MECHANISMS OF SYNAPTIC AND INTRINSIC MODULATION OF
GRANULE CELLS IN THE RAT OLFACTORY BULB
By
RICHARD TODD PRESSLER
Submitted in partial fulfillment of the requirements
For the degree Doctor of Philosophy
Thesis Adviser: Dr. Ben W. Strowbridge
Department of Neurosciences
CASE WESTERN RESERVE UNIVERSITY
August, 2006 CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the dissertation of
______
candidate for the Ph.D. degree *.
(signed)______(chair of the committee)
______
______
______
______
______
(date) ______
*We also certify that written approval has been obtained for any proprietary material contained therein. Table of contents Table of contents...... iii List of figures...... iv Acknowledgements...... v List of Abbreviations...... vi Abstract...... vii Chapter 1: Introduction...... 1 Basic circuit description of the olfactory bulb...... 2 Diversity of interneurons in the olfactory bulb...... 10 Centrifugal input to the olfactory bulb...... 14 Chapter 2: Blanes cells mediate persistent feedforward inhibition onto granule cells in the olfactory bulb...... 30 Summary...... 31 Introduction...... 32 Results...... 35 Discussion...... 50 Experimental Procedures...... 61 Chapter 3: Muscarinic receptor modulation of afterpotentials and firing modes in olfactory bulb granule cells...... 87 Introduction...... 88 Materials and Methods...... 92 Results...... 94 Discussion...... 104 Chapter 4: Discussion...... 123 Enhancement of excitability by an intrinsic activity-evoked Afterdepolarization...... 125 Afterdepolarizations as a mechanism for generating persistent firing...... 133 Similarities and differences between olfactory bulb circuitry and retinal circuitry...... 136 Factors that modulate granule cell activity...... 138 Strategies to determine the contributions of these two inputs onto granule cells inintact and behaving animals...... 144 Future Directions...... 145 Chapter 5: Bibliography...... 151
iii List of Figures Figure 1-1 The spatial map of odorant receptors onto olfactory bulb glomeruli...... 24 Figure 1-2 The basic circuit of the olfactory bulb...... 26 Figure 1-3 Original description of the olfactory bulb Blanes cell...... 28 Figure 2-1 Multiple Cell Types in Granule Cell Layer of the Olfactory Bulb...... 66 Figure 2-2 Visualization of Blanes Cell Morphology with Two-Photon Microscopy...... 68 Figure 2-3 Blanes Cells Are GABAergic and Innervate Granule Cells...... 70 Figure 2-4 Afterdepolarizations in Blanes Cells...... 73 Figure 2-5 Calcium-Dependent Afterdepolarizations in Blanes Cells...... 75 Figure 2-6 Afterdepolarizations Are Blocked by ICAN Antagonists...... 77 Figure 2-7 Brief Depolarizations Trigger Persistent Firing in Blanes Cells...... 79 Figure 2-8 Prolonged Hyperpolarization Stops Persistent Firing...... 82 Figure 2-9 Synaptic Stimulation Activates Blanes Cells and Evokes Long-Lasting Inhibition onto Granule Cells...... 84 Figure 3-1 Carbachol reveals an afterdepolarization in granule cells...... 111 Figure 3-2 Calcium-dependent afterdepolarizations in granule cells...... 113 Figure 3-3 Pharmacology of mACh receptor activation in granule cells...... 115 Figure 3-4 Increased concentrations of carbachol enhance granule cell excitability...... 117 Figure 3-5 Carbachol enhances granule cell output onto mitral cells...... 119 Figure 3-6 During Carbachol application, brief depolarizations in granule cells can trigger persistent firing...... 121 Figure 4-1: A schematic showing the olfactory bulb circuit...... 149
iv Acknowledgements
I would like to thank my thesis advisor, Dr. Ben Strowbridge, for the years of advice and painstaking instruction in electrophysiological methods, and neuroscience history. Since my days as an undergraduate in his laboratory, he has been shaping my critical thinking skills and pushing me to go the extra distance in pursuit of answering the interesting questions, and for that I will always be eternally grateful. I would also like to thank the members of my thesis committee, Dr. Hillel Chiel, Dr. Stefan Herlitze, Dr. Diana Kunze, and Dr. Iain Robinson for their assistance and helpful suggestions in my research.
v List of Abbreviations
ACSF: artificial cerebrospinal fluid AMPA: alpha-amino-3-hydroxy-5-methylisoxazole-4-propionate AMPAR: AMPA receptor AP: action potential 4-AP: 4-aminopyridine, fast K channel blocker BAPTA: O,O’-Bis(2-aminophenyl)ethyleneglycol-N,N,N’,N’-tetraacetic acid tetrapotassium salt D-APV: D-2-amino-5-phosphonovalerate, NMDA receptor antagonist cAMP: cyclic adenosine monophosphate cGMP: cyclic guanosine monophosphate EGTA: O,O’-Bis(2-aminoethyl)ethyleneglycol-N,N,N’,N’-tetraacetic acid, slow Ca chelator EPL: external plexiform layer EPSC: excitatory postsynaptic current EPSP: excitatory postsynaptic potential GABA: gamma-aminobutyric acid, the neurotransmitter in granule cells GC: granule cell GCL: granule cell layer IPSC: inhibitory postsynaptic current IPSP: inhibitory postsynaptic potential LOT: lateral olfactory tract, axons of mitral cells MC: mitral cell MCL: mitral cell layer NBQX: 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo[f]quinoxaline-7-sulfonamide, AMPA receptor antagonist NMDA: N-methyl D-aspartate NMDAR: NMDA receptor NMG: N-methyl D-glucamine OB: olfactory bulb TEA: tetraethylammonium TTX: tetrodotoxin, blocks voltage activated Na channels VDCC: voltage dependent Ca2+ channel
vi MECHANISMS OF SYNAPTIC AND INTRINSIC MODULATION OF GRANULE
CELLS IN THE RAT OLFACTORY BULB
Abstract
By
Richard Todd Pressler
Mitral cell activity during olfactory behavior varies with respect to spatial location in the olfactory bulb, and temporally evolves during olfaction (Kay and Laurent, 1999). The main inhibitory input onto mitral cells originates from granule cells, axonless interneurons in the olfactory bulb, which sculpt and pattern mitral cell output through dendrodendritic inhibitory synapses (Price and Powell, 1970b). This dendrodendritic microcircuit can mediate recurrent inhibition onto mitral cells (Jahr et al., 1980), as well as lateral inhibition between mitral cells (Isaacson and Strowbridge, 1998). Additionally, this microcircuit plays a large part in generating gamma-frequency oscillations in the olfactory bulb during olfactory behavior (Lagier et al., 2004). Little data exists on the modulatory inputs onto granule cells from other olfactory bulb interneurons and from brain regions outside of the olfactory bulb. Previous anatomical studies suggest that granule cells receive GABAergic inputs from other olfactory bulb interneurons (Price and
Powell, 1970b). Additionally, granule cells receive significant cholinergic input from the horizontal limb of the diagonal band (Price, 1969).
vii First, I defined the intrinsic electrophysiological properties and synaptic inputs of a class
of olfactory bulb interneuron that was previously unstudied. This interneuron, termed the
Blanes cell based on their morphological similarity to olfactory bulb interneurons
discussed in a classic neuroanatomical investigation (Blanes, 1890), have interesting intrinsic properties. Suprathreshold current injections can generate large amplitude afterdepolarizations in these cells. Additionally, these cells can enter a persistent firing mode following transient stimuli that is reset by membrane potential hyperpolarization.
The Blanes cell is GABAergic, and can provide persistent inhibitory inputs onto granule cells, which may modulate granule cell firing for minutes in duration. Next, I investigated the effect of cholinergic receptor activation in granule cells. Cholinergic input to the olfactory bulb, from the horizontal limb of the diagonal band, plays a large role in sensory discrimination (Doty et al., 1999, Linster et al., 2001) and olfactory perceptual learning (Wilson et al., 2004). Cholinergic receptor activation in granule cells reveals a large amplitude stimulus evoked afterdepolarization. This afterdepolarization enhances granule cell excitability as well as inhibition onto mitral cells. These two inputs onto granule cells have interesting functional consequences on granule cell firing and olfactory bulb output.
viii
Chapter 1
Introduction
1 Olfaction is the sense by which organisms parse their surroundings for molecular clues suspended in the air (or water) that enable them to retrieve important information about their environment. Olfaction is, phylogenetically, a very old sense (Shepherd and Greer,
1998), and is often crucial for the survival of many organisms. Olfaction serves to inform an organism about the locations and properties of potential food sources, approaching dangerous predators, and mates. Odors can also convey a large amount of information about the physical state of a prey organism, in regards to how sick or injured it is, and even its level of excitement or anxiety. Olfactory information can also provide a number of other informative characteristics to an environment, as most organic and inorganic objects give off scents that form distinct odors perceptions. At its core, the work of olfaction involves the transformation of a chemical depiction of an odor into a neural code, composed of a sequence of action potentials that accurately codes for this odor.
Basic circuit description of the olfactory bulb
The perception of olfaction begins with air- or water-suspended odorant molecules entering the nares and diffusing across mucus on the nasal epithelium to encounter the cilia of odorant receptor neurons. I will now provide a summary of a number of experiments that have contributed to this view. Bronshtein and Minor (Bronshtein and
Minor, 1977) showed that treating a frog’s olfactory mucosa with the detergent Triton X-
100 destroyed the cilia of the olfactory receptor neuron, and disrupted the electroolfactogram in these frogs (Bronshtein and Minor, 1977). Electrical activity in neurons can be recorded by placing an electrode into the body of neurons and observing
2 the changes in potential difference between the recorded potential in the electrode and that of a separate ground electrode. The electroolfactogram (EOG) is a measure of the electrical activity in populations of odorant receptor neurons following odor application.
Ottoson (Ottoson, 1956) was the first to coin the phrase “electroolfactogram”. He noted that this potential could be eliminated when odorant receptor neurons degenerated as a result of severing of severing their axons. Additionally, the recorded electrical activity was negative at the surface of the olfactory epithelium, which suggests that the olfactory receptor neurons were depolarizing in response to odorant application. Because
Bronshtein and Minor were able to show that Triton X-100 application selectively degenerated the odorant receptor neuron cilia and not the cell body, and these results were reversible when the odorant receptor neuron cilia regenerated, this experiment suggests that the olfactory receptor sites are located on the cilia of the olfactory receptor neurons.
Pace and colleagues showed that the odorant receptor itself is a G-protein coupled receptor (Pace et al., 1985). They were able specifically to dissect out the cilia of the olfactory receptor neurons, and then through differential centrifugation, they were able to obtain a cilia-enriched fraction. They found that this fraction had enhanced adenylate cyclase activity when compared to the rest of the brain, and the adenylate cyclase could be further enhanced by odorant exposure such as citral and L-carvone. Finally, they showed that the adenylate cyclase activity could be suppressed by GDPβS application.
Pace and colleagues were thus able to show that cilia from odorant receptor neurons in rat
3 olfactory epithelium contain receptor proteins activated by odorants that activate
adenylate cyclase and produce cAMP like other G-protein coupled receptors.
Nakamura and Gold made the first electrophysiological recordings from the cAMP activated channel thought to be downstream of the odorant receptors (Nakamura and
Gold, 1987). They examined excised patches of plasma membrane obtained from dissociated toad olfactory receptor neurons. The channels that are expressed on the plasma membrane are selectively activated by low levels of cyclic nucleotides (for cAMP and cGMP, K1/2 values are between 1.6 and 3.3 μM). The channel activated by the cyclic
nucleotides was cation-permeable, had a reversal potential of about 0 mV (-5 ± 3 mV;
mean ± S.D) in Ringer’s (Ringer, 1883) in the patch pipette and a solution similar to the
intracellular solution in the bath, and rectified at more positive potentials. With this
information-- the fact that the odorant receptors are expressed on the olfactory receptor
neuron cilia, and the high probability that they were G-protein coupled receptors that
depolarize olfactory receptor neurons via a cAMP activated cation conductance, Buck
and Axel were able to determine the gene family responsible for the olfactory receptor
proteins (Buck and Axel, 1991). Expressed on the surface of individual odorant receptor
neurons are many copies of a single G-protein coupled odorant receptor (Buck and Axel,
1991) that bind to particular epitopes on the odorant, and convert the binding of this
chemical ligand through second-massager interactions to neuronal depolarization and
action potential firing. There are approximately 1000 unique odorant receptor proteins
expressed in the rat olfactory bulb alone (Buck, 1996).
4 One of the first hurdles the olfactory system needs to overcome is the ability to code and
differentiate a set of different odorants that number in the hundreds of thousands (Mori
and Yoshihara 1995) into limited “coding space” provided by the olfactory receptor
family. To begin to solve this problem, single odorants activate many different types of
G-protein coupled odorant receptors. Malnic and colleagues (Malnic et al. 1999)
approached this problem by depositing mouse olfactory receptor neurons from the nasal
epithelium onto slides, and then loading these cells with fura-2, a calcium indicator. With this method they could test a series of odorants systematically. Then, after testing the evoked responses in these neurons to odors comprised of a single-class of alcohol or other aliphatic compounds, Malnic and colleagues performed single cell PCR on these neurons to ascertain the expressed odorant receptor. Multiple epitopes expressed on the odorant activate different receptors (Malnic et al. 1999) and because of some cross over in the selectivity of the receptors for specific odorant epitopes, it is not the case that an odor representation is simply the activation of one subclass of odorant receptor (Zhao et al. 1998). This view has been challenged recently for components of natural olfactory stimuli (Lin et al., 2006).
The odorant receptor neurons project to the olfactory bulb in a very stereotyped way
(Mori et al. 1999) (figure 1-1). Odorant receptor neurons send their axons to specialized structures in the olfactory bulb called “glomeruli”, which are essentially glial sheaths that contain the synapses that these axons make with the primary dendrites of mitral cells, the principal cells of the olfactory bulb. Odorant receptor neurons that express the same type of odorant receptor project to two glomeruli per olfactory bulb (Vassar et al. 1994).
5 Additionally, odorant receptor neurons are arranged into four zones in the nasal
epithelium, and this spatial arrangement is projected onto the olfactory bulb such that the two dimensional spatial arrangement is wrapped around the surface of the olfactory bulb
(Mori et al. 1985; Ressler et al. 1993; Vassar et al. 1993; Vassar et al. 1994). This glomerular mapping maintains the relative spatial location of these receptors, and is important for olfactory processing (Shepherd and Greer, 1998). As discussed previously, wide-ranging glomerular activity results from the input originating from large numbers of different odorant receptors. However, with natural stimuli, a recent investigation indicates olfactory encoding operates in a different manner. Lin and colleagues (Lin at al., 2006) have recently shown in mice that natural olfactory stimuli, such as bobcat urine or coffee, only activate a small number of glomeruli, which suggests that natural stimuli such as this selectively activates a small fraction of odorant receptors. Additionally, when imaging the activity in olfactory bulb glomeruli during the application of these same odors after molecular separation by gas chromatography so that only single odor components were applied to the odorant receptors at any given time, only one or a few glomeruli became activated by each odor component. This suggests that natural stimuli activate a small subset of odorant receptors, and not a broad amount of odorant receptors.
One alternate interpretation however, is that glomerular activation is not reflecting all of the odorant receptor activation; as the inhibitory synapses that exist between glomeruli
(Shepherd and Greer, 1998) could be sharpening the resulting glomerular activation.
Several distinct levels of processing occur in the olfactory bulb (figure 1-2). A number of ultrastructural studies examining the putative synapses via electron microscopy identified
6 the basic olfactory bulb circuit. At the glomerular level, olfactory receptor neurons make glutamateric synapses onto mitral cell primary dendrites. Additionally, there are classes of periglomerular and juxtaglomerular cells that modulate these synapses, as well as laterally inhibit the mitral cells of neighboring glomeruli. Mitral cells, in turn, project to olfactory cortex and other higher brain areas (Shepherd and Greer, 1998). However, mitral cells also have long secondary dendrites, which make multiple dendrodendritic synapses with olfactory bulb granule cells (Rall et al. 1966; Price and Powell 1970a;
Price and Powell 1970b; Price and Powell 1970d). Granule cells are small axonless
GABAergic interneurons, and are the most populous cell in the olfactory bulb (Price and
Powell 1970a).
In addition to these studies, a number of electrophysiological studies complement this classic ultrastructural evidence. These electrophysiological experiments were largely
based on theoretical, EM, and extracellular recordings (Rall et al. 1966) and
ultrastructural evidence (Price and Powell 1970b) suggesting that action potentials
backpropagating down the mitral cell secondary dendrite will cause excitatory transmitter
release from the mitral cell dendrite onto the granule cell gemmule. The granule cell
gemmule will then depolarize and release inhibitory transmitter back onto the original
mitral cell dendrite. The granule cell gemmule is a specialized dendritic process that is
larger than spines typically found on most CNS neurons. The gemmule is the process that
contains glutamate receptors as well as VDCCs and synaptic vesicles containing GABA.
Jahr and Nicholl (1982) were the first to provide conclusive evidence that mitral cell
dendrites were pre-synaptic to granule cells, and that granule cells provide a source of
7 recurrent inhibition onto mitral cells. They used sharp electrodes to record voltage transients in turtle mitral cells. When stimulating orthodromically (olfactory nerve stimulation) and antidromically (lateral olfactory tract stimulation) they observed IPSPs in mitral cells. Additionally, they could generate IPSPs in mitral cells following suprathreshold current injection, and when sodium channel based action potentials were blocked by tetrodotoxin application, direct stimulation in the mitral cell evoked large amplitude calcium spikes that were sufficient to create inhibitory potentials in mitral cells. Other investigators further defined the role of the dendrodendritic synapse in olfactory bulb processing. Using whole cell voltage clamp recordings from rat olfactory bulb slices, Issacson and Strowbridge (Issacson and Strowbridge, 1998), using NMDA receptor specific antagonists and magnesium-free ACSF were able to show that NMDA receptors activation is critical to dendrodendritic inhibition. Additionally, using paired mitral cell recordings, Issacson and Strowbridge were able to show that mitral cells, through interactions with granule cell mediated dendrodendritic synapses, can laterally inhibit other mitral cells. Additionally, Halabisky et al. (2000), showed that the calcium entry through granule cell NMDA receptors is sufficient to evoke GABA release at dendrodendritic synapses.
Investigations of olfactory bulb activity during olfaction indicate an olfactory stimulus also evokes temporally dynamic activity among olfactory bulb neurons. In vivo recordings from olfactory bulb mitral cells in rats (Friedrich and Laurent 2001) and antennal lobe projection neurons (analogous olfactory structures in insects) in locusts
(Laurent et al., 1996) show significant changes in activity during, and after, odorant
8 application. The activity in these neurons during olfactory behavior shows a dynamic and temporal evolution of action potentials that can change many times during the odor application as well as continuing to evolve for many seconds after the termination of the odor. The activity of mitral cells was not sharper than the activity of glomeruli, and the sparseness of mitral cell activity failed to increase over time (Friedrich and Laurent
2001). This result is in stark contrast to what one would expect if the function of the olfactory bulb were to take an existing spatially oriented olfactory input and merely
sharpen it via granule cell mediated lateral inhibition.
Olfactory behaviors generate synchronized high frequency activity oscillations in
olfactory bulb neurons (Laurent, et al. 2001; Kay 2003). These high frequency oscillations were originally described by Lord Adrian (Adrian 1950) in olfactory bulb recordings done from anaesthetized rabbits stimulated with odors. He found that odorant stimulation not only evoked increased rates of firing in olfactory bulb neurons, but also produced synchronized local field potential oscillations during the duration of the odorant stimulus. Indeed, the inhibitory currents mediated by granule cell dendrodendritic inhibition also mediate high frequency oscillations important for mitral cell action potential patterning and olfactory coding (Lagier et al. 2004). Lagier and colleagues demonstrated this in rat olfactory bulb slices by creating synchronized local field potential oscillations with stimulating in the olfactory nerve layer. Then, they were able to show that in olfactory bulb slices that the synchronized local field potential oscillations
existed in slices that contained large cuts in them, severing the granule cell dendrites (and
the location of the dendrodendritic inhibitory synapse) from the granule cell layer. These
9 results suggest that the reciprocal and lateral inhibition in granule cells may perform
other functions besides sharpening olfactory input.
Stopfer and colleagues trained honeybees in an odor identification task, and recorded
oscillations in the antennal lobe during the odorant assessment. When picrotoxin (a
GABA receptor antagonist) was superfused into the antennal lobe shortly before the odor
assessment, it prevents the synchronization of these oscillations while not disturbing the temporal activity evolution in the principal output neurons. During periods of
GABAergic synaptic transmission blockade by picrotoxin, odor evoked activity in these structures becomes markedly desynchronized and these organisms suffer from poor odor discrimination among structurally similar odorants (Stopfer et al. 1997). Additionally, picrotoxin superfusion prevented the discrimination of similar odors (for example, 1- hexanol vs 1-octanol) while not affecting the level of discrimination between structurally dissimilar odors. Thus, local interneuron mediated inhibition onto principal cells directs field potential oscillations in second order olfactory structures, and the local interneuron
mediated inhibition is important for olfactory encoding.
Diversity of interneurons in the olfactory bulb
Aside from granule cells and cells in the glomerular layer, there are other olfactory bulb
interneurons whose contribution to the olfactory bulb circuit is unknown. Many
morphologically distinct interneuronal subtypes have been demonstrated in the
mammalian olfactory bulb using Golgi and immunocytochemical methods. At least seven
10 non-granule cell types exist in the granule cell layer alone (Golgi cells, Blanes cells, vertical cells of Cajal, horizontal cells, bi-tufted neurons, deep stellate cells, and deep short axon cells; (Schneider and Macrides 1978; Lopez-Mascaraque, De Carlos et al.
1986; Kakuta, Oda et al. 1998; Alonso, Brinon et al. 2001). The morphology of these
cells were described in early neuroanatomical studies performed by Ramon y Cajal and
colleagues in the late 1800s.
The seven different cell types in the granule cell layer have different dendritic
morphologies, somata locations, and axonal projections. Blanes cells resemble multipolar
“short-axon cells” (figure 1-3; Lopez-Mascaraque et al., 1986; Schneider and Macrides,
1978). Short axon cells were named because these neurons had incompletely filled axons
in the original neuroanatomical studies in Golgi-impregnated sections. Golgi staining
typically doesn’t label the entire axonal process, however, and when visualized with other
intracellular labels (such as biocytin) the axons of these neurons can actually be quite
large, so this name is somewhat of a misnomer. “Short axon” cells in the glomerular layer
actually have quite long axons, as illustrated in a recent study that employed direct
intracellular fills (Aungst et al., 2003). These cells have multiple dendrites radiating out of the soma in a stellate fashion. The dendrites of these neurons are very spiny. Blanes cells differ from granule cells in that they have an axon, and the axon that emanates from these neurons typically originated from the proximal portion of one of the larger dendrites, or from a region of the soma in very close proximity to the largest dendrite.
11 Golgi cells are grouped in the multipolar “short-axon” category and are very similar in
most respects to Blanes cells, with the major exception being their dendrites are mostly spineless (Schneider and Macrides 1978; Lopez-Mascaraque et al., 1986). Additionally, most of the dendritic processes of these cell types are restricted to the granule cell layer, but in rare cases, the distal portion of the axon has passed through the mitral cell layer and has entered the deep portion of the EPL (Schneider and Macrides 1978).
Vertical cells of Cajal are another type of olfactory bulb interneuron, and these cells somata are more localized near the mitral cell layer, almost at the border between the mitral cells and the granule cells. These neurons have dendrites that are oriented perpendicular to the dendrites of mitral cells. These cells have dendrites that are sparsely spiny in contrast with those of Blanes cells and Granule cells. Vertical cells of Cajal also possess an axon that extends into the EPL, and does not ramify over large distances
(Schneider and Macrides, 1978; Alonso et al., 2001). Horizontal cells differ from vertical cells of Cajal in that they possess horizontally oriented dendrites (as the name suggests) that extend into the mitral cell layer and very shallow portions of the granule cell layer.
These cells are similar to vertical cells of Cajal in that they also project axons in a very similar manner as those neurons, in that their axon extends into the EPL, and does not ramify over large distances (Schneider and Macrides, 1978; Alonso et al., 2001).
Horizontal cells have sparsely spiny dendrites, and their dendritic extents are similar and their somata are spatially restricted to the same locations as vertical cells of Cajal. The dendritic and axonal extent of vertical cells of Cajal and horizontal cells are spatially restricted when compared to those of Blanes and Golgi cells. Deep stellate cells possess
12 somata located in a deeper sublayer of the granule cell layer (Alonso et al., 2001). At a
first approximation they resemble Golgi or Blanes cells, except they have a significantly
reduced dendritic extent, and in comparison, a much larger axonal projection. Except for
the number of primary dendrites, which is lower in deep short-axon cells, deep short- axon cells are very similar to deep stellate cells-- these cells are very small bipolar neurons with short, stubby dendrites. The classic neuroanatomical studies performed by
Ramon y Cajal (Ramon y Cajal, 1995) also describe a number of bi-tufted and superficial granule cell layer stellate cells that are absent from recent reports describing the morphological characteristics of neurons in the olfactory bulb.
Another method to differentiate populations of interneurons is to define them on the basis
of calcium-binding protein expression. The binding of antibodies specific to calcium-
binding proteins can identify subpopulations of GCL interneurons from the more common granule cells, and can link gene expression to cell morphology. Parvalbumin staining has been especially useful in discriminating horizontal cells, bi-tufted cells, vertical cells of Cajal and Golgi cells from granule cells (Kakuta et al. 1998).
Neuropeptide-Y, NADPH, and somatostatin staining have also been useful at identifying short-axon cells, Golgi cells, Blanes cells, and granule cells (Davis et al., 1982; Scott et al., 1987; Alonso et al., 2001). In a number of these immocytochemical investigations, it was difficult for the authors to differentiate between the Golgi cell and the Blanes cell.
These cell types have similarly sized somata and dendritic morphologies, but Blanes cells have far more dendritic spines and slightly different axonal projections. Peptide immuno- staining is well suited to label the soma and proximal dendrite, but is unable to resolve
13 fine structural details like the presence of dendritic spines and axon morphology. It’s also
interesting to note that the majority of the cell types in the granule cell layer, the granule cells, horizontal cells, vertical cells of Cajal, Golgi cells, Blanes cells, and deeper short axon cells all stained positively for neurocalcin, calbindin, and calretinin (Alonso et al,.
2001).
