The Pennsylvania State University

The Graduate School

Department of Food Science

CHEMESTHESIS AND BITTERNESS OF NATURAL AND SYNTHETIC ANTI-

INFLAMMATORY STIMULI

A Thesis in

Food Science

by

Samantha Bennett

! 2012 Samantha Bennett

Submitted in Partial Fulfillment of the Requirements for the Degree of

Master of Science

May 2012

ii The thesis of Samantha Bennett was reviewed and approved* by the following:

John Hayes Assistant Professor of Food Science Thesis Advisor

Joshua Lambert Assistant Professor of Food Science

John Coupland Professor of Food Science

John Floros Professor of Food Science Head of the Department of Food Science

*Signatures are on file in the Graduate School

iii ABSTRACT

Recently, Gibson’s ecological approach to classifying chemosensation has regained

popularity. In his model, sensations are not classified on the basis of anatomy, but rather by the

function they serve. The idea of a “chemofensor” complex expands the Gibsonian model to include chemesthetic sensations, as well as the classical bitter and sour tastes and rancid or rotten

odors. This comes at a fitting time when more unpalatable, but healthful compounds are incorporated into our diet. In particular, the discovery of oleocanthal, a compound responsible for the pungency of virgin olive oils, has reopened issues of qualitatively assessing the characteristics

of chemesthetic compounds. This thesis includes a discussion of the difficulties surrounding a

human behavioral approach to measuring chemesthesis and bitterness, including perceptual similarity, temporal considerations, and sensory adaptation. Major experimental findings include:

Study 1- Differences between sub-qualities of irritation from olive oil, ibuprofen, and capsaicin

were quantified, implicating that olive oil shares qualities with both ibuprofen and capsaicin.

Study 2- Behavioral response to compounds that share structural similarity to oleocanthal and

ibuprofen were used to identify functional parts of the molecules as well as further characterize the oral sensations from commonly used pain medications. The data from tyrosol and oleuropein suggest the functional end of oleocanthal does not include the phenol group, but may require the aldehyde groups at the other end of the molecule. Study 3- Ibuprofen and naproxen are similar to

each other (high irritation, some bitterness), but not acetaminophen (high bitterness, little

irritation). This result mimics what would be expected structurally. Finally, in Study 4- we explored the role of fat for masking unpleasant oral sensations from ibuprofen in dairy products.

We observed a modest reduction in irritation with increasing milk fat, but bitterness was unaffected. Unexpectedly, the reduction cannot be explained by partitioning into the fat phase, so

more sophisticated analyses are needed to understand the binding mechanisms responsible.

iv

TABLE OF CONTENTS

LIST OF FIGURES ...... vi

LIST OF TABLES...... viii

ACKNOWLEDGEMENTS...... ix

Chapter 1 The importance of chemesthesis in flavor ...... 1

Sidebar ...... 1 Introduction...... 2 Methodologial Concerns...... 4 Promising Techniques...... 8 Conclusions...... 9

Chapter 2 Preface...... 10

Chapter 3 Chemestethic sub-qualities of capsaicin, ibuprofen and olive oil...... 14

Introduction...... 15 Materials and methods ...... 19 Results...... 22 Discussion...... 27

Chapter 4 Structure-activity relationships between olive oil phenolics: tyrosol, oleuropein, and oleocanthal ...... 31

Introduction...... 31 Materials and methods ...... 34 Results and Discussion ...... 36

Chapter 5 Common over-the-counter pain medications differ in their primary perceptual qualities...... 39

Introduction...... 39 Materials and methods ...... 41 Results and Discussion ...... 43

Chapter 6 Physiochemical masking of irritation and bitterness with milk fat...... 46

Introduction...... 47 Materials and methods ...... 50 Results and Discussion ...... 54

Chapter 7 Conclusions and further steps ...... 60

v References...... 64

Appendix A: Chapter 2 Copyright Permissions ...... 69

Appendix B: Chapter 5 Supplemental Data...... 72

vi

LIST OF FIGURES

Figure 1-1: Difference scores of time intensity curves for cooling, mint flavor, and bitterness of two gum samples. The control contains peppermint extract and the treatment contains a novel compound intended to increase cooling. While bitterness is higher in the treatment gum for the first two minutes of chewing, it decreases to be no different from the control as cooling peaks after 3 minutes ...... 5

Figure 1-2: Time-intensity ratings of capsaicin after exposure to a non-desensitizing stimulus (citric acid) and a desensitizing stimulus (capsaicin) daily for ten days. The graph shows that the group exposed to capsaicin showed significantly lower intensity ratings and faster recovery than the control group...... 6

Figure 1-3: Shown is a box plot of group intensity ratings for irritation qualities of olive oil. The data suggests that participants had a difficult time distinguishing between the qualities, as mean intensities are similar for all qualities...... 8

Figure 3-1: Chemical structures of (a) the non-steroidal anti-inflammatory drug ibuprofen (b) (")-deacetoxy-dialdehydic ligstroside aglycone [oleocanthal] and (c) capsaicin, a known agonist of the TRPV1 channel...... 16

Figure 3-2: Intensity ratings (group means with standard errors) for ‘overall irritation’ in the throat for each stimulus. Capsaicin was significantly different from olive oil and ibuprofen at all time points, while olive oil and ibuprofen did not differ significantly at any time point. Abbreviations on the right axis correspond to labels given on the gLMS presented to participants: BD= Barely Detectable, W=Weak, M=Moderate, S=Strong, and VS=Very Strong...... 24

Figure 3-3: Correlations between Imax values for overall irritation for each stimulus. (A) Ibuprofen correlated with both olive oil (p <0.0001) and capsaicin (p=0.038). (B) Capsaicin was also correlated with olive oil (p=0.005). Similar results were seen for the AUC values ...... 25

Figure 3-4: Percentages for the number of times each of the sub-qualities was described as the ‘predominant sensation’ for the irritation from a particular stimulus. Note that this measure does not directly reflect the intensity of the sensations. Capsaicin was rated as burning 56% of the time, but a disjointed scale is used to highlight differences between the other sub-qualities...... 26

Figure 4-1: Figure from Ref [1] showing structural analogs of oleocanthal created to assess stimulation of TRPA1 in rat trigeminal neurons...... 33

vii Figure 4-2: Chemical structures of phenolic compounds from extra virgin olive oil (a) oleocanthal (b) tryrosol and (c) oleuropein. Take particular note of the structural similarity between these compounds ...... 34

Figure 4-3: Intensity ratings (group means) for ‘irritation’ and ‘bitterness’ from tyrosol (top), oleuropein (middle), and ibuprofen sodium (bottom). Tyrosol and oleuropein show essentially no response for either irritation or bitterness, suggesting the functional part of the oleocanthal molecule does not include the phenol group present on tyrosol. Ibuprofen acts as a positive control for both irritation and bitterness ...... 37

Figure 5-1: Chemical structures of the non-steroidal anti-inflammatory drugs (a) ibuprofen (b) naproxen and (c) acetaminophen. Ibuprofen and naproxen are COX-2 inhibitors, while acetaminophen acts via a different mechanism ...... 41

Figure 5-2: Intensity ratings (group means) for ‘irritation’, ‘bitterness’, and ‘sourness’ from acetaminophen (top), naproxen sodium (middle), and ibuprofen sodium (bottom). None of the stimuli demonstrated considerable sourness. Interestingly, naproxen and ibuprofen (which share structural similarities) each have higher irritation ratings than bitterness, while acetaminophen (which differs structurally) has much higher bitterness and very little irritation. This suggests different techniques for masking may be required ...... 44

Figure 6-1: The effect of milk fat on irritation (A) and bitterness (B) from ibuprofen. A significant effect of fat was seen for irritation between the skim and half-and-half samples, but no significant effects were seen for bitterness...... 56

Figure B-1: Ibuprofen standard curve used for quantification of ibuprofen remaining in the continuous phase of milk products after 24-hour equilibration ...... 72

viii

LIST OF TABLES

Table 2-1 List of irritation sub-qualities and definitions provided to participants during task orientation. Adapted from [2, 3]...... 16

Table 5-1: Physical composition of milk products. Top shows milk products presented to panelists 1-34, where bottom was presented to panelists 35-50. Two batches were required due to the length of the experiment and quality concerns for the products...... 47

Table B-1: Total Solids and fat content of milk samples used in the pilot experiment in Chapter 5...... 66

ix

ACKNOWLEDGEMENTS

To John and my committee: For always having your doors or inboxes open for my questions and concerns. I could not have asked for a more rewarding graduate experience and each of you have contributed to that in numerous ways. Thank you.

To Alissa, Nadia, and Meriel: I have learned so much from each of you. I truly would not have made it through this without our fights about each other’s accents, the quote board, random conversations on chat, and the never-ending supply of baked goods. Thanks for letting me be

"Mama Sam" when I really needed it.

To Mom and Dad: Thank you for always having high expectations of me, but still letting me call when I am walking home late at night or need to know where to take my dry cleaning. Most importantly, thank you for being living examples that our destiny is only what we make of our circumstances.

To Aaron: For supporting me from afar and reminiscing with me for hours when I was missing

Madison like crazy. Thank you for always forcing me to be the best person I can be, even when it seems too hard. I know this is the start of many more years of accomplishments together.

1

Chapter 1

The importance of chemesthesis in flavor

Published previously as

Bennett, S., and Hayes, J., Chemesthesis and Flavor, in The World of Food Ingredients.

February 2012, CNS Media BV.: Singapore.

Sidebar

In trying to precisely define and categorize human perceptions arising from food, the classical approach is to fall back on anatomy. Within this scheme, odors and aromas come from the nose, and tastes come from the mouth. But how then do we classify the burn of a nice curry, the tingle of carbonation, or the warmth of a tequila shot? Such sensations are fundamental to the experience of these foods and beverages, yet a physiologist (and Wikipedia) will tell you that the burn of chiles or the tingle of carbonation are not tastes. Why not? Because they are carried to the central nervous system (CNS) by non-taste nerves like the trigeminal, glossopharyngeal, and vagus nerves. Alas, this logic becomes rather circular, because taste nerves are defined as those that carry taste sensations, namely sweet, sour, salty, savory/umami, and bitter. One appealing solution is to redefine terms so that the common usage of ‘taste’ is equivalent to the technical term ‘flavor’, and not the narrow, technical definition of ‘taste’. Unfortunately, different segments of the food industry differ on the meaning of ‘flavor’. Some use flavor to refer the overall integrated perception of taste, touch and smell (as we do here), while others may use it to describe just the volatile odor active components of food, or the retronasal delivery of these volatiles. For

2 the remainder of this article, we will define the word flavor as referring to the integrated perceptual experience arising from taste, smell and touch input.

Alternatively, instead of falling back on anatomy, one can define perceptual systems in terms of their ecological purpose. In this scheme, the physiology becomes secondary to function.

Echoing earlier work by Gibson, Green recently extended this idea beyond smell and taste to include chemesthesis. Within this framework, sensations can be classified as ‘appetitive’ (e.g., salty and sweet help satisfy bodily need), or ‘chemofensive’ (i.e., toxin avoidance via bitter and irritating sensations) regardless whether the signal comes from gustatory, olfactory, or somatosensory nerves [4]. This scheme neatly avoids the false distinction between state (volatile versus non-volatile) or delivery route (through the nose versus through the mouth). Whether such a functional categorization system will gain traction is unclear.

Introduction

You’ve probably heard it before, “80% of what we taste is really smell”. This common statement is usually touted as an explanation for why nothing tastes good when you have a stuffy nose or why children pinch their nose to choke down an unpalatable food. The reality is that the story is much more complicated than this purported 80/20 split. If you have ever enjoyed the burn from a spicy meal or the tingle of a carbonated beverage, this fact should be readily apparent.

These sensations form a class of sensations beyond taste and smell that is best termed chemesthesis. While sometimes called ‘trigeminal sensations,’ other cranial nerves besides the trigeminal nerve (CN V) play a key role in transmitting these sensations. Pungent burning compounds are one major source of these sensations, although chemesthesis also includes menthol cooling, the tingle of soft drinks, and the buzz of Szechuan buttons, so irritation is also too limiting a term. In any case, these sensations make a critical contribution to the flavor of food.

3 Evolutionarily, our perception of the sweetness of sugar is thought to have developed to help us identify and ingest energy dense foods, while salty taste was important for finding and ingesting the physiologically required ion sodium [5]. Conversely, bitterness and chemesthesis are often associated with secondary metabolites that plants produce to prevent predation.

Interestingly, many of these molecules are associated with positive health effects in humans at appropriate doses (e.g. phytonutrients or phytochemicals) [6]. Thus, demand is increasing for new products that naturally contain or are fortified with these compounds. This in turn places renewed emphasis on the importance of chemesthesis and chemesthetic ingredients for the food manufacturer.

The term chemesthesis was coined to address issues that arose from using terminology such as ‘the common chemical sense,’ ‘trigeminal sensitivity,’ or simply ‘irritation’. The benefit of using chemesthesis as opposed to older terms is that it encompasses thermal, tactile, or painful sensations caused by stimuli without relying on a particular ‘sense’ to define the sensation [7].

Chemesthesis is fundamentally different from taste sensations (e.g., sweetness, bitterness), most notably due to the time course of sensation (both onset and dissipation) as well as adaptation and desensitization phenomenon. Tastes are perceived relatively quickly and disappear in a matter of seconds while burn can take a while to build and may not dissipate for minutes. Whether this is due to receptor access (e.g., depth in the tissue) or other biophysical properties (e.g., diffusion rates) is unclear. Chemesthesis also differs from smell and taste in regard to adaptation and desensitization.

We have all walked into an odor filled room like a bakery or locker room and noticed that the smell seems to disappear after a few minutes. If we leave and re-enter ten minutes later, the smell has returned. Physically, the concentration of odorant in the room hasn’t changed – instead, we have adapted. Smell and taste (under certain conditions) typically show this sort of adaptation, where perceived intensity declines over time with exposure. In contrast, chemesthetic agents

4 often exhibit sensitization, where repeated stimulation results in progressively stronger sensations. This can be readily observed when the salsa on your chips gets hotter with each bite.

