Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The role of lysine methyltransferase Ehmt2/G9a in mesenchymal development Zhang, Regan-Heng 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_february_zhang_reganheng.pdf [ 324.85MB ]
Metadata
JSON: 24-1.0340781.json
JSON-LD: 24-1.0340781-ld.json
RDF/XML (Pretty): 24-1.0340781-rdf.xml
RDF/JSON: 24-1.0340781-rdf.json
Turtle: 24-1.0340781-turtle.txt
N-Triples: 24-1.0340781-rdf-ntriples.txt
Original Record: 24-1.0340781-source.json
Full Text
24-1.0340781-fulltext.txt
Citation
24-1.0340781.ris

Full Text

The role of lysine methyltransferase EHMT2/G9a in mesenchymal development by Regan-Heng Zhang H.B.Sc., University of Toronto, 2009 A thesis submitted in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY in The Faculty of Graduate and Postdoctoral Studies (Medical Genetics) The University of British Columbia (Vancouver) January 2017© Regan-Heng Zhang, 2017ABSTRACT The euchromatic histone methyltransferase 2 (Ehmt2), aka. G9a, is responsible for methylating histone H3 at lysine 9. However, it is a multifaceted gene whose functions sometimes exceed histone-tail modifications, and the nature and importance of its regulatory effects differ vastly between different tissue systems. A number of lines of evidence hint at its critical role in the developmental biology and tissue biology of mesenchymal tissues. In this work I assessed the role of Ehmt2 in skeletal muscle, adipose tissues, and craniofacial development. Previous publications employing immortalised cell lines have proposed that Ehmt2 is an important inhibitor of myogenic differentiation. In addition to the well-known repressive effects of H3K9 dimethylation, for which Ehmt2 is largely responsible, it was postulated that it can directly methylate MYOD, a master regulator of myogenesis, and lead to repression of myogenic cell differentiation. In a mouse model to validate this, I found conditional knockout muscle stem cells activated, proliferated, and differentiated normally without Ehmt2. Knockout mice under the control of Myod-Cre developed and regenerated normally after acute injury to leg muscle, refuting the previous theory that Ehmt2 is required for myogenesis. The global loss of H3K9 dimethylation in normal myogenesis also signalled that this histone tail modification is largely irrelevant in skeletal muscle. Despite the gene’s dispensability in a number of developmental tissues, Ehmt2-/- mice die early during embryogenesis. In order to uncover the role of Ehmt2 in other mesenchymal tissues, I generated a transgenic mouse line to conditionally delete Ehmt2 during mesoderm and neural crest development. I found that the loss of Ehmt2 in the Pdgfra developmental lineage resulted in striking yet highly reproducible craniofacial malformations. !iiAdipose tissue is also an important topic in the understanding of Ehmt2, especially since previous publications have found its importance in white adipose tissue, and its homologue, Ehmt1/GLP, to be required for brown adipose tissue specification and activation. The PdgfraCre Ehmt2floxed/null mouse model revealed novel in vivo insight for the role in which Ehmt2 limits lipid accumulation and is required for normal brown adipose development. 
!iiiPREFACE This dissertation is an original intellectual product of RH Zhang. All of the work presented henceforth was conducted in Rossi Lab, The Biomedical Research Centre, at The University of British Columbia, Point Grey campus. All projects and associated methods were approved by The University of British Columbia’s Animal Care and Use Program and all experiments were performed according to the Canadian Council on Animal Care (CCAC) regulations. RH Zhang was the lead investigator, responsible for all major areas of concept formation, data collection and analysis, as well as manuscript composition. F M V Rossi was the supervisor and contributed throughout all projects in concept formation. R N Judson was involved in the project detailed in Chapter 2, and contributed to the design of in vitro experiments. All other technical contributions come from members of The Biomedical Research Centre. All first person accounts henceforth are made by RH Zhang. A version of Chapter 2 has been published [Zhang, RH., R. N. Judson, D. Y. Liu, J. Kast, and F. M. V. Rossi. (2016). The lysine methyltransferase Ehmt2/G9a is dispensable for skeletal muscle development and regeneration. Skeletal Muscle 6:22.]. Permission to reprint original figures and tables used in this dissertation that were published elsewhere was requested from each publisher and all requests were exempted based on noncommercial and educational use. 
!ivTABLE OF CONTENTS Abstract .............................................................................................................................................................................................................. iiPreface ................................................................................................................................................................................................................ iiiTable of Contents ............................................................................................................................................................................................. ivList of Tables ..................................................................................................................................................................................................... viiiList of Figures ................................................................................................................................................................................................... ixAcknowledgements ........................................................................................................................................................................................ xivChapter 1: Introduction .................................................................................................................................... 1Histone methylation and epigenetic mechanisms ........................................................................................................................... 1Ehmt2/G9a: protein functions ............................................................................................................................................................. 6H3K9 methylation ............................................................................................................................................................................ 6Protein-protein interactions .......................................................................................................................................................... 8Transcription activation .................................................................................................................................................................. 8Non-histone methylation ............................................................................................................................................................... 10Chemical inhibition and demethylase ......................................................................................................................................... 11The Ehmt gene family: Ehmt1/G9a-like protein ............................................................................................................................. 11Established animal studies of Ehmt1/2 .............................................................................................................................................. 15Embryonic stem cells ....................................................................................................................................................................... 15Kleefstra syndrome ........................................................................................................................................................................... 16Cognition and learning ................................................................................................................................................................... 17Mental health and drug addiction ................................................................................................................................................ 19Immune response ............................................................................................................................................................................. 21Leukemia and hematopoiesis ........................................................................................................................................................ 22Heart .................................................................................................................................................................................................... 23Aim and scope of this work ................................................................................................................................................................... 24Skeletal muscle: brief background ........................................................................................................................................................ 25Epigenetics of myogenesis .............................................................................................................................................................. 34Craniofacial development: brief background ................................................................................................................................... 35Adipose tissue development: brief background ............................................................................................................................... 39!vDevelopmental origins .................................................................................................................................................................... 40What is the mesenchyme: developmental and adult perspectives ............................................................................................... 43Chapter 2: The role of Ehmt2/G9a in skeletal muscle development and regeneration .................................... 50Introduction: current understanding of Ehmt2 in skeletal muscle ............................................................................................. 50Expression of Ehmt2 during embryonic myogenesis .............................................................................................................. 51Results ......................................................................................................................................................................................................... 52Developmental biology ................................................................................................................................................................... 52Generation and efficiency of developmental knockout of Ehmt2/G9a in skeletal muscle ..................................... 52Viability of Ehmt2/G9a knockout mice .............................................................................................................................. 56Neonatal growth analysis of Ehmt2/G9a knockout mice ............................................................................................... 56Structural integrity of skeletal muscle in Ehmt2/G9a knockout mice ......................................................................... 57Muscle stem cells ............................................................................................................................................................................... 58Ehmt2/G9a knockout satellite cell activation .................................................................................................................... 58Ehmt2/G9a knockout satellite cell proliferation ............................................................................................................... 60Ehmt2/G9a knockout myogenic differentiation ............................................................................................................... 60Ehmt2/G9a expression during myogenic differentiation ............................................................................................... 62Regeneration ...................................................................................................................................................................................... 62Ehmt2/G9a expression during skeletal muscle regeneration ......................................................................................... 62Myoblast activation and differentiation during regeneration ......................................................................................... 63Skeletal muscle regeneration in Ehmt2/G9a developmental knockout mice ............................................................ 65Skeletal muscle regeneration in Ehmt2/G9a adult satellite cell knockout mice ........................................................ 66Discussion .................................................................................................................................................................................................. 69The role of H3K9me2 in skeletal muscle development and regeneration .......................................................................... 69Potential compensatory mechanisms by Ehmt1/GLP ............................................................................................................ 70Contradictions between models ................................................................................................................................................... 71Ehmt2-mediated methylation of MYOD is uncertain ............................................................................................................ 71Methods and materials ........................................................................................................................................................................... 74Mice and animal care ....................................................................................................................................................................... 74Acute muscle injury .......................................................................................................................................................................... 74Cell culture and immunocytochemistry ..................................................................................................................................... 74Histology ............................................................................................................................................................................................. 75!viOptimization of purification of satellite cells ............................................................................................................................. 75Optimizing efficiency in knockout models ................................................................................................................................ 77A novel, fast and accurate method of measuring gene knockout efficiency ................................................................ 77Tamoxifen induction ................................................................................................................................................................ 80Myofibre size measurement ........................................................................................................................................................... 81Statistics ............................................................................................................................................................................................... 82Datasets of histological measurements ................................................................................................................................ 82Genetic knockout efficiency measurement ......................................................................................................................... 85Growth pattern analysis ........................................................................................................................................................... 86Chapter 3: The role of Ehmt2/G9a in congenital anomalies ............................................................................. 87Introduction: current understanding of Ehmt1/2 in Kleefstra syndrome and craniofacial development ........................ 87Results ......................................................................................................................................................................................................... 87General characterisation of the Pdgfra developmental lineage ............................................................................................ 87Tracing the Pdgfra lineage in skeletal muscle ............................................................................................................................ 97Tracing the Pdgfra lineage in hematopoiesis and the endothelium ..................................................................................... 99Viability of Pdgfra-Cre Ehmt2 knockout mice ........................................................................................................................ 100Cartilagenous bone development in Pdgfra-Cre Ehmt2 knockout mice .......................................................................... 102Ehmt2 is required during the closure of the fontanelles ......................................................................................................... 105Ehmt2 likely dictates ossification and cell composition at cranial sutures ......................................................................... 108Skull deformation in Pdgfra-Cre Ehmt2 knockout mice ...................................................................................................... 111Ehmt2 influences cell density in the brain .................................................................................................................................. 112Discussion .................................................................................................................................................................................................. 113Comparison with Wnt1-Cre Ehmt2 knockout ........................................................................................................................ 113Pdgfra-Cre Ehmt2 knockout mice as a model of Kleefstra syndrome ............................................................................... 114Methods and materials ........................................................................................................................................................................... 115Micro CT ............................................................................................................................................................................................ 115Chapter 4: The role of Ehmt2/G9a in adipose tissues ....................................................................................... 116Introduction: current understanding of Ehmt1/2 in adipose tissue development and homeostasis .................................. 116Results ......................................................................................................................................................................................................... 118Tracing the Pdgfra lineage in white adipose tissue ................................................................................................................... 118!viiAdipogenic differentiation of Ehmt2 knockout adipocyte progenitors .............................................................................. 119Ehmt2 is required for normal brown adipose tissue development ...................................................................................... 121Brown adipose tissue maintenance in Pdgfra-Cre Ehmt2 knockout mice ....................................................................... 124Ehmt2 influences the development of dermal adipose tissue ................................................................................................ 125Discussion .................................................................................................................................................................................................. 127Methods and materials ........................................................................................................................................................................... 128Histology ............................................................................................................................................................................................. 128Lipid vacuole size measurement ................................................................................................................................................... 128Chapter 5: Conclusion ....................................................................................................................................... 129Summary .................................................................................................................................................................................................... 129General discussion ................................................................................................................................................................................... 130Cellular variation of Ehmt2 function .......................................................................................................................................... 130Redundancy in the function of Ehmt2/G9a and Ehmt1/GLP ............................................................................................. 131Ehmt2/G9a as a regulator of terminal differentiation ............................................................................................................. 132Pdgfra as a marker of a critical developmental lineage ........................................................................................................... 133Broadened view of epigenetics ...................................................................................................................................................... 133Future directions ...................................................................................................................................................................................... 133Tissue maintenance .......................................................................................................................................................................... 133Brain development ........................................................................................................................................................................... 134Other mesenchymal tissues ............................................................................................................................................................ 134Non-histone post-translational modifications .......................................................................................................................... 135Therapeutic relevance ...................................................................................................................................................................... 135Significance of this work ......................................................................................................................................................................... 136Developmental biology ................................................................................................................................................................... 136Genetic modelling ............................................................................................................................................................................ 136Value of negative findings ............................................................................................................................................................... 136Bibliography ..................................................................................................................................................................................................... 138!viiiLIST OF TABLES Table 2.1 Summary of whole-mount in situ hybridisation results showing the expression of Myod, Ehmt2, and Ehmt1 during muscle development 51 ........................................................................................................................................................................................................................Table 3.1 Comparison of phenotypes of different Ehmt2 and Ehmt1 mutations in human and mice 115
.........................................................!ixLIST OF FIGURES Fig 1.1 Ehmt2 protein domains 10 ......................................................................................................................................................................................Fig 1.2 Ehmt gene family. A phylogenetic tree generated using the neighbour joining method showing the evolutionary relationship between Drosophila EHMT and the mouse and human orthologs of EHMT1 and EHMT2/G9a 12 .............................................................Fig. 1.3 Compensatory relationshiop between Ehmt2 and Ehmt1, based on viability of knockout mice 14 ...................................................Fig. 1.4 Summary of existing research on Ehmt2 in different tissue systems 15 ......................................................................................................Fig. 1.5 Summary of existing research on Ehmt2 in black and the focus of this work in red 25 ..........................................................................Fig. 1.6 Illustration of skeletal muscle tissue structure, cell types, and major regulatory factors 27 ....................................................................Fig. 1.7 Historical understanding of muscle, bone, and fat development 43 .............................................................................................................Fig. 2.1 C57BL/6 mice with Ehmt2-floxed allele bred to Myod-Cre 53 ....................................................................................................................Fig. 2.2 Ehmt2 domains and the Ehmt2-floxed targeted deletion area 53 ................................................................................................................Fig. 2.3 Ehmt2 deletion efficiency in Myod-Cre Ehmt2-floxed mice, as measured by Ehmt2 functional allele frequency in FACS-purified satellite cells. 54 .........................................................................................................................................................................................................Fig. 2.4 Immunofluorescence detection of EHMT2 on myofiber, isolated from wildtype and conditional knockout mice. 55 ..................Fig. 2.5 Relative abundance of H3K9me2 in whole skeletal muscle tissue lysate of wildtype and knockout mice, normalized to histone H3 55 ...........................................................................................................................................................................................................................................Fig. 2.6 a) Number of live births from n ≥ 3 mating pairs of Myod-Cre Ehmt2-floxed mice. b) Neonatal weight of wildtype and knockout mice at D0-D1. 56 .................................................................................................................................................................................................Fig. 2.7 Growth curve of wildtype and knockout mice. Regression analysis fitted a linear model for each group. 57 ....................................!xFig. 2.8 a) Weight of whole tibialis anterior muscle (p > 0.05). b) Masson’s trichrome stain of histological sections of the tibialis anterior muscle of adult mice. c) Myofiber size measurement by cross-sectional area  58 ....................................................................................................Fig. 2.9 a) Immunofluorescence detection and quantification of MYOD+ cells at 72h after myofibre isolation (p > 0.05). b) Immunofluorescence detection and quantification of PAX7+ cells at 72h after myofibre isolation (p > 0.05) 59 ...........................................Fig. 2.10 Immunofluorescence detection and quantification of EdU+ cells at 4h after EdU treatment (p > 0.05). 60 ....................................Fig. 2.11 a) Immunofluorescence detection and quantification of MYOG+ (myogenin) cells at 4h, 24h, and 48h during differentiation. b) Immunofluorescence detection of myosin heavy chain (MYH2) in myotubes at 48h during differentiation, myoblast fusion index calculated as % nuclei inside myosin-expressing myotubes. 61 ....................................................................................................................................Fig. 2.12 Immunofluorescence detection and quantification of EHMT2+ cells at 4h, 24h, and 48h during differentiation. 62 ..................Fig. 2.13 Ehmt2 gene expression in satellite cells during adult skeletal muscle regeneration. Satellite cells were purified by FACS at D1, D5, and D7, after notexin-induced TA muscle injury. 63 ..............................................................................................................................................Fig. 2.14 Myogenic gene expression in satellite cells during adult skeletal muscle regeneration 64 ....................................................................Fig. 2.15 a) Schematic diagram of muscle injury timeline for Myod-Cre Ehmt2-floxed mice. b) Masson’s trichrome stain of histological sections of the tibialis anterior muscle of adult Myod-Cre Ehmt2-floxed mice at 21 days after injury and myofiber size measurement by cross-sectional area 65 .......................................................................................................................................................................................................Fig. 2.16  YFP reporter expression as a measure of CreERT2 induction efficiency and specificity 67 ................................................................Fig 2.17 Pax7-CreERT2 Ehmt2-floxed/null mice and deletion efficiency 68 .........................................................................................................Fig. 2.18 a) Schematic diagram of leg injury timeline for Pax7-CreERT2 Ehmt2-floxed/null mice. b) Myofiber size measurement by cross-sectional area and Masson’s trichrome stain of histological sections of the tibialis anterior muscle of induced adult Pax7-CreERT2 Ehmt2-floxed/null mice at 21 days after injury 69 ........................................................................................................................................Fig. 2.19 LC-MS results from Ling, Bharathy, et al. (2012), showing the methylated MYOD peptide in cells transfected with wildtype Ehmt2/G9a 73 ...........................................................................................................................................................................................................................Fig. 2.20 Workflow of allele quantification for measuring knockout efficiency. 79 .................................................................................................!xiFig. 2.21 Comparison of knockout efficiency between different tamoxifen delivery methods in Ub-CreERT2 and Pax7-CreERT2. 81 .Fig. 2.22 Standard curve of ddPCR signal ratio to Ehmt2 functional allele frequency correlation 86 ................................................................Fig. 3.1 TdTomato reporter expression of mouse embryos at Theiler stage 23 (E14.5) 88 .....................................................................................Fig. 3.2 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of head. 89 .......................................................................................................................................................................................................................................Fig. 3.3 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of brain. 90 ......................................................................................................................................................................................................................................Fig. 3.4 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of liver and lungs. 91 .....................................................................................................................................................................................................................Fig. 3.5 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of lower abdomen, penis, and lower limbs. 92 .......................................................................................................................................................................Fig. 3.6 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Top: coronal view of lumbar vertebrae, left: lumbar vertebrae details and intervertebral discs. 93 ............................................................................................................Fig. 3.7 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of neck. 94 .......................................................................................................................................................................................................................................Fig. 3.8 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of scapula and head of humerus. 95 .........................................................................................................................................................................................Fig. 3.9 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of forelimb. 96 ................................................................................................................................................................................................................................Fig. 3.10 Flow cytometry showing percentage of fibro-adipogenic progenitors expressing tdTomato (left) and percentage of satellite cells expressing tdTomato (right) in digested skeletal muscle tissue. 98 .....................................................................................................................!xiiFig. 3.11 Flow cytometry showing percentage of the hematopoietic compartment (PTPRC+) expressing the YFP reporter (left) and percentage of endothelial cells (PECAM+) expressing the YFP reporter (right), in digested subcutaneous fat and skeletal muscle tissues. 