The Highly Parallel Homology Directed Repair Assay and the Analysis of BRCA1 Variants

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Muhtadi Muhammad Islam, B.E.

Biomedical Sciences Graduate Program

The Ohio State University

2016

Dissertation Committee:

Jeffrey Parvin, MD, PhD, Advisor

Michael Freitas, PhD

Kun Huang, PhD

Amanda Toland, PhD

Copyrighted by

Muhtadi Muhammad Islam

2016 Abstract

DNA damage and repair are processes that have been linked to cancer, both in carcinogenesis and in the treatment of cancer. We explore two different DNA repair factors, BRCA1 and HDAC10, in two distinct studies. BRCA1 is a highly penetrant gene in both breast and ovarian cancer and with several functions. We analyze BRCA1 in a novel assay that scores the homologous recombination repair function of several thousand BRCA1 variants in parallel. Our goal is to create a structure function map of

BRCA1 at the amino acid resolution to not only prove the novel highly parallel functional assay, but also to provide clinically relevant data of BRCA1 variants with respect to homologous recombination repair. HDAC10 is a histone deacetylase with only one known function, stimulating homologous recombination repair of double strand breaks.

Utilizing genome databases, we found that HDAC10 is deleted in up to 10% of ovarian tumors. We also found that low HDAC10 expression correlated with platinum therapy sensitivity. We demonstrated general HDAC inhibition enhanced cisplatin therapy in primary ovarian carcinomas. Additionally, HDAC10 specific inhibition sensitized a BRCA1 deficient ovarian cancer cell line to both DNA damage and cisplatin therapy. We suggest that HDAC10 is an important agent in the mechanisms of general HDAC inhibition as well as a potential new target for BRCA1 deficient ovarian tumors.

Dedication

This document is dedicated to the people and families who are struggling with cancer.

iii

Acknowledgements

The primary driving force who supported me throughout my research was Dr.

Jeffrey Parvin. His support was unwavering throughout the years, despite the few triumphs amongst the many setbacks. My committee has provided me with invaluable feedback and regularly reinvigorated my drive through their excitement at the potential of my work. My colleagues have been a continual source of both technical and emotional support, something I will dearly miss. My parents and sisters have been remarkably patient with me during these years and always believed in my ability even when I had my doubts. Finally, to my wife, who has been my partner in all my trials and tribulations these last few years. Andria has done so much more than simply share my burdens, she transformed them time and again so they were less heavy for both of us.

This is something I am grateful for beyond words. Thank you all.

Vita

May 2003 ...... Centerville High School

April 2007 ...... B.E. Chemical Engineering

University of Dayton

August 2007-present ...... Graduate Research Assistant

Department of Biomedical Informatics

The Ohio State University

Publications

Zhang, J., Lu, K., Xiang, Y., Islam, M., Kotian, S., Kais, Z., ... & Huang, K. (2012). Weighted frequent gene co-expression network mining to identify genes involved in genome stability. PLoS Comput Biol, 8(8), e1002656. Starita, L. M., Young, D. L., Islam, M., Kitzman, J. O., Gullingsrud, J., Hause, R. J., ... & Fields, S. (2015). Massively parallel functional analysis of BRCA1 RING domain variants. Genetics, 200(2), 413-422.

Field of Study

Major Field: Biomedical Sciences

Table of Contents

Abstract ...... ii

Dedication ...... iii

Acknowledgements ...... iv

Vita ...... v

List of Figures ...... ix

List of Tables...... xi

Chapter 1: A Review of DNA Repair and Cancer ...... 1

1.1 A Recent History of Molecular Biology ...... 2

1.2 BRCA1 and Breast Cancer ...... 4

1.3 The link between DNA damage and cancer ...... 7

1.4 DNA Repair review ...... 9

1.4a Non-homologous End Joining ...... 14

1.4b Homologous Recombination ...... 15

1.5 BRCA1 ...... 17

1.6 Cancer treatment ...... 20

1.7 Summary of studies presented ...... 23

Chapter 2: Highly Parallel Homology Directed Repair ...... 25

2.1 Abstract ...... 26 vi

2.2 Introduction...... 26

2.3 Materials and Methods ...... 28

2.3a Homology Directed Repair Assay ...... 28

2.3b HeLa DR-FRT generation ...... 30

2.3c HeLa BRCA1-variant cell line generation ...... 34

2.3d Highly Parallel Homology Directed Repair Assay ...... 35

2.3e Construction of barcoded variant libraries of BRCA1 ...... 36

2.3f Sequencing to link the BRCA1 N-terminal variants to their barcode ...... 37

2.3g Sequencing barcodes from genomic DNA ...... 38

2.3h Determining a functional score for each BRCA1 variant ...... 39

2.3i Statistics ...... 42

2.4 Validation Results ...... 42

2.5 BRCA1 variants ...... 51

2.6 Discussion ...... 57

Chapter 3: HDAC10 as a potential therapeutic target in ovarian cancer ...... 61

3.1 Abstract ...... 62

3.2 Introduction...... 62

3.3 Materials and Methods ...... 64

3.3a Cell Culture and Reagents ...... 64

3.3b Homology Directed Repair Assay (HDR) ...... 65

3.3c Comet Assay ...... 65

vii

3.3d MTT Assay ...... 67

3.3e Statistics ...... 67

3.4 Results ...... 68

3.5 Discussion ...... 80

3.6 Tables ...... 85

Chapter 4: Summary and Future Directions ...... 86

4.1 DNA repair and cancer ...... 87

4.2 BRCA1 ...... 88

4.2a Highly Parallel Homology Directed Repair ...... 88

4.2b BRCA1 Future Directions ...... 89

4.3 HDAC10 ...... 91

4.3a HDAC10 Results ...... 91

4.3b HDAC10 Future Directions ...... 93

4.4 BRCA1 and HDAC10 ...... 93

References...... 95

Appendix A: HeLa DR-[Library] Generation Protocol ...... 106

Appendix B: Highly Parallel Homology Directed Repair Assay (HP-HDR) Protocol ...... 112

viii

List of Figures

Figure 1 Breast cancer incidence ...... 5

Figure 2 Breast cancer stage and survival ...... 6

Figure 3 Simplified DSB repair ...... 12

Figure 4 BRCA1 major domains ...... 19

Figure 5 Homology Directed Repair Assay ...... 29

Figure 6 Flp-FRT recombination and HP-HDR ...... 32

Figure 7 β-Galactosidase activity of HeLa DR-FRT ...... 33

Figure 8 HDR of single integration HeLa BRCA1 cells ...... 43

Figure 9 HeLa BRCA1 WT and C61G HDR and sort ...... 45

Figure 10 Barcode WT BRCA1 Integration Efficiency ...... 48

Figure 11 Comparison of mixed HP-HDR and single HDR ...... 50

Figure 12 BRCA1 HDR and Domains ...... 52

Figure 13 BRCA1 variants clinical and HDR comparison ...... 53

Figure 14 Error Prone library 2nd integration overview ...... 56

Figure 15 HDAC10 and platinum status in ovarian cancer patients ...... 70

Figure 16 HDR of cisplatin and HDAC inhibitors in combination ...... 72

Figure 17 UWB1.289 Ovarian cells comet assay with HDAC10 depletion ...... 74-75

Figure 18 MTT Assay of UWB1.289 cells treated with cisplatin ...... 77 ix

Figure 19 MTT assay in primary ovarian cancer cells ...... 79

List of Tables

Table 1 Breast cancer compared to common types of cancer ...... 5

Table 2 Nucleic acid oligomer sequences used in HP-HDR ...... 41

Table 3 HeLa WT BRCA1 Barcode Colony Sequences ...... 47

Table 4 siRNA sequences for HDAC10 study ...... 85

Chapter 1: A Review of DNA Repair and Cancer

1.1 A Recent History of Molecular Biology

Molecular biology has experienced a number of breakthrough discoveries and developments over the last century. Through a continuous and concerted effort to identify the structures and sequences of subcellular components, the field of molecular biology has expanded from identifying cells to organelles to central dogma. In the latest chapter, the natural evolution of available technology has created a divergence.

Molecular biology revolutionized biological sciences starting in the 1930s with the dissection of pathways, one molecule at a time. Now, the development and potential of big data analyses are a paradigm shift away from the traditional testing of single pathways and functions and the first step into the realm of massively parallel studies.

The first player in the bioinformatics field was naturally genomics, highlighted by the human genome project (HGP) (NHGRI, 2015). Completed in 2003, the data from

HGP and other genomic studies were analyzed in genome wide association studies

(GWAS). These studies associate variants in the genome with disease (Ku, Loy, Pawitan,

& Chia, 2010). Although there have been a number of important findings, the discovery rates have been lower than expected (Ku et al., 2010; Maher, 2008). The next step was to move beyond DNA studies and into the downstream, more function based units of the cell, such as transcriptome sequencing and proteomics, as well as meta-studies compiling the findings from many different informatics studies. For example, the cancer genome atlas (TCGA) is a concerted effort to compile DNA, RNA, protein, and clinical annotations into a centralized and public database (National Cancer Institute, 2016).

While the bioinformatics studies continue to develop, the traditional pathways continued to progress. New tools emerged, such as yeast two-hybrid, RNA silencing, chromatin immunoprecipitation (ChIP), any number of functional assays, CRISPR editing, and many others (Fields & Song, 1989; Gilmour & Lis, 1984; Jinek et al., 2012; R. C. Lee,

Feinbaum, & Ambros, 1993; Mojica, Diez-Villasenor, Garcia-Martinez, & Soria, 2005;

Pierce, Johnson, Thompson, & Jasin, 1999; Wightman, Ha, & Ruvkun, 1993). However, the scope remained typically in single variant studies. Studying another variant necessitated a biological replicate, a laborious process that prohibits studying more than a handful of variants in any given experiment.

The scope and depth of informatics analyses and functional assays are complementary. Informatics focuses on widening the scope of a study to every molecule but with less depth, while functional assays test the effect a variant has on a physiological function, but only a single variant at a time. Technologies developed in the informatics fields can potentially parallelize the traditional single variant functional assays. Similarly, using the framework of a single variant experiment can finally bring functional and potential clinical associations to the results of informatics technologies.

Just in the past few years, we have developed such tools to test variants. We have successfully tested several thousands of variants in a yeast two-hybrid protein-protein interaction assay and a bacteriophage enzyme activity assay (Starita et al., 2015). These new studies analyze the function of breast cancer 1 (BRCA1) variants, which serves as the starting point for this study. As its name hints, certain mutations of BRCA1 are

3 major risk factors for the development of breast cancer. A study that directly links

BRCA1 functions to breast cancer could prove invaluable to the management of a major disease.

1.2 BRCA1 and Breast Cancer

A primary driving factor for using BRCA1 as the protein of interest in this study is two-fold: BRCA1 has a robust DNA repair function as well as a strong clinical link with breast cancer (Futreal et al., 1994; Ransburgh, Chiba, Ishioka, Toland, & Parvin, 2010).

Excluding non-melanoma skin cancer, breast cancer has overtaken all other cancers to become the most diagnosed cancer in the U.S., despite less than 1% of breast cancer cases occurring in men (Table 1 and Figure 1) (National Cancer Institute, 2015). The lifetime risk of a woman developing breast cancer is 1 in 8, substantially greater than any other cancer. Additionally, breast cancer is the second most common cause of death from cancer in most women, and the most common cause for death from cancer among Hispanic women (CDC, 2012). Early detection, as with many cancers, is important for treatment options and survival (Figure 2) (National Cancer Institute,

2015).

Estimated New Estimated Deaths Common Types of Cancer Cases 2015 2015 1. Breast Cancer (Female) 231,840 40,290 2. Lung and Bronchus Cancer 221,200 158,040 3. Prostate Cancer 220,800 27,540 4. Colon and Rectum Cancer 132,700 49,700 5. Bladder Cancer 74,000 16,000 6. Melanoma of the Skin 73,870 9,940 7. Non-Hodgkin Lymphoma 71,850 19,790 8. Thyroid Cancer 62,450 1,950 9. Kidney and Renal Pelvis Cancer 61,560 14,080 10. Endometrial Cancer 54,870 10,170 Table 1 Breast cancer compared to common types of cancer Despite the vast majority of breast cancers occurring in women, breast cancer has become the most commonly diagnosed cancer. (National Cancer Institute, 2015)

Figure 1 Breast cancer incidence Breast cancer accounts for 14% of all new cancer diagnoses, more than any other cancer. (National Cancer Institute, 2015)

Figure 2 Breast cancer stage and survival A. The majority of breast cancers are localized or regional when diagnosed. B. Localized and regional breast cancers have much better survival than metastasized cancer (National Cancer Institute, 2015).

Although most of breast cancer cases are sporadic, 5-10% are hereditary

(Campeau, Foulkes, & Tischkowitz, 2008). Mutations in the BRCA1 and BRCA2 genes account for 20-25% of hereditary breast cancer (Easton, 1999). Just as important, the penetrance of a harmful BRCA1 mutation is very high, conferring a 55-65% risk of developing breast cancer by an age of 70 years (Antoniou et al., 2003; S. Chen &

Parmigiani, 2007). The population that carries a harmful BRCA1 mutation has the unique option of prophylactically preventing breast cancer. One treatment, bilateral prophylactic mastectomies, surgically removes as much of the at risk tissue as possible.

As with any surgery, it is invasive, but the effectiveness of the procedure could be as high as a 90% reduction of risk (Hartmann et al., 1999). This prophylactic plan, however, only applies to mutations that are known to carry significant cancer risk. A major goal of this study is to provide decision-making data to the remaining variants of uncertain significance (VUS) as well as synthetic variants by studying the DNA repair function of each variant.

1.3 The link between DNA damage and cancer

The logical understanding of the link between DNA damage and cancer has been understood for a long time. In fact, even before the structure of DNA was known, abnormal distributions of chromosomes in cancer cells led to a theory that a chromosomal change could be responsible for carcinogenesis (Boveri, 2008). While

Boveri made this discovery in 1902, the English translation was published in 2008.

Almost immediately after the seminal paper by Watson and Crick on the structure of

DNA, there was speculation that changes in DNA could be responsible for cancer

(Burdette, 1955; Watson & Crick, 1953). Although the link between DNA repair and cancer was theorized very early, it was not until later that evidence supported the proposed causal nature of DNA damage and repair to cancer. The simplest evidence of this link is the association of tumor suppressors and DNA repair factors: BRCA1, BRCA2, p53, among others, (Futreal et al., 1994; Kastan et al., 1992; Patel et al., 1998; Scully,

Chen, Ochs, et al., 1997). On the mutagenesis side, one of the driving forces for mutation is DNA damage, particularly when repaired using a more error-prone pathway.

The conclusion is that DNA damage can lead to mutation, which can lead to cancer. The relationship between DNA damage and cancer is easily seen in the association of mutagens with cancer, such as UV exposure or cigarette smoke (Cleaver, 1968;

Leanderson & Tagesson, 1992). Many important mutations linked with cancer are found in genes that encode proteins that are involved in the DNA response or repair pathways. One of the key discoveries of this association is the identification of the proteins important in Xeroderma Pigmentosum. Xeroderma Pigmentosum patients were discovered to be unable to repair UV damage, which could explain their extremely high incidence of skin cancer, but not other cancers (Cleaver, 1969). The ensuing discovery that the genetic changes of Xeroderma Pigmentosum, mutations in the XPA,

XPB, XPC, ERCC4, and other genes involved in the nucleotide-excision repair pathway, provide a compelling argument that DNA damage is the direct cause of the incredibly high rate of skin cancer in these patients (Lehman, 2011). Although many new

8 associations between DNA repair and cancer are being discovered, disease penetrance for any one variant is often relatively low, possibly due to redundancies in tumor suppressor functions (Tomlinson, Houlston, Montgomery, Sieber, & Dunlop, 2012).

However, in the case of BRCA1, the fairly high penetrance and large section of familial cancers means there is a unique opportunity to link one or more of its functions to cancer directly. As shown later, we have data that suggests BRCA1 mediated DNA repair is the major function linked to BRCA1 cancer risk, another prompt to explore its structure and function in detail.

1.4 DNA Repair review

The function of interest for BRCA1 is homologous recombination, one of several

DNA double-strand break repair pathways. To understand the impact of homologous recombination repair it is important to understand its context in the DNA damage response system.

DNA damage can occur from a variety of factors. Some DNA damage occurs intrinsically, such as from replication stress. Extrinsic factors are also a major source of

DNA damage. Prominent sources of DNA damage from external sources are UV or ionizing radiation, for example, X-ray radiation (Ciccia & Elledge, 2010). UV damages the

DNA by forming pyrimidine dimers. Ionizing radiation, unlike UV, generates double stranded breaks (DSB). Double strand breaks can also occur from unresolved single stranded breaks that the cell tries to replicate, as well as from free radicals and other sources (Khanna & Jackson, 2001). Single strand repair generally involves a system of 9 detecting the error, removing the problematic bases and/or structures, and then use the second DNA strand as a template to resolve the error with no loss in sequence integrity

(Sancar, Lindsey-Boltz, Unsal-Kacmaz, & Linn, 2004). Double strand breaks are mechanistically more challenging than a single strand lesion. Single strand lesions can be repaired using the complementary strand as a template, but without a complementary strand available in DSBs, cells must utilize repair machinery to either repair the break directly, carrying risk of a mutation, or find a homologous double stranded DNA template to provide sequence data for both broken strands (Goodarzi &

Jeggo, 2013). The major repair mechanisms that correct for double strand breaks are known as non-homologous end joining (NHEJ), which does not utilize a sequence template, and homologous recombination (HR), which does (Figure 3). Although homologous recombination can refer to other cellular processes not related to DSB repair, such as those which occur during meiosis, this study refers to the repair mechanism of HR. In addition to HR and NHEJ, there at least two more double strand break repair pathways: microhomology mediated non-homologous end joining (MMEJ, also known as alt-NHEJ) and single strand annealing (SSA) (Ciccia & Elledge, 2010;

Goodarzi & Jeggo, 2013). MMEJ and SSA are sometimes classified as alternatives to

NHEJ and HR, respectively, and although the mechanistic pathways do overlap, this study will treat them as distinct repair pathways. Since HR can use an undamaged sister chromatid as a template in its repair process, it is less error prone than the other pathways (Moynahan & Jasin, 2010). How cells determine the mechanism with which to

10 repair double strand breaks is still actively being investigated, but the presence of an undamaged sister chromatid is important to utilize the HR pathway (Johnson & Jasin,

2000, 2001).

The first step of each pathway is to detect the double strand break. As shown below, detection at least a partially redundant function with multiple complexes able to detect, localize, and begin recruitment of downstream factors. Additionally, the complexes can be associated with a specific repair pathway. The MMEJ and SSA pathways share some early DSB detection mechanisms with HR, but NHEJ detection occurs independently and nearly simultaneously. Importantly, though early factors for

HR and NHEJ can be present at the DSB, the cell does not commit to any particular repair pathway until later (J. S. Kim et al., 2005).

Figure 3 Simplified DSB repair This figure is a simplified diagram of the two major DSB repair pathways. A. DSBs repaired via HR involve a complex system of detection, end resection, and strand invasion. HR conserves sequence integrity due to the presence of an undamaged sister chromatid template. B. NHEJ involves detection and end- protection to repair DSBs. NHEJ is more error prone than HR. Adapted from (Ciccia & Elledge, 2010).

