저작자표시 - 비영리 - 변경금지 2.0 대한민국 이용자는아래의조건을따르는경우에한하여자유롭게 이저작물을복제, 배포, 전송, 전시, 공연및방송할수있습니다. 다음과같은조건을따라야합니다 : 저작자표시. 귀하는원저작자를표시하여야합니다. 비영리. 귀하는이저작물을영리목적으로이용할

Transcription

1 저작자표시 - 비영리 - 변경금지 2.0 대한민국 이용자는아래의조건을따르는경우에한하여자유롭게 이저작물을복제, 배포, 전송, 전시, 공연및방송할수있습니다. 다음과같은조건을따라야합니다 : 저작자표시. 귀하는원저작자를표시하여야합니다. 비영리. 귀하는이저작물을영리목적으로이용할수없습니다. 변경금지. 귀하는이저작물을개작, 변형또는가공할수없습니다. 귀하는, 이저작물의재이용이나배포의경우, 이저작물에적용된이용허락조건을명확하게나타내어야합니다. 저작권자로부터별도의허가를받으면이러한조건들은적용되지않습니다. 저작권법에따른이용자의권리는위의내용에의하여영향을받지않습니다. 이것은이용허락규약 (Legal Code) 을이해하기쉽게요약한것입니다. Disclaimer

2 Shank2 Snf7-3, Behavioral and molecular studies on the neurodevelopmental disorders using Shank2 and Snf7-3 knock out mice

3

4

5

6

7

8 ABSTRACT Behavioral and molecular studies on the neurodevelopmental disorders using Shank2 and Snf7-3 knock out mice Hyopil Kim School of Biological Sciences The Graduate School Seoul National University Neurodevelopmental disorders are defined as a group of conditions caused by developmental deficits with onset in the relatively early period of lifetime, which produce cognitive and behavioral abnormalities including learning, memory, locomotion, attention and social functions according to the fifth edition of Diagnostic and Statistical Manual of Mental Disorders (DSM-5). They are classified into disorders such as intellectual disabilities, autism spectrum disorder (ASD), attention deficit hyperactivity disorder (ADHD), obsessive compulsive disorders and tic disorders by specific behavioral and mental symptoms, and in many cases patients of neurodevelopmental disorders share some symptoms. However, the causes and physiological mechanisms of each disorder is not well understood yet, and even in one category of disorder, it seems that various distinct mechanisms are implicated in it. Among the neurodevelopmental disorders, ASD is characterized by deficits of sociability, social communication, repetitive behaviors and restricted interests. Because of its high prevalence, genetic analyses of ASD patients have been done worldwide and these studies indicate that many synaptic genes including SH3 and multiple ankyrin repeat domains protein 2 (SHANK2) are involved in ASD. Concomitantly with these studies, two Shank2 knock out mice line with deletion of exon 6 and 7 (Shank2 KO e67) and with deletion of exon 7 (Shank2 KO e7) were characterized 1

9 as autistic behaviors. However, when I analyzed the transcriptome of the two lines, expression of Gabra2 gene, encoding -aminobutyric acid (GABA) A receptor subunit α2 (GABAAα2) and inhibitory neurotransmission were reduced only in Shank2 KO e67. The excitatory neurotransmission was normal, so the excitatory and inhibitory balance (E/I balance) was impaired. Restoring this inhibitory neurotransmission could rescue spatial memory deficits of Shank2 KO e67 mice, while it did not affect the spatial memory deficits of Shank2 KO e7 mice. In addition to the work, I investigated if the endo-lysosomal pathway is involved in the neurodevelopmental processes. Interestingly my colleague found that when the expression of Snf7-3 protein, a component of endosomal sorting complexes required for transport III (ESCRT- III), was reduced in developing neuron culture, the dendrite outgrowth was exaggerated. To examine this phenotype is involved with the symptoms of neurodevelopmental disorders, I generated conventional and conditional KO mice of Snf7-3, and the mice displayed hyperactivity which is the main symptom of ADHD. Furthermore, in the conditional KO mice, object location memory was also impaired. In addition, since the E/I balance of neurotransmission is often disturbed in the model mice of neurodevelopmental disorders, I assessed this possibility and found that the frequency of spontaneous excitatory neurotransmission was elevated in the KO mice of hippocampus.... Keyword: Neurodevelopmental disorders, ASD, ADHD, Shank2, Snf7-3, E/I balance Student number:

10 CONTENTS Abstract... i List of Figures... v Chapter I. Introduction Background... 8 Purpose of study Chapter II. Distinct mechanisms accounted for behavioral deficits of two Shank2 KO lines Introduction Experimental Procedures Results Discussion Chapter III. Role of Snf7-3 in neurodevelopmental disorders regulating the dendritic developmental processes Introduction Experimental Procedures Results Discussion

11 Chapter IV. Conclusion...78 Acknowledgements...80 References

12 LIST OF FIGURES Figure 1. Representative images of three-chamber test and Morris-water maze 12 Figure 2. Strategies to delete Shank2 by targeting exon 6 and 7 or only exon Figure 3. Confirmation of Shank2 expression in Shank2 KO e67 and e7 mice in RNA sequencing Figure 4. Differential gene expression profiles of Shank2 e67 or e7 KO lines (WT vs KO) samples 29 Figure 5. Decreased Gabra2 expression in mrna and protein level. 35 Figure 6.Decreased inhibitory GABAergic function and rescue of it by L838,417 in Shank2 e67 KO mice.. 37 Figure 7. Recording of spontaneous mepsc or mipsc, and paired-pulse facilitation ratios (PPR).. 39 Figure 8. Effects of L838,417 treatment on impaired social behaviors of Shank2 KO e67 mice. 41 Figure 9.Effects of L838,417 treatment on spatial memory deficits of Shank2 KO e67 mice. 43 Figure 10. Effects of L838,417 treatment on spatial memory deficits of Shank2 KO e7 mice.. 45 Figure 11. Effects of L838,417 on anxiety like behavior and locomotor activity of Shank2 KO e67 mice.. 47 Figure 12. Differential role of Snf7-2 and Snf7-3 in dendritic developments. 54 Figure 13. Generation of conventional and conditional Snf7-3 KO lines Figure 14. Reduced expression of Snf7-3 in generated KO lines Figure 15. Locomotion, anxiety and social behaviors of 8 to 24 weeks old Snf7-3 KO mice Figure 16. Hippocampus dependent spatial and contextual fear memories of 8 to 24 weeks old Snf7-3 KO mice.. 67 Figure 17. Locomotion, anxiety and social behaviors of more than 25 weeks old Snf7-3 KO 5

13 mice. 69 Figure 18. Hippocampus dependent spatial and contextual fear memories of more than 25 weeks old Snf7-3 KO mice 70 Figure 19. Dendrite growth and complexity in the hippocampus and mpfc of conditional Snf7-3 KO mice 72 Figure 20. Spontaneous mepsc frequency and amplitude in the hippocampus and mpfc of Snf7-3 conditional KO mice 73 Table 1.Differentially expressed genes in Shank2 KO e67 mice Table 2. Differentially expressed genes in Shank2 KO e7 mice Table 3. Gene ontology (GO) analysis on DEGs of Shank2 KO e67 mice Table 4. GO analysis on DEGs of Shank2 e7 mice