However, this phenomenon also depends on the time course of stimulation. With a longer break between exposures, chemesthetic agents can show desensitization, where the perceived intensity is dramatically lower. Unlike taste adaptation, this desensitization can last days, potentially explaining why your friend the diehard chile-head can eat hot sauce that has you reaching for the garden hose. These unique temporal characteristics introduce methodological difficulties when designing sensory tasks for products containing chemesthetic ingredients like chiles, menthol, carbon dioxide (carbonation), mustard, and horseradish.

Methodological Concerns

Temporal factors in chemesthesis present two main issues. Depending on what type of information you are interested in, the decision of how and when to take intensity measurements of a quality can be critically important. If you are interested in the maximum intensity, the time required to reach maximum intensity, or the decay time, multiple measurements are needed to capture this time sensitive data. That is, using a single rating at single time point will probably miss the optimum point and provide misleading information. Time-intensity curves are commonly used as they describe intensity across time, either continuously or in discrete intervals, and provide resolution to study rate of onset, time to maximum, maximum intensity, and decay.

Figure 1-1 shows smoothed time intensity curves for cooling, mint flavor, and bitterness in two gum samples. The treatment gum included a novel compound that purportedly caused more cooling. What we see here is that the time-intensity curves were critical to get the whole story about how this ingredient would affect consumer perception. Had we taken a single time point rating after 60 seconds of chewing, we would have concluded that the treatment gum

5 provided about the same amount of cooling as the control, but had higher bitterness. With the time-intensity measurements, we see that as chewing progresses, the bitterness of the treatment gum declines to be no different from the control as cooling increases. These details are critically important when marketing a new product! This could be the difference between marketing claims of “long-lasting cooling” versus “a burst of cooling from the first chew”.

Differences in Gum Flavor Over Time (Treatment - Control) 10

9 Cooling Mint Flavor 8 Bitterness 7

6

5

4

3

2

1

0

−1 Mean Difference (Treatment - Control)- (Treatment MeanDifference −2

−3

−4

−5 0 30 60 90 120 150 180 210 240 270 300 Time (s)

Figure 1-1 Difference scores of time intensity curves for cooling, mint flavor, and bitterness of two gum samples. The control contains peppermint extract and the treatment contains a novel compound intended to increase cooling. While bitterness is higher in the treatment gum for the first two minutes of chewing, it decreases to be no different from the control as cooling peaks after 3 minutes.

The dissipation time of chemesthetic agents can be much longer than that of tastants, highlighting the critical need to consider potential carryover effects. Imagine testing three snack chip formulations in a consumer test. If the samples are presented too close together, without ample time to cleanse the palate or allow the sensation to subside, sensitization effects may

6 present in the data. That is, the perception of chile heat will be inflated in subsequent samples, which may impact liking scores. Allowing several minutes to ‘cool off’ between samples is often enough to solve this problem, but too much time can also be problematic. In particular, forced choice methods can become untenable if the required lag between samples is so high that the memory of the first sample fades. Long lag times can also become problematic in rating based tasks, if the lag results in desensitization. In this case, subsequent ratings would be lower than expected. In either case, the ratings may provide data that do not reflect real eating experience if the person eats faster or slower than in the laboratory.

Figure 1-2 Time-intensity ratings of capsaicin after exposure to a non-desensitizing stimulus (citric acid) and a desensitizing stimulus (capsaicin) daily for ten days. The graph shows that the group exposed to capsaicin showed significantly lower intensity ratings and faster recovery than the control group (Hayes, unpublished data).

Moreover, it is well known that individuals can differ greatly in their perception of chemesthetic agents with prior experience and exposure (Figure 1-2), and our laboratory is actively exploring whether basic physiological differences might also play a role. The existence

7 of such variation implies that no two consumers will experience the same food product in exactly the same way.

Research on the qualitative aspects of chemesthetic sensations is surprisingly understudied, especially since we all know that the burning of a chile pepper is much different than the burn of cinnamon or mustard [8]. In part, this lack of data is due to complications in applying traditional quantitative methodologies used to describe products or ingredients, as they rely on calibrating a panel of trained judges to use words or terms in common manner. Due to the unique temporal aspects and adaptation issues described above, the number of samples that can be presented in any one training session are limited. Also, training typically involves the use of reference compounds to standardize panelist language used to describe sensations. These reference compounds are usually perceptually simple – that is, they represent only a single taste or odor quality. For instance, solutions of sucrose may be provided to illustrate ‘sweetness’. This presents a problem for chemesthetic agents because few of them are strictly ‘burning,’ ‘stinging,’ or ‘numbing’ without having additional perceptions associated with them. Even the prototypical chemesthetic stimulus capsaicin suffers from this, as it can have a bitter side taste in many individuals [9]. Another issue is that the hydrophobic nature of many chemesthetic stimuli requires the addition of additional ingredients (like surfactants or emulsifiers) to keep them in solution. How these ingredients affect the perceptual quality of the stimulus is unknown and undesired when training panelists.

Finally, as seen in Figure 1-3, even with verbal or written descriptions of irritation qualities, panelists are confused about how to differentiate these sensations. Still, these perceptual differences between pungent compounds are critically important for determining how these ingredients will affect the liking and intake of foods formulated with chemesthetic compounds.

So, how do we proceed if the current methodologies are flawed?

8

Figure 1-3 Shown is a box plot of group intensity ratings for irritation qualities of olive oil. The data suggests that participants had a difficult time distinguishing between the qualities, as mean intensities are similar for all qualities. (Unpublished data from [10]).

Promising Techniques

Check-all-that-apply (CATA) techniques coupled with discrete-interval time-intensity ratings may be a useful way to collect semi-qualitative data. This enables us to supplement valuable time-intensity data by asking individuals to use a number of descriptors to describe what they are experiencing at that particular moment [10]. For example, if we were evaluating the heat intensity of a new curry, we might ask the panelist to rate the overall intensity and then indicate if they were feeling ‘burning,’ ‘warming,’ ‘stinging,’ and/or ‘tingling’. One could even include taste or odor qualities in the CATA task, but care must be taken to avoid overwhelming the participant with too many options. This task would then be repeated after every intensity rating and the frequency of the use of each descriptor can be measured. Note that this differs from the temporal dominance of sensation (TDS) approach because it does not assume that one sensation

9 overwhelms and masks all others. It may also be tempting to couple CATA approaches with a hedonic task (i.e. ask how much your participants liked each of the samples), but you’ll find that repeatedly switching between analytical and affective modes may be very difficult for participants and result in data that are compromised in both domains.

Conclusions

As the food culture in the United States shifts to incorporate spicy dishes from Latin

America and Asia, and more emphasis is placed on including healthful foods and ingredients in the diet, the importance of chemesthetic ingredients will only continue to grow. Unfortunately, many traditional methodologies that have been used to assess tastes or odors fall short when used with chemesthetic compounds due to the adaptation effects and unique temporal properties associated with them. New methods that appreciate the time-sensitive nature of these stimuli while incorporating some qualitative data, like CATA tasks, may help us move forward.

10

Chapter 2

Preface

The “Mediterranean Diet” has received a lot of attention over the last decade as a healthy counterpoint to the “Western Diet,” which has been implicated in the increased prevalence of obesity (e.g. [11]), cardiovascular disease (e.g. [12]), and type II diabetes (e.g. [13]) seen in affluent developed nations. The apparent health benefits of the Mediterranean Diet have been partially attributed to the consumption of virgin olive oils, both due to the favorable fatty acid profile and the biologically active phenolic compounds naturally present in virgin olive oils [14].

In particular, the discovery of oleocanthal, an endogenous phenolic compound in olive oil, has resulted in a flurry of research to better its unique characteristics, as oleocanthal not only has anti- inflammatory COX-2 inhibition in vivo [15], but is also responsible for the distinct throat pungency from high quality oils [16, 17]. These studies include attempts to characterize its sensory characteristics [10, 16], receptor mechanisms [1], and biological functions (e.g. [18]).

The oleocanthal story started with an interesting perceptual connection that was made in the early 2000s when Gary Beauchamp, the Director of the Monell Chemical Senses Center, tasted freshly pressed extra virgin olive oil for the first time outside the context of food. The freshly pressed oil caused localized throat pungency very similar to the pungency caused by liquid formulations of the non-steroidal, anti-inflammatory drug (NSAID), ibuprofen, which he and his colleagues had studied previously. In their 2001 paper, they describe the time course and qualitative aspects of oral sensations from ibuprofen, including evidence that the irritation is localized to the throat [3], which is atypical for most irritants. This perceptual similarity

(confirmed in 2009 when the sensory qualities of oleocanthal were verified experimentally [16]),

11 led Beauchamp and his colleagues to test the pharmacological properties of oleocanthal to determine if they were similar to those of ibuprofen, namely COX-2 inhibition. What they found was at equivalent doses in vitro, oleocanthal is a more potent COX-2 inhibitor than ibuprofen, but is also a COX-1 inhibitor, like aspirin. Because oleocanthal is non-selective for COX-2 inhibition, it may show some of the side effects common in aspirin use (i.e. gastrointestinal ulcers or bleeding), but this has yet to be determined.

Due to the potential health benefits of oleocanthal consumption (e.g. “an aspirin a day”), a number of experiments were carried out to determine the mechanism of action of oleocanthal and ibuprofen, including their sensory receptor [1]. This was particularly important work because, at this time, oleocanthal and ibuprofen were the only chemesthetic stimuli known to cause localized irritation. Other oral irritants like capsaicin, piperine, and cinnamaldehyde are known to burn throughout the oral cavity (e.g. [19, 20]) and elsewhere.

Oleocanthal and ibuprofen were shown to selectively activate the human nociceptor, transient receptor potential cation channel, member A1 (TRPA1) in HEK 293 cells, in addition to exciting only those rodent sensory neurons that were expressing functional TRPA1 [1].

Information about other TRPA1 agonists such as cinnamaledehye and allyl isothiocyanate from foods and environmental pollutants like crotonaldehyde and acrolein (e.g. [21-24]) was available from recent advances in molecular biology and human and animal behavioral studies. The !, "- unsaturated aldehydes (e.g. cinnamaldehyde) and electrophillic TRPA1 agonists (e.g. allyl isothiocyanate) have been shown to activate TRPA1 via a mechanism that involves the covalent modification of cysteine residues on the N-terminus of the protein, while the mechanism for non- electrophillic agonists (e.g. carvacrol) is yet to be determined [21]. To rule out this mechanism for oleocanthal (despite it also containing an !, "-unsaturated aldehyde) and ibuprofen, Beauchamp and colleagues conducted a study to measure the response of mutant TRPA1 channels (containing serine residues in place of the reactive cysteines) to oleocanthal, ibuprofen, cinnamaldehyde, allyl

12 isothiocyanate, and a non-electrophillic TRPA1 agonist, 2-aminoethoxydiphenyl borate (2-APB).

The response to cinnamaldehye and allyl isothiocyanate was obliterated, while response for oleocanthal, ibuprofen, and 2-APB did not change between normal and mutant TRPA1, suggesting an alternative mechanism of action [1]. In addition to this alternative mechanism, the authors suggest that oleocanthal and ibuprofen are unique in their selectivity for TRPA1, unlike other agonists that are known to activate more than one TRP receptor. This evidence, coupled with immunohistochemistry from human tissue implicating that TRPA1 expression is restricted to the throat in humans [1], suggests that despite structural dissimilarity, oleocanthal and ibuprofen elicit this distinct pungency through a TRPA1-dependent mechanism. What that mechanism may be remains to be determined.

Another interesting point raised by the perceptual connection between oleocanthal and ibuprofen is the idea that the sensory properties of a compound may be able to predict its function or drug potency in the body. This was an idea first proposed by Fischer in the mid-1960s, suggesting that taste sensitivity to the bitterness of sodium saccharinate (measured by Weber ratio) could predict the intensity of drug-effects from Psiocybin (a potent psychedelic drug being used to treat schizophrenia) [25]. While this may be an attractive theoretical approach, true applicability (using human bioassay to predict drug potency) will likely never gain favor due to safety and time concerns. This idea can be useful, though, when considering flavor masking strategies for pharmaceuticals, as similar flavor profiles will require similar masking techniques.

This implies that if we work backward and relate known biological function (e.g. NSAIDs vs. anti-histamines) to specific and distinct flavor profiles, we can target masking technologies to a specific drug class.

Despite published methods for the synthesis of oleocanthal [26], a commercial source is still not readily available. As such, studies of its behavior are severely limited, particularly because extra virgin olive oil is an extremely complex stimulus [10]. What these studies have

13 rekindled, though, is an appreciation for the complexity involved in testing chemesthesis and bitterness behaviorally in humans. Additionally, studies of ibuprofen may serve a dual function, first to make generalizations back to oleocanthal, but also to expand our understanding of the sensory characteristics of pharmaceuticals for the study of flavor masking technologies.

Our approach here was to fill in some of the gaps from the existing literature on oleocanthal and ibuprofen, as well as test a partitioning hypothesis for applicability in flavor masking of oral pharmaceuticals.

14

Chapter 3

Chemesthetic sub-qualities of capsaicin, ibuprofen and olive oil

Published previously as

Bennett, S. M. and J. E. Hayes (2012). "Differences in the Chemesthetic Subqualities of

Capsaicin, Ibuprofen, and Olive Oil." Chemical Senses. doi: 10.1093/chemse/bjr129

Chemesthetic sensations elicited by ibuprofen, extra virgin olive oil and capsaicin were compared to quantify perceptual differences between known agonists of TRPA1 and TRPV1.

Extra virgin olive oil contains a phenolic compound, oleocanthal, which is thought to share unique chemesthetic qualities with the non-steroidal anti-inflammatory drug, ibuprofen. Pilot work suggested participants had difficulty distinguishing between multiple chemesthetic sub- qualities (e.g. burn, sting, itch, tickle, etc) in a multi-attribute rating task. Here we assessed overall irritation via direct scaling, and a check-all-that-apply (CATA) task was used to collect information about chemesthetic sub-qualities over time. Replicated ratings were collected at discrete intervals using the generalized labeled magnitude scale (gLMS) to generate time- intensity curves; maximum intensity (Imax) and area under the curve (AUC) were extracted for each participant. Intensity responses varied substantially across participants, and within a participant the relationship was strongest between ibuprofen and olive oil. However, there were also positive, albeit weaker, correlations between capsaicin and ibuprofen and capsaicin and olive oil. The correlation found between olive oil and capsaicin may suggest the presence of unknown

TRPV1 agonists in olive oil. This view was also supported by the qualitative data: capsaicin was

15 described most often as burning and warm/hot, whereas ibuprofen was numbing and tickling.