The DSB detection step for HR has previously been associated with ataxia telangiectasia mutated (ATM), but more recent evidence shows that other factors play upstream of ATM, including Poly(ADP-Ribose) Polymerase 1 (PARP1) and the Mre11-

Rad50-Nbs1 (MRN) complex (J. S. Kim et al., 2005; J. H. Lee & Paull, 2005; Schreiber,

Dantzer, Ame, & de Murcia, 2006). Historically, the ATM gene was identified first due to its mutation association in patients with the disease ataxia telangiectasia, who have a predisposition to cancer and radiosensitivity (Savitsky et al., 1995). The association of ataxia telangiectasia and cancer was a critical factor that lead to the eventual study of the repair function of ATM and downstream effectors (Harper & Elledge, 2007). More recent studies suggest that it is actually the PARP and MRN complex that does the initial

DSB detection for HR, MMEJ, and SSA. The HR and SSA pathways are then mediated by the recruitment of ATM and its signaling cascade (Haince et al., 2007; J. H. Lee & Paull,

2005). Once ATM has been activated, there are a number of factors that are immediately recruited to the double strand break. Checkpoint kinase 2 (Chk2), a cell cycle regulator and tumor suppressor is phosphorylated in complex with ATM (van Gent,

Hoeijmakers, & Kanaar, 2001). Additionally, H2AX (member X of the histone H2A family) involved in recruitment is also phosphorylated by ATM at the site of the break, forming

H2AX (Paull et al., 2000).

Independently, the early NHEJ factors are also recruited to a DSB site, however, these factors do not commit the repair pathway to NHEJ. Ku proteins in complex with

DNA-PK are almost immediately present in response to double strand breaks, but their

13 presence is not deterministic of NHEJ (J. S. Kim et al., 2005). In an additional redundancy mechanism, DNA-PK phosphorylates H2AX as well (van Gent et al., 2001).

Another DSB detection protein, PARP1, is associated with HR, SSA, and MMEJ.

PARP1 function was originally associated with single stranded DNA repair, though it is now known to be involved in multiple double strand break repair pathways, in part via recruitment of BARD1 protein, as well as MRN recruitment and ATM phosphorylation

(Haince et al., 2007; Li & Yu, 2013; Sharma et al., 2015).

At this stage, the cell must commit to either NHEJ or one of the other three pathways. While the exact mechanism with which the repair pathway is selected is unknown, the cell cycle stage is very important. Cells are more likely to undergo homologous recombination in S phase and G2 phase and mostly undergo NHEJ in G1 phase (Saleh-Gohari & Helleday, 2004). The molecular mechanism is not perfectly understood, but sister chromatids only present in S and G2 phases provide an identical and undamaged double strand template for homologous recombination (Johnson &

Jasin, 2000). One known pathway mediator is p53 binding protein 1 (53BP1), which promotes NHEJ and inhibits HR (Bunting et al., 2010).

1.4a Non-homologous End Joining

The cell may not have the correct conditions to perform homologous recombination. Particularly if there are no sister chromatids, as seen in G1 phase cells, there is unlikely to be any acceptable sequence homology available nearby (Johnson &

Jasin, 2000; Saleh-Gohari & Helleday, 2004). In these cases, the cell typically undergoes

NHEJ. The basis of NHEJ is to simply recreate the bonds that were broken in the damage event. While this is a simpler approach, it runs a risk of ligating the DNA with one or more missing bases, as the damaging agent may have affected more than a single base pair (Moynahan & Jasin, 2010). As stated above, the NHEJ proteins Ku70 and Ku80 bind to the broken ends of the double strand break within seconds, where they begin to recruit other NHEJ specific proteins (J. S. Kim et al., 2005). The Ku proteins recruit and complex with DNA-PK, which activate the catalytic subunit DNA-PKc (Mahaney, Meek, &

Lees-Miller, 2009). DNA-PKc plays a critical role in stabilizing the DSB and preventing end resection, thus committing the cell to the NHEJ pathway (Meek, Dang, & Lees-

Miller, 2008). DNA-PKc then recruits XRCC4 and Ligase IV to line up the DNA strands to enzymatically recreate the broken phosphodiester bonds (Nick McElhinny, Snowden,

McCarville, & Ramsden, 2000). While carrying a risk of sequence data loss, this is the major DNA repair pathway of exogenous double strand breaks when there is no available template, such as in G1 phase cells (Moynahan & Jasin, 2010; Sancar et al.,

2004).

1.4b Homologous Recombination

If the cell has determined that NHEJ is not the preferred repair mechanism, the first step that eliminates the NHEJ pathway is the resection of some DNA to generate single stranded tails. MRN already present at the site recruits CtIP. CtIP, in conjunction with the Mre11 part of the MRN complex, begins resecting the 5’ strand of DNA on both

15 sides of the break (Williams, Williams, & Tainer, 2007; Zhong et al., 1999). If the resection is limited, the cell can undergo MMEJ. If BRCA1 has also been recruited to the

DSB and there is more extensive resection, the ssDNA tails are then stabilized by RPA

(Wold, 1997). At this stage, the DSB will be repaired with either the SSA or HR pathways. In SSA, RAD52 proteins directly anneal homologous ssDNA sequences (Stark,

Pierce, Oh, Pastink, & Jasin, 2004). In HR, RPA is replaced with Rad51 proteins to coat the single strand overhangs in complex with BRCA2, PALB2, and BRCA1 (S. C. West,

2003). The Rad51-ssDNA structure activates the ATR pathway and aligns with nearby double stranded DNA, which will serve as the template for the ensuing repair, a process known as strand invasion (Z. Chen, Yang, & Pavletich, 2008). At this point, the Rad51- ssDNA complex invades the intact double stranded DNA on both ends of the double strand break. They create a specialized double Holliday junction such that the intact double stranded DNA can serve as the template for both strands of the damaged DNA

(Holliday, 1964). This DNA complex involves four strands of DNA, two damaged strands, and two template strands. Enzymes such as DNA polymerase , DNA Ligase 1, GEN1, and MUS81 synthesize DNA past the break site and ligate each damaged strand to the resected strand on the other side of the break (X. B. Chen et al., 2001; Goetz, Motycka,

Han, Jasin, & Tomkinson, 2005; Ip et al., 2008; Maloisel, Fabre, & Gangloff, 2008). The formation can then resolve into two intact double stranded DNA molecules. If no crossover event occurs, this process conserves sequence integrity. In contrast, a mutation in the machinery of HR can compromise sequence integrity and confer

16 significant cancer risk. A potential example of this is BRCA1, where mutations causing loss of function could direct cells into a more error-prone pathway such as NHEJ, eventually leading to tumorigenesis.

1.5 BRCA1

BRCA1 has many known functions, including function as a DNA repair factor and an E3 ubiquitin ligase (Starita & Parvin, 2003). This particular project is focused on the

DNA repair function of BRCA1, namely its effect in homologous recombination. BRCA1 function in DNA repair is critical, demonstrated by a robust effect on a homology directed repair assay, which will be described later. Exactly how BRCA1 modulates its role in homologous recombination is likely multimodal. As mentioned earlier, BRCA1 is present in the complex that performs the 5’ resection committing the DSB to homologous recombination (Huen, Sy, & Chen, 2010; Zhong et al., 1999). Additionally, immunofluorescence studies show BRCA1 localizing near H2AX (Paull et al., 2000).

Another BRCA1 known interaction is with partner and localizer of BRCA2 (PALB2).

PALB2 importantly binds to BRCA2 and mediates the RAD51 response, responsible for stabilizing the ssDNA post end resection and the subsequent homology search in the HR repair of a DSB (Scully, Chen, Plug, et al., 1997; Sy, Huen, & Chen, 2009).

Though a role for BRCA1 in DNA repair and in cancer risk has been well established, studying the mutation variants of BRCA1 can be challenging. The BRCA1 protein contains 1863 amino acid residues. Variants that have uncertain clinical significance (VUS) have discovery rates are difficult to estimate, ranging between 17

Myriad’s reported 2.1% to as high as 15%; rates can even be over 20% when limited to specific populations (Eggington et al., 2014; Lindor, Goldgar, Tavtigian, Plon, & Couch,

2013). The VUS rates may seem low, but will number in the thousands for individuals tested each year. The challenge to provide relevant data for these variants is impractical using traditional techniques. This study aims to develop a new technique to target the BRCA1 regions where VUS are concentrated and study them with a per amino acid approach.

Figure 4 BRCA1 major domains The N-terminal RING domain interacts with BARD1. The central region of BRCA1 is a DNA binding domain (DBD). A coiled coil (CC) domain interacts with PALB2. Serine clustered domains (SCD) provides ATM phosphorylation sites. The BRCT domain at the C-terminus interacts with a number of proteins, including CtIP.

BRCA1 structure has two primary domains of interest for homologous recombination function, an N-terminal RING domain and a C-terminal BRCT domain

(Figure 4). The RING domain has known E3 ubiquitin ligase function, while the BRCT domain is involved in protein-protein interaction (Starita & Parvin, 2003). The RING domain complexes with BARD1 in a stable manner (Meza, Brzovic, King, & Klevit, 1999;

Yu, Chini, He, Mer, & Chen, 2003). Both domains, however, exhibit a high number of genetic variants that are high risk for cancer (Ransburgh et al., 2010; Starita et al., 2015).

Additionally, the coiled coil (CC) domain interacts with partner and localizer of BRCA2

(PALB2). A serine cluster domain (SCD) overlaps with the CC domain and provides phosphorylation sites for ATM (Roy, Chun, & Powell, 2012). The central region of BRCA1 has also been shown to bind to DNA (Paull, Cortez, Bowers, Elledge, & Gellert, 2001).

This study aims to comprehensively assess every RING domain variant, synthetic or otherwise, for function in HR.

1.6 Cancer treatment

Homologous recombination repair is importantly tied to the carcinogenesis of cancer. Paradoxically, repair mechanisms like homologous recombination are also targets for cancer therapy (Helleday, Petermann, Lundin, Hodgson, & Sharma, 2008).

There are many forms of cancer treatments: surgery, radiation, chemotherapy, immunotherapy, targeted, and more (Sudhakar, 2009). Drugs that target DNA are typically considered chemotherapy drugs (Chabner & Roberts, 2005; Helleday et al.,

2008). This study focuses on chemotherapy treatments.

Chemotherapy drugs typically attack proliferating cells directly. Since cancer cells have overcome host defense mechanisms, chemotherapy drugs rely on selectively targeting the mechanisms inherent to cancer cells, such as their uncontrolled proliferation. Although these drugs are selective, most chemotherapies do have significant toxicities to the patient (Chabner & Roberts, 2005). In fact, the earliest chemotherapy drug was nitrogen mustard used to treat non-Hodgkins lymphoma, using data obtained from the autopsies of soldiers who died from sulfur mustard gas during

World War I (GILMAN, 1963). Not long after, Sydney Farber and Harriett Kilte developed methotrexate, a folate analog (FARBER & DIAMOND, 1948). Still in use today, methotrexate was a major breakthrough in the treatment of cancer. For example, methotrexate was shown to cure choriocarcinoma, marking the first time a drug was able to cure a solid tumor (LI, HERTZ, & BERGENSTAL, 1958). At a basic level, methotrexate inhibits the DNA replication by inhibiting dihydrofolate reductase, a critical enzyme in the synthesis of thymidine (Jolivet, Cowan, Curt, Clendeninn, &

Chabner, 1983). Similar drugs have since been developed and discovered, such as 5- fluorouracil.

The next class of chemotherapy drugs was discovered from natural products.

Catharanthus roseus and other Vinca genus plants produce alkaloid compounds such as vincristine. These drugs inhibit microtubule polymerization, first discovered for their anti-tumor effects by Eli Lilly (Bensch & Malawista, 1968; JOHNSON, ARMSTRONG,

GORMAN, & BURNETT, 1963). Inhibiting microtubules disrupts mitosis mechanisms that

21 allows vincristine and similar drugs to select preferentially against rapidly dividing cancer cells. Similarly, paclitaxel, an antimitotic that inhibits microtubule disassembly, was derived from the bark of the Pacific yew tree. Another naturally derived anti-tumor drug is irinotecan, a camptothecin analog derived from a Chinese ornamental tree (Wall et al., 1966). Irinotecan inhibits topoisomerase I, an enzyme responsible for unwinding

DNA necessary for replication (Sawada, Yokokura, & Miyasaka, 1995). Several additional natural agents have been developed, often targeting similar processes.

Another breakthrough in cancer chemotherapy came from platinum compounds and the development of cisplatin. Barnett Rosenberg was studying the effects of platinum in bacteria and tested the compounds on mouse models, which demonstrated tumor inhibition (Rosenberg, VanCamp, Trosko, & Mansour, 1969). This was eventually translated into cisplatin, a potent anti-tumor agent (Bosl et al., 1986). Cisplatin is an intra-strand cross-linker of DNA. Additionally, the cisplatin molecules themselves act as bulky adducts, directly interfering with DNA polymerases as well as activating repair responses (Jung & Lippard, 2007). This DNA damage triggers multiple different pathways. Many of the pathways are involved in the DNA damage response and result in apoptosis of cancer cells. Although nucleotide excision repair is a major component of cisplatin response, this study focuses on the role of homologous recombination repair and targeting HR to enhance cisplatin sensitivity. Previous studies have shown that cancers and cells that are deficient in HR are sensitive to cisplatin treatment (Wang et al., 2005). Similarly, mouse models deficient in HR exhibited higher sensitivity to

22 cisplatin therapy (Raaphorst, Leblanc, & Li, 2005). This study explores the potential of inhibiting homologous recombination to enhance cisplatin sensitivity.

A relatively recent class of anti-tumor pharmaceuticals are histone deacetylase inhibitors (HDACi). Histone deacetylases (HDACs) are a class of enzymes that remove acetyl groups from lysine amino acids. However, they have a broad range of target substrates and function (Gregoretti, Lee, & Goodson, 2004; Spange, Wagner, Heinzel, &

Krämer, 2009). There are at least 18 different HDAC proteins, classified into five general classes: I, IIA, IIB, III, and IV. Class III HDACs do not require zinc for activity (Gregoretti et al., 2004). This study focuses on HDAC10. Although not much is known about HDAC10, it is established that HDAC10 plays a role in HR (Kotian, Liyanarachchi, Zelent, & Parvin,

2011). This study explores the potential of HDACi and HDAC10 inhibition to sensitize cancer cells to cisplatin therapy, possibly mediated through HDAC10 HR repair function.

1.7 Summary of studies presented

This dissertation explores two distinct studies that are linked through their basis in homologous recombination repair. In one study we will develop a new tool to simultaneously analyze many BRCA1 variants and their function in homology directed repair. Using the several known BRCA1 variants in both HR function as well as clinical risk of breast cancer, we wish to study how well correlated BRCA1 HR function and cancer risk truly are. Ultimately, we hope to provide at least HR data to patients who carry variants with unknown significance. We also hope to prove a novel technique of

23 analyzing protein structure-function that can be adapted to other proteins and/or functions.

The other study focuses on HDAC10 and its potential as a target for cancer therapy. HDACi are already used in certain cancers such as cutaneous T-cell lymphoma

(Bertino & Otterson, 2011). In this study we explore how HDACi might be used to enhance cisplatin therapy mediated through the effects of HDAC10 and HR.

Chapter 2: Highly Parallel Homology Directed Repair

Muhtadi M. Islam*1, Lea Starita*2, Tapahsama Banerjee1, Stanley Fields2, Jay Shendure2, and Jeffrey D. Parvin1

1The Ohio State University, Department of Biomedical Informatics 2The University of Washington, Department of Genome Sciences *Contributed equally

This chapter is finalizing results and analysis for submission for publication.

Author contributions: Muhtadi Islam and Tapahsama Banerjee adapted HeLa cells for variant library integration. Muhtadi Islam integrated BRCA1 variant libraries into adapted HeLa cells, cloned the expression vector for BRCA1, and performed homology directed repair assays and sorting. Lea Starita cloned barcoded BRCA1 variant libraries, generated sub-assembled links of the variant libraries to barcodes, performed DNA sequencing of genomic DNA from homology directed repair results, and analyzed sequencing data. Jeffrey Parvin, Stanley Fields, and Jay Shendure advised on experiments and results.

2.1 Abstract

We developed a novel technique for analyzing the structure function relationship of a given protein at the amino acid resolution. Utilizing a proven homology directed repair assay (HDR) and massively parallel DNA sequencing, we have adapted a functional assay designed to test a single variant into a highly parallel assay that can test several thousand variants simultaneously. When paired with efficient mutagenesis, the highly parallel homology directed repair assay (HP-HDR) can potentially generate a structure function map of a given protein and function at the amino acid resolution. We are using BRCA1, a DNA repair protein with hundreds of clinically important variants, to prove our HP-HDR assay, as well as provide function data for variants of BRCA1 that have unknown clinical significance.

2.2 Introduction

The scientific history of DNA has seen significant advances in just a few decades.

From first recognizing chromosomes as the code through which inheritance might be passed, to the seminal structural discovery of Watson and Crick, to Taq polymerase being discovered from bacteria in Yellowstone National Park, our ability to understand and manipulate DNA had already seen tremendous growth in just 30 years (Boveri,

2008; Chien, Edgar, & Trela, 1976; Watson & Crick, 1953). The next generation of technologies scaled DNA analysis capability by several orders of magnitude.

Spearheaded by the human genome project, the cost and efficiency of massively parallel

DNA sequencing dropped at an incredible rate (NHGRI, 2015). The affordability of

26 whole genome sequencing has enabled investigators to begin scanning the human genome for all types of genetic links to disease. The genome wide association studies

(GWAS) has generated hundreds to thousands of candidate variants that could be predictive of disease. A major shortcoming of found associations is that the variants are typically low penetrance (Ku et al., 2010). These variants of unknown significance (VUS), when there is not enough genetic evidence, present a challenge that goes beyond DNA.

As exome sequencing and proteomics attempts to apply the massively parallel techniques used in GWAS to RNA and proteins, testing protein function is a more direct approach to determine the impact of a given VUS.

The massively parallel fields have seen a rapid decline in cost and efficiency. In contrast, the functional tests available have grown in diversity. The ability to inhibit and/or express proteins enabled development of a number of functional assays testing anything from protein-protein interactions in yeast to editing genomes in model organisms (Fields & Song, 1989; Jinek et al., 2012). However, one thing has remained consistent: the functional assays are limited to test a handful of variants at a time.

Our goal is to apply the concepts of massively parallel genomics to the functional tests and adapt a high throughput functional assay. Specifically, this project aims to adapt a proven homology directed repair assay to be able to analyze thousands of variants simultaneously. Using a well-documented and penetrant gene, BRCA1, another major goal is to provide functional impact to the VUS in BRCA1. Additionally, the scope of the parallelized HDR goes beyond just VUS, which must exist in the population to be 27 identified, and provides functional impact to synthetic variants that may become VUS in the future.

2.3 Materials and Methods

2.3a Homology Directed Repair Assay

The HDR assay challenges a cell with a specific double strand DNA break that expresses functional green fluorescence protein (GFP) when repaired via homologous recombination. HeLa cells have been adapted with a stable integration of a DR-GFP construct (Ransburgh et al., 2010). The DR-GFP construct is two GFP sequences, neither of which can express a functional GFP (Pierce et al., 1999). The first GFP sequence contains an I-SceI cut site and a stop codon just downstream of the cut site. The second

GFP sequence is an incomplete GFP sequence, missing the 5’ end of the sequence.

Under the correct conditions, a cell repairing a double strand break at the I-SceI cut site will utilize the downstream GFP sequence as a template to repair the break, and replace the downstream stop codon with the original GFP sequence. This provides a reliable method with which to test cellular repair of DSBs with homologous recombination under various treatments (Figure 5).

Homology Directed Repair Assay

I-SceI site

GFP off

SceGFP iGFP

Deplete test protein (eg BRCA1) by RNAi; Express I-SceI to generate DSB; Homologous Recombination

GFP on

iGFP GFP+

Cells

control RNAi BRCA1 RNAi

Figure 5 Homology Directed Repair Assay Cells that are able to repair the DSB break generated by I-SceI express GFP. Cells where HR is inhibited, such as the depletion of BRCA1, do not express GFP.

The HR pathway has many important players, but few are as well documented or well known as BRCA1. BRCA1 is a highly penetrant gene that accounts for a relatively high share of hereditary breast cancer. Although BRCA1 has multiple functions, we focus on its effect in homologous recombination. Utilizing an siRNA targeting the 3’ untranslated region of the endogenous WT BRCA1, a 10-fold reduction is observed in the number of cells that convert to GFP-positive (Parvin, Chiba, & Ransburgh, 2011;

Towler et al., 2012). This enables us to express a variant of BRCA1 and test if it can rescue the cells ability to repair with HR and convert to GFP-positive.

