code to add selenocysteine into their proteins. The basic genetic code used by all organisms on Earth specifies twenty amino acids, along with a few stop codons. In order to add a 21st amino acid to this code, cells reinterpret UGA stop codons in special cases. But this causes a problem: how does a cell know when to read UGA as "stop" and when to read it as "selenocysteine"? This is done by using a special signal sequence that is located after the UGA codon. In bacteria, this signal is immediately after the UGA codon, in the coding portion of the messenger RNA. In our cells, the signal is much further away at the end of the coding sequence. As shown on the next page, this sequence is recognized by a special elongation factor that delivers the selenocysteine tRNA to the ribosome at just the right moment.” This mechanism is called translational recoding. The description of this mechanism is not new343. Goodsell goes on to note, “About 15 selenoproteins have been found in human cells. These include several deiodinase enzymes that are essential for the generation of thyroid hormones, several glutathione peroxidases that are important for the conversion and detoxification of compounds with reactive oxygen atoms, and several proteins of unknown function.” Gromer et al. have provided more information on these selenium containing proteins344. “ Xu et al. have described an entire seleno protein family of rare amino acids in considerable detail, including locating some of these rare amino acids in relation to their common brethren345. Their figure 2 shows a common binding site (lysine) across a wide variety of selenium protein derivatives. Their figure 6 shows a potential transition from serine to selenocysteine (Sec). Their 31P-NMR spectra show the location of one of these unusual amino acids in the context of other amino acids, Figure 8.5.9-6. The presence of monoselenophosphate (SeP) is clearly displayed in the left-most three frames and it is absent from the right-most. The complete meaning of this display can be found in the paper. The point of interest here is that the yield in the processes leading to this display was not high, but the ability of NMR to identify the product clearly was excellent. Three digit accuracy was obtained in labeling the SeP at +23.2 ppm. The height of the material in the upper global views could easily be missed, but the signal to noise ratio in the lower views is quite adequate.

342Goodsell, D. (2008) Selenocysteine Synthase PDB protein data bank doi: 10.2210/rcsb_pdb/mom_2008_8 ePub Version

343Böck A. and Stadtman T. (1988) Selenocysteine, a highly specifi c component of certain enzymes, is incorporated by a UGA-directed co-translational mechanism. Biofactors vol 1, pp 245–250

344Gromer, S. Eubel, J. Lee, B. & Jacob, J. (2005) Human selenoproteins at a glance. Cell Molec Life Sci vol 62, pp 2414-2437

345Xu, X.-M. Carlson, B. Mix, H. et al. (2007) Biosynthesis of Selenocysteine on Its tRNA in Eukaryotes PloS vol 5(1), pp 0096-0104 274 Neurons & the Nervous System

8.5.9.6 Proposed mechanism of sweet protein perception

Figure 8.5.9-6 In vitro NMR spectroscopic analysis showing SeP presence. Top row; conventional 31P-NMR of four samples prepared by different paths. Lower row; expanded views showing excellent signal-to-noise. No SeP was created by the reaction leading to the right-most frames. See original article for details. From Xu et al., 2007.

Although not as well supported as the main hypothesis of gustation, primarily because of limitations in analytic instrumentation, the bonding of the sweet proteins to the nominal G-path GR appears quite defendable. This is particularly true if the laboratory record is broadened to allow for the presence of rare amino acids in brazzein, and presumably all sweet proteins, rather than just the 20 common amino acids identified by most present laboratory instrumentation.

The following discussion is meant to be global, but where necessary it may be restricted to the perception of sweetness by humans and old world monkeys. In general, no change in perceived taste quality was found with changes in concentration of brazzein or its derivatives. Signal Generation & Processing 8- 275 As noted above, the present state of the art is quite limited in this leading edge research area. it is proposed that the perceived sweetness of certain proteins is due to three features of their interaction with the G-path GR’s of the system; • the sweet proteins exhibit glycophores located along the external periphery of their molecular space, generally described as a point along a β-turn of a flexible loop, • these glycophores participate in an AH,B,X relationship with the GR’s, • the AH,B,X relationship involves a DACB between two orbitals of the glycophore with a d-value approximating 2.82 Angstrom and the dispersion centroid, X, is separated from the AH,B orbitals by the same distance as for other super-sweet glycophores, • the very high dipole potential of these proteins is the major contributor to the great perceived sweetness of these proteins and the high signal response of the stage 1 sensory neurons. - - - -

The work of Goodsell and others at the beginning of the 21st Century to identify the 21st (common) amino acid has gone a long way toward rewriting how the genetic code is actually read, to include reinterpreting the stop codes as conditional, a mechanism known as translational recoding. However, this area remains provisional. Assadi-Porter et al. (2000a) have asserted that brazzein can be synthesized in the laboratory with good yield using totally conventional techniques (without introducing a translational recoding ) and a codon of ATG for the modified amino acid, pGlu or (pidolic acid) as shown in Figure 8.5.9-7. The conventional glutamic amino acid is formed using the codon GAA.

- - - -

Figure 8.5.9-7 Amino acid sequence of brazzein and DNA sequence of synthetic gene. A; top line, the pGLU (pyroglutamic acid) residue is shown at position 1 of the complete brazzein protein. Second line, brazzein absent pGlu (des-pGlu1). The bridges represent four disulfide bond pairings. C; the DNA sequence of the synthetic gene used for heterologous production of brazzein in E. coli. The stop code is TAG. Excerpt from Asadi-Porter et al., 2000a. 276 Neurons & the Nervous System

[xxx provide a discussion here. The challenge is to locate and document the necessary orbital configuration(s) along the amino acid sequence. In the case of brazzein, there is general agreement that the glycophore is located along the β-turn of the flexible loop between locations 34 and 45 or an equivalent location between residues 30 and 35 (as suggested by the arrows in the above figure from Jin. The amino chains in these regions, (SGECFYDEKRNL and KHARSG) offer a variety of side chains which could support the required d-value and dispersion centroid distance. The simplest configurations would be represented by modifications to glycine (G) through β-turn of a flexible loop hydroxylation and serine (S) through demethylenation that provided an oxygen orbital attached to the second carbon resulting in a glycophore of HO-C-C-O and a third orbital at the dispersion centroid. In this case, the residues would be modified similarly to the pGlu (pE) residue at the N-terminus. The X orbital is likely to be an N associated with the a distant peptide bond. Additional analysis will be required to identify this orbital since the distance to the dispersion centroid will strongly depend on the local curvature of the amino acid chain. Figure 8.5.9-8 shows the situation under discussion. The left frame shows the nominal schematic of a peptide chain. Each shaded group is planar and includes a peptide bond. This bond has considerable similarity to a double bond and does not allow free rotation. However, as noted, the bonds outside of the planar groups are free to rotate. As a result, the ultimate bond angles along the chain are subject to other constraints as noted below for the α-helix. The oxygen-carbon bond shown within each planar group is also more constrained than indicated, suggestive of a dual bond. In this application, it is suggested the left-most R-group is another oxygen atom with possibly an additional ligand. The two left-most oxygen atoms are at the appropriate distance to form a glycophore of the AH,B type. It is proposed that a “super” glycophore of the AH,B,X type can be formed in conjunction with a nitrogen located farther to the right along the chain. The distance between the AH,B orbitals and the X orbital is highly dependent on the angles of rotation about the non-planar carbon.

Figure 8.5.9-8 Options available for the dispersion centroid supporting a glycophore in an α-helix. Left, a schematic of the peptide chain focusing on the individual planar groups (shown shaded), the out-of-plane R-groups and the points of rotation between the planar groups. Right, the nominal spacing of the same structure in its minimal energy α-helix arrangement. The residue side-chains are not shown. See text. A collage modified from Lehninger, 1972. Signal Generation & Processing 8- 277

The right frame shows the minimum energy configuration of the schematic peptide structure, the α-helix. The frame from Lehninger has been modified to include the rigidized oxygen atom of the planar structure on the left and the out-of-plane oxygen on the right. Using the nominal spacings shown, it is clear there are a number of nitrogen orbitals that could support a dispersion centroid at a distance from the AH,B orbitals that satisfies the requirements the Shallenberger team and Kier discussed in Section 8.5.3. It is not likely that the appropriate distance to X can be achieved within a minimum energy α-helix. Under relaxed conditions of a less curved, peptide chain, virtually any required dispersion distance necessary to support the AH,B,X configuration could be provided. The forming of hydrogen bonds over extended intra-chain distances is well recognized in protein chemistry. Based on the above structural considerations, the challenge is to find an amino acid structure that can provide an oxygen orbital attached to the left-most (out-of-plane) carbon of the left frame. None of the normally discussed 20 amino acids forming proteins meet this requirement. However there are other common amino acids as well as several rare amino acids that might provide the necessary orbital, or like in the case of pGlu, an amino acid might be modified within the protein after its expression following its DNA code using the mechanism of translational recoding.

It is possible that threonine (T) could provide the oxygen orbital via its hydroxyl group, proline (P) could provide the necessary orbital via the nitrogen in its ring structure or alanine could provide the desired structure via a conceptually simple methyl-hydroxyl transferase. However, the simpler solution would be to modify either glycine (G) by hydroxylation or serine (S) by demethylenation. While no threonine is found within the brazzein peptide chain, the other three candidates do occur within the zones expected to include the glycophore of brazzein (P12, S14, S34 or G35). The latter three candidates have been marked in the above brazzein peptide chain from Caldwell et al.

- - - - A set of images of brazzein are available from the European PDB bank under code 1BRZ. A 3D projection of brazzein is available at http://www.ebi.ac.uk/pdbe-apps/nmrviewer/cgi-bin/index.py?page=main&pdbId=1brz&wiz= cluster&type=2

It provides an amazing level of detail with the oxygen and other atoms shown in color. The ball and stick model suggests the various cartoon models only show a computed backbone, not necessarily the actual location of the amino acid residues. - - - -

It is not obvious to this investigator whether the conventional laboratory identification procedures, which are generally based on relative diffusion rates for the individual amino acids would indicate clearly the presence of a rare amino acid in place of a glycine or serine. The description of rare amino acids by Lehninger in 1972 or earlier (page 72) appears applicable at this point. “It is likely that other rare amino acids of proteins will be discovered. . . that they will be derivatives of presently known common amino acids and that they will be limited in occurrence to single proteins or to a class of proteins.” The six sweet proteins could easily be such a class. The material of Shen et al. and Xu et al. in the previous section support these earlier observations. Most of the rare forms described by Lehninger involve chemical additions. Lehninger describes an azaserine on page 73 approaching the desired amino acid form but not normally found in proteins. Derivatives of nearly all the well-known 20 amino acids have been isolated and described. There are at least 220 (some say 500346) rare amino acids, mostly found in plants. At least 25 human proteins include selenocysteine (Sec), a rare amino acid, in their primary structure347.

346Wagner, I. & Musso, H. (1983) New Naturally Occurring Amino Acids. Angew. Chem Int. Ed. Engl. 22 (22): 816–828. doi:10.1002/anie.198308161*

347 Kryukov GV, Castellano S, Novoselov SV, Lobanov AV, Zehtab O, Guigo R, et al. (2003) Characterization of mammalian selenoproteomes. Science vol 300, pp1439–1443 278 Neurons & the Nervous System

Kryukov et al have been very specific concerning identifying the rare amino acids. “In the universal genetic code, 61 codons encode 20 amino acids, and 3 codons are terminators. However, the UGA codon has a dual function in that it signals both the termination of protein synthesis and incorporation of the amino acid selenocysteine (Sec) (1–3). Available computational tools lack the ability to correctly assign UGA function. Consequently, there are numerous examples of misinterpretations of UGA codons as both Sec codons (4) and terminators (5, 6), including annotations of the human genome (7, 8), where no selenoproteins have been correctly predicted. With 18 human selenoprotein genes previously discovered (3), the estimates of the actual number of such genes vary greatly (9). All previously characterized selenoproteins except selenoprotein P (10) contain single Sec residues that are located in enzyme-active sites and are essential for their activity. Thus, misidentification of UGA codons leads to a loss of crucial biological and functional information. Sec is cotranslationally incorporated into nascent polypeptides in response to UGA codons when a specific stem-loop structure, designated the Sec insertion sequence (SECIS) element, is present in the 3 untranslated regions (UTRs) in eukaryotes.” They have provided a tabulation in their figure 2 of false positive reporting at locations occupied by a selenoprotein. Analysis of proteins by chromatography is probably the ultimate example of overlooking the facts using Bayesian experimentation. The nominal chromatograph protocol assumes there are only 20 amino acids in the material under examination and assigns names to each residue based on its relative position with this restricted order of 20 amino acids. The conventional chromatography technique offers no opportunity to uncover the presence of rare amino acids.

Figure 8.5.9-9 shows the output of a typical automatic recording chromatograph. Note the presence of ammonia as a reference among the other 17 residues labeled by their conventional amino acid names. Where do the other 200-500 rare amino acics appear in this representation.

Figure 8.5.9-9 The relative positions of the amino acid residues in chromatography Where does selenocysteine (Sec) appear in this presentation? Where do the other rare amino acids appear? The area under each peak is proportional to the amount of each amino acid in the mixture. See text. From Lehninger, 1972.

Based on recent activity in a broad group of laboratories (only addressed briefly above), it appears quite likely that at least one rare amino acid has been formed and inserted into a β-turn (or equivalent) of the six known sweet proteins thereby allowing their participation in transduction by DACB at the conventional GR of the G-pathway.

The following two sections will focus on the requirements for DACB between a sweet protein and a GR, and the requirements/affects of the location and charge associated with a specific residue Signal Generation & Processing 8- 279 on the preceived intensity of the stimulant. It will be assumed in those sections that each sweet protein only supports one glycophore. The initial discussion will bound the problem of DACB binding and the identification of the residue forming the dispersion centroid, X. The following discussion will bound the potential for changing the dipole potential of the stimulant and the resulting perceived sweetness intensity. A limitation of prior analyses has been the Bayesian assumption that proteins only contain a mixture of the 20 common amino acids, and the position enunciated by Caldwell (1998, 1st paragraph) that the sweet proteins contain no carbohydrates. A limitation of the current analysis is its dependence on the frequently qualitative statements concerning the replacement of one residue by another in experiments. Caldwell et al. stated the generality, supported by citations, that “Chemical modification of the N-terminus, cysteine, hisitdine, arginine, lysine, tyrosine, and glutamate or aspartate residues of brazzein have been found to reduce or abolish its sweet taste.” This statement appears counter to that of Jin based on neuron recordings, “Asp29Ala, Asp29Asn, and Glu41Lys elicited strong responses with fast onset and with a temporal profile reminding of sucrose rather than WT brazzein, while neither mutant Glu36Ala or His31Ala gave a response.” Ratios are more useful than comments that one is sweeter (or less sweet) than the other variant.

It is postulated that a change from an AH,B,X configuration to an AH,B configuration may cause a major change in perceived sweetness (by as much as 10,000:1), but not necessarily a total loss of perceived sweetness. Thus separating the change in the bonding mechanism should be kept separate from the perception mechanism.

