Florida State University Libraries

Electronic Theses, Treatises and Dissertations The Graduate School

2014 Characterization of Human MacroH2A Through Targeting Shawn C. Moseley

Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected] FLORIDA STATE UNIVERSITY

COLLEGE OF ARTS AND SCIENCES

CHARACTERIZATION OF HUMAN MACROH2A THROUGH GENE TARGETING

By

SHAWN C. MOSELEY

A Thesis submitted to the Department of Biological Science in partial fulfillment of the requirements for the degree of Master of Science

Degree Awarded: Spring Semester, 2014 Shawn C. Moseley defended this thesis on March 31, 2014. The members of the supervisory committee were:

Brian Chadwick Professor Directing Thesis

Jonathan Dennis Committee Member

Karen McGinnis Committee Member

The Graduate School has verified and approved the above-named committee members and certifies that the thesis has been approved in accordance with university requirements.

ii

ACKNOWLEDGMENTS

I would like to thank my advisor Dr. Brian Chadwick, as well as Dr. Jonathan Dennis and Dr. Karen McGinnis for their advice, assistance, and support.

I would also like to thank all of the members of the Chadwick lab, past and present, for their support.

iii

TABLE OF CONTENTS

List of Tables ...... vi List of Figures ...... vii Abstract ...... ix INTRODUCTION AND BACKGROUND ...... 1 GENERATING RESOURCES FOR GENOME ENGINEERING ...... 3 Introduction ...... 3 Generating Zinc Finger Nucleases (ZFNs) ...... 4 Generating Constructs ...... 8 Methods ...... 11 TARGETING MACROH2A1 AND MACROH2A2 ...... 13 Targeting MacroH2A1 ...... 13 Targeting MacroH2A2 ...... 18 Targeting MacroH2A2 with TALENs ...... 19 Retargeting MacroH2A1 ...... 22 Methods ...... 26 EXPRESSION ANALYSIS OF MACROH2A1 KNOCKOUTS ...... 30 Expression Analysis of Clone-8 ...... 30 Expression Analysis of Clone-21 ...... 35 Expression Analysis of Clone-64 ...... 45 Methods ...... 48 TRANSGENE RESCUE OF MACROH2A1...... 50 Generating Clone-21 Derived MacroH2A1 Rescues ...... 50 Generating Clone-64 Derived MacroH2A1 Rescues ...... 53 Methods ...... 53 CHROMATIN ANALYSIS IN MACROH2A1 MUTANTS ...... 57 Chromatin Immunoprecipitation of CCL2 ...... 57 Analysis of the Inactive X by Immunofluorescence and FISH ...... 58 Methods ...... 62 CONCLUSIONS AND DISCUSSION ...... 70

APPENDIX ...... 72

iv

A. PRIMERS ...... 72 REFERENCES ...... 75 BIOGRAPHICAL SKETCH ...... 79

v

LIST OF TABLES

1 of Interest – Set 1...... 32

2 Summary of Quantitative RT-PCR Trends for MacroH2A1 Knockouts Compared to RNAseq ...... 34

3 Genes of Interest – Set 2 ...... 34

4 Summary of Quantitative RT-PCR Trends for all MacroH2A1 Targets Compared to RNAseq ...... 37

5 Genes of Interest – Set 3 ...... 37

6 Summary of Quantitative RT-PCR Trends for MacroH2A1 Clone-21 Compared to Microarray...... 38

7 Genes of Interest – Set 4 ...... 39

8 Summary of RT-PCR Trends for Gene Set 4 ...... 39

9 Gene Expression Changes Comparing Different Passage Controls to Clone-21 ...... 41

10 Genes of Interest – Set 5 ...... 42

11 Summary of RT-PCR Trends for Gene Set 5 ...... 43

12 Genes of Interest – Set 6 ...... 44

vi

LIST OF FIGURES

1 MacroH2A ZFN Luciferase Assays ...... 6

2 MacroH2A2 ZFN T7EI Assay ...... 7

3 MacroH2A1 (Site 2) Homology Mediated Repair Assay ...... 8

4 Structure of the Synthetic Exon Promoter Trap Plasmid (pSEPT) ...... 9

5 Schematic Illustration of the MacroH2A1 Targeting Strategy ...... 10

6 MacroH2A1 Targeting Screen ...... 14

7 Sequence Analysis of NHEJ Induced Mutation of MacroH2A1 Exon-2 ...... 15

8 Anti-macroH2A1 Immunofluorescence ...... 16

9 Western Blot Analysis of MacroH2A1 Knockouts ...... 17

10 Quantitative RT-PCR Analysis of MacroH2A1 Knockouts ...... 17

11 Potential MacroH2A2 Exon-2 Start Sites ...... 19

12 MacroH2A2 TALEN FLASH Assembly ...... 20

13 MacroH2A2 TALEN Cut Assays...... 21

14 MacroH2A1 Retargeting Screen ...... 22

15 Quantitative RT-PCR Analysis of New MacroH2A1 Knockout ...... 23

16 Additional Western Blots of MacroH2A1 Knockouts ...... 24

17 Anti-macroH2A2 Immunofluorescence ...... 25

18 MacroH2A1 Sequence Alignment ...... 30

19 Scatter Plot of Gene Expression Changes in MacroH2A1 Knockout ...... 31

20 Quantitative RT-PCR Analysis on Genes of Interest – Set 1 ...... 33

21 Quantitative RT-PCR Analysis on Genes of Interest – Set 2 ...... 36

22 Quantitative RT-PCR Analysis on Genes of Interest – Set 3 ...... 38

vii

23 RT-PCR Analysis on Genes of Interest – Set 4 ...... 39

24 Venn Diagram Comparing Two Different RPE1 Controls ...... 40

25 RT-PCR Analysis on Genes of Interest – Set 5 ...... 43

26 Genes Which Have Decreased Expression in Clone-21 ...... 45

27 Comparison of Gene Expression Changes in Clone-21 and Clone-64 ...... 46

28 Comparison of Gene Expression Changes in Clones 8, 21, 64, and 70 ...... 47

29 Myc Rescue PCR Screening ...... 51

30 Myc Rescue Immunofluorescence Screening ...... 51

31 MacroH2A1 Isoform Expression in Each Myc Clone...... 52

32 MacroH2A1 Clone-21 Derived MYC Rescues ...... 54

33 MacroH2A1 Clone-64 Derived MYC Rescues ...... 55

34 Primer Pairs for Quantitative ChIP of CCL2 ...... 57

35 Quantitative ChIP Analysis of CCL2 with MacroH2A1 and MacroH2A2 ...... 59

36 Quantitative ChIP Analysis of CCL2 with H3K4me3 and H3K9me3 ...... 60

37 Anti-EZH2 Immunofluorescence ...... 63

38 Anti-H3K4me2 and Anti-HP1β Immunofluorescence ...... 63

39 Anti-H3K9me3 and Anti-H3K27me3 Immunofluorescence ...... 64

40 Anti-H3K27me3 and Anti-H1 Immunofluorescence ...... 64

41 Anti-SMCHD1 and Anti-HBix1 Immunofluorescence ...... 65

42 XIST RNA FISH and CEP-X DNA FISH ...... 66

43 PLS3 and Chr8 DNA FISH ...... 67

44 Model of H3K9me3 Spreading ...... 71

viii

ABSTRACT

Macro histone H2A (macroH2A) is a variant of core histone H2A that differs primarily by an extensive carboxy-terminal tail of unknown function that makes up two-thirds of the ’s mass. The histone variant is distributed throughout the nucleus, but in female mammalian cells, it has been found to be associated with the inactive X chromosome (Xi) in a local accumulation referred to as a macrochromatin body. The association of macroH2A with facultative heterochromatin of the Xi is suggestive of a role for the variant in gene silencing. MacroH2A1 was the first form of macroH2A discovered and is encoded by the H2AFY gene. Two splice isoforms exist due to two alternate versions of exon-6, giving rise to macroH2A1.1 and macroH2A1.2. A second form of macroH2A encoded by H2AFY2 gene, known as macroH2A2, shares 80% amino acid identity with macroH2A1 and also accumulates at Xi, suggesting the possibility of functional redundancy between the two . In order to further investigate macroH2A, we have generated knockouts of macroH2A1 and targeted a single allele of macroH2A2 in a human female telomerase immortalized retinal pigment epithelial cell line (RPE1). Targeted clones were generated by exchanging exon-2 of one or both alleles with a promoter-trap construct containing a promoterless neomycin selection cassette flanked by arms of homology designed to the intronic sequences immediately adjacent to exon-2. The selection cassette contains a splice-acceptor, internal ribosome entry site and polyadenylation signal, which when exchanged with exon-2 results in the inclusion of the neomycin coding sequence in the resulting truncated messenger RNA and its subsequent translation, allowing for correct targeting to be selected for through neomycin resistance. To enhance targeting, Zinc Finger Nucleases (ZFNs) were engineered to create a double strand break at or close to exon-2. In order to target the genes, cells were co-nucleofected with the targeting construct and ZFNs before seeding cells in media containing neomycin. Single cell clones were screened for correct targeting and loss of macroH2A assessed by Western blotting. In order to explore the impact of macroH2A1 loss, RNA was extracted from a macroH2A1 knockout as well as parental RPE1 and changes to the transcriptome assessed by massively paralleled sequencing of complementary DNA (RNAseq). Genes that showed a significant change in expression between the wildtype and knockout cells were selected for further study. Quantitative Chromatin Immunoprecipitation (qChIP) was performed on an affected gene to evaluate any local chromatin changes due to

ix macroH2A1 loss. Additionally, to examine the possibility that macroH2A1 splice isoforms fulfill different roles, full-length MYC-tagged expression constructs for macroH2A1.1 and macroH2A1.2 were reintroduced into cells and rescue of wild-type expression assessed for genes that displayed altered expression in response to macroH2A1 loss.

x

CHAPTER ONE

INTRODUCTION AND BACKGROUND

Histones are basic proteins that package DNA into structural units called nucleosomes, which in turn are the fundamental structural units of chromatin (Kornberg, 1974). Nucleosomes are composed of two molecules each of four core histones (H2A, H2B, H3 and H4). All histones contain a central structural motif called the histone fold domain that is flanked by amino and carboxy terminal tails (Arents et al., 1991). These tails can be covalently modified to affect the local chromatin environment, altering access to the underlying DNA (Zentner and Henikoff, 2013). In addition to covalent modification, variants of the core histones exist (Sarma and Reinberg, 2005). These variant histones can be substituted for their core counterparts to perform additional functions (Talbert and Henikoff, 2010).

In contrast to core histones, which are transcribed and assembled into nucleosomes during DNA replication in the synthesis phase of the cell cycle (Gunjan et al., 2005), transcription and incorporation of variants into chromatin is independent of the cell cycle. Evidence indicates that histone variants are essential for the epigenetic control of gene expression and other cellular responses (Talbert and Henikoff, 2010). To date, the most variants have been identified for H2A, including H2A.X (Mannironi et al., 1989), H2A.Z (Hatch and Bonner, 1988), H2A.Bbd (Chadwick and Willard, 2001b) and macroH2A (Chadwick and Willard, 2001a; Costanzi and Pehrson, 2001; Pehrson and Fried, 1992).

The first macroH2A variant discovered was macroH2A1 (Pehrson and Fried, 1992). MacroH2A1 is encoded by the H2AFY gene located on human chromosome 5q31.1 (henceforth referred to as macroH2A1 gene). At its amino terminus, macroH2A1 is collinear with and 65% identical to core H2A, but in addition, it possesses an extensive carboxy terminal tail that accounts for two- thirds of the protein's mass. The function of the carboxy terminal tail is unknown, although the tail contains the founding member of the macrodomain, a motif found in numerous proteins that are thought to regulate mono ADP-ribosylation (Feijs et al., 2013; Rosenthal et al., 2013). In addition, splice variant macroH2A1.1 contains a putative leucine zipper motif (Pehrson and Fried, 1992), a domain that is typically involved in protein-protein interactions (Landschulz et al., 1988) whereas splice variant macroH2A1.2 lacks this domain (Pehrson and Fried, 1992).

1

A second macroH2A variant was discovered that shares 80% amino acid identity with macroH2A1 (Chadwick and Willard, 2001a; Costanzi and Pehrson, 2001). This variant was classified as macroH2A2, and is expressed from a paralogous gene, H2AFY2, that is located on human chromosome 10q22.1 (henceforth referred to as the macroH2A2 gene).

All isoforms of macroH2A are enriched within the territory of the Xi in female cells (Chadwick and Willard, 2001a; Costanzi and Pehrson, 1998; Costanzi and Pehrson, 2001). The Xi is the result of X chromosome inactivation (XCI), the mammalian form of dosage compensation that serves to balance the levels of X-linked gene expression between the sexes (Lyon, 1961) by repackaging the chosen Xi into facultative heterochromatin, shutting down most gene expression (Carrel and Willard, 2005). Consequently, macroH2A is believed to be involved in maintaining gene silencing. Association of macroH2A1 with the Xi is dependent on the long non-coding RNA (lncRNA) X-inactive specific transcript (XIST), a central factor in XCI (Lee and Bartolomei, 2013), as conditional deletion of Xist in mouse, results in loss of macroH2A from the Xi (Csankovszki et al., 1999).

In addition to maintaining gene silencing, macroH2A has been shown to be directly involved in inhibiting transcription. This is done by inhibiting transcription factor binding and SWI/SNF- induced nucleosome sliding (Angelov et al., 2003). Other functions include blocking initiation of RNA polymerase II (RNAP II) transcription in vitro (Doyen et al., 2006).

MacroH2A has also been shown to block cell reprogramming. When macroH2A expression is reduced, the frequency in which cells can be reprogrammed to induced pluripotent stem cells (iPS cells) are increased (Gaspar-Maia et al., 2013).

Differential expression of macroH2A1 isoforms is linked to cancer prognosis. Under normal conditions macroH2A1.1 is found to be expressed mainly in differentiated, non-proliferating cells, whereas macroH2A1.2 is more generally expressed. Colon cancer cells, which have down- regulated macroH2A1.1, have shown to have poor patient prognosis, while higher levels of macroH2A1.1 predict survival (Sporn and Jung, 2012).

2

CHAPTER TWO

GENERATING RESOURCES FOR GENOME ENGINEERING

Introduction The past 5 years has seen major advances in the ability to generate knockouts of genes in human cells; a phenomenon that previously was largely restricted to mouse embryonic stem cells due to the high efficiency of homologous recombination in this cell type. This advance has been made possible through the ability to generate double-strand breaks at desired target sequences through the construction of custom nucleases that are directed to the sequence of choice. Double-strand breaks are repaired by one of two ways. First, the DNA ends can be blunted and re-ligated through non-homologous end-joining (NHEJ) (Lieber, 2010). This is an imprecise repair pathway that results in frequent insertions and deletions at the break site (Indels). Alternatively, the sister chromatid can be used as a template for repair through homologous recombination. This second method can be manipulated by providing an exogenous repair template that has homology to the sequences flanking the break. When used as a repair template, any desired sequence cloned between the regions of homology will also be introduced into the damage site through homology-mediated repair.

Two well-characterized methods to introduce specific double-strand breaks are based on the same basic principal, involving custom DNA binding proteins tethered at the carboxy terminus to an obligate heterodimer nuclease called FokI. In both strategies, a DNA binding protein is designed to one strand, with a second DNA binding protein designed to bind to the opposite DNA strand, downstream of the first. When the two proteins bind, the FokI nuclease is positioned such that it can dimerize and cut within the DNA sequence between the two DNA binding sites. The first approach makes use of zinc fingers, a common DNA binding motif that recognize specific 3 bp sequences (Carroll, 2011). Stitching together three zinc fingers results in a protein that recognizes a 9 bp sequence. Combined with the protein binding to the opposite strand, as well as the spacer sequence for nuclease dimerization, these proteins, or ZFN’s can recognize a specific 18 bp sequence. The second approach is based on transcription activator-like effector (TALE) proteins from the plant pathogen Xanthomonas spp. TALE proteins contain a DNA binding domain that consists of a 34 amino acid tandem repeat (Joung and Sander, 2013).

3

Each repeat recognizes a single . The specificity of recognizing a G, A, T of C is provided through two amino acids within the domain called the repeat variable domain (RVD). Therefore, stitching together a specific order of repeat domains with defined RVDs generates a highly specific DNA binding domain. Like the ZFN strategy, nuclease fused TALE proteins, or TALENs, are generated to recognize a 16-18 bp sequence on each strand, with a spacer of ~16 bp between the binding sites, resulting in specific recognition of a 48 bp binding site.

Generating Zinc Finger Nucleases (ZFNs) In order to create ZFNs for the macroH2A genes, potential binding sites needed to be identified. For both isoforms, potential binding sites were identified within exon-2 using Zinc Finger Targeter (ZiFiT) (Sander et al., 2010; Sander et al., 2007). This is an ideal location because exon-2 contains the start codon, and disruption this far upstream should prevent any truncated protein from being functional.

