sign in sign up
<<
Home , Caudate nucleus, Claustrum, Islands of Calleja, Nucleus accumbens, Olfactory bulb, Olfactory receptor

INVESTIGATION OF THE ELECTROPHYSIOLOGICAL PROPERTIES

OF THE MAJOR CELL TYPES

IN THE RAT

By

ELIZABETH CHEN CHIANG

Submitted in partial fulfillment of the requirement

For the degree Doctor of Philosophy

Thesis Advisor: Dr. Ben W. Strowbridge

Department of

CASE WESTERN RESERVE UNIVERSITY

January, 2008

Case Western Reserve University School of Graduate Studies

We hereby approve the thesis/dissertation of

Elizabeth Chen Chiang candidate for the PHD degree*.

signed Jerry Silver Chair of the Committee

Ben Strowbridge

Gary Landreth

Ruth Siegel

Lynn Landmesser

October 5, 2007

*We also certify that written approval has been obtained for any proprietary material contained within.

ii

To my parents, Dalen and Ru-fang Chiang, for teaching me the value of and education from a young age

and always supporting my academic pursuits

iii Table of Contents

Title Page…………………………………………………………………………………..i

Committee Signatures……………………………………………………………………..ii

Dedication………………………………………………………………………………...iii

Table of Contents…………………………………………………………………………iv

List of Tables…………………………………………………………………………...... vi

List of Figures…………………………………………………………………………....vii

Acknowledgements……………………………………………………………………...... x

List of Abbreviations…………………………………………………………………..…xi

Abstract……………………………………………………………………………….…xiv

Chapter 1: Introduction……………………………………………………………………1

Basic description of the …………………………………………2

Structure of the olfactory tubercle………………………………………………...4

Potential circuitry of the olfactory tubercle………………………………………7

The olfactory tubercle as part of the ventral ……………………………13

Chapter 2: Diversity of neural signals mediated by multiple burst-firing mechanisms in rat olfactory tubercle ……………………………………………………………36

Introduction………………………………………………………………………37

Methods…………………………………………………………………………..39

Results……………………………………………………………………………43

Discussion………………………………………………………………………..54

Chapter 3: Simulation of the principle currents of the nonregenerative bursting cell of the rat olfactory tubercle……………………………………………………………………102

Introduction……………………………………………………………………..103

Methods and Results……………………………………………………………105

iv Discussion………………………………………………………………………128

Chapter 4: Discussion…………………………………………………………………..177

Two mechanisms of generating burst firing……………………………………178

Role of in bursting cells…………………………………………….…179

Comparing the tubercle bursting cell to the hippocampal bursting cell……..…183

Function of bursting activity and plasticity………………………………….…185

The tubercle as part of the ventral striatum………………………………….…186

Circuitry of the olfactory tubercle……………...……………………………….187

The tubercle as a site of action for anti-psychotics and ……..191

The tubercle is a sensory, limbic, and motor structure…………………………194

Chapter 5: Future Directions…………………………………………………………...196

Circuitry of the olfactory tubercle……………………………………………....197

Role for bursting cells………………………………………………………..…199

Function of receptors and cocaine in the tubercle………………..….201

Conclusion…………………………………………………………………...…207

Chapter 6: Bibliography………………………………………………………………...213

v List of Tables

Table 2-1 Membrane properties of tubercle neurons…………………………………….61

Table 5-1 Summary of results from Luskin and Price (1983) regarding the olfactory tubercle…………………………………………………………………………………208

vi List of Figures

Figure 1-1 Schematic diagram showing the ventral structures contained in tubercle slices………………………………………………………………………….18

Figure 1-2 Nissl stain of corona slice of rat olfactory tubercle…………………………..20

Figure 1-3 Coronal slices of rat stained for proteins related to glutamate……………………………………………………………………..22

Figure 1-4 Regional distribution of VGLUT3 mRNA and protein……………………...24

Figure 1-5 Immunohistochemical localization of VGLUT3 on a coronal rat brain section taken at the level of the tubercle and the accumbens………………………...27

Figure 1-6 Coronal slice of the rat brained stained with acetylcholinesterase…………..29

Figure 1-7 Diagram of ChAT+ and GAD+ neurons in the tubercle……………………..32

Figure 1-8 Comparison of the distribution of D1 (A), D2, (B), and D3(C) mRNA in the rat brain……………………………………………………………………....34

Figure 2-1 Schematic diagram showing the ventral structures contained in tubercle slices……………………………………………………………….…………62

Figure 2-2 Graded responses in three types of olfactory tubercle neurons………………64

Figure 2-3 Morphology of regular spiking tubercle neurons…………………………….66

Figure 2-4 Morphology of intermittently discharging and bursting tubercle neurons…...68

Figure 2-5 Correlations of burst properties with stimulus strength……………………...70

Figure 2-6 Quantitative analysis of burst properties……………………………………..72

Figure 2-7 Bursting cell response to depolarizing ramp stimuli…………………………74

Figure 2-8 Cs increases firing during depolarizing steps in bursting cells………………76

Figure 2-9 Low Ca ACSF increases firing in bursting cells……………………………..78

Figure 2-10 TTX does not block depolarizing plateau potentials………………………..80

Figure 2-11 Reduction of input resistance during afterhyperpolarization of bursting cells…………………………………………………………………………..82

vii Figure 2-12 Comparison of response of regenerative and nonregenerative bursting cells…………………………………..…………………………………..…..84

Figure 2-13 Differences in burst properties between regenerative and nonregenerative bursting cells…………………………………………………………………86

Figure 2-14 Plateau potential evoked by graded steps…………………………………..88

Figure 2-15 Membrane potential of plateaus are constant………………………………90

Figure 2-16 Plateau potentials seen with brief stimuli………………………………..…92

Figure 2-17 Low Ca abolishes plateau potentials…………………………………….….94

Figure 2-18 Plasticity of burst responses in nonregenerative cells………………………96

Figure 2-19 Brief depolarizing steps suppresses later activity…………………………..98

Figure 2-20 Responses of tubercle neurons to -like periodic input………….…100

Figure 3-1 Morphology of bursting tubercle …………………………………...133

Figure 3-2 Figure 3-2: Montage of 2-photon images of a nonregenerative bursting neuron in the MFL………………………………………………………………….135

Figure 3-3 Schematic diagram of morphology of neuron model……………………….137

Figure 3-4 Correlations of burst properties with stimulus strength………………….…139

Figure 3-5 One mathematical model of IH with different distributions across model cell……………………………………………………………………….….141

Figure 3-6 Five different mathematical models of IH………………………………..…143

Figure 3-7 Comparison of three experimentally recorded cells with the simulation…...145

Figure 3-8 Comparison of the control simulation to the simulation after removal of IH.147

Figure 3-9 Different spatial distributions of T, L and N current in the model neuron…149

Figure 3-10 Variation in the distribution of T current with L and N current in the and ……………………………………………………………..…151

Figure 3-11 Variation in the magnitude of T current………………………………...... 153

Figure 3-12 Nonregenerative bursting cells at different membrane potentials…………155

Figure 3-13 Graded response of the simulation run in control conditions and run with the removal of calcium activated potassium current…………………………...157

viii Figure 3-14 Graded response of the simulation run in control conditions and run with the removal of M current…………………………………………………….…159

Figure 3-15 Effect of a low calcium environment on the model cell………………..…161

Figure 3-16 Response of model to simulated condition of the addition of TTX…….....163

Figure 3-17 Comparison of the effects of TTX on recorded cells vs. the simulated model cell…………………………………………………………………………..165

Figure 3-18 Response of model to TTX compared to TTX and Low Ca ACSF……….167

Figure 3-19 Response of model to brief stimuli……………………………………..…169

Figure 3-20 2 Inhibition of firing following a long stimulus…………………………..171

Figure 3-21 Inhibition of firing following a brief stimulus…………………………….173

Figure 3-22 Loss of suppression of firing with loss of CaK current…………………...175

Figure 5-1 Diagram of neurons of a brain slice of the olfactory tubercle………………209

Figure 5-2 Hypothesized circuit of the olfactory tubercle……………………………...211

ix Acknowledgements

I would like to thank my thesis advisor, Dr. Ben Strowbridge, for years of advice and

guidance. I would also like to thank the members of my thesis committee for their

helpful suggestions in my research. Lastly, I’d like to thank the past and present

members of the Strowbridge lab for their assistance, discussions, and companionship.

x List of Abbreviations

ACSF: artificial cerebrospinal fluid

AP:

4-AP: 4-aminopyridine, fast K channel blocker

BAPTA: O,O’-Bis(2-aminophenyl)ethyleneglycol-N,N,N’,N’-tetraacetic acid tetrapotassium salt cAMP: cyclic adenosine monophosphate

ChAT: choline acetyltransferase

DCL: Dense Cell Layer

D: dopamine

DNPI: differentiated-associated Na+ dependent inorganic phosphate cotransporter

EGTA: O,O’-Bis(2-aminoethyl)ethyleneglycol-N,N,N’,N’-tetraacetic acid, slow Ca chelator

EPSC: excitatory postsynaptic current

EPSP: excitatory postsynaptic potential

GABA: gamma-aminobutyric acid, the neurotransmitter in granule cells

GAD: decarboxylase

GC:

HRP: horse radish peroxidase

HVA: high voltage activated

IC:

IF: intermittently-firing

xi IPSC: inhibitory postsynaptic current

IPSP: inhibitory postsynaptic potential

ISI: inter-spike interval

LOT: lateral , of mitral cells

LTD: long term

LTP: long term potentiation

LVA: low voltage activated

MFL: multiform layer

ML: molecular layer mRNA: messenger ribonucleic acid

NBQX: 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo[f]quinoxaline-7-sulfonamide, AMPA antagonist

NMDA: N-methyl D-aspartate

NMDAR: NMDA receptor

OB:

6-OHDA: 6-hydroxydopamine

OT: olfactory tubercle

Rin: input resistance

RS: regularly-spiking

TEA: tetraethylammonium

TTX: tetrodotoxin, blocks voltage activated Na channels

VAChT: vesicular transporter

VDCC: voltage dependent Ca2+ channel

xii VGLUT: vesicular glutamate uptake transporter

VTA:

xiii Investigation of the Electrophysiological Properties

of the Major Cell Types in the Rat Olfactory Tubercle

Abstract

by

ELIZABETH CHEN CHIANG

Olfactory information is processed by a diverse group of interconnected regions. Most efforts to define the cellular mechanisms involved in processing olfactory information have been focused on understanding the function of the olfactory bulb, the primary second-order olfactory region, and its principal target, the piriform .

However, the olfactory bulb also projects to other targets, including the rarely-studied olfactory tubercle, a ventral brain region recently implicated in regulating cocaine-related reward behavior. We used whole-cell patch clamp recordings from rat tubercle slices to define the intrinsic properties of neurons in the dense and multiform cell layers. We find three common firing modes of tubercle neurons: regular-spiking, intermittent-discharging and bursting. Regular-spiking neurons are typically spiny-dense-cell-layer cells with pyramidal-shaped, dendritic arborizations. Intermittently-discharging and bursting neurons comprise the majority of the deeper multiform layer and share a common morphology: multipolar, sparsely-spiny cells. Rather than generating all-or-none stereotyped discharges, as observed in many brain areas, bursting cells in the tubercle generate depolarizing plateau potentials that trigger graded but time-limited intermittent discharges. We find two distinct subclasses of bursting cells that respond similarly to

xiv step stimuli but differ in the role transmembrane Ca currents play in their intrinsic

behavior. We also created a numerical model of the bursting cell to examine the currents that allow the unique bursting pattern of firing. We varied the currents until we found a response that was similar to that of recordings. Then we ran simulations to test what the

response of the model neuron in environments mimicking Cs, low Ca, and TTX. We also

looked at the response to brief stimuli and two pulse stimuli. Experiments and the

mathematical model both lead us to assert that the role of calcium in nonregenerative

bursting tubercle neurons appears to be to decrease excitability by triggering Ca-activated

K currents. Nonregenerative bursting cells exhibit a prolonged refractory period following even short discharges suggesting that they may function to detect transient

events.

xv

Chapter 1

Introduction

1 Olfaction is the oldest sensory system since it involves the basic response to a

chemical stimulus suspended in the air or water from the environment of the organism.

Olfaction is still the primary used by a number of organisms and is necessary for

their survival. In addition, it is often used to find potential food sources as well as

potential mates. Much of is actually due to olfaction, as a complex mixture of molecules rise from food entering the mouth to the back of the during chewing and swallowing. Throughout evolution, the olfactory system has developed into several dedicated brain structures.

In this introduction, I will review the olfactory system as an introduction to the olfactory tubercle. I will then describe the location, structure, and organization of the tubercle. The cell types described through previous studies utilizing Golgi staining

methods will be reviewed. After which, I will discuss the inputs and projections of the

tubercle, followed by what is known about the used in the tubercle.

Dopamine and its binding to different dopamine receptors in the tubercle will be

highlighted. In particular, the dopaminergic input and the dopamine receptors in the

tubercle may be associated with the possible role of the tubercle in neuropsychiatric

disorders as well as cocaine addiction. Cocaine addiction has been attributed to the

accumbens just dorsal to the tubercle. Finally, the possible classification of the tubercle as part of ventral striatum and the will be discussed.

Basic description of the olfactory system

Humans are able to perceive thousands of different odorants which inform us of

availability of food, potential danger, or even what flowers are growing in the garden.

2 We may associate some with grandma’s house and others with our dog. Our

of begins with odorant molecules suspended in the air or water vapor

entering our nostrils and diffusing into the mucus on the neuroepithelium in the back of

the nasal cavity. These odorant molecules then bind to receptors on the cilia of olfactory

sensory neurons.

The contains olfactory sensory neurons, supporting cells, and

basal stem cells. The olfactory is a bipolar cell with an apical

and a basal . The extends to the epithelial surface where it expands into a knob with cilia that protrude into the mucus coating the epithelium. The cilia of the olfactory neuron have specific receptors for odorants. An axon projects from the basal pole through the bony cribiform plate above the nasal cavity to the olfactory

bulb. The olfactory sensory neurons are unique compared to other neurons in that they

have a short lifespan and are constantly replaced from the basal stem cell population.

Sensory information from the epithelium is transmitted to the olfactory bulbs.

There are five types of neurons in the olfactory bulb; mitral neurons, tufted neurons,

periglomerular , granule cells, and Blanes cells. The mitral neurons and

tufted relay neurons project to the olfactory cortex while the periglomerular interneurons

encircle specialized structures in the olfactory bulb called glomeruli. Odorant receptor

neurons that express the same type of odorant receptor project to the same glomeruli.

The sensory olfactory neuron axons on the dendrites of the olfactory bulb

neurons in the glomeruli, with one axon making in only one . The

primary dendrite of each mitral and tufted relay neuron is confined to a single

glomerulus. The axons of several thousand sensory neurons converge on the dendrites of

3 20-50 mitral neurons in the glomerulus. The sensory neuron axons of many different areas of the epithelium converge onto one glomerulus. Each responds to multiple odorants, but mitral cells connected to different glomruli generally respond to different sets of odorants.

Axons of mitral and tufted relay neurons of the olfactory bulb project through the lateral olfactory tract to the olfactory cortex. The olfactory cortex consists of the anterior olfactory nucleus, the prifirom cortex, parts of the amygdale, the olfactory tubercle, and part of the . Of these structures, there have been many studies on the , the entorhinal cortex, and the amygdale but very little is known about the olfactory tubercle.

Structure of the olfactory tubercle

The olfactory tubercle is a prominent structure on the ventral side of the brain. It is bounded by the lateral olfactory tract and the piriform cortex on the lateral border. The dorsal border is not easily defined due to the striatal bridges that lie between the and the tubercle. A diagram of the olfactory tubercle can be seen in Figure 1-

1.

The three layer laminar pattern of the olfactory tubercle was described by Ramon y Cajal (Calleja, 1893). The three layers consist of the molecular layer, also referred to as the plexiform layer, the dense cell layer, and the multiform layer, also referred to as the polymorph layer. These three layers are clearly visible when viewing a Nissl stain as seen in Figure 1-2. The molecular layer lies between the dense cell layer and the pial surface. Few cell bodies are found in the molecular layer, and the majority of nuclei

4 stained in this layer are glial cells (Millhouse and Heimer, 1984). The dense cell layer

(DCL) consists of tightly packed cell bodies resembling a layer. The cell bodies in the DCL are similar in size and shape. Dorsal to the dense cell layer is the multiform layer which consists of cells that vary in size and shape. The multiform layer is the thickest layer of the tubercle. Medium sized cells contribute to bridges, referred to as the strital bridges, which connect the multiform layer with the nucleus accumbens.

The olfactory tubercle also contains compact clusters of small granule cells called the Islands of Calleja which are found in the multiform layer. The of the granule cells in these islands have been studied by Halliwell (Halliwell and Horne,

1995). The granule cells in the islands have neurons that measure 7-12 µm in diameter and have one to three short primary dendrites that stay within the island. These cells are

GABAergic with axons that do not appear to leave the island. Granule cells may be electrically coupled through gap junctions (Ribak and Fallon, 1982).

The most common cell found in the tubercle is the medium-sized densely spined neurons. These neurons have cell bodies that are free of spines but have several dendrites which are covered with spines. The cell bodies tend to vary between round, fusiform, and triangular in shape. There are a few primary dendrites that are generally short and spineless, which then divide into multi-branched dendritic trees. The lateral extent of the dendritic trees is 200 µm (Millhouse and Heimer, 1984). These cells make up the majority of cells found in the dense cell layer and, in the past, have been referred to as pyramidal cells. The arborization of the dendritic trees vary such that Ramon y Cajal

(1955) wrote that these cells are “… more irregular than in any other cortical region.”

5 Heimer reported that the axons of medium sized densely spined neurons in the

dense cell layer originated either from the soma, the dorsal primary dendrites, or in some

rare cases, a ventral dendrite. The axon was generally found to travel dorsally and

somewhat caudally toward the multiform layer. Each axon generates several collaterals

arising near the origin of the axon and remaining within the tubercle, generally within the dendritic tree of the neuron. (Millhouse and Heimer, 1984)

Heimer also described other types of neurons with varying cell body shape, their dendritic arborization, and density of spines on dendrites (Millhouse and Heimer, 1984).

He described crescent neurons which have cell bodies that resemble a crescent or are semi-lunar shaped with thicker dendrites. These crescent neurons are located in the dense cell and multiform layers. Spindle cells have long, thin dendrites that are nearly spine free with few branches. Their soma size is similar to that of medium densely spined cells, and they tend to have two primary dendrites that divide once or twice. A group of medium sized neurons have long, thin, and sparsely spined or spine free dendrites. These cells are generally found in the multiform layer, though some were seen in the dense cell layer.

There are also a number of small cells described by Heimer. (Millhouse and

Heimer, 1984) The dwarf cell sits near pial surface. However, Heimer suggests that these cells are not in the molecular layer but are in the DCL at regions where the DCL is near the pial surface and the molecular layer is small to non-existent. The radiate cell, found mostly in DCL, has several short dendrites that radiate from the soma. There is also a spine-rich small cell found in the molecular layer. These cells have spine-free primary dendrites that quickly divide into branches that are richly covered with spines.

6 Large neurons were found both in the DCL and the multiform layer. The large neurons in DCL were spine-poor with round, conical, or ellipsoidal somas measuring 25-

35 microns. Several primary dendrites divided into secondary dendrites and have one or

two branch points were characteristic of these cells. The large neurons in the multiform

layer had polygonal, triangular, or fusiform somas that could be 60 µm in length. These

neurons had two to four primary dendrites that divide near their origin to produce long

secondary dendrites.

The striatal bridges consists mainly of medium spiny cells that are generally

bipolar with two dendritic shafts that branch into more dendrites. The striatal bridges are

generally 60-80 µm wide column with the dendrites of several cells compressed together.

Potential circuitry of the olfactory tubercle

The olfactory tubercle was classified as part of the olfactory system due to the

input from the olfactory bulb through the mitral cell axons in the lateral olfactory tract

(LOT). The LOT is a band of fibers that can be seen on the surface of the cortex and lies

on the lateral border of the olfactory tubercle. Afferent fibers branch from the LOT over

the surface of the tubercle and terminate in the outer molecular layer. (Allison, 1953;

Price and Sprich, 1975) If the olfactory bulb is lesioned, terminal degeneration of neurons

in the olfactory tubercle is seen. (Carlson et al, 1982) The sudden massive loss of

olfactory input as a result of the olfactory bulb lesion causes neurons receiving this input

to degenerate due to a transneuronal effect. Tracing olfactory bulb projections through

the degeneration method, the autoradiographic technique and the HRP method show that

7 the olfactory tubercle receives substantial input from the olfactory bulb (Carlson et al.,

1982) .

The tubercle receives input from a wide variety of sources outside of the olfactory

bulb. The ventral tegmental area (VTA) sends dopaminergic input to the olfactory tubercle which gives the tubercle a possible role in reward behavior (Voorn et al., 1986).

Another non-sensory input to the olfactory tubercle is the projection from the

horizontal limb of the diagonal band nucleus (Mesulam et al., 1983; Zaborszky et al.,

1986). The of the , a memory structure, has also been shown

to project to the tubercle (Kunzle, 2005). The olfactory tubercle is also a main projection

site of the prelimbic cortex and the medial frontal cortex: areas involved in planning and initiating of movement (Vertes, 2004). The nucleus of the tractus solitarius and the

nucleus gemini have been found to send projections to the multiform layer of the

olfactory tubercle (Ruggiero et al, 1998, Price et al, 1991).

There has been disagreement over whether the olfactory tubercle sends

projections to the olfactory bulb Carmicheal et al. (1982). published that retrograde

tracers injected into the olfactory bulb do not label the olfactory tubercle, but Levy et al.

(1999) published that injection of a retrograde fluorescent tracer into the olfactory bulb

labeled the olfactory tubercle. If the olfactory tubercle does not project to the olfactory

bulb, it will be the only olfactory area that receives input from other olfactory areas

including the bulb that does not send reciprocal output to the bulb.

Golgi stains showed axons of the medium sized cells of the dense cell layer reaching to the ventral pallidum (Millehouse and Heimer, 1984). Different tracer studies have given possible projection sites for the olfactory tubercle. Injections of anterograde

8 tracer in the mediodorsal thalamic nucleus label the medium and large sized cell in the

multiform layer of the tubercle. Tracing studies show a projection from the multiform

layer of the olfactory tubercle to the posterior half of the lateral with axons

confined to the in the rostral hypothalamus (Price et al, 1991).

The neurotransmitters used in the tubercle remain unknown. Determining the neurotransmitter for a neuron is best done by paired recordings. However, paired recordings require some knowledge of the circuitry of the neurons in order to successfully isolate one neuron projecting to another. The next approach is to stain for antibodies for neurotransmitters or for proteins related to neurotransmitter production and release. Some neurotransmitters, like GAD or GABA, have an antibody that consistently stains cells that use that neurotransmitter. Other neurotransmitters, like glutamate, are much more difficult to stain to find which cells use that molecule as a neurotransmitter.

It is also important to differentiate between staining dendrites and axons that may project from other areas of the brain to the stained area versus staining the cells where the cell body is located in the area of interest.

Since glutamate is an , it is found in every neuron and, therefore, a poor marker of whether or not gluatamate is used as a neurotransmitter in that neuron.

Vesicular glutamate uptake transporter 1 accumulates neurotransmitter glutamate into synaptic vesicles, making it a better marker for a neuron. VGLUT1 immunoreactivity was intense in the rat olfactory tubercle, but studies did not identify the morphology of the neurons that labeled VGLUT1 positive (Kaneko, Fujiyama and Hioki,

2002). Figure 1-3 shows the labeling for VGLUT1, and Figure 1-4 compares labeling of

VGLUT1, VGLUT2 and VGLUT3. Cell bodies in the olfactory tubercle multiform layer

9 also labels positive for VGLUT3 transcript and VGLUT3 protein, as seen in Figure 1-5

(Herzog, et al., 2004). VGLUT1 is expressed by glutamatergic neurons of the cerebral and cerebellar cortices, the hippocampus, and the , whereas VGLUT3 is found in cholinergic neurons in the caudate and serotonergic neurons in the raphe nucleus. (Kaneko, et al., 2002; Herzog, et al., 2004) The tubercle expressing both

VGLUT1 and VGLUT3 groups it both with cortical areas and basal ganglia.

The high level staining for acetylcholinesterase and acetylcholine indicate that the olfactory tubercle must have either cholinergic innervation or cholinergic cells. Figure 1-

6 shows the dark staining of the tubercle with acetylcholinesterase. To determine which is the case for the tubercle, staining must differentiate between cell bodies and axons and dendrites. Choline Acetyltransferase (ChAT) is an synthesized in the cell body and then transferred to the nerve terminal where it joins acetyl CoA to choline to form acetylcholine. Vesicular Acetylcholine Transporter (VAChT) is a transporter for acetylcholine that is present in cholinergic nerve terminals (Weihe, 1996). Neuron cell bodies, preterminal axons, and terminal-like structures were stained with both ChAT and

VAChT antibodies (Ichikawa, 1997; Roghani, 1998; Talbot et al., 1988). Cholinergic interneurons of the olfactory tubercle and Islands of Calleja are visualized with VAChT antibody staining (Schafer, 1998; Talbot et al., 1988). ChAT cRNA showed dense labeling in neurons in the olfactory tubercle (Lauterborn, 1993).

The OT densely labels for GABAergic fibers, and GAD mRNA is abundant in the olfactory tubercle (Veenman and Reiner, 1994; Feldblum, 1993). One study reported that

GABAergic cell bodies represent 10-15% of cells in the rat olfactory tubercle (Cloez,

1996). Another author asserted that the densely spiny cells, which are the most common

10 cells, stain GAD+ (Gritti et al., 1993). Figure 1-7 shows the distribution of cell bodies

staining GAD+, as well as for cell bodies staining ChAT+. In the nucleus accumbens, the medium spiny cells are believed to be GABAergic. Since the number of medium spiny cells in the tubercle far exceeds 10-15%, it seems unlikely that all medium spiny

cells are GABAergic. The granule cells of the Islands of Calleja fluoresce green in GFP

GABA transgenic mice and may represent a GABAergic population in the tubercle

(Shipley, personal communication).

Another important neurotransmitter in the tubercle is dopamine. Dopamine is not

used as a neurotransmitter by tubercle cells, but the tubercle receives dopaminergic input

from the ventral tegmental area. Electrical stimulation of the dopaminergic pathway

elicits dopamine release in the tubercle of rats (Suaudchagny et al., 1989). The effects of

dopamine are mediated by receptors belonging to several classes. These classes have been differentiated by their pharmacology and signal transduction mechanisms.

Generally, dopamine receptors are divided into three classes: D1, D2 and D3. The division between the D1 and D2 classes was first generally accepted, with D3 added later

(Kebabian and Calne, 1979). The D3 receptor shares 52% similarity with the D2 receptor but present pharmacology that is distinct from the D2 receptor (Mengod, Goudsmt, et al.,

1992). Knowing more about the dopamine receptors in the olfactory tubercle is important to help elucidate the effects of dopamine in the tubercle.