Olive oil shared characteristics with both capsaicin (warm/hot) and ibuprofen (tickle).

Introduction

The quality of extra virgin olive oils is primarily assessed using a standardized procedure developed by the International Olive Council (IOC). This method uses trained panelists to detect and rate bitter, pungent, and fruity characteristics of olive oils to establish their commercial grade

[27]. Interestingly, pungency and bitterness are considered positive qualities by experts (but not consumers [27]), as they speak to the composition and concentration of the phenolic fraction in the oil. This desirable pungency is attributed to the phenolic compound oleocanthal (‘oleo’ for olive, ‘canth’ for sting, and ‘al’ for aldehyde) [15, 17]. Despite structural dissimilarity, oleocanthal has been shown to have similar sensory and pharmacological properties to the non- steroidal anti-inflammatory drug (NSAID) ibuprofen (Figure 3-1). Both compounds cause a distinct pungency that is restricted to the throat [15]. This is peculiar because most chemesthetic stimuli elicit irritation throughout the oral cavity and other mucosal tissues, although the intensity of sensation may differ by site [19, 28].

16

(a)!

(b)!

(c)!

Figure 3-1 Chemical structures of (a) the non-steroidal anti-inflammatory drug ibuprofen (b) (!)- deacetoxy-dialdehydic ligstroside aglycone [oleocanthal] and (c) capsaicin, a known agonist of the TRPV1 channel.

Because of the cardioprotective effects of olive oil in the Mediterranean Diet, oleocanthal bioactivity has received substantial attention as an alternative causative mechanism, as cardioprotection might arise from beneficial anti-inflammatory action rather than a favorable fatty acid profile [14].This is because oleocanthal and ibuprofen have similar pharmacological effects.

Both are COX-1 and COX-2 inhibitors, and oleocanthal’s IC50 value for COX-2 inhibition is nearly ten times lower than that of ibuprofen [15], suggesting oleocanthal may be more potent at equivalent concentrations. Due to these potential health benefits, efforts have been made to determine how individuals differ in their perception of oleocanthal [16] and to explore the extent of its protective effects (e.g. [18]). However, no published work to date has compared the sensory properties of oleocanthal and ibuprofen behaviorally in the same population.

Here, we quantified the overall intensity of oropharyngeal irritation elicited by extra virgin olive oil and ibuprofen as well as their predominant and secondary irritation sub-qualities;

17 capsaicin was also included as a control stimulus. We reasoned that if oleocanthal and ibuprofen share a common receptor, they should be described by the same sub-qualities and the intensity of the irritancy elicited by these compounds should be correlated with each other across individuals.

Indeed, the localized irritation from oleocanthal and ibuprofen has been attributed to the specificity of these compounds for the TRPA1 receptor [1]. Moreover, if that mechanism is independent of TRPV1, the irritancy from olive oil and ibuprofen should not correlate with the irritancy elicited by the prototypical TRPV1 agonist capsaicin, similar to the independence of bitterness observed for tastants that act via different receptors (e.g. [29]).

The applied sensory science literature is replete with reliable and consistent methods that can be used to collect quantitative data on the qualitative differences between stimuli. Such methods often involve calibrating a panel of trained judges and using these judges as analytical sensors. Unfortunately, these techniques are not well suited to chemesthetic stimuli. These limitations have been described in detail elsewhere [30]. Briefly, the training of panelists is extremely difficult due to issues of sensitization, desensitization, and the time course of sensation that limit the number of samples that can be presented in a given session (e.g. [31, 32]). Also, with tastants, reference compounds that are strictly sweet, sour, bitter, salty, or umami exist and are easy to prepare for training. Many chemesthetic stimuli are hydrophobic and so require preparation with ethanol and surfactants, which may or may not affect the integrity of the stimulus. Additionally, we and others have found that the diffuse nature of chemesthetic sensations makes them hard to characterize [20]. Even with verbal or written descriptions of irritation sub-qualities (e.g. [2]), it seems that panelists find it difficult to associate a description with an actual oral sensation. Thus, it becomes a circular problem: meaningful qualitative data is difficult to collect without training and training is difficult to perform due to the nature of the stimuli and lack of training standards (reference compounds). Nonetheless, behavioral work is critical to determine how our perceptions are ultimately related to recently elucidated molecular

18 mechanisms, as it provides a context for the study of natural ligands found in foods. Because a trained panel approach is not feasible, novel approaches for collecting qualitative or semi- qualitative data on irritants are required to move forward in this area.

In a pilot study that aimed to quantify and correlate multiple oral irritation sub-qualities from capsaicin, olive oil, and ibuprofen over time, we found the magnitude of irritancy from olive oil was correlated with ibuprofen within individuals. However, there was also a positive, albeit weaker relationship between olive oil and capsaicin irritancy, implying that olive oil may also contain an unknown TRPV1 agonist. Additionally, scatter matrices of individual responses across the sub-qualities suggested the task might have been overly complex for participants. Contrary to the dumping effects that are well documented in time intensity studies for classical tastants and odorants (e.g. [33, 34]), we found the opposite: a ‘smearing’ bias. We suspect that when untrained participants struggle to distinguish between different types of irritancy (e.g. burning, tickle, itch), they spread their ratings across multiple attribute scales. Whether this resulted from providing too many different response options (seven) or the unfamiliar nature of the chemesthetic sensation was unclear. It was also evident that there could be a response bias, as the quality ratings occurred in a fixed order (e.g. [35]), with ‘burn’ always occurring first. Thus, more work was needed to further explore the qualitative aspects of the irritancy from olive oil, ibuprofen and capsaicin. In the simplified task described here, participants rated overall irritation intensity, and then indicated the predominant sensation and levels of each of the sub-qualities used in the pilot study. Also, the order in which the sub-qualities were presented was pseudo-randomized to avoid any order bias for the first and last positions.

Our primary goal was to compare the irritation (i.e. chemesthetic sub-qualities) of capsaicin, ibuprofen and olive oil within the same group of individuals. A secondary goal was to explore individual differences in irritation intensity. It was not the purpose of this experiment to elucidate the receptor mechanisms of ibuprofen and oleocanthal, but instead to supplement the

19 current molecular data, by providing behavioral evidence that the chemesthetic nature of olive oil and ibuprofen is distinctive from other more commonly studied irritants.

Materials and methods

Subjects

Reportedly healthy, non-smoking adults (n=37; 10 men; aged 18-45 years) were recruited from the Penn State community. Procedures were IRB approved, informed consent was obtained, and participants were paid for their time. All data were collected in a one-on-one setting at the

Sensory Evaluation Center at Penn State.

Stimuli

The test stimuli were 5 mL samples of 2.5% (w/v) (121.2 mM) USP grade ibuprofen

(Spectrum, CAS# 15687-27-1), 75 mg/L (~0.246 mM) natural capsaicin (65% capsaicin/35% dihydrocapsaicin, Sigma Aldrich, CAS# 404-86-4), and commercially available extra virgin olive oil (Fruttato Colavita) held at 35°C and presented in 30 mL plastic medicine cups. Solutions of ibuprofen and capsaicin were prepared in canola oil (Wegman’s, State College, PA), as canola more closely mimics the fatty acid composition of olive oil (compared to corn oil). Stimuli concentrations were selected from previously published reports [3, 36] and revised based on intensity data from the pilot study to ensure approximately equal irritation intensity. All samples were presented in randomized order, and labeled with random 3-digit blinding codes.

20 Procedure

Participants were asked to refrain from eating and the use of chemesthetic agents (i.e. toothpaste, mouthwash, spicy food) for at least two hours prior to their session. Before beginning the test, participants were oriented to the generalized Labeled Magnitude Scale (gLMS) [37] using a list of 15 imagined or remembered sensations that included both oral and non-oral items

(Hayes, Allen & Bennett, under review). Scale instructions encouraged participants to make ratings in a generalized context by indicating that the top of the scale should reflect their

‘strongest sensation of any kind’. (The modifier ‘imaginable’ is not needed to generalize the scale; see discussion in [37]). During training, participants were also introduced to a list of seven irritation sub-qualities and their definitions (Table 3-1). The list of sub-qualities and definitions was visible to participants throughout the entire test.

Table 3-1 List of irritation sub-qualities and definitions provided to participants during task orientation. Adapted from [2, 3].

Irritation sub-qualities and definitions Burning the sensation that commonly results from exposure to very high temperatures (i.e. thermal burns), skin abrasions (e.g. run or floor burns) or chemical irritants (e.g. alcohol); may or may not be accompanied by a thermal sensation

Stinging/Pricking sharp sensations similar to those produced by an insect bite— other than itching—or a pinprick; may be constant (stinging) or brief (pricking)

Itching the sensation that provokes the desire to scratch

Tingling a lively pins-and-needles sensation

Warm/Hot sensations of mild (warm) or extreme (hot) heating

Numbness the diffuse (e.g. ‘fuzzy’) sensation produced during the onset of an anesthetic (e.g. novocaine); it is NOT the complete absence of sensation

Tickle the sensation in the back of the mouth or throat that when weak causes the urge to clear the throat and when strong causes coughing

21

To evaluate irritation localized to the throat, the stimulus delivery method was based on two previously published reports [15, 16]. Briefly, participants were instructed to place the 5 mL oil sample in their mouth and tilt their head back to allow the oil to reach the throat. Then, they were instructed to allow the oil to sit at the back of the throat for 5 seconds before swallowing in two stages (swallowing, then immediately swallowing again). Swallowing in two stages purportedly ensures that the stimulus is distributed to the whole surface of the throat. This method is designed specifically to localize the stimulus exposure to the throat and minimize contact in the rest of the oral cavity. The participant’s first rating was made immediately after the second swallow. Discrete-interval time-intensity ratings for ’Overall irritation in your throat’ were collected every 30 seconds for 180 seconds using Compusense® five (Guelph, ON, Canada).

Participants were asked to keep the sub-qualities of irritation in mind while they rated. Then, before rinsing with water, participants indicated their ‘Predominant Sensation’ from the list of seven sub-qualities and endorsed each sub-quality as ‘No Sensation,’ ‘Low,’ or ‘High.’ After rating, participants were allowed to rinse with 35°C RO water ad libitum, but were asked to sit quietly with their mouth closed to maintain a constant temperature in the mouth. A minimum inter-stimulus interval (ISI) of an additional 180 seconds was enforced between each sample. If a participant had any residual irritation at the end of the 3 min ISI (6 min after initial sample presentation), they were given more water and asked to wait until all irritation had subsided before proceeding to the next sample. A total of six stimuli (3 samples x 2 replicates) were presented within a single session (~45 min). When designing the experiment, we considered issues of participant fatigue and desensitization/sensitization that would arise from presenting six samples within a session. However, we decided against splitting testing across days, both because we were worried about the increased variability that would arise across multiple days, and because prior reports indicate ibuprofen does not sensitize or desensitize across trials [3]; since

22 they act on the same receptor, we presume oleocanthal response similarly does not desensitize.

Moreover, even if some small degree of sensitization or desensitization occurred for capsaicin, the presentation order was counterbalanced, so this would only introduce noise and not a systematic bias into our data.

Data Analysis

All analyses were conducted in SAS 9.2 (Cary, NC). To characterize the qualitative aspects of each stimulus, the number of endorsements for predominant quality for all replicates were summed and expressed as a percentage of total possible responses (i.e. 2 reps x 37 people = 74 total responses for a descriptor). Because our secondary hypothesis involved comparing individual differences between participants across compounds, we extracted two scaffolding parameters from the discrete interval time intensity functions: the maximum intensity

(Imax) and the total area under the curve (AUC). The AUCs were calculated using the trapezoid rule [38]: the sum of a series of isosceles trapezoids, each with an area equal to [(height1 + height2) ⁎ width]/2. These summary parameters (Imax and AUC) were compared across stimuli using linear regression. Additionally, replicate means at each time point were analyzed via repeated measures mixed model ANOVAs.

Results

Individual differences in perception were seen for responses to capsaicin, olive oil, and ibuprofen irritation (not shown), consistent with previous reports [3, 16, 39] and our pilot work.

Group means of overall irritation were used to evaluate the relationships between overall temporal patterns of each stimulus across the group. These results are shown in Figure 3-2. As

23 expected, in two-way (stimulus by time) repeated measures ANOVA, there was a significant main effect of time [F (12,1296) = 193.8, p<0.0001]. There was also a significant interaction of stimulus by time [F (24,1296) = 19.1, p<0.0001]; pairwise comparisons (Tukey-Kramer) revealed capsaicin ratings were higher than ibuprofen and olive oil at all time points (all p’s <0.0001) while olive oil and ibuprofen ratings never differed from each other (all p’s >0.1). Thus, although the olive oil and ibuprofen irritation levels were successfully brought into the same range, the capsaicin concentration was still a little high, resulting in an Imax that was closer to ‘very strong’ than just above ‘moderate’ for the other two stimuli. Visual inspection of the curves also suggests the irritation from olive oil and ibuprofen decays at a faster rate than the capsaicin irritation, particularly in the first 110 s.

24

Figure 3-2 Intensity ratings (group means with standard errors) for ‘overall irritation’ in the throat for each stimulus. Capsaicin was significantly different from olive oil and ibuprofen at all time points, while olive oil and ibuprofen did not differ significantly at any time point. Abbreviations on the right axis correspond to labels given on the gLMS presented to participants: BD= Barely Detectable, W=Weak, M=Moderate, S=Strong, and VS=Very Strong.

