Molecular analysis of GTF2IRD1: a implicated in the neurobehavioural features of Williams-Beuren Syndrome

Jocelyn Widagdo BSc (Hons Class I)

A thesis submitted to the University of New South Wales in fulfillment of the requirements for the degree of Doctor of Philosophy

Faculty of Medicine University of New South Wales Sydney, Australia February 2011 COPYRIGHT STATEMENT

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed

( 05 j 5- o t \ Date

AUTHENTICITY STATEMENT

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’

Signed

Date ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed JOCELYN WIDAGDO

Date ACKNOWLEDGEMENTS

There are so many people who have given me supports in many ways during the past four years of this challenging yet most rewarding experience. Firstly, thank you to Prof Edna Hardeman, who has given me continual supports and encouragement since I was an undergraduate summer scholar, and which I have no doubt, will keep going beyond this day. I am grateful for your mentoring, attentiveness, and critical mind that always push me to find ways to reach for the 'answer. To Dr Stephen Palmer, 1 am greatly thankful for your dedicated supervision, invaluable advice from project direction to lab technical matters, and your support during the intense thesis writing. Yes, it has come to an end!

Thank you to Kylie Taylor who is always happy to help with any experiments and lab matters. To all members of the Oncology Research Unit and the bigger Cellular Genetic Medicine Unit, thank you for the friendship and making work so much more enjoyable. Thanks to Prof Peter Gunning for the many enthusiastic discussions, constant support and direction to the project. Thanks to Renee Szokolai for keeping our lab safe and running efficiently. Thanks to Christine Lucas who can always bring laughter and to Josephine Joya for being the mother figure in the lab.

I would like to thank the past members of the lab and colleagues at the Children’s Medical Research Institute (CMRI) at Westmead for getting me started in the early years of my research life. Special thanks to Dr Enoch Tay for the supervision and introducing me to the research world and to Kata Popovic for the friendship and assistance. Thanks to Dr Christine Smyth and staffs at the Westmead Millenium Institute for the tireless help with the seemingly impossible FACS project! I would also like to thank all our colleagues who have kindly provided me with the reagents which would otherwise not be available for my research. I would also like to acknowledge the financial support from the University of Sydney (University Postgraduate Award) and the CMRI (supplementary scholarship), which then continued by the Faculty of Medicine, University of New South Wales.

To Victor Anggono, my fiancee and mentor, thank you for your faith and dedication and endless support throughout this journey. It wasn't easy, but we were where we were, learning what we have to. Lastly, I am blessed for having supportive and loving big families and I thank God for this. PUBLICATIONS

PUBLISHED ARTICLES

Palmer SJ, Santucci N, Widagdo J, Bontempo SJ, Taylor KM, Tay ES, Hook J, Lemckert F, Gunning PW, Hardeman EC (2010) Negative autoregulation of GTF2IRD1 in Williams-Beuren Syndrome via a novel DNA binding mechanism. J Biol Chem 285: 4715-4724.

MANUSCRIPTS IN PREPARA TION

Howard ML, Tay E, Palmer SJ, Widagdo J, Taylor KM, Popovic K, Gunning PW, Spitzer MW, Murphy M, Arrthurson GJ, Pang TYC, Renoir T, Hannan AJ, Hardeman EC (201 1) Mutation of the Williams-Beuren syndrome-associated Gtflirdl causes altered GABAergic function and a unique behavioural phenotype including heightened vocalization. Submitted to Molecular Genetics (under review).

Widagdo J, Taylor KM, Bontempo SJ, Hardeman EC, Palmer SJ (2011) Novel interaction between -related protein GTF21RD1 and a chromatin- associated protein, ZMYM5, is modulated by SUMOylation.

Palmer SJ, Taylor KM, Santucci N, Widagdo J, Chan A, Yeo JL, Reynaud B, Guven K, Joya J, Adams M, Gunning PW, Hardeman EC (2011) GTF2IRD2 from the Williams-Beuren critical region encodes a nuclear/cytoplasmic protein that antagonizes GTF2IRD1 and TFII-I.

CONFERENCE ABSTRA CTS

Hardeman EC, Widagdo J, Howard ML, Hannan AJ, Gunning PW and Palmer SJ (2011) GTF2IRD1, implicated in the Williams syndrome cognitive profile, interacts with chromatin-regulating . Talk presented at The Australian Neuroscience Society annual meeting, Auckland, New Zealand. Palmer SJ, Widagdo J, Taylor KM, Bontempo S, Gunning PW and Hardeman EC (2010) Specific interactions of the Williams-Beuren syndrome-associated protein GTF2IRD1 suggest a role in the chromatin reorganization. Poster presented at The American Society of Human Genetics annual meeting, Washington DC, USA.

Widagdo J, Palmer SJ, Taylor KM, Bontempo S and Hardeman EC (2010) GTF2IRD1, a Williams Syndrome-related protein, is a novel SUMO substrate. Poster presented at the OzBio2010, Melbourne, VIC.

Widagdo J, Tay ESE, Howard ML, Hannan AJ, Gunning PW, Hardeman EC and Palmer SJ (2010) GTF2IRD1, a Williams Syndrome-related protein, interacts with chromatin-regulating proteins: a novel role in epigenetics. Poster presented at the Gordon Research Conference: Molecular and Cellular Neurobiology, Hong Kong.

Palmer SJ, Santucci N, Widagdo J, Bontempo S, Taylor KM, Tay ESE, Gunning PW and Hardeman EC (2009) The biochemistry and evolution of the Williams Beuren syndrome-associated gene GTF2IRD1. Talk presented at The American Society of Human Genetics annual meeting, Honolulu, Hawaii.

Widagdo J, Tay ESE, Popovic K and Hardeman EC (2009) Molecular function of GTF2IRD1, a factor implicated in Williams-Beuren Syndrome. Poster presented at The Australian Neuroscience Society annual meeting, Canberra, ACT.

Widagdo J, Tay ESE and Hardeman EC (2007) A potential role for the serine protease inhibitor Serpina3n in the anxiety-like phenotype of Williams Syndrome. Poster presented at Westmead Association Research Meeting Symposium, Westmead Hospital, We st me ad, NSW.

Widagdo J, Tay ESE and Hardeman EC (2007) Towards the molecular mechanism of Gtf'Iirdl in the neurocognitive pathology of Williams syndrome and human cognition/behaviour. Presented at the 7,h IBRO World Congress of Neuroscience, Melbourne, VIC. (Presenting authors are positioned first) ABSTRACT

Background: Williams-Beuren syndrome (WBS) is a complex neurodevelopmental disorder that results from a hemizygous deletion involving up to 28 within 7q 11.23. WBS is characterised by a set of physical abnormalities and specific cognitive and behavioural features that are collectively called the Williams syndrome cognitive profile (WSCP). Genotype-phenotype correlations in patients with atypical deletions have implicated two members of the GTF2I protein family,

GTF2IRD1 and TFI1-I, in the main aspects of the WSCP. We and others have generated

Gtflirdl knockout mouse lines that show developmental and neurological abnormalities, including social and non-social anxiety-related behavioural changes, reminiscent of the WSCP. By analysing using a Gtf2irdl knockout/LacZ knockin mouse, we identified a major overlap of Gtfiirdl expression in

GABAergic cell types, the major inhibitory neurons in the . However, the molecular function of GTF2IRD1 is poorly understood. GTF2IRD1 is generally considered to be a DNA-binding with a number of proposed target genes that were originally identified in yeast one-hybrid experiments, but none have been confirmed in vivo, except for the GTF2IRD1 gene itself, which has been shown to be directly regulated by its own protein product.

Aims: The aims of this thesis were to understand the molecular and cellular mechanisms of GTF2IRD1 function by investigating: i) potential target genes of

GTF2IRD1 in the brain through microarray analyses; ii) novel protein partners of

GTF2IRD1 via yeast two-hybrid screening and iii) post-translational regulation of the

GTF21RD1 protein by SUMOylation.

Results: i) A combination of gene expression profiling and bioinformatics analysis was utilised to search for genes that may be directly or indirectly regulated by GTF2IRD1.

Firstly, microarray analyses of various brain regions from the Gtf2irdl knockout mice revealed unexpectedly subtle differences in their transcriptional profiles in comparison with wild type controls. Since Gtflirdl is not ubiquitously expressed in the brain, the cellular heterogeneity of the brain regions may hamper the identification of cell-specific changes of gene regulation caused by the loss of Gtf2irdl. Gtflirdl expression overlaps significantly with GABAergic neurons, thus, we developed a way to isolate GABAergic cells. Using fluorescence-activated cell sorting, GABAergic neurons were isolated from

Gtf2irdl knockout mice that were also carrying the Gadl-GFP knockin allele, which expresses GFP under the control of the Gadl gene that encodes decarboxylase 1, the rate-limiting enzyme in GABA synthesis. Gene expression profiles from these FAC sorted GFP-positive cells did not ultimately result in the identification of any further direct gene targets of GTF21RD1 regulation but it did create a valuable reference resource that can be mined in future studies. As an alternative approach, an in silico search was performed to identify genes which contain the GTF2IRD1 recognition sequences in their upstream regions. Although, many genes were identified that contained such potential domains in a conserved fashion, none of these were found to correspond to genes that showed dysregulation in either of the microarray analyses, ii) Yeast two-hybrid screens for GTF2IRD1-interacting proteins led to the identification of several novel proteins. These proteins were subjected to a systematic analysis to determine the interaction domain in GTF21RD1. Many interactions were mapped to the repeat domains of GTF2IRD1, formerly only considered to be DNA binding domains.

This finding has an important impact for understanding the reason for the duplication of the repeat domains during chordate evolution. A subset of these interacting proteins,

SETD6, ZMYM5 and DCAF6, has been associated with histone modification pathways.

SETD6 and ZMYM5 were shown to co-immunoprecipitate and co-localise with

GTF2IRD1 in the nucleus when expressed in mammalian cells lending further support to these interactions. The functional implications of these interactions need to be investigated further but they suggest a potential role for GTF2IRD1 in the epigenetic regulation of chromatin. vii iii) Confirmation of a previously identified interaction of GTF2IRD1 with a SUMO E3 ligase (PIASX) was established and these findings were extended to the identification of a direct interaction between GTF2IRD1 and the SUMO E2 ligase UBC9 suggesting that

GTF2IRD1 may be regulated by SUMOylation. Biochemical analyses showed that

GTF2IRD1 is conjugated by endogenous SUMO at Lys-495, which lies within a highly conserved SUMO consensus motif. SUMOylation of GTF21RD1 had no detectable effect on its nuclear localisation, protein stability or transcriptional control of the

Gtf2irdl upstream region, but enhanced its interaction with the chromatin-associated protein, ZMYM5.

Conclusions: These studies have revealed some new insights into GTF2IRD1 function at the molecular level. A lack of evidence for major transcriptional change in the of Gtjlirdl knockout mice and for direct target genes other than itself, has led to a reappraisal of the initial proposed functions of GTF21RD1 as a conventional DNA- binding transcription factor. Instead, a role in chromatin remodeling is suggested by its novel interaction with proteins whose functions are associated with histone modifications. Moreover, SUMOylation of GTF2IRD1, which enhances its interaction with ZMYM5, supports the accumulating evidence for the importance of SUMOylation in chromatin biology. Lastly, epigenetic control through chromatin remodeling in the nervous system is rapidly growing as an important topic in the development of cognitive function, learning and memory, and a variety of neuropsychiatric disorders.

The work in this thesis suggests that GTF2IRD1 may play role(s) in various aspects of brain function through chromatin modification and the disruption of such mechanisms may explain the underlying cause of the neurobehavioural features of WBS.

viii ABBREVIATIONS

3-AT 3-amino-1,2,4-triazole Ade adenine ADHD attention deficit hyperactive disorder APS ammonium persulfate ATP adenosine triphosphate BSA bovine serum albumin ChIP chromatin immunoprecipitation CNS central nervous system DDO double dropout DMEM Dulbecco's modified Eagle's medium DNA deoxyribonucleic acid DNAse deoxyribonuclease ECL enhanced chemiluminescence EDTA ethylenediaminetetraacetic acid EGTA ethyleneglycol-bis(P-aminoethyl ether)-tetraacetic acid FACS fluorescence activated cell sorting GABA y-amino butyric acid GAD glutamate decarboxylase GFP green fluorescent protein GUR GTF2IRD1 upstream region HC1 hydrochloric acid HAT histone acetyltransferase HDAC histone deacetylase HKMT histone methyltransferase PI is Histidine IgG immunoglobulin G IP immunoprecipitation IPTG isopropyl p-D-thiogalactopyranoside KC1 potassium chloride kDa kilodalton Leu leucine LiAc lithium acetate Lys lysine LZ MW molecular weight NaCI sodium chloride NEM N-ethylmaleimide NLS Nuclear localisation signal ORF open reading frame PBS phosphate-buffered saline PCR polymerase chain reaction PFA paraformaldehyde PMSF phenylmethylsulfonylfluoride QDO quadruple dropout qRT-PCR quantitative real time PCR RD repeat domain RIPA radio-immunoprecipitation assay buffer RNA ribonucleic acid SDS sodium dodecyl sulphate SDS-PAGE sodium dodecyl sulphate polyacrylamide gel electrophoresis SEM standard error of the mean SIM SUMO-interacting motif SSC saline-sodium citrate buffer SUMO small ubiquitin-like modifier SVAS supravalvular aortic stenosis TBE tris-borate-EDTA buffer TEMED tetraethylmethylenediamine Trp tryptophan UV ultra violet w/v weight/volume WBS Williams-Beuren syndrome WSCP Williams syndrome cognitive profile Y2H Yeast two-hybrid

Note on gene, protein and mouse nomenclature: Gene and protein symbols are in italics and non-italics, respectively. Mouse genes are in lower case, but start in capitals. Human genes are all in capitals.

Mouse nomenclature is as per the Standardized Genetic Nomenclature for Mice (International Committee on Standardized Genetic Nomenclature for Mice). For example, the superscript letters in the name, Gtf2irdl'm,Hni', denote the first targeted mutation of the Gtf2irdl gene produced in the Hardeman laboratory. Table of Contents

ORIGINALITY STATEMENT...... ii ACKNOWLEDGEMENTS...... iii PUBLICATIONS...... iv ABSTRACT...... vi ABBREVIATIONS...... ix Table of Contents...... xi CHAPTER 1 - INTRODUCTION...... 1 1.1 Linking gene, brain and behaviour...... 1 1.1.1 Regions of the brain and their specialised functions...... 1 1.1.2 The neurons and the synapses...... 3 1.1.3 Neurotransmitters and their receptors...... 4 1.1.3.1 Glutamate receptors...... 6 1.1.3.2 GABA receptors...... 7 1.1.3.3 The excitation-inhibition balance...... 8 1.1.4 CNS development: From neurogenesis to synaptogenesis...... 9 1.1.5 Adaptive changes to neuronal circuitry during development and adulthood ...... 10 1.1.6 Gene regulation in the CNS...... 10 1.1.6.1 Transcription factors in neural development and differentiation...... 11 1.1.6.2 Transcriptional control of activity-dependent genes...... 12 1.1.7 Epigenetic regulation in the CNS...... 14 1.1.7.1 DNA methylation...... 16 1.1.7.2 Histone modifications...... 18 1.1.8 Understanding the genetic and epigenetic basis of human cognition...... 25 1.1.8.1 Single gene-cognition linkage: lessons from SNPs...... 25 1.1.8.2 Genetic and epigenetic alterations in neurocognitive disorders...... 26 Rett syndrome...... 27 Angelman and Prader-Willi Syndrome...... 27 Rubinstein-Taybi syndrome...... 28 1.2 Williams-Beuren syndrome as a model for the genetic basis of human cognition and behaviour...... 29 1.2.1 The genetic basis of WBS...... 29 1.2.2 The physical characteristics...... 31 1.2.3 The cognitive profile...... 33 1.2.4 The behavioural profile...... 35 1.2.5 Other neurological features...... 37 1.2.6 Neurobiology in WBS...... 38 1.2.6.1 Brain structure and cytoarchitectonic abnormalities...... 38 1.2.6.2 Dysfunctions of neural activities...... 39 1.2.6.3 Biochemical abnormalities...... 41 1.2.7 Duplications of the 7q 11.23 chromosomal region and the consequences of positive gene dosage imbalance...... 41 1.2.8 Linking genotype to phenotype...... 43 1.2.8.1 Genes linked to the Williams syndrome cognitive profile...... 43 1.2.8.2 Studies of atypical deletions link disruption of GTF2IRD1 and GTF21 to the main features of the WSCP...... 46 1.2.8.3 Further analysis of the genes behind the craniofacial abnormalities .48 1.3 The TFII-I protein family...... 50 1.3.1 Structural features andevolutionary relationships of members of the TFII- 1 protein family...... 50 1.3.2 TF1I-1...... 52 1.3.2.1 The proposed functions of TFII-I...... 52 1.3.2.2 Expression analysis of TFII-I...... 56 1.3.2.3 Analyses of Gtf2i knockout mice...... 57 1.3.3 GTF2IRD1...... 59 1.3.3.1 Molecular analysis...... 60 1.3.3.2 Expression analysis...... 64 1.3.3.3 Analyses of Gtf2irdl mouse models...... 66 1.3.4 GTF2IRD2...... 72 1.3.4.1 Genomic sequence and transcript analysis...... 72 1.3.4.2 Protein analysis...... 73 1.4 Aims of this thesis...... 74 CHAPTER 2 - MATERIALS AND METHODS...... 86 2.1 General Materials...... 86 2.1.2 Protein Biochemistry...... 86 2.1.3 Cell Biology...... 87 2.1.4 Yeast reagents...... 87 xii 2.1.5 Plasmids...... 87 2.1.6 Antibodies...... 90 2.2 Molecular Biology...... 91 2.2.1 Agarose gel electrophoresis...... 91 2.2.1 Polymerase chain reaction amplification...... 91 2.2.2 Genomic DNA extraction from mouse tail samples...... 92 2.2.3 Purification of DNA from gel or solution...... 92 2.2.4 Ligation of DNA fragments...... 92 2.2.5 Transformation of E. coli...... 93 2.2.6 Purification of plasmid DNA...... 93 2.2.7 Restriction endonuclease digestion...... 94 2.2.8 Klenow fill-in reaction...... 94 2.2.9 Site directed mutagenesis...... 94 2.2.10 DNA sequencing...... 95 2.2.11 Total RNA extraction...... 96 2.2.12 RNA-formaldehyde gel electrophoresis...... 96 2.2.13 lllumina array analysis...... 97 2.2.14 Northern Blotting...... 98 2.2.15 Reverse transcription (first strand cDNA synthesis)...... 99 2.2.16 Quantitative real time-PCR...... 100 2.3 Protein Biochemistry...... 100 2.3.1 SDS-polyacrylamide gel electrophoresis...... 100 2.3.2 Western blotting...... 101 2.3.2.1 Reprobing blots...... 102 2.3.2.2 Densitometry analysis of blots...... 102 2.3.3 Total protein extraction from tissues or cells...... 103 2.3.4 Extraction of histone proteins from tissues...... 103 2.3.5 BCA assay...... 104 2.3.6 Immunoprecipitation...... 104 2.4 Cell Biology...... 105 2.4.1 FACS-array experiment...... 105 2.4.1.1 Animals...... 105 2.4.1.2 Tissue and disaggregated cells sample preparation...... 106

xiii 2.4.1.3 FACS / RNA extraction...... 107 2.4.2 Cell culture...... 107 2.4.3 Transient transfection...... 108 2.4.3.1 GeneJuice...... 108 2.4.3.2 Calcium phosphate transfection...... 108 2.4.4 Immunofluorescence...... 109 2.4.5 Protein degradation rate analysis...... 109 2.4.6 Luciferase reporter assay...... 110 2.5 Yeast two-hybrid assay...... 110 2.5.1 Small-scale yeast transformations...... 110 2.5.2 Y2H library screen...... Ill 2.5.3 Plasmid rescue from yeast...... 113 2.5.4 Identification of library clones...... 114 CHAPTER 3 - GENE EXPRESSION ANALYSIS OF GTF2IRD1 KNOCKOUT MOUSE BRAIN...... 118 3.1 Overview of experimental approaches...... 119 3.2 Brain region-specific microarray analysis...... 121 3.2.1 Comparison of gene expression profiles from different brain regions ... 122 3.2.2 Analysis of the differentially expressed genes by qRT-PCR...... 123 3.2.3 Analysis of transcript and protein from the Actl6b gene...... 124 3.2.4 Actl6b and genes closely linked to the mutant Gtf2irdl retain an un­ recombined 129Sv/J genomic region...... 124 3.2.5 Discussion of the brain region-specific microarray analyses...... 126 3.3 Isolation of GABAergic neurons from the Gtf2irdl KO mouse brain and mRNA microarray analysis profiles...... 127 3.3.1 Mice used for GABAergic neuron isolation...... 128 3.3.2 Optimisation of the FACS isolation of GABAergic neurons...... 128 3.3.3 Transcript profile of isolated GABAergic neurons from Gtf2irdl KO and WT mice...... 131 3.3.4 Discussion of GABAergic neuron-specific microarray analyses....134 3.4 In silico analysis of regulatory regions containing the putative GTF2IRD1 binding sites...... 135 3.5 Overall discussion...... 136 CHAPTER 4 - IDENTIFICATION OF NOVEL PROTEIN PARTNERS OF GTF2IRD1 ...... 154 xiv 4.1 Novel protein interactions identified in yeast two-hybrid library screening .. 156 4.2 Mapping the interacting domains of GTF2IRD1...... 158 4.3 Proteins involved in histone modification...... 160 4.3.1 SETD6...... 161 4.3.2 ZMYM5...... 162 4.3.3 Preliminary analyses of global histone H3K14ml and H4K12mel methylation patterns in brain regions of Gtf2irdl KO mice...... 164 4.4 Proteins involved in other pathways...... 165 4.4.1 ARMCX5...... 165 4.4.2 KPNA2 (karyophrenin or -a2)...... 167 4.5 Discussion...... 169 CHAPTER 5 - GTF2IRD1 IS SUMOYLATED IN VIVO...... 187 5.1 Interaction of GTF2IRD1 with SUMO E2 and E3 ligases...... 191 5.2 GTF2IRD1 is SUMOylated in vivo...... 192 5.3 GTF2IRD1 is SUMOylated at Lys-495 contained within the evolutionarily well-conserved SUMO motif...... 194 5.4 SUMOylation and subcellular localisation of GTF2IRD1 ...... 195 5.5 SUMOylation and GTF2IRD1 protein stability...... 196 5.6 Transcriptional activity of GTF2IRDI is independent of SUMOylation...... 197 5.7 GTF2IRD1 interaction with ZMYM5 is modulated by SUMOylation...... 198 5.8 Discussion...... 199 CHAPTER 6 - OVERVIEW AND FUTURE DIRECTIONS...... 219 6.1 Overview...... 219 6.1.1 Changes in the gene transcriptional profile of Gtf2irdl knockout mice.219 6.1.2 Novel protein partners of GTF2IRD1...... 221 6.1.3 GTF2IRD1 as a target for SUMOylation in vivo...... 223 6.2 Future Directions...... 224 REFERENCE LIST...... 231 APPENDICES...... 270

xv CHAPTER 1 - INTRODUCTION

1.1 Linking gene, brain and behaviour

The brain is the most complex organ in the human body, and expresses approximately

80% of all the genes in the genome (Chaudhari & Hahn, 1983; Johnson et al, 2009;

Lein et al, 2007). The brain is also made up by the largest diversity of cell types of any organ. These include broad classes of neuronal and non-neuronal or glial cells, each with highly specific gene expression profiles. Neurogenetics research aims to link the genetic basis of neurobiology to the function and morphology of the brain, towards understanding the mechanisms that underpin the normal and diseased functions of the brain (Walter et al, 2009).

1.1.1 Regions of the brain and their specialised functions The human brain is divided into three major structural regions based on its embryonic development: the forebrain (prosencephalon), the midbrain (mesencephalon) and the hindbrain (rhombencephalon).

The forebrain consists of the cerebral cortex, the limbic system, and the basal ganglia.

The cerebral cortex, which makes up the largest part of the human brain, regulates mental activities and higher cognitive functions. The appearance of the cerebral cortex in human and other large is characterised by its surface pattern comprised of sulci (the grooves) or fissures (for deeper grooves) and gyri (the bumps) (Figure 1.1 .A).

The outer cerebral hemispheres are divided into four lobes named for the overlying cranial bones: the frontal, parietal, occipital, and temporal lobes (Figure 1.1.C). The frontal lobe is mainly involved in the control of behaviour and movement, the parietal in somatic sensation and spatial cognition, the occipital in vision, and the temporal in auditory activities as well as learning, memory and emotion. The limbic system is a group of interconnected structures in the centre of the cerebral hemisphere and is associated with olfaction, emotion, motivation, behaviour and various autonomic functions which control the body physiology (Figure 1.1.B). The hippocampus is important in memory and learning of specific external stimuli. The thalamus relays most of the information reaching the cerebral cortex from the rest of the central nervous system (CNS). The hypothalamus controls behavioural, endocrine and autonomic functions. The amygdala is well known for its role in emotional and fear conditioning. Although not prominently known as part of the limbic system, the olfactory system responsible for odour recognition is heavily interconnected with parts of the limbic system as the olfactory tracts (axons from the olfactory bulb) project to the hypothalamus and the olfactory cortex (made up of the piriform cortex and the amygdala).

The basal ganglia are large subcortical structures comprising several interconnected nuclei (clusters of functionally similar groups of neurons) in the forebrain, midbrain and thalamic regions, including the striatum (caudate putamen, nucleus accumbens), the subthalamic nucleus, the globus pallidus, and the substatia nigra. They are thought to be involved in motor functions that coordinate the movements of the limbs.

The midbrain and the hindbrain are part of the brainstem, which extends from the lower areas of the forebrain to the spinal cord (Figure 1.1.B). The midbrain which connects the hindbrain and the forebrain is involved in auditory and visual responses and motor function. The hindbrain extends from the spinal cord and is composed of the metencephalon (pons and cerebellum) and myelencephalon (medulla oblongata). The cerebellum is especially important in processing various kinds of motor functions such as motor coordination and learning, while considerable evidence also suggests a role in cognition and emotion (Kandel et al, 1991; Kolb & Whishaw, 2001; Purves et al, 2004;

Timmann et al, 2010).

2 1.1.2 The neurons and the synapses

Neurons are the smallest functional units of the brain. They generate active electrical signals in order to communicate with each other. Each typically has three morphologically defined regions: (1) the cell body {soma or perikaryon), which contains the nucleus and the major cytoplasmic organelles; (2) dendrites, which emanate from the cell body to receive inputs; and (3) the axon, a fine tubular process that is responsible for transmitting neural information out of the cell body. Depending on the neuronal type, the dendritic tree or arborisation may have a characteristic shape as well as varying number of spines (small protrusions) along the dendritic shaft, such that some dendrites appear spiny (as in pyramidal cells) or non-spiny (as in most interneurons) (Figure 1.1.D). The dendritic spines (also known as ‘spine heads") substantially increase the number of possible contacts between neurons. The axon is generally smooth and in most cases, extends much farther from the cell body than the dendritic arbor. In vertebrates, many axons are insulated by a surrounding myelin sheath which facilitates rapid impulse conduction (Squire, 2008).

Based on the type of neurotransmitter they release, neurons can be classified either as excitatory glutamatergic or inhibitory GABAergic neurons (Figure 1.1.D). About 80% of all cortical neurons are pyramidal cells which release the excitatory neuroransmitter glutamate (Peters & Kara, 1985). Pyramidal neurons are the main projection neurons of the cortex and extend their axons to other areas in the brain (Contamina & Boada, 1992;

Miller et al, 1993). The remaining 20% of the cortical neurons release the inhibitory neurotransmitter y-aminobutyric acid (GABA) and are therefore called GABAergic interneurons (DeFelipe, 2002). The axonal projections of cortical interneurons are limited to the cortex only, hence, these neurons are also known as the inhibitory local circuit neurons. Cortical interneurons have been classified according to their axonal or dendritic morphology (for instance, basket, chandelier, bitufted and bipolar cells), their intrinsic neuronal firing properties, their network interconnections and the neurochemical markers, such as neuropeptides (17% express somatostatin, 1-2%

3 express neuropeptide Y) and calcium-binding proteins that they express (Figure 1.1.D;

DeFelipe, 1993; Jones, 1981; Markram et al, 2004; Obst & Wahle, 1997). The great diversity of the inhibitory interneurons that make up a relatively small fraction of the total cortical neurons could be crucial for optimising the 'task-specific' microcircuit in the cortex (Markram et al, 2004).

On average, each of the 100 billion neurons in the human brain makes around 10,000 synaptic contacts at which a series of electrical and chemical signals are transmitted through the connected networks. The highly specialised site of contact between neurons is called the synapse. Each synapse is composed of (1) a presynaptic (axonal) element, separated by (2) a cleft to a closely opposed (3) postsynaptic element on a dendrite or cell body. The presynaptic element is characterised by the swellings ('boutons') found either along the axons or at their tips. A concentrated amount of synaptic vesicles are contained in the presynaptic element, also known as the ‘active zone', with a small number ready to dock and fuse at the plasma membrane upon activation by an action potential resulting in the release of neurotransmitter into the synaptic cleft (Figure

1.1 .E; Dresbach et al, 2001; Garner et al, 2000). The postsynaptic reception apparatus is highly identifiable at the ultrastructural level by the presence of an electron-dense thickening called the postsynaptic density (PSD). The PSD contains a cluster of postsynaptic receptors and ion channels at high density (Scannevin & Huganir, 2000).

Dendritic spines are enriched with elements of the PSD; variable spine shape and volume is thought to be correlated with the strength and maturity of each spine-synapse, and has long been hypothesised as the basis of memory (Bailey & Kandel, 1993; Yuste

& Bonhoeffer, 2001).

1.1.3 Neurotransmitters and their receptors

Neurotransmitters are the chemicals that transmit signals between neurons by moving across synapses. In order to be classified as a neurotransmitter, a molecule must: (i) be synthesized in the neuron, (ii) be available and released in sufficient quantity to produce

4 an effect on the postsynaptic neuron, (iii) have the same effect when externally administered as when acting endogenously, and (iv) have a biochemical mechanism for inactivation.

The process of chemical transmission of a neurotransmitter involves (1) synthesis in the presynaptic neuron, (2) storage in a synaptic vesicle in the presynaptic terminal, (3) release into the synaptic cleft upon stimulation or action potential firing, (4) interaction with the target receptors on the postsynaptic neuron, and (5) rapid termination of its synaptic action by diffusion, enzymatic degradation, or re-uptake of the transmitter

(Figure 1.1.E).

The three major categories of neurotransmitters are amino acids, monoamines, and peptides. Amino acids and monoamine neurotransmitters are called small-molecule neurotransmitters and are also known as the classical neurotransmitters. Amino acid neurotransmitters include glutamate, y-aminobutyric acid (GABA) and glycine.

Glutamate and GABA have been described as the major “workhorse” neurotransmitters as they are the major excitatory and inhibitory neurotransmitters in the brain, respectively. Examples of monoamine neurotransmitters include acetylcholine, dopamine, epinephrine, norepinephrine, histamine, and serotonin. A large class of peptide transmitters (neuropeptides) have also been identified to date, including substance P, vasopressin, somatostatin, prolactin, and neuropeptide Y (Burbach, 2010).

Both glutamate and GABA mediate their actions by activating ionotropic or metabotropic receptors. Ionotropic receptors or ligand-gated ion channels consist of a binding site for a neurotransmitter and a channel that allows the movement of ions across the membrane. Metabotropic receptors are linked to a G-protein coupled to the which can either regulate an ion channel or an enzyme by activating a series of biochemical steps (Kolb & Whishaw, 2001).

5 1.1.3.1 Glutamate receptors

Three subclasses of ionotropic glutamate receptors are named after their selective agonist, AMPA, NMDA, and kainate receptors. Individual receptors are multimeric assemblies of subunits, and a total of 16 functional subunits (transcribed from 16 genes) are known to make up the ionotropic glutamate receptor family (Tolle et al, 1993). The metabotropic glutamate receptors consist of at least eight different subtypes (mGluRl- mGluR8).

NMDA receptors (GluNl, GluN2A-GluN2D and GluN3A-B) are known to be partly responsible for aspects of development, learning and memory, and neuronal dysfunction due to brain injury. A unique property of NMDA receptors is their Ca“ permeability; by permitting a significant influx of Ca“ , the activation of calcium signalling cascades are triggered, further affecting a variety of processes in the neuron. Excess Ca" is also toxic to neurons, so hyperactivity of NMDA receptors has been shown to contribute to a variety of neurodegenerative disorders (Hardingham, 2009). The role of NMDA receptors in the activity-dependent increase (long-term potentiation or LTP) or decrease (long-term depression or LTD) of synaptic strength in the hippocampus has been studied extensively (reviewed in Citri & Malenka, 2008). These synaptic plasticity phenomena have been proposed as one of the underlying mechanisms in learning and memory processes (reviewed in Malenka & Bear, 2004). Similarly, AMPA receptors

(GluA 1-GluA4), which mediate the majority of fast synaptic transmissions also play crucial roles in LTP and LTD (Malinow & Malenka, 2002). Kainate receptors (GluKl-

GluK5) have been shown to control neurotransmission where at the presynaptic terminal they modulate both inhibitory and excitatory neurotransmitter release, and at the postsynapse, they mediate excitatory neurotransmission (Bleakman, 1999; Lerma,

2006).

Hypofunctional glutamatergic transmission is one of the proposed mechanistic causes of schizophrenia. There is substantial evidence that the glutamate level in cerebrospinal

6 fluid is decreased by as much as 50% and there is a significant increase in the mRNA level of the GluN2D subunit of the NMDA receptor in the prefrontal areas. This is thought to be a compensatory response to the decreased glutamatergic function in the prefrontal cortical regions of schizophrenic brains (Akbarian et al, 1996).

1.1.3.2 GABA receptors

GABA receptors which respond to the neurotransmitter GABA can be divided into

GABAa receptors (ionotropic), GABAb receptors (metabotropic), and the more recently identified GABAc receptors (ionotropic) which resemble GABAa receptors.

GABAa receptors consist of a pentameric assembly of distinct subunits from twenty- one known subunits (al-6, pi-4, yl-4, 5, s, 7i, 0, pl-3). The majority of GABAa receptors are co-assemblies of two a-, two P-, and one y-subunit. Some GABAa subunits are expressed during embryonic development (a2, a3, and a5); whereas, others are expressed in the early period postnatal or in the adult brain (Lujan et al, 2005; Vicini & Ortinski, 2004). Although the significance of the differentially expressed

GABAa receptor subunits is not completely understood, subunit switching during development in various regions of the brain has been documented and implicated in establishing the proper functional GABAergic transmission (Davis et al, 2000; Fritschy et al, 1994; Liu & Wong-Riley, 2006; Ogris et al, 2006). Mechanistically, these ionotropic receptors form channels permeable to chloride ions (Cl ) and gating of these channels permits Cl influx into the cell, which hyperpolarises the membrane. Opening of these channels also increases the resting membrane conductance, thus shunting any excitatory current Bowing into the cell. Two important classes of anxiolytic drugs, benzodiazepines and barbiturates, both bind to GABAa receptors and enhance the Cl" influx through these channels in response to GABA (Frazer et al, 1994).

GABAa receptors have been shown to underlie sedation, sleep induction, and myorelaxation, as well as anxiety, seizures, and amnesia. Disorders linked with an

7 altered GABAAergic system, or which can be efficiently treated with GABAAergic drugs, include alcoholism, Angelman's syndrome, anxiety disorders, autism, depression, mania, premenstrual syndrome, schizophrenia and sleep disorders (Korpi &

Sinkkonen, 2006).

GABAb receptors are voltage-gated channels (potassium or calcium) that work with intracellular G-proteins and hence, mediate a slower inhibition than GABAa receptors.

Two subtypes of the receptors, GABAbi and GABAb2, assemble as heterodimers and link through their intercellular C-termini (Kammerer et al, 1999; Kuner et al, 1999). impaired GABAb receptor function is associated with pathophysiological effects including pain, epilepsy, spasticity and cognitive deficits (Couve et al, 2000). GABAb receptor agonists, for example baclofen, display a number of pharmacological effects acting through the central and peripheral nervous systems including antispasticity, anti­ nociception, epileptogenesis and anti-bronchoconstrictor activity (Bowery et al, 2002).

1.1.3.3 The excitation-inhibition balance The effect of neurotransmission can be either excitatory or inhibitory, depending on whether the resulting response in the postsynaptic neuron is driven towards a point above or below its firing threshold. Excitatory synaptic transmission is driven mainly by glutamatergic synapses; whereas, inhibitory synaptic transmission involves GABAergic and glycinergic signaling. A functional balance between excitatory and inhibitory system, through appropriate expression and function of neurotransmitters and their receptors, is established during development and maintained throughout life (Cline,

2005; Lujan et al, 2005; Turrigiano & Nelson, 2004). Deregulation of this balance is associated with a number of nervous system disorders such as epilepsy, mental retardation and autism (Eichler & Meier, 2008).

8 1.1.4 CNS development: From neurogenesis to synaptogenesis

Cells in the nervous system arise from the neural ectoderm which contains multipotent precursor populations which, in turn differentiate into neurons and microglial cells

(astroyctes and oligodendrocytes). Key events during CNS development include the process of neurogenesis or the birth of neurons (also takes place in the adult mammalian

CNS), followed by neuronal proliferation, migration, differentiation and finally, synapse formation (synaptogenesis) and circuit refinement (reviewed in Gotz & Huttner, 2005;

Ming & Song, 2005),

There is now convincing evidence that separate pools of progenitor cells in distinct regions of the brain give rise to different types of neurons and that the function of these neurons is determined by their final resting place after migration (Kriegstein & Noctor,

2004; Marin & Rubenstein, 2001; Sidman & Rakic, 1973). This is consistent with the developmental formation of the two distinct major neuronal populations in the cortex.

The glutamatergic pyramidal neurons result from a population of radially migrating neurons that originate in the ventricular zone of the pallium or the neocortex. The second population consists of tangentially migrating neurons that originate in the ventricular zone of the subpallium that gives rise to inhibitory GABAergic interneurons

(Jimenez et al, 2002). Disordered neuronal migration is known to underlie several human brain disorders affecting cortical development. Collectively, such conditions account for up to one third of severe epilepsy cases, as well as specific disorders associated with severe mental retardation (Brodtkorb et al, 1992; Palmini et al, 1991).

Neuronal circuit refinement is initiated by the process of synaptogenesis. In principle, synaptogenesis takes place in two consecutive steps. First, the initial contact between an axonal growth cone and a specific target cell is made. Given the large number of different neuronal cell types in the brain, a complex mechanism must exist to mediate target cell recognition. Secondly, specific pre- and postsynaptic protein components are recruited to the initial contact site, forming a functional synapse. Both processes seem to

9 be driven to some degree by various classes of cell-adhesion molecules which so far have been shown to include at least three different systems: the cadherin and the cadherin-like neuronal receptors, the ephrin-Eph receptors, and the neurexin-neuroligin system (Brose, 1999; Waites et cil, 2005).

1.1.5 Adaptive changes to neuronal circuitry during development and adulthood

From the early postnatal period through to old age, sensory experiences lead to formation of long-lasting memories and experience-induced adaptations. The cellular mechanisms that underlie human responses to sensory experiences can be explained by experience-driven changes in neuronal connectivity. A large number of studies have provided evidence that neuronal activity plays a continuing role in the dynamic processes of (i) dendritic outgrowth, (ii) synaptic maturation, (iii) synapse elimination and (iv) synaptic plasticity (Flavell & Greenberg, 2008; Saneyoshi et al, 2010). The molecular basis underpinning these morphological and functional changes involves the activation of calcium-sensitive signalling cascades. Sensory input initiates neurotransmitter release at synapses which leads to membrane depolarisation and calcium influx into the neurons. In recent years there has been extensive effort aimed at understanding how the rise in intracellular calcium level influences the function of several transcription factors, which in turn control the expression of a large number of neuronal activity-regulated genes (see next section).

1.1.6 Gene regulation in the CNS

During neural development and differentiation a highly regulated genetic program regulated by multiple transcription factors unfolds to ensure proper activation or inactivation of genes in the appropriate cell type. As the complexity of the brain develops into adulthood, neuronal cells undergo constant adaptive changes in response to extrinsic environmental cues and signalling. Much of this molecular neuronal plasticity is brought about by a group of activity-regulated transcription factors.

10 1.1.6.1 Transcription factors in neural development and differentiation

Numerous studies investigating molecular mechanisms of neuronal cell type specification have agreed upon a key principle, that is, transcription factors act in combination to determine cell fate and establish the neural network (Lee & Pfaff, 2001;

Ma, 2006). Cell fate specification towards either neural stem cell maintenance or differentiation into post-mitotic neurons or glia is achieved by the balance of expression of antagonistic transcription factors (reviewed in Bertrand et al, 2002; Sugimori et al,

2007). Expression of these key determinant factors in neural stem cells has been experimentally shown to have the capacity to convert the stem cells into specific cell types.

Specification of neuronal identity from progenitor cells into either a glutamatergic

(excitatory neuron) or GABAergic (inhibitory neuron) phenotype is mediated by pro- neural transcription factors, many of which belong to the homeodomain and basic helix- loop-helix (bHLH) classes of transcription factors (Anderson et al, 1997; Bertrand et al,

2002; Lee et al, 1995; Schuurmans & Guillemot, 2002). The , a bHLH family of proteins, drive the specification of glutamatergic pyramidal neurons in the cerebral cortex (Schuurmans et al, 2004). TLX1 and TLX3 (T-cell leukemia translocation protein 1 and 3) promote specification of glutamatergic neurons and antagonise the function of LBXI (ladybird-like homeobox 1), which promotes GABAergic differentiation in spinal cord neurons. Examination of brain development in 77x5 knockout mice showed increased GABAergic differentiation, while in Lbxl knockout mice, the presumptive GABAergic neurons were transformed into glutamatergic neurons. Interestingly, normal glutamatergic differentiation is restored in mice that lack expression of both TLX3 and LBXI, suggesting that LBXI defines a basal GABAergic differentiation state, and TLX3 acts to antagonise LBXI to promote glutamatergic differentiation (Cheng et al, 2004; Cheng et al, 2005). Another bHLH-containing protein MASH1 is indispensable for appropriate specification of

GABAergic neurons in the ventral telencephalon (Berninger et al, 2007; Casarosa et al, 1999; Parras et al, 2002); while PTF1A (pancreas-specific transcription factor la)

defines the GABAergic neurons in the cerebellum (Hori et al, 2008). In the absence of

PTF1A, only glutamatergic neurons are formed in the cerebellum; whereas, ectopic

expression of PTF1A in glutamatergic precursors is sufficient to switch neurons to the

GABAergic phenotype (Hoshino et al, 2005). In addition, the Dlx (distal-less

homeobox) genes, which are important for basal ganglia differentiation, are also

essential for the development, as well as the functional longevity of neocortical,

olfactory bulb and hippocampal GABAergic interneurons (Anderson et al, 1999; Cobos

et al, 2005).

Some neuronal development programs rely on more complex inductive interactions. For

example, the association of neurogenin-2 with OLIG2 in the spinal cord promotes motor

neuron differentiation. The pro-GABAergic specification factor MASH1 is also found

to be essential for the birth of serotonergic neurons in the hindbrain as well as

noradrenergic neurons (Hirsch et al, 1998; Pattyn et al, 2004).

It is therefore clear that the establishment of excitatory, inhibitory and other transmitter-

specific neural phenotypes is heavily controlled by a set of temporospatial-specific

transcription factors that pattern the developing brain.

1.1.6.2 Transcriptional control of activity-dependent genes

In brain, activity-dependent genes are defined as those whose transcription can be triggered by extracellular stimuli. The first discovery of such genes was made through

observing the rapid induction of c-fos proto-oncogene transcription upon addition of growth factors in 3T3 fibroblast (Greenberg & Ziff, 1984). The increased transcription of c-fos occurs within approximately 5 minutes and becomes undetectable within 30

minutes later. The term ‘immediate early gene’ (IEG) is given to genes that are activated rapidly and transiently without requiring new protein synthesis, such as c-fos

(Flavell & Greenberg, 2008). The expression of c-fos and other IEGs is now routinely

12 used to mark neurons that have been recently electrically activated and is thus, a powerful technique for imaging regions of the brain utilised in behavioural tasks

(Kaczmarek & Chaudhuri, 1997; Perez-Villalba et al, 2005; Pothuizen et al, 2009). In contrast, genes that are induced or repressed more slowly (within hours) and are dependent on new protein synthesis have been described as ‘late-response genes' (Hong et al, 2004). These activity-dependent genes play a direct role in regulating synaptic function in neurons and their activation in the brain often leads to several forms of behavioural, learning and memory deficits in animals (Brown et al, 1996; Fleischmann et al, 2003; Korte et al, 1995; Tzingounis & Nicoll, 2006). Therefore, transcriptional regulators of these genes have also become the subject of intense investigation. The

CREB (cAMP response element binding protein) and (myocyte enhancer factor-

2) protein families were amongst the first identified and now stand out as the prototype of activity-dependent transcription factor systems.

cAMP response element binding protein (CREB) CREB was isolated as a protein that bound to the cAMP response element in the somatostatin (Stt) gene promoter and to the calcium response element (CRE) in the c-fos promoter (Gonzalez & Montminy, 1989; Sheng et al, 1988). Gene regulation by CREB in response to neuronal activity is pivotal in mediating long-term memory and synaptic plasticity (Benito & Barco, 2010; Impey et al, 1996). Calcium influx through NMDA receptors or L-type calcium channels activates calcium-dependent kinase CamKIl or

CamKIV, which phosphorylates CREB on Ser-133 (Gonzalez & Montminy, 1989;

Sheng et al, 1991). The phosphorylated CREB binds to coactivator CBP/p300, which has an intrinsic histone acetyltransferase activity (discussed in section 1.1.7.2) and activates the transcription of its downstream target genes. Many CRE-containing target genes have been identified that are regulated by CREB, including BDNF which encodes brain-derived neurotrophic factor, a key player in neuronal plasticity (Tao et al, 1998).

Numerous studies have elucidated the role of BDNF in a large number of distinct processes during neural development (reviewed in Chourbaji et al, 2010; Cunha et al,

13 2010; Santos et al, 2010). To date, extensive work supports the diverse and complex functions of CREB and its closely related factors CREM (cAMP response element modulator) and ATF-1 (activating transcription factor 1), ranging from development to plasticity to disease (reviewed in Lonze & Ginty, 2002).

Myocyte enhancer factor 2 (MEF2)

Family members of MEF2 (A-D) were first identified as regulators of muscle-specific gene expression. Subsequently, they were found to also regulate IEGs such as nur77 and c-jun in diverse cell types (reviewed in McKinsey et al, 2002). In the brain, they regulate many activity-regulated genes with diverse functions at synapses including Arc,

Homerla and BDNF (Flavell et al, 2006). The genes encoding the MEF2 family of transcription factors are themselves negatively regulated by class II HDACs. Upon membrane depolarisation and calcium influx, these HDACs are phosphorylated by

CamKs and exported out of the nucleus, relieving MEF2 repression and allowing for the activation of MEF2-dependent transcription (Chawla et al, 2003; McKinsey et al,

2002). In addition, the activity of the MEF2 protein is regulated by calcium-dependent post-translational modifications including , SUMOylation and acetylation (Shalizi et al, 2006). By eliciting the production of the synaptic proteins Arc and Homerla, MEF2 negatively regulates the number of excitatory synapses in hippocampal neurons (Flavell et al, 2006). The mechanism by which MEF2 regulates genes in the brain is similar to its mechanism in muscle, and this is logical since both brain and skeletal muscle are distinct forms of electrically excitable tissues (reviewed in

Shalizi & Bonni, 2005).

1.1.7 Epigenetic regulation in the CNS

In various biological systems, and particularly in the highly plastic brain, the dynamic of epigenetic regulation is rapidly being recognised as the mechanism which translates environmental stimuli into changes in cell function. In general, “epigenetic changes” refer to the stable and heritable modifications of chromatin (the DNA and its

14 associated histone proteins) that are independent of the underlying DNA sequence, which help determining the phenotypic traits of cells (Berger et al, 2009). However, this term is generally used in a broader sense when studying of epigenetic processes in the

CNS since fully differentiated neurons are post-mitotic and hence, epigenetic modifications are seldom ‘heritable’ (Franklin & Mansuy, 2010; Riccio, 2010).

In recent years a wealth of new insights into epigenetic processes has been gained from investigating epigenetic ‘marks' and molecular machineries that regulate chromatin.

Chromatin is the state in which DNA is packaged within the cell. The fundamental unit of chromatin is the nucleosome which is composed of an octamer of the four core histones (H3, H4, H2A, H2B) around which 147 base pairs of DNA are wound.

Chromatin is extremely flexible and capable of varying the accessibility of transcriptional machinery to the DNA. Heterochromatin, a highly condensed form, is largely inaccessible and thus correlates with inhibition of gene expression. In contrast, euchromatin represents a more open form and is associated with activation of gene expression. Changes in chromatin structure are achieved by three distinct mechanisms:

ATP-dependent chromatin remodelling, DNA methylation and histone modification

(Kouzarides, 2007). The next section focuses on the basic mechanisms of DNA methylation and histone modification which are highly interdependent and often recruit

ATP-dependent chromatin remodelling to assert their functional impact on chromatin architecture (Figure 1.2) (Cedar & Bergman, 2009). The dynamic nature of epigenetic modifications has recently generated intense interest in the field of neuroscience as a candidate explanation for the phenomenon of experience-regulated neuronal plasticity.

Accumulating evidence suggests that both DNA methylation and histone modification link extracellular signals to neuronal gene regulation, and ultimately, to cellular function. Their importance is highlighted by the cognitive and behavioural defects that occur in mouse models as well as human disorders when a component of the

DNA/histone modifying machinery is perturbed (Franklin & Mansuy, 2010).

15 1.1.7.1 DNA methylation

DNA methylation is generally considered the most stable epigenetic mark and hypermethylation of DNA is associated with gene transcriptional repression. DNA methylation involves addition of a methyl group on the 5' position of cytosine in CpG dinucleotides (cytosine next to guanine separated by a phosphate). In human somatic cells methylated cytosine accounts for 1% of total DNA bases, equivalent to 70-80% of all CpG dinucleotides in the genome (, 1986; Ehrlich et al, 1982). CpG islands, clusters of CpG motifs which are found in the promoters of approximately 60% of human genes, are generally protected from methylation (Antequera & Bird, 1993; Bird,

1986). Many CpG islands are located in genes that have a tissue-restricted expression pattern and during development a small, but significant proportion of CpG islands become methylated which normally leads to a stably silent promoter. Such developmentally programmed CpG-island methylation is involved in genomic imprinting and inactivation (Bird, 2002; Bogdanovic & Veenstra, 2009;

Robertson, 2005). In cancer cells, global hypermethylation of CpG islands is thought to underlie the epigenetic inactivation of tumor supressor genes and thus is regarded as one of the molecular bases of cancer biology (reviewed in Sincic & Herceg, 2010).

DNA methylation is catalysed by DNA methyltransferases (DNMTs) and involves the recruitment of methyl-CpG binding domain proteins (MBPs). Members of the DNMT family include DNMT1, which functions to maintain the DNA methylation pattern and

DNMT3A and DNMT3B which function as de novo methyltransferases (Figure 1.2.A;

Jeltsch, 2006). DNA methylation is now known to be much more dynamic than it was thought previously; CpG methylation and demethylation have been shown to cycle within relatively short time scales of approximately 100 minutes (Kangaspeska et al,

2008; Metivier et al, 2008). Demethylation involves deamination of cytosine bases followed by replacement of the ‘damaged' cytosine with an unmethylated cytosine by members of the base excision repair system (Gehring et al, 2009; Niehrs, 2009). Despite an intense search, DNA demethylases have yet to be identified.

16 DNA methylation in neurons

Genes encoding the DNA methyltransferases DNMT1 and DNMT3A are expressed at high levels in the developing and adult nervous system (Feng et al, 2010). Conditional knockout mice that lack both genes showed abnormal long-term plasticity in the hippocampal CA1 region together with deficits in learning and memory. Normally these genes are upregulated in the hippocampus following contextual fear conditioning (the paradigm of hippocampal-dependent associative memory). Administration of DNMT inhibitors zebularine and 5-aza-2-deoxycytidine blocks memory consolidation and hippocampal LTP (Levenson et al, 2006; Miller & Sweatt, 2007). Thus, accumulating evidence supports the role of DNA methylation in learning and memory (reviewed in

Day & Sweatt, 2010).

Analysis of DNA methylation in the mouse brain upon administration of a DNMT general inhibitor has identified alterations in the methylation status of DNA in the regulatory regions of the neuronal plasticity-promoting genes reelin and BDNF

(Levenson et al, 2006). BDNF is also recognised as a principle target of MeCP2, a methyl-CpG-binding domain protein (Chen et al, 2003; Martinowich et al, 2003). DNA methylation represses transcription in part by recruiting the binding of MBPs. MeCP2 loss-of-function and gain-of-function knockout and transgenic mouse models have perturbations in BDNF as well as other gene transcriptional consequences, and have altered neurological phenotypes (see later section on Rett syndrome; Chahrour et al,

2008). Deficiency of MeCP2 in neurons causes a widespread increase in histone H3 acetylation, suggesting that rather than acting as a gene-specific transcriptional , MeCP2 recruits co-repressor complexes leading to repression of histone acetylation in regions that are rich in methylated DNA (Shahbazian et al, 2002; Skene et al, 2010).

In the hippocampus, neural activity also raises intracellular GADD45b (growth arrest and DNA-damage-inducible, beta) which has been shown to possess DNA demethylase

17 activity in vitro (Ma et al, 2009). In this study, demethylation of the BDNF and Fgf-1

(fibroblast growth factor-1) promoters was shown to be dependent on GADD45b in vitro; although, others suggest that GADD45b acts as a cofactor that promotes DNA demethylation through DNA repair (Barreto et al, 2007; Rai et al, 2008). Mice that lack

Gadd45b showed a significant reduction in activity-dependent adult neurogenesis in vivo and a deficit in dendritic growth in vitro, further evidence that demonstrates the importance of DNA methylation in neurons (Ma et al, 2009).

1.1.7.2 Histone modifications

While the core histones are predominantly globular, their N-terminal “tails” are unstructured and contain multiple target sites for post-translational modifications

(PTMs). At least eight distinct types of modifications have been found on histones, including lysine acetylation, lysine methylation and serine//tyrosine phosphorylation, which are the best-studied of the PTM types (Kouzarides, 2007)

(Figure 1.2.B). Most of these modifications are known to be dynamic and histone­ modifying enzymes that add and remove these modifications have been identified. The multiple PTMs mapped on the same or different histone tails have shown extensive crosstalk, acting either in a synergistic or antagonistic manner (Jenuwein & Allis, 2001).

This led to the ‘histone code' hypothesis which predicts that PTMs on histones, alone or in combination, function to direct specific and distinct DNA-templated programs.

Although the best characterised functions of histone modifications are in gene transcriptional control, they are also involved in other biological tasks such as DNA repair, DNA replication and chromosome condensation (Kouzarides, 2007). In general, there are two proposed mechanisms by which a histone modification may exert its function: (i) by affecting higher-order chromatin structure or (ii) by recruiting non­ histone proteins which have the ability to further modify chromatin via their enzymatic activities such as ATP-dependent chromatin remodelling (Kouzarides, 2007). Global analyses of histone modifications in different biological systems have been greatly

18 aided by the use of modification-specific antibodies in chromatin immunoprecipitations

(ChIP) coupled to gene array technology (ChIP on CHIP) or high-throughput sequencing (ChIP-Seq). Instead of a uniform distribution throughout the genome,

PTMs on histones, especially histone acetylation and lysine methylation, have been related to specific patterns of occupancies (discussed in the next sections).

Histone acetylation

Acetylation involves the addition of an acetyl group from acetyl-coenzyme A to the £ - amino side chain of a lysine. By neutralising the basic charge of the lysine residue, histone acetylation provides the potential to unfold chromatin by weakening the bonds between different histones in adjacent nucleosomes or between histones and DNA

(Boeger et al, 2003; Reinke & Horz, 2003). Direct evidence for this physical property has been demonstrated in vitro by biophysical analysis of recombinant nucleosomes that have undergone H4K16 acetylation, showing a negative impact on the formation of higher-order chromatin structure (Shogren-Knaak et al, 2006). Understandably, therefore, histone acetylation is mainly associated with actively transcribed genes.

Genome wide analyses suggest that (i) acetylation is enriched at specific sites in the promoter and at the 5' end of coding regions and (ii) within the promoter there are two nucleosomes flanking the initiation sites that are hypoacetylated at certain key

(Liu et al, 2005; Millar & Grunstein, 2006; Millar et al, 2006). Another well- characterised mechanism by which histone acetylation functions is in providing a docking site for the bromodomains of chromatin-associated proteins (Hassan et al,

2007; Kanno et al, 2004; Yang, 2004).

Addition and removal of acetyl groups on histone lysine residues are brought about by the opposing actions of histone acetyltransferases (HATs) and histone deacetylases

(HDACs). HATs and HDACs generally act as part of multiprotein complexes. Proteins present in the complex may influence the substrate specificity and lysine preferences of these enzymes (Shahbazian & Grunstein, 2007). HATs are divided into three main

19 families: Gcn5-related N-acetyltransferase (GNAT), MYST (for its founding members

MOZ, YBF2/SAS3, SAS2, and TIP60), and CREB-binding protein (CBP)/p300

(Sterner & Berger, 2000). In general, these enzymes modify more than one lysine; for example, TIP60, a member of the MYST family, is known to catalyse acetylation of multiple lysines on H4 (K5, K8. K12, K16) and H3K14 (Kimura & Horikoshi, 1998).

Similarly, HDAC enzymes do not appear to show specificity for particular substrates, acting as mediators of transcriptional repression. Mammalian HDACs can be divided into three different classes according to sequence and structure homologies. Class I

HDACs (HDAC1, 2, 3, 8) are ubiquitously expressed with predominantly nuclear localisation, except HDAC3 which can also localise to the cytoplasm. In contrast, class

II HDACs (HDAC4, 5, 6, 7, 9, 10) can undergo stimulus-dependent nucleo-cytoplasmic shuttling. The non-classical class III HDACs or Sirtuin family share no homology with class I and II HDACs; their deacetylase activity depends on the co-factor NAD+ and they are found in all subcellular compartments (de Ruijter et aU 2003; Michan &

Sinclair, 2007; Yang & Seto, 2008)

Histone lysine methylation

Histones can be methylated on either lysine (K) or (R) residues by the catalytic activities of histone lysine methyltransferase (HKMTs) or protein arginine methyltransferases (PRMTs), respectively. More is known about lysine methylation and a large number of HKMTs have now been identified since the discovery of the first

HKMT, SUV39H1, over a decade ago (Rea et al, 2000). Lysine side chains may be mono-, di-, or tri-methylated with differing functional consequences and many have been associated with either active or repressive gene transcriptional marks, central to the idea of the ‘histone code' (Bannister & Kouzarides, 2005; Lachner & Jenuwein, 2002).

As an example, monomethylated H3K4 (H3K4me) has been shown to be associated with transcription factor binding at enhancers, trimethylated H3K4 (H3K4me3) with

20 transcription start sites (TSS), whereas dimethylated H3K4 (H3K4me2) with both the

TSS and enhancers (Barski et al, 2007; Heintzman et al, 2007).

Heterochromatin, the condensed inactive form, is associated with high levels of certain methylated sites: H3K9, H3K.27, and H4K20. Specifically, H3K9me3 and H4K20me3 are enriched in pericentric heterochromatin, while H3K27me3 is enriched in the inactive X-chromosome (Martens et al, 2005; Plath et al, 2003; Schotta et al, 2004).

Euchromatin is associated with trimethylations at H3K4 and H3K36. In budding yeast,

H3K4me3 is present at the 5’ end of active genes and is found associated with the initiating form of RNA polymerase II. In contrast, H3K36me3 is found predominantly at the 3' end of active genes and associated with the elongating form of RNA polymerase II (reviewed in Bernstein et al, 2007).

Unlike acetylation, addition of a relatively small methyl group does not neutralise the charge of lysine or arginine, hence, methylation is unlikely to cause significant impact on the chromatin structure. Instead, it often creates binding sites for effector proteins which recognise the modification. For example, methylation of H3K9 creates a binding site for the chromo domain of heterochromatin protein 1 (HP1), which together with proteins, mediate gene repression (Bannister et al, 2001; Lachner et al,

2001).

In general, HKMTs are specific to their target substrates. Originally identified in

Drosophila, the proteins [Suppressor of variegation [Su(var)], Enhancer of zeste [E(z)] and Trithorax] all contain the evolutionarily conserved SET domain. This peptide motif has been shown to catalyse the attachment of methyl groups to specific lysine residues on histone tails (Jenuwein et al, 1998). Over 60 members of human SET domain- containing proteins have been identified; although, not all have known histone lysine methyltransferase activity (Smith & Denu, 2009). SUV39HI, the first mammalian

HKMT, was identified to act specifically on H3K9 (Rea et al, 2000). To date, all except

21 one (Min et al, 2003) of the histone lysine residues known to be methylated are modified by SET domain-containing proteins (reviewed in Bannister & Kouzarides,

2005; Qian & Zhou, 2006).

In contrast, few histone demethylases have been identified and assigned to specific histone target residues. LSD1 (lysine-specific demethylase 1) acts to demethylate

H3K4, thus repressing transcription (Shi et al, 2004). A family of Jumonji C proteins have also been identified as histone demethylases, with members of the family targeting demethylation of a range of histone lysine residues (Tsukada et al, 2006; Tsukada et al,

2010).

Histone modifications in neurons

Changes in histone PTMs, in particular histone acetylations, have been extensively documented within the regulatory regions of genes induced by neuronal activity

(reviewed in Dulac, 2010; Riccio, 2010). For example, the histone acetyltransferase

CBP is an important factor for CREB-dependent transcription. CBP is recruited to

CREB-regulated genes upon calcium-dependent phosphorylation by CamKIV

(Hardingham et al, 1999; Hu et al, 1999; Impey et al, 2002). Interestingly, the route of calcium entry (either NMDA receptors or L-type Ca2t channels) differentially controls the association of CBP with neuronal gene promoters (Hardingham et al, 1999). The essential role of HAT activity in CBP-mediated neuronal plasticity is evident in the long-term memory defects of a transgenic mouse line carrying a dominant negative

CBP that blocks endogenous HAT activity (Korzus et al, 2004). Genome wide studies of the occupancy of CBP and p300 in brain or cultured neuronal cells have revealed its interaction with noncoding regions in and around gene promoters (Kim et al, 2010;

Visel et al, 2009). CBP and p300 were also found in association with monomethylated

H3K4, an epigenetic mark for enhancers. Using a ChIP-Seq based assay, CBP recruitment to enhancers was shown to increase significantly after membrane depolarisation (Kim et al, 2010).

22 A recent genome-wide study shows that altered histone acetylation is associated with age-dependent memory impairment in mice (Peleg et al, 2010). Specifically, H4K12 acetylation is deregulated in aged mice during learning tasks. This leads to a failure to activate the hippocampal gene expression program associated with memory consolidation.

The histone acetylation program in the brain is also regulated by a subset of HDACs, most of which are developmentally regulated (Broide et al, 2007). For example, within the class 1 group, HDAC1 expression is restricted to stem cells and glia; whereas,

HDAC2 is mostly found in differentiated neurons. HDAC1 has been implicated in axonal demyelination while HDAC2 has been strongly implicated in the regulation of memory formation (Guan et al, 2009; Kim & Casaccia, 2010). Complementary mouse models that either lack or have elevated levels of HDAC2 showed the contrasting phenotypes of increased and impaired emotional and spatial working memory, respectively (Guan et al, 2009). These cognitive phenotypes were accompanied by alterations in the amount of synapse formation which is known to correlate with learning and memory. Using ChIP-qPCR analyses, HDAC2 was shown to occupy the promoter regions of a large number of genes involved in synaptic plasticity or activity- dependent processes. This study, as well as many others now suggest that HDAC2 negatively regulates memory. An inhibitor that specifically targets HDAC2 has been proposed as a potential clinical therapeutic for treatment of neurodegenerative diseases

(Fischer et al, 2007; Guan et al, 2009; Vecsey et al, 2007).

The functional significance of histone methylation in neuronal systems is increasingly being elucidated. The protein orthologs of the Drosophila Trithorax and Polycomb proteins share the highly conserved histone methyltransferase SET domain, which act as epigenetic activators and silencers of the homeobox (Hox) gene clusters during vertebrate development, respectively (Hanson et al, 1999). These two counteracting sets of proteins are large multiprotein complexes which mediate methylation of H3K4 and

23 H3K27, respectively. Members of Polycomb (such as BM11) and Trithorax (such as

MLL1) groups have been shown to be required for neural stem cell renewal and neurogenesis in the mouse postnatal brain, respectively (Fasano et aL 2007; Lim et al,

2009).

The roles of other histone lysine methyltransferases such as GLP/G9a and SETDB1 have also been elucidated in specific cognitive and behavioural pathways. The two related proteins GLP and G9a exist endogenously as a GLP/G9a heteromer which has been shown to be essential for H3K9 mono- and di-methylation which mediates transcriptional silencing (Tachibana et al, 2005). Conditional inactivation of GLP/G9a in the postnatal brain leads to activation of non-neuronal and early neuron progenitor genes in neurons of the adult brain, suggesting its role in maintaining transcriptional homeostasis in neuronal cells (Schaefer et al, 2009). In addition to a marked reduction in the level of H3K9me2 in neuronal chromatin, behavioural analyses of the GLP/G9a conditional knockout mice revealed defects in learning, motivation and environmental adaptation, but there were no apparent structural abnormalities of brain development.

The impact of GLP/G9a and H3K9me2 on neuronal gene expression has also been implicated in a mouse model for drug addiction (Maze et al, 2010). Repeated cocaine administration in mice resulted in the specific downregulation of G9a in neurons of the nucleus accumbens, which led to detectable global reduction of H3K9me2 in the brain.

The changes were related to aberrant transcriptional activation of numerous genes known to regulate dendritic spine plasticity in the nucleus accumbens, the molecular basis of cocaine addiction (Robinson & Kolb, 2004). SETDBI (Set domain, bifurcated

1) is a specific H3K9 methyltransferase associated with a repressive mark in adult brain neurons. Overexpression of SETDBI in the adult forebrain of mice resulted in antidepressant-like behavioural phenotypes through repression of the NMDA receptor subunit gene GluN2B (Jiang et al, 2010).

24 1.1.8 Understanding the genetic and epigenetic basis of human cognition

In the complex CNS, invaluable lessons have been learned from the identification of genes affected by single nucleotide polymorphisms (SNPs) that cause specific, often rare, cognitive abnormalities in . On the other hand, the more prominent neurodevelopmental disorders are frequently caused by genetic mutations or chromosomal microdeletions affecting a subset of genes, which may involve perturbations at both the genetic and epigenetic levels. Ongoing research aimed at linking the molecular details of SNPs, genetic or epigenetic alterations to the neurobehavioural phenotypic outcomes in these disoreders provides insight into the role of specific genes and more recently, the epigenetic mechanisms underlying human cognition and behaviour.