Local circuit synaptic interactions between these different types of granule cell layer neurons could have a number of interesting functional consequences. Granule cells receive inhibitory synapses from target cells originating from inside the olfactory bulb
(Price and Powell 1970a; Price and Powell 1970d), and Price and Powell suggest that at least a subset of these inputs originate from “short axon cells” in the granule cell layer.
Previous literature has indicated that these short axon cells resemble Blanes or Golgi cells
(Price and Powell, 1970b). These inputs target the proximal aspect of the apical dendrites
of granule cells, but the strength, and the specific neurotransmitter phenotype of these
inputs were unknown.
Centrifugal input to the olfactory bulb
The four principal inputs from extrinsic brain regions originate from the olfactory cortex,
locus ceruleus, raphe nuclei and a cholinergic basal forebrain region labeled the
horizontal limb of the diagonal band (Shepherd and Greer, 1998, Price, 1969; Price and
Powell, 1970b; Price and Powell, 1970c; Macrides et al., 1981; Luskin and Price, 1982;
Kasa et al., 1995). The olfactory cortex is the main brain region responsible for the
14 feedback extrinsic inputs to the olfactory bulb, accounting for roughly 90 percent
(Shepherd and Greer, 1998; Macrides et al., 1981) of the non-olfactory receptor neuron
extrinsic inputs. These recurrent fibers tend to synapse most in the granule cell layer, onto
local interneurons (Price and Powell 1970c; Price and Powell 1970d). These fibers may
mediate contextual information about odors, and the olfactory cortex sends recurrent
connections back to the olfactory bulb to influence odor processing. A study by Kay and
Laurent (Kay and Laurent, 1999) showed the power of odor related reward expectation to
alter mitral cell activity in an awake and behaving rat. When rats were conditioned to
expect a sucrose solution reward following the application of a specific odor, in vivo
mitral unit activity during the application of the odor decreased significantly immediately before drinking. When these same odors were later paired with a bitter tasting solution, mitral cell unit activity remained unchanged throughout the task. These behaviorally associated modulations in firing rate became behavioral indicators and predicted a rat trained in these tasks would attempt to consume the flavored liquid.
Projections from piriform cortex presumably mediate the activity disparity in mitral cells between these conditioning paradigms. However, other subcortical areas also provide a sizable portion of innervations to the olfactory bulb. The horizontal limb of the diagonal band sends the second largest amount of fibers to the olfactory bulb. This is primarily a cholinergic nucleus, though roughly 30% of the fibers that emanate from this location towards the olfactory bulb are GABAergic (Brashear, Zaborszky et al. 1986; Zaborszky,
Carlsen et al. 1986). Inputs from these nuclei make synapses onto granule cell dendrites and gemmules in the GCL and EPL, as well as synapsing onto periglomerular cells.
15 These nuclei, in addition to other cholinergic basal forebrain nuclei, are instrumental in regulating wake/ sleep cycles as well as attention. The locus ceruleus and raphe nuclei are noradrenergic and serotonergic respectively, and constitute a much smaller percentage of
extrinsic input to the olfactory bulb in comparison to olfactory cortex and the cholinergic
basal forebrain (Shepherd and Greer, 1998; Macrides et al., 1981). These other brain
regions send fibers that synapse onto cells in the granule cell layer, as well as targets
within the glomeruli themselves, such as mitral cell primary dendrites and olfactory receptor neuron axon terminals (Shepherd and Greer, 1998).
The cholinergic basal forebrain, in particular, the horizontal limb of the diagonal band
provides the second largest amount of input to the olfactory bulb (Macrides et al., 1981).
In a macroscopic behavioral sense, attention is the mental procedure responsible for
focusing on a specific episodic event or sensation (Sarter and Bruno 2000; Sarter,
Hasselmo et al. 2005). Attention is the cognitive direction required to optimize goal-
directed behavior and cognitive processes (Sarter et al., 2006). The ability to provide
selective attention to a specific stimulus, and the ability to perceive important but weak
stimuli, as well as process these sensory signals despite distracting stimuli or a noisy
signal, are some of the primary goals of attentional effort. In the following sections, I will
illustrate a few of the studies that provide evidence for the necessity of the cholinergic
basal forebrain in the normal functioning of attentional processing, including its role in
the enhancement of sensory inputs, and how it relates to olfaction in particular.
16 The cholinergic basal forebrain is the crucial element in generating attentional effort. To examine the role of the cholinergic basal forebrain in mediating attention, Muir and colleagues (Muir et al., 1994) investigated deficits in a visual discrimination task. They trained rats to identify a brief visual stimulus presented randomly in one of five locations and provided a food reward for correct responses. Then, they introduced selective lesions to a region of the cholinergic basal forebrain known to innervate the neocortex, the nucleus basalis, by giving the rats bilateral AMPA injections to this brain region, or, in the sham-operated rats, a phosphate buffer. The AMPA injections produced a 70% reduction of ChAT activity in this brain region, while leaving neighboring noncholinergic brain regions intact. Three weeks following the injections the rats with the selective legions performed significantly worse in the behavioral task in both choice accuracy and response latency. Additionally, in the lesion group, a systemic injection of physostigmine
(an anticholinesterase) 20 minutes prior to the behavioral, test significantly improved discrimination and performance in a dose-dependent fashion. This experiment was one of the first to show that basal forebrain cholinergic system is important for sustained attentional function. Cholinergic input also increased sensory cortical representation.
Penschuck and colleagues (Penschuck et al., 2002) conducted an investigation in rats to determine if the cortical representation of the input from a single whisker is affected by cholinergic receptor activation. They performed these experiments in vivo, and used intrinsic signal measurements to measure neuronal activation following whisker stimulation. They discovered that carbachol (a mAChR agonist) application to the exposed cortex led to an enhancement the whisker-representation significantly, and scopalmine (an MChR antagonist) application to the exposed cortex led to a reduction in
17 whisker-representation. The cholinergic basal forebrain modulates attentional
performance at the level of sustained-attention, and at the level of sensory-signal
processing.
Previous investigations into the effects of cholinergic input on olfactory bulb activity
have generated contradictory results. Inokuchi et al. indicated that stimulation in the
horizontal limb of the diagonal band inhibited most mitral cells while exciting most
granule cells (Inokuchi et al. 1987), and Kunze et al. reported the opposite result (Kunze
et al. 1992)—possibly confounding these investigations are the mixed transmitter
phenotypes present in the horizontal limb of the diagonal band.
As discussed above, neuroanatomical studies indicated that there are large amounts of
cholinergic innervation to granule cells from the horizontal limb of the diagonal band,
and, granule cells also express muscarinic acetylcholine receptors (Hunt and Schmidt
1978; Cortes and Palacios 1986; Crespo, Blasco-Ibanez et al. 2000). Investigations involving whole cell recordings from granule cells during application of muscarinic acetylcholine agonists are sparse, and one study by Castillo et al. (Castillo et al. 1999) described granule cell inhibition by carbachol- though one caveat in this study was the relatively high concentration of carbachol used.
Granule cells express a large amount of M1 receptors, which often mediate excitation in other neurons by reducing specific potassium conductances (Dutar and Nicoll, 1988), I sought to examine the effect of activating those receptors. Granule cell dendrites are
18 densely spiny and extend far into the EPL where they synapse onto many mitral cells.
The modulation of granule cell activity could then greatly affect olfactory bulb output.
While the precise role of cholinergic input to the olfactory bulb is yet unknown, Ravel
and colleagues have shown that if cholinergic activity in the olfactory bulb is blocked,
short-term odor learning is impaired (Ravel et al. 1992; Ravel et al. 1994), suggesting
cholinergic input to the olfactory bulb may also mediate a form of short term olfactory
memory. In this investigation Ravel and colleagues trained rats to perform a delayed
matching or nonmatching to sample task where the rats had to discriminate between two
dissimilar odors. Before each trial, one of the two odors was randomly assigned as the
reward odor, and the rats were subjected to the target odor and then after a measured
delay (either 4 or 30 seconds), the rat was trained to choose a hallway corresponding to that target odor and received a food reward. If the rat chose incorrectly, no reinforcement was delivered. Once the rats learned the task, one group of rats received an infusion of scopolamine (a mAChR antagonist) in both olfactory bulbs 20 min prior to the start of the task, while another group received an infusion of saline. Additionally, in some of the rats that received scopolamine infusions, the cannula was placed in the deep granule cell layer, and in others, it was placed in the glomerular layer. When the delay period was set to 4 seconds, there was no difference between the control, sham, and scopolamine groups. However, when the delay period was set to 30 seconds, there was a significant reduction in the ability for the rat to choose the target odor correctly in the scopolamine infusion group with the cannulas placed in the deep granule cell layer. This finding is implicates mAChR activation in the olfactory bulb, specifically granule cell layer interneurons, is important for the short-term olfactory memory.
19
When cholinergic inputs to the olfactory bulb have been selectively lesioned, animals are
unable to differentiate between similar odorants (Linster et al., 2001). Linster and colleagues trained rats to expect a food reward underneath a small dish scented with the
test odor, and to choose that dish over another dish in the test area that was unscented. All
of the odors used in this study were either alcohols, aldehydes, or fatty acids that differed
in carbon length. In the test trials, the rats were given an unscented dish, and a dish
scented with a compound of a different carbon length than the test odor. Throughout
these different trials, Linster and colleagues recorded the time duration that the rats spent
digging underneath the respective dish. Then, one group of rats received a bilateral
injection of 192 Ig-saporin in the horizontal limb of the diagonal band region to
selectively lesion the cholinergic neurons projecting to the olfactory system, while
another group received a saline injection. Two weeks after the injections, Linster and
colleagues resumed the behavioral task, and observed that lesioned rats behaved much
like the sham operated rats, except they spent significantly more time digging under structurally similar odors-- especially odors that differed in only one or a few carbon atoms. The principal conclusion of this study was that a reduction in the cholinergic input to the olfactory system increases odorant generalization. Understanding how cholinergic receptor activation affects granule cell firing properties will greatly enhance our understanding of how this particular type of input modulates the activity in this brain region.
20 Inputs that originate from brain regions that are extrinsic to the olfactory bulb play a large role in shaping olfactory coding. The ability of the olfactory bulb to encode odorant sensory information, and indeed, in olfactory behaviors in general, are highly regulated by attention, mood, behavioral state, and context dependence. (Kay and Laurent 1999;
Sarter and Bruno 2000; Sarter et al. 2005) In addition, olfaction is also highly involved in regulating, in turn, all of these other behavioral states. Olfactory exploratory behaviors are heavily influenced by attentional factors, and this is especially obvious to anyone who has ever seen a dog sniff a novel object rich in olfactory diversity like some gym socks or
rotting vegetable matter. These inputs from other brain regions are responsible for
altering the parameters of olfactory sensory coding during these different behavioral and
contextual states, and during these different states information encoding changes in a
context dependant or state dependant manner.
In other systems, small numbers of modulatory fibers such as these cholinergic fibers
provide a significant effect on circuit modulation. Previous investigators have produced
some very interesting work describing the ability of a relatively small number of extrinsic
inputs to affect greatly the output of a brain region by modulating the activity of neurons
that impinge on large numbers of principal output cells. Freund (Freund and Antal, 1988)
provided neuroanatomical evidence showing how a small number of fibers entering the
hippocampus could provide a large functional effect on hippocampal output. Freund was
able to show that relatively sparse septal GABAergic inputs form basket-like arrays of en
passant synapses with the soma and proximal dendrites of GABA-containing
interneurons in the hippocampus. Because of the location of the GABAergic synapses
21 onto these interneurons, septal inhibitory projections tightly control the output of these neurons. The GABA containing interneurons participated in providing inhibitory input onto excitatory principal cells and regulated the activity at three locations in hippocampus—the CA1, the CA3 and the dentate gyrus. The septum therefore has a critical role in determining hippocampus activity, as the phasic inhibition originating from the septum can provide discrete domains of disinhibition onto hippocampal output neurons, which would increase principal cell excitability and may facilitate LTP in these cells. In the olfactory bulb, the dendrodendritic synapse mediates lateral inhibition between mitral cells (Isaacson and Strowbridge, 1998), as well as reciprocal inhibition of mitral cell dendrites (Jahr and Nicoll, 1982). The dendrodendritic microcircuit is malleable to influence from modulatory inputs from other interneurons receiving input from mitral cells, or from neurons outside of the olfactory bulb. The impact of these factors remains to be a yet unanswered and potentially important question.
The primary focus of this thesis is to investigate experimentally two different inputs onto granule cells—inhibitory inputs from other olfactory bulb interneurons, and cholinergic inputs from the horizontal limb of the diagonal band. I want to determine electrophysiologically their modulatory effect on granule cell intrinsic properties, and their role in shaping olfactory bulb activity. As discussed previously, neuroanatomical evidence suggests that olfactory bulb interneurons make inhibitory projections onto granule cells (Price and Powell, 1970d). First, I set out to determine the morphological properties of the inhibitory interneuron that synapses onto granule cells. I also wanted to examine the electrophysiological properties and synaptology of this interneuron so I
22 could understand its position in the circuit framework of the olfactory bulb. This work
has recently been published (Pressler and Strowbridge, 2006). Next, I sought to
determine the intrinsic electrophysiological effect of chonlinergic input to granule cells
on granule cell excitability. I also wanted to examine the effects of this cholinergic modulation of granule cell activity on mitral cell input, to understand how cholinergic
input to granule cells affects olfactory bulb activity.
23
Figure 1-1 From Mori et al., 1999
24 Figure 1-1: A schematic showing the spatial mapping of odorant receptors onto olfactory bulb glomeruli. All the odorant receptors expressing the same odorant receptor protein express to two glomeruli per olfactory bulb, and the location of the glomeruli conform to the spatial map of the odorant receptors in the nasal epithelium.
25
Figure 1-2 From Mori et al., 1999
26 Figure 1-2: A schematic showing the canonical olfactory bulb circuit. Olfactory sensory neurons in the nasal epithelium innervate mitral cells in specialized structures called glomeruli in the olfactory bulb. At the level of the glomeruli, synaptic interactions between mitral cells and peri-glomerular cells preprocess the olfactory information.
Mitral cells also make dendrodendritic synapses with granule cells, small axon-less interneurons. Mitral cells then send axons to higher order olfactory areas. Note the absence of the other interneuronal subtypes in the olfactory bulb in this representation. In this figure, OSN labels the olfactory sensory neurons, GL labels the glomeruli layer, PG labels periglomerular cells, T labels tufted cells, M labels mitral cells, and Gr labels granule cells. Excitatory synapses are white arrows, and inhibitory synapses are black arrows.
27
Figure 1-3 From Blanes, 1890
28 Figure 1-3: A figure showing one of the other olfactory bulb interneurons, the Blanes cell, labeled with a “d” in this figure, which is of interest in this story. Note the presence of spiny dendrites and a presumptive axon.
29
Chapter 2
Blanes Cells Mediate Persistent Feedforward Inhibition onto Granule Cells in the
Olfactory Bulb
30 Summary
Inhibitory local circuits in the olfactory bulb play a critical role in determining the firing patterns of output neurons. However, little is known about the circuitry in the major plexiform layers of the olfactory bulb that regulate this output. Here we report the first electrophysiological recordings from Blanes cells, large stellate-shaped interneurons located in the granule cell layer. We find that Blanes cells are GABAergic and generate large ICAN-mediated afterdepolarizations following bursts of action potentials. Using paired two-photon guided intracellular recordings, we show that Blanes cells have a presumptive axon and monosynaptically inhibit granule cells. Sensory axon stimulation evokes barrages of EPSPs in Blanes cells that trigger long epochs of persistent spiking; this firing mode was reset by hyperpolarizing membrane potential steps. Persistent firing in Blanes cells may represent a novel mechanism for encoding short-term olfactory information through modulation of tonic inhibitory synaptic input onto bulbar neurons.
31 Introduction
Local circuits are the fundamental building blocks that define how different CNS regions process afferent information. Local circuits often are described in terms of specific patterns of synaptic interactions (e.g., feedback and feedforward inhibition, recurrent excitation, lateral inhibition) that are iterated across modules within a brain region
(Douglas and Martin, 1991, Shepherd and Greer, 1998 and Stepanyants et al., 2004). In the olfactory bulb, the second-order olfactory brain region, sensory information is relayed from olfactory receptor neurons to both principal neurons (mitral and tufted cells) and to a variety of glomerular-layer interneurons (Shepherd and Greer, 1998). The computational work of the olfactory bulb presumably is mediated in large part by stereotyped synaptic interactions that take place in the inner and outer plexiform layers and that are replicated in the local circuits linked to the 1800 olfactory glomeruli that
receive olfactory receptor subtype specific afferents (Mori et al., 1999 and Shepherd and
Greer, 1998). However, little is known about the local circuits that operate in these
plexiform and output layers of the olfactory bulb, aside from the reciprocal
dendrodendritic synapse formed between mitral cells and axonless granule cells (Jahr and
Nicoll, 1980, Price and Powell, 1970b and Rall and Shepherd, 1968). This dendritic
microcircuit mediates both feedback and lateral inhibition (Isaacson and Strowbridge,
1998 and Schoppa et al., 1998) of mitral cells but is strongly dependent on NMDAR
activation and therefore is tonically attenuated by extracellular Mg2+ ions. Conventional
inhibitory local circuits that involve transmitter release from axon terminals have not
32 been demonstrated electrophysiologically in the major plexiform layers of the olfactory bulb.
This paucity of functional data regarding the local circuitry in most of the olfactory bulb contrasts with the abundance of anatomical data that demonstrates the presence of a wide variety of presumptive interneurons in this brain region. In the granule cell layer alone, at least seven types of non-granule cells have been demonstrated using Golgi and immunocytochemical methods (Alonso et al., 2001; Kakuta et al., 1998; Lopez-
Mascaraque et al., 1986; Schneider and Macrides, 1978). The electrophysiological properties and synaptic connections made by these interneurons have not been explored previously. Recent work on interneuronal subtypes in the glomerular input layer (Aungst et al., 2003) has significantly revised the canonical basic circuit for the olfactory bulb, which now includes long-range (interglomerular) excitatory synaptic projections made by so-called “short axon” cells. Understanding the functional consequences of interneuronal diversity in the olfactory bulb is a critical step toward the goal of deciphering the computational operations performed by this major brain region.
In this study we sought to define the properties and synaptic connections made by one subpopulation of interneurons that is readily identifiable under infrared differential interference contrast (IR-DIC) microscopy: large stellate-shaped neurons in the granule cell layer. We used whole-cell patch-clamp recording and two-photon imaging in acute brain slices to determine whether these cells represent a novel cell type that is anatomically and electrophysiologically distinct from granule cells. We found that these
33 stellate-shaped cells are a GABAergic cell type that innervate granule cells and generate large depolarizing afterpotentials following transient stimulation. We refer to these neurons as “Blanes cells,” reflecting their morphological similarity to cells first described in Golgi-stained olfactory bulb sections (Blanes, 1898). We found that brief depolarizing stimuli trigger persistent activity in Blanes cells that can be reset by membrane potential hyperpolarization. Persistent spiking in Blanes cells is a previously unrecognized feature of the olfactory bulb and may represent an intrinsic form of short-term memory that enables brief sensory experiences to modulate activity in local circuits over a time span of seconds to minutes.
34 Results
In addition to granule cells, IR-DIC microscopy revealed other neurons in the granule cell layer of rat olfactory bulb slice with cell bodies that are much larger than granule cells
(mean size viewed under IR-DIC illumination = 20.0 × 12.6 μm for large cells and 11.8 ×
9.7 μm for granule cells). A subpopulation of these large cells—termed Blanes neurons, reflecting their dendritic morphology (see below)—were present throughout the extent of the granule cell layer (GCL) and often were located between stripes of granule cells
(Figure 2-1A). We used whole-cell patch-clamp recording to determine whether these large cells represent a novel cell type. Current-clamp recordings from large cells located in the granule cell layer (n = 547 cells in this study) typically showed large-amplitude overshooting action potentials (63.0 ± 1.5 mV; n = 26) and high input resistances (260 ±
28 MΩ; n = 19), suggesting that they represent a potentially novel cell type and not a subpopulation of injured granule cells.
Large cells in the GCL differed from nearby granule cells in their intrinsic electrophysiological properties (Figure 2-1B). The mean input resistance of Blanes neurons was significantly lower than granule cells (260 ± 28 MΩ versus 463 ± 63 MΩ; p
< 0.001). The mean action potential threshold in Blanes cells was also significantly more hyperpolarized than granule cells (−41.1 ± 0.8 mV, n = 26 versus −27.2 ± 1.3 mV for granule cells, n = 7; p < 0.001). The mean resting membrane potential in Blanes cells was more depolarized than that of granule cells (−53.9 ± 1.1 mV, n = 22 versus −65.7 ± 2.1 mV, n = 12; p < 0.001). Other intrinsic properties were similar in Blanes cells and
35 granule cells, including action potential width (2.09 ± 0.09 ms, n = 26 versus 2.09 ± 0.16
ms, n = 7 granule cells) and the amplitude of the afterhyperpolarization following a single
action potential (16.2 ± 0.8 mV, n = 26 large cells versus 18.1 ± 2.9 mV, n = 7 granule
cells). A subpopulation of large cells (23%) showed a pronounced depolarizing notch
during the spike afterhyperpolarization (Figure 2-1B, inset). Other large cells showed a
less prominent inflection at the same time following the spike; granule cells never displayed depolarizing notches or inflections following action potentials. Both granule
cells and Blanes cells fired tonically with depolarizing current injection.
We used the Neurolucidia reconstruction system (Microbrightfield, Williston, VT) and
two-photon microscopy to define the dendritic and axonal morphology of Blanes cells.
Twenty-two Blanes neurons were filled with neurobiotin (0.5%; Molecular Probes) through the patch pipette and processed using the ABC peroxidase method to reveal dendritic and axonal processes. All 22 visualized Blanes neurons exhibited afterdepolarizations and persistent firing following suprathreshold 500 ms depolarizing steps, two characteristics of Blanes neurons described below. Eight visually and electrophysiologically identified granule cells also were filled with neurobiotin and processed using the ABC method. Figure 2-1C illustrates the dendritic morphology of
one granule cell and two typical Blanes neurons. Occasionally, processes of Blanes cells
entered the mitral cell layer (see middle panel in Figure 2-1C; 3/9 reconstructed cells);
however, we never observed processes that entered the external plexiform layer (0/9
cells; Figure 2-1D1). Most granule cells visualized had an apical dendritic process that
ramified in the external plexiform layer and had frequent large spines associated with
36 reciprocal dendrodendritic synapses (Rall et al., 1966). Blanes neurons, by contrast,
rarely had a prominent apically directed process. Morphometric analyses revealed
significant differences between Blanes and granule cell soma size (117.7 ± 11.9 μm2
versus 47.4 ± 8.0 μm2; p < 0.001) and the number of primary dendrites (4.6 ± 0.6 versus
2.0 ± 0.3; p < 0.002; Figure 2-1D1). Blanes neurons and granule cells also differed in the
extent of their dendritic aborization as quantified by Sholl analysis (Figure 2-1D2).
Blanes cells generated significantly more of their dendritic aborization within 100 μm of the cell body than did granule cells. While we observed small dendritic spines in neurobiotin-filled Blanes cells, these spines were not as large or numerous as the spines on the distal dendrites of granule cells. Two-photon microscopy of Alexa594-filled
Blanes cells also revealed some spines on primary dendrites (Figures 3-2A–3-2C). Most
Blanes neurons had a presumptive axonal process that emanated from a primary dendrite and ramified in the granule cell layer (Figures 2D and 2E). Occasionally, the axonal plexus entered the mitral cell layer; we never observed presumptive axonal processes in the external plexiform layer. Axon-like processes typically branched at right angles, did not taper, and were devoid of spines under two-photon visualization (Figure 2-2D).
We used paired recordings to determine the transmitter phenotype and to define the synaptic targets of Blanes cells. Because both neurobiotin visualization and two-photon fluorescence imaging revealed a presumptive axon in Blanes cells that ramified in the granule cell layer, we focused our search for potential postsynaptic targets on granule cells. We utilized two-photon imaging to determine the specific region within the GCL where the Blanes cell axon bifurcates (see Experimental Procedures). We then tested 18
37 Blanes/granule cell pairs using the two-photon axon visualization method. Of these paired recordings, 17% (3 of 18) generated monosynaptic IPSPs onto granule cells with onset latencies <2.5 ms. The cell bodies of the three monosynaptic pairs that we found using two-photon-guided recordings were separated by 400, 560, and 1100 μm. We found no monosynaptic pairs that generated EPSPs onto granule cells, and we never observed a reverse projection (from granule to Blanes cell). We also recorded from 83
Blanes/granule cell pairs with cell bodies separated by <80 μm using conventional IR-
DIC visualization methods. Of these 83 pairs, we found one monosynaptically coupled
Blanes/granule cell pair (mean onset latency = 1.81 ± 0.04 ms; n = 96 PSPs).
Figure 2-3A illustrates a typical Blanes-to-granule cell paired recording. In this paired recording, single Blanes cell action potentials generated short latency hyperpolarizations
(mean latency = 1.51 ± 0.04; n = 20 IPSPs; Figure 2-3A) with a low (6%) failure rate.
The mean failure rate was 7.2% ± 2.8% in our population of four paired recordings. As shown in Figure 2-3B, Blanes cell-evoked IPSPs reversed polarity near the chloride equilibrium potential (reversal potential = −78.9 ± 0.4 mV; n = 3 postsynaptic cells), suggesting that Blanes cells make GABAergic synapses onto granule cells. The mean amplitude of the IPSP recorded in this granule cell at −55 mV was −1.8 ± 0.1 mV (Figure
2-3C1; n = 101 IPSPs). The average IPSP latency in our experiments was 1.74 ± 0.15 ms
(n = 4 paired recordings; range 1.46–2.13 ms) and the average amplitude at −55 mV was
−1.84 ± 0.04 mV (Figure 2-3C2; n = 3). We used the specific GABAA receptor antagonist gabazine (10 μM) to test this hypothesis. Gabazine reversibly blocked Blanes cell-evoked IPSPs in all three monosynaptic Blanes/granule cell paired recordings
38 (Figure 2-3D). In one Blanes/granule cell pair, we also confirmed that blockers of
ionotropic glutamate receptors (5 μM NBQX and 50 μM d-APV) did not affect the IPSP
evoked in action potentials in the Blanes cells (data not shown). These results
demonstrate that Blanes cells monosynaptically inhibit granule cells in the olfactory bulb
by activating GABAA receptors.
