ADULT STEM CELL POPULATIONS AMAR BHARAT DESAI. Submitted in partial fulfillment of the requirements. For the degree of Doctor of Philosophy

Transcription

1 THE MULTIFACETED ROLE OF EXONUCLEASE 1 IN DNA REPAIR AND ADULT STEM CELL POPULATIONS By AMAR BHARAT DESAI Submitted in partial fulfillment of the requirements For the degree of Doctor of Philosophy Dissertation Advisor: Dr. Stanton L. Gerson, M.D. Department of Pharmacology Case Western Reserve University May, 2014

2 CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES We hereby approve the thesis/dissertation of Amar Bharat Desai candidate for the Doctor of Philosophy degree *. (signed) Derek Taylor (chair of the committee) Stanton L. Gerson Anthony Berdis Guangbin Luo Marvin Nieman Scott Welford (date)december 5 th, 2013 *We also certify that written approval has been obtained for any proprietary material contained therein.

3 Dedication To my parents, Bharat and Madhu Desai, who emigrated from India in 1979 with $42 in their pockets and the dream of a better life for their children. To this day they live with a simple mantra that constantly inspires me: work hard, love your friends and family, and treat people kindly.

4 Table of Contents Table of Contents List of Figures Acknowledgments i iv vii Abstract 1 Chapter 1: Introduction and Background The DNA Damage Response Cell Cycle Checkpoints DNA Repair Pathways Hematopoietic Stem Cells Quiescence and HSC Cell Cycling The HSC Niche DNA Repair in Hematopoietic Stem Cells Cancer Stem Cells CD133 as a Marker for CSCs DNA Repair in CD133 Cells Exonuclease Exo1 in DNA Mismatch Repair Exo1 in Homologous Recombination Statement of Purpose 27 i

5 Chapter 2: Exonuclease 1 is a critical mediator of survival during DNA double strand break repair in non-quiescent hematopoietic stem and progenitor cells Abstract Introduction Materials and Methods Results Discussion Acknowledgements 51 Chapter 3: CD133+ cells contribute to radioresistance via altered regulation of DNA repair genes in human lung cancer cells Abstract Introduction Materials and Methods Results Discussion Acknowledgements 76 Chapter 4: Exo1 Independent DNA Mismatch Repair Involves Multiple Compensatory Nucleases Abstract Introduction Materials and Methods 91 ii

6 4.4 Results Discussion Acknowledgments 104 Chapter 5: Discussion and Future Directions The Relationship between DNA Repair and Cell Cycle in HSCs Proposed Role of HR and HSC Cell Cycle Status in Fanconi Anemia DNA Repair and Cancer Stem Cell Specific Therapies DNA Repair Protein Targets in Clinical Trials Maintenance of Genomic Stability via Compensatory DNA Repair Pathways 130 Appendix 136 References 147 iii

7 List of Figures Figure 1.1: Overview of DNA damaging lesions and repair pathways 4 Figure 1.2: Overview of Hematopoiesis and Lineage Cell Commitment 12 Figure 1.3: Schematic Rationale for Cancer Stem Cell Specific Therapy 18 Figure 1.4: Proposed model for Exonuclease 1 catalytic vs. structural role in multiple cellular processes 25 Figure 2.1: Exo1 mut fibroblasts and hematopoietic progenitors display DSB repair defects 52 Figure 2.2: Exo1 mut mice exhibit no defects in competitive repopulation, serial transplant, or niche occupancy 54 Figure 2.3: Exo1 mut HSCs at steady state are not more IR sensitive than WT HSCs 56 Figure 2.4: Exo1 mut mice become IR sensitive after 5-FU induced HSC cell cycle entry 58 Figure 2.5: Poly IC mediated cell cycle entry also results in HSC IR hypersensitivity in Exo1 mut mice only 60 Figure 3.1: The CD133 marker promotes IR resistance in A549 cells but not H1229 cells 77 Figure 3.2: CD133+ cells display basally upregulated DNA repair genes in A549 cells only 79 iv