2.3b HeLa DR-FRT generation

Expressing specific variants of BRCA1 in the standard HDR assay is fairly straightforward. Modified pcDNA3 expression vectors with BRCA1 cDNA are able to rescue the number of cells expressing GFP to the same level as observed with wild type

BRCA1 or with control siRNA depletions. The parallel adaption of the HDR assay required a very different approach. Since the assay works at a cellular level, it was imperative that only a single variant of BRCA1 could be expressed in any given cell. If two plasmids enter the cell, we would be unable to score which variant was responsible for any positive HDR activity. When the goal was working with libraries of BRCA1 variants, we faced a significant technical challenge. Our solution was to adapt the HeLa

DR-13-9 cells once again. We stably integrated a flipase recognition target (FRT) sequence in the cells (Thermo Fisher Scientific Inc.). The FRT constructed also included a

LacZeo gene, giving the cells zeocin resistance and β-galactosidase activity (Figure 6 A).

After selecting clones that exhibited a single band via southern blot, we used a mammalian β-galactosidase activity assay (Gal-Screen, Thermo Fisher Scientific Inc.) and selected a clone that showed excellent β-galactosidase expression, indicating the construct was integrated in an active locus (Figure 7). β-galactosidase activity was assessed according to manufacturer instructions using a kit from Thermo Fisher

Scientific Inc. (Catalog #T1032). Clonal expansion of the selected colony established our

HeLa DR-FRT cell line. HeLa DR-FRT cells provide the framework of using flipase (Flp) to integrated plasmid DNA in a process known as Flp-FRT recombination. Flp-FRT recombination allowed us to efficiently integrate tens to hundreds of thousands of

BRCA1 variants in parallel into our modified HeLa cell line.

Figure 6 Flp-FRT recombination and HP-HDR A. FRT site integration and Flp-FRT recombination schematic ©Thermo Fisher Scientific Inc. B. HeLa BRCA1 variant library integration. C. HeLa BRCA1 variant library HP-HDR

β-Galactosidase Activity Assay

8000000 7588435

7000000

6000000

5000000 Units

4000000 Light

3000000 Relative

2000000

1000000

226712 15 33 3648 0 No cells No substrate HeLa-DR HeLa-DR-FRT HeLa-DR-FRT BRCA1 integration

Figure 7 β-Galactosidase activity of HeLa DR-FRT HeLa DR-FRT shows enhanced β-galactosidase activity from integrated LacZeo expression. BRCA1 integration demonstrates expected loss of β-galactosidase activity.

BRCA1 integration utilizing Flp-FRT recombination are site specific integrations to ensure consistent expression of any given variant. Additionally, the hygromycin resistance will only express if it has been integrated into the HeLa genome at the appropriate FRT site. The FRT construct that is already inserted in the host genome includes the start ATG codon for the hygromycin resistance. The rest of the hygromycin resistance sequence, which is delivered to the host genome via the BRCA1-containing pcDNA5 vector ensures appropriate integration for expressing BRCA1. When integrated into the host genome via the Flp-FRT alignment and recombination, the cell can begin to express BRCA1 and hygromycin resistance. We also modified the pcDNA5 vector with a

5’ untranslated region (5’ UTR) sequence that includes an intron, derived from a rabbit

β-globin intron, which improves BRCA1 expression level.

2.3c HeLa BRCA1-variant cell line generation

Seventy to eighty million HeLa DR-FRT cells in ten 10 cm tissue culture plates are transfected with 200 µg of pOG44 to express a modified flipase enzyme and 100 µg pcDNA5/FRT cloned with BRCA1 variant library. pOG44 flipase has reduced activity at

37 C. Plasmids are diluted in 10 mL Opti-MEM and incubated for 5 minutes. 300 µl of

Lipofectamine 2000 is diluted in 10 mL Opto-MEM for 5 minutes. The lipofectamine and plasmid dilutions are then combined and incubated for 20 minutes. The mixture is then applied directly to cells. After 24 hours, cells are tryspinized and transferred to ten 15 cm tissue culture dishes. Four to eight hours later, the cells are moved from a 37 C humidified incubator to a 30 C humidified incubator. The cells will incubate at 30 C 34 overnight. This temperature shift improves the activity of the flipase enzyme. The cells are then incubated at 37 C for an additional 24 hours. Approximately 72 hours after the initial transfection, the cells are trypsinized and transferred to twenty 15 cm plates containing selection media. Selection media is conditioned with 50% culture media over healthy cells for 24-72 hours. The media is filter sterilized with a 22 µm filter.

Combined with fresh media supplemented with 10% fetal bovine serum, hygromycin B reagent is added to a concentration of 550 µg/mL. Cells are incubated at 37 C for 24 hours. The plates are washed with sterile phosphate buffered saline (PBS) and the selection media is replaced. Cells are then incubated for 48 hours at 37 C. The cells are washed again with PBS and the selection media is replaced. This is repeated every 48 hours until cell colonies are visible without a microscope. Cells continued to be cultured until colonies become tightly clustered with cells centrally when observed with a phase contrast microscope. Colonies were then counted, trypsinized, resuspended in 20 mL culture media, and mixed thoroughly. 75% of the resuspension is frozen in 1 mL aliquots. 5 mL of resuspended colony mixture is plated onto three 15 cm plates (3 mL,

1.5 mL, 0.5 mL). Incubate for 24 hrs. Begin passaging plate that is the closest to a confluent monolayer of cells (Figure 6 B).

2.3d Highly Parallel Homology Directed Repair Assay

A confluent 10 cm plate of HeLa BRCA1 variant cell line is trypsinized and resuspended in 10 mL of culture media. 65 l of the suspension was plated in each of

48 wells across two 24-well tissue culture plates. 24 hours later, each well is transfected 35 with 30 pmol of siRNA and 1.5 l oligofectamine. Oligofectamine was diluted with 6 l

Opti-MEM and siRNA was diluted with 25 l Opti-MEM for 5 minutes. The dilutions were then combined and incubated for an additional 30 minutes. The transfection mixture is then applied directly to the cells. 24 hours later, the cells are trypsinized and transferred to four 6-well tissue culture plates, then incubated for another 24 hours.

Each well is then transfected with 50 pmol of siRNA, 3 g pCBASceI (for I-SceI expression), and 3 l Lipofectamine 2000. Plasmid and siRNA are diluted in 125 l Opti-

MEM and lipofectamine is diluted in 125 l as well, then incubated for 5 minutes. The dilutions are combined and incubated for an additional 20 minutes. The transfection mixture is then applied directly to cells. Four to six hours later, the culture media is replaced with fresh media. After 72 hours, the cells are sorted using fluorescent activated cell sorting (FACS) (Figure 6 C). GFP expressing cells and GFP negative cells populations are then processed for genomic DNA. Genomic DNA (gDNA) is extracted with a Qiagen DNeasy Blood & Tissue Kit according to manufacturer instructions. DNA preparations are then amplified for sequencing.

2.3e Construction of barcoded variant libraries of BRCA1

For the error prone library, BRCA1 1-302 was amplified by PCR and used as template for mutagenesis by error prone PCR using the Mutazyme II (Agilent). Sanger sequencing of individual clones was used to empirically determine cycle number and input template concentration to maximize clones with single mutations (18 and 22 cycles, 100 ng template amplicon). Codon swap libraries of BRCA1 1-302 were 36 constructed using pUC19_HA-BRCA1_1-302 as a template using a previously described method (Jain & Varadarajan, 2014). For each codon, mutagenic primers were ordered with machine-mixed NNK bases at the 5’ end of the sense oligonucleotide from IDT. Pool

1 has codon swaps in amino acids 2-96, pool 2 has codon swaps in amino acids 97-192 and pool 3 has codon swaps in amino acids 193-302. Mutagenized BRCA1 1-302 was then cloned into the HindIII/EcoRI sites of pcDNA5-FRT/TO_Intron_BRCA1_303-1863 that had a second EcoRI site in the gene for hygromycin resistance destroyed and a multiple cloning site added at the MluI/NruI sites for the barcode. A 16 base, degenerate barcode was then cloned into the NotI/SbfI sites of the multiple cloning site.

There are approximately 160,000 barcoded variants in the library made by error prone mutagenesis and 25,000 for each of the codon swap libraries.

2.3f Sequencing to link the BRCA1 N-terminal variants to their barcode

The error prone library was sequenced using a previously published Illumina- based, sub-assembly method resulting in 46,000 high quality assemblies (Hiatt,

Patwardhan, Turner, Lee, & Shendure, 2010; Starita et al., 2015). The variants and barcodes of the codon swap libraries were sequenced by single molecule real time

(SMRT) sequencing (Pacific Biosciences). To prepare the circular SMRT-bell templates, the intervening sequence between the barcode and the BRCA1 N-terminal variable region was removed by restriction digest, followed by end-repair and blunt-end ligation

(Travers, Chin, Rank, Eid, & Turner, 2010). The ligations were transformed into E. coli to remove concatamers. The plasmids were then cut with SbfI and EcoRI to release the

37 barcode and BRCA1 N-terminal variable region. Hairpin SMRT-bell oligonucleotides with priming sequences were ligated to the fragments, purified and quantified by

BioAnalyzer (Agilent). Each library was loaded in four lanes of a Pacific Biosciences RS II.

To obtain high quality sequences, base call files (bax files converted to bam) for each lane of the sequencer were analyzed by the Circular Consensus Sequencing 2 algorithm

(Pacific Biosciences). Consensus sequences were filtered further to remove low quality reads by requiring a minimum PHRED score of 20 at each base and requiring that the majority of variants associated with each barcode have identical sequences. If there was not a majority for any variant, the sequence with the highest average score was used.

Barcodes with variants that did not meet these thresholds were removed from analysis.

2.3g Sequencing barcodes from genomic DNA

Barcodes were amplified from genomic DNA from the GFP positive and negative samples and prepared for Illumina sequencing using a two-step PCR protocol. In order to specifically amplify barcodes from the genomic DNA and not contaminating plasmid the primers for the first PCR reaction were complementary to the SV-40 promoter in the genomic FRT cassette and to sequence directly 3’ of the integrated barcodes and amplify a 3750 base region (SV40_F and pc5_bc_nexteraR). Up to four g of genomic

DNA was used as PCR template over sixteen 50 µl PCR reactions containing the above primers and KAPA Robust hot start ready mix (KAPA Biosystems). We assume that the

HeLa genome is triploid and weighs 9 pg, therefore we estimate that four µg of genomic

DNA corresponds to 444,444 genome equivalents. In the case that the genomic DNA

38 concentration was too low to add 250 ng per PCR reaction, the DNA was spread over eight PCR reactions. We performed 15 cycles of amplification using the following conditions: 95 ˚C for 5 minutes, followed by 95 ˚C for 1 minute, 72 ˚C for 4 minutes. PCR reactions were pooled and primers were removed from the first PCR reaction using 0.5

X Ampure (Agilent) clean up and eluted in 1/10 of the original PCR volume. Half of the product from the first PCR reaction was used to seed the second PCR reactions. The second PCR had primers that are complementary to a sequence 5’ of the barcode and the tail of the pc5_bc_nexteraR primer (pc5bc_p5_F and example index primer,

NexV2ad2_A1), they also have Illumina p5 and P7 sequences and nine base indices that uniquely identify each sample, these primers amplify a 250 base pair sequence. The second PCR reactions were 1/8 of the first PCR volume and contained KAPA Robust hot start ready mix and 0.5X SYBR Gold (Invitrogen) and were amplified under the following conditions 95 ˚C for 5 minutes, followed by 95 ˚C for 20 seconds, 72 ˚C for 40 seconds.

These reactions were monitored using MiniOpticon qPCR machine (Bio Rad) and removed during exponential amplification. PCR reactions were pooled and primers were removed using a 1X Ampure clean up. Amplicons were quantified, multiplexed and the

16-base barcode and indices were sequenced on a Nextseq500 (Illumina). All samples had at least 10 million reads.

2.3h Determining a functional score for each BRCA1 variant

The 16 base barcodes were converted back BRCA1 N-terminal variants. The frequency at which each variant occurs in each sample is calculated. The frequency for

39 each variant in the GFP+ population is divided by its frequency in the GFP- population.

These ratios are then normalized to the performance of wild type.

Name Sequence SV40_F GAA GTA GTG AGG AGG CTT TTT TGG AGG CTA CC pc5_bc_nexteraR GGC TCG GAG ATG TGT ATA AGA GAC AGG CAT CCC ATG CAT CGT CAT CGT CAT CGC ATC AG pc5bc_p5_F AATGATACGGCGACCACCGAGATCTACACGAGCAAA ATTTAAGCTACAACAAGG NexV2ad2_A1 CAAGCAGAAGACGGCATACGAGATTACGAAGTCGT CTCGTGGGCTCGGAGATGTGTATAAGAGACAG siRNA GL2 sense 5’- CGU ACG CGG AAU ACU UCG A - 3’ siRNA GL2 antisense 5’- UCG AAG UAU UCC GCG UAC G - 3’ siRNA 3’ UTR BRCA1 sense 5’- GCU CCU CUC ACU CUU CAG U - 3' siRNA 3’ UTR BRCA1 antisense 5’- ACU GAA GAG UGA GAG GAG C - 3' siRNA coding sequence BRCA1 5’- GGU UUC AAA GCG CCA GUC A - 3' sense siRNA coding sequence BRCA1 5’- UGA CUG GCG CUU UGA AAC C - 3’ antisense Table 2 Nucleic acid oligomer sequences used in HP-HDR siRNA sequences have an additional dTdT at the 3’ end of the sequence

2.3i Statistics

Error bars represent standard error of mean. Spearman’s rank order test correlation and significance were calculated using R statistics software.

2.4 Validation Results

The initial step to developing the highly parallel HDR was to validate its results.

This starts with our controls. For the assay to work, expressing WT BRCA1 resistant to the 3’ UTR BRCA1 siRNA was needed to rescue the cells and establish a positive control.

To test this, we integrated WT BRCA1 cDNA in the HeLa DR-FRT cell line. We then depleted the WT cell line of endogenous BRCA1 and challenged the cells with the I-SceI endonuclease. As shown in Figure 8, the integrated WT BRCA1 was able to rescue the cells to nearly 68% of the non-depleted cells. Now that we had established a positive control, we needed to prove a negative control. We integrated a C61G mutation of

BRCA1. We previously have shown that the high risk C61G mutation is defective in DNA repair (Ransburgh et al., 2010). The C61G BRCA1 cell line showed no improvement at all compared to endogenous BRCA1 depleted cells, consistent with the standard HDR assay results, as seen in Figure 8.

100%

90% Empty 80% WT BRCA1 70%

control C61G

60% WT:C61G mix 50% normalized

40% GFP 30% %

20%

10%

0% 3' UTR siRNA Coding Sequence siRNA

Figure 8 HDR of single integration HeLa BRCA1 cells WT BRCA1 integration rescued nearly 68% compared to controls, while C61G exhibited very little rescue. Coding sequence siRNA targeting both endogenous and integrated BRCA1 inhibited WT rescue to just over 26%.

With a positive and negative control, the next step was to determine if we could sort WT cells and C61G cells using their HDR GFP reporter. 50% of each WT and C61G cell lines were mixed and run through the HDR protocol (Figure 8). The standard HDR ends with a flow cytometry analysis of the number of GFP cells in the population, but this time we used the sorting function of fluorescent activated cell sorting (FACS). Using

FACS, we separated GFP positive cells from GFP negative cells. We predicted that the

GFP positive cells would only contain WT cells, while the GFP negative cells would contain a mix of WT and C61G mutant cells. After a genomic DNA purification of the sorted cells, Sanger sequencing revealed that the GFP positive population was preferentially wild type, while the GFP negative population contained exhibited the mixed bases expected of a WT and C61G mutation mix as shown in Figure 9. These results lead us to begin exploring scaling and more complex sorting.

Figure 9 HeLa BRCA1 WT and C61G HDR and sort A. The WT cysteine codon is a TGT sequence while the loss of function glycine variant is GGT. HDR result for control siRNA vs 3’ UTR BRCA1 siRNA during sort. B. Sanger sequencing demonstrated the GFP positive cells that in the 3’ UTR BRCA siRNA selected for WT sequence.

The critical advantage adapting our assay to a highly parallel version is the capability to test several thousand variants simultaneously. We then tested the capacity of our variant integration. Our collaborator, Dr. Lea Starita, provided a ‘library’ of WT

BRCA1. This library had the same BRCA1 cDNA, however, a noncoding region of DNA was used to generate a 65,000 unique DNA sequences. This ‘barcode’ DNA string was used to study and identify our systems integration capabilities. We proceeded to integrate the WT library. The targeted libraries used later in HP-HDR only account for

2000-6000 variants. However, the barcode library of over 65,000 sequences allow for internal redundancies (multiple barcodes are linked to a single BRCA1 variants). After many optimizations, we were able to observe an estimated 5,000 colonies resistant to hygromycin selection per 10 million cells integrated. We isolated approximately 24 of these colonies, of which 19 survived. Each colony was then sequenced to ensure only a single barcode was present (Table 3). The remaining colonies were pooled into a single cell line. We then extracted the genomic DNA to determine how many of the barcodes could be sequenced using Illumina DNA Seq. As shown in Figure 10, applying a filter of a minimum of 10 reads per barcode closely matches our estimated number of colonies.

Increasing the read filter reduces the number of barcodes that were detected, but increases confidence.

Sample Barcode Mixed bases in electrophoretogram Mix A12 (illumina index) Colony C ATTTAGGGACAGGATG No Colony F TGTAGATTTGGGATGA No Colony G CCATGCTCAAGTCGAG No Colony H ATTTAGGGACAGGATG Same as C Colony H2 TTGGTGATTGTTACGG No Colony P CCGTACTGTACGCCAG No Colony K2 AGCAGCCAGGCGTGGA No Colony L TACGCATTTAGTAGTC No Colony N2 CCATGCTCAAGTCGAG Same as G Colony P2 CCGTACTGTACGCCAG Same as P Colony R GTGCGGGTTGTCCAGC No Colony S GCGTTACGTGAGTGAC No Colony T ATTGATGAGAATTTGG No Colony T2 ATTGATGAGAATTTGG Same as T Colony V GCCGGGCTCGCGACTG No Colony I ATTTAGGGACAGGATG Same as C Table 3 HeLa WT BRCA1 Barcode Colony Sequences HeLa cells integrated with WT BRCA1 including barcode sequence were isolated during clonal selection. 19 clones were established and PCR successfully amplified the barcode region for sequencing from 16 clones. No mixed bases were present in any clonal population, validating only a single integration event took place for each colony.

Figure 10 Barcode WT BRCA1 Integration Efficiency HeLa cells integrated with WT BRCA1 including barcode sequence in 10 million HeLa DR-FRT cells were pooled after an estimated 5,000 colonies were counted during hygromycin B selection. Illumina sequencing shows that at a filter of approximately 10 reads per barcode, approximately 5000 unique barcodes are detected. Increasing the filter reduces to number of unique barcodes but improves the confidence of the barcode reads.

Although we established that we can integrate 50,000 barcodes simultaneously, it is just as critical that any given cell only expresses a single variant. The isolated colonies are clonal populations where a single barcode should be seen in DNA sequencing. Sequencing the genomic DNA at the barcode region using Sanger sequencing showed no mixed bases, demonstrating that the colonies were only expressing a single variant (Table 3).

The final validation experiment was to simultaneously integrate 20 BRCA1 variants that we had previously generated and for which we had known HDR data. This tested the ability of the protocol to sort complex mixtures of BRCA1 variants. Each of the 20 variants selected had a mutation in the first 100 bases of BRCA1, such that a next generation DNA sequence analysis would yield frequency of that particular variant within the sorted GFP population. After sorting, the GFP positive frequencies were normalized against the GFP negative frequencies and compared to standard HDR. As shown in Figure 11, the simultaneous HDR in our highly parallel HDR correlated well with standard HDR results.