Assadi (2000, pp 259-) Most of the mutants in the flexible loop region around Arg43 were found to decrease, but not completely eliminate, the sweetness of the protein.

Jin also noted “mutations of residues close to one another in the protein sequence led to remarkable differences in sweetness.”

Jin and other investigators have reported multiple situations where modification of WT brazzein led to higher perceived sweetness; a result suggesting the precise form of WT brazzein may not be optimized for sweetness alone.

Following the earlier discussions, the transduction mechanism will be divided into two steps; the mechanism of gustaphore/GR selection leading to signaling along the G-pathway, and the mechanism of dipole potential measurement by the first amplifier of individual G-pathway stage 1 sensory neurons. 8.5.9.6.1 Details of proposed sweet protein selection

This section will focus on a comparison of the findings of Jin et al, Caldwell et al. and Assadi-Porter et al. within a broader realm consistent with the inclusion of rare amino acids in the sweet proteins.

The first order conclusions are that; • the AH,B glycophore of brazzein is found between residues 40 and 43, most likely at residue 41 • a dispersion center, X, is also located in this range and most probably found at location 43 (probably involving Arg43, an amino acid residue incorporating a highly charged guanidinium group). • the slow rise and fall in the perception of sweetness from the sweet proteins is a result of their large size and relatively low average diffusion velocity in the face of Brownian motion by the smaller constituents of saliva. • a rare amino acid is found at location 43 (not generally identified by conventional–and automated–chromatography techniques). • the highly polar character of brazzein makes almost any modification in the amino acid sequence outside of locations 40 and 43 significant in modifying the dipole potential of the molecule and thereby its perceived sweetness (as discussed in the next section). • 280 Neurons & the Nervous System

The conventional amino acid sequence of interest for brazzein is Phe38, Tyr39 Asp40, Glu41, Lys42, Arg43, Asn44 with the two residues in italic shown to expand the margins of the poorly defined set. However, it is possible that any of these amino acids may in fact be a rare amino acid not found within our past Bayesian assumption. - - - - - Summary positions related to brazzein

Assadi: two regions of the protein are critical for the sweetness of brazzein; a region that includes the N – and C–termini, which are located close to one another, and a region that includes the flexible loop around Arg43. Jin: residues 29-33 and residues 39-43, plus residue 36 connecting these stretches as well as the N – and C–terminals determine the sweetness of brazzein. Jin: WT brazzein has no bitter after taste or any other secondary taste quality Jin: the taste quality associated with brazzein did not change specificity with concentration.

Caldwell: One site is near the N-terminus and contains athe unique histidine, His31, which is within 12 Angstrom of Arg33 and near examples of other residues of the types impoicates as critical for sweet taste, including tyrosines (residues 8, 24, and 51), lysines (residues 3, 5, 6, 27 and 30), glutamates (residues 9 and 53), and aspartates (residues 2 and 50). The second candidate site centers around the other arginine residue, Arg43, . .

It is generally agreed that sweetness of the sweet proteins is lost when unfolded.

- - - -

Relevance of location 4

Assadi; disruption of one disulfide bridge by mutation of Cys4 is sufficient to abolish the sweetness.

- - - -

Relevance of location 29

Jin: mutation of Asp29 to Asp29Ala, Asp29Asn or Asp29Lys markedly increased sweetness over WT. The notation describes substitution of the second residue in place of the first.

- - - -

Relevance of location 30

Jin: mutation of Lys30 to Lys30Asp removed all sweetness at the concentration used. - - - - Relevance of location 31

Jin: a mutation from His31 to Ala31 introduced a significant loss in sweetness (although a previous experiment had found an increase in sweetness threshold by this change).

- - - - Relevance of location 33 Signal Generation & Processing 8- 281

Jin: mutation of Arg33 to Arg33Ala removed all sweetness at the concentration used. - - - - Relevance of location 36 Jin: mutations from Glu36 to Glu36Ala, Glu36Gln or Glu36Lys destroyed the perception of sweetness.

- - - - Relevance of location 41 Caldwell: did not address the perceptual significance of replacing Glu41. Jin changes at location 41 from Glu41 to Glu41Lys could dramatically increase sweetness. - - -

Relevance of location 43

Jin noted a mutation at location 43 from Arg43 to Arg43Ala abolished the sweetness (?or just downgraded 1t to regular sweetness?)

- - - -

Jin: a mutation at location 29 increased the sweetness significantly over brazzein. Jin: mutations at locations 30 or 33 removed all sweetness.

- - - - Relevance of mutations to the N-terminus

Jin: omitting pGlu1 increased the perceived sweetness of brazzein by 2:1. Jin: addition of Ala to the N-terminus or mutation of the side chain from Asp to Asn did not change the sweetness.

Assadi: replacement of pGlu1 with Ala1 increases the sweetness of the protein by a factor of 4.

- - - - Relevance of mutations to the C-terminus Assadi: the deletion of the normal C-terminal residue or the addition of residues at the C-terminal is detrimental to the sweetness of brazzein. - - - - - Concluding analysis The fabrication of what appears to be a fully functional brazzein from a synthetic gene code using a TAG stop codon supports the conclusion that only the 20 common amino acids plus pGlu are required to create brazzein. How the pGlu was created from the amino acid gene triplets during synthetic gene preparation and expression in E. coli is not clear. Assadi-Porter et al. use the codon ATG for this residue. 282 Neurons & the Nervous System

The above assertions are in general agreement. Disruption of the N – and C– termini, or possibly the bridge between them reduces the perceived sweetness significantly but does not abolish it. However some augmentations of the N –terminal actually raises the perceived sweetness by a small factor. Disruption of the Arg43 site also reduces perceived sweetness significantly but does not abolish it. Only disruption of the Arg43 is reported to abolish sweetness.

[xxx edit ] It is proposed that such hydrogen bonding as found between residues of helical and sheet forms of the protein can form equally easily between the residues of the various single strands and a distinct G-path GR. Depending on the curvature of the chain, Any of the nitrogen orbitals could represent the X element in the AH,B,X relationship. For a configuration such as the β-turn of brazzein and potentially other sweet proteins, involving less tight twisting, the distance to an appropriate nitrogen orbital is quite likely. 8.5.9.6.2 The AH,B,X relationship in sweet proteins

At least one glycophore has been identified in brazzein that is compatible with its super sweetness in the context of the hypothesis of this work. Because of the complexity of the protein, there may be more.

The introduction of the sweet proteins into the repertoire of glycophores requires modification of the previous hypothesis related to their structural geometry. For the first time, although also observed among the organic picrophores, the structural requirement does not include the presence of only two carbon atoms between the AH,B orbitals, nor does it require they maintain a nominal equat-trans relationship within a single ligand. The requirement is loosened to only require the two orbitals to be on the external surface of a complex structure and maintain an average distance between the orbitals of 2.82 Angstrom plus or minus a value to be determined, but at least ±5% at the half amplitude of the efficacy function.

This relaxation of the requirement makes it even more difficult to recognize the glycophores based on any schematic description (Fischer diagram, etc.) or even 2D ball and stick structure. The glycophores in complex molecules such as the sweet proteins can only be visualized using 3D graphic representations. See Sections 8.5.3.2 & 8.5.3.3.

Figure 8.5.9-10 shows the only glycophore of brazzein identified to date in a space-filled 3D representation to illustrate its surface location and the location of the key orbitals. Several other features are identified primarily for orientation purposes. Leu28 is to the immediate upper right of Asp29.

Note the relatively small area involved in this glycophore relative to the total volume of this “small” sweet protein. While possible to locate the glycophore in a ball and stick representation of this molecule, it is a visually demanding task and very difficult to display for pedagogical purposes. Signal Generation & Processing 8- 283

Figure 8.5.9-10 Location of the glycophore of brazzein on an annotated space-filled model. The oxygen orbitals of both Asp25 & Asp29 are assumed to be from two separate resonant carboxylic ligands in the Astex representation. The dispersion centroid can be assumed to be either the hydroxyl group of Tyr24 or the phenol ring of that same residue. See text.

The potential orbitals of the β-turn (such as in Glu41, Lys42, Glu41, Arg43 etc) are on the outside of the turn and too widely spaced to be glycophores. However, they may contribute significantly to the dipole potential of the stimulant discussed in a following section (particularly the guanidinium group in Arg43). Figure 8.5.9-11 shows the area of the glycophore described above in a schematic format. The defined glycophore is not represented by a simple ligand, or even a contiguous small group of orbitals and carbon atoms. It is actually formed by elements of three different residues operating in electrolytic consort. The two closest oxygen orbitals of Asp25 and Asp29 act as the AH,B group while the electrostatic field of either the hydroxyl group or the phenol ring of the Tyr24 residue acts as the dispersion centroid, X, exciting the sensitive area of the G-path GR. See Section xxx. The Astex viewer assumes the resonant condition for the carboxyl group when preparing its representation. If a polar representation was assumed, the d-value calculated between the two oxygen orbitals shown might be marginally different. 284 Neurons & the Nervous System

Figure 8.5.9-11 Schematic of glycophore involving Tyr24, Asp25 & Asp29. Astex represents the carboxylic acid groups of Asp25 & Asp29 as in resonance. If they are not, the dimensions of the AH,B,X triangle could change marginally. The two oxygen orbitals closest to each other are used in defining AH,B. The oxygen orbital of the hydroxyl group in Tyr24 is shown as the dispersion centroid, X.in this figure. The phenol ring of Tyr24 may be or contribute to the overall dispersion centroid. The shaded areas represent planar structures associated with the peptide bonds. See text.

It is important to note that any structural modification to the molecule that changes the AH,X or B,X dimensions of a super-sweet glycophore will reduce its role to that of a common sweet glycophore. Any structural modification leading to a change in the baseline, AH,B is likely to cause a loss in the DACB to the GR and complete extermination of the perception of sweetness. Thus a reversion of an AH,B,X form of sweetener may cause its reversion to an AH,B sweetener with a loss of perceived sweetness of as much as 2000:1 but without total extermination of perceived sweetness. Loss of the AH,B relationship leads to total abolition of the perceived sweetness associated with that glycophore. A casual reference in the literature to the total loss of perceived sweetness may only apply super-sweetness at the concentration used in the tests, and not a truly total loss of sweetness at concentrations compatible with other normal sweeteners.

Jin et al. have highlighted the variability of the super-sweetness of brazzein with respect to Asp29 and Lys30. Their assertion that substitution of the residue asparagine (Asn) for Asp29 leads to increased sweetness is compatible with replacing one oxygen of the carboxylic acid group with an amide group. An increase in sweetness after substitution of Ala for Asp29 suggests there may be more than one glycophore present in brazzein. Substitution of Lys for Asp29 introduces the possibility of destroying the initial glycophore and creating a replacement. It is quite possible that other glycophores are present in the super-sweet stimulant brazzein. However, the area around His31 (residues 29 to 33) was explored and showed little promise. Locations within the α-helix, within strand 2 of the molecule and along the β-turn are not promising on structural grounds. The required dimensions are difficult to achieve within these regions. Residues extending outward from the mean path shown in cartoons of brazzein tend to have orbitals too widely space to form the necessary geometric AH,B,X relationship. 8.5.9.6.3 Quantifying the AH,B,X dimensions in proteins.

Figure 8.5.9-12 from Caldwell et al. illustrates a problem in describing the precise locations and dimensions features within complex molecules such as proteins. Thermal forces cause a constant Signal Generation & Processing 8- 285 “dancing” within the molecule, not unlike the Brownian motion associated with individual molecules in a solution or gel. It is typical for the various 3D representation programs to present a “cartoon” representing the average of typically 43 or more conformers (conformations at individual moments in time) describing a given protein. In an accompanying “stick” version of such a cartoon, Caldwell et al. suggest a possible glycophore of brazzein may be found near the Arg33 and Arg43 residues at lower left and upper right in this figure. It is not likely these residues are involved in the glycophore DACB process but they may contribute considerably to the dipole potential of this molecule.

Figure 8.5.9-12 43 superimposed conformers of brazzein in solution. From Caldwell et al., 1998. 286 Neurons & the Nervous System

Figure 8.5.9-13 compares the available parameters for AH,B,X for various glycophores.

Figure 8.5.9-13 Expanded table of AH,B,X parameters for the glycophores of taste. Paired gustaphores are shown where more than one AH,B,X conformation are of potential interest. Many potential AH,B,X configurations are omitted here. The boxed values are the most likely GR values for the G-path. All of the A–B values shown are within the estimated ranges of Kier and Shallenberger from the 1960's. The values for sweetness relative to sucrose are very rough and usually provided in semantic rather than tabular form.

The table shows two potential GR’s for the G-path, both phosphatidylgalactose (PtdGal) . The first involves A=O3 and B=O4 as the primary DACB participant as suggested by Kier and by Shallenberger et al. in the 1960's (Section 8.5.5). Jmol suggests this configuration has a d = 2.8 Angstrom. The second involves A = O2 and B=O3 as the primary DACB participant with a d=2.86 Angstrom. Both values are bracketed by the values originally proposed by Kier and by Shallenberger et al. and both are within 5% of the value of 2.92 Angstrom exhibited by many artificial sweeteners. The first candidate GR appears more compatible with the AH,B,X configuration of most artificial sweeteners with an angle A near 120 degrees. This nominal G-path gustatory receptor (GR) is shown in Figure 8.5.5-9 Galactose and the artificial and protein sweeteners can exhibit multiple AH,B,X configurations. Only those configurations closest to the values of Kier and Shallenberger et al. are shown here for comparison. Aspartame provides two different potential AH,B.X glycophores as noted in the figure. Aspartame and brazzein using the OH group as X show a good match for the A–B and B–X distances and the angle, pX. The pB is different when X = OH in brazzein. Similarly, the A–B distance to the centroid of the phenol ring of Brazzein is the same as the A–B distances for aspartame and brazzein to the OH group. These ranges may help determine the effective diameter of the dispersion centroid(s). Signal Generation & Processing 8- 287

8.5.10 The dipole potential (DP) in the perceived intensity of gustation

Once a stimulant has become associated with the appropriate gustatory path through a DACB, with a GR, the response of the sensory neuron associated with that GR becomes important. As discussed in Section xxx, the response involves an electrolytic amplifier within the stage 1 sensory neuron that measures a change in potential at its input relative to the surrounding saliva. This change in potential is caused by the dipole potential of the stimulant molecule. The dipole potential of the stimulant molecule is described in terms of the electrostatic potential of the molecule in the vicinity of its AH,B relative to the integrated potential of the rest of the molecule in contact with the saliva (resulting from its aggregate charge distribution). Every sensory neuron contains two electrolytic amplifiers or Activa. Typically, the Activa forming the first amplifier is in electrical contact with its GR and exhibits an input range of about xxx 15 milli-Volts and an input threshold of tens of microvolts. The resulting usable dynamic signal range, in the absence of adaptation, is typically on the order of 200:1. The typical natural glycophore (of the AH,B type) introduces a dipole potential less than the available 15 mV range. However, some glycophores of the AH,B,X type can introduce a larger potential change that results in saturation of the first amplifier. This may occur by two distinct mechanisms; first, the dipole potential of the stimulant may be quite large and/or second, the stimulant may place a charged region in the vicinity of the dispersion centroid of the GR that results in a large change in the dipole potential of the GR itself. The combination of these two mechanisms can easily cause a saturation in the input Activa.