1.1.8.1 Single gene-cognition linkage: lessons from SNPs

Millions of single nucleotide polymorphisms (SNPs) have been identified in the , which constitute the most frequent DNA sequence variant. SNPs can affect the promoter, splice site, or coding sequence, and thus have the potential to change the expression or biochemistry of proteins (Goldberg & Weinberger, 2004).

A common SNP has been identified in the gene encoding BDNF, which results in a

Val66Met amino acid substitution and biochemically alters the intracellular trafficking and secretion of BDNF from neurons (reviewed in Cohen-Cory et al, 2010). The

Val66Met variant is associated with lower neuronal integrity and synaptic abundance in the hippocampus of affected individuals and impaired verbal episodic memory in a task involving memory for semantically structured stories. This further emphasises the important role of BDNF in human memory and hippocampal function (Egan et al,

2003).

SNPs in the cholinergic receptor al nicotinic receptor gene (CHRNA7) on chromosome

15q 14 have been associated with attention deficit hyperactivity disorder (ADF1D)

25 (Freedman et al, 1997). Alterations in the CHRNA7 promoter sequence affect the abundance of CHRNA7 protein, hence affecting nicotinic processing in the hippocampus and related cortices, which in turn affect processing in this cognitive paradigm. This inhibitory deficit in attentional processing has been implicated in symptoms such as hallucinations and disorganisation in schizophrenia (Leonard et al,

2002). ADHD has also been associated with polymorphisms in catecholamine-related genes, including dopamine type-4 receptor (DRD4), dopamine transporter (DAT), catechol-O-methyltransferase (COMT), serotonin 2A receptor (HTR2A) and monamine oxidase-A (MAOA). These genes are all involved in diverse aspects of aminergic signalling in the brain by affecting dopamine, norepinephrine or serotonin. Thus, catecholamimetic drugs have been used as the basic treatment for disorders of arousal and attention (reviewed in Fossella et al, 2002).

A remarkable linkage has been made between speech and language impairments and rare mutations in a single gene, FOXP2 (Lai et al, 2001; Lennon et al, 2007). Numerous investigations involving neuroimaging, animal models (both mouse and songbird), molecular studies of gene function, and population and evolutionary studies have provided evidence supporting the role of FOXP2 in the development of brain regions responsible for fine motor control and how its disruption has an exceptionally severe impact on speech development (reviewed in Fisher & Scharff, 2009).

1.1.8.2 Genetic and epigenetic alterations in neurocognitive disorders

A number of neurodevelopmental disorders that result in specific cognitive and behavioural phenotypes due to both genetic and epigenetic abnormalities have been identified. These include Rett syndrome, Angelman syndrome, Prader-Willi syndrome,

Rubinstein-Taybi syndrome and potentially, Williams-Beuren syndrome. The contribution of epigenetic mechanism to these disorders may involve dysregulation of key components of the epigenetic machinery or epigenetic alterations in the expression of genes important for cognition and behaviour (Franklin & Mansuy, 2010).

26 Rett syndrome

Mutations in the X-linked gene MeCP2 causes Rett syndrome, a progressive neurodevelopmental disorder which affects 1 in 15,000 births (Amir et al, 1999;

Chahrour & Zoghbi, 2007). The onset of disease begins in early childhood with mental retardation, behavioural changes, late movement disturbances, loss of speech and hand skills, ataxia, apraxia, irregular breathing and frequent seizures (Amstrong, 2002;

Naidu, 1997). Patients show abnormal electroencephalography in the final stage of the disease. The neuropathology includes signs of neurodegenerative and metabolic disorders, decreased brain growth and reduced neuronal density with thinned dendrites in some cortical layers and abnormalities in the substantia nigra, indicative of deficient synaptogenic development. Neurometabolic changes include reduced levels of dopamine, serotonin, noradrenalin, choline acetyltransferase, nerve growth factor, glutamate and other amino acids, as well as various receptors (Amstrong, 2002; Huppke & Gartner, 2005).

The major defects in brain function and the late onset of phenotypes observed in Rett syndrome have also been observed in MeCP2 mutant mouse models (Chahrour et al,

2008; Shahbazian et al, 2002). The complexity of the disease phenotypes reflects the broad impact of MeCP2 predominantly in gene silencing associated with CpG methylation, such as negatively regulating BDNF in neurons (Chahrour et al, 2008).

Angelman and Prader-Willi Syndrome

Angelman syndrome (AS) and Prader-Willi syndrome (PWS) have considerably distinct phenotypes; although, the genes involved in these syndromes have been mapped to the same chromosome region 15q 11-13. The clinical aspects of AS include characteristic behaviours of ataxia and jerkiness, severe learning difficulties, dysmorphic facial features, loss of linguistic ability and frequent seizures. Their hyperactivity and easily provoked laughter lead to this condition being named as “happy puppet” syndrome 27 (Clayton-Smith & Laan, 2003). The clinical features of PWS consist of mental retardation and learning deficiencies, infantile hypotonia, poor suck reflex, growth retardation, delayed sexual development and hyperphagia (obsessive eating). The behavioural features include high rates of depression, mood liability, obsessions and compulsions, possibly relating to the food related difficulties (State & Dykens, 2000).

The 15q 11-13 region affected in these syndromes is subject to the epigenetic mechanism of genomic imprinting (Fryer, 1997; Wilkinson et aU 2007). Imprinted genes are differentially methylated depending on their parental origin. A child inheriting a 4 Mb hemizygous deletion of the critical region of the paternally derived chromosome

15 will develop PWS; whereas, a child inheriting a similar deletion of the maternally derived chromosome 15 will develop AS (Fryer, 1997). Haploinsufficiency of UBE3A, one of the 13 genes in the 4 Mb critical region which encodes an E3 ubiquitin ligase, causes disruptions to the level of important synaptic proteins (such as Arc and Ephexin

5), which is thought to be a major contributor to the cognitive dysfunctions of AS and PWS (Fryer, 1997; Greer et al, 2010).

Rubinstein-Taybi syndrome

Mutations in the gene encoding the histone acetyltransferase CBP (CREB-binding protein) are responsible for Rubinstein-Taybi syndrome (RTS), an autosomal dominant neurological disease with an estimated prevalence of 1 in 125,000 individuals (Petrij et al, 1995; Roelfsema & Peters, 2007). RTS patients suffer from intellectual disabilities, growth retardation and skeletal abnormalities, including craniofacial dysmorphology.

The impaired CBP function is associated with decreased HAT activity, which impacts on CREB-dependent and other pathways (Roelfsema & Peters, 2007). Mice lacking

CBP die prenatally, but CBP heterozygous mutant mice survive to adulthood with several symptoms reminiscent of RTS, including long-term memory and learning deficits (Alarcon et al, 2004). Abnormal chromatin acetylation patterns have been observed in the brains of CBP heterozygous mice due to the defective HAT function of

28 CBP. Administration of the HDAC inhibitor SAHA reversed the memory defect observed in a fear conditioning test of the mice, demonstrating the importance of histone acetylation as means of gene control in the mammalian brain (Alarcon et al,

2004).

1.2 Williams-Beuren syndrome as a model for the genetic basis of human cognition and behaviour

Williams-Beuren syndrome (WBS) is a complex human neurodevelopmental disorder with an estimated incidence of between 1 in 20,000 (Greeberg, 1990) to 1 in 7,500 live births (Stromme et al, 2002). It is usually caused by a hemizygous 1.5-1.8 Mb deletion of chromosome 7q 11.23 (Ewart et al, 1993a), resulting in a range of distinctive physical, cognitive and behavioural phenotypes. Cases are generally sporadic; however, familial cases of WBS with an autosomal dominant mode of inheritance have been reported (Morris et al, 1993).

1.2.1 The genetic basis of WBS The WBS region at human chromosome 7ql 1.23 is flanked by low-copy repeat (LCR) blocks that are thought to be directly responsible for the recurrent rearrangements in this region (Figure 1.3; Valero et al, 2000). LCR sequences have been found within human chromosomal regions that are subject to other spontaneous deletions, including 15q 1 1-

13 (Angelman/Prader-Willi syndrome) and 22q 11 (Velocardiofacial/DiGeorge syndrome) (Stankiewicz & Lupski, 2002).

The LCRs at 7ql 1.23 consist of three main blocks named “A”, “B" and “C” which are located at the centromeric, medial, and telomeric end of the WBS region (Figure 1.3,

Bayes et al, 2003). The LCRs are thought to create the WBS deletion by illegitimate recombination and unequal exchange and this is likely because the LCRs show (1) a large continuous homology (-100 kb for each block and >300 kb for the extended

29 centromeric and telomeric arrays) and (2) an extremely high degree of sequence conservation (~98%). In 95% of the cases, the deletion involves the breakpoints residing between the directly oriented and most highly similar repeats of block B in the centromeric and medial LCRs, giving arise to the 1.55 Mb deletion with 28 genes that have been identified in the region to date (reviewed in Schubert, 2009). In a rarer case, a

1.84 Mb deletion resulted from recombination between the centromeric and medial A blocks (Bayes et al, 2003; Peoples et al, 2000; Valero et al, 2000).

The non-allelic homologous recombination mediated by the LCRs could occur through intrachromosomal (same chromosome) or interchromosomal (different ) rearrangement. In each case, the chromosomes are thought to loop in a way that allows the alignment of two non-allelic LCRs, leading to recombination and excision of the intervening DNA (Bayes et al, 2003). Interchromosomal non-allelic homologous recombination will theoretically lead to deletions and reciprocal duplications of the genomic region between the LCRs at equal frequencies. However, to date, much fewer cases of the reciprocal duplication have been reported (see section 1.2.7). The higher observed frequency of the deletions could be explained by a higher rate of intrachromatid recombination which would lead to a deletion and a reciprocal acentric chromosome fragment, but not to a duplication (Bayes et al, 2003). It is also believed that the lack of a well-defined phenotypic spectrum associated with reciprocal WBS duplication also hampers the identification of individuals with duplication (Osborne,

2008; Turner et al, 2008). Analysis of the parents of WBS patients has shown that approximately one third carry the genomic inversion (without deletion) of WBS region

(Hobart et al, 2010; Osborne et al, 2001). The genomic variant does not cause significant clinical symptoms, but is thought to predispose the chromosome to subsequent unequal recombination during the next round of meiosis (Tam et al, 2008).

30 1.2.2 The physical characteristics

WBS was first discovered in the early 1960s from a number of reports associating a cardiovascular condition known as supravalvular aortic stenosis (SVAS) with a distinctive facial appearance and a growth and developmental retardation (Williams et al, 1961). Since then, a wider range of clinical phenotypes have been noted, including infantile hypercalcaemia, musculoskeletal abnormalities, hoarse voice, dental problems, gastrointestinal problems, and urinary tract abnormalities such as enuresis (Morris &

Mervis, 2000).

Cardiovascular abnormalities

Cardiovascular disease accounts for most of the morbidity and mortality of WBS; although, with the typically early intervention that is now commonly experienced, the overall mortality is relatively low (Collins et al, 2010). The cardiovascular abnormalities have an overall prevalence of 80%, in which SVAS, the congenital narrowing of the ascending aorta, is the most common manifestation. SVAS often develops in combination with other vascular stenoses (narrowing of the blood vessels), including pulmonary artery and renal artery stenosis (Donnai & Karmiloff-Smith, 2000;

Eronen et al, 2002). SVAS is also recognised either as an isolated sporadic or an autosomal dominant genetic disorder caused by mutations in the elastin gene (ELN), which lies within the WBS deletion region (Curran et al, 1993; Ewart et al, 1994; Ewart et al, 1993b; Metcalfe et al, 2000). Elastin in elastic fibres provides strength and elasticity to tissues such as heart, skin, and major blood vessels, including the aorta and pulmonary arteries. Accordingly, haploinsufficiency of elastin causes a range of vascular problems with a variable severity in WBS individuals.

Craniofacial and dental abnormalities

The facial appearance of individuals with WBS is distinctive and the dysmorphology involves both soft tissue and skeletal components. The major skeletal features include a shorter anterior cranial base, a steeper angle of the mandibular plane, an unusual

31 proportion of upper to lower and posterior to anterior facial dimensions, and a relatively small chin button. The soft tissue characteristics typically consist of full lips and a wide smile, full cheeks, periorbital fullness, short nose and prominent earlobes (Mass &

Belostoky, 1993; Morris et al, 1988). Overall, these features of craniofacial abnormalities contribute to the characteristic “elf-like” appearance of WBS patients

(Figure 1.4.A; Jones & Smith, 1975). Three-dimensional imaging, which generates surface models and identifies discriminating features of the face, has been used to define the characteristic features of a typical WBS face and can now be used as a diagnostic tool (Hammond et al, 2005).

In addition to the altered facial features, WBS patients generally have small, widely spaced teeth and dental malocclusions. These are attributable to the underlying skeletal features, such as the high mandibular plane angle and relative protrusion of the maxilla as well as to the underlying soft tissue, such as the posture and muscle tone of the lips.

Other dental problems include malformed teeth, missing permanent teeth and localised enamel hypoplasia (Axelsson, 2005; Mass & Belostoky, 1993).

Growth retardation The incidence of prenatal growth deficiency in WBS is approximately 50-70%. Feeding problems, prolonged colic, gastroesophageal reflux and constipation lead to failure to thrive in 80% of infants as well as poor weight gain and linear growth (Morris et al,

1988). Data from all age groups showed that 30-40% of WBS individuals fall below the third percentile of standard growth curves, but it is thought to have a multi-factorial cause (Pankau et al, 1992). The frequency of vascular stenoses as well as hypercalcaemia, which is documented in 15% of individuals with WBS (most commonly found in the first 18 months, but may also recur in adulthood), may contribute to the growth delay. In addition, possible parent-of-origin effects have to be taken into account (Francke, 1999). Puberty is often early and has been associated with the cause of the short final adult height (Cherniske et ah 1999).

32 1.2.3 The cognitive profile

The cognitive profile of WBS is of particular interest because of the striking contrasts in the major brain functions. In general, WBS individuals have relatively preserved auditory memory, language and facial recognition abilities, but have profound impairments in their spatial processing and a lower IQ. In some cases, spontaneous musical interests and exceptional levels of musical aptitude have been reported

(Carrasco et al, 2005).

Studies measuring the overall level of intellectual ability of WBS individuals have shown that the mean IQ of WBS individuals is considerably lower than normal age- matched controls. This result has been consistent in studies that used full scale IQ tests which measure verbal, nonverbal reasoning and spatial abilities (mean IQ = 51-70;

Bellugi et al., 2000) and those that used the Kaufman Brief Intelligence Test (K-BIT), which does not assess spatial constructive ability. The spread of IQ scores was found to be broad ranging from severe mental retardation to low average intelligence, but the mean IQs indicated mild mental retardation (Mervis & Robinson, 2000).

When testing cognitive performance in WBS subjects, many studies recognise the importance of comparing WBS to other syndromes, such as Down Syndrome (DS), in order to provide age-matched and IQ-matched control subjects. Three areas of intellectual abilities that have been studied most extensively with reference to DS are discussed below: visuospatial construction, language and face recognition.

1.2.3.1 Visuospatial construction deficit

The visuospatial construction deficit of WBS patients is an inability to conceptualise and reproduce shapes and forms. This can be demonstrated by asking WBS individuals to engage in tasks involving drawing and block design (Morris & Mervis, 2000). These tasks highlight the tendency of WBS subjects to focus on extreme detail at the expense of the whole or global picture. This result is opposite to the age- and IQ-matched DS

33 subjects who produce simple or poor detail in their drawings, but the global configuration is maintained (Figure 1.4.B; Bellugi et al. 2001).

1.2.3.2 Preserved language

The earliest stages of language development in WBS children are usually delayed

(Nazzi et al. 2003), but once acquired, language has been considered to be a relative strength in the WBS cognitive profile (reviewed in Brock, 2007; Mervis & Becerra,

2007). Comparative studies, such as those done through story-telling, with age- and IQ- matched DS subjects indicate that the WBS counterparts have stronger grammar and a more expressive, richer and unusual vocabulary (Bellugi et al. 1999b; Mervis &

Robinson, 2000). It should be noted, however, that stronger grammar was not significantly different when compared to age- and IQ-matched individuals with other intellectual disabilities, suggesting a possible extreme weakness of DS in grammatical development (Grant et al. 2002; Udwin & Yule, 1990). The strongest aspect of WBS vocabulary is in the use of receptive concrete language (understanding of objects, actions and descriptors). Poorer performance was found in relational/conceptual vocabulary (spatial, temporal, qualitative and dimensional terms), in agreement with the visuospatial construction deficit (Mervis & Becerra, 2007).

Although WBS individuals are widely recognised for their strong drive for social interaction (see section 1.2.4.1), these subjects have pragmatic/socio-communicative language anomalies during conversational interactions that are similar to those manifested in autism spectrum disorders (Laws & Bishop, 2004; Philofsky et al. 2007).

Non-verbal communication elements, such as pointing, gestures, facial expression, eye contact and body language seemed to be executed differently by WBS individuals compared to control individuals (Harris et al. 1997; Klein-Tasman et al. 2007). In addition, an excessive usage of unusual language was found, such as cliches, idioms, and figurative speech, which often appeared inappropriate to conversational context and

34 this may indicate a poor underlying knowledge of the meaning of such words and phrases (Reilly et al, 2004; Thomas et al, 2010; Udwin & Yule, 1990).

1.2.3.3 Enhanced facial recognition ability

Despite the severe spatial processing deficit, WBS individuals exhibit good facial recognition ability. This is evident in complex tests involving discrimination of unfamiliar faces in different conditions of lighting and orientation, which challenge the subjects' recognition, classification and memory (Bellugi et al, 2000). Good facial processing is accompanied by a strong attention bias toward faces. Eye-tracking devices used during face recognition tasks revealed that WBS individuals showed prolonged face-gazing behaviour and a prolonged time taken to disengage from viewing faces compared to viewing objects (Porter et al, 2010; Riby et al, 2008); these findings are highly relevant to the heightened sociability characteristic of WBS (see section 1.2.4.1).

In conclusion, it has been proposed that the three cognitive skills above (spatial construction, language and face processing) follow distinct trajectories of development in WBS, in contrast to the relatively even developmental trajectories of cognitive impairment shown in the subjects with DS (Figure 1.4.C; Bellugi et al, 1999b).

1.2.4 The behavioural profile

A unique component of the behavioural profile that distinguishes WBS from other developmental disorders, such as autism and DS, is the hypersociability. Paradoxically, this decrease in social anxiety is contrasted with increased non-social anxiety manifested in high rates of phobias. Symptoms of attention deficit-hyperactivity disorder (ADHD) are also part of the classical behavioural features of WBS, which could also be exacerbated by the additional feature of sleep disturbances (Einfeld et al,

1997; Goldman et al, 2009; Sarimski, 1996).

35 1.2.4.1 Hypersociability

“Everybody in the world is my friend” is a quote from a WBS patient which reflects the notion of an over-friendly, outgoing, engaging and highly affectionate personality, typical of WBS individuals in their social interactions. This characteristic is extremely noticeable and spontaneously referred to in around 90% of the cases (Carrasco et al,

2005). Studies measuring distinct aspects of the hypersociable behaviour in WBS have shown that this trait is highly uniform across a wide range of ages and multiple experimental paradigms (Jones et al, 2000). The social behaviour of WBS individuals is apparent from an early age and is displayed through an abnormal degree of fearlessness towards strangers, thus, becoming a major concern for the parents. The unusual attraction to people, faces, and especially the eye are also features of the striking social- behaviour phenotype of WBS individuals (Porter et al, 2010; Riby et al, 2009; Tager-

Flusberg et al, 2003).

Using a test that assessed the judgment of approachability of various unfamiliar faces, it was found that WBS subjects tended to rate approachability much higher than the normal controls (Bellugi et al, 1999a). Similar characteristics of unusual friendliness has been found in patients with bilateral amygdala damage in which, like WBS, facial identification is also preserved (Adolphs et al, 1999). However, the difference between the two conditions is that subjects with WBS manage to differentiate between trustworthy and untrustworthy looking faces, but find a positive bias toward approachability (Frigerio et al, 2006; Porter et al, 2007); whereas, those with bilateral amygdala damage failed to differentiate between the trustworthy and untrustworthy looking faces. Therefore, amygdala damage impairs the normal ability to judge potentially dangerous or threatening people (based on appearance), while WBS enhances a strong tendency to approach anyone regardless of their judged state of mind.

An alternative explanation that is gathering supporting evidence involves frontal lobe dysfunction. Consistent with WBS, some patients with frontal lobe damage are able to detect threatening faces, but are unable to avoid approaching the subject and hence, are

36 referred to as impaired in the response-inhibition pathway (Green & Phillips, 2004).

People with WBS also display classic signs of frontal lobe dysfunction including impulsive and inappropriate behaviour (Einfeld et al, 1997). Functional MRI neuroimaging studies have implicated abnormalities in the linked pathway of amygdala and prefrontal cortex as the proposed underlying mechanism of social interaction anomalies in WBS (see section 1.2.6.2; Meyer-Lindenberg et al, 2005a; Munoz et al,

2010).

1.2.4.2 Non-social anxiety

Between 50 to 90% of adolescents and adults with WBS meet the diagnostic criteria for anxiety disorder, phobic disorder, ADHD or a combination thereof (Davies et al, 1998).

This high rate of anxiety in individuals with WBS is consistent with many other reports

(Cherniske et al, 2004; Dykens, 2003; Leyfer et al, 2009), which persist over time and are not related to age or to IQ scores (Leyfer et al, 2006; Woodruff-Borden et al, 2010).

A specific phobia of hyperacusis - oversensitivity and fear of loud noises, such as thunderstorms or fireworks, is common in WBS from infancy (Blomberg et al, 2006; Gothelf et al, 2006; Johnson et al, 2001; Levitin et al, 2005; Nigam & Samuel, 1994).

Findings by Gothelf et al. (2006) also suggested that a combination of cochlear hearing loss and auditory nerve dysfunction, which alters the perception of loudness in the afferent auditory system, underlie the hyperacusis in WBS.

1.2.5 Other neurological features

Although motor function in WBS has received little attention, children with WBS have frequently been reported with delayed developmental motor milestones, including difficulties in fine and gross motor coordination, inability to perform ‘tandem' walking, abnormalities in tone and ‘awkward' gait (Chapman et al, 1996; Trauner et al, 1989).

The first detailed analysis of the gait patterns in WBS subjects showed a reduced gait speed and stride length, but the stepping frequency (cadence) was disproportionately increased compared to the control group (Hocking et al, 2008). The gait abnormalities

37 are in line with the mild extrapyramidal signs reported in WBS, which include choreoathetosic, dystonic and involuntary movements, which increases with age

(Gagliardi et al, 2007). This neurological dysfunction may be related to the structural and biochemical abnormalities of the cerebellum, discussed below (Chiang et al, 2007;

Jones et al, 2002; Rae et al, 1998).

1.2.6 Neurobiology in WBS

The neurobiological profile of WBS has been explored through a combination of studies including (i) brain structure and cytoarchitecture (through 3D computer graphic coupled magnetic resonance imaging, MR1; voxel-based morphometry, VBM; and brain autopsy), (ii) neural activities (through event-related potentials, ERPs and functional

MRI, fMRI), and (iii) brain biochemistry (through magnetic resonance spectroscopy,

MRS).

1.2.6.1 Brain structure and cytoarchitectonic abnormalities Autopsy and high-resolution MRI studies suggest an overall decreased brain size (11- 13%) in WBS (Jernigan & Bellugi, 1990; Reiss et al, 2000). A clear decrease in the total cerebral volume is mainly due to a significant reduction in the cerebral white matter (~20%), while cerebral gray matter is less affected (Reiss et al, 2000; Schmitt et al, 2001; Thompson et al, 2005). When total brain volume is controlled for, marked reductions have been found in parieto-occipital lobe volume relative to the preserved frontal lobe, which in turn may cause the differences in cerebral shape (Eckert et al,

2005; Reiss et al, 2004b; Schmitt et al, 2001), in the corpus callosum (Schmitt et al,

2001) and brainstem (Reiss et al, 2000). Accompanying these reductions are preservations in other areas. Heschl's gyrus in the limbic area of the primary auditory cortex is not only preserved in volume, but enlarged in size, which might have relevant implications to the strength in the auditory, short-term memory, language and musicality functions (Levitt et al, 2004). The cerebellum is larger relative to the smaller

38 cerebrum, and preservation in the limbic area of the temporal lobes, including the

amygdala has been noted (Chiang et al, 2007; Jones et al, 2002).

Studies using fMRI provide converging evidence that the neural organisation in WBS is

abnormal. The neural abnormalities described in WBS that may be responsible for the

visuospatial construction deficit include the significantly smaller, more closely packed

neurons in the regions of the primary visual cortex (Galaburda et al, 2002) and a

significant reduction of grey matter density within several cortical and subcortical

regions comprising the visual system (Boddaerta et al, 2005; Reiss et al, 2004a).

1.2.6.2 Dysfunctions of neural activities

While fMRI provides the opportunity to localise regional neural activation in response

to a stimulus, ERPs provide information about the timing and temporal sequence of

neural events and to some extent, the location of neural activity. The technique has been

applied during language-associated and face-processing stimuli and the results indicate

that there are significantly different ERPs or unique brain-wave markers in WBS

subjects compared to those of normal subjects. These studies suggest that there is an

abnormal functional organisation of the neural systems that subserve the higher

cognitive functions, such as language and face processing (Bellugi et al, 1999b; Mills et

al, 2000).

A hypothesis that seeks to explain the function of the primate visual cortex promotes the

existence of a ventral or “what" stream which is associated with object processing -

such as with form, colour and face-processing functions - and a dorsal or “where"

stream in the tempo-parietal junction which has been associated with the spatial-

integrative and motion-processing functions (Van Essen et al, 1992). As expected,

fMRI studies during different visual tasks has shown an isolated hypoactivation in the

parietal portion of the dorsal stream areas of WBS subjects compared to the controls, with no difference found in the ventral stream (Meyer-Lindenberg et al, 2004). There

39 was also a significant reduction of grey matter volume in the parietal-occipital sulcus, the region immediately adjacent to the hypoactivated area, suggesting a structural- functional connection.

fMRI has also been used to investigate the neural systems behind the preserved face processing skill. One such study showed that WBS subjects had more extensive activation in the right fusiform gyrus, anterior cingulate, and subcortical regions

(anterior thalamus and caudate), while control individuals showed greater activation in the primary and secondary visual cortices (Mobbs et al, 2004). The authors suggested that during face and gaze processing, WBS individuals have impairment in the visual cortical regions, but may have used the frontal and temporal regions as a compensatory mechanism.

The amygdala in the limbic system which is involved in the processing of emotions, fear responses and pleasure, is a major focus of neuropsychological and neuroimaging studies in seeking to understand the neural basis of social disinhibition in WBS.

Combined fMRI and ERP studies showed that WBS individuals exhibit both heightened amygdala reactivity to images of happy facial expressions and reduced reactivity for images of threatening/ fearful facial expressions, relative to matched normal controls

(Haas et al, 2010; Meyer-Lindenberg et al, 2005a). Greater amygdala activation in response to images of threatening/fearful scenes was also found in WBS compared to control subjects and is thought to be a potential neural mechanism for high non-social anxiety in WBS (Meyer-Lindenberg et al, 2005a). The abnormalities in the amygdala activation are also associated with altered prefrontal cortex regulation (Meyer-

Lindenberg et al, 2005a; Munoz et al, 2010). The prefrontal amygdala dysregulation hypothesis is supported by neuropsychological studies which indicate that the increased tendency of WBS individuals to approach strangers is due to abnormalities in cognitive judgement, rather than a failure to detect potential threats caused by amygdala impairment alone (Porter et al, 2007). This is consistent with the relative preservation of

40 amygdala size in individuals with WBS. Nevertheless, alterations in amygdala function remains a popular hypothesis as fMRI studies have also shown that right-amygdala activation during music processing is significantly increased in WBS subjects compared to controls, which could account for the increased expressiveness in the context of music (Levitin et al, 2003).

1.2.6.3 Biochemical abnormalities

In vivo assays of N-acetyl aspartate (NAA) through MRS imaging is often used to measure cellular integrity and synaptic abundance in the human brain (Pan &

Takahashi, 2005). The amount of NAA in WBS brains was found to be reduced in the cerebellum, consistent with the morphological difference observed in the cerebellum

(Jones et al, 2002; Rae et al, 1998). Similar reductions of NAA were also found in the hippocampus and put together with previous finding on the reduced blood flow in this region, indicate an overall depression of hippocampal energy metabolism and synaptic activity (Meyer-Lindenberg et al, 2005b). The deficit in hippocampal function may explain several cognitive defects of WBS, including memory and learning, in the verbal and spatial domains (Nichols et al, 2004; Vicari et al, 2005).

1.2.7 Duplications of the 7qll.23 chromosomal region and the consequences of positive gene dosage imbalance

While clear physical features and neurobehavioural phenotypes are recognisable in

WBS patients with the typical 7ql 1.23 deletion, fewer cases have been described and less distinctive phenotypes have been associated with those who carry a reciprocal duplication of 7q 11.23 region. Therefore, it is likely that some cases are still undiagnosed. A similar situation exists in other genetic disorders such as in duplications versus deletions of 22ql 1.2 (Ensenauer et al, 2003) and 1 q21.1 (Brunetti-Pierri et al,

2008) where milder pathological consequences tend to arise with gene duplications compared with the reciprocal deletions.

41 To date, a total of 27 individuals with duplications of 7q 11.23 region have been described (reviewed in Van der Aa et al, 2009). These patients have been variably associated with normal to mild facial dysmorphism, developmental delay, normal to moderate mental retardation, hypotonia (14 out of 20 examined), and epilepsy (6 out of

27) (reviewed in Merla et al, 2010; Osborne & Mervis, 2007; Van der Aa et al, 2009).

The prominent features associated with individuals with a duplicated WBS region are autistic behaviour (11 out of 27, with variable degree) and speech delay (27 out of 27)

(Van der Aa et al, 2009). Autistic features include poor eye contact and poor social interaction. The consistent phenotype of speech delay often involves severe impairment of both expressive and receptive language. These phenotypes contrast with the classical

WBS features, which are characterised by the overfriendly or gregarious personality and verbal strength with unusually expressive language. Thus, it is likely that the chromosomal region 7ql 1.23 contains dosage-sensitive gene(s) which influence social behaviour and human speech development. Interestingly, one case of triplication of the

WBS region has been reported, leading to tetraploidy in this domain, which led to the development of mental retardation, a severe expressive language delay, behavioural problems and dysmorphisms (Beunders et al, 2010). The clinical description suggested an exacerbation of the symptoms seen in the duplication cases, further supporting the connection between the pathological consequences and increasing the dosage of genes in the WBS common deletion region.

Limited neuroimaging studies have been performed on individuals with duplications and so far no consistent features have been found; although, abnormal development of the left temporal lobe has been documented in a patient with a reciprocal duplication

(Torniero et ai 2007; Van der Aa et al, 2009). The cerebral cortex also showed a simplified gyral pattern and increased cortical thickness. Cortical dysplasia manifesting as a thickening of the right perisylvian cortex is a documented feature of classic WBS

(Thompson et al, 2005). Systematic analysis of the neurobiological profile in

42 individuals with duplications and WBS could potentially lead to a better understanding of how specific brain abnormalities correlate with dysfunctions in language skill development and social interaction.

1.2.8 Linking genotype to phenotype

1.2.8.1 Genes linked to the Williams syndrome cognitive profile

Out of the contiguous array of genes located within the WBS region, only the elastin gene ELN has been firmly associated with the phenotype of SVAS. This was made possible as incidences of small internal mutations in the ELN gene correlate with the cardiovascular abnormalities in human patients (Tassabehji & Urban, 2006). As ELN is located at the centre of the WBS deletion region and SVAS is a consistent feature of

WBS, this information served as the first clue for the location of a WBS microdeletion

(Ewart et al, 1993a). The critical role of ELN in the regulation of vascular morphogenesis has also been demonstrated in studies of an Eln knockout mouse model

(Faury et al, 2003; Flammond et al, 2005; Li et al, 1998).

In general, strategies to map genotype to specific phenotypes of WBS have relied on firstly, studies of WBS patients with smaller atypical deletions and secondly, analysis of single gene knockout mouse models. As the common 1.55 Mb WBS deletion (also known as the WBS critical region or WBSCR) occurs in approximately 95% of cases due to the rearrangement mechanism predisposed by the flanking LCRs, identification of atypical deletions is rare and only accounts for 2% to 3% of cases (Antonell et al,

2005). Nevertheless, molecular and clinical analyses of these patients allow associations to be made between specific gene deletions and phenotypic outcomes.

Mouse lines carrying specific mutations of the orthologous genes from the WBS region have been generated, which provide valuable insights into the specific pathogenic outcomes. A mouse model resembling the WBS human deletion has also been generated by targeted deletion of the conserved syntenic region of mouse chromosome 5G in two 43 portions: a proximal deletion (PD) missing 7 genes from Gtf2i to Limkl (the telomeric region of the WBSCR) and a distal deletion (DD) missing the rest of the genes from

Limkl to Fkbp6 (the centromeric region of WBSCR) (Li et al, 2009). Inter-breeding of the two mice generates double heterozygous mice (D/P), which model the complete human deletion with the exception of Limkl which is included in both deletions, hence, the gene is homozygously deleted in D/P mice.

The behavioural and cognitive symptoms of WBS are collectively known as the

Williams Syndrome cognitive profile (WSCP). The advantage of WBS as a model for understanding the genetics behind specific human cognition and behaviour lies in the distinctive patterns of abilities and deficits that are present in the WSCP. Genes from the WBSCR that have been identified as having a potential role in the brain include

LIMK1, FZD9, STX1A, CL1P2, GTF21RD1 and GTF21. The possible contributions of dysregulation of these genes to the WSCP will be discussed with respect to both the human atypical deletions and analysis of mutant mouse models.

LIMK1 A link was initially made between LIMK.1 (LIM kinase 1) haploinsufficiency and the poor visuospatial constructive abilities of WBS individuals based on two families with small deletions involving only ELN and LIMK1 resulting in SVAS and some features of the WSCP (Frangiskakis et al, 1996). LIMK1 encodes a novel protein kinase with LIM domains which regulate aspects of the actin cytoskeleton and is strongly expressed in the brain. Limkl" mice show significant abnormalities in dendritic spine morphology and synaptic function, including the hippocampal LTP (Meng et al, 2002). Behavioural tests on these mice revealed altered fear responses and deficiencies in spatial learning.

However, other reports of individuals who have this gene deleted but lack the typical spatial constructive impairment and other cognitive deficits, contradict these findings

(see next section on ‘atypical deletions’) (Antonell et al, 2010; Morris et al, 2003;

Tassabehji et al, 1999b). Re-investigation of the individuals carrying the deletion of

44 ELN and LIMK1 only, using a much more comprehensive range of spatial tests revealed that the visuospatial impairment was not as clear-cut as the spatial difficulties observed in the typical WBS group. Nevertheless, characterisation of Limkl knockout mice clearly demonstrates its role in neurodevelopment. Although deletion of LIMK1 alone is not sufficient to cause WSCP features, hemizygous deletion of LIMK1 may have an additive effect with other genes within the WBS deletion region in the neurological etiology of WBS (Gray et al, 2006).

STX1A

STX1A encodes syntaxin 1A protein, a component of the SNARE (soluble N- ethylmaleimide-sensitive fusion protein attachment protein receptor) synaptic apparatus, and therefore, was thought to be a good candidate for WSCP features (Nakayama et al,

1998). Two mouse mutant lines of Stxla have been reported (Fujiwara et al, 2006;

McRory et al, 2008). In both lines, heterozygosity did not cause any apparent abnormalities in neuronal function or behaviour or cognition in the mice. In one of the lines, the homozygous null mice died in utero (McRory et al, 2008), while the other was viable, but showed a specific hippocampal learning impairment (Fujiwara et al, 2006).

These results suggest that reduced levels of STX1A protein in the brain does not affect its normal function, but the absence of the protein leads to abnormalities during early development. To date, no studies of atypical deletions have implicated hemizygous deletion of STX1A in any features of the WSCP (Antonell et al, 2010; Wu et al, 2002).

Using a different approach, Gao et al. (2010) performed a quantitative and correlative analysis between expression levels of WBS genes in and around the WBS deletion from lymphoblastoid cell lines of 65 patients and their intelligence scores. This analysis showed that variation in the level of STX1A correlates most significantly with intelligence in WBS. However, how this finding relates to the neural gene expression remains enigmatic.

FZD9

45 FZD9 (Frizzled drosophila homolog 9) encodes a transmembrane receptor of the WNT signalling protein, which is selectively expressed in the hippocampus throughout life

(Wang et al, 1997). Fzd9 null mutant mice showed severe deficits in tests for visuospatial learning/memory. Large increases in apoptotic cell death in the developing dentate gyrus were also found, but only persisted into the first postnatal week of life, while heterozygous mice show a milder difference in neuroanatomy. Both groups of mice had a diminished seizure threshold (Zhao et al, 2005). Again, atypical deletion studies have not linked hemizygosity of FZD9 with any features of the WSCP.

CLIP2

CLIP2 (CAP-Gly domain-containing linker protein 2) belongs to a family of cytoplasmic linker proteins which have been implicated in the regulation of microtubule dynamics (De Zeeuw et al, 1997). CLIP2 is highly expressed in the dendrites and cell bodies of many neurons in the brain. CLIP2 heterozygous and homozygous mice showed features reminiscent of WBS, which included mild growth deficiency, brain abnormalities, hippocampal dysfunction and particular deficits in motor coordination (Hoogenraad et al, 2002). The neurodevelopmental abnormalities seen in these mice suggest a possible role for CLIP2 in WBS. Indeed, CLIP2 together with GTF2IRD1 and

GTF2I have been implicated in the main features of the WSCP in a number of atypical deletion studies (described in the next section).

1.2.8.2 Studies of atypical deletions link disruption of GTF2IRD1 and GTF2I to the main features of the WSCP

Accumulating evidence from atypical deletions implicates genes at the telomeric end of the WBS deletion, GTF2IRD1 and GTF2I, as the genes behind the main aspects of the

WSCP.

Several patients with atypical deletions spanning all of the genes between ELN and

GTF2I have been described with the full WSCP (Botta et al, 1999; Heller et al, 2003).

46 In contrast, patients diagnosed with SVAS due to deletions extending from the centromeric LCR to near the CLIP2 gene (including LIMK1), but leaving the genes at the telomeric end intact (see Figure 1.3), appear to have an “incomplete WSCP” with mild or no mental retardation, preserved visuospatial cognition and mild facial anomalies (Ferrero et al, 2010; Gagliardi et al, 2003; Karmiloff-Smith et al, 2003;

Tassabehji et al, 2005). They displayed a mild language delay, an IQ higher than WBS controls and more developed visuospatial skills. This suggests that the haploinsufficiency of CLIP2 may contribute to some of the cognitive impairment, but not other characteristics of the WSCP such as behavioural abnormalities and visuospatial impairment, which can be explained by those genes telomeric of CLIP2,

GTFI and GTF21RD1.

GTF2IRD1 and GTF2I have been linked to visual spatial functioning from the study of a patient with an atypical deletion up to and including CLIP2 who has no visuospatial processing impairment (Hirota et al, 2003). Similar deletions that extended further to

GTF2IRDI were identified in five families with visuospatial construction impairments but no global mental retardation (mean IQ is above the mean for classical WBS) was detected (Morris et al, 2003). These findings suggest a role for GTF2I in global mental retardation while GTF2IRD1 is linked to visuospatial processing. However, it is still unclear whether hemizygosity of GTF2IRD1 alone is sufficient to cause the visuospatial construction deficit as family with a deletion from CLDN3 to GTF2IRD1 had a very mild defect in this cognitive aspect (Antonell et al, 2010).

Genomic deletions that potentially discriminate between the roles of GTF2IRD1 and

GTF2I in the neuropathology of WBS have been screened for using high-resolution arrays (Dai et al, 2009). An individual with a telomeric breakpoint downstream of

GTF21RDI was identified; SVAS, typical facial features, and a visuospatial construction deficit were found, but it was clear that the child did not have the social drive characteristic of WBS children of her age. She also performed better than the

47 typical WBS in cognitive tests. This finding therefore suggests a potential role for

GTF2I in the reduced social anxiety phenotype and further supports its contribution to cognitive function.

There are several caveats to take into account in interpreting the genotype-phenotype correlations in patients with atypical deletions, which include the exact location of the breakpoint, the effect that the deletion has on the expression of non-deleted genes (for example, the possible removal of long-range regulatory sequences), inconsistent phenotyping methods, age and ascertainment bias. Nevertheless, overall clinical and genetic mapping data suggest that GTF2IRD1 and GTF2I contribute to the craniofacial features, global intellectual impairments, the visuospatial construction deficit and the hypersociability in WBS. Furthermore, some of these features have now acquired supporting evidence from monogenic mouse mutations of Gtf2irdl (see section 1.3.3.3) and Gtf2i (see section 1.3.2.3).

1.2.8.3 Further analysis of the genes behind the craniofacial abnormalities The genetic basis of the craniofacial dysmorphology of WBS remains highly debatable, but the genes CL1P2, GTF21RD1 and GTF21 remain the best candidates to date. The correlation is supported by the fact that craniofacial abnormalities are seen in individuals with atypical deletions spanning from ELN to the end of the telomeric region (Botta et a/, 1999; Heller et al, 2003), but not in individuals where this region is preserved (Karmiloff-Smith et ah 2003; Tassabehji et ah 2005; Tassabehji et ah 1999c; van Hagen et al, 2007). An atypical deletion that includes CL1P2 and a partial deletion of GTF21RD1 (preserving both copies of GTF2I) showed “mild'5 facial features (Morris et ah 2003; Tassabehji et ah 2005), similar to when CLIP2 is partially deleted

(preserving both copies of GTF21RD1 and GTF21), indicating a potential additive effect of the three genes in the craniofacial dysmorphology (Gagliardi et ah 2003; Howald et ah 2006).

48 The second line of evidence comes from the study of Gtf2irdlTg

Gtf2irdl in mice generated by others show no gross craniofacial abnormalities (Palmer et al, 2010; Young et al, 2008). There were similar findings in the PD mice (Li et al,

2009), which include a heterozygous deletion of Gtf2irdl, Gtf2i and CLIP2; whereas, craniofacial dysmorphisms were clearly present in the DD line of mice and the combined D/P mice, thus contradicting the proposed role for GTF2IRD1 in WBS facial dysmorphology. From mouse model studies, the other candidate gene that has been implicated in craniofacial development from the centromeric region of the WBS deletion (DD in the mouse model) is BAZ1B. Mice homozygous for a Bazlb mutation died postnatally and their skulls were found to be shortened due to a reduction of the parietal and nasal bones as well as hypoplasia of the lower jaw, features that were also displayed by the heterozygotes to a slightly less extent (Ashe et al 2008).

Therefore, current data from mouse models and human atypical deletion studies show discrepancies in which genes are likely to be responsible for craniofacial abnormalities in WBS. It is possible that BAZ1B and genes within the WBS telomeric region all have some role to play in this phenotypic aspect. It has been emphasised that careful standardised analysis of facial features must be maintained to measure the varying degrees of the dysmorphism displayed by people with smaller atypical deletions

(Morris, 2006).

49 1.3 The TFII-I protein family

1.3.1 Structural features and evolutionary relationships of members of the TFII-I protein family

The three members of the TFII-I protein family are encoded by the genes GTF2I,

GTF2IRD1, and GTF2IRD2, which are clustered in humans at the telomeric end of the critical WBS region (Figure 1.3, Bayarsaihan et ah 2002; Franke et ah 1999; Osborne et al, 1999; Perez Jurado et ah 1998; Tassabehji et ah 1999a; Tipney et al, 2004). A partially truncated copy of GTF2I has also been identified within the centromeric LCR block, which is not deleted in WBS, but appears to be a pseudogene with no protein product, thus termed, GTF2IP (Perez Jurado et ah 1998).

Members of the TFII-I protein family are characterised by their unique I-repeat motifs, and the presence of a leucine zipper (LZ) motif at the N-terminus, which is thought to mediate hetero/homodimerisation (Figure 1.5, Hinsley et ah 2004; Vinson et ah 2002).

The I-repeats, also referred to as the repeat domains (RDs), were predicted to form a helix-loop-helix (HLH) structure due to broad similarities to basic HLH transcription factors (Roy, 1997). However, it was noted that if they do form a HLH structure, the loop region is unusually large containing approximately 40 residues (Hinsley et al,

2004) instead of the typical 5-15 residues (Ferre-D'Amare et al, 1993). A three- dimensional model has been solved by NMR stretoscopy using RD-5 of the murine

TFII-I. The structure of the RD does not resemble a typical HLH structure (Doi-

Katayama et ah 2007). It consists of four helices (helix 1-4), two anti-parallel beta strands (sheet 1-2), conferring an aaPaPa topology with a defined loop of two beta- turns between pi and a3. The structure has a hydrophobic core with an overall negative charge distribution on its surface. This structure is classified as a “new fold', not previously described in protein databases. Sequence alignment of all six repeat domains of TFII-I suggests that the other RDs have a similar fold to that of RD-5, although each

50 domain may have different surface electrostatic charge distribution caused by the non- conserved residues (Doi-Katayama et al, 2007).

All three genes encoding the TFII-1 family occupy adjacent locations in mammalian genomes (apart from GTF2IRD2 which is separated by the NCFI gene) and all show good conservation of synteny between mouse and human and all other placental mammals (unpublished data; Bayarsaihan et al, 2002). The adjacent locations and the high shared by the three genes, in particular, the approximately

70% homology shared by the RDs (Franke et al, 1999), strongly indicates that the genes have arisen by duplication and divergence from a common ancestor (Hinsley et al,

2004; Makeyev et al, 2004). Phylogenetic analysis suggests that the RDs are likely to have arisen by internal duplication; sequence information from Ciona intestinalis suggests that this family began as a single ancestral gene with a single RD (Palmer et al, unpublished). In teleost fish, a single gene with multiple RDs is evident and is most similar to GTF2IRD1. Subsequently, two genes corresponding to GTF2IRD1 and

GTF2I can be found in genomic sequence from Xenopus tropicalis. GTF21RD2 is not present in non-mammalian vertebrates or in marsupials such as the opossum

(unpublished data). It seems to have arisen in the eutherian (placental ) lineage

(Tipney et al, 2004). Therefore, GTF2IRD1 seems to be most closely related to the common ancestor which underwent gene duplication and divergence giving rise to

GTF2I, followed by GTF2IRD2 (Palmer et al, unpublished). High sequence homology of approximately 75% identity is shared between the two later evolved proteins, TFII-I and GTF2IRD2, particularly, at their N-termini, suggesting that GTF2IRD2 arose as a duplication of the GTF2I gene (Tipney et al, 2004).

Bioinformatics analysis also showed that the less conserved inter-repeat domain regions of TFII-I family members contain a number of predicted regulatory elements for each protein (Hinsley et al, 2004). For example, a PEST sequence which is rich in

(P), glutamic acid (E), serine (S) and threonine (T) and generally acts as a signal for

51 protein degradation (Rogers et al, 1986), as well as a SUMO (small ubiquitin-like

modifier) motif (see Chapter 5), have been identified between RD2 and RD3 of

GTF21RD1 and between RD1 and RD2 of TFI1-I and GTF2IRD2 (Hinsley et al, 2004).

1.3.2 TFII-I

TF1I-I which is also known by its accepted gene symbol as GTF21 (general transcription

factor 21), is a widely expressed protein and the first member of the TFII-I family

identified (Roy et al, 1991). Four alternatively spliced isoforms of TFII-I (a, (3, y, A)

have been characterised with subcellular localisation distributed between the nucleus

and cytoplasm; although, nuclear localisation is predominantly observed in

overexpression analyses (Cheriyath & Roy, 2000; Hakre et al, 2006). In murine

fibroblasts, TFII-I-A and -p isoforms have been shown to be primarily cytoplasmic and

nuclear, respectively, and the two have opposing activities on growth factor-induced

gene expression (Hakre et al, 2006). Dimerisation of the different isoforms has also

been demonstrated and it is postulated that the combinations of isoforms allow for a

wider range of physical and functional outcomes (Cheriyath & Roy, 2000). All isoforms contain six RDs, an N-terminal LZ motif, a basic region which could act as a DNA

binding domain, and a functional NLS (Figure 1.5, Cheriyath & Roy, 2001).

1.3.2.1 The proposed f unctions of TFII-I

TFII-I was initially isolated as a component of the basal transcriptional machinery that

binds to the core promoter initiator (Inr) element and an E-box upstream element to activate transcription (Roy, 1997; Roy et al, 1993b; Roy et al, 1991). However, the

striking properties of TFII-I shown by numerous subsequent reports are its active roles

in signal transduction pathways. TFII-I was also isolated as BAP-135, a substrate for

Bruton's tyrosine kinase (Btk), an important regulator in B cell differentiation, calcium

signalling and phospholipid metabolism (Yang & Desiderio, 1997). In resting B cells,

Btk binds and sequesters TFII-I in the cytoplasm, but upon ligand activation of the B cell receptor (BCR), TFII-I is tyrosine-phosphorylated by Btk and released for 52 translocation to the nucleus (Novina et al, 1999). The activity of TF1I-I is known to be regulated by phosphorylation at serine and tyrosine residues, mediated by multiple kinases upon activation of distinct signalling pathways further supporting its potential for multifunctional activities (Cheriyath et al, 2002; Kim & Cochran, 2001; Novina et al, 1998; Sacristan et al, 2009).

The roles of TFI1-1 in linking signal transduction events to nuclear gene transcription was first demonstrated by the positive regulation of c-fos by TFII-I upon addition of growth-promoting and mitogenic stimuli, such as serum, EGF (epidermal growth factor), PDGF (platelet-derived growth factor) and TPA (12-O-tetradecanoylphorbol

13-acetate) (Grueneberg et al, 1997; Kim et al, 1998; Kim & Cochran, 2000). As noted previously, c-fos is the prototype gene for immediate early genes (lEGs) which are silent at the basal level, but whose expression is rapidly activated in response to a wide variety of cellular stimuli (see section 1.1.6.2) (Cochran et al, 1984; Greenberg & Ziff,

1984; Kruijer et al, 1984). As regulators of and differentiation signals, c-fos and other IEGs are widely implicated as proto-oncogenes, but in neurons they regulate activity-dependent expression of genes important for neuronal plasticity (reviewed in

Loebrich & Nedivi, 2009; Okuno, 2010). The c-fos promoter harbors multiple binding sites for a number of proteins such as SRF (), STAT1, and

STAT3, which form complexes with TFII-I (Grueneberg et al, 1997; Kim et al, 1998).

The transcriptional activation of c-fos is also known to be dependent on the Ras/Rho signalling pathway (Hill & Treisman, 1995; Wang et al, 1998). The endogenous potent inhibitor of the Rho GTPases, pl90RhoGap, has also been shown to interact with TFII-I via an evolutionarily conserved FF domain thus sequestering TFII-I in the cytoplasm

(Jiang et al, 2005). However, upon PDGF receptor-mediated phosphorylation of the FF domain, TFII-I is discharged from pl90RhoGap and translocates to the nucleus where it can activate the transcription of c-fos or potentially, other inducible genes (Jiang et al,

2005). Therefore, a model was developed whereby nuclear translocation of TFII-1 is

53 regulated by distinct signalling mechanisms which typically involve its cytoplasmic sequesteration and phosphorylation (Roy, 2007).

In addition to c-fos regulation, TFII-I has also been reported as a transcriptional regulator of other genes in response to signal transduction events. In angiogenesis, TFII-

I binds to an Inr and three regulatory E-box elements in the TATA-less promoter of

VEGFR-2 (vascular endothelial growth factor receptor-2) (Jackson et al, 2005; Wu &

Patterson, 1999). This regulation was then shown to be antagonised by GATA2. The balance of TFII-EGATA2 regulation on VEGFR-2 is controlled by pl90RhoGap, whose activity is influenced by mechanical forces conveyed by the extracellular matrix

(Jackson et al, 2005). In the endoplasmic reticulum (ER)-induced pathway, Src- dependent phosphorylation of TFII-I at Y248 was shown to regulate transcription of

Grp78 (glucose regulated protein 78), a member of the family of stress-induced chaperones, via its ER-stress response element (Hong et al, 2005). In the cell cycle,

TFII-I is recruited to the core-promoter of cyc/in Dl and activates its gene transcription, which is required for the Gl-S phase transition (Desgranges et al, 2005).

Overexpression of TFII-I accelerates the progression of entry to and exit from S phase; whereas, upon cell cycle arrest induced by ionizing radiation which is known to activate , TFII-I is degraded by ubiquitiylation in a p53-dependent manner. These findings suggest that TFII-I acts as a mediator of cell division and proliferation in response to mitogen signalling (Desgranges & Roy, 2006).

TFII-I also has been shown to play a role in the cyclic guanosine 3,,5’-monophosphate

(cGMP)-dependent protein kinase I (PKGI) pathway, a soluble cytoplasmic serine/threonine kinase and one of the major mediators of nitric oxide (NO)/cGMP- triggered signal transduction (reviewed in Francis et al, 2010; Hofmann et al, 2009).

PKGI is widely expressed in vascular smooth muscle, endothelial and neuronal cells, where it has been implicated in myocardial contractility, platelet aggregation, endothelial cell permeability and synaptic plasticity by regulating downstream gene

54 expression both at the transcriptional and post-transcriptional levels (Feil et al, 2005a;

Feil et al, 2005b). Interaction of TFII-I with PKGI|3 leads to TFII-I phosphorylation at

Ser-371 and Ser-743, which results in the positive transactivation of c-fos (Casteel et al,

2002). More recently, PKGip was shown to regulate the BMP (bone morphogenetic protein) signalling pathway, a crucial regulator of joint tissue homeostasis, embryogenesis and CNS development (Liu & Niswander, 2005; Lories & Luyten, 2005;

Varga & Wrana, 2005). PKGip interacts and phosphorylates BMP receptor II, but upon

BMP-2 ligand stimulation, PKGip dissociates from the receptors and associates with

Smad complexes that allow its translocation into the nucleus where it recruits TFII-I for transcriptional regulation of BMP target genes (Schwappacher et al, 2009).

In addition to its gene transcriptional regulatory functions, TFII-I has been implicated in the regulation of agonist-induced calcium entry into the cytoplasm. Using a phosphopeptide that includes the Btk phosphorylation site of TFII-I, Caraveo et al.

(2006) identified phospholipase C-y (PLCy), an important enzymatic regulator of intracellular calcium level, as an interacting partner. Non-enzymatic activity of PLCy has been shown to act by binding to transient receptor potential 3 (TRPC3), a member of TRP ion channel family, via the split pleckstrin homology (PH) domains which enhances plasma membrane insertion of TRPC3 channel, thus increasing calcium influx

(van Rossum et al, 2005). A phosphotyrosine residue of TFII-I was found to bind to the

Src homology 2 (SH2) domain of PLCy and as well, a PH-like domain in TFII-I binds to the split PH domain of PLCy thus competing with TRPC3 for binding to PLCy.

Consequently, TFII-I reduces agonist-induced calcium entry by suppressing the surface accumulation of TRPC3 (Caraveo et al, 2006). As calcium signalling is involved in virtually all cellular processes, this finding opens a further potential physiological role for TFII-I.

Although studies of TFII-I have provided us with a wealth of data on the many different molecular pathways it may be involved in, the breadth of the findings create difficulties for a coherent molecular identity. A possible unifying hypothesis is that TFII-I seems to

55 act as a ‘scaffolding' transcription factor linking signalling molecules (such as PKGI[3, pl90RhoGap and BMP) and the coordination of multiple transcription factors (such as

SRF and STAT) which together, result in the signal-transduced gene regulations

(Hinsley et al, 2004).

1.3.2.2 Expression analysis ofTFII-I

Investigation of GTF2I expression at the transcript and protein level at various developmental stages suggests that its expression is maximal during embryonic development, but remains detectable throughout postnatal life and adulthood (Danoff et al, 2004; Enkhmandakh et al, 2004; Fijalkowska et al, 2010).

During the pre-implantation phase of murine embryogenesis, TFI1-I is detected in both the cytoplasm and nucleus of the zygote through to the blastocysts stage (E3.5-4.5), where it is also localised in the cytoplasm and nuclei of the inner-cell mass (ICM) and the external trophectoderm cells (Enkhmandakh et al. 2004). Post-implantation studies of mouse embryos showed that TFII-I is also continuously expressed from the beginning of organogenesis, from E8 to El6, with varying intensity and subcellular localisation at different stages (Fijalkowska et al, 2010). At E8, TFII-I is largely cytoplasmic in the ectoderm and mesoderm; whereas, at E9 a shift in localisation to the nucleus was detected in the neuroectoderm and the developing eye (the inner and outer retinal layer and lens primordium). During organ differentiation from Ell to E15, strong expression of TFII-I can be detected in the developing brain (both nuclear and cytoplasmic), heart (cytoplasmic), lung or airway epithelium (nuclear and cytoplasmic), and the gut villi (nuclear and cytoplasmic). It is also detected in the skin and hair follicles (cytoplasmic) and the submandibular gland (nuclear) (Fijalkowska et al, 2010).

During postnatal development of the mouse brain, in situ hybridisation and immunohistochemical analyses revealed TFII-I expression in marked regional centres

(Danoff et al, 2004). The highest expression was detected in the Purkinje cells of the cerebellum with immunoreactivity present in the cell body and nucleus as well as in the

56 dendrites extending into the molecular layer. In the hippocampus, strong expression was detected in both the pyramidal cells and interneurons of CA1-3 and the dentate gyrus, and similarly in the cell bodies and nuclei as well as in the processes of the neurons.

Using isoform-specific antibodies, the a, p, y isoforms were detected only in the interneurons, suggesting that the A isoform is predominantly expressed in the pyramidal cells. Weaker levels of TFII-I were also noted in the olfactory bulb and large neurons of the cortex (Danoff et al, 2004).

Therefore, the significant expression of TFII-I during early development and the neuronal-specific expression in the brain are consistent with a proposed role in the developmental and neurocognitive aspects of WBS.

1.3.2.3 Analyses of Gtf2i knockout mice

Three different Gtf2i mutant mouse lines have recently characterised. Two lines,

Gtf2iG,By8 (Enkhmandakh et al, 2009) and Gtf2i(,lBux (Sakurai et al, 2010) were generated by gene-trap mutagenesis, each through the insertion of a LacZ-neo cassette into intron 3 and immediately after exon 3 of Gtf2i, respectively. The reduction and total absence of TFII-I protein in the heterozygous and homozygous null mice, respectively, were demonstrated in the Gtf2iClBia line. In both lines, the homozygous knockout mice were embryonic lethal with apparent neural tube defects evident at

El0.5. However, the homozygous Gtf2iJ,Byg embryos showed more severe phenotypes, including severe bleeding (haemorrhage), significant growth retardation and dilated pericardial sacs, or arrested heart looping. The heterozygous mice from both lines were viable, but again, the heterozygous Gtf2i(,t(XE029)Byg mice were more severely affected. Ten percent were significantly smaller than the control mice; whereas, heterozygous Gtf2i(,,Bm mice were indistinguishable from their wildtype littermates.

Some of the heterozygous Gtf2i’uXh()29)Byg mice were also identified with craniofacial defects including defects in the cranial vault and some missing craniofacial bones.

57 The third mutant mouse line, Gtf2i"n (no laboratory code yet assigned), was generated by a targeted deletion of exon 2 (containing the start codon) through insertion of a PGK-neo cassette (Lucena et al, 2010). The gene mutation resulted in the translation of a truncated protein lacking the first 140 amino acids (A140GTF21) due to the alternative use of a second in-frame ATG within exon 5. The predominant nuclear localisation of the truncated TFII-I was not affected, but due to the loss of the N- terminal LZ domain, A140GTF2I was unable to homo-/heterodimerise. This further impacted on the ability to form higher-order protein complexes and consequently, there was a loss of some DNA regulatory activities, such as failure to drive the expression of cyclin Dl. The impact on the cell cycle was shown in MEF cells derived from the mutant mice which displayed a slower growth rate and re-entry into cell cycle when exposed to mitogenic stimuli. The overall phenotypic consequences of the N-terminal deletions in the mice were milder than those caused by the total knockout of the gene in the other two gene-trap mutant lines. The Gtf2i""( k’x2> homozygous mice survived; although, their birth rate was lower than the expected Mendelian frequency and only 8% developed normally to adulthood. A craniofacial dysmorphology was apparent in the

Gtf2i"n homozygous and heterozygous mice, with a short, symmetric snout and apparent midface hypoplasia (Lucena et al, 2010).

In an attempt to relate the neurological phenotypic consequences of TFII-I haploinsufficiency in mice with WBS, behavioural analyses were performed on heterozygous mice from the Gtf2iGlBwc and Gtf2i'm mouse lines. Heterozygous mice from both lines had no apparent differences in the gross morphology of the brain, but their behavioural profiles were quite distinct. Gtf2i"n<]ex2> heterozygotes showed a higher level of anxiety as indicated by decreased exploratory activity despite normal motor function and a low threshold for sound intolerance (Lucena et al, 2010). However, the most prominent behavioural feature of Gtf2iGtBux heterozygotes was altered social behaviour. The mice displayed increased social interaction with strangers and social

58 interest was retained by a lack of habituation to familiar partners. There were no differences in non-social anxiety observed in this line of mice (Sakurai et al, 2010).

Summary of Gtf2i knockout mouse models

Taken together, mouse models of Gtf2i mutations demonstrate an absolute requirement for TFII-I during embryonic development (Table 1.1), consistent with its spatial and temporal expression (see section 1.3.2.2). Mice with hemizygous deletions of Gtf2i are viable, but display neurological and behavioural changes (Table 1.1). The different phenotypic traits observed in the mouse models highlights the need to generate multiple mouse lines to unambiguously define the roles of TFII-I in cognition and behaviour.

However, since the full-length protein was absent in the Gtf2iCjlBia line; whereas, a truncated TFII-I protein was present in Gtf2i,m< ]ex2> mouse line, it can be argued that the former mouse line provides a more relevant model of the hemizygous gene deletion in

WBS. The increased sociability evident in the Gtf2iJtBux heterozygous mice suggests that GTF2I could be the gene at the telomeric end of WBS region which is responsible for the hypersociability in WBS individuals. No genes within the centromeric half of WBS region seems to be responsible for this phenotype as altered sociability was only displayed by the PD mouse (Limkl-Gt/2i deletion), and not DD mouse (Trim50-Limkl deletion) (Figure 1.3; Li et al, 2009).

1.3.3 GTF2IRD1

The protein encoded by GTF2IRD1 was initially called MusTRDl (muscle TFII-I repeat domain-containing protein 1) as it was first isolated in a yeast one-hybrid screen for factors that could bind to the DNA enhancer element of the human slow-muscle- fiber gene TroponinI slow (TNNI1) (O'Mahoney et al, 1998). Three independent yeast one-hybrid screens subsequently identified human, mouse, and frog homologs of the protein which were named GTF3 (Calvo et al, 2001), BEN (Bayarsaihan & Ruddle,

2000), and XWBSCR11 (Ring et al, 2002) and which bound to the upstream region of

TNNIf Hoxc8, and goosecoid, respectively.

59 GTF2IRD1 contains similar features to the TF1I-J proteins: an LZ domain at the N- terminus, five RDs (or six RDs in some mouse isoforms), and an NLS at the C-terminus

(Figure 1.5). Expression of recombinant GTF2IRD1 in mammalian cells reveals a predominantly nuclear localisation. The functional role of the LZ domain in homodimerisation of GTF2IRD1 has been demonstrated biochemically (unpublished data; Vullhorst & Buonanno, 2003). Furthermore, electrophoretic mobility shift assay

(EMSA) experiments revealed a higher-order, protein-DNA complex which was abolished when the LZ domain of the protein was deleted (Palmer et al, 2010). The RDs have mainly been implicated in DNA binding, each with variable affinity for specific

DNA sequences (Polly et al, 2003; Vullhorst & Buonanno, 2005). The NLS has been shown to be necessary for the nuclear localisation of GTF2IRD (Yan et al, 2000).

Eleven isoforms of GTF2IRD1 that arise from alternative splicing have been identified in the mouse (Tay et al, 2003) and at least two isoforms have been identified in human

(Franke et al, 1999; Osborne et al, 1999; Tassabehji et al, 1999a) with a possible third isoform indicated through database searches (Thompson et al, 2007). To date, the use of antibodies to detect endogenous GTF2IRD1 has been problematic and hence, there is no data to indicate whether protein isoforms exist.

1.3.3.1 Molecular analysis

DNA binding profile

A comparison of the DNA bait sequences bound by GTF2IRD1 in the four yeast one- hybrid screens mentioned above (Bayarsaihan & Ruddle, 2000; Calvo et al, 2001;

O'Mahoney et al, 1998; Ring et al, 2002), revealed a common core binding sequence of

GGATTA. This core recognition motif has then been verified by a number of DNA binding studies (Polly et al, 2003; Thompson et al, 2007; Vullhorst & Buonanno, 2003;

Vullhorst & Buonanno, 2005). In unbiased random oligonucleotide library screens

(SELEX procedure), the same core binding sequence or its inverted motif, TAATCC

60 has consistently been identified as the preferred target binding sequence of GTF2IRD1

(Thompson et al, 2007; Vullhorst & Buonanno, 2005).

Studies using a series of truncated GTF2IRD1 or GST fusion proteins containing individual RDs have indicated that RD2, 3, 4, and 5 all have DNA binding activities; although, RD3 shows little or no selectivity for DNA recognition motifs (Polly et al,

2003; Vullhorst & Buonanno, 2003; Vullhorst & Buonanno, 2005). The strongest DNA binding affinity has been demonstrated by RD4. This is consistent with a unique region of basic residues in RD4 not conserved in other RDs of GTF2IRD1 that could conceivably serve to stabilize the DNA-protein interaction by binding to the phosphate backbone of DNA. This region is partially conserved in RD2 (Hinsley et al, 2004), which has relatively strong affinity, albeit weaker than RD4, for the specific cognate

DNA binding sequence (Polly et al, 2003). DNA binding activity of RD1 has not been observed in any of these studies.

Transcriptional repression has emerged as a dominant feature of GTF2IRD1 from transactivation assays and transgenic mice. Transactivation assays reveal that

GTF2IRD1 mediates a strong level of transcriptional repression on the expression of

GGATTA-containing reporter plasmids and mutations of the motif are sufficient to abrogate the repression by GTF2IRD1 (Palmer et al, 2010; Thompson et al, 2007).

Although GTF2IRD1 protein has been implicated in the regulation of the genes it was initially associated with in the yeast one-hybrid screens: TNNll (Calvo et al, 2001;

O'Mahoney et al, 1998), HoxcS (Bayarsaihan & Ruddle, 2000), and goosecoid (Ring et al, 2002), these findings lack supporting evidence from in vivo data. The bait sequences used in all of the yeast one-hybrid studies contain trimerised binding motifs for

GTF2IRD1 which is typically done to maximize the chance of trapping potential prey proteins, but could conceivably result in an artificial increase in affinity of GTF21RD1 binding. EMSA analysis using in vitro translated GTF2IRD1 and a single GGATTA

DNA probe failed to detect a GTF2IRDEDNA complex, compared to those formed by

61 GTF2IRD1 with probes containing dimerised or trimerised recognition sequences

(Palmer et al, 2010). Therefore, future studies need to address whether the affinity of

GTF21RD1 for the single GGATTA site present in the proposed targets is sufficient to be of biological importance.