Figure 11 Comparison of mixed HP-HDR and single HDR Variants of BRCA1 that were mixed, sorted, and analyzed in the HP-HDR assay. Scores were correlated against single variant HDR. Significance and Spearman correlation calculated with R.

2.5 BRCA1 variants

Next, we began to work with BRCA1 variant libraries. Although the highly parallel HDR system is designed to handle several thousand variants simultaneously, efficient mutagenesis is equally important in how efficient our design can be. At the amino acid level, missense mutations alone account for over 35,000 unique single amino acid variants of full-length BRCA1 (1862 amino acids with 19 possible variations). To allow for both more effective mutagenesis and more robust parallel HDR results, we chose to restrict the location of the mutagenesis. Based on our standard HDR data, we have a catalog of BRCA1 mutants and their effect on HDR. It is important to note that the high risk clinical variants correlate perfectly with loss of function in HDR (Ransburgh et al., 2010; Starita et al., 2015). This provides our rationale to intensely analyze thousands BRCA1 variants in HDR, enabled by highly parallel HDR. Previous data from numerous studies corroborate that there are two domains in BRCA1 that are important for both HDR and clinical risk: the N-terminal RING domain and the C-terminal BRCT domains (Figure 12 and Figure 13) (Lu et al., 2015; Ransburgh et al., 2010; Starita et al.,

2015; Towler et al., 2012). For this study, we chose to analyze the RING domain.

Figure 12 BRCA1 HDR and Domains Loss of function mutations in BRCA1 are clustered in the RING and BRCT domains. Adapted from (Lu et al., 2015).

Figure 13 BRCA1 variants clinical and HDR comparison A. Most BRCA1 variants are VUS. Synthetic variants do not exist in the population yet. Pathogenic variants are clustered in the RING and BRCT domains. B. Every tested pathogenic variant is deleterious in the HDR assay, and every neutral variant is neutral in the HDR assay. Variants that result in loss of function for HDR are similarly clustered in the RING and BRCT domains.

Dr. Lea Starita at the University of Washington, has been closely collaborating with us throughout the development of the highly parallel HDR. Dr. Starita was able to create two RING domain missense mutation libraries. One was generated via error- prone PCR, and the other using customized primers and an automated cloning system.

These libraries utilize a DNA barcoding system that can be used to identify a specific variant. This key breakthrough allows us quickly identify which variants exist in GFP positive and negative sorted cells.

We have successfully integrated both RING domain libraries. We estimate approximately 35,000 integration events using the error prone library, and nearly

200,000 integration events using the primer generated library. The number of colonies exceeds the number of possible variants (approximately 6,000 variants per library), and should ensure capture of nearly all the possible amino acid variants. The error-prone library was integrated twice, once with approximately 20,000 integration events, and a second time with approximately 15,000 integration events. The primer generated library was integrated three times, and each integration used a library that targeted approximately 100 amino acid region of the RING domain. Data from these integrations are currently being analyzed, but we have promising results so far. We have begun to validate the number of integrations in the second error-prone integration experiment.

After converting the barcodes into the corresponding missense mutations, we believe we have fairly good coverage of the RING domain.

The error-prone library after sorting has generated promising benchmarks as well. We have some preliminary results from the second integration of the error-prone library. GFP positive population demonstrated a 75% reduction in nonsense mutations.

Conversely, WT sequences sorted with a slight preference to the GFP positive populations (Figure 14).

A Error-prone library (2nd integration) Overview

Type of Variant noGFP GFP WT 37.9% 41% SYN WT 6.9% 8.2% Single missense 7.0% 6.8% 1st 100 AA Single Missense 15.4% 17.3% 100-300 AA Multiple missense 30.8% 26.1% STOP 1.9% 0.5%

GFP WT

SYN

100-300

RING

multiple

noGFP STOP

0% 20% 40% 60% 80% 100%

Percentage of reads from each sample

Figure 14 Error Prone library 2nd integration overview A. We see slight preferences after sorting for WT, synthetic WT (SYN WT) for the GFP positive population, as expected. Mutation variants sort slightly preferentially in the GFP negative population. Most importantly, nearly 75% of the nonsense mutations are sorting in the GFP negative population. B. A graphical representation of panel A.

2.6 Discussion

Translational research is a primary driving force behind molecular biology. One of the main reasons we selected BRCA1 as the first protein to study in the HP-HDR assay was due to its very important clinical application. We have tested over a hundred

BRCA1 variants in the standard HDR assay. Not counting splicing differences, we have observed that BRCA1 variants that were clinically pathogenic never rescued HDR function. Similarly, every clinically benign variant always rescued HDR (Figure 13).

Specific to the RING domain, only 22 mutations have been classified, 19 pathogenic and

3 benign (Starita et al., 2015). In previous studies, we analyzed 45 RING domain variants

(Ransburgh et al., 2010; Starita et al., 2015; Towler et al., 2012). 8 of the 19 known pathogenic mutants have been tested in the HDR assay, with an average of 19% rescue of HDR and a maximum of 33%. R71G is a clinically pathogenic mutant that affects splicing. Since BRCA1 variant expression is from cDNA, a limitation of HDR and HP-HDR are mutations that affect splicing. The three benign mutants scored much higher, with a mean of 88% rescue and a minimum of 77%. We defined a cutoff of 53%, splitting the difference between the pathogenic mean and the benign mean, similar to an analysis done for BRCA2 (Guidugli et al., 2014). Using this cutoff and accounting for splicing variants, the HDR assay has 100% sensitivity and 100% specificity for clinical risk. HDR scores were used as the “gold standard” in parallel assays studying E3 ubiquitin ligase activity and BARD1 interaction (Starita et al., 2015).

BRCA1 function in HDR has thus far been shown to be its major determinant of cancer risk. We have previously studied BRCA1 in collaboration with Dr. Lea Starita in a similar highly parallel fashion (Starita et al., 2015). This study analyzed a RING domain variant library in two different assays: an E3 ubiquitin ligase assay and a yeast two- hybrid assay. The E3 ubiquitin ligase assay was tested using a phage display assay that selects for protein variants capable of autoubiquitination (Christensen, Brzovic, & Klevit,

2007; Starita et al., 2013). A similar assay analyzed BRCA1 and BARD1 interaction using a yeast two-hybrid assay. These parallel assays were referenced against the single variant HDR results to determine the impact of E3 ubiquitin ligase and BARD1 binding activity. There was relatively low linear correlation, and slightly improved support vector regression model correlation (Starita et al., 2015).

We believe the evidence provided so far indicates that a HP-HDR assay is the best method to determine the impact of VUS in BRCA1. Although E3 ubiquitin ligase activity and BARD1 binding are important to BRCA1 function, they do not intrinsically test for a pathway in the way a functional assay such as HDR tests for. Ubiquitin ligase and even protein interactions could have redundancies in the cell that can be compensated for and thus result in false-positive results when assessing clinical impact.

However, the strength of the HDR and similar pathway assays are that there are fewer redundancies and compensation mechanisms which the assay cannot account for. A

BRCA1 variant that scored poorly in the HDR assay means that the cell population could

58 not utilize any bypass or other compensatory mechanisms and still repair DSBs with HR.

We believe this type of functional analysis can best estimate clinical impact.

Another study that focused on BRCA1 variants was a complementation assay that integrated BRCA1 variants in mouse embryonic stem cells (Bouwman et al., 2013).

This study used a similar Flp-mediated recombination integration of BRCA1 variants.

However, this is a standard Flp-mediated recombination intended to integrate a single variant of BRCA1 per experiment. Indeed, it takes approximately eight weeks to test up to 20 variants in the study’s workflow (Bouwman et al., 2013). In contrast, the HP-HDR is designed to analyze up to 50,000 variants in 4-6 weeks. Another shortcoming of the single-variant approach is that it is impractical to analyze synthetic variants. As a result, single-variant functional assays are limited in scope as reactive studies that are incapable of providing relevant laboratory data to clinicians when a new variant surfaces in the population. Highly or massively parallel assays have the scale to test for synthetic variants that do not yet exist in the population. Studies such as HP-HDR have the potential to generate lookup tables for future patients at a fraction of the cost of single- variant assays.

HP-HDR was designed to initially focus on analyzing BRCA1, a large protein with known effects in HDR and correlating clinical risk. However, the assay is also flexible in nature, with several potential libraries that can be studied in the future. The next logical choice is the BRCT domain of BRCA1. Similar to the RING domain, variants that carry pathogenic risk cluster in the BRCT domain and similarly do not score well in the 59 standard HDR assay (Ransburgh et al., 2010; Starita et al., 2015; Towler et al., 2012).

Another prime library is BARD1. BARD1 is particularly interesting because despite its strong association with BRCA1, relatively few BARD1 mutations have known pathogenic risk (C. Lee et al., 2015). These libraries can be relatively quickly integrated into the

HeLa DR-FRT cells and tested for HDR function in a highly parallel fashion.

Finally, the concepts of how we developed HP-HDR are also applicable to other functional assays. Only three core properties are necessary to develop a highly parallel version of a functional assay: it must operate at a cellular level, there must be a way to sort positive and negative results (such as FACS), and there must be a way to identify which variants exist in each population (such as direct sequencing or barcodes). These basic properties apply to many reporter-based assays already, including other repair assays that test for NHEJ, SSA, or MMEJ. We believe that these parallelized functional assays are not only a breakthrough in how we study protein structure-function but also a key component in the effort towards determining the clinical relevance of detected variants.

Chapter 3: HDAC10 as a potential therapeutic target in ovarian cancer

Muhtadi M. Islam*, Tapahsama Banerjee*, Colin Z. Packard, Selvendiran Karuppaiyah,

David Cohn, and Jeffrey D. Parvin

*Contributed equally

This chapter is in preparation for submission for publication.

Author contributions: Muhtadi Islam analyzed public databases essential for Figure 15 and integrated the results from other experiments and wrote the manuscript.

Tapahsama Banerjee performed the proliferation assays. Colin Z. Packard performed the homology directed repair assays. Selvendiran Karuppaiyah and David Cohn provided primary cell lines from ovarian tumors. Jeffrey D. Parvin supervised the experiments and advised on the analysis of results.

3.1 Abstract

We analyzed histone deacetylase 10 (HDAC10) for function within the DNA damage response in BRCA1-null ovarian cancer cells as well as evaluated the potential of general HDAC inhibitors in primary ovarian carcinoma cells. HDAC10 had previously been shown to be highly stimulatory to the process of homology directed repair in HeLa cells, and in this study we investigated if the loss of HDAC10 could impact cancer. Using

The Cancer Genome Atlas (TCGA) dataset, we found that deep deletions in HDAC10 occurred in 5-10% of ovarian cancer tumors. We hypothesized that the loss of HDAC10 would sensitize cells to platinum therapy. From the TCGA data we found that low

HDAC10 mRNA levels correlated with platinum sensitivity of the tumors. Cell proliferation and DNA damage assays in a BRCA1-null ovarian carcinoma cell line demonstrated reduced DNA repair capacity and sensitization of platinum therapy.

Similarly, primary ovarian carcinoma cells demonstrated a sensitization of platinum therapies when treated with HDAC inhibitors. From the results of this study, we suggest that the inhibition of HDAC10 may potentiate the platinum therapies in BRCA1 deficient ovarian tumors.

3.2 Introduction

Histone modifications have been central in the understanding of post- translational modifications and their effects on the regulation of gene expression

(ALLFREY, FAULKNER, & MIRSKY, 1964; Kouzarides, 2007). Histone acetylation is a reversible process, governed by two classes of enzymes: histone acetyl transferases

(HATs) and histone deacetylases (HDACs). The acetylation of lysine on histone proteins is generally stimulatory to mRNA synthesis (Brownell et al., 1996). One suggested mechanism is the neutralization of the positively charged lysine residue by acetylation loosens the interaction of the DNA with the underlying nucleosome, providing access to transcriptional machinery (Bannister & Kouzarides, 2011). Alternatively, the transcriptional control could be modulated through direct protein interactions, or a combination of the above (Torok & Grant, 2004). HDACs catalyze the reverse reactions, contributing to transcriptional repression by rendering the chromatin to a state that is functionally silenced (Taunton, Hassig, & Schreiber, 1996). HDAC proteins have five phylogenetic classifications: I, IIA, IIB, III, and IV (Gregoretti et al., 2004). In this study we focus on the class IIB HDAC10.

Although originally classified as ‘histone’ deacetylases, these enzymes are not limited to histone substrates. Acetylated lysine on non-histone proteins have been shown to regulate multiple cellular functions, which can be reversed by HDACs

(Choudhary et al., 2009). Most findings about HDACs focus on these factors as repressors of transcription, but our understanding of the roles of HDACs continues to expand (Glozak, Sengupta, Zhang, & Seto, 2005; Heinzel et al., 1997; Jones et al., 1998;

Nan et al., 1998). As an example, HDAC1, HDAC2, and HDAC3 have all shown tumor- suppressive genomic stability function (Bhaskara et al., 2010; Heideman et al., 2013).

HDAC9 and HDAC10 have been shown to stimulate homologous recombination in HeLa

63 cells (Kotian et al., 2011). The DNA repair function of HDAC10 prompted us to explore a possible association with cancer.

In this study we find that HDAC10 is either deleted or expressed at low level in a subset of ovarian cancer tumors. Additionally, we find a significant correlation with platinum therapy sensitivity and low levels of HDAC10 mRNA within the same tumor samples. Based on our results from the in vitro studies, we suggest that inhibition of

HDAC10 may potentiate the platinum-based therapy response of ovarian BRCA1 deficient tumors.

3.3 Materials and Methods

3.3a Cell Culture and Reagents

HeLa DR-13-9 cells utilized for homology directed repair have been previously described (Parvin et al., 2011) and cultured using standard HeLa culturing protocols.

UWB1.289 ovarian carcinoma cells were purchased from ATCC and cultured according to manufacturer specifications. HDAC inhibitors trichostatin A (TSA) and suberanilohydroxamic acid (SAHA) were purchased from Sigma-Aldritch. HDAC10 and control siRNAs were synthesized and purchased from Integrated DNA Technologies.

Sequences for the siRNAs are listed in Table 4. MTT reagent, 3-(4,5-dimethylthiazol-2- yl)-2,5-diphenyltetrazolium bromide and comet assay lysis buffer were purchased from

Trivigen. SYBR Green used in the comet assay was purchased from Bio-Rad.

3.3b Homology Directed Repair Assay (HDR)

The HDR assay we utilized has been previous described and characterized (Parvin et al., 2011; Ransburgh et al., 2010; Towler et al., 2012). 300-400 thousand HeLa cells were plated in each well of a 6-well tissue culture dish (~9.5 cm2) on day 1. On Day 2, 3

µg of pCBASce1 plasmid containing I-Sce1 endonuclease was transfected using 3-5 µl of

Lipofectamine 2000, first diluted in 125 µl Opti-MEM each. Four to six hours later, the culture media was replaced with fresh media containing 0, 75, 150, or 300 nM of trichostatin A (TSA) or 0, 0.5, 1.0, or 2.0 µM of suberanilohydroxamic acid (SAHA).

Cisplatin was added to each of the TSA and SAHA treatments at 0, 100, and 300 nM concentrations. Cells were then incubated for 72 hours. The cells in each treatment sample were then treated with trypsin and resuspended in 0.5-1 mL PBS. 10,000 cells were counted using a BD Biosciences FACSCalibur instrument available at the Ohio State

University Comprehensive Cancer Center Analytical Flow Cytometry core laboratory.

The 10,000 cells were gated using an area of the forward scatter-side scatter plot to optimize live cell counting. The number of cell expressing green fluorescent protein

(GFP) was recorded for each sample and the GFP percentage was normalized against untreated samples. The experiment was repeated in triplicate.

3.3c Comet Assay

The comet assay measures eukaryotic cell DNA damage using gel electrophoresis

(Nandhakumar et al., 2011; Tice et al., 2000). Damaged DNA is electrophoresed away from cells suspended in agarose, forming a comet tail shape that can be visualized and

65 quantified. UWB1.289 cells were plated in 6-well tissue plates (~9.5 cm2 per well). Once the cells reached a confluence of approximately 60%, they were transfected with control siRNA targeting luciferase (GL2), or siRNA targeting two different HDAC10 sequences (HDAC10-1 and HDAC10-2). 100 pmol of each siRNA was diluted in 250 µl of

Opti-MEM, and 5 µl of Oligofectamine was similarly diluted in 250 µl of Opti-MEM and the transfection was carried out according to manufacturer specifications. 48 hours later, a second transfection was repeated with the same volumes and incubation times.

Five days after the initial transfection, the cells were exposed to four grays of x-ray ionizing radiation. The cells were then incubated for four hours, then detached from the plate with trypsin, mixed with culture media, pelleted at low speed, and resuspended in 1 mL of PBS. 5 µl of the suspension was diluted with 45 µl melted low melting point agarose and applied to a microscope slide. Slides were refrigerated to solidify the agarose, and then incubated in comet assay lysis buffer at 4 C for 45 minutes. The lysis buffer was then aspirated, replaced with electrophoresis buffer, and equilibrated in this buffer for 25 minutes. The cells are then placed in an electric field of

21 volts for 45 minutes. DNA is then precipitated using a precipitation buffer and incubated at room temperature for 25 minutes followed by aspiration and drying overnight. SYBR Green I solution (0.1 ml) was added to each slide for 30 minutes, and the solution was then gently removed and dried for 10-20 minutes. Slides were imaged with an Axiocamera instrument and analyzed with CometScore software. Each treatment and slide was prepared in triplicate.

3.3d MTT Assay

UWB1.289 MTT assays follow similar cell culture and transfection procedures as the comet assay until day five. On day five, or 48 hours after the second transfection,

2000 cells of each treatment were counted and plated in a 96-well plate (~0.143 cm2).

Cells were cultured for 72 hours in a 96 well plate.

Primary ovarian cancer cells were counted and 1000 cells were plated per well of a 96-well tissue culture plate. 0, 5, 10, and 20 µM cisplatin and 0 and 75 nM TSA were applied to both platinum sensitive and platinum resistant primary ovarian cells.

Three days after the start MTT assay, 10 µl of 12 mM MTT stock solution (~10% of the well volume) was added to each well. The cells were incubated at 37 C for four hours, after which cell membranes were disrupted with a 10% sodium dodecyl sulfate

(SDS), 0.01M hydrochloric acid (HCl) solution. After a final four-hour incubation at 37

C, the plate is shaken to ensure mixing and the absorbance is measured at 600 nm.

Each treatment was performed in triplicate.

3.3e Statistics

Error bars represent the standard error of mean. P-values represent two-sided student’s t-tests unless otherwise indicated. Two-way ANOVA analysis was used to test for interaction between TSA and cisplatin treatments in sensitive and resistant primary ovarian tumor cells. ANOVA P-values are interaction P-values.

3.4 Results

The starting point of this study is the analysis of HDAC10 in TCGA data (National

Cancer Institute, 2016). For the TCGA analysis, we utilized the visualization and analysis web tool cBioPortal (Cerami et al., 2012; Gao et al., 2013). We initially screened genetic changes to HDAC10 across multiple tumor types, including a large ovarian serous cystadenocarcinoma data set. This ovarian dataset had two different gene copy analyses and indicated a high rate of HDAC10 deletion. A TCGA provisional dataset indicated that 311 samples had 10% of tumors with a deep deletion of the HDAC10 gene. The same dataset analyzed in 2011 with 316 samples indicated about 5% of tumors with a deep deletion of HDAC10. Both gene copy analyses indicate ovarian tumors had the highest HDAC10 deletion rates out of all the available cancer datasets.

The dataset was also analyzed for loss of BRCA1. Although deep deletions of BRCA1 was relatively rare, approximately 10% of the tumors had a nonsense BRCA1 mutation. Two tumor samples had both an HDAC10 deletion and BRCA1 nonsense mutation.