ZFNs were constructed using the Oligomerized Pool Engineering (OPEN) protocol (Maeder et al., 2008; Maeder et al., 2009) and several left and right ZFNs were isolated. The basic principal of this procedure is that a library is generated consisting of tens of thousands of zinc finger combinations that have the potential to bind your DNA sequence of interest fused to a transcriptional activator domain. These are introduced into a reporter strain that carries the DNA target sequence of interest for the left ZFN or right ZFN cloned upstream of a selectable marker. Bacteria are then seeded onto large agar plates containing a gradient of selection. Zinc finger combinations with the highest binding specificity are able to drive expression of the highest level of resistance and therefore can grow in the most stringent region of the plate. The bacterial colonies able to survive in the most stringent of selection were chosen for further analysis. Once the zinc finger plasmids were isolated, the binding ability of these ZFNs were further assessed using a beta-Galactosidase (β-Gal) assay; a bacterial 2-hybrid procedure that tests the ability of the selected zinc fingers fused to the yeast Gal11P domain to recruit RNA polymerase II fused to a Gal4 domain and activate beta-Galactosidase expression. Zinc fingers that perform well in both assays are then cloned into the FokI expression construct as mature ZFNs that can be assessed for their ability to bind and cut target DNA in vivo.

4

ZFNs which were able to bind effectively were then chosen to be tested in pairs by luciferase assay. This assay consisted of cloning the ZFN target site between the promoter and firefly luciferase reporter gene in the pGL3 vector from PROMEGA. The construct is then introduced into cells, with or without different combinations of left and right ZFN expression constructs as well as a Renilla luciferase expression construct. Changes to the levels of firefly luciferase were assayed, and normalized to Renilla activity to control for transfection efficiency. Each possible pair combination was tested. As shown below (Figure 1), macroH2A1 ZFN pairs showed little cutting activity, while macroH2A2 pairs looked more promising.

Pairs which most effectively reduced luciferase activity were transfected into the human embryonic kidney derived cell line 293T via lipofection and assayed for the ability to introduce a double strand break at exon-2. 293T cells were chosen for this assay because they can be transfected by lipofection at very high efficiency.

These lipofected cells were first tested for NHEJ using the T7 endonuclease I (T7EI) assay, which can detect heteroduplex formation. If a double strand break occurs, the cell will attempt to repair it by NHEJ. This usually results in the addition or removal of bases at the site of the break. DNA isolated from control and transfected cells is then used as a template for PCR. After the PCR, if cutting did occur, the resulting product will be a mixture of wild-type target DNA and DNA containing indels, the proportion of which will be related to the relative efficiency of the cutting process. The PCR product is then denatured and allowed to slowly anneal. Consequently, the re-annealed PCR products will include wild-type sequence hybridized to wild-type sequence, mutated sequence hybridized to mutated sequence and wild-type sequence hybridized to mutant sequence. The latter group will not be able to perfectly base pair due to the indels, and as such, a stretch of heteroduplex will be generated where no base pairing is occurring. This heteroduplex is recognized by T7EI, which will cut at the site and break the PCR product in two. The T7EI reactions are then separated on a high percentage agarose gel, and any cut PCR will migrate ahead of the non-cut DNA and the amount of cut DNA will reflect how efficient the ZFNs are at cutting. As can be seen in Figure 2, an additional DNA band can be seen migrating ahead of (below) the main PCR product in the sample derived from cells transfected with both the left and right ZFNs, but not either alone or the mock transfected sample.

5

MacroH2A1 Site 1 ZFN Luciferase Assay 1.2

1

0.8

0.6

0.4

0.2

0

MacroH2A2 Site 1 ZFN Luciferase Assay 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0

Figure 1. MacroH2A ZFN Luciferase Assays. Graphs indicating cutting ability of different ZFN pairs designed to cut either macroH2A1 (top) or macroH2A2 (bottom). Mock control transfections are highlighted in green. Negative controls using ZFN designed to unrelated target sites are shown in red. Each sample is normalized to luciferase activity detected for the pGL3 vector containing the macroH2A1 or macroH2A2 target sequence. Transfected DNA samples are indicated below each column. Blue columns show data for transfections with each left or right ZFN alone or different combinations of left and right. Reduced Luciferase activity correlates with higher cutting activity. Data represents results from the luciferase assay performed in triplicate.

6

Figure 2. MacroH2A2 ZFN T7EI Assay. Ethidium bromide stained agarose gel image showing results of the T7EI heteroduplex analysis assay. Samples include mock transfection and cells transfected with the left or right ZFN alone or both. The PCR product was denatured, allowed to slowly reanneal before incubation with (+) or without (-) T7EI endonuclease. Evidence of cutting is indicated by Indels immediately below the main product. The well of interest is indicated with a red arrow.

In addition to checking for NHEJ, ZFN pairs were also tested for the ability to introduce the neomycin selection cassette at the endogenous locus in 293T cells. This was achieved by transfecting ZFN pairs alongside the macroH2A targeting construct and then PCR screening to detect whether successful targeting of macroH2A1 or macroH2A2 has occurred. The principals of this assay are described in more detail below. Once a pair was identified that could successfully target the macroH2A1 or 2 gene in 293T, then the most efficient ZFN pair along with the targeting construct were introduced into RPE1 cells using a proprietary form of electroporation termed Nucleofection. Unlike 293T cells, lipofection efficiency in RPE1 cells is very low, necessitating the use of Nucleofection. RPE1 cells were chosen because, unlike 293T, and most other cancer derived cell line used in cell biology, RPE1 cells are diploid and essentially equivalent to primary cells with the exception that they overexpress human telomerase and therefore are immortal, allowing them to be single cell cloned. The most efficient set found for macroH2A2 was identified as Left Finger 2 and Right Finger 10. No efficient binding pairs were identified for macroH2A1.

7

Due to the lack of success with macroH2A1 ZFNs, a new target site (Site-2) was chosen, and the procedure was repeated. After performing similar assays (Data not shown, Figure 3), an effective site 2 ZFN pair was identified as Left Finger 10 and Right Finger 10.

Figure 3. MacroH2A1 (Site 2) Homology Mediated Repair Assay. Ethidium bromide stained agarose gel images showing PCR results to detect correct integration of the pSEPT-containing targeting construct at exon-2 of the macroH2A1 gene. Results are shown for PCR between a primer located upstream of the targeting construct to the pSEPT cassette (left image) and for a primer located downstream of the targeting construct to the pSEPT cassette (right image). Primers used and the expected product size are indicated to the left of each gel image. Samples tested are indicated at the top of the gel image. Samples include no-template control (Water), mock transfection (293T), DNA from a previously targeted clone (Clone C2), cells transfected with the targeting construct alone or targeting construct with various indicated left and right ZFN combinations.

Generating Constructs In addition to designing functional ZFNs, which can cause double strand breaks inside exon-2 of both macroH2A1 and macroH2A2, a targeting construct was also designed. The purpose of this construct is to provide a repair template which disrupts the target gene open reading frame and provides a means to select and enrich for correctly targeted cells. The constructs were made using the synthetic exon promoter trap plasmid (pSEPT) (Topaloglu et al., 2005) in conjunction with arms of homology to either macroH2A1 or macroH2A2 cloned into the left and right multiple cloning sites (Figure 4). The pSEPT vector contains a promoterless neomycin cassette which is used for selection. The benefit of using a promoterless cassette is that it can only be 8 transcribed and translated when inserted in the correct orientation downstream of an active promoter, reducing background caused by random integration events. Flanking the cassette are lox-P site-specific recombination recognition sites that allow for easy removal of the cassette and neomycin gene if desired. Removal of this cassette allows Neomycin to be used again in future targeting. The vector also contains an internal ribosome entry sequence (IRES) that is recognized by scanning ribosomes and permits efficient translation of the neomycin coding sequence. Upstream of the IRES is a splice acceptor sequence that will be spliced to the splice donor site of upstream exons, in this case that of exon-1 for both macroH2A1 and macroH2A2. Finally, the construct contains a polyadenylation sequence immediately downstream of the neomycin cassette that signals for transcriptional termination.

Figure 4. Structure of the Synthetic Exon Promoter Trap Plasmid (pSEPT). Schematic illustration of the pSEPT vector that was used in the macroH2A1 and macroH2A2 targeting constructs. Arms of homology to either the macroH2A1 gene or macroH2A2 gene were cloned into the multiple cloning sites indicated by the vertical lines at either edge of the map. Blue arrows indicate key features of the cassette. The downward kink in the line represents the location of the intron sequence.

In order to screen for correctly targeted clones, PCR primers were designed immediately upstream of the left homology arm and downstream of the right homology arm (Figure 5). These primers are used in combination with primers in the pSEPT vector. PCR products can only be generated if the pSEPT vector has exchanged with exon-2. Additionally, a third set of primers

9 was designed within the left and right homology arms, outside of the loxP sites. These were named the flox primers because they can be used to determine whether the construct has been successfully removed when cells are treated with Cre recombinase. The primers also serve a second purpose. In wild type cells, the primers will generate a small product of approximately 300bp. In clones that have had both alleles of macroH2A1/2 targeted, the template will no longer have the 300bp region and as such the product will include the full pSEPT sequence giving a large band. In those cells in which only one allele has been targeted, the resultant PCR product can be sequenced to see if the remaining allele has been disrupted by NHEJ.

Figure 5. Schematic Illustration of the MacroH2A1 Targeting Strategy. Top section shows the genomic exon-2 locus for macroH2A1 flanked by introns 1 and 2. Above this is the targeting construct consisting of left and right homology arms flanking the pSEPT cassette (red box). The lightning strike indicates the site of a double strand break generated by the ZFN. Repair of the damage using the targeting construct introduces the pSEPT cassette into the genomic locus, replacing exon-2. Red arrows indicate the location of primers used to screen for correct targeting, with one set located in pSEPT and the other set located up or downstream of the targeting construct homology arms. Blue arrows indicate the location of primers used for Cre-Lox excision confirmation and screening for double-targeted clones or NHEJ mutations.

10

Methods Zinc Finger Nucleases Zinc Finger Nucleases (ZFNs) were created following the Oligomerized Pool Engineering (OPEN) Protocol designed by the Joung lab (Maeder et al., 2009). Potential Zinc Finger binding sites were identified by the Zinc Finger Consortium’s ZiFiT Targeter software package (Sander et al., 2010; Sander et al., 2007). The genomic DNA sequences for both H2AFY and H2AFY2 genes were entered into the software and potential binding sites were located. Suitable sites were identified within exon-2 of macroH2A1 and immediately downstream of exon-2 in macroH2A2.

Beta-Galactosidase Assay The β-Gal assay was performed essentially as described in the Oligomerized Pool Engineering (OPEN) Protocol (Maeder et al., 2009).

Luciferase Assay In order to test the engineered nucleases’ ability to cut by luciferase assay, each of the ZFN half- sites were cloned into the Promega pGL3-Promotor Vector and DNA purified using a Macherey- Nagel Midiprep kit. 500ng of the pGL3 construct, 250ng of each of the ZFN pair combinations, and 20ng of a Renilla construct were transfected into 293T cells using Life Technologies Lipofectamine 2000 Reagent. Lipofections were performed in triplicate by mixing 4µl of Lipofectamine 2000 with the above constructs and incubating for 20 minutes in 200µl of serum free media. Each 200µl mix was transferred to 24-well plates containing 5x104 293T cells in 500µl DMEM media per well. The samples were then incubated at 37°C overnight.

The following day, lipofected cells were washed with 250µl of phosphate buffered saline (PBS) and lysed for 15min in 100µl of Promega Passive Lysis Buffer (PLB). The GloMax 20/20 luminometer was primed with Promega Luciferase Assay Reagent II (LAR II) and Stop & Glo® Reagent. After cell lysis was complete, 20µl of lysate was transferred to a 1.5ml tube and the luminescence of each transfection was measured after the luminometer added 100µl LAR II. Once measured, 100µl Stop & Glo® was administered which stopped the previous reaction and measured the amount of renilla present. This allowed for normalization between each sample.

11

T7 Endonuclease I Assay In order to perform the T7EI assay, ZFN pairs were transfected into 293T cells as described above, except 800ng of each ZFN was used. The following day, each transfection in the 24 well plate was transferred to the larger wells of a 6 well plate. These cells were permitted to grow for an additional day and then a genomic DNA isolation was performed using a Macherey-Nagel Genomic DNA isolation kit.

After isolating DNA, a 40 cycle PCR was performed with either the macroH2A1 or macroH2A2 Flox primers to amplify across the target region. The PCR product was cleaned with a Macherey- Nagel PCR cleanup kit and eluted in water. Half of the elution was used as the control and the other half was cut with the T7EI enzyme for 30 minutes at 37°C. After incubation, the DNA was ethanol precipitated at -80°C for 20 minutes and re-eluted in water. The elution was run on a 2% agarose gel at 90 volts for 45 minutes.

12

CHAPTER THREE

TARGETING MACROH2A1 AND MACROH2A2

Targeting MacroH2A1 After identifying ZFN pairs which effectively cut at both the macroH2A1 and macroH2A2 genes, targeting was attempted in a human female telomerase immortalized retinal pigment epithelial cell line (RPE1). RPE1 cells were selected as the target cell line of choice because telomerase immortalized cells are karyotypically, morphologically and phenotypically similar to the parental primary cells with the added advantage of unlimited cell division potential (Bodnar et al., 1998). The latter feature is essential for single cell cloning which is required for gene targeting experiments. Like most primary cells, the efficiency of transfection of RPE1 by lipid or calcium based methods is very low. Therefore, in order to introduce our ZFN and targeting constructs into as many cells as possible, we turned to Nucleofection; a proprietary form of electroporation.

Post Nucleofection, targeted cells were selected for by exposing cells to Neomycin over a period of 2-3 weeks, and surviving colonies were screened by PCR. Several outcomes arise from the targeting screen. Some clones, despite being neomycin resistant, are not targeted at the gene of interest, reflecting random integration of the targeting construct into a transcriptionally active region of the genome. Some proportion of clones will be positive for either the left or right targeting screen only and reflect clones that have undergone a more extensive deletion or rearrangement. These clones are ignored. The majority of positives reflect clones in which one allele has been successfully replaced with the neomycin construct and the second allele remains intact. These are heterozygous targeted clones that give positive PCR products for the left and right screens as well as the flox PCR. Some clones that are targeted at one allele give the appearance of being heterozygous targeted, but can be an actual knockout due to the ZFN disrupting the non-replaced allele by NHEJ. The NHEJ repair is imperfect and results in Indels at the cut site, leading to the insertion or loss of DNA. If large enough, the mutation on the second allele can be detected by a noticeable shift in size for the flox PCR, or for smaller Indels these can be detected by sequencing of the PCR product. The final type of clones isolated are those in which both alleles are replaced by the neomycin construct. These clones are positive for the left

13 and right targeting screen, but negative for the flox PCR. Figure 6 below is an example of a screening PCR with two double targeted clones indicated.

Figure 6. MacroH2A1 Targeting Screen. Ethidium bromide stained agarose gel showing a representative example of a colony screen. Primers used for the left screening PCR (top panel) and right screening PCR (middle panel) are indicated to the left of each image. The Flox PCR is indicated at the bottom. Samples include various clones, a positive PCR control (Clone-C2), and negative controls for the parental cell genomic DNA (RPE1) and no template control (Water). The results for Clone-8 (one allele targeted by pSEPT and the second disrupted by a deletion induced by NHEJ) and Clone-21 (both alleles replaced by pSEPT) are indicated at the bottom.

One clone, named Clone-21, was found to have both alleles targeted. This was confirmed using the flox PCR screen across Exon-2. This clone was missing the small product, which means both

14 alleles have been targeted. While this was the only clone with both alleles targeted, two other clones, named Clone-8 (above) and Clone-70 (not shown), displayed flox PCR products of unusual size. The PCR products of these two clones were sequenced, and the nature of the mutation generated by NHEJ determined. The NHEJ disrupted allele of Clone-8 is characterized by an 86bp deletion and 30bp insertion, resulting in a shift to the reading frame by one base. Clone-70 shows a 77bp insertion and introduces a new in-frame TAA stop codon almost immediately at the ZFN site (Figure 7)

Figure 7. Sequence Analysis of NHEJ Induced Mutation of MacroH2A1 Exon-2. Image shows the results of sequence analysis of macroH2A1 Exon-2 alleles disrupted by NHEJ for Clone-8 (top) and Clone-70 (bottom). Features for each clone relative to the wild-type exon-2 sequence are indicated.

Loss of macroH2A1 was confirmed in clones 8, 21 and 70 by immunostaining and western blotting (Figures 8, 9). Quantitative reverse transcription PCR (qRT-PCR) indicates that macroH2A1 transcript can still be detected in these clones (Figure 10). This is not unexpected as active transcription of the gene is necessary for production of the neomycin resistance gene product, and at least for clones 8 and 70, transcription through the allele disrupted by NHEJ is unimpeded (Figure 7).