The D1 receptor has been found in high densities in the caudate-putamen, the

accumbens, and the olfactory tubercle including the Islands of Calleja (Mengod, Villaro

et al., 1992). It is also seen in the and the entorhinal cortex. The presence of

the D1 receptor in these areas could represent that D1 receptors are expressed by the

11 projecting cells and not intrinsic cells of the area. D2 receptors were seen in the caudate-

putamen, accumbens, and the straitum. In the striatum, D2 receptors are expressed by

cells intrinsic to the area.

The olfactory tubercle labels positive for D1, D2, and D3 receptors as seen in

Figure 1-8. The dense cell layer of the olfactory tubercle showed high levels of D2

receptors, while the Islands of Calljea showed no D2 transcripts (mRNA)

(Landwehrmeyer et al., 1993). The D3 receptor is located selectively in the limbic region

and is thought to play a role in psychiatric diseases (Sokoloff et al., 1992). One of the effects of chronic antidepressant treatments is a selective increase of dopamine D3 receptor expression (Lammers, 2000). Both the olfactory tubercle and the islands of

Calleja show high levels of staining for D3 receptors. Other areas that stain for D3

receptors include the anterior nucleus accumbens, the medial mamillary nucleus of the

bed nucleus of the stria terminialis, and the cerebellum. The anterior nucleus accumbens

and the olfactory tubercle are significant as the only areas expressing D1, D2, and D3

receptors. The Islands of Calleja express both D1 and D3 but do not express D2

receptors (Mengold, 1992).

Specific manipulations have affected the dopaminergic signal in the tubercle.

Removal of olfactory bulb input to the olfactory tubercles causes sprouting of

dopaminergic axons in the tubercle. There is an increase in density

but not affinity in the olfactory tubercle. (Lingham and Gottesfeld, 1986) Systemic α-

methyl-para-tyrosine, NSD-1015, reserpine, and d,l-apomorphine decrease dopamine

release in the olfactory tubercle evoked by electrical stimulation of the dopaminergic

pathway (medial forebrain bundle) in rats. Systemic pargyline, nomifensine,

12 , , sulpiride, and haloperidol enhanced dopamine released

in the tubercle. (Chagny et al., 1989)

The olfactory tubercle has often been found as a potential site of action for

drugs. The therapeutic effect of some antipsychotic drugs are thought to be

due to the interaction with dopamine D2-like receptors. Haloperidol, raclopride,

eticlopride, and nemonapride are D2 antagoinsts that affect binding at the D2 receptor

(Aassie et al., 2006). Another hypothesis is that dopamine-sensitive adenylate cyclase is

the target of action of antipsychotic drugs. (Clement-Cormier et al., 1974) Dopamine-

sensitive adenylate cyclase was studied in the , accumbens, , median

eminence, the cortex, and the olfactory tubercle. (Krieger, 1980) Activity of dopamine-

sensitive adenylate cyclase was highest in molecular layer and DCL, lowest in multiform

layer, and non-existent in the islands of Calleja.

The olfactory tubercle as part of the ventral striatum

Compared to the extensive number of investigations focused on the nucleus

accumbens, the olfactory tubercle receives little . However, recent studies have

provided evidence for the involvement of the olfactory tubercle in a number of functions, including cocaine addiction. The nucleus accumbens receives strong inputs from the hippocampus and and projects to the ventral pallidum and .

Like the nucleus accumbens, the tubercle also receives input from the hippocampus and projects to the ventral pallidum. The ventral pallidum is an important link in the mediation of both locomotor and orofacial activity. Both the nucleus accumbens and the

13 olfactory tubercle receive mesolimbic dopaminergic neurotransmission from the ventral tegemental area.

The mechanism of addiction for cocaine is still unknown, but cocaine is believed to cause a rewarding affect by inhibiting dopamine (Einhorn et al, 1988). Acute cocaine (20 and 40 mg/kg, i.p.) was found to decrease D2 receptor mRNA expression.

Chronic cocaine (20 mg/kg, i.p. x 15 days), however, produced an increase in D2 receptor mRNA levels in this region which was detected 24 h, but not 7 days, after withdrawal (Spyraki and Sealfon, 1993). Pharmacological blockage of dopamine receptors removes the rewarding effect of cocaine in animals.

The nucleus accumbens is strongly thought of as the brain structure mediating the rewarding effects of psychomotor stimulates including cocaine. Cocaine administration to animals causes an increase of extracellular dopamine in the accumbens. Lesion of dopaminergic terminals in the accumbens decreases the rewarding effects of systemic cocaine administration (Roberts et al, 1977). Injection of into the nucleus acumens also decreases the rewarding effects of cocaine. However, injection of cocaine directly into the accumbens gives a marginal, if any, rewarding effect.

When cocaine is injected into the olfactory tubercle, it causes locomotion and rearing. Rats will learn to self-administer cocaine injections to the tubercle but not the medial core or shell of the accumbens nor the ventral pallidum or dorsal striatum. Rats also learn to discriminate between two levers to choose the lever that leads to cocaine infusion in the tubercle. Co-administration of D1 or D2 receptor antagonists with cocaine decreased self-administration rates. Animals also showed conditioned place preference after receiving injections to the tubercle but did now show preference to the cocaine-

14 associated compartment when cocaine was injected into the shell of the accumbens or the

ventral pallidum. Cocaine and significantly increased utilization in the

olfactory tubercle. These results implicate the tubercle’s involvement in the rewarding

effects of cocaine and possible other drugs of abuse (Ikemoto, 2003). The nucleus

accumbens has been proposed to constitute the neural interface between the limbic and

motor systems (Mogenson et al., 1980). It is possible that the tubercle is also involved,

choosing the behavioral response to olfactory cues.

The connection of the olfactory tubercle with the nucleus accumbens is not a new

discovery. Heimer has argued that the corpus striatum of the rat extends to the ventral

surface of the rat brain to include the nucleus accumbens and the olfactory tubercle

(Heimer, 1978; Heimer and Wilson, 1975; Millhouse and Heimer, 1984). He uses the

term ventral striatum to identify this area to emphasize that these structures are the

ventral part of the striatum and associated with the nucleus accumbens, caudate nucleus,

and the putamen. This designation would classify the olfactory tubercle as part of the

basal ganglia instead of as a cortical area.

The basal ganglia regulate planning and execution of movement by providing

feedback circuits that regulate cortical and brain stem motor areas. It receives excitatory, glutaminergic projections from various areas of cortex, as well as dopaminergic projections from the and serotonergic input from . The basal ganglia projects to the motor areas of cortex via the thalamus, but it does not project directly to projection neurons in the brain stem or have significant output to .

This brain structure has been found to be necessary for smooth movement and posture.

15 Degenerative diseases of basal ganglia include Parkinson’s and Huntington’s which

involve involuntary movements and abnormalities of posture.

The basal ganglia consist of the striatum, the (or pallidum), the

substantia nigra, and the . The striatum consists of the caudate

nucleus, the putamen, and the ventral striatum (which includes the nucleus accumbens).

The caudate nucleus and putamen are divided by the on the posterior

end. The caudate nucleus, the putamen, and the nucleus accumbens have a common

embryological origin.

The striatum appears as a large ‘striated’ mass due to the dispersed fiber bundles,

representing corticofugual and corticopetal projections. Many of these projection fibers

gradually collect to form the internal capsule, which divides the stratium into the caudate

nucleus and the putamen. The cytoarchitecture of the caudate nucles and the putamen are

similar. The two receive their neocortical input from different parts of ,

with the putamen receiving more input from the sensory-motor cortex and the caudate

receiving input from regions known as the associative cortex.

The division between the nucleus accumbens and the olfactory tubercle is made

up of the fibers of the olfacto-diencephalic projection system and deep olfactory

radiation. Between bundles of fibers, the striatal bridges, that are made up of spiny

medium neurons, connect the olfactory tubercle and the nucleus accumbens. The

accumbens and the tubercle have embryological, cytological, and histochemical features in common with the rest of striatum. When looking at acetylcholinesterase (AChE) stained sections of the rat brain, one readily sees why the accumbens and the tubercle appear to be an extension of the striatum. Although the densely spined neurons of the

16 tubercle’s dense cell layer have been called pyramidal neurons due to the tubercle being classified as a cortical area, these neurons resemble the spined ‘striatal’ neurons found in the caudate putamen. The spiny neurons in the tubercle have a larger variability in their dendritic pattern than striatal spiny neurons. The spiny tubercle neurons also share the characteristic pattern of striatal spiny neurons have: axons with pronounced local collaterization in the immediate area of the cell. The large majority of medium spiny neurons have been shown to be GAD positive (Oertel and Mugnaini, 1983). The large aspiny neurons, which are found in the tubercle, show similarities with the large neurons of the accumbens and caudate putamen.

Another similarity between the nucleus accumbens and olfactory tubercle is that they both project to the ventral pallidum. (Heimer, 1978; Heimer and Wilson 1975)

Common histochemistry, morphology of neurons, and projection targets make the nucleus accumbens and the olfactory tubercle more similar than previously perceived.

The classification of the tubercle as a part of the ventral striatum may give clearer insight on the functional role of the olfactory tubercle.

17

Figure 1-1

18 Figure 1-1: Schematic diagram showing the ventral structures contained in tubercle slices.

The tubercle is bounded by the nucleus of the horizontal limb of the diagonal band

(HDB), piriform cortex and the nucleus accumbens. The lateral olfactory tract (LOT) is clearly visible in tubercle slices as it is cut through in cross section. The tubercle is a three layered structure consisting of the molecular layer (ML), the dense cell layer

(DCL) and the multiform layer (MFL). The deep multiform layer in the tubercle contains multiple tightly packed clusters of small cells (islands of Calleja, IC) and linear bundles of neuronal processes called strital bridges (Str. bridges) The strital bridges separate bundles of fibers which separate the tubercle from the nucleus accumbens.

19

Figure 1-2 Modified from Paxinos and Watson, 1998

20 Figure 1-2: Nissl stain of coronal slice of rat olfactory tubercle

Top: Diagram labeling the ventral structures of coronal slices of the rat brain corresponding to the Nissl stain below.

Bottom: Nissl stain of coronal slice of the rat brain corresponding to diagram on top. The olfactory tubercle is enclosed in the red circle. The dark band of staining indicted by the arrow represents the cell bodies that make up the dense cell layer. Note how the dense

cell layer connects with the pyramidal layer of the piriform cortex, which lies just lateral

to the olfactory tubercle. The dark stained structures in the multiform (arrowheads) are

the islands of Calleja. The molecular layer has few cell bodies.

CB cell bridge

DEn dorsal endopiriform nucleus

HDB horizontal diagonal band

ICj islands of Calleja

LAcbSh lateral accumbens nucleus, shell

LSS lateral stripe of the striatum

Pir piriform cortex

Tu olfactory tubercle

VP ventral pallidum

Modified from Paxinos and Watson, 1998

21

Figure 1-3 Modified from Pal et al., 1999

22 Figure 1-3: Coronal slices of rat brain stained for proteins related to

neurotransmitter glutamate.

Top: Coronal slice of a rat brain stained for differentiation-associated Na+ dependent inorganic phosphate cotransporter (DNPI). This is a protein related to the

neurotransmitter of glutamate. Positive staining indicates the presence of glutamate that

acts functionally as a neurotransmitter. Note the positive staining the multiform layer of

the olfactory tubercle.

Bottom: Coronal slice of a rat brain stained for Vesicular 1

(VGluT1), a protein related to neurotransmitter glutamate. The molecular layer of the

tubercle is darkly labeled, most likely due to the glutamatergic input that the tubercle receives. The multiform layer of the tubercle also labels positively for VLGUT 1.

ac anterior commisure

Acc nucleus accumbens

OT olfactory tubercle

Pir piriform cortex

VP ventral pallidum

Modified from Pal et al., 1999

23

Figure 1-4 From Herzog et al., 2004

24 Figure 1-4: Regional distribution of VGLUT3 mRNA and protein

Comparison of VGLUT1, VGLUT2, and VGLUT3 distribution. Autoradiograms

obtained by in situ hybridization performed with [S35]labeled antisense oligonucleotides

specific of VGLUT3 nucleic sequence on coronal sections from the rat brain are in the

first column (A, E, I and M). Immunoautoradiographic labeling of VGLUT1-3 were obtained by incubating adjacent coronal sections from the frontal rat brain with affinity-

purified anti-VGLUT3 (second column, B, F, J, N), anti-VGLUT1 (third column, C, G,

K, O), anti-VGLUT2 (fourth column, D, H, L, P) antisera and then to anti-rabbit [125I]

IgG. This experiment illustrates the very different distribution pattern of the three

VGLUTs. Notice in F how the olfactory tubercle (Tu), the nucleus accumbens (Acb),

and the caudate putamen (CPu) are darkly labeled in contrast to the other brain structures.

In all sections, the tubercle appears to show a similar density of expression as the nucleus accumbens and caudate putamen.

Acb nucleus accumbens

AcbC nucleus accumbens core

AcbSh nucleus accumbens shell

AON anterior olfactory nucleus

BST bed nucleus of the

cc corpus callosum

Cg

CPu caudate putamen

25 Cx cerebral cortex

HDB horizontal diagonal band

IG indusium griseum

LS lateral septum

MS medial septum

Pir piriform cortex

Tu olfacotry tubercle

TT tenia tecta

VDB vertical diagonal band

VP ventral pallidum

From Herzog et al., 2004

26

Figure 1-5 Modified from Herzog et al., 2004

27 Figure 1-5: Immunohistochemical localization of VGLUT3 on a coronal rat brain section taken at the level of the tubercle and the accumbens

Coronal slice of the rat brain was stained for VGLUT3. Note that the olfactory tubercle and parts of the nucleus accumbens shell are both heavily stained. The unstained area between the olfactory tubercle and the nucleus accumbens are the fiber bundles that separate the two structures. The positive expression for VGLUT3 help visualize the strital bridges that link the olfactory tubercle with the nucleus accumbens. The piriform cortex next to the olfactory tubercle shows very weak expression for VGLUT3.

ac

AcbC nucleus accumbens core

AcbSh nucleus accumbens shell

Pir piriform cortex

Tu olfactory tubercle

Modified from Herzog et al., 2004

28

Figure 1-6 Modified from Paxinos and Watson, 1998

29 Figure 1-6: Acetylcholinesterase stain of coronal slice of rat olfactory tubercle

Left: Coronal slice of the rat brain stained with acetylcholinesterase. Note the continuity of the staining between the caudate putamen, the accumbens, the ventral pallidum, and the tubercle. In contrast, the piriform cortex adjacent to the olfactory tubercle, is not stained darkly with acetylcholinesterase and a clear boundary can be seen between these two brain structures with this stain.

Right: Diagram for naming the structures seen in this section of the rat brain.

AcbC accumbens nucleus, core

AcbSh accumbens nucleus, shell

AID agranula , dorsal part

AIV agranula insular cortex, ventral part

Cg1 cingulate cortex area 1

Cl

CPucaudate putamen (striatum)

DEn dorsal endopiriform nucleus

DP dorsal peduncular cortex

DTT dorsal tenia tecta

Gl granular insular cortex

ICj islands of Calleja

Il

30 LSS lateral stripe of the striatum

M1 primary motor cortex

M2 secondary motor cortex

Pir piriform cortex

PrL prelimbic cortex

SHi septohippocampal nuclues

SiJ primary sematosensory cortex, jaw region

SL semilunar nucleus

Tu olfactory tubercle

VO ventral orbital cortex

VP ventral pallidum

Modified from Paxinos and Watson, 1998

31

Figure 1-7 Modified from Gritti et al., 1993

32 Figure 1-7: Diagram of ChAT+ and GAD+ neurons in the tubercle.

Atlas figure drawn from Nissl-stained sections (with the aid of a computerized image analysis system) to determine where ChAT+ (open circles) and GAD+ (closed red triangles) neurons were located. The olfactory tubercle (OTu) showed neurons that stained both ChAT+ and GAD+ indicating the presences of cholinergic neurons and gabaergic neurons. These neurons appear to be more dense in the multiform layer with

GAD+ cells more dorsal than ChAT+ cells.

Modified from Gritti et al., 1993

33

Figure 1-8 Modified from Mengod et al., 1992

34 Figure 1-8: Comparison of the distribution of D1 (A), D2, (B), and D3(C) mRNA in the rat brain.

Pictures A-C are photomicropgrahs from film autoradiograms of adjacent sections hybridized with 32P-labelled oligonucleotides. Picture D is a cresyviolet stained section

corresponding to C. The olfactory tubercle (circled in red) expresses D1, D2, and D3

mRNA. Picture A shows a continuity of D1 expression between the striatum and the

tubercle. The olfactory tubercle is one of the few areas of the rat brain to express D3

mRNA.

Acb, nucleus accumbens

CPu, caudate-putamen

ICj, Island of Calleja

ICjM, Island of Calleja magna

Tu, tuberculum olfactorium

Bar equals 3 mm.

Modified from Mengod et al., 1992

35

Chapter 2

Diversity of neural signals mediated by multiple burst-firing mechanisms

in rat olfactory tubercle neurons

36 Introduction

The olfactory bulb is the primary conduit that enables chemosensory inputs to reach cortical areas (Shepherd and Greer, 1998). However, projections from the bulb

target a wide variety of brain areas, including the piriform cortex as well as several

poorly-understood regions, such as the anterior olfactory cortex, agranular insula cortex

and olfactory tubercle (Luskin and Price, 1983; Heimer et al., 1985; Shepherd, 2004).

Even among these “secondary” targets of the olfactory bulb, the olfactory tubercle (the

anterior perforated substance in ) stands out both for how little is known about its

cellular and synaptic organization and for the enigmatic structures found there, such as

the islands of Calleja and the striatal bridges. Few previous studies have reported

intracellular recordings from tubercle neurons in either brain slices or in vivo, and these

have focused on neurons within islands of Calleja (Halliwell and Horne, 1995, 1998). No

previous studies have applied modern patch-clamp methods to the tubercle. Millhouse

and Heimer (1984) used Golgi staining techniques to define the major cell types in the

tubercle, which include spiny neurons located predominately in the dense cell layer

(DCL; Figure 2-1) and sparsely-spiny, multipolar neurons located in the deeper

multiform layer (MFL). The elemental properties of tubercle neurons (intrinsic

, neurotransmitter, synaptic targets) have not been defined, though a

subpopulation of tubercle cells, and most granule cells within islands of Calleja, appear to

be GABAergic (Gritti et al., 1993). Even the fundamental question of whether the

tubercle is primarily an olfactory brain region or is, instead, a component of the limbic

system or basal ganglia is unresolved (Haberly and Price, 1978; Luskin and Price, 1983;

37 Millhouse and Heimer, 1984; Heimer et al., 1985). Also unknown is whether the neurons in the tubercle generate distinctive responses when driven with slow, phasic input during sniffing. Previous work in the olfactory bulb (Balu et al., 2004; Balu and Strowbridge,

2007) demonstrated that mitral cells express specific ionic currents that enable them to phase-lock to sniffing-like periodic inputs with very high temporal precision. It is not known whether tertiary olfactory targets, like the olfactory tubercle, also have intrinsic properties that are linked to the temporal dynamics of neuron activation.

While the functional properties of tubercle neurons (and the tubercle itself) are mysterious, new behavioral and neurochemical findings suggest that the tubercle may be a critical brain region that mediates the rewarding effects of cocaine. Recent work that reexamined how lesions of the ventral forebrain affect self-administration of psychomotor stimulants, such as and cocaine, points to the tubercle’s significance. Classic work on this question implicated the nucleus accumbens in mediating the stimulant-driven reward, since relatively large lesions in this region diminish amphetamine self-administration by rodents (Lyness et al., 1979) and conditioned place-preference (Spyraki et al., 1982). Direct injections of amphetamine into the nucleus accumbens supports self-administration (Hoebel et al., 1983) and conditioned place-preference (Carr and White, 1986). While these results often have been generalized to include cocaine, Ikemoto (2003) found that relatively small lesions of the olfactory tubercle, but not of the nucleus accumbens, disrupt cocaine-induced conditioned place-preference. The same group also found that rats will self-administer cocaine into the olfactory tubercle more readily than into the core of nucleus accumbens

38 and that the rewarding aspects of cocaine were abolished by co-infusion of dopamine

receptor antagonists into the tubercle (Ikemoto, 2003; Ikemoto and Donahue, 2005).

While understanding how the tubercle modulates behavioral reward and the role

that dopamine receptors play in this system are critical, long-term questions, these issues

cannot be resolved without first defining the intrinsic properties of the major cell types in

the tubercle. These functional elements may represent both potential sites of action for

cocaine and therapeutic targets to combat addiction. In this study, we define the intrinsic properties of tubercle neurons and find three broad classes of intrinsic responses in these neurons: regular-spiking, intermittently-discharging and bursting. The first two neuronal classes closely resemble similarly-named cells in other cortical regions. Bursting cells in the tubercle, however, are unusual and can generate graded, time-limited clusters of action potentials that may function to signal sudden changes in synaptic input. We also find two subclasses of bursting cells, both located predominately in the MFL, that differ dramatically in the underlying cellular mechanism responsible for their intrinsic bursts and in their short-term, intrinsic plasticity.

Methods

Slice preparation and recording

Olfactory tubercle slices (300 µm thick) were prepared by making coronal slices from anesthetized (ketamine, 140 mg/kg, i.p.) P14-24 Sprague-Dawley rats using a modified

Leica (Nussloch, Germany) VT1000S vibratome. An artificial cerebrospinal fluid

(ACSF) dissection solution with reduced Ca was used when preparing and storing slices.

39 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 chilled to 4°C during slicing.

Brain slices were incubated in a 30°C water bath for 30 min and then maintained at room

temperature. During experiments, 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.

Whole-cell patch-clamp recordings were made from neurons visualized under IR-DIC

optics, using either an Axioskop 1 FS (Carl Zeiss, Thornwood, NY) or BX51WI

(Olympus, Center Valley, PA) fixed-stage upright microscope and a Multiclamp 700A or

Axopatch 1D amplifier (both from Axon Instruments/Molecular Devices, Sunnyvale,

CA). We avoided recording from granule cells located in or near islands of Calleja in the

MFL. Because of this selection bias away from small, granule-like cells, our results may

not represent the full extent of variation in intrinsic properties among MFL cells. Patch

pipettes (typically 4-6 MΩ resistance) contained (in mM) K-methylsulfate 140, NaCl 4,

HEPES 10, EGTA 0.2, MgATP 4, Na3GTP 0.3 and phosphocreatine 10. In some

experiments, this internal solution was supplemented with a polar intracellular marker

(Alexa594, 100 µM or Neurobiotin, 0.5 %). All chemicals were obtained from Sigma

except for Alexa594 hydrazide (Molecular Probes/Invitrogen, Eugene, OR),

Neurobiotion (Vector Labs, Burlingame, CA) and TTX (Calbiochem/EMD, San Diego,

CA).

Neuronal imaging

40 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, Lexington,

MA). Intracellularly-loaded fluorescent dyes were excited at 830 nm through a 60× water-immersion objective (Olympus). Emitted light was detected through an epifluorescent light path that included a 700DCLPXR dichroic mirror, a BG39 emission filter (both from Chroma Technology, Rockingham, VT) and a cooled PMT detector module (H7422P-40; Hamamastu, Bridgewater, NJ). Photomultiplier output was

converted into an analog voltage by a high-bandwidth current preamplifier (SR-570;

Stanford Research Systems, Sunnyvale, CA). 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, Danbury, CT)

and a Uniblitz shutter (Vincent Associates, Rochester, NY). In most experiments, the

output of the Mira laser was attenuated by 90% to 95%.

In some experiments, slices containing labeled neurons were fixed (4%

paraformaldehyde) and either viewed with a standard, wide-field epifluorescence

microscope (Axioskop 2, Carl Zeiss) after being mounted in Prolong (Invitrogen; for

Alexa593-filled cells) or processed to visualize Neurobioton labeling using the ABC

peroxidase method (Horikawa and Armstrong, 1988) and using the Vector Elite kit

(Vector Labs) and the Neurolucida 3D reconstruction system (Microbrightfield,

Williston, VT). Neurobiotin-filled cells were used predominately to assess the overall dendritic arborization. The Alexa-filled neurons were used to assess fine morphological details (e.g., dendritic spines), especially when imaged using 2-photon microscopy.

41 Soma area was estimated by measuring the length of orthogonal major and minor axes of

the best-fit ellipse. Spine density was estimated by averaging the density at four different

dendritic locations in maximal projections of 2-photon Z-stack images (mean dendritic length examined = 54 µm).

Data acquisition and analysis

Electrophysiological data were recorded and analyzed using custom software written in

Visual Basic 6 (Microsoft, Redmond, WA) and Origin 7.5 (OriginLab, Northampton,

MA). Current and voltage records were low-pass filtered at 2 kHz and then digitized at 5 kHz, using a 16-bit A/D converter (ITC-18, Instrutech, Port Washington, NY). Input resistance was assessed by measuring the maximal response to hyperpolarizing current steps that caused 5-10 mV hyperpolarizations. Action potential properties (amplitude,

width, latency-to-peak-spike-AHP, AHP amplitude) were calculated by custom software

written in Visual Basic. Neurons were included in this study if they had overshooting

action potentials and input resistances > 100 MΩ. The average input resistance across

our population of tubercle cells was 263.8 ± 14.5 MΩ. The tendency for some tubercle cells to discharge primarily during the initial phase of long (2+ sec duration) depolarizing steps was assessed using a spike clustering ratio (number of spikes evoked during the first

500 ms / number of spikes evoked during first 2000 ms). This metric would yield a value of 0.25 for tonically-discharging neurons and 1.0 for neurons that only fired during the initial 500 ms. The metric was averaged over multiple (mean = 5.8 ± 0.2) trials from each neuron. In figure 2-8 and figure 2-10A, we positioned grey rectangles at visually- identified inflexions in the membrane potential record to mark the temporal extent of depolarizing plateau potentials. Pharmacological agents were applied by changing the

42 perfusion solution. Voltages presented are not corrected for the liquid junction potential.

Unless noted, statistical significance was assessed using the Student’s t-test. Data are presented as mean ± S.E.M.

Results

Intrinsic firing patterns of olfactory tubercle neurons

We recorded from 176 rat olfactory tubercle neurons that fit within our selection criteria. Neurons in the dense cell layer (DCL) and multiform layer (MFL) of the rat olfactory tubercle could be classified into three broad groups based on their intrinsic firing responses: regular-spiking (RS), intermittently-firing (IF) and bursting. The molecular layer (ML) is sparsely populated with neurons and was not included in this study. Responses to graded depolarizing steps from each cell type are shown in Figure 2-

2A.