With the scope of the informatics analysis narrowed to ovarian cancer, we analyzed characteristics of HDAC10 loss within the dataset. Although DNA repair genes are linked to cancer, they are also commonly targets of cancer therapeutics. The uncontrolled cell division of cancers makes DNA a prime target for disrupting the multiple processes needed to sustain the proliferation. Cisplatin is an interstrand crosslinker of DNA, interfering with mitosis as well as initiating the apoptosis response of the DNA damage response pathway. Since HDAC10 has been shown to be involved in

DNA repair, the first characteristic we evaluated was platinum sensitivity. We hypothesized that patients who were deficient in HDAC10 would be more sensitive to platinum therapy. As shown in Figure 15, no patients who had deep deletions of

HDAC10 were resistant to platinum therapy. 66.2% of shallow deletions were sensitive to platinum therapy, and 57.1% of diploid or amplified HDAC10 patients were resistant to platinum therapy. These results indicate the possibility that the loss of HDAC10 helps sensitize cells to platinum therapy, however the sample size of the deep deletion patients is fairly small.

Data regarding HDAC10 DNA copy numbers in cisplatin sensitive tumors were complemented by transcriptional analysis. The HDAC10 mRNA levels correlated with platinum sensitivity in the patients. In the subset of tumors in which platinum status was available, 62 patients were resistant to platinum therapy, while 128 patients were sensitive to platinum therapy. The mean of HDAC10 mRNA levels in the resistant sample was significantly higher than the sensitive sample (Figure 15). Tumor data suggested that HDAC10 copy loss or low expression were correlated with maintaining sensitivity to cisplatin. This result initiated our investigation of HDAC10 and HDAC inhibitor effects in a number of cell culture assays with an emphasis on ovarian cancer cells and platinum therapy.

Figure 15 HDAC10 and platinum status in ovarian cancer patients A. Platinum status of HDAC10 copy numbers. No patients with deep deletions of HDAC10 exhibited platinum therapy resistance, while patients with shallow, diploid, or amplified HDAC10 copy number exhibited platinum resistance in 34% of patients. B. mRNA levels of HDAC10 are significantly higher in platinum resistant status patients. P-value calculated using a student’s t-test.

We have previously found that HDAC10 is stimulatory for homologous recombination (Kotian et al., 2011). Cisplatin has multiple mechanisms of action through which it modulates its cytotoxic effect, including DNA crosslinking, and we tested whether cisplatin inhibited DNA repair by homology directed repair (HDR) and whether HDAC inhibition would enhance the effect cisplatin had on HDR. HDAC inhibitors are a class of compounds that typically inhibit multiple HDACs. Since the inhibition of HR is specific to HDAC 9 and 10, we focused on the inhibitors that include inhibition of HDAC10 (Kotian et al., 2011). The HDAC inhibitors we selected were trichostatin A (TSA) and suberanilohydroxamic acid (SAHA). SAHA does inhibit HDAC10, but to a lesser degree than its inhibition of other HDACs. There is currently no HDAC10 specific inhibitor. Cisplatin inhibited homologous recombination in our HDR assay at concentrations as low as 100 nM cisplatin. The effect of cisplatin inhibition of HDR had been observed before (Raaphorst et al., 2005; Wang et al., 2005). Importantly, HDAC inhibition further reduced HDR activity at high concentrations (Figure 16). This result suggests a possible mechanism mediated through HDAC10 for enhancing cisplatin in its clinical usage of inducing apoptosis.

Figure 16 HDR of cisplatin and HDAC inhibitors in combination A. Trichostatin A (TSA) HDAC inhibitor and cisplatin both reduce HDR. *Significant inhibition of HDR due to TSA when normalized to cisplatin. B. SAHA HDAC inhibitor exhibits similar behavior to TSA, except at high concentrations of SAHA which completely abolishes any HDR function. *Significant inhibition of HDR due to SAHA when normalized to cisplatin. Significance calculated with student’s t-test.

We have previously established that HDAC10 stimulates HR in HeLa cells, consistent with the results of the effects of cisplatin and HDAC inhibition in the HDR assay. In the following experiment we tested whether HDAC10 was involved in DNA repair with ovarian cancer cells. To study this, we utilized a comet assay in a cancer cell line. Using the BRCA1-null ovarian carcinoma cell line UWB1.289, we exposed cells treated with HDAC10 siRNA to ionizing radiation, allowed four hours for recovery, and compared the level of unrepaired DNA damage against controls. Cells were embedded in agarose in neutral pH and DNA with double-strand breaks yielding fragments sufficiently small to electrophorese out of the nucleus and form a comet tail was measured. The tail moments of the cells that had been transfected with either of two different non-overlapping siRNAs targeting HDAC10 were significantly larger than the tail moments from cells transfected with the control siRNA. The larger tail moment indicated more unrepaired DNA damage four hours post irradiation than in control transfected cells, as shown in Figure 17. The BRCA1-null UWB1.289 cell line was sensitive to depletion of HDAC10. Since HDAC10 and BRCA1 both function in the HR pathway, it is possible that this ovarian cancer cell line has partially compensated for the deficiency in BRCA1, and HDAC10 inhibition caused the compensation to fail.

UWB1.289 cells did not show significant DNA damage after ionizing radiation under control conditions despite being BRCA1 negative, while other cell lines that have BRCA1 deficiency exhibit significant DNA damage in a comet assay (B. Y. Chen et al., 2014).

Figure 17 UWB1.289 Ovarian cells comet assay with HDAC10 depletion A. Comet assay of UWB1.289 cells 4 hours post X-Ray irradiation (4 Grays). Cells were transfected with either control siRNA (GL2), or two different siRNAs each targeting HDAC10. Tail moments of cells that were transfected with HDAC10 siRNA are greater than control cells.

Figure 17 UWB1.289 Ovarian cells comet assay with HDAC10 depletion B. CometScore software based quantification of comet assay shows significant persistence in DNA damage when depleted with HDAC10 siRNA over control GL2 siRNA. P-values calculated using a student’s t-test. C. Western blot of HDAC10 depletion.

We hypothesized that the inhibition of HDAC10 would sensitize the UWB1.289 cells to the DNA damage and cytotoxic effects of platinum therapy. To test this hypothesis, we performed an MTT assay. MTT is a measure of metabolically active cells, which we use to measure proliferative activity. We transfected the cells with two rounds of siRNA specific for HDAC10 or control, and 48 hours after the second transfection, we incubated the cells in the presence of different concentrations of cisplatin. After three days of treatment with cisplatin at a range of concentrations from

0 to 80 µM, we performed the MTT assay. As shown in Figure 18, when treating

UWB1.289 with cisplatin, cells transfected with HDAC10 siRNA had significantly lower proliferation as compared to cells transfected with control siRNA. This result demonstrates the potential enhancement the inhibition of HDAC10 can have on cisplatin in BRCA1 deficient ovarian tumors.

Figure 18 MTT Assay of UWB1.289 cells treated with cisplatin UWB1.289 cells were transfected with control siRNA or HDAC10 siRNA. 48 hours after transfection, the cells were treated with 3 concentrations of cisplatin and cultured for 72 hours. Absorbance was measured after MTT reaction. Cells that were transfected with HDAC10 siRNA have significantly less metabolic activity, an indication that HDAC10 depletion enhances cisplatin cytotoxicity. *Significant enhancement of cisplatin cytotoxicity calculated using a student’s t-test.

Since decreased activity of HDAC10 enhanced the sensitivity of cells to cisplatin, we tested whether generally functioning HDAC inhibition enhances cisplatin sensitivity in primary ovarian cancer cells. These cells are primary cells from ascites of patients with ovarian tumors. These cells can only be grown for three passages, making it impossible to use the siRNA transfection; instead, we were able to culture these cells in the presence of HDAC inhibitors and cisplatin. We performed MTT assays in primary tissues cells, treating cells with various concentrations of cisplatin (0, 5, 10, and 20 µM) and TSA

(0 or 75nM). As shown in Figure 19, TSA enhanced cisplatin sensitivity in both resistant and primary cells. This result confirms a similar analysis of SAHA and paclitaxel in ovarian cancer cell lines (Dietrich et al., 2010).

Figure 19 MTT assay in primary ovarian cancer cells A. TSA enhances cisplatin resistant cells. B. Cisplatin sensitive cells show cisplatin enhancement with TSA. Significance was calculated using two-way ANOVA analysis to test for interaction. ANOVA interaction P- value is listed.

Based on these results we suggest that an HDAC10 specific therapeutic target could be developed to further study the potential of enhancing cisplatin sensitivity and increase the total response rate of BRCA1 deficient ovarian carcinomas to first line therapies.

3.5 Discussion

Platinum therapies are a cornerstone in the chemotherapy treatment of many cancers. In ovarian cancers, first line chemotherapy is generally platinum based therapy

(Agarwal & Kaye, 2003). Initial response is often promising, 80% of patients have a significant reduction in tumor size. 40-60% patients have a complete response, with no detectable tumor after initial therapy. Unfortunately, most patients relapse, with a median survival of only 18 months (Greenlee, Hill-Harmon, Murray, & Thun, 2001). Our results indicated that HDAC inhibition of HR could enhance the first line platinum therapy and increase survival. Several clinical trials are already underway of HDAC inhibitors, including adjuvant therapy with platinum therapies. SAHA is currently in clinical trials specifically in conjunction with platinum based therapies in ovarian cancers. These studies are in progress and do not have published data or patient outcomes. Preclinical studies corroborate our finding of platinum sensitization in ovarian cancer cells using other ovarian cells lines when treated with SAHA (Dietrich et al., 2010). We propose a novel finding that HDAC10 inhibition enhances the platinum sensitization of BRCA1 deficient ovarian carcinoma cells. This finding can direct current

80 clinical trials of pharmacological HDAC inhibitors as well as explore the development of an HDAC10 isoform specific agent.

A major issue with platinum therapy and ovarian cancer is relapse and platinum resistance. Part of this issue is the limited therapeutic window of platinum therapy due to its toxicity, especially nephrotoxicity (Arany & Safirstein, 2003). Similarly, current

HDAC inhibitors in conjunction with platinum therapy have been found to have significant toxicities (Matulonis et al., 2015; Mendivil et al., 2013). One possible reason of the toxicity could be most clinical HDAC inhibitors are not isoform specific. Although the broad HDAC inhibition could maximize cytotoxicity, it can also shrink the therapeutic window of HDAC inhibitors. Additionally, it is important to note that most HDAC inhibitors do not inhibit or have only low levels of inhibition of HDAC10 (A. C. West &

Johnstone, 2014). Isoform specific HDAC inhibitors are rare, although HDACs 3, 6, and 8 specific inhibitors are in preclinical studies (Falkenberg & Johnstone, 2014). Our results suggest that a development of an HDAC10 isoform specific inhibitor could improve both the toxicity of current HDAC inhibitors and further increase platinum sensitivity in

BRCA1 deficient ovarian carcinoma tumors with a larger therapeutic window.

HDAC inhibitors have become an important class of drugs in the search for new therapies against cancer among other diseases. Several clinical trials are currently analyzing HDAC inhibitors as anti-tumor agents (A. C. West & Johnstone, 2014). Most of these inhibitors are broad spectrum inhibitors, affecting both class I and class II HDACs.

Although many HDAC inhibitor studies are in progress, most of these are based on

81 empirical data. The exact mechanisms through which these agents are producing their cytotoxic effects are currently not well understood.

Valproic acid (VPA) has been used to treat neurological disease such as epilepsy for more than 30 years. However, VPA is specific to class I HDACs, and has no effect on the class II HDAC10 (A. C. West & Johnstone, 2014). VPA had no effect in the HDR assay, supporting the concept that HDAC10 is the HDAC protein associated with the DNA repair activity (Kotian et al., 2011). VPA is currently in phase I and phase II clinical trials for leukemias and cervical cancer (Minucci & Pelicci, 2006). However, VPA does have neurological toxicity side effects, which limits its therapeutic window.

Romidepsin is FDA approved for the treatment of cutaneous T-cell chronic lymphoma (CTCL). Similar to VPA, romidepsin is a class I and II inhibitor, however it does not affect HDAC10 (A. C. West & Johnstone, 2014). Romidepsin is in several clinical trials, several studying its effect in conjunction with other therapies, including platinum based therapies (Bertino & Otterson, 2011).

The final FDA approved HDAC inhibitor is SAHA, also known as vorinostat.

Approved for the treatment of CTCL, SAHA is currently being studied in multiple phase I and phase II studies (Modesitt, Sill, Hoffman, Bender, & Group, 2008; A. C. West &

Johnstone, 2014). SAHA is a broad spectrum HDAC inhibitor, affecting both class I and II

HDACs. Unlike the other FDA approved HDAC inhibitors, it does inhibit HDAC10, albeit to a reduced effect compared to other HDACs (A. C. West & Johnstone, 2014). Out of

82 the current clinical trials, SAHA has potential to sensitize cancer cells to platinum therapy mediated by inhibition of HDAC10 (Ramalingam et al., 2010; Ramalingam et al.,

2007).

Although our analysis was limited to ovarian cells, the DNA repair mechanisms through which HDAC10 potentially modulates cisplatin sensitivity could be present in other cell lines. A recent clinical trial demonstrated SAHA enhanced the efficacy of platinum therapy in patients with non-small-cell lung cancer (Ramalingam et al., 2010;

Ramalingam et al., 2007). Similarly, platinum sensitivity is not mediated solely by

HDAC10. This is evidenced by the multiple trials where HDAC inhibition is in combination with platinum therapy as well as preclinical studies demonstrating other

HDACs mediating platinum sensitivity (M. G. Kim et al., 2012; Minucci & Pelicci, 2006;

Stronach et al., 2011).

HDAC inhibition carries tremendous potential in search for pharmacological solutions against cancer. HDAC inhibitors affect a large number of pathways and have already been proven to improve patient outcomes in certain cancers. Although there is an ever-growing library of work that is analyzing the mechanisms through which HDAC and their inhibition modulate their effects, many of their functions are still unknown

(Minucci & Pelicci, 2006). HDAC10 in particular is poorly understood, only recently being discovered to be involved in DNA repair (Kotian et al., 2011). Similarly, few of the current pharmacological solutions affect HDAC10 as strongly as other HDACs (A. C. West

& Johnstone, 2014). Our results provide evidence to explore HDAC10 specific inhibition 83 as an adjuvant therapy to platinum therapies, especially for BRCA1-deficient ovarian carcinoma patients.

3.6 Tables

Name Sequence GL2 sense 5’- CGU ACG CGG AAU ACU UCG ATT - 3’ GL2 antisense 5’- UCG AAG UAU UCC GCG UAC GTT - 3’ HDAC10-1 sense 5’ - CGG AGU CAG UGU GCA UGA CAG UAC A - 3’ HDAC10-1 antisense 5’- UGU ACU GUC AUG CAC ACU GAC UCC G - 3’ HDAC10-2 sense 5’- UCA CUG CAC UUG GGA AGC UCC UGU A - 3’ HDAC10-2 antisense 5’ - UAC AGG AGC UUC CCA AGU GCA GUG A - 3’ Table 4 siRNA sequences for HDAC10 study Sequences of siRNAs in study. dTdT are present on the 3’ end of every oligo.

Chapter 4: Summary and Future Directions

4.1 DNA repair and cancer

The studies in this document analyzed two DNA repair factors, BRCA1 and

HDAC10. Although BRCA1 and HDAC10 are both DNA repair factors, the context of how we study them represent two very different roles DNA repair factors have in cancer.

BRCA1 is a highly penetrant gene in which some mutations signify substantial lifetime risk of developing cancer (National Cancer Institute, 2015). We studied BRCA1 function with a novel DNA repair assay in an effort to provide functional data to BRCA1 variants that have unknown clinical significance. In the case of BRCA1, we studied DNA repair in the framework of identifying variants of a DNA repair factor that could potentially cause cancer, empowering the patient to consider prophylactic treatments. HDAC10 is a histone deacetylase with relatively few known functions. Despite only being recently identified as a DNA repair factor, our initial genetic screen found ovarian tumors exhibited a relatively high rate of HDAC10 deletion. More importantly, low HDAC10 expression was correlated with increased platinum therapy sensitivity in tumor samples

(Cerami et al., 2012; National Cancer Institute, 2016). We studied HDAC10 and general

HDAC inhibition as targets to improve the therapy of ovarian cancer patients. In the case of HDAC10, we studied how DNA repair factors can be targeted in the treatment of cancer.

DNA repair and cancer maintain a paradoxical relationship. The loss of DNA repair creates mutations that increase the chance of carcinogenesis. Simultaneously,

DNA repair factors are also targets in the treatment of cancer.

4.2 BRCA1

We studied BRCA1 in the context of DNA repair factors as tumor suppressors.

Specific BRCA1 mutations are associated with substantial lifetime risk of developing breast cancer. However, there are many BRCA1 variants that carry uncertain clinical significance. Outside of BRCA1 variants that exist in the population, thousands of BRCA1 missense mutations similarly have no known clinical risk (synthetic variants). Although synthetic BRCA1 variants have limited clinical significance individually, future VUS will be drawn from the current population of synthetic BRCA1 variants.

4.2a Highly Parallel Homology Directed Repair

We developed a novel assay to analyze as many synthetic and VUS variants of

BRCA1 in homologous recombination repair. We adapted a proven functional assay in

HeLa cells, the homology directed repair assay (Ransburgh et al., 2010), to simultaneously test a library population of BRCA1 missense mutation variants. We utilized flipase and flipase recognition target (Flp-FRT) recombination to perform parallel integrations of BRCA1 libraries generated in collaboration with Dr. Lea Starita at the

University of Washington. The FRT sequence ensures single variant integration and prevents position-effect variegation. The single variant integration serves both to ensure copy number expression effects are controlled but more importantly to ensure the DSB break challenge presented to the cell tests a single BRCA1 variant per cell. If multiple BRCA1 variants exist in any given cell, any functional BRCA1 sequence will promote HR repair, generating false positives. Utilizing fluorescent activated cell

88 sorting, we collected cells able to repair a DSB with HR and cells not able to repair the

DSB with HR. Using next generation DNA sequencing, we identified the BRCA1 variants in the HR competent and HR deficient populations. This novel assay, the highly parallel homology directed repair assay, is currently testing multiple RING domain BRCA1 variant libraries. In the traditional HDR assay, each cell essentially acts as an experimental replicate, and each experiment can analyze millions of replicates. In the novel HP-HDR assay, we can now simultaneously analyze thousands of variants with hundreds to thousands of replicates per variant.

4.2b BRCA1 Future Directions

The immediate future involves finishing the analysis of the RING domain BRCA1 libraries that have been sorted with HP-HDR and sequenced. Our next study would be to similarly analyze the BRCT domain of BRCA1. BRCT shares the clinical characteristic of the RING domain that pathogenic BRCA1 mutations cluster in the RING and BRCT domains. The RING domain library analysis was the culmination of several years of continuous optimization of each step of the assay. We are confident that the BRCT domain analysis can benefit from these optimizations and the analysis can be performed in a fraction of the time.

BRCA1 provides several known mutations, both pathogenic and benign, that serve as internal controls for our HP-HDR analysis. Once the complete assay has been proven, we hope to use it for other repair factors. An excellent candidate is the known

BRCA1 interactor BARD1. BARD1 shares homology with BRCA1 RING and BRCT domains 89 and exists in vivo as a heterodimer with BRCA1. Despite this, relatively few BARD1 mutations are associated with cancer. Recently we have studied multiple BARD1 mutations in HDR (C. Lee et al., 2015). Analysis of BARD1 with HP-HDR would provide a more complete structure-function analysis of BARD1. The analysis could identify important BARD1 structures or amino acids that are sensitive to mutation with regard to

HR function.