The dipole potential of a molecule has not played a major role in chemical research prior to the 1990's. However, a related dipole moment is commonly addressed and it is frequently demonstrated in undergraduate chemistry programs. Commonly, the moment arm of such a dipole moment is described relative to an arbitrary geometrical axis. The lack of a dipole moment under conditions of molecular symmetry are also commonly addressed. Here, the moment arm of the moment is not of interest, only the aggregate charge described above and associated with that moment arm is of interest. The separation of the charge by a distance equal to the moment arm can be expressed as a potential.

Beginning in the 1990's, the significance of the dipole potential of biological membranes has become of considerable interest. In the following, both the composite dipole potential of the bilayer membrane and the dipole potential of only the outer layer are critically important.

One can conceive of a gustaphore bonding to a specific GR but not introducing any net change in the dipole potential presented to the input of the 1st Activa of the sensory neuron. In this case, no signal would be generated at the output of the Activa and there would be no perception of taste. This situation appears to be common among the “anti- sweeteners” and potentially other gustatory channel blockers.

The fact that a phospholipid of the outer bilayer exhibits a measurable dipole potential implies that it exhibits a finite (although quite high) resistance value.

In the following discussions, it is critically important that the specific character of a given molecule be appreciated. As an example, the description of a molecule as a glyco-lipid is entirely inadequate. In the case of the outer layer of the external lemma of a sensory neuron, it will typically consist of a phospholipid such as phosphatidylserine (PtdSer), acting as a GR, that is susceptible to DACB linking to an organic acid. The change in the dipole potential from that of PtdSer alone and when stimulated by the organic acid is of critical importance in perceiveing the intensity of the stimulant. 8.5.10.1 Background

The study of the dipole potential of biological membranes experienced a renaissance in the mid 1990's. Significant changes occurred in the area of computational chemistry which allowed for much more clarity in interpreting the results of Langmuir Trough Experiments. 288 Neurons & the Nervous System

Brockman provided an excellent introduction appropriate to the time period348. His Abstract highlights many of the conflicting concepts proposed in that era. In discussing some of these, he does review the status of phosphatidylcholine (PtdCho), phosphatidylinositol (PtdIns) and other phospholipid of significance in gustation. In a section labeled Perspective, he notes, “Measurements of ΔV in monolayers have been made for over 60 years and in bilayer membranes for over 20 years. However, the use of these measurements for probing the regulation of biological phenomena has been minimal.” Maggio gave an even more extensive introduction relevant to gustation at nearly the same time349. He opened with the observation concerning his subject, glycosphingolipids (GSL), “A hallmark of the structure of GSLs is the extraordinary variety given by the different number and type of carbohydrate residues that, joined by glycosidic linkages, are linked to the hydroxyl group in carbon 1' of the ceramide moiety. While providing some valuable background, much of the terminology used in these papers has been replaced in the interim. There is considerable graphical data but a shortage of detailed schematics and relationships. Much of this early work involved less than ideally formed bilayer membranes. Transport of heavy ions through less than perfect artificial biological membranes provides virtually useless information.

Data on the dipole potential of fully elaborated gustaphores in solution is quite rare if not non- existent. No concerted effort has been found to record this parameter as it relates to gustation. Such measurements are commonly made on other organic molecules, particularly due to the recent focus on liquid crystal display chemistry and the electrolytic character of the biological membrane.

It may become necessary to adopt a more explicit name for the dipole potential associated with a stimulant when in a conductive environment such as saliva. While the dipole potential of the phospholipids of the GR sensed by the first Activa are constrained by their liquid crystalline structure, that of the stimulant is not. The stimulant is nearly surrounded by a conductive fluid. The dipole potential of interest here is that scalar potential of the stimulant presented at the AH,B interface with the GR with respect to the saliva surround.

Beitinger et al. have reported the dipole characteristics of a variety of gangliosides, some of which may also be applicable to the globosides350. Their measurements were made using an upper air interface and a lower water interface by floating their material on a water pool. They note,

“Gangliosides are characteristic glycosphingolipids containing different numbers of negatively charged sialic acids. These molecules being particularly: abundant in nerve cell membranes of vertebrates are components of the outer bilayer leaflet and might be intimately involved in various cellular biological events.” “More than 2000 articles on gangliosides were published during the past decade, a lot of them with special regard to the large hydrophilic sugar moiety and its great potential for hydrogen-bonding. These reports give evidence that the polar headgroup determines the physicochemical properties of these molecules. These reports give evidence that the polar headgroup determines the physicochemical properties of these molecules.

348Brockman, H. (1994) Dipole potential of lipid membranes Chem Phys Lipids vol 73, pp 57-79

349Maggio, B. (1994) The Surface Behavior of Glycosphingolipids in Biomembranes: A New Frontier of Molecular Ecology Prog Biophys Molec Biol vol. 62, pp. 55-117

350Beitinger, H. Vogel, V. Mobius, D. & Rahmann, H. (1989) Surface potentials and electric dipole moments of ganglioside and phospholipid monolayers: contribution of the polar headgroup at the water/lipid interface Biochim Biophys Acta, vol 984, pp 293-300 Signal Generation & Processing 8- 289 Several studies (t3C-NMR, ESR, X-ray diffraction) on the orientation of the sugar headgroup indicate that the polar headgroup is relatively rigid, and in mixed phospholipid/ganglioside bilayers fully extended and approximately perpendicular to the interface.”

“The formation of a monolayer can be measured as a change in surface pressure and surface potential. The surface potential is proportional to the change of the normal component of the dipole density with regard to the pure water surface. Numerous reports exist of measurements with respect to the total surface potential and the overall dipole moment, respectively, of simple giycosphingolipids and gangliosides.” “The phospholipids used for the experiments have been obtained with a degree of purity higher than 99%. The probes showed in each case a single spot in HPTLC and were used without further purification.” “In summary, we have given the sign and the magnitude of the potential drop across the Water/lipid interface for different ganglioside, phospholipid and sulfatide monolayers. The apparent dipole moment per ganglioside headgroup differs from other lipids. The most remarkable result is that among gangliosides the size and the number of charges do not cause large changes in the potential drop. Our potential data imply that the main purpose of nature to vary in biomembranes the number of charges per ganglioside headgroup, e.g., from one to three, is not to change the apparent potential drop across the headgroup region.”

“The headgroup potential of the pure ganglioside monolayers reaches minus several hundred millivolts which can influence the electrostatic properties of neuronal bilayer surfaces strongly.”

While the Beitinger et al. data is not directly applicable to gustation, it provides a very good scenario of the applicability of their techniques to such research. In some cases, their surface potential may correspond to the relevant dipole potential of the AH,B configuration gustaphores.

Their demonstration that many of the fatty acid side chains of the phospholipids of interest are conductive, and in fact exhibit a dipole potential is of critical importance. Repeating some of their experiments with dioleoylphosphatidylcholine (DOPC), dipalmitoylphosphatidylethanolamine (DPPE), dipalmitoylphosphatidylserine (DPPS), dioleoylphosphatidic acid (DOPA), but using the phospholipids defined in this work should be very rewarding. Their values for DPPS may be directly applicable to the PtdSer (the GR of the A-Path) of this work.

Figure 8.5.10-1 shows a portion of their Table II for a pH of 7.4 and 20 ± 0.5 C. Note the sharp division between the surface potential values associated with their gangliosides and the phospholipids alone (dashed line). The data at pH 7.4 is particularly relevant to the gustatory modality. What they describe as gangliosides appear to be a combination of a phospholipid and a sugar as opposed to their simple phospholipids (DPPE and Figure 8.5.10-1 Abridged Table II of DPPS). GM1; monosialic ganglioside. GD1a and GD1b; Beitinger et al., surface potentials of a disialic gangliosides. GT1b; trisialic ganglioside. GMix; variety of relevant chemicals. See text for a mixture of sialic gangliosides. C18; stearic acid. abbreviations. See their extensive caption for more information. From Beitinger et al., Note the dipole potential order of the gangliosides 1989. at their concentrations; GD1b>GM1>GD1a>GT1b. 290 Neurons & the Nervous System

The values given for the phospholipids are for single layers. When arranged in a bilayer, the potentials oppose each other resulting in a significantly lower net value, i.e., a DPPE/DPPS bilayer exhibits a net dipole potential on the order of 270 mV at a pressure of 30 milliNewtons/meter (mN/m). A bilayer of exclusively DPPE gives a net dipole potential nominally equal to zero. The value for C18, an 18 carbon atom fatty acid is included to illustrate how much of the dipole potential of a single phospholipid can be attributed to the tail group as opposed to the head group. Assuming the DPPE and DPPS employ the same tail group, the head group of the phospholipid is the controlling factor in the net dipole potential of the monolayer. 8.5.10.1.1 Background related to the phospholipid structure of the GR

The Beitinger et al paper has been widely cited and spawned considerable research of interest here. Zheng & Vanderkooi provided an early introduction to the study of lipid bilayers in 1992351. They open with a particularly important comment, “It has been known for several years that the passive permeability of lipid bilayers is considerably greater for anions than for cations.” This expresses a conventional chemical viewpoint and is based on the conventional wisdom drawn from experiments with tetraphenylborate in two forms. They do note, “The bilayer translocation rate and partition coefficient are both several orders of magnitude larger for the anion tetraphenylboron (TPB-) than they are for the structurally similar tetraphenylphosphonium (TPP+) cation.” Importantly, Zheng & Vanderkooi did not cite the theoretical and experimental work of Anderson & Fuchw of 1975 related to the use of tetraphenylborate in their experiments and discussed below. It should be made clearer that the transport of positive ions and negative ions through a uniform bilayer membrane is virtually impossible while the transport of electrons alone through such a membrane is a quite straight forward electrolytic process. The difference in the rate of transmission of electrons in the two directions is due to the underlying mechanisms of transport, leading to the well established concept of electrons and “holes.”

They have described their method of calculating the dipole potential of a bilayer of either PtdCho or PtdEtn. They have provided a skeletal model of the molecules they are talking about and the important statement, “The P-N vector in the polar head group does not lie parallel to the bilayer plane as has often been claimed, but makes an angle of 150 with the bilayer plane in DLPE (PtdEtn), and 90 and 250, respectively, in the two independent DMPC (PtdCho) molecules. This gives the result that the z component of the head group dipole is nonzero with the negative end of the dipole pointing toward the bilayer, thereby causing the head group contribution to the internal potential to be negative.” Their model appears to omit the water molecules found between the two bilayers but does include the water on the outer surfaces of their bilayers. See Section xxx. Their figures 2 & 3 and table 1 describes their calculated potential fields within the bilayer and these are similar to the measured values [xxx be specific and cite sources ]. Their figures 1, 2 & 3 have been combined in Figure 8.5.10-2. The left frame illustrates their interpretation of their 1,2-dimyristoyl-sn-glycero-3-phosphocholine (or PtdCho as abbreviated in this work). “The P-N vector in the polar head group does not lie parallel to the bilayer plane as has often been claimed, but makes an angle of 15 degrees with the bilayer plane in DLPE (τ in the Fig.), and 9 degrees and 25 degrees, respectively, in the two independent DMPC molecules. This gives the result that the z component of the head group dipole is nonzero with the negative end of the dipole pointing toward the bilayer, thereby causing the head group contribution to the internal potential to be negative. The acyl esters make a positive contribution to the internal potential, but the magnitude of this contribution is only one quarter to one third that of the polar head group, as can be seen from Table 1.”

351Zheng, C. & Vanderkooi, G. (1992) Molecular origin of the internal dipole potential in lipid bilayers: calculation of the electrostatic potential Biophys. J Biophys Soc vol 63, pp 935-941 Signal Generation & Processing 8- 291

Figure 8.5.10-2 Lipid and bilayer chains from Zheng & Vanderkooi. Left; a nominally standard representation of a phospholipid (they labeled DMPC) showing two different glyceride chains and two distinct angles of interest. Center; their representation of a bilayer of DMPC:2H2O and the electrostatic potentials associated with that bilayer. Right; their representation of DLPE:HAc and the electrostatic potential associated with that bilayer. Note the variation in scales. See text. From Zheng & Vanderkooi, 1992.

The center frame shows a bilayer of their DMPC hydrated on its external surfaces but not showing the water normally found at the interface between the two DMPC layers in biological specimens (Section xxx). Above the schematic are the potential plots for a unit charge brought to a position adjacent to or interior to the bilayer. The left scales provide energy values typically used by physicists and the right scales provide voltage potentials of more immediate interest here. The right frame shows similar information for 1,2-dilauroyl-DL-phosphatidylethanolamine:acetic acid (DLPE:HAc). Again, it does not show the water expected to be found at the interface between the two bilayers. Note the different fatty acids used in their DLPE. The figure at lower right is considerably more complex than that at lower center, suggesting a more complex mixture of fatty acids associated with a less pure sample of DLPE. Zheng & Vanderkooi address this complication in their methods and results sections. Figure 8.5.10-3 reproduces their Table 1 for purposes of later comparison. While their calculated values are not precisely applicable to the biological situation, they do describe relative values that may be useful in the discussions in the following sections. The calculated values for just the phospholipid groups are nearly an order of magnitude lower than that of other theoretical and laboratory investigators. Their conclusions must be interpreted in the context of other experimental results. 292 Neurons & the Nervous System

Majewski et al. have provided significant information on the dipole potentials of a single layer of a bilayer membrane based largely on X-ray experiments352. Their figures 1 & 6 are reproduced here as Figure 8.5.10- 4. Note the angle of the head group of the PtdEtn in agreement with the comments of Sheng & Vanderkooi. Their abstract notes, at the concentrations and conditions specified, “the two components do not phase-separate and no evidence for domain formation was observed. X-ray Figure 8.5.10-3 Atypical calculated components scattering measurements reveal that GM1 is of the electrostatic potential at the bilayer accommodated within the host DPPE midplane. See text. From Zheng & Vanderkooi, monolayer and does not distort the 1992. hexagonal in-plane unit cell or out-of-plane two-dimensional (2-D) packing compared with a pure DPPE monolayer. The oligosaccharide headgroups were found to extend normally from the monolayer surface, and the incorporation of these glycolipids into DPPE monolayers did not affect hydrocarbon tail packing (fluidization or condensation of the hydrocarbon region). This is in contrast to previous investigations of lipopolymer-lipid mixtures, where the packing structure of phospholipid monolayers was greatly altered by the inclusion of lipids bearing hydrophilic polymer groups.”