15

Figure 8. Anti-macroH2A1 Immunofluorescence. (A) Indirect immunofluorescence (IF) showing the distribution of macroH2A1 (red) in parental RPE1 cells, Clone-8, Clone-70 and a subclone of Clone-8 in which the neomycin cassette was removed by Cre recombinase (C8 –Neo Sc5). (B) Indirect IF showing the distribution of macroH2A1 in parental RPE1 (green) and Clone-21. DNA is counterstained with DAPI (blue).

16

Figure 9. Western Blot Analysis of MacroH2A1 Knockouts. A western blot (left) was performed on the knockouts using anti-macroH2A1. As expected Clone-21 with both alleles targeted shows no macroH2A1 protein. In addition, both Clone-8 and Clone-70, which have one of their alleles disrupted by NHEJ, also do not make detectable amount of protein. A heterozygous targeted clone (Clone-3) appears to have approximately half the amount of macroH2A1 protein compared to RPE1 control. The band indicating macroH2A1 protein is indicated with a red arrow. A Coomassie loading control is shown on the right.

Figure 10. Quantitative RT-PCR Analysis of MacroH2A1 Knockouts. Messenger RNA expression levels for both macroH2A1 (left) and macroH2A2 (right) were accessed in the macroH2A1 knockouts using quantitative RT-PCR. A heterozygous targeted clone (Clone-3) appears to have approximately half the amount of macroH2A1 expression compared to RPE1 control. MacroH2A1 expression is nearly undetectable in double targeting Clone-21 and reduced in Clone-8 and Clone-70, indicating some read- through of the NHEJ allele. However, the NHEJ alleles are destabilized compared to the single targeted heterozygous Clone-3. Levels of macroH2A2 appear comparable in Clone-3 and Clone-8, but are approximately 2 fold increased in Clone-21 and 2 fold decreased in Clone-70.

17

Targeting MacroH2A2 Targeting of the macroH2A2 gene was performed using a similar strategy. However, no suitable ZFN sites were identified within exon-2, but instead were limited to intron-2 54 bp distal of the exon. This added a challenge to the targeting. For macroH2A1, disruptions of an allele by NHEJ directly disrupted the exon sequence and reading frame. However, for macroH2A2, a large NHEJ-induced deletion at the ZFN site is required to disrupt the exon. Consequently, while many clones showed successful introduction of the neomycin cassette at one allele, Indels generated at the second allele were not sufficiently large enough to disrupt the reading frame. Therefore, targeting was dependent upon replacing both alleles with the neomycin cassette. No double-targeted clones of this type were obtained for the first targeting attempt.

Further targeting efforts were performed in hopes of obtaining a double target. However, each of these attempts were unsuccessful. Therefore, alternate strategies were devised.

For the first approach, a new targeting construct was generated that replaced the neomycin cassette with either the blasticidin or zeocin resistance genes. These were used in combination with the ZFN pair on either parental RPE1 cells or on a clone that were already targeted at one allele by the neomycin cassette (Clone-H5). No RPE1 or Clone-H5 colonies were obtained with the zeocin construct. In contrast, successfully targeted blasticidin clones were obtained from parental RPE1, but not from Clone-H5.

Targeting using the neomycin cassette was more successful than the blasticidin cassette. Therefore, the second targeted strategy involved first removing the neomycin cassette from Clone-H5 to make the cells neomycin sensitive, followed by retargeting. This was possible because the pSEPT construct contains LoxP site-specific recombination recognition sequences flanking the neomycin gene. Clone-H5 was treated with Cre recombinase expressing Adenovirus to excise the DNA between the LoxP sites. Single cell clones were isolated and tested by PCR for excision of the neomycin gene and sensitivity to neomycin. A clone lacking the neomycin cassette (Clone-H5 -Neo) was used for retargeting using the original targeting construct and ZFNs. Despite the success targeting the first allele with neomycin, no clones were isolated from Clone-H5 –Neo after several targeting attempts.

18

Targeting MacroH2A2 with TALENs Since developing ZFNs to target macroH2A2, the use of TALENs in genome engineering was reported (Cermak et al., 2011). TALENs are more versatile than ZFNs due to fewer restrictions in the choice of target site. As such, target sites within exon-2 of the macroH2A2 were available and could be used for targeting with TALENs.

Numerous TALEN target sites were available in exon-2 that could be used with the existing neomycin targeting construct. In addition to the start codon used for full-length macroH2A2, there are three additional in-frame ATG sequences located downstream (Figure 11). These other start codons could potentially be used if NHEJ disrupts the standard ATG. As such, the TALENs were designed to cut within the last of these ATG sequences to avoid any possibility of functional truncated protein being generated from an alternate translational start site.

Figure 11. Potential MacroH2A2 Exon-2 Start Sites. The macroH2A2 Exon-2 sequence, highlighting in red, the original ATG start site, along with in-frame downstream start sites. If the first start site is disrupted by NHEJ, the remaining three could potentially be used. Therefore, TALENs were designed to cut the last ATG to improve the likelihood of NHEJ creating a knockout.

The above sequence was put into the ZiFiT Targeter software, which in addition to identifying ZFN cut sites, can also identify potential TALEN cut sites (Reyon et al., 2012). The software identified two potential TALEN sets that will cut in the region specified (Figure 12). The TALEN FLASH modules were cloned in the order shown below before cloning into the final plasmid (either JDS71 or JDS74).

19

Figure 12. MacroH2A2 TALEN FLASH Assembly. The assembly and sequence of each TALEN pair are indicated. Each FLASH module needed is shown at the top and are cloned together in the order shown before cloning into the final vector (bottom).

In order to test the activity of the macroH2A2 TALENs, we took advantage of the presence of a restriction endonuclease recognition sequence within the TALEN cutting site. In this assay, TALEN pairs that can cut their target sequence will, through NHEJ, disrupt the enzyme recognition sequence. Genomic DNA is isolated from cells transfected with the either the TALEN pair or mock transfection, and the region containing the TALEN binding sites is amplified by PCR. The PCR product is then subjected to digestion with the corresponding restriction enzyme. In the control sample, the PCR product is cut into two fragments, whereas in the TALEN exposed sample, any PCR for which the restriction site has been disrupted will fail to cut, resulting in some full-length PCR product post-digestion. The activity of the TALENs combined with the transfection efficiency is reflected by the amount of uncut PCR product. Figure 13A suggests that perhaps the macroH2A2 TALENs have some activity.

An additional assay that is used to detect NHEJ is the Surveyor™ assay (Qiu et al., 2004), that takes advantage of the mismatch repair activity of the CEL nuclease. As with the T7EI assay, the Surveyor™ assay recognizes and cuts heteroduplex DNA. This assay was performed on genomic

20

DNA isolated from TALEN transfected and mock transfected cells. However, there was no noticeable difference between the mock transfection and the TALEN transfections (Figure 13B). These data suggest that the TALEN pairs cannot efficiently cause a double strand break at the desired target site.

A B

Figure 13. MacroH2A2 TALEN Cut Assays. (A) Ethidium bromide stained agarose gel showing MacroH2A2 TALEN Restriction Digest Assay. In this assay, NHEJ is detected by loss of the restriction enzyme site. (B) Ethidium bromide stained agarose gel showing Surveyor™ Assay. In this assay, NHEJ is detected by the cutting of heteroduplexes formed when mismatches caused by NHEJ occur. No additional bands are present when compared with the 293T mock transfection, indicating that the TALENs are not active.

Nevertheless, given the potential for some activity based on the restriction enzyme assay, the different TALEN pairs and the same targeting construct used in the ZFN targeting experiments were introduced into RPE1 cells by Nucleofection. Cells were exposed to Neomycin, and despite the appearance of a small number of colonies, none appeared to be successfully targeted (data not shown). Therefore, it was concluded that the TALENs were not active enough for targeting.

21

Retargeting MacroH2A1 To validate and replicate observations made in the macroH2A1 knockout, Clone-21 (See Chapter 4), we decided to attempt to isolate an additional independent double –targeted clone. RPE1 cells were once again targeted as before. Screening identified a single clone (Clone-64) that was positive for the left and right targeting PCRs, but negative for the flox PCR, suggesting that, like Clone-21, this new clone had both copies of exon-2 exchanged with the neomycin cassette (Figure 14). Quantitative RT-PCR showed that, like Clone-21, macroH2A1 mRNA levels in Clone-64 were very low compared to parental RPE1 cells (Figure 15). In addition, we also screened the clone for macroH2A2 levels. Given how similar the macroH2A1 and macroH2A2 proteins are (Chadwick and Willard, 2001b; Costanzi and Pehrson, 2001), it is possible that the loss of one gene is compensated for by the other. Intriguingly, Clone-64 appeared not to be expressing macroH2A2 (Figure 15). Therefore, Clone-64 appears not to generate any macroH2A protein and is equivalent to a macroH2A1 and macroH2A2 knockout that we had attempted to generate experimentally.

Figure 14. MacroH2A1 Retargeting Screen. Ethidium bromide stained agarose gel images showing the results of an additional macroH2A1 targeting attempt. The Flox PCR shown on the left indicates that Clone-64 has lost Exon-2. This is indicated by the loss of the smaller band. In order to confirm proper targeting, the Left and Right Homology Arm PCRs were also performed confirming that like Clone-21, Clone-64 has both alleles properly targeted.

22

One possible explanation for this is that immunofluorescent staining of parental RPE1 cells for macroH2A2 revealed that some cells express macroH2A2 and others do not, whereas all cells express macroH2A1 (data not shown). Therefore, Clone-64 likely originates from targeting of a macroH2A2 negative RPE1 cell.

Figure 15. Quantitative RT-PCR Analysis of New MacroH2A1 Knockout. Messenger RNA expression levels for both macroH2A1 isoforms, as well as macroH2A2, were accessed in Clone-21 and new Clone-64 using quantitative RT-PCR. In addition to the reduced levels of macroH2A1 expression in Clone-64, macroH2A2 levels were essentially undetectable. This is in stark contrast compared to macroH2A1 knockout Clone-21.

Western blotting confirmed that Clone-64 is a genuine macroH2A1 knockout (Figure 16) and that, in agreement with the qRT-PCR data, this clone is not generating any macroH2A2 protein. The relatively low levels of macroH2A2 in the parental RPE1 cells more likely represent variable levels of macroH2A2 between cells, and not an overall low level of expression in all cells. In order to confirm this supposition, immunofluorescence (IF) to macroH2A2 was performed on RPE1, Clone-21 and Clone-64 (Figure 17). These data verified that Clone-64 does not express macroH2A2 and that the majority of RPE1 cells (>90%) do not express macroH2A2.

However, qRT-PCR data for RPE1 obtained almost 2 years earlier showed high levels of macroH2A2 expression (Figure 10) and immunofluorescence showed a higher proportion of cells expressing macroH2A2 (data not shown). This population of RPE1 was what was used for 23 the first successful round of macroH2A2 targeting, yet macroH2A2 targeting attempts in the interim were much less efficient resulting in few to no clones. Given that macroH2A2 must be expressed in order for targeting by the promoter-trap construct, it seems likely that the during the extended culture of RPE1 cells, either expression of macroH2A2 was gradually shut down in the majority of cells, or that those cells not expressing macroH2A2 had a growth advantage and slowly took over as the predominant cell in the culture. Consequently, targeting of macroH2A2 might prove more successful if an early passage RPE1 culture is used.

Figure 16. Additional Western Blots of MacroH2A1 Knockouts. Western blots were performed on all macroH2A1 knockouts, one macroH2A1 single target (Clone-3) and two macroH2A2 single targets (Clone-17 and Clone-H5). These were performed to confirm that Clone-64 is a true macroH2A1 knockout (top), as well as confirm the loss of macroH2A2 in that clone (middle). The blots were stripped and rehybridized with H4 as a loading control (bottom). The RPE1 control shows an unexpected reduction in macroH2A2 levels compared with many of the other clones.

24

Figure 17. Anti-macroH2A2 Immunofluorescence. Indirect IF to macroH2A2 was performed on RPE1, Clone-21 and Clone-64. Expression of macroH2A2 in RPE1 was only present in a small number of cells, whereas macroH2A2 appears to be expressed in all Clone-21 cells. The lack of macroH2A2 expression in Clone 64 is consistent with the Western Blot data.

25

Methods Nucleofection All of the targeting, except for the macroH2A1 retargeting, was nucleofected as follows: 1x106 RPE1 cells collected and resuspended into Lonza Type R nucleofection reagent along with 1µg of SmaI linearized targeting construct and 500ng of each ZFN. This mix was put into a cuvette and nucleofected in the Lonza Nucleofector™ II Device. For macroH2A1, the Site 2 Left ZF10 and Right ZF10 were used along with the macroH2A1 construct. For macroH2A2, the Site 1 Left ZF2 and Right ZF10-4 were used along with the macroH2A2 construct. Nucleofected cell were transferred to a 100mm dish with 10ml of prewarmed DMEM/F12 media and recovered overnight at 37°C. The following day, the cells were collected from the dish and transferred to four 96-well plates with media containing selection. When using Neomycin selection, 500µg/ml was used for macroH2A1 targeting, and 300µg/ml was used for macroH2A2 targeting. For Zeocin selection, 100µg/ml was used, and for Blasticidin selection, 10µg/ml.

The macroH2A1 retargeting was performed with the new Lonza 4D-Nucleofector™ System. The protocol was performed as above except cells were resuspended in Lonza Type SF solution. Additionally, the cells were allowed to recover immediately after nucleofection by putting 500µl of prewarmed DMEM/F12 media into the cuvette and incubating for 10 minutes at 37°C before transferring to the 100mm dish containing 10ml of media.

Screening Cells were incubated at 37°C for 3-4 weeks to allow the emergence of single cell clones. Any wells with colony formation were transferred to 6 well plates so that the colonies could grow in number. When the wells reached confluency the cells were collected, with a small amount (10%) seeded back into the well, and the rest used for DNA isolation. PCR screens were performed on each one of the isolated clones. Primers amplifying between the pSEPT vector and genomic DNA on the left and right sides were used to confirm successful insertion of the targeting construct at the correct location. In addition, the Flox PCR was performed which amplified across exon-2. If a small product is absent from this PCR then it indicates that both alleles have been successfully targeted.

26

Immunofluorescence Cells, which were seeded onto slides the previous day, were fixed in a formaldehyde solution (3.7% formaldehyde, 0.1% Triton X-100 in 1x PBS) for 10 minutes. The slides were washed twice with 1x PBS and incubated with blocking solution (PBS containing 0.1% Tween-20 (PBS- T) and supplemented with 3% bovine serum albumin (BSA)) for 30 minutes. After blocking, the slides were washed twice with 1x PBS and incubated with the primary antibody for 1 hour at room temperature. Primary antibodies were diluted 1:200 (anti-macroH2A1 or anti-macroH2A2) in PBS-T containing 1% BSA. Slides were washed three times in 1x PBS before addition of the secondary antibody diluted 1:200 in PBS-T containing 1% BSA. Slides were washed once more, before fixing and washing as above. Samples were mounted in Vectashield containing 4’, 6- diamidino-2-phenylindole (DAPI).

Western Blotting For the first western blot performed on macroH2A1 clones 3, 8, 21, and 70, insoluble protein extracts were made. This was done by harvesting each clone from a 100mm dish and pelleting the cells. The pellet was washed with PBS and resuspended in lysis buffer (1% IGEPAL with protease inhibitors AEBSF and Pepstatin-A in PBS). The samples were incubated on ice for 10 minutes then pelleted. The insoluble fraction was resuspended in digest buffer containing MNase and incubated at 37°C for 15 minutes. Loading buffer containing BME was added and heated to 75°C for 4 minutes. 15µl was separated on a polyacrylamide gel for 40 minutes at 200 volts then stained with Coomassie dye. This provided approximate concentrations so that loading volumes could be adjusted for the western blot.

The samples were then run on a new polyacrylamide gel using the same conditions as above, but with the following volumes: RPE1 (15µl), mH2A1-3 (10µl), mH2A1-8 (10µl), mH2A1-21 (8µl), and mH2A1-70 (20µl). The gel was then transferred to blotting buffer (2.5g of Non-fat Dry Milk in 50 ml of PBS-Tween) for 15 minutes and the polyvinylidene difluoride (PVDF) blotting membrane was pre-wet in 100% methanol before being put in blotting buffer for 15 minutes. Blotting was done using an Invitrogen apparatus and transferred at 30v for 2 hours at 4°C.

The blot was put in blocking buffer for 30 minutes at room temperature then incubated with a 1:800 dilution of macroH2A1 primary antibody for 1 hour. It was then rinsed twice with washing

27 solution (1% IGEPAL in PBS-Tween) and a 1:10,000 dilution of anti-rabbit secondary antibody was incubated for an additional hour. The blot was washed as before then treated with ECL solution then exposed to film.