A longer term goal would be to adapt the concepts of parallel functional assays to similar assays. The three core components of HP-HDR are a functional assay that operates at the cellular level, a reporter signal that can be used to sort the cells, and a variant library coupled with a protocol to identify variants in a function sorted population. Several GFP reporter based DNA repair assays are viable candidates to be adapted for parallel analysis (Bennardo, Cheng, Huang, & Stark, 2008; Pierce et al.,

1999; Towler et al., 2012; Zhuang, Jiang, Willers, & Xia, 2009).

Finally, the gene of interest of this study is BRCA1, which was selected in part due its clinical relevance. Although we have evidence that BRCA1 HDR function is specific and selective with regards to clinical risk, a clinical trial is required assign actual clinical value to any test. We hope to submit an observational clinical trial analyzing the results of the HP-HDR assay and assessing the predictive power of HP-HDR scores in determining pathogenicity. Importantly, the assessment would be for both known existing variants and future variants that are already analyzed in HP-HDR as a synthetic

90 variant. An observational clinical trial is a longer term goal as the statistics require patient data in order to estimate risk.

4.3 HDAC10

Histone deacetylase 10 has only recently been implicated in DNA repair (Kotian et al., 2011). However, general HDAC inhibitors are already in clinical trials for the treatment of cancer (Matulonis et al., 2015; A. C. West & Johnstone, 2014). We studied if HDAC10 was associated with cancer and if HDAC10 could be a target for cancer therapy.

4.3a HDAC10 Results

Our initial study analyzed HDAC10 genetic changes in the cancer database, TCGA.

A large ovarian serous cystadenocarcinoma tumor dataset (n>550) indicated that

HDAC10 is deleted in 5-10% of ovarian tumors, relatively much higher than other available tumor datasets (Cerami et al., 2012). The first line treatment of ovarian carcinomas is typically platinum based therapy (Agarwal & Kaye, 2003). Platinum based therapy causes intrastrand DNA crosslinks and bulky adducts on DNA that interfere with signals and processes needed for cell proliferation (Jung & Lippard, 2007). Previous studies have shown that inhibition of HR enhanced the efficacy of cisplatin, a platinum therapy drug (Wang et al., 2005). Analyzing HDAC10 expression data, we found a significant correlation between tumors that were sensitive to platinum therapy with less

HDAC10 expression (Cerami et al., 2012).

Sub-lethal concentrations of cisplatin inhibited HR in our HDR assay. More importantly, this inhibition could be further enhanced using HDAC inhibitors. The enhanced cisplatin inhibition of HDR suggested a potential mechanism of action of general HDAC inhibitors in cancer therapy is mediated through HDAC10 and HR. We performed a comet assay on UWB1.289 ovarian carcinoma cells. This BRCA1 deficient ovarian cancer cell line exhibited insignificant DNA damage after ionizing radiation.

Depleting the cell line of HDAC10 demonstrated significant DNA damage after ionizing radiation. This result suggested that HDAC10 can sensitize BRCA1-deficient ovarian cells to ionizing radiation. We postulated that the UWB1.289 cancer cells may have compensated for the DNA repair functions of BRCA1. Additionally, this repair function of the cell line is still inhibited by depletion of HDAC10. Indeed, HDAC10 depletion of

UWB1.289 cells demonstrated significant enhancement of cisplatin efficacy as shown in an MTT assay. We lastly performed two additional MTT assays using primary ovarian cells, one assay using cells derived form a cisplatin sensitive tumor and the other using cells from a cisplatin resistant tumor. Unfortunately, we were unable to specifically inhibit HDAC10 since the cells could not be cultured for more than two to three passages. We used general HDACi and demonstrated that HDACi will enhance cisplatin sensitivity in both cisplatin resistant and sensitive primary ovarian cancer cells. Based on these results we proposed that HDAC10 inhibition is a potential adjuvant to platinum therapy in BRCA1 deficient ovarian cancers.

4.3b HDAC10 Future Directions

General HDAC inhibitors are currently being studied in combination with platinum therapy in ovarian cancers in a phase II clinical trial (Modesitt et al., 2008).

Ideally, a modification to the study would analyze efficacy of HDAC inhibition with respect to BRCA1 status in the ovarian tumors. This would not change the selection parameters of the clinical trial nor the overall outcome, but could potentially identify a sub-population of ovarian tumors that are sensitized to HDAC inhibition in combination with cisplatin due to HR inhibition.

Another study we would like to explore is to characterize HDAC10 inhibition with respect to BRCA1. The UWB1.289 BRCA1 deficient cell line demonstrated insignificant

DNA damage after ionizing radiation. Although the UWB1.289 cells may have compensated for the loss of BRCA1, persistent DNA damage upon depletion of HDAC10 could still indicate that HDAC10 mediates HR function in a BRCA1-independent manner.

Clinically speaking, this could sensitize other DNA repair targeting therapies, such as

PARP inhibition. Experimentally, we would like to explore the effects BRCA1 and

HDAC10 depletion in combination would have in a variety of repair assays and MTT proliferation assays.

4.4 BRCA1 and HDAC10

The studies presented in this dissertation analyze two different DNA repair factors and in two very different studies. In the HP-HDR study, we wished to transform how we study proteins, leveraging next generation technologies and concepts in a 93 functional assay with significant specificity and sensitivity to clinical risk. We developed a novel method to study thousands of protein variants in a single experiment with respect to a high level cellular function in HR repair. The implications of this assay are far reaching, from transforming how we study protein structure-function to directly providing present and future patients carrying BRCA1 VUS with functional data.

HDAC10 told a different story. HDAC10 has few known functions outside of DNA repair but is targeted by pharmaceutical HDAC inhibitors that have been approved from some cancers and in clinical trials for more. The HDAC10 study progressed from DNA repair factor to genetic link to potential target in cancer therapy. The impact of this study could be seen immediately, potentially affecting the analysis of an on-going phase

II clinical trial. Longer term, the elucidation of HDAC10 function could provide critical insight into the mechanisms of HDAC inhibitor and they can best be used in the fight against cancer.

References

Agarwal, R., & Kaye, S. B. (2003). Ovarian cancer: strategies for overcoming resistance to chemotherapy. Nat Rev Cancer, 3(7), 502-516. doi:10.1038/nrc1123

ALLFREY, V. G., FAULKNER, R., & MIRSKY, A. E. (1964). ACETYLATION AND METHYLATION OF HISTONES AND THEIR POSSIBLE ROLE IN THE REGULATION OF RNA SYNTHESIS. Proc Natl Acad Sci U S A, 51, 786-794.

Antoniou, A., Pharoah, P. D., Narod, S., Risch, H. A., Eyfjord, J. E., Hopper, J. L., . . . Easton, D. F. (2003). Average risks of breast and ovarian cancer associated with BRCA1 or BRCA2 mutations detected in case Series unselected for family history: a combined analysis of 22 studies. Am J Hum Genet, 72(5), 1117-1130. doi:10.1086/375033

Arany, I., & Safirstein, R. L. (2003). Cisplatin nephrotoxicity. Semin Nephrol, 23(5), 460-464.

Bannister, A. J., & Kouzarides, T. (2011). Regulation of chromatin by histone modifications. Cell Res, 21(3), 381-395. doi:10.1038/cr.2011.22

Bennardo, N., Cheng, A., Huang, N., & Stark, J. M. (2008). Alternative-NHEJ is a mechanistically distinct pathway of mammalian chromosome break repair. PLoS Genet, 4(6), e1000110. doi:10.1371/journal.pgen.1000110

Bensch, K. G., & Malawista, S. E. (1968). Microtubule crystals: a new biophysical phenomenon induced by Vinca alkaloids. Nature, 218(5147), 1176-1177.

Bertino, E. M., & Otterson, G. A. (2011). Romidepsin: a novel histone deacetylase inhibitor for cancer. Expert Opin Investig Drugs, 20(8), 1151-1158. doi:10.1517/13543784.2011.594437

Bhaskara, S., Knutson, S. K., Jiang, G., Chandrasekharan, M. B., Wilson, A. J., Zheng, S., . . . Hiebert, S. W. (2010). Hdac3 is essential for the maintenance of chromatin structure and genome stability. Cancer Cell, 18(5), 436-447. doi:10.1016/j.ccr.2010.10.022

Bosl, G. J., Gluckman, R., Geller, N. L., Golbey, R. B., Whitmore, W. F., Herr, H., . . . Bains, M. (1986). VAB-6: an effective chemotherapy regimen for patients with germ-cell tumors. J Clin Oncol, 4(10), 1493-1499.

Bouwman, P., van der Gulden, H., van der Heijden, I., Drost, R., Klijn, C. N., Prasetyanti, P., . . . Jonkers, J. (2013). A high-throughput functional complementation assay for classification of BRCA1 missense variants. Cancer Discov, 3(10), 1142-1155. doi:10.1158/2159- 8290.cd-13-0094

Boveri, T. (2008). Concerning the origin of malignant tumours by Theodor Boveri. Translated and annotated by Henry Harris. J Cell Sci, 121 Suppl 1, 1-84. doi:10.1242/jcs.025742

Brownell, J. E., Zhou, J., Ranalli, T., Kobayashi, R., Edmondson, D. G., Roth, S. Y., & Allis, C. D. (1996). Tetrahymena histone acetyltransferase A: a homolog to yeast Gcn5p linking histone acetylation to gene activation. Cell, 84(6), 843-851.

Bunting, S. F., Callén, E., Wong, N., Chen, H. T., Polato, F., Gunn, A., . . . Nussenzweig, A. (2010). 53BP1 inhibits homologous recombination in Brca1-deficient cells by blocking resection of DNA breaks. Cell, 141(2), 243-254. doi:10.1016/j.cell.2010.03.012

Burdette, W. J. (1955). The significance of mutation in relation to the origin of tumors: a review. Cancer Res, 15(4), 201-226.

Campeau, P. M., Foulkes, W. D., & Tischkowitz, M. D. (2008). Hereditary breast cancer: new genetic developments, new therapeutic avenues. Hum Genet, 124(1), 31-42. doi:10.1007/s00439-008-0529-1

CDC. (2012). Breast Cancer Statistics. Retrieved from http://www.cdc.gov/cancer/breast/statistics/index.htm

Cerami, E., Gao, J., Dogrusoz, U., Gross, B. E., Sumer, S. O., Aksoy, B. A., . . . Schultz, N. (2012). The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov, 2(5), 401-404. doi:10.1158/2159-8290.CD-12- 0095

Chabner, B. A., & Roberts, T. G. (2005). Timeline: Chemotherapy and the war on cancer. Nat Rev Cancer, 5(1), 65-72. doi:10.1038/nrc1529

Chen, B. Y., Huang, C. H., Lin, Y. H., Huang, C. C., Deng, C. X., & Hsu, L. C. (2014). The K898E germline variant in the PP1-binding motif of BRCA1 causes defects in DNA Repair. Sci Rep, 4, 5812. doi:10.1038/srep05812

Chen, S., & Parmigiani, G. (2007). Meta-analysis of BRCA1 and BRCA2 penetrance. J Clin Oncol, 25(11), 1329-1333. doi:10.1200/JCO.2006.09.1066

Chen, X. B., Melchionna, R., Denis, C. M., Gaillard, P. H., Blasina, A., Van de Weyer, I., . . . McGowan, C. H. (2001). Human Mus81-associated endonuclease cleaves Holliday junctions in vitro. Mol Cell, 8(5), 1117-1127.

Chen, Z., Yang, H., & Pavletich, N. P. (2008). Mechanism of homologous recombination from the RecA-ssDNA/dsDNA structures. Nature, 453(7194), 489-484. doi:10.1038/nature06971

Chien, A., Edgar, D. B., & Trela, J. M. (1976). Deoxyribonucleic acid polymerase from the extreme thermophile Thermus aquaticus. J Bacteriol, 127(3), 1550-1557.

Choudhary, C., Kumar, C., Gnad, F., Nielsen, M. L., Rehman, M., Walther, T. C., . . . Mann, M. (2009). Lysine acetylation targets protein complexes and co-regulates major cellular functions. Science, 325(5942), 834-840. doi:10.1126/science.1175371

Christensen, D. E., Brzovic, P. S., & Klevit, R. E. (2007). E2-BRCA1 RING interactions dictate synthesis of mono- or specific polyubiquitin chain linkages. Nat Struct Mol Biol, 14(10), 941-948. doi:10.1038/nsmb1295

Ciccia, A., & Elledge, S. J. (2010). The DNA damage response: making it safe to play with knives. Mol Cell, 40(2), 179-204. doi:10.1016/j.molcel.2010.09.019 96

Cleaver, J. E. (1968). Defective repair replication of DNA in xeroderma pigmentosum. Nature, 218(5142), 652-656.

Cleaver, J. E. (1969). Xeroderma pigmentosum: a human disease in which an initial stage of DNA repair is defective. Proc Natl Acad Sci U S A, 63(2), 428-435.

Dietrich, C. S., 3rd, Greenberg, V. L., DeSimone, C. P., Modesitt, S. C., van Nagell, J. R., Craven, R., & Zimmer, S. G. (2010). Suberoylanilide hydroxamic acid (SAHA) potentiates paclitaxel-induced apoptosis in ovarian cancer cell lines. Gynecol Oncol, 116(1), 126- 130. doi:10.1016/j.ygyno.2009.09.039

Easton, D. F. (1999). How many more breast cancer predisposition genes are there? Breast Cancer Res, 1(1), 14-17.

Eggington, J. M., Bowles, K. R., Moyes, K., Manley, S., Esterling, L., Sizemore, S., . . . Wenstrup, R. J. (2014). A comprehensive laboratory-based program for classification of variants of uncertain significance in hereditary cancer genes. Clin Genet, 86(3), 229-237. doi:10.1111/cge.12315

Falkenberg, K. J., & Johnstone, R. W. (2014). Histone deacetylases and their inhibitors in cancer, neurological diseases and immune disorders. Nat Rev Drug Discov, 13(9), 673-691. doi:10.1038/nrd4360

FARBER, S., & DIAMOND, L. K. (1948). Temporary remissions in acute leukemia in children produced by folic acid antagonist, 4-aminopteroyl-glutamic acid. N Engl J Med, 238(23), 787-793. doi:10.1056/NEJM194806032382301

Fields, S., & Song, O. (1989). A novel genetic system to detect protein-protein interactions. Nature, 340(6230), 245-246. doi:10.1038/340245a0

Futreal, P. A., Liu, Q., Shattuck-Eidens, D., Cochran, C., Harshman, K., Tavtigian, S., . . . et al. (1994). BRCA1 mutations in primary breast and ovarian carcinomas. Science, 266(5182), 120-122.

Gao, J., Aksoy, B. A., Dogrusoz, U., Dresdner, G., Gross, B., Sumer, S. O., . . . Schultz, N. (2013). Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal, 6(269), pl1. doi:10.1126/scisignal.2004088

GILMAN, A. (1963). The initial clinical trial of nitrogen mustard. Am J Surg, 105, 574-578.

Gilmour, D. S., & Lis, J. T. (1984). Detecting protein-DNA interactions in vivo: distribution of RNA polymerase on specific bacterial genes. Proc Natl Acad Sci U S A, 81(14), 4275-4279.

Glozak, M. A., Sengupta, N., Zhang, X., & Seto, E. (2005). Acetylation and deacetylation of non- histone proteins. Gene, 363, 15-23. doi:10.1016/j.gene.2005.09.010

Goetz, J. D., Motycka, T. A., Han, M., Jasin, M., & Tomkinson, A. E. (2005). Reduced repair of DNA double-strand breaks by homologous recombination in a DNA ligase I-deficient human cell line. DNA Repair (Amst), 4(6), 649-654. doi:10.1016/j.dnarep.2005.02.004

Goodarzi, A. A., & Jeggo, P. A. (2013). The repair and signaling responses to DNA double-strand breaks. Adv Genet, 82, 1-45. doi:10.1016/b978-0-12-407676-1.00001-9

Greenlee, R. T., Hill-Harmon, M. B., Murray, T., & Thun, M. (2001). Cancer statistics, 2001. CA Cancer J Clin, 51(1), 15-36.

Gregoretti, I. V., Lee, Y. M., & Goodson, H. V. (2004). Molecular evolution of the histone deacetylase family: functional implications of phylogenetic analysis. J Mol Biol, 338(1), 17-31. doi:10.1016/j.jmb.2004.02.006

Guidugli, L., Carreira, A., Caputo, S. M., Ehlen, A., Galli, A., Monteiro, A. N., . . . consortium, E. (2014). Functional assays for analysis of variants of uncertain significance in BRCA2. Hum Mutat, 35(2), 151-164. doi:10.1002/humu.22478

Haince, J. F., Kozlov, S., Dawson, V. L., Dawson, T. M., Hendzel, M. J., Lavin, M. F., & Poirier, G. G. (2007). Ataxia telangiectasia mutated (ATM) signaling network is modulated by a novel poly(ADP-ribose)-dependent pathway in the early response to DNA-damaging agents. J Biol Chem, 282(22), 16441-16453. doi:10.1074/jbc.M608406200

Harper, J. W., & Elledge, S. J. (2007). The DNA damage response: ten years after. Mol Cell, 28(5), 739-745. doi:10.1016/j.molcel.2007.11.015

Hartmann , L. C., Schaid , D. J., Woods , J. E., Crotty , T. P., Myers , J. L., Arnold , P. G., . . . Jenkins , R. B. (1999). Efficacy of Bilateral Prophylactic Mastectomy in Women with a Family History of Breast Cancer. New England Journal of Medicine, 340(2), 77-84. doi:doi:10.1056/NEJM199901143400201

Heideman, M. R., Wilting, R. H., Yanover, E., Velds, A., de Jong, J., Kerkhoven, R. M., . . . Dannenberg, J. H. (2013). Dosage-dependent tumor suppression by histone deacetylases 1 and 2 through regulation of c-Myc collaborating genes and p53 function. Blood, 121(11), 2038-2050. doi:10.1182/blood-2012-08-450916

Heinzel, T., Lavinsky, R. M., Mullen, T. M., Söderstrom, M., Laherty, C. D., Torchia, J., . . . Rosenfeld, M. G. (1997). A complex containing N-CoR, mSin3 and histone deacetylase mediates transcriptional repression. Nature, 387(6628), 43-48. doi:10.1038/387043a0

Helleday, T., Petermann, E., Lundin, C., Hodgson, B., & Sharma, R. A. (2008). DNA repair pathways as targets for cancer therapy. Nat Rev Cancer, 8(3), 193-204. doi:10.1038/nrc2342

Hiatt, J. B., Patwardhan, R. P., Turner, E. H., Lee, C., & Shendure, J. (2010). Parallel, tag- directed assembly of locally derived short sequence reads. Nat Methods, 7(2), 119-122. doi:10.1038/nmeth.1416

Holliday, R. (1964). A mechanism for gene conversion in fungi. Genetical Research, 5(02), 282- 304.

Huen, M. S., Sy, S. M., & Chen, J. (2010). BRCA1 and its toolbox for the maintenance of genome integrity. Nat Rev Mol Cell Biol, 11(2), 138-148. doi:10.1038/nrm2831

Ip, S. C., Rass, U., Blanco, M. G., Flynn, H. R., Skehel, J. M., & West, S. C. (2008). Identification of Holliday junction resolvases from humans and yeast. Nature, 456(7220), 357-361.

Jain, P. C., & Varadarajan, R. (2014). A rapid, efficient, and economical inverse polymerase chain reaction-based method for generating a site saturation mutant library. Anal Biochem, 449, 90-98. doi:10.1016/j.ab.2013.12.002 98

Jinek, M., Chylinski, K., Fonfara, I., Hauer, M., Doudna, J. A., & Charpentier, E. (2012). A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science, 337(6096), 816-821. doi:10.1126/science.1225829

JOHNSON, I. S., ARMSTRONG, J. G., GORMAN, M., & BURNETT, J. P. (1963). THE VINCA ALKALOIDS: A NEW CLASS OF ONCOLYTIC AGENTS. Cancer Res, 23, 1390-1427.

