<<

Increased activity of GABAA receptors contributes to postanesthetic memory deficits

by

Agnieszka A. Zurek

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Physiology University of Toronto

© Copyright by Agnieszka A. Zurek, 2015

Increased activity of GABAA receptors contributes to postanesthetic memory deficits

Agnieszka A. Zurek Doctor of Philosophy Department of Physiology University of Toronto 2015 Abstract

General anesthetics are widely used to allow patients to tolerate surgery and to sedate patients in intensive care units. Anesthesia causes long-term memory deficits in laboratory animals, suggesting it could contribute to memory loss in patients. However, the mechanisms underlying postanesthetic memory deficits are unknown.

Most anesthetics cause neurodepression by allosterically increasing the activity of γ- aminobutyric acid type A (GABAA) receptors. In particular, positive allosteric modulation of

α5 subunit-containing GABAA (α5GABAA) receptors contributes to the acute, amnestic effects of the anesthetic . Once the anesthetic has been eliminated, allosteric modulation of GABAA receptors is rapidly reversed and it is assumed that GABAA receptors do not contribute to memory deficits that persist after anesthesia. However, previous work from our lab suggests that α5GABAA receptors may play a role in postanesthetic memory loss, as treatment with a that inhibits these receptors before anesthesia prevents memory deficits. I hypothesized that exposure to anesthetics results in a sustained increase in

α5GABAA activity, which causes memory deficits.

ii

First, my data showed that α5GABAA receptors are required for postanesthetic memory deficits as wild-type, but not Gabra5 null-mutant mice (Gabra5-/-), exhibit impaired memory performance on the object recognition memory task. These deficits are evident for 24 hours after isoflurane anesthesia and persist for up to 1 week after exposure to the injectable anesthetic etomidate.

The mechanism underlying these memory deficits is likely a sustained increase in the function of α5GABAA receptors as a single, in vivo exposure to the anesthetic etomidate (8 mg/kg i.p.) or isoflurane (1.3%, 1 h) increases a tonic inhibitory current that is mediated by

α5GABAA receptors and increases cell-surface expression of α5GABAA receptors for least 1

week with full recovery by 2 weeks. Memory deficits in wild-type mice are reversed after anesthesia by inhibiting α5GABAA receptor activity with the L-655,708.

Collectively, these results refute the widely-held belief that the function of GABAA

receptors returns to baseline after the anesthetic agent has been eliminated. Furthermore, the

data present a mechanism and a plausible treatment strategy for postanesthetic memory

deficits.

iii

Acknowledgments

I would like to thank my supervisor and mentor Dr. Beverley Orser, without you none of this would be possible. Thank you for your guidance, support, unwavering patience, and all the opportunities that allowed me to develop as a scientist and as a person. I admire your adventurous spirit in science and life. I thank my Supervisory Committee Drs. Evelyn Lambe and Paul Frankland for your advice, support and kindness over the years. Thank you also to Drs. Melanie Woodin, Mike Salter and Milton Charlton for your mentorship and counsel on my project.

Thank you to lab alumni, Drs. Loren Martin and Robert Bonin, for your friendship and for always sharing your wisdom with me, especially in times of need. I am very grateful for the contributions of the entire Orser lab. Thank you Dr. Dianshi Wang for being the lab sage, the crusader for perfection, and the patient teacher. Thank you Ella Czerwinska for all your jokes, hugs, and zucchinis, you made me feel at home from day one. Thank you Anine (Jieying) Yu for your immense generosity and help on my project. Thank you to all the summer students I’ve had the pleasure to work with over the years; Burç Aydin, Mohammed Haijha, Erica Bridgwater, Zeenia Aga and Eric Salter – you all taught me so much. Also thank you to everyone who has been a part of the lab family, Dr. Paul Whissell, Irene Lecker, William To, Dr. Sinziana Avramescu, Dr. Antonello Penna, Dr. Gang Li, Dr. Stephen Kemp, Dr. Junhui Wang, Sean Haffey, and Fariya Mostafa. You have taught me the power of teamwork and have made the lab feel like a second home.

Most importantly, I would like to thank my family. I thank my parents, Beata and Zdzislaw Zurek, for your love, constant encouragement, and interest in my work, especially over the last decade of university. I will be forever grateful. Lastly, thank you to my partner-in-crime, Peter Dziak, for your love, understanding, and always reassuring me that “yes, I can”.

iv

List of Contributions

Agnieszka Zurek produced all of the results with the exception of those listed below. The behavioural data presented in Chapter 4 were collected with the assistance of Ms. Erica Bridgwater, who performed the experiment with . In Chapter 5, the behavioural studies of mice treated with a sedative dose of etomidate were performed with the assistance of Ms. Erica Bridgwater. Studies of cell-surface expression were performed by Ms. Jieying Yu, Dr. Gang Li, and Mr. Tom Chang. Recordings from cultured neurons were performed with the assistance of Dr. Antonello Penna and Dr. Dianshi Wang. The recordings in astrocyte-neuron and microglia-neuron cocultures were performed by Mr. Sean Haffey and Ms. Irene Lecker. The experiments presented in Appendix 1 were performed with the help of Ms. Zeenia Aga. Dr. Dianshi Wang and Dr. Beverley Orser helped with the writing of the articles published in Anesthesia and Analgesia and in The Journal of Clinical Investigation, modified versions of these articles are presented in Chapters 4 and 5.

v

Table of Contents

Acknowledgments...... iv

Table of Contents ...... vi

List of Tables ...... xi

List of Figures ...... xii

List of Abbreviations ...... xv

List of Appendices ...... xix

Chapter 1. Thesis Structure ...... 1

1.1 General Overview ...... 1

1.2 Hypothesis and Specific Aims ...... 2

1.2.1 Hypothesis...... 2

1.2.2 Specific Aims ...... 2

1.3 Thesis Structure ...... 3

Chapter 2. General Introduction ...... 5

2.1 Postoperative Cognitive Dysfunction ...... 5

2.1.1 Incidence ...... 5

2.1.2 Cognitive Domains Affected by POCD ...... 6

2.1.3 Effects on patient outcome...... 7

2.1.4 Risk factors and potential causes of POCD in patients...... 8

2.2 Memory deficits after surgery and anesthesia in animal models ...... 12

2.2.1 Memory deficits after surgery in animal models ...... 14

2.2.2 Memory deficits after anesthesia in animal models ...... 18

2.3 Proposed mechanisms of postanesthetic memory deficits in animals ...... 18

2.3.1 and activation of apoptotic pathways ...... 19

vi

2.3.2 Impaired neurogenesis ...... 21

2.3.3 Alzheimer Disease-related mechanisms ...... 21

2.3.4 Potential GABAA receptor-dependent mechanisms ...... 24

2.4 GABA and GABAA receptor-mediated inhibition ...... 24

2.4.1 GABA Synthesis and release ...... 24

2.4.2 GABA transport and metabolism...... 26

2.4.3 GABA receptors...... 27

2.4.4 General overview of GABAA receptors ...... 28

2.4.5 Subunit composition of GABAA receptors ...... 31

2.4.6 GABAA receptor-associated ...... 32

2.4.7 GABAA receptor mediated inhibition ...... 36

2.4.8 Synaptic GABAA receptors ...... 40

2.4.9 Extrasynaptic GABAA receptors ...... 42

2.4.10 The physiological role of α5GABAA receptors ...... 46

2.4.11 The role of α5GABAA receptors in pathophysiology ...... 48

2.5 Pharmacology of GABAA receptors ...... 51

2.5.1 Effects of general anesthetics on GABAA receptors ...... 53

2.5.2 Pharmacology of α5GABAA receptors ...... 58

2.6 Structure and Function of the Hippocampus ...... 59

2.6.1 Structure ...... 59

2.6.2 The role of the hippocampus in learning and memory ...... 63

2.6.3 LTP in the hippocampus ...... 64

2.6.4 GABAA receptors in the hippocampus ...... 67

2.7 Summary ...... 69

Chapter 3. General Materials and Methods ...... 70

vii

3.1 Study approval ...... 70

3.2 Experimental animals...... 70

3.3 Anesthesia ...... 70

3.4 Preparation of Pharmacological Agents used In Vivo ...... 72

3.5 Behaviour ...... 73

3.5.1 Handling ...... 73

3.5.2 Object recognition ...... 73

3.6 Electrophysiology in Brain Slices ...... 78

3.6.1 Preparation of Brain Slices ...... 78

3.6.2 Extracellular Recordings ...... 79

3.6.3 Whole-cell recordings in brain slices ...... 80

3.7 Cell-surface and total expression ...... 81

3.7.1 Hippocampal Slice Preparation ...... 81

3.7.2 Solutions used for surface biotinylation ...... 83

3.7.3 Biotinylation ...... 83

3.7.4 Western Blot ...... 86

3.7.5 Analysis of cell-surface and total expression data ...... 88

3.8 Electrophysiology in cell culture ...... 88

3.8.1 Preparation of cell cultures ...... 88

3.8.2 Whole-cell recordings in cell culture ...... 90

3.9 Statistical analyses ...... 90

Chapter 4. Inhibition of α5GABAA receptors restores recognition memory after isoflurane general anesthesia...... 92

4.1 Introduction ...... 92

4.2 Methods...... 94

4.2.1 Animal Model ...... 94

viii

4.2.2 Anesthesia ...... 94

4.2.3 Novel Object Recognition...... 95

4.2.4 Drug Treatment ...... 97

4.2.5 Statistical Analysis ...... 98

4.3 Results ...... 102

4.3.1 1-minute and 1-hour Memory Performance ...... 102

4.3.2 L-655,708 reverses memory deficits after isoflurane ...... 108

4.3.3 Memory performance 72 h after isoflurane ...... 112

4.3.4 Memory performance of Gabra5-/- mice 24 h after isoflurane ...... 115

4.3.5 Prevention of postanesthesia memory deficits ...... 119

4.3.6 Memory performance 24 h after sevoflurane...... 120

4.4 Discussion ...... 123

Chapter 5. Sustained increase in α5GABAA receptor function impairs memory after anesthesia ...... 128

5.1 Introduction ...... 128

5.2 Methods...... 129

5.2.1 Experimental animals...... 129

5.2.2 Anesthesia...... 129

5.2.3 Novel object recognition memory assay...... 130

5.2.4 Electrophysiology in hippocampal slices...... 131

5.2.5 Primary cell culture...... 133

5.2.6 Whole-cell voltage-clamp recordings in cell culture...... 134

5.2.7 Cell-surface biotinylation...... 134

5.2.8 Statistical analyses...... 136

5.3 Results ...... 136

5.4 Discussion ...... 169

ix

Chapter 6. General Discussion ...... 172

6.1 Summary ...... 172

6.2 Discussion ...... 173

6.2.1 The role of astrocytes ...... 174

6.2.2 Dexmedetomidine ...... 175

6.2.3 Pharmacological regulation of GABAA receptor expression ...... 177

6.2.4 Regulation of GABAA receptor expression ...... 178

6.3 Future directions ...... 179

6.4 Summary ...... 185

Appendix 1. Mutations in Gabra5 are associated with autism-related deficits ...... 187

Appendix 2. Thesis-relevant work published by candidate ...... 206

Appendix 3. Additional publications resulting from my doctoral studies...... 208

References ...... 209

x

List of Tables

Table 5.1……………………………………………………………………………………….145

xi

List of Figures

Figure 2.1 Perioperative factors that affect the likelihood of POCD...... 11

Figure 2.2 A simple classification scheme for the mechanisms that contribute to POCD. .... 13

Figure 2.3 The GABAA receptor...... 30

Figure 2.4 A schematic drawing representing phasic and tonic inhibition mediated by

GABAA receptors as recorded from a neuron during whole-cell voltage-clamp recording. .. 45

Figure 2.5 The trisynaptic pathway of the hippocampus...... 60

Figure 3.1 The object recognition paradigm...... 77

Figure 4.1 The timelines of experimental protocols...... 100

Figure 4.2 Normal 1-minute memory 24 hours after isoflurane anesthesia...... 104

Figure 4.3 Impaired 1-hour recognition memory 24 hours after isoflurane anesthesia...... 106

Figure 4.4 Memory deficits can be reversed by inhibition of α5GABAA receptors 24 hours after isoflurane anesthesia...... 110

Figure 4.5 Normal recognition memory 72 hours after isoflurane anesthesia...... 113

Figure 4.6 Gabra5-/- mice exhibit no recognition memory deficits 24 hours after isoflurane anesthesia...... 117

Figure 4.7 Memory deficits can be prevented by inhibition of α5GABAA receptors prior to anesthesia...... 121

Figure 5.1 A sedative dose of etomidate impairs memory for 72 hours in WT mice...... 138

Figure 5.2 Plasticity of fPSPs is reduced 24 hours after sedation with etomidate...... 140

xii

Figure 5.3 Plasticity at Schaffer collateral-CA1 synapses 1 week after sedation with etomidate...... 142

Figure 5.4 Miniature inhibitory postsynaptic currents are not affected by etomidate treatment...... 144

Figure 5.5 A sedative dose of etomidate causes a sustained increase in α5GABAA receptor- mediated tonic current for at least 1 week...... 146

Figure 5.6 Treatment of astrocytes with etomidate is necessary and sufficient to trigger a sustained increase in tonic current in neurons...... 149

Figure 5.7 The cell-surface expression of α5GABAA receptors is increased for 1 week after treatment of mice with a sedative dose of etomidate...... 152

Figure 5.8 Etomidate increases cell-surface expression of β3 subunits but does not change the cell-surface expression of δ and α1 subunits 24 hours after treatment...... 153

Figure 5.9 Reversal of memory impairment after a sedative dose of etomidate...... 156

Figure 5.10 Plasticity is not impaired 24 hours after treatment of Gabra5-/- mice with a sedative dose of etomidate...... 158

Figure 5.11 An anesthetizing dose of etomidate impairs memory performance on the object recognition task for at least 1 week...... 161

Figure 5.12 An anesthetizing dose of etomidate impairs plasticity for 24 hours in slices from WT mice...... 162

Figure 5.13 An anesthetizing dose of etomidate increases a tonic inhibition and cell-surface

expression of α5GABAA receptors 24 hours after treatment...... 163

Figure 5.14 A brief, sedative dose of isoflurane does not affect the expression of α5 and δ

GABAA receptor subunits...... 166

xiii

Figure 5.15 An anesthetizing dose of isoflurane increases the tonic current and cell-surface expression of α5GABAA receptors 24 hours after treatment of mice...... 167

xiv

List of Abbreviations

5HT3 5-hydroxytryptamine (5-HT) 3 (receptor)

α5IA α5GABAA receptor-selective inverse agonist

α5GABAA α5 subunit-containing γ-aminobutyric acid subtype A (receptor)

δGABAA δ subunit-containing γ-aminobutyric acid subtype A (receptor)

ACSF Artificial cerebrospinal fluid

AD Alzheimer Disease

AMPA α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate

ANOVA Analysis of variance

AP2 Adaptin complex 2

APV (2R)-amino-5-phosphonovaleric acid

ASD Autism spectrum disorder

ATP

BAPTA 1,2-bis(o-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid

BCA Bicinchoninic acid (assay)

Bcl-2 B cell lymphoma 2 ()

BIC

CA1 Cornu Ammonis area 1

CA3 Cornu Ammonis area 3

CaMKII Calcium/calmodulin-dependent protein kinase II cAMP cyclic adenosine monophosphate

CNQX 6-cyano-7-nitroquinoxaline-2,3-dione

xv

CREB cAMP response element-binding (protein)

D-AP5 D(-)aminophosphopentanoic acid

ECF Extracellular fluid

EDTA Ethylenediaminetetraacetic acid

EEG Electroencephalogram

EGTA Ethylene glycol-bis(2-aminoethylether)-N,N,N′,N′-tetraacetic acid

EPSP Excitatory postsynaptic potential

ERK Extracellular signal-regulated kinase fPSP Field postsynaptic potential

GABA γ-Aminobutyric acid

GABAA γ-Aminobutyric acid subtype A (receptor)

GABAB γ-Aminobutyric acid subtype B (receptor)

GABA-T γ-Aminobutyric acid transaminase

Gabra5-/- α5 subunit-containing γ-aminobutyric acid subtype A deletion

Gabrd-/- δ subunit-containing γ-aminobutyric acid subtype A gene deletion

GAD Glutamate decarboxylase

GABARAP GABAA receptor-associated protein

GAT γ-Aminobutyric acid transporter

GDP Guanosine diphosphate

GFAP Glial fibrillary acidic protein

GIRK G protein-coupled inwardly-rectifying potassium channels

GODZ Golgi-specific DHHC finger protein

GTP Guanosine triphosphate

xvi

HAP1 Huntingtin associated protein 1

HEK Human embryonic (cell)

HEPES 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

Hz Hertz i.p. Intraperitoneal

ICF Intracellular fluid

IL-1β Interleukin-1β

IL-1R Interleukin 1 receptor

IL-6 Interleukin-6

IPSC Inhibitory postsynaptic current

KCC2 K+ - Cl- co-transporter

LORR Loss of righting reflex

LTD Long-term depression

LTP Long-term potentiation

MAC Minimum alveolar concentration

MAPK Mitogen-activated protein kinase mIPSC Miniature inhibitory postsynaptic current mOsm Milliosmoles mRNA Messenger ribonucleic acid mV Millivolt

NKCC1 Na-K-Cl cotransporter

NMDA N-methyl-D-aspartic acid

P38-MAPK P38 mitogen-activated protein kinase

xvii pA Picoampere

PAM Positive allosteric modulator

PET Positron emission tomography pF Picofarad

PKA Protein kinase A

PKC Protein kinase C

Plic-1 Proteins linking integrin-associated protein and cytoskeleton

PLP Pyridoxal phosphate

POCD Postoperative cognitive dysfunction pS Picosiemens

S100B S100 calcium binding protein B

TEA Tetraethylammonium

TM Transmembrane

TNFα Tumor necrosis factor α

TTX Tetrodotoxin

USV Ultrasonic vocalization

VGAT Vesicular γ-aminobutyric acid transporter

VGCC Voltage-gated calcium channel

Vm Membrane potential

WT Wild type

xviii

List of Appendices

Appendix 1. Mutations in Gabra5 are associated with autism-related deficits.

Appendix 2. Thesis-relevant work published by candidate.

Appendix 3. Additional publications resulting from my doctoral studies.

xix 1

Chapter 1. Thesis Structure 1.1 General Overview

Cognitive dysfunction is common in patients in the postoperative period (Monk et al. 2008).

Specifically, cognitive deficits occur in approximately a third of patients at hospital discharge

and can persist for months after surgery (Moller et al. 1998; Monk, et al. 2008). The causes

of memory deficits after surgery and anesthesia are likely multifactorial however, studies using animal models have identified anesthetics as a major contributing factor (Culley et al.

2003; Culley et al. 2004; Saab et al. 2010). It is unknown how anesthetics trigger long- lasting memory deficits.

Most commonly used general anesthetics cause the desirable endpoints of immobility, hypnosis, and amnesia by allosterically enhancing the activity of inhibitory γ-aminobutyric acid type A (GABAA) receptors in the brain (Rudolph et al. 2004). Once the anesthetic is

eliminated, it is assumed that GABAA receptor function returns to baseline and GABAA

receptors no longer impair cognition. In this thesis, I examine the molecular mechanisms that

underlie memory deficits that occur after exposure to an anesthetic in adult animals. My goal

is to determine whether there are lasting effects of anesthetics on memory performance and on the function of GABAA receptors, particularly the function of α5 subunit-containing

GABAA (α5GABAA) receptors.

α5GABAA receptors are highly expressed in the hippocampus and have been shown to modulate learning and memory processes (Crestani et al. 2002; Martin et al. 2010).

Reducing the activity of these receptors, either pharmacologically or by reducing their

expression genetically in a mouse model improves performance on certain memory tasks

2

(Martin, et al. 2010). Conversely, pharmacologically enhancing receptor activity impairs

learning on contextual fear memory and Morris Water Maze spatial memory tasks (Cheng et

al. 2006). Previous studies from our lab show that α5GABAA receptors are necessary for the acute amnesia produced by low, sedative doses of the anesthetic etomidate (Cheng, et al.

2006). Additionally, postanesthetic memory deficits that occur after isoflurane anesthesia can be prevented by pretreating mice with a drug that pharmacologically inhibits α5GABAA

receptors (Saab, et al. 2010). These results suggest that these receptors trigger memory

deficits that persist after the anesthetic has been eliminated (Saab, et al. 2010).

1.2 Hypothesis and Specific Aims 1.2.1 Hypothesis

I hypothesized that α5GABAA receptors are necessary to trigger postanesthetic memory deficits and exposure to an anesthetic results in a sustained increase in α5GABAA

receptor activity, which causes memory deficits. I tested these hypotheses by addressing the

following specific aims.

1.2.2 Specific Aims

1. Determine whether pharmacologically or genetically inhibiting α5GABAA receptors

during an acute exposure to an anesthetic prevents postanesthetic memory deficits.

2. Determine whether pharmacologically inhibiting α5GABAA receptors after anesthesia

reverses deficits in learning and memory.

3. Determine whether the expression or the current generated by α5GABAA receptors is

enhanced after anesthesia.

3

1.3 Thesis Structure

In Chapter 2, I present an overview of the relevant literature. In Chapter 3, the methods are

described in considerable detail to ensure the reproducibility of my work. In Chapter 4, the

results address Aims 1 and 2, which seek to determine whether inhibition of α5GABAA

receptors can both prevent and reverse memory deficits after anesthesia. In Chapter 4, I

examined the effects of the anesthetics isoflurane and sevoflurane on memory performance

using the object recognition memory task. I found that both anesthetics impaired memory 24

hours after treatment and isoflurane impaired memory for at least 48 hours. Gabra5 null-

mutant mice, which do not express the α5GABAA receptor (Gabra5-/-), exhibited normal learning and memory after isoflurane. Pharmacological inhibition of α5GABAA receptors 10

minutes before anesthesia prevented memory deficits and inhibition of α5GABAA receptors

24 hours after anesthesia reversed any impairment in learning and memory. This work presented a potential treatment for deficits in learning and memory after anesthesia in adult animals. This work was published in Anesthesia and Analgesia (Zurek et al. 2012).

In Chapter 5 the results address Aim 3 and identify the mechanism by which anesthetics that act on GABAA receptors cause memory deficits. I found that sedative and anesthetic doses of the GABAergic anesthetic etomidate impaired memory and plasticity in wild-type (WT) mice but not Gabra5-/- mice. These memory deficits were associated with an

increase in α5GABAA receptor-mediated tonic inhibitory current in the hippocampus. In vitro

experiments showed that the increased tonic current required the presence of astrocytes. The

increase in tonic current was caused by an increase in cell-surface expression of α5GABAA

receptors. The increased tonic current and cell surface expression persisted for at least 1

week after treatment and recovered to baseline by 2 weeks. Similarly, an anesthetizing dose

4

of isoflurane increased tonic current and cell-surface expression of α5GABAA receptors 24 hours after treatment. This study is published in The Journal of Clinical Investigation (Zurek et al. 2014).

5

Chapter 2. General Introduction 2.1 Postoperative Cognitive Dysfunction 2.1.1 Incidence

Many physicians have heard anecdotal reports from family members that their relative was

“never quite the same” after surgery. Indeed, one of the first retrospective studies of memory

loss after surgery was published more than 50 years ago (Bedford 1955). Large-scale studies have now confirmed that a surprising number of patients experience cognitive deficits in the days and months after surgery. This deterioration in cognitive performance is commonly referred to as postoperative cognitive dysfunction (POCD) (Caza et al. 2008; Moller, et al.

1998). The highest incidence of POCD occurs after cardiac surgery: 53% of patients exhibit

deficits at hospital discharge and 42% of patients still show deficits 5 years later (Newman et

al. 2001). Among adults undergoing noncardiac surgery, the incidence of POCD is lower.

The multi-centre International Study of POCD (ISPOCD1), demonstrated that 23% of

patients aged 60 to 69 years old, and 29% of patients over age 70, experienced POCD one

week after major non-cardiac surgery (Moller, et al. 1998). Three months later, 7% of

patients in the 60 to 69 age group and 14% of patients in the 70 and older age group still

experienced POCD (Moller, et al. 1998). In a subsequent study of 1064 patients who

underwent major noncardiac surgery, 36.6% of young adults (18-39 years old), 30.4% of

middle-aged (40-59 years old) and 41.4% of elderly (60 years or older) experienced POCD at

hospital discharge (Monk, et al. 2008). Three months after surgery, 5.7% of young, 5.6% of

middle-aged and 12.7% of elderly still had POCD. Even relatively brief surgical procedures

can be associated with cognitive deficits. For example, in one study, almost half (47%) of

6

elderly patients who underwent minor surgical procedures under general anesthesia exhibited

memory deficits that persisted for at least 24 hours (Rohan et al. 2005).

In children, anesthesia may have lasting effects on neural development and cognition.

Anesthesia in conjunction with surgery results in impaired executive function, as measured by choice reaction time, and also causes retrograde amnesia in pediatric patients (Millar et al.

2006; Quraishi et al. 2007). Even a brief, general anesthetic administered during a dental procedure is associated with impaired visual memory 24 hours later (Millar et al. 2014). An analysis of a large database shows that these surgeries may have long-lasting side-effects

(Wilder et al. 2009). Exposure to general anesthesia during cesarean delivery has been linked to an increased risk of the child developing autism (Chien et al. 2014). Furthermore, the incidence of learning disabilities later in life is almost double in patients who underwent multiple episodes of anesthesia before 4 years of age when compared with children who never received an anesthetic (Wilder, et al. 2009). However, interpreting the results of psychometric tests in pediatric patients is confounded by pre-existing medical conditions and developmental changes in the young brain. Current initiatives by the International

Anesthesia Research Society (IARS) are focused on determining how to increase the safety of anesthesia in pediatric patients, as well as to identify the cognitive domains that are most affected after anesthesia and the mechanisms underlying these impairments. While the effects of anesthetics on the developing brain are important, they are not the subject of this thesis so they will not be reviewed in detail.

2.1.2 Cognitive Domains Affected by POCD

In adult patients, psychometric tests, such as the delayed word recall memory test and the

backward digit span test, administered before and after surgery have identified memory and

7

executive function as the most vulnerable cognitive domains (Price et al. 2008; Silverstein et

al. 2007). At 3 months after surgery, memory loss was more prevalent than deficits in

executive function: 42% of POCD patients showed impairment on learning and memory

tasks, 26% showed deficits in executive function, and 9% showed both memory and

executive deficits (Price, et al. 2008). These deficits impair the ability of patients to perform

the activities of daily life in the early postoperative period and for months after surgery.

2.1.3 Effects on patient outcome

A significant decline in cognition after surgery has long-term consequences and is associated

with a reduction in the quality of life, premature retirement, an increase in the likelihood that

patients will require social assistance (Steinmetz et al. 2009) and even premature death

(Newman, et al. 2001; Phillips-Bute et al. 2006). These deficits have practical significance

particularly in the early postoperative period when patients require explicit recall for important information, such as postoperative instructions.

In daily life, deficits in executive function impair patients’ organizational skills and affect their ability to plan and schedule appointments (Price, et al. 2008). Although memory

deficits are more frequent, adult patients with impaired executive function experience greater

functional impairment and require assistance with instrumental activities of daily living such

as meal preparation, commuting, shopping, housework, as well as assistance with finances

and managing their (Price, et al. 2008). Due to the profound impact of POCD

on patient outcome, it is important to identify and minimize risk factors for POCD.

8

2.1.4 Risk factors and potential causes of POCD in patients

The risk factors associated with POCD can be broadly categorized as patient-related, surgical

and anesthetic-related. Age and preexisting medical conditions are the most significant

patient-related risk factors. Advanced age appears to be a particularly prominent risk factor

(Monk, et al. 2008). Importantly, among elderly patients, the incidence of POCD does not seem to decrease from 6 weeks (54.3%) to 1 year (46.1%), which indicates that deficits are not easily resolved in this age group (McDonagh et al. 2010). In elderly patients, pre-

existing mild cognitive impairment increases the likelihood of POCD (Bekker et al. 2010).

Similarly, surgery exacerbates memory deficits in patients with Alzheimer disease (AD) and

increases the levels of amyloid β oligomers in the plasma and cerebrospinal fluid (Evered et

al. 2009). However, it remains unclear whether anesthesia and surgery trigger permanent

cognitive decline in elderly patients (Avidan et al. 2009). In fact, a recent study shows that when patients were followed for 11 years after the ISPOCD1 and ISPOCD2 studies, POCD

was not significantly associated with a diagnosis of dementia (Steinmetz et al. 2013).

Other patient-related risk factors for POCD include a low level of education at the

time of surgery, previous cerebrovascular accident and prior abuse (Ancelin et al.

2001; Gigante et al. 2011; Heyer et al. 2000; Hudetz et al. 2007; Monk, et al. 2008;

Newman, et al. 2001; Rudolph et al. 2008; Stockton et al. 2000; Tsai et al. 2008). In

addition, patients with metabolic syndrome, which is characterized by a high body mass

index or abdominal obesity, hypertension, insulin resistance and hyperglycemia, are at

increased risk of developing POCD (Hudetz et al. 2011; Hudetz et al. 2011). One week after

elective cardiac or non-cardiac surgery, patients with metabolic syndrome exhibit greater

9

impairment on tests of memory and executive function than patients that do not have the

disorder (Hudetz, et al. 2011; Hudetz, et al. 2011).

Many perioperative or “surgical” factors affect cognitive performance in the early postoperative period. The type of surgery seems to affect the incidence of POCD, with cardiac surgery posing the highest risk, followed by major intraabdominal and thoracic surgeries (Moller, et al. 1998). Additional perioperative factors that are positively correlated with POCD are the duration of anesthesia, postoperative , respiratory complications

during surgery, administered before surgery, a longer hospital stay, delirium

during the hospital stay and opioids administered within 24 hours of cognitive testing

(Moller, et al. 1998; Monk, et al. 2008).

Interestingly, increasing the duration of anesthesia, but not the choice of technique

(general versus regional anesthesia), is associated with POCD in patients (Newman et al.

2007). However, interpreting the results of clinical studies is problematic because most patients undergoing regional anesthesia also receive a sedative dose of a general anesthetic, such as . Several studies suggest that the incidence of POCD in the early postoperative period is greater after general anesthesia (Papaioannou et al. 2005; Weber et al. 2009), but the majority of evidence indicates that cognitive deficits occur independent of the anesthetic technique (Rasmussen et al. 2003). Patients also experience POCD independent of whether they underwent general anesthesia with total intravenous anesthesia using such as propofol, or with inhalational anesthesia using drugs such as desflurane or isoflurane (Rohm et al. 2006). However, the specific anesthetic agent may affect the likelihood of a patient developing POCD. A recent study showed that in a group of patients undergoing lower extremity or abdominal surgery, anesthesia with desflurane and spinal

10

tetracaine was associated with POCD, whereas spinal anesthesia alone or spinal anesthesia

with inhaled isoflurane were not associated with POCD (Zhang et al. 2012). To devise an

anesthetic regimen that will minimize the incidence of POCD, the relative neurotoxicity of

different anesthetic agents and the mechanisms by which they impair cognition must be the

subject of future studies.

The cause of POCD is undoubtedly multi-factorial and multiple factors impair

cognition at different postoperative time points (Figure 2.1). For example, the increase in

stress-related cortisol levels adversely influences performance on psychometric tests

immediately after surgery (Rasmussen et al. 2005; Wiklund et al. 2009). Other factors that

may cause reduced cognitive acuity in the early postoperative period include opioid analgesics (Monk, et al. 2008), postoperative hypothermia (Tan et al. 2010), sleep disturbances during the hospital stay (Gogenur et al. 2007) and the novel hospital environment. Moreover, medical illness and surgical trauma stimulate an immunological response and the release of proinflammatory cytokines, which can last for days and is associated with neurocognitive decline after cardiopulmonary bypass surgery (Hudetz et al.

2011; Ramlawi et al. 2006). Increased serum concentrations of the proinflammatory cytokines interleukin-6 (IL-6) and C reactive protein were associated with cognitive dysfunction after cardiopulmonary bypass surgery (Hudetz, et al. 2011). These factors are known to impair cognition and may act synergistically to cause POCD (Figure 2.1).

Furthermore, while surgery and anesthesia last only for minutes to hours, these interventions may trigger sustained changes that impair cognition for weeks to months. Animal models are required to study the pathophysiological mechanisms that underlie POCD.

11

Figure 2.1 Perioperative factors that affect the likelihood of POCD.

Preoperative, intraoperative and postoperative factors interact to modify and exacerbate

POCD. For example, hypothermia, stress, sleep deprivation contribute to POCD only during the early postoperative period whereas the effects of analgesics last for days. The effects of anesthesia and postoperative inflammation may affect cognition even months after surgery.

12

2.2 Memory deficits after surgery and anesthesia in animal models

It is impossible to identify specific factors that cause POCD in studies of patients who have undergone surgery. For example, the effects of anesthesia cannot be disentangled from the myriad of other perioperative factors that alter cognition. Consequently, animal models are needed. Many of the animal studies conducted to date have focused on memory behaviours, because learning and memory deficits are primary features of POCD in patients. Also, well- validated experimental assays to study memory in animal models are available and can be used to study the effects of surgery and anesthesia on memory performance.

The results of animal studies show that the mechanisms of POCD can be broadly divided into nonanesthetic, surgical causes and anesthetic-related causes (Figure 2.2). An important surgical cause is inflammation that results from the surgical trauma and the disease that necessitated surgery. Other nonanesthetic factors, such as disrupted sleep patterns and novel hospital environments, are known to disrupt memory performance but will not be considered here. Here, I will review studies that examine memory deficits after surgery and anesthesia or after anesthesia alone in animal models. The mechanisms by which anesthetics cause memory deficits will be reviewed in particular detail as they are the focus of this thesis.

13

Figure 2.2 A simple classification scheme for the mechanisms that contribute to POCD.

The mechanisms can be broadly classified into nonanesthetic and anesthetic-related causes, which have been shown to be triggered by exposure to anesthesia alone in animal models.

14

2.2.1 Memory deficits after surgery in animal models

In an effort to understand the surgical causes of POCD, rodent models have been subjected to surgery, anesthesia and analgesia regimens similar to those that are used for treating human patients. Laparotomy, partial hepatectomy and tibial osteotomy surgeries have been performed in rodents and memory performance has been studied at different time intervals after surgery (Cao et al. 2010; Chi et al. 2013; Cibelli et al. 2010).

Rodents exhibit retrograde and anterograde memory deficits after surgery as evidenced by impaired recall of tasks learned before surgery and impaired learning of new tasks after surgery. In aged rats (22-24 months) partial hepatectomy triggers retrograde spatial memory impairment in the Morris Water Maze task for up to 7 days (Cao, et al.

2010). Older rats also exhibit impaired anterograde memory after surgery and anesthesia; for example, impaired spatial memory in the radial arm maze after laparotamy (Chi, et al. 2013), impaired contextual fear memory after orthopedic surgery and impaired working memory in the Y-maze after orthopedic surgery (Hu et al. 2014). Aged mice (14 months) similarly exhibit impaired anterograde spatial memory when they are trained on the Y-maze 1 or 3 days after partial hepatectomy (Li et al. 2013). These deficits are not confined to aged animals. Adult rats (3-6 months) show impaired retrograde spatial memory 24 hours postoperatively (Cao, et al. 2010). Similarly, adult (3-3.5 month old) mice also exhibit retrograde memory impairment on hippocampus-dependent memory tasks 3 days after orthopedic surgery (Cibelli, et al. 2010). Specifically, mice were trained 30 minutes before tibial osteotomy surgery to associate a specific context and auditory tone with an aversive foot shock in a memory task called delay fear conditioning (Cibelli, et al. 2010). When reintroduced to the training context 3 days later for testing, mice exhibited impaired

15

hippocampus-dependent contextual fear memory but no impairment in hippocampus- independent cued fear memory that was cued by the auditory tone (Cibelli, et al. 2010). In these surgical models of POCD, mice do not exhibit memory deficits after anesthesia alone

(Cao, et al. 2010; Cibelli, et al. 2010). However, this may be due to the mnemonic demands

of the task that is used to detect a deficit. While surgery might trigger memory deficits on

easy tasks, subtle memory deficits that can occur after anesthesia may be detected only on

more demanding tasks; for example tasks with a shorter training time, a longer retention

interval, or tasks such as object recognition that do not utilize any external appetitive or

aversive stimuli to motivate the animal to perform the task.

2.2.1.1 Postoperative Inflammation and Memory Loss

A major cause of memory deficits after surgery is postoperative inflammation. Surgery-

induced trauma, anesthesia and the stress response can all activate the peripheral immune

system, causing the release of proinflammatory cytokines (Buchanan et al. 2008; Caza, et al.

2008; Frank et al. 2011; Hietbrink et al. 2006). Increases in cytokine levels in the plasma

and the brain can impair memory performance. After major surgery, the level of the main

proinflammatory cytokines, including tumor necrosis factor-α (ΤΝF-α), interleukin 1β (IL-

1β) and interleukin 6 (IL-6), are increased in rodents (Cibelli, et al. 2010; Ni Choileain et al.

2006; Terrando et al. 2010). Initially, TNF-α and IL-1β are released from activated macrophages and monocytes at the site of injury as part of the acute-phase response. These

factors, in turn, stimulate the production of more cytokines, especially IL-6 (Allan et al.

2005; Ni Choileain and Redmond 2006; Yirmiya et al. 2011). The cytokines originating

from the periphery exert effects on the central nervous system through both direct and

indirect pathways to cause neuroinflammation and stimulate the de novo production of

16

cytokines by glial cells within the brain (Dantzer et al. 2008; Konsman et al. 2002).

Neuroinflammation results in the activation of microglia, which secrete cytokines, eicosanoids, excitatory amino acids, oxidative radicals and neurotoxins such as amyloid-β

(Hanisch et al. 2007), all of which can contribute to cytopathophysiological changes and cognitive deficits after surgery. At high levels, cytokines released after surgery can disrupt the blood-brain barrier (Hu, et al. 2014), which renders the brain more vulnerable to toxicity from amino acids, toxins and anesthetics that are present in the circulation.

Rodent models have shown that peripheral surgery can induce neuroinflammation and consequently, postoperative memory deficits. In rats, partial hepatectomy induces a neuroinflammatory response as brain levels of IL-1β and IL-6 were increased (Cao, et al.

2010). This study also reveals an age-related increase in susceptibility to cognitive impairment, with aged animals exhibiting a greater increase in the expression of proinflammatory cytokines and glial cell activation, as evidenced by increased expression of the glial cell markers glial fibrillary acidic protein (GFAP) and S100 calcium binding protein

B (S100B) (Cao, et al. 2010). Similarly, after orthopedic surgery, mice exhibit microglial activation and increased expression of the cytokines IL-1β, IL-6 and TNF-α in the brain

(Cibelli, et al. 2010; Terrando, et al. 2010). TNF-α acts upstream of IL-1β and initiates the release of other cytokines (Terrando, et al. 2010). The level of TNF-α in the blood is transiently increased 30 minutes after surgery, whereas plasma levels of IL-1β and IL-6 are increased 6 to 24 hours after surgery (Terrando, et al. 2010).

The increase in cytokine levels is paralleled by memory deficits and likely contributes to their etiology. Memory for contextual and trace fear conditioning tasks (hippocampus-

17

dependent tasks) is impaired in mice with increased cytokine levels after orthopedic surgery

(Cibelli, et al. 2010; Terrando, et al. 2010). Importantly, no memory deficits were observed

when the interleukin-1 receptor (IL-1R) was inhibited pharmacologically after orthopedic

surgery and the deficits were absent in IL-1R null-mutant mice (Cibelli, et al. 2010). This

important finding indicates that cytokines, through direct actions on IL-1R, contribute to

postoperative cognitive deficits (Cibelli, et al. 2010). Inflammation could also be limited in

vivo by prophylactic administration of a monoclonal antibody to TNF-α (Terrando, et al.

2010). After administration of the antibody, the expression of IL-1β and microgliosis in the brain were reduced and postoperative memory impairment was prevented (Terrando, et al.

2010).

Given the putative causal role of IL-1β in the genesis of memory deficits after surgery, it seems reasonable to simply block the cytokine receptor to prevent memory deficits. However, IL-1β contributes to the patient’s immune system function. Blocking the

IL-1R might increase the risk of and inhibit wound healing (Fleischmann et al.

2006). Also, IL-1β plays a physiological role in memory processes, whereby low basal levels of this cytokine promote memory (Avital et al. 2003). Since high levels of IL-1β that

occur during inflammation cause memory loss, future studies are required to identify the intracellular pathways through which cytokines act to modulate the expression of transcription factors and the activity of neurotransmitter receptors to cause learning and memory deficits. These downstream pathways could be targeted as specific treatment

strategies for memory deficits that do not interfere with normal wound-healing.

18

2.2.2 Memory deficits after anesthesia in animal models

Preclinical studies have overwhelmingly shown that anesthesia in the absence of surgery

triggers cognitive deficits that persist for days to weeks (Culley, et al. 2003; Culley, et al.

2004). This thesis will focus on a mechanism by which anesthetic exposure triggers lasting

memory deficits in an adult murine model.

In adult and aged rodents, a single anesthetic can cause long-lasting impairments on

memory tasks (Culley et al. 2004; Culley, et al. 2004). For example, young and aged rats

show when they are trained on the radial arm or Morris Water Maze

spatial memory tasks 1 to 2 weeks after exposure to isoflurane or isoflurane in combination

with nitrous oxide (Culley, et al. 2004; Lin et al. 2011). Mice exposed to a single isoflurane

treatment or multiple sevoflurane treatments also exhibit anterograde memory deficits in

contextual fear conditioning tasks (Le Freche et al. 2012; Saab, et al. 2010). Retrograde

memory impairment also occurs in adult mice on an object recognition task acquired before

sevoflurane anesthesia and in aged (18-month old) rats on a spatial memory task that was

acquired 2 hours before isoflurane anesthesia (Culley, et al. 2003; Wiklund, et al. 2009).

These rodent models have been used to study the lasting neurotoxic effects of anesthesia and

to investigate possible mechanisms that may underlie postanesthetic memory deficits.

2.3 Proposed mechanisms of postanesthetic memory deficits in animals

In the healthy, adult brain, the mechanisms by which anesthetics trigger long-lasting memory deficits are only beginning to be elucidated. Microarray analysis shows that the mRNA expression of over 400 is changed 48 hours after sevoflurane anesthesia (2.5% for 6 hours) in young-adult, 6 week-old rats (Pan et al. 2011). The expression of genes involved in

19

apoptosis, cell metabolism, signal transduction and vesicular release is altered (Pan, et al.

2011). Studies using adult and aged rodents have identified several mechanisms that may

contribute to memory loss after anesthesia, including inflammation and apoptosis, impaired

neurogenesis, and an increase in the production of Alzheimer-related proteins.

2.3.1 Inflammation and activation of apoptotic pathways

Exposure to certain anesthetics, such as isoflurane, triggers an increase in expression of

markers of neuroinflammation (Kong et al. 2013; Lin and Zuo 2011; Wu et al. 2012).

Exposure to isoflurane, at clinically relevant concentrations, activates inflammatory signaling

pathways in the adult brain as evidenced by increased levels of mRNA and protein for the

proinflammatory cytokines TNF-α, IL-6 and IL-1β (Kong, et al. 2013; Lin and Zuo 2011;

Wu, et al. 2012). Specifically, exposure of adult (5-8 month old) mice to 1.4% isoflurane for

2 hours results in increased mRNA and protein expression of TNF-α and IL-6 6 to 24 hours

after anesthesia and increased expression of IL-1β 12 to 24 hours after anesthesia (Wu, et al.