For the second western blot performed on macroH2A1 clones 3, 8, 21, 64 and 70 and macroH2A2 clones 17 and H5, acid extraction of histones was performed. This was done by harvesting each clone from a 100mm dish and pelleting. The pellet was resuspended in TEB buffer (0.5% Triton X100, 1mM AEBSF, 0.02% NaN3, and 5mM NaButyrate in 1x PBS) to lyse the cells. The cells were lysed on ice for 10 minutes, swirling every minute, then pelleted at 4°C, 2000 rpm for 10 minutes. The supernatant was discarded and the pellet washed with half the volume of TEB and pelleted again, removing and discarding the supernatant. The pellet was then resuspended in 0.2N (0.2M) HCl. The samples were rotated overnight at 4°C.

The following day the samples were pelleted at 2000 rpm for 10 minutes at 4°C. The supernatant was transferred to a fresh tube and the concentration determined using the Nanodrop. An equivalent volume of loading buffer was added to each sample and heated at 80°C for 4 minutes before placing on ice. Approximately 30µg per lane, along with 10µl of Biorad Kaleidoscope pre-stained marker, and 10µl of Biotinylated protein ladder from Cell Signaling Technology was loaded onto a polyacrylamide gel and separated for 40 minutes at 200 volts.

The gel was then transferred to a Bio-rad Trans-Blot® Turbo™ Transfer System. This was done by placing the bottom stack of a pre-made Bio-Rad transfer pack into the transfer system cassette. The PVDF membrane is located at the top of this stack. The gel is oriented on top of the PVDF membrane and the top stack is placed on top of the gel. The cassette was then inserted into the transfer system and run for 25 volts, 2.5 amps for 10 minutes followed by 13 volts, 2.5 amps for 5 minutes.

The PVDF membrane was then prepared as described previously (both anti-macroH2A1 and anti-macroH2A2 primary antibodies were added at a 1:800 dilution). The membrane was exposed to film then stripped in 20ml of stripping solution rotating overnight at 50°C.

28

The following day, the PVDF membrane was once again prepared as done previously but this time a control anti-H4 primary antibody was used. This ensured that roughly equal amounts of histones were loaded onto each gel.

29

CHAPTER FOUR

EXPRESSION ANALYSIS OF MACROH2A1 KNOCKOUTS

Expression Analysis of Clone-8 Several independent clones were confirmed not to generate macroH2A1 protein (Clones 8, 21, 70 and later, 64). We sought to determine changes in gene expression (both up and down) that accompanied the loss of macroH2A1. Due to concerns over functional redundancy between macroH2A1 and macroH2A2, the level of macroH2A2 mRNA was determined in the mutants and compared to that of the parental RPE1 cells (Figure 10). Clone-8 showed comparable macroH2A2 levels and was selected along with parental RPE1 for whole transcriptome sequencing (RNA-Seq).

Clone-8 has exon-2 replaced with the pSEPT cassette at one allele, and most of exon-2 is deleted at the second allele due to NHEJ (Figure 7). Unless the NHEJ disrupted allele impacts the stability of the mRNA, we would expect that macroH2A1 mRNA to be expressed, but lacking exon-2. Analysis of the RNA-Seq data indicates that macroH2A1 mRNA lacking exon-2 is indeed present in Clone-8 (Figure 18).

Figure 18. MacroH2A1 Sequence Alignment. A sequence alignment of RPE1 and macroH2A1 knockout Clone-8 cDNA to the macroH2A1 gene. Each peak represents exonic sequence aligning to the macroH2A1 gene. Clone-8 shows a loss of sequence alignment to exon-2. This shows that while some macroH2A1 mRNA is being made it is lacking exon-2.

30

These data revealed a total of 1,438 genes that were found to have a greater than 2 fold change in expression levels (Figure 19).

Figure 19. Scatter Plot of Gene Expression Changes in MacroH2A1 Knockout. Gene expression changes that have occurred in macroH2A1 knockout Clone-8 compared to RPE1. Genes in red have not changed significantly (<2 fold), while genes in black have gone up or down significantly (>2 fold). Axes are represented in RPKM (Reads per kilo base per million) values, with the higher the value, the higher the confidence.

In order to validate the RNA-Seq data, primers suitable for qRT-PCR were designed to 15 genes that showed the greatest change in expression (Table 1). With the exception of MEMO1-Like, the qRT-PCR data was consistent with the RNA-Seq data for Clone-8, and most were consistent for the other macroH2A1 mutants Clone-21 and Clone-70 (See Figure 20 and Table 2).

These data suggest that the loss of macroH2A1 is likely responsible for most of these gene changes. However, RNA-Seq analysis was also performed on RNA isolated from an unrelated targeted gene BAZ1B, and a comparison of the RNA-Seq data obtained from the macroH2A1 knockout to that of the BAZ1B knockout indicated that many of the genes that showed altered expression levels are shared between the mutants relative to the parental RPE1.

31

Therefore, it is possible that some gene expression changes are unrelated to macroH2A1 loss, but are instead associated with the targeting process (i.e. response to the presence of ZFN or incubation in neomycin or the presence of the neomycin resistance protein). Consequently, those genes that respond comparably between macroH2A1 and BAZ1B mutants were excluded from further analysis, reducing the overall number of genes that show macroH2A1-specific change to less than 200. Of these genes, several were selected for further analysis (Table 3). Unlike before, these genes do not demonstrate such dramatic changes in gene expression levels.

Once again, all knockouts were analyzed by qRT-PCR. Two additional clones were included in the analysis. The first was a clone that is targeted at only one allele of macroH2A1 (a heterozygous mutant), with the second allele intact (Clone-3). Clone-3 produces reduced levels of the macroH2A1 protein (Figures 9 and 10). The second clone is a subclone of Clone-8 in which the neomycin cassette had been removed through Ad-Cre treatment.

Table 1. Genes of Interest – Set 1.

Gene Description Fold Change Trend ATP-binding cassette, sub-family B (MDR/TAP), ABCB5 139.7 up member 5 AP000689.1 MEMO1 Like -847.7 down ARSH arylsulfatase family, member H 24.6 up C14orf169 chromosome 14 open reading frame 169 860.3 up COL11A1 collagen, type XI, alpha 1 28.2 up CYSLTR1 cysteinyl leukotriene receptor 1 86.0 up FMO3 flavin containing monooxygenase 3 38.5 up GABRE gamma-aminobutyric acid (GABA) A receptor, epsilon -139.9 down GPR143 G protein-coupled receptor 143 -183.7 down HSD17B2 hydroxysteroid (17-beta) dehydrogenase 2 30.4 up PLXNA4 plexin A4 -166.6 down PTGFRN prostaglandin F2 receptor negative regulator -102.9 down serpin peptidase inhibitor, clade B (ovalbumin), SERPINB2 -60.7 down member 2 serpin peptidase inhibitor, clade G (C1 inhibitor), SERPING1 39.5 up member 1 TIMP3 TIMP metallopeptidase inhibitor 3 -147.4 down

32

Figure 20. Quantitative RT-PCR Analysis on Genes of Interest – Set 1. Quantitative RT-PCR analysis on a log scale of 15 genes selected from the RNAseq showing the greatest changes in gene expression levels. These genes were accessed in Clone-8 to verify the expression trends observed in the RNAseq, and were also accessed in two other macroH2A1 knockouts (Clone-21 and Clone-70) to determine whether these trends were clone specific or similar across all knockouts. Out of the 15 genes shown above, at least 10 appear to have similar trends among all macroH2A1 knockouts.

33

Table 2. Summary of Quantitative RT-PCR Trends for MacroH2A1 Knockouts Compared to RNAseq.

Gene C-8 Seq. C-8 Trend C-21 Trend C-70 Trend ABCB5 Up Up Up Up ARSH Up Up No change Up C14orf169 Up Up Up Up COL11A1 Up Up Up Up CYSLTR1 Up Up Up Up FMO3 Up Up Up Up GABRE Down Down Down Down GPR143 Down Down Down Down HSD17B2 Up Up Up Up MEMOP1 (AP000689.1) Down Up Up Up PLXNA4 Down Down Down Down PTGFRN Down Down Down Up SERPINB2 Down Down Down Down SERPING1 Up Up Up Down TIMP3 Down Down Down Up

Table 3. Genes of Interest – Set 2.

Gene Description Fold Change Trend ARMCX2 Armadillo Repeat Containing, X-Linked 2 -3.7 Down CCL3 chemokine (C-C motif) ligand 3 5.7 Up CNN1 calponin 1, basic, smooth muscle 10.1 Up CXCR7 C-X-C chemokine receptor type 7 2.1 Up CYBA cytochrome b-245, alpha polypeptide 7.5 Up FAM131C family with sequence similarity 131, member C 2.6 Up FAM198B family with sequence similarity 198, member B 2.5 Up FGF11 fibroblast growth factor 11 3.6 Up G0S2 G0/G1switch 2 -11.0 Down GPX3 glutathione peroxidase 3 (plasma) -11.5 Down KCTD16 potassium channel tetramerization domain containing 16 6.0 Up LSR lipolysis stimulated lipoprotein receptor -2.2 Down MYD88 myeloid differentiation primary response 88 -41.2 Down NCF2 neutrophil cytosolic factor 2 7.8 Up QPRT quinolinate phosphoribosyltransferase -9.9 Down SCD stearoyl-CoA desaturase (delta-9-desaturase) 2.0 Up SH3RF3 SH3 domain containing ring finger 3 3.0 Up TGM2 transglutaminase 2 -5.0 Down ZMAT3 zinc finger, matrin-type 3 2.4 Up

34

Consistent with our observations for the first set of genes that were tested, expression changes in Clone-8 by qRT-PCR were comparable to the RNA-Seq data. However, only some of these genes show the same expression trends across all macroH2A1 knockouts (Figure 21). Furthermore, some genes show apparent differences between the -Neo subclone and Clone-8, suggesting that the neomycin resistance gene product could influence gene expression. Those genes which showed similar trends (up in all knockouts, or down in all knockouts) were considered to be a direct result of macroH2A1 loss (Table 4).

Expression Analysis of Clone-21 To complement this analysis and to identify consistent gene expression changes between mutants, we performed expression microarray analysis for the double-targeted Clone-21, an independent macroH2A1 knockout. The expectation was that the results would be comparable to the RNA-Seq data for Clone-8, validating gene expression changes caused by macroH2A1’s loss. However, much to our surprise, the results were actually very contradictory.

Over 3400 genes showed a greater than 2-fold expression difference between the parental RPE1 cells and Clone-21; more than twice as many as the RNA-Seq showed between RPE1 and Clone- 8. In addition, many of the genes which exhibited either an upward or downward change in expression on the RNA-Seq showed either no significant change or the opposite trend on the microarray. Therefore, it appears that there is substantial heterogeneity in gene expression between clones that are not a direct result of macroH2A1 loss.

As was done with the RNA-Seq data, several of these genes were tested by qRT-PCR to confirm the reliability of the microarray data (Table 5). However, unlike with the RNA-Seq data, trends in gene expression were not consistent. Of the 14 genes analyzed by qRT-PCR, 8 did not agree with the microarray results (Figure 22, Table 6).

Next, genes with large changes in gene expression were chosen to be tested by endpoint RT-PCR (Table 7). Genes with large changes in expression levels may actually represent genes which have gone from off to on (or vice versa). The results are shown in Figure 23 and summarized in Table 8. Of the 6 genes tested, only one gene (FMO3), showed a clear on/off relationship.

35

512.00 256.00 128.00 64.00 32.00 16.00 8.00 4.00 2.00 1.00 0.50 0.25 0.13 Relative Normalized Expression Expression Normalized Relative 0.06 0.03 0.02 0.01 ARMCX2 CCL3 CNN1 CXCR7 CYBA FAM131C FAM198B FGF11 G0S2 GPX3

RPE1 mH2A1 # 3 mH2A1 # 8 mH2A1 # 8 -Neo mH2A1 #21 mH2A1 #70

16.00 8.00 4.00 2.00 1.00 0.50 0.25 0.13 0.06 0.03 Relative Normalized Expression Expression Normalized Relative 0.02 0.01 0.00 KCTD16 LSR MYD88 NCF2 QPRT SCD SH3RF3 TGM2 ZMAT3

RPE1 mH2A1 # 3 mH2A1 # 8 mH2A1 # 8 -Neo mH2A1 #21 mH2A1 #70

Figure 21. Quantitative RT-PCR Analysis on Genes of Interest – Set 2. Quantitative RT-PCR analysis on 19 more genes selected from the RNAseq. These genes were selected based on being unique to the macroH2A1 knockout and not shared with a BAZ1B knockout. Once again, these genes were accessed in Clone-8 to verify the expression trends observed in the RNAseq, and were also accessed in two other macroH2A1 knockouts (Clone-21 and Clone-70) to determine whether these trends were clone specific or similar across all knockouts.

36

Table 4. Summary of Quantitative RT-PCR Trends for all MacroH2A1 Targets Compared to RNAseq.

C8 -Neo. Gene C8 Seq. C3 Trend C8 Trend C21 Trend C70 Trend Trend ARMCX2 Down Down Down Up Up Down CCL3 Up Up Up No Change Up Up CNN1 Up Up Up Up Up Up CXCR7 Up Up Up Up Up No Change CYBA Up Up Up Down Down Down FAM131C Up Up Up No Change Up Up FAM198B Up Up Up Up Up Up FGF11 Up Up Up Up Up Up G0S2 Down Down Down Down Down Down GPX3 Down Down Down Down Down No Change KCTD16 Up Up Up Up Up Up LSR Down Down Down Down Down Down MYD88 Down Up Down Up Up Up NCF2 Up Down Up No Change Down No Change QPRT Down Up Down Up Up Up SCD Up Up Up Down Up Up SH3RF3 Up Down Up Up Up Down TGM2 Down Up Down Up Up Down ZMAT3 Up Up Up Up Up Up

Table 5. Genes of Interest – Set 3.

Gene Symbol Description Fold Change (array) Trend CCL2 Chemokine (C-C Motif) Ligand 2 184 Up CTSZ cathepsin Z 100 Up GALC Galactosylceramidase 60 Up GATA3 GATA binding protein 3 -70 Down HOXA3 homeobox A3 -7 Down HOXB6 homeobox B6 -15 Down HOXB7 homeobox B7 -4 Down HOXB8 homeobox B8 -6 Down RGS3 regulator of G-protein signaling 3 2 Up SLC25A20 solute carrier family 25, member 20 2 Up ZFP93 zinc finger protein 93 3 Up ZNF114 zinc finger protein 114 3 Up ZNF141 zinc finger protein 141 29 Up ZNF253 zinc finger protein 253 6 Up

37

Figure 22. Quantitative RT-PCR Analysis on Genes of Interest – Set 3. Quantitative RT-PCR analysis on a linear scale of 15 genes selected from the microarray. These genes were selected based on having consistent trends between the RNAseq and the microarray. The expression level of each gene in RPE1 parental is normalized to 1.

Table 6. Summary of Quantitative RT-PCR Trends for MacroH2A1 Clone-21 Compared to Microarray.

Gene Symbol Clone-21 Microarray Clone-21 qRT-PCR CCL2 Up Down CTSZ Up Down GALC Up Down GATA3 Down Down HOXA3 Down Down HOXB6 Down Down HOXB7 Down Down HOXB8 Down Down RGS3 Up Down SLC25A20 Up Down ZFP93 Up Down ZNF114 Up Down ZNF141 Up Down ZNF253 Up Up

38

Table 7. Genes of Interest – Set 4. Genes showing large expression changes on Microarray

Gene Symbol Description Fold Change (array) Trend amyloid beta (A4) precursor protein-binding, APBB1IP 111 Up family B, member 1 interacting protein DPYSL4 dihydropyrimidinase-like 4 108 Up FMOD Fibromodulin -136 Down FOXE1 forkhead box E1 (thyroid transcription factor 2) 88 Up NRXN3 neurexin 3 40 Up OCIAD2 OCIA domain containing 2 68 Up

Figure 23. RT-PCR Analysis on Genes of Interest – Set 4. Ethidium bromide stained agarose gel showing RT-PCR analysis on 6 genes with significant fold changes represented on the microarray. RT- PCR was selected to access whether a clear on to off (or off to on) pattern emerges. Only FMOD has a clear pattern of on in RPE1 to off in Clone-8. Multiple primer sets were developed and tested above for possible use in qRT-PCR. A GAPDH loading control is shown last.

Table 8. Summary of RT-PCR Trends for Gene Set 4.