Note carefully; the group labeled GM1 is distinctly different than the group labeled DPPE. C The upper fatty acid of GM1 terminates with an NH ligand while the upper fatty acid of DPPE terminates with an Oxygen atom. C The lower fatty acids are significantly different in the two cases. C The ganglioside of GM1 has replaced a very critical ligand incorporating the PO3 group and the NH3 group, both highly polar in character. C In their lower frame, a molecular schematic is shown containing a five-sided ring that is not present in their upper schematics or discussed in their text. [xxx check ] C The schematic in the lower frame shows a mixture of two diglyceride materials in a ratio of 4:2. As a result, what they describe as a ganglioside is actually a glycan/diglyceride, a member of the cerebroside family and not generally derived from a phospholipid, whereas their DPPE is a phospholipid containing different glycerides. Their paper includes much other data. However, when discussing gustation, these differences make comparing their data with other data in the literature difficult.

352Majewski, J. Kuhl, T. Kjaer, K. & Smith, G. (2001) Packing of Ganglioside-Phospholipid Monolayers: An X-Ray Diffraction and Reflectivity Study Biophys J vol 81, pp 2707–2715 Signal Generation & Processing 8- 293

Figure 8.5.10-4 Chemical structure of DPPE & GM1 (a specific PtdEtn & a “ganglioside”). Top; the chemical structure of the two materials. Bottom, the schematic of the mixed materials “DPPE” and “GM1.” in a ratio of 4:2 at an air-water interface. The “gangliosides” incorporate a five-sided ring not present in the upper schematics. See text. From Majewski et al., 2001.

Peterson et al provided information on the dipole potential of a variety of phospholipids incorporating various fatty acid chains, including their test configuration and specific information 294 Neurons & the Nervous System regarding substitutions of sulfur along the fatty acid chains of DPPC353. Their focus was on the electrolytic properties of a single monolayer as they might affect protein binding to the material. Their actual results are of little apparent value here. Figure 8.5.10-5 reproduces their Table 1. “The absolute dipole potential of unlabeled and sulfur-containing DPPC membranes, φd, was measured by the CR method (Pickar and Benz, 1978). The values recorded for pure lipids (DPPC, DHPC, GMO; Table 1) were found to be in good agreement with values reported earlier (Pickar and Benz, 1978; Gawrisch et al., 1992). φd of DPhPC was estimated to be (228± 5) mV.” The top two entries in this table indicate the range of the dipole potential achieved by merely varying the character of the fatty acids forming the diglyceride tail. Both of these entries would be labeled phosphatidylcholine (PtdCho) in most academic journal papers.

Shamberger & Clarke provided additional information concerning the dipole potential of a single phospholipid layer and its use in a bilayer similar to the biological equivalent354. They addressed several questions and inconsistencies regarding previous theory of,and laboratory values for dipole potentials related to a membrane. They conclude with a very important statement, “The electric field produced by the dipole potential in the membrane interface is extremely 8 9 Figure 8.5.10-5 The dipole potential of different lipids large, i.e., 10 – 10 V/m (Brockman measured with the CR method See text. From Peterson et 1994; Cafiso, 1995; Clarke, 2001), al., 2002. which is significantly larger than that caused by a typical total membrane potential (e.g., a 100-mV membrane potential produces an electric field stength of ~2.5 x 107 V/m).” Restated, the internal dipole potentials related to a biological bilayer are far larger than the net dipole potential of the package.

Their conclusions in 2002 are important. After discussing the use of tetraphenylarsonium (TPA ) and tetraphenylborate (TPB) as potentially positive and negative ions capable of penetrating biological membranes, they noted “The dipole potential values determined in this way are generally 100 to 200 mV higher (Hladky and Haydon, 1973; Beitinger et al., 1989; Smaby and Brockman, 1990) than those previously determined using hydrophobic ions on lipid bilayers. For dioleoylphosphatidylcholine (DOPC), for example, Beitinger et al. (1989) have determined values of 420 and 431 mV at pH 7.4 using two different buffer systems. For egg yolk lecithin (predominant component DOPC) Hladky and Haydon (1973) determined a value of 441 mV. The conductance measurements of Pickar and Benz (1978) using hydrophobic ion yielded, on the other hand, a value of 224 mV for DOPC (Table 3). This discrepancy between bilayer and monolayer values of the dipole potential has been known for many years, but as yet no generally accepted explanation for it has been found.” They conclude relatively optomistically, “Although the absolute value of the dipole potential can still not be precisely defined, due to the uncertainty in the calculated values of the hydration energies of the hydrophobic ions, the calculations carried out here demonstrate that relatively small differences in the hydration energies of TPB , TPP , and TPA can easily account for the

353Peterson, U. Mannock, D. Lewis, R. Pohl, P. McElhaney, R. & Pohl, E. (2002) Origin of membrane dipole potential: Contribution of the phospholipid fatty acid chains Chem Phys Lipids vol 117, pp 19–27

354Schamberger, J. & Clarke, R. (2002) Hydrophobic Ion Hydration and the Magnitude of the Dipole Potential Biophys J vol 82, pp 3081–3088 Signal Generation & Processing 8- 295 differences between dipole potential values previously reported from monolayer and bilayer measurements.” Their discussion did not demonstrate the validity of this statement. TPA amd TPB are very large ions, typically 10 Angstrom diameter based on Jmol 3D representations as opposed to the calculated sizes they provided in their Table 1. They did attempt to rationalize the 25% difference in their values calculated using their volume data versus their area data. No physical evidence of pores on the order of 10 Angstrom have been reported in the biological literature. In the absence of identifiable pores, the likelihood of their penetration of a natural, or properly prepared analog of a, bilayer membrane is extremely unlikely. Anderson & Fuchs have performed a much more comprehensive study of tetraphenylborate in lipid bilayers355. Their experiments focus on the transient “transport” of charges through lipid bilayers. They did not demonstrate any net flow of these materials all of the way through the membrane. In 2005, Elana Pohl appears to have repeated much of the material in the 2002 Peterson et al. paper as part of a review356. The paper appeared in an obscure publication that is difficult to locate. The Google Scholar leads to a publisher’s expanded Abstract that includes a figure 3 requiring an unidentified pore to explain any transport of potassium through the membrane. The above summaries show the complexity and the exploratory character of recent work in this area. No substantial results have been shown indicating that physical ions, particularly of the heavy alkali or alkali metal atoms, are able to pass through the lemma of a sensory neuron as part of the gustatory process. 8.5.10.1.2 Background related to the DP of a stimulant in saliva

With the recent appearance of high quality 3D molecule modeling, many of the better programs are now capable of presenting electrostatic fields created by a stimulant. These fields are generally limited to the free-space condition and are not necessarily indicative of the molecule when present in solution. While saliva and water may not normally be considered good electrical conductors, they are at the impedance level associated with the neural system in general, and the gustatory modality in particular.

Thus, the dipole potential for a stimulant is that potential presented at the AH,B interface with the appropriate GR when surrounded by saliva acting as the local electrical ground environment. A conceptually similar interface is shown in figure 2a of Venanzi & Venanzi (1992) for amiloride.

The electrostatic fields (of given strength) predicted by these computational chemistry programs generally look like balloons slightly larger than but generally following the contours of the atoms of the stimulant molecule as in Figure 8.5.10-6 for aspartame. With distance (and resulting lower field strength), the balloons become spherical. While aspartame is a man-made sweetener, it is not considered a super-sweetener. The dimensions in this figure are A–B = 2.91 Angstrom, B–X = 5.359 Angstrom, A–X = 4.426 Angstrom and angle A = ~90 degrees. It should be noted that the charge pattern is not defined by points but by loci. However, the stimulant will not form the required DACB relationship unless its actual AH,B spacing is within the tolerance required for this parameter by the GR. The tolerance on the locus of the dispersion centroid, C, has not been determined but in the case of aspartame, its C is probably outside of the acceptable locus for the sweet GR.

355Andersen, O. & Fuchs, M. (1975) Potential Energy Barriers to Ion Transport Within Lipid Bilayers Studies with Tetraphenylborate Biophys J vol 15, pp 795-830

356Pohl, E. (2005) Dipole Potential of Bilayer Membranes Advances Planar Lipid Bilayers Liposomes vol 1, pp 77–100 296 Neurons & the Nervous System Signal Generation & Processing 8- 297 298 Neurons & the Nervous System

Figure 8.5.10-6 Electrostatic field of aspartame showing the AH,B,X relationship in this man-made sweetener. Free space representation. See text for dimensions. The charge cloud associated with the hexane ring is shown in a light pink in the lower frame. Drawn using DS3.5 & aspartame from http://kaist.ac.uk . Signal Generation & Processing 8- 299

In the presence of a conducting surround, the electrostatic field external to the molecule is forced to conform to the geometrical shape of the solvent/molecule interface except where the stimulant molecule is in close proximity to a relevant GR or other non-solvent structure. 8.5.10.1.3 Background related to the dispersion centroid relative to AH,B,X ADD

Van der Heijden and his team recognized early in their work that the location of the dispersion centroid, X in the AH,B,X relationship between a stimulant and its appropriate GR was not likely to be a distinct point357. “To describe stereochemical requirements more precisely, new conceptual parameters were introduced, namely α, δ and ω (minimum, optimum and maximum distances between these third binding sites and the atoms A, H and B of the AH-B moieties respectively, especially appropriate for homologous series) and the S value (shortest distance between the position of an atom and the plane formed by the atoms A, H and B of the AH-B moiety).

On page 58 of the paper (1985a), the van der Heijden team had stressed that their set of sweeteners was not :pure;

“The literature on the five series of sweeteners (nitroanilines, sulphamates, oximes, isocoumarins and dipeptides) was searched for sweetness values and for details as to whether these substances are denoted as tasteless or bitter, because in the latter situation their possible sweetness potency may be masked by the bitter taste.”

Many of these chemical classes also exhibit other types of gustaphores.

357Van der Heijden, A. van der Wei, H. & Peer, H. (1985a) Structure-activity relationships in sweeteners. I. Nitroanilines, sulphamates, oximes, isocoumarins and dipeptides Chem Senses vol 10(1), pp.57-72 300 Neurons & the Nervous System

[xxx following para need editing for continuity and overlap ] Van der Heijden et al, 1985a, provided a 3D histogram for a group of sweeteners that was further perfected in the 1985b paper and reproduced there and in the 1993 paper. It was supported by considerable calculations based on their exploratory investigations. Figure 8.5.10-7 shows possible alternate interpretations of their 1993 histogram. Focusing only on the glycophores; the principle axis of interest is drawn through the points A and B, and the median of the distance between point A and point B is taken as 0.282 nm (2.82 Angstrom). Using the old axes for simplicity, instead of performing a rotation of the data set (Section 8.5.2.3 xxx on MDS), it appears there is an association between the data points and a plane near 45 degrees to the original axes. It is proposed the angle should be defined relative to the plane of the DACB relationship between the two pairs of orbitals shown. The long alternate axis and the four orbitals remain in the x-y plane of the original axes (but are rotated by about 23 degrees). This interpretation is similar to that illustrated in [Figure 8.5.5-8 of Section 8.5.5.1]. That figure suggests the four orbitals in the DACB relationship are generally perpendicular to the plane of the glycophore and the GR. The distance and angle to the point X then suggests the loci of the most sensitive area of the GR to dipole potential modification.

The specific stimulants included in this data set are not material here but are detailed in the 1985a and 1993 papers. In general, they are “1, sulphamates (site 1); 2, isocoumarins (site 2); 3, oximes + nitroanilines (site 3); 4, dipeptides (site 3).

As noted by the van der Heijden team, the location of the dispersion point is best described by a loci centered on the locations in their 3D histograms. They did not indicate that their specified points were statistically precise with small standard deviations.

The 1985b paper extended their investigation358; “The previously introduced conceptual parameters (α, δ, ω and S) to describe the stereochemical requirements for organic compounds to taste sweet, were now applied to another series of sweeteners and to some well-known potent substances.” They also provided a selection of “recalculated” dispersion point locations Figure 8.5.10-7 3D histogram of centroid of X for a group of stimulants. Note potential alternate axes for gustaphores in general ( page 79). Their o abstract closes with the interesting assertion, and nominal 45 angle of the plane containing a “It is remarkable that the average δ positions majority of the points (loci) to the notional plane of the DACB relationship. The distance between belonging to sweeteners with similar AH-B A and B’ in this relationship is 0.27 nm. All moieties are located very close to each distances in nm. See text. Modified from van der other.” Heijden et al., 1993 In 1987, the Van der Heijden team provided a major review of their earlier work359. Their studies are now dated but are clearly suggestive of and compatible with the concept of a locus rather than a point describing the sensitivity of a GR to a charge associated with a gustaphore. They did not clearly differentiate between stimulants

358Van der Heijden, A. van der Wei, H. & Peer, H. (1985b) Structure-activity relationships in sweeteners. II. Saccharins, acesulfames, chlorosugars, tryptophans and ureas Chem Senses vol 10(1), pp 73-88

359Van der Heijden, H. van der Heijden, A. Peer, H. (1987) Sweeteners Food Rev Internat vol 3(3), pp 193-268 Signal Generation & Processing 8- 301 to the four distinct GR’s and resultant gustatory paths. They also defined a plane determined by the three atoms A, H & B and provided a distance S for their dispersion points from that plane. Their concept requires more study as to whether their concept applied to a stimulant when in an AH,B,X relationship with a GR involving a DACB with the appropriate GR or whether it applied to only the stimulant. It is not clear there is any relevance of the location of the H in determining a relevant plane, and even where this atom is at a given instant. It may require further analysis to confirm or deny the following points; • The pair of orbitals labeled A & B probably relate to the glycophore but might relate to the GR. • The loci described in the above discussion may relate to an electron-rich or electrophobic area, such as a dual bond carbon or a ring structure, rather than a single atom. • The loci described for the glycophore may not correspond to the location of the feature of the corresponding GR. The loci may describe a location that is pushing the charge pattern of the GR from a location near its side. 8.5.10.2 Results of the net change of a DP due to a AH,B type stimulant

[ Section 8.5.1 presented both a physical and electrolytic schematic of the first amplifier (Activa) within a stage 1 sensory neuron dedicated to gustation. Section 8.5.5.1 included a more detailed circuit schematic highlighting an “capacitative” relationship between the gustaphore and the GR. Figure 8.5.10-8 expands on the electrolytic schematic to show the proposed stimulation by a AH,B type stimulant, applicable to all four of the gustatory paths. The right-most molecular group of the type 4 outer lemma and the AH,B type stimulant have been expanded to allow a more detailed discussion.

First, note the net potential between the saliva and the hydronium channel between the bilayers is quite small (a nominal 24 mV). This small potential is due primarily to the opposing potentials of the two phospholipids forming the outer layer of the bilayer lemma. The precise potential requires detailed knowledge of these two phospholipids and the characteristics of the receptor ligand (or gustatory receptor, GR). As noted by xxx, merely describing these molecules as phospholipids is inadequate. The precise character of the glycerides in their tail structure must be known if the precise dipole of this structure is to be specified.

The net dipole potential is also dependent on any hydrated water forming a boundary layer on the active surface of the GR.

It is suggested here without justification that one of the purposes of the saliva is to act as a wetting agent for food brought into the oral cavity. In this role, it would also act as a wetting agent to the tissue of the oral cavity and effectively eliminate any hydrated boundary layer in the position indicated by the dotted box at upper right. If the above suggestion is accepted, this box and any contribution to the net dipole potential due to water associated with the boundary layer adjacent to the GR surface can be eliminated from further discussion.