2012). Similarly, exposure of adult and aged rats to isoflurane results in an upregulation of

IL-1β (Kong, et al. 2013; Lin and Zuo 2011).

Elevated cytokine levels after isoflurane may cause neurotoxicity by stimulating

apoptosis and are associated with increased expression of the proapoptotic enzyme caspase-3

and neuronal loss in the CA1 subfield of the hippocampus (Kong, et al. 2013; Lin and Zuo

2011). Certain brain regions that are crucial for the acquisition of memory in rodents,

including the CA1 and dentate gyrus subfields of the hippocampus, as well as the olfactory

bulb, are particularly vulnerable to anesthetic-induced apoptotic cell death (Deng et al. 2014;

Hofacer et al. 2013; Lin and Zuo 2011). Anesthetic-induced apoptosis is likely calcium- dependent as the anesthetics isoflurane and halothane both trigger an increase in intracellular

20

calcium in hippocampal neurons in brain slices (Kindler et al. 1999). Isoflurane has also been shown to trigger the release of calcium from intracellular stores in the endoplasmic reticulum via inositol triphosphate receptors (Wei et al. 2008). Furthermore, isoflurane- induced apoptosis can be attenuated by calcium chelators such as 1,2-bis(o- aminophenoxy)ethane-N,N,N',N'-tetraacetic acid (BAPTA) in cultured H4 human neuroglioma cells, and in vivo in adult mice by blocking the calcium-permeable NMDA receptor with the partial antagonist memantine (Zhang et al. 2008).

Anesthetic-induced apoptosis is correlated with behavioral memory deficits (Kong, et

al. 2013; Lin and Zuo 2011; Mawhinney et al. 2012). In adult rats, an increase in caspase-3

activation 16 hours after isoflurane anesthesia is associated with reduced neuronal density in

the CA1 region of the hippocampus and impaired performance in terms of contextual fear

memory and in terms of spatial memory in the Barnes maze 2 weeks after anesthetic

exposure (Lin and Zuo 2011). Apoptosis, as measured by the number of caspase-3 –positive

cells, is also strongly correlated with poor performance on the Morris water maze task 48

hours after anesthesia (Valentim et al. 2010). In aged (18-month-old) mice, caspase-3 activation in the cortex and in the hippocampus is associated with spatial memory impairments in the Morris water maze even 3 months after isoflurane and nitrous oxide anesthesia (Mawhinney, et al. 2012). Reducing inflammation with the antibiotic and anti- inflammatory drug minocycline, administered to rats before isoflurane, decreases the expression of IL-1β and caspase-3, reduces apoptotic cell death and improves spatial memory performance on the Morris water maze (Kong, et al. 2013). Similarly, pretreatment of aged rats with the non-steroidal anti-inflammatory drug parecoxib, before sevoflurane anesthesia, improves performance on the Morris water maze test (Gong et al. 2012). Together, these

21

findings suggest that inflammation and apoptosis caused by some anesthetics contributes to

long-term cognitive deficits in the adult brain. It is important to note that not all anesthetics

have been shown to cause inflammation and certain anesthetic and sedative agents, such as

xenon and dexmedetomidine, have actually been shown to reduce release of proinflammatory

cytokines (Sanders et al. 2009; Tanabe et al. 2014; Vizcaychipi et al. 2011).

2.3.2 Impaired neurogenesis

In addition to causing apoptotic neuronal death, isoflurane anesthesia also impairs cognition

by reducing the proliferation of new neurons in the adult brain. In young, 3 month-old rats, exposure to a 3 hour anesthesia with the drug propofol, increases the number of reactive

astrocytes, which express S100B, and decreases the number of new, differentiating neurons

(Erasso et al. 2013). In addition to reducing the number of new neurons labeled with S-phase

marker bromodeoxyuridine, propofol also reduces the length of dendritic branches (Krzisch

et al. 2013). In older, 12 month-old rats, isoflurane similarly increases the number of

activated astrocytes and reduces the number of differentiating neurons (Erasso, et al. 2013).

The reduction in neurogenesis likely has functional implications for cognition. Aged rats,

exposed to sevoflurane anesthesia not only exhibit a reduction in the number of nestin-

expressing progenitor cells and doublecortin protein-expressing immature neurons, but also

exhibit a decrease in the expression of phosphorylated, cAMP response element- binding

(CREB) protein, which is important for both memory consolidation and induction of

plasticity in the hippocampus (Xiong et al. 2013).

2.3.3 Alzheimer Disease-related mechanisms

Exposure to anesthesia also triggers the production of Alzheimer disease-related proteins including neurofibrillary tangles and β-amyloid plaques. Neurofibrillary tangles are

22

composed of aggregates of hyperphosphorylated tau protein while amyloid plaques are

formed upon cleavage, by β- and γ-secretases, of amyloid precursor protein into amyloid-β peptides, which then form large protein aggregates (Masters et al. 2011). The process of amyloid-β aggregation is thought, in turn, to stimulate the release of cytokines from the neuroglia and to produce reactive oxygen species that trigger apoptotic neuronal death

(Masters and O'Neill 2011).

In healthy, adult animals, inhalational anesthetics such as isoflurane increase amyloid-

β production and phosphorylation of tau protein (Liu et al. 2014; Xie et al. 2008; Zhen et al.

2009). For example, treatment of mice with isoflurane alone or isoflurane in combination with nitrous oxide causes an increase in the expression of the enzyme β-secretase, an increase in levels of amyloid precursor protein N-terminus fragments, and a subsequent increase in amyloid-β protein 6 hours after anesthesia (Xie, et al. 2008; Zhen, et al. 2009). The increase in expression of β-secretase and amyloid-β persists for at least 24 hours after anesthesia (Xie, et al. 2008). In aged rats, the anesthetic enflurane similarly causes an increase in expression of amyloid-β (Liu and Weng 2014). The anesthetic sevoflurane increases amyloid-β levels in

the brain by facilitating transport of the protein from the blood into the brain and by reducing

the degradation of newly generated amyloid-β (Liu et al. 2013). Specifically, sevoflurane

increases the expression of the receptor-for-advanced-glycation-end-products, which is involved in transporting amyloid-β from the circulation into the brain, and reduces the expression of lipoprotein receptor related protein -1, which moves amyloid-β, across the blood brain barrier, from the brain and into the blood (Liu, et al. 2013). Sevoflurane also reduces the mRNA and protein expression of the enzymes insulin-degrading enzyme and neprilysin, both of which degrade newly generated amyloid-β in the brain (Liu, et al. 2013).

23

Anesthetics, such as enflurane, also contribute to Alzheimer disease pathology by

increasing the hyperphosphorylation of the tau protein, hence enabling the formation of

neurofibrilliary tangles (Liu and Weng 2014). Phosphorylation of the tau protein is also

increased after exposure of adult (5-6 month-old) mice to sevoflurane (Le Freche, et al.

2012), and anesthetic-induced hypothermia can further enhance tau phosphorylation by

decreasing the activity of the enzyme protein phosphatase 2A (Planel et al. 2008; Planel et

al. 2007).

In transgenic animal models of Alzheimer disease, exposure to anesthetic exacerbates

Alzheimer disease pathology. In comparison to WT mice, amyloid-β expression is greatly

increased after multiple exposures to isoflurane in transgenic Tg2576 mice that overexpress

the human amyloid precursor protein double Swedish mutation, which has been linked to

Alzheimer disease (Perucho et al. 2010). These transgenic mice also exhibit increased microglial activation, a process that promotes the release of proinflammatory cytokines, and increased apoptotic neurodegeneration as evidenced by an increase in the ratio of proapoptotic Bax protein to anti-apoptotic Bcl2 protein, as well as an increase in the number

of apoptotic neurons detected with terminal deoxynucleotidyl transferase dUTP nick end

labeling (TUNEL) (Perucho, et al. 2010). The presenilin-1 mutation, which has been

associated with Alzheimer disease in patients, also results in an increased susceptibility to

isoflurane-induced neuronal death (Liang et al. 2008). Isoflurane further exacerbates

Alzheimer disease pathology by increasing tau phosphorylation in the brainstem, spinal cord

and cortex of a transgenic mouse model of tauopathy (Planel et al. 2009). In the healthy and

Alzheimer disease-compromised brain, the accumulation of Alzheimer disease proteins after

anesthesia may contribute to cognitive deficits.

24

2.3.4 Potential GABAA receptor-dependent mechanisms

The preliminary data that stimulated my work, of which I was a co-author, first showed that

inhibitory neurotransmission mediated by GABAA receptors may play a role in memory deficits after anesthesia (Saab, et al. 2010). In this study, we show that in adult mice learning and memory deficits persist for at least 48 hours while motor and sensory function recover within minutes after cessation of isoflurane anesthesia (1.3% for 1 hour) (Saab, et al. 2010).

Specifically, mice were trained on the fear conditioning memory task either 1 hour or 24 hours after anesthesia and were tested 30 minutes after training (to assess short-term memory) and 48 hours after training (to assess long-term memory) (Saab, et al. 2010). Short- term and long-term memory are impaired when mice are trained on the task 1 hour after anesthesia, whereas only short-term memory is impaired when mice are trained 24 hours after anesthesia (Saab, et al. 2010). Importantly, these memory deficits can be prevented by pretreatment with the drug L-655,708, a highly selective inverse agonist of α5 subunit- containing GABAA receptors (Quirk et al. 1996). These results strongly suggest that

α5GABAA receptors play a role in the genesis of memory deficits after anesthesia. In this

thesis I will examine the mechanisms by which GABAA receptors contribute to memory

deficits in the postanesthetic period. Since GABAA receptors are the focus of this thesis,

GABAA receptor-mediated inhibition will be described below.

2.4 GABA and GABAA receptor-mediated inhibition 2.4.1 GABA Synthesis and release

In 1950, γ-aminobutyric acid (GABA; NH2CH2CH2 CH2COOH) was identified in protein-

free extracts of rat, rabbit, guinea pig, frog and human brain (Roberts et al. 1950). Since its

discovery, GABA has been established as the primary inhibitory neurotransmitter in the

25

mammalian central nervous system. GABA is a small amino acid with no net charge at

physiological pH. It is synthesized from the amino acid glutamate in a reaction that is

catalyzed by glutamic acid decarboxylase (GAD). GAD exists in two isoforms, GAD65 and

GAD67, which are encoded by two separate genes and differ in their sequence, molecular

weight, and distribution within neurons (Kaufman et al. 1991). Both isoforms require the

presence of the cofactor pyridoxal phosphate (PLP) for their enzymatic activity. GAD67 is

present in the cytoplasm throughout the neuron, whereas GAD65 is concentrated at axon

terminals (Kaufman, et al. 1991). GAD67 is nearly always bound to its activating cofactor

PLP and may produce GABA that is used as a source of energy in the cell or released

through non-vesicular mechanisms (Kaufman, et al. 1991). In contrast, the membrane-

associated GAD65 is only half-saturated with PLP and produces GABA for use as

neurotransmitter at inhibitory synapses (Kaufman, et al. 1991). In fact, at presynaptic

terminals GAD65 is part of a complex with vesicular GABA transporter (VGAT) and other proteins thereby facilitating the transport of newly synthesized GABA into synaptic vesicles

(Jin et al. 2003).

In order for GABA to be released from presynaptic terminals it must be packaged into synaptic vesicles. GABA is transported into synaptic vesicles by VGAT (Chaudhry et al.

1998). VGAT is present at axon terminals of GABAergic interneurons and glycinergic

neurons in the brain and spinal cord and is necessary for the loading of vesicles with

inhibitory neurotransmitters (Chaudhry, et al. 1998). The release of GABA from synaptic

vesicles is stimulated by depolarization of the presynaptic neuron, which causes activation of

voltage-gated calcium channels and calcium-dependent fusion of vesicles with the cell

membrane, which allows the release of vesicle contents into the synapse. After vesicular

26 release, GABA levels in the synapse reach approximately 1.5 to 1.8 millimolar (mM)

(Barberis et al. 2004). Release of GABA also occurs from nonvesicular sources, for example through reverse activity of GABA transporters or release from astrocytes through

Bestrophin-1 channels (Lee et al. 2010; Song et al. 2013). These sources can also contribute to ambient GABA levels in the extracellular space. However, GABA concentrations in the synaptic cleft rapidly decrease and concentrations detected in the extracellular space with in vivo microdialysis in the hippocampus range from 0.2-0.8 micromolar (µM) (Lerma et al.

1986; Tossman et al. 1986).

2.4.2 GABA transport and metabolism

Following release into the extracellular space, GABA levels decrease in one of four ways.

First, GABA binds to receptors on the postsynaptic membrane. Second, GABA diffuses away from the synapse and may undergo reuptake into neurons where it may be repackaged into vesicles (Scimemi 2014). Alternatively, GABA is transported into astrocytes (Martin et al. 1998; Scimemi 2014). Lastly, extracellular GABA can be transported into cells and degraded (Madsen et al. 2008).

Reuptake of GABA into neurons and astrocytes occurs through active transport via

GABA transporters (GATs). In addition to the vesicular GABA transporter, four additional

GABA transporters exist, GAT1-4. GATs are part of the Solute Carrier 6 transporter family that uses sodium and chloride gradients to transport transmitters into cells (Scimemi 2014).

For every GABA molecule, two sodium ions and one chloride ion are also transported into the cell (Radian et al. 1983). GAT1 and GAT3 are the most abundant transporters and are present throughout the brain, whereas GAT2 and GAT4 are expressed in the meninges

(Radian et al. 1990). The transporters also differ in their distribution in different cell types.

27

GAT1 is mostly expressed on presynaptic axon terminals and GAT1-specific staining overlaps with staining for the GABA-synthesizing enzyme GAD67 (Minelli et al. 1995). On

the other hand, GAT3 is expressed in astrocytes in rodents and in astrocytes and

oligodendrocytes in humans (due to species-specific differences in nomenclature, GAT3 in

humans and rats is the same as GAT4 in mice) (Pow et al. 2005). In the thalamus, the

distance from the synapse may determine which transporter is expressed and regulates the

levels of ambient GABA. GAT1 is localized perisynaptically while GAT3 is expressed in

extrasynaptic regions (Beenhakker et al. 2010). Indeed, the intensity of GAT1 staining

decays rapidly as the distance from GABAergic synapses, labeled with gephyrin or VGAT,

increases (Beenhakker and Huguenard 2010). This suggests that GAT1-mediated reuptake

may be used to recycle GABA for neurotransmission.

Following reuptake, GABA can be degraded by the catabolic enzyme GABA

transaminase (GABA-T) (Madsen, et al. 2008). GABA-T degrades GABA to succinic

semialdehyde, which is an intermediate in the Kreb’s cycle. Thus, GABA can be used for the

metabolic needs of the cell (Madsen, et al. 2008). However, arguably the most important

effect of extracellular GABA is its inhibitory effect on neuronal excitability that is mediated

by GABA binding to GABA receptors on the neuronal membrane.

2.4.3 GABA receptors

At inhibitory synapses in the central nervous system, GABA mediates the majority of

inhibitory neurotransmission by binding to GABA receptors on the postsynaptic membrane.

GABA receptors can be classified as 1) ionotropic GABAA receptors that mediate a fast

response to GABA (Farrant et al. 2005) or 2) G-protein coupled, metabotropic GABAB

receptors that mediate a slow response to GABA (Padgett et al. 2010). GABAB receptors are

28

made up of GABAB1 and GABAB2 heterodimers. Upon agonist binding at the GABAB

receptor, the exchange of GDP to GTP occurs and initiates dissociation of the G protein

dimer into Gαi/o and Gβγ subunits (Padgett and Slesinger 2010). Gαi/o affects second messenger systems within the cell by inhibiting the enzyme adenylyl cyclase and thereby reducing levels of cyclic adenyl monophosphate (cAMP) in the cell (Padgett and Slesinger

2010). The Gβγ subunit can inhibit P, Q and N-type voltage-gated calcium channels and activate G protein-gated inwardly rectifying K+ (GIRK) channels, thereby preventing Ca2+

influx and stimulating K+ efflux to cause neuronal hyperpolarization (Padgett and Slesinger

2010). GABAB receptors will not be considered further as GABAA receptors are the focus of

this thesis.

2.4.4 General overview of GABAA receptors

GABAA receptors are ligand-gated, ionotropic receptors that are part of the Cys-loop

superfamily of receptors that also includes nicotinic acetylcholine receptors, glycine

receptors, 5-hydroxytryptamine 3 (5HT3) receptors and zinc-activated channels (Lester et al.

2004). Although the human GABAA receptor, a β3 homopentamer that is not found

endogenously, was recently crystallized while bound to a novel agonist (Miller et al. 2014),

our current understanding of GABAA receptor structure is based on the crystal structure of

the nicotinic acetylcholine receptor. The GABAA receptor is a pentameric channel with the five subunits forming a central channel pore (Figure 2.3). Each subunit of the GABAA

receptor has a long, extracellular N-terminal domain, four transmembrane α-helical domains

(TM1-4) and an extracellular C-terminal domain (Olsen et al. 2009). The N-terminal domain is important for ligand binding, with five binding pockets present at the five interfaces between subunits. There are additional binding sites within each subunit with each binding

29

site surrounded by the 4 transmembrane domains (Olsen and Sieghart 2009). Most

importantly, the two GABA binding sites are located between the α and β subunits (Figure

2.3). The N-terminal domain is also important for the binding of other allosteric modulators and for the assembly of subunits into GABAA receptors (Olsen and Sieghart 2009). The

intracellular loop between TM3 and TM4 is the site of regulation by phosphorylation and

binding of anchoring proteins that sequester the receptor at the membrane (Olsen and

Sieghart 2009). The TM2 domain forms the wall of the channel pore (Olsen and Sieghart

2009), through which the entry of anions into the cell mediates neuronal inhibition in mature

neurons.

30

Figure 2.3 The GABAA receptor.

The GABAA receptor is a pentameric, ligand-gated . The five subunits are

arranged around a central channel pore. The typical configuration of subunits is 2 α subunits,

2 β subunits and one γ subunit. In some populations of extrasynaptic receptors the γ subunit

is replaced by a δ subunit. The GABA binding sites are located at the extracellular interface

between α and β subunits. In adult neurons, GABA binding to the receptors triggers channel

opening and chloride (Cl−) influx into the cell.

31

2.4.5 Subunit composition of GABAA receptors

GABAA receptors are composed of five subunits that are arranged in various combinations.

There are 19 mammalian genes encoding GABAA receptor subunits: α1-6, β1-3, γ1-3, δ, ε, θ,

π, and ρ1-3. While some subunits are expressed throughout the central nervous system, the

expression of others is more restricted. For example, ρ1-3 subunits are expressed only within the retina and GABAA receptors comprised of these subunits have previously been

categorized into their own class as GABAC subunits (Boue-Grabot et al. 1998). The mRNA

for ρ subunits has also been found in the superior colliculus, lateral geniculate nucleus and

cerebellar Purkinje cells (Boue-Grabot, et al. 1998). The subunits α1, β1, β2, β3, and γ2 are

found throughout the brain. In contrast, the expression of α2, α3, α4, α5, α6, and γ1 is more

concentrated in specific brain regions (Olsen and Sieghart 2009).

Although theoretically hundreds of subunit combinations are possible, only a couple

of dozen have been shown to exist in native neurons (Olsen and Sieghart 2009). To

determine which subunits partner together, immunocytochemical and electron microscopy

studies have been used to study the colocalization of subunits. These studies have revealed that most receptors are composed of α, β, and γ subunits (Fritschy et al. 1992; Somogyi et al.

1996). In addition, coimmunoprecipitation of GABAA receptors from brain tissue shows that

receptors are typically composed of 2α, 2β and 1γ or 1δ subunit as two different α or β

subunits could be coimmunoprecipitated from brain tissue or colocalized together in tissue

slices, but two different γ subunits or a γ and δ could not be precipitated or colocalized

together (Bohlhalter et al. 1996; Olsen and Sieghart 2009). In the brain, the most common

combination of subunits is α1βxγ2 (Olsen and Sieghart 2009). While many receptors contain two identical α subunits and two identical β subunits, colocalization studies in

32

cerebellar granule cells showed that individual receptors can contain α1 and α6 subunits

(Nusser et al. 1998). Most receptors have a γ or δ subunit, however, mounting evidence suggests that receptors comprised of only α and β subunits exist. These αβ receptors have been detected in δ and α1 null-mutant mice (Ogris et al. 2006; Tretter et al. 2001). Further evidence is that a subset of α4 receptors is not colocalized with γ or δ subunits (Bencsits et al.

1999). αβ receptors have also been identified using whole-cell electrophysiology.

Specifically, a tonic conductance that is insensitive to benzodiazepines is recorded, hence the receptors that mediate the conductance do not contain the γ subunit, however, they also do

not contain the δ subunit as they have a much lower single channel conductance than δ-

containing receptors (11 pS versus 25-28 pS in δGABAA receptors) (Mortensen et al. 2006).

The subunit composition is crucial to determining how the GABAA receptor function is

regulated, as different subunits bind different GABAA receptor-associated proteins and their activity is differentially regulated by protein kinases and phosphotases.

2.4.6 GABAA receptor-associated proteins

The cytosolic loop between transmembrane domains 3 and 4 of each subunit is subject to

modification by kinases, phosphatases and the binding of GABAA receptor-associated proteins (Chen et al. 2007). Hence, it is important for the regulation of GABAA receptor

function and expression. A number of GABAA receptor cofactors have been identified through yeast two-hybrid assays that are used to identify protein-protein interactions (Chen and Olsen 2007). The most studied cofactor is GABAA receptor-associated protein

(GABARAP), which binds to the γ1-3 subunits (Chen and Olsen 2007). It is likely involved

in intracellular trafficking of GABAA receptors as immunohistochemistry and electron microscopy studies have found that it is colocalized with GABAA receptors in the

33

endoplasmic reticulum, Golgi apparatus and on intracellular vesicles (Chen and Olsen 2007).

Additionally, GABARAP interacts with other factors that are involved in GABAA receptor

trafficking and anchoring at the cell membrane including gephyrin, N-ethylmaleimide sensitive factor, which is important for membrane fusion of intracellular vesicles, and tubulin

(Chen and Olsen 2007).

Several other proteins have been implicated in the intracellular trafficking of GABAA

receptors. Brefeldin-A inhibited GDP/GTP exchange factor2 binds to the intracellular loop of

all β subunits (Chen and Olsen 2007). It promotes GDP to GTP exchange on ADP- ribosylation factors, which in turn activates these factors and initiates membrane budding of the Golgi apparatus and likely promotes the trafficking of receptor proteins from the Golgi apparatus towards the cell surface (Chen and Olsen 2007; Lüscher et al. 2004). Similarly,

the protein Golgi-specific DHHC zinc finger protein (GODZ, or alternatively, palmitoyl

acyltransferase-DHHC-3) facilitates receptor trafficking as it binds to the γ2 subunit and

mediates its palmitoylation (Chen and Olsen 2007; Keller et al. 2004; Lüscher and Keller

2004). This modification promotes intracellular trafficking of the receptor and helps stabilize

it at the cell membrane.

Other proteins that interact with GABAA receptors include Proteins Linking Integrin-

associated protein and Cytoskeleton (Plic-1), Adaptin complex (AP2) and Huntingtin

Associated Protein (HAP1). Plic-1 binds to the intracellular loop of α1, α2, α3, α6 and β1-3 subunits and prevents the degradation of intracellular GABAA receptors through ubiquitin-

dependent mechanisms (Bedford et al. 2001). AP2 binds to β and γ subunits and mediates

clathrin-mediated endocytosis of receptors (Kittler et al. 2008). HAP1 binds to the β1

34

subunit and likely also regulates trafficking of GABAA receptors as genetic inhibition of

HAP1 resulted in reduced GABAA receptor expression (Kittler et al. 2004).

Arguably, the most important GABAA receptor-associated proteins are those that

sequester GABAA receptors at the cell membrane. Thus far, two proteins have been

identified: gephyrin and radixin. Gephyrin is as an anchoring protein at glycinergic and

GABAergic synapses and it is important for clustering of inhibitory receptors at the synapse

(Tretter et al. 2012). Gephyrin null-mutant mice exhibit a significant reduction in GABAA

receptor clustering with clusters of α2 and γ2 subunits completely absent in hippocampal

neurons (Kneussel et al. 1999). Gephyrin is made up of 3 domains; N-terminal G domain,

the C-terminal E domain, and the C domain that links the other two domains (Tretter, et al.

2012). The E domain contains the binding site for both glycine and GABAA receptors

(Tretter, et al. 2012; Tyagarajan et al. 2014). Only 4 GABAA receptor subunits have been

colocalized with gephyrin, α1-3, and γ2, and it is believed that gephyrin binds to the α subunit (Tretter, et al. 2012; Tyagarajan and Fritschy 2014). In fact the gephyrin binding

domain was identified on the α2 subunit, as deletion of this domain prevents recombinant α2

subunits from clustering with gephyrin in cultured neurons (Tretter et al. 2008). In contrast,

introduction of the gephyrin-binding sequence into α6 subunits, which do not normally

associate with gephyrin, or into an unrelated protein, promotes the clustering of the proteins

with gephyrin in cultured neurons (Tretter, et al. 2008). While only α1-3 subunits have been

confirmed to bind to gephyrin, other subunits may interact as well. In one study, 34% of

GABAA receptor clusters containing the α5 subunit colocalized with gephyrin and not the

other anchoring protein, radixin, which will be discussed below (Loebrich et al. 2006). The

G terminal domain is important for spontaneous aggregation of gephyrin into trimers that

35

form a scaffold, which binds to the actin sytoskeleton and helps sequester receptors at the

plasma membrane (Tyagarajan and Fritschy 2014).

The other major anchoring protein that has been identified is radixin. Radixin binds exclusively to α5 subunits and anchors them to the cytoskeleton (Loebrich, et al. 2006).

Radixin was identified as a binding partner to the α5 subunit using a yeast two-hybrid assay that also showed it does not interact with any other GABAA receptor subunits (Loebrich, et al. 2006). The N-terminus of radixin interacts with the cell membrane and phosphorylation of the C-terminus enables a conformational change in the protein that exposes the F-actin binding site and thereby anchors α5 subunits to the actin cytoskeleton (Tretter, et al. 2012).

In contrast to gephyrin, immunohistochemistry studies show that 87% of radixin is extrasynaptic (Loebrich, et al. 2006). When colocalization of α5 and radixin was examined,

7.5% of clusters were located synaptically and 92.5% were located extrasynaptically

(Loebrich, et al. 2006).

These two anchoring proteins are likely not the only ones present in neurons. For

example, radixin is colocalized to only 57% of all α5GABAA receptor puncta suggesting that

the remaining α5 subunit puncta are either located in intracellular pools or are anchored at the

membrane by gephyrin or other unidentified proteins (Loebrich, et al. 2006). Additional

results from studies of gephyrin null-mutant mice also point to the existence of other

anchoring proteins. For example, clustering of α1 subunits is intact in gephyrin knock-out

mice, suggesting that gephyrin-independent mechanisms of receptor clustering on the

postsynaptic membrane may exist (Kneussel et al. 2001).

36

2.4.7 GABAA receptor mediated inhibition

Ligand-gated ion channels, such as GABAA receptors, generate a current, which can be

measured using whole-cell electrophysiology, by allowing ions to pass through the channel

pore and across the membrane (Hille 2001). Ion channels can move cations into the cell or

anions out of the cell to generate an inward current that increases the membrane potential and

can depolarize the cell (Hille 2001). Alternatively, ion channels can move cations out of the

cell and anions into the cell to generate an outward current that decreases the membrane

potential and can hyperpolarize the cell (Hille 2001). The movement of ions or the

conductance through a single, open ion channel can be described with Ohm’s law, where I is

current, g is conductance and V is voltage (Hille 2001).

I = gV

The current (I) through the ion channel is proportional to the membrane potential, or

voltage (V). For a specific membrane potential (V), a channel with a high conductance (g)

will increase the current to a greater extent than a channel with low conductance (Hille

2001). Whether an ion moves into or out of the cell is determined by the electrochemical

gradient. Specifically, the driving force for a specific ion to move across the membrane is

determined by the difference between the membrane potential (Vm) and the equilibrium

potential for that ion (E) (Hille 2001). Therefore, the driving force can be expressed as Vm –

E.

The equilibrium potential for an ion is determined exclusively by the concentrations of that ion on either side of the cell membrane. At the equilibrium potential, the net movement of the ion across the cell membrane is zero (Hille 2001). In other words, when Vm

37

= E there is no net driving force for the ion. The equilibrium potential of an ion can be

calculated using the Nernst equation (Hille 2001). For example, for the chloride ion the

Nernst equation is:

- - Where ECl is the equilibrium or reversal potential for the Cl ion, R is the

thermodynamic gas constant, T is the temperature in Kelvin, z is the valence of the ion (-1 for

chloride), F is the Faraday constant (the amount of charge in coulombs in 1 mole of ion),

- [Cl]o is the concentration of Cl ion outside of the cell and [Cl]i is the concentration of the ion

inside the cell.

Since the cell membrane is permeable to many ions, the equilibrium (or reversal)

potential of the cell needs to take account many ions and can be calculated with the Goldman

equation (Hille 2001).

In this equation Pion is the relative permeability of each ion, K represents potassium ions and

Na represents sodium ions.

- - GABAA receptors are permeable to two anions, Cl and HCO3 (Bormann 1988).

Hence, the equilibrium potential of GABAergic (EGABA) neurotransmission is dependent on

- these ions. Under baseline conditions, GABAA receptors are 5 times more permeable to Cl

38

− - than to HCO3 . Consequently, EGABA is much closer to the equilibrium potential of Cl

(Bormann 1988). As the equilibrium potential of Cl- is critically dependent on intracellular

chloride concentrations, it is important to note the factors that regulate intracellular Cl-.

The intracellular Cl- concentrations are determined by the transport of Cl- via cation-

chloride cotransporters in the cell membrane. In immature neurons, transport is dominated by

the sodium-potassium chloride cotransporter (NKCC1), which transports 1 Na+, 1 K+ and 2

Cl- ions into the cell (Ben-Ari et al. 2012). This results in increased levels of Cl- inside the

cell relative to outside the cell and a positive equilibrium potential for Cl- (Ben-Ari, et al.

- 2012). As a result, GABAA receptor activation leads to efflux of Cl from the neuron and

depolarization (Ben-Ari, et al. 2012).

During neural development, the expression of different chloride transporters in the

brain changes (Ben-Ari, et al. 2012). In the 40th gestational week in humans and after

approximately two postnatal weeks in rodent neurons, the expression of the transporter

potassium-chloride cotransporter (KCC2) begins to increase and the expression of NKCC1

begins to decrease (Ben-Ari, et al. 2012). The expression of KCC2 determines the Cl-

gradient in mature neurons (Rivera et al. 1999). Similarly to NKCC1, KCC2 is also a

symporter that moves two different ions in the same direction across the membrane.

Specifically, KCC2 moves K+ and Cl- ions out of the cell (Rivera, et al. 1999). As a result,

the relative concentration of Cl- inside the cell is lower than that outside the cell and the

- - equilibrium potential for Cl is negative. Activation of GABAA receptors results in Cl influx

into the cell (Ben-Ari, et al. 2012). This influx of Cl- in mature neurons is inhibitory either

through membrane hyperpolarization or shunting inhibition. If the GABA equilibrium

potential (which is -65 to -70 millivolts in hippocampal neurons) is lower than the resting

39

- membrane potential, influx of Cl through GABAA receptors will result in hyperpolarization.

If EGABA is equal to the resting membrane potential or between the resting membrane

potential and the potential required for action potential generation, Cl- influx will shunt

excitatory inputs (Staley et al. 1992). Shunting inhibition reduces the depolarizing effect of

excitatory events by reducing the input resistance of the cell membrane and thus, reducing

the amplitude of excitatory postsynaptic effects (Staley and Mody 1992). Experimental blockade of GABAergic neurotransmission in neurons results in a lower threshold potential for generating action potentials and an increased frequency of action potentials generated at the threshold potential (Farrant and Nusser 2005).

- - During intense activation of GABAA receptors, rapid influx of Cl can exceed Cl

extrusion, which can lead to a temporary accumulation of Cl- in the cell (Isomura et al. 2003;

- - Staley et al. 1995). This accumulation collapses the Cl gradient and the extrusion of HCO3 ions through GABAA receptors becomes dominant, thereby causing EGABA to shift toward the

more positive EHCO3- (Grover et al. 1993; Isomura, et al. 2003). As a result, a biphasic response to GABA can be observed; an initial hyperpolarization followed by depolarization.

Additionally, in mature neurons GABAA receptor activation can be depolarizing under

pathological conditions in which the function or expression of KCC2 is downregulated

(Kahle et al. 2008). This thesis will focus on GABAA receptor activation in mature,

hippocampal neurons where it is expected that normal Cl- extrusion occurs via KCC2 and

GABAA receptor activation is inhibitory.

Depending on their subunit composition, GABAA receptors are localized at varying

distances from the synapse and mediate either phasic, synaptic inhibition or tonic inhibition

(Figure 2.4).

40

2.4.8 Synaptic GABAA receptors

GABAA receptors located at the synapse mediate phasic inhibition and are activated by transient increases in GABA that is released from presynaptic terminals (Farrant and Nusser

2005). The inward flow of anions following the activation of synaptic GABAA receptors is

called an inhibitory postsynaptic current (IPSC) (Figure 2.4) (Farrant and Nusser 2005).

Studies using electron microscopy and immunogold labeling of subunits have identified

subunits that are primarily located at the synapse. Receptors containing α1-3 subunits in

conjunction with one of the γ1-3 subunits are typically located at the synapse (Farrant and

Nusser 2005). Common receptor configurations within the synapse are α1β2/3γ2, α2β2/3γ2,

and α3β2/3γ2 (Farrant and Nusser 2005). While γ2-containing receptors are expressed

throughout the brain, the expression of γ1 and γ3 is low and restricted with γ1 expressed in

the pallidum, substantia nigra and amygdala, and the γ3 subunit expressed primarily in the

hippocampus and cerebral cortex (Farrant and Nusser 2005). Notably, none of the receptor

subtypes have been found exclusively at the synapse, with puncta of every subunit expressed

to some extent perisynaptically (Farrant and Nusser 2005).

Vesicular mechanisms of GABA release can produce thousands of GABA molecules

per vesicle and raise the concentration of GABA in the synapse to 1.5-1.8 mM (Mozrzymas

et al. 2003). GABA concentrations in the synapse are reduced rapidly, within approximately

100 µs, as GABA diffuses away from the synapse. GABAA receptors can exist in one of

three states: closed, open (GABA bound and ions flow through the channel) and desensitized

(GABA is bound but the channel is closed) (Farrant and Nusser 2005). The transition

between states depends on the presence of GABA and the kinetic properties of the receptor,

41

which are determined by the subunit composition and phosphorylation state of the receptor

(Farrant and Nusser 2005).

The properties of the channel can be measured electrophysiologically in terms of the

concentration of GABA that is required to produce a half-maximal response (EC50), the rate of onset of the current after GABA binding, the rate and length of desensitization in the

presence of GABA, the rate of deactivation after GABA is removed, the mean open and

closed times and the single-channel conductance (Farrant and Nusser 2005). Two molecules of GABA are required to fully activate the GABAA receptor. The release of a single vesicle

results in a miniature IPSCs (mIPSC) that can be recorded from the postsynaptic cell (Farrant

and Nusser 2005). mIPSCs have a fast rise time in the range of a couple hundred

microseconds (Farrant and Nusser 2005). The decay time of IPSCs is primarily dependent

on the closing of the channel after GABA has dissociated (Farrant and Nusser 2005). The potency of GABA at different receptors is most strongly affected by the α subunit. In general, synaptic receptors have a lower affinity for GABA and hence higher EC50 concentrations are

required to produce the half-maximal response (Farrant and Nusser 2005). Synaptic α1-3

and γ-containing receptors have EC50 values of 120 to 160 µM of GABA (Karim et al.

2013). The kinetics of the channel are also dependent on the subunit composition, with activation rates highest in α3-containing receptors, followed by α1 and then α2 (Lavoie et al.

1997). Similarly, the α subunit influences deactivation rate (Lavoie, et al. 1997). For example, α1GABAA receptors deactivate five times faster than α2GABAA receptors

(McClellan et al. 1999). Desensitization affects how quickly GABAA receptors can open

again. Synaptic GABAA receptors that contain the γ subunit desensitize much faster and for

longer than extrasynaptic receptors that contain the δ subunit (Farrant and Nusser 2005).

42

Desensitization is also influenced by the α subunit, with the fastest desensitization in

α1GABAA receptors (Tia et al. 1996).

2.4.9 Extrasynaptic GABAA receptors

Extrasynaptic GABAA receptors mediate tonic inhibition, which as the name implies, is relatively constant within the neuron. In electrophysiological experiments, the current can only be revealed by applying a GABAA receptor antagonist, which then reduces the holding

current that is required to clamp the neuron at a certain membrane potential and reduces electrical noise associated with random opening of these channels (Figure 2.3) (Bright et al.

2013). Tonic inhibition can be measured throughout the brain, in the hippocampus, thalamus, hypothalamus, layers I-V of the cortex, the amygdala, pons, medulla, spinal cord, olfactory bulb and retina (Lee et al. 2014). The magnitude of tonic inhibition varies across brain regions, ranging from 0.5 picoamperes (pA) to approximately 40 pA, and is based on the subunit complement of extrasynaptic receptors and the concentrations of ambient GABA

(Lee and Maguire 2014).

The receptors that mediate tonic inhibition contain the δ subunit (in combination with

α4, α6 or α1 and a β subunit) instead of a γ subunit, or alternatively contain an α5 subunit paired with two β3 subunits and a γ2 subunit (Farrant and Nusser 2005). The δ subunit has

not been detected at synapses, but has been detected extrasynaptically on the soma and

dendrites (Nusser, et al. 1998). Similarly, about 75% of α5GABAA receptors are located at extrasynaptic sites (Loebrich, et al. 2006). The majority of tonic inhibition is mediated by

α4,δ-containing GABAA receptors, with the exception of tonic inhibition in the cerebellum,

which is mediated by α6,δ-containing receptors, and α1,δ-containing receptors that contribute

to tonic inhibition in the dentate gyrus of the hippocampus (Chandra et al. 2006; Glykys et

43 al. 2007; Hanchar et al. 2005). In addition, in the CA1 region of the hippocampus,

α5GABAA receptors are the primary mediators of tonic inhibition (Caraiscos et al. 2004).

The receptors that mediate tonic inhibiton are located on principal neurons and on interneurons (Lee and Maguire 2014). In the mature brain, tonic inhibition of principal neurons reduces the excitability of these principal neurons, whereas tonic inhibition of interneurons increases the excitability of principal neurons by disinhibiting the principal neurons (Lee and Maguire 2014).

Extrasynaptic receptors are activated by extracellular, ambient GABA in the nanomolar range (Farrant and Nusser 2005). Ambient GABA originates from spillover from spontaneous or action potential-mediated synaptic release, reverse activity of GABA transporters and non-vesicular release from astrocytes (Glykys et al. 2007; Lee, et al. 2010;

Song, et al. 2013). GABA spillover is likely not the principal contributor to tonic inhibition as tonic currents can be recorded even when action potentials and calcium entry into presynaptic neurons, which are both needed for vesicular release of neurotransmitter, are blocked pharmacologically (Farrant and Nusser 2005). In comparison to synaptic receptors, extrasynaptic receptors have a much higher affinity for GABA and thus, a lower EC50 with

EC50 values ranging from the nanomolar range in δGABAA receptors to 30 µM in α5GABAA receptors (Karim, et al. 2013). Currents mediated by δGABAA receptors have faster rise times than those mediated by GABAA receptors that contain the γ subunit (Haas et al. 1999).

The activation of individual extrasynaptic receptors persists for much longer than that of synaptic receptors, as extrasynaptic receptors are much slower to desensitize (Bianchi et al.

2002; Haas and Macdonald 1999). The high affinity of extrasynaptic receptors for GABA, the resistance to desensitization and hence, the prolonged open time, all cause a much larger

44

charge transfer through extrasynaptic receptors than through synaptic receptors over a fixed period of time (Bai et al. 2001).

α5GABAA receptors will be the focus of this thesis as they are highly expressed in the

hippocampus (Pirker et al. 2000) and have been implicated in normal learning and memory

processes (Martin, et al. 2010). Throughout the brain, α5GABAA receptors comprise less

than 5% of all GABAA receptors, however, in the hippocampus they make up 25% of all

GABAA receptors (Pirker, et al. 2000). While the majority of α5GABAA receptors are

located extrasynaptically, several studies show that α5GABAA receptors can be expressed

both extrasynaptically and synaptically on the dendrites of hippocampal pyramidal neurons

and are enriched in postsynaptic-density fractions of cultured neurons (Christie et al. 2002;

Loebrich, et al. 2006; Serwanski et al. 2006). In the CA1 region, α5GABAA receptors

mediate the majority of tonic inhibition, as the tonic current is reduced by approximately

88% in Gabra5-/- neurons (Caraiscos, et al. 2004). In addition, synaptic α5GABAA

receptors contribute to IPSCs, which in hippocampal pyramidal neurons can be reduced by

the α5GABAA receptor-selective inverse agonist (α5IA) (Ali et al. 2008). In comparison to

synaptic α1GABAA receptors, α5GABAA receptors generate a lower amplitude current (over

2.4-fold lower), desensitize slower (approximately 2.3-fold slower) and have a higher affinity

for GABA (Caraiscos, et al. 2004; Yeung et al. 2003). The role of these receptors in

physiology and pathophysiology will be considered below.

45

Figure 2.4 A schematic drawing representing phasic and tonic inhibition mediated by

GABAA receptors as recorded from a neuron during whole-cell voltage-clamp recording.

The baseline is the holding current required to clamp the cell at a given resting membrane potential, for example -60 mV. Phasic inhibition mediated by synaptic receptors is recorded as miniature inhibitory postsynaptic currents (mIPSCs). mIPSCs represent the response of a population of GABAA receptors to the release of neurotransmitter from a single vesicle.

Application of the competitive GABAA receptor antagonist bicuculline inhibits both synaptic

receptors and hence, mIPSCs, as well as extrasynaptic receptors that mediate tonic inhibition.

The amplitude of the tonic inhibitory current can be measured as a shift from baseline after

application of bicuculline.

46

2.4.10 The physiological role of α5GABAA receptors

As the expression of α5GABAA receptors is highest in the hippocampus (Pirker, et al. 2000),

their function in this brain region and in learning and memory behaviours has been

extensively studied. In the CA1 region of the hippocampus, tonic inhibition is mediated primarily by α5GABAA receptors (Caraiscos, et al. 2004). α5GABAA receptors regulate the

excitability of hippocampal principal cells, as the depolarizing current that is required to

generate an action potential is two times greater in WT than in Gabra5-/- neurons (Bonin et

al. 2007). α5GABAA receptors also modulate hippocampal network activity, with larger

amplitude γ oscillations observed in Gabra5-/- mice (Towers et al. 2004). These fast oscillations (20-80 Hertz) in hippocampal electroencephalogram (EEG) activity may play a role in learning and memory processes (Towers, et al. 2004). Further support for the role of

α5GABAA receptors in memory formation comes from the study of long-term potentiation

(LTP) (Martin, et al. 2010). LTP of excitatory glutamatergic synaptic transmission is considered to be the network substrate of memory. α5GABAA receptors regulate the

threshold of stimulation that is required to induce LTP (Martin, et al. 2010). Specifically,

α5GABAA receptors inhibit the induction of LTP when hippocampal slices are simulated at

10 ‒ 20 Hertz (Hz), a frequency range that is associated with the acquisition of new memory in vivo (Buzsaki 2005; Martin, et al. 2010). After stimulation of the Schaffer collateral pathway in the hippocampus at 10 Hz, slices from Gabra5-/- express LTP, whereas slices from WT mice exhibit long-term depression (LTD) of synaptic efficacy (Martin, et al.