Linear regression was used to compare individual Imax and AUC values for each stimulus (Figure 3-3). Unexpectedly, significant correlations were seen between the Imax ratings of all three stimuli. Nonetheless, the effect size for the ibuprofen-olive oil relationship (r = +0.60, p <0.0001) was larger than the ibuprofen-capsaicin relationship (r = +0.34, p = 0.038). The size of the capsaicin-olive oil relationship fell in between these values (r = +0.45, p = 0.005), suggesting olive oil may contain an unknown TRPV1 agonist. Analysis of the AUC values (not shown) revealed similar results: ibuprofen–olive oil (r=0.72; p < 0.001), ibuprofen–capsaicin

(r=0.37; p=0.025) and capsaicin–olive oil (r=0.49; p=0.002).

25

A

B

Figure 3-3 Correlations between Imax values for overall irritation for each stimulus. (A) Ibuprofen correlated with both olive oil (p <0.0001) and capsaicin (p=0.038). (B) Capsaicin was also correlated with olive oil (p=0.005). Similar results were seen for the AUC values (see text).

26 Frequency counts (expressed as percentages of total responses) were plotted to show the relative relationship of the sub-qualities described as the ‘predominant sensation’ for each stimulus (Figure 3-4). While it is clear from this plot that the simplified CATA task was successful in pulling apart the irritation profiles of the three stimuli, the plot also suggests capsaicin is a much cleaner stimulus than ibuprofen or olive oil. By ‘cleaner’, we mean that that there was general agreement that the predominant sub-quality of capsaicin was captured by two perceptually similar sub-qualities. Figure 3-4 shows the tendency of capsaicin to be described predominantly by ‘burning’ and ‘warm/hot’, and ibuprofen as evoking ‘numbness’ and ‘tickle’.

The qualitative aspects of extra virgin olive oil show similarities to both capsaicin (warm/hot) and ibuprofen (tickle), as would be expected if olive oil contains both TRPA1 and TRPV1 agonists.

Figure 3-4 Percentages for the number of times each of the sub-qualities was described as the ‘predominant sensation’ for the irritation from a particular stimulus. Note that this measure does not directly reflect the intensity of the sensations. Capsaicin was rated as burning 56% of the time, but a disjointed scale is used to highlight differences between the other sub-qualities.

27 Discussion

General Findings

In this laboratory-based study of adults, we confirm that olive oil and ibuprofen irritancy vary across individuals. We extend prior work by demonstrating that response to both covaries in vivo –individuals who show reduced sensory response to ibuprofen also tend to show reduced response to olive oil. Consistent with the idea that oleocanthal and ibuprofen act via a capsaicin independent mechanism, we found that the strength of the relationship between ibuprofen and olive oil was stronger than the relationships between these stimuli and capsaicin. Unexpectedly however, we did find a positive relationship between olive oil response and capsaicin, suggesting olive oil contains an unknown TRPV1 agonist. This is also supported by qualitative data on the perceptual sub-qualities of each stimulus, as olive oil shared attributes of both capsaicin and ibuprofen.

The anticipated predominant irritation sub-quality for ibuprofen was sting, itch, or tickle

[3]. In contrast, capsaicin is typically associated with burning and warming (e.g. [19]), although side tastes like bitterness have also been reported [9, 19]. These qualitative differences in vivo are consistent with in vitro data showing capsaicin and ibuprofen activate TRPV1 and TRPA1, respectively [1, 40]. We anticipated that olive oil would therefore show a response pattern similar to ibuprofen as the irritancy of olive oil has been attributed to oleocanthal, a known TRPA1 agonist. Instead, we found that the qualitative aspects of olive oil were intermediate between ibuprofen (predominantly tickling, and tingling and numbing) and capsaicin (burning, hot/warming). This suggests that although oleocanthal may be the major source of irritancy in olive oil [16, 17], it may not be the only one. Indeed, homovanillic and vanillic acids, as well as a number of structurally similar compounds, have been identified in the minor phenolic fraction of

28 extra virgin olive oils [41]. The mere presence of these vanilloids in olive oil is not sufficient to conclude they actively contribute to the pungency of olive oil, as they may be present at levels well below human detection thresholds (i.e. below the ‘window of perception’), but it is not unreasonable to imagine they may play a secondary role considering how little oleocanthal or capsaicin are required to initiate a sensory response.

As such, the presence of vanilloids in olive oil likely accounts for the correlation between the irritation elicited by capsaicin and olive oil observed here. This relationship was initially unexpected, as we had hypothesized there would be no association between capsaicin response and the other two stimuli, due to the ubiquity of TRPV1 mediated irritation in contrast to the locus specific nature of oleocanthal and ibuprofen irritancy. Previous human work had shown olive oil irritancy did not covary with the irritancy from carbon dioxide [16], a known TRPA1 agonist [42]. Subsequently, oleocanthal was shown to activate TRPA1, at least in rat trigeminal ganglion [1]. The lack of a correlation between carbon dioxide and olive oil evoked sensations in humans in spite of a common receptor can be explained in one of three ways. It may be that carbon dioxide is promiscuous, activating more than one TRP receptor. Alternatively, it may have a mechanism that is more complicated than a single agonist-receptor relationship [43]. Finally, it may reflect the idea that oleocanthal acts on TRPA1 via a unique mechanism that is not shared with other known TRPA1 agonists like allyl isothiocyanate or cinnamaldehyde and is somehow specific to channels present in the throat [1]. The trigeminal nerve appears capable of expressing

TRPA1, as pure oleocanthal burns in the nose [1], so the reason for the lack of burn in the mouth is unclear.

Although the most parsimonious explanation for the capsaicin-olive oil correlation is the presence of vanilloids in olive oil, another potential explanation would be coexpression of

TRPA1 and TRPV1 in vivo. In rodent trigeminal neurons, coexpression of TRPV1 has been reported in 100% of TRPA1 expressing neurons [22, 23]. Thus, differences in the number of

29 neurons that dually express TRPV1 and TRPA1 across people could also account for the correlation between capsaicin and the other two stimuli. Additionally, TRPV1 exclusive neurons may be thermospecific labeled lines that transduce sensations associated with heat pain, whereas

TRPA1 acts as a more diffuse generalized chemofensor system intended to notify the body of chemical toxins. This later role is usually attributed to bitterness [44] although evolutionarily functional similarities between bitterness and chemesthesis have been discussed previously [45].

This would also help explain the difficultly our participants had in characterizing a specific percept (sub-quality) from the ibuprofen and olive oil.

Using time-intensity data alone to explore agonist-receptor relationships can be complicated by biophysical factors. In contrast to single timepoint ratings, generation of time intensity curves is more influenced by tissue access and retention of the compounds, as well as cellular adaptation. Each of these becomes even more complicated when more than one stimulus is presented within a session. Differential access and retention can result from the compound’s lipophilicity; the degree of lipophilicity is often expressed as Log P. Reported Log P values for capsaicin, ibuprofen, and oleocanthal are 3.8 [46], 3.5 [47], and 1.5 [48] respectively. Log P values greater than 1 indicate that a compound is more lipophillic than hydrophillic. In this experiment, stimuli were presented in oil, so highly lipophillic molecules may be less likely to partition out of the lipid matrix and into the aqueous salivary environment. This may influence how and, more critically when these molecules reach their receptors.

Limitations and Conclusions

Here, a natural product, extra virgin olive oil, was used instead of pure oleocanthal in solution. This limits our ability to speak directly to the nature of percepts arising from oleocanthal

30 in humans. However, this also increases our generalizability toward real foods and thus dietary habits and ingestive behavior.

This work also provides qualitative perceptual data that can only be obtained behaviorally. In contrast to dumping that typically occurs when response options are overly restricted, pilot work revealed evidence of a smearing bias where participants struggled somewhat to characterize the sub-qualities arising from ibuprofen. Whether this might be improved with some sort of panelist training, or reflects the diffuse nature of the ibuprofen/oleocanthal percept is unclear. Previously, reduced intensity ratings (compression) have been observed in rating tasks when too many response options are provided to participants [49]. We do not anticipate this is the case with the check-all-that-apply task used here, but more work is needed to confirm that CATA approaches are more robust in this respect. Here, capsaicin was found to be predominantly but not exclusively burning and warming and ibuprofen was numbing and tickling; olive oil was intermediate, sharing qualities with each.

In summary, both qualitative and quantitative data suggest olive oil contains vanilloid compounds that actively contribute to the perceived irritancy of olive oil. We confirm that olive oil and ibuprofen irritancy each vary across individuals, and demonstrate that this variable response covaries in vivo.

31

Chapter 4

Structure-activity relationships between olive oil phenolics:

tyrosol, oleuropein, and oleocanthal

Introduction

Classically, structure-activity studies have been employed for purposes such as predicting the boiling point of newly synthesized organic molecules or to predict the toxicity of a novel drug in the body (e.g. [50, 51]). When applying this type of approach in chemosensory research, animal bioassays are often used to predict human sensory response. For instance, studies of airborne pollutants and irritants showed correlation between decreased rodent respiration rate and eye, nose, and throat irritation in humans, a relationship that led to the development of the human

Threshold Limit Value (TLV). TLV is the amount of a pollutant to which a worker can be chronically exposed to without adverse health effects (e.g. [52, 53]). This idea has also been used previously for agonists of the transient receptor potential cation channel subfamily V member1

(TRPV1). Experiments using TRPV1 knockout mice indicate that the channel is essential for transducing nociceptive, inflammatory, and hypothermic effects of vanilloid compounds; the channel also plays a role in acute thermal nocicepton and thermal hyperalgesia following injury

[54]. TRPV1 agonists have shown promise as analgesic drugs due to their ability to desensitize

TRPV1 to nociceptive stimuli. Unfortunately, many potentially therapeutic TRPV1 agonists also elicit the characteristic pungency associated with capsaicin and piperine (the ‘spicy’ components of chili pepper and black pepper, respectively). Rodent behavioral bioassays that quantify pungency (through eye swipes), coupled with calcium flux studies or other measures of TRPV1

32 activation have been critical in identifying novel compounds that may provide clinical efficacy without undesirable pungency [55].

In humans, to fully understand structure-function relationships, it is not sufficient to stop at receptor activation (e.g. calcium imaging), as receptor cell depolarization is just the first step in the process of perception. A half century ago, Fischer proposed that taste activity, specifically bitterness, could be used as a behavioral assay of pharmacological activity [25]. This idea gained new traction with the identification of a novel anti-inflammatory compound in olive oil called oleocanthal, as this compound was shown to be responsible for the pungency associated with high quality extra virgin olive oils [15, 17].

Recently, Breslin, Beauchamp, and colleagues conducted a structure-activity study [1] with synthesized oleocanthal (both + and – enantiomers) and 10 derivatives. Among the derivatives, 5 compounds kept the tail end of oleocanthal intact and changed or removed the hydroxyphenolethanol group (Figure 4-1 Left) and 5 compounds modified the tail end of the molecule, conserving the hydroxyphenolethanol moiety (Figure 4-1 Right).

33

Figure 4-1 Figure taken from [1] showing structural analogs of oleocanthal created to assess stimulation of TRPA1 in rat trigeminal neurons.

The findings from the Breslin study indicated that only the compounds that conserved both aldehyde groups and the double bond (compounds A1-A7, above) on the tail end of the molecule were able to activate TRPA1 in rat trigeminal neurons. TRPA1, like TRPV1, has been shown to be involved in chemosensation and pain transduction as well as the inflammatory action of environmental pollutants and cough (e.g. [22, 56, 57]). Selective activation of TRPA1 by oleocanthal has been shown to correlate with the unique irritation profile of oleocanthal. Here, we aimed to confirm that the relationship between the presence of the aldehyde groups and activation of TRPA1 holds behaviorally in humans. This was achieved by testing irritation and bitterness from tyrosol and oleuropein, compounds that share structural similarities to oleocanthal (Figure

4-2). Our hypothesis was that tyrosol and oleuropein presented at levels naturally occurring in

34 extra virgin olive oil would not cause irritation (as they do not have the free aldehydes present on oleocanthal), confirming that they do not activate TRPA1 like oleocanthal.

a)

b)

c)

Figure 4-2 Chemical structures of phenolic compounds from extra virgin olive oil (a) oleocanthal (b) tyrosol and (c) oleuropein. Take particular note of the structural similarity between these compounds.

Materials and methods

Subjects

Reportedly healthy, non-smoking adults (n=14; 5 men; aged 18-45 years) were recruited from the graduate student population in the Food Science Department at Penn State. Procedures were approved by the local Institutional Review Board, informed consent was obtained, and

35 participants were paid for their time. All data were collected in a one-on-one setting with the first author at the Sensory Evaluation Center at Penn State.

Stimuli

Aqueous solutions of tyrosol and oleuropein were prepared to reflect concentrations naturally occurring in extra virgin olive oils [58]. This was 20 mg/L for tyrosol

(hydroxyphenolethanol, Fluka, CAS: 501-94-0) and 5 mg/L for oleuropein glycoside (Sigma,

CAS: 32619-42-4). 1% (w/v) ibuprofen sodium (Fluka, CAS: 31121-93-4) was presented as a positive control.

Procedure

Prior to testing, the task was described to participants and they were oriented to a generalized Labeled Magnitude Scale (gLMS) [37] that would be used to indicate the intensity of

“bitterness”, and “irritation” resulting from the stimuli. Participants were asked to rate 15 imagined or remembered sensations (both oral and non-oral items) on the scale (Hayes, Allen, and Bennett, under review). Instructions encouraged them to make these ratings in a generalized context by using the top of the scale to indicate their ‘strongest sensation of any kind’.

Participants were asked to refrain from eating and the use of chemesthetic agents (i.e. toothpaste, mouthwash, spicy food) for at least two hours prior to testing. Both scale orientation questions and test questions were presented to the participant in the Plus module of Compusense five, version 5.2 (Guelph, ONT).

Discrete interval time-intensity ratings for “bitterness” and “irritation” were collected for all stimuli every 15 seconds for 60 seconds. Intensity measurements were only taken for 60

36 seconds because our main interest in these studies was to determine the maximum intensity and relative contributions of these sensations and not necessarily their full time course. This also reduced session time (approximately 20 minutes) and participant fatigue. A total of six samples were presented per session (3 stimuli x 2 replicates). A minimum inter-stimulus interval (ISI) of an additional 90 seconds was enforced between each sample. If a participant had any residual irritation at the end of the ISI, they were given more water and asked to wait until all sensations had subsided before proceeding to the next sample.