Note the phospholipid facing the saliva is a phospholipid esterified to the GR. It is not a ganglioside as investigated by Beitinger et al. and some later investigators. The presence of the phosphate moiety probably plays a major role in determining the dipole potential of the outer layer and insuring the inner layer and outer layer together exhibit a small net dipole potential. If the outer phospholipid is a mirror image of the inner phospholipid, and there is no boundary layer at the active surface of the GR, the small net potential of 24 mV as shown could be due entirely to the dipole potential of the GR when esterified to the phospholipid. In the case of the A-Path, this would be the dipole potential of serine (Ser) when esterified to Ptd.

Assuming the quiescent potential of –24 mV is due primarily to the dipole potential of the GR, it becomes clear that upon forming a DACB with a stimulant as indicated by the O–H- - O bonds at upper right, the net dipole potential of the pair would be reflected in a change in the quiescent potential at the base of the Activa. If the net change was to a more positive potential, 302 Neurons & the Nervous System the current through the Activa would be increased as indicated in the input characteristic of the Activa shown at lower left. This would result in an increased current through the type 2 lemma forming the Activa as represented in the lower right output characteristic.

Figure 8.5.10-8 Detailed electrolytic sensory neuron operation in AH.B situation. The last phospholipid/GR/stimulant on the rigth in the type 4 region has been expanded to show the details required. The characteristics at lower left show the operating performance of the Activa shown at upper left. See Text.

This operating methodology is easily defended using the Electrolytic Theory of the Neuron and is far more parsimonious than any equivalent chemical theory of gustation. No unknown, or un- demonstrable chemical reactions are required. In addition as noted above, the temporal response of the excitation/de-excitation process from a kinetics perspective is limited to the formation and breaking of a coordinate chemistry bond. Such bonding and unbonding are quantum-mechanical events requiring a higher level of differential equation, than a first order kinetic equation, to describe their temporal characteristics. Note that a very highly polar stimulant at high concentration can cause a saturation within the affected gustatory pathways. The subject will not be able to perceive the quality of such a stimulant compared to similar stimulants until its concentration is reduced and the signal pathway returns to its normal operating regime. Signal Generation & Processing 8- 303 To appreciate the perceived difference between a family of similar stimulants, the use of a glycan similar to that shown in [Figure 8.5.10-4] that does not hydrolyze when introduced into the saliva is particularly attractive. A glycan can exhibit only a single glycophore while simultaneously exhibiting a dipole potential determined by the number of sugar residues present in its overall structure and their arrangement. As Majewski et al. note, the sialic acid sugars are highly polar. 8.5.10.3 Details of proposed net change in DP by AH,B,X stimulants

It has long been known that both complex glycophores and complex picrophores represent super gustaphores. In particular, the artificial sweeteners represent glycophores that are extremely effective in stimulating the G-path GR’s. It has been established by Kier and by Shallenberger et al. that these sweeteners exhibit an electrostatic potential at a location about 5 Angstrom away from the AH,B group and at an angle to the perpendicular line bisecting the distance between the two orbitals. It has been proposed that this electrostatic potential can cause a change in the electrostatic potential within the GR when brought into close alignment to the GR as a result of DACB. As noted above, the challenge is to locate the position of the “charge” near the surface of the glycophore and how it impacts the GR. It appears the super sweeteners can utilize an electrostatic charge that can vary witin a range not yet determined. However, the range encountered in the known super-sweeteners suggests the location of the “dispersion point” centroid is within the GR and not more remote as in the phosphate group of the phospholipid esterified to the GR.

[xxx combine next two paragraphs ] Noting the gross arrangement of the gustaphore and phospholipid in [Figure 8.5.5-2] and the potential impact of the gustaphore on the electrostatic potential of the phospholipid in [Figure 8.5.10-2]. the tail-to-tail arrangement of the two phospholipids of a bilayer lemma leads to a bucking circuit where the two phospholipids oppose each other. . Each lipid may exhibit a potential on the order of 250 mV but the net potential across them applied to the base of the 1st Activa in the absence of stimulation may be on the order of 24 mV. As noted above, stimulants typically cause a positive change in the net potential at the base of the 1st Activa. If the stimulant should also cause a significant positive going change in the dipole potential of the outer bilayer, the total net change at the base of the 1st Activa would be considerably enhanced.

The conclusion can be drawn that the electrostatic potential of the super-sweet glycophore can cause a significant change in the dipole potential of the GR alone and the resultant net change in the dipole potential presented to the base of the Activa is a combination of the change in dipole potential represented by the glycophore alone and the change in dipole potential of the GR caused by a change in the charge density near the dispersion centroid. This change is illustrated conceptually in Figure 8.5.10-9. The labels on the left apply equally to the features on the right. The change in the potential applied to the Activa base in the AH,B configuration is due entirely to the change in dipole potential (a vectorial sum) due to the coupling of the GR to the potential of the stimulant, VGS1. In the AH,B,X configuration, the change in the net potential can involve a change in the dipole potential of the GR, ΔVGR and the dipole potential of the stimulant, VGS2 added vectorially. This change can be so large that the concentration of the stimulant must be reduced to maintain operation of the gustatory modality in the linear range. 304 Neurons & the Nervous System

- - - - -

[xxx add introductory sentence ] Sugars have many polar hydroxyl groups (–OH) and are overall highly polar. One of the –OH groups is usually found in a –CH2OH group. Calculation of the overall dipole potential of the sugars is complicated. If a significantly polar group of another molecule is brought close to one of the hydroxyl groups, the electrical charge distribution near the hydroxyl, and therefore the dipole potential of the sugar, can be changed significantly. This appears to be phenomenon brought into play by the “super sweeteners.” The optimum sugar moiety for use in the receptor will have its various hydroxyl groups in optimum location relative to the AH,B group. Determining this degree of optimization requires careful Figure 8.5.10-9 The schematic difference between examination of a 3-D representation of the the AH,B and AH,B,X configurations during DACB sugar. If located appropriately, the moiety binding. Left; AH,B stimulation as illustrated in the can be optimally esterified with previous figure. Right’ the enhanced stimulation phosphatidic acid to provide the desired associated with the AH,B,X configuration. The optimum receptor. [xxx cite or consolidate proximity of the locus of charge concentration of with the figure “Concept: a stacked the gustaphore to the dispersion centroid of the situation. . . “ GR is a critical parameter. The most likely sweetness receptor appears to be phosphatidyl galactose based on its known presence in many situations involving lemma. Figure 8.5.10-10 shows galactose projected on a plane including the axis between the presumed AH,B participants, O-3 and O-4. It shows two triangles that are not in this plane. There dimensions are therefore distorted in the figure. [xxx need to give distances for these two triangles. ] To be effective in the AH,B,X relationship, the three points of each triangle must be accessible by a potential stimulant (no significant other structure interfering in the plane of the triangle). It appears galactose can accommodate two different AH,B,X relationships within the present tolerances on the required dimensions of the triangles. Both the ring oxygen and O-2 are susceptible to an applied electronic force from the stimulant, thereby changing the dipole potential of the overall phosphatidyl galactose molecule. This makes galactose a preferred candidate for the super sweet (as well as the sweet) receptor. Sucralose, a super sweetener (600:1 relative to glucose), is a disaccharide wherein the Figure 8.5.10-10 Galactose in the plane of O-3 & most likely glycophore involves the O–2 and O-4. Two potential AH,B,X triangles are shown. O- 1 is the normal site for esterification to the phosphatidic acid. See text. Signal Generation & Processing 8- 305 O–3 atoms since O–4 has been replaced by chlorine. Van der Heijden addressed the question of the AH,B,X configuration in considerable detail in 1985360,361. He continued to use the line between A and H as his baseline rather than the A,B baseline preferred here. (see figure 8.5.3-3) His figure 2 in the first paper may be important. He shows the location of the field of interest at distance X may lie at a distinct distance from any specific atom, suggesting it may be a specific location in the molecular electrostatic potential (MEP) that is critically important. The papers are discussed in more detail in Section 8.6.10. In using the A,H baseline, Van der Heijden addressed several cases where there was no second, B, orbital present. This introduces an entirely different framework from that of Shellenberger and colleagues and the extension of their concept pursued here.

8.5.10.4 Details of proposed of sweet protein DP measurements

The sweet proteins are known to be highly charged and polar. In the case of brazzein, it is likely that many of the substitutions that have been performed in the laboratory have cause significant changes in the dipole potential of these stimulants; where the dipole potential is the electrolytic potential introduced at the GR caused by the potential difference between the region of the glycophore orbitals and the rest of the stimulant relative to the solvent. [xxx cover adjustments due to location and specific residue selection along remainder of chain. [xxx review Assadi papers on charge ] [xxx address glu36 here 8.5.11 Correlating genes to gustatory receptors

Dahanukar et al. have made some important observations concerning the gustatory genes of Drosophila in an extensive paper362. They have associated Gr5a with the sensing of a disaccharide sugar, trehalose. Such disaccharides are typically broken down before sensing in gustation. They also noted, “Gr5a-labeled neurons are responsive not only to trehalose, but to sucrose and other sugars.”

Subsequently, they find that, “Gr5a is required for detection of a small subset of sugars including trehalose. We generate deletion mutants lacking Gr64a and find that it is required for response to a complementary subset of sugars. Strikingly, flies lacking both Gr5a and Gr64a do not show electrophysiological or behavioral responses to any tested sugar. These results demonstrate that the sugars divide into two classes that are dependent either on Gr5a or on Gr64a for their responses.” They go on, “Classic physiological and biochemical studies led to the proposal of a ‘‘fructose’’ site in sugar-sensing neurons. Our studies provide a molecular and genetic identity to this site: fructose response is completely abolished by loss of Gr64a and is completely restored by the addition of a Gr64a transgene.” Whether humans involve two separate sweetness receptors is not yet clear. Either or both of these genes can be considered directly associated with the proposed human sweetness receptor, proposed here to be PtdGal, phosphatidyl galactose.

They have associated gene Gr66a with the bitter sensory channel. This gene can be directly associated with the proposed bitter receptor, Ptd3'Og, phosphatidyl 3'-O-aminoacyl glycerol.

360Van der Heijden, A. van der Wei, H. & Peer, H. (1985) Structure-activity relationships in sweeteners. I. Chem Sens vol 10(1) pp 57-72

361Van der Heijden, A. van der Wei, H. & Peer, H. (1985) Structure-activity relationships in sweeteners. II. Chem Sens vol 10(1) pp 73-88

362Dahanukar, A. Lei, Y-T. Kwon, J. & Carlson, J. (2007) Two Gr genes underlie sugar reception in drosophila Neuron vol 56, pp 503–516 306 Neurons & the Nervous System

8.5.11.1 Rationalization of the lipid versus protein versus sugar debate in gustation EMPTY

8.5.12 Clinical (medical) disorders affecting taste

Bromley and Doty, writing in Doty, have illustrated the limited understanding of the taste modality from a clinical perspective363. They list over a dozen medical specialties involved in trying to understand taste disorders. They list nearly a dozen clinically recognized subdisorders and about forty medical conditions contributing to or mediating taste disorders. They offer no matrix or other technique leading to the underlying causes of these disorders. Costanzo et al., in the following chapter, discuss head injuries and taste, concluding that the predominant reports of loss of taste through head injury actually involve the more easily explained loss of olfactory sensation. Shallenberger described two patients being treated for hypoparathyroidism that were unable to perceive the sensation of sweetness (Section 8.5. 1.1.6). Assuming the glandular diagnosis was correct, the lack of ability to form PtdGal may account for a common problem among the geriatric population, particularly women. Functional loss of the GR 2 receptor could result in the sugars being perceived as sour, bitter, and potentially salty as well; it would depend on the width of the efficacy function of these other receptor channels.

8.5.13 Man-made taste sensors EMPTY

Julia Tsitron et al364.have recently reported on a Bayesian model of a man-made gustatory sensor focused on the reception of only four chemicals related to glucose. [See reference papers xxx. ]

8.5.14 Overview of the complete gustatory modality–stage 2 and higher ADD

8.5.14.1 Gross gustatory signal paths within the human cerebrum

Yamamato has provided good material on the gross organization of the stage 4 signal paths within the human brain, and associated paths in several mammals365. The material is extensive and will not be reviewed here. He writes using the global expression cortical gustatory area (CGA). Figure 8.5.14-1 shows the general areas associated with gustation in the cerebrum. It is not clear whether these are the initial stage 4 information extraction areas or later association areas. The mappings are old, dating from the 1930-40's.

363Bromley, S. & Doty, R. (2003) Clinical disorders affecting taste: evaluation and management In Doty, R. ed. Handbook of Olfaction and Gustation. NY: Marcel Dekker Chapter 44

364Tsitron, J. Ault, A. Broach, J. & Morozov, A. (2011) Decoding Complex Chemical Mixtures with a Physical Model of a Sensor Array PLoS Comput Biol vol 7(10): e1002224. doi:10.1371/journal.pcbi.1002224

365Yamamato, T. (1989) Role of the cortical gustatory area in taste discrimination In Cagan, R. ed. Neural Mechanisms in Taste Boca Raton, FL: CRC Press Chap. 9 Signal Generation & Processing 8- 307 “A; Cortical areas eliciting orolingual sensations following electrical stimulation. Solid circles indicate the point where electrical stimulation elicited taste sensations. Encircled area with stripes, tongue sensory area; area within dashed line, mouth sensory area. B; Three clinical cases of brain damage showing taste impairment: (1) bullet wound which elicited ageusia to four basic taste qualities, (2) bullet wound which elicited hypogeusia, mainly to sweet and sour tastes, (3) hematoma which elicited hypogeusia, mainly to salty and bitter tastes.” His figure 4 conceptual representation does not recognize the role of Lewis acids in gustation and follows the conventional view of the time in using the inorganic nocent, HCl, as the reference for the acid channel of gustation.

8.5.15 Confirmation of the hypothesis EMPTY 8.5.15.1 Role of inositol as an organic taste enhancer

Stone & Oliver has noted the remarkable ability of the organic molecule, inositol, as a “taste enhancer.366” This observation is strong confirmation of the choice within this theory to adopt muco-inositol as esterified to phosphatidic acid as the primary receptor of the N-Path sensory neurons.

This places inositol, as a gustaphore, in the optimum position to form a dimer with the sensory receptor. This dimer exhibits minimum stress in its d-value relative to the optimum d-value of the receptor and insures maximum transduction efficacy for this gustaphore.

[xxx 8.5.1.1.1, 8.5.1.6.7, 8.5.3, 8.5.4.4 and Fig 8.5.4-8 before condensation ]

End Section 8.5

Figure 8.5.14-1 Gross stage 4 gustatory information extraction areas in humans ADD. See text. CS; central sulcus. SS; Sylvian sulcus. From Yamamato, 1989.