14 CHAPTER I INTRODUCTION 7

15 BACKGROUND Causes of neurodevelopmental disorders Neurodevelopmental disorders are defined as a group of pathological conditions including various deficits in intellectual abilities, attention, locomotion and social functions with onset in the relatively early period of lifetime. They are not convergent to one disorder and even in same disorder, the symptoms and causes are quite different. Consistently with the complex property of neurodevelopmental disorders, various environmental and genetic factors have been identified implicated in the disorders. Environmental factors include prenatal exposure to valproic acids, gut microbiota and maternal obesity, increasing the incidence of neurodevelopmental disorders, especially ASD (Buffington et al., 2016; Li et al., 2017a; Schneider and Przewłocki, 2005). Genetic studies have identified hundreds genes related with a lot of biological processes, such as synaptic scaffolding proteins, neurotransmitter receptors, adherent molecules, cytoskeletal proteins and signaling molecules. Although it is difficult to find out general mechanism for neurodevelopmental disorders, it seems that the factors cause developmental delays and abnormal brain connectivity. For example, lots of genes are found to be interact with signaling proteins of Rho GTPases pathways which is critical for synapse development (Govek et al., 2005). In addition, mutations in genes involved in regulation of mrna translation of synaptic proteins are also found to be related with neurodevelopmental disorders, including fragile X mental retardation protein (FMRP), tuberous sclerosis protein 1/2 and eukaryotic translation initiation factor 4E (eif4e) (Gkogkas et al., 2013; Schneider and Przewłocki, 2005; The Dutch-Belgian Fragile et al., 1994). Mutations in the genes cause dysregulation of protein synthesis, altering synaptic density and neurotransmission properties. 8

16 E/I balance Fine tuning of activation or inhibition of a specific population of neurons is important for the adequate development and function of brain, and a neuron can regulate the activity of another neuron by excitatory or inhibitory neurotransmission. The most well-known excitatory and inhibitory neurotransmitter is glutamate and GABA respectively. There are two major ionotropic glutamate receptors, -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (Gkogkas et al.) receptor and N-methyl-D-aspartate (NMDA) receptor, that can activate a neuron after binding glutamate by generating positive sodium or calcium ion influx. Similarly, GABA can inhibit a neuron with GABA receptors (GABA-R) which are divided into GABAA and GABAB receptors, by generating negative chloride influx. Thus, if there are alterations in the neurotransmitter release and the number or conductance of the receptors, the E/I balance can be disturbed and this condition possibly result in abnormal pathophysiological conditions. Indeed, many mouse models of neurodevelopmental disorders such as Shank2 KO and Cntnap2 KO mice have disturbed E/I balance, and social and cognitive functions can be regulated by manipulating E/I balance in specific brain areas (Lee et al., 2017; Peñagarikano et al., 2011; Schmeisser et al., 2012; Won et al., 2012; Yizhar et al., 2011). Hyperactivity and neurodevelopmental disorders Hyperactivity is a main symptom of ADHD which is a major neurodevelopmental disorder. The term of hyperactivity means exaggerated motions and locomotion in both vertical and horizontal directions. Although the hyperactivity itself is not a main diagnosis criteria of the other neurodevelopmental disorders, in many cases, patients and animal models of neurodevelopmental disorders, including autism display the hyperactivity. For example, autistic mouse models such as Shank2 KO, Cntnap2 KO and Scn1a +/-, are found to be hyperactive (Han et al., 2012; Peñagarikano et al., 2011; Schmeisser et al., 2012; Won et al., 2012). In addition, patients of autism or tic disorders are characterized by repetitive behaviors which are regulated by the cortico-striatal circuit that also regulates locomotion (Kim et al., 2016). 9

17 RNA sequencing Finding molecular targets of specific biological processes is very important to understand the molecular mechanisms of the processes. There are many methods to test if a candidate molecule is a really important molecule, but it is not easy to set the candidate molecules. Moreover, nowadays, a pathological condition is thought to be resulted from dynamic alterations in a molecular profile rather than from alteration in one molecule. For the reasons, RNA sequencing is a widely used technique that provides a RNA transcriptome profile, and by comparing the transcriptome, we can get a clue for actual important molecules. For that purpose, mrna is extracted from a sample and fragmented, and copied into cdna. This cdna library is sequenced by short-read sequencing. Each reads are aligned to a reference genome and quantitative analysis about gene expression can be made from the data. Three-chamber test Three-chamber test is a behavioral test which addresses sociability and social recognition (Figure 1A). The test has been widely used to investigate social behaviors of animal models since when it was developed by Jacqueline N. Crawley (Moy et al., 2004). The test is composed of two phases. At the first phase, a mouse is placed in a center chamber of sequential three chambers. In the two chambers of outside, a social target, stranger mouse, or a non-social target, such as a little block, is presented in a wired cup respectively. A subject mouse spends time exploring the targets, and normal mice usually display much higher exploration to the social target. If a mouse has a problem in sociability, it would not spend much time with the stranger mouse. At the second phase, the non-social target is replaced with new stranger mouse and in this phase, normal mice spend more time with a novel stranger, because they have instinct to explore novel social target. If a mouse has poor social recognition ability, the mouse cannot distinguish the novel stranger from the prior one. Autistic mice usually display abnormal sociability or social recognition in the test (Han et al., 2012; Peñagarikano et al., 2011; Schmeisser et al., 2012; Won et al., 2012). Morris-water maze Morris-water maze is developed by Richard G. Morris in 1981 as a test for spatial memory, and 10

18 many reports have shown that the spatial memory is dependent on proper hippocampal function in the task (Morris et al., 1982; Vorhees and Williams, 2006). In Morris-water maze, a platform (mouse cannot see the platform because the plat form is hidden under the surface of opaque water) is prepared in a water pool and specific spatial cues are presented around the pool. Then a mouse is placed in the pool and the mouse try to escape from the pool, since mice do not like being soaked in water and mice feel more comfortable on the hard platform. In the training session, mice were guided to find the platform so they learn the location of the platform based on the spatial cues. After this training sessions, a probe test is performed in which mouse is exposed to the pool again without platform, and the time spent near the platform position is assessed (Figure 1B). 11

19 Figure 1. Representative images of three-chamber test and Morris-water maze (A) Representative images of three-chamber test (B) Representative images of Morris-water maze 12