Johnson, R. D., & Jasin, M. (2000). Sister chromatid gene conversion is a prominent double- strand break repair pathway in mammalian cells. Embo j, 19(13), 3398-3407. doi:10.1093/emboj/19.13.3398

Johnson, R. D., & Jasin, M. (2001). Double-strand-break-induced homologous recombination in mammalian cells. Biochem Soc Trans, 29(Pt 2), 196-201.

Jolivet, J., Cowan, K. H., Curt, G. A., Clendeninn, N. J., & Chabner, B. A. (1983). The pharmacology and clinical use of methotrexate. N Engl J Med, 309(18), 1094-1104. doi:10.1056/NEJM198311033091805

Jones, P. L., Veenstra, G. J., Wade, P. A., Vermaak, D., Kass, S. U., Landsberger, N., . . . Wolffe, A. P. (1998). Methylated DNA and MeCP2 recruit histone deacetylase to repress transcription. Nat Genet, 19(2), 187-191. doi:10.1038/561

Jung, Y., & Lippard, S. J. (2007). Direct cellular responses to platinum-induced DNA damage. Chemical reviews, 107(5), 1387-1407.

Kastan, M. B., Zhan, Q., el-Deiry, W. S., Carrier, F., Jacks, T., Walsh, W. V., . . . Fornace, A. J. (1992). A mammalian cell cycle checkpoint pathway utilizing p53 and GADD45 is defective in ataxia-telangiectasia. Cell, 71(4), 587-597.

Khanna, K. K., & Jackson, S. P. (2001). DNA double-strand breaks: signaling, repair and the cancer connection. Nat Genet, 27(3), 247-254. doi:10.1038/85798

Kim, J. S., Krasieva, T. B., Kurumizaka, H., Chen, D. J., Taylor, A. M., & Yokomori, K. (2005). Independent and sequential recruitment of NHEJ and HR factors to DNA damage sites in mammalian cells. J Cell Biol, 170(3), 341-347. doi:10.1083/jcb.200411083

Kim, M. G., Pak, J. H., Choi, W. H., Park, J. Y., Nam, J. H., & Kim, J. H. (2012). The relationship between cisplatin resistance and histone deacetylase isoform overexpression in epithelial ovarian cancer cell lines. J Gynecol Oncol, 23(3), 182-189. doi:10.3802/jgo.2012.23.3.182

Kotian, S., Liyanarachchi, S., Zelent, A., & Parvin, J. D. (2011). Histone deacetylases 9 and 10 are required for homologous recombination. J Biol Chem, 286(10), 7722-7726. doi:10.1074/jbc.C110.194233

Kouzarides, T. (2007). Chromatin modifications and their function. Cell, 128(4), 693-705. doi:10.1016/j.cell.2007.02.005

Ku, C. S., Loy, E. Y., Pawitan, Y., & Chia, K. S. (2010). The pursuit of genome-wide association studies: where are we now? J Hum Genet, 55(4), 195-206. doi:10.1038/jhg.2010.19

Leanderson, P., & Tagesson, C. (1992). Cigarette smoke-induced DNA damage in cultured human lung cells: role of hydroxyl radicals and endonuclease activation. Chem Biol Interact, 81(1-2), 197-208.

Lee, C., Banerjee, T., Gillespie, J., Ceravolo, A., Parvinsmith, M. R., Starita, L. M., . . . Parvin, J. D. (2015). Functional Analysis of BARD1 Missense Variants in Homology-Directed Repair of DNA Double Strand Breaks. Hum Mutat, 36(12), 1205-1214. doi:10.1002/humu.22902

Lee, J. H., & Paull, T. T. (2005). ATM activation by DNA double-strand breaks through the Mre11- Rad50-Nbs1 complex. Science, 308(5721), 551-554. doi:10.1126/science.1108297

Lee, R. C., Feinbaum, R. L., & Ambros, V. (1993). The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell, 75(5), 843-854.

Li, M., & Yu, X. (2013). Function of BRCA1 in the DNA damage response is mediated by ADP- ribosylation. Cancer Cell, 23(5), 693-704. doi:10.1016/j.ccr.2013.03.025

LI, M. C., HERTZ, R., & BERGENSTAL, D. M. (1958). Therapy of choriocarcinoma and related trophoblastic tumors with folic acid and purine antagonists. N Engl J Med, 259(2), 66-74. doi:10.1056/NEJM195807102590204

Lindor, N. M., Goldgar, D. E., Tavtigian, S. V., Plon, S. E., & Couch, F. J. (2013). BRCA1/2 sequence variants of uncertain significance: a primer for providers to assist in discussions and in medical management. Oncologist, 18(5), 518-524. doi:10.1634/theoncologist.2012-0452

Lu, C., Xie, M., Wendl, M. C., Wang, J., McLellan, M. D., Leiserson, M. D., . . . Ding, L. (2015). Patterns and functional implications of rare germline variants across 12 cancer types. Nat Commun, 6, 10086. doi:10.1038/ncomms10086

Mahaney, B. L., Meek, K., & Lees-Miller, S. P. (2009). Repair of ionizing radiation-induced DNA double-strand breaks by non-homologous end-joining. Biochem J, 417(3), 639-650. doi:10.1042/BJ20080413

Maher, B. (2008). Personal genomes: The case of the missing heritability Nature (Vol. 456, pp. 18-21). England.

Maloisel, L., Fabre, F., & Gangloff, S. (2008). DNA polymerase delta is preferentially recruited during homologous recombination to promote heteroduplex DNA extension. Mol Cell Biol, 28(4), 1373-1382. doi:10.1128/MCB.01651-07

Matulonis, U., Berlin, S., Lee, H., Whalen, C., Obermayer, E., Penson, R., . . . Horowitz, N. (2015). Phase I study of combination of vorinostat, carboplatin, and gemcitabine in women with recurrent, platinum-sensitive epithelial ovarian, fallopian tube, or peritoneal cancer. Cancer Chemother Pharmacol, 76(2), 417-423. doi:10.1007/s00280-015-2813-9

Meek, K., Dang, V., & Lees-Miller, S. P. (2008). DNA-PK: the means to justify the ends? Adv Immunol, 99, 33-58. doi:10.1016/s0065-2776(08)00602-0

Mendivil, A. A., Micha, J. P., Brown, J. V., Rettenmaier, M. A., Abaid, L. N., Lopez, K. L., & Goldstein, B. H. (2013). Increased incidence of severe gastrointestinal events with first- line paclitaxel, carboplatin, and vorinostat chemotherapy for advanced-stage epithelial 100

ovarian, primary peritoneal, and fallopian tube cancer. Int J Gynecol Cancer, 23(3), 533- 539. doi:10.1097/IGC.0b013e31828566f1

Meza, J. E., Brzovic, P. S., King, M. C., & Klevit, R. E. (1999). Mapping the functional domains of BRCA1. Interaction of the ring finger domains of BRCA1 and BARD1. J Biol Chem, 274(9), 5659-5665.

Minucci, S., & Pelicci, P. G. (2006). Histone deacetylase inhibitors and the promise of epigenetic (and more) treatments for cancer. Nat Rev Cancer, 6(1), 38-51. doi:10.1038/nrc1779

Modesitt, S. C., Sill, M., Hoffman, J. S., Bender, D. P., & Group, G. O. (2008). A phase II study of vorinostat in the treatment of persistent or recurrent epithelial ovarian or primary peritoneal carcinoma: a Gynecologic Oncology Group study. Gynecol Oncol, 109(2), 182- 186. doi:10.1016/j.ygyno.2008.01.009

Mojica, F. J., Diez-Villasenor, C., Garcia-Martinez, J., & Soria, E. (2005). Intervening sequences of regularly spaced prokaryotic repeats derive from foreign genetic elements. J Mol Evol, 60(2), 174-182. doi:10.1007/s00239-004-0046-3

Moynahan, M. E., & Jasin, M. (2010). Mitotic homologous recombination maintains genomic stability and suppresses tumorigenesis. Nat Rev Mol Cell Biol, 11(3), 196-207. doi:10.1038/nrm2851

Nan, X., Ng, H. H., Johnson, C. A., Laherty, C. D., Turner, B. M., Eisenman, R. N., & Bird, A. (1998). Transcriptional repression by the methyl-CpG-binding protein MeCP2 involves a histone deacetylase complex. Nature, 393(6683), 386-389. doi:10.1038/30764

Nandhakumar, S., Parasuraman, S., Shanmugam, M. M., Rao, K. R., Chand, P., & Bhat, B. V. (2011). Evaluation of DNA damage using single-cell gel electrophoresis (Comet Assay). J Pharmacol Pharmacother, 2(2), 107-111. doi:10.4103/0976-500X.81903

National Cancer Institute. (2015). SEER Stat Fact Sheets: Female Breast Cancer. Retrieved from http://seer.cancer.gov/statfacts/html/breast.html

National Cancer Institute. (2016). The Cancer Genome Atlas.

NHGRI. (2015). All About The Humane Genome Project. Retrieved from http://www.genome.gov/27562713

Nick McElhinny, S. A., Snowden, C. M., McCarville, J., & Ramsden, D. A. (2000). Ku recruits the XRCC4-ligase IV complex to DNA ends. Mol Cell Biol, 20(9), 2996-3003.

Pacific Biosciences. pbccs - Generate Accurate Consensus Sequences from a Single SMRTbell. Retrieved from https://github.com/PacificBiosciences/pbccs

Parvin, J., Chiba, N., & Ransburgh, D. (2011). Identifying the effects of BRCA1 mutations on homologous recombination using cells that express endogenous wild-type BRCA1. J Vis Exp(48). doi:10.3791/2468

Patel, K. J., Yu, V. P., Lee, H., Corcoran, A., Thistlethwaite, F. C., Evans, M. J., . . . Venkitaraman, A. R. (1998). Involvement of Brca2 in DNA repair. Mol Cell, 1(3), 347-357.

101

Paull, T. T., Cortez, D., Bowers, B., Elledge, S. J., & Gellert, M. (2001). Direct DNA binding by Brca1. Proc Natl Acad Sci U S A, 98(11), 6086-6091. doi:10.1073/pnas.111125998

Paull, T. T., Rogakou, E. P., Yamazaki, V., Kirchgessner, C. U., Gellert, M., & Bonner, W. M. (2000). A critical role for histone H2AX in recruitment of repair factors to nuclear foci after DNA damage. Curr Biol, 10(15), 886-895.

Pierce, A. J., Johnson, R. D., Thompson, L. H., & Jasin, M. (1999). XRCC3 promotes homology- directed repair of DNA damage in mammalian cells. Genes Dev, 13(20), 2633-2638.

Raaphorst, G. P., Leblanc, M., & Li, L. F. (2005). A comparison of response to cisplatin, radiation and combined treatment for cells deficient in recombination repair pathways. Anticancer Res, 25(1A), 53-58.

Ramalingam, S. S., Maitland, M. L., Frankel, P., Argiris, A. E., Koczywas, M., Gitlitz, B., . . . Belani, C. P. (2010). Carboplatin and Paclitaxel in combination with either vorinostat or placebo for first-line therapy of advanced non-small-cell lung cancer. J Clin Oncol, 28(1), 56-62. doi:10.1200/JCO.2009.24.9094

Ramalingam, S. S., Parise, R. A., Ramanathan, R. K., Ramananthan, R. K., Lagattuta, T. F., Musguire, L. A., . . . Belani, C. P. (2007). Phase I and pharmacokinetic study of vorinostat, a histone deacetylase inhibitor, in combination with carboplatin and paclitaxel for advanced solid malignancies. Clin Cancer Res, 13(12), 3605-3610. doi:10.1158/1078- 0432.CCR-07-0162

Ransburgh, D. J., Chiba, N., Ishioka, C., Toland, A. E., & Parvin, J. D. (2010). Identification of breast tumor mutations in BRCA1 that abolish its function in homologous DNA recombination. Cancer Res, 70(3), 988-995. doi:10.1158/0008-5472.CAN-09-2850

Rosenberg, B., VanCamp, L., Trosko, J. E., & Mansour, V. H. (1969). Platinum compounds: a new class of potent antitumour agents. Nature, 222(5191), 385-386.

Roy, R., Chun, J., & Powell, S. N. (2012). BRCA1 and BRCA2: different roles in a common pathway of genome protection. Nat Rev Cancer, 12(1), 68-78. doi:10.1038/nrc3181

Saleh-Gohari, N., & Helleday, T. (2004). Conservative homologous recombination preferentially repairs DNA double-strand breaks in the S phase of the cell cycle in human cells. Nucleic Acids Res, 32(12), 3683-3688. doi:10.1093/nar/gkh703

Sancar, A., Lindsey-Boltz, L. A., Unsal-Kacmaz, K., & Linn, S. (2004). Molecular mechanisms of mammalian DNA repair and the DNA damage checkpoints. Annu Rev Biochem, 73, 39- 85. doi:10.1146/annurev.biochem.73.011303.073723

Savitsky, K., Bar-Shira, A., Gilad, S., Rotman, G., Ziv, Y., Vanagaite, L., . . . Shiloh, Y. (1995). A single ataxia telangiectasia gene with a product similar to PI-3 kinase. Science, 268(5218), 1749-1753.

Sawada, S., Yokokura, T., & Miyasaka, T. (1995). Synthesis and antitumor activity of A-ring or E- lactone modified water-soluble prodrugs of 20 (S)-camptothecin, including development of irinotecan hydrochloride trihydrate (CPT-11). Curr Pharm Design, 1, 113-132.

Schreiber, V., Dantzer, F., Ame, J. C., & de Murcia, G. (2006). Poly(ADP-ribose): novel functions for an old molecule. Nat Rev Mol Cell Biol, 7(7), 517-528. doi:10.1038/nrm1963 102

Scully, R., Chen, J., Ochs, R. L., Keegan, K., Hoekstra, M., Feunteun, J., & Livingston, D. M. (1997). Dynamic changes of BRCA1 subnuclear location and phosphorylation state are initiated by DNA damage. Cell, 90(3), 425-435.

Scully, R., Chen, J., Plug, A., Xiao, Y., Weaver, D., Feunteun, J., . . . Livingston, D. M. (1997). Association of BRCA1 with Rad51 in mitotic and meiotic cells. Cell, 88(2), 265-275.

Sharma, S., Javadekar, S. M., Pandey, M., Srivastava, M., Kumari, R., & Raghavan, S. C. (2015). Homology and enzymatic requirements of microhomology-dependent alternative end joining. Cell Death Dis, 6, e1697. doi:10.1038/cddis.2015.58

Spange, S., Wagner, T., Heinzel, T., & Krämer, O. H. (2009). Acetylation of non-histone proteins modulates cellular signalling at multiple levels. Int J Biochem Cell Biol, 41(1), 185-198. doi:10.1016/j.biocel.2008.08.027

Starita, L. M., & Parvin, J. D. (2003). The multiple nuclear functions of BRCA1: transcription, ubiquitination and DNA repair. Curr Opin Cell Biol, 15(3), 345-350.

Starita, L. M., Pruneda, J. N., Lo, R. S., Fowler, D. M., Kim, H. J., Hiatt, J. B., . . . Klevit, R. E. (2013). Activity-enhancing mutations in an E3 ubiquitin ligase identified by high- throughput mutagenesis. Proc Natl Acad Sci U S A, 110(14), E1263-1272. doi:10.1073/pnas.1303309110

Starita, L. M., Young, D. L., Islam, M., Kitzman, J. O., Gullingsrud, J., Hause, R. J., . . . Fields, S. (2015). Massively Parallel Functional Analysis of BRCA1 RING Domain Variants. Genetics, 200(2), 413-422. doi:10.1534/genetics.115.175802

Stark, J. M., Pierce, A. J., Oh, J., Pastink, A., & Jasin, M. (2004). Genetic steps of mammalian homologous repair with distinct mutagenic consequences. Mol Cell Biol, 24(21), 9305- 9316. doi:10.1128/MCB.24.21.9305-9316.2004

Stronach, E. A., Alfraidi, A., Rama, N., Datler, C., Studd, J. B., Agarwal, R., . . . Gabra, H. (2011). HDAC4-regulated STAT1 activation mediates platinum resistance in ovarian cancer. Cancer Res, 71(13), 4412-4422. doi:10.1158/0008-5472.CAN-10-4111

Sudhakar, A. (2009). History of Cancer, Ancient and Modern Treatment Methods. J Cancer Sci Ther, 1(2), 1-4. doi:10.4172/1948-5956.100000e2

Sy, S. M., Huen, M. S., & Chen, J. (2009). PALB2 is an integral component of the BRCA complex required for homologous recombination repair. Proc Natl Acad Sci U S A, 106(17), 7155- 7160. doi:10.1073/pnas.0811159106

Taunton, J., Hassig, C. A., & Schreiber, S. L. (1996). A mammalian histone deacetylase related to the yeast transcriptional regulator Rpd3p. Science, 272(5260), 408-411.

Tice, R. R., Agurell, E., Anderson, D., Burlinson, B., Hartmann, A., Kobayashi, H., . . . Sasaki, Y. F. (2000). Single cell gel/comet assay: guidelines for in vitro and in vivo genetic toxicology testing. Environ Mol Mutagen, 35(3), 206-221.

Tomlinson, I. P., Houlston, R. S., Montgomery, G. W., Sieber, O. M., & Dunlop, M. G. (2012). Investigation of the effects of DNA repair gene polymorphisms on the risk of colorectal cancer. Mutagenesis, 27(2), 219-223. doi:10.1093/mutage/ger070

103

Torok, M. S., & Grant, P. A. (2004). Histone acetyltransferase proteins contribute to transcriptional processes at multiple levels. Adv Protein Chem, 67, 181-199. doi:10.1016/S0065-3233(04)67007-0

Towler, W. I., Zhang, J., Ransburgh, D. J., Toland, A. E., Ishioka, C., Chiba, N., & Parvin, J. D. (2012). Analysis of BRCA1 Variants in Double-Strand Break Repair by Homologous Recombination and Single-Strand Annealing. Hum Mutat. doi:10.1002/humu.22251

Travers, K. J., Chin, C. S., Rank, D. R., Eid, J. S., & Turner, S. W. (2010). A flexible and efficient template format for circular consensus sequencing and SNP detection. Nucleic Acids Res, 38(15), e159. doi:10.1093/nar/gkq543 van Gent, D. C., Hoeijmakers, J. H., & Kanaar, R. (2001). Chromosomal stability and the DNA double-stranded break connection. Nat Rev Genet, 2(3), 196-206. doi:10.1038/35056049

Wall, M. E., Wani, M., Cook, C., Palmer, K. H., McPhail, A. a., & Sim, G. (1966). Plant antitumor agents. I. The isolation and structure of camptothecin, a novel alkaloidal leukemia and tumor inhibitor from camptotheca acuminata1, 2. Journal of the American Chemical Society, 88(16), 3888-3890.

Wang, J. Y., Ho, T., Trojanek, J., Chintapalli, J., Grabacka, M., Stoklosa, T., . . . Reiss, K. (2005). Impaired homologous recombination DNA repair and enhanced sensitivity to DNA damage in prostate cancer cells exposed to anchorage-independence. Oncogene, 24(23), 3748-3758. doi:10.1038/sj.onc.1208537

Watson, J. D., & Crick, F. H. (1953). Molecular structure of nucleic acids. Nature, 171(4356), 737- 738.

West, A. C., & Johnstone, R. W. (2014). New and emerging HDAC inhibitors for cancer treatment. J Clin Invest, 124(1), 30-39. doi:10.1172/JCI69738

West, S. C. (2003). Molecular views of recombination proteins and their control. Nat Rev Mol Cell Biol, 4(6), 435-445. doi:10.1038/nrm1127

Wightman, B., Ha, I., & Ruvkun, G. (1993). Posttranscriptional regulation of the heterochronic gene lin-14 by lin-4 mediates temporal pattern formation in C. elegans. Cell, 75(5), 855- 862.