The firing rate in step-evoked discharges in RS neurons (n = 36) initially adapts, then becomes tonic throughout the remainder of the step stimulus. This adaptation pattern is shown in the plot of instantaneous firing-frequency in Figure 2-2B, left. Step- evoked discharges in these tubercle neurons closely resemble the firing behavior of regular-spiking neocortical pyramidal cells (McCormick et al., 1985) and, therefore, were classified as regular-spiking. The vast majority of neurons recorded in the DCL (82.6 %) were regular-spiking, though this discharge pattern was encountered occasionally in the

MFL (12.6 % of MFL cells). IF neurons (n = 32) also discharged throughout a 2-sec depolarizing step stimulus but did not fire tonically, except for short (~100-300 ms)

43 epochs. Responses in these neurons were dominated by long (>100 ms) pauses, giving rise to their classification name. IF neurons were encountered most frequently in the multiform layer (21.5 % of all MFL neurons) and occasionally in the DCL (8.7 % of all

DCL neurons). IF neurons had significantly greater coefficients of variation (CV) of the inter-spike intervals assayed in responses to 2-sec step responses (mean CVISI = 0.59 ±

0.09) than did RS neurons (0.18 ± 0.02; p < 0.001; Table 2-1). Intermittent discharges often are associated with GABAergic interneurons in other brain regions (McCormick et al., 1985), though occasionally this intrinsic behavior occurs in excitatory principal neurons, such as mitral cells in the olfactory bulb (Balu et al., 2004).

In bursting neurons (n = 98), action potential discharges were concentrated largely in the initial 500 ms of 2-sec depolarizing steps (bursting cells fired for 555 ± 58 ms (n =

98 cells) after step onset; mean discharge duration (first-to-last spike interval) = 289.6 ±

41.5 ms) and were followed by a pronounced afterhyperpolarization that began during the step response (Figure 2-2A, right). Bursting tubercle neurons had a greater initial spike- clustering ratio (0.88 ± .02; see Methods) than did either RS (0.30 ± 0.01) or IF tubercle cells (0.45 ± 0.05; both significantly less than bursting neurons; p < 0.001; Table 2-1).

Rather than generating all-or-none bursts, tubercle bursting neurons’ discharges were modulated with step amplitude (Figure 2-2A). Firing within burst discharges was typically intermittent, not tonic (mean CVISI = 0.52 ± 0.03; significantly greater than RS cells; p <0.001; not significantly different from intermittent neurons; p > 0.05; see also

Figure 2-5). As described below, functional tests can subdivide this category of tubercle neurons into two classes: regenerative and nonregenerative bursting cells. The cellular mechanisms and functional relevance of discharges in both types of bursting cells are

44 described below. Most bursting neurons were found in the MFL (65.9 % of MFL

neurons) with a minority found in the DCL (8.7 % of DCL neurons). Neurons in the

DCL were relatively homogenous with most classified as RS. By contrast, the multiform

layer was heterogeneous and included all three intrinsic cell types. A small percentage of

tubercle cells (5.7 %; 10 of 176) had firing patterns that contained elements of both IF

and bursting neurons (intermittent spikes clustered near the beginning of the step that did

not trigger a burst AHP) and were excluded from further analysis.

As shown in Table 2-1, many intrinsic properties (input resistance, membrane

time constant, action potential amplitude and width, latency-to-maximum-spike AHP)

were similar across the three major cell types in the tubercle. We observed a modest, but

statistically significant, difference in resting membrane potential between bursting (-62.2

± 1.1 mV) and RS cells (-69.5 ± 3.6 mV; p < 0.05). We also noted a slightly depolarized action potential threshold in RS neurons (-34.6 ± 1.9 mV), compared with both IF (-39.4

± 1.2 mV; p < 0.05) and bursting tubercle neurons (-39.2 ± 0.6 mV; p < 0.05). Spike- evoked AHP responses varied dramatically in tubercle cells, even among neurons in the same category (for example, see Figures 2-7, 2-8, and 2-9). On average, bursting neurons had significantly smaller spike AHP amplitudes (-7.8 ± 0.6 mV) than did either RS (-11.7

± 1.6; p < 0.001) or IF cells (-11.7 ± 0.9; p < 0.05). A majority of bursting neurons (80.8

%) had a “sag” in response to hyperpolarizing steps, most likely due to IH. We observed similar membrane potential sags in hyperpolarizing step responses less frequently in both

RS (42.9 %) and IF (38.9 %) cells.

Morphological correlates of olfactory tubercle neurons

45 We next sought to define the morphological properties of RS, IF and bursting tubercle neurons. We visualized 34 tubercle neurons filled with either Alexa594 (n = 26) or Neurobiotin (n = 8) through the patch pipette. All visualized RS neurons (n = 11) were spiny (3.0 ± 0.7 spines / 10 µm), multipolar neurons (mean number of processes emanating from the soma = 5.8 ± 0.7; Figure 2-3). The mean cell body area of DCL neurons was 172 ± 30.7 µm2. RS cells located in the DCL (n = 4) had extensive dendritic arborization in both the ML and MFL layers (Figure 2-3A-B). Morphological features of many RS cells (e.g., Figure 2-3B) suggest a similarity to superficial neocortical pyramidal cells with an apical dendrite that was oriented perpendicular to the DCL.

Electrophysiologically, all visualized DCL neurons were regular-spiking (Figure 2-3B, insert).

We recorded from 30 filled MFL neurons, including 10 IF cells and 13 bursting neurons. We found that IF and bursting neurons had generally similar morphologies

(Figure 2-4) that tended to have both fewer spines (1.0 ± 0.5 and 1.7 ± 0.1 spines / 10 µm for IF and bursting, respectively) and fewer primary dendrites (3.3 ± 0.4 and 2.5 ± 0.3) than did RS cells. Grouped together, IF and bursting MFL cells had significantly fewer spines (1.2 ± 0.3; p < 0.05) and fewer primary processes (2.9 ± 0.3; p < 0.001) than did

RS tubercle neurons. The cell bodies of IF neurons (mean area = 235.7 ± 30.3 µm2)

tended to be larger than the cell bodies of either bursting (179.6 ± 21.3 µm2) or RS (172.3

± 30.7 µm2) neurons, though these differences were not statistically significant (p > 0.05).

The dendritic arborization of non-RS cells was diverse and included sparsely-spiny cells whose dendrites were oriented primarily parallel to the DCL (Figure 2-4A1) and perpendicular to the DCL (Figure 2-4A2). One reconstructed MFL neuron had a sparsely

46 spiny process that entered a striatal bridge and terminated within the nucleus accumbens

(Figure 2-4B). These results suggest that the three distinct intrinsic firing patterns of tubercle neurons comprise two distinct morphological types: spiny, pyramidal-like neurons that discharge in a RS firing pattern and sparsely-spiny neurons that show IF and bursting discharge patterns.

Mechanism of intrinsic bursts in tubercle neurons

The absence of all-or-none, stereotyped discharges in tubercle bursting neurons raises the possibility that aspects of the burst response may represent attributes of the stimulus, such as intensity or slope. While the mean firing frequency within the discharge was correlated with step amplitude in the example shown in Figure 2-5, this relationship existed only through part of the stimulus range. Responses to larger steps deviated from this linear relationship (Figure 2-6; mean R2 = 0.69 ± 0.08; n = 11 cells).

We found a more robust correlation between stimulus amplitude and the first interspike interval (expressed as an instantaneous frequency), in the example shown in Figure 2-6

(R2 = 0.92) and in the population of 11 bursting cells tested systematically (mean R2 =

0.91 ± 0.01). Surprisingly, given the intermittent nature of the firing within the discharge

(mean CVISI = 0.52 ± 0.03) and the relatively weak correlation with mean firing- frequency, we also found a strong correlation between step amplitude and the total number of spikes evoked (R2 = 0.97 in the example shown in Figure 2-6; mean R2 = 0.88

± 0.02 for the population of 11 bursting cells tested). These results are summarized in

Figure 2-6B and suggest that stimulus intensity in bursting cells may be represented by the first interspike interval within the burst response.

47 Bursting tubercle neurons reliably discharged near the peak of ramp stimuli, such

as those shown in Figure 2-7A1. Both mean firing-frequency and the instantaneous

firing-frequency, determined by the initial two spikes, correlated well with the slope of

the ramp stimulus (R2 = 0.76 ± 0.08 for mean frequency and 0.91 ± 0.04 for

instantaneous frequency; n = 5). As shown in enlargements in Figure 2-7A2, the phase

relationship between the end of the ramp stimulus and the burst discharge was not

constant; steeper ramps triggered bursts near the end of the ramp while discharges began

before the peak stimulus intensity in less steep ramps. The ability of bursting neurons to

detect both rapidly- (steps) and slowly-changing stimuli (ramps) suggests a possible role for these neurons in signaling the rate of change in the firing of presynaptic neurons.

It is likely that the graded nature of discharges in bursting OT neurons is caused by a combination of active currents. Rather than being evoked by an underlying Ca spike, discharges in bursting OT neurons often appear on top of a steady, depolarizing plateau potential, such as shown in Figure 2-7A2 and Figure 2-8. As shown below, the graded nature of OT burst discharges is likely due to voltage-dependent modulation in

this underlying plateau potential. Burst-generating plateau potentials in the tubercle often

continue beyond the last action potential (Figure 2-8, top trace) before they are

terminated abruptly by the afterhyperpolarization.

Graded intrinsic bursts in tubercle neurons appear to result from the interactions

among at least four active currents: IH, IAHP, subthreshold Na current and low-threshold

Ca current. Most (80.8 %) tubercle bursting neurons exhibit membrane potential sag during steady hyperpolarizations, presumably reflecting the slow activation of IH during

the step response. The same current can diminish following depolarizing steps,

48 especially during the initial response, due to a transient reduction in input resistance

(Maccaferri and McBain, 1996). Consistent with this model, we found that bath application of the IH blocker Cs (4 mM) facilitated discharges in bursting neurons (Figure

2-8; mean number of action potentials evoked in Cs = 224.7 ± 61.7 % of control; significantly greater than control; p < 0.05; n = 6 cells). This increase in excitability was paralleled by a reduction in the membrane potential sag in response to hyperpolarizing steps (see insert in Figure 2-8) and was reversible upon washout of Cs. In most tubercle bursting cells tested (5 of 6), depolarizing steps still evoked time-limited burst responses in Cs (Figure 2-8, middle trace; number of APs in burst responses = 164.3 ± 15.3 % of control; n = 5 cells with plateau responses in Cs), suggesting that the underlying plateau potentials are modulated, but not mediated, by IH.

In all nonregenerative bursting tubercle neurons tested (6/6), blockade of transmembrane Ca currents by perfusion with a low Ca / high Mg extracellular solution converted the burst response into a prolonged discharge that persisted throughout most of the 2-sec depolarizing step (number of APs evoked = 310 ± 86 % of control; significantly greater than control; p < 0.05; n = 6; Figure 2-5C). Low Ca ACSF also reduced the normally prominent burst AHP response in these cells (see arrow in Figure 2-

9). Surprisingly, in regenerative bursting cells (5 of 5 cells tested; see Figure 2-17), reduction of Ca currents low with low Ca ACSF had the opposite effect—decreasing excitability and the number of spikes evoked by the step stimulus. Because of this difference in response to low Ca ACSF, the mechanisms underlying bursting in regenerative and nonregenerative cells will be considered separately.

49 Tetrodotoxin (TTX) blocked Na-based action potentials and reduced, but did not

abolish, the underlying depolarizing plateau potentials in nonregenerative bursting cells

(5 of 5 cells tested; Figure 2-10A). The amplitude and duration of these TTX-resistant

plateau potentials were graded with stimulus amplitude (Figure 2-10A, right traces),

suggesting that voltage-gated Na current is not required to generate plateau potentials in nonregenerative bursting cells. While low Ca extracellular solution increased excitability in nonregenerative bursting cells under control conditions (Figure 2-9), the same treatment reduced excitability and eliminated plateau responses evoked in TTX (Figure

2-10B), suggesting that low-threshold Ca channels may contribute to the underlying plateau potential. A role for low-threshold Ca channels also is suggested by the ability of weak, subthreshold depolarizations to trigger both plateau potentials and AHP responses in control conditions (Figure 2-10A, bottom left trace). In these examples, steady-state plateau potentials were triggered by <20 mV depolarizations from rest (to approximately

-55 mV), within the range of typical T-type low-threshold Ca channels (Randall and

Tsien, 1997). The burst AHP response was associated with a transient decrease in input resistance (to 71.2 ± 7.0 % of control; n = 4; Figure 2-11) that reversed polarity at -99 mV (Figure 2-11, insert), consistent with the activation of a Ca-activated K current.

Presumably, the counterintuitive effect of low Ca ACSF we find in control conditions

(increasing excitability and prolonging discharges) reflects the critical role Ca-activated

K currents play in truncating the burst discharge.

Together, our experiments suggest that low-threshold Ca currents underlie the depolarizing plateau response in bursting tubercle neurons and that IH and subthreshold

Na current function to enhance this response. Presumably, the inward current caused by

50 low-threshold Ca channels is opposed by K channels, generating the periods of steady- state depolarization that trigger burst discharges. While the identity of the channels involved in “flattening” the plateau response is not known, the plateau depolarization itself appears to be terminated by a Ca-activated K current.

Regenerative bursting tubercle neurons

In a minority of bursting tubercle neurons (21.4%; 21 of 98), short-duration (25 –

100 ms) stimuli could trigger regenerative discharges that outlasted the step

depolarization (Figure 2-12A). Responses from similar depolarizing stimuli applied to a

nonregenerative bursting cell are shown in Figure 2-12B. Both regenerative and

nonregenerative generated self-limiting discharges in response to 2-sec duration

depolarizing steps (mean duration = 329.7 ± 87.5 and 263.8 ± 40.5 ms, respectively;

mean number of spikes = 3.7 ± 0.5 and 3.7 ± 0.3, respectively) and thus were categorized

as bursting. While discharges generated by regenerative and nonregenerative bursting

cells were similar, bursts were initiated at significantly longer latencies in regenerative

cells (119.9 ± 26.8 versus 51.2 ± 7.3 ms; p < 0.01; Figure 2-13) and required less

depolarizing current from the same membrane potential (mean efficiency = 81.1 ± 32

versus 30.9 ± 4.9 spikes/nA; p < 0.05). Discharges in nonregenerative bursting cells also

had slightly but significantly higher initial clustering ratios (0.92 ± 0.02) than

regenerative cells (0.79 ± 0.06; p < 0.01; Figure 2-13). All regenerative bursting cells

were located in the MFL (12.6 % of MFL neurons), while nonregenerative bursting cells

occurred in both the DCL (8.7 % of DCL neurons) and MFL (53.3 % of MFL neurons).

51 Regenerative bursting cells generated prolonged plateau potentials that appeared

to be initiated by one or two action potentials (Figure 2-14). In most regenerative cells,

depolarizing steps could elicit a stereotyped plateau potential, with the latency of the

initial action potential graded with stimulus intensity. In regenerative cells that fired

doublets, the plateau depolarization often was terminated by a single action potential

(Figure 2-15). The prolonged periods without firing during the plateau response in

regenerative tubercle cells are reminiscent of “silent plateaus” recently reported in

subthalamic neurons (Kass and Mintz, 2006). However, unlike subthalamic cells, some

regenerative tubercle cells appear to have multiple stable plateau potentials evident upon

graded depolarization (right trace in Figure 2-15) or repeated steps (Figure 2-16A) from

the same resting potential. In the example shown in Figure 2-16A, a single action

potential appeared to trigger a transition from a depolarized plateau potential to a second, more hyperpolarized, potential. Plateau depolarizations initiated by brief steps could be truncated by hyperpolarizing pulses (Figure 2-16B) and were abolished in low Ca ACSF

(6 of 6 cells tested; Figure 2-17), suggesting that regenerative Ca currents contribute to these intrinsic responses.

Short-term plasticity of intrinsic responses in tubercle neurons

The large AHP that terminates the burst response in nonregenerative cells also imparts a dramatic, short-term plasticity to burst responses in tubercle neurons. As shown in Figure 2-18A, a single burst could completely inhibit the response to a subsequent, identical depolarizing step evoked 3 sec later (n = 12 cells). Bursting neurons appear to recover completely approximately 10 sec after the initial response,

52 with partially-recovered responses evoked at intermediate intervals. The suppression of burst responses is not caused exclusively by the membrane hyperpolarization associated with the AHP, since altering the bias current to match the membrane potential to the same voltage reached during the peak of the AHP response had only a minor effect on the burst discharge (Figure 2-18A, top right trace). This form of two-pulse plasticity was not observed in regenerative bursting cells (Figure 2-18B). Figure 2-18C summarizes the degree of two-pulse inhibition in recordings from 8 regenerative and 12 nonregenerative bursting cells.

The short-term plasticity of the burst response did not depend on the duration of the depolarizing stimulus used to trigger the burst. Discharges evoked by 500 ms steps could completely suppress burst responses to 2 sec depolarizing steps (Figure 2-19A).

Subthreshold depolarizing stimuli partially inhibited subsequent burst responses (Figure

2-19A, second trace), suggesting that the mechanism responsible for triggering the short- term plasticity has a low threshold. The suppression of the burst response by a depolarizing pre-pulse was not absolute. As shown in Figure 2-19B, the suppressed burst response could be recovered by increasing the step amplitude. These results argue that relatively weak depolarizations (including subthreshold depolarizations) can activate intrinsic mechanisms that dampen the responsiveness of tubercle neurons by raising their burst-discharge threshold over a period of several seconds.

Two-pulse inhibition in bursting cells dramatically alters neuronal responses to slow, phasic excitatory input. We applied trains of alpha functions (tau = 100 ms, interval = 400 ms) that mimic the normal pattern of glomerular activation in the olfactory bulb during sniffing (Charpak et al., 2001; Margrie and Schaefer, 2003), and have been

53 used previously to define intrinsic currents that govern mitral cell responses to periodic

input (Halabisky and Strowbridge, 2003; Balu et al., 2004; Balu and Strowbridge, 2007),

to the four classes of olfactory tubercle cells we have identified. As shown in Figure 2-

20A, both RS and IF cells discharge during each stimulus cycle, except for near-threshold responses in IF cells. (At very low stimulus amplitudes, IF cells fire only a single spike on the first alpha function.) While RS cells show weak adaptation in the number of spikes generated by each alpha function in the train, both regenerative and nonregenerative bursting cells tended to adapt completely and typically fired only during the first 1-2 cycles. The pronounced adaptation in nonregenerative bursting cells appeared to be due to the afterhyperpolarization following the burst discharge and could not be overcome with larger amplitude phasic stimuli. By contrast, adaptation over multiple cycles in regenerative cells was reduced with stronger stimuli (Figure 2-20A, top right trace), suggesting that adaptation in these cells may reflect the ability of intrinsic voltage-dependent Ca currents to amplify weak excitatory inputs. In support of this hypothesis, we found that low Ca ACSF reversibly abolished burst responses triggered by near-threshold trains of alpha functions in regenerative cells (Figure 2-20B). Responses to similar stimuli in IF cells differed dramatically from bursting cells and were potentiated following the response to the initial cycle (mean number of spikes triggered by sEPSP2 in IF cells = 30.5 ± 1.2 versus 13.1 ± 2.2 spikes triggered by sEPSP1; p <

0.001; n = 8). Figure 2-20C summaries the modulation of discharges during alpha function trains in bursting, IF and RS tubercle cells.

Discussion

54 We made three principal conclusions in this study. First, we found three types of firing patterns of neurons located in the dense and multiform cells layers that could be revealed by recording responses to 2-s depolarizing steps from the same membrane potential: regular-spiking, intermittently-discharging and bursting. Most neurons located in the DCL were regular-spiking, while we found all three firing modes in the MFL.

Second, we found a strong correlation between regular discharges and spiny, pyramidal cell-like dendritic morphology. By contrast, both bursting and IF neurons appeared to share a common sparsely-spiny, bipolar or multipolar dendritic morphology. The related morphology of bursting and IF neurons raises the possibility that these two intrinsic phenotypes may represent variations (or differential modulation) of a single cell type. In support of that hypothesis, we found that burst discharges are neither all-or-none nor regular/tonic but, instead, are intermittent. However, the wide variety of dendritic morphologies we encountered in intracellular fills of MFL neurons precludes an unequivocal answer at this time. Finally, we found two distinct mechanisms that generate burst responses in tubercle neurons. The most common burst discharge, termed

nonregenerative, involves both voltage-gated Na and Ca currents, is modulated by IH, and

is terminated by Ca-activated K current. In a minority of cells, a regenerative Ca current

amplified weak depolarizing inputs and appeared to generate bursts directly. These

different bursting mechanisms could be separated by their sensitivity to blockade of Ca

currents and by using tests for short-term intrinsic plasticity.

Relationship to previous work on the olfactory tubercle

Previous work using Golgi staining (Millhouse and Heimer, 1984) defined two

common morphologies of tubercle cells, medium-sized densely spiny cells and larger,

55 spine-poor neurons. Both cell types occurred in DCL and MFL layers, though most DCL

neurons appeared to be densely spiny neurons. Our work suggests that most medium

densely spiny tubercle neurons are regular-spiking while the spine-poor neurons consist

primarily of IF and bursting neurons. While the transmitter used by either cell type has

not yet been directly established using paired recordings, many DCL neurons are

immunoreactive for glutamic acid decarboxylase (Gritti et al., 1993), suggesting that they

are GABAergic. This transmitter identity is consistent with the morphological parallels

between spiny tubercle neuron and medium spiny neurons in the striatum, which are

GABAergic, and the dense cholinesterase staining pattern that clearly links the tubercle

with structures in the basal ganglia (Heimer et al., 1985). Like spiny tubercle neurons, most striatal medium spiny neurons are RS, though some fire tonically (Venance and

Glowinski, 2003). Relatively few studies have defined the intrinsic properties of tubercle cells using intracellular recordings. Halliwell and colleagues (Halliwell and Horne, 1995,

1998) recorded from granule cells in islands of Calleja and demonstrated

coupling within this neuronal population. The same group also used field recordings in

tubercle brain slices to investigate the synaptic circuits activated by extracellular stimulation in the ML and MFL (Owen and Halliwell, 2001).

Multiple mechanisms generate intrinsic bursts in tubercle neurons

The two types of bursting tubercle neurons we found could be separated by both pharmacological tests (reducing Ca currents using low Ca ACSF; Figure 2-9 and 2-17) and by using two-step protocols to test for short-term intrinsic plasticity (Figure 2-18).

Preliminary experiments testing the effects of blockers of specific types of Ca channels on the burst response have yielded complex results, suggesting that time-limited bursts in

56 tubercle cells may result from multiple Ca and Ca-activated currents. A parallel study is

underway in which the primary Ca currents are dissected pharmacologically and the

resulting data used to generate a realistic computer model of bursting cells. Our present

results suggest that the principal difference between the two bursting cell types relates to

the function of Ca currents. In the more-commonly-encountered nonregenerative

bursting cells, Ca currents play a relatively minor role in generating the burst itself but

are critical in terminating the discharge (through Ca-activated K currents). Reducing Ca

currents in these cells increases excitability and enables the discharge to continue

throughout the stimulus. The time-limited nature of the burst response largely reflects the

recruitment of a large-amplitude AHP response mediated by these K channels. Similar

counterintuitive effects of Ca currents that function primarily to decrease excitability (by triggering AHP responses) have been reported in other neurons, including hippocampal pyramidal cells (Madison and Nicoll, 1982). Since relatively weak depolarizations (to ~ -

50 mV; Figure 2-10A) appear to trigger AHP responses in tubercle neurons, it is possible that Ca influx through low-threshold Ca currents may trigger the AHP response either directly or by amplifying weak stimuli to activate high-threshold Ca channels. Burst discharges in these neurons appear to be potentiated by subthreshold Na currents, consistent with recent work on bursting in hippocampal neurons (Yue et al., 2005), and are dampened by IH.

Regenerative bursting neurons resemble the classic bursting phenotype of CA3 pyramidal neurons, including the ability of brief depolarizations to trigger prolonged depolarizing envelopes (Wong and Prince, 1981). These depolarizing responses could be truncated by brief hyperpolarizing stimuli in both hippocampal (Wong and Prince, 1981)

57 and tubercle neurons (Figure 2-16B). Bursts were not observed after Ca currents were reduced with low Ca ACSF, suggesting that regenerative Ca currents contribute to the burst response. Calcium currents, including low-threshold Ca currents (Figure 2-17), play an important role in amplifying inputs and enable regenerative cells to respond to very weak depolarizing stimuli that would not evoke a response in nonregenerative bursting cells. The mechanism of burst termination also differs between the two types of bursting cells with Ca-activated K currents playing a less dominant role in truncating discharges in regenerative cells than in nonregenerative cells. Given the significant differences in the underlying mechanisms, the similarity in the overall burst discharges between regenerative and nonregenerative tubercle cells was surprising and may indicate a common underlying role for low-threshold Ca currents. The difference between regenerative and nonregenerative bursting cells may reflect differences in the density or distribution of these low-threshold Ca currents, as well as the contribution of other types of channels.

Many regenerative bursting cells also displayed pronounced plateau potentials that were apparent during long pauses between action potentials. This discharge pattern is reminiscent of recently described “silent plateau” responses in subthalamic neurons

(Kass and Mintz, 2006) and may reflect one or more stable “up” states generated by active conductances in the different soma-dendritic compartments. Since many “silent plateaus” in both tubercle cells (e.g., Figure 2-15) and subthalamic neurons (Kass and

Mintz, 2006) appeared to be terminated by a single action potential, it is possible that either the rapid depolarization or, more likely, the spike AHP can disrupt the local plateau potential.

58 Potential significance of intrinsic properties of tubercle neurons

RS and IF neurons are found in many diverse brain regions often associated with

excitatory principal cells and GABAergic interneurons, respectively, especially in

cortical areas (Shepherd, 2004). However, this correlation is not universal. In the

olfactory system, glumatergic mitral cells in the olfactory bulb discharge intermittently

(Chen and Shepherd, 1997; Balu et al., 2004). The pyramidal cell-like morphology of

spiny RS neurons, as well as their location in the DCL, is suggestive of an excitatory,

glutamatergic phenotype, paralleling the pyramidal cells found in the major cell layers of

the neighboring piriform cortex (Shepherd, 2004) while immunocytochemical methods

suggest that at least a subpopulation of DCL and MFL neurons are GABAergic (Gritti et al., 1993). Determination of which transmitter is released by the major tubercle cell types will likely require either paired recordings or more detailed co-labeling studies.

Of the three firing modes we identified in tubercle neurons, the graded bursting appears to be the most unusual and interesting. The pronounced short-term intrinsic refractory period we found for MFL nonregenerative bursting cells also was surprising and presents an obvious potential target for modulation by centrifugal afferents (Heimer

et al., 1985). Also surprising were the large differences in the pattern of responses to

slow phasic (sniffing-like) input among the different classes of tubercle neurons.

Nonregenerative bursting cells discharged only in response to the first or second phasic

input while the other types of tubercle neurons followed each slow, simulated EPSP

(sEPSP). Olfactory bulb mitral cells, by contrast, fail to respond to the first slow sEPSP

in a train, but then follow subsequent sEPSP reliably (Balu et al., 2004; Balu and

Strowbridge, 2007). The results from the present study suggest that different subtypes of

59 tubercle cells likely play very different roles in processing periodic olfactory input. This

difference is especially pronounced for IF and bursting cells which show an opposite

modulation in their responses during trains of sniffing-like excitatory input.