Gene Symbol Clone-21 Microarray Clone-21 RT-PCR APBB1IP Up No Change DPYSL4 Up Down FMOD Down Down FOXE1 Up No Change NRXN3 Up No Change OCIAD2 Up No Change

39

One factor that might account for the discrepancies in the expression data is the parental RPE1 cells. Performing qRT-PCR analysis of cDNA prepared from RNA isolated from different batches of RPE1 was not consistent (Data not shown). One RNA source was obtained from an early passage of RPE1, whereas the other was isolated from cells that had been in culture for an extensive period and subjected to several periods of storage at -80°C. Both the RNAseq and the quantitative and endpoint RT-PCRs shown above have been performed with cDNA prepared from RNA isolated from an early passage of RPE1.

A separate microarray analysis was performed with RNA isolated from an early passage of RPE1 (referred to as the 3rd hybridization). Figure 24 shows a Venn diagram comparing the gene expression changes between Clone-21 and each RPE1 control. Genes which overlap were consistent between the two RPE1 passages and were considered to be significant and any non- overlapping genes were excluded from further analysis. Below is a list of genes accessed previously, comparing the original 1st hybridization control with the 3rd hybridization control (Table 9). Of the 21 genes originally chosen for analysis, 9 are not consistent.

Figure 24. Venn Diagram Comparing Two Different RPE1 Controls. Two different RPE1 sources were used as a control. Clone-21 was compared to each of these two controls separately and then a Venn diagram was created comparing the gene expression changes in common. A represents Clone-21 vs RPE1 3rd hybridization (early passage). B represents Clone-21 vs RPE1 1st hybridization (later passage).

40

Table 9. Gene Expression Changes Comparing Different Passage Controls to Clone-21 Gene Symbol RPE1 (1st Hyb.) vs Clone-21 RPE1 (3rd Hyb.) vs Clone-21 APBB1IP Up Up CCL2 Up No Change CIAS1 Down Down CTSZ Up No Change DPYSL4 Up No Change FMOD Down Down FOXE1 Up Up GALC Up Up GATA3 Down Down HOXA3 Down No Change HOXB6 Down Down HOXB7 Down Down HOXB8 Up No Change NRXN3 Up No Change OCIAD2 Up No Change RGS3 Up No Change SLC25A20 Up No Change ZNF114 Up Up ZNF141 Up Up ZNF253 Up Up ZNF93 Up Up

A total of 668 genes are in common between Clone-21 compared to both RPE1 controls. In light of these new data, genes were also compared across both RNAseq and microarray sources. A small number of genes did exhibit consistent trends between the RNA-Seq and the microarray (233 total). It was presumed that this small number of genes may represent the only genes directly affected by the loss of macroH2A1 and that the other gene changes were artifacts of either the analyses, targeting, or variation between passages.

In comparing genes that were consistent between the RNAseq and microarray, more genes were selected for analysis. Two genes, EMR1 and NPTX1, were consistent between the RNAseq and only one microarray control, but were also accessed (Table 10). These genes were selected to be tested by endpoint RT-PCR (Figure 25). Out of 21 genes, 14 appeared consistent, 2 inconsistent, and 5 could not be determined by endpoint analysis (Table 11).

41

Table 10. Genes of Interest – Set 5

Gene RPE1 vs. RPE1 (1st Hyb.) RPE1 (3rd Hyb.) Description Trend Symbol Clone-8 vs. Clone-21 vs. Clone-21 Acyl-CoA Dehydrogenase, Long ACADL 7.3 17.2 9.4 Up Chain ANXA10 annexin A10 -3.7 -5.1 -9.0 Down chitinase 3-like 1 (cartilage CHI3L1 -5.0 -25.4 -2.7 Down glycoprotein-39) COL17A1 collagen, type XVII, alpha 1 -2.3 -2.5 -2.5 Down COL1A1 collagen, type I, alpha 1 3.4 3.5 3.6 Up COL25A1 collagen, type XXV, alpha 1 7.3 3.7 2.4 Up collagen, type IV, alpha 3 COL4A3 4.7 16.7 9.4 Up (Goodpasture antigen) COL4A4 collagen, type IV, alpha 4 3.4 6.9 8.3 Up CYSLTR1 cysteinyl leukotriene receptor 1 86.0 2.9 5.9 Up DPYSL4 dihydropyrimidinase-like 4 2.2 247.2 108.3 Up egf-like module containing, mucin- EMR1 -36.8 - -42.2 Down like, hormone receptor-like 1 gamma-aminobutyric acid (GABA) A GABRE -139.9 -17.0 -9.1 Down receptor, epsilon GIMAP2 GTPase, IMAP family member 2 -4.1 -6.0 -8.6 Down interleukin 1 receptor accessory IL1RAPL1 -4.1 -18.8 -20.6 Down protein-like 1 potassium channel, subfamily K, KCNK1 -4.3 -25.5 -9.9 Down member 1 KIF5C kinesin family member 5C 8.5 4.1 5.8 Up MT1F Metallothionein 1F -2.1 -14.2 -9.0 Down NPTX1 neuronal pentraxin I -11.7 - -35.5 Down protein kinase, cGMP-dependent, PRKG1 9.3 2.5 15.4 Up type I retrotransposon gag domain RGAG4 -2.9 -21.8 -11.4 Down containing 4 sterile alpha motif domain SAMD9 -3.8 -68.3 -55.5 Down containing 9 SLAMF7 SLAM family member 7 -109.9 -17.9 -1.2 Down ZNF528 zinc finger protein 528 -94.2 -1.1 -6.4 Down

42

Figure 25. RT-PCR Analysis on Genes of Interest – Set 5. Ethidium bromide stained agarose gel showing RT-PCR analysis on 23 genes which showed consistent trends between RNAseq and Microarray data. Multiple primer sets were developed and tested above for some genes for possible use in qRT-PCR. A GAPDH loading control is shown last.

Table 11. Summary of RT-PCR Trends for Gene Set 5.

Clone-21 Clone-21 RT- Clone-21 Clone-21 RT- Gene Symbol Gene Symbol Microarray PCR Microarray PCR ACADL Up Up GABRE Down Down ANXA10 Down Down GIMAP2 Down Down CHI3L1 Down Down IL1RAPL1 Down Up COL17A1 Down Up KCNK1 Down Down COL1A1 Up Up KIF5C Up Down COL25A1 Up N/A MT1F Down Down COL4A3 Up N/A NPTX1 Down Down COL4A4 Up N/A PRKG1 Up N/A CYSLTR1 Up Up RGAG4 Down Down DPYSL4 Up N/A SAMD9 Down Down EMR1 Down Down

43

Using the RNAseq data as a guide, genes that showed a marked decrease in expression in the absence of macroH2A1 were identified (Table 12). Because macroH2A1 is generally considered to be a gene repressor (Costanzi and Pehrson, 1998), genes that are substantially reduced in expression or switched off, could be interpreted as genes that are dependent on macroH2A1 for their own expression. Alternatively, reduced expression of these genes might be an indirect effect, caused by activation of a repressor or silencing of an activator in macroH2A1’s absence.

Table 12. Genes of Interest – Set 6. Genes showing large decreases in expression across all analyses.

Gene RPE1 vs. RPE1 (1st Hyb.) RPE1 (3rd Hyb.) Description Trend Symbol Clone-8 vs. Clone-21 vs. Clone-21 chitinase 3-like 1 (cartilage CHI3L1 -5 -15 -2.5 Down glycoprotein-39)

GIMAP2 GTPase, IMAP family member 2 -4 -9 -4 Down

immunoglobulin superfamily, IGSF3 -2 -34 -17 Down member 3 Junctional Protein Associated With KIAA1462 On to off -15 -6 Down Coronary Artery Disease

MT1F metallothionein 1F -2 -9 -2 Down

PLXNA4 plexin A4 -167 -86 -34 Down

retrotransposon gag domain RGAG4 -3 -11 -5 Down containing 4 sterile alpha motif domain SAMD9 -4 -50 -16 Down containing 9

Quantitative RT-PCR analysis confirms that these genes all indeed go down in the macroH2A1 knockout Clone-21, and therefore appears that these reductions in expression are the result of macroH2A1’s loss (Figure 26). However, further study is needed to determine whether the downregulation of these genes are a direct effect of the loss of macroH2A1 or an indirect effect.

44

Figure 26. Genes Which Have Decreased Expression in Clone-21. Quantitative RT-PCR on a linear scale examining genes which have decreased expression levels in macroH2A1 knockout Clone-21 compared to RPE1 control. Genes with very low expression in Clone-21 may indicate that they have gone from on to off.

Expression Analysis of Clone-64 The primary reason for obtaining a second double-targeted clone from an independent targeting experiment was to validate gene expression changes observed in clones obtained from the first targeting experiment. Replicating the data would strengthen confidence in the observations made, especially given the discrepancies described so far.

Once again, qRT-PCR data revealed that many of the gene expression changes observed for Clone-21 were not consistent in Clone-64 (Figure 27). However, the following genes were consistent: ABCB5, ARMCX2, CCL2, CCL3, CHI3L1, CIAS1, FAM198B, FMO3, GABRE, HOXB8, HSD17B2, KIAA1462, MT1F, PLXNA4, SAMD9, SCD, and ZNF114.

These genes were then tested in all macroH2A1 knock-out clones by qRT-PCR (Figure 28, data not shown) and among them only CCL2, FMO3, GABRE, KIAA1462, PLXNA4, and SAMD9 were consistent.

45

Figure 27. Comparison of Gene Expression Changes in Clone-21 and Clone-64. A series of qRT- PCRs comparing expression level changes between RPE1 and Clone-21 and Clone-64. All of the genes tested previously were tested on new Clone-64. Many genes did were not consistent between the two clones.

46

Figure 28. Comparison of Gene Expression Changes in Clones 8, 21, 64, and 70. Genes which showed consistent trends between Clone-21 and Clone-64 were tested again using all of the macroH2A1 knockouts. Only genes which showed consistent trends among all knockouts are shown above.

One possible explanation for the variability between Clone-21 and Clone-64 is redundancy between macroH2A1 and macroH2A2 function. As described above, Clone-64 does not express macroH2A2 whereas Clone-21 does and therefore the loss of macroH2A1 could conceivably be compensated for by macroH2A2. This could explain why some genes show consistent expression levels among all of the knockouts, except for Clone-64 (data not shown).

This could be tested in the future by one of several approaches:

(1) Targeting macroH2A2 in Clone-21, (2) Reducing macroH2A2 levels in Clone-21 by RNA interference, or (3) Reintroducing a macroH2A2 transgene into Clone-64.

47

Methods RNAseq Analysis Total RNA was isolated from cells using the Qiagen RNeasy Mini kit. Library preparation and Illumina sequencing was performed through the services offered by HudsonAlpha Institute for Biotechnology (Huntsville, AL). Analysis by TopHat software was also performed by HudsonAlpha Institute for Biotechnology.

Microarray Analysis Total RNA was treated with DNase I for 1 hour at 37°C then cleaned up using the Qiagen RNeasy Mini kit. RNA was checked for quality by running a small amount on an agarose gel. In addition, the presence of DNA was checked for by performing a standard PCR using MYT1 and DXZ4 primers and running for 40 cycles.

Single-stranded cDNA was made with Applied Biosystems High Capacity cDNA Reverse Transcription kit. 2µg of RNA was used, and each sample was placed into the thermocycler at 25°C for 10 minutes, 37°C for 120 minutes, then 85°C for 10 minutes. To denature RNA/cDNA hybrids, the samples were heated to 95°C for 1 minute, placed on ice, then treated with RNaseA for 30 minutes at 37°C. The cDNA was cleaned up using the Qiagen PCR MinElute kit and eluted into 12µl of water.

The cDNA was hybridized to a Nimblegen microarray by the FSU Biology Core Facility. Microarray analysis was performed using the DNASTAR Arraystar 4 software package.

Quantitative RT-PCR Analysis In order to perform qRT-PCR analysis, cDNA had to be first synthesized. This was done by treating 4µg of RNA with DNase I for 20 minutes then deactivated by adding EDTA and incubating at 70°C for 10 minutes. After DNase treatment, the RNA was split into two tubes, one containing a mix with the reverse transcriptase (RT), and the other without. Random hexamers were used as primers. Each mix was placed into the thermocycler at 37°C for 1 hour, 42°C for 30 minutes, then 70°C for 10 minutes. Both the RT and the no RT mixes were diluted 1:20 in water.

48

SYBR Green compatible primers were used for qRT-PCR analysis. RTprimerDB was used to search for published primers of genes of interest. Genes which did not have published primers available had primers designed using predicted primers from Massachusetts General Hospital’s Primerbank software or designed primers using the NCBI’s PrimerBLAST software. See Appendix for primer sequences.

For each reaction, 1µl of each primer was added to 10µl of abm EvaGreen 2X qPCR MasterMix, along with 2µl of diluted cDNA and brought to 20µl with water. Quantitative RT-PCR was performed in a Bio-Rad CFX-96 with an initial step of 95°C for 10 minutes, then 35-40 cycles of 95°C for 15 seconds then 60°C for 1 minute. The plate was read after each cycle. Analysis of PCR results was performed using the Bio-Rad CFX Manager Software. Each sample was analyzed in triplicate and GAPDH was used to normalize between each sample.

49

CHAPTER FIVE

TRANSGENE RESCUE OF MACROH2A1

Generating Clone-21 Derived MacroH2A1 Rescues The macroH2A1 mutant clones provide a useful cell-based model to investigate the function of this unusual histone variant. Two alternatively spliced isoforms of the macroH2A1 gene contribute to macroH2A1 protein in a cell (Pehrson and Fried, 1992). To investigate the role of the macroH2A1 splice isoforms, expression constructs containing the full open reading frame of either macroH2A1.1 or macroH2A1.2, with an in-frame 3’ fusion to the coding sequence of the Myc-tag epitope (carboxy-terminal tagged), were reintroduced into macroH2A1 mutant cells. Differential expression of the macroH2A1.1 and macroH2A1.2 isoforms correlate with cancer prognosis (Sporn and Jung, 2012). One possible explanation for this could be that the two isoforms regulate different subsets of genes. Therefore, restoration of macroH2A1.1 or macroH2A1.2 alone could identify downstream target genes for each isoform.

Both the macroH2A1.1 and macroH2A1.2 cDNAs are cloned into the multiple cloning site of the pcDNA3.1-CT-Myc/His-A vector (Chadwick et al., 2001), and are under the control of the strong cytomegalovirus (CMV) promoter. The construct also contains a neomycin resistance gene to permit isolation of stable clones through selection. However, targeting of the endogenous macroH2A1 gene with the pSEPT construct resulted in clones being neomycin resistant, and therefore an alternative selection was required for the isolation of stably rescued clones. Alternatively, increasing the dose of neomycin might be sufficient to allow selection of a second neomycin construct.

To test this idea, a kill curve was performed, whereby macroH2A1 knockout Clone-8 cells were seeded into a 6 well plate and were exposed to varying levels of neomycin. Even at the highest neomycin concentration, cells grew to confluency unimpeded (Data not shown). Therefore, we sought to transfer the expression cassette to a vector carrying an alternate selectable marker gene.

Thus, the CMV promoter, macroH2A1.1/1.2 cDNA and Myc epitope were transferred from pcDNA3.1-CT-Myc/His-A into pCMV6-A-BSD using AscI and XbaI; a mammalian expression vector carrying the blasticidin resistance gene.

50

Constructs were introduced into Clone-21 by Nucleofection, and cells exposed to blasticidin before the isolation of clones. Individual clones were then assessed for the expression of the transgene by RT-PCR using a primer-pair specific to the transgene: one primer located in the alternatively spliced 1.1 or 1.2 exon 6 and the second primer within the 3’ untranslated region of the expression vector (Figure 29). Positive clones were then assessed for macroH2A1.1-Myc and macroH2A1.2-Myc by immunofluorescence with a mouse monoclonal antibody that is specific to the Myc-tag (Figure 30).

Figure 29. Myc Rescue PCR Screening. Ethidium bromide stained agarose gel images showing representative examples of Myc rescue clone screening. Clones are screened for both the insertion of the myc tagged macroH2A1.1 isoform, and for insertion of the myc tagged macroH2A1.2 isoform. MacroH2A1.1 clones are screened with the macroH2A1.2 primers (and vice versa) as negative controls.

Figure 30. Myc Rescue Immunofluorescence Screening. Indirect IF showing the distribution of Myc- tagged macroH2A1 in Clone-21 derived rescues. The top panel (blue) shows DAPI staining of the nuclei. The bottom panel (green) shows nuclei stained by IF with a mouse monoclonal antibody that is specific to the Myc-tag. Any cells with green stained nuclei are expressing one of the Myc-tagged macroH2A1 isoforms.

51

Several independent clones for each construct were isolated from both Clone-21 and Clone-64 and macroH2A1 isoform expression was assessed by qRT-PCR (Figure 31)

Figure 31. MacroH2A1 Isoform Expression in Each Myc Clone. (A) Quantitative RT-PCR analysis of macroH2A1 isoform expression and macroH2A2 expression of Clone-21 derived myc clones. (B) Quantitative RT-PCR analysis of macroH2A1 isoform expression and macroH2A2 expression of Clone- 64 derived myc clones.