100 ....................................................................................................................................................................................................................................Fig. 3.12 a) Genotype distribution of live births from Pgfra-Cre Ehmt2-floxed mating pairs. Expected frequencies are based on Mendelian ratios. b) Whole-mount light microscopy of embryo littermates: Pdgfra-Cre Ehmt2-f/wt and Pdgfra-Cre Ehmt2-f/f, at Theiler stage 20 (E12.5). 101 ....................................................................................................................................................................................................Fig. 3.13 Confocal microscopy of tdTomato expression in the radius bone of Pdgfra-Cre Rosa-tdTomato embryo at Theiler stage 23 (E14.5). 103 ..................................................................................................................................................................................................................................Fig. 3.14 a) Micro-CT of Pdgfra-Cre Ehmt2-f/f neonate, dorsal and lateral view of vertebrae, pelvis, and femur. b) Skeletal micro-CT of Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, lateral view. 104 ........................................................................................................................Fig. 3.15 a) Confocal microscopy showing tdTomato expression in the calvaria, in Pdgfra-Cre Rosa-tdTomato embryo at Theiler stage 23 (E14.5). Transverse section of superior part of head. b) Micro-CT of the frontal, parietal, occipital bones, and the upper palate in Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, inferior view. Right image is enlarged due to microcephaly. 105 .....................................Fig. 3.16 Skeletal micro-CT showing the fontanelles in Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, superior view. 106 ..................Fig. 3.17 Skull micro-CT showing the fontanelles in Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, superior view. 107 .......................Fig. 3.18 a) Coronal sections of the head of Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, alcian blue and picrosirius red co-stained, P2. b) Coronal sections showing frontal bones at the location of the frontal suture of littermates, alcian blue and picrosirius red co-stained, P2. c) Coronal sections of the frontal suture of litter mates, H/E, P7. d) Coronal sections of the frontal suture of litter mates, H/E, P14. 109 ..................................................................................................................................................................................................................Fig. 3.19 a) Micro-CT of the head of Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates. b) Cranial length measurement from the top of the incisors to the planum occipitale, p < 0.05, n = 6. 111 ...........................................................................................................................................Fig. 3.20 a) Confocal microscopy of tdTomato expression in spinal cord and dorsal root ganglion of Pdgfra-Cre Rosa-tdTomato b) Confocal microscopy of tdTomato expression in brain and tissues of the head. c) Coronal section of brain at the medial longitudinal fissure, H/E, P2. 113 ...................................................................................................................................................................................................................!xiiiFig. 4.1 a) Flow cytometry of digested subcutaneous adipose tissue from Pdgfra-Cre Rosa-tdTomato mice, showing non-blood and non-endothelial cells expressing tdTomato (red), compared to control (blue). b) Flow cytometry showing adipose progenitor cells (LY6A+) expressing tdTomato. 119 .......................................................................................................................................................................................Fig. 4.2 Primary adipogenic progenitors from Pdgfra-Cre Ehmt2-f/f mice differentiate faster.. 120 ....................................................................Fig. 4.3 Thoracic transverse sections of Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates at P2, showing interscapular and body wall brown adipose tissues, H/E. 122 .............................................................................................................................................................................................Fig. 4.4 a) Immunohistochemistry of perilipin in interscapular brown adipose tissues in Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates at P2. b) Frequency distribution of lipid vacuole size, measured by cross sectional area. 123 ............................................................Fig. 4.5 Brown adipose tissue in WT and KO compared to white adipose tissue 123 ...............................................................................................Fig. 4.6 Subcutaneous white adipose tissue appear similar between WT and KO 124 .............................................................................................Fig. 4.7 a) Schematic diagram of the generation of Ubc-CreERT2 Ehmt2-f/f mice. b) Timeline of Cre induction and tissue collection. c) H/E stain of interscapular brown adipose tissue. 125 ...................................................................................................................................................Fig. 4.8 a) Confocal microscopy of bone, muscle, dermal, and epidermal layers in the upper limb of Pdgfra-Cre Rosa-tdTomato embryo, showing tdTomato expression concentrated in the dermis. Theiler stage 23 (E14.5), transverse section. b) Coronal section of the scalp of Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, H/E, P2. 126 ............................................................................................................Fig. 5.1 Schematic diagram showing different conditional knockout studies of Ehmt2 in mice. Contributions from this work are in red. 130
..........................................................................................................................................................................................................................................!xivACKNOWLEDGEMENTS I thank my research supervisor Dr Fabio M V Rossi for crucial guidance in this research and supporting one of the most friendly close-knit work environments at The University of British Columbia. I would like to thank graduate committee members Dr Pamela Hoodless, Dr Keith Humphries, and Dr Martin Hirst for guidance, and the External and University Examiners for taking the time to review this dissertation. I offer my enduring gratitude to all members and alumni of the Rossi Lab for technical support and scientific advice, especially Dr Robert N. Judson, Ms Lin Yi, Mr Chihkai Chang, Ms Vittoria Canale, Ms Claudia Hopkins, Dr Bernhard Lehnertz, Dr Norihisa Higashihori, Dr Dario Lemos, Dr Pretheeban Thavaneetharajah, Dr Leslie S Alfaro, Dr Anuradha Natarajan, Dr Farshad Babaei, Ms Christine Eisner, Dr Elena Groppa, Dr Coral-Ann Lewis, Dr Marcela Low, Ms Joey Nguyen, Dr Hesham Soliman, Ms Gloria Loi, Mr David Y. Liu, Mr Alan Wong, Mr Ryan Cheng, and Mr Alvin Tsuei. I would also like to thank researchers at The Biomedical Research Centre Dr Jürgen Kast and Dr Kelly M McNagny for guidance and scientific input. The Animal Unit staff at the centre were instrumental in supporting my in vivo research, especially Ms Helen Merkens, Ms Krista Ranta, Mr Wei Yuan, and Mr Jerry Chen. I also owe tremendous help from Mr Andy Johnson and Mr Justin Wong of UBC Flow Cytometry, Mr Takahide Murakami of the BRC Genotyping Unit, Ms Ingrid Barta of the UBC Histology Lab, Mr John Schipilow and Dr Nancy Ford of the Centre for High-Throughput Phenogenomics. Mr Rick White and Mr Yi Huang from the Statistical Consulting and Research Laboratory (SCARL) taught me and helped me design statistical approaches for my projects, through the Statistical Opportunity for Students (SOS) program. I thank Dr Rosemary J Redfield for my training and experience in teaching. I thank my colleagues who served on the Trainee Communications Committee of the Stem Cell Network for supporting my ideas, Dr Natalie Farra, Dr Marleen Eijkholt, Dr Zubin Master, Dr Eva Szabo, Dr Alessandra Pasut, Dr Krishna Panchalingam, Dr Kelly McClellan, Dr Leslie S Alfaro, Mr Branavan Manoranjan, Dr Ashley Sanders, Dr Janet M Rothberg, Ms Shelly Benjaminy, Ms Sneha Balani, Dr Geoffrey Clarke, Dr Ben Paylor, as well as staff members of the Stem Cell Network Dr Paul Cassar, Ms Shannon Sethuram, Ms Rebecca Cadwalader, and Ms Lisa Willemse. I would also thank my wonderful students enrolled in Useful Genetics and Genetics for Life. I thank my graduate program administration, especially Ms Cheryl Bishop and Dr Ann M Rose. I thank my graduate program mentor Dr George Chung, and my classmates for collaborative learning, Dr Katharina Rothe, Ms Kasia Stepien, Ms Jennifer Grants, Ms Kate Slowski, Dr Colúm Connolly, Dr Sheng Liu, Ms Sarah Lepage, Ms Karen Jacob, Dr Peter Zhang, Dr David Lin, Ms Courteney Lai, Ms Christine Yang, Ms Marie Morimoto, and Dr Anna FY Poon. I thank the Canadian Institutes of Health Research and the Stem Cell Network for scholarships and grant funding of this research. !xv1. Introduction Histone methylation and epigenetic mechanisms Back in those gloomy rainy days when there was barely anything to do outside the portable classrooms, which made elementary schools look like barnyards, my nine-year-old self had enjoyed admiring the twins in pretty dresses across the desk. Sporting long, curly hair on the sides of almond-shaped faces, they looked astonishing, and exactly the same. “Their parents are going to take them to acting school”, someone whispered. They shared an inner language, as so often happens with twins; they had jokes that only the other twin understood. They even smelled the same. But if you knew them, you’d notice the striking differences. One was gregarious. She made friends easily. She was impervious to insults. The other was reserved, quieter, and more brittle. Looking at millions of identical twins in the world, this provokes a puzzling question: Why are identical twins different? Because, you might answer, fate impinges differently on their bodies. As they go through life, one twin falls down the crumbling stairs of her country-side house and breaks her ankle; the other scalds her thigh on a tipped cup of coffee in a bustling European station. Each acquires the wounds, calluses, and memories of chance and fate. But how are these changes recorded, so that they persist over the years? We know that the genome can manufacture identity; the trickier question is how it gives rise to difference. Looking at the thirty-seven trillion cells in the human body, this is also a question that had long troubled geneticists and cell biologists: if all the cells in the body have the same genome, how does one become a nerve cell, say, and another a blood cell, which looks and functions very differently? !1In the nineteen-forties, Conrad Waddington, a British embryologist, had proposed an ingenious answer: cells acquired their identities just as humans do—by letting nurture (environmental signals) modify nature (genes). For that to happen, Waddington concluded, an additional layer of information must exist within a cell—a layer that hovered, ghostlike, above the genome. This layer would carry the “memory” of the cell, recording its past and establishing its future, marking its identity and its destiny but permitting that identity to be changed, if needed. He termed the phenomenon “epigenetics”—“above genetics” (Waddington, 1942). Waddington, ardently anti-Nazi and fervently Marxist, may have had more than a scientific stake in this theory. The Nazis had turned a belief in absolute genetic immutability (“a Jew is a Jew”) into eugenics, a state-mandated program of sterilization and mass murder. By affirming the plasticity of nature (“everyone can be anyone”), a Marxist could hope to eradicate such innate distinctions and achieve a radical collective good. It was not until years after the complete DNA sequencing of the human genome did scientists finally came to the conclusion that Waddington did have the right idea. A neuron in the brain is a neuron, and not a lymphocyte, because a specific set of genes is turned “on” and another set of genes is turned “off.” The genome is not a passive blueprint: the selective activation or repression of genes allows an individual cell to acquire its identity and to perform its function. When one twin breaks an ankle and acquires a gash in the skin, wound-healing and bone-repairing genes are turned on, thereby recording a scar in one body but not the other. The nucleosome is an important clue to understanding epigenetics. Genes are typically carried in long, continuous chains of DNA: one such chain can carry hundreds of thousands of genes. But a chain of DNA does not typically sit naked in animal cells; it is wrapped tightly around a core of proteins called histones, or in the words of the celebrated Dr Siddhartha Mukherjee (2016), “like a python coiled around a skewer of marshmallows” (p. 26). And the chemical changes in histones !2could force the DNA coils to open, thereby allowing genes to be transcribed. As it turns out, a huge gamut of proteins have been discovered that could modify histones, and the marks they make on histones include many different types of methylation, acetylation, phosphorylation, ubiquitination, etc. These modifications give rise to the concept of heritable epigenetic information stored in the histone code. Dr David Allis made a clear summary of the features of these mechanisms: Two features of histone modifications are notable. First, changing histones can change the activity of a gene without affecting the sequence of the DNA. And, second, the histone modifications are passed from a parent cell to its daughter cells when cells divide. A cell can thus record “memory”, and not just for itself but for all its daughter cells. (Mukherjee, 2016, p. 26) One category of histone modification, perhaps the most widely studied, is histone methylation, which refers to the attachment of a methyl group to a lysine or arginine residue on the tail of one of the four histone proteins: H2A, H2B, H3, and H4. Methylation of histones can either increase or decrease transcription of genes, depending on which amino acids in the histones are methylated, and how many methyl groups are attached. These variables result in many different histone marks; currently, researchers have discovered over forty unique types of histone methylation. Trimethylation of lysine 4 on histone H3 (H3K4me3), for example, is commonly associated with active transcription, being highly enriched at active promoters near transcription start sites (Liang et al., 2004). Trimethylation of lysine 27 (H3K27me3), on the other hand, recruits complexes that facilitates the formation of heterochromatin, which has tightly packed nucleosomes, therefore suppressing transcription. Not only are there many unique histone methylation marks, but also their relationship with histone methyltransferases, the producer of such marks, are complex. Some of these enzymes are !3highly specific, such as SETD2, which is only known for methylating lysine 36 of histone H3. Others, such as CARM1, are more promiscuous, methylating arginines 17, 26, and 42 of histone H3. However, they are all similar in that they catalyze the transfer of one to three methyl groups from S-Adenosyl methionine (SAM) onto lysine or arginine residues. The histone methyltransferases can be grouped by their specificity to either lysine or arginine. The lysine-specific transferases are further broken down into whether or not they have a SET domain or a non-SET domain. SET domain was originally identified as part of a larger conserved region present in the Drosophila Trithorax protein and was subsequently identified in the Drosophila Su(var)3-9 and 'Enhancer of zeste' proteins, from which the acronym SET is derived. These domains specify exactly how the enzyme catalyzes the transfer of the methyl group from SAM to the target residue (Rice et al., 2003). In addition to methyltransferases, there are also demethylases, acetylases, deacetylases, etc, that can also modify the histone code. Non-histone systems, too, that could scratch different kinds of code on the genome were identified (some of these discoveries predating the identification of histone modifications). One involved the addition of a methyl group to DNA itself, which typically occurs in a CG dinucleotide context (i.e. where a cytosine is followed by a guanine), in a process called DNA methylation. The most heavily methylated parts of the genome tend to be dampened in their activity, as the methylation may physically impede the binding of transcriptional proteins to the gene, and may be bound by proteins known as methyl-CG-binding domain proteins (MBDs). MBD proteins then recruit additional proteins to the locus, such as histone modifiers and other chromatin remodeling proteins that can modify histones, thereby forming heterochromatin. In the medical sciences, researchers are most excited about the epigenetic implications for cancer. In some cancers, such as leukemias, malignant cells have markedly aberrant patterns of DNA !4methylation or histone modification. Clearly, there’s a signal that epigenetic information is important for a cancer cell, but it is still uncertain if a drug can safely change the epigenome of a cancer cell without touching a normal cell. Other researchers have looked at how epigenetics might change behaviours—not just cellular memory and identity but an organism’s memory and identity. The neuroscientist and psychiatrist Eric Nestler, whose group studies addiction, gave mice repeated injections of cocaine, and found that the histones were altered in the reward-recognizing region of the brain. When the histone modification was chemically blocked, the mice were less likely to become addicted (Damez-Werno et al., 2016; Heller et al., 2016). In 2004, a team of researchers at McGill University noticed that rats raised by low-nurturing mothers were likely to be notably stresssed as young adults (Fish et al., 2016). The memory of childhood neglect in rats appears to be related to epigenetic changes: a gene that acts as a set point for stress—an anxiety rheostat—is dampened in these poorly nurtured rats, resulting in higher levels of stress hormones. McGill researchers went on to study the brains of human beings who were abused as children and later committed suicide, and found similar epigenetic alterations. In 2008, the first consensus definition of the epigenetic trait, "stably heritable phenotype resulting from changes in a chromosome without alterations in the DNA sequence", was made at a Cold Spring Harbor meeting (Berger et al., 2009). However, epigenetics could be considered in a broader sense, including mechanisms that aren’t directly related to chromatin structure. For example, microRNAs are non-coding RNAs that can prevent the translation of a messenger RNA, thereby turning off gene expression. Another example involve prions, which are proteins folded in infectious conformational states that can induce phenotypic change, without affecting the transcription or translation processes. In the broadest sense, many cell signalling mechanisms can also be viewed as epigenetic mechanisms when post-translational modifications of non-histone proteins, such as phosphorylation, are passed from parental cells to daughter cells. New research has reported a handful of non-histone methylation that can have profound regulatory influence !5on gene expression; some kinases and regulatory proteins involved in phosphorylation are substrates of methylation. Similar to protein phosphorylation, non-histone protein methylation has been shown to have key roles in the signalling pathways that regulate growth, proliferation, differentiation and cell fate, or those that are associated with pathological conditions such as cancer (Biggar & Li, 2014). Collectively, epigenetic mechanisms thus far have emerged as crucial components of diverse biological processes. The full breadth of this relatively new field of study cannot possibly be covered in detail in this introduction, and I would recommend the textbook titled Epigenetics, by Allis, Caparros, Jenuwein, and Reinberg (2015) for a comprehensive background. Ehmt2/G9a protein functions The focus of this doctoral dissertation is a gene, called euchromatic histone-lysine N-methyltransferase 2 (Ehmt2) (MGI: 2148922), also known as G9a (Brown, Campbell, and Sanderson 2001). First characterised by Tachibana et al., (2001, 2002) as an epigenetic regulator due to its preferential methylation of histone H3 at lysine 9, subsequent studies in mouse has found that this gene has multiple interesting functions (Fig. 1.1), playing different regulatory roles in different mammalian tissue systems. H3K9 methylation Ehmt2 was characterised as a histone tail modifier due to its prominent SET domain, a conserved sequence commonly found in Su(var)3-9 and 'Enhancer of zeste' family of proteins, which are well known for catalyzing mono-, di, or tri-methylation of histone H3 at lysine residues, thereby regulating chromatin structure in many model organisms (Rea et al. 2000, Herz et al. 2013). Ehmt2 was initially found to act preferentially at lysines 9 and 27 of histone H3 (Tachibana et al. !62001). Ehmt2null/null mouse embryos failed to develop, and their embryonic stem cells showed a drastic reduction in H3K9 mono- and di-methylation (H3K9me1 and H3K9me2) but did not show reduced H3K9 trimethylation (H3K9me3) (Tachibana et al. 2002, 2005, Peters et al. 2003, Rice et al. 2003), suggesting that Ehmt2 is the key mediator of H3K9me1/2. H3K9me3 is primarily enriched in pericentric heterochromatin (Rice et al. 2003), and is mediated by Setdb1 during proviral silencing (Matsui et al. 2010), and by Suv39h1 and Suv39h2 at major satellite repeats (Lehnertz et al. 2003). In contrast, H3K9me1 is usually associated with actively transcribed genes (Barski et al. 2007), and H3K9me2 is associated with repressed euchromatic regions of the genome (Peters et al. 2003). Subsequently, a number of studies found Ehmt2 interacting with transcriptionally repressive protein complexes and it was required to mediate H3K9me2 at specific genes (Ueda et al. 2006, Mozzetta et al. 2014, Fritsch et al. 2010, Shi et al. 2003, Ogawa et al. 2002, Nishio and Walsh 2004), solidifying our understanding of its function as a transcription repressor through the production of the epigenetic mark H3K9me2. This aspect of Ehmt2 acting as a corepressor has been reported as an important mechanism in lineage-specific differentiation, as its H3K9me2 production can be targeted to specific genes by associating with transcriptional repressors and corepressors. It is a critical player in the regulation of genes required for normal embryonic development (Tachibana et al. 2002). The importance of this histone mark is evident in embryonic stem cells after differentiation starts, as the areas of H3K9me2-marked chromatin vary, depending on the specific differentiated cell lineages (Wen et al. 2009). To prevent the expression of neuronal genes in non-neuronal tissues, H3K9me2 mediates the repressive activity of REST (Roopra et al. 2004). Recently validated in mouse T cell differentiation, EHMT2 is required for CD4+ T cells to differentiate into Th2 cells by repressing transcription at lineage-promiscuous loci (Lehnertz et al. 2010). Thus, EHMT2 mediated H3K9me2 marks are largely believed to function as transcription repressors. Other examples of !7its corepressor function are associated with CDP/cut (Nishio & Walsh 2004), and Blimp-1/PRDI-BF1 (Gyory et al. 2004). Protein-protein interactions In addition, EHMT2 also contains an ANK repeat domain, which is well known for enabling protein-protein interactions (Mosavi et al. 2004). Not only does it allow for interaction with aforementioned corepressors, the ANK domain also allows it to bind to DNA methyltransferases (DNMT) (Dong et al. 2008). Through this interaction, EHMT2 reveals an alternative mechanism for gene repression, by DNA methylation, which is found in the silencing of Pou5f1 (Oct3/4) in embryonic stem cells (Epszteijn-Litman et al. 2008). Another interesting aspect of this domain is its ability to interact with H3K9me1 and H3K9me2, the very marks that EHMT proteins are known to generate (Collins et al. 2008), suggesting that EHMT2 may ‘read’ and ‘write’ these histone tail marks. Transcription activation The third, and the most interesting domain function, allows Ehmt2 to act as a gene activator in addition to its repressive roles, by interacting with coactivators. It can be a coactivator in nuclear receptor mediated transcription regulation (Lee et al. 2006). Nuclear receptors (NR) are transcription factors that respond to ligand binding, and recruit coactivator complexes to the promoters of target genes. EHMT2 acts in synergy with the p160 nuclear receptor coactivator GRIP1, the protein arginine methyltransferase CARM1, and the histone acetyltransferase p300 to activate the transcription of targets of the androgen receptor (AR) (Lee et al. 2006). This appears to be in conflict with Ehmt2’s aforementioned repressive roles. Lee et al., (2006) further reveals that the coactivation function is independent of EHMT2’s methyltransferase activity, and that the methyltransferase activity is completely inhibited when H3K9 acetylation (Lee et al. 2006, !8Chaturvedi et al. 2009) or H3K4 methylation (Wang et al. 2001) is present to allow it to function purely as an activator. In addition, the estrogen receptor α (ERα) is another nuclear receptor that involves EHMT2 coactivator function. Upon stimulation by the natural estrogen estradiol (E2) on ERα, EHMT2 is recruited directly to target gene areas. The ankyrin domain was initially believed to be responsible for the coactivating function by binding coactivators such as GRIP1, but Purcell et al., (2011) found that the activation of some target genes required neither the ankyrin domain nor the SET methyltransferase domain. Although previously undocumented, Lee et al., (2006) found that the fragment of residues 72–333 at the N-terminus contains an autonomous transcriptional activation domain. This unknown domain is sufficient by itself to bind to the ERα ligand-binding domain and act as an activator to upregulate E2-dependent target genes (Purcell et al. 2011). Interestingly, the residues 152–383 correspond to an unannotated domain in the Pfam-B database of domain families (http://pfam.sanger.ac.uk/) (Fig. 1.1), which was generated by the protein sequence-clustering algorithm ADDA (http://ekhidna.biocenter.helsinki.fi/sqgraph/pairsdb/index_html). Another example of this coactivator activity is EHMT2’s interaction with the transcription activator NF-E2/p45 in erythroid cells, where it activates the βmaj globin locus by establishing a preinitiation complex, through binding of RNA polymerase II (Chaturvedi et al. 2009). Currently, the gene activation role of EHMT2 lacks understanding, but Lee et al., (2006) and Chaturvedi et al., (2009) theorise that promoter context and/or regulatory environment could control whether EHMT2 functions as a repressor or an activator. !9   Non-histone methylation Futhermore, two N-terminal lysine residues of Ehmt2 can be self-methylated (Sampath et al. 2007, Chang et al. 2011). Methylation of these sites is required for protein–protein interactions (reviewed in Lanouette et al. 2014). In addition to self-methylation, the SET domain in EHMT2 and other histone methyltransferases have the potential to methylate non-histone polypeptides (Sampath et al. 2007, Huang & Berger 2008). Biggar and Li (2015) have concluded from many biochemical studies that Ehmt2 may mono- or dimethylate a minimum of 17 non-histone proteins, including itself. The most interesting of these was found by Ling, Bharathy, et al., (2012), in which EHMT2 methylated MYOD in a mechanism to repress myogenic differentiation. To be discussed further in section 1.2, this was the first time and only instance that the MYOD protein was found to be methylated, thereby potentially revealing a novel mechanism of non-histone epigenetic regulation of myogenesis. Thus, the multitude of domain functions in the large EHMT2 protein allow for the potential to play completely different roles in different contexts. transcription activation ankyrin repeat SETH3K9me1 H3K9me2 H3K9me3 DNMT H3K9me1 H3K9me2GRIP1 ERNFE2Fig. 1.1 Ehmt2 protein domains. The ankyrin repeat and SET domains are annotated by PFAM-A. The transcription activation domain is proposed by Lee et al. (2006) and Purcell et al. (2011), and is predicted by PFAM-B.!10Chemical inhibition and demethylation The discovery of two chemical inhibitors against the methyltransferase activity of EHMT2 greatly facilitated studies to evaluate the importance of the SET methyltransferase domain apart from other functions of EHMT2. The first one, Bix01294, has been shown to be effective as it prevents the SET methyltransferase activity and was successfully used to enhance the efficiency of reprogramming gene expression (Epsztejn-Litman et al. 2008, Shi et al. 2008, Imai et al. 2010), but it is toxic to primary cells at the concentrations required for EHMT2 inhibition (Vedadi et al. 2011) and therefore of limited utility. A more potent second compound was described, UNC0638, which displayed significantly lower toxicity than BIX0129421 and was therefore amenable to use in primary cells (Vedadi et al. 2011). Performing the opposite task as Ehmt2 is Phf8 (PHD finger protein 8), which is known to demethylate H3K9me2 and affect rRNA synthesis in the context of craniofacial development (Feng et al., 2010; Zhu et al., 2010; Qi et al., 2010). However, it is not specific, as it is also known to demethylate at histone H4 lysine 20 (Liu et al., 2010). The Ehmt gene family: ehmt1/g9a-like protein Ehmt2 is not the only mediator of H3K9 methylations. A closely related gene, Ehmt1, also known as G9a-like protein (GLP), can carry out the same methylation, as it also contains the SET domain and ANK repeats domain in almost the same region of the protein. Although located on separate chromosomes—in mice, Ehmt2 is on chromosome 17, Ehmt1 is on chromosome 2; in humans, they are on chromosomes 9 and 6, respectively—the two protein sequences share a high degree of homology. Mouse embryonic stem cells obtained from Ehmt1null embryos showed the same reduction in H3K9me1 and H3K9me2, but not H3K9me3 (Tachibana et al. 2002, 2005, Peters et al. 2003, Rice et al. 2003). Furthermore, Tachibana et al., (2002) found Ehmt2 and Ehmt1 are !11present in a heterodimeric complex, and that this conformation is required for their catalytic activity. The shared specificity and functional relationship suggests that the two genes constitute a single gene sub-family within the larger SET gene family, which is supported by phylogenetic analysis (Fig. 1.2). Ehmt1 and Ehmt2 are both present in many mammals, including humans, cows, sheep, monkeys, etc.; fish, including zebrafish, killifish (Nothobranchius furzeri), minnows (Cyprinodon variegatus), etc; frogs (Xenopus tropicalis); reptiles, including pit vipers (Protobothrops mucrosquamatus), geckos (Gekko japonicus), alligators (Alligator mississippiensis), etc.; and chickens (Gallus gallus). However, in Drosophila (fruit flies), only one ortholog exists, called Ehmt or G9a (Kramer et al. 2001, Mis, Ner, and Grigliatti 2006, Stabell et al. 2006). Genome-wide analysis in Ehmt mutant (loss of function) Drosophila larvae revealed that H3K9me2 was reduced at approximately 5% of the euchromatic genome, and affected up to one third of all fly genes and showed a preference for the 5’ and 3’ ends of genes. The prevalence of the Ehmt2 and Ehmt1 binary family across vertebrate classes, and the single ortholog in invertebrates, suggest that the gene duplication event, from which the two genes likely arose, may have occurred early in the evolution of vertebrates.   Fig. 1.2 Ehmt gene family. A phylogenetic tree generated using the neighbour joining method showing the evolutionary relationship between Drosophila EHMT and the mouse and human orthologs of EHMT1 and EHMT2/G9a. Analysis was performed with the Vector NTI software (Invitrogen). The scale bar indicates phylogenetic distance. Adapted from Kramer et al. (2011).!12Although Ehmt2 and Ehmt1 are thought to have a largely overlapping function and possess similar domain structures, the importance of the two are not equal. One interesting property of these proteins is their ability to interact with the very histone modifications that they deposit. Ehmt2 and Ehmt1 show affinity for H3K9me1/2 via the ANK repeat domain (Collins et al. 2008). One study showed that the heterochromatinization of Pou5f1 (Oct3/4) in response to retinoic acid is dependent on the ability of Ehmt1 to interact with H3K9me1. The binding of Ehmt1 to H3K9me1/2 increased its ability to dimethylate nearby lysine residues, and loss of this binding ability delayed pluripotent gene silencing in retinoic acid induced embryonic stem cells (Liu et al. 2015). Knock-in mice containing mutations in the ANK repeat domain had decreased the affinity of Ehmt1 for H3K9me1/2, and survived gestation but died shortly after due to congenital defects (Liu et al. 2015). However, loss of the entire Ehmt1 SET domain, and thus its methyltransferase activity, did not affect mouse viability (Inagawa et al. 2013), indicating that the affinity of Ehmt1 for H3K9me1/2 was more critical for life than its function as a lysine methyltransferase. Remarkably, analogous mutations in the Ehmt2 ANK domain had no notable effect on mouse development (Liu et al. 2015). Kramer (2016) proclaims that ‘these recent studies certainly challenge the dogma that Ehmt2 and Ehmt1 have the same function’ and summarised the two as having different molecular focus, within a complementary relationship. Figure 1.3 illustrates that the ANK domain of Ehmt1 can compensate for its homologue in Ehmt2, but not the other way around, while the SET domain of Ehmt2 can compensate for its homologue in Ehmt1, but not the other way around. !13  As findings about the histone code proliferated in the late 1990s and 2000s in the new era of epigenetics, most of the Ehmt1/2 studies in animal models focused on H3K9me2 as a repressive mechanism of gene regulation. And since Ehmt2 is the predominant methyltransferase over its homologue, I gave my attention to this gene and the existing plethora of interesting characterisations in a variety of biomedical aspects, including germ cell development, heart development, lymphocyte development, leukemia, drug addiction, cognition, adaptive behaviours, etc. (Fig. 1.4), which I will detail in the following pages. Fig. 1.3 Compensatory relationshiop between Ehmt2 and Ehmt1, based on viability of knockout mice. Adapted from Kramer (2016).!14Established animal studies of ehmt1/2   Embryonic stem cells One of the main evidence for the importance of Ehmt2 is its requirement in embryonic development. Ehmt2 knockout mice die during the early periods of gestation and have a global loss of H3K9me2 (Tachibana et al. 2002, 2005). Cultured embryonic stem (ES) cells derived from these embryos grow and divide normally but showed defects upon induction of differentiation with retinoic acid (RA) (Tachibana et al. 2002). Normally, during RA-induced differentiation, pluripotency markers such as Nanog and Pou5f1 (Oct3/4) become transcriptionally silenced. To keep them silent, the promoter regions need to be heterochromatinized, which is dependent on Ehmt2-mediated H3K9me2 and DNA methylation (Epsztejn-Litman et al. 2008, Feldman et al. 2006). Epsztejn-Litman et al., (2008) found that Ehmt2 recruited DNA methyltransferases by its ANK domain in addition to producing H3K9me2. As a result, Ehmt2 knockout cells had prolonged expression of Nanog and Pou5f1 (Oct3/4), and reverted to a pluripotent state, Ehmt2/G9aES cell differentiationDong et al. 2008, Yamamizu et al. 2012germ cell developmentTachibana et al. 2007heart developmentInagawa et al. 2013lymphocyte developmentThomas et al. 2008, Lehnertz et al. 2010leukemiaLehnertz et al. 2014drug addictionMaze et al. 2010cognition & bahaviourSchaefer et al. 2009, Kramer et al. 2011embryogenesisTachibana et al. 2002e i urFig. 