2010). Thus, the tonic current generated by α5GABAA receptors in WT mice increases the

threshold of stimulation required to induce LTP.

47

Behavioural studies support the role of α5GABAA receptors in memory processes.

Two genetic models have been used to study these receptors in learning and memory; mice

with a point mutation in the Gabra5 gene (α5H105R) that decreases α5GABAA receptor

expression in hippocampal pyramidal neurons, and Gabra5-/- mice (Crestani, et al. 2002). In

comparison to WT controls, α5H105R mice exhibit superior performance on the trace fear

conditioning task, as they spend 20% more time freezing in response to the aversive context

(Crestani, et al. 2002). Additionally, α5H105R mice but not WT mice, learn an appetitive

trace conditioning task (Yee et al. 2004). The α5H105R mice also exhibit reduced latent

inhibition and reduced prepulse inhibition of the auditory startle response (Gerdjikov et al.

2008; Hauser et al. 2005). Similarly, Gabra5-/- mice show superior performance on the

demanding trace fear conditioning task and on the spatial, Morris water maze test than WT

mice (Collinson et al. 2002; Martin, et al. 2010). Pharmacological inhibition of α5GABAA

receptors also improves performance of WT mice on trace fear conditioning, radial arm maze

and Morris Water Maze memory tasks (Dawson et al. 2006; Koh et al. 2013; Martin, et al.

2010). Importantly, these pharmacological studies demonstrate that α5GABAA receptors are

important for the encoding or acquisition as well as the recall of memory (Collinson et al.

2006). Inhibiting α5GABAA receptors by treating mice with the inverse agonist α5IA-II

immediately before learning or during memory recall, but not during memory consolidation

(the retention period), improves performance on the Morris water maze (Collinson, et al.

2006). Conversely, pharmacological enhancement of α5GABAA receptor activity impairs memory performance (Cheng, et al. 2006). Specifically, very low, amnestic doses of the anesthetic etomidate enhance tonic inhibition in the hippocampus, impair memory

48

acquisition on the fear conditioning and Morris water maze tasks, and prevent the induction

of LTP in WT but not Gabra5-/- mice (Cheng, et al. 2006; Martin et al. 2009).

2.4.11 The role of α5GABAA receptors in pathophysiology

Since α5GABAA receptors are important for normal learning and memory processes, their

activity or expression may be altered in pathological conditions. In humans and animals,

α5GABAA receptors have been implicated in a variety of diseases including Down

syndrome, autism, stroke and inflammation (Clarkson et al. 2010; Martinez-Cue et al. 2013;

Wang et al. 2012).

In humans, is caused by trisomy of 21 and results in

mental retardation. The Ts65Dn mouse model of Down syndrome has reduced LTP in the

hippocampus, an increased number of inhibitory synapses and impaired memory performance on the Morris water maze and object recognition tasks (Braudeau et al. 2011;

Martinez-Cue, et al. 2013). α5GABAA receptors may mediate these deficits in learning and memory in the Ts65Dn model, as treatment of Ts65Dn mice with the α5GABAA receptor

inverse agonist α5IA reverses deficits in spatial and recognition memory (Braudeau, et al.

2011). In addition, treatment with the inverse agonist enhances LTP, increases

neurogenesis and reduces the number of GABAergic synaptic boutons (Martinez-Cue, et al.

2013). These studies suggest that α5GABAA receptors may contribute to aberrant plasticity

and synaptic morphology in the Down syndrome model.

Another developmental disorder that may be in part mediated by α5GABAA receptors

is autism. Linkage and association studies in human patients have demonstrated that

mutations in human chromosome 15q11e13, which encodes several GABAA receptor genes

49 including Gabra5, are significantly associated with the incidence of autism (Cook et al.

1998; McCauley et al. 2004; Menold et al. 2001; Shao et al. 2003). Additionally, studies of post-mortem tissue show that autistic patients have a reduced expression of α1, α2, α3, α4,

α5, β1, and β3 subunits of the GABAA receptor (Blatt et al. 2001; Fatemi et al. 2010; Fatemi et al. 2009; Mori et al. 2012; Oblak et al. 2009). Furthermore, a positron-emission tomography (PET) study of autistic patients demonstrates reduced radioligand binding to

α5GABAA receptors (Mendez et al. 2013), suggesting reduced expression of this receptor population in the autistic brain. A reduction in α5GABAA receptor expression may indeed contribute to autism-like symptoms. A reduction in α5GABAA receptor expression increases neuronal excitability (Bonin, et al. 2007) and preclinical studies show that an increased ratio of excitation to inhibition is characteristic of autism-spectrum disorders and epilepsy

(Rubenstein et al. 2003).

In a pilocarpine rat model of epilepsy, α5GABAA receptor expression was reduced in the CA1 and CA3 where there was extensive neuronal death, but was significantly increased in the dentate gyrus (Fritschy et al. 1999). The increase in α5GABAA receptor expression may serve as a compensatory mechanism for the increased excitability during status epilepticus (Fritschy, et al. 1999). However, reducing α5GABAA receptor activity with inverse agonists does not have proconvulsant effects (Atack et al. 2006) and thus far, polymorphisms in Gabra5 have not been associated with inherited forms of epilepsy (Ma et al. 2006).

α5GABAA receptors have also been associated with several psychiatric disorders.

Associations between specific Gabra5 alleles and the incidence of bipolar disorder have been reported in Greek and Japanese populations (Otani et al. 2005; Papadimitriou et al. 1998).

50

In addition, polymorphisms in Gabra5 have been associated with unipolar depression (Oruc

et al. 1997). Negative symptoms of , which include anhedonia and lack of

motivation that often occur in depression, are also associated with α5GABAA receptor

expression in the prefrontal cortex and the hippocampus (Asai et al. 2008). Specifically, one

study shows that binding of the highly-selective α5GABAA receptor ligand [(11)C]Ro15-

4513 in these brain regions was negatively correlated with the expression of negative

symptoms in schizophrenic patients (Asai, et al. 2008). Studies in preclinical animal models

support the role of α5GABAA receptors in schizophrenia. In a methylazoxymethanol

mouse model, a positive allosteric modulator of α5GABAA receptors reduces locomotor

hyperactivity and spontaneous firing of dopaminergic neurons in the ventral tegmental area

to baseline levels (Gill et al. 2011). A negative allosteric modulator of α5GABAA receptors,

RO4938581, also reduces cognitive deficits in a neonatal-phencyclidine model of

schizophrenia-like behaviour (Redrobe et al. 2012). Additional studies must be performed to

determine the precise role of α5GABAA receptors in neuropsychiatric disorders and whether

this receptor population may be targeted for treatment of specific symptoms.

Cognitive deficits in a mouse model of ischemic stroke may also be in part mediated

by α5GABAA receptors (Clarkson, et al. 2010). Ischemic stroke enhances tonic inhibition in

layer II cortical pyramidal neurons in the peri-infarct region for 3 to 14 days (Clarkson, et al.

2010). When mice are treated with the α5GABAA receptor-selective inverse agonist L-

655,708 for several days after stroke, recovery of motor function is greatly improved

(Clarkson, et al. 2010). The enhanced tonic inhibition may be triggered by inflammation in

the peri-infarct region. I co-authored a study that showed that neuroinflammation caused by a

systemic injection of the proinflammatory cytokine IL-1β, or the endotoxin

51

lipopolysaccharide, triggers an increase in α5GABAA receptor-mediated tonic inhibition and

increased cell-surface expression of α5GABAA receptors in the hippocampus (Wang, et al.

2012). In addition, inflammation causes learning and memory deficits on the contextual fear

conditioning task and reduces LTP in WT but not Gabra5-/- mice (Wang, et al. 2012).

Undoubtedly, inflammation triggered by surgical trauma or anesthesia also causes memory

deficits in part through α5GABAA receptor- dependent mechanisms.

2.5 Pharmacology of GABAA receptors

In this section, I will briefly review the drugs that bind to GABAA receptors and affect their

activity.

The agonist binding sites are located at the extracellular N-terminal interface between

α and β subunits. In addition to the endogenous agonist GABA, , which is derived

from the mushroom Amanita muscaria, serves as an exogenous agonist that binds to the

GABA binding sites on all GABAA receptors. Antagonists of GABAA receptors decrease the

response of GABAA receptors to agonist, thus they have affinity but no efficacy at GABAA

receptors. Competitive antagonists of GABAA receptors, which bind to the GABA binding site, are (SR-95531) and bicuculline (Bright and Smart 2013). At low concentrations, gabazine blocks synaptic receptors that generate IPSCs but not extrasynaptic receptors that generate tonic inhibition (Bai, et al. 2001). Bicuculline blocks all GABAA

receptors and was used in this thesis to inhibit both synaptic and extrasynaptic receptors in

order to reveal a tonic inhibitory current. The GABAA antagonist is a non-

competitive open channel blocker that blocks the channel pore. The zinc ion also reduces the

amplitude of mIPSCs and inhibit GABAA receptors, with greatest effects at GABAA

receptors that do not contain the γ2 subunit (Barberis et al. 2011). GABAA receptor activity

52

is also subject to positive and negative allosteric modulation by endogenous and exogenous

ligands (Johnston 1996). are endogenous positive allosteric modulators of

GABAA receptor activity, with greatest enhancement of activity at δ-containing GABAA

receptors (Wohlfarth et al. 2002).

A variety of sedative and hypnotic drugs cause their behavioural effects by acting as

positive allosteric modulators at GABAA receptors. For example, two potential sites have

been identified as important for the positive allosteric modulation of GABAA receptor

activity by ethanol; S270 on TM2 of the α1 subunit and S265 on TM2 of the β1 subunit of

the GABAA receptor (Mihic et al. 1997). Benzodiazepines are positive allosteric modulators

that bind between the α and γ subunits and enhance the activity of GABAA receptors in the presence of GABA. They cause anxiolytic, sedative and hypnotic effects (Olsen and Sieghart

2009). Classical benzodiazepines such as or preferentially bind to α1,

α2, α3 or α5-containing GABAA receptors and do not bind to α4 or α6-containing receptors

(Olsen and Sieghart 2009). Other sedatives, such as , which also binds to the

binding site, have highest affinity for α1 receptors and lower affinity for α2

and α3 receptors (Olsen and Sieghart 2009). Endogenous benzodiazepine-like compounds may also exist (Rye et al. 2012). Cerebrospinal fluid collected from hypersomnolent patients potentiates currents evoked by GABA in human embryonic kidney (HEK) cells that are

transfected with recombinant GABAA receptors and this effect is blocked by the

benzodiazepine-site antagonist (Rye, et al. 2012). Common injectable and

inhalational anesthetics including propofol, etomidate and isoflurane can act as direct

agonists and positive allosteric modulators of GABAA receptors and will be discussed below.

53

2.5.1 Effects of general anesthetics on GABAA receptors

The majority of clinically used anesthetics produce their neurodepressive endpoints by

increasing GABAA receptor activity (Rudolph and Antkowiak 2004). Acute administration of an anesthetic causes amnesia (loss of memory), sedation (decreased motor activity, reduces arousal), hypnosis (unconsciousness, in mice loss of righting reflex) and immobility

(lack of movement in response to noxious stimuli, in mice lack of response to toe pinch)

(Rudolph and Antkowiak 2004). These endpoints are mediated by anesthetics inhibiting distinct brain regions, for example anesthetic action on the hippocampus contributes to amnesia, anesthetic action at the spinal cord causes immobility and anesthetic action in subcortical structures such as the thalamus and midbrain reticular formation contributes to the sedative and hypnotic effects (Rudolph and Antkowiak 2004). In addition, each

GABAergic anesthetic causes each endpoint by acting on specific subtypes of GABAA

receptors. Some of these anesthetics will be reviewed here.

Propofol is a common injectable anesthetic and a positive allosteric modulator of

GABAA receptors that increases the duration and amplitude of mIPSCs by increasing the frequency of channel opening, decreasing deactivation and reducing desensitization of

receptors (Bai et al. 1999). A recent study, using a photolabelled form of propofol identified

the propofol binding site on H267 of the β3 subunit, between TM1 and TM2 domains with

the binding site open to the channel pore (Yip et al. 2013).

2.5.1.1 Isoflurane

In this thesis, the inhalational anesthetic isoflurane was used because it targets GABAA

receptors, it is rapidly eliminated and most importantly, it is commonly used in clinical

practice and experimental neuroscience.

54

Isoflurane is an ether-like anesthetic that is structurally similar to other common

anesthetics such as enflurane, sevoflurane and desflurane (Campagna et al. 2003). The

isoflurane binding site has been identified on S270 in the TM2 domain of the α1 subunit

(Jenkins et al. 2001; Nishikawa et al. 2002). However, this amino acid is conserved across

all α subunits and isoflurane affects the activity of GABAA receptors that contain other α subunits, including α5GABAA receptors (Caraiscos et al. 2004). Isoflurane is a positive allosteric modulator at GABAA receptors that enhances both synaptic and extrasynaptic inhibition. At low concentrations (25 µM), isoflurane selectively enhances tonic inhibition mediated by α5GABAA receptors and at high concentrations (EC50 300 µM) it enhances

IPSCs (Caraiscos, et al. 2004).

Distinct receptor subtypes are responsible for the different endpoints caused by

isoflurane. Knock-in mice with the mutations S270H and L277A in the α1 subunit require

higher concentrations of isoflurane to exhibit loss of righting reflex than WT mice, thereby implicating the α1 subunit in the hypnotic actions of the drug (Sonner et al. 2007). The acute, amnestic effects of isoflurane are likely mediated by α4-containing and β3-containing

GABAA receptors, as null mutant mice that do not express one of these subunits exhibit

normal memory acquisition on the contextual fear conditioning assay during isoflurane

treatment (Rau et al. 2009; Rau et al. 2011). Since the isoflurane binding pocket that was

identified on the α1subunit is conserved across all α subunits, other populations of GABAA

receptors, including α5GABAA receptors, may also contribute to the acute amnestic effects of

isoflurane. However, null mutant mice for other α subunits of the GABAA receptor have not

been tested to determine whether other receptor subtypes are also necessary for the acute,

amnestic effect of isoflurane.

55

Following inhalation, isoflurane undergoes minimal biodegradation or metabolism (<

0.2%) and nearly all isoflurane can be recovered in expired gases (Holaday et al. 1975).

Levels of isoflurane in the brain decline rapidly and are at undetectable or trace levels

(0.0095%) 24 hours after anesthesia (Saab, et al. 2010). The dose of an inhalational

anesthetic is typically measured as a minimum alveolar concentration (MAC). One MAC is

the concentration required to prevent movement in response to a painful stimulus in 50% of

subjects (Merkel et al. 1963). Hence, 1 MAC is the ED50 for immobility with isoflurane. In

mice, 1 MAC of isoflurane is equivalent to approximately 1.3% of isoflurane in inhaled gases

(Sonner et al. 2000).

2.5.1.2 Etomidate

Etomidate is an intravenous anesthetic that is highly selective for GABAA receptors, and

displays little or no effect on other receptor populations (Belelli et al. 1997). The binding site

for etomidate has been identified to be on N289 in TM2 domain of the β3 subunit and it faces

towards the channel pore (Belelli, et al. 1997). Etomidate is a positive allosteric modulator

and at higher concentrations can be a direct agonist of GABAA receptors, triggering the opening of the channel in the absence of GABA (Belelli, et al. 1997).

Transgenic mice have been used to determine the subtypes of GABAA receptors that

are required for etomidate’s neurodepressive actions. Mice with the β3N265M point mutation

are immune to the hypnotic (loss of righting reflex) and immobilizing (toe pinch) effects of

etomidate (Jurd et al. 2003). Mice with the β2N265M mutation are resistant to the

hypothermic and sedative properties of etomidate (Reynolds et al. 2003). Gabra5-/- mice are

resistant to the amnestic effects of etomidate and this action is likely mediated by α5β3γ2

receptors in the hippocampus (Cheng, et al. 2006).

56

In patients, following a single bolus injection, the onset of etomidate’s effects occurs

within 30 to 60 seconds and the hypnotic effects last for 2 to 10 minutes (Forman 2011).

Concentrations of etomidate required for hypnosis in patients are approximately 200 ng/ml or

1 µM in the blood (Forman 2011). In mice, a bolus injection produces hypnosis for 20 to 30

minutes with an ED50 (for loss of righting reflex) of approximately 10-12 mg/kg and an

ED100 of 20 mg/kg (Cheng, et al. 2006). The duration of etomidate’s hypnotic effects is short as etomidate concentrations in the blood decline rapidly due to the drug being redistributed into peripheral tissues (Forman 2011). The decline can be divided into 3 half-lives, the shortest is 2 minutes as etomidate redistributes into highly perfused tissues, the second is 20 minutes as etomidate redistributes into muscle and the last, longest, half-life for terminal metabolism is approximately 3.9 hours (Van Hamme et al. 1978). Etomidate’s short duration of action after a bolus injection can also be attributed to its rapid conversion to inactive metabolites by esterases in the plasma and (Forman 2011). The esterases hydrolyze etomidate into carboxylic acid and ethanol (Forman 2011). The majority of etomidate metabolism occurs in the liver and the metabolites are excreted by the kidneys in urine

(Forman 2011).

Etomidate is typically administered by bolus injection, as sustained infusions of the drug or chronic treatment can interfere with steroid synthesis and thus cause adrenocortical suppression, leading to negative side-effects such as hypotension and hyponatremia (Forman

2011).

2.5.1.3 Dexmedetomidine: A non-GABAergic anesthetic

Some sedatives and anesthetics, such as dexmedetomidine and ketamine, do not act on

GABAA receptors to achieve their neurodepressive effects. In this thesis dexmedetomidine

57

was used as an active comparator, meaning it had a similar sedative effect as a low, sedative

dose of etomidate but it did not affect the activity of GABAA receptors.

Dexmedetomidine is an agonist at α2-adrenergic receptors (Doze et al. 1989).

Dexmedetomidine primarily activates G-protein coupled, presynaptic, α2-adrenergic receptors and reduces the release of norepinephrine neurotransmitter from presynaptic terminals through two mechanisms: first, it activates G-protein gated potassium channels leading to hyperpolarization of the presynaptic neuron and second, it reduces the activity and thereby calcium entry through voltage-gated calcium channels (Jorm et al. 1993; Nacif-

Coelho et al. 1994).

α2-Adrenergic receptors in the locus coerulus, a brain area important for maintaining wakefulness, are important for the sedative and hypnotic actions of dexmedetomidine

(Correa-Sales et al. 1992), whereas receptors in the spinal cord are important for the analgesic properties of the drug (Kalso et al. 1991).

In humans, a single bolus dose causes sedation within seconds, with peak sedation at

10 minutes after injection and persisting for, on average, 195 minutes (Belleville et al.

1992). Dexmedetomidine undergoes hepatic metabolism and the metabolites are excreted in the urine with a terminal elimination half-life of 1.5 to 2 hours in humans (Dyck et al. 1993).

During surgery, administration of dexmedetomidine seems to decrease the dose of isoflurane required to maintain anesthesia and the dose of opioids required for analgesia (Aantaa et al.

1997; Aantaa et al. 1990; Acevedo-Arcique et al. 2014). Here, dexmedetomidine will be used to determine whether non-GABAergic anesthetics cause lasting memory deficits and changes in GABAA receptor physiology.

58

2.5.2 Pharmacology of α5GABAA receptors

α5GABAA receptors are typically composed of α5β3γ2 subunits (Sur et al. 1998). The

presence of the β3 subunit makes these receptors vulnerable to positive allosteric modulation

by propofol and etomidate, which bind to sites on the β3 subunit (Bai, et al. 2001; Belelli, et al. 1997; Cheng, et al. 2006). In addition, these receptors are sensitive to positive allosteric modulation by classical benzodiazepines such as diazepam and flunatrazepam but insensitive to modulation by zolpidem, which preferentially modulates α1 subunit-containing (α1βγ2)

GABAA receptors (Olsen and Sieghart 2009; Yeung, et al. 2003).

Inverse agonists at the benzodiazepine site act as negative allosteric modulators that reduce the activity of GABAA receptors. Several benzodiazepine inverse agonists have been

shown to be selective for α5GABAA receptors, including L-655,708, α5IA, MRK-016, PWZ-

029 and RO4938581 (Atack et al. 2009; Ballard et al. 2009; Dawson, et al. 2006; Quirk, et

al. 1996; Savic et al. 2008). These drugs are often used to selectively reduce the activity of

α5GABAA receptors in vitro and in vivo (Quirk, et al. 1996). The benzodiazepine inverse

agonist L-655,708 was used in this thesis. The affinity of L-655,708 for α5GABAA receptors

is 50 to 100-fold greater than for α1GABAA receptors, α2GABAA receptors or α3GABAA

receptors (Quirk, et al. 1996). In vitro, L-655,708 inhibits 51% of the GABAA receptor-

mediated current at a low, EC20 dose of GABA (Chambers et al. 2003). In vivo, a single

injection of L-655,708 (1 mg/kg) causes 64% occupancy of the α5GABAA receptor (Atack, et al. 2006). L-655,708 reaches peak levels in the rat brain within 15 minutes and has a half- life of 30 minutes (Atack et al. 2006). In vivo treatment with L-655,708 improves learning on certain tasks and lowers the threshold of stimulation required to induce LTP in the hippocampus (Martin, et al. 2010).

59

2.6 Structure and Function of the Hippocampus 2.6.1 Structure

The hippocampus is part of the limbic system and it is located in the medial temporal lobe.

The curved shape of the hippocampus forms the dentate gyrus and Ammon’s horn, which is further subdivided into the cornus ammonis 1 and 3 (CA1 and CA3) (Neves et al. 2008). The basic, functional anatomy of the hippocampus is usually described as the trisynaptic loop, which is characterized by unidirectional flow of information and synapses in the three regions of the hippocampus; the dentate gyrus, the CA3 and the CA1 (Neves, et al. 2008)

(Figure 2.4). First, the perforant pathway carries sensory information from layers II and III of the entorhinal cortex (Neves, et al. 2008) and perforant path axons synapse onto the dendrites of granule cells in the dentate gyrus. Second, granule cells project their axons along the mossy fiber pathway to the CA3, where they form synapses on the apical dendrites of

CA3 pyramidal neurons (Neves, et al. 2008). Third, CA3 pyramidal neurons project their axons along the Schaffer collateral pathway and synapse onto CA1 pyramidal neurons

(Neves, et al. 2008). These projections are not solely ipsilateral, as Schaffer collateral axons also project to contralateral CA3 and CA1 pyramidal neurons (Neves, et al. 2008).

60

Figure 2.5 The trisynaptic pathway of the hippocampus.

Perforant pathway axons from the entorhinal cortex synapse on dentate gyrys granule cells. Granule cells project their axons along the mossy fiber pathway and synapse onto cornus ammonis 3 (CA3) pyramidal neurons. Axons of CA3 pyramidal neurons form the

Schaffer collateral pathway and synapse onto cornus ammonis 1 (CA1) pyramidal neurons.

61

The principal cells of the hippocampus are organized in a laminar fashion. The

electrophysiological studies in this thesis focus on the CA1 region and hence, the laminar

organization of this region will be described in detail. The cell bodies of CA1 pyramidal neurons are located in the stratum pyramidale layer of the CA1 (Spruston et al. 2006).

Electron microscopy studies show that each pyramidal neuron has approximately 30,000 dendritic spines (Spruston and McBain 2006). The basal dendrites of CA1 pyramidal neurons occupy the stratum oriens and form synapses with Schaffer collateral axons

(Spruston and McBain 2006). The proximal apical dendrites are in the stratum radiatum and

also receive input from the Schaffer collateral pathway (Spruston and McBain 2006). In

addition to receiving inputs from the trisynaptic pathway, distal apical dendrites in the

stratum lacunosum-moleculare receive input from layer III of the entorhinal cortex via the

perforant pathway (Spruston and McBain 2006). Similarly, CA3 pyramidal neurons receive

direct input from layer II cells of the entorhinal cortex (Spruston and McBain 2006). CA1

neurons also receive inputs from the thalamus and basolateral nucleus of the amygdala at

distal dendrites (Spruston and McBain 2006). Other neurotransmitters, beyond excitatory

glutamate and inhibitory GABA, also modulate the activity of CA1 pyramidal neurons

(Spruston and McBain 2006). Dopaminergic afferents from the ventral tegmental area and noradrenergic afferents from the locus coerulus synapse on distal apical dendrites in the stratum lacunosum moleculare (Spruston and McBain 2006). The CA1 also receives

cholinergic input from the septum and serotonergic input from the raphe nuclei (Spruston and

McBain 2006).

Axons of CA1 pyramidal neurons project to the subiculum and to layer V of the

entorhinal cortex (Spruston and McBain 2006). Additionally, septal axons originating from

62 medial portions of the hippocampus project to the retrosplenial, perirhinal and Broca’s area of the cortex (Spruston and McBain 2006). Axons originating from neurons in the more temporal or lateral regions of the CA1 project to the medial frontal cortex, olfactory bulb, nucleus accumbens, and the basal nucleus of the amygdala (Spruston and McBain 2006).

In contrast to principal cells of the hippocampus, hippocampal interneurons do not have a laminar organization and are present throughout the hippocampus. All interneurons in the hippocampus release GABA as their primary neurotransmitter (Spruston and McBain

2006). These neurons typically have short axons that project onto dendrites and soma of principal cells (Spruston and McBain 2006). There is a great diversity of interneurons in the hippocampus with at least 14 types that are classified according to morphology and many more that are classified according to the neurochemical markers that they express (for e.g. parvalbumin, somatostatin or cholecystokinin) (Spruston and McBain 2006). Different populations of interneurons innervate different portions of the pyramidal neuron and to modulate the efficacy of excitatory inputs.

The primary input to the hippocampus, the entorhinal cortex, receives inputs from the perirhinal cortex and the parahippocampal cortex (Bird et al. 2008). Inputs from the perirhinal cortex carry information from the ventral visual processing or “what” stream that carries information about the shape and features of an object (Bird and Burgess 2008). The parahippocampal cortex receives information from the dorsal visual processing or “where” stream that carries information about an object’s position in space (Bird and Burgess 2008).

The “what” and “where” information that is integrated by the hippocampus forms the basis of episodic memory (Bird and Burgess 2008).

63

2.6.2 The role of the hippocampus in learning and memory

The first evidence of the importance of the hippocampus in episodic memory in humans

came from the famous patient H.M. who had a large portion of his medial temporal lobe

removed in order to treat his epileptic (Scoville et al. 1957). While the surgery eliminated his seizures, he was also left with anterograde amnesia such that he was unable to remember events that occurred after the surgery as well as some retrograde amnesia for events that occurred before the surgery (Scoville and Milner 1957).

Animal studies that have used lesion or pharmacological strategies to inactivate the hippocampus, have further proved the importance of this region in learning and memory

(Bird and Burgess 2008). The hippocampus seems to be very effective at remembering the

“where” in episodic memories. Pyramidal neurons in the CA1 region are called “place cells” and respond to the location of an animal in a given environment (O'Keefe et al. 1971).

Specifically, each neuron fires action potentials only when the animal is in a particular

location of the environment (the place cell field), for example the north-west corner of a

room. The place cell field for each neuron can be remapped when spatial cues in the

environment change (Neves, et al. 2008). Memory of events, or episodic memory, requires

the function of the hippocampus. In rodents models, this includes the memory of an aversive

stimulus in a specific context in the contextual fear conditioning task, memory of a set of

objects in a given environment in the object recognition task and memory for the location of

an escape platform in a pool of cool water in the Morris Water Maze task (Bird and Burgess

2008). Lesions to the hippocampus cause impaired memory performance on these tasks in

rodents (Bird and Burgess 2008).

64

The memory task that is used in many studies presented in this thesis is the object recognition assay. This task relies on a rodent’s innate preference for novel objects (Bevins et al. 2006). During training on the task mice are presented with two identical objects (Bevins and Besheer 2006). Following a retention period, mice are expected to recall the familiar, previously shown object (Bevins and Besheer 2006). If the mouse is able to recall the previously shown object, it will spend more time exploring the novel object (Bevins and

Besheer 2006). The recognition of familiar objects depends on the perirhinal cortex, which provides input to the entorhinal cortex, the primary source of afferents to the hippocampus

(Winters et al. 2010). Recognition memory also depends on the hippocampus and can be impaired in patients with hippocampal damage (Bird and Burgess 2008; Eichenbaum et al.

2007; Wixted et al. 2010) .

The object recognition test was selected for the purposes of this thesis for several reasons. First, it can be used to probe subtle memory deficits, whereby learning and recall of the task is not motivated by aversive or appetitive stimuli such as electric foot shock or food reward (Ennaceur 2010). Second, it is a well-validated model for high throughput screening of drug-induced effects (Ennaceur 2010). Lastly, each mouse can be trained and tested on the task repeatedly (Ennaceur 2010), therefore allowing us to test the time course of recovery of learning and memory function after anesthesia.

2.6.3 LTP in the hippocampus

Memory formation is thought to be dependent on lasting changes in synaptic strength at specific synapses in the hippocampus. Long-term potentiation (LTP) is a sustained enhancement of excitatory neurotransmission in the hippocampus that is thought to underlie learning and memory (Bliss et al. 1993). LTP was first discovered by Bliss and Lomo when

65

they showed that high-frequency electrical stimulation of the perforant pathway in

anesthetized rabbits resulted in a sustained increase in the amplitude of excitatory postsynaptic potentials (EPSPs) in the dentate gyrus that lasted for hours or days (Bliss et al.

1973).

The relationship between LTP and memory has been examined in rodent models using

pharmacological agents that seem to block both memory formation and the induction of LTP.

For example, administration of the N-methyl-D-aspartate (NMDA) receptor inhibitor (2R)-

amino-5-phosphonovaleric acid (APV) prevents the induction of LTP in the CA1 region of

the hippocampus and impairs learning on the spatial Morris water maze task (Morris et al.

1986).

In this thesis, LTP will be studied at Schaffer-collateral synapses with CA1 pyramidal

neurons. At baseline, EPSPs are evoked in the stratum radiatum of the CA1 by electrically

stimulating the Schaffer collateral pathway. LTP can be induced through high frequency

stimulation (for e.g. 100 Hz, 600 pulses) of the Schaffer collateral pathway (Bliss et al.

2006). Typical LTP has 3 phases: post-tetanic potentiation, early-LTP (E-LTP) and late or

long-lasting-LTP (L-LTP). Each phase is distinguished by its duration and the mechanisms

that underlie it. These mechanisms have been reviewed by Bliss et al. and will be briefly

discussed below (Bliss, et al. 2006).

Post-tetanic potentiation occurs within seconds of high frequency of stimulation and

can last for tens of minutes. It is a presynaptic phenomenon caused by an increase in

intracellular calcium concentrations in the presynaptic terminal and consequently, increased

release of neurotransmitter (Bliss, et al. 2006). The second phase of LTP, E-LTP is

66

dependent on increased activity of excitatory, glutamatergic receptors and increased

trafficking of these receptors to the cell surface (Collingridge et al. 1995). The induction of

E-LTP requires the depolarization of the postsynaptic neuron and calcium entry through

NMDA receptors in the cell membrane (Collingridge and Bliss 1995). Upon glutamate

binding at glutamatergic α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors, Na+ influx and K+ efflux through AMPA receptors helps depolarize the neuron.

Depolarization, promotes the dissociation of the Mg2+ ion that blocks the NMDA receptor

pore, thereby allowing calcium entry via the NMDA receptor. This calcium entry is

important, as mitigating this increase in intracellular calcium with calcium buffers prevents

the induction of LTP (Bliss, et al. 2006). The calcium ions bind to calmodulin, which then activates calcium/calmodulin-dependent protein kinase II (CaMKII) (Malenka 2003).

CaMKII phosphorylates AMPA receptors in the cell membrane to increase their conductance

(Malenka 2003). It also phosphorylates AMPA receptors sequestered in intracellular reservoirs to promote their trafficking to the cell membrane (Malenka 2003). L- LTP, is

dependent on the synthesis of new proteins that are involved in synapse formation (Reymann

et al. 2007). These proteins include glutamate receptors and kinases such as CaMKII

(Reymann and Frey 2007). L-LTP depends on the calcium dependent activation of protein

kinase A (PKA), which translocates to the nucleus and phosphorylates the CREB

transcription factor, which then initiates the transcription of proteins involved in the

formation of excitatory synapses (Reymann and Frey 2007). The activation of extracellular-

regulated kinase-mitogen-activated protein kinase is also thought to contribute to increased

transcription of new proteins (Bliss, et al. 2006; Reymann and Frey 2007). Consequently,

67

due to the formation of new synapses and synthesis of new proteins L-LTP is long-lasting

and can last for months after induction.

In addition to LTP, excitatory synapses in the CA1 can undergo long-term depression

(LTD) of synaptic activity (Bliss, et al. 2006). LTD is triggered by low frequency stimulation (1 Hz) over a prolonged period of time (Bliss, et al. 2006). Similar to LTP, LTD in the CA1 is also dependent on calcium entry into the cell (Bliss, et al. 2006). However, it is

dependent on slow, small increases in calcium through a specific subtype of NMDA receptor

that contains the NR2B subunit (Bliss, et al. 2006). LTD requires the activation of protein

phosphotases, which dephosphorylate AMPA receptors at the cell surface and lead to their

internalization (Bliss, et al. 2006).

Previous synaptic activity can influence whether LTP or in LTD are induced at a

specific synapse (Abraham et al. 1996). This concept whereby the history of synaptic

plasticity influences subsequent plasticity is called metaplasticity (Abraham and Bear 1996).

In one model, previous high levels of activity at a specific synapse make it more difficult to

induce LTP and thereby increase the stimulation threshold that is required to induce it

(Cooper et al. 2012). Whereas, low levels of activity, make it more likely that subsequent

stimulation will induce LTP and thus, reduce the stimulation threshold for the induction of

LTP (Cooper and Bear 2012). In the CA1 region, the threshold stimulation at which LTP can

be induced is modulated by α5GABAA receptor activity (Martin, et al. 2010).

2.6.4 GABAA receptors in the hippocampus

The principal subtypes of GABAA receptors in the hippocampus are α1βγ2, α2βγ2, α3βγ2,

α4βγ2, α5βγ2, and αβδ. The single channel conductance of each of these receptors is

68

approximately 25-30 picosiemens (pS) (Pavel et al. 2006). In the dentate gyrus, α2, α4, α5,

β1, β3 and δ subunits are the predominant subunits expressed in granule cells (Pavel, et al.

2006) and the majority of tonic inhibition in the dentate gyrus is mediated by α4βδ and α1βδ

receptors (Pavel, et al. 2006). Pyramidal neurons in the CA1 express α1, α2, α4, α5, β2, β3,

γ1 and γ2 subunits (Pavel, et al. 2006). The majority of synaptic inhibition in the CA1 region

is mediated by α1βγ2 subunit-containing GABAA receptors and the majority of tonic inhibition is mediated by α5β3γ2 subunit-containing receptors (Caraiscos, et al. 2004;

Farrant and Nusser 2005; Kasugai et al. 2010). However, some α5GABAA receptors are

located synaptically as the amplitude of spontaneous inhibitory postsynaptic currents is

reduced in CA1 pyramidal neurons in brain slices from Gabra5-/- mice (Collinson, et al.

2002). Different complements of GABAA receptors are present at different locations within

the same pyramidal neuron. For example, the α5 subunit is present in the dendrites of

pyramidal neurons, whereas the expression of the α2 subunit is greatest at the axon initial

segment (Fritschy et al. 1998).

A CA1 pyramidal cell has approximately 1700 synapses with GABAergic

interneurons (Pavel, et al. 2006). On the axon initial segment, there are approximately 25

synapses per 50 µm and inhibition at these synapses can prevent the initiation of an action

potential (Pavel, et al. 2006). Different types of interneurons synapse at each stratum of the

CA1. For example, parvalbumin-expressing fast-spiking interneurons synapse onto

α1GABAA receptor-containing synapses, parvalbumin-expressing basket cells synapse onto

α2GABAA receptor-containing synapses and the expression of α5GABAA receptors seems to

be highest at synapses innervated by bistratified interneurons (Nyiri et al. 2001; Somogyi et

al. 2005).

69

In the CA1 region, tonic inhibition mediated by extrasynaptic α5GABAA receptors regulates neuronal excitability (Bonin, et al. 2007). Furthermore, at the network level,

α5GABAA receptors regulate the induction of LTP (Martin, et al. 2010). Specifically, expression of α5GABAA receptors shifts the threshold to the right, thus a higher stimulation frequency is required to induce LTP at Schaffer-collateral CA1 synapses in WT than in

Gabra5-/- mice (Martin, et al. 2010). At a systems level, the expression of α5GABAA regulates learning and memory processes (Martin, et al. 2010).

2.7 Summary

In summary, POCD is a major clinical problem and currently there are no treatments available (Monk, et al. 2008). Exposure to an anesthetic alone is known to cause memory deficits (Culley, et al. 2004; Culley, et al. 2004), however it is not clear how anesthetics, which act on specific receptors, trigger sustained changes in cognition. GABAA receptors are the principal targets of most general anesthetics (Rudolph and Antkowiak 2004) and are the focus of this thesis. In particular, extrasynaptic GABAA receptors that contain the α5 subunit are important for normal learning and memory, are highly expressed in the hippocampus, and are required for the acute, amnestic effects of anesthetics during anesthesia (Cheng, et al.

2006; Martin, et al. 2010; Pirker, et al. 2000). In this thesis, I will examine whether

α5GABAA receptors are also necessary for memory deficits that persist after anesthesia. In addition, I will examine whether a single exposure to an anesthetic triggers sustained changes in GABAA receptor-mediated current or GABAA receptor expression in the hippocampus and whether these changes may account for postanesthetic memory deficits.

70

Chapter 3. General Materials and Methods 3.1 Study approval

All experimental procedures were approved by the Animal Care Committee of the University of Toronto and performed in accordance with guidelines from the Canadian Council on

Animal Care. The methods have been described in extensive detail to ensure the reproducibility of the results reported in this thesis.

3.2 Experimental animals

The Gabra5-/- mice were generated using a C57BL/6J and Sv129Ev background, as described previously (Collinson, et al. 2002). Gabra5-/- mice breed normally and have a normal lifespan (Collinson, et al. 2002). Mice were bred from homozygote parents and after weaning (at postnatal day 21) were housed in groups of two to four littermates per cage under standard conditions. Mice were supplied with food and water ad libitum. A circadian cycle of

14 hours light/10 hours dark was maintained in the housing room, and all experiments were performed during the light phase between zeitgeber time 2 and 10. For all behavioural tests, age-matched 2- to 6-month-old, male, WT and Gabra5-/- mice were studied. For electrophysiological and biochemical experiments 1- to 4-month old male mice were used.

For electrophysiological studies of plasticity, 1 month old mice were used. The experimenter was blinded to the drug treatment of individual mice.

3.3 Anesthesia Mice were treated with one of four anesthetics and their corresponding vehicles. Mice were treated with a sedative dose (0.7%, 20 min) or anesthetizing dose (1.3%, 1 h) of isoflurane or the corresponding vehicle (70% air, 30% O2, 20 min or 1 h). For experiments with

71

sevoflurane, mice were treated with sevoflurane (2.3%; 1 MAC) or vehicle gas (70% air,

30% O2) for 1 hour. In Chapter 5, mice were treated with the injectable drugs etomidate and

dexmedetomidine. To induce a brief sedation, a bolus intraperitoneal (i.p.) injection of

etomidate (8 mg/kg, i.p.) or dexmedetomidine (200 μg/kg, i.p.) was administered.

Physiological saline (i.p.) was used as the vehicle for sedative doses of etomidate and

dexmedetomidine and all sedative doses were administered at a volume of 0.2 ml/30 g (6.67

ml/kg). Etomidate (8 mg/kg i.p.) produces a brief 15 to 20 minute sedation during which the

mouse does not lose righting reflex and does not display spontaneous locomotion.

Dexmedetomidine (200 μg/kg, i.p.) produces a similar level of sedation lasting

approximately 2 hours. An anesthetizing dose of etomidate (20 mg/kg, i.p.) and its

corresponding vehicle (propylene glycol 26% volume/volume in physiological saline and

30% O2, respectively) was also used. The sedating and anesthetizing doses of etomidate and

isoflurane were selected from the literature to approximate the ED50 and ED100 for the loss of

righting reflex (LORR), respectively (Cheng, et al. 2006; Sonner, et al. 2000; Sonner, et al.

2007).

During anesthetic treatment, each mouse was placed in an airtight acrylic chamber (27

cm wide × 10 cm deep × 10 cm high) that had been preflushed with an anesthetic gas

mixture, if appropriate, or vehicle gas (30% O2), delivered at 1 L/min. The concentrations of isoflurane, O2, and expired CO2 in the chamber were continuously analyzed with a commercial gas analyzer (Datex Ohmeda, Mississauga, Ontario). Mice were not tested for

immobility with the tail pinch assay or for LORR during anesthesia to avoid unnecessary

stimulation.

72

To prevent hypothermia, the temperature of the chamber was maintained at 35ºC with

a heating blanket that was placed under the acrylic chamber. This anesthesia regimen does

not cause hypoxia or hypothermia (Saab, et al. 2010). In a subset of isoflurane and

etomidate-treated mice, transcutaneous oxygen saturation was measured at a frequency of 15

Hz with a mouse pulse oximetry sensor (MouseOx, Starr Life Sciences Corp., Allison,

Pennsylvania) that was placed on a shaved area of the throat over the carotid arteries. None

of the anesthetic treatments caused hypoxia as oxygen saturation remained above 98% during

etomidate (20 mg/kg i.p.) and isoflurane (1.3%, 1 hour) anesthesia.

3.4 Preparation of Pharmacological Agents used In Vivo

All drugs were prepared fresh each day to minimize degradation and contamination of

solutions. Etomidate at 2 mg/mL (Hospira Healthcare Corporation, Saint-Laurent, Quebec)

was stored in a sterile, vaccum-sealed vial in 35% propylene glycol v/v. Etomidate was

diluted with physiological saline to produce the desired dose of 8 mg/kg or 20 mg/kg. The

low dose of etomidate was administered at a volume of 0.2 ml/30 g mouse (6.67 ml/kg) and

the high dose at 0.4 ml/30 g mouse (13.33 ml/kg). Dexmedetomidine (Hospira Healthcare

Corporation, Saint-Laurent, Quebec) was also diluted in physiological saline to the

appropriate concentration that would yield a dose of 200 ug/kg. L-655,708 (Tocris

Bioscience, R&D Systems Inc., Minneapolis, Minnesota) was dissolved in 100% DMSO to a

concentration of 5 mM. This solution was further diluted in physiological saline to yield

final doses of 0.35, 0.5 or 0.7 mg/kg. All drugs were administered by intraperitoneal

injection.