Results and Discussion

Group means were calculated for each quality, at each time point, and plotted to show temporal patterns of bitterness and irritation from the stimuli (Figure 4-3).

37

Figure 4-3 Intensity ratings (group means) for ‘irritation’ and ‘bitterness’ from tyrosol (top), oleuropein glycoside (middle), and ibuprofen sodium (bottom). Tyrosol and oleuropein show essentially no response for either irritation or bitterness, suggesting the functional part of the oleocanthal molecule does not include the phenol group present on tyrosol. Ibuprofen acts as a positive control for both irritation and bitterness. Abbreviations on the right axis correspond to labels given on the gLMS presented to participants: BD= Barely Detectable, W=Weak, M=Moderate, and S=Strong.

Both tyrosol and oleuropein showed little to no sensory response (bitterness or irritation) for most panelists at the concentrations presented. Group means were at or below 6 on the gLMS, which corresponds to ‘weak’. Even this high of a rating may be due to a phenomenon known as

‘end use avoidance’ where participants are reluctant to use the extremes of the scale due to preconceived notions about the nature of the study (see [59]). Unfortunately, this result may be

38 due to the fact that the tyrosol and oleuropein levels in olive oil fall below their detection threshold (published values are not available in the literature, and we were unable to determine them experimentally as we did not have ethics board approval to increase the concentrations beyond those naturally found in olive oil.) Regardless, this confirms our hypothesis that these compounds do not contribute individually to the pungency of extra virgin olive oils. However, additivity or synergy among multiple sub-threshold stimuli could still cause a perceptible sensation, as has been shown for wine [60]. Our data also support the conclusion from other studies that the functional groups of oleocanthal are likely the aldehyde groups present on the tail end of the molecule and not the hydroxyphenolethanol moiety [1].

Functional assays like this work may help to build to a TRPA1 pharmacophore in the future. A pharmacophore is "an ensemble of steric and electronic features that is necessary to ensure the optimal supramolecular interactions with a specific biological target and to trigger (or block) its biological response" [61]. Currently, it is not possible to model the pharmacophore in silico because the crystal structure of TRPA1 remains to be determined.

39

Chapter 5

Common over-the-counter pain medications differ in their primary perceptual qualities

Introduction

Flavor has been shown to be a critical predictor of ingestive behavior [62]. Pungency and bitterness are often considered negative flavor qualities in foods and can lead to decreased consumption of vegetables (see [6]) and rejection of medications [63] and other beneficial compounds and foods, especially by children. Sensory evaluation aimed at classifying and quantifying these sensations can lead to improved formulations or strategies for increasing palatability of pharmaceutical agents and fortified foods. Interestingly, in extra virgin olive oil, pungency and bitterness are considered positive qualities by experts (but not consumers [27]), as they speak to the composition and concentration of the phenolic fraction in the oil, specifically the compound oleocanthal. The pungency from oleocanthal is unique in that it acts solely in the back of the throat without eliciting a response on other oral surfaces [16]. Only one other compound is known to elicit a similar regionalized response, the anti-inflammatory drug ibuprofen [3]. This perceptual connection prompted work to demonstrate that oleocanthal shares some of ibuprofen’s biological effects, namely COX-2 inhibition [15]. Using a human bioassay to detect this unique pungency may then be a quick way to identify other compounds with similar biological effects, or to identify functional groups that are important for their activity. Humans also are able distinguish between irritation, bitterness, and sourness, helping us to qualitatively describe sensations from these complex stimuli.

40 Alternatively, molecular modeling could be used to demonstrate the potential for receptor-agonist interaction, essentially eliminating the need for an experimental approach altogether. Oleocanthal and ibuprofen were recently shown to activate the transient receptor potential channel A1 (TRPA1) [1], so presumably other compounds that share this specificity might also share the unique flavor qualities of these compounds. Unfortunately, receptor crystallization is extremely difficult and time-consuming, especially for membrane proteins such as those in the TRP family. To date, TRPA1 has not been crystallized from any organism; the absence of this information makes it impossible to model the receptor or its pharmacophore in silico. Thus, bioassays in cultured cells or intact animals are still required.

The present study uses a human behavioral approach (a human bioassay) to examine the chemosensory qualities of three commonly used, over-the-counter pain medications. Because bitterness, irritation, and sourness arise from different receptors and different cranial nerves, understanding the perceptions these drugs elicit can inform how to apply existing masking technologies to increase palatability or to engineer new methods for masking. Here, ibuprofen was used as a known bitter and irritating compound [3, 10] and naproxen was chosen both because it is a known COX-2 inhibitor and also because it shares structural similarities to ibuprofen. Based on the relationship observed between oleocanthal and ibuprofen, naproxen was therefore hypothesized to have similar perceptual qualities to ibuprofen. The third OTC medication, acetaminophen, shows a mechanism of action that may involve inhibition of the nitric oxide pathway mediated by a variety of neurotransmitter receptors including N-methyl-D- aspartate and substance P, but does not include COX-2 inhibition [64]. Based on this differential pharmacological mechanism, we hypothesized that it would have different chemosensory qualities. To the best of our knowledge, no reports of the qualitative profiles for acetaminophen or naproxen exist in the peer-reviewed literature. Structures of the three compounds can be found below (Figure 5-1).

41

a)

b)

c)

Figure 5-1 Chemical structures of the non-steroidal anti-inflammatory drugs (a) ibuprofen (b) naproxen and (c) acetaminophen. Ibuprofen and naproxen are COX-2 inhibitors, while acetaminophen acts via a different mechanism.

Materials and Methods

Subjects

Reportedly healthy, non-smoking adults (n=20; 6 men; aged 18-45 years) were recruited from the Penn State community. Five of participants from a separate study on tyrosol, oleuropein, and ibuprofen (previous chapter) also participated in this experiment. Procedures were approved by the local Institutional Review Board, informed consent was obtained, and participants were paid for their time. All data were collected in a one-on-one setting at the Sensory Evaluation

Center at Penn State.

42 Stimuli

Aqueous solutions of acetaminophen (Sigma, USP, CAS: 103-90-2), naproxen sodium

(Sigma, USP, CAS: 36159-34-2), and ibuprofen sodium (Fluka, USP, CAS: 31121-93-4) were prepared at 1% (w/v). This concentration has been shown to produce a measurable sensory response from ibuprofen sodium [3] and the molecular weights of ibuprofen, naproxen, and acetaminophen do not differ appreciably (206.3, 230.3, and 151.2 g/mol, respectively). Dosage was capped to ensure that the participants would not cumulatively receive more than a single

OTC dose of the medications. All samples and rinse water were held at 35°C until served. The samples were presented to participants in 30 mL plastic medicine cups.

Procedure

Prior to testing, the task was described to participants and they were oriented to a generalized Labeled Magnitude Scale (gLMS) [37] that would be used to indicate the intensity of

“bitterness,” “irritation,” and “sourness” resulting from the stimuli. At the beginning of the session, participants were asked to rate 15 imagined or remembered sensations (both oral and non-oral items) on the scale (Hayes, Allen, and Bennett, Under Review). Instructions encouraged them to make these ratings in a generalized context by using the top of the scale to indicate their

‘strongest sensation of any kind’. A reminder email asked them to refrain from eating and the use of chemesthetic agents (i.e. toothpaste, mouthwash, gum, spicy food) for at least two hours prior to testing. Both scale orientation and test questions were presented to the participant in the Plus module of Compusense five, version 5.2 (Guelph, ONT).

Discrete interval time-intensity ratings for “bitterness,” “irritation,” and “sourness” were collected for all stimuli every 15 seconds for 60 seconds. Participants were asked to rate

43 “sourness” to avoid dumping effects [34], as we thought the carboxylic acid moiety on naproxen and ibuprofen might potentially impart some sourness to the sample. Intensity measurements were taken for 60 seconds only, as our main interest in this study was to determine the maximum intensity and relative contributions of these sensations rather than their full time intensity profile.

This also served to reduce session time (approximately 20 minutes) and participant inattention. A total of six samples were presented per session (3 stimuli x 2 replicates). A minimum inter- stimulus interval (ISI) of 90 seconds was enforced between each sample. If a participant had any residual irritation at the end of the ISI, they were given more water and asked to wait until all sensations had subsided before proceeding to the next sample.

Results and Discussion

Group means were calculated for each quality, at each time point, and plotted to show temporal patterns of bitterness, sourness, and irritation (Figure 5-2).

44

Figure 5-2 Intensity ratings (group means) for ‘irritation’, ‘bitterness’, and ‘sourness’ from acetaminophen (top), naproxen sodium (middle), and ibuprofen sodium (bottom). None of the stimuli demonstrated considerable sourness. Interestingly, naproxen and ibuprofen (which share structural similarities) each have higher irritation ratings than bitterness, while acetaminophen (which differs structurally) has much higher bitterness and very little irritation. This suggests different techniques for masking may be required.

Visual inspection of the curves shows that naproxen and ibuprofen share similar flavor profiles, as irritation predominates followed by bitterness, with minimal sourness. The steric effects of the additional benzene ring on naproxen may have actually increased the binding affinity of naproxen to TRPA1, as sensory response for irritation was consistently higher than that for ibuprofen. According to receptor theory, increasing the binding affinity will decrease the dissociation constant (KD) and allow naproxen to stay associated with the receptor for a longer

45 time, theoretically increasing the response. This may be expected when considering the additional bulk present on the oleocanthal molecule (Figure 4-2). Additional studies that quantify the KD values for naproxen, ibuprofen, and oleocanthal in vitro with isolated TRPA1 are needed to confirm this hypothesis.

Acetaminophen clearly showed much higher bitterness than either naproxen or ibuprofen, especially immediately after administration. While some irritation was indicated for acetaminophen, our pilot work suggests that this may be artificial inflation of ratings due to the lack of a hedonic response option. As the samples are quite unpleasant overall, the panelists may be unwilling to rate any of the negative qualities at zero. This may also represent participants’ hesitance to use the extremes of the scale, a bias known as ‘end-use avoidance’ (see [59]).

We have seen here, as elsewhere [9], that participants are able to pull apart their ratings of bitterness and irritation. This is important because others have shown bitterness and chemesthesis are perceptually similar [45] and current thinking is that they have a similar functional role, helping us to reject potential toxins [4]. Regardless of this ecological view, the sensations are transduced by different receptors, on different cell types, and are carried to the central nervous system by different nerves. The implication of this for flavor masking strategies is that multiple strategies for knocking out the unpleasant flavor may be necessary with a molecule that can elicit both irritation and bitterness. For instance, here, a strategy that focuses solely on bitterness reduction (i.e. antagonism of T2R receptors) will miss the irritation response from ibuprofen and naproxen, resulting in a product that is still rejected by the consumer.

Moving forward, it will be critical to assess the sensory qualities of other commonly used children’s medications and develop more flavor masking methods that involve physical partitioning or barriers between the molecule of interest and the sensory receptor (e.g. microencapsulation, cyclodextrin complexes).

46

Chapter 6

Physiochemical masking of irritation and bitterness with milk fat

Currently in press as

Bennett, S. M., Zhou, L., and J. E. Hayes (2012). "Using milk fat to reduce the irritation

and bitter taste of ibuprofen." Chemosensory Perception.

Bitterness and irritation elicited by pharmaceutically active molecules remain problematic for pediatric medications, fortified foods and dietary supplements. Few effective methods exist for reducing these unpalatable sensations, negatively impacting medication compliance and intake of beneficial phytonutrients. A physiochemical approach to masking these sensations may be the most successful approach for generalizability to a wide range of structurally and functionally unique compounds. Here, solutions of the non-steroidal anti- inflammatory drug, ibuprofen, were prepared in milk products with varying fat content. Our hypothesis, based on other reports of similar phenomena, was that increasing the fat content would cause ibuprofen to selectively partition into the fat phase, thereby reducing interaction with sensory receptors and decreasing aversive sensations. Quantification of the aqueous concentration of ibuprofen was performed using an isocratic HPLC method coupled with an external standard curve. Sensory testing showed a modest but significant decrease (~20%) in irritation ratings between the skim milk (0% fat) and the half-and-half (11% fat) samples, indicating that fat may contribute to a reduced sensory response. Bitterness was not reduced, remaining constant over all fat levels. The HPLC results indicate a constant amount of ibuprofen remained in the aqueous phase regardless of fat level, so a simple partitioning hypothesis cannot explain the reduced

47 irritancy ratings. Association of ionized ibuprofen with continuous phase solutes such as unabsorbed protein should be explored in future work.

Introduction

The acceptability and palatability of oral pediatric medications is an ongoing problem for parents and medical professionals. Rejection of unpalatable medications compromises compliance with medical regimes and can lead to harm of the child [63]. Unfortunately, many biologically or pharmaceutically active compounds taste bitter and/or irritate the mouth or throat, making these issues important for adults, either directly or indirectly. Similar issues also confront food manufacturers who wish to fortify foods with bioactive ingredients.

It is widely accepted in the chemosensory literature that suppression of bitter tastes can occur through one of three mechanisms. Central cognitive suppression relies on the actual perception of an opponent taste quality for the suppression to occur. The best illustration of this effect was shown in a series of experiments by Lawless [65]. In one, sucrose sweetness was inhibited by the bitterness of phenylthiocarbamide (PTC) for tasters of PTC, but not for non- tasters. In a separate split-tongue experiment, the bitterness of quinine on one side of the tongue was decreased by 20% when sucrose was flowed over the other side. Together these suggest a central mechanism for suppression that is not explained by chemical interactions at the receptor, as the sucrose and quinine were physically separated in the second experiment.

Alternatively, peripheral suppression implies modification of binding at the receptor either through altering the shape of the receptor (as has been shown for sodium, lithium, and zinc ions [66, 67]) or direct antagonism of the receptor (e.g. [68]). Note that these are both independent of a perceived taste quality, like ‘saltiness.’ The specificity of direct receptor antagonism is simultaneously desirable and problematic. This approach may have limited

48 practical utility because many bitter compounds activate more than one receptor, so even if you were to successfully antagonize a single bitter receptor, other receptors may provide functional recovery, eliciting an undesirable sensory response [69].

Interestingly, neither central nor peripheral suppression have been wholly effective for all bitter molecules and even a mixture of approaches may fall short at providing real applicability to the food and pharmaceutical industries, as shown by Keast and colleagues [70]. Thus, it is not surprising that the current methods used in current pediatric formulations (addition of sweeteners such as sucrose, glycerin, sugar alcohols, or high intensity sweeteners) have been relatively unsuccessful in increasing the palatability of liquid pharmaceuticals.

As an alternative, using what we know about the physical characteristics of the molecule in question (i.e. polarity) we may be able to either provide a physical barrier between the agonist and receptor or manipulate its ability to access the receptor by providing an environment that is more “attractive” than the aqueous salivary layer surrounding the receptor. This last method makes intuitive sense to food scientists, as similar techniques are applied to control flavor release and moisture migration in foods. We believe this technique may also have the most utility for the reduction of bitter and irritating sensations across a number of structurally and functionally diverse pharmaceutical agents.

There are a variety of ways to physiochemically encapsulate or block a bitterant or irritant from interacting with a receptor. Cyclodextrins have been used to form complexes that allow hydrophobic molecules to enter a protective pocket on the inside of the cyclodextrin, while sugar molecules on the outside make the overall complex water-soluble [71]. More simply yet is an interesting phenomenon where the introduction of fat into the system may increase or decrease the amount of perceived bitterness. That is, for the bitter compound caffeine, adding fat has been shown to intensify the bitter taste [72]; while for quinine, increasing the fat content of the sample causes a reduction in bitterness [73]. This may be expected due to the relative hydrophobicity of

49 the two compounds. Caffeine is predominantly water-soluble (LogP =-0.07 [74]) and as such, may selectively partition into the aqueous phase of the sample, creating a local concentration of caffeine in the aqueous salivary environment that is much higher than the ‘true’ molarity of the solution. Alternatively, we would expect quinine, which is predominantly hydrophobic (LogPs of

2.82 and 3.52 [75]), to partition selectively into the fat phase thereby reducing access to the aqueous boundary layer adjacent to the receptor. Additionally, increasing the fat content in an emulsion system was also shown to significantly increase the detection threshold of quinine, regardless of the type of fat used [76]. This thinking is not new, as Lawless and students suggested a similar explanation for why capsaicin thresholds were significantly higher when presented in soybean oil, and suprathreshold ratings of irritation intensity were significantly lower in oil [36]. While theoretically appealing, these reports have postulated these mechanistic explanations without empirically quantifying the concentration of the target compound in each phase. Here, we provide HPLC data to quantitate the amount of ibuprofen that remains in the aqueous phase of the ibuprofen-dairy solutions after a 24-hour equilibration period, to gain additional insight into the relationship between fat concentration and bitterness and chemesthetic intensity. Our motivation for using dairy products was two fold. First, milk is a naturally occurring stable emulsion, precluding the need to formulate a model system. Second, milk is readily available in different fat levels, meaning that a successful result would provide caregivers an immediately deployable means to improve oral medication palatability.

In a pilot study where untrained participants (n=28) were asked to make single time point ratings of ‘overall irritation’ and ‘bitterness’ from ibuprofen in milk products with increasing fat content, no significant effect of fat was seen for either attribute, though a trend for lower irritation with increasing fat content was visible. Upon further inspection of the data, there was evidence of a clear first-position bias [59]. That is, regardless of the nature of the sample, the first sample presented to the participant was rated as the most intense of the series. Seventy percent of

50 participants gave the first sample their highest rating on a generalized Labeled Magnitude Scale

(gLMS), where only 1/3rd would be expected by chance. We concluded that the novelty of ibuprofen/dairy samples had to be overcome before panelists were capable of making unbiased judgments in a rating task. We attribute this response to the incongruity associated with these compounds being presented in milk, which is under normal circumstances considered a bland or even refreshing beverage. In their textbook, Lawless and Heymann [59] suggest a possible remedy for first position effects may be to present a ‘dummy’ sample first to absorb the psychological effects before proceeding with the test stimuli of interest. We decided to use this approach in a follow-up experiment where the first sample presented to each participant was quinine in whole milk. Quinine was chosen as the dummy stimulus because it is bitter and unpleasant, but not irritating. Our intention was to reduce fatigue by presenting a non- chemesthetic compound.

Materials and Methods

Subjects

Reportedly healthy, non-smoking adults (n=50; 13 men; aged 18-45 years) were recruited from the Penn State community. Procedures were approved by the local Institutional Review

Board, written informed consent was obtained, and participants were paid for their time. All sensory data were collected one-on-one by the lead author at the Sensory Evaluation Center at

Penn State.

51 Stimuli

All samples were prepared in commercially available skim milk, whole milk, and half- and half purchased from Penn State’s Berkey Creamery. The stimuli were 10 mL samples of

2.50% (w/v) (121.2 mM) USP grade ibuprofen sodium (Fluka, CAS# 31121-93-4) and 0.41 mM kosher quinine hydrochloride (SAFC, CAS# 6119-47-4) in milk. We considered including a water only control to rule out effects of dairy proteins, but decided against it, both because the sample would be visually distinct from the other 4 samples, and because the absence of lactose and dairy volatiles in the water only condition would fundamentally alter perception of the sample [77]. Commercial half-and-half does contain a small amount of disodium phosphate as an emulsifier. How this addition may affect the partitioning behavior of ibuprofen was not determined in this study

The total solids and fat content of the samples were determined using CEM SMART

System5 Moisture/Solids Analyzer and CEM Trac Fat Analysis System (Mathews, NC) following manufacturers instructions, and are provided in Table 6-1. The ibuprofen concentration used was chosen based on work in our laboratory, which indicated this concentration would give irritation ratings between ‘moderate’ and ‘strong’ on a generalized Labeled Magnitude Scale

(gLMS) [10]. All samples were held in 30 mL plastic medicine cups, at 4°C until presented to the participant. All samples were presented in randomized order and labeled with random 3-digit blinding codes. A dummy ‘warm-up’ sample containing quinine, and three test samples (total n=4) were presented per session. Replicates were not obtained due to concerns regarding maximal daily dosing.

52 Table 6-1 Physical composition of milk products. Top shows milk products presented to panelists 1-34, where bottom was presented to panelists 35-50. Two batches were required due to the length of the experiment and quality concerns for the products. Product Total Solids (%) Fat (%) Skim milk 9.06 0.12 Whole milk 12.01 3.40 Half-n-half 18.70 10.58

Product Total Solids (%) Fat (%) Skim milk 9.04 0.36 Whole milk 11.98 3.42 Half-n-half 18.78 10.76

Procedure

Sensory Methods

Participants were asked to refrain from eating and the use of chemesthetic agents (i.e. toothpaste, mouthwash, spicy food) for at least two hours prior to their session. Before beginning the test, participants were oriented to a gLMS [37] using a list of 15 imagined or remembered sensations that included both oral and non-oral items (Hayes, Allen, and Bennett, Under Review).

Scale instructions and orientation encouraged participants to make ratings in a generalized context by indicating that the top of the scale should reflect their ‘strongest sensation of any kind.’ Both scale orientation questions and test questions were presented to the participant in the

Plus module of Compusense five, version 5.2 (Guelph, ONT).

To evaluate the samples, participants were asked to make a single rating of ‘overall irritation in the throat’ and ‘bitterness’ on a gLMS immediately after swallowing the sample.

Participants were instructed to place the 10 mL sample in their mouth and then tilt their head back to allow the sample to reach the throat. They were then instructed to allow the sample to sit at the

53 back of the throat for 5 seconds before swallowing in two stages (swallowing, then immediately swallowing again). Swallowing in two stages purportedly ensures that the stimulus is distributed to the whole surface of the throat. This method has been used previously (e.g. [10, 16]) and is designed specifically to help localize the stimulus exposure to the throat. The participant’s first rating was made immediately after the second swallow. After rating, participants were allowed to rinse with 4°C RO (reverse osmosis) water ad libitum. A minimum inter-stimulus interval (ISI) of an additional 180 seconds was enforced between each sample. If a participant needed more time to recover at this point, they were given more water and asked to indicate when they felt ready to continue. A dummy ‘warm-up’ sample containing quinine, and three test samples were presented per session. Total session time was approximately 20 minutes.

Ibuprofen Analysis by HPLC

Ibuprofen sodium (2.50 % w/v) was introduced into dairy samples (skim milk, whole milk, and half-and-half) and allowed to equilibrate for 24 h at 4!C. Ibuprofen in the continuous phase was separated by filtration using Amicon Ultra 0.5 mL centrifugal filters with 10 kDa cutoff from EMD Millipore (Billerica, MA). The size of this filter would be expected to preclude both the fat phase and many proteins. The filtrate was diluted 1:2000 using methanol and then filtered over 0.45 µm PTFE syringe filters prior to HPLC analysis. Samples were introduced using a Shimadzu 20ADvp temperature-controlled autosampler (4 ºC) and separation achieved using a reverse phase Supelcosil LC-18 (4.6 x 150 mm, 5 µm; Supelco Inc., Bellefonte, PA).

Ibuprofen was eluted using an isocratic method with a mobile phase of 0.1% v/v formic acid in a

74% v/v methanol in water solution. The injection volume was 20 µL and the flow rate held at 1 mL/min. Ibuprofen was detected at 220 nm using a Shimadzu SPD-20AV UV-Vis detector and quantitation based on an external standard curve.

54 Statistical Analysis

Data were analyzed using SAS 9.2 (Cary, NC). For sensory data, repeated measures main effects ANOVA were performed via proc mixed, with participants as a random effect, assuming compound symmetry for the covariance structure. Planned comparisons across individual samples were tested via unadjusted t-tests. The quinine ‘warm-up’ sample was excluded from primary analysis a priori. HPLC data were analyzed via 1 –way ANOVA in proc mixed, assuming compound symmetry for the covariance structure.

Results and Discussion

Sensory Data

For the irritation ratings, we performed repeated-measures ANOVA with fat level and sample position as factors: the main effect of fat level was marginal [F(2,96)=2.56; p=0.083] while sample position showed no effect [F(2.96)=2.32; p=0.104]. Planned comparisons via t-tests indicated that the half-and-half samples were significantly less irritating [t96 = -2.24; p = 0.027] that skim milk samples. Irritancy in whole milk was intermediate between the low and high fat samples, although the differences with skim [t96 = -1.36; p = 0.18] and half-and-half [t96 = -0.89; p

= 0.38] were not significant. As shown in Figure 6-1A, the mean irritation ratings in half-and-half samples were nearly 6 points lower than the skim milk sample on a gLMS, although both were still in the “moderate” to “strong” range of the scale. In a separate analysis, all ibuprofen samples were significantly more irritating than the quinine ‘warm-up’ sample, indicating the participants could successfully distinguish between bitterness and irritancy.

Bitterness ratings are shown in Figure 6-1B. Repeated-measures ANOVA with fat level and sample position as factors indicated no effect of fat level [F(2,96)=1.58; p=0.21] while

55 sample position was significant [F(2,96)=3.71; p=0.028]. The sample received last was less bitter than the second to last sample [t96 = 2.72; p = 0.007]; no other effects of position were observed

(p’s <0.15).

56

A

Effect of Fat Content on Bitterness from Ibuprofen

Skim

p = .10

Whole p = .78

p = .17

1/2 & 1/2

0 10 20 30 40 Perceived Irritation (gLMS) B

Figure 6-1 The effect of milk fat on irritation (A) and bitterness (B) from ibuprofen. A significant effect of fat was seen for irritation between the skim and half-and-half samples, but no significant effects were seen for bitterness.

57 Previous work suggests untrained participants are capable of pulling apart bitterness and pungency from capsaicin and other oral irritants [9]. Here, the lack of a hedonic response option for the samples may have caused affective responses be dumped [34] into both the bitterness and irritation ratings. Based on our experience with these stimuli, we believe dumping may have reduced the apparent effect size. Traditionally, it would be considered inappropriate to change the cognitive task and ask for both affective and intensity ratings in untrained participants [59], but due to the unfamiliar and unpleasant nature of the stimuli, the added cognitive load may be more than offset by the avoidance of dumping. Additional work is needed to clarify this trade-off.

Alternatively, a trained panel approach could be used, although the lack of perceptually clean reference compounds would complicate the training process [10, 20, 78].

Analysis of continuous phase ibuprofen

After an addition of 2.50% w/v ibuprofen, ibuprofen concentrations remaining in the fraction not bound to protein or fat were 1.66±0.04%, 1.49±0.05%, and 1.66±0.01% w/v ibuprofen in skim milk, whole milk, and half-and-half, respectively. In one-way ANOVA, the amount of ibuprofen remaining in the aqueous phase differed across fat level [F(2,6)=5.0; p=0.05], but not in the manner we anticipated. The amount of ibuprofen was significantly lower in whole milk than either skim milk [t6=2.74; p=0.03] or half-and-half [t6=2.74; p=0.03].

Although unexpected, this finding does not change the overall interpretation: the irritancy reduction observed in the human sensory data could not be easily explained by partitioning into the lipid phase.

The pH of the delivery media selected may have influenced this lack of partitioning.

Ibuprofen is a propionic acid derivative with a pKa of 5.2 leading to pH-dependent increases in aqueous solubility [79] due to ionization which would also occur at the pH common to milk (pH

58 6.7). Different results may have been achieved in low pH milk products, like yogurt or kefir, where ibuprofen would remain protonated and thus more hydrophobic. Ionic and non-ionic surfactants have also been shown to increase ibuprofen aqueous solubility [80] with proteins and phospholipids in dairy products, potentially causing similar effects. Our results suggest that the ionized ibuprofen may be interacting with continuous phase solutes, such as individual whey proteins or casein micelles, but dialysis and additional HPLC analysis would be required to be certain. Additionally, it may be that n-octanol/water partition coefficients (those used to generate the hypotheses of this experiment) are not good predictors of partitioning behavior in milk fat

[81].