20 PURPOSE OF THIS STUDY 1. Distinct mechanisms accounted for behavioral deficits of two Shank2 KO lines Despite to the high prevalence of ASD, there is no general effective treatment for ASD and lots of researchers are still trying to understand the mechanisms of ASD and develop effective treatments for the disorder. However, the heterogeneity of ASD makes it difficult to study and treat the disorder. For the purpose of studying the neuronal and molecular mechanisms of ASD, genetic analyses of ASD patients have been done globally. Interestingly, these studies indicated that many synaptic genes are involved with ASD including SCN1A, CNTNAP2 and NRXN1A (Kim et al., 2008; Strauss et al., 2006; Weiss et al., 2003). Among the synaptic genes involved in ASD, genes encoding SH3-Ankirin and proline rich synaptic associated protein, Shank, are strongly associated with neuropsychiatric disorders (Guilmatre et al., 2014). Furthermore, it was found that copy number variations in SHANK2 gene is involved with ASD and mental retardation (Berkel et al., 2010). Based on this, a group of researchers, investigated if Shank2 deleted mice display autistic behaviors and if so, what the underlining mechanism of the behaviors is. For that purpose, our laboratory investigated autistic behaviors of Shank2 KO e67 and showed that the KO mice have social impairment, repetitive jumping behavior and intellectual disabilities. Furthermore, the neurotransmission through N-methyl-D-aspartate receptors (NMDA-R) is weakened in the mouse and enhancing the function of NMDA-R ameliorated the social deficit of the Shank2 KO e67 mice (Won et al., 2012). Consistently with the findings, another group reported that another Shank2 KO mouse, Shank2 KO e7 mouse, also displays autistic like phenotypes including social deficits. However, interestingly, the NMDA-R function was not weakened in the KO mice, rather enhanced (Schmeisser et al., 2012). Thus, I hypothesized that their behaviors may be resulted from 13

21 different molecular mechanisms even though the behavioral patterns are quite similar. Researchers made ASD model mice based on the candidates, and they studied physiological and molecular characteristics of the models. However, the phenotypes of the models were not convergent even if they have similar deficits related with ASD (Kim et al., 2016). Since the causes of ASD are not well known, and the symptoms of ASD are quite variable, I do not think that one golden solution exists for ASD. Specific molecular mechanisms underlying a specific behavioral deficit should be studied and treatment also be done according to the mechanisms. Moreover, it is possible that different molecular mechanisms underpin same behavioral deficits. Therefore, to understand the mechanisms of ASD we need non-biased studies, and should know detailed causes of a specific phenotype to treat ASD efficiently. The goal of this study is to compare molecular and physiological profiles of the two Shank2 KO lines, and find out which mechanism is accounted for each behavioral deficit of the lines. In addition, the final goal is to rescue the deficits based on the mechanism. 2. Role of Snf7-3 in regulating the neurodevelopmental processes Snf7-3 is a protein which is a member of Snf7 family, and also known as chromatin-modifying protein/charged multivesicular body protein 4C (Chmp4C). Snf7 proteins are components of endosomal sorting complex required for transport III (ESCRT-III), and the complex is involved in the endo-lysosomal pathway, regulating multivesicular body (MVB) formation (Schuh and Audhya, 2014). ESCRT-III is composed of Vps20, Snf7, Vps24 and Vps2 and they form a spiral structure in membrane of endosome or plasma membrane, so influencing MVB formation, cytokinesis and virus budding (Henne et al., 2011). In addition to that, a drosophila homolog of Snf7, Shrub, is involved in the pruning of dendrites in developing neurons, regulating the morphology of the neurons (Loncle et al., 2015; Sweeney et al., 2006). However, it is not well-known that if Snf7 or ESCRT-III has a role in neurodevelopmental processes in mouse. Mouse has two paralogues of Snf7, Snf7-2 and Snf7-3, and my colleague showed that in primary neuron culture, dendritic development was 14

22 decreased when she reduced the expression of Snf7-2 (Lee et al., 2007). Interestingly, according to her, reducing the expression of Snf7-3 caused the opposite phenomenon increasing the dendritic development. Furthermore, Snf7-3 was most highly expressed on DIV6 in mouse primary culture, when the development of culture neurons is vigorous, including dendrite outgrowth and maturation. Thus I became interested in the protein and thought that it might be implicated in neurodevelopmental disorders. Supporting this, a genetic study of human patients indicated that (SNPs) in CHMP7, a homologue of SNF7, may be involved in ADHD (Neale et al., 2010). Therefore, I generated Snf7-3 KO mice and addressed if the depletion of Snf7-3 would affect the regulation of dendrite complexity in vivo. In addition, I examined if the Snf7-3 KO mice display some physiological and behavioral features of neurodevelopmental disorders. 15

23 CHAPTER II Distinct mechanisms accounted for behavioral deficits of two Shank2 KO lines 16

24 INTRODUCTION ASD is diagnosed based on social deficits, communication deficits, stereotypic behaviors and restricted interests according to DSM-5. In addition to the primary symptoms, lots of ASD patients have intellectual disabilities and 25 75% of them have learning and memory problems (O'Brien and Pearson, 2004). Furthermore, consistently with the term of Spectrum, the patients show various range of cognitive deficits and the severity of the symptom is also divergent. For instance, some patients are hyperactive, while others are hypoactive. Some patients show microencephaly, while others show megalencephaly (Fombonne et al., 1999). For the last decades, the number of ASD patients have been increased dramatically and recent surveys said that the prevalence of ASD is 1 in 42 children in United States and 1 in 38 children in South Korea (Kim et al., 2011; Zablotsky et al., 2015). As the most frequently occurring neurodevelopmental disorder, it demands tremendous social payment for treating the patients. However, despite the high prevalence of ASD, there is no effective treatment for the disorder. Just risperidone and aripiprazole are approved by FDA for treating ASD patient to reduce irritability, while they are not that effective for main symptoms of ASD. ASD is thought to be caused by both genetic and environmental factors. Nowadays, some environmental factors implicated in ASD have been uncovered, such as administration of valproic acid during pregnancy or alteration in the balance of the gut microbiome (Christensen et al., 2013; Li et al., 2017a). On the other hand, ASD is highly heritable disorder and a lot of studies have discovered ASD associated genes by genome analyses. About 20% of ASD cases are involved with genetic and epigenetic variations including alterations in chromosomal structures, copy number variations and mutations in coding sequences (Leblond et al., 2014). According to Simons Foundation Autism Research Initiative, SFARI, over a thousand of ASD risk genes are scored and estimated. Among them, genes encoding synaptic proteins are one of the most highlighted gene group, including ion channels, adhesion molecules and scaffolding proteins, such as SCN1A, CNTNAP2 17

25 and NRXN1A (Kim et al., 2008; Strauss et al., 2006; Weiss et al., 2003). Based on the genetic studies of human patients, dozens of animal models were generated by mimic the mutations of the candidate genes, and some animals actually displayed social deficits and repetitive behaviors which are main symptoms of ASD (Etherton et al., 2009; Han et al., 2012; Peñagarikano et al., 2011). SHANK encoding SH3 and multiple ankyrin repeat domains protein, Shank, is one of the genes which is implicated in ASD. Shank protein is a major synaptic scaffolding protein in excitatory synaptic spines, make a network between glutamate receptors such as NMDA-R and mglur, and actin structures. Mammalian SHANK family of genes is composed of SHANK1, SHANK2 and SHANK3. All the SHANK genes are dynamically involved in neuropsychiatric diseases and ASDs. About 1~2 % of patients with ASD or intellectual disabilities patients have mutations in the SHANK genes (Leblond et al., 2014). Especially, it was found that copy number variations in SHANK2 gene is involved with ASDs and mental retardation (Berkel et al., 2010). Based on this, a group of researchers including our lab, investigated if Shank2 deleted mice display autistic behaviors and if so, what the underlining mechanism of the behaviors is. For that purpose, we investigated autistic behaviors of Shank2 KO e6-7 and showed that the KO mice have social impairment, repetitive jumping behavior and intellectual disabilities. Furthermore, the neurotransmission through N-methyl-D-aspartate receptors (NMDA-R) is weakened in the mouse and enhancing the function of NMDA-R ameliorated the social deficit of the Shank2 KO e6-7 mice (Won et al., 2012). Consistently with the findings, another group reported that another Shank2 KO mouse, Shank2 KO e7 mouse, also displays autistic like phenotypes including social deficits. However, the NMDA-R function was not weakened in the KO mice, rather enhanced (Schmeisser et al., 2012). This implies that the two KO lines may have different molecular mechanisms for their behavioral and physiological symptoms, and different treatment will work. Studying this differences would help to understand various mechanisms for autistic behaviors. To compare the molecular differences of the two KO lines I conducted unbiased RNA sequencing in the hippocampus and analyzed the transcriptomes of them. Interestingly, I found that the expression of Gabra2 gene which encoding GABAAα2, was reduced in Shank2 e67 KO 18