73

3.5 Behaviour 3.5.1 Handling

To reduce variability in learning and memory performance caused by acute stress during the

conditioning and testing phases of the study, each mouse was handled for 5-10 minutes per

day for 5 days before the start of the behavioural experiments. Mice were transported from

the holding room to the testing room for handling. I ensured that the same transportation

route and cart was used each day. Lighting of the testing room during handling was identical

to the lighting used for training and testing, as described below. To minimize novelty- induced stress, mice were handled in the same order each day and gloves were changed between mice. If intraperitoneal injections are required immediately before training or testing, habituation to the restraint required for an injection decreases the stress associated with the restraint and increases exploratory behaviour in the object recognition task.

3.5.2 Object recognition

Anterograde memory was studied after exposure to an anesthetic with the novel object

recognition task. The novel object recognition assay relies on the natural preference of

rodents to explore novel rather than familiar objects (Ennaceur et al. 1988). The test involves a training phase, a retention delay, and a testing phase (Ennaceur and Delacour 1988).

During testing, the mouse is expected to remember the objects that were presented during

training and spend more time exploring novel objects (Bevins and Besheer 2006) (Figure

3.1).

Object recognition was assessed in a 20 cm L × 20 cm W × 30 cm H opaque chamber.

As described by Bevins and Besheer, decreasing the area of the box or increasing the height

can result in increased exploration during the task (Bevins and Besheer 2006). Training was

74

conducted in a dimly lit room, with all overhead lights turned off and a 60 Watt desk lamp placed on a table approximately 50 cm from the chamber and pointed towards the ceiling. To reduce ultrasonic noise, care was taken to ensure that incandescent, rather than fluorescent lighting, was used in the testing room. Movement and interaction with the objects was recorded with a video camera mounted on a tripod and centered above the chamber.

Each mouse was habituated to the chamber on the day before training (within 24 hours of training). During habituation, each mouse was placed in the empty chamber, with no objects present, for 15 minutes. The inside of the chamber and the table were cleaned with

70% ethanol between each mouse. To increase exploration of the objects during training, additional habituation may be performed. Specifically, during the handling phase of the experiment, each mouse may be placed in the empty object recognition chamber for 1 to 2 minutes each day during handling.

Mice were assigned to be trained with one pair of sample objects. Pilot studies were performed to confirm that there was no inherent preference for any of the objects. Objects were chosen to be similar in size, but different in shape and with a comparable amount of textural features. Some examples of objects used are inter-locking blocks, toy cars and action

figures. I ensured that all objects were washable and made of a durable substrate that could

not be easily chewed by the mice. Additionally, the set of objects and the position of the

familiar and novel objects in the test chamber were counterbalanced throughout the

experiments. No external motivational factors, such as food deprivation, appetitive or

aversive stimuli, were used.

75

During the training phase, each mouse was placed in the chamber and allowed to

explore the two identical sample objects for 10 minutes. The mouse was then returned to its

home cage for a retention period. Retention delays can vary between 1 minute and 3 hours,

with longer delays increasing the difficulty of the task. Similarly, a shorter training period

(for example 6 minutes versus 10 minutes) will increase the difficulty of the task. Mice

waiting to be trained and tested were kept in their home cages in the testing room throughout the day. The experimenter was present in the room during training and testing. When drug injections were required they were also performed in the testing room. At times injections coincided with training and testing of other mice in this study.

During the testing phase, the mouse was reintroduced to the training context and presented with one familiar sample object and one novel object (Ennaceur and Delacour

1988). The familiar object was placed in the same position within the chamber during training and testing. All of the mouse’s movements were video-recorded, and the time spent exploring each object was scored manually. Exploratory behaviour was defined as sniffing, licking, or touching the object while facing it (Bevins and Besheer 2006). Usually, mice will explore the novel object significantly more than the familiar object (Bevins and Besheer

2006; Ennaceur and Delacour 1988). This bias toward novelty is interpreted as “recognition” or recall of the familiar object. Learning was deemed to have taken place if the time spent with the novel object was greater than the time spent with the familiar object. Additionally, memory was assessed by measuring the proportion of total exploration time that was spent exploring the novel object and calculating a discrimination ratio, where the discrimination ratio was the time spent exploring the novel object divided by the total time spent exploring both objects (Bevins and Besheer 2006). Mice that spent a greater proportion of time with

76

the novel object, as evidenced by a discrimination ratio greater than the chance value of 0.5,

were deemed to have remembered the familiar object (i.e., the object to which they had

previously been exposed). Typical discrimination ratios that indicate learning range from

0.61 to 0.72 (Bevins and Besheer 2006).

Animals that did not interact with each object (interaction time of less than 1 second with each object) during the test period were excluded. In addition, animals for which the discrimination ratio deviated from the mean discrimination ratio by 2 standard deviations or more were also excluded from the analysis. To determine whether the treatments affected locomotor activity or exploration, the total time spent exploring both objects was measured during the training phase and during the testing phase (Bevins and Besheer 2006).

77

Figure 3.1 The object recognition paradigm.

During training on the object recognition task, mice were exposed to two identical, sample objects. Following a retention delay (1 hour in most studies), mice were reintroduced to the context for testing. During testing, mice were exposed to one familiar, previously shown object and one novel object. Mice were video-recorded and the time spent exploring both objects was measured. Memory was assessed by measuring the proportion of total exploration time that mice spent exploring the novel object and calculating a discrimination ratio (time spent exploring the novel object divided by the total time spent exploring both objects).

78

3.6 Electrophysiology in Brain Slices 3.6.1 Preparation of Brain Slices

Mice were treated with anesthetic or vehicle at various time points before sacrifice. On the

experimental day, the mouse was transferred from its home cage to a transfer cage with some of the home cage bedding. The mouse was then carried from the animal facility to the

laboratory and allowed to acclimatize for at least 30 minutes prior to sacrifice. The mouse

was sacrificed by live decapitation using a guillotine. After live decapitation, brains were

removed and placed in ice-cold, oxygenated (95% O2, 5% CO2) artificial cerebrospinal fluid

(ACSF) that contained (in mM): 124 NaCl, 3 KCl, 1.3 MgCl2, 2.6 CaCl2, 1.25 NaH2PO4, 26

NaHCO3 and 10 D- with the solution osmolarity adjusted to 300–310 milliosmoles

(mOsm). The ACSF was continuously bubbled with carbogen (95% O2, 5% O2). Coronal

brain slices (350 μm) were prepared with a VT1200S vibratome (Leica, Deerfield, Illinois).

The slices were allowed to recover for at least 1 hour at room temperature (23–25 °C) before being transferred to a submersion recording chamber, where they were perfused with ACSF at 3–4 ml/min.

Recordings were performed in a submersion recording chamber on the stage of an upright light microscope (BX151W1, Olympus, Center Valley Pennsylvania) equipped with a water immersion objective and differential-interference contrast optics. All recordings were performed at room temperature. Electrophysiological recordings were acquired with a

Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, California) controlled with pClamp 9.0 software (Molecular Devices) and a Digidata 1322 digitizer interface (Molecular

Devices).

79

3.6.2 Extracellular Recordings

For extracellular recording of field postsynaptic potentials (fPSPs), the pipettes were filled

with ACSF and placed in the stratum radiatum of the CA1. The concentric, bipolar, tungsten

stimulation electrode (Rhodes Medical Instruments, Summerland, California) was placed in the stratum radiatium approximately 50 µm from the stratum pyramidale at a depth of approximately 100 µm. For all electrophysiological recordings, recording electrodes were made from borosilicate glass capillary tubes (1.5 mm outer diameter, World Precision

Instruments, Sarasota, Florida) using a Narishige, vertical, two-stage electrode puller

(Narishige, Tokyo, Japan). The recording electrode was placed 100-150 µm from the stimulation electrode in the stratum radiatum and approximately 10-20 µm from the stratum pyramidale. Baseline fPSPs were evoked by stimulating the Schaffer collateral pathway with

0.1 millisecond pulses, every 20 seconds (0.05 Hz), delivered by the stimulation electrode.

The slope of fPSPs was measured at baseline for at least 10 minutes with a stimulation intensity that produced a half-maximal response. Slices were then stimulated at 20 Hz (600 pulses), and fPSPs were monitored for 60 minutes after stimulation. The 20 Hz stimulation protocol was selected because the activity of extrasynaptic α5GABAA receptors modifies the

plasticity of fPSPs under these experimental conditions (Martin, et al. 2010). Changes in the

slope of the fPSP were measured using Clampfit 9.0 software (Molecular Devices,

Sunnyvale, California) and were used as an indication of synaptic plasticity. The average of

the slope during the last 5 minutes of recording was compared with the average of the slope

of fPSPs during the baseline recording. Post-tetanic depression was measured during the first

2 minutes after stimulation. Short-term depression was measured during the first 15 minutes

after stimulation. Paired pulse facilitation was measured during the 20 Hz stimulation (inter-

80

stimulus interval of 50 ms). Specifically, the slope of the fPSP after the first stimulus pulse

(fPSP1) and the second stimulus pulse (fPSP2) were measured. The paired pulse ratio was

calculated by dividing the slope of fPSP2 by the slope of fPSP1.

3.6.3 Whole-cell recordings in brain slices

For whole-cell voltage-clamp recordings, the recording electrode (2–3 MΩ) was filled with the intracellular solution containing (in mM): 140 CsCl, 10 HEPES, 11 EGTA, 4 Mg2ATP, 2

MgCl2, 1 CaCl2, 2 TEA (pH adjusted to 7.3 with CsOH and the osmolarity adjusted to 290–

295 mOsm). The pipette offset was compensated for when the recording electrode entered

the bath. The electrode was lowered onto the stratum pyramidale in the CA1 region of the

hippocampus using the “blind-patch” technique. A tight gigaohm seal ( > 1 GΩ) was formed

on the cell body of the neuron. Pipette capacitance was compensated before obtaining a

whole cell configuration.The whole cell configuration was obtained by applying negative

pressure to the recording electrode and rupturing the cell membrane. This creates electrical

access to the entire cell and results in the replacement of the intracellular contents with the

artificial intracellular solution in the patch pipette. The series resistance was monitored using

a 10 mV hyperpolarizing voltage step in Multiclamp software (Molecular Devices,

Sunnyvale, California). Currents were sampled at 10 kHz. All cells were recorded at a

holding potential of −60 mV and automatic capacitance compensation from Multiclamp

software was applied. Exogenous GABA (5 μM) was added to ACSF. The addition of

GABA (5 μM) to the perfusate will result in GABA concentrations similar to physiological

levels of extracellular GABA (0.2–0.25 µM) measured in vivo with microdialysis because of

active reuptake of GABA in brain slices (Bright and Smart 2013). The following drugs were

also bath applied throughout the recordings: the competitive NMDA receptor antagonist

81

APV (20 μM), the competitive AMPA/ antagonist CNQX (20 μM; both

from Abcam Inc, Toronto, Ontario), and the voltage-gated sodium channel blocker TTX (300

nM, Alomone Labs Ltd., Jerusalem, Israel).

Miniature inhibitory postsynaptic potentials (mIPSCs) were measured during a stable

baseline recording of 2 minutes. mIPSC parameters (amplitude, inter-event interval, rise

time, decay time, area) were measured using Minianalysis software. Cumulative probability

plots for inter-event interval and amplitude were made from data from 125 events per cell

and compared using the Kolmogorov-Smirnov test. To measure the tonic current, the

GABAA receptor competitive antagonist, bicuculline methiodide (10 μM) or the α5GABAA

receptor-selective inverse agonist at the benzodiazepine site L-655,708 (200 nM) were applied. The tonic current was quantified by measuring the change in the holding current from segments containing no synaptic events.

3.7 Cell-surface and total expression 3.7.1 Hippocampal Slice Preparation

To minimize stress, which may affect GABAA receptor expression, mice were moved from

the animal facility to the laboratory at least 30 minutes prior to sacrifice. Mice were kept in

their home cages in a room separate from the dissection room. The dissection area and all

instruments were cleaned with ethanol and rinsed with water. Then the instruments were

cleaned with a single-use disinfectant hydroxgen peroxide wipe and rinsed with water again

to ensure that all blood and odors were removed. The experimenter changed gloves between

each dissection.

82

The ACSF used for slice preparation was identical to the ACSF used for

electrophysiology. The ACSF was prepared fresh on the day of the dissection. The ACSF

was prepared and placed in the freezer in several Erlenmeyer flasks (one for each dissection)

1 hour prior to the dissection to freeze the ACSF to a thick, ice slurry. Fifteen minutes before the dissection, the ACSF was removed from the freezer, placed on ice and bubbled with carbogen (95% O2, 5% O2) for 15 minutes.

Mice were carried individually, on the experimenter’s hand, to the dissection area to

minimize stress. The mouse was then held by the tail and allowed to walk into the guillotine

off the experimenter’s hand. Mice were sacrificed by live decapitation. The brain was

removed quickly (under 30 seconds) and immersed in carbogenated, ACSF slurry for 3 to 5

minutes.

The brain was then briefly removed from the ACSF slurry and glued onto the

vibratome stage as quickly as possible. The ACSF slurry was gently poured onto the

vibratome stage so as not to disturb the brain. The ACSF continued to be carbogenated

during the cutting procedure. The brain was sliced into coronal brain slices (350 µm thick) at

a vibration amplitude of 1.00 mm and a speed of 0.2 mm/s. Care was taken not to damage any part of the slices. The hippocampus was microdissected from each slice with a scalpel

and transferred individually to a solution containing fresh ice-cold, carbogenated ACSF.

Slices from each mouse were placed in separate wells. Approximately, 10 to 12 hemi-slices

of hippocampus were obtained (whole hippocampus for each mouse). If more slices were

obtained from one mouse, slices were removed from other samples so that approximately

equal volumes of tissue were obtained for each mouse. The ACSF in the vibratome tray was

changed between each mouse dissection.

83

3.7.2 Solutions used for surface biotinylation

The Dulbecco PBS (DPBS) contained 1X PBS, 1mM CaCl2, and 0.5mM MgCl2, pH 7.4, and

was purchased from Gibco (Grand Island, NY, USA). The lysis buffer contained: 20 mM

HEPES, 150 mM NaCl, 2 mM EDTA, 1% (v/v) Triton X-100, 0.1% (w/v) sodium dodecyl

sulfate (SDS) and cOmplete Protease Inhibitor Cocktail (Roche, Laval, QC, Canada) and

Phosphostop phosphatase inhibitor (Roche, Laval, QC, Canada), pH 7.4. Protease and

phosphatase inhibitors were added to the buffer immediately before the buffer was used. The

phosphate buffered saline (PBS) used for washing in biotinylation experiments contained

1.37 M NaCl, 27 mM KCl, 81 mM Na2HPO4, 14.7 mM KH2PO4, pH 7.4.

The modified Tris-buffered saline (TBS) used for the quenching step of biotinylation

contained the following: 25 mM Tris-Cl, 137 mM NaCl, 5 mM KCl, 2.3 mM CaCl2, 0.5 mM

MgCl2 and had a pH of 7.4. The agarose wash buffer contained 1X PBS and 0.05% (w/v)

SDS, pH 7.4. The elution buffer contained: 50 mM Tris-Cl, 2% (w/v) SDS, 2 mM DTT, pH

7.4. The DTT was added immediately before use.

3.7.3 Biotinylation

The surface biotinylation and Western blot protocols were developed by Jieying (Anine) Yu

and are reported in her MSc thesis (Jieying Yu, 2014). After microdissection, hippocampal

slices were allowed to stabilize in ice cold ACSF for 15 minutes to stop trafficking of receptors and to minimize cell death. Next, slices were transferred into smaller wells in a 12

well plate and kept on ice. A pipette was used to carefully and quickly aspirate buffer away.

During this step, the slices were not allowed to dry and DPBS (3-4 ml) was added

immediately at the side of each well. Slices were washed once with ice cold DPBS. Sulpho-

NHS-SS-biotin (0.75 mg/ml) was prepared and used immediately as it is rapidly hydrolyzed.

84

Carefully, and quickly, DPBS was aspirated and 1.5 mL of ice cold, sulpho- NHS-SS-biotin

in DPBS (0.75 mg/ml) was immediately added. Slices were incubated with gentle agitation

for 30 minutes at 4 ºC. Next, the biotin was aspirated and slices were incubated again with

biotin (1.5 ml/well at 0.75 mg/ml, 30 minutes at 4 ºC) that was freshly prepared immediately

before incubation. This second incubation with biotin was done to ensure complete

biotinylation. After the 30 minute incubation period, the biotin solution was aspirated and

excess biotin was quenched by gently washing slices 10 times in ice-cold, modified TBS.

Samples were placed in cold eppendorf tubes and ice cold lysis buffer with protease and phosphatase inhibitors was added to lyse the samples. For each sample, 500 μL of lysis buffer was added and triturated 10 times slowly with a 200 μL pipette. Then, each sample was sonicated at 70% output with ten 1-second pulses (to ensure that the homogenate did not get too hot, each sample was sonicated for 1 second, then placed on ice for 2 seconds and this was repeated 10 times). The samples were then rotated for 45 minutes at 4 ºC. Next, samples were centrifuged at 18,000 g at 4 ºC for 20 minutes to remove insoluble material.

The supernatant was collected into cold eppendorf tubes without disturbing the insoluble pellet. The supernatant was centrifuged again at 18,000 g for 20 minutes at 4 ºC.

The supernatant was isolated and the bicinchoninic acid assay (BCA) (Bio-Rad, Hercules,

CA, USA) was performed on the samples to determine protein concentration. A standard curve made through the serial dilution of a standard BSA sample was run with each assay so that the concentration and yield of protein in the unknown samples can be calculated. All the standard and unknown protein samples were run in triplicates on a 96 well plate. The colorimetric results were detected using a spectrophotometer (BioTek, Winooski, Vermont,

USA) at 562 nm. An aliquot of the supernatant was retained and this fraction was “total

85

protein” and was used to measure total protein expression. The aliquot was stored at -20 ºC if it was being used the next day or at -80 ºC for longer storage.

Next, the surface protein was isolated. The High Capacity Neutravidin agarose resin used was purchased from Pierce (Rockford, IL, USA). The resin “beads” were washed 3 times with DPBS and restored with equal volumes of DPBS. Equal volumes of beads were placed into clean eppendorf tubes (one for each sample) and allowed to settle. A fixed amount (0.8 mg) of the supernatant was incubated overnight with 200 μl of 50% slurry High

Capacity Neutravidin agarose beads (Pierce, Rockford, IL, USA) at 4 ºC with constant rotation. For High Capacity Neutravidin beads, 400 μg of protein for 100 μL of beads was used, whereas for normal neutravidin beads 150 μg of protein for 100 μL of beads was used.

The agarose and protein slurry for each sample was completely transferred into individual spin columns (Qiagen, Toronto, Ontario), one for each sample, for washing. To ensure that all beads were transferred to the spin column, the eppendorf tube was washed out with 1ml

PBS with 0.05% SDS and this wash solution was added to column. The spin columns were centrifuged at 2000 g at 4 ºC for 1 minute to remove protein that was not bound to neutravidin. The eluted lysate is non-bound protein and was discarded. The bound protein was in the stationary agarose fraction of the spin column. The agarose beads were washed immediately, 12 times with 600 μl PBS with 0.05% SDS. The beads were washed by adding the wash buffer to the beads, letting the beads settle at room temperature for 1 minute and then centrifuging at 2000 g for 1 minute. The flow-through was discarded. After the final wash, all beads were scraped into an eppendorf tube and 350 μl of elution buffer containing 5 mM DTT was added. The eppendorf tubes containing the agarose and the elution buffer were incubated at room temperature for 30 minutes with constant rotation. Then the tubes were

86

incubated at 37 ºC for 10 minutes. Next, the tubes were rotated again at room temperature for

20 minutes. The spin columns were placed in clean eppendorf tubes and the corresponding

samples (agarose beads and elution buffer) were added to their respective columns. The

samples were centrifuged at 10,000 g for 5 minutes. The flow-through was kept since it contained the isolated surface proteins.

The protein concentrations of the surface protein samples were determined using the

DC™ Protein Assay (Bio-Rad, Hercules, California). All the standard and unknown surface protein samples were run in triplicates on a 96 well plate. The colorimetric results were detected using a spectrophotometer (BioTek, Winooski, Vermont) at 670 nm. The total and surface proteins were subjected to SDS polyacrylamide gel electrophoresis (SDS page) immediately as GABAA receptors are sensitive to degradation after freeze-thaw cycles.

3.7.4 Western Blot

Equal amounts of total or surface protein were loaded for each condition onto a 10% SDS

polyacrylamide gel. Typically, 8 μg of total hippocampal protein and 8 μg of surface

hippocampal protein were loaded. Equal amounts of each sample were used for same gel

comparisons. Since not all samples have the same protein concentration, samples were

prepared and diluted in distilled water so that the protein concentration was identical in each

sample. For every 4 μl of homogenous sample, 1 μl of 5X sample buffer with 5% (v/v)

betamercaptoethanol was used in the sample preparation, the final sample buffer

concentration in the sample contained (in mM): 10% (w/v) sucrose, 10% (w/v) SDS, 62.5

Tris-Cl, trace bromophenol blue (for color), pH 6.8. The final prepared samples were incubated at 37 ºC for 15 minutes. 15 μl of each sample was loaded onto a 10% SDS

polyacrylamide gel with a stacking layer containing: 125 mM Tris-Cl, 0.1% (w/v) SDS,

87

0.05% (w/v) ammonium persulfate, 0.001% (v/v) TEMED, 4% polyacrylamide/bis (37.5:1), pH 6.8, and a resolving layer containing 375 mM Tris-Cl, 0.1% (w/v) SDS, 0.05% (w/v) ammonium persulfate, 0.0005% (v/v) TEMED, 10% polyacrylamide/bis (37.5:1), pH 8.8.

The gels were run in a mini electrophoresis system (Bio-Rad, Hercules, California) in running buffer containing: 25 mM Tris-Base, 192 mM glycine, 0.1% (w/v) SDS.

Gels were run at 35 V through the stacking layer and 100 V through the resolving layer. The electrophoresis was stopped when the bromophenol blue dye has just run out of the gel. The gels were lightly washed with distilled water to remove the SDS from the running buffer and were left to set in 4 ºC transfer buffer containing (in mM): 25 Tris-Base,

192 glycine, 20 % (v/v) methanol. The transfer sandwich was made with two sponges presoaked with transfer buffer, blotting paper presoaked with transfer buffer, and the gel with a nitrocellulose blotting membrane (Pall, Pensacola, FL, USA) presoaked in transfer buffer on top. All these layers were held together in a transfer cassette. The assembled cassette was placed in the electrophoresis transfer box containing transfer buffer. The gels were transferred over 12 hours at 0.12 mA at 4 ºC.

To block unspecific binding, the nitrocellulose membrane, which contained the transferred proteins was incubated at room temperature for 1.5 hours or overnight at 4 ºC with 5% milk dissolved in TBS-Tween containing 50 mM Tris-Cl, 150 mM NaCl, and 0.1%

(v/v) Tween-20, pH 7.4. Next, a primary antibody which probes for the protein of interest was applied to the membrane for 2 hours at room temperature. The primary antibody was diluted to various concentrations depending on the antibody in 3% (w/v) BSA in TBS-

Tween. The primary antibodies used in different experiments included anti-GABAA receptor

α1 antibody 1:1000 (Millipore, Billerica, Massachusetts, USA), α5 antibody 1:1000

88

(PhosphoSolutions, Aurora, Colorado), β3 1:1000 (Thermo Scientific, Rockford, Illinois), or

δ 1:1000 (Millipore, Billerica, MA, USA), anti-β-actin antibody 1:2000 (Millipore, Billerica,

Massachusetts), anti-Na+/K+ ATPase antibody 1:5000 (Developmental Studies Hybridoma

Bank, Iowa City, Iowa). The membrane was then washed 3 times, for 15 minutes each time, with TBS-Tween. A horseradish peroxidase- conjugated (HRP-conjugated) secondary antibody was applied for 1.5 hours at room temperature or overnight at 4 ºC. The membrane was then washed 3 times, for 15 minutes each time, with TBS-Tween. Enhanced chemiluminescence substrate (Thermo Scientific, Rockford, Illinois) was applied to the membrane for 3 to 5 minutes, and the chemiluminescence signal from the membrane was captured with the Image Station 2000R (Kodak, USA) or the Chemidoc XRS+ system (Bio-

Rad, Hercules, California).

3.7.5 Analysis of cell-surface and total expression data

All samples were loaded onto two replicate blots in order to control for transfer errors and the

densities of each sample were averaged over the two blots. Receptor bands were normalized

to their respective loading controls, which were Na+/K+ ATPase for surface protein, and β- actin for total protein. Blots containing surface protein were probed for β-actin to determine the purity of isolated biotinylated surface protein. Data are presented as a percentage of the mean of the control group.

3.8 Electrophysiology in cell culture 3.8.1 Preparation of cell cultures

Primary cultures of hippocampal neurons were prepared from Swiss Webster mice (Charles

River, Montreal, Canada), as described previously (MacDonald et al. 1989). Briefly, fetal

89

pups (embryonic day 18) were removed from mice euthanized by cervical dislocation. The

hippocampi were dissected from each fetus and placed in an ice-cooled culture dish. Neurons

were then dissociated by mechanical titration using two Pasteur pipettes (tip diameter, 150–

200 μm) and plated on 35-mm culture dishes at a density of approximately 1 × 106 cells/ml.

The culture dishes were coated with collagen or poly-D-lysine (Sigma-Aldrich Co., St. Louis,

Missouri). For the first 5 days in vitro, cells were maintained in minimal essential media

supplemented with 10% fetal bovine serum and 10% horse serum (Life Technologies, Grand

o Island, New York). The neurons were cultured at 37 C in a 5% CO2-95% air environment.

After the cells had grown to confluence, 0.1 ml of a mixture of 4 mg 5-fluorodeoxyuridine and 10 mg uridine in 20 ml minimal essential media was added to the extracellular solution to reduce the number of dividing cells. Subsequently, the media was supplemented with 10% horse serum and changed every 3 or 4 days. Neurons were maintained in culture for 14 to 21 days prior to recording. To prepare microglia-neuron cocultures, cortical microglia were isolated from embryonic mice as described previously (Deierborg 2013). Cells were maintained as described above and media was changed every 3 to 4 days. Once cell confluence was achieved, microglia were separated from the mixed glial cultures by gently shaking the dishes (200 rpm for 2 hours at 37 °C) then centrifuging the supernatant. The

pellet was suspended in neurobasal media and applied directly over cultured hippocampal

neurons that had been grown for 10 to 14 days in vitro. For astrocyte culture, cortical

astrocytes were isolated from embryonic day 18 mouse embryos as described previously

(Albuquerque et al. 2009). Cells were allowed to grow to confluence in minimal essential

media and 10% fetal bovine serum (FBS; Life Technologies, Grand Island, New York) for 14

d. Cells were then enzymatically dissociated with trypsin–EDTA (0.05%; Life Technologies,

90

Grand Island, New York), and passaged three times to obtain a nearly pure astrocytic culture.

Astrocytes were then plated at a density of 25,000 cells per dish. For astrocyte-neuron

coculture, astrocyte cell suspension was placed over hippocampal neurons cultured at 14

days in neurobasal media. Astrocytes were monitored visually to ensure survival and

confluence for the duration of the experiment. For all reported results, data were acquired

from cells from at least three different dissections.

3.8.2 Whole-cell recordings in cell culture

Whole-cell recordings were performed as described for brain slices with several exceptions.

The extracellular solution contained the following (in mM): 140 NaCl, 2.0 KCl, 1.3 CaCl2, 1

MgCl2, 25 HEPES, and 28 glucose (pH 7.4, 320–330 mOsm). To measure the amplitude of

the tonic current, exogenous GABA (0.5 μM) was added to the extracellular solution and the

change in holding current was measured during application of bicuculline (20 μM). GABA

(0.5 µM) is similar to physiological levels of extracellular GABA that occur in vivo (Bright

and Smart 2013).

3.9 Statistical analyses

Results are presented as means ± standard error of the mean (SEM). An unpaired Student’s t

test was used to compare groups where appropriate. For comparing three or more groups, a

one-way analysis of variance followed by Dunnett’s test or a two-way analysis of variance followed by Tukey’s Honestly Significant Difference was applied. The Kolmogorov-

Smirnov test and Shapiro-Wilk test were used to validate the assumption of normality. In

cases where the assumption of normality was not met for one or more groups, the Kruskal-

Wallis or Mann-Whitney U test were employed. Statistical testing was performed with two

91

statistical software packages: Statistical Package for the Social Sciences (SPSS version 17.0,

IBM Corporation, Armonk, New York) and GraphPad Prism software, version 4.0

(GraphPad Software, San Diego, California). A P value less than 0.05 was considered statistically significant.

92

Chapter 4. Inhibition of α5GABAA receptors restores recognition memory after isoflurane general anesthesia 4.1 Introduction

General anesthetics and benzodiazepines are administered to millions of patients each year to

allow them to tolerate surgery (Weiser et al. 2008). Unfortunately, these neurodepressive

drugs may cause cognitive deficits that persist much longer than would be expected on the

basis of their pharmacokinetic properties. For example, up to 47% of elderly patients who

have undergone anesthesia for noncardiac surgery exhibit cognitive deficits at the time of

hospital discharge (Rohan, et al. 2005). The duration of anesthesia has been shown to be an independent predictor of cognitive dysfunction in the early postoperative period (Moller, et

al. 1998). Such cognitive deficits are associated with poor long-term outcome, yet no specific treatments have been developed to date (Price, et al. 2008). A current research priority is to understand the neurobiological basis of postanesthetic cognitive deficits, including the types of memory that are susceptible to disturbance by anesthetics and the molecular mechanisms underlying these deficits.

GABAA receptors are principal targets for most inhaled anesthetics (Caraiscos, et al.

2004; Hemmings et al. 2005). The prototypic volatile anesthetic isoflurane interacts with a

putative binding cavity on the GABAA receptor to allosterically increase receptor function

(Caraiscos, et al. 2004). In particular, increased activity of GABAA receptors containing the

α5 subunit (α5GABAA receptors) is thought to contribute to acute, desirable memory

blockade during anesthesia (Cheng, et al. 2006; Martin, et al. 2009).

93

At the cellular level, both onset of and recovery from isoflurane modulation of

GABAA receptors occur on a time scale of milliseconds to seconds (Caraiscos, et al. 2004).

In humans, the rate of uptake of isoflurane is rapid (onset 3 to 5 minutes) (Hendrickx et al.

2003). The elimination of isoflurane also occurs within minutes with an initial, fast, 5 minute component and a slower 15 minute component (Lu et al. 2008). Similarly, in laboratory animals, 97% of the isoflurane is eliminated from the brain within 270 minutes (Strum et al.

1986). Surprisingly, despite relatively rapid elimination, isoflurane has been shown to cause

anterograde and retrograde memory deficits that persist for days to months in laboratory

animals (Culley, et al. 2003; Culley, et al. 2004).

Using a mouse model, we previously showed that isoflurane causes deficits in fear- associated learning and memory that last for at least 24 to 48 hours (Saab, et al. 2010).

These postanesthetic memory deficits can be prevented by pretreating the mice with the drug

L-655,708 30 minutes before administration of isoflurane (Saab, et al. 2010). L-655,708 is an imidazobenzodiazepine that acts at the benzodiazepine site of the GABAA receptor to

reduce receptor affinity for GABA and to reduce the opening of the integral

(Quirk, et al. 1996). The affinity of L-655,708 is 50-fold greater for α5GABAA receptors

than for other receptor subtypes (Quirk, et al. 1996). Thus, our previous results are interpreted as showing that preventing the activation of α5GABAA receptors during

isoflurane anesthesia prevents postanesthetic deficits in fear memory.

The above results raise the following critical question: Can memory impairment that

occurs after isoflurane be reversed by inhibiting α5GABAA receptors? The main aims of this

study were to determine whether isoflurane causes deficits in anterograde recognition

memory and whether L-655,708, administered after isoflurane anesthesia restores memory

94

back to baseline levels. Additionally, I examined whether memory with a short retention time

(1 minute) and a longer retention time (1 hour) were equally impaired after isoflurane. I also

measured the time required for spontaneous recovery of recognition memory. Furthermore,

to determine whether the expression of α5GABAA receptors is necessary for the development of memory deficits after isoflurane, we studied genetically modified Gabra5-/- mice.

Finally, to determine whether other volatile anesthetics also impair recognition memory,

learning and memory were examined 24 hours after exposure to another ether-like anesthetic, the commonly used inhalational agent sevoflurane.

4.2 Methods 4.2.1 Animal Model

Experiments were approved by the Animal Care Committee of the University of Toronto.

The Gabra5-/- mice were generated using a C57BL/6J and Sv129Ev background, as

described previously (Collinson, et al. 2002). For all behavioural tests, age-matched 3- to 4-

month-old male WT and Gabra5−/− mice were studied. To reduce variability in learning and

memory performance caused by acute stress, each mouse was handled for at least 10 minutes

per day for 5 days before the start of the behavioural experiments. The experimenter was

blinded to the drug treatment of individual mice.

4.2.2 Anesthesia

Mice were assigned to treatment with isoflurane (1.3%; 1 MAC) or vehicle gas (70% air,

30% O2) for 1 hour. During treatment, each mouse was placed in an airtight acrylic chamber

(27 cm wide × 10 cm deep × 10 cm high) that had been preflushed with the anesthetic gas

mixture or the vehicle gas, delivered at 1 L/minute. The concentrations of isoflurane, O2, and

95

expired CO2 in the chamber were continuously analyzed with a commercial gas analyzer

(Datex Ohmeda, Mississauga, Ontario, Canada). To prevent hypothermia, the temperature of

the chamber was maintained at 35ºC with a heating blanket, as previously described (Saab, et al. 2010). Following exposure to isoflurane or the vehicle gas, the mouse was removed from the chamber and was allowed to recover for 1 hour under a heat lamp before being returned to its home cage. This anesthesia regimen is known not to cause hypoxia or hypothermia

(Saab, et al. 2010). Behavioural testing was performed 24 or 72 hours after discontinuation

of treatment. At that point, motor function had fully recovered, and the sedative and

analgesic actions of isoflurane had dissipated (Saab, et al. 2010). We have previously shown

that the concentration of isoflurane in the brain at 24 h after anesthesia, as measured with gas

chromatography, is undetectable or at trace levels (0.0095 %) (Saab, et al. 2010). For experiments with sevoflurane, mice were treated with sevoflurane (2.3%; 1 MAC) or vehicle gas (70% air, 30% O2) for 1 hour.

4.2.3 Novel Object Recognition

Memory performance was studied after exposure to an inhaled anesthetic with the novel

object recognition task. Object recognition was assessed in a 20 cm × 20 cm × 30 cm opaque

chamber in a dimly lit room. Movement and interaction with the test objects (interlocking

building blocks or toy cars) was recorded with a video camera mounted above the chamber.

Each mouse was habituated to the chamber for 15 minutes on the day before testing. Mice

were assigned to be trained with one pair of sample objects. Pilot studies were performed to

confirm that there was no inherent preference for either of the objects. Additionally, the set

of objects and the position of the familiar and novel objects in the test chamber were

counterbalanced throughout the experiments.

96

The timelines and general experimental protocols are outlined in Figure 4.1. Mice

were trained on the object recognition memory paradigm 24 or 72 hours after exposure to

isoflurane or vehicle gas (Fig. 4.1 A-C). During the training phase, each mouse was placed in

the chamber and allowed to explore the two identical sample objects for 10 minutes. After

either 1 minute or 1 hour the mouse was reintroduced to the same context and was exposed to

one familiar sample object and one novel object. Memory deficits after a short retention

delay (1 minute) suggest impairment of encoding, whereas memory deficits after a longer

delay (1 hour) implicate the processes of memory consolidation and/or retention. All of the

mouse’s movements were video-recorded, and the time spent exploring each object was

scored manually. Exploratory behaviour was defined as sniffing, licking, or touching the object while facing it (Bevins and Besheer 2006). Learning was deemed to have taken place

if the time spent with the novel object was greater than the time spent with the familiar

object. Additionally, memory performance was assessed by calculating a discrimination

ratio, where the discrimination ratio was the time spent exploring the novel object divided by

the total time spent exploring both objects (Bevins and Besheer 2006).

Animals that did not interact with each object (interaction time of 0 s) during the test period were excluded, as described previously (Bertaina-Anglade et al. 2006). Mice meeting this criteria included those treated with isoflurane + vehicle injection (n = 2), vehicle gas + L-

655,708 (0.7 mg/kg; n = 2), and mice treated with isoflurane + L-655,708 (0.7 mg/kg; n = 3).

In addition, animals for which the discrimination ratio deviated from the mean discrimination ratio by 2 standard deviations or more were also excluded from the analysis, as described previously (Palanisamy et al. 2011). In total, 2 animals were excluded during the analysis

phase, one from the group that received vehicle gas plus vehicle injection and the other from

97

the group that received isoflurane plus L-655,708 (0.7 mg/kg).

To determine whether the treatments affected locomotor activity or exploration, the total time spent exploring both objects was measured during the training phase. This analysis was undertaken in response to a reviewer’s comments and several of the 247 potential videos were not available due to file corruption or the files were unavailable. Videos were not analyzed for mice in the following groups: WT control in the treatment experiment (n = 1), isoflurane-treated in the treatment experiment (n = 2), Gabra5-/- control (n = 1), Gabra5-/-

isoflurane-treated (n = 1), Gabra5-/- treated with vehicle gas and L-655,708 (n = 2),

Gabra5-/- mice treated with isoflurane and L-655,708 (n = 1), control group in the sevoflurane experiment (n = 3) and sevoflurane-treated group (n = 2).

4.2.4 Drug Treatment

Selective antagonists for α5GABAA receptors are currently not available; however, the

inverse agonist L-655,708 preferentially decreases the activity of α5GABAA receptors. The

doses of L-655,708 used in this study were selected on the basis of in vivo binding data, pharmacokinetic analyses, and previous memory studies (Atack, et al. 2006; Chambers, et al.

2003). In experiments to study reversal of memory deficits by L-655,708, doses of this agent

(0.35 mg/kg or 0.70 mg/kg, i.p) or vehicle (90% saline, 10% dimethylsulfoxide [DMSO], i.p.) were administered 23.5 hours after exposure to isoflurane or vehicle gas and 30 minutes

before training in the object recognition paradigm (Figure 4.1B). This time schedule was

selected so that all mice in the treatment cohort would be studied 24 h after isoflurane

anesthesia. In the prevention experiments, WT and Gabra5-/- mice were treated with L-

655,708 (0.7 mg/kg) or vehicle (90% saline, 10% DMSO, i.p.) injection 10 minutes before

administration of isoflurane, sevoflurane or vehicle gas (Fig. 4.1 D,E). The 0.7 mg/kg dose

98 administered 10 minutes before anesthesia has been studied previously, with no apparent effect on the potency of the anesthetic, as measured by the tail pinch assay (Saab, et al.

2010), and no generalized effects on fear-associated learning (Martin, et al. 2010). In previous studies, L-655,708 (0.7 mg/kg) caused 60% to 70% occupancy of α5GABAA receptors in vivo 30 minutes after i.p. injection, with limited binding to other GABAA receptors (Atack, et al. 2006; Atack, et al. 2006). To control for the effect of injection, control mice tested 1 minute after training, or trained and tested 72 hours after anesthesia, were given an injection of vehicle (90% saline, 10% DMSO, i.p.) 30 minutes before behavioural training (Figure 4.1A,C).

4.2.5 Statistical Analysis

Results are presented as means ± standard error of the mean (SEM). To determine whether subjects within a group remembered the familiar object, the mean time spent with the two objects was compared using a one-tailed Student’s t test, which assumed equal variances. To determine whether the memory performance between treatment groups differed, data were compared with an unpaired, two-tailed Student’s t test for equal variances or a two-way analysis of variance (ANOVA) (L-655,708 dose × gas), as appropriate. Post hoc analyses, when required, were conducted with Tukey’s Honestly Significant Difference (HSD) test.

The Shapiro-Wilk test was used to validate the assumption of normality (all P > 0.09). The discrimination ratios of all but two groups were normally distributed; the discrimination ratios of the control group in the Gabra5-/- experiment as well as the control group in the sevoflurane experiment had non-normal distributions. In these cases, the Kruskal-Wallis and

Mann-Whitney U test were used to corroborate the findings of the parametric statistics. The homogeneity of variances was verified for each group with the Levene test, which indicated

99 that the variance of discrimination ratios in each experiment was comparable between groups

(P > 0.13). Statistical testing was performed with two statistical software packages:

Statistical Package for the Social Sciences (SPSS version 17.0, IBM Corporation, Armonk,

New York) and GraphPad Prism software, version 4.0 (GraphPad Software, San Diego,

California). A P value less than 0.05 was considered statistically significant.

100

Figure 4.1 The timelines of experimental protocols.

(A) WT mice that were exposed to isoflurane anesthesia (1 hour, 1 MAC) or vehicle gas received an injection of a vehicle 23.5 hours later. To assess 1-minute memory, these mice were trained on the object recognition paradigm 24 hours after isoflurane and tested 1 minute after training. (B) To assess 1-hour memory and to determine whether memory deficits could be reversed, WT mice were exposed to isoflurane anesthesia and received an injection of either vehicle or L-655,708 (0.35 mg/kg or 0.70 mg/kg) 23.5 hours after isoflurane. This group was trained on the object recognition paradigm 24 hours after isoflurane and tested 1 hour after training. (C) Timeline of experimental treatment in which WT mice were tested 72 hours after isoflurane. Mice received a vehicle injection 71.5 hours after exposure to isoflurane or vehicle gas and 30 minutes before behavioural training and memory was assessed 1 hour after training. (D) Timeline of experimental treatment in which memory was tested in Gabra5-/- mice. Mice received injections of L-655,708 (0.70 mg/kg) or vehicle 10 minutes prior to isoflurane anesthesia. Mice were trained on the object recognition paradigm

24 hours after isoflurane and memory was tested 1 hour after training. (E) Timeline of experimental treatment in which prevention of isoflurane-induced memory deficits was assessed in WT mice. Mice received injections of L-655,708 (0.70 mg/kg) or vehicle 10 minutes before isoflurane. Mice were trained on the object recognition paradigm 24 hours after isoflurane and memory was tested 1 hour after training. (F) WT mice were exposed to sevoflurane and received a vehicle injection 23.5 hours later. Sevoflurane-treated mice were trained on the object recognition task 24 hours after anesthesia and tested 1 hour after training.

101

102

4.3 Results 4.3.1 1-minute and 1-hour Memory Performance

Mice were treated with isoflurane or vehicle gas for 1 hour and memory performance was

studied the following day. Mice were tested either 1 minute (1-minute memory) or 1 hour (1- hour memory) after training on the task. To assess memory performance within a treatment group, the time the mouse spent with the novel object was compared to the time spent with the familiar object. To compare memory performance between treatment groups, the discrimination ratios were compared. Control WT mice that were exposed to vehicle gas and a vehicle injection demonstrated normal 1-minute memory as they spent more time exploring the novel object than the familiar object (novel vs. familiar object, t = 3.30, df = 9, P =

0.005; Figure 4.2A). Isoflurane-treated mice also demonstrated normal 1-minute memory

(novel vs. familiar object, t = 2.64, df = 9, P = 0.013; Figure 4.2A). The discrimination ratios for 1-minute memory for control and isoflurane-treated groups were similar (0.68 ± 0.05 vs.