52

Next, several clones were selected to access rescue of expression for each affected gene by performing quantitative RT-PCR (qRT-PCR).

Reintroducing macroH2A into the knockouts was expected to have one of the following outcomes for each gene:

- Gene expression is not rescued by either macroH2A1.1 or macroH2A1.2 - Gene expression is rescued by macroH2A1.1 and macroH2A1.2 - Gene expression is rescued by either macroH2A1.1 or macroH2A1.2 only.

Analysis of the qRT-PCR data revealed that some genes are preferentially rescued by one isoform. Genes that appear to be fully or partially rescued by macroH2A1.1 only are ACADL, CCL3, COL11A1, and CYSLTR1, while both isoforms appear to rescue FMO3, SAMD9, and SERPINB2. (See Figure 32, Genes which did not show rescue are not shown). The NPTX1 gene appears to be rescued by macroH2A1.2, but data obtained for the two macroH2A1.2-Myc clones is not consistent, complicating the analysis.

Generating Clone-64 Derived MacroH2A1 Rescues In order to attempt to validate these observations, the macroH2A1.1-Myc and macroH2A1.2- Myc constructs were introduced into the macroH2A1 knockout Clone-64, and blasticidin resistant clones were isolated. In sharp contrast to the rescue attempts in Clone-21, Clone-64 showed no evidence of rescue (Figure 33). However, at this time, we were unaware of the fact that Clone-64 lacks both macroH2A1 and macroH2A2, which may account for the conflicting observations.

Methods Generating Rescues The pCMV6 vector containing myc tagged macroH2A1.1 and macroH2A1.2 splice forms were independently nucleofected into the macroH2A1 knockout Clone-21. 1µg of plasmid DNA was used for the nucleofection and was performed as previously described. The blasticidin selection did not seem very efficient, and multiple colonies appeared in each of the 96 wells within one week, however, when an attempt was made to subclone any of the colonies they died off. It appears that the expression construct is expressed transiently and provides short term resistance 53 without integrating. The construct has to be integrated in order to create a stably expressing MYC clone that maintains blasticidin resistance. To overcome this, 24 wells were transferred from the 96 well plates to 24 well plates, and fresh blasticidin was added.

Figure 32. MacroH2A1 Clone-21 Derived MYC Rescues. Several genes tested by quantitative RT-PCR analysis on Clone-21 derived MYC rescues for return of wildtype expression. The examples shown are the only genes which appeared to demonstrate complete or partial return to wildtype levels in the rescue clones.

54

As expected, most of the cells died and a few single cell clones emerged in the 24 well plates. These were then transferred to 6 well plates and DNA isolated as described before. This process was repeated on Clone-64 to generate Clone-64 derived rescues.

Figure 33. MacroH2A1 Clone-64 Derived MYC Rescues. Several genes were tested by quantitative RT-PCR analysis on Clone-64 derived MYC rescues for return of wildtype expression; however, no genes exhibited rescue in any of the Clone-64 derived myc rescues. The examples shown are the genes which were rescued in Clone-21 derived myc clones as a comparison.

55

Screening Clones were screened first by genomic PCR using DNA isolated as previously described, with primers that amplify between exon 6 and the myc vector. Once positive clones were identified, RNA was isolated, and cDNA was made as described in the previous chapter. RT-PCR analysis was then done on those clones to confirm that the isoform was being transcribed. Finally, expression was quantified using qRT-PCR performed as described in the previous chapter.

Accessing Genes for Expression Rescue Previously done qRT-PCRs were repeated, comparing a few select rescue clones with that of RPE1 and macroH2A1 Clone-21 or Clone-64. Quantitative RT-PCR was performed as described in the previous chapter.

Immunofluorescence Immunofluorescence was performed as previously described in Chapter 3.

56

CHAPTER SIX

CHROMATIN ANALYSIS IN MACROH2A1 MUTANTS

Chromatin Immunoprecipitation of CCL2 Gene expression analysis of Clone-64 has revealed very little consistency between it and Clone- 21. Only a few genes tested by qRT-PCR were consistent among all of the knockouts, which include CCL2, FMO3, GABRE, KIAA1462, PLXNA4, and SAMD9

Of those genes, CCL2 was chosen for analysis by Chromatin Immunoprecipitation (ChIP) coupled with quantitative PCR (qChIP). CCL2 is composed of three exons that encompass <2kb of human chromosome 17q12, and its expression is substantially reduced relative to parental RPE1 cells in Clone-21 and undetectable in Clone-64 (Figure 28). Fourteen pairs of primers were designed at regular intervals throughout CCL2 and up to 6kb upstream (Figure 34). Twelve of the fourteen primer sets were confirmed as suitable for qPCR based on their efficiency and melt-curve profiles (Sets 3 and 6 were unsuitable). ChIP was performed with the following:

- macroH2A1 and macroH2A2 - To test if macroH2A1 and macroH2A2 are features of CCL2 chromatin and strengthen the supposition that changes in CCL2 expression are directly due to macroH2A1 loss and not a downstream indirect effector. - Histone H3 trimethylated at lysine-4 (H3K4me3) – a histone modification associated with active promoters (Guenther et al., 2007) to monitor changes at the CCL2 promoter in the mutant clones. - Histone H3 trimethylated at lysine-9 (H3K9me3) – a histone modification associated with gene silencing (Bernstein et al., 2006) to monitor changes in repressive chromatin at the CCL2 locus in the mutant clones.

Figure 34. Primer Pairs for Quantitative ChIP of CCL2. The area upstream and at CCL2 amplified by each primer pair. These primers will be used for quantitative ChIP analysis.

57

The macroH2A1 ChIP in parental RPE1 indicates that macroH2A1 is present throughout the interval upstream of the CCL2 gene (Figure 35). Within the gene body, macroH2A1 is clearly present in the first intron (F11.R11). Additional signals above background (indicated by the pre- bleed negative serum samples) can be detected at other locations in the gene body. However, these signals are comparable to that seen for Clone-21, which lacks macroH2A1. One explanation for this discrepancy is that macroH2A1 is very similar to macroH2A2 (Chadwick and Willard, 2001b; Costanzi and Pehrson, 2001) and consequently macroH2A1 antisera can weakly cross react with macroH2A2. Clone-21 expresses macroH2A2, as does RPE1, but Clone- 64 does not. The fact that Clone-64 does not show these macroH2A1 qChIP signals supports this being cross-reactivity. As anticipated, macroH2A1 signal upstream of CCL2 is lost in Clones-21 and 64.

The macroH2A2 ChIP indicates that macroH2A2 is primarily restricted to the F1.R1 region in parental RPE1 but that Clone-21 has elevated levels of macroH2A2 both upstream of and throughout the body of CCL2 (Figure 36). These data suggest that perhaps macroH2A2 is compensating for the loss of macroH2A1. Notably, Clone-64 that lacks both macroH2A1 and macroH2A2 shows no consistent signals above background throughout the interval. Interestingly, Clone-21 that appears to have acquired higher levels of macroH2A2 throughout the interval retains higher levels of CCL2 compared to Clone-64 that has no macroH2A protein. One interpretation of this observation is that CCL2 expression is dependent on macroH2A.

H3K4me3 levels are relatively high throughout the interval, with highest levels in the gene body. Consistent with reduced CCL2 expression, H3K4me3 levels are lower in Clone-21 and lower still in Clone-64. In contrast, heterochromatin levels, defined by the H3K9me3 qChIP, are higher in Clone-21 and highest in Clone-64. One interpretation of these data are that macroH2A1 protects CCL2 from the spread of H3K9me3 heterochromatin, a role that can be partially compensated by macroH2A2.

Analysis of the Inactive X Chromosome by Immunofluorescence and FISH The seminal observation that linked macroH2A1 with gene silencing was the elevated levels of the protein at the territory of the Xi (Costanzi and Pehrson, 1998), and subsequent determination

58

macroH2A1 ChIP 0.20% RPE1 IP 0.18% RPE1 Neg

0.16% mH2A1-21 IP mH2A1-21 Neg 0.14% mH2A1-64 IP 0.12% mH2A1-64 Neg

0.10%

0.08%

Percentage of Input Input of Percentage 0.06%

0.04%

0.02%

0.00% CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 F1.R1 F2.R2 F4.R4 F5.R5 F7.R7 F8.R8 F9.R9 F10.R10 F11.R11 F12.R12 F13.R13 F14.R14

macroH2A2 ChIP 0.30% RPE1 IP RPE1 Neg 0.25% mH2A1-21 IP mH2A1-21 Neg 0.20% mH2A1-64 IP mH2A1-64 Neg

0.15%

Percentage of Input Input of Percentage 0.10%

0.05%

0.00% CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 F1.R1 F2.R2 F4.R4 F5.R5 F7.R7 F8.R8 F9.R9 F10.R10 F11.R11 F12.R12 F13.R13 F14.R14

Figure 35. Quantitative ChIP Analysis of CCL2 with MacroH2A1 and MacroH2A2. The length of the CCL2 gene, and 6kb upstream, was accessed by qChIP for both macroH2A1 and macroH2A2 to access whether they are features of CCL2 chromatin. MacroH2A1 is present throughout CCL2, especially at the F1.R1 region, and the promoter region (F11.R11). In the absence of macroH2A1, macroH2A2 appears to perform a similar role.

59

H3K4me3 ChIP 250 RPE1 IP

mH2A1-21 IP 200 mH2A1-64 IP

150

100 Relative to Negative Negative to Relative

50

0 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 F1.R1 F2.R2 F4.R4 F5.R5 F7.R7 F8.R8 F9.R9 F10.R10 F11.R11 F12.R12 F13.R13 F14.R14

H3K9me3 ChIP 120 RPE1 IP

100 mH2A1-21 IP

mH2A1-64 IP 80

60

40 Relative to Negative Negative to Relative

20

0 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 CCL2 F1.R1 F2.R2 F4.R4 F5.R5 F7.R7 F8.R8 F9.R9 F10.R10 F11.R11 F12.R12 F13.R13 F14.R14

Figure 36. Quantitative ChIP Analysis of CCL2 with H3K4me3 and H3K9me3. The length of the CCL2 gene, and 6kb upstream, was accessed by qChIP for both H3K4me3 (a histone modification associated with active promoters) and H3K9me3 (a histone modification associated with gene silencing). As expected, RPE1 shows the greatest amount of H3K4me3 and least amount of H3K9me3 while Clone- 64 shows the opposite arrangement. Clone-21 follows that of Clone-64 but to a lesser extent.

60 that macroH2A2 subnuclear localization is indistinguishable (Chadwick and Willard, 2001a; Costanzi and Pehrson, 2001). Indeed, loss of Xist in mouse destabilizes gene silencing at the Xi coupled with loss of macroH2A1 association with the chromosome territory (Csankovszki et al., 1999). Therefore, we sought to determine if loss of macroH2A1 was coupled to perturbation of the Xi territory by examining changes to markers of the Xi. Immunofluorescence was performed on RPE1 cells alongside Clone-21 and Clone-64. The distribution of the following markers was performed:

- Enhancer of Zeste 2 (EZH2) – a histone methyltransferase responsible for histone H3 trimethylated at lysine-27 (H3K27me3) (Cao et al., 2002). EZH2 is a member of the polycomb repressive complex, PRC2, and is enriched at the Xi (Plath et al., 2003). - Structural maintenance of hinge domain-1 (SMCHD1) – a known marker of the Xi (Nozawa et al., 2013) - Histone H3 dimethylated at lysine-2 (H3K4me2) – A euchromatin marker that is absent from the Xi (Chadwick and Willard, 2002). - H3K9me3 – a feature of the human Xi (Chadwick and Willard, 2004). - H3K27me3 – the repressive chromatin modification caused by EZH2 - Heterochromatin protein 1 beta (HP1β) – a chromatin protein that recognizes H3K9me3 (Bannister et al., 2001; Lachner et al., 2001) and is enriched at the Xi (Chadwick and Willard, 2003). - Heterochromatin protein 1 binding on the inactive X chromosome (HBiX1) – a SMCHD1 and HP1 binding protein that is enriched at the Xi (Nozawa et al., 2013) - Histone H1 – linker histone that is enriched at the territory of the Xi (Chadwick and Willard, 2003). - XIST – a bonafide marker of the Xi chromosome territory (Brockdorff et al., 1992; Brown et al., 1992; Clemson et al., 1996)

The distribution of all chromatin proteins, histone modifications and XIST was comparable in Clone-21 relative to parental RPE1 (See Figures 37 - 42) suggesting that macroH2A1 is not necessary to maintain the Xi territory.

61

However, Clone-64 showed a distinct absence of any Xi feature, including localization of XIST RNA. Given that Clone-64 does not express macroH2A1 or macroH2A2, the results obtained for Clone-21 could reflect functional redundancy between the two proteins. Alternatively, it is possible that the Xi has been lost from Clone-64.

In order to test for this possibility, DNA fluorescence in situ hybridization (FISH) was performed using a direct labeled probe for the human X centromere (CEP-X). If Clone-64 retains the Xi, then we would anticipate observing two CEP-X signals in RPE1, Clone-21 and Clone-64.

Two CEP-X signals could be detected in most RPE1 and Clone-21 nuclei, but Clone-64 showed a mixture of nuclei with either 1 or 2 signals. This could indicate loss of the Xi in some, but not all cells. Alternatively, this could reflect loss of the Xi in all cells, but that cells showing two signals are tetraploid.

To address this, FISH was performed using an X-linked probe to the t-Plastin gene (PLS3) and a second direct labeled probe to chromosome 8. If the Xi is lost in all cells, we would expect twice as many chromosome 8 signals regardless of cells being diploid or tetraploid. However, if the Xi is lost in a proportion of cells, then we would expect those cells showing two X signals to have two chromosome 8 signals. Our FISH data consistently showed twice as many chromosome 8 signals to the X, indicating the Xi has indeed been lost in Clone-64. Furthermore, those nuclei with two X signals had four chromosome 8 signals, confirming them as tetraploid (Figure 43)

Methods Quantitative ChIP Chromatin was isolated from RPE1 clones at a density of 1.4x107 cells and fixed in 1% formaldehyde then sonicated to create sheared chromatin of the desired size. This was done by sonicating for 5 minutes (using 30 second ON/OFF cycles), then vortexing the samples and spinning them down at 4°C. This 5 minute sonication procedure was performed for 3 more rounds. Insoluble debris was removed by pelleting and the supernatant transferred to 6 individual tubes at 200µl each.

62

Figure 37. Anti-EZH2 Immunofluorescence. Indirect IF of RPE1 and macroH2A1 knockouts Clone-21 and Clone-64. The middle panel shows DNA stained with DAPI. The right panel shows nuclei stained by IF with an EZH2 antibody, and the left panel is a merge of the two signals.

Figure 38. Anti-H3K4me2 and Anti-HP1β Immunofluorescence. Indirect IF of RPE1 and macroH2A1 knockouts Clone-21 and Clone-64. The middle panel shows nuclei stained by IF with an H3K27me2 antibody, while the right panel shows nuclei stained by IF with an HP1β antibody. The left panel shows a merge of the two signals, along with a blue DAPI signal.

63

Figure 39. Anti-H3K9me3 and Anti-H3K27me3 Immunofluorescence. The above images show nuclei of RPE1 and macroH2A1 knockouts Clone-21 and Clone-64. The middle panel shows nuclei stained by IF with an H3K9me3 antibody, while the right panel shows nuclei stained by IF with an H3K27me3 antibody. The left panel shows a merge of the two signals, along with a blue DAPI signal.

Figure 40. Anti-H3K27me3 and Anti-H1 Immunofluorescence. Indirect IF of RPE1 and macroH2A1 knockouts Clone-21 and Clone-64. The middle panel shows nuclei stained by IF with an H3K27me3 antibody, while the right panel shows nuclei stained by IF with an H1 antibody. The left panel shows a merge of the two signals, along with a blue DAPI signal.

64

Figure 41. Anti-SMCHD1 and Anti-HBix1 Immunofluorescence. Indirect IF of RPE1 and macroH2A1 knockouts Clone-21 and Clone-64. The middle panel shows nuclei stained by IF with an SMCHD1 antibody, while the right panel shows nuclei stained by IF with an HBix1 antibody. The left panel shows a merge of the two signals, along with a blue DAPI signal.

For each sample, the 200µl of chromatin was split into two tubes, one for the IP and one for the negative control. In each tube, 100µl of chromatin was diluted to 1ml using ChIP dilution buffer (containing 1% triton X-100, 2mM EDTA, 150mM NaCl, 20mM Tris pH8, and protease inhibitors). Protein A beads were added and incubated by rotating for 1 hour at 4°C in order to block non-specific binding. The beads were removed, and 4µl of primary antibody was added to the IP sample and 4µl of rabbit serum was added to the negative control. Each sample was incubated by rotating overnight at 4°C.