26 mice while it was intact in Shank2 KO e7 mice. GABAA receptors are ligand-gated chloride channels and consist of two α subunits, two β subunits and one γ subunit. The subcellular localization is dependent on their composition and GABAA receptors containing α1 3 are mainly synaptic, while receptors containing α4 6 subunits are mainly extrasynaptic (Belelli et al., 2009). Especially GABAAα2 is important for regulating neuronal activities and implicated in emotional processes such as anxiety and depression, and the CNS development (Dixon et al., 2008; Gonzalez-Nunez, 2015). In relation with ASD, alterations of GABAergic signaling or disturbed E/I balance is commonly found in the neurodevelopmental disorders (Braat and Kooy, 2015; Han et al., 2014; Han et al., 2012; Robertson et al., 2016). Several mouse models harboring ASD-associated mutations display altered ratio of E/I ratio (Cui et al., 2008; Gkogkas et al., 2013; Han et al., 2012; Houbaert et al., 2013; Lee et al., 2014; Tabuchi et al., 2007). In addition, copy-number variations (CNVs) in chromosome 15q11 13 which includes genes encoding GABAA receptor subunits, have been found in patients of a type of ASD, Prader-Willi/Angelman syndrome. (Hogart et al., 2007; Scoles et al., 2011). Furthermore, it was found that GABA receptor subunits are less expressed in ASD patients by post-mortem studies (Fatemi et al., 2014; Oblak et al., 2011). Thus I hypothesized that this difference might be accounted for the differences of the Shank2 KO lines. Consistently with the reduced expression of Gabra2, GABA-R mediated synaptic transmission was reduced in e6-7 KO, while the excitatory AMPA receptor (AMPA-R) mediated synaptic transmission was normal, so alternating the E/I ratio. Then I applied an agonist of GABAA receptor including the alpha 2 subunit, L838,417, to see if it can rescue some behavioral deficits of the Shank2 e6-7 KO mice. Treatment of L838,417 did not affect the social impairment of the two Shank2 KO mice. However, the treatment of L838,417 reverses spatial memory deficits in Shank2 e6-7 KO mice, while it had no effects in Shank2 e7 KO mice. These findings would be attributed to understanding the heterogeneity of ASD and developing effective treatments for each patient. 19

27 EXPERIMENTAL PROCEDURES Animals I used two Shank2 KO mouse lines, Shank2 KO e67 and Shank2 KO e7, for following experiments. They were targeted and deleted exon6 and exon7, or only exon7 respectively (Figure 2). Each line had been backcrossed to C57Bl/6N and C57Bl/6J respectively. To generate homogeneous KO mice and WT littermates, I crossed heterogeneous male with heterogeneous female mice. For the RNA sequencing, western blot and electrophysiological experiments, 4 to 5 weeks old male mice were used and for the behavioral experiments, 8 to 15 weeks old male mice were used. Food and water were provided ad libitum and all the mice were kept on a 12 h light/dark cycle. All the experimental procedures were done during the light phase of the cycle. The Institutional Animal Care and Use Committee of Seoul National University approved the animal care and experiments. 20

28 Figure 2. Strategies to delete Shank2 by targeting exon 6 and 7 or only exon 7 (A) KO strategy to delete Shank2 by targeting exon 6 and 7 (B) KO strategy to delete Shank2 by targeting exon 7 21

29 RNA sequencing After anesthetization with isoflurane, hippocampi of Shank2 KO e67 or Shank2 KO e7 mice with their respective littermate WT mice was extracted. To make RNA libraries, the hippocampus was grinded and total RNA was separated by TRIZOL (Invitrogen). 5 μg of poly-a mrna was processed by Illumina Truseq RNA Sample Prep Kit with poly-t oligo-attached magnetic beads. To make cdna libraries, reverse transcription was performed using Superscript II reverse transcriptase (Life Technologies). The adaptor-ligated libraries were separated in gel electrophoresis and extracted using QIAquick gel extraction kit (Qiagen). These libraries were sequenced on Illumina HiSeq 2500 (NICEM, Seoul National University) in the paired-end sequencing mode (2 101 bp reads), and these raw reads were mapped onto the mouse genome reference mm10 using GSNAP (version ). Uniquely and properly mapped read pairs were used for further analysis. Gene expression level was calculated based on the RPKM (reads per kilobase of exon per million mapped reads) measure. To identify differentially expressed genes (DEGs) between the WT and KO groups, cuffdiff module in the Cufflinks package (version v2.1.1) was used and in this analysis DEGs were defined as those with changes of at least 1.5-fold between samples at a false discovery rate of 10%. Finally, PANTHER (Protein ANalysis THrough Evolutionary Relationships) Tools ( and GO term enrichment analysis were used to divide the DEGs into biological and functional protein classes. Quantitative real time-pcr (qrt-pcr) cdna samples were generated from RNA samples used for RNA sequencing. qrt-pcr was performed using ExTaqII SYBR Green Master Mix (Takara) in ABI7300 (Applied Biosystems) system. Primer sequences for each gene are listed below. Shank1 5 CCGCTACAAGACCCGAGTCTA 3 5 CCTGAATCTGAGTCGTGGTAGTT 3 Shank3 5 CCGGACCTGCAACAAACGA 3 5 GCGCGTCTTGAAGGCTATGAT 3 Gria1 5 GTCCGCCCTGAGAAATCCAG 3 5 CTCGCCCTTGTCGTACCAC 3 22