While morphology and intrinsic behavior can be used to define the major cell

types in a brain region, a functional understanding of this area also requires a description

of the synaptic connectivity between cell types. Unfortunately, the presumptive axon

appeared to be severed in most of our intracellular fills. This finding may help explain

our difficulty in activating tubercle neurons synaptically using extracellular stimulation

using cornal slices. A similar paucity of axonal labeling was reported following Golgi staining (Millhouse and Heimer, 1984). Additional work utilizing intracellular fills in

different slice orientations will likely be necessary to determine how bursting and IF cells are synaptically interconnected with other tubercle cells. Since MFL tubercle cells appear to integrate both olfactory information, through molecular layer synaptic inputs, and inputs from the basal ganglia and the , through synapses on dendrites within the MFL (Haberly and Price, 1978; Luskin and Price, 1983; Heimer et al., 1987), it is appealing to speculate that the these two classes of inputs may selectively activate different plateau potentials in MFL neurons. Using fast Ca imaging and focal stimulation methods (Balu et al., 2007), it should be possible to test whether the different

“up states” we record in bursting cells (e.g., Figure 2-15 and 2-16A) reflect regenerative currents localized within different dendritic compartments and if these intrinsic responses are regulated by centrifugal modulators such as dopamine.

60

Table 2-1 Membrane properties of tubercle neurons

61

Figure 2-1

62 Figure 2-1: Schematic diagram showing the ventral structures contained in tubercle slices

The tubercle is bounded by the nucleus of the horizontal limb of the diagonal band

(HDB), piriform cortex and the nucleus accumbens (Nuc. Accumbes). The lateral olfactory tract (LOT) is clearly visible in tubercle slices as it is cut through in cross section. The deep, multiform layer of the tubercle contains multiple tightly-packed clusters of small cells (Islands of Callela, IC) and linear bundles of neuronal processes called striatal bridges (Str. Bridges).

63

Figure 2-2

64 Figure 2-2: Graded responses in three types of olfactory tubercle neurons

(A) Responses to graded depolarizing current steps in three olfactory tubercle cells

(Regular Spiking, Intermittently Firing, and Bursting). The initial voltage in the episode is indicated (in mV) above each trace and the laminar location of the cell body (DCL or

MFL, in this study) below each trace. Bursting tubercle cells often develop near steady- state depolarizing plateau potentials.

(B) Plots of the instantaneous firing frequency (the reciprocal of the inter-spike interval;

Inst Freq) versus time for the three top-most responses in A.

65

Figure 2-3

66 Figure 2-3: Morphology of regular spiking tubercle neurons

(A) Neurolucidia reconstruction of a neurobiotin-filled regular spiking neuron in the dense cell layer (DCL) with a dendritic arborization that was oriented perpendicular to the DCL and entered both the molecular layer (ML) and the multiform layer (MFL).

(B) 2-Photon image of another regular spiking DCL neuron showing a pyramidal-shaped dendritic arborization and numerous dendritic spines. Image is a composite made from multiple maximum projection Z-stack compressions. Inset shows adapting discharge pattern recorded in this neuron to a 2-second depolarizing step. The ventral surface of the brain is at the bottom of both images.

67

Figure 2-4

68 Figure 2-4: Morphology of intermittently discharging and bursting tubercle neurons

(A) Neurolucidia reconstruction of two neurobiotin-filled MFL cells, one with a dendritic arborization oriented parallel to the DCL (A1) and the other largely oriented perpendicular to the DCL (A2).

(B) Montage of 2-photon images of an intermittently discharging MFL neuron with a long process that entered the nucleus accumbens.

(C) Montage of 2-photon images of a nonregenerative bursting neuron in the MFL.

Responses to depolarizing steps in the cells shown in B and C are shown in the insets.

Both cells shown in B and C are sparsely spiny though the there are spines evident on the long process that transverses a striatal bridge to enter the nucleus accumbens.

69

Figure 2-5

70 Figure 2-5: Correlations of burst properties with stimulus strength

Enlargements of responses of a nonregenerative bursting tubercle neuron to graded depolarizing steps. Burst firing occurs in an intermittent firing pattern rather than a regular firing pattern. The interspike interval (ISI) is not constant between action potentials. The duration of firing remains fairly constant with increasing stimuli but the frequency of firing increases as the stimulus amplitude increases.

71

Figure 2-6

72 Figure 2-6: Quantitative analysis of burst properties

(A) Plots of the relationship between instantaneous frequency (reciprocal of the first inter-spike interval), mean frequency through the burst, burst duration and the number of

spikes in the burst against step amplitude. Solid lines represent best linear fit with

correlation (R2) indicated within each trace. Dashed line indicates best linear fit within

the indicated subset of points.

(B) Summary of the correlation (R2) between the four measures presented in A and step

amplitude over a population of 11 nonregenerative bursting neurons.

73

Figure 2-7

74 Figure 2-7: Bursting cell response to depolarizing ramp stimuli

(A1) Bursting cells discharge near the peak of depolarizing ramp stimuli. Open arrowhead indicates plateau depolarization triggered by fastest ramp stimuli. Response of the same bursting cell to step stimuli from -70 mV shown in insert. Inset calibrations are 20 mV and 500 ms.

(A2) Enlargements of responses in A1 showing phase advance of burst response relative to the end of ramp stimuli (sold arrows). Open arrowhead indicates plateau depolarization triggered by fastest ramp stimuli. Responses aligned by first spike timing

(vertical dashed line).

75

Figure 2-8

76 Figure 2-8: Cs increases firing during depolarizing steps in bursting cells

Bath Application of CsCl (4mM) increases firing during depolarizing steps. However, depolarizing plateau potentials (indicated by grey rectangles) are not abolished by Cs in the same cell. Inset shows blockade of membrane potential sag following hyperpolarizing steps by Cs. Inset calibrations are 20 mV and 500 ms.

77

Figure 2-9

78 Figure 2-9: Low Ca ACSF increases firing in bursting cells

Reducing Ca currents by switching to a low (0.5mM) Ca / high (6 mM) Mg ACSF abolished the burst afterhyperpolarization and facilitated firing throughout most of the 2 sec duration step. Time (in min) after switch to low Ca ACSF indicated above each trace.

Solid arrow indicates onset of reduction in burst AHP at 2 min.

79

Figure 2-10

80 Figure 2-10: TTX does not block depolarizing plateau potentials

(A) Depolarizing plateau potentials do not require Na current. Responses to graded

depolarizing current steps before and after bath application of TTX (1 µM). Horizontal

dashed lines indicate peak amplitude of response in TTX at lowest two step amplitudes.

Duration of plateau potentials indicated by grey rectangles.

(B) Effect of low Ca/ high Mg ACSF on response in TTX in a nonregenerative bursting

cell. In TTX alone, the response of the bursting cell continues to show the distinctive

depolarization followed by hyperpolarization. In TTX and low Ca, the response of the

bursting cell is flattened at a membrane potential that is between the most depolarized

potential and the most hyperpolarized potential during the response to the stimulus in

TTX alone.

81

Figure 2-11

82 Figure 2-11: Reduction of input resistance during afterhyperpolarization of

bursting cells

Burst afterhyperpolarization is associated with a reduction in input resistance. Inset

shows superimposed burst AHP responses initiated at membrane potentials from -57

(bottom trace) to -103 mV (top trace). Burst AHP response reversed polarity -99 mV, consistent with activation of K current.

83

Figure 2-12

84 Figure 2-12: Comparison of response of regenerative and nonregenerative bursting

cells

(A) Responses to long- and short-duration depolarizing steps in regenerative bursting

tubercle cells. In regenerative bursting cells, a time-limited response is seen in a long

duration step. Brief (50 ms) depolarizing pulses triggered prolonged depolarization in

regenerative bursting cells.

(B) Responses to long- and short-duration depolarizing steps in nonregenerative bursting

tubercle cells. Brief (50ms) depolarizing pulses did not trigger any prolong depolarization as seen in regenerative bursting cells.

85

Figure 2-13

86 Figure 2-13: Differences in burst properties between regenerative and

nonregenerative bursting cells

Summary of difference in burst latency, efficiency (number of action potentials evoked / depolarizing step amplitude) and initial clustering ratio for 2 sec steps in regenerative and nonregenerative bursting cells. Horizontal arrow indicate mean initial clustering ratio for populations of regular-spiking, intermittent, and bursting (regenerative and nonregenerative) cells. * p<0.05; **p<0.01

87

Figure 2-14

88 Figure 2-14: Plateau potential evoked by graded steps.

Constant-amplitude plateau potential evoked by graded depolarizing steps in a regenerative bursting cell.

89

Figure 2-15

90 Figure 2-15: Membrane potential of plateaus are constant

Silent plateaus at constant membrane potentials (indicated by dashed horizontal line) in

response to graded amplitude depolarizing steps. Inset shows regenerative plateau potential (open arrow) triggered by a brief depolarizing step. Calibrations in inset are 20 mV and 50 ms.

91

Figure 2-16

92 Figure 2-16: Plateau potentials seen with brief stimuli

(A) Bursting tubercle cell with two plateau potentials. Three responses triggered by identical depolarizing pulses. The right trace shows one single stimulus producing a response with two plateau potentials, one before the action potential was fired and one after the action potential was fired. The AHP from the action potential most likely caused the membrane potential to switch to the more hyperpolarized plateau potential.

The two different plateau potentials seen in the right response are at similar membrane potentials as seen the in left and center responses.

(B) Brief hyperpolarizing pulses can terminate regenerative plateau response in bursting tubercle cells. After a small hyperpolarizing pulse, regenerative depolarization occurs to depolarize the cell to its previous plateau potential. A larger hyperpolarizing pulse will terminate the regenerative plateau response.

93

Figure 2-17

94 Figure 2-17: Low Ca abolishes plateau potentials

Effect of switching to a low Ca (0.5 mM) / high Mg (6mM) ACSF solution (“Low Ca”) in a regenerative bursting cell. Low calcium causes the regenerative plateau potential to no longer occur and reduces the number of action potentials fired. Times indicated are minutes after extracellular solution change.

95

Figure 2-18

96 Figure 2-18: Plasticity of burst responses in nonregenerative cells

(A) Responses to a second depolarizing step are abolished or diminished with inter-step intervals less than 10 sec in nonregenerative bursting cells. Dotted liens indicate resting membrane potential. Dashed line on top trace indicates membrane potential (-80mV)

immediately before the second step was applied. Top right trace demonstrates that dc hyperpolarization to -80 mV does not abolish burst response. Inter-step intervals

indicated below second step response on each trace.

(B) Regenerative bursting cells show little two-step plasticity at 3 sec inter-step intervals.

(C) Summary of two-step plasticity in regenerative (n=8; open bars) and nonregenerative

(n=12; filled bars) bursting cells. ** p<0.01.

97

Figure 2-19

98 Figure 2-19: Brief depolarizing steps suppresses later activity

(A) Brief (500 ms) duration depolarizing steps diminished burst responses evoked by 2

sec current steps. Even subthreshold responses (second trace from left) modulate burst

responses evoked 1 sec later.

(B) Inhibition of burst responses by depolarizing pre-pulses can be overcome by increasing step amplitudes. Larger step responses indicated by asterisk.

99

Figure 2-20

100 Figure 2-20: Responses of tubercle neurons to sniffing-like periodic input

(A) Responses of RS, intermittent, nonregenerative and regenerative bursting cells to a train of four alpha functions (100 ms tau, 400 ms interval) at three different intensities.

Regenerative bursting cells showed a very low threshold for triggering burst responses that spanned two alpha function stimuli. Response thresholds were greater in each of the other three cell types.

(B) Effect of low Ca / high Mg ACSF on threshold responses in a regenerative bursting cell.

(C) Summary of number of spikes evoked by alpha function in RS cells (n = 6), intermittent cells (n = 8) and bursting (n = 7 nonregenerative and 5 regenerative).

Discharges are plotted as percent of total number of action potentials triggered by all four alpha functions.

101

Chapter 3

Simulation of the principle currents of the nonregenerative bursting cell

of the rat olfactory tubercle

102 Introduction

The olfactory tubercle contains neurons with unique burst firing patterns where the number of action potentials in each burst is graded to the amount of stimulus. These bursts of action potentials followed by suppressed activity may play an important role in neural signaling. What types of currents underlie the behavior of the bursting cell? What is the distribution of these currents across the soma and dendrites? Can the firing pattern be reproduced through the combination of simulated currents? To further examine the possible mechanisms that lead to the bursting behavior of these unique cells, I constructed a model of the electrophysiological properties of the bursting neuron in the rodent olfactory tubercle.

Rodent olfactory tubercle cells have the interesting property of being able to generate a high frequency burst of action potentials after depolarization followed by a period where activity is suppressed. The unique electrophysiological features of individual tubercle neurons may be important in filtering signal transmission. Recordings have revealed a variety of neural signals in the olfactory tubercle, including firing patterns that are novel and differ from those reported in any other brain region. Two different types of cells recorded in the tubercle fire a number of action potentials concentrated in the initial 500 ms of a 2 second depolarizing step followed by a long hyperpolarization that began during the step response. These two cell types differ in their response to a short (50 ms or less) current step, the temporal timing of action potentials, and their responses to more intense stimuli.

One way to test the hypothesis concerning the contribution of various ionic currents present in the tubercle neuron to their electrophysiological properties is the

103 success with which an accurate numerical model of the properties of these different currents is able to reconstruct the electrophysiological behavior of the neuron.

Mathematical models of neurons are valuable because they allow one to see the possible effects after manipulation of a specific ionic current. Specific ionic currents can be increased, decreased, or eliminated.

One group of neurons, termed regenerative bursting cells, have a low threshold calcium current (T-type current) that triggers an action potential doublet. When these cells are given a short (50 ms or shorter) current step, they display a regenerative depolarization following the end of the step. If a more intense long step is injected, it is possible for the cell to fire action potentials after the initial burst of action potentials.

The second group, termed nonregenerative bursting cells, is the focus of this second study. These cells show no regenerative plateaus when given a short step. When the intensity of a long step is increased, the cells can not be induced to fire additional action potentials. The number of action potentials fired increased as the step amplitude increased but action potentials continued to discharge within the first 500 ms of the step.

The inter-spike interval (ISI) between action potentials decreased with lager step amplitudes and the frequency of discharges increased. The pattern of firing during burst discharges were intermittent and not tonic.

Burst type firing commonly seen in other neurons are generally all or none discharges and are caused by voltage-activated calcium currents channels (Wong &

Prince, 1978; Llinas, 1988; Huguenard, 1996). The burst firing of tubercle bursting cells appears to be due to some modulation that is dependent on the voltage potential of the cell, as opposed to being evoked by a large influx of calcium.

104

Methods and Results

Summary of Experimental Evidence

Ionic currents flow thorough the of neurons through ionic channels which act as macromolecular pores. There is a huge diversity of ionic channels, even within the channels for one particular type of current. Some channels are specific while others are not. Some are affected by voltage, some are affected by ligands, and some are affected by both ligands and voltage. Channels have different kinetics with specific activation rates, inactivation rates, and deactivation rates which alter the ion flow through these channels at different voltages or if various ligands are present. Because of the complexity of different channels, we choose instead to focus on specific currents. These currents are likely composed of the ionic flow of a particular ion through multiple types of channels.

Compared to previously published reports of neurons that discharge with burst firing, calcium appears to be less of a driving force for burst firing in nonregenerative bursting cells. In other neurons that fire a burst of action potentials, the action potentials generally are on a calcium spike such that a reduction of calcium reduces firing. In the tubercle nonregenerative bursting cell, the calcium currents are important in terminating the discharge through calcium-activated potassium currents. This calcium-activated potassium current limits the number of discharges by hyperpolarizing the membrane potential to below the threshold for action potentials. A reduction in calcium current increases the number of action potentials fired, presumably by decreasing the activation

105 of potassium current. Tubercle burst discharges are potentiated by subthreshold

current and dampened by IH.

We tested the role four types of currents play in modulating the intrinsic physiology of the olfactory tubercle cell: hyperpolarizing current (IH), calcium currents, calcium-activated potassium currents, and sodium current. The reason we focused on these four currents will be explained below.

The majority of tubercle bursting neurons exhibit membrane potential sag during

steady hyperpolziations, a characteristic of IH. The sag seen is due to the slow activation

of IH during a steady hyperpolarizing step. IH has slow deactivation kinetics which can enhance firing during the initial portion of a depolarizing step (Ghamari-Langroudi and

Bourque, 2000). Bath application of IH blocker Cs (Ghamari-Langroudi and Bourque,

1995) facilitated discharges in bursting cells as well as eliminated the sag during a steady hyperpolarizing step. Plateau potentials were still seen in Cs suggesting that these are not mediated by IH.

Calcium current plays a clear role in the plateau potential seen in bursting neurons. Reducing transmembrane calcium current by perfusion with a low calcium/ high magnesium extraceullar solution caused bursting cells to prolong the time interval of

discharges such that action potentials would fire throughout the 2 second step. The low

calcium/ high magnesium solution decreased the magnitude of the initial depolarization

and the potential of the plateau was decreased compared to control. However, the later

half of the current step was more depolarized in the low calcium/high magnesium

solution than in control. This suggests that a calcium current is partially responsible for

the initial depolarization and that a voltage dependent calcium channel plays a role in

106 terminating plateau potentials in nonregenerative bursting cells, most likely through a

calcium-activated potassium current.

Potassium currents are responsible for hyperpolarization of the membrane

potential. Potassium channels generate the fast afterhyperpolarization response at the end

of each action potential as well as at the end of the entire current step. When a high

potassium bath solution was applied (causing a decrease in the driving force of potassium

current), the slow AHP was reduced.

Sodium channels play an important role in the generation of action potentials.

Experiments with TTX showed that TTX abolished Na+ based action potentials and

diminished the underlying depolarizing plateau potential indicating a subthreshold

sodium current that contributes to steady state depolarizing plateaus.

These results suggest that calcium currents underlie the depolarizing plateau

response. Sodium current supports the depolarization for the plateau potential while IH

decreases excitability. The depolarizing plateau appears to be terminated by calcium-

activated potassium current.

Morphology of neurons

We have neurobiotin reconstructions and images from alexa filled neurons from

13 nonregenerative bursting neurons. Analysis from experimentally recorded neurons

revealed that bursting neurons had 1.7 +/- 0.1 spines / 10 µm and 2.5 +/- 0.3 primary

dendrites. The cell body area averaged 179.6 +/- 21.3 µm2 with measurements made after slices were fixed. The cell body measured during recording showed typical soma

107 area to be 211.8 +/0 10.0 µm2 (n=32) with a range of 113 µm to 320 µm. Figure 3-1 and

3-2 shows images of bursting cells filled with either neurobiotin or alexa-594.

The dendritic morphology was chosen as a simplistic summary of the dendritic branching of imaged neurons. This multi-compartmental model is composed of a soma and two dendrites, with each dendrite branching once as seen in Figure 3-3. Although this is an obvious simplification of the morphology of the tubercle neurons, it is representative of the average morphology. Since the spatial distributions of the various ionic currents are unknown, a more detailed morphology would not provide an increase in information that could be extracted from the model. Other models have used single compartments and still been able to replicate the basic electrophysiological properties suggesting that complicated dendritic morphology is not essential to modeling basic electrophysiological features (McCormick and Hugenard, 1992).

The soma of the model neuron is 24 µm x 15 µm with a cm of 1 and an axial resistance (Ra) of 150. The dimensions of the soma were chosen to approximate the dimension of the soma as seen in recorded cells. There are two primary dendrites that extend in opposite directions from the soma. Each dendrite branches once. Dendrite 1 is

200 µm in length and 1.5 µm in diameter before branching into two smaller dendrites of

150 µm in length and 0.8 µm in diameter, Dendrite 2 is 400 µm in length and 2 µm in diameter before it branches into two smaller dendrites of 100 µm in length and 1 µm in diameter.

The model neuron is a simplified version of real tubercle neurons. Real OT cells often have more branching than what is portrayed in the model. The length of the dendrites before and after branching in recorded cells is very variable. The average

108 number of processes actually visualized as leaving the soma is 3.11 +/- 1.76 and the

mode is 2. (n=9) The size of the soma in a real cell varies in both size and shape. Most

bursting cells have an ellipse shaped soma but some had soma bodies that were more

triangular in shape. The long axis of the soma ranged from 9 to 36 µm. While many of

the soma’s in recorded cells appear elliptical, some are more triangular in shape. The model soma is elliptical in shape and represents an average area of a recorded cell.

Implementation of each major current into the model

We utilized mathematical models of the voltage dependence and kinetics of several ionic currents in an attempt to simulate the basic electrophysiological properties of nonregenerative bursting neurons in the olfactory tubercle. The model incorporated 1) a fast Na+ current INafast, 2) a persistent, depolarization-activated Na current INap 3) a

mixed calcium current ICa 4) a calcium-activated K+ current ICaK 5) a slowly inactivating and depolarization-activated K current IKDR 6) a hyperpolarization-activated cation current , Ih and 7) a leak channel Ileak. The end result allowed a model that had the characteristic properties of a nonregenerative bursting cell as seen in Figure 3-4.

The effects of various ionic current on the eletrophysiological properties of bursting neurons were initially investigated through examining the effect of each current individually on the membrane potential. The model for several currents was based on data from recordings obtained from neurons from various regions of the brain in different animal species, not just tubercle cells. Although it is likely that there are differences among these ionic currents of hippocampal, cortical, dorsal lateral geniculate nucleus, thalamic, or sympathetic ganglion neurons, the inclusion of these currents in the present

109 model was necessary to make a better fit of the current clamp behavior of tubercle

neurons.

In many of the models, each current modeled was based on a channel controlled by at least one gate. All gates of an individual channel must be open for ion conduction through the channel. The gate opening and closing rates are voltage-dependent. Hodgkin and Huxley (1952) dynamics were used to predict the time-dependent changes of each gate. The time constant τ was determined as a function of voltage and the time- dependent changes for each gate at a given action potential was determined by knowing the voltage-dependent values of the gate∞ and τgate.

gatet = gate∞ = (gate∞ – gatet-1) exp (- ∆t/τgate)

The Hodgkin and Huxley nomenclature for Na current denotes m for the activation gate

and h for the inactivation gate. M and h are continuous variables between 0 and 1.

ĝ = mNh

Currents are obtained from Ohm’s law

I = ĝ * gmax * (E – Eeq)

where gmax is the maximum conductance, E is the membrane potential, and Eeq is the equilibrium (reversal) potential for the current.

To find the appropriate values for gmax and EEq, we must determine the value of

τm, τn, h∞ and m∞. Functions were derived to describe of τm, τh, h∞ and m∞ as a function of membrane potential based on experimental results. The standard Boltzman equation was used to describe both inactivation and activation:

110

Where V1/2 is the membrane potential in which half of the channels are activated, K is the

slope factor related to the balance of charge that must be displace to switch between

closed and permissive states, and N is a power factor that influences the amplitude-time

course of the current (e.g., such as the “delay” of activation in voltage clamp). The

power factor N may be thought of as the number of activation (or inactivation) gates,

where each gate must be open for the channel to be conductive. The rate constants for

activation and inactivation were assumed to be continuous functions of voltage.

Activation rates were obtained from best fitted Hogdkin and Huxley.

where τm, is the activation time constant, τh, is the inactivation time constant, N is the

power factor, I∞ represents the maximal current level that would be reached in the

absence of inactivation (τh = ∞). For inactivation, the rates were determined by fitting exponential current decay. Because closing of each channel requires only one of the activation gates to change state, the time constant of decay of the tail current (τtail) will be

1/N of that for deactivation of single m gates (τm).

IH

111 H Current was one of the first channels selected because of the ability to isolate

the effects of this current from other current. IH has the most dramatic effect on the

membrane potential during steady hyperpolarizing steps, where most other mechanisms

play a large role in the membrane potential during a depolarizing step.

I examined models published in previous papers that simulate the IH current. I

then tested various distributions of this mechanism in the model of the bursting cell. The

distributions tested comprised of inserting the channel into the soma only, soma and

proximal dendrites, soma and all dendrites, proximal and distal dendrites, or distal

dendrites only. I simulated a hyperpolarizing current step for each different mechanism

type. I then compared the results with experimental data collected from bursting cells.

Figure 3-5 shows the effect of one mechanism IH with different distributions across the

model cell. This was done for each different IH mechanism. Figure 3-6 shows an example of 5 different IH mechanisms to show their variability.

I found the McCormick and Huguenard (1992) IH model inserted to the soma only

to yield the results closest to actual experimental results. Their model was originally

designed for the lateral geniculate nucleus relay neuron. The IH model is based on

Hodgkin and Huxley channel dynamics and includes activation gates but no inactivation gates. The data for IH were obtained from neurons in adult guinea pig dorsal lateral geniculate nucleus using switched-single-electrode voltage-clamp recordings with sharp

o electrodes in slices maintained in vitro at 35.5 C. The time constant of activation of IH were fitted with single exponential functions and were modeled as a bell-shaped function:

τm = 1 / ([exp(-14.59 – 0.08 Vm) + exp (1.87 - 0.0701 Vm)].

m∞ = 1.0/(1+exp((v + 80 )/ 5.5))

112 τm= (1.0/(exp(-15.02 - 0.086*v)+exp(-1.5195 + 0.0701*v))) /rate_k

rate_k adjust for temperature

Figure 3-7 compares the model with traces from the recordings of three different

bursting cells. Figure 3-8 shows the effect of IH current by comparing the model with IH

to the same model after removing IH.

Calcium Channels

Calcium channels have two major roles, electrogenic and signaling. While calcium’s regulatory roles as an intracellular messenger is important, the model focuses on calcium’s electrogenic roles including the ability of calcium channels to shape regenerative action potentials. Calcium current has a clear role in the regenerative bursting cell but what role does it play in the nonregenerative bursting cell?

To find what voltage-gated calcium channels were involved, I tested a large number of L Type, N Type, P Type, and T Type calcium channels in a number of different distributions across the model buster cell. Initially, it appeared that T-type current might be responsible for the initial depolarization. When L current or N current was added to the cell, no initial depolarization with plateau potential was seen. However, when T-type current was added, an initial depolarization followed by a hyperpolarization was seen. The hyperpolarizaton was due to the deactivation of the T-type current. Figure

3-9 shows examples of different spatial distributions of added T, L and N current.

However, the addition of T current did not cause a the initial depolarization to have a plateau potential with the same duration as seen experimentally. I then experimented with the addition of L and N currents to a model that had T current. Figure

113 3-10 shows the effect of adding L and N current to the soma and dendrites with the T

current in different spatial distributions. Figure 3-11 shows the effect of L and N current

in the soma only and not the dendrites with the T current in varying distributions In search through various combinations of voltage gated calcium channels, I was unable to find a distribution of any type of combination of these channels that produced the same calcium-mediated depolarization as the tubercle bursting cell.

LVA Ca channels may open at voltages near RMP. LVA Ca channels usually

have rapid, voltage-dependent inactivation and are therefore not available when the cell is maintained at depolarized holding potentials. HVA Ca channels may lack rapid inactivation and can be recorded in isolation form LVA currents by starting from depolarized holding potentials. (Llinas and Yarom, 1981) Calcium current amplitudes peak in HVA channels at approximately +10 mV. (Fenwick et al., 1982) Recordings can be done with the resting membrane potential held at a depolarized voltage near -40 mV that causes LVA currents to be inactivated but HVA currents are no inactivated. Small voltage changes (less than 10 mV) would affect LVA channels but not HVA channels which require larger voltage changes to activate. The response of the neuron to stimuli that only effect LVA can be subtracted from stimuli that would activate both LVA and

HVA Ca currents to see the HVA Ca response. Another way to isolate HVA Ca currents is to block the LVA currents with specific drugs such as ω-phonetoxin-IIA, verapamil, omega-conotoxin GVIA, and omega-agatoxin-IVA.