8 Figure 3.3: CD133 contributes to radioresistance in cells surviving multiple IR doses in both A549 and H1229 cells via upregulation of DNA repair genes 81 Figure 3.4: CD133+ cells display upregulated DNA repair genes in both A549 and H1299 following IR exposure 83 Figure 3.5: Rad51 and Exo1 silencing in A549 cells abrogates the IR expansion phenotype and induces CD133 IR sensitivity 85 Figure 4.1: Exo1 mut MEFs demonstrate temozolomide sensitivity and repair G/T mismatches in vitro 105 Figure 4.2: Exo1 mut mice do not gain a hematopoietic competitive advantage after temozolomide treatment in vivo 107 Figure 4.3: Exo1 mut MEFs demonstrate upregulated gene expression of multiple nucleases after temozolomide treatment 109 Figure 4.4: shrna mediated silencing of Artemis, Fan1, and Mre11 in Exo1 mut MEFs results in mild temozolomide resistance and decreased MMR capacity 111 Figure 4.5: Combination silencing of Artemis/Fan1, Artemis/Mre11 and Fan1/Mre11 in Exo1 mut MEFs demonstrate increased temozolomide resistance and decreased MMR capacity 113 Figure 4.6: Triple silencing of Artemis/Fan1/Mre11 demonstrates significant MMR loss in both WT and Exo1 MEFs 115 Figure 5.1: Proposed model for Exo1mut mouse HSC Response to 5-FU + IR 119 Figure 5.2: Proposed model for studying CSC cell cycle activation and combination v

9 treatment 128 Figure 5.3: Proposed studies to identify additional components involved in Exo1 independent MMR 134 Appendix 1: Exo1 mut mice demonstrate normal bone marrow characteristics at steady state 137 Appendix 2: Measurement of 5-FU Induced SKL Cell Cycle Entry and SKL Measurement after 5-FU and IR in WT and Exo1 mut Mice 138 Appendix 3: CD133+ cells expand following ionizing radiation in human lung cancer cell lines 139 Appendix 4: Purity of magnetically sorted CD133 cells 140 Appendix 5: Confirmation of gene expression data using western blotting 141 Appendix 6: Confirmation of lentiviral knockdown of Exo1 and Rad51 in A549 cells 142 Appendix 7: Confirmation of real-time PCR data using western blotting 143 Appendix 8: WT and Exo1 mut MEF temozolomide sensitivity after transduction with scrambled shrna vector 144 Appendix 9: Verification of MMR phenotype using additional shrna targeting vector 145 vi

10 Acknowledgements It was during an 8 week summer internship at the University of Alabama at Birmingham that I first established my love for biomedical science. I had just completed my sophomore year at Kenyon College in Gambier Ohio, and I spent the summer with Dr. Debasish Chattopadhyay using x-ray crystallography to study the DHFR-TS protein complex of the parasite T. cruzi. It was there that I witnessed firsthand the creativity, ingenuity, and camaraderie that good science brings out, and I knew that I had found my career path. I have several people to thank for guiding me professionally along this path. I was introduced to the CWRU Pharmacology department after a really insightful summer internship in the lab of Dr. Amy Wilson-Delfosse. From there my thesis advisor Dr. Stanton Gerson taught me the importance of telling a good story to go along with good science, and he trusted me enough to give me freedom to pursue a wide range of scientific interests. My thesis committee of Dr. Tony Berdis, Dr. Guangbin Luo, Dr. Steven Sanders, Dr. Derek Taylor, and Dr. Scott Welford were instrumental in providing new perspectives to each of my projects, and my fellow members of the Gerson lab created a strong working environment. The collaborative nature of the CWRU Pharmacology department and Cancer Center resulted in building relationships with many faculty members that I hope to continue even after I leave Case. Finally there really isn t enough room to describe the unconditional love and support I have received from my family through this journey. My parents Bharat and Madhu and my brothers Anand and Ashish have been my cheerleaders since day one vii

11 and I will forever be grateful for having such a loving foundation of people by my side. Most importantly however has been my fiancé Deval, who is the single most supportive, caring and kind person I have ever met, and it was her positivity and sense of humor that helped me get through the rollercoaster that is graduate school. Thank you all for everything, this was by no means a singular journey, and I will forever be grateful for this support network! viii