0.67 ± 0.04, t = 0.026, df =18, P = 0.979, 95% CI -0.12 to 0.12; Figure 4.2B). Also, time spent exploring both sample objects during the training phase did not differ between the groups (45.21 ± 8.05 s vs. 49.28 ± 9.60 s, t = 0.32, df = 18, P = 0.750, 95% CI -30.39 to

22.26 s; Figure 4.2C). However, since the 95% confidence interval for the difference between the means is quite wide, this experiment may have been underpowered to detect differences in exploratory behaviour during training.

Control WT mice demonstrated normal 1-hour memory as they spent more time exploring the novel object than the familiar object (novel vs. familiar object, t = 4.00, df =

30, P < 0.001; Figure 4.3A). In contrast, isoflurane-treated mice exhibited deficits in 1-hour

103 memory as they exhibited no preference for the novel object (novel vs. familiar object, t =

0.40, df = 30, P = 0.345, 95% CI -2.24 to 1.51 s; Figure 4.3A). The discrimination ratio was lower for the isoflurane-treated group than for the control group (0.51 ± 0.03 vs. 0.66 ± 0.03, t = 3.66, df = 60, P < 0.001; Figure 4.3B) and was similar to that predicted by chance (0.5).

The impairment of short-term memory performance in the isoflurane-treated mice could not be attributed to differences in exploratory behaviour during the training phase (control 74.02

± 7.54 s vs. isoflurane 62.3 ± 9.44 s, t = 0.97, df = 58, P = 0.336, 95% CI -12.47 to 35.91 s;

Figure 4.3C).

104

Figure 4.2 Normal 1-minute memory 24 hours after isoflurane anesthesia.

One day after isoflurane exposure, mice were trained on the object recognition paradigm and tested 1 minute after training. (A) Time spent with novel and familiar objects during testing.

Both control (exposed to vehicle gas; n = 10) and isoflurane-treated (exposed to 1 hour, 1.3%

isoflurane; n = 10) mice demonstrated normal memory performance as they spent more time

exploring the novel than the familiar object. (B) The discrimination ratios (time spent with

novel object/time spent with both objects) of control and isoflurane-treated mice. The dotted line represents a chance level of interaction with the novel object (discrimination ratio = 0.5).

(C) Time spent exploring identical sample objects during training. Data are represented as mean ± SEM. * denotes significance at P < 0.05.

105

106

Figure 4.3 Impaired 1-hour recognition memory 24 hours after isoflurane anesthesia.

One day after isoflurane exposure, mice were trained on the object recognition paradigm and

tested 1 hour later. (A) Time spent with novel and familiar objects during testing. Control

(exposed to vehicle gas; n = 31) mice demonstrated normal memory performance as they

spent more time exploring the novel than the familiar object, whereas isoflurane-treated

(exposed to 1 hour, 1.3% isoflurane; n = 31) mice spent an equal amount of time with each

object and hence, impaired memory performance. (B) The discrimination ratios (time spent

with novel object/time spent with both objects) of control and isoflurane-treated mice. The

dotted line represents a chance level of interaction with the novel object (discrimination ratio

= 0.5). (C) Time spent exploring identical sample objects during training. Data are represented as mean ± SEM. * denotes significance at P < 0.05.

107

108

4.3.2 L-655,708 reverses memory deficits after isoflurane

We next sought to determine whether L-655,708 reversed the short-term memory impairment detected 24 hours after isoflurane anesthesia. Mice were exposed to isoflurane, followed by

L-655,708 (0.35 mg/kg or 0.7 mg/kg, i.p.) or vehicle administered 23.5 hours after anesthesia and 30 minutes before behavioural training. Memory was assessed 1 hour after training. Mice in the control group (vehicle gas + vehicle injection) spent more time with the novel object than with the familiar object (t = 4.00, df = 30, P < 0.001; Figure 4.4A). As shown in Figure

4.3, mice exposed to isoflurane only (isoflurane + vehicle injection) exhibited memory deficits on the 1-hour memory task and spent a similar amount of time with the novel and familiar objects (t = 0.400, df = 30, P = 0.346, 95% CI -2.24 to 1.51 s; Figure 4.4A). L-

655,708 restored normal memory performance in groups that were exposed to isoflurane

(effect of isoflurane × L-655,708 F2, 102 = 3.59, P = 0.032; Figure 4.4B). A low dose of L-

655,708 (0.35 mg/kg) increased the proportion of time that isoflurane-treated mice spent with

the novel object (discrimination ratio, isoflurane + L-655,708, 0.68 ± 0.03 vs. isoflurane +

vehicle injection, 0.51 ± 0.03, P < 0.05, Tukey’s HSD; Figure 4.4B). Both control and

isoflurane-treated mice that received L-655,708 at 0.35 mg/kg learned the task and spent

more time with the novel object than with the familiar object (control, novel vs. familiar

object, t = 3.53, df = 9, P = 0.003; isoflurane + low-dose L-655,708, novel vs. familiar

object, t = 4.85, df = 10, P < 0.001; Figure 4.4A).

We also tested whether a higher dose of L-655,708 (0.7 mg/kg) also reversed the

memory deficit. This higher dose of L-655,708 failed to reverse memory deficits in

isoflurane-treated mice (discrimination ratio, isoflurane + vehicle injection 0.54 ± 0.04;

isoflurane + high-dose L-655,708 0.54 ± 0.05) and both vehicle and L-655,708 injected

109

groups that were exposed to isoflurane spent similar amounts of time spent with the novel

and familiar objects (Isoflurane + vehicle injection, novel vs. familiar object, t = 1.42, df = 9,

P = 0.094, 95% CI -1.63 to 7.10 s; isoflurane + high dose L-655,708, novel vs. familiar

object, t = 1.29, df = 8, P = 0.117, 95% CI -1.60 to 5.67 s), although the sample sizes were small for this comparison.

The memory performance could not be attributed to changes in exploratory behaviour

during training, as treatment with isoflurane and L-655,708 did not influence the amount of time that mice spent with both objects (effect of isoflurane, F1,99 = 0.15, P = 0.703 ; effect of

L-655,708, F2,99 = 2.40, P = 0.096; effect of isoflurane + L-655,708, F2,99 = 2.07 , P = 0.132;

Figure 4.4C).

110

Figure 4.4 Memory deficits can be reversed by inhibition of α5GABAA receptors 24 hours after isoflurane anesthesia.

Mice were trained on the object recognition paradigm 24 hours after anesthesia and memory was tested 1 hour after training. Mice received either an injection of vehicle (10% DMSO) or the α5GABAAR-selective inverse agonist L-655,708 (0.35 mg/kg, i.p.) 23.5 hours after anesthesia and 30 minutes before behavioural training. Control = vehicle gas (30% O2, 70% air, 1 hour) and vehicle injection (n = 31); Isoflurane = isoflurane (1.3%, 1 hour) and vehicle injection (n = 31); L-655,708 = vehicle gas and L-655,708 (0.35 mg/kg, i.p.; n = 10);

Isoflurane + L-655,708 = isoflurane (1.3%, 1 hour) and L-655,708 (0.35 mg/kg, i.p.; n = 11)

(A) Time spent with novel and familiar objects during testing. (B) Discrimination ratios of

Control, L-655,708, and Isoflurane+L-655,708 groups demonstrate learning. The dotted line represents a chance level of interaction with the novel object (discrimination ratio = 0.5). (C)

Time spent exploring identical sample objects during training. Data are represented as mean

± SEM. * denotes significance at P < 0.05.

111

112

4.3.3 Memory performance 72 h after isoflurane

To determine whether memory deficits persisted beyond the first 24 hours, mice were studied

72 hours after isoflurane treatment. Mice were trained on the object recognition task 72 hours

after isoflurane anesthesia and tested 1 hour later. At that time point, control and isoflurane-

treated mice showed normal recognition memory, as evidenced by a preference for the novel

object in both groups (control, novel vs. familiar object, t = 2.86, df = 9, P = 0.009; isoflurane, novel vs. familiar object; t = 2.45, df = 9, P = 0.018; Figure 4.5A). The discrimination ratios were similar between the two groups (control, 0.65 ± 0.05 vs. isoflurane, 0.60 ± 0.04; t = 0.787, df = 18, P = 0.441, 95% CI -0.08 to 0.17; Figure 4.5B), which indicates that learning and recognition memory recovered by 72 hours after isoflurane treatment. The normal memory performance could not be attributed to differences in exploratory activity between groups as both control and isoflurane-treated mice spent a similar amount of time exploring both objects during the training phase (control, 45.92 ±

9.44 s vs. isoflurane, 46.65 ± 6.57 s; t = 0.06, df = 18, P = 0.95, 95% CI -24.89 to 23.43 s;

Figure 4.5C). However, since the 95% confidence interval for the difference between the means is quite wide, this experiment may have been underpowered to detect differences in exploratory behaviour during training.

113

Figure 4.5 Normal recognition memory 72 hours after isoflurane anesthesia.

Mice were trained on the object recognition paradigm 72 hours after anesthesia and memory was tested 1 hour after training. (A) Time spent with novel and familiar objects during testing. Control (exposed to vehicle gas; n = 10) and Isoflurane (exposed to 1 hour, 1.3% isoflurane; n = 10) mice demonstrated normal memory performance as they spent more time exploring the novel than the familiar object. (B) The discrimination ratios (time spent with novel object/time spent with both objects) of Control and Isoflurane-treated mice. The dotted line represents a chance level of interaction with the novel object (discrimination ratio = 0.5).

(C) Time spent exploring identical sample objects during training. Data are represented as mean ± SEM. * denotes significance at P < 0.05.

114

115

4.3.4 Memory performance of Gabra5-/- mice 24 h after isoflurane

Based on our previous study, which showed that L-655,708 administered prior to anesthesia can prevent memory deficits (Saab, et al. 2010), and the results presented above, we predicted that mice lacking α5GABAA receptors would not exhibit postanesthesia memory

deficits. To test this postulate, Gabra5-/- mice were trained on the object recognition

paradigm 24 hours after anesthesia and tested 1 hour later. The performance of control

Gabra5-/- mice and control WT mice did not differ significantly (discrimination ratio 0.66 ±

0.05 vs. 0.74 ± 0.05; t = 1.24, df = 19, P = 0.230, 95% CI -0.23 to 0.06). Control Gabra5-/-

mice spent more time with the novel object than with the familiar object (novel vs. familiar

object, t = 2.56, df = 11, P = 0.013; Figure 4.6A). As predicted, Gabra5-/- mice exposed to

isoflurane also showed a preference for the novel object (novel vs. familiar object, t = 2.51,

df = 11, P = 0.015; Figure 4.6A). Isoflurane did not cause significant impairment of memory

performance in Gabra5-/- mice 24 hours after anesthesia (discrimination ratio, control, 0.66

± 0.05; isoflurane, 0.62 ± 0.05; effect of isoflurane, F1,47 = 0.38, P = 0.544; Figure 4.6B).

Gabra5-/- mice treated with L-655,708 (0.70 mg/kg) 10 minutes before exposure to

isoflurane or vehicle gas also learned the task and preferred the novel object over the familiar

object (L-655,708, novel vs. familiar object, t = 2.10, df = 11, P = 0.030; Isoflurane + L-

655,708, novel vs. familiar object, t = 3.83, df = 10, P = 0.002). L-655,708 did not affect the memory performance of Gabra5-/- mice exposed to vehicle gas or isoflurane (discrimination ratio, 0.64 ± 0.05 vs. 0.62 ± 0.05; effect of L-655,708, F1,47 = 0.02, P = 0.90; Figure 4.6B).

No significant interactions were observed (effect of isoflurane × L-655,708, F1,46 = 0.02, P =

0.41; Figure 4.6B). Nonparametric analysis also revealed no differences in discrimination

ratios between treatment groups (χ2 = 1.80, df = 3, P = 0.616). Additionally, neither

116 isoflurane nor L-655,708 influenced exploratory behaviour during training (effect of isoflurane, F1,43 = 0.04, P = 0.851; effect of L-655,708, F1,43 = 1.17, P = 0.285; effect of isoflurane × L-655,708, F1,43 = 0.66, P = 0.423; Figure 4.6C).

117

Figure 4.6 Gabra5-/- mice exhibit no recognition memory deficits 24 hours after isoflurane anesthesia.

Mice were trained on the object recognition paradigm 24 hours after anesthesia and memory was tested 1 hour after training. Mice received either an injection of vehicle (10 % DMSO) or the α5GABAA receptor-selective inverse agonist L-655,708 (0.70 mg/kg, i.p.) 10 minutes before gas exposure. Control = vehicle gas (30% O2, 70% air, 1 hour) and vehicle injection

(n = 12); Isoflurane = isoflurane (1.3%, 1 hour) and vehicle injection (n = 12); L-655,708 = vehicle gas and L-655,708 (0.70 mg/kg, i.p.; n = 12); Isoflurane + L-655,708 = isoflurane

(1.3%, 1 hour) and L-655,708 (0.70 mg/kg, i.p.; n = 11) (A) Time spent with novel and familiar objects during testing. All groups all spent more time with the novel than the familiar object. (B) The discrimination ratios of all groups demonstrate normal memory performance. The dotted line represents a chance level of interaction with the novel object

(discrimination ratio = 0.5). (C) Time spent exploring identical sample objects during training. Data are represented as mean ± SEM. * denotes significance at P < 0.05.

118

119

4.3.5 Prevention of postanesthesia memory deficits

L-655,708, administered before isoflurane, has been shown to prevent deficits in fear

memory in the early postanesthetic period (Saab, et al. 2010). We sought to determine

whether the deficit in recognition memory could also be prevented by administering L-

655,708 before isoflurane as a positive control. L-655,708 (0.7 mg/kg) was injected and 10

minutes later, mice were exposed to isoflurane. Memory performance was studied 24 hours

after anesthesia. Control mice showed the predicted preference for the novel object (time

spent with novel vs. familiar object, t = 3.43, df = 8, P = 0.045, Figure 4.7A; discrimination

ratio 0.74 ± 0.04, Figure 4.7B). In contrast, mice exposed to isoflurane did not prefer the

novel object (time spent with novel vs. familiar object, t = 0.64, df = 10, P = 0.268, 95% CI -

4.90 to 8.85 s; Figure 4.7A), and the discrimination ratio was 0.53 ± 0.05, similar to a chance

level of interaction with the novel object (Figure 4.7B). Again, isoflurane administered to

WT mice 24 hours before training impaired their performance in the object recognition task

(effect of isoflurane, F1,38 = 10.39, P = 0.003; Figure 4.7B).

Mice treated with L-655,708 before exposure to vehicle gas showed normal learning and preference for the novel object (time spent with novel vs. familiar object, t = 4.91, df = 9,

P < 0.001; Figure 4.7A). Mice treated with L-655,708 before exposure to isoflurane

preferred the novel object (time spent with novel vs. familiar object, t = 2.48, df = 8, P =

0.019; Figure 4.7A). L-655,708 alone did not significantly enhance or diminish

discrimination ratios across any of the groups (effect of L-655,708, F1,39 = 1.59, P = 0.215;

Figure 4.7B). There was no significant interaction between L-655,708 and isoflurane (F1,39 =

0.614, P = 0.439; Figure 4.7B) although the study may have been underpowered to detect a difference between groups. Neither isoflurane nor L-655,708 influenced the amount of time

120

that mice spent with both objects during training (effect of isoflurane, F1,38 = 1.43, P = 0.240;

effect of L-655,708, F1,38 = 1.54, P = 0.223; effect of isoflurane × L-655,708, F1,38 = 0.10, P

= 0.752; Figure 4.7C).

4.3.6 Memory performance 24 h after sevoflurane

Finally, to determine whether the postanesthetic impairment in recognition memory occurs after exposure to another commonly used inhaled anesthetic, mice were treated with sevoflurane and trained on the object recognition task 24 hours later. Mice were tested 1 hour after training. Control, vehicle-gas treated mice demonstrated normal memory performance and preferred the novel object (novel 33.92 ± 5.03; familiar 20.03 ± 3.03, t = 5.00, df = 9, P

< 0.001). In contrast, sevoflurane-treated mice spent similar amounts of time with the novel

and familiar object and hence, did not learn the task (novel 15.41 ± 2.25; familiar 13.13 ±

1.95, t = 1.07, df = 9, P = 0.157, 95% CI -2.56 to 7.12 s). The discrimination ratio for the

sevoflurane-treated group was lower than that for the control group and was similar to that

predicted by chance (sevoflurane 0.53 ± 0.03 vs. control 0.63 ± 0.02, t = 2.22, df =18, P =

0.039). A nonparametric analysis revealed that the difference between the discrimination

ratios of control and sevoflurane-treated groups only approached significance (Z = -1.66, P =

0.096). Exploratory behaviour during the training phase was not affected by exposure to

sevoflurane (sevoflurane 23.45 ± 3.88 s vs. control 36.46 ± 4.77 s, t =2.14, df = 13, P =

0.052, 95% CI -0.14 to 26.16 s).

121

Figure 4.7 Memory deficits can be prevented by inhibition of α5GABAA receptors prior to anesthesia.

Mice were trained on the object recognition paradigm 24 hours after anesthesia and memory

was tested 1 hour after training. Mice received either an injection of vehicle (10 % DMSO)

or the α5GABAA receptor-selective inverse agonist L-655,708 (0.70 mg/kg, i.p.) 10 minutes

before gas exposure. Control = vehicle gas (30% O2, 70% air, 1 hour) and vehicle injection

(n = 9); Isoflurane = isoflurane (1.3%, 1 hour) and vehicle injection (n = 11); L-655,708 = vehicle gas and L-655,708 (0.70 mg/kg, i.p.; n = 10); Isoflurane + L-655,708 = isoflurane

(1.3%, 1 hour) and L-655,708 (0.70 mg/kg, i.p.; n = 9) (A) Time spent with novel and familiar objects during testing. All groups spent more time with the novel than the familiar object. (B) The discrimination ratios of all groups during testing. The dotted line represents a chance level of interaction with the novel object (discrimination ratio = 0.5). (C) Time spent exploring identical sample objects during training. Data are represented as mean ± SEM. * denotes significance at P < 0.05.

122

123

4.4 Discussion

Our results show that isoflurane-treated mice exhibited memory deficits on a recognition

memory task with a 1 hour retention period, when compared with vehicle-treated controls as evidenced by the lower discrimination ratios. In contrast, 1-minute memory was not impaired

24 hours after isoflurane as the discrimination ratios were similar between isoflurane- and

vehicle-treated mice. A low dose of the α5GABAA receptor-selective inverse agonist, L-

655,708, administered 24 hours after isoflurane fully reversed the memory deficits. Changes in memory performance could not be attributed to changes in exploratory activity as exposure to isoflurane or L-655,708 did not alter the time that mice spent exploring objects.

Consequently, all mice had an equal opportunity to perceive and learn the characteristics of

the objects. Memory deficits resolved within 72 hours. The expression of α5GABAA

receptors was necessary for the isoflurane-induced deficits in recognition memory to occur,

as Gabra5-/- mice exhibited no memory impairment. Finally, recognition memory deficits

also occurred 24 hours after sevoflurane anesthesia.

The most novel and important finding of this study is that a low dose of L-655,708

(0.35 mg/kg) administered 24 hours after isoflurane completely reversed the deficit in

recognition memory. This result was unexpected, given the widely believed mechanism by

which volatile anesthetics block memory. The concentration-dependent suppression of

memory during acute exposure to isoflurane has been attributed, at least in part, to increased

activity of GABAA receptors (Rau, et al. 2009; Rau, et al. 2011). Isoflurane and other volatile anesthetics, including desflurane and sevoflurane, potentiate GABAA receptor

function and the resulting increase in chloride conductance reduces neuronal excitability

(Hemmings, et al. 2005). In brain networks, such as the CA1 subfield of the hippocampus,

124

the enhanced chloride conductance prevents the synaptic plasticity that subserves memory

formation (Martin, et al. 2010). Once the anesthetic has been eliminated, it is assumed that

memory recovers. However, in these experiments, memory performance was impaired 24

hours after exposure to isoflurane, when the concentration of isoflurane in the brain has

declined to the limits of detection (0.0095%) (Saab, et al. 2010). This low concentration of

isoflurane is orders of magnitude less than the concentration (0.4%) required for memory

blockade during anesthesia (Alkire et al. 2004) and far lower than the concentration required

to allosterically potentiate GABAA receptor activity (Caraiscos, et al. 2004). Taken together,

the available data suggest that a simple interaction between isoflurane and GABAA receptors

could not account for the memory deficits at 24 hours.

An analogous and surprising long-term, behavioural effect of the intravenous

anesthetic ketamine on cognitive function has been reported (DiazGranados et al. 2010).

Ketamine is a noncompetitive antagonist of the NMDA subtype of glutamate receptors

(Orser et al. 1997). A single dose of ketamine causes long-lasting effects that persist after the drug has been eliminated, specifically a rapid and sustained reversal of depression that lasts for weeks to months (Price et al. 2009). The sustained effect of ketamine involves the rapamycin intracellular signalling pathway, which increases synaptic signaling proteins and the number and function of synapses in the cortex (Duman et al. 2012).

Modulation of α5GABAA receptors by isoflurane plays a crucial role in initiating the

memory deficits that were evident at 24 hours, as genetic and pharmacological inhibition of

these receptors prevented memory impairment. L-655,708, administered after isoflurane,

may counteract an unrecognized increase in the function or expression of α5GABAA

receptors that persists after anesthesia in the absence of isoflurane binding to the receptor.

125

Alternatively, L-655,708 may cause a nonspecific compensatory enhancement of memory

processes. The molecular mechanisms that are triggered during periods of high GABAergic

activity during anesthesia, and cause α5GABAA receptor-dependent memory deficits, remain

to be determined.

On the basis of the current study, we propose that inhibition of α5GABAA receptors is

a plausible strategy for reversing memory deficits after general anesthesia in patients. Inverse

agonists that preferentially target α5GABAA receptors lack the adverse effects that typify nonselective GABAA receptor antagonists, such as anxiogenesis and seizures (Atack, et al.

2006). Several human trials have studied this class of drugs (Wallace et al. 2011). The inverse agonist α5IA, which is functionally selective for α5GABAA receptors, attenuated ethanol-induced impairment of word recall when administered before ethanol and was well tolerated by human volunteers (Nutt et al. 2007). Also, the α5GABAA receptor–selective

inverse agonist RG1662 (Roche Pharmaceuticals, Basel, Switzerland) is currently

undergoing phase 1 clinical trials for treatment of cognitive deficits in patients with

Alzheimer disease (Wallace, et al. 2011). Our results suggest that the dose of inverse agonist

must be selected carefully, as a high dose administered immediately before learning may

actually impair memory performance (Savic et al. 2008). α5GABAA receptors have a high affinity for L-655,708, which reduces receptor activity even at low doses (Quirk, et al.

1996). However, at higher doses, L-655,708 may affect the activity of other receptor populations that have a lower affinity for the drug. Higher doses may exert agonist-like

effects on non-α5GABAA receptors, thus increasing the activity of GABAA receptors and

impairing memory performance (Savic, et al. 2008).

Several of our results are consistent with previous studies. The interaction times and

126

discrimination ratios for object recognition measured under baseline, control conditions in

the current study were comparable to those reported by others for rodents (Bertaina-Anglade,

et al. 2006). Also, the memory deficits observed in isoflurane-treated mice were consistent

with a deficit in retrograde memory observed in mice treated with sevoflurane (2.6% for 2

hours) and then conditioned with two object learning sessions (Wiklund, et al. 2009).

Notably, we observed that isoflurane impaired 1-hour recognition memory, whereas 1-minute memory remained intact. These results suggest that isoflurane spares the perception and encoding of information but disrupts consolidation of memory into long-term storage or memory retrieval. Our results predict that patients exposed to isoflurane could exhibit normal recall for immediate events but might exhibit deficits in recall for events after a longer delay.

Similar effects have been found after exposure to the benzodiazepines and diazepam: object recognition was intact when rats were tested 10 minutes after training but impaired when they were tested 1 hour after training (Bertaina-Anglade, et al. 2006).

The current study raises many additional important questions for future study. It remains to be determined whether isoflurane triggers downstream events that impair memory through processes that initially require the interaction between isoflurane and α5GABAA

receptors. A causal role for these receptors in memory deficits is supported by results that

show inhibition and genetic deletion of α5GABAA receptors prevents the deficits. In

addition, although memory deficits resolved spontaneously within 72 hours, it will be critical to determine whether higher doses of isoflurane (due to higher concentrations and/or longer durations of treatment) prolong the memory deficit. Also, it must be determined whether factors that impair recognition memory performance, such as age and inflammation, exacerbate isoflurane-induced memory loss. Previous studies have shown that age

127 exacerbates postanesthetic memory loss; for example, aged rats exhibit impaired anterograde memory for up to 2 weeks after anesthesia, whereas adult rats are no longer impaired at that time point (Culley, et al. 2004). The object recognition task is a versatile experimental model to study such interactions, as it requires no appetitive or aversive reinforcement, and it has the potential for high throughput (Bevins and Besheer 2006).

In summary, isoflurane impairs recognition memory for at least 24 hours in an ethologically relevant paradigm. Furthermore, α5GABAA receptors are necessary for the development of postanesthetic memory deficits and are a potential therapeutic target for restoring memory after general anesthesia.

128

Chapter 5. Sustained increase in α5GABAA receptor function impairs memory after anesthesia 5.1 Introduction

Each year, more than 234 million surgical procedures are performed under anesthesia worldwide (Weiser, et al. 2008). A large proportion of patients exhibit cognitive impairment,

including memory deficits, after surgery and anesthesia (Monk, et al. 2008). Such

postoperative cognitive deficits are present in approximately 37% of young adults and 41%

of elderly patients at hospital discharge and in 6% of young adults and 13% of elderly

patients at 3 months after surgery (Monk, et al. 2008). These deficits are associated with

poor patient outcomes, including reduced quality of life, loss of independence and increased

mortality (Monk, et al. 2008; Price, et al. 2008).

The cause of postoperative cognitive dysfunction is multifactorial. For example,

inflammation triggered by surgical trauma and underlying medical conditions appears to

contribute to cognitive deficits in both human patients and laboratory animals (Cibelli, et al.

2010; Peng et al. 2013). General anesthetics may also play a causal role, given that the

duration of anesthesia is positively correlated with the incidence of postoperative cognitive

deficits in patients (Moller, et al. 1998). In addition, studies using rodent models have shown

that a single exposure to an anesthetic can cause retrograde and anterograde memory deficits

that persist for days to weeks (Crosby et al. 2005; Culley, et al. 2003). The mechanisms by

which anesthetics cause persistent memory deficits in adult animals remain poorly

understood.

129

Most general anesthetics used in clinical practice act as direct, positive allosteric

modulators of inhibitory GABAA receptors (Rudolph and Antkowiak 2004). During

anesthesia, increased activity of GABAA receptors contributes to the desired neurodepressive

properties of these drugs, including amnesia (Cheng, et al. 2006). Once the anesthetic is

eliminated, positive allosteric modulation of GABAA receptor function is rapidly reversed, on

a time scale of seconds (Belelli et al. 1996). Consequently, it has been assumed that receptor

activity returns to baseline and that these receptors do not contribute to undesirable memory

deficits that persist after anesthesia. Here, I tested the hypothesis that even a brief exposure to

an anesthetic triggers a sustained increase in GABAA receptor function and that this increase

causes persistent memory deficits.

5.2 Methods 5.2.1 Experimental animals.

Gabra5-/- mice and WT mice (C57BL/6J × SvEv129) were housed in the Animal Care

Facility, University of Toronto. For all behavioural tests, age-matched 3- to 5-month-old male WT and Gabra5-/- mice were used. For electrophysiological and cell-surface biotinylation experiments, 1- to 4-month-old male mice were studied. Researchers were blinded to all drug conditions. In addition, researchers were blinded to the genotype for all behavioural experiments.

5.2.2 Anesthesia.

Mice were treated with sedating doses of etomidate (8 mg/kg, i.p.), dexmedetomidine (200

μg/kg, i.p.) or isoflurane (0.7%, 20 min). Physiological saline (i.p.) was used as the vehicle

for sedative doses of etomidate and dexmedetomidine, and 30% O2 (20 minutes) mixed with

130

air was used as the vehicle for isoflurane. Anesthetizing doses of etomidate (20 mg/kg, i.p.)

or isoflurane (1.3%, 1hour) and their corresponding vehicles (propylene glycol 26% v/v in physiological saline and 30% O2, respectively) were also used. The sedating and

anesthetizing doses of etomidate and isoflurane were selected from the literature to

approximate the ED50 and ED100 for the loss of righting reflex, respectively (Cheng, et al.

2006; Sonner, et al. 2000; Sonner, et al. 2007).

To prevent hypoxia during anesthesia, each mouse was placed in an air-tight acrylic

chamber (27 cm × 10 cm × 10 cm) that was flushed with supplemental oxygen and medical

air (70% air, 30% O2) delivered at a flow of 1 L/min. In a subset of mice, transcutaneous oxygen saturation was measured at a frequency of 15 Hz with a mouse pulse oximetry sensor

(MouseOx, Starr Life Sciences Corp., Allison PA) that was placed on a shaved area of the throat over the carotid arteries. None of the anesthetic treatments caused hypoxia as oxygen saturation remained above 98% during etomidate (20 mg/kg i.p.) and isoflurane (1.3%, 1 h).

To prevent hypothermia, the temperature of the chamber was maintained at 35°C with a heating blanket (Saab, et al. 2010).

5.2.3 Novel object recognition memory assay.

Mice were handled for at least 10 minutes daily for 5 days before the start of behavioural

experiments. Object recognition was assessed in a 20 cm × 20 cm × 30 cm opaque chamber

in a dimly lit room. Each mouse was habituated to the chamber for 15 minutes one day

before testing. During the training phase, the mouse was allowed to explore two identical

objects for 10 minutes. The mouse was then returned to its home cage for a retention period

of 1 hour. The mouse was reintroduced to the training context and presented with one

familiar object and one novel object for 5 minutes. Movement and interaction with the

131 objects was recorded with a video camera that was mounted above the chamber and exploratory behaviour was measured by a blinded observer. Exploratory behaviour was defined as sniffing, licking, or touching the object while facing the object. Memory was assessed by measuring the discrimination ratio (i.e. the ratio of time spent exploring the novel object to the time spent exploring both objects). Animals that did not interact for a minimum of 1 second with each object during the test period were excluded. Total interaction time with both objects was compared between groups to determine whether the treatments affected locomotor activity or exploration during the testing phase. Each mouse treated with dexmedetomidine, etomidate (8 mg/kg, i.p.) or the corresponding vehicle was trained and tested 3 times; 24 hours, 72 hours, and 1 week after anesthetic exposure. Each mouse treated with etomidate (20 mg/kg, i.p.) was trained and tested at 4 times; 24 hours, 72 hours, 1 week and 2 weeks. Different pairs of objects were used for each training session. A separate group of mice was treated with the inverse agonist for α5GABAA receptors, L-655,708 (0.5 mg/kg i.p.) or vehicle (2% DMSO in physiological saline) 30 minutes before training on the object recognition task.

5.2.4 Electrophysiology in hippocampal slices.

After live decapitation, brains were removed and placed in ice-cold, oxygenated (95% O2,

5% CO2) artificial cerebrospinal fluid (ACSF) that contained (in mM): 124 NaCl, 3 KCl, 1.3

MgCl2, 2.6 CaCl2, 1.25 NaH2PO4, 26 NaHCO3 and 10 D-glucose with the solution osmolarity adjusted to 300–310 mOsm. Coronal brain slices (350 μm) were prepared with a

VT1200S vibratome (Leica, Deerfield, Illinois). The slices were allowed to recover for at least 1 hour at room temperature (23–25 °C) before being transferred to a submersion recording chamber, where they were perfused with ACSF at 3–4 ml/min. All recordings were

132

performed at room temperature using a MultiClamp 700B amplifier (Molecular Devices,

Sunnyvale, California) controlled with pClamp 9.0 software via a Digidata 1322A interface

(Molecular Devices, Sunnyvale, California).

For extracellular recording of field postsynaptic potentials (fPSPs), the pipettes were

filled with ACSF and placed in the stratum radiatum of the CA1. The Schaffer collateral

pathway was stimulated with 0.1 ms pulses delivered by a concentric bipolar tungsten

electrode (Rhodes Medical Instruments, Summerland, California). To record plasticity of

fPSPs, the slope of fPSPs was measured at baseline for at least 10 minutes with a stimulation

frequency of 0.05 Hz and using a stimulation intensity that produced a half-maximal response. Slices were then stimulated at 20 Hz (600 pulses), and fPSPs were monitored for

60 minutes after stimulation. The average of the last 5 minutes of recording was compared with the average of the baseline fPSPs. Post-tetanic depression was measured during the first

2 minutes after stimulation. Short-term depression was measured during the first 15 minutes after stimulation. Paired pulse facilitation was measured during the 20 Hz stimulation (inter- stimulus interval of 50 ms). The slope of the fPSP after the first stimulus pulse (fPSP1) and the second stimulus pulse (fPSP2) were measured. The paired pulse ratio was calculated by dividing the slope of fPSP2 by the slope of fPSP1.

For whole-cell voltage-clamp recordings, the pipettes (2–3 MΩ) were filled with the intracellular solution containing (in mM): 140 CsCl, 10 HEPES, 11 EGTA, 4 Mg2ATP, 2

MgCl2, 1 CaCl2, 2 TEA (pH 7.3 with CsOH, 290–300 mOsm). Currents were sampled at 10 kHz. All cells were recorded at a holding potential of −60 mV and automatic capacitance compensation was applied. To measure the tonic current, the GABAA receptor competitive

antagonist, bicuculline (10 μM) or L-655,708 (200 nM), was applied. The tonic current was

133

quantified by measuring the change in the holding current from segments containing no

synaptic events. Exogenous GABA (5 μM) was added to ACSF.

5.2.5 Primary cell culture.

Primary cultures of hippocampal neurons were prepared from Swiss Webster mice (Charles

River, Montreal, Canada), as described previously (MacDonald, et al. 1989). For the first 5

days in vitro, cells were maintained in minimum essential media supplemented with 10%

fetal bovine serum and 10% horse serum (Life Technologies, Grand Island, New York). The

o neurons were cultured at 37 C in a 5% CO2-95% air environment. After the cells had grown to confluence, 0.1 ml of a mixture of 4 mg 5-fluorodeoxyuridine and 10 mg uridine in 20 ml minimum essential media was added to the extracellular solution to reduce the number of dividing cells. Subsequently, the media was supplemented with 10% horse serum and changed every 3 or 4 days. Neurons were maintained in culture for 14 to 21 days prior to

recording. To prepare microglia-neuron cocultures, cortical microglia were isolated from

embryonic mice as described previously (Deierborg 2013). Cells were maintained as

described above and media was changed every 3 to 4 days. Once cell confluence was

achieved, microglia were separated from the mixed glial cultures by gently shaking the

dishes (200 revolutions per minute for 2 hours at 37° C) then centrifuging the supernatant.

The pellet was suspended in neurobasal media and applied over cultured hippocampal

neurons. For astrocyte culture, cortical astrocytes were isolated from embryonic day 18

mouse embryos as described previously (Albuquerque, et al. 2009). Cells were allowed to

grow to confluence in minimum essential media and 10% fetal bovine serum (FBS; Life

Technologies, Grand Island, New York) for 14 days. Cells were then enzymatically

dissociated with trypsin–EDTA (0.05%; Life Technologies, Grand Island, New York), and

134

passaged three times to obtain a nearly pure astrocytic culture. Astrocytes were then plated at

a density of 25,000 cells per dish. For astrocyte-neuron coculture, astrocyte cell suspension

was placed over hippocampal neurons. For all reported results, data were acquired from cells

from at least three different dissections.

5.2.6 Whole-cell voltage-clamp recordings in cell culture.

Whole-cell recordings were performed as described above with several exceptions. The

extracellular solution contained the following (in mM): 140 NaCl, 2.0 KCl, 1.3 CaCl2, 1

MgCl2, 25 HEPES, and 28 glucose (pH 7.4, 320–330 mOsm). Etomidate (0.25 μM or 1 μM)

or vehicle solution was used to treat the culture dish for 1 hour. The concentrations of 0.25

μM and 1 μM of etomidate were selected because they correspond to low, sedative and

anesthetizing doses of etomidate, respectively (Alkire and Gorski 2004; Benkwitz et al.

2007). The medium was then removed and replaced with fresh culture media. Recordings were performed 24 hours later. To prepare conditioned supernatant media, astrocyte cultures

were treated with etomidate (1 μM) for 1 hour. The media was then removed and replaced with fresh culture media. The astrocytes were kept in the fresh media for 2 hours before the conditioned media was collected and applied to neuronal culture for 24 hours. To measure the amplitude of the tonic current, exogenous GABA (0.5 μM) was added to the extracellular solution and the change in holding current was measured during application of bicuculline

(20 μM).

5.2.7 Cell-surface biotinylation.

Coronal hippocampal slices (350 μm) were prepared 24 hours, 1 week or 2 weeks after in vivo treatment with etomidate or isoflurane and placed in oxygenated ACSF (95% O2, 5%

CO2) for 1 hour to recover. Slices were then placed on ice and incubated twice with 0.75

135

mg/ml NHS-SS-biotin (Thermo Scientific, Rockford, Illinois) dissolved in DPBS (Gibco,

Burlington, Ontario, Canada) for 30 minutes each time. Excess biotin was quenched and

removed by washing slices 6 times with ice cold modified TBS containing (in mM): 25 Tris-

Cl, 137 NaCl, 1 KCl, 2.3 CaCl2, pH 7.4. Slices were then placed in lysis buffer (pH 7.4)

containing complete protease inhibitor cocktail (Roche, Laval, Quebec, Canada) for

homogenization. Insoluble material was removed by centrifugation. Bicinchoninic acid assay

(Bio-Rad, Hercules, California) was performed to determine protein concentration. The

supernatant lysates were incubated with Hi-Capacity NeutrAvidin beads (Thermo Scientific,

Rockford, Illinois) for 16-18 hours at 4°C. The beads were washed with PBS containing

0.05% SDS. Bound material was eluted with elution buffer containing (in mM): 50 Tris-Cl,

2% SDS, 2 DTT; protein concentration was determined using DC™ Protein Assay (Bio-Rad,

Hercules, California) and subjected to SDS-PAGE analysis. A western blot analysis with

anti-GABAA receptor antibodies for α1 (Abcam, Cambridge, Massachusetts or Millipore,

Billerica, Massachusetts), α5 (PhosphoSolutions, Aurora, Colorado), β3 (Thermo Scientific,

Rockford, Illinois), or δ (Millipore, Billerica, Massachusetts) was performed. Anti-β-actin

antibody (Millipore, Billerica, Massachusetts) and anti-Na+/K+ ATPase antibody

(Developmental Studies Hybridoma Bank, Iowa City, Iowa) were also used. Blots were imaged using the Chemidoc XRS+ system (Bio-Rad, Hercules, California) and quantified using Image Lab software (Bio-Rad, Hercules, California). All samples were loaded onto two replicate blots in order to control for transfer errors. Receptor bands were normalized to their respective loading controls, which were Na+/K+ ATPase for surface protein and β-actin for

total protein. Blots containing surface protein were probed for β-actin to determine the purity

136

of isolated biotinylated surface protein. Data in each experiment are presented as a percentage of the mean of the control group.

5.2.8 Statistical analyses.

Results are presented as means ± standard error of the mean (SEM). An unpaired Student’s t

test was used to compare groups where appropriate. For comparing three or more groups, a

one-way analysis of variance followed by Dunnett’s test was applied. The Kolmogorov-

Smirnov test and Shapiro-Wilk test were used to validate the assumption of normality. In

cases where the assumption of normality was not met for one or more groups, the Mann-

Whitney U test was employed. Statistical Package for the Social Sciences (SPSS version

17.0, IBM Corporation, Armonk, New York) and GraphPad Prism software, version 4.0

(GraphPad Software, San Diego, California) were used. A P value less than 0.05 was

considered statistically significant.

5.3 Results

First, we investigated whether a single exposure to the injectable anesthetic etomidate causes

postanesthetic memory deficits in mice using the novel object recognition memory assay. We

selected etomidate because it preferentially binds to GABAA receptors and is rapidly metabolized to inactive metabolites (Belelli, et al. 1997; Forman 2011). Mice were treated with a low, sedative dose of etomidate (8 mg/kg, i.p.). This dose approximates the half-

maximal effective dose (ED50) for loss of the righting reflex (LORR), a widely used

surrogate measure for anesthetic-induced loss of consciousness in animal models (Cheng, et

al. 2006; McCarren et al. 2013; Rudolph and Antkowiak 2004). Memory was impaired at

24 hours and 72 hours but not at 1 week after treatment with etomidate (Figure 5.1A). In

137 contrast, no memory deficits were observed in mice treated with the active comparator, dexmedetomidine (200 μg/kg, i.p.), a sedative α2-adrenergic receptor agonist that does not target GABAARs (Sanders et al. 2007) (Figure 5.1A). Memory performance was not confounded by sedation or by reduced exploratory behaviour, as the total interaction time of mice with both objects was similar in all groups (Figure 5.1B).

138

Figure 5.1 A sedative dose of etomidate impairs memory for 72 hours in WT mice.

(A) Memory performance on the novel object recognition task after etomidate (8 mg/kg, i.p) or dexmedetomidine (200 μg/kg, i.p.) and (B) total interaction time with both objects during

testing (n = 9-12, one-way ANOVA at each time point, Dunnett’s post-test). Ctrl: vehicle

control, Etom: etomidate, Dex: dexmedetomidine. Data are shown as mean ± SEM.

***P<0.001.

139

Next, to determine whether synaptic plasticity, a cellular correlate of memory, was

impaired after etomidate (8 mg/kg), ex vivo slices were prepared from etomidate- and

vehicle-treated mice 24 hours and 1 week after in vivo drug treatment. Field postsynaptic

potentials (fPSPs) were recorded at Schaffer collateral-CA1 synapses in hippocampal slices that were stimulated at a threshold frequency (20 Hz) for inducing synaptic potentiation

(Martin, et al. 2010). Under these conditions, synaptic plasticity is sensitive to changes in

GABAA receptor activity (Martin, et al. 2010; Steele et al. 1999). Twenty-four hours after treatment, the potentiation of fPSPs was significantly lower in slices from etomidate-treated mice than in slices from vehicle-treated mice (60 minutes after stimulation; control 108% versus etomidate-treated 132% of baseline respectively; Figure 5.2A). In addition, post- tetanic depression of fPSPs (2 minutes after stimulation; Figure 5.2B) and short-term

depression of fPSPs (15 minutes after stimulation; Figure 5.2C) occurred in slices from

etomidate-treated but not vehicle-treated mice. Paired-pulse facilitation, a presynaptic form

of short-term plasticity (Schulz et al. 1994), was similar in the two groups, which suggests

that there were no differences in the release of neurotransmitters from presynaptic terminals

(Figure 5.2D). Since memory performance recovered 1 week after etomidate, we studied plasticity at the 1 week time point. One week after etomidate (8 mg/kg), there was no

significant difference in potentiation of fPSPs between groups, although there was a trend

towards a reduction (Ctrl 127% vs Etom 113% of baseline, Figure 5.3A). At 1 week, post- tetanic depression and short-term depression were no longer observed (Figure 5.3B,C).

Paired-pulse facilitation was also similar in both groups (Figure 5.3D). Thus, potentiation of

fPSPs was reduced 24 h after etomidate and recovered by 1 week.

140

Figure 5.2 Plasticity of fPSPs is reduced 24 hours after sedation with etomidate.

(A) Plasticity at Schaffer collateral-CA1 synapses 24 hours after treatment with etomidate (8 mg/kg, i.p.). Insets: representative traces recorded before (1) and 60 min after (2) 20 Hz stimulation. Right panel summarizes data for the last 5 min of recording 24 hours (n = 7 slices) after treatment with etomidate. (B) Post-tetanic depression and (C) short-term depression are observed in slices from etomidate-treated mice but (D) paired-pulse facilitation is not affected by etomidate treatment. fPSP: field postsynaptic potential. An unpaired, two-tailed Student’s t-test was used to compare between groups. Data are shown as mean ± SEM. * P< 0.05, *** P< 0.001.

141

142

Figure 5.3 Plasticity at Schaffer collateral-CA1 synapses 1 week after sedation with etomidate.

The Schaffer collateral pathway was stimulated at 20 Hz (600 pulses) and fPSPs were recorded for 60 minutes. (A) Data for the last 5 minutes of recording 1 week (n = 9-10 slices) after treatment with etomidate (8 mg/kg, i.p.). (B) Post-tetanic depression, (C) short-term depression and (D) paired-pulse facilitation are similar in slices from vehicle and etomidate- treated mice. fPSP: field postsynaptic potential. An unpaired, two-tailed Student’s t-test was used to compare between groups (P > 0.05 for each comparison). Data are shown as mean ±

SEM.

143

To determine whether GABAA receptor activity was persistently increased after

etomidate (8 mg/kg), we recorded miniature inhibitory postsynaptic currents (mIPSCs) and a

tonic inhibitory current in CA1 pyramidal neurons (Farrant and Nusser 2005). Twenty-four

hours after etomidate, the amplitude, frequency and time course of mIPSCs were unchanged,

suggesting no change in the activity of postsynaptic GABAARs (Figure 5.4 and Table 1). In

contrast, the tonic current was increased to 168% of control (Figure 5.5A, B). The increase in

tonic current persisted at 72 hours and 1 week, but not at 2 weeks (Figure 5.5B). Treatment with dexmedetomidine, the anesthetic that did not impair memory performance, caused no increase in tonic current (Figure 5.5C).

Tonic current in CA1 pyramidal neurons is generated primarily by α5GABAA

receptors (Caraiscos, et al. 2004). To determine whether α5GABAA receptors contributed to

the increased tonic current, slices were perfused with the α5GABAA receptor-selective

inverse agonist L-655,708 (200 nM) (Quirk, et al. 1996). The L-655,708-sensitive current

was increased in slices from etomidate-treated mice (Figure 5.5D). Furthermore, tonic

current was unchanged in slices from Gabra5-/- mice 24 hours after etomidate (Figure 5.5E).

144

Figure 5.4 Miniature inhibitory postsynaptic currents are not affected by etomidate treatment.

(A) Representative recordings of mIPSCs in CA1 pyramidal neurons 24 hours after

etomidate (8 mg/kg, i.p.). (B) Averaged traces from control (red) and etomidate-treated

(black) mice. (C) The cumulative amplitude (P = 0.89) and (D) the cumulative frequency (P

= 0.25) distributions (Kolmogorov-Smirnov test; 125 events).

145

Amplitude Frequency Rise-time Decay-time Area

(pA) (Hz) (ms) (ms) (pA·ms)

Ctrl 35.49 ± 1.46 3.24 ± 0.53 4.01 ± 0.24 10.92 ± 0.75 372.56 ± 16.92

Etom 38.52 ± 1.19 3.24 ± 0.29 3.67 ± 0.22 9.96 ± 0.63 390.92 ± 19.33

Table 5.1 Parameters of GABAergic miniature inhibitory postsynaptic currents 24 h after etomidate. Data are shown as mean ± SEM. n = 10-11, P > 0.05, An unpaired, two-tailed

Student’s t-test was used to compare between control, vehicle-treated (Ctrl) and etomidate- treated (Etom) groups.

146

Figure 5.5 A sedative dose of etomidate causes a sustained increase in α5GABAA receptor- mediated tonic current for at least 1 week.

(A) Representative traces of tonic current recorded in CA1 pyramidal neurons in slices from

WT mice 24 hours after treatment of mice with vehicle or etomidate (8 mg/kg, i.p.). BIC,

bicuculline 10 μM. Tonic current recorded from (B) WT slices 24 hours - 2 weeks after

treatment of mice with etomidate (8 mg/kg i.p., n = 6-19, one-way ANOVA, Dunnett’s post- test), (C) WT slices 24 hours after treatment of mice with dexmedetomidine (200 µg/kg i.p.,

n = 6) and (D) from WT slices measured after application of L-655,708 (L6, 200 nM) 24

hours after etomidate (n = 10-11). (E) Tonic current measured in Gabra5-/- slices 24 h after etomidate (n = 6). Unpaired, two-tailed Student’s t-test unless otherwise indicated. Data are shown as mean ± SEM. *P < 0.05, **P < 0.01.

147

148

We next asked whether etomidate acts directly on neurons to increase tonic current.

Etomidate did not change tonic current in cultured hippocampal neurons 24 hours after

treatment (1 µM, 1 hour, Figure 5.6A). Since glial–neuron interactions might contribute to

the increase in tonic current observed in ex vivo slices, neuron-glia cocultures were treated with etomidate. Etomidate did not change the tonic current in microglia-neuron cocultures 24

hours after treatment (1 μM, 1 hour, Figure 5.6B). However, the tonic current was increased

in neurons cocultured with astrocytes 24 hours after etomidate treatment (0.25 µM or 1 µM,

1 hour, Figure 5.6C). To determine whether etomidate acting on astrocytes was sufficient to

increase the tonic current in neurons, conditioned media was collected from astrocytes

cultured alone and treated with etomidate (1 μM, 1 hour). The astrocyte-conditioned media was then applied to hippocampal neurons for 24 hours. Under these conditions, the tonic current in neurons was increased to 136% of control (Figure 5.6D), suggesting that etomidate treatment of astrocytes stimulates the release of soluble factors that increase tonic inhibition in hippocampal neurons. Together, these results show that treatment of astrocytes with etomidate was necessary and sufficient to trigger an increase in tonic current in hippocampal neurons.

149

Figure 5.6 Treatment of astrocytes with etomidate is necessary and sufficient to trigger a sustained increase in tonic current in neurons.

(A) Tonic current 24 h after etomidate treatment (1 μM, 1 hour) in (A) cultured hippocampal neurons (n = 21), and (B) neurons in microglia-neuron cocultures (n = 19). (C) Tonic current in neurons in astrocyte-neuron cocultures 24 hours after etomidate treatment (0.25 μM, 1 hour, n = 6; 1 μM, 1 hour, n = 10). The representative traces are obtained from astrocyte- neuron cocultures treated with 1 μM etomidate. (D) Tonic current in neurons treated with conditioned media from etomidate-treated astrocytes (n = 19-21). ACM, astrocyte- conditioned media. Unpaired, two-tailed Student’s t-test unless otherwise indicated. Data are shown as mean ± SEM. *P < 0.05, ***P < 0.001.

150

151

We postulated that etomidate enhanced the tonic current by increasing cell-surface expression of α5GABAA receptors in the hippocampus. The cell-surface expression of α5

subunits was indeed increased to 128% of control at 24 hours and to 130% of control 1 week

after treatment of mice with etomidate; however, levels returned to baseline by 2 weeks

(Figure 5.7). The total expression of α5 subunits was unchanged at all time points (Figure

5.7). Cell-surface expression of β3 subunits, which partner with α5 subunits to form GABAA

receptors (Farrant and Nusser 2005), was also increased at 24 hours but not 1 or 2 weeks

after etomidate (Figure 5.8A). We measured the cell-surface and total expression of δ

subunits, which similarly to α5 subunits form receptors that contribute to tonic inhibition in the hippocampus, and α1 subunits, which contribute to synaptic inhibition in the hippocampus (Farrant and Nusser 2005). The cell-surface and total expression of δ subunits

and α1 subunits was unchanged 24 hours after treatment of mice with etomidate (Figure

5.8B,C). Thus, etomidate selectively increased the cell-surface expression of α5GABAA

receptors in the hippocampus.

152

Figure 5.7 The cell-surface expression of α5GABAA receptors is increased for 1 week after

treatment of mice with a sedative dose of etomidate.

Western blots of (A) cell-surface and (B) total expression in hippocampal slices 24 hours, 1

week and 2 weeks after etomidate (8 mg/kg, i.p.) treatment. Separate blots were performed

for each time point (unpaired, two-tailed Student’s t-test for each time point; 24 h, n = 6

mice; 1 wk, n = 5 mice; 2 wk n = 3 mice). NKA, Na+/K+ ATPase. MW, molecular weight

(kDa). Data are shown as mean ± SEM. Unpaired, two-tailed Student’s t-test. * P < 0.05,

**P<0.01.

153

Figure 5.8 Etomidate increases cell-surface expression of β3 subunits but does not change

the cell-surface expression of δ and α1 subunits 24 hours after treatment.

Western blots of cell-surface and total expression of (A) β3 subunits (n = 3 mice), (B) δ subunits (Surface n = 6 mice, Total n = 3 mice), and (C) α1 subunits (n = 3 mice). NKA,

Na+/K+ ATPase. MW, molecular weight (kDa). Samples collected at different time points

were run on separate gels. An unpaired, two-tailed Student’s t-test was used to compare

between groups at each time point. Data are shown as mean ± SEM. *P < 0.05.

154

155

We next explored whether pharmacological or genetic inhibition of α5GABAA receptors reverses memory deficits after etomidate. Previous studies from our lab show that inhibition of α5GABAA receptors after isoflurane treatment can reverse memory deficits in the object recognition task (Zurek, et al. 2012). Treatment with L-655,708 (0.5 mg/kg, i.p.)

30 minutes before training on the novel object recognition task completely reversed the memory deficits after etomidate, whereas L-655,708 alone did not alter performance (Figure

5.9A). Also, no memory deficits were observed in Gabra5-/- mice treated with etomidate or dexmedetomidine (Figure 5.9B). The total interaction time with the objects was similar in all groups (Figure 5.9C,D). Consistent with these behavioural results, postsynaptic and presynaptic plasticity at Schaffer collateral-CA1 synapses was not impaired in slices from

Gabra5-/- mice treated with etomidate (Figure 5.10A-D).

156

Figure 5.9 Reversal of memory impairment after a sedative dose of etomidate.

(A) Effects of etomidate (8 mg/kg i.p.) on memory performance in WT mice treated with L-

655,708 (L6, 0.5 mg/kg) 23.5 hours after etomidate and 30 minutes before training on the object recognition task (n = 6-11, one-way ANOVA, Dunnett’s post-test). (B) Memory performance in Gabra5-/- mice 24 hours, 72 hours and 1 week after etomidate etomidate (8 mg/kg i.p.) or dexmedetomidine (200 µg/kg) (n = 8-13, one-way ANOVA, Dunnett’s post- test at each time point). Data are shown as mean ± SEM. Unpaired, two-tailed Student’s t-test unless otherwise indicated. **P<0.01.

157

158

Figure 5.10 Plasticity is not impaired 24 hours after treatment of Gabra5-/- mice with a sedative dose of etomidate.

(A) Plasticity at Schaffer collateral-CA1 synapses 24 hours after etomidate (8 mg/kg, i.p.).

Insets: representative traces recorded before (1) and 60 min after (2) 20 Hz stimulation. Right panel shows summarized data for the last 5 min of recording (n = 7-9). (B) Post-tetanic depression and (C) short-term depression are not observed and (D) paired-pulse facilitation is similar between groups (n = 7-8, for all analyses of fPSPs an unpaired, two-tailed Student’s t- test was used). fPSP: field post-synaptic potential. Data are shown as mean ± SEM.

159

160

We next studied a higher, anesthetizing dose of etomidate. Mice were treated with

etomidate (20 mg/kg, i.p.), which is the ED100 dose for LORR (Cheng, et al. 2006). Object recognition memory was impaired for 1 week but recovered by 2 weeks after treatment

(Figure 5.11A). This dose did not alter total interaction time with the objects (Figure 5.11B).

The longer duration of memory impairment after the higher (20 mg/kg) versus the lower

(8 mg/kg) dose of etomidate, suggests a dose-dependent effect. Similar to the sedative dose, treatment with etomidate (20 mg/kg) reduced potentiation of fPSPs (Figure 5.12A), induced post-tetanic depression and short-term depression of fPSPs, but did not change paired-pulse facilitation 24 hours after treatment (Figure 5.12B-D). Furthermore, etomidate (20 mg/kg) increased the tonic current and cell-surface expression of α5 subunits in the hippocampus but did not alter the expression of α1 subunits 24 hours after treatment (Figure 5.13A-C).

161

Figure 5.11 An anesthetizing dose of etomidate impairs memory performance on the object

recognition task for at least 1 week.

(A) Memory performance in WT mice (n = 9-10) 24 hours, 72 hours, 1 week and 2 weeks

after etomidate anesthesia (20 mg/kg, i.p.). (B) Total interaction times with both objects are

not affected by etomidate treatment. An unpaired, two-tailed Student’s t-test was used to compare groups at each time point. Data are shown as mean ± SEM. *P < 0.05.

162

Figure 5.12 An anesthetizing dose of etomidate impairs plasticity for 24 hours in slices from WT mice.

(A) Plasticity at Schaffer collateral-CA1 synapses 24 hours after etomidate (20 mg/kg i.p.).

Insets: representative traces recorded before (1) and 60 min after (2) 20 Hz stimulation. Right

panel shows summarized data for the last 5 minutes of recording (n = 7-9). (B) Post-tetanic depression and (C) short-term depression are observed and (D) paired-pulse facilitation is similar between groups (n = 6-7, for all analyses of fPSPs an unpaired, two-tailed Student’s t- test was used). fPSP: field post-synaptic potential. Data are shown as mean ± SEM. *P <

0.05, ***P < 0.001.

163

Figure 5.13 An anesthetizing dose of etomidate increases a tonic inhibition and cell-surface expression of α5GABAA receptors 24 hours after treatment.

(A) Tonic current recorded from CA1 pyramidal neurons in WT brain slices from etomidate- treated (20 mg/kg, i.p.) or vehicle (Ctrl)-treated mice (n = 7-8). (B-C) Cell-surface and total expression of (B) α5 subunits (n = 4) and (C) α1 subunits (n = 4) in WT hippocampal slices

24 hours after treatment of mice with etomidate (20 mg/kg, i.p.). NKA, Na+/K+ ATPase.

MW, molecular weight (kDa). Data are shown as mean ± SEM. Unpaired, two-tailed

Student’s t-test unless otherwise indicated. *P<0.05.

164

165

Lastly, we investigated whether an inhaled anesthetic caused a similar increase in

tonic current and cell-surface expression of α5GABAA receptors. For these experiments, we

selected isoflurane because it is widely used in clinical practice, it acts on GABAA receptors

and is at undetectable or trace levels in the brain 24 hours after treatment (Saab, et al. 2010).

A low, sedative dose of isoflurane (0.7%, 20 min) (Sonner, et al. 2007), caused no change in

the cell-surface expression of α5 subunits or δ subunits 24 hours after treatment (Figure

5.14). In contrast, a higher, anesthetizing dose of isoflurane (Sonner, et al. 2000) (1.3%, 1 hour) increased the tonic current to 200% of control and the cell-surface expression of α5 to

134% of control, but the expression of δ subunits was unchanged (Figure 5.15).

166

Figure 5.14 A brief, sedative dose of isoflurane does not affect the expression of α5 and δ

GABAA receptor subunits.

Western blots of cell-surface and total expression of (A) α5 subunits and (B) δ subunits 24 h after a sedative dose of isoflurane (0.7%, 20 min; n = 3 mice). NKA, Na+/K+ ATPase. MW, molecular weight in kDa. An unpaired, two-tailed Student’s t-test was used to compare between groups. Data are shown as mean ± SEM.

167

Figure 5.15 An anesthetizing dose of isoflurane increases the tonic current and cell-surface expression of α5GABAA receptors 24 hours after treatment of mice.

(A) Tonic current recorded from CA1 pyramidal neurons 24 hours after treatment of WT mice with an anesthetizing dose of isoflurane (1.3%, 1 hour; Mann-Whitney U test, n = 9). (B and C) Western blots of cell-surface and total expression of (B) α5 subunits (n = 6) and (C) δ subunits (n = 5) 24 hours after treatment of mice with an anesthetizing dose of isoflurane

(1.3%, 1 hour). NKA, Na+/K+ ATPase. MW, molecular weight in kDa. An unpaired, two- tailed Student’s t-test was used to compare between groups. Data are shown as mean ± SEM.

*P < 0.05.

168

169

5.4 Discussion

Our findings present the first evidence that shows even a brief in vivo exposure to a

GABAergic general anesthetic triggers a sustained increase in tonic current and cell-surface

expression of α5GABAA receptors in the hippocampus of adult mice. This increase in

α5GABAA receptor activity in turn causes deficits in anterograde memory as treatment with

the α5GABAA receptor inhibitor L-655,708 reverses memory deficits. In contrast, the

anesthetic dexmedetomidine, which targets adrenergic receptors rather than GABAA

receptors, causes no change in the amplitude of the tonic current or memory performance.

These results refute the assumption that the activity of target receptors for GABAergic

anesthetics returns to baseline after the anesthetic has been eliminated, as the function of

α5GABAA receptors in the hippocampus was increased for days after a single exposure to

anesthetic.

Treatment with the higher, anesthetizing dose of etomidate produced longer-lasting

memory deficits than treatment with the low, sedative dose. This indicates a dose-dependent

effect and may imply that patients who receive a higher cumulative dose of anesthesia (due

to longer duration of anesthesia) are at greater risk for memory deficits.

Interestingly, at 1 week after etomidate (8 mg/kg), memory performance recovered,

yet the tonic current remained elevated. The sustained increase in tonic current may trigger

compensatory changes at the cellular, network and system levels that may contribute to the

recovery of memory performance, given that homeostatic plasticity has been widely

demonstrated in the hippocampus (Pozo et al. 2010; Vitureira et al. 2013). Indeed our results suggest such compensatory changes can occur. One week after etomidate, tonic

170

current and cell-surface expression remained elevated, whereas synaptic plasticity had

recovered and memory performance had already returned to baseline. Future studies are

required to identify potential compensatory changes in GABAergic or other neurotransmitter

systems that may be triggered after anesthesia (Borges da Silveira et al. 2013; Lynch et al.

2014; Nikiforuk et al. 2013; Stern et al. 2013). These compensatory changes may result in

the recovery of plasticity and memory performance at the 1 week timepoint while tonic

current and surface expression of α5GABAA receptors remain elevated.

Notably, the sedative dose of etomidate but not isoflurane increased cell-surface expression of α5GABAA receptors. The difference between anesthetics most likely arises

from the shorter duration of action of isoflurane compared with etomidate. Isoflurane

undergoes minimal metabolism (< 0.2%) and nearly 100% can be recovered in expired gas

(Holaday, et al. 1975), whereas etomidate is metabolized by plasma and liver esterases and

the metabolites are excreted through the kidneys with a half-life of approximately 4 hours

(Forman 2011).

These results, which demonstrate a critical role for α5GABAA receptors in memory

behaviours are consistent with previous studies. In particular, performance on certain

memory tasks such as trace fear, appetitive conditioning, as well as the Morris Water Maze

spatial memory test is enhanced in Gabra5-/- mice and in mice expressing a point mutation

that causes reduced expression of α5GABAA receptors (Collinson, et al. 2002; Crestani, et

al. 2002; Yee, et al. 2004). Conversely, pharmacologically increasing α5GABAA receptor

activity impairs memory performance and reduces synaptic plasticity (Cheng, et al. 2006).

171

In conclusion, to the best of my knowledge this the first evidence for a previously unrecognized long-term effect of general anesthetics on the function and cell-surface expression of α5GABAA receptors. Additional studies are required to determine whether the sustained increase in α5GABAA receptor activity after anesthesia is triggered by the initial direct allosteric actions of anesthetics on GABAA receptors or by other mechanisms and to identify the intracellular pathways that are involved in increasing surface expression in neurons.

172

Chapter 6. General Discussion 6.1 Summary

The objective of this thesis was to determine whether GABAA receptors are involved in the etiology of POCD. The results show that α5GABAA receptors in the hippocampus are necessary for postanesthetic memory deficits. Specifically, WT but not Gabra5-/- mice exhibited impaired recognition memory for days after exposure to the GABAergic anesthetics isoflurane, sevoflurane or etomidate. The non-GABAergic anesthetic dexmedetomidine did not cause memory impairments. In WT mice, memory deficits could be prevented by pharmacologically inhibiting α5GABAA receptors before anesthesia and reversed by inhibiting α5GABAA receptors 24 hours after anesthesia. Memory deficits were associated with increased tonic inhibition mediated by α5GABAA receptors and increased cell-surface expression of α5GABAA receptors. The increase in tonic inhibition in neurons in vitro was triggered by the anesthetic acting on astrocytes. The increased tonic inhibition and increased expression of functional α5GABAA receptors persisted for at least 1 week and recovered to baseline 2 weeks after treatment of mice with etomidate.

In recent years, studies have identified potential “off target” mechanisms of anesthetic neurotoxicity (Lin and Zuo 2011; Xie, et al. 2008). That is, mechanisms, such as apoptosis or accumulation of Alzheimer disease-related proteins, which are unrelated to the receptor targets through which anesthetics exert their desired neurodepressive effects. Prior to the publication of our data it was unknown whether anesthetics trigger sustained changes in inhibitory neurotransmission that cause lasting deficits in cognition.

173

The significance of these data are two-fold. First, these data present a potential

mechanism for postanesthetic-induced memory deficits and a plausible treatment strategy.

Second, the data refute the assumption that anesthetics only affect the activity of their target

receptors while present at binding sites on the receptor. This presents a major shift in our

understanding of how anesthetics affect brain physiology. My results show that the effects of

etomidate on α5GABAA receptors extend beyond the initial allosteric actions of the drug on

the receptor and trigger lasting changes in receptor expression that persist after the anesthetic

has been eliminated. This refutes current dogma, which assumes that the effects of

anesthetics on the brain are rapidly reversible.

6.2 Discussion

These results are in line with previous studies that demonstrated that α5GABAA receptors are important for learning and memory (Martin, et al. 2010). Specifically, previous studies have shown that pharmacologically inhibiting these receptors improves memory performance

(Martin, et al. 2010), whereas enhancing receptor activity with the positive allosteric modulator etomidate impairs memory performance (Cheng, et al. 2006).

The results suggest a role of α5GABAA receptors in the initiation or triggering of

postanesthetic memory deficits as deficits in recognition memory could be prevented by

inhibiting α5GABAA receptors with the drug L-655,708 before anesthesia. These results

support previous results from our lab that show L-655,708, when administered before

isoflurane anesthesia, can prevent deficits in contextual fear memory (Saab, et al. 2010).

Whether the initial allosteric interaction between the anesthetic and α5GABAA receptors

triggers memory deficits and an upregulation of α5GABAA receptor expression remains to be

determined.

174

In vitro data presented in Chapter 5 shows that the anesthetic etomidate acts on

astrocytes to trigger a sustained increase in tonic inhibition in hippocampal neurons. It is

tempting to speculate that allosteric modulation of α5GABAA receptors on astrocytes could initiate signaling cascades that lead to increased expression of α5GABAA receptors on

neurons.

6.2.1 The role of astrocytes

Astrocytes are part of the tripartite synapse and express neurotransmitter receptors, including

GABAA receptors, and are able to synthesize and release neurotransmitters (Araque et al.

2014). Activation of GABAA receptors on astrocytes depolarizes the astrocyte membrane and

thereby stimulates calcium influx (Angulo et al. 2008). This, in turn, can cause vesicular

release of neurotransmitters, peptides and other signaling molecules (Angulo, et al. 2008).

Since the treatment of cultured neurons with medium from etomidate-treated astrocytes was

sufficient to trigger a sustained increase in tonic current, it suggests that astrocytes release

soluble factors into the medium that increase tonic inhibition in neurons.

Astrocytes seem to preferentially affect tonic inhibition mediated by extrasynaptic receptors versus synaptic neuronal inhibition (Yoon et al. 2011). Astrocytic release of

GABA through Bestrophin 1 channels is a source of ambient GABA that activates extrasynaptic receptors (Lee, et al. 2010). In the cerebellum, the amount of GABA in

astrocytes is positively correlated with the amplitude of the tonic inhibitory current (Yoon, et al. 2011). It is possible, that depolarization of astrocytes by etomidate leads to increased

GABA release and enhanced tonic inhibition. However, this does not explain the increase in cell surface expression of α5GABAA receptors that is observed after etomidate treatment.

175

Astrocytes also release other neurotransmitters including glutamate, ATP, adenosine,

D-serine and proinflammatory cytokines such as TNF-α, which may bind to receptors on neurons to stimulate an increase in cell-surface expression of GABAA receptors (Perea et al.

2009). Previous studies from our laboratory show that the cytokine interleukin-1β can

increase α5GABAA receptor expression and α5GABAA receptor-mediated tonic inhibitory

current (Wang, et al. 2012). While activated microglia release similar proinflammatory

factors (Hanisch and Kettenmann 2007), etomidate did not trigger a sustained increase in

tonic current in microglia-neuron cocultures.

The interactions between astrocytes and neurons are just beginning to be characterized

and the precise effects of anesthetics on astrocytes are largely unknown. Partial hepatectomy

under isoflurane anesthesia in rats stimulates the release of the cytokines IL-6 and IL-1β and

increases glial cell activation, as evidenced by an increase in the expression of glial cell

markers GFAP and S100B (Cao, et al. 2010; Li, et al. 2013). Isoflurane treatment alone

reduces calcium signaling in astrocytes of awake, restrained mice (Thrane et al. 2012) and

reduces the expression of the cytoskeletal proteins α-tubulin and GFAP in cultured astrocytes

(Culley et al. 2013). In one study, plating isoflurane-treated astrocytes with neurons reduced

axonal growth in the neurons by approximately 30% (Ryu et al. 2014). Together these

studies suggest that isoflurane may exert some of its neurotoxic effects through its actions on

astrocytes.

6.2.2 Dexmedetomidine

Interestingly, the non-GABAergic sedative dexmedetomidine did not cause memory deficits

on the object recognition task or trigger an increase in tonic inhibition. In the immature brain,

dexmedetomidine has been shown to prevent neuronal apoptosis and memory deficits

176

(Sanders et al. 2010; Sanders, et al. 2009). Specifically, pretreatment of postnatal day 7 rats with dexmedetomidine prevents isoflurane (0.75%, 6 hours)-induced caspase-3 activation, neuronal apoptosis, activation of the kinases c-Jun N-terminal kinase and p38 mitogen-

activated protein kinase (p38-MAPK), as well as activation of nuclear factor kappa-light-

chain-enhancer of activated B cells (Sanders, et al. 2010; Sanders, et al. 2009).

Dexmedetomidine pretreatment also prevents deficits in trace fear memory induced by

isoflurane treatment (Sanders, et al. 2009). Interestingly, the authors showed that in young,

immature rodents, administration of the GABAA receptor antagonist gabazine did not prevent

isoflurane-induced apoptosis, indicating that the proapoptotic effects of isoflurane occurred

independently of GABAA receptor activation (Sanders, et al. 2009). This suggests that the there may be distinct mechanisms that cause memory deficits after anesthesia; GABAA

receptor-independent apoptosis, and the α5GABAA receptor-dependent mechanism that is

described in this thesis. Additionally, the mechanisms may differ between the immature brain

and the mature, adult brain.

In the immature brain, dexmedetomidine is thought to mediate its antiapoptotic effects

by acting on postsynaptic α2-adrenergic receptors to activate the extracellular signal-

regulated kinase (ERK)-Bcl-2 antiapoptotic pathway (Sanders, et al. 2010). Indeed, exposure

of young rats to isoflurane decreases phosphorylated ERK and Bcl-2 expression in the brain

and dexmedetomidine cotreatment can prevent this decrease (Sanders, et al. 2010).

Dexmedetomidine has been shown to be neuroprotective in other pathophysiological

conditions such as ischemia and traumatic brain injury (Hoffman et al. 1991; Schoeler et al.

2012). Some of the neuroprotective effects of dexmedetomidine in the brain may be due to

the drug’s effect on astrocytes (Zhang et al. 2014). In cultured astrocytes, dexmedetomidine

177

reduces astrocytic release of proinflammatory cytokines after lipopolysaccharide-induced

inflammation (Zhang, et al. 2014) and pretreatment with dexmedetomidine before

splenectomy surgery and anesthesia prevents increases in IL-1β, TNF-α, caspase-3 and

proapoptotic Bax protein in mice (Qian et al. 2014). These anti-inflammatory and anti-

apoptotic effects of dexmedetomidine may at least partially explain why the anesthetic does not cause memory deficits.

6.2.3 Pharmacological regulation of GABAA receptor expression

Previous studies show that GABAA receptor expression can be regulated by pharmacological

agents (Uusi-Oukari et al. 2010). Chronic treatment with GABAergic drugs alters GABAA

receptor expression and either upregulates or downregulates subunit expression, depending

on the drug, the subunit subtype, and the brain region that is being studied (Uusi-Oukari and

Korpi 2010). For example, chronic treatment with the benzodiazepine diazepam

downregulates the expression of α1 subunits and upregulates the expression of α4 subunits in

the cortex (Uusi-Oukari and Korpi 2010). Similarly, chronic treatment with either diazepam

or ethanol (84 days), increases the expression of mRNA for the α5 subunit in the

frontoparietal cortex and the hippocampus (Uusi-Oukari and Korpi 2010). Even a single, 24

hour exposure to the benzodiazepine can decrease cell-surface expression of

α2GABAA receptors and reduce the amplitude of mIPSCs, as the benzodiazepine treatment

increases lysosomal degradation of the receptors (Jacob et al. 2012). Anesthesia with

propofol and , or pentobarbital and midazolam, increases mRNA levels of the α4

subunit of the GABAA receptor (Sekine et al. 2006). However, the doses of anesthetics used

in this study were quite high and since multiple drugs were used it is unclear which one

contributed to the change in mRNA expression. Also, it is unknown whether the mRNA was

178

translated into protein and led to increased surface expression of functional GABAA

receptors (Sekine, et al. 2006).

6.2.4 Regulation of GABAA receptor expression

Since we observed an increase in surface but not total expression of α5GABAA receptors, I expect that the increased surface expression is due to either increased trafficking of

α5GABAA receptors to the cell surface or reduced endocytosis of α5GABAA receptors, rather than de novo synthesis of new receptors.

GABAARs can be rapidly translocated from internal reservoirs to the cell surface after

neuronal activity (Barnes 2000; Lüscher et al. 2011; Vithlani et al. 2011). Phosphorylation

of specific GABAA receptor subunits promotes either retention of the receptors in the cell

membrane or increased trafficking of these receptors to the cell surface (Kittler et al. 2002).

Protein kinase B and protein kinase C (PKC) increase surface expression of GABAA

receptors by phosphorylating S410 on the β2 subunit and thereby preventing endocytosis

(Kittler, et al. 2002; Terunuma et al. 2004). Protein kinase A (PKA) and PKC phosphorylate

S408 on the β3 subunit, and PKC also phosphorylates S443 on the α4 subunit, to both reduce endocytosis and promote increased cell-surface expression (Kittler et al. 2003). In contrast, phosphorylation of the β1 subunit by PKA or dephosphorylation of S408 on the β3 subunit by PP2A increases internalization of these receptors and decreases cell-surface expression

(Kittler, et al. 2002).

Two kinases have been implicated in the trafficking of α5GABAA receptors to the

cell-surface. One known mechanism by which this receptor trafficking occurs is dependent

2+ 2+ on an increase in the intracellular concentration of calcium ([Ca ]i) due to entry of Ca from

the extracellular space via L-type voltage-gated calcium channels (VGCCs) (Saliba et al.

179

2+ 2+ 2012; Vithlani, et al. 2011). An increase in [Ca ]i activates Ca /calmodulin-dependent

protein kinase II (CaMKII) (Wayman et al. 2008). Active, and phosphorylated CaMKII, then

phosphorylates the serine residue 383 on the β3 subunit (β3S383) of GABAARs that are assembled and ready to be trafficked to the cell membrane (Saliba, et al. 2012). The phosphorylation of β3S383 promotes accumulation of α5- and β3-containing extrasynaptic

GABAARs on the surface of hippocampal neurons (Saliba, et al. 2012).

The p38-MAPK pathway has also been implicated in increased α5GABAA receptor

expression (Wang, et al. 2012). Specifically, treatment of neurons or mice with the cytokine

IL-1β increased α5GABAA receptor-mediated tonic current and cell-surface expression of

α5GABAA receptors (Wang, et al. 2012). The increase in tonic current was prevented by

inhibitors of p38-MAPK (Wang, et al. 2012).

It is possible that the observed increase in surface expression after etomidate or

isoflurane treatment is due to increased trafficking of receptors to the cell-surface through

CaMKII-dependent, p38MAPK-dependent or through yet unidentified signaling pathways.

6.3 Future directions

I have presented results that demonstrate the effect of a brief, single exposure to anesthetic

impairs memory and increases α5GABAA receptor cell surface expression. However, the

mechanism by which etomidate acts on astrocytes to increase α5GABAA receptor cell surface

expression in neurons is not known and should be the subject of future studies.

Since our results demonstrate that the treatment of astrocytes with etomidate causes

the astrocytes to release a soluble factor that increases tonic inhibition in neurons, it is important to determine first, how anesthetics act on astrocytes to induce the release of the soluble factor and second, to identify the soluble factor. First, it should be conclusively

180

determined whether GABAA receptors on astrocytes are sensitive to positive allosteric

modulation by etomidate. Using whole-cell electrophysiology in cultured astrocytes, currents

evoked by GABA could be measured in the absence or presence of etomidate. To determine

whether α5GABAA receptors contribute to the GABA evoked current in astrocytes, L-

655,708 should be applied during electrophysiological recordings to determine whether a

portion of the GABA-evoked current can be blocked. Additionally, immunohistochemistry

should be performed to determine whether α5GABAA receptors are expressed in cultured

astrocytes. To determine whether activation of GABAA receptors on astrocytes is necessary

for the observed increase in tonic current on neurons, astrocytes may be co-treated with bicuculline and etomidate before application of the astrocyte-conditioned medium to neurons. Additionally, to determine whether activation of α5GABAA receptors in particular is sufficient to trigger the increase in tonic current in neurons, astrocytes can be co-treated with L-655,708 and etomidate before electrophysiological recordings. Also, astrocytes collected from Gabra5-/- mice may be treated with etomidate and the conditioned medium then applied to WT neurons to determine whether α5GABAA receptors on the astrocytes are

necessary for the increase in tonic current in neurons. Also, it should be determined whether

physiological activation of astrocytes with high concentrations of GABA is sufficient to

increase the tonic current in neurons or whether pharmacological enhancement of GABAA

receptor activity on astrocytes with etomidate is necessary to increase the tonic current in

neurons.

Astrocytes can become “activated”, in a process called reactive gliosis, under a variety

of pathological conditions such as hypoxia, ischemia, traumatic brain injury and

neurodegenerative diseases (Pekny et al. 2005). Astrocyte activation is characterized by

181

structural and functional changes in the astrocytes, specifically, a hypertrophy of cellular

processes and an upregulation of cytoskeletal proteins such as GFAP (Pekny and Nilsson

2005). Future experiments should determine whether etomidate treatment activates astrocytes

by determining whether GFAP expression is increased after etomidate treatment using

Western blot or immunohistochemistry. Astrocytes are the main sources of innate

inflammatory mediators in the brain and their activation promotes the release of

proinflammatory cytokines (Ransohoff et al. 2012). Although the soluble factor that

increases GABAA receptor activity has not been identified, the cytokine IL-1β can be

released from astrocytes (Ransohoff and Brown 2012) and is known to increase the activity

and cell-surface expression of α5GABAA receptors (Wang, et al. 2012). Activated astrocytes

can also release reactive oxygen species (Pekny and Nilsson 2005), which are also known to increase tonic inhibition (Penna et al. 2014).

Studying the astrocyte-conditioned medium could help identify the soluble factor.

Mass spectroscopy analysis of the conditioned medium could help determine whether

expression of specific proteins is higher in medium from etomidate-treated versus vehicle- treated astrocytes. This strategy could identify potential soluble factors. The soluble factor

could be a neurotransmitter such as the amino acid glutamate, or a protein that is vulnerable

to denaturation such as a cytokine or a neurotrophic factor (Perea, et al. 2009). Boiling the astrocyte-conditioned medium prior to applying it to neurons would help determine whether the soluble factor is a protein as amino acids, including neurotransmitters such as GABA, are stable at temperatures around 100ºC (Ito et al. 2006). If the soluble factor is identified to be, for example, IL-1β, co-treatment of astrocytes with etomidate and the specific antagonist for

IL-1 receptor should prevent a subsequent increase in tonic current and α5GABAA receptor

182

cell-surface expression. Since some anesthetics, such as isoflurane, are known to increase the

release of proinflammatory factors (Lin and Zuo 2011), it would be interesting to note

whether anesthetic-induced inflammation, for example through the release of cytokines from

neuroglia, plays a role in enhancing GABAA receptor function after anesthesia. Additional

studies should examine whether IL-1 receptor null mutant mice or WT mice treated with

TNF-α antibody are resistant to the etomidate-induced increase in α5GABAA receptor cell-

surface expression and memory deficits. These studies would provide further insight into

astrocyte-neuron communication; specifically they would identify proteins that are released

by astrocytes and act on neurons to increase the activity of GABAA receptors.

In neurons, CaMKII and p38-MAPK signaling pathways have been implicated in the regulation of α5GABAA receptor trafficking to the cell-surface (Saliba, et al. 2012; Wang, et

al. 2012). Western blot experiments should be used to determine whether the levels of

activated, phosphorylated forms of these kinases are increased in brains from etomidate-

treated mice. To prevent dephosphorylation of target sites during the preparation of brain

tissue, focused microwave irradiation could be used to preserve the in vivo protein

phosphorylation (O'Callaghan et al. 2004). The kinases that are involved in increased

receptor expression could also be isolated through experiments that employ selective kinase

or phosphatase inhibitors. For example, neuron-astrocyte cocultures could be co-treated with

etomidate and the p38-MAPK inhibitor SB203580. Additional experiments should address

the mechanism by which the receptor number is increased. For example, it should be

determined whether the increase in cell-surface expression is due to increased trafficking of

receptors to the cell-surface or a reduction in endocytosis and hence, enhanced retention of

receptors in the membrane. Endocytosis can be measured by labelling GABAA receptors with

183

a pH-sensitive fluorescent dye (Brady et al. 2014). As receptors are endocytosed, the

flurorescence is reduced due to a lower pH in the endocytic vesicle (Brady, et al. 2014).

These results would not only further elucidate the mechanism by which α5GABAA receptors

are upregulated after anesthesia but would also provide insight into the mechanisms that

underlie the trafficking of extrasynaptic α5GABAA receptors.

While the absence of memory deficits in Gabra5-/- mice suggests that α5GABAA

receptors are solely responsible for postanesthetic memory deficits, it is possible that changes

in other neurotransmitter systems also occur. Compensatory changes in other

neurotransmitter systems may occur as a result of an increase in α5GABAA receptor

expression, as homeostatic plasticity has been widely demonstrated in the hippocampus

(Pozo and Goda 2010). Future studies should examine whether exposure to anesthetic

changes the activity or expression of glutamatergic AMPA and NMDA receptors that are

necessary for memory formation and LTP (Bliss, et al. 2006). Interestingly, one research

group found that in a mouse model of ischemic stroke, motor recovery could be improved by

treatment with either the α5GABAA receptor-selective inverse agonist L-655,708 or a positive allosteric modulator of AMPA receptors, alluding to the possibility that both of these receptor populations could have a significant contribution to cognitive recovery (Clarkson, et al. 2010; Clarkson et al. 2011).

The mechanisms by which anesthetics cause memory deficits has been studied extensively in the early postnatal period in rodents and non-human primates (Jevtovic-

Todorovic et al. 2013). Some of the mechanisms that occur in the immature brain, such as apoptosis and impaired neurogenesis (Jevtovic-Todorovic, et al. 2013), also occur in the adult brain (Erasso, et al. 2013; Lin and Zuo 2011). It is unknown whether the mechanism

184

described in this thesis also occurs in the immature brain. Future studies should determine

whether α5GABAA receptor cell-surface expression and activity is enhanced in the brains of mice that are exposed to isoflurane or etomidate in the early postnatal period. Additionally, it should be determined whether Gabra5-/- mice, or WT mice treated with L-655,708, are immune to anesthetic-induced memory deficits and neurotoxicity in the early postnatal period. These results would be important as they would demonstrate whether this mechanism of memory impairment occurs irrespective of developmental stage and may even present a potential prevention strategy for anesthetic-induced neurotoxicity in the pediatric brain.

The implications of my work pose additional questions for future studies. While a single exposure to anesthetic causes an increase in α5GABAA receptor function, the effect of

multiple exposures to anesthesia is not known. If activation of α5GABAA receptors is

required to trigger memory deficits, does the increased expression of these receptors increase

vulnerability to memory deficits during a second anesthetic exposure? Indeed, rats exhibit

greater memory deficits after multiple exposures versus a single exposure to anesthesia.

(Murphy et al. 2013). Future studies should examine whether multiple exposures produce

greater impairment in memory performance or impaired performance for a longer period of

time than a single exposure to anesthesia. Additionally, does the increase in α5GABAA

receptor expression after anesthesia affect the dose required to cause amnesia at subsequent

treatments? Future studies should examine whether a lower dose of anesthetic will be

required to cause amnesia during a second anesthetic exposure when more target receptors are present on the cell membrane.

Lastly, I have shown that inverse agonists that are selective for α5GABAA receptors can reverse memory deficits after anesthesia. This may prove to be an effective treatment

185

strategy for postanesthetic memory deficits. L-655,708 and related inverse agonists are

particularly attractive as therapies because they lack the proconvulsant properties of

nonselective GABAA receptor antagonists (Atack, et al. 2006). This desired property is attributed to their low efficacy and preferential inhibition of α5GABAA receptors (Quirk, et al. 1996). Further, α5GABAA receptors have a low level of expression and are found only

within select regions of the brain (Pirker, et al. 2000). In humans, α5GABAA receptor- selective inverse agonists have been successfully shown to improve word recall after consumption of ethanol (Nutt, et al. 2007) and these agents are now being tried in clinical trials for the treatment of Alzheimer disease (Wallace, et al. 2011). L-655,708 can restore memory when it is administered after anesthesia and would not interfere with the desirable neurodepressive properties of the anesthetic during surgery. Future studies should also examine the effectiveness of L-655,708 at treating postanesthetic memory deficits in non- human primates and eventually, in human patients. Additionally, since tritiated versions of

α5GABAA receptor-selective inverse agonists have been used in humans (Mendez, et al.

2013), PET studies of human patients should examine whether increases in α5GABAA

receptor expression can be used as a biomarker of postoperative cognitive deficits.

6.4 Summary

In summary, in this thesis I have shown first, that α5GABAA receptors are necessary for

postanesthetic memory deficits as pharmacologically or genetically inhibiting α5GABAA

receptors during an acute exposure to an anesthetic prevents postanesthetic memory deficits.

Second, the cell-surface expression and the current generated by α5GABAA receptors are

enhanced after anesthesia. Lastly, α5GABAA receptors are potential treatment targets for

186

postanesthetic memory deficits as pharmacologically inhibiting α5GABAA receptors after anesthesia reverses deficits in learning and memory.