If the initial depolarization was due to a LVA Ca channel, that the characteristic plateau potential would change shape as the cell is held at a more depolarized level.

Experimentally, this was not the case with bursting cells, the plateau potential did not

114 alter when the cell was held at different resting membrane potentials. As seen in Figure

3-12, the nonregenerative bursting cell can be held at resting membrane potentials ranging from -49 mV to -82 mV and still have the characteristic depolarizing plateau.

This differs from the regenerative bursting cell which when it is held at -55 mV, a large change is seen. T current has been found to be mostly inactivated at -50 mV (Carbone and Lux, 1984). Nonregenerative bursting cells found in the tubercle do not typically have a rebound spike following a hyperpolarizing step, another sign indicating a lack of a low threshold calcium current that is inactivated at depolarized membrane potentials

(Burlhis and Aghajanian, 1987).

Coexistence of several types of calcium channels seems to be predominant in most cells (Matteson and Armstrong 1986, Tsien et al 1987). Different HVA channels are classified for their voltage dependence, activation range, inactivation range, decay rate, deactivation rate, and pharmacology. However, attempts to define electrophysiological criteria for each calcium channel have proved to be challenging.

There exists a large number of calcium channel subtypes with properties so similar that it is difficult to distinguished them by the criteria developed so far (Snutch et al., 1990).

Ca channels my dramatically change their open-time distribution and will gate with very different kinetics, a phenomenon known as this is called mode switching. Drugs, hormones, or even patterns of depolarization can cause mode switching (Pietrobon and

Hess, 1990; Artalejo et al., 1990).

The calcium current in the olfactory tubercle is likely a combination of currents using several different channels. I was unable to replicate the current clamp of tubercle cells using previously published models of voltage-gated calcium channels. The

115 complexity of the interaction of these channels makes it difficult to find a number of

calcium channels that together yield the type of current fluctuations seen in experimentally. Therefore, I created a calcium current model that represents the summed effects of several calcium channels with the activation and inactivation rates necessary to

match the experimental data. The current is based on Hodgkin and Huxley dynamics.

Since I did not perform voltage-clamp recordings, the values for m∞ h∞, τm, and τh were chosen to best match the initial depolarizing plateau. The values were first set for a T current and then altered to match recorded data.

m∞= 1.0 / ( 1 + exp(-(v+40)/12) )

h∞ = 1.0 / ( 1 + exp((v+60)/5) )

τm = ( 30 + 1.0 / ( exp((v+22)/15) + exp(-(v+62)/10) ) ) / phi_m

τh = ( 500 + 1.0 / ( exp((v+44)/3) + exp(-(v+397)/52) ) ) / phi_h

phi_m and phi_h adjust for temperature differences.

The difference in m∞ was a small shift of the curve to the left. The difference only affected voltages that were more depolarized than -46 mV, and the difference was less than 10% from voltages between -40 mV and 10 mV. The h∞ was slightly shifted to

the right with changes effective between -25 mV and 2 mV.

The changes increased the duration of activation and inactivation. The τm activation differs from the T current created by Huguenard in a few ways; 1) activation

116 rates over all voltages was much higher, 2) the peak activation was shifted to the right to

a depolarized voltage, and 3) the difference between peak activation and low activation

was smaller (McCormick and Huguenard, 1992). The τh was changed such that the inactivation rates higher and the whole curve as shifted to the right. There was also a

20% decrease in the difference between peak rates and low rates. Both the activation and inactivation rates were increased in the model presented here compared to Huguenard’s

T-current model causing channels to activate and inactivate faster over all voltages.

T current from Huguenard

m_inf = 1.0 / ( 1 + exp(-(v+52)/7.4) )

h_inf = 1.0 / ( 1 + exp((v+80)/5.0) )

tau_m = ( 3 + 1.0 / ( exp((v+27)/10) + exp(-(v+102)/15) ) ) / phi_m

tau_h = ( 85 + 1.0 / ( exp((v+48)/4) + exp(-(v+407)/50) ) ) / phi_h

Potassium Channels

There are a large number of different potassium currents that have been found, with some cells having more than five different potassium currents. These include voltage-gated K currents, calcium-gated K currents, leak K currents, resting K currents, and receptor-operated K currents. In this model, three different potassium currents were used, a delayed rectifier, a calcium-activated potassium channel, and a M current.

We used the KDR delayed rectifier potassium channel published by Traub et al.

(1991) in his model of a CA3 hippocampal pyramidal neuron. Traub based the current on

117 experimental data from Sah et al (1988) which were done on adult guinea pig hippocampus neurons done at 22-24 degrees Celsius. Traub then scaled the kinetics of the channel to the recordings done in his lab. This current does not inactivate. KDR current helps shape the shape of the action potential, mainly affecting its width and peak amplitude.

Traub uses the Hodgkin-Huxley equations, defining rate functions

m∞= α/(α+β)

τm = 1/(α+β)

α = (0.016(35.1 – V))/(exp((35.1 – V)/5)-1)

β = 0.25 * exp ((20-V)/40)

We tested a number of published models of calcium-activated potassium channels and eventually chose to use a model created by Miyasho (2001) which was based on work presented by Deschutter and Bower (1994). In the model presented by Deschutter and Bower, the voltage threshold, speed of the kinetics, and the half-activation calcium concentration was altered compared to a previous described calcium-activated potassium channel to produce results in their model that matched experimental data. Miyasho, then used Deschutter and Bower’s mathematical model for their calcium-activated potassium channel and made alterations until they found results that matched the experimental data from the tubercle. We used one of Miyasho’s models with one change made where we made the z∞ smaller.

118

m∞ = 25/(25+ 0.075/exp((v+5)/10))

h∞ = 1/(1+ 5/(ca*1000))

τm = (1 - exp(-dt*(25+b)))

τh= (1 - exp(-dt/10))

dt = time interval step

The effect of ICaK can clearly be seen when ICaK is taken out of the model. Figure

3-13 shows the increased duration of discharges when ICaK is not present to suppress activity. At lower intensities of current injection, ICaK suppresses any discharge. At current injections large enough to cause a discharge in the model with ICaK, the model

without ICaK will fire throughout the step. The model suggests that ICaK provides a role in

suppressing activity after the initial burst of action potentials. However, it does not

appear that ICaK is necessary for the depolarizing plateau potential which is still seen in the model after ICaK is taken out.

M current is a voltage-dependent potassium current that is modulated by muscarinic Ach receptors (Brown, 1980). IM has been found in found in hippocampal

neurons, sympathetic ganglia, and olfactory cortex. It is activated at voltages near the

action potential threshold voltage and plays a dominant role in regulating neuronal

excitability. IM activates and deactivates slowly and mono-exponentially and does not inactivate (Shahidullah, 2005). The classical role of IM is to produce spike frequency adaptation and it seems to be a main factor in medium AHP following a single spike or spike burst. IM will attenuate depolarizing inputs, especially those of long duration. In

119 hippocampal pyramidal cells, IM underlies the early phase of spike frequency adaptation.

(Madison and Nicoll, 1984)

During a steady hyperpolarization, IM will turn off and reduce the magnitude of

hyperpolarization, contributing to sag. M current has been found to be modulated by

Ach, a neurotransmitter which is abundant in the tubercle. M current can be inhibited by

acetylcholine and other muscarinic agonists) and has been found to be reduced by

. (Dutar and Nicoll, 1988; Colino and Halliwell, 1987)

The model for the IM was based on recordings in bullfrong sympathetic ganglion

cells. (Yamada, Koch & Adams, 1989) The model was made by Arthur Howeling for

MyFirstNEURON. The current does not inactivate and uses Hodgkin and Huxley

dynamics.

τm = 1000.0/(3.3*(exp((v+35)/20)+exp(-(v+35)/20))) / tadj

m∞ = 1.0 / (1+exp(-(v+35)/10))

tadj allows for temperature adjustment

In the model, the addition of IM causes the duration of the firing to decrease. The effect, seen in Figure 3-14, is quite prominent and the addition of IM is necessary to prevent the duration of firing to become so long that it fires throughout the current step.

It also lowers the excitability of the cell at lower current steps. IM does not change the plateau after firing since a lower intensity steps, this plateau is seen even after IM is

removed.

120 The role of voltage-gated sodium channels is primarily the upstroke of the action

potential. There are no somatic spikes when this current is blocked by tetrodotoxin.

.(Linas and Sugimori 1980) The model of the persistent sodium channel and the fast

sodium channel was published by Dursteqitz et al (2000) and was based on recordings

from prefrontal pyramidal cells in layer 5 of the in rats. They use

Hodgkin and Huxley dynamics with activation and inactivation.

m∞ = αm /(αm+βm)

τm = 1/ (αm + βm)

h∞ = αh / (αh+βh)

τh = 1/ (αh + βh)

Persistent Na

αm = (-0.2816*va)/(-1+exp(-va/9.3))

βm = (0.2464*vb)/(-1+exp(vb/6))

αh = (2.8e-5+DA_alphahfactor)*(exp(-vc/4.0248))

βh = (0.02+DA_betahfactor)/(1+exp(-vd/148.2589))

Fast Na

αm = -0.2816*(-9.3 + va*0.5)

βm = 0.2464*(6 + vb*0.5)

αh =0.098*(20 + vc*0.5)

121 βh =1.4*(10 + vd*0.5)

The application of TTX was modeled by the removal of sodium current. Sodium

current is necessary for action potentials as well as depolarization at the subthreshold

level. This is further discussed in the section on comparing the model to experimental

results.

A leak channel was created to maintain the resting membrane potential by

compensating for subthreshold inward currents. This leak channel does not have any

temperature dependence.

Calcium buffering properties in the tubercle neurons are not known. We used a

model of calcium diffusion that allowed for calcium ion accumulation with radial and

longitudinal diffusion. To model the radial diffusion of calcium, we used a series of 3

concentric shells around a cylindrical core. The outermost shell is half as thick as the

other compartments. This technique is described in The NEURON book by Carnevale

and Hines (2005).

The soma area was determined by the actual morphology found from Alexa filled

neurons. Intracellular recordings from tubercle cells were obtained at 30oC. However, when the cell was modeled at 30oC, the kinetics of the channels were faster than what

was seen experimentally. This was solved by setting the temperature in the model to

20oC. The extracellular concentration of calcium used in slice experiments was 2.5 mM and therefore was adopted here in an attempt to model results obtained from intracellular recordings in slices

122 Model Response

The model produced a neuron with a firing pattern consisting of an initial depolarization followed by hyperpolarization. The model fires with a 250 pA stimulus in current clamp. When a larger current is injected, more AP’s fire until ~400 pA are injected. At the 400 pA, the number of action potentials decrease. This is likely do to sodium channel inactivation where the channels do not close and are not available to reopen and fire another action potential.

The simulations show regular firing of action potentials during the firing interval.

This is likely do to the “ideal environment” of the model. The model lacks normal perturbations that are normal while performing whole cell patch clamp experiments.

Other models that show intermittent type of firing generally add some sort of perturubation to change the baseline membrane potential irrespective of the current step.

(Rubin and Cleland, 2006)

Simulations were then run using similar circumstances as used in actual recordings to compare the response of simulation in different test environments to actual recordings. Simulations were done in a low calcium environment, an environment representing the addition of TTX, and an environment representing the addition of Cs.

The model was based on the response to two second long steps. We then looked at the response of the model to short steps and two pulse steps and compared the results with recordings.

The model was run in a low calcium ACSF. Although experimental recordings used a low calcium ACSF of [Ca2+] = 0.5 mM, this concentration produced little change

123 in the model. To decrease the calcium current to a magnitude large enough to see a

difference in the traces, setting [Ca2+] = 0.1 was necessary.

When bursting cells were placed in low calcium ACSF, the number of action potentials fired increased and the duration of firing also increased. The envelope of the trace would flatten and lose the distinctive initial depolarizing plase followed by hyperpolarization. When the model cell was simulated in a low calcium ACSF, less activity was seen (Figure 3-15). While in normal ACSF, the model cell fired action potentials with a 250 pA step, no action potentials were fired when the model cell was simulated in low calcium ACSF. Action potentials were fired at 300 pA step but the number of action potentials was decreased compared to control. This decrease was still seen a 350 pA and 400 pA until firing in both the control and low calcium continued throughout the step at 450 pA. This was opposite of the expected behavior of the model.

In the simulation, TTX was modeled by the removal of the sodium channels.

When TTX is applied to bursting cells experimentally, the bursting cells still show the classic envelope of an initial depolarization followed by hyperpolarization. The simulation continued to show an initial depolarization followed by hyperpolarization in a similar manner as seen in experimental results (Figure 3-16). The cell did not depolarize as much in TTX compared to control in the model. This shows the sodium component of subthreshold depolarization, which is also seen experimentally (Figure 3-17).

Experimentally, TTX also reduces the amount of hyperpolarization after the initial depolarizing plateau potential. This pattern is not seen in the simulations.

When both TTX and low calcium ACSF are applied experimentally, the envelope flattens to appear more like the step stimulus. The membrane potential during the step is

124 between the depolarized membrane potential of the initial depolarization and the

membrane potential after hyperpolarization during the current step. The slow AHP seen

after the current step also decreases in magnitude suggesting that calcium-activated potassium current mediates this response.

The same conditions were tested in the model with different results than was seen

experimentally (Figure 3-18). The addition of the low calcium ACSF and TTX caused

the initial depolarization to be less than in TTX alone. However, after the initial

depolarization, there was still some hyperpolarization with the membrane potential at the

end of the step of the cell in TTX and low Ca ACSF being the same as the membrane

potential in TTX alone. In experiments, TTX tended to keep the cell more depolarized

after the plateau potential compared to control but this is not seen in the simulations. In

the simulations, TTX tended to decrease the duration of the depolarizing plateau.

When Cs, which is known to block IH, is applied to a bursting cell, the cell

becomes more excitable and facilitated discharges of action potentials. The cell will fire

action potentials with less current injected compared to control as well as fire more action

potentials with the same current injected compared to control. The mean number of

action potentials evoked in Cs = 224.7 ± 61.7 % of control (significantly greater than

control; p < 0.05; n = 6 cells). The cell continues to show an initial depolarization

followed by hyperpolarization. In some recordings of bursting cells in a Cs bath, a large

stimulus will cause the cell to fire throughout the step with more AP’s at the beginning of

the step.

Application of Cs was simulated in the model by the removal of IH. However, Cs

also blocks other potassium channels and those effects would not be reflected in the

125 simulations. The results that are seen in the simulation mirror the results seen

experimentally (Figure 3-19). With no IH, AP’s will fire with a 200 pA stimulus compared to zero action potentials fired in control conditions. The number of discharges in the simulation at a higher stimulus increases with the subtraction of IH and the duration

of firing is longer. At 250 pA, there are 5 spikes in control conditions compared to 7

spikes without IH. At 300 pA, there are 12 spikes in control conditions compared to 15

spikes without IH and the duration of discharges is 15% longer compare to control.

Two different types of cells which fire a burst of action potentials were found in

the olfactory tubercle, the regenerative bursting cell and the nonregenerative bursting

cell. The regenerative bursting cell preferentially fired at the beginning of a current step,

often with a doublet of action potentials. Unlike nonregenerative bursting cells, regenerative bursting cells would to fire throughout the step if the magnitude of the stimulus was increased. Regenerative bursting cells also showed a regenerative plateau potential to short (50 msec or less) steps. Nonregenerative bursting cells did not display any regenerative plateau when tested with a short step. When the model is given a short current step, no regenerative plateau is seen, which confirms that the parameters for the model are specific for the nonregenerative bursting cell and not the regenerative bursting cell (Figure 3-20).

After a nonregenerative bursting cell fires, it suppresses further activity for a period of time greater than 10 seconds. In recordings of actual cells, if a second current injection is done 3 seconds after an initial injection, no action potentials will fire. If the second injection is delayed to 4 or 5 seconds, some action potentials will fire during the second step but less than the control situation. (4 AP’s fired after a 3 second delay, 5

126 AP’s fired after a 5 second delay, and 5 AP’s fired after an 8 second delay.) Full recovery of firing occurs after approximately 10 seconds.

The model was tested to see if this two pulse inhibition would be reproduced.

After a one second delay, no action potentials were seen during the second step (Figure 3-

20). Action potentials fired after a 3 and 5 second delay but the number and duration was

less than control. Even after an 8 sec delay, the number of discharges was less than half

of control. This decrease in firing is not due to a lower membrane potential from the

slow AHP since a cell held to the same lower membrane potential will fire more action

potentials compared to if a current step was made within the 10 seconds prior. The slow

ADP of the model is does not have the same magnitude of experimentally recorded cells,

indicating that more potassium current maybe be involved experimentally.

In experiments, nonregenerative bursting cells were also tested with a short (500 ms) depolarizing step before a long 2 second depolarizing step. Even if no action potentials were fired during this short step, activity in the long step was suppressed evidenced by less discharges. Experimentally, if action potentials were fired during the short step, no action potentials were fired during the long 2 second step after a delay.

This behavior was only partially replicated in the model a seen in Figure 3-21. If

a short step preceded a long step and no discharges occurred during the short step, the

following long step will fire normally. In the model, if there is no depolarization

preceding the 2 second step, 13 AP’s are fired. If there is a short step with no AP’s, then

8 AP’s fired during the 2 second step following the short step. As the small step becomes

larger in magnitude, the number of discharges during the long step does moderately

decrease to 5 action potentials compared to control (13 AP’s). However, the number of

127 discharges during the long step does not change from 5 AP’s regardless if there are 4, 9 or 17 action potentials fired during the short step.

The results suggest that while the calcium-activated potassium current used in the

simulations does explain a decrease in activity following firing, it is likely that other

potassium currents are involved that were not represented in the model. The smaller

magnitude of the slow AHP in the model compared to experimental results is consistent

with this idea. Figure 3-22 shows simulations run with the calcium-activated potassium

current taken out of the model resulting in no suppression of activity is seen during the

second stimulus.

Discussion

In this study, we created a neural model to replicate the nonregenerative bursting

cell of the olfactory tubercle. First, we summarized the experimental evidence and

specifically discussed the properties of the bursting cell that we wished to replicate. We

examined the morphology of bursting cells and determined what general features should

be represented in the model. Then we analyzed each of the potential currents that might

play a role in the firing pattern of the bursting cell. Finally, we ran simulations of the

model cell in different environmental conditions including low calcium, TTX, low

calcium and TTX, and Cs. We also looked at simulations with paired-pulses and

compared results with that seen in recordings of actual cells.

Comparison of simulations to actual recorded data

128 Many aspects of the model fit well with the experimental data. The simulation of

graded steps closely matches what is seen experimentally. The model showed results to

short steps similar to that of nonregenerative bursting cells and not that of regenerative bursting cells. The model with the removal of IH to represent Cs in the simulation

showed effects that correlated with what is seen when Cs is applied experimentally.

Several aspects of the model did not replicate the experimental data well. One of the most notable differences is the comparison of experimental recordings of the cell in low calcium ACSF compared to simulations with the model cell in a low calcium environment. Instead of an increase in firing as seen experimentally, the simulations showed a decrease in firing. This is most likely due to less depolarization initially with the depolarizing plateau having a larger effect than the non-activation of calcium- activated potassium channels in the simulation. In experimental data, the calcium- activated potassium channel may play a larger role than what is accounted for in the simulations. The decrease in calcium in the simulation did not seem to result in a large change in the effect of the calcium-activated potassium current. This also would explain the simulation results of the two pulse experiments.

The simulations of TTX were more varied in their success of matching experimental results. While the simulation showed similar traces to what is seen experimentally, upon close examination, differences are found. TTX did not decrease the duration of the depolarizing plateau as was seen in the simulations. It did, however, change the overall shape of the trace throughout the step such that initially, it was less

depolarized while later, it was more depolarized compared to control, not including spike

discharges. TTX in simulations seems to have only affected the initial depolarization and

129 not later responses. All spike discharges were eliminated in both experiments and the

simulation.

The timing of the action potentials in the model is very regular with interspike

intervals being the same between each spike for a given step. While some adaptation is

seen, most firing in simulations is tonic with consistent ISIs. Recordings show an intermittent pattern of firing with varying ISIs between each discharge. The model is numerical and regular in nature whereas the baseline in recordings is constantly perturbed with noise in the signal. Other models that show intermittent firing by adding

subthreshold oscillations to the baseline (Rubin and Cleland, 2006).

Experimentally, most bursting cells will not fire throughout a 2 second step in

control conditions, even when injected with very large amounts of currents. In the

model, if enough current is added, the model cell will fire action potentials throughout the

step. Most models have the limitation of working only within a range of stimuli. This

model is no exception. In a living cell, the cell may respond in a sigmoidal fashion such

that at the extremes, the cell responds differently. The model implemented is more

linear in design and will continue on even at points where such a large current injection

would “kill” living cells. While recording, living cells will become “leaky” over time

causing different behaviors than a mathematically model could emulate.

Future experiments to address weaknesses of the model

The model suggests that some combination of calcium channels can generate a

depolarizing plateau which may result in a burst of discharge. There are still a variety of

calcium channels where the kinetics remains unknown. The calcium current in the

130 nonregenerative bursting neuron could not be modeled using current published kinetics

on calcium channels. Therefore, it is possible that the kinetics of the calcium channels in

these neurons is one that has yet to be studied.

Current clamp behavior is complicated because it shows the resulting behavior of

a combination of many different currents from a number of channels what have different

spatial distributions through the cell. The current created for the model represents the

functional consequence of the summation of all the calcium currents. In modeling, this is

done often. Robert Traub set a precedent when he combined a variety of potassium

currents into one current for his model of the cortical neuron (Traub, 1991).

The calcium current could by studied using voltage clamp. In voltage clamp, the

ionic currents are measured while the voltage is varied in a controlled manner. Since

voltage determines the activity of most channels, voltage clamp studies could elucidate

the activity of the calcium channels in these neurons. In current clamp, the situation is far

more complex because the membrane potential is allowed to vary spontaneously along

with the membrane currents, while only the current injected into the cell is controlled.

Since all experiments on bursting cells were done in current clamp, our knowledge of

specific currents at different voltages is limited. Because we have been unable to identify

a bursting cell by morphology alone , cells would need to be recorded in current clamp to

characterize the type of cell before switching to voltage clamp.

Inhibition, resulting from a short depolarization, that lasts for many seconds was clearly seen in the nonregenerative bursting cell. The mechanism that causes this response in the cell most likely plays an important role in neural signaling within the tubercle and the to brain structure that the tubercle projects to. There are several potential

131 avenues that may cause this long inhibition. It may be that system of calcium buffering

or calcium diffusion within these neurons keeps calcium-activated potassium channels

active for longer than is typically seen in other cells. Another possibility is that there is a new type of calcium-activated potassium channel that has long lasting affects, possibly due to lower inactivation or deactivation than seen in currently studied calcium-activated potassium channels. It is also possible that a calcium signaling cascade, or another signaling cascade caused by depolarization, is responsible for the effect.

Future experiments could be done to differentiate between these mechanisms.

The rise of internal free calcium could be detected by measuring the absorbance changes of a calcium indicator dye injected into the cell. Voltage clamp studies could be done to attempt to study the calcium-activated potassium channel. Transgenic mice with calcium-activated potassium channels knocked out or otherwise altered could be recorded from to see the difference in the electrophysiology of the nonregenerative bursting cell.

132

Figure 3-1

133 Figure 3-1: Morphology of bursting tubercle neurons

(A) Neurolucida reconstruction of neurobotin-filled MFl bursting neurons. Neurons

were filled with neurobiotin during recording. After recording, brain slices were fixed

and stained to reveal the neurobiotin injected cell. Reconstructions were made using

Neurolucida. The top reconstruction shows atypical morphology compared to other bursting cells by having many dendrites originating from the soma and shorter dendrites.

The bottom reconstruction shows more typical morphology with two dendrites originating from the soma that each branch twice.

(B) Photographs of MFL bursting neurons filled with Alexa-594. Neurons were filled with Alexxa-594 during recording. After recording, brain slices were transferred to a dissection microscope with a zenon light and camera in order to take photographs of the filled neurons. Neurons appear to have few, if any, spines. Neurons have two or three dendrites originating from the soma which have few branches.

134

Figure 3-2

135 Figure 3-2: Montage of 2-photon images of a nonregenerative bursting neuron in the

MFL

Neurons were filled with Alexa-594 during recording. Multiple overlapping images were scanned using the 2-photon laser and fitted together. Note the similarities in morphology of the bursting cell. Most neurons appear to have 2 or 3 dendrites which branch once or twice. Dendrites are sparsely spiny or apsiny. However, the top rightmost image indicates that there are some bursting neurons that have more dendritic branching.

136

Figure 3-3

137 Figure 3-3: Schematic diagram of morphology of neuron model

The soma has two dendrites, each dendrite branches once. The left dendrite is 200 µm in length before branching into two smaller dendrites of 150 µm in length. The right dendrite is 400 µm in length before it branches into two smaller dendrites of 100 µm in length. The soma is and ellipse that measures 24 µm x 15 µm.

138

Figure 3-4

139 Figure 3-4: Correlations of burst properties with stimulus strength

A) Responses to graded depolarizing current steps in the nonregenerative bursting

olfactory tubercle cell in experimental recordings.

(B1) Responses to graded depolarizing current steps in the simulation of the olfactory tubercle nonregenerative bursting cell with all the currents present. Note that the simulation shows the characteristic depolarization followed by hyperpolarization as seen in experimental recordings. After the initial bursting of action potentials, the simulation also shows the depolarized plateau potential.

(B2) Enlargement of responses for simulation of model neuron to graded depolarizing steps. The largest step stimulus produced a smaller number of action potentials than the second largest stimulus, something not typically seen in experimental recordings. The initial voltage in the episode is indicated (in mV) above each trace. The initial voltage in each simulation is -70 mV.

140

Figure 3-5

141 Figure 3-5: One mathematical model of IH with different distributions across model

cell

A single current can impact the overall responses depending on where the current is

distributed over the cell. One mathematical model of hyperpolarizing current was

inserted in only the soma, the soma and dendrites, only the dendrites, only the distal

dendrites, and in the soma and proximal dendrites. The different distributions caused

different rates of depolarization during a hyperpolarizing step. The initial voltage of each

simulation is -70 mV. Each trace indicates the locations of IH for that specific trace. The bottom right compares the traces by overlapping them. Variability in the rate of depolarization can be affected by the location of the current.

142

Figure 3-6

143 Figure 3-6: Five different mathematical models of IH

There exist several different published mathematical models for hyperpolarizing current,

five of which are shown here. The currents vary in the magnitude of response to the

hyperpolarizing step, the rate of depolarization, and whether or not action potentials fired

during the rebound. The initial voltage of each simulation is -70 mV. Each trace indicates a different mathematical model of IH. The bottom right compares the traces by

overlapping them. There is some variability in the magnitude and rate of depolarization

between different mathematical models of the same current. The spatial distributions of

the current were the same in all simulations

144

Figure 3-7

145 Figure 3-7 Comparison of three experimentally recorded cells with the simulation

The top row shows the response of three nonregenerative bursting cells to a depolarizing step. The initial voltage is indicated above each trace in mV. The bottom row shows the response of the same cells to a hyperpolarizing step. While all three nonregenerative bursting cells showed depolarization during the time course of a hyperpolarizing step, the rate of depolarization had some variance. The bottom right compares the three hyperpolarizing step traces with that of the simulation.