12 The Multifaceted Role of Exonuclease 1 in DNA Repair and Adult Stem Cell Populations Abstract By AMAR BHARAT DESAI The DNA damage response is composed of multiple signaling and repair pathways which together constitute an important tool cells utilize to preserve genomic stability. Repair pathways have shown to be critical in many cell types, including in human cancers where upregulation of DNA repair is believed to contribute to therapy resistance. In normal human development it is vital to stem cell populations, including in the hematopoietic system, where maintenance of genomic stability is necessary for development and normal immune function. The precise proteins and pathways responsible for maintaining cellular damage responses are often cell type specific, and identification of critical repair enzymes continues to yield promising therapeutic targets for a plethora of human conditions. The 5 3 nuclease Exonuclease 1 (Exo1) has been implicated in several cellular processes including DNA mismatch and double strand break repair. Its upregulation has been characterized in human cancers of the breast, lung, colon, bladder and others. Here we examine the role of Exo1 in multiple contexts, including in hematopoietic stem cells (HSC), where we describe Exo1 loss in the HSC damage response both in quiescent and active settings. We demonstrate that while Exo1 and homologous recombination (HR) are dispensable for HSCs at steady state, stress induced cell cycle entry results in an 1

13 HSC reliance on Exo1 mediated HR. We also explore the importance of DNA repair pathways and Exo1 in human lung cancer stem cells using the CD133 marker, and characterize the potential for Exo1 silencing as a cancer stem cell specific therapy. We find that upon prior exposure to double-strand break therapy, CD133+ cells are activated and rely on multiple DNA repair proteins including Exo1, and that this reliance contributes to radiation resistance. Finally we mechanistically describe the role of Exo1 in DNA Mismatch Repair (MMR) and identify a compensatory Exo1 independent pathway that cells adopt to minimize genomic instability involving the additional 5 3 nucleases Artemis, Fan1, and Mre11. Collectively our findings provide deeper insight into the DNA repair dependence of multiple stem cell populations and characterize the importance of Exonuclease 1 both in stem cell maintenance and as a potential therapeutic target. 2

14 Chapter 1- Introduction to DNA Repair and Stem Cell Populations 1.1 The DNA Damage Response DNA damage response pathways are essential for repairing genomic insults cells are exposed to daily. These insults can arise endogenously from sources such as metabolic byproducts, spontaneous hydrolysis, and reactive oxygen species, or exogenously from sources including ionizing or ultraviolet radiation, as well as various chemical agents [1] [2]. It is estimated that each cell in the human body undergoes tens of thousands of DNA-damaging events daily which compromise genetic integrity, and if left unrepaired can interfere with crucial cellular processes including DNA replication and RNA transcription, or introduce potentially oncogenic mutations. The nature of the lesions can vary from base modifications, to single and double strand breaks, to DNA crosslinks, but cells have developed distinct and conserved DNA response pathways to repair each form of damage (Figure 1.1, adapted from [3]). After induction of DNA damage cell cycle checkpoints are activated in order to halt cell cycle progression and allow the cell time to repair the damage. It also results in initiation of transcriptional profiles which enhance DNA repair capacity and in the event of severe damage, initiate apoptosis to prevent damaged cells from reproducing[4]. 3

15 Figure 1.1 Overview of DNA damaging lesions and repair pathways (A) Genomic DNA constantly encounters endogenous and exogenous DNA damage. The typical sources of damage and the consequences to DNA are depicted as well as common repair pathways responsible for resolving these lesions. (B) The consequences of DNA damage on cell cycle checkpoints and two major outcomes of damage induction: inhibition of critical cellular processes which result in cell death, and accumulation of mutations and translocations that can result in oncogenesis. Figure adapted from Hoeijmakers Nature 2001, 411:

16 Figure 1.1 Overview of DNA damaging lesions and repair pathways Hoeijmakers Nature 2001, 411:

17 1.1.1 Cell Cycle Checkpoints The cell cycle consist of four major stages: the G1 (Gap-1) phase necessary for cell growth and preparation for DNA synthesis, the S (synthesis) phase in which total genome replication occurs, G2 (Gap-2) phase in which additional growth occurs as well as preparation for cell division, and the M (mitosis) phase in which cells separate the duplicated chromosomes into two daughter cells. Cell cycle checkpoints are regulatory pathways to ensure completion of cell cycle events and scan DNA for damage before moving forward with DNA synthesis and cell replication. In the DNA damage response the key mediators of checkpoint induction are the serine-threonine kinases ataxia telangiectasia, mutated (ATM) and ATM and Rad3-related (ATR) [5] which are responsible for activation of all downstream events. While ATM and ATR have similar cellular substrates they are activated in response to different forms of DNA damage, with ATM directing the response to DNA double strand breaks and ATR directing the response to replication stalls and crosslink damage[6]. The G1 checkpoint is the best understood and occurs immediately prior to entry into S phase, thus preventing replication of damaged DNA. Critical to this is the phosphorylation of the tumor suppressor p53 by the kinases ATM and Chk2, which results in accumulation of p53. This activates the transcriptional activity of p53 and the accumulation of the cyclin-dependent kinase inhibitor p21, which in turn suppresses cyclin E/Cdk2 kinase activity and causes G1 arrest. The S-phase checkpoint functions to decrease the rate of DNA synthesis after damage induction. It is also initiated by ATM 6