366Stone, H. & Oliver, S. (1966) Beidler’s theory and human taste stimulation Percept Psychophysics vol 1, pp 358-360 308 Neurons & the Nervous System

Table of Contents 1 August 2016

8 Stage 1 & 2, Signal Generating & Processing Neurons...... 2 8.5 The gustatory modality ...... 2 8.5.1 Background for and summary--the gustatory modality hypothesis ...... 3 8.5.1.1 Background ...... 3 8.5.1.1.1 Historical documentation...... 3 8.5.1.1.2 Major problems with the RSC Jmol & JSmol Libraries . . 6 8.5.1.1.3 How have the taste sensations been defined?...... 7 8.5.1.2 Anatomy of the peripheral portion of the gustatory modality . . 8 8.5.1.2.1 Morphology of the gustatory modality...... 10 8.5.1.2.2 The morphology of the taste bud & sensory neuron ...... 16 8.5.1.2.3 The structure of the “sweet” lemma of the microvilli ...... 19 8.5.1.3 Electrophysiology of the sensory neurons ...... 22 8.5.1.3.1 The electrophysiology of the “sweet” gustatory sensory neuron...... 22 8.5.1.4 Initial block diagram of the modality ...... 24 8.5.1.4.1 A proposed top level architecture of the gustatory modality ...... 26 8.5.1.5 The chemistry most important to gustation ...... 26 8.5.1.5.1 The chemical families of carbohydrates involved in gustation ...... 27 8.5.1.5.2 Chemical families pertinent to gustatory receptor identification ...... 27 8.5.1.5.3 The special case of the saturated aliphatic alcohols and aldehydes...... 30 8.5.1.5.4 The unique role of the hydrated sodium ion in gustation ...... 31 8.5.1.5.5 Inorganic acids and astringents excite the nocent (pain) modality ...... 31 8.5.1.5.6 Definition of specific stereo–molecular structures ....31 8.5.1.5.7 Equilibrium in the context of gustation–a brief review ...... 33 8.5.1.5.8 The change in free energy associated with a DACB –a brief review...... 36 8.5.1.5.9 The primary question regarding the transduction mechanism of gustation...... 37 8.5.1.6 Summary of the gustatory modality hypothesis...... 37 8.5.1.6.1 Defining the gustatory perception space ...... 45 8.5.1.6.2 A 3D olfactory perception space with calibrated scales ...... 46 8.5.1.6.3 The proposed qualitative 3D gustatory perception space ...... 51 8.5.1.6.4 An equivalent quantitative 3D olfactory perception space EMPTY ...... 59 8.5.1.6.5 The gustatory response versus molarity of stimuli ....59 8.5.1.6.6 The structural constraints on tasting sweet...... 60 8.5.1.6.7 Extending the chemoreception concepts of Shallenberger, Kier and Beets ...... 61 8.5.1.6.8 Proteins as stimulants and/or gustaphores...... 63 8.5.1.6.9 Tests of the Electrolytic hypothesis of gustation...... 63 8.5.1.7 A dichotomy: the labeled-line and across-neuron-pattern theories ...... 63 8.5.1.8 Initial identification of human genetic differences...... 66 8.5.1.9 Renewal of the gustatory sensory neurons...... 66 8.5.2 Analysis of perceived gustatory sensations–MDS and other techniques ...... 66 Signal Generation & Processing 8- 309 8.5.2.1 Dendrographic representation ...... 67 8.5.2.2 ROC analysis...... 68 8.5.2.3 Employing MDS techniques in understanding gustation ...... 69 8.5.2.3.1 Background relative to the MDS technique–map ,making ...... 69 8.5.2.3.2 Dimensionality–selecting the number of dimensions ...... 75 8.5.2.3.3 Multidimensional scaling applied to gustation ...... 76 8.5.2.3.4 Strange representations due to dimension reduction ...... 78 8.5.2.3.5 2D MDS representation applied to gustation ...... 79 8.5.2.3.6 Initial considerations related to entropy in gustation ...... 79 8.5.2.4 Changes in multidimensional presentation EMPTY...... 80 8.5.2.4.1 Rotation and displacement of MDS axes...... 80 8.5.2.4.2 Conclusions from analysis of the data ...... 80 8.5.2.4.3 The basis functions of gustation...... 85 8.5.2.4.4 The proposed human sensory space of gustation . . . 88 8.5.2.4.5 The family of Neural Response Functions...... 89 8.5.2.5 Conclusions from analysis of the MDS technique & examples EMPTY ...... 91 8.5.2.5.1 The representation of an MDS dataset in the preferred form...... 91 8.5.2.5.2Failure to accommodate hidden variables in MDS representations ...... 92 8.5.3 The 2-step hypothesis of gustatory transduction...... 92 8.5.3.1 Previous theories of gustatory transduction ...... 95 8.5.3.2 The AH,B & AH,B,X coordination chemistry of the gustatory channel...... 95 8.5.3.3 Proposed transduction mechanism in gustation.....102 8.5.3.4 Proposed selection mechanism for “desireable/sweet” RENAME...... 105 8.4.4.6xxx Sugars as potential GR’s in gustation ...... 110 8.5.4 The initial selection operation of the gustatory sensory receptors .....111 8.5.4.1 Operation of the “sweet” gustatory sensory neuron ...... 112 8.5.4.1.1 Review of the historical database BRIEF...... 113 8.5.4.1.2 Chemical identification of large classes of sugars (saccharides) ...... 113 8.5.4.1.3 The unique “sugar alcohols” or glyco-alcohols .....115 8.5.4.1.4 The unique “sugar acids” or glyco-acids ...... 115 8.5.4.1.5 Sweetness antagonists (inhibitors) ...... 116 8.5.4.2 Operation of the “super sweet” sensory neuron & AH,B,X....116 8.5.4.2.1 The unique non– saccharide sweeteners EMPTY . . . 118 8.5.4.3 Operation of the “acidic” gustatory sensory neuron...... 118 8.5.4.3.1 Background ...... 118 8.5.4.3.2 Proposed acidic channel sensory neuron receptor ...... 119 8.5.4.3.3 The polarized forms of carboxylic acids...... 122 8.5.4.3.4 The perception of carboxylic acid derivatives as acids ...... 123 8.5.4.3.5 The perception of inorganic acids as nocents–HCl ...... 124 8.5.4.4 Operation of the “alkaline” gustatory sensory neuron...... 125 8.5.4.4.1 The details/confusion related toPtdIns...... 128 8.5.4.4.2 The details/confusion related to muco-inositol.....130 8.5.4.4.3 The gustaphores of the inositol ion ...... 134 8.5.4.4.4 The perception of sodium as sweet at low concentrations ...... 134 8.5.4.5 Operation of the “bitter” gustatory sensory neuron...... 135 8.6.10.3 The detailed nomenclature of the picric channel stimulants ...... 135 8.6.10.3.1 Review of the historical database...... 135 310 Neurons & the Nervous System

8.6.10.3.2 Summarizing the picrophores of taste ...... 137 8.6.10.3.3 Amarogentin, artabsin and quinine ...... 139 8.6.10.3.4 The triterpenes ...... 140 8.5.4.5.1 Review of diverse bitter gustants and gustaphores ...... 141 8.5.4.5.2 Potential picric channel receptors ...... 149 8.5.4.5.3 Hydrated organic molecules as picric channel gustaphore ...... 150 8.5.4.5.4 hydrated hydrogen sulfide as an inorganic picric channel gustaphore ...... 150 8.5.4.5.6 OBSOLETE MATERIAL ON PICROPHORE/RECEPTOR MATCH ...... 151 8.5.4.6 Operation of the “super-bitter” sensory neurons...... 153 8.5.4.7 Summary of the proposed receptor d-values CONSOLIDATE ...... 158 8.5.4.8 Other gustaphores...... 159 8.5.4.8.1 CaCl2 & MgCl2 as gustaphores or nocents ...... 159 8.5.4.8.2 The thio moieties as stimulants...... 161 8.5.4.8.3 The “water” gustatory sensory response...... 163 8.5.4.8.4 The “browned flavors” sensory response ...... 163 8.5.4.8.5 The role of amines & amino acids in the taste sensation TIE 8.6.2.6.6...... 164 8.5.4.8.6 The phenols and aliphatic-aromatics ...... 164 8.5.4.8.7 The non-hydroxyl guanidines...... 165 8.5.4.8.8 Procaine and other local anesthetics ...... 167 8.5.4.8.9 Nutmeg and Mace...... 167 8.5.4.8.10 Heterocyclics–caramel and butterscotch...... 167 8.5.4.8.11 Heterocyclics–the pyridines...... 167 8.5.4.9 The putative “umami” sensory response ...... 168 8.5.4.9.1 History of umami...... 168 8.5.4.9.2 Recent literature on umami...... 169 8.5.4.9.3 The underlying mechanism–the perception of umami ...... 170 8.5.4.10 The putative “non esterified fat” sensory response...... 172 8.5.4.11 The mints as nocents instead of gustants...... 173 8.5.5 The vernier (intensity) operation of the gustatory modality...... 174 8.5.5.1 Background ...... 174 8.5.5.1.1 Dipole potential and related dipole moment ...... 176 8.5.5.1.2 Dipole potential calculation ...... 178 8.5.5.1.3 Molecular electrostatic potential profiles ...... 179 8.5.5.2 Analog intensity variation due to gustaphores...... 181 8.5.5.2.1 Two distinct response–concentration characteristics for sweeteners...... 183 8.5.5.2.2 Structure of simple artificial and Anti-sweeteners . . . 183 8.5.5.2.3 Structure of super sweeteners–acesulfame & saccharine ...... 184 8.5.5.2.4 Potential dispersion centroids of super sweeteners ...... 189 8.5.5.2.5 The super-bitter (picric) stimulants EMPTY...... 190 8.5.5.3 Defining the gustatory receptor (standing alone) ...... 190 8.5.5.3.1 The nominal gustatory receptor EDIT ...... 193 8.5.5.3.2 The electrolytic properties of the receptors...... 194 8.5.5.3.3 Description of the operation of the sensory neuron in gustation...... 197 8.5.5.4 The perception versus stimulus intensity function ...... 199 8.5.6 Electrophysiology of gustation–the Excitation/De-excitation equation ...... 199 8.5.6.1 The impulse response of the stage 1 neuron...... 200 8.5.6.1.1 Character of the DACB phenomenon...... 200 Signal Generation & Processing 8- 311 8.5.6.1.2 Circuit description of the gustatory sensory neuron with GR REFOCUS EDIT...... 200 8.5.6.1.3 Analog waveforms generated by stage 1 neurons EMPTY REFOCUS...... 206 8.5.6.1.4 The generic Excitation/De-excitation equation applied to gustation EMPTY...... 208 8.5.6.1.5 Circuit parameters of gustatory transduction EDIT . . 208 8.5.6.2 The square pulse response of the stage 1 neuron...... 210 8.5.6.2.1 Circuit parameters of gustatory adaptation...... 210 8.5.6.3 Chemical kinetics at the receptor/gustaphore interface....211 8.5.6.3.1 The Beidler equation of chemical kinetics in transduction ...... 211 8.5.6.3.2 Transduction kinetics of the amino acids...... 215 8.5.6.3.3 Comparing the human perceived response to simple sugars ...... 218 8.5.6.3.4 Comparing the human perceived response between sweeteners ...... 220 8.5.6.3.5 Mechanisms of sweet taste transduction from Simon ...... 220 8.5.6.3.6 Solubility of the natural sugars from Andersen et al...... 221 8.5.7 Antagonists (blocking agents) & adaptation in the gustatory modality ...... 223 8.5.7.1 Generic blocking agents ...... 224 8.5.7.2 Gymnemic acids as G-path blockers...... 225 8.5.7.3 Affect of amiloride on the monkey & other species...... 226 8.5.7.3.1 Current problem related to the available archives and visualizers...... 227 8.5.7.4 Adaptation and/or suppression by antagonists in gustation EMPTY ...... 228 8.5.8 Analysis of the literature based on the hypothesis ...... 228 8.5.8.1 Reinterpretation of Smith et al. of 1983 using hamsters ...... 229 8.5.8.2 Reinterpretation/expansion of Rohse & Belitz of 1991 ...... 232 8.5.8.3 Reinterpretation of Hellekant et al. of 1997 using M. mulatta ...... 233 8.5.8.4 The investigations of Hellekant et al. of using chimpanzees . . 239 8.5.8.5 Reinterpretation of the 2002 paper of Danilova et al...... 240 8.5.8.6 Reinterpretation of the Giza & Scott 1991 paper ...... 242 8.5.8.7 Reinterpretation of the review by Spector & Travers of 2005 ...... 245 8.5.8.8 Reinterpretation of the nociceptor data from Kashiwagura et al. 1980 ...... 245 8.5.8.9 The conditioned taste aversion MDS data of Chang & Scott–1984 ...... 248 8.5.9 Extending the hypothesis to include the sweet proteins ...... 250 8.5.9.1 The study of protein mutations in human & primate gustation ...... 251 8.5.9.1.1 The folding of proteins...... 254 8.5.9.1 Searching glycophore location based on complex protein theory ...... 255 8.5.9.2 Amino acid sequencing of the sweet proteins...... 256 8.5.9.3 The high electrical charge of the peptides...... 262 8.5.9.4 Studies related to monellin...... 263 8.5.9.5 Recent experiments to characterize rare amino acids...... 266 8.5.9.6 Proposed mechanism of sweet protein perception...... 267 8.5.9.6.1 Details of proposed sweet protein selection...... 274 8.5.9.6.2 The AH,B,X relationship in sweet proteins ...... 277 8.5.9.6.3 Quantifying the AH,B,X dimensions in proteins...... 280 8.5.10 The dipole potential (DP) in the perceived intensity of gustation.....282 8.5.10.1 Background ...... 282 8.5.10.1.1 Background related to the phospholipid structure of the GR ...... 285 312 Neurons & the Nervous System

8.5.10.1.2 Background related to the DP of a stimulant in saliva ...... 290 8.5.10.1.3 Background related to the dispersion centroid relative to AH,B,X ADD...... 291 8.5.10.2 Results of the net change of a DP due to a AH,B type stimulant ...... 294 8.5.10.3 Details of proposed net change in DP by AH,B,X stimulants ...... 295 8.5.10.4 Details of proposed of sweet protein DP measurements....297 8.5.11 Correlating genes to gustatory receptors ...... 298 8.5.11.1 Rationalization of the lipid versus protein versus sugar debate in gustation EMPTY ...... 298 8.5.12 Clinical (medical) disorders affecting taste...... 298 8.5.13 Man-made taste sensors EMPTY...... 299 8.5.14 Overview of the complete gustatory modality–stage 2 and higher ADD ...... 299 8.5.14.1 Gross gustatory signal paths within the human cerebrum . . . 299 8.5.15 Confirmation of the hypothesis EMPTY ...... 299 8.5.15.1 Role of inositol as an organic taste enhancer...... 299