146

Figure 3-8

147 Figure 3-8: Comparison of the control simulation to the simulation after removal of

IH

Application of CsCl is modeled by the removal of IH resulting in increased firing during depolarizing steps. The left column shows the graded response of the simulation in control conditions. The right column shows the graded response of the simulation with the IH current removed. Removal of IH increases the excitability and the number of action potentials fired. Depolarizing plateau potentials are not abolished by removal of

IH. Inset calibrations are 20 mV and 1 sec.

148

Figure 3-9

149 Figure 3-9: Different Spatial distributions of T, L and N current in the model neuron

The left most column shows the simulation run with the T current in varying locations and no L or N current. The middle column shows the simulation run with the L current in varying distributions and no T or N current. The right most column shows the

simulation run with the N current in varying distributions and no T or L current. The

diagram shows the distribution of the current. For example, in the left most column, the

T current in the soma only is in red, the T current in the dendrites only is in green and the

T current throughout the soma and dendrites is in blue. No T current is in black. The

addition of T current appears to be necessary for an initial depolarizing peak.

150

Figure 3-10

151 Figure 3-10 Variation in the distribution of T current with L and N current in the

soma and dendrites

L current and N current are in both the soma and the dendrites. No T current is in black, the T current in the soma only is in red, the T current in the dendrites only is in green and the T current throughout the soma and dendrites is in blue. T current appears to play a

role in the initial depolarization. The apparent hyperpolarization is due to T-current

inactivation.

152

Figure 3-11

153 Figure 3-11: Variation in the magnitude of T current

A comparison of four simulations of the model with the magnitude of T current increased. T current was in the soma only while the L current and N current were in the dendrites only. Increasing the magnitude of T current depolarized the resting membrane potential and resulted in a depolarization in the initial peak during a current stimulus, but it did not dramatically increase the duration of the peak.

154

Figure 3-12

155 Figure 3-12: Nonregenerative Bursting Cells at Different Membrane Potentials

Each row represents one neuron that was given a current step stimulus at different resting membrane potentials. The amplitude of the stimulus varies between traces. The initial voltage in each episode indicated (in mV) above each trace. The characteristic burst of discharges followed by a hyperpolarization is seen at membrane potentials between -49 mV and -90 mV indicating that the response is not voltage dependent. This suggests that currents necessary to produce the bursting firing pattern do not inactivate at resting membrane potentials between -50 mV and -90 mV.

156

Figure 3-13

157 Figure 3-13: Graded response of the simulation run in control conditions and run

with the removal of calcium-activated potassium current

The left column shows the simulation in control conditions with all currents present. The

right column shows the simulation with the removal of ICaK. The removal of ICaK causes the model neuron to fire more action potentials with a smaller magnitude current injection. Without ICaK, the cell will fire throughout the step with the same stimulus that caused the control to reach the threshold for firing. Removal of ICaK did not remove the hyperpolarization after the initial depolarized plateau and the depolarized plateau is still observed with small current stimuli.

158

Figure 3-14

159 Figure 3-14: Graded response of the simulation run in control conditions and run with the removal of M current

The left column shows the simulation in control conditions with all currents present. The right column shows the simulation with the removal of M current. The removal of M current causes the model neuron to fire more action potentials. Action potentials are fired with a smaller threshold of current stimuli. The depolarized plateau is still observed after the burst of action potentials with small magnitude current stimuli.

160

161 Figure 3-15: Effect of a low calcium environment on the model cell

Graded response of the simulation run in control conditions and in low calcium conditions. In experiments, calcium currents were reduced by switching to a low (0.5 mM) Ca / high (6 mM) Mg ACSF which caused facilitated firing throughout most of the

2 sec step. Simulations with a calcium environment of 0.5 mM showed no macroscopic differences compared to the control calcium environment of 2.5 mM. The calcium environment was reduced to 0.1 mM in order to see the effect of a low calcium environment. In simulations, a low calcium environment reduced firing, requiring a larger current amplitude to induce firing and with firing durations shorter than in control conditions.

162

Figure 3-16

163 Figure 3-16: Response of model to simulated condition of the addition of TTX

(A) Graded response of a nonregenerative bursting cell in control conditions is shown in

black. Red indicates the graded responses at the same stimulus for the cell after the

application of TTX. The cell is initially less depolarized in TTX but then hyperpolarizes

less than control conditions after the initial depolarizing plateau.

(B) Application of TTX is modeled by the removal of fast sodium current and persistent

sodium current. Graded responses of the model cell in control conditions is show in black. To simulate TTX, the sodium currents were removed from the model. The TTX simulations are shown in red. The initial depolarization is less than control conditions and the depolarizing plateau is shorter than in control conditions. Depolarizing plateau potentials do not require Na current.

164 TTX

Figure 3-17

165 Figure 3-17: Comparison of the effects of TTX on recorded cells vs. the simulated model cell

Graded response of nonregenerative bursting cell in TTX shown in black. Graded responses of the model simulated in a TTX environment in red. At smaller depolarization, the simulation is less depolarized than the experimental data. At larger depolarization, the initial depolarizing plateau is similar but the simulation remains more depolarized than experimental results after the plateau.

166

Figure 3-18

167 Figure 3-18: Response of model to TTX compared to TTX and Low Ca ACSF

(A) Experimental results of a nonregenerative bursting cell in TTX (black) and in TTX and low Ca ACSF (blue). The addition of low Ca ACSF causes the depolarizing plateau to flatten. When the two traces are overlapped, it can be seen that the amplitude of the initial depolarizing plateau is decreased and the amplitude of the later hyperpolarization is also decreased to give a more depolarized level. The slow AHP is also decreased in low Ca ACSF.

(B) Simulations without sodium current (black) and simulations without sodium current

simulated in a low Ca environment (blue). The initial depolarizing plateau is decreased

but not completely blocked in the simulation. The slow AHP is not affected by the low

Ca environment.

168

Figure 3-19

169

Figure 3-19: Response of model to brief stimuli

Responses to a very brief (50 ms) stimulus do not show any depolarization or depolarizing plateau potential extending after the end of the stimulus in simulations. This indicates that the model is representative of the nonregenerative bursting cell and not the

regenerative bursting cell.

170

Figure 3-20

171 Figure 3-20: 2 Inhibition of firing following a long stimulus

Responses to a second depolarizing step are abolished with an inter-step interval of 1 sec and diminished with inter-step intervals of 2 to 8 sec in the simulation. Dashed lines indicate resting membrane potential. Dotted line indicates membrane potential immediately (mV) before the second step was applied. Top right trace demonstrates that hyperpolarization to -72 mV does not abolish the burst response. The suppression of activity in the second step is due to the activity in the first step.

172

Figure 3-21

173 Figure 3-21: Inhibition of firing following a brief stimulus

Brief (500 ms) duration depolarizing steps diminished burst responses evoked in 2 sec current steps with an inter-step interval of 1 sec. When the brief first step was small in amplitude with no discharges, there was little effect on the longer second step. When the brief first step showed subthreshold response or discharges, the longer second step fired fewer discharges.

174

Figure 3-22

175 Figure 3-22: Loss of suppression of firing with loss of CaK current

(A) Response to a second depolarizing step is unchanged with an inter-step interval of 1

second with the removal of calcium-activated potassium current. This suggests that

calcium-activated potassium current is part of the mechanism that causes suppression of activity after a burst of action potentials are fired.

(B) Brief (500 ms) duration depolarizing steps did not diminished burst responses evoked

in 2 second current steps with an inter-step interval of 1 second with the removal of

calcium-activated potassium current. This suggests that subthreshold depolarization is

enough to cause calcium-activated potassium current to suppress further activity in

experimental recordings.

176

Chapter 4

Discussion

177 In this thesis, I set out to determine the intrinsic properties of the major cell types of the olfactory tubercle neurons. I found three basic firing patterns of neurons in the dense cell layer and multiform layer: regular-spiking, intermittently-discharging, and bursting firing. I characterized the intrinsic properties of these neurons, including the unusual bursting cells, which are further subdivided into regenerative bursting cells and nonregenerative bursting cells. I also examined the dendritic morphology of these cells.

In addition, I modulated calcium currents using different external solutions to differentiate between the two types of bursting cells. I used drugs like cesium and TTX to learn more about the currents that play a role in producing bursting firing. I then used the information to develop a model, using the NEURON software system, of the nonregenerative bursting cell. The model allowed manipulation of currents involved in order to reproduce the behavior seen experimentally.

Two mechanisms of generating burst firing

The regenerative and nonregenerative bursting cells show that there are at least two different mechanisms for generating burst firing in tubercle neurons. Both mechanisms appear to use calcium but in different ways. The regenerative bursting neuron seems to be more similar to the bursting cells found in the hippocampus pyramidal neurons. The depolarization appears to be due to calcium current through voltage-gated calcium channels. Removal of calcium causes a decrease in activity and causes the regenerative depolarizing plateau to disappear. The regenerative depolarizing plateau is recovered upon the re-addition of the calcium to solution. This suggests that

178 these cells have calcium channels that open when the cell depolarizes, and calcium current will continue for several milliseconds after the stimulus.

The nonregenerative bursting cells react quite differently to a low calcium solution. Instead of decreased activity, these cells show an increase of activity with more spiking at regular intervals instead of a burst of action potentials. I believe this response is due to a strong calcium-activated potassium current in these neurons. The initial calcium partially contributes to the depolarization immediately following depolarization due to a current step. This calcium then causes the opening of potassium channels which hyperpolarize the cell. These calcium-activated potassium channels appear to have a long lasting effect for several seconds.

Role of calcium in bursting cells

In both types of bursting cells, the burst of firing causes calcium entry into the cell. This calcium entry may have a variety of second messenger functions as well as opening calcium-activated potassium channels. The types of calcium channels that allow calcium entry into tubercle bursting cells have not yet been studied. My research focused on the electrophysiology of tubercle cells and what role their activity may play in the neural circuitry of the tubercle. I did not investigate the subtypes of the calcium channels that are responsible for the firing activity, since this was outside the scope of my thesis work. However, this is a potential area of future interest that may be studied by those interested in channel kinetics and activity.

In hippocampal CA1 pyramidal neurons, repetitive synaptic activity induces an influx of calcium entry that is necessary for long-term potentiation (LTP), a form of

179 longer term plasticity. It is thought that the main influx of calcium is through NMDA

receptors on postsynaptic spines (Sabatini et al., 2001). In granule cells, repetitive

stimulation causes calcium entry and long-lasting depolarization mediated by protein

kinase A. In the cerebellum, long term depression (LTD) is induced when both parallel

fiber synapses and synapses on the Purkinje cell are simultaneously activated (Miyata et al., 2000). Activation of mGluRs on the spines of the parallel-fiber-

Purkinje cell synapse leads to IP3 production. The IP3 produced binds to receptors on the smooth ER causing release of intracellular calcium, which is, in some synapses, necessary and sufficient to trigger LTD (Miyata et al., 2000). These examples show that there are many signaling cascades involving calcium and ion channels that can produce long lasting effects to neural circuitry.

When creating the model for the nonregenerative bursting cell, I first attempted to use the channel kinetics of previously published models on different types of calcium currents. I found that these calcium currents had kinetics that were too fast compared to the 200-500 ms of the initial depolarizing plateau seen in bursting cells. In order to

produce a model of a neuron that behaved like the recorded cell, I had to create a calcium

current. This created calcium current could represent either a novel calcium channel type

that has yet to be described or studied, or more likely, it could represent a complex

combination of existing channels. Calcium channels have been found in different states

and may have different dynamics when in different states. The number of possible

combinations of multiple calcium channels, each varying between states, is too large to

find the correct combination by trial and error. Another variable is the spatial distribution

of these multiple types of calcium channels. The simulation suggested that the majority

180 of currents were located in the soma. However, the burst activity may initiate in the

dendrites, and the examples of where dendritic calcium flux plays an important role in

signaling has already been mentioned.

Calcium imaging could help determine the location of spike initiation by

detecting where the greater change in calcium current occurs. Calcium imaging could be

done with local calcium channel blockers directed at different sites on the dendrite and soma. One can test for which LVA and HVA channels may be located at different dendritic departments by using specific blockers for these channels and looking at the change in calcium flux through calcium imaging. For example, if a LVA blocker was used on a dendrite and there was no change in calcium entry into the cell, then it is likely that LVA channels play a less prominent role in that part of the dendrite. However, if a

HVA blocker was used on the same site and a large decrease in calcium entry was seen, it is likely that HVA channels are present at that location on the dendrite and play a role in calcium signaling.

The burst firing is terminated by hyperpolarizing after potential (AHP), mediated

by one or more calcium-dependent potassium currents. A calcium-activated potassium

channel was necessary to add to the model neuron for the AHP seen in experimental

recordings. However, I believe that the calcium-activated potassium current in the

nonregenerative bursting cell has a larger effect on the overall membrane potential in the

neuron than seen in the model. While the model did have some changes in the two pulse

experiments, showing possible plasticity, the results were not as evident as in

experimental recordings. In experimental recordings, reduced activity was seen for

periods of greater than 10 seconds following an initial burst, and the amount of reduction

181 was graded over time. After 3 seconds, no activity was elicited; after 4 seconds, one

action potential was elicited; and after 5 seconds, 2 action potentials were elicited. In the

model, there were four action potentials elicited after a 3 second delay instead of the 13

action potentials elicited in control conditions. After a 5 second and 8 second delay, 5

action potentials were elicited.

Experimental recordings also showed that a short depolarization small enough to

not elicit any action potentials still subsequently reduced activity. This may be due to a

calcium current entering the cell at a lower membrane potential than is required for

sodium channel activation leading to an action potential. If a second stimulus was large

enough, action potentials could still be elicited, demonstrating that the reduced activity is

unlikely due to inactivation of sodium channels, though modulation of sodium channels

could still be involved. Performing these plasticity experiments with a calcium dye

inside the neuron in order to measure the amount of calcium current into the cell would

give insight into the role calcium plays.

There is still much to be learned about the calcium currents in the tubercle. An

important future study would be to describe the currents generating bursts and after potentials based on voltage-clamp data. Unfortunately, there is no way to determine if a cell is a nonregenerative bursting neuron without patch clamp recording. Therefore, to do voltage clamp recordings on these cells would require first current clamp recording from them and then switching to voltage clamp. This requires loading the pipette with current clamp solution in the tip followed by voltage clamp. Then, one needs to be able to patch the cell, characterize it in the few minutes before the internal solution switches,

182 and then successfully switch from current clamp to voltage clamp. While technically

possible, this would be a difficult set of experiments to perform.

There are likely many channels involved in the generation of a burst of firing, and

these channels are likely gated by multiple factors besides membrane potential.

Understanding the variables that change the states of different ionic channels would lead to a better understanding of what effects impact cell signaling. I believe for people studying calcium channels, that the olfactory tubercle may be an area that would provide

novel calcium channel kinetics. The tubercle may also be a good candidate for studying

calcium-binding proteins or calcium-activated signaling cascades.

Another important aspect to study is how localized ionic currents result in a

spatial and temporal distribution of membrane potential. Does the calcium depolarization

arise in the dendrites or the soma? Is there amplification of the depolarization in the

dendrites or the soma? What is the mechanism of the amplification? Do action

potentials initiate in the dendrite or does most of the depolarization move down the

dendrite to the soma before an action potential can be formed? How does input from

other cells affect calcium depolarization? Where are calcium-activated potassium

channels located? Are they spread throughout the cell or do they shunt the depolarization

at the soma? How does input from other cells affect the hyperpolarization caused by

calcium-activated potassium current?

Comparing the tubercle bursting cell to the hippocampal bursting cell

Bursting cells are often associated with hippocampal pyramidal cells, which fire

bursts of up to eight spikes at intervals of 5-10 ms (Wong and Prince, 1981). These bursts

183 consist of a series of action potentials with one or more slow calcium-mediated action

potentials and lasts for 30-50 ms. The burst is followed by a long AHP that may last for

about 1 second. In the cells, triggering a burst is much like triggering an action potential

in other cells, in that it is an all-or none event. The cell either fires a burst or does not fire

at all. The burst represents a form of amplification, both in total membrane current and

duration. Hippocampal bursting cells tend to fire a burst of action potentials separated

by longer intervals with reduced firing.

How do the burst firing of olfactory tubercle neurons compare to hippocampal

pyramidal neurons? In the case of the nonregenerative bursting cell, these cells will fire a

single action potential (more often two action potentials) and do not always fire a group of 5-8 spikes. The nonregenerative bursting cell fires graded bursts, with an increased number of spikes associated with a larger stimulus. This burst firing is not an all-or-none event and therefore does not represent the same form of amplification as burst firing in hippocampal cells. The bursts are followed by a long AHP that lasts for several seconds, much longer than that seen in hippocampal cells. The activity in the nonregenerative

bursting cell is followed by a long interval of reduced firing of over 10 seconds. This

AHP has implications for longer term plasticity for large time periods. The suppression

of firing blocks the ability to transmit an electrical signal to the postsynaptic neuron. The

length of the suppression of firing is something that might be modulated by

neurotransmitters, , or hormones. This modulation could lead into

changing the neural transmission of the signal to induce a different motor behavior.

On the other hand, regenerative bursting cells tend to fire a doublet where the

inter-spike interval between the first and second action potential is relatively constant. A

184 larger stimulus will produce more action potentials, so like the nonregenerative tubercle bursting cells, regenerative tubercle bursting cells are graded bursters. When given a large enough current, action potentials will continue to fire even after the initial burst, a characteristic of regenerative bursting cells but not nonregenerative bursting cells. With a short stimulus, the regenerative bursting cell may continue to stay depolarized and fire additional action potentials, thus acting to amplify the original stimulus. The AHP is much smaller with regenerative bursting cells, such that the firing of the initial burst does not suppress later activity. A regenerative bursting cell can repeatedly fire and does not show the reduced firing seen in hippocampal pyramidal cells and nonregenerative tubercle cells.

Function of bursting activity and plasticity

Burst firing allows for amplification of signal. If the bursting cell synapses onto a cell with a time constant longer than the inter-spike interval during the burst, then the

EPSPs will summate. A single EPSP may be subthreshold for triggering an action potential in the postsynaptic cell, but the summation of multiple EPSPs due to burst firing may elicit activity. When given phasic input that resembles sniffing, regenerative bursting cells had an elongated pronounced response, an example of amplification of signal.

The nonregenerative bursting cells discharged only in response to the first response, which can act to block any elongated or latent response to a signal stimulus.

This kind of action may narrow the timeframe where action potentials are allowed to signal the next targeted cell. In this way, burst firing with reduced activity between

185 bursts may also be used as a filter. Once the nonregenerative bursting cell fires, there is

reduced activity for several seconds afterwards. Therefore, even if there is incoming

activity from cells connected to the bursting cell, the bursting cell will not fire and will

not transmit a signal to the cells it synapses onto. Edge detectors like those found in the

retina may benefit from this type of filtering of information (Maturana and Frenk, 1963).

The nonregenerative bursting cell is a good edge detector, as seen with the experiments

where the current was injected gradually like a ramp instead of a step. The burst of

activity occurred at the edge, followed by reduced activity.

The bursting cell is a potential target for modulation. Experiments have shown

that changing the calcium concentration in the ACSF changes the firing pattern in

bursting cells. We have already discussed how calcium can play an important role in

signaling cascades. Modulating the calcium channels of the bursting cells could

modulate the firing of bursting cells.

The tubercle as part of the ventral striatum

The striatum in most is divided by the internal capsule into two areas: the dorsomedial caudate nucleus and the ventrolateral putamen. Cell bridges that penetrate the internal capsule connect the caudate nucleus and putamen. There are no structural or connectional criteria which can serve to entirely distinguish the nucleus accumbens from the striatum. The olfactory tubercle is separated from the nucleus accumbens by the boundary line created by the fiber layer of olfactory radiations.

Despite this boundary, the nucleus accumbens and tubercle are connected by cell bridges.

186 Low powered views of the Nissl stained sections show small to medium sized

cells in both the nucleus accumbens and the olfactory tubercle multiform layer. The

principle cell of the nucleus accumbens and the striatum is the medium spiny cell

(O’Donnell and Grace, 1993). The striatal medium spiny cell has dendritic morphology

that is very similar to the tubercle medium spiny cell, with the difference being that the

tubercle dendrites are more varied in the pattern of arborization and the number of

dendrites. The multiform layer of the olfactory tubercle contains the islands of Calleja

which are not seen in the more dorsal nucleus accumbens. The tubercle also contains a

cell layer of densely packed cell bodies, which is not found anywhere in striatum

(Millhouse and Heimer, 1984). While there are many similarities, there are still some

obvious differences between the striatum and the olfactory tubercle.

Both the nucleus accumbens and caudate putamen receive afferents from a variety of diverse brain regions making it difficult for a principle afferent relationship. Most afferents to the nucleus accumbens originate from structures within the .

Most sensory, motor, limbic, and association areas of the endocortex project to the caudate putamen. Likewise, the olfactory tubercle also receives afferents from a range of brain areas, including limbic areas like the prefrontal cortex, sensory areas like the olfactory bulb, and memory areas like the hippocampus.

Circuitry of the tubercle

One of the main objectives of my research was to discover the intrinsic circuitry of the olfactory tubercle, including finding out which cells are the projection cells and which cells are the interneurons. Interneurons often have a fast-spiking firing pattern like

187 that seen in the medium spiny cells. However, the medium spiny cells have been labeled as the projection cells in striatal areas. In the cortex, the projection cells are the pyramidal cells. The medium spiny cells in the dense cell layer were once classified as pyramidal cells.

In the striatum, there are cells that are described as large cholinergic interneurons that are aspiny and have long dendrites with few branches (Bolam et al., 1984). Bursting cells tend to be large with fewer dendrites and little to no spines, with morphology similar

to the large cholinergic interneurons of the striatum. I would have predicted that the

bursting cells to be projection neurons, since the bursting pattern of firing appeared to be

a great mechanism for filtering information. However, if the olfactory tubercle mirrors

the structure of the striatum, the bursting cells may be interneurons that shape the output

of the medium spiny cells, and the medium spiny cells would be the projection neurons.

One of the limitations of working with slices is that processes are often cut off. In

our experiments, the axon was cut off such that we could not see where the axon

projected. We only have one example of a bursting cell that sent a dendrite to the nucleus

accumbens through the striatal bridges. Therefore, an important future study would be to

discover where the different types of cells project to in order to learn the intrinsic

circuitry of the tubercle. It is also important to learn which cells are projection cells and

what areas of the brain are the projection targets. This could be done through recording

and filling the cell using whole brain recording techniques. The tubercle’s location on

the ventral exterior of the brain makes it a perfect candidate for whole brain recording.

While the tubercle is thought to project to the ventral pallidum, there may be other

projections that are currently unknown (Millehouse and Heimer, 1984). Much of the

188 information I have gathered on the tubercle was found by studies that did not study the tubercle itself but rather studied the structure that the tubercle is connected with. A comprehensive study of the projections of the tubercle will be necessary to better understand the role the tubercle plays as a brain structure. A series of experiments where

HRP is injected in the tubercle and the charting of HRP labeled cells following injection should be done to find all the brain structures that project to the tubercle. HRP should be injected in several locations within the tubercle, and the resulting retrograde cell labeling should be visualized. The injection sites should divide the tubercle by layers: the multiform layer versus the DCL, as well as lateral tubercle versus medial tubercle and rostral tubercle versus caudal tubercle. One may find that the lateral tubercle projects to different sites than the medial tubercle. The dopaminergic input into the tubercle is greater in the medial side of the tubercle, and it is this region that is believed to be most involved in the reinforcing effects of cocaine (Ikemoto, 2005). There may be other surprising projections to the tubercle that have yet to be discovered. It was only recently found that the hippocampus projects to the tubercle. The variety of different brain structures that project to the tubercle indicate its role as encompassing more than olfactory processing.

Another large question posed in understanding the circuitry of the tubercle is to know what neurotransmitters are used by which neurons in the tubercle. What neurotransmitters do the projection cells and the interneurons use? What neurotransmitters do the regenerative and nonregenerative bursting cells use? Are they the same or different? Understanding what neurotransmitters are used by these neurons is necessary in understanding synaptic connectivity and plasticity. On the most basic

189 level, it is important to know whether a synapse is excitatory or inhibitory. This is

necessary to know whether activity in the presynaptic neuron causes greater or less

activity in the postsynaptic neuron. Knowing the neurotransmitter each type of neuron

uses will have a great impact on understanding the neural circuitry in the olfactory

tubercle. Learning what neurotransmitters are in the projection cells of the tubercle will

give a greater understanding about what role the tubercle serves in the brain as a whole.

Studying the neurotransmitters in the tubercle can be approached by a number of different methods. One method to learn more about neurotransmitters present in a brain region is to label for proteins or mRNA specific to a certain transmitter. A review of work that published stains for neurotransmitters was given in the introduction. The conclusion of that review was that there is still much to be understood about neurotransmitters in the tubercle. For example, the olfactory tubercle has been stained for GAD mRNA with 15-20% of cell bodies staining positive (Cloez, 1996). If the medium spiny cells of the tubercle were really the same as the medium spiny cells in the striatum, which are known to be GABAergic, one would expect a higher percentage of

OT cell bodies to stain for GAD. This result could be due to a large number of

GABAergic cells, which do no not stain positive. One way to test these results would be to stain coronal slices that contain both the striatum and the olfactory tubercle and compare the percentage of cell bodies stained. If a low percentage of cell bodies in the striatum stain GAD+, this indicates the possibility of many false negatives where

GABAergic cells did not stain GAD+.

Another way to determine if the medium spiny neurons are GABAergic would be to do a combined electrophysiological recording and immunohistochemistry approach.

190 Cells could be injected with an antibody to GAD that fluoresces only if it binds to its

target. If the cell is GAD+, it should begin to florescence during the recording period.

This can be monitored with 2-photon microscopy to measure the amount of florescence,

as well as the spatial distribution of florescence. Such an antibody may not yet exist but I

believe that future technology will make such an antibody a reality.

The most conclusive way to find what neurotransmitters are used by interneurons

is to perform paired recordings between interneurons and their target cells. Once a pair is

found, specific blockers to different neurotransmitters can be applied until the blocker

that decreases the response significantly is found. Admittedly, this is a difficult

experiment to perform. However, the use of 2-photon imaging to fill a neuron and

follow the axon to its target makes this experiment a greater possibility. Other labs have

done experiments to characterize the pharmacological profile of postsynaptic responses to determine which neurotransmitter is used at each synapse in other brain areas (Dugue et al., 2005).

The tubercle is a site of action for anti-psychotics and cocaine addiction

The function of the olfactory tubercle remains a mystery. There is data to link the olfactory tubercle with the site of action for the therapeutic effect of anti-psychotics as well as the site of action for cocaine addiction. The role of the tubercle in these two

important clinical correlations should be more thoroughly examined. What synaptic

changes occur in the tubercle when an antipsychotic or cocaine is administered? There

are many approaches of how this could be studied. Recordings could be done while

or cocaine is administered in the bath solution.