1.4 Summary of existing research on Ehmt2 in different tissue systems.!15suggesting that it was required for the maintenance of the differentiated identity (Yamamizu et al. 2012, Epsztejn-Litman et al. 2008, Feldman et al. 2006). Ehmt1, on the other hand, is involved in embryonic development in a different way. This is the case where Ehmt1 was discovered to use its ANK repeat domain to ‘read’ the H3K9 methylation code. The loss of this function lead to congenital defects associated with the musculoskeletal system that caused an inability to eat, eventually leading to starvation (Liu et al. 2015). The same function in Ehmt2 is completely dispensable. The SET domain in Ehmt1 is also dispensable in embryonic development (Inagawa et al. 2013). Surprisingly, the sole ortholog of Ehmt2 and Ehmt1 in Drosophila, Ehmt/G9a, is entirely dispensable for development (Seum et al. 2007, Kramer et al. 2011). The contrast with mouse suggests that the gene family’s evolution was coupled with increasing involvement in embryonic development. Kleefstra syndrome Although the developmental role of Ehmt1 was only recently elucidated in mice, its importance was first known in humans, as the genetic basis of Kleefstra Syndrome (OMIM #610253). First reported in 2004, deletions in 9q34.3 were associated with patients who display congenital anomalies, including brachy(micro)cephaly, flat face with hypertelorism, synophrys, anteverted nares, everted lower lip, carp mouth with macroglossia, severe mental retardation, hypotonia, epileptic seizures, and heart defects (Harada et al. 2004, Iwakoshi et al. 2004, Stewart et al. 2004, Neas et al. 2005). It was later found that most of these cases were due to loss of function mutations in EHMT1, which was happloinsufficient (Kleefstra et al. 2006, 2009, Yatsenko et al. 2009, Verhoeven et al. 2011). Although most were de novo mutations, two unrelated families were !16found to have disease inheritance. In both cases, affected children inherited an interstitial 9q34.3 deletion from a mildly affected mother who was somatically mosaic for the deletion. Willemsen et al., (2011) noted the implications for genetic counselling and emphasized that multiplex ligation-dependent probe amplification (MLPA) may be necessary to detect mosaicism. In cases not caused by EHMT1 loss of function, Kleefstra et al., (2012) identified de novo mutations in four epigenetic regulators that are related to EHMT1, namely MBD5, MLL3, SMARCB1, and NR1I3. In Drosophila, MBD5, MLL3, and NR1I3 cooperate with EHMT1, while SMARCB1 was known to interact directly with MLL3. Much of the craniofacial phenotype was replicated in mouse models of Ehmt1 mutation. Balemans et al., (2014) reported that Ehmt1wt/null mice showed brachycephalic crania, a shorter or bent nose, and hypertelorism. At P28, researchers found significant upregulation of Runx2 and several other bone related genes. Furthermore, it was associated with decreased H3K9me2 in the promoter regions of these genes. Liu et al., (2015) further pointed the underlying mechanism to the ANK repeat domain of Ehmt1, using a mouse model with a three-amino acid mutation that disabled the H3K9me2 code-reading function. The involvement of Ehmt2 in this aspect was previously unknown, and will be discussed in chapter 3. Cognition and learning In addition to the information about mental retardation in Kleefstra Syndrome, a great number of studies uncovered the roles of Ehmt2 and Ehmt1 in the nervous system. Firstly, H3K9me2 is dynamically regulated in the brain in response to contextual fear conditioning, a classic associative memory assay for rats that is performed by pairing a neutral sensory input (a novel !17context or cage) to an electrical shock. H3K9me2 increased in the hippocampus 1 hour after contextual fear conditioning, however this increase was also observed in response to context alone (Gupta et al. 2010, Gupta-Agarwal et al. 2012). In contrast, in the entorhinal cortex (EC), H3K9me2 was specifically increased in reponse to fear conditioning and not context alone. Chemical inhibition of Ehmt2 in the hippocampus immediately before fear conditioning lead to reduced memory, whereas in the EC it enhanced memory. H3K9me2 was also increased in the lateral amygdyla (LA) of rats in response to auditory fear conditioning, a memory paradigm that pairs a tone with an eletrical shock. Blocking this increase through chemical inhibition of Ehmt2 decreases memory. Enhancing LA-specific H3K9me2 by inhibition of the H3K9me2 demethylase LSD1 increased memory (Gupta-Agarwal et al. 2014), suggesting a direct association between H3K9me2 levels in the LA and memory formation. These studies illustrate the dynamic regulation of H3K9me2 in the brain in response to environmental cues encountered during fear conditioning and demonstrate an acute role for Ehmt2 in memory formation. Evidence of a role for Ehmt1 in memory formation has been found through genetic manipulation in mice. Postnatal conditional knockout of Ehmt1 in the mouse forebrain caused loss of fear-induced memory, as well as reduced exploratory behaviour, reduced awareness of danger, and reduced reward seeking behaviour. These behavioural defects were accompanied by increased brain expression of a number of genes that were normally repressed in neurons. This may indicate that Ehmt1 is required for epigenetic repression of non-neuronal genes in neuronal tissues (Schaefer et al. 2009). In the Ehmt1wt/null  genetic mouse model of Kleefstra Syndrome, there was also reduced exploration, increased anxiety, reduced dendritic branching in the hippocampus, deficits in fear distinction learning, and deficits in spatial object recognition (Balemans et al. 2010, 2013, 2014). These mice also had increased expression of skeletal genes in the brain, which was accompanied by reduced H3K9me2 at these gene, suggesting Ehmt1 may be required to suppress non-neuronal genes in the brain. !18Further evidence solidifying an evolutionarily conserved role for Ehmt2 and Ehmt1 in complex brain function came from analysis of fruit flies with loss of function mutations in the Drosophila Ehmt/G9a ortholog, which resulted in defects in sensory dendrite branching, larval locomotory behaviour, non-associative learning, and courtship memory. Aside from these phenotypes, Ehmt mutant flies were completely normal in most aspects of neural development, suggesting that the gene regulates specific adult cellular processes in flies and is not a general regulator of the nervous system. Mental health and drug addiction Ehmt2 is important in the behavioural response of mice to chronic drug abuse. Chronic cocaine exposure in mice leads to the development of addiction-like behaviours, such as preference for a chamber that is associated with injection of cocaine over a chamber associated with a saline injection. The nucleus accumbens (NAc) is central to the brains reward circuitry and is critical in the development of drug preference. Repeated cocaine exposure caused reduction of Ehmt2, Ehmt1, and H3K9me2 in the NAc. Overexpression of Ehmt2 in the NAc increased H3K9me2 levels and decreased the development of cocaine preference. Knockdown of Ehmt2 or treatment with chemical inhibitors decreased H3K9me2 levels and increased cocaine preference (Maze et al. 2010). A similar effect was observed in mice that were chronically exposed to morphine, which caused downregulation of Ehmt2 and H3K9me2. Genome-wide H3K9me2 analysis showed enrichment in genes involved in glutamate receptor signalling (Sun et al. 2012). The methyl CpG DNA binding protein, MeCP2, may mediate reduction of Ehmt2 expression in response to morphine. MeCP2 was upregulated in the NAc and bound to the Ehmt2 promoter in this context (Zhang et al. 2014). !19A chronic social defeat stress assay in mice can induce antisocial behaviour and decreased reward seeking, which is thought to be analogous to mental depression in humans. Like chronic cocaine and morphine exposure, prolonged social defeat stress caused a subtle decrease in Ehmt2 and H3K9me2 in the NAc of mice. Post mortem brains of clinically depressed humans also had decreased Ehmt2 and H3K9me2. Repeated cocaine exposure in mice can increase their sensitivity to social stress. This type of response is intended to model increased rates of depression observed amongst drug addicts. Activation of Ehmt2 immediately before cocaine exposure prevented sensitivity towards stress, and the reverse was observed in knockdown studies. The mechanism appeared to be dependent on increased activity of BDNF signalling through the RAS/MAPK pathway. Ehmt2 bound and repressed Ras in the NAc (Covington et al. 2011). This study suggested that Ehmt2 can affect behaviour by altering the signalling potential of neurons, either increasing or decreasing the sensitivity to different stimuli. Further studies used neuronal subtype specific knockout in the striatum in mice. Knockout of Ehmt2 and Ehmt1 in dopamine receptor 1 (dr1) or dr2 expressing neurons altered the response of mice to specific dr1 and dr2 agonists, respectively, but did not otherwise affect general behaviour (Schaefer et al. 2009). Interestingly, conditional knockout of Ehmt2 in drd2 cells, but not drd1, caused increased cocaine addiction (Maze et al. 2014). Remarkably, conditional knockout of Ehmt2 in drd2 neurons caused a shift in the transcriptional identity, so that drd2 cells reduced expression of some drd1-enriched genes. Ehmt2 deficient drd2 neurons also projected to locations normally occupied by drd1 cells (Maze et al. 2014). This study suggests a role for Ehmt2 in neuronal subtype specification within a common brain region and further elucidated the complex role of Ehmt2 in the behavioural response to drug abuse. !20Immune response Several studies suggest a role for Ehmt2 in the regulation of immunity. Naive T helper (Th) cells can differentiate into several different subclasses of Th cells, which produce different cytokines depending on the infectious agent that stimulates their differentiation. Conditional knockout of Ehmt2 in mouse Th cells increased susceptibility to the gastrointestinal parasite Trichuris muris, through disruption of Th2 cell differentiation. In culture, Ehmt2 deficient Th cells failed to express the appropriate Th2 cytokines upon stimulation and showed inappropriate activation of Il-17 (Lehnertz et al. 2010). Ehmt2 deposited H3K9me2 at the Il-17 gene in naive Th cells, and in the absence of Ehmt2, the cells showed increased tendency towards Th17 differentiation (Antignano et al. 2014). Thus in mice, the optimal cytokine response to infection requires Ehmt2-mediated repression and activation of cell type specific cytokine genes. Ehmt2 may also play a role in the general repression of type I interferon, which is a potent proinflammatory cytokine that facilitates innate and adaptive immune responses against viruses and bacterial pathogens (Stetson and Medzhitov 2006). At steady state, IFN in the organism is largely produced by dendritic cells (Barchet, Cella, and Colonna 2005). Fang et al. (2012) found that H3K9me2, mediated by Ehmt2, is low in dendritic cells, compared to fibroblasts, at the IFN-β gene (Ifnb1) and other IFN-inducible genes. When Ehmt2 is knocked-out in fibroblasts, H3K9me2 is lost at the Ifnb1 gene, resulting in its increased capacity for IFN production, suggesting that this is an epigenetic mechanism that determines cell type–specific differences in IFN expression. However, low H3K9me2 did not directly lead to activation of genes, but rather established their expression potential. Kramer (2016) argues that Ehmt2 and H3K9me2 may be required to buffer genetic responsiveness to immune challenge. This idea is supported by a recent study investigating the role of Ehmt/G9a in the innate immune response to virus in Drosophila. Ehmt mutant flies die more rapidly than controls upon exposure to virus, however mutant flies do !21not show an increase in viral load, indicating that their ability to fight off infection is not compromised (Merkling et al. 2015). Innate viral immunity in flies is initiated by JAK/STAT signalling and downstream activation of transcription (Dostert et al. 2005). Some JAK/STAT-induced genes, such as Domeless and Socs36E, show reduced H3K9me2 in Ehmt mutant flies (Merkling et al. 2015, Kramer et al. 2011). These same genes show a hyperactive transcriptional response to virus (Merkling et al. 2015). Thus, Drosophila Ehmt appears to be required in flies to buffer the transcriptional activation of viral response genes. The mechanisms through which an overactive response to virus causes harm is not known, however the buffering effect of Ehmt on viral-induced transcription appears to be essential for mediating tolerance to infection. Leukemia and hematopoiesis Inhibition of Ehmt1 and 2 delays the differentiation of human hematopoietic stem cells (HSCs) ex vivo (Chen et al. 2012), which suggests they may have roles in early hematopoiesis. In conditional knockout models of Ehmt2 in the murine hematopoietic compartment, Lehnertz et al. (2014) found no discernible function for Ehmt2 in hematopoietic stem cells, despite it acting as a selective regulator of fast proliferating myeloid progenitors. However, in mouse models of acute myeloid leukemia (AML), loss of Ehmt2 significantly delays disease progression and reduces leukemia stem cell (LSC) frequency. This function was further connected to Ehmt2’s methyltransferase activity and its interaction with the leukemogenic transcription factor HoxA9 and provided evidence that primary human AML cells are sensitive to Ehmt2 inhibition. A clinical potential was thus established of Ehmt2 inhibition as a means to counteract the proliferation and self-renewal of AML cells by attenuating HoxA9-dependent transcription. !22Heart Both Ehmt2 and Ehmt1 appear to be essential for heart development. In mice with cardiomyocyte-specific conditional knockout of Ehmt1 and knockdown of Ehmt2, severe cardiac defects (atrioventricular septal defects, AVSD) were coupled with the loss of H3K9me2. Inagawa et al. (2013) observed upregulation of several non-cardiac genes in the hearts of these mice, suggesting that Ehmt2 and Ehmt1 are again playing a repressive role, via H3K9me2, to ensure that only the correct genes are expressed for the correct cell identity. However, in adult cardiac progenitor cells, chemical inhibition of Ehmt2 promoted cell expansion without changing their phenotype or differentiation potential (Kaur et al. 2016). It is possible that Ehmt2 maintains the cardiomyocyte specificity, which is important during development, but in the adult, it could be blocked to encourage a more stem cell state. Furthermore, Ehmt2 and Ehmt1 have roles in heart diseases. In a rat model of dilated cardiomyopathy, Ehmt2 RNA interference (RNAi) was applied to primary neonatal cardiomyocytes. The results indicated that Ehmt2 placed a repressive force on the expression of cell adhesion molecules, which are highly expressed in dilated cardiomyopathy and contribute to chronic degeneration in cardiac incompetence. Thus, Chen et al. (2015) suggest that Ehmt2 could ameliorate dilated cardiomyopathy, indicating its potential for use in the treatment of the disease. In a study by Han et al. (2016), expression of Ehmt1 & 2 was activated by pathological stress on the heart to form a repressive complex. They found that Brg1 (nucleosome remodelling factor) recruited Ehmt2, Dnmt3a, and Dnmt3b to carry out H3K9me2 and DNA methylation on a key molecular motor gene Myh6, thereby impairing cardiac contraction. !23Aim and scope of this work As one can see, there is already a large and growing body of knowledge about Ehmt1 & 2, portraying these genes as playing diverse and often contrasting roles in different tissue systems. The list does not end here, multiple new studies are focusing on their involvement in skeletal muscle and adipose tissues, for example. The current understanding of those, along with more details about Ehmt2 in cranial development, are the subject of this study. The research described in the ensuing chapters aims to contribute to the growing body of understanding of Ehmt2. Using different conditional and inducible conditional knockout models in mice, this work aims to describe phenotypic changes caused by the lack of Ehmt2, and thus uncover its role in each particular tissue and cell type. The results pay particular attention to developmental biology and describe effects at the level of tissue morphology and cellular behaviour. Chapter 2 is devoted to the study of Ehmt2 in skeletal muscle development as well as tissue regeneration. Chapter 3 focuses on the role of Ehmt2 in congenital anomalies, with specific emphasis on craniofacial development. Ehmt2 in adipose tissue development is covered in Chapter 4. !24  SKELETAL MUSCLE: brief background One of the tissues of interest in this work is skeletal muscle. Skeletal muscle is the voluntary one of the three types of muscles, the other two being cardiac and smooth muscles. It works with nerves, connective tissue, and vasculature to provide the body with structure and the capability to perform voluntary, precise movements. Due to this indispensable role, many intricate histological, biochemical, and biological studies have been done to characterize skeletal muscle. These studies have culminated in the large body of knowledge we currently have regarding skeletal muscle. Each muscle is encased within a layer of connective tissue called epimysium. Within it, there are multiple muscle bundles (fascicle), each is bound by a layer of connective tissue called perimysium, which contains multiple muscle fibres. A muscle fibre, or myofibre, is a long, tubular, multi-nucleated cell, also known as a myocyte. The muscle fibres, supplied by capillaries and Ehmt2/G9aES cell differentiationDong et al. 2008, Yamamizu et al. 2012germ cell developmentTachibana et al. 2007heart developmentInagawa et al. 2013lymphocyte developmentThomas et al. 2008, Lehnertz et al. 2010leukemiaLehnertz et al. 2014drug addictionMaze et al. 2010cognition & bahaviourSchaefer et al. 2009, Kramer et al. 2011embryogenesisTachibana et al. 2002skeletal muscle developmentmuscle regenerationcraniofacial developmentbrown adipose tissue developmentehaviourFig. 1.5 Summary of existing research on Ehmt2 in black and the focus of this work in red.!25innervated by the somatic nervous system, are separated by connective tissue called endomysium, which lies above the cell membrane. The insides of muscle fibres are filled with structures called myofibrils, the basic rod-like units of a muscle cell. They are composed of long proteins including actin, myosin, and titin, and other proteins that hold them together. These proteins are organised into thick and thin filaments called myofilaments, which repeat along the length of the myofibril in sections called sarcomeres. In vertebrate animals, all skeletal muscles are striated muscle, in which the filaments are organised in transverse bands (Fig. 1.6). Muscles contract by sliding the thick (myosin) and thin (actin) filaments along each other. !26   The development of skeletal muscle begins in the somites, which arises from the vertebrate mesoderm in embryo. The mesoderm generates almost all of the organs between the ectodermal wall and the endodermal tissues. The trunk mesoderm of a neurula-stage embryo can be subdivided into four regions, one of which, flanking the notochord on both sides is the paraxial, or somitic, mesoderm, which appears to be specified by the antagonism of BMP signalling by the Noggin protein. As the primitive streak regresses and the neural folds begin to gather at the centre of the embryo, the cells in this region will form somites. Satellite cellFAP/MSCPICPericytePerimysiumBasal laminaMuscle fibreMyonucleusBlood vesselFig. 1.6 Illustration of skeletal muscle tissue structure, cell types, and major regulatory factors.!27Mature somites contain three major compartments: the sclerotome, which forms the vertebrae and rib cartilage; the myotome, which forms the musculature of the back, rib cage, and ventral body wall; and the dermamyotome, which also contains skeletal muscle progenitor cells (including muscle progenitors that migrate into the limbs), as well as the cells that generate the dermis of the back. One of the key agents in determining where somites form is the Notch signalling pathway (see Aulehla & Pourquie, 2008). Cells located at the somite boundary instruct the cells anterior to them to epithelialise and separate. Non-boundary cells will not induce border formation, but they can acquire boundary-forming ability if an activated Notch protein is electroporated into them (Sato et al. 2002). To produce the segmentally defined pattern of somites, Notch activity and gene expression must oscillate. Critical gene expression that oscillates in response to Notch signalling include Hairy1 and Mesp2. Then, to carry out fissure formation and somite separation, the cells express Eph tyrosine kinases and their ligands, the ephrin proteins. These are able to elicit cell-cell repulsion between the posterior somite and migrating neural crest cells. Somite formation is also synchronized on both sides of the embryo (Palmeirim et al. 1997), which depends on a “clock-and-wave” mechanism in which an oscillating signal generated by the Notch pathway sets the periodicity of the process (the “clock”), and a rostral-to-caudal gradient provides a moving “wave” of FGF activity. Each oscillation of FGF organises groups of presomitic cells that will then segment together (see Maroto et al. 2012). Although all somites look identical, they will form different structures. For instance, the somites that form the cervical vertebrae of the neck and lumbar vertebrae of the abdomen are not capable of forming ribs; ribs are generated only by the somites that form the thoracic vertebrae, and this specification of thoracic vertebrae occurs very early in development. In contrast to the early commitment of the positional identity of the presomitic mesoderm along the anterior-posterior !28body axis, the commitment of the cells within a somite toward a particular cell fate occurs relatively late, after the somite has already formed. The dermamyotome comes from some of the epithelial portions of the somite. Fate mapping with chick-quail chimeras (Ordahl and Le Douarin 1992; Brand-Saberi et al. 1996; Kato and Aoyama 1998) has revealed that the dermamyotome is arranged into three regions. The cells in the two lateral portions of this epithelium (i.e., the dorsomedial and ventrolateral lips closest to and farthest from the neural tube, respectively) are the myotomes and will form muscle cells. Muscle precursor cells will migrate beneath the dermamyotome to produce a lower layer of muscle precursor cells, the myoblasts. Those myoblasts in the myotome closest to the neural tube form the centrally located primaxial muscles, which include the intercostal musculature between the ribs and the deep muscles of the back; those myoblasts formed in the region farthest from the neural tube produce the abaxial muscles of the body wall, limbs, and tongue. Various transcription factors, such as Prox1, distinguish the primaxial and abaxial muscles. The central portion of the dermamyotome gives rise to a third population of muscle cells (Gros et al. 2005; Relaix et al. 2005). This part of the somite also undergoes an epithelial-mesenchymal transition, during which the FGF signals from the myotome activate the transcription of the Snail2 gene in the central dermamyotome cells. After cell division, the ventral daughter cell joins the other myoblasts from the myotomes, while the other daughter cell locates dorsally, becoming a precursor of the dermis. The muscle precursor cells that delaminate from the epithelial plate to join the primary myotome cells remain undifferentiated, and they proliferate rapidly to account for most of the myoblast cells. While most of these progenitor cells differentiate to form muscles, some remain !29undifferentiated and surround the mature muscle cells. These undifferentiated cells become the satellite cells responsible for postnatal muscle growth and muscle regeneration. All the skeletal musculature in the vertebrate body with the exception of the head muscles comes from the dermamyotome of the somite. The major transcription factors associated with (and causing) muscle development are the myogenic regulatory factors (MRFs). This family of transcription factors include Myod, Myf5, Myog (Myogenin), and Myf6 (aka MRF4). These proteins share a homologous basic helix-loop-helix (bHLH) domain required for DNA binding and also for interaction with the E-protein family of transcription factors. MRF-E protein heterodimers then bind a specific CANNTG sequence, otherwise known as the E-box, on the promoters of many muscle-specific genes (Parker et al. 2003). Although E-proteins are not exclusively expressed in skeletal muscle, MRFs are muscle specific. This specificity of MRFs in combination with their timed expression allows MRFs to specifically direct myogenesis. All four MRFs are considered master regulators of muscle in that they can each activate the complete myogenic program when introduced to nonmyogenic cells. They bind to and activate genes that are critical for muscle function. For instance, the MYOD proten appears to directly activate the muscle-specific creatine phosphokinase gene by binding to the DNA immediately upstream from it (Lassar et al. 1989). MYOD also directly activates its own gene transcription. The MRFs act during different stages of myogenic progression (Zammit et al. 2004). Much of what is known about the roles of MRFs comes from studies performed using knockout animals. The presence of multipotent, putative myogenic progenitors, but complete lack of muscle in Myf5:Myod1 double knockout mice suggests an early role for Myf5 and Myod1 in myogenic precursor specification (Rudnicki et al. 1993). The severe lack of differentiated myofibers in Myogenin knock-out mice, despite the presence of myogenic progenitors, along with increased Myogenin expression and deficient myogenesis in Myf6 null mice suggest that the latter two !30MRFs, Myf6 and Myogenin, play a role in myogenic differentiation (Hasty et al. 1993; Nabeshima et al. 1993; Patapoutian et al. 1995; Rawls et al. 1995; Venuti et al. 1995). In general, Myf5 and Myod1 are termed “early MRFs” regulating myogenic specification and Myf6 and Myogenin are the “late MRFs” regulating myogenic differentiation (Francetic and Li 2011). However, the coordination of this process is more complex than this, as environmental signals to activate the satellite cells are mediated by a plethora of extracellular signals from other cells in the vicinity, through growth factors, cytokines, and cell-cell interactions (Paylor et al. 2011; Gopinath & Rando 2008). Studies using transplantation and knockout mice indicate that the primaxial myoblasts from the medial portion of the somite are induced by factors from the neural tube—probably Wnt1 and Wnt3a from the floor plate of the neural tube (Munsterberg et al. 1995; Stern et al. 1995; Borycki et al. 2000). These induce the Pax3-containing cells of the somite to activate the Myf5 gene in the primaxial myotome. Myf5 (in concert with Mef2 and either Six1 or Six4) activates the Myog and Myf6 genes, whose proteins activate the muscle-specific batteries of genes (Buckingham et al. 2006). As the myoblasts mature, they migrate along fibronectin cables, and eventually align, fuse, and elongate to become the deep muscles of the back, connecting to the developing vertebrae and ribs (Deries et al. 2010; 2012). The abaxial myoblasts that form the limb and ventral body wall musculature arise from the lateral edge of the somite. Two conditions appear necessary to produce these muscle precursors: 1) the presence of Wnt signals and 2) the absence of BMPs (Marcelle et al. 1997; Reshef et al. 1998). Wnt proteins (especially Wnt7a) are made by the epidermis (Cossu et al. 1996; Pourquie et al. 1996; Dietrich et al. 1998), but the BMP4 made by the adjacent lateral plate mesoderm would normally prevent muscles from forming. Recent studies (Gerhart et al. 2006; 2011) have found that the somites have attached at their tips a population of cells that secrete the BMP inhibitor Noggin. !31Once BMP is inhibited, Wnt7 can induce Myod in the competent dermamyotome cells, which activates the battery of MRF proteins that generate the muscle precursor cells. Another population of cells—the migrating neural crest cells—also affects myotome development in the somite. As trunk neural crest cells begin to migrate through the early somite, the Delta-expressing neural crest cells touch the Notch-containing membranes of the primaxial myotome cells, which helps induce Myf5 expression in those cells (Rios et al. 2011). Moreover, the neural crest cells secrete neuregulin-1, a paracrine factor that prevents the premature differentiation of myoblasts into muscle cells (Ho et al. 2011). Several myoblasts align together and fuse their cell membranes to form a myofibre (Konigsberg 1963; Mintz and Baker 1967; Richardson et al. 2008). Studies on mouse embryos show that by the time a mouse is born, it has the adult number of myofibres, and that these multinucleated myofibres grow during the first week after birth by the fusion of singly nucleated myoblasts (Ontell et al. 1988; Abmayr and Pavlath 2012). After the first week, muscle cells can continue to grow by the fusion of satellite cells into existing myofibres and by an increase in contractile proteins within the myofibres. The first step in myoblast fusion requires the cells to exit the cell cycle, and this is thought to be accomplished by Hedgehog signals (Osborn et al. 2011). Next, the myoblasts secrete fibronectin and other proteins onto their extracellular matrices and bind to it through α5β1 integrin, a major receptor for these extracellular matrix components (Menko and Boettiger 1987; Boettiger et al. 1995; Sunadome et al. 2011). The signal from the integrin-fibronectin attachment is critical for instructing myoblasts to differentiate into myofibres. The third step is the alignment of the myoblasts into chains. This step is mediated by cell membrane glycoproteins, including several cadherins (Knudsen 1985; Knudsen et al. 1990). Recognition and alignment between cells take !32place only if the cells are myoblasts. In the fourth step, calcium ions are critical, and fusion can be achieved by calcium transporters, such as A23187, that carry calcium ions across cell membranes (Shainberg et al. 1969; David et al. 1981). Fusion appears to be mediated by a set of metalloproteinases called meltrins. Yagami-Hiromasa et al. (1995) found that one of these, meltrin-α, is expressed in myoblasts at about the same time that fusion begins. As the myoblasts become capable of fusing, Myog becomes active. It binds to the regulatory region of several muscle-specific genes and activates their expression. Thus, whereas Myod and Myf5 are active in the lineage specification of muscle cells, Myog appears to mediate muscle cell differentiation (Bergstrom and Tapscott 2001). Cell fusion ends with the re-sealing of the newly apposed membranes. This is accomplished by proteins such as myoferlin and dysferlin, which stabilise the membrane phospholipids (Doherty et al. 2005). After the original fusion of myoblasts, the myofibre secretes the paracrine factor interleukin 4 (IL4). Horsely et al. (2003) found that IL4 secreted by new myofibres recruits other myoblasts to fuse with the tube, thereby forming the mature myofibre. The number of myofibres in the embryo and the growth of these fibres after birth appear to be negatively regulated by myostatin, a member of the TGF-β family (McPherron et al. 1997; Lee 2004). Myostatin is made by the developing and adult skeletal muscle and most probably works in an autocrine fashion. The human skeletal muscle presents one of the best example systems of adult tissue regeneration. In healthy individuals, muscle tissue repairs after damage, and regenerates by a series of steps of cellular mechanisms known as adult myogenesis. As it is mechanistically different from embryonic myogenesis (Wang and Conboy 2010), the study of adult myogenesis focuses on exploring the regenerative process and capability of adult stem cells, which are important in regenerative medicine.  !33The principal stem cell in adult myogenesis is the satellite cell, which resides between the myofibers and basement membrane of the muscle bundle (Buckingham et al. 2006). Quiescent satellite cells are identified by their expression of Pax7, a paired homeobox transcription factor partly responsible for survival and specification of the myogenic cell lineage (Seale et al. 2000). More importantly, it is a required marker of embryonic cells that will become the adult satellite cell population (Lepper & Fan 2010). Following traumatic myofiber damage or temporal progression of myopathy, these satellite cells become activated and readily proliferate, differentiate, and give rise to myoblasts, which fuse with damaged myofibers or form new myofibers. Ablation of Pax7-expressing cells results in the depletion of satellite cells and failure of muscle regeneration (Lepper et al. 2011). Epigenetics of myogenesis Recent focus on epigenetic research in myogenesis has yielded important insights into the mechanism of the myogenic program regulation in satellite cells and transient amplifying myoblasts. Mechanisms of epigenetics in adult myogenesis can involve chromatin remodelling, histone modification, DNA methylation, and silencing by microRNA. As a part of the epigenetic code, modifications on the unstructured ends of histone proteins, including acetylation, methylation, phosphorylation, and ubiquitination, may regulate the transcription of genes needed for myogenesis (Klose & Zhang 2007, Ruthenburg et al. 2007). Such regulation is crucial for the three fate decisions of satellite cells: proliferation, differentiation, and self-renewal (Olguin et al. 2007). After committed myogenic progenitors are produced, changes in gene expression also dictate the steps of myoblast proliferation, migration, differentiation, and fusion. Current studies try to uncover the relationship between histone modifications, the muscle specific regulatory factors (Myf5, Myod, Myogenin, Myf6), and the principal signalling pathways (MAPK, IGF1, PKB/AKT). Sartorelli et al. (1999), Polesskaya et al. (2000), Berkes & Tapscott (2005) have found that !34histone acetyltransferases can interact with Myod, a key regulator in the undifferentiated proliferating myoblast (Sartorelli & Caretti 2005, Tapscott 2005), and acetylate the histones associated with muscle specific genes, thereby activating the transcription of those genes. Tao et al. (2011) reported that Set7 interacts with Myod to activate genes, which is required for myoblast differentiation. Set7 is a histone methyltransferase responsible for methylation of histone H3 at Lys4 (H3K4me). Knockdown of Set7 or expression of a dominant-negative Set7 mutant impairs skeletal muscle differentiation, accompanied by a decrease in levels of H3K4me. This histone modification is mutually exclusive to H3K9me by Suv39h1, which silences muscle-specific genes. Other marks such as H3K27me3 by EZH2 is responsible for the repression of genes that are not specific to myogenesis, thus favouring differentiation. Without EZH2, satellite cells start to express non-satellite cell genes and would lose their identity (Juan et al. 2011). Recent genome-wide analyses have uncovered dynamic epigenetic changes during myogenesis (Asp et al. 2011), including the loss of H2B ubiquitination (Vethantham et al. 2012). ChIP-seq analyses have also revealed the importance of bivalent domains containing both H3K4me1 and H3K27ac in regulating muscle enhancers during myogenesis (Asp et al. 2011). These data also point to Myod as playing a key role in the recruitment of chromatin modifying enzymes and transcription factors to activate such enhancers (Blum et al. 2012; Blum and Dynlacht 2013). Craniofacial development: brief background Another tissue system investigated in this work is the development of the tissues of the head, especially craniofacial bones. Two distinct lineages generate these bones—the paraxial mesoderm and the cranial neural crest. Unlike the paraxial mesoderm described in the previous section, the neural crest is derived from the ectoderm; yet it is so important that it has sometimes been called the “fourth germ layer” (see Hall 2009). It has even been said, somewhat hyperbolically, that “the only interesting thing about vertebrates is the neural crest” (Thorogood 1989). Certainly, the !35emergence of the neural crest is one of the pivotal events of animal evolution, as it led to the jaws, face, skull, and sensory ganglia of the vertebrates (Northcutt and Gans 1983). The neural crest is a transient structure. Adults do not have a neural crest, nor do late-stage vertebrate embryos. Rather, the cells of the neural crest undergo an epithelial-mesenchymal transition from the dorsal neural tube, after which they migrate extensively to generate a prodigious number of differentiated cell types. Although neural crest cells are not apparent until they emigrate from the neural tube, induction of these cells first occurs during early gastrulation, at the border between the presumptive epidermis and the presumptive neural plate (the region that will form the central nervous system). Then as neural crest cells disperse throughout the body, they enter different tissues and form different cell types. The neural crest can be divided into four main anatomical regions, each with characteristic derivatives and functions. Only one of these, the cranial (cephalic) neural crest produces the craniofacial mesenchyme, which differentiates into the cartilage, bone, cranial neurons, glia, pigment cells, and connective tissues of the face. These cells also enter the pharyngeal arches and pouches to give rise to thymic cells, the odontoblasts of the tooth primordia, and the bones of the middle ear and jaw. Of these, only the cells from the midbrain and rhombomeres 1 and 2 of the hindbrain migrate to the first pharyngeal arch (the mandibular arch), forming the jawbones as well as the incus and malleus bones of the middle ear. These cells will also differentiate into neurons of the trigeminal ganglion—the cranial nerve that innervates the teeth and jaw—and will contribute to the ciliary ganglion that innervates the ciliary muscle of the eye. These neural crest cells are also pulled by the expanding epidermis to generate the frontonasal process, the bone-forming region that becomes the forehead, the middle of the nose, and the primary palate. Thus, these particular neural crest cells generate much of the facial skeleton. !36Bones form in two major ways. In one way, mesenchyme becomes cartilage and the cartilage is replaced by bone; this is called endochondral ossification. The other type of bone formation, where mesenchyme forms bones directly, is called intramembranous ossification. Both the paraxial mesoderm-derived and the neural crest-derived mesenchyme undergo intramembranous ossification in forming the face and the skull. The pathway to intramembranous bone begins when cranial neural crest and paraxial mesodermal cells, under the influence of BMPs from the head epidermis, proliferate and condense into compact nodules. High levels of BMPs induce these nodules to become cartilage, whereas lower levels of BMPs induce them to become pre-osteoblast progenitor cells that express the Runx2 transcription factor and the mRNA for collagens II and IX. Later, these cells downregulate Runx2 and begin expressing the osteopontin gene, giving them a phenotype similar to that of a developing chondrocyte; thus, this stage is called a chondrocyte-like osteoblast. Under the influence of Indian hedgehog (which it secretes and probably receives in an autocrine fashion), the chondrocyte-like osteoblast becomes a mature osteoblast—a committed bone precursor cell (Abzhanov et al. 2007). The osteoblasts secrete a collagen-proteoglycan osteoid matrix that is able to bind calcium. Osteoblasts that are embedded in the calcified matrix become osteocytes (bone cells). As calcification proceeds, bony spicules radiate out from the region where ossification began. Furthermore, the entire region of calcified spicules becomes surrounded by compact mesenchymal cells that form the periosteum (a membrane of cells that surrounds bone). The cells on the inner surface of the periosteum also become osteoblasts and deposit matrix parallel to the existing spicules. In this manner, many layers of bone are formed. The vertebrate skull, or cranium, is composed of the neurocranium (skull vault and base) and the viscerocranium (jaws and other pharyngeal arch derivatives). Skull bones are derived from both the cranial neural crest and the paraxial mesoderm (Le Lievre 1978; Noden 1978; Evans and !37Noden 2006). While the neural crest origin of the viscerocranium has been well documented, the contributions of cranial neural crest cells to the largely mesoderm-derived skull vault are more controversial. In 2002, Jiang et al. constructed transgenic mice that expressed β-galactosidase only in their cranial neural crest cells. When the embryonic mice were stained for β-galactosidase, the cells forming the anterior portion of the head—the nasal, frontal, alisphenoid, and squamosal bones—turned blue; the parietal bone of the skull did not. The boundary between neural crest-forming head bone and mesoderm-forming head bone is between the frontal and parietal bones (Yoshida et al. 2008). Although the specifics may vary among the vertebrate groups, in general the front of the head is derived from the neural crest while the back of the skull is derived from a combination of neural crest-derived and mesodermal bones. The neural crest contribution to facial muscle mixes with the cells of the paraxial mesoderm, such that facial muscles probably also have dual origins (Grenier et al. 2009). Given that the neural crest forms our facial skeleton, it follows that even small variations in the rate and direction of cranial neural crest cell divisions will determine what we look like. Moreover, since we look more like our biological parents than our friends do (at least, we hope this is true), such small variations must be hereditary. The regulation of our facial features is probably coordinated in large part by numerous paracrine growth factors. BMPs (especially BMP3) and Wnt signalling cause the protrusion of the frontonasal and maxillary processes, giving shape to the face (Brugmann et al. 2006; Schoenebeck et al. 2012). FGFs from the pharyngeal endoderm are responsible for the attraction of the cranial neural crest cells into the arches as well as for patterning the skeletal elements within the arches. Fgf8 is both a survival factor for the cranial neural crest cells and is critical for the proliferation of the cells forming the facial skeleton (Trokovic et al. 2003; 2005; Creuzet et al. 2004; 2005). The FGFs work in concert with BMPs, sometimes activating them and sometimes repressing them (Lee et al. 2001; Holleville et al. 2003; Das and Crump 2012). !38FGFs also work in concert with Sonic hedgehog (Shh; Haworth et al. 2004), which is critical for the proper growth of the facial midline. Shh is also crucial for shaping the neural crest derivatives of the head. The epithelia (both neural and epidermal) of the dorsal part of the frontonasal process secrete FGF8, while the ventral epithelia of the frontonasal process secrete SHH. The crest-derived mesenchyme between the epithelia receives both signals. Where these signals meet is where a bird’s beak cartilage grows out; if the region of frontonasal process containing the FGF/SHH boundary is inverted in the chick, an upside-down beak forms (Hu et al. 2003; Abzhanov and Tabin 2004). Variants in some regulators and targets of paracrine pathways have been shown to cause normal facial variation, as have variants of the Pax3 gene, which is expressed in the cranial neural crest. Adipose tissue development: brief background The third tissue system investigated in this work is the adipose tissue. Our understanding of adipose tissue biology has progressed rapidly since the turn of the century. White adipose tissue has emerged as a key determinant of healthy metabolism and metabolic dysfunction; it has an important role in buffering nutrient availability and demand by storing excess calories and preventing the toxic accumulation of excess nutrients in non-adipose tissues (Lelliott and Vidal-Puig 2004). It also communicates with metabolically relevant organs by secreting so-called adipokines as part of a dynamic endocrine system that regulates nutrient partitioning into peripheral tissues (Scherer 2006). This realisation is paralleled only by the confirmation that humans can have heat-dissipating brown adipose tissue, an important contributor to energy balance and a possible therapeutic target for the treatment of metabolic disease. Brown adipose tissue maintains core body temperature in reponse to cold stress by generating heat, a process known as thermogenesis (Cannon and Nedergaard 2004). Like the unilocular white adipocytes, the multilocular brown adipocytes also accumulate and store lipids. However, they are distinct !39from white adipocytes in that their more abundant mitochondria are enriched with uncoupling protein 1 (UCP1), which uncouples substrate oxidation from ATP production so that heat is produced (Cannon and Nedergaard 2004). The revelation that adult humans possess such a potent metabolic tissue was accompanied by the discovery that UCP1-expressing thermogenic adipocytes can be found in white adipose tissue in the form of beige adipocytes (Loncar, Afzelius and Cannon 1988; Young, Arch and Ashwell 1984). In rodents, chronic cold exposure increases thermogenic capacity by also recruiting beige or ‘brite’ (brown in white) cells in white adipose tissue, resulting in ‘browning’ (Loncar, Afzelius and Cannon 1988; Young, Arch and Ashwell 1984). UCP1 expression and uncoupled respiration in stimulated beige adipocytes could equal that of stimulated brown adipocytes (Wu et al. 2012; Shabalina et al. 2013). In summary, when considering adipose tissue involvement in the regulation of energy balance, the contribution of three different ‘shades’ of fat—white, brown, and beige—should be considered. Developmental origins In the human embryo primitive fat lobules become visible between gestational weeks 14-16. Adipose tissue depots in the head and neck are the first to develop, followed by the trunk and then the limbs, and by 28 weeks they can be detected in all presumptive visceral and subcutaneous white adipose tissue locations (Poissonnet, Burdi, and Garn 1984). In mice, subcutaneous adipose tissue develops between embryonic days 14-18, whereas visceral adipose tissue develops postnatally (Wang et al. 2013). With the exception of facial adipose tissue, which derives from the neural crest (Billon et al. 2007), adipose tissue is generally considered to derive from the mesoderm (Billon and Dani 2012). Lineage tracing studies in mice have recently shown that visceral adipose tissue develops from Wilms tumour gene (Wt1)-expressing cells found specifically in the lateral plate mesoderm. By comparison, Wt1-expressing cells make no contribution to subcutaneous adipose tissue leaving the exact embryological origins of this depot !40unknown (Chau et al. 2014). These findings suggest that individual white adipose tissue depots arise from different anatomical regions of the mesoderm. There is further heterogeneity amongst individual visceral depots with regards to the contribution of Wt1 cells, demonstrating the complexity of white adipose tissue as a multi-depot organ. Mesenchymal stem cells give rise to cells of the adipocyte lineage through a process involving an early commitment step followed by terminal adipocyte differentiation. In humans the number of mature white adipocytes remains relatively constant throughout adult life, reflecting the fine balance between adipocyte death and adipogenesis. It has been estimated that approximately 10% of white adipocytes turnover each year (Spalding et al. 2008) and interestingly this rate was reported not to be influenced by age or changes in energy balance. The adipogenic differentiation of preadipocytes is tightly regulated by a well-defined transcription factor cascade (Rosen and MacDougald 2006) involving the transient expression of early transcriptional regulators (CEBPβ, CEBPδ) followed by the induction of critical adipogenic regulators (CEBPα, CEBPγ) which coordinate the expression of many adipocyte genes. By genetically marking PPARγ-expressing cells, Tang et al. (2008) traced the source of these progenitors to the adipose tissue vasculature. Further studies have also pointed to a perivascular or endothelial contribution to the adipose lineage (Gupta et al. 2012; Tran et al. 2012). However, there are also other non-adipose sources of progenitor cells to be considered; bone marrow-derived haematopoietic stem cells, for instance, appear capable of migrating to adipose tissue depots and undergoing adipogenic differentiation (Crossno et al. 2006; Tomiyama et al. 2008). Lineage-tracing studies have shown that brown adipocytes derive from the same progenitors (Myf5+ and Pax7+) as myocytes, which originate in the paraxial mesoderm, and that brown adipocytes can potentially derive from skeletal muscle satellite cells in adulthood (Seale et al. 2000; Lepper and Fan 2010; Yin et al. 2013; Seale et al. 2008). Moreover, it is known that !41expression of some myogenic microRNAs (miRNAs) persists in mature brown adipocytes (Timmons et al. 2007; Walden et al. 2009). Given the absence of the myogenic gene expression signature in white adipocytes and their precursors, it was expected that white adipocytes would derive preferentially from a Myf5- lineage (Seale et al. 2008; Timmons et al. 2007; Walden et al. 2009). However, this view was recently challenged when conditional depletion of Pten driven by Myf5Cre caused overgrowth of brown adipose tissue, but also a paradoxical overgrowth of specific white adipose tissue depots and a loss of others (Sanchez-Gurmaches et al. 2012). Subsequent lineage-tracing studies have demonstrated the presence of some Myf5+ adipocyte precursors in white adipose tissue, indicating that white adipocyte precursors can derive from both Myf5+ and Myf5- lineages (Sanchez-Gurmaches et al. 2012; Shan et al. 2013), although it might be speculated that those from the Myf5+ lineage become white adipocytes with beige potential. Unexpectedly, lineage-tracing studies using the endothelial marker vascular endothelial cadherin or the pre-adipocyte marker Zfp432 also suggested that some brown and white adipocytes could have endothelial origins, despite a recent study finding that multiple endothelial markers did not label adipocytes (Rodeheffer et al. 2008; Tran et al. 2012; Gupta et al. 2012; Berry and Rodeheffer 2013). Given that beige adipocytes can be derived from white adipocyte precursors in vitro by chronic treatment with PPARγ agonists, it is likely that beige adipocytes share the Myf5- lineage origin of most white adipocytes (Sharp et al. 2012; Petrovic et al. 2010; Digby et al. 1998; Petrovic et al. 2008). Beige adipocytes arise de novo in white adipose tissue in response to adrenergic stimulation. Subsequent lineage tracing studies indicated that self-renewing, tissue-resident Pdgfra+ precursors were a significant source of these newly formed beige adipocytes (Lee et al. 2012). !42What is the mesenchyme: developmental and adult perspectives Before exploring the findings of this work, it is important to discuss the concept of the mesenchyme, which can be interpreted in various different ways in human biology. The word “mesenchyme” itself often refers to the simple idea of “soft” tissues that can make up supporting structures. In this sense, it does not specify any particular developmental lineage, and can include all fibroblasts, vasculature, muscle, fat, cartilage, and bones. Historically, these tissues were thought to be all mesoderm-derived, which is no longer accurate given the current understanding of the neural crest and epithelial–mesenchymal transition. Thus, in developmental biology, this terminology is best kept as a loose and broad concept, and it is more meaningful to specify the development of individual tissues, such as skeletal muscle, calvarial bones, and brown adipose tissue, covered in the following chapters.     Mesodermbone fatmuscleParaxial mesodermFig. 1.7 Historically, skeletal muscle, bone, and fat tissues were all considered to have derived from the paraxial mesoderm. However, modern understanding concludes that they have similarities but also many differences in their developmental origin, due to the discovery of the neural crest cell lineage and multiple origins of different fat depots.!43This terminology is relevant to this study not only in the historical understanding of the mesenchymal tissues, but also due to the project’s focus on skeletal muscle, bone, and fat, which, save for the head region, all have significant developmental contributions from the paraxial mesoderm (Fig. 1.7). They have similarities but modern understanding has revealed many differences in their developmental lineage. As detailed in previous sections, a large portion of skeletal muscle and bones of the body are derived from the paraxial mesoderm. The major issue in this concept is the neural crest cell lineage, which has an ectodermal origin, and gives rise to some of the tissues in the head region. Thus it is important to consider the neural crest cell lineage when studying mesenchymal tissues of the head. As mentioned in the previous section, fat is a more complex tissue type due to its multiple developmental origins; brown fat is paraxial mesoderm derived, but visceral white fat is derived from the lateral plate mesoderm. Subcutaneous white fat, which is the other major type  in humans, might be derived from any mesodermal region, and is currently unclear if endothelial cells, which are derived from the lateral plate mesoderm, is a significant contributor. Mesenchymal stem cells (MSC), on the other hand, is a completely different concept, which is usually considered in adult tissues. They are multipotent cells, with fibroblast morphologies, that can differentiate into a variety of cell types. The specific types of cells they may develop into is still debated as MSCs in different tissues were found to have different potentials. These cells are important in tissue maintenance, repair, and regeneration. They may have a distinct developmental origin from the tissues they reside in. One interesting marker employed in this work, called Pdgfra (platelet-derived growth factor alpha), has a relationship with the mesenchymal concept in both developmental biology and adult tissues. Its expression is widespread in the mesoderm during embryogenesis, but in the adult, its expression is highly specific. In adult skeletal muscle, for example, resident MSCs are identified by !44Pdgfra, but other cells are not. Joe et al. (2010) reported it as an excellent marker of stem cells residing in muscle tissue, critical to tissue regeneration, that have fibrogenic and adipogenic potentials. Thus, developmentally, Pdgfra may contribute to many individual mesenchymal tissues, but in the adult, it is a much more specific marker of MSCs. One of the mouse models employed in this work makes use of Pdgfra-Cre to drive the conditional knockout of Ehmt2 during development, but it is not an inducible conditional model, thus it is logical to present a brief introduction to Pdgfra here as a developmental marker. Platelet-derived growth factor (Pdgf) was first characterised as the major mitogenic component of serum (Kohler and Lipton 1974; Ross et al. 1974) and is known to initiate DNA synthesis in a wide variety of mesenchymal cells grown in culture (Raines et al. 1992). Subsequent studies have shown that Pdgf is, in addition, capable of modifying cell behaviour in a variety of ways (Ataliotis and Mercola 1997). Its signalling is mediated by two distinct, high-affinity receptors termed Pdgfr-α and -β (Yarden et al. 1986; Claesson-Welsh et al. 1989; Matsui et al. 1989). They are transmembrane proteins with intrinsic tyrosine kinase activity. The extracellular portion comprises five immunoglobulin-like domains and the intracellular portion contains the catalytically active kinase that is interrupted by a hydrophobic kinase insert region. This latter feature is characteristic of members of the class III (Ullrich and Schlessinger 1990) or Pdgfr family of receptor tyrosine kinases, which also includes KIT and FMS/CSF1R. Pdgfra and Kit are clustered together on chromosome 5 in mice and chromosome 4 in humans. Pdgfra genes from rat, mouse, and human are 90-95% similar, and they share 73% similarity with Xenopus Pdgfra. There is evidence that Pdgf-like molecules may function even in invertebrates such as Hydra and sea urchins (Hanai et al. 1987; Ramachandran et al. 1993; 1995; Govindarajan et al. 1995). !45PDGFRA binds both PDGF-A and -B chains with high affinity (Seifert et al. 1989). Our current understanding of the ligand-receptor interaction is that dimeric PDGF molecules initiate or stabilise receptor dimer formation (Bishayee et al. 1989; Heldin et al. 1989), which then triggers an intracellular signalling cascade. Dimerisation of PDGF receptors is accompanied by transphosphorylation between dimer partners of specific tyrosine residues within the intracellular domain (Heldin et al. 1989; Kelly et al. 1991). These phosphorylated tyrosines serve as high-affinity binding sites for a variety of membrane-associated and cytoplasmic proteins that contain src homology 2 (SH2) domains (Claesson-Welsh 1994; Kazlauskas 1994). It is largely these SH2 domain-containing proteins that mediate transmission of the PDGF signal inside the cell. Binding of particular SH2 domain-containing proteins to phosphotyrosines is regulated by flanking amino acid sequences (Songyang et al. 1993) that determine whether a given SH2 domain-containing protein can interact with the phosphorylated growth factor receptor. PDGFs are secreted, soluble proteins that might be expected to signal over long distances in the embryo. However, the analysis of ligand and receptor distribution in vivo suggests that both paracrine and autocrine signalling can occur, depending on the developmental stage or tissue. Furthermore, the patterns of ligand and receptor expression suggest that switching between paracrine and autocrine signalling occurs within particular tissues. Firstly, Pdgfa and Pdgfra are synthesised as maternal transcripts in mouse and Xenopus embryos (Mercola et al. 1988; 1990; Rappolee et al. 1988; Palmieri et al. 1992; Jones et al. 1993). In Xenopus embryos, mRNA levels of ligand and receptor decline during the initial stages of development, then increase shortly after the onset of zygotic transcription but in a spatially restricted pattern. !46In mice, ligand and receptor proteins are evenly distributed in two-cell- and blastocyst-stage embryos (Palmieri et al. 1992). At embryonic day 7.5, Pdgfra is produced in the mesoderm, with the exception of the primitive streak, whereas Pdgfa mRNA is present in adjacent ectodermal and endodermal layers (Orr-Urtreger et al. 1992; Orr-Urtreger and Lonai 1992; Palmieri et al. 1992). The patterns of production described previously suggest that at early stages of development, Pdgf signalling may be acting in an autocrine manner to stimulate cell proliferation. During the early periods of morphogenesis, it seems more likely that it acts in a paracrine manner to influence the migration of mesenchymal cells. After somitogenesis begins, Pdgfra production is no longer confined to mesodermal derivatives, but also in the neural crest, which is ectodermally derived (Morrison-Graham et al. 1992; Orr-Urtreger et al. 1992; Orr-Urtreger and Lonai 1992; Palmieri et al. 1992; Schatteman et al. 1992; Ho et al. 1994). In mice, it is produced widely throughout the cranial neural crest and is present in most, if not all, neural crest cells in the branchial arches (Morrison-Graham et al. 1992). However, neuronal derivatives of the neural crest do not produce PDGFRA prior to E16 (Morrison-Graham et al. 1992; Orr-Urtreger et al. 1992; Orr-Urtreger and Lonai 1992; Schatteman et al. 1992), although high levels of PDGFR-α and -β production are seen in neonatal rats (Eccleston et al. 1993). These production patterns suggest that the direct influence of PDGF is confined to only a subset of neural crest derivatives. Pdgfra mRNA is present throughout the presomitic mesoderm and the newly formed somite as epithelialization occurs, but as differentiation proceeds PDGFR-α production is maintained only in the sclerotome and downregulated in the dermamyotome (Morrison-Graham et al. 1992; Orr-Urtreger and Lonai 1992; Schatteman et al. 1992). Later still, the dermatome begins to produce PDGFR-α again, but the myotome does not, and instead produces PDGF-A (Orr-Urtreger and Lonai 1992). PDGF-A production is also widespread throughout the epidermal ectoderm of the !47mouse, thus leading to the apposition of receptor-producing cells in the dermis and sclerotome with ligand-producing cells in the epidermis and myotome. It appears that neurons, glia, and their precursors can all produce PDGFR-α depending on the region of the CNS and the time of development. For example, oligodendrocyte precursors probably produce PDGFR-α in vivo (Pringle and Richardson 1993) as they are known to do in vitro (Hart et al. 1989), and retinal ganglion neurons also produce PDGFR-α (Mudhar et al. 1993). PDGF-A is abundantly produced in neurons in the late (E18) mouse embryo (Yeh et al. 1991) but is also produced in glial cells in regions such as the developing optic nerve (Mudhar et al. 1993). However, a degree of uncertainty exists partly because many of these studies using in situ hybridization are dated and often lack sufficient detail to confidently assign labelling to specific cell types and also because of some discrepancies from different laboratories. PDGF-A is produced in the surface ectoderm and muscle masses of the mouse limb, whereas PDGFR-α is present throughout the limb bud mesenchyme and then in the perichondrium of the developing bone as production is lost from the condensing chondrogenic mesenchyme (Orr-Urtreger et al. 1992; Schatteman et al. 1992). PDGF may serve to prevent differentiation and/or promote the proliferation of chondrocytes in the limb because it can prevent chondrocyte differentiation in an in vitro system (Chen et al. 1992). This inhibition may be mediated by the production of tumor growth factor β (TGF-β) family members in the limb because TGF-β is known to stimulate production of PDGF-A and -B in microvascular endothelial cells (Kavanaugh et al. 1988). Intriguingly, both PDGF-A and PDGFR-α are produced in the apical ectodermal ridge (AER) of the limb bud (Morrison-Graham et al. 1992; Orr-Urtreger and Lonai 1992; Potts and Carrington 1993) along with many other growth factors. The role of PDGF here remains unclear but the AER !48is essential for normal limb development (Tickle and Eichele 1994). However, defects of the limb have not been reported in Patch mouse mutant embryos that lack the Pdgfra gene. The only exception to this has been in mice of a particular genetic strain carrying the Patch mutation in which the overall phenotype was more severe than previously described and limbs appeared poorly formed (Orr-Urtreger et al. 1992).!492. The role of Ehmt2/G9a in skeletal muscle development and regeneration INTRODUCTION: CURRENT UNDERSTANDING OF EHMT2 IN SKELETAL MUSCLE Despite Ehmt2 being capable of regulating a diverse range of cellular and biological functions, as described in the previous section, little is known about its role in skeletal muscle. Working with the widely used murine cell line C2C12, Ling, Bharathy, et al. (2012) recently highlighted Ehmt2 as a possible regulator of myogenic differentiation. Using in vitro overexpression and knockdown strategies, Ehmt2 was shown to act as an inhibitor of myotube formation. These experiments also showed that Ehmt2 without its SET methyltransferase domain did not have any effect in myoblasts, suggesting that methylation is its only relevant function in myogenesis. Biochemical analyses suggested that EHMT2 has the capability to directly methylate MYOD at K104 (Ling, Bharathy, et al. 2012) revealing a novel non-histone methylation mechanism that inhibits myogenic differentiation. Besides this, experiments employing in vitro cell lines also found Ehmt2 to be required for Bhlhe41/Sharp1-mediated repression of myogenesis; Bhlhe41/Sharp1 is a basic helix-loop-helix transcription factor belonging to the Hairy/Enhancer of Split subfamily. The correlation between Bhlhe41/Sharp1 overexpression and H3K9me2 on the Myog (myogenin) promoter lead to the proposal that the presence of Ehmt2 and Bhlhe41/Sharp1 together leads to transcriptional repression of the myogenic differentiation machinery. Thus, recent publications using the C2C12 cell line suggests a Bhlhe41/Sharp1-dependent mechanism inhibiting myogenesis that is controlled by Ehmt2 both through direct methylation of MYOD and the repressive H3K9me2 modification of Myod target genes (Ling, Gopinadhan, et al. !502012; Wang et al. 2013). In spite of these initial findings, whether Ehmt2 plays an equivalent role in a more complex biological system such as in vivo skeletal muscle development or regeneration is yet to be validated. Expression of Ehmt2 during embryonic myogenesis Although Ehmt2 is known to be important for embryogenesis, its expression in embryonic muscle was unclear. I performed an analysis on the EMBRYS database, which is a previously published collection of transcriptomic whole-mount in situ hybridisation results of mouse embryos, with emphasis on myogenic development (Yokoyama et al., 2009). After querying Ehmt2, Ehmt1, and various myogenic markers, I found that Ehmt2 was localised, with medium intensity, in the somites and in the limb bud, but not in newly formed embryonic muscle (Table 2.1).   Ehmt2 coincided with the first appearance of Myod expression in somites at E9.5. In contrast to the expression of Myod in the embryonic muscle tissue, Ehmt2 expression was not found. There was no localised signals of Ehmt1/GLP, homologue of Ehmt2, in any of these tissues. Furthermore, Ehmt2 and Ehmt1 were not detected in the heart, liver, pharyngeal arch, eye, somite limb bud muscleE9.5 E10.5 E11.5 E9.5 E10.5 E11.5Myod medium high high none medium high highEhmt2 medium medium none medium medium medium noneEhmt1 none none none none none none noneTable 2.1 Summary of whole-mount in situ hybridisation results showing the expression of Myod, Ehmt2, and Ehmt1 during muscle development. Data extracted from the EMBRYS database (Yokoyama et al. 2009).!51genitalia, olfactory placode, otic pit, brain, or spinal cord. The data suggest that Ehmt2 could be relevant to skeletal muscle development due to its expression in the developing limb bud. In the study described in this chapter, I generated transgenic mouse strains to genetically delete Ehmt2 during muscle development as well as in adult satellite cells. I found that proliferation and differentiation of satellite cells was not influenced by the absence of Ehmt2. Knocking out Ehmt2 also failed to result in significant consequences for skeletal muscle development and in adult muscle regeneration in vivo. Thus, Ehmt2 is completely dispensable for the normal functioning, maintenance and damage response of murine skeletal muscle. RESULTS Developmental biology Generation and efficiency of developmental knockout of Ehmt2/G9a in skeletal muscle To examine the role of Ehmt2 in myogenesis in vivo, I first established a transgenic mouse model in which the Ehmt2 gene was conditionally knocked out in the skeletal muscle lineage. Mice with loxP-flanked Ehmt2 alleles (Ehmt2floxed) were crossed to mice with a Cre recombinase gene knocked-in to the Myod locus (MyodCre) (Fig. 2.1). The Ehmt2floxed allele was previously generated by Lehnertz et al. (2010). The loxP sites flanked exons 4 to 20, the deletion of which causes a frameshift mutation that places the downstream coding sequence out of frame, leaving a truncated peptide that is quickly degraded (Fig. 2.2). As mentioned in chapter 1, all skeletal muscle fibres and adult satellite cells are derived from Myod-expressing myogenic progenitor cells during embryogenesis (Wood et al., 2013). Thus, the F2 generation resulting from this genetic cross contained individuals that were MyodCre/wt Ehmt2floxed/floxed , which expressed Cre !52recombinase in all myogenic progenitor cells during embryogenesis, and resulted in loss of both alleles of Ehmt2 (Ehmt2null/null) in the entire skeletal muscle lineage. These individuals are henceforth referred to as the knockout group (KO), which were compared to the wildtype group (WT) that had no deletion of Ehmt2.         To verify the efficiency of the conditional knockout, I performed an exon-specific allele-counting assay using digital droplet PCR to measure functional allele frequency (see Methods for details). In FACS-purified satellite cell population from control mice (Myodwt/wt Ehmt2floxed/floxed), the Ehmt2 functional allele frequency was 100%; whereas in the knockout mice (Myodwt/Cre Ehmt2floxed/floxedMyodCre/wtFig. 2.1 C57BL/6 mice with Ehmt2floxed allele was bred to a strain expressing Cre recombinase under the control of the Myod promoter, which was generated by Kanisicak et al. (2009).transcription activation ankyrin repeat SET1 249 976 1263targeted deletion frameshiftFig. 2.2 C57BL/6 mice with the Ehmt2floxed allele were previously generated by Lehnertz et al. (2010). In this strain, Cre-mediated targeting of the loxP sites results in a genomic deletion of exons 4-20 and a frameshift within all exons downstream of exon 20.!53Ehmt2floxed/floxed), the functional allele frequency was reduced to 2.9-7.9% (95% CI) (Fig 2.3. This confirms that the conditional genetic deletion was 94.6% efficient in terms of absolute allele count.     These genetic results were consistent with immunostaining quantification of EHMT2 protein in satellite cells. In wildtype individuals, EHMT2 was present in activated satellite cells whereas no detectable staining was present in knockout mice (Fig. 2.4). Furthermore, western blot analysis of whole skeletal muscle lysates from the knockout mice showed reduction of H3K9me2 levels compared to wildtype (Fig. 2.5), congruent with previous reports that H3K9me2 is diminished in Ehmt2null/null models. wt/wt wt/f f/f020406080100Ehmt2 genotypefunctional allelefrequency (%)Fig. 2.3 Ehmt2 deletion efficiency in Myod-Cre Ehmt2-floxed mice, as measured by Ehmt2 functional allele frequency in FACS-purified satellite cells. Analysis by gDNA allele counting using ddPCR, n ≥ 5.!54       Thus, by analysing the gene deletion at the DNA level, the protein level, and in biochemical activity, it was clear that this conditional knockout was highly efficient. WTKODAPI EHMT2myoblast myoblastmyoblast myoblastmyonucleus myonucleusmyoblast myoblastmyonucleus myonucleusFig. 