End Section 8.5 ...... 300 Signal Generation & Processing 8- 313

Chapter 8 List of Figures 8/1/16 Figure 8.5.1-1 Semi-schematic representation of the tongue...... 9 Figure 8.5.1-2 Proposed schematic of the gustatory system...... 11 Figure 8.5.1-3 Schematic of mammalian neural paths in gustation...... 13 Figure 8.5.1-4 Relative magnitude estimates for locations on the human tongue ADD...... 14 Figure 8.5.1-5 Diagram of types of lingual gustatory papillae ...... 15 Figure 8.5.1-6 Morphological features of the mammalian taste bud ...... 17 Figure 8.5.1-7 Tracing of a gustatory sensory neuron showing internal lemma ...... 18 Figure 8.5.1-8 Two views of the colax and microvilli of a gustatory sensory neuron ...... 19 Figure 8.5.1-9 The electrophysiology of the gustatory “sweet” microvilli MOD ...... 21 Figure 8.5.1-10 Candidate cytology & electrophysiology schematics of the gustatory sensory neuron DUMMY ADD...... 23 Figure 8.5.1-11 Provisional block diagram of the gustatory modality ADD...... 25 Figure 8.5.1-12 A normalized etiology of chemicals relating to proposed gustatory sensory receptors...... 29 Figure 8.5.1-13 Conformational notation applied to a Newman representation ...... 33 Figure 8.5.1-14 The equilibrium constant in different contexts ADD...... 35 Figure 8.5.1-15 Summary: sensory receptors of gustatory modality...... 41 Figure 8.5.1-16 Lemma sub-type designations for the sensory receptors of gustation ADD . . . 42 Figure 8.5.1-17 UPDATE XXX’s Summary: performance of the gustatory modality ...... 43 Figure 8.5.1-18 The first-order gustaphores of taste ...... 44 Figure 8.5.1-19 The one-dimensional effectivity graph of gustatory performance...... 46 Figure 8.5.1-20 Alternate representations of a 1D parameter in a 3D perception space ....48 Figure 8.5.1-21 Potential taste sensation space for a mammalian species ...... 50 Figure 8.5.1-22 Preferred MDS space based on a right-hand rule...... 52 Figure 8.5.1-23 Citations & parameters of recent MDS investigations ADD DATA to Smith . . . 53 Figure 8.5.1-24 The transition from behavioral to fundamental perspectives in gustation....58 Figure 8.5.1-25 A representative summated recording from the chorda tympani nerve.....61 Figure 8.5.1-26 Caricature of extended multipoint model of sugar-receptor coupling...... 62 Figure 8.5.1-27 Records of total chorda tympani responses to water and taste stimuli...... 65 Figure 8.5.2-1 Cluster dendrogram of 31 hamster PbN neurons...... 68 Figure 8.5.2-2 Creation of a 2D MDS representation based on a matrix tabulation...... 70 Figure 8.5.2-3 Creation of a 2D MDS with absolute scales ...... 71 Figure 8.5.2-4 Scatter diagrams and their application to gustation ...... 74 Figure 8.5.2-5 The Kruskal stress index as a percentage ...... 75 Figure 8.5.2-6 Multidimensional scaling of gustatory stimulants in monkey ADD ...... 77 Figure 8.5.2-7 Two-dimensional histogram of hamster taste preferences...... 79 Figure 8.5.2-8 The foundation for the chromaticity diagram of tetrachromatic vision...... 82 Figure 8.5.2-9 A taste sensation space based on an incomplete experimental database paths ...... 83 Figure 8.5.2-10 REWRITE A two-dimensional taste space with alternate axes applied ...... 86 Figure 8.5.2-11 Proposed taste sensation space for a mammalian species...... 88 Figure 8.5.2-12 Proposed 2-D neural response function of the molecule X, NRFX...... 90 Figure 8.5.3-1 XXX modified multidimensional analysis based on the polar head of lipids EMPTY ...... 94 Figure 8.5.3-2 Proposed coordination chemistry of the G-Path sensory neurons...... 96 Figure 8.5.3-3 Comparison of AH,B,X geometries...... 98 Figure 8.5.3-4 Sweet tasting non-sugars and the their AH,B relationships ...... 100 Figure 8.5.3-5 Proposed sensitivity of the desirable/”sweet” sensory channel ADD ...... 102 Figure 8.5.3-6 The numbering system of the simple sugars ...... 106 Figure 8.5.3-7 Two gangliosides associated with the neural system...... 108 Figure 8.5.4-1 The “super sweet” tripartite glycophores of two sugars ...... 117 Figure 8.5.4-2 Proposed phosphatidylserine shown in polar form...... 120 Figure 8.5.4-3 Serine as the potential sensory receptor ligand...... 120 Figure 8.5.4-4 A gas-phase carboxylic acid dimer EDIT...... 121 Figure 8.5.4-5 Potential d-values for phosphatidyl serine...... 121 Figure 8.5.4-6 The sodium ion at hydration levels of 2 and 6...... 126 Figure 8.5.4-7 A potential hydrated sodium “dimer...... 127 Figure 8.5.4-8 Muco-inositol phosphate and a fully hydrated sodium...... 129 Figure 8.5.4-9 Representations of a six-member ring from Glusker et al., 1994 ...... 131 314 Neurons & the Nervous System

Figure 8.5.4-10 Stereo-isomers of inositol ...... 132 Figure 8.5.4-11 Chair conformations of muco-inositol ...... 133 Figure 8.5.4-12 Muco-inositol conformation ...... 133 Figure 8.5.4-13 Most potent of various classes of bitter compounds...... 136 Figure 8.5.4-14 The premier gustant of the bitter or P-channel of gustation...... 138 Figure 8.5.4-15 Ionic forms of picric acid from the literature...... 138 Figure 8.5.4-16 Quinine as a stimulant of the P–channel of gustation ...... 140 Figure 8.5.4-17 A typical triterpene...... 141 Figure 8.5.4-18 The diversity of historically bitter compounds EDIT...... 142 Figure 8.5.4-19 Denatonium benzoate & saccharide, the most bitter compounds ADD & EDIT ...... 143 Figure 8.5.4-20 Denatonium benzoate from Jmol ...... 144 Figure 8.5.4-21 Alternate structures for the denatonium family ...... 145 Figure 8.5.4-22 Quinine, the preferred bitter taste in the laboratory ...... 146 Figure 8.5.4-23 Corilagin, a tannic acid & a complex sugar ester ...... 146 Figure 8.5.4-24 Caffeine, 1,3,7-trimethylxanthine and its parent ...... 147 Figure 8.5.4-25 Two “most intriguing” examples of anisaldoxime...... 148 Figure 8.5.4-26 Potential phosphatidyl aspartic acid receptor and picric acid...... 149 Figure 8.5.4-27 Candidate sensory receptor performance for the “bitter” channel...... 151 Figure 8.5.4-28 Candidate picric sensory receptor and quinine coordinate bonding RESCALE ...... 152 Figure 8.5.4-29 Potential P-path picrophores and receptors...... 153 Figure 8.5.4-30 The bitter and super-bitter stimulants of van der Heijden EDIT...... 154 Figure 8.5.4-31 2D representations of Lucidenic acid D1 ...... 155 Figure 8.5.4-32 Lucidenic acid D1 (HMDB 38199) from Jmol ...... 156 Figure 8.5.4-33 Variants of lucidenic acid ADD...... 157 Figure 8.5.4-34 Potential AH,B,X geometries for amarogentin and artabsin...... 158 Figure 8.5.4-35 Proposed summary d-values for the gustatory receptors...... 159 Figure 8.5.4-36 Calcium cation fully coordinated with water...... 160 Figure 8.5.4-37 The potential gustaphores of CaCl2 and MgCl2 ADD ...... 161 Figure 8.5.4-38 The d-values associated with orbital pairs in the amino acid, cysteine .....162 Figure 8.5.4-39 Phenythiocarbamide as presented in stick and 3D form ...... 163 Figure 8.5.4-40 Quanidine, a stimulant with three gustaphores...... 166 Figure 8.5.4-41 The guanidine derivative, SC-45647 as a super-sweetener ...... 166 Figure 8.5.4-42 Structure of procaine with d-values as represented using the Jmol visualizer ...... 167 Figure 8.5.4-43 Effectiveness of 12 pyridines in stimulating the chemoreceptors of crayfish ...... 168 Figure 8.5.4-44 A selected group of stimulants represented as “umami” type ...... 171 Figure 8.5.4-45Stimulants associated with umami due to their multiple gustaphores EDIT . . . 171 Figure 8.5.5-1 The many lone pairs of electrons in amiloride available for dual coordinate bonding ...... 176 Figure 8.5.5-2 Computed molecular electrostatic potential maps of amiloride...... 177 Figure 8.5.5-3 The mechanism(s) of intensity determination (including dispersion) in gustation ...... 181 Figure 8.5.5-4 The structural form of xylitol, a sweet alcohol EMPTY...... 183 Figure 8.5.5-5 Acesulfame and saccharine in dry and solvated forms ...... 184 Figure 8.5.5-6 Acesulfame as a conventional and super sweetener...... 185 Figure 8.5.5-7 Aspartame as a super-sweetener...... 186 Figure 8.5.5-8 Aspartame showing the axis of its AH,B,X “supersweet” glycophore ...... 186 Figure 8.5.5-9 Aspartame bonding to the picric sensory receptor ADD...... 187 Figure 8.5.5-10 The structure of saccharin ...... 188 Figure 8.5.5-11 ED A dimer situation with galactose as both sensory receptor and stimulant ...... 189 Figure 8.5.5-12 Concept: A stacked situation with the sensory receptor galactose interfacing with the glycophore of an undefined stimulant...... 190 Figure 8.5.5-13 The serine ligand coordinate bonding with organic acids and amino acids ...... 191 Figure 8.5.5-14 EDIT A dimer situation with galactose as both sensory receptor and stimulant ...... 192 Signal Generation & Processing 8- 315 Figure 8.5.5-15 The galactose-based receptor in context with the sensory microvilli...... 194 Figure 8.5.5-16 The proposed molecular operation of the microtubules in GRN’s ...... 196 Figure 8.5.5-17 Effective potential at the base of the 1st amplifier (Activa) of the GRN.....198 Figure 8.5.6-1 Framework for impulse response versus square pulse analyses ADD...... 200 Figure 8.5.6-2 Candidate circuitry of the gustatory sensory neurons...... 201 Figure 8.5.6-3 Proposed cytological and electrolytic description of the olfactory sensory neuron ADD & MODIFY...... 204 Figure 8.5.6-4 Summated chorda tympani and glossopharyngeal nerves during taste stimulation ...... 207 Figure 8.5.6-5 Odorant induced currents at various holding potentials from a newt, Cynops pyrrhogaster ADD...... 209 Figure 8.5.6-6 The characteristics of the E/D response in gustation...... 210 Figure 8.5.6-7 Comparison of the Beidler and Dzendolet equations...... 213 Figure 8.5.6-8 Parametric graph of the kinetics of gustation by Dzendolet ADD...... 215 Figure 8.5.6-9 Beidler equation plotting data for amino acids, C/R versus C...... 216 Figure 8.5.6-10 Amino acid slope values, maximum responses & equilibrium constants ....217 Figure 8.5.6-11 The perceived response of humans to polysaccharides of nominal concentration based on their dry weight...... 219 Figure 8.5.6-12 Solubility of the natural saccharides versus temperature ...... 222 Figure 8.5.7-1 Lactisole, a sweetness blocker ...... 226 Figure 8.5.7-2 Amiloride (DB00594) as represented by a Jmol file ...... 228 Figure 8.5.8-1 Three-dimensional space showing the location of 18 stimuli...... 232 Figure 8.5.8-2 Distribution of 23 stimuli in a 3D space based on data from 47 CT fibers .....236 Figure 8.5.8-3 Distribution of 18 stimuli in a 3D space based on data from 33 NG fibers ....237 Figure 8.5.8-4 A presentation based on a 3D MDS analysis with a limited gustaphore set . . . 241 Figure 8.5.8-5 Changes in three-dimensional taste spaces of one rat caused by amiloride ...... 244 Figure 8.5.8-6 Integral of stage 3 action potential recovered from the glossopharyngeal nerve of the bullfrog ...... 247 Figure 8.5.8-7 Three dimensional MDS spaces representing a CTA group and a control group...... 250 Figure 8.5.9-1 Characteristics of proteins perceived as sweet by humans ADD ...... 253 Figure 8.5.9-2 Diagram showing the position of mutations on the brazzein molecule ...... 258 Figure 8.5.9-3 The amino acid sequence of brazzein with three positions marked...... 259 Figure 8.5.9-4 A partial 3D ball and stick representation of the b-turn in brazzein EDIT...... 260 Figure 8.5.9-5 Complete amino acid chain for monellin...... 265 Figure 8.5.9-6 In vitro NMR spectroscopic analysis showing SeP presence...... 267 Figure 8.5.9-7 Amino acid sequence of brazzein and DNA sequence of synthetic gene . . . 268 Figure 8.5.9-8 Options available for the dispersion centroid supporting a glycophore in an a-helix ...... 270 Figure 8.5.9-9 The relative positions of the amino acid residues in chromatography...... 273 Figure 8.5.9-10 Location of the glycophore of brazzein on an annotated space-filled model ...... 278 Figure 8.5.9-11 Schematic of glycophore involving Tyr24, Asp25 & Asp29...... 279 Figure 8.5.9-12 43 superimposed conformers of brazzein in solution...... 280 Figure 8.5.9-13 Expanded table of AH,B,X parameters for the glycophores of taste ...... 281 Figure 8.5.10-1 Abridged Table II of Beitinger et al., surface potentials...... 284 Figure 8.5.10-2 Lipid and bilayer chains from Zheng & Vanderkooi ...... 286 Figure 8.5.10-3 Atypical calculated components of the electrostatic potential at the bilayer midplane...... 286 Figure 8.5.10-4 Chemical structure of DPPE & GM1 (a specific PtdEtn & a “ganglioside”) . . . 288 Figure 8.5.10-5 The dipole potential of different lipids measured with the CR method .....288 Figure 8.5.10-6 Electrostatic field of aspartame showing the AH,B,X relationship ...... 291 Figure 8.5.10-7 3D histogram of centroid of X for a group of stimulants ...... 293 Figure 8.5.10-8 Detailed electrolytic sensory neuron operation in AH.B situation ...... 295 Figure 8.5.10-9 The schematic difference between the AH,B and AH,B,X configurations during DACB ...... 296 Figure 8.5.10-10 Galactose in the plane of O-3 & O-4...... 297 Figure 8.5.14-1 Gross stage 4 gustatory information extraction areas in humans ADD...... 299 316 Neurons & the Nervous System