Afterdepolarizations in Blanes Neurons
We focused our study on a subset of large neurons in the GCL, which we term “Blanes
cells” based on their dendritic morphology (compare Figure 2-1 in Schneider and
Macrides, 1978, with Figure 2-1C in the present study). In addition to the morphological differences discussed in the previous section, Blanes neurons and granule cells differed in their responses to short trains of action potentials. While granule cells typically responded to a burst of action potentials with a large-amplitude afterhyperpolarization
(AHP), all Blanes cells visualized following intracellular staining (n = 46) generated an afterdepolarization (ADP) rather than an AHP following similar stimuli. In addition to generating ADPs, Blanes cells could fire persistently at depolarized membrane potentials
(see below). Across our entire population of large neurons in the GCL recorded under current-clamp conditions, 72% (361 of 498, including the 46 cells that were intracellularly stained) generated ADPs and could fire persistently. Preliminary studies of the subpopulation of large cells that showed an AHP rather than an ADP following bursts of action potentials revealed a range of different dendritic morphologies (e.g., long
39 horizontally or vertically oriented dendrites), suggesting that multiple interneuronal subtypes exist within the GCL (Schneider and Macrides, 1978).
Single spikes reliably triggered ADPs in Blanes cells; ADPs were graded in amplitude
with additional spikes (Figure 2-4A). Afterdepolarizations evoked by bursts of 8 to 12
action potentials also have different kinetics than ADP responses to single spikes (Figure
2-4A2). Afterdepolarizations were voltage dependent and were not present when the
Blanes neuron was held at −80 mV (tested with increased step amplitude to trigger a
similar number of action potentials evoked by steps at −60 mV; ADP amplitude
significantly different between −60 and −80 mV; 4.01 ± 0.45 mV and −0.20 ± 0.69 mV, respectively; p < 0.001; Figure 2-4B). When Blanes cells were held at more depolarizing potentials (−65 to −50 mV), ADPs triggered multiple action potentials in Blanes cells.
The firing frequency in the resulting train decremented during the response (see left panel on Figure 2-4B1), suggesting that Blanes cell spiking was driven directly by the ADP. A different firing mode (discussed below) characterized by persistent, nonadapting spiking also was observed in Blanes cells following stronger stimulation.
Afterdepolarizations in Blanes cells were Ca dependent and were reversibly blocked by a low Ca (0.5 mM), high Mg (6 mM) external solution (ADP reduced to 3.1% of control; n
= 4; p < 0.02; Figure 2-4C). We used the inorganic channel blocker Cd to determine whether Ca influx through voltage-dependent Ca channels (VDCCs) is required to trigger an ADP. In all five Blanes cells tested, 200 μM Cd blocked the ADP (ADP reduced to
−27.7% ± 11.6% of control; n = 4; p < 0.001; Figure 2-4D) and revealed a residual
40 afterhyperpolarization. The ADP was unaffected by the L-type VDCC blockers
nifedipine (50 μM; 89.4% ± 4.6% of control; n = 5 cells) and nitrendipine (20 μM; 90.0%
± 7.0% of control; n = 5 cells). The ADP also was unaffected by the L-type VDCC
modulator S-(-)-Bay K 8644 (1 μM; 96.0 ± 7.0; n = 5 cells). We then used EGTA to
determine whether the ADP resulted from an intracellular signaling action of Ca or
instead reflected an inward current mediated by voltage-dependent Ca channels. Five
Blanes cells were recorded using pipettes whose tips contained low EGTA solution (0.2
mM) and were backfilled with the same internal solution supplemented with 10 mM
EGTA. In all five cells, depolarizing steps evoked ADPs immediately after break-through
into the whole-cell recording mode. These responses gradually declined over 30–40 min
until they reached steady-state levels (at 5.9% ± 17% of control; Figure 2-5A). The ADP
response showed only a modest decay (to 83% ± 9% of control amplitude; n = 3 cells)
over a similar recording period (mean 40 min) in interleaved control experiments using
electrodes backfilled with the low EGTA internal solution. The ADP amplitude in high
EGTA experiments was significantly reduced compared with control experiments using
0.2 mM EGTA at 40 min (p < 0.01; summarized in inset in Figure 2-5A). These results
suggest that the ADP was due to an EGTA-sensitive Ca signaling pathway activated by
Ca entry through a non-L-type VDCCs.
Next, we used the photolyzable Ca chelator NP-EGTA (Ellis-Davies and Kaplin, 1994 and Hall and Delaney, 2002) to test whether a transient increase in intracellular Ca is
sufficient to trigger an afterdepolarization in Blanes cells. In a majority of the neurons
with large cell bodies recorded in the granule cell body layer (18 of 34), brief exposure to
41 UV illumination (400–2000 ms) evoked an inward current when the cell was voltage clamped near rest (−60 to −80 mV). In a minority of large cells, UV exposure evoked an outward current (10 of 34 cells) or no response (6 of 34 cells), suggesting that there may be multiple subtypes of large neurons in the granule cell layer. The mean amplitude of the response to a 2 s photolysis of NP-EGTA at −40 mV was −16.2 ± 2.8 pA (n = 12 cells
with inward currents). As shown in Figure 2-5C1, varying the duration of light exposure
from 100 to 1000 ms generated a graded series of inward currents. Two second light
exposures to a Blanes cell recorded with a control internal solution that did not contain
NP-EGTA failed to generate a response (n = 5 cells; Figure 2-5C2). The light-activated
inward current was associated with a 68.0% ± 7.0% decrease in input resistance
measured using small depolarizing test pulses applied during maximal inward current (10
mV, 100 ms duration; n = 5 cells; p < 0.01; data not shown). The light-activated response
reversed polarity at −10.7 ± 5.8 mV (n = 9 cells; Figure 2-5D1) and had a modest voltage
dependency (Figure 2-5D2), consistent with the properties described previously for the
Ca-activated nonselective cation current (ICAN) in other cell types (Haj-Dahmane and
Andrade, 1998, Haj-Dahmane and Andrade, 1999 and Hall and Delaney, 2002).
Our results thus far suggest that a Ca-dependent response underlies the
afterdepolarization in Blanes cells. We next sought to determine whether this response
reflects the activation of a Ca-dependent inward current (e.g., ICAN) or is mediated by a
transient decrease in a resting K current. Depolarizing responses mediated by transient
reductions in K currents are highly sensitive to changes in extracellular K concentration
(Li and Hatton, 1997). As shown in Figure 2-6A, doubling the extracellular K
42 concentration (from 3 to 6 mM) did not affect the amplitude or time course of the ADP
response in Blanes cells (98.4% ± 2.9% of control; n = 5; tested after adjusting the DC
current to return the neuron to the original resting membrane potential), suggesting that
the ADP is not mediated by a decrease in K currents. Increasing extracellular K
concentration did reduce the fast afterpolarization that immediately follows the step
offset (Figure 2-6A, arrow), suggesting that multiple responses probably shape the
afterdepolarization responses in Blanes cells. We also found that the specific M-current
blocker XE-991 (30 μM) depolarized Blanes cells but did not affect the burst-evoked
ADP (101.2% ± 9.9% of control; n = 8; data not shown). Calcium spikes evoked in the
combination of TTX (1 μM), TEA (25 mM), and 4-AP (100 μM) also triggered ADPs
(Figure 2-6B, middle trace), which displayed the same voltage dependence as ADPs evoked under control conditions (n = 12). These results suggest that the ADP was not mediated by a decrease in standing outward currents. In contrast with the results obtained by altering [K], reducing extracellular Na ions by partially substituting with the impermanent ion NMG (99 mM) strongly attenuated ADPs triggered by Ca spikes in
Blanes cells (9.7% ± 14.5% of control; n = 7; p < 0.001; Figure 2-6B). The reduction in
ADPs in low Na ACSF was fully reversible upon washout of the NMG solution (data not shown). Low Na ACSF also increased the amplitude of the fast AHP that immediately followed the step offset, suggesting that ADP and this hyperpolarizing response overlap.
Nonselective cation currents often mediate spike- and burst-evoked afterdepolarizations in neurons in other brain areas (Egorov et al., 2002, Haj-Dahmane and Andrade, 1998,
Hall and Delaney, 2002 and Partridge et al., 1994). We used flufenamic acid (FFA), a
43 partially selective ICAN blocker (Ghamari-Langroudi and Bourque, 2002 and Partridge
and Valenzuela, 2000), to test whether Blanes neuron ADPs are mediated by ICAN.
Because preliminary experiments demonstrated a nonspecific action of FFA on fast Na currents, we tested FFA on ADPs evoked by short trains of Ca spikes in TTX, 4-AP, and
TEA. Under these conditions we found that 200 μM FFA significantly reduced the ADP amplitude (to 31.0% ± 13.5% of control; n = 7; p < 0.002; Figures 6C and 6D). As shown in Figure 2-6C, this action was fully reversible upon washout of FFA. Flufenamic acid decreased the number of Ca spikes evoked by the current step. However, afterdepolarization amplitude was still significantly reduced by FFA (to 61.6% ± 6.3% of control; n = 7; p < 0.001) even after we controlled for this action by increasing the current step amplitude to trigger the same number of Ca spikes (Figure 2-6C). Together, these data point to ICAN as the underlying mechanism responsible for ADPs in Blanes
cells. However, the absence of selective ICAN antagonists or applicable molecular tools
precludes definitively ruling out other potential Ca-sensitive mechanisms for the ADP.
Persistent Spiking in Blanes Neurons
In addition to evoking Ca-dependent afterdepolarizations, depolarizing current steps also
triggered prolonged periods of tonic firing in Blanes neurons held near firing threshold.
The transition between afterdepolarization, which could trigger several irregularly spaced
action potentials, and persistent firing was governed by both membrane potential (Figure
2-7A) and stimulus amplitude (Figure 2-7B1). Near spike threshold, the duration of the
ADP-triggered spike train was highly sensitive to membrane potential and increased as
44 the cell was held closer to firing threshold (Figure 2-7A). Depolarizing stimuli that
reliably triggered persistent firing in Blanes cells failed to elicit prolonged firing in
granule cells and instead evoked afterhyperpolarizations (Figure 2-7B2; n = 53 cells). The
frequency of persistent firing in a Blanes cell was relatively invariant and not affected by
changes in the step amplitude (Figure 2-7C1). While we occasionally observed persistent
firing triggered by single action potentials, in most cells a 100–500 ms duration
suprathreshold depolarizing step was required to trigger persistent firing. The frequency
of persistent firing varied considerably among Blanes neurons (Figure 2-7C2; mean firing
frequency = 6.5 Hz; range 3.1–13.4 Hz; n = 61 cells). However, the frequency of tonic
firing was relatively constant with repeated trials in the same cell (CV = 0.10 ± .01; n =
13 cells). Persistent firing also could be triggered by uncaging Ca in Blanes cells loaded with NP-EGTA (2 mM; Figure 2-7D). Exposure to UV light alone did not depolarize
Blanes cells or initiate persistent firing in control experiments without NP-EGTA added
to the internal solution (n = 3 cells; Figure 2-7D, bottom trace). As shown in Figure 2-7E,
persistent firing in Blanes cells evoked periodic IPSPs in monosynaptically coupled
granule cells (mean IPSP latency to single spikes = 1.54 ± 0.03 ms; mean IPSP latency
following individual spikes during persistent firing = 1.47 ± 0.05 ms).
Persistent firing was reversibly abolished in low Ca/high Mg ACSF (Figure 2-8A; n = 4)
and was unaffected by the L-type voltage-dependent Ca channel blocker nifedipine (50
μM; n = 5; data not shown). These results suggest that non-L-type voltage-dependent Ca channels contribute to the initiation or maintenance of persistent firing. The ability of low
Ca ACSF to block persistent firing was not likely due to blockade of synaptic
45 transmission, as persistent firing was unaffected by antagonists of iontropic glutamate (5
μM NBQX and 25 μM d-APV; n = 28) or GABAA receptors (10 μM gabazine or 50 μM picrotoxin; n = 4).
Once triggered, persistent firing typically persisted for many minutes before spontaneously returning to the pre-step membrane potential. The average duration of tonic firing was 16.5 ± 5.1 min (range 1.5–44.1 min; n = 9). As shown in Figure 2-8B1, tonic firing did not stop abruptly but rather gradually decreased over 1–2 min. This gradual decrease in average firing rate was due in part to increasingly frequent pauses between runs of spikes at the original firing frequency. Enlargements of these pauses often revealed small depolarizing peaks (Figure 2-8B2). The intervals between the
preceding action potential and these depolarizing peaks were similar to the intervals
between successive action potentials during the initial tonic firing phase, suggesting that
tonic firing frequency may be governed by the kinetics of the spike afterpolarization. We
recorded persistent firing lasting at least 10 s in the majority of Blanes cells tested systematically (58 of 79 cells). Persistent firing could generally be triggered when the
membrane potential was within 10–15 mV (mean 11.8 mV; n = 23 cells) below the action
potential threshold. Most Blanes cells we recorded had resting membrane potentials that
were within this window (mean RMP = −53.9 ± 1.1 mV; mean AP threshold = −41.1 ±
0.8 mV).
Persistent firing was not affected by brief hyperpolarizing current injections that
temporarily blocked spiking (Figure 2-8C; firing maintained in 10 of 10 cells after 1 s
46 duration hyperpolarizing pulses with a mean amplitude of −24 mV; firing maintained in 5
of 6 cells after 3–6 s duration pulses with a mean amplitude of −43 mV) but was blocked
by a prolonged period (60 s) of membrane potential hyperpolarization (Figure 2-8D;
persistent firing abolished in 9 of 10 trials). The average duration of persistent firing
following a 60 s hyperpolarization was significantly reduced compared with interleaved
control trials without hyperpolarizing current injection (Figure 2-8D2; firing duration =
1.43 min with hyperpolarization versus 16.5 min in control trials; n = 10 and 9,
respectively; p < 0.005). Additional experiments will be needed to understand why
prolonged periods of moderate hyperpolarization were more effective at blocking
persistent activity than brief, large-amplitude hyperpolarizing steps.
Excitatory synaptic input to Blanes cells was an effective trigger for persistent firing.
Blanes cells receive excitatory input that can be activated by extracellular stimulation in the granule cell layer (Figure 2-9A2; 11 of 11 cells tested) or glomerular layer (9A4; 10 of
10 cells tested). The average amplitude of the GCL-evoked EPSP in Blanes cells
recorded at −80 mV was 4.2 ± 0.6 mV in control conditions and 0.1 ± 0.2 mV in 25 μM d-APV and 5 μM NBQX (EPSP amplitude significantly reduced; p < 0.001; n = 4 cells).
The mean EPSP onset latency was short (1.70 ± 0.06 ms; n = 4 cells) following the GCL shock, suggesting that Blanes cells were monosynaptically excited by this stimulus.
Trains of GCL layer stimuli generated summated EPSPs (see lower panel in Figure 2-
9A2) that reliably triggered persistent firing in Blanes cells recorded in both cell-attached
(mean firing frequency = 6.3 ± 1.1 Hz; n = 7) and whole-cell modes (6.7 ± 2.1 Hz; n = 4
cells; Figure 2-9A3). Single shocks to the glomerular layer often triggered prolonged
47 EPSPs in Blanes cells that also could trigger persistent firing when the membrane
potential was near firing threshold (Figure 2-9A4; 6 of 6 cells tested). These data suggest
that glutamatergic synaptic inputs can effectively trigger persistent firing in Blanes cells.
The same extracellular stimulation protocols that reliably triggered persistent firing in
Blanes cells also triggered prolonged barrages of inhibitory postsynaptic responses in
granule cells. Figure 2-9B illustrates persistent IPSCs recorded in voltage-clamped
granule cells held at 0 mV to minimize the amplitude of spontaneous EPSCs. Granule
cell layer stimulation (ten shocks at 50 Hz; 100 μA) evoked periodic IPSCs in granule
cells (Figure 2-9B, insert; IPSC frequency = 6.2 Hz) and approximately the same
frequency that Blanes cells discharge following similar extracellular stimuli. Periodic
IPSCs evoked by GCL stimulation were abolished by both the GABAA receptor
antagonist gabazine (Figure 2-9C1; 10 μM) and by ionotropic glutamate receptor
antagonists (Figure 2-9C2; 5 μM NBQX + 50 μM d-APV). The variation in IPSC
amplitude and occasional missed oscillation cycles (Figures 9B and 9C) are consistent
with the quantal nature of these synaptic responses. Based on their similar periodic
nature, it is likely that these inhibitory barrages in granule cells resulted from persistent
activity in one or a small number of Blanes cells. Finally, we tested whether persistent synaptic activity had functional effects on granule cells. In 4 of 4 granule cells tested in which GCL stimulation evoked inhibitory barrages, the number of action potentials evoked by test current steps was significantly decreased when the step was preceded by a
GCL tetanus (to 54.1% ± 11.1% of control; the reduction in number of spikes was statistically significant; p < 0.02; Figure 2-9D). This reduction in excitability was
48 reversed by gabazine (4 of 4 cells), suggesting that this functional inhibition of granule cells was due to persistent inhibitory input to granule cells.
49 Discussion
In this study, we report for the first time the electrophysiological properties of Blanes
cells, large stellate-shaped interneurons in the granule cell layer of the mammalian
olfactory bulb. We used paired recordings to show that Blanes cells use GABA as a
neurotransmitter and inhibit granule cells. Unlike granule cells, which have pronounced
afterhyperpolarizations following repetitive spiking, Blanes cells generate depolarizing
afterpotentials that allow them to continue firing following a depolarizing stimulus. When
held near their action potential threshold, transient stimuli can trigger persistent spiking
in Blanes cells that can be reset by hyperpolarizing current steps. Persistent activity in
these feedforward GABAergic interneurons modulates tonic inhibition on granule cells and therefore is positioned to disynaptically regulate mitral and tufted cell activity.
Heterogeneous Interneuronal Subtypes in the Olfactory Bulb
A large variety of interneuronal subtypes has been demonstrated in the mammalian olfactory bulb using Golgi and immunocytochemical staining methods. On the basis of
morphology, at least seven non-granule cell types have been described in the granule cell
layer (Golgi cells, vertical cells of Cajal, stellate cells, short axon cells, horizontal cells,
bitufted neurons, and Blanes cells; Alonso et al., 2001, Kakuta et al., 1998, Lopez-
Mascaraque et al., 1986 and Schneider and Macrides, 1978). However this classification
system has not been adopted widely, and some authors have used both “short axon cells”
and “stellate cells” to represent non-granule cells located in the granule cell layer. Many
of the non-granule cells in the granule cell layer are immunoreactive for GABA or GAD,
50 two markers for GABAergic neurons (Gracia-Llanes et al., 2003 and Ohm et al., 1990),
although it is not clear whether all seven non-granule cell types identified thus far are
GABAergic. Similarly, many cell types located in the granule cell layer express one or
more calcium binding proteins (Alonso et al., 2001 and Kakuta et al., 1998), though
presently there is no clear correlation between neuronal subtype and the presence of a
particular protein marker. Also complicating the anatomical segregation of granule cell
layer interneurons is the observation that granule cells themselves express a variety of
GABAergic markers and calcium binding proteins (Alonso et al., 2001).
The major morphological features of the cell type we describe in this report (large multipolar neurons with stellate-like dendritic morphology) resemble three previously described granule cell layer neurons: Golgi cells, stellate cells, and Blanes cells. Of these three, the closest match for our cells appears to be Blanes cells (Blanes, 1898, Kakuta et al., 1998, Nakajima et al., 1996, Schneider and Macrides, 1978 and Scott et al., 1987), based on dendritic morphology and the presence of dendritic spines (see Figure 2-1 in
Schneider and Macrides, 1978). Stellate cells are similar in size to Blanes cells but are typically described in the deep part of the granule cell layer and may correspond to a subpopulation of VIP-immunoreactive interneurons (Nakajima et al., 1996). Golgi cells lack dendritic spines and also are typically reported in the deep granule cell layer (Kakuta et al., 1998; Price and Powell, 1970d; Schneider and Macrides, 1978). While our population of spiny non-granule interneurons appears relatively homogeneous in terms of intrinsic properties and persistent firing modes, we have also recorded from other large neurons in the granule cell layer that are distinct from Blanes cells. When intracellularly
51 stained, many of these interneurons correspond to either vertical cells of Cajal or horizontal cells (Alonso et al., 2001, Kakuta et al., 1998 and Schneider and Macrides,
1978). Our preliminary results suggest that most large non-granule cells in the GCL, including Blanes cells, receive fewer spontaneous synaptic inputs than granule cells.
However, we also observed apparent differences in the rates and kinetics of spontaneous inputs among these cells, suggesting that defining the complex pattern of synaptic innervation to these cells will require further study.
Our results provide multiple lines of evidence that suggest that Blanes cells monosynaptically innervate granule cells. Experiments using paired recordings demonstrate that single action potentials in Blanes cells are associated with short-latency
IPSPs in granule cells that are blocked by GABAA receptor antagonists. The short onset latency (mean 1.7 ms) along with the high reliability (>92% average success rate), low latency jitter, and insensitivity to glutamate receptor antagonists are all consistent with activation of a monosynaptic inhibitory pathway and are inconsistent with disynaptic inhibition. The ability to identify monosynaptically coupled Blanes/granule cell pairs using two-photon microscopy raises the possibility of visualizing the location of this synaptic connection on the granule cell dendrite in future studies. Additional ultrastructural experiments then will be necessary to define the morphological properties of the putative synaptic contacts identified using multiphoton microscopy (e.g., number of active zones, number of synapses per postsynaptic cell, number of postsynaptic neurons per Blanes cell).
52 Cellular Mechanism of Afterdepolarization and Persistent Firing
One of the most obvious electrophysiological differences between Blanes cells and granule cells is the large afterdepolarization evoked by suprathreshold activity in Blanes cells. This response does not appear to be mediated by polysynaptic activity, as the ADP is present after Na-based action potentials are blocked with TTX and is resistant to glutamate receptor antagonists.
The sensitivity of the ADP to the blockade of Ca influx (with either low Ca ACSF or Cd) and to intracellular perfusion with the Ca chelator EGTA suggests that it is triggered by a rise in intracellular Ca. Two potential explanations have been proposed for Ca-dependent depolarizing afterpotentials in other cell types: a transient reduction in standing outward
K current (Li and Hatton, 1997, Roper et al., 2003 and Roper et al., 2004) and Ca-
activated nonselective cation currents (Bourque, 1986, Ghamari-Langroudi and Bourque,
2002, Haj-Dahmane and Andrade, 1998 and Partridge et al., 1994). We found no
compelling evidence for transient reductions in standing K currents following
depolarizing steps in Blanes cells. The ADP response was unaffected by blockers of H-,
D-, and M-type currents; ADPs also could be evoked in slices bathed in high
concentrations of TEA, which would be expected to block delayed rectifier K currents.
The ADP was unaffected by raising the extracellular K concentration from 3 to 6 mM,
which should passively decrease ADP responses mediated by transient decreases in
standing K currents.
53 Several independent lines of evidence suggest that the ADP response in Blanes cells is
mediated by ICAN. The reversible blockade of the ADP by FFA is consistent with an ICAN mechanism (Ghamari-Langroudi and Bourque, 2002 and Partridge and Valenzuela,
2000), as is the reduction in ADP amplitude in low Na ACSF and the voltage dependence we observe to the ADP response (Haj-Dahmane and Andrade, 1998, Haj-Dahmane and
Andrade, 1999 and Bourque, 1986). However, FFA appears to have other effects in olfactory bulb slices (e.g., it reduced the number of Ca spikes evoked by depolarizing steps), and it only partially reduced the ADP even when applied at relatively high concentrations. While not definitive, our results suggest that at least one component of the ADP response in Blanes cells is mediated by ICAN. Also consistent with the ICAN hypothesis is our observation that uncaging Ca in Blanes cells activates an inward current that reverses polarity near 0 mV (Hall and Delaney, 2002 and Partridge et al., 1994).
Interestingly, the voltage dependence that we observed when we activated ICAN directly
through photolysis of NP-EGTA in Blanes cells was not as pronounced as the voltage
dependence of the ADP activated by depolarizing steps. The different degree of voltage
dependence we found in these two protocols may reflect the functional effects of other
currents besides ICAN that are activated by the depolarizing step and which oppose the
ADP at hyperpolarized membrane potentials.
Interestingly, olfactory granule cells in semi-intact frog forebrain also appear to have
ICAN-mediated afterdepolarizations (Hall and Delaney, 2002). The presence of large
AHPs in rat granule cells recorded in in vitro brain slices may reflect the absence of tonic
cholinergic input to the olfactory bulb; in many brain regions, AHPs evoked by bursts of
54 action potentials are blocked following muscarinic receptor activation (Egorov et al.,
2002 and Haj-Dahmane and Andrade, 1999). Our preliminary data suggest that this is
also true for granule cells in the rat olfactory bulb and that bath perfusion with carbachol
can reveal latent ADP responses in granule cells (T. Inoue, R.T.P., and B.W.S.,
unpublished data).
While the afterdepolarization is likely necessary to initiate persistent firing, the cellular
mechanism that maintains persistent firing in Blanes cells appears to rely on triggering a stable intrinsic up state. Several lines of evidence suggest that a Ca-dependent process is necessary to initiate the persistent firing up state in Blanes cells. We found that increasing intracellular Ca by photolysis of NP-EGTA triggered persistent firing, while manipulations that blocked Ca accumulations blocked persistent firing initiated by transient depolarization of Blanes cells. Surprisingly, interrupting persistent firing by injecting a 1–6 s duration hyperpolarizing current does not abolish firing when the step is removed. These two results suggest that persistent firing does not result from a simple regenerative system in which Ca influx from one action potential evoked an ADP response that triggered the next spike. Rather, the frequency of persistent firing in Blanes cells appears to be set through other biophysical or signaling systems that remain to be defined. The occasional membrane potential “notches” that are apparent when persistent firing begins to fail suggest that the kinetics of intrinsic spike afterpotentials may help determine the frequency of persistent firing. Periods of persistent firing do not result from a decrease in action potential threshold (data not shown). Rather, persistent firing appears to reflect the presence of a steady depolarizing current. This tonic depolarizing current
55 may be generated through a plateau potential in a distal dendritic compartment, slow Ca-
dependent modulation of leak conductances, or ICAN activation by Ca efflux from a
subcellular compartment.
Persistent firing in Blanes cells appears to be closely related to a similar phenomenon in
entorhinal cortical neurons (Egorov et al., 2002). Entorhinal pyramidal cells generate
large AHPs following action potential bursts. When these AHPs are blocked by
carbachol, the resulting ADP can trigger persistent firing. Blanes cells lack large AHP
responses and exhibit persistent firing under normal pharmacological conditions.
Initiation of persistent firing in entorhinal neurons is blocked by FFA (Egorov et al.,
2002), suggesting that the ICAN current responsible for the ADP may contribute to the persistent firing up state. Unlike Blanes cells, entorhinal neurons can fire persistently at different frequencies; strong depolarizing or hyperpolarizing current steps can increase or decrease firing frequency. The underlying intrinsic mechanism that allows Blanes cells
(and perhaps entorhinal cells) to fire persistently appears to be distinct from the synaptic mechanisms that mediate prolonged periods of activity in neocortical neurons
(McCormick et al., 2003 and Sanchez-Vives and McCormick, 2000). These neocortical up states are driven by barrages of excitatory synaptic input and periodically alternate with intervals in which the membrane potential is hyperpolarized. Persistent firing in entorhinal pyramidal cells (Egorov et al., 2002) and olfactory Blanes cells appears to operate at a single-cell level and is insensitive to antagonists of excitatory neurotransmitter receptors.
56 Functional Significance of Blanes Cells and Persistent Firing in the Olfactory Bulb
Our study demonstrates that Blanes cells represent a new class of bulbar interneurons that
inhibit granule cells by releasing GABA from axon terminals. There have been no
previous reports of local circuit synaptic responses mediated by axonal transmitter release
from interneurons located in the mitral or granule cell layers of the olfactory bulb. Blanes
cells may regulate the output cells of the olfactory bulb through at least three
mechanisms. First, Blanes cells mediate feedforward inhibition onto granule cells (Figure
2-3 and Figure 2-7E, and 9B–9D) and thus are in a position to control mitral and tufted
cell activity through disinihibition. We directly demonstrated that the same synaptic
stimuli that triggered persistent firing in Blanes cells evoked tonic GABAA receptor- mediated inhibition onto granule cells. These inhibitory barrages onto granule cells are unlikely to result from persistent activation of mitral cells, as we have previously shown that mitral cells do not receive prolonged excitation following trains of GCL stimuli
(Halabisky and Strowbridge, 2003). The close similarity between the frequency of persistent firing in Blanes cells and spontaneous IPSCs in granule cells strongly suggests that these inhibitory barrages arose from one or a small population of activated Blanes cells. We also showed directly that persistent firing in a Blanes cells evoked a persistent
train of inhibitory synaptic responses in monosynaptically coupled granule cells. Periodic
inhibitory barrages had powerful effects on granule cell excitability and reduced the
average number of action potentials triggered by a test stimulus by half. Our results
suggest that Blanes cells play an important role in regulating the excitability of inhibitory
granule cells and therefore are well positioned to control the strength of recurrent and
lateral inhibition onto mitral and tufted cells. Next, Blanes cells also can influence
57 activity in the olfactory bulb by regulating activation (or inactivation) of active currents in granule cells. Inhibitory postsynaptic potentials in granule cells may interact with voltage-dependent Ca currents expressed by these currents. Recently, Egger et al. (Egger et al., 2003) demonstrated that granule cells express low-threshold T-type Ca currents. In other cell types, GABAergic IPSPs can generate rebound discharges by deinactivating T- type Ca currents (McCormick and Bal, 1997). Additional studies will be required to determine whether Blanes cell-mediated IPSPs play a similar role in olfactory granule cells. Finally, because Blanes cells may potentially innervate hundreds of granule cells, spiking activity in Blanes cells may represent a novel mechanism to generate synchronous activity in subpopulations of olfactory bulb neurons. Synchronization may result from correlated rhythmic inhibition of large groups of granule cells activity, which then could lead to synchronization of mitral and tufted cells through periodic disinhibition or by triggering rebound spikes.
Persistent firing in Blanes cells is likely to have several consequences within the olfactory bulb. Persistent firing can be triggered by single shocks to the glomular layer, tetanic extracellular stimulation in the granule cell layer, or by direct depolarization. Slow phasic EPSP-like waveforms (Balu et al., 2004 and Halabisky and Strowbridge, 2003), similar to the membrane potential oscillations recorded in mitral and granule cells during normal respiration (Cang and Isaacson, 2003 and Margrie and Schaefer, 2003), also are an effective trigger for persistent firing (R.T.P. and B.W.S., unpublished data). Persistent activation of Blanes cells triggered by sensory stimulation will activate tonic feedforward inhibition onto granule cells and is likely to dampen granule cell responses on subsequent
58 respiratory cycles. Cang and Isaacson (2003) observed a sudden damping of granule cell responses following the initial inspiration cycle. It is not clear from these studies whether this effect, which is not observed in mitral cells, reflects persistent activation of Blanes cells or is due instead to intrinsic properties or signaling events in granule cells. Persistent activity also may represent a mechanism for recording short-term events, such as strong transient activation of a specific glomular-linked circuit, within the network dynamics of the olfactory bulb. Friedman and Strowbridge (2003) found that repetitive tetanic stimulation of olfactory bulb slices caused a long lasting increase in oscillatory activity.
These persistent oscillations, which were broadly tuned and centered in the γ band, were reflected in both field potential recordings and in an increase in tonic inhibitory input to mitral cells. Because this form of tetanic stimulation is an effective trigger for persistent firing in Blanes cells, it is intriguing to speculate that activity in these cells may contribute to these network oscillations.
Persistent firing in Blanes cells is remarkably robust. Even strong hyperpolarizing membrane steps (50 mV for 2–3 s) did not abolish persistent firing, suggesting that, once initiated, persistent firing is likely to be unaffected by subsequent inhibitory synaptic inputs. Prolonged hyperpolarization (60 s) did block persistent firing, but it is unclear whether barrages of hyperpolarizing inputs would ever be presented to Blanes cells over this long duration. Rather, it seems likely that under physiological conditions, persistent firing is limited by the duration of periodic sensory stimuli that summate to bring the membrane potential within the membrane potential window where persistent firing is possible. The prolonged EPSPs triggered in Blanes cells by single glomerular layer
59 shocks (Figure 2-9A) suggest that an inspiration-like pattern of excitation will exhibit a large degree of temporal summation, perhaps allowing Blanes cells to maintain their firing between inspiration cycles.
60 Experimental Procedures
Olfactory bulb slices (300 μm thick) from P14–P24 Sprague-Dawley rats were made using a modified Leica (Nussloch, Germany) VT1000S vibratome as previously described (Friedman and Strowbridge, 2003 and Isaacson and Strowbridge, 1998). An artificial cerebrospinal fluid (ACSF) solution with reduced Ca was used when preparing and storing slices. This solution contained 124 mM NaCl, 2.6 mM KCl, 1.23 mM
NaH2PO4, 3 mM MgSO4, 26 mM NaHCO3, 10 mM dextrose, and 1 mM CaCl2, equilibrated with 95% O2/5% CO2 and was chilled to 4°C during slicing. Olfactory bulb slices were incubated in a 30°C water bath for 30 min and then maintained at room temperature. During experiments, olfactory bulb slices were superfused with ACSF that contained 124 mM NaCl, 3 mM KCl, 1.23 mM NaH2PO4, 1.2 mM MgSO4, 26 mM
NaHCO3, 10 mM dextrose, and 2.5 mM CaCl2, equilibrated with 95% O2/5% CO2 and warmed to 30°C. All drugs except for TTX (Calbiochem) and o–nitrophenyl EGTA (NP-
EGTA, Molecular Probes) were obtained from Sigma (St. Louis, MO).
Whole-cell current-clamp recordings were made from olfactory neurons visualized with infrared-differential interference contrast (IR-DIC) optics using either Axioskop 1 FS
(Carl Zeiss) or BX51WI (Olympus) fixed-stage upright microscopes and either Axopatch
1D or Axoclamp 2B amplifiers (Axon instruments). Electrodes used for whole-cell recordings (5–7 MΩ) were pulled from thin-wall capillary tubes with filament (WPI) and contained 140 mM K-methylsulfate, 4 mM NaCl, 10 mM HEPES, 0.2 mM EGTA, 4 mM
MgATP, 0.3 mM Na3GTP, 10 mM phosphocreatine for current-clamp experiments, and
61 115 mM Cs-methanesulfonate, 25 mM TEA-methanesulfonate, 10 mM HEPES, 5 mM
QX-314, 4 mM NaCl, 4 mM MgATP, 1 mM EGTA, 0.3 mM Na3GTP, 10 mM
phosphocreatine for voltage-clamp recordings in granule cells. In several experiments,
the EGTA concentration was raised to 10 mM to increase Ca buffering capacity.
In the experiments shown in Figures 2-5C and 2-5D, we used a different internal solution
for Ca uncaging under voltage-clamp conditions (118 mM Cs-methanesulfonate, 25 mM
TEA-methanesulfonate, 10 mM HEPES, 4 mM NaCl, 4 mM MgATP, 0.3 mM Na3GTP,
10 mM phosphocreatine, 2 mM NP-EGTA) and an external solution that contained TTX
(1 μM). Control responses (Figure 2-5C2) were obtained using the same internal solution with no added NP-EGTA. The Ca uncaging results shown in Figure 2-7D were obtained using a current-clamp internal solution (140 mM K-methanesulfonate, 8 mM NaCl, 10 mM HEPES, 4 mM MgATP, 0.3 mM NaGTP, 10 mM phosphocreatine, 2 mM NP-
EGTA.) We used a 150 W Xe short-arc lamp (Opti-Quip) controlled by a fast electronic shutter (Uniblitz) to photolyse NP-EGTA (Ellis-Davies and Kaplin, 1994), which was loaded by one to three depolarizing steps to 0 mV (400 ms duration) immediately before the exposure. The output of the Xe lamp was directed toward the microscope objective
(Zeiss 63× water-immersion, NA = 0.9) using a long-pass dichroic mirror designed for epifluorescent detection of fura-2 (400DCLP, Omega Optical). In most experiments, we used arc lamp exposures of 100–2000 ms focused on the cell body region to activate Ca- dependent responses.
62 Neurons were visualized with either the ABC peroxidase method (Horikawa and
Armstrong, 1988) following intracellular Neurobiotin labeling (0.5%) and paraformaldehyde fixation or by using two-photon imaging in living slices following intracellular labeling with Alexa 594 (100 μM; Molecular Probes). We used the
Neurolucida (Microbrightfield) to generate the neuronal reconstructions shown in Figure
2-1C and the morphometric data in Figure 2-1D. Live imaging experiments utilized a
custom two-photon microscope based on the Verdi V10 pump laser, Mira 900 Ti-
sapphire laser (both from Coherent, Santa Clara, CA) and a high-speed XY galvanometer
mirror system (6210; Cambridge Technology). Intracellularly loaded fluorescent dyes
were excited at 830 nm through a 60× water-immersion objective (Olympus). Emitted
light was detected through a light path that included a 700DCLPXR dichroic mirror, a
BG39 emission filter (both from Chroma Technology), and a cooled PMT detector
module (H7422P-40; Hamamastu). Photomultiplier output was converted into an analog
voltage by a high-bandwidth current preamplifier (SR-570; Stanford Research Systems).
Custom Visual Basic software written by BWS controlled the scanning system and image analysis functions. Laser beam intensity was controlled electronically through a Pockels
cell attenuator (ConOptics) and a shutter (Uniblitz). In most experiments, the output of
the Mira laser was attenuated by 90%–95%.
We used two-photon microscopy to facilitate paired recordings by first recording from the presynaptic neuron with an electrode containing Alexa 594 (100 μM, Molecular
Probes). After waiting 20–25 min for the dye to diffuse through distal processes, we used
a fast scanning mode (3000 lines per second) to identify the axon (the nontapering,
63 spineless process that typical emerged from one of the primary dendrites of the Blanes
cell) and to follow this process until it started to bifurcate. In most cells, we could readily
identify the axonal plexus within 30–40 min after forming a whole-cell recording on the
Blanes cell. We then were able to record from between 1 and 5 potential target cells in this region before mechanical movements caused by repeated placements of the patch- clamp electrode deteriorated the Blanes cell recording. We determined whether postsynaptic responses evoked by Blanes cell action potentials were monosynaptic by measuring IPSP onset latency; monosynaptic responses had short, relatively invariant onset latencies that varied between 0.8 and 2.2 ms. All monosynaptic Blanes-to-granule cell monosynaptic IPSPs that we recorded also had high unitary probability (>90% assessed by single action potentials evoked in the Blanes cell every 2 s.)
Series resistance was typically <15 MΩ and was routinely compensated by >80% in voltage-clamp experiments. Inhibitory postsynaptic currents were recorded in granule cells at a holding potential of 0 mV in the experiment shown in Figure 2-9C to minimize the amplitude of spontaneous EPSCs. Voltage and current records were low-pass filtered at 2 kHz and sampled at 5 kHz using a ITC-18 16-bit analog-to-digital converter
(Instrutech) using custom software written in Visual Basic 6 (Microsoft). Origin 7.5
(OriginLab) and custom Visual Basic routines were used to analyze data. Synaptic responses were evoked using either tungsten monopolar electrodes (FHC) connected to a constant-current stimulus isolation unit (WPI) or by directly activating a presynaptic neuron. We determined the average amplitude of GCL-evoked EPSPs onto Blanes cells using minimal stimulation and averaging the response amplitude in nonfailure trials.
64 Pharmacological agents were applied by switching the perfusion solution. Voltages presented are not corrected for the liquid junction potential. Statistical significance was determined using the Student's t test. Data are presented as mean ± SEM.
65
Figure 2-1
66 Figure 2-1: Multiple Cell Types in Granule Cell Layer of the Olfactory Bulb
(A1) Schematic diagram of the olfactory bulb. Layers indicated are glomerular layer
(Glom), external plexiform layer (EPL), mitral cell layer (MCL), and granule cell layer
(GCL). (A2) Infrared DIC image of the granule cell layer in a rat olfactory bulb slice. The
asterisk indicates the cell body of a large non-granule Blanes neuron in the granule cell.
Scale bar, 20 μm.
(B) Responses to hyperpolarizing and depolarizing current injections in granule and
Blanes cells.
(C) Reconstructions of neurobiotin-labeled granule and Blanes neurons. Presumptive
axons on Blanes cells are indicated by asterisks.
(D1) Comparison of soma size, number of primary dendrites, and percentage of cells with
processes that enter the external plexiform layer between granule cells and Blanes cells.
We never observed a Blanes cell process enter the EPL (no bar shown in graph), whereas
86% of intracellularly stained granule cells had processes in the EPL. p < 0.002, p <
0.001. (D2) Sholl analysis of dendritic branching pattern in Blanes cells (open circles) and
granule cells (closed circles). p < 0.05.
67
Figure 2-2
68 Figure 2-2. Visualization of Blanes Cell Morphology with Two-Photon Microscopy
(A) Two-photon image of a dendritic morphology of Blanes neuron filled with 100 μM
Alexa 594 through the patch pipette. Scale bar, 20 μm.
(B and C) Enlargements of dendritic segments with dendritic spines from locations
indicated on panel (A).
(D) Enlargement of presumptive axon. Scale bar on panels (B)–(D) is 5 μm.
(E1) 2P image of soma/proximal dendrite region of a different Blanes cell filled with
Alexa 594. Diamond symbol indicates presumptive main axon and axon collateral. (E2)
2P image of axon plexus in the GCL formed by the axonal processes indicated by the diamond symbol in (E1). The center of the axon plexus was 250 μm away from the cell
body shown in (E1). Scale bar, 20 μm.
69
Figure 2-3
70 Figure 2-3. Blanes Cells Are GABAergic and Innervate Granule Cells
(A) Granule cell responses to 20 successive Blanes cell action potentials. Blanes cells
reliably evoked hyperpolarizing IPSPs in a granule cells held at −55 mV. Average
response in granule shown in middle trace.
(B1) Blanes cell-evoked IPSPs reversed polarity near the chloride equilibrium potential.
Granule cell postsynaptic responses recorded at −51 mV, −70 mV, and −90 mV; IPSPs evoked by two action potentials separated by 50 ms. (B2) Plot of mean IPSP amplitude
versus membrane potential in three Blanes-granule cell pairs (mean reversal potential =
−78.3 mV).
(C1) Amplitude distribution of Blanes cell-evoked IPSPs (mean amplitude = −1.76 ± 0.1 mV; n = 101 presynaptic action potentials, including 6 failures). (C2) Onset latency
distribution for the same Blanes/granule paired recording (mean latency = 1.54 ± 0.03
ms; n = 88).
(D) Monosynaptic Blanes cell-evoked IPSPs are reversibly blocked by the GABAA
receptor antagonist gabazine (10 μM). Top traces show averages of five consecutive
responses at the times indicated by the open arrows in the bottom time plot. Blanes cell-
evoked IPSPs in this paired recording were not affected by NBQX (5 μM) and d-APV
(50 μM; data not shown).
71 Responses shown in (A) and (B)–(D) were recorded from different Blanes/granule cell pairs.
72
Figure 2-4
73 Figure 2-4. Afterdepolarizations in Blanes Cells
(A) Afterdepolarizations evoked by graded duration suprathreshold current steps;
responses enlarged and aligned to step offset in (A2).
(B1) Responses to depolarizing steps at different membrane potentials. Adapting firing triggered by ADPs in Blanes cells (left panel). Afterdepolarization is voltage dependent and is not present at −80 mV. Step amplitude increased at −80 mV to trigger the same
number of action potentials as at −65 mV. The inset shows superposition and
enlargement of responses at −65 and −80 mV. (B2) Summary plot of afterpotential
amplitude at −80 mV and at −62 mV from six experiments. p < 0.001.
(C) Decreasing Ca influx by switching to a low Ca (0.5 mM), high Mg (6 mM) external
solution reversibly abolishes the ADP response. Current step = 50 pA.
(D) The ADP also is blocked by 200 μM CdCl2. Current step = 200 pA.
74
Figure 2-5
75 Figure 2-5. Calcium-Dependent Afterdepolarizations in Blanes Cells
(A) Afterdepolarization is abolished by the intracellular Ca chelator EGTA. Responses
shown 2 min following breakthrough to whole-cell recording mode (six APs) and after 38
min (seven APs). Pipette tip filled with low EGTA (0.2 mM) intracellular solution; high
EGTA (10 mM) solution used to backfill pipettes. Lower graph shows reduction in ADP
amplitude during recording. The inset shows summary graph of percent ADP reduction
with high EGTA backfill experiments (n = 4) and with interleaved control experiments
(backfilled with 0.2 mM EGTA; n = 3). Current step = 200 pA. p < 0.01.
(B) Photolyzing NP-EGTA (2 mM) evokes an inward current in Blanes cells voltage
clamped at −60 mV. The UV illumination timing is indicated by the horizontal bar above
the trace. Blanes cells were transiently depolarized to 0 mV 2–3 s before each period of
UV illumination to maximize the concentration of Ca bound to NP-EGTA. (Uncaging
responses shown in subsequent traces begin after this step depolarization.)
(C1) Responses to a series of graded duration UV exposures (100, 250, 400, 1000 ms long). (C2) A 1 s duration UV exposure fails to elicit a response in a Blanes cell filled
with a control internal solution without any added NP-EGTA.
(D1) Responses to UV exposure in Blanes cells loaded with NP-EGTA reverse polarity
near 0 mV. (D2) Summary plot of the peak amplitude of the responses to UV exposure at
different holding potentials in nine voltage-clamped Blanes cells.
76
Figure 2-6
77 Figure 2-6. Afterdepolarizations Are Blocked by ICAN Antagonists
(A) Increasing extracellular [K] concentration from 3 to 6 mM does not affect
afterdepolarization. Superimposed responses recorded in 3 and 6 mM [K] shown in right
panel. Elevated [K] did reduce the amplitude of the fast afterhyperpolarization
immediately following the step depolarization (arrow in middle trace). Responses to the
step depolarizations are shown in inserts above each trace.
(B) Afterdepolarizations were evoked by a train of Ca spiked after Na and K channels
were blocked using bath application of TTX (1 μM), TEA (25 mM), and 4-AP (100 μM).
The amplitude of the ADP evoked by Ca spikes was reduced in low Na ACSF. Responses before and after switching to low Na ACSF are superimposed on the right panel.
(C) Plot of the reduction in ADP amplitude by the ICAN blocker FFA (200 μM). In most cells, FFA decreased the number of Ca spikes triggered by the depolarizing steps. We increased the amplitude of the depolarizing step in several episodes (open circles) so as to trigger the same number of Ca spikes as in control. Traces above plot show control ADP,
response in FFA to a larger amplitude step [+ FFA (inc stim)], and following washout of
FFA.
(D) Summary of experiments with elevated [K], NMG substitution, and FFA. p < 0.002.
78
Figure 2-7
79 Figure 2-7. Brief Depolarizations Trigger Persistent Firing in Blanes Cells
(A) Responses to constant step depolarizations at different membrane potentials
(indicated above trace). Persistent firing evoked at the most depolarized membrane
potential (−58.5 mV).
(B1) Transient and persistent firing modes evoked at the same membrane potential by
graded depolarizing steps (60, 80, 100 pA). (B2) Similar depolarizing current steps (100 pA) evoked an afterhyperpolarization in granule cells.
(C1) Frequency of persistent firing was not affected by modulating the amplitude of the
initiating depolarizing step. (C2) Distribution of the frequencies of persistent firing in 61
Blanes cells activated by step depolarization. The mean firing frequency in this
population of Blanes cells was 6.5 Hz.
(D) Persistent firing evoked by uncaging Ca. Light exposure (UV, duration indicated by
bar above trace) in a Blanes cells recorded with NP-EGTA (2 mM) in the internal
solution depolarized the neuron and initiated tonic firing. No response was observed to
light exposure in a control experiment (below) in which NP-EGTA was not added to the
internal solution. Persistent firing was induced by a depolarizing step in this cell near the
end of the record.
80 (E1) Paired recording between a Blanes cell (bottom trace) and a monosynaptically
coupled granule cell held at −50 mV (top trace). Single Blanes cell action potentials
consistently evoked a short-latency (1.54 ± 0.03 ms; same paired recording as in Figure
2-3C) IPSP in this granule cell. A 500 ms long depolarizing step to the Blanes cell
evoked a barrage of IPSPs in the granule and triggered persistent firing in the Blanes cell.
The resulting persistent firing at 5.9 Hz in the Blanes cell triggered persistent inhibition
in the granule cell. (E2) Gabazine (10 μM) blocked spike-evoked IPSPs in this Blanes- granule cell paired recording.
81
Figure 2-8
82 Figure 2-8. Prolonged Hyperpolarization Stops Persistent Firing
(A) Persistent firing is reversibly blocked by switching to a low Ca (0.5 mM), high Mg (6 mM) external solution. Current step = 50 pA.
(B1) Plot of time course of persistent firing evoked by a 500 ms depolarizing step.
Example traces from time points indicated on plot. (B2) Enlargements of steady-state
period of persistent firing (bottom trace) and the firing pattern immediately before
persistent firing ceased (top trace). Cessation of persistent firing was associated with
increased interspike intervals that revealed spike-evoked ADPs at the time of the
expected spike.
(C) Persistent firing evoked by a 1 s depolarizing step is unaffected by transient 3 s
hyperpolarization. Average firing rate indicated above trace segments indicated by solid
lines.
(D1) Persistent firing in another Blanes cell is abolished by prolonged hyperpolarization
(60 s). (D2) Summary of effect of prolonged (45–60 s) hyperpolarization on mean duration of persistent firing on ten Blanes cells. Prolonged hyperpolarization significantly
reduced the average firing duration compared with interleaved control experiments (n = 9
cells) without tonic hyperpolarizing current applied. p < 0.005.
83
Figure 2-9
84 Figure 2-9. Synaptic Stimulation Activates Blanes Cells and Evokes Long-Lasting
Inhibition onto Granule Cells
(A1) Schematic drawing of stimulation experiment with stimulus electrodes placed in
either the glomerulus layer or the granule cell layer. (A2) Excitatory synaptic response to
a single shock in the GCL was blocked by NBQX (5 μM) and d-APV (25 μM). Lower
panel shows response to a train of GCL stimuli at 20 Hz. The second and subsequent
GCL-evoked EPSPs triggered action potentials. (A3) Persistent firing evoked by tetanic
stimulation in the granule cell layer (ten shocks at 50 Hz). Responses recorded in both cell-attached (top) and whole-cell configuration (bottom) from the same Blanes cell. (A4)
Single shocks in the glomerular layer triggered prolonged EPSPs in Blanes cells (top) that could evoke persistent firing at increased stimulus intensity (bottom).
(B) Tetanic stimulation in the GCL evokes periodic large-amplitude IPSCs recorded in voltage-clamped granule cells held near the EPSC reversal potential (0 mV). Both periodic IPSCs and spontaneous EPSCs were recorded at hyperpolarized membrane potentials (−30 to −90 mV). Episodes acquired at 2 min intervals. The inset shows inter-
IPSC interval histogram compiled from 281 events detected in the same Blanes cell.
GCL-evoked inhibitory barrages were periodic with a preferred interval of 173 ms (5.8
Hz) with a smaller second peak at the second harmonic (358 ms).
(C1) Responses to four consecutive GCL tetani in a granule cells held at 0 mV. Periodic
IPSCs evoked by this stimulus were blocked by the GABAA receptor antagonist
85 gabazine (10 μM; bottom trace). (C2) Persistent inhibition in granule cells evoked by a
GCL tetanus also was blocked by NBQX (5 μM) and d-APV (25 μM).
(D) GCL tetani inhibited granule cell responses to test depolarizing stimuli (40 pA
depolarizing step, 1 s duration). This form of functional inhibition onto granule cells was blocked by gabazine (10 μM), suggesting that is was mediated by persistent firing in
inhibitory neurons. The rastergram above the traces represents the timing of step-evoked action potentials in seven consecutive trials. No spikes were triggered by the depolarizing step on 4/7 trials following the GCL tetanus. The GCL stimulating electrode was placed at least 200 μm away from the cell body in all experiments.
86
Chapter 3
Muscarinic receptor modulation of afterpotentials and firing modes in olfactory
bulb granule cells
87 Introduction
The olfactory bulb (OB) is the second order olfactory brain area, and extensive
processing of odorant information occurs in this region before transmitting to olfactory cortex and other brain regions. Odorant receptors expressed on the cilium of olfactory
receptor neurons in the nasal epithelium mediates olfactory information transduction
(Buck and Axel, 1991).The odorant receptor neurons project to the OB in a spatially
conserved manner (Mori et al. 1999). Odorant receptor neurons send their axons to
specialized structures in the OB called “glomeruli”, which are essentially glial sheaths
that contain the synapses that these axons make with the primary dendrites of mitral cells,
the principal output cells of the OB. The pattern of this projection maintains the spatial
location of the odorant receptor neurons in the nasal epithelium. Mitral cells, in turn, project to olfactory cortex and other higher brain areas (Shepherd and Greer, 1998).
Mitral cells also have long secondary dendrites, which make multiple dendrodendritic synapses with OB granule cells. The main inhibitory input that modulates mitral cell firing originates from granule cells-- small axonless GABAergic interneurons, considered
the most populous cell in the OB (Price and Powell, 1970a). It is through the
dendrodendritic synapse that granule cells are able to exert feedback inhibition onto mitral cells (Jahr and Nicoll, 1980), as well as mediating lateral inhibition between mitral cells (Isaacson and Strowbridge, 1998). Previous work on the OB described the synaptology and intrinsic properties of mitral, granule cells, with special regard to the dendrodendritic synapse and its role in information processing (Lagier et al. 2004).
Additionally, recent functional studies have defined roles for a few other OB
88 interneurons, namely periglomerular neurons (Aungst JL 2003) and Blanes cells (Pressler and Strowbridge, 2006). Unfortunately, little else is known about the local circuits within
the OB, including the functional significance of centrifugal inputs to OB from other brain
regions.
Afferents to the OB that govern the degree of behavioral and context dependence largely synapse onto granule cells and other granule cell layer interneurons, and are likely to play key roles in the modulation of granule cell activity, and therefore, granule cell mediated mitral cell feedback and lateral inhibition. The horizontal limb of the diagonal band, for example, provides cholinergic and GABAergic input to OB granule cells in the form of descending fibers that terminate specifically on granule cell proximal dendrites (Kasa et al., 1995), and on granule cell gemmules (Kasa et al., 1995)—the location of the dendrodendritic microcircuit. The horizontal limb of the diagonal band is a brain region found in the cholinergic basal forebrain—the area of the brain responsible for providing attentional resources to sensory and motor behaviors. Selective ablation of the cholinergic basal forebrain in living specimens generates a number of deficits, including a reduction in olfactory discrimination and differentiation (Linster et al., 2001) and even the prevention of the formation of olfactory learning. Despite the anatomical evidence (Kasa et al., 1995; Macrides et al., 1981) describing this input, there are no studies that illustrate, using intracellular recordings, the effects these cholinergic inputs would have on granule cell intrinsic properties or excitability.
89 A number of investigators have studied the effect the horizontal limb of the diagonal
band and/or cholinergic input has in the OB (Inokuchi et al., 1981; Kunze et al., 1992),
with some mixed results. Inokuchi et al. indicated that stimulation in the horizontal limb of the diagonal band inhibited most mitral cells while exciting most granule cells
(Inokuchi et al., 1981), whereas Kunze et al. reported the opposite result (Kunze et al.,
1992)—possibly confounding these investigations are the mixed transmitter phenotypes present in the horizontal limb of the diagonal band. We decided to study cholinergic activation in granule cells as a means to resolve this situation. Investigations involving whole cell recordings from granule cells during application of muscarinic acetylcholine agonists in vitro are sparse, and one study by Castillo and colleagues (Castillo et al.,
1999) described a reduction in granule cell firing rate, but an increase in granule cell spontaneous transmitter release following the application of large concentrations of carbachol (50 μM) and other m2AChR agonists. This study lacked intracellular recordings from granule cells and a mechanistic explanation for the decrease in extracellularly monitored granule cell firing rate. Additionally, a recent publication from
Ghatpande and colleagues (Ghatpande et al., 2006) indicated that the increase in spontaneous transmitter release following the application of carbachol (50 μM) was due to the activation of M1AChRs and not M2 AChRs. Given that granule cells express a large amount of M1AChRs (Cortes and Palacios, 1986; Le Jeune et al., 1995; Le Jeune et al., 1996), and that M1AChRs activation can result in excitatory actions in other neurons
(Dutar and Nicoll, 1988), it is likely that m1AChR agonists would have a different effect on granule cell excitability than that indicated in the Castillo et al. investigation.
90 In this study, we sought to determine the intrinsic electrophysiological effect of cholinergic input to granule cells—with special regard to cholinergic modification of granule cell excitability. Next, we sought to determine the receptor subtype responsible for the cholinergic modification of granule cell properties. Finally, we defined the functional significance for granule cell output, with special regard to changes in granule cell output that would influence OB circuit dynamics while granule cells receive cholinergic modulation.
91 Materials and Methods
Horizontal slices (300 μm thick) from the olfactory bulbs of 14- to 21- day old Sprague-
Dawley rats were placed in a submerged recording chamber on the stage of an upright
microscope (Zeiss Axioskop FS) and continuously perfused at 30 °C with standard
artificial cerebrospinal fluid (in mM: 124 NaCl, 3 KCl, 1.23 NaH2PO4, 1.2 MgSO4, 26
NaHCO3, 10 dextrose, 2.5 CaCl2). Individual granule and mitral cells were visualized
using IR/DIC optics. Whole-cell current clamp recordings were made using an Axopatch
1D amplifier using a K-methylsulfate based internal solution (in mM: 140 K
methylsulfate, 4 NaCl, 10 HEPES, 0.2 EGTA, 4 MgATP, 0.3 Na3GTP, 10
phosphocreatine) for current clamp recordings and a Cs-methylsulfate based internal
solution (in mM: 115 Cs-methanesulfonate, 25 TEA-methanesulfonate, 10 HEPES, 5
QX-314, 4 NaCl, 4 MgATP, 1 EGTA, 0.3 Na3GTP, 10 phosphocreatine) for voltage
clamp recordings.
Large stellate-shaped interneurons in the GCL could be readily differentiated from the
more numerous granule cells using IR-DIC optics in live slices. In most cases, current-
clamp recordings from neurons with large somata (>15 μm) in the GCL showed common
characteristics of Blanes cells (ADP, persistent firing; see Results below).
In some experiments we used a different internal solution for Ca uncaging under voltage
clamp conditions (in mM: 118 Cs-methanesulfonate, 25 TEA-methanesulfonate, 10
HEPES, 4 NaCl, 4 MgATP, 1 EGTA, 0.3 Na3GTP, 10 phosphocreatine, 2 NP-EGTA)
92 and an extracellular solution that contained TTX (1 μM). Control responses were
obtained using the same internal solution with no added NP-EGTA. We used a 150 W Xe
short-arc lamp (Opti-Quip) controlled by a fast electronic shutter (Uniblitz) to photolyse
NP-EGTA (Ellis-Davies and Kaplin, 1994), which was loaded by one depolarizing step to 0 mV (250 ms duration) immediately before the exposure. The output of the Xe lamp was directed toward the microscope objective (Zeiss 63x water-immersion, NA = 0.9) using a long-pass dichroic mirror designed for epifluorescent detection of fura-2
(400DCLP, Omega Optical). In most experiments, we used arc lamp exposures of 100-
2000 ms focused on the cell body region to activate Ca-dependent responses.
Series resistance was typically <15 MΩ and was routinely compensated by >80% in voltage-clamp experiments. Inhibitory postsynaptic currents were recorded in mitral cells at a holding potential of -70 mV in the experiment shown in Figure 2-5. Voltage and current records were low-pass filtered at 2 kHz and sampled at 5 kHz using an ITC-18
16-bit analog-to-digital converter (Instrutech) using custom software written in Visual
Basic 6 (Microsoft). Origin 7.5 (OriginLab) and custom Visual Basic routines were used to analyze data. Synaptic responses were evoked using either tungsten monopolar electrodes (FHC) connected to a constant-current stimulus isolation unit (WPI).
Pharmacological agents were applied by switching the perfusion solution. Voltages presented are not corrected for the liquid junction potential. Statistical significance was determined using the Student’s t-test. Data are presented as mean ± SEM.
93 Results
Carbachol reveals activity evoked afterdepolarizations in granule cells
A short burst of action potentials robustly evokes large amplitude (-3.7 ± 0.9 mV, n= 10)
afterhyperpolarizations (AHPs) in depolarized (to -60 mV) granule cells (Figure 3-1 A).
However, carbachol (1-2 μM, a nonselective muscarinic acetylcholine receptor agonist),
when added to the external solution, converted a large amplitude (3.0 ± 0.8 mV, n= 10),
stimulus evoked ADP (Figure 3-1A). Next, we recorded from granule cells with an
intracellular solution containing 10 mM BAPTA to determine if the large amplitude
stimulus evoked ADP in the presence of carbachol is the result of events downstream of
Ca signaling, or, if the inward current due to Ca influx generates the ADP. Granule cells
using the BAPTA intracellular solution did not express an ADP with carbachol
application (n= 3) and the afterpotential is significantly lower when compared to control
in carbachol (p < 0.01). This suggests that a Ca specific signaling pathway generates the
ADP and current due to Ca entry is not responsible for the stimulus evoked ADP.
Afterdepolarizations are used in the nervous system to enhance synaptic input by
extending or amplify depolarizing currents (Pressler and Strowbridge, 2006; Haj-
Dahmane and Andrade, 1998; Haj-Dahmane and Andrade, 1999; and Hall and Delaney,
2002). An afterdepolarizing nonselective ICAN cation current is responsible for a number
of these ADPs in other brain areas (Egorov et al., 2002; Partridge et al., 1994).
Additionally, other investigations have shown that large amplitude ICAN currents exist in amphibian olfactory bulb granule cells (Hall and Delaney, 2002). Next, we used
94 flufenamic acid (100 μM), a partially selective ICAN blocker (Ghamari-Langroudi and
Bourque, 2002; Partridge and Valenzuela, 2000) to see if the stimulus evoked ADP is composed of ICAN current. Flufenamic acid reduces the amplitude of the ADP in granule
cells by 34.6 ± 9.1% (n = 3). It is likely that mammalian olfactory bulb granule cells also
express an ICAN current, much like the recently described Blanes cells (Pressler and
Strowbridge, 2006). However, unlike Blanes cells, the ADP in granule cells is not apparent until after carbachol application. Like the ICAN current expressed in other
neurons including amphibian granule cells (Hall and Delaney, 2002) and olfactory bulb
Blanes cells (Pressler and Strowbridge, 2006), the stimulus evoked ADP in granule cells
is outward rectifying (Figure 3-1C1), and is significantly larger at -60 mV compared to -
80 mV (0.4 ± 0.4 mV, n= 10, p < 0.01). By contrast, similar current steps in Blanes cells,
another olfactory bulb interneuron located in the granule cell layer, produces large
amplitude afterdepolarizations (ADPs) even in the absence of muscarinic agonists (Figure
3-1 D). Figure 3-1C2 is a summary plot illustrating that carbachol application does not
significantly affect the amplitude of these large amplitude Blanes cell ADPs (7.4 ± 0.9
mV in control; compared to 6.6 ± 1.0 in 2 μM CCh, n=7).
Our results suggest that the carbachol activated stimulus evoked ADP in granule cells is a voltage sensitive current that is activated by intracellular Ca, and is probably an ICAN
current given its attenuation by flufenamic acid. To test if an ICAN current is responsible
for the ADP we examined the effects of Cd application on the ADP. In all five of the
granule cells tested, Cd application abolished the carbachol induced ADP. Figure 3-2A
shows a representative granule cells initial AHP, its ADP following carbachol application
95 (2 μM), and the subsequent elimination of the ADP following additional Cd application
(200 μM). The average control stimulus evoked afterpotential amplitude is -5.1 ± 0.3 mV
(n=44), in carbachol (2 μM) the afterpotential became a significantly larger ADP with an
amplitude of 4.1 ± 0.7 mV (n=19, p<0.001), and Cd significantly decreased the size of
the ADP to -0.4 ± 0.7 mV (n=5, p< 0.01). These results suggest that the ADP in granule
cells is thus dependent on calcium entry.
Next, we used the photolyzable intracellular Ca chelator NP-EGTA to see if intracellular
Ca release is sufficient to generate an ADP in voltage clamped granule cells. In 10 out of
14 granule cells (small cells with a soma diameter of roughly 10 μm in the GCL) tested,
photolyzing NP-EGTA with brief UV exposures (100-2000 ms) evoked an inward
current when the presumptive granule cell was held at -40 mV. In one of these cells, brief
UV exposure elicited an outward current at this membrane potential, and in three of the
cells UV exposure failed to elicit any current. As illustrated in Figure 3-2C1 increasingly
longer shutter durations evoked increasingly larger inwards currents at -40 mV.
Additionally, a 1000 ms duration UV exposure onto granule cells that did not contain
NP-EGTA failed to generate a response (Figure 3-2C2, n=4). Figure 3-2D shows example traces of UV uncaged Ca current at several different membrane potentials. The current generated by the Ca uncaging reversed near to 0 mV. Figure 3-2E shows a summary plot of the currents evoked at these membrane potentials (n=6). The UV-activated current is also associated with a decrease in input resistance (57.7 ± 5.9 %, n=5 cells, p < 0.01), measured by inserting small test pulses at the maximal point of UV-activated current (10 mV for 100 ms, n=4 cells). In addition to outwardly rectifying, being reduced by
96 flufenamic acid, blocked by cadmium, and blocked by intracellular BAPTA, transient increases in intracellular calcium also specifically activate this depolarizing current, and this depolarizing current corresponds to a conductance increase. Together these data suggest that ICAN is responsible for stimulus evoked ADP in granule cells exposed to carbachol.
The pharmacology of muscarinic acetylcholine receptor activation
Next, we determined which muscarinic acetylcholine receptor (mAChR) is responsible for the AHP reduction and ADP enhancement in granule cells. Carbachol is a nonselective muscarinic acetylcholine receptor agonist, and as Figure 3-3A illustrates, the
ADP that exists following carbachol application can be blocked by further application of atropine (5 μM, n=1), a nonselective mAChR antagonist. Next, we tested M2 AChR involvement by attempting to reduce the size of the CCh activated ADP by applying
AFDX-116 (1 μM), an M2 AChR antagonist (Figure 3-3B). AFDX-116 did not significantly affect the size of the ADP in five cells tested (4.2 ± 1.2 mV in 2 μM CCh and 1 μM AFDX-116, vs. 4.1 ± 0.7 mV in 2 μ CCh). Next, to test M1AChR involvement in the CCh revealed ADP, we applied pirenzepine (10 μM), a selective M1AChR antagonist (Figure 3-3C). Pirenzepine significantly reduced the size of the ADP (bringing it to 0.3 ± 0.6 mV, n=5, p < 0.01). To show that M1AChR activation is sufficient to convert the stimulus evoked AHP to an ADP, we then applied MCN-A-343 (100 μM) to the bath solution (Figure 3-2D). MCN-A-343 significantly changed the afterpotential from an AHP to the ADP (from -5.1 ± 0.3 mV to 3.8 ± 1.7 mV, n=5, p < 0.0001) to a
97 similar extent as CCh application. These data show that in addition to ICAN blockers, Ca channel blockers, and intracellular calcium chelators, the CCh revealed stimulus evoked
ADP in granule cells can be reduced by further application of M1AChR, but not M2
AChR selective antagonists. And as confirmed by MCN-A-343 application, the stimulus evoked ADP in CCh is the result of M1 AChR activation.
Carbachol enhances granule cell excitability
The effect of carbachol on granule cell properties have been investigated recently using a concentration of carbachol at relatively high (50 μM) concentrations. Little is known about the effects of lower concentrations on olfactory bulb granule cell activity. We therefore compared 2 and 50 μM concentrations of carbachol on granule cell afterpotentials and intrinsic properties. Examples illustrated in Figure 3-4A show representative traces of the stimulus evoked AHP in the control state, the CCh (2 μM) revealed ADP, and then the enhanced ADP and reduced firing after increased CCh (50
μM) application. A summary plot of the average AHP (-5.1 ± 0.3 mV, n = 44), ADP in 2
μM CCh (4.1 ± 0.7 mV, n = 19) and ADP in 50 μM CCh (8.1 ± 1.3, n = 8) is illustrated in
Figure 3-4B. Compared to the 2 μM CCh revealed ADP, the ADP in 50 μM is significantly larger (p < 0.01), and is significantly larger than the control AHP (p <
0.001). One interesting effect of 50 μM CCh application is the number of action potentials generated by the brief current step decrement considerably. It is likely that the reduction of the number of action potentials is not because the increased concentration of
CCh is lowering the intrinsic excitability of the granule cell, but in fact, perhaps the
98 higher concentration of CCh is vastly enhancing granule cell excitability, and as a result
enters a period of depolarization action potential block. Next, we set out to measure the
intrinsic excitability of granule cells and it also increases dramatically in 50 μM CCh when compared to 2 μM CCh and control. Figure 3-4C shows the number of action potentials generated by a brief 50 pA current step in control, 2 μM CCh, and 50 μM CCh.
The number of action potentials generated by these small steps increases proportionally with the concentration of CCh applied to the extracellular solution. A summary plot of the mean values in control (1.7 ± 0.5, n=12), 2 μM CCh (3.1 ± 0.6 spikes, n=7), and 50
μM CCh (4.5 ± 0.4 spikes, n=7) is shown in Figure 3-4D. By using smaller current steps,
we observed the increase in action potential generation before the cell entered
depolarizing block. Lastly, another effect of the increased concentration of CCh on granule cell intrinsic properties is that during larger current depolarizations (square steps of 70 pA or greater), the evoked action potentials suffered a marked decrease in amplitude. Figure 3-4E shows an enlargement of the representative traces originally illustrated in Figure 3-4A. Note the reduced number of action potentials in the 50 μM
CCh example, as well as the reduced action potential amplitude. Figure 3-4F is a summary plot of mean action potential amplitudes in control (n = 11), 2 μM CCh (n = 6)
and 50 μM CCh (n=8). The mean amplitudes of the first, second, and third action
potentials are indicated by the white, hatched, and gray bars respectively. The mean
amplitudes for the first, second, and third control action potentials are 59.5 ± 2.4 mV,
54.3 ± 1.8 mV, and 52.8 ± 2.2 mV respectively. The mean amplitudes for the first,
second and third 2 μM CCh action potentials are 57.0 ± 1.8 mV, 50.6 ± 1.6 mV, and 47.8
± 1.8 mV respectively. In addition, lastly, the mean amplitudes for the first, second, and
99 third 50 μM CCh action potentials are 57.4 ± 2.7 mV, 34.9 ± 7.2 mV, and 30.2 ± 6.7
respectively. The second and third mean action potential amplitudes in the 50 μM CCh
condition are significantly lower than those in the control condition (p < 0.01). The
reduction in action potential amplitude while under 50 μM CCh application may occur
because the ADP depolarizes the neuron to such a degree that the neuron is not
sufficiently hyperpolarized by the single spike AHP, and thus the sodium channels are
unable to recover from inactivation.
Our results thus far suggest that M1 AChR receptor activation reveals a current, likely to
be an ICAN current, that is selectively active at more depolarized membrane potentials.
Stimulus evoked ADPs effectively extend the duration of depolarizing input, and can also
enhance the amplitude of depolarizing input as well. Additionally, calcium-activated
nonselective cation currents increase cellular excitability in other neurons (Pressler and
Strowbridge, 2006; Egorov et al., 2002). In this section, we investigated how the CCh
induced ADP in granule cells affects granule cell output onto mitral cells, and to what
extent the CCh induced ADP contributes to granule cell excitability. To address these
questions, we first recorded from a voltage clamped mitral cell, and stimulated in an on- beam glomerulus with a tungsten stimulating electrode. An on-beam glomerulus is a glomerulus closest to the mitral cell, and is typically the target of the mitral cells apical dendrite. In Figure 3-5A, we stimulated four times at a frequency similar to inhalation frequency (100 μA for 200 μs, 2.5 Hz). The glomerulus stimulation evoked a significantly larger inward current. The current integral of a three second time window following the last stimulation reflects this change (Figure 3-5C). In four cells, the mean
100 integrated current increased from 198.4 ± 76.8 nA * s in the control condition to 478.4 ±
75.1 nA * s (p < 0.05).
Next, we examined the intrinsic granule cell contribution to this excitability change. We injected current waveforms that are similar to the depolarizations that granule cells receive from mitral cells (Cang and Isaacson, 2003) at the same frequency as in Figure 3-
5A in order to examine any changes in granule cell excitability due to CCh application.
Figure 3-5B shows two representative example traces of granule cell action potential firing due to these simulated EPSPs in control, and in CCh (2 μM). As shown in Figure
3-5D, the action potentials generated by the simulated EPSPs significantly increase in
CCh (14.3 ± 1.3 in control vs. 29.4 ± 6.6 in CCh, n = 8, p < 0.05). Additionally, later simulated EPSPs generate proportionally more action potentials than earlier simulated
EPSPs. Figure 3-5E is a summary plot of these data. Earlier simulated EPSPs account for
27.2 ± 1.7 % of these action potentials in control and 20.6 ± 0.9 % of these action potentials in CCh (n=8, p < 0.01). Additionally, the last simulated EPSPs account for
21.5 ± 1.6 % of these action potentials in control and 27.4 ± 1.5 % of these action potentials in CCh (n=8, p < 0.05). The shift in simulated EPSP generated action potentials to later EPSPs is most likely because of an accumulation of ADP current at later periods in the stimulus train. Thus, in addition to stimulus evoked afterdepolarizations, CCh application increases granule cell excitability. In addition, the increased number of action potentials generated during trains of simulated EPSPs, especially the increased proportion of action potentials generated during later EPSPs reflect the increased GABA release onto mitral cells. This suggests, as seen with the
101 mitral cell voltage clamp recordings of currents generated by glomerular stimulation, that a significant portion of the increased transmitter release in CCh is due to intrinsic excitability changes in granule cells, in addition to the effects that CCh might have on increasing transmitter release from granule cells (Castillo et al., 1999; Ghatpande et al.,
2006).
Persistent spiking in granule cells
In addition to converting the stimulus evoked AHP to ADP, granule cells that receive carbachol application can sometimes enter a persistent firing state reminiscent of layer five entorhinal cortex pyramidal cells (Egorov et al., 2002) and olfactory bulb Blanes cells (Pressler and Strowbridge, 2006). In Figure 3-6A two representative traces showing a granule cell, with an activity evoked AHP, and a Blanes cell, with an activity evoked persistent firing mode in control conditions. Granule cells, in the absence of pharmacological agents, normally have a large amplitude spike evoked AHP following current injections, and are therefore unable to respond to spike-evoking depolarizations with persistent firing. Blanes cells, however, normally generate an ADP following suprathreshold depolarizing stimulation that can easily translate into spike-evoked persistent firing at more depolarized potentials. Figure 3-6B is a comparison between activity evoked afterdepolarizations in granule cells (in 2 μM CCh) and Blanes cells (in control conditions). Both interneurons can generate spike evoked ADPs large enough to trigger additional action potentials, and, at slightly more depolarized resting membrane potentials, can generate persistent firing. However, the persistent firing in Blanes cells is
102 more stable than the persistent firing in granule cells during CCh application. The calculated coefficient of variation of instantaneous frequencies in granule cell and Blanes cell firing are significantly different (0.24 ± 0.08, n=13 for granule cells vs. 0.12 ± 0.05, n=11 for Blanes cells; p < 0.001), and this difference is illustrated in Figure 3-6C. While
Blanes cells can fire action potentials at a specific frequency for many minutes (Pressler and Strowbridge, 2006), granule cells have considerably more variation in firing frequency over a similar time window. Finally, in Figure 3-6D, a summary plot describes the difference between the resting membrane potential and persistent firing threshold in
Blanes cells (-53.9 ± 1.1 mV, n = 22; and -50.7 ± 0.9 mV, n=23) and granule cells (-74.3
± 1.6 mV, n=15; and -51.7 ± 1.0 mV, n=15). The resting membrane potentials of Blanes cells and granule cells are significantly different (p < 0.001), and this difference reflects one of the main differences in the execution of the persistent firing state in these two cells. While Blanes cells have a resting membrane potential only a few mV from the persistent firing threshold, granule cells normally have a resting membrane potential that are greater than tens of mV more hyperpolarized depolarized, thus restricting considerably the times when these neurons can enter a stimulus evoked persistent firing state.
103 Discussion
In this study, we present the first evidence showing muscarinic acetylcholine receptor
control of stimulus evoked after potentials in olfactory bulb granule cells. Muscarinic
acetylcholine receptor activation reduces the stimulus evoked AHP normally present in
granule cells, and reveals a stimulus evoked ADP. We also show that these stimulus
evoked ADPs in granule cells are similar to the ADPs that exist in another type of
olfactory bulb interneuron—the Blanes cell. The primary difference between granule
cells and Blanes cells in native conditions behind that granule cells normally have large
amplitude AHP instead of an ADP following depolarizing current injections. We used
mAChR receptor agonists and antagonists to show that M1 AChR activation is
responsible for the stimulus evoked ADP. The granule cell ADP also improves granule
cell excitability—the number of action potentials generated by suprathreshold current
injections and simulated EPSPs increases in mAChR agonists. Additionally, mitral cell recurrent and lateral inhibition is increased in the presence of mAChR agonists. Lastly,
when granule cells are held at depolarized membrane potentials close to firing threshold, depolarizing current injections can trigger persistent firing in carbachol. Our results indicate that cholinergic input to granule cells functions to increase granule cell excitability, enabling granule cells to enhance recurrent and lateral inhibition onto mitral cells, and continue firing action potentials after brief transient depolarizing stimuli.
104 The AHP to ADP conversion in granule cells is M1 AChR dependent
Our results provide evidence that suggest that M1 AChR activation in granule cells
converts the stimulus evoked AHP to an ADP. Activation of M1 AChRs typically act
through a G-coupled receptor complex that upregulates the production of phospholipase
C and inositol triphosphate which can also cause a marked increase in intracellular Ca.
The conversion of the stimulus evoked AHP to an ADP is probably mediated by an
increase in intracellular Ca which enables an inward current, likely to be ICAN, to
overcome the AHP. It could also be possible that the stimulus evoked AHP in granule
cells is mediated, at least in part, by M-current. The subsequent attenuation of the M-
current by mAChR activation would enhance the ADP as well. By contrast, M2 AChRs
typically act through a separate G-protein coupled receptor complex that reduces cAMP
levels and typically results in inhibitory effects.