The following day, Protein A beads were added each sample and incubated by rotating for 2 hours at 4°C. The beads were then washed twice with low wash buffer (0.1% SDS, 1% Triton X- 100, 2 mM EDTA, 150 mM NaCl, 20 mM Tris pH 8, protease inhibitors), once with high wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 500 mM NaCl, 20 mM Tris pH 8, protease inhibitors), then twice with TE buffer. The beads were incubated in 100µl of elution buffer by rotating for 15 minutes at room temperature. The elution was transferred to a new tube and repeated one additional time. Finally, 8µl of 5M NaCl was added to each tube and incubated at 65

Figure 42. XIST RNA FISH and CEP-X DNA FISH. (A) RNA FISH performed on RPE1 as well as macroH2A1 knockouts Clone-21 and Clone-64. Nuclei are stained with DAPI (blue), while the inactive X is stained with XIST (green). (B) DNA FISH performed on RPE1 as well as macroH2A1 knockouts Clone-21 and Clone-64. Nuclei are stained with DAPI (blue), while each X chromosome is stained with CEP-X (red). The combination of both RNA and DNA fish shows the Clone-64 has lost the inactive X.

6 C overnight. After incubating, the samples were treated with RNase A for 30 minutes, then Proteinase K for 1 hour. Samples were diluted 1:300 with water

To perform quantitative ChIP, 1µl of each primer was added to 10µl of abm EvaGreen 2X qPCR MasterMix, along with 2µl of DNA and brought to 20µl with water. Quantitative PCR was performed in a Bio-Rad CFX-96 with an initial step of 95°C for 10 minutes, then 35-40 cycles of 95°C for 15 seconds then 60°C for 1 minute. The plate was read after each cycle. Analysis of PCR results was performed using the Bio-Rad CFX Manager Software. Each sample was analyzed in triplicate, and an input diluted 1:2000 was used to normalize between each sample.

66

Figure 43. PLS3 and Chr8 DNA FISH. DNA FISH performed on RPE1 as well as macroH2A1 knockout Clone-64. Nuclei are stained with DAPI (blue), while each X chromosome is stained with anti- PLS3 (green), and each chromosome 8 is stained with anti-Chr8 (red). This is useful for identifying whether a cell is diploid or polyploid. 67

Primers were designed using the NCBI’s PrimerBLAST software. Primers were designed to amplify approximately 6000bp upstream of the gene and then every 300bp until reaching the end of the gene, for a total of 14 primer sets. The primers are as follows:

F1- GCTGCTCAGTATTTTCCCACC R1 – AGGAGGGGTCACATTTTGGG

F2 – TTAGATGACCCCAGCACAGG R2 – TGTGATGAGGGGGTCAAGGT

F3 – CCTGTGCAGTCCTCTACCTG R3 – TGCTGGTTTGCTTATTTCTGGG

F4 – GGGACAAGTGAACCGCAGAA R4 – TCCTCTGCATGAACCTTGGTC

F5 – CCACTCACTTCTCTCACGCC R5 – CTGTCTGCCTCCCACTTCTG

F6 – AAAACCCGAAGCATGACTGGA R6 – GCCATCTCACCTCATCTTCCA

F7 – GGAAACATCCTGGGTGGGAG R7 – AGGGAGTCAGGTATGGTGCT

F8 – GGCACAACTGAGGAATGAAGTC R8 – CAGGCAGCAGTGGACAAGA

F9 – TGCTGTCTATGCCTTTGTCCA R9 – AGCATCATAGAAGCCTAGCAGA

F10 – GACCCCCTGCTTCCCTTTC R10 – GCTGCTGTCTCTGCCTCTT

F11 – TGAACCCCAAATCCAGCTCC R11 – GGAAACAGCCTACACACCCA

F12 – GTCTTCTCCTGCCTGCCTTT R12 – CAGGTGACTGGGGCATTGAT

F13 – CTCCTTCCACCTGCGTTCC R13 – GGGGAGGGAGGACAGAGTT

F14 – TGTTGATGTGAAACATTATGCCT R14 – CCAGGGGTAGAACTGTGGTT

Immunofluorescence Immunofluorescence was performed as described in the Chapter 3. The primary antibody dilutions are as follows: 1:100 (Cell Signaling (D2C9) Rabbit anti-EZH2), 1:200 (Upstate Biotech 07-449 Rabbit anti-H3K27me3, Upstate Biotech 07-030 Rabbit anti-H3K4me2), 1:400 (Chemicon Mouse anti-HP1-beta), and 1:500 (Santa Cruz (AE-4) Mouse anti-Histone H1).

DNA FISH Probes were heated to 75°C for 8 minutes before placing at 37°C for 30-60 minutes to block repeats. Slides were fixed for 8 minutes in 3.7% formaldehyde, 0.1% triton X-100 PBS solution.

68

They were then washed in PBS and dehydrated through 70% and 100% ethanol, 2 minutes each at room temperature, then air dried. The slides were transferred to a denaturing solution (70% formamide, 2x SSC) at 83°C and incubated for 10 minutes. The slides were once again dehydrated as above. Next, 12µl of probe was added and a 22mm sq. cover-slip was sealed onto the slide using rubber cement. The slides were incubated at 37°C overnight in a humidified chamber. The following day, the rubber cement was removed, and coverslip is soaked off in the first wash. The first two washes were for 8 minutes each at 42°C in 50% formamide, 2xSSC and the final wash was for 8 minutes at 42°C in 2xSSC. DAPI was added, and the slide was sealed with a coverslip.

RNA FISH Probes were heated to 72°C for 10 minutes before placing at 37°C for 30-60 minutes to block repeats. Slides were fixed for 10 minutes in 3.7% formaldehyde, 0.1% triton X-100 PBS solution. They were then washed in PBS and dehydrated through 70%, 80%, and 100% ethanol, 2 minutes each at –20°C, then air dried. Next, 10-15µl of the probe was added and a 22mm sq. cover-slip was sealed onto the slide using rubber cement. The slides were incubated at 37°C overnight in a humidified chamber. The following day, the rubber cement was removed, and coverslip was soaked off in the first wash. The first two washes were for 2 minutes each at RT in 50% formamide, 2xSSC. The third wash was for 3 minutes at 37°C in 50% formamide, 2xSSC and the final wash was for 3 minutes at 37°C in 2xSSC. DAPI was added, and the slide was sealed with a coverslip.

69

CHAPTER SEVEN

CONCLUSIONS AND DISCUSSION

In spite of the fact that there was much inconsistency between the RNAseq data and the microarray expression data, at least one gene, CCL2, was identified that appears to be directly impacted by macroH2A loss. According to our ChIP data, under normal conditions, macroH2A1 is present throughout the CCL2 gene in RPE1 cells. However, in its absence, macroH2A2 appears to take its place. As shown in Figure 28, CCL2 is actively transcribed in RPE1; however, when macroH2A1 is lost there is a reduction of CCL2 expression. Additionally, when both macroH2A1 and macroH2A2 are lost the gene is completely shut off. In stark contrast to macroH2A acting strictly as a transcriptional repressor, these data seem to indicate that macroH2A can, in some instances, have the opposite role. This suggests a novel function for macroH2A and we have theorized that it can do this by preventing the spread of neighboring heterochromatin. Figure 44 illustrates a model to explain how macroH2A could assist in preventing the spread of the heterochromatic feature H3K9me3.

These data suggest that macroH2A2 can indeed compensate for macroH2A1 providing some level of functional redundancy. In our model, in RPE1 cells, macroH2A1 prevents the spread of H3K9me3, ensuring that the CCL2 gene remains on. In the absence of macroH2A1, the presence of macroH2A2 is able to prevent the spread of H3K9me3, albeit less effectively than when both macroH2A proteins are present. When macroH2A1 and macroH2A2 are both lost, heterochromatin is free to spread throughout the CCL2 gene turning it completely off.

As shown previously with the immunofluorescence and FISH on Clone-21, losing macroH2A1 does not appear to affect the distribution of chromatin proteins, histone modifications or XIST, suggesting that macroH2A1 is not necessary to maintain the Xi territory. However, one question which remains unanswered is whether this can be attributed to functional redundancy with macroH2A2. Unfortunately, this cannot be tested in Clone-64, because it has lost the Xi. A knockout of macroH2A1 and macroH2A2 would have to be obtained in RPE1 cells which still retained both X chromosomes to adequately answer this question.

70

Figure 44. Model of H3K9me3 Spreading. In RPE1 parental cells, macroH2A1 is distributed throughout CCL2. The presence of macroH2A1 can prevent spreading of H3K9me3 heterochromatin and the gene remains on. In Clone-21, which is lacking macroH2A1, macroH2A2 compensates for, but is less efficient at preventing the spread of heterochromatin, which results in reduced expression of CCL2. In Clone-64, which is lacking both macroH2A1 and macroH2A2, heterochromatin is free to spread, shutting down transcription of CCL2.

71

APPENDIX A

PRIMERS

Quantitative RT-PCR Primers Gene Forward Primer Reverse Primer ABCB5 CCGGATGATGGCTTTATCATGGTGG TGGTGGTCCCGAACAAAACAGGC ACADL TCATGCAGCTGGAGACAGTT TTGGCAAAACAGTTGCTCAC ADAMTS16 CCCTGTGGTGCCATCGTATT GAAGACCAGTCCGACCAGTG ANXA10 CTAGCATTTGGGCATCCATT TCCCTGAGGTTAACAATTACCA APBB1IP AGAGCAGTCGAGATGGGTGA GCTCTGGGTGGATTAGGGTC ARMCX2 AGTTCTGCTGAGGCTTAGG CCTTAGGATGAAGCTCTAACTT ARSH GTGCTGCTACGGTAATAACTCAG GCTGCGAGATGCTGGGTAA C14orf169 CCATGCCGCCAGATCACTTT CAGCATCGAATCCAGGTCAG CA9 GCGACGCAGCCTTTGAAT CCACTCCAGCAGGGAAGGA CCL2 ACTCTCGCCTCCAGCATGAA GGGAATGAAGGTGGCTGCTA CCL3 CGGTGTCATCTTCCTAACCA GACATATTTCTGGACCCACTC CHI3L1 GGCTCCATGAAAATCGTAGG GGGAAGGTCACCATTGACAG CIAS1 TCTAGAGGACCTTGAAGATG AAGTGATCTGCCTTCTCCAT CIAS1 (Set 2) TTACCTGCGAGGCAACACTC TGACGTGAGGTTGCAGTTGT CNN1 TTGAGAACACCAACCATACACA TTCCGCTCCTGCTTCTCTG COL11A1 GTCAGCGTGGTCCAACGGGTC GCCATCGCCACCTGAAGTGC COL17A1 GGCTGTGAACACTGGCGTT AGGTCATCGCTCTGCACACTT COL1A1 CACACGTCTCGGTCATGGTA AAGAGGAAGGCCAAGTCGAG COL25A1 TGTCCAGGAGGACCAGATTC GAATCGCAAGAGAAGCACCT COL4A3 GCACTGGCCTTTGTCTTTACA CAGGTGCTCCTGCTGCC COL4A4 CCTTTCTCTCCTGAAAGCCC TGTGTTCCTGAAAAGGGGTC CXCR7 CACTGCTACATCTTGAACCT AGAAGATGAGGTGTGTGACTTT CYBA GCGGTCCCAAGGATGGTG GCGTGTTTGTGTGCCTGCT CYSLTR1 AAGTCCGTGGTCATAACCTTGT TCTGGGTACATAAGTCACGCT CYSLTR1 (Set 2) TGGTGATCAATGCCTTTTACG GCTTCTGAGAACAAACGCAA DPYSL4 AGAATGAGTTCGTCGCGGTG CGCCCCTTCCTTGGGTAAAA DPYSL4 (Set 2) GTCATTCACGATCCTCCCAC CTACCAGAGACCCCCAGGAG EMR1 CAAGGTCCAGGTCAGGAGAA ATAATCGCTGCTGGCTGAAT ENG CAACATGCAGATCTGGACCAC CTTTAGTACCAGGGTCATGGC FAM131C CCGTTGGTGGTCTGGAAGC CTCCAGACTGTGTCATTGGCA FAM198B GTACACAGGCATCTTCCTTG CATCATCATGAGTGGTCCAAG FGF11 CGTGTGGTCACCATCCAGA CGTACAGGACGTAGTAATTCTC FMO3 GGCCATCATTGGAGCTGGTGTGA GCCCTCCTCTGCATGGTCTGAAA FMOD AACTTGAGAGACAAAATGCAGTG TAGTAGGTGGACTGCTGGCT FOXE1 TACTAACTGCCCTCGCTCTG TGTGGTGCCCGCTAGTTTAG G0S2 CCAGAGCCGAGATGGAAAC CGTACAGCTTCACCATCTTC

72

GABRE GCGCCCTGGCATTGGAGAGAA AGGCGTTCGTCGTACCAGGTCT GABRE (Set 2) CGTACCAGGTCTGGGAGAAG CATTGGAGAGAAGCCCACTG GALC CCTGCATGAAGATGAGCTGTTCACA TAGGTACTTGGGAAGGGCTGGGA GATA3 CTGCTTCATGGATCCCTACC GATGGACGTCTTGGAGAAGG GIMAP2 CCCACCAGGATGATTCTCAG ATCAGGAACACCAATGGACC GPR143 TTCCGTGTCCAGGTGTGAGCG TTCGCCACGAGAACCAGCAGC GPX3 GACAAGAGAAGTCGAAGATG CTTCCTGTAGTGCATTCAGTT HSD17B2 TGTCAGCAGCATGGGAGGAGGG CACAGCCGCCTTTGATGAGCCA IGSF3 AGGGCTCCCACATCACTATCT GAAGGCAGGTAAATGGACCAC IL1RAPL1 GACCAGTCAGTGCATCCATC GCTCCGATTCCACACTTGAT KCNK1 GGCCTTACCTCCATCTGACA TGGAACTGGGACTTCACCTC KCNN2 CACCCAATCCAGATTCCAGAG CTTTGGGGAGAAAAAGTTTGG KCNN2 (Set 2) GCAGCTGCCAATGTACTCAG CCACCAAAGTGTTTGCTTGGT KCTD16 AGATCTGGTCAAGCTACACTG GCTAGATGTGGAGAGGTCAT KIAA1462 CGGCTTCGTTCAGTACATTCC TGATTTATCGTCAAGCTTTATGTCTTC KIF5C ATGTCTTCGACAGAGTGCTACC ACGCAAAAATCGTCCCGTTAT KLF13 CAGGACTGCAACAAGAAGTT CGCTTGTCGCAGATAGG LOC401074 AAACCAACGGAGCCCTGAGA AGCCAACAAAGGAATGACGC LSR CGTGTGGCTTCTGCTTAGCA AGGCTGGAAGAGGATCACC MEMO1P1 CGAGTGGTCTGCCGAGAAGCC AGCTGTGCATTCAGCTGCGGT MT1F TGCATTTGCACTCTTTGCAC AGTCCAGTCTCTCCTCGGCT MYD88 TCAAGTACAAGGCAATGAAGA GTAGTCGCAGACAGTGATGA NCF2 CCTTCCAGCCATTCTTCATTC TGGAGGCACTCTTCAGTTATG NPTX1 GCTCTTCTTCACCTTGGCAT GATCAGCGAGCTCGAGAAAG NRXN3 ACTCGGGACAACAGTAATACCCAC CTGGGCTAAGCCAGCCATATAG NTNG2 GTGATGCGCCTGAAGGACTA ACACTCGTTGCTGCATAGGT OCIAD2 CACCCAGGGACTAGTCTACCA CGTGCAACTTTGGGCAATGA OR2M4 GGACTCTACTACGGTGCTGC CCTTCTGTAGTGCCCTGAACA PLXNA4 ACGGTGCAGGTGGTGGACCC GGGACTCTGGTGAGCTGCCTCT PLXNA4 (Set 2) AGAACCACTCTCTGGCCTTTG ATGTAGAGTTGCTCGTGGTCC PRKG1 GATCAATGGCCCAGAGTTTT TACCATGGGTCCAGGAAAAG PSCDBP TCATTATCCTGCTTCTCCACAG TCAAATGCTAGCAGACACGG PTGFRN GCGCTCCTGTCGTTGGCTCTTT TGACCAGCTCAGTGCCCACCA QPRT ACCTCTGCTCATCTCAGTTTC GCTCATTATCACCGCAGAAC RGAG4 TCTGAGGCGTCAGGAAATCTC CCCGCGAAGGGCATTTAATTC SAMD9 TGCAGTGGGTGAATAGGTGG GGAAGGTTAAGTTGCTTTGCCA SAMD9 (Set 2) AAGTTGCTTTGCCATTCTGA TTGCAGTGGGTGAATAGGTG SCD TCGGATATCGTCCTTATGACA AGGAGTGGTGGTAGTTGTG SERPINB2 AGGATGGTCCTGGTGAATGCTGT GGTGTGCGCTGAGCCGAGTT SERPING1 CTGCTCGGGGCTGGGGAGAA CCCTTCAGGGCCTGGTGGACA SH3RF3 CGGTGCTCAGGAGAGTGGAT GCTTGGCGGAGTCATTGAG SLAMF7 GAACCGACCAGCTCTTTCAC AATATGGCTGGTTCCCCAAC SLC14A1 CTCCTGAGAAGCACTCTCCCT AGCAGGAGATGCCCAGTTTTC