Limitations and Conclusions

Ibuprofen was used as a model compound for this work because it is safe, commonly used, easy to obtain, and known to cause both irritation and bitterness. However, a compound with a readily ionizable group may not be ideal for a partitioning study. Additionally, there are drawbacks to using milk as the delivery vehicle instead of a model emulsion system, but our goal here was a one of translational utility —to determine if there was an easy, at-home method caregivers could use to increase the palatability of oral pharmaceuticals. The 20% reduction observed here might be due to viscosity changes, as the viscosity of the skim and whole milk samples were not matched to that of half-and-half. Increasing viscosity has been shown to correlate with decreased perception of sweetness [82], bitterness [83], and capsaicin pungency

[84]. In any case, the reduction observed is not easily explained by a partitioning of the compound in to the fat phase. Follow-up analyses to more completely characterize the distribution of ibuprofen within the food matrix may provide more insight to the mechanism by which this reduction occurs.

59 Acknowledgements

This manuscript was completed in partial fulfillment of the requirements for a Master of

Science degree at the Pennsylvania State University by S.M.B. The authors warmly thank Bonnie

C. Ford for assistance with total solids and fat determination of our milk samples and Lisa Zhou for assistance with HPLC analysis and thoughtful input on the manuscript.

Funding

This work was funded by funds from the Pennsylvania State University and a grant from the National Institutes of Health Institute of Deafness and Communication Disorders to the corresponding author [grant DC010904].

60

Chapter 7

Conclusions and further steps

To further explore the unique irritation elicited by oleocanthal and ibuprofen, a series of related theoretically informed sensory experiments were undertaken to determine qualitative differences between these and other oral irritants, identify relationships between these compounds and some that are structurally or functionally similar, and apply this knowledge in an attempt to increase palatability of a common pharmaceutical agent. Below are the major experimental findings of these studies:

• Irritation sub-qualities from extra virgin olive oil (EVOO) suggest that olive oil

contains both agonists of the TRPA1 and TRPV1 cation channels—EVOO is described

predominantly as tickling and warming/hot, whereas capsaicin is burning and

warming/hot and ibuprofen is mostly tickling and numbing, but is quite diffuse.

• Studies of tyrosol and oleuropein suggest that neither compound elicits appreciable

bitterness or irritation at levels similar to levels found in EVOO. This confirms other

reports that they do not contribute to the pungency of EVOO. It also agrees with

structure-activity data that suggests the activity of oleocanthal involves the aldehyde

groups and not the phenol group present in tyrosol and oleuropein

• Similarly, naproxen shows a chemosensory profile resembling that of ibuprofen and not

acetaminophen, suggesting that the shared carboxylic acid moiety may be responsible for

the irritant properties of ibuprofen and naproxen. Also, acetaminophen is extremely bitter

and not irritating, suggesting that the sensory response to these compounds is mediated

by different receptor systems. This has major implications for masking technologies

61 because an approach that focuses solely on bitterness reduction may be unsuccessful for

compounds that are also irritating. We suggest methods that prevent the active molecule

from reaching the receptor either through a physical barrier or selective partitioning in the

medium.

• A preliminary study of ibuprofen showed that milk fat was effective for masking

irritation although the effect size was small. In contrast, fat level did not reduce

bitterness. HPLC data suggest that simple fat/aqueous phase partitioning does not

explain the sensory data. Additional studies are needed to better understand the chemical

phenomena underlying this effect.

While these studies have been useful for behaviorally validating relationships that have been identified in vitro, they have also generated more questions that need to be explored further.

Due to the issues we found in using a rating task with a large number of response options, we designed a modified check-all-that apply (CATA) methodology and coupled it with discrete interval, time-intensity scaling to create a qualitative irritation profile for capsaicin, ibuprofen, and olive oil. This method should be refined and used to explore irritation properties of other compounds including traditional irritants like cinnamaldehyde, CO2, and menthol, but also less studied irritants like naproxen and other pharmaceuticals. These new analyses should be coupled with existing work that focused on spatial and temporal characteristics of oral irritants (e.g. [8,

19, 20]).

While great strides have been made to fully understand the unique irritation and receptor mechanisms of oleocanthal and ibuprofen, it is still uncertain how these structurally dissimilar molecules each elicit this distinctive response and how other compounds (e.g. cinnamaldehyde, acrolein, crotonaldehyde) containing an ", #-unsaturated aldehyde (like oleocanthal) do not. The claim for oleocanthal is that it is more specific for TRPA1 (while the other common TRPA1

62 agonists have been shown to activate other TRP receptors as well [24, 85]) and also that oleocanthal activates TRPA1 through an alternative mechanism. Other electrophillic agonists of

TRPA1 have been shown to predominantly activate TRPA1 via covalent modification of cysteine residues located on the N-terminal cytoplasmic domain of the receptor [21]. This mechanism was ruled out for oleocanthal by testing activity on mutant TRPA1, where there was substitution of serine for two reactive cysteine residues. This analysis was not reported for ibuprofen [1].

So, while oleocanthal and ibuprofen both selectively activate TRPA1 in HEK 293 cells expressing human TRP channels and immunohistochemistry on human tissues implicates restriction of TRPA1 to tissues in the upper pharynx and nasal epithelium, this is only good evidence that ibuprofen and oleocanthal’s shared sensory response may be due to a similar mechanism of action and not what the mechanism might be. A structure-activity study of oleocanthal (see Chapter 3) indicated that both aldehyde groups on the tail end of the molecule are required for activity, but again the same analysis for ibuprofen, which does not contain even a single aldehyde moiety, was not reported [1]. As we identify more compounds that elicit this unique sensory profile (as we have done here for naproxen), we will get closer to truly understanding the relationship between ibuprofen and oleocanthal. I suspect that due to the potential health benefits of oleocanthal consumption, work in this area will continue to be a ‘hot topic’ for some time.

Finally, while the effect of milk fat on irritation and bitterness reduction was not as pronounced as we expected, there is still evidence that other physiochemical techniques such as microencapsulation may be valuable tools for improving the flavor of liquid, oral pharmaceuticals. Additional work qualitatively evaluating sensations from other medications and studies coupling physiochemical, peripheral, and central cognitive suppression mechanisms may help increase medication compliance in children or help improve the flavor of dietary supplements. A trained panel could be a valuable tool for studying these sensations, as training

63 and practice can help to decouple hedonic response and the intensity of bitterness and irritation, but it may also be extremely difficult due to the nature of chemesthetic stimuli (see Chapter 1).

64

References

1. Peyrot des Gachons, C., et al., Unusual pungency from extra-virgin olive oil is attributable to restricted spatial expression of the receptor of oleocanthal. J Neurosci, 2011. 31(3): p. 999-1009. 2. Cliff, M.A. and B.G. Green, Sensitization and desensitization to capsaicin and menthol in the oral cavity: interactions and individual differences. Physiol Behav, 1996. 59(3): p. 487-94. 3. Breslin, P.A.S., T.N. Gingrich, and B.G. Green, Ibuprofen as a Chemesthetic Stimulus: Evidence of a Novel Mechanism of Throat Irritation. Chemical Senses, 2001. 26(1): p. 55-65. 4. Green, B.G., Chemesthesis and the Chemical Senses as Components of a "Chemofensor Complex". Chemical Senses, 2011. 5. Chandrashekar, J., et al., The receptors and cells for mammalian taste. Nature, 2006. 444(7117): p. 288-294. 6. Drewnowski, A. and C. Gomez-Carneros, Bitter taste, phytonutrients, and the consumer: a review. The American Journal of Clinical Nutrition, 2000. 72(6): p. 1424-1435. 7. Green, B.G., Mason, J.R., Kare, M.R. ed., Irritation. Chemical Senses. Vol. 2. 1990, New York: Marcel-Dekker. 8. McDonald, S., Barrett, P., and Bond, L., What Kind of Hot Is It?, in Perfumer and Flavorist. 2010, Allured Publishing Corp.: Carol Stream, IL. 9. Green, B.G. and J.E. Hayes, Capsaicin as a probe of the relationship between bitter taste and chemesthesis. Physiology & Behavior, 2003. 79(4-5): p. 811-821. 10. Bennett, S.M. and J.E. Hayes, Differences in the Chemesthetic Subqualities of Capsaicin, Ibuprofen, and Olive Oil. Chemical Senses, 2012. 11. Popkin, B.M., The Nutrition Transition and Obesity in the Developing World. The Journal of Nutrition, 2001. 131(3): p. 871S-873S. 12. Fung, T.T., et al., Association between dietary patterns and plasma biomarkers of obesity and cardiovascular disease risk. The American Journal of Clinical Nutrition, 2001. 73(1): p. 61-67. 13. Hu, F.B., Globalization of Diabetes. Diabetes Care, 2011. 34(6): p. 1249-1257. 14. Cicerale, S., et al., Chemistry and Health of Olive Oil Phenolics. Critical Reviews in Food Science and Nutrition, 2008. 49(3): p. 218-236. 15. Beauchamp, G.K., et al., Phytochemistry: ibuprofen-like activity in extra-virgin olive oil. Nature, 2005. 437(7055): p. 45-6. 16. Cicerale, S., et al., Sensory Characterization of the Irritant Properties of Oleocanthal, a Natural Anti-Inflammatory Agent in Extra Virgin Olive Oils. Chemical Senses, 2009. 34(4): p. 333-339. 17. Andrewes, P., et al., Sensory properties of virgin olive oil polyphenols: identification of deacetoxy-ligstroside aglycon as a key contributor to pungency. J Agric Food Chem, 2003. 51(5): p. 1415-20. 18. Pitt, J., et al., Alzheimer's-associated Abeta oligomers show altered structure, immunoreactivity and synaptotoxicity with low doses of oleocanthal. Toxicol Appl Pharmacol, 2009. 240(2): p. 189-97. 19. Lawless, H.T. and D.A. Stevens, Responses by humans to oral chemical irritants as a function of locus of stimulation. Percept Psychophys, 1988. 43(1): p. 72-8.

65 20. Cliff, M. and H. Heymann, Descriptive Analysis of Oral Pungency. Journal of Sensory Studies, 1992. 7(4): p. 279-290. 21. Macpherson, L.J., et al., Noxious compounds activate TRPA1 ion channels through covalent modification of cysteines. Nature, 2007. 445(7127): p. 541-545. 22. Bautista, D.M., et al., TRPA1 Mediates the Inflammatory Actions of Environmental Irritants and Proalgesic Agents. Cell, 2006. 124(6): p. 1269-1282. 23. Bautista, D.M., et al., Pungent products from garlic activate the sensory ion channel TRPA1. Proceedings of the National Academy of Sciences of the United States of America, 2005. 102(34): p. 12248-12252. 24. Xu, H., et al., Oregano, thyme and clove-derived flavors and skin sensitizers activate specific TRP channels. Nat Neurosci, 2006. 9(5): p. 628-35. 25. Fischer, R., et al., Weber Ratio in Gustatory Chemoreception; an Indicator of Systemic (Drug) Reactivity. Nature, 1965. 207(5001): p. 1049-1053. 26. Smith, A.B., J.B. Sperry, and Q. Han, Syntheses of (‚àí)-Oleocanthal, a Natural NSAID Found in Extra Virgin Olive Oil, the (‚àí)-Deacetoxy-Oleuropein Aglycone, and Related Analogues. The Journal of Organic Chemistry, 2007. 72(18): p. 6891-6900. 27. Delgado, C. and J.-X. Guinard, How do consumer hedonic ratings for extra virgin olive oil relate to quality ratings by experts and descriptive analysis ratings? Food Quality and Preference, 2011. 22(2): p. 213-225. 28. Green, B.G. and M.T. Schullery, Stimulation of bitterness by capsaicin and menthol: differences between lingual areas innervated by the glossopharyngeal and chorda tympani nerves. Chem Senses, 2003. 28(1): p. 45-55. 29. Horne, J., et al., Bitter Taste of Saccharin and Acesulfame-K. Chemical Senses, 2002. 27(1): p. 31-38. 30. Green, B.G., Psychophysical Measurement of Oral Chemesthesis, in Methods in Chemosensory Research, S.A. Simon, Nicolelis, M.A.L., Editor. 2002, CRC Press Boca Raton. p. 3-19. 31. Lawless, H., Oral chemical irritation: psychophysical properties. Chemical Senses, 1984. 9(2): p. 143-155. 32. Karrer, T. and L. Bartoshuk, Capsaicin desensitization and recovery on the human tongue. Physiology & Behavior, 1991. 49(4): p. 757-764. 33. Frank, R., N. van der Klaauw, and H. Schifferstein, Both perceptual and conceptual factors influence taste-odor and taste-taste interactions. Attention, Perception, & Psychophysics, 1993. 54(3): p. 343-354. 34. Clark, C.C. and H.T. Lawless, Limiting response alternatives in time-intensity scaling: an examination of the halo-dumping effect. Chemical Senses, 1994. 19(6): p. 583-594. 35. Green, B.G., et al., Taste mixture interactions: Suppression, additivity, and the predominance of sweetness. Physiology & Behavior, 2010. 101(5): p. 731-737. 36. Lawless, H.T., C. Hartono, and S. Hernandez, Thresholds and Suprathreshold Intensity Functions for Capsaicin in Oil and Aqueous Based Carriers. Journal of Sensory Studies, 2000. 15(4): p. 437-477. 37. Snyder, D., K. Fast, and L. Bartoshuk, Valid Comparisons of Suprathreshold Sensations. Journal of Consciousness Studies, 2004. 11: p. 96-112. 38. Matthews, J.N., et al., Analysis of serial measurements in medical research. BMJ, 1990. 300(6719): p. 230-5. 39. Green, B.G., Regional and Individual Differences in Cutaneous Sensitivity to Chemical Irritants: Capsaicin and Menthol. Cutaneous and Ocular Toxicology, 1996. 15(3): p. 277- 295.