Additionally, M1 AChR activation has also been shown to reduce the contribution of
certain kinds of intracellular Ca evoked AHPs in CA1 pyramidal neurons (Dutar and
Nicoll, 1988), probably through the same G-protein mediated pathway that results in a marked increase in intracellular Ca. At least some proportion of the AHP in granule cells is generated by slow, Ca evoked AHPs (Pressler and Strowbridge, data not shown). It is also likely that M1 AChR mediated reduction of these currents reveals the ICAN mediated
ADP, either by reducing the net stimulus evoked hyperpolarization, or, by removal of
potential shunting that could gate the ICAN mediated ADP. Additionally not much is
known about the channel correlate of ICAN. It is also likely that M1 AChR activation can
105 modulate this current directly via the same G-protein mediated pathway mentioned earlier. Future experiments will be necessary to define the modulation of ICAN and the granule cell AHP by cholinergic input.
Other investigations into cholinergic inputs onto granule cells have determined that carbachol, when acting at higher concentrations, functions to enhance GABA release from granule cells (Castillo et al. 1999, Ghatpande et al., 2006) while, simultaneously, inhibiting granule cell firing (Castillo et al., 1999). One team of investigators (Castillo et al., 1999) has determined that M2 AChR activation is responsible for both effects, while the more recent investigation by Ghatpande and colleagues (Ghatpande et al., 2006) has determined that, at least the increase in spontaneous transmitter release from granule cells is mediated by M1 AChR activation. There are several explanations for the discrepancy between the results of this paper, that carbachol acting through M1 AChRs, enhances granule cell excitability and that higher concentrations of carbachol, acting through M2
AChRs, decreases granule cell firing rate. First, it is likely that the higher concentration of carbachol enhances granule cell excitability to the degree that granule cells become hyperexcitable and enter depolarization block after they receive inputs that would have otherwise generated a train of non-decrementing action potentials. This effect of carbachol could be mistaken as an inhibitory effect in extracellular recordings. Secondly, it is also possible that the higher concentration of carbachol activates a larger proportion of M2 AChR receptors which inhibit the cell through a different second messenger system independent of the stimulus evoked ADP generated by M1 AChR activation.
Future experiments should address the intrinsic electrophysiological changes that occur
106 during M2 AChR activation in granule cells to determine the mechanism and function of
the possible inhibitory potentials reducing granule cell spiking.
Calcium activated nonselective cation current (ICAN) is responsible for the
afterdepolarization in granule cells
Carbachol, a mAChR agonist, and MCN-A-343, a selective m1AChR agonist, convert
the stimulus evoked AHP normally present in granule cells into an ADP. This response does not seem to be mediated by synaptic activity as granule cells are GABAergic (Jahr
and Nicoll, 1982). The carbachol evoked ADP in granule cells’ sensitivity to intracellular
Ca chelation (by an intracellular solution containing BAPTA), and by Ca blockade (by
Cd application) suggests that it is activated by a rise in intracellular Ca. Because
amphibian olfactory bulb granule cells (Hall and Delaney, 2002) express ICAN currents,
we found it likely that the Ca activated depolarizing currents in granule cells were also
ICAN mediated. In fact, uncaging of intracellular Ca by photolyzing Ca-bound NP-EGTA
generated a depolarizing current in voltage clamped granule cells with a corresponding decrease in input resistance. This suggests that the Ca-activated depolarizing current in granule cells is unlikely to be mediated by a Ca-activated reduction in standing K current.
The carbachol evoked ADP in granule cells is reversibly reduced by flufenamic acid, a compound known for its ability to selectively reduce ICAN currents (Ghamari-Langroudi and Bourque, 2002 and Partridge and Valenzuela, 2000). The voltage dependence of the stimulus evoked ADP is consistent with an ICAN current mediating the response. Also, the
107 inward current evoked by the Ca uncaging reversed near -20 mV, and its rectification at positive membrane potentials is also consistent with an ICAN current mediated ADP.
The carbachol revealed ADP in granule cells is also likely to underlie the persistent firing
state that can occur when granule cells receive suprathreshold stimulation at membrane
potentials near firing threshold. The persistent firing mode in granule cells is reminiscent
of the persistent firing mode in Blanes cells (Pressler and Strowbridge, 2006; in the
absence of pharmacological agents) and entorhinal cortical neurons (Egorov et al., 2002;
in the presence of carbachol). Persistent firing frequency in granule cells is likely
determined by the interactions between sodium and potassium currents, and single spike
evoked ICAN. In contrast to the persistent firing in other neurons generated by an intrinsic
mechanism (Egorov et al., 2002) however, granule cells and Blanes cells do not seem to
“encode” additional hyperpolarizing or depolarizing inputs as changes in firing rate. In
contrast, Blanes cells are remarkably robust in their persistent firing (Pressler and
Strowbridge, 2006). Granule cells do not share the same level of persistent firing frequency stability however, and seem to alter their firing frequency based on stochastic factors not yet illuminated by the current study.
Functional significance of cholinergic input to olfactory bulb granule cells and olfactory bulb processing
The cholinergic source of input to granule cells comes from the horizontal limb of the diagonal band. The horizontal limb of the diagonal band is part of the cholinergic basal
108 forebrain, a brain region that provides cholinergic input to a large portion of cortical and subcortical brain structures, and these inputs serve an important function in many sensory
processing tasks (Muir et al., 1994; Penschuck et al., 2002) including odor discrimination
(Linster et al., 2001). Inputs from the horizontal limb of the diagonal limb of the diagonal
band constitute the second largest portion of the centrifugal inputs to the olfactory bulb,
after piriform cortex (Macrides et al., 1981). Portions of the cholinergic inputs from this brain region to the olfactory bulb specifically terminate on granule cell proximal dendrites and gemmules (Kasa et al., 1995). Prior to this study there have been no
investigations that provide insight into the effects of mAChR stimulation on granule cell
intrinsic properties.
Cholinergic input to the granule cell makes the cell more excitable by enhancing the
duration and amplitude of stimulus-evoked depolarizations. Additionally, under special
circumstances, this afterdepolarization could enable the granule cell to enter a persistently
firing mode. Granule cells, while receiving cholinergic input, exhibit increased
excitability and stand to have increased dendrodendritic inhibitory output onto mitral
cells. The cholinergic input enhanced granule cell mediated mitral-mitral cell lateral
inhibition is likely to improve odor contrast during attention-driven olfactory behaviors.
Additionally, at depolarized membrane potentials brought on by barrages of EPSPs from
mitral cells, granule cells could enter a persistent firing state similar to the persistent
firing state found in Blanes cells. Blanes cells readily enter this persistent firing state as
their resting membrane potential is only a few millivolts away from the persistent firing
threshold, but granule cells normally rest at membrane potentials far more hyperpolarized
109 than their persistent firing threshold. Granule cells, by contrast, are relatively unlikely
ever to reach a point where they can become persistently active. In order to reach this
firing mode, they would need to receive a barrage of EPSPs in order to elevate them to the persistent firing threshold, and this level of voltage depolarization would not last long
enough for any sustained firing like we see in Blanes cells or entorhinal cortical neurons.
It is much more likely that the persistent firing mode in granule cells instead serves to generate additional action potentials when they receive input from mitral cells, especially on later EPSPs. Since it has been shown by Desmaisons and colleagues (Desmaisons et al., 1999) that inhibitory inputs to mitral cells can generate rebound spikes, it is possible that the additional spikes in granule cells would actually enhance mitral cell output. In the olfactory bulb, local circuits driven by mitral cell activity generated by olfactory receptor neuron input to the glomerulus that is repeated as a module throughout this brain region.
It is likely that the cholinergic input from the basal forebrain functions to synchronize granule cells, as they would all be receiving a similarly patterned cholinergic input, possibly aligning the activity of multiple glomerular modules of mitral cell output. Blanes cells, as they have long axons that end in large axon plexuses, are more likely to act as a way to disinhibit neighboring glomerular modules of mitral cells and to pattern granule cell output, especially during periods of cholinergic innervation and the resultant excitability increase in granule cells.
110
Figure 3-1
111 Figure 3-1. Carbachol reveals an afterdepolarization in granule cells
(A) Afterhyperpolarization evoked by depolarizing current step in a granule cell.
Carbachol (2 μM) application reveals an afterdepolarization evoked by the same
depolarizing current step. The afterdepolarization is then abolished by the intracellular Ca
chelator BAPTA. (B1) FFA (100 μM) application reduces the size of the evoked
afterdepolarization in CCh. Responses enlarged in (B2). (C1) Summary of afterpotential
amplitudes in control, CCh (2 μM) at -60 mV, CCh (2 μM) at -80 mV, and CCh (2 μM)
at -60 mV and intracellular BAPTA. (C2) Summary of afterpotential amplitude in Blanes cells in control and CCh (2 μM) at -60 mV. (D) The afterdepolarization in Blanes cells is
unaffected by CCh (2 μM) application.
112
Figure 3-2
113 Figure 3-2. Calcium-dependent afterdepolarizations in granule cells
(A) Cadmium application abolishes the granule cell activity evoked afterdepolarization in
CCh (2 μM). (B) Summary plot of granule cell afterpotential amplitude in control, CCh
(2 μM, n=19), and CCh (2 μM) with Cadmium (200 μM, n =5). **p < 0.001, *p < 0.01.
(C1) Photolyzing NP-EGTA (2 mM) evokes an inward current in granule cells. The horizontal line above the current trace indicates the duration of the UV exposure.
Responses to a series of graded duration UV exposures (100, 250, 500, and 1000 ms duration). (C2) A 1000 ms duration UV exposure fails to elicit a response in a granule cell
filled with control internal solution that lacks NP-EGTA. (D) Currents evoked by
photolyzing NP-EGTA at different membrane potentials indicate that the uncaging currents reversal is near 0 mV. (E) Summary plot of the peak amplitude of the responses to UV exposure at different membrane potentials in voltage clamped granule cells.
114
Figure 3-3
115 Figure 3-3. Pharmacology of mACh receptor activation in granule cells
(A) The afterdepolarization in granule cells revealed by CCh (2 μM) is blocked by the addition of atropine (5 μM). (B) The further addition of AFDX-116 (1 μM, n=5) did not affect the CCH revealed afterdepolarization. The inset shows superposition and enlargement of the afterdepolarizations before and after adding AFDX-116 to the extracellular solution. (C) The granule cell CCh revealed afterdepolarization is abolished by pirenzepine (10 μM, n=5). The inset shows a superposition and enlargement of the afterdepolarization in CCH and the reduced afterpotential in CCh and pirenzepine. (D)
Application of MCN-A-343 (100 μM, n=5) can also reveal the latent afterdepolarization in granule cells. The inset shows a superposition and enlargement of the control afterpotential and the afterdepolarization revealed by the addition of MCN-A-343. (E1)
Summary chart showing the average afterpotential amplitude in control (n=44), CCh (2
μM, n=19), and MCN-A-343 (100 μM, n=5). **p < 0.0001. (E2) Summary chart showing the average afterpotential amplitude in CCh (2 μM) after the further addition of AFDX-
116 (1 μM, n=5), pirenzepine (10 μM, n=5), and cadmium (200 μM, n=5). *p < 0.01.
116
Figure 3-4
117 Figure 3-4. Increased concentrations of carbachol enhance granule cell excitability
(A) Representative stimulus evoked afterpotentials in control, CCh (2 μM) and CCh (50
μM, in a different cell). (B) Summary chart showing afterpotential amplitude in control
(n=44), CCh (2 μM, n=19) and CCh (50 μM, n=8). ***p < 0.001, **p<0.01. (C)
Representative traces showing action potentials generated by a 50 pA current step in control, CCh (2 μM) and CCh (50 μM). (D) Summary chart showing the average number of action potentials generated by 50 pA steps in control (n=12), CCh (2 μM, n=7) and
CCh (50 μM, n=7). ***p < 0.001, *p < 0.05. (E) Enlargement of the representative traces shown in (A). (F) Summary chart showing average action potential amplitude in control
(n=11), CCh (2 μM, n=6), and CCh (50 μM, n=8). The white bar represents the average amplitude of the first action potential, the hatched bar represents the average amplitude of the second action potential, and the gray bar represents the average amplitude of the third action potential. **p < 0.01.
118
Figure 3-5
119 Figure 3-5. Carbachol enhances granule cell output onto mitral cells
(A) Responses in a voltage-clamped mitral cell to four consecutive shocks (100 μA for 1 ms) in the glomerular-layer. Note the evoked current increase in CCh (2 μM). (B)
Responses of a granule cell to simulated EPSPs (α; τ = 100 ms) in control and after CCh
(2 μM) application. Note the activity increased in CCh (2 μM). (C) Summary chart showing the integrated current (nA *s) for 3 s following the last shock in the control and
CCh. CCh application significantly increased the size of the shock evoked inward current
(n=4). *p < 0.05. (D) Summary chart showing the action potentials generated by simulated EPSPs increased significantly in CCh. The number of action potentials generated by simulated EPSPs in CCh (2 μM) increased significantly (n=8). *p < 0.05.
(E) Summary chart showing the percentage of action potentials generated during each simulated EPSP in control and CCh (2 μM; n=8). Note the increased in action potentials generated by later simulated EPSPs. **p < 0.01, *p < 0.05.
120
Figure 3-6
121 Figure 3-6. During Carbachol application, brief depolarizations in granule cells can
trigger persistent firing
(A) Comparison of the afterpotential responses of granule cells and Blanes cells to depolarizing current injections when held near firing threshold. (B) Responses to constant step depolarizations at different membrane potentials. Persistent firing can be evoked at the most depolarizing membrane potentials in granule cells (in CCh, 2 μM) and Blanes cells (in control conditions). The numbers above the persistent firing examples denote averages of instantaneous firing frequency during the period indicated by the bold line.
(C) A comparison of persistent firing in a representative granule cell and Blanes cell. The
CV of instantaneous frequencies of granule cell and Blanes cell firing are significantly different (0.24 ± 0.08, n=13 for granule cells vs. 0.12 ± 0.05, n=11 for Blanes cells; p <
0.001). (D) A comparison of the resting membrane potential and persistent firing threshold in Blanes cells and granule cells.
122
Chapter 4
Discussion
123 In summary, the principal conclusions are as follows: 1) olfactory bulb Blanes cells have
an ICAN mediated intrinsic ADP that can generate persistent firing in response to a suprathreshold stimulus, 2) Blanes cells monosynaptically inhibit granule cells and prevent granule cell action potential generation, 3) olfactory bulb granule cells typically
have a large amplitude stimulus evoked AHP, but M1 AChR activation reveals an ADP
in these cells. The activation of M1 AChRs enhances granule cell excitability in response
to depolarizing current steps and simulated EPSPs and 4) during periods of M1 AChR
activation, granule cells also have ICAN mediated ADPs that can generate persistent firing
under defined circumstances.
In this thesis, I set out to define the source and functional properties of the major synaptic
inputs onto granule cells. I defined a novel inhibitory input onto granule cells that
originates from a previously uncharacterized olfactory bulb interneuron-- the Blanes cell.
I characterized the intrinsic properties of these neurons as well as their unusual
afterdepolarization and persistent firing mode. This pathway, when Blanes cells become
persistently active, evokes periodic persistent IPSPs onto granule cells that sculpt granule
cell activity and tonically inhibit them over long periods.
I also defined the modulatory properties of muscarinic cholinergic input to granule cells.
This pathway originates from centrifugal inputs from the horizontal limb of the diagonal
band and can be mimicked by carbachol application to an olfactory bulb slice.
Cholinergic inputs appear to function to down-modulate the primary intrinsic
afterhyperpolarization response in granule cells. Carbachol application also reveals a
124 latent afterdepolarization in granule cells, which enhances excitability and can even lead
to persistent firing in certain situations. This underlying stimulus evoked
afterdepolarization that was very similar mechanistically to the afterdepolarization in
Blanes cells under native conditions. This discussion of these results is divided into five major sections—(1) a discussion of the function and mechanisms of activity evoked afterdepolarizations and persistent firing modes in olfactory bulb neurons, (2) a discussion of two classes of granule cell input that influence granule cell activity and the differing functional roles these inputs play in olfactory bulb function, (3) a discussion of
how the circuitry of the mammalian olfactory bulb compares to other mammalian CNS
sensory processing areas, like the retina, (4) a discussion of the similarities and
differences between intrinsic and synaptic modulation input, as well as opportunities to
test the involvement of these inputs in a computational model of the olfactory bulb, and
(5) a discussion of ways to ascertain the contributions of these inputs on olfactory coding
in intact and behaving animals.
Enhancement of excitability by an intrinsic activity-evoked afterdepolarization
The synthesis of the extracellular inputs (in the form of receptor or signaling molecule activation), active conductances, and passive membrane properties govern the output properties of a neuron. One of the clear characteristics that separates Blanes cells from most other bulbar neurons is the presence of a large afterdepolarization immediately following suprathreshold activity. Granule cells can also generate a large afterdepolarization following suprathreshold activity, but they can only do so in certain
125 situations (as previously discussed in the results section). In granule cells, typically suprathreshold stimulation generates large slow AHPs.
Potassium channel mediated AHPs can take a number of forms. The three categories of
AHPs are fast, medium, and slow—so named because of their time course of action.
Many different potassium currents can contribute to these three different AHPs. Common currents that generate AHPs in other cells are IM, IC, IBK, ISK, and IAHP. The IC and IBK currents are both Ca activated, voltage dependent potassium currents responsible for the fast AHP (Adams et al., 1982; Barrett at al., 1982; Marty, 1981). These currents require a large concentration of intracellular Ca to become active (1-10 μM), and in neurons are usually situated near Ca channels (Schreiber and Salkoff, 1997; Cox et al., 1997; Roper et al., 2003) These currents are mainly responsible for spike-repolarization as well as a brief window of hyperpolarization immediately following the action potential. This window of hyperpolarization typically does not extend for longer than 20-50 ms (Lancaster and
Adams, 1986). The IM current, a medium AHP, activates independent of intracellular Ca,
and activates and deactivates slowly (Adams et al., 1982). Typically M-currents have a
low activation potential (typically -60 to -70 mV), so that any suprathreshold
depolarization that occurs in this window will result in the activation of IM and the current
will drive the cell towards its resting membrane potential (Adams et al., 1982; Storm
1989; Brown and Griffith, 1983; Jones, 1985; Madison et al., 1987; Adams and Brown,
1982). M-current is described as being responsible for “medium AHP”, and is noticeably
attenuated by cholinergic input—specifically M1 AChR activation (Hamilton et al. 1997;
Jones, 1985) M2 AChR activation (Dutar and Nicoll 1988). The last two potassium
126 currents, ISK and IAHP are also Ca dependent like the IBK current, except these currents do
not exhibit any voltage dependence (Pennefather et al., 1985). These currents are
activated by significantly lower concentrations of intracellular Ca, typically 50-900 nM,
and are not commonly found in close proximity to Ca channels (Constanti and Sim, 1986;
Lancaster and Adams, 1986; Roper et al., 2003; Lancaster et al., 1991; Park 1994). These
currents are responsible for the stimulus evoked slow AHP (Lancaster et al., 1991; Park
1994; Pennfather et al., 1985; Lancaster and Adams, 1986; Constanti and Sim, 1987; Sah,
1995). These currents activate and deactivate slowly, over the time course of seconds
rather than milliseconds (Lancaster and Adams, 1986). These currents, because of their
slow time courses, are not involved in action potential repolarization and instead provide
the stimulus evoked inhibition in these neurons, which inhibits spiking after bursts of
action potentials (Lancaster and Adams, 1986). Muscarinic cholinergic receptor
activation, specifically, M1 AChRs in hippocampal CA1 pyramidal cells (Dutar and
Nicoll, 1988) can selectively abolish this current. Additionally, in CA1 pyrmidal neurons,
norepinephrine (Lancaster and Adams, 1986), serotonin (Andrade and Nicoll, 1987), and
histamine (Haas and Konnerth, 1983) also selectively reduce the IAHP current.
In granule cells, it is not clear what current or combination of current is responsible for
the stimulus evoked AHP. In the results section of this thesis I present evidence showing
that M1 AChR activation converts the granule cells’ slow AHP to an ADP. The action of this muscarinic receptor could be blocking the action of the AHP directly, and thus revealing a latent ICAN current. This result also suggests that IAHP and not ISK mediate the
AHP (as Gulledge and Stuart, 2005 shows m1AChR activation can lead to intracellular
127 Ca release which activates ISK and hyperpolarize the neuron). However, an alternative
explanation is that M1 AChR activation causes intracellular Ca release that enhances the
ADP, and causes this current to overshadow the competing AHP currents. Because Ca regulates a number of K channels that govern stimulus evoked AHPs, in this instance, Ca can be thought of as a second messenger ion that regulates degrees of inhibition. One way to test the difference between these two likely mechanisms of M1 AChR activation is to
look at the individual contributions to these two different Ca activated currents. Phorbol
esters, through their ability to increase protein kinase C, have been shown to block IAHP
by other investigators (Dutar and Nicoll, 1988). By doing a current clamp experiment at a relatively positive membrane potential (-50mV), in the prescence of phorbol esters (5 μM phorbol 12,13-diacetate) one could measure the relative change of the ICAN before and
after carbachol application. Then, to test the change in IAHP current due to carbachol, you
could repeat the current clamp recordings from a granule cell, without phorbol esters in
the bath solution, and observe the spike evoked AHP before and after carbachol
application.
In olfactory bulb granule cells, intracellular Ca also activates an ADP, so in this cell type,
Ca may be modulating the degree of post-stimulus activation in these cells. Activity-
evoked afterdepolarizations take a transient input and extend it in amplitude and time
course. In this way, the activity-evoked afterdepolarizations can function to prolong the
synaptic input to a neuron, and thereby facilitate the neuronal output. Additionally, if the
currents that generate the afterdepolarization are differentially expressed in dendritic
compartments, the afterdepolarization could also be well suited to amplify inputs to
128 specific dendrites (or dendritic compartments), and even allow the cell to create divergent
output modes depending on the location of the depolarizing inputs to its location in the
dendritic tree. Cholinergic input to granule cells could also selectively activate specific dendritic compartments, and the summation of these two factors-- one intrinsic, and one
extrinsic, could have a powerful effect on granule cell output. For example, if certain
classes of excitatory inputs only innervate the large Blanes cell dendrite that is oriented
towards the mitral cell layer, and this dendrite is the primary source of ICAN current in
these cells, then this could be a neuroanatomical solution to specifically amplifying one
set of inputs onto these neurons. To a first approximation, an activity-evoked
afterdepolarization can function as an output gain amplifier; its presence in the neuron
serves to extend the firing duration and increase the firing frequency of the host neuron.
Activity-evoked afterdepolarizations are a common feature found in many different types
of neurons that are iterated in other systems using different strategies. Either intrinsic
active conductances generate these afterdepolarizations, or by recurrent synaptic
feedback generates these afterdepolarizations— each mode of generation could have
important functional consequences. In cat motor neurons, for example, the active
properties of L-type calcium channel activation produce a large activity evoked
afterdepolarization (Russo and Hounsgaard 1999). This afterdepolarization produces a
plateau potential in motor neurons that generate additional action potentials following
transient stimuli. These L-type calcium channel mediated plateau potentials appear to recruit additional motor neurons during movement tasks. In layer V entorhinal cortex,
pyramidal neurons typically express an AHP following transient current injection;
129 however, the application of the nonspecific muscarinic receptor agonist carbachol reveals a large afterdepolarization (Egorov et al. 2002). In carbachol, the cells are capable of firing persistently in a similar manner as Blanes cells. Entorhinal cells are also able to augment their firing frequency based on subsequent inputs once they have entered the persistent firing state. Transient depolarizing inputs cause these neurons to fire faster, and
transient hyperpolarizing inputs can reduce firing rate— as a result the persistent firing rate of these neurons is closely determined by the history of depolarizing and hyperpolarizing inputs. This property could account for the ability of the entorhinal cortex to function in its role of information retrieval and storage, as the entorhinal cortex
is the primary interface linking the neocortex to the hippocampus, and the hippocampal lesions can produce profound memory deficits (Morris et al., 1982). Afterdepolarization- like currents can also be observed in goldfish area I neurons involved in eye movements.
The frequency of firing in these position control is an indication of eye position in the nasal to temporal axis (Aksay, Baker et al. 2000). When a neuron in this region receives a direct hyperpolarizing current injection to prevent it from firing action potentials, the membrane potential depolarizes and hyperpolarizes in square-like deflections, corresponding almost exactly with eye saccades. An interesting difference between these neurons, and the other neurons previously described above, is that the step changes in current and the increased firing frequency that is associated with changes in eye position are all governed by the eye position circuitry, as intracellular depolarizing current injections failed to elicit plateau potentials or sustained firing.
130 The afterdepolarization in Blanes cells is most similar to the plateau in motorneurons
(Russo and Hounsgaard 1999) and the carbachol revealed ADP in layer V entorhinal
cortical pyramidal neurons (Egorov et al., 2002). Results from my thesis support that ICAN
current in Blanes cells mediates the stimulus-evoked afterdepolarization, and as a result,
intracellular calcium release activates the afterdepolarization. It can be activated by
suprathreshold stimulation (see chapter 2), and can thereby serve to extend the duration
and amplitude of the suprathreshold stimulus by providing additional depolarizing
current. The ICAN current has a long time course of activation, which enables the cell to receive additional depolarization for a time far outlasting the initial transient input (see
chapter 2).
Because ICAN current is partially outwardly rectifying (see Figure 2-5), an interesting emergent property of this current is that it will only provide the additional depolarizing current during periods of relatively depolarized membrane potentials (typically greater than -70 mV), which suggests this afterdepolarization can also function as a limited
coincidence detector. In essence, it can enable extended periods of increased excitability
only after the cell receives suprathreshold input at more depolarized potentials. Blanes
cells have resting membrane potentials well into the range at which the ICAN mediated
afterdepolarization can become active, so that the ADP amplifies the response of nearly
every suprathreshold input. Hypothetically, specifically timed IPSPs onto Blanes cells
could shunt or terminate the ADP. The afterdepolarization in granule cells also seems to
be ICAN mediated, like Blanes cells (see results section). One notable difference is that
granule cells have a much more hyperpolarized resting membrane potential (-75 mV), so
131 granule cells would typically need to be depolarized considerably in order for the
afterdepolarization to be evident. Additionally, like layer V entorhinal cortical neurons
(Egorov et al. 2002), granule cell afterdepolarizations are revealed during periods of
active cholinergic input from the basal forebrain, suggesting that attentional modulation
of olfactory output is at least, in part, to modulate granule cell excitability.