Negative autoregulation of GTF2IRD1

Despite all of the proposed target genes for GTF2IRD1, the only bona fide gene target for GTF2IRD1, as shown by in vivo data (Palmer et al, 2010), is GTF2IRD1. The region immediately upstream of the transcription start site in mouse and human is highly conserved. The GTF2IRD1 upstream region (GUR), defined as a 104-bp region sharing approximately 82% sequence homology between vertebrate species including fish, frog, mouse and human, contains three copies of the GTF2IRD1 recognition sequence

(GGATTA), with 100% identity conserved between all species. EMSA experiments show that GTF2IRD1 binds to the GUR sequence with high affinity and provides a working model for the GTF2IRD1 protein:DNA interaction, whereby at least two

GGATTA recognition sequences, no further than 57 bp apart, are required for detectable binding affinity (Figure 1.5.C; Palmer et al, 2010).

Luciferase transactivation assays demonstrate that GTF2IRD1 can repress the expression of the GUR in a reporter construct, thus supporting negative autoregulation of its own gene. This autoregulatory feedback mechanism leads to an increased level of transcript in Gtf2irdl knockout mice and dosage compensation of GTF21RD1 transcript in lymphoblastoid cell lines derived from WBS patients (Palmer et al, 2010). The latter finding is consistent with previous gene expression studies of individuals with the typical WBS deletion (Antonell et al, 2010; Collette et al, 2009; Merla et al, 2006).

Most genes that are hemizygously deleted in WBS have an average relative expression level of approximately half that of controls as expected, but GTF2IRD1 showed no significant change in average relative expression level between control and patient samples. These findings question whether there is haploinsufficiency of GTF2IRD1

62 protein and further call into question its role in the neurocognitive pathology of WBS.

However, whether the same feedback mechanism occurs in the brains of WBS patients is unknown and cannot be assessed. A similar gene expression analysis done on fibroblast skin cells of WBS patients showed a significant reduction of GTF2IRD1 transcript (Merla et al, 2006) and therefore, the degree of autoregulation would seem to be vary depending on cell type.

Interacting protein partners of GTF2IRD1

The identification of GTF2IRD1 interacting partners is very limited. Currently, only three proteins have been shown to associate with GTF2IRD1: retinoblastoma (Yan et al,

2000), PIASxP (Tussie-Luna et al, 2002b), and HDAC3 (Tussie-Luna et al, 2002a).

The relevance of the interactions to the function of GTF2IRD1 is not clear. GTF21RD1 was independently isolated as a retinoblastoma (Rb)-interacting protein, an important regulator of vertebrate cell cycle and development, but the functional importance of this interaction remains unknown (Yan et al, 2000). In yeast two-hybrid screens, PIASxP was identified as a common interacting partner of both TFII-I and GTF2IRD1 (Tussie-

Luna et al, 2002b). PIASxP appeared to augment the impact of TFII-I on c-fos transcription while alleviating the repressive effect of GTF2IRDI on TFII-I transcriptional regulation. However, it was also noted that PIASxP alone mediates the activation of c-fos making these findings ambiguous. Similarly, the histone deacetylase

HDAC3, which was initially identified as a TFII-I interacting partner (Wen et al, 2003), also interacted with GTF2IRD1 as shown by a co-immunoprecipitation assay (Tussie-

Luna et al, 2002a). Although the intrinsic activity of HDACs in gene transcriptional repression through chromatin remodelling fits with the observed properties of

GTF2IRD1, further evidence supporting this interaction, such as mapping the regions of interaction between the two proteins, has not been demonstrated.

63 1.3.3.2 Expression analysis

Northern blot analyses revealed that GTF2IRD1 is widely expressed in mouse and human tissues (Bayarsaihan & Ruddle, 2000; Calvo et al, 2001; Franke et al, 1999;

Osborne et al, 1999; Tassabehji et al, 1999a). Similar to the expression pattern of

GTF2I in mice, higher levels of expression were detected in developing embryonic tissues compared to adult tissues, but in general the abundance of Gtf2irdl transcript is lower than that of Gtf2i.

The expression of Gtf2irdl during embryonic development and in adult tissues has been extensively analysed using a LacZ reporter knockout/knockin mouse line,

Gtf2irdl,m2(LacZ>Hrd (Palmer et al, 2007). During organogenesis, regionalised expression of the LacZ reporter was detected, most notably in the forebrain and the gut of the lower trunk of embryos at E9.5, and in the heart and tissues around the umbilical region at

E10.5.

In the developing brain (between El3.5 to 18.5), the expression was strongest in the pituitary, the thalamic and hypothalamic nuclei, the roof of the midbrain and hindbrain, and a diffuse low level signal was also found throughout most parts of the brain. In the adult mouse brain, strong expression was detected in the Purkinje cell layer of the cerebellum, the mitral and granule cell layer of the olfactory bulb, and neurons in the piriform cortex (Figure 1.6.A; Howard et al, submitted). Weaker expression was detected in other regions including the cortex, caudate putamen (striatum), the amygdala, the septum, the pallidum, the thalamic nuclei, superior and inferior colliculi, substantia nigra and the pons.

Substantial overlap between p-galactosidase activity and GAD67-expressing cells suggested a predominance of Gtflirdl expression in the inhibitory GABAergic neurons.

GAD67, which is encoded by the GAD1 gene, is one of the two isoforms of glutamic acid decarboxylase (GAD), the rate limiting enzyme for GABA synthesis. This

64 observation was confirmed by analysis of brains from a cross between Gtf2irdl"n2(lMc/)lln1 and Qa(jj"n,

Gtf2irdl/LacZ-expressing GAD67-GFP positive neurons were mapped to the Purkinje neurons in the cerebellum, granule cells in the olfactory bulb, the reticular thalamic nucleus, caudate putamen, globus pallidus, the lateral septum, the intercalated masses and the central nucleus of the amygdala, the inferior and superior colliculi and the substantia nigra (Figure 1.6.B). Dysregulation of the GABAergic system in the

Gtf2irdl-expressing GABAergic neurons may underpin some of the phenotypic consequences seen in Gtf2irdltmll,nl knockout mice and hence, WBS individuals (see next section). Non-GABAergic neurons that express Gtf2irdl include the glutamatergic mitral cells of the olfactory bulb and pyramidal cells of layer 11 of the piriform cortex

(Howard et al, submitted).

In developing and adult peripheral and sensory nerves, expression of the LacZ reporter was also detected, such as the dorsal root ganglia, the olfactory epithelium, the retina of the eye, and the cochlea of the inner ear. Consistent with other reports, GtfUrdl, along with Gtf2i and Gtf2ird2, are all expressed in the developing tooth buds (Ohazama &

Sharpe, 2007).

In skeletal muscle, Gtf2irdl expression is present during myogenesis (fusion of myoblasts into multi-nucleated myotubes or fibres), but is then downregulated in the mature fibres of adult muscle (Palmer et al, 2007). The decrease in expression as development proceeds is consistent with previous reports (Calvo et al, 2001)

Intense staining was also detected in the adipocytes of brown adipose tissue (BAT), beginning from El7.5. This correlates with the onset of mitochondrial biogenesis in

BAT. Significant expression persists during adulthood in the interscapular region.

65 It is noteworthy that another report based on immunohistochemical analysis using an anti-GTF2IRDl antibody has demonstrated a reasonable overlap in findings compared with the pattern shown by the LacZ knockin mice, particularly, in the developing nervous system and developing teeth (Bayarsaihan et al, 2003). However, the lack of specificity of the antibody as shown by the multiple bands detected in Western blot analysis may explain the many discrepancies between the two studies. For example, immunohistochemical analysis detected high expression of GTF2IRD1 in the liver, but the LacZ reporter activity was barely detected in the liver of Gtf2irdltm2(Lac/>Hn1 mice.

Results from the LacZ-based analysis of Gtf2irdl expression are consistent with in situ hybridisation data available from the atlas of mouse gene expression patterns

(www.genepaint.org).

1.3.3.3 Analyses of Gtf2irdl mouse models

As with Gtf2i, there are several knockout mouse models of Gtf2irdl, each harbouring different genetic mutations, and having different phenotypic traits.

Insertional knockout mouse model The first reported mouse model, Gtf2irdlT8l66'8Ssi, was identified through a fortuitous c- transgene insertion which induced a ~40-kb genomic deletion, starting downstream from the CLIP2 gene up to and including the first exon of Gtf2irdl (Durkin et al, 2001; Tassabehji et al, 2005). Disruption of the 5' end of Gtf2irdl was shown to significantly reduce its mRNA level in the transgenic mice, while the expression of

CLIP2 was unaffected. The homozygous mice were viable and had a normal lifespan, except they were prone to liver-specific carcinomas due to overexpression of the c-myc transgene (Durkin et al, 2001). Tassabehji et al. (2005) reported that all Gtf2irdl null mice from this line showed phenotypes relevant to WBS, including growth retardation and craniofacial abnormalities which involve periorbital fullness and a short snout.

Homozygous null mice were also shown to have increased brain ventricle volume, but

66 extensive behavioural analysis was not performed on these mice (van Hagen et ah

2007) .

Targeted knockout mouse models

Two other Gtf2irdl knockout mice, CD\x[\CRx\29]-Gtf2irdJlml 1,0 (Young et al, 2008) and C57BL/6-Gtf2irdl'm,Hrd (Palmer et ah 2010), were generated by targeting exon 2

(containing the start codon) of Gtf2irdl. During the characterisation of both mouse lines, analysis of transcript production indicated that both mutant alleles generated an altered transcript that would theoretically not be capable of producing GTF2IRD1 protein because the initial ATG is out of frame. However, to check this rigorously, mammalian expression plasmids were generated containing the cDNA equivalents of wild type Gtf2irdl mRNA or the mRNA generated by the Gtf2irdl'm,Hnl allele and these were transfected into COS7 cells. Analysis of protein extracts from these transfected cells showed that translation of GTF2IRD1 with an N-terminal truncation was possible at a low efficiency (3%), as a result of initiation from the second ATG (Figure 1.6.C;

Palmer et ah 2010). Since the autoregulatory mechanism controlling Gtf2irdl gene expression (see section 1.3.3.1) would result in a doubling of the output from the mutant allele in homozygous mutant mice (Figure 1.6.C.a), the truncated peptide is estimated to be present at 6% of normal GTF2IRD1 protein levels in homozygous knockout mouse tissues. Homozygous mice from both Gt/2irdl knockout targeted lines were viable, but had mild reductions in body weight. Both were physically indistinguishable from their wildtype littermates and neither displayed the craniofacial abnormalities described in the Gtf2irdlT8(A,bl~xlyL>,66'SSsl homozygotes (Palmer et ah 2010; Young et ah

2008) . The gross brain morphology of both mice appeared normal although the two lines displayed specific behavioural features.

The Gtf2irdltm,Lw mutant mice had reduced anxiety and aggression, and engaged in more social interactions, which points to less inhibited behaviours in social and non­ social contexts (Young et ah 2008). Neurochemical analysis of the brain tissues of the

67 null mice identified a significant elevation of serotonin metabolites, 5- hydroxyindoleacetic acid (5-HIAA) in the prefrontal cortex; although, the level of serotonin (5-HT) itself was unaffected. Electrophysiological analysis showed that 5-HT elicited a larger inhibitory outward current in layer V pyramidal neurons in the prefrontal cortex of Gtf2irdl""'iro homozygous null mice than in the wildtype controls

(Proulx et al, 2010). The differential response was shown to be specifically mediated through the 5-HTia receptor subtype, although the mechanistic role of GTF2IRD1 in this system is unknown as no changes in expression level of 5-HT receptors or the 5-

HT-metabolising enzyme - monoamine oxidase (MAO-A) were detected (Proulx et al,

2010). It is widely known that the serotonergic pathway plays an important role in mood regulation and has been a major target for the treatment of depression and anxiety disorders (Fuller, 1991; Graeff et al, 1996; Fesch, 2005). For example, 5-HTia receptor knockout mice have been shown to exhibit greater anxiety (Heisler et al, 1998; Parks et al, 1998; Ramboz et al, 1998); whereas, overexpression of the receptor results in decreased anxiety in mice (Kusserow et al, 2004). Therefore, dysregulation of the serotonergic pathway in the prefrontal cortex is an appealing potential cellular mechanism that could contribute to the anxiety-related features of WBS.

Despite the similar genetic deletion strategies used in the Gtf2irdllm,L,° and

Gtf2irdl"nlHn/ mouse lines, extensive behavioural analyses of the Gtf2irdl"",l!u/ knockout mice revealed several different phenotypes (Howard et al, submitted). These knockout mice displayed slight abnormalities in their social interaction, but no changes to the non-social anxiety were evident in tests for global non-social anxiety.

Homozygous null mice displayed a motor control and coordination deficit and increased hyperactivity during the dark cycle, which could be related to the ADHD symptoms in

WBS. Most strikingly, a novel audible vocalisation was emitted by the homozygous null mice in response to water immersion and other anxiety-provoking stimuli, such as physical restraint. In response to the stressful stimuli, all mice regardless of their genotype produced normal audible and/or ultrasonic vocalisation (USV). However, in

68 each context Gtf2irdl"",Hni null mice emitted significantly more normal audible and

USV vocalisations, as well as the novel audible vocalisation, which has not been previously described in wildtype mice (Howard et al, submitted). These findings led to an interpretation of the vocalisations as a response to selective stressful stimuli, or environment-provoked anxiety, as no changes to the basal level of non-social anxiety were detected. This phenotype could be related to the excessive worrying and specific phobias often observed in WBS patients (Dykens, 2003; Einfeld et al, 1997).

Furthermore, pharmacological approaches used to investigate the function of the

GABAergic system revealed that the mice responded differently to GABA antagonist and agonist drugs (Howard et al, submitted). GABA is the major inhibitory neurotransmitter in the brain which has been strongly implicated in anxiety and neurological disorders involving disruptions in the excitation-inhibition balance

(reviewed in section 1.1.3). Therefore, the aberrant GABAergic activity in specific regions of the brain is thought to relate to the differences in the provoked-anxiety response of the mice (Howard et al, submitted).

Gene trap knockout mouse model The fourth mouse model generated through a gene trap mutagenesis screen, Gtf2irdlc'lBy8, was characterised alongside the Qtj2i(''

(Enkhmandakh et al, 2009). This mutant mouse line carried an insertion of the LacZ- neo cassette in intron 22 of Gtf2irdl. As with the Gtf2i gene trap model, this mouse line displayed more severe phenotypes than the other three Gtfiirdl mouse models described above. Homozygous null mice were embryonic lethal with hemorrhages and severe delayed development apparent from E9.5. The heterozygous mouse displayed growth retardation and a subset of animals demonstrated craniofacial and severe skeletal abnormalities, such as kyphosis, a pronounced curvature of the spine (Enkhmandakh et al, 2009). It is possible that the severity of the gene trap lines could result from a dominant-negative effect of the p-galactosidase-fused peptide. This peptide was predicted to disable the translocation of GTF2IRD1 to the nucleus due to the insertion

69 of the cassette upstream of the NLS. Although the nuclear function of GTF2IRD1 as a transcription factor is generally presumed, its potential activity in the cytoplasm is also possible as it is known that TF11-I plays a role in the cytoplasm, independent of its gene transcriptional function (see section 1.3.2.1; Caraveo et al, 2006). It is possible that the presence of the mutant peptide in the cytoplasm could cause an off target effect.

Summary of Gtf2irdl knockout mouse models

The impact of haploinsufficiency of GTF2IRD1 in WBS using single gene deletion mouse models has been complicated by the lack of consistencies found in the different mouse models which are probably caused by the different genetic background of the mice, and more importantly, the precise nature of the genetic mutation and the biochemical consequences of each mutant (Osborne, 2010). In theory, the phenotypic features of the monogenic Gtjlirdl heterozygous knockout mice should be limited to or milder than the phenotypic range displayed by the PD mice which involved heterozygous chromosomal deletion from Litnkl to Gtf2i (Li et al, 2009). Overall, some conclusions can be drawn on the Gtf2irdl genotype-phenotype correlation based on the

Gtf2irdllmlUo and Gt/2irdl""lllll/ mouse lines which showed some consistent phenotypic features with the PD mice. In reference to these three mouse lines, Gtf2irdl heterozygosity results in the mild growth retardation, but not the craniofacial abnormalities or the learning and memory deficits observed in WBS, which seem to require the deletion of gene(s) at the centromeric end of WBS region, as these phenotypes were found in the DD mouse line but not in the PD mice (Table 1.1). The motor coordination deficit and increased level of non-social anxiety displayed by the

PD mouse line can be explained by Gtf2irdl, as these phenotypes were shown in

Gtf2irdllmlllnl mouse line, although, to varying degrees. There was no motor deficit and the reverse - reduced level of anxiety displayed by the Gtf2irdl""l,J0 mouse line, which could indicate mouse strain-specific effects, as both PD and in Gtj2irdl"n,,,ni lines were maintained on the C57BL/6, whereas Gtf2irdl,mlL>0 line is on an outbred background.

70 CLIP2, which sits within the PD interval, has also been implicated in the motor coordination dysfunction (Hoogenraad et al, 2002; van Hagen el al, 2007).

In addition, the social disinhibition in the PD mouse line and in WBS may only be partially caused by Gtf2irdl mutation as it has only been observed in the Gtf2irdl"nlUo mouse line. Hemizygous deletion of Gtf2i is highly likely to contribute to this phenotype as Gtf2iGtBm heterozygous mice have been described with an enhanced social behaviour and a lack of habituation to repeated social stimuli, reminiscent to those in

WBS individuals (Sakurai et al, 2010). Therefore, the two members of the GTF2I family may synergistically regulate the neurological mechanism of social inhibition.

Other informative mouse models

In view of the initial discovery of GTF2IRD1 as a regulator of muscle-specific genes, the role of GTF2IRD1 in skeletal muscle tissue was investigated using a transgenic mouse line which expresses human GTF2IRD1 specifically in the skeletal muscle (Issa et al, 2006). Analogous to neuronal specification during brain development, proper specification of different muscle fibre types is important for functional skeletal muscle in adult life. Fibre type determination is thought to occur through a combination of intrinsic genetic programming (DiMario & Stockdale, 1997; Pin et al, 2002) and external workload and neuronal stimulation (reviewed in Pette, 2001). GTF2IRD1 transgene expression resulted in normal fetal muscle fibre patterning despite expression of the transgene during early muscle differentiation; however, the mice displayed progressive loss of slow fibres at the postnatal and adult stage with an abnormal increased conversion into fast fibres (Issa et al, 2006). Molecular analysis showed that many slow fibre-specific genes were repressed in the transgenic mice; whereas, fast fibre-specific gene expression was either unaffected or enhanced. This observation fits with previous findings of GTF2IRD1 as a protein that has transcriptional repressive capabilities (Calvo et ah 2001; O'Mahoney et al, 1998; Polly et ah 2003). However, the molecular mechanism of how GTF2IRD1 regulates slow fibre-specific gene expression

71 is not fully understood. Analysis of the Gtf2irdl,mlllul knockout mice did not reveal any abnormal muscle fibre type patterns at any stage examined (Palmer et al, unpublished data). This may relate to the fact that the normal level of GTF2IRD1 in adult mouse skeletal muscle is very low (Calvo et al, 2001; Tay et al, 2003), hence, abolishing its expression would not cause any significant consequences. Therefore, the fibre type shift seen in adult transgenics may be due to ectopic expression of GTF21RD1 at a time when it is normally absent.

1.3.4 GTF21RD2

1.3.4.1 Genomic sequence and transcript analysis

GTF21RD2 is believed to be the most recent member of the GTF21 gene family (see section 1.3.1). Its involvement in WBS is complex due to the multiple copies of

GTF21RD2 genes located within the LCRs (Bayes et al, 2003). With the lack of LCRs in the mouse genome, only one Gtf2ird2 gene is present which is highly homologous to the human ortholog. In human, two GTF21RD2 genes are believed to be transcribed since cDNA clones from each variant have been identified (Makeyev et al, 2004;

Tipney et al, 2004). The two gene loci GTF21RD2/a and GTF2IRD2/3, map to the medial- and telomeric-B blocks, respectively (Figure 1.3). Despite the fact that the great majority of patients have the common 1.55 Mb deletion, there are multiple recombination hotspots within the B-block. Variations in the precise breakpoint within the medial B-block determine whether the two genes, NCFI and GTF21RD2, are deleted or retained. High-resolution genetic mapping studies estimated that one third of the typical WBS patients lose GTF21RD2, while two thirds retain this gene (Bayes et al,

2003; Del Campo et al, 2006). Moreover, a larger 1.84 Mb deletion which extends to

GTF21RD2 in the telomeric-B block has been identified in a smaller number of WBS patients (Bayes et al, 2003; Tipney et al, 2004). Hence, the protein level of GTF2IRD2 in WBS patients needs to be closely investigated in order to pinpoint the potential contribution of GTF2IRD2 to the pathology of WBS. In addition, a pseudogene copy termed GTF21RD2P which contains a frameshift mutation has been identified from 72 human genomic sequence and is located within the centromeric B block (Tipney et al,

2004). A pseudogene of GTF21, GTF21P, is also present within the same block (Perez

Jurado et al, 1998).

RT-PCR analysis revealed a broad tissue expression pattern of Gt/2ird2 in the mouse, but relatively strong expression was detected in heart, brain and liver by Northern blotting (Tipney et al, 2004). Differential expression levels of GTF2IRD2/a and

GTF2IRD2/3 in humans have not been determined.

1.3.4.2 Protein analysis

As a member of the TFII-1 protein family, GTF21RD2 has an LZ domain at the N- terminus, but contains only two RDs. In addition, a CHARLIE8 transposon element constitutes the C-terminal half of the protein (Tipney et al, 2004). The two RDs in

GTF2IRD2 are most similar to RD1 and RD6 of TFII-1 (Makeyev et al, 2004). The

CHARLIE8 domain is unlikely to retain any transposase activity, but the high sequence conservation of this domain in the human and mouse suggests that the element plays an important role in the function of the protein (Tipney et al, 2004).

Immunofluorescence analysis revealed that the sub-cellular distribution of GTF2IRD2 is distinct from GTF2IRD1 and TFII-I (Palmer et al., manuscript in preparation). When expressed in cell lines, GTF2IRD2 localises to distinct nuclear foci and microtubules in the cytoplasm, but a low diffuse signal can be detected in the nucleoplasm, unlike

GTF2IRD1 and TFII-I. The identity of the nuclear puncta does not overlap with any of known nuclear proteins found in speckles. The overall distribution of GTF2IRD2 requires both the N-terminal and C-terminal domains of the protein. Transfection of the

C-terminally truncated protein shifted its localisation to a broad cytoplasmic pattern; whereas, transfection of the CHARLIE8 domain alone localised exclusively to small number of large spherical nuclear bodies (Palmer et al, manuscript in preparation).

73 GTF2IRD1 transgene expression in mouse muscle shifted the fibre type from slow to fast; however, the Gtf2ird2 transgenic mouse displayed the opposite phenotype with profound shifts in fibre type from fast to slow (Issa et al, 2006). The antagonistic actions of GTF2IRD1 and GTF2IRD2 in vivo is in line with the co-localisation and co- immunoprecipitation of the two proteins when co-transfected in cell lines (Palmer et al, manuscript in preparation).

1.4 Aims of this thesis

Evidence from the mouse models supports a role for GTF2IRD1 and TFII-I dysregulation in some of the neurological and physical defects in WBS. However, a poor understanding of their precise function, especially GTF2IRD1, at the molecular level is preventing progress in determining its mechanistic role in the CNS. It was the overall aim of this thesis to improve the current knowledge of the molecular and cellular mechanisms of GTF2IRD1 function. The specific aims of this study, utilising the Gtf2irdl knockout mouse and in vitro biochemical assays, were:

I. To investigate the role of GTF2IRD1 in regulating gene transcription in the

mouse brain (Chapter 3);

II. To screen for novel interacting protein partners of GTF2IRD1 (Chapter 4);

III. To characterise a novel post-translational mechanism, SUMOylation of

GTF2IRD1, at the biochemical level and its potential function at the cellular

level (Chapter 5).

74 Figure 1.1. The specialised functions of the brain are underscored by complex networks of highly responsive nerve cells, the neurons.

(A-C) Gross anatomy of the human brain (adapted from Purves et al, 2004). (A) The forebrain includes the cerebral cortex, the limbic system, and the basal ganglia. The cerebral hemisphere which makes up the largest proportion of the brain is characterised by patterns of sulci (depressions or fissures) and gyri (convolutions) on its surface. (B) Midsagittal view of the forebrain, which exposes the limbic system at the centre of the cerebral hemisphere, and the brainstem. (C) The four lobes that make up the major divisions of the cerebral cortex are shown from the exterior and midsagittal view. (D) The two main classes of neurons in the cerebral cortex according to the neurotransmitter they release, glutamate and GABA (y-aminobutyric acid) (adapted from Jones, 1981). Excitatory glutamatergic neurons (left) primarily include the pyramidal neurons with their highly characterised spiny dendrites and typically long extended axons spanning different layers of the cortex. Also shown is a non-pyramidal excitatory stellate cell (a). In contrast, the inhibitory GABAergic neurons (right) are typically characterised by smooth (non-spiny) dendrites, and according to their morphology, can be divided into a number of subclasses: local axon arcades cell (b), double bouquet cell (c), basket cells (d, h), chandelier cells (e), bitufted cells (f) and the neurogliaform cell (g). (E) Schematic diagram illustrating the process of synaptic transmission which takes place in the synapse: (1) synthesis of neurotransmitter in the soma of the presynaptic neuron, (2) storage in a synaptic vesicle in the presynaptic terminal, (3) release into the synaptic cleft upon stimidation or action potential firing, (4) interaction with the target receptors on the postsynaptic neuron, and (5) rapid termination of synaptic action by diffusion, enzymatic degradation, or re-uptake of the transmitter (adaptedfrom URL: http://quest.nasa.gov/neuron). B

Cingulate Central Precentral Postcentral sulcus sulcus gyrus Diencephalon gyrus Parieto­ Parieto­ occipital sulcus occipital Cingulate sulcus gvrus

Calcarine Corpus sulcus callosum

Lateral (Sylvian) fissure Anterior commissure Preoccipital f Midbrain notch Brainstem Brainstem -j Pons Spinal cord L Medulla

E Frontal Parietal lobe lobe Myelin

Occipital Temporal lobe lobe Synaptic- Frontal Parietal vesicle lobe lobe Transmitter 1$ molecules Across

Ions Transmitter Postsynaptic Occipital receptor current flow Temporal lobe lobe

Excitatory glutamatergic Inhibitory GABAergic neurons neurons

GLU Figure 1.2. Epigenetic mechanisms targeting DNA and histones regulate chromatin remodelling.

(A) DNA methylation (reprinted from Day and Sweatt, 2010): (a) The wrapping of DNA around an octamer of highly basic histone proteins forms chromatin, the organisation of genetic material inside a eukaryotic ; (b) DNA methyltransferase (DNMT) catalyses the transfer of a methyl group at the 5 'position ofpyrimidine ring of a cytosine base, and can be categorised into two groups: (c) De novo methyltransferase (dnDNMT), such as DNMT3A and DNMT3B, which initiate methylation on previously non-methylated cytosines, and maintenance of methyltransferase (mDNMT), such as DNMT1, which methylates hemi- methylated DNA at the complementary strand. (B) Histone modifications (reprinted from Franklin and Mansuy, 2010). Schematic diagram of four core histones, H2A, H2B, H3, and H4, showing the major post-translational modifications targeting amino acid residues particularly at their N-terminal tails. Methylation (M) can be either mono-, di- or tri-methv/ation. A

DNA methylation Cytosine Histone nh2 octamer DNMT

Methylated ^ DNA

Histone 5' —CGCGCG-- 3' tails 3' — GCGCGC— 5'

5' —CGCGCG— 3' 3' —GCGCGC— 5' mDNMT

De novo Maintenance methylation methylation

B Centromeric Medial Telomeric LCR LCR LCR Common 1.55 Mb deletion/duplication

BAZlB VPS37D WBSCR27 WBSCR16

FKBP6 BCL7B CLDN4I WBSCR28 RFC2 STX1A UMK1 CEN — — TEL FZD9 CUP2

CLDN3 MIR590 GTF2SRD2 ABHD11 BF4H NCFl / WBSCR22

I >T >■» > ^—< K I WO 0 WHW POM121 NSUNS\ FKBP6 GTF2I | GTF2IRD2 WBSCR13L* ] PMS2L' TRIMSO NCF1 STAG3L* The Williams-Beuren syndrome region on chromosome 7q11.23 Expert Reviews in Molecular Medicine 2007 Published by Cambridge University Press

Figure 1.3. The WBS deletion region on human chromosome 7q 11.23.

Schematic diagram shows the presence of highly homologous blocks of low copy repeats (LCRs): A, B, and C, at the centromeric, medial and telomeric regions which predispose the region to genomic rearrangement during meiosis (adapted from Osborne and Mervis, 2007). Unequal homologous recombination mediated by misalignment of the centromeric- and medial-B blocks results in the common 1.55 Mb deletion or potentially, duplication of the region. Genes and their corresponding orientation within this deleted region are indicated by the green arrows. NCFl and GTF21RD2 in blue are variably deleted depending on the location of the telomeric breakpoint. Figure 1.4. Distinguishing features ofWBS individuals.

(A) The friendly faces of WBS children (reprinted from Jarvinen-Pasley, et al, 2008). The facial characterisation typically includes broad brow, full nasal tip and flat nasal bridge, full cheeks, full lips and a wide smile. (B) The severe spatial construction deficit in WBS, distinct to that in Down syndrome, is visually conveyed in both drawing and block design tasks (reprinted from Bellugi et al, 1999). WBS individuals focus on the extreme details at the expense of the whole picture; whereas, DS subjects are able to construct the global configuration, but with poor internal details. (C) Developmental trajectories ofWBS compared to DS in three cognitive domains: language, spatial cognition and face processing (adapted from Bellugi et al, 1999). While all three trajectories show a uniform delay in DS, distinctly different trajectories are displayed in WBS. The greatest delay in the WBS cognitive profile is within their spatial processing, but much better language and facial recognition processing is eventually acquired. B Williams syndrome Down syndrome (poor on global organization) (poor on internal detail)

living roof two-storey Free house door A windows drawing ffl BB aooo V&ZS doordoc , of a house swimmingsvvimumiu uwupool. n windows Example sidewalk

^gn ^ 1 $ % esaa Model

yyyy Local- y y r7 global J y Uubtib%nu A task yyyt Model V

C Williams syndrome Down syndrome

Faces 14 n Vocabulary uT cu ^>; 10 1 Faces 03CD 03 Spatial Spatial Vocabulary 03 3 6- C 0)

7 11 15 19 23 Age at test (years) Age at test (years) A

GTF2IRD1 RD1 RD2 RD3 RD4 RD5

TFII-I RD1 RD2 RD3 RD4 RD5 RD6

GTF2IRD2 RD1 RD2 CHARLIE8 TRANSPOSON

B

0VLY3gj ?GRAS vfflpflgYE RL LRES Jaf Jaeydpr EHSHRjj NSRYAj ?GL D HMHeByR RI AC D jjPF jCTfGVP EERHsf JFNTRYAi IGI3 E fBfTOy 3 RF LMH fjSL 5NC|GIAK" iHEASNSl SfNRRYGE JGI RYPHcByR RI R S N iiPFj| IctIgsqnE VAD sQggEgFNERYG ?GL NRPgl^YR LI RD 3 gD up F S "nt|dih RARE c

RD

TAATCC ATTAGG CCTAAT CCTAAT Monomer

TAATCC GGATTA GGATTA ATTAGG CCTAAT CCTAAT Dimer

Figure 1.5. Structural features of the TFII-I family members and working model of DNA binding mediated by the repeat domains of GTF2IRDI.

(A) Primary structures of the TFII-I family members GTF21RD1, TFII-I, and GTF2IRD2. Common features of all proteins include a leucine zipper (LZ) at their N-terminal end and multiple I-repeat domains (RDs). A CHARLIE8 transposon element has inserted at the C-terminal region of GTF2IRD2, the most recently evolved member of the family. (B) Amino acid alignment of the five repeat domains (RDs) in human GTF2IRD1 shows high sequence homology shared by the repeats. See Appendix A for the full protein sequence of GTF2IRD1. (C) Model of the direct DNA binding mechanism of GTF2IRD1 proposed in the study of its autoregulatory feedback mechanism (adapted from Palmer et al, 2010). The GTF2IRD1 upstream region contains three GGATTA recognition sequences whereby high affinity binding is formed by the simultaneous interactions of the RDs of the proteins with at least two closely positioned and same-oriented GGATTA sites. Formation of the dimer is dependent on the presence of the LZ domain. Figure 1.6. Brain expression mapping, transcript and protein analysis of Gtf2irdl in the mutant mouse.

(A) Detection of p-galactosidase activity from the Gtf2irdl locus in a sagital section of the Gtf2irdltm2(LacZ)Hrd adult mouse brain (reprinted from Howard et al, submitted). The expression pattern is regionalised with intense staining detected in the cerebellum and olfactory bulb and scattered expression elsewhere. (B) Co-localisation analysis of GFP fluorescence (a, b, c, d, e) and lacZ staining (a b \ c\ d’, e j in sections of the brain from a Gtf2irdltm2

■s % 5 fill 3 .5 E S' a; ;§ Jamo "*»• i! !s 5 3 i*f

ii }P!jap joioiai S. £•it g

■3 Ajotuaiu 5 3 >8 Sumjeai

• | Ajaixuv 3 S & s n•1 ^56 sanj|eououqe lepos 3 2 3 = £ 5 at If? 5 J3M10 -s | 3 g HI * 3 E 1 “ S 1 i S I *

A§o|oqdJOtu Illl I! uiejg 3 1111 55 Hi; -= i done.

A8o|oqdJOtusAp not ••n jepejojueo - s

3 n/d 3 uoijepjejaj su qjMOJ9 -3

-3

3 mutation; 3 s ? I i « «r ! Ii ° 3 | Hi the 3 O ? - 111J sSS 0 al % fi

3 c 3 p of

£ -S 1 *1 2 2 i?H 3 x& s *o UJ OO O'. z I III! I $ Hi 6 at C Q. i lethality a >* in +* •Si or *5 3 <5 a> :L % « I I &J0 c c § „ 0 .2 m 11 f e ? •fi ?

0J n severity 4-* *- Si* >•“ a> s 8 § C a ^ 2 S Q> 11 S S s 1 ™ ao S I 1 I 3 aJ _ -g £ s sfli S ; S “ § 3 the 3 i s - £ ._ || s s O'i —2 5 i S I ii $ § 12 u 5 U II II to •2 3*. s & O | O | o | ° ii 8-§ 8*§ 3 |8 |8 11= ? 7: « 9 ? m Is Hill N £ r* N £ due - £ I 11112 II II 11! o 2 5 SIS 3 3 £ | o' ’--i II

*N 5 applicable § i if >/ ;? 1 -3 s 3 * 8 11 fi £ 8 •b: I s 4 o' £ s s § o ill a. t y> o o o CHAPTER 2 - MATERIALS AND METHODS

2.1 General Materials

2.1.1 Molecular Biology

Agarose was from Amersco (Solon, OH). Ampicillin, bromophenol blue, EDTA, ethidium bromide, kanamycin and TRl-reagent were from Sigma (St. Louis, MO).

Bacto-agar, Bacto-tryptone and yeast extract were from BD Biosciences (Franklin

Lakes, NJ). Glycerol, glacial acetic acid, isopropanol, morpholinopropanesulphonic acid

(MOPS), sodium acetate and Tris-base were from Scharlau (Barcelona, Spain).

Chloroform and ethanol were from Merck (Darmstadt, Germany). Phenol/ chloroform/ isoamyl alcohol mix (1:1:1) was from Acros (Geel, Belgium). DNA ladder standards were from Fermentas (Ontario, Canada). All the restriction enzymes used, dNTPs mix and T4-ligase were from New England Biolabs (Ipswich, MA). Wizard Plus SV mini- preps and gel extraction/ Polymerase chain reaction (PCR) purification kits were from

Promega (Madison, WI). Plasmid midi and maxi kits were from Qiagen (Hilden,

Germany).

2.1.2 Protein Biochemistry

Acrylamide:bisacrylamide (37.5:1), Coomassie Blue G-250, glycine, pre-stained molecular weight (MW) markers and 20% sodium dodecyl sulfate (SDS) were from

Bio-Rad Laboratories (Hercules, CA). Complete EDTA-free protease inhibitor tablets, protein A- and G-agarose were from Roche Diagnostics (Mannheim, Germany). Acetic acid, glycine, hydrochloric acid, methanol and sodium hydroxide were from Chem-

Supply (Gillman, SA, Australia). Ammonium persulfate (APS) was from Promega.

Nonidet P40 (NP-40) was from BDH chemicals (Kilsyth, VIC, Australia). (3- mercaptoethanol (PME), EDTA, EGTA, glucose, magnesium chloride, magnesium sulfate, N-ethylmaleimide (NEM), phenylmethylsulfonylfluoride (PMSF), potassium chloride, sodium azide, sodium deoxycholate, sodium fluoride, N,N,N',N'- 86 Tetramethyylethylenediamine (TEMED), trichloroacetic acid, Triton X-100, Tween-20

(polyoxyethylenesorbitan monolaurate) and all other reagents were from Sigma (St.

Louis, MO). Enhanced Chemiluminescence Substrate was from PerkinElmer (Waltham,

MA). SuperSignal West Pico chemiluminescent substrate was from Pierce (Rockford,

IL). Polyvinylidene fluoride(PVDF, 0.45 pm) membrane was from Millipore (Billerica,

MA). All reagents used were of analytical reagent grade or molecular biology grade.

2.1.3 Cell Biology

Tissue culture plastics were from Falcon (Franklin Lakes, NJ). Coverslips were from

Menzel-Glazer (Braunschweigh, Germany). Dulbecco’s Modified Eagle's Medium

(DMEM), fetal bovine serum (FBS), L-glutamine, penicillin/streptomycin, and trypsin-

EDTA were from Thermo Scientific HyClone (Logan, UT). Phosphate-buffered saline

(PBS) was from MP Biomedicals (Solon, OH). Lipofectamine Plus and Opti-MEM were from Invitrogen (Carlsbad, CA). GeneJuice was from Merck. For neuronal cell preparation, bovine serum albumin (BSA), deoxyribonuclease I (DNAse), dimethylsulphoxide (DMSO), HEPES buffer and poly-D-lysine were from Sigma.

2.1.4 Yeast reagents

All dropout supplements: single (-Leu or-Trp), double (-Leu, -Trp), quadruple (-Leu, -

Trp, -His, -Ade) dropouts and X-gal substrate were from Clontech (Mountain View,

CA). Minimal synthetic defined (SD) base was from BD Biosciences. Adenine hemisulfate, 3-amino-1,2,4-triazole (3-AT), lithium acetate, PEG 3350 and YPD medium were from Sigma.

2.1.5 Plasmids

All cloning-related PCR reactions were done using Phusion DNA polymerase

(Finnzymes, NEB) according to manufacturer's recommendation. All constructs were verified by DNA sequencing.

87 Mammalian expression plasmids

GTF2IRD1 expression plasmids were generated by Dr. Stephen Palmer. pCDNA-

GTF2IRD1 was made by subcloning the human GTF2IRD1 isoform lal coding sequence into HindlU and Xho\ sites of the pCDNA3.1 vector (Invitrogen). The epitope-tagged plasmid, pMyc-GTF2IRDl was made by inserting a synthesized oligonucleotide linker fragment encoding the Myc epitope into the HindlU site of pcDNA-GTF2IRDl, while pGFP-GTF2IRDl was generated by subcloning the open reading frame (ORF) sequence of GTF2IRD1 into HindlU and Sail sites of pEGFP-C3

(Clontech). The luciferase reporter plasmid, pGL3-GUR, has been described previously

(Palmer et ai 2010).

The SUMO motif K495R and K271R single mutants were generated by standard splicing by overlap extension PCR method (section 2.2.8) using pcDNA-GTF2IRDl as the template. The combined 2KR (K495R/K271 R) double mutant was generated by cutting the pCDNA3.1- K271R plasmid with HindlU and Age I and inserting into the same sites upstream of pCDNA3.1-K495R backbone. HA-SUMO (pMT3-HA-SUM01) containing the mature form of SUMO-1 with pre-exposed C-terminal double glycine and tagged with HA-epitope has been previously described (Perdomo et al, 2005). The plasmid encoding UBC9, pCDNA3.1-UBC9 was generated by Dr Stephen Palmer by amplifying the UBC9 ORF sequence from mouse cDNA and cloning into BamHl and

EcoRl sites of pCDNA3.1. The Ets-1 expression plasmid was from Dr J. Pimanda

(University of New South Wales, Sydney, Australia). The SUMO-GTF21RD1 fusion plasmid was generated by PCR amplification of the SUMO coding region from the pMT3-HA-SUM01 and inserting it in-frame into the HindlU site of the pcDNA-

GTF2IRD1 plasmid to create a plasmid encoding an N-terminal SUMO fusion protein.

Epitope-tagged expression plasmids encoding the various proteins identified in the yeast two-hybrid library are listed in Appendix B-l. Full-length coding regions of each gene were amplified either from corresponding constructs in the rescued pGADT7 yeast

88 plasmids or from mouse cDNA samples using primers listed in Appendix B-2. The fragments were subcloned into the epitope-tagged vectors using the indicated restriction sites. pMyc-HDAC3 was subcloned directly from pGADT7-HDAC3 using EcoRI and

BamH\ sites.

Yeast two-hybrid plasmids

For yeast two-hybrid assays, bait and prey sequences were cloned into pGBKT7 and pGADT7 vectors, respectively (Matchmaker system, Clontech, Appendix B-4) and the primers used in the cloning procedure are listed in Appendix B-3. The pGBKT7 bait plasmid contains the coding sequence under test fused in-frame to the Gal4-BD (DNA binding) domain while the pGADT7 prey plasmid contains the candidate coding sequence fused in-frame to the Gal4-AD (DNA activation) domain. pGBKT7 plasmids containing the full-length human GTF2IRD1 or specific domains of the protein

(excluding pGBKT7-LZ, pGBKT7-RD5, and pGBKT7-Sumo2) and pGADT7-PlASx were generated and described previously (Bontempo, 2009). The alignment of the

GTF2IRD1 amino acid sequence with the locations of the various domains are highlighted in Appendix A. pGADT7-UBC9 was generated by PCR amplification of the fragment encoding UBC9 from the pcDNA3.1-UBC9 plasmid, cut with EcoRI and the fragment was inserted into pGADT7 which had beendigested with Ndel, blunted by klenow-filling and then digested with EcoRX.

Plasmids containing C-terminal truncations of GTF2IRD1: pGBKT7-Trl (encoding amino acids 1-932 according to human protein sequence #Q9UHL9), pGBKT7-Tr2 (aa

1-909), PGBKT7-Tr3 (aa 1-787) and pGBKT7-Tr4 (aa 1-432) were generated by PCR amplification using primers listed in Appendix B-3 and pGBKT7-GTF2IRDl plasmid as template. A stop codon was introduced at the 3' end of each fragment and all were cloned into the EcoR\ and Sail sites of the pGBKT7 plasmid. pGBKT7-2KR, which contains a GTF2IRD1 K495R/K271R double mutation, was generated by PCR

89 amplification of the ORF from pCDNA3.1-2KR and insertion of the cut fragment into the EcoR\ and Sail sites of the pGBKT7 vector.

Prey plasmids containing the full-length coding sequence of the various proteins identified in our Y2H library screens were amplified from mouse cDNA samples and cloned in-frame to the Gal4-AD domain in the pGADT7 plasmid (Appendix B-3). The full-length Setd6 clone was generated using a two-step PCR amplification: the N- terminal half was amplified from a mouse liver cDNA sample (Fragment A) and the C- terminal half was amplified from the Y2H library clone (Fragment B). Fragment A was amplified using the ttgaattcatggcggcccccgccaagcg (containing EcoRl site) and

ATGGGCTGAGTAGCCACCATCCG primer set. Fragment B was amplified using

AATACTATTCCATCGTCCTG and AAGCCTGCTTAATTTGTGAGTTCC primers and inserted into the SmaI site of pGADT7. Fragment A was then subcloned into the EcoRl and Xbal sites of the pGADT7-Fragment B plasmid.

2.1.6 Antibodies Anti-GTF2IRD1 (M-19), anti-GFP (FL), and anti-HA (Y-ll) antibodies were from

Santa Cruz Biotechnology (Santa Cruz, CA). Anti-GFP (IP grade) was from Abeam

(Cambridge, MA). Anti-p-actin, anti-Myc (9E10), anti-SUMO-1 (C-terminal), and anti- a-tubulin (DM1 A) antibodies were from Sigma. Anti-TFII-I and anti-Histone H3

(96C10) antibodies were from Cell Signalling (Denvers, MA). Anti-GAPDH and anti-

H3K4mel (ChIP grade) were from Millipore. The antibody against Baf53b was a gift from Prof. Gerald Crabtree (Stanford University, California). Anti-SETD6, anti-

H3K12meland anti-H4K12mel antibodies were gifts from Dr Thomas Jenuwein (Max

Planck Institute, Germany).

Horse radish peroxidase (HRP)-conjugated secondary antibodies used were as follows: goat anti-mouse (DAKO), donkey anti-rabbit (Amersham) and donkey anti-goat (Santa

Cruz). All Alexa-Fluor-conjugated secondary antibodies used were from Molecular

90 Probes (Invitrogen), which includes goat anti-mouse (448), goat anti-mouse (555), goat anti-rabbit (448), goat anti-rabbit (555) and donkey anti-goat (555).

2.2 Molecular Biology

2.2.1 Agarose gel electrophoresis

DNA was analysed and purified through separation by size on agarose gels. To make a

1% agarose gel, 1 g of agarose was added to 100 mL of IX Tris-borate-EDTA (TBE) buffer (89 mM Tris-HCl pH 8.4, 89 mM boric acid and 2 mM EDTA) in a conical flask and heated in a microwave oven until the agarose was fully dissolved. The solution was cooled to 50°C and ethidium bromide solution was added to a final concentration of 0.1 pg/mL and then poured into a sealed gel casting platform and a comb was inserted. The gel was allowed to set, the comb was carefully removed and the gel was then placed into an electrophoresis tank filled with IX TBE buffer. DNA samples, mixed with loading dye (0.04% Bromophenol Blue and 10% glycerol), were loaded onto the gel along with a 100 bp or 1 kb DNA ladder (GeneRuler, Fermentas) as size markers. Typically, the gel was run at a constant voltage of 100 V for 40-70 min (varied with the size of the DNA fragments and the percentage of the gel). DNA was visualised using a

UV transilluminator and an image was taken using the gel documentation system

(BioRad, Hercules, CA).

2.2.1 Polymerase chain reaction amplification

Most PCR reactions were performed in a total volume of 20 pi containing Lucigen

EconoTaq PLUS Green master mix (Middleton, WI), 10 pmol of each forward and reverse primers, and approximately 50 ng of DNA template. Genotyping PCR reactions were performed using Epicentre Biotechnologies MasterAmp 2X PCR Premix D

(Madison, WI), 0.2 pi of Taq DNA polymerase (Roche), 10 pmol each of the forward and reverse primers and approximately 50 ng of genomic DNA template. For the purposes of cloning ORF cDNAs, the high-fidelity DNA Polymerase Phusion 91 (Finnzymes) was used according to the manufacturer's recommended protocol. All PCR reactions were carried out on a thermal cycler (BioRad) under the following conditions: an initial 95°C denaturation step for 3 min was followed by 35 cycles of denaturation

(95°C for 30 s), annealing (54-62°C for 1 min) and elongation steps (72°C at 1 min/kb).

2.2.2 Genomic DNA extraction from mouse tail samples

DNA Lysis buffer consisting of 100 mM Tris-HCl pH 8.5, 5 mM EDTA, 0.2% SDS,

200 mM NaCl and 100 pg/mL proteinase K (NEB) was added to the clipped tail piece and incubated at 65°C overnight on a rotating platform. On the next day, any undigested tissues were spun down at 13,000 rpm for 5 min. The supernatant was mixed with 500 pi of isopropanol and centrifuged at 13,000 rpm for 5 min to precipitate the genomic

DNA. The DNA pellet was washed with 70% ethanol and centrifuged again at 13,000 rpm for 5 min. Ethanol was removed and the DNA pellet was air-dried for 5 to 10 min and resuspended in 50-100 pi of water. To genotype the mice, 1 pi of each DNA sample was used in a PCR reaction using KOScreenF2 (5’-GCATCTCTGGAACTTGAGTTG-

3'), KOScreenR (5’-GACTTCCTGTAGGTGCTAGGTG-3’) and MDEX2R (5'- ATGGAGTCTAAGGCGGACAC-3') as described previously (Palmer et al, 2010).

2.2.3 Purification of DNA from gel or solution

DNA fragments or PCR products were purified from gel or solution using the Wizard

SV gel and PCR Clean-Up system (Promega). To purify DNA fragments from agarose gels, bands of interest were visualised under UV light and were excised with a sterile scalpel blade. Gel pieces were transferred to a clean microcentrifuge tube and processed according to the manufacturer's instructions.

2.2.4 Ligation of DNA fragments

For blunt-ended vector DNA preparation, dephosphorylation was carried out prior to ligation to prevent self-religation. In a total volume of 50 pL, the purified vector DNA

92 was incubated with 2 (liL of Antartic phosphotase (NEB) and lx NEB buffer. The reaction was incubated at 37°C for 15 min and inactivated for 55°C for 10 min.

Typically, ligations were carried out in a total volume of 10 pL containing the purified vector and insert DNA (1:3 molar ratio) mixed with 1 pL of lOx ligation buffer and 1 pL of T4 DNA ligase (NEB). The ligation mixture was incubated at 4°C overnight.

Transformation of competent E.coli (section 2.2.5) was then carried out using 5 pL of the ligation reaction. Colonies were selected and screened for the presence of insert

DNA by restriction endonuclease digestion. The correct orientation and sequence of the insert was confirmed by DNA sequencing (section 2.2.10).

2.2.5 Transformation of E. coli

Transformation was performed using competent XL-1 blue E. coli cells, which had been prepared using the rubidium chloride supercompetant method (Inoue et ah 1990) and stored at -80°C. Plasmid DNA (10-50 ng) or ligation products (5 pL) were pipetted into a microfuge tube and mixed with 50 pL of freshly thawed XL-1 blue competent cells by gently flicking the tube. The tubes were incubated for 30 min on ice, subjected to heat shock in a 42°C water bath for 45 sec and returned to ice for 2 min. SOC medium (0.5 mL, Invitrogen) was added to the cells and incubated for 90 min at 37°C with shaking at

200 pm. Between 50-200 pL of the culture was plated onto pre-warmed LB-agar plate

(10 g/L Bacto-tryptone, 5 g/L yeast extract, 10 g/L NaCl, 1.5% agar) containing the appropriate antibiotic (100 pg/mL for ampicillin; 10 pg/mL for kanamycin) and the resistant colonies were allowed to grow in a 37°C incubator overnight.

2.2.6 Purification of plasmid DNA

Small quantities of plasmid DNA for analysis or cloning purposes were purified from 2 mL overnight bacterial culture using the Wizard Plus SV miniprep DNA purification system according to the manufacturer's protocols (Promega). Large quantities of plasmid DNA for transfection purposes were purified from 200 mL of overnight 93 bacterial culture with the Qiagen Plasmid Midi kit according to the manufacturer's protocols. The concentration and purity of plasmid DNA was measured using aNanoDrop UV spectrophotometer (Thermo Scientific) at 260/280 nm wavelengths and by visualisation on an agarose gel.

2.2.7 Restriction endonuclease digestion

A typical restriction digest was set up in total volume of 30 pL containing plasmid

DNA (1-3 pg), 1 pL of restriction enzyme (40U, NEB) and 3 pi of 10X buffer (NEB).

Reaction mixture was incubated in 37°C water bath, ranging between 30 min to overnight according to the quantity of DNA, efficiency of the enzyme and to reduce self-religation of cut vectors. Digested DNA products were analysed by agarose gel electrophoresis if necessary, purified and then extracted (section 2.2.3).

2.2.8 Klenow fill-in reaction Blunt-ended DNA vectors were generated through fill-in reactions using DNA

Polymerase I Large Klenow (Promega). Typically, plasmid DNA (at least 3 pg) was digested with an appropriate restriction enzyme in a 50 pL volume, heat-inactivated at

65°C for 20 min and then the single strand overhangswere filled in by adding 1 unit of

Klenow enzyme and 2.5 pL of 10 mM dNTP mixture. The reaction was incubated for

15 min at room temperature and the DNA was purified using the Promega PCR purification kit.

2.2.9 Site directed mutagenesis

GTF21RD1 lysine to argenine mutants were created by the standard splicing by overlap extension (SOE) PCR-based mutagenesis protocol (Ho et al, 1989). In the first step, two separate PCRs were used to generate two primary PCR products, designated fragment A and B. The primary PCR products were to have an overlapping region of at least 17 bases between two internal primers which contained the desired mutation. In the second

94 round of PCR, the overlapped regions between fragment A and B annealed to each other and acted as template to amplify the combined fragment C using the same outer primers. Unique restriction sites must also be present in the outer primers to enable subcloning of the PCR fragment into vector of choice.

PCRs were done using Phusion DNA polymerase (Finnzymes, NEB). The first round of PCR was done using a standard PCR protocol according to the manufacturer's recommendation except that no primers were added, thus allowing replication of DNA from the overlapping regions. The outer primers were added at the end of the 2nd cycle to facilitate amplification of the complete fragment containing the new mutation. A small portion of the PCR product was then checked on agarose gel and the rest was purified from solution using Promega gel extraction/PCR purification kit. The mutated fragment was digested with the appropriate enzymes and subcloned into a vector. DNA sequencing was performed to verify the desired mutation.

2.2.10 DNA sequencing DNA sequencing was carried out at the Ramaciotti Centre (UNSW) using BigDye terminator chemistry. PCR reactions were carried out prior to submitting samples, which consisted of 1 pL of BigDye terminator (ABI), 100-500 ng template (plasmid),

3.2 pmol primer and 4 pi of 5X buffer in a total volume of 20 pL. Cycle sequencing reactions consisted of 25 rounds of: 96°C for 10 sec, 50°C for 5 sec, and 60°C for 4 min. Ethanol precipitation was then carried out to remove excess BigDye terminators.

Briefly, the cycle-sequencing product (20 pL) was mixed with 16 pL water and 64 pL non-denatured 95% EtOH in a 1.5 mL microfuge tube and vortexed. The mixture was incubated at room temperature for 15 min, then centrifuged for 20 min at 13,000 rpm.

The supernatant was carefully aspirated, washed with 250 pL 70% ethanol and spun for

10 min at 13,000 rpm. The supernatant was carefully aspirated and samples were dried by placing the tube with the lid open in a heat block at 90°C for 1 min. Dried samples were kept at -20°C until ready for analysis.

95 2.2.11 Total RNA extraction

For RNA isolation from brain tissues, mice were sacrificed by cervical dislocation and the brain was carefully dissected out in cold-PBS and immediately frozen in liquid nitrogen. Small portions of the tail were collected for genotyping purposes. Brain tissue was homogenised in TRl-reagent (10 mL per gram of tissue; Sigma) for 1 min using a

Polytron homogeniser. Chloroform (1 to 5 volume ratio to TRl-reagent) was added to the homogenate, vortexed, chilled on ice for 15 min and then centrifuged at 12,000 g for

15 min at 4°C. The resulting aqueous layer was collected into a fresh 15 mL propylene tube and mixed with an equal volume of 70% ethanol and immediately passed through an RNAeasy mini column (Qiagen). The column was washed with the supplied wash buffers according to manufacturer's instructions and the total RNA was eluted with 40-

60 pi of RNase-free water.

For RNA isolation from cell cultures grown on a 100 mm dish, cells were washed once with cold PBS before removal by aspiration. One mL of TRl-reagent was added, and the lysed cell mixture was collected into an RNase-free microfuge tube using a cell scraper. The cells were homogenised on ice by passing them through a 23-gauge needle attached to a sterile plastic syringe8 times. The mixture was then treated in the same way as the brain samples to extract the RNA, which was finally resuspended with 30-50 pi of RNase-free water.

The quantity and quality of extracted RNA were measured using a NanoDrop UV spectrophotometer. Samples isolated through the RNAeasy mini column gave a higher quality of RNA, which is essential for the downstream microarray experiment.

2.2.12 RNA-formaldehyde gel electrophoresis

Gel apparatus and glassware were cleaned with RNase-Zap (Ambion, Austin, TX) and disposable RNase-free filtered tips and tubes were used to minimise RNA degradation. 96 Fora small gel (total volume of 100 mL), 1% (w/v) agarose was dissolved and boiled in

70 mL of IX MOPS buffer (40 mM MOPS, 10 mM sodium acetate, 1 mM EDTA disodium, pH 7 adjusted with glacial acetic acid) and 10 mL of filter-sterilised water. In the fume hood, 20 mL of formaldehyde was added into the molten agar solution, cooled and poured into the casting tray containing a comb. The gel was allowed to set before being transferred to the electrophoresis tank filled with IX MOPS running buffer.

Typically, 2 to 5 pg of RNA was needed to assess the quality of RNA. For each RNA sample, loading buffer containing 2.5 pi of 10X MOPS, 3.5 pi of formaldehyde, 10 pi of deionised formamide, 1 pi of ethidium bromide (1 mg/mL) and 1 pi of Bromophenol

Blue was added. The RNA/buffer mixtures were heated at 65°C for 5 min to denature any secondary structure of the RNA, cooled on ice for 2 min and then loaded onto the gel. The gel was run at 70-100 V until the dye had migrated as far as 2/3 the length of the gel. RNA bands were visualised on a UV transilluminator. Intact total RNA have sharp 28S and 18S rRNA bands with an intensity ratio of 2:1.

2.2.13 Illumina array analysis For analysis of the different regions of the brain, 2 pg of total RNA per sample was submitted for microarray analysis processed by the Institute of Molecular Biosciences

(IMB) microarray facility at the University of Queensland (Brisbane, QLD, Australia).

Microarray analysis of the FACS-isolated GABAergic cells was carried out by the

Australian Genome Research Facility (AGRF, Melbourne, VIC, Australia). Each facility carried out the common routine for Illumina array processing including an RNA quality check (integrity, purity and concentration), reverse transcription to generate cDNA. amplification to yield cRNA and hybridisation and scanning on the Illumina

BeadChip Sentrix Mouse-6 (containing >45,000 transcript-specific probe sequences per array). The raw data received, which had already been normalised using BeadStudio software, were further analysed using GeneSpring GX software (Agilent Technologies,

Santa Clara, CA). For the FACS-array study, t-test Benjamini-Hochberg data significance analysis was performed on datasets obtained from array triplicates from 97 each genotype group, and the analysis was performed using GeneSifter

(Geospiza, Seattle, WA). However, it is important to note that these samples were not independent replicates and therefore, the variance being analysed only derives from technical components and not from true biological variance.

2.2.14 Northern Blotting

For northern analysis, RNA samples were separated by electrophoresis on an agarose- formaldehyde gel prepared as described above (section 2.2.12). To determine the size of the transcript of interest, 500ng of RNA markers (Promega) was loaded onto the gel in the same manner as the RNA samples. The gel was then photographed next to a ruler to record the distance travelled by different molecular weight species. The gel was soaked in RNase-free water for 3 x 5 min to remove the formaldehyde, then washed in 10X

SSC (1.5 M sodium chloride, 0.15 M sodium citrate, pH 7.0) transfer buffer with gentle agitation for 15 min. A long rectangular Whatman filter paper was placed on a solid glass support so that both of its ends formed a wick in the reservoirs of transfer buffer.

The capillary blot was set up on the paper wick in the following order: 2 x Whatman paper - gel (facing down) - Hybond N+ membrane (Amersham Biosciences) - 2 x

Whatman paper - papertowels - glass plate - weight. Four strips of parafilm were placed around the capillary blot to prevent “short circuiting" of the buffer around the gel. The transfer was allowed to proceed overnight.

The resulting membrane with attached RNA was air-dried for 2 hr, UV cross-linked, soaked in 2X SSC briefly and then pre-hybridised in Church buffer (Church & Gilbert,

1984) (0.5 M disodium hydrogen orthophosphate, 1% BSA, 7% SDS, 1 mM EDTA, pH

7.2) for at least 1 hr at 65°C with rotation. DNA probes were either PCR-products or

DNA restriction fragments cut from plasmids.. To detect the Actl6b transcript, a region of the cDNA was amplified using the forward primer 5' GGCTATTGATATCATTCCTCC and the reverse primer 5' CTGGCGATGAGCTTAAGAC. The PCR product was purified using a gel extraction kit (Promega) and used as a template in another round of PCR and

98 purified again from solution. DNA fragments used for probes in Northern blots were labelled using the DECAprime II labelling kit (Ambion) according to the manufacturer's instructions. In a total volume of 14 pL, 50 ng of template DNA, 20 pmol of the random decamer primer and water were mixed, denatured at 95°C for 5 min and immediately chilled on ice for 5 min. A reaction mix containing 5 pL of 5X reaction buffer minus dCTP, 1 pi of the Klenow fragment of DNA Polymerase I

(5units) and 5 pi of 3000 Ci/mmol [a-,2P] dCTP was added to the denatured samples and incubated at 37°C for 20 min. At the end of incubation time, 25 pi of TE buffer was added and the sample was passed through a ProbeQuant G-50 micro column

(Amersham Biosciences) to remove any unincorporated nucleotides. The filtered probes were incubated at 95°C for 4 min to separate the double stranded fragments and then pipetted into the Church buffer in the hybridisation tube containing the membrane for an overnight hybridisation with rotation at 65°C. The next day, the radioactive solution was discarded appropriately. The membrane was washed three times in high stringency wash buffer (0.1X SSC, 0.1% SDS) on a rotating platform at 65°C for 15 min each wash. The membrane was removed from the tube, wrapped in cling wrap and exposed to X-ray film overnight (longer if required) before developing using the X-ray film developer CP 1000 (Agfa, Mortsel, Belgium). To strip and re-hybridise the membrane with a different probe; 250 mL of boiling 0.5% SDS was poured onto the membrane in a washing box and incubated with agitation at room temperature until the solution cooled down. The SDS solution was discarded and the membrane was kept wet with 50 mL of 5X SSC until it was ready to be re-probed.

2.2.15 Reverse transcription (first strand cDNA synthesis)

In an RNase free microfuge tube, 20 pmol of Oligo dT primers (Promega) and 1 pg of total RNA were mixed in a total volume of 5 pi and heated at 65°C for 5 min. The mixture was immediately chilled on ice for 5 min. A master mix consisting of 4 pi of

5X reverse transcriptase reaction buffer, 40 pmol of dNTPs, 2.4 pi of 25 mM MgCE,

3.6 pi of water and 1 pi of ImProm-II™ reverse transcriptase (Promega) was added to 99 the RNA-oligo dT mix. The reaction was allowed to proceed for 2 hr at 37°C. 1 pi of the product was used for subsequent quantitative real-time PCR (qRT-PCR) reactions.

2.2.16 Quantitative real time-PCR

PCR primers to amplify regions of each transcript were designed using Primer Blast

(NCBI) and tested with a standard PCR assay to check visually. Quantitative RT-PCR reactions were set up using the QuantiTect SYBR green kit (Qiagen) based on SYBR green chemistry and performed on the Rotor Gene 6000 (Corbett Life Science) or

Stratagene MX3005 QPCR system. Standards for absolute quantitative analysis were prepared by amplifying the PCR product, gel extraction from several lanes in order to generate approximately 500ng of product and then serial dilutions were prepared. Each reaction was set up in triplicate in a 20 pL total volume, consisting of 2x QuantiTect

SYBR green PCR master mix, 5 pL DNase-free water, 5 pL of cDNA (“unknown’') or

DNA standard. The same set of reactions were set up for the housekeeping genes,

Gapdh and Hprt, in order to normalise against the levels of the gene of interest. Data were analysed in Microsoft Excel by deriving estimates of starting template quantities from the standard curve, normalising these values against the housekeeping genes and performing a t-test analysis for the different groups of samples.

2.3 Protein Biochemistry

2.3.1 SDS-polyacrylamide gel electrophoresis

SDS-polyacrylamide gel electrophoresis (PAGE) was set-up and carried out using the

BioRad mini-Protean 3 cell system. Depending on the protein(s) of interest, acrylamide solution for the resolving gel (pH 8.8) contained either 7.5%, 10%, 12% or 15% acrylamide:bisacrylamide, 0.1% SDS, 0.375 M Tris-HCI pH 8.8. Gel polymerization was initiated by the addition of 10% APS (final concentration 0.1%) and TEMED (final concentration 0.1%). Once poured, gels were overlayed with ethanol to prevent them from drying out and allowed to polymerise for at least 30 min. Ethanol was discarded

100 and a stacking solution containing 4.4% acrylamide:bisacrylamide, 0.125 M Tris-HCl pH 6.8 and 0.1% SDS was poured. Polymerisation was initiated as above. The appropriate comb was inserted and the stacking gel was allowed to set for at least 15 min.

The gels were placed in a gel electrophoresis tank, containing running buffer (0.025 mM Tris-base, 0.19 M glycine, 0.1% SDS). Samples containing a final concentration of

375 mM Tris-HCl pH 6.8, 2% SDS, 4 mg/mL Bromophenol Blue, 10% glycerol, 2mM

EGTA and 8% (3-mercaptoethanoI were heated at 95°C for 5 min before loaded into the wells of the gels. Pre-stained protein marker was loaded into the outside lane(s).

If required, the gels were stained in coomassie staining solution (0.125% Coomassie

Blue G-250, 10% glacial acetic acid, 45% methanol) until the desired staining had been obtained, then placed in destaining solution (40% methanol, 10% acetic acid). Gels were washed twice with deionised water for a total of 30 min to remove any acetic acid before drying between two sheets of cellophane.

2.3.2 Western blotting Proteins were separated by SDS-PAGE on 7.5%, 10%, 12% or 15% acrylamide mini­ gels at 120 V for 90 min. Broad range pre-stained MW markers (Bio-Rad) were loaded into the outside lane(s). Western blotting was performed according to the method of

Towbin (Towbin et aU 1979). After electrophoresis, the gel was equilibrated for 10 min in transfer buffer (25 mM Tris-HCl pH 8.3, 192 mM glycine, 20% methanol) by gentle shaking. The gel was assembled in a cassette in a tray filled with ice-cold transfer buffer in the following order: sponge pad, 2 pieces of Whatman filter paper, the gel, PVDF membrane (methanol pre-soaked), another 2 pieces of filter paper and sponge pad. The assembled blot cassette was then placed in the blotting tank containing ice-cold transfer buffer and the transfer was carried out under a constant voltage of 80 V for 2 h or 35 V overnight in the cold room. After transfer was completed, the membrane was blocked in

101 5% skim milk, made up in Tris-buffered saline (200 mM NaCI, 150 mM Tris-HCl, pH

7.4) containing 0.1% v/v Tween-20 (TBS-T) for either 1 h at room temperature or overnight at 4°C. The membrane was washed twice with washing solution (TBST). The membrane was then incubated with primary antibody for 2 h at room temperature or overnight at 4°C, followed by 5 washes for 5 min each in washing solution. The membrane was then incubated with HRP-conjugated secondary antibody (1:10,000) for

1 h at room temperature followed by 5 washes for 10 min each in washing solution.

After removal from solution, the enhanced chemiluminescence (ECL) mixture

(containing equal amount of the peroxide and luminal solutions) was poured onto the membrane and incubated for 2-5 min. Excess ECL reagent was removed by placing the membrane in between 2 pieces of filter paper. The blot was then exposed to Super-RX film (Fujifilm) for various amount of time, ranging from 5 sec to 1 h. The film was then developed and fixed by an automated X-ray film developer CP 1000 (Agfa).

2.3.2.1 Reprobing blots For reprobing of Western blots, the PVDF membrane was stripped by incubating it in acidic glycine-SDS buffer (25 mM glycine-HCl, pH 2.0, 1% SDS) for 30 min with agitation at room temperature. The acidic buffer was removed thoroughly by washing in deionised water, followed by incubation in TBST for 10 min. To proceed with the subsequent immunolabelling, the membrane was blocked in 5% skim milk solution, followed by the same procedures described above.

2.3.2.2 Densitometry analysis of blots

Films with the appropriate exposure were scanned using an Epson scanner.

Immunoreactive bands were quantified using Image J (National Institutes of Health,

Bethesda, MD).

102 2.3.3 Total protein extraction from tissues or cells

All procedures were performed on ice and all solutions were chilled to 4°C before use

unless otherwise indicated. Total protein extracts from mouse brain were prepared by

homogenising tissues in 5 mL ice-cold R1PA buffer (20 mM Tris pH 7.4, 1% Triton X-

100, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA) per gram of tissue using a Dounce

homogeniser. Protease inhibitors, consisting of EDTA-free complete protease inhibitor

cocktail (1 tablet per 50 mL), 1 mM PMSF and 20 pg/mL leupeptin, were added just

prior to homogenisation. The homogenate was centrifuged twice at 14,000 rpm for 30

min at 4°C and the cleared protein supernatant was collected. Total cellular protein

extracts from COS-7, C2C12 or HEK293 cells grown on 100 mm dish were prepared by

scraping cells into 1 mL of ice-cold RIPA buffer or 1% Triton lysis buffer (1% Triton

X-100, 1 mM EDTA, 1 mM EGTA in PBS). Cells were further lysed by incubating at

4°C for 20 min with rotation and sonicated briefly. Protein samples (supernatant) were

collected by centrifuging lysates on a bench top centrifuge at 14,000 rpm for 20 min at

4°C. Lysates were stored at -20°C or used for subsequent immunoprecipitation or

Western blotting.

2.3.4 Extraction of histone proteins from tissues

This protocol was adapted from previous reports (Gupta et al, 2010; Shechter et al,

2007). In studying the function of histones, histone variants and histone modifications,

acid extraction of histone proteins is routinely used, based on the basic nature of

histones, which means that they are insoluble in acidic solutions. All procedures were

performed on ice and all solutions were chilled to 4°C before use unless otherwise

indicated. All centrifugation steps were performed at 4°C. Brain tissues were

homogenised in ice-cold homogenisation buffer (250 mM sucrose, 50 mM Tris pH 7.5,

25 mM KC1, 0.9 mM sodium butyrate, protease inhibitors cocktail and PMSF), using

six strokes of the glass pestle. Homogenates were centrifuged at 7,700 x g for 1 min and the supernatant (cytoplasmic fraction) was aspirated. The pellet (nuclear fraction) was then resuspended well in 500 pL of 0.2 M NH2SO4, incubated for 30 min or overnight

103 with rotation at 4°C and centrifuged at 14.000 x g for 10 min. The supernatant containing histones was transferred to a fresh tube, and precipitated with 250 pL of

100% trichloroacetic acid containing 4 mg/mL sodium deoxycholate for 30 min on ice.

Acid extracted histone proteins were then collected by centrifugation at 14,000 x g for

30 min. The supernatant was discarded, and the protein pellet was washed with 1 mL of acidified acetone (0.1% HC1) followed by 1 mL of acetone for 5 min each and collected by centrifugation at 14,000 x g for 5 min in between washes. Histone pellets were air- dried at room temperature for 10-20 min, and resuspended in 10 mM Tris (pH 8.0) and stored at -80°C until analysed by Western blotting.

2.3.5 BCA assay

The BCA assay was performed using Pierce BCA protein assay kit in a 96-well clear bottom plate according to manufacturer's instruction. Briefly, a working reagent (200 pL per well) was made up by mixing reagent A and B in a 50 : 1 ratio. Samples (20 pL, either diluted or undiluted) or standards were distributed into the microplate well.

Standards (0 for blank, 2, 5, 10, 20 and 30 pg of protein) were prepared using a 2 mg/mL bovine serum albumin stock solution and made up to 20 pL with the same diluent as the samples. The plate was incubated for 30 min at 37°C and allowed to cool down at room temperature. Absorbance at 562 nm was read using a SpectraMax microplate reader (Molecular Devices). The protein quantity for each sample was calculated by plotting the reading of each sample against the absorbance readings of the known standards.

2.3.6 Immunoprecipitation

All procedures were performed at 4°C unless otherwise stated and incubations were done with constant gentle rotation. Cell lysates were pre-cleared by incubation with 25 pL of protein A- or protein G-Sepharose beads (pre-washed with PBS and left as 50% slurry, Roche) for 1 h. Meanwhile, approximately 5 pg of antibody was coupled to 25 pL of either protein A- or protein G-sepharose for one hour. Pre-cleared lysates were 104 separated from the beads by centrifugation at 3,000 rpm for 3 min and 50 pL of each lysate was saved as the input sample, while the rest were incubated with the antibody- bound beads overnight. Beads were washed with the lysis buffer three times, with centrifugation at 3000 rpm for 3 min each time. The immunoprecipitated proteins were eluted by boiling the beads in 2X Laemmli SDS sample buffer at 95°C in a heating block for 10 min. Input cell lysates and one fifth of the immunoprecipitated samples were separated on SDS-PAGE gels and analysed by Western blotting.

2.4 Cell Biology

2.4.1 FACS-array experiment

2.4.LI Animals

The Gtf2irdl mutant mouse line, C57BL/6-Gtf2irdl"”,l,ul (Palmer et al, 2007) has been backcrossed 5 generations onto the C57BL/6 background and maintained by heterozygous mating in the Chidren's Medical Research Institute Bioservices Facility (up to December 2008) and Australian BioResources (January 2009 onwards).

Genotyping was performed by collecting either tail or ear punch samples for DNA extraction and subsequent PCR amplification as described in section 2.2.2. Studies were carried out on homozygous null mice and age- and sex-matched littermate wild type control mice, with the exception of those studies where heterozygous mutant mice are specified. The mating of C57BL/6-Gadl'm,(EGFP)Tan'a (Tamamaki et al, 2003) and

Gtf2irdltmIHrd to generate C51BL/6-Gadltml(EGFP)Tama/Gtf2irdl,mlHrd knock-in/knockout mice is described in detail in Chapter 3, section 3.3.1. Genotyping of the

Gadl,m,(hGPP)Tama allele was determined by observing GFP expression in the brain, which was visible through the semi-transparent skull of newborn pups, using a UV-light coupled with EGFP-goggles. All experiments were approved by the AEC of the

Children's Medical Research Institute or the University of New South Wales and conducted in accordance with the NHMRC Australian code of practice for the care and use of animals for scientific purposes, the NHMRC Guidelines for the generation.

105 breeding, care and use of genetically modified and cloned animals for scientific purposes, and the NSW Animal Research Act.

2.4.1.2 Tissue and disaggregated celts sample preparation

GAD67-GFP positive cells were isolated from P9-P11 brains of Gadllml(EGFP)Tama/

Gtf2irdl"n,Hni mice, that were homozygous KO or WT at the Gtf2irdl allele. For each sorting, whole brain tissue was dissociated to a single cell suspension using Neuronal

Tissue Dissociation Kit (Miltenyi Biotec) according to the manufacturer's instructions.

Tissue and cell preparations were kept under sterile condition. Briefly, isolated brains were placed on a petri dish containing HBSS buffer and cut into small pieces using scalpel blade. The pieces were then collected into a 15 mL tube and subjected to enzymatic digestion according to the manufacturer's protocol. For mechanical dissociation, three glass pipettes were prepared, one untouched while the other two were flame-polished to give a medium bore and another with fine bore. Digested tissue was then mechanically dissociated by pipetting up and down ten times slowly, first with the wide-tipped glass pipette, followed by the medium and then the small bore pipettes, until no tissue pieces remained. After each trituration, tissue was incubated in the 37°C water bath for 10 min, except after the last trituration, after which the cells were applied to a 40 pm cell strainer (BD Falcon) and washed with 10 ml of HBSS buffer. Cells were collected by centrifugation at 1000 rpm for 5 min at room temperature. The cell pellet was gently resuspended in 4.4 ml sorting buffer (10 mM HEPES, 5 mM EDTA, 25 ug/ml DNase I, lx Pen/Strep in L15-CC>2 medium) and equally divided into 4x sterile

FACS tubes (BD Falcon) to lower the cell density in the suspended sample which can lead to clogging of the FACS machine.. Samples were transferred on ice and processed immediately by FACS as prolonged intervals would result in decreased GFP positive cell recovery. Propidium iodide (PI) (20 pg/ml) used to label the dead cells is added just prior to sorting.

106 2.4.1.3 FACS / RNA extraction

Cells were sorted on a FACSVantage cell sorter (Becton Dickinson) for their FSC

(forward scatter), SSC (side scatter), phycoerythrin or PE (detecting PI) and fluorescein-

5-isothiocyanate or FITC (detecting EGFP), using 100 pm nozzle; performed by The

Flow Facility at Westmead Millenium Institute (Westmead). Only viable cells (PI negative) and high EGFP expression at equal to or greater than 103 5 were gated for collection. In the FACS collecting chamber, cells were gathered in a fresh tube containing 500 pi of sorting buffer (with 1% BSA). After sorting, cells were immediately placed back on ice until all of the cells collected per animal (typical yield was between 200,000-250,000 cells) had been achieved. Cells were pooled together according to genotype, centrifuged at 1,000 rpm for 5 min at room temperature, and the buffer was discarded. Cells were processed immediately for RNA extraction using the

Picopure RNA isolation kit (Arcturus) according to the manufacturer's protocol, including the recommended DNase treatment step (Qiagen). Isolated RNA was kept at -

80°C until use. The quality and relative concentration were confirmed on a Bioanalyzer

Nanochip (Agilent Technology) prior to the Illumina microarray processing, performed by the Australian Genomic Research Facility (AGRF), Melbourne.

2.4.2 Cell culture

Cell lines used for experiments in this thesis included COS-7, Hela, N1H-3T3, and

HEK293 cells. All cell lines were maintained in 100 mm culture dishes in DMEM supplemented with 10% FBS, L-glutamine (2 mM), and IX penicillin/streptomycin at

37°C in a humidified 5% CO2 incubator. When confluent, the culture medium was removed by aspiration and the cells were rinsed with warmed PBS. Cells were then incubated in 1 mL of 0.25% trypsin-EDTA solution at 37°C until the cells detached from the dish. Growth medium (9 mL) was added to dilute the trypsin solution, and cells were transferred to a sterile tube and centrifuged at 1,000 rpm for 5 min. The supernatant was carefully aspirated and the pellet was resuspended in growth medium

107 and was further diluted (1:5 to 1:10) with fresh culture medium in new cell culture dishes.

2.4.3 Transient transfection

2.4.3.1 Gene Juice

COS-7, NIH-3T3 or HeLa cells were plated the day before transfection so the confluence reached 70-80% on the day of transfection. Cells were transfected with 1.5 pg DNA per well (6-well plate) or 6 pg DNA per 100 mm dish. GeneJuice (2:1 ratio with DNA) was first diluted into 100 pL (6-well plate) or 800 pL (100 mm dish) Opti-

MEM and incubated for 5 min at room temperature. DNA was then added into the mixture, mixed gently by pipetting, and incubated for 15 min at room temperature.

DNA/GeneJuice mixture was added drop-wise to cells in complete growth medium and the dish was gently rocked to ensure even distribution of DNA/GeneJuice complexes.

Cells were incubated for 24 h in the incubator before harvesting for immunoprecipitation or Western blot analysis.

2.4.3.2 Calcium phosphate transfection

HEK293 cells were transiently transfected by calcium phosphate precipitation method.

The day before transfection, one confluent dish was split in a 1:5 dilution to new 100 mm dishes. Cells should be around 50-60% confluent on the day of transfection. For each transfection, DNA (typically, 5 pg per construct) and sterile water were added to a total volume of 450 pL. Fifty microlitres of 2.5 M CaCE was added to each tube. Then,

500 pL of sterile 2X HBS (274 mM NaCl, 10 mM KC1, 1.4 mM NaiHPCrt, 15 mM D- glucose and 42 mM HEPES, pH 7.05) was added dropwise to the DNA mixture and incubated for 20 min at room temperature. Transfection solution was added gently to

HEK293 cells prepared previously and media change was performed 6-9 h later. Cells were harvested 48 h post-transfection for immunoprecipitation and Western blot analysis.

108 2.4.4 Immunofluorescence

Cells were plated on 20 mm glass coverslips in a 12-well plate. Cells were maintained

and/or transiently transfected as described above and were washed once in PBS prior to

fixation.

For paraformaldehyde (PFA) fixation, cells were incubated with 4% PFA for 15 min at

room temperature followed by 3 x 5 min washes with PBS. Cells were then

permeabilised with 0.25% Triton X-100 (in PBS) for 10 min. For methanol fixation,

cells were incubated in ice-cold methanol for 10 min at -20°C followed by one wash

with 0.1% Triton X-100 (in PBS) and another wash with PBS. Following fixation, cells

were incubated in blocking buffer (10% BSA in PBS) for 1 h at room temperature. Cells

were then incubated with primary antibodies (diluted in 1% BSA) overnight at 4°C.

Cells were washed five times with PBS and then incubated in the dark with a 1:500

dilution of Alexa Fluor-conjugated secondary antibodies (Molecular Probe, Invitrogen)

for I h at room temperature. Coverslips were washed five more times with PBS before

they were mounted on microscope slides with a drop of anti-fade solution (23.3 mg/mL

DABCO, 20 mM Tris-HCl pH 8.0, 90% glycerol). Images were acquired using a Zeiss

epifluorescence microscope with a 40X or oil-immersion 60X, 100X objective with

appropriate filter sets and a cooled CCD camera. Images were further processed for

illustrative purposes using Adobe Photoshop.

2.4.5 Protein degradation rate analysis

COS-7 or HEK293 cells were transiently transfected with plasmid encoding the protein to be analysed. After 24 hours, cells were pre-incubated with either MG 132 (20 pM) to

block protein degradation via the 26S proteosome pathway or vehicle (DMSO) for 30

min. Cycloheximide was incubated with cells at a final concentration of 100 pg/ml for

varying lengths of time. At each time point, cells were lysed with 2X Laemmli SDS sample buffer, sonicated to shear the homogenate and incubated at 95°C for 5 min. 109 2.4.6 Luciferase reporter assay

COS-7 cells were transiently transfected with 0.5 pg of pGL3-basic (control) or pGL-

GUR luc+ gene reporter-containing plasmid, in addition to 1 pg of the tested activating plasmid (pCDNA3.1-GTF2IRDl) or pCDNA3.1 alone. Transfections were done at least in duplicate per condition in a 12-well plate(s) using GeneJuice. After 24 hours, luciferase activities were assayed with the Luciferase Assay System (Promega) according to the recommended protocol. Briefly, cells were gently washed in PBS and lysed in 100 pL of IX passive lysis buffer (PLB). The dish was rocked horizontally several times to ensure even lysis. Lysates were scrapedto remove all material and transferred into microfuge tubes and kept on ice. Tubes were vortexed 10-15 sec and centrifuged at 4°C at maximum speed for 1 min. The supernatant was transferred into fresh tubes, kept on ice or stored at -80°C. Luminescence reading was carried out using an opaque 96-well plate. Luciferase assay reagent (LAR, 100 pL per well) was brought down to ambient temperature before adding into each lysate (20 pL) in each well.

Luminescence was read immediately using a SpectraMax microplate reader (Molecular

Devices). To normalise the luciferase activity to the amount of total lysate, a BCA assay was also performed using 20 pL of each sample (section 2,3.5).

2.5 Yeast two-hybrid assay

Many routines for culturing and handling yeast, as well as media recipes, were obtained from the Yeast Protocol Handbook (Clontech Laboratories, Inc.).

2.5.1 Small-scale yeast transformations

S. cerevisiae strain AH 109 was transformed using the lithium acetate (LiAc)/ salmon sperm DNA (ssDNA)/ polyethylene glycol (PEG) protocol (Gietz & Woods, 2002). For

10 transformations, a single colony of 2-3 mm in diameter (not more than 3 weeks old) was innoculated into 5 mL of YPDA medium (50 g/L YPD broth supplemented with 60 mg/L adenine hemisulfate) and incubated at 30°C with shaking at 200 rpm overnight.

1 10 The next morning, the number of cells was counted using haemocytometer and should at least be lxlO7 cells per mL. Yeast culture was diluted (normally between 1:5 to 1:10) with fresh pre-warmed YPDA medium to make a concentration of 5x10(1 cells per mL.

The culture was grown for another 3-5 hours until the cell titer reached at least 2 xlO7 cells per mL (2 cell divisions). Yeast cells were harvested by centrifugation in a 50 mL tube at 1,000 x g for 10 min using a benchtop centrifuge. The medium was discarded and cells were washed with 25 mL sterile water and centrifuged again at 1,000 x g for 5 min. The supernatant was discarded and resuspended well in 1 mL sterile water before pippetting• • 100 pL (approximately 1x10 Q cells) into microfuge tubes for each transformation. Excess liquid was removed from the cells by gently centrifuging the tubes at 8,000 rpm for 30 sec. Transformation mixture containing (per transformation):

240 pL of 50% PEG-3350 (w/v), 36 pL of 1M LiAc, 10 pL of salmon sperm DNA

(boiled at 95°C for 10 min prior to use), plasmid DNA (100 ng per plasmid) and sterile water was prepared to a total volume of 360 pL. Yeast cells and transformation mixture containing the plasmid(s) were vortexed well for approximately 1 min. Cells were incubated at 30°C for 30 min, followed by a 20 min heat shock step at 42°C. Cells were pelleted by centrifugation at 8,000 rpm for 30 sec and resuspended in 500 pL of sterile water. Half of the cell suspension was plated on an appropriate selective medium (6.7 g/L SD base, pH 5.8 containing IX drop-out supplement, 2% glucose and 2% w/v agar).

Yeast clones containing protein that interacts with the bait were selected on SD medium that lacks leucine (Leu), tryptophan (Trp), adenine (Ade) and histidine (His), referred to as quadruple dropout (QDO) medium. The stringency of the interaction was further assayed by the addition of 3-aminotriazole (3-AT) at a final concentration between 5-20 mM. Plates were then incubated at 30°C for 3-5 days.

2.5.2 Y2H library screen

The flow chart in Figure 2.1 summarises the procedure used for the Y2H library screen.

The screen was carried out using the Matchmaker Mate & Plate System - universal mouse (normalised) library (Clontech) according to the user manual. The bait (pGBKT7-GTF2IRDl) was transformed into the AH 109 yeast strain using the

LiAc/ssDNA/PEG method described above. This bait was tested for auto-activation and toxicity in a similar mating procedure at a smaller scale by using the control pGADT7-T transformed Y187 yeast strain, which resulted in no detectable activation of the selective markers. Meanwhile, a positive control mating was set up using pGBK.T7-53

[AH 109] and pGADT7-T [Y187] mating which resulted in a strong interaction between the two proteins (p53 and SV40 large T-antigen). Mated diploids were selected on double dropout (DDO or SD -Leu, -Trp) medium while interactions between the hybrid proteins were selected on QDO (SD -Leu, -Trp, -Ade, -His) medium.

For the large-scale mating, a concentrated bait culture was prepared by innoculating a fresh (2-3 mm) colony of pGBKT7-GTF2IRDl [AH 109] into 50 mL of SD medium that lacks tryptophan (-Trp). Culture was allowed to grow overnight at 30°C with shaking at 250-270 rpm until the OD6oo reached 0.8 or total number of cells was at least

1 x 10s*. Cells were pelleted by centrifugation at 1,000 x g for 5 min. Cells were

o resuspended in 4-5 mL of SD -Trp medium to a final density of >1x10 cells per mL. 1 mL of the library strain was rapidly thawed to room temperature in a water bath and was combined with the bait strain in a sterile 2L flask. To allow for slow mating, 45 mL of 2X YPDA medium (lOOg/L YPD broth, 60 mg/L adenine hemisulfate with 50 pg/mL kanamycin) was added and incubated at 30°C for 20-24 hour with shaking at 30-50 rpm. Cells were centrifuged at 1,000 x g for 10 min, washed twice with 50 mL of 0.5X

YPDA (with 50 pg/mL kanamycin) and rinsed through the 2L Bask each time to collect any remaining cells in the flask. Washed cells were resuspended well in 10 mL of 0.5X

YPDA/Kan medium. The total volume was measured to distribute the culture equally over 45 x 150 mm plates (QDO medium). Plates were incubated at 30°C for up to 6 days with the growth of colonies checked daily after 3 days. Colonies were re-streaked into fresh QDO plates containing X-a-galactosidase (1 mL of 20 mg/mL stock per 500 mL medium) to verify interactions that are genuine. When confirmed positive, the yeast colonies were subjected to plasmid isolation for gene identification and further analysis.

112 2.5.3 Plasmid rescue from yeast

The protocol to isolate plasmids from yeast was adapted from a previous report

(Hoffman & Winston, 1987). The prey library plasmids were rescued from yeast cells grown on QDO (with or without X-a-gal). A healthy yeast colony was grown overnight to saturation in 2 mL of QDO medium at 30°C at 250 rpm. Cells were collected by centrifugation at 1,000 x g for 10 min. The supernatant was removed, the cells were resuspended in 0.2 mL of Buffer A (2% Triton X-100, 1 % SDS, 100 mM NaCl, 10 mM

Tris-HCl pH 8.0, 1 mM EDTA) and transferred into a microfuge tube containing approximately 150 pL volume of glass beads (425-600 pm, Sigma). A mixture (0.2 mL) of phenol:chloroform:isoamyl alcohol (25:24:1) was added into the tubes and vortexed for 2 min to break up the cell wall of the yeast. Samples were centrifuged for 5 min at

14,000 rpm. The DNA aqueous layer was collected into a fresh tube and incubated with

240 pL of 100% ethanol on ice for 30 min. Precipitated DNA was centrifuged for 15 min at 14,000 rpm at 4°C. Supernatant was discarded and DNA was washed twice with ice-cold 70% ethanol, centrifuged at 14,000 rpm for 5 min at room temperature each time. After the second wash, the DNA pellet was air-dried for 5 min and then resuspended in 40 pL of sterile water. The isolated plasmid DNA was transformed into electrocompetent E. coli (DH5a) cells by electroporation for greater efficiency. Briefly,

1 pL of DNA was mixed with 50 pL of DH5a in a microfuge tube and was then transferred into a pre-ch i 1 led electroporation cuvette (1 mm gap, Bio-Rad).

Electroporation was performed at 1.8 kV, 200 ohms, and 25 pFd using Gene Pulsar

Xcell electroporation system (BioRad). The cuvette was placed back immediately onto ice and cells were recovered by adding 200 pL of SOC medium and incubating the cells at 37°C for 1 h at 250 rpm. The cells were plated on an LB-Ampicillin agar plate at

37°C overnight.

113 2.5.4 Identification of library clones

The prey library plasmid, selected by ampicillin, was isolated by mini-prep and analysed by restriction digests with XhoMEcoRX and/or Nde\/Cla\ to identify potential duplicate clones. Plasmids were sequenced using T7 and/or 3'AD sequencing primers

(Matchmaker Yeast Two-Hybrid System, Clontech), which flank the prey library inserts

(section 2.2.10).

114 Transform bait in AH 109 yeast Lithium acetate/PEG protocol (section 2.5.1)

Auto-activation test Figure 2.2.A-C of the bait

Mate bait and library on QDO plates (Section 2.5.2) * Pick and re-streak colonies for Figure 2.2. D stringency test

QDO + x-a-gal medium (and optional, 3-AT)

Grow mini yeast culture to rescue for the prey plasmid QDO medium (Section 2.5.3) * Transform plasmid into E.coli (DH5a) Selected on LB-Amp for colonies containing the prey plasmid Restriction digest analysis to eliminate Sequencing of the identical clones isolated plasmid Figure 2.2. E (Section 2.5.4) * Back transformation into the yeast to confirm interaction Lithium acetate/PEG protocol

Figure 2.1 Flowchart diagram of the Y2H library screening procedures Figure 2.2 Representative images of assays performed within Y2H library screening.

(A),(B) Auto-activation test of the bait pGBKT7 -GTF2IRD1 in AH 109 was performed by mating with Y187 strain transformed with a negative prey, pGADT7-T which expresses Gal4-AD-SV 40 large T- antigen. Positive mating was demonstrated by pGBKT7-p53 [AH 109] interaction with pGADT7-T [Y187]. The mated yeast cultures were plated on media that lacked Leu, Trp, Ade and His (QDO) with (B) or without (A) x-a-galactosidase (x-a-gal) which select for the positive interacting diploids by survival or by colour in the presence of x-a- gal. (C) The absence of cell toxicity effect was shown by growing the mated diploids in double dropout (DDO) medium lacking Leu and Trp, with the correct activation of the blue colour in the presence of positive interaction. (D) Stringency test of a batch of yeast colonies identified in the library screening was performed by re-growing on QDO + x-a-gal medium. Positive clones were subjected to plasmid rescue. (E) Restriction digest analysis of a batch of isolated plasmids using EcoRI and Xhol (top) or Ndel and Clal (bottom) which are found in the multiple restriction sites of the library prey plasmids. Identical digest pattern of both reactions indicating potential identical clones. Positive Positive control control

QDO

Cell viability or toxicity test

DDO + x-a-gal CHAPTER 3 - GENE EXPRESSION ANALYSIS OF GTF2IRD1 KNOCKOUT MOUSE BRAIN

GTF2IRD1 has been widely regarded as a transcription factor since it was discovered as a sequence-specific DNA binding protein in yeast one-hybrid screens (see Chapter 1, section 1.3.3.1). The DNA binding capability of GTF2IRD1 is mainly mediated through the repeat domains (RDs), each of which possess variable affinity for a specific DNA recognition sequence (Palmer et al, 2010; Polly et al, 2003; Vullhorst & Buonanno,

2003; Vullhorst & Buonanno, 2005). GTF2IRD1 has been shown to bind multmerised regions taken from the upstream regulatory elements of Troponin 1 slow (Calvo et ai

2001; O'Mahoney et al, 1998), Hoxc8 (Bayarsaihan & Ruddle, 2000) and Goosecoid

(Ring et al, 2002). While these genes are regarded as target genes of GTF2IRD1, much of the evidence has come from in vitro binding assays or cell culture transactivation assays. The physiological relevance of GTF2IRDUs involvement with these genes has not been demonstrated.

Gene expression profiling by microarray analysis has been used in an attempt to identify candidate target genes of GTF2IRD1 in two previous studies. Chimge et al. (2007) investigated the impact of GTF2IRD1 over-expression in cultured mouse embryonic fibroblasts. This analysis identified approximately 2,000 genes, of diverse biological function, that showed large magnitude alterations in expression level (1.7- to 22-fold).

Using a gene-trap-based Gtf2ircll mutant mouse line, Enkhmandakh et al. (2009) performed a microarray comparison of mRNA from mutant and wild type embryos. 38 up-regulated and 498 down-regulated genes were identified that did not cluster into functional gene ontology categories. Neither study resulted in the unambiguous identification of GTF2IRD1 gene targets.

At the outset of our studies, we hypothesised that GTF2IRD1 may be directly regulating a number of genes that mediate important cellular and physiological processes 118 underlying normal behavioural and cognitive functions of a brain. In accordance with the proposed role of GTF2IRD1 hemizygosity in the behavioural and neurocognitive pathology of WBS, expression mapping of Gtf2irdl has demonstrated prominent and regionalised expression in the central nervous system (Howard et al., submitted; see

Chapter 1). In the postnatal and adult mouse brain, neuronal expression of Gtf2irdl overlaps extensively, but not exclusively with the GABAergic neuronal populations

(Figure 1.6.). Several Gtf2irdl mutant mouse models have been generated and characterised that have behavioural and neurological phenotypes, albeit with variable phenotypic consequences. This variability is most likely due to the different genetic mutation and background strain of each mouse model (see Chapter 1, section 1.3.3.3).

Throughout this study, we utilised the C57BL/6-GtJ2irdl"",l,ul knockout mouse line as a model to investigate molecular pathway(s) that are regulated by GTF2IRD1 in the brain.

These mice display a motor coordination deficit, a novel audible vocalisation phenotype involving increased use of both audible and ultrasonic vocalisations in stressful environments, which could be linked to excessive worrying and specific phobias in

WBS patients. In addition, the mutant mice are hyperactive during the dark cycle which is reminiscent of the ADHD symptoms in WBS patients (Howard et al, submitted; see section 1.3.3.3). Furthermore, differential responses of the knockout mice to GABA receptor antagonist and agonist drugs indicate an underlying molecular alteration to the

GABAergic system.

In this chapter, we aimed to identify genes expressed in the brain that are regulated by

GTF2IRD1 at the transcriptional level using an unbiased microarray analysis. This method allows us to quantitatively compare the genome-wide gene expression profiling in the Gtf2irdl homozygous knockout (KO) versus wildtype (WT) brain.

3.1 Overview of experimental approaches

Microarray analysis is a powerful tool for identifying gene expression networks underlying complex disorders; however, specific problems do arise when dealing with 1 19 the complexity and heterogeneity of brain tissue. It is known that there are a significant proportion of low abundance transcripts that display restricted spatial and temporal expression in the brain, which in turn dictate specific aspects of neuronal phenotypes

(Chaudhari & Hahn, 1983; Levitt et al, 1997; McConnell, 1995). Thus, subtle changes in transcript levels in these diverse cell populations would likely be masked in microarray experiments when using whole brain tissues. This problem is further magnified by the complex cellular heterogeneity, which includes non-neuronal and diverse types of neuronal cells.

The problem of the ‘dilution effect' was also evident in our previous microarray analyses using whole brain tissue. These experiments involved pooled samples collected from 5 mice of each genotype (wildtype, heterozygous and homozygous KO) at 8-12 weeks of age, the age at which the mouse behavioural studies were conducted (Tay thesis, 2005). In two consecutive independent microarray experiments (conducted by the AGRF, Melbourne, using the Affymetrix platform), only 7 common genes were identified as up- or down-regulated in either heterozygous or homozygous mutant mice compared to the wild type controls (Widagdo thesis, 2006). Further analyses by qRT-

PCR and northern blotting confirmed the up-regulation of one gene, Serpina3n, in both heterozygous and homozygous KO mouse brains. SERPINA3N is a serine protease inhibitor that has no known neuronal function but has homology with Neuroserpin, which has been shown to be protective against neurotoxicity (Lebeurrier et al, 2005).

In this chapter, two major experimental approaches were taken to analyse the gene expression profile in the brains of Gtf2irdl KO mice (Figure 3.1). The first (section 3.2) involved gene expression analyses of specific regions of the brain (cortex, cerebellum, olfactory bulb and hippocampus) isolated from either WT or homozygous KO mice.

This approach should reduce the complexity of the sample by focusing on functional domains and would allow comparison between these major brain regions. However, this approach still faces the risk of the dilution effect described above, as Gtflirdl is largely

120 in the GABAergic neurons. Thus, the second approach (section 3.3) involved microarray analysis of purified GABAergic neuronal population from Gtf2irdl""",ni homozygous KO and WT control mice. Detailed experimental descriptions and findings of each approach are described in the next sections, lllumina Sentrix Mouse-6 BeadChip arrays were used throughout this study. The RNA quality control, processing (reverse transcription and labelling of cDNA) and array hybridisations were carried out by microarray facilities as described in Chapter 2 (Materials and Methods).

3.2 Brain region-specific microarray analysis

By restricting the microarray analysis to specific brain regions, it was felt that a more detailed analysis was possible, as well as providing a means to cross-compare results.

However, as large microarray studies can be costly, total RNA samples were extracted from various brain regions of multiple animals (n=7 per genotype) and pooled for array processing. To minimise biological variations between subjects in each pooled sample, all mice were sex- (males) and age-matched (postnatal day 10-12) and for each dissection at least a pair of WT and homozygous KO mice were collected from the same litter to minimise between-litter variation. An early postnatal developmental time point was selected since an upregulation of Gtflirdl expression has been observed in the

Gtf2irdl,m2(LacZ)Hni mice at approximately postnatal day 7 (Tay E, unpublished data).

The pooled samples submitted for the microarray analyses contained equal proportions of individual samples, representing total RNA samples from each brain region and each genotype: cortex (WT, KO), hippocampus (WT, KO), olfactory bulbs (WT, KO) and cerebellum (WT, KO). Three of the brain regions were hybridised simultaneously on the same chip for logistical reasons and the cerebellum samples were hybridised at a later time point.

121 3.2.1 Comparison of gene expression profiles from different brain regions

Following data pre-processing and normalisation (done by Institute of Molecular

Biosciences microarray facility at the University of Queensland), data filtering was performed to identify genes that were differentially expressed between genotypes in different brain regions. An initial filter cut-off of a minimum 2-fold change yielded a total of 32 different entities in the three regions: 4 in the cortex, 16 in the hippocampus, and 21 in the olfactory bulb, and 3 entities were found to be common in all three regions

(Figure 3.2). One of these 3 common entities corresponded to the gene Actl6b and the other two corresponded to Zfp68 (Table 3.1). The rest of the entities were dysregulated in either one or two brain regions only (Table 3.1). Due to the redundancy of probes

(50-mer oligonucleotides) present on the Illumina chip, multiple probes corresponding to the same gene were identified. The same analysis using data obtained from the microarray experiment of the cerebellum region identified 10 differentially expressed entities (data not shown), 3 of which corresponded to the same oligonucleotides of

Actl6b and Zfp68, identified in the other brain regions.

Those genes that showed the same direction in their dysregulation, either increased or decreased expression levels (not necessarily differentially expressed by >2-fold) could potentially be genuine consequences of Gtf2irdl loss. Examples of such genes include

Calnl, Prkag2 and Taf6 (Table 3.1). In addition, an increased level of mutant Gtf2irdl transcript was detected in all brain regions from the Gtf2irdl"",llll/ homozygous KO mice (data not shown), confirming our previous result on the autoregulatory negative feedback mechanism of GTF2IRD1 (Palmer et al, 2010). A subtle, but increased level of Serpina3n was also observed in the cortex and olfactory bulb of Gtj2irdl""lu,d KO mice (~1.4-fold), in agreement with the previous whole brain screen.

Using a 2-fold change filter, overall changes in the transcript profiles in different brain regions of Gtjlirdl KO mice were subtle and involved a small number of genes. For

122 this reason, data analysis using the gene ontology tool for gene categorisation was not deemed to be necessary.

3.2.2 Analysis of the differentially expressed genes by qRT-PCR

The microarray experiments showed that Actl6b (also known as BAF53b) and Zfp68 were down-regulated by 2- to 3-fold across all regions of the brain in KO mice. To confirm the changes in expression level of these genes, quantitative real-time PCR

(qRT-PCR) was performed using cDNA samples synthesised by reverse transcription from: (a) individual RNA samples which were submitted as a pooled sample for the array analysis (depending on the availability or starting amount of the individual sample) and (b) RNA samples extracted from a subset of sex- and age-matched mice that had been newly acquired.

Primers for qRT-PCR analyses of Zfp68 (zfp68F and zfp68R) and Actl6b (BAF53bF and BAF53bR) were designed to amplify cDNA regions overlapping with the identified

50-mer probes, which were frequently located in the 3' untranslated region (UTR) of the transcripts. The qRT-PCR analysis of Zfp68 did not reveal a significant change in expression level in any brain regions of the Gtf2irdl KO mice compared to the WT controls (Figure 3.3). In contrast, expression of Actl6b was shown to be reduced by 2- to

3-fold, similar to the level of reduction shown by the microarray experiment. The down- regulation was confirmed in all regions of the brain of the homozygous null, as well as in the hippocampus of Gtf2irdl heterozygous KO mice (Figure 3.3).

Other potential candidate genes, including Calnl, Prkag2, Taf6 and Ttr, were also analysed by qRT-PCR using primers designed against each gene. No significant differences in the expression level of any of these genes were evident between Gtf2irdl

KO and WT mice in any of the brain regions analysed (data not shown).

123 3.2.3 Analysis of transcript and protein from theAct/6b gene

After confirming the microarray data on ActI6b by qRT-PCR, we investigated the expression level of its protein product which is widely referred to as BAF53b. Western blot analysis was performed on proteins extracted from various brain regions

(cerebellum, olfactory bulb and cortex) derived from WT, heterozygous, and homozygous null Gtflirdl mice using an antibody against BAF53b (Olave et al, 2002).

Surprisingly, we did not find significant changes in the level of BAF53b in brain regions of the Gtflirdl KO mice compared to the WT controls (Figure 3.4.A). Equal loading of all samples was confirmed by post-hoc Coomassie staining of the membranes

(data not shown).

This result prompted a reinvestigation of the changes at the transcript level previously shown by qRT-PCR analysis using the independent method of Northern blot analysis.

Pooled total RNA samples from the cerebellum of WT, heterozygous and homozygous

KO mice were used for the analysis. Consistent with the Western blot result, no evidence of differential expression of Actl6b transcript was found between genotypes

(Figure 3.4.B). These data suggested that our initial findings may be confounded by the qRT-PCR assay and the primer set used. To further interrogate our initial results, a new set of PCR primers was designed targeting the coding region rather than the 3'UTR of

Actl6b. Consistent with the Western and Northern blotting data, no changes in Actl6b expression level could be detected by this qPCR assay (Figure 3.4.C). This confirmed that the decreased level of Actl6b detected by qRT-PCR assay and the microarray could only be attributed to the 3'UTR.

3.2.4 Actl6b and genes closely linked to the mutant Gtf2irdl locus retain an un­ recombined 129Sv/J genomic region

It was intriguing to find that a number of the dysregulated genes identified by the microarray experiment were clustered within 5 Mb region on chromosome 5 G2, surrounding the Gtflirdl genomic locus (Chr 5: 134.833-134.932 Mb). These genes

124 include Ca/nl (Chr 5: 130.845 Mb), Actl6b (Chr 5: 137.994 Mb), Mospd3 (Chr 5:

138.038 Mb), Taf6 (Chr 5: 138.620 Mb) and Zfp68 (Chr 5: 139.046 Mb) (Figure 3.5.A).

It is also interesting to note that Prkag2 and 1500035N22Rik are located very closely with each other on chromosome 5 (Table 3.1). This led us to hypothesise that the microarray had detected quantitative differences in these genes because the original mutation had been developed in embryonic stem cells of 129Sv/J origin (Nagy et al,

1993) and then backcrossed on the C57BL/6J background. At the time of these experiments, only 5 generations of backcross had elapsed and therefore it is likely that a region surrounding the mutant Gtf2irdl allele would have selectively retained its

129Sv/J character. Therefore, single nucleotide polymorphisms (SNPs) in the 3'UTR of the candidate genes may have been enough to create differential hybridisation to the probe as a result of small mismatches.

To test this hypothesis, we sequenced the 3,UTR of the Actl6b transcript amplified from

WT and K.0 cDNA derived from the original brain RNA samples. The sequencing analysis confirmed the presence of two single base-pair mismatches within the identified array probe and the reverse primer used in the initial qPCR analysis (Figure

3.5.B). The nucleotides in the KO transcripts, which were specific to the 129Sv/J background, and those in the WT transcripts, which were specific to the C57BL/6J background, have been documented in the SNPs database (www.informatics.jax.oru;

Figure 3.5.C). In contrast, no SNPs were found within another 50-mer Illumina Actk6b probe, which is located just upstream of the stop codon, and in accordance with the hypothesis, did not indicate a reduction of Actl6b level in the microarray analysis.

Similarly, we also found that the two Illumina probes against Zfp68 also contain one

SNP (rs29551605) within their overlapping sequence (data not shown). Therefore, we conclude that the reduction in Actl6b and Zfp68 expression detected by initial microarray and qPCR analyses were artifacts due to the presence of SNPs in the mutant allele positioned within the array probe and qPCR primers.

125 3.2.5 Discussion of the brain region-specific microarray analyses

Overall, the brain region-specific microarray analyses revealed relatively subtle differences between the gene expression profiles of Gtf2irdl KO and the WT mice. Out of -45,000 transcript-specific probes in the lllumina mouse array, only 32 probes representing 29 transcripts were identified as >2-fold differentially expressed in the cortex, the hippocampus and the olfactory bulb of Gtf2irdl homozygous KO mice. The least number (3 transcripts) were identified in the cortex and all three were explicable by the SNP alterations. This may reflect the sparse expression of Gtf2irdl in this region, but in addition, the cerebral cortex is the largest and anatomically the most complex part of the brain. It has been experimentally shown that the more complex the tissue is, the less likely the ability to detect subtle differences in gene expression by microarray

(Sandberg et al, 2000).

Our initial screening identified two prominent candidate genes, Actl6b and Zfp68, that were dysregulated across all brain regions of the Gtflirdl KO mice. BAF53b is a member of a neuron-specific chromatin remodelling complex Brg/Brm associated factor

(BAF) and has an important role in dendritic development (Olave et al, 2002; Wu et al, 2007), making it an attractive downstream target gene of GTF2IRD1. Flowever, we were not able to validate the reduced expression of Actl6b both at the transcript and protein levels in our subsequent experiments, due to array probe mismatch and SNPs present in the KO mice. A number of other genes clustered within the chromosomal region and identified as differentially expressed by these microarray analyses were also likely to be caused by incomplete genomic recombination during backcrossing of the

KO animals. The issues encountered in this study demonstrated; however, the high sensitivity of the lllumina array in detecting differences caused by single mismatches, as well as highlighting the reproducibility of findings in different tissue isolations. Gtflirdl KO mice used throughout this study have gone through 5 rounds of meiotic recombination (N5 backcross generation) and therefore, are estimated to contain

96.9% of C57BL/6J genomic DNA and the remaining 3.1% would be of 129Sv/J

126 genomic DNA origin, randomly distributed throughout the genome (Wong. 2002).

However, the distribution is not random around the Gtflirdl allele which is selected at each generation by PCR genotyping assays. As the frequency of recombination between the two loci is roughly proportional to the length of DNA that separates them, it makes sense that the genes in relative close proximity to the Gtflirdl locus would have resulted in a lower probability of recombination between these alleles. A 1% recombination frequency is equivalent to 1 centimorgan of genetic distance, which is roughly equivalent to an average of 2 Mb of DNA in the mouse. Therefore at the N5 generation, one might predict that approximately 20 cM could be preserved either side of the Gtflirdl locus, equivalent to a 40 Mb region of genomic DNA.

3.3 Isolation of GABAergic neurons from the Gtflirdl KO mouse brain and mRNA microarray analysis profiles

The minimal changes in gene expression profiles from specific regions of the brain suggested that a more sensitive assay may be required to avoid a “dilution effect" when employing microarray analysis in a complex tissue such as the brain. This problem, which has been described as “finding needles in molecular haystacks" by Lewandowski et al. (2005), illustrates the difficulty in detecting subtle changes of low abundance mRNAs that are present in specific subsets of the neuronal population. Therefore, to increase the sensitivity of our array-based transcript profiling, we focussed our effort on a specific neuronal population that has been shown to overlap substantially with the expression of Gtflirdl', the GABAergic neurons.

The Gacu">,l(h('H‘)lama mouse line characterised by Tamamaki et al. (2003) is widely used in studies that require a reliable in vivo marker of GABAergic neurons because it encodes the GAD67 enzyme which is required for the synthesis of GABA. In our expression analysis study, this mouse line was obtained and crossed with the

Gtflirdl"u2n'xi mouse line which revealed a substantial overlap between both reporters in the brain - GFP for the GABAergic neurons and LacZ staining for Gtflirdl 127 expression (Howard et al, submitted; see Chapter 1). For the GABAergic-specific gene expression study described in this section, the Qa(jj,m,Tama m0use was crossed with the C57BL/6-G(/2/>c//""///'

3.3.1 Mice used for GABAergic neuron isolation

In order to obtain GFP-positive mice with either the wildtype or mutant Gtf2irdl allele, a pair of double heterozygous Gadl""l(EGFP)Tama/Gtf2irdl"",Hrcl mutant mice were crossed. As homozygous null mice from the Qa(jj,m,(h(jhP>lama mutant line die in utero due to the embryonic lethality of GAD67 loss, the breeding produced litters of mice that were wildtype at the Gadl locus (GAD67-GFP negative) or heterozygous for Gcid]""l(hCjhP>Tama (GAD67-GFP positive). Segregation of the Gtf2irdl allele should lead to the inheritance of homozygous WT, heterozygous WT/K.O and homozygous KO in a

1:2:1 ratio (Figure 3.6.A).

3.3.2 Optimisation of the FACS isolation of GABAergic neurons

At the time of this analysis, very few studies involving purification and array-profiling of specific neuronal populations isolated from mouse brain had been reported (Arlotta et al, 2005; Lobo et al, 2006). Neurons are post-mitotic cells and those isolated from the post-natal mouse brain are known to be fragile. In addition, the processing time from tissue dissociation and single cell isolation could induce loss of cell viability and stress that leads to transcriptional changes. To avoid these problems, careful optimisation of

128 protocols and quality controls needed to be in place to ensure isolation of high quality

RNA samples required for microarray analysis.

The procedure was divided into two parts for optimisation:

(i) Dissociation of brain tissue by enzymatic and mechanical processes - the main goal of this step is to generate a single-cell suspension of neurons with high viability. This technique was optimised prior to commencing cell sorting by FACS.

(ii) Cell sorting and RNA extraction - this procedure involves selecting the appropriate sorting buffer, collection buffer and the RNA extraction methodology to maximise the recovery and integrity of cells and the isolated RNA. Optimisation of these procedures required collection of cells, unlike the analytical flow cytometry used in part (i).

The quality of the pre- and post-sorted cells was examined by flow cytometry and microscopy-based analyses. Analytical flow cytometry (performed using the BD

FACSCanto) displayed the profile of the single cell suspension based on the standard light scattering parameters - forward scatter (correlates to the particle size) and side scatter (correlates to the cell granularity), as well as the fluorescence detection of EGFP and propidium iodide (PI) signals. PI intercalates into the DNA of dead cells, but is excluded from live cells, hence serving as a marker of cell viability. Cell sorting

(performed on the BD FACSVantage cell sorter) was done using these similar parameters. Also, cell morphology was visualised and imaged by fluorescence microscopy. Detailed protocols and reagents used in the FACS procedure are described in Chapter 2.

Tissue dissociation (pre-sorting optimisation)

Preparation of a single cell suspension from brain tissue is a crucial step for the success of FACS-based cell isolation. Post-natal brain tissue dissociation must be executed in the gentlest and most efficient way possible in order to recover the maximum number of all neuronal cell types. This procedure involves an enzymatic digestion, which degrades

129 the extracellular adhesion proteins that maintain the structural integrity of the tissue.

This, in turn, facilitates the second step of mechanical dissociation.

Two protocols were tested that utilised different sets of reagents: (i) a protocol used in preparing cerebellar granule cells for primary culture from postnatal day 7 mouse brains

(Anggono et al, 2008) and (ii) the neural tissue dissociation kit (NTDK) manufactured by Miltenyi Biotec. Although both protocols use trypsin for the enzymatic digestion of brain tissues, they differ in the compositions of the buffering media (specific ingredients are not disclosed in NTDK). When comparing these two protocols side by side, a higher yield was achieved using the NTDK (data not shown). FACS analysis of these samples detecting the EGFP and PI signals showed a low percentage of Pi-positive cells in the

EGFP-positive and negative populations, suggesting isolation of intact neuronal cells

(Figure 3.7). The background fluorescence was determined using dissociated brain cells from GAD67-GFP negative mice. FACS analysis of NTDK-dissociated cells from

GAD67-GFP mice, revealed a cluster of EGFP-positive cells (>1(E ') in the GAD67-

GFP positive sample, which were absent in the negative control sample (Figure 3.7).

The morphology and fluorescence of the single cell suspension was analysed further by microscopy, which revealed that most cells appeared rounded and unattached, typical of neuronal cells dissociated from early postnatal mouse brain samples. At this stage, we were able to observe scattered population of GFP-labelled GABAergic neurons among the much higher proportion of non-fluorescent, non-GABAergic neurons in the GAD67-

GFP positive sample. Some neurons retained their neurites, indicating the mildness and suitability of the dissociation protocol.

Cel! sorting and RNA extraction

Following multiple trials, it was determined that the L-15 buffered media used by Lobo et al. (2006) achieved the best recovery and viability of the sorted cells (data not shown). Gating criteria for the live sorting includes gating by cell size to exclude debris

- forward versus side scatter characteristics, followed by doublet discrimination gating

130 to exclude any unwanted GFP-negative cells that may adhere to the sorted EGFP- positive cells - based on area versus height signal ratio which tends to increase with aggregated cells (Figure 3.8.A). A conservative gating for the highly fluorescent GFP- positive cells was set at >10°, which typically represents 3.0-3.4% of total events detected (Figure 3.8.A). Comparison of pre- and post-sorted cells showed a high purity of GFP-positive GABAergic neurons obtained by the sorting routine (Figure 3.8.B).

Approximately 250,000 cells were routinely isolated from one brain using the FACS protocol. The sorted cells were lysed directly after sorting and processed using the

PicoPure RNA extraction kit (Arcturus), which on average yielded approximately 25 ng of RNA per brain. Due to the limited amount of RNA obtained from cells collected by the FACS procedure, multiple rounds of sorting were performed to collect sufficient

RNA for the array analysis (5 animals from each genotype - WT and homozygous KO).

RNA samples underwent a quality control check for integrity and concentration by the microarray facility (The Australian Genome Research Facility, Melbourne). The pooled

WT and KO RNA samples were divided into 3 replicates and all were hybridised simultaneously to a single Sentrix 6 chip.

3.3.3 Transcript profile of isolated GABAergic neurons from GtJUrdl KO and

WT mice

In contrast to the microarray data obtained by the previous approach, there were many genes identified as differentially expressed in the GABAergic neurons of Gtf2irdl KO mice compared to the WT control. Data filtering revealed a total of 552 genes (listed in

Appendix C-l) that were selected based on a change>3 -fold with a level of confidence of greater than 95% or p-value<0.05 (based on the triplicate array hybridisation). Gene ontology analysis revealed that the majority of genes identified fell into the category of protein binding functionality (~50%), followed by catalytic activity (-26%) while a relatively small subset was identified to have transcriptional regulator activity (-2%)

(Figure 3.9).

131 A manual assessment of the list of differentially regulated genes revealed some that are known to be expressed in and/or play roles in the GABAergic pathway, listed below.

Npy (neuropeptide Y) and Sst (somatostatin)

Neuropeptide Y and somatostatin are two neuropeptides expressed in subsets of

GABAergic neurons and are widely used as molecular markers of GABAergic neurons.

Both transcripts were down-regulated (~3-fold) in GABAergic neurons of Gtf2irdl KO mice.

Gria3 (AMPA-type glutamate receptor subunit 3)

Glutamate receptor 3 (GluA3) is known to be abundant in parvalbumin-expressing interneurons in the cortex and hippocampus. Gria3 knockout mice have reduced motor coordination and exploratory activity (hypoactive), as well as enhanced seizure susceptibility (Steenland et al, 2008). Gria3 was upregulated by ~3-fold (based on two probes) in Gtf2irdl KO mice. Slclal (solute carrier family 1)

Slclal or EEAC1 (excitatory amino acid carrier 1) serves as the predominant neuronal glutamate transporter, which co-localises to inhibitory GABAergic neurons. Expression of EEAC1 in the presynaptic GABAergic terminals is thought to be responsible for the uptake of glutamate, which subsequently contributes to re-synthesis of GABA. Down- regulation of EEAC1 leads to electrographic seizures (Sepkuty et al, 2002). Slclal was upregulated by 4- to 6-fold (based on two probes) in Gtflirdl KO mice.

Npas4 (neuronal PAS domain protein 4)

Npas4 was recently identified as a transcription factor that plays a role in the development of inhibitory synapses by regulating the expression of activity-dependent genes which control the number of GABA-releasing synapses that form on excitatory neurons (Lin et al, 2008). Npas4 was upregulated by 3.8-fold in the Gtjlirdl KO mice.

132 Dysregulation of these genes in the GABAergic neurons of Gtflirdl KO mice remains to be investigated fully as further tests by qRT-PCR were difficult since the RNA from the sorted cells was limiting, with the cell isolation protocol yielding only 25 ng per brain. A less rigorous, but available option was to examine the expression level of these genes in the RNA samples collected from brain regions as discussed earlier (section

3.2). It was hoped that some subtle changes could be detected in spite of the "‘dilution effect”. Quantitative RT-PCR analyses of Gria3, Npas4 and Slclal showed no significant changes in their expression levels in the cortical regions of Gtf2irdl KO mouse brain (Figure 3.9.B). It is estimated that GABAergic interneurons constitute approximately 20% of the total cortical neuronal population and therefore, changes in gene expression specific to these GABAergic neurons may be too dilute to detect in the context of the whole cortex (DeFelipe et al, 2002). Therefore, the results are inconclusive and a re-appraisal of the FACS procedure is needed, which at the time of this study was extremely difficult due to time constraints and limited resources. In future experiments it will be critical to obtain greater amounts of RNA from

GABAergic neurons than 25 ng per brain.

Other groups of genes highly represented in the list were also noted. Firstly, a number of myelin (the insulating sheath around the axon) associated genes were found to be differentially regulated (mostly up-regulated). These include Mog (myelin oligodendrocyte glycoprotein; up-regulated 17-fold), Mag (myelin-associated glycoprotein; up-regulated 11-fold), Mobp (myelin-associated oligodendrocytic basic protein; up-regulated 4-fold) and Mbp (myelin basic protein; up-regulated 4-fold). The genes are expressed at high levels in non-neuronal (glial) cells, such as oligodendrocytes or Schwann cells. This suggests that there was some contamination of the GABAergic neurons with non-neuronal cell types.

However, as expected, the cluster of genes on chromosome 5 close to Gtf2irdl, including Actl6b, Zfp68 and Taf6, were again identified as down-regulated in this

133 microarray analysis, indicating the sensitivity and reproducibility of some parts of the assay in spite of the presence of artefacts due to potential contamination.

3.3.4 Discussion of GABAergic neuron-specific microarray analyses

The analysis of gene expression profiles of GAD67-GFP-positive cells from Gtf2irdl

KO and WT control mice was carried out in order to identify genes that may be dysregulated in the GABAergic neuronal population known to express Gtf2irdl. This cell-specific microarray approach gets around the problem that arises from the “dilution effect” which potentially masked changes in gene expression in our initial brain microarray analyses. In this study, we observed a much larger subset of genes with differential expression than was identified in the previous study. Thus, our aim was partially achieved in this experiment. However, due to the technical difficulty inherent in isolating neuronal populations from the post natal mouse brain it is clear that a number of limitations remain, which need to be solved for future studies.

Firstly, a major difficulty was encountered in collecting sufficient RNA for the microarray pool and for follow-up analysis of candidates by qRT-PCR. Without independent biological replicates, data obtained from a single pooled sample may be compromised due to variable gene expression amongst the individual animals, enzymatic dissociation, FACS sorting and RNA extraction quality. With the large number of genes found to be dysregulated in this study, it is important to verify the changes of each gene expression. This leads to the second problem whereby, confirming tests by RT-PCR analysis could not be performed without re-undertaking the FACS procedure. Therefore, this study acts as a guide rather than as evidence for changes in transcript levels. Improvements in the technical and experimental design are required in order to exploit the data fully.

Our microarray analyses ended in 2008, after which Hardt et al. (2008) reported a procedure involving FACS purification of GABAergic neurons from the

134 Qa(j]lm,

defined GABAergic neurons from the cortex, olfactory bulb, striatum, and cerebellum

of the GAD67-GFP knock-in mouse. A methodology similar to ours was described

including single-cell suspension using the NTDK (Miltenyi Biotech), with the exception

that a papain-based enzyme mix was used instead of trypsin. In addition, different RNA

isolation and amplification methodologies were established which appeared to be the

key to the success of the study. The authors isolated and performed PCR amplification

of mRNA from 10,000 cells per FACS procedure, which generated sufficient cDNA for

individual array hybridisation. This allowed analysis of independent biological

replicates which permitted greater statistical power. Some interesting insights were

gained from comparing the gene expression profiles of the regional specific GABAergic

neurons, which were further validated by in situ hybridization and qRT-PCR. The

relative success achieved with these experimental techniques suggests that the protocols

can be adapted for use in future analyses of the Gtflirdl KO mouse brain.

3.4 In silico analysis of regulatory regions containing the putative GTF2IRD1 binding sites

Although neither of the microarray-based studies presented above identified genes with

statistically significant different expression in the Gtf2irdl KO mice, a large amount of

data has been generated which could be mined as a resource in future work. In a

separate attempt to identify genes which could be directly regulated by GTF2IRD1, we

used an in silico analysis to search for sequences in the genome which contain the

predicted binding sites of GTF2IRD1 in their regulatory regions. Using the mechanism

by GTF2IRD1 binds to its own promoter (Palmer et al, 2010), an algorithm was

designed to search for conserved binding elements in genomic DNA according to the

following criteria: (i) two binding sequences (GGATTA or the reverse complementary

TAATCC) must be present no further than 60 nucleotides apart from each other; (ii) the elements in (i) must be conserved in human and mouse orthologous regions; and (iii) the region must lie within 5 kb of a gene transcription start site. We then reasoned that if 135 the identified genes were regulated by GTF2IRD1 in vivo, then their expression levels are likely to be changed in the GtJ2irdlKO mice. The bioinformatics analysis described here was performed by Dr Bruno Gaeta at the School of Computer Science and

Engineering, UNSW.

Using the above criteria, searches within a 5-kb region upstream of the transcription start site of all genes in the mouse database identified 101 different transcripts

(Appendix C-2). Using the first microarray analysis (the regional brain analyses), the expression values of these individual transcripts in the Gtf2irdl KO and littermate control mice were manually examined. In particular, a weak differential expression of

>1.5-fold with consistent changes (either up- or down-regulation) in at least two brain regions of the Gtflirdl KO mice were considered as a ‘sign* of dysregulation. Using these criteria, we did not identify any genes from the list with altered gene expression.

3.5 Overall discussion

Efforts to find target genes that may be directly or indirectly regulated by GTF2IRD1 using a combination of genome-wide expression profiling and bioinformatics analysis have not yielded definitive results.

Although there were limitations to each experimental approach, it is possible that our initial hypothesis for GTF2IRD1 function, as a transcription factor, may not be correct.

GTF2IRD1 as a DNA-binding transcription factor derives from the various experiments that have been referenced (Bayarsaihan & Ruddle, 2000; Polly et al, 2003; Ring et al,

2002). However, its effect on gene transcription could be much more subtle than expected. This is reflected in the relatively subtle neurological phenotypes of the

Gtf2irdl'm,,,ul KO mice, many of which are only apparent in certain environmental or stimulus-dependent contexts.

136 Our recent refinement of the pattern of Gtf2irdl expression in the brain will allow us to determine the specific region(s) of the brain in which gene dysregulation is most likely to occur. These regions may include the striatum (caudatoputamen), and the superior and inferior colliculi in the anterior half of the midbrain (see Chapter 1, Figure 1.6), all of which can be identified and isolated using the stereotaxic atlas of mouse brain coronal sections. The importance of targeting the right brain region for gene expression analysis has been emphasised in studies of MECP2, the methyl-CpG-binding protein, where mutations in the gene are known to cause Rett syndrome (Chahrour & Zoghbi,

2007). Microarray analysis of whole brains from Mecp2 KO and WT mice revealed only subtle differences in gene expression (Tudor et al, 2002). However, by identifying a specific region, the hypothalamus, which is associated with specific neurological phenotypes of the mutant mice, as well as of Rett syndrome patients, led the investigators to perform microarray analysis using RNA extracted from this region

(Chahrour et al, 2008). This study identified a wide range of genes dysregulated in the hypothalamic system of the Mecp2 mutant mice, leading to greater insights into the gene regulatory mechanism of MECP2.

Developmental time points are also an important consideration in relation to the function of GTF2IRD1 in the brain and its impact on gene regulation. Analysis of

Gtf2irdl expression based on LacZ production has shown that Gtf2irdl is abundantly expressed in the fetal brain in a pattern that is generally similar to the adult, although there are some distinctive differences (Palmer et al, 2007) (discussed in section 1.3.3.2).

LacZ analysis of the early post-natal brain shows a pattern that is almost identical to the adult. However, transcript distribution and quantity does not take into account regulation of GTF2IRD1 at the protein level and the influence of interacting signaling networks and co-factors and this could have a strong influence on its impact as a transcription factor. Behavioural phenotypes were first noted in the adult homozygous knockout animals and, therefore, the first microarray experiments were conducted on brain tissue from adult mice, which resulted in a very small number of potential gene

137 targets. As a means to explore whether developmental stage was an important factor in gene regulation, the second set of microarray experiments were conducted using early post-natal mice. As a result of work shown in this thesis, it is now thought that the action of GTF2IRD1 may be cell context-dependent and, therefore, future microarray work will focus on specific neuronal cell types that express Gtf2irdl in a uniform manner, such as the medium spiny neurons of the striatum. If specific gene targets can be identified, their regulation by GTF2IRD1 will be studied throughout development.

In the next chapter, we demonstrate that GTF2IRD1 interacts with proteins that directly or indirectly associated with chromatin modifying functions. It is therefore possible that

GTF2IRD1 may have little direct gene regulatory function in vivo, apart from auto­ regulation of its own transcript (Palmer et al, 2010).

138 Gene expression analysis of Gtf2ird1 KO and WT mice

(Section 3. Section 3.3)

Tissue approach Cell-specific approach Brain region analyses FACS isolation of CABAergic neurons

RNA isolation

Microarray analysis i Bioinformatics analysis i Confirmation of dysregulated genes

Figure 3.1. Schematic diagram outlining the two different approaches taken to investigate the gene expression profde of the Gtf2irdl KO mouse brain.

Gene expression studies by microarray analysis were carried out using (A) major regions of the post natal mouse brain from homozygous null Gtf2irdl KO mice and WT mice and more sensitively, using (B) GABAergic neurons isolated from the entire brain by fluorescence activated cell sorting (FACS). Microarray data were analysed using a bioinformatics program and genes showing dysregulated expression as determined by the selection criteria were assessed further, mainly by quantitative real-time PCR. Table 3.1 Identity of genes with a minimum 2-fold change in expression in the brain regions of Gtf2irdl KO mice.

Gene Description Cortex Hippo Olf Bulb Cer Gene locus symbol 1500035N22Rik 2.18down 1.39down 1.87down 1.8 down Chr 5: 24,491,660- 24,503,986 9130024FllRik (Large intergenic non-coding RNA) 1.02down 1.87up 2.68down 1.08 down Chr 1

**Actl6b Actin-like 6B 2.66down 2.47 down 2.66down 3.35 down Chr 5: 137,994,745 -138,010,810 *Calnl Calneuron 1 1.03down 1.16down 2.77down 1.38 down Chr 5: 130,845,346 -131,316,515 Cga Glycoprotein hormones, alpha 1.08 down 1.04 up 6.92down 1.02 down Chr 4 subunit Chll Cell adhesion molecule with 1.06up 1.09up 2.53down 1.45 down Chr 6 homology to L1CAM ctgf Connective tissue growth factor 1.07down 2.30up l.llup 1.28 down Chr 10

Ctgf As above l.Olup 2.02up 1.17 up 1.29 down Chr 10

F830002E14Rik 1.29up l.Oldown 4.27 up 1.96 down

Gh Growth hormone 1.12up 1.07up 18.41down 1.03 up Chr 11 Gpr88 G-protein coupled receptor 88 1.07down 3.46up 2.24down 1.40 down Chr 3 Hspb3 Heat shock protein 3 1.16down 2.70up 1.08up 1.33 down Chr 13 Indo Indoleamine-pyrrole 2 1.03down 1.75up 2.45down 1.09 down Chr 8 Itpka Inositol 1,4,5-trisphosphate 3- 1.03down l.Oldown 2.31down 1.41 down Chr 2 kinase A Kcnq5 Transcribed locus, strongly similar 1.69down 2.35down 1.79down 2.12 down Chr 1 to XP_532200.2 PREDICTED: similar to potassium voltage-gated channel, KQT-like subfamily LOC385855 1.03up 1.37 up 2.17down 1.47 down *Mospd3 Motile sperm domain containing 3 1.44up l.Oldown 2.11up 1.30 up Chr 5: 138,037,873 -138,042,286 Prkag2 Protein kinase, AMP-activated, 1.70down 2.34down 1.76down 2.01 down Chr 5: 24,368,562- gamma 2 non-catalytic subunit 24,606,460 Prl Prolactin 1.07 down 1.05down 2.51down 1.13 down Chr 13 Ramp3 Receptor (calcitonin) activity l.Olup 2.31up 1.24down 1.04 down Chr 11 modifying protein 3 Ramp3 As above l.OOup 2.19up 1.36down 1.33 down Chr 11

Rpell Phosphatase and actin regulator 1 l.OSup 1.02down 2.76down 1.17 down Chr 13 S100a8 S100 calcium binding protein A8 1.09up 2.12up 1.40down 1.42 down Chr 3 (calgranulin A) Sla Mus musculus src-like adaptor l.OOup 1.36up 2.16down 1.09 down Chr 15 *Taf6 TAF6 RNA polymerase II, TATA box 1.58down 1.94down 2.03down 2.70 down Chr 5: 138,619,845 binding protein (TBP)-associated -138,628,679 factor Tcrb-V13 T-cell receptor beta, variable 13 1.09down 7.41up 2.62down 1.31 down Chr 6 Tcrb-V8.2 T-cell receptor beta, variable 8.2 1.13down 6.38up 1.83down 1.07 down Chr 6 Trh Thyrotropin releasing hormone 1.02up 2.08up 1.07up 1.17 down Chr 6 Tshb Thyroid stimulating hormone 1.14up 1.06up 4.90down 1.48 down Chr 3 Ttr Transthyretin 1.30down 2.59down 3.03up 4.64 down Chr 18 **Zfp68 protein 68 3.18down 2.74down 2.78down 3.58 down Chr 5: 139,045,844 -139,060,989 *'Zfp68 Zinc finger protein 68 2.52down 2.51down 2.18down 4.99 down as above

Genes corresponding to those shown in the Venn diagram (Figure 3.2) are listed alphabetically with the fold-change values and direction of the change (up- or down-regulation) found in each specific brain region of KO mice relative to their WT littermates; (*) indicate the three transcript probes found to be commonly dysregulated by > 2-fold in all brain regions. In some cases, the same transcript was identified by more than one probe, resulting in multiple entities. Gene loci were retrieved from Ensembl mouse database; (#) indicates cluster of genes located within 5 Mb region around the Gtf2irdl mutant locus on chromosome 5. Olfactory Bulb: Cortex: Fold change «> 2.0 Fold change «> 2.0 21 entities 4 entities

Identifier Gene name

NM_013844 Zinc finger protein 68

NM_031404 Actin-like 6B

NM_013844 Zinc finger protein 68

Hippocampus: Fold change *> 2.0 16 entities

Figure 3.2. Venn diagram of genes with a minimum 2-fold change in expression level in brain regions of Gtf2irdl KO mice.