66 40. Caterina, M.J., et al., The capsaicin receptor: a heat-activated ion channel in the pain pathway. Nature, 1997. 389(6653): p. 816-24. 41. Harwood, A., The Handbook of Olive Oil. 1 ed. 1999: Springer. 42. Wang, Y.Y., R.B. Chang, and E.R. Liman, TRPA1 is a component of the nociceptive response to CO2. J Neurosci, 2010. 30(39): p. 12958-63. 43. Komai, M. and B.P. Bryant, Acetazolamide specifically inhibits lingual trigeminal nerve responses to carbon dioxide. Brain Research, 1993. 612(1-2): p. 122-129. 44. Glendinning, J.I., Is the bitter rejection response always adaptive? Physiology & Behavior, 1994. 56(6): p. 1217-1227. 45. Lim, J. and B.G. Green, The Psychophysical Relationship between Bitter Taste and Burning Sensation: Evidence of Qualitative Similarity. Chemical Senses, 2007. 32(1): p. 31-39. 46. Iida, T., et al., TRPV1 activation and induction of nociceptive response by a non-pungent capsaicin-like compound, capsiate. Neuropharmacology, 2003. 44(7): p. 958-967. 47. Stuer-Lauridsen, F., et al., Environmental risk assessment of human pharmaceuticals in Denmark after normal therapeutic use. Chemosphere, 2000. 40(7): p. 783-793. 48. Peyrot Des Gachons, C., Sperry, J. B., Bryant, B., Breslin, P.A.S., Smith, A.B., Beauchamp, G.K., Use of the Irritating Principal Oleocanthal in Olive Oil, As Well As Structurally and Functionally Similar Compounds. 2011: United States. p. 26. 49. van der Klaauw, N.J. and R.A. Frank, Scaling component intensities of complex stimuli: The influence of response alternatives. Environment International, 1996. 22(1): p. 21-31. 50. Genter, M.B., et al., Olfactory toxicity of methimazole: dose-response and structure- activity studies and characterization of flavin-containing monooxygenase activity in the Long-Evans rat olfactory mucosa. Toxicol Pathol, 1995. 23(4): p. 477-86. 51. Kaschula, C.H., et al., Structure-Activity Relationships in 4-Aminoquinoline Antiplasmodials. The Role of the Group at the 7-Position. Journal of Medicinal Chemistry, 2002. 45(16): p. 3531-3539. 52. Alarie, Y., Bioassay for evaluating the potency of airborne sensory irritants and predicting acceptable levels of exposure in man. Food and Cosmetics Toxicology, 1981. 19(0): p. 623-626. 53. Alarie, Y., Dose-response analysis in animal studies: prediction of human responses. Environ Health Perspect, 1981. 42: p. 9-13. 54. Clapham, D.E., L.W. Runnels, and C. Strubing, The TRP ion channel family. Nat Rev Neurosci, 2001. 2(6): p. 387-96. 55. Ursu, D., et al., Pungency of TRPV1 agonists is directly correlated with kinetics of receptor activation and lipophilicity. European Journal of Pharmacology, 2010. 641(2–3): p. 114-122. 56. Bessac, B.F. and S.-E. Jordt, Breathtaking TRP Channels: TRPA1 and TRPV1 in Airway Chemosensation and Reflex Control. Physiology, 2008. 23(6): p. 360-370. 57. Taylor-Clark, T.E., et al., TRPA1: A potential target for anti-tussive therapy. Pulmonary Pharmacology & Therapeutics, 2009. 22(2): p. 71-74. 58. Saitta, M., et al., Minor compounds in the phenolic fraction of virgin olive oils. Food Chemistry, 2009. 112(3): p. 525-532. 59. Lawless, H.T. and H. Heymann, Context Effects and Biases in Sensory Judgment Sensory Evaluation of Food. 2010, Springer New York. p. 203-225. 60. Singleton V, L. and C. Noble A, Wine Flavor and Phenolic Substances, in Phenolic, Sulfur, and Nitrogen Compounds in Food Flavors. 1976, AMERICAN CHEMICAL SOCIETY. p. 47-70.

67 61. Wermuth, C.G., Ganellin, C.R., and Mitscher, L.A., Glossary of Terms Used in Medicinal Chemistry. Pure and Applied Chemistry, 1998. 70(5): p. 1129-1143. 62. IFICF, Food and Health Survey: Consumer Attitudes Toward Food Safety, Nutrition, & Health, I.F.I.C. Foundation, Editor. 2011. 63. Mennella, J.A. and G.K. Beauchamp, Optimizing oral medications for children. Clin Ther, 2008. 30(11): p. 2120-32. 64. Bj#$rkman, R., et al., Acetaminophen blocks spinal hyperalgesia induced by NMDA and substance P. Pain, 1994. 57(3): p. 259-264. 65. Lawless, H.T., Evidence for neural inhibition in bittersweet taste mixtures. J Comp Physiol Psychol, 1979. 93(3): p. 538-47. 66. Breslin, P.A.S. and G.K. Beauchamp, Suppression of Bitterness by Sodium: Variation Among Bitter Taste Stimuli. Chemical Senses, 1995. 20(6): p. 609-623. 67. Keast, R.S.J., The Effect of Zinc on Human Taste Perception. Journal of Food Science, 2003. 68(5): p. 1871-1877. 68. Greene, T.A., et al., Probenecid Inhibits the Human Bitter Taste Receptor TAS2R16 and Suppresses Bitter Perception of Salicin. PLoS One, 2011. 6(5): p. e20123. 69. Ley, J., Masking Bitter Taste by Molecules. Chemosensory Perception, 2008. 1(1): p. 58- 77. 70. Keast, R. and P. Breslin, Bitterness Suppression with Zinc Sulfate and Na-Cyclamate: A Model of Combined Peripheral and Central Neural Approaches to Flavor Modification. Pharmaceutical Research, 2005. 22(11): p. 1970-1977. 71. Challa, R., et al., Cyclodextrins in drug delivery: An updated review. AAPS PharmSciTech, 2005. 6(2): p. E329-E357. 72. Keast, R.S.J., Modification of the bitterness of caffeine. Food Quality and Preference, 2008. 19(5): p. 465-472. 73. Metcalf, K.L. and Z.M. Vickers, Taste Intensities of Oil-in-Water Emulsions with Varying Fat Content. Journal of Sensory Studies, 2002. 17(5): p. 379-390. 74. Biagi, G.L., et al., Study of the lipophilic character of xanthine and adenosine derivatives : I. RM and log P values. Journal of Chromatography A, 1990. 498(0): p. 179-190. 75. Eros, D., et al., Reliability of logP predictions based on calculated molecular descriptors: a critical review. Curr Med Chem, 2002. 9(20): p. 1819-29. 76. Thurgood, J.E. and S. Martini, Effects of Three Emulsion Compositions on Taste Thresholds and Intensity Ratings of Five Taste Compounds. Journal of Sensory Studies, 2010. 25(6): p. 861-875. 77. Hayes, J.E. and V.B. Duffy, Revisiting Sugar-Fat Mixtures: Sweetness and Creaminess Vary with Phenotypic Markers of Oral Sensation. Chemical Senses, 2007. 32(3): p. 225- 236. 78. Bennett, S.M., and Hayes, John E., Chemesthesis and Flavor, in The World of Food Ingredients. 2012, CNS Media BV: Singapore. p. 44-46. 79. Hansen, N.T., et al., Prediction of pH-Dependent Aqueous Solubility of Druglike Molecules. Journal of Chemical Information and Modeling, 2006. 46(6): p. 2601-2609. 80. Stephenson, B.C., et al., Experimental and theoretical investigation of the micellar- assisted solubilization of ibuprofen in aqueous media. Langmuir, 2006. 22(4): p. 1514-25. 81. Scheytt, T., et al., 1-Octanol/Water Partition Coefficients of 5 Pharmaceuticals from Human Medical Care: Carbamazepine, Clofibric Acid, Diclofenac, Ibuprofen, and Propyphenazone. Water, Air, & Soil Pollution, 2005. 165(1): p. 3-11. 82. Arabie, P. and H. Moskowitz, The effects of viscosity upon perceived sweetness. Attention, Perception, & Psychophysics, 1971. 9(5): p. 410-412.

68 83. Moskowitz, H.R. and P. Arabie, Taste Intensity as a Function of Stimulus Concentration and Solvent Viscosity. Journal of Texture Studies, 1970. 1(4): p. 502-510. 84. Baron, R.F. and M.P. Penfield, Capsaicin Heat Intensity - Concentration, Carrier, Fat Level, and Serving Temperature Effects. Journal of Sensory Studies, 1996. 11(4): p. 295- 316. 85. Macpherson, L.J., et al., More than cool: promiscuous relationships of menthol and other sensory compounds. Mol Cell Neurosci, 2006. 32(4): p. 335-43.

69

Appendix A: Chapter 3 Copyright Permissions

OXFORD UNIVERSITY PRESS LICENSE TERMS AND CONDITIONS Feb 15, 2012

This is a License Agreement between Samantha M Bennett ("You") and Oxford University Press

("Oxford University Press") provided by Copyright Clearance Center ("CCC"). The license consists of your order details, the terms and conditions provided by Oxford University Press, and the payment terms and conditions.

All payments must be made in full to CCC. For payment instructions, please see information listed at the bottom of this form. License Number 2847830679758

License date Feb 14, 2012

Licensed content publisher Oxford University Press

Licensed content publication Chemical Senses

Licensed content title Differences in the Chemesthetic Subqualities of Capsaicin, Ibuprofen, and Olive Oil:

Licensed content author Samantha M. Bennett, John E. Hayes

Licensed content date 01/25/2012

Type of Use Thesis/Dissertation

Institution name

Title of your work CHEMESTHESIS AND BITTERNESS OF NATURAL AND SYNTHETIC ANTI-INFLAMMATORY STIMULI

70

Publisher of your work n/a

Expected publication date May 2012

Permissions cost 0.00 USD

Value added tax 0.00 USD

Total 0.00 USD

Total 0.00 USD Terms and Conditions

STANDARD TERMS AND CONDITIONS FOR REPRODUCTION OF MATERIAL FROM AN OXFORD UNIVERSITY PRESS JOURNAL 1. Use of the material is restricted to the type of use specified in your order details. 2. This permission covers the use of the material in the English language in the following territory: world. If you have requested additional permission to translate this material, the terms and conditions of this reuse will be set out in clause 12. 3. This permission is limited to the particular use authorized in (1) above and does not allow you to sanction its use elsewhere in any other format other than specified above, nor does it apply to quotations, images, artistic works etc that have been reproduced from other sources which may be part of the material to be used. 4. No alteration, omission or addition is made to the material without our written consent. Permission must be re-cleared with Oxford University Press if/when you decide to reprint. 5. The following credit line appears wherever the material is used: author, title, journal, year, volume, issue number, pagination, by permission of Oxford University Press or the sponsoring society if the journal is a society journal. Where a journal is being published on behalf of a learned society, the details of that society must be included in the credit line. 6. For the reproduction of a full article from an Oxford University Press journal for whatever purpose, the corresponding author of the material concerned should be informed of the proposed use. Contact details for the corresponding authors of all Oxford University Press journal contact can be found alongside either the abstract or full text of the article concerned, accessible from www.oxfordjournals.org Should there be a problem clearing these rights, please contact [email protected] 7. If the credit line or acknowledgement in our publication indicates that any of the figures, images or photos was reproduced, drawn or modified from an earlier source it will be necessary for you to clear this permission with the original publisher as well. If this permission has not been obtained, please note that this material cannot be included in your publication/photocopies. 8. While you may exercise the rights licensed immediately upon issuance of the license at the end of the licensing process for the transaction, provided that you have disclosed complete and accurate details of your proposed use, no license is finally effective unless and until full payment is received from you (either by Oxford University Press or by Copyright Clearance Center (CCC)) as provided in CCC's Billing and Payment terms and conditions. If full payment is not received on a timely basis, then any license preliminarily granted shall be deemed automatically revoked and shall be void as if never granted. Further, in the event that you breach any of these terms and conditions or any of CCC's Billing and Payment terms and conditions, the license is automatically revoked and shall be void as if never granted. Use of materials as described in a revoked license, as well as any use of the materials beyond the scope of an unrevoked

71 license, may constitute copyright infringement and Oxford University Press reserves the right to take any and all action to protect its copyright in the materials. 9. This license is personal to you and may not be sublicensed, assigned or transferred by you to any other person without Oxford University Press’s written permission. 10. Oxford University Press reserves all rights not specifically granted in the combination of (i) the license details provided by you and accepted in the course of this licensing transaction, (ii) these terms and conditions and (iii) CCC’s Billing and Payment terms and conditions. 11. You hereby indemnify and agree to hold harmless Oxford University Press and CCC, and their respective officers, directors, employs and agents, from and against any and all claims arising out of your use of the licensed material other than as specifically authorized pursuant to this license. 12. Other Terms and Conditions: v1.4 If you would like to pay for this license now, please remit this license along with your payment made payable to "COPYRIGHT CLEARANCE CENTER" otherwise you will be invoiced within 48 hours of the license date. Payment should be in the form of a check or money order referencing your account number and this invoice number RLNK500719307. Once you receive your invoice for this order, you may pay your invoice by credit card. Please follow instructions provided at that time.

Make Payment To: Copyright Clearance Center Dept 001 P.O. Box 843006 Boston, MA 02284-3006

For suggestions or comments regarding this order, contact RightsLink Customer Support: [email protected] or +1-877-622-5543 (toll free in the US) or +1-978-646-2777. Gratis licenses (referencing $0 in the Total field) are free. Please retain this printable license for your reference. No payment is required.

72

Appendix B: Chapter 6 Supplemental Data

Milk Sample Analysis

Table B-1 Total Solids and fat content of milk samples used in the pilot experiment in Chapter 5. Product Total Solids (%) Fat (%) Skim milk 9.07 0.16 Whole milk 12.02 3.33 Half-n-half 18.82 10.84

Analytical Analysis of Ibuprofen

Ibuprofen Standard Curve 500×103 7762 + 21585*x 450 r = 0.9999

400

350

300

250 Area

200

150

100

50

0 0 2 4 6 8 10 12 14 16 18 20 22 Ibuprofen Concentration (ppm)

Figure B-1 Ibuprofen standard curve used for quantification of ibuprofen remaining in the continuous phase of milk products after 24-hour equilibration.