In Blanes cells, the intrinsic afterdepolarization amplifies and prolongs nearly every
suprathreshold input at membrane potentials more depolarized than -70 mV. It is
unknown if Blanes cells preferentially express ICAN current in different dendritic
locations, as differential ICAN density along the dendritic arborizations could have
interesting functional consequences in terms of the effects that different excitatory inputs
could have. One could test this by examining the contribution of ICAN current along the
soma and dendrites of Blanes cells by UV-photolyzing Ca-bound NP-EGTA. By using
voltage-clamp recordings from many neurons with NP-EGTA and Alexa 594 in the
intracellular solution, one could visualize the dendritic arborizations, and then examine the I CAN currents generated by Ca-uncaging at specific locations along the soma and
dendritic arborizations. One could use this method to obtain a description of channel
density along these cellular processes. Suprathreshold excitatory input onto Blanes cells
(presumably from mitral cell axon collaterals), would generate afterdepolarizations in
Blanes cells that could produce extended periods of a depolarized membrane potential that is closer to the action potential threshold in these cells. The added depolarization insures that additional excitatory inputs in this period could be substantially smaller and still bring the Blanes cell to produce action potentials. The afterdepolarization in Blanes
132 cells would enhance even a few spikes from mitral cells, and thus produce much more
inhibitory activity onto the granule cells that are postsynaptic to the Blanes cell.
By contrast, the afterdepolarization in granule cells would function differently. Granule
cells receive large amplitude slow time course EPSPs from mitral cells in vivo (Cang and
Isaacson 2003). During an EPSP barrage, granule cells typically depolarize to fire a
number of action potentials. While receiving cholinergic input, granule cells that receive
these EPSP barrages from mitral cells would fire additional spikes, as resulting
depolarization from the EPSPs will bring the cells sufficiently depolarized as to activate
the ICAN mediated afterdepolarization. The enhanced firing during EPSPs would likely
produce more spikes at later EPSPs (see chapter 3), because of the summation of the
spike-evoked afterdepolarization. The function of the afterdepolarization in granule cells
is, in this case, to distribute the action potentials towards the end of the barrage of EPSPs
(for a circuit schematic in normal conditions and during cholinergic input, consult Figure
4-1).
Afterdepolarizations as a mechanism for generating persistent firing
Working memory acts as a type of “temporary” informational storage where small amounts of information are “held”, briefly, in order to facilitate the completion of objectives (Funahashi et al., 1989; Deco and Rolls, 2003; Durstewitz et al., 2000). The first evidence of the electrophysiological properties of working memory came from studies in awake and behaving monkeys (Fuster and Alexander 1971; Kubota and Niki
133 1971) where, in vivo electrophysiological recordings of neurons in the prefrontal cortex showed persistently firing neurons during the delay period in a delayed response task.
These neurons that are persistently active during the delay period, and it is likely that the persistent activity provides the mechanism enabling this cortical area to maintain the visual stimulus throughout the delay period. Persistent firing has received a lot of attention because, as Goldman-Rakic put it, “it is thought to be the cellular basis of working memory” (Goldman-Rakic 1995). In addition to the prefrontal cortex, the goldfish ocular-motor saccade circuit and the layer V entorhinal cortical neurons (as mentioned above) operate along similar lines. In the goldfish saccade circuit, the firing rate of the persistently active position cells are thought to encode the “eye position” of the goldfish eyes—as the firing rate increases or decreases, the goldfishes eyes actually move accordingly. With the layer V entorhinal cortical neuron example, it is not yet known what the actual behavioral result of the varying frequencies in the persistent activity. However, the ability of these neurons to encode additional depolarizing or hyperpolarizing currents as step increases in persistent firing is thought to be physiologically important, especially considering its place in the entorhinal cortex— one of the main input regions to the hippocampus.
The afterdepolarization present in granule cells and Blanes cells can lead to a persistent firing state in these cells. The persistent firing in Blanes cells and granule cells is different from these other examples of persistent firing in other systems. Granule cells fire persistently because of an intrinsic intracellular mechanism similar to those cells in the entorhinal cortex. Blanes cells however, can fire persistently even without the further
134 addition of cholinergic input. In contrast to the persistent firing in other systems however,
granule cells and Blanes cells do not seem to “encode” additional hyperpolarizing or
depolarizing inputs as changes in firing rate. On the contrary, Blanes cells (and to a lesser
extent granule cells), are remarkably robust in their persistent firing. Because the
olfactory bulb is a second-order brain structure responsible for olfactory information encoding, it is likely that the persistent firing mode in Blanes cells represents not an
olfactory version of working memory, but, a temporary information storage mechanism,
at the encoding level, which is responsible for inhibiting and sculpting granule cell output
on a long time scale. Persistent inhibitory inputs onto Blanes cells, resembling those used
in the results section to terminate persistent firing, could function to reset the ability of
Blanes cells to maintain the memory of depolarizing stimuli. Because Blanes cells
typically have long axons (as long as 1 mm in some cases), and, because Blanes cells
typically have large axon plexuses (with diameters surpassing 250 μm), it is likely that
the principal action of their persistent activity would be to inhibit phasically large groups
of granule cells that are found in neighboring glomerular modules. If Blanes cells do
indeed receive input from mitral cells (which is likely given the results of stimulation
experiments, see chapter 2), it is also likely that they could be turned on by mitral cells in
neighboring glomeruli. Since every glomerulus receives innervation from odorant
receptor neurons expressing the same odorant receptor, it is likely that Blanes cells are
receiving converging input telling them of a specific combination of odor epitopes, and
they are remembering this information for many minutes by phasically inhibiting granule
cells over the course of many minutes. The phasic inhibition of granule cells in this
manner may also function to disinhibit mitral cells over the same time scale.
135
Granule cells, by contrast, are relatively unlikely ever to reach a point where they can become persistently active. Granule cells typically exhibit a resting membrane potential significantly hyperpolarized to their persistent firing threshold (see chapter 3). In order to reach this firing mode, they would need to receive a barrage of EPSPs in order to elevate them to the persistent firing threshold, and this level of voltage depolarization would not be long enough for any long sustained firing level. It is much more likely that the persistent firing mode in granule cells instead functions as an intrinsic mechanism to generate additional action potentials when they receive input from mitral cells, especially on later EPSPs. Additionally, it has been shown by Desmaisons and colleagues
(Desmaisons, 1999) that transient inhibitory inputs to mitral cells can generate rebound spikes. One possibility is that the additional spikes in granule cells would actually go to enhance mitral cell output.
Similarities and differences between olfactory bulb circuitry and retinal circuitry
Both the olfactory bulb and the retina are regions of the central nervous system necessary for the initial processing and encoding of sensory information. While the retina has a great number of circuit complexities making it as diverse as the olfactory bulb, the overall circuit organization in these two brain regions is similar. The retina is organized in modules associated with small receptive fields of spatially defined input, in this case, visual input in the form of photons of light instead of chemical odorants. Cone and or rod cells transduce the visual information into neural code by briefly ceasing inhitory input
136 onto bipolar cells after receiving photon stimulation (Sterling and Demb, 2004). Bipolar neurons, labeled “on” or “off” bipolar neurons based on their input onto retinal ganglion cells in response to light (“on”, excitatory) or darkness (“off”, inhibitory). Retinal ganglion cells thus retain this naming system based on their response to light or darkness, and then convey this information to higher brain regions. Retinal ganglion cells also receive input that mediates horizontal interactions between neighboring retinal ganglion cells and bipolar cells from amacrine cells. These kinds of inputs enable retinal ganglion cells to integrate inputs from neighboring neurons to encode spatially complex aspects to visual input such as edges and contours. The retina—just like the olfactory bulb, is organized into modules based on spatially defined input from the neurons that transducer the stimulus. Additionally, the principal cells of these two brain regions receive top-down input from sensory neurons, as well as horizontal inputs that enable neurons to provide contrast between similar but different inputs. Interestingly, both retinal amacrine cells and olfactory bulb granule cells are similar in that these neurons lack axons, and mediate horizontal interactions between principal cells through dendritic transmitter release.
These remarkable similarities imply that local circuits that utilize functionally relevant synaptic interactions such as recurrent excitation, feedback inhibition, feed-forward inhibition and lateral inhibition, are assembled and iterated within brain regions and across the CNS. Another interesting emergent property of dividing functionally distinct brain regions into modules of synaptically relevant local circuits is that it is possible to use a small number of densely innervating inputs to modulate these circuits as contextual properties change. The retina, in addition to the olfactory bulb, receives a large amount of modulatory inputs that regulate the firing of principal neurons. The results of this thesis
137 show a functionally interesting feed-forward synapse from Blanes cells onto granule
cells-- cells that mediate horizontal interactions between principal neurons in the
olfactory bulb. One hypothesis generated from this finding, and from the similarities in
organizational principles governing the retina and the olfactory bulb, is that the retina
also has a class of interneuron that creates feed-forward inhibitory inputs onto amacrine
cells.
Factors that modulate granule cell activity
The mechanism of using a relatively small number of cholinergic and GABAergic afferents to inhibitory interneurons to affect significant changes on principal cell output is repeated in many brain regions (Freund and Antal 1988; Freund and Meskenaite 1992).
Much like in the hippocampus and cortex, inhibitory interneurons of the olfactory bulb receive input from neurons in the cholinergic basal forebrain. These olfactory bulb interneurons, the granule cells, make synapses with many mitral cells (Price and Powell
1970b), and inputs onto individual granule cells that modulate their activity could actually exert powerful effects on olfactory bulb processing, as the number of mitral cells
you could influence scales up dramatically.
In this thesis I present results that describe two separate inputs onto granule cells—
GABAergic inhibitory inputs originating from Blanes cells located within the olfactory
bulb, and centrifugal cholinergic inputs. These two inputs onto granule cells modulate
granule cell firing in different ways and have different functions. The activity of mitral
138 cells is a representation, in neural code, of olfactory encoded information. Granule cells
provide the principal inhibitory input to mitral cells (Ezeh et al., 1993), and they refine
and sculpt mitral cell output, which is refining and sculpting the individual mitral cells
contribution to olfactory code. The main subject of this portion of the discussion will be
on the topic of what these different inputs are doing to granule cell activity.
Granule cells receive cholinergic input from the horizontal limb of the diagonal band, a
part of the cholinergic basal forebrain that is composed of cholinergic and GABAergic
neurons that send centrifugally oriented projections to other limbic and cortical brain
regions. Cholinergic input originates from the horizontal limb of the diagonal band, and
innervates cells in every layer of the olfactory bulb. Granule cell somata, proximal
dendrites, and gemmules are a significant source of cholinergic innervation. Cholinergic
fibers also synapse onto periglomerular and juxtaglomerular cells in the glomerular layer,
mitral cell somata and dendrites in the EPL and mitral cell layer, and short axon cells in
the granule cell layer (Kasa et al., 1995). The cholinergic projections to the olfactory bulb
are best understood in the context of the cholinergic projections to the surrounding limbic
brain structures. While doing investigative sniffing, the sniff frequency entrains to the slow wave theta rhythm present in the hippocampus, and a preferred activity latency relationship is formed between olfactory bulb output, olfactory cortex activity, and
hippocampal activity (Macrides et al., 1982). This preferred latency relationship comes
about as a way to link the activity in the hippocampus and other limbic structures to the
output from the olfactory bulb, as a means to enhance olfactory processing and anaylsis
(Macrides et al., 1982).
139
As discussed in the introduction, the cholinergic basal forebrain is the centralized brain
region that modulates the processing in these different brain regions, and because of the cholinergic basal forebrains importance in attentional regulation, it is well suited to do so.
The cholinergic basal forebrain, especially the horizontal limb of the diagonal band, also receives sizable input from the olfactory system—these brain regions regulate each other in a reciprocal manner. In addition to odor discrimination deficits in horizontal limb of the diagonal band lesioning studies, there is a pronounced deficit in the formation of olfactory associated memories (Wilson et al. 2004). This implies that not only are cholinergic inputs important for the coding of olfactory information, but they are important as well for the consolidation of this information into long-term storage.
As discussed previously, cholinergic input to the granule cell increases cell excitability, and under defined circumstances could enable the granule cell to enter a persistently firing mode. Granule cells receive cholinergic input from the horizontal limb of the diagonal band primarily onto the primary dendritic tree, its arborizations, and even directly onto individual gemmules. Periods of intense cholinergic input could generate the changes in excitability described in the results section, but it is also possible that the specific inputs to the gemmules could have a different effect. Ghatpande et al. (2006) defined the effect that cholinergic input has on spontaneous transmitter release from granule cells that was originally described by Castillo et al. (Castillo et al. 1999).
Ghatpande and colleagues noted that M1ACh receptor specific agonists cause calcium release from stores that significantly increase mIPSC frequency onto mitral cells.
140 Cholinergic activation of the dendritic tree might produce the excitability changes noted in the results, while specific activation of the gemmules might function more specifically to increase transmitter-release probability. Granule cells that receive cholinergic input exhibit increased excitability and are likely to have increased dendrodendritic inhibitory output onto mitral cells. Perhaps this cholinergic input enhances granule cell mediated mitral-mitral cell lateral inhibition, and it is this increase in lateral inhibition that represents itself in odor perception as enhanced contrast between glomeruli. Fibers that synapse onto gemmules may accomplish a similar function but on a more restricted scale.
In any event, it seems clear from the results of this study that cholinergic input to granule cells would serve to enhance dendrodendritic inhibition onto mitral cells.
Granule cells also, as discussed in the results section, receive inhibitory input from
Blanes cells. Blanes cell mediated IPSPs onto granule cells are different from the cholinergic inputs they receive from the basal forebrain, as the Blanes cell mediated inputs are inputs originating from the olfactory bulb, and are the result of feedforward inhibitory inputs from mitral cells. If the activity of mitral cells is a representation of olfactory code, then Blanes cells could be activated by certain combinations of olfactory characteristics, which then selectively inhibit a large population of granule cells involved in dendrodendritic inhibition and lateral inhibition with separate compartments of glomerular columns. As discussed previously in the results, Blanes cells can readily enter a persistently active state, as their resting membrane potential is very close to persistent firing threshold. Single shocks in the glomerular layer can cause Blanes cells to fire persistently, and brief trains of stimulation in the granule cell layer is also reliable at
141 generating persistent firing in Blanes cells. Blanes cell activation and resulting persistent
feed forward inhibition onto granule cells is a normal functioning property of olfactory
encoding. Blanes cells activated by specific combinations of mitral cells phasically
inhibit granule cells in order to reduce granule cell mediated dendrodendritic and lateral inhibition onto mitral cells during specific intervals, or to regulate granule cell firing to
specific frequency domains. The location of the inhibitory synapse from Blanes cells
could also have some functional importance, as it might serve to accomplish a number of
important functions in addition to phasic inhibitory input. If the Blanes cell inhibitory
input falls mainly onto the proximal dendrite, it could serve to shunt action potentials
from propagating into the granule cell dendritic tree, in effect uncoupling the different
dendritic branches from one-another. In other brain areas, there is precedence for focal
dendritic inputs coactivating several synapses and generating a NMDA-driven dendritic
spikes (Rhodes, 2006). Additionally, Zelles and his colleagues have shown that somatic
action potentials in granule cells propagate into the dendritic arborizations to produce
large Ca transients (Zelles et al., 2006). Blanes cells inhibitory inputs could play an
important role in regulating these different types of dendritic spikes. If the Blanes cell
inhibitory input synapses onto granule cell dendritic branch points, this input could
function specifically to shunt some branches from other branches—thus enabling somatic
action potentials from still activating some branches, while keeping other branches
electrically isolated. In addition to these properties, the primary function of Blanes cell
mediated persistent inhibitory synapses onto granule cells may be to transiently disinhibit
mitral cell output by inhibiting granule cells dendrodendritic and lateral inhibition, thus
142 producing temporally aligned periods of increased activity in mitral cell output
determined by Blanes cell firing frequency.
These two inputs to granule cells are under different originating sources of control, and they function to accomplish different functional objectives in olfactory processing. The cholinergic projection onto granule cells from the cholinergic basal forebrain may enhance lateral inhibition between mitral cells (see Figure 4-1). By contrast, the inhibitory projection from Blanes cells might decrease lateral inhibition between mitral cells, thereby reducing contrast between mitral cell outputs. A common theme in the nervous system is the use of populations of inhibitory interneurons to modulate and sculpt the activity of principal neurons (Shepherd and Greer, 1998). These two separate inputs to granule cells may exploit the olfactory bulb circuitry in precise ways to exert these differences. The activity of mitral cells recorded in vivo from animals doing olfactory behaviors evolves over time, alternating between periods of intense firing and inhibition. The interplay between Blanes cell mediated persistent disinhibition of granule cell activity, and cholinergic basal forebrain mediated excitability likely contribute to the temporal evolution of mitral cell activity. It is likely that the slow phasic cholinergic input from the basal forebrain functions to synchronize large groups of granule cells, as they would all be receiving a similarly patterned cholinergic input, revealing ICAN mediated ADPs in these cells, which may enable these cells to take advantage of these intrinsic currents and synchronize at faster time scales. Blanes cells, as with their long axons that end in large axon plexuses, are likely to act in a way to disinhibit neighboring glomerular modules of mitral cells by providing phasic inhibitory input to groups of
143 granule cells. In essence, this increase in intrinsic granule cell excitability would be amplifying in both duration, and time course, the “inhibitory regimes” in the odor evoked mitral cell firing pattern. While, the disinhibitory effect that Blanes cells have on mitral cells would amplify the excitatory portions of the temporally evolving activity.
Testing the precise contribution of Blanes cells to odor evoked mitral cell activity will be challenging without a genetic method for labeling or selectively knocking out these neurons. Identifying the role of Blanes cells in modulating olfactory bulb output, however, can be approached computationally. Individual neurons can be created using a standardized two compartment model, with contributions of simulated potassium channels and sodium channels that generate similar firing properties and making them mimic the main subclasses of olfactory bulb neurons—the mitral cells, granule cells, and
Blanes cells. Simple olfactory bulb models built up from modules involving mitral cells sending primary dendrites to the same glomeruli, and the granule cells that laterally connect these modules, can be generated relatively easily. Then, the model would gain one additional level of complexity by adding low numbers of artificial Blanes cells to the circuit, and giving these Blanes cells diverse excitatory synapses from mitral cells, and then giving these Blanes cells inhibitory synapses onto columns of granule cells. To investigate the contribution of Blanes cells to the function of this circuit, simply record the activity of the mitral cells in this circuit with, and without active Blanes cells.
Strategies to determine the contributions of these two inputs onto granule cells in intact and behaving animals
144
To test the effect the cholinergic input to granule cells has for olfactory bulb output, one
could first start by training rats to perform a delayed response odor identification task.
Then, one could place electrodes into the olfactory bulb in such a way that it would be
possible to monitor local field potential oscillations from mitral cells, as well as mitral
cell and granule cell unit recordings. It would then be possible to correlate the temporally
evolving mitral cell activity with olfactory bulb oscillations and granule cell activity. One
could then compare these results to rats that have had their cholinergic basal forebrain lesioned, and then determine the effect of cholinergic input to granule cells in intact rats,
and how this input effects mitral cell output as well as olfactory bulb oscillations. It
would be possible to test the contribution of Blanes cells in a similar manner. However, instead of lesioning the cholinergic basal forebrain, the synapse that Blanes cells make onto granule cells could be negated by the injection of a GABA antagonist, gabazine, into the granule cell layer of the olfactory bulb. Because Blanes cells do not appear to send axons out of the granule cell layer, these gabazine injections should prevent the Blanes
cells from modulating granule cell activity over large stretches of the olfactory bulb. In
order to determine the effect of Blanes cell input onto granule cells in intact rats, and how
this input effects mitral cell output as well as olfactory bulb oscillations, one could
compare the results of these recordings before, and after, gabazine administration.
Future Directions
145 A number of testable predictions can also be generated from the main conclusions of this
thesis. The complete description of how Blanes cells fit into the canonical olfactory bulb
circuit is currently incomplete. Blanes cells receive barrages of EPSPs from glomerular-
layer focal stimulation, as well as stimulus-evoked EPSPs from granule cell layer focal stimulation (see chapter 2). Additionally, stimulus trains in the granule cell layer can generate persistent IPSPs onto granule cells. While my work implies that Blanes cells receive input from mitral cells, presumably from mitral cell axon collaterals, this has yet to be shown directly. Paired recordings between mitral cells and Blanes cells could establish the source of the excitatory input onto Blanes cells. Also, paired recordings between Blanes cells and granule cells could show that inputs from a single Blanes cell are sufficient to prevent action potential generation in granule cells. Because of the large
Blanes cell axon plexus (see chapter 2), it is likely that dozens of granule cells receive
synaptic input from a single Blanes cell. One of the conclusions from this thesis is that
the periodic inhibitory input from Blanes cells could synchronize groups of granule cells.
One could test this by making synaptically coupled paired recordings between a Blanes cell and two (or more) postsynaptic granule cells, and then activating granule cells with simulated (or stimulus evoked) EPSPs, and measuring action potential synchrony with and without Blanes cell mediated persistent IPSPs. Additionally, one could test this hypothesis by bulk loading a Ca dye in granule cells, and then noting the change in the dynamics of Ca fluorescence of granule cells mediated by, or in response to, single
Blanes cells activation. Another interesting aspect to this experiment would be to monitor granule cell activity in response to glomerulus stimulation, and then measure the
146 difference of that population activity when current-clamped Blanes cells fire persistent action potentials in conjunction with glomerulus stimulation.
In this thesis two different concentrations of cholinergic agonists were used to determine the effect of activating these receptors on granule cell intrinsic properties. To determine the precise amount of acetylcholine released in the olfactory bulb when awake, beheaving animals engage in delayed response tasks (such as the one illustrated in Ravel et al.,
1994), one could insert probes or electrodes into the olfactory bulb directly, and measure acetylcholine concentrations, as well as changes in concentrations over time, using in vivo microdialysis and in vivo voltammetry (Laplante et al., 2004).
When rats are investigating odors during odorant discrimination tasks, a preferred latency develops between hippocampal theta rhythm, and sniffing frequency (Macrides et al.,
1982). There is strong temporal correlation between other sensory-motor actions and limbic theta rhythm (Forbes and Macrides, 1984). One possible explanation for these temporal correlations is that the cholinergic basal forebrain, with its periodic input to the hippocampus, olfactory bulb, and other brain regions in the limbic system and in the neocortex, is rhythmically modulating these brain regions. The inputs these brain areas receive are not fast enough to enable the phase-aligned activity these regions have during tasks that require attentional regulation. Instead, it is likely that the input from the cholinergic basal forebrain enables these structures to self-synchronize. To test the impact of the cholinergic basal forebrain in phase-aligning the hippocampus and the olfactory bulb during odorant discrimination tasks, one could implant field electrodes in the
147 olfactory bulb and hippocampus and monitor extracellularly the population activity in
these regions. Then, one could measure the preferred phase latency in these control rats,
and in other rats that have received selective cholinergic basal forebrain lesions
(generated by IG-saporin injections to the horizontal limb of the diagonal band and
medial septum).
Lastly, granule cells that receive M1 AChR activation are more excitable and respond to
square depolarizing steps and simulated EPSPs with increased action potential firing (see chapter 3). As a result, it is likely that granule cells that receive cholinergic input would increase the degree of lateral inhibition between mitral cells. To test this hypothesis, one
could perform paired mitral cell recordings in a manner similar to that described by
Isaacson and Strowbridge (1998), and measure the degree of lateral inhibition before and
during granule cell M1 AChR activation.
148
Figure 4-1
149 Figure 4-1: A schematic showing the olfactory bulb circuit. ON indicates olfactory nerve inputs, MT indicates a mitral/tufted cell, GC indicates a granule cell, and LOT indicates the projections to the lateral olfactory tract. (A) Stylized responses in mitral/tufted cells and a granule cell to olfactory nerve input. (B) The same circuit as in
(A), but the granule cell is receiving cholinergic input.
150
Chapter 5
Bibliography
151 Adams PR, and Brown DA (1982), “Synaptic inhibition of the m-current: slow excitatory
post-synaptic potential mechanism in bullfrog sympathetic neurons”, J Physiol 332:263-
272.
Adams PR, Brown DA, Constanti A (1982), “M-currents and other potassium currents in
bullfrog symapathetic neurons”, J Physiol 330: 537-572.
Adrian, E. D. (1950). "The electrical activity of the mammalian olfactory bulb."
Electroencephalogr Clin Neurophysiol 2: 377-388.
Aksay E., Baker R., Seung H.S., Tank D.W. (2000). "Anatomy and discharge properties
of pre-motor neurons in the goldfish medulla that have eye-position signals during
fixations." J Neurophysiol 84(2): 1035-49.
Alonso J.R., Brinon J.G., Crespo C., Bravo I.G., Arevalo R., Aijon J. (2001). "Chemical organization of the macaque monkey olfactory bulb: II. Calretinin, calbindin D-28k, parvalbumin, and neurocalcin immunoreactivity." J Comp Neurol 432(3): 389-407.
Andrade R and Nicoll R (1987), “Pharmacologically distinct actions of serotonin on single pyramidal neurons of the rat hippocampus recorded in vitro”, J Physiol 394: 99-
124.
152 Arnold H.M., Burk J.A., Hodgson E.M., Sarter M., Bruno J.P. (2002). "Differential
cortical acetylcholine release in rats performing a sustained attention task versus
behavioral control tasks that do not explicitly tax attention." Neuroscience 114(2): 451-
60.
Aungst J.L., Heyward P.M., Puche A.C., Karnup S.V., Hayar A., Szabo G., Shipley M.T.
(2003), Centre-surround inhibition among olfactory bulb glomeruli, Nature 426, pp. 623–
629.
Balu R., Larimer P., Strowbridge B.W. (2004), Phasic stimuli evoke precisely timed
spikes in intermittently discharging mitral cells, J. Neurophysiol. 92, pp. 743–753.
Barrett JN, Magleby KL, and Pallotta BS (1982), “Properties of single calcium-activated
potassium channels in cultured rat muscle”, J Physiol 331: 211-230
Blanes T. (1898), “Sobre algunos puntos dudoses de la estructura del bulbo olfatorio”,
Revta Trimest Microgr. 3, pp. 99–127.
Bourque C.W. (1986), “Calcium-dependent spike after-current induces burst firing in
magnocellular neurosecretory cells”, Neurosci. Lett. 70, pp. 204–209.
Brashear H.R., Zaborszky L., Heimer L. (1986), "Distribution of GABAergic and
cholinergic neurons in the rat diagonal band", Neuroscience 17(2): 439-51.
153
Bronshtein A.A. and Minor A.V. (1977), “Regeneration of olfactory flagella and restoration of the electroolfactogram following application of Triton X-100 to the olfactory mucosa of frogs”, Tsitologiia 19(1): 33-9.