30 Gria2 5 TTCTCCTGTTTTATGGGGACTGA 3 5 CTACCCGAAATGCACTGTATTCT 3 Gria3 5 ACCATCAGCATAGGTGGACTT 3 5 ACGTGGTAGTTCAAATGGAAGG 3 Grin1 5 TGGCCGTGTGGAATTCAATG 3 5 TTGTGGGCTTGACATACACG 3 Grin2a 5 GACATCCACGTTCTTCCAGTT 3 5 GGCGTCCTCAAAAGAGGTGT 3 Grin2b 5 GCCATGAACGAGACTGACCC 3 5 GCTTCCTGGTCCGTGTCATC 3 Dlg4 5 ACCAGAAGAGTATAGCCGATTCG 3 5 GGTCTTGTCGTAGTCAAACAGG 3 Dlgap1 5 AAACCGATGTCTGTCTATTGGGA 3 5 GGGCCAGGACACGAATTGT 3 Grm1 5 TGGAACAGAGCATTGAGTTCATC 3 5 CAATAGGCTTCTTAGTCCTGCC 3 Grm5 5 CCCAGCACAAGTCGGAAATAG 3 5 TGTCTGGTTGGGGTTCTCCTT 3 Gabra2 5 CCAAGTGTCATTCTGGCTGA 3 5 GCCCATCCTCTTTTTGTGAA 3 Cycling conditions were 95 C for 30 s, followed by 40 cycles of 95 C for 5 s and 60 C for 31 s. GAPDH was used as an internal control for each gene. L838,417 preparation L838,417 (Sigma, L9169) was dissolved in DMSO at 10 mg/ml and aliquots of L838,417 was stored at 20 C. To administrate L838,417 in the behavioral experiments, 10 mg/ml of L838,417 was diluted in saline to a final concentration of 0.02 mg/ml. DMSO diluted in saline at a concentration of 0.2% was used as a vehicle mg/ml of L838,417 and 0.2% DMSO were also stored at 20 C and thawed immediately before to use. 5 ml/kg (0.1 mg/kg) of L838,417 or vehicle was injected intraperitoneally for the behavioral experiments. Drug injection was randomized by other experimenters who did not perform behavior experiments. Behavioral Tests Elevated zero maze (EZM) test A round-shaped (inner diameter: 50 cm, outer diameter: 60 cm) maze which is made of white Plexiglass and elevated 40 cm from the floor, was used. The maze was composed of 4 arms, 2 of open arms and 2 of closed arms with 20 cm walls on both sides. A mouse was placed on the center 23

31 of one of closed arm of the maze, and the movement of the mouse was tracked for 5 min by using a tracking program (EthoVision 9.0, Nodulus), under bright light. If a mouse fell onto the floor, the mouse was excluded from the analysis. 30 min before the test, L838,417 or vehicle was injected and experimenters were blinded to the genotypes and identity of injected drugs. Open-field test A mouse was placed in the middle of a square white plexiglass box ( cm) under dim light. The movement of the mouse and location in the box was tracked using a tracking program (EthoVision 9.0, Nodulus) for 20 min. The center was defined as the middle area that occupied 1/2 of the total area. 30 min before the test, L838,417 or vehicle was injected and experimenters were blinded to the genotypes and identity of injected drugs. Morris water maze (MWM) test Mice were handled for 3 min on 4 consecutive days before performing the test. During training session, a mouse was placed into a white opaque water-filled (22 23 C) tank (140 cm diameter, 100 cm height) placed in a room with multiple spatial cues under dim light. The tank was divided into 4 virtual fan-shaped quadrants, which target quadrant (TQ), opposite quadrant (OQ), adherent quadrant 1 (AQ1) and adherent quadrant 2 (AQ2). In a fixed location of TQ, a 10 cm diameterplatform was placed. On the training days, mice were pseudo-randomly released at the edge of the maze facing the inner wall of the tank and trained to reach the platform for 60 s. If the mice failed to reach the platform for 60 s, they were guided to the platform and maintained on the platform for 10 s. When the mice successfully reached the platform and stayed on the platform for more than 1 s, mice were rescued from the maze after 10 s. Mice were trained with 4 trials per day to be released at all the quadrants and the trial interval was 2 min. During the 6 consecutive training days, 30 min before the first trial, L838,417 or vehicle was injected. The day after the final day of training, a probe test was performed. In the probe test, the platform was removed and a mouse was released at the center of the pool. The mouse was tracked for 60 s with a tracking program (EthoVision 9.0, Nodulus) and, mean distance from the platform, the number of platform location crossing and time spent in each quadrant were analyzed. Experimenters were blinded to 24

32 the genotypes and identity of injected drugs. Three-chamber test A three-chamber apparatus was made up of successive three chambers with a door between the faced two chambers (two outer chambers: 20 x 15 x 25 cm and an inner chamber: 20 x 10 x 25 cm). Stranger mice were habituated in a metal-wired cage placed in the outer chamber for 10 min on 4 consecutive days. Shank2 KO mice and their littermate WT mice (test mice) were also habituated in the apparatus with the doors open on two consecutive days and they had not met with the stranger mice. 24 hours after the habituation, the test mouse was placed in the inner chamber with the doors open. After 10 min, the test mouse was guided to the inner chamber and the doors were closed. A wired cup with the stranger mouse (stranger) and a wired cup with yellow block (object) were introduced into each of the outer chamber, and then the doors were opened to test sociability. The movement of the test mouse was tracked for 10 min with a tracking program (EthoVision 9.0, Nodulus). 30 min before the test, L838,417 or vehicle was injected and experimenters were blinded to the genotypes and identity of injected drugs. Experimenters were blinded to the genotypes and identity of injected drugs. Electrophysiology After anesthetization with isoflurane, mice were decapitated and their brains were extracted. For spontaneous/miniature and evoked IPSC experiments, 300 μm thick hippocampal slices were sectioned using a vibrating blade microtome (Leica VT1200S) and incubated in a chamber at 28 C for at least 1 h for recovery. After the recovery period, the slices were placed in a recording chamber at 25 C and perfused (1 1.5 ml/min) with oxygenated artificial cerebrospinal fluid (ACSF: 124 mm NaCl, 2.5 mm KCl, 1 mm NaH2PO4, 25 mm NaHCO3, 10 mm glucose, 2 mm CaCl2, and 2 mm MgSO4, 290 mosm). For the spontaneous/miniature recording, pyramidal cells of CA1 area were recorded by whole-cell patch clamp using a glass electrode (3 4 MΩ) filled with internal solutions. One of internal solution was chosen depending on experimental conditions between Cs-gluconate internal solution containing 100 mm Cs-gluconate, 5 mm NaCl, 10 mm HEPES, 10 mm EGTA, 20 mm TEA-Cl, 3 mm QX-314, 4 mm MgATP, and 0.3 mm Na3GTP 25

33 ( mosm, ph adjusted to 7.2 with CsOH) and High-Cl internal solution containing 145 mm KCl, 5 mm NaCl, 10 mm HEPES, 10 mm EGTA, 10 mm QX-314, 4 mm MgATP, and 0.3 mm Na3GTP ( mosm ph adjusted to 7.2 with KOH). For the evoked IPSC recording, the Schaffer collateral (SC) pathway was stimulated every 20 s using concentric bipolar electrodes (MCE-100; Kopf Instruments). For E/I ratio recording, evoked excitatory currents through AMPA-R and inhibitory currents through GABA-R were recorded in the same CA1 pyramidal cells. AMPA-R-mediated currents were recorded by holding at 40 mv in AP5-containing ACSF and then switched to ACSF with AP5 and CNQX to measure GABA-R-mediated currents at holding potential of 0 mv from the same cells. About 200 μm away from the recorded cells, electronic stimulations were given to obtain approximately 300 pa of AMPA current. Experimenters were blinded to the genotypes and treatment conditions and treatment of L838,417 or vehicle was randomized by other experimenters who did not perform electrophysiological experiments. Western Blot After anesthetization with isoflurane, hippocampi of Shank2 KO e6-7 or Shank2 KO e7 with their respective littermate WT mice was extracted. Then the hippocampus in the lysis solution (30 mm Tris-Cl ph 7.4, 4 mm EDTA, 1 mm EGTA, Protease inhibitor cocktail) was homogenized with metal beads using a vibrating homogenizer (Qiagen). The homogenates were centrifuged at 500 g for 5 min at 4 C to remove nucleus fraction. The supernatant was centrifuged at 100,000 g for 1 h at 4 C and then the pellet was lysed in 0.5% Triton X-100 added lysis solution. The lysates were carefully loaded onto the surface of 1 M sucrose and then centrifuged again at 100,000 g for 1 h at 4 C. Then, total protein amounts were quantified in the supernatants by the Bradford assay. Equal amounts of proteins were electrophoresed on SDS polyacrylamide gels and separated proteins were transferred onto a nitrocellulose membrane. After blocking with 5% skim milk in Tris-buffered saline plus Triton X-100 (TBST) for 2 h at room temperature, membranes were incubated with rabbit anti-gabaa-r α2 antibody (1: 2,000, R&D Systems) overnight at 4 C. After washing with TBST buffer, the membranes were treated with an HRP-conjugated secondary antibody for 2 h at room temperature. The protein bands by HRP with enhanced 26