2.4 Immunofluorescence detection of EHMT2 on myofiber, isolated from wildtype and conditional knockout mice.H3H3K9me2normalizedH3K9me2intensityWT KO01234Fig. 2.5 Relative abundance of H3K9me2 in whole skeletal muscle tissue lysate of wildtype and knockout mice, normalized to histone H3.!55Viability of Ehmt2/G9a knockout mice Knockout and control group progenies from the F2 generation cross were born at expected Mendelian frequencies (Fig. 2.6a) and neonatal weights were similar between both groups (Fig. 2.6b).      Neonatal growth analysis of Ehmt2/G9a knockout mice Growth patterns of knockout and control group progenies were charted by body weight during the first 42 days (Fig. 2.7). Regression and statistical analyses revealed that the patterns of growth were not significantly different (see Methods for details). Cre+ f/f Cre+ f/wt Cre- f/f Cre- f/wt01020304050no. of progenya expectedobservedmale female0.00.51.01.52.0b WTneonatalweight (g)KOns nsFig. 2.6 a) Number of live births from n ≥ 3 mating pairs of Myod-Cre Ehmt2-floxed mice. b) Neonatal weight of wildtype and knockout mice at D0-D1. (p > 0.05)!56   Structural integrity of skeletal muscle in Ehmt2/G9a knockout mice Mature skeletal muscle in these mice are of similar size, and showed no difference upon histological examination. Wildtype and knockout tissues were compared by myofiber size, measuring the cross-sectional area of myofibers, as well as the weight of dissected tibialis anterior muscle (Fig. 2.8). 10 20 30 400510152025age (days)weight (g)WTHETKON.S.Fig. 2.7 Growth curve of wildtype and knockout mice. Regression analysis fitted a linear model for each group. A sum-of-squares F test was performed on a shared model to test the null hypothesis that one model fits all groups (p > 0.05).!57       Muscle stem cells Ehmt2/G9a knockout satellite cell activation As previous reports have suggested Ehmt2 is an important regulator of C2C12 myogenesis, I next assessed whether it plays a similar role in satellite cell and primary myoblast cultures ex vivo. To analyse satellite cell activation I quantified immunofluorescent staining of PAX7 and MYOD in myofibre-associated satellite cells from wildtype and the aforementioned conditional knockout mice after 72 hours in culture. Since these cells were analysed on cultured myofibres, I was able to visualise the process of cell activation in a gradual manner; satellite cells cultured without its myofibre niche would spontaneously activate at a much faster rate. I have observed in trial experiments where nearly all satellite cells express MYOD by 72 hours, if purified by FACS and cultured simply on matrigel. Subtle differences in activation might be missed if the cells activate too quickly. In comparison, cell lines such as C2C12 would be an even less reliable candidate for WT KO010203040TA weight (mg)nsa bWTKOWT KO05001000150020002500average myofiber CSA (μm2 )cnsFig. 2.8 a) Weight of whole tibialis anterior muscle (p > 0.05). b) Masson’s trichrome stain of histological sections of the tibialis anterior muscle of adult mice. c) Myofiber size measurement by cross-sectional area (p > 0.05).!58this study, due to their constant myogenic activation. In my quantification of satellite cells cultured in their myofibre niche, the abundance of PAX7 nad MYOD expressing cells between wildtype and knockout are similar (Fig. 2.9), thus indicating that satellite cells lacking Ehmt2 are activated normally.      DAPIMYODWT KOaWT KO020406080100% MYOD+ cellsN.S.DAPIPAX7WT KObWT KO020406080100% Pax7+ cells% PAX7+ cellsN.S.Fig. 2.9 a) Immunofluorescence detection and quantification of MYOD+ cells at 72h after myofibre isolation (p > 0.05). b) Immunofluorescence detection and quantification of PAX7+ cells at 72h after myofibre isolation (p > 0.05).!59Ehmt2/G9a knockout satellite cell proliferation To analyse satellite cell proliferation I performed a 4 hour 5-ethynyl-2'-deoxyuridine (EdU) pulse on myofibre-associated satellite cells from wildtype and conditional knockout mice after 72 hours in culture. No significant differences in EdU incorporation were detected between control and knockout groups (Fig. 2.10). Again the use of satellite cells cultured in their myofibre niche allowed for more sensitive detection of differences in proliferation between cells than FACS-purified satellite cells. C2C12 cells would be even more unsuitable since they were reported to have a much shorter doubling time than realistic myoblasts (Cheng et al., 2014).    Ehmt2/G9a knockout myogenic differentiation Next I assessed the requirement of Ehmt2 for satellite cells to undergo myogenic differentiation. Satellite cell derived myofibers from WT and KO mice were expanded to confluence, induced to differentiate and then analysed by immunostaining of MYOG and myosin heavy chain after 4, 24 aDAPIEdUWT KOWT KO020406080100% EdU+ cells% EdU+ cellsN.S.Fig. 2.10 Immunofluorescence detection and quantification of EdU+ cells at 4h after EdU treatment (p > 0.05).!60and 48 hours. Under differentiating conditions, MYOG-expressing cells increased, but no significant differences in the percentages of cells expressing MYOG were observed between control and knockout myoblasts (Fig. 2.11a). Similarly, I found no significant differences in myogenic fusion index (ratio of fused nuclei found in myosin-expressing cells to total nuclei) following 48 hours of differentiation (Fig. 2.11b), providing further support that deletion of Ehmt2 does not have significant effects on the progress or timing of myogenic differentiation in primary myoblasts.      DAPIMYOGWT KOa4H 24H 48H020406080100time in differentiation medium% MYOG+ cells% MYOG+ cellstime in differentiation mediamyogCre X G9A1020304050EXPECTEDOBSERVEDno. of progenyWTKb WT KODAPIMYH2WT KO0204060myoblast fusion index (%)myoblastfusion index (%)N.S.Fig. 2.11 a) Immunofluorescence detection and quantification of MYOG+ (myogenin) cells at 4h, 24h, and 48h during differentiation. b) Immunofluorescence detection of myosin heavy chain (MYH2) in myotubes at 48h during differentiation, myoblast fusion index calculated as % nuclei inside myosin-expressing myotubes.!61Ehmt2/G9a expression during myogenic differentiation To further detect any differences between WT and KO cells during myogenesis, I monitored the expression level of Ehmt2 at 4, 24, and 48 hours during differentiation. The expression did not show major changes as the cells differentiated from the WT satellite cells. In the KO group, the population remained EHMT2 negative, suggesting that cells in which genetic knockout wasn't efficient did not gain any advantage during differentiation (Fig. 2.12). Lack of dynamic changes in expression of the gene was a sign that it did not likely play a role in the dynamic changes during myogenic differentiation.    Regeneration Ehmt2/G9a expression during skeletal muscle regeneration As part of a larger study on cell population behaviour during muscle regeneration, we performed RNA sequencing of primary myoblasts at different stages of the regenerative process. This was done after purifying myoblasts by FACS from wildtype mice, after acute injury, by an intramuscular injection of notexin (a snake venom toxin) in the tibialis anterior muscle. Regeneration is quickly activated after injury, most new fibres are formed at 7 days, and the DAPIEHMT2WT KO4H 24H 48H020406080100time in differentiation medium% G9A+ cells% EHMT2+ cellstime in differentiation mediamyogCre X G9A1020304050EXPECTEDOBSERVEDno. of progenyWTKFig. 2.12 Immunofluorescence detection and quantification of EHMT2+ cells at 4h, 24h, and 48h during differentiation.!62process is complete at 2-3 weeks. The RNA sequencing results confirmed current understanding of key myogenic regulators. However, Ehmt2 and Ehmt1 did not show dynamic changes during the process. For example, when compared to the drastic up-regulation of Myog during regeneration, Ehmt2 and Ehmt1 remained low and stable (Fig. 2.13). If Ehmt2 blocks myogenic differentiation, as previously reported, we would expect to see down-regulation as an enabling mechanism during regeneration. This piece of surprising evidence further lead me to expect the dispensability of Ehmt2 in a skeletal muscle tissue regeneration model.    Myoblast activation and differentiation during regeneration To confirm the findings from the ex vivo studies, I analysed the myogenic programme in vivo, by measuring gene expression of Pax7, Myod, Myf5, Myog, and Myosin (Myh1) in FACS-purified D1 D5 D70100200300MyogEhmt2Ehmt1FPKMFig. 2.13 Ehmt2 gene expression in satellite cells during adult skeletal muscle regeneration. Satellite cells were purified by FACS at D1, D5, and D7, after notexin-induced TA muscle injury. Gene expression was analyzed by RNA sequencing (Illumina MiSeq).!63myoblasts, at 3 days after intramuscular injection of notexin. No significant differences were oberserved between the WT and KO groups (Fig. 2.14), confirming that the knockout did not alter the myogenic programme following acute injury.    Pax7WT KO050100150200mRNA gene expression (%)n.s.MyodWT KO050100150mRNA gene expression (%)n.s.Myf5WT KO050100150mRNA gene expression (%) n.s.myosinWT KO050100150mRNA gene expression (%) n.s.MyogWT KO050100150 n.s.mRNA gene expression (%)Fig. 2.14 Myogenic gene expression in satellite cells during adult skeletal muscle regeneration (p > 0.05). Satellite cells were purified by FACS at 3 days after notexin-induced TA muscle injury.!64Skeletal muscle regeneration in Ehmt2/G9a developmental knockout mice I next evaluated the requirement of Ehmt2 in the response of skeletal muscle to acute injury in vivo. I began with the aforementioned MyodCre Ehmt2floxed/floxed model, which deletes Ehmt2 in all skeletal muscles and satellite cells during their development. Following the aforementioned acute injury, we quantified the cross-sectional area of centrally nucleated myofibers as a measure of regeneration (Murphy et al., 2011). Despite a very slight trend toward an increased number of the largest fibers in KO samples, no statistically significant difference was found between WT and KO mice in the distribution of myofiber size at 7, 14 or 21 days post injury, indicating comparable regenerative capacities (Fig. 2.15).   D7a MyodCreEhmt2f/fD14 D21D0notexininjury histologyhistologyhistologyD7WTKOD14 D21b!65   Skeletal muscle regeneration in Ehmt2/G9a adult satellite cell knockout mice Myod-driven CRE leads to target deletion early in development, and could therefore trigger compensatory effects that mask the regulatory role of Ehmt2 in adult regenerative myogenesis. One way to mitigate this risk is by restricting the gene deletion within the adult satellite cell population. As discussed in Chapter 1, satellite cells in the adult is specifically marked by the expression of Pax7. Hence, I explored a few different Pax7-driven Cre lines, and measured efficiency using a Rosa-YFP reporter. By flow cytometry, I identified the adult satellite cell population using cell surface markers and found that the Pax7-Cre line produced by Keller et al. (2004) had 100% labelling efficiency (Fig. 2.16). This is not surprising, as this is a constitutively active Cre construct, which would be expressed under the Pax7 promoter in somites during embryonic muscle development. Since I was only interested in the adult satellite cells, this model would not fully address my concern. Inducible Cre systems, on the other hand, would allow temporal control of the activity of CRE. Using the same reporter, I analysed the labelling 400600800100012001400160018002000220024002600280030003200340036003800400042004400460048000246810biological reps grouped binsWTKOrelative frequency (%)fibre size (μm2)myogCre X G9ACre+ F/FCre+ F/WTCre- F/FCre- F/WT01020304050EXPECTEDOBSERVEDgenotypeno. of progenyWTKFig. 2.15 a) Schematic diagram of muscle injury timeline for Myod-Cre Ehmt2-floxed mice. b) Masson’s trichrome stain of histological sections of the tibialis anterior muscle of adult Myod-Cre Ehmt2-floxed mice at 21 days after injury and myofiber size measurement by cross-sectional area (p > 0.05).!66efficiency of two inducible Pax7-Cre lines, generated by Nishijo et al. (2009) and Lepper et al. (2009), which are tamoxifen-inducible CreER and CreERT2 constructs. Adult mice were induced for one week, and I waited one more week before analysis. I found by flow cytometry that the Pax7-CreERT2 generated by Lepper et al. (2009) was more efficient in labelling satellite cells (Fig. 2.16). I also confirmed the specificity of the labelling by flow cytometry; nearly all YFP-expressing cells from the Pax7-CreERT2 mice (Lepper et al., 2009) were identified as satellite cells by surface markers.     To produce an efficient inducible conditional knockout of Ehmt2, I bred the Pax7-CreERT2 line to the Ehmt2floxed/null mice, and induced adult mice with tamoxifen. To confirm the definitive efficiency of gene deletion, I purified satellite cells by FACS and employed the aforementioned exon-specific allele-counting assay (see Methods section), which found the Ehmt2 functional Comparison of Pax7-Cre labelling efficiencyNISHIJO PAX7LEPPER PAX7KELLER PAX7020406080100Strain of Pax7-CreYFP+ satellite cellsa100 101 102 103 104APC-A: A7 APC-A100101102103104PE-Cy7-A: VCAM PE-Cy7-A96.1bFig. 2.16 a) YFP reporter expression as a measure of CreERT2 induction efficiency. A mouse line carrying a tamoxifen-activated CreERT2 recombinase knocked-in to the Pax7 locus (Lepper et al. 2009), was used for generating inducible conditional knockout. YFP expressing cells were quantified within the satellite cell population by flow cytometry, using surface markers, one week after the end of tamoxifen treatment. Wildtype mice (not shown), mice carrying Pax7-CreERT (Nishijo et al. 2009), and mice carrying a constitutively active Pax7-Cre (Keller et al. 2004) were used as controls and for comparison. b) YFP reporter expression as a measure of CreERT2 induction specificity. YFP+ cells were gated in flow cytometry showing markers of satellite cells (Integrin alpha-7 and VCAM). The number refers to % of total YFP+ cells with satellite cell markers.!67allele frequency had been reduced to 27.2±4.2% (SEM) (Fig. 2.17). The results indicated that I had generated a model in which Ehmt2 was deleted only in satellite cells and only in adulthood.    Using this adult satellite cell specific knockout model, I analysed the capacity of muscle regeneration. Acute muscle injury by notexin was performed at 7 days after the final CreERT2 induction on mice harbouring Pax7CreERT2 and Ehmt2floxed/null alleles. At 7, 14, and 21 days after the injury no significant differences in myofiber size distribution were observed between the control and knockout groups (Fig. 2.18, data not shown), suggesting lack of Ehmt2 in adult satellite cells does not significantly affect repair and regeneration in vivo. wt/wt f/null020406080100Ehmt2 genotypefunctional allelefrequency (%)Ehmt2floxed/nullPax7CreERT2/wta bFig. 2.17 a) Schematic diagram of Pax7-CreERT2 Ehmt2-floxed/null mice. b) Ehmt2 deletion efficiency by Pax7-CreERT2 as measured by Ehmt2 functional allele frequency in FACS-purified satellite cells. Analysis by allele quantification using ddPCR, n ≥ 3.!68    DISCUSSION The role of H3K9me2 in skeletal muscle development and regeneration To date, no genome-wide analysis of the EHMT2 mediated H3K9me2 in myogenic cells exists. Dynamic changes in H3K9me2 have been reported at specific gene bodies and regulatory regions (Zylicz et al., 2015; Renneville et al., 2015), suggesting that modulating this epigenetic mark may affect gene transcription. However, in our experiments lack of Ehmt2 led to a dramatic drop in the global levels of detectable H3K9me2, as described in section 2.2.1.2, in the absence of any D12D5D0notexininjurytamoxifena Pax7CreERT2Ehmt2f/nullD19 D26 D33histologyhistologyhistologyD14 D21WTKOWT KO0500100015002000250030003500average myofiber CSA (μm2 )nsFig. 2.18 a) Schematic diagram of leg injury timeline for Pax7-CreERT2 Ehmt2-floxed/null mice. b) Myofiber size measurement by cross-sectional area and Masson’s trichrome stain of histological sections of the tibialis anterior muscle of induced adult Pax7-CreERT2 Ehmt2-floxed/null mice at 21 days after injury. p > 0.05.!69effects on skeletal muscle development. This indicates that EHMT2 activity is mostly non-redundant, and questions the importance of EHMT2-mediated histone modifications in myogenesis in particular, and in the control of differentiation in general. Potential compensatory mechanisms by Ehmt1/GLP Although we have shown here that the loss of EHMT2’s histone methylation function in our model was not compensated, the possibility exists that its potential interaction with myogenic regulators could be compensated by another gene, such as its close homolog Ehmt1 (GLP). These two genes are highly similar in structure, as their protein products contain the same catalytic domain for lysine methylation (SET domain) and a set of ankyrin repeats for protein-protein interaction. These two enzymes are known to form heterodimers (Tachibana et al., 2002), but also play unique roles depending on the cell type and developmental stage (Kramer, 2015). Using domain-specific mutations, the EHMT1 ankyrin repeats, but not the EHMT2 ankyrin repeats, were found to be required for mouse viability (Liu et al., 2015), suggesting this domain function in EHMT2 could be compensated by EHMT1. On the other hand, the SET domain in EHMT1 is dispensable for mouse viability (Inagawa et al., 2013), suggesting EHMT2 could compensate for EHMT1’s methyltransferase activity (Fig. 1.3). In our Ehmt2 knockout cells, even though the histone methyltransferase activity, a function shared by both proteins, is not compensated by EHMT1, its ankyrin repeat dependent protein-protein interactions may be compensated. Both proteins have been reported to potentially methylate a number of non-histone targets beyond MYOD (Herz, Garruss, and Shilatifard, 2013; Lanouette et al., 2014), and it is unknown if EHMT1 could replace EHMT2 in binding and methylating these targets. To address this concern in the future, a conditional double knockout of Ehmt2 and Ehmt1 in myogenesis would be required.  !70Contradictions between models These results are surprising given the previous reports suggesting an important role for Ehmt2 in negatively regulating myogenic differentiation of C2C12 cells, an immortalized myogenic line. In particular, as detailed in section 1.x, it was demonstrated that siRNA knockdown of Ehmt2 led to enhanced and/or premature differentiation of C2C12 myoblasts (Ling, Bharathy, et al., 2012). These results were not consistent with observations in the current study when examining the differentiation capacity of satellite cells lacking Ehmt2, which showed no premature differentiation, no alterations in myogenic fusion, and normal proliferation. This discrepancy in findings could stem from the fundamental differences between the biological models being analysed; unlike C2C12 cells, which were derived from a different strain of mouse (C3H) (Yaffe and Saxel, 1977), immortalised (Blau et al., 1985), and have a much shorter doubling time (Cheng et al., 2014); the primary cells in our study were from a C57BL/6 strain background, not serially passaged, and were analysed in their myofibre niche during proliferation and on matrigel during differentiation, in addition to the use of an in vivo model. These differences may be particularly relevant in the case of Ehmt2, as Lehnertz et al. (2014) have recently reported, in another tissue system, that its absence leads to drastically different effects in transformed compared to natural hematopoietic cells. EHMT2-mediated methylation of MYOD is uncertain What remains unclear is the status of EHMT2-mediated methylation of MYOD at lysine 104 (Ling, Bharathy, et al., 2012) in the in vivo model. Ling, Bharathy, et al. reported the identification of this residue by mass spectrometry of peptides resulting from the digestion of MYOD with trypsin (2012) (Fig. 2.20). However, the reported MYOD peptides, ACKACKRKTT and its methylated forms are preceded by a tryptophan residue, and they do not appear to be obtainable !71by trypsin digestion alone, or by any commonly used digestion method. The reported MYOD peptide is also not found in the tandem mass spectrometry proteomics repository Peptide Atlas (https://db.systemsbiology.net/sbeams/cgi/shortURL?key=1mk36ybs).  More intriguingly, the proposed mechanism was based on LC-MS results, which show the different methylation states of the MYOD peptide were separated by only 1 m/z unit each. Methylation adds 14 Da to the peptide mass; thus, each peptide would have to carry 14 charges on 10 residues. The LC-MS results could not possibly correspond to methylation of the reported MYOD peptide. The uncertainty of MYOD methylation, together with the dispensability of EHMT2-mediated H3K9me2 in vivo, cast doubts on the proposed role, by Ling, Gopinadhan, et al. (2012), of Ehmt2 in Bhlhe41/Sharp1-mediated regulation of myogenesis. Nevertheless, SUMOylated Bhlhe41/Sharp1, proposed by Wang et al. (2013), may still regulate Myod and downstream targets through an alternative, non-EHMT2, non-H3K9me2 mechanism. !72      …CLLWACKACKRKTTNADR…Fig. 2.19 LC-MS results from Ling, Bharathy, et al. (2012), showing the methylated MYOD peptide in cells transfected with wildtype Ehmt2/G9a. Protein sequence of MYOD is shown here with the peptide in question in red and the methylated lysine residue in purple. The peptide is preceded by a tryptophan residue. !73METHODS AND MATERIALS Standard in vivo methods Mice and animal care C57BL/6 mouse harbouring the Ehmt2floxed allele were previously generated by Lehnertz et al. (2010). In this strain Cre mediated targeting of the Ehmt2floxed allele results in a genomic deletion from exon 4 to exon 20 and a frameshift mutation that places the downstream coding sequence out of frame. All other transgenic strains used herein were generated by other groups (Kanisicak et al., 2009; Lepper, Conway, and Fan, 2009; Srinivas et al., 2001), and obtained from The Jackson Laboratory. Inducible Cre recombinase was activated by intraperitoneal injection of tamoxifen dissolved in corn oil (250mg per kg of body weight per day) for 5 consecutive days, followed by 7 days without any treatment to allow sufficient activation. Mice were housed in a pathogen-free facility, and all experiments were performed according to the Canadian Council on Animal Care (CCAC) regulations. Acute muscle injury The tibialis anterior (TA) muscle of 8-12 week-old control or experimental mice was injected with the myotoxin notexin (7ul) as previously described (Joe et al., 2010). Cell culture and immunocytochemistry Viable single myofibers were isolated from the extensor digitorum longus (EDL) muscle of 6-8 week old mice following dissociation with collagenase I as previously described (Collins and Zammit, 2009) Myofibers and their associated satellite cells were maintained ex vivo for up to 72 hours in high glucose Dulbecco’s modified eagle medium (DMEM) supplemented with 20% v/v !74FBS, 0.5% v/v chick embryo extract and pen-strep. Following culture, single myofibers were fixed in 4% PFA and then stained overnight with the following primary antibodies: mouse anti-Pax7 (DSHB), mouse anti-MyoD (Dako, clone 5.8A), mouse anti-Myogenin (DSHB, clone FD5). FACS-sorted cells were grown in high-glucose DMEM, supplemented with 2.5 ng/ml bFGF (Invitrogen) 20% v/v FBS and 10% v/v horse serum. This medium is hereafter referred to as ‘growth medium’. Cells were seeded in tissue-culture-treated plastics coated with Matrigel (BD Biosciences). Media was changed every 24-48 hours. To induce myogenic differentiation confluent myoblasts were cultured in DMEM supplemented with 5% horse serum for up to 96 hours, before being fixed in 4% PFA and stained overnight with mouse anti-Myosin (DSHB, clone MF20). Histology Tibialis anterior muscles were dissected from mice, fixed in 4% paraformaldehyde overnight followed by 70% ethanol overnight, and then embedded in paraffin following standard protocols. Tissues were cut with a microtome in a cross sectional orientation through the entire length of the muscle. Cross sections of 5 mm thickness were then mounted onto glass slides (Thermo Fisher Scientific, USA) and stained with Masson's trichrome or Picrosirius red following standard protocols. Optimization of purification of satellite cells Skeletal muscle tissue was prepared as described previously in Joe et al. (2010), which involved two rounds of enzymatic digestions and washes. Cell preparations were then incubated with primary antibodies for 30 min at 4°C in supplemented PBS containing 2mM EDTA and 2% fetal !75bovine serum at ~3×107 cells per ml. I used the following monoclonal conjugated primary antibodies: anti-PECAM-1 (CD31) (clones MEC13.3, Becton Dickenson, and 390, Cedarlane Laboratories), anti-PTPRC (CD45) (clone 30-F11, Becton Dickenson), anti-LY6A/E (Sca-1) (clone D7, eBiosciences), anti-VCAM (produced in-house), and anti-Integrin alpha-7 (produced in-house). Antibody dilution and staining volume were determined experimentally. Where necessary, biotinylated primary antibodies were detected using streptavidin coupled to phycoerythrin (PE), allophycocyanin (APC), phycoerythrin-cyanine 7 tandem complex (PE-Cy7) or fluorescein isothiocyanate (FITC) (Caltag). To assess viability, cells were stained with propidium iodide (1 μg/ml) and Hoechst 33342 (2.5 μg/ml) and resuspended at ~1×107 cells/ml immediately before flow cytometry analysis or sorting. Myself and others in our team have experimentally determined that the addition of VCAM as a marker greatly enhanced the reliability and quality of satellite cell purification from mouse skeletal muscle. In all experiments where satellite cells are analysed by flow cytometry or purified by FACS, the population is defined as PECAM negative, PTPRC negative, LY6A/E negative, Integrin alpha-7 positive, and VCAM positive. Since this is a very small population compared to the large amount of debris resulting from the tissue digestion, the time required to purify by FACS can be very long, usually 45 minutes for only 100,000 cells. The time consumption not only increased the cost, but also decreased the viability of the cells for culturing, since the purification took place in a serum-free and growth factor-free condition. To increase the efficiency of this process, I developed an additional step to remove unwanted cells and some of the debris that arose from the tissue digestion, before performing FACS. After cell preparations were stained with antibodies and washed, they were enriched using the EasySep selection kits (Stem Cell Technologies). The PE-selection kit, for example, was used !76to label the PECAM, PTPRC, and LY6A/E positive cells through the PE fluorophore with magnetic beads. Strong magnets were then used to separate these cells from the sample. The kit was designed for positive selection, but by using it in a negative selection manner, it becomes more suitable for my purpose; it did not completely remove unwanted cells, but it had the side-effect of also removing some debris. Thus, combining magnetic cell separation and FACS, I was able to achieve purification with higher efficiency. Optimizing efficiency in knockout models A novel, fast and accurate method of measuring gene knockout efficiency Due to the need for different conditional knockout models in this project, I had a major concern about the efficiency of these systems. Many researchers agree that the usefulness of each new conditional knockout model relies on its efficiency. However, there is no definitive measure of genetic knockout efficiency in our discipline; as current research often use an incomplete combination of gene expression assays, immunoblots, immunocytochemistry, and immunohistochemistry to demonstrate the efficiency of a model. I find that the caveats associated with these methods do not fully reflect the true efficiency, especially in inducible conditional knockouts, where 100% gene deletion is rarely achieved. Measuring the mRNA transcripts of the gene, for example, would not be accurate due to the fact that many cells retain copies of the transcripts regardless of the presence of the gene at the DNA level. This is especially true for vital gene products that need to be produced very quickly, and rely on translational control of existing mRNA transcripts. Measuring the protein abundance of the gene could also be inadequate if the expression level was low, as translational and post-translational regulations are unaffected by the presence or absence of the gene. I reasoned that the most definitive way to demonstrate the efficiency of gene deletion was a direct measure of the gene at the DNA level. For this project, I !77developed a method that uses digital PCR to quantify the allele copies in a population of purified cells. At least 10,000 FACS-purified satellite cells per sample were used for measuring the efficiency of Ehmt2 conditional knockouts. Cells were lysed and purified for genomic DNA, of which 50ng per sample was mixed with digital droplet PCR supermix (Bio-rad), and two Taqman Copy Number Assays for a duplex (FAM and VIC) digital droplet PCR assay (Fig. 2.21). The ‘functional assay’ is a Taqman Copy Number Assay with FAM dye (Thermo Fisher Scientific #4400291 Mm00466045_cn) that detects Ehmt2 in a region from intron 14 to exon 15, which is found only in the wildtype or floxed alleles (functional alleles) of the gene. The ‘reference assay’ is a Taqman Copy Number Assay with VIC dye (Thermo Fisher Scientific #4400291 Mm00466690_cn) that detects Ehmt2 in a region from intron 25 to intron 26, which is found in any null, wildtype, or floxed alleles of the gene. Droplets of the mixture were generated according to standard digital droplet PCR protocol (Bio-rad), and ran in a thermocycler for 40 PCR cycles. PCR products were read droplet-wise in duplex (FAM and VIC) following standard protocol (Bio-rad), and a signal ratio was calculated by dividing the absolute copy number of the functional assay to the copy number of the reference assay. The signal ratio was then used to interpolate the functional allele frequency (%) from a known standard curve. The standard curve of signal ratios was produced by performing ddPCR using genomic DNA that were mixed at known proportions from Ehmt2wt/wt and Ehmt2null/null mouse embryonic fibroblasts. The results of this measurement is either presented as the frequency of functional alleles in each experimental sample, or as the efficiency of gene deletion, which is calculated as the reciprocal of the functional allele frequency. This method takes advantage of the digital PCR's capability at measuring absolute copy numbers, as opposed to the relative quantification of real-time PCR. This method demonstrates, as reported !78in the results subsection, that protein level quantifications overestimated the true efficiency of the gene deletion event. I am currently in collaboration with others to compare the genetic efficiency of inducible conditional knockout models across different lines and cell types, and my method has revealed that many models have lower efficiencies than researchers initially believed. The results will be available in a future publication.    One concern with this method, however, is that it measures the allele frequency in a population, not genotype frequency. For exapmle, a cell population from a floxed/floxed mouse may have a functional allele frequency of 20% after induction, but it does not mean 80% of cells are null/null and 20% of the cells remain unaffected; the actual genotype frequencies could be estimated by the corresponding Hardy–Weinberg proportions: 64% null/null, 32% floxed/null, and 4% floxed/floxed. Nevertheless, the allele frequency is still a good standardised measure that can be applied and compared across different tissues, conditional constructs, and genetic targets. On the other hand, if the model is designed to require the Cre recombinase to act on only one allele, this Fig. 2.20 Workflow of allele quantification for measuring knockout efficiency.!79uncertainty in data interpretation can be avoided; this is the main reason I used floxed/null mice in the experimental group for inducible conditional knockouts. Tamoxifen induction Since I observed a great degree of variation between different lines of similar knockout models. I attempted to assess if the dose and delivery method of the induction agent, tamoxifen, was affecting knockout efficiency. Widely used method of CreER induction by tamoxifen involves dissolving the drug in corn or sunflower oil and delivered by daily intraperitoneal injection. Absorption in the peritoneum is slow, as I have found significant remaining ascites upon dissection one week after the end of induction. My colleagues and I have also encountered a number of cases of scrotal hernia resulting from this method of delivery. Furthermore, tamoxifen is known to be toxic (GHS08 hazard), and in mice, I have found doses to be fatal beyond 250mg/kg per day. However, it is uncertain if fatalities were due to drug toxicity, oil ascites, or a combination of both factors. I hypothesized that the efficiency of CreER-mediated gene deletion was dose-dependent, which is one reason some researchers in the field of developmental biology favoured oral gavage as a delivery method for tamoxifen, as it allowed for higher doses without the risks of ascites or hernia. I have tested this method in different lines of CreER mice and compared it with intraperitoneal injection, and found that in some cases oral gavage with a higher dose yields a slightly higher and more reliable gene deletion efficiency, as measured by functional allele counting (Fig. 2.22). However, none of the differences were statistically significant. Nevertheless, the oral gavage delivery route still has advantages, as no ascites or scrotal hernia were observed in mice induced with this method. !80   Myofibre size measurement In the myogenesis field, the use of cross-sectional area of myofibres in histology has long been established as a measure of skeletal myofibre size. The challenges to automated quantification are the identification of individual myofibres and the processing of large numbers of images to cover a wide area of tissue injury. I have developed a protocol for semi-automated cross-sectional area quantificaiton using the Nikon NIS-Elements microscope software. Automated large image stitching can capture the entire section, which eliminates the possibility of human bias in selecting images. For automated identification of myofibres, it was important to have high contrast between the fibres and interstitial space in an image. Thus, I employed Picrosirius red staining instead of the traditional H/E or Masson's trichrome for skeletal muscle, producing images of faintly yellow fibres and crimson red basal lamina. IP OG IP OG020406080100efficiency (%)Ub-Cre Pax7-CreERIP OG020406080100efficiency (%)n.s.n.s.n.s.180mg/kg250mg/kg180mg/kg250mg/kgKOHET250mg/kgFig. 