18 and the kinase Chk2 which phosphorylates the phosphatase Cdc25A resulting in its ubiquitin-dependent degradation. This prevents activation of the Cdk2/Cyclin E and Cdk2/Cyclin A complexes and prevents DNA synthesis. Finally the G2 checkpoint allows for cell cycle halting prior to the chromosome segregation step so that the damaged DNA is not passed along to daughter cells. The G2 checkpoint is also modulated by the Chk proteins after ATM/ATR dependent activation which phosphorylate the phosphatase Cdc25c, resulting in its sequestration in the cytoplasm. This prevents the activation of the Cdc2/Cyclin B1 complex and blocks cell entry into mitosis. Combined these checkpoints are crucial for maintaining genomic stability by preventing division of damaged cells and allowing time for proper repair by DNA repair pathways[4, 5] DNA Repair Pathways There are multiple forms of DNA lesions that can be introduced to cells. For instance ionizing radiation (IR) introduces predominantly double strand break (DSB), ultraviolet radiation introduces DNA crosslinks, internal reactives oxygen species introduce single strand breaks (SSB) which can be converted to DSBs, and chemotherapeutic agents can introduce a variety of lesions including alkyl groups or base dimers. Each of these forms of damage is repaired by a specific and highly conserved repair pathway[7]. Base Excision Repair Base excision repair (BER) is responsible for correcting non-bulky damage to DNA bases caused by oxidation, deamination, or complete loss of the DNA base. Defects in BER are associated with genomic instability and oncogenesis, with mutations in Pol-B 7

19 observed in up to 30% of cancers. BER excises the damaged base through DNA glycosylase activity, producing an apyrimidic/apurinic (AP) site. This AP site is processed by the enzyme APE1 which leads to recruitment of DNA polymerase beta and DNA ligase to reseal the nick[8, 9]. Mismatch Repair DNA mismatch repair (MMR) is responsible for repairing base:base mismatches or insertion/deletion loops caused by faulty replication or recombination. Defects in MMR are also associated with genomic instability and oncogenesis, with mutations in MMR proteins Mlh1 and Msh2 strongly associated with the pathogenesis of hereditary nonpolyposis colorectal cancer (HNPCC) [10]. Introduction of a mispaired base is recognized by the heterodimer MutSα which results in the recruitment of higher order complexes including the heterodimer MutLα. This results in recruitment of the nuclease Exo1 to excise the mispaired base, with DNA polymerase delta and DNA ligase resealing the DNA[11, 12]. Nucleotide Excision Repair Nucleotide excision repair (NER) is responsible for resolving multiple forms of DNA damage including pyrimidine dimers, intrastrand crosslinks, and bulky helix-altering chemical adducts. Multiple human diseases are associated with loss of NER including xeroderma pigmentosum and Cockayne syndrome[13]. NER requires a huge array of proteins and two global NER pathways exist depending on the location of the damaged DNA. After recognition of an altered helix DNA structure by multiple sensing complexes, 8

20 XPA and replication protein A (RPA) bind the site of the injury recruiting the helicases XPB and XPD which unwind the DNA, the endonucleases XPG and XPF which excise the damage DNA, and DNA Pol delta epsilon which resynthesizes the intact double helix[14]. Double Strand Break Repair DNA double strand breaks are perhaps the most dangerous lesion in cells because they can severely impair DNA replication, transcription, cause chromosomal translocations, and a single DSB can result in cell death. DSBs are caused by a variety of factors including exogenous ionizing radiation or chemicals, or endogenous reactive oxygen species or metabolic byproducts. Defects in DSB repair are associated with a plethora of human diseases including the well-known association of BRCA mutations with breast cancer. DSB repair occurs through two highly conserved pathways: Nonhomologous end-joining (NHEJ), and homologous recombination (HR). While HR is the higher fidelity pathway that uses a homologous sister chromatid for repair it can only be performed during the S and G2 phases due to sister chromatid availability, and thus NHEJ is the predominant DSB pathway in cells. NHEJ is more error prone and functions to simply ligate the two ends together. It relies on recognition of the DSB by the heterodimer Ku70/Ku80, activation of the NHEJ machinery by the DNA-dependent protein kinase DNA-PK, and DNA Ligase to reseal the break. HR on the other hand undergoes an initial resection step after DSB recognition by the core complex of Mre11, Nbs1, and Rad50 (MRN) and the nucleases Exo1 and CTIP. This results in recruitment of Rad51 to single stranded DNA along with BRCA1, BRCA2, and other proteins which 9

21 search the undamaged DNA for the homologous repair template. DNA polymerases, resolvases and Ligase I subsequently reseal the break for this high fidelity repair[15, 16]. The importance of these pathways are made clear by the human diseases described above caused by defective DNA repair, the fact that these pathways are conserved and very tightly regulated, and the fact that DNA repair proteins are increasingly being exploited as therapeutic targets due to their upregulation in many human cancers[17-19]. 1.2 Hematopoietic Stem Cells The role of DNA repair and genomic stability in the hematopoietic stem cell (HSC) population has been well studied and appears critical for long term function. HSCs are rare primitive cells in the blood and bone marrow that are responsible for the continuous synthesis of at least eight separate lineages of differentiated blood cells. These cells include all subtypes of the myeloid (erythrocyte, megakaryocyte, basophil, eosinophil, neutrophil, macrophage) and lymphoid (B-cell, T-cell) subtypes (Figure 1.2, adapted from [20]) [21]. HSCs are responsible for the natural turnover of billions of cells daily and also must compensate for hematological stresses including blood loss and infection. The primitive hematopoietic population can be divided into long-term selfrenewing HSCs, short-term self-renewing HSCs, and non self-renewing multipotent progenitor cells, with long term HSCs localized to a specialized microenvironment known as the HSC niche and residing in a mainly quiescent (G0) state to minimize potential genomic instability [22] [23] [24]. Because HSCs are conserved throughout the 10

22 lifetime of an organism and responsible for the constant reconstitution of immune cells, minimizing HSC stress and DNA damage is critical to long term preservation, with HSC damage potentially resulting in bone marrow failure or leukemogenesis [25]. 11

23 Figure 1.2 Overview of Hematopoiesis and Lineage Cell Commitment HSCs are responsible for the development and continual replenishment of all cells in the immune system. This occurs in the HSC niche, a specialized microenvironment in the bone marrow. Primitive long-term and short-term HSCs have self-renewal properties and give rise to multipotent committed progenitor cells of the myeloid and lymphoid lineages. These in turn give rise to all subtypes of immune cells including lymphocytes, erythrocytes, monocytes, and neutrophils. Figure adapted from Passegue PNAS, :

24 Figure 1.2 Overview of Hematopoiesis and Lineage Cell Commitment Passegue PNAS, :

25 1.2.1 Quiescence and HSC Cell Cycling As mentioned above, long term HSCs reside in a specialized microenvironment in a mainly quiescent state with very little cell cycling. This quiescent state is achieved after periods of active cell cycle and proliferation during fetal development in order to generate the blood system. During fetal life the main function of HSCs is to generate suitable red blood cell numbers for oxygen distribution and to develop the immune system. Thus early in development approximately ~95% of HSCs are in active cycle. Between 3 and 4 weeks of age in mice the HSC population transitions from that of mixed proliferative status to a largely quiescent population which remains for all of adulthood. Kinetic tracking studies of adult mouse HSCs have demonstrated that adult HSCs with high multi-lineage differentiation capability and self-renewal capacity harbor much slower cell cycle kinetics than other hematopoietic progenitor cells and that they can be induced upon cell stress. It is estimated that upwards of 80% of HSCs are in a G0 quiescent state with an additional ~20% in G1, while only ~5% are in the S/G2/M phase. However it is estimated that ~99% of long term HSCs divide at least once every 57 days which suggests that quiescence is not a static state[22, 24, 26] The HSC Niche The HSC niche is believed to be anatomically located in the endosteum and contains a heterogeneous population of cells that interact with HSCs and play a role in the longevity and regulation of these cells[27]. Osteoblastic lineage cells have been found to be critical to HSC regulation, as depletion of osteoblasts results in 3-10 fold 14

26 reduction in HSC numbers with additional associated defects[28]. Mesenchymal stem cells (MSCs) are also implicated in niche regulation as cotransplantation of MSCs with HSCs has been shown to improve donor engraftment and HSC self renewal[29]. Osteoclasts, endosteal monocytes, macrophages as well as vascular and perivascular stromal[30-32] cells have also been identified in niche maintenance with additional cellular factors such as N-cadherin and B-integrin important in anchoring HSCs in the niche[33]. Several groups have been able to use in vivo imaging to characterize the spatial relationship between these cell types in the bone[34]. To further demonstrate the complexity of the niche, multiple signaling pathways including the Wnt/B catenin pathway, TGF-B, notch, VEGF, and cytokine/receptor signaling such as SDF1/CXCR4, TPO/Mpl, and SCF/c-Kit have been identified as playing some role in niche regulation[23, 27, 35]. While characterization of molecules involved in niche maintenance is ongoing, several identified components have already been discovered which may play therapeutic roles in bone marrow transplantation and hematopoietic diseases DNA Repair in Hematopoietic Stem Cells Several studies characterizing the effects of defective DNA damage in HSCs have demonstrated the importance of multiple repair pathways in HSC longevity. Reese et. al (2003) used an MMR deficienct Msh2-/- mouse to show that loss of MMR resulted in methylating agent tolerance as well as defective HSC function, demonstrated through competitive repopulation and serial transplantation, stating that loss of MMR function causes stem cell exhaustion due to accumulation of genomic instability [36]. Navarro et. 15

27 al (2006) studied the HSC effects in a Fanconi anemia mouse model Brca2 Δ27/Δ27 defective in crosslink repair and found high levels of chromosomal instability, hypersensitivity to ionizing radiation, defects in competitive repopulation, niche occupancy and a severe hematopoietic proliferation defect [37]. Mouse models of defective NHEJ have convincingly shown that loss of this DSB repair pathways has deleterious effects on HSCs as Nijnik et. al (2007) use a DNA Ligase IV mutant (Lig4 Y288C ) mouse to show impaired HSC function as well as loss of HSC quiescence [38]. Similarly Qing et. al (2012) have shown that mutant mice of the NHEJ protein DNA-PKcs (SCID mice) also demonstrate hematopoietic defects as measured through loss of competitive repopulation function and ability to retain the HSC niche [39]. Many others including Bender et. al (2002), Rossi et. al (2007), and Farres et. al (2013) have demonstrated the importance that intact DNA repair pathways have on HSCs during both homeostasis and periods of activation [40] [41] [42]. However the specific repair proteins and pathways critical for HSC maintenance during quiescence or active cell cycle have yet to be elucidated. 1.3 Cancer Stem Cells In addition to HSCs, a subset of stem cells initially discovered in hematopoietic malignancies has been shown to rely on DNA repair signaling. These cancer stem cells (CSCs) were initially identified in acute myelogenous leukemia but have now been identified in many cancer types including brain, breast, liver, pancrease, colon, lung, and others. The characterization of CSCs in recent years has shown that these cells are often 16

28 therapy resistant and can self-renew indefinitely to give rise to differentiated daughter cells. CSCs are believed to play a critical role in cancer therapy due to their implications in tumor relapse and treatment failure (Figure 1.3, adapted from [43]). Multiple cell surface and enzymatic markers have been established to identify these cells within a tumor, with the classic confirmation of stemness shown through the ability to reconstitute a tumor with low cell numbers[44-47]. As mentioned, CSCs were first identified in AML by Lapidot et. al (1994) when leukemia initiated cells transplanted into SCID mice were shown to home to the bone marrow and recreate the disease observed in the original patients. The frequency of these cells was found to be one engraftment unit in 250,000 cells and further fractionation of the cell populations identified the cell surface markers CD34+/CD38- as greatly enriching for these CSCs[48]. This was followed by CSC identification in a solid tumor by Al-Hajj et. al (2003) in breast (CD44+/CD24-) and by Singh et. al (2003) in brain (CD133) [49, 50]. 17

29 Figure 1.3 Schematic Rationale for Cancer Stem Cell Specific Therapy Cancer stem cells have been characterized as highly malignant cells within a tumor that gives rise to the bulk of differentiated tumor cells. While conventional chemotherapy traditionally shrinks tumors by killing the bulk tumor cells, evidence has shown that CSCs are able to survive therapy via several pro-survival mechanisms. This can result in eventual repopulation of the tumor and disease relapse. CSC specific therapies have been proposed to target factors contributing to cell survival and therapy resistance in order to kill CSCs in addition to the bulk tumor. Figure adapted from Pardal Nature Reviews Cancer, :

30 Figure 1.3 Schematic Rationale for Cancer Stem Cell Specific Therapy Pardal Nature Reviews Cancer, :

31 1.3.1 CD133 as a Marker for CSCs CD133, a transmembrane glycoprotein member of the prominin family has been found to be a CSC marker in many tumor types including those of the brain, colon, lung, and prostate. It was discovered initially by Yin et. al (1997) on hematopoietic stem cells and by others including Asahara et. al (1997) on circulating endothelial progenitors[51, 52]. Multiple groups have shown that CD133 can enrich CSCs approximately 200-fold from human tumor tissue and that in certain tumor types CD133+ cells demonstrate limitless self-renewal, therapy resistance, and can fully recapitulate the human disease when transplanted into immune deficient mice[53, 54]. CD133 has three reported isoforms (CD ) and loss of CD133-2 has been associated with a gain in terminal differentiation, suggesting its role in maintaining stemness[55]. Further evidence has shown that CD133+ colon cancer and melanoma cells remain largely in the G0/G1 stage of the cell cycle and while the exact function of CD133 is not well understood, many cell survival and oncogenic pathways have been implication along with CD133 expression including mtor, Wnt, PI3K-AKT, and CXCR4-SDF1[56-58]. In patients multiple groups have demonstrated that CD133 expression in tumor samples correlates with poor prognosis including in gliomas, colon cancer, and lung cancer[59-61] DNA Repair in CD133 Cells The role of DNA repair in CSCs and in particular CD133+ cells has also been shown to be important for survival and therapy resistance. Bao et. al (2006) used glioma models to demonstrate that CD133+ cells expand in response to ionizing radiation, are 20

32 significantly more radioresistant than CD133- cells, and that this is in part due to enhanced DNA repair capacity (measured by comet tail resolution and measurement of p-chk1 levels). A specific inhibitor of Chk1 combined with IR was found to induce CD133+ sensitivity, confirming the importance of DNA repair and cell cycle checkpoints in CD133+ resistance[62]. Others including Ropolo et. al (2009), Mihatsch et. al (2011), and Short et. al (2011) have demonstrated the importance of DNA repair proteins in CD133+ resistance and have identified potential DNA proteins which could serve as therapeutic targets to induce sensitivity[63-65]. However the precise pathways and repair proteins critical to CD133+ function have yet to be fully determined. 1.4 Exonuclease 1 The 5 3 nuclease Exonuclease 1 (Exo1) is a member of the Rad2 family of structure specific nucleases consisting of flap endonuclease 1 (Fen1) which participates in the processing of Okazaki fragments, gap endonuclease 1 (Gen1) which is involved in Holliday junction resolution, and xeroderma pigmentosum complementation group G (XPG) which processes DNA bubble structures[66]. Rad2 nuclease family members share an N-terminal catalytic core nuclease domain but differ in DNA substrate specificity.[67] Exo1, which also possesses 5 flap endonuclease activity, has been implicated in multiple DNA repair processes and is thought to play a critical role in both MMR and HR. It additionally has been implicated in somatic hypermutation (SHM), telomere maintenance, and male and female meiosis, although its precise role in each of these pathways has yet to be characterized[68-70]. In mice and humans Exo1 expression 21

33 varies in multiple tissues, with mouse expression highest in testis, spleen, lung and kidney, and human expression highest in testis, bone marrow, thymus, and pancreas [71] Exo1 in DNA Mismatch Repair Human Exo1 was first characterized for its 5 3 exonuclease function in MMR, requiring a nick 5 to the mismatch in order to perform hydrolysis on double stranded DNA[72, 73] Its interactions with multiple MMR proteins including the MutLα and MutSα complexes have been shown to be important for its exonuclease activity in MMR, with binding of Exo1 to MutSα in an ATP dependent manner necessary for its activity[11, 74-76]. Unlike mutations in other critical MMR proteins Exo1 mutations in humans have not conclusively been linked to hereditary nonpolyposis colorectal cancer (HNPCC), [77-79] although mouse models of mutant Exo1 have demonstrated increased incidences of microsatellite instability (a measure of MMR failure), increased mutation rates, and increased tumorigenesis[69, 80]. While Exo1 is currently the only nuclease that has been implicated in MMR, several studies have suggested that its exonuclease activity is not essential for MMR function. Kadyrov et. al (2009) studied mouse Exