Preclinical Validation of Endocannabinoid System Drug Targets for the Treatment of Dravet Syndrome

Preclinical validation of endocannabinoid system drug targets for the treatment of Dravet syndrome

Dilara Bahceci

A thesis submitted to fulfil requirements for the degree of Doctor of Philosophy

Faculty of Medicine and Health The University of Sydney 2020

Acknowledgements They say it takes a village to raise a child, and I believe this sentiment rings true for the completion of a PhD thesis. Although I am the sole author on this document, this thesis would not have been possible without the input of countless people, many who will have to go unnamed but not unappreciated. Firstly, I’d like to acknowledge our benefactors, Barry and Joy Lambert, because without their generosity and inspiration, the Lambert Initiative for Cannabinoid Therapeutics and this research would not exist, and cannabinoid research in Australia would have been stunted. I would like to sincerely thank all the members of the Lambert Initiative, prior and present, for creating an environment that pushes you to achieve your best and helps you not only survive the process but somehow enjoy it too. I have come to appreciate the environment of high- expectations and high-pressure that has forged me in to the confident and capable researcher I am today. I’m especially grateful to everyone for their comradery and generosity with knowledge, guidance, kindness, encouragement, support and for creating so many wonderful memories that I will carry with me forevermore. I’d especially like to express my deepest gratitude for my supervisor Jonathon Arnold for his mentorship in teaching me the ropes of academia and research. More so though, I’d like to thank Jonathon for his consistent patience, compassion and support in the face of life’s unexpected adversities. Thank you for prioritising my personal health and safety, whilst simultaneously encouraging me to persevere so I did not sacrifice my potential in the process. I could not have done this without you and I will always be grateful. Words cannot express the gratitude I have for Lyndsey Anderson, my unofficial supervisor and science sister for her close mentorship. Thank you for being so generous with your time in providing lessons on science and life. I’d also like to acknowledge A/Prof Jennifer Kearney and Nicole Hawkins for their experimental assistance and sharing of knowledge, in addition to their intellectual insight and advice. I would also like to thank my associate supervisor, and our fearless leader, Iain McGregor for his academic guidance and support. I want to acknowledge all female scientists, prior and present, near and far. In a career that is inherently challenging, the existence of additional unjust and uncontrollable barricades can be disheartening. However, seeing women who have persevered in the face of adversity to achieve your desires is inspiring and motivating. Thank you for paving the way. I’d also like to thank all the female scientists who currently surround me, my science sisters, for providing a sense of community, solidarity and support. We are stronger together. Thank you to my friends and family. Despite not knowing what I actually do or understanding why I am still a student at my age, you have supported me unconditionally throughout this defining journey. I will never be able to express my appreciation for all the love and support that came in many forms, but I will continue trying. Lastly, I offer my gratitude to the mice that contributed to this research. Thank you.

Table of Contents Statement of originality and author attribution ...... i Acknowledgements ...... ii Table of Contents ...... i Publications ...... iii List of figures ...... iv List of tables ...... vi Abbreviations ...... vii Abstract ...... xi Chapter 1 Introduction ...... 1 1. Introduction to cannabis and cannabinoids ...... 2 2. Cannabis-based treatments for epilepsy...... 4 3. The endocannabinoid system ...... 7 3.1 Cannabinoid receptors 1 and 2 ...... 7

3.2 Endocannabinoids ...... 9

3.3 CB1-mediated signalling ...... 11

3.4 G protein-coupled receptor 55 (GPR55) ...... 13

4. Targeting the endocannabinoid system for epilepsy ...... 17 5. Epilepsy...... 24 6. Ion channels ...... 27 7. Voltage-gated sodium channels ...... 28

8. SCN1A mutations and NaV1.1 impairment ...... 30 9. Dravet syndrome ...... 32 9.1 Current treatment strategies for Dravet syndrome ...... 35

9.2 Scn1a+/- mouse models of Dravet syndrome ...... 40

10. Genetic modifiers: a strategy to identify drug targets ...... 42 11. Aims ...... 44 11.1 Specific aims...... 44

Chapter 2 Validation of CB1 receptors as an anticonvulsant target in a Scn1a+/- mouse model of Dravet Syndrome ...... 48 Introduction ...... 49 Materials and methods ...... 53 Results ...... 61 Discussion ...... 69 Supplementary materials ...... 76 Chapter 3 Targeting Gpr55 as a treatment strategy in a Scn1a+/- mouse model of Dravet syndrome ...... 77 Introduction ...... 78 Materials and Methods ...... 80 Results ...... 85 Discussion ...... 92 Supplementary materials ...... 98 Chapter 4 Adolescent behavioral abnormalities in a Scn1a+/- mouse model of Dravet syndrome ...... 102 Adolescent behavioral abnormalities in a Scn1a+/− mouse model of Dravet syndrome ...... 103 1. Introduction ...... 104 2. Material and methods ...... 105 3. Results ...... 111 4. Discussion ...... 118 Declaration of competing interest ...... 122 Acknowledgments...... 122 References ...... 122 Chapter 5 General discussion...... 124 Conclusion ...... 142 Chapter 6 References ...... 143

iii

Publications

Bahceci, D, Anderson, LL, Occelli Hanbury Brown, CV, Zhou, C & Arnold, JC 2019,

'Adolescent behavioral abnormalities in a Scn1a+/− mouse model of Dravet syndrome', Epilepsy

& Behavior, pp. 106842.

Clarke, DJ, Sarkissian, L, Todd, SM, Suraev, AS, Bahceci, D, Brzozowska, N & Arnold,

JC 2019, 'Nrg1 deficiency modulates the behavioural effects of prenatal stress in mice', Prog

Neuropsychopharmacol Biol Psychiatry, vol. 88, pp. 86-95.

Todd, SM, Zhou, C, Clarke, DJ, Chohan, TW, Bahceci, D & Arnold, JC 2017, 'Interactions between cannabidiol and Delta(9)-THC following acute and repeated dosing: Rebound hyperactivity, sensorimotor gating and epigenetic and neuroadaptive changes in the mesolimbic pathway', Eur Neuropsychopharmacol, vol. 27, no. 2, pp. 132-145.

List of figures

Chapter 1

Figure 1. Phytocannabinoid biosynthesis and degradation routes and products for the 11 types of

phytocannabinoids, from Berman et al. (2018)...... 3

Figure 2. The synthesis and metabolism of anandamide and 2-AG, adapted from Parsons and Hurd

(2015) ...... 8

Figure 3. Cooperation between the GPCRs orthosteric and allosteric binding sites to induce intracellular protein coupling, from Scott & Kendall, 2017………………………………………22

Figure 4. The expanded ILAE 2017 operational classification of seizure types, modified from

Fisher et al. (2017b)...... 25

Figure 5. A schematic representation of the structure of the α and β subunits of voltage-gated

sodium channel, adapted from George (2005)...... 29

Figure 6. Schematic representation of reduced SCN1A expression in Dravet syndrome, with permission from Doohan (2019)………………………………………………………………….34

Chapter 2

Figure 1. CB1 receptor mRNA and protein expression in WT and Scn1a+/- mice on 129 and F1

(129 x B6) background strains...... 61

Figure 2. Hippocampal concentrations of endocannabinoids in WT and Scn1a+/- mice from 129

and F1 (129 x B6) background strains...... 63

Figure 3. Hippocampal transcript expression of 2-AG’s synthesizing and catabolising enzymes in

WT and Scn1a+/- mice from 129 and F1 (129 x B6) background strains...... 65

Figure 4. Efficacy of GAT229, a CB1 receptor PAM, and ABX-1431, a MAGL inhibitor, against

hyperthermia-induced seizures in F1.Scn1a+/- mice...... 67

Chapter 3

Figure 1. Hippocampal expression of Gpr55 transcript was elevated in F1 seizure-susceptible

mice...... 85

Figure 2. Hippocampal concentrations of LPIs in WT and Scn1a+/- mice on 129 and F1 background

strains...... 86

Figure 3. Hippocampal mRNA expression of the best characterized LPI synthetic and catabolic

enzymes in the Scn1a+/- mouse model...... 88

Figure 4. Heterozygous Gpr55 deletion was anticonvulsant and restored the survival deficit in

Scn1a+/- mice...... 90

Supplementary Figure 1. Unsuccessful validation of three commercially-available primary

antibodies for Gpr55 using wildtype (WT) and Gpr55-/- (KO) tissue, and human GPR55-

expressing cells...... 98

Chapter 4

Figure 1. Sequence of behavioral tests ...... 107

Figure 2. Scn1a+/- mice displayed anxiogenic-like behavior in the open field test...... 112

Figure 3. Scn1a+/- mice displayed normal behavior in the elevated plus maze and the marble

burying test...... 113

Figure 4. Scn1a+/− mice displayed impaired preference for social novelty compared with WT mice

...... 114

Figure 5. Scn1a+/- mice displayed better novel object recognition and impaired spatial memory

compared to WT mice ...... 115

Figure 6. Scn1a+/- mice exhibited an acoustic startle and an atypical fear response ...... 117

List of tables

Chapter 1

Table 1. Drugs used in the treatment of Dravet syndrome and their proposed mechanisms of action...... 35

Chapter 2

Table 1. Analyte mass transitions used for HPLC/MS/MS quantification...... 59

Table 2. Hippocampal concentrations of fatty acid amides in WT and Scn1a+/- mice from

129 and F1 (129 x B6) background strains, as determined by HPLC/MS/MS...... 64

Table 3. Plasma and brain concentrations of GAT229 at the time of testing against hyperthermia-induced seizures...... 68

Table 4. Plasma and brain concentrations of ABX-1431 at the time of testing against hyperthermia-induced seizures...... 69

vii

Abbreviations

129 129S6/SvEvTac

2-AG 2-arachidonoylglycerol

ABHD12 α/β-hydrolase-12

ABHD6 α/β-hydrolase-6

ACEA arachidonyl-2'-chloroethylamide

AEA N-arachidonoylethanolamide, anandamide

AGPAT8 1-acylglycerol-3-phosphate acyltransferase 8

ANOVA analysis of variance

ATF-2 activating transcription factor 2

B6 C57BL/6 cAMP cyclic adenosine monophosphate levels

CB1 cannabinoid receptor 1

CB2 cannabinoid receptor 2

CBC cannabichromene

CBD cannabidiol

CBE cannabielsoin

CBG cannabigerol

CBL cannabicyclol

CBN cannabinol

CBND cannabinodiol

CBT cannabitriol

viii

CE collision energy

CNS central nervous system

DAG diacylglycerols

DAGL diacylglycerol lipase ddRT-PCR digital-drop real-time PCR

DHEA docosahexaenoylethanolamide

DI discrimination index

DSI/DSE depolarization-induced suppression of inhibition/excitation

EDTA ethylenediaminetetraacetic acid

EEG electroencephalogram

ERK1/2 extracellular signal-regulated kinase

ESI+ electrospray ionization in positive mode

FAAH fatty acid amine hydrolase

FABP5 fatty-acid binding protein 5

GEFS+ generalised epilepsy with febrile seizures plus

GIRK inwardly-rectifying potassium channels

GPR55 G protein-coupled receptor 55

GTCS generalized tonic-clonic seizure

HPLC/MS/MS high-performance liquid chromatography tandem mass spectrometry i.p. intraperitoneal

ILAE International League Against Epilepsy

IPSC induced pluripotent stem cell

LEA dihomo-γ-inolenoylethanolamine

LPI L-α-lysophosphatidylinositol

LPIAT1 LPI-acyltransferase-1

MAGL monoacylglycerol lipase

MAPK mitogen-activated protein kinase

MBOAT7 membrane-bound O-acyltransferase 7

MES maximal electroshock

MRM multiple reaction monitoring

MSI/MSE metabotropic-induced suppression of inhibition/excitation

MTBE methyl tert-butyl ether

NAPE N-arachidonoyl phosphatidylethanolamine

NAPE-PLD N-acyl phosphatidylethanolamine phospholipase D

NaV voltage-gated sodium

NaV1.1 voltage-gated sodium channel, type I, alpha subunit

NFAT nuclear factor of activated T-cells

NF-ĸβ nuclear factor kappa-light-chain-enhancer of activated B cells

NHMRC National Health and Medical Research Council

NMDA non-N-methyl-D-aspartate

OEA oleoylethanolamine

P postnatal day

PACAP-27 pituitary adenylate cyclase activating peptide-27

PAM positive allosteric modulator

PA-PLA1 phosphatidic acid-preferring phospholipase A1

PCR polymerase chain reaction

PEA palmitoylethanolamide

PI phosphatidylinositol

PLC phospholipase C

PPI prepulse inhibition of startle

PTZ pentylenetetrazol

RhoA-ROCK Ras homolog family member A - Rho-associated protein kinase

RIPA radioimmunoprecipitation assay

SEM standard error of the mean

SLE supported-liquid extraction

SSI/SSE synaptically-evoked suppression of inhibition/excitation

SUDEP sudden unexpected death in epilepsy

TRP transient receptor potential

TRPV1 transient receptor potential vanilloid 1

WHO World Health Organization

WT wildtype

Δ8-THC (−)-Δ8-trans-tetrahydrocannabinol

Δ9-THC (−)-Δ9-trans-tetrahydrocannabinol

Abstract

Dravet syndrome is a severe, early-onset epileptic encephalopathy associated with multiple seizure types, behavioural impairments, and significant risk of sudden unexplained death in epilepsy (SUDEP). Patients exhibit poor prognosis due to pharmacoresistance, with no specific therapeutic options for the associated behavioural comorbidities that have a greater impact on patients’ quality of life than seizures. Thus, there is an urgent need for the development of better therapeutic options that more comprehensively address the multiple aspects of Dravet syndrome.

Over 80% of patients with Dravet syndrome carry mutations in the SCN1A gene, which disable voltage-gated sodium channels located on inhibitory neurons and lead to global hyperexcitability through disinhibition. The genetic aetiology of Dravet syndrome has been harnessed to develop mouse models that employ heterozygous deletion of Scn1a (Scn1a+/-) and recapitulate many features of the condition, including hyperthermia-induced seizures, spontaneous seizures, premature death and associated comorbidities. Additionally, the Scn1a+/- mouse model has been pharmacologically validated for the preclinical screening of potential therapeutics.

In Chapters 2 and 3 we utilise a strategy that exploits the background strain-dependent determination of the epilepsy phenotype in a Dravet syndrome mouse model to uncover potential novel drug targets. Scn1a+/- mice on a 129S6/SvEvTac (129) background display a normal phenotype, whereas forward-crossing 129.Scn1a+/- mice with C57BL/6 mice to generate Scn1a+/- mice on a mixed F1(129 x C57BL/6) background strain uncovers the severe seizure phenotype.

This implies the presence of phenotype-modifying genes (genetic modifiers) that may serve as therapeutic targets. By comparing gene transcription and functional protein levels between the seizure-resistant and seizure-susceptible strains of mice, differentially-expressed molecular targets are identified as potential genetic modifiers. Then, the phenotype-modifying ability of these

xii candidate targets can be determined by evaluating their impact on the epilepsy phenotypes in the

Scn1a+/- mouse model through pharmacological or genetic manipulation; thus, providing causal evidence for their potential therapeutic significance. A preliminary RNA-seq analysis that compared the transcriptome expression between the seizure-resistant and seizure-susceptible strains highlighted that the genes encoding the cannabinoid type 1 (CB1) receptor (Cnr1) and G protein-coupled receptor 55 (Gpr55; Gpr55) might be genetic modifiers, as mice on the seizure- susceptible background strain expressed less Cnr1 mRNA but more Gpr55 mRNA compared to mice on the seizure-resistant background strain.

Drugs targeting the endocannabinoid and Gpr55-lysophosphatidylinositol (LPI) systems hold potential for treating Dravet syndrome as these systems are known to regulate neuronal excitability. Our investigation of the endocannabinoid system focussed on the CB1 receptor, which is the main cannabinoid receptor of the brain, and its two main endogenous ligands, anandamide and 2-AG. While the Gpr55-LPI system investigations centred around the Gpr55 receptor and its most selective, abundant, potent and efficacious endogenous ligands, 2-arachidonoyl LPI and 1- stearoyl LPI. Enhanced endocannabinoid signalling is typically associated with reduced neuronal excitability, whereas increased Gpr55 signalling was shown to enhance neuronal excitability.

Moreover, potentiating endocannabinoid signalling is known to have anticonvulsant properties in conventional preclinical seizure models, while there is accumulating evidence to suggest Gpr55 antagonism may be anticonvulsant. Yet, these systems have not been specifically investigated for the treatment of paediatric epilepsies like Dravet syndrome. Therefore, this thesis utilises a Scn1a+/- mouse model of Dravet syndrome to explore the endocannabinoid and Gpr55-LPI systems for novel druggable targets that may have the potential to address not only the seizure phenotypes, but also the behavioural abnormalities associated with Dravet syndrome.

xiii

The work in Chapter 2 investigates the endocannabinoid system of a Dravet syndrome mouse model to determine its therapeutic potential. We confirmed that seizure-susceptible mice had reduced hippocampal expression of CB1 receptor mRNA and protein with quantitative PCR

(qPCR) and western blotting, respectively. We also measured lower hippocampal concentrations of the endogenous cannabinoid 2-AG in seizure-susceptible mice compared to seizure-resistant mice using high-performance liquid chromatography tandem mass spectrometry (HPLC/MS/MS).

Moreover, when we examined the hippocampal expression of 2-AG’s metabolising enzymes with qPCR, we identified that mRNA expression of diacylglycerol lipase beta (Dagl-β) was selectively increased in the phenotypic F1.Scn1a+/- mice. Dagl-β is the enzyme responsible for synthesising

2-AG in the brain’s resident immune microglial cells, which highlights it as a potential therapeutic target that warrants further investigation. To explore the functional significance of the observed endocannabinoid system alterations, we investigated the phenotype-modifying potential of pharmacologically potentiating endocannabinoid signalling using ABX-1431, a monoacylglycerol lipase (MAGL) inhibitor that increases 2-AG levels, and with GAT229, a CB1 receptor positive allosteric modulator. We demonstrated that both methods of enhancing endocannabinoid signalling had anticonvulsant actions against hyperthermia-induced seizures. The work in Chapter

2 supports that augmenting the endocannabinoid system, putatively through CB1 receptors, has anticonvulsant effects and warrants the further examination of targeting the 2-AG-CB1 receptor axis in the treatment of Dravet syndrome.

The research in Chapter 3 explores the expression and therapeutic captivity of the Gpr55-

LPI system in a Dravet syndrome mouse model. We confirmed with qPCR that hippocampal mRNA expression of Gpr55 was significantly greater in mice on the seizure-susceptible background strain, then demonstrated with HPLC/MS/MS that hippocampal concentrations of the

xiv

LPI analogues were overall less in the seizure-susceptible strain of mice. Moreover, we observed that heterozygous deletion of Scn1a reduced hippocampal 2-arachidonoyl LPI concentrations in mice specifically on the seizure-resistant but not seizure-susceptible background strain, suggesting potential maladaptation in seizure-susceptible mice. Furthermore, examination of LPI synthetic and degradative enzymes with qPCR revealed that heterozygous deletion of Scn1a downregulated the mRNA expression of membrane Bound O-Acyltransferase Domain Containing 7 (Mboat7), which has been associated with epilepsy and behavioural impairments, and requires further investigation. As selective and brain penetrant Gpr55 antagonists do not exist, Chapter 3 then sought to genetically validate whether downregulation of Gpr55 could influence the seizure and mortality phenotypes of the Dravet syndrome mouse model. We demonstrated that heterozygous deletion of Gpr55 was anticonvulsant against hyperthermia-induced seizure and rescued the

+/- premature mortality in the phenotypic Scn1a mice. Therefore, the data presented in Chapter 3 supports Gpr55 as a novel anticonvulsant drug target to pursue in the treatment of Dravet syndrome.

The behavioural comorbidities associated with Dravet syndrome are a detrimental aspect of the condition that are critically unaddressed with current treatment strategies. Therefore, the research in Chapter 4 expands the utility of the Scn1atm1Kea mouse model by characterising behavioural impairments in order to provide a screening platform for the development of drugs that specifically address this aspect of Dravet syndrome. The behavioural phenotypes of other mouse models of Dravet syndrome have been characterised only during adulthood. The work presented in Chapter 4 advances on the other models by characterizing for the first time the behavioural phenotype during adolescence, which better models paediatric Dravet syndrome patients. The behaviour of adolescent Scn1a+/- and WT mice were compared across several

xv behavioural domains that as assessed motor function (open field test), sociability and social recognition memory (three-chambered social preference and social interaction tests), memory function (novel object recognition, Barnes maze, fear conditioning paradigm), anxiety-related behaviour (elevated plus maze and open field thigmotaxis), startle reflex and sensorimotor gating

(prepulse inhibition of startle tests), and repetitive compulsive behaviour (marble burying test).

We demonstrated that adolescent Dravet syndrome mice displayed increased anxiety-related thigmotactic behaviour, atypical fear expression, blunted acoustic startle responses, and impaired social recognition and spatial memory. This work has established a behavioural assessment platform to investigate the therapeutic potential of pharmacological agents, and to assess the impact of any genetic modifier candidates. The absence of additional and more severe behavioural modalities that have previously been reported for adult Dravet syndrome mice suggests that cognitive decline may be progressive, further supporting the necessity for an early-intervention platform.

Chapter 5 summarises the main findings of this thesis and discusses implications of these findings. Furthermore, we suggest future directions to expand on the work presented in this thesis in order to assist with the development of effective therapies for the management of all aspects of

Dravet syndrome.

Chapter 1

Introduction

1. Introduction to cannabis and cannabinoids

‘Cannabis’ is a colloquial term referring to plants from the Cannabis genus, which are commonly treated as a single, highly diverse species. Historically, cannabis has evolved alongside humans for millennia with human interest in the plant arising due to its material, spiritual, recreational and medicinal utilities (Bonini et al., 2018). Cannabis is most commonly known for its psychoactive use and abuse as consumption of cannabis is associated with feelings of relaxation, euphoria, spontaneous laughter, increased sociability and increased appetite. However, cannabis can also induce dissociation, paranoia, psychosis, cognitive impairments and lead to dependence.

Due to its propensity to be abused, cannabis use was prohibited in the early 20th century (Bonini et al., 2018). Despite its restriction, cannabis remains the most commonly consumed illicit drug worldwide (Peacock et al., 2018). The legalisation of cannabis for medical and recreational use has been increasing worldwide and attitudes towards cannabis are becoming more positive

(Carliner et al., 2017). This has also contributed to the increase in research regarding the medicinal benefits of cannabis, especially for epilepsy.

Figure 1. Phytocannabinoid biosynthesis and degradation routes and products for the 11 types of phytocannabinoids, from Berman et al. (2018).

The 11 subclasses are (−)-Δ9-trans-tetrahydrocannabinol (Δ9-THC), (−)-Δ8-trans- tetrahydrocannabinol (Δ8-THC), cannabigerol (CBG), cannabichromene (CBC), cannabidiol (CBD), cannabinodiol (CBND), cannabielsoin (CBE), cannabicyclol (CBL), cannabinol (CBN), cannabitriol (CBT), and miscellaneous types (no structure). The most prevalent acid components are presented for each type of phytocannabinoid. The letters represent the structural components of CBGA; (A) a resorcinol (benzene-1,3-diol) ring, (B) a terpene moiety, (C) an alkyl chain, and (D) a carboxylic group. The numbers represent the conversion methods.

Investigations into the chemical composition of the cannabis plant has identified over 560 chemical constituents, with 120 being classified as plant-derived cannabinoids

(phytocannabinoids) based on structural similarity to (−)-Δ9-trans-tetrahydrocannabinol (Δ9-

THC), the main psychoactive component in cannabis (ElSohly et al., 2017). Phytocannabinoids are defined as products isolated from the cannabis plant that exhibit the typical C21 terpenophenolic skeleton, including derivatives and transformation products. Phytocannabinoids have been categorised into 11 general types (Figure 1): (−)-Δ9-trans-tetrahydrocannabinol (Δ9-

THC), (−)-Δ8-trans-tetrahydrocannabinol (Δ8-THC), cannabigerol (CBG), cannabichromene

(CBC), cannabidiol (CBD), cannabinodiol (CBND), cannabielsoin (CBE), cannabicyclol (CBL), cannabinol (CBN), cannabitriol (CBT), and other miscellaneous types (ElSohly et al., 2017).

Identification, separation and synthesis of the phytocannabinoids has allowed investigations into their pharmacology, with Δ9-THC and CBD being the best characterised. Δ9-THC is the most abundant phytocannabinoid in street cannabis and is responsible for the profound psychotropic effects that drive recreational use. Although it displays therapeutic benefits, its intoxicating effects render it undesirable for clinical use (Dos Santos et al., 2014; Swift et al., 2013). However, CBD and other phytocannabinoids are more desirable therapeutic candidates as they lack psychotropic effects and display improved toxicity while maintaining therapeutic benefits (Mechoulam et al.,

2007).

2. Cannabis-based treatments for epilepsy

Historically, the earliest use of cannabis for medicinal purposes was documented to 2700

BCE in Asia, with reports of cannabis improving seizure control (Friedman & Sirven, 2017). The prohibition of cannabis during the early 20th century significantly hindered research into its therapeutic potential. Recently however, research into the medicinal benefits of cannabis and

5 cannabinoids has reignited, primarily due to the emergence of anecdotal reports claiming dramatic benefits of cannabis-based products for children with treatment-resistant epilepsy (Porter &

Jacobson, 2013; Press, Knupp & Chapman, 2015; Suraev et al., 2018; Suraev et al., 2017). One striking example was a case study reporting that the use of cannabis extract oil, high in CBD, drastically reduced the seizure incidence of a child with Dravet syndrome from “50 convulsive seizures per day to now 2–3 nocturnal convulsions per month” (Maa & Figi, 2014).

These anecdotal reports identified that CBD was a commonality in these cannabis products; which led to preclinical and clinical investigation of CBD in severe refractory epilepsies. CBD was shown to significantly reduce seizures when administered in conjunction with conventional anticonvulsant medications in three phase III trials, in addition to demonstrating an acceptable safety profile (Devinsky et al., 2017; Devinsky et al., 2019; Devinsky et al., 2018b). Consequently,

CBD has been approved by US and European drug approval agencies for the treatment of refractory childhood epilepsies, including Dravet syndrome and Lennox-Gastaut syndrome. The exact mechanisms by which CBD is improving seizure control has not been ascertained and is likely multifaceted due to CBD’s multimodal nature. One likely mechanism is the potentiation of concurrently administered conventional anticonvulsants. In patients with Dravet syndrome, CBD was shown to elevate plasma concentrations of other anticonvulsants and their metabolites, particularly with clobazam, presumedly via inhibition of their metabolising enzymes cytochrome

P450s (Geffrey et al., 2015; Morrison et al., 2019). Work from our own lab supported a pharmacokinetic interaction between CBD and clobazam, while also demonstrating a pharmacodynamic interaction through GABAA receptors in a mouse model of Dravet syndrome

(Anderson et al., 2019a). However, further evaluation of the clinical trials assessing adjuvant CBD in refractory paediatric epilepsies suggest that CBD has anticonvulsant activity independent of its

6 interaction with clobazam (Bialer & Perucca, 2020; Devinsky et al., 2020; Savage et al., 2020).

CBD has also demonstrated direct anticonvulsant actions through increasing inhibitory and decreasing excitatory neuronal activity in a Dravet syndrome patient-derived induced pluripotent stem cell (IPSC) model (Sun & Dolmetsch, 2018). Additionally, CBD’s anticonvulsant activity in a mouse model of Dravet syndrome was attributed, at least partially, to inhibition of G-protein receptor 55 (GPR55) (Anderson et al., 2019a; Kaplan et al., 2017). CBD has additional molecular targets that could be contributing to its anticonvulsant efficacy (Table 1) but these yet to be confirmed for Dravet syndrome. Based on the intersection between CBD concentrations observed in patients and the concentrations at which CBD’s pharmacological engagements are observed, functional antagonism of GPR55 receptors, desensitization of transient receptor potential cation channel subfamily V member 1 (TRPV1) receptors and inhibition of adenosine transport are

CBD’s mechanisms of action that have been proposed to be responsible for the control of epileptic seizures in humans (Gray & Whalley, 2020). For example, CBD’s anticonvulsant activity in a mouse model of generalised seizures was attributed to its interaction with Trpv1 receptors as CBD exhibited a blunted anticonvulsant response in Trpv1 knock-out mice (Gray et al., 2019). Despite

CBD’s efficacy in some cases, not all patients responded to the treatment. Furthering our understanding of the pharmacology behind CBD’s anticonvulsant molecular mechanisms would better inform and direct the development of future cannabis-based anticonvulsants.

CBD is only one of many phytocannabinoids present in the cannabis plant and there is emerging preclinical evidence to show that some of these phytocannabinoids, such as cannabidivarin (CBDV), also elicit anticonvulsant actions (Hill et al., 2012; Huizenga, Sepulveda-

Rodriguez & Forcelli, 2019). Additionally, there is evidence to suggest that the phytocannabinoids may produce improved anticonvulsant actions in combination rather than individually (Hill et al.,

2013; Pamplona, da Silva & Coan, 2018). These findings suggest that the cannabis plant could hold great therapeutic potential for the treatment of epilepsy and warrants thorough investigation if cannabinoid pharmacology. Many of the cannabinoids elicit their pharmacological actions through the endogenous cannabinoid (endocannabinoid) system. Therefore, there is also a need to improve our understanding of the endocannabinoid system and its role in the aetiology of epilepsy, as this may also yield druggable molecular targets for improved anticonvulsant therapeutics.

3. The endocannabinoid system

The endocannabinoid system has been implicated in many physiological systems; including energy metabolism, cardiovascular and reproductive functions, inflammation, glaucoma, cancer, liver and musculoskeletal disorders, appetite, learning and memory, anxiety, depression, schizophrenia, stroke, multiple sclerosis, neurodegeneration, addiction, and epilepsy (Zou &

Kumar, 2018). The endocannabinoid system formally consists of two main receptors, cannabinoid type 1 (CB1) and cannabinoid type 2 (CB2) receptors, and two main endogenous ligands

(endocannabinoids), N-arachidonoyl-ethanolamine (anandamide) and 2-arachidonoylglycerol (2-

AG). However, there are several other receptors, including G protein-coupled receptors, ion channel and nuclear receptors, that also interact with endocannabinoids and thus might be considered part of the broader endocannabinoid system. Given the current thesis is devoted to the study of childhood epilepsy, our overview of the endocannabinoid system will largely be confined to the central nervous system (CNS).

3.1 Cannabinoid receptors 1 and 2

CB1 and CB2 receptors are both seven-transmembrane G protein-coupled receptors that are encoded by the genes CNR1 and CNR2, respectively. CB1 receptors are most highly expressed

8 in the brain but also are present in adipose tissue, and peripheral organs at varying levels

(Fagerberg et al., 2014). Within the CNS, CB1 receptors are the most abundant G protein-coupled receptor, with highest expression in the basal ganglia, hippocampus, cortex and cerebellum in humans and rodents (Howlett & Abood, 2017).

In the brain, CB1 receptors almost always present on presynaptic GABAergic interneuron synapses while only being found at low density on glutamatergic synapses (Howlett & Abood,

2017; Kano, 2014; Kawamura et al., 2006). Moreover, CB1 receptors are almost exclusively expressed on cholecystokinin-expressing interneurons in relation to parvalbumin-, calretinin- and calbindin-expressing interneurons (Marsicano & Lutz, 1999). Due to this selectivity, despite the higher expression at GABAergic synapses, CB1 receptor-positive interneurons represent a smaller portion of all GABAergic interneurons (10-20%) and therefore there may be more CB1 receptor- sensitive ea lexcitatory synapses than CB1 receptor-sensitive GABAergic synatpses in the brain

(Katona, 2015). There is some evidence to show CB1 receptors are also present at much lower but still functionally-relevant levels on astrocytes, but there is some controversy around this (Oliveira da Cruz et al., 2016; Stella, 2010). CB2 receptors, on the other hand are mainly expressed peripherally in organs associated with the immune system (Fagerberg et al., 2014). The expression of CB2 receptors in the CNS remains controversial with conflicting evidence regarding its low expression on neurons and/or microglia (Chen et al., 2017; Li & Kim, 2017).

Figure 2. The synthesis and metabolism of anandamide and 2-AG, adapted from Parsons and Hurd (2015). Endocannabinoids are synthesised in the post-synaptic neuron in response to activation. Anandamide is synthesised from N-arachidonoyl-phosphatidylethanolamine (NAPE) via phospholipase D (NAPE-PLD), while 2-AG is hydrolysed from diacylglycerol (DAG) via the α and β DAG lipase (DAGL) isoforms. Endocannabinoids travel in a retrograde manner across the synaptic cleft to stimulate CB1 receptors located on presynaptic GABAergic or glutamatergic neurons. Endocannabinoid signalling is terminated with the degradation of the endocannabinoids after cellular uptake. 2-AG is hydrolysed predominantly with monoacylglycerol lipase (MAGL) but also with α/β-hydrolase domain-6 (ABHD6) into glycerol and arachidonic acid (AA). Whereas, anandamide is hydrolysed by fatty acid amide hydrolase (FAAH) into ethanolamine and AA.

3.2 Endocannabinoids

The arachidonic acid derivatives anandamide and 2-AG are the best characterised endocannabinoids. Anandamide was identified in 1992 and was named based on the Sanskrit word for ‘bliss’ (Devane et al., 1992). When assessed pharmacologically, anandamide did not replicate all the effects of Δ9-THC. This prompted explorations for other endocannabinoids, which led to

10 the discovery of 2-AG in 1995 (Mechoulam et al., 1995). A schematic depiction of endocannabinoid synthesis and metabolism can be found in Figure 2. Within the CNS, endocannabinoid synthesis occurs in the postsynaptic neuron, in an ‘on-demand’ or ‘activity- dependent’ manner as it is triggered by increased intracellular Ca2+ concentrations (Di Marzo et al., 1994). Anandamide is synthesized from N-arachidonoyl phosphatidylethanolamine (NAPE) through cleavage by N-acyl phosphatidylethanolamine phospholipase D (NAPE-PLD) (Di Marzo et al., 1994). While 2-AG is synthesized from diacylglycerol (DAG) by the diacylglycerol lipases

(DAGLs), DAGL-α and DAGL-β (Murataeva, Straiker & Mackie, 2014). Although both isoforms synthesise 2-AG, the α isoform is the predominant synthesiser as the β isoform is predominantly expressed on microglia (Viader et al., 2016), and Dagl-β knock-out mice produce normal global

2-AG concentrations and exhibit normal endocannabinoid signalling (Tanimura et al., 2010;

Wilkerson et al., 2016).

Following intracellular synthesis, the endocannabinoids pass through the plasma membrane with evidence to support both passive diffusion and active transport (Chicca et al.,

2012; Fowler, 2013; Nicolussi & Gertsch, 2015). Endocannabinoids then undergo ‘retrograde signalling’, whereby they travel back across the synapse, at least partially transported by protein fatty-acid binding protein 5 (FABP5) (Haj-Dahmane et al., 2018), to bind to CB1 receptors located on the presynaptic neuron. Afterwards, the endocannabinoids enter the presynaptic cell where they are rapidly catabolized by several intracellular enzymes. The degradation of the two endocannabinoids is also achieved via distinct pathways. Monoacylglycerol lipase (MAGL) is the enzyme responsible for ~85% of 2-AG’s hydrolysis into arachidonic acid and glycerol, while α/β- hydrolase-6 (ABHD6) and α/β-hydrolase-12 (ABHD12) account for ~9% and ~4% of the degradation, respectively (Blankman, Simon & Cravatt, 2007). Anandamide on the other hand is

11 hydrolysed predominantly by fatty acid amine hydrolase (FAAH) into arachidonic acid and ethanolamine (Cravatt et al., 1996). FABP5 has the additional function of assisting anandamide’s degradation due to intracellular shuttling of anandamide to FAAH (Bjorklund et al., 2014;

Kaczocha et al., 2014).

Anandamide and 2-AG display sophisticated selectivity in their regulation of several brain functions. For example, through the use of selective gene-deletion mouse models and pharmacological studies, it has been established that 2-AG is the main endocannabinoid responsible for the retrograde suppression of neural activity and synaptic plasticity, while anandamide serves a more “fine-tuning” function (Luchicchi & Pistis, 2012). For this reason, 2-

AG is sometimes referred to as the “main cannabinoid”. This notion is supported by 2-AG being present in concentrations that are orders of magnitude more abundant than anandamide; along with

2-AG displaying full agonism at CB1 receptors, while anandamide is a partial agonist at

CB1 receptors (Sugiura et al., 2006).

Additional to the endocannabinoids exist structurally related fatty acids amides, including oleoylethanolamine (OEA) and palmitoylethanolamide (PEA), which can also influence neuronal activity. PEA, for example, has been shown to increase concentrations and efficacy of anandamide and 2-AG, while also displaying intrinsic anticonvulsant properties (Iannotti, Di Marzo &

Petrosino, 2016; Petrosino et al., 2016; Post et al., 2018).

3.3 CB1-mediated signalling

Both CB1 and CB2 receptors predominantly couple to the Gαi/o proteins (Zou & Kumar,

2018). CB1 receptors, however, have also been shown to couple to Gαz, Gαq/11, and Gα12/13 in certain cell types and under certain conditions (Diez-Alarcia et al., 2016; Prather et al., 2000).

Cannabinoid receptor signalling mediated by Gαi/o proteins leads to inhibition of adenylyl cyclase, which supresses cellular cyclic adenosine monophosphate (cAMP) levels and consequently the activity of protein kinase A (Howlett et al., 2002). In addition, Gαi/o is linked to activation of inwardly-rectifying potassium channels (GIRK) or mitogen-activated protein kinase signalling cascades (Howlett et al., 2002). Moreover, presynaptic CB1 receptors coupled to Gαi/o inhibit N- type and P/Q type voltage-gated calcium channels. Overall, these actions reduce membrane excitability, which in turn inhibits the release of neurotransmitters such glutamate and GABA into the synapse (Farrell & Soltesz, 2019; Soltesz et al., 2015).

Interestingly, CB1 receptors are more efficiently coupled to G proteins at glutamatergic synapses than at GABAergic synapses (Steindel et al., 2013), implying that despite the receptor expression data demonstrating greater CB1 receptor expression on GABAergic synapses, CB1 receptor-mediated control over glutamatergic synapses may be stronger. This is supported by numerous accounts where stimulation of CB1 receptors resulted in suppression of excitatory transmission in the hippocampus (Kawamura et al., 2006; Sugaya et al., 2016; Takahashi &

Castillo, 2006). Auxiliary to GαI/o-mediated signalling, cannabinoid receptors activation also associate with β-arrestins 1 and 2, which divert signalling along β-arrestin-mediated pathways and are responsible for desensitisation and internalisation of the receptors (Zou & Kumar, 2018).

Interestingly, CB1 receptors exhibited biased signalling patterns dependent on the ligand bound

(Diez-Alarcia et al., 2016). The ligand-specific biased signalling, in conjunction with the varied cellular and regional expression of cannabinoid receptors, demonstrates how the endocannabinoid system can elicit specific regulation and homeostasis of multiple biological functions.

As activation of CB1 receptors reduce neuronal excitability and consequently neurotransmitter release in an activity-dependent manner, the endocannabinoid system essentially

13 functions as a “circuit-breaker” in the brain, regulating and dampening excess neuronal activity where needed. The endocannabinoid system has been shown to mediate multiple forms of synaptic plasticity that may explain its role in controlling learned behaviour, memory function and neuroadaptive changes that might occur as a result of disease including brain hyperexcitability observed in epilepsy. There are multiple forms of synaptic plasticity that have been characterised to be mediated by endocannabinoid-mediated retrograde signalling in response to different patterns of neural stimulation. The first type is depolarization-induced suppression of inhibition/excitation

(DSI/DSE), which is the transient suppression of neurotransmitter release in response to strong neuronal activation which suppresses inhibitory or excitatory signalling dependent on the synapse

(Kreitzer & Regehr, 2001; Pitler & Alger, 1992). Interestingly, inhibitory synapses are more sensitive to depolarization-induced suppression of synaptic transmission than excitatory synapses

(Ohno-Shosaku et al., 2002). Other forms of retrograde signalling are short-term depression, which is also referred to as metabotropic-induced suppression of inhibition/excitation (MSI/MSE) or synaptically-evoked suppression of inhibition/excitation (SSI/SSE) (Kano et al., 2009). Finally, endocannabinoids also mediate long-term depression, which is a long-lasting inhibition of synaptic plasticity at both excitatory and inhibitory synapses (Chevaleyre & Castillo, 2003; Gerdeman,

Ronesi & Lovinger, 2002). This neural-regulating function of CB1-mediated signalling suggests it may be useful in dampening the neuronal hyperactivity elicited by epileptic brains.

3.4 G protein-coupled receptor 55 (GPR55)

As mentioned earlier, there are additional receptors to CB1 and CB2 receptors that show cannabinoid reactivity. GPR55 is the receptor with the most evidence to support its role in the endocannabinoid system and will be discussed in detail. GPR55 was originally referred to as the third cannabinoid receptor due to its affinity and activation by cannabinoids with no known

14 endogenous ligand at the time (Ryberg et al., 2007). Interestingly, GPR55 binds to cannabinoids despite only sharing 13.5% and 14.4% sequence homology with CB1 and CB2 receptors, respectively (Ross, 2009). Additionally, the formation of CB1-GPR55 heteromers supports a strong interaction between CB1 and GPR55 receptors and further implicates a place for GPR55 in the broader endocannabinoid system (Kargl et al., 2012; Marichal-Cancino et al., 2013; Martinez-

Pinilla et al., 2014).

However, GPR55 left orphan status with the discovery of its endogenous ligand, the lipid signalling molecule L-α-lysophosphatidylinositol (LPI) (Oka et al., 2007; Oka et al., 2009), which does not react at cannabinoid receptors (Kapur et al., 2009). This discovery established the GPR55-

LPI system, which has since been implicated in a number of pathologies. The peripheral physiological functions of GPR55 receptors have been more extensively studied, with GPR55 being investigated as a target for vasculature health, metabolism, gastrointestinal function, bone health, inflammation, cancer and pain (Alhouayek, Masquelier & Muccioli, 2018; Simcocks et al.,

2018). Centrally, GPR55 is currently being investigated for its function in anxiety, Parkinson’s disease and pain (Marichal-Cancino et al., 2017).

GPR55 has also been shown to be activated by the “endocannabinoid-like” fatty acids PEA and OEA, and by various peptides, most notably pituitary adenylate cyclase activating peptide-27

(PACAP-27) (Foster et al., 2019; Ryberg et al., 2007). However, the full implications and physiological functions of these endogenous ligands at this receptor are unknown as our understanding of GPR55’s function and its role in various physiological processes are still developing. Unsurprisingly, both Δ9-THC and CBD have affinity for the GPR55 receptor, but while Δ9-THC activates GPR55, CBD was an antagonist at GPR55 (Lauckner et al., 2008; Ryberg et al., 2007; Whyte et al., 2009). Moreover, there are also numerous synthetic compounds that

15 modulate GPR55, including CBD derivatives and synthetic cannabinoid agonists (Simcocks et al.,

2018). Although selective agonists and antagonists have been developed for GPR55, they do not readily cross the blood brain barrier, which further limits investigation of GPR55’s CNS function

(Heynen-Genel et al., 2010; Kargl et al., 2012; Kotsikorou et al., 2013).

GPR55 receptors have been shown to couple to Gα12/13 and/or Gαq proteins in a ligand- dependent manner and activate intracellular cascades that have functionally opposite outcomes to

CB1/CB2 receptors (Henstridge et al., 2009; Lauckner et al., 2008; Ryberg et al., 2007).

Stimulation of the Gα12/13 subunit activates the Ras homolog family member A - Rho-associated protein kinase (RhoA-ROCK) pathway. This stimulates the phosphorylation of p38 mitogen- activated protein kinase (MAPK) and then subsequently increases phosphorylation of activating transcription factor 2 (ATF-2). ROCK also activates phospholipase C (PLC), which increases intracellular Ca2+ concentrations and triggers Nuclear factor of activated T-cells (NFAT)-induced nuclear translocation (Henstridge et al., 2009; Lauckner et al., 2008; Oka et al., 2010; Ryberg et al., 2007). Alternatively, Gαq-mediated PLC activation results in the release of diacylglycerol and

Ca2+, which result in phosphorylation of protein kinase C and extracellular signal-regulated kinases

(ERK1/2) and subsequently activates cAMP-response element binding protein (CREB) and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-ĸβ) (Henstridge et al., 2009; Lauckner et al., 2008). Moreover, GPR55 also exhibits β-arrestin-mediated receptor internalization (Kapur et al., 2009). Generally, stimulation of GPR55 and activation of these G-protein-mediated pathways ultimately increase cell excitability and alter gene expression.

Recently, there has been evidence showing that GPR55 can modulate neuronal excitability, suggesting that GPR55 may hold therapeutic potential in the treatment of epilepsy. Peripherally,

GPR55 receptors are located in the stomach, intestines, spleen and bone marrow; while in the

16 brain, GPR55 is expressed in the basal ganglia, hippocampus, hypothalamus, frontal cortex and cerebellum (Henstridge et al., 2011; Ryberg et al., 2007; Wu et al., 2013). Centrally, GPR55 has been shown to be expressed on both neurons and glia, with hippocampal GPR55 located presynaptically on glutamatergic neurons (Kallendrusch et al., 2013; Ryberg et al., 2007;

Sylantyev et al., 2013). The sub-cellular locations or knowledge of GPR55 receptor distribution on specific neuron types in other regions of the brain has not yet been characterised.

Based on GPR55 distribution in the hippocampus and its activation resulting in increased cell excitability, it is consistent that GPR55 activation promoted neural excitability and glutamate release in the hippocampus (Hurst et al., 2017; Kramar et al., 2017; Sylantyev et al., 2013). In contrast, PEA enhanced GABAergic transmission in the striatum in a GPR55-dependent manner, as determined with electrophysiological recordings from mouse brain slices (Musella et al., 2017).

This contrasting effect may be due to GPR55 being expressed on GABAergic neurons in the striatum but is yet to be investigated and confirmed. The functional outcome of GPR55 activation in other brain regions is yet to be determined, but GPR55 does not appear to be a receptor critical for neuronal function based on Gpr55 deletion (Gpr55-/-) mouse models. Gpr55-/- mice exhibited intact short-term and long-term synaptic plasticity (Hurst et al., 2017; Wu et al., 2013). Moreover, when assessed in a comprehensive array of behavioural tests, Gpr55-/- mice exhibited reduced locomotor activity and metabolism, while displaying comparable performance in cognitive and mood-related behaviours (Bjursell et al., 2016; Wu et al., 2013).

Despite GPR55 not displaying a critical role in neural functioning, it may still hold therapeutic potential as the GPR55-LPI system may be altered in disease states and because seizure activity can be attenuated by altering metabolism and mediating inflammation, other GPR55- mediated functions (Clanton et al., 2017; Dey et al., 2016; Korff & Dale, 2017; Rho, 2017;

Simeone, Simeone & Rho, 2017). Moreover, CBD was recently demonstrated to have anticonvulsant activity against multiple seizure types in two mouse models of Dravet syndrome, and CBD’s anticonvulsant efficacy was suggested attributed, at least partly, to antagonism of

GPR55 (Anderson et al., 2019a; Kaplan et al., 2017). Hence, GPR55 may be contributing to the endocannabinoid system’s control of neurotransmission and it may be implicated in the pathophysiology of epilepsy. Therefore, GPR55 will be investigated as potential anticonvulsant target for Dravet syndrome in this thesis.

4. Targeting the endocannabinoid system for epilepsy

Due to the endocannabinoid system functioning as a regulator of neuronal transmission, especially in an activity-dependent manner, it is not surprising that the endocannabinoid system could play a role in the pathophysiology and subsequently the treatment of epilepsy disorders.

There is evidence to suggest pathophysiological reorganisation of the endocannabinoid system in epilepsy conditions (De Caro et al., 2017; Katona, 2015). For example, reduced mRNA expression of the 2-AG synthesising DAGL- was observed in post-mortem hippocampal sections from patients with temporal lobe epilepsy, suggesting patients had reduced 2-AG concentrations

(Romigi et al., 2010). Additionally, temporal lobe epilepsy patients also exhibited significantly lower anandamide concentrations in cerebrospinal fluid, and significantly downregulated hippocampal CB1 receptor mRNA, specifically in glutamatergic synapses (Ludanyi et al., 2008).

Moreover, variants of the gene encoding DAGL- were found to be significantly associated with seizures and neurodevelopmental disorders (Smith et al., 2017), implicating 2-AG alterations in the pathophysiology of epilepsy conditions.

Animal models of seizure and epilepsy have also provided evidence to suggest that the endocannabinoid system is altered by seizures. Seizure activity appears to upregulate CB1 receptor expression in the hippocampus, as elevated hippocampal CB1 protein expression was reported one year after pilocarpine-induced seizures in one study and 3 h after kainic acid-induced seizures in another study (Bojnik et al., 2012; Wallace et al., 2003). Moreover, hyperthermia-induced seizures also upregulated CB1 receptors in the hippocampus of juvenile rats, and upregulation was shown to be specific to interneurons (Chen et al., 2007; Chen et al., 2003; Feng et al., 2016). Interestingly,

CB1 receptor activation was critical for the adulthood seizure-susceptibility that developed in response to hyperthermia-induced seizures, suggesting that CB1 receptor activation can promote epileptogenesis and potentially be proconvulsant (Chen et al., 2007; Feng et al., 2016). This may be a barricade to the development CB1 receptor-related therapeutics for epilepsy that requires careful consideration and exploration when pursuing cannabinoid-related therapeutics.

Changes in endocannabinoid concentrations in response to seizures seem to be more varied and this may be due to the activity-dependent, transient nature of endocannabinoid signalling. For example, kainic acid-induced seizures have been associated with increased hippocampal concentrations of only anandamide, both anandamide and 2-AG or neither endocannabinoid in rats

(Lerner et al., 2017; Marsicano et al., 2003; Wettschureck et al., 2006). Elevated hippocampal 2-

AG concentrations have also been reported in rats shortly after pilocarpine-induced seizures

(Wallace et al., 2003). Moreover, age also seems to influence the endocannabinoid system’s response to seizure. One study demonstrated age-dependent alteration in endocannabinoid levels in response to kainic acid-induced seizures, with P14 rats exhibiting increased hippocampal anandamide and 2-AG levels, while adult rats had reduced anandamide and 2-AG levels (Fezza et al., 2014). Although this does complicate the story, it also suggests that there may be beneficiary

19 effects of endocannabinoid system modulation specific to certain aged population, i.e. paediatric vs adulthood epilepsy disorders.

The role of CB1 receptors in seizure aetiology has been investigated using both genetic and pharmacological methods. CB1 receptor knock-out mice do not display innate spontaneous seizures (Rowley et al., 2017) but have displayed increased susceptibility to and severity in kainic acid-induced and pilocarpine-induced seizure models (Kow et al., 2014; Marsicano et al., 2003;

Sugaya et al., 2016). Moreover, the specific deletion of CB1 receptors in the hippocampus was sufficient to increase susceptibility to kainic acid-induced seizures, highlighting the integral role of cannabinoid signalling in the hippocampus (Monory et al., 2006). Consistent with the CB1 receptor-mediated “circuit-breaking” function was the finding that conditional deletion of CB1 receptors from glutamatergic but not GABAergic neurons increased seizure susceptibility in a kainic acid-induced seizure model and a kindling model of temporal lobe epilepsy (Monory et al.,

2006; von Ruden et al., 2015b). Interestingly, despite greater CB1 receptor expression at inhibitory synapses, global CB1 receptor deletion exacerbates seizures, suggesting that CB1 receptors seem to have a stronger effect at glutamatergic synapses.

Considering endocannabinoid signalling functions as a “circuit-breaker” in conjunction with the

CB1 receptor deletion studies demonstrating proconvulsant effects, there is considerable evidence to support the anticonvulsant potential of potentiating endocannabinoid signalling. Thus, there is a plethora of preclinical evidence to suggest that modulation of the endocannabinoid system may hold anticonvulsant potential and was reviewed in detail by Rosenberg, Patra and Whalley (2017).

Due to the multi-component nature of the endocannabinoid system, there are numerous pharmacological approaches that could be employed to augment endocannabinoid signalling. The most obvious approach is direct CB1 receptor agonism. In experimental acute models of seizure

20 and chronic models of epileptogenesis, CB1 receptor agonism typically had anticonvulsant effects.

Rosenberg et al. (2017) summarised that CB1 receptor-activating drugs, including Δ9-THC, displayed anticonvulsant actions in 34 of the 50 assessments (68.0%) in acute seizure models such as the maximal electroshock and pentylenetetrazol models (Rosenberg, Patra & Whalley, 2017).

CB1 receptor-activating drugs were also anticonvulsant in 25 of the 34 incidences (73.5%) testing in models of chronic epileptogenesis, including the pilocarpine-induced and amygdala kindling models (Rosenberg, Patra & Whalley, 2017). For example, the CB1 receptor agonist WIN55,212 was anticonvulsant in the maximal electroshock and pentylenetetrazol acute models of seizure and in the pilocarpine model of status epilepticus in rodents (Suleymanova et al., 2016; Vilela et al.,

2013; Wallace et al., 2001). Interestingly, CB1 receptor agonism had mixed effects on seizure outcomes in 4 of the acute seizure assessments and 2 of the chronic seizure models. For example,

WIN55,212 was anticonvulsant at a low dose (0.5mg/kg), but proconvulsant at a higher dose (5 mg/kg) in a kainic acid seizure model (Rudenko et al., 2012). Moreover, there were no significant effects in 7 of the acute seizure model assessments and 5 of the chronic epilepsy assessments

(Rosenberg, Patra & Whalley, 2017).

Although CB1 receptor agonists mostly proved anticonvulsant, there were incidences of

CB1 receptor agonism displaying proconvulsant activity. For example, the selective CB1 receptor agonists WIN55,212 and arachidonyl-2'-chloroethylamide (ACEA), reduced seizure thresholds and increased seizure duration in the pentylenetetrazol seizure model in rats (Vilela et al., 2013).

Likewise, Δ9-THC reduced latency to pentylenetetrazol-induced seizures in pregnant female mice and reduced the threshold to seizure in the 60 Hz electroshock model (de Salas-Quiroga et al.,

2015; Karler & Turkanis, 1980). More recently, synthetic cannabinoid receptor agonists (SCRAs) have become popular drugs of abuse that have been associated with serious toxicity, including

21 seizures (Adamowicz et al., 2017; Louh & Freeman, 2014; Schep et al., 2015). Pharmacological evaluation of these compounds in mice confirmed that they do indeed display intrinsic proconvulsant activity and increase susceptibility to seizures through CB1 receptor agonism

(Funada & Takebayashi-Ohsawa, 2018; Kevin et al., 2019; Malyshevskaya et al., 2017; Wilson et al., 2019).

Due to the potential CB1 receptor-mediated toxicity, additional methods of promoting endocannabinoid signalling for anticonvulsant purposes are being investigated. One approach is to increase endocannabinoid concentrations, which can be achieved through inhibition of anandamide’s and 2-AG’s hydrolysing enzymes, FAAH and MAGL/ABHD6, respectively; along with inhibition of anandamide’s synaptic re-uptake. Again, from the afore-mentioned review by

Rosenberg et al. (2017), pharmacologically increasing endocannabinoid concentrations was generally found to be anticonvulsant, with 9 out of 16 cases showing endocannabinoid-enhancing drugs to have anticonvulsant effects in acute seizure models. Moreover, these agents displayed mixed effects in 2 cases and had no significant effect in 5 cases of the total 16 assessments in acute seizure models, therefore generally proving anticonvulsant (Rosenberg, Patra & Whalley, 2017).

In chronic models of epilepsy, elevating endocannabinoid concentrations was summarised to be anticonvulsant in 7 cases and proconvulsant in 1 of the total 8 assessments (Rosenberg, Patra &

Whalley, 2017). Overall, indirect modulation of CB1 receptors may be a safer approach to augmenting endocannabinoid signalling. Specifically, MAGL inhibitors that elevate 2-AG concentration have been demonstrated to display anticonvulsant effects in pentylenetetrazol, status epilepticus and kindling models of seizure and epilepsy in rodents (Terrone et al., 2018; von Ruden et al., 2015a; Zareie et al., 2018). As the pharmacological activity of MAGL inhibitors are dependent on the presence of 2-AG, MAGL inhibitors theoretically function to enhance the 2-AG-

22 mediated “circuit-breaking” function. Moreover, as 2-AG is only produced on-demand where neural activity is present, MAGL inhibition has the added advantage of eliciting its actions in a spatiotemporally specific and necessary manner.

Figure 3. Cooperation between the GPCRs orthosteric and allosteric binding sites to induce intracellular protein coupling, from Scott & Kendall, 2017. Schematic illustration showing how binding of the allosteric modulator to an allosteric binding site distinct from the orthosteric binding site may influence binding of the orthosteric ligand to its respective site and how both ligands may impact binding of intracellular proteins to their respective binding sites. β-Arrestin binding can involve different isoforms not indicated and other G protein isoforms may also be impacted. Further, some ago-allosteric modulators can function without an orthosteric compound.

More recently, the development of CB1 receptor positive allosteric modulators (PAMs) has opened up another avenue of endocannabinoid system modulation that avoids direct CB1 receptor agonism and is yet to be assessed for its anticonvulsant potential. Allosteric modulators

23 act by binding to an allosteric site, i.e. at a site distinct from the orthosteric site, that theoretically has no intrinsic efficacy but induces conformational changes in the receptor that increases the affinity and/or intrinsic efficacy of an orthosteric ligand (Figure 3; Price et al., 2005).

This pharmacological approach is therapeutically superior to CB1 receptor agonism because, like MAGL inhibitors, it should enhance the endocannabinoid system’s intrinsic “circuit- breaking” function in an on-demand, region specific manner, whilst avoiding adverse off-target effects or overstimulation. This has been demonstrated by CB1 receptor PAMs, ZCZ011 and

GAT229, reducing neuropathic pain in mouse models without inducing cannabimimetic adverse effects (Ignatowska-Jankowska et al., 2015; Slivicki et al., 2018). There are no published studies investigating the anticonvulsant potential of CB1 receptor PAMs, but theoretically these hold anticonvulsant potential. Thus, the anticonvulsant potential of positively modulating the allosteric site of CB1 receptors will be investigated in this thesis.

Interestingly, endocannabinoid system modulators have been assessed only in experimental seizure and epilepsy models that use adult rodents and predominantly in models that induce seizures in healthy animals. Due to the developmental changes that occur in the endocannabinoid system (Meyer, Lee & Gee, 2018) and variation of endocannabinoid signalling caused by disease state (Kaur, Ambwani & Singh, 2016), it is possible that endocannabinoid system modulation could have different effects in different patient populations. Thus, there is a need to assess the therapeutic potential of potentiating endocannabinoid signalling in paediatric models of epilepsy, including Dravet syndrome, that will be explored in this thesis.

5. Epilepsy

Epilepsy is the most common, chronic neurological condition with a prevalence rate of 7.6 per 1,000 persons worldwide (Fiest et al., 2017). “Epilepsy” is defined by the International League

Against Epilepsy (ILAE) as “a disorder of the brain characterized by an enduring predisposition to generate epileptic seizures and by the neurobiologic, cognitive, psychological, and social consequences of this condition” (Fisher et al., 2005). While an “epileptic seizure” is defined as “a transient occurrence of signs and/or symptoms due to abnormal excessive or synchronous neuronal activity in the brain” (Fisher et al., 2005). The definition of “epilepsy” has been updated to be more encompassing and now epilepsy is diagnosed if a condition meets any of the following criteria:

“(1) at least two unprovoked (or reflex) seizures occurring > 24 h apart; (2) one unprovoked (or reflex) seizure and a probability of further seizures similar to the general recurrence risk (at least

60%) after two unprovoked seizures, occurring over the next 10 years; (3) diagnosis of an epilepsy syndrome” (Fisher et al., 2014).

Figure 4. The expanded ILAE 2017 operational classification of seizure types, modified from Fisher et al. (2017b).

Seizures can present in many different forms and are classified based on initial manifestation then typical features observed during the seizure, as outlined in Figure 5 (Fisher et al., 2017a; Fisher et al., 2017b). Seizures can be divided into three categories based on initial manifestation: focal, generalised and unknown. Focal seizures, formerly called partial seizures, originate in neuronal networks limited to part of one cerebral hemisphere, whereas generalised seizures originate from cortical or subcortical regions of both hemispheres. If the source of seizure activity cannot be identified, then it is classified as having unknown onset. Ancillary to onset, seizures are further classified based on features observed during the seizure. For focal seizures, characterisation is divided based on level of awareness, then on motor or non-motor onset.

Whereas for generalised seizures, which are the most commonly observed type of seizure, seizures are divided into motor or non-motor (absence) seizures. Generalised motor seizures can then be categorised as tonic-clonic (formerly referred to as grand mal seizures), myoclonic, atonic or other types that are less common. Tonic-clonic seizures involve bilateral stiffening of limbs during the tonic phase, followed by repetitive, rhythmic convulsions of muscles during the clonic phase.

Myoclonic seizures, on the other hand, consist of sudden, brief movements that can involve contraction of one muscle or several muscles. Whereas, atonic seizures, also referred to as “drop seizures”, involve loss of body tone and can result in falls. Finally, generalised non-motor seizures, which are more commonly referred to as absence seizures and were formerly known as petit mal seizures, manifest as episodes of unresponsiveness or lapse of consciousness and can occur with head nodding or eye blinking.

Seizures conceptually result from transient imbalance between excitatory (glutamatergic) and inhibitory (GABAergic) activity within a neuronal network that causes abnormal enhanced synchrony, resulting in network dysfunction (Fisher et al., 2005). This imbalance in transmission

27 can arise from numerous sources. The aetiologies of epilepsy are grouped by the ILAE into six categories: genetic, structural, infectious, metabolic, immune and unknown (Scheffer et al., 2017), with genetic origins remaining as the leading proposed aetiology (Thomas & Berkovic, 2014).

Epilepsies are classified as being genetic in origin if the condition results directly from a single

(monogenetic) or number of mutations (polygenetic), with genetic mutations being inherited or occurring de novo (Thomas & Berkovic, 2014). Moreover, it should be noted that even epilepsies categorized as acquired, such as those resulting from trauma, stroke, neoplasm, infection or congenital malformations, can have genetic contributions (Thomas & Berkovic, 2014). Majority of the mutations associated with epilepsy disorders occur in genes encoding ion channels, with ion channel gene mutations accounting for approximately 25 to 56% of the mutations identified for paediatric refractory epilepsies (Anwar et al., 2020; Kaplan, Isom & Petrou, 2016; Liu et al., 2018;

Oyrer et al., 2018; Wang et al., 2017a).

6. Ion channels

Ion channels are pore-forming membrane proteins that control the flux of ions across cellular membranes. There are many different types of ion channels, each with specific functions that alter neuronal excitability, including establishing action potentials and maintaining homeostasis. Since ion channels are critical to the excitability of neurons, they appear to play an important role in epileptogenesis. Ion channels can be activated by various stimuli including ligands, temperature or mechanically, and those activated by changes in membrane voltage are characterised as voltage-gated ion channels. Then they are characterised by the ions that flux through the channel, which include sodium ions, potassium ions, calcium ions and protons, and the channels can be selective or non-selective for specific ion types. Of all ion channel types,

28 voltage-gated sodium channels are the biggest contributors to the aetiology of epilepsy (de Lera

Ruiz & Kraus, 2015; Kaplan, Isom & Petrou, 2016).

7. Voltage-gated sodium channels

Voltage-gated sodium channels (NaV) are responsible for the generation and propagation of action potential in excitable nerves, skeletal muscle and cardiac cells. Structurally, NaV channels are a membrane-bound protein complex comprising of one pore-forming -subunit (260 kDa) bound to one or two smaller, auxiliary β-subunits (30-45 kDa) (Figure 5) (Catterall, 2000; Kaplan,

Isom & Petrou, 2016; Meisler & Kearney, 2005; Wang, Ou & Wang, 2017). There are ten different

-subunit proteins (NaV1.1-NaV1.9 and NaVX; encoded by genes SCN1A-SCN5A, SCN8A-

SCN11A and SCN7A, respectively) and 5 β-subunits (β1, β1B, β2-β4; encoded by genes SCN1B-

SCN4B, with β1B being a splice variant of SCN1B). Structurally, each -subunit is comprised of four structurally homologous domains (D1-4), which are composed of six transmembrane  helical segments (S1-S6) (Figure 5). The S4 segment act as a voltage sensor as it contains a high concentration of positively charged arginine residues. When assembled into its 3D confirmation, the -subunit creates an ion-conducting pore, with ion permeability selectivity controlled by the

S5-S6 pore loop of each domain (Catterall, 2000; Kaplan, Isom & Petrou, 2016; Meisler &

Kearney, 2005; Wang, Ou & Wang, 2017). The β-subunits are auxiliary and modulate the kinetics and voltage-dependence of -subunits, while also participating in non-conducting roles, such as channel localisation, cell-to-cell and cell-to-matrix interactions, and modulating pharmacological interactions with compounds (Catterall, 2000; Kaplan, Isom & Petrou, 2016; Meisler & Kearney,

2005; Wang, Ou & Wang, 2017).

Figure 5. A schematic representation of the structure of the α and β subunits of voltage-gated sodium channel, adapted from George (2005). The α-subunit, marked in red, is composed of 4 homologous domains (D1-D4), which themselves are comprised of six transmembrane  helical segments (S1-S6). The S4 segment act as a voltage sensor (shown in black) and ion selectivity is determined by the S5-S6 pore loop of each domain, shown in turquoise. The domains assemble in the membrane to form a Na+-permeable pore. The auxiliary β subunit is shown in blue.

The -subunits of voltage-gated sodium channels that are mostly expressed in the CNS are

NaV1.1, NaV1.2, NaV1.3, NaV1.5, and NaV1.6, while the NaV1.7, NaV1.8, and NaV1.9 -subunits are mainly distributed in the periphery (Wang, Ou & Wang, 2017). Within neurons, NaV channels are expressed ubiquitously throughout neuronal architecture but with higher density at the nodes of Ranvier and axon initial segments, supporting their crucial role in action potential initiation and propagation (Catterall, 2000; Kaplan, Isom & Petrou, 2016; Meisler & Kearney, 2005; Wang, Ou

& Wang, 2017).

Activation of NaV channels are responsible for the initiation of action potentials and subsequently contribute to the stimulation or activation of neurons. When a neuronal membrane is at its normal resting potential, the pore of NaV channels are in a closed state, ready to be opened.

Once the membrane voltage surrounding the NaV channels becomes positive enough to reach the activation threshold potential, NaV channels undergo conformational change, opening the ion pore and allowing an influx of positively-charged sodium ions into the cell. This causes depolarization of the neuronal membrane, and the voltage of the neuronal membrane surrounding the channels becomes more positive and in turn activates neighbouring NaV channels by the same mechanism..

Once enough NaV channels are activated and there is a large influx of sodium ions to surpass the threshold level, an action potential is generated which results in a sharp rise in membrane voltage.

Action potentials travels down the neuronal axons to the axon terminal, where they trigger the release of neurotransmitters. Within milliseconds after activation, NaV channels close and enter an inactivated state. During inactivation, NaV channels are non-responsive to membrane voltage and are responsible for the refractory period of action potentials, during which it is impossible for another action potential to occur (Catterall, 2000; Kaplan, Isom & Petrou, 2016; Meisler &

Kearney, 2005; Wang, Ou & Wang, 2017). As NaV channels play such a crucial role in activating and regulating neuronal excitability, it is no surprise that changes in the function of these channels by mutation or drug action can influence global neuronal activity. As mentioned earlier, mutations in genes encoding NaV channels are prevalent in epilepsy, and mutations in SCN1A, the gene encoding the α1-subunit (Nav1.1), constitutes the greatest number of epilepsy-related mutations identified (Kaplan, Isom & Petrou, 2016; Meisler et al., 2001).

8. SCN1A mutations and NaV1.1 impairment

Most SCN1A mutations identified effect one copy of the gene and render half the NaV1.1 subunits absent or functionally impaired and are associated with a broad spectrum of epilepsy disorders. Within the brain, NaV1.1 is expressed almost ubiquitously with relatively high expression observed in the thalamus, superior colliculus, inferior colliculus, deep cerebellar nuclei,

31 pons, medulla, and spinal cord; whereas lower expression was observed in the hippocampus, cerebral cortex, and cerebellum (Ogiwara et al., 2007). NaV1.1 is predominantly distributed on initial axonal segments of parvalbumin-expressing GABAergic inhibitory neurons (interneurons), with some expression on somatostatin-expressing interneurons, calretinin- expressing interneurons and excitatory neurons (Dutton et al., 2013; Ogiwara et al., 2007). As NaV1.1 channels are located on GABAergic neurons and are responsible for action potential initiation and propagation, NaV1.1 impairment results in impaired sodium currents and neuronal firing, consequently impairing the functioning of GABAergic neurons; as demonstrated by heterozygous deletion of Scn1a in mice

(Kaplan et al., 2017; Rubinstein et al., 2015; Tsai et al., 2015; Yu et al., 2006).

There is evidence to support a genotype-phenotype correlation between the severity or sequence-locations of SCN1A mutations, which determine NaV1.1 protein impairment, and the severity of the epilepsy phenotype observed (Marini et al., 2011). For example, single amino substitutions, which are mostly missense mutations, cause milder NaV1.1 impairment and are more frequently associated with the milder form of epilepsy, generalised epilepsy with febrile seizure plus (GEFS+). Whereas nonsense or frame-shift type mutations, that cause truncation of the

NaV1.1 protein, are prevalent in the severe intractable epileptic encephalopathy, Dravet syndrome

(Catterall, Kalume & Oakley, 2010; Kaplan, Isom & Petrou, 2016; Marini et al., 2007; Meisler et al., 2001). Moreover, mutations in the pore-forming region of NaV1.1, a region integral for channel function, are more prevalent in patients with Dravet syndrome than in patients with GEFS+ (Kanai et al., 2004). Further supporting the genotype-phenotype correlation was the report of an infantile epileptic encephalopathy more severe than Dravet syndrome being linked to a severe gain-of- function mutation of SCN1A that caused depolarization-block and a more profound NaV1.1 impairment (Berecki et al., 2019). However, this genotype-phenotype correlation is not always

32 observed, with the same SCN1A mutations resulting in Dravet syndrome in some patients and

GEFS+ in others (Escayg et al., 2000; Guerrini et al., 2010; Suls et al., 2010). This incomplete penetrance of SCN1A mutations suggests other factors can influence epilepsy severity and highlight potential avenues to pursue in the development of novel treatment strategies for SCN1A- mediated epileptic encephalopathies.

9. Dravet syndrome

SCN1A mutations are the leading cause of Dravet syndrome, previously known as severe myoclonic epilepsy in infants, which is a rare but catastrophic form of intractable infant-onset epilepsy that has an incidence of 1 per 15, 700 to 1 per 45, 800 around the world (Bayat, Hjalgrim

& Moller, 2015; Brunklaus et al., 2012; Ishii et al., 2017; Krueger & Berg, 2015; Wu et al., 2015).

Dravet syndrome typically presents with febrile (generalised and unilateral, clonic or tonic–clonic) seizures within the first year of life in otherwise healthy infants. Seizure types progress to include generalised tonic-clonic, myoclonic, atypical absences, and focal impaired awareness seizures

(Dravet, 2011).

Additional to seizures, delays in cognitive, behavioural and motor development manifest in patients with Dravet syndrome during the second year of life (Dravet, 2011). As Dravet syndrome patients get older, seizures tend to become less frequent and less severe (Genton,

Velizarova & Dravet, 2011) but the behavioural comorbidities can worsen and persist well into adulthood (Battaglia et al., 2016; Olivieri et al., 2016). The observed comorbidities include hyperactivity and attention disorders, motor disorders, intellectual disabilities, speech difficulties, and features of autism spectrum disorder including; asocial, impulsive and perseverative behaviours (Acha et al., 2015; Brunklaus, Dorris & Zuberi, 2011; Genton, Velizarova & Dravet,

2011; Lagae et al., 2018; Olivieri et al., 2016; Sinoo et al., 2019; Takayama et al., 2014; Villeneuve et al., 2014). Patients with Dravet syndrome also exhibit neuropsychological deficits in both visual and auditory sensorimotor processing (Acha et al., 2015; Chieffo et al., 2016), which may precede and contribute to the associated attention, language and cognitive impairments (Brunklaus, Dorris

& Zuberi, 2011; Genton, Velizarova & Dravet, 2011; Sinoo et al., 2019; Turner et al., 2017;

Villeneuve et al., 2014).

The presence of these behavioural comorbidities, especially hyperactivity and inattention symptoms, are stronger predictors of worse quality of life than seizure frequency (Brunklaus,

Dorris & Zuberi, 2011; Lagae et al., 2018; Sinoo et al., 2019; Wolff, Casse-Perrot & Dravet, 2006).

Tragically, Dravet syndrome is also associated with a high rate of childhood mortality

(approximately 15% by the age of 18), with a particularly high incidence of sudden unexpected death in epilepsy (SUDEP) (Cooper et al., 2016; Shmuely et al., 2016).

Over 80% of Dravet syndrome patients carry a mutation in SCN1A with around 90% of these mutations arising de novo, and the majority of the identified mutations were truncating mutations, followed by splice-site or missense mutations (De Jonghe, 2011; Depienne et al., 2009;

Marini et al., 2011). As all these mutations result in NaV1.1 channel malfunction or absence, it suggests that NaV1.1 haploinsufficiency is responsible for the majority of Dravet syndrome cases.

NaV1.1 haploinsufficiency leads to the impaired function of GABAergic neurons (Figure 6), as demonstrated by reduced GABAergic function recorded from post-mortem patient cortical tissue explants (Ruffolo et al., 2018). Moreover, patient‐derived IPSC studies have confirmed that

SCN1A mutations impaired sodium currents and action potentials predominantly in inhibitory neurons; confirming that network hyperexcitability is resulting from disinhibition (Higurashi et al., 2013; Liu et al., 2013; Maeda et al., 2016; Sun et al., 2016).

Imaging studies in Dravet syndrome patients have reported structural abnormalities are observed with variable incidence (11% - 71% cases) and the most commonly reported abnormalities were sclerosis and brain atrophy, predominantly in the hippocampus and cortex

(Brunklaus et al., 2012; Gaily et al., 2013; Perez et al., 2014; Siegler et al., 2005; Striano et al.,

2007). These reports highlight cortical and hippocampal regions as regions of interest in the pathology of Dravet syndrome. Furthermore, an interesting observation made in majority of these studies was that patients initially presented with normal MRIs but then progressed to exhibit structural abnormalities. This progressive pathology is consistent with Dravet syndrome being an epileptic encephalopathy and supports the need for early pharmacological intervention.

Figure 6. Schematic representation of reduced SCN1A expression in Dravet syndrome, with permission from Doohan (2019). (A) In a healthy patient, Nav1.1 is predominantly expressed on inhibitory interneurons. Inhibitory neurons function to maintain balance between excitatory and inhibitory neurotransmission. (B) In Dravet syndrome there is a heterozygous loss of Nav1.1 expression, leading to reduced interneuron function. The reduced function of interneurons disrupts the balance between inhibitory and excitatory neurotransmission resulting in hyperexcitability.

having multimodal actions. Clobazam is a GABAA receptor PAM (Sankar, 2012), while valproic acid increases GABA concentrations through inhibiting its degradation enzymes, in addition to inhibiting voltage-gated sodium, potassium, and calcium channels, and upregulating SCN1A expression (Ghodke-Puranik et al., 2013; Zhang et al., 2018). Topiramate is also a GABAA PAM, in addition to an inhibitor of non-N-methyl-D-aspartate (NMDA) glutamate receptors and an inhibitor of voltage-dependent sodium and calcium channels (Latini et al., 2008; White et al.,

2000). Benzodiazepines are known to lose anticonvulsant efficacy due to GABAA receptor internalisation. Interestingly, stiripentol is also a GABAA receptor PAM but it binds at a site separate to that of benzodiazepines that does not result in GABAA receptor internalisation

(Grosenbaugh & Mott, 2013). Moreover, stiripentol inhibits the synaptosomal uptake of GABA

(Poisson et al., 1984) and inhibits lactate dehydrogenase, which is hypothesised to reduce neuronal excitability by restricting lactate concentrations, the fuel source for neuron metabolism (Sada et al., 2015). Furthermore, stiripentol has the additional action of inhibiting CYP450 enzymes, consequently increasing the concentrations of concurrent anticonvulsant medications, including clobazam (Giraud et al., 2006). Despite the multimodal actions of the conventional anticonvulsants, Dravet syndrome patients still fail to respond completely to these treatments, highlighting a dire need for the development of better therapeutic options.

Two recent therapeutic developments are fenfluramine and cannabidiol, which were discovered serendipitously. Fenfluramine was originally approved as an anoretic agent but was withdrawn in 1997 due to cardiovascular-associated adverse effects. However, before its withdrawal, fenfluramine proved successful in a trial for refractory epilepsy with self-induced seizures (Boel & Casaer, 1996). When a significant portion of the responders were discovered to carry SCN1A mutations, subsequent trials were conducted in Dravet syndrome patients. Recently,

38 fenfluramine was shown to be an effective anticonvulsant in Dravet syndrome patients in phase III clinical trials when administered as an adjuvant therapy; with patients maintaining a safe cardiovascular profile (Lagae et al., 2019; Schoonjans et al., 2017a; Schoonjans et al., 2017b).

Interestingly fenfluramine’s anticonvulsant actions are proposed to be through augmenting serotonergic signalling, as its primary mode of action is to induce serotonin release from vesicles and consequently enhance synaptic serotonin concentrations (Sourbron et al., 2017). The other novel non-conventional medication to be recently approved for the treatment of Dravet syndrome is cannabidiol, which was described earlier in section 2.

Most anticonvulsant medications can only address the symptoms of epilepsy and not the causal mechanisms. As most Dravet syndrome cases are due to NaV1.1 haploinsufficiency, therapies that augment NaV1.1 might prove efficacious. There are advances in pharmacotherapy being developed to utilize anti-natural antisense transcripts (antagoNAT), which are single-strand oligonucleotide molecules that enhance gene transcription. There is currently an antagoNAT agent called CUR-1916 (also known as OPK88001) being developed by OPKO Health, which aims to upregulate transcription of the healthy copy of SCN1A. CUR-1916 received orphan drug status by the FDA in 2017 and OPKO Health reported intentions for clinical trials but no progress has been reported since. Although the concept of upregulating NaV1.1 is alluring, it may not be a perfect treatment. CUR-1916 significantly reduced spontaneous seizure frequency and duration, while also increasing the temperature thresholds for hyperthermia-induced seizures in a mouse model of

Dravet syndrome (Hsiao et al., 2016). However, it did not completely protect against hyperthermia-induced seizures or spontaneous seizures. Furthermore, CUR-1916’s efficacy was only assessed in mice that express a milder phenotype. That is, the mice were tested at 7-11 weeks,

39 an age at which 70% of the Dravet syndrome mice expressing a severe phenotype have already died. Future studies are needed to assess the agent’s efficacy during the vulnerable period.

Further challenges for the antagoNATs clinical development is that these agents need to be delivered via the invasive intrathecal route (Hsiao et al., 2016). This is likely due to the oligonucleotides being too large to cross the blood brain barrier and having poor bioavailability due to rapid degradation by circulating enzymes (Wahlestedt, 2013). AntagoNATs plasma stability was improved by the addition of a phosphorothioate group, however phosphorothioate-containing oligonucleotides may increase systemic inflammation, leading to fever in some patients

(Wahlestedt, 2013). While this may be tolerated by some populations, fevers are known triggers for seizures in Dravet syndrome patients. It is apparent that there are still numerous hurdles and still much development required for Scn1a-antagoNAT therapies.

To further complicate the treatment strategies of Dravet syndrome, there are also contraindicated pharmacotherapies. The use of sodium channel blockers, such as lamotrigine, carbamazepine or phenytoin, should be avoided as they can exacerbate seizures and can induce lasting negative effects on cognitive outcomes (de Lange et al., 2018; Genton, 2000; Shi et al.,

2015). It is evident that current treatment strategies for Dravet syndrome are complicated and lack efficacy in addressing the condition’s multiple aspects. Most anticonvulsant drugs are designed as broad-spectrum agents, i.e. to treat all types of seizures. However, seizures and epilepsy disorders can arise from many different aetiologies that cause different pathophysiological mechanism.

Although there has been progress in the development of anticonvulsant drugs that utilise novel mechanisms of action and display improved safety profile, they have not reduced the prevalence of drug resistant epilepsy disorders, including Dravet syndrome. The pharmacoresistance observed in patients with Dravet syndrome suggests that seizures are resulting from different or additional

40 pathophysiological mechanisms to those targeted by current therapeutics. Although novel therapeutics such as fenfluramine and CBD have proven useful, they were discovered serendipitously. A targeted and physiologically-informed strategy is required to develop improved therapeutics for Dravet syndrome. Expanding the understanding of the underlying pathophysiological mechanisms of the disorder, in addition to understanding how variability in phenotype can arise from identical genetic alterations, could provide a platform for the discovery of novel and precise drug targets that this thesis will explore.

9.2 Scn1a+/- mouse models of Dravet syndrome

As haploinsufficiency of SCN1A is the major cause of Dravet Syndrome, numerous genetic

+/- mouse models that utilize haploinsufficiency of NaV1.1 (Scn1a ) have been generated. These mouse lines include targeted deletion of Scn1a exon 1 (Scn1atm1Kea) (Miller et al., 2014) and exon

26 (Scn1atm1Wac) (Yu et al., 2006), along with specific point insertion of human mutations R1407X

(Scn1atm1Kzy) (Ogiwara et al., 2007) and E1099X (Scn1atm1.1Swl) (Tsai et al., 2015). Scn1a+/- mouse models recapitulate many of the seizure-related features seen in Dravet syndrome patients. Scn1a+/- mice display increased susceptibility to febrile seizures and exhibit spontaneous seizures and spontaneous death, particularly during the 3rd and 4th postnatal weeks of life (Cao et al., 2012;

Hawkins et al., 2017; Kalume et al., 2007; Miller et al., 2014; Mistry et al., 2014; Oakley et al.,

2009; Ogiwara et al., 2007; Tsai et al., 2015; Yu et al., 2006). Two of the mouse models that assessed behaviour identified that Scn1a+/- models reproduced many of the comorbidities observed in patients, including hyperactivity, stereotyped behaviours, lowered sociability, poor spatial learning, memory deficits and impaired sleep (Han et al., 2012; Ito et al., 2013; Kalume et al.,

2015). As the Scn1a+/- model faithfully recapitulates many aspects of what Dravet syndrome

patients exhibit, it has provided a useful model to investigate the impact of NaV1.1 impairment on the pathophysiologies of the disease.

The use of mouse models have informed on NaV1.1 expression patterns, revealing its expression being predominantly on GABAergic interneurons (Dutton et al., 2013; Ogiwara et al.,

2007). Moreover, the contribution of NaV1.1 channels to the seizure phenotype was supported by

NaV1.1 expression being upregulated during the second and third postnatal weeks, which coincides with the period of spontaneous seizure susceptibility observed in the Scn1a+/- mice (Cheah et al.,

+/- 2013). Electrophysiological studies using the Scn1a mouse models have confirmed that NaV1.1 haploinsufficiency causes network hyperexcitability at the network level as recorded by field potentials from brain slices (Liautard et al., 2013). They have also confirmed impaired function in interneurons from the hippocampus, neocortex and cerebellum (Kalume et al., 2007; Kaplan et al.,

2017; Ogiwara et al., 2013; Rubinstein et al., 2015; Tai et al., 2014; Tsai et al., 2015; Yu et al.,

2006). Interestingly, one study identified increased excitability from cortical excitatory pyramidal neurons (Mistry et al., 2014) while other studies report pyramidal neurons to be unaffected

(Rubinstein et al., 2015; Yu et al., 2006). Further supporting the theory that loss of Scn1a from interneurons is the underlying pathophysiological mechanism was work showing that conditional deletion of Scn1a on GABA interneurons globally or specifically in cortical and hippocampal neurons, was sufficient to reproduce the febrile- and spontaneous-seizure phenotypes, as well as the premature lethality phenotype (Cheah et al., 2012; Kuo et al., 2019; Ogiwara et al., 2013).

Whereas, Scn1a deletion from excitatory neurons had no detectable effects (Dutton et al., 2013;

Ogiwara et al., 2013). Together these studies support impaired interneuron function to be the leading cause of hyperexcitability through disinhibition.

Moreover, Scn1a+/- mice respond to common anticonvulsant medications in a manner similar to what is observed clinically. For example, sodium channel blockers exacerbated seizures in these mice while clobazam was the most effective anticonvulsant (Hawkins et al., 2017).

Additionally, as mentioned earlier, CBD’s efficacy was also replicated in the Scn1a+/- mice

(Anderson et al., 2019a; Kaplan et al., 2017). Therefore, in addition to improving understanding

+/- of the pathophysiology associated with NaV1.1 impairment, the Scn1a mouse models also provide a validated platform to assess the efficacy of potential therapeutic agents.

10. Genetic modifiers: a strategy to identify drug targets

One strategic approach to identifying novel drug targets is the identification of genetic modifiers. As described earlier, identical SCN1A mutations can lead to a range of epilepsy symptoms and variability in their severity. The variability observed in the symptomatic presentation of patients with Dravet syndrome has also been observed in Scn1a+/- mouse models.

The severity of the epilepsy phenotype is critically sensitive to the background strain of the mice with the Scn1a alternations. Heterozygous Scn1a deletion on a 129 background strain

(129.Scn1a+/-) yields a very mild to non-existent seizure phenotype. However, crossing

129.Scn1a+/- mice with wildtype (WT) C57BL/6 mice to generate Scn1a+/- mice on a mixed F1

(129 x C57BL/6) background strain uncovers a severe seizure phenotype (Miller et al., 2014;

Mistry et al., 2014; Yu et al., 2006). Moreover, the seizure phenotype is exacerbated with continuous backcrossing on to a congenic C57BL/6 background strain (Miller et al., 2014;

+/- Rubinstein et al., 2015; Yu et al., 2006). The minimal phenotype expression of Scn1a mice on the 129 background is corroborated by these mice exhibiting significantly less impaired, if not normal, sodium current density in inhibitory neurons (Mistry et al., 2014; Rubinstein et al., 2015).

The sensitivity of the Dravet syndrome phenotype to background strain indicates that non-Scn1a

43 genes are influencing phenotype severity. Genes secondary to the mutation that alter phenotype severity are referred to as genetic modifiers and hold potential as novel therapeutic targets for the treatment of Dravet syndrome.

There have been several genetic modifiers of Scn1a that have already been identified and validated. Potential genetic modifiers were identified in a preliminary RNA-seq analysis that compared the hippocampal transcriptome between the seizure-resistant and seizure-susceptible background strains of a Dravet syndrome mouse model (Hawkins et al., 2019; Hawkins &

Kearney, 2016). After confirmation of varied expression between background strains, the ability of the target gene to influence phenotype can be established through genetic or pharmacological investigations. For example, Gabra2, which codes for the 2 subunit of the GABAA receptor, was less expressed in mice on the F1 seizure-susceptible background strain compared to the 129 seizure-resistant strain, suggesting it may influence phenotype (Hawkins et al., 2016). Then, the

GABAA receptor’s influence on the phenotype was confirmed when positive allosteric modulation of the GABAA receptors, with clobazam or AZD7325, improved susceptibility to febrile seizures and spontaneous seizures (Hawkins et al., 2017; Hawkins et al., 2016; Nourbakhsh, Atabaki &

Roohbakhsh, 2018). Another example is Scn8a, encoding the voltage-gated sodium channel - subunit NaV1.6, which was also more expressed in the seizure-susceptible strain and was proved to be a modifier of Scn1a when pharmacological inhibition of NaV1.6 with GS967 significantly improved survival and spontaneous seizures (Anderson et al., 2017; Martin et al., 2007).

It is likely that there are additional genes, from many neurological systems, that are genetic modifiers of Scn1a. As the endocannabinoid system functions to regulate neuronal activity, there may be molecular constituents in this system that could be influencing the Dravet syndrome

44 phenotype. Utilizing the Scn1a+/- mouse model to investigate the interaction between the endocannabinoid system and NaV1.1 impairment could highlight mechanisms by which cannabinoids are eliciting anticonvulsant actions and reveal new cannabinoid-based drug targets for the directed development of cannabinoid-based medicines.

11. Aims

The global objective to this thesis is to assist in the development of novel therapeutics for the more comprehensive treatment of Dravet syndrome by taking a multi-faceted, physiologically- directed approach that is centred around the Scn1a+/- mouse model. We aim to identify and validate molecular targets by first expanding the understanding of the broader endocannabinoid system in the Scn1a+/- mouse model, and then by employing genetic and pharmacological techniques to determine the impact of potential molecular targets on the epilepsy phenotype. Moreover, we aim to improve the drug-testing platform of the mouse model to include assessment of the associated behavioural comorbidities.

11.1 Specific aims

1. Validation of CB1 receptors as an anticonvulsant target in a Scn1a+/- mouse

model of Dravet Syndrome (Chapter 2)

The endocannabinoid system plays a major role in regulating neuronal excitability, predominantly through CB1 receptors. Pharmacological potentiation of the endocannabinoid system has shown promise in numerous preclinical models of acute seizure and epilepsy but is yet to be investigated in paediatric epilepsy models, including Dravet syndrome models. A preliminary

RNA-seq investigation of the hippocampal transcriptome of a Scn1a+/- mouse model highlighted

Cnr1, the gene encoding the CB1 receptor, as a potential genetic modifier as it was less expressed in mice on the seizure-susceptible strain. Therefore, in Chapter 2 we aim to assess if Cnr1 could be a genetic modifier of Scn1a. We aim to determine if there are alterations in molecular component of the endocannabinoid system that are consistent with increased excitability in the F1 seizure-susceptible strain of the Scn1a+/- mouse model by measuring the mRNA and protein expression of key constituents with ddRT PCR and western blotting, respectively; and endogenous ligand concentrations will be assessed with mass spectrometry. Additionally, we aim to assess if potentiation of endocannabinoid signalling with a CB1 receptor PAM or MAGL inhibitor will display anticonvulsant efficacy against the hyperthermia-induced seizure phenotype of F1.Scn1a+/- mice. We hypothesise that the mice on the seizure-susceptible background strain will have less expression of endocannabinoid system constituents and that potentiation of the endocannabinoid signalling will elicit anticonvulsant actions.

2. Targeting Gpr55 as a treatment strategy in a Scn1a+/- mouse model of Dravet

syndrome (Chapter 3)

There is growing evidence to suggest that Gpr55 can modulate neuronal excitability, including the anticonvulsant actions of CBD being attributed to GPR55 antagonism. Moreover,

Gpr55 was also suggested as a potential genetic modifier of Scn1a in the preliminary RNA-seq study as it was expressed more in the hippocampus of seizure-susceptible mice than seizure- resistant mice. Therefore, in Chapter 3 we aim to investigate if Gpr55 could be a modifier of the

Scn1a+/- phenotype. Our first objective is to determine if there are differences in the Gpr55-LPI system of the Scn1a+/- mouse model by measuring the gene expression and ligand concentrations of key constituents of the Gpr55-LPI system using ddRT-PCR and mass spectrometry,

46 respectively. Then, we aim to examine the influence of Gpr55 on the Dravet syndrome phenotype by evaluating the impact of heterozygous deletion of Gpr55 on the hyperthermia-induced seizures, spontaneous seizures and the premature mortality phenotype in Scn1a+/- mice. We predict to observe changes in key components of the Gpr55-LPI system that would be consistent with increased Gpr55 signalling in the F1 seizures-susceptible mice, and therefore hypothesise that genetic down-regulation of Gpr55 will improve the Dravet syndrome phenotype.

3. Characterising the adolescent behavioural phenotype of a Scn1a+/- mouse

model (Chapter 4)

In addition to seizures, patients with Dravet syndrome also exhibit developmental delays in attention, emotional, and cognitive functions that severely impact on the quality of and are not specifically addressed by current therapeutic strategies. The behavioural phenotype of other

Scn1a+/- mouse models have been characterised to be consistent with impairments seen in patients but behavioural phenotypes have only been characterised in adult mice. The behavioural phenotype of the mouse model we utilise has not been characterised at any stage of development.

Characterisation of the impact of heterozygous deletion of Scn1a adolescents will establish a platform to assess the therapeutic effects of earlier pharmacological intervention. Thus, in Chapter

4, we aim to assess the impact of heterozygous Scn1a deletion on adolescent behaviour by comparing the behaviour of Scn1a+/- and WT mice across several behavioural domains. We will assess motor function (open field test), sociability and social recognition memory (three- chambered social preference and social interaction tests), memory function (novel object recognition, Barnes maze, fear conditioning paradigm), anxiety-related behaviour (elevated plus maze and open field thigmotaxis), startle reflex and sensorimotor gating (prepulse inhibition of

47 startle tests), and repetitive compulsive behaviour (marble burying test). We hypothesise that

Scn1a+/- mice will display impairments in numerous behavioural modalities measured.

Chapter 2

Validation of CB1 receptors as an anticonvulsant target in

a Scn1a+/- mouse model of Dravet Syndrome

Introduction

Dravet syndrome is a severe, intractable form of paediatric epilepsy with patients exhibiting various seizure-types, behavioural comorbidities and a significant risk for sudden unexplained death in epilepsy (SUDEP) (Cooper et al., 2016; Dravet et al., 2005; Dravet & Oguni,

2013; Shmuely et al., 2016). Current anticonvulsant treatments for Dravet syndrome aren’t sufficient and there is a dire need for the development of novel therapeutics (Cooper et al., 2016;

Knupp & Wirrell, 2018). Over 80% of patients have loss-of-function mutations in the SCN1A gene, encoding the 1 subunit of the voltage-gated sodium channel (NaV1.1) (De Jonghe, 2011;

Depienne et al., 2009; Marini et al., 2011). As NaV1.1 is predominantly expressed on inhibitory neurons (Dutton et al., 2013; Ogiwara et al., 2007), its loss-of-function results in global network hyperexcitability through disinhibition (Higurashi et al., 2013; Liu et al., 2013; Maeda et al., 2016;

Sun et al., 2016).

Despite many of the SCN1A mutations resulting in the same functional outcomes, there is considerable variability in symptom severity amongst patients (Brunklaus et al., 2012; Olivieri et al., 2016; Ragona et al., 2011). This incomplete penetrance of the Scn1a phenotype is also observed in mouse models of Dravet syndrome where the severity of the Scn1a+/- phenotype is critically sensitive to the background strain. Scn1a+/- mice on the 129 background strain (129.Scn1a+/-) are phenotypically similar to their wildtype (WT) counterparts. However, crossing 129.Scn1a+/- mice with C57BL/6 mice to create a mixed (129 X C57BL/6J) F1 background strain produces

F1.Scn1a+/- mice that recapitulate many of the symptoms observed in patients. F1.Scn1a+/- mice exhibit increased susceptibility to febrile seizures, spontaneous seizures, reduced lifespan, and behavioural comorbidities (Hawkins et al., 2017; Miller et al., 2014; Mistry et al., 2014; Chapter

4). This strain-dependent presentation of the Dravet syndrome phenotype in mice suggests that

50 phenotype is influenced by non-Scn1a genes, which are termed ‘genetic modifiers’ and may serve as novel drug targets.

One physiological system that may host druggable targets is the endogenous cannabinoid

(endocannabinoid) system, which functions in the central nervous system (CNS) as a regulator of neuronal activity (Kawamura et al., 2006; Sugaya et al., 2016; Takahashi & Castillo, 2006). There are two main cannabinoid receptors, cannabinoid receptor type 1 (CB1) that is distributed predominantly in the CNS, and cannabinoid receptor type 2 (CB2) that is mainly located peripherally (Howlett & Abood, 2017). The main endogenous ligands of CB1 and CB2 receptors are N-arachidonoylethanolamide (anandamide, AEA) and 2-arachidonoylglycerol (2-AG)

(Devane et al., 1992; Mechoulam et al., 1995). In response to post-synaptic neuronal activation, anandamide and 2-AG are synthesised by N-arachidonoyl phosphatidylethanolamine (NAPE) and diacylglycerol lipase (DAGL) isoforms α and β, respectively (Di Marzo et al., 1994; Murataeva,

Straiker & Mackie, 2014).

Endocannabinoids travel retrogradely across the synaptic cleft to activate CB1 receptors, which are located presynaptically on both on glutamatergic and GABAergic neurons (Howlett &

Abood, 2017; Kano, 2014; Kawamura et al., 2006). CB1 receptors are G protein coupled receptors coupled predominantly to Gαi/o proteins (Howlett et al., 2002). Stimulation of the Gαi/o proteins results in numerous intracellular actions, including suppression of cAMP, stimulation of potassium channels and inhibition of inward-facing voltage-gated calcium channels, which ultimately reduce presynaptic cell excitability and inhibit further neurotransmitter release (Farrell & Soltesz, 2019;

Howlett et al., 2002; Soltesz et al., 2015). Anandamide and 2-AG are then rapidly metabolised by fatty acid amine hydrolase (FAAH) and mainly monoacylglycerol lipase (MAGL), respectively

(Blankman, Simon & Cravatt, 2007; Cravatt et al., 1996).

As the endocannabinoid system regulates neuronal activity, we hypothesized that there may be endocannabinoid-related molecular targets that could hold therapeutic potential for the treatment of Dravet syndrome. Further support to investigate the endocannabinoid system was provided by an RNA-seq analysis of the Scn1a+/- model, which showed significantly less hippocampal expression of Cnr1, the gene encoding CB1 receptors, in mice on the seizure- susceptible F1 background strain (129S6/SvEvTac x C57BL6/J) compared to mice on the

129S6/SvEvTac seizure-resistant background strain (Hawkins et al., 2019). Theoretically, reduced

CB1 receptor expression in the F1 mice could be decreasing regulation of neuronal transmission.

This could be contributing to the excessive neuronal hyperexcitability and the severe seizure phenotype observed in F1.Scn1a+/- mice, making it a potential genetic modifer.

Based on the neuronal type that CB1 receptors are located on, stimulation of CB1 receptors can inhibit the release of glutamate or GABA (Farrell & Soltesz, 2019; Lee, Foldy & Soltesz,

2010; Neu, Foldy & Soltesz, 2007; Soltesz et al., 2015), which can lead to anticonvulsant or proconvulsant effects. This contrasting effect of CB1 receptor stimulation has been observed in preclinical models of seizure. For example, the synthetic CB1 receptor agonists WIN 55,212 and

ACEA were anticonvulsant in pentylenetetrazol and maximal electroshock acute models of seizure and in the pilocarpine model of status epilepticus in rodents (Suleymanova et al., 2016; Vilela et al., 2013; Wallace et al., 2001). Whereas CB1 receptor agonism has also demonstrated intrinsic proconvulsant effects and increased susceptibility in conventional seizure models, albeit predominantly with synthetic and potent agonists (Funada & Takebayashi-Ohsawa, 2018; Kevin et al., 2019; Malyshevskaya et al., 2017; Vilela et al., 2013; Wilson et al., 2019).

To avoid the proconvulsant effects of CB1 receptor agonism, endocannabinoid signalling can be pharmacologically enhanced indirectly and specifically. One such approach is the elevation

52 of endocannabinoid concentrations through inhibition of their degradative enzymes. For example,

MAGL inhibitors that elevate 2-AG concentration have been demonstrated to display anticonvulsant effects in pentylenetetrazol, status epilepticus and kindling models of seizure and epilepsy in rodents (Terrone et al., 2018; von Ruden et al., 2015a; Zareie et al., 2018). MAGL inhibition has the added advantage of enhancing endocannabinoid signalling in spatiotemporally restricted manner as its effects are dependent on the presence of 2-AG, and 2-AG is only produced on-demand where neural activity is present, which would occur at glutamatergic synapses in a hyperexcitable network.

Another preferred pharmacological approach is positive allosteric modulation (PAM) of

CB1 receptors. PAMs bind to a site on the receptor separate to the orthosteric binding site that elicits no intrinsic activation of the receptor (Wootten, Christopoulos & Sexton, 2013). Instead,

PAMs increase the efficacy and activity of orthosteric ligands and therefore their function is specific and limited by the presence of orthosteric ligands (Wootten, Christopoulos & Sexton,

2013). Like MAGL inhibitors, CB1 receptor PAMs potentiate the intrinsic neural regulating effect of the endocannabinoid system. This has been demonstrated by the CB1 receptor PAMs ZCZ011 and GAT229 reducing neuropathic pain in mouse models without inducing cannabimimetic adverse effects (Ignatowska-Jankowska et al., 2015; Slivicki et al., 2018). As such, MAGL inhibitions and CB1 receptor positive allosteric modulation are compelling pharmacological approaches to increasing CB1 receptor-mediated neuromodulation that has not yet been explored in Dravet syndrome, or other paediatric epilepsy preclinical models.

The present chapter aimed to determine if Cnr1 could be a genetic modifier in a Scn1a+/- mouse model of Dravet syndrome. We measured CB1 mRNA and protein expression, along with endocannabinoid concentrations in the Scn1a+/- mouse model of Dravet syndrome, to identify

53 reduced CB1 receptor expression and reduced 2-AG concentrations in the F1 seizure-susceptible strain of mice. After identifying a potential double deficit in the endocannabinoid system of seizure susceptible mice, we hypothesized that targeting these deficits could be anticonvulsant. Thus, we investigated and demonstrated that pharmacological potentiation of endocannabinoid signalling with a CB1 receptor PAM, or MAGL inhibitor was anticonvulsant against hyperthermia-induced seizures in F1.Scn1a+/- mice.

Materials and methods

Mice

All research and animal care procedures were approved by the University of Sydney

Animal Ethics Committee in agreement with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes and by the Northwestern University Animal Care and Use

Committee in accordance with the National Institutes of Health Guide for the Care and Use of

Laboratory Animals. Mice heterozygous for Scn1a (Scn1a+/-) were generated by targeted deletion of Scn1a exon 1 as previously described (Miller et al., 2014) and obtained from The Jackson

Laboratory (MMRRC Stock No: 37107-JAX) for the Australian studies. Scn1a+/- mice on the

129S6/SvEvTac (129) background (129.Scn1a+/-) were maintained by continuous backcrossing.

129.Scn1a+/- mice were crossed with C57BL/6J (B6) mice to generate [129 x B6] F1.Scn1a+/- mice and WT mice. Genotyping was performed as previously described (Miller et al., 2014). Gene transcription studies were conducted at Northwestern University, U.S.A, where the mouse facility operated on a 14 h light/10 h dark cycle. Whereas, all other work too place at the University of

Sydney, Australia, where the facility was kept on a 12 h light/dark cycle. All animals had ad libitum access to food and water and were group housed in individually ventilated cages.

Digital-drop real-time PCR (ddRT-PCR)

Transcript expression of target genes were determined using ddRT-PCR and performed as previously described (Hawkins et al., 2019; Hawkins et al., 2016). Briefly, at P24, cortical and hippocampal samples were dissected from WT and Scn1a+/− mice on both the 129 and F1 (129 x

B6) background strains. To achieve sufficient mRNA concentrations for analyses, primary pools of mRNA were created by combining tissue from 3 – 4 mice, with at least one male and female in each group to control for potential sex-dependent differences. Primary mRNA pools were used to assess Cnr1 expression. For analysis of the 2-AG metabolizing enzymes, RNA superpools were created by combining 4-3 primary RNA pools (n = 12 – 16 mice/superpool). Total RNA was isolated before first-strand cDNA was synthesized and ddRT-PCR was performed using TaqMan

Gene Expression Assays (Life Technologies) for mouse Cnr1 (Mm01212171_s1); Mgll

(Mm00449274_m1); Dagla (Mm00813830_m1); Daglb (Mm00523381_m1); Abhd6

(Mm00481199_m1); Abhd12 (Mm00470489_m1); Tbp Mm00446971_m1. After amplification, a

QX200 droplet reader with QuantaSoft v1.6.6.0320 software was used to analyse droplets and transcript levels of target genes were expressed as a ratio relative to Tbp concentrations.

Western blot

Western blot analyses were performed as previously described (Hawkins et al., 2019;

Hawkins et al., 2016) with modifications. Briefly, animals were euthanized via cervical dislocation and decapitation. Brains were rapidly extracted and snap-frozen in liquid nitrogen. Hippocampal sections were dissected from brains thawed on ice and snap-frozen again in liquid nitrogen. Frozen hippocampal sections were homogenized by hand in ice cold 0.32 M sucrose, 5 mM Tris (pH 7.4) supplemented with protease inhibitor (Merck 4693159001) before being centrifuged for 10 min at

2,000 g, at 4°C. Supernatants were collected and centrifuged for 40 min at 20,000 g, at 4°C. The pellets were re-homogenised in 5 mM Tris (pH 8.2) and 1 mM ethylenediaminetetraacetic acid

(EDTA) supplemented with protease inhibitor. Samples were centrifuged again for 40 min at

20,000 g, at 4°C, then the procured pellet was resuspended in a solution of 50 mM Tris (pH 7.5),

10 mM (EDTA) supplemented with protease inhibitor (Roche, 4693159001). Protein lysates (20

µg) were denatured in 2x laemmli sample buffer (Biorad, 1610737) and 5% beta mercaptoethanol for 15 min at room temperature, before being separated on a 10% SDS-page gel and transferred to nitrocellulose membranes. Membranes were probed for CB1 using a rabbit polyclonal antibody

(1:200, Abcam, ab23703) and for β-tubulin using a mouse monoclonal antibody (1:500, Sigma-

Aldrich, T5201) for 1 h at room temperature, then with secondary antibodies, anti-Rabbit Alexa

Fluor 800 (A32735) and anti-Mouse Alexa Fluor 680 (A28183) at 1:20,000 for 30 min at room temperature. The probed membranes were imaged with an Odyssey imaging system (LI-COR

Biosciences). Signal-intensity analysis of western blots was performed using Image Studio Lite software (LI-COR Biosciences) and data was presented as the signal intensity of CB1 as a ratio of

β-tubulin.

Drugs

To investigate the anticonvulsant efficacy of potentiating endocannabinoid signalling, we elected to use the CB1 positive allosteric modulator (PAM) with GAT229 (Mitjavila et al., 2018) and the irreversible MAGL inhibitor ABX-1431 (Cisar et al., 2018). S-(-)-3-(2-Nitro-1- phenylethyl)-2-phenyl-1H-indole (GAT229) was generously gifted by Assoc. Prof. Ganesh

Thakur (Northeastern University, USA), while 1,1,1,3,3,3-hexafluoropropan-2-yl-4-[[2- pyrrolidin-1-yl-4-(trifluoromethyl)phenyl]methyl]-piperazine-1-carboxylate (ABX1431) was purchased from WuXi AppTec (Shanghai, China) and Diazepam was purchased from Sigma

Aldrich (St Louis, USA). We investigated the anticonvulsant potential of a GAT229 at 30 and 100 mg/kg and ABX-1431 at 3, 10 and 30 mg/kg (Cisar et al., 2018; Mitjavila et al., 2018). Our starting dose of 30 mg/kg for GAT229 was based on the structurally-similar but more potent CB1 PAM,

ZCZ011, eliciting in vivo effects at 40 mg/kg (Ignatowska-Jankowska et al., 2015). Whereas the

ABX-1431 dosing regimen was based off work demonstrating its ability to inhibit Magl at a 4 mg/kg dose when administered orally in mice (Cisar et al., 2018). All drugs were prepared in a vehicle of ethanol:Tween80:0.9% saline (1:1:18) fresh on the day of experimentation and were administered via intraperitoneal (i.p.) injection in an injection volume of 10 mL/kg. Experimental time points of 30 min for GAT229 and 60 min for ABX-1431 administration were based on previously determined time-to-peak plasma and brain concentrations or effect from the literature or our pharmacokinetic studies (Berger et al., 2012; Kaczocha et al., 2014).

Endocannabinoid and drug analyses using high-performance liquid chromatography tandem mass spectrometry (HPLC/MS/MS)

Analysis of endocannabinoid and structurally-related lipids in brain tissue were determined via LC/MS/MS as previously described (Clarke et al., 2017; Stuart et al., 2013). Briefly, male mice at P24 were euthanized at the same time of day via cervical dislocation. Brains were rapidly extracted and flash frozen with liquid nitrogen. Hippocampal sections were dissected from brains thawed on ice and re-frozen in liquid nitrogen. Samples were stored in the -80C freezer until analysed. Homogenised hippocampal samples underwent solid phase extraction using 500 mg C18 solid-phase extraction columns (Agilent Technologies, Santa Clara, CA) to isolate the lipids, 2-

AG), AEA, PEA, OEA, LEA and DHEA. Elutants (20 μL) were separated in a C18 Zorbax reversed-phase analytical column (Agilent Technologies, Santa Clara, CA) and ionized using a

Shimadzu 8030 triple quadrupole mass spectrometer via positive electrospray ionization through

a multiple reaction monitoring method. Internal standard arachidonoyl ethanolamide-d4 (d4-AEA) was purchased from Tocris Bioscience (St. Louis, MO), whereas standards for 2-AG and AEA,

PEA, OEA, LEA and DHEA were purchased from Cayman Chemical (Ann Arbor, MI).

Hyperthermia-induced seizures

Hyperthermia-induced seizure experiments were conducted as previously described in both male and female F1.Scn1a+/- mice from P14 to P16 (Hawkins et al., 2017). Mice received an i.p. injection of vehicle, GAT229 or ABX-1431. Then after a pre-determined experimental time-point, the mice were fitted with a rectal probe that was connected to a temperature controller, which was connected to a heat lamp, and allowed to acclimatize for 5 min. The baseline body temperature of the mice was recorded before their core body temperature was elevated with a heat lamp at a rate of 0.5°C every 2 min until the onset of the first clonic convulsion with loss of posture or until

42.5°C was reached. Mice that reached 42.5°C were held at temperature for 3 min and if no seizure occurred, they were considered seizure-free. Body temperatures of seizure onset were recorded and then the mice were euthanized via isoflurane inhalation before plasma and brain samples collected and stored at -80°C until assayed.

Analytical methods with HPLC/MS/MS

Plasma and brain concentrations of GAT229 and ABX-1431 in experimental mice were assayed with HPLC/MS/MS and were performed based on previously described methods with modifications (Anderson et al., 2019a; Anderson et al., 2019b). Briefly, plasma samples (50 μL aliquots) were spiked with 5 μL internal standard (2 μg/mL diazepam) and extraction was achieved by vortex-mixing with 4 X volume of acetonitrile. The organic layer was isolated by centrifugation

(4000 g for 10 min at 4°C) and evaporated to dryness under nitrogen at room temperature. For

GAT229, residues were reconstituted in methanol (150 μL) and 0.1% formic acid (250 μL) in water; while for ABX-1431, residues were reconstituted in acetonitrile (90 μL) and 0.1% formic acid in water (300 μL). Then supported-liquid extraction (SLE) was performed and extracts were eluted in methyl tert-butyl ether (MTBE) and dried to completeness under nitrogen at room temperature before GAT229 samples were reconstituted in 50 μL of 0.1% formic acid in water and methanol (2:3 v/v); while ABX-1431 samples were reconstituted in 0.1% formic acid in water and acetonitrile (1:1 v/v) for HPLC/MS/MS analysis.

For brain analyses, brain hemispheres were homogenized in 10 X methanol: water (1:1) and the supernatant layer was isolated by centrifugation (20,000 g for 30 min at 4°C), supplemented with acetonitrile (300 μL), spiked with 15 μL internal standard (2μg/mL diazepam) and vortexed for 10 min. Samples were first centrifuged (10 min at 18,000 g and 4°C) before supernatants were collected and centrifuged again (2 h at 18,000 g and 4°C). Filtrates were collected and dried to completeness under nitrogen, reconstituted for SLE and underwent SLE as described above. For GAT229, SLE eluents were reconstituted in 50 μL of 0.1% formic acid in water and methanol (2:3 v/v); whereas for ABX-1431, SLE eluents were reconstituted in 0.1% formic acid in water and methanol (2:3 v/v) for analysis via HPLC/MS/MS.

GAT229 and ABX-1431 concentrations were analysed using a Shimadzu Nexera ultra- high-performance liquid chromatograph coupled to a Shimadzu 8030 triple quadrupole mass spectrometer (Shimadzu Copr., Kyoto, Japan). For GAT229, mobile phases 0.1% formic acid in water (A) and methanol (B) were delivered with a flow rate of 0.3 L/min as follows: mobile phase

B was first held at 40% for 1 min, then increased from 40% to 90% over 2 min, held at 90% for

3.5 min, decreased back to 40% over 0.5 min and equilibrated at 40% for 0.5 min. For ABX-1431, mobile phases 0.1% formic acid in water (A) and acetonitrile (B), were delivered at a flow of 0.6

Statistical analyses

Statistical analyses were performed using SPSS software, version 24 (IBM Corp, Armonk,

New York USA). The ddRT-PCR transcript, protein and endocannabinoid and related-lipid data were analysed using two-way ANOVA, which compared main effects of background strain and genotype. If a strain by genotype interaction effect was identified, then planned Bonferroni’s comparisons were employed (129.WT vs 129.Scn1a+/-; F1.WT vs F1.Scn1a+/-; 129.WT vs F1.WT;

129.Scn1a+/- vs F1.Scn1a+/-). For hyperthermia-induced seizures, threshold temperatures were compared using the time to event analysis (log-rank Mantel-Cox). For analytical analyses of experimental samples, plasma and brain concentrations for GAT229 were compared with unpaired t-tests; while ABX-1431 samples were compared with a one-way ANOVA, followed by Tukey’s post hoc tests. No significant sex differences were observed across all observable measures, and so groups were collapsed across sexes. Comparisons were statistically significant if p < 0.05.

F1 seizure-susceptible background strain mice displayed reduced hippocampal

Cnr1 mRNA and CB1 receptor protein than 129 seizure-resistant strain mice.

Seizure susceptible F1.Scn1a+/- mice start to develop spontaneous seizures between the 3rd and 4th post-natal weeks (Miller et al., 2014; Mistry et al., 2014). Thus, we examined whether hippocampal Cnr1 transcript and CB1 protein expression were altered at P24 using ddRT-PCR and western blotting, respectively. Cnr1 mRNA expression was modestly but significantly lower in the F1 seizure-susceptible strain compared to the 129 seizure-resistant strain (two-way

ANOVA: strain, F 1, 31 = 7.178, p = 0.012). Heterozygous deletion of Scn1a did not influence Cnr1 transcript expression as there was no effect of genotype (F 1, 31 = 0.007, p = 0.986), nor a strain by genotype interaction (F 1, 31 = 1.495, p = 0.231). We then sought to confirm these transcriptional changes at the functional protein level using western blotting of the CB1 receptor in hippocampal samples (Figure 1b-c). CB1 protein concentrations relative to β-tubulin correlated with transcript expression as there was significantly less protein expression in F1 mice compared to 129 mice

(two-way ANOVA: strain, F1, 27 = 5.513, p = 0.026), with no effect of genotype (F1, 27 = 3.801, p =

0.062) nor a strain by genotype interaction (F1, 27 = 1.583, p = 0.219).

Figure 2. Hippocampal concentrations of endocannabinoids in WT and Scn1a+/- mice from 129 and F1 (129 x B6) background strains. Hippocampal concentrations of endocannabinoids (a) 2-AG and (b) AEA as determined by HPLC/MS/MS. Blank columns represent WT data while coloured columns represent Scn1a+/- data. Data is represented as the mean + SEM with individual circles representing individual mouse data (n = 14 – 16). Statistical comparison between groups were made using a two-way ANOVA.

F1 seizure-susceptible background strain mice displayed lower hippocampal 2-

AG concentrations than 129 seizure-resistant mice.

Next we evaluated hippocampal concentrations of the two main endocannabinoids, 2-AG and anandamide (Figure 2), along with the structurally-related fatty acid amides, PEA, OEA, LEA and DHEA (Table 2) using HPLC/MS/MS. Hippocampal 2-AG concentrations were significantly lower in F1 seizure-susceptible mice compared to 129 seizure-resistant mice (Figure 2a; two-way

ANOVA: strain, F1, 56 = 4.7, p = 0.035) with no main effect of genotype (F1, 56 = 0.025, p = 0.875) and no strain by genotype interaction (F1, 56 = 1.161, p = 0.286). Alternatively, hippocampal

Figure 3. Hippocampal transcript expression of 2-AG’s synthesizing and catabolising enzymes in WT and Scn1a+/- mice from 129 and F1 (129 x B6) background strains. (a) Schematic diagram of 2-AG metabolism highlighting the genes coding the synthesizing and degradation enzymes. Hippocampal transcript expression of 2-AG’s synthesizing enzymes, (b) Dagl-α (Dagla) and (c) Dagl-β (Daglb); and 2-AG’s metabolizing enzymes, (d) Magl (Mgll), (e) Abhd6 (Abhd6) and (f) Abhd12 (Abhd12) as determined with ddRT-PCR and expressed as a ratio of Tbp. Data is presented as the mean + SEM with triangles representing mRNA superpools (n = 3 – 6 biological replicates). Blank columns represent WT data while coloured columns represent Scn1a+/- data. Data were analysed using a two-way ANOVA with planned Bonferroni comparisons, *p < 0.05.

2-AG metabolizing enzymes are less expressed in the hippocampus of seizure- susceptible mice and Daglb mRNA expression is selectively increased seizure- susceptible F1.Scn1a+/- mice

After identifying altered 2-AG concentrations between the seizure-resistant and seizure- susceptible background strains, we measured the hippocampal transcript expression for 2-AG’s synthetic and metabolizing enzymes using ddRT-PCR (Figure 3). We did not detect any differences in the expression of 2-AG’s main synthesizing enzyme Dagla (Figure 3b). However, we found that heterozygous Scn1a deletion selectively increased Daglb mRNA expression, but only in mice on the seizure-susceptible F1 strain (Figure 3c). This was supported by a significant strain by genotype interaction (two-way ANOVA: F1, 11 = 5.198, p = 0.044), with no main effects of strain (F1, 11 = 0.812, p = 0.387) or genotype (F1, 11 = 2.11, p = 0.174). Planned Bonferroni comparisons confirmed F1.Scn1a+/- mice had significantly more Daglb mRNA expression than

F1.WT mice (p = 0.016) and 129.Scn1a+/- mice (p = 0.033). We also identified that heterozygous

Scn1a deletion significantly reduced the degrative enzyme Mgll’s mRNA expression across both background strains (Figure 3d; two-way ANOVA: genotype, F1, 11 = 6.728, p = 0.025; strain, F1,

11 = 0.660, p = 0.434, strain by genotype interaction, F1, 11 = 1.925, p = 0.193). Whereas Abhd6 mRNA expression was significantly reduced in F1 mice compared to 129 mice (Figure 3e; two- way ANOVA: strain, F1, 11 = 11.02, p = 0.007), with no effect of genotype (F1, 11 = 0.908, p =

0.361) and no strain by genotype interaction (F1 ,11 = 0.255, p = 0.624). We did not observe any statistically significant differences in Abhd12 mRNA expression (Figure 3f).

Figure 4. Efficacy of GAT229, a CB1 receptor PAM, and ABX-1431, a MAGL inhibitor, against hyperthermia-induced seizures in F1.Scn1a+/- mice. Threshold temperature for GTCS induced by hyperthermia following administration of (a) vehicle or GAT229 (30 or 100 mg/kg); and (b) vehicle or ABX-1431 (3, 10 or 30 mg/kg). Data is shown as average temperatures of seizure induction + SEM, with open circles representing individual mouse data (n = 14 – 16 mice). Data was analysed with log-rank Mantel-Cox with mean temperature thresholds that are significantly greater than the vehicle group depicted in green, *p < 0.05 compared to vehicle.

Positive allosteric modulation of the CB1 receptor and inhibition of MAGL were anticonvulsant against hyperthermia-induced seizures.

The anticonvulsant efficacy of GAT229, a CB1 receptor PAM, and ABX-1431, a MAGL inhibitor, were assessed against hyperthermia-induced seizures in F1.Scn1a+/- mice (Figure 4).

Administration of 30 and 100 mg/kg of GAT229 equivalently and significantly elevated the temperature threshold for GTCS compared to vehicle control mice (Figure 4a, log-rank Mantel‐

Cox test: p = 0.035 and p = 0.024, respectively). Analysis of GAT229 concentrations from

70 emphasis on CB1 receptors. We confirmed that Cnr1 mRNA and functional CB1 protein concentrations were significantly lower in the hippocampus of F1 seizure-susceptible mice compared to seizure-resistant 129 strain mice. Moreover, we found F1 mice had significantly less hippocampal 2-AG concentrations, with no differences in anandamide concentrations between the strains. After detecting a potential deficit in the endocannabinoid system of susceptible F1 mice, we assessed the anticonvulsant potential of augmenting the endocannabinoid system and demonstrated that positive allosteric modulation of CB1 receptors and MAGL inhibition were anticonvulsant against hyperthermia-induced seizures in the phenotypic F1.Scn1a+/- mice. These data provide first evidence in support of Cnr1 being a putative genetic modifier of Scn1a.

We hypothesised that less hippocampal CB1 receptor mRNA and protein expression, along with less hippocampal 2-AG concentrations in the F1 mice may suggest reduced hippocampal endocannabinoid signalling in F1 mice compared to 129 mice. As 2-AG regulates neuronal excitability through CB1 receptors (Gao et al., 2010; Tanimura et al., 2010), deficient endocannabinoid signalling on excitatory neurons may contribute to the heightened neuronal excitability and seizures observed in F1.Scn1a+/- mice. We provide evidence in support of the endocannabinoid system double deficit hypothesis by showing that pharmacological potentiation of endocannabinoid signalling is anticonvulsant in Dravet syndrome mice. That is, both positive allosteric modulation of CB1 receptors with GAT229 and the irreversible MAGL inhibitor ABX-

1431 were anticonvulsant against hyperthermia-induced seizures in F1.Scn1a+/- mice. To unequivocally demonstrate that that CB1 receptors mediate the anticonvulsant effects of GAT229 and ABX1431, future studies could attempt to reverse the effects of these compounds using CB1 receptor antagonists. However, this may be problematic as CB1 antagonists have themselves

71 demonstrated proconvulsant actions in numerous conventional seizure models (Hsu et al., 2012;

Perescis et al., 2017; Wallace et al., 2003).

As we only studied hippocampal CB1 receptor mRNA and protein expression in homogenate hippocampal tissue samples, we cannot distinguish between CB1 receptor expression on excitatory pyramidal neurons versus inhibitory interneurons (Howlett & Abood, 2017; Kano,

2014; Kawamura et al., 2006). Future studies could use single-cell sequencing or double-labelling immunofluorescence to evaluate the CB1 receptor mRNA or protein expression on specific neuronal populations in Scn1a+/- mice. CB1 receptors appear to be expressed more abundantly on inhibitory interneurons (Katona, 2015), so CB1 receptor activation would be anticipated to disinhibit neuronal excitability leading to proconvulsant effects. However, here the CB1 receptor

PAM GAT229 and the MAGL inhibitor ABX-1431 both had anticonvulsant effects. Therefore, it appears more likely that the predominate effects of CB1 receptor activation is to reduce neuronal hyperexcitability through actions on CB1 receptors expressed on excitatory pyramidal neurons.

This might be because CB1 receptors are more efficiently coupled to G proteins at excitatory synapses (Katona, 2015; Steindel et al., 2013). Moreover, CB1 receptors are almost exclusively expressed on cholecystokinin-expressing interneurons, which only account for 10-20% of all

GABAergic neurons, and are different to the parvalbumin and somatostatin interneurons impacted by partial genetic deletion of Scn1a (Marsicano & Lutz, 1999; Whissell et al., 2015). Future electrophysiological experiments could be employed to disentangle the effects of GAT229 and

ABX-1431 on excitatory and inhibitory transmission in Scn1a+/- mice.

This is the first demonstration of a CB1 receptor PAM having anticonvulsant activity.

GAT229 displays no intrinsic activity at CB1 receptors, but instead potentiates CB1 receptor signalling in the presence of orthosteric ligands (Laprairie et al., 2017). Thus, the anticonvulsant

72 efficacy of GAT229 would suggest it potentiates the recruitment of the endocannabinoid system to counter seizure-associated hyperexcitability. Future studies are needed to show that seizures increase endocannabinoid transmission. The efficacy of GAT229 plateaued at the 100 mg/kg dose as its anticonvulsant effect was no greater than the effects of the 30 mg/kg dose. The plateau in efficacy may be due to GAT229 being poorly soluble and having poor absorption, as demonstrated by both doses yielding comparable plasma and brain concentrations. Unfortunately, due to the nature of Dravet syndrome being a paediatric disease, experiments are conducted on juvenile mice and we therefore cannot deliver drugs by central routes of administration. Consistent with the observation that F1 strain mice have less brain 2-AG concentrations, an alternative explanation for the limited anticonvulsant efficacy of GAT229 is that its positive allosteric modulation is limited by the restricted endocannabinoid tone in F1.Scn1a+/- mice. One possible approach to surmount this issue is to observe whether greater anticonvulsant effects can be attained through the co- administration of GAT229 with ABX1431 to increase concentrations of the orthosteric ligand 2-

AG.

This Chapter provides novel evidence that a MAGL inhibitor has anticonvulsant effects in a mouse model of intractable, childhood epilepsy. Our results are consistent with MAGL inhibitors displaying anticonvulsant properties in other conventional seizure and epilepsy models (Sugaya et al., 2016; Terrone et al., 2018; von Ruden et al., 2015a; Zareie et al., 2018). ABX-1431 was selected as our test compound because it is a highly selective irreversible MAGL inhibitor that has good oral bioavailability, which is conducive to our future experiments that will deliver drug in mouse chow to assess effects on spontaneous seizures and survival (Cisar et al., 2018). There is also a chance for more rapid translation as ABX1431 has passed Phase 1 safety trials and is currently in Phase II clinical trials for the treatment of Tourette’s syndrome (NCT03625453).

Disappointingly, ABX-1431 demonstrated an inverted-U-shaped dose-response in the present study, with the middle dose level proving anticonvulsant but the lower and higher doses failing to demonstrate efficacy. We predict that ABX-1431 should dose-dependently increase brain 2-AG concentrations, as oral doses of ABX-1431 (> 2 mg/kg) dose-dependently increased 2-AG levels more than 3-fold (Cisar et al., 2018). However, we need to confirm that ABX-1431 increased brain

2-AG concentrations in our experimental mice, which might provide some insight into the inverted

U-shaped dose-response.

Assuming a dose-response increase in 2-AG concentrations, one possible explanation for loss of efficacy at the higher ABX-1431 dose is that the excessive 2-AG concentrations impinge upon CB1 receptors expressed on GABA interneurons. This would result in suppression of inhibitions (disinhibition) that would counter and potentially override the anticonvulsant excitatory-suppressing effects of 2-AG observed at the lower dose. This hypothesis is supported by other MAGL inhibitors, including JZL184 and KML29, inhibit excitatory neurotransmission at low doses but inhibiting inhibitory transmission at higher doses (Wang et al., 2017b). Another possible explanation for ABX-1431’s loss of anticonvulsant efficacy at the higher dose is through high 2-AG concentrations exerting actions other excitatory targets. For example, 2-AG can behave as a negative allosteric modulator of adenosine A3 receptors and an agonist at GPR55 and TRPV1 receptors (Lane et al., 2010; Petrosino et al., 2016; Ryberg et al., 2007), which are all mechanism that could increase neuronal excitability (Kaplan et al., 2017; Naziroglu & Ovey, 2015; Shirazi et al., 2014; Swiader, Kotowski & Luszczki, 2014).

The current Chapter also investigated the expression of 2-AG’s synthetic and degradative enzymes at the transcript level. Heterozygous Scn1a deletion downregulated 2-AG’s main degradative enzyme, Mgll, as well as another catabolising enzyme Abhd6 in F1 mice. Together,

74 these expression patterns imply that F1.Scn1a+/- mice may have increased 2-AG concentrations.

However, our finding that the F1 background strain displayed less hippocampal 2-AG concentrations mismatched our findings of changes in 2-AG synthetic and metabolizing enzymes.

This inconsistency could occur due to tissue samples being collected in different laboratories with different sampling times, as it is well-established that endocannabinoids concentrations fluctuate with circadian rhythms (Liedhegner, Sasman & Hillard, 2014; Valenti et al., 2004). Future studies could assess whether changes in these enzymes occur at a functional protein level, in conjunction with measuring endocannabinoid concentrations in the same animals.

A notable finding from this work was the increase of Daglb mRNA expression selectively in the phenotypic F1.Scn1a+/- mice. Although both the  and β isoforms of DAGL synthesise 2-

AG, the β isoform is almost exclusively centrally expressed in microglia and doesn’t overly influence global 2-AG levels as demonstrated in Daglb knock-out mice (Tanimura et al., 2010;

Viader et al., 2016; Wilkerson et al., 2016). Microglia are the immunocompetent cells of the CNS

(Kettenmann et al., 2011), with microglial 2-AG concentrations increasing in response to injury and eliciting a protective role (Hohmann et al., 2019; Kreutz et al., 2009; Kreutz et al., 2007; Stella,

2009; Viader et al., 2016; Walter et al., 2003). Additionally, Dagl-β enzymes function is indirectly responsible for the production of inflammatory cytokines, including PGE2, IL-6, TNF (Hsu et al., 2012). The increased expression of Daglb in F1.Scn1a+/- mice then suggests that these mice may have increased microglial 2-AG concentrations, which may be occurring in response to neuronal damage. While the increased Daglb mRNA expression could be due to an increase in

Daglb on microglia, it could also be due to an increase in microglia cell density. Future work could provide a more detailed appraisal of immune function and neuroinflammation in Dravet syndrome mice. Assessment of endocannabinoid concentrations from isolated microglia may not be feasible,

75 but triple-immunolabelling with probes for microglia (Iba-1), Dagl-β and CB2 receptors would inform on microglial cell count and morphology (Carlisle et al., 2002).

In summary, we present evidence for a potential double deficit in the endocannabinoid system of seizure-susceptible F1 Dravet syndrome mice with reduced expression of hippocampal

CB1 receptors and 2-AG concentrations. Moreover, we demonstrated potentiation of endocannabinoid signalling through positive allosteric modulation of CB1 receptors and MAGL inhibition was anticonvulsant against hyperthermia-induced seizures in F1.Scn1a+/- mice, supporting Cnr1 as a putative genetic modifier of Scn1a. Future work extending upon these experiments should investigate the efficacy of augmenting endocannabinoid signalling on the other phenotypes displayed by the Scn1a+/- mice, including spontaneous seizures, premature mortality and behavioural comorbidities.

Supplementary materials

Supplementary Figure 1. An example of a western blot used to evaluate hippocampal cannabinoid receptor type 1 (CB1) expression in WT and Scn1a+/- mice on 129 and F1 (129 x B6) background strains. A cellulose membrane was imprinted with hippocampal protein fractions that were then probed with both CB1 (1:200, Abcam, ab23703) and β-tubulin (1:500,

Sigma-Aldrich, T5201) antibodies labelled with immunofluorescent secondary antibodies (Alexa

Fluor 800 (A32735) and 680 (A28183), respectively), before being imaged at (a) 700 nm, (b) 800 nm wavelengths. (c) Both 700 nm and 800 nm images overlayed. F, female; M, male; WT, wildtype; 129 and F1 (129 x B6) represent background strains; white arrow points to CB1 protein

(predicted molecular weight: 53 kDa). β-tubulin’s predicted molecular weight is 50 kDa.

Chapter 3

Targeting Gpr55 as a treatment strategy in a Scn1a+/-

mouse model of Dravet syndrome

Introduction

Dravet syndrome is a catastrophic and intractable childhood epilepsy with patients experiencing poor quality of life due to unmanaged seizures, associated behavioural comorbidities and significant risk of SUDEP (Cooper et al., 2016; Dravet et al., 2005; Dravet & Oguni, 2013).

Dravet syndrome patients are refractory to current treatments, emphasizing the need to develop novel therapeutics (Knupp & Wirrell, 2018). Over 80% of patients with Dravet syndrome express loss-of-function mutations in SCN1A, which encodes the α1-subunit of voltage gated sodium channel (NaV1.1) that is predominantly expressed on inhibitory neurons (De Jonghe, 2011;

Depienne et al., 2009; Marini et al., 2011); thereby resulting in hyperexcitability through disinhibition. However, carrying a mutation in SCN1A does not guarantee development of a severe epilepsy phenotype (Kearney, 2011; Meisler & Kearney, 2005), suggesting additional factors influence disease severity.

Incomplete penetrance of the Scn1a phenotype is also observed in the Scn1a+/- mouse model of Dravet syndrome. Heterozygous deletion of Scn1a on a 129 background strain shows no overt phenotype but gene deletion on a mixed F1 (129 x C57BL/6) background yields a severe epilepsy phenotype corresponding to what is observed in patients, with febrile seizures, spontaneous seizures, premature mortality, along with behavioural and cognitive impairments (Miller et al.,

2014; Mistry et al., 2014; Yu et al., 2006; Chapter 4). Phenotypic Scn1a+/- mice also respond to pharmacological agents in a way that is similar to patients, further validating the reliable use of this model for drug development (Anderson et al., 2019a; Hawkins et al., 2017; Kaplan et al.,

2017). The sensitivity of the Dravet syndrome phenotype to background strain in mice indicates that non-Scn1a genes are influencing the severity of the phenotype, which are termed ‘genetic modifier’ genes and may serve as novel drug targets. G protein-couple receptor 55 (Gpr55) was

79 identified as a genetic modifier candidate when its mRNA was shown to be more highly expressed in the hippocampus of seizure-susceptible F1 (129S6/SvEvTac x C57BL/6J) mice than the seizure- resistant 129S6/SvEvTac mice in an RNA-seq analysis of a Scn1a+/- mouse model (Hawkins et al.,

2019).

GPR55 is being investigated as a target for pathologies including cancer, metabolism inflammation and pain (Alhouayek, Masquelier & Muccioli, 2018; Simcocks et al., 2018), but our understanding of its function in the CNS is still developing. We know that Gpr55 receptors are expressed in brain regions including the hippocampus and cortex (Henstridge et al., 2011; Ryberg et al., 2007; Wu et al., 2013). Moreover, they couple to Gα12/13 and/or Gαq proteins in a ligand- dependent manner with their activation leading to an increase in cell excitability, predominantly through increased intracellular Ca2+ concentrations (Henstridge et al., 2009; Lauckner et al., 2008;

Oka et al., 2010; Ryberg et al., 2007). GPR55’s endogenous ligands include L-α- lysophosphatidylinositol (LPI) analogues, which are specific for GPR55 (Kapur et al., 2009; Oka et al., 2007; Oka et al., 2009), along with the endocannabinoids anandamide and 2-AG, their congener palmitoylethanolamide (PEA), and by various peptides including pituitary adenylate cyclase activating peptide-27 (PACAP-27) (Foster et al., 2019; Ryberg et al., 2007).

Supporting Gpr55’s role as a genetic modifier and its potential to influence the Dravet syndrome phenotype is its ability to modulate neuronal excitability. Within the hippocampus,

Gpr55 receptors have been shown to be distributed presynaptically on glutamatergic neurons

(Sylantyev et al., 2013; Wu et al., 2013), with their activation stimulating glutamatergic release and subsequently promoting excitatory transmission (Deliu et al., 2015; Hurst et al., 2017; Kramar et al., 2017; Lauckner et al., 2008; Sylantyev et al., 2013). Moreover, inhibition of Gpr55 has been attributed as the anticonvulsant mechanim of cannabidiol (CBD), which has proven anticonvulsant

80 in three phase III trials for patients with Dravet syndrome (Devinsky et al., 2017; Devinsky et al.,

2019; Devinsky et al., 2018a) and in two Scn1a+/- mouse models (Anderson et al., 2019a; Kaplan et al., 2017).

In this study we aim to utilise a Scn1a+/- mouse model of Dravet syndrome to measure key constituents of the Gpr55-LPI system with quantitative PCR and mass spectrometry. Then we aim to investigate the contribution of Gpr55 to the severe epilepsy phenotype by examine the impact of heterozygous Gpr55 deletion on key Scn1a+/- phenotype features. We hypothesise that Gpr55 mRNA expression will be greater in F1 seizure-susceptible mice and that genetic downregulation of Gpr55 expression will attenuate features of the Dravet syndrome phenotype.

Materials and Methods

Animals

All research and animal care procedures were approved by the University of Sydney Animal

Ethics Committee in agreement with the Australian Code of Practice for the Care and Use of

Animals for Scientific Purposes and by the Northwestern University Animal Care and Use

Committee in accordance with the National Institutes of Health Guide for the Care and Use of

Laboratory Animals. Scn1a+/- mice were generated by targeted deletion of Scn1a exon 1 as previously described (Miller et al., 2014) and were maintained by continuous backcrossing to the

129S6/SvEvTac (129) background strain (129.Scn1a+/-). For the Australian studies, 129.Scn1a+/- mice were obtained from The Jackson Laboratory (MMRRC Stock No: 37107-JAX). To generate mice for experiments, 129.Scn1a+/- mice were crossed with C57BL/6J (B6) resulting in [129 x B6]

F1.Scn1a+/- mice and wildtype mice (WT). Mice with homozygous Gpr55 deletion on a F1 (129 x

B6, 50:50) strain background (F1.Gpr55-/-) were received as a gift from Prof Ken Mackie, Indiana

University, and were rederived (Australian BioResources) using B6 mice to generate mice heterozygous for Gpr55 on a N2 (129 x B6, 75:25) background strain (N2.Gpr55+/-). The gene transcription studies were performed at Northwestern University, U.S.A, where the mouse facility operated on a 14 h light/10 h dark cycle. All other work was performed at the University of Sydney,

Australia, where the facility was kept on a 12 h light/dark cycle. All animals were group housed in IVC cages and had ad libitum access to food and water. For Scn1a and Gpr55 genotyping, primers and PCR conditions were performed as previously described (Miller et al., 2014; Wu et al., 2013, respectively).

Quantitative digital droplet PCR (ddRT-PCR)

Confirmation of transcript expression with ddRT-PCR for our genes of interest was performed as previously described (Hawkins et al., 2019; Hawkins et al., 2016). Briefly, hippocampal and cortical samples were dissected from male and female WT and Scn1a+/− littermates from both 129 and F1 background strains at P24. In order to sufficiently increase RNA yield for gene analysis, tissue from 3 – 4 mice, with at least one male and female in each group to control for potential sex-dependent differences, were combined to generate primary RNA pools. Primary pools were used to evaluate Gpr55 mRNA expression. Primary RNA pools (3 – 4) were combined into super

RNA pools (n = 12 – 16 mice/superpool) to evaluate Pla2g4a, Mboat7, Ddhd1 and Lclat1 mRNA expression. First-strand cDNA was synthesized then ddRT-PCR was performed using TaqMan

Gene Expression Assays (Life Technologies) for mouse Ddhd1 (Mm00616337_m1); Gpr55

(Mm02621622_s1); Lclat1 (Mm01237235_m1); Mboat7 (Mm00513220_m1); Pla2g4a

(Mm00447040_m1); Tbp (Mm00446971_m1). After amplification, droplets were analysed and relative transcript levels of target genes were expressed as a ratio of the gene to Tbp concentrations.

Cell culture

HEK wildtype cells and those expressing human GPR55 (hGPR55) were obtained from the

Medical University of Graz (Kargl et al., 2012). Cells were cultured in Dulbecco’s Modified

Eagle’s Medium (Sigma-Aldrich) supplemented with 10% foetal bovine serum (Sigma-Aldrich) and 100 U penicillin plus 100 μg/mL streptomycin (Gibco) and grown at 37°C and 5% CO2.

Additionally, hGPR55 cells were grown under selection (800 μg/mL G418; Invitrogen). When confluent, cells were lysed in RIPA plus complete protease inhibitor (Roche) with constant agitation for 35 min at 4°C. Cells were centrifuged at 18,000 g for 30 min (4°C) and the supernatant was stored at -20°C until analysed.

LPI analysis

Quantification of 1-stearoyl (18:0) and 2-arachidonoyl (20:4) LPI were determined via high- performance liquid chromatography tandem mass spectrometry (HPLC/MS/MS). Male WT and

Scn1a+/− littermates (P24) from both 129 and F1 backgrounds were euthanized via cervical dislocation between 12:00 and 12:15. Brains were rapidly extracted and flash frozen with liquid nitrogen. Hippocampal sections were dissected on ice then snap-frozen in liquid nitrogen and stored at -80°C until analysed. Hippocampal dissections were homogenized in methanol (3000 µL) and spiked with 17:1 LPI as an internal standard. Homogenates were centrifuged at 20,000 g for

20 min at 4°C. Supernatants underwent solid phase extraction using 500 mg C18 solid-phase extraction columns (Agilent Technologies, Santa Clara, CA) to collect lipids in an elution of 65%,

85% and 100% methanol. Elutants (20 μL) were separated using a C18 Zorbax reversed-phase analytical column (Agilent Technologies, Santa Clara, CA) coupled to a Shimadzu 8030 triple quadrupole mass spectrometer. The mass spectrometer was operated in positive electrospray

83 ionization mode with multiple reaction monitoring. LPI analogues, 1-stearoyl (18:0) and 2- arachidonoyl (20:4) were quantified by comparing samples to standards prepared with known amounts of each analogue. Total LPI content was determined by combining concentrations from each of the elutions. Internal standards, 17:1 Ly.so PI, 18:0 Lyso PI and 20:4 Lyso PI, were purchased from Avanti Polar Lipids (USA).

Genetic deletion of Gpr55 in Scn1a+/- mice

To investigate the contribution of Gpr55 to the Scn1a+/- phenotype, Scn1a+/- mice with or without heterozygous deletion of Gpr55 were generated. N2.Gpr55+/- mice were crossed with

+/- +/- 129.Scn1a mice to generate Scn1a mice on a mixed 129:B6 (62.5:37.5) background strain.

Scn1a+/- ; Gpr55+/+ and Scn1a+/- ; Gpr55+/- mice were used for experiments. Different cohorts of mice were used for hyperthermia-induced seizure testing to those used to assess spontaneous seizures and survival.

Hyperthermia-induced seizures

Hyperthermia-induced seizures were performed in mice aged P14 – 16 as previously described (Hawkins et al., 2017) using a rodent temperature regulator (TCAT-2DF, Physitemp

Instruments, Inc, Clifton, NJ) reconfigured with a Partlow 1160 + controller (West Control

Solutions, Brighton, UK) that was connected to a heat lamp and RET-3 rectal temperature probe.

Briefly, core body temperature was elevated using an overhead heatlamp by 0.5°C every 2 min until a generalized tonic-clonic seizure (GTCS) occurred or 42.5°C was reached. If body temperature reached 42.5°C, then it was maintained for 3 min and if no GTCS occurred within this time then the mouse was considered seizure-free.

Spontaneous seizure frequency and survival monitoring

Spontaneous seizure frequency monitoring was performed as previously described

(Hawkins et al., 2017). At P18, a single, brief GTCS was induced by hyperthermia as described above then immediately terminated by rapidly cooling body temperature back down to 37°C with the use of a cooling pad. Spontaneous GTCS were then quantified over 60 h of continuous video recording from 12:00 on P19 through to 00:00 on P22. The number of spontaneous GTCS were quantified from the video recordings by a researcher blind to the conditions. On P22 the mice were returned to their home-cages and survival was monitored until P30.

Statistical analysis

SPSS software was used to perform statistical analyses, version 24 (IBM Corp, Armonk,

New York USA). For ddRT-PCR transcript and LPI analyses, statistical comparisons were made using two-way ANOVA. If a strain by genotype interaction effect was identified, then Bonferroni’s planned comparisons were employed (129.WT vs 129.Scn1a+/-; F1.WT vs F1.Scn1a+/-; 129.WT vs

F1.WT; 129.Scn1a+/- vs F1.Scn1a+/-). Log-rank Mantel-Cox was used to compare data from hyperthermia-induced seizure and survival experiments. For spontaneous seizures, statistical comparisons were conducted using Fisher’s exact test (proportion seizure-free and proportion of hindlimb seizures) and Mann-Whitney U test (seizure frequency). No significant sex differences were observed across all observable measures, so groups were collapsed across sexes.

Comparisons were deemed statistically significant if p < 0.05.

Results

Figure 1. Hippocampal expression of Gpr55 transcript was elevated in F1 seizure- susceptible mice. Relative hippocampal Gpr55 transcript expression of WT (blank columns) and Scn1a+/- (coloured columns) mice on 129 and F1 genetic background strains. Expression was determined by ddRT-PCR and expressed as a ratio to Tbp (n = 7 – 11 biological replicates). Data is represented as the mean + SEM, with squares representing primary RNA pools (3 – 4 mice/primary RNA pool). Statistical comparisons between groups were made using a two-way ANOVA.

Hippocampal Gpr55 transcript expression is elevated in the F1 seizure- susceptible mouse strain

In order to confirm RNA-seq findings reported by Hawkins et al. (2019), we conducted ddRT-PCR to measure Gpr55 mRNA expression in the hippocampus of WT and Scn1a+/- mice on both 129 and F1 genetic backgrounds. Hippocampal Gpr55 mRNA expression was significantly higher in mice on the F1 seizure-susceptible strain compared to mice on the 129 seizure-resistant

strain (Figure 1; two-way ANOVA: strain, F1,31 = 63.102, p < 0.0001). There was no effect of genotype (F1,31 = 0.811, p = 0.375) nor a strain by genotype interaction (F1,31 = 0.000, p = 0.986).

We then aimed to evaluate whether this difference in mRNA expression would correlate with a change in Gpr55 protein expression. Unfortunately, we were unable to address this question as three commercially-available GPR55 antibodies lacked specificity for Gpr55 (see Supplementary material).

Figure 2. Hippocampal concentrations of LPIs in WT and Scn1a+/- mice on 129 and F1 background strains. Hippocampal concentrations of (a) 2-arachidonoyl LPI and (b) 1-stearoyl LPI, the main endogenous ligands for Gpr55, in WT (blank columns) and Scn1a+/- (coloured columns) mice on 129 and F1 genetic background strains were determined with HPLC/MS/MS (n = 7 – 9 mice). Data is shown as the mean + SEM, with circles representing individual mice. Statistical comparisons between groups were made using a two-way ANOVA, followed planned Bonferroni’s comparisons with *p < 0.05.

Hippocampal concentrations of 2-arachidonoyl LPI and 1-stearoyl LPI were lower in F1 mice and heterozygous Scn1a deletion selectively reduced 2- arachidonoyl LPI on the 129 strain.

While there are 6 LPI analogues (1-palmitoyl (16:0), 1-stearoyl (18:0), 1- or 2-oleoyl (18:1),

2-linoleoyl (18:2) and 2-arachidonoyl (20:04) LPI) known to activate Gpr55 (Oka et al., 2009), 2- arachidonoyl LPI and 1-stearoyl LPI are the most abundant in rodent brains (Inloes, Jing & Cravatt,

2018; Oka et al., 2009). Additionally, they are the most potent and efficacious ligands for human

GPR55-expressing cells (Oka et al., 2009). Therefore, we measured the hippocampal concentrations of these two main LPI analogues, 2-arachidonyl and 1-stearoyl LPI, using HPLC-

MS/MS. We determined that heterozygous Scn1a deletion selectively reduced hippocampal 2- arachidonoyl LPI concentrations but only in mice on the seizure-resistant 129 background strain

(Figure 2a). This was supported by a significant strain by genotype interaction (two-way

ANOVA: F1, 28 = 5.675, p = 0.024), with no main effects of strain (F1, 28 = 0.618, p = 0.438) or genotype (F1, 28 = 1.954, p = 0.173). Planned Bonferroni comparisons confirmed that 2- arachidonoyl LPI concentrations were significantly greater in 129.WT mice compared to

129.Scn1a+/- mice (p = 0.015), and compared to F1.WT mice (p = 0.038). In comparison, a strain- dependent effect was observed for 1-stearoyl LPI, with significantly lower concentrations in the

F1 mice compared to the 129 mice (Figure 2b; two-way ANOVA: main effect of strain, F1, 29 =

5.902, p = 0.022; no main effect of genotype, F1, 29 = 1.635, p = 0.211; and no strain by genotype interaction, F1, 29 = 0.853, p = 0.184).

acid-preferring phospholipase A1 (PA-PLA1; Ddhd1) and phospholipase A2 (cPLA2; Pla2g4a), respectively. While 2-arachidonoyl LPI and 1-stearoly LPIs are converted back to PI through 1- acylglycerol-3-phosphate acyltransferase 8 (AGPAT8; Lclat1) and membrane-bound O- acyltransferase 7 (MBOAT7; Mboat7), which is also known as LPI-acyltransferase-1 (LPIAT1). Transcript expression of (b) Pla2g4a, (c) Mboat7, (d) Ddhd1 and (e) Lclat1 in WT (blank columns) and Scn1a+/- (coloured columns) mice on 129 and F1 genetic background strains. Expression was determined using ddRT-PCR and expressed as a ratio of Tbp (n = 3 – 5 biological replicates). Data represents the mean + SEM, with triangles representing mRNA superpools (n = 12 – 16 mice/superpool). Statistical comparisons between groups were made using a two-way ANOVA.

Heterozygous Scn1a deletion decreased hippocampal expression of the LPI metabolising enzyme Mboat7.

Since significant differences in LPI levels were observed across strains, we aimed to determine whether this resulted from differences in the synthesis or degradation. We evaluated the mRNA expression of the best-characterised enzymes involved in the synthesis and metabolism of these LPI analogues using ddRT-PCR (Figure 3a). Briefly, 2-arachidonoyl LPI is synthesized by

PA-PLA1, which is encoded by Ddhd1, while 1-stearoly LPI in synthesised by cPLA2, encoded by

Pla2g4a (Alhouayek, Masquelier & Muccioli, 2018). Then, 2-arachidonoyl LPI and 1-stearoly

LPIs are converted back to their precursor by AGPAT8, encoded by Lclat1, and MBOAT7, encoded by Mboat7, respectively (Alhouayek, Masquelier & Muccioli, 2018). We found no effects of strain or genotype in the mRNA expression of Pla2g4a or Lclat1 (Figures 3b and e, respectively). Although Ddhd1 mRNA expression appears to be greater in the F1 seizure- susceptible strain compared the 129 seizure-resistant strain, the difference was not statistically significant (Figure 3d; two-way ANOVA, strain: F1, 11 = 3.826, p = 0.076). However,

90 heterozygous deletion of Scn1a significantly reduced Mboat7 mRNA expression (Figure 3c; two- way ANOVA: genotype, F1, 10 = 6.932, p = 0.025), with no effect of strain (F1, 10 = 0.28, p = 0.609) nor a strain by genotype interaction (F1, 10 = 0.123, p = 0.733, respectively).

Figure 4. Heterozygous Gpr55 deletion was anticonvulsant and restored the survival deficit in Scn1a+/- mice. (a) Threshold temperatures for hyperthermia-induced GTCS of individual Scn1a+/- mice with WT expression of Gpr55 (Scn1a+/- ; Gpr55+/+) or heterozygous deletion of Gpr55 (Scn1a+/- ;Gpr55+/-). Heterozygous deletion of Gpr55 significantly increased the temperature threshold for GTCS of Scn1a+/- mice (green, open symbols). Data is shown as average temperatures of GTCS ± SEM, with open circles representing individual mice (n = 14 – 16 mice, log-rank Mantel-Cox, *p < 0.05). (b) Spontaneous GTCS frequency of individual animals (open symbols) quantified over a 60 h period (n = 16 – 18 mice, Fisher’s exact test). Expression of Gpr55

91 had no effect on spontaneous seizure frequency or the proportion of mice that had spontaneous seizures. (c) Total number of spontaneous GTCS with or without tonic hindlimb extension. Heterozygous Gpr55 expression significantly the proportion of seizures that progressed to tonic hindlimb extension (n = 16 – 18 mice, Fisher’s exact test, *p < 0.05). (d) Survival curves comparing Scn1a+/- ; Gpr55+/+ and Scn1a+/- ; Gpr55+/ mice. Survival of Scn1a+/- mice was significantly improved with heterozygous Gpr55 deletion (green line) (n = 16 – 18 mice, log-rank Mantel-Cox, p < 0.05).

Heterozygous Gpr55 deletion has anticonvulsant effects against hyperthermia- induced seizures and restores survival in Scn1a+/- mice.

We utilized Gpr55+/- mice to evaluate whether the increased Gpr55 transcript expression of seizure-susceptible F1 mice contributed to the severe epilepsy phenotype of F1.Scn1a+/- mice. We generated Scn1a+/- mice that had WT expression of Gpr55 (Scn1a+/- ; Gpr55+/+) or heterozygous deletion of Gpr55 (Scn1a+/- ; Gpr55+/-) and then assessed the effect on hyperthermia-induced seizures, spontaneous seizures and survival. We observed that heterozygous Gpr55 deletion elicited an anticonvulsant effect against hyperthermia-induced seizures in Scn1a+/- mice. Scn1a+/-

; Gpr55+/- mice had a significantly increased temperature thresholds for GTCS compared to

Scn1a+/- ; Gpr55+/+ mice (Figure 4a; log-rank Mantel-Cox test: p = 0.038).

We then examined whether heterozygous Gpr55 deletion would be anticonvulsant against spontaneous seizures (Figure 4b). Spontaneous seizures were observed in 39% (7/18) of Scn1a+/-

; Gpr55+/+ mice with an average frequency of 0.08 GTCS per hour, compared to 19% (3/16) of

Scn1a+/- ; Gpr55+/- mice with an average frequency of 0.02 GTCS per hour (Figure 4b). While it appeared as though heterozygous Gpr55 deletion reduced the number of Scn1a+/- mice that

2 exhibited spontaneous seizures, it was not statistically significant (Fisher's Exact test: χ 1 = 1.655,

92 p = 0.099). Nor was there a difference in overall seizure frequency (Mann-Whitney U test: U =

115, p = 0.217). In addition to quantifying seizure frequency, we also monitored the effect of heterozygous Gpr55 deletion on seizure severity. Each GTCS seizure was scored for progression to the most severe stage of full tonic hindlimb extension (hindlimbs at 180 to torso). Interestingly, despite Scn1a+/- ; Gpr55 +/- mice exhibiting a smaller number of total seizures, a significantly larger proportion of the total seizures advanced to hindlimb extension (10/23 = 0.435) compared to

+/- +/+ 2 Scn1a ; Gpr55 mice (18/83 = 0.217) (Figure 4c; Fisher's exact test: χ 1 = 4.4, p = 0.036).

Finally, we assessed the effect of heterozygous Gpr55 deletion on survival of Scn1a+/- mice.

Survival of Scn1a+/- mice was significantly improved with heterozygous deletion of Gpr55

(Figure 4d; Mantel-Cox log-rank test: p = 0.025). All Scn1a+/- ; Gpr55+/- mice (n = 16) survived to P30 compared to only 72% (13/18) of the Scn1a+/- ; Gpr55+/+ mice.

Discussion

Gpr55 was highlighted as a potential genetic modifier when an RNA-seq analysis reported more than double the mRNA expression in the seizure-susceptible F1 background strain compared to the seizure-resistant 129 background strain in a Scn1a+/- mouse model (Hawkins et al., 2019).

In support of Gpr55’s potential to influence the Dravet syndrome phenotype is Gpr55 receptor activation promoting hippocampal excitatory transmission (Hurst et al., 2017; Kramar et al., 2017;

Sylantyev et al., 2013), and CBD’s anticonvulsant efficacy being attributed to Gpr55 inhibition

(Kaplan et al., 2017). We hypothesised that the increased Gpr55 mRNA expression observed in the F1 may be increasing excitatory neurotransmission, thereby exacerbating the neuronal hyperexcitability and contributing to the severe seizure phenotype observed in F1.Scn1a+/- mice.

Therefore, this chapter further investigated the potential of Gpr55 as a genetic modifier in the

Scn1a+/- mouse model of Dravet syndrome.

We determined that mice on the mixed F1 seizure-susceptible background strain express significantly more hippocampal Gpr55 transcript than mice on the 129 seizure-resistant strain, reaffirming that Gpr55 may be a genetic modifier of Scn1a. Additionally, we demonstrated that hippocampal concentrations of 2-arachidonoyl and 1-stearoyl LPI, the major endogenous ligands of Gpr55, were altered in distinct strain- and genotype-dependent manners. Moreover, we also demonstrated that heterozygous deletion of Gpr55 was anticonvulsant against hyperthermia- induced seizures and ameliorated the survival deficit normally observed in the F1.Scn1a+/- mice.

We found significantly less 1-stearoyl LPI levels in mice on the F1 genetic background compared to those on a 129 background strain. This strain-dependent reduction was only apparent in WT mice for 2-arachidonoyl LPI. Intriguingly, on the 129 background strain specifically, significantly less 2-arachidonoyl LPI was measured in Scn1a+/- mice than in WT mice. We hypothesise that the observed reduction of LPI concentrations in Scn1a+/- mice on the 129 background may be an adaptive mechanism to counter the neuronal hyperexcitability caused by

NaV1.1 haploinsufficiency. Reduced concentrations of 2-arachidonoyl LPI would likely lead to less Gpr55 activation and consequently less network excitability. F1.Scn1a+/- mice did not exhibit this same reduction in 2-arachidonoyl LPI levels compared to F1.WT mice, which in the context of the increased Gpr55 expression of F1 mice could contribute to the severe epilepsy phenotype observed in F1.Scn1a+/- mice. Future studies might investigate whether administration of 2- arachidonoyl LPI or a brain-penetrant GPR55 agonist would triggers seizures in 129.Scn1a+/- mice.

We wanted to determine whether the differences that we observed in LPI levels were the result of inherent differences in expression of the synthesizing or metabolizing enzymes.

Therefore, this chapter also assessed transcript expression of the best-characterised LPI- modulating enzymes. We report that heterozygous deletion of Scn1a reduced expression of

Mboat7, an enzyme that metabolizes 1-stearoyl LPI, irrespective of background strain. However, this does not correlate with the differences observed with 1-stearoyl LPI, which showed a strain- dependent effect with reduced levels in F1 mice. The other enzymes examined showed no significant differences in gene expression.

Although differences in mRNA expression of these enzymes does not account for the observed differences in LPI levels, it does not discount that there could be differences in protein expression. Future experiments could determine whether mRNA expression of these enzymes correlates with a functional change in protein expression. In this chapter we chose to investigate the expression of the major synthesizing and metabolizing enzymes for LPI but the complexity of the pathways involved in LPI metabolism may mean we missed additional key enzymes. For example, LPI analogues can also degraded by lysophospholipase A isoforms 1 and 2 (Wepy et al.,

2019), an isoform of phospholipase C specific for LPI for which the gene has not yet been identified (Murase & Okuyama, 1985; Tsutsumi et al., 1994; Ueda et al., 1993), autotaxin, a lysophospholipase D isoform (Aoki et al., 2002; Xie & Meier, 2004), and a/b hydrolase domain 6

(Thomas et al., 2013).

One noteworthy finding from the work in this chapter was that heterozygous Scn1a deletion downregulated Mboat7 transcript expression irrespective of background strain. As one of 1- stearoyl LPI’s degrative enzymes, the reduction in Mboat7 transcript expression did not correspond with the reduced 1-stearoyl LPI concentrations observed. Mboat7 is an LPI-specific acyltransferase involved in phospholipid remodelling of membranes (Gijon et al., 2008).

Specifically, Mboat7 transfers arachidonic acid to 1-stearoyl LPI to form 1-stearoyl-2- arachidonoyl PI, an important membrane constituent (Anderson et al., 2013; Lee et al., 2012).

Mboat7 has also been implicated in the synthesis and metabolism of additional lipid mediators,

95 including other species of LPI, PIs, and phosphatidylinoethanolamines (PEs) (Anderson et al.,

2013; Lee et al., 2012). This suggests that the hippocampal concentrations of these other lipid mediators may be altered by heterozygous Scn1a deletion and calls for a more comprehensive lipidomic investigation of the Scn1a+/- mouse model that future work could address.

Interestingly, loss-of-function gene variants of MBOAT7 have been associated with a form of intellectual disability that is accompanied by autism spectrum disorder and epilepsy in humans

(Jacher et al., 2019; Johansen et al., 2016; Yalnizoglu et al., 2019). Moreover, mice with homozygous Mboat7 deletion exhibited cortical and hippocampal atrophy due to disordered cortical lamination and premature death (Anderson et al., 2013; Lee et al., 2012). Dravet syndrome patients and Scn1a+/- mouse models have also reported associated behavioural comorbidities including cognitive impairments, along with cortical atrophy (particularly white matter) and decreased lifespan (Brunklaus et al., 2012; Dravet, 2011; Genton, Velizarova & Dravet, 2011; Han et al., 2012; Ito et al., 2013; Perez et al., 2014; Striano et al., 2007; Chapter 4).

The reduced Mboat7 expression resulting from heterozygous deletion of Scn1a might contribute to the epilepsy phenotype observed in Scn1a+/- mice. However, the reduction in Mboat7 was observed in both the seizure-resistant 129 and the seizure-susceptible F1 strains, suggesting that reduced Mboat7 is less likely to contribute to seizure phenotypes since the overt seizure phenotype is absent in 129.Scn1a+/- mice. However, the behavioural phenotype of 129.Scn1a+/- mice has not been assessed and it is possible that 129.Scn1a+/- mice could exhibit behavioural impairments in the absence of a seizure phenotype. This theory is supported by evidence demonstrating that the Dravet syndrome seizure and behavioural-deficit phenotypes arise from

NaV1.1 haploinsufficiency in differing regions of the brain (Bender et al., 2013; Cheah et al.,

2012). Future work could directly examine the impact of Mboat7 expression on the behavioural phenotypes of Scn1a+/- mice.

Finally, we sought to examine the impact of heterozygous Gpr55 deletion on the epilepsy phenotype observed in the seizure-susceptible Scn1a+/- mice. Indeed, we found that heterozygous deletion of Gpr55 in Scn1a+/- mice on a mixed 129:B6 (62.5:37.5) background was anticonvulsant against hyperthermia-induced seizures and significantly improved survival. Heterozygous deletion of Gpr55, however, had no significant effect on spontaneous seizures. These findings align with

CBD, a Gpr55 antagonist, demonstrating anticonvulsant efficacy against hyperthermia-induced seizures (Anderson et al., 2019a; Kaplan et al., 2017). Our findings are also consistent with inhibition of Gpr55 reducing the frequency of action potential outputs and increasing the frequency of spontaneous inhibitory postsynaptic currents in granule cells from Scn1a+/- hippocampal slices

(Kaplan et al., 2017).

Although we found that heterozygous deletion of Gpr55 was anticonvulsant against hyperthermia-induced seizures and improved survival of Scn1a+/- mice, there was a major limitation to these studies. Unfortunately, the Gpr55+/- mice we acquired were not congenic on a

B6 background but rather were on a mixed 129:B6 (25:75) genetic background. As such, when these mice were crossed with 129.Scn1a+/- mice for seizure experiments, the genetic background of the progeny would have varied ranging from 50 to 75% of 129. Since the background strain directly influences the severity of the Scn1a+/- phenotype and congenic 129.Scn1a+/- mice exhibit no overt phenotype, the anticonvulsant effect of heterozygous Gpr55 deletion may have been overestimated and the increased presence of modifier loci from the 129 background strain could contribute to the observed anticonvulsant effect. This would also explain why we observed a lower rate of spontaneous seizures and reduced mortality in the Scn1a+/- ; Gpr55+/+ mice compared to

97 those previously reported for Scn1a+/- mice (Hawkins et al., 2017; Miller et al., 2014; Mistry et al.,

2014). In order to unambiguously determine the whether Gpr55 is a genetic modifier of Scn1a+/- mice, the genetic-cross experiments would need to be repeated using Gpr55+/- mice on a congenic

B6 background.

In summary, we present evidence that suggests Gpr55 may be a genetic modifier of the

Scn1a+/- phenotype. We confirmed that seizure-susceptible F1 mice express significantly more hippocampal Gpr55 mRNA and report reduced hippocampal concentrations of 1-stearoyl LPI. We also report that 2-arachidonoyl LPI was reduced by heterozygous deletion of Scn1a but only on the seizure-resistant 129 background strain. The work in this chapter also demonstrated that heterozygous deletion of Gpr55 was anticonvulsant against hyperthermia-induced seizures and rescued survival in seizure-susceptible Scn1a+/-. Although there are limitations to the research in this chapter, our findings further reinforce the view that Gpr55 may be a novel therapeutic target in the management of Dravet syndrome.

Evaluation of Gpr55 protein expression was unsuccessful

We aimed to examine whether the strain-dependent difference in Gpr55 mRNA expression translated to a functional difference in protein expression. Unfortunately, assessing Gpr55 protein expression was unsuccessful due to commercially-available Gpr55 primary antibodies proving inadequate (Supplementary figure 1). We tested three commercially-available antibodies

(ThermoFisher, Abcam and Cayman Chemical), none of which exhibited specificity or selectivity for Gpr55. The predicted protein fraction for Gpr55 was 37 kDa with possibly heavier fractions occurring due to post-translational modifications. To validate the antibodies, we used tissue from wildtype and Gpr55-null mice, with genotypes confirmed using PCR (Supplementary figure 1a).

The first antibody that we used was from ThermoFisher (720285) and it produced strong signals at ~80 kDa in the hippocampus (Supplementary figure 1b.i, 5-40 µg) and at ~30 kDa in whole- brain lysate (Supplementary figure 1b.ii, 100 µg); however, the signal was equally present in both WT and Gpr55-/- tissue.

The second antibody tested was from Abcam (Ab203663), which produced a strong signal at ~35 kDa in whole-brain lysates (Supplementary figure 1c, 100 µg), but again was present in the Gpr55-/- lysates. The third antibody tested was from Cayman Chemical (10224) and it produced a signal at ~42 kDa in whole-brain lysate (30-100 µg) but was again present in Gpr55-/- tissue

(Supplementary figure 1d.i-ii). A Gpr55 blocking peptide effectively blocked all the signals that were produced by the Cayman antibody (Supplementary figure 1d.iii). For further validation, we also tested the Cayman antibody on WT HEK cells and HEK cells expressing human GPR55 receptors (Supplementary figure 1d.v, 100 µg). There was a weak signal around the 75 kDa mark that was specific to the GPR55-positive cells that could potentially be a GPR55 dimer. We attempted to denature the potential GPR55 dimer but saw no change after incubation at 60°C for

100

15 min; and all signals were blocked by the Gpr55 blocking peptide (Supplementary figure

1d.vi).

Western blot materials and methods

Western blot analyses were performed as previously described (Hawkins et al., 2019;

Hawkins et al., 2016) with modifications. Briefly, protein from hippocampal and whole-brain sections were isolated from Gpr55+/+ and Gpr55-/- adult mice. Tissue samples were homogenised by hand in ice cold 0.32 M sucrose, 5 mM Tris (pH 7.4) supplemented with protease inhibitor

(Roche) before being centrifuged for 10 min at 2,000G, at 4°C. Supernatants were collected and centrifuged for 40 min at 20,000 G, at 4°C. The pellets were re-homogenised in radioimmunoprecipitation assay (RIPA) solution (5 mM Tris (pH 8.2) and 1 mM

Ethylenediaminetetraacetic acid (EDTA)) supplemented with protease inhibitor (+). Samples were centrifuged again for 40 min at 20,000 G, at 4°C, then the procured pellet was resuspended in a solution of 50 mM Tris (pH 7.5), 10 mM EDTA supplemented with protease inhibitor. For both cells and tissue, protein levels were quantified using Bradford reagent (Bio-Rad) on a NanoDrop

One. 5-100 μg of protein lysate was denatured with laemmli sample buffer (Bio-Rad) and 5% beta mercaptoethanol for 15 min at room-temperature, or 15 min at 60°C where specified. Proteins were separated on a 10% SDS-page gel and transferred to nitrocellulose membranes. Membranes were probed for GPR55/Gpr55 using rabbit polyclonal antibodies (1:200; ThermoFisher

Scientific, 720285; Abcam, Ab203663; Cayman chemical, 10224) and for β-tubulin using a mouse monoclonal antibody (1:500, Sigma-Aldrich, T5201) for 1 h at room temperature. Thereafter, membranes were incubated with secondary antibodies, anti-Rabbit Alexa Fluor 800 (Thermo

Fisher Scientific, A32735) and anti-Mouse Alexa Fluor 680 (Thermo Fisher Scientific, A28183) at 1:20,000 for 30 min at room temperature. The probed membranes were imaged with an Odyssey

101 imaging system (Licor) and relative protein levels were determined by densitometry using

ImageStudio software (Licor).

102

Chapter 4

Adolescent behavioral abnormalities in a Scn1a+/- mouse

model of Dravet syndrome

104

1. Introduction Approximately 80% of patients with Dravet syndrome have a mutation in SCN1A, the gene Dravet syndrome is a rare and catastrophic form

encoding the α1-subunit of the voltage-gated neuronal of intractable childhood epilepsy with an incidence of

sodium channel (NaV1.1) [19]. SCN1A mutations approximately 1 in 15,000 to 40,000 [1–5] and a high associated with Dravet syndrome are typically severe premature mortality rate [6,7]. Dravet syndrome missense and nonsense mutations that result in typically presents with febrile seizures within the first haploinsufficiency [20]. Mice with heterozygous year of life in previously healthy infants, which deletion of Scn1a (Scn1a+/− mice) recapitulate many progress to both febrile and afebrile aspects of Dravet syndrome, such as susceptibility to seizures and the development of psychomotor and hyperthermia-induced seizures, the development of cognitive impairments [8]. Commonly reported spontaneous seizures, and reduced lifespan [21–23]. behavioral comorbidities that are associated with The present study sought to phenotype the behavior of Dravet syndrome include hyperactivity and attention the Scn1a+/− mouse line developed by targeted deletion disorders, motor disorders, intellectual disabilities, of exon 1 (Scn1atm1Kea), which has not yet been along with autism spectrum disorder-associated asocial characterized [22]. Previous studies have phenotyped and compulsive behaviors [9– 16]. Patients with Dravet the behavior of adult mice with heterozygous loss-of- syndrome also exhibit deficits in auditory sensorimotor function alleles, developed by targeted deletion of processing [13,17], which may precede and contribute Scn1a exon 26 (Scn1atm1Wac) or knock-in of the Scn1a- to the associated attention, language, and cognitive R1407X mutation (Scn1atm1Kzy), and described impairments [9–11,16,18]. Behavioral comorbidities hyperactivity, stereotyped behaviors, lowered are stronger predictors than seizure frequency of a sociability, poor spatial learning, and memory deficits lower quality of life [15,16], highlighting a dire need [24,25]. In the current study, we examined the for treatments that specifically target behavioral deficits aforementioned behaviors, as well as repetitive, during early development. compulsive behaviors, acoustic startle responses, and

sensorimotor gating, which have not been investigated

105 in Dravet syndrome mice. Furthermore, we assessed for Scn1a+/− mice, unless otherwise specified, with the whether any behavioral abnormalities in Scn1a+/− mice same mice being used for all experiments. Mice were were present during adolescence to more closely model genotyped as previously described [22] and were group the onset of developmental delays in attention, housed (3–4 mice per cage), unless otherwise specified, emotional, and cognitive function observed in patients in individually ventilated cages in a specific pathogen- with Dravet syndrome. Identifying behavioral free facility with food and water available ad libitum. impairments in adolescence would provide a platform All experiments were conducted during the light phase to test early intervention therapeutics aiming to correct of a 12-hour light/dark cycle. the behavioral disturbances. 2.2. Behavioral testing paradigm 2. Material and methods A schematic depiction of the sequence of behavioral tests can be seen in Fig. 1. The test mice 2.1. Mice were all individually handled for 3 min for three All research and animal care procedures were consecutive days before testing began. Behavioral approved by the University of Sydney Animal Ethics testing of age-matched WT and Scn1a+/− mice began at Committee and were in agreement with the Australian postnatal days (P) 34–37 during adolescence and after Code of Practice for the Care and Use of Animals for the vulnerable mortality period (50% of F1.Scn1a+/− Scientific Purposes. Scn1atm1Kea mice were purchased mice to P30) [22]. The behavioral tests were performed from the Jackson Laboratory (stock 37107-JAX; Bar in the following test order: 1) three-chambered social Harbor, MA, USA) and maintained by continuous preference, 2) open-field, 3) social interaction, 4) novel backcrossing to 129S6/SvEvTac (129.Scn1a+/−). To object recognition, 5) Barnes maze, 6) elevated plus generate mice for experiments, 129.Scn1a+/− mice were maze, 7) acoustic startle, 8) prepulse inhibition of startle crossed with wildtype (WT) C57BL/6J, resulting in F1 (PPI), 9) marble burying, and 10) contextual fear mice on a mixed 129S6/SvEvTac × C57BL/6J conditioning. Four days prior to behavioral testing, all background (Scn1a+/−). Mice used for experiments were mice were individually housed to enhance social male littermates with n = 11 for WT mice and n = 10

106 motivation in the three-chambered social preference aggressive or sexual interactions. Each trial was and the social interaction tests [26]. Mice remained conducted for 10 min. During the habituation phase individually housed until after the novel object (trial 1), the test mouse was placed in the empty arena. recognition test at which point they were returned to Social preference was tested during trial 2, where an group housing with their former cage mates. unfamiliar mouse was caged in one chamber and the

Additionally, mice were habituated to the testing room other cage was left empty. Preference for social for at least 60 min prior to each experiment. All arenas novelty was assessed during trial 3, with one chamber and equipment were cleaned with 80% ethanol in containing the mouse from trial 2 (familiar mouse) between trials. None of the mice exhibited overt seizure and the other a stranger mouse (novel mouse). All behavior during any of the video-recorded behavioral trials were captured with a video recording system tests. and tracking software (TopScan; Clever Sys Inc.,

U.S.A.) that measured distance traveled and time 2.3. Three-chambered social preference spent in a 5-centimeter radius of the cages. Social The three-chambered social preference test mice used for this experiment were age-matched male was performed as previously described with C57BL/6J mice habituated to testing conditions by modifications [27]. The apparatus consisted of a long being contained in experimental wire cages for 20 min rectangular blue Perspex box (60 × 42 × 22 cm) on three consecutive days prior to testing. One divided into three identical 20 cm long chambers Scn1a+/− mouse escaped from the arena during trial 3 separated by two transparent partitions, each with a (social novelty) so was excluded from trial 3 analyses. square opening (10 × 10 cm), with the light intensity in the center of the arena at 10 lx. The two chambers on either end of the apparatus contained a wire cage 2.4. Open-field test

(5 × 5 × 5 cm) that was used to hold a mouse. The The open-field test was conducted in a blue, wire cage allowed exposure to visual, auditory, Perspex, 40 × 40 × 40 cm arena (10 lx). Mice were olfactory, and some tactile stimuli while preventing placed into the arena for 10 min and video-recorded

107 by an overhead camera. TopScan was used to measure genotype using scoring software (ODLog; Macropod total distance traveled in the whole arena and Software Pty Ltd., Australia). differentiate activity that occurred in the inner (70%) 2.6. Novel object recognition test and outer (30%) regions. Time spent in the center of The novel object test assesses recognition the arena was used to assess anxiety-like behavior memory and was performed as previously described

[28]. with modifications in the open-field arena [32]. During

the familiarization phase, two identical objects were 2.5. Social interaction adhered to the floor of the arena on opposite corners, The social interaction test was performed as and mice were placed in the arena for 10 min. Mice previously described with modifications [29,30]. were returned to their home cages for 60 min before a Immediately following the open-field test, a stranger second 10-minute trial. Prior evaluation of object mouse was introduced into the arena for 10 min. The recognition performance in a Scn1atm1Wac mouse model stranger mice were male A/JArc mice (Australian utilized a 24-hour intertrial interval [24]. Here, we Resources Centre, Australia), aged 5–7 weeks old. An chose a 60-minute intertrial interval based on research overhead camera was used to record each session. demonstrating improved object recognition using a 60-

Fig 1 Sequence of behavioral tests Behavioral testing began at postnatal days 34–37 (average 35 5 ± 0 2) Three-chambered social preference (3CHSP) and prepulse inhibition of startle (PPI)

Total duration and frequency of social contact, minute delay compared with a 24-hour delay [32]. including sniffing, adjacent lying, following, In the second trial, mice were presented with the crawling under/over, and mutual grooming [29,31], familiar object from the previous trial and a novel were manually recorded by a researcher blinded to object. The objects used were polypropylene Falcon®

108 tubes (50 mL, 30 × 115 mm), filled with clean bedding holes led to false escape boxes (6 × 6 × 3 cm). The maze and Corning® U-shaped polystyrene cell culture flasks was located 70 cm off the ground and was placed under

(75 cm2) covered in red electrical tape. The objects and bright lights (1350 lx) with spatial cues placed on the their locations varied among animals and were walls and an overhead video camera recording each counterbalanced across genotypes. Sessions were session. The paradigm employed here was based on the recorded by an overhead camera, and TopScan (Clever paradigm utilized by Ito et al. [25] demonstrating

Sys Inc., U.S.A.) was used to measure total distance impairment in spatial learning and memory in traveled and time spent within a 5-centimeter radius of Scn1atm1Kzy mice. Mice were habituated to the Barnes each object. Additionally, for the test trial, time spent maze for 5 min, and to the escape box for 1 min, the day directly interacting with the objects, defined as nosing, before training. Mice received three training sessions sniffing, circling, rearing on, and climbing on the per day (5 min), with an intertrial period of 20 min, for objects, were manually recorded by a researcher four consecutive days. During training, mice started in blinded to group using ODLog scoring software. The a tube (15 × 15 × 20 cm) in the center of the maze then discrimination index (DI) was calculated as [(novel had 5 min to search for the escape box. The location of object interaction time − familiar object interaction the escape box remained consistent for the same animal time) / (total interaction time) ∗ 100], with DI ≥ 25 but was varied among animals with the location being indicating preference for the novel object [33]. 2.7. counterbalanced across conditions. If a mouse failed to

Barnes maze test find the target box within the 5-minute trial, it was

guided to the hole by the researcher. When the mouse Spatial learning and memory were assessed entered the escape box, it remained there for 1 min using the Barnes maze (San Diego Instruments, U.S.A.) before being returned to its homecage. On day 5, a [34], which consisted of a white circular platform (91 probe test was performed for 3 min in the absence of cm diameter) that had 20 circular holes (5 cm diameter) the escape box. Any-MAZE tracking software around its perimeter. An escape box (17 × 13 × 7 cm) (Stoelting Co., U.S.A.) measured distance traveled, was located under one of the holes, while the other

109 latency to find the target zone, time in the target zone, startle stimulus (120 dB, 40 ms) with a randomized and number of entries into the target zone. intertrial interval of 12–30 s. Immediately following the

acoustic startle test, PPI was assessed. Prepulse 2.8. The elevated plus maze inhibition of startle consisted of 50 trials comprised of The elevated plus maze was used to assess 10 repetitions each of the following: no stimuli, startle anxiety-like behavior [35] and was performed in low- stimulus alone (120 dB for 40 ms), and PPI trials with light conditions (150 lx). The arena consisted of four 3 different prepulse intensities presented in a arms (4.5 × 30 cm) made of blue Perspex and arranged pseudorandomized order with randomized intertrial in a “plus shape”. Two of the opposing arms were intervals of 12– 30 s. Prepulse inhibition of startle trials enclosed by walls (30 × 14.5 cm), and the arena was consisted of a 20-millisecond prepulse stimulus (74, 78, elevated 41 cm off the floor. At the commencement of and 82 dB) that were presented 100 ms before a 40- the test session, mice were placed in the center of the millisecond startle stimulus (120 dB). The mean startle arena and video-recorded from above for 10 min. amplitude during the 100 ms following the startle TopScan (Clever Sys Inc., U.S.A.) was used to measure stimulus was recorded as the startle response. The total distance traveled, the number of entries, and the percentage of PPI was calculated as [(startle response time spent in open and closed arms. 120 dB) − (prepulse + pulse response) / startle response

120 dB ∗ 100]. 2.9. Acoustic startle response and prepulse inhibition of startle 2.10. Marble burying

The PPI test was used to assess the startle reflex Marble burying was used to assess repetitive, and auditory sensorimotor gating. Tests were conducted compulsive-like behavior [37]. The test was conducted in SR lab chambers (San Diego Instruments, U.S.A.) in opaque Perspex cages (38 cm × 21 cm × 13 cm)

(50 lx) as previously described with modifications [36]. containing normal bedding (4.5 cm deep) overlaid with

First, the acoustic startle response was independently 20 glass marbles (15 mm diameter) equidistant in a 4 × assessed by exposing mice to 30 presentations of the 5 arrangement. Cages were set in a room with the light

110 intensity set to 15 lx. For testing, a single mouse was memory. This was repeated 24 h after to assess long- placed in the cage with the lid closed for 60 min. At the term memory. All trials were video-recorded, and end of the trial, the mouse was removed from the cage, FreezeScan (Clever Sys Inc., U.S.A.) software was used and a researcher blinded to genotype counted the to measure freezing behavior. number of marbles buried. Marbles with more than 2/3 2.12. Statistical analysis of its surface covered was counted as buried. Bedding Statistical analyses were performed using was replaced, and marbles were cleaned with 80% GraphPad Prism 7 software (GraphPad Software, La ethanol, rinsed with water, and dried between trials. Jolla California USA) and SPSS version 24 (IBM Corp,

2.11. Contextual fear conditioning Armonk, New York USA). Normality of distribution

Fear conditioning was conducted in shock and equality of variance were assessed using the chambers (Clever Sys Inc., U.S.A.) (5 lx) as previously D'Agostino & Pearson normality test and the F test, described with modification [38]. The behavioral respectively. If these parameters were violated, then paradigm employed was based on prior demonstration data were analyzed using nonparametric tests. The of impairments in fear memory in adult Scn1atm1Wac performance of WT and Scn1a+/− mice was generally mice [24]. Briefly, a mouse was placed in the chamber compared using unpaired parametric t-tests or for 198 s before receiving a single, footshock (2 s, 0.5 nonparametric Mann–Whitney U tests. For the three- mA). The mouse remained in the chamber for another chambered social preference tests, paired parametric t-

60 s before it was returned to its homecage. The first 3 tests or nonparametric Wilcoxon matched-pairs signed min of the habituation period was used as a measure of rank tests were employed to compare preferences. The baseline freezing, whereas the 60 s immediately data for the training phase of the Barnes maze and fear following the footshock was used to assess immediate conditioning were analyzed using repeated two-way fear response. Thirty minutes after conditioning, the analysis of variance (ANOVA), followed by Sidak's mice were reexposed to the shock-paired context in the post hoc tests. Prepulse inhibition of startle data were absence of shock for 3 min to assess short-term analyzed using nonparametric aligned ranks

111 transformation (ART) ANOVA by first aligning and genotype interaction, F9, 171 = 1.737, p = 0.0841). The ranking data for each variable using ARTool (version impact of Scn1a+/− genotype on anxiety-like behavior

1.6.2, Washington USA) then analyzing differences was also evaluated in the open-field test by between variables with a factorial ANOVA. The Log- comparing the time spent in the inner 70% of the rank Mantel–Cox test was used to analyze the ability to arena (Fig. 2c and d, representative traces). Overall, locate the target zone during the probe test of the Barnes Scn1a+/− mice spent significantly less time in the maze because one animal failed to enter the target zone. center region of the arena than WT mice, and they

Differences were deemed statistically significant when engaged in more “wall-hugging” thigmotactic p < 0.05. behavior (Fig. 2e, unpaired t-test: t19 =2.115, p =

0.044). Time-course analysis showed that while both

genotypes of mice increased time 3. Results

3.1. Scn1a+/− mice displayed normal locomotor activity but anxiogenic-like thigmotactic behavior in the open- field test

Heterozygous Scn1a deletion did not influence locomotor activity as both WT and Scn1a+/− mice traveled comparable distances during the open- field test (Fig. 2a, unpaired t-test: t19 = 0.054, p =

0.957). Assessing the time course of distance traveled per minute showed that both groups of mice similarly habituated to the open field with a reduction in locomotor activity over time (Fig. 2b, repeated two- way ANOVA: minutes, F9, 171 = 33.54, p b 0.0001; genotype, F1, 19 = 0.003, p = 0.957; minute by

112

Fig. 2. Scn1a+/− mice displayed anxiogenic-like behavior in the open-field test. (a) Total distance traveled during a 10-minute open-field test by wildtype (WT) and Scn1a+/− mice and (b) distance traveled per minute in the test. Representative locomotor traces from a (c) WT and (d) Scn1a+/− mouse. (e) Total time spent in the center region of the open-field arena, defined as the inner 70%, and (f) time spent in center during each minute of the test. Data represent means± standard error mean (SEM) with open circles representing individualmice, n = 10–11 and were analyzed using unpaired t-tests and repeated two-way ANOVA. *p < 0.05.

+/− spent in the center region over time, Scn1a mice minutes, F9, 171 = 3.593, p = 0.0004; genotype , F1, 19 =

spent significantly less time in the center over the 4.645, p = 0.044; minute by genotype interaction, F9,

entire course of the test, suggesting greater anxiety- 171 = 0.275, p = 0.098).

related behavior (Fig. 2f, repeated two-way ANOVA:

113

Fig. 3. Scn1a+/−mice displayed normal behavior in the elevated plus maze and the marble burying test. (a) Time spent on the open arms, (b) the ratio of time spent on the open arms to time spent on the closed arms, and (c) the number of entries made to the open arms of the elevated plus maze. (d) Number of marbles that were buried during the marble burying test of repetitive, compulsive behavior. Bar graphs represent means + SEM and were analyzed using unpaired t-tests. For the box-and-whiskers plot, the horizontal line within the box indicates the median, the ‘+’ represents the mean, the boundaries of the box indicate the 25th to 75th percentile, and the whiskers represent the minimum and maximum values, and data were analyzed with a Mann–Whitney U test. Open circles represent individual mice, n = 10–11.

3.2. Scn1a+/− mice did not exhibit altered anxiety-like measures (unpaired t-test: t19 = 1.877, p = 0.076, Mann– behavior in the elevated plus maze or repetitive, Whitney test: U = 35, p = 0.173 and unpaired t-test: compulsive behavior in the marble burying test

unpaired t-test: t19 = 1.917, p = 0.071, respectively). We Despite Scn1a+/− mice displaying anxiogenic- also measured repetitive, compulsive behavior in the like behavior in the open-field test, this finding was not marble burying test. There was no significant difference reproduced in the elevated plus maze. Total time spent between WT and Scn1a+/− mice on the number of on the open arms (Fig. 3a), the time spent on the open marbles buried (Fig. 3c, unpaired t-test: t19 = 0.712, p = arms as a ratio of the time spent on the closed arms (Fig. 0.485). 3b), and the entries to open arms (Fig. 3c) appeared larger in the Scn1a+/− mice than the WT mice, but this effect did not reach statistical significance for any of the

114

Fig. 4. Scn1a+/− mice displayed impaired preference for social novelty compared with WT mice. (a) The total test time spent engaged in and (b) the frequency of social interactions (SI) during a social interaction test, with n = 10–11, unpaired t-tests. (c) Time spent in a 5-centimeter radius around an empty cage (white boxes) or a cage holding a stranger mouse (light blue boxes) in the three-chambered social preference test, with n = 10–11, WT: Wilcoxon matched-pairs signed rank test, Scn1a+/−: paired t- test. (d) Time spent in a 5-centimeter radius around a cage holding a familiar mouse (light blue boxes) or a novel mouse (dark blue boxes) in the three-chambered social preference test, with n = 9–11, paired t-tests. *p < 0.05. Data represent means + SEM with open circles representing individual mice. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3.3. Scn1a+/− mice displayed similar sociability but chambered social preference test. We first measured impaired preference for social novelty compared with preference for a cage holding a stranger mouse over an WT mice empty cage. Interestingly, we found that both WT and We conducted the social interaction test to Scn1a+/− mice lacked preference for the stranger mouse; examine whether the Scn1a+/− genotype affects social as both groups spent an equivalent time in proximity to behavior. There was no difference between WT and the stranger mouse compared with the empty cage (Fig. Scn1a+/− mice in the total time (Fig. 4a, unpaired t-test: 4c, WT, Wilcoxon matched-pairs signed rank test: Z = t19 = 1.216, p = 0.239) or frequency of social

10, p = 0.700; paired t-test: t10 = 0.17, p = 0.868; interactions with a novel mouse (Fig. 4b, unpaired t- +/− Scn1a , paired t-test: t9 = 1.4, p = 0.195, respectively). test: t19 = 0.308, p = 0.761). We also sought to examine We then investigated social recognition, by measuring whether heterozygous Scn1a deletion affects more the preference for a novel mouse over a familiar mouse subtle measures of social behavior using the three- (Fig. 4d). As expected, WT mice displayed a preference

115

Fig. 5. Scn1a+/− mice displayed better novel object recognition and impaired spatial memory compared with WT mice. (a) Discrimination index (DI) scores are displayed for WT and Scn1a+/− mice from the novel object recognition test, n = 10–11, paired t-tests. DI scores > 25 demonstrate preference for the novel object, while DI scores <−25 demonstrate preference for the familiar object; these thresholds are represented by dotted lines. (b) The average latency to locate the escape box during the training phase of the Barnes maze, with n = 10, repeated measures ANOVA. (c) Latency to and (d) distance to first entry to target zone, with n = 10, Mantel–Cox test. (e) Shows the total distance traveled in the probe test, n = 10, Mann–Whitney U test. (f) Shows the % of time spent in the target zone, Mann–Whitney U test, and (g) the % of correct entries into the target zone in the probe test of the Barnes maze, with n = 10, unpaired t-tests. For the box-and-whiskers plots, the horizontal line within the box indicates the median, the ‘+’ represents the mean, the boundaries of the box indicate the 25th to 75th percentile, and the whiskers represent the minimum and maximum values. Otherwise, data represent means ± SEM with open circles representing individual mice. *p < 0.05. for the novel mouse, spending significantly more time 3.4. Scn1a+/− mice displayed better novel object recognition and subtle impairments in spatial memory in proximity to the novel mouse than the familiar mouse compared with WT mice +/− (paired t-test: t10 = 3.074, p = 0.012). Scn1a mice, We utilized the novel object recognition test to however, showed no preference and spent an equivalent evaluate recognition memory, in which a DI score of time exploring the novel mouse to that of the familiar >25 indicates a preference for the novel object [33]. mouse (paired t-test: t8 = 1.032, p = 0.332). Interestingly, Scn1a+/− mice showed preference for the

novel object with a mean DI score of 26.4 ± 7.4, which

was significantly greater than the mean DI score of WT

116 mice who tended to prefer the familiar object (−17.5 ± (Fig. 5e; Mann–Whitney test: U = 41, p = 0.529). This

+/− 13.9) (Fig. 5a; unpaired t-test: t19 = 2.74, p = 0.013). suggests that Scn1a mice exhibited a subtle spatial

Total time spent interacting with both objects was memory deficit during the probe test. No significant equivalent between groups (WT = 83.5 ± 17.3 s; differences, however, were found between genotypes in

+/− Scn1a = 90.2 ± 26.4 s; unpaired t-test: t19 = 0.218, p = the percentage of time spent in the target zone (Fig. 5f;

0.83). Mann–Whitney test: U = 39, p = 0.436) or correct

entries made into the target zone (Fig. 5g unpaired t- Spatial learning and memory were assessed test: t = 0.77, p =0.452). using the Barnes maze. During the training phase, both 17 groups showed similar learning of the task evidenced 3.5. Scn1a+/− mice exhibited an acoustic startle deficit by reduced latencies to locate the escape box across with intact sensorimotor gating, and an atypical fear response compared with WT mice training days (Fig. 5b; repeated two-way ANOVA: trial Acoustic startle, PPI, and contextual fear day, F3, 54 = 21.12, p b 0.0001; genotype: F1, 18 = 0.742, conditioning paradigms were utilized to evaluate startle p = 0.4; trial day × genotype: F3, 54 = 0.54, p = 0.657). response, auditory sensorimotor gating, and fear While it appeared that Scn1a+/− mice displayed longer memory of WT and Scn1a+/− mice, respectively. During latencies to first enter the target zone, this did not reach the acoustic startle test, startle amplitude was measured statistical significance (Fig. 5c; Mantel–Cox Log-rank across 30 repeated presentations of a 120 decibel 2 +/− test: χ 1 = 3.196, p = 0.074). However, Scn1a mice stimulus. Scn1a+/− mice displayed significantly reduced covered a significantly greater distance than WT mice acoustic startle responses compared with WT mice to this first entry into the target zone, with one Scn1a+/− across all trials (Fig. 6a, repeated two-way ANOVA: mouse failing to enter the target zone (Fig. 5d; Mantel– trials, F29, 551 = 1.537, p = 0.038; genotype, F1, 19 = 8.936, 2 Cox Log-rank test: χ 1 = 5.381, p = 0.02). This could p = 0.007; trials by not be explained by increased locomotor activity, as there was no differences between WT and Scn1a+/− mice in the total distance traveled during the probe test

118 and a trial by genotype interaction, p = 0.002). Baseline mouse models of Dravet syndrome have demonstrated freezing behavior prior to footshock was not different various behavioral abnormalities, but these studies between WT and Scn1a+/− mice (Sidak's post hoc test: p were restricted to adult mice [24,25]. Here, we provide

= 0.992). Immediately following the footshock, WT data demonstrating behavioral abnormalities in mice exhibited the expected increase in freezing Scn1atm1Kea mice during adolescence. Adolescent behavior compared with baseline (Sidak's post hoc test: Scn1a+/− mice exhibited altered behavior across various p < 0.001). However, this fear response was absent in domains, including increased anxiety-like behavior, the Scn1a+/− mice (Sidak's post hoc test: p = 0.996). subtle impairments in spatial memory, blunted acoustic

Accordingly, the WT mice showed significantly higher startle reflexes, and atypical fear responses. In addition, freezing responses than Scn1a+/− mice immediately our study demonstrates that heterozygous Scn1a postshock (Sidak's post hoc test: p = 0.002). During the deletion modulated novelty preference depending on reexposure trial that took place 30 min and 24 h after the nature of the stimuli. Scn1a+/− mice showed an footshock, both WT and Scn1a+/− mice displayed enhanced preference for novel, inanimate stimuli but an significantly elevated freezing behavior compared with impaired preference for novel, social stimuli. baseline (Sidak's post hoc test: p < 0.001 for all Adolescent Scn1a+/− mice did not exhibit a comparisons), which did not differ significantly hyperactive phenotype in the open-field test, a between genotypes (Sidak's post hoc test: p = 0.128 and phenotype reported in adult mice [24,25]. However, p = 0.595, respectively). Scn1a+/− mice displayed increased anxiety-like

4. Discussion behavior as gauged by increased time being spent in the peripheral regions of the open-field test arena [28,39]. Identifying behavioral abnormalities present in Interestingly, the anxiogenic phenotype of the Scn1a+/− adolescent Scn1a+/− mice would better model the early mice was not reproduced in the elevated plus maze. onset of attentional, emotional, and cognitive Consistent with this model-dependent effect on impairments that are observed in patients with Dravet anxiety-like behavior, the Scn1atm1Kzy mouse model of syndrome. Studies in Scn1atm1Wac and Scn1atm1Kzy

119

Dravet syndrome also exhibited increased anxiety in preference. This indicates that Scn1a+/- mice may the open-field test but decreased anxiety in the elevated display social recognition memory deficits. Other plus maze [25]. Collectively, these results imply that mouse models of Dravet syndrome displayed social anxiety-related behavior observed in the open-field test recognition deficits, along with sociability deficits may have distinct qualities to that measured in the [24,25]. In the current study, neither the WT nor elevated plus maze. Alternatively, the increased Scn1a+/− mice showed a preference for a novel mouse exploration of the open arms in the elevated plus maze over a novel inanimate object. The mixed genetic might reflect increased fear-motivated escape behavior, (129S6/SvEvTac × C57BL/ 6J) background of the which has been observed following stress exposure in Scn1a+/− mice may contribute to this unexpected

C57BL/6J mice [40]. Future studies might expand on performance. C57BL/6 mice display clear preferences the characterization of anxiety-related behaviors in for social stimuli [41]. The social preference of

Scn1atm1Kzy mice using other models such as the light– 129S6/SvEvTac mice remains unreported; however, a dark test or zero maze that may help clarify the nature different 129 substrain (129S1/SvJ) lacks social of these behaviors in these mice. preference [41].

Our results show that heterozygous Scn1a Unexpectedly, WT mice demonstrated poor deletion selectively impairs preference for social object discrimination in the novel object recognition novelty without affecting general social behavior. task with a tendency to favor exploration of the familiar

Scn1a+/− mice displayed normal general social object. In another study, C57BL/6 mice displayed behavior, as there were no differences between WT and robust recognition of novel objects [32], while

Scn1a+/− mice in the total time of social interaction or 129S6/SvEvTac mice failed to exhibit novel object frequency of social behaviors. Whereas in the three- recognition [42]. Therefore, as the Scn1a+/− mice used chambered social preference test, WT mice displayed a in this study were on a mixed background clear preference for a novel mouse over a familiar (129S6/SvEvTac × C57BL/6J), the introduction of

+/ mouse, when the Scn1a − mice did not show such a 129S6/SvEvTac genetics into the F1 mixed background

120 strain may have countered the usual performance of gating, as measured in the PPI paradigm, was normal.

C57BL/6J mice on this task. Surprisingly, the Scn1a+/− However, it is possible that if the PPI test was repeated mice displayed a significant preference for the novel in adult Scn1a+/− mice, an impairment in sensorimotor object during the novel object recognition test. While gating may be observed since deficits in PPI have been the Scn1a+/− mice displayed intact spatial learning on identified in adult patients with temporal lobe epilepsy the Barnes maze during acquisition, a subtle spatial and autism spectrum disorder [45,46]. memory impairment was observed during the probe test Here, we show that adolescent Scn1a+/− mice as Scn1a+/− mice covered a significantly greater display a remarkably blunted acoustic startle distance to locate the target zone than WT mice. In adult response. Impaired acoustic startle reflexes may mouse models of Dravet syndrome, more pronounced indicate defects in hearing [47]. As all the prepulse impairments in learning and memory were reported. stimuli promoted a prepulse intensity-dependent tm1Kzy Scn1a and Scn1atm1Wac mice both made more errors attenuation of the startle response to a loud acoustic in locating the target, while Scn1atm1Wac mice also stimulus, it appears that Scn1a+/− mice do not have any displayed an increased latency to find the target during grossly impaired hearing function. This is consistent both the learning phase and probe test of the Barnes with patients with Dravet syndrome having normal maze [24,25]. auditory function [48]. It is possible that the reduced

Repetitive, compulsive behavior and startle response in the Scn1a+/− mice is due to a sensorimotor processing deficits occur in patients with brainstem dysfunction since this brain region autism and Dravet syndrome [11,13,43,44]. However, mediates the startle reflex [49]. Indeed, patients with object-oriented compulsive behavior and sensorimotor Dravet syndrome have reduced white matter volumes gating have never been examined in any mouse model in the brainstem, which could suggest impaired of Dravet syndrome. Adolescent Scn1a+/− mice brainstem function [50]. Furthermore, conditional exhibited normal marble burying behavior, a measure knockout of Scn1a in inhibitory neurons promoted of repetitive, compulsive behavior. Sensorimotor hyperexcitability of neurons in the brainstem [51].

121

Whether such effects of heterozygous Scn1a deletion transferred or enhanced through observation or social similarly affect the brainstem regions mediating interaction [55,56]. startle reflex is unknown and requires further Our findings of impaired social recognition investigation. memory, intact novel object recognition, and increased

In the contextual fear conditioning paradigm, anxiety-like behavior in adolescent Scn1a+/− mice are adolescent Scn1a+/− mice displayed severely reduced congruent with other behavioral studies conducted in freezing behavior immediately following shock adult mouse models of Dravet syndrome, aged 10–28 exposure, which was not observed in other Scn1a+/− weeks and 6– 8 months [24,25]. By examining the adult mouse models [24,52]. One possible explanation for Scn1a+/− phenotype, these studies have neglected the the reduced freezing behavior is that Scn1a+/− mice early onset of the behavioral abnormalities observed in have impaired pain perception. Pain sensitivity has patients with Dravet syndrome. The present study not been assessed in any of the mouse models of demonstrates that behavioral deficits are present in

Dravet syndrome, although NaV1.1 channels adolescent mice, thus more closely modeling the early contribute to sensory perception of mechanical pain onset of developmental delays observed in young

[53,54], and SCN1A gain-of-function mutations have patients with Dravet syndrome. While clinical studies been linked to familial hemiplegic migraine. have reported that behavioral comorbidities associated

However, impaired pain perception would not explain with Dravet syndrome are present in young patients, expression of fear memory 30 min and 24 h these behavioral deficits appear to worsen with age postshock, suggesting successful acquisition of long- [9,10,14,57]. A future study is required to characterize term fear memory. One explanation for the long-term the behavioral phenotypes of Scn1atm1Kea mice in fear expression is that contextual fear may have been adulthood, which would allow us to track the acquired vicariously through ‘social transfer’ when developmental progression of impaired cognitive, mice were returned to group-housed cages between social, emotional, and motor function. If the impact of trials. This is supported by contextual fear being NaV1.1 haploinsufficiency on behavioral impairments

122 is progressive, then early intervention treatments may Acknowledgments be required to attain the best outcomes in improving The authors would like to acknowledge Ivan quality of life for patients with Dravet syndrome. Low for technical assistance. This work was supported

by the Lambert Initiative for Cannabinoid Declaration of competing interest Therapeutics, a philanthropically funded center for Associate Professor Jonathon Arnold is Deputy medicinal cannabis research at the University of Academic Director of the Lambert Initiative for Sydney, and the Australian National Health and Cannabinoid Therapeutics, a philanthropically funded Medical Research Council (NHMRC) [grant number research center at the University of Sydney. He has GNT1161571]. The authors gratefully acknowledge served as an expert witness in various medicolegal Barry and Joy Lambert for their continued support. In cases involving cannabis and in 2018, was a temporary addition, we thank Katelyn Lambert for inspiring our advisor to the World Health Organization (WHO) on work on novel cannabinoid therapies for childhood their review of cannabis and the cannabinoids. His epilepsy. research is funded by the Lambert Initiative and the

Australian National Health and Medical Research References

Council (NHMRC). A/Prof Arnold and Dr. Lyndsey [1] Krueger J, Berg AT Incidence of Dravet syndrome in a US population Pediatr Neurol Briefs 2015;29:92 [2] Wu YW, Sullivan J, McDaniel SS, Meisler MH, Walsh EM, Li SX, et al Incidence of Dravet Anderson hold patents on cannabinoid therapies syndrome in a US population Pediatrics 2015;136:e1310–5 [3] Ishii A, Watkins JC, Chen D, Hirose S, Hammer MF Clinical implications of SCN1A missense and truncation variants in a large Japanese cohort with Dravet syndrome Epilepsia 2017;58:282– (PCT/AU2018/05089 and PCT/AU2019/050554). The 90 [4] Bayat A, Hjalgrim H, Moller RS The incidence of SCN1A-related Dravet syndrome in Denmark is 1:22,000: a population-based study from 2004 to 2009 Epilepsia 2015; 56:e36-9 [5] Brunklaus A, Ellis R, Reavey E, Forbes GH, Zuberi SM Prognostic, clinical and demographic remaining authors have no conflicts of interest. We features in SCN1A mutation-positive Dravet syndrome Brain 2012;135: 2329–36 [6] Shmuely S, Sisodiya SM, Gunning WB, Sander JW, Thijs RD Mortality in Dravet syndrome: a review Epilepsy Behav 2016;64:69–74 confirm that we have read the Journal's position on [7] Cooper MS, McIntosh A, Crompton DE, McMahon JM, Schneider A, Farrell K, et al Mortality in Dravet syndrome Epilepsy Res 2016;128:43–7 [8] Ba ttaglia D, Ricci D, Chieffo D, Guzzetta F Outlining a core neuropsychological phenotype for issues involved in ethical publication and affirm that Dravet syndrome Epilepsy Res 2016;120:91–7 [9] Brunklaus A, Dorris L, Zuberi SM Comorbidities and predictors of health-related quality of life in Dravet syndrome Epilepsia 2011;52:1476–82 this report is consistent with those guidelines. [10] Genton P, Velizarova R, Dravet C Dravet syndrome: the long-term outcome Epilepsia 2011;52(Suppl 2):44–9 [11] Villeneuve N, Laguitton V, Viellard M, Lepine A, Chabrol B, Dravet C, et al Cognitive and adaptive evaluation of 21 consecutive patients with Dravet syndrome Epilepsy Behav 2014;31:143–8 [12] Takayama R, Fujiwara T, Shigematsu H, Imai K, Takahashi Y, Yamakawa K, et al Long-term course of Dravet syndrome: a study from an epilepsy center in Japan Epilepsia 2014;55:528– 38 [13] Acha J, Perez A, Davidson DJ, Carreiras M Cognitive characterization of children with Dravet syndrome: a neurodevelopmental perspective Child Neuropsychol 2015;21: 693–715

123

[14] Olivieri G, Battaglia D, Chieffo D, Rubbino R, Ranalli D, Contaldo I, et al Cognitive-behavioral [36] Brzozowska NI, Smith KL, Zhou C, Waters PM, Cavalcante LM, Abelev SV, et al Genetic profiles in teenagers with Dravet syndrome Brain Dev 2016;38:554–62 deletion of P-glycoprotein alters stress responsivity and increases depression-like behavior, [15] Lagae L, Brambilla I, Mingorance A, Gibson E, Battersby A Quality of life and comorbidities social withdrawal and microglial activation in the hippocampus of female mice Brain Behav associated with Dravet syndrome severity: a multinational cohort survey Dev Med Child Neurol Immun 2017;65:251–61 2018;60:63–72 [37] Angoa-Perez M, Kane MJ, Briggs DI, Francescutti DM, Kuhn DM Marble burying and nestlet [16] Sinoo C, de Lange IM, Westers P, Gunning WB, Jongmans MJ, Brilstra EH Behavior problems shredding as tests of repetitive, compulsive-like behaviors in mice J Vis Exp 2013:50978 and health-related quality of life in Dravet syndrome Epilepsy Behav 2019;90:217–27 [38] Smith KL, Kassem MS, Clarke DJ, Kuligowski MP, Bedoya-Perez MA, Todd SM, et al [17] Chieffo D, Battaglia D, Lucibello S, Gambardella ML, Moriconi F, Ferrantini G, et al Disorders Microglial cell hyper-ramification and neuronal dendritic spine loss in the hippocampus and of early language development in Dravet syndrome Epilepsy Behav 2016;54:30–3 medial prefrontal cortex in a mouse model of PTSD Brain Behav Immun 2019;80:889–99 [18] Turner SJ, Brown A, Arpone M, Anderson V, Morgan AT, Scheffer IE Dysarthria and broader [39] Simon P, Dupuis R, Costentin J Thigmotaxis as an index of anxiety in mice Influence of motor speech deficits in Dravet syndrome Neurology 2017;88:743–9 dopaminergic transmissions Behav Brain Res 1994;61:59–64 [19] Depienne C, Trouillard O, Saint-Martin C, Gourfinkel-An I, Bouteiller D, Carpentier W, et al [40] Mozhui K, Karlsson RM, Kash TL, Ihne J, Norcross M, Patel S, et al Strain differences in stress Spectrum of SCN1A gene mutations associated with Dravet syndrome: analysis of 333 patients responsivity are associated with divergent amygdala gene expression and glutamate-mediated J Med Genet 2009;46:183–91 neuronal excitability J Neurosci 2010;30:5357–67 [20] Marini C, Scheffer IE, Nabbout R, Suls A, De Jonghe P, Zara F, et al The genetics of Dravet [41] Moy SS, Nadler JJ, Young NB, Perez A, Holloway LP, Barbaro RP, et al Mouse behavioral syndrome Epilepsia 2011;52(Suppl 2):24–9 tasks relevant to autism: phenotypes of 10 inbred strains Behav Brain Res 2007;176:4–20 [21] Mistry AM, Thompson CH, Miller AR, Vanoye CG, George Jr AL, Kearney JA Strainand age- [42] Wolf A, Bauer B, Abner EL, Ashkenazy-Frolinger T, Hartz AMA Comprehensive behavioral dependent hippocampal neuron sodium currents correlate with epilepsy severity in Dravet test battery to assess learning and memory in 129S6/Tg2576 mice PLoS One 2016;11:e0147733 syndrome mice Neurobiol Dis 2014;65:1–11 [43] Martin P, Rautenstraubeta B, Abicht A, Fahrbach J, Koster S Severe myoclonic epilepsy in [22] Miller AR, Hawkins NA, McCollom CE, Kearney JA Mapping genetic modifiers of survival in infancy — adult phenotype with bradykinesia, hypomimia, and perseverative behavior: report of a mouse model of Dravet syndrome Genes Brain Behav 2014;13:163–72 five cases Mol Syndromol 2010;1:231–8 [23] Anderson LL, Absalom NL, Abelev SV, Doohan PT, Low I, Martin LJ, et al Coadministered [44] Patterson PH Modeling autistic features in animals Pediatr Res 2011;69:34r–40r cannabidiol and clobazam: preclinical evidence for both pharmacodynamic and pharmacokinetic [45] Kiziltan ME, Alpaslan BG, Ozkara C, Uzan M, Gunduz A Role of mesial temporal lobe interactions Epilepsia 2019;00:1–11 structures in sensory processing in humans: a prepulse modulation study in temporal lobe [24] Han S, Tai C, Westenbroek RE, Yu FH, Cheah CS, Potter GB, et al Autistic-like behaviour in epilepsy Exp Brain Res 2018;236:3297–305 Scn1a+/− mice and rescue by enhanced GABA-mediated neurotransmission Nature [46] Takahashi H, Kamio Y Acoustic startle response and its modulation in schizophrenia and autism 2012;489:385–90 spectrum disorder in Asian subjects Schizophr Res 2018;198:16–20 [25] Ito S, Ogiwara I, Yamada K, Miyamoto H, Hensch TK, Osawa M, et al Mouse with Nav1 1 [47] Parham K, Willott JF Acoustic startle response in young and aging C57BL/6J and CBA/J mice haploinsufficiency, a model for Dravet syndrome, exhibits lowered sociability and learning Behav Neurosci 1988;102:881–6 impairment Neurobiol Dis 2013;49:29–40 [48] Villas N, Meskis MA, Goodliffe S Dravet syndrome: characteristics, comorbidities, and [26] Varlinskaya EI, Spear LP, Spear NE Social behavior and social motivation in adolescent rats: caregiver concerns Epilepsy Behav 2017;74:81–6 role of housing conditions and partner's activity Physiol Behav 1999;67: 475–82 [49] Davis M, Parisi T, Gendelman DS, Tischler M, Kehne JH Habituation and sensitization of startle [27] Yang M, Silverman JL, Crawley JN Automated three-chambered social approach task for mice reflexes elicited electrically from the brainstem Science 1982;218 688–90 Curr Protoc Neurosci 2011;56:8 26 1–8 26 16 [50] Perez A, Garcia-Penton L, Canales-Rodriguez EJ, Lerma-Usabiaga G, Iturria-Medina Y, Roman [28] Seibenhener ML, Wooten MC Use of the open field maze to measure locomotor and anxiety- FJ, et al Brain morphometry of Dravet syndrome Epilepsy Res 2014;108: 1326–34 like behavior in mice J Vis Exp 2015:e52434 [51] Kuo FS, Cleary CM, LoTurco JJ, Chen X, Mulkey DK Disordered breathing in a mouse model [29] Boucher AA, Hunt GE, Micheau J, Huang X, McGregor IS, Karl T, et al The schizophrenia of Dravet syndrome Elife 2019;8 susceptibility gene neuregulin 1 modulates tolerance to the effects of cannabinoids Int J [52] Rubinstein M, Westenbroek RE, Yu FH, Jones CJ, Scheuer T, Catterall WA Genetic background Neuropsychopharmacol 2011;14:631–43 modulates impaired excitability of inhibitory neurons in a mouse model of Dravet syndrome [30] Todd SM, Zhou C, Clarke DJ, Chohan TW, Bahceci D, Arnold JC Interactions between Neurobiol Dis 2015;73:106–17 cannabidiol and Delta(9)-THC following acute and repeated dosing: rebound hyperactivity, [53] Osteen JD, Herzig V, Gilchrist J, Emrick JJ, Zhang C, Wang X, et al Selective spider toxins sensorimotor gating and epigenetic and neuroadaptive changes in the mesolimbic pathway Eur reveal a role for the Nav1 1 channel in mechanical pain Nature 2016;534:494–9 Neuropsychopharmacol 2017;27:132–45 [54] Salvatierra J, Castro J, Erickson A, Li Q, Braz J, Gilchrist J, et al NaV1 1 inhibition can reduce [31] Silverman JL, Yang M, Lord C, Crawley JN Behavioural phenotyping assays for mouse models visceral hypersensitivity JCI Insight 2018;3 of autism Nat Rev Neurosci 2010;11:490–502 [55] Jeon D, Kim S, Chetana M, Jo D, Ruley HE, Lin SY, et al Observational fear learning involves [32] Sik A, van Nieuwehuyzen P, Prickaerts J, Blokland A Performance of different mouse strains affective pain system and Cav1 2 Ca2+ channels in ACC Nat Neurosci 2010;13:482–8 in an object recognition task Behav Brain Res 2003;147:49–54 [56] Knapska E, Mikosz M, Werka T, Maren S Social modulation of learning in rats Learn Mem [33] Vogel-Ciernia A, Wood MA Examining object location and object recognition mem ory in mice 2010;17:35–42 Curr Protoc Neurosci 2014;69:8 31 1–17 [57] Berkvens JJ, Veugen I, Veendrick-Meekes MJ, Snoeijen-Schouwenaars FM, Schelhaas HJ, [34] Rosenfeld CS, Ferguson SA Barnes maze testing strategies with small and large rodent models Willemsen MH, et al Autism and behavior in adult patients with Dravet syndrome (DS) J Vis Exp 2014:e51194 Epilepsy Behav 2015;47:11–6 [35] Walf AA, Frye CA The use of the elevated plus maze as an assay of anxiety-related behavior in rodents Nat Protoc 2007;2:322–8

124

Chapter 5

General discussion

125

The research conducted in this thesis advances our understanding of the endocannabinoid and GPR55-LPI systems and their influence on the phenotype of the Scn1a+/- mouse model of

Dravet syndrome. Results within the thesis identifies and validates CB1 and Gpr55 receptors as potential therapeutic targets for the treatment of Dravet syndrome. The research presented in

Chapter 2 demonstrated that mice with the F1 seizure-susceptible background strain had reduced hippocampal CB1 expression and 2-AG concentrations, and that potentiation of endocannabinoid signalling with a MAGL inhibitor and CB1 receptor PAM was anticonvulsant against hyperthermia-induced seizures. The work conducted in Chapter 3 confirmed that hippocampal

Gpr55 mRNA expression was increased and that LPI concentrations were reduced in the F1 seizure-susceptible strain compared to the 129 seizure-resistant strain of mice. We also reported in Chapter 3 that heterozygous deletion of Gpr55 in Scn1a+/- mice improved seizure susceptibility and rescued premature mortality. Additionally, the work in Chapter 4 expanded the current knowledge of the behavioural impacts of heterozygous Scn1a deletion during adolescence and showed that Scn1a+/- mice had multiple deficits in measures of social, attentional and cognitive function. This provides a strong platform for future investigations into new therapies aimed at correcting the developmental delays in cognitive and emotional functioning observed in Dravet syndrome.

The endocannabinoid system plays an integral role in regulating neuronal excitability, and there is a plethora of preclinical research that has investigated endocannabinoid system modulation in various animal models of seizures and epilepsy (Gaston & Szaflarski, 2018; Katona, 2015;

Rosenberg, Patra & Whalley, 2017; Soltesz et al., 2015). Moreover, a preeminent mechanism for the anticonvulsant effects of CBD is antagonism of the GPR55 receptor, a receptor once thought to be the 3rd cannabinoid receptor but now known to be activated by the lipid mediator LPI (Oka

126 et al., 2009). Yet, specific targeting of endocannabinoid and GPR55-LPI system components as a therapeutic strategy for Dravet syndrome or other paediatric epilepsies had not yet been investigated. As discussed in Chapter 1, epilepsy disorders may all produce seizures, but their aetiology and underlying pathophysiology can vary. This knowledge suggests that anticonvulsant efficacy of endocannabinoid or GPR55-LPI system modulation in one preclinical seizure or epilepsy model may not necessarily transfer to another. Therefore, the work in this thesis aimed to investigate the endocannabinoid and GPR55-LPI systems in a Dravet syndrome mouse model by utilising the ‘genetic modifier’ approach. As the Dravet syndrome phenotype in Scn1a+/- mice is critically-dependent on background strain genetics, the ‘genetic modifier’ approach compares the transcriptome of seizure-susceptible and seizure-resistant mice to detect divergences in gene expression that may influence the epilepsy phenotype, and ultimately identify druggable targets.

Thus, by comparing the expression of genes involved in the endocannabinoid and GPR55-LPI systems between the two background strains we identified a number of gene candidates and selected Cnr1 and Gpr55 as the genes to be investigated in Chapters 2 and 3 of this thesis, respectively.

The work in Chapter 2 identified a potential double deficit in the F1 seizure-susceptible mice with reduced expression of CB1 receptor mRNA and protein, along with less 2-AG levels in the hippocampus. As the endocannabinoid system is as a regulator of excess neuronal activity (Zou

& Kumar, 2018), we hypothesised that our observed strain-dependent variability may be contributing to the hyperexcitability observed in the Scn1a+/- mice on the seizure-susceptible background strain. We therefore investigated and demonstrated that pharmacological potentiation of endocannabinoid signalling with a CB1 receptor PAM and MAGL inhibitor were anticonvulsant against hyperthermia-induced seizures. One way to strengthen the role of the endocannabinoid

127 system as a modifier of Scn1a disease would be to assess the impact of Cnr1 or Dagl deletion on the phenotype of the Scn1a+/-mice. However, this method of validation is ethically undesirable as we predict this would exacerbate the seizure phenotype. Although the work conducted in Chapter

2 supports the anticonvulsant potential of potentiating the endocannabinoid system, putatively through activation of CB1 receptors, future research will need confirm that CB1 receptor activation was indeed the mechanism of action. This may prove difficult as antagonism of CB1 receptors has previously been shown to increase seizure activity (Rosenberg, Patra & Whalley, 2017). However, if a CB1 receptor antagonist was shown to have no intrinsic activity in the Scn1a+/- mouse model, then it could be co-administered with a CB1 receptor PAM or MAGL inhibitor to determine its ability to disrupt these agents’ anticonvulsant effects and thus infer if their effects are indeed mediated by the CB1 receptor.

The work in Chapter 2 was limited in that it only evaluated the efficacy of a CB1 receptor

PAM and MAGL inhibitor against the hyperthermia induced seizures. It is important to assess the robustness of the anticonvulsant efficacy of these compounds against other epilepsy-relevant endpoints available in the model, such as the spontaneous seizure and survival phenotypes, as these phenotypes might be subserved by different neurochemical mechanisms (Hawkins et al., 2017).

Drugs that enhance endocannabinoid transmission being effective against hyperthermia-induced seizures is not surprising given prior research showing that CB1 receptor blockade exacerbated acute febrile seizures in rats (Feng et al., 2016). However, whether endocannabinoid stimulating agents also have anticonvulsant efficacy against spontaneous seizures may not be so straightforward.

In Dravet syndrome, children usually first present with febrile seizures that progress to afebrile spontaneous seizures (Dravet, 2011). Therefore, to replicate this in our model, we induce

128 a febrile seizure before examining spontaneous seizures (Hawkins et al., 2017). Interestingly, febrile seizures upregulate CB1 receptor expression, mostly at inhibitory synapses, resulting in disinhibition of neuronal excitability (Chen et al., 2003). Moreover, febrile seizures caused CB1 receptor-mediated adulthood seizure-susceptibility, as demonstrated by SR141716, a CB1 receptor antagonist, preventing the development of adult seizure susceptibility (Feng & Chen, 2016). Thus, endocannabinoid system stimulating agents might actually have proconvulsant effects against spontaneous seizures preceded by a febrile seizure. Future research could then investigate the effects of CB1 receptor PAMs or MAGL inhibitors against spontaneous seizures that have or have not been preceded by an induced febrile seizure. Moreover, it would be useful to explore how the endocannabinoid system responds to febrile seizures in the short and long-term, by examining alterations of endocannabinoid system components at both excitatory and inhibitory synapses.

It would also be important for future studies to address whether the compounds maintain their efficacy with repeated administration. Repeated stimulation of CB1 receptors has been demonstrated to cause downregulation and/or desensitisation of CB1 receptors, which has led to the development of tolerance of efficacy. This was seen with the repeated administration of CB1 receptor agonist WIN55,212 losing efficacy against seizures (Blair et al., 2009) and the irreversible

MAGL inhibitor JZL184 losing its analgesic effect (Schlosburg et al., 2010). Further support for the need to investigate chronic administration is demonstrated by the chronic use of high doses of

JZL184 proving proconvulsant in a mouse model of pilocarpine induced temporal lobe epilepsy

(Ma et al., 2014).

Future work extending on Chapter 2’s results could also investigate how endocannabinoid concentrations change in response to seizures. An improvement on measuring endocannabinoid concentrations from tissue homogenates would be to assess endocannabinoids in vivo with either

129 microextraction or microdialysis techniques that both utilise probes to extract analytes from the surrounding interstitial space. With solid-phase microextraction, the probe is coated with an

“extraction medium” that directly absorbs surrounding organic compounds, before being removed then desorbed to collect analytes. Whereas, with microdialysis, the probe is comprised of a semipermeable surface that has a constant flow of perfusate moving through it that can absorb molecules of desired size and solubility from the surrounding space before dialysate being discharged for collection (Chefer et al., 2009). As endocannabinoids are lipophilic, they exhibit poor solubility in standard perfusate solutions. However, the addition of the oligosaccharide β- cyclodextrin has improved the solubility of lipids in the perfusate, improving sampling efficiency of endocannabinoids by tenfold (Walker et al., 1999). β-cyclodextrin is a cylindrically shaped molecule with a hydrophobic core that binds endocannabinoids, while the hydrophilic outer surface of the complex promoting solubility. For both microdialysis and microextraction, the collected analytes can then be analysed with any analytical technique. These techniques have been successfully utilised to analyse endocannabinoid concentrations and could be applied to future research to further illuminate the role of these mediators in childhood epilepsy models (Aslam et al., 2019; Serrano et al., 2018; Sustkova-Fiserova et al., 2017; Zestos & Kennedy, 2017).

These in vivo sampling techniques are advantageous because they allow repeated sampling from live animals and gaining a better temporal understanding of endocannabinoid concentrations would better direct therapeutic developments. Improved temporal resolution of endocannabinoid levels is important as the fluctuating concentrations of endocannabinoid concentrations throughout the day could have differing functional implications for regulation of excitability. For example, 2-

AG concentrations diminish during the dark phase of the light/dark cycle in multiple brain region in rats (Valenti et al., 2004), while Dravet syndrome mice are more likely to exhibit spontaneous

130 seizures and experience SUDEP during their dark phase (Teran et al., 2019). This suggests that

MAGL inhibition may be a better served through administration in the dark phase. Moreover, these techniques have the ethical advantage of requiring less animals, which would also provide cleaner data by reducing inter-animal variability while avoiding any post-mortem accumulation of endocannabinoids (Sugiura, Yoshinaga & Waku, 2001). Furthermore, these techniques could be modified to extend to the quantification of additional lipid mediators, such as LPIs.

A notable finding from the work conducted in Chapter 2 was the increase of Daglb mRNA expression selectively in the phenotypic F1.Scn1a+/- mice. DAGL-β is expressed almost exclusively in microglia (Viader et al., 2016), therefore, if transcriptional expression translates to protein, our findings suggest that F1.Scn1a+/- mice may have increased microglial cell 2-AG activity. Research indicates that microglial cell 2-AG levels increase in response to cell damage to elicit neuroprotection via CB2 receptors, which exhibit increased expression on activated microglia (Hohmann et al., 2019; Kreutz et al., 2009; Kreutz et al., 2007; Stella, 2009; Viader et al., 2016; Walter et al., 2003). Future research could use multi-probe immunofluorescence staining of Dravet syndrome mouse brain tissue with antibody probes for Dagl-β and CB2 receptors to quantify expression of these proteins, in conjunction with a probe for Iba-1 to detect microglia number and morphology. If protein expression correlates with transcription, then the impact of

Daglb on the Dravet syndrome phenotype could be assessed genetically by crossing Daglb knockout mice with Dravet syndrome mice (Tanimura et al., 2010; Viader et al., 2016; Wilkerson et al., 2016) or pharmacologically with specific Dagl-β inhibitors (Luk et al., 2018; Wilkerson et al., 2016).

Dagl-β is a compelling target as the activation of microglia has been implicated in the process of epileptogenesis (Devinsky et al., 2013; Eyo, Murugan & Wu, 2017). For example,

131 selective activation of microglia resulted in mice developing severe early-onset spontaneous seizures and premature death during post-natal weeks 3 and 4 (Zhao et al., 2018). This is noteworthy because Dravet syndrome mice develop spontaneous seizures and premature death during this exact time-frame (Miller et al., 2014; Mistry et al., 2014). Moreover, although preliminary, this finding does suggest that F1.Scn1a+/- mice may be exhibiting neuroinflammation in the hippocampus. Increased inflammatory response to seizure insults facilitates epileptogenesis, as demonstrated by the pharmacological or genetic inhibition of interleukin 1β receptor preventing the development of seizure susceptibility (Feng et al., 2016). Additionally, many anticonvulsants exert therapeutic efficacy via anti-inflammatory actions in various seizure models (Aronica et al.,

2017), while many cannabinoids, including CBD, exhibit anti-inflammatory activity (Nagarkatti et al., 2009; van Niekerk, Mabin & Engelbrecht, 2019). Therefore, whether neuroinflammation occurs in the Dravet syndrome mouse model remains an open question and its investigation might provide useful traction on understanding potential anticonvulsant mechanisms of cannabinoids.

Another useful investigation that would greatly advance our understanding of the pathophysiology of the Dravet syndrome phenotype would be comparing the transcriptome of different neuronal cell types using single-cell sequencing (Cadwell et al., 2017; Ho et al., 2018).

This technique can investigate the transcriptome from a single cell and can therefore be utilised to compare differences between the WT and Scn1a+/- mice from both seizure-susceptible and seizure- resistant background strains of different cell types, including different neuronal populations, astrocytes and microglia. For instance, Gpr55 receptors are also expressed on microglia with

Gpr55 activation reducing the number of activated microglia and demonstrating neuroprotective effects in a model of excitotoxicity (Kallendrusch et al., 2013; Pietr et al., 2009). Therefore, this

132 technique could be useful for investigating the expression of multiple targets of interest in a cell- specific manner to better inform on their possible contribution to the Dravet syndrome phenotype.

Chapter 3 of the thesis investigated the role of the GPR55-LPI system in the Scn1a+/- mouse model. It showed that the F1 seizure-susceptible mice had increased hippocampal Gpr55 transcript expression and overall reduced LPI concentrations. Interestingly, heterozygous deletion of Scn1a only reduced hippocampal LPI concentrations in mice on the seizure-resistant 129 background strain and not in mice on the seizure-susceptible F1 strain. This suggests that the GPR55-LPI system in the seizure-susceptible strain may not be able to adapt to the impact of heterozygous deletion of Scn1a. We also demonstrated that heterozygous deletion of Gpr55 was anticonvulsant against hyperthermia-induced seizures and ameliorated the survival deficit normally observed in the phenotypic Scn1a+/- mice, albeit with a technical limitation requiring these experiments to be repeated with Gpr55 knockout mice that are congenic to the C57BL/6 background strain.

A background strain can be made congenic with continuous backcrossing, that can be accelerated with the assistance of background strain sequencing (Wong, 2002). However, a potential limitation with this method is the conservation of the original 129 genetics on loci that influence phenotype. This has previously been observed by our collaborator Dr Jennifer Kearney

(Northwestern University, Chicago) who backcrossed a background strain to 99.9% congenic

C57BL/6 with the remaining 0.1% of 129 genetic material segregating to survival-influencing loci

(unpublished data). A better approach would be to generate a new line of Gpr55 knockout mice directly on the C57BL/6 background using CRISPR-Cas9 technology (Wefers et al., 2017).

Nevertheless, there is research demonstrating that the selective Gpr55 antagonist, CID16020046, reduced the frequency of action potential outputs and increased the frequency of spontaneous inhibitory postsynaptic currents in granule cells from Scn1a+/- hippocampal slices (Kaplan et al.,

133

2017). Therefore, despite the limitations of the work conducted in Chapter 3, when taken in conjunction of prior work showing GPR55 antagonism reduced neuronal hyperexcitability, the data presented here putatively supports Gpr55 as a new anticonvulsant drug target for Dravet syndrome.

A novel interesting finding from Chapter 3 that requires further investigation is that heterozygous Scn1a deletion downregulated the mRNA expression of Mboat7. This enzyme is involved in phospholipid remodelling and is known to degrade 1-stearoyl LPI (Anderson et al.,

2013; Lee et al., 2012), while also being involved in the remodelling of other lipid mediators such as other species of LPI, PIs, and phosphatidylinoethanolamines (PEs) (Anderson et al., 2013; Lee et al., 2012). Gene variants of MBOAT7 in humans has been associated with intellectual disability, autism spectrum disorder and epilepsy (Jacher et al., 2019; Johansen et al., 2016; Yalnizoglu et al., 2019), while Mboat7 gene deletion in rodents causes brain atrophy and premature death

(Anderson et al., 2013; Lee et al., 2012). It is therefore possible that diminished MBOAT7 expression in Scn1a+/- mice might be linked to brain atrophy, especially in white matter, and cognitive impairments observed in Dravet syndrome patients (Brunklaus et al., 2012; Dravet,

2011; Genton, Velizarova & Dravet, 2011; Han et al., 2012; Ito et al., 2013; Perez et al., 2014;

Striano et al., 2007).

Downregulation of Mboat7 mRNA expression in both seizure-susceptible and seizure- resistant mouse strains suggests it is not critically important to the seizure phenotype. However, as the impact of heterozygous deletion of Scn1a on cognitive and emotional function has not been examined in the 129 strain, Mboat7 may still be implicated in the behavioural comorbidities.

Future work would first need to confirm alterations at the functional protein level and if confirmed, investigate any potential implications. Unfortunately, there are currently no pharmacological tools

134 available to directly potentiate the activity of MBOAT7 and observe its effect on behavioural phenotypes. However, a comprehensive lipidomic analysis of the effects of Mboat7 downregulation might reveal other druggable targets to revert the metabolic impact of Mboat7 deficiency.

The work in Chapter 2 and 3 assessed the endocannabinoid and Gpr55-LPI systems at a single age and in a single brain region. We chose P24 as this is when susceptible mice exhibit spontaneous seizures, however the Dravet syndrome phenotype changes through development

(Miller et al., 2014; Mistry et al., 2014). As detailed in Chapter 1, susceptibility to febrile seizures is prominent in the earlier years of life before afebrile spontaneous seizures develop. Seizures can resolve or lessen with age, while the behavioural deficits that begin in childhood worsen with age.

The endocannabinoid system also changes throughout development, as demonstrated by age- related changes in receptor expression and endocannabinoid concentrations (Kwok et al., 2017;

Lu & Mackie, 2016; Meyer, Lee & Gee, 2018). Therefore, a temporal understanding of dynamic changes in the endocannabinoid system in the Dravet syndrome mouse model may better inform how the endocannabinoid system contributes to the different staged of pathophysiology of this condition and might better inform therapeutic strategies that aim to target endocannabinoid signalling.

The work in this thesis solely focused on the hippocampus because it is a region that is integral to the Dravet syndrome phenotype. NaV1.1 haploinsufficiency significantly impairs neuronal excitability the hippocampus (Liautard et al., 2013; Rubinstein et al., 2015; Tsai et al.,

2015), and deletion of Scn1a in mice on a C57Bl6J background selectively in the hippocampus was sufficient for the development of thermal seizures and spatial memory deficits during adulthood (Stein et al., 2019). However, the study conducted by Stein et al. (2019) could not

135 reproduce other behavioural deficits such as hyperactivity or social deficits, signifying that other brain regions must be important to the phenotype. Future research could therefore expand investigation to additional brain regions including the cortex, which is also implicated in the aetiology of the epilepsy phenotype as demonstrated by impaired cortical inhibitory transmission and through forebrain-specific Scn1a deletion studies (Cheah et al., 2012; Han et al., 2012; Kalume et al., 2007; Kaplan et al., 2017; Ogiwara et al., 2013; Rubinstein et al., 2015; Tai et al., 2014; Tsai et al., 2015; Yu et al., 2006). Supporting the need to expand investigation of the endocannabinoid and GPR55-LPI system to additional brain regions is research demonstrating that the CB1 and

GPR55 receptor function can vary between brain regions. For example, Gpr55 activation increased excitatory transmission in the hippocampus but promoted inhibitory transmission in the striatum

(Hurst et al., 2017; Kramar et al., 2017; Musella et al., 2017; Sylantyev et al., 2013). Moreover, as we saw improved survival in the Scn1a mice with heterozygous deletion of Gpr55, it would be informative to assess expression and function of Gpr55 in brain regions that may be implicated in

SUDEP, such as the breathing centres in the brainstem (Kuo et al., 2019).

Gaining a better understanding of CB1 and Gpr55 receptor expression and function throughout development and in response to seizures would provide vital information on how the broader endocannabinoid system could be contributing to and responding to NaV1.1 impairment- induced hyperexcitability. Transcript expression of candidate receptors can be measured via quantitative PCR or in situ hybridisation, and protein expression can be measured with western blotting or immunohistochemical techniques. These methods require the use of many mice as they provide only single time-points of information per sample. Moreover, we couldn’t successfully measure Gpr55 protein expression with western blotting due to commercially-available antibodies

136 proving nonspecific, thus there is a need to develop and validate improved Gpr55-sepcific antibodies.

Another technique that could be utilised to measure receptor expression in vivo that also allows sampling across multiple time-points is a system that utilises fibre photometry optimised for assessment of transcription. This technique combines fibre photometry, which utilises the implantation of optical fibres in select brain regions with the viral delivery of fluorescence or bioluminescence reporters that are targeted to genes of interest and stimulated by the transcription process. This technique was demonstrated to reliably measure the expression of genes that are markers of circadian rhythms across multiple days (Mei et al., 2018). Therefore, the transcription of our candidate genes, for example Cnr1 or Gpr55, can be directly measured in a regionally- specific manner, which can be utilised to measure expression throughout development, as well as in response to seizures and epileptogenesis. Although this technique is limited in only measuring gross regional expression, it would inform on the temporal response to seizures and epileptogenesis. Then further investigation with double- or triple-fluorescent imaging of candidate receptors (e.g. CB1 and Gpr55) and markers for specific neuronal populations, such as glutamate decarboxylate (GAD65/67) for GABAergic neurons and vesicular glutamate transporter 1

(VGLUT1) for glutamatergic neurons, could further provide information on cell-specific expression of the candidate molecular targets.

More detailed functional mechanisms of our candidate molecular targets could be explored in future studies utilising electrophysiological recordings from ex vivo brain slices of Dravet syndrome mice. As activation of CB1 receptors inhibits neuronal cell activity, CB1 receptor function can be directly assessed using whole-cell patch-clamp recordings from either excitatory or inhibitory post-synaptic neurons (Farrell & Soltesz, 2019; Soltesz et al., 2015). As

137 endocannabinoid signalling is stimulated by depolarisation, the evaluation of excitatory or inhibitory post-synaptic potentials in response to depolarising stimuli in Scn1a+/- mice blain slices will inform on efficacy of depolarisation-induced suppression of neuronal activity, and therefore cannabinoid receptor-mediated suppression of excitatory activity. Confirmation of CB1 receptor- mediated activity can be determined by repeating the experiment in the presence of a CB1 receptor antagonist such as AM251. The function of Gpr55 receptors, on the other hand, can be determined from whole-cell patch-clamp recordings from excitatory or inhibitory postsynaptic neurons conducted in the presence of applied Gpr55 endogenous ligand (LPI) or synthetic ligands (e.g. O-

1602) and confirmed with Gpr55 antagonists (e.g. CID16020046) (Sylantyev et al., 2013).

In addition to seizures, Dravet syndrome patients exhibit developmental delays in attentional, emotional and cognitive functions. These behavioural comorbidities are stronger predictors of a lower quality-of-life than seizure frequency (Lagae et al., 2018; Sinoo et al., 2019), highlighting a dire need for treatments that specifically target behavioural deficits. The adult behavioural phenotype of Scn1a+/- mice recapitulates many of the associated behavioural comorbidities reported in humans (Han et al., 2012; Ito et al., 2013). The work conducted in

Chapter 4 aimed to characterise for the first time the behavioural phenotype of a Scn1a+/- mouse model during adolescence. This characterisation more closely models the behavioural impairments observed in paediatric Dravet syndrome patients and establishes a platform to test early- intervention therapeutics. We assessed the impact of heterozygous deletion of Scn1a in mice on an array of behavioural measures and report that adolescent Scn1a+/- mice exhibited normal locomotor activity, marble burying behaviour, sociability and sensorimotor gating, but increased anxiety-related thigmotactic behaviour, atypical fear expression, blunted acoustic startle responses, and impaired social recognition and spatial memory.

138

Cannabis-based therapies may prove to have only modest effects on seizure control but may have profound effects on the non-seizure symptoms associated with Dravet syndrome, ultimately improving the overall quality of life of these patients. CBD’s impact on non-seizure symptoms in Dravet syndrome patients has not been directly assessed in clinical trials. However, there are multiple observational studies suggesting that CBD treatment improved quality of life measures, such as sleep, alertness and mood in paediatric epilepsy patients (McCoy et al., 2018;

Porter & Jacobson, 2013; Stockings et al., 2018). Although these are just observational studies, they are supported by preclinical research showing that CBD reversed social deficits in adult

Scn1a+/- mice in the three-chambered social preference test and in the standard social interaction test (Kaplan et al., 2017). Future work extending on this thesis could utilise the behavioural assessment platform established here to assess the behavioural impact of early-intervention treatment with CBD and other cannabinoids.

This behavioural platform could also be utilised to examine the impact of Cnr1, Gpr55 or

Mboat7, and any other potential genetic modifiers, on the behavioural phenotypes of Scn1a+/- mice.

The endocannabinoid system has been well-established to be involved in many behavioural modalities, including learning and memory (Kruk-Slomka et al., 2017; Mechoulam & Parker,

2013; Riedel & Davies, 2005), suggesting that CB1 receptor-mediated endocannabinoid signal potentiation could be therapeutically beneficial. Whereas, pharmacological and genetic studies in rodents have shown Gpr55 influences various behavioural functions including locomotor activity, sociability, as well as fear, recognition and spatial memory (Alavi et al., 2016; Hurst et al., 2017;

Kramar et al., 2017; Marichal-Cancino et al., 2018; Meadows et al., 2016; Rahimi, Hajizadeh

Moghaddam & Roohbakhsh, 2015). Collectively these reports suggest that Gpr55 activation may have detrimental effects on behaviour and that Gpr55 inhibition may mitigate behavioural

139 impairments, as observed by CBD ameliorating social deficits in Scn1a+/- mice (Kaplan, Isom &

Petrou, 2016). Pharmacological and genetic knockdown studies could be implemented to assess the influence of novel target genes on the Scn1a+/- behavioural phenotype to validate new drug targets that extend beyond seizure control.

An interesting observation from this work was that the behavioural deficits reported in adult mouse models of Dravet syndrome, which include hyperactivity, impaired sociability and significant fear and spatial memory deficits were not replicated in our study (Han et al., 2012; Ito et al., 2013). This may be due to the age of the mice at testing, as the behavioural impairments might progressively worsen with age, which has been reported in Dravet syndrome patients

(Berkvens et al., 2015; Brunklaus, Dorris & Zuberi, 2011; Genton, Velizarova & Dravet, 2011;

Olivieri et al., 2016). The adult behavioural phenotype of these mice could be evaluated to determine if impairments are progressive which would reaffirm the need to explore early intervention therapeutics. However, another possibility for the inability to replicate prior behavioural characterisations of Scn1a+/- mice could be the genetic background strain of the mice that were investigated. The Scn1a+/- mice tested here were on a mixed F1 (129S6/SvEvTac x

C57BL/6J) background, whereas studies in other lines were conducted in mice on a congenic

C57BL/6 background (Han et al., 2012; Ito et al., 2013). Severity of the Dravet syndrome seizure phenotype is catastrophically sensitive to C57BL/6 background-strain genetics, with increasing the proportion of C57BL/6 genetics increasing the severity of the seizure phenotype (Hawkins et al., 2016; Miller et al., 2014; Mistry et al., 2014; Rubinstein et al., 2015). Therefore, it is possible that the partial C57BL/6 background genetics used here might explain the less pronounced behavioural impairments that were observed in Chapter 4.

140

Future work could extend upon the characterisation of the non-seizure behavioural phenotypes of Scn1a+/- mice reported in Chapter 4. For example, characterising the sleep phenotype in our Scn1a+/- mice could be explored as patients report high prevalence of sleep disturbances (~75%) that negatively impact on quality of life (Dhamija et al., 2014; Licheni et al.,

2018). We hypothesise that we may observe impaired sleep behaviour in our mouse model as impaired sleep quality and homeostasis has already been reported in other Scn1a+/- mice (Kalume et al., 2015). This is another area where cannabis-based therapies may prove beneficial as there is evidence to suggest that CBD may improve sleep quality (Babson, Sottile & Morabito, 2017;

Fleury-Teixeira et al., 2019; Pane & Sacca, 2019; Porter & Jacobson, 2013), and because the therapeutic use of cannabinoids for sleep is a growing area of research (Babson, Sottile &

Morabito, 2017).

A limitation of the work in Chapter 4 was that we did not assess behaviour in female mice.

During our investigation, the female Scn1a+/- mice displayed increased mortality compared to males that was not statistically significant. Despite most work on Scn1a+/- mouse models reporting no sex-differences (Anderson et al., 2019a; Hawkins et al., 2017), a recent study reported a sex- dependent effect on mortality with female mice displaying a significantly higher mortality rate than male mice (Niibori et al., 2019). Moreover, in support of sex differences was a recent study reporting social impairment in male but not female Scn1a+/- mice (Williams et al., 2019). This suggests that there may be subtle differences in rearing conditions that may influence sex- dependent effects. However, the translational significance of this finding is difficult to determine as there have been no sex-dependent differences reported in the prevalence of Dravet syndrome patients (Bayat, Hjalgrim & Moller, 2015; Brunklaus et al., 2012; Ishii et al., 2017; Krueger &

Berg, 2015; Wu et al., 2015), nor for the rate of mortality observed in Dravet syndrome patients

141

(Cooper et al., 2016). There has been, however, one study reporting female Dravet syndrome patients display more severe symptoms (Ohmori et al., 2003). Future work could investigate whether our model does indeed display sex-dependent effects on survival and behavioural impairments. The identification of sex-dependent effects may be a consideration that needs to be addressed in the development of therapeutics that are effective for both sexes.

The ‘genetic modifier’ approach has identified additional molecular targets associated with the endocannabinoid system to those explored in this thesis that may hold therapeutic potential.

For example, we have identified that mRNA expression of transient receptor potential cation channel subfamily V member 1 (TRPV1) receptors, was significantly increased in F1 seizure- susceptible mice (data not shown). TRPV1 is a non-selective cation channel that is activated by endogenous and exogenous vanilloids, extracellular acidification, heat and anandamide (Bevan,

Quallo & Andersson, 2014; Zygmunt et al., 1999). TRPV1 activation enhances neuronal excitability by increasing intracellular concentrations of cations, which increase membrane conductance and consequently neurotransmitter release (Fawley, Hofmann & Andresen, 2014;

Naziroglu, 2015). Interestingly, Trpv1 knock-out mice and Trpv1 antagonists, such as capsazepine and AMG-9810, have proven anticonvulsant in pentylenetetrazol-induced seizure models (Jia et al., 2015; Kong et al., 2014; Naderi et al., 2015; Shirazi et al., 2014; Suemaru et al., 2018).

Therefore, the influence of Trpv1 on the Dravet syndrome phenotype is currently being investigated. Moreover, the identification of numerous molecular targets that can impact on the

Scn1a+/- phenotype supports and further validates the use of the ‘genetic modifier’ approach in novel drug development.

142

Conclusion

The research presented in this thesis expands our understanding of the Dravet syndrome mouse model by providing insight into how the broader endocannabinoid system could be influencing phenotype severity as well as providing a better understanding of the behavioural impairments that present during adolescence. This research highlights CB1 and Gpr55 receptors as potential modifiers of the Dravet syndrome phenotype. We demonstrated less CB1 and more

Gpr55 receptor mRNA expression in mice on the F1 seizure-susceptible background strain compared to mice on the 129 seizure-resistant strain. Moreover, we validated the anticonvulsant potential of these targets by pharmacologically potentiating CB1 receptor activity and with heterozygous deletion of Gpr55. The work in this thesis has also identified additional molecular targets, Dagl-β and Mboat7, that may influence the Scn1a+/- phenotype and warrant further investigation. Lastly, we characterised the behavioural impact of heterozygous Scn1a deletion in adolescent mice to establish a platform that better models the behavioural impairments observed in paediatric Dravet syndrome patients. We identified numerous deviations in measures of social, cognitive and attentional behaviours but did not reproduce all impairments previously reported for adult Scn1a+/- mice. This suggests that developmental delays may be progressive and there may be benefit to early therapeutic intervention. By using the Scn1a+/- mouse model to deepen our understanding of the Dravet syndrome condition, the work in this thesis will assist in the development of novel therapeutics for a more comprehensive treatment of Dravet syndrome.

143

Chapter 6

References

144

Acha, J, Perez, A, Davidson, DJ & Carreiras, M 2015, 'Cognitive characterization of children with Dravet syndrome: A neurodevelopmental perspective', Child Neuropsychol, vol. 21, no. 5, pp. 693-715.

Adamowicz, P, Gieron, J, Gil, D, Lechowicz, W, Skulska, A & Tokarczyk, B 2017, 'The effects of synthetic cannabinoid UR-144 on the human body-A review of 39 cases', Forensic Sci Int, vol. 273, pp. e18-e21.

Alavi, MS, Hosseinzadeh, H, Shamsizadeh, A & Roohbakhsh, A 2016, 'The effect of O-1602, an atypical cannabinoid, on morphine-induced conditioned place preference and physical dependence', Pharmacol Rep, vol. 68, no. 3, pp. 592-7.

Alhouayek, M, Masquelier, J & Muccioli, GG 2018, 'Lysophosphatidylinositols, from Cell Membrane Constituents to GPR55 Ligands', Trends Pharmacol Sci.

Anderson, KE, Kielkowska, A, Durrant, TN, Juvin, V, Clark, J, Stephens, LR & Hawkins, PT 2013, 'Lysophosphatidylinositol-acyltransferase-1 (LPIAT1) is required to maintain physiological levels of PtdIns and PtdInsP(2) in the mouse', PLoS One, vol. 8, no. 3, pp. e58425.

Anderson, LL, Absalom, NL, Abelev, SV, Low, IK, Doohan, PT, Martin, LJ, Chebib, M, Mcgregor, IS & Arnold, JC 2019a, 'Coadministered cannabidiol and clobazam: Preclinical evidence for both pharmacodynamic and pharmacokinetic interactions', Epilepsia.

Anderson, LL, Hawkins, NA, Thompson, CH, Kearney, JA & George, AL, Jr. 2017, 'Unexpected Efficacy of a Novel Sodium Channel Modulator in Dravet Syndrome', Sci Rep, vol. 7, no. 1, pp. 1682.

Anderson, LL, Low, IK, Banister, SD, Mcgregor, IS & Arnold, JC 2019b, 'Pharmacokinetics of Phytocannabinoid Acids and Anticonvulsant Effect of Cannabidiolic Acid in a Mouse Model of Dravet Syndrome', J Nat Prod.

Anwar, H, Khan, QU, Nadeem, N, Pervaiz, I, Ali, M & Cheema, FF 2020, 'Epileptic seizures', Discoveries (Craiova), vol. 8, no. 2, pp. e110.

Aoki, J, Taira, A, Takanezawa, Y, Kishi, Y, Hama, K, Kishimoto, T, Mizuno, K, Saku, K, Taguchi, R & Arai, H 2002, 'Serum lysophosphatidic acid is produced through diverse phospholipase pathways', J Biol Chem, vol. 277, no. 50, pp. 48737-44.

Aronica, E, Bauer, S, Bozzi, Y, Caleo, M, Dingledine, R, Gorter, JA, Henshall, DC, Kaufer, D, Koh, S, Loscher, W, Louboutin, JP, Mishto, M, Norwood, BA, Palma, E, Poulter, MO, Terrone, G, Vezzani, A & Kaminski, RM 2017, 'Neuroinflammatory targets and treatments for epilepsy validated in experimental models', Epilepsia, vol. 58 Suppl 3, pp. 27-38.

Aslam, M, Feleder, C, Newsom, RJ, Campeau, S & Musteata, FM 2019, 'In vivo monitoring of rat brain endocannabinoids using solid-phase microextraction', Bioanalysis, vol. 11, no. 16, pp. 1523-1534.

145

Babson, KA, Sottile, J & Morabito, D 2017, 'Cannabis, Cannabinoids, and Sleep: a Review of the Literature', Curr Psychiatry Rep, vol. 19, no. 4, pp. 23.

Bahceci, D, Anderson, LL, Occelli Hanbury Brown, CV, Zhou, C & Arnold, JC 2019, 'Adolescent behavioral abnormalities in a Scn1a+/− mouse model of Dravet syndrome', Epilepsy & Behavior, pp. 106842.

Battaglia, D, Ricci, D, Chieffo, D & Guzzetta, F 2016, 'Outlining a core neuropsychological phenotype for Dravet syndrome', Epilepsy Res, vol. 120, pp. 91-7.

Bayat, A, Hjalgrim, H & Moller, RS 2015, 'The incidence of SCN1A-related Dravet syndrome in Denmark is 1:22,000: a population-based study from 2004 to 2009', Epilepsia, vol. 56, no. 4, pp. e36-9.

Bender, AC, Natola, H, Ndong, C, Holmes, GL, Scott, RC & Lenck-Santini, PP 2013, 'Focal Scn1a knockdown induces cognitive impairment without seizures', Neurobiol Dis, vol. 54, pp. 297-307.

Berecki, G, Bryson, A, Terhag, J, Maljevic, S, Gazina, EV, Hill, SL & Petrou, S 2019, 'SCN1A gain of function in early infantile encephalopathy', Ann Neurol, vol. 85, no. 4, pp. 514-525.

Berger, WT, Ralph, BP, Kaczocha, M, Sun, J, Balius, TE, Rizzo, RC, Haj-Dahmane, S, Ojima, I & Deutsch, DG 2012, 'Targeting fatty acid binding protein (FABP) anandamide transporters - a novel strategy for development of anti-inflammatory and anti-nociceptive drugs', PLoS One, vol. 7, no. 12, pp. e50968.

Berkvens, JJ, Veugen, I, Veendrick-Meekes, MJ, Snoeijen-Schouwenaars, FM, Schelhaas, HJ, Willemsen, MH, Tan, IY & Aldenkamp, AP 2015, 'Autism and behavior in adult patients with Dravet syndrome (DS)', Epilepsy Behav, vol. 47, pp. 11-6.

Berman, P, Futoran, K, Lewitus, GM, Mukha, D, Benami, M, Shlomi, T & Meiri, D 2018, 'A new ESI-LC/MS approach for comprehensive metabolic profiling of phytocannabinoids in Cannabis', Sci Rep, vol. 8, no. 1, pp. 14280.

Bevan, S, Quallo, T & Andersson, DA 2014, 'TRPV1', Handb Exp Pharmacol, vol. 222, pp. 207- 45.

Bialer, M & Perucca, E 2020, 'Does cannabidiol have antiseizure activity independent of its interactions with clobazam? An appraisal of the evidence from randomized controlled trials', Epilepsia, vol. 61, no. 6, pp. 1082-1089.

Bjorklund, E, Blomqvist, A, Hedlin, J, Persson, E & Fowler, CJ 2014, 'Involvement of fatty acid amide hydrolase and fatty acid binding protein 5 in the uptake of anandamide by cell lines with different levels of fatty acid amide hydrolase expression: a pharmacological study', PLoS One, vol. 9, no. 7, pp. e103479.

146

Bjursell, M, Ryberg, E, Wu, T, Greasley, PJ, Bohlooly, YM & Hjorth, S 2016, 'Deletion of Gpr55 Results in Subtle Effects on Energy Metabolism, Motor Activity and Thermal Pain Sensation', PLoS One, vol. 11, no. 12, pp. e0167965.

Blair, RE, Deshpande, LS, Sombati, S, Elphick, MR, Martin, BR & Delorenzo, RJ 2009, 'Prolonged exposure to WIN55,212-2 causes downregulation of the CB1 receptor and the development of tolerance to its anticonvulsant effects in the hippocampal neuronal culture model of acquired epilepsy', Neuropharmacology, vol. 57, no. 3, pp. 208-18.

Blankman, JL, Simon, GM & Cravatt, BF 2007, 'A comprehensive profile of brain enzymes that hydrolyze the endocannabinoid 2-arachidonoylglycerol', Chem Biol, vol. 14, no. 12, pp. 1347-56.

Boel, M & Casaer, P 1996, 'Add-on therapy of fenfluramine in intractable self-induced epilepsy', Neuropediatrics, vol. 27, no. 4, pp. 171-3.

Bojnik, E, Turunc, E, Armagan, G, Kanit, L, Benyhe, S, Yalcin, A & Borsodi, A 2012, 'Changes in the cannabinoid (CB1) receptor expression level and G-protein activation in kainic acid induced seizures', Epilepsy Res, vol. 99, no. 1-2, pp. 64-8.

Bonini, SA, Premoli, M, Tambaro, S, Kumar, A, Maccarinelli, G, Memo, M & Mastinu, A 2018, 'Cannabis sativa: A comprehensive ethnopharmacological review of a medicinal plant with a long history', J Ethnopharmacol, vol. 227, pp. 300-315.

Brunklaus, A, Dorris, L & Zuberi, SM 2011, 'Comorbidities and predictors of health-related quality of life in Dravet syndrome', Epilepsia, vol. 52, no. 8, pp. 1476-82.

Brunklaus, A, Ellis, R, Reavey, E, Forbes, GH & Zuberi, SM 2012, 'Prognostic, clinical and demographic features in SCN1A mutation-positive Dravet syndrome', Brain, vol. 135, no. Pt 8, pp. 2329-36.

Cadwell, CR, Scala, F, Li, S, Livrizzi, G, Shen, S, Sandberg, R, Jiang, X & Tolias, AS 2017, 'Multimodal profiling of single-cell morphology, electrophysiology, and gene expression using Patch-seq', Nat Protoc, vol. 12, no. 12, pp. 2531-2553.

Cao, D, Ohtani, H, Ogiwara, I, Ohtani, S, Takahashi, Y, Yamakawa, K & Inoue, Y 2012, 'Efficacy of stiripentol in hyperthermia-induced seizures in a mouse model of Dravet syndrome', Epilepsia, vol. 53, no. 7, pp. 1140-5.

Carliner, H, Brown, QL, Sarvet, AL & Hasin, DS 2017, 'Cannabis use, attitudes, and legal status in the U.S.: A review', Prev Med, vol. 104, pp. 13-23.

Carlisle, SJ, Marciano-Cabral, F, Staab, A, Ludwick, C & Cabral, GA 2002, 'Differential expression of the CB2 cannabinoid receptor by rodent macrophages and macrophage-like cells in relation to cell activation', International Immunopharmacology, vol. 2, no. 1, pp. 69-82.

147

Catterall, WA 2000, 'From ionic currents to molecular mechanisms: the structure and function of voltage-gated sodium channels', Neuron, vol. 26, no. 1, pp. 13-25.

Catterall, WA, Kalume, F & Oakley, JC 2010, 'NaV1.1 channels and epilepsy', J Physiol, vol. 588, no. Pt 11, pp. 1849-59.

Cheah, CS, Westenbroek, RE, Roden, WH, Kalume, F, Oakley, JC, Jansen, LA & Catterall, WA 2013, 'Correlations in timing of sodium channel expression, epilepsy, and sudden death in Dravet syndrome', Channels (Austin), vol. 7, no. 6, pp. 468-72.

Cheah, CS, Yu, FH, Westenbroek, RE, Kalume, FK, Oakley, JC, Potter, GB, Rubenstein, JL & Catterall, WA 2012, 'Specific deletion of NaV1.1 sodium channels in inhibitory interneurons causes seizures and premature death in a mouse model of Dravet syndrome', Proc Natl Acad Sci U S A, vol. 109, no. 36, pp. 14646-51.

Chefer, VI, Thompson, AC, Zapata, A & Shippenberg, TS 2009, 'Overview of brain microdialysis', Current protocols in neuroscience, vol. Chapter 7, pp. Unit7.1-Unit7.1.

Chen, D-J, Gao, M, Gao, F-F, Su, Q-X & Wu, J 2017, 'Brain cannabinoid receptor 2: expression, function and modulation', Acta pharmacologica Sinica, vol. 38, no. 3, pp. 312-316.

Chen, K, Neu, A, Howard, AL, Foldy, C, Echegoyen, J, Hilgenberg, L, Smith, M, Mackie, K & Soltesz, I 2007, 'Prevention of plasticity of endocannabinoid signaling inhibits persistent limbic hyperexcitability caused by developmental seizures', J Neurosci, vol. 27, no. 1, pp. 46-58.

Chen, K, Ratzliff, A, Hilgenberg, L, Gulyás, A, Freund, TF, Smith, M, Dinh, TP, Piomelli, D, Mackie, K & Soltesz, I 2003, 'Long-Term Plasticity of Endocannabinoid Signaling Induced by Developmental Febrile Seizures', Neuron, vol. 39, no. 4, pp. 599-611.

Chevaleyre, V & Castillo, PE 2003, 'Heterosynaptic LTD of Hippocampal GABAergic Synapses: A Novel Role of Endocannabinoids in Regulating Excitability', Neuron, vol. 38, no. 3, pp. 461-472.

Chicca, A, Marazzi, J, Nicolussi, S & Gertsch, J 2012, 'Evidence for bidirectional endocannabinoid transport across cell membranes', The Journal of biological chemistry, vol. 287, no. 41, pp. 34660-34682.

Chieffo, D, Battaglia, D, Lucibello, S, Gambardella, ML, Moriconi, F, Ferrantini, G, Leo, G, Dravet, C, Mercuri, E & Guzzetta, F 2016, 'Disorders of early language development in Dravet syndrome', Epilepsy Behav, vol. 54, pp. 30-3.

Cisar, JS, Weber, OD, Clapper, JR, Blankman, JL, Henry, CL, Simon, GM, Alexander, JP, Jones, TK, Ezekowitz, RaB, O'neill, GP & Grice, CA 2018, 'Identification of ABX-1431, a Selective Inhibitor of Monoacylglycerol Lipase and Clinical Candidate for Treatment of Neurological Disorders', J Med Chem, vol. 61, no. 20, pp. 9062-9084.

148

Clanton, RM, Wu, G, Akabani, G & Aramayo, R 2017, 'Control of seizures by ketogenic diet- induced modulation of metabolic pathways', Amino Acids, vol. 49, no. 1, pp. 1-20.

Clarke, DJ, Stuart, J, Mcgregor, IS & Arnold, JC 2017, 'Endocannabinoid dysregulation in cognitive and stress-related brain regions in the Nrg1 mouse model of schizophrenia', Prog Neuropsychopharmacol Biol Psychiatry, vol. 72, pp. 9-15.

Cooper, MS, Mcintosh, A, Crompton, DE, Mcmahon, JM, Schneider, A, Farrell, K, Ganesan, V, Gill, D, Kivity, S, Lerman-Sagie, T, Mclellan, A, Pelekanos, J, Ramesh, V, Sadleir, L, Wirrell, E & Scheffer, IE 2016, 'Mortality in Dravet syndrome', Epilepsy Res, vol. 128, pp. 43-47.

Cravatt, BF, Giang, DK, Mayfield, SP, Boger, DL, Lerner, RA & Gilula, NB 1996, 'Molecular characterization of an enzyme that degrades neuromodulatory fatty-acid amides', Nature, vol. 384, no. 6604, pp. 83-7.

De Caro, C, Leo, A, Citraro, R, De Sarro, C, Russo, R, Calignano, A & Russo, E 2017, 'The potential role of cannabinoids in epilepsy treatment', Expert Rev Neurother, vol. 17, no. 11, pp. 1069-1079.

De Jonghe, P 2011, 'Molecular genetics of Dravet syndrome', Dev Med Child Neurol, vol. 53 Suppl 2, pp. 7-10.

De Lange, IM, Gunning, B, Sonsma, ACM, Van Gemert, L, Van Kempen, M, Verbeek, NE, Nicolai, J, Knoers, N, Koeleman, BPC & Brilstra, EH 2018, 'Influence of contraindicated medication use on cognitive outcome in Dravet syndrome and age at first afebrile seizure as a clinical predictor in SCN1A-related seizure phenotypes', Epilepsia, vol. 59, no. 6, pp. 1154-1165.

De Lera Ruiz, M & Kraus, RL 2015, 'Voltage-Gated Sodium Channels: Structure, Function, Pharmacology, and Clinical Indications', J Med Chem, vol. 58, no. 18, pp. 7093-118.

De Petrocellis, L, Ligresti, A, Moriello, AS, Allara, M, Bisogno, T, Petrosino, S, Stott, CG & Di Marzo, V 2011, 'Effects of cannabinoids and cannabinoid-enriched Cannabis extracts on TRP channels and endocannabinoid metabolic enzymes', Br J Pharmacol, vol. 163, no. 7, pp. 1479-94.

De Salas-Quiroga, A, Diaz-Alonso, J, Garcia-Rincon, D, Remmers, F, Vega, D, Gomez-Canas, M, Lutz, B, Guzman, M & Galve-Roperh, I 2015, 'Prenatal exposure to cannabinoids evokes long-lasting functional alterations by targeting CB1 receptors on developing cortical neurons', Proc Natl Acad Sci U S A, vol. 112, no. 44, pp. 13693-8.

Deliu, E, Sperow, M, Console-Bram, L, Carter, RL, Tilley, DG, Kalamarides, DJ, Kirby, LG, Brailoiu, GC, Brailoiu, E, Benamar, K & Abood, ME 2015, 'The Lysophosphatidylinositol Receptor GPR55 Modulates Pain Perception in the Periaqueductal Gray', Mol Pharmacol, vol. 88, no. 2, pp. 265-72.

149

Depienne, C, Trouillard, O, Saint-Martin, C, Gourfinkel-An, I, Bouteiller, D, Carpentier, W, Keren, B, Abert, B, Gautier, A, Baulac, S, Arzimanoglou, A, Cazeneuve, C, Nabbout, R & Leguern, E 2009, 'Spectrum of SCN1A gene mutations associated with Dravet syndrome: analysis of 333 patients', J Med Genet, vol. 46, no. 3, pp. 183-91.

Devane, WA, Hanus, L, Breuer, A, Pertwee, RG, Stevenson, LA, Griffin, G, Gibson, D, Mandelbaum, A, Etinger, A & Mechoulam, R 1992, 'Isolation and structure of a brain constituent that binds to the cannabinoid receptor', Science, vol. 258, no. 5090, pp. 1946- 9.

Devinsky, O, Cross, JH, Laux, L, Marsh, E, Miller, I, Nabbout, R, Scheffer, IE, Thiele, EA & Wright, S 2017, 'Trial of Cannabidiol for Drug-Resistant Seizures in the Dravet Syndrome', N Engl J Med, vol. 376, no. 21, pp. 2011-2020.

Devinsky, O, Nabbout, R, Miller, I, Laux, L, Zolnowska, M, Wright, S & Roberts, C 2019, 'Long- term cannabidiol treatment in patients with Dravet syndrome: An open-label extension trial', Epilepsia, vol. 60, no. 2, pp. 294-302.

Devinsky, O, Patel, AD, Cross, JH, Villanueva, V, Wirrell, EC, Privitera, M, Greenwood, SM, Roberts, C, Checketts, D, Vanlandingham, KE & Zuberi, SM 2018a, 'Effect of Cannabidiol on Drop Seizures in the Lennox-Gastaut Syndrome', N Engl J Med, vol. 378, no. 20, pp. 1888-1897.

Devinsky, O, Patel, AD, Thiele, EA, Wong, MH, Appleton, R, Harden, CL, Greenwood, S, Morrison, G & Sommerville, K 2018b, 'Randomized, dose-ranging safety trial of cannabidiol in Dravet syndrome', Neurology, vol. 90, no. 14, pp. e1204-e1211.

Devinsky, O, Thiele, EA, Wright, S, Checketts, D, Morrison, G, Dunayevich, E & Knappertz, V 2020, 'Cannabidiol efficacy independent of clobazam: meta-analysis of four randomized- controlled trials', Acta Neurol Scand.

Devinsky, O, Vezzani, A, Najjar, S, De Lanerolle, NC & Rogawski, MA 2013, 'Glia and epilepsy: excitability and inflammation', Trends Neurosci, vol. 36, no. 3, pp. 174-84.

Dey, A, Kang, X, Qiu, J, Du, Y & Jiang, J 2016, 'Anti-Inflammatory Small Molecules To Treat Seizures and Epilepsy: From Bench to Bedside', Trends Pharmacol Sci, vol. 37, no. 6, pp. 463-484.

Dhamija, R, Erickson, MK, St Louis, EK, Wirrell, E & Kotagal, S 2014, 'Sleep abnormalities in children with Dravet syndrome', Pediatr Neurol, vol. 50, no. 5, pp. 474-8.

Di Marzo, V, Fontana, A, Cadas, H, Schinelli, S, Cimino, G, Schwartz, JC & Piomelli, D 1994, 'Formation and inactivation of endogenous cannabinoid anandamide in central neurons', Nature, vol. 372, no. 6507, pp. 686-91.

Diez-Alarcia, R, Ibarra-Lecue, I, Lopez-Cardona, AP, Meana, J, Gutierrez-Adan, A, Callado, LF, Agirregoitia, E & Uriguen, L 2016, 'Biased Agonism of Three Different Cannabinoid Receptor Agonists in Mouse Brain Cortex', Front Pharmacol, vol. 7, pp. 415.

150

Doohan, P 2019, ‘Identification of Cnr1 as a Putative Genetic Modifier in a Mouse Model of Dravet Syndrome’, (Unpublished honor’s thesis). The University of Sydney.

Dos Santos, RG, Hallak, JE, Leite, JP, Zuardi, AW & Crippa, JA 2014, 'Phytocannabinoids and epilepsy', J Clin Pharm Ther.

Dravet, C 2011, 'The core Dravet syndrome phenotype', Epilepsia, vol. 52 Suppl 2, pp. 3-9.

Dravet, C, Bureau, M, Oguni, H, Fukuyama, Y & Cokar, O 2005, 'Severe myoclonic epilepsy in infancy: Dravet syndrome', Adv Neurol, vol. 95, pp. 71-102.

Dravet, C & Oguni, H 2013, 'Dravet syndrome (severe myoclonic epilepsy in infancy)', Handb Clin Neurol, vol. 111, pp. 627-33.

Dutton, SB, Makinson, CD, Papale, LA, Shankar, A, Balakrishnan, B, Nakazawa, K & Escayg, A 2013, 'Preferential inactivation of Scn1a in parvalbumin interneurons increases seizure susceptibility', Neurobiol Dis, vol. 49, pp. 211-20.

Elsohly, MA, Radwan, MM, Gul, W, Chandra, S & Galal, A 2017, 'Phytochemistry of Cannabis sativa L', Prog Chem Org Nat Prod, vol. 103, pp. 1-36.

Escayg, A, Macdonald, BT, Meisler, MH, Baulac, S, Huberfeld, G, An-Gourfinkel, I, Brice, A, Leguern, E, Moulard, B, Chaigne, D, Buresi, C & Malafosse, A 2000, 'Mutations of SCN1A, encoding a neuronal sodium channel, in two families with GEFS+2', Nat Genet, vol. 24, no. 4, pp. 343-5.

Eyo, UB, Murugan, M & Wu, LJ 2017, 'Microglia-Neuron Communication in Epilepsy', Glia, vol. 65, no. 1, pp. 5-18.

Fagerberg, L, Hallstrom, BM, Oksvold, P, Kampf, C, Djureinovic, D, Odeberg, J, Habuka, M, Tahmasebpoor, S, Danielsson, A, Edlund, K, Asplund, A, Sjostedt, E, Lundberg, E, Szigyarto, CA, Skogs, M, Takanen, JO, Berling, H, Tegel, H, Mulder, J, Nilsson, P, Schwenk, JM, Lindskog, C, Danielsson, F, Mardinoglu, A, Sivertsson, A, Von Feilitzen, K, Forsberg, M, Zwahlen, M, Olsson, I, Navani, S, Huss, M, Nielsen, J, Ponten, F & Uhlen, M 2014, 'Analysis of the human tissue-specific expression by genome-wide integration of transcriptomics and antibody-based proteomics', Mol Cell Proteomics, vol. 13, no. 2, pp. 397-406.

Farrell, JS & Soltesz, I 2019, 'Plants come to mind: phytocannabinoids, endocannabinoids and the control of seizures', Addiction, vol. 114, no. 8, pp. 1343-1345.

Fawley, JA, Hofmann, ME & Andresen, MC 2014, 'Cannabinoid 1 and transient receptor potential vanilloid 1 receptors discretely modulate evoked glutamate separately from spontaneous glutamate transmission', J Neurosci, vol. 34, no. 24, pp. 8324-32.

Feng, B & Chen, Z 2016, 'Generation of Febrile Seizures and Subsequent Epileptogenesis', Neurosci Bull, vol. 32, no. 5, pp. 481-92.

151

Feng, B, Tang, Y, Chen, B, Xu, C, Wang, Y, Dai, Y, Wu, D, Zhu, J, Wang, S, Zhou, Y, Shi, L, Hu, W, Zhang, X & Chen, Z 2016, 'Transient increase of interleukin-1beta after prolonged febrile seizures promotes adult epileptogenesis through long-lasting upregulating endocannabinoid signaling', Sci Rep, vol. 6, pp. 21931.

Fezza, F, Marrone, MC, Avvisati, R, Di Tommaso, M, Lanuti, M, Rapino, C, Mercuri, NB, Maccarrone, M & Marinelli, S 2014, 'Distinct modulation of the endocannabinoid system upon kainic acid-induced in vivo seizures and in vitro epileptiform bursting', Mol Cell Neurosci, vol. 62, pp. 1-9.

Fiest, KM, Sauro, KM, Wiebe, S, Patten, SB, Kwon, CS, Dykeman, J, Pringsheim, T, Lorenzetti, DL & Jette, N 2017, 'Prevalence and incidence of epilepsy: A systematic review and meta- analysis of international studies', Neurology, vol. 88, no. 3, pp. 296-303.

Fisher, JL 2009, 'The anti-convulsant stiripentol acts directly on the GABA(A) receptor as a positive allosteric modulator', Neuropharmacology, vol. 56, no. 1, pp. 190-7.

Fisher, RS, Acevedo, C, Arzimanoglou, A, Bogacz, A, Cross, JH, Elger, CE, Engel, J, Jr., Forsgren, L, French, JA, Glynn, M, Hesdorffer, DC, Lee, BI, Mathern, GW, Moshe, SL, Perucca, E, Scheffer, IE, Tomson, T, Watanabe, M & Wiebe, S 2014, 'ILAE official report: a practical clinical definition of epilepsy', Epilepsia, vol. 55, no. 4, pp. 475-82.

Fisher, RS, Cross, JH, D'souza, C, French, JA, Haut, SR, Higurashi, N, Hirsch, E, Jansen, FE, Lagae, L, Moshe, SL, Peltola, J, Roulet Perez, E, Scheffer, IE, Schulze-Bonhage, A, Somerville, E, Sperling, M, Yacubian, EM & Zuberi, SM 2017a, 'Instruction manual for the ILAE 2017 operational classification of seizure types', Epilepsia, vol. 58, no. 4, pp. 531-542.

Fisher, RS, Cross, JH, French, JA, Higurashi, N, Hirsch, E, Jansen, FE, Lagae, L, Moshe, SL, Peltola, J, Roulet Perez, E, Scheffer, IE & Zuberi, SM 2017b, 'Operational classification of seizure types by the International League Against Epilepsy: Position Paper of the ILAE Commission for Classification and Terminology', Epilepsia, vol. 58, no. 4, pp. 522-530.

Fisher, RS, Van Emde Boas, W, Blume, W, Elger, C, Genton, P, Lee, P & Engel, J, Jr. 2005, 'Epileptic seizures and epilepsy: definitions proposed by the International League Against Epilepsy (ILAE) and the International Bureau for Epilepsy (IBE)', Epilepsia, vol. 46, no. 4, pp. 470-2.

Fleury-Teixeira, P, Caixeta, FV, Ramires Da Silva, LC, Brasil-Neto, JP & Malcher-Lopes, R 2019, 'Effects of CBD-Enriched Cannabis sativa Extract on Autism Spectrum Disorder Symptoms: An Observational Study of 18 Participants Undergoing Compassionate Use', Front Neurol, vol. 10, pp. 1145.

Foster, SR, Hauser, AS, Vedel, L, Strachan, RT, Huang, XP, Gavin, AC, Shah, SD, Nayak, AP, Haugaard-Kedstrom, LM, Penn, RB, Roth, BL, Brauner-Osborne, H & Gloriam, DE 2019, 'Discovery of Human Signaling Systems: Pairing Peptides to G Protein-Coupled Receptors', Cell, vol. 179, no. 4, pp. 895-908.e21.

152

Fowler, CJ 2013, 'Transport of endocannabinoids across the plasma membrane and within the cell', Febs j, vol. 280, no. 9, pp. 1895-904.

Friedman, D & Sirven, JI 2017, 'Historical perspective on the medical use of cannabis for epilepsy: Ancient times to the 1980s', Epilepsy Behav, vol. 70, no. Pt B, pp. 298-301.

Funada, M & Takebayashi-Ohsawa, M 2018, 'Synthetic cannabinoid AM2201 induces seizures: Involvement of cannabinoid CB1 receptors and glutamatergic transmission', Toxicol Appl Pharmacol, vol. 338, pp. 1-8.

Gaily, E, Anttonen, AK, Valanne, L, Liukkonen, E, Traskelin, AL, Polvi, A, Lommi, M, Muona, M, Eriksson, K & Lehesjoki, AE 2013, 'Dravet syndrome: new potential genetic modifiers, imaging abnormalities, and ictal findings', Epilepsia, vol. 54, no. 9, pp. 1577-85.

Gao, Y, Vasilyev, DV, Goncalves, MB, Howell, FV, Hobbs, C, Reisenberg, M, Shen, R, Zhang, MY, Strassle, BW, Lu, P, Mark, L, Piesla, MJ, Deng, K, Kouranova, EV, Ring, RH, Whiteside, GT, Bates, B, Walsh, FS, Williams, G, Pangalos, MN, Samad, TA & Doherty, P 2010, 'Loss of retrograde endocannabinoid signaling and reduced adult neurogenesis in diacylglycerol lipase knock-out mice', J Neurosci, vol. 30, no. 6, pp. 2017-24.

Gaston, TE & Szaflarski, JP 2018, 'Cannabis for the Treatment of Epilepsy: an Update', Curr Neurol Neurosci Rep, vol. 18, no. 11, pp. 73.

Geffrey, AL, Pollack, SF, Bruno, PL & Thiele, EA 2015, 'Drug-drug interaction between clobazam and cannabidiol in children with refractory epilepsy', Epilepsia, vol. 56, no. 8, pp. 1246- 51.

Genton, P 2000, 'When antiepileptic drugs aggravate epilepsy', Brain Dev, vol. 22, no. 2, pp. 75- 80.

Genton, P, Velizarova, R & Dravet, C 2011, 'Dravet syndrome: the long-term outcome', Epilepsia, vol. 52 Suppl 2, pp. 44-9.

George, AL, Jr. 2005, 'Inherited disorders of voltage-gated sodium channels', J Clin Invest, vol. 115, no. 8, pp. 1990-9.

Gerdeman, GL, Ronesi, J & Lovinger, DM 2002, 'Postsynaptic endocannabinoid release is critical to long-term depression in the striatum', Nat Neurosci, vol. 5, no. 5, pp. 446-51.

Ghodke-Puranik, Y, Thorn, CF, Lamba, JK, Leeder, JS, Song, W, Birnbaum, AK, Altman, RB & Klein, TE 2013, 'Valproic acid pathway: pharmacokinetics and pharmacodynamics', Pharmacogenet Genomics, vol. 23, no. 4, pp. 236-41.

Gijon, MA, Riekhof, WR, Zarini, S, Murphy, RC & Voelker, DR 2008, 'Lysophospholipid acyltransferases and arachidonate recycling in human neutrophils', J Biol Chem, vol. 283, no. 44, pp. 30235-45.

153

Giraud, C, Treluyer, JM, Rey, E, Chiron, C, Vincent, J, Pons, G & Tran, A 2006, 'In vitro and in vivo inhibitory effect of stiripentol on clobazam metabolism', Drug Metab Dispos, vol. 34, no. 4, pp. 608-11.

Gray, RA, Stott, CG, Jones, NA, Di Marzo, V & Whalley, BJ 2019, 'Anticonvulsive Properties of Cannabidiol in a Model of Generalized Seizure Are Transient Receptor Potential Vanilloid 1 Dependent', Cannabis and Cannabinoid Research, vol. 5, no. 2, pp. 145-149.

Gray, RA & Whalley, BJ 2020, 'The proposed mechanisms of action of CBD in epilepsy', Epileptic Disorders, vol. 22, no. S1, pp. S10-S15.

Grosenbaugh, DK & Mott, DD 2013, 'Stiripentol in refractory status epilepticus', Epilepsia, vol. 54 Suppl 6, pp. 103-5.

Guerrini, R, Cellini, E, Mei, D, Metitieri, T, Petrelli, C, Pucatti, D, Marini, C & Zamponi, N 2010, 'Variable epilepsy phenotypes associated with a familial intragenic deletion of the SCN1A gene', Epilepsia, vol. 51, no. 12, pp. 2474-7.

Haj-Dahmane, S, Shen, RY, Elmes, MW, Studholme, K, Kanjiya, MP, Bogdan, D, Thanos, PK, Miyauchi, JT, Tsirka, SE, Deutsch, DG & Kaczocha, M 2018, 'Fatty-acid-binding protein 5 controls retrograde endocannabinoid signaling at central glutamate synapses', Proc Natl Acad Sci U S A, vol. 115, no. 13, pp. 3482-3487.

Han, S, Tai, C, Westenbroek, RE, Yu, FH, Cheah, CS, Potter, GB, Rubenstein, JL, Scheuer, T, De La Iglesia, HO & Catterall, WA 2012, 'Autistic-like behaviour in Scn1a+/- mice and rescue by enhanced GABA-mediated neurotransmission', Nature, vol. 489, no. 7416, pp. 385-90.

Hawkins, NA, Anderson, LL, Gertler, TS, Laux, L, George, AL, Jr. & Kearney, JA 2017, 'Screening of conventional anticonvulsants in a genetic mouse model of epilepsy', Ann Clin Transl Neurol, vol. 4, no. 5, pp. 326-339.

Hawkins, NA, Calhoun, JD, Huffman, AM & Kearney, JA 2019, 'Gene expression profiling in a mouse model of Dravet syndrome', Exp Neurol, vol. 311, pp. 247-256.

Hawkins, NA & Kearney, JA 2016, 'Hlf is a genetic modifier of epilepsy caused by voltage-gated sodium channel mutations', Epilepsy Res, vol. 119, pp. 20-3.

Hawkins, NA, Zachwieja, NJ, Miller, AR, Anderson, LL & Kearney, JA 2016, 'Fine Mapping of a Dravet Syndrome Modifier Locus on Mouse Chromosome 5 and Candidate Gene Analysis by RNA-Seq', PLoS Genet, vol. 12, no. 10, pp. e1006398.

Henstridge, CM, Balenga, NA, Ford, LA, Ross, RA, Waldhoer, M & Irving, AJ 2009, 'The GPR55 ligand L-alpha-lysophosphatidylinositol promotes RhoA-dependent Ca2+ signaling and NFAT activation', Faseb j, vol. 23, no. 1, pp. 183-93.

Henstridge, CM, Balenga, NA, Kargl, J, Andradas, C, Brown, AJ, Irving, A, Sanchez, C & Waldhoer, M 2011, 'Minireview: recent developments in the physiology and pathology of

154

the lysophosphatidylinositol-sensitive receptor GPR55', Mol Endocrinol, vol. 25, no. 11, pp. 1835-48.

Heynen-Genel, S, Dahl, R, Shi, S, Milan, L, Hariharan, S, Bravo, Y, Sergienko, E, Hedrick, M, Dad, S, Stonich, D, Su, Y, Vicchiarelli, M, Mangravita-Novo, A, Smith, LH, Chung, TDY, Sharir, H, Barak, LS & Abood, ME 2010, 'Screening for Selective Ligands for GPR55 - Agonists', Probe Reports from the NIH Molecular Libraries Program, National Center for Biotechnology Information (US), Bethesda (MD).

Higurashi, N, Uchida, T, Lossin, C, Misumi, Y, Okada, Y, Akamatsu, W, Imaizumi, Y, Zhang, B, Nabeshima, K, Mori, MX, Katsurabayashi, S, Shirasaka, Y, Okano, H & Hirose, S 2013, 'A human Dravet syndrome model from patient induced pluripotent stem cells', Mol Brain, vol. 6, pp. 19.

Hill, AJ, Mercier, MS, Hill, TD, Glyn, SE, Jones, NA, Yamasaki, Y, Futamura, T, Duncan, M, Stott, CG, Stephens, GJ, Williams, CM & Whalley, BJ 2012, 'Cannabidivarin is anticonvulsant in mouse and rat', Br J Pharmacol, vol. 167, no. 8, pp. 1629-42.

Hill, TD, Cascio, MG, Romano, B, Duncan, M, Pertwee, RG, Williams, CM, Whalley, BJ & Hill, AJ 2013, 'Cannabidivarin-rich cannabis extracts are anticonvulsant in mouse and rat via a CB1 receptor-independent mechanism', Br J Pharmacol, vol. 170, no. 3, pp. 679-92.

Ho, H, Both, M, Siniard, A, Sharma, S, Notwell, JH, Wallace, M, Leone, DP, Nguyen, A, Zhao, E, Lee, H, Zwilling, D, Thompson, KR, Braithwaite, SP, Huentelman, M & Portmann, T 2018, 'A Guide to Single-Cell Transcriptomics in Adult Rodent Brain: The Medium Spiny Neuron Transcriptome Revisited', Front Cell Neurosci, vol. 12, pp. 159.

Hohmann, U, Pelzer, M, Kleine, J, Hohmann, T, Ghadban, C & Dehghani, F 2019, 'Opposite Effects of Neuroprotective Cannabinoids, Palmitoylethanolamide, and 2- Arachidonoylglycerol on Function and Morphology of Microglia', Front Neurosci, vol. 13, pp. 1180.

Howlett, AC & Abood, ME 2017, 'CB1 and CB2 Receptor Pharmacology', Adv Pharmacol, vol. 80, pp. 169-206.

Howlett, AC, Barth, F, Bonner, TI, Cabral, G, Casellas, P, Devane, WA, Felder, CC, Herkenham, M, Mackie, K, Martin, BR, Mechoulam, R & Pertwee, RG 2002, 'International Union of Pharmacology. XXVII. Classification of cannabinoid receptors', Pharmacol Rev, vol. 54, no. 2, pp. 161-202.

Hsiao, J, Yuan, TY, Tsai, MS, Lu, CY, Lin, YC, Lee, ML, Lin, SW, Chang, FC, Liu Pimentel, H, Olive, C, Coito, C, Shen, G, Young, M, Thorne, T, Lawrence, M, Magistri, M, Faghihi, MA, Khorkova, O & Wahlestedt, C 2016, 'Upregulation of Haploinsufficient Gene Expression in the Brain by Targeting a Long Non-coding RNA Improves Seizure Phenotype in a Model of Dravet Syndrome', EBioMedicine, vol. 9, pp. 257-277.

155

Hsu, KL, Tsuboi, K, Adibekian, A, Pugh, H, Masuda, K & Cravatt, BF 2012, 'DAGLbeta inhibition perturbs a lipid network involved in macrophage inflammatory responses', Nat Chem Biol, vol. 8, no. 12, pp. 999-1007.

Huizenga, MN, Sepulveda-Rodriguez, A & Forcelli, PA 2019, 'Preclinical safety and efficacy of cannabidivarin for early life seizures', Neuropharmacology, vol. 148, pp. 189-198.

Hurst, K, Badgley, C, Ellsworth, T, Bell, S, Friend, L, Prince, B, Welch, J, Cowan, Z, Williamson, R, Lyon, C, Anderson, B, Poole, B, Christensen, M, Mcneil, M, Call, J & Edwards, JG 2017, 'A putative lysophosphatidylinositol receptor GPR55 modulates hippocampal synaptic plasticity', Hippocampus, vol. 27, no. 9, pp. 985-998.

Iannotti, FA, Di Marzo, V & Petrosino, S 2016, 'Endocannabinoids and endocannabinoid-related mediators: Targets, metabolism and role in neurological disorders', Prog Lipid Res, vol. 62, pp. 107-28.

Ignatowska-Jankowska, BM, Baillie, GL, Kinsey, S, Crowe, M, Ghosh, S, Owens, RA, Damaj, IM, Poklis, J, Wiley, JL, Zanda, M, Zanato, C, Greig, IR, Lichtman, AH & Ross, RA 2015, 'A Cannabinoid CB1 Receptor-Positive Allosteric Modulator Reduces Neuropathic Pain in the Mouse with No Psychoactive Effects', Neuropsychopharmacology, vol. 40, no. 13, pp. 2948-59.

Inloes, JM, Jing, H & Cravatt, BF 2018, 'The Spastic Paraplegia-Associated Phospholipase DDHD1 Is a Primary Brain Phosphatidylinositol Lipase', Biochemistry, vol. 57, no. 39, pp. 5759-5767.

Ishii, A, Watkins, JC, Chen, D, Hirose, S & Hammer, MF 2017, 'Clinical implications of SCN1A missense and truncation variants in a large Japanese cohort with Dravet syndrome', Epilepsia, vol. 58, no. 2, pp. 282-290.

Ito, S, Ogiwara, I, Yamada, K, Miyamoto, H, Hensch, TK, Osawa, M & Yamakawa, K 2013, 'Mouse with Nav1.1 haploinsufficiency, a model for Dravet syndrome, exhibits lowered sociability and learning impairment', Neurobiol Dis, vol. 49, pp. 29-40.

Jacher, JE, Roy, N, Ghaziuddin, M & Innis, JW 2019, 'Expanding the phenotypic spectrum of MBOAT7-related intellectual disability', Am J Med Genet B Neuropsychiatr Genet.

Jia, YF, Li, YC, Tang, YP, Cao, J, Wang, LP, Yang, YX, Xu, L & Mao, RR 2015, 'Interference of TRPV1 function altered the susceptibility of PTZ-induced seizures', Front Cell Neurosci, vol. 9, pp. 20.

Johansen, A, Rosti, RO, Musaev, D, Sticca, E, Harripaul, R, Zaki, M, Caglayan, AO, Azam, M, Sultan, T, Froukh, T, Reis, A, Popp, B, Ahmed, I, John, P, Ayub, M, Ben-Omran, T, Vincent, JB, Gleeson, JG & Abou Jamra, R 2016, 'Mutations in MBOAT7, Encoding Lysophosphatidylinositol Acyltransferase I, Lead to Intellectual Disability Accompanied by Epilepsy and Autistic Features', Am J Hum Genet, vol. 99, no. 4, pp. 912-916.

156

Kaczocha, M, Rebecchi, MJ, Ralph, BP, Teng, YH, Berger, WT, Galbavy, W, Elmes, MW, Glaser, ST, Wang, L, Rizzo, RC, Deutsch, DG & Ojima, I 2014, 'Inhibition of fatty acid binding proteins elevates brain anandamide levels and produces analgesia', PLoS One, vol. 9, no. 4, pp. e94200.

Kallendrusch, S, Kremzow, S, Nowicki, M, Grabiec, U, Winkelmann, R, Benz, A, Kraft, R, Bechmann, I, Dehghani, F & Koch, M 2013, 'The G protein-coupled receptor 55 ligand l- alpha-lysophosphatidylinositol exerts microglia-dependent neuroprotection after excitotoxic lesion', Glia, vol. 61, no. 11, pp. 1822-31.

Kalume, F, Oakley, JC, Westenbroek, RE, Gile, J, De La Iglesia, HO, Scheuer, T & Catterall, WA 2015, 'Sleep impairment and reduced interneuron excitability in a mouse model of Dravet Syndrome', Neurobiol Dis, vol. 77, pp. 141-54.

Kalume, F, Yu, FH, Westenbroek, RE, Scheuer, T & Catterall, WA 2007, 'Reduced sodium current in Purkinje neurons from Nav1.1 mutant mice: implications for ataxia in severe myoclonic epilepsy in infancy', J Neurosci, vol. 27, no. 41, pp. 11065-74.

Kanai, K, Hirose, S, Oguni, H, Fukuma, G, Shirasaka, Y, Miyajima, T, Wada, K, Iwasa, H, Yasumoto, S, Matsuo, M, Ito, M, Mitsudome, A & Kaneko, S 2004, 'Effect of localization of missense mutations in SCN1A on epilepsy phenotype severity', Neurology, vol. 63, no. 2, pp. 329-34.

Kano, M 2014, 'Control of synaptic function by endocannabinoid-mediated retrograde signaling', Proceedings of the Japan Academy, Series B, vol. 90, no. 7, pp. 235-250.

Kano, M, Ohno-Shosaku, T, Hashimotodani, Y, Uchigashima, M & Watanabe, M 2009, 'Endocannabinoid-mediated control of synaptic transmission', Physiol Rev, vol. 89, no. 1, pp. 309-80.

Kaplan, DI, Isom, LL & Petrou, S 2016, 'Role of Sodium Channels in Epilepsy', Cold Spring Harb Perspect Med, vol. 6, no. 6.

Kaplan, JS, Stella, N, Catterall, WA & Westenbroek, RE 2017, 'Cannabidiol attenuates seizures and social deficits in a mouse model of Dravet syndrome', Proc Natl Acad Sci U S A.

Kapur, A, Zhao, P, Sharir, H, Bai, Y, Caron, MG, Barak, LS & Abood, ME 2009, 'Atypical responsiveness of the orphan receptor GPR55 to cannabinoid ligands', J Biol Chem, vol. 284, no. 43, pp. 29817-27.

Kargl, J, Balenga, N, Parzmair, GP, Brown, AJ, Heinemann, A & Waldhoer, M 2012, 'The cannabinoid receptor CB1 modulates the signaling properties of the lysophosphatidylinositol receptor GPR55', J Biol Chem, vol. 287, no. 53, pp. 44234-48.

Karler, R & Turkanis, SA 1980, 'Subacute cannabinoid treatment: anticonvulsant activity and withdrawal excitability in mice', Br J Pharmacol, vol. 68, no. 3, pp. 479-84.

157

Katona, I 2015, 'Cannabis and Endocannabinoid Signaling in Epilepsy', Handb Exp Pharmacol, vol. 231, pp. 285-316.

Kaur, R, Ambwani, SR & Singh, S 2016, 'Endocannabinoid System: A Multi-Facet Therapeutic Target', Curr Clin Pharmacol, vol. 11, no. 2, pp. 110-7.

Kawamura, Y, Fukaya, M, Maejima, T, Yoshida, T, Miura, E, Watanabe, M, Ohno-Shosaku, T & Kano, M 2006, 'The CB1 cannabinoid receptor is the major cannabinoid receptor at excitatory presynaptic sites in the hippocampus and cerebellum', J Neurosci, vol. 26, no. 11, pp. 2991-3001.

Kearney, JA 2011, 'Genetic modifiers of neurological disease', Curr Opin Genet Dev, vol. 21, no. 3, pp. 349-53.

Kettenmann, H, Hanisch, UK, Noda, M & Verkhratsky, A 2011, 'Physiology of microglia', Physiol Rev, vol. 91, no. 2, pp. 461-553.

Kevin, RC, Anderson, L, Mcgregor, IS, Boyd, R, Manning, JJ, Glass, M, Connor, M & Banister, SD 2019, 'CUMYL-4CN-BINACA Is an Efficacious and Potent Pro-Convulsant Synthetic Cannabinoid Receptor Agonist', Front Pharmacol, vol. 10, pp. 595.

Knupp, KG & Wirrell, EC 2018, 'Treatment Strategies for Dravet Syndrome', CNS Drugs, vol. 32, no. 4, pp. 335-350.

Kong, WL, Min, JW, Liu, YL, Li, JX, He, XH & Peng, BW 2014, 'Role of TRPV1 in susceptibility to PTZ-induced seizure following repeated hyperthermia challenges in neonatal mice', Epilepsy Behav, vol. 31, pp. 276-80.

Korff, CM & Dale, RC 2017, 'The Immune System in Pediatric Seizures and Epilepsies', Pediatrics, vol. 140, no. 3.

Kotsikorou, E, Sharir, H, Shore, DM, Hurst, DP, Lynch, DL, Madrigal, KE, Heynen-Genel, S, Milan, LB, Chung, TD, Seltzman, HH, Bai, Y, Caron, MG, Barak, LS, Croatt, MP, Abood, ME & Reggio, PH 2013, 'Identification of the GPR55 antagonist binding site using a novel set of high-potency GPR55 selective ligands', Biochemistry, vol. 52, no. 52, pp. 9456-69.

Kow, RL, Jiang, K, Naydenov, AV, Le, JH, Stella, N & Nathanson, NM 2014, 'Modulation of pilocarpine-induced seizures by cannabinoid receptor 1', PLoS One, vol. 9, no. 4, pp. e95922.

Kramar, C, Loureiro, M, Renard, J & Laviolette, SR 2017, 'Palmitoylethanolamide Modulates GPR55 Receptor Signaling in the Ventral Hippocampus to Regulate Mesolimbic Dopamine Activity, Social Interaction, and Memory Processing', Cannabis Cannabinoid Res, vol. 2, no. 1, pp. 8-20.

Kreitzer, AC & Regehr, WG 2001, 'Retrograde inhibition of presynaptic calcium influx by endogenous cannabinoids at excitatory synapses onto Purkinje cells', Neuron, vol. 29, no. 3, pp. 717-27.

158

Kreutz, S, Koch, M, Bottger, C, Ghadban, C, Korf, HW & Dehghani, F 2009, '2- Arachidonoylglycerol elicits neuroprotective effects on excitotoxically lesioned dentate gyrus granule cells via abnormal-cannabidiol-sensitive receptors on microglial cells', Glia, vol. 57, no. 3, pp. 286-94.

Kreutz, S, Koch, M, Ghadban, C, Korf, HW & Dehghani, F 2007, 'Cannabinoids and neuronal damage: differential effects of THC, AEA and 2-AG on activated microglial cells and degenerating neurons in excitotoxically lesioned rat organotypic hippocampal slice cultures', Exp Neurol, vol. 203, no. 1, pp. 246-57.

Krueger, J & Berg, AT 2015, 'Incidence of Dravet Syndrome in a US Population', Pediatr Neurol Briefs, vol. 29, no. 12, pp. 92.

Kruk-Slomka, M, Dzik, A, Budzynska, B & Biala, G 2017, 'Endocannabinoid System: the Direct and Indirect Involvement in the Memory and Learning Processes-a Short Review', Mol Neurobiol, vol. 54, no. 10, pp. 8332-8347.

Kuo, FS, Cleary, CM, Loturco, JJ, Chen, X & Mulkey, DK 2019, 'Disordered breathing in a mouse model of Dravet syndrome', Elife, vol. 8.

Kwok, CH, Devonshire, IM, Imraish, A, Greenspon, CM, Lockwood, S, Fielden, C, Cooper, A, Woodhams, S, Sarmad, S, Ortori, CA, Barrett, DA, Kendall, D, Bennett, AJ, Chapman, V & Hathway, GJ 2017, 'Age-dependent plasticity in endocannabinoid modulation of pain processing through postnatal development', Pain, vol. 158, no. 11, pp. 2222-2232.

Lagae, L, Brambilla, I, Mingorance, A, Gibson, E & Battersby, A 2018, 'Quality of life and comorbidities associated with Dravet syndrome severity: a multinational cohort survey', Dev Med Child Neurol, vol. 60, no. 1, pp. 63-72.

Lagae, L, Sullivan, J, Knupp, K, Laux, L, Polster, T, Nikanorova, M, Devinsky, O, Cross, JH, Guerrini, R, Talwar, D, Miller, I, Farfel, G, Galer, BS, Gammaitoni, A, Mistry, A, Morrison, G, Lock, M, Agarwal, A, Lai, WW & Ceulemans, B 2019, 'Fenfluramine hydrochloride for the treatment of seizures in Dravet syndrome: a randomised, double- blind, placebo-controlled trial', The Lancet, vol. 394, no. 10216, pp. 2243-2254.

Lane, JR, Beukers, MW, Mulder-Krieger, T & Ijzerman, AP 2010, 'The endocannabinoid 2- arachidonylglycerol is a negative allosteric modulator of the human A3 adenosine receptor', Biochem Pharmacol, vol. 79, no. 1, pp. 48-56.

Laprairie, RB, Kulkarni, PM, Deschamps, JR, Kelly, MEM, Janero, DR, Cascio, MG, Stevenson, LA, Pertwee, RG, Kenakin, TP, Denovan-Wright, EM & Thakur, GA 2017, 'Enantiospecific Allosteric Modulation of Cannabinoid 1 Receptor', ACS Chem Neurosci, vol. 8, no. 6, pp. 1188-1203.

Latini, G, Verrotti, A, Manco, R, Scardapane, A, Del Vecchio, A & Chiarelli, F 2008, 'Topiramate: its pharmacological properties and therapeutic efficacy in epilepsy', Mini Rev Med Chem, vol. 8, no. 1, pp. 10-23.

159

Lauckner, JE, Jensen, JB, Chen, HY, Lu, HC, Hille, B & Mackie, K 2008, 'GPR55 is a cannabinoid receptor that increases intracellular calcium and inhibits M current', Proc Natl Acad Sci U S A, vol. 105, no. 7, pp. 2699-704.

Lee, HC, Inoue, T, Sasaki, J, Kubo, T, Matsuda, S, Nakasaki, Y, Hattori, M, Tanaka, F, Udagawa, O, Kono, N, Itoh, T, Ogiso, H, Taguchi, R, Arita, M, Sasaki, T & Arai, H 2012, 'LPIAT1 regulates arachidonic acid content in phosphatidylinositol and is required for cortical lamination in mice', Mol Biol Cell, vol. 23, no. 24, pp. 4689-700.

Lee, SH, Foldy, C & Soltesz, I 2010, 'Distinct endocannabinoid control of GABA release at perisomatic and dendritic synapses in the hippocampus', J Neurosci, vol. 30, no. 23, pp. 7993-8000.

Lerner, R, Post, J, Loch, S, Lutz, B & Bindila, L 2017, 'Targeting brain and peripheral plasticity of the lipidome in acute kainic acid-induced epileptic seizures in mice via quantitative mass spectrometry', Biochim Biophys Acta, vol. 1862, no. 2, pp. 255-267.

Li, Y & Kim, J 2017, 'Distinct roles of neuronal and microglial CB2 cannabinoid receptors in the mouse hippocampus', Neuroscience, vol. 363, pp. 11-25.

Liautard, C, Scalmani, P, Carriero, G, De Curtis, M, Franceschetti, S & Mantegazza, M 2013, 'Hippocampal hyperexcitability and specific epileptiform activity in a mouse model of Dravet syndrome', Epilepsia, vol. 54, no. 7, pp. 1251-61.

Licheni, SH, Mcmahon, JM, Schneider, AL, Davey, MJ & Scheffer, IE 2018, 'Sleep problems in Dravet syndrome: a modifiable comorbidity', Dev Med Child Neurol, vol. 60, no. 2, pp. 192-198.

Liedhegner, ES, Sasman, A & Hillard, CJ 2014, 'Brain region-specific changes in N- acylethanolamine contents with time of day', J Neurochem, vol. 128, no. 4, pp. 491-506.

Liu, J, Tong, L, Song, S, Niu, Y, Li, J, Wu, X, Zhang, J, Zai, CC, Luo, F, Wu, J, Li, H, Wong, AHC, Sun, R, Liu, F & Li, B 2018, 'Novel and de novo mutations in pediatric refractory epilepsy', Mol Brain, vol. 11, no. 1, pp. 48.

Liu, Y, Lopez-Santiago, LF, Yuan, Y, Jones, JM, Zhang, H, O'malley, HA, Patino, GA, O'brien, JE, Rusconi, R, Gupta, A, Thompson, RC, Natowicz, MR, Meisler, MH, Isom, LL & Parent, JM 2013, 'Dravet syndrome patient-derived neurons suggest a novel epilepsy mechanism', Ann Neurol, vol. 74, no. 1, pp. 128-39.

Louh, IK & Freeman, WD 2014, 'A 'spicy' encephalopathy: synthetic cannabinoids as cause of encephalopathy and seizure', Crit Care, vol. 18, no. 5, pp. 553.

Lu, HC & Mackie, K 2016, 'An Introduction to the Endogenous Cannabinoid System', Biol Psychiatry, vol. 79, no. 7, pp. 516-25.

160

Luchicchi, A & Pistis, M 2012, 'Anandamide and 2-arachidonoylglycerol: pharmacological properties, functional features, and emerging specificities of the two major endocannabinoids', Mol Neurobiol, vol. 46, no. 2, pp. 374-92.

Ludanyi, A, Eross, L, Czirjak, S, Vajda, J, Halasz, P, Watanabe, M, Palkovits, M, Magloczky, Z, Freund, TF & Katona, I 2008, 'Downregulation of the CB1 cannabinoid receptor and related molecular elements of the endocannabinoid system in epileptic human hippocampus', J Neurosci, vol. 28, no. 12, pp. 2976-90.

Luk, J, Lu, Y, Ackermann, A, Peng, X, Bogdan, D, Puopolo, M, Komatsu, DE, Tong, S, Ojima, I, Rebecchi, MJ & Kaczocha, M 2018, 'Contribution of diacylglycerol lipase beta to pain after surgery', J Pain Res, vol. 11, pp. 473-482.

Lyseng-Williamson, KA & Yang, LPH 2008, 'Spotlight on Topiramate in Epilepsy', CNS Drugs, vol. 22, no. 2, pp. 171-174.

Ma, L, Wang, L, Yang, F, Meng, XD, Wu, C, Ma, H & Jiang, W 2014, 'Disease-modifying effects of RHC80267 and JZL184 in a pilocarpine mouse model of temporal lobe epilepsy', CNS Neurosci Ther, vol. 20, no. 10, pp. 905-15.

Maa, E & Figi, P 2014, 'The case for medical marijuana in epilepsy', Epilepsia, vol. 55, no. 6, pp. 783-6.

Maeda, H, Chiyonobu, T, Yoshida, M, Yamashita, S, Zuiki, M, Kidowaki, S, Isoda, K, Yamakawa, K, Morimoto, M, Nakahata, T, Saito, MK & Hosoi, H 2016, 'Establishment of isogenic iPSCs from an individual with SCN1A mutation mosaicism as a model for investigating neurocognitive impairment in Dravet syndrome', J Hum Genet, vol. 61, no. 6, pp. 565-9.

Malyshevskaya, O, Aritake, K, Kaushik, MK, Uchiyama, N, Cherasse, Y, Kikura-Hanajiri, R & Urade, Y 2017, 'Natural ((9)-THC) and synthetic (JWH-018) cannabinoids induce seizures by acting through the cannabinoid CB1 receptor', Sci Rep, vol. 7, no. 1, pp. 10516.

Marichal-Cancino, BA, Fajardo-Valdez, A, Ruiz-Contreras, AE, Mendez-Diaz, M & Prospero- Garcia, O 2017, 'Advances in the Physiology of GPR55 in the Central Nervous System', Curr Neuropharmacol, vol. 15, no. 5, pp. 771-778.

Marichal-Cancino, BA, Fajardo-Valdez, A, Ruiz-Contreras, AE, Mendez-Diaz, M & Prospero- Garcia, O 2018, 'Possible role of hippocampal GPR55 in spatial learning and memory in rats', Acta Neurobiol Exp (Wars), vol. 78, no. 1, pp. 41-50.

Marichal-Cancino, BA, Manrique-Maldonado, G, Altamirano-Espinoza, AH, Ruiz-Salinas, I, Gonzalez-Hernandez, A, Maassenvandenbrink, A & Villalon, CM 2013, 'Analysis of anandamide- and lysophosphatidylinositol-induced inhibition of the vasopressor responses produced by sympathetic stimulation or noradrenaline in pithed rats', Eur J Pharmacol, vol. 721, no. 1-3, pp. 168-77.

Marini, C, Mei, D, Temudo, T, Ferrari, AR, Buti, D, Dravet, C, Dias, AI, Moreira, A, Calado, E, Seri, S, Neville, B, Narbona, J, Reid, E, Michelucci, R, Sicca, F, Cross, HJ & Guerrini, R

161

2007, 'Idiopathic epilepsies with seizures precipitated by fever and SCN1A abnormalities', Epilepsia, vol. 48, no. 9, pp. 1678-1685.

Marini, C, Scheffer, IE, Nabbout, R, Suls, A, De Jonghe, P, Zara, F & Guerrini, R 2011, 'The genetics of Dravet syndrome', Epilepsia, vol. 52 Suppl 2, pp. 24-9.

Martin, P, De Witte, PaM, Maurice, T, Gammaitoni, A, Farfel, G & Galer, B 2020, 'Fenfluramine acts as a positive modulator of sigma-1 receptors', Epilepsy Behav, vol. 105, pp. 106989.

Marsicano, G, Goodenough, S, Monory, K, Hermann, H, Eder, M, Cannich, A, Azad, SC, Cascio, MG, Gutierrez, SO, Van Der Stelt, M, Lopez-Rodriguez, ML, Casanova, E, Schutz, G, Zieglgansberger, W, Di Marzo, V, Behl, C & Lutz, B 2003, 'CB1 cannabinoid receptors and on-demand defense against excitotoxicity', Science, vol. 302, no. 5642, pp. 84-8.

Marsicano, G & Lutz, B 1999, 'Expression of the cannabinoid receptor CB1 in distinct neuronal subpopulations in the adult mouse forebrain', Eur J Neurosci, vol. 11, no. 12, pp. 4213-25.

Martin, MS, Tang, B, Papale, LA, Yu, FH, Catterall, WA & Escayg, A 2007, 'The voltage-gated sodium channel Scn8a is a genetic modifier of severe myoclonic epilepsy of infancy', Hum Mol Genet, vol. 16, no. 23, pp. 2892-9.

Martinez-Pinilla, E, Reyes-Resina, I, Onatibia-Astibia, A, Zamarbide, M, Ricobaraza, A, Navarro, G, Moreno, E, Dopeso-Reyes, IG, Sierra, S, Rico, AJ, Roda, E, Lanciego, JL & Franco, R 2014, 'CB1 and GPR55 receptors are co-expressed and form heteromers in rat and monkey striatum', Exp Neurol, vol. 261, pp. 44-52.

Mccoy, B, Wang, L, Zak, M, Al-Mehmadi, S, Kabir, N, Alhadid, K, Mcdonald, K, Zhang, G, Sharma, R, Whitney, R, Sinopoli, K & Snead, OC, 3rd 2018, 'A prospective open-label trial of a CBD/THC cannabis oil in dravet syndrome', Ann Clin Transl Neurol, vol. 5, no. 9, pp. 1077-1088.

Meadows, A, Lee, JH, Wu, CS, Wei, Q, Pradhan, G, Yafi, M, Lu, HC & Sun, Y 2016, 'Deletion of G-protein-coupled receptor 55 promotes obesity by reducing physical activity', Int J Obes (Lond), vol. 40, no. 3, pp. 417-24.

Mechoulam, R, Ben-Shabat, S, Hanus, L, Ligumsky, M, Kaminski, NE, Schatz, AR, Gopher, A, Almog, S, Martin, BR, Compton, DR & Et Al. 1995, 'Identification of an endogenous 2- monoglyceride, present in canine gut, that binds to cannabinoid receptors', Biochem Pharmacol, vol. 50, no. 1, pp. 83-90.

Mechoulam, R & Parker, LA 2013, 'The endocannabinoid system and the brain', Annu Rev Psychol, vol. 64, pp. 21-47.

Mei, L, Fan, Y, Lv, X, Welsh, DK, Zhan, C & Zhang, EE 2018, 'Long-term in vivo recording of circadian rhythms in brains of freely moving mice', Proceedings of the National Academy of Sciences, vol. 115, no. 16, pp. 4276.

162

Meisler, MH, Kearney, J, Ottman, R & Escayg, A 2001, 'Identification of epilepsy genes in human and mouse', Annu Rev Genet, vol. 35, pp. 567-88.

Meisler, MH & Kearney, JA 2005, 'Sodium channel mutations in epilepsy and other neurological disorders', J Clin Invest, vol. 115, no. 8, pp. 2010-7.

Meyer, HC, Lee, FS & Gee, DG 2018, 'The Role of the Endocannabinoid System and Genetic Variation in Adolescent Brain Development', Neuropsychopharmacology, vol. 43, no. 1, pp. 21-33.

Mijangos-Moreno, S, Poot-Aké, A, Arankowsky-Sandoval, G & Murillo-Rodríguez, E 2014, 'Intrahypothalamic injection of cannabidiol increases the extracellular levels of adenosine in nucleus accumbens in rats', Neuroscience Research, vol. 84, pp. 60-63.

Miller, AR, Hawkins, NA, Mccollom, CE & Kearney, JA 2014, 'Mapping genetic modifiers of survival in a mouse model of Dravet syndrome', Genes Brain Behav, vol. 13, no. 2, pp. 163-72.

Mistry, AM, Thompson, CH, Miller, AR, Vanoye, CG, George, AL, Jr. & Kearney, JA 2014, 'Strain- and age-dependent hippocampal neuron sodium currents correlate with epilepsy severity in Dravet syndrome mice', Neurobiol Dis, vol. 65, pp. 1-11.

Mitjavila, J, Yin, D, Kulkarni, PM, Zanato, C, Thakur, GA, Ross, R, Greig, I, Mackie, K & Straiker, A 2018, 'Enantiomer-specific positive allosteric modulation of CB1 signaling in autaptic hippocampal neurons', Pharmacol Res, vol. 129, pp. 475-481.

Monory, K, Massa, F, Egertova, M, Eder, M, Blaudzun, H, Westenbroek, R, Kelsch, W, Jacob, W, Marsch, R, Ekker, M, Long, J, Rubenstein, JL, Goebbels, S, Nave, KA, During, M, Klugmann, M, Wolfel, B, Dodt, HU, Zieglgansberger, W, Wotjak, CT, Mackie, K, Elphick, MR, Marsicano, G & Lutz, B 2006, 'The endocannabinoid system controls key epileptogenic circuits in the hippocampus', Neuron, vol. 51, no. 4, pp. 455-66.

Morrison, G, Crockett, J, Blakey, G & Sommerville, K 2019, 'A Phase 1, Open-Label, Pharmacokinetic Trial to Investigate Possible Drug-Drug Interactions Between Clobazam, Stiripentol, or Valproate and Cannabidiol in Healthy Subjects', Clin Pharmacol Drug Dev.

Murase, S & Okuyama, H 1985, 'A membrane-bound phospholipase C with an apparent specificity for lysophosphatidylinositol in porcine platelets', J Biol Chem, vol. 260, no. 1, pp. 262-5.

Murataeva, N, Straiker, A & Mackie, K 2014, 'Parsing the players: 2-arachidonoylglycerol synthesis and degradation in the CNS', Br J Pharmacol, vol. 171, no. 6, pp. 1379-91.

Musella, A, Fresegna, D, Rizzo, FR, Gentile, A, Bullitta, S, De Vito, F, Guadalupi, L, Centonze, D & Mandolesi, G 2017, 'A novel crosstalk within the endocannabinoid system controls GABA transmission in the striatum', Sci Rep, vol. 7, no. 1, pp. 7363.

163

Naderi, N, Shafieirad, E, Lakpoor, D, Rahimi, A & Mousavi, Z 2015, 'Interaction between Cannabinoid Compounds and Capsazepine in Protection against Acute Pentylenetetrazole- induced Seizure in Mice', Iran J Pharm Res, vol. 14, no. Suppl, pp. 115-20.

Nagarkatti, P, Pandey, R, Rieder, SA, Hegde, VL & Nagarkatti, M 2009, 'Cannabinoids as novel anti-inflammatory drugs', Future Med Chem, vol. 1, no. 7, pp. 1333-49.

Naziroglu, M 2015, 'TRPV1 Channel: A Potential Drug Target for Treating Epilepsy', Curr Neuropharmacol, vol. 13, no. 2, pp. 239-47.

Naziroglu, M & Ovey, IS 2015, 'Involvement of apoptosis and calcium accumulation through TRPV1 channels in neurobiology of epilepsy', Neuroscience, vol. 293, pp. 55-66.

Neu, A, Foldy, C & Soltesz, I 2007, 'Postsynaptic origin of CB1-dependent tonic inhibition of GABA release at cholecystokinin-positive basket cell to pyramidal cell synapses in the CA1 region of the rat hippocampus', J Physiol, vol. 578, no. Pt 1, pp. 233-47.

Nicolussi, S & Gertsch, J 2015, 'Endocannabinoid transport revisited', Vitam Horm, vol. 98, pp. 441-85.

Niibori, Y, Lee, S, Minassian, BA & Hampson, DR 2019, 'Sexually Divergent Mortality and Partial Phenotypic Rescue after Gene Therapy in a Mouse Model of Dravet Syndrome', Hum Gene Ther.

Nourbakhsh, F, Atabaki, R & Roohbakhsh, A 2018, 'The role of orphan G protein-coupled receptors in the modulation of pain: A review', Life Sci, vol. 212, pp. 59-69.

Oakley, JC, Kalume, F, Yu, FH, Scheuer, T & Catterall, WA 2009, 'Temperature- and age- dependent seizures in a mouse model of severe myoclonic epilepsy in infancy', Proc Natl Acad Sci U S A, vol. 106, no. 10, pp. 3994-9.

Ogiwara, I, Iwasato, T, Miyamoto, H, Iwata, R, Yamagata, T, Mazaki, E, Yanagawa, Y, Tamamaki, N, Hensch, TK, Itohara, S & Yamakawa, K 2013, 'Nav1.1 haploinsufficiency in excitatory neurons ameliorates seizure-associated sudden death in a mouse model of Dravet syndrome', Hum Mol Genet, vol. 22, no. 23, pp. 4784-804.

Ogiwara, I, Miyamoto, H, Morita, N, Atapour, N, Mazaki, E, Inoue, I, Takeuchi, T, Itohara, S, Yanagawa, Y, Obata, K, Furuichi, T, Hensch, TK & Yamakawa, K 2007, 'Nav1.1 localizes to axons of parvalbumin-positive inhibitory interneurons: a circuit basis for epileptic seizures in mice carrying an Scn1a gene mutation', J Neurosci, vol. 27, no. 22, pp. 5903- 14.

Ohmori, I, Ohtsuka, Y, Ouchida, M, Ogino, T, Maniwa, S, Shimizu, K & Oka, E 2003, 'Is phenotype difference in severe myoclonic epilepsy in infancy related to SCN1A mutations?', Brain Dev, vol. 25, no. 7, pp. 488-93.

164

Ohno-Shosaku, T, Tsubokawa, H, Mizushima, I, Yoneda, N, Zimmer, A & Kano, M 2002, 'Presynaptic cannabinoid sensitivity is a major determinant of depolarization-induced retrograde suppression at hippocampal synapses', J Neurosci, vol. 22, no. 10, pp. 3864-72.

Oka, S, Kimura, S, Toshida, T, Ota, R, Yamashita, A & Sugiura, T 2010, 'Lysophosphatidylinositol induces rapid phosphorylation of p38 mitogen-activated protein kinase and activating transcription factor 2 in HEK293 cells expressing GPR55 and IM-9 lymphoblastoid cells', J Biochem, vol. 147, no. 5, pp. 671-8.

Oka, S, Nakajima, K, Yamashita, A, Kishimoto, S & Sugiura, T 2007, 'Identification of GPR55 as a lysophosphatidylinositol receptor', Biochem Biophys Res Commun, vol. 362, no. 4, pp. 928-34.

Oka, S, Toshida, T, Maruyama, K, Nakajima, K, Yamashita, A & Sugiura, T 2009, '2- Arachidonoyl-sn-glycero-3-phosphoinositol: a possible natural ligand for GPR55', J Biochem, vol. 145, no. 1, pp. 13-20.

Oliveira Da Cruz, JF, Robin, LM, Drago, F, Marsicano, G & Metna-Laurent, M 2016, 'Astroglial type-1 cannabinoid receptor (CB1): A new player in the tripartite synapse', Neuroscience, vol. 323, pp. 35-42.

Olivieri, G, Battaglia, D, Chieffo, D, Rubbino, R, Ranalli, D, Contaldo, I, Dravet, C, Mercuri, E & Guzzetta, F 2016, 'Cognitive-behavioral profiles in teenagers with Dravet syndrome', Brain Dev, vol. 38, no. 6, pp. 554-62.

Oyrer, J, Maljevic, S, Scheffer, IE, Berkovic, SF, Petrou, S & Reid, CA 2018, 'Ion Channels in Genetic Epilepsy: From Genes and Mechanisms to Disease-Targeted Therapies', Pharmacological Reviews, vol. 70, no. 1, pp. 142-173.

Pamplona, FA, Da Silva, LR & Coan, AC 2018, 'Potential Clinical Benefits of CBD-Rich Cannabis Extracts Over Purified CBD in Treatment-Resistant Epilepsy: Observational Data Meta- analysis', Front Neurol, vol. 9, pp. 759.

Pane, C & Sacca, F 2019, 'The use of medical grade cannabis in Italy for drug-resistant epilepsy: a case series', Neurol Sci.

Parsons, LH & Hurd, YL 2015, 'Endocannabinoid signalling in reward and addiction', Nat Rev Neurosci, vol. 16, no. 10, pp. 579-94.

Peacock, A, Leung, J, Larney, S, Colledge, S, Hickman, M, Rehm, J, Giovino, GA, West, R, Hall, W, Griffiths, P, Ali, R, Gowing, L, Marsden, J, Ferrari, AJ, Grebely, J, Farrell, M & Degenhardt, L 2018, 'Global statistics on alcohol, tobacco and illicit drug use: 2017 status report', Addiction, vol. 113, no. 10, pp. 1905-1926.

Perescis, MF, De Bruin, N, Heijink, L, Kruse, C, Vinogradova, L, Luttjohann, A, Van Luijtelaar, G & Van Rijn, CM 2017, 'Cannabinoid antagonist SLV326 induces convulsive seizures and changes in the interictal EEG in rats', PLoS One, vol. 12, no. 2, pp. e0165363.

165

Perez, A, Garcia-Penton, L, Canales-Rodriguez, EJ, Lerma-Usabiaga, G, Iturria-Medina, Y, Roman, FJ, Davidson, D, Aleman-Gomez, Y, Acha, J & Carreiras, M 2014, 'Brain morphometry of Dravet syndrome', Epilepsy Res, vol. 108, no. 8, pp. 1326-34.

Petrosino, S, Schiano Moriello, A, Cerrato, S, Fusco, M, Puigdemont, A, De Petrocellis, L & Di Marzo, V 2016, 'The anti-inflammatory mediator palmitoylethanolamide enhances the levels of 2-arachidonoyl-glycerol and potentiates its actions at TRPV1 cation channels', Br J Pharmacol, vol. 173, no. 7, pp. 1154-62.

Pietr, M, Kozela, E, Levy, R, Rimmerman, N, Lin, YH, Stella, N, Vogel, Z & Juknat, A 2009, 'Differential changes in GPR55 during microglial cell activation', FEBS Lett, vol. 583, no. 12, pp. 2071-6.

Pitler, TA & Alger, BE 1992, 'Postsynaptic spike firing reduces synaptic GABAA responses in hippocampal pyramidal cells', J Neurosci, vol. 12, no. 10, pp. 4122-32.

Poisson, M, Huguet, F, Savattier, A, Bakri-Logeais, F & Narcisse, G 1984, 'A new type of anticonvulsant, stiripentol. Pharmacological profile and neurochemical study', Arzneimittelforschung, vol. 34, no. 2, pp. 199-204.

Porter, BE & Jacobson, C 2013, 'Report of a parent survey of cannabidiol-enriched cannabis use in pediatric treatment-resistant epilepsy', Epilepsy Behav, vol. 29, no. 3, pp. 574-7.

Post, JM, Loch, S, Lerner, R, Remmers, F, Lomazzo, E, Lutz, B & Bindila, L 2018, 'Antiepileptogenic Effect of Subchronic Palmitoylethanolamide Treatment in a Mouse Model of Acute Epilepsy', Front Mol Neurosci, vol. 11, pp. 67.

Prather, PL, Martin, NA, Breivogel, CS & Childers, SR 2000, 'Activation of cannabinoid receptors in rat brain by WIN 55212-2 produces coupling to multiple G protein alpha-subunits with different potencies', Mol Pharmacol, vol. 57, no. 5, pp. 1000-10.

Press, CA, Knupp, KG & Chapman, KE 2015, 'Parental reporting of response to oral cannabis extracts for treatment of refractory epilepsy', Epilepsy Behav, vol. 45, pp. 49-52.

Price, MR, Baillie, GL, Thomas, A, Stevenson, LA, Easson, M, Goodwin, R, Mclean, A, Mcintosh, L, Goodwin, G, Walker, G, Westwood, P, Marrs, J, Thomson, F, Cowley, P, Christopoulos, A, Pertwee, RG & Ross, RA 2005, 'Allosteric modulation of the cannabinoid CB1 receptor', Mol Pharmacol, vol. 68, no. 5, pp. 1484-95.

Ragona, F, Granata, T, Dalla Bernardina, B, Offredi, F, Darra, F, Battaglia, D, Morbi, M, Brazzo, D, Cappelletti, S, Chieffo, D, De Giorgi, I, Fontana, E, Freri, E, Marini, C, Toraldo, A, Specchio, N, Veggiotti, P, Vigevano, F, Guerrini, R, Guzzetta, F & Dravet, C 2011, 'Cognitive development in Dravet syndrome: a retrospective, multicenter study of 26 patients', Epilepsia, vol. 52, no. 2, pp. 386-92.

Rahimi, A, Hajizadeh Moghaddam, A & Roohbakhsh, A 2015, 'Central administration of GPR55 receptor agonist and antagonist modulates anxiety-related behaviors in rats', Fundam Clin Pharmacol, vol. 29, no. 2, pp. 185-90.

166

Rho, JM 2017, 'How does the ketogenic diet induce anti-seizure effects?', Neurosci Lett, vol. 637, pp. 4-10.

Riedel, G & Davies, SN 2005, 'Cannabinoid function in learning, memory and plasticity', Handb Exp Pharmacol, no. 168, pp. 445-77.

Romigi, A, Bari, M, Placidi, F, Marciani, MG, Malaponti, M, Torelli, F, Izzi, F, Prosperetti, C, Zannino, S, Corte, F, Chiaramonte, C & Maccarrone, M 2010, 'Cerebrospinal fluid levels of the endocannabinoid anandamide are reduced in patients with untreated newly diagnosed temporal lobe epilepsy', Epilepsia, vol. 51, no. 5, pp. 768-72.

Rosenberg, EC, Patra, PH & Whalley, BJ 2017, 'Therapeutic effects of cannabinoids in animal models of seizures, epilepsy, epileptogenesis, and epilepsy-related neuroprotection', Epilepsy Behav, vol. 70, no. Pt B, pp. 319-327.

Ross, RA 2009, 'The enigmatic pharmacology of GPR55', Trends in Pharmacological Sciences, vol. 30, no. 3, pp. 156-163.

Rowley, S, Sun, X, Lima, IV, Tavenier, A, De Oliveira, ACP, Dey, SK & Danzer, SC 2017, 'Cannabinoid receptor 1/2 double-knockout mice develop epilepsy', Epilepsia, vol. 58, no. 12, pp. e162-e166.

Rubinstein, M, Westenbroek, RE, Yu, FH, Jones, CJ, Scheuer, T & Catterall, WA 2015, 'Genetic background modulates impaired excitability of inhibitory neurons in a mouse model of Dravet syndrome', Neurobiol Dis, vol. 73, pp. 106-17.

Rudenko, V, Rafiuddin, A, Leheste, JR & Friedman, LK 2012, 'Inverse relationship of cannabimimetic (R+)WIN 55, 212 on behavior and seizure threshold during the juvenile period', Pharmacol Biochem Behav, vol. 100, no. 3, pp. 474-84.

Ruffolo, G, Cifelli, P, Roseti, C, Thom, M, Van Vliet, EA, Limatola, C, Aronica, E & Palma, E 2018, 'A novel GABAergic dysfunction in human Dravet syndrome', Epilepsia, vol. 59, no. 11, pp. 2106-2117.

Russo, EB, Burnett, A, Hall, B & Parker, KK 2005, 'Agonistic properties of cannabidiol at 5-HT1a receptors', Neurochem Res, vol. 30, no. 8, pp. 1037-43.

Ryberg, E, Larsson, N, Sjogren, S, Hjorth, S, Hermansson, NO, Leonova, J, Elebring, T, Nilsson, K, Drmota, T & Greasley, PJ 2007, 'The orphan receptor GPR55 is a novel cannabinoid receptor', Br J Pharmacol, vol. 152, no. 7, pp. 1092-101.

Sada, N, Lee, S, Katsu, T, Otsuki, T & Inoue, T 2015, 'Epilepsy treatment. Targeting LDH enzymes with a stiripentol analog to treat epilepsy', Science, vol. 347, no. 6228, pp. 1362- 7.

Sankar, R 2012, 'GABA(A) receptor physiology and its relationship to the mechanism of action of the 1,5-benzodiazepine clobazam', CNS Drugs, vol. 26, no. 3, pp. 229-44.

167

Savage, TE, Sourbron, J, Bruno, PL, Skirvin, LA, Wolper, ES, Anagnos, CJ & Thiele, EA 2020, 'Efficacy of cannabidiol in subjects with refractory epilepsy relative to concomitant use of clobazam', Epilepsy Res, vol. 160, pp. 106263.

Scheffer, IE, Berkovic, S, Capovilla, G, Connolly, MB, French, J, Guilhoto, L, Hirsch, E, Jain, S, Mathern, GW, Moshe, SL, Nordli, DR, Perucca, E, Tomson, T, Wiebe, S, Zhang, YH & Zuberi, SM 2017, 'ILAE classification of the epilepsies: Position paper of the ILAE Commission for Classification and Terminology', Epilepsia, vol. 58, no. 4, pp. 512-521.

Schep, LJ, Slaughter, RJ, Hudson, S, Place, R & Watts, M 2015, 'Delayed seizure-like activity following analytically confirmed use of previously unreported synthetic cannabinoid analogues', Hum Exp Toxicol, vol. 34, no. 5, pp. 557-60.

Schlosburg, JE, Blankman, JL, Long, JZ, Nomura, DK, Pan, B, Kinsey, SG, Nguyen, PT, Ramesh, D, Booker, L, Burston, JJ, Thomas, EA, Selley, DE, Sim-Selley, LJ, Liu, QS, Lichtman, AH & Cravatt, BF 2010, 'Chronic monoacylglycerol lipase blockade causes functional antagonism of the endocannabinoid system', Nat Neurosci, vol. 13, no. 9, pp. 1113-9.

Schoonjans, A, Paelinck, BP, Marchau, F, Gunning, B, Gammaitoni, A, Galer, BS, Lagae, L & Ceulemans, B 2017a, 'Low-dose fenfluramine significantly reduces seizure frequency in Dravet syndrome: a prospective study of a new cohort of patients', Eur J Neurol, vol. 24, no. 2, pp. 309-314.

Schoonjans, AS, Marchau, F, Paelinck, BP, Lagae, L, Gammaitoni, A, Pringsheim, M, Keane, MG & Ceulemans, B 2017b, 'Cardiovascular safety of low-dose fenfluramine in Dravet syndrome: a review of its benefit-risk profile in a new patient population', Curr Med Res Opin, vol. 33, no. 10, pp. 1773-1781.

Scott, CE & Kendall, DA 2017, 'Assessing Allosteric Modulation of CB(1) at the Receptor and Cellular Levels', Methods in enzymology, vol. 593, pp. 317-342.

Serrano, A, Pavon, FJ, Buczynski, MW, Schlosburg, J, Natividad, LA, Polis, IY, Stouffer, DG, Zorrilla, EP, Roberto, M, Cravatt, BF, Martin-Fardon, R, Rodriguez De Fonseca, F & Parsons, LH 2018, 'Deficient endocannabinoid signaling in the central amygdala contributes to alcohol dependence-related anxiety-like behavior and excessive alcohol intake', Neuropsychopharmacology, vol. 43, no. 9, pp. 1840-1850.

Shi, XY, Tomonoh, Y, Wang, WZ, Ishii, A, Higurashi, N, Kurahashi, H, Kaneko, S, Hirose, S & Epilepsy Genetic Study Group, J 2015, 'Efficacy of antiepileptic drugs for the treatment of Dravet syndrome with different genotypes', Brain Dev.

Shirazi, M, Izadi, M, Amin, M, Rezvani, ME, Roohbakhsh, A & Shamsizadeh, A 2014, 'Involvement of central TRPV1 receptors in pentylenetetrazole and amygdala-induced kindling in male rats', Neurol Sci, vol. 35, no. 8, pp. 1235-41.

Shmuely, S, Sisodiya, SM, Gunning, WB, Sander, JW & Thijs, RD 2016, 'Mortality in Dravet syndrome: A review', Epilepsy Behav, vol. 64, no. Pt A, pp. 69-74.

168

Siegler, Z, Barsi, P, Neuwirth, M, Jerney, J, Kassay, M, Janszky, J, Paraicz, E, Hegyi, M & Fogarasi, A 2005, 'Hippocampal sclerosis in severe myoclonic epilepsy in infancy: a retrospective MRI study', Epilepsia, vol. 46, no. 5, pp. 704-8.

Simcocks, AC, O’keefe, L, Hryciw, DH, Mathai, ML, Hutchinson, DS & Mcainch, AJ 2018, 'GPR55', in Choi, S (ed.), Encyclopedia of Signaling Molecules, Springer International Publishing, Cham.

Simeone, TA, Simeone, KA & Rho, JM 2017, 'Ketone Bodies as Anti-Seizure Agents', Neurochem Res, vol. 42, no. 7, pp. 2011-2018.

Sinoo, C, De Lange, IM, Westers, P, Gunning, WB, Jongmans, MJ & Brilstra, EH 2019, 'Behavior problems and health-related quality of life in Dravet syndrome', Epilepsy Behav, vol. 90, pp. 217-227.

Slivicki, RA, Xu, Z, Kulkarni, PM, Pertwee, RG, Mackie, K, Thakur, GA & Hohmann, AG 2018, 'Positive Allosteric Modulation of Cannabinoid Receptor Type 1 Suppresses Pathological Pain Without Producing Tolerance or Dependence', Biol Psychiatry, vol. 84, no. 10, pp. 722-733.

Smith, DR, Stanley, CM, Foss, T, Boles, RG & Mckernan, K 2017, 'Rare genetic variants in the endocannabinoid system genes CNR1 and DAGLA are associated with neurological phenotypes in humans', PLoS One, vol. 12, no. 11, pp. e0187926.

Soltesz, I, Alger, BE, Kano, M, Lee, SH, Lovinger, DM, Ohno-Shosaku, T & Watanabe, M 2015, 'Weeding out bad waves: towards selective cannabinoid circuit control in epilepsy', Nat Rev Neurosci, vol. 16, no. 5, pp. 264-77.

Sourbron, J, Smolders, I, De Witte, P & Lagae, L 2017, 'Pharmacological Analysis of the Anti- epileptic Mechanisms of Fenfluramine in scn1a Mutant Zebrafish', Front Pharmacol, vol. 8, pp. 191.

Stein, RE, Kaplan, JS, Li, J & Catterall, WA 2019, 'Hippocampal deletion of NaV1.1 channels in mice causes thermal seizures and cognitive deficit characteristic of Dravet Syndrome', Proc Natl Acad Sci U S A, vol. 116, no. 33, pp. 16571-16576.

Steindel, F, Lerner, R, Haring, M, Ruehle, S, Marsicano, G, Lutz, B & Monory, K 2013, 'Neuron- type specific cannabinoid-mediated G protein signalling in mouse hippocampus', J Neurochem, vol. 124, no. 6, pp. 795-807.

Stella, N 2009, 'Endocannabinoid signaling in microglial cells', Neuropharmacology, vol. 56, pp. 244-253.

Stella, N 2010, 'Cannabinoid and cannabinoid-like receptors in microglia, astrocytes, and astrocytomas', Glia, vol. 58, no. 9, pp. 1017-30.

Stockings, E, Zagic, D, Campbell, G, Weier, M, Hall, WD, Nielsen, S, Herkes, GK, Farrell, M & Degenhardt, L 2018, 'Evidence for cannabis and cannabinoids for epilepsy: a systematic

169

review of controlled and observational evidence', J Neurol Neurosurg Psychiatry, vol. 89, no. 7, pp. 741-753.

Striano, P, Mancardi, MM, Biancheri, R, Madia, F, Gennaro, E, Paravidino, R, Beccaria, F, Capovilla, G, Dalla Bernardina, B, Darra, F, Elia, M, Giordano, L, Gobbi, G, Granata, T, Ragona, F, Guerrini, R, Marini, C, Mei, D, Longaretti, F, Romeo, A, Siri, L, Specchio, N, Vigevano, F, Striano, S, Tortora, F, Rossi, A, Minetti, C, Dravet, C, Gaggero, R & Zara, F 2007, 'Brain MRI findings in severe myoclonic epilepsy in infancy and genotype- phenotype correlations', Epilepsia, vol. 48, no. 6, pp. 1092-6.

Stuart, JM, Paris, JJ, Frye, C & Bradshaw, HB 2013, 'Brain levels of prostaglandins, endocannabinoids, and related lipids are affected by mating strategies', Int J Endocrinol, vol. 2013, pp. 436252.

Suemaru, K, Yoshikawa, M, Aso, H & Watanabe, M 2018, 'TRPV1 mediates the anticonvulsant effects of acetaminophen in mice', Epilepsy Res, vol. 145, pp. 153-159.

Sugaya, Y, Yamazaki, M, Uchigashima, M, Kobayashi, K, Watanabe, M, Sakimura, K & Kano, M 2016, 'Crucial Roles of the Endocannabinoid 2-Arachidonoylglycerol in the Suppression of Epileptic Seizures', Cell Rep, vol. 16, no. 5, pp. 1405-1415.

Sugiura, T, Kishimoto, S, Oka, S & Gokoh, M 2006, 'Biochemistry, pharmacology and physiology of 2-arachidonoylglycerol, an endogenous cannabinoid receptor ligand', Progress in Lipid Research, vol. 45, no. 5, pp. 405-446.

Sugiura, T, Yoshinaga, N & Waku, K 2001, 'Rapid generation of 2-arachidonoylglycerol, an endogenous cannabinoid receptor ligand, in rat brain after decapitation', Neurosci Lett, vol. 297, no. 3, pp. 175-8.

Suleymanova, EM, Shangaraeva, VA, Van Rijn, CM & Vinogradova, LV 2016, 'The cannabinoid receptor agonist WIN55.212 reduces consequences of status epilepticus in rats', Neuroscience, vol. 334, pp. 191-200.

Suls, A, Velizarova, R, Yordanova, I, Deprez, L, Van Dyck, T, Wauters, J, Guergueltcheva, V, Claes, LR, Kremensky, I, Jordanova, A & De Jonghe, P 2010, 'Four generations of epilepsy caused by an inherited microdeletion of the SCN1A gene', Neurology, vol. 75, no. 1, pp. 72-6.

Sun, Y & Dolmetsch, RE 2018, 'Investigating the Therapeutic Mechanism of Cannabidiol in a Human Induced Pluripotent Stem Cell (iPSC)-Based Model of Dravet Syndrome', Cold Spring Harb Symp Quant Biol, vol. 83, pp. 185-191.

Sun, Y, Pasca, SP, Portmann, T, Goold, C, Worringer, KA, Guan, W, Chan, KC, Gai, H, Vogt, D, Chen, YJ, Mao, R, Chan, K, Rubenstein, JL, Madison, DV, Hallmayer, J, Froehlich- Santino, WM, Bernstein, JA & Dolmetsch, RE 2016, 'A deleterious Nav1.1 mutation selectively impairs telencephalic inhibitory neurons derived from Dravet Syndrome patients', Elife, vol. 5.

170

Suraev, A, Lintzeris, N, Stuart, J, Kevin, RC, Blackburn, R, Richards, E, Arnold, JC, Ireland, C, Todd, L, Allsop, DJ & Mcgregor, IS 2018, 'Composition and Use of Cannabis Extracts for Childhood Epilepsy in the Australian Community', Sci Rep, vol. 8, no. 1, pp. 10154.

Suraev, AS, Todd, L, Bowen, MT, Allsop, DJ, Mcgregor, IS, Ireland, C & Lintzeris, N 2017, 'An Australian nationwide survey on medicinal cannabis use for epilepsy: History of antiepileptic drug treatment predicts medicinal cannabis use', Epilepsy Behav, vol. 70, no. Pt B, pp. 334-340.

Sustkova-Fiserova, M, Charalambous, C, Havlickova, T, Lapka, M, Jerabek, P, Puskina, N & Syslova, K 2017, 'Alterations in Rat Accumbens Endocannabinoid and GABA Content during Fentanyl Treatment: The Role of Ghrelin', Int J Mol Sci, vol. 18, no. 11.

Swiader, MJ, Kotowski, J & Luszczki, JJ 2014, 'Modulation of adenosinergic system and its application for the treatment of epilepsy', Pharmacol Rep, vol. 66, no. 3, pp. 335-42.

Swift, W, Wong, A, Li, KM, Arnold, JC & Mcgregor, IS 2013, 'Analysis of cannabis seizures in NSW, Australia: cannabis potency and cannabinoid profile', PLoS One, vol. 8, no. 7, pp. e70052.

Sylantyev, S, Jensen, TP, Ross, RA & Rusakov, DA 2013, 'Cannabinoid- and lysophosphatidylinositol-sensitive receptor GPR55 boosts neurotransmitter release at central synapses', Proc Natl Acad Sci U S A, vol. 110, no. 13, pp. 5193-8.

Tai, C, Abe, Y, Westenbroek, RE, Scheuer, T & Catterall, WA 2014, 'Impaired excitability of somatostatin- and parvalbumin-expressing cortical interneurons in a mouse model of Dravet syndrome', Proc Natl Acad Sci U S A, vol. 111, no. 30, pp. E3139-48.

Takahashi, KA & Castillo, PE 2006, 'The CB1 cannabinoid receptor mediates glutamatergic synaptic suppression in the hippocampus', Neuroscience, vol. 139, no. 3, pp. 795-802.

Takayama, R, Fujiwara, T, Shigematsu, H, Imai, K, Takahashi, Y, Yamakawa, K & Inoue, Y 2014, 'Long-term course of Dravet syndrome: a study from an epilepsy center in Japan', Epilepsia, vol. 55, no. 4, pp. 528-38.

Tanimura, A, Yamazaki, M, Hashimotodani, Y, Uchigashima, M, Kawata, S, Abe, M, Kita, Y, Hashimoto, K, Shimizu, T, Watanabe, M, Sakimura, K & Kano, M 2010, 'The Endocannabinoid 2-Arachidonoylglycerol Produced by Diacylglycerol Lipase α Mediates Retrograde Suppression of Synaptic Transmission', Neuron, vol. 65, no. 3, pp. 320-327.

Teran, FA, Kim, Y, Crotts, MS, Bravo, E, Emaus, KJ & Richerson, GB 2019, 'Time of Day and a Ketogenic Diet Influence Susceptibility to SUDEP in Scn1a (R1407X/+) Mice', Front Neurol, vol. 10, pp. 278.

Terrone, G, Pauletti, A, Salamone, A, Rizzi, M, Villa, BR, Porcu, L, Sheehan, MJ, Guilmette, E, Butler, CR, Piro, JR, Samad, TA & Vezzani, A 2018, 'Inhibition of monoacylglycerol lipase terminates diazepam-resistant status epilepticus in mice and its effects are potentiated by a ketogenic diet', Epilepsia, vol. 59, no. 1, pp. 79-91.

171

Thomas, G, Betters, JL, Lord, CC, Brown, AL, Marshall, S, Ferguson, D, Sawyer, J, Davis, MA, Melchior, JT, Blume, LC, Howlett, AC, Ivanova, PT, Milne, SB, Myers, DS, Mrak, I, Leber, V, Heier, C, Taschler, U, Blankman, JL, Cravatt, BF, Lee, RG, Crooke, RM, Graham, MJ, Zimmermann, R, Brown, HA & Brown, JM 2013, 'The serine hydrolase ABHD6 Is a critical regulator of the metabolic syndrome', Cell Rep, vol. 5, no. 2, pp. 508- 20.

Thomas, RH & Berkovic, SF 2014, 'The hidden genetics of epilepsy-a clinically important new paradigm', Nat Rev Neurol, vol. 10, no. 5, pp. 283-92.

Tsai, MS, Lee, ML, Chang, CY, Fan, HH, Yu, IS, Chen, YT, You, JY, Chen, CY, Chang, FC, Hsiao, JH, Khorkova, O, Liou, HH, Yanagawa, Y, Lee, LJ & Lin, SW 2015, 'Functional and structural deficits of the dentate gyrus network coincide with emerging spontaneous seizures in an Scn1a mutant Dravet Syndrome model during development', Neurobiol Dis, vol. 77, pp. 35-48.

Tsutsumi, T, Kobayashi, T, Ueda, H, Yamauchi, E, Watanabe, S & Okuyama, H 1994, 'Lysophosphoinositide-specific phospholipase C in rat brain synaptic plasma membranes', Neurochem Res, vol. 19, no. 4, pp. 399-406.

Turner, SJ, Brown, A, Arpone, M, Anderson, V, Morgan, AT & Scheffer, IE 2017, 'Dysarthria and broader motor speech deficits in Dravet syndrome', Neurology, vol. 88, no. 8, pp. 743-749.

Ueda, H, Kobayashi, T, Kishimoto, M, Tsutsumi, T & Okuyama, H 1993, 'A possible pathway of phosphoinositide metabolism through EDTA-insensitive phospholipase A1 followed by lysophosphoinositide-specific phospholipase C in rat brain', J Neurochem, vol. 61, no. 5, pp. 1874-81.

Valenti, M, Vigano, D, Casico, MG, Rubino, T, Steardo, L, Parolaro, D & Di Marzo, V 2004, 'Differential diurnal variations of anandamide and 2-arachidonoyl-glycerol levels in rat brain', Cell Mol Life Sci, vol. 61, no. 7-8, pp. 945-50.

Van Niekerk, G, Mabin, T & Engelbrecht, AM 2019, 'Anti-inflammatory mechanisms of cannabinoids: an immunometabolic perspective', Inflammopharmacology, vol. 27, no. 1, pp. 39-46.

Viader, A, Ogasawara, D, Joslyn, CM, Sanchez-Alavez, M, Mori, S, Nguyen, W, Conti, B & Cravatt, BF 2016, 'A chemical proteomic atlas of brain serine hydrolases identifies cell type-specific pathways regulating neuroinflammation', Elife, vol. 5, pp. e12345.

Vilela, LR, Medeiros, DC, Rezende, GH, De Oliveira, AC, Moraes, MF & Moreira, FA 2013, 'Effects of cannabinoids and endocannabinoid hydrolysis inhibition on pentylenetetrazole- induced seizure and electroencephalographic activity in rats', Epilepsy Res, vol. 104, no. 3, pp. 195-202.

Villeneuve, N, Laguitton, V, Viellard, M, Lepine, A, Chabrol, B, Dravet, C & Milh, M 2014, 'Cognitive and adaptive evaluation of 21 consecutive patients with Dravet syndrome', Epilepsy Behav, vol. 31, pp. 143-8.

172

Von Ruden, EL, Bogdanovic, RM, Wotjak, CT & Potschka, H 2015a, 'Inhibition of monoacylglycerol lipase mediates a cannabinoid 1-receptor dependent delay of kindling progression in mice', Neurobiol Dis, vol. 77, pp. 238-45.

Von Ruden, EL, Jafari, M, Bogdanovic, RM, Wotjak, CT & Potschka, H 2015b, 'Analysis in conditional cannabinoid 1 receptor-knockout mice reveals neuronal subpopulation-specific effects on epileptogenesis in the kindling paradigm', Neurobiol Dis, vol. 73, pp. 334-47.

Wahlestedt, C 2013, 'Targeting long non-coding RNA to therapeutically upregulate gene expression', Nat Rev Drug Discov, vol. 12, no. 6, pp. 433-46.

Walker, JM, Huang, SM, Strangman, NM, Tsou, K & Sanudo-Pena, MC 1999, 'Pain modulation by release of the endogenous cannabinoid anandamide', Proc Natl Acad Sci U S A, vol. 96, no. 21, pp. 12198-203.

Wallace, MJ, Blair, RE, Falenski, KW, Martin, BR & Delorenzo, RJ 2003, 'The endogenous cannabinoid system regulates seizure frequency and duration in a model of temporal lobe epilepsy', J Pharmacol Exp Ther, vol. 307, no. 1, pp. 129-37.

Wallace, MJ, Wiley, JL, Martin, BR & Delorenzo, RJ 2001, 'Assessment of the role of CB1 receptors in cannabinoid anticonvulsant effects', Eur J Pharmacol, vol. 428, no. 1, pp. 51- 7.

Walter, L, Franklin, A, Witting, A, Wade, C, Xie, Y, Kunos, G, Mackie, K & Stella, N 2003, 'Nonpsychotropic cannabinoid receptors regulate microglial cell migration', J Neurosci, vol. 23, no. 4, pp. 1398-405.

Wang, J, Lin, ZJ, Liu, L, Xu, HQ, Shi, YW, Yi, YH, He, N & Liao, WP 2017a, 'Epilepsy-associated genes', Seizure, vol. 44, pp. 11-20.

Wang, J, Ou, SW & Wang, YJ 2017, 'Distribution and function of voltage-gated sodium channels in the nervous system', Channels (Austin), vol. 11, no. 6, pp. 534-554.

Wang, Y, Gu, N, Duan, T, Kesner, P, Blaskovits, F, Liu, J, Lu, Y, Tong, L, Gao, F, Harris, C, Mackie, K, Li, J, Tan, Q, Hill, MN, Yuan, Z & Zhang, X 2017b, 'Monoacylglycerol lipase inhibitors produce pro- or antidepressant responses via hippocampal CA1 GABAergic synapses', Mol Psychiatry, vol. 22, no. 2, pp. 215-226.

Wefers, B, Bashir, S, Rossius, J, Wurst, W & Kuhn, R 2017, 'Gene editing in mouse zygotes using the CRISPR/Cas9 system', Methods, vol. 121-122, pp. 55-67.

Wepy, JA, Galligan, JJ, Kingsley, PJ, Xu, S, Goodman, MC, Tallman, KA, Rouzer, CA & Marnett, LJ 2019, 'Lysophospholipases cooperate to mediate lipid homeostasis and lysophospholipid signaling', J Lipid Res, vol. 60, no. 2, pp. 360-374.

Wettschureck, N, Van Der Stelt, M, Tsubokawa, H, Krestel, H, Moers, A, Petrosino, S, Schutz, G, Di Marzo, V & Offermanns, S 2006, 'Forebrain-specific inactivation of Gq/G11 family G

173

proteins results in age-dependent epilepsy and impaired endocannabinoid formation', Mol Cell Biol, vol. 26, no. 15, pp. 5888-94.

Whissell, PD, Cajanding, JD, Fogel, N & Kim, JC 2015, 'Comparative density of CCK- and PV- GABA cells within the cortex and hippocampus', Frontiers in neuroanatomy, vol. 9, pp. 124-124.

White, HS, Brown, SD, Woodhead, JH, Skeen, GA & Wolf, HH 2000, 'Topiramate modulates GABA-evoked currents in murine cortical neurons by a nonbenzodiazepine mechanism', Epilepsia, vol. 41 Suppl 1, pp. S17-20.

Whyte, LS, Ryberg, E, Sims, NA, Ridge, SA, Mackie, K, Greasley, PJ, Ross, RA & Rogers, MJ 2009, 'The putative cannabinoid receptor GPR55 affects osteoclast function in vitro and bone mass in vivo', Proc Natl Acad Sci U S A, vol. 106, no. 38, pp. 16511-6.

Wilkerson, JL, Ghosh, S, Bagdas, D, Mason, BL, Crowe, MS, Hsu, KL, Wise, LE, Kinsey, SG, Damaj, MI, Cravatt, BF & Lichtman, AH 2016, 'Diacylglycerol lipase beta inhibition reverses nociceptive behaviour in mouse models of inflammatory and neuropathic pain', Br J Pharmacol, vol. 173, no. 10, pp. 1678-92.

Williams, AD, Kalume, F, Westenbroek, RE & Catterall, WA 2019, 'A More Efficient Conditional Mouse Model of Dravet Syndrome: Implications for Epigenetic Selection and Sex- Dependent Behaviors', J Neurosci Methods, pp. 108315.

Wilson, CD, Tai, S, Ewing, L, Crane, J, Lockhart, T, Fujiwara, R, Radominska-Pandya, A & Fantegrossi, WE 2019, 'Convulsant Effects of Abused Synthetic Cannabinoids JWH-018 and 5F-AB-PINACA Are Mediated by Agonist Actions at CB1 Receptors in Mice', J Pharmacol Exp Ther, vol. 368, no. 2, pp. 146-156.

Wirrell, EC 2016, 'Treatment of Dravet Syndrome', Can J Neurol Sci, vol. 43 Suppl 3, pp. S13-8.

Wolff, M, Casse-Perrot, C & Dravet, C 2006, 'Severe myoclonic epilepsy of infants (Dravet syndrome): natural history and neuropsychological findings', Epilepsia, vol. 47 Suppl 2, pp. 45-8.

Wong, GT 2002, 'Speed congenics: applications for transgenic and knock-out mouse strains', Neuropeptides, vol. 36, no. 2-3, pp. 230-6.

Wootten, D, Christopoulos, A & Sexton, PM 2013, 'Emerging paradigms in GPCR allostery: implications for drug discovery', Nat Rev Drug Discov, vol. 12, no. 8, pp. 630-44.

Wu, CS, Chen, H, Sun, H, Zhu, J, Jew, CP, Wager-Miller, J, Straiker, A, Spencer, C, Bradshaw, H, Mackie, K & Lu, HC 2013, 'GPR55, a G-protein coupled receptor for lysophosphatidylinositol, plays a role in motor coordination', PLoS One, vol. 8, no. 4, pp. e60314.

174

Wu, YW, Sullivan, J, Mcdaniel, SS, Meisler, MH, Walsh, EM, Li, SX & Kuzniewicz, MW 2015, 'Incidence of Dravet Syndrome in a US Population', Pediatrics, vol. 136, no. 5, pp. e1310- 5.

Xie, Y & Meier, KE 2004, 'Lysophospholipase D and its role in LPA production', Cell Signal, vol. 16, no. 9, pp. 975-81.

Yalnizoglu, D, Ozgul, RK, Oguz, KK, Ozer, B, Yucel-Yilmaz, D, Gurbuz, B, Serdaroglu, E, Erol, I, Topcu, M & Dursun, A 2019, 'Expanding the phenotype of phospholipid remodelling disease due to MBOAT7 gene defect', J Inherit Metab Dis, vol. 42, no. 2, pp. 381-388.

Yu, FH, Mantegazza, M, Westenbroek, RE, Robbins, CA, Kalume, F, Burton, KA, Spain, WJ, Mcknight, GS, Scheuer, T & Catterall, WA 2006, 'Reduced sodium current in GABAergic interneurons in a mouse model of severe myoclonic epilepsy in infancy', Nat Neurosci, vol. 9, no. 9, pp. 1142-9.

Zareie, P, Sadegh, M, Palizvan, MR & Moradi-Chameh, H 2018, 'Anticonvulsive effects of endocannabinoids; an investigation to determine the role of regulatory components of endocannabinoid metabolism in the Pentylenetetrazol induced tonic- clonic seizures', Metab Brain Dis, vol. 33, no. 3, pp. 939-948.

Zestos, AG & Kennedy, RT 2017, 'Microdialysis Coupled with LC-MS/MS for In Vivo Neurochemical Monitoring', Aaps j, vol. 19, no. 5, pp. 1284-1293.

Zhang, Z, Wang, Z, Zhang, B & Liu, Y 2018, 'Downregulation of microRNA155 by preoperative administration of valproic acid prevents postoperative seizures by upregulating SCN1A', Mol Med Rep, vol. 17, no. 1, pp. 1375-1381.

Zhao, X, Liao, Y, Morgan, S, Mathur, R, Feustel, P, Mazurkiewicz, J, Qian, J, Chang, J, Mathern, GW, Adamo, MA, Ritaccio, AL, Gruenthal, M, Zhu, X & Huang, Y 2018, 'Noninflammatory Changes of Microglia Are Sufficient to Cause Epilepsy', Cell Rep, vol. 22, no. 8, pp. 2080-2093.

Zou, S & Kumar, U 2018, 'Cannabinoid Receptors and the Endocannabinoid System: Signaling and Function in the Central Nervous System', Int J Mol Sci, vol. 19, no. 3.

Zygmunt, PM, Petersson, J, Andersson, DA, Chuang, H, Sorgard, M, Di Marzo, V, Julius, D & Hogestatt, ED 1999, 'Vanilloid receptors on sensory nerves mediate the vasodilator action of anandamide', Nature, vol. 400, no. 6743, pp. 452-7.