Williams, R. S., Williams, J. S., & Tainer, J. A. (2007). Mre11-Rad50-Nbs1 is a keystone complex connecting DNA repair machinery, double-strand break signaling, and the chromatin template. Biochem Cell Biol, 85(4), 509-520. doi:10.1139/O07-069

Wold, M. S. (1997). Replication protein A: a heterotrimeric, single-stranded DNA-binding protein required for eukaryotic DNA metabolism. Annu Rev Biochem, 66, 61-92. doi:10.1146/annurev.biochem.66.1.61

Yu, X., Chini, C. C., He, M., Mer, G., & Chen, J. (2003). The BRCT domain is a phospho-protein binding domain. Science, 302(5645), 639-642. doi:10.1126/science.1088753

Zhong, Q., Chen, C. F., Li, S., Chen, Y., Wang, C. C., Xiao, J., . . . Lee, W. H. (1999). Association of BRCA1 with the hRad50-hMre11-p95 complex and the DNA damage response. Science, 285(5428), 747-750.

104

Zhuang, J., Jiang, G., Willers, H., & Xia, F. (2009). Exonuclease function of human Mre11 promotes deletional nonhomologous end joining. J Biol Chem, 284(44), 30565-30573. doi:10.1074/jbc.M109.059444

105

Appendix A: HeLa DR-[Library] Generation Protocol

Purpose: To generate a stable integration of a library of variants into the HeLa DR-FRT cell line.

Materials: pcDNA 5/FRT/TO/[library] plasmid cloned with variant library and 5’ intron, AmpR, HygroR pOG44 vector with AmpR

HeLa DR-FRT cells with broken GFP with I-SceI cut site, FRT sequence, ZeoR, PuroR

HeLa cells

Media for HeLa DR-FRT

DMEM (high glucose)

1% Pen/Strep

1% Glutamax

1% Sodium Pyruvate

10% Bovine Serum

Puromycin (1.5 µg/ml)

Zeocin (150 µg/ml)

Media for HeLa (unless otherwise specificied)

DMEM (high glucose)

1% Pen/Strep

1% Glutamax

10% Bovine Serum

Opti-MEM

106

Lipofectamine 2000

TrypLE Express

DMSO

Hygromycin B (50 mg/ml)

Integrating a library into HeLa DR-FRT cells is a reagent intensive protocol. You will need to have at least ten 15 cm plates of HeLa cells growing in log phase without selection. This is in parallel to the rest of experiment. Additionally, you can go through a Maxi-Prep of pOG44 plasmid per experiment, depending on yield. Make sure you have enough before starting the experiment.

Day 1 Seeding (Friday): Take a 100% confluent 10 cm dish of HeLa DR-FRT cells. Wash with 5 mL PBS

twice. Using 1 mL TrypLE Express, dissociate cells and dilute into 9 mL media for a total of 10 mL

suspension. Split 1 mL of suspension into ten 10 cm dishes that each contain 9 mL media. Gently

shake each plate form side to side and incubate in a humidified incubator at 37 °C for 48 hours.

Day 3 Transfection (Sunday): Cells should be about 70-80% confluent. If they are not, wait an extra day.

Draw off media and wash with 5 mL sterile PBS per plate. Replace with 10 mL transfection media

per plate.

Transfection:

1. Dilute 210 µg pcDNA5/FRT/TO/[library] plasmid and 105 µg pOG44 plasmid in 10.5 mL

Opti-MEM and incubate for 5 minutes

2. Dilute 315 µl Lipofectamine 2000 in 10.5 mL Opti-MEM and incubate for 5 minutes.

3. Combine both dilutions and gently tap the tube to mix. Incubate 15-20 minutes.

4. For each plate, add a volume of the mixture that equals to 2 µg pcDNA5/FRT, 1 µg

pOG44, 30 µl Lipofectamine 2000 and 20 mL Opti-MEM

5. Incubate overnight at 37 °C

107

Day 4 Express flipase (Monday): Replace the media with fresh transfection media and incubate overnight

at 37 °C. Set an empty humidified incubator to run at 30 °C and let temperature stabilize

overnight.

Day 5 Integration (Tuesday): In the morning, wash each plate with 5 mL sterile PBS, twice. With 1 mL

TrypLE Express each plate, dissociate the cells and transfer the cells into ten 15cm plates

containing 19 mL HeLa media (no selection factors, recommend Pen/Strep). Gently shake each

plate from side to side and incubate for 6-8 hours. After 6-8 hours, place the plates in the 30 °C

incubator.

Day 6 Express hygromycin resistance (Wednesday): Move the plates back into the 37 °C incubator and

incubate overnight. The 30 °C incubator can now be changed back to 37 °C if needed.

Begin preparation to make selection media for the next day

1. Wash ten 15 cm plates of log-phase HeLa cells with 10 mL sterile PBS.

Note: HeLa cells should be at least 30% confluent. Ideally, they are also not more than 75%

confluent

2. Add 25 mL HeLa media that contains Pen/Strep but no selection factors and is supplemented

with 10% FBS (rather than 10% BS).

Day 7 Selection (Thursday): Prepare 440 mL selection media and dissociate transfected HeLa cells to begin selection.

Selection media:

1. Draw 22 mL of media from each 15 cm plate of log-phase HeLa cells that was prepared the

day before. Replace with fresh media.

2. Add 220 mL of fresh HeLa media that contains Pen/Strep and no selection factors, and is

supplemented with 10% FBS

3. Add 4.4 mL of 50 ug/ml Hygromycin B stock solution.

4. Mix well

108

5. Filter sterilize with a 22 µm vacuum filter

6. Add 10 mL of selection media to twenty 15 cm plates

Dissociate and split transfected HeLa DR-FRT cells

7. Wash each 15 cm plate with 10 ml sterile PBS, twice

8. Dissociate transfected HeLa DR-FRT cells with 2.5 mL of TrypLE Express per plate by

incubating for 2 minutes at 37 °C.

9. Within 5 minutes of adding TrypLE Express, add 17.5 mL of selection media to each plate of

dissociated cells to create 20 mL of suspension. Mix thoroughly by pipetting up and down

until suspension appears fairly homogenous.

10. Transfer 10 mL of the suspension into one of the prepared 15 cm plates. Repeat with the

remaining 10 mL suspension with another prepared 15 cm plate. Each plate of transfected

cells will be transferred into two 15 cm plates for a total of twenty 15 cm plates with HeLa

DR-FRT cells in selection.

11. Incubate overnight at 37 °C

Day 8 Continued Selection (Friday): Prepare another 440 mL of selection media, wash and replace media for cells in selection.

Selection media:

1. Draw 22 mL of media from each 15 cm plate of log-phase HeLa cells that was prepared the

day before. Replace with fresh media.

2. Add 220 mL of fresh HeLa media that contains Pen/Strep and no selection factors, and is

supplemented with 10% FBS

3. Add 4.4 mL of 50 ug/ml Hygromycin B stock solution.

4. Mix well

5. Filter sterilize with a 22 µm vacuum filter

Wash and replace media for selecting cells

109

6. Gently wash each 15 cm plate with 10 ml sterile PBS, twice

Note: You may begin to see dead cells dissociate from the plate substrate, this is normal. Do

not try to shear the cells off the substrate with PBS. Gently add 10 mL sterile PBS to a corner

of the plate and rock back and forth, and gently aspirate the PBS. Repeat.

7. Gently add 20 mL of selection media to each 15 cm plate

8. Incubate overnight at 37 °C

Day 10+ Continued selection (Sunday and after): Observe cells every 48 hours after day 8. Dead cells should be dissociating off of the substrate for a few days. If there are visible dead cells floating, repeat

Day 8 steps. After 4-8 days, colonies should be visible under a microscope. Continue observation. Once the colonies are just visible without a microscope and no dead cells are present, replace the media with

HeLa media supplemented with 10% FBS, Pen/Strep, and 550 µg/mL Hygromycin B. Change media every

3-4 days until colonies are readily visible without a microscope. Using a microscope, if the center of most colonies is tightly packed with cells compared to the colony periphery, count the colonies and dissociate the cells. Plate about 50% of the colonies and freeze the remaining colonies.

Count, dissociation, freeze:

1. Estimate the number of colonies by counting a selected 25 cm2 area of 3 different plates.

Extrapolate to calculate total number colonies.

Note: The colony distribution is not always even. Try to select plates that have

representative distributions of colonies, and areas on those plates that are similarly

representative. I see about 50,000 colonies total typically, but the range has been 15,000-

150,000.

2. Thoroughly wash each plate with 10 mL sterile PBS buffer, twice.

3. Dissociate the cells from each plate with 3 mL TrypLE Express and incubate at 37 °C for 2-5

minutes.

110

Note: Colonies may be very difficult to dissociate. A 1 mL pipetman may be used to

dissociate colonies from substrate and break up clumps from colonies. If the colonies do not

dissociate from the substrate after 5-10 minutes even with a 1 mL pipetman, use a cell

scraper to gently dissociate the colonies from the substrate.

4. Collect all dissociated cells into 75 mL of HeLa media, for a total of 100 mL cell suspension

(for all plates), split between two 50 mL Falcon tubes.

5. Centrifuge the cells for 5 minutes at 1000 RPM

6. Decant supernatant. Be careful not to dislodge the pellet.

7. Combine both pellets with 10 mL of HeLa media that contains 10% FBS, 550 µg/mL

Hygromycin B, and Pen/Strep. Pipet up and down multiple times to mix thoroughly.

8. Prepare three 10 cm plates with the following volumes of HeLa media that contains 10% FBS,

550 µg/mL Hygromycin B, and Pen/Strep: 9.5 mL, 8.5 mL, 7 mL.

9. Add 0.5 mL suspension to the plate with 9.5 mL media, 1.5 mL suspension to the plate with

8.5 mL media, and 3 mL suspension to the plate with 7 mL suspension. Incubate at 37 °C

until a plate is confluent, then passage normally.

10. Centrifuge the remaining 5 mL of suspension for 5 minutes at 1000 RPM

11. Suspend the pellet in 5 mL HeLa freezing media (2.5 mL HeLa media that contains 10% FBS,

550 µg/mL Hygromycin B, and Pen/Strep, 2 mL FBS, 0.5 mL DMSO).

12. Aliquot 1 mL of suspension into each of 5 cryogenic tubes.

13. Slowly freeze aliquots insulated with Styrofoam in -80 °C freezer

14. Store frozen cells long term in vapor phase liquid nitrogen.

111

Appendix B: Highly Parallel Homology Directed Repair Assay (HP-HDR) Protocol

The purpose of this assay is to sort a library of variants into a functional HDR population and unrepaired population (containing both functional and non-functional variants). The HP-HDR is an extension of HDR, utilizing stably integrated variant libraries instead of expression vectors to assay variants.

Materials: pCBASceI vector with I-SceI, AmpR

HeLa DR-[Library] cells with broken GFP with I-SceI cut site, PuroR, HygroR

Media for HeLa DR-[Library]:

DMEM (high glucose)

1% Pen/Strep

1% Glutamax

1% Sodium Pyruvate

10% Bovine Serum

Puromycin (1.5 µg/ml)

Hygromycin B (550 µg/ml)

Opti-MEM

Lipofectamine 2000

Opti-MEM

Sorting buffer

1x Phosphate Buffered Saline (PBS) (Ca2+/Mg2+ free)

5 mM EDTA 112

25mM HEPES pH 7.0

1% Fetal Bovine Serum (Heat-Inactivated), dialyzed against Ca2+/Mg2+ free PBS

Filter sterilized with a 0.2 µm filter

Store at 4 °C

All siRNA at 20 uM (20 pmol/µl)

For transfections, follow Invitrogen/Thermo Fischer Scientific Oligofectamine and Lipofectamine 2000 protocol for timing, but use the volumes/amount in this protocol.

Use an siRNA that targets the 3’ UTR of the endogenous gene of interest. Use the luciferase targeting siRNA GL2 and a coding sequence targeting siRNA as controls

Typically, one 24-well plate is used per HeLa DR-[Library]

Day 1 Seeding (Saturday):

1. Begin with a 10 cm dish at least 95% confluent

2. Trypsinize in 1 mL for 2 minutes and stop by adding 9 ml DMEM/BS

3. Add 65 µl of suspension to each well of a 24-well plate

4. Incubate at 37 °C overnight

Day 2 Oligofectamine depletion (Sunday):

1. Cells should be 70-80% confluent

2. Wash cells once with 500 µl sterile PBS per well, replace with transfection media (HeLa media

with no selection factors or antibiotics)

3. Take 30 pmol siRNA (1.5 µl) and dilute in 25 µl Opti-MEM (26.5 µl), per well

a. Incubate for 5 minutes

4. Take 1.5 µl Oligofectamine and dilute in 6 µl Opti-MEM (7.5 µ), per well

113

a. Incubate for 5 minutes

5. Mix siRNA/Opti with Oligofectamine/Opti (34 µl per well)

a. Incubate for 30 minutes

6. Recommended well distribution: 1 coding sequence siRNA, 8 GL2 siRNA, 15 3’ UTR siRNA

7. Add mixture to each well with transfection media

8. OPTIONAL: Change media 4-6 hours later

9. Incubate overnight at 37 °C

Day 3 Transfer (Monday):

1. Wash cells with 500 µl sterile PBS per well, twice

2. Dissociate cells from the well with 300 µl TrypLE Express incubated at 37 °C for 2-5 minutes

3. Using a 1 mL pipetman, pipet the cells up and down and transfer each well to a well in a 6-well

dish containing 2 mL HeLa media

4. Incubate overnight at 37 °C

Day 4 Lipofectamine depletion and I-SceI expression (Tuesday):

1. Cells should be 60-80% confluent

2. Wash cells with 1 mL sterile PBS, once

3. Replace media with HeLa transfection media (no antibiotics or selection factors)

4. Take 50 pmol siRNA (2.5 µl) and dilute in 125 µl OptiMEM with 3 µg of pCBASceI per well

a. Incubate for 5 minutes

5. Take 3 µl Lipofectamine and dilute in 125 µl OptiMEM (130 µl) per well

a. Incubate for 5 minutes

6. Mix pCBASceI/siRNA/Opti with Lipo/Opti

a. Incubate for 30 minutes

7. Add mixture to each well with transfection media

8. Be sure to use same siRNA for each well

114

a. Recommended well distribution: 1 coding sequence siRNA, 8 GL2 siRNA, 15 3’ UTR

siRNA

9. Change media 4-6 hours later

10. Make sure to schedule flow cytometry if using the core facility by today. If doing the

recommended protocol, schedule 30 minutes of analysis as early as possible for Day 6, and 4

hours of sorting for Day 7 at least 24 hours after the end of the Day 6 flow (this gives you a

chance to cancel the appointment if the Day 6 flow results are poor)

Day 5 Transfer (Wednesday)

1. Wash cells with 1 mL sterile PBS, twice

2. Dissociate cells from substrate with TrypLE Express and combine each treatment and transfer

into 10 or 15 cm plates.

a. If using recommended well distribution:

i. Dissociate cells from 7 wells of GL2 siRNA transfected cells and from 14 wells of

3’ UTR siRNA transfected cells with 0.5 mL TrypLE Express incubated at 37 °C

for 2-5 minutes

ii. Transfer 7 wells of GL2 siRNA transfected cells into a 10 cm dish with 10 mL of

HeLa media and 14 3’ UTR siRNA transfected wells into a 15 cm dish with 20 mL

of HeLa media

iii. Replace media in the 3 remaining wells in the 6 well plates with 2 mL of HeLa-

[Library] media for each well

b. NOTE: It is a good idea to leave 1 well from each treatment to check GFP expression the

day before sorting.

3. Incubate overnight at 37 °C

Day 6 Flow cytometry check (Thursday)

1. Wash the cells in the 6 well plates (if reserved from Day 5) with 1 mL sterile PBS, twice

115

2. Dissociate cells from substrate with 0.5 mL TrypLE Express incubate at 37 °C for 2-5 minutes

3. Transfer cells into to a flow cytometry tube containing 1 mL HeLa media for each well

4. Perform a flow cytometry analysis count analyzing GFP expression and gating for live cells. Count

to 10,000 cells

a. Note: Try to do this early in the morning. If there are any expression problems, then you

may be able to cancel the next day’s flow cytometry sorting appointment. You need at

least 24 hours to cancel flow cytometry sorting at the flow cytometry core facility to

avoid paying for the time reserved.

b. Hela-BRCA1 libraries typically have the following GFP positive populations:

i. GL2 control: 7-9% GFP positive

ii. 3’ UTR (5787 siRNA): 4-7% GFP positive

iii. Coding sequence (2616 siRNA): 1-2% GFP positive

Day 7 Flow Cytometry Sorting (FACS) (Friday)

1. NOTE: Prepare one treatment at a time, the flow takes more than 1 hour to per treatment to

complete. For example, prepare the 3’ UTR plate first and give the cells to the flow cytometry

core. Do not prepare the GL2 plate until the 3’ UTR cells have almost finished sorting. This is

because the cells will aggregate over time and preparing the cells too early before sorting will

decrease the FACS efficiency.

2. Wash cells with sterile PBS, twice

Recommended distribution:

a. Wash 10 cm plate of GL2 siRNA transfected cells with 5 mL sterile PBS, twice

b. Wash 15 cm plate of 3’ UTR siRNA transfected cells with 5 mL sterile PBS, twice

3. Dissociate cells from plate substrate with TrypLE Express and suspend in sorting buffer

a. For 10 cm plate:

i. Wash plate with 5 mL sterile Ca2+/Mg2+ free PBS, twice

116

ii. Use 1 mL TrypLE Express and incubate at 37 °C for 2 minutes

iii. Thoroughly break up all the cells using a 1 mL pipetman. Try to get the cell

suspension as homogenous as possible

iv. Transfer cells into at LEAST 4 mL of sorting buffer. At this point I would

recommend 7-9 mL, sorting efficiencies have been much better when well

diluted, but further optimization may be needed. Transfer the cells into a 15

mL Falcon tube containing sort buffer.

v. Save 10% of the sample as input and store on ice

b. For 15 cm plate:

i. Wash 15 cm plate with 10 mL sterile Ca2+/Mg2+ free PBS, twice

ii. Use 2.5 mL TrypLE Express and incubate at 37 °C for 2 minutes

iii. Thoroughly break up all the cells using a 1 mL pipetman. Try to get the cell

suspension as homogenous as possible

iv. Transfer cells into at LEAST 7 mL of sorting buffer. At this point I would

recommend 10-14 mL, sorting efficiencies have been much better when well

diluted, but further optimization may be needed. Transfer the cells into a 15

mL Falcon tube containing sort buffer.

v. Save 10% of the sample as input and store on ice

4. Take cells to flow cytometry core on the 10th floor BRT

5. Tell the flow technician who is assisting you that you need to sort the cells based on GFP

expression. As many GFP positive cells should be collected as possible, and around 2 million GFP

negative cells should be collected. Remember to provide the flow core with HeLa-[Library] media

(~9 mL per treatment is sufficient)

117

6. IMPORTANT: Have the flow technician move the CIP up approximately 1 cm from the bottom of

the tube and set the agitation to the highest setting in the control software. This will prevent a

lot of clogging that would otherwise occur.

7. Check on the status of the flow cytometry every 30 to 60 minutes to see when the next

treatment needs to be prepared

8. Pick up collected cells and store on ice until genomic DNA extraction

9. Genomic DNA extraction

a. Follow the QIAGEN instructions provided with the DNEasy Blood & Tissue Kit until the

final 2 steps

b. Follow the Cultured cells protocol, found within the Animal Blood (Spin Column)

Protocol

c. The first step of the protocol is to take input, GFP positive, and GFP negative samples

and centrifuge them for 5 min at 300g. Balance with PBS if needed.

d. Do the optional RNAse A treatment as described in the protocol. RNAse A in the correct

stock concentration should be available in Maxi Prep kits (they are not included in the

DNEasy Blood & Tissue Kit)

e. The last 2 steps use buffer AE. Instead use buffer EB, same volumes.

i. Use 200 µL Buffer EB to elute, and for the 2nd elution, use a fresh tube to not

dilute the first.

ii. If sending gDNA to be sequenced, send the 1st elution.

f. Store gDNA at -20 °C

118