(Active) SUBJECT INDEX (using advanced indexing option) 3D . 6, 7, 32, 33, 39, 46-48, 51, 56, 59, 61, 62, 71-73, 80, 81, 83, 87, 129, 141, 144-146, 148, 149, 152, 154, 155, 158, 163, 165, 166, 169, 174, 176, 177, 179, 227, 230, 231, 233-238, 241, 242, 244, 248, 249, 256, 259, 260, 271, 277, 280, 289, 290, 293 3-D...... 83, 297 9(2) + 2...... 18 99%...... 284 acidophore ...... 39, 44, 111, 158, 165, 169, 171, 172, 227, 228, 234-238, 241, 251, 252, 259, 262 across-neuron...... 64, 174 action potential...... 26, 67, 68, 73, 200, 201, 205, 206, 208, 235, 238, 240, 246, 247 Activa . 20, 21, 38, 88, 102, 103, 109, 112, 113, 121, 123, 137, 169, 174, 181, 182, 196-199, 201-205, 208, 210, 248, 282, 283, 294-296 active diode...... 201 adaptation.... 38, 59, 67, 69, 73, 79, 89, 92, 113, 137, 163, 181, 197, 199, 210, 212, 223, 228, 246- 248, 282 adaptation amplifier...... 137 AH,B . . . 5, 19, 39, 41, 44, 45, 58, 61, 62, 95, 97-102, 107-109, 112, 113, 116, 119, 120, 125, 127, 142, 144-146, 148, 151, 152, 154-156, 158, 164-166, 169, 170, 178, 181, 183-187, 189-192, 194, 197, 198, 205, 220, 221, 224, 233, 234, 253, 257, 260, 263, 268, 269, 271, 274, 277-284, 290, 291, 293-297 AH,B,G...... 62 AH,B,X . . . 39, 44, 45, 62, 95, 98, 102, 116, 142, 145, 152, 154-156, 158, 166, 178, 181, 183-187, 190, 198, 205, 220, 221, 224, 234, 253, 257, 263, 268, 269, 271, 274, 277, 279-282, 291, 293, 295-297 aliphatic-aromatics ...... 164 amiloride...... 5, 58, 78, 175-178, 199, 221, 223, 224, 226-228, 237, 238, 240, 242-245, 290 ammonia ...... 272 amplification ...... 200 amygdala...... 10, 14, 15, 249 anomeric carbon ...... 114 arginine ...... 216, 217, 273, 275 ascorbic acid...... 6 association areas...... 299 atomic force microscopy...... 19 attention...... 2, 92, 125, 176, 184, 257, 258 axon segment ...... 4 axoplasm ...... 205, 206, 208 azeotrope...... 112, 231 bar code...... 238, 257 Bayesian ...... 5, 63, 67, 221, 230, 272-274, 299 bilayer.... 20, 21, 27, 39, 42, 58, 93, 94, 106, 109, 112, 121, 151, 193, 195, 198, 203, 282-287, 289, 290, 294, 296 bilayer membrane...... 151, 193, 198, 282, 285, 287, 289 Bombyx mori...... 5 boundary layer ...... 294 brazzein...... 57, 63, 251-253, 256-263, 266, 268, 269, 271, 273-281, 297 Brownian motion ...... 263, 274, 280 C/D ...... 5, 21, 211 calibration...... 90, 213, 250 cAMP...... 5, 206 camphor...... 174 capsaicin ...... 229 Central Limit Theorem...... 73, 91 cerebrum ...... 299 chirality ...... 111, 174 cis- ...... 32, 38, 40, 110 Signal Generation & Processing 8- 317 computation ...... 202 computational...... 69, 99, 122, 129, 144, 174, 177-179, 184, 187, 266, 272, 282, 290 computational chemistry...... 99, 122, 129, 144, 282, 290 confirmation...... 113, 133, 134, 155, 180, 299, 300 continuum ...... 45, 76 coordinate bond.... 38, 39, 42, 61, 97, 109, 120, 121, 124, 132-134, 138, 141, 145, 146, 148, 151, 176, 177, 182, 186, 190, 191, 194, 198, 200, 223 coordinate chemistry ...... 34-37, 41, 93, 102, 119, 125, 131, 213, 295 cribriform plate...... 205 cross section...... 19 cross-section...... 88, 203 Cu(I)...... 114 Cu(II) ...... 114 DACB..... 30, 31, 36-38, 47, 58, 83, 102-104, 116, 119, 120, 123, 124, 148, 150, 153, 155, 162, 165, 166, 174, 176-180, 183, 197, 200, 201, 212, 217, 218, 221, 224, 226, 228, 233, 245, 251, 252, 261, 262, 268, 273, 279-282, 290, 293, 294, 296 data base...... 82, 85, 154 database ...... 7, 60, 68, 72, 83, 85, 91, 113, 135, 203, 227, 249 Debye...... 178, 202 denatonium...... 95, 143-145, 236 dendrolemma ...... 5, 90, 110, 112, 196 determinants ...... 263 DG...... 33, 34, 37 dihedral...... 32 diode...... 21, 109, 182, 196, 201, 209 diol...... 32, 40, 159, 180, 218, 219, 222 dipole moment ...... 28, 58, 112, 122, 123, 148, 176-178, 180, 195, 197, 202, 203, 282, 284 dipole potential . . 21, 28, 38, 58, 95, 98, 101, 102, 121, 138, 140, 176-179, 182, 183, 185, 193, 195, 197, 198, 261, 262, 268, 273, 274, 278, 280, 282-285, 288-290, 293-297 disparity...... 147 dispersion point ...... 97, 98, 293, 296 DNA...... 266, 268, 271 dopamine...... 223 dulcal ...... 212 dynamic range ...... 137, 197, 213 E/D...... 3, 61, 80, 113, 210, 211, 248 electrometer ...... 107, 195 electrostenolytic process...... 21, 196, 203, 217, 225 enantiomers...... 115, 190 entropy...... 34, 54, 55, 78-80, 111, 218, 234, 244, 245 epiglottis...... 15 equilibrium ...... 33-36, 102, 103, 113, 114, 200, 211, 212, 215, 217-219, 254, 255 equilibrium constant...... 35, 36, 103, 113, 212, 217, 219 exothermic animals...... 211 expanded . . 6, 24, 33, 48, 62, 71, 78, 92, 103, 143, 180, 181, 205, 210, 213, 233, 244, 267, 281, 290, 294, 295 feedback ...... 15 Fehling’s reagent...... 114 free energy...... 33, 34, 36, 212, 217, 218, 254 GABA...... 97 ganglion neuron ...... 10, 12 ganglioside...... 106-111, 283, 284, 287, 288, 294 genetics ...... 66, 143, 147, 256 genome ...... 266, 272 globoside ...... 19, 21, 93, 95, 110, 112, 196 glomeruli...... 12, 31, 205 gluconic acid...... 115 glucophore...... 33, 102, 117, 139, 180 glutamate...... 7, 8, 22, 39, 49, 51, 53, 58, 59, 89, 93, 169-171, 225, 228, 237, 243, 273 glycol...... 32, 38, 40, 41, 44, 51, 53, 58, 60, 238, 242 318 Neurons & the Nervous System glycophore . . 5, 32, 33, 39, 40, 44, 45, 49, 58, 61, 98, 99, 108, 112, 115, 141, 158, 169, 171, 184-186, 189-191, 226, 234-236, 238, 241, 255-257, 260-262, 268-271, 273, 274, 277-280, 282, 293-297 GPCR...... 98 guanidine...... 166, 177, 178, 234 gustophore...... 158 g-protein...... 5, 22 Hodgkin Condition ...... 3 hole ...... 141 homogeneous ...... 151 homologs ...... 162 hormone...... 255 Huckel Rule...... 233 hydrogen bond . . . 30, 36, 45, 58, 83, 95, 119, 121, 127, 134, 149, 150, 164, 170, 180, 200, 233, 262 hydrogen sulfide ...... 86, 150 hydronium...... 21, 83, 111, 195, 196, 294 hydronium liquid crystal ...... 195 hypothalamus ...... 10, 14, 15, 249 h-neurons ...... 76, 234, 235 inositol...... 5, 42, 51, 53, 61, 93, 111, 115, 128-130, 132-134, 170, 171, 183, 227, 299, 300 intramolecular hydrogen bond...... 164 inverting ...... 180 IUB ...... 128, 129, 131 IUBMB...... 129, 132 IUPAC...... 115, 128, 129, 132 just-noticeable difference ...... 34 Kruskal stress...... 56, 73-75, 91, 235-239, 241 labeled line...... 45 labeled-line...... 64, 174, 231, 240 Langmuir...... 282 latency ...... 195, 197, 206 limbic system ...... 10 London dispersion ...... 60, 96 LOT...... 283 lysine ...... 252, 267, 273 major nerves...... 3 marker...... 65 masking...... 224 MDS....4, 26, 46-49, 51-53, 55-57, 59, 67, 69-75, 78-81, 84, 86-92, 229-231, 234-237, 239-242, 245, 248-250, 293 microtubule ...... 196 microvilli ...... 11, 18-21, 38, 39, 88, 92, 93, 109, 110, 112, 121, 147, 151, 169, 190-192, 194, 225 miraculin...... 112, 226, 240 modulation...... 10, 91, 113 multi-dimensional...... 43, 76, 80, 81, 84, 87, 113, 169 myelin ...... 30, 107 myelinated...... 4, 10, 12 Myelination...... 12, 16, 17 N1...... 145, 177, 244, 255 N2...... 255 narrow band ...... 46, 88 natrophore...... 32, 39, 40, 44, 161, 169, 171, 185, 191, 223, 226, 227, 234, 237, 238, 241 navigation ...... 260 neural coding...... 8, 56, 60, 78, 85, 208 neural response function ...... 89, 90, 92 neurotransmitter...... 22, 92 neuro-facilitator...... 169, 223 nocent modality ...... 31, 83 Node of Ranvier...... 4, 10, 12, 17 noise ...... 33, 34, 54, 74, 91, 103, 137, 216, 218, 220, 267 Signal Generation & Processing 8- 319 n-neurons ...... 76, 84 n-type ...... 196 orbital . 31, 44, 45, 97, 99, 111, 139, 140, 145, 148, 152, 153, 156, 158, 162, 165, 166, 223, 251, 260, 269, 271, 277, 279, 297 orbitals..... 31, 32, 38, 39, 44, 45, 59, 62, 103, 104, 116, 140, 144, 149, 152-155, 157-159, 162, 165, 167, 172, 173, 177, 189, 223, 233, 251, 259, 268, 269, 271, 277-280, 293, 294, 296, 297 orbitofrontal cortex...... 10 oskonatory ...... 173 O–H- -O...... 149 P/D equation ...... 218 pain...... 3, 31 parametric...... 114, 208, 213, 215 parietal lobe...... 10, 14, 78 patch-clamp ...... 206 pedestal ...... 245, 246 percept...... 14, 87, 216, 300 perceptual space...... 47, 49, 56, 78 picrophore .... 39, 44, 59, 112, 137-141, 149, 151, 153-158, 161, 185, 187, 192, 223, 226, 228, 238, 241, 242, 245 piezoelectric...... 4 pnp ...... 196, 205 poditic...... 208 polyprotic...... 35, 36 pons...... 11 protocol ...... 6, 49, 51, 59, 60, 67, 69, 80, 81, 103, 199, 210, 218, 226, 242, 249, 272 pulse-to-pulse...... 59 pyrazine...... 175, 178 pyridine...... 168, 184, 185 p-type ...... 196 quantum-mechanical...... 3, 4, 92, 94, 156, 197, 199-201, 295 rafts ...... 18-20 reading...... 248 residue...... 110, 218, 252, 259-262, 268-270, 272-276, 278, 279, 297 resonance...... 114, 121, 133, 144, 279 saliency map ...... 14, 26, 56, 78, 82, 229, 244 SAR...... 62, 112, 118, 224 second messenger ...... 221 signal-to-noise ...... 267 stage 0 ...... 174, 246-248 stage 1 . . 2, 24, 26, 67, 73, 78, 116, 174, 199-201, 206, 210, 226, 229, 230, 238, 240, 246, 263, 268, 274, 282, 294 stage 1A ...... 246 stage 1B ...... 246 stage 2 ...... 11, 13, 24, 26, 38, 67, 78, 116, 174, 200, 226, 229, 230, 235, 238, 246 stage 3 . 4, 12, 16, 17, 26, 47, 56, 58, 59, 67, 68, 73, 107, 168, 200, 201, 211-213, 229, 230, 235, 246, 247 stage 3A ...... 246 stage 4 ...... 5, 10, 13, 26, 45, 47, 48, 56, 67, 69, 299 stage 5 ...... 7, 14, 26, 45, 56, 80, 137 stage 6 ...... 7 stress...... 34, 56, 72-75, 91, 185, 235-239, 241, 300 structural chemistry...... 194 sugar alcohol...... 5, 115, 174, 234 synapse...... 4, 201 SYSTAT ...... 67 s-neurons...... 76 temporal lobe ...... 10, 14, 26 tests of the hypothesis...... 63 tetrahedron ...... 83, 84, 87, 97 thalamus...... 10, 14 threshold...... 60, 63, 66, 136, 137, 142, 143, 151, 153, 154, 200, 219, 242, 262, 275, 282 320 Neurons & the Nervous System topography ...... 14 topology...... 111 torsion ...... 32 transduction . . 3, 5, 6, 8, 9, 16, 19, 20, 22, 28, 37, 38, 43, 59, 61, 63, 66, 81, 88, 92-95, 102, 109, 111, 112, 118, 119, 121, 135, 141, 142, 148, 149, 175, 177, 181, 195, 196, 199, 200, 203, 204, 206-208, 211, 213, 215, 216, 218, 220, 221, 225, 232, 240, 246-248, 261, 263, 266, 273, 274, 300 transition dipole moment ...... 202, 203 translation...... 169, 266 trans- ...... 32, 38, 40, 51, 53, 58, 110, 132-134, 173, 218, 219, 222, 261 trigeminal nerve...... 11, 159 type 1 ...... 21, 194 type 2 ...... 20, 21, 109, 112, 295 type 4 ...... 20, 21, 42, 88, 93, 109, 111, 147, 191, 195, 198, 225, 294, 295 umami...... 7, 8, 16, 37, 39, 51-53, 58, 59, 68, 85, 89, 168-172, 225, 240 V2...... 39 Van der Waals ...... 96 verification ...... 199 white matter...... 30, 110 Wikipedia ...... 11, 96, 114, 131, 143, 167, 174, 225, 226 xxx . . 1-3, 5, 8, 11, 19, 24-26, 31, 33, 37, 39, 43, 51, 56, 60, 61, 68, 72, 73, 87, 94, 101-103, 106, 107, 109, 111-113, 122, 124, 134, 143, 145, 147, 150, 151, 161, 163, 168, 175, 189, 191, 194-198, 204, 205, 208, 211, 218, 220, 221, 225, 228, 229, 232, 234, 240, 248, 252, 255, 256, 260, 261, 263, 265, 278, 282, 285, 286, 293, 294, 299 xylitol ...... 183, 184, 234 [xxx . . . 2, 14, 15, 20, 21, 26, 32, 37, 39, 43, 44, 47, 50, 51, 61, 64, 67, 76, 80, 84, 86, 91, 92, 100, 106, 109-111, 113-115, 118, 128, 129, 134, 135, 137, 142, 145-147, 149, 151, 161, 162, 164, 174, 182-185, 190-192, 199-202, 205, 208, 210, 215, 226, 228, 233, 234, 237, 242, 245, 251, 252, 259, 261, 269, 277, 285, 287, 293, 296-298, 300