2.21 a, b) Comparison of knockout efficiency between different tamoxifen delivery methods in Ub-CreERT2 and Pax7-CreERT2. Efficiency (%) was measured by Ehmt2 functional allele frequency. p > 0.05.!81Statistics Datasets of histological measurements The aforementioned cross-sectional measurements generated large amounts of data that is difficult to interpret without advanced methods of statistical analysis. Six TA samples were collected from each experimental group. Thus, the biological sample size was defined as 6 per group. Five sections were produced from each sample, in order to analyse all parts of the length of the sample. The number of myofibres that could be measured in each section was variable, ranging from 2000 to 3000 myofibres, depending on the size of the section. Thus, each biological sample had up to 15,000 measurements, and the measurements varied from 300 μm2 to 5000 μm2. Each experimental group, then, had up to 90,000 measurements. A common practice in the field is to calculate the mean value of each sample from 15,000 measurements, and then presenting the mean value of the group. It is a logical way to present the data, but the propagation of errors is often overlooked as one calculates the mean of means, by simply using a group SEM based on the variation between the 6 sample means of the group. This resulted in small error bars for each group as it discarded the large variation within each sample, and lead to potentially false conclusions about statistical significance. I wish to correct this practice in our field by describing the complete propagation of errors that should be required for this kind of data presentation, which I did in this project by calculating the pooled variance for each experimental group. Sample variance (s2) was first calculated from the standard deviation (s) from the thousands of measurements. Then, since each sample had different numbers of measurements, their contribution to the group's variance was weighted. Thus, the pooled variance (sp2) was the weighted average of the sample variances, defined as: !82  where the samples were indexed i=1, ..., k; ni is the number of measurements done on sample i; si is the standard deviation of measurements within sample i; and sp is the standard deviation of the group. The use of (ni-1) weighting factors instead of ni comes from Bessel's correction. The final standard error of the mean, for each experimental group, is   where N is the number of biological samples in the group. Proper propagation of errors in this type of presentation of the data results in very large error bars. In most cases, the differences between groups would not be statistically significant. In the aforementioned data I collected, for example, there would never be statistical significance even if the group means were extremely different. This type of presentation is simply ineffective as the natural variation of measurements within each sample was too large that it often masked group differences. The only reason that it lead to stastical significance in published studies in the past was the improper propagation of errors, which was evident in studies as recent as the high-impact publication by Tierney et al. (2014). However, this is not to suggest that all past findings in the field were wrong, as the same conclusions might still be made with more recent and robust methods of analysis. In my project, it was especially important to uncover any masked differences, since the WT and KO groups appeared without differences upon first glance. sp2 =∑ki=1(ni − 1)si2∑ki=1(ni − 1)SEM =√∑ki=1(ni − 1)si2∑ki=1(ni − 1)N!83To more effectively interpret this kind of data, a newer method has emerged in the field, which structured the dataset in a frequency distribution. Measurements were assigned to bins based on the value, and in one of my datasets, each bin spanned 200 μm2. This allowed us to visualise and compare the relative abundance of small myofibres and large myofibres across samples and groups. It also yielded meaningful standard errors within each bin for each experimental group. However, this method posed new challenges in the determination of statistical significance; the change resulting from an experimental treatment or genetic deletion could shift the frequency distribution curve, such that the bins at the low and high ends might yield significance, but the bins in the middle might not. The simple statistical tests for each bin cannot conclude the significance of the overall difference between groups. The appropriate approach should be the permutation ANOVA, which would make a direct comparison between the profiles of frequency distribution from each group. It needs a measure of dissimilarity between two profiles, which could be the sum of absolute or squared differences between frequencies of all bins. The permutation test would compute the p-value that represents how extreme the observed dissimilarity is among all its possible values in the given dataset. One method to implement this permutation approach in my datasets employed the Community Ecology Package ('vegan'), developed by Oksanen (2015), in the R statistical environment, although various other options might exist. The reason for choosing this tool is that it was designed to solve a very similar problem in ecological studies, where researchers needed to compare the frequency distribution of a set of species of organisms between different communities or environments, just as how I needed to compare the frequency distribution of classes/bins of myofibres between different mice. In package 'vegan', the function 'adonis' is a Permutational Multivariate Analysis of Variance Using Distance Matrices, also referred to as PerMANOVA, by Anderson (2001; McArdle and Anderson, 2001). It randomised all !84measurements, ignoring their respective original WT or KO group labels, assigned each measurement into one of two arbitrary groups, and recorded a measure of dissimilarity between the resultant two profiles. By repeating this process 10,000 times, the function tested the likelihood of randomly achieving the dissimilarity I observed between the WT and KO profiles, using the data I collected. There are alternative options for the aforementioned statistical tests, such as fitting the data on a linear mixed-effects (LME) model and using ANOVA, for instance. Nevertheless, all methods suffer from a small sample size. For a time-consuming and costly in vivo experiment, it is difficult to increase the number of biological samples. The method I chose would in theory have much more impressive statistical power if the sample size was large. Furthermore, the statistical considerations discussed here and the methods of presenting the dataset are not only useful in myofibre size measurements, but in all types of object-based measurements using microscopy. For example, the same methods could be employed for measuring the size of cells, the diameter of blood vessels, the expression level/intensity of a marker in cells, etc., all of which could have immense variation within each sample. Genetic knockout efficiency measurement The standard curve was produced by performing ddPCR using genomic DNA that were mixed at known proportions from Ehmt2wt/wt and Ehmt2null/null mouse embryonic fibroblasts (Fig. 2.23). The standard curve had a pearson coefficient of 0.993, and a statistical test of linearity yielded a p-value of 0.0001. The curve was plotted with 95% confidence intervals. The allele frequencies of experimental samples were determined by interpolation from the standard curve, along with the confidence intervals. !85   Growth pattern analysis To correctly evaluate neonatal growth, it was important to compare the growth pattern between mice, instead of simple comparisons of absolute weight. This provided a holistic view of weight gain over a few weeks, and eliminated the possibility of drawing conclusions from day-to-day weight fluctuations. Mouse weight measurements plotted against age were subjected to a few regression analyses, which fitted linear, quadratic, cubic, quartic, and quintic function models. A sum-of-squares F test was performed on a shared model to test the null hypothesis that one model fits all groups. How well the shared model fitted the data points of all groups determined a p-value, which was used to conclude the differences in growth pattern between groups. None of the analyses revealed significant differences between groups. Thus, the linear model, which was the simplest, was chosen and presented.0 20 40 60 80 1000.00.51.0% known Ehmt2wt gDNAddPCR signal ratioFig. 2.22 Standard curve of ddPCR signal ratio to Ehmt2 functional allele frequency correlation; n ≥ 3, p = 0.0001.!863. The role of Ehmt2/G9a in congenital anomalies INTRODUCTION: CURRENT UNDERSTANDING OF EHMT1/2 IN KLEEFSTRA SYNDROME AND CRANIOFACIAL DEVELOPMENT Although I have clarified in detail the dispensability of Ehmt2 in the development of skeletal muscle, its role in other tissues remains unclear. Loss of function mutations in Ehmt1, homologue of Ehmt2, form the genetic basis of syndromic congenital defects (see Chapter 1). Combined with the fact that Ehtm2null/null genotype is not viable, there is suspicion that Ehmt2 should still play a critical role in development, likely in multiple tissues. In order to uncover the role of Ehmt2 in other mesenchymal tissues and its catalytic compensation of Ehmt1, I generated a transgenic mouse line to conditionally delete Ehmt2 during mesoderm and neural crest development. I found that the loss of Ehmt2 in the Pdgfra lineage resulted in striking yet highly reproducible craniofacial malformations. RESULTS General characterisation of the Pdgfra developmental lineage To study the role of Ehmt2 in multiple mesenchymal tissues, I first characterised the Pdgfra developmental lineage using a line of PdgfraCre RosatdTomato mice, which I developed by crossing two transgenic lines. As mentioned in Chapter 1, Pdgfra is expressed early during the development of the mesoderm and neural crest. Thus, it isn't surprising to find widespread tdTomato reporter expression in most tissues of the embryo. At Theiler stage 23 (equivalent to E14.5), the entire embryo is lit by the strong illuminating tdTomato, in contrast to Pdgfrawt RosatdTomato embryos, which demonstrates the effectiveness of the reporter (Fig. 3.1). Transverse !87sections showed strong tdTomato labelling in calvarial bones, sinus, retina, lungs, connective tissues of the limbs, dorsal root ganglion, penis, connective tissues surrounding cartilagenous bones. As expected, tdTomato labelling was not found in any epithelial cells, which formed from the ectoderm (Fig. 3.2-3.9).    Fig. 3.1 TdTomato reporter expression of mouse embryos at Theiler stage 23 (E14.5). Left: Pdgfra-Cre Rosa-tdTomato, right: Pdgfra-wt Rosa-tdTomato. Image taken from dissection microscope, lateral view.!88   Fig. 3.2 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of head.!89   Fig. 3.3 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of brain.!90   Fig. 3.4 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of liver and lungs.!91   Fig. 3.5 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of lower abdomen, penis, and lower limbs.!92   
vertebraenucleus pulposusFig. 3.6 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Top: coronal view of lumbar vertebrae, left: lumbar vertebrae details and intervertebral discs.!93   Fig. 3.7 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of neck.!94  
Fig. 3.8 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of scapula and head of humerus.!95   Of interesting note, one area that showed strong labelling was inside the developing nucleus pulposus (Fig. 3.6), which form the centres of intervertebral discs. The nucleus pulposus is derived from cells of the notochord. The notochord, which arises from the mesodermal layer, is a key player in signalling and coordinating embryogenesis. However, vertebrate evolution had Fig. 3.9 TdTomato reporter expression of Pdgfra-Cre Rosa-tdTomato mouse embryo at Theiler stage 23 (E14.5). Transverse view of forelimb.!96resulted in the replacement of the notochord by the vertebral column (Stemple, 2005), and in mammals, the nucleus pulposus is a unique tissue, being the only remnant of the notochord that is retained into adulthood (McCann et al., 2012). The solid rod-like notochord was formed by E10.5 (McCann et al., 2012), and this further suggest that Pdgfra is expressed by early embryonic cells. Tracing the Pdgfra lineage in skeletal muscle In adult mice, lineage tracing also revealed Pdgfra-expressing cells contributed to cells of skeletal muscle. In digested leg muscles from adult PdgfraCre RosatdTomato mice, flow cytometry showed that most satellite cells, defined as PECAM- PTPRC- LY6A/E- ITGA7+ VCAM+, expressed tdTomato; and most fibro-adipogenic progenitors, defined as PECAM- PTPRC- ITGA7- LY6A/E+, also expressed tdTomato (Fig. 3.10). This confirmed that different stem cells populations in adult skeletal muscle were developed from a common lineage that expressed Pdgfra during the development of the mesoderm. !97   100 101 102 103 104100101102103104FITC-A100 101 102 103 104100101102103104PE-Cy7-A100 101 102 103 104100101102103104PE-Cy7-A81100 101 102 103 104100101102103104FITC-A100 101 102 103 104100101102103104PE-Cy7-A100 101 102 103 104100101102103104PE-Cy7-A87.9Cell populations in Pdgfra-Cre tdTomato skeletal muscleIntegrin alpha-7 Integrin alpha-7Integrin alpha-7Tomato TomatoIntegrin alpha-7FAP satellite cellFig. 3.10 Flow cytometry showing percentage of fibro-adipogenic progenitors expressing tdTomato (left) and percentage of satellite cells expressing tdTomato (right) in digested skeletal muscle tissue.!98Tracing the Pdgfra lineage in hematopoiesis and the endothelium In adult tissues, lineage tracing also revealed that the entire hematopoietic compartment and the endothelium are derived from Pdgfra-expressing cells. I analysed adult PdgfraCre RosaYFP mice by flow cytometry (Fig. 3.11). In digested skeletal muscle samples, all hematopoietic cells were identified by the surface marker PTPRC (CD45), were predominantly YFP-positive. Unsurprisingly, the same phenomenon was observed in digested subcutaneous adipose tissue. This finding was suspected, since it was known that the aorta-gonad-mesonephros (AGM) region of the mesoderm, beginning at E9.5, contributes significantly to the stem cells of adult hematopoiesis (Medvinsky and Dzierzak, 1996). Furthermore, endothelial cells, identified by PECAM (CD31), were also predominantly YFP-positive. This is congruent with the understanding that the lateral plate mesoderm contributes significantly to angioblasts and the development of the endothelium (Gilbert, 2013). Thus, both the muscle and adipose tissues showed that circulating hematopoietic cells and endothelial cells of these tissues were developed from the Pdgfra lineage. !99    Viability of PdgfraCre Ehmt2 knockout mice Since the Pdgfra developmental lineage is a good candidate to study the role of Ehmt2 in multiple mesenchymal tissues, I crossed the Ehmt2floxed mice with the PdgfraCre line. The number of live births of Ehmt2 KO and heterozygous mice were much lower than expected, suggesting that many died prenatally (Fig. 3.12a). Most KO mice that are born were runted and do not survive until weaning. This is in sharp contrast to the birth rates observed in the Ehmt2 knockout in skeletal muscle. Since Ehmt2null/null resulted in early embryonic death, it is possible that the PdgfraCre Ehmt2floxed/floxed conditional knockout revealed an intermediate severity of the outcome, as it did not result in complete inviability, but the Pdgfra lineage was involved in enough tissue systems to affect birth rates and neonatal survival. At Theiler stage 20 (E12.5), I observed semi-reabsorbed KO fetal tissue (Fig. 3.12b). However, the earliest embryonic death resulting from this conditional knockout is unknown. With special care, very few mice were able to survive after weaning. The range of outcomes of the knockout mice suggest that the efficiency of the conditional knockout FAT MUSCLEPTPRC (CD45)PDGFRA-CRE-YFPWTTransgenicFAT MUSCLEPECAM (CD31)PDGFRA-CRE-YFPLIVESpecimen_001_1B5F 4.fcsEvent Count: 7000100 101 102 103 104FITC-A100101102103104APC-A70.73.611.6324.1LIVESpecimen_001_1B6F 4.fcsEvent Count: 8414100 101 102 103 104FITC-A100101102103104APC-A0.02400.5199.5ENDOTHELIUMHEMATOPOIETICLIVESpecimen_001_1B5F 5.fcsEvent Count: 5451100 101 102 103 104FITC-A100101102103104APC-A5.89 14.952.426.8LIVESpecimen_001_1B6F 5.fcsEvent Count: 5595100 101 102 103 104FITC-A100101102103104APC-A24.9 0075.1LIVESpecimen_001_1B5F 6.fcsEvent Count: 6793100 101 102 103 104FITC-A100101102103104PE-Cy7-A5.89 51.233.98.95LIVESpecimen_001_1B6F 6.fcsEvent Count: 8228100 101 102 103 104FITC-A100101102103104PE-Cy7-A56.9 0.0610.02443SCA-1FAT MUSCLE MUSCLEFAT FATLIVESpecimen_001_1B5M 2.fcsEvent Count: 4160100 101 102 103 104FITC-A100101102103104APC-A3.92 34.820.5 33.8LIVESpecimen_001_1B6M B_002.fcsEvent Count: 4288100 101 102 103 104FITC-A100101102103104APC-A35.4 1.459.6 0.07LIVESpecimen_001_1B6M B_003.fcsEvent Count: 3585100 101 102 103 104FITC-A100101102103104APC-A8.01 0.1190.7 0.28LIVESpecimen_001_1B5M 3.fcsEvent Count: 4708100 101 102 103 104FITC-A100101102103104APC-A0.51 6.1429.1 59.2SATELLITE CELLSWTTransgenicFig. 3.11 Flow cytometry showing percentage of the hematopoietic compartment (PTPRC+) expressing the YFP reporter (left) and percentage of endothelial cells (PECAM+) expressing the YFP reporter (right), in digested subcutaneous fat and skeletal muscle tissues.!100might also be a confounding factor. I have subsequently bred PdgfraCre Ehmt2floxed/null mice, and allele counting with genomic DNA samples would address the efficiency issue in the future.       BP2-4, 6-8 (n=69)Cre+ G9a f/fCre+ G9a f/wt Cre-020406080100EXPECTEDOBSERVEDgenotype% of progenyFig. 3.12 a) Genotype distribution of live births from Pgfra-Cre Ehmt2-floxed mating pairs. Expected frequencies are based on Mendelian ratios. b) Whole-mount light microscopy of embryo littermates: Pdgfra-Cre Ehmt2-f/wt and Pdgfra-Cre Ehmt2-f/f, at Theiler stage 20 (E12.5).!101Cartilagenous bone development in Pdgfra-Cre Ehmt2 knockout mice In PdgfraCre RosatdTomato mice, lineage tracing showed that the developing bones of limbs expressed tdTomato at Theiler stage 23 (E14.5) (Fig. 3.13). Thus, it is relevant to study this lineage to delineate any role Ehmt2 might have in bone development. I conducted micro-CT on select PdgfraCre Ehmt2floxed/null specimens that lived beyond the first two weeks, and found no obvious sign of defect in the body wall and limb bones (Fig. 3.14). !102  
Fig. 3.13 Confocal microscopy of tdTomato expression in the radius bone of Pdgfra-Cre Rosa-tdTomato embryo at Theiler stage 23 (E14.5).!103     
Fig. 3.14 a) Micro-CT of Pdgfra-Cre Ehmt2-f/f neonate, dorsal and lateral view of vertebrae, pelvis, and femur. b) Skeletal micro-CT of Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, lateral view.!104Ehmt2 is required during the closure of the fontanelles In normal mice, the anterior fontanelle is expected to close at birth due to the growth of the frontal and parietal bones, which have a different developmental origin than long bones, as calvarial bones are derived from neural crest cells. In my analysis of lineage tracing, calvarial bones were strongly labelled (Fig. 3.15a), confirming that they developed from the Pdgfra lineage neural crest cells. In live-born KO mice, I used specimen micro-CT to observe a consistent defect where the anterior fontanelle remains open (Fig. 3.15b, 3.16). Even in mice cared beyond weaning, the anterior fontanelle showed no sign of closing. No KO mice were born with a closed anterior fontanelle, in contrast to every WT littermate having closed fontanelles (Fig. 3.17). There was a degree of variation in the extent of the opening, as the posterior fontanelle also remained open in some severe cases.   !105       Fig. 3.15 a) Confocal microscopy showing tdTomato expression in the calvaria, in Pdgfra-Cre Rosa-tdTomato embryo at Theiler stage 23 (E14.5). Transverse section of superior part of head. b) Micro-CT of the frontal, parietal, occipital bones, and the upper palate in Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, inferior view. Right image is enlarged due to microcephaly.Fig. 3.16 Skeletal micro-CT showing the fontanelles in Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, superior view.!106f/wt f/ff/wt f/f   25B125B2CONTROL5mmFONTANELLE CLOSUREKO25B325C128A225B2Fig. 3.17 Skull micro-CT showing the fontanelles in Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, superior view.!107Ehmt2 likely dictates ossification and cell composition at cranial sutures As seen in the CT images, most of the cases of open fontanelles extended to the frontal and saggital sutures. Only a small portion of the frontal sutures were in contact, with the rest far from fusion. The saggital sutures also remained open, in some cases completely affecting the posterior fontanelle. However, the lambdoid suture was not affected, and the coronal suture was mostly in contact. One explanation for the open fontanelles was that the frontal and saggital sutures failed to meet, possibly as a result of a defect in the growth of the frontal and parietal bones. Since it is well known in humans and in mice that the anterior fontanelle and posterior frontal suture fuses at the end of calvarial bone development (Bradley et al., 1995), this observation could be explained by a severe delay in cranial bone growth. It is important to note here that unlike cartilagenous bones, the bones of the skull develop by intramembranous ossification. The difference between these mechanisms might underlie the lack of similar defects in cartilagenous bones. Interestingly, since the frontal bones are neural crest-derived, and the parietal bones are mesoderm-derived, the defect observed in the anterior fontanelle lies at the boundary between the neural crest and mesoderm development (see Chapter 1). The closure of the fontanelle is the final step in cranial bone development, and this location reveals that the impact of Ehmt2 may be greatest in late stage intramembranous bone development. The sutures in this case were relevant, since recent publication by Zhao et al. (2015) revealed that a population of mesenchymal stem cells (MSC), which were responsible for the growth and repair of calvarial bones, reside within the adult cranial sutures. The report showed a clear relationship between the lack of MSCs with craniosynostosis. Thus, cranial suture defect might be a sign of MSC abnormality. Interestingly, Pdgfra-expressing cells in adult mice are known to be fibro-adipogenic progenitors, a population of multipotent tissue-resident MSCs found in skeletal muscle (Joe et al., 2008), heart (Soliman, personal communication, 2015), lungs (Lo, personal !108communication, 2015), and bone marrow of long bones (Eisner and Nguyen, personal communication, 2016). With osteogenic potential, these cells suggest that one additional hypothesis would be the loss of Ehmt2 in the Pdgfra lineage disrupted the dynamics of resident MSCs, which led to failure for the bones to meet at the suture. Neonatal specimens at P2 were prepared in transverse sections and stained with alcian blue and picrosirius red. At the frontal suture, both WT and KO samples stained red, although in different shades, suggesting differences in tissue type (Fig. 3.18a, b). This revealed that a layer of cells existed at the suture in the KO but did not form ossified bone that would be visible by micro-CT. H/E staining further showed the fusion of the suture in WT mice was completed within the first week, with the disappearance of the cell-rich layer of connective tissue. However, in KO mice, these cells persisted (Fig. 3.18c, d). This layer, likely containing MSCs, might provide an important insight in the role of Ehmt2 in the growth of the frontal bones and formation of the suture. Staining with more detailed markers would be required to identify cells in the layer.    !109f/wt f/fAB-PSRa          Fig. 3.18 a) Coronal sections of the head of Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, alcian blue and picrosirius red co-stained, P2. b) Coronal sections showing frontal bones at the location of the frontal suture of littermates, alcian blue and picrosirius red co-stained, P2. c) Coronal sections of the frontal suture of litter mates, H/E, P7. d) Coronal sections of the frontal suture of litter mates, H/E, P14.!110f/wt f/fAB-PSRf/wt, P2 f/f, P2f/wt, P7 f/f, P7bcdSkull deformation in Pdgfra-Cre Ehmt2 knockout mice In addition to the lack of fontanelle closure, the KO mice also showed obvious microcephaly and shortening of the skull. Using micro CT, I measured the length of the skull from the top of the incisors to the planum occipitale, which showed significant shortening in KO compared to WT specimens (Fig. 3.19).       Data 1WT KO0510152025cranial length (mm)Fig. 3.19 a) Micro-CT of the head of Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates. b) Cranial length measurement from the top of the incisors to the planum occipitale, p < 0.05, n = 6.!111Ehmt2 influences cell density in the brain One of the hallmarks of Kleefstra syndrome is severe mental retardation. Previous reports on neuronal conditional knockout of Ehmt2 and Ehmt1 had revealed learning disability but no obvious physical disabilities. The Pdgfra lineage tracing showed strong labelling of the dorsal root ganglion (Fig. 3.20a), which is congruent with the understanding that it is expressed by Schwann cells of the peripheral nervous system (Peggy Assinck, personal communication, 2014). In the central nervous system, it is believed to be a marker of oligodendrocytes, and I observed some labelling in the brain (Fig. 3.20b). H/E sections of the brains of PdgfraCre Ehmt2floxed/null mice showed an astonishing density of cell nuclei compared to WT (Fig. 3.20c). However, specific cell types needed to be identified for any further conclusion. !112         DISCUSSION Comparison with Wnt1Cre Ehmt2floxed/floxed Previously, Higashihori et al. (personal communication, 2010) had studied a related mouse model, in which Ehmt2 was deleted under the control of the Wnt1 locus, resulting in a conditional knockout in the developing neural crest. In this model, mice died perinatally, and also spinal corddorsal root ganglionWT KOP2 medial longitudinal fissureFig. 3.20 a) Confocal microscopy of tdTomato expression in spinal cord and dorsal root ganglion of Pdgfra-Cre Rosa-tdTomato b) Confocal microscopy of tdTomato expression in brain and tissues of the head. c) Coronal section of brain at the medial longitudinal fissure, H/E, P2.!113a bchad similar deformations of the skull. Some had open fontanelles. Since Wnt1 is a well known neural crest marker, and Pdgfra is expressed in the neural crest and the mesoderm, the consistent reproduction of similar phenotypes in PdgfraCre Ehmt2floxed/null confirmed that the role of Ehmt2 in skull development lies within the Pdgfra lineage. In Wnt1Cre Ehmt2floxed/floxed, staining with osteogenic markers revealed delayed ossification of craniofacial bones. At E14.5 and 15.5 the ossification centres had prolonged expression of Twist combined with a 24h delay in Runx2 and osteopontin expression. Runx2 is a master regulator of osteoblast differentiation (Komori et al. 1997). Twist is an antagonist of Runx2 (Bialek et al. 2004), and is expressed in pre- but not mature osteoblasts. Osteopontin (Opn/secreted phosphoprotein-1) is an extracellular matrix protein expressed in bone. The suggestion that Ehmt2 might be regulating ossification is promising, since the suture area in PdgfraCre Ehmt2floxed/null showed connective tissue without ossification. Pdgfra-Cre Ehmt2 knockout mice as a model for Kleefstra syndrome As described in Chapter 1, Kleefstra syndrome is caused by a heterozygous loss-of-function mutation in Ehmt1/GLP. Many of the skull abnormalities of the syndrome were evident in the PdgfraCre Ehmt2floxed/null model. By comparison with the human disease and many other conditional knockouts in mice, my model fitted well on a spectrum of developmental outcomes (Table 3.1). Since mice with a mutated Ehmt2 ankyrin-repeat domain appeared normal, and that full Ehmt2null/null mice die early in embryogenesis, my data suggest that the syndromic congenital anomalies were likely due to the lack of lysine methyltransferase activity of Ehmt2, and that this function could not be compensated by Ehmt1. !114   METHODS AND MATERIALS Micro CT Standard specimen micro CT scanning was performed according to the skeletal bone protocol at the Centre for High-Throughput Phenogenomics at The University of British Columbia. 3D renderings were isosurface renderings made using Amira and MicroView. Cranial length measurements were carried out in MicroView. G9A 3AM/MNEURONG9A GLP KOGLPWT/NULLHUMANGLPWT/NULLGLP 3AM/MWNT1G9A KO PDGFRAG9A KO G9A GLP NULL/NULLnormallearning disabilitydeath at D2death at D0death at E9.5-12.525B125B2 CONTROL5mmFONTANELLE CLOSUREKO25B325C128A225B2Table 3.1 Comparison of phenotypes of different Ehmt2 (G9a) and Ehmt1 (GLP) mutations in human and mice. “3A M/M” refers to a homozygous three-amino acid mutation in the ankyrin repeat domain that leads to loss of domain function. At the far right of the spectrum, full loss of either Ehmt2 or Ehmt1 results in the earliest embryonic death.!1154. The role of Ehmt2/G9a in adipose tissues INTRODUCTION: CURRENT UNDERSTANDING OF EHMT1/2 IN ADIPOSE TISSUE DEVELOPMENT AND HOMEOSTASIS In mice, Ehmt2 and Ehmt1 play a critical role in the differentiation of adipose cells. In immortalized brown preadipocytes, H3K9me2 is enriched over the entire PPARγ gene (Wang et al. 2013). Upon induction of adipogenesis in these cells, H3K9me2 and Ehmt2 levels decrease, leading to activation of PPARγ expression. Loss of Ehmt2 in cultured preadipocytes causes premature activation of PPARγ expression and increased adipogenesis. In parallel, Ehmt2 appears to activate expression of the negative adipogenic regulator Wnt10a. Ehmt2 is found at the Wnt10a promoter and can activate its expression independent of its methyltransferase activity. Mice with conditional knockout of Ehmt2, under the control of Fabp4-Cre (fatty acid binding protein 4, adipocyte, aka aP2) in differentiating preadipocytes, are obese, with increased expression of adipogenic markers (Wang et al. 2013). Fabp4 is a direct PPARγ target gene and is robustly induced when preadipocytes have differentiated into immature adipocytes (Cho et al. 2009). For an in vivo model to study brown fat, a conditional knockout of Ehmt1 was made by Ohno et al. (2013), who used Myf5-Cre to drive the gene deletion. They found that mice were born with substantially smaller interscapular brown adipose tissue and smaller brown adipocytes. Myogenic genes lost H3K9me2 and were ectopically expressed, which is a sign that Ehmt1 is critical in determining brown fat vs muscle cell fate. Ohno et al. (2013) also used Adiponectin-Cre (Adipoq) to drive the conditional knockout of Ehmt1 in both brown and white fat. Adiponectin is a much later developmental marker than Myf5, and is expressed !116in both mature brown and white adipocytes. With this model, the group tested adaptive thermogenesis by cold challenge and chemical activation of the β3-adrenoceptor pathway, which are well known activators of brown adipose tissue for heat generation. They found that Ehmt1 is required for adaptive thermogenesis. The researchers further found the lack of Ehmt1 led to enlarged brown adipocytes, obesity and insulin resistance much more quickly than wildtype with high fat diet. Using in vitro methods, the researchers point to Ehmt1 as a stabiliser of the Prdm16 transcriptional complex as the underlying mechanism to positively regulate brown fat activation. These studies highlight seemingly contrasting roles of Ehmt1/2 in brown vs white fat, and tissue development vs energy homeostasis. Ehmt2 in white fat prevents adipogenic differentiation. Ehmt1 in brown fat promotes brown adipocyte fate and activation. Taken together, however, a plausible theory can be drawn, if we assume that Ehmt1 and 2 have similar roles. During development, Ehmt1/2 repress white adipocyte cell fate in one lineage, promotes brown adipocytes in another lineage. In the adult, Ehmt1/2 repress lipid storage through PPARγ, and promotes thermogenesis through Prdm16. As we associate lipid accumulation with white fat, and thermogenesis with brown fat, then Ehmt1 and 2 appear to be an agent that promotes the good kind while repressing the bad kind of fat, which makes it a promising subject in developing therapies for obesity and related diseases. However, there are still a few details that are not in agreement, such as H3K9me2 repressing lipid storage in immortalized brown preadipocytes and Ehmt1 knockout brown adipocytes being smaller than wildtype. Studies focused on Ehmt1 in brown fat and Ehmt2 in white fat, but Ehmt2 in brown adipose tissue has not be explored. As another major mesenchymal tissue, adipose tissue is also an important topic in the understanding of Ehmt2, especially since previous publications have found its importance in white adipose tissue, and its homologue, Ehmt1, to be required for brown adipose tissue specification and activation. However, no in vivo study has been done regarding Ehmt2 in brown !117adipose tissue. For this study, I continued to analyse the PdgfraCre Ehmt2floxed/null mouse model, as both brown and white adipose tissues derive this lineage. I report in this chapter the role in which Ehmt2 limits lipid accumulation and is required for normal brown adipose development. RESULTS Tracing the Pdgfra lineage in white adipose tissue To evaluate the role of Ehmt2 in adipose tissue, I employed the Pdgfra lineage, since the majority of adipose tissue develop from the mesoderm (see Chapter 1). Using flow cytometry, I analysed the subcutaneous adipose tissues from PdgfraCre RosaYFP and PdgfraCre RosatdTomato mice, which confirmed that tissue-resident adipogenic progenitors (PTPRC- PECAM- LY6A/E+) from adult mice are predominantly derived from the Pdgfra lineage (Fig. 4.1). !118      Adipogenic differentiation of Ehmt2 knockout adipocyte progenitors Adipogenic progenitors purified by FACS from neonatal PdgfraCre Ehmt2floxed/null mice were cultured for three days in proliferative conditions. Then, they were cultured in adipogenic differentiation conditions for 5 days and 9 days. The KO cells accumulated lipid vacuoles at enormously faster rates than WT cells (Fig. 4.2). 100 101 102 103 104Hoechst-A0100200300400# Cells47100 101 102 103 104PE-Cy7-A: SCA PE-Cy7-A100101102103104APC-A: LIN APC-A20.9 16.336.826Q1: SCA PE-Cy7-A—, LIN APC-A+100 101 102 103 104PE-A: TOMATO PE-A020406080100% of Max100 101 102 103 104PE-A: TOMATO PE-A020406080100% of MaxQ3: SCA PE-Cy7-A+, LIN APC-A—100 101 102 103 104PE-A: TOMATO PE-A020406080100% of MaxQ4: SCA PE-Cy7-A—, LIN APC-A—100 101 102 103 104PE-A: TOMATO PE-A020406080100% of MaxQ2: SCA PE-Cy7-A+, LIN APC-A+FATLY6A (SCA1)PDGFRA-CRE-YFPLIVESpecimen_001_1B5F 4.fcsEvent Count: 7000100 101 102 103 104FITC-A100101102103104APC-A70.73.611.6324.1LIVESpecimen_001_1B6F 4.fcsEvent Count: 8414100 101 102 103 104FITC-A100101102103104APC-A0.02400.5199.5ENDOTHELIUMHEMATOPOIETICLIVESpecimen_001_1B5F 5.fcsEvent Count: 5451100 101 102 103 104FITC-A100101102103104APC-A5.89 14.952.426.8LIVESpecimen_001_1B6F 5.fcsEvent Count: 5595100 101 102 103 104FITC-A100101102103104APC-A24.9 0075.1LIVESpecimen_001_1B5F 6.fcsEvent Count: 6793100 101 102 103 104FITC-A100101102103104PE-Cy7-A5.89 51.233.98.95LIVESpecimen_001_1B6F 6.fcsEvent Count: 8228100 101 102 103 104FITC-A100101102103104PE-Cy7-A56.9 0.0610.02443SCA-1FAT MUSCLE MUSCLEFAT FATLIVESpecimen_001_1B5M 2.fcsEvent Count: 4160100 101 102 103 104FITC-A100101102103104APC-A3.92 34.820.5 33.8LIVESpecimen_001_1B6M B_002.fcsEvent Count: 4288100 101 102 103 104FITC-A100101102103104APC-A35.4 1.459.6 0.07LIVESpecimen_001_1B6M B_003.fcsEvent Count: 3585100 101 102 103 104FITC-A100101102103104APC-A8.01 0.1190.7 0.28LIVESpecimen_001_1B5M 3.fcsEvent Count: 4708100 101 102 103 104FITC-A100101102103104APC-A0.51 6.1429.1 59.2SATELLITE CELLSWTTransgenicFig. 