187

Appendix 1. Mutations in Gabra5 are associated with autism-related deficits

Introduction

Autism spectrum disorders (ASDs) are complex neurodevelopmental conditions

characterized by impaired social interactions, deficits in social communication and repetitive

behaviours. The overall prevalence of ASDs in the United States has been estimated at

approximately 1 in 88 children and up to 30% of cases can be explained by a genetic cause

(Anagnostou et al. 2014; Jiang et al. 2013). ASDs are associated with mutations in hundreds

of genes and genomic microarrays are now used clinically to identify chromosomal

abnormalities in patients (Anagnostou, et al. 2014; Jiang, et al. 2013).

In particular, mutations in chromosome 15q11e13 locus have been estimated to occur

in 1-4% of ASD patients and may contribute to ASD symptoms (Schroer et al. 1998). This

chromosomal region contains the genes Gabra5, Gabrb3 and Gabrg3 that encode the α5, β3

and γ3 subunits of the γ-aminobutyric acid type A (GABAA) receptor, respectively.

Deficiencies or deletions of this region cause neurodevelopmental disorders such as Prader-

Willi syndrome and (Nicholls et al. 2001), whereas maternally

inherited duplications are associated with the core symptoms of autism. Linkage and

association studies in human patients have demonstrated that mutations in 15q11e13 are

significantly associated with the incidence of autism (Cook, et al. 1998; McCauley, et al.

2004; Menold, et al. 2001; Shao, et al. 2003). Specifically, several single-nucleotide

polymorphisms (SNPs) in the Gabrb3 gene and one in the Gabra5 gene have been linked to

ASDs (Buxbaum et al. 2002; McCauley, et al. 2004).

188

Studies of post-mortem tissue from adult humans suggest that ASD patients have a reduced density of GABAA receptors and reduced expression of specific GABAA receptor subunits, including α1, α2, α3, α4, α5, β1, and β3 subunits (Blatt, et al. 2001; Fatemi, et al.

2010; Fatemi, et al. 2009; Mori, et al. 2012; Oblak, et al. 2009). Furthermore, a recent positron-emission tomography (PET) study of autistic subjects demonstrated reduced radioligand binding to α5GABAA receptors (Mendez, et al. 2013). In particular, radioligand binding was significantly reduced in the nucleus accumbens, a brain area that is involved in the perception of social reward, thereby suggesting that reduced expression of α5GABAA receptors in critical brain regions could cause some of the core features of autism (Mendez, et al. 2013).

α5GABAA receptors are composed of two α5 subunits, two β3 subunits and one γ2 subunit and are primarily anchored at extrasynaptic sites in the neuronal membrane by the protein radixin (Rdx gene) and at synaptic sites by the protein gephyrin (GPHN gene).

Activity of α5GABAA receptors modulates the intrinsic excitability of neurons and in the hippocampus α5GABAA receptors regulate normal learning and memory processes (Bonin, et al. 2007; Martin, et al. 2010). However, it is unknown whether α5GABAA receptors contribute to the behavioural deficits that occur in ASDs.

We hypothesized that an absence of Gabra5 expression would cause autism-like behavioural deficits. We tested wild-type (WT) and Gabra5 null mutant (Gabra5-/-) mice on a battery of behavioural assays to assess social behaviour, communication and repetitive behaviour. Additionally, we examined an exome database of human ASD patients for mutations in the Gabra5 gene and the Rdx gene, which encodes the main anchoring protein for α5GABAA receptors.

189

Methods

Experimental animals

All experimental procedures were approved by the Animal Care Committee of the University

of Toronto and performed in accordance with guidelines from the Canadian Council on

Animal Care. The generation, genotyping, and characterization of Gabra5–/– mice have

been previously described (Collinson, et al. 2002). For all behavioural tests, age-matched 3-

to 4-month-old male mice were used. Mice were housed in groups and were supplied with

food and water ad libitum. A circadian cycle of 14 hours light, 10 hours dark was maintained

in the housing room, and all experiments were performed during the light phase. To reduce variability in behavioural performance caused by acute stress, mice were handled for 5

minutes daily for 3 days before behavioural testing. Mice were randomly assigned to

treatment groups, and the experimenter was blinded to the genotype prior to video scoring.

Behavioural phenotyping assays

To test for abnormalities in social interaction the three-chamber social preference test and the

social proximity test were used. To test for communication deficits, ultrasonic vocalizations

were measured in isolated pups and the pup retrieval test was performed. To assay repetitive

behaviour, time spent self-grooming was measured. A summary of behavioural tests can be found in Figure 1 and these methods have been reviewed previously (Silverman et al. 2010).

190

Appendix Figure 1. 1 Behavioural phenotyping assays used to assess autism-like deficits in mice.

191

Three-chamber social preference test

The test was performed in a transparent, plastic, three-chamber apparatus (each chamber: 25 cm × 25 cm × 25 cm) with retractable doors between the center chamber and the left and right chambers. Each mouse completed two identical habituation sessions: one 24 hours prior to testing and the second immediately before testing. During habituation, the mouse was placed in the center chamber for 10 minutes with the doors to the other chambers closed.

Following this 10 minute period, the mouse was allowed to explore all 3 chambers for 15 minutes. Time spent in each chamber was measured to determine any innate preference.

Transparent, plastic cylinders (11 cm diameter, height 20 cm) with small perforations for scent transmission were placed in the the left and right chambers. The cylinders were designed to provide the mouse with visual and olfactory access without tactile access to the objects enclosed in the cylinders. During testing, a conspecific (stranger, non-cagemate of the test mouse) was placed in one transparent plastic cylinder and a novel object (toy car) was placed in the second transparent cylinder. The mouse used as the conspecific was habituated to the transparent, plastic cylinder during two habituation sessions: one 24 hours prior to testing and the second immediately before testing. The test mouse was placed in the center chamber and the doors to the left and right chambers were opened. The test mouse was allowed to explore all 3 chambers for 15 minutes. The test sessions were video-recorded and the videos were scored manually. The time spent in each chamber was measured.

Social proximity test

Mice were transported from their housing room to the experimental room 20 minutes prior to testing. During testing, two non-cagemate mice of the same genotype were simultaneously placed in a transparent, rectangular chamber (7cm × 14 cm × 20cm) for 10 minutes. After

192

each trial, mice were removed from the apparatus and placed in a transfer cage for 30

minutes to prevent any scent transfer to the home cage. Testing was conducted under dim red

light and test trials were video-recorded. The videos were scored manually for incidence of

the following behaviours in each mouse: Nose tip-to-nose tip contact (nose tip and/or vibrissae contact the nose tip and/or vibrissae of the other mouse); Nose-to-head (nose or vibrissae contacts the dorsal, lateral, or ventral surface of other mouse's head); Nose-to-

anogenital (nose or vibrissae contacts the base of the tail or anus of the other mouse); Crawl

over (forelimbs cross the midline of the dorsal surface of the other mouse); Crawl under

(head crosses the midline of the ventral surface of the other mouse); and upright (mouse

displays a reared posture oriented towards the other mouse with head and/or vibrissae

contact).

Ultrasonic vocalizations

Ultrasonic vocalizations (USVs) were recorded from pups on postnatal day 8. Each pup was

separated from the dam and immediately transported from the home cage to the procedure

room. The procedure room was illuminated by two 40 Watt red bulbs. Pups were placed in a

shallow plastic beaker (height 6 cm, diameter 4 cm) in a sound-attenuating chamber (40 cm ×

25 cm × 30 cm). The ultrasound detector D1000X (Pettersson Elektronik AB, Uppsala,

Sweden) was suspended 12 cm above the floor of the beaker. After a 5 second habituation

period, USV emissions were recorded for 4 minutes at a sampling frequency of 250 kHz.

Weight and temperature of the pups were recorded after the USV recording. USVs were

analyzed using Avisoft Bioacoustics software for the latency to emit the first call, total

number of calls, call length and total duration of calls. The spectrogram was scored

manually.

193

Pup retrieval

The latency for the dam to retrieve 5 pups to the nest was measured on pup postnatal days 6-

8. The dam and the breeding male were separated from the pups and placed in separate

holding cages for 3 minutes. Five pups were scattered in the home cage opposite to the nest.

The dam was returned to the nest and the time to retrieve the pups was measured with a stopwatch.

Self-grooming

Duration of self-grooming was measured in the home-cage for a 15 minute period. The testing room was lit with a dim, red light.

Exome data from human ASD probands

The Gabra5 locus on human and the Rdx locus on human chromosome 11 were examined for coding sequence variants in next-generation exome sequencing data from

306 Canadian ASD probands, as previously described (Lionel et al. 2013). After target enrichment utilizing the Agilent SureSelect V3 50 Mb Human All Exon kit (Agilent

Technologies, Santa Clara, CA, USA), paired-end sequencing was conducted on Life

Technologies SOLiD4 and SOLiD5500XL (Life Technologies) platforms. Protocols for sequencing and target capture followed specifications from the manufacturers. The generated paired end reads were mapped to the reference (UCSC hg19 build) using

BFAST. MarkDuplicates (Picard tools version 1.35; http:// icard.sourceforge.net) was used to remove duplicate paired end reads and the subsequent alignments were refined using local realignment in colorspace implemented in SRMAversion 0.1.15. GATK version1.1.28 was used for the calling of SNPs. Those novel non-synonymous variants detected by Sanger and

NGS exome sequencing, which were not previously reported in the SNP database (dbSNP)

194

build 137, were validated bidirectionally using Sanger re-sequencing in the case and in

samples from both parents, when available. Damaging missense single nucleotide variants

were defined as those that SIFT and PolyPhen-2 prediction software predicted to be damaging via the Variant Effect Predictor.

Statistical analysis

Results are presented as means ± standard error of the mean (SEM). An unpaired Student’s t

test was used to compare groups where appropriate. Data from the social approach test were

analyzed with a two-way ANOVA followed by Tukey’s post-hoc test. The Shapiro-Wilk test

was used to validate the assumption of normality. In cases where the assumption of normality

was not met for one or more groups, the Mann-Whitney U test was employed. GraphPad

Prism software, version 4.0 (GraphPad Software, San Diego, California, USA) was used. A P value less than 0.05 was considered statistically significant.

195

Results

Social behaviour is impaired in Gabra5-/- mice

To examine whether preference for social stimuli is affected in Gabra5-/- mice, mice were

tested on the three-chamber social approach test. Normal social preference is defined as more

time spent in the chamber containing the novel conspecific than in the chamber containing a

novel inanimate object. During habituation on the task, the time spent in each empty chamber

was measured to rule out side preference. WT and Gabra5-/- mice spent equal amounts of time in the empty, left and right chambers indicating no inherent side preference. However, both genotypes spent less time in the center chamber (Chamber: F(2,75) = 358.3, p < 0.0001;

Fig. 2B). During habituation, no differences between genotypes were observed (Genotype:

F(1,75) = 0.32, P > 0.05, Interaction F(2,75) = 1.46, P > 0.05; Figure 2A). During testing,

WT and Gabra5-/- mice spent more time in the chamber containing the conspecific than in the chamber containing the object (chamber: F(2,75) = 34.28, p < 0.001; Figure 2B). No

differences were detected between genotypes (genotype: F(1,75) = 0.05, p > 0.05,

interaction: F(2,75) = 0.13, p > 0.05; Figure 2B). These results suggest that WT and Gabra5-

/- mice both exhibit normal preference for a social stimulus.

Since impaired or reduced social contact is a key feature of ASDs, we measured the

number of reciprocal social contacts between a mouse and a conspecific in the social

proximity test. In comparison to WT mice, Gabra5-/- mice exhibited fewer social contacts

with a conspecific (t = 2.28, p = 0.031, n = 12 -14; Figure 2C). In particular, the number of

nose-to-nose (t = 4.68, p < 0.0001) and nose-to-head contacts (t=4.14, p = 0.0004) was

significantly reduced in Gabra5-/- mice (Figure 2D). The number of nose-to-anogenital contacts, crawling over the conspecific, crawling under the conspecific, or upright postures

196 was similar between genotypes (Figure 2B). Together these results indicate that Gabra5-/- mice exhibit impaired social behaviour as they engage in fewer social contacts with a conspecific.

197

Appendix Figure 1. 2

Gabra5-/- mice exhibit fewer social interactions than WT mice. (A) No inherent preference

for left or right chambers exhibited by WT or Gabra5-/- mice (two-way ANOVA genotype vs. chamber, P > 0.05). (B) Normal social preference in WT and Gabra5-/- mice (two-way

ANOVA genotype vs. chamber, P > 0.05). (C) Gabra5-/- mice exhibited fewer contacts with a conspecific than WT mice during the social proximity test. (D) During the social proximity test, Gabra5-/- mice exhibited fewer nose-to-nose and nose-to-head contacts than WT mice.

WT and Gabra5-/- mice exhibited a similar number of nose-to-anogenital, crawling over and under the conspecific or upright rearing. An unpaired, Student’s t-test was used to compare the number of each type of contact exhibited by each genotype. Data are presented as mean ±

SEM. * P < 0.05, *** P < 0.001.

198

199

Communication is impaired in Gabra5-/- mice

To measure deficits in communication, ultrasonic vocalizations (USVs) were measured.

USVs are emitted by mice in social situations. For example, infant mice (postnatal days 6-8) that are separated from the dam emit ultrasonic vocalizations (USVs) that serve to elicit pup retrieval by the dam. Significant differences in USVs were detected between Gabra5-/- and

WT pups. The latency to emit the first USV was greatly increased in Gabra5-/- pups in comparison to WT pups (t = 3.27, p = 0.0061, n = 7-8; Figure 3A). In addition, the total number of USV calls (t = 2.47, p = 0.0278, n = 7-8; Figure 3B) and the total calling time

(Mann-Whitney U = 10.0, p = 0.04, n = 7-8; Figure 3C) were reduced in Gabra5-/- pups, suggesting impaired communication in Gabra5-/- pups. The average length of an individual

USV was similar in both genotypes (t = 1.14, P = 0.274, n = 7-8; Figure 3D).

To determine whether reduced vocalization by Gabra5-/- pups had functional implications, the amount of time required for the dam to retrieve 5 pups to the nest was measured after a 3 minute separation. The latency to retrieve all 5 pups was significantly increased in Gabra5-/- versus WT mice (t = 2.487, p = 0.023, n = 9-10; Figure 3E).

200

Appendix Figure 1. 3

Impaired communication and pup retrieval in Gabra5-/- mice. (A) Postnatal day pups days 6-

8 were isolated from the dam and ultrasonic vocalizations were measured over 4 minutes.

Gabra5-/- mice exhibited a longer latency to emit the first vocalization. Gabra5-/- mice also exhibited (B) fewer calls and (C) reduced total calling time. (D) The length of individual calls was similar between genotypes. (E) Pups (P6-8) were separated from the dam for 3 minutes and scattered opposite the nest. The time required for the dam to retrieve 5 pups to the nest was measured. Gabra5-/- mice required more time to retrieve 5 pups to the nest. An unpaired Student’s t-test was used to compare between genotypes. Data are presented as mean ± SEM. * P < 0.05, ** P < 0.01.

201

Repetitive behaviour is increased in Gabra5-/- mice

Since repetitive behaviours are a core feature of ASD, we measured the amount of time each mouse spent engaged in self-grooming; a common spontaneous motor stereotypy in mice.

Gabra5-/- mice spent more time self-grooming than WT mice during the observation period

(WT 5.44 ± 1.08 s versus Gabra5-/- 10.78 ± 1.23 s, t=3.25, p = 0.005, n = 9), indicating an increase in this repetitive behaviour in Gabra5-/- mice.

Missense mutations in Gabra5 and Rdx occur in ASD patients

To determine whether mutations in Gabra5 exist in autistic probands, the Gabra5 locus on chromosome 15 was examined for coding sequence variations in next-generation exome sequencing data collected from 306 Canadian ASD probands. Since functional α5 subunit- containing GABAA receptors are anchored in the neuronal membrane by the protein radixin, the Rdx locus on chromosome 11 was also screened.

Missense coding variants were identified at 2 positions on the Gabra5 locus (Table 1).

Each of these variants occurred in single ASD cases in male probands. One of the variants in

Gabra5 was predicted to be functionally damaging by Polyphen and Sift prediction software

(Table 1). Four missense coding variants were identified on the Rdx locus (Table 1). One of the variants (hg 19 chr11:110,104,062) was present in 3 male probands, while the remaining variants were present in single ASD cases, 2 male and 1 female (Table 1). Two variants on the Rdx locus were predicted to be damaging by Polyphen and Sift prediction software (Table

1).

202

Gene Position Pro- Codon Substitution Inheritance Polyphen Sift bands change Prediction Prediction

Gabra5 chr15: 1 M Gtc/Atc V204I Maternal 0.005; 0.41; benign tolerated 27,182,361

Gabra5 chr15: 1 M gGg/gCg G113A Maternal 0.991; 0.04; probably damaging 27,128,545 damaging

Rdx chr11: 1 M aCc/aTc T516I Paternal 0.998; 0.02; probably damaging 110,104,002 damaging

Rdx chr11: 1 M Cct/Act P471T Maternal 0.585; 0.51; possibly tolerated 110,104,138 damaging

Rdx chr11: 1 F Gat/Cat D197H Heterozygous 0.999; 0; in both probably damaging 110,128,601 damaging

Rdx chr11: 3 M gCt/gTt A496V 1 Paternal 0.999; 0.52; probably tolerated 110,104,062 2 Maternal damaging

Table 1. Genetic details of ASD probands with missense mutations in Gabra5 and Rdx.

203

Discussion

The results of this study demonstrate that mutations in Gabra5 can cause autism-like deficits.

Specifically, a global deletion of Gabra5 causes fewer social contacts, reduced ultrasonic vocalizations, impaired pup retrieval and increased repetitive grooming behaviour in mice.

Additionally, missense mutations occur in the Gabra5 gene and in the related Rdx gene in

ASD probands. These mutations are predicted to be functionally damaging and may contribute to the etiology of ASD in these patients.

Gabra5-/- mice exhibited normal social preference, yet fewer social contacts in the

social proximity test. Since ASDs produce a spectrum of impairment that ranges from

moderate to severe, it is not surprising that some social behaviours were impaired while other

behaviours, such as preference for social stimuli, were preserved. The behavioural

impairments exhibited by Gabra5-/- mice are similar to deficits exhibited by other mouse

models of autism. Known mouse models such as the Tsc1 mutant mouse, the Shank1 null-

mutant mouse, and the BTBR mouse model of autism, exhibit fewer social contacts in the

social proximity test, fewer ultrasonic vocalizations, and an increased time spent grooming

(Defensor et al. 2011; Silverman et al. 2010; Tsai et al. 2012; Woehr et al. 2011; Young et

al. 2010). Similarly, to other mouse models of autism, previous studies show that mice with

reduced expression of α5GABAA receptors exhibit exaggerated startle responses and

specifically, impaired prepulse inhibition of these responses (Hauser, et al. 2005).

As predicted, Gabra5-/- mice displayed autism-like deficits. Previous studies support

the role of Gabra5 in ASDs and have shown reduced mRNA expression of α5 in post-

mortem brains of ASD patients and reduced radioligand binding to α5 in vivo (Fatemi, et al.

2010; Mendez, et al. 2013). Additionally, exonic deletions in the gene encoding the protein

204

gephyrin, which anchors α5GABAA receptors at synaptic sites on the neuronal membrane,

have been associated with an increased risk of autism, schizophrenia and seizures (Lionel, et

al. 2013).

Autism-like behavioural deficits in Gabra5-/- mice may be caused by a disruption in

the balance between excitation and inhibition in the brain. One emerging principle is that

ASD symptoms are caused by an increase in the ratio of excitation to inhibition (E/I balance)

in the brain (Rubenstein and Merzenich 2003). In fact, optogenetically elevating the E/I

balance results in autism-like deficits in social behaviour in mice (Yizhar et al. 2011). The

E/I balance can also be pathologically elevated by increasing excitatory neurotransmission or

alternatively, reducing inhibitory neurotransmission mediated by GABAA receptors. Indeed,

an absence of Gabra5 expression in Gabra5-/- mice reduces neuronal inhibition and

increases the excitability of individual neurons (Bonin, et al. 2007). This increase in excitability has been observed in other autism models and could cause ASD symptoms by increasing the E/I ratio in neural networks.

The results of this study pose implications for the diagnosis and treatment of ASD.

First, during diagnostic testing future genomic microarrays might screen for single nucleotide polymorphisms in Gabra5 and Rdx. Mutations in these genes may cause reduced Gabra5

expression that could also be detected by radioligand binding in a PET exam, as

demonstrated previously (Mendez, et al. 2013). Second, α5GABAA receptors could be

potential treatment targets for ASD symptoms. It is conceivable, that patients with reduced

Gabra5 expression could be treated pharmacologically with positive allosteric modulators

(PAMs) of α5GABAA receptors that would increase the activity of existing α5GABAA

receptors and thereby alleviate some ASD symptoms. Selective PAMs of α5GABAA

205 receptors could be used safely as they lack the sedative properties of non-selective PAMs of

GABAA receptors (Gill, et al. 2011). Future studies should determine the effectiveness of

α5GABAA receptor-selective PAMs in treating autism-like symptoms in other preclinical models of ASDs.

206

Appendix 2. Thesis-relevant work published by candidate

1. Zurek A.A., Yu J., Wang D-S., Haffey S.C., Bridgwater E.M., Penna A., Lecker I.,

Lei G., Salter E.W.R., & Orser B.A. (2014) Sustained increase in GABAA receptor

function impairs memory after anesthesia. The Journal of Clinical Investigation; 124

(12): 5437-5441.

2. Zurek A.A. & Orser B.A. (2014) New Vistas in Anesthesiology: Understanding

anesthesia-induced memory loss. Essentials of Pharmacology for Anesthesia, Pain

Medicine and Critical Care. Springer Science and Business Media. Eds. A.D. Kaye et

al., p.847-857.

3. Zurek A.A., Bridgwater E.M., & Orser B.A. (2012) Inhibition of α5GABAA

receptors restores recognition memory after general anesthesia. Anesthesia and

Analgesia; 114 (4): 845-55.

4. Saab B.J., Maclean A.J., Kanisek M., Zurek A.A., Martin L.J., Roder J.C., & Orser

B.A. (2010) Short-term memory impairment after isoflurane in mice is prevented by

the α5 γ-aminobutyric acid type A receptor inverse agonist L-655,708.

Anesthesiology; 113 (5): 1061-71.

5. Martin L.J., Zurek A.A., MacDonald J.K., Roder J.C., Jackson M.F., & Orser B.A.

(2010) α5GABAA receptor activity sets the threshold for long-term potentiation and

207 constrains hippocampus-dependent memory. Journal of Neuroscience; 30 (15): 5269-

82.

208

Appendix 3. Additional publications resulting from my doctoral studies.

1. Diaz M.R., Vollmer C.C., Zamudio-Bulcock P.A., Vollmer W., Blomquist S.,

Morton R.A., Everett J.C., Zurek A.A., Yu J., Orser B.A., & Valenzuela C.F.

(2014) Repeated intermittent alcohol exposure during the third trimester-equivalent

increases expression of the GABAA receptor δ subunit in cerebellar granule neurons

and delays motor development in rats. Neuropharmacology; 79: 262-74.

2. Bonin R.P., Zurek A.A., Yu J., Bayliss D.A., & Orser B.A. (2013)

Hyperpolarization-activated current (Ih) is reduced in hippocampal neurons from

Gabra5-/- mice. PLoS One; 8 (3): e58679.

3. Wang D.S., Zurek A.A., Lecker I., Yu J., Abramian A.M., Avramescu S., Davies

P.A., Moss S.J., Lu W-Y., & Orser B.A. (2012) Memory deficits induced by

inflammation are regulated by α5 subunit-containing GABAA receptors. Cell Reports;

2 (3): 488-96.

209

References

Aantaa R, Jaakola ML, Kallio A, Kanto J. Reduction of the minimum alveolar concentration of isoflurane by dexmedetomidine. Anesthesiology.1997;86(5):1055-1060.

Aantaa R, Kanto J, Scheinin M, Kallio A, Scheinin H. Dexmedetomidine, an α2- adrenoceptor agonist, reduces anesthetic requirements for patients undergoing minor gynecologic surgery. Anesthesiology.1990;73(2):230-235.

Abraham WC, Bear MF. Metaplasticity: the plasticity of synaptic plasticity. Trends Neurosci.1996;19(4):126-130.

Acevedo-Arcique CM, et al. Lidocaine, dexmedetomidine and their combination reduce isoflurane minimum alveolar concentration in dogs. PLoS One.2014;9(9).

Albuquerque C, Joseph DJ, Choudhury P, MacDermott AB. Dissection, plating, and maintenance of cortical astrocyte cultures. Cold Spring Harb Protoc.2009;(8):pdb.prot5273.

Ali AB, Thomson AM. Synaptic α5 subunit-containing GABAA receptors mediate IPSPs elicited by dendrite-preferring cells in rat neocortex. Cerebral Cortex.2008;18(6):1260-1271.

Alkire MT, Gorski LA. Relative amnesic potency of five inhalational anesthetics follows the Meyer-Overton rule. Anesthesiology.2004;101(2):417-429.

Allan SM, Tyrrell PJ, Rothwell NJ. Interleukin-1 and neuronal injury. Nature reviews.2005;5(8):629-640.

Anagnostou E, et al. Autism spectrum disorder: advances in evidence-based practice. CMAJ.2014;186(7):509-519.

Ancelin ML, et al. Exposure to anaesthetic agents, cognitive functioning and depressive symptomatology in the elderly. Br J Psychiatry.2001;178:360-366.

Angulo MC, Le Meur K, Kozlov AS, Charpak S, Audinat E. GABA, a forgotten gliotransmitter. Prog Neurobiol.2008;86(3):297-303.

Araque A, et al. Gliotransmitters travel in time and space. Neuron.2014;81(4):728-739.

Asai Y, et al. GABAA/Benzodiazepine receptor binding in patients with schizophrenia using [11C]Ro15-4513, a radioligand with relatively high affinity for α5 subunit. Schizophr Res.2008;99(1-3):333-340.

210

Atack JR, et al. L-655,708 enhances cognition in rats but is not proconvulsant at a dose selective for α5-containing GABAA receptors. Neuropharmacology.2006;51(6):1023- 1029.

Atack JR, et al. In vitro and in vivo properties of 3-tert-butyl-7-(5-methylisoxazol-3-yl)-2-(1- methyl-1H-1,2,4-triazol-5-ylmethoxy)- pyrazolo[1,5-d]-[1,2,4]triazine (MRK-016), a GABAA receptor α5 subtype-selective inverse agonist. J Pharmacol Exp Ther.2009;331(2):470-484.

Atack JR, et al. Rat pharmacokinetics and pharmacodynamics of a sustained release formulation of the GABAA α5-selective compound L-655,708. Drug Metab Dispos.2006.

Avidan MS, et al. Long-term cognitive decline in older subjects was not attributable to noncardiac surgery or major illness. Anesthesiology.2009;111(5):964-970.

Avital A, et al. Impaired interleukin-1 signaling is associated with deficits in hippocampal memory processes and neural plasticity. Hippocampus.2003;13(7):826-834.

Bai D, Pennefather PS, MacDonald JF, Orser BA. The general anesthetic propofol slows deactivation and desensitization of GABAA receptors. J Neurosci.1999;19(24):10635- 10646.

Bai DL, et al. Distinct functional and pharmacological properties of tonic and quantal inhibitory postsynaptic currents mediated by γ-aminobutyric acid A receptors in hippocampal neurons. Mol Pharmacol.2001;59(4):814-824.

Ballard TM, et al. RO4938581, a novel cognitive enhancer acting at GABAA α5 subunit- containing receptors. Psychopharmacology.2009;202(1-3):207-223.

Barberis A, Petrini EM, Cherubini E. Presynaptic source of quantal size variability at GABAergic synapses in rat hippocampal neurons in culture. Eur J Neurosci.2004;20(7):1803-1810.

Barberis A, Petrini EM, Mozrzymas JW. Impact of synaptic neurotransmitter concentration time course on the kinetics and pharmacological modulation of inhibitory synaptic currents. Front Cell Neurosci.2011;5(6).

Barnes EM, Jr. Intracellular trafficking of GABA(A) receptors. Life Sci.2000;66(12):1063- 1070.

Bedford FK, et al. GABAA receptor cell surface number and subunit stability are regulated by the ubiquitin-like protein Plic-1. Nat Neurosci.2001;4(9):908-916.

Bedford PD. Adverse cerebral effects of anaesthesia on old people. Lancet.1955;2(AUG6):259-263.

211

Beenhakker MP, Huguenard JR. Astrocytes as gatekeepers of GABAB receptor function. J Neurosci.2010;30(45):15262-15276.

Bekker A, et al. Does mild cognitive impairment increase the risk of developing postoperative cognitive dysfunction? Am J Surg.2010;199(6):782-788.

Belelli D, Callachan H, Hill-Venning C, Peters JA, Lambert JJ. Interaction of positive allosteric modulators with human and Drosophila recombinant GABA receptors expressed in Xenopus laevis oocytes. Br J Pharmacol.1996;118(3):563-576.

Belelli D, Lambert JJ, Peters JA, Wafford K, Whiting PJ. The interaction of the general anesthetic etomidate with the γ-aminobutyric acid type A receptor is influenced by a single amino acid. Proc Natl Acad Sci.1997;94(20):11031-11036.

Belleville JP, Ward DS, Bloor BC, Maze M. Effects of intravenous dexmedetomidine in humans. I. Sedation, ventilation, and metabolic rate. Anesthesiology.1992;77(6):1125- 1133.

Ben-Ari Y, Khalilov I, Kahle KT, Cherubini E. The GABA excitatory/inhibitory shift in brain maturation and neurological disorders. Neuroscientist.2012;18(5):467-486.

Bencsits E, Ebert V, Tretter V, Sieghart W. A significant part of native γ-aminobutyric acid A receptors containing α4 subunits do not contain γ or δ subunits. J Biol Chem.1999;274(28):19613-19616.

Benkwitz C, et al. Determination of the EC50 amnesic concentration of etomidate and its diffusion profile in brain tissue: Implications for in vitro studies. Anesthesiology.2007;106(1):114-123.

Bertaina-Anglade V, Enjuanes E, Morillon D, Drieu la Rochelle C. The object recognition task in rats and mice: A simple and rapid model in safety pharmacology to detect amnesic properties of a new chemical entity. J Pharmacol Toxicol Methods.2006;54(2):99-105.

Bevins RA, Besheer J. Object recognition in rats and mice: a one-trial non-matching-to- sample learning task to study 'recognition memory'. Nat Protoc.2006;1(3):1306-1311.

Bianchi MT, Haas KF, MacDonald RL. α1 and α6 subunits specify distinct desensitization, deactivation and modulation of GABAA receptors containing the δ subunit. Neuropharmacology.2002;43(4):492-502.

Bird CM, Burgess N. The hippocampus and memory: insights from spatial processing. Nat Rev Neurosci.2008;9(3):182-194.

Blatt GJ, et al. Density and distribution of hippocampal neurotransmitter receptors in autism: an autoradiographic study. J Autism Dev Disord.2001;31(6):537-543.

212

Bliss T, Collingridge GL, Morris R. 2006. Synaptic plasticity in the hippocampus. In The Hippocampus Book. New York: Oxford University Press.

Bliss TV, Lomo T. Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path. J Physiol.1973;232(2):331-356.

Bliss TVP, Collingridge GL. A synaptic model of memory - long-term potentiation in the hippocampus. Nature.1993;361(6407):31-39.

Bohlhalter S, Weinmann O, Mohler H, Fritschy JM. Laminar compartmentalization of GABAA-receptor subtypes in the spinal cord: an immunohistochemical study. J Neurosci.1996;16(1):283-297.

Bonin RP, Martin LJ, MacDonald JF, Orser BA. α5GABAA receptors regulate the intrinsic excitability of mouse hippocampal pyramidal neurons. J Neurophysiol.2007;98(4):2244-2254.

Borges da Silveira CK, Furini CRG, Benetti F, Monteiro SdC, Izquierdo I. The role of histamine receptors in the consolidation of object recognition memory. Neurobiol Learn Mem.2013;103:64-71.

Bormann J. Patch-clamp analysis of GABA- and glycine-gated chloride channels. Adv Biochem Psychopharmacol.1988;45:47-60:47-60.

Boue-Grabot E, et al. Expression of GABA receptor rho subunits in rat brain. J Neurochem.1998;70(3):899-907.

Brady ML, Moon CE, Jacob TC. Using an α-bungarotoxin binding site tag to study GABAA receptor membrane localization and trafficking. 2014;(85):e51365.

Braudeau J, et al. Specific targeting of the GABAA receptor α5 subtype by a selective inverse agonist restores cognitive deficits in Down syndrome mice. J Psychopharmacol.2011;25(8):1030-1042.

Bright DP, Smart TG. Methods for recording and measuring tonic GABAA receptor- mediated inhibition. Front Neural Circuits.2013;7.

Buchanan JB, Sparkman NL, Chen J, Johnson RW. Cognitive and neuroinflammatory consequences of mild repeated stress are exacerbated in aged mice. Psychoneuroendocrinology.2008;33(6):755-765.

Buxbaum JD, et al. Association between a GABRB3 polymorphism and autism. Mol Psychiatry.2002;7(3):311-316.

Buzsaki G. Theta rhythm of navigation: link between path integration and landmark navigation, episodic and semantic memory. Hippocampus.2005;15(7):827-840.

213

Campagna JA, Miller KW, Forman SA. Mechanisms of actions of inhaled anesthetics. N Engl J Med.2003;348(21):2110-2124.

Cao XZ, et al. Postoperative cognitive deficits and neuroinflammation in the hippocampus triggered by surgical trauma are exacerbated in aged rats. Prog Neuropsychopharmacol Biol Psychiatry.2010;34(8):1426-1432.

Caraiscos VB, et al. Tonic inhibition in mouse hippocampal CA1 pyramidal neurons is mediated by α5 subunit-containing γ-aminobutyric acid type A receptors. Proc Natl Acad Sci.2004;101(10):3662-3667.

Caraiscos VB, et al. Selective enhancement of tonic GABAergic inhibition in murine hippocampal neurons by low concentrations of the volatile anesthetic isoflurane. J Neurosci.2004;24(39):8454-8458.

Caza N, Taha R, Qi Y, Blaise G. The effects of surgery and anesthesia on memory and cognition. Prog Brain Res.2008;169:409-422.

Chambers MS, et al. Identification of a novel, selective GABAA α5 receptor inverse agonist which enhances cognition. J Med Chem.2003;46(11):2227-2240.

Chandra D, et al. GABAA receptor α4 subunits mediate extrasynaptic inhibition in thalamus and dentate gyrus and the action of . Proc Natl Acad Sci.2006;103(41):15230-15235.

Chaudhry FA, et al. The vesicular GABA transporter, VGAT, localizes to synaptic vesicles in sets of glycinergic as well as GABAergic neurons. J Neurosci.1998;18(23):9733- 9750.

Chen ZW, Olsen RW. GABAA receptor associated proteins: a key factor regulating GABAA receptor function. J Neurochem.2007;100(2):279-294.

Cheng VY, et al. α5GABAA receptors mediate the amnestic but not sedative-hypnotic effects of the general anesthetic etomidate. J Neurosci.2006;26(14):3713-3720.

Chi H, et al. Postoperative pain impairs subsequent performance on a spatial memory task via effects on N-methyl-D-aspartate receptor in aged rats. Life Sci.2013;93(25-26):986- 993.

Chien LN, Lin HC, Shao YH, Chiou ST, Chiou HY. Risk of autism associated with general anesthesia during cesarean delivery: a population-based birth-cohort analysis. J Autism Dev Disord.2014;26:26.

Christie SB, Miralles CP, de Blas AL. GABAergic innervation organizes synaptic and extrasynaptic GABAA receptor clustering in cultured hippocampal neurons. J Neurosci.2002;22(3):684-697.

214

Cibelli M, et al. Role of interleukin-1β in postoperative cognitive dysfunction. Ann Neurol.2010;68(3):360-368.

Clarkson AN, Huang BS, Macisaac SE, Mody I, Carmichael ST. Reducing excessive GABA- mediated tonic inhibition promotes functional recovery after stroke. Nature.2010;468(7321):305-309.

Clarkson AN, et al. AMPA receptor-induced local brain-derived neurotrophic factor signaling mediates motor recovery after stroke. J Neurosci.2011;31(10):3766-3775.

Collingridge GL, Bliss TV. Memories of NMDA receptors and LTP. Trends Neurosci.1995;18(2):54-56.

Collinson N, Atack JR, Laughton P, Dawson GR, Stephens DN. An inverse agonist selective for α5 subunit-containing GABAA receptors improves encoding and recall but not consolidation in the Morris water maze. Psychopharmacology.2006;188(4):619-628.

Collinson N, et al. Enhanced learning and memory and altered GABAergic synaptic transmission in mice lacking the α5 subunit of the GABAA receptor. J Neurosci.2002;22(13):5572-5580.

Cook EH, et al. Linkage-disequilibrium mapping of autistic disorder, with 15q11-13 markers. Am J Hum Genet.1998;62(5):1077-1083.

Cooper LN, Bear MF. The BCM theory of synapse modification at 30: interaction of theory with experiment. Nat Rev Neurosci.2012;13(11):798-810.

Correa-Sales C, Rabin BC, Maze M. A hypnotic response to dexmedetomidine, an α2 agonist, is mediated in the locus coeruleus in rats. Anesthesiology.1992;76(6):948- 952.

Crestani F, et al. Trace fear conditioning involves hippocampal α5GABAA receptors. Proc Natl Acad Sci.2002;99(13):8980-8985.

Crosby C, Culley DJ, Baxter MG, Yukhananov R, Crosby G. Spatial memory performance 2 weeks after general anesthesia in adult rats. Anesth Analg.2005;101(5):1389-1392.

Culley DJ, Baxter M, Yukhananov R, Crosby G. The memory effects of general anesthesia persist for weeks in young and aged rats. Anesth Analg.2003;96(4):1004-1009.

Culley DJ, Baxter MG, Crosby CA, Yukhananov R, Crosby G. Impaired acquisition of spatial memory 2 weeks after isoflurane and isoflurane-nitrous oxide anesthesia in aged rats. Anesth Analg.2004;99(5):1393-1397.

Culley DJ, Baxter MG, Yukhananov R, Crosby G. Long-term impairment of acquisition of a spatial memory task following isoflurane-nitrous oxide anesthesia in rats. Anesthesiology.2004;100(2):309-314.

215

Culley DJ, et al. Isoflurane affects the cytoskeleton but not survival, proliferation, or synaptogenic properties of rat astrocytes in vitro. Br J Anaesth.2013;110(1).

Dantzer R, O'Connor JC, Freund GG, Johnson RW, Kelley KW. From inflammation to sickness and depression: when the immune system subjugates the brain. Nat Rev Neurosci.2008;9(1):46-56.

Dawson GR, et al. An inverse agonist selective for α5 subunit-containing GABAA receptors enhances cognition. J Pharmacol Exp Ther.2006;316(3):1335-1345.

Defensor EB, et al. A novel social proximity test suggests patterns of social avoidance and gaze aversion-like behavior in BTBR T plus tf/J mice. Behav Brain Res.2011;217(2):302-308.

Deierborg T. Preparation of primary microglia cultures from postnatal mouse and rat brains. Methods Mol Biol.2013;1041:25-31.

Deng M, et al. Brain regional vulnerability to anaesthesia-induced neuroapoptosis shifts with age at exposure and extends into adulthood for some regions. Br J Anaesth.2014;113(3):443-451.

DiazGranados N, et al. Rapid resolution of suicidal ideation after a single infusion of an N- Methyl-D-Aspartate antagonist in patients with treatment-resistant major depressive disorder. J Clin Psychiatry.2010;71(12):1605-1611.

Doze V, Chen BX, Li Z, Maze M. Pharmacologic characterization of the receptor mediating the hypnotic action of dexmedetomidine. Acta Vet Scand Suppl.1989;85:61-64.

Duman RS, Li N, Liu R-J, Duric V, Aghajanian G. Signaling pathways underlying the rapid antidepressant actions of ketamine. Neuropharmacology.2012;62(1):35-41.

Dyck JB, Maze M, Haack C, Vuorilehto L, Shafer SL. The pharmacokinetics and hemodynamic effects of intravenous and intramuscular dexmedetomidine hydrochloride in adult human volunteers. Anesthesiology.1993;78(5):813-820.

Eichenbaum H, Yonelinas AP, Ranganath C. The medial temporal lobe and recognition memory. Annu Rev Neurosci.2007;30:123-152.

Ennaceur A. One-trial object recognition in rats and mice: methodological and theoretical issues. Behav Brain Res.2010;215(2):244-254.

Ennaceur A, Delacour J. A new one-trial test for neurobiological studies of memory in rats. 1: Behavioral data. Behav Brain Res.1988;31(1):47-59.

Erasso DM, Camporesi EM, Mangar D, Saporta S. Effects of isoflurane or propofol on postnatal hippocampal neurogenesis in young and aged rats. Brain Res.2013;12:1-12.

216

Evered LA, et al. Plasma Amyloid beta(42) and Amyloid beta(40) Levels Are Associated With Early Cognitive Dysfunction After Cardiac Surgery. Annals of Thoracic Surgery.2009;88(5):1426-1432.

Farrant M, Nusser Z. Variations on an inhibitory theme: phasic and tonic activation of GABAA receptors. Nat Rev Neurosci.2005;6(3):215-229.

Fatemi SH, et al. mRNA and protein levels for GABAA α4, α5, β1 and GABAB R1 receptors are altered in brains from subjects with autism. J Autism Dev Disord.2010;40(6):743- 750.

Fatemi SH, Reutiman TJ, Folsom TD, Thuras PD. GABAA receptor downregulation in brains of subjects with autism. J Autism Dev Disord.2009;39(2):223-230.

Fleischmann RM, et al. Safety of extended treatment with anakinra in patients with rheumatoid arthritis. Ann Rheum Dis.2006;65(8):1006-1012.

Forman SA. Clinical and molecular pharmacology of etomidate. Anesthesiology.2011;114(3):695-707.

Frank MG, Watkins LR, Maier SF. Stress- and glucocorticoid-induced priming of neuroinflammatory responses: potential mechanisms of stress-induced vulnerability to drugs of abuse. Brain Behav Immun.2011;25(1):21.

Fritschy JM, et al. Five subtypes of type A γ-aminobutyric acid receptors identified in neurons by double and triple immunofluorescence staining with subunit-specific antibodies. Proc Natl Acad Sci U S A.1992;89(15):6726-6730.

Fritschy JM, Johnson DK, Mohler H, Rudolph U. Independent assembly and subcellular targeting of GABAA-receptor subtypes demonstrated in mouse hippocampal and olfactory neurons in vivo. Neurosci Lett.1998;249(2-3):99-102.

Fritschy JM, Kiener T, Bouilleret V, Loup F. GABAergic neurons and GABAA-receptors in temporal lobe epilepsy. Neurochem Int.1999;34(5):435-445.

Gerdjikov TV, et al. Hippocampal α5 subunit-containing GABAA receptors are involved in the development of the latent inhibition effect. Neurobiol Learn Mem.2008;89(2):87- 94.

Gigante PR, et al. Polymorphisms in complement component 3 (C3F) and complement factor H (Y402H) increase the risk of postoperative neurocognitive dysfunction following carotid endarterectomy. J Neurol Neurosurg Psychiatry.2011;82(3):247-253.