Gene expression microarray analyses were performed on three different brain regions of Gtf2irdl KO and WT mice at postnatal day 10-12 (n-7'), including cortex, hippocampus and olfactory bulb. Pairwise comparisons of normalised array data from the two genotype groups were performed for each distinct brain region. Data filtering was applied to extrapolate transcript probes (“entities ”) found to be up- or down- regulated by > 2-fold in Gtf2irdl KO mice in each brain region. Identities of all genes contained in this diagram are shown in Table 3.1. Asterisks (*) indicate the three transcript probes found to be commonly dysregulated in all regions of the brain analysed, which corresponds to genes shown in the box . Figure 3.3. Quantitative RT-PCR analyses confirmed the down- regulation of Actl6by but not of Zfp68 in Gtf2irdl KO mouse brain.

Quantitative RT-PCR analyses using primers targeting the 3’ untranslated region of Zfp68 (A-D) and Act/6b (E-H) were performed on cDNA samples synthesised from total RNA samples extracted from the regions of the brain: cortex (A and E; ex), cerebellum (B and F; cer), hippocampus (C and G; hip) and olfactory bulbs (D and H; olfbulb) of wild type (WT), Gtf2irdl heterozygous (het, only obtained in the hippocampal region) and homozygous null KO mice. Each value was normalised to housekeeping genes GapDH or /3-actin. Data were statistically analysed using t-test for two-group comparisons (cx, cer, and olf bulb) or one-way ANOVA test for the three-group comparison (hip). *p-value <0.05; ***p-value <0.0005. A E Zfp68 Act 6b r »♦

—

cx Cx B F Zfp68 Actl6b 50000

=>. 40000 S 6000- us

jb 4000-

S’ 20000- *■■■ 2000 ■ 15 10000-

Cer Cer C G Zfp68 Actl6b 50000

1 40000 -

S 20000- 20000-

Hip Hip D H Zfp 68 ActlSb zoom

jb 20000-

■ i .if- ^ Olf Bulb Of Bulb A B WT Het Null WT Het Null (kb)

1.90-

1.38-

Northern blot Olf bulb

28s

18s

RNA Ethidium

Olf Bulb

Figure 3.4. Down-regulation of Actl6b in Gtf2irdl KO mice was not observed by Western blotting or Northern blotting.

(A) Three regions of the brain were dissected from wild type (WT), Gtf2irdl heterozygous (Het) and homozygous (Null) KO mice: cerebellum (Cer), olfactory bulb (Olf bulb) and cortex. Total protein extracts were analysed by Western blotting using an anti-BAF53b antibody. Equal loading was confirmed by Coomassie staining of the membranes (not shown). (B) Northern blot analysis of total RNA extracted from the cerebellum of wild type (WT), Gtf2irdl heterozygous (Het) and homozygous (Null) KO mice. The mouse Act/6b transcript was detected at the expected size of 1.6 kb. Approximate loading is shown by the gel electrophoresis image of the gel (RNA Ethidium). (C,D) Re-evaluation of Actl6b expression level by RT-qPCR analysis using primers targeting the coding region of the transcript. Figure 3.5. Dysregulated genes physically close or linked to the Gtf2irdl locus in the mutant allele retained 129Sv/J genomic DNA.

(A) Approximate location of differentially expressed genes on mouse chromosome 5G2 which were identified by microarray analysis of the various brain regions of Gtf2irdl KO mice. Diagram is not drawn to scale and details of other existing genes in the intergenic regions have been excluded for simplicity. (B) Sequence alignment of the 3 'portion of the Act 16b transcript amplified from WT and KO cDNA samples showing the presence of SNPs within the 3 UTR array probe. Dots represent identical nucleotide bases. Locations of the primers, Baf53bF and Baf53bR, used in the first qPCR assay are highlighted by the arrows. Illumina array probes highlighted in yellow are found downstream and upstream of the stop codon (boxed). The SNP IDs are shown next to the single base changes between C57BL/6J to 129Sv/J mouse strains (source: irvrvr. informatics, iax. ore). (C) DNA sequencing chromatograms showing region containing the two SNPs in the 3 'UTR of the Actl6b transcript in WT and KO mice. 134.6 138 139 (Mb)

Gtf2ird2 Ncfl Gtf2i Gtf2irdl Actl6b Mospd3 Taf6 Zfp68

Mus musculus Chr 5G2

B Baf53bF 1180 1190 1200 1210 1220 1230 1240 1250 1260 ___ I____ I ------I____ I ------I------I ------I------I ------I____ I ------I------I ------I------I ------I . ... I ------I------I WT1 AGAAGACCCC ACCGAGCATG CGTCTTAAGC TCATCGCCAG CAACAGCACC ATGGAACGCA AGTTCAGCCC CTGGAJ TGGA GGCTCCATCT WT2 ...... r WT3...... KOI...... K02...... K03......

1270 1280 1290 1300 1310 1320 1330 1340 1350 -- I--- I --- I--- I --- t--- I --- I--- I --- I--- I --- I--- I --- I--- I --- I--- I --- TGGCCTCACT GGGCACA'TTC CAGCAGATGT GGATCTCCAA ACAGGAATAC GAGGAGGGAG GGAAGCAGTG CGTGGAGCGG AAGTGCCCCi

Baf53bR 1380 1390 1400 1410 1420 1430 1440 .. I ____ I------I ------1------I _____I------I ------I------I ------I------I ------I------I GTC CACCCCATCC ACGGAGAAAC GCCAGAGGGG CGTTCTACCA GCCAGAAATC CAAAAGCGCT

...... C...... A...... C...... A...... C...... A...... rs3150207fxTcl rs3150208fo7Al 1450 1460 1470 1480 1490 1500 ------I------I ------I------I _____I------I _____I------I ------I------I ------I------I GAACTCTCCC TTTCCCATTA CCCTCTTCTC TCTTACCCAC ATCTTCtTAG ACTGTGATGT

rs3150209[c/Tl

C Wildtype Knockout

r AC CCG0CCAC CCCAT C CAcJgJg AG AA. r ACC C(jc]cCAC CCCAT CCAcQrAGAA. rs3150207 rs3150208 rs3150207 rs3150208 Figure 3.6. Diagram of procedures involved in the FACS array study.

(A) Breeding of double heterozygous Gadltml(EGFp)Tama/ Gtf2irdl,mIHrd mutant mice used to generate the litters of mice used in the FACS array experiment: Gadl-GFP heterozygous (GFP positive) which were either WT or homozygous null for Gtf2irdl. Expected ratio of genotypes resulting from the breeding is shown based on Mandelian genetics, except that the homozygous null Gad 1tm 1 (EGFP>Tama mice died in utero, resulting in a 2:1 proportion of GFP-positive (Gadl-GFP +/-) to GFP- negative (Gadl-GFP -/-) mice. (B) Flow chart of the FACS array procedure from brain dissection and isolation of GFP-expressing GABAergic neurons to the microarray and bioinformatics analyses. 25% 50% 25% ' 1 25% 50% 25%

GtfUrdl Gtf2irdl Gtf2irdl Gtf2irdl Gtf2irdl Gtf2irdl +/+

Gadl -GFP +/- Gadl-GFP -/-

B

Single cell suspension * Technical r ^ optimisation FACS isolation L______A

QC of isolated Total RNA cells extraction Fluorescent microscopy *

RNA QC and processing Performed by microarray facility at AGRF

lllumina array

Bioinformatics analysis I I Confirmation > by qRT-PCR A GAD67-GFP positive

B GAD67-GFP negative

50 100 150 200 250 50 100 150 200 250 FSC-A FSC-A ' , 000>

Figure 3.7. Microscopic and flow cytometry analyses of GAD67-GFP positive and negative cell populations isolated from the mouse brains.

(A,B) Representative images and two-dimensional dot plots of single cell suspensions prepared from brains of GAD67-GFP positive (A) and negative (B) littermates at post natal day P10. Microscopy images showing GFP- overlayed dark and bright field. Dot plots showing forward scatter (FSC) versus propidium iodide (PI) signal (left) and FSC versus GFP signal of cells analysed by flow cytometry (right). GFP-positive cells are located in the Q2-2 area of the latter plot in (A). Figure 3.8. Isolation of GAD67-GFP positive GABAergic neurons by FACS.

(A) Representative FACS plots of single cell suspensions prepared from GAD67-GFP positive mice. Dot plots showing selection of cell population to be sorted based on multiple parameters: PI 'gating based on the FSC versus SSC (side scatter characteristic) plot in (a), which is used to exclude small debris or large aggregated cells, which was then further gated for P2 ' in (b) to exclude the doublets or aggregated cells based on FSC-A (area) versus FSC-H (height) plot, and finally gated for

‘P4' based on GFP versus FSC plot in (d) that selects for GFP expression. The relatively small number of dead cells is indicated by the uptake of propidium iodide (PI) shown on the GFP versus PI plot in (c). The relative proportion of GFP positive (>103 5) to GFP negative cells is shown by the histogram (e). Representative data from the gated populations are shown (f). (B) Representative images of cells taken before and after sorting showing GFP alone (left) and DAP I and GFP overlaid channel (right). GFPFL1-A C 3.4 36.5 36.0 100.0 %Total 9.4 36.5 98.8 %Parent

339 b 3,602 3,646 10,000 ^Events GFPFL1-A P2

HBP4 pi 3

Events

H

All

I Population

I

Post-sorting a

-oss A Figure 3.9. Microarray analysis of RNA isolated from GABAergic neurons.

(A,B) Gene ontology’ analysis performed on genes with changes in expression > 3-fold at p-va/ue <0.05 based on the standard domain of "molecular function" and "biological processes". (C-F) Quantitative RT-PCR analyses of genes known to be expressed in GABAergic neurons, Gria3, S/cl a 1 and Npas4, revealed no significant differences in their expression levels in the cortex of wild type (WT) and Gtf2irdl homozygous (Null) KO mice. A

Molecular function catalytic activity (26.05%)|

transporter activity (7.98%)|

molecular transducer activity (7.56%)j enzyrne regulator activity (3.36%)| transcription regulator activity (Z.10%)| k,- structural molecule activity (1.26%)1 electron carrier activity (Q.84%)| - transIation reguIator activity (0.84%)| au^liary transport protein activity (0.42%)

B

Biological processes establishment of localization (7.42%) ] -multicellular organismal process (7.23%) I developmental process (7.03%)| localization (8:20%}f response to stimulus (3.32%)[ regulation of biological process (9 96%^ biological adhesion (2 54%] positive regulation of biological biological regulation (10.74%)^ process (1.95%) negative regulation of bioiogicai process (176%) .growth (1,17%)j

metabolic process (12.70%)|-

hmhhk - locomotion (0.59%) [ multi-organism process (0.39%)l

-cellularprocess (21 48%)|

Gria3 Slc1a1 Npas4 CHAPTER 4 - IDENTIFICATION OF NOVEL PROTEIN PARTNERS OF GTF2IRD1

Owing to its predicted structure as a transcription factor, most studies on GTF2IRD1 to date have focused on its DNA binding and gene regulatory ability. Yet, a conventional way to study functions of proteins, including transcription factors, is through identifying proteins that interact directly or indirectly with them in the form of macromolecular protein complexes. Evolutionary analysis has focused on the repeat domains (RDs) as the most conserved features shared among the GTF2I family of proteins, although their functions are not clearly understood. In regards to 3D structural data, the RDs were originally predicted to form a helix-loop-helix motif typical of many transcription factors. However, this view was disputed by a study using NMR spectroscopy on a repeat domain which classified the RD as a ‘new* fold (Doi-Katayama et al, 2007). As discussed in Chapter 1, in vitro binding assays showed that the DNA binding capability of GTF2IRD1 seems to be mediated by certain RDs, particularly RD4 and to a lesser extent RD5 and RD2 (Polly et al, 2003; Vullhorst & Buonanno, 2003). Therefore, we predict that some of these RDs, in particular RD1 which has no demonstrated DNA binding capability, may be mediating protein-protein interactions and this could explain why members of the GTF2I protein family have undergone internal duplications to multiply their motif during evolution.

TFII-I, the evolutionarily-related protein of GTF21RD1, has been reported to engage in multiple protein-protein interactions involved in different levels of cellular functions, leading to its title as a ‘multifunctional transcription factor'. TFII-I was initially identified as a basal transcription factor as well as an interacting partner of the helix- loop-helix activator USF in the general transcriptional machinery (Roy et al, 1991).

Independently, TFII-I was cloned and identified as an interacting protein for serum response factor (SRF) and homeodomain protein Phoxl which together, stimulate the transcription of c-fos in vivo (Grueneberg et al, 1997). Many more reports followed 154 these studies, adding players to the current multifaceted protein network of TF1I-1.

Some of these putative interacting proteins include: Myc, a nuclear proto-oncoprotein and transcription factor which regulates cell proliferation and differentiation (Roy et al,

1993a); cGMP-dependent protein kinase 1(3 (PKGIp) which not only phosphorylates

TF11-1 and enhances its transcriptional activity, but is also thought to be recruited and thus, phosphorylates other proteins present in the multiprotein complexes TFII-I participates in (Casteel et al, 2005; Casteel et al, 2002); and a Rho inhibitor pl90RhoGap (GRLF), an important regulator of the capillary network formation which together with TFII-I were shown to regulate the vascular endothelial growth factor receptor VEGFR2 (Jiang et al, 2005; Mammoto et al, 2009). Finally, in a recent study, an unanticipated non-transcriptional and cytoplasmic role for TFII-I was uncovered. It was shown that the phosphorylated form of TFII-I binds to phospholipase C-y (PLC-y) and competes with the binding of TRPC3, a cell surface calcium channel (Caraveo et al,

2006). Despite the numerous reports on TFII-I properties, a unifying molecular mechanism on how TFII-I functions at the cellular level remains elusive.

In contrast to TFII-I, few studies have reported direct interactions of GTF2IRD1 with other protein partners. Currently, only three proteins have been associated with

GTF2IRD1: retinoblastoma (Yan et al, 2000), PIASX (Tussie-Luna et al, 2002b), and

HDAC3 (Tussie-Luna et al, 2002a), and in all, the extent of investigations have not fully elucidated the mechanism of actions or the functional implications of the interactions. In the next chapter, we show that the role of the PIASX interaction with

GTF2ID1 is functionally important for the SUMOylation of GTF2IRD1.

Factors that may have contributed to the slow progress of the biochemical characterisation of GTF2IRD1 include the overall low basal level of protein and/or the lack of antibody that can successfully detect endogenous GTF2IRD1. To circumvent these problems, in this chapter we endeavoured to understand the molecular pathways

GTF2IRD1 functions in by identifying its potential protein partners using an unbiased

155 yeast two-hybrid (Y2H) library screening strategy. This study led to the identification of novel proteins that interact with GTF21RD1 in yeast, some of which have been confirmed when co-expressed in mammalian cells. The domains of GTF2IRD1 that mediate the interaction with each of these proteins have also been systematically mapped. Among these partners, we identified a group of proteins involved in histone modification, suggesting a potential role of GTF2IRD1 in chromatin remodelling.

4.1 Novel protein interactions identified in yeast two-hybrid library screening

To perforin the Y2F1 library screening, we used the universal normalised mouse library

(Matchmaker Mate & Plate system, Clontech) and the full-length GTF2IRD1 as a bait.

The bait, pGBKT7-GTF2IRDI, contains Gal4 DNA-BD (binding domain) fused to the open reading frame (ORF) of GTF21RD1, whilst the library and all Y2H protein preys made for subsequent interaction tests are fused downstream of Gal4 DNA-AD

(activation domain) in pGADT7 plasmids (Appendix B). When the bait and prey proteins interact, the Gal4 DNA-BD and Gal4 DNA-AD are brought to close proximity, enabling the activation of Gal4-responsive reporter genes in AH 109 yeast strains.

Activation of two reporter genes, HIS3, ADE2, allows yeast to grow on selection media that lack the histidine and adenine amino acids. Meanwhile, activation of MELl allows secretion of a-galactosidase (a-gal) resulting in blue yeast colonies when grown in the presence of the chromogenic substrate, X-a-gal, providing an additional means of selection.

Screening of positive interactions was conducted on a high stringency culture medium that lacked Leu, Trp, Ade and His amino acids (quadruple drop out, QDO). Prior to the library screening, we confirmed that the bait did not autonomously self-activate the reporter genes in the presence of non-interacting prey, pGADT7-T, as shown by negative survival on the QDO medium (Figure 2.2.A.B). On the other hand, the reporter genes were properly turned on in the selective media in the presence of a positive 156 control interaction (p53 and T-antigen mating). The mated diploids in both instances were able to survive on the double drop out medium (DDO, SD -Leu, -Trp) demonstrating that both bait and prey plasmids were contained in the diploids and the negative interaction was not due to a toxicity issue caused by GTF2IRD1 expression

(Figure 2.2.C).

Approximately 2 million diploid clones were screened in each library screen on QDO selective media by counting the colonies from DDO plates. Yeast colonies that grew on

QDO medium were subjected to another stringency test by re-spotting onto another

QDO medium with x-a-gal (Figure 2.2.D) prior to isolating the plasmids from the yeast clones. It was noted that all colonies did turn blue although with varying intensities, indicating variability in the strength of interactions. When comparing the strength of interactions between different clones, an additional stringency test was performed by using QDO medium plus 3-amino-1,2,4-triazole (3-AT), a competitive inhibitor to the

His3 protein in yeast (data not shown). Two consecutive screens of the same library gave rise to a total of 52 colonies that passed the stringency tests and were sent for plasmid sequencing (The Ramaciotti Centre, UNSW). The sequencing results were subjected to basic local alignment tool (BLAST, NCB1) searches against the mouse transcript database. Altogether, there were 14 different transcripts with gene coding sequences positioned in-frame to the Gal4 DNA-AD sequence in the prey plasmid

(Table 4.1). Some genes were present in more than one clone, either as identical or non­ identical clones that represent different lengths of a transcript.

To rule out any false positives due to non-specific escape from the selection strategy, the 14 different interacting prey plasmids were back-transformed into the parental

AH 109 yeast strain along with either pGBKT7-GTF2IRD 1 or pGBKT7 as bait plasmid

(Figure 4.1). This assay also determines whether the prey protein specifically interacts with GTF2IRD1 and/or whether it also interacts non-specifically with the Gal4 DNA-

BD domain in the pGBKT7 plasmid. Indeed, this experiment showed that 3 out of 14

157 prey proteins (AUP1, TRIM27, and COPS5) showed survival on QDO plates after transformation with pGBKT7, suggesting their interaction was with the Gal4 DNA-BD

(Figure 4.1). FHAD1 seemed to interact weakly with the pGBKT7 control, but only at background level and the interaction was not reproduced in a subsequent test (Figure

4.2.B, see FHAD1 versus control). We regarded this as a genuine interaction with

GTF2IRD1 in yeast. EXOC3 failed to interact with both pGBKT7-GTF2IRDl and pGBKT7 bait on the QDO medium, suggesting that the yeast containing this plasmid escaped selection by non-specific means.

Interaction of the prey proteins with the Gal4 DNA-BD of pGBKT7 plasmid does not imply that the prey may not have a genuine interaction with GTF2IRD1. Therefore, to test this possibility with TRIM27, we switched the prey and bait combination by subcloning TRIM27 coding sequence into the pGBKT7 plasmid and tested its interaction with GTF2IRD1 in pGADT7 prey plasmid to remove the possibility of Gal4 activation via binding to the Gal4 DNA-BD. This experiment revealed that no interaction was established between pGBKT7-TRIM27 and pGAD T7-GTF2IRD1 (data not shown), confirming that the clone was identified due to interaction of TRIM27 with the Gal4 DNA-BD of the bait plasmid. The status of AUP1 and COPS5 interactions with GTF2IRDI has not been investigated.

4.2 Mapping the interacting domains of GTF2IRD1

We then investigated the regions of GTF2IRD1 that interact with each of the confirmed interacting prey. In particular, we wondered if the protein interactions are mediated by the RDs or other evolutionarily conserved regions in GTF2IRD1, such as the C-terminal end and regions that contain SUMOylation motifs (studied in Chapter 5). To answer this question, we cloned various peptide domains of equal length into pGBKT7: the leucine zipper at the N-terminus (LZ), repeat domain 1 to 5 (RD1, RD2, RD, RD4, and RD5), the C-terminal end (C-term), and regions containing the first and second conserved

SUMO motifs (Sumol and Sumo2) (Figure 4.2.A.; Appendix A). In designing these 158 domain-specific baits, we took into account the importance of the evolutionary conserved residues and the consistent length of each fragment (88 amino acids). Prior to performing the interaction mapping, each bait was tested for autoactivation by testing for possible interaction with the empty pGADT7 prey plasmid. All baits were negative for autoactivation, except the C-terminal region (data not shown). Therefore, interactions that occur through the C-terminus region, which includes the nuclear localisation signal (NLS) and the serine stretch cannot be unambiguously determined.

To address this issue, we generated GTF21RD1 C-terminal serial truncation constructs in pGBKT7 as indicated in Fig. 4.2.A.

Y2H analysis of GTF21RD1 region-specific interactions with the isolated prey proteins, including ARMCX5, TAF1B, HSF2BP, SETD6, DCAF6, FHAD1, ZMYM5 and

OPHN1 are shown in Figure 4.2.B. As discussed above, any interactions with the C- terminus region needs to be investigated using the C-terminal deletion constructs

(Figure 4.2.C). Proteins with markedly reduced binding or failure to bind to the first and all subsequent truncations of GTF21RD1 suggest that their interactions are likely to be mediated by the last 27 residues of the C-terminal region, for example DCAF6. On the other hand, interactions whose intensities were not affected by any of the C-terminal truncations indicated that their interactions were mostly likely mediated by the N- terminal region of GTF21RD1, such as the RDs.

The qualitative interpretation of the mapping experiment is summarised in Table 4.2.

The analysis was semi-quantitative as all diploid yeast colonies plated on the QDO medium had a similar number of cells, ensuring a similar growth of the yeast colonies on DDO medium (not shown). Furthermore, the conformational complexity of the protein in vivo cannot be examined by this assay as binding with each single domain is examined at a time, thus the outcome of a combinatorial interaction with multiple domains could not be predicted. Nevertheless, these data provided insight into the protein-protein interaction capacity of individual regions of GTF2IRD1.

159 Overall, all prey proteins interacted with most regions of GTF2IRD1, except none were able to interact with RD5, the serine stretch or the leucine zipper region. The latter finding was expected as the leucine zipper is known to mediate dimerisation of the same or similar molecules (Landschulz et al, 1988). Dimerisation of GTF2IRD1, mediated by the leucine zipper, has been shown to be functionally important for a greater and more flexible binding to multiples DNA binding sites (Palmer et al, 2010). Interestingly, of all the RDs, RD1 which has demonstrated no DNA binding ability scored the most number of interactions with other proteins. And surprisingly, the Sumol region, which contains the conserved regulatory SUMO motif in GTF2IRD1, was a site of multiple protein interactions. This suggests that SUMOylation may play an important role in regulating the protein-protein interactions of GTF2IRD1 (see Chapter 5).

4.3 Proteins involved in histone modification

Three of the proteins identified in the screen have an association with histone modification - SETD6, ZMYM5 and DCAF6. As a result, our interest has focussed on a potential role for GTF1RD1 in chromatin remodelling.

Post-translational modifications scattered along the N-terminal tails of histones is a central mechanism underlying chromatin and transcriptional regulation (see Chapter 1 - section 1.1.7.2). Described as the ‘histone code’, the various histone modifications can act individually or in combination to direct specific function, such as gene activation or repression (Jenuwein & Allis, 2001). The most well-studied histone modifications are lysine acetylation and lysine methylation. The enzymes that dynamically catalyse the addition and removal of the acetyl- or the methyl- group are histone acetyltransferases and deacetylases, or histone methyltransferases and demethylases, respectively. In the section below some background literature and the preliminary work on SETD6 and

ZMYM5 are discussed.

160 4.3.1 SETD6

SETD6 (SET-domain containing protein 6) belongs to the Suppressor of variegation

[Su(var)], Enhancer of zeste [E(z)] and Trithorax (SET)-domain methyltransferase protein superfamily (Dillon et al, 2005). Originally identified in Drosophila proteins, the evolutionarily conserved SET domain has been shown to catalyse attachment of methyl groups to specific lysine residues on histone tails (Jenuwein et al, 1998).

SUV39H1 was the first mammalian histone lysine methyltransferase (HKMT) identified to act specifically on H3K9 (lysine 9 of histone H3) (Rea et al, 2000). To date, all except one (Min et al, 2003) of the histone lysine residues known to be methylated are modified by SET domain-containing proteins. Each lysine residue may be either mono-, di-, or tri-methylated. adding to the highly complex, but precise control by methylation which is the most specific histone modification mechanism identified to date.

The only functional characterisation of SETD6 has been published in a non-peer- reviewed book (Richter, 2009). In this study, SETD6 was shown to be a histone lysine mono-methyltransferase with the ability to modify H1K26, H2AK5, H3K14, H4K5,

H4K12 residues, an unusually broad substrate specificity for an HKMT enzyme.

Furthermore, all of these specific lysine residues have not been shown previously to be methylated, limiting any predictions that can be made about the function of the histone lysine methylation. Interestingly, each of these lysine residues has been shown to be a target for acetylation, all of which would be mediated by the histone acetyltransferase

TIP60 (Kimura & Horikoshi, 1998). This led the authors to hypothesise that the two enzymes SETD6 and TIP60 may modify histones in an antagonistic manner by competing for the same lysine residues (Richter, 2009).

To further characterise the interaction between SETD6 and GTF21RD1, as well as determine its functional importance, we cloned the full-length ORF of mouse SETD6

(protein ID #Q9CWY3, 473 amino acids in length, 53 kDa in molecular mass) into a myc-tagged mammalian expression vector. Firstly, we examined the subcellular

161 localisation of transiently transfected myc-SETD6 and GFP-GTF2IRD1 in COS-7 and

3T3 cells by immunocytochemistry. The immunofluorescence analysis shows the presence of myc-SETD6 in the nucleus where it co-localises to a large extent with GFP-

GTF21RD1, both in COS-7 (Figure 4.3.B) and 3T3 cells (data not shown).

Subsequently, following transfection of the same expression plasmids in COS-7 cells, we subjected cell lysates to a co-immunoprecipitation assay to confirm the interaction of SETD6 and GTF2IRD1 in mammalian cells. Figure 4.3.C shows an immunoblot confirming the interaction by first using an antibody specific to GFP to immunoprecipitate GFP-GTF21RD1-interacting protein complexes, followed by an antibody to myc that detected myc-SETD6. The same band was not detected in the control lanes when a GFP construct was used instead or when myc-SETD6 was absent in cell lysates, suggesting that the interaction between GFP-GTF2IRD1 and myc-

SETD6 was not a result of non-specific interactions that may have occurred with the

GFP-tag or any endogenous proteins, respectively. Therefore, our experiment confirmed that over-expressed SETD6 localises to the nucleus in a pattern that overlaps with over­ expressed GTF2IRD1 and these proteins physically interact in mammalian cells.

4.3.2 ZMYM5

ZMYM5 or ZNF237 belongs to the MYM-type family of zinc finger proteins, five of which have been identified in the human genome (Figure 4.4.A, Smedley et ai 1999;

Sohal et al, 2000). Each protein contains tandem repeats of unique zinc fingers [Cys-Xi-

Cys-Xi9_22-Cys-X3-Cys-Xi3_i9-Cys-X2-Cys-Xi9_25-Cys-X3-Cys] named ‘MYM" for myeloproliferative and mental retardation. The repeats have been shown to function in protein-protein interactions (Gocke & Yu, 2008). ZMYM2 (or ZNF198) is disrupted in the chromosomal translocation t(8; 13)(p 11 ;q 12) which leads to human myeloproliferative disorder; whereas, ZMYM3 (or ZNF261) is a candidate gene for X- linked mental retardation on Xq 13.1 (van der Maarel et al, 1996; Xiao et al, 1998).

Intriguingly, both of these proteins have been identified in a complex with TFI1-I and a

162 histone deacetylase, HDAC-2, in HeLa nuclear extract (Hakimi et al, 2003). The MYM proteins also share homology with the protein encoded by Woe in Drosophila which was shown to function in chromatin remodelling (Raffa et al, 2005).

Comparing the ZMYM family members by sequence homology, ZMYM5 is related to

ZMYM2 and both genes are located adjacent to each other on human chromosome 13, suggesting that they have arisen by duplication and divergence (Sohal et al, 2000). The most significant homology shared by the two proteins is in their MYM domains,

ZMYM5 contains four MYM-domains and ZMYM2 has ten, and in their N-terminal region. The human ZMYM5 gene has 5 isoforms resulting from alternative splicing

(Sohal et al, 2000). The two different clones isolated from our mouse cDNA library screen were of different lengths, but overlapped with each other and corresponded to the longest isoform in human (Figure 4.4.A).

No experimental characterisation of the ZMYM5 protein has been reported. In order to analyse its relationship with GTF2IRD1 in mammalian cells, we cloned the full-length mouse sequence of ZMYM5 (protein ID #Q3U2E2-1,627 amino acids, molecular mass of 70 kDa) into an epitope-tagged mammalian expression plasmid. Due to the presence of the same restriction sites that are available in the cloning site of the myc-tagged plasmid and in the open reading frame of Zmym5, the coding sequence was cloned into a GFP-tagged plasmid. Immunocytochemical experiments have shown that ZMYM2 is present in nuclei in association with chromatin (Gocke & Yu, 2008; Ollendorff et ah

1999), prompting us to determine if ZMYM5 has a similar cellular distribution.

Immunocytochemical analysis of COS-7 cells transfected with GFP-ZMYM5 revealed the presence of ZMYM5 in punctate structures within the nucleus. Co-transfection with myc-GTF21RDl in the cells showed that GTF2IRD1 co-localises in small nuclear dots with some nucleoplasmic localisation observed (Figure 4.4.B). Both proteins seem to be excluded from the nucleolus.

163 We then performed co-immunoprecipitation assays to confirm the interaction of

ZMYM5 and GTF2IRD1 in mammalian cells. First, an antibody to GFP was used to

immunoprecipitate GFP-ZMYM5 protein complexes, an antibody to myc confirmed the

presence of myc-GTF2IRDl in the immunoprecipitated band (Figure 4.4.C). No protein

was detected when GFP-ZMYM5 was not transfected or in the control GFP

transfection, excluding the possibility of a non-specific interaction. Thus, this

experiment demonstrated that ZMYM5 and GTF2IRD1 could interact in the nuclei of

mammalian cells.

4.3.3 Preliminary analyses of global histone H3K14ml and H4K12mel methylation

patterns in brain regions of Gtjlirdl KO mice

Since GTF21RD1 is associated with various histone modifying proteins, we

hypothesised that specific histone modifications may be dysregulated in the Gtf2irdl

knockout mouse, thus leading to a number of physical and behavioural phenotypic changes (Howard et al., submitted). The impact of histone modification in an organism can take place on a global level or at specific loci in the genome. There is no direct evidence of how the reduction in or loss of GTF2IRD1 may affect histone modifications

in the mouse. However, considering Gtf2irdl knockout mice are viable and show relatively subtle cognitive and behavioural phenotypic abnormalities, it is likely that if histone modification(s) are affected, they do not cause widespread gene dysregulation.

Instead, the impact may be restricted to specific gene sets, which in the brain result in the neurocognitive phenotypes of the mice. This would be consistent with our results from gene expression profiling of the knockout mouse brain (Chapter 3). A challenge that remains is the identification of GTF2IRD1 target genes. The only gene for which there is strong evidence of in vivo regulation is Gtflirdl itself (Palmer et al, 2010).

Bioinformatic analysis revealed that approximately 100 genes contain putative

GTF2IRD1 DNA binding motifs conserved in the mouse and human genome; however, these genes do not appear to be dysregulated in our array analyses (Palmer et al, 2010).

164 A method used to detect global changes in specific histone modifications is to analyse total cell lysates or purified histone extracts by Western blotting using antibodies that recognise the specific modifications. On the basis of the interaction of GTF2R1D1 with

SETD6, we first looked for changes in the monomethylation status of Lys-14 of histone

3 and Lys-12 of histone 4 using anti H3K14mel and H4K12mel antibodies (Richter,

2009). To compare the levels of these modifications in the brains of Gtf2irdl knockout and wildtype mice, histones in whole brain lysates (from cerebellum and olfactory bulbs, regions with high Gtflirdl expression) were enriched by acid extraction

(described in section 2.3.4). Analysis of histone extracts on a 15% SDS-PAGE gel showed an enrichment and separation of histone proteins with the expected banding pattern (Figure 4.5.A). The extracts were then subjected to Western blotting analysis using antibodies against H3K14mel and Ft4K12mel (Figure 4.5.B) and re-probed using a pan histone H3 antibody for normalisation. Densitometric analyses of the blots revealed that there were no significant differences in the global level of both histone modifications, H3K14mel and H4K12mel, in brain regions analysed from both groups of mice, although there was a trend toward a reduction of H4K12mel level in the olfactory bulbs of Gtf2irdl knockout mice (Figure 4.5.C). These data suggest that if the hypothesis involving GTF2IRD1 in chromatin modification is correct, the histone modification(s) affected in Gtf2irdl knockout mice may not be occurring on a global level, but on specific loci in the genome.

4.4 Proteins involved in other pathways

4.4.1 ARMCX5

Armadillo (ARM)-repeat proteins make up a iarge family of proteins and have been implicated in a wide range of cellular processes, including signalling, transport, and cytoskeletal regulation (reviewed in Tewari et al, 2010). Each ARM repeat is made up of a conserved 42 amino acid motif which folds into three a-helices. Tandem repeats of multiple ARM units in turn fold together to form a superhelical structure with a broad

165 binding capacity for many protein partners, which is reflected in the versatile functions of the ARM-repeat-containing proteins. Although ARM-repeat proteins may not share high amino acid sequence identity, they do have remarkably conserved three- dimensional folds (Tewari et al, 2010).

First identified in the Drosophila melanogaster, the mutation of the segment polarity gene Arm was shown to cause the 'armadillo'-like appearance of the embryo

(Riggleman et al, 1989). Its homologue in mammals was then identified as P-catenin, which is both an adhesion protein and a key signalling protein in the Wnt/p-catenin pathway (reviewed in Conacci-Sorrell et al, 2002). ARMCX5 is one among six proteins of the ARM-repeat family (ARMCX1-6) encoded by genes located on the X chromosome, which are poorly characterised. ARMCX proteins have fewer numbers of

ARM repeats than most of the classical ARM family members. For example, p-catenin contains twelve ARM repeats and most of the other members contain a minimum of six repeats (Hatzfeld, 1999). In contrast, ARMCX5 and ARMCX 1 contain three and two

ARM repeats, respectively, while only a single ARM repeat is present in ARMCX2 and

ARMCX3 (Kurochkin et al, 2001). Sequence analysis suggests that ARMCX 1-3 and

ARMCX6, but not ARMCX5, contain a transmembrane domain at their N-termini, a hydrophobic cluster that potentially targets these proteins to the plasma membrane

(Kurochkin et al, 2001; Mou et al, 2009). ARMCX3 has been shown to be present in the mitochondrial outer membrane (Mou et al, 2009).

In order to investigate the specificity of GTF2IRD1 binding to ARMCX5 and not to the

ARMCX family members in general, we cloned the full-length ORF of ARMCX5, as well as the closely related ARMCX 1 and ARMCX 3 into pGADT7 prey plasmid and then tested their interaction with GTF2IRD1 in a Y2H assay. We confirmed a positive interaction between GTF2IRD1 and ARMCX5 in this assay, but GTF2IRD1 failed to interact with ARMCX 1 or ARMCX 3 which suggests that either the interaction with

ARMCX5 does not occur via the ARM repeats or GTF2IRD1 can only interact with the

166 ARM repeats of ARMCX5 (Figure 4.6.A). Furthermore, we mapped the interaction domains of ARMCX5 on GTF2IRD1 and found that ARMCX5 can bind to the first three RDs within GTF2IRD1 (RD1-3, Figure, 4.2.B). It is interesting to note that

ARMCX5 contains three ARM repeats, which may potentially form a one to one interaction with each of GTF2IRD1 RDs, RD1-3.

We investigated the subcellular localisation of myc-ARMCX5 in comparison to the known mitochondrial-associated localisation pattern of ARMCX3.

Immunocytochemical analysis of 3T3 cells transfected with myc-ARMCX3 shows a cytoplasmic punctate staining pattern similar to that published by Mou et al. (2009)

(Figure 4.6.C). In contrast, our immunocytochemical analysis revealed a predominantly uniform cytoplasmic staining of myc-ARMCX5, not associated with any cytoplasmic structures. The staining was excluded from the nucleus where GTF2IRD1 normally resides (Figure 4.6.B). Co-expression of myc-ARMCX5 and GFP-GTF2IRD1 did not affect GTF2IRD1 localisation (Figure 4.6.B). Therefore, at this stage, we are uncertain how the interaction between the two proteins can take place when localised in such different cellular compartments. The null hypothesis is that the interaction does not occur naturally and the interaction is an artifact of the Y2H assay. Another possibility is that certain stimuli may trigger translocation of ARMCX5 into the nucleus where functional interaction with GTF2IRD1 occurs. Recently, ARMCX1 was shown to be transcriptionally regulated by the Wnt/p-catenin signalling pathway (Iseki et al, 2010).

The final alternative is that some GTF2IRD1 is also retained in the cytoplasm by

ARMCX5.

4.4.2 KPNA2 (karyophrenin or importin-a2)

In eukaryotic cells, nuclear proteins are actively transported into the nucleus from the cytoplasm where nascent proteins are being synthesised by the ribosome. The transport through the nuclear pore complex is bidirectional and the shuttling of proteins in and out of the nucleus is mediated largely by the karyopherin family of transport receptors,

167 also known as (for nuclear import) or exportins (for nuclear export). These proteins are able to associate with components of the nuclear pore complex and bind to cargo molecules that need to be translocated across the nuclear membrane (reviewed in

Conti & Izaurralde, 2001). There are at least six importin-a and twenty two importin-(3 proteins identified so far in the human proteome (Chook & Blobel, 2001). The classical nuclear import mechanism involves the recognition of a canonical nuclear localisation signal (NLS) by importin-a, which then heterodimerises with importin-p, which docks the cargo bound complex onto the nucleoporins of the nuclear pore complex (Gorlich et al, 1996; Moroianu et al, 1995). Once in the nucleus, importin-P binding to RanGTP causes the dissociation of importin-a and the release of the cargo.

Importin-a consists of an N-termina! importin-p binding (IBB) domain followed by a tandem stretch of ten armadillo repeats that bind the canonical NLS (Marfori et al,

2010). Phylogenetic analysis has suggested that importin-a or importin-like proteins are the oldest common ancestors for ARM-repeat proteins (Tewari et al, 2010). Canonical

NLSs are characterised by a cluster of several positively charged residues or two of such motifs (bipartite NLS) separated by a linker of 10-12 amino acid residues

(Dingwall & Laskey, 1991). GTF2IRD1 contains an NLS characterised by a highly conserved stretch of basic amino acids at its C-terminal end (KRKRKRV, Appendix A), functionally important for GTF2IRD1 nuclear localisation. Deletion of the NLS causes

GTF2IRD1 mislocalisation into the cytoplasm (Yan et al, 2000). Thus, an obvious functional implication for K.PNA2 or importin-a2 as an interacting partner is to mediate the nuclear localisation of GTF2IRD1 through interaction with its NLS. Indeed, our interaction mapping using the Y2H assay supports the hypothesis that KPNA2 binds to the NLS of GTF2IRD1 (Figure 4.2.C). Firstly, KPNA2 failed to interact with any specific domains of GTF2IRD1, except the C-terminal region which contains the NLS

(Figure 4.2.B). This result was further confirmed by serial deletion of GTF2IRD1 C- terminal tails. The interaction of KPNA2 with GTF2IRD1 was lost when the NLS (and

RD5) was deleted from the C-terminus in the Tr3 construct (Figure 4.2.C). Therefore,

168 our finding strongly suggests that KPNA2, as a member of the importins, mediates the active nuclear transportation of GTF2IRD1. A future experiment to test whether

KPNA2 is the sole mediator in this function would be to knockdown KPNA2 using small interfering RNAs in cells that express GTF2IRD1 and analyse the impact on

GTF2IRD1 localisation.

4.5 Discussion

In this chapter, we report the identification and confirmation of novel protein partners of

GTF2IRD1 using the Y2H assay. We have also mapped the domain(s) of GTF21RD1 that interact with each of the confirmed prey proteins. Surprisingly, we found that virtually all domains within GTF2IRD1 are able to serve as protein-protein interaction domains. Our results challenge the current dogma of the evolutionary-conserved RDs of

GTF2IRD1 being solely DNA-binding domains. It also led us to speculate that

GTF2IRD1 may serve other physiological functions beyond its known role as a transcriptional regulator of its own promoter.

One unifying theme that emerged from our Y2H screens is the interaction of

GTF2IRD1 with a group of proteins that are involved in chromatin remodelling, specifically, those associated with histone modification. These proteins include SETD6,

ZMYM5 and DCAF6.

The identification of SETD6 as GTF2IRD1-interacting partner is interesting. The lysine methyltransferase catalytic domain, SET domain, was first recognised as the conserved sequence in Drosophila Trithorax and Polycomb proteins. Trithorax and Polycomb function antagonistically to each other via chromatin regulation mechanism to tightly regulate the expression of multiple Hox genes, the developmental regulators of anterior- posterior segmentation along the body axis (Hanson et al, 1999). Mutations in the

Trithorax and Polycomb genes in fruit flies and in mice resulted in altered expression of the Hox genes spatially and/or temporally and increased the frequency of abnormal 169 axial patterning, consistent with the phenotypes of Hox gene mutant animals (Grimaud et al, 2006; Soshnikova & Duboule, 2009). During our initial characterisation of the

Gtf2irdl knockout mouse line, originally on a 129/SvJ background, a subtle posterior transformation was observed (Tay, 2005). Specifically, Gtf2irdl +/- and -/- (129/SvJ) mice had a significantly higher incidence of lumbar vertebra loss (sixth lumbar) accompanied by the gain of an extra sacral vertebra (first sacral). This phenotype is consistent with a potential relationship between GTF2IRD1 function and Hox gene regulation. Of interest, this phenotype was not present when the mutant allele was backcrossed onto the C57BI6 background; all mice, regardless of their genotype, showed a normal patterning of the vertebrae (data not shown). We determined that wildtype 129/SvJ mice are prone to the specific axial transformation phenotype described above, although at a significantly lower frequency. Therefore, the presence of the phenotype appears to require a natural predisposition to posterior transformations and a reduction in or absence of Gtf2irdl enhances the penetrance of the phenotype.

The identification of ZMYM5 as a novel interacting protein of GTF2IRD1 was not totally surprising due to the previously reported association of ZMYM2/3 with TFII-I in

HDAC-containing repressor complexes (Hakimi et al, 2003). ZMYM proteins were shown to act as adaptor proteins mediating protein-protein interactions through its

MYM-type zinc fingers, thus, maintaining the integrity of the HDAC complex (Gocke

& Yu, 2008). Independently, Wen et al. (2000; 2003) identified TFII-I within an endogenously purified HDAC-3 immunocomplex and characterised this interaction.

Furthermore, exogenously expressed GTF2IRD1 was shown to interact with HDAC-3 via a co-immunoprecipitation assay ((Tussie-Luna et al, 2002a). Therefore, we suggest that the association of TFII-I and GTF2IRD1 with HDACs may be mediated by

ZMYM2/3 and ZMYM5, respectively.

By cloning the full-length ORF for SETD6 and ZMYM5 into epitope-tagged plasmids, we have confirmed the predicted nuclear localisation of these proteins in cultured cells.

170 It should be noted that these experiments relied on the expression of recombinant tagged proteins in cultured cells and, therefore, may be subject to experimental artifacts. The presence of the tag peptide may influence the behaviour of the GTF2IRD1 protein and slight differences in the observed sub-nuclear localisation of GTF21RD1 could be due to the technical constraints of observing GFP directly versus detection of the Myc-tag via indirect immunofluorescence, with its inherent antibody penetration problems. The ability of each of these proteins to interact with GTF21RD1 has also been confirmed by co-immunoprecipitation assays. The next crucial step of this study is to characterise the functional importance of these interactions.

Although further work has not been carried out in terms of DCAF6 (DDBI and CUL4- associated factor 6) interaction with GTF21RD1, if true, the involvement of DCAF proteins in the regulation of the ‘histone code' may fit our current hypothesis. DCAF proteins, which in common contain WD40 domain, is known as the substrate binding module for the CUL4-DDB1 ubiquitin E3 ligase machinery (Angers et al, 2006; He et al, 2006; Higa et al, 2006; Jin et al, 2006; reviewed in Lee & Zhou, 2007). By recruiting specific substrates, CUL4-DDB1 complex regulates various processes, including cell-cycle progression, DNA replication and DNA damage response (Petroski

& Deshaies, 2005). Proteomics and structural studies revealed that the highly conserved double ‘DxR’ box motifs in WD40 repeat proteins could act as the interacting docking sites on DDBI (Angers et al, 2006; Higa et al, 2006). Although the functions of DCAFs are still poorly characterised, several WD40-repeat proteins have been associated with histone methylation, including WDR5 (Han et al, 2006; Wysocka et al, 2005) and

RBBP5 (Dou et al, 2005; Ruthenburg et al, 2006), both of which are associated with

H3K4 methylation through direct or indirect binding with the enzyme that mediates the histone modification, MLL (Odho et al, 2010). The CUL4-DDB1 complex has also been associated with methylated histone H3, including H3K4, K9 and K27 methylation

(Higa et al, 2006). Since the regulatory effect of WDR5 and RBBP5 was specific to

171 H3K4 inethylation alone, it was suggested that other WD-repeat proteins that make up

CUL4-DDB1 could be associated with other histone modifications (Higa et al, 2006).

So what are the possible functional implications of GTF21RD1 interaction with proteins involved in the chromatin remodelling pathway? There are two possible main hypotheses, which are not mutually exclusive. Firstly, interactions with the histone modifying proteins may underpin the mechanism by which GTF2IRD1 regulates its target genes in vivo, which thus far has only included its own gene. The GUR, as a genomic target of GTF2IRD1, presents a good model to investigate the involvement of histone modifications underlying the negative autoregulation of GTF2IRD1 transcription. The association of specific genomic regions with proteins, including histones with specific modifications can be analysed by chromatin immunoprecipitation

(ChIP) assay using specific antibodies against such modified histones. The enrichment of fragments bound to the immunoprecipitated complex can then be quantified by qPCR analysis (kChIP-qPCR'). However, an important question in this line of investigation is which specific histone modifications are involved? Currently, scant information exists about the specific proteins that we isolated in our screen; there is little functional data on their histone modifying activity. ZMYM5's association with histone acetylation is assumed on the basis of its homology with the closely related protein ZMYM2. On the other hand, the activity of SETD6 as a histone lysine methyltransferase has been shown to target a number of different histone lysine residues. Ideally, to identify those that may be associated with the GUR, a more comprehensive set of residue-specific anti­ methylated or acetylated histone antibodies are required to perforin ChlP-qPCR assays.

It remains possible that through interaction with the complex of histone-modifying proteins, GTF2IRD1 may be recruited and regulates key genomic sites that are yet to be identified. In the previous chapter, the potential gene regulatory function of GTF2IRD1 was investigated by microarray-based analyses of the knockout mice. As discussed, the impact of GTF2IRD1 loss on the overall gene expression in the major regions of the

172 brains in which it is expressed appeared to be subtle, although the possibility of region- and/or timing-specific gene regulation by GTF2IRD1 remains to be investigated.

Further characterisation of ZMYM5, SETD6 and DCAF6, would also greatly aid our prediction on how histone regulation may underpin GTF21RD1 function. Although we currently predict that the main impact of GTF21RD1 function is in transcriptional regulation, other possible functions impacted by chromatin regulations include DNA repair, replication, and chromatin condensation (reviewed in Kouzarides, 2007). Only if we understand the system through which GTF2IRD1 functions, can we make more specific predictions concerning the expected outcomes of GTF2IRD1 loss.

The second hypothesis is that GTF21RD1 may be a substrate protein by interacting with those proteins which have (such as SETD6) or are part of the histone modifying complexes (such as ZMYM5). Histone modifying proteins have been increasingly reported to target non-histone proteins, especially histone lysine methyltransferases, and thus, now widely termed as protein lysine methyltransferases (Huang & Berger, 2008).

Very recently, SETD6 was shown to mediate lysine methylation of the NF-kB subunit

RelA at Lys-310, modification which rendered RelA inert and attenuated RelA-driven transcriptional programs (Levy et al, 2011). It is then possible that the purpose for

SETD6 interaction is to target GTF2IRD1 for protein lysine methylation. This possibility can be investigated through an in vitro lysine-methylation assay or mass- spectrometry (Fingerman et al, 2008). DCAF6, member of the DCAFs family that has been identified as substrate binding modules for the CUL4-DDB1 ubiquitin ligase complex, may also interact and direct GTF2IRD1 for ubiquitylation. Although both scenarios from the two hypotheses are equally likely and/or may co-exist, it seems unlikely that the sole purpose of the novel interactions identified herein is to target and to regulate GTF2IRD1 for various post-translational modifications, as this would again leaves us with the big question of what is the function of GTF2IRD1?

173 A potential separate pathway through which ARMCX5 functions and physically interacts with GTF2IRD1 also warrants further investigations. Our immunocytochemical analysis revealed that ARMCX5 transfected in cells is localised in the cytoplasm. Interestingly, the Human Protein Atlas, which shows antibody detection of ARMCX5 endogenous protein, revealed that it is present as speckles in the cytoplasm as well as in the nucleus of three different cell lines (www.proteinatlas.org,

Berglund et al, 2008). Thus, ARMCX5 could potentially be shuttled in and out of the nucleus, depending on the specific signalling in the cells. Future experiments need to investigate the effect of different signalling modulators on the cellular localisation of

ARMCX5, for example, by using an activator of the Wnt/p-catenin pathway.

Finally, the remaining group of proteins that we identified in our screen have not been explored, hence, not discussed in detail in this chapter (Table 4.1). There is limited knowledge about these proteins, either in terms of functions or subcellular localisation, which we can construct hypothesis on how they may be functionally related to

GTF21RDI. The time constraint of this study restricted our initial characterisations to those proteins for which a clear testable hypothesis was available. However, the identification of additional interacting protein partners presents an opportunity to enlarge our testable predictions concerning the molecular function of GTF2IRD1, and consequently, members of the GTF2I protein family which are evolutionarily, and almost certainly, functionally related.

174 Table 4.1. Identities of library clones recovered in yeast two-liybrid screen of the mouse universal library using the full-length GTF2IRD1 as a bait.

Accession mRNA title Gene GTF2IRD1 binding number symbol specificity confirmed?

NM_001009575.3 Armadillo repeat containing, X-linked 5 Armcx5 Yes

NM 007517.3 Ancient ubiquitous protein 1 Aupl Not specific, binds to Gal4 DNA-BD

NM_013715.1 C0P9 (constitutive photomorphogenic) Cops5; Not specific, binds to homolog, subunit 5 (Arabidopsis CSN5; Jabl; Gal4 DNA-BD thaliana) Mov34; Sgn5 NM_026493.3 Centrosome and spindle pole Csppl Yes associated protein 1

NM_028759.1 DDB1 and CUL4 associated factor 6 Dcaf6 Yes

NM 177333.3 Exocyst complex component 3 Exoc3 Not specific, binds to Gal4 DNA-BD

NMJL77868.3 Forkhead-associated (FHA) Fhadl Yes phosphopeptide binding domain

XM 904809.3 PREDICTED: heat shock transcription Hsf2bp Yes factor 2 binding protein

NM 052976.3 Oligophrenin 1 Ophnl Yes

XM 001481215.1 PREDICTED: similar to nuclear pore­ Kpna2 Yes targeting complex component of 58 kDa, transcript variant 2 (LOC100043906) NM 001035123.3 SET domain containing 6 Setd6 Yes

NM_020614.1 TATA box-binding protein (TBP)- Taflb Yes associated factor, RNA polymerase l,B

NM 009054.2 Tripartite motif-containing 27 Trim27 Not specific, binds to Gal4 DNA-BD

NM_144842.3 Zinc finger, MVM-type 5 Zmym5 Yes

NCBI Basic Local Alignment Search tool (BLAST) was used to search the identity of each sequence against the mouse Re/Seq mRNA database. Only plasmid clones that contain an

open reading frame fused in-frame to the Gal4 DNA-AD sequence are shown, alphabetically sorted. The results of back-transformation of each protein prey with GTF2IRD1 or empty bait plasmid in yeast (Figure 4.1) are indicated in the last column. pGBKT7 control pGBKT7- GTF2IRD1

A*\o& &

pGBKT7 control pGBKT7- GTF2IRD1

pGBKT7 control pGBKT7- GTF2IRD1

Figure 4.1. Confirmation of GTF2IRD1 interacting proteins by re­ transformation with the rescued prey plasmid.

The binding specificity of each prey protein to GTF2IRD1 was analysed by back transforming the isolated yeast clones as preys (indicated above) with the pGBKT7 control or the pGBKT7-GTF2IDl as bait plasmid into the AH 109 yeast strain. Double transformants grown on DDO medium (not shown) were re-spotted on QDO+x-a-gal medium (shown above). Positive interactions are represented by those that grew on the latter medium and additionally displayed blue colour. Only those that interacted with pGBKT7-GTF2IRD1 and not pGBKT7 were of interest as those that were positive in both indicate that the interaction is mediated by the Gal4 DNA-BD of the bait plasmid and not GTF2IRD1. Figure 4.2. The mapping of interaction domains of GTF2IRD1 with the various proteins identified in the Y2H screen.

(A) Schematic diagram of human GTF2IRD1 which consists of a leucine zipper (LZ) at the N-terminus, five repeat domains (RDs), two regions containing SUMO motifs, a nuclear localisation signal (NLS), a stretch of serine residues, and a conserved region at the C-ter minus. Lines drawn above correspond to the various GTF2IRD1 regions cloned into pGBKT7 plasmid used as baits in mapping the interaction domains with other proteins. A detailed amino acid sequence and alignment of GTF2IRD1 orthologs is provided in Appendix A. Arrows below the diagram point to the approximate positions of truncations in the C-terminal truncated baits (Trl- 4). (B) The ability of each prey protein (indicated on the left) to interact with various bait plasmids encoding the full-length (FL) or specific regions of GTF2IRD1 (indicated above) were shown by the survival of transformants on QDO+x-a-gal medium. (C) The same assay was performed using C-terminal truncations of GTF2IRD1 as the baits. Human GTF2IRD1 (944 AA)

RDl Sumo2 RD2 Sumol RD3 RD4 RD5 C-term

RD1 RD2 RD3 RD4 RD5 t t / \ Tr4 Tr3 Tr2 Trl

Leucine zipper Serine repeat % C-terminus I (LZ) ^RD)631 d0ma'n |‘SUMO’motif |NLS J I

A (V Prey

ARMCX5 Table 4.2. Summary of GTF2IRD1 interaction domain mapping.

.GTF2IRD1 Serine LZ RD1 RD2 RD3 RD4 RD5 Sumol Sumo2 NLS C-terminus Interactor repeat

ARMCX5

TAF1B

HSF2BP

SETD6

KPNA2

DCAF6

OPHN1

FHAD1

ZMYM5

RD1 RD2 RD3 RD4 RD5 JU I Leucine zipper (LZ) Repeat domain (RDi j SUMO'motif | NLS | Serine repeat |c-terminus

Data summarised from Figure 4.2. B-C shows the regions within GTF2IRD1 that interact with each interacting prey protein in the Y2H assay. Interactions (X) with specific regions were either observed directly or indirectly, in regards to the C-terminal domain, the serine repeat and NLS by the negative inference of results obtained with the C-terminal truncated baits. Asteriskf*) marks those observed with noticeably weaker interaction. Figure 4.3. Nuclear co-localisation and interaction of SETD6 and GTF2IRD1 were confirmed in mammalian cells.

(A) Schematic diagram of mouse SETD6 protein which contains the catalytic methyltransferase SET domain and a RuBisCO large subunit methyltransferase (LSMT) domain which is thought to mediate the binding to its substrate. The diagram is drawn approximately to scale, based on the currently available information for the protein in the UniProtKB database (EMBL-EBI, Cambridge, UK). The line drawn above represents the portion of SETD6 present in the Y2H clone identified in the screen. (B) Myc-SETD6 and GFP-GTF2IRD1 were detected in the nuclei of COS-7 cells. Immunofluorescence staining was performed on methanol-fixed cells, mvc- SETD6 was detected with an anti-myc and an Alexa-555 fluorescent secondary antibody, GFP-GTF2IRD1 was visualised by its GFP fluorescence and the cellular nucleic acid was stained with DAP1. Yellow indicates an overlap in the location of the two proteins. (C) Interaction of myc-SETD6 and GFP-GTF2IRD1 was confirmed by co-immunoprecipitation of protein complexes in cell lysates of COS-7 cells transfected with the indicated plasmids. Lysates were subjected to immunoprecipitation with a GFP antibody and immunoblotted for myc and GFP. Asterisk (*) denotes IgG heavy chains. A Y2H clones BMP* LSMT-SBD l C 473 aa * SET domain Rubisco LSMT substrate-binding domain

B

Merge MYC-SETD6 GFP-GTF2IRD1 DAPI

C GFP + GFP-GTF2IRD1 - + + MYC-SETD6 + +

Blot: MYC

Blot: GFP

Blot: MYC 60 Figure 4.4. Nuclear co-localisation and interaction of ZMYM5 and GTF2IRD1 were confirmed in mammalian cells.

(A) Schematic diagram depicting ZMYM5 (also known as ZNF237) in comparison to the other members of the MYM-type zinc finger proteins family (adapted from Gocke and Yu, 2008). Lines drawn above ZMYM5 represent the approximate sequence coverage of the clones isolated from the Y2H screen. Scale bar represents approx. 100 aa. (B) GFP-ZMYM5 and myc-GTF2IRDl were detected in nuclear puncta in COS-7 cells. Immunofluorescence staining was performed as described in previous Figure 4.3. Yellow indicates co-localisation of the two proteins. (C) Interaction of GFP-ZMYM5 and myc-GTF2IRDl were confirmed by co- immunoprecipitation of protein complexes in cell lysates of COS-7 cells transfected with the indicated plasmids. Cell extracts were subjected to immunoprecipitation with a GFP antibody and immunoblotted for myc and GFP. Asterisk (*) denotes IgG chains. ZMYM2 (ZNF198) N a 111 ■ i ■ a i « Y2H clones

ZMYM5 (ZNF237) N

ZMYM3 (ZNF261) N

ZNF262 N » HI llll II I

ZNF258

100 aa Proline/valine-rich |mym Cre-like domain domain

Merge MYC-GTF2IRD1 GFP-ZMYM5 DAPI

GFP - - + GFP-ZMYM5 - + - MYC-GTF2IRD1 + + +

Blot: MYC

IP: GFP Blot: GFP

Blot: MYC Figure 4.5. Analysis of histones extracted from Gtf2irdl KO vs wild type brain regions reveal no significant differences in global H3K14 and FI4K12 methylation patterns.

(A) The acid extraction of histone proteins from mouse brain lysates resulted in an enrichment of histones (HI, H3, H2A, H2B and H4) from the whole­ cell lysates (WL) and cytoplasmic (C) fractions of the tissue. (B) Histone extracts from the olfactory’ bulbs (Olf Bulb) and cerebella (Cer) of wildtype (WT) and Gtf2irdl knockout (KO) mice were subjected to Western blot analyses using antibodies against H3K14 and H4K12 monomethylations, and histone H3 as a control. (C) The relative amount of each histone methylation was quantified by densitometric analysis of Western blots (WT n=3, KO n-2). Unpaired t-test was performed on each category, none was found to be statistically significant, although a trend toward a reduction of H4K12me level was seen in the olfactory bulbs of KO mice. A kDa WL WL C H H

— H1

/ H3 <- H2A ^ H2B — H4

B Olf Bulb Cer WT KO WT KO H3K14me1

Histone H3

H4K12me1

Histone H3

C

H3K14me1 -Olf Bulb H3K14me1 -Cer

WT KO

H4K12me1 -Olf Bulb H4K12me1 -Cer

WT KO WT KO A Bait: GTF2IRD1

Prey:

QDO + x-a-gal

DDO

Merge MYC-ARMCX5 GFP-GTF2IRD1

Figure 4.6. Y2H analysis suggests that GTF2IRD1 and ARMCX5 can interact however the two proteins do not appear to co-localise in cells.

(A) The full-length cDNA sequences of Armcx5, Armcxl, and Armcx3 were cloned into the pGADT7 prey plasmid and each was tested for the ability to interact with GTF2IRD1 by Y2H assay. A positive interaction between ARMCX5 and GTF2IRD1 was shown on QDO+x-a-gal medium, while viability of all transformants were shown by their survival on DDO medium. (B) Immunocytochemical analysis shows a predominantly cytoplasmic localisation of mvc-tagged ARMCX5 in contrast to the nuclear localisation of GFP-tagged GTF2IRD1 in 3T3 cells. (C) Unlike ARMCX5, myc-tagged ARMCX3 is specifically targeted to the mitochondrial outer membrane, consistent with a previous report (Mou et al, 2009). CHAPTER 5 - GTF2IRD1 IS SUMOYLATED IN VIVO

Post-translational modifications (PTMs) of proteins allow rapid and dynamic regulation of their cellular activities and functions. This is reflected in studies of numerous proteins targeted by phosphorylation, the best known PTM, as well as acetylation and ubiquitylation, all of which participate in most cellular activities. SUMOylation was discovered as a covalent linkage of an 11 kDa SUMO (small ubiquitin-like modifier) protein to the GTPase activating protein RanGAPl in mammalian cells (Mahajan el al,

1997; Matunis el al, 1996). SUMOylation of RanGAPl changes its localisation from the cytoplasm to the nuclear pore via interaction with the nucleoporin RanBP2. Initial characterisations of the SUMO protein also came from studies that identified SUMO as a binding partner for human RAD51 and RAD52 (Shen et al, 1996), FAS (Okura el al,

1996) and PML (Boddy et al, 1996) in yeast two-hybrid assays. SUMOylation has been predominantly regarded as a modification that affects nuclear proteins; although, there are now an increasing number of studies reporting SUMOylation of cytoplasmic and membrane proteins (reviewed in Martin el al, 2007; Watts, 2004; Wilkinson et al,

2010).

SUMO proteins share less than 20% amino acid sequence identity with ubiquitin; however, their three-dimensional folded structures are very similar (Bayer et al, 1998).

SUMO proteins are highly conserved and ubiquitously expressed in all eukaryotes.

Simple eukaryotes, such as yeast and invertebrates, have a single SUMO gene while complex eukaryotes, such as plants and vertebrates, have several SUMO genes (Johnson et al, 1997; Jones et al, 2001; Kamitani et al, 1998; Tanaka et al, 1999). The loss of the

SUMO gene Smt3 in the budding yeast Saccharomyces cerevisiae leads to a loss of cell viability (Meluh & Koshland. 1995); whereas, deletion of Pmt3, the SUMO gene in

Schizosaccharomycespomke, leads to severe growth impairment (Tanaka et al, 1999).

187 The human genome contains four SUMO genes (Su & Li, 2002): SUMO-1, SUMO-2,

SUMO-3, and the recently identified SUMO-4. SUMO-1 shares approximately 50% similarity to SUMO-2 and SUMO-3. The conjugated forms of SUMO-2 and SUMO-3 are almost identical, only differing by three amino acid residues at the N-terminus and are yet to be functionally differentiated. SUMO-2/3 are capable of forming chains of variable lengths on substrate proteins through an internal acceptor lysine (Tatham et al,

2001). SUMO-1 does not have an equivalent lysine residue and thus, is not likely to act as a link in elongating chains in vivo. However, SUMO-1 can act as the terminator for the elongating chains of SUMO-2/3. The conjugation capacity of SUMO-4 remains in question as the maturation process of SUMO-4 is inhibited by a unique proline residue

(Pro-90) at the position required for interaction and cleavage by SUMO protease (see below) (Owerbach et ah 2005).

Analogous to the ubiquitin-1 ike mechanism, SUMO conjugation and deconjugation involves a cascade of enzymatic processes (Figure 5.1.A). The first step is the proteolytic cleavage of an inactive SUMO protein by a family of SENPs

(sentrin/SUMO-specific proteases), which expose its C-terminal di-glycine residues, a prerequisite for its conjugation to the lysine residues in target proteins. Next, the mature

SUMO protein is activated by the ATP-dependent El activating enzyme, a heterodimer of SAE1 (SUMO-activating enzyme El) and SAE2 in mammals (Gong et ah 1999).

This step results in the formation of a thioester bond between the active cysteine of

SAE2 and the C-terminal glycine residue of SUMO. SUMO is then transferred to the active site cysteine of the E2 conjugating enzyme UBC9, forming another thioester linkage (Rodriguez et al, 2001; Sampson et al, 2001). UBC9 is the only known SUMO

E2 conjugating enzyme, in contrast to the multiple E2s that participate in the ubiquitin pathway. UBC9 has a strong overall positive charge and can recognise and bind directly to the consensus SUMOylation motif TKXD/E, where T (psi) indicates a large hydrophobic residue, which may be either I, L, or V; and X is any amino acid (Bernier-

Villamor et al, 2002). Finally, UBC9 transfers SUMO to the substrate where an

188 isopeptide bond is formed between the C-terminal glycine residue of SUMO and lysine side chain of the target. This process is typically facilitated by E3 ligases; although, this is not essential. SUMO conjugation is reversible and it can be cleaved from target proteins by SENPs, the enzymes which are also responsible for the cleaveage of SUMO precursors, thus recycling the SUMO moiety (Hay, 2007).

In contrast to the ubiquitylation pathway where the activities of E3 ligases are essential for the modification of specific substrates, SUMOylation of proteins can be achieved with just SUMO El and E2 enzymes alone, at least in vitro. Multiple proteins with

SUMO E3 ligase activity have been reported and the first family identified, the Siz proteins, were isolated in yeast (Johnson & Gupta, 2001) and their mammalian relatives, the PIAS proteins (protein inhibitor of activated STAT), were identified later (Johnson

& Gupta, 2001; Kotaja et al, 2002; Schmidt & Muller, 2002; Takahashi et al, 2001).

These proteins contain a SP-RING (Siz/PIAS RING) motif, similar to the RING domain in ubiquitin E3 ligases. SUMO E3 ligases act as scaffolds by binding both UBC9 and the target proteins, as well as binding the UBC9-conjugated SUMO moiety non- covalently via a SUMO interacting motif (SIM).