4.1 a) Flow cytometry of digested subcutaneous adipose tissue from Pdgfra-Cre Rosa-tdTomato mice. Live cells (Hoechst+) were categorized by expression of PECAM (CD31), PTPRC (CD45), and LY6A (SCA1). The lower four histogram charts reflect each of the four quadrants of the gating strategy shown in the top right plot, showing abundance of cells expressing tdTomato in transgenic mice (red) compared to control mice (blue). This shows many blood cells and endothelial cells in fat are traced by Pdgfra-Cre Rosa-tdTomato. Adipocyte progenitors (LY6A+) are entirely derived from this lineage. b) Flow cytometry of digested subcutaneous adipose tissue from Pdgfra-Cre Rosa-YFP showing adipose progenitor cells (LY6A+) expressing YFP. PECAM+ cells and PTPRC+ cells were excluded from this gate.!119ab   WT KO050010001500avg vacuole area per 100 nuclei***WTKOFig. 4.2 Primary adipogenic progenitors from Pdgfra-Cre Ehmt2-f/f mice differentiate faster. Cells were purified by FACS and induced for differentiation for 9 days. Left: phase contrast microscopy. Right: oil red-O and hematoxylin co-stained. Bottom: Quantification of average vacuole size by oil red-O area, normalised to nuclei.!120Ehmt2 is required for normal brown adipose tissue development The PdgfraCre Ehmt2floxed/null mice also revealed a drastic abnormality in brown adipose tissue. At P2, the brown adipose tissue of KO neonates had much higher lipid content than WT, and far fewer nuclei (Fig. 4.3). To quantify this observation, I stained the transverse sections against perilipin, which lines the lipid vacuoles. The data revealed a significant disparity between brown adipocytes of KO and WT neonates (Fig. 4.4). Thus, Ehmt2 is required for normal brown adipose tissue development; without it, the tissue's morphology resembled an intermediate between white and brown adipose tissue (Fig. 4.5). The finding here is not congruent with those of its homologue in Ohno et al. (2013), in which Myf5Cre Ehmt1floxed/floxed mice have smaller brown adipocytes than wildtype. The brown adipocytes here is clearly larger in the knockout. However, white adipose tissue does not appear to be affected (Fig. 4.6). !121   PdgfraCre G9afloxed/nullbrown fat at D2body weight 1.3gWTKOFig. 4.3 Thoracic transverse sections of Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates at P2, showing interscapular and body wall brown adipose tissues, H/E.!122       0 50 100 150 2000510152025Data 2vacuole cross sectional areafrequency (%)WTKOFig. 4.4 a) Immunohistochemistry of perilipin in interscapular brown adipose tissues in Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates at P2. b) Frequency distribution of lipid vacuole size, measured by cross sectional area.Ehmt2floxed/floxedPdgfraCre/wtWT brown fat KO brown fat WT white fatFig. 4.5 Interscapular brown adipose tissue in WT and KO is compared to subcutaneous white adipose tissue at P2, H/E. The morphology of brown adipose tissue in KO is intermediate between normal brown and normal white adipose tissue.!123WTKO   Brown adipose tissue maintenance in Pdgfra-Cre Ehmt2 knockout mice Although Ehmt2 was required for brown fat development, it is unknown if it is required for maintenance. To evaluate this, I generated another model, where Ehmt2floxed was crossed with UbcCreERT2. Under the control of the ubiquitin C locus, this inducible conditional knockout allowed me to delete Ehmt2 during neonatal maintenance. At 18 days after the start of tamoxifen induction, no difference in brown adipose morphology was detected between the KO and the WT mice (Fig. 4.7). WT white fat KO white fatFig. 4.6 Subcutaneous white adipose tissue appear similar between WT and KO mice. P2, H/E.!124   Ehmt2 influences the development of dermal adipose tissue Like other adipose tissues, dermal adipose tissue was also verified to be derived from the Pdgfra lineage. Microscopy of PdgfraCre RosatdTomato embryos further demonstrated strong labelling of dermal adipose tissue. In PdgfraCre Ehmt2floxed/null mice, the dermis was surprisingly thin (Fig. 4.8). This is unexpected, since I discovered that adipocytes were larger in the KO. However, it is difficult to conclude at this point, since Ehmt2 might play different roles in other cell types in the dermis. Ehmt2floxed/floxedUbcCreERT2/wt D5D0tamoxifenD18histology4-5 weeks oldWT HET KOFig. 4.7 a) Schematic diagram of the generation of Ubc-CreERT2 Ehmt2-f/f mice. b) Timeline of Cre induction and tissue collection. c) H/E stain of interscapular brown adipose tissue.!125    bonemuscledermisepidermisWT KOP2 scalpFig. 4.8 a) Confocal microscopy of bone, muscle, dermal, and epidermal layers in the upper limb of Pdgfra-Cre Rosa-tdTomato embryo, showing tdTomato expression concentrated in the dermis. Theiler stage 23 (E14.5), transverse section. b) Coronal section of the scalp of Pdgfra-Cre Ehmt2-f/wt and Ehmt2-f/f littermates, H/E, P2.!126DISCUSSION In this study, I have confirmed the developmental lineage of white adipose tissues coming entirely from the Pdgfra lineage. The loss of Ehmt2 in this lineage resulted in a dramatic change in the potential of tissue resident adipogenic progenitors. The data shows that Ehmt2 is involved in the repression of adipogenic differentiation. The loss of Ehmt2 in this lineage also resulted in abnormal brown adipose tissue development; knockout mice were born with very high lipid content in interscapular brown adipose tissues. However, the role of Ehmt2 in the maintenance of brown adipose tissue is still uncertain. In order to clarify the active postnatal involvement of Ehmt2 in brown fat over time will need analyses that assess the storage of fats during diet change or other transformations. One potential problem is that the elevated lipid accumulation found in cultured KO adipocyte progenitors did not manifest in morphological changes in subcutaneous white adipose tissue in vivo, despite an obvious effect in brown adipose tissue. It is possible that the effects of the knockout is mainly on the rate of lipid accumulation during development, and that cannot be observed in terminally differentiated white adipocytes. Again, in vivo analyses at different times during diet change may yield answers to this problem. In addition to confirming previous finding about the role of Ehmt2 in white adipose tissue differentiation using tissue resident progenitor cells, this study is the first time that Ehmt2’s involvement in brown fat is assessed in vivo. Similar to the role of Ehmt1, I found that Ehmt2 is required for brown adipocyte fate during development. Without it, the tissue over-accumulate lipids and acquire a phenotype that resembles neither brown nor white adipose tissue. This similarity to Ehmt1 suggests that Ehmt2 might be promoting thermogenesis through similar mechanisms, such as through Prdm16. This also reveals the complex relationship between Ehmt1 !127and 2, as it suggests neither is able to compensate for the loss of the other. This makes the adipose tissue an ideal subject to uncover the degree of redundancy in the targets of Ehmt1 and 2, as well as the importance of their heterodimerization. METHODS AND MATERIALS Histology Adipose tissue were either analysed intact in neonates, or dissected from mice, fixed in 4% paraformaldehyde overnight followed by 70% ethanol overnight, and then embedded in paraffin following standard protocols. Tissues were cut with a microtome in a cross sectional orientation. Cross sections of 5 mm thickness were then mounted onto glass slides (Thermo Fisher Scientific, USA) and stained with H/E or perilipin following standard protocols. Lipid vacuole size measurement Lipid vacuoles size was measured in a similar fashion as myofibre size, described in Chapter 2. Cross sectional area was measured using immunohistochemistry of perilipin, which clearly delineates the boundaries of lipid vacuoles. Statistical analysis of the frequency distribution used the same permutation approached described in Chapter 2, which employed the Community Ecology Package ('vegan'), developed by Oksanen (2015). !1285. Conclusion    SUMMARY In this work I have described three studies employing four mouse conditional knockout models aimed at expanding the current understanding of Ehmt2 in the development of different tissue systems. In the attempt to validate existing theories about Ehmt2’s role in skeletal muscle, I have described in Chapter 2 that those theories are false in mouse models. In Chapter 3, I explored the role of Ehmt2 in the development of the Pdgfra lineage, and found that it is absolutely required for Ehmt2/G9aES cell differentiationDong et al. 2008, Yamamizu et al. 2012germ cell developmentTachibana et al. 2007heart developmentInagawa et al. 2013lymphocyte developmentThomas et al. 2008, Lehnertz et al. 2010leukemiaLehnertz et al. 2014drug addictionMaze et al. 2010cognition & bahaviourSchaefer et al. 2009, Kramer et al. 2011embryogenesisTachibana et al. 2002skeletal muscle developmentmuscle regenerationcraniofacial developmentbrown adipose tissue developmentFig. 5.1 Schematic diagram showing different conditional knockout studies of Ehmt2 in mice. Contributions from this work are in red.  !129not requirednot requiredrequiredrequirednormal craniofacial development and a large contributing factor to viability. Lastly I reported, in Chapter 4, that Ehmt2 plays a critical role in adipogenesis and the development of brown fat. GENERAL DISCUSSION The expansion of understanding about Ehmt2 since its discovery has given the outline of a large regulatory gene that has many different functions. Previous research using knockout models of this gene in ES cells, heart, cancer, brain, immune system, etc., has found Ehmt2 with a set of roles that changes between cell types and tissue systems. The expansive characterisations presented in this dissertation adds skeletal muscle development, regeneration, craniofacial development, and brown fat development, to the repertoire of tissue-specifc knockout models. Some of the surprising findings, including the complete dispensability of Ehmt2 in myogenesis, and the highly peculiar calvarial bone malformations, demonstrate extreme differences in how this gene behaves in different cell lineages. We know that Ehmt2 can affect cell fate choices, but this work reinforces the concept that its role in specifically regulating cell differentiation, is highly lineage-dependent, or in other words, is the ideal candidate to illuminate the complexity of mammalian developmental biology. There are, however, some commonalities between the roles of Ehmt2 in different cell types; for example, its push for intramembraneous ossification in calvarial bones and lipid accumulation in adipose tissues, reveal similar properties of favouring terminal differentiation. Cellular variation of Ehmt2/G9a function The different cell types investigated in this work show different roles of Ehmt2. In preadipocytes, I confirmed that Ehmt2 prevents adipogenic differentiation and found novel insights into its ability to limit the lipid storage of brown adipose tissues in neonatal mice. This demonstrated that in brown adipocytes in vivo, Ehmt2 has a role congruent to that of Ehmt1 (Ohno et al. 2013). In the   !130head, Ehmt2 promoted normal late stage cranial bone growth in mice, which confirms findings from Higashihori et al. (personal communication 2010) that Ehmt2 likely mediates key regulatory functions to promote intramembraneous ossification. In skeletal muscle progenitors and satellite cells, however, Ehmt2 does not contribute any significant function. Thus, this research shows evidence that Ehmt2 performs different functions as cells from the mesoderm develop into three different tissues. Redundancy in the function of Ehmt2/G9a and Ehmt1/GLP The complex compensatory relationship between Ehmt2 and Ehmt1, which are homologues and form heterodimers, has been well studied in ES cells and various loss-of-function animal models (Fig. 1.3). The SET domain, which produces the methylation, in Ehmt2 is the more dominant one, and can compensate for the loss of the SET domain in Ehmt1, but not the other way around. However, the reverse is true for the ankyrin repeat domain, as Ehmt1 uses it to read H3K9 methylation marks and is an absolutely required function, but the one in Ehmt2 is not. The consequence of losing both of these domains in either one gene would be failure of development (Table 3.1). Results from this study appears mostly in line with this current understanding. In normal development of the brown adipose tissue, the requirement of Ehmt1 was previously described. The findings described in Chapter 4 suggests an independent requirement for Ehmt2, and that neither of the two genes can compensate for the loss of the other to restore normal development. Ehmt1 was found to act through Prdm16, but the molecular mechanism for Ehmt2 in brown fat has not been identified. If the two share common molecular targets and cannot compensate for the loss of the other, then it would be possible that the heterodimerization and differential protein domain dominance could be at play in brown fat. The only concern is the lack of a phenotype in   !131white adipose tissue in the Pdgfra-lineage Ehmt2 knockout. However, I would argue that it is unlikely to be due to compensation by Ehmt1; since Fabp4-lineage Ehmt2 knockout mice were obese (Fabp4 is a specific marker during preadipocyte differentiation), it could be that it requires observation at a later age. Ehmt2/G9a as a regulator of terminal differentiation Although dispensable in skeletal muscle, this study found Ehmt2 to be critical in brown adipose development and cranial bone development. Seemingly different on the surface, yet there are some characteristics that are common in these two findings. In cranial bones, this work, along with others, contributes to the idea that Ehmt2 supports osteoblast ossification. In brown fat, it is possible to be playing a role similar to Ehmt1 in preventing PPARγ-directed adipogenesis. These two appear to be in opposite directions in the differentiation programmes of these progenitor cells, but in both cases, Ehmt2 resembles part of a downstream machinery that is utilized by a master regulator. I would argue that it is an unlikely candidate to serve as the initiator in the regulation of terminal differentiation in these tissues, mostly because it associates with completely different proteins in different cell types, and it has no DNA-binding capability on its own. Instead, it is far more likely to be directed by key regulators of the signalling pathways of differentiation in each cell type, such as the PPARγ pathway in adipose tissue and the Runx2 pathway in cranial bones. Interestingly, in skeletal muscle, Bharathy, Ling, & Taneja (2013) proposed a mechanism for Ehmt2 that fits exactly in this concept, in which Ehmt2 is a tool in the Sharp1-Myod relationship for regulating the timing of terminal myoblast differentiation. Unfortunately the proposal is unsupported by in vivo data. Other lines of research have hinted at various known regulators in adipose tissue and bone that Ehmt2 could be associated with, which are mentioned in Chapters 3 and 4, and the next step in this research is to identify protein-protein interactions to establish the mechanism.   !132Pdgfra as a marker of a critical developmental lineage The studies presented in this dissertation that involved the use of Pdgfra as a driver of conditional knockouts revealed that Ehmt2 has far more influence in this cell lineage than others. Since this lineage traces the origins of many mesenchymal tissues, it provides an insight about Ehmt2 that could generalise this gene’s functions as gravitating towards mesenchymal tissues. Broadened view of epigenetics By strengthening the concept of a highly dynamic multifunctional methyltransferase, this work helps to bring light to epigenetic mechanisms beyond the histone code. Having multiple epigenetic capabilities within one molecule, Ehmt2 is the poster child of a holistic view of epigenetics. FUTURE DIRECTIONS Tissue maintenance Although it is clear that Ehmt2 is dispensable in adult skeletal muscle regeneration, its role in maintaining brown adipose tissue needs more studies, which should analyse the impact of the loss of Ehmt2 as mice matures, ages, and changes diet. I hypothesize that, since Ehmt2 is reported to limit lipid storage via regulating PPARγ and Wnt10a (Wang et al. 2013), it is complementing Ehmt1 in promoting a brown adipocyte identity. It remains to be tested whether Ehmt2 has the same targets as Ehmt1 in brown fat. Furthermore, Ehmt2 may also have a role in the homeostasis and repair of cranial bones in the adult. It is unclear if suture-resident MSCs were impacted in the mouse model described in chapter 3 to produce the fontanelle defect. The Gli1+ MSCs, which are proposed by Zhao et al.   !133(2015) to be responsible for cranial bone repair, would be a good candidate to study to better understand Ehmt2 in the repair process. Brain development Perhaps the most clinically relevant aspect of modelling Kleefstra syndrome using Ehmt1 and Ehmt2 is mental retardation, a key manifestation of Ehmt1 loss-of-function in humans. A good number of studies were done using neuron-specific Ehmt1/2 knockouts, concluding findings in dendritic and synaptic defects, behavioural abnormality, and drug addiction (see Chapter 1). In contrast, the Pdgfra-Cre Ehmt2 knockout model in this study is mesoderm and neural crest specific, thus not affecting the neuronal lineage; yet in Chapter 3, histology of the brain in knockout mice appeared drastically abnormal. Is mental retardation in Kleefstra syndrome a result of cranial bone malformation, or are they separable defects? The knockout mice used in this study models the cranial bone defects of human disease, showing smaller cranial volume (see Chapter 3). It is possible that the cranial volume alone is the source of mental retardation and requires further exploration, in which case the Pdgfra-Cre Ehmt2 knockout is an ideal tool. If not, then Ehmt1/2 may have entirely novel effects in brain development, since Pdgfra lineage contributes to oligodendrocytes and is a specific marker of adult oligodendrocytes. Myelinating oligodendrocytes in the brain is important in the process of learning. Recent findings of missense mutations in the ankyrin repeat domain of Ehmt1 and the SET domain of Ehmt2 in autism subjects makes this an exciting topic for future research (Balan et al., 2014). Other mesenchymal tissues Given the discoveries about Ehmt2 in osteoblasts, adipocytes, and its possible involvement in suture MSCs, the next logical step is to study inducible conditional knockout models of Ehmt2 in other mesenchymal lineages. The Pdgfra marker used in this work is an excellent marker of MSCs   !134in the adult (see chapter 1). Currently, we lack information about Ehmt2’s status in the development and tissue maintenance of cartilage, long bones, endothelium, and dermis, all of which are also derivatives of the paraxial mesoderm. Some of these aspects would have large implications for stem cell biology; for instance, some skin-derived precursor cells (SKPs) have neural crest origins, and can generate both neural and mesodermal progeny. Studying the role of Ehmt2 in such a context could help stem cell biologists understand the regulatory mechanisms in the fate decisions of adult multipotent stem cells. In addition, an inducible conditional reporter model of Ehmt2 would greatly aid time-specific studies of developmental tissues. Non-histone post-translational modifications Although Chapter 2 has highlighted the uncertainty of Ehmt2’s function in non-histone methylation in the context of myogenesis, there is a growing body of work that identified many non-histone substrates of Ehmt2’s SET methylatransferase domain (Biggar & Li, 2015). Many of these methylations are postulated to impact cell signalling, however, most still require in vivo validation to clarify the degree of influence of this mechanism. Therapeutic relevance Whether or not Ehmt2 has a profound implication for obesity and diabetes remains an open question. Disease modelling using diet, induced thermogenesis, and long-term glucose response are immediate options if suitable transgenic mice are available. An inducible, brown/white adipose tissue specific driver of the Ehmt2 knockout would be ideal. More molecular understanding of Kleefstra syndrome would be desired, but most of the clinically relevant aspects would focus on the large growing body of research of Ehmt1/2 in neuroscience and mental health, which are largely outside of the scope of this work. Additionally, technology is   !135becoming more available, which, in the not-so-distant future, will greatly expand prenatal screening. I remain hopeful of genetic counselling becoming useful for even rare syndromes such as Kleefstra syndrome, but also mindful that a new era of mass-scale biotechnology will require new regulatory frameworks of health information. SIGNIFICANCE OF THIS WORK Developmental biology As a wide survey of Ehmt2 in the developmental biology of multiple tissue systems, this work has solidified the view that Ehmt2 is an incredibly versatile gene, as it can act as repressor or activator of different cellular programmes. It has made Ehmt2 and its product, H3K9me2, a complex yet fascinating genetic regulator. The research described here forms an important foundation for future understanding of stem cells and tissue repair. Genetic modelling This work not only produced a disease model of Kleefstra syndrome, but also used the development of the various genetic mouse models as opportunities of technical innovation. It has established a new standard in the evaluation of genetic knockout models and statistical methods that address some of the largely ignored statistical pitfalls in this field. Value of negative findings In clinical research, there is a wellknown growing discrepancy between the investment in therapeutic development and the rate of approval of new therapies. For the cause of many failed drug trials, London and Kimmelman (2015) point to the lack of negative publications in the medical sciences. Stalled projects are significantly less likely to be published than positive   !136findings. The lack of information of negative findings has led to uninformed investments in projects that were certain to fail. Part of this problem begins in basic medical science research, where the discrepancy is exacerbated by citation distortions that disfavour negative publications (Greenberg 2009). For example, there has been a widely-held belief that β amyloid, a protein accumulated in the brain in Alzheimer’s disease, is produced by and injures skeletal muscle of patients with inclusion body myositis. However, there were actually more primary publications critical of the theory than those supporting it. Due to slight bias favouring positive data and massive cascading effects of reviews, related publications, secondary and tertiary publications citing those primary publications, the belief about β amyloid generated an unfounded authority of claims and became mainstream. Combined with the mechanism of impact factors, this created a vicious cycle where the distortion becomes amplified with time. The study presented in chapter 2 is the first in vivo validation of the role of Ehmt2 in skeletal muscle, and it found that previous modelling using a widely accepted immortalised cell line did not reflect in vivo cellular behaviour at all. However, a number of reviews in the field has already concluded that Ehmt2 plays a critical role in myogenesis and that it held great therapeutic potential. Without the publication of this study, the cascading effects would lead to significantly more research funding invested in a fallacy. Thus, in the words of London and Kimmelman (2015), “Negative results advance drug discovery by helping to improve our understanding of both diseases and treatments” (p. 2).  !137Bibliography Abmayr, S. M. and G. K. Pavlath. 2012. Myoblast fusion: Lessons from flies and mice. Development 139: 641–656. Abzhanov, A. and C. J. Tabin. 2004. Shh and Fgf8 act synergistically to drive cartilage outgrowth during cranial development. Dev. Biol. 273: 134–148. Abzhanov, A., D. R. Cordero, J. Sen, C. J. Tabin and J. A. Helms. 2007. Cross- regulatory interactions between Fgf8 and Shh in the avian frontonasal prominence. Congenital Anomalies 47: 136–148.  Alexander Medvinsky, Elaine Dzierzak. (1996). Definitive Hematopoiesis Is Autonomously Initiated by the AGM Region. Cell, Volume 86, Issue 6, p897–906. Allis, C. D., Caparros, ML., Jenuwein, T., & Reinberg, D. (Eds.). (2015). Epigenetics (2nd ed.). New York: Cold Spring Harbor Laboratory Press. Anderson, M. J. 2001. A new method for non-parametric multivariate analysis of variance. Austral Ecology , 26 : 32–46. Asp P, Blum R, Vethantham V, Parisi F, Micsinai M, Cheng J, Bowman C, Kluger Y, Dynlacht BD. Genome-wide remodeling of the epigenetic landscape during myogenic differentiation. Proc Natl Acad Sci U S A 2011 May 31;108(22):E149-58. Ataliotis, P. & Mercola, M. (1997). Distribution and functions of platelet-derived growth factors and their receptors during embryogenesis. Int Rev Cytol, 172:95-127. Aulehla, A., & Pourquie, O. (2008). Oscillating signaling pathways during embryonic development. Curr Opin Cell Biol., 20(6), 632-7. Balan S, Iwayama Y, Maekawa M, Toyota T, Ohnishi T, Toyoshima M, Shimamoto C, Esaki K, Yamada K, Iwata Y, Suzuki K, Ide M, Ota M, Fukuchi S, Tsujii M, Mori N, Shinkai Y, & Yoshikawa T. (2014). Exon resequencing of H3K9 methyltransferase complex genes, EHMT1, EHTM2 and WIZ, in Japanese autism subjects. Mol Autism, 5(1), 49. Barski, A., Cuddapah, S., Cui, K., Roh, T.-Y., Schones, D.E., Wang, Z., et al. 2007. High-Resolution Profiling of Histone Methylations in the Human Genome. Cell, 129: 823–837. doi:10.1016/j.cell.2007.05.009. PMID:17512414. Berger SL, Kouzarides T, Shiekhattar R, Shilatifard A (2009). "An operational definition of epigenetics". Genes Dev. 23 (7): 781–3. doi:10.1101/gad.1787609. PMC 3959995free to read. PMID 19339683 Bergstrom, D. A. and S. J. Tapscott. 2001. Molecular distinction between specification and differentiation in the myogenic basic helix-loop-helix transcription factor family. Mol. Cell. Biol. 21: 2404–2412. Berkes CA, Tapscott SJ. MyoD and the transcriptional control of myogenesis. Semin Cell Dev Biol 2005 Aug-Oct;16(4-5):585-595. Berry, R. & Rodeheffer, M. S. Characterization of the adipocyte cellular lineage in vivo. Nature Cell Biol. 15, 302–308 (2013). Bharathy, N., Ling, B. M., & Taneja, R. (2013). Epigenetic regulation of skeletal muscle development and differentiation. Subcell Biochem., 61, 139-50. !138Billon N, Iannarelli P, Monteiro MC, Glavieux-Pardanaud C, Richardson WD, Kessaris N, Dani C, Dupin E. (2007). The generation of adipocytes by the neural crest. Development, 134(12), 2283–2292. Billon, N. & Dani, C. (2012). Developmental origins of the adipocyte lineage: new insights from genetics and genomics studies, Stem Cell Rev, 8(1), 55–66. Bishayee S, Majumdar S, Khire J, Das M. Ligand-induced dimerization of the platelet-derived growth factor receptor. Monomer-dimer interconversion occurs independent of receptor phosphorylation. J Biol Chem. 1989 Jul 15;264(20):11699–11705. Blau HM, Pavlath GK, Hardeman EC, Chiu CP, Silberstein L, Webster SG, Miller SC, Webster C. Plasticity of the differentiated state. Science 1985 Nov 15;230(4727):758-766. Blum R, Dynlacht BD. The role of MyoD1 and histone modifications in the activation of muscle enhancers. Epigenetics 2013 Aug;8(8):778-784. Blum R, Vethantham V, Bowman C, Rudnicki M, Dynlacht BD. Genome-wide identification of enhancers in skeletal muscle: the role of MyoD1. Genes Dev 2012 Dec 15;26(24):2763-2779. Boettiger, D., M. Enomoto-Iwamoto, H. Y. Yoon, U. Hofer, A. S. Menko and R. Chiquet-Ehrismann. 1995. Regulation of integrin a5b1 affinity during myogenic differentiation. Dev. Biol. 169: 261–272. Borycki AG, Brunk B, Tajbakhsh S, Buckingham M, Chiang C, Emerson CP, Jr. Sonic hedgehog controls epaxial muscle determination through Myf5 activation. Development 1999 Sep;126(18):4053-4063. Borycki, A., A. M. Brown and C. P. Emerson Jr. 2000. Shh and Wnt signaling pathways converge to control Gli gene activation in avian somites. Development 127: 2075–2087. Bradley, J. P., Levine, J. P., Roth, D. A., McCarthy, J. G., Longaker, M. T. (1995). Studies in cranial suture biology: IV. temporal sequence of posterior frontal cranial suture fusion in the mouse. Plastic and Reconstructive Surgery, 98(6):1039-1045. Brand-Saberi, B., J. Wilting, C. Ebensperger and B. Christ. 1996. The formation of somite compartments in the avian embryo. Int. J. Dev. Biol. 40: 411–420. Brugmann, S. A., M. D. Tapadia and J. A. Helms. 2006. The molecular origins of species-specific facial pattern. Curr. Top. Dev. Biol. 73: 1–42. Buckingham, M., L. Bajard, P. Daubas, M. Esner, M. Lagha, F. Relaix and D. Rocancourt. 2006. Myogenic progenitor cells in the mouse embryo are marked by the expression of Pax3/7 genes that regulate their survivial and mygoenic potential. Anat. Embryol. 211: S51–S56. Cannon, B. & Nedergaard, J. Brown adipose tissue: function and physiological significance. Physiol. Rev. 84, 277–359 (2004). Chau YY, Bandiera R, Serrels A, Martínez-Estrada OM, Qing W, Lee M, Slight J, Thornburn A, Berry R, McHaffie S, Stimson RH, Walker BR, Chapuli RM, Schedl A, Hastie N. Visceral and subcutaneous fat have different origins and evidence supports a mesothelial source, Nat. Cell Biol. 16 (4) (2014) 367–375. Chen P, Carrington JL, Paralkar VM, Pierce GF, Reddi AH. (1992). Chick limb bud mesodermal cell chondrogenesis: inhibition by isoforms of platelet-derived growth factor and reversal by recombinant bone morphogenetic protein. Exp Cell Res, 200(1):110-7. !139Chen, G., Wang, X., Zhang, Y., Ru, X., Zhou, L., & Tian, Y. (2015). H3K9 histone methyltransferase G9a ameliorates dilated cardiomyopathy via the downregulation of cell adhesion molecules. Mol Med Rep, 11(5), 3872-9. Cheng CS, El-Abd Y, Bui K, Hyun YE, Hughes RH, Kraus WE, Truskey GA. Conditions that promote primary human skeletal myoblast culture and muscle differentiation in vitro. Am J Physiol Cell Physiol 2014 Feb 15;306(4):C385-95. Claesson-Welsh L, Eriksson A, Westermark B, Heldin CH. cDNA cloning and expression of the human A-type platelet-derived growth factor (PDGF) receptor establishes structural similarity to the B-type PDGF receptor. Proc Natl Acad Sci U S A. 1989 Jul;86(13):4917–4921. Claesson-Welsh, L. (1994). Platelet-derived growth factor receptor signals. J Biol Chem, 269(51):32023-6. Collins CA, Zammit PS. Isolation and grafting of single muscle fibres. Methods Mol Biol 2009;482:319-330. Cossu, G., R. Kelly, S. Tajbakhsh, S. Di Donna, E. Vivarelli and M. Buckingham. 1996. Activation of different myogenic pathways: myf-5 is induced by the neural tube and MyoD by the dorsal ectoderm in mouse paraxial mesoderm. Development 122: 429–437. Creuzet, S., B. Schuler, G. Couly and N. M. Le Duoarin. 2004. Reciprocal relationships between Fgf8 and neural crest cells in facial and forebrain development. Proc. Natl. Acad. Sci. USA 101: 4843–4847. Creuzet, S., G. Couly and N. M. Le Douarin. 2005. Patterning the neural crest derivatives during development of the vertebrate head: Insights from avian studies. J. Anat. 207: 447–459. Crossno, J. T. Jr., Majka SM, Grazia T, Gill RG, Klemm DJ. Rosiglitazone promotes development of a novel adipocyte population from bone marrow-derived circulating progenitor cells, J. Clin. Invest. 116 (12) (2006) 3220–3228. Damez-Werno DM, Sun H, Scobie KN, Shao N, Rabkin J, Dias C, Calipari ES, Maze I, Pena CJ, Walker DM, Cahill ME, Chandra R, Gancarz A, Mouzon E, Landry JA, Cates H, Lobo MK, Dietz D, Allis CD, Guccione E, Turecki G, Defilippi P, Neve RL, Hurd YL, Shen L, Nestler EJ. (2016). Histone arginine methylation in cocaine action in the nucleus accumbens. Proc Natl Acad Sci U S A. 113(34):9623-8. Das, A. and J. G. Crump. 2012. BMPs and id2a act upstream of Twist1 to restrict ectomesenchyme potential of cranial neural crest. PLoS Genet. 8: e1002710. David, J. D., W. M. See and C. A. Higginbotham. 1981. Fusion of chick embryo skeletal myoblasts: Role of calcium influx preceding membrane union. Dev. Biol. 82: 297–307. Deries, M., A. B. Gonçalves, R. Vaz, G. G. Martins, G. Rodrigues and S. Thorsteinsdóttir. 2012. Extracellular matrix remodeling accompanies axial muscle development and morphogenesis in the mouse. Dev. Dyn. 241: 350–364. Deries, M., R. Schweitzer and M. J. Duxson. 2010. Developmental fate of the mammalian myotome. Dev. Dyn. 239: 2898–2910. Dietrich, S., F. R. Schubert, C. Healy, P. T. Sharpe and A. Lumsden. 1998. Specification of hypaxial musculature. Development 125: 2235–2249. Digby, J. E. et al. Thiazolidinedione exposure increases the expression of uncoupling protein 1 in cultured human preadipocytes. Diabetes 47, 138–141 (1998). Doherty, K. R. and 7 others. 2005. Normal myoblast fusion requires myoferlin. Development 132: 5565–5575. !140Dong KB, Maksakova IA, Mohn F, Leung D, Appanah R, Lee S, Yang HW, Lam LL, Mager DL, Schubeler D, Tachibana M, Shinkai Y, Lorincz MC. DNA methylation in ES cells requires the lysine methyltransferase G9a but not its catalytic activity. EMBO J 2008 Oct 22;27(20):2691-2701. Eccleston, P. A., Funa, K., & Heldin, C. H. (1993). Expression of platelet-derived growth factor (PDGF) and PDGF alpha- and beta-receptors in the peripheral nervous system: an analysis of sciatic nerve and dorsal root ganglia. Dev Biol, 155(2):459-70. Epsztejn-Litman S, Feldman N, Abu-Remaileh M, Shufaro Y, Gerson A, Ueda J, Deplus R, Fuks F, Shinkai Y, Cedar H, Bergman Y. De novo DNA methylation promoted by G9a prevents reprogramming of embryonically silenced genes. Nat Struct Mol Biol 2008 Nov;15(11):1176-1183. Evans, D. J. R. and D. M. Noden. 2006. Spatial relations between avian craniofacial neural crest and paraxial mesoderm cells. Dev. Dyn. 235: 1310–1325. Feng W, Yonezawa M, Ye J, Jenuwein T, Grummt I. (2010). PHF8 activates transcription of rRNA genes through H3K4me3 binding and H3K9me1/2 demethylation. Nat Struct Mol Biol, 17(4), 445-50. Fish, E. W., Shahrokh, D., Bagot, R., Caldji, C., Bredy, T., Szyf, M. and Meaney, M. J. (2004), Epigenetic Programming of Stress Responses through Variations in Maternal Care. Annals of the New York Academy of Sciences, 1036: 167–180. Francetic, T., & Li, Q. (2011). Skeletal myogenesis and Myf5 activation. Transcription, 2(3), 109-114. Gerhart J, Hayes C, Scheinfeld V, Chernick M, Gilmour S, George-Weinstein M. 2011. Myo/Nog cell regulation of bone morphogenetic protein signaling in the blastocyst is essential for normal morphogenesis and striated muscle lineage specification. Dev. Biol. 359: 12–25. Gerhart, J., J. Elder, C. Neely, J. Schure, T. Kvist, K. Knudsen and M. George-Weinstein. 2006. MyoD-positive epiblast cells regulate skeletal muscle differentiation in the embryo. J. Cell Biol. 175: 283–292. Gilbert, S. F. (2013). Developmental Biology, 10th ed. Sinauer Associates, Sunderland, MA. Gopinath SD, Rando TA. (2008). Stem cell review series: aging of the skeletal muscle stem cell niche. Aging Cell, 7(4), 590-8. Govindarajan V, Ramachandran RK, George JM, Shakes DC, Tomlinson CR. (1995). An ECM-bound, PDGF-like growth factor and a TGF-alpha-like growth factor are required for gastrulation and spiculogenesis in the Lytechinus embryo. Dev Biol, 172(2):541-51. Grenier, J., M. A. Teillet, R. Grifone, R. G. Kelly and D. Duprez. 2009. Relationship between neural crest cells and cranial mesoderm during head muscle development. PLoS One 4: e4381. Gros, J., M. Manceau, V. Thomé, and C. Marcelle. 2005. A common somitic origin for embryonic progenitors and satellite cells. Nature 435: 954–958.  