Brown DA, and Griffith WH (1983), “Calcium-activated outward current in voltage- clamped hippocampal neurons of the guinea-pig”, J Physiol 337:287-301.
Brumberg J.C., Nowak L.G., McCormick D.A. (2000), "Ionic mechanisms underlying repetitive high-frequency burst firing in supragranular cortical neurons", J Neurosci
20(13): 4829-43.
Buck L., and Axel R. (1991), “A novel multigene family may encode odorant receptors: a
molecular basis for odor recognition”, Cell 65(1): 175-187.
Buck L. (1996), “Information coding in the vertebrate olfactory system”, Annu Rev
Neurosci 19, pp. 517-544
Butt A.E., Noble M.M., Rogers J.L., Rea T.E. (2002), "Impairments in negative
patterning, but not simple discrimination learning, in rats with 192 IgG-saporin lesions of
the nucleus basalis magnocellularis", Behav Neurosci 116(2): 241-55.
154 Cang and Isaacson, (2003). “In vivo whole-cell recording of odor-evoked synaptic
transmission in the rat olfactory bulb”, J. Neurosci. 23, pp. 4108–4116.
Castillo P.E., Carleton A., Vincent J.D., Lledo P.M. (1999). "Multiple and opposing roles
of cholinergic transmission in the main olfactory bulb." J Neurosci 19(21): 9180-91.
Constanti A, and Sim JA (1987), “Calcium-dependent potassium conductance in guinea-
pig olfactory cortex neurons in vitro”, J Physiol 387:173-194.
Cortes, R. and J. M. Palacios (1986). "Muscarinic cholinergic receptor subtypes in the rat
brain. I. Quantitative autoradiographic studies." Brain Res 362(2): 227-38.
Cox DH, Cui J, Aldrich RW (1997), “Allosteric gating of a large conductance CA-
activated K+ channel”, J Gen Physiol 110:257-281.
Crespo C., Blasco-Ibanez J.M., Brinon J.G., Alonso J.R., Dominguez M.I., Martinez-
Guijarro F.J. (2000), "Subcellular localization of M2 muscarinic receptors in GABAergic
interneurons of the olfactory bulb." Eur J Neurosci 12(11): 3963-74.
Davis B.J., Burd G.D., Macrides F. (1982). "Localization of methionine-enkephalin,
substance P, and somatostatin immunoreactivities in the main olfactory bulb of the hamster." J Comp Neurol 204(4): 377-83.
155 Deco, G. and E. T. Rolls (2003). "Attention and working memory: a dynamical model of
neuronal activity in the prefrontal cortex." Eur J Neurosci 18(8): 2374-90.
Desmaisons D, Vincent JD, Lledo PM, “Control of action potential timing by intrinsic
subthreshold oscillations in olfactory bulb output neurons”, J Neurosci 19(24): 10727-
10737.
Douglas R.J. and Martin K.A., (1991), “A functional microcircuit for cat visual cortex”,
J. Physiol. 440, pp. 735–769.
Doty, R.L., Bagla R., and Kim, N. (1999), “Physostigmine enhances performance on an odor mixture discrimination test. Physiol. Behav. 65:801-804.
Durstewitz D, Seamans J, Sejnowski, T (2000), “Neurocomputational models of working memory”, Nature Neuroscience, 3:1184-1191
Dutar, P. and R. A. Nicoll (1988). "Classification of muscarinic responses in hippocampus in terms of receptor subtypes and second-messenger systems: electrophysiological studies in vitro." J Neurosci 8(11): 4214-24.
Egger V., Svoboda K., Mainen Z.F. (2003), “Mechanisms of lateral inhibition in the olfactory bulb: efficiency and modulation of spike-evoked calcium influx into granule cells”, J. Neurosci. 23 (2003), pp. 7551–7558.
156
Egorov A.V., Hamam B.N., Fransen E., Hasselmo M.E., Alonso A.A. (2002), “Graded
persistent activity in entorhinal cortex neurons”, Nature 420, pp. 173–178.
Ellis-Davies G.C. and Kaplin J.H., (1994), “Nitrophenyl-EGTA, a photolabile chelator
that selectively binds Ca2+ with high affinity and releases it rapidly upon photolysis”,
Proc. Natl. Acad. Sci. USA 91 (1994), pp. 187–191.
Everitt, B. J. and Robbins T. W. (1997), "Central cholinergic systems and cognition."
Annu Rev Psychol 48: 649-84.
Ezeh P, Wellis D, Scott J (1993), “Organization of inhibition in the rat olfactory bulb external plexiform layer”, J Neurophys 70: 263-274.
Freund, T. F. and Antal M. (1988), "GABA-containing neurons in the septum control
inhibitory interneurons in the hippocampus." Nature 336(6195): 170-3.
Freund, T. F. and Meskenaite V. (1992), "gamma-Aminobutyric acid-containing basal
forebrain neurons innervate inhibitory interneurons in the neocortex." Proc Natl Acad Sci
U S A 89(2): 738-42.
157 Friedman D. and Strowbridge B.W. (2003), “Both electrical and chemical synapses mediate fast network oscillations in the olfactory bulb”, J. Neurophysiol. 89 (2003), pp.
2601–2610.
Friedrich, R. W. and Laurent G. (2001), "Dynamic optimization of odor representations by slow temporal patterning of mitral cell activity." Science 291(5505): 889-94.
Funahashi S, Bruce C, Goldman-Rakic PS (1989), “Mnemonic coding of visual space in the monkey’s dorsolateral prefrontal cortex”, J Neurophys 61(2):331-349.
Fuster, J. M. and Alexander G.E. (1971). "Neuron activity related to short-term memory."
Science 173(997): 652-4.
Ghamari-Langroudi m. and Bourque C.W. (2002), ”Flufenamic acid blocks depolarizing afterpotentials and phasic firing in rat supraoptic neurons”, J. Physiol. 545, pp. 537–542.
Ghatpande A, Sivaraaman K, Vijayaraghavan S (2006), “Store calcium mediates cholinergic effects on mIPSCs in the rat main olfactory bulb”, J Neurophysiology 95:
1345-1355/
Goldman-Rakic, P. S. (1995). "Cellular basis of working memory." Neuron 14(3): 477-
85.
158 Gracia-Llanes F.J., Crespo C., Blasco-Ibanez J.M., Marques-Mari A.I. and Martinez-
Guijarro F.J. (2003), “VIP-containing deep short-axon cells of the olfactory bulb
innervate interneurons different from granule cells”, Eur. J. Neurosci. 18, pp. 1751–1763.
Haas HL and Koneerth A (1983), “Histamine and noradrenaline decrease calcium-
activated potassium conductance in hippocampal pyramidal cells”, Nature 302(5907):
432-434.
Haj-Dahmane S. and Andrade R., (1998), “Ionic mechanism of the slow
afterdepolarization induced by muscarinic receptor activation in rat prefrontal cortex”, J.
Neurophysiol. 80, pp. 1197–1210.
Haj-Dahmane S. and Andrade R., (1999), “Muscarinic receptors regulate two different calcium-dependent non-selective cation currents in rat prefrontal cortex”, Eur. J.
Neurosci. 11, pp. 1973–1980.
Halabisky B. and Strowbridge B.W., (2003), “Gamma-frequency excitatory input to granule cells facilitates dendrodendritic inhibition in the rat olfactory bulb”, J.
Neurophysiol. 90, pp. 644–654.
159 Hall B.J. and Delaney K.R. (2002), “Contribution of a calcium-activated non-specific
conductance to NMDA receptor-mediated synaptic potentials in granule cells of the frog olfactory bulb”, J. Physiol. 543, pp. 819–834.
Hamilton S.E., Loose M.D., Qi M., Levey A.I., Hille B., McKnight G.S., Idzerda R.L.,
Nathanson N.M. (1997), "Disruption of the M1 receptor gene ablates muscarinic
receptor-dependent M current regulation and seizure activity in mice." Proc Natl Acad
Sci U S A 94(24): 13311-6.
Hille, 2001 Hille 2001, Potassium Channels and Chloride Channels. In: Hille, Ion
Channels and Excitable Membranes, Sinauer associates, Sunderland MA. (2001), pp.
131-168.
Horikawa K. and Armstrong W.E. (1988), “A versatile means of intracellular labeling:
injection of biocytin and its detection with avidin conjugates”, J. Neurosci. Methods 25,
pp. 1–11.
Hunt, S. and Schmidt J. (1978), "Some observations on the binding patterns of alpha-
bungarotoxin in the central nervous system of the rat." Brain Res 157(2): 213-32.
Inokuchi A., Restrepo J.P., Snow J.B. Jr. (1987). "Effect of stimulation of the horizontal
limb of the diagonal band on rat olfactory bulb neuronal activity." Am J Otolaryngol
8(4): 205-10.
160
Isaacson J.S. and Strowbridge B.W., (1998), “Olfactory reciprocal synapses: dendritic signaling in the CNS”, Neuron 20, pp. 749–761.
Jahr C.E. and Nicoll R.A. (1980), “Dendrodendritic inhibition: demonstration with intracellular recording”, Science 207, pp. 1473–1475.
Johnston D., Magee J. C., Colbert C. M., Christie B. R. (1996), “Active properties of neuronal dendrites”, Annu Rev Neurosci.19:165-86.
Johnston and Amaral, (1998), Hippocampus. Synaptic Organization of the Brain. G. M.
Shepherd. New York, Oxford University Press: 417-458.
Jones, SW (1985), “Muscarinic and peptidergic excitation of bull-frog sympathetic neurons”, J Physiol 366: 63-87.
Juliano S.L., Ma W., Eslin D. (1991), "Cholinergic depletion prevents expansion of topographic maps in somatosensory cortex." Proc Natl Acad Sci U S A 88(3): 780-4.
Kakuta S., Oda S., Takayanagi M., Kishi K. (1998), “Parvalbumin immunoreactive neurons in the main olfactory bulb of the house musk shrew, Suncus murinus”, Brain
Behav. Evol. 52, pp. 285–291.
161 Kasa P., Hlavati I., Dobo E., Wolff A., Joo F., Wolff J.R. (1995), "Synaptic and non- synaptic cholinergic innervation of the various types of neurons in the main olfactory bulb of adult rat: immunocytochemistry of choline acetyltransferase." Neuroscience
67(3): 667-77.
Kay, L.M. (2003), "Two species of gamma oscillations in the olfactory bulb: dependence on behavioral state and synaptic interactions." J Integr Neurosci 2(1): 31-44.
Kay, L. M. and Laurent G. (1999), "Odor- and context-dependent modulation of mitral cell activity in behaving rats." Nat Neurosci 2(11): 1003-9.
Kubota, K. and Niki H.(1971), "Prefrontal cortical unit activity and delayed alternation performance in monkeys." J Neurophysiol 34(3): 337-47.
Kunze W.A., Shafton A.D., Kem R.E., McKenzie J.S. (1992). "Intracellular responses of olfactory bulb granule cells to stimulating the horizontal diagonal band nucleus."
Neuroscience 48(2): 363-9.
Lagier S., Carleton A., Lledo P.M. (2004). "Interplay between local GABAergic interneurons and relay neurons generates gamma oscillations in the rat olfactory bulb." J
Neurosci 24(18): 4382-92.
162 Lancaster B, Adams PR (1986). “Calcium-dependent current generating the afterhyperpolarization of hippocampal neurons”, J Neurophys 55(6):1268-1282.
Lancaster B, Nicoll R, Perkel, DJ (1991), “Calcium activates two types of potassium
channels in rat hippocampal neurons in culture”, J Neurosci 11(1): 23-30.
Laplante F., Stevenson C.W., Gratton A., Srivastava L.K., Quirion R. (2004), “Effects of
neonatal ventral hippocampal lesion in rats on stress-induced acetylcholine release in the
prefrontal cortex”, J Neurochem 91: 1473-1482.
Laurent G., Wehr W., Davidowitz H. (1996), “Temporal representations of odors in an
olfactory network”, J. Neurosci 16(12): 3837-3847.
Laurent, G., Stopfer M., et al. (2001). "Odor encoding as an active, dynamical process:
experiments, computation, and theory." Annu Rev Neurosci 24: 263-97.
Le Jeune H, Aubert I, Jourdan F, Quirion R (1995), “Comparative laminar distribution of
various autoradiographic cholinergic markers in adult rat main olfactory bulb”, J Chem
Neuroanatomy 9:99-112.
Le Jeune H, Aubert I, Jourdan F, Quirion R (1996), “Developmental profiles of various
cholinergic markers in the rat main olfactory bulb using quantitative autoradiography”, J
Comp Neurol 373: 433-450.
163
Li Z. and Hatton G.I., (1997), “Reduced outward K+ conductances generate depolarizing after-potentials in rat supraoptic nucleus neurones", J. Physiol. 505, pp. 95–106.
Lin D.Y., Shea A.D., Katz L.C. (2006), “Representation of natural stimuli in the rodent main olfactory bulb”, Neuron 50, pp. 937-949
Linster C, Garcia PA, Hasselmo ME, Baxter MG (2001), “Selective loss of cholinergic neurons projecting to the olfactory system increases perceptual generalization between similar, but not dissimilar, odorants.
Lopez-Mascaraque L., De Carlos J.A. and Valverde F.(1986), “Structure of the olfactory bulb of the hedgehog (Erinaceus europaeus): description of cell types in the granular layer”, J. Comp. Neurol. 253, pp. 135–152.
Luskin, M. B. and Price J. L. (1982). "The distribution of axon collaterals from the olfactory bulb and the nucleus of the horizontal limb of the diagonal band to the olfactory cortex, demonstrated by double retrograde labeling techniques." J Comp Neurol 209(3):
249-63.
Macrides F., Davis B.J., Youngs W.M., Nadi N.S., Margolis F.L. (1981). "Cholinergic and catecholaminergic afferents to the olfactory bulb in the hamster: a neuroanatomical, biochemical, and histochemical investigation." J Comp Neurol 203(3): 495-514.
164
Macrides F., Eichenbaum H.B., Forbes W.B. (1982). "Temporal relationship between
sniffing and the limbic theta rhythm during odor discrimination reversal learning." J
Neurosci 2(12): 1705-17.
Madison DV, Lancaster B, and Nicoll R (1987), “Voltage clamp analysis of cholinergic action in the hippocampus”, J Neurosci 7(3):733-741
Malnic B, Hirono J, Sato T, Buck LB, (1999) "Combinatorial receptor codes for odors"
Cell 96(5): 713-23.
Margrie T.W. and Schaefer A.T. (2003), “Theta oscillation coupled spike latencies yield computational vigour in a mammalian sensory system”, J. Physiol. 546, pp. 363–374.
Marty A (1981), “Ca-dependent K channels with large unitary conductance in chromaffin cell membranes”, Nature 291(5815):497-500.
McCormick D.A. and Bal T., (1997), “Sleep and arousal: thalamocortical mechanisms”,
Annu. Rev. Neurosci. 20 (1997), pp. 185–215.
McCormick D.A., Shu Y., Hasenstaub A., Sanchez-Vives M., Badoual M., Bal T.,
“Persistent cortical activity: mechanisms of generation and effects on neuronal excitability”, Cereb. Cortex 13 (2003), pp. 1219–1231.
165
Mercado E., Bao S., Orduna I., Gluck M.A., Merzenich M.M. (2001). "Basal forebrain
stimulation changes cortical sensitivities to complex sound." Neuroreport 12(10): 2283-7.
Miles R, Wong RKS (1986), “Excitatory synaptic interactions between CA3 neurons in
the guinea-pig hippocampus”, J Physiol 373:397-418
Mori K., Fujita S.C., Imamura K., Obata K. (1985). "Immunohistochemical study of
subclasses of olfactory nerve fibers and their projections to the olfactory bulb in the
rabbit." J Comp Neurol 242(2): 214-29.
Mori, K. and Y. Yoshihara (1995). "Molecular recognition and olfactory processing in
the mammalian olfactory system." Prog Neurobiol 45(6): 585-619.
Mori K., Nagao H. and Yoshihara Y. (1999), “The olfactory bulb: coding and processing
of odor molecule information”, Science 286, pp. 711–715.
Morris RG, Garrud P, Rawlins JN, O’Keefe J (1982), “Place navigation impaired in rats
with hippocampal lesions”, Nature 297(5868):681-683.
Muir J.L., Everitt B.J., Robbins T.W., “AMPA-induced excitotoxic lesions of the basal forebrain: a significant role for the cortical cholinergic system in attentional function”, J
Neurosci 14(4): 2313-2326
166
Nakajima T., Okamura M., Ogawa K., Taniguchi K. (1996), “Immunohistochemical and enzyme histochemical characteristics of short axon cells in the olfactory bulb of the golden hamster”, J. Vet. Med. Sci. 58, pp. 903–908.
Ohm T.G., Muller H., Ulfig N. Braak E.(1990), “Glutamic-acid-decarboxylase-and parvalbumin-like-immunoreactive structures in the olfactory bulb of the human adult”, J.
Comp. Neurol. 291, pp. 1–8.
Ottoson D. (1956), “Analysis of the electrical activity of the olfactory epithelium”, Acta
Physiologica Scandinavica 35 (Suppl 122):1–83.
Pace U, Hanski E, Salomon Y, Lancet D (1985), “Odorant-sensitive adenylate cyclase may mediate olfactory receptorion”. Nature 316 (6025):255-258
Park YB (1994), “Ion selectivity and gating of small conductance Ca2+-activated K+ channels in cultured rat adrenal chromaffin cells”, J Physiol 481(3): 555-570.
Partridge L.D., and Valenzuela C.F., (2000), “Block of hippocampal CAN channels by flufenamate”, Brain Res. 867, pp. 143–148.
Partridge L.D., Muller T.H. and Swandulla D. (1994), “Calcium-activated non-selective channels in the nervous system”, Brain Res. Brain Res. Rev. 19, pp. 319–325.
167
Pennefather P, Lancaster B, Adams PR, and Nicoll R (1985), “Two distinct Ca-dependent
K currents in bullfrog sympathetic ganglion cells”, PNAS 82:3040-3044.
Penschuck S., Chen-Bee C.H., Prakash N., Frostig R.D. (2002). "In vivo modulation of a
cortical functional sensory representation shortly after topical cholinergic agent
application." J Comp Neurol 452(1): 38-50.
Pressler RT and Strowbridge BW (2006), “Blanes cells mediate persistent feedforward
inhibition onto granule cells in the olfactory bulb”, Neuron 49(6):889-904.
Price, J. L. (1969). "The origin of the centrifugal fibres to the olfactory bulb." Brain Res
14(2): 542-5.
Price, J. L. and Powell T. P. (1970a), "The morphology of the granule cells of the
olfactory bulb." J Cell Sci 7(1): 91-123.
Price J.L. and Powell T.P. (1970b), “The synaptology of the granule cells of the olfactory
bulb”, J. Cell Sci. 7, pp. 125–155.
Price, J. L. and Powell T. P. (1970c), "An electron-microscopic study of the termination
of the afferent fibres to the olfactory bulb from the cerebral hemisphere." J Cell Sci 7(1):
157-87.
168
Price J.L. and Powell T.P. (1970d), “The mitral and short axon cells of the olfactory bulb”, J. Cell Sci. 7, pp. 631–651.
Rall W. and Shepherd G.M. (1968), “Theoretical reconstruction of field potentials and dendrodendritic synaptic interactions in olfactory bulb”, J. Neurophysiol. 31, pp. 884–
915.
Rall W., Shepherd G.M., Reese T.S. and Brightman M.W. (1966), “Dendrodendritic synaptic pathway for inhibition in the olfactory bulb”, Exp. Neurol. 14, pp. 44–56.
Ramon y Cajal, S (1995), “Histology of the nervous system, translated by Swanson, N. and Swanson, L.”, Oxford University Press.
Ravel N., Elaagouby A., Gervais R. (1994), "Scopolamine injection into the olfactory bulb impairs short-term olfactory memory in rats." Behav Neurosci 108(2): 317-24.
Ravel N., Vigouroux M., Elaagouby A., Gervais R. (1992). "Scopolamine impairs delayed matching in an olfactory task in rats." Psychopharmacology (Berl) 109(4): 439-
43.
Ressler K.J., Sullivan S.L., Buck L.B. (1993). "A zonal organization of odorant receptor gene expression in the olfactory epithelium." Cell 73(3): 597-609.
169
Rhodes P (2006), “The properties and implications of NMDA spikes in neocortical
pyramidal cells”, J Neurosci. 26(25):6704-6715.
Ringer S. (1883), “A further contribution regarding the influence of the different
constituents of the blood on the contraction of the heart”, J Physiol, 4: 29-42.
Roper P., Callaway J., Shevchenko T., Teruyama R. and Armstrong W. (2003), “AHP's,
HAP's and DAP's: how potassium currents regulate the excitability of rat supraoptic
neurones”, J. Comput. Neurosci. 15, pp. 367–389.
Roper P., Callaway J. and Armstrong W. (2004), “Burst initiation and termination in phasic vasopressin cells of the rat supraoptic nucleus: a combined mathematical, electrical, and calcium fluorescence study”, J. Neurosci. 24, pp. 4818–4831.
Russo RE and Hounsgaard J (1999). "Dynamics of intrinsic electrophysiological
properties in spinal cord neurones." Prog Biophys Mol Biol 72(4): 329-65.
Sah P (1995), “Properties of channels mediating the apamin-insensitive
afterhyperpolarization in vagal motorneurons”, J Neurophys 74(4):1772-1776.
170 Sah P, and McLachlan E (1991), “Ca2+-activated K+ currents underlying the
afterhyperpolarization in guinea pig vagal neurons: A role for Ca2+-activated Ca2+ release”, Neuron 7(2): 257-264.
Sanchez-Vives M.V. and McCormick D.A. (2000), “Cellular and network mechanisms of rhythmic recurrent activity in neocortex”, Nat. Neurosci. 3, pp. 1027–1034.
Sarter, M. and Bruno J. P. (2000). "Cortical cholinergic inputs mediating arousal, attentional processing and dreaming: differential afferent regulation of the basal forebrain by telencephalic and brainstem afferents." Neuroscience 95(4): 933-52.
Sarter M., Hasselmo M.E., Bruno J.P., Givens B. (2005). "Unraveling the attentional functions of cortical cholinergic inputs: interactions between signal-driven and cognitive modulation of signal detection." Brain Res Brain Res Rev 48(1): 98-111.
Sarter M, Gehring W., Kozak R. (2006), “More attention must be paid: The neurobiology of attentional effort.” Brain Res Brain Res Rev 51: 156-160
Schneider S.P. and Macrides F. (1978), “Laminar distributions of internuerons in the main olfactory bulb of the adult hamster”, Brain Res. Bull. 3, pp. 73–82.
Schreiber M, Salkoff, L (1997), “A novel calcium-sensing domain in the BK channel”,
Biophysical Journal 73:1355-1363.
171
Schoppa N.E., Kinzie J.M., Sahara Y., Segerson T.P. Westbrook G.L. (1998),
Dendrodendritic inhibition in the olfactory bulb is driven by NMDA receptors, J.
Neurosci. 18, pp. 6790–6802.
Scott J.W., McDonald J.K. and Pemberton J.L. (1987), “Short axon cells of the rat olfactory bulb display NADPH-diaphorase activity, neuropeptide Y-like immunoreactivity, and somatostatin-like immunoreactivity”, J. Comp. Neurol. 260, pp.
378–391.
Shepherd G.M. and Greer C.A. (1998), Olfactory Bulb. In: G.M. Shepherd, Editor,
Synaptic Organization of the Brain, Oxford University Press, New York, pp. 159–204.
Shipley MT and Adamek GD (1984), “The connections of the mouse olfactory bulb: a study using orthograde and retrograde transport of wheat germ agglutinin conjugated to horseradish peroxidase”, Brain Res. Bull. 12(6): 669-688.
Stepanyants A., Tamas G., Chklovskii D.B. (2004), “Class-specific features of neuronal wiring”, Neuron 43 (2004), pp. 251–259.
Schneider, S. P. and Macrides F. (1978). "Laminar distributions of internuerons in the main olfactory bulb of the adult hamster." Brain Res Bull 3(1): 73-82.
172 Scott J.W., McDonald J.K., Pemberton J.L. (1987). "Short axon cells of the rat olfactory
bulb display NADPH-diaphorase activity, neuropeptide Y-like immunoreactivity, and
somatostatin-like immunoreactivity." J Comp Neurol 260(3): 378-91.
Shepherd, G. M. and Greer, C.A.(1998). Olfactory Bulb. Synaptic Organization of the
Brain. G. M. Shepherd. New York, Oxford University Press: 159-204.
Stepanyants, A., G. Tamas, et al. (2004). "Class-specific features of neuronal wiring."
Neuron 43(2): 251-9.
Sterling P., and Demb J.B. (2004), Retina. Synaptic Organization of the Brain. G.M.
Shepherd. New York, Oxford University Press: 217-269.
Stopfer M., Bhagavan S., Smith B.H., Laurent G. (1997). "Impaired odour discrimination
on desynchronization of odour-encoding neural assemblies." Nature 390(6655): 70-4.
Storm JF (1989), “An after-hyperpolarization of medium duration in rat hippocampal
pyramidal cells”, J Physiol 409: 171-190.
Storm, J. F. (1990). "Potassium currents in hippocampal pyramidal cells." Prog Brain Res
83: 161-87.
173 Vassar R., Chao S.K., Sitcheran R., Nunez J.M., Vosshall L.B., Axel R. (1994).
"Topographic organization of sensory projections to the olfactory bulb." Cell 79(6): 981-
91.
Vassar R., Ngai J., Axel R. (1993). "Spatial segregation of odorant receptor expression in
the mammalian olfactory epithelium." Cell 74(2): 309-18.
Wilson D.A., Fletcher M.L., Sullivan R.M. (2004). "Acetylcholine and olfactory
perceptual learning." Learn Mem 11(1): 28-34.
Zaborszky L., Carlsen J., Brashear H.R., Heimer L. (1986). "Cholinergic and GABAergic
afferents to the olfactory bulb in the rat with special emphasis on the projection neurons
in the nucleus of the horizontal limb of the diagonal band." J Comp Neurol 243(4): 488-
509.
Zelles T, Boyd J, Hardy A, Delaney K (2006), “Branch-specific Ca2+ influx from Na+- dependent dendritic spikes in olfactory granule cells”, J Neurosci 26(1):30-40.
Zhao H., Ivic L., Otaki J.M., Hashimoto M., Mikoshiba K., Firestein S. (1998).
"Functional expression of a mammalian odorant receptor." Science 279(5348): 237-42.
174