Gill KM, Lodge DJ, Cook JM, Aras S, Grace AA. A novel α5GABAAR-positive allosteric modulator reverses hyperactivation of the dopamine system in the MAM model of schizophrenia. Neuropsychopharmacology.2011;36(9):1903-1911.

217

Glykys J, Mody I. The main source of ambient GABA responsible for tonic inhibition in the mouse hippocampus. J Physiol.2007;582(Pt 3):1163-1178.

Glykys J, et al. A new naturally occurring GABAA receptor subunit partnership with high sensitivity to ethanol. Nat Neurosci.2007;10(1):40-48.

Gogenur I, et al. Impact of sleep and circadian disturbances in urinary 6-sulphatoxymelatonin levels, on cognitive function after major surgery. J Pineal Res.2007;43(2):179-184.

Gong M, et al. Parecoxib mitigates spatial memory impairment induced by sevoflurane anesthesia in aged rats. Acta Anaesthesiol Scand.2012;56(5):601-607.

- Grover LM, Lambert NA, Schwartzkroin PA, Teyler TJ. Role of HCO3 ions in depolarizing GABAA receptor-mediated responses in pyramidal cells of rat hippocampus. J Neurophysiol.1993;69(5):1541-1555.

Haas KF, Macdonald RL. GABAA receptor subunit γ2 and δ subtypes confer unique kinetic properties on recombinant GABAA receptor currents in mouse fibroblasts. J Physiol.1999;514(1):27-45.

Hanchar HJ, Dodson PD, Olsen RW, Otis TS, Wallner M. Alcohol-induced motor impairment caused by increased extrasynaptic GABAA receptor activity. Nat Neurosci.2005;8(3):339-345.

Hanisch UK, Kettenmann H. Microglia: active sensor and versatile effector cells in the normal and pathologic brain. Nat Neurosci.2007;10(11):1387-1394.

Hauser J, et al. Hippocampal α5 subunit-containing GABAA receptors modulate the expression of prepulse inhibition. Mol Psychiatry.2005;10(2):201-207.

Hemmings HC, Jr., et al. Emerging molecular mechanisms of general anesthetic action. Trends Pharmacol Sci.2005;26(10):503-510.

Hendrickx JFA, Dishart MK, De Wolf AM. Isoflurane and desflurane uptake during liver resection and transplantation. Anesth Analg.2003;96(2):356-362.

Heyer EJ, et al. Severe pain confounds neuropsychological test performance. J Clin Exp Neuropsychol.2000;22(5):633-639.

Hietbrink F, Koenderman L, Rijkers G, Leenen L. Trauma: the role of the innate immune system. World J Emerg Surg.2006;1:15.

Hille B. 2001. Ion channels of excitable membranes. Edition 3. Sunderland, MA: Sinauer Associates Inc. 1-814 pp.

Hofacer RD, et al. Cell age-specific vulnerability of neurons to anesthetic toxicity. Ann Neurol.2013;73(6):695-704.

218

Hoffman WE, Kochs E, Werner C, Thomas C, Albrecht RF. Dexmedetomidine improves neurologic outcome from incomplete ischemia in the rat. Reversal by the α2- adrenergic antagonist atipamezole. Anesthesiology.1991;75(2):328-332.

Holaday DA, Fiserova-Bergerova V, Latto IP, Zumbiel MA. Resistance of isoflurane to biotransformation in man. Anesthesiology.1975;43(3):325-332.

Hu N, et al. Involvement of the blood-brain barrier opening in cognitive decline in aged rats following orthopedic surgery and high concentration of sevoflurane inhalation. Brain Res.2014;10:13-24.

Hudetz JA, Gandhi SD, Iqbal Z, Patterson KM, Pagel PS. Elevated postoperative inflammatory biomarkers are associated with short- and medium-term cognitive dysfunction after coronary artery surgery. J Anesth.2011;25(1):1-9.

Hudetz JA, et al. Postoperative cognitive dysfunction in older patients with a history of alcohol abuse. Anesthesiology.2007;106(3):423-430.

Hudetz JA, Patterson KM, Amole O, Riley AV, Pagel PS. Postoperative cognitive dysfunction after noncardiac surgery: effects of metabolic syndrome. Journal of Anesthesia.2011;25(3):337-344.

Hudetz JA, Patterson KM, Iqbal Z, Gandhi SD, Pagel PS. Metabolic syndrome exacerbates short-term postoperative cognitive dysfunction in patients undergoing cardiac surgery: results of a pilot study. J Cardiothorac Vasc Anesth.2011;25(2):282-287.

Isomura Y, et al. Synaptically activated Cl- accumulation responsible for depolarizing GABAergic responses in mature hippocampal neurons. J Neurophysiol.2003;90(4):2752-2756.

Ito M, Gupta LP, Masuda H, Kawahata H. Thermal stability of amino acids in seafloor sediment in aqueous solution at high temperature. Org Geochem.2006;37(2):177-188.

Jacob TC, et al. Benzodiazepine treatment induces subtype-specific changes in GABAA receptor trafficking and decreases synaptic inhibition. Proc Natl Acad Sci.2012;109(45):18595-18600.

Jenkins A, et al. Evidence for a common binding cavity for three general anesthetics within the GABAA receptor. J Neurosci.2001;21(6).

Jevtovic-Todorovic V, et al. Anaesthetic neurotoxicity and neuroplasticity: an expert group report and statement based on the BJA Salzburg Seminar. Br J Anaesth.2013;111(2):143-151.

Jiang Y-h, et al. Detection of clinically relevant genetic variants in autism spectrum disorder by whole-genome sequencing. Am J Hum Genet.2013;93(2):249-263.

219

Jin H, et al. Demonstration of functional coupling between γ-aminobutyric acid (GABA) synthesis and vesicular GABA transport into synaptic vesicles. Proc Natl Acad Sci.2003;100(7):4293-4298.

Johnston GA. GABAA receptor pharmacology. Pharmacol.Ther.1996;69(3):173-198.

Jorm CM, Stamford JA. Actions of the hypnotic anaesthetic, dexmedetomidine, on noradrenaline release and cell firing in rat locus coeruleus slices. Br J Anaesth.1993;71(3):447-449.

Jurd R, et al. General anesthetic actions in vivo strongly attenuated by a point mutation in the GABAA receptor β3 subunit. Faseb J.2003;17(2):250-252.

Kahle KT, et al. Roles of the cation-chloride cotransporters in neurological disease. Nat Clin Pract Neurol.2008;4(9):490-503.

Kalso EA, Poyhia R, Rosenberg PH. Spinal antinociception by dexmedetomidine, a highly selective alpha 2-adrenergic agonist. Pharmacol Toxicol.1991;68(2):140-143.

Karim N, et al. Potency of GABA at human recombinant GABAA receptors expressed in Xenopus oocytes: a mini review. Amino Acids.2013;44(4):1139-1149.

Kasugai Y, et al. Quantitative localisation of synaptic and extrasynaptic GABAA receptor subunits on hippocampal pyramidal cells by freeze-fracture replica immunolabelling. Eur J Neurosci.2010;32(11):1868-1888.

Kaufman DL, Houser CR, Tobin AJ. 2 forms of the γ-aminobutyric-acid synthetic enzyme glutamate-decarboxylase have distinct intraneuronal distributions and cofactor interactions. J Neurochem.1991;56(2):720-723.

Keller CA, et al. The γ2 subunit of GABAA receptors is a substrate for palmitoylation by GODZ. J Neurosci.2004;24(26):5881-5891.

Kindler CH, Eilers H, Donohoe P, Ozer S, Bickler PE. Volatile anesthetics increase intracellular calcium in cerebrocortical and hippocampal neurons. Anesthesiology.1999;90(4):1137-1145.

Kittler JT, et al. Regulation of synaptic inhibition by phospho-dependent binding of the AP2 complex to a YECL motif in the GABAA receptor γ2 subunit. Proc Natl Acad Sci.2008;105(9):3616-3621.

Kittler JT, McAinsh K, Moss SJ. Mechanisms of GABAA receptor assembly and trafficking - Implications for the modulation of inhibitory neurotransmission. Mol Neurobiol.2002;26(2-3):251-268.

Kittler JT, Moss SJ. Modulation of GABAA receptor activity by phosphorylation and receptor trafficking: implications for the efficacy of synaptic inhibition. Curr Opin Neurobiol.2003;13(3):341-347.

220

Kittler JT, et al. Huntingtin-associated protein 1 regulates inhibitory synaptic transmission by modulating γ-aminobutyric acid type A receptor membrane trafficking. Proc Natl Acad Sci.2004;101(34):12736-12741.

Kneussel M, et al. Gephyrin-independent clustering of postsynaptic GABAA receptor subtypes. Mol Cell Neurosci.2001;17(6):973-982.

Kneussel M, et al. Loss of postsynaptic GABAA receptor clustering in gephyrin-deficient mice. J Neurosci.1999;19(21):9289-9297.

Koh MT, Rosenzweig-Lipson S, Gallagher M. Selective GABAA α5 positive allosteric modulators improve cognitive function in aged rats with memory impairment. Neuropharmacology.2013;64:145-152.

Kong F, et al. Minocycline attenuates cognitive impairment induced by isoflurane anesthesia in aged rats. PLoS One.2013;8(4).

Konsman JP, Parnet P, Dantzer R. Cytokine-induced sickness behaviour: mechanisms and implications. Trends Neurosci.2002;25(3):154-159.

Krzisch M, et al. Propofol anesthesia impairs the maturation and survival of adult-born hippocampal neurons. Anesthesiology.2013;118(3):602-610.

Lavoie AM, Tingey JJ, Harrison NL, Pritchett DB, Twyman RE. Activation and deactivation rates of recombinant GABAA receptor channels are dependent on α-subunit isoform. Biophys J.1997;73(5):2518-2526.

Le Freche H, et al. Tau phosphorylation and sevoflurane anesthesia: an association to postoperative cognitive impairment. Anesthesiology.2012;116(4):779-787.

Lee S, et al. Channel-mediated tonic GABA release from glia. Science.2010;330(6005):790- 796.

Lee V, Maguire J. The impact of tonic GABAA receptor-mediated inhibition on neuronal excitability varies across brain region and cell type. Front Neural Circuits.2014;8(3).

Lerma J, Herranz AS, Herreras O, Abraira V, Martin del Rio R. In vivo determination of extracellular concentration of amino acids in the rat hippocampus. A method based on brain dialysis and computerized analysis. Brain Res.1986;384(1):145-155.

Lester HA, Dibas MI, Dahan DS, Leite JF, Dougherty DA. Cys-loop receptors: new twists and turns. Trends Neurosci.2004;27(6):329-336.

Li RL, et al. Postoperative impairment of cognitive function in old mice: a possible role for neuroinflammation mediated by HMGB1, S100B, and RAGE. J Surg Res.2013;185(2):815-824.

221

Liang G, et al. A presenilin-1 mutation renders neurons vulnerable to isoflurane toxicity. Anesthesia and Analgesia.2008;106(2):492-500.

Lin D, Zuo Z. Isoflurane induces hippocampal cell injury and cognitive impairments in adult rats. Neuropharmacology.2011;61(8):1354-1359.

Lionel AC, et al. Rare exonic deletions implicate the synaptic organizer Gephyrin (GPHN) in risk for autism, schizophrenia and seizures. Hum Mol Genet.2013;22(10):2055-2066.

Liu H, Weng H. Up-regulation of Alzheimer's disease-associated proteins may cause enflurane anesthesia induced cognitive decline in aged rats. Neurol Sci.2014;35(2):185-189.

Liu Y, Gao M, Ma L, Zhang L, Pan N. Sevoflurane alters the expression of receptors and enzymes involved in Aβ clearance in rats. Acta Anaesthesiol Scand.2013;57(7):903- 910.

Loebrich S, Bahring R, Katsuno T, Tsukita S, Kneussel M. Activated radixin is essential for GABAA receptor α5 subunit anchoring at the actin cytoskeleton. Embo J.2006;25(5):987-999.

Lu CC, et al. Pharmacokinetics of isoflurane in human blood. Pharmacology.2008;81(4):344-349.

Lüscher B, Fuchs T, Kilpatrick CL. GABAA receptor trafficking-mediated plasticity of inhibitory synapses. Neuron.2011;70(3):385-409.

Lüscher B, Keller CA. Regulation of GABAA receptor trafficking, channel activity, and functional plasticity of inhibitory synapses. Pharmacol Ther.2004;102(3):195-221.

Lynch G, Cox CD, Gall CM. Pharmacological enhancement of memory or cognition in normal subjects. Front Syst Neurosci.2014;8:90-90.

Ma S, et al. Mutations in GABRA1, GABRA5, GABRG2 and GABRD receptor genes are not a major factor in the pathogenesis of familial focal epilepsy preceded by febrile seizures. Neurosci Lett.2006;394(1):74-78.

MacDonald JF, Mody I, Salter MW. Regulation of N-methyl-D-aspartate receptors revealed by intracellular dialysis of murine neurones in culture. J Physiol.1989;414:17-34:17- 34.

Madsen KK, Larsson OM, Schousboe A. Regulation of excitation by GABA neurotransmission: focus on metabolism and transport. Results Probl Cell Differ.2008;44:201-221.

Malenka RC. Synaptic plasticity and AMPA receptor trafficking. Ann N Y Acad Sci.2003:1- 11.

222

Martin DL, Barke KE. Are GAD(65) and GAD(67) associated with specific pools of GABA in brain? Perspect Dev Neurobiol.1998;5(2-3):119-129.

Martin LJ, Oh GH, Orser BA. Etomidate targets α5 γ-aminobutyric acid subtype A receptors to regulate synaptic plasticity and memory blockade. Anesthesiology.2009;111(5):1025-1035.

Martin LJ, et al. α5GABAA receptor activity sets the threshold for long-term potentiation and constrains hippocampus-dependent memory. J Neurosci.2010;30(15):5269-5282.

Martinez-Cue C, et al. Reducing GABAA α5 receptor-mediated inhibition rescues functional and neuromorphological deficits in a mouse model of Down syndrome. J Neurosci.2013;33(9):3953-3966.

Masters SL, O'Neill LA. Disease-associated amyloid and misfolded protein aggregates activate the inflammasome. Trends Mol Med.2011;17(5):276-282.

Mawhinney LJ, et al. Isoflurane/nitrous oxide anesthesia induces increases in NMDA receptor subunit NR2B protein expression in the aged rat brain. Brain Res.2012;11:23-34.

McCarren HS, Moore JT, Kelz MB. Assessing changes in volatile general anesthetic sensitivity of mice after local or systemic pharmacological intervention. J Vis Exp.2013;(80):e51079-e51079.

McCauley JL, et al. A linkage disequilibrium map of the 1-Mb 15q12 GABAA receptor subunit cluster and association to autism. Am J Med Genet B - Neuropsychiatr Genet.2004;131B(1):51-59.

McClellan AM, Twyman RE. Receptor system response kinetics reveal functional subtypes of native murine and recombinant human GABAA receptors. J Physiol.1999;515(Pt 3):711-727.

McDonagh DL, et al. Cognitive function after major noncardiac surgery, apolipoprotein E4 genotype, and biomarkers of brain injury. Anesthesiology.2010;112(4):852-859.

Mendez MA, et al. The brain GABA-benzodiazepine receptor α-5 subtype in autism spectrum disorder: A pilot C-11 Ro15-4513 positron emission tomography study. Neuropharmacology.2013;68:195-201.

Menold MM, et al. Association analysis of chromosome 15 GABAA receptor subunit genes in autistic disorder. J Neurogenet.2001;15(3-4):245-259.

Merkel G, Eger EI, 2nd. A comparative study of halothane and halopropane anesthesia including method for determining equipotency. Anesthesiology.1963;24:346-357.

Mihic SJ, et al. Sites of alcohol and volatile anaesthetic action on GABAA and glycine receptors. Nature.1997;389(6649):385-389.

223

Millar K, et al. The effects of brief sevoflurane-nitrous oxide anaesthesia upon children's postoperative cognition and behaviour. Anaesthesia.2006;61(6):541-547.

Millar K, et al. Children's cognitive recovery after day-case general anesthesia: a randomized trial of propofol or isoflurane for dental procedures. Paediatr Anaesth.2014;24(2):201-207.

Miller PS, Aricescu AR. Crystal structure of a human GABAA receptor. Nature.2014;512(7514):270-275.

Minelli A, Brecha NC, Karschin C, DeBiasi S, Conti F. GAT-1, a high-affinity GABA plasma membrane transporter, is localized to neurons and astroglia in the cerebral cortex. J Neurosci.1995;15(11):7734-7746.

Moller JT, et al. Long-term postoperative cognitive dysfunction in the elderly ISPOCD1 study. ISPOCD investigators. International Study of Post-Operative Cognitive Dysfunction. Lancet.1998;351(9106):857-861.

Monk TG, et al. Predictors of cognitive dysfunction after major noncardiac surgery. Anesthesiology.2008;108(1):18-30.

Mori T, et al. Evaluation of the GABAergic nervous system in autistic brain: I-123- SPECT study. Brain Dev.2012;34(8):648-654.

Morris RG, Anderson E, Lynch GS, Baudry M. Selective impairment of learning and blockade of long-term potentiation by an N-methyl-D-aspartate receptor antagonist, AP5. Nature.1986;319(6056):774-776.

Mortensen M, Smart TG. Extrasynaptic αβ subunit GABAA receptors on rat hippocampal pyramidal neurons. J Physiol.2006;577(Pt 3):841-856.

Mozrzymas JW, Zarnowska ED, Pytel M, Mercik K. Modulation of GABAA receptors by hydrogen ions reveals synaptic GABA transient and a crucial role of the desensitization process. J Neurosci.2003;23(22):7981-7992.

Murphy KL, Baxter MG. Long-term effects of neonatal single or multiple isoflurane exposures on spatial memory in rats. Front Neurol.2013;4(87).

Nacif-Coelho C, Correa-Sales C, Chang LL, Maze M. Perturbation of ion channel conductance alters the hypnotic response to the α2-adrenergic agonist dexmedetomidine in the locus coeruleus of the rat. Anesthesiology.1994;81(6):1527- 1534.

Neves G, Cooke SF, Bliss TV. Synaptic plasticity, memory and the hippocampus: a neural network approach to causality. Nat Rev Neurosci.2008;9(1):65-75.

Newman MF, et al. Longitudinal assessment of neurocognitive function after coronary-artery bypass surgery. N Engl J Med.2001;344(6):395-402.

224

Newman S, Stygall J, Hirani S, Shaefi S, Maze M. Postoperative cognitive dysfunction after noncardiac surgery - A systematic review. Anesthesiology.2007;106(3):572-590.

Ni Choileain N, Redmond HP. Cell response to surgery. Arch Surg.2006;141(11):1132-1140.

Nicholls RD, Knepper JL. Genome organization, function and imprinting in Prader-Willi and Angelman syndromes. Annu Rev Genomics Hum Genet.2001;2:153-175.

Nikiforuk A, Fijal K, Potasiewicz A, Popik P, Kos T. The 5-hydroxytryptamine (serotonin) receptor 6 agonist EMD 386088 ameliorates ketamine-induced deficits in attentional set shifting and novel object recognition, but not in the prepulse inhibition in rats. J Psychopharmacol.2013;27(5):469-476.

Nishikawa K, Jenkins A, Paraskevakis I, Harrison NL. Volatile anesthetic actions on the GABAA receptors: contrasting effects of α1(S270) and β2(N265) point mutations. Neuropharmacology.2002;42(3):337-345.

Nusser Z, Sieghart W, Somogyi P. Segregation of different GABAA receptors to synaptic and extrasynaptic membranes of cerebellar granule cells. J Neurosci.1998;18(5):1693-1703.

Nutt DJ, Besson M, Wilson SJ, Dawson GR, Lingford-Hughes AR. Blockade of alcohol's amnestic activity in humans by an α5 subtype benzodiazepine receptor inverse agonist. Neuropharmacology.2007.

Nyiri G, Freund TF, Somogyi P. Input-dependent synaptic targeting of α2-subunit-containing GABAA receptors in synapses of hippocampal pyramidal cells of the rat. Eur J Neurosci.2001;13(3):428-442.

O'Callaghan JP, Sriram K. Focused microwave irradiation of the brain preserves in vivo protein phosphorylation: comparison with other methods of sacrifice and analysis of multiple phosphoproteins. J Neurosci Methods.2004;135(1-2):159-168.

O'Keefe J, Dostrovsky J. The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat. Brain Res.1971;34(1):171-175.

Oblak A, Gibbs TT, Blatt GJ. Decreased GABAA receptors and benzodiazepine binding sites in the anterior cingulate cortex in autism. Autism Res.2009;2(4):205-219.

Ogris W, et al. Investigation of the abundance and subunit composition of GABAA receptor subtypes in the cerebellum of α1-subunit-deficient mice. J Neurochem.2006;96(1):136-147.

Olsen RW, Sieghart W. GABAA receptors: subtypes provide diversity of function and pharmacology. Neuropharmacology.2009;56(1):141-148.

Orser BA, Pennefather PS, MacDonald JF. Multiple mechanisms of ketamine blockade of N- methyl-D-aspartate receptors. Anesthesiology.1997;86(4):903-917.

225

Oruc L, et al. Positive association between the Gabra5 gene and unipolar recurrent major depression. Neuropsychobiology.1997;36(2):62-64.

Otani K, et al. The GABA type A receptor α5 subunit gene is associated with bipolar I disorder. Neurosci Lett.2005;381(1-2):108-113.

Padgett CL, Slesinger PA. GABAB receptor coupling to G-proteins and ion channels. Adv Pharmacol.2010;58:123-147.

Palanisamy A, et al. Rats exposed to isoflurane in utero during early gestation are behaviorally abnormal as adults. Anesthesiology.2011;114(3):521-528.

Pan Z, et al. The effects of sevoflurane anesthesia on rat hippocampus: a genomic expression analysis. Brain Res.2011;24:124-133.

Papadimitriou GN, et al. Association between the GABAA receptor α5 subunit gene locus (GABRA5) and bipolar affective disorder. Am J Med Genet.1998;81(1):73-80.

Papaioannou A, Fraidakis O, Michaloudis D, Balalis C, Askitopoulou H. The impact of the type of anaesthesia on cognitive status and delirium during the first postoperative days in elderly patients. Eur J Anaesthesiol.2005;22(7):492-499.

Pavel O, William W, Rolf S. 2006. Molecular mechanisms of synaptic function in the hippocampus: neurotransmitter exocytosis and glutamatergic, GABAergic, and cholinergic transmission. In The Hippocampus Book. New York: Oxford University Press.

Pekny M, Nilsson M. Astrocyte activation and reactive gliosis. Glia.2005;50(4):427-434.

Peng L, Xu L, Ouyang W. Role of peripheral inflammatory markers in postoperative cognitive dysfunction (POCD): A meta-analysis. PLoS One.2013;8(11).

Penna A, et al. Hydrogen peroxide increases GABAA receptor-mediated tonic current in hippocampal neurons. J Neurosci.2014;34(32):10624-10634.

Perea G, Navarrete M, Araque A. Tripartite synapses: astrocytes process and control synaptic information. Trends Neurosci.2009;32(8):421-431.

Perucho J, et al. Anesthesia with isoflurane increases amyloid pathology in mice models of Alzheimer's Disease. J Alzheimers Dis.2010;19(4):1245-1257.

Phillips-Bute B, et al. Association of neurocognitive function and quality of life 1 year after coronary artery bypass graft (CABG) surgery. Psychosom Med.2006;68(3):369-375.

Pirker S, Schwarzer C, Wieselthaler A, Sieghart W, Sperk G. GABAA receptors: immunocytochemical distribution of 13 subunits in the adult rat brain. Neuroscience.2000;101(4):815-850.

226

Planel E, et al. Acceleration and persistence of neurofibrillary pathology in a mouse model of tauopathy following anesthesia. Faseb Journal.2009;23(8):2595-2604.

Planel E, et al. Anesthesia-induced hyperphosphorylation detaches 3-repeat tau from microtubules without affecting their stability in vivo. J Neurosci.2008;28(48):12798- 12807.

Planel E, et al. Anesthesia leads to tau hyperphosphorylation through inhibition of phosphatase activity by hypothermia. J Neurosci.2007;27(12):3090-3097.

Pow DV, et al. Differential expression of the GABA transporters GAT-1 and GAT-3 in brains of rats, cats, monkeys and humans. Cell Tissue Res.2005;320(3):379-392.

Pozo K, Goda Y. Unraveling mechanisms of homeostatic synaptic plasticity. Neuron.2010;66(3):337-351.

Price CC, Garvan CW, Monk TG. Type and severity of cognitive decline in older adults after noncardiac surgery. Anesthesiology.2008;108(1):8-17.

Price RB, Nock MK, Charney DS, Mathew SJ. Effects of intravenous ketamine on explicit and implicit measures of suicidality in treatment-resistant depression. Biol. Psychiatry.2009;66(5):522-526.

Qian XL, et al. Dexmedetomidine improves early postoperative cognitive dysfunction in aged mice. Eur J Pharmacol.2014;20(14):00793-00796.

Quirk K, et al. [3H]L-655,708, a novel ligand selective for the benzodiazepine site of GABAA receptors which contain the α5 subunit. Neuropharmacology.1996;35(9- 10):1331-1335.

Quraishi SA, Girdharry TD, Xu SG, Orkin FK. Prolonged retrograde amnesia following sedation with propofol in a 12-year-old boy. Pediatr Anesth.2007;17(4):375-379.

Radian R, Kanner BI. Stoichiometry of sodium- and chloride-coupled γ-aminobutyric acid transport by synaptic plasma membrane vesicles isolated from rat brain. Biochemistry.1983;22(5):1236-1241.

Radian R, Ottersen OP, Stormmathisen J, Castel M, Kanner BI. Immunocytochemical localization of the GABA transporter in rat-brain. J Neurosci.1990;10(4):1319-1330.

Ramlawi B, et al. C-Reactive protein and inflammatory response associated to neurocognitive decline following cardiac surgery. Surgery.2006;140(2):221-226.

Ransohoff RM, Brown MA. Innate immunity in the central nervous system. J Clin Invest.2012;122(4):1164-1171.

227

Rasmussen LS, et al. Does anaesthesia cause postoperative cognitive dysfunction? A randomised study of regional versus general anaesthesia in 438 elderly patients. Acta Anaesthesiol Scand.2003;47(3):260-266.

Rasmussen LS, et al. Is peri-operative cortisol secretion related to post-operative cognitive dysfunction? Acta Anaesthesiol Scand.2005;49(9):1225-1231.

Rau V, et al. γ-Aminobutyric acid type A receptor α4 subunit knockout mice are resistant to the amnestic effect of isoflurane. Anesth Analg.2009;109(6):1816-1822.

Rau V, et al. γ-Aminobutyric acid type A receptor β3 subunit forebrain-specific knockout mice are resistant to the amnestic effect of isoflurane. Anesth Analg.2011;113(3):500- 504.

Redrobe JP, et al. Negative modulation of GABAA alpha5 receptors by RO4938581 attenuates discrete sub-chronic and early postnatal phencyclidine (PCP)-induced cognitive deficits in rats. Psychopharmacology.2012;221(3):451-468.

Reymann KG, Frey JU. The late maintenance of hippocampal LTP: requirements, phases, 'synaptic tagging', 'late-associativity' and implications. Neuropharmacology.2007;52(1):24-40.

Reynolds DS, et al. Sedation and anesthesia mediated by distinct GABAA receptor isoforms. J Neurosci.2003;23(24):8608-8617.

Rivera C, et al. The K+/Cl- co-transporter KCC2 renders GABA hyperpolarizing during neuronal maturation. Nature.1999;397(6716):251-255.

Roberts E, Frankel S. γ-aminobutyric acid in brain - Its formation from glutamic acid. J Biol Chem.1950;187(1):55-63.

Rohan D, et al. Increased incidence of postoperative cognitive dysfunction 24 hr after minor surgery in the elderly. Can J Anaesth.2005;52(2):137-142.

Rohm KD, Piper SN, Suttner S, Schuler S, Boldt J. Early recovery, cognitive function and costs of a desflurane inhalational vs. a total intravenous anaesthesia regimen in long- term surgery. Acta Anaesthesiol. Scand.2006;50(1):14-18.

Rubenstein JLR, Merzenich MM. Model of autism: increased ratio of excitation/inhibition in key neural systems. Genes Brain Behav.2003;2(5):255-267.

Rudolph JL, et al. Delirium is associated with early postoperative cognitive dysfunction. Anaesthesia.2008;63(9):941-947.

Rudolph U, Antkowiak B. Molecular and neuronal substrates for general anaesthetics. Nat Rev Neurosci.2004;5(9):709-720.

228

Rye DB, et al. Modulation of vigilance in the primary hypersomnias by endogenous enhancement of GABAA receptors. Sci Transl Med.2012;4(161):3004685.

Ryu YK, Khan S, Smith SC, Mintz CD. Isoflurane impairs the capacity of astrocytes to support neuronal development in a mouse dissociated coculture model. J Neurosurg Anesthesiol.2014;26(4):363-368.

Saab BJ, et al. Short-term memory impairment after isoflurane in mice is prevented by the α5 γ-aminobutyric acid type A receptor inverse agonist L-655,708. Anesthesiology.2010;113(5):1061-1071.

Saliba RS, Kretschmannova K, Moss SJ. Activity-dependent phosphorylation of GABAA receptors regulates receptor insertion and tonic current. Embo J.2012;31(13):2937- 2951.

Sanders RD, Maze M. α2-adrenoceptor agonists. Curr Opin Investig Drugs.2007;8(1):25-33.

Sanders RD, et al. Dexmedetomidine provides cortical neuroprotection: impact on anaesthetic-induced neuroapoptosis in the rat developing brain. Acta Anaesthesiol Scand.2010;54(6):710-716.

Sanders RD, et al. Dexmedetomidine attenuates isoflurane-induced neurocognitive impairment in neonatal rats. Anesthesiology.2009;110(5):1077-1085.

Savic MM, et al. PWZ-029, a compound with moderate inverse agonist functional selectivity at GABAA receptors containing α5 subunits, improves passive, but not active, avoidance learning in rats. Brain Res.2008;7:150-159.

Savic MM, et al. Are GABAA receptors containing α5 subunits contributing to the sedative properties of benzodiazepine site agonists? Neuropsychopharmacology.2008;33(2):332-339.

Schoeler M, et al. Dexmedetomidine is neuroprotective in an in vitro model for traumatic brain injury. BMC Neurol.2012;12(20):1471-2377.

Schroer RJ, et al. Autism and maternally derived aberrations of chromosome 15q. Am J Hum Genet.1998;76(4):327-336.

Schulz PE, Cook EP, Johnston D. Changes in paired-pulse facilitation suggest presynaptic involvement in long-term potentiation J Neurosci.1994;14(9):5325-5337.

Scimemi A. Structure, function, and plasticity of GABA transporters. Front Cell Neurosci.2014;8(161).

Scoville WB, Milner B. Loss of recent memory after bilateral hippocampal lesions. J Neurol Neurosurg Psychiatry.1957;20(1):11-21.

229

Sekine S, et al. Changes in expression of GABAA α4 subunit mRNA in the brain under anesthesia induced by volatile and intravenous anesthetics. Neurochem Res.2006;31(3):439-448.

Serwanski DR, et al. Synaptic and nonsynaptic localization of GABAA receptors containing the α5 subunit in the rat brain. J Comp Neurol.2006;499(3):458-470.

Shao YJ, et al. Fine mapping of Autistic disorder to chromosome 15q11-q13 by use of phenotypic subtypes. Am J Hum Genet.2003;72(3):539-548.

Silverman JL, Tolu SS, Barkan CL, Crawley JN. Repetitive self-grooming behavior in the BTBR mouse model of autism is blocked by the mGluR5 antagonist MPEP. Neuropsychopharmacology.2010;35(4):976-989.

Silverman JL, Yang M, Lord C, Crawley JN. Behavioural phenotyping assays for mouse models of autism. Nat Rev Neurosci.2010;11(7):490-502.

Silverstein JH, Timberger M, Reich DL, Uysal S. Central nervous system dysfunction after noncardiac surgery and anesthesia in the elderly. Anesthesiology.2007;106(3):622- 628.

Somogyi P, Fritschy JM, Benke D, Roberts JD, Sieghart W. The γ2 subunit of the GABAA receptor is concentrated in synaptic junctions containing the α1 and β2/3 subunits in hippocampus, cerebellum and globus pallidus. Neuropharmacology.1996;35(9- 10):1425-1444.

Somogyi P, Klausberger T. Defined types of cortical interneurone structure space and spike timing in the hippocampus. J Physiol.2005;562(1):9-26.

Song I, et al. Different transporter systems regulate extracellular GABA from vesicular and non-vesicular sources. Front Cell Neurosci.2013;7(23).

Sonner JM, Gong D, Eger EI. Naturally occurring variability in anesthetic potency among inbred mouse strains. Anesth Analg.2000;91(3):720-726.

Sonner JM, et al. Effect of isoflurane and other potent inhaled anesthetics on minimum alveolar concentration, learning, and the righting reflex in mice engineered to express α1 γ-aminobutyric acid type A receptors unresponsive to isoflurane. Anesthesiology.2007;106(1):107-113.

Spruston NLS, McBain C. 2006. Structural and functional properties of hippocampal neurons. In The Hippocampus Book. New York: Oxford University Press.

Staley KJ, Mody I. Shunting of excitatory input to dentate gyrus granule cells by a depolarizing GABAA receptor-mediated postsynaptic conductance. J Neurophysiol.1992;68(1):197-212.

230

Staley KJ, Soldo BL, Proctor WR. Ionic mechanisms of neuronal excitation by inhibitory GABAA receptors. Science.1995;269(5226):977-981.

Steele PM, Mauk MD. Inhibitory control of LTP and LTD: Stability of synapse strength. J Neurophysiol.1999;81(4):1559-1566.

Steinmetz J, Christensen KB, Lund T, Lohse N, Rasmussen LS. Long-term consequences of postoperative cognitive dysfunction. Anesthesiology.2009;110(3):548-555.

Steinmetz J, Siersma V, Kessing LV, Rasmussen LS. Is postoperative cognitive dysfunction a risk factor for dementia? A cohort follow-up study. Br J Anaesth.2013;110(1):28.

Stern SA, Alberini CM. Mechanisms of memory enhancement. Wiley Interdiscip Rev Syst Biol Med.2013;5(1):37-53.

Stockton P, Cohen-Mansfield J, Billig N. Mental status change in older surgical patients - Cognition, depression, and other comorbidity. Am J Geriatr Psychiatry.2000;8(1):40- 46.

Strum DP, Johnson BH, Eger EI, 2nd. Elimination of anesthetics from rabbit brain. Science.1986;234(4783):1586-1588.

Sur C, Quirk K, Dewar D, Atack J, McKernan R. Rat and human hippocampal α5 subunit- containing gamma-aminobutyric acid A receptors have α5β3γ2 pharmacological characteristics. Mol Pharmacol.1998;54(5):928-933.

Tan W, et al. Tau hyperphosphorylation is associated with memory impairment after exposure to 1.5% isoflurane without temperature maintenance in rats. Eur J Anaesthesiol.2010;27(9):835-841.

Tanabe K, Matsushima-Nishiwaki R, Kozawa O, Iida H. Dexmedetomidine suppresses interleukin-1beta-induced interleukin-6 synthesis in rat glial cells. Int J Mol Med.2014;34(4):1032-1038.

Terrando N, et al. Tumor necrosis factor-α triggers a cytokine cascade yielding postoperative cognitive decline. Proc Natl Acad Sci.2010;107(47):20518-20522.

Terunuma M, et al. GABAA receptor phospho-dependent modulation is regulated by phospholipase C-related inactive protein type 1, a novel protein phosphatase 1 anchoring protein. J Neurosci.2004;24(32):7074-7084.

Thrane AS, et al. General anesthesia selectively disrupts astrocyte calcium signaling in the awake mouse cortex. Proc Natl Acad Sci.2012;109(46):18974-18979.

Tia S, Wang JF, Kotchabhakdi N, Vicini S. Distinct deactivation and desensitization kinetics of recombinant GABAA receptors. Neuropharmacology.1996;35(9-10):1375-1382.

231

Tossman U, Jonsson G, Ungerstedt U. Regional distribution and extracellular levels of amino acids in rat central nervous system. Acta Physiol Scand.1986;127(4):533-545.

Towers SK, et al. α5 Subunit-containing GABAA receptors affect the dynamic range of mouse hippocampal kainate-induced gamma frequency oscillations in vitro. J Physiol.2004;559(Pt 3):721-728.

Tretter V, et al. Targeted disruption of the GABAA receptor δ subunit gene leads to an up- regulation of γ2 subunit-containing receptors in cerebellar granule cells. J Biol Chem.2001;276(13):10532-10538.

Tretter V, et al. The clustering of GABAA receptor subtypes at inhibitory synapses is facilitated via the direct binding of receptor α2 subunits to gephyrin. J Neurosci.2008;28(6):1356-1365.

Tretter V, et al. Gephyrin, the enigmatic organizer at GABAergic synapses. Front Cell Neurosci.2012;6(23).

Tsai PT, et al. Autistic-like behaviour and cerebellar dysfunction in purkinje cell Tsc1 mutant mice. Nature.2012;488(7413):647-+.

Tsai SJ, et al. Interleukin-1β (C-511T) genetic polymorphism is associated with cognitive performance in elderly males without dementia. Neurobiol Aging.2008.

Tyagarajan SK, Fritschy JM. Gephyrin: a master regulator of neuronal function? Nat Rev Neurosci.2014;15(3):141-156.

Uusi-Oukari M, Korpi ER. Regulation of GABAA receptor subunit expression by pharmacological agents. Pharmacol Rev.2010;62(1):97-135.

Valentim AM, et al. Lower isoflurane concentration affects spatial learning and neurodegeneration in adult mice compared with higher concentrations. Anesthesiology.2010;113(5):1099-1108.

Van Hamme MJ, Ghoneim MM, Ambre JJ. Pharmacokinetics of etomidate, a new intravenous anesthetic. Anesthesiology.1978;49(4):274-277.

Vithlani M, Terunuma M, Moss SJ. The dynamic modulation of GABAA receptor trafficking and its role in regulating the plasticity of inhibitory synapses. Physiol Rev.2011;91(3):1009-1022.

Vitureira N, Goda Y. The interplay between Hebbian and homeostatic synaptic plasticity. J Cell Biol.2013;203(2):175-186.

Vizcaychipi MP, et al. Xenon pretreatment may prevent early memory decline after isoflurane anesthesia and surgery in mice. PLoS One.2011;6(11):3.

232

Wallace TL, Ballard TM, Pouzet B, Riedel WJ, Wettstein JG. Drug targets for cognitive enhancement in neuropsychiatric disorders. Pharmacol Biochem Behav.2011;99(2):130-145.

Wang D-S, et al. Memory deficits induced by inflammation are regulated by α5-subunit- containing GABAA receptors. Cell Reports.2012;2(3):488-496.

Wayman GA, Lee YS, Tokumitsu H, Silva A, Soderling TR. Calmodulin-kinases: modulators of neuronal development and plasticity. Neuron.2008;59(6):914-931.

Weber CF, et al. Impact of general versus local anesthesia on early postoperative cognitive dysfunction following carotid endarterectomy: GALA study subgroup analysis. World J Surg.2009;33(7):1526-1532.

Wei H, et al. The common inhalational anesthetic isoflurane induces apoptosis via activation of inositol 1,4,5-trisphosphate receptors. Anesthesiology.2008;108(2):251-260.

Weiser TG, et al. An estimation of the global volume of surgery: a modelling strategy based on available data. Lancet.2008;372(9633):139-144.

Wiklund A, et al. Object memory in young and aged mice after sevoflurane anaesthesia. Neuroreport.2009;20(16):1419-1423.

Wilder RT, et al. Early exposure to anesthesia and learning disabilities in a population-based birth cohort. Anesthesiology.2009;110(4):796-804.

Winters BD, Saksida LM, Bussey TJ. Implications of animal object memory research for human amnesia. Neuropsychologia.2010;48(8):2251-2261.

Wixted JT, Squire LR. The role of the human hippocampus in familiarity-based and recollection-based recognition memory. Behav Brain Res.2010;215(2):197-208.

Woehr M, Roullet FI, Hung AY, Sheng M, Crawley JN. Communication impairments in mice lacking Shank1: reduced levels of ultrasonic vocalizations and scent marking behavior. PLoS One.2011;6(6).

Wohlfarth KM, Bianchi MT, Macdonald RL. Enhanced neurosteroid potentiation of ternary GABAA receptors containing the δ subunit. J Neurosci.2002;22(5):1541-1549.

Wu X, et al. The inhalation anesthetic isoflurane increases levels of proinflammatory TNF- alpha, IL-6, and IL-1beta. Neurobiol Aging.2012;33(7):1364-1378.

Xie ZC, et al. The common inhalation anesthetic isoflurane induces caspase activation and increases amyloid β-protein level in vivo. Ann Neurol.2008;64(6):618-627.

Xiong WX, et al. Impaired spatial learning and memory after sevoflurane-nitrous oxide anesthesia in aged rats is associated with down-regulated cAMP/CREB signaling. PLoS One.2013;8(11).

233

Yee BK, et al. GABAA receptors containing the α5 subunit mediate the trace effect in aversive and appetitive conditioning and extinction of conditioned fear. Eur J Neurosci.2004;20(7):1928-1936.

Yeung JY, et al. Tonically activated GABAA receptors in hippocampal neurons are high- affinity, low-conductance sensors for extracellular GABA. Mol Pharmacol.2003;63(1):2-8.

Yip GMS, et al. A propofol binding site on mammalian GABAA receptors identified by photolabeling. Nat Chem Biol.2013;9(11):715-720.

Yirmiya R, Goshen I. Immune modulation of learning, memory, neural plasticity and neurogenesis. Brain Behav Immun.2011;25(2):181-213.

Yizhar O, et al. Neocortical excitation/inhibition balance in information processing and social dysfunction. Nature.2011;477(7363):171-178.

Yoon BE, et al. The amount of astrocytic GABA positively correlates with the degree of tonic inhibition in hippocampal CA1 and cerebellum. Mol Brain.2011;4(42):1756- 6606.

Young DM, Schenk AK, Yang S-B, Jan YN, Jan LY. Altered ultrasonic vocalizations in a tuberous sclerosis mouse model of autism. Proc Natl Acad Sci.2010;107(24):11074- 11079.

Zhang B, et al. The effects of isoflurane and desflurane on cognitive function in humans. Anesth Analg.2012;114(2):410-415.

Zhang GH, et al. Isoflurane-induced caspase-3 activation is dependent on cytosolic calcium and can be attenuated by memantine. J Neurosci.2008;28(17):4551-4560.

Zhang X, et al. Dexmedetomidine inhibits tumor necrosis factor-α and interleukin 6 in lipopolysaccharide-stimulated astrocytes by suppression of c-Jun N-terminal kinases. Inflammation.2014;37(3):942-949.

Zhen Y, et al. Nitrous oxide plus isoflurane induces apoptosis and increases β-amyloid protein levels. Anesthesiology.2009;111(4):741-752.

Zurek AA, Bridgwater EM, Orser BA. Inhibition of α5 γ-Aminobutyric acid type A receptors restores recognition memory after general anesthesia. Anesth Analg.2012;114(4):845- 855.

Zurek AA, et al. Sustained increase in α5GABAA receptor function impairs memory after anesthesia. J Clin Invest.2014;124(12):5437-5441.