Gupta RK, Mepani RJ, Kleiner S, Lo JC, Khandekar MJ, Cohen P, Frontini A, Bhowmick DC, Ye L, Cinti S, Spiegelman BM. Zfp423 expression identifies committed preadipocytes and localizes to adipose endothelial and perivascular cells, Cell Metab. 15 (2) (2012) 230–239. Gustafsson MK, Pan H, Pinney DF, Liu Y, Lewandowski A, Epstein DJ, Emerson CP, Jr. Myf5 is a direct target of long-range Shh signaling and Gli regulation for muscle specification. Genes Dev 2002 Jan 1;16(1):114-126. !141Hall, B. K. 2009. The Neural Crest and Neural Crest Cells in Vertebrate Development and Evolution. Springer, New York. Han, P., Li, W., Yang, J., Shang, C., Lin, C. H., Cheng, W., Hang, C. T., Cheng, H. L., Chen, C. H., Wong, J., Xiong, Y., Zhao, M., Drakos, S. G., Ghetti, A., Li, D. Y., Bernstein, D., Chen, H. S., Quertermous, T., & Chang, C.P. (2016). Epigenetic response to environmental stress: Assembly of BRG1-G9a/GLP-DNMT3 repressive chromatin complex on Myh6 promoter in pathologically stressed hearts. Biochim Biophys Acta, 1863(7B), 1772-81. Hanai, K., Kato, H., Matsuhashi, S., Morita, H., Raines, E. W., & Ross, R. (1987). Platelet proteins, including platelet-derived growth factor, specifically depress a subset of the multiple components of the response elicited by glutathione in Hydra. J Cell Biol, 104(6):1675-81. Hart IK, Richardson WD, Heldin CH, Westermark B, Raff MC. (1989). PDGF receptors on cells of the oligodendrocyte-type-2 astrocyte (O-2A) cell lineage. Development, 105(3), 595-603. Hasty, P., Bradley, A., Morris, J. H., Edmondson, D. G., Venuti, J. M., Olson, E. N., et al. (1993). Muscle deficiency and neonatal death in mice with a targeted mutation in the myogenin gene. Nature, 364(6437), 501-506. Haworth, K. E., C. Healy, P. Morgan and P. T. Sharpe. 2004. Regionalisation of early head ectoderm is regulated by endoderm and prepatterns the orofacial epithelium. Development Heldin CH, Ernlund A, Rorsman C, Rönnstrand L. Dimerization of B-type platelet-derived growth factor receptors occurs after ligand binding and is closely associated with receptor kinase activation. J Biol Chem. 1989 May 25;264(15):8905–8912. Heller EA, Hamilton PJ, Burek DD, Lombroso SI, Peña CJ, Neve RL, Nestler EJ. (2016). Targeted Epigenetic Remodeling of the Cdk5 Gene in Nucleus Accumbens Regulates Cocaine- and Stress-Evoked Behavior. J Neurosci. 36(17):4690-7. Herz HM, Garruss A, Shilatifard A. SET for life: biochemical activities and biological functions of SET domain-containing proteins. Trends Biochem Sci 2013 Dec;38(12):621-639. Ho L, Symes K, Yordán C, Gudas LJ, Mercola M. (1994). Localization of PDGF A and PDGFR alpha mRNA in Xenopus embryos suggests signalling from neural ectoderm and pharyngeal endoderm to neural crest cells. Mech Dev. 1994 Dec;48(3):165-74. Ho, A. T. V., S. Hayashi, D. Bröhl, F. Auradé, R. Rattenbach and F. Relaix. 2011. Neural crest cell lineage restricts skeletal muscle progenitor cell differentiation through Neuregulin1-ErbB3 signaling. Dev. Cell 21: 273–287. Holleville, N., A. Quilhac, M. Bontoux, and A.-H. Monsoro-Burq. 2003. BMP signals regulate Dlx5 during early avian skull development. Dev. Biol. 257: 177–189.  Horsley, V., K. M. Jansen, S. T. Mills and G. K. Pavlath. 2003. IL-4 acts as a myoblast recruitment factor during mammalian muscle growth. Cell 113: 483–494.  Hu, D., R. S. Marcucio, and J. A. Helms. 2003. A zone of frontonasal ectoderm regulates patterning and growth in the face. Development 130: 1749–1758. Huang J, Berger SL. The emerging field of dynamic lysine methylation of non-histone proteins. Curr Opin Genet Dev 2008 Apr;18(2):152-158. Huang J, Dorsey J, Chuikov S, Perez-Burgos L, Zhang X, Jenuwein T, Reinberg D, Berger SL. G9a and Glp methylate lysine 373 in the tumor suppressor p53. J Biol Chem 2010 Mar 26;285(13):9636-9641. !142Inagawa M, Nakajima K, Makino T, Ogawa S, Kojima M, Ito S, Ikenishi A, Hayashi T, Schwartz RJ, Nakamura K, Obayashi T, Tachibana M, Shinkai Y, Maeda K, Miyagawa-Tomita S, Takeuchi T. Histone H3 lysine 9 methyltransferases, G9a and GLP are essential for cardiac morphogenesis. Mech Dev 2013 Nov-Dec;130(11-12):519-531. Jiang, X., S. Iseki, R. E. Maxson, H. M. Sucov and G. Morriss-Kay. 2002. Tissue origins and interactions in the mammalian skull vault. Dev. Biol. 241: 106–116. Joe AW, Yi L, Natarajan A, Le Grand F, So L, Wang J, Rudnicki MA, Rossi FM. Muscle injury activates resident fibro/adipogenic progenitors that facilitate myogenesis. Nat Cell Biol 2010 Feb;12(2):153-163. Jones SD, Ho L, Smith JC, Yordan C, Stiles CD, Mercola M. (1993). The Xenopus platelet-derived growth factor alpha receptor: cDNA cloning and demonstration that mesoderm induction establishes the lineage-specific pattern of ligand and receptor gene expression. Dev Genet, 14(3):185-93. Juan AH, Derfoul A, Feng X, Ryall JG, Dell'Orso S, Pasut A, Zare H, Simone JM, Rudnicki MA, Sartorelli V. (2011). Polycomb EZH2 controls self-renewal and safeguards the transcriptional identity of skeletal muscle stem cells. Genes Dev, 25(8), 789-94. Kanisicak O, Mendez JJ, Yamamoto S, Yamamoto M, Goldhamer DJ. Progenitors of skeletal muscle satellite cells express the muscle determination gene, MyoD. Dev Biol 2009 Aug 1;332(1):131-141. Kato, N. and H. Aoyama. 1998. Dermomyotomal origin of the ribs as revealed by extirpation and transplantation experiments in chick and quail embryos. Development 125: 3437–3443. Kaur, K., Yang, J., Edwards, J. G., Eisenberg, C. A., & Eisenberg, L. M. (2016). G9a histone methyltransferase inhibitor BIX01294 promotes expansion of adult cardiac progenitor cells without changing their phenotype or differentiation potential. Cell Prolif, 49(3), 373-385. Kavanaugh WM, Harsh GR 4th, Starksen NF, Rocco CM, Williams LT. (1988). Transcriptional regulation of the A and B chain genes of platelet-derived growth factor in microvascular endothelial cells. J Biol Chem, 263(17):8470-2. Kazlauskas, A. (1994). Receptor tyrosine kinases and their targets. Curr Opin Genet Dev, 4(1):5-14. Kelly JD, Haldeman BA, Grant FJ, Murray MJ, Seifert RA, Bowen-Pope DF, Cooper JA, Kazlauskas A. Platelet-derived growth factor (PDGF) stimulates PDGF receptor subunit dimerization and intersubunit trans-phosphorylation. J Biol Chem. 1991 May 15;266(14):8987–8992. Klose, R. J. & Zhang, Y. (2007). Regulation of histone methylation by demethylimination and demethylation. Nat Rev Mol Cell Biol., 8(4), 307-18 Knudsen, K. A. 1985. The calcium-dependent myoblast adhesion that precedes cell fusion is mediated by glycoproteins. J. Cell Biol. 101: 891–897. Knudsen, K. A., S. A. McElwee and L. Myers. 1990. A role for the neural cell adhesion molecule, NCAM, in myoblast interaction during myogenesis. Dev. Biol. 138: 159–168. Kohler, N. & Lipton, A. (1974). Platelets as a source of fibroblast growth-promoting activity. Exp Cell Res, 87(2):297-301. Komori T, Yagi H, Nomura S, Yamaguchi A, Sasaki K, Deguchi K, Shimizu Y, Bronson RT, Gao YH, Inada M, Sato M, Okamoto R, Kitamura Y, Yoshiki S, Kishimoto T. (1997). Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell, 89(5):755-64. !143Konigsberg, I. R. 1963. Clonal analysis of myogenesis. Science 140: 1273–1284. Kramer JM, Kochinke K, Oortveld MA, Marks H, Kramer D, de Jong EK, Asztalos Z, Westwood JT, Stunnenberg HG, Sokolowski MB, Keleman K, Zhou H, van Bokhoven H, Schenck A. Epigenetic regulation of learning and memory by Drosophila EHMT/G9a. PLoS Biol 2011 Jan 4;9(1):e1000569. Kramer JM. Regulation of cell differentiation and function by the euchromatin histone methyltranserfases G9a and GLP. Biochem Cell Biol 2016 Feb;94(1):26-32. Lanouette S, Mongeon V, Figeys D, Couture JF. The functional diversity of protein lysine methylation. Mol Syst Biol 2014 Apr 8;10:724. Lassar, A. B. (2009). The p38 MAPK family, a pushmi-pullyu of skeletal muscle differentiation. J Cell Biol, 187(7), 941-943. Le Lièvre, C. S. 1978. Participation of neural crest-derived cells in the genesis of the skull in birds. J. Embryol. Exp. Morphol. 47: 17–37. Lee DY, Northrop JP, Kuo MH, Stallcup MR. Histone H3 lysine 9 methyltransferase G9a is a transcriptional coactivator for nuclear receptors. J Biol Chem 2006 Mar 31;281(13):8476-8485. Lee, S. H., K. K. Fu, J. N. Hui, and J. M. Richman. 2001. Noggin and retinoic acid transform the identity of avian facial prominences. Nature 414: 909–912. Lee, S. J. 2004. Regulation of muscle mass by myostatin. Annu. Rev. Cell Dev. Biol. 20: 61–86. Lee, Y.-H., Petkova, A. P., Mottillo, E. P. & Granneman, J. G. In vivo identification of bipotential adipocyte progenitors recruited by β3-adrenoceptor activation and high-fat feeding. Cell Metab. 15, 480–491 (2012). Lehnertz B, Northrop JP, Antignano F, Burrows K, Hadidi S, Mullaly SC, Rossi FM, Zaph C. Activating and inhibitory functions for the histone lysine methyltransferase G9a in T helper cell differentiation and function. J Exp Med 2010 May 10;207(5):915-922. Lehnertz B, Pabst C, Su L, Miller M, Liu F, Yi L, Zhang RH, Krosl J, Yung E, Kirschner J, Rosten P, Underhill TM, Jin J, Hebert J, Sauvageau G, Humphries RK, Rossi FM. The methyltransferase G9a regulates HoxA9-dependent transcription in AML. Genes Dev 2014 Feb 15;28(4):317-327. Lehnertz B, Ueda Y, Derijck AA, Braunschweig U, Perez-Burgos L, Kubicek S, Chen T, Li E, Jenuwein T, Peters AH. Suv39h-mediated histone H3 lysine 9 methylation directs DNA methylation to major satellite repeats at pericentric heterochromatin. Lelliott, C. & Vidal-Puig, A. J. Lipotoxicity, an imbalance between lipogenesis de novo and fatty acid oxidation. Int. J. Obes. Relat. Metab. Disord. 28, S22–S28 (2004). Lepper C, Conway SJ, Fan CM. Adult satellite cells and embryonic muscle progenitors have distinct genetic requirements. Nature 2009 Jul 30;460(7255):627-631. Lepper, C. & Fan, C. Inducible lineage tracing of Pax7-descendant cells reveals embryonic origin of adult satellite cells. Genesis 48, 424–436 (2010) Lepper, C., Partridge, T. A., & Fan, CM. (2011). An absolute requirement for Pax7-positive satellite cells in acute injury-induced skeletal muscle regeneration. Development, 138(17), 3639-46. !144Liang, G., Lin, J. C., Wei, V., Yoo, C., Cheng, J. C., Nguyen, C. T., Weisenberger, D. J., Egger, G., Takai, D., Gonzales, F. A., & Jones, P. A. (2004). Distinct localization of histone H3 acetylation and H3-K4 methylation to the transcription start sites in the human genome. Proc Natl Acad Sci U S A. 2004 May 11;101(19):7357-62. Ling BM, Bharathy N, Chung TK, Kok WK, Li S, Tan YH, Rao VK, Gopinadhan S, Sartorelli V, Walsh MJ, Taneja R. Lysine methyltransferase G9a methylates the transcription factor MyoD and regulates skeletal muscle differentiation. Proc Natl Acad Sci U S A 2012 Jan 17;109(3):841-846. Ling BM, Gopinadhan S, Kok WK, Shankar SR, Gopal P, Bharathy N, Wang Y, Taneja R. G9a mediates Sharp-1-dependent inhibition of skeletal muscle differentiation. Mol Biol Cell 2012 Dec;23(24):4778-4785. Liu N, Zhang Z, Wu H, Jiang Y, Meng L, Xiong J, Zhao Z, Zhou X, Li J, Li H, Zheng Y, Chen S, Cai T, Gao S, Zhu B. Recognition of H3K9 methylation by GLP is required for efficient establishment of H3K9 methylation, rapid target gene repression, and mouse viability. Genes Dev 2015 Feb 15;29(4):379-393. Liu W, Tanasa B, Tyurina OV, Zhou TY, Gassmann R, Liu WT, Ohgi KA, Benner C, Garcia-Bassets I, Aggarwal AK, Desai A, Dorrestein PC, Glass CK, Rosenfeld MG. (2010). PHF8 mediates histone H4 lysine 20 demethylation events involved in cell cycle progression. Nature, 466(7305), 508-512. Loncar, D., Afzelius, B. A. & Cannon, B. Epididymal white adipose tissue after cold stress in rats. I. Nonmitochondrial changes. J. Ultrastruct. Mol. Struct. Res. 101, 109–122 (1988). Marcelle, C., M. R. Stark and M. Bronner-Fraser. 1997. Coordinate actions of BMPs, Wnts, Shh and Noggin mediate patterning of the dorsal somite. Development 124: 3955–3963. Maroto, M., R. A. Bone, and J. K. Dale. 2012. Somitogenesis. Development 139: 2453–2456. Matsui T, Heidaran M, Miki T, Popescu N, La Rochelle W, Kraus M, Pierce J, Aaronson S. Isolation of a novel receptor cDNA establishes the existence of two PDGF receptor genes. Science. 1989 Feb 10;243(4892):800–804. Matsui T, Leung D, Miyashita H, Maksakova IA, Miyachi H, Kimura H, Tachibana M, Lorincz MC, Shinkai Y. Proviral silencing in embryonic stem cells requires the histone methyltransferase ESET. Nature. 2010 Apr 8;464(7290):927-31. doi: 10.1038/nature08858. Epub 2010 Feb 17. Matthew R. McCann, Owen J. Tamplin, Janet Rossant, Cheryle A. Séguin. 2012. Tracing notochord-derived cells using a Noto-cre mouse: implications for intervertebral disc development. Disease Models and Mechanisms, 5: 73-82 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. Essential role of the histone methyltransferase G9a in cocaine-induced plasticity. Science 2010 Jan 8;327(5962):213-216. McArdle, B. H. and M. J. Anderson. 2001. Fitting multivariate models to community data: A comment on distance-based redundancy analysis. Ecology, 82: 290–297. McPherron, A. C., A. M. Lawler and S. J. Lee. 1997. Regulation of skeletal muscle mass in mice by a new TGF-beta superfamily member. Nature 387: 83–90. Menko, A. S. and D. Boettiger. 1987. Occupation of the extracellular matrix integrin is a control point for myogenic differentiation. Cell 51: 51–57. !145Mercola M, Wang CY, Kelly J, Brownlee C, Jackson-Grusby L, Stiles C, Bowen-Pope D. Selective expression of PDGF A and its receptor during early mouse embryogenesis. Dev Biol. 1990 Mar;138(1):114–122. Mercola, M., Melton, D. A., & Stiles, C. D. (1988). Platelet-derived growth factor A chain is maternally encoded in Xenopus embryos. Science, 241(4870):1223-5. Mintz, B. and W. W. Baker. 1967. Normal mammalian muscle differentiation and gene control of isocitrate dehydrogenase synthesis. Proc. Natl. Acad. Sci. USA 58: 592–598. Mis J, Ner SS, Grigliatti TA (2006) Identification of three histone methyltransferases in Drosophila: dG9a is a suppressor of PEV and is required for gene silencing. Mol Genet Genomics 275: 513–526. Morrison-Graham K, Schatteman GC, Bork T, Bowen-Pope DF, Weston JA. (1992). A PDGF receptor mutation in the mouse (Patch) perturbs the development of a non-neuronal subset of neural crest-derived cells. Development, 115(1):133-42. Mosavi LK, Cammett TJ, Desrosiers DC, Peng ZY. The ankyrin repeat as molecular architecture for protein recognition. Protein Sci 2004 Jun;13(6):1435-1448. Mudhar HS, Pollock RA, Wang C, Stiles CD, Richardson WD.(1993). PDGF and its receptors in the developing rodent retina and optic nerve. Development, 118(2):539-52. Mukherjee, S. (2016). Same but different. The New Yorker, May 2, 2016, 24-30. Münsterberg, A. E., J. Kitajewski, D. A. Bumcroft, A. P. McMahon and A. B. Lassar. 1995. Combinatorial signaling by sonic hedgehog and Wnt family members induce myogenic bHLH gene expression in the somite. Genes Dev. 9: 2911–2922. Murphy MM, Lawson JA, Mathew SJ, Hutcheson DA, Kardon G. Satellite cells, connective tissue fibroblasts and their interactions are crucial for muscle regeneration. Development 2011 Sep;138(17):3625-3637. Nabeshima, Y., Hanaoka, K., Hayasaka, M., Esumi, E., Li, S., & Nonaka, I. (1993). Myogenin gene disruption results in perinatal lethality because of severe muscle defect. Nature, 364(6437), 532-535. Noden, D. M. 1978. The control of avian cephalic neural crest cytodifferentiation. I. Skeletal and connective tissue. Dev. Biol. 67: 296–312. Northcutt, R. G. and C. Gans. 1983. The genesis of neural crest and epidermal placodes: A reinterpretation of vertebrate origins. Q. Rev. Biol. 58: 1–28. Oksanen, J. (2015). Multivariate Analysis of Ecological Communities in R: vegan tutorial. http://vegan.r-forge.r-project.org Olguin, H. C., Yang, Z., Tapscott, S. J., & Olwin, B. B. (2007). Reciprocal inhibition between Pax7 and muscle regulatory factors modulates myogenic cell fate determination. J Cell Biol, 177(5), 769-79. Ontell, M., D. Hughes and D. Bourke. 1988. Morphometric analysis of the developing mouse soleus muscle. Am. J. Anat. 181: 279–288. Ordahl, C. P. and N. Le Douarin. 1992. Two myogenic lineages within the developing somite. Development 114: 339–353. !146Orr-Urtreger A, Bedford MT, Do MS, Eisenbach L, Lonai P. (1992). Developmental expression of the alpha receptor for platelet-derived growth factor, which is deleted in the embryonic lethal Patch mutation. Development. 1992 May;115(1):289-303. Orr-Urtreger A, Lonai P. (1992). Platelet-derived growth factor-A and its receptor are expressed in separate, but adjacent cell layers of the mouse embryo. Development. 1992 Aug;115(4):1045-58. Osborn, D. P., K. Li, Y. Hinits and S. M. Hughes. 2011. Cdkn1c drives muscle differentiation through a positive feedback loop with Myod. Dev. Biol. 350: 464–475. Palmeirim, I., D. Henrique, D. Ish-Horowicz and O. Pourquié. 1997. Avian hairy gene expression identifies a molecular clock linked to vertebrate segmentation and somitogenesis. Cell 91: 639–648. Palmieri, S. L., Payne, J., Stiles, C. D., Biggers, J. D., & Mercola, M. (1992). Expression of mouse PDGF-A and PDGF alpha-receptor genes during pre- and post-implantation development: evidence for a developmental shift from an autocrine to a paracrine mode of action. Mech Dev, 39(3):181-91. Parker, M. H., Seale, P., & Rudnicki, M. A. (2003). Looking back to the embryo: defining transcriptional networks in adult myogenesis. Nat Rev Genet, 4(7), 497-507. Patapoutian, A., Yoon, J. K., Miner, J. H., Wang, S., Stark, K., & Wold, B. (1995). Disruption of the mouse MRF4 gene identifies multiple waves of myogenesis in the myotome. Development, 121(10), 3347-3358. Paylor, B., Natarajan, A., Zhang, RH., and Rossi, F. M. V. (2011). Nonmyogenic cells in skeletal muscle regeneration. In G. Pavlath (Ed.), Current Topics in Developmental Biology, 96:139-65. Peters, A.H.F.M., Kubicek, S., Mechtler, K., O’Sullivan, R.J., Derijck, A.A.H.A., Pérez-Burgos, L., et al. 2003. Partitioning and plasticity of repressive histone methylation states in mammalian chromatin. Mol. Cell, 12: 1577–1589. doi: 10.1016/S1097-2765(03)00477-5. PMID:14690609. Petrovic N, Walden TB, Shabalina IG, Timmons JA, Cannon B, Nedergaard J. (2010). Chronic peroxisome proliferator-activated receptor gamma (PPARgamma) activation of epididymally derived white adipocyte cultures reveals a population of thermogenically competent, UCP1-containing adipocytes molecularly distinct from classic brown adipocytes. J Biol Chem, 285(10):7153-64. Petrovic, N., Shabalina, I. G., Timmons, J. A., Cannon, B. & Nedergaard, J. Thermogenically competent nonadrenergic recruitment in brown preadipocytes by a PPARg agonist. Am. J. Physiol. Endocrinol. Metab. 295, E287–E296 (2008). Poissonnet, C. M., Burdi, A. R., & Garn, S. M. (1984). The chronology of adipose tissue appearance and distribution in the human fetus, Early Hum. Dev. 10 (1–2) (1984) 1–11. Polesskaya A, Duquet A, Naguibneva I, Weise C, Vervisch A, Bengal E, Hucho F, Robin P, Harel-Bellan A. CREB-binding protein/p300 activates MyoD by acetylation. J Biol Chem 2000 Nov 3;275(44):34359-34364. Potts, J. D. & Carrington, J. L. (1993). Selective expression of the chicken platelet-derived growth factor alpha (PDGF alpha) receptor during limb bud development. Dev Dyn, 198(1):14-21. Pourquié O, Fan CM, Coltey M, Hirsinger E, Watanabe Y, Bréant C, Francis-West P, Brickell P, Tessier-Lavigne M, Le Douarin NM. 1996. Lateral and axial signals involved in somite patterning: A role for BMP4. Cell 84: 461–471. !147Pringle, N. P. & Richardson, W. D. (1993). A singularity of PDGF alpha-receptor expression in the dorsoventral axis of the neural tube may define the origin of the oligodendrocyte lineage. Development, 117(2), 525-33. Purcell DJ, Jeong KW, Bittencourt D, Gerke DS, Stallcup MR. A distinct mechanism for coactivator versus corepressor function by histone methyltransferase G9a in transcriptional regulation. J Biol Chem 2011 Dec 9;286(49):41963-41971. Qi HH, Sarkissian M, Hu GQ, Wang Z, Bhattacharjee A, Gordon DB, Gonzales M, Lan F, Ongusaha PP, Huarte M, Yaghi NK, Lim H, Garcia BA, Brizuela L, Zhao K, Roberts TM, Shi Y. (2010). Histone H4K20/H3K9 demethylase PHF8 regulates zebrafish brain and craniofacial development. Nature, 466(7305):503-507. Raines EW, Ross R. Compartmentalization of PDGF on extracellular binding sites dependent on exon-6-encoded sequences. J Cell Biol. 1992 Jan;116(2):533–543. Ramachandran RK, Govindarajan V, Seid CA, Patil S, Tomlinson CR. (1995). Role for platelet-derived growth factor-like and epidermal growth factor-like signaling pathways in gastrulation and spiculogenesis in the Lytechinus sea urchin embryo. Dev Dyn, 204(1):77-88. Ramachandran RK, Seid CA, Lee H, Tomlinson CR. (1993). PDGF-BB and TGF-alpha rescue gastrulation, spiculogenesis, and LpS1 expression in collagen-disrupted embryos of the sea urchin genus Lytechinus. Mech Dev, 44(1):33-40. Rappolee, D. A., Brenner, C. A., Schultz, R., Mark, D., & Werb, Z. (1988). Developmental expression of PDGF, TGF-alpha, and TGF-beta genes in preimplantation mouse embryos. Science, 241(4874):1823-5. Rawls, A., Morris, J. H., Rudnicki, M., Braun, T., Arnold, H. H., Klein, W. H., et al. (1995). Myogenin's functions do not overlap with those of MyoD or Myf-5 during mouse embryogenesis. Dev Biol, 172(1), 37-50. Rea, S., Eisenhaber, F., O’Carroll, D., Strahl, B. D., Sun, Z. W., Schmid, M., Opravil, S., Mechtler, K., Ponting, C. P., Allis, C. D., and Jenuwein, T. (2000). Regulation of chromatin structure by site-specific histone H3 methyltransferases. Nature, 406(6796):593-599. Relaix, F., D. Rocancourt, A. Mansouri and. Buckingham. 2005. A Pax3/Pax7-dependent population of skeletal muscle progenitor cells. Nature 435: 948–953. Renneville A, Van Galen P, Canver MC, McConkey M, Krill-Burger JM, Dorfman DM, Holson EB, Bernstein BE, Orkin SH, Bauer DE, Ebert BL. EHMT1 and EHMT2 inhibition induces fetal hemoglobin expression. Blood 2015;126(16):1930-1939. Reshef, R., M. Maroto and A. B. Lassar. 1998. Regulation of dorsal somitic cell fates: BMPs and Noggin control the timing and pattern of myogenic regulator expression. Genes Dev. 12: 290–303. Rice, Judd C.; Briggs, Scott D.; Ueberheide, Beatrix; Barber, Cynthia M.; Shabanowitz, Jeffrey; Hunt, Donald F.; Shinkai, Yoichi; Allis, C.David (2003). "Histone Methyltransferases Direct Different Degrees of Methylation to Define Distinct Chromatin Domains". Molecular Cell. 12 (6): 1591–1598. Richardson, B. E., S. J. Nowak and M. K. Baylies. 2008. Myoblast fusion in fly and vertebrates: New genes, new processes and new perspectives. Traffic 9: 1050–1059. Rios, A. C., O. Serralbo, D. Salgado and C. Marcelle. 2011. Neural crest regulates myogenesis through the transient activation of Notch. Nature 473: 532–535. !148Rodeheffer, M. S., Birsoy, K. & Friedman, J. M. Identification of white adipocyte progenitor cells in vivo. Cell 135, 240–249 (2008). Rosen, E. D. & MacDougald, O. A. (2006). Adipocyte differentiation from the inside out, Nat. Rev. Mol. Cell Biol. 7 (12) (2006) 885–896. Ross R, Glomset J, Kariya B, Harker L. A platelet-dependent serum factor that stimulates the proliferation of arterial smooth muscle cells in vitro. Proc Natl Acad Sci U S A. 1974 Apr;71(4):1207–1210. Rudnicki, M. A., Schnegelsberg, P. N., Stead, R. H., Braun, T., Arnold, H. H., & Jaenisch, R. (1993). MyoD or Myf-5 is required for the formation of skeletal muscle. Cell, 75(7), 1351-1359. Ruthenburg, A. J., Allis, C. D., & Wysocka, J. (2007). Methylation of lysine 4 on histone H3: intricacy of writing and reading a single epigenetic mark. Mol Cell, 25(1), 15-30. Sampath SC, Marazzi I, Yap KL, Sampath SC, Krutchinsky AN, Mecklenbrauker I, Viale A, Rudensky E, Zhou MM, Chait BT, Tarakhovsky A. Methylation of a histone mimic within the histone methyltransferase G9a regulates protein complex assembly. Mol Cell 2007 Aug 17;27(4):596-608. Sanchez-Gurmaches, J. et al. PTEN loss in the Myf5 lineage redistributes body fat and reveals subsets of white adipocytes that arise from Myf5 precursors. Cell Metab. 16, 348–362 (2012). Sartorelli V, Juan AH. Sculpting chromatin beyond the double helix: epigenetic control of skeletal myogenesis. In: Pavlath GK, editor. Myogenesis. United States: Elsevier Inc; 2011. p. 57-83. Sartorelli V, Puri PL, Hamamori Y, Ogryzko V, Chung G, Nakatani Y, Wang JY, Kedes L. Acetylation of MyoD directed by PCAF is necessary for the execution of the muscle program. Mol Cell 1999 Nov;4(5):725-734. Sartorelli, V. & Caretti, G. (2005). Mechanisms underlying the transcriptional regulation of skeletal myogenesis. Curr Opin Genet Dev, 15(5), 528-35. Sato, Y., Yasuda., K., & Takahashi, Y. (2002). Morphological boundary forms by a novel inductive event mediated by Lunatic fringe and Notch during somitic segmentation. Development, 129(15), 3633-44. Schaefer A, Sampath SC, Intrator A, Min A, Gertler TS, Surmeier DJ, Tarakhovsky A, Greengard P. Control of cognition and adaptive behavior by the GLP/G9a epigenetic suppressor complex. Neuron 2009 Dec 10;64(5):678-691. Schatteman GC, Morrison-Graham K, van Koppen A, Weston JA, Bowen-Pope DF. (1992). Regulation and role of PDGF receptor alpha-subunit expression during embryogenesis. Development, 115(1):123-31. Scherer, P. E. Adipose tissue: from lipid storage compartment to endocrine organ. Diabetes 55, 1537–1545 (2006). Schoenebeck JJ, Hutchinson SA, Byers A, Beale HC, Carrington B, Faden DL, Rimbault M, Decker B, Kidd JM, Sood R, Boyko AR, Fondon JW III, Wayne RK, Bustamante CD, Ciruna B, Ostrander EA. (2012). Variation of BMP3 contributes to dog breed skull diversity. PLoS Genet, 8(8), e1002849. Seale P, Bjork B, Yang W, Kajimura S, Chin S, Kuang S, Scimè A, Devarakonda S, Conroe HM, Erdjument-Bromage H, Tempst P, Rudnicki MA, Beier DR, Spiegelman BM. (2008). PRDM16 controls a brown fat/skeletal muscle switch. Nature, 454, 961–967. !149Seale P, Sabourin LA, Girgis-Gabardo A, Mansouri A, Gruss P, Rudnicki MA. (2000). Pax7 is required for the specification of myogenic satellite cells. Cell, 102(6):777-86. Seifert RA, Hart CE, Phillips PE, Forstrom JW, Ross R, Murray MJ, Bowen-Pope DF. Two different subunits associate to create isoform-specific platelet-derived growth factor receptors. J Biol Chem. 1989 May 25;264(15):8771–8778. Shabalina IG, Petrovic N, de Jong JM, Kalinovich AV, Cannon B, Nedergaard J. UCP1 in brite/beige adipose tissue mitochondria is functionally thermogenic. Cell Rep. 5, 1196–1203 (2013). Shainberg, A., G. Yagil and D. Yaffe. 1969. Control of myogenesis in vitro by Ca2+ concentration in nutritional medium. Exp. Cell Res. 58: 163–167. Shan T, Liang X, Bi P, Zhang P, Liu W, Kuang S. Distinct populations of adipogenic and myogenic Myf5-lineage progenitors in white adipose tissues. J. Lipid Res. 54, 2214–2224 (2013). Sharp LZ, Shinoda K, Ohno H, Scheel DW, Tomoda E, Ruiz L, Hu H, Wang L, Pavlova Z, Gilsanz V, Kajimura S. Human BAT possesses molecular signatures that resemble beige/brite cells. PLoS ONE 7, e49452 (2012). Songyang Zhou, Steven E. Shoelson, Manas Chaudhuri, Gerald Gish, Tony Pawson, Wayne G. Haser, Fred King, Tom Roberts, Sheldon Ratnofsky, Robert J. Lechleider, Benjamin G. Neel, Raymond B. Birge, J.Eduardo Fajardo, Margaret M. Chou, Hidesaburo Hanafusa, Brian Schaffhausen, Lewis C. Cantley. (1993). SH2 domains recognize specific phosphopeptide sequences. Cell, 72(5):767-78. Spalding KL, Arner E, Westermark PO, Bernard S, Buchholz BA, Bergmann O, Blomqvist L, Hoffstedt J, Näslund E, Britton T, Concha H, Hassan M, Rydén M, Frisén J, Arner P. Dynamics of fat cell turnover in humans, Nature 453 (7196) (2008) 783–787. Srinivas S, Watanabe T, Lin CS, William CM, Tanabe Y, Jessell TM, Costantini F. Cre reporter strains produced by targeted insertion of EYFP and ECFP into the ROSA26 locus. BMC Dev Biol 2001;1:4. Stabell M, Eskeland R, Bjorkmo M, Larsson J, Aalen RB, et al. (2006) The Drosophila G9a gene encodes a multi-catalytic histone methyltransferase required for normal development. Nucleic Acids Res 34: 4609–4621. Stemple, D. L. 2005. Structure and function of the notochord: an essential organ for chordate development. Development, 132(11):2503-12. Stephanie E. Brown, R. Duncan Campbell, Christopher M. Sanderson. (2001). Novel NG36/G9a gene products encoded within the human and mouse MHC class III regions. Mammalian Genome 12, 916–924 (2001). Stern, H. M., A. M. C. Brown and S. D. Hauschka. 1995. Myogenesis in paraxial mesoderm: Preferential induction by dorsal neural tube and by cells expressing Wnt-1. Development 121: 3675–3686. Sunadome, K., T. Yamamoto, M. Ebisuya, K. Kondoh, A. Sehara-Fujisawa and E. Nishida. 2011. ERK5 regulates muscle cell fusion through Klf transcription factors. Dev. Cell. 20: 192–205. Tachibana M, Nozaki M, Takeda N, Shinkai Y. Functional dynamics of H3K9 methylation during meiotic prophase progression. EMBO J 2007 Jul 25;26(14):3346-3359. Tachibana M, Sugimoto K, Nozaki M, Ueda J, Ohta T, Ohki M, Fukuda M, Takeda N, Niida H, Kato H, Shinkai Y. G9a histone methyltransferase plays a dominant role in euchromatic histone H3 lysine 9 methylation and is essential for early embryogenesis. Genes Dev 2002 Jul 15;16(14):1779-1791. !150Tachibana M1, Sugimoto K, Fukushima T, Shinkai Y. (2001). Set domain-containing protein, G9a, is a novel lysine-preferring mammalian histone methyltransferase with hyperactivity and specific selectivity to lysines 9 and 27 of histone H3. J Biol Chem, 276(27):25309-17. Tajbakhsh S, Borello U, Vivarelli E, Kelly R, Papkoff J, Duprez D, Buckingham M, Cossu G. Differential activation of Myf5 and MyoD by different Wnts in explants of mouse paraxial mesoderm and the later activation of myogenesis in the absence of Myf5. Development 1998 Nov;125(21):4155-4162. Tang W, Zeve D, Suh JM, Bosnakovski D, Kyba M, Hammer RE, Tallquist MD, Graff JM. (2008). White fat progenitor cells reside in the adipose vasculature, Science, 322(5901):583-6. Tao, Y., Neppl, R.L., Huang, Z.-P., Chen, J., Tang, R.-H., Cao, R., Zhang, Y., Jin, S.-W., Wang, D.-Z., 2011. The histone methyltransferase Set7/9 promotes myoblast differentiation and myofibril assembly. J. Cell Biol. 194, 551–565. Tapscott, S. J. (2005). The circuitry of a master switch: Myod and the regulation of skeletal muscle gene transcription. Development, 132(12), 2685-95. Thomas LR, Miyashita H, Cobb RM, Pierce S, Tachibana M, Hobeika E, Reth M, Shinkai Y, Oltz EM. Functional analysis of histone methyltransferase g9a in B and T lymphocytes. J Immunol 2008 Jul 1;181(1):485-493. Thorogood, P. 1989. Review of developmental and evolutionary aspects of the neural crest. Trends Neurosci. 12: 38–39. Tickle C, Eichele G. (1994). Vertebrate limb development. Annu Rev Cell Biol, 10:121-52. Tierney MT, Aydogdu T, Sala D, Malecova B, Gatto S, Puri PL, Latella L, Sacco A. (2014). STAT3 signaling controls satellite cell expansion and skeletal muscle repair. Nature Medicine, 20(10): 1182-6. Timmons, J. A. et al. Myogenic gene expression signature establishes that brown and white adipocytes originate from distinct cell lineages. Proc. Natl Acad. Sci. USA 104, 4401–4406 (2007). Tomiyama K, Murase N, Stolz DB, Toyokawa H, O'Donnell DR, Smith DM, Dudas JR, Rubin JP, Marra KG. Characterization of transplanted green fluorescent protein + bone marrow cells into adipose tissue, Stem Cells 26 (2) (2008) 330–338. Tran KV, Gealekman O, Frontini A, Zingaretti MC, Morroni M, Giordano A, Smorlesi A, Perugini J, De Matteis R, Sbarbati A, Corvera S, Cinti S. The vascular endothelium of the adipose tissue gives rise to both white and brown fat cells, Cell Metab. 15 (2) (2012) 222–229. Trokovic, N., R. Trockovic, P. Mai and J. Partanen. 2003. Fgfr1 regulates patterning of the pharyngeal region. Genes Dev. 17: 141–153. Trokovic, N., R. Trokovic and J. Partinen. 2005. Fibroblast growth factor signalling and regional specification of the pharyngeal ectoderm. Int. J. Dev. Biol. 49: 797–805. Ullrich A, Schlessinger J. Signal transduction by receptors with tyrosine kinase activity. Cell. 1990 Apr 20;61(2):203–212. Venuti, J. M., Morris, J. H., Vivian, J. L., Olson, E. N., & Klein, W. H. (1995). Myogenin is required for late but not early aspects of myogenesis during mouse development. J Cell Biol, 128(4), 563-576. Vethantham V, Yang Y, Bowman C, Asp P, Lee JH, Skalnik DG, Dynlacht BD. Dynamic loss of H2B ubiquitylation without corresponding changes in H3K4 trimethylation during myogenic differentiation. Mol Cell Biol 2012 Mar;32(6):1044-1055.  !151Waddington, C.H. (1942). Endeavour 1, 18–20. Walden, T. B., Timmons, J. A., Keller, P., Nedergaard, J. & Cannon, B. Distinct expression of muscle-specific microRNAs (myomirs) in brown adipocytes. J. Cell. Physiol. 218, 444–449 (2009). Wang QA, Tao C, Gupta RK, Scherer PE. Tracking adipogenesis during white adipose tissue development, expansion and regeneration, Nat. Med. 19 (10) (2013) 1338–1344. Wang Y, Shankar SR, Kher D, Ling BM, Taneja R. Sumoylation of the basic helix-loop-helix transcription factor sharp-1 regulates recruitment of the histone methyltransferase G9a and function in myogenesis. J Biol Chem 2013 Jun 14;288(24):17654-17662. Wang, J. & Conboy, I. (2010). Embryonic vs. adult myogenesis: challenging the 'regeneration recapitulates development' paradigm. J Mol Cell Biol., 2(1), 1-4. Wood WM, Etemad S, Yamamoto M, Goldhamer DJ. MyoD-expressing progenitors are essential for skeletal myogenesis and satellite cell development. Dev Biol 2013 Dec 1;384(1):114-127. Wu, J. et al. Beige adipocytes are a distinct type of thermogenic fat cell in mouse and human. Cell 150, 366–376 (2012). Yaffe D, Saxel O. Serial passaging and differentiation of myogenic cells isolated from dystrophic mouse muscle. Nature 1977 Dec 22-29;270(5639):725-727. Yagami-Hiromasa, T., T. Sato, T. Kurisaki, K. Kamijo, Y.-I. Nabeshima and A. Fujisawa-Sehara. 1995. A metalloprotease-disintegrin participating in myoblast fusion. Nature 377: 652–656. Yamamizu K, Fujihara M, Tachibana M, Katayama S, Takahashi A, Hara E, Imai H, Shinkai Y, Yamashita JK. Protein kinase A determines timing of early differentiation through epigenetic regulation with G9a. Cell Stem Cell 2012 Jun 14;10(6):759-770. Yarden Y, Escobedo JA, Kuang WJ, Yang-Feng TL, Daniel TO, Tremble PM, Chen EY, Ando ME, Harkins RN, Francke U, et al. Structure of the receptor for platelet-derived growth factor helps define a family of closely related growth factor receptors. Nature. 1986 Sep 18;323(6085):226–232. Yeh HJ, Ruit KG, Wang YX, Parks WC, Snider WD, Deuel TF. (1991). PDGF A-chain gene is expressed by mammalian neurons during development and in maturity. Cell, 64(1):209-16. Yin, H. et al. MicroRNA-133 controls brown adipose determination in skeletal muscle satellite cells by targeting Prdm16. Cell Metab. 17, 210–224 (2013). Yoshida, T., P. Vivatbutsiri , G. Morriss-Kay, Y. Saga and S. Iseki. 2008. Cell lineage in mammalian craniofacial mesenchyme. Mech. Dev. 125: 797–808. Young, P., Arch, J. R. & Ashwell, M. Brown adipose tissue in the parametrial fat pad of the mouse. FEBS Lett. 167, 10–14 (1984). Zammit, P. S., Golding, J. P., Nagata, Y., Hudon, V., Partridge, T. A., & Beauchamp, J. R. (2004). Muscle satellite cells adopt divergent fates: a mechanism for self-renewal? J Cell Biol, 166(3), 347-357. Zhao, H., Feng, J., Ho TV., Grimes, W., Urata M., & Chai Y. (2015). The suture provides a niche for mesenchymal stem cells of craniofacial bones. Nature Cell Biology, 17(4):386-396. !152Zhu Z, Wang Y, Li X, Wang Y, Xu L, Wang X, Sun T, Dong X, Chen L, Mao H, Yu Y, Li J, Chen PA, Chen CD. (2010). PHF8 is a histone H3K9me2 demethylase regulating rRNA synthesis. Cell Res. 20 (7): 794-801. Zylicz JJ, Dietmann S, Günesdogan U, Hackett JA, Cougot D, Lee C, Surani MA. Chromatin dynamics and the role of G9a in gene regulation and enhancer silencing during early mouse development. eLife 2015 Nov 9;4. pii: e09571. doi: 10.7554/eLife.09571. !153

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0340781/manifest

Comment

Related Items