191 Dopamine is known to be part of the mechanism for cocaine addiction. If

dopamine receptors are blocked by injection of dopamine antagonists into the accumbens,

the rewarding effects of cocaine are reduced (Maldano et al., 1993; McGregor and

Roberts, 1993; Baker et al., 1998). Also, lesions of the dopaminergic terminals in the accumbens reduced the rewarding effects of systemic cocaine administration (Roberts et al., 1977, 1979). Since the olfactory tubercle has been found to be more involved than the accumbens in cocaine addiction, I believe similar results would be found after lesioning dopaminergic terminals of the tubercle or injection of dopamine antagonists to the tubercle. It is possible that the dopaminergic lesions of the accumbens would show the effect of decreased cocaine addiction because such lesions would also lesion the dopaminergic projections that go through the accumbens to reach the tubercle. I predict that if the olfactory tubercle was lesioned with 6-OHDA, then rats would no longer show signs of cocaine addiction like conditioned place preference and two lever discrimination.

To further study the role of dopamine receptors in cocaine addiction, an experiment that could be performed is to inject dopamine antagonists specific to either D1, D2, or D3 receptors in the tubercle while administering local injection of cocaine into the tubercle and looking for the effects of cocaine. Since the tubercle is one of the few areas of the brain that is rich in D3 receptors, this may elucidate how D3 receptors differ from D1 and

D2 receptors.

Less is known about the therapeutic mechanism of anti-psychotics. Anti- psychotic drugs (also called neuroleptic drugs) are used to treat , manic states, and delirium. There are two main modes of action for antipsychotic drugs: drugs that block the activity of dopamine receptors in the brain and drugs that block the activity

192 of serotonin receptors in the brain. Since the tubercle contains both dopamine receptors and serotonin receptors, the therapeutic effects could be from activity at either or a combination of both types of receptors (Cloez-Tayarani, 1995; Mijnster et al., 1997;

Voorn et al., 1986; Zazpe et al., 1994).

Evidence that anti-psychotics produce changes in the tubercle have been found in a number of studies. Levels of D2 receptors in rat olfactory tubercle showed the most significant difference after chronic haloperidol treatment of all brain areas tested (Joyce,

2001). Amisulpride increases dopamine release in the rat olfactory tubercle (Scatton et al, 1997). D3 mRNA levels were found to increase in the tubercle after administration of haloperidol, pimozide, and sulpiride (Wang et al., 1996). Another study showed that risperidone, prothipendyl, ORG 5222, sertindole, and olanzapine all preferentially acted in the tubercle and not the nucleus accumbens (Cools et al., 1995). Would anti- psychotics still have a therapeutic effect in animals where the olfactory tubercle is completely lesioned or where the dopaminergic input was lesioned? The antipsychotic clozapine was no longer effective after 6-OHDA destroyed the dopaminergic nerve endings in the nucleus accumbens (Huff, 1980). It is possible that destroying the dopaminergic nerve ending in the accumbens also destroyed the dopaminergic fibers traveling to the tubercle. I predict that if the tubercle dopaminergic fibers were destroyed with 6-OHDA, then antipsychotics would become less effective indicating that the tubercle is a site of action for these antipsychotics.

The homologue of the olfactory tubercle is called the anterior perforated substance. The anterior perforated substance is a small region of gray matter located immediately posterior to the gyrus rectus. The many holes in this region are from the

193 large number of small, penetrating that transverse the section of the brain.

Dopamine concentrations in the anterior perforated substance were elevated by 95% in a group of 50 schizophrenic patients compared with controls (Bird et al, 1979). Although the human homologue of the olfactory tubercle is small, the role antipsychotic play in the tubercle may still have large relevance to the treatment of schizophrenia in humans.

The tubercle is a sensory, limbic, and motor structure

I believe that the tubercle integrates olfactory input and the limbic system to affect motor behavior. Its role in the rewarding effects of cocaine as well as it being the site of action for antipsychotics suggest that the tubercle plays a role beyond olfaction.

Injections of cocaine directly into the tubercle induce motor activity, including locomotion and rearing, emphasizing the tubercle as a motor area (Ikemoto, 2002).

Lesions to the VTA of rats lead to a number of motor behavioral changes including an increase in horizontal locomotor activity, a decrease in exploratory behavior, and the disappearance of freezing reaction (Tassin et al., 1978). 6-OHDA lesions to rat nucleus accumbens decrease spontaneous motor activity. I believe that the dopaminergic input from the VTA to both the nucleus accumbens and the olfactory tubercle modulates motor behavior.

The limbic system is associated with , , and emotional association with memory. Traditional limbic structures include the amygdala, hippocampus, and hypothalamus. I believe that the olfactory tubercle will become associated with the limbic system as the cingulated gyrus, prefrontal cortex, and nucleus accumbens have become. While the nucleus accumbens has received attention for its

194 involvement in reward, , and addiction, I assert that the olfactory tubercle also

deserves recognition for its involvement in these same areas.

The tubercle will be found to be involved in integrating olfactory sensation with

positive or negative associations. Future studies will also show that the olfactory tubercle

is the brain area where association is made between odor stimulants (including food and

sex ) and the behavior elicited by those odor stimulants. For example, the

tubercle will be involved in the detection of a food odor, as well as to the positive

association of food that will cause motor behavior to follow the odor and find the food.

The tubercle is an understudied brain structure that plays an important role in affecting the behavior of animals that depend on their olfactory .

195

Chapter 5

Future Directions

196 Circuitry of the olfactory tubercle

Projection fibers from the olfactory bulb generally terminate in the outer lamina

of the molecular layer of olfactory cortical areas (Luskin and Price, 1983). Although the

olfactory tubercle may not be a cortical area, it still has a three layered structure like

olfactory cortical areas. The olfactory tubercle’s outer molecular layer is relatively free

of cell bodies and is mostly composed of fibers much like layer 1 of the piriform cortex.

The dense cell layer of the olfactory tubercle parallels the pyramidal layer (layer 2) of the

neighboring piriform cortex and they often connect. The multiform layer of the tubercle corresponds roughly with layer 3 of piriform cortex. It is likely that the projection fibers from the olfactory bulb terminate in the molecular layer of the olfactory tubercle (Price,

1973).

The neurons in the dense cell layer (DCL) have dendrites that reach ventrally into the molecular layer and are likely to receive input from olfactory bulb projection fibers in the molecular layer. The majority of DCL neurons fire with the regular spiking pattern.

There are a minority of DCL neurons that fire intermittently. There are some neurons in the multiform layer which have longer dendrites that also reach into the molecular layer while many other multiform layer cells do not have dendrites that extend to the molecular layer. The dense cell layer neurons also have dendrites extending dorsally into the multiform layer where they are likely to synapse with cells found in the multiform layer.

A complex system of associated fibers interconnects the different olfactory cortical areas (Luskin and Price, 1983). Luskin and Price (1983) assert that “these fibers arise from all parts of olfactory cortex except the olfactory tubercle,” which differentiates the tubercle from other olfactory cortical areas. The authors defined olfactory cortical

197 areas to include the olfactory tubercle, as well as other cytoarchitechtonically distinct

areas including the ventral tenia tecta, anterior cortical nucleus of the amygdale and the

. To review the findings from Luskin and Price, see Table 5-1.

The first column indicates where the retrograde label was injected. The second, third and forth column indicates the amount of labeling seen in each of the three layers of the olfactory tubercle.

When HRP was injected into the molecular layer of the olfactory tubercle, the only area labeled was the multiform layer of the tubercle. No other part of olfactory cortex showed any transported label after injection into the molecular layer of the tubercle. Injection of HRP in to the multiform layer transported label to the lateral hypothalmus, the ventral tegmental area and layer I of the anterior olfactory nucleus.

This research suggests that the tubercle receives projections in all three layers, with the majority of input from the olfactory bulb terminating in the molecular layer

(Luskin and Price, 1983). It is likely that these axons synapse onto the dendrites of medium spiny cells found in the dense cell layer, and possibly on the dendrites of a few cells whose cell bodies are found in the multiform layer. The projection cells of the tubercle are most likely in the multiform layer and are not the medium spiny cells found in the dense cell layer.

I hypothesize that input to the tubercle from other olfactory areas, like the olfactory bulb and piriform cortex, causes activity in the medium spiny cells of the dense cell layer. (Figure 5-2) These cells fire in either a regular pattern or an intermittent pattern to stimulate cells in the multiform layer. There are also some regular spiking cells found in the multiform layer which I believe are interneurons that connect only with other

198 tubercle neurons. The regular spiking cells that are found in the multiform layer may be neurons receiving input that comes to the multiform layer of the olfactory tubercle like input from the ventral tenia tecta, periamygdaloid cortex, nucleus of the LOT or the anterior cortical nucleus of the amygdale. The medium spiny cells of the dense cell layer and the multiform layer form synapse onto the bursting cells and intermittently firing cells. The bursting cells may be interneurons that mediate firing activity of the intermittently firing projection cells or it is possible that the bursting cells are the projection cells.

Role for bursting cells

What may be the functional role of the bursting firing pattern? Unlike bursting cells found in other areas of the brain, the non-regenerative bursting cells would not be good candidates to amplify a signal. Non-regenerative bursting cells fire one burst of action potentials and then suppress further activity for several seconds afterwards.

Therefore, non-regenerative bursting cells are good cells to filter a signal to a specific temporal window. The bursting cell may serve to amplify the initial input, and then suppress further input that follows that initial input. Burst firing could help to determine if two inputs arrive at the same time. If a stimulus appears and the neuron responds by continuous firing, then observing the current activity (and not past activity) would not distinguish between new stimuli or persistent stimuli. However, if a stimulus appears and the neuron fires a few times and then stops firing, than activity of the neuron signals the onset of the stimulus or the presence of new stimuli.

199 In this way, bursting cells may act as edge detectors. Edge detection is a concept that often focuses on visual images. One of the goals of edge detect is to mark the points where there is a sharp change in the brightness of stimuli, which often reflects some change in the what is being viewed. The sharp change could be due to a change in the material property like the edge of an object against a background. The ability to detect the edge can filter out less relevant data.

Edge detection via the olfactory tubercle bursting neurons may enable specific functions in the olfactory system. For example, if suddenly a new smell is detected, the non-regenerative burster cell may fire upon the novel detection in order to make the time point when the odor was first detected. It could also be a mechanism in order to sample the odor only every few seconds. Odor plumes are often several different molecules mixed together. As an animal moves toward the source of an odor, the composition of the molecules making up that the odor changes. It may be advantageous to take a sampling of the odor every 10 seconds instead of with every sniff so that the difference between the two samples is larger. If one is presented every step of a gradient, one may not detect the change over time. However, by taking samples every few seconds instead of every few hundred milliseconds, the change in time will be more robust.

If the olfactory tubercle is the interface between olfaction and motor behavior, it could be that the bursting cells filter olfactory input into motor output. When looking at the flight path of a moth following an odor, one can see that the moth regularly changes his flight direction back and forth (Willis and Aras, 1998). The suppression of activity from the bursting cell may be involved in signaling when a new sample of the odor

200 should be taken and the decision to switch directions should occur. The bursting tubercle

cells may determine when new stimuli should be processed and action taken.

Function of dopamine receptors and cocaine in the tubercle

The olfactory tubercle receives a dense dopaminergic input from the ventral tegmental area via the mesolimbicortical pathways. These pathways are thought to be

important for cognitive function and motivation (Willner and Scheel-Kruger, 1991).

Dopaminergic input to the striatum is through the nigrostrital pathway which originates in

the substantia nigra. The nigrostrital pathway is thought to be involved in extrapyramidal

motor function. Administration of psychostimulants like amphetamine and cocaine

increases dopamine release in mesolimbic areas whereas withdrawal of reduces dopamine

release.

In general, dopamine receptors are divided up into two families. D1 and D5

dopamine receptors are grouped into one family called the D1-like receptor types and D2,

D3 and D4 dopamine receptors are grouped into a second family called the D2-like

receptor types. D1-like receptor subtypes generally stimulate adenylate cyclase activity

whereas D2-like receptor subtypes inhibit adenylate cyclase activity. Little is known

about D5 receptors, partially because there are no D5-specific drugs.

The olfactory tubercle is one of the few areas of the rat brain that expresses D1,

D2, and D3 dopamine receptors. D3 dopamine receptor expression appears to be

preferentially expressed in the mesocorticolimbic system. D3 dopamine receptor mRNA

and protein expression are both found in the nucleus accumbens, the olfactory tubercle,

and the islands of Calleja. In general, protein expression and mRNA expression for D3

201 receptors are high colocalized, implying that the D3 receptors are primarily on the soma

and proximal dendrites (Levesque et al., 1992).

D3 receptors have been found to interact with both D1 and D2 receptors (Scarselli

et al., 2001; Zeng et al., 2006). D3 receptor stimulation inhibits locomotion. However,

when D2 and D2 dopamine receptors are activated at the same time, rodent locomotion is

stimulated (Dalia et al., 1998; Dreher and Jackson, 1989). D1 and D3 receptors are

frequently co-expressed (Schwartz et al., 1998) while D2 and D3 receptors less often

have overlapping expression (Bouthenet et al., 1991). D1 and D3 receptors have opposite

regulatory effects on ERK activity and Fos expression (Zhang et al., 2004)

D3 receptor occupation by an agonist appears to inhibit neuronal firing (Liu et al,

1994). One possibility is that D3 receptor activation causes activation of potassium

currents. If non-regenerative bursting cells have D3 receptors, it is possible that the

activation of D3 increases the ability of non-regenerative bursting cells to suppress firing

activity. During recording, there were a small group of cells that exhibited both

intermittent and non-regenerative bursting like qualities. During smaller stimuli, the cells

appeared to be non-regenerative bursting cells but the suppression of activity during the

later half of a long 2 second step could be overcome by a larger stimulus. The existence

of these cells shows the possibility of modulating the amount of suppression of activity

that occurs in a bursting cell after the initial burst of action potentials fire.

Kuzhikandathil and Oxford (1999) found that activation of human D3 receptors

inhibit P/Q type calcium channels but appear not to affect L type calcium channels. The

ability of D3 dopamine receptor to regulate the activity of different calcium channel

subtypes may allow D3 receptor to modulate synaptic transmission. Calcium channel

202 dynamics play an important role in the firing pattern of both regenerative and non- regenerative bursting cells which are unique to the olfactory tubercle. The activation of

D3 receptors on the bursting cells could alter the calcium current that flows through P/Q type calcium channels. If the regenerative bursting cells express P/Q type calcium channels, this may reduce the ability of regenerative calcium channels to amplify signal since it is likely that by inhibiting calcium channels, there would be less calcium current.

The effect of inhibiting P/Q type calcium channels in non-regenerative bursting cells is likely to be more complicated. Calcium currents are involved in the initial depolarization causing the burst of action potentials. However, it is possible that calcium-activated potassium channels also indirectly trigger the hyperpolarization during a long step stimulus and active suppression of activity following a burst of action potentials. Reduction of calcium current may decrease potassium current flow causing these cells to be more active. In this way, D3 dopamine receptor activation may cause non-regenerative bursting cells to fire more action potentials or to decrease the amount of suppression an inactivity seen after a burst of action potentials is fired.

In the nucleus accumbens, dopamine depresses inhibitory synaptic transmission by reducing calcium influx into the presynaptic terminal (Nicola and Malenka, 1997).

Since the nucleus accumbens and the olfactory tubercle appear to have many properties in common, it is possible that dopamine in the olfactory tubercle also depresses inhibitory synaptic transmission. The principal cell in the nucleus accumbens is the medium spiny cells which are Gabaergic. Dopamine may cause a decrease in medium spiny cell activity which would lift the amount of inhibition that these cells impose on their targets.

Although it is unknown if the medium spiny cells of the tubercle are Gabaergic, stains for

203 GAD show that there are Gabaergic neurons in the tubercle (Gritti et al., 1993). It is possible that the medium spiny cells of the tubercle are Gabaergic and inhibitory.

Reduction of calcium flux in the presynaptic terminal in these cells may also cause a decrease in activity and therefore, lift inhibition on the cells that the medium spiny cell targets onto. Calcium flux is generally depolarizing and causes the cell to be active. The decrease of calcium flux could potentially decrease the amount of action potentials fired and therefore, decrease synaptic transmission.

Nicola and Malenka (1997) also found that dopamine depresses excitatory transmission independently of calcium influx in the nucleus accumbens. D2 receptor agonists decrease the amplitude of EPSCs recorded from medium spiny neurons in the nucleus accumbens (Umemiya and Raymond, 1997). The decrease of excitatory transmission may be linked with the decrease in the amplitude of EPSCs. This could be due to activation of other potassium currents like M current. The model of the non- regenerative tubercle cell suggested that M current may be involved in limited the number of action potentials fired during a burst and the suppression of firing after the burst. An increase of potassium currents would lead to decreased excitatory transmission

if the membrane potential was hyperpolarized below threshold.

Research has suggested that the olfactory tubercle plays a role in mediating the

rewarding effects of cocaine (Ikemoto, 2003). There are multiple ways that the olfactory tubercle may be involved. The tubercle may be involved the reinforcing the feeling of pleasure with cocaine or in producing motor behavior to seek out cocaine due to the association of rewarding effects of cocaine. Due to the input from premotor areas and the connection of the tubercle with the ventral striatum, a motor area, it is like that the

204 tubercle is involved with initiating behavior to seek more pleasurable experiences like that of more cocaine administration.

Unlike many other illicit drugs, cocaine is often snorted instead of injected into the bloodstream. It is thought that cocaine is absorbed by the large network of blood capillaries under the nasal mucosa into the blood stream. The absorbed cocaine would then enter systemic circulation and be distributed throughout the brain. However, there is evidence that substances applied nasally may enter the brain directly from the nasal cavity through the olfactory system (Chow et al., 1999). When cocaine is applied nasally, plasma and blood concentrations of cocaine remain elevated for significantly longer than when cocaine is injected into the blood stream directly. Also, the tissue to plasma ratio of cocaine in the olfactory bulb is three times greater with nasal administration than intravenous administration (Chow et al, 1999).

It is possible that cocaine travels in the axons of mitral cells to other parts of the brain including the olfactory tubercle. Once cocaine is in the olfactory tubercle, it would increase the amount of extracellular dopamine which with then bind to dopamine receptors. What changes in the olfactory tubercle may occur due to the presence of cocaine and greater concentrations of extracellular dopamine?

Expression of to stimulants is believed to be a result of changes in the nucleus accumbens (Wolf, 1998). Initiation of sensitization has been found to involve the D1 dopamine receptor and activation within the ventral tegemntum and substantia nigra. Ikemoto’s (2003) research clarified that while the nucleus accumbens is likely the site of action for the rewarding effects of amphetamines, it is the olfactory tubercle that plays a role in the rewarding effect of cocaine. The

205 olfactory tubercle has D1 receptors and gluatamate receptors so a similar signaling cascade that occurs in the nucleus accumbens for amphetamines may occur in the

tubercle for cocaine.

However, I hypothesize that D3 receptor activation in the olfactory tubercle is

critical to cocaine addiction. Studies have linked cocaine addiction with the D3 receptor.

Xi and Garnder (2007) found that systemic administration of D3 specific receptor

antagonist NRB 2904 inhibits cocaine self administration. One injection of the D3

receptor antagonist blocked cocaine self-administration for 1-2 days after a single

injection. Dopamine antagonists injected into the olfactory tubercle also blocks the

rewarding effects of systemic cocaine administration (Ikemoto, 2003). It is possible that the action of blocking the D3 receptor may be enough to block cocaine addiction. D1 and D2 receptor activation may not be involved in cocaine addiction.

D3 receptors have a higher affinity for dopamine than D1 or D2 receptors

(Sokoloff et al, 1992). Therefore, when dopamine is first released in an area, the D3 receptors will be stimulated before D1 or D2 receptors. If there are low levels of dopamine, than it is possible that only D3 receptors will be activated and D1 and D2 receptors will remain quiescent. At higher levels of dopamine, a larger percentage of D3 receptors will be bound to dopamine compared to D1 or D2 receptors.

Cocaine administration increases dopamine levels for prolong periods (Jones et al,

1996). Dopamine levels found after cocaine administration showed an average concentration of 750 nM (Zetterstorm et al, 1983). At this concentration, the receptor occupancy was be 96% for D3 receptor, 25% for D1 and 27% for D2, reinforcing the much higher affinity for D3 receptor to dopamine than D1 or D2 receptors.

206 I believe future experiments may show that D3 receptors in the tubercle will be involved in mediating the physiological changes that lead to cocaine addition. D3 receptor activation will be found to cause change in calcium currents in the bursting cells of the tubercle and alter synaptic transmission. Inhibitory and excitatory transmission will both be affected to change the balance of neural activity in the olfactory tubercle.

Conclusion

Although input from the olfactory bulb to the tubercle occurs primarily in the molecular layer, the olfactory tubercle receives input from other olfactory areas in all three layers. Medium spiny cells are likely to be the main receiver of input in the tubercle. Intermittently firing cells or bursting cells are likely to be the projection cells of the tubercle. The bursting cells may serve to filter input in a specific temporal window and act as edge detectors. The dopaminergic input to the tubercle potentially mediates the rewarding effects of cocaine through the activation of D3 dopamine receptors. D3 dopamine receptors are likely to be located on the soma and proximal dendrites.

Activation of these receptors appears to inhibit neuronal firing, possibly through affecting specific calcium channels and potassium currents. Cocaine applied nasally may be carried through the axon of mitral cells to the olfactory tubercle to begin the signaling cascade that leads to addiction. The D3 receptors of the olfactory tubercle may serve as a novel target for drugs to treat cocaine addiction.

207

Area retrograde Molecular Layer DCL Multiform Layer label injected in Ventral pallidum moderately labeled unlabeled labeled LOT densely labeled lightly labeled lightly labeled piriform cortes densely labeled lightly labeled rarely labeled lateral entorhinal labeled on medial unlabeled labeled on medial areal side side ventral tenia tecta rarely labeled rarely labeled moderately labeled periamygdaloid unlabeled labeled moderately labeled cortex nucleus of the LOT moderately labeled moderately labeled moderately labeled anterior cortical moderately labeled ventral side labeled moderately labeled nucleus of the amygdale

Table 5-1 Summary of results from Luskin and Price (1983) regarding the olfactory tubercle (interpreted from published figures)

208

Figure 5-1

209 Figure 5-1 Diagram of neurons of a brain slice of the olfactory tubercle

Neurolucida reconstructions from cells filled with neurobiotin are shown in the layer of

the tubercle they were found in. The firing pattern of each cell is shown with a representative response to a two second long positive current injection. Neurons in the

DCL tend to be bipolar with many branching dendrites that are spiny and are regular spiking. Neurons in the multiform layer have are more variable in their morphology and

tend to be intermittently firing or burst firing. The nucleus accumbens is dorsal to the olfactory tubercle and the piriform cortex is lateral to the olfactory tubercle ML

Molecular layer; DCL Dense Cell layer; MFL Multiform Layer; LOT Lateral Olfactory

Tract.

210

Figure 5-2

211 Figure 5-2 Hypothesized circuit of the olfactory tubercle

Input from the olfactory bulb to the olfactory tubercle synapses onto dendrites in the

molecular layer. Olfactory bulb axons most likely synapse onto the dendrites of the medium spiny cells of the dense cell layer. The dense cell layer cells tend to be regularly spiking or intermittently firing. These cells then project to cells in the multiform layer, which tend to be intermittently firing or bursting. The projection cell of the tubercle is hypothesized to be a neuron found in the multiform layer and may be intermittently firing or bursting. The tubercle also receives input from other sources besides the olfactory bulb. I hypothesize that the input from the VTA targets cells in the multiform layer and research the tubercle through the nucleus accumbens. Other input to the tubercle may arrive from the molecular layer or arrive dorsally to target neurons in the multiform layer.

212

Chapter 6

Bibliography

213 Allison, AC. The morphology o the olfactory system in the . Biol Rev 28:195- 244, 1953.

Artalejo CR, Ariano MA, Perlman RL, Fox AP. Activation of facilitation calcium channels in chromaffin cells by D2 dopamine receptors through a cAMP/-dependent mechanism. Nature (Lond) 348:239-242, 1990.

Assie MB, Dominguez H, Consul-Denjean N, Newman-Tancredi A. In vivo occupancy of dopamine D-2 receptors by antipsychotic drugs and novel compounds in the mouse striatum and olfactory tubercles. Naunyn-Schmiedebergs Archives of Pharmacology 373: 441-450, 2006.

Baker DA, Fuchs RA, Specio SE, Khroyan TV, Neisewander JL. Effects of intraaccumbens administration of SCH-23390 on cocaine induced locomotion and conditioned place preference. Synapse 30:181–193, 1998.

Balu R, Strowbridge BW. Opposing inward and outward conductances regulate rebound discharges in olfactory mitral cells. J Neurophysiology 97:1959-1968, 2007.

Balu R, Larimer P, Strowbridge BW. Phasic stimuli evoke precisely timed spikes in intermittently discharging mitral cells. J Neurophysiology 92:743-753, 2004.

Balu R, Pressler RT, Strowbridge BW. Multiple modes of synaptic excitation of olfactory bulb granule cells. J 27:5621-5632, 2007.

Bird ED, Spokes EG, Iverson LL. Increased dopamine concentration in limbic areas of brain from patients dying with schizophrenia. Brain 102:347-260, 1979.

Bolam JP, Wainer BH, Smith AD. Characterization of cholinergic neurons in the rat neostritaum: A combination of choline acetyltransferase immunocytochemistry golgi- impregnation and electron microscopy. Neuroscience 12:711-718, 1984.

Bouthenet ML, Souil E, Martres MP, Sokoloff P, Giros B, Schwartz JC. Locolization of dopamine D3 receptor mRNA in the rat brain using in situ hybridization histochemistry: comparison with dopamine D2 recepto rmRNA. Brain Res 564: 203-219, 1991.

Bowery B, Rothwell LA, Seabrook GR. Comparison between the pharmacology of dopamine receptors mediating the inhibiton of cell firing in rat brain slices through the substantia nigra and the ventral tegmental area. Br J Pharmacol 112:873- 880, 1994.

Brown DA, Adams PR. Muscrarinic suppression of a novel voltage sensitive K current in a neuron. Nature 283:673-676, 1980.

214 Burlhis TM, Aghajanian GK. Pacemaker potentials of serotonergic dorsal raphe neurons: contribution of a low-threshold calcium conductance. Synaction potentialse 1:562-588, 1987.

Calleja, C. La region olfactoria del cerebro. Madrid: N. Moya, 1983.

Carbone E, Lux HD. A low voltage-activated, fully inactivating ca channel in vertebrate sensory neurons. Nature 310:501-502, 1984.

Carlson J, De Olmos J, Heimer L Tracing of two-neuron pathways in the olfactory system by the aid of transnuronal degeneration: projections to the amygdaloid body and hippocampal formation. J Comp Neurology 208:196-208, 1982 .

Carmichael ST, Clugnet MC, Price JL Central olfactory connections in the macaque monkey. J Comp Neurology 346:403-34, 1994.