SIM is characterised by stretches of three or four hydrophobic residues, isoleucine, leucine, or valine residues plus one acidic/polar residue at position 2 or 3 [(V/I/L)-X-

(V/I/L)-(V/l/L)] or [(V/I/L)-(V/I/L)-X-(V/I/L)]. The hydrophobic core forms a P-strand that can fit in either orientation into a hydrophobic groove between an a-helix and a P- strand in SUMO (Hecker et al, 2006; Song et al, 2004). A method often employed to screen for non-covalent SUMO interacting proteins is to utilise a construct that encodes a SUMO polypeptide with mutated C-terminal glycine residues and thus, cannot be conjugated to the lysine side chain of substrate proteins (Hecker et al, 2006;

Makhnevych et al, 2009). To date, many SIM-containing proteins have been identified and their direct interactions with SUMO and SUMO-conjugated proteins contributes to

189 the versatile functional impact of SUMO in biological pathways (reviewed in Garcia-

Dominguez & Reyes, 2009; Kerscher, 2007).

In general, the impact of SUMOylation on a given protein target cannot be predicted, but it has been shown to affect the stability, subcellular localisation or transcriptional regulation activity of a protein. At the molecular level, SUMOylation exert its functions through: (1) masking of existing binding sites for another protein, (2) creation of a new binding surface that can recruit other binding partners either by direct non-covalent interaction with the SUMO moiety or via a novel interaction domain created at the

SUMO-substrate interface, and (3) conformational change in the SUMOylated substrate which may directly regulate its function (Geiss-Friedlander & Melchior, 2007;

Wilkinson & Henley, 2010).

We investigated the possibility of GTF21RD1 as a novel SUMO substrate based on several lines of evidence. Firstly, PIASXp, a member of the SUMO E3 ligase family of proteins, was identified as an interacting partner of GTF2IRD1 and TFI1-1 in a yeast two-hybrid screen (Tussie-Luna et al, 2002b). Secondly, two members of the GTF2I protein family have been identified in broad proteomic screens for SUMO substrates:

TFII-I was identified in a screen of proteins derived from HEK293 cells (Zhao et al,

2004) and GTF2IRD2 was identified in C8161 melanoma cells (Ganesan et al, 2007).

Finally, in silico sequence analysis revealed the presence of a highly conserved SUMO consensus motif and a second less well-conserved site in the GTF2IRD1 protein sequence (Figure 5.3).

In this chapter we report that GTF2IRD1 is SUMOylated in mammalian cells based on several lines of evidence from biochemical analyses. Furthermore, we show that

SUMOylation of GTF2IRD1 affects neither its subcellular localisation nor its transcriptional repression activity; we propose that it may play an important role in

190 regulating its protein-protein interactions, such as enhancing the interaction with a chromatin-associated and SIM-containing protein, ZMYM5.

5.1 Interaction of GTF2IRD1 with SUMO E2 and E3 ligases

To initiate our investigation, we tested for potential interactions between GTF2IRD1 and UBC9, the E2 SUMO conjugating enzyme, and GTF2IRD1 and PIASX, a SUMO

E3 ligase, using Y2H assays. The Y2H experiment verified the interaction between

GTF2IRD1 and PIASX as reported previously (Figure 5.1 .B, Tussie-Luna et al, 2002b).

In addition, we demonstrated a positive interaction between GTF2IRD1 and UBC9

(Figure 5.1.B). As negative controls, no interactions were observed in transformed yeast with GTF2IRD1, PIASX or UBC9, and empty bait or prey plasmids, showing that the interactions of GTF2IRD1 with PIASX and UBC9 are specific. All transformed yeast survived on DDO medium confirming successful transformation with each pair of plasmids and the viability of the transformants prior to restreaking on the QDO-x-a-gal medium (Figure 5.1 .C).

In silico analysis of GTF2IRD1 revealed the presence of two SUMO consensus motifs

(TKXE) (Figure 5.2). Alignment of orthologous GTF2IRDI primary protein sequence from human, mouse, frog and fugu fish showed a perfect conservation of the first

SUMO motif (494IKIE497, ‘sumof) throughout evolution. Meanwhile, the second

SUMO motif ( LKQE , ‘sumo2') is less well-conserved as the first hydrophobic residue preceding the lysine has been substituted with an amphiphatic glycine residue in fugu. If functional, this site could be a more recently acquired SUMO regulatory motif.

To map regions of GTF2IRD1 that are responsible for binding with PIASX and UBC9, a series of bait plasmids that contain various GTF2IRD1 subdomains were screened against both proteins using the Y2H system (Figure 5.3.A). GTF2IRD1 interaction with

PIASX appeared to be mediated by the C-terminal region and not with any other subdomains tested (Figure 5.3.B). However, it is worth noting that in the initial auto- 191 activation assessment of all subdomain plasmids, the C-terminal region bait could interact with the Gal4-DNA BD in the pGBKT7 plasmid, thus, making this result ambiguous (Chapter 4). Therefore, a series of GTF2IRD1 C-terminal truncation constructs were generated to test the binding site of PIASX (Figure 5.3.A: Trl, Tr2,

Tr3). Y2H assay with these bait plasmids showed that the GTF2IRD1 interaction with

PIASX was completely abolished with all of these truncated baits, but not with full- length GTF2IRD1 (Figure 5.3.C). All yeast colonies grew on the DDO medium confirming the presence of both bait and prey plasmids as well as their viability prior to selection. This result not only verified the interaction of PIASX with the C-terminal tail of GTF2IRD1, but also mapped the interaction to the last twenty-seven amino acids.

The two SUMO motif-containing regions, sumol and sumo2, interacted strongly with

UBC9 (Figure 5.3.D) which was expected since UBC9 is capable of directly recognising and binding to the canonical SUMO consensus motif (Bernier-Villamor et al, 2002). UBC9 also appeared to interact with the C-terminal region of GTF2IRD1

(Figure 5.3.D), but positive interactions with all GTF2IRD1 C-terminal truncation plasmids suggests that the former result was merely due to the auto-activation capability of C-terminal bait plasmid (Figure 5.3.E). Evidence of a weak interaction between

UBC9 and the leucine zipper of GTF2IRD1 was also detected, which was unexpected as there are no SUMO-like motifs in this region.

5.2 GTF2IRD1 is SUMOylated in vivo

To investigate whether GTF2IRD1 can be SUMOylated in mammalian cells, transient transfections of full-length GTF2IRD1 and HA-tagged SUMO-1 (HA-SUMO) with or without UBC9 were performed in COS-7 cells. Cells were lysed under denaturing conditions in SDS sample buffer and analysed by Western blotting. It is generally recognised that the WBS field lacks good antibodies that can easily detect endogenous

GTF2IRD1, perhaps due to low affinity and the very low abundance of endogenous protein in cells. However, we have shown recently that the anti-GTF2IRDl antibody, 192 M-19 (Santa Cruz Biotechnology), can reliably detect over-expressed protein in cultured cells and on Western blots of protein extracts, as well as endogenous

GTF2IRD1 in immunoprecipitated protein extracts (Palmer et al, 2010). Using this antibody, over-expressed human GTF21RD1 protein was detected at the expected molecular weight of approximately 110 kDa (Figure 5.4.A). Co-expression of

GTF2IRD1 and HA-SUMO yielded an additional higher molecular weight band of approximately 150 kDa, which is typical of the shift caused by the slower migration of

SUMO-conjugated proteins on a SDS-PAGE gel. Moreover, multiple higher molecular weight GTF2IRD1 bands became apparent when UBC9 was also added (Figure 5.4.A).

None of these immunoreactive higher molecular weight bands were present in cells transfected with GTF2IRD1 alone (lane 1). To confirm these bands as HA-SUMO- conjugated proteins, the membrane was stripped and re-probed with a specific antibody against the HA-epitope. The immunoblot showed detection of several high molecular weight bands; two of the most prominent ones aligned perfectly with the novel bands detected by the GTF21RD1 antibody (Figure 5.4.A, right panel). The increase in the amount of SUMO-conjugated proteins detected in cells transfected with UBC9 is consistent with its activity as a SUMO E2 conjugating enzyme. The ‘laddering' effect of SUMOylation on the gel may result from either mono SUMO-conjugations at multiple lysine sites or conjugation of polychains of SUMO proteins onto specific lysine targets. In a later section, GTF2IRD1 is found to be predominantly SUMOylated on one site, hence, favouring the latter hypothesis. Although SUMO-1 is thought not capable of forming polychains in vivo, formation of SUMO chains for SUMO-1 has been observed in vitro (Pichler et al, 2002; Yang et al, 2006). By running the cell lysates on a higher percentage gel, the 11 kDa unconjugated HA-SUMO monomers were detected (Figure 5.4.A, bottom panel).

We extended our biochemical analysis by investigating whether GTF2IRD1 can be

SUMOylated by the endogenous cellular machinery. COS-7 cells are known to contain low amounts of SUMO-1 protein (Ayaydin & Dasso, 2004) and hence, HEK293 cells

193 which contain a large pool of SUMO-1 protein were used for subsequent experiments

(Su & Li, 2002). To increase the possibility of detecting SUMOylated GTF2IRD1, an inhibitor of cysteine peptidases, N-ethylmaleimide (NEM) was used in the cell lysis buffer to prevent SUMO cleavage by SENPs. HEK293 cells overexpressing GTF2IRD1 were lysed in the presence or absence of NEM and immunoprecipitated with anti-

GTF2IRD1. Western blotting analysis using the anti-GTF21RDl antibody showed a higher molecular weight band of approximately 150 kDa (Figure 5.4.B). This high molecular weight shift is NEM-dependent, consistent with the notion that SUMOylated proteins may be cleaved by NEM-sensitive SENPs during cell lysis. The identity of this higher molecular weight band as SUMOylated GTF2IRD1 was further confirmed by immunoblotting with a specific antibody against SUMO-1, which also revealed two additional bands similar to the polychain detected in the previous assay (Figure 5.4.B).

Together, these data indicate that GTF2IRD1 can be poly-SUMOylated in the presence of endogenous SUMO proteins. Moreover, co-expression of the GFP-tagged PIASX further enhanced the level of GTF21RD1 SUMOylation as shown by immunoblots with anti-GTF2IRDl and anti-SUMO-1 antibodies (Figure 5.4.B). This data confirms the functional role of PIASX as a SUMO E3 ligase for GTF2IRD1 in the SUMOylation pathway. Collectively, these experiments show that GTF21RD1 is a novel target of protein SUMOylation in vivo with specificity likely defined by the SUMO E3 ligase

PIASX.

5.3 GTF2IRD1 is SUMOylated at Lys-495 contained within the evolutionarily well-conserved SUMO motif

The lysine residues within the two conserved consensus motifs in GTF2IRD1 may be the sites of SUMO-conjugation in mammalian cells (Figure 5.3). To map the functional site(s) for GTF21RD1 SUMOylation, mutations of Lys-495 and Lys-271 into arginine residues (K495R/K271R or 2KR) were introduced by site-directed mutagenesis. These mutants lack the 8-amino functional group of lysine residues for the covalent attachment of SUMO moiety, but retain the overall positive charge of the amino acid. 194 First, we analysed the effect of the double mutant GTF2IRD1 2KR in COS-7 cells. Co­ expression of wildtype GTF2IRD1 and HA-SUMO resulted in higher molecular weight

SUMOylated GTF2IRD1 bands, shown by immunoblots of cell lysates or immunoprecipitated GTF21RD1 with specific antibodies against GTF2IRD1 and the

HA-epitope (Figure 5.5.A). A reciprocal immunoprecipitation experiment using anti-

HA antibodies confirmed this finding. Conversely, co-expression of the GTF2IRD1

2KR mutant with HA-SUMO completely abolished SUMOylation of GTF21RD1 in

COS-7 (Figure 5.5.A). A similar result was obtained with HEK293 cells. The endogenous SUMO-1 failed to conjugate efficiently to the GTF2IRD1 2KR mutant in comparison to the wildtype protein. (Figure 5.5.B).

To determine if both lysine residues K495 and K271 are sites for GTF2IRD1

SUMOylation, constructs containing individual mutations were generated. Western blot analysis of COS-7 cells transfected with either GTF2IRD1 WT, K495R, K271 R or 2KR mutants with HA-SUMO revealed that conjugation of SUMO was detected on

GTF21RD1 WT and the K.271R mutant, but was very weak on the K495R and 2KR mutants (Figure 5.6). This suggests that SUMO-conjugation of GTF2IRD1 predominantly occurs at Lys-495 residue which is consistent with the fact that it sits within the mostly highly conserved SUMO consensus motif (Figure 5.3, sumol).

5.4 SUMOylation and subcellular localisation of GTF2IRD1

Having established that GTF21RD1 is SUMOylated at Lys-495 in mammalian cells, we next investigated the functional significance of SUMO-modification of GTF2IRD1 by making use of the GTF2IRD1 2KR mutant as a molecular tool. Each of the possible outcomes known to be affected by protein SUMOylation was explored. Firstly, many proteins acquire a SUMO-dependent alteration in their subcellular localisation.

Plasmids encoding GFP tagged GTF2IRD1 or the 2KR mutant together with HA-

SUMO were transfected into NIH3T3 cells, a cell type typically used in subcellular localisation studies. Immunofluorescence analysis showed nuclear localisation of the 195 GFP-GTF2IRD1 2KR, which was indistinguishable from GFP-GTF2IRD1 WT (Figure

5.7). Co-expression of HA-SUMO in the same cells with GTF2IRD1 or the 2KR mutant revealed no observable differences in their subnuclear localisation pattern (Figure 5.7).

Similar results were obtained when the same experiments were performed in COS-7 and

HeLa cells (not shown). These data suggest that SUMOylation does not affect the subcellular localisation of GTF2IRD1 in mammalian cells.

5.5 SUMOylation and GTF2IRD1 protein stability

In some cases the lysine residues of SUMO substrate proteins can also be sites for ubiquitylation, which regulates the stability of proteins through the proteasomal degradation pathway. Although the SUMO and ubiquitin systems were initially found to be competitive, the two systems can also act in synergy to regulate substrate proteins

(reviewed in Geoffroy & Hay, 2009; Ulrich, 2005). To investigate the potential role of

SUMOylation in regulating the protein stability of GTF21RD1, the normal degradation rate of GTF2IRD1 was examined. COS-7 cells transfected with a GTF21RD1 expression plasmid were treated with cycloheximide, a protein synthesis inhibitor, and the stability of existing proteins were measured over time by Western blot analysis. As a control for the assay, the degradation rate of Ets-1 as a short-lived protein was analysed

(Ji et ai 2006). The analysis showed that following cycloheximide addition, GTF2IRD1 underwent a relatively slow degradation rate in COS-7 cells compared to Ets-1 (Figure

5.8.A and B, DMSO treatment). Protein degradation via the 26S proteosome pathway can be inhibited by MG 132 and its addition resulted in an increased stability of the Ets-

1 protein, but had little effect on GTF2IRD1 protein level (Figure 5.8.A and B, MG 132 treatment). Next, the degradation rates of GTF2IRD1 and the 2KR mutant, which cannot be SUMOylated, were compared. If SUMOylation plays a role in controlling the degradation rate of GTF2IRD1, the profile of the mutant will be different to that of the wildtype protein. A trend in decreased protein stability was detected in the 2KR mutant when compared to the wildtype GTF21RD1; however, the differences were not statistically significant at any time points (Figure 5.8.C). Overall, this result suggests 196 that GTF2IRD1 turnover is slow, at least in COS-7 cell line and that SUMOylation of

GTF21RD1 does not play a significant role in regulating its protein stability.

5.6 Transcriptional activity of GTF2IRD1 is independent of

SUMOylation

Transcription factors can be regulated by SUMOylation. For example, Sp3 functions as a repressor in its SUMO-modified form; whereas, SUMO-deficient Sp3 is a strong transcriptional activator (Sapetschnig et al, 2002). The impact of SUMOylation on the transcriptional regulatory activity of GTF21RD1 was analysed by utilising the previously established luciferase transactivation assay in which the GTF2IRD1 upstream region (GUR) is fused to a luciferase reporter (pGL3-GUR) and can be strongly repressed by the binding of GTF2IRD1 protein (Palmer et al, 2010). Consistent with this study, co-transfection of pGL3-GUR and GTF2IRD1 expression plasmids in

COS-7 cells led to an approximately 97% reduction in luciferase activity compared to cells transfected with pGL3-GUR plasmid alone (Figure 5.9). Co-transfection of pGL-

GUR and F1A-SUMO led to an approximately 50% reduction in luciferase activity which is consistent with the reported transcriptional inhibitory effect of SUMOylation

(Gill, 2005). The repressive activity of GTF2IRDI on the pGL3-GUR luciferase construct was not altered by the addition of HA-SUMO or by the disruption of the lysine acceptor sites in the GTF21RD1 2KR mutant (Figure 5.9). There are two potential problems with this assay. Firstly, only a minor portion of the exogenous GTF21RD1 is

SUMOylated and secondly, the exogenous SUMO proteins have the capacity to conjugate with any substrates in the cell. We attempted to address this by expressing a constitutive SUMOylated form of GTF2IRD1 generated by fusing the SUMO ORF in frame at the N-terminus of the GTF21RD1 ORF (Ouyang et al, 2009). The effect of the

SUMO-GTF2IRD1 fusion protein on GUR activity was comparable to that of

GTF2IRD1 alone (Figure 5.9). These experiments indicate that GTF2IRD1 transcriptional activity as assayed by its direct regulation of the GUR, is independent of its regulation by SUMOylation. 197 5.7 GTF2IRD1 interaction with ZMYM5 is modulated by

SUMOylation

Like other PTMs, SUMO-conjugation of protein substrates has the potential to alter the shape and charge of protein surfaces and thereby modify their interactions with other proteins. In the previous chapter, we reported a set of novel protein partners of

GTF2IRD1 and mapped the sites of interactions of these proteins to specific regions of

GTF2IRD1 (Chapter 4, Table 4.2). To our surprise, many of these proteins, including

TAF1B, HSF2BP, FHAD1 and ZMYM5, interacted with the critical SUMO motif- containing region of GTF2IRD1 (sumol). Of particular interest is that the interaction of

ZMYM5 with GTF2IRD1 mapped exclusively to the sumol region. Although F1SF2BP has a similar binding profile to ZMYM5, we are yet to confirm its interaction with

GTF2IRD1 in mammalian cells. It is theoretically possible that these proteins might be interacting with the regions that flank the SUMO site. However, there are no recognisable motifs in this region of GTF2IRD1 other than the SUMO consensus motif.

Hence, it is likely that the protein partners bind to the sumol region either because

SUMO-modified GTF2IRD1 provides a site of interaction for these proteins, or because these proteins interact with the SUMO moiety of SUMOylated GTF2IRD1 non- covalently through a SIM-SUMO interaction. In both cases, an assumption is that the

SUMO motif in sumol could be recognised by the endogenous SUMOylation machinery in the Y2H system and thus, conjugated by the yeast SUMO protein (Smt3).

Although we have yet to show that GTF2IRD1 is SUMOylated by Smt3 in yeast, this is possible due to high conservation of the SUMO machinery between Sacchciromyces cerevisiae and mammalian cells (Johnson, 2004). The lethality of SmtS disruption in S. cerevisiae can be rescued by the expression of human SUMO-1 protein, underscoring the functional overlap of the two orthologs (Takahashi et al, 1999).

To explore the effect of GTF2IRD1 SUMOylation on its binding to ZMYM5, we performed a co-immunoprecipitation assay and analysed the strength of interaction 198 between the two proteins while varying the SUMOylation status of GTF2IRD1. Firstly, it was found that the GTF2IRD1 2KR mutant interaction with ZMYM5 is comparable to that of GTF2IRD1 alone indicating that SUMOylation of GTF2IRD1 is not essential for the interaction (Figure 5.10.A). However, when both SUMOylated and non-

SUMOylated forms of GTF21RD1 are present in transfected cells, an increased level of

SUMO-GTF21RD1 was observed in the ZMYM5 immunoprecipitant (Figure 5.10.A).

Quantitative analysis showed that the significantly higher percentage of the

SUMOylated GTF2IRD1 co-immunoprecipitated with GFP-ZMYM5 in comparison to that present in the total cell lysate (Figure 5.10.B). Hence, these data support the hypothesis that SUMOylation of GTF2IRD1 increases the affinity of ZMYM5 binding.

We did not observe SUMOylation ofZMYM5 modification in Western blot analysis or co-immunoprecipitation assays (Figure 5.10.A, IP: GFP panel), consistent with the lack of a strong SUMO consensus motif in its protein sequence. However, ZMYM5 and its related protein ZMYM2 contain a SIM that has been shown to interact non-covalently with both SUMO-1 and SUMO-2 (Hecker et ai 2006). Therefore, we hypothesise that the SIM region of ZMYM5 preferentially interacts with SUMOylated GTF2IRD1, although it is not an essential requirement as ZMYM5 is also able to interact with the non-SUMOylated GTF2IRD1 and the GTFI2RD1 2KR mutant.

5.8 Discussion

It is generally difficult to identify SUMO substrates in vivo due to the low stoichiometry of SUMO modification and the highly reversible nature of SUMOylation by the SENPs, which often leads to less than 1% of the total protein being SUMOylated at any given time (Johnson, 2004). This is consistent with our observation where the proportion of

SUMOl-modified GTF2IRD1 in HEK.293 cells is very low relative to the unmodified protein and the modification is undetectable in the absence of NEM, a peptidase inhibitor (Figure 5.4.B and 5.5.B). Nevertheless, disruptions in the SUMO regulation can lead to a profound biological impact and increasing links are made between SUMO 199 and human diseases (Bawa-Khalfe & Yeh, 2010; Wang & Schwartz, 2010; Wilkinson et al, 2010; Yan et al, 2010). In this chapter, we have convincingly established the biochemical characterisation of GTF21RD1 as a novel target for protein SUMOylation through its highly conserved SUMO motif.

Furthermore, our data addresses the functional relevance of PIASX as an E3 SUMO ligase for GTF2IRD1. The binding region for PIASX was mapped to a cluster of highly conserved amino acids at the C-terminal end of GTF2IRD1 using Y2H analysis. It will be necessary to determine if deletion of this region prevents GTF2IRDUs interaction with PIASX in mammalian cells and in doing so, inhibits E3 ligase activity for

GTF2IRD1 SUMOylation. This seems likely as an E3 ligase activity involves the binding to both the substrate protein and the SUMO-loaded E2 ligase thus bringing the two molecules closer together for a more efficient SUMO-conjugation onto the target protein. However, there is also a possibility that PIASX may have a different role in the regulation of GTF2IRD1. PIAS proteins can act as transcriptional regulators and have a role in the nuclear architecture by binding to the nuclear matrix via its N-terminal SAP

(scaffold attachment factor-A/B/acinus/PIAS) domain (Palvimo, 2007; Rytinki et al,

2009; Sharrocks, 2006). A report by Tussie-Luna et al. (2002b) suggests that the interaction of PIASX with GTF2IRD1 and TFII-I is mediated by the N-terminal region of PIASX, which contains the SAP domain, as well as a ‘PINIT' domain which seems to be important in subcellular localisation of PIAS proteins (Palvimo, 2007; Rytinki et al, 2009; Sharrocks, 2006). The authors then suggested that PIASX acts as a cofactor for TFII-I transcriptional activity, such as for the activation of c-fos, but the precise mechanism by which this occurs has not been investigated.

Our data suggests that this modification plays an important role in fine-tuning

GTF2IRD1 protein-protein interactions. It was shown that several GTF2IRD1 interacting partners bind via the conserved SUMO motif-containing region (sumol) of

GTF2IRD1 (Table 4.2). The conformational change in GTF2IRD1 imposed by

200 SUMOylation may reveal or mask the binding surface with other proteins in vivo. Thus, it is possible that the binding of other interacting partners that do not directly bind to sumol region in GTF2IRD1 may also be regulated by this mechanism. As such, each interaction with GTF2IRD1 will need to be individually assessed. In particular, our analyses of the ZMYM5 interaction with GTF21RD1 in co-immunoprecipitation assays suggest that its binding is enhanced when GTF21RD1 is SUMOylated, most likely through a SIM-SUMO interaction (Hecker et al, 2006). A means to test this hypothesis could involve mutating all hydrophobic residues within the SIM of ZMYM5 to alanine residues (Hecker et al, 2006) and assessing the ability of this mutant protein to interact with GTF2IRD1, or specifically with the SUMO region of GTF2IRD1 in a Y2H assay.

Interestingly, Gocke and Yu (2008) reported that a similar SIM-mediated binding in

ZMYM2 is important for its interaction with SUMOylated HDAC1. The MYM-type zinc fingers in ZMYM2 are postulated to mediate protein-protein interactions in the

HDAC1-containing co-repressor complex while stabilising the complex on chromatin.

SUMOylation of HDAC1 enhances its binding to ZMYM2 which interacts non- covalently with the SUMO moeity.

Our findings that GTF2IRD1 interacts with proteins implicated in chromatin regulation

(Chapter 4) and is a target protein for SUMOylation, are relevant to the well- documented roles of SUMO in regulating chromatin biology. SUMOylation has been widely correlated with transcriptional repression and there is a remarkable number and diversity of chromatin modifying enzymes and proteins associated with SUMO- dependent transcriptional repression (Gill, 2005; Stielow et al, 2008). The general consensus for SUMO-dependent repressive mechanisms is that SUMOylation of transcription factors creates new interaction surfaces which recruit chromatin modifying enzymes such as histone deacetylases (HDACs) or co-repressor proteins such as DAXX

(Lin et al, 2006). The recruitment often involves SUMO-SIM interactions (Ouyang &

Gill, 2009). Thus, we predict that SUMOylation of GTF2IRD1 might play a role in mediating its autoregulatory repressive activity, presumably through a mechanism

201 where SUMO-modified GTF2IRD1 recruits a chromatin-modifying protein such as a

HDAC via ZMYM5. However, no evidence of such a mechanism was seen in the regulation of the GUR (Figure 5.9). Expression of GTF2IRD1 has a strong repressive activity on the GUR, which appears to be unaffected by its SUMOylation status. Since the interaction between GTF2IRD1 and ZMYM5 is not exclusively dependent on

GTF2IRD1 SUMOylation, then the association of GTF2IRD1 with a putative HDAC or chromatin modifying complex, which could be bridged by ZMYM5, may also not be significantly affected. To establish the level of importance of ZMYM5 and

SUMOylation in GTF2IRD1-associated repressive activity, it may be necessary to utilise more subtle assay systems that rely less on levels of over-expressed proteins.

A number of questions arise from the finding that GTF21RD1 is a novel substrate of protein SUMOylation. Firstly, is GTF2IRD1 also conjugated by SUMO-2/3 in vivo?

Although SUMO-2/3 tend to be conjugated to different target proteins in vivo, this possibility remains to be investigated (Saitoh & Hinchey, 2000). Secondly, is SUMOylation of GTF21RD1 cell-cycle dependent? Studies on some proteins such as

SUMOylation of PCNA (DNA-polymerase processivity factor) was found to occur in yeast exclusively during S-phase and disappears at later stages of the cell cycle most likely due to deconjugation (Hoege et al, 2002). This could be addressed by synchronising HEK293 cells and assaying the SUMOylation status of GTF2IRD1 at different stages of the cell cycle. Another question is which SENP enzyme targets the deconjugation of SUMO-GTF21RD1? In mammals, there are six SENPs (SENP1-3, 5-

7) which have different specificity and selectivity for SUMO maturation or deconjugation activity (reviewed in Mukhopadhyay & Dasso, 2007; Yeh, 2009).

Identification of the specific SENP responsible for GTF2IRD1 SUMO-deconjugation would add to our understanding of the dynamic regulation of GTF2IRD1 SUMOylation and its downstream effects.

202 A complex interplay between SUMOylation and other post-translational modifications

is often found on the same protein. This study represents the first characterised PTM of

GTF2IRD1 and other possible PTMs are yet to be investigated. For example, the

competition between SUMOylation and ubiquitylation on lysine residues is a potential

mechanism for the stability of target proteins. Many substrate proteins are also co­

regulated by SUMOylation and acetylation at lysine residues, such as MEF2A (Sahalizi

et al, 2007) and H 1C-1 (Stankovic-Valentin et al, 2007), both of which are

antagonistically regulated by the two modifications. In neurons, the signal-dependent

SUMOylation-acetylation switch in MEF2A promotes or inhibits dendritic

differentiation via MEF2-dependent gene transcription (Shalizi et al, 2006). Analyses of

a group of these substrate proteins revealed a common proline residue at the C-terminus

of the SUMO core motif (H^KXEP). Intriguingly, the functionally important SUMO

motif in GTF2IRD1 (sumol) is followed by the proline residue (I4bKlEP) which is

highly conserved throughout evolution (Figure 5.3), suggesting a high likelihood of the

proline acting as a SUMOylation-acetylation switch in GTF21RD1.

In addition, there is often cooperation between SUMOylation and phosphorylation where the negative charge conferred by phosphorylation of a serine residue often facilitates SUMO modification of the target lysine (Wilkinson & Henley, 2010). A phosphorylation-dependent SUMOylation motif (PDSM) has been described as

^KXEXXSP, with a proline-directed phosphorylatable serine residue located at the C- terminus of the SUMO motif (Hietakangas et al, 2006). No such serine residue is present in the region adjacent to the SUMO motif in GTF2IRD1. However, phospho- sites in GTF2IRD1 have been identified in large scale studies of phosphoproteins by mass spectrometry, including Ser-448 and Ser-654 (both conserved in human, mouse and xenopus, but not in fugu) (Beausoleil et al, 2004; Dephoure et al, 2008; Matsuoka et al, 2007; Olsen et al, 2006). Future studies aimed at identifying further PTMs on

GTF2IRD1 will aid our understanding of GTF2IRD1 cellular functions.

203 Figure 5.1. Interaction of GTF2IRD1 with SUMO E2 conjugating enzyme and E3 ligase.

(A) The SUMOylation pathway involves a cascade of enzymatic reactions (reprinted from Martin et al, 2007). SUMO proteins are matured by the hydrolysis activity of specific SUMO proteases (SENPs), which are then activated in an ATP-dependent process by SUMO El activating enzymes: SAE1/SAE2 dimers. Activated SUMO is transferred to the active cysteine of the E2 conjugating enzyme UBC9, which can directly recognise the canonical SUMO motif (PKXE; P Marge hydrophobic residue, KMysine, X=any residue, E=g/utamic acid) on substrate proteins and catalyse the transfer of SUMO. Alternatively, conjugation can occur in conjunction with an E3 ligase, a nonessential enzyme that confers substrate specificity. Conjugation involves formation of an isopeptide bond between SUMO and the target lysine of the substrate protein which can be cleaved by the SENP family, releasing free SUMO which can participate in the next cycle of SUMOylation of subsrates. (B) GTF2IRDI interacts with UBC9 and PIASX in a Y2H assay. The AH 109 yeast strain was co-transformed with the indicated bait and prey plasmids. Double transformed yeast colonies were selected on DDO medium and re-streaked on QDO + x-a-galactosidase medium where positive interactions were evident by the blue yeast colonies grown on this medium. Transformations with pGBKT7 and pGADT7 plasmids were used to test non-specific interaction with the Gal4 DNA-BD and AD domains of the bait and prey, respectively. (C) Survival of the yeast on DDO medium indicates the success of each transformation and the viability of corresponding diploids. A

SUMO hydrolase .—activity ;SENP)------| Maturation 1

SUMO V Activation isopeptidase activity \ ADP + Pi

| De-SUMOylation|

SUMO Modulation of target Transfer to E2 function conjugating enzyme (UBC9)

SUMO conjugation to target protein

Target H'KXE

Target recognition by UBC9/E3

B GTF2IRD1 C + UBC9

GTF2IRD1 + PIASX BKT7 PIASX

GTF2IRD pGBKT7 + pGADT7 + UBC9 QDO + x-a-gal DDO sumo 2 sumol C-term

Sumol -•!••• Human RA1t IlBljAAGNARB-PKGSMSEDCGPi PDPSPTSEEMTDSMPG M--LTDKGLSE Mouse RD; ■ si.l: a GTAWi-DMSSVSEDCGP DPSPTSEEMTDSLPG M--PADKC-PSE Frog KRI hnihfH-iksmfdqrifd P------EALAECVKS PTGKVCDDRPCD Fugu DKicHpapagrhr' Islmgahklsvpt EA PE------PPAVCR

Sumo2

Human ----- GSRDC GLHGQAPKVPPQDLPPTATSS S apscplap—sdBglsrpmpepkatgaqdfsMcgqkptgpg- Mouse ----- GSRDC GLHGQAS KV APQDLT PTAT P S S: vptcpltp—sJgmgwpvpephvpstodfsHcgqtpagpag Frog --SATKVt PDTSQCPRL--QRLPNCEHPLS vhsnmlgtnavsqblvhrpsaen nhdfsdfsBcgqqspvag- Fugu AWTASGHGPVKAAPPAGSTSSSATPSSSV |l^FKAMSFLS------§ A K D RAA S WTSGTDKM S VKjHaA AEE TRFG- VKXE

C-terminal ----|-----|---- |-----|-----|-----|-----|----- Human Hr a) BB:jDAKVPAKij|s-irWWWaj|s|&* s|s|ggr SSSSSSSN Mouse SSSSSS-N Frog teNSSSSSSSSSN Fugu I I |]I

Figure 5.2. Sequence alignments of regions containing SUMO motifs and the conserved C-terminal region in GTF2IRD1 orthologs.

ClustalW alignments of GTF2IRD1 in human, mouse, frog and fugu showing conservation of amino acid residues within regions that are involved in SUMOylation of GTF2IRD1. The first two regions, sumol and sumo2, contain the canonical SUMO consensus motifs (FKXE), except in the sumo2 region of the fugu sequence which has lost the W residue. The importance of sumol is emphasised by the perfect conservation of residues embedded in its SUMO motif. Also highlighted in the C-terminal region is the highly conserved last 27 amino acids which interact with PI ASX (Figure 5.3). Figure 5.3. Interaction of PIASX with the C-terminal region and UBC9 with the SUMO-motif containing regions of GTF2IRD1.

(A) Schematic diagram of human GTF2IRD1 with the various representative Y2H bait plasmids indicated above containing subregions or C-terminal truncations of GTF2IRD1 (Trl,2,3) generated to map the interaction sites with different protein partners (described in more detail in Figure 4.2). (B,C) Mapping of GTF21RD1 interaction regions with PIASx in Y2H assays. (B) Double transformations were performed using pGADT7-PIASX and the indicated bait plasmids from GTF2IRD1 containing either the full-length (FL) or the specific domains, or the empty vector pGBKT7 control (ctr). Double transformants were selected on DDO medium and re-spotted on QDO medium. Positive interactions are shown by the survival of transformed yeast on QDO media. (C) Similar experiment using the C-terminal truncation series of GTF2IRD1 to confirm the requirement of the C-terminal region for the interaction with PIASX. Note that deletion of the last section of the C-terminal end (Trl) and further truncations (Tr2, Tr3) abolished GTF2IRD1 interaction with PIASX. (D, E) Mapping of GTF2IRD1 interaction regions with UBC9 in Y2H assays. Yeast transformations were performed using pGADT7-UBC9 and the indicated bait plasmids from GTF2IRD1 as above. UBC9 predominantly interacts with regions of GTF2IRDI which contain the canonical SUMO motifs, sumo l and sumo2. Note that the interaction with the C-terminal region shown in (D) was not verified in (E) as all C-terminal truncated baits of GTF2IRD1 were able to interact with UBC9. A

LZ RD1 sumo2 RD2 sumol RD3 RD4 RD5 C-term

Tr3 Tr2 Trl

B

RD2 RD3 RD4 sumol C-term RD5

D

RD2 RD3 RD4 sumol C-term RD5 Figure 5.4. GTF2IRD1 is SUMOylated in vivo.

(A) Western blots showing protein extracts from COS-7 cells transfected with expression constructs for GTF2IRD1, HA-tagged SUMO and UBC9, and probed with anti-GTF2IRDl (M-19) and anti-HA tag antibodies. Unmodified GTF21RD1 protein was detected at approximately 120kDa as expected (arrowhead) and novel bands associated with GTF2IRD1 SUMOvlation are detectable at higher molecular weights (asterisks). Unconjugated HA-SUMO protein was detected at around 11 kDa (bottom panel). (B) Western blot analyses of immunoprecipitated GTF2IRD1 (IP) or whole cell extracts (input) from HEK293 cells transfected with expression constructs for GTF2IRD1 and GFP-PIASX, in the presence or absence of N-ethylmaleimide (NEM). Co-immunoprecipitation of SUMO-1 with GTF2IRD1 proteins in the presence of NEM are enclosed by the brackets. Detection of unconjugated monomers of the endogenous SUMO-1 proteins is shown on the bottom panel. GTF2IRD1 + + + HA-SUMO - + + UBC9 - + Stripped and reprobed kDa

*

*

100 -

15- Unconjugated SUMO

Blot: GTF2IRD1 Blot: HA (SUMO)

IP: GTF2IRD1

+ NEM + NEM GTF2IRD1 + + + + + + GFP-PIASX - + - +

SUMO-GTF2IRD1

Blot: GTF2IRD1 Blot: SUM01

Blot: GFP 100 (PIASX) Input 15 - Blot: SUM01 Figure 5.5. Mutation of lysine residues K271 and K495 to arginine abolishes GTF2IRD1 SUMOylation.

(A) Western blots showing protein extracts of COS-7 cells transfected with plasmids encoding GTF2IRD1 (wildtype, WT) or the double mutant K271R/K495R (2KR) and HA-SUMO, immunoprecipitated with anti- GTF21RD1 or anti-HA antibodies, and probed with the same or the reciprocal antibody. Both GTF2IRD1 (arrowhead) and SUMO- conjugated GTF2IRDJ (asterisks) are detectable in the input lysates. Cells were lysed in the presence of NEM prior to immunoprecipitation. (B) Western blot analyses of immunoprecipitated GTF2IRD1 or input lysate of HEK293 cells transfected with plasmids encoding GTF21RD1 (WT) or GTF2IRD1 2KR and probed with anti-GTF2IRDl and anti- SUMO-1 antibodies. In the presence of NEM, co-immunoprecipitation of GTF2IRD1, but not its 2KR mutant, with endogenous SUMO-1 protein was detected (asterisks). Input IP: GTF2IRD1 IP: GTF2IRD1 IP: HA

GTF2IRD1 WT WT 2KR WT WT 2KR WT WT 2KR WT WT 2KR HA-SUMO - ++ + + _ + + + +

250

150

100 Blot: GTF2IRD1 Blot: GTF2IRD1 Blot: HA Blot: GTF2IRD1

■ unconjugated SUMO

Blot: HA (SUMO)

B Input IP: GTF2IRD1 Input IP: GTF2IRD1 GTF2IRD1 WT WT 2KR WT WT 2KR WT WT 2KR WT WT 2KR NEM . + + - + + + + + +

kDa

250.

150- 100- «B* Blot: GTF2IRD1

Blot: SUMO-1 GTF2IRD1 ^ ^ ^ HA-SUMO - + + + +

Blot: GTF2IRD1

Blot: HA (SUMO)

Figure 5.6. Lys-495 is the main ligation site for GTF2IRD1 SUMOylation.

Western blot showing protein extracts from COS-7 cells transfected with a plasmid encoding GTF2IRD1 (WT), the single lysine to arginine mutants (GTF2IRD1 K495R or K271R) or the combined mutant (GTF2IRD1 2KR), probed with the anti-GTF2IRDl antibody. SUMO conjugation was detectable on the GTF2IRD1 WT and the K271R mutant, but not on the K495R or 2KR mutants of GTF2IRD1. Approximately equal level of unconjugated monomers of the transfected HA-SUMO proteins are shown in the bottom panel. GFP-GTF2IRD1 HA (SUMO) MERGE /DAPI

Figure 5.7. SUMOylation does not affect the nuclear localisation of GTF2IRD1.

NIH3T3 cells transiently transfected with GFP-tagged GTF2IRD1 or the 2KR mutant with or without HA-SUMO. The nuclear localisation pattern of GTF2IRD1 is shown by GFP fluorescence (gf'een channel, left column). HA- SUMO was detected by immunostaining with anti-HA and Alexa-555 antibodies (red channel, middle panel). DAPI staining (blue) to visualise the nuclei is shown in the merged images (right panels). Figure 5.8. SUMOylation does not significantly alter the protein stability of GTF2IRD1.

(A) Western blot showing protein extracts from COS-7 cells collected at different time points (hour, h) following treatment with the protein synthesis inhibitor cycloheximide (CHX). Cycloheximide was added at 24 hours after transfection with the GTF2IRD1 -expressing plasmid and in the presence of a proteosome degradation inhibitor MG 132 or DMSO control. The blot was probed with a GAPDH antibody as a loading control. (B) A similar experiment was performed to assess the stability of ETS-1, a rapidly degraded protein which served as a positive control for the CHX assay. (C) Graph showing densitometric analysis of Western blots of GTF2IRD1 or GTF2IRD1 2KR normalised to the corresponding GAPDH value from CHX experiments described in (A) without MG 132 pretreatment. Normalised values are expressed as a percentage of the maximum value at the 0 h time point in each group. Plotted data represents mean ± S.E.M. (n=3). Data analysed by unpaired t-test analysis showed no statistical differences in the level of the two proteins at any time points (note: p-value = 0.0531 at 1 h). DMSO MG132 + CHX (h) 0 1247 21 0 1247 21

GTF2IRD1

GAPDH

DMSO MG132 + CHX (h) 0 0.5 1 2 4 7 0 0.5 1 2 4 7

Ets-1

GAPDH

£

0 a; CM Ll_ 1 (D o 0) > _0)

“O

Time (h) C 5-r 3 0

i i I i t r 1 2 3 4 5 6 7

pGL3-GUR + + + + + + +

pCDNA3.1-GTF2IRD1 + +

pCDNA3.1-GTF2IRD1-2KR + +

pMT3-HA-SUM0 + + +

pCDNA3.1-SUMO- + GTF2IRD1

Figure 5.9. SUMOylation does not affect the transcriptional repressive activity of GTF2IRD1.

Luciferase activity of the GTF2IRD1 upstream region (GUR)-luciferase reporter in extracts from COS-7 cells transfected with pGL3-GUR in combination with other plasmids indicated was measured 24 hours post-transfection. The amount of DNA transfected was kept constant at 1.5 pg by adjusting the pCDNA3.1 empty vector content. Lucifer ase activity was expressed as a percentage relative to the maximal activity detected in cells transfected with pGL3-GUR alone. Data represents mean S.E.M. from two independent experiments (each with 3 replicates). A GFP + GFP-ZMYM5 - + + + + GTF2IRD1 WTWT2KRWT 2KR HA-SUMO - - - + +

Blot: H GTF2IRD1 IP: GFP

Blot: GFP (ZMYM5)

INPUT Blot: GTF2IRD1

B

8 ro 20-

i o 10-

IP GFP (Zmym5)

Figure 5.10. GTF2IRD1 SUMOylation increases ZMYM5 binding affinity.

(A) Immimoprecipitation (with anti-GFP antibody) of lysates from HEK293 cells transfected with GFP-ZMYM5, GTF2IRD1 (WT or 2KR) with or without HA- SUMO expression plasmids, followed by Western blot analysis of the immunoprecipitated and input fractions using anti-GTF2lRDl and GFP antibodies. The input and co-immunoprecipitated unmodified GTF21RD1 proteins (arrowhead) or the SUMO-conjugated forms (asterisks) are indicated. (B) Densitometric analysis of the Western blots shows the percentage (%) of SUMOylated GTF2IRD1 out of the total amount of GTF21RD1 that interacted with GFP-ZMYM5 (lane 4 of IP:GFP panel) in the co-immunoprecipitation assays, in comparison to the initial % detected in the input lysates (lane 4 of input panel). Paired t-test analysis was done for each independent experiment (n=5), **p-value = 0.0002 CHAPTER 6 - OVERVIEW AND FUTURE DIRECTIONS

6.1 Overview

Neurogenetics research is a powerful tool that can be employed to understand the genetic basis of human behaviour and cognition by examining what happens to brain function when one gene or a group of genes are disrupted in human conditions, such as

WBS, or in corresponding animal models. The key to such powerful analyses lies within the ability to connect well-characterized neurological consequences with the causal genetic insult. Then the task of unravelling the molecular and cellular basis of the defect becomes a well-defined and finite aim. GTF2IRD1 and the evolutionarily related protein TFII-I have been strongly implicated in the neuropathology of the Williams syndrome cognitive profile. Clinical data from WBS patients carrying atypical deletions, together with evidence from monogenic mouse models of Gtf2irdl mutation, provide ample support for the role of GTF2IRDI in specific neurological aspects of the disease. However, the molecular function of GTF2IRD1 is still poorly understood. The work shown in this thesis provides some new insights into the molecular mechanisms of

GTF2IRD1 function and regulation (Figure 6.1) and the key findings are discussed below.

6.1.1 Changes in the gene transcriptional profile of Gtflirdl knockout mice

At the outset of this work, GTF2IRD1 was widely regarded as a DNA-binding transcription factor and we, therefore, sought to identify genes that are differentially regulated in the brains of Gtf2irdl""l,lul knockout mice which could potentially be downstream target genes for GTF2IRD1 regulation (Chapter 3). The project was undertaken using two microarray-based approaches: a brain region-specific analysis based on isolation of RNA from four major regions and a cell-specific analysis that

219 utilised FAC sorted GABAergic neurons as the test population. In the first method, a relatively small subset of genes were identified as showing a greater than two-fold change in expression and it emerged that those showing the most substantial changes of expression levels in all brain regions of the KO mice could be attributed to hybridization artefacts that arose due to SNPs in the genomic sequence of alleles closely linked to the mutant Gtf2irdl gene. Other candidate genes located outside this region showed weaker changes of expression level, which were not consistently displayed in all brain regions, and to date, none have been confirmed by independent assays. It was assumed that some of the weakness in the degree of transcriptional change could be due to a dilution effect caused by the cellular heterogeneity of the brain region under study.

Therefore, the second methodology held a much greater potential to identify genes dysregulated by GTF2IRD1 loss by targeting the analysis to the GABAergic neurons, in which Gtf2irdl is highly expressed. Many more genes were identified by this study as being differentially expressed but validation of these microarray data has been hampered by the insufficient yields of RNA from each FACS isolation and it is likely that this screen was susceptible to contamination problems and possible artefacts caused by the time taken to process each sample. Two aspects of this study have been considerably improved since the project started, hence warranting its reinvestigation in the near future. Firstly, we now have a more comprehensive map of Gtjlirdl expression in the brain, including a catalogue of the domains that overlap with specific populations of GABAergic neurons. This means that a combination of region-specific isolation in combination with FACS could generate a very pure population of similar neuronal types that all express Gtf2irdl. Isolation of the striatum would be a good candidate for such a strategy. Secondly, protocols for FACS isolation of neurons from postnatal mouse brain and RNA purification are more readily available. In particular, a very similar procedure using the same model system, Gad 11 Tama, to isolate GABAergic neurons for a downstream microarray analysis has been reported with positive outcomes (Hardt et al,

2008). In previous work, we had defined clear rules for high affinity DNA binding of

GTF2IRD1, which included two independent occurrences of the GGATTA motif not

220 more than 60bp apart (Palmer et al. 2010). This allowed us to perforin in silico genome­ wide search to identify genes that could potentially be regulated by GTF21RD1. By looking for instances of the motif in regions that were conserved and not more than 5kb from transcription start sites, we identified approximately 100 genes that could be considered as candidates. However, based on the brain region microarray dataset, none of these genes seemed to show much difference in expression between KO and WT samples.

Overall, the unexpected outcome of this project suggests that if GTF21RD1 acts as a transcriptional regulator, then its role is probably more subtle and complex than originally conceived. A similar result was obtained in the analysis of another Gt/2irdl

KO model (Lucy Osborne: personal communication) suggesting that our experience is not due to methodological problems. Therefore, it is conceivable that it does not act as a conventional transcription factor that binds to cis-regulator DNA sequences in order to direct transcription of a specific and consistent set of target genes. Findings from the proteomics screen discussed in the next section show that GTF21RD1 could be involved in the epigenetic regulation of chromatin, which is a dynamic and context-dependent mechanism. Under these circumstances, GTF2IRD1 might have a more modulatory role in transcriptional regulation that varies according to cell type and cell status and this would be consistent with the observed microarray data.

6.1.2 Novel protein partners of GTF2IRD1

The key findings of the yeast two-hybrid library screen (Chapter 4) include the discovery of a total of nine novel candidate proteins that were shown to interact with

GTF2IRD1. Supporting evidence for the validity of these interactions is at different stages. SETD6 and ZMYM5 have been shown to occupy the nucleus and to interact with GTF2IRD1 in mammalian cells. However, the other interactions are yet to be validated. The interacting regions of the new protein partners have been mapped onto various conserved domains of GTF2IRD1. The repeat domains (RDs), previously

221 unexplored for their protein-binding capacity, turned out to be main sites for protein- protein interactions as well as DNA binding domains, with the exception of RD5. Some proteins also interact with regions containing consensus SUMOylation sequences as well as the highly conserved C-terminal region of GTF2IRD1.

A subset of the novel proteins, including SETD6, ZMYM5, and DCAF6, probably function as histone modifying proteins, which suggests a potential novel role of

GTF21RD1 in chromatin regulation. Preliminary investigations into the levels of H3K14 and H4K12 monomethylation; modifications which have been associated with SETD6, showed no significant changes in the brains of Gtf2irdl KO mice at a global level. A more comprehensive assessment of histone modifications will be required in future studies (discussed in the next section). KPNA2 is an importin family member and we found that it binds exclusively to the nuclear localisation signal of GTF21RD1.

Therefore, it is likely that this protein is responsible for the transport of GTF2IRD1 into the nucleus.

Other candidate proteins that have no known or predicted function, such as the armadillo repeat-containing ARMCX5 or the forkhead-domain containing FHAD1 will need to be examined in a stepwise fashion. It will be very important to establish more supporting evidence for the biological relevance of these interactions, by doing co- immunopreciptation and co-localisation studies, before embarking on strategies to determine their function. However, it is interesting to speculate based on the small amount of evidence available. It is conceivable that ARMCX5 has some sort of signalling role based on its homology to p-catenin and the possibility of a protein that could shuttle between the cytoplasm and the nucleus would open up a whole new area of investigation. Also, examination of FHAD1 expression shows that, in the brain, it is confined to the choroid plexus (Alan brain atlas): a non-neuronal cell type in which

GTF2IRD1 is also found. It is conceivable that a defect in the choroid plexus could lead to abnormalities in the production of cerebrospinal fluid, which could explain the

222 increased brain ventricle volume observed in some of the Gtf2irdl mutants (van Hagen et al, 2007). In this way, the analysis of protein interactions of GTF2IRD1 has opened up new hypotheses, which is beneficial for the project as a whole.

6.1.3 GTF21RD1 as a target for SUMOylation in vivo

This project arose from the finding that two components of the SUMOylation enzymatic pathway, UBC9 (E2 conjugating enzyme) and PIASX (E3 ligase) were capable of direct interaction with GTF2IRD1 at recognisable SUMO-motif-containing regions and the conserved C-terminal region, respectively. It was then shown that GTF2IRD1 is

SUMOylated in vivo, predominantly at Lys-495 and the level of SUMOylation is enhanced by PIASX overexpression. SUMOylation does not appear to affect the subcellular localisation, protein stability and transcriptional activity of GTF2IRD1.

However, the physical interaction of ZMYM5 is detectably enhanced with the

SUMOylated form of GTF2IRD1, most likely by the non-covalent interaction of its

SUMO interacting motif (SIM) with the SUMO moiety of the modified GTF2IRD1.

This finding is the first demonstration of a post-translational modification on

GTF2IRD1 and suggests that tight regulatory control of GTF2IRD1 function is an important aspect of its role. When considered in conjunction with the demonstrated auto-regulation of the Gtf2irdl gene (Palmer et al. 2010), such tight regulation at the protein and transcript level suggests that the gene could easily be susceptible to dosage sensitivity in the haploid state, as experienced in WBS. In addition, the fact that

GTF2IRD1 is SUMOylated also provides some small clues to function. SUMOylation is mainly restricted to nuclear proteins and is a growing aspect of chromatin biology, which is consistent with the demonstrated interactions with DCAF6, ZMYM5 and

SETD6.

223 6.2 Future Directions

The results reported in this thesis have provided some new insights into the regulation of GTF21RD1 function at the molecular level. In particular, the identification of novel protein partners of GTF2IRD1 has provided new avenues for future research. The following section discusses some significant new questions that arise as a consequence of the findings made in this thesis and suggests approaches that could be employed in investigating those questions.

How can we further interrogate the role of GTF21RD1 in gene transcription?

The functional consequences of GTF21RD1 loss, at the molecular level and in terms of the neurological phenotypes displayed by the KO mouse models, appears to be relatively subtle. However, one of the most interesting phenotypes of C57BL/6-

Gtf2irdl"",Hni mice is their novel vocalisation as a response to environmentally induced stress — which is thought to be a heightened anxiety-like phenotype (Howard et al, submitted). This observation raises the question of whether the activity of GTF2IRD1 is context (promoter/cell-type/signal)-dependent. Activation of TFII-I in a context- dependent manner has been well demonstrated, whereby specific signalling cues (such as angiogenesis or stress) leads to its phosphorylation and/or cytoplasmic sequestration.

This, in turn, activates its gene transcription activity, either through direct binding to gene regulatory elements (such as the Inr and E-box elements of VEGFR-2) or by interacting with other transcription factors (such as the interaction with SRF and STAT proteins on the c-fos promoter) (reviewed in Roy, 2007).

From our in silico analysis, we concluded that GTF2IRD1 sequence-specific gene regulation cannot be predicted by its DNA binding properties. A future experiment that would substantiate this finding is to identify genomic regions that GTF2IRD1 is associated to in vivo by using chromatin immunoprecipitation (ChlP)-based assay. In

ChIP, antibodies are used to immunoprecipitate specific proteins, which may be bound either directly or indirectly to regions of genomic DNA. Parallel sequencing of these

224 DNA fragments in a procedure known as ChIP-Seq would allow for identification of the bound sequence motifs, as well as a more accurate list of gene targets for GTF2IRD1

(through wild type and Gtf2irdl KO mice comparison) as the sequencing reaction is not dependent on probe hybridisation, as would be the case for its probe-dependent array- based predecessor (ChlP-chip) (Park, 2009). It would then be informative to correlate the ChIP-Seq patterns to gene expression data from the microarray analyses. If the expression of a gene correlates with the binding of GTF2IRD1 to a nearby regulatory sequence, then the gene is a good candidate for being a direct target gene of GTF2IRD1.

Technically, the success of any ChIP experiment depends on the quality of the antibody used, thus, the sensitivity and specificity of the GTF2IRD1 antibody must first be resolved. Ultimately, the use of ChIP-Seq technology is central to investigations that examine the role of GTF2IRD1 in chromatin regulation, which is described in the next section.

Recent advances in Gtf2irdl expression mapping in the mouse brain (Howard et al, submitted) will allow us to make better judgements about the exact brain regions that we should focus on for further microarray analysis (also discussed in Chapter 3). It is also worth noting that any future microarray analysis will be performed using the congenic N10 Gtf2irdl KO mice in order to minimise artefacts that were encountered in the project due to retention of parts of the 129-derived genome around the Gtf2irdl allele.

The potential role of GTF2IRD1 in histone modification

If GTF2IRD1 does indeed play a role in chromatin regulation (Chapter 4) then a major challenge will be to determine which of the multiple post-translational modification mark(s), that are laid down on histone tails, are involved in the role of GTF2IRD1

(Figure 1.2). Although SETD6 is known to be a lysine methyltransferase, its activity towards specific histone lysine residues has not been definitively demonstrated.

Similarly, the histone deacetylase, or potentially other chromatin regulating enzyme(s),

225 that ZMYM5 associates with, have yet to be revealed experimentally.. If this information becomes available, genome-wide characterisation of the histone modification(s) involved can be mapped using the ChlP-Seq procedure (subject to the availability of ChIP-grade antibodies that recognise the specific modification). The remarkable progress in high-throughput sequencing platforms allows spatial resolution at the nucleosome level and an accurate identification of the DNA sequence elements enriched at the specific chromatin mark. Parallel ChlP-Seq analysis should also be performed on chromatin purified from Gtf2irdl KO mouse tissue, to identify changes in the histone modification profile. Furthermore, ChlP-Seq profiles of the chromatin modification can be compared to the ChlP-Seq profile of GTF2IRD1 binding sites, which should indicate some overlapping patterns if GTF2IRD1 and the histone- modifying proteins are indeed recruited to the same loci in the genome.

On the basis of one study on SETD6 (Richter, 2009), the levels of H3K14mel and

H4K12mel were measured in Gtf2irdl KO mice. No basal changes to the levels of these chromatin marks were observed in the different brain regions of Gtf2irdl KO mice. However, altered chromatin plasticity, including changes to histone acetylation and methylation levels have been associated to prolonged neural activity (reviewed in

Borrelli et al, 2008), such as memory and learning, which were demonstrated in the brains of mice following fear conditioning training (Gupta et al, 2010; Peleg et al,

2010). Therefore, in future experiments it would be worthwhile considering an analysis of the levels of histone modifications in the brain tissues of mice following specific behavioural tests; for example, in response to anxiety-provoking tests in Gtf2irdl KO mice. It is likely that the induced changes would take place in association with regulatory regions of specific genes, particularly those that are activated or repressed in response to anxiety. Interestingly, mice carrying a conditional deletion of the gene encoding KAP-1 (KRAB-associated protein 1) in brain have anxiety-like and stress- induced impairment. It was shown that neurons in the forebrain had decreased

226 H3K9me3 and increased H3 and H4 acetylation at the promoter of the KAP-1 target gene, Mkrn3 (Jakobsson et ai 2008).

Further biochemical analysis of GTF2IRD1

Many proteins identified as novel partners of GTF2IRD1 in this study have poorly defined functions, and thus, require a lot of further work to be of benefit to GTF2IRD1 research. However, it is also likely that there are other interacting proteins yet to be identified, which could be identified using Y2H screens of different cDNA libraries, or by other proteomics methodologies, such as immunoprecipitation of GTF2IRD1- containing complexes and mass spectrometry analysis. However, immunoprecipitation of endogenous GTF2IRD1 could be technically very difficult due to its low abundance.

This could be circumvented by transfecting an epitope-tagged GTF2IRD1 into cell lines before affinity purification using an antibody against the epitope. Chromatin remodelling proteins, including histone acetyltransferases and deacetylases, histone methyltransferases and demethylases, are often assembled into multi-subunit complexes. For example, ZMYM2, which is evolutionarily-related to ZMYM5, is a member of a transcriptional co-repressor complex that contains a lysine demethylase (LSD1), the REST co-repressor (Co-REST), and HDAC1/2. Formation of this LSD1-

Co-REST-HDAC complex on chromatin is thought to be stabilised by ZMYM2 (Gocke

& Yu, 2008). Therefore, it is also possible that GTF2IRD1 could be part of a multi- subunit chromatin remodelling complex, and other unknown members of the complex could be identified by mass-spectrometry.

Further analysis of GTF2IRD1 at the cellular level

Recombinant GTF2IRD1 expressed in cell lines is predominantly detected in the nucleus. Furthermore, by manipulation of fixation techniques it is possible to see nuclear foci where GTF21RD1 is located at greater concentrations. In non-synchronized cell cultures, it is possible to see some variability in these patterns, which would prompt an investigation of sub-cellular distribution during the cell cycle using synchronized

227 cultures. In addition, the interaction of GTF2IRD1 with KPNA2 should be investigated further as it may provide the mechanism for GTF2IRD1 nuclear import. This question could be addressed by examining sub-cellular co-localisation and by treating cells with siRNA blocking agents targeted against KPNA2. Efforts to identify endogenous proteins that co-localise with GTF2IRD1 foci have so far been unsuccessful, but will be continually endeavoured based on the types of proteins identified in the proteomics approaches and on hypotheses that arise from the molecular studies.

Concluding remarks

WBS presents a unique opportunity for researchers to study the genetic, and possibly the epigenetic mechanisms, underlying aspects of human cognition, mood and behaviour. In the work described in this thesis, we have taken a bottom-up approach by aiming to characterise the molecular function of GTF2IRD1, one of the two GTF2I family genes that are strongly implicated in WBS. An added benefit to such an approach is in what it can tell us about the function of TF1I-I, the protein encoded by GTF2I. A global view based on the current literature would suggest that these two proteins have very divergent functions. However, since both of these genes are derived from a common ancestor that existed at some time prior to the radiation of the vertebrates and there is still a significant degree of sequence similarity, it seems likely that they should share some functions in common. Our finding of an interaction with ZMYM5, as the corollary of the TFII-I interaction with a ZMYM2-containing complex is an example of such potential functional overlaps. Since TFII-I is also a major player in the cause of

WBS, it is of high importance to determine the function of TFII-I and GTF2IRD1 and any potential synergy between these two proteins. Therefore, as our molecular analysis of GTF2IRD1 proceeds by all of the means discussed, we will assess the properties of recombinant TFII-I also.

Perhaps the most exciting finding in this thesis is the novel interaction between

GTF2IRD1 and proteins that are involved in chromatin remodelling, the main

228 instrument of epigenetic control. While neurogenetic research has provided much of our understanding of the genetic regulation of the nervous system, ‘neuroepigenetic’' research has just begun, and is rapidly gaining ground as one of the important mechanisms that guide dynamic plasticity and maintenance of long-lasting neuronal responses (Borrelli et al, 2008). Finally, with the characterisation of another chromatin regulator contained within the WBS critical region, BAZ1B (Ashe et al, 2008; Kitagawa et al, 2011), it is tempting to think that WBS could be a disease of epigenome regulation. Therefore, it will also be important to examine the possible synergy between

GTF2IRD1, TFII-I and BAZ1B, to see if the combination of the whole is greater than the sum of the parts.

229 Stimulus

Figure 6.1 The working model of GTF2IRD1 function

GTF2IRD1 interactions with proteins involved in chromatin remodelling, including SETD6 (in histone lysine methylation) and ZMYM5 (associated with histone deacetylation) may regulate specific regions in the genome, such as the promoter of its own gene, which contains multiple GTF2IRD1 binding sites (GTF2IRD1 upstream region; GUR). Epigenetic mechanisms are known to be highly influenced by environmental factors, especially those in the nervous system, which could be guiding the dynamic plasticity processes and long- lasting neuronal responses (Borrelli et al, 2008). Thus, it is possible that GTF2IRD1 s role in chromatin regulation, especially in the CNS, could be triggered by environmental stimuli. In addition, SUMOvlation which predominantly targets Lys-495 of GTF2IRD1 appears to have a positive effect on its interaction with ZMYM5, which is likely to be due to the non-covalent interaction between the SUMO moiety and the SUMO interacting motif of ZMYM5. REFERENCE LIST

Adolphs R. Tranel D, Hamann S, Young AW, Calder AJ, Phelps EA, Anderson A, Lee GP, Damasio AR (1999) Recognition of facial emotion in nine individuals with bilateral amygdala damage. Neuropsychologia 37: 1111-1117

Akbarian S, Sucher NJ, Bradley D, Tafazzoli A, Trinh D, Hetrick WP, Potkin SG, Sandman CA, Bunney WE, Jr, Jones EG (1996) Selective alterations in gene expression for NMDA receptor subunits in prefrontal cortex of schizophrenics. J Neurosci 16: 19-30

Alarcon JM, Malleret G, Touzani K, Vronskaya S, Ishii S, Kandel ER, Barco A (2004) Chromatin acetylation, memory, and LTP are impaired in CBP+/- mice: a model for the cognitive deficit in Rubinstein-Taybi syndrome and its amelioration. Neuron 42: 947-959

Amir RE, Van den Veyver IB, Wan M, Tran CQ, Francke U, Zoghbi HY (1999) Rett syndrome is caused by mutations in X-linked MECP2, encoding methyl-CpG-binding protein 2. Nat Genet 23: 185-188

Amstrong DD (2002) Neuropathology of Rett syndrome. Ment Retard Dev Disabil Res Rev 8: 12-16

Anderson DJ, Groves A, Lo L, Ma Q, Rao M, Shah NM, Sommer L (1997) Cell lineage determination and the control of neuronal identity in the neural crest. Cold Spring Harb Symp Quant Biol 62: 493-504

Anderson S, Mione M, Yun K, Rubenstein JL (1999) Differential origins of neocortical projection and local circuit neurons: role of Dlx genes in neocortical interneuronogenesis. Cereb Cortex 9: 646-654

Angers S, Li T, Yi X, MacCoss MJ, Moon RT, Zheng N (2006) Molecular architecture and assembly of the DDBI-CUL4A ubiquitin ligase machinery. Nature 443: 590-593

Anggono V, Cousin M, Robinson P (2008) Styryl dye-based synaptic vesicle recycling assay in cultured cerebellar granule neurons. Methods Mol Biol: 333-345

Antequera F, Bird A (1993) Number of CpG islands and genes in human and mouse. Proc Natl AcadSci USA 90: 11995-11999

Antonell A, de Luis O, Domingo-Roura X, Perez-Jurado LA (2005) Evolutionary mechanisms shaping the genomic structure of the Williams-Beuren syndrome chromosomal region at human 7ql 1.23. Genome Res 15: 1179-1188

Antonell A, Del Campo M, Magano LF, Kaufmann L, de la Iglesia JM, Gallastegui F, Flores R, Schweigmann U, Fauth C, Kotzot D, Perez-Jurado LA (2010) Partial 7ql 1.23 deletions further implicate GTF2I and GTF2IRD1 as the main genes responsible for the Williams-Beuren syndrome neurocognitive profile. J Med Genet 47: 312-320

Arlotta P, Molyneaux BJ, Chen J, Inoue J, Kominami R, Macklis JD (2005) Neuronal subtype- specific genes that control corticospinal motor neuron development in vivo. Neuron 45: 207- 221

Ashe A, Morgan DK, Whitelaw NC, Bruxner TJ, Vickaryous NK, Cox LL, Butterfield NC, Wicking C, Blewitt ME, Wilkins SJ, Anderson GJ, Cox TC, Whitelaw E (2008) A genome- 231 wide screen for modifiers of transgene variegation identifies genes with critical roles in development. Genome Biol 9: R182

Axelsson S (2005) Variability of the cranial and dental phenotype in Williams syndrome. Swed Dent JSuppl 170: 3-67

Ayaydin F, Dasso M (2004) Distinct in vivo dynamics of vertebrate SUMO paralogues. Mol Biol Cell 15: 5208-5218

Bailey CH, Kandel ER (1993) Structural changes accompanying memory storage. Annu Rev Physiol 55: 397-426

Bannister AJ, Kouzarides T (2005) Reversing histone methylation. Nature 436: 1103-1106

Bannister AJ, Zegerman P, Partridge JF, Miska EA, Thomas JO, Allshire RC, Kouzarides T (2001) Selective recognition of methylated lysine 9 on histone H3 by the HP1 chromo domain. Nature 410: 120-124

Barreto G, Schafer A, Marhold J, Stach D, Swaminathan SK, Handa V, Doderlein G, Maltry N, Wu W, Lyko F, Niehrs C (2007) Gadd45a promotes epigenetic gene activation by repair- mediated DNA demethylation. Nature 445: 671-675

Barski A, Cuddapah S, Cui K, Roh TY, Schones DE, Wang Z, Wei G, Chepelev 1, Zhao K (2007) High-resolution profiling of histone methylations in the human genome. Cell 129: 823- 837

Bawa-Khalfe T, Yeh ET (2010) SUMO Losing Balance: SUMO Proteases Disrupt SUMO Homeostasis to Facilitate Cancer Development and Progression. Genes Cancer 1: 748-752

Bayarsaihan D, Bitchevaia N, Enkhmandakh B, Tussie-Luna Ml, Leckman JF, Roy A, Ruddle F (2003) Expression of BEN, a member of TFII-1 family of transcription factors, during mouse pre- and postimplantation development. Gene Expr Patterns 3: 579-589

Bayarsaihan D, Dunai J, Greally JM, Kawasaki K, Sumiyama K, Enkhmandakh B, Shimizu N, Ruddle FH (2002) Genomic organization of the genes Gtf2ird 1, Gtf2i, and Ncfl at the mouse chromosome 5 region syntenic to the human chromosome 7ql 1.23 Williams syndrome critical region. Genomics 79: 137-143

Bayarsaihan D, Ruddle FH (2000) Isolation and characterization of BEN, a member of the TFII- 1 family of DNA-binding proteins containing distinct helix-loop-helix domains. Proc Natl Acad Sci USA 91: 7342-7347

Bayer P, Arndt A, Metzger S, Mahajan R, Melchior F, Jaenicke R, Becker J (1998) Structure determination of the small ubiquitin-related modifier SUMO-1. J Mol Biol 280: 275-286

Bayes M, Magano LF, Rivera N, Flores R, Perez Jurado LA (2003) Mutational mechanisms of Williams-Beuren syndrome deletions. Am J Med Genet 73: 131-151

Beausoleil S, Jedrychowski M, Schwartz D, Elias J, Villen J, Li J, Cohn M, Cantley L, Gygi S (2004) Large-scale characterization of HeLa cell nuclear phosphoproteins. Proc Natl Acad Sci USA 101: 12130 - 12135

Bellugi U, Adolphs R, Cassady C, Chiles M (1999a) Towards the neural basis for hypersociability in a genetic syndrome. Neuroreport 10: 1653-1657

232 Bellugi U, Korenberg JR. Klima ES (2001) Williams syndrome: an exploration of neurocognitive and genetic features. Clin Neurosci Res 1: 217-229

Bellugi U, Lichtenberger L, Mills D, Galaburda A, Korenberg JR (1999b) Bridging cognition, the brain and molecular genetics: evidence from Williams syndrome. Trends Neurosci 22: 197- 207

Bellugi U, Lichtenberger L, Jones W, Lai Z, St.George M (2000) The neurocognitive profile of Williams syndrome: A complex pattern of strengths and weaknesses. J Cogn Neurosci 12: 7-29

Benito E, Barco A (2010) CREB's control of intrinsic and synaptic plasticity: implications for CREB-dependent memory models. Trends Neurosci 33: 230-240

Berger SL, Kouzarides T, Shiekhattar R, Shilatifard A (2009) An operational definition of epigenetics. Genes Dev 23: 781-783

Berglund L, Bjorling E, Oksvold P, Fagerberg L, Asplund A, Szigyarto CA, Persson A, Ottosson J, Wernerus H, Nilsson P, Lundberg E, Sivertsson A, Navani S, Wester K, Karnpf C, Hober S, Ponten F, Uhlen M (2008) A genecentric Human Protein Atlas for expression profiles based on antibodies. Mol Cell Proteomics 7: 2019-2027

Bernier-Villamor V, Sampson DA, Matunis MJ, Lima CD (2002) Structural basis for E2- mediated SUMO conjugation revealed by a complex between ubiquitin-conjugating enzyme Ubc9 and RanGAPl.Ce// 108: 345-356

Berninger B, Guillemot F, Gotz M (2007) Directing neurotransmitter identity of neurones derived from expanded adult neural stem cells. Eur J Neurosci 25: 2581-2590

Bernstein BE, Meissner A, Lander ES (2007) The mammalian epigenome. Cell 128: 669-681

Bertrand N, Castro DS, Guillemot F (2002) Proneural genes and the specification of neural cell types. Nat Rev Neurosci 3: 517-530

Beunders G, van de Kamp JM, Veenhoven RH, van Hagen JM, Nieuwint AW, Sistermans EA (2010) A triplication of the Williams-Beuren syndrome region in a patient with mental retardation, a severe expressive language delay, behavioural problems and dysmorphisms. J Med Genet 47: 271-275

Bird A (2002) DNA methylation patterns and epigenetic memory. Genes Dev 16: 6-21

Bird AP (1986) CpG-rich islands and the function of DNA methylation. Nature 321: 209-213

Bleakman D (1999) Kainate receptor pharmacology and physiology. Cell Mol Life Sci 56: 558- 566

Blomberg S, Rosander M, Andersson G (2006) Fears, hyperacusis and musicality in Williams syndrome. Res Dev Disabil 27: 668-680

Boddaerta N, Mochelc F, Meressea I, Seidenwurmd D, Cachiaa A, Brunellea F, Lyonnetc S, Zilbovicius M (2005) Parieto-occipital grey matter abnormalities in children with Williams syndrome. Neuroimage 30: 721-725

233 Boddy MN, Howe K, Etkin LD, Solomon E, Freemont PS (1996) PIC 1, a novel ubiquitin-like protein which interacts with the PML component of a multiprotein complex that is disrupted in acute promyelocytic leukaemia. Oncogene 13: 971-982

Boeger H. Griesenbeck J, Strattan JS, Romberg RD (2003) Nucleosomes unfold completely at a transcriptionally active promoter. Mol Cell 11: 1587-1598

Bogdanovic O, Veenstra GJ (2009) DNA methylation and methyl-CpG binding proteins: developmental requirements and function. Chromosoma 118: 549-565

Bontempo SJ (2009) A molecular analysis of the Williams Syndrome associated gene, GTF2IRD1. BSc (Med) (Hons) Thesis, Department of Anatomy and Histology, University of Sydney, Sydney

Borrelli E, Nestler EJ, Allis CD, Sassone-Corsi P (2008) Decoding the epigenetic language of neuronal plasticity. Neuron 60: 961-974

Botta A, Novelli G, Mari A, Novelli A, Sabani M, Korenberg J, Osborne LR, Digilio MC, Giannotti A, Dallapiccola B (1999) Detection of an atypical 7q 11.23 deletion in Williams syndrome patients which does not include the STX1A and FZD3 genes. J Med Genet 36: 478- 480

Bowery NG, Bettler B, Froestl W, Gallagher JP, Marshall F, Raiteri M, Bonner TI, Enna SJ (2002) International Union of Pharmacology. XXXIII. Mammalian y -aminobutyric acidB receptors: Structure and function. Pharmacol Rev 54: 247-264

Brock J (2007) Language abilities in Williams syndrome: a critical review. Dev Psychopathol 19: 97-127

Brodtkorb E, Nilsen G, Smevik O, Rinck PA (1992) Epilepsy and anomalies of neuronal migration: MRI and clinical aspects. Acta Neurol Scand 86: 24-32

Broide RS, Redwine JM, Aftahi N, Young W, Bloom FE, Winrow CJ (2007) Distribution of histone deacetylases 1-11 in the rat brain. J Mol Neurosci 31: 47-58

Brose N (1999) Synaptic cell adhesion proteins and synaptogenesis in the mammalian central nervous system. Naturw’issenschaften 86: 516-524

Brown JR, Ye H, Bronson RT, Dikkes P, Greenberg ME (1996) A defect in nurturing in mice lacking the immediate early gene fosB. Cell 86: 297-309

Brunetti-Pierri N, Berg JS, Scaglia F, Belmont J. Bacino CA, Sahoo T, Lalani SR, Graham B, Lee B, Shinawi M, Shen J, Kang SH, Pursley A, Lotze T, Kennedy G, Lansky-Shafer S, Weaver C, Roeder ER, Grebe TA, Arnold GL, Hutchison T, Reimschisel T, Amato S, Geragthy MT, Innis JW, Obersztyn E, Nowakowska B. Rosengren SS, Bader PI, Grange DK, Naqvi S, Garnica AD, Bernes SM, Fong CT, Summers A, Walters WD, Lupski JR, Stankiewicz P, Cheung SW, Patel A (2008) Recurrent reciprocal 1 q21.1 deletions and duplications associated with microcephaly or macrocephaly and developmental and behavioral abnormalities. Nat Genet 40: 1466-1471

Burbach JPH (2010) Neuropeptides from concept to online database vvww.neuropeptides.nl. Eur J Pharmacology’ 626: 27-48

234 Calvo S, Vullhorst D, Venepally P, Cheng J, Karavanova I, Buonanno A (2001) Molecular dissection of DNA sequences and factors involved in slow muscle-specific transcription. Mol Cell Biol 21: 8490-8503

Caraveo G, van Rossum DB, Patterson RL, Snyder SH, Desiderio S (2006) Action of TF1I-I outside the nucleus as an inhibitor of agonist-induced calcium entry. Science 314: 122-125

Carrasco X, Castillo S, Aravena T, Rothhammer P. Aboitiz F (2005) Williams syndrome: Pediatric, neurologic, and cognitive development. Pediatr Neurol 32: 166-172

Casarosa S, Fode C, Guillemot F (1999) Mashl regulates neurogenesis in the ventral telencephalon. Development 126: 525-534

Casteel DE, Boss GR, Pilz RB (2005) Identification of the interface between cGMP-dependent protein kinase Ibeta and its interaction partners TFI1-I and IRAG reveals a common interaction motif. J Biol Chem 280: 38211-38218

Casteel DE, Zhuang S, Gudi T, Tang J, Vuica M, Desiderio S, Pilz RB (2002) cGMP-dependent protein kinase 1 beta physically and functionally interacts with the transcriptional regulator TFII-I. J Biol Chem 277: 32003-32014

Cedar H. Bergman Y (2009) Linking DNA methylation and histone modification: patterns and paradigms. Nat Rev Genet 10: 295-304

Chahrour M, Jung SY, Shaw C, Zhou X, Wong ST, Qin J, Zoghbi HY (2008) MeCP2, a key contributor to neurological disease, activates and represses transcription. Science 320: 1224- 1229

Chahrour M, Zoghbi HY (2007) The story of Rett syndrome: from clinic to neurobiology. Neuron 56: 422-437

Chapman CA, du Plessis A, Pober BR (1996) Neurologic findings in children and adults with Williams syndrome. J Child Neurol 11: 63-65

Chaudhari N, Hahn WE (1983) Genetic Expression in the Developing Brain. Science 220: 924- 928

Chawla S, Vanhoutte P, Arnold FJ, Huang CL, Bading H (2003) Neuronal activity-dependent nucleocytoplasmic shuttling of HDAC4 and HDAC5. JNeurochem 85: 151-159

Chen WG, Chang Q, Lin Y, Meissner A, West AE, Griffith EC, Jaenisch R, Greenberg ME (2003) Derepression of BDNF transcription involves calcium-dependent phosphorylation of MeCP2. Science 302: 885-889

Cheng L, Arata A, Mizuguchi R, Qian Y, Karunaratne A, Gray PA, Arata S, Shirasawa S, Bouchard M, Luo P, Chen CL, Busslinger M, Goulding M, Onimaru H, Ma Q (2004) Tlx3 and Tlxl are post-mitotic selector genes determining glutamatergic over GABAergic cell fates. Nat Neurosci 7: 510-517

Cheng L, Samad OA, Xu Y, Mizuguchi R. Luo P, Shirasawa S, Goulding M, Ma Q (2005) Lbxl and Tlx3 are opposing switches in determining GABAergic versus glutamatergic transmitter phenotypes. Nat Neurosci 8: 1510-1515

235 Cheriyath V, Desgranges ZP, Roy AL (2002) c-Src-dependent transcriptional activation of TFII-I. J Biol Chem 277: 22798-22805

Cheriyath V, Roy AL (2000) Alternatively spliced isoforms of TFII-I. J Biol Chem 275: 26300- 26308

Cheriyath V, Roy AL (2001) Structure-function analysis of TFII-I. Roles of the N-terminal end, basic region, and I-repeats. J Biol Chem 276: 8377-8383

Cherniske EM, Carpenter TO, Klaiman C, Young E, Bregman J, Insogna K, Schultz RT, Pober BR (2004) Multisystem study of 20 older adults with Williams syndrome. Am J Med Genet A 131:255-264

Cherniske EM, Sadler LS, Schwartz D, Carepenter TO, Pober B (1999) Early puberty in Williams syndrome. Clin Dysmorphol 8: 117-121

Chiang MC, Reiss AL, Lee AD, Bellugi U, Galaburda AM, Korenberg JR, Mills DL, Toga AW, Thompson PM (2007) 3D pattern of brain abnormalities in Williams syndrome visualized using tensor-based morphometry. Neuroimage 36: 1096-1109

Chimge NO, Mungunsukh O, Ruddle F, Bayarsaihan D (2007) Expression profiling of BEN regulated genes in mouse embryonic fibroblasts. J Exp Zool B Mol Dev Evol 308: 209-224

Chook YM, Blobel G (2001) Karyopherins and nuclear import. Curr Opin Struct Biol 11: 703- 715

Chourbaji S, Brandwein C, Gass P (2010) Altering BDNF expression by genetics and/or environment: Impact for emotional and depression-like behaviour in laboratory mice. Neurosci Biohehav Rev 35: 599-61 1

Church GM, Gilbert W (1984) Genomic sequencing. Proc Natl Acad Sci USA 81: 1991-1995

Citri A, Malenka RC (2008) Synaptic plasticity: multiple forms, functions, and mechanisms. Neuropsychopharmacology 33: 18-41

Clayton-Smith J, Laan L (2003) Angleman syndrome: a review of the clinical and genetic aspects. J Med Genet 40: 87-95

Cline HT (2005) Synaptogenesis: A balancing act between excitation and inhibition. Curr Biol 15: R203-205

Cobos I, Calcagnotto ME, Vilaythong AJ, Thwin MT, Noebels JL, Baraban SC, Rubenstein JLR (2005) Mice lacking Dlxl show subtype-specific loss of interneurons, reduced inhibition and epilepsy. Nat Neurosci 8: 1059-1068

Cochran BH, Zullo J, Verma IM, Stiles CD (1984) Expression of the c-fos gene and of an fos- related gene is stimulated by platelet-derived growth factor. Science 226: 1080-1082

Cohen-Cory S, Kidane AH, Shirkey NJ, Marshak S (2010) Brain-derived neurotrophic factor and the development of structural neuronal connectivity. Dev Neurobiol 70: 271-288

Collette JC, Chen XN, Mills DL, Galaburda AM, Reiss AL, Bellugi U, Korenberg JR (2009) William's syndrome: gene expression is related to parental origin and regional coordinate control. J Hum Genet 54: 193-198

236 Collins RT, 2nd, Kaplan P, Somes GW, Rome JJ (2010) Cardiovascular abnormalities, interventions, and long-term outcomes in infantile Williams syndrome. JPediatr 156: 253-258

Conacci-Sorrell M, Zhurinsky J, Ben-Ze'ev A (2002) The cadherin-catenin adhesion system in signaling and cancer. J Clin Invest 109: 987-991

Contamina P, Boada E (1992) Time of origin of cortico-collicular projection neurons in the rat visual cortex. J Comp Neurol 323: 570-576

Conti E, Izaurralde E (2001) Nucleocytoplasmic transport enters the atomic age. Curr Opin Cell Biol 13: 310-319

Couve A, Moss SJ, Pangalos MN (2000) GABAB receptors: a new paradigm in G protein signaling. Mol Cell Neurosci 16: 296-312

Cunha C, Brambilla R, Thomas KL (2010) A simple role for BDNF in learning and memory? Front Mol Neurosci 3

Curran ME, Atkinson DL, Ewart AK, Morris CA, Leppert MF, Keating MT (1993) The elastin gene is disrupted by a translocation associated with supravalvular aortic stenosis. Cell 73: 159- 168

Dai L, Bellugi U, Chen XN, Pulst-Korenberg AM, Jarvinen-Pasley A, Tirosh-Wagner T, Eis PS, Graham J, Mills D, Searcy Y, Korenberg JR (2009) Is it Williams syndrome? GTF2IRD1 implicated in visual-spatial construction and GTF21 in sociability revealed by high resolution arrays. Am J Med Genet A 149A: 302-314

Danoff SK, Taylor HE, Blackshaw S, Desiderio S (2004) TFII-I, a candidate gene for Williams syndrome cognitive profile: parallels between regional expression in mouse brain and human phenotype. Neuroscience 123: 931-938

Davies M, Udwin O, Howlin P (1998) Adults with Williams syndrome. Preliminary study of social, emotional and behavioural difficulties. Br J Psychiatry 172: 273-276

Davis AM, Penschuck S, Fritschy JM, McCarthy MM (2000) Developmental switch in the expression of GABA(A) receptor subunits alpha(l) and alpha(2) in the hypothalamus and limbic system of the rat. Brain Res Dev Brain Res 119: 127-138

Day JJ, Sweatt JD (2010) DNA methylation and memory formation. Nat Neurosci 13: 1319- 1323 de Ruijter AJM, van Gennip AH, Caron HN, Kemp S, van Kuilenburg ABP (2003) Histone deacetylases (HDACs): characterization of the classical HDAC family. Biochem T370: 737-749

De Zeeuw Cl, Hoogenraad CC, Goedknegt E, Hertzberg E, Neubauer A, Grosveld F, Galjart N (1997) CLIP-115, a Novel Brain-Specific Cytoplasmic Linker Protein, Mediates the Localization of Dendritic Lamellar Bodies. Neuron 19: 1187-1199

DeFelipe J (1993) Neocortical neuronal diversity: Chemical heterogeneity revealed by colocalization studies of classic neurotransmitters, neuropeptides, calcium-binding proteins, and cell surface molecules. Cereb Cortex 3: 273-289

DeFelipe J (2002) Cortical interneurons: From Cajal to 2001. Prog Brain Res 136: 215-238

237 DeFelipe J, Alonso-Nanclares L, Arellano JI (2002) Microstructure of the neocortex: Comparative aspects. J Neurocytol 31: 299-316

Del Campo M, Antonell A, Magano LF, Munoz FJ, Flores R, Bayes M, Perez Jurado LA (2006) Hemizygosity at the NCF1 gene in patients with Williams-Beuren Syndrome decreases their risk of hypertension. Am J Hum Gen 78: 533-542

Dephoure N, Zhou C, Villen J, Beausoleil S, Bakalarski C, Elledge S, Gygi S (2008) A quantitative atlas of mitotic phosphorylation. Proc Natl Acad Sci USA 105: 10762 - 10767

Desgranges ZP, Ahn J, Lazebnik MB, Ashworth T, Lee C, Pestell RC, Rosenberg N, Prives C, Roy AL (2005) Inhibition of TFIl-I-dependent cell cycle regulation by p53. Mol Cell Biol 25: 10940-10952

Desgranges ZP, Roy AL (2006) TFII-I: connecting mitogenic signals to cell cycle regulation. Cell Cycle 5: 356-359

Dillon SC, Zhang X, Trievel RC, Cheng X (2005) The SET-domain protein superfamily: protein lysine methyltransferases. Genome Biol 6: 227

DiMario JX, Stockdale FE (1997) Both myoblast lineage and innervation determine fiber type and are required for expression of the slow myosin heavy chain 2 gene. Dev Biol 188: 167-180

Dingwall C, Laskey RA (1991) Nuclear targeting sequences—a consensus? Trends Biochem Sci 16: 478-481

Doi-Katayama Y, Hayashi F, Inoue M, Yabuki T, Aoki M, Seki E, Matsuda T, Kigawa T, Yoshida M, Shirouzu M, Terada T, Hayashizaki Y, Yokoyama S, Hirota H (2007) Solution structure of the general transcription factor 21 domain in mouse TFII-I protein. Protein Sci: ps.072792007

Donnai D, Karmiloff-Smith A (2000) Williams Syndrome: from genotype through to the cognitive phenotype. Am J Med Genet 97: 164-171

Dou Y, Milne TA, Tackett AJ, Smith ER, Fukuda A, Wysocka J, Allis CD, Chait BT, Hess JL, Roeder RG (2005) Physical association and coordinate function of the H3 K4 methyltransferase MLL1 and the H4 K16 acetyltransferase MOF. Cell 121: 873-885

Dresbach T, Qualmann B, Kessels MM, Garner CC, Gundelfinger ED (2001) The presynaptic cytomatrix of brain synapses. Cell Mol Life Sci 58: 94-116

Dulac C (2010) Brain function and chromatin plasticity. Nature 465: 728-735

Durkin ME, Keck-Waggoner CL, Popescu NC, Thorgeirsson SS (2001) Integration of a c-myc transgene results in disruption of the mouse GtfZirdl gene, the homologue of the human GTF2IRD1 gene hemizygously deleted in Williams-Beuren syndrome. Genomics 73: 20-27

Dykens EM (2003) Anxiety, fears, and phobias in persons with Williams syndrome. Dev Neuropsychol 23: 291-316

Eckert MA, Hu D, Eliez S, Bellugi U, Galaburda A, Korenberg J, Mills D, Reiss AL (2005) Evidence for superior parietal impairment in Williams syndrome. Neurology’ 64: 152-153

238 Egan MF, Kojima M. Callicott JH, Goldberg TE, Kolachana BS, Bertolino A, Zaitsev E, Gold B, Goldman D, Dean M (2003) The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell 112: 257-269

Ehrlich M, Gama-Sosa MA, Huang LH, Midgett RM, Kuo KC, McCune RA, Gehrke C (1982) Amount and distribution of 5-methylcytosine in human DNA from different types of tissues of cells. Nucl Acids Res 10: 2709-2721

Eichler SA, Meier JC (2008) E-I balance and human diseases - from molecules to networking. Front Mol Neurosci 1: 2

Einfeld SL, Tonge BJ, Florio T (1997) Behavioral and emotional disturbance in individuals with Williams syndrome. Am JMent Retard 102: 45-53

Enkhmandakh B. Bitchevaia N, Ruddle F, Bayarsaihan D (2004) The early embryonic expression of TF1I-I during mouse preimplantation development. Gene Expr Patterns 4: 25-28

Enkhmandakh B, Makeyev AV, Erdenechimeg L. Ruddle FH, Chimge NO. Tussie-Luna MI, Roy AL, Bayarsaihan D (2009) Essential functions of the Williams-Beuren syndrome- associated TFII-I genes in embryonic development. Proc Natl Acad Sci USA 106: 181-186

Ensenauer RE, Adeyinka A, Flynn HC, Michels VV, Lindor NM, Dawson DB, Thorland EC, Lorentz CP, Goldstein JL, McDonald MT, Smith WE, Simon-Fayard E, Alexander AA, Kulharya AS, Ketterling RP, Clark RD, Jalal SM (2003) Microduplication 22q 11.2, an emerging syndrome: clinical, cytogenetic, and molecular analysis of thirteen patients. Am J Hum Genet 73: 1027-1040

Eronen M, Peippo M, Hiippala A, Raatikka M, Arvio M, Johansson R, Kahkonen M (2002) Cardiovascular manifestations in 75 patients with Williams syndrome. J Med Genet 39: 554- 558

Ewart AK, Jin W, Atkinson D, Morris CA, Keating MT (1994) Supravalvular aortic stenosis associated with a deletion disrupting the elastin gene. J Clin Invest 93: 1071-1077

Ewart AK, Morris CA, Atkinson D, Jin W, Sternes K, Spallone P, Stock AD, Leppert M, Keating MT (1993a) Hemizygosity at the elastin locus in a developmental disorder, Williams syndrome. Nat Genet 5: 11-16

Ewart AK, Morris CA, Ensing GJ, Loker J, Moore C, Leppert M, Keating M (1993b) A human vascular disorder, supravalvular aortic stenosis, maps to . Proc Natl Acad Sci U S A 90: 3226-3230

Fasano CA, Dimos JT, Ivanova NB, Lowry N, Lemischka IR, Temple S (2007) shRNA knockdown of Bmi-1 reveals a critical role for p21-Rb pathway in NSC self-renewal during development. Cell Stem Cell 1: 87-99

Faury G, Pezet M, Knutsen RH, Boyle WA, Heximer SP, McLean SE, Minkes RK, Blumer KJ, Kovacs A, Kelly DP. Li DY, Starcher B, Mecham RP (2003) Developmental adaptation of the mouse cardiovascular system to elastin haploinsufficiency. J Clin Invest 112: 1419-1428

Feil R, Feil S, Hofmann F (2005a) A heretical view on the role of NO and cGMP in vascular proliferative diseases. Trends Mol Med 11: 71-75

239 Feil R, Hofmann F, Kleppisch T (2005b) Function of cGMP-dependent protein kinases in the nervous system. Rev Neurosci 16: 23-41

Feng J, Zhou Y, Campbell SL, Le T, Li E, Sweatt JD. Silva AJ, Fan G (2010) Dnmtl and Dnmt3a maintain DNA methylation and regulate synaptic function in adult forebrain neurons. Nat Neurosci 13: 423-430

Ferre-D'Amare AR, Prendergast GC, Ziff EB, Burley SK (1993) Recognition by Max of its cognate DNA through a dimeric b/HLH/Z domain. Nature 363: 38-45

Ferrero GB, Howald C, Micale L, Biamino E, Augello B, Fusco C, Turturo MG, Forzano S, Reymond A, Merla G (2010) An atypical 7q 11.23 deletion in a normal IQ Williams-Beuren syndrome patient. Eur J Hum Genet 18: 33-38

Fijalkowska I, Sharma D, Bult CJ, Danoff SK (2010) Expression of the transcription factor, TF1I-L during post-implantation mouse embryonic development. BMC Res Notes 3: 203

Fingerman 1M, Du HN, Briggs SD (2008) In Vitro Histone Methyltransferase Assay. CSH ProtocZ: nihpa83349

Fischer A, Sananbenesi F, Wang X, Dobbin M, Tsai LH (2007) Recovery of learning and memory is associated with chromatin remodelling. Nature 447: 178-182

Fisher SE, Scharff C (2009) FOXP2 as a molecular window into speech and language. Trends Genet 25: 166-177

Flavell SW, Cowan CW, Kim TK, Greer PL, Lin Y, Paradis S, Griffith EC, Hu LS, Chen C, Greenberg ME (2006) Activity-dependent regulation of MEF2 transcription factors suppresses excitatory synapse number. Science 311: 1008-1012

Flavell SW, Greenberg ME (2008) Signaling mechanisms linking neuronal activity to gene expression and plasticity of the nervous system. Annu Rev Neurosci 31: 563-590

Fleischmann A, Hvalby O, Jensen V, Strekalova T, Zacher C, Layer LE, Kvello A, Reschke M, Spanagel R, Sprengel R, Wagner EF, Gass P (2003) Impaired long-term memory and NR2A- type NMDA receptor-dependent synaptic plasticity in mice lacking c-Fos in the CNS. ./ Neurosci 23: 9116-9122

Fossella J, Sommer T, Fan J, Wu Y, Swanson JM, Pfaff DW, Posner MI (2002) Assessing the molecular genetics of attention networks. BMC Neurosci 3: 14

Francis SH, Busch JL, Corbin JD, Sibley D (2010) cGMP-dependent protein kinases and cGMP phosphodiesterases in nitric oxide and cGMP action. Pharmacol Rev 62: 525-563

Francke U (1999) Williams-Beuren syndrome: genes and mechanisms. Hum Mol Genet 8: 1947-1954

Frangiskakis JM, Ewart AK, Morris CA, Mervis CB, Bertand J, Robinson BF, Klein BP, J EG, Everett LA, Green ED, Proschel C, Gutowski NJ, Noble M, Atkinson DL, Odelberg SJ, Keating MT (1996) LIM-kinasel hemizygosity implicated in impaired visuospatial constructive cognition. Cell 86: 296-304

240 Franke Y, Peoples RJ, Francke U (1999) Identification of GTF2IRD1, a putative transcription factor within the Williams-Beuren syndrome deletion at 7q 11.23. Cytogenet Cell Genet 86: 296-304

Franklin TB, Mansuy IM (2010) The prevalence of epigenetic mechanisms in the regulation of cognitive functions and behaviour. Curr Opin Neurobiol 20: 441-449

Frazer A, Molinoff P, Winokur A (1994) Biological bases of brain function and disease, New York: Raven Press, Ltd.

Freedman R. Coon FI, Myles-Worsley M, Orr-Urtreger A, Olincy A, Davis A, Polymeropoulos M, Holik J, Hopkins J, Hoff M, Rosenthal J, Waldo MC, Reimherr F, Wender P, Yaw J, Young DA, Breese CR, Adams C, Patterson D, Adler LE, Kruglyak L, Leonard S, Byerley W (1997) Linkage of a neurophysiological deficit in schizophrenia to a chromosome 15 locus. Proc Natl AcadSci USA 94: 587-592

Frigerio E, Burt DM, Gagliardi C, Cioffi G, Martelli S, Perrett DI, Borgatti R (2006) Is everybody always my friend? Perception of approachability in Williams syndrome. Neuropsychologia 44: 254-259

Fritschy JM, Paysan J, Enna A, Mohler H (1994) Switch in the expression of rat GABAA- receptor subtypes during postnatal development: an immunohistochemical study. JNeurosci 14: 5302-5324

Fryer A (1997) Angelman and Prader-Willi syndromes. Curr Paediatrics 7: 242-245

Fujiwara T, Mishima T, Kofuji T, Chiba T, Tanaka K, Yamamoto A, Akagawa K (2006) Analysis of knock-out mice to determine the role of HPC-l/syntaxin 1A in expressing synaptic plasticity. J Neurosci 26: 5767-5776

Fuller RW (1991) Role of serotonin in therapy of depression and related disorders. J Clin Psychiatry 52 Suppl: 52-57

Gagliardi C, Bonaglia MC, Selicorni A, Borgatti R, Giorda R (2003) Unusual cognitive and behavioural profile in a Williams syndrome patient with atypical 7ql 1.23 deletion. J Med Genet 40: 526-530

Gagliardi C, Martelli S, Burt MD, Borgatti R (2007) Evolution of neurologic features in Williams syndrome. Pediatr Neurol 36: 301-306

Galaburda AM, Holinger DP, Bellugi U, Sherman GF (2002) Williams syndrome: neuronal size and neuronal-packing density in primary visual cortex. Arch Neurol 59: 1461-1467

Ganesan AK, Kho Y, Kim SC, Chen Y, Zhao Y, White MA (2007) Broad spectrum identification of SUMO substrates in melanoma cells. Proteomics 7: 2216-2221

Gao MC, Bellugi U, Dai L, Mills DL, Sobel EM, Lange K, Korenberg JR (2010) Intelligence in Williams Syndrome Is Related to STX1A, Which Encodes a Component of the Presynaptic SNARE Complex. PLoSONE 5: el0292

Garcia-Dominguez M, Reyes JC (2009) SUMO association with repressor complexes, emerging routes for transcriptional control. Biochim Biophys Acta 1789: 451-459

241 Garner CC, Kindler S, Gundelfinger ED (2000) Molecular determinants of presynaptic active zones. Curr Opin Neurobiol 10: 321-327

Gehring M, Reik W, Henikoff S (2009) DNA demethylation by DNA repair. Trends Genet 25: 82-90

Geiss-Friedlander R, Melchior F (2007) Concepts in sumoylation: a decade on. Nat Rev Mol Cell Biol 8: 947 - 956

Geoffroy MC, Hay RT (2009) An additional role for SUMO in ubiquitin-mediated proteolysis. Nat Rev Mol Cell Biol 10: 564-568

Gietz RD, Woods RA (2002) Transformation of yeast by Li Ac/ SS carrier DNA/ PEG Method. Methods in Enzymology 35: 87-96

Gill G (2005) Something about SUMO inhibits transcription. Curr Opin Genet Dev 15: 536-541

Gocke CB. Yu H (2008) ZNF198 stabilizes the LSDl-CoREST-HDACl complex on chromatin through its MYM-type zinc fingers. PLoS ONE 3: e3255

Goldberg TE, Weinberger DR (2004) Genes and the parsing of cognitive processes. Trends Cogn Sci 8: 325-335

Goldman SE, Malow BA, Newman KD, Roof E, Dykens EM (2009) Sleep patterns and daytime sleepiness in adolescents and young adults with Williams syndrome. J Intellect Disabil Res 53: 182-188

Gong L, Li B, Millas S, Yeh ET (1999) Molecular cloning and characterization of human AOS1 and UBA2, components of the sentrin-activating enzyme complex. EEBS Lett 448: 185-189

Gonzalez GA, Montminy MR (1989) Cyclic AMP stimulates somatostatin gene transcription by phosphorylation of CREB at serine 133. Cell 59: 675-680

Gorlich D, Henklein P, Laskey RA, Hartmann E (1996) A 41 amino acid motif in importin- alpha confers binding to importin-beta and hence transit into the nucleus. EMBO J 15: 1810- 1817

Gothelf D, Farber N, Raveh E, Apter A, Attias J (2006) Hyperacusis in Williams syndrome: Characteristics and associated neuroaudiologic abnormalities. Neurology 66: 390-395

Gotz M, Huttner WB (2005) The cell biology of neurogenesis. Nat Rev Mol Cell Biol 6: 777- 788

Graeff FG, Guimaraes FS, De Andrade TG, Deakin JF (1996) Role of 5-HT in stress, anxiety, and depression. Pharmacol Biochem Behav 54: 129-141

Grant J, Valian V, Karmiloff-Smith A (2002) A study of relative clauses in Williams syndrome. J Child Lang 29: 403-416

Gray V, Karmiloff-Smith A, Funnel! E, Tassabehji M (2006) In-depth analysis of spatial cognition in Williams syndrome: A critical assessment of the role of the L1MK1 gene. Neuropsychologia 44: 679-685

242 Greeberg F (1990) Williams syndrome professional symposium. Am J Med Genet Suppl 6: 85- 88

Green MJ, Phillips ML (2004) Social threat perception and the evolution of paranoia. Neurosci Biobehav Rev 28: 333-342

Greenberg ME, Ziff EB (1984) Stimulation of 3T3 cells induces transcription of the c-fos proto­ oncogene. Nature 311: 433-438

Greer PL, Flanayama R, Bloodgood BL, Mardinly AR, Lipton DM, Flavell SW, Kim TK, Griffith EC, Waldon Z, Maehr R, Ploegh HL, Chowdhury S, Worley PF, Steen J, Greenberg ME (2010) The Angelman Syndrome protein Ube3A regulates synapse development by ubiquitinating arc. Cell 140: 704-716

Grimaud C, Negre N, Cavalli G (2006) From genetics to epigenetics: the tale of Polycomb group and trithorax group genes. Chromosome Res 14: 363-375

Grueneberg DA, Henry RW, Brauer A, Novina CD, Cheriyath V, Roy AL, Gilman M (1997) A multifunctional DNA-binding protein that promotes the formation of serum response factor/homeodomain complexes: identity to TFII-I. Genes Dev 11: 2482-2493

Guan JS, Haggarty SJ, Giacometti E, Dannenberg JH, Joseph N, Gao J, Nieland TJ, Zhou Y, Wang X, Mazitschek R, Bradner JE, DePinho RA, Jaenisch R, Tsai LH (2009) HDAC2 negatively regulates memory formation and synaptic plasticity. Nature 459: 55-60

Gupta S, Kim SY, Artis S, Molfese DL, Schumacher A, Sweatt JD, Paylor RE, Lubin FD (2010) Histone methylation regulates memory formation. J Neurosci 30: 3589-3599

Haas BW, Hoeft F, Searcy YM, Mills D, Bellugi U, Reiss A (2010) Individual differences in social behavior predict amygdala response to fearful facial expressions in Williams syndrome. Neuropsychologia 48: 1283-1288

Hakimi M-A, Dong Y, Lane WS, Speicher DW, Shiekhattar R (2003) A Candidate X-linked Mental Retardation Gene Is a Component of a New Family of Histone Deacetylase-containing Complexes. J Biol Chem 278: 7234-7239

Hakre S, Tussie-Luna MI, Ashworth T, Novina CD, Settleman J, Sharp PA, Roy AL (2006) Opposing functions of TFII-I spliced isoforms in growth factor-induced gene expression. Mol Cell 24: 301-308

Hammond P, Hutton TJ, Allanson JE, Buxton B, Campbell LE, Clayton-Smith J, Donnai D, Karmiloff-Smith A, Metcalfe K, Murphy KC, Patton M, Pober B, Prescott K, Scambler P, Shaw A, Smith AC, Stevens AF, Temple IK, Hennekam R, Tassabehji M (2005) Discriminating power of localized three-dimensional facial morphology. Am J Hum Genet 77: 999-1010

Han Z, Guo L, Wang H, Shen Y, Deng XW, Chai J (2006) Structural basis for the specific recognition of methylated histone H3 lysine 4 by the WD-40 protein WDR5. Mol Cell 22: 137- 144

Hanson RD, Hess JL, Yu BD, Ernst P, van Lohuizen M, Berns A, van der Lugt NM, Shashikant CS, Ruddle FH, Seto M, Korsmeyer SJ (1999) Mammalian Trithorax and polycomb-group homologues are antagonistic regulators of homeotic development. Proc Natl Acad Sci USA 96: 14372-14377

243 Hardingham GE (2009) Coupling of the NMDA receptor to neuroprotective and neurodestructive events. Biochem Soc Tram 37: 1147-1160

Hardingham GE, Chawla S, Cruzalegui FH, Bading H (1999) Control of recruitment and transcription-activating function of CBP determines gene regulation by NMDA receptors and L- type calcium channels. Neuron 22: 789-798

Hardt O, Scholz C, Kusters D, Yanagawa Y, Pennartz S, Cremer H, Bosio A (2008) Gene expression analysis defines differences between region-specific GABAergic neurons. Mol Cell Neurosci 39: 418-428

Harris NGS, Bellugi U, Bates E, Jones W, Rossen M (1997) Contrasting profiles of language development in children with williams and down syndromes. Developmental Neuropsychology 13: 345 - 370

Hassan AH, Awad S, Al-Natour Z, Othman S, Mustafa F, Rizvi TA (2007) Selective recognition of acetylated histones by bromodomains in transcriptional co-activators. Biochem J 402: 125-133

Hatzfeld M (1999) The armadillo family of structural proteins. Int Rev Cytol 186: 179-224

Hay RT (2007) SUMO-specific proteases: a twist in the tail. Trends Cell Biol 17: 370-376

He YJ, McCall CM, Hu J, Zeng Y, Xiong Y (2006) DDB1 functions as a linker to recruit receptor WD40 proteins to CUL4-ROC1 ubiquitin ligases. Genes Dev 20: 2949-2954

Hecker CM. Rabiller M, Haglund K, Bayer P, Dikic 1 (2006) Specification of SUMOl- and SUM02-interacting motifs../ Biol Chem 281: 16117-16127

Heintzman ND, Stuart RK, Hon G, Fu Y, Ching CW, Hawkins RD, Barrera LO, Van Calcar S, Qu C, Ching KA, Wang W. Weng Z, Green RD, Crawford GE, Ren B (2007) Distinct and predictive chromatin signatures of transcriptional promoters and enhancers in the human genome. Nat Genet 39: 31 1-318

Heisler LK, Chu HM, Brennan TJ, Danao JA, Bajwa P, Parsons LH, Tecott LH (1998) Elevated anxiety and antidepressant-like responses in serotonin 5-HT1A receptor mutant mice. Proc Natl AcadSci USA 95: 15049-15054

Heller R, Rauch A, Luttgen S, Schroder B, Winterpacht A (2003) Partial deletion of the critical 1.5 Mb interval in Williams-Beuren syndrome. J Med Genet 40: e99

Hietakangas V, Anckar J, Blomster HA, Fujimoto M, Palvimo JJ, Nakai A, Sistonen L (2006) PDSM, a motif for phosphorylation-dependent SUMO modification. Proc Natl Acad Sci USA 103: 45-50

Higa LA, Wu M, Ye T, Kobayashi R. Sun H, Zhang H (2006) CUL4-DDB1 ubiquitin ligase interacts with multiple WD40-repeat proteins and regulates histone methylation. Nat Cell Biol 8: 1277-1283

Hill CS, Treisman R (1995) Transcriptional regulation by extracellular signals: mechanisms and specificity. Cell 80: 199-211

Hinsley TA, Cunliffe P, Tipney HJ, Brass A, Tassabehji M (2004) Comparison of TFI1-I gene family members deleted in Williams-Beuren syndrome. Protein Sci 13: 2588-2599

244 Hirota H, Matsuoka R, Chen XN, Salandanan LS, Lincoln A, Rose FE, Sunahara M, Osawa M, Bellugi U, Korenberg J (2003) Williams Syndrome deficits in visual spatial processing linked to GTF21RD1 and GTF2I on chromosome 7ql 1.23. Genet Med 5: 311-321

Hirsch MR, Tiveron MC, Guillemot F, Brunet JF, Goridis C (1998) Control of noradrenergic differentiation and Phox2a expression by MASH1 in the central and peripheral nervous system. Development 125: 599-608

Ho SN, Hunt HD. Horton RM, Pullen JK, Pease LR (1989) Site-directed mutagenesis by overlap extension using the polymerase chain reaction. Gene 77: 51-59

Hobart HH, Morris CA, Mervis CB, Pani AM, Kistler DJ, Rios CM, Kimberley KW, Gregg RG, Bray-Ward P (2010) Inversion of the Williams syndrome region is a common polymorphism found more frequently in parents of children with Williams syndrome. American Journal of Medical Genetics Part C: Seminars in Medical Genetics 154C: 220-228

Hocking D, Rinehart N, McGinley J, Bradshaw J (2008) Gait function in adults with Williams syndrome. Exp Brain Res

Hoege C, Pfander B, Moldovan G-L, Pyrowolakis G, Jentsch S (2002) RAD6-dependent DNA repair is linked to modification of PCNA by ubiquitin and SUMO. Nature 419: 135-141

Hoffman CS, Winston F (1987) A ten-minute DNA preparation from yeast efficiently releases autonomous plasmids for transformation of Escherichia coli. Gene 57: 267-272

Hofmann F, Bernhard D, Lukowski R, Weinmeister P (2009) cGMP Regulated Protein Kinases (cGK). In cGMP: Generators, Effectors and Therapeutic Implications, Schmidt HHHW, Hofmann F, Stasch J-P (eds), Vol. 191, pp 137-162. Springer Berlin Heidelberg

Hong M, Lin MY, Fluang JM, Baumeister P, Hakre S, Roy AL, Lee AS (2005) Transcriptional regulation of the Grp78 promoter by endoplasmic reticulum stress: role of TFII-I and its tyrosine phosphorylation. J Biol Chem 280: 16821-16828

Hong SJ, Li H, Becker KG, Dawson VL, Dawson TM (2004) Identification and analysis of plasticity-induced late-response genes. Proc Natl Acad Sci USA 101: 2145-2150

Hoogenraad CC, Koekkoek B, Akhmanova A, Krugers H, Dortland B, Miedema M, van Alphen A, Kistler WM, Jaegle M, Koutsourakis M, Van Camp N, Verhoye M, van der Linden A, Kaverina I, Grosveld F, De Zeeuw Cl, Galjart N (2002) Targeted mutation of Cyln2 in the Williams syndrome critical region links CLIP-115 haploinsufficiency to neurodevelopmental abnormalities in mice. Nat Genet 32: 116-127

Hori K, Cholewa-Waclaw J, Nakada Y, Glasgow SM, Masui T, Henke RM, Wildner H, Martarelli B, Beres TM, Epstein JA, Magnuson MA, Macdonald RJ, Birchmeier C, Johnson JE (2008) A nonclassical bHLH Rbpj transcription factor complex is required for specification of GABAergic neurons independent of Notch signaling. Genes Devil: 166-178

Hoshino M, Nakamura S, Mori K, Kawauchi T, Terao M, Nishimura YV, Fukuda A, Fuse T, Matsuo N, Sone M, Watanabe M, Bito H, Terashima T, Wright CV, Kawaguchi Y, Nakao K, Nabeshima Y (2005) Ptfla, a bHLH transcriptional gene, defines GABAergic neuronal fates in cerebellum. Neuron 47: 201-213

245 Howald C, Merla G, Digilio MC, Amenta S, Lyle R, Deutsch S, Choudhury U, Bottani A, Antonarakis SE, Fryssira H, Dallapiccola B, Reymond A (2006) Two high throughput technologies to detect segmental aneuploidies identify new Williams-Beuren syndrome patients with atypical deletions. JMed Genet 43: 266-273

Hu SC, Chrivia J, Ghosh A (1999) Regulation of CBP-mediated transcription by neuronal calcium signaling. Neuron 22: 799-808

Huang J, Berger S (2008) The emerging field of dynamic lysine methylation of non-histone proteins. Curr Opin Genet Dev 18: 152 - 158

Huppke P, Gartner J (2005) Molecular diagnosis of Rett syndrome. J Child Neurol 20: 732-736

Impey S, Fong AL, Wang Y, Cardinaux JR, Fass DM, Obrietan K, Wayman GA, Storm DR, Soderling TR, Goodman RH (2002) Phosphorylation of CBP mediates transcriptional activation by neural activity and CaM kinase IV. Neuron 34: 235-244

Impey S, Mark M, Villacres EC, Poser S, Chavkin C, Storm DR (1996) Induction of CRE- mediated gene expression by stimuli that generate long-lasting LTP in area CA1 of the hippocampus. Neuron 16: 973-982

Inoue H, Nojima H, Okayama H (1990) High efficiency transformation of Escherichia coli with plasmids. Gene 96: 23-28

Iseki H, Takeda A, Andoh T, Takahashi N, Kurochkin IV, Yarmishyn A, Shimada H, Okazaki Y, Koyama I (2010) Human Arm protein lost in epithelial cancers, on chromosome X 1 (ALEX1) gene is transcriptionally regulated by CREB and Wnt/beta-catenin signaling. Cancer Sci 101: 1361-1366

Issa LL, Palmer SJ, Guven KL, Santucci N, Hodgson VRM, Popovic K, Joya JE, Hardeman EC (2006) MusTRD can regulate postnatal fiber-specific expression. Dev Biol 293: 104-115

Jackson TA, Taylor HE, Sharma D, Desiderio S, Danoff SK (2005) Vascular endothelial growth factor receptor-2: Counter-regulation by the transcription factors, TFII-1 and TFII-IRD1. J Biol Chem 280: 29856-29863

Jakobsson J, Cordero MI, Bisaz R, Groner AC, Busskamp V, Bensadoun JC, Cammas F, Losson R, Mansuy 1M, Sandi C, Trono D (2008) KAP1 -mediated epigenetic repression in the forebrain modulates behavioral vulnerability to stress. Neuron 60: 818-831

Jeltsch A (2006) Molecular enzymology of mammalian DNA methyltransferases. Curr Top Microbiol Immunol 301: 203-225

Jenuwein T, Allis CD (2001) Translating the Histone Code. Science 293: 1074-1080

Jenuwein T, Laible G, Dorn R. Reuter G (1998) SET domain proteins modulate chromatin domains in eu- and heterochromatin. Cell Mol Life Sci 54: 80-93

Jernigan TL, Bellugi U (1990) Anomalous brain morphology on magnetic resonance images in Williams syndrome and Down syndrome. Arch Neurol 47: 529-533

Ji Z, Degerny C, Vintonenko N, Deheuninck J, Foveau B, Leroy C, Coll J, Tulasne D, Baert J- L, Fafeur V (2006) Regulation of the Ets-1 transcription factor by sumoylation and ubiquitinylation. Oncogene 26: 395-406

246 Jiang W, Sordella R, Chen GC, Hakre S, Roy AL, Settleman J (2005) An FF domain-dependent protein interaction mediates a signaling pathway for growth factor-induced gene expression. Mol Cell 17: 23-35

Jiang Y, Jakovcevski M, Bharadwaj R, Connor C, Schroeder FA, Lin CL, Straubhaar J, Martin G, Akbarian S (2010) Setdbl histone methyltransferase regulates mood-related behaviors and expression of the NMDA receptor subunit NR2B. J Neurosci 30: 7152-7167

Jimenez D, Lopez-Mascaraque LM, FValverde F, De Carlos JA (2002) Tangential migration in neocortical development. Dev Biol 244: 155-169

Jin J, Arias EE, Chen J, Harper JW, Walter JC (2006) A family of diverse Cul4-Ddbl- interacting proteins includes Cdt2, which is required for S phase destruction of the replication factor Cdt 1. Mol Cell 23: 709-721

Johnson ES (2004) Protein modification by SUMO. Annu Rev Biochem 73: 355-382

Johnson ES, Gupta AA (2001) An E3-like factor that promotes SUMO conjugation to the yeast septins. Cell 106: 735-744

Johnson ES, Schwienhorst I, Dohmen RJ, Blobel G (1997) The ubiquitin-like protein Smt3p is activated for conjugation to other proteins by an Aoslp/Uba2p heterodimer. EMBO J 16: 5509- 5519

Johnson LB, Comeau M, Clarke KD (2001) Hyperacusis in Williams syndrome. J Otolaryngol 30: 90-92

Jones D, Crowe E, Stevens T, Candido EP (2001) Functional and phylogenetic analysis of the ubiquitylation system in : ubiquitin-conjugating enzymes, ubiquitin- activating enzymes, and ubiquitin-like proteins. Genome Biol 3: research0002.000l - research0002.0015

Jones EG (1981) Anatomy of cerebral cortex: Columnar input-output organization. In The cerebral cortex, Schmitt FO, Worden, F.G., and Dennis, S.G. (ed), pp 99-235. Cambridge, MA: MIT Press

Jones KL, Smith DW (1975) The Williams elfin facies syndrome. A new perspective. J Pediatr 86: 718-723

Jones W, Bellugi U, Lai Z, Chiles M, Reilly J, Lincoln A, Adolphs R (2000) Hypersociability in Williams syndrome. JCogn Neurosci 12: 30-46

Jones W, Hesslink J, Courchesne E, Dunean T, Matsuda K, Bellugi U (2002) Cerebellar abnormalities in infants and toddlers with Williams syndrome. Dev Med Child Neurol 44: 688- 694

Kaczmarek L, Chaudhuri A (1997) Sensory regulation of immediate-early gene expression in mammalian visual cortex: implications for functional mapping and neural plasticity. Brain Res Brain Res Rev 23: 237-256

Kamitani T, Kito K, Nguyen HP, Fukuda-Kamitani T, Yeh ETH (1998) Characterization of a second member of the Sentrin family of Ubiquitin-like proteins. J Biol Chem 273: 11349-11353

247 Kammerer RA, Frank S, Schulthess T, Landwehr R, Lustig A, Engel J (1999) Heterodimerization of a Functional GABAB Receptor Is Mediated by Parallel Coiled-Coil a- Helicest. Biochemistry 38: 13263-13269

Kandel ER, Schwartz JH, Jessell TM (1991) Principles of neural science, 3rd edn. New York: Elsevier.

Kangaspeska S, Stride B. Metivier R, Polycarpou-Schwarz M, Ibberson D, Carmouche RP, Benes V, Gannon F, Reid G (2008) Transient cyclical methylation of promoter DNA. Nature 452: 112-115

Kanno T, Kanno Y, Siegel RM, Jang MK, Lenardo MJ, Ozato K (2004) Selective recognition of acetylated histones by bromodomain proteins visualized in living cells. Mol Cell 13: 33-43

Karmiloff-Smith A, Grant J, Ewing S, Carette MJ. Metcalfe K, Donnai D, Read AP. Tassabehji M (2003) Using case study comparisons to explore genotype-phenotype correlations in Williams-Beuren syndrome. J Med Genet 40: 136-140

Kerscher O (2007) SUMO junction - what's your function? EMBO Rep 8: 550-555

Kim DW, Cheriyath V. Roy AL, Cochran BH (1998) TFI1-I enhances activation of the c-fos promoter through interactions with upstream elements. Mol Cell Biol 18: 3310-3320

Kim DW, Cochran BH (2000) Extracellular signal-regulated kinase binds to TF1I-I and regulates its activation of the c-fos promoter. Mol Cell Biol 20: 1140-1148

Kim DW, Cochran BH (2001) JAK2 activates TF1I-1 and regulates its interaction with extracellular signal-regulated kinase. Mol Cell Biol 21: 3387-3397

Kim JY, Casaccia P (2010) HDAC1 in axonal degeneration: A matter of subcellular localization. Cell Cycle 9: 3680-3684

Kim TK, Hemberg M, Gray JM, Costa AM, Bear DM, Wu J, Harmin DA, Laptewicz M, Barbara-Haley K, Kuersten S, Markenscoff-Papadimitriou E, Kuhl D, Bito H, Worley PF, Kreiman G, Greenberg ME (2010) Widespread transcription at neuronal activity-regulated enhancers. Nature 465: 182-187

Kimura A, Horikoshi M (1998) Tip60 acetylates six lysines of a specific class in core histones in vitro. Genes Cells 3: 789-800

Kitagawa H, Fujiki R, Yoshimura K, Oya H, Kato S (2011) Williams syndrome is an epigenome-regulator disease. Endorine J

Klein-Tasman BP, Mervis CB, Lord C, Phillips KD (2007) Socio-communicative deficits in young children with Williams Syndrome: Performance on the Autism Diagnostic Observation Schedule. Child Neuropsychology 13: 444 - 467

Kolb B, Whishaw IQ (2001) An introduction to brain and behaviour, New York: Worth.

Korpi ER. Sinkkonen ST (2006) GABA(A) receptor subtypes as targets for neuropsychiatric drug development. Pharmacol Ther 109: 12-32

248 Korte M, Carroll P, Wolf E, Brem G, Thoenen H, Bonhoeffer T (1995) Hippocampal long-term potentiation is impaired in mice lacking brain-derived neurotrophic factor. Proc Natl Acad Sci USA 92: 8856-8860

Korzus E, Rosenfeld MG, Mayford M (2004) CBP histone acetyltransferase activity is a critical component of memory consolidation. Neuron 42: 961-972

Kotaja N, Karvonen U, Janne OA, Palvimo JJ (2002) P1AS proteins modulate transcription factors by functioning as SUMO-1 ligases. Mol Cell Biol 22: 5222-5234

Kouzarides T (2007) Chromatin modifications and their function. Cell 128: 693-705

Kriegstein AR, Noctor SC (2004) Patterns of neuronal migration in the embryonic cortex. Trends Neurosci 27: 392-399

Kruijer W, Cooper JA, Hunter T, Verma 1M (1984) Platelet-derived growth factor induces rapid but transient expression of the c-fos gene and protein. Nature 312: 711-716

Kuner R, Kohr G, Griinewald S, Eisenhardt G, Bach A, Kornau H-C (1999) Role of Heteromer Formation in GABAB Receptor Function. Science 283: 74-77

Kurochkin IV, Yonemitsu N, Funahashi SI, Nomura H (2001) ALEX1, a novel human armadillo repeat protein that is expressed differentially in normal tissues and carcinomas. Biochem Biophys Res Comm 280: 340-347

Kusserow H, Davies B, Hortnagl H, Voigt I, Stroh T, Bert B. Deng DR, Fink H, Veh RW, Theuring F (2004) Reduced anxiety-related behaviour in transgenic mice overexpressing serotonin 1A receptors. Brain Res Mol Brain Res 129: 104-1 16

Lachner M, Jenuwein T (2002) The many faces of histone lysine methylation. Curr Opin Cell Biol 14: 286-298

Lachner M, O'Carroll D, Rea S, Mechtler K, Jenuwein T (2001) Methylation of histone H3 lysine 9 creates a binding site for HP1 proteins. Nature 410: 116-120

Lai CS, Fisher SE, Hurst JA, Vargha-Khadem F, Monaco AP (2001) A forkhead-domain gene is mutated in a severe speech and language disorder. Nature 413: 519-523

Landschulz WH, Johnson PF, McKnight SL (1988) The leucine zipper: a hypothetical structure common to a new class of DNA binding proteins. Science 240: 1759-1764

Laws G, Bishop D (2004) Pragmatic language impairment and social deficits in Williams syndrome: a comparison with Down's syndrome and specific language impairment. Int J Lang Commun Disord 39: 45-64

Lebeurrier N, Liot G, Lopez-Atalaya JP, Orset C, Fernandez-Monreal M, Sonderegger P, Ali C, Vivien D (2005) The brain-specific tissue-type plasminogen activator inhibitor, neuroserpin, protects neurons against excitotoxicity both in vitro and in vivo. Mol Cell Neurosci 30: 552-558

Lee J, Zhou P (2007) DCAFs, the missing link of the CUL4-DDB1 ubiquitin ligase. Mol Cell 26: 775-780

249 Lee JE, Hollenberg SM, Snider L, Turner DL, Lipnick N, Weintraub H (1995) Conversion of Xenopus ectoderm into neurons by NeuroD, a basic helix-loop-helix protein. Science 268: 836- 844

Lee SK, Pfaff SL (2001) Transcriptional networks regulating neuronal identity in the developing spinal cord. Nat Neurosci 4 Suppl: 1183-1191

Lennon PA. Cooper ML, Peiffer DA, Gunderson KL, Patel A, Peters S, Cheung SW, Bacino CA (2007) Deletion of 7q31.1 supports involvement of FOXP2 in language impairment: clinical report and review. Am J Med Genet A 143A: 791-798

Leonard S, Gault J, Hopkins J, Logel J, Vianzon R, Short M, Drebing C, Berger R, Venn D, Sirota P, Zerbe G, Olincy A, Ross RG, Adler LE, Freedman R (2002) Association of promoter variants in the a7-nicotinic acetylcholine receptor subunit gene with an inhibitory deficit found in schizophrenia. Arch Gen Psychiatry 59: 1085-1096

Lerma J (2006) Kainate receptor physiology. Curr Opin Pharmacol 6: 89-97

Lesch KP (2005) Genetic alterations of the murine serotonergic gene pathway: the neurodevelopmental basis of anxiety. Handh Exp Pharmacol: 71-112

Levenson JM, Roth TL, Lubin FD, Miller CA, Huang IC, Desai P, Malone LM, Sweatt JD (2006) Evidence that DNA (cytosine-5) methyltransferase regulates synaptic plasticity in the hippocampus../ Biol Chem 281: 15763-15773

Levitin DJ, Cole K, Lincoln A, Bellugi U (2005) Aversion, awareness, and attraction: investigating claims of hyperacusis in the Williams syndrome phenotype. J Child Psychol Psychiatry 46: 514-523

Levitin DJ, Menon V, Schmitt JE, Eliez S, White CD, Glover GH, Kadis J, Korenberg JR, Bellugi U, Reiss AL (2003) Neural correlates of auditory perception in Williams syndrome: An tIVIRI study. Neuroimage 18: 74-82

Levitt P, Barbe MF, Eagleson KL (1997) Patterning and specification of the cerebral cortex. Annu Rev Neurosci 20: 1 -24

Levitt P, Eagleson KL, Powell EM (2004) Regulation of neocortical interneuron development and the implications for neurodevelopmental disorders. Trends Neurosci 27: 400-406

Levy D, Kuo AJ, Chang Y, Schaefer U, Kitson C, Cheung P, Espejo A, Zee BM, Liu CL, Tangsombatvisit S, Tennen Rl, Kuo AY, Tanjing S, Cheung R, Chua KF, Utz PJ, Shi X, Prinjha RK, Lee K, Garcia BA, Bedford MT, Tarakhovsky A, Cheng X, Gozani O (2011) Lysine methylation of the NF-kappaB subunit RelA by SETD6 couples activity of the histone methyltransferase GLP at chromatin to tonic repression of NF-kappaB signaling. Nat Immunol 12: 29-36

Lewandowski NM, Small SA (2005) Brain microarray: finding needles in molecular haystacks. J Neurosci 25: 10341-10346

Leyfer O, Woodruff-Borden J, Mervis CB (2009) Anxiety Disorders in Children with Williams Syndrome, Their Mothers, and Their Siblings: Implications for the Etiology of Anxiety Disorders. J Neurodev Disord 1: 4-14

250 Leyfer OT, Woodruff-Borden J, Klein-Tasman BP, Fricke JS, Mervis CB (2006) Prevalence of psychiatric disorders in 4 to 16-year-olds with Williams syndrome. Am J Med Genet B Neuropsychiatr Genet 141B: 615-622

Li DY, Faury G, Taylor DG, Davis EC, Boyle WA, Mecham RP. Stenzel P, Boak B, Keating MT (1998) Novel arterial pathology in mice and humans hemizygous for elastin. J Clin Invest 102: 1783-1787

Li HH, Roy M, Kuscuoglu U, Spencer CM, Hahn B, Harrison KC, Bayle JH, Splendore A, Ding F, Meltzer LA, Wright E, Paylor R, Deisseroth K, Francke U (2009) Induced chromosome deletions cause hypersociability and other features of Williams-Beuren syndrome in mice. EMBO Mol Med 1: 50-65

Lim DA, Huang YC, Swigut T, Mirick AL, Garcia-Verdugo JM, Wysocka J, Ernst P, Alvarez- Buylla A (2009) Chromatin remodelling factor Mil 1 is essential for neurogenesis from postnatal neural stem cells. Nature 458: 529-533

Lin DY, Huang YS, Jeng JC, Kuo HY, Chang CC, Chao TT, Ho CC, Chen YC, Lin TP, Fang HI, Hung CC, Suen CS, Hwang MJ, Chang KS, Maul GG, Shih HM (2006) Role of SUMO- interacting motif in Daxx SUMO modification, subnuclear localization, and repression of sumoylated transcription factors. Mol Cell 24: 341-354

Lin Y, Bloodgood BL, Hauser JL, Lapan AD, Koon AC, Kim T-K, Hu LS, Malik AN, Greenberg ME (2008) Activity-dependent regulation of inhibitory synapse development by Npas4. Nature 455: 1198-1204

Liu A, Niswander LA (2005) Bone morphogenetic protein signalling and vertebrate nervous system development. Nat Rev Neurosci 6: 945-954

Liu CL, Kaplan T, Kim M, Buratowski S, Schreiber SL, Friedman N, Rando OJ (2005) Single- nucleosome mapping of histone modifications in S. cerevisiae. PLoS Biol 3: e328

Liu Q, Wong-Riley MT (2006) Developmental changes in the expression of GABAA receptor subunits alphal, alpha2, and alpha3 in brain stem nuclei of rats. Brain Res 1098: 129-138

Lobo MK, Karsten SL, Gray M, Geschwind DH, Yang XW (2006) FACS-array profiling of striatal projection neuron subtypes in juvenile and adult mouse brains. Nat Neurosci 9: 443-452

Loebrich S, Nedivi E (2009) The function of activity-regulated genes in the nervous system. Physiol Rev 89: 1079-1103

Lonze BE, Ginty DD (2002) Function and regulation of CREB family transcription factors in the nervous system. Neuron 35: 605-623

Lories RJ, Luyten FP (2005) Bone morphogenetic protein signaling in joint homeostasis and disease. Cytokine Growth Factor Rev 16: 287-298

Lucena J, Pezzi S, Aso E, Valero MC, Carreiro C, Dubus P, Sampaio A, Segura M, Barthelemy I, Zindel MY, Sousa N, Barbero JL, Maldonado R, Perez-Jurado LA, Campuzano V (2010) Essential role of the N-terminal region of TFII-I in viability and behavior. BMC Med Genet 11: 61

Lujan R. Shigemoto R, Lopez-Bendito G (2005) Glutamate and GABA receptor signalling in the developing brain. Neuroscience 139: 567-580

251 Ma DK, Jang MH, Guo JU, Kitabatake Y, Chang ML, Pow-Anpongkul N, Flavell RA, Lu B, Ming GL, Song H (2009) Neuronal activity-induced Gadd45b promotes epigenetic DNA demethylation and adult neurogenesis. Science 323: 1074-1077

Ma Q (2006) Transcriptional regulation of neuronal phenotype in mammals. J Physiol 575: 379-387

Mahajan R, Delphin C, Guan T, Gerace L, Melchior F (1997) A small ubiquitin-related polypeptide involved in targeting RanGAPl to nuclear pore complex protein RanBP2. Cell 88: 97-107

Makeyev AV, Erdenechimeg L, Mungunsukh O, Roth JJ, Enkhmandakh B, Ruddle FH, Bayarsaihan D (2004) GTF2IRD2 is located in the Williams-Beuren syndrome critical region 7ql 1.23 and encodes a protein with two TFII-I-like helix-loop-helix repeats. Proc Natl Acad Sci USA 101: 11052-11057

Makhnevych T, Sydorskyy Y, Xin X, Srikumar T, Vizeacoumar FJ, Jeram SM, Li Z, Bahr S, Andrews BJ, Boone C, Raught B (2009) Global Map of SUMO Function Revealed by Protein- Protein Interaction and Genetic Networks. Mol Cell 33: 124-135

Malenka RC, Bear MF (2004) LTP and LTD: an embarrassment of riches. Neuron 44: 5-21

Malinow R, Malenka RC (2002) AMPA receptor trafficking and synaptic plasticity. Annu Rev Neurosci 25: 103-126

Mammoto A, Connor KM, Mammoto T, Yung CW, Huh D, Aderman CM, Mostoslavsky G, Smith LE, Ingber DE (2009) A mechanosensitive transcriptional mechanism that controls angiogenesis. Nature 457: 1103-1108

Marfori M, Mynott A, Ellis JJ, Mehdi AM, Saunders NF, Curmi PM, Forwood JK, Boden M, Kobe B (2010) Molecular basis for specificity of nuclear import and prediction of nuclear localization. Biochim Biophys Acta

Marin O, Rubenstein JLR (2001) A long, remarkable journey: Tangential migration in the telencephalon. Nat Rev Neurosci 2: 780-790

Markram H, Toledo-Rodriguez M, Wang Y, Gupta A, Silberberg G, Wu C (2004) Interneurons of the neocortical inhibitory system. Nat Rev Neurosci 5: 793-807

Martens JH, O'Sullivan RJ, Braunschweig U, Opravil S, Radolf M, Steinlein P, Jenuwein T (2005) The profile of repeat-associated histone lysine methylation states in the mouse epigenome. EMBO J 24: 800-812

Martin S, Wilkinson KA, Nishimune A, Henley JM (2007) Emerging extranuclear roles of protein SUMOylation in neuronal function and dysfunction. Nat Rev Neurosci 8: 948-959

Martinowich K, Hattori D, Wu H, Fouse S, He F, Hu Y, Fan G, Sun YE (2003) DNA methylation-related chromatin remodeling in activity-dependent BDNF gene regulation. Science 302: 890-893

Mass E, Belostoky L (1993) Craniofacial morphology of children with Williams syndrome. Cleft Palate Craniofac J30: 343-349

252 Matsuoka S, Ballif BA, Smogorzewska A, McDonald ER, 3rd, Hurov KE, Luo J, Bakalarski CE, Zhao Z, Solimini N, Lerenthal Y, Shiloh Y, Gygi SP, Elledge SJ (2007) ATM and ATR substrate analysis reveals extensive protein networks responsive to DNA damage. Science 316: 1160-1166

Matunis MJ, Coutavas E, Blobel G (1996) A novel ubiquitin-like modification modulates the partitioning of the Ran-GTPase-activating protein RanGAPl between the cytosol and the nuclear pore complex. J Cell Biol 135: 1457-1470

Maze I, Covington HE, 3rd, Dietz DM, LaPlant Q, Renthal W, Russo SJ, Mechanic M, Mouzon E, Neve RL, Haggarty SJ, Ren Y, Sampath SC, Hurd YL, Greengard P, Tarakhovsky A, Schaefer A, Nestler EJ (2010) Essential role of the histone methyltransferase G9a in cocaine- induced plasticity. Science 327: 213-216

McConnell S (1995) Strategies for the generation of neuronal diversity in the developing central nervous system. JNeurosci 15: 6987-6998