73

SLCO1A2 GGAAGCTTTGAGATTGGAAA TTCATATCGGTTCATGAGGA SLCO1B3 TCCATCAATTAAACCAGCAAGA AGACGCTGCAATGGATTCAA SPRY1 GACCGCTCTGCAAACCACTG TGTTCCTTATCCAGGGCAGG TGM2 GCCACTTCATTTTGCTCTTCAA TCCTCTTCCGAGTCCAGGTACA TIMP3 GACATCGTGATCCGGGCCAAGG GTAGACCAGCGTGCCGAAGGG TNFAIP6 TCATGTCTGTGCTGCTGGATG GGGCCCTGGCTTCACAA TNFRSF21 CCATGAAGTCCCTTCCTCCA TCACGTCTTCCTTCCCCCTT TRO TAGGAACGGAGATACAGAGTGG GGAAGTGGTCCTGCCTCTTT ZMAT3 GATGAGAAGCAAGGTCTTCAA GTGGACCAGGTGGATCATCT ZNF528 CAAGGCTTGTTTCCTGCATTA ATAGCAAACGATCCAGACGG

RT-PCR Primers Gene Forward Reverse DPYSL4 CGGTTCACACTACTGGAGCA TCACCGCGACGAACTCATTC EMR1 GACCCACACGGAAACCAAAC CTGCACTTGTACCTCCCTGAC FMOD ACTTGAGAGACAAAATGCAGTGGA GAGAAGACCTTCCTGCCCAC FOXE1 CCAGTACTAACTGCCCTCGC TGAGCGCGATGTAGCTGTAG GABRE AGATCACCATGCCCAACCAG GGGCACGGCTATTGATACGA IL1RAPL1 AAGACTGTTGTGGGGAACGG ACCAGTCAGTGCATCCATCG KCNK1 CTGGGCTACTTGCTCTACCTG CATCTGACAAGGGCACGGT NRXN3 ACGCGGAACTACATCAGCAA AGAGCGCTAAGTGTGTGCTT OCIAD2 TCATCATGGCTTCAGCGTCT CGTGCAACTTTGGGCAATGA OR2M4 CATAGTGCTTGCAGCTGTCC GCAGCACCGTAGTAGAGTCC SAMD9 CAGTTTGGTATCAGAATGGCAAAG TGGCTCTCGAATGCACTTCTT SLAMF7 GGAAGATCCAGCAAATACGGTT GCCCTGGGATGGAATGAAAC

74

REFERENCES

Angelov, D., Molla, A., Perche, P.Y., Hans, F., Cote, J., Khochbin, S., Bouvet, P., and Dimitrov, S. (2003). The Histone Variant MacroH2A Interferes with Transcription Factor Binding and SWI/SNF Nucleosome Remodeling. Mol Cell 11, 1033-1041.

Arents, G., Burlingame, R.W., Wang, B.C., Love, W.E., and Moudrianakis, E.N. (1991). The nucleosomal core histone octamer at 3.1 A resolution: a tripartite protein assembly and a left- handed superhelix. Proceedings of the National Academy of Sciences of the United States of America 88, 10148-10152.

Bannister, A.J., Zegerman, P., Partridge, J.F., Miska, E.A., Thomas, J.O., Allshire, R.C., and Kouzarides, T. (2001). Selective recognition of methylated lysine 9 on histone H3 by the HP1 chromo domain. Nature 410, 120-124.

Bernstein, E., Duncan, E.M., Masui, O., Gil, J., Heard, E., and Allis, C.D. (2006). Mouse polycomb proteins bind differentially to methylated histone H3 and RNA and are enriched in facultative heterochromatin. Molecular and cellular biology 26, 2560-2569.

Bodnar, A.G., Ouellette, M., Frolkis, M., Holt, S.E., Chiu, C.P., Morin, G.B., Harley, C.B., Shay, J.W., Lichtsteiner, S., and Wright, W.E. (1998). Extension of life-span by introduction of telomerase into normal human cells. Science 279, 349-352.

Brockdorff, N., Ashworth, A., Kay, G.F., McCabe, V.M., Norris, D.P., Cooper, P.J., Swift, S., and Rastan, S. (1992). The product of the mouse Xist gene is a 15 kb inactive X-specific transcript containing no conserved ORF and located in the nucleus. Cell 71, 515-526.

Brown, C.J., Hendrich, B.D., Rupert, J.L., Lafreniere, R.G., Xing, Y., Lawrence, J., and Willard, H.F. (1992). The human XIST gene: analysis of a 17 kb inactive X-specific RNA that contains conserved repeats and is highly localized within the nucleus. Cell 71, 527-542.

Cao, R., Wang, L., Wang, H., Xia, L., Erdjument-Bromage, H., Tempst, P., Jones, R.S., and Zhang, Y. (2002). Role of histone H3 lysine 27 methylation in Polycomb-group silencing. Science 298, 1039-1043.

Carrel, L., and Willard, H.F. (2005). X-inactivation profile reveals extensive variability in X- linked gene expression in females. Nature 434, 400-404.

Carroll, D. (2011). Genome engineering with zinc-finger nucleases. Genetics 188, 773-782.

Cermak, T., Doyle, E.L., Christian, M., Wang, L., Zhang, Y., Schmidt, C., Baller, J.A., Somia, N.V., Bogdanove, A.J., and Voytas, D.F. (2011). Efficient design and assembly of custom TALEN and other TAL effector-based constructs for DNA targeting. Nucleic acids research 39, e82.

75

Chadwick, B.P., Valley, C.M., and Willard, H.F. (2001). Histone variant macroH2A contains two distinct macrochromatin domains capable of directing macroH2A to the inactive X chromosome. Nucleic acids research 29, 2699-2705.

Chadwick, B.P., and Willard, H.F. (2001a). Histone H2A variants and the inactive X chromosome: identification of a second macroH2A variant. Hum Mol Genet 10, 1101-1013.

Chadwick, B.P., and Willard, H.F. (2001b). A novel chromatin protein, distantly related to histone H2A, is largely excluded from the inactive X chromosome. J Cell Biol 152, 375-384.

Chadwick, B.P., and Willard, H.F. (2002). Cell cycle-dependent localization of macroH2A in chromatin of the inactive X chromosome. The Journal of cell biology 157, 1113-1123.

Chadwick, B.P., and Willard, H.F. (2003). Chromatin of the Barr body: histone and non-histone proteins associated with or excluded from the inactive X chromosome. Human molecular genetics 12, 2167-2178.

Chadwick, B.P., and Willard, H.F. (2004). Multiple spatially distinct types of facultative heterochromatin on the human inactive X chromosome. Proceedings of the National Academy of Sciences of the United States of America 101, 17450-17455.

Clemson, C.M., McNeil, J.A., Willard, H.F., and Lawrence, J.B. (1996). XIST RNA paints the inactive X chromosome at interphase: evidence for a novel RNA involved in nuclear/chromosome structure. The Journal of cell biology 132, 259-275.

Costanzi, C., and Pehrson, J.R. (1998). Histone macroH2A1 is concentrated in the inactive X chromosome of female mammals. Nature 393, 599-601.

Costanzi, C., and Pehrson, J.R. (2001). MacroH2A2, a new member of the MacroH2A core histone family. J Biol Chem 276, 21776-21784.

Csankovszki, G., Panning, B., Bates, B., Pehrson, J.R., and Jaenisch, R. (1999). Conditional deletion of Xist disrupts histone macroH2A localization but not maintenance of X inactivation. Nat Genet 22, 323-324.

Doyen, C.M., An, W., Angelov, D., Bondarenko, V., Mietton, F., Studitsky, V.M., Hamiche, A., Roeder, R.G., Bouvet, P., and Dimitrov, S. (2006). Mechanism of polymerase II transcription repression by the histone variant macroH2A. Molecular and cellular biology 26, 1156-1164.

Feijs, K.L., Forst, A.H., Verheugd, P., and Luscher, B. (2013). Macrodomain-containing proteins: regulating new intracellular functions of mono(ADP-ribosyl)ation. Nature reviews Molecular cell biology 14, 445-453.

Gaspar-Maia, A., Qadeer, Z.A., Hasson, D., Ratnakumar, K., Leu, N.A., Leroy, G., Liu, S., Costanzi, C., Valle-Garcia, D., Schaniel, C., et al. (2013). MacroH2A histone variants act as a barrier upon reprogramming towards pluripotency. Nature communications 4, 1565.

76

Guenther, M.G., Levine, S.S., Boyer, L.A., Jaenisch, R., and Young, R.A. (2007). A chromatin landmark and transcription initiation at most promoters in human cells. Cell 130, 77-88.

Gunjan, A., Paik, J., and Verreault, A. (2005). Regulation of histone synthesis and nucleosome assembly. Biochimie 87, 625-635.

Hatch, C.L., and Bonner, W.M. (1988). Sequence of cDNAs for mammalian H2A.Z, an evolutionarily diverged but highly conserved basal histone H2A isoprotein species. Nucleic acids research 16, 1113-1124.

Joung, J.K., and Sander, J.D. (2013). TALENs: a widely applicable technology for targeted genome editing. Nature reviews Molecular cell biology 14, 49-55.

Kornberg, R.D. (1974). Chromatin structure: a repeating unit of histones and DNA. Science 184, 868-871.

Lachner, M., O'Carroll, D., Rea, S., Mechtler, K., and Jenuwein, T. (2001). Methylation of histone H3 lysine 9 creates a binding site for HP1 proteins. Nature 410, 116-120.

Landschulz, W.H., Johnson, P.F., and McKnight, S.L. (1988). The leucine zipper: a hypothetical structure common to a new class of DNA binding proteins. Science 240, 1759-1764.

Lee, J.T., and Bartolomei, M.S. (2013). X-inactivation, imprinting, and long noncoding RNAs in health and disease. Cell 152, 1308-1323.

Lieber, M.R. (2010). The mechanism of double-strand DNA break repair by the nonhomologous DNA end-joining pathway. Annual review of biochemistry 79, 181-211.

Lyon, M.F. (1961). Gene action in the X-chromosome of the mouse (Mus musculus L.). Nature 190, 372-373.

Maeder, M.L., Thibodeau-Beganny, S., Osiak, A., Wright, D.A., Anthony, R.M., Eichtinger, M., Jiang, T., Foley, J.E., Winfrey, R.J., Townsend, J.A., et al. (2008). Rapid "open-source" engineering of customized zinc-finger nucleases for highly efficient gene modification. Mol Cell 31, 294-301.

Maeder, M.L., Thibodeau-Beganny, S., Sander, J.D., Voytas, D.F., and Joung, J.K. (2009). Oligomerized pool engineering (OPEN): an 'open-source' protocol for making customized zinc- finger arrays. Nat Protoc 4, 1471-1501.

Mannironi, C., Bonner, W.M., and Hatch, C.L. (1989). H2A.X. a histone isoprotein with a conserved C-terminal sequence, is encoded by a novel mRNA with both DNA replication type and poly A 3'-processing signals. Nucleic Acids Res 17, 9113-9126.

Nozawa, R.S., Nagao, K., Igami, K.T., Shibata, S., Shirai, N., Nozaki, N., Sado, T., Kimura, H., and Obuse, C. (2013). Human inactive X chromosome is compacted through a PRC2- independent SMCHD1-HBiX1 pathway. Nature structural & molecular biology 20, 566-573.

77

Pehrson, J.R., and Fried, V.A. (1992). MacroH2A, a core histone containing a large nonhistone region. Science 257, 1398-1400.

Plath, K., Fang, J., Mlynarczyk-Evans, S.K., Cao, R., Worringer, K.A., Wang, H., de la Cruz, C.C., Otte, A.P., Panning, B., and Zhang, Y. (2003). Role of histone H3 lysine 27 methylation in X inactivation. Science 300, 131-135.

Qiu, P., Shandilya, H., D'Alessio, J.M., O'Connor, K., Durocher, J., and Gerard, G.F. (2004). Mutation detection using Surveyor nuclease. BioTechniques 36, 702-707.

Reyon, D., Tsai, S.Q., Khayter, C., Foden, J.A., Sander, J.D., and Joung, J.K. (2012). FLASH assembly of TALENs for high-throughput genome editing. Nature biotechnology 30, 460-465.

Rosenthal, F., Feijs, K.L., Frugier, E., Bonalli, M., Forst, A.H., Imhof, R., Winkler, H.C., Fischer, D., Caflisch, A., Hassa, P.O., et al. (2013). Macrodomain-containing proteins are new mono-ADP-ribosylhydrolases. Nature structural & molecular biology 20, 502-507.

Sander, J.D., Maeder, M.L., Reyon, D., Voytas, D.F., Joung, J.K., and Dobbs, D. (2010). ZiFiT (Zinc Finger Targeter): an updated zinc finger engineering tool. Nucleic acids research 38, W462-468.

Sander, J.D., Zaback, P., Joung, J.K., Voytas, D.F., and Dobbs, D. (2007). Zinc Finger Targeter (ZiFiT): an engineered zinc finger/target site design tool. Nucleic acids research 35, W599-605.

Sarma, K., and Reinberg, D. (2005). Histone variants meet their match. Nature reviews Molecular cell biology 6, 139-149.

Sporn, J.C., and Jung, B. (2012). Differential regulation and predictive potential of MacroH2A1 isoforms in colon cancer. The American journal of pathology 180, 2516-2526.

Talbert, P.B., and Henikoff, S. (2010). Histone variants--ancient wrap artists of the epigenome. Nature reviews Molecular cell biology 11, 264-275.

Topaloglu, O., Hurley, P.J., Yildirim, O., Civin, C.I., and Bunz, F. (2005). Improved methods for the generation of human gene knockout and knockin cell lines. Nucleic acids research 33, e158.

Zentner, G.E., and Henikoff, S. (2013). Regulation of nucleosome dynamics by histone modifications. Nature structural & molecular biology 20, 259-266.

78

BIOGRAPHICAL SKETCH

Education/Training

DEGREE INSTITUTION AND LOCATION MM/YY FIELD OF STUDY (if applicable) Florida State University, B.S. 07/05 – 05/09 Biological Science Tallahassee FL Florida State University, 07/09 – 08/11 Cell and Molecular Biology Tallahassee FL Florida State University, M.S. 08/11 – 05/14 Cell and Molecular Biology Tallahassee FL

Personal Statement I have always been fascinated with how things work. Since childhood, I have been taking things apart attempting to figure them out. After suffering for many years during my childhood with an undiagnosed congenital defect of my ureter, I naturally became interested in the molecular mechanisms of disease, and I was determined to find ways to prevent diseases from being passed on to my children. Since then, I’ve known that I wanted to have an active role in genetic engineering. The work I’ve done in the Chadwick Lab has been extremely valuable in allowing me to become proficient in these goals. This includes gene targeting in live human cells. I hope that these skills help me one day assist in curing disease, especially childhood diseases.

Positions and Employment 2009 Laboratory Assistant, Yu Lab, Biological Science, Florida State University, FL 2009-2011 Laboratory Technician, Chadwick Lab, Biological Science, Florida State University, FL 2011-2014 Graduate Student, Chadwick Lab, Biological Science, Florida State University, FL

79

Selected Peer-Reviewed Publications Horakova, A. H., S. C. Moseley, C. R. McLaughlin, D. C. Tremblay, and B. P. Chadwick. 2012. The macrosatellite DXZ4 mediates CTCF-dependent long-range intrachromosomal interactions on the human inactive X chromosome. Human Molecular Genetics. 21(20):4367-4377

Moseley, S. C., R. Rizkallah, D. C. Tremblay, B. R. Anderson, M. M. Hurt, and B. P. Chadwick. 2012. YY1 associates with the macrosatellite DXZ4 on the inactive X chromosome and binds with CTCF to a hypomethylated form in some male carcinomas. Nucleic Acids Research 40:1596–1608.

Tremblay, D. C., S. Moseley, and B. P. Chadwick. 2011. Variation in array size, monomer composition and expression of the macrosatellite DXZ4. PLoS ONE 6:e18969.

Tremblay, D. C., G. Alexander, Jr., S. Moseley, and B. P. Chadwick. 2010. Expression, tandem repeat copy number variation and stability of four macrosatellite arrays in the . BMC Genomics 11:632.

80