Carnevale NT, Hines ML. The NEURON Book. Cambrige: Cambridge University Press, 2005.

Carr GD, White NM. Anatomical disassociation of amphetamine's rewarding and aversive effects: an intracranial microinjection study. (Berl) 89:340-346, 1986.

Charpak S, Mertz J, Beaurepaire E, Moreaux L, Delaney K. Odor-evoked calcium signals in dendrites of rat mitral cells. Proc Natl Acad Sci USA 98:1230-1234, 2001.

Chen WR, Shepherd GM. Membrane and synaptic properties of mitral cells in slices of rat olfactory bulb. Brain Res 745:189-196, 1997.

Chow HS, Chen Z, Matsuura GT. Direct Transport of Cocaine from the Nasal Cavity to the Brain Following Intranasal Cocaine Administration in Rats. J Pharm Sci 90: 1729- 1735, 2001.

Clement-Cormier YC, Kebabian JW, Petzold GL, Greengard P. Dopamine-sensitive adenylate cyclase in mammalian brain: a possible sit of action of antipsychotic drugs. Proc Natl Acad Sci USA 71:1113-1117, 1974.

Cloez-Tayarani I, Wusher N, Huere M, Fillion G. Cellular Localization of 5HT1B receptor mRNA in the rat olfactory tubercle. Brain Res Mol Brain Res 36: 337-42, 1996.

Colino A, Halliwell JV. Differential modulation of 3 separate K-condutances in hippocampal CA1 neurons by serotonin. Nature 328:73-77, 1987.

Cools AR, Prinssen EPM, Ellenbroek BA. The olfactory tubercle as a site of action of neuroleptics with an atypical profile in the Paw test – Effect of Risperidone,

215 prothiphendyl, Org-5222, Sertindole and Olanzapine. Psychopharmoclogy 119:428-439, 1995.

Dalia A, Uretsky NJ, Wallace LJ. Dopaminergic agonists administered into the nucleus accumbens: effects on extracellular glutamate and on locomotor activity. Brain Res 788:111-117, 1998.

Deschutter E, Bower JM, An active membrane model of the cerebellar purkinjje cells 1. simulation of current clamp in a slice. J Neuropyshiology 71:375-400, 1994.

Dreher JK, Jackson DM. Role of D1 and D2 dopamine receptors in mediating locomotor activity elicited from the nucleus accumbens of rats. Brain Res, 487: 267-277, 1989.

Duque GP, Dumoulin A, Triller A, Dieudonne S. Target-dependent use of co-released inhibitor transmitters at central synapses. J of Neuroscience 25:6490-6498, 2005.

Durstewitz D, Seamans JK, Sejnowski TJ. Dopamine-mediated stabilization of delay- period activity in a network model of prefrontal cortex. J Neurophysiology 83:1733- 1750, 2000.

Einhorn LC, Johansen PA, White FJ. Electrophysiological effects of cocaine in the mesoaccumbens dopamine system: studies in the ventral tegmental area. J Neuroscience 8:100-112, 1988.

Feldblum S, Erlander MG, Tobin AJ. Different Distributions of GAD(65) and GAD(67) messenger-RNAs Suggest that the 2 glutamate decaroxylases play distinctive functional roles. J of Neurosci Res 34:689-706, 1993.

Fenwick EM, Marty A, Neher E. Sodium and calcium channels in bovine chromaffin cells. J. Physiol. (Lond.) 331: 599-635, 1982.

Ghamari-Langroudi M, Bourque CW. Caesium blocks depolarizing after-potentials and phasic firing in rat supraoptic neurones. J Physiology 510:165-175, 1998.

Ghamari-Langroudi M, Bourque CW. Excitatory role of the hyperpolarization-activated inward current in phasic and tonic firing of rat supraoptic neurons. J Neuroscience 13:4855-63, 2000.

Gritti I, Mainville L, Jones BE. Codistribution of GABA- with acetylcholine-synthesizing neurons in the of the rat. J Comp Neurology 329:438-457, 1993.

Haberly LB, Price JL. Association and commissural fiber systems of the olfactory cortex of the rat. J Comp Neurology 178:711-740, 1978.

216 Halabisky B, Strowbridge BW. Gamma-frequency excitatory input to granule cells facilitates dendrodendritic inhibition in the rat olfactory Bulb. J Neurophysiology 90:644- 654, 2003.

Halliwell JV, Horne AL. Membrane properties of the granule cells of the islands of Calleja of the rat studied in vitro. J Physiology (Lond) 487:421-440, 1995.

Halliwell JV, Horne AL. Evidence for enhancement of gap junctional coupling between rat island of Calleja granule cells in vitro by the activation of dopamine D3 receptors. J Physiol (Lond) 506:175-194, 1998.

Heimer L. The olfactory cortex and ventral striatum. In: Limbic mechanisms, edited by Livingston, KE and Hornykiewicz, O. New York: Plenum Press, 1978, p. 95-187.

Heimer L, Alheid G, Zaborszky L. Basal Ganglia. In: The rat nervous system: Volume 1 Forebrain and Midbrain, edited by Paxinos, G. London: Academic Press, 1985, p. 37-86.

Heimer L, Wilson RD. The subcortical projections of : Similarities in the neural associations of the hippocampus, the piriform cortex and the neocortex. In: Golgi centennial symposium proceedings, edited by Santini, M. Ney York: Raven Press, 1975, p. 177-193.

Heimer L, Zaborszky L, Zahm DS, Alheid GF. The ventral striatopallidothalamic projection: I. The striatopallidal link originating in the striatal parts of the olfactory tubercle. J Comp Neurology 255:571-591.

Herzog E, Gilchrist J, Gras C, Muzerelle A, Ravassard P, Giros B, Gaspar P, El Mestikawy S. Localization of VGLUT3, the vesicular glutamate transporter type 3, in the rat brain. Neuroscience 123:983-1002, 2004.

Hines ML, Carnevale NT. The NEURON simulation environment. Neural Comput 9: 1179–1209, 1997.

Hines ML, Carnevale NT. NEURON: a tool for neuroscientists. Neuroscientist 7: 123– 135, 2001.

Hodgkin A, Huxley A. A quantitative description of membrane current and its application to conduction and excitation in nerve. J Physiology 117:500–544, 1952.

Hoebel BG, Monaco AP, Hernandez L, Aulisi EF, Stanley BG, Lenard L. Self-injection of amphetamine directly into the brain. Psychopharmacology (Berl) 81:158-163, 1983.

Horikawa K, Armstrong WE. A versatile means of intracellular labeling: injection of biocytin and its detection with avidin conjugates. J Neuroscience Methods 25:1-11.

Houweling, A.R. and Sejnowski, T.J.. MyFirstNEURON, available from

217 http://www.cnl.salk.edu/~arthur/MyFirstNEURON.html, last visited September 12, 2007.

Houweling, A.R. and Sejnowski, T.J.. MyFirstNEURON, available from ModelDB http://modeldb.yale.edu via accession number 3808.

Huguenard, J. R. (1996). Low-threshold calcium currents in neurons. Annual Review of Physiology 58:329-348, 1996.

Huff, RM, Adams, RN. Dopamine release in N. acumens and striatum by clozapine: Simultaneous monitoring by in vivo electrochemistry. 19:587-590, 1980.

Ichikawa T, Ajiki K, Matsuura J, Misawa H. Localization of two cholinergic markers, choline acetyltransferase and esicular acetylcholine transporter in the central nervous system of the rat: in situ hybridization histochemistry and immunohistochemsitry. J of Chemical Neuroanatomy 13:23-39, 1997.

Ikemoto S. Involvement of the olfactory tubercle in cocaine reward: intracranial self- administration studies. J Neuroscience 23:9305-9311, 2003.

Ikemoto S, Ventral strital of locomotor activity induced by cocaine, D- amphetamine, dopamine and D-1/D-2 agonists. Neuroscience 113:939-955, 2002.

Ikemoto S, Donahue KM. A five-minute, but not a fifteen-minute, conditioning trial duration induces conditioned place preference for cocaine administration into the olfactory tubercle. Synapse 56:57-59, 2005.

Jones SR, Lee TH, Wightman Rm, Ellinwood EH. Effects of intermittent and continous cocaine administration on doap-mine release and uptake regulation in the striatum: in vitro voltammetric assessment. Psychopharamcology (Berlin) 126: 331-338, 1996.

Joyce JN. D2 but not D3 receptors are elevated after 9 or 11 months chronic haloperidol treatment: influence of withdrawl period. Synapse 40:137-44, 2001.

Kaneko T, Fujiyama F, Hioki H. Immunohistochemical localization of candidates for vesicular glutamate transporters in the rat brain. J Comp Neurology 444:39-62, 2002.

Kass JI, Mintz IM (2006) Silent plateau potentials, rhythmic bursts, and pacemaker firing: three patterns of activity that coexist in quadristable subthalamic neurons. Proc Natl Acad Sci USA 103:183-188, 2006.

Krieger NR. Localization of dopamine-sensitive adenylate-cyclase within the rat olfactory tubercle. Brain Res 1983:383-391, 1980.

218 Kunzle H. An extrahippocampal projection from the dentate gyrus to the olfactory tubercle. BMC Neuroscience 6: 38, 2005.

Kuzhikandathil and Oxford (1999) Kuzhikandathil EV, Oxford GS. Activation of Human D3 Dopamine Receptor Inhibits P/Q-Type Calcium Channels and Secretory Activity in AtT-20 Cells. J Neuroscience 19: 1698-1707, 1999.

Lammers CH, Diaz J, Scwartz JC, Sokoloff P. Selective increase of dopamine D-3 receptor as a common effect of chronic antidepressant treatments. Molecular Psychiatry 5:378-388, 2000.

Landwehrmeyer B, Mengod G, Palacios JM. Differential visualization of dopamine-D2 and dopamine-D3 receptor-sites in rat-brain – A Comparative-study using in situ hybridization histochemistry and ligand-binding autoradiography. European Journal of Neuroscience 5:145-153, 1993.

Lauterborn JC, Isackson PJ, Montalvo R, Gall CM. In situ hybridization localization of choline-acetyltransferase messenger-RNA in adult-rat brain and spinal cord. Molecular Brain Res 17: 59-69, 1993.

Levesque D, Diaz J, Pilon C, Martes MP, Giros B, Souil E, Schott D, Morgat JL, Schwartz JC, Sokoloff P. Identification, characterization, and localization of the dopamine D3 receptor in rat brain using 7-[3H]hydroxyl-N,N-di-n-propyl-2- aminotetralin. Proc Natl Acad Sci USA 89:8155-8159, 1992.

Levy F, Meurisse M, Ferreira G, Thibault J, Tillet Y. Afferents to the rostral olfactory bulb in sheep with special emphasis on cholinergic, noradrenergic and serotonergic connections. J Chem Neuroanatomy 16:245-63, 1999.

Linas, R. The intrinsic electrophysiological properties of mammalian neurons insights into central nervous system function. Science 242:1654-1664, 1988.

Linas R, Yarom Y. Properties and distribution of ionic conductances generating electroresponsiveness of mammalian Interior olivary neurons in vitro. J Physiology 316:569-584, 1981.

Lingham RB, Gottesfeld Z. Deafferentation elicits increased dopamine-sensitive adenylate-cylcase and receptor-binding in the olfactory tubercle. J of Neuroscience 6:2208-2214, 1986.

Liu JC, Cox RF, Greif GJ, Freedman JE, Wazczak BL. The putative dopamine D3 receptor agoinst 7-OH-DPAT: lack of mesolimbic selectivity. Eur J Pharmacology 264: 269-278, 1994.

219 Luskin MB, Price JL (1983) The topographic organization of associational fibers of the olfactory system in the rat, including centrifugal fibers to the olfactory bulb. J Comp Neurology 216:264-291, 1983.

Lyness WH, Friedle NM, Moore KE. Destruction of dopaminergic nerve terminals in nucleus accumbens: effect on d-amphetamine self-administration. Pharmacol Biochem Behav 11:553-556, 1979.

Maccaferri G, McBain CJ. The hyperpolarization-activated current (Ih) and its contribution to pacemaker activity in rat CA1 hippocampal stratum oriens-alveus interneurones. J Physiology (Lond) 497:119-130, 1996.

Madison DV, Nicoll RA. Noradrenaline blocks accommodation of pyramidal cell discharge in the hippocampus. Nature 299:636-638, 1982.

Madison DV, Nicoll RA. Control of the Repetitive Discharge of Rat Ca1 Pyramidal Neurons in vitro. J Physiology (London) 354:319-331, 1984.

Maldonado R, Robledo P, Chover AJ, Caine SB, Koob GF. D1 dopamine receptors in the nucleus accumbens modulate cocaine self administration in the rat. Pharmacol Biochem Behav 45:239–242, 1993

Margrie TW, Schaefer AT. Theta oscillation coupled spike latencies yield computational vigour in a mammalian sensory system. J Physiology (Lond) 546:363-374, 2003.

Matteson DR, Armstrong CM. Properties of 2 types of calcium channels in clonal pituitary-cells. J of General Physiology 87:161-182, 1986.

Maturana HR, Frenk S. Direction movement and Horizontal Edge Detectors in the Pigeon Retina, Science 15:977-979, 1963.

McCormick DA, Connors BW, Lighthall JW, Prince DA. Comparative electrophysiology of pyramidal and sparsely spiny stellate neurons of the neocortex. J Neurophysiology 54:782-806, 1985.

McCormick DA, Hugenard JR. Simulation of the currents involved in rhythmic oscillations in thalamic relay neurons, J Neurophysiology 68:1373-1383, 1992.

McCormick DA, Hugenard JR. A model of the electrophysiological properties of thalamocortical relay neurons. J Neurophysiology 68:1384-1400, 1992.

McGregor A, Roberts DC. Dopaminergic antagonism within the nucleus accumbens or the amygdala produces differential effects on intravenous cocaine self-administration under fixed and progressive ratio schedules of . Brain Res 624:245–252, 1993.

220 Mengod G, Goudsmt E, Probst A, Palacios JM. Insitu hybridization histochemistry in the human hypothalamus. Prog in Brain Res 93: 45-55, 1992.

Mengod G, Villaro MT, Landwehrmeyer GB, Martinezmir MI, Niznik HB, Sunahara RK, Seeman P, ODowd BF, Probst A, Palacios JM. Visualization of dopamine D1, D2, and D3 receptor messenger-RNAs in human and rat-brain. Neurochemistry Intentional 20: SS33-S43 Suppl. S, 1992.

Mesulam MM, Mufson EJ, Wainer B, Levey AI. Central cholinergic pathways in the rat: an overview based on an alternative nomenclature. Neuroscience 10:1185-1201, 1983. Millhouse OE, Heimer L. Cell configurations in the olfactory tubercle of the rat. J Comp Neurology 228:571-597, 1984.

Mijnster MJ, Raimundo AGV, Koskuba K, Klob H, Docter GJ, Groenewegen JH, Voorn P. Regional and cellular distrubtion of serotonin 5-hydroxytryptamine(2a) receptor mRNA in the nucleus accumbens, olfactory tubercle and caudate putamen of the rat. J Comp Neurology 389:1-11, 1997.

Miyasho T, Takagi H, Suzuki H, Watanabe S, Inoue M, Kudo Y, Miyakawa H. Low- threshold potassium channels and a low-threshold calcium channel regulate calcium spike firing in the dendrites of cerebellar Purkinje neurons: a modeling study. Brain Res 891: 106-15, 2001.

Miyata M, Finch EA, Khiroug L, Hashimoto K, Hayasaka S, Oda SI, Inouye M, Takagishi Y, Agustine GJ, Kano M. Local calcium release in dendritic spines required for long-term synaptic depression. Neuron 29:233-244, 2000.

Mogenson GJ, Jones DL, Yim CY, From motivation to action – functional interface between the limbic system and the . Progress in Nerurobiology 14: 69-97, 1980.

Momiyama T, Amano T, Todo N Sasa M Inhibition by a putative antipsychotic quinolinone derivative (OPC-14597) of dopaminergic neurons in the ventral tegmental area. Eur J Pharmacology 310: 1-8, 1996.

Nicola SM, Malenka RC. Dopamine depresses excitatory and inhibitory synaptic transmission by distinct mechanisms in the nucleus accumbens. J Neuroscience, 17: 5697-5710, 1997.

O’Donnell P, Grace AA. Physiological and Morphological Properties of Accumbens Core and Shell Neurons Recorded in Vitro. Synapse 13: 135-160, 1993.

Oertel WH, Mugnaini, E. Two classes of GABAergic neurons represent the majority of neostrital neurons in the rat. Soc Neurosci Abst 9,14, 1983.

221 Owen GS, Halliwell JV. Electrophysiological characterization of laminar synaptic inputs to the olfactory tubercle of the rat studied in vitro: modulation of glutamatergic transmission by cholinergic agents is pathway-specific. Eur J Neuroscience 13:1767- 1780, 2001.

Pietrobon D and Heess P. Novel mechanism of voltage-dependent gating in L-type calcium channels. Nature (Lond) 346:651-655, 1990.

Price J. An autoradiographic study of complementary laminar patterns of termination of afferent fibers to the olfactory cortex. J Comp Neurology 150: 87-108, 1973.

Price JL, Slotnick BM, Revial MF. Olfactory Projections to the Hypothalamus. J Comp Neurology 206:447-461, 1991.

Price JL, Sprich WW. Observations on the lateral olfactory tract of the rat. J Comp Neurology 162 321-336, 1975.

Ramon Y Cajal, S. Studies on the Cerebral Cortex, translated by L.M. Kraft. London, Lloyd-Luke, LTD, 1955, xi, 179 pp.

Randall AD, Tsien RW. Contrasting biophysical and pharmacological properties of T- type and R-type calcium channels. Neuropharmacology 36:879-893, 1997.

Ribak CE, and Fallon JH The Island of Calleja Complex of Rat Basal Forebrain 1 Light and Electron-Microscopic Observations. J Comp Neurology 205: 207-218, 1982.

Roberts DCS, Corcoran ME, Fibiger HC. On the role of ascending catecholaminergic systems in intravenous self-administration of cocaine. Pharmacol Biochem Behav 6:615– 620, 1977.

Roberts DCS, Koob GF, Klonoff P, FibigerHC. and recovery of cocaine self- administration following 6-hydroxydopamine lesions of the nucleus accumbens. Pharmacol Biochem Behav 12:781–787, 1979

Roghani A, Shirzadi A, Butcher LL, Edwards RH. Distribution of the Vesicular Transporter For Acetylcholine in the Rat CNS. Neuroscience 82:1195-1212, 1998.

Rubin DB, Cleland TA. Dynamical mechanisms of odor processing in olfactory bulb mitral cells. J Neurophysiology 96:555-568, 2006.

Ruggiero DA, Anwar S, Kim J, Glickstein SB. Visceral afferent pathways to the thalamus and olfactory tubercle: Behavioral implications. Brain Res 799:159-171, 1998.

Sabatini BL, Maravall M, Svoboda K. Ca2+ signaling in dendritic spines. Current Opinion in Neurobiology 11:349-356, 2001.

222 Sah P, Gibb AJ, Gage PW. Potassium current activated by depolarization of dissociated neurons from adult guinea pig hippocampus. J Gen Physiology 92: 263-278, 1988.

Sah P, Gibb AJ, Gage PW. The sodium current underlying action-potentials in guinea-pig hippocampal CA1 neurons. J Gen Physiology 91:373-398, 1988.

Scarselli M, Novi F, Schallmach E, Lin R, Bragli A, Colzi A, Griffon N, Corsini Gu, Sokoloff P, Levenson R, Vogel Z, Maggio R. D2/D3 dopamine receptor heterodimers exhibit unique functional properties. J Biol Chem 276: 30308-30314, 2001.

Scatton B, Claustre Y, Cudennec A, Oblin A, Perrault G, Sanger DJ, Schoemaker H. Amisulpride: from animal pharmacology to therapeutic action. Int Clini Psychopharmacol Suppl 2 S29-36, 1997

Schafer MK, Eiden LE, and Weihe E Cholinergic Neurons and terminal Fields Revealed by Immunohistochemistry for the Vesicular Acetylcholine Transporter. Neuroscience 84: 331-359, 1998.

Schwartz JC, Diaz J, Bordet R, Griffon N, Perachon S, Pilon C, Ridray S, Sokoloff P. Functional implications of multiple dopamine receptor subtypes: the D1/D3 receptor coexistence. Brain Res Brain Res Rev 26:236-242, 1998.

Shahidullah M, Santarelli LC, Wen H, Levitan IB. Expression of a calmodulin-binding KCNQ2 potassium channel fragment modulates neuronal M-current and membrane excitability. Proceedings of the National Academy of Sciences of the United States of America 102:16454-16459, 2005.

Shepherd GM. The synaptic organization of the brain. New York: Oxford Univ Press, 2004.

Shepherd GM, Greer CA. Olfactory Bulb. In: Synaptic Organization of the Brain, 4th Edition, edited by Shepherd GM. New York: Oxford University Press, 1998, p.159-204. Sokoloff P, Martres MP, Giros B, Bouthenet ML, Schwartz, JC. The 3rd dopamine receptor (D3) as a novel target for antipsychotic. Biochemical Pharmacology 43: 659- 666, 1992.

Snutch TP, Leonard JP, Gilbert MM, Lester HA, Davidson N. Rat-brain expresses a heterogenous family of calcium channels. Proceedings of the National Academy of Sciences of the United States of America 87:3391-3395, 1990.

Sokoloff P, Martes MP, Giros B, Bouthenet Ml, Schwartz JC. The third dopamine receptor (D3) as a novel target for antipsychotics. Biochem Pharmacol 43: 659-666, 1992.

Spain WJ, Schwindt PC, Crill WE. Anomalous rectification in neurons from sensorimotor cortex invitro. J Neurophysiology 57:1555-1576, 1987.

223

Spyraki C, Fibiger HC, Phillips AG. Dopaminergic substrates of amphetamine-induced place preference conditioning. Brain Res 253:185-193, 1982.

Spyraki C, Sealfon SC. Regulation of dopamine-d2 receptor messenger-RNA expression in the olfactory tubercle by cocaine. Molecular Brain Research 19:313-317, 1993.

Stevens JR. Schizophrenia and dopamine regulation in the mesolimbic system, Trends Neurol Sci. 102-105, 1979.

Suaud-Chagny MF, Buda M, Gonon FG. Pharmacology of electrically evoked dopamine release studied in the rat olfactory tubercle by invivo electrochemistry. Eur J of Pharmacology 164: 273-283, 1989.

Suaud-Chagny MF, Chergui K, Chovet G, Gonon F. Relationship between dopamine release in the rat nucleus accumbens and the discharge activity of dopamingeric neurons during local in vivo application of amino acids in the ventral tegmental area. Neuroscience 49: 63-72, 1992.

Talbot K., Woolf NJ, Butcher LL. Feline islands of Calleja complex: II. Cholinergic and cholinesterasic features. J Comp Neurology 275:580-603, 1988.

Tassin, JP, Stinus L, Simon H, Blanc G, Thierry AM, Lemoal M, Cardo B, Glowinski J. Relationship between the locomotor hyperactivity induced by A10 lesions and the destruction of the frontocortical dopaminergic innervation in the rat. Brain Res 141:267- 281, 1978.

Traub RD, Wong RKS, Miles R, Michelson H. A Model of Ca3 Hippocampal Pyramidal Neuron Incorporating Voltage-clamp data on Intrinsic Conductances. J Neurophysiology 66: 635-650, 1991.

Tsien RW, Hess P, McClewskey EW, Rosenberg RL. Calcium channels - Mechanisms of selectivity, permeation, and block. Annual Review of Biophsyics and Biophysical Chemistry 16:265-290, 1987.

Umemiya M, Raymond L. Dopaminergic modulation of excitatory postsynaptic currents in rat neostriatal neurons. J Neurophysiol 78: 1248-1255, 1997.

Veenman CL, Riner A. The distribution of GABA containing perikarya, fibers and terminas in the forebrain and midbrain of pigeons, with particular reference to the basal ganglia and its projection targets. J Comp Neurology 339:209-250, 1994.

Venance L, Glowinski J. Heterogeneity of spike frequency adaptation among medium spiny neurones from the rat striatum. Neuroscience 122:77-92, 2003.

224 Vertes, RP. Differential projections of the infralimbic and prelimbic cortex in the rat. Synapse 51:32-58, 2004.

Voorn P, Jorritsmabyham B, Vandijk C, Buijs RM. The Dopaminergic Innervation of the Ventral Striatum in the Rat – A Light-Microscopic and Electron-Microscopic Study with Antibodies Against Dopamine. J Comp Neurology 251:84-99, 1986.

Wang W, Hahn KH, Bishop JF, Gao DQ, Jose PA, Mouradian MM. Up-regulation of D3 dopamine receptor mRNA by neuroleptics. Synapse 23:232-235, 1996.

Weihe E, TaoCheng JH, Schafer MKH, Erickson JD, Eiden LE. Visualization of the vesicular acetylcholine transporter in cholinergic nerve terminals and its targeting to a specific population of small synaptic vesicles. Proceeding of the National Academy of Sciences USA 93:3547-3552, 1996.

Willis MA, Arbas EA Variability in odor-modulated flight by moths. J Comps Physiol A 182: 191-202, 1998.

Willner P, Scheel-Kruger J The mesolimbic Dopamine System: From Motivation to Action. Chichester: John Wiley & Soons, 1991.

Wolf ME The role of exciatory amino acids in behavioral sensitization to psychomotor stimulant. Prog Neurobiology 54: 679-720, 1998.

Wong RK, Prince DA. Afterpotential generation in hippocampal pyramidal cells. J Neurophysiology 45:86-97, 1981.

Wong, R. K. & Prince, D. A. Participation of calcium spikes during intrinsic burst firing in hippocampal neurons. Brain Research 159:385-390, 1978.

Xi ZX, Garnder EL. Pharmacological actions of NGB 2904, a selective D3 receptor antagonist, in animal models of drug addiction. CNS Drug Rev 13: 240-59, 2007.

Yamada WM, Koch C, Adams PR. Multiple channels and calcium dynamics. In: Methods in Neural Modeling: From Synapses to Networks, edited by Koch,C, Segev I. MIT Press, 1989, p. 97-133.

Yue C, Remy S, Su H, Beck H, Yaari Y. Proximal persistent Na+ channels drive spike afterdepolarizations and associated bursting in adult CA1 pyramidal cells. J Neuroscience 25:9704-9720, 2005.

Zaborszky L, Carlsen J, Brashear HR, Heimer L. 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 Neurology 243:488-509, 1986.

225 Zazpe A, Artaiz I, Delrio J. Role of 5-HT3 receptors in basal and K+-evoked dopamine release from rat olfactory tubercle and striatal slices. British J of Pharm. 113:968-972, 1994.

Zetterstrom T, Sharp T, Marsden CA, Ungerstedt U, In vivo measurement of dopamine and its metabolites by intracerebral dialysis: changes after d-amphetamine. J Neurochem 41: 1769-1773, 1983.

Zhang L, Lou D, Jiao H, Zhang D, Wang X, Xia Y, Zhang J, Xu M. Cocaine-induced intracellular signaling and gene expression are oppositely regulated by dopamine D1 and D3 receptors. J Neuroscience 24: 3344-3354, 2004.

226