McKinsey TA, Zhang CL, Olson EN (2002) MEF2: a calcium-dependent regulator of cell division, differentiation and death. Trends Biochem Sci 27: 40-47

McRory JE, Rehak R, Simms B, Doering CJ, Chen L, Hermosilla T, Duke C, Dyck R, Zamponi GW (2008) Syntaxin 1A is required for normal in utero development. Biochem Biophys Res Commun 375: 372-377

Meluh PB, Koshland D (1995) Evidence that the MIF2 gene of Saccharomyces cerevisiae encodes a centromere protein with homology to the mammalian centromere protein CENP-C. Mol Biol Cell 6: 793-807

Meng Y, Zhang Y, Tregoubov V, Janus C, Cruz L, Jackson M, Lu WY, MacDonald JF, Wang JY, Falls DL, Jia Z (2002) Abnormal spine morphology and enhanced LTP in LIMK-I knockout mice. Neuron 35: 121-133

Merla G, Brunetti-Pierri N, Micale L, Fusco C (2010) Copy number variants at Williams- Beuren syndrome 7ql 1.23 region. Hum Genet 128: 3-26

Merla G, Howald C, Henrichsen CN, Lyle R, Wyss C, Zabot M, Antonarakis S, Reymond A (2006) Submicroscopic deletion in patients with Williams-Beuren syndrome influences expression levels of the nonhemizygous flanking genes. Am J Hum Genet 70: 332-341

Mervis CB, Becerra AM (2007) Language and communicative development in Williams syndrome. Ment Retard Dev Disabil Res Rev 13: 3-15

Mervis CB, Robinson BF (2000) The Williams syndrome cognitive profile. Brain Cogn 44: 604-628

Metcalfe K, Rucka AK, Smoot L, Hofstadler G, Tuzler G, McKeown P, Siu V, Rauch A, Dean J, Dennis N, Ellis 1, Reardon W, Cytrynbaum C, Osborne L, Yates JR, Read AP, Donnai D, Tassabehji M (2000) Elastin: mutational spectrum in supravalvular aortic stenosis. Eur J Hum Genet 8: 955-963

Metivier R, Gallais R, Tiffoche C, Le Peron C, Jurkowska RZ, Carmouche RP, Ibberson D. Barath P, Demay F, Reid G, Benes V, Jeltsch A, Gannon F, Salbert G (2008) Cyclical DNA methylation of a transcriptionally active promoter. Nature 452: 45-50

253 Meyer-Lindenberg A, Hariri AR, Munoz KE, Mervis CB, Mattay VS, Morris CA, Berman KF (2005a) Neural correlates of genetically abnormal social cognition in Williams syndrome. Nat Neurosci 8: 991-993

Meyer-Lindenberg A, Kohn P, Mervis CB, Kippenhan JS, Olsen RK, Morris CA, Berman KF (2004) Neural basis of genetically determined visuospatial construction deficit in Williams syndrome. Neuron 43: 623-631

Meyer-Lindenberg A, Mervis CB, Sarpal D, Koch P, Steele S, Kohn P, Marenco S, Morris CA, Das S, Kippenhan S, Mattay VS, Weinberger DR, Berman KF (2005b) Functional, structural, and metabolic abnormalities of the hippocampal formation in Williams syndrome. J Clin Invest 115: 1888-1895

Michan S, Sinclair D (2007) Sirtuins in mammals: insights into their biological function. Biochem J 404: 1-13

Millar CB, Grunstein M (2006) Genome-wide patterns of histone modifications in yeast. Nat Rev Mol Cell Biol 7: 657-666

Millar CB, Xu F, Zhang K, Grunstein M (2006) Acetylation of H2AZ Lys 14 is associated with genome-wide gene activity in yeast. Genes Dev 20: 711-722

Miller B, Chou L, Finlay BL (1993) The early development of thalamocortical and corticothalamic projections. J Comp Neurol 335: 16-41

Miller CA, Sweatt JD (2007) Covalent modification of DNA regulates memory formation. Neuron 53: 857-869

Mills D, Alvarez TD, St George M. Appelbaum LG, Bellugi U, Neville H (2000) Electrophysiological studies of face processing in Williams syndrome. J Cogn Neurosci 12: 47- 64

Min J, Feng Q, Li Z, Zhang Y, Xu RM (2003) Structure of the catalytic domain of human DOT1L, a non-SET domain nucleosomal histone methyltransferase. Cell 112: 711-723

Ming GL, Song H (2005) Adult neurogenesis in the mammalian central nervous system. Annu Rev Neurosci 28: 223-250

Mobbs D, Garrett AS, Menon V, Rose FE, Bellugi U, Reiss AL (2004) Anomalous brain activation during face and gaze processing in Williams syndrome. Neurology> 62: 2070-2076

Moroianu J, Hijikata M, Blobel G, Radu A (1995) Mammalian karyopherin alpha 1 beta and alpha 2 beta heterodimers: alpha 1 or alpha 2 subunit binds nuclear localization signal and beta subunit interacts with peptide repeat-containing nucleoporins. Proc Natl Acad Sci USA 92: 6532-6536

Morris C, Dilts C, Demsey S, Leonard C, Blackburn B (1988) The natural history of Williams syndrome: physical characteristics. JPediatr 113: 318-326

Morris CA (2006) Genotype-phenotype correlations. In Williams syndrome, C.A. M, Lenhoff HM, Wang PP (eds), pp 59-82. Baltimore, MD Johns Hopkins University Press

Morris CA, Mervis CB (2000) Williams Syndrome and related disorders. Annu Rev Genomics Hum Genet 1: 461-484

254 Morris CA, Mervis CB, Hobart HH, Gregg RG, Bertrand J, Ensing GJ, Sommer A, Moore CA, Hopkin RJ, Spallone PA, Keating MT, Osborne L, Kimberley KW, Stock AD (2003) GTF2I hemizygosity implicated in mental retardation in Williams syndrome: Genotype-phenotype analysis of five families with deletions in the Williams syndrome region. Am J Med Genet 123A: 45-59

Morris CA, Thomas IT, Greenberg F (1993) Williams Syndrome: autosomal dominant inheritance. Am J Med Genet 47: 478-481

Mou Z, Tapper AR, Gardner PD (2009) The armadillo repeat-containing protein, ARMCX3, physically and functionally interacts with the developmental regulatory factor SoxlO. J Biol Chem 284: 13629-13640

Mukhopadhyay D, Dasso M (2007) Modification in reverse: the SUMO proteases. Trends Biochem Sci 32: 286-295

Munoz KE, Meyer-Lindenberg A, Hariri AR, Mervis CB. Mattay VS, Morris CA, Berman KF (2010) Abnormalities in neural processing of emotional stimuli in Williams syndrome vary according to social vs. non-social content. Neuroimage 50: 340-346

Nagy A, Rossant J, Nagy R, Abramow-Newerly W, Roder JC (1993) Derivation of completely cell culture-derived mice from early-passage embryonic stem cells. Proc Natl Acad Sci USA 90: 8424-8428

Naidu S (1997) Rett syndrome. A disorder affecting early brain growth. Ann Neurol 42: 3-10

Nakayama T, Matsuoka R, Kimura M, Hirota H, Mikoshiba K, Shimizu Y, Shimizu N, Akagawa K (1998) Hemizygous deletion of the HPC-l/syntaxin 1A gene (STX1A) in patients with Williams syndrome. Cytogenet Cell Genet 82: 49-51

Nazzi T, Paterson S, Karmiloff-Smith A (2003) Early word segmentation by infants and toddlers with Williams syndrome. Infancy 4: 251 - 271

Nichols S, Jones W, Roman MJ, Wulfeck B, Delis DC, Reilly J, Bellugi U (2004) Mechanisms of verbal memory impairment in four neurodevelopmental disorders. Brain Lang 88: 180-189

Niehrs C (2009) Active DNA demethylation and DNA repair. Differentiation 77: 1-11

Nigam A, Samuel PR (1994) Hyperacusis and Williams syndrome. J Laryngol Otol 108: 494- 496

Novina CD, Cheriyath V, Roy AL (1998) Regulation of TFII-1 activity by phosphorylation. J Biol Chem 273: 33443-33448

Novina CD, Kumar S, Bajpai U, Cheriyath V, Zhang K, Pillai S, Wortis HH, Roy AL (1999) Regulation of nuclear localization and transcriptional activity of TFII-1 by Bruton's tyrosine kinase. Mol Cell Biol 19: 5014-5024

O'Mahoney J, Guven KL, Joya JE. Robinson S, Wade RP, Hardeman EC (1998) Identification of a novel slow-muslce-fiber enhancer binding protein, MusTRDl. Mol Cell Biol 18: 6641- 6652

255 Obst K, Wahle P (1997) Phenotype specification of cortical neurons during a period of molecular plasticity. Eur JNeurosci 9: 2571-2580

Odho Z, Southall SM, Wilson JR (2010) Characterization of a novel WDR5-binding site that recruits RbBP5 through a conserved motif to enhance methylation of histone H3 lysine 4 by mixed lineage leukemia protein-1. J Biol Chem 285: 32967-32976

Ogris W, Lehner R, Fuchs K, Frtmuller B, Hoger H, Homanics GE, Sieghart W (2006) Investigation of the abundance and subunit composition of GABAa receptor subtypes in the cerebellum of a 1-subunit-deficient mice. J Neurochem 96: 136-147

Ohazama A, Sharpe PT (2007) TFII-I gene family during tooth development: candidate genes for tooth anomalies in Williams syndrome. Dev Dyn 236: 2884-2888

Okuno H (2010) Regulation and function of immediate-early genes in the brain: Beyond neuronal activity markers. Neurosci Res

Okura T, Gong L, Kamitani T, Wada T, Okura I, Wei CF, Chang HM, Yeh ET (1996) Protection against Fas/APO-1- and tumor necrosis factor-mediated cell death by a novel protein, sentrin. J Immunol 157: 4277-4281

Olave I, Wang W, Xue Y, Kuo A, Crabtree GR (2002) Identification of a polymorphic, neuron- specific chromatin remodeling complex. Genes Dev 16: 2509-2517

Ollendorff V, Guasch G, Isnardon D, Galindo R. Birnbaum D, Pebusque MJ (1999) Characterization of F1M-FGFR1, the fusion product of the myeloproliferative disorder- associated t(8; 13) translocation. J Biol Chem 274: 26922-26930

Olsen JV, Blagoev B, Gnad F, Macek B, Kumar C, Mortensen P, Mann M (2006) Global, in vivo, and site-specific phosphorylation dynamics in signaling networks. Cell 127: 635-648

Osborne LR (2008) Genomic rearrangements in the spotlight. Nat Genet 40: 6-7

Osborne LR (2010) Animal models of Williams syndrome. Am J Med Genet C Semin Med Genet 154C: 209-219

Osborne LR, Campbell T, Daradich A, Scherer SW, Tsui LC (1999) Identification of a putative transcription factor gene (WBSCR11) that is commonly deleted in Williams-Beuren syndrome. Genomics 57: 279-284

Osborne LR, Li M, Pober B, Chitayat D, Bodurtha J, Mandel A, Costa T, Grebe T, Cox S, Tsui LC, Scherer SW (2001) A 1.5 million-base pair inversion polymorphism in families with Williams-Beuren syndrome. Nat Genet 29: 321-325

Osborne LR, Mervis CB (2007) Rearrangements of the Williams-Beuren syndrome locus: molecular basis and implications for speech and language development. Expert Rev Mol Med 9: 1-16

Ouyang J, Gill G (2009) SUMO engages multiple to regulate chromatin structure and transcription. Epigenetics 4: 1-5

Ouyang J, Valin A, Gill G (2009) Regulation of transcription factor activity by SUMO modification. In SUMO Protocols, pp 141-152. Humana Press

256 Owerbach D, McKay EM, Yeh ET, Gabbay KEI, Bohren KM (2005) A proline-90 residue unique to SUMO-4 prevents maturation and sumoylation. Biochem Biophys Res Commun 337: 517-520

Palmer SJ, Santucci N, Widagdo J, Bontempo SJ, Taylor KM, Tay ES, Hook J, Lemckert F, Gunning PW, Hardeman EC (2010) Negative autoregulation of GTF2IRD1 in Williams-Beuren Syndrome via a novel DNA binding mechanism. J Biol Chem 285: 4715-4724

Palmer SJ, Tay E, Santucci N, Cue Bach T, Hook J, Lemckert F, Jamieson R, Gunnning P, Hardeman E (2007) Expression of Gtf2ird 1, the Williams syndrome-associated gene, during mouse development. Gene Expr Patterns 7: 396-404

Palmini A, Andermann F, Olivier A, Tampieri D, Robitaille Y, Andermann E, Wright G (1991) Focal neuronal migration disorders and intractable partial epilepsy: A study of 30 patients. Ann Neurol 30: 741-749

Palvimo JJ (2007) PIAS proteins as regulators of small ubiquitin-related modifier (SUMO) modifications and transcription. Biochem Soc Trans 35: 1405-1408

Pan JW, Takahashi K (2005) Interdependence of N-acetyl aspartate and high-energy phosphates in healthy human brain. Ann Neurol 57: 92-97

Pankau R, Partsch C-J, Gosch A, Oppermann H, Wessel A (1992) Statural growth in Williams- Beuren syndrome. Eur JPediatr 151: 751-755

Park PJ (2009) ChlP-seq: advantages and challenges of a maturing technology. Nat Rev Genet 10: 669-680

Parks CL, Robinson PS, Sibille E, Shenk T, Toth M (1998) Increased anxiety of mice lacking the serotonin 1A receptor. Proc Natl Acad Sci USA 95: 10734-10739

Parras CM, Schuurmans C, Scardigli R, Kim J, Anderson DJ, Guillemot F (2002) Divergent functions of the proneural genes Mashl and Ngn2 in the specification of neuronal subtype identity. Genes Dev 16: 324-338

Pattyn A, Simplicio N, van Doorninck JH, Goridis C, Guillemot F, Brunet JF (2004) Ascll/Mashl is required for the development of central serotonergic neurons. Nat Neurosci 7: 589-595

Peleg S, Sananbenesi F, Zovoilis A, Burkhardt S, Bahari-Javan S, Agis-Balboa RC, Cota P, Wittnam JL, Gogol-Doering A, Opitz L, Salinas-Riester G, Dettenhofer M, Kang H, Farinelli L, Chen W, Fischer A (2010) Altered histone acetylation is associated with age-dependent memory impairment in mice. Science 328: 753-756

Peoples R, Franke Y. Wang YK, Perez-Jurado L, Paperna T, Cisco M, Francke U (2000) A physical map, including a BAC/PAC clone contig, of the Williams-Beuren syndrome deletion region at 7ql 1.23. Am J Hum Genet 66: 47-68

Perdomo J, Verger A, Turner J, Crossley M (2005) Role for SUMO Modification in Facilitating Transcriptional Repression by BKLF. Mol Cell Biol 25: 1549-1559

Perez-Villalba A, Teruel-Marti V, Ruiz-Torner A, Olucha-Bordonau F (2005) The effect of long context exposure on cued conditioning and c-fos expression in the rat forebrain. Behav Brain Res 161: 263-275

257 Perez Jurado L, Wang Y, Peoples R, Coloma A, Cruces J, Francke U (1998) A duplicated gene in the breakpoint regions of the 7q 11.23 Williams- Beuren syndrome deletion encodes the initiator binding protein TFII-I and BAP-135, a phosphorylation target of BTK. Hum Mol Genet 7: 325-334

Peters A, Kara DA (1985) The neuronal composition of area 17 of rat visual cortex. I. The pyramidal cells. J Comp Neurol 234: 218-241

Petrij F, Giles RH, Dauwerse HG, Saris JJ, Hennekam RC, Masuno M, Tommerup N, van Ommen GJ, Goodman RH, Peters DJ, et al. (1995) Rubinstein-Taybi syndrome caused by mutations in the transcriptional co-activator CBP. Nature 376: 348-351

Petroski MD, Deshaies RJ (2005) Function and regulation of cullin-RING ubiquitin ligases. Nat Rev Mol Cell Biol 6: 9-20

Pette D (2001) Historical Perspectives: plasticity of mammalian skeletal muscle. J Appl Physiol 90: 1119-1124

Philofsky A, Fidler DJ, Hepburn S (2007) Pragmatic language profiles of school-age children with autism spectrum disorders and Williams syndrome. Am J Speech Lang Pathol 16: 368-380

Pichler A, Gast A, Seeler JS, Dejean A, Melchior F (2002) The nucleoporin RanBP2 has SUMOl E3 ligase activity. Cell 108: 109-120

Pin CL, Hrycyshyn AW, Rogers KA, Rushlow WJ, Merrifield PA (2002) Embryonic and fetal rat myoblasts form different muscle fiber types in an ectopic in vivo environment. Dev Dyn 224: 253-266

Plath K, Fang J, Mlynarczyk-Evans SK, Cao R, Worringer KA, Wang H, de la Cruz CC, Otte AP, Panning B, Zhang Y (2003) Role of histone H3 lysine 27 methylation in X inactivation. Science 300: 131-135

Polly P, Haddadi LM, Issa LL, Subramaniam N, Palmer SJ, Tay ES, Hardeman EC (2003) hMusTRDlal represses MEF2 activation of the troponin 1 slow enhancer. J Biol Chem 278: 36603-36610

Porter MA, Coltheart M, Langdon R (2007) The neuropsychological basis of hypersociability in Williams and Down syndrome. Neuropsychologia 45: 2839-2849

Porter MA, Shaw TA, Marsh PJ (2010) An unusual attraction to the eyes in Williams-Beuren syndrome: a manipulation of facial affect while measuring face scanpaths. Cogn Neuropsychiatry 15: 505-530

Pothuizen HH, Davies M, Albasser MM, Aggleton JP, Vann SD (2009) Granular and dysgranular retrosplenial cortices provide qualitatively different contributions to spatial working memory: evidence from immediate-early gene imaging in rats. Eur JNeurosci 30: 877-888

Proulx E, Young EJ, Osborne LR, Lambe EK (2010) Enhanced prefrontal serotonin 5-HT(lA) currents in a mouse model of Williams-Beuren syndrome with low innate anxiety. J Neurodev Disord 2: 99-108

Purves D, Augustine GJ. Fitzpatrick D. Hall WC, Lamantia AS, McNamara JO, Williams SM (2004) Neuroscience, 3rd edn.: Sinauer Associates.

258 Qian C, Zhou MM (2006) SET domain protein lysine methyltransferases: Structure, specificity and catalysis. Cell Mol Life Sci 63: 2755-2763

Rae C, KarmiIoff-Smith A, Lee MA, Dixon RM, Grant J, Blamire AM, Thompson CH, Styles P, Radda GK (1998) Brain biochemistry in Williams syndrome: Evidence for a role of the cerebellum in cognition? Neurology 51: 33-40

Raffa GD, Cenci G, Siriaco G, Goldberg ML, Gatti M (2005) The putative Drosophila transcription factor woe is required to prevent telomeric fusions. Mol Cell 20: 821-831

Rai K, Huggins IJ, James SR, Karpf AR, Jones DA, Cairns BR (2008) DNA demethylation in involves the coupling of a deaminase, a glycosylase, and gadd45. Cell 135: 1201-1212

Ramboz S, Oosting R, Amara DA, Rung HF, Blier P, Mendelsohn M, Mann JJ, Brunner D, Hen R (1998) Serotonin receptor 1A knockout: an animal model of anxiety-related disorder. Proc Natl Acad Sci USA 95: 14476-14481

Rea S, Eisenhaber F, O'Carroll D, Strahl BD, Sun ZW, Schmid M, Opravil S, Mechtler K, Ponting CP, Allis CD, Jenuwein T (2000) Regulation of chromatin structure by site-specific histone H3 methyltransferases. Nature 406: 593-599

Reilly J, Losh M, Bellugi U, Wulfeck B (2004) "Frog, where are you?" Narratives in children with specific language impairment, early focal brain injury, and Williams syndrome. Brain Lang 88: 229-247

Reinke H, Horz W (2003) Histones are first hyperacetylated and then lose contact with the activated PH05 promoter. Mol Cell 11: 1599-1607

Reiss AL, Eckert MA, Rose FE, Karchemskiy A, Kesler S, Chang M, Reynolds MF, Kwon H, Galaburda A (2004a) An experiment of nature: Brain anatomy parallels cognition and behavior in Williams syndrome. JNeurosci 24: 5009-5015

Reiss AL, Eckert MA, Rose FE, Karchemskiy A, Kesler S, Chang M, Reynolds MF, Kwon H, Galaburda A (2004b) An experiment of nature: Brain anatomy parallels cognition and behavior in Williams syndrome. J Neurosci 24: 5009-5015

Reiss AL, Stephan EJ, Schmitt E, E. S, Lai Z, Jones W, Bellugi U (2000) Neuroanatomy of Williams syndrome: A high resolution MRI study. JCogn Neurosci 12: 65-73

Riby DM, Doherty-Sneddon G, Bruce V (2008) Atypical unfamiliar face processing in Williams syndrome: what can it tell us about typical familiarity effects? Cogn Neuropsychiatry 13: 47-58

Riby DM, Doherty-Sneddon G, Bruce V (2009) The eyes or the mouth? Feature salience and unfamiliar face processing in Williams syndrome and autism. Q J Exp Psychol (Colchester) 62: 189-203

Riccio A (2010) Dynamic epigenetic regulation in neurons: enzymes, stimuli and signaling pathways. Nat Neurosci 13: 1330-1337

Richter M (2009) Setd6: A novel histone lysine mono-me thy l transfer as e, Saarbrucken, Germany: Sudwestdeutscher Verlag fur Hoschschulschriften.

259 Riggleman B, Wieschaus E, Schedl P (1989) Molecular analysis of the armadillo locus: uniformly distributed transcripts and a protein with novel internal repeats are associated with a Drosophila segment polarity gene. Genes Dev 3: 96-113

Ring C, Ogata S, Meek L, Song J, Ohta T, Miyazono K, Cho KW (2002) The role of a Williams-Beuren syndrome-associated helix-loop-helix domain-containing transcription factor in activin/nodal signaling. Genes Dev 16: 820-835

Robertson KD (2005) DNA methylation and human disease. Nat Rev Genet 6: 597-610

Robinson TE, Kolb B (2004) Structural plasticity associated with exposure to drugs of abuse. Neuropharmacology 47 Suppl 1: 33-46

Rodriguez MS, Dargemont C, Hay RT (2001) SUMO-1 conjugation in vivo requires both a consensus modification motif and nuclear targeting. J Biol Chem 276: 12654-12659

Roelfsema JH, Peters DJ (2007) Rubinstein-Taybi syndrome: clinical and molecular overview. Expert Rev Mol Med 9: 1-16

Rogers S, Wells R, Rechsteiner M (1986) Amino acid sequences common to rapidly degraded proteins: the PEST hypothesis. Science 234: 364-368

Roy AL (1997) Cloning of an Inr and E-box binding protein, TFII-I that interacts physically and functionally with USF1. EMBO J16: 7091-7104

Roy AL (2007) Signal-induced functions of the transcription factor TFI1-1. Biochim Biophys Acta 1769: 613-621

Roy AL, Carruthers C, Gutjahr T, Roeder RG (1993a) Direct role for Myc in transcription initiation mediated by interactions with TFII-I. Nature 365: 359-361

Roy AL, Malik S, Meisterernst M, Roeder RG (1993b) An alternative pathway for transcription initiation involving TFII-I. Nature 365: 355-359

Roy AL, Meisterernst M, Pognonec P, Roeder RG (1991) Cooperative interaction of an initiator-binding transcription initiation factor and the helix-loop-helix activator USF. Nature 354: 245-248

Ruthenburg AJ, Wang W, Graybosch DM, Li H, Allis CD, Patel DJ, Verdine GL (2006) Histone H3 recognition and presentation by the WDR5 module of the MLL1 complex. Nat Struct Mol Biol 13: 704-712

Rytinki MM, Kaikkonen S, Pehkonen P, Jaaskelainen T, Palvimo JJ (2009) PIAS proteins: pleiotropic interactors associated with SUMO. Cell Mol Life Sci 66: 3029-3041

Sacristan C, Schattgen SA, Berg LJ, Bunnell SC, Roy AL, Rosenstein Y (2009) Characterization of a novel interaction between transcription factor TFII-I and the inducible tyrosine kinase in T cells. Eur J Immunol 39: 2584-2595

Sahalizi A, Bilimoria PM, Stegmuller J, Gaudilliere B. Yang Y, Shuai K, Bonni A (2007) PlASx is a MEF2 SUMO E3 ligase that promotes postsynaptic dendritic morphogenesis. The Journal of Neuroscience 27: 10037-10046

260 Saitoh H, Hinchey J (2000) Functional heterogeneity of small ubiquitin-related protein modifiers SUMO-1 versus SUMO-2/3. J Biol Chem 275: 6252-6258

Sakurai T, Dorr NP, Takahashi N, Mclnnes LA. Elder GA, Buxbaum JD (2010) Haploinsufficiency of Gtf2i, a gene deleted in Williams Syndrome, leads to increases in social interactions. Autism Res

Sampson DA, Wang M, Matunis MJ (2001) The small ubiquitin-1 ike modifier-1 (SUMO-1) consensus sequence mediates Ubc9 binding and is essential for SUMO-1 modification. J Biol Chem 276: 21664-21669

Sandberg R, Yasuda R, Pankratz DG, Carter TA, Del Rio JA, Wodicka L, Mayford M, Lockhart DJ, Barlow C (2000) Regional and strain-specific gene expression mapping in the adult mouse brain. Proc Natl Acad Sci USA 97: 11038-11043

Saneyoshi T, Fortin DA, Soderling TR (2010) Regulation of spine and synapse formation by activity-dependent intracellular signaling pathways. Curr Opin Neurobiol 20: 108-115

Santos AR, Comprido D, Duarte CB (2010) Regulation of local translation at the synapse by BDNF. Prog Neurobiol 92: 505-516

Sapetschnig A, Rischitor G, Braun H, Doll A, Schergaut M, Melchior F, Suske G (2002) Transcription factor Sp3 is silenced through SUMO modification by P1AS1. EMBO Jl\\ 5206- 5215

Sarimski K (1996) Specific eating and sleeping problems in Prader-Willi and Williams-Beuren syndrome. Child Care Health Dev 22: 143-150

Scannevin RH, Huganir RL (2000) Postsynaptic organization and regulation of excitatory synapses. Nat Rev Neurosci 1: 133-141

Schaefer A, Sampath SC, Intrator A, Min A, Gertler TS, Surmeier DJ, Tarakhovsky A, Greengard P (2009) Control of cognition and adaptive behavior by the GLP/G9a epigenetic suppressor complex. Neuron 64: 678-691

Schmidt D, Muller S (2002) Members of the PIAS family act as SUMO ligases for c-Jun and p53 and repress p53 activity. Proc Natl Acad Sci USA 99: 2872-2877

Schmitt JE, Eliez S, Bellugi U, Reiss AL (2001) Analysis of cerebral shape in Williams syndrome. Arch Neurol 58: 283-287

Schotta G, Lachner M, Sarma K, Ebert A, Sengupta R, Reuter G, Reinberg D, Jenuwein T (2004) A silencing pathway to induce H3-K.9 and H4-K20 trimethylation at constitutive heterochromatin. Genes Dev 18: 1251-1262

Schubert C (2009) The genomic basis of the Williams-Beuren syndrome. Cell Mol Life Sci 66: 1178-1197

Schuurmans C, Armant O, Nieto M, Stenman JM, Britz O, Klenin N, Brown C, Langevin LM, Seibt J, Tang H, Cunningham JM, Dyck R, Walsh C, Campbell K, Polleux F, Guillemot F (2004) Sequential phases of cortical specification involve Neurogenin-dependent and - independent pathways. EMBO J 23: 2892-2902

261 Schuurmans C, Guillemot F (2002) Molecular mechanisms underlying cell fate specification in the developing telencephalon. Curr Opin Neurobiol 12: 26-34

Schwappacher R, Weiske J, Heining E, Ezerski V, Marom B. Henis YI, Fluber O, Knaus P (2009) Novel crosstalk to BMP signalling: cGMP-dependent kinase I modulates BMP receptor and Smad activity. EMBOJ28: 1537-1550

Sepkuty JP, Cohen AS, Eccles C, Rafiq A, Behar K, Ganel R, Coulter DA, Rothstein JD (2002) A neuronal glutamate transporter contributes to neurotransmitter GABA synthesis and epilepsy. J Neurosci 22: 6372-6379

Shahbazian M, Young J, Yuva-Paylor L, Spencer C, Antalffy B, Noebels J, Armstrong D, Paylor R, Zoghbi H (2002) Mice with truncated MeCP2 recapitulate many Rett syndrome features and display hyperacetylation of histone H3. Neuron 35: 243-254

Shahbazian MD, Grunstein M (2007) Functions of site-specific histone acetylation and deacetylation. Annu Rev Biochem 76: 75-100

Shalizi A, Gaudilliere B, Yuan Z, Stegmuller J, Shirogane T, Ge Q, Tan Y, Schulman B, Harper JW, Bonni A (2006) A calcium-regulated MEF2 sumoylation switch controls postsynaptic differentiation. Science 311: 1012-1017

Shalizi AK, Bonni A (2005) brawn for brains: the role of MEF2 proteins in the developing nervous system. Curr Top Dev Biol 69: 239-266

Sharrocks AD (2006) P1AS proteins and transcriptional regulation—more than just SUMO E3 ligases? Genes Dev 20: 754-758

Shechter D, Dormann HL, Allis CD, Hake SB (2007) Extraction, purification and analysis of histones. Nat Protoc 2: 1445-1457

Shen Z, Pardington-Purtymun PE, Comeaux JC, Moyzis RK, Chen DJ (1996) UBL1, a human ubiquitin-1 ike protein associating with human RAD51/RAD52 proteins. Genomics 36: 271-279

Sheng M, Dougan ST, McFadden G, Greenberg ME (1988) Calcium and growth factor pathways of c-fos transcriptional activation require distinct upstream regulatory sequences. Mol Cell Biol 8: 2787-2796

Sheng M, Thompson MA, Greenberg ME (1991) CREB: a Ca(2+)-regulated transcription factor phosphorylated by calmodulin-dependent kinases. Science 252: 1427-1430

Shi Y, Lan F, Matson C, Mulligan P, Whetstine JR, Cole PA, Casero RA (2004) Histone demethylation mediated by the nuclear amine oxidase homolog LSD1. Cell 119: 941-953

Shogren-Knaak M, Ishii H, Sun JM, Pazin MJ, Davie JR, Peterson CL (2006) Histone H4-K16 acetylation controls chromatin structure and protein interactions. Science 311: 844-847

Sidman RL, Rakic P (1973) Neuronal migration with special reference to developing human brain: A review. Brain Res 62: 1-35

Sincic N, Herceg Z (2010) DNA methylation and cancer: ghosts and angels above the genes. Curr Opin Oncol

262 Skene PJ, Illingworth RS, Webb S, Kerr AR, James KD, Turner DJ, Andrews R, Bird AP (2010) Neuronal MeCP2 is expressed at near histone-octamer levels and globally alters the chromatin state. Mol Cell 37: 457-468

Smedley D, Hamoudi R, Lu YJ, Cooper C, Shipley J (1999) Cloning and mapping of members of the MYM family. Genomics 60: 244-247

Smith BC, Denu JM (2009) Chemical mechanisms of histone lysine and arginine modifications. Biochim Biophys Acta 1789: 45-57

Sohal J, Reiter A, Goldman J, Cross N (2000) Cloning of ZNF237, a novel member of the MYM gene family that maps to human chromosome 13q 11 —>q 12. Cytogenet Cell Genet 89: 24-28

Song J, Durrin LK, Wilkinson TA, Krontiris TG, Chen Y (2004) Identification of a SUMO- binding motif that recognizes SUMO-modified proteins. Proc Natl Acad Sci USA 101: 14373- 14378

Soshnikova N, Duboule D (2009) Epigenetic regulation of vertebrate Hox genes: a dynamic equilibrium. Epigenetics 4: 537-540

Squire LR (2008) Fundamental neuroscience, 3rd edn.: Academic Press.

Stankiewicz P, Lupski JR (2002) Genome architecture, rearrangements and genomic disorders. Trends Genet 18: 74-82

Stankovic-Valentin N, Deltour S, Seeler J, Pinte S, Vergoten G, Guerardel C, Dejean A, Leprince D (2007) An acetylation/deacetylation-SUMOylation switch through a phylogenetically conserved psiKXEP motif in the tumor suppressor HIC1 regulates transcriptional repression activity. Mol Cell Biol 27: 2661-2675

State M, Dykens EM (2000) Genetics of childhood disorders: XV. Prader-Willi syndrome: genes, brain, and behavior. J Am Acad Child Adolesc Psychiatry 39: 797-800

Steenland HW. Kim SS, Zhuo M (2008) GluR3 subunit regulates sleep, breathing and seizure generation. Eur JNeurosci 27: 1166-1173

Sterner DE, Berger SL (2000) Acetylation of histones and transcription-related factors. Microbiol Mol Biol Rev 64: 435-459

Stielow B, Sapetschnig A, Kruger I, Kunert N, Brehm A, Boutros M, Suske G (2008) Identification of SUMO-dependent chromatin-associated transcriptional repression components by a genome-wide RNAi screen. Mol Cell 29: 742-754

Stromme P, Bjornstad PG, Ramstad K (2002) Prevalence estimation of Wililams syndrome. J Child Neurol 17: 269-271

Su HL, Li SS (2002) Molecular features of human ubiquitin-like SUMO genes and their encoded proteins. Gene 296: 65-73

Sugimori M, Nagao M, Bertrand N, Parras CM, Guillemot F, Nakafuku M (2007) Combinatorial actions of patterning and HLH transcription factors in the spatiotemporal control of neurogenesis and gliogenesis in the developing spinal cord. Development 134: 1617-1629

263 Tachibana M, Ueda J, Fukuda M, Takeda N, Ohta T, Iwanari H, Sakihama T, Kodama T, Flamakubo T, Shinkai Y (2005) Histone methyltransferases G9a and GLP form heteromeric complexes and are both crucial for methylation of euchromatin at H3-K9. Genes Dev 19: 815- 826

Tager-Flusberg H, Plesa-Skwerer D, Faja S, Joseph RM (2003) People with Williams syndrome process faces holistically. Cognition 89: 11-24

Takahashi Y, Iwase M, Konishi M, Tanaka M, Toh-e A, Kikuchi Y (1999) Smt3, a SUMO-1 homolog, is conjugated to Cdc3, a component of septin rings at the mother-bud neck in budding yeast. Biochem Biophys Res Comm 259: 582-587

Takahashi Y, Kahyo T, Toh EA, Yasuda H, Kikuchi Y (2001) Yeast Ulll/Sizl is a novel SUM01/Smt3 ligase for septin components and functions as an adaptor between conjugating enzyme and substrates. J Biol Chem 276: 48973-48977

Tam E, Young E, Morris CA, Marshall C, Loo W, Scherer S, Mervis C, Osborne L (2008) The common inversion of the Williams-Beuren syndrome region at 7ql 1.23 does not cause clinical symptoms. Am J Med Gen A 146A: 1797-1806

Tamamaki N, Yanagawa Y, Tomioka R, Miyazaki J, Obata K, Kaneko T (2003) Green fluorescent protein expression and colocalization with calretinin, parvalbumin, and somatostatin in the GAD67-GFP knock-in mouse. J Comp Neurol 567: 60-79

Tanaka K, Nishide J, Okazaki K, Kato H, Niwa O, Nakagawa T, Matsuda H, Kawamukai M, Murakami Y (1999) Characterization of a fission yeast SUMO-1 homologue, Pmt3p, required for multiple nuclear events, including the control of telomere length and chromosome Segregation. Mol Cell Biol 19: 8660-8672

Tao X, Finkbeiner S, Arnold DB, Shaywitz AJ, Greenberg ME (1998) Ca2+ influx regulates BDNF transcription by a CREB family transcription factor-dependent mechanism. Neuron 20: 709-726

Tassabehji M, Carette M, Wilmot C, Donnai D, Read AP, Metcalfe K (1999a) A transcription factor involved in skeletal muscle gene expression is deleted in patients with Williams syndrome. Eur J Hum Genet 7: 737-747

Tassabehji M, Hammond P, Karmiloff-Smith A, Thompson P, Thorgeirsson SS, Durkin ME, Popescu NC, Hutton T, Metcalfe K, Rucka A, Stewart H, Read AP, Maconochie M, Donnai D (2005) GTF21RD1 in craniofacial development of humans and mice. Science 310: 1184-1187

Tassabehji M, Metcalfe K, Karmiloff-Smith A, Carette MJ (1999b) Williams Syndrome: Use of chromosomal microdeletions as a tool to dissect cognitive and physical phenotypes. Am J Hum Genet 64: 118-125

Tassabehji M, Metcalfe K, Karmiloff-Smith A, Carette MJ (1999c) Williams Syndrome: Use of chromosomal microdeletions as a tool to dissect cognitive and physical phenotypes. Am J Hum Genet 64: 118-125

Tassabehji M, Urban Z (2006) Congenital heart disease: Molecular diagnostics of supravalvular aortic stenosis. Methods Mol Med 126: 129-156

264 Tatham MH, Jaffray E, Vaughan OA, Desterro JM, Botting CH, Naismith JH, Hay RT (2001) Polymeric chains of SUMO-2 and SUMO-3 are conjugated to protein substrates by SAE1/SAE2 and Ubc9. J Biol Chem 276: 35368-35374

Tay E (2005) Characterisation of a novel family of transcription factors, the MusTRDs. PhD (Medicine) Thesis, Muscle Development Unit, The Children's Medical Research Institute, University of Sydney, Sydney

Tay ES, Guven KL, Subramaniam N, Polly P, Issa LL, W. GP, Hardeman EC (2003) Regulation of alternative splicing of Gtf2irdl and its impact on slow muscle promoter activity. Biochem J374: 359-367

Tewari R, Bailes E, Bunting KA, Coates JC (2010) Armadillo-repeat protein functions: questions for little creatures. Trends Cell Biol 20: 470-481

Thomas MS, Van Duuren M, Purser HR, Mareschal D, Ansari D, Karmiloff-Smith A (2010) The development of metaphorical language comprehension in typical development and in Williams syndrome. J Exp Child Psychol 106: 99-114

Thompson PD, Webb M, Beckett W, Hinsley T, Jowitt T, Sharrocks AD, Tassabehji M (2007) GTF2IRD1 regulates transcription by binding an evolutionarily conserved DNA motif'GUCE'. FEBS Letters 581: 1233-1242

Thompson PM, Lee AD, Dutton RA, Geaga JA, Hayashi KM, Eckert MA, Bellugi U, Galaburda AM, Korenberg JR, Mills DL, Toga AW, Reiss AL (2005) Abnormal cortical complexity and thickness profiles mapped in Williams syndrome. JNeurosci 25: 4146-4158

Timmann D, Drepper J, Frings M, Maschke M, Richter S, Gerwig M, Kolb FP (2010) The human cerebellum contributes to motor, emotional and cognitive associative learning. A review. Cortex 46: 845-857

Tipney HJ, Hinsley TA, Brass A, Metcalfe K, Donnai D, Tassabehji M (2004) Isolation and characterisation of GTF2IRD2, a novel fusion gene mapping to the Williams-Beuren syndrome critical region. Eur J Hum Genet 12: 551-560

Tolle TR, Berthele A, Zieglgansberger W, Seeburg PH, Wisden W (1993) The differential expression of 16 NMDA and non-NMDA receptor subunits in the rat spinal cord and in periaqueductal gray. J Neurosci 13: 5009-5028

Torniero C, dal la Bernardina B, Novara F, Vetro A, Ricca I, Darra F, Pramparo T, Guerrini R, Zuffardi O (2007) Cortical dysplasia of the left temporal lobe might explain severe expressive- language delay in patients with duplication of the Williams-Beuren locus. Eur J Hum Genet 15: 62-67

Towbin H, Staehlin T, Gordon J (1979) Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc Natl Acad Sci USA 76: 4350-4354

Trauner DA, Bellugi U, Chase C (1989) Neurologic features of Williams and Down syndromes. Pediatr Neurol 5: 166-168

Tsukada Y, Fang J, Erdjument-Bromage H, Warren ME, Borchers CH, Tempst P, Zhang Y (2006) Histone demethylation by a family of JmjC domain-containing proteins. Nature 439: 811-816

265 Tsukada Y, Ishitani T, Nakayama KI (2010) KDM7 is a dual demethylase for histone H3 Lys 9 and Lys 27 and functions in brain development. Genes Dev 24: 432-437

Tudor M, Akbarian S, Chen RZ, Jaenisch R (2002) Transcriptional profiling of a mouse model for Rett syndrome reveals subtle transcriptional changes in the brain. Proc Natl Acad Sci USA 99: 15536-15541

Turner DJ, Miretti M. Rajan D, Fiegler H, Carter NP, Blayney ML, Beck S, Hurles ME (2008) Germline rates of de novo meiotic deletions and duplications causing several genomic disorders. Nat Genet 40: 90-95

Turrigiano GG, Nelson SB (2004) Homeostatic plasticity in the developing nervous system. Nat Rev Neurosci 5: 97-107

Tussie-Luna MI, Bayarsaihan D, Seto E, Ruddle FH, Roy AL (2002a) Physical and functional interactions of histone deacetylase 3 with TFII-I family proteins and PIASxp. Proc Natl Acad Sci USA 99: 12807-12812

Tussie-Luna MI, Michel B, Hakre S, Roy AL (2002b) The SUMO ubiquitin-protein isopeptide ligase family member Mizl/PIASxbeta /Siz2 is a transcriptional cofactor for TFII-I. J Biol Chem 277: 43185-43193

Tzingounis AV, Nicoll RA (2006) Arc/Arg3.1: linking gene expression to synaptic plasticity and memory. Neuron 52: 403-407

Udwin O, Yule W (1990) Expressive language of children with Williams syndrome. Am J Med Genet Suppl 6: 108-1 14

Ulrich HD (2005) Mutual interactions between the SUMO and ubiquitin systems: a plea of no contest. Trends Cell Biol 15: 525-532

Valero MC, de Luis O, Cruces J, Perez Jurado LA (2000) Fine-scale comparative mapping of the human 7q I 1.23 region and the orthologous region on mouse chromosome 5G: The low-copy repeats that flank the Williams-Beuren syndrome deletion arose at breakpoint sites of an evolutionary inversion(s). Genomics 69: 1-13

Van der Aa N, Rooms L, Vandeweyer G, van den Ende J, Reyniers E, Fichera M, Romano C, Delle Chiaie B, Mortier G, Menten B, Destree A, Maystadt I, Mannik K, Kurg A, Reimand T, McMullan D, Oley C, Brueton L, Bongers EMHF, van Bon BWM, Pfund R, Jacquemont S, Ferrarini A, Martinet D, Schrander-Stumpel C, Stegmann APA, Frints SGM, de Vries BBA, Ceulemans B, Kooy RF (2009) Fourteen new cases contribute to the characterization of the 7ql 1.23 microduplication syndrome. Eur J Med Genet 52: 94-100 van der Maarel SM, Scholten IH, Huber I, Philippe C, Suijkerbuijk RF, Gilgenkrantz S, Kere J, Cremers FP. Ropers HH (1996) Cloning and characterization of DXS6673E, a candidate gene for X-linked mental retardation in Xq 13.1. Hum Mol Genet 5: 887-897

Van Essen DC, Anderson CH. Felleman DJ (1992) Information processing in the primate visual system: an integrated systems perspective. Science 255: 419-423 van Hagen JM, van der Geest JN, van der Giessen RS, Lagers-van Haselen GC, Eussen HJ, Gille JJ, Govaerts LC, Wouters CH, de Coo IF, Hoogenraad CC, Koekkoek SK, Frens MA, van Camp N, van der Linden A, Jansweijer MC, Thorgeirsson SS, De Zeeuw Cl (2007)

266 Contribution of CYLN2 and GTF2IRD1 to neurological and cognitive symptoms in Williams Syndrome. Neurobiol Dis 26: 112-124 van Rossum DB, Patterson RL, Sharma S, Barrow RK, Kornberg M, Gill DL. Snyder SH (2005) Phospholipase Cgammal controls surface expression of TRPC3 through an intermolecular PH domain. Nature 434: 99-104

Varga AC, Wrana JL (2005) The disparate role of BMP in stem cell biology. Oncogene 24: 5713-5721

Vecsey CG, Hawk JD, Lattal KM, Stein JM, Fabian SA, Attner MA, Cabrera SM, McDonough CB, Brindle PK, Abel T, Wood MA (2007) Histone deacetylase inhibitors enhance memory and synaptic plasticity via CREB:CBP-dependent transcriptional activation. J Neurosci 27: 6128- 6140

Vicari S, Bellucci S, Carlesimo GA (2005) Visual and spatial long-term memory: differential pattern of impairments in Williams and Down syndromes. Dev Med Child Neurol 47: 305-311

Vicini S, Ortinski P (2004) Genetic manipulations of GABAa receptor in mice make inhibition exciting. Pharmacol Ther 103: 109-120

Vinson C, Myakishev M, Acharya A, Mir AA, Moll JR, Bonovich M (2002) Classification of human B-Z1P proteins based on dimerization properties. Mol Cell Biol 22: 6321-6335

Visel A, Blow MJ, Li Z, Zhang T, Akiyama JA, Holt A, Plajzer-Frick 1, Shoukry M, Wright C, Chen F, Afzal V, Ren B, Rubin EM, Pennacchio LA (2009) ChlP-seq accurately predicts tissue- specific activity of enhancers. Nature 457: 854-858

Vullhorst D, Buonanno A (2003) Characterisation of general transcription factor 3, a transcription factor involved in slow muscle-specific gene expression. J Biol Chem 278: 8370- 8379

Vullhorst D, Buonanno A (2005) Multiple GTF2I-like repeats of general transcription factor 3 exhibit DNA binding properties. Evidence for a common origin as a sequence-specific DNA interaction module. J Biol Chem 280: 31722-31731

Waites CL, Craig AM, Garner CC (2005) Mechanisms of vertebrate synaptogenesis. Annu Rev Neurosci 28: 251-274

Walter E, Mazaika PK, Reiss AL (2009) Insights into brain development from neurogenetic syndromes: evidence from fragile X syndrome, Williams syndrome, Turner syndrome and velocardiofacial syndrome. Neuroscience 164: 257-271

Wang J, Schwartz RJ (2010) Sumoylation and regulation of cardiac gene expression. Circ Res 107: 19-29

Wang Y, Falasca M, Schlessinger J, Malstrom S, Tsichlis P, Settleman J, Hu W, Lim B, Prywes R (1998) Activation of the c-fos serum response element by phosphatidyl inositol 3-kinase and rho pathways in HeLa cells. Cell Growth Differ 9: 513-522

Wang YK, Samos CH, People R. Perez Jurado LA, Nusse R, Francke U (1997) A novel human homologue of the Drosophila frizzled wnt receptor gene binds wingless protein and is in the Williams syndrome deletion at 7ql 1.23. Hum Mol Genet 6: 465-472

267 Watts FZ (2004) SUMO modification of proteins other than transcription factors. Seminars in Cell X Developmental Biology 15: 211-220

Wen Y-D, Perissi V, Staszewski LM, Yang W-M, Krones A, Glass CK, Rosenfeld MG, Seto E (2000) The histone deacetylase-3 complex contains corepressors. Proc Natl Acad Sci US A 91: 7202-7207

Wen YD, Cress WD, Roy AL, Seto E (2003) Histone deacetylase 3 binds to and regulates the multifunctional transcription factor TFI1-I. J Biol ChemllS: 1841-1847

Widagdo J (2006) Towards the molecular mechanism of GTF2IRD1 in the neurocognitive pathology of Williams-Beuren Syndrome and human cognition/ behaviour BSc (Bioinformatics) (Hons) Thesis, Muscle Development Unit, The Children's Medical Research Institute, Department of Pathology, University of Sydney, Sydney

Wilkinson KA, Henley JM (2010) Mechanisms, regulation and consequences of protein SUMOylation. Biochem J 428: 133-145

Wilkinson KA, Nakamura Y, Henley JM (2010) Targets and consequences of protein SUMOylation in neurons. Brain Res Rev 64: 195-212

Wilkinson LS, Davies W, Isles AR (2007) Genomic imprinting effects on brain development and function. Nat Rev Neurosci 8: 832-843

Williams J, Barratt-Boyes B, Lowe J (1961) Supravalvular aortic stenosis. Circulation 24: 1311-1318

Wong GT (2002) Speed congenics: applications for transgenic and knock-out mouse strains. Neuropeptides 36: 230-236

Woodruff-Borden J, Kistler DJ, Henderson DR, Crawford NA, Mervis CB (2010) Longitudinal course of anxiety in children and adolescents with Williams syndrome. Am J Med Genet C Semin Med Genet 154C: 277-290

Wu J, Lessard J, Olave I, Qiu Z, Ghosh A, Graef I, Crabtree G (2007) Regulation of dendritic development by neuron-specific chromatin remodeling complexes. Neuron 56: 94-108

Wu Y, Patterson C (1999) The Human KDR/flk-1 Gene Contains a Functional Initiator Element That Is Bound and Transactivated by TFII-I. J Biol Chem 274: 3207-3214

Wu YQ, Bejjani BA, Tsui LC, Mandel A, Osborne LR, Shaffer LG (2002) Refinement of the genomic structure of STX1A and mutation analysis in nondeletion Williams syndrome patients. Am J Med Genet 109: 121-124

Wysocka J, Swigut T, Milne TA, Dou Y, Zhang X, Burlingame AL, Roeder RG, Brivanlou AH, Allis CD (2005) WDR5 associates with histone H3 methylated at K4 and is essential for H3 K4 methylation and vertebrate development. Cell 121: 859-872

Xiao S, Nalabolu SR. Aster JC, Ma J, Abruzzo L, Jaffe ES, Stone R, Weissman SM, Hudson TJ, Fletcher JA (1998) FGFR1 is fused with a novel zinc-finger gene, ZNF198, in the t(8;13) leukaemia/lymphoma syndrome. Nat Genet 18: 84-87

Yan D, Davis FJ, Sharrocks AD, 1m HJ (2010) Emerging roles of SUMO modification in arthritis. Gene 466: 1-15

268 Yan X, Zhao X, Qian M, Guo N, Gong X, Zhu X (2000) Characterization and gene structure of a novel retinoblastoma-protein-associated protein similar to the transcription regulator TFII-1. Biochem J345: 749-757

Yang M, Hsu C-T, Ting C-Y, Liu LF, Hwang J (2006) Assembly of a polymeric chain of SUMOl on human Topoisomerase 1 in vitro. J Biol Chem 281: 8264-8274

Yang W, Desiderio S (1997) BAP-135, a target for Bruton's tyrosine kinase in response to B cell receptor engagement. Proc Natl Acad Sci U S A 94: 604-609

Yang XJ (2004) Lysine acetylation and the bromodomain: a new partnership for signaling. Bioessays 26: 1076-1087

Yang XJ, Seto E (2008) The Rpd3/Hdal family of lysine deacetylases: from and yeast to mice and men. Nat Rev Mol Cell Biol 9: 206-218

Yeh ET (2009) SUMOylation and De-SUMOylation: wrestling with life's processes. J Biol Chem 284: 8223-8227

Young E, Lipina T, Tam E, Mandel A, Clapcote S, Bechard A, Chambers J, Mount H, Fletcher P, Roder J, Osborne L (2008) Reduced fear and aggression and altered serotonin metabolism in Gtf2irdl-targeted mice. Genes, Brain and Behavior 7: 224-234

Yuste R, Bonhoeffer T (2001) Morphological changes in dendritic spines associated with long­ term synaptic plasticity. Annu Rev Neurosci 24: 1071-1089

Zhao C, Aviles C, Abel RA, Almli CR, McQuillen P, Pleasure SJ (2005) Hippocampal and visuospatial learning defects in mice with a deletion of frizzled 9, a gene in the Williams syndrome deletion interval. Development 132: 2917-2927

Zhao Y, Kwon SW, Anselmo A, Kaur K, White MA (2004) Broad spectrum identification of cellular small ubiquitin-related modifier (SUMO) substrated proteins. J Biol Chem 279: 20999- 21002

269 Appendix A. Amino acid alignment of GTF2IRD1 protein in human, mousey Xenopus and Fugu.

The coloured bars indicate a perfect conservation of the residues across all species. Functional domains are underlined and labeled. Note that each functional domains (leucine zipper, LZ; repeat domains, RDs; small ubiquitin-like modifier, SUMO motif; nuclear localisation signal, NLS) are highly conserved across species. 100 ■ I .... I ■ I human -OB VPTNGCH( CCPDR PDR WNSAFTR—| Uig ffl R PPWFPPWKD------.msw R pLWND------mouse -Ofl IPTNGCGSER WNSTFAR Xenopus -Yffl GSLNNSRPNL WPSSIPG4 I AHKKD----- Fugu SC® VRA GSRGE- —VPLVSA 1H Nil RE.SIHTQSIP

130 140 150 160 170 180 190 20< • I--- I ---- I---- I human RSSLEHG SDVYLLRKMV mouse RSSLEQC SDVYLTLQKMV Xenopus HVPLGPT SDIYLLRKMV Fugu AEGETRD QITNGGESAAi iii m Mi in Hi tim RD11

210 220 230 240 250 260 270 280 290 300 • • I --- I---- I human IPKGSRD CGLHGQAEPKV PPQDLPPTAT mouse PKGSRD CGLHGQASKV•QAS APQDLTPTAT Xenopus PTSATKV TPETSQCHRL IQE Fugu VSAAWTA SGHGPVKAAP PAGSTSSSAT

RD1

) 330 340 350 360 370 380 390 ------I------i human MPEPKATGAQ DFSfs Qk KJKP TGPG-GPLIQ HVH mouse VPEPHVP.STQ K}TP AGPA-GPLIQ NVHA5 Xenopus VQ1 PVAG-SSLRQ NVLi Fugu AS1WTS TDKM S’ TRP'IDTPRVS WRSHLgBittiHfM mmmIbI RD2

440 450 460 . . . I--- | human LEPAS pQdtsaevs Sri mouse LEPAS TSTEVCVC Xenopus SDSSS CSEN

RD2

530 540 550 560 570 590 ■ I • • • I • • • I . |------| ------| . I human DPSPjj S------Tip 7HLPS rjj.SGY-MEM-nfc>S Y V mouse dpspB s------>HLPS S)SGY MEM~ Xenopus GTSSEM S VKLIRLES EDPDLIQITV PGTSNETSGV KPKLESEDPD LIQIAV LAP S)PSYVLDTG Fugu -EAGV0 G------IP PAVCRVS-

610 620 630 640 650 660 670 680 690 700 ------|------| ------I------| ------|------| ------|------| . . human — LTDKGI.SE DARPt*RPVE DSH

710 720 730 740 750 760 770 78< 7 90 800 | | 1 human vTASSLGFSP PALP PE RD 3 v DPLVfJ --ESL Ff-V rFQ0HYD a mouse 3w DRLVJ ETP IQC ►GLCQMYN T Xenopus evtlE —ETN Trap SFQraSFD A Fugu Wfa!': ,S®S ::::3g rr pvteJ* GAALS K0F^ktQcle s 9m 11 3 5 3KE £! RD3 RD4

810 850 860 870 880 890 human S1 —ml::::1 mouse yte jjEj&TKII, TT HEA KTT RPRRL'/C'DTW QPjJiS 25 Xenopus Fugu I ^ HB9§5»prrv tllkkayasi n------BaSBrSy

RD4 RD5

910 920 930 940 950 960 970 980 990

• •• I -- I DAKVrAKDSS mouse n PXVPPETDS KKL KKVTIR Xenopus EPPXIILM KT «------Fugu QACmTUCtXMA

RD5 RD6™

1010 1020 1030 1040 1050 1060 1070 1080 1090

------|------( ------|------| ------|------| ------|------I ------I------I ------I------| ------|------| ------|------I ------|------I ------I------I human ESSES mouse EGVKELFKEK J ®AI LITO VX.VTYKI.tRD SPDAVEVKGL FDDIPFRNPM TYDIURLEKJ LKAREUVRMV riMCXQEr ® VCNMAKVTAK DKI Xenopus Fugu ------RD6 NLS

1120 1130 I I mouse Xenopus Fugu nil Appendix B-l. Table of various mammalian expression plasmids used in transient transfection experiments. Cloning sites Resistance Plasmid Vector (573'end) gene

pMyc-ARMCX5 pRK5 (Myc) So/I / Not\ Amp

pMyc-ARMCXl EH364 (pCDNA3.1, Myc-tagged) Xho\/ Xba\ Amp

pMyc-ARMCX3 EH364 (pCDNA3.1, Myc-tagged) Xho\/Xba\ Amp

EH318 (pEGFP-Cl, GFP excised, EcoR\/(Xba\) pMyc-TAFIB Kan Myc inserted) blunted EH318 (pEGFP-Cl, GFP excised, EcoR\/(Xba\) pMyc-SETD6 Kan Myc inserted) blunted EH318 (pEGFP-Cl, GFP excised, pMyc-SETD7 EcoR\/Xba\ Kan Myc inserted)

pGFP-ZMYM5 pEGFP-Cl (Clontech) Xho\/Kpn\ Amp

EH318 (pEGFP-Cl, GFP excised, pMyc-HDAC3 EcoR\/BamH\ Kan Myc inserted)

Each plasmid contains the full-length ORF of the transcript and was amplified either from its corresponding plasmid in pGADT7 (Appendix B-3) or from mouse cDNA sample. Taflb was re-amp lifted from the isolated yeast two-hybrid library clone. All plasmids are N-terminally epitope tagged by inserting the gene coding sequence into the indicated vector.

Appendix B-2. Table of primers used for cloning various mammalian expression plasmids

Forward primer (5' to 3') Reverse primer (5' to 3') name pMvc- GGGGATCGTCGACCATGATTGGCTCTAAAACT GGAT T GCGGCC GCT CAGAGTTTGAGTATT AAT C ARMCX5 pMyc- TTCTCGAGATGGGTCGCACCCGGGAAGCTGG TGTTGTCTAGAACAGATCTAAAATTATAATTTTG ARMCX1 pMyc- TTCTCGAGATGGGCTACGCCAGGAAAGTAGG GCTCTAGAGTTATTCCTGACTCTTTGGGAAC ARMCX3

pMyc-TAFlB ttgaattcatgaatgtggaggaggtgaaag ttctcgagtcagtgtcttcttcctttcttg

pMyc-SETD6 ttgaattcatggcggcccccgccaagcg aagcctgcttaatttgtgagttcc

pMyc-SETD7 ttgaattcaccatggatagcgacgatgag tttctagacttcagctccacttgataccactcg

pGFP- TTCTCGAGACATGGAAGCTCATCTCGCAG TTCCCGGGTACCTATGTGTACAGTAGCTGCACAC ZMYM5

Each plasmid contains the full-length ORF of the transcript and was amplified either from its corresponding plasmid in pGADT7 (Appendix B-3) or from mouse cDNA sample. Taflb was re-amplified from the isolated yeast two-hybrid library clone. All plasmids are N-terminally epitope tagged by inserting the gene coding sequence into the indicated vector. Appendix B-3. Table of plasmids used in generating Y2H constructs

Cloning sites Forward primer Reverse primer Plasmid (5' / 3'end) (5' to 3') (5' to 3') pGBKT7- EcoR\/Sal\ ATGAATTCACCATGGCCTTGCTG ATCTCGAGCTAGTAATTAAGAGGT GTF2IRD1* GGTAGCGCTGTGACGTCCCC CCCGGGAGCTGCACGTTCAGG pGBKT7-LZ EcoR\/Sal\ TTGAATTCAACGGCTGCGGACCC TTCTCGAGATCCTTCCACGGGGGC GAC CC pGBKT7-RDl* EcoR\/Sal\ TTGAATTCGGCTCAGATGTGTAC TTCTCGAGCCGCCCGCCATCCTCA CTTCTG AGTGG pGBKT7-RD2* EcoR\/Sal\ TTGAATTCACCGAGGACATCAAC TTGAATTCACCGAGGACATCAACA ACGCTC CGCTC pGBKT7-RD3* EcoR\/Sal\ TTGAATTCGGTGACGTGATCCGG TTCTGGAGTCCCTCAGTGAGCAGC CCCCTG TCGGG pGBKT7-RD4* E coR I/So/I TTCTGGAGTCCCTCAGTGAGCAG TTCTCGAGTGGGATGAGTCCTTGG CTCGGG AAAGG pGBKT7-RD5# EcoR\/Sal\ TTGAATTCCTCGGGGAGAAGGTG TTCTGGAGTTCTGCAAAGGGTTGG ATCCTG AGCTG pGBKT7-Sumol* EcoR\/Sal\ TTCTGGAGTTCTGCAAAGGGTTG TTGTCGACCTCACTCAGACCTTTG GAGCTG TC pGBKT7-Sumo2# EcoR\/Sal\ TTGAATTCGGGTCACGGGACTGT TTCTCGAGACCAGGCCCAGTGGGC GGCCTG TTCTG pGBKT7-Trl ATGTCGACCTAGTTGGCCGATGCC ACTGAATC pGBKT7-Tr2 ATGTCGACCTAATTTCCTTCCGAG ATGAATTCACCATGGCCTTGCTG ACCCGC EcoR\/Sal\ pGBKT7-Tr3 GGTAAGCGCTGTGACGTCCCC ATGTCGACCTACTTTGGGATGAGT CCTTGG pGBKT7-Tr4 ATGTCGACCTAGAAAATTCGCTCA TCAAAC pGADT7-ARMCX5 EcoR\ /(BamH\) TTGAATTCATGATTGGCTCTAAA TGACTAGCATCCATAAAATGG blunted ACTAAAAGG PGADT7-ARMCX1 Nde 1/ (Xba\) TTCATATGGGTCGCACCCGGGAA TGTTGTCTAGAACAGATCTAAAAT blunted GCTG TATAATTTTG PGADT7-ARMCX3 Nde\/ (Xba\) TTCATATGGGCTACGCCAGGAAA GCTCTAGAGTTATTCCTGACTCTT blunted GTAG TGGGAAC pGADT7-SETD6 EcoR\f See section 2.1.5 blunted PGADT7-HDAC3 EcoR\/ BamH\ TTGAATTCATGGCCAAGACCGTG TTGGATCCTTAAATCTCCACATCG GCCTATTTC CTTTCCTTG

All inserts were amplified by PCR using the listed primers. *Constructs were made by Ms Sara Bontempo; constructs were made by Ms Kylie Taylor. Appendix B-4. Plasmid maps of pGBKT7 (bait) and pGADT7 (prey) vectors used in Y2H assays (Ciontech Matchmaker system)

Hindlll (738) Hindlll 16544)

MCS .... Hind III (1480) Hindlll / / r SV40NLS\ \ (1606) j ; Amp pGADT7 gal«adM

1) 8°kb

\ PUC r" Hind I 1------(22801

Matchmaker 5 0HA-B0 Vector T7 SequencmflPrnrer "g i-sen Seeing A~p re' ^ GAU DNA Binding Domain '856 MATCHMAKER 5 AO 10 Insert Screening Am pinner TCA TCG GAA GAG AGT AGT AAC AAA GGT CAA AGA CAG TTG ACT GTA TCG CCG GAA TTT CTA TTC GAT GAT GAA GAT ACC CCA CCA AAC CCA AAA AAA GAG ATC TTT AAT ACG ACT “ II 14 Bgl ______T7 Secuenemq Pnmer _ * T — , S7^ HAtpRay Ti GTA ATA CCA CTC ACT ATA GGG CGA GCC GCC ATC MS GAG GAG CAG AAG CTG ATC TCA GAG GAG GAC CtS CAC TAT AGGGCGAGC GCC GCC ATG GAG TAC CCA TAC GAC GTA CCA GAT TAC GCT START A/col 1281 1*9 CAT ATG GCC ATG GAG GCC GAA TTC CCG GGG ATC CGT CGA CCTGCAGCG GCC GCA TAACTAGCATAACCCC CAT ATG GCC ATG GAG GCC AGT GAA TTC CAC CCG GGT GGG CAT CGA TAC GGG ATC CAT NiteV Ncoi Sfi\ fcoRI Small Sal I Psl I {J? S10P Nde I n^0| Sfi I fcoR I SmaTT cia I BamHI XmaI Xma I • T7 Terminator 2026 ______TTGGGGCCTCTAAACGGGTCTTGAGGGGTTfrrTGCGCGCTTGCAGCCAAGCTAATTCCGGGCGAATTTCTTATGATTT CGA GCT CGA GCT GCA GAfGAATCGTAGATACTGAAAAACCCCGCAAGTTCACTTC STOP tat* Sac I Xho I Pst I * MATCHMAKER 3 A0 ID Insert Scrsmmo Amofimei 1430 2061 ATGATTTTTATTATTAAATAAGTTATAAAAAAAATAAGTGTATACAAATTTTAAAGTGACTCTTAGGTTTTAAAACGAAAA AACTGTGCATCGTGCACCATCT 3 DNA-BD Sequencing Pr.mer 3 AD Sequencing Pnmer

Multiple cloning sites (MCS) are shown below the diagrams. Unique restriction sites are in bold. Appendix B-5. Table of primers used for sequencing of the general cloning vectors

Vector name Forward primer Reverse primer

(name; 5'-> 3') (name; 5'-> 3')

pGBKT7 T7: 3'BDSeq:

TAATACGACTCACTATAGGGC TTTTCGTTTTAAAACCTAAGAGT

pGADT7 T7 3'ADSeq:

As above AGATGGTGCACGATGCACAG

pCDNA3.1 T7 BgFIr:

As above AGAAGGCACAGTCGAGGCTG

pEGFP CMVEndF: LivCoIR:

CGGGACTTTCCAAAATGTCG AAACCACAACTAGAATGCAG

pRK5 (Myc) pRK5F: pRK5R:

GGTGACACTATAGAATAACATCCAC CCCGATCGATCCAGACATGATAAG

Appendix B-6. List of primers used in qPCR analyses

Transcript Forward primer (5'-> 3') Reverse primer (5'-> 3')

Baf53bF: Baf53bR: Actl6b GCAAGTTCAGCCCCTGGA CGTTTCTCCGTGGATGGG

Act! 6b F: Actl6bR: Act 16 b GCAAGTTCAGCCCCTGGA CGTTTCTCCGTGGATGGG

Zfp68F: Zfp68R: Zfp68 GCTTTCCACCAGAAAACAAA ACGGTGAAGGAAAACCTTGA

Cain IF: CalnlR: Calnl ATGCCGTTCCACCATGTAACC GCAGTTGGCTAATATCTCCGTG

Prkag2F: Prkag2R: Prkag2 TGGATCTGGAAAGGCTATGC TTTCCCTTCAACTGCATGAC

Taf6F: Taf6R: Taf6 TGTGTCACATGTGCTGACCT CAAATCCAACTTTTATTCTGAGG

Sic la IF: SIclalR: SI cl a 1 GGAAGGTAACCGAATGCTCA CCACCTGATAAGGGGACAGA

Npas4F: Npas4R: Npas4 GAGTCTCACTTCCTCACCGC CAGAGTCCGTCGGAATTGAA

Gria3F: Gria3R: Gria3 TCAAGCTATGGGAGTTTGGC ATGGCAATGAGGTTCGGTAA

All primers were designed to target the mouse mRNA sequence.