34 chemiluminescence reagents were detected and analyzed using the ChemiDoc (Bio-Rad) program. Statistics For the Morris water maze, the one-way ANOVA (analysis of variance) was performed to examine whether the mice spent significantly more time in the target quadrant than in the other three quadrants. Effects of L838,417 on different genotypes were analyzed using two-way ANOVA followed by Bonferroni post-hoc test. When two groups were compared, the unpaired two-tailed t-test was used. All data are represented as mean ± SEM and the statistical analyses were performed with Graphpad 5.0 software. 27

35 RESULTS Comparison of transcriptomes of the two Shank2 KO lines It had been interesting to me that Shank2 KO e67 mice and Shank2 KO e7 mice have opposite physiological characteristics related with NMDA-R functions. Thus I decided to compare the two KO lines further using RNA sequencing technique. Since the previous differences in NMDA-R functions of the two KO lines were examined in the hippocampus, I made cdna libraries from the hippocampus of the two KO mice and their respective WT littermates. At first, RNA sequencing confirmed that exon 6 and 7 or exon 7 of Shank2 gene was completely deleted in the two KO lines respectively (Figure 3). Next, the mrna expression profile of each KO line was generated compared to their WT littermates in a hierarchical clustering analysis and 93 and 59 genes are identified as DEGs respectively in Shank2 KO e67 and Shank2 KO e7 (Figure 5, Tables 1 and 2). Furthermore, a GO enrichment analysis indicated that these DEGs were involved in lots of biological processes, including synaptic functions (Tables 3 and 4). These data showed that the two Shank2 KO lines were differently affected by the genomic modifications in Shank2 gene, and suggested the possibility of further differences in synaptic functions other than NMDA-R related functions. 28

36 Figure 3. Confirmation of Shank2 expression in Shank2 KO e67 and e7 mice in RNA sequencing (in collaboration with Dr. Jae-Hyung Lee) Figure 4. Differential gene expression profiles of Shank2 e67 or e7 KO lines (WT vs KO) (A) Volcano plots showing up-regulated (Kroon et al.) and down-regulated (blue) DEGs, as the x-axis and the y-axis representing the magnitude of fold changes (log2 transformed) and adjusted p-value (-log2) respectively. (B) A heatmap of expression levels of DEGs in four different groups (log2rpkm; reads per kilobase of exon per million mapped reads). (in collaboration with Dr. Jae-Hyung Lee) 29

37 Table 1. Differentially expressed genes in Shank2 KO e67 mice (in collaboration with Dr. Jae-Hyung Lee) 30

38 Table 2. Differentially expressed genes in Shank2 KO e7 mice (in collaboration with Dr. Jae-Hyung Lee) 31

39 Table 3. Gene ontology (GO) analysis on DEGs of Shank2 KO e67 mice (in collaboration with Dr. Jae-Hyung Lee) 32

40 Table 4. GO analysis on DEGs of Shank2 KO e7 mice (in collaboration with Dr. Jae-Hyung Lee) 33

41 Among the DEGs, mrna expression of Gabra2 was significantly down-regulated in Shank2 KO e67 mice, but not in Shank2 KO e7 mice. Based on this result, I performed quantitative RT- PCR to evaluate the mrna expression of genes related to postsynaptic functions (Figure 5A, B). I confirmed that the mrna expression of Gabra2 was significantly reduced only in Shank2 KO e67 mice. In addition, mrna expression of a gene encoding the subunit 1 of NMDA-R (Grin1) was reduced in Shank2 KO e67 mice, but not in Shank2 KO e7 mice. mrna expression of a gene encoding the subunit 1 of AMPA-R (Gria1) and a gene encoding a postsynaptic density protein, Dlg4, were increased in Shank2 KO e7 mice, but not in Shank2 KO e67 mice. However, the alterations in the mrna expression of Grin1, Gria1 and Dlg4 were not that dramatic and I focused Gabra2. To evaluate the GABAAα2 protein level, I performed western blot analysis in the hippocampus of the Shank2 KO e67 mice. Since GABAAα2 is primarily located in postsynaptic area, I extracted the synaptic fraction from the hippocampus and found that GABAAα2 was significantly decreased in Shank2 KO e67 mice compared to their WT littermates (Figure 5C, D). In addition, we found another difference between the two Shank2 KO lines. An unknown transcript of Shank2 is expressed only in Shank2 KO e67 mice, which starts from center of exon 8 and expanded to exon 16 without SH3 and PDZ domains (data not shown). However, since it was not certain whether the transcript is really expressed as a protein, I focused on the difference level of Gabra2 expression. 34

42 Figure 5. Decreased Gabra2 expression in mrna and protein level (A, B) qrt-pcr analyses to evaluate the expression level of postsynaptic genes from the hippocampus of Shank2 KO e67 and e7 mice (WT: n = 3, KO: n = 3, unpaired t-test *P < 0.05, **P < 0.01). (C, D) Reduced expression of GABAAα2 in the hippocampus of Shank2 KO e67 mice. Pan-cadherin was used as internal control and the expression levels of GABAA α2 subunit are normalized by the intensity of (WT: n = 8, KO: n = 6, unpaired t-test **P < 0.01). (in collaboration with Dr. Nam-Kyung Yu Dr. Hyoung-Gon Ko) 35

43 Functional identification of decreased GABAA2 in Shank2 KO e67 mice I found that the expression of GABAAα2 was reduced only in the Shank2 KO e67 mice. GABAA receptors are localized in both synaptic and extrasynaptic sites and primarily involved in the inhibitory neurotransmission in the cortex and hippocampus (Benarroch, 2007). Thus I expected that the inhibitory transmission of Shank2 KO e67 mice was impaired, changing the normal E/I balanced states. To examine it, GABAergic inhibitory currents were recorded from CA1 pyramidal neurons using whole-cell patch clamp while the Schaffer collateral pathway was stimulated with varying intensities (input-output relationship). Consistently with the decreased expression of GABAα2, the amplitude of inhibitory currents was slightly, but significantly reduced in the Shank2 KO e67 mice compared to their WT littermates (Figure 6A) On the other hand, the input-output relationship of inhibitory currents of the Shank2 KO e7 mice was comparable to their WT littermates (Figure 6B). The reduced inhibitory transmission in Shank2 KO e67 mice could be resulted from general impairment of neurotransmission. Thus, the AMPA-R mediated excitatory transmission was also examined and AMPA-R currents was not affected by Shank2 KO in both lines (Figure 6C). As a result, the E/I ratio was increased in the Shank2 KO e67 mice, while the ratio was not altered in the Shank2 KO e7 mice (Figure 6D). To further confirm whether the change in inhibitory currents are resulted from an impairment in GABAA functions by reduced GABAα2, a GABAA receptor agonist, especially containing α2 subunit, L838,417, was treated during the patch clamp recording. The AMPA-R mediated currents were not affected, while the reduced GABA-R currents and E/I ratio of Shank2 KO e67 mice were restored being comparable to those of WT littermates. The GABA-R currents of WT mice were not increased by L838,417 and it might be resulted from the ceiling effect (Figure 6C, D). 36

44 Figure 6. Decreased inhibitory GABAergic function and rescue of it by L838,417 in Shank2 e67 KO mice (A) Decreased GABAergic current in input-output relationships of CA1 neurons of Shank2 KO e67 mice compared to WT littermates (WT-Vehicle: n = 27 cells/6 animals; KO-Vehicle: n = 28 cells/6 animals; two-way ANOVA, genotype x intensity interaction, *P < 0.05). (B) Normal GABAergic current in input-output relationships of CA1 neurons of Shank2 KO e7 mice (WT: n = 21 cells/4 animals; KO: n = 20 cells/4 animals; two-way ANOVA, genotype x intensity interaction, P = 0.743). (C) Representative traces (left) and plot (right) of AMPAR or GABAR mediated currents from CA1 neurons of Shank2 KO e67 mice after treatment with vehicle or L838,417 (GABAR currents, WT-Vehicle vs. KO-Vehicle, unpaired t-test, **P < 0.01; KO- Vehicle vs. KO-L838,417, unpaired t-test, **P < 0.01). Scale bar: 200 pa, 100 msec. (D) Increased E/I ratios in CA1 neurons in Shank2 KO e67 mice compared to WT littermates. The E/I ratio was rescued by L838,417 treatment (WT-Vehicle: n = 12 cells/5 animals; KO-Vehicle: n = 13 cells/5 animals; WT-L838,417: n = 9 cells/4 animals; KO-L838417: n = 13 cells/5 animals; two-way ANOVA, effect of genotype, *P < 0.05, Bonferroni posttests, KO-Vehicle vs. KO- L838,417, **P < 0.01). NS, not significant. (in collaboration with Dr. Chae-Seok Lim) 37

45 In addition to the evoked transmissions, spontaneous miniature EPSC (mepsc) and IPSC (mipsc) were also recorded in CA1 pyramidal cells since in many neurodevelopmental disorders, the frequency or amplitude of the spontaneous postsynaptic currents are also perturbed (Gkogkas et al., 2013; Han et al., 2012; Kim et al., 2016; Lee et al., 2017). The frequency or amplitude of mepsc and mipsc of Shank2 KO were not significantly different from those of WT littermates (Figure 7A, B). Presynaptic release probability was also examined by paired-pulse ratio (PPR) in the CA1 pyramidal cells at two stimulation intervals (70 ms and 150 ms) and at both intervals, the PPR was not different between Shank2 KO e67 mice and WT littermates (Figure 7C). 38

46 Figure 7. Recording of spontaneous mepsc or mipsc, and paired-pulse facilitation ratios (PPR) (A, B) Normal spontaneous and miniature inhibitory postsynaptic currents (sipsc and mipsc) in Shank2 KO e67 mice compared with WT littermates (sipsc, WT: n = 26; KO: n = 25; mipsc, WT: n = 21; KO: n = 18 cells). Scale bar: 300 pa, 2 s. (C) Normal paired-pulse facilitation ratios (PPR) in Shank2 KO e67 mice compared with WTlittermates (70 ms, WT: n = 13; KO: n = 14; 150 ms, WT: n = 9; KO: n = 12 cells; two-way ANOVA,genotype: F1,44 = 0.03, P = 0.855). NS, not significant (in collaboration with Dr. Chae-Seok Lim) 39

47 Effects of L838,417 in the social behaviors of Shank2 KO e67 mice. Alterations in E/I balance are frequently observed phenomena in the animal models of neurodevelopmental disorders (Lee et al., 2017; Peñagarikano et al., 2011; Schmeisser et al., 2012; Won et al., 2012; Yizhar et al., 2011), and suggested as one of underlying physiological mechanism of the cognitive deficits of the disorders. Previously, Shank2 KO e67 mice are characterized by autistic behaviors such as social deficits and repetitive jumping behaviors, and enhancing NMDA-R function partially improved social behaviors of Shank2 KO e67 mice (Won et al., 2012). To estimate if restoring the E/I balance by enhancing GABAA-R function also could rescue the social deficits of the mice, I performed three-chamber test after treating L838,417 in the Shank2 KO e67 mice. Consistently with the previous report, the sociability of Shank2 KO e67 mice was significantly impaired compared to that of WT littermate mice, however, the impaired sociability was not rescued by the L838,417 treatment (Figure 8). The result suggests that the decreased GABAA-R function may not attribute to the social deficits in the Shank2 KO e67 mice. 40

48 WT-Vehicle WT-L838,417 KO-Vehicle KO-L838,417 Exploration time (s) * * 0 Object Stranger Object Stranger Object Stranger Object Stranger Figure 8. Effects of L838,417 treatment on impaired social behaviors of Shank2 KO e67 mice L838,417 did not affect the sociability of Shank2 KO e67 mice in the three-chamber test (WT- Vehicle: n = 9; WT-L838,417: n = 8; KO-Vehicle: n = 8; KO-L838,417: n = 7; unpaired t-test, *P < 0.05). 41

49 Effects of L838,417 in spatial memory deficits of Shank2 KO e67 mice. Individuals of ASD are suffer from various cognitive deficits such as anxiety, learning and memory deficits, as well as the impaired social behaviors. Shank2 KO e67 mice also displays spatial learning and memory deficits (Won et al., 2012). I focused on the spatial learning and memory, since intellectual disabilities are frequently accompanied with the ASD patients. In addition, imbalances of the E/I balance in the hippocampus have also been reported to be involved in the learning and memory deficits (Cui et al., 2008; Han et al., 2014; Han et al., 2012; Houbaert et al., 2013; Lee et al., 2014). I examined whether the spatial learning and memory deficits of Shank2 KO e67 mice in Morriswater maze could be restored by the L838,417 treatments. As previously reported, vehicle treated Shank2 KO e67 mice did not acquire the position of the hidden platform well compared to WT littermate mice during the training sessions (Figure 9A). L838,417 treated Shank2 KO e67 mice also could not find the platform well and the latency to find the platform was not significantly different from that of vehicle treated Shank2 KO e67 mice. L838,417 treated WT littermate mice learned the position of the platform well. In the probe test, WT littermate mice spent significantly more time in the target quadrant where the platform was located during the training sessions, while vehicle treated Shank2 KO e67 mice randomly went around the pool and spent similar time in each quadrant. On the other hand, L838,417 treated Shank2 KO e67 mice spent significantly more time in the target quadrant (Figure 9B). Moreover, the mean distance from the center of the platform was significantly higher in the vehicle treated Shank2 KO e67 mice group compared to the L838,417 treated group (Figure 9C). The mean swimming velocity was comparable in all groups, suggesting that these results were not coming from the alterations on locomotive properties (Figure 9D). 42

50 Figure 9. Effects of L838,417 treatment on spatial memory deficits of Shank2 KO e67 mice (A) Learning curve of Shank2 KO e67 mice in the Morris water maze task (WT-Vehicle vs KO- Vehicle, two-way ANOVA, genotype: F1,90 = 17.74, P < 0.001; KO-Vehicle vs KO-L838,417, two-way ANOVA, treatment: F1,100 = 1.704, P > 0.05). (B) Time spent in each quadrant during probe test, 24 h after training (WT-Vehicle: n = 9; WT-L838,417: n = 8; KO-Vehicle: n = 11; KO-L838,417: n = 11; one-way ANOVA of KO-Vehicle, NS; one-way ANOVA of KO-L838,417, ***P < 0.001). TQ: target, OQ: opposite, AQ1: right, AQ2: left quadrant. (C) Enhanced spatial memory of Shank2 e67 KO mice in mean distance from the platform during probe test by L838,417 (WT-Vehicle: n = 9; WT-L838,417: n = 8; KO-Vehicle: n = 11; KO-L838,417: n = 11, two-way ANOVA followed by Bonferroni posttests, *P < 0.05). (D) Swimming velocity of Shank2 KO e67 mice was comparable with WT littermates during probe test (WT-Vehicle: n = 9; WT-L838,417: n = 8; KO-Vehicle: n = 11; KO-L838,417: n = 11, unpaired t-test). NS, not significant. 43

51 I examined spatial memory of Shank2 KO e7 mice in the Morris-water maze task either, to address whether the mice also show deficits in spatial learning and memory. In spite of the normal E/I ratio of them, the Shank2 KO e7 mice also showed a tendency of deficit during the training days (Figure 10A). Furthermore, vehicle treated Shank2 KO e7 mice randomly swam around the four quadrants, while WT littermates spent significantly more time in the target quadrant during the probe test. The results indicated that the spatial memory of Shank2 KO e7 mice is also impaired (Figure 10B). If the rescue effect of L838,417 in Shank2 KO e67 mice was really resulted from the restoring of E/I balance, I expected that the drug should have no memory rescue effect in Shank2 KO e7 mice. As I expected, when I treated L838,417 in Shank KO e7 mice, contrary to e6-7 KO mice, the mice did not show a preference for the target quadrant, still randomly swimming around the four quadrants during the probe test (Figure 10B). In addition, the mean distance from platform position was comparable between the L838,417-treated Shank2 KO e7 mice group and the vehicle treated group, suggesting that L838,417 did not rescue the impaired spatial memory of Shank2 KO e7 mice (Figure 10C). As in the Shank2 KO e67 mice, L838,417 treatment did not affect the swimming velocity (Figure 10D). Taken together, these results suggest that the spatial memory deficits of Shank2 KO e7 mice was not resulted from the GABAA-R functions, but from other mechanisms, while the alteration of GABAA-R functions is attributable for the spatial memory deficits of Shank2 KO e67 mice. 44

52 Figure 10. Effects of L838,417 treatment on spatial memory deficits of Shank2 KO e7 mice (A) Learning curve of Shank2 KO e7 mouse line during training in the Morris water maze task (WT-Vehicle vs KO-Vehicle, two-way ANOVA, effect of genotype, P = 0.146; KO-Vehicle vs KO-L838,417, two-way ANOVA, effect of treatment: P = 0.639). (B) Time spent in each quadrant during probe test, 24 h after training (WT-Vehicle: n =10; WT-L838,417: n = 6; KO-Vehicle: n = 12; KO-L838,417: n = 11; one-way ANOVA of WT-Vehicle, *** P < 0.001; WT-838,417, *** P < 0.001; KO-Vehicle, P = 0.168; KO-L838,417, P = 0.961). TQ: target, OQ: opposite, AQ1: right, AQ2: left quadrant. (C) L838,417 did not improve mean distance from the platform of Shank2 e7 during probe (WT-Vehicle: n = 10; WT-L838,417: n = 6; KO-Vehicle: n = 12; KO-L838,417: n = 11, two-way ANOVA, effect of genotype, P < 0.01; effect of treatment, P = 0.404). (D) Swimming velocity of Shank2 KO e7 mice was comparable with WT littermates during probe test (WT- Vehicle: n = 10; WT-L838,417: n = 6; KO-Vehicle: n = 12; KO-L838,417: n = 11, unpaired t-test). (in collaboration with Jaehyun Lee) 45

53 Effects of L838,417 in emotional status and locomotion of Shank2 KO e67 mice. In addition to the behavioral tests involved in sociability and spatial memory, I performed elevated zero-maze task and open-field test to examine effects of L838,417 on the emotional states or locomotion (Figure 11 A, B). In the elevated zero-maze, time in open area of vehicle treated Shank2 KO e67 mice was significantly different with that of vehicle treated WT littermates. L838,417 treated Shank2 KO e67 mice also spent comparable time with the vehicle treated Shank2 KO e67 group. Similarly, in the open-field test, the center duration of L838,417 treated Shank2 KO e67 group was not different from that of vehicle treated KO group. These results suggest that L838,417 treatment did not affect the anxiety of the Shank2 KO e67 mice. In relation with locomotion, in both elevated zero-maze and open-field test, the hyperactivity of Shank2 KO e67 was tended to reduce by L838,417 but it was not statistically significant. Taken together, our results suggest that restoring GABAergic synaptic transmission and E/I balance using L838,417 could ameliorate the spatial memory deficit of Shank2 e6-7 KO mice, without affecting the social or emotional behaviors of the KO mice. 46

54 Figure 11. Effects of L838,417 on anxiety like behavior and locomotor activity of Shank2 KO e67 mice (A) No changes in basal anxiety level of Shank2 e6-7 KO mice and WT littermates following administration of L838,417 in the elevated zero-maze (EZM) test (left, WT-Vehicle: n = 11; WTL838,417: n = 10, KO-Vehicle: n = 8; KO-L838,417: n = 7; two-way ANOVA, interaction, F1,32 = 0.339, P = 0.564) while locomotor activity was significantly increased compared with WT littermates (right, WT-Vehicle vs. KO-Vehicle, unpaired t-test, * P < 0.05). (B) In the openfield test, Shank2 e6-7 KO mice showed significantly decreased time spent in the center region compared with WT littermates (left, WT-Vehicle: n = 11; WT-L838,417: n = 10; KO-Vehicle: n = 8; KO-L838,417: n = 7, KO-Vehicle vs KO-L838,417, unpaired t-test, * P < 0.05), while locomotor activity was significantly increased compared with WT littermates (right, WT-Vehicle vs KO-Vehicle, unpaired t-test, *** P < 0.001), which was partially rescued by L838,417 administration (KO-Vehicle vs. KO-L838,417, unpaired t-test, P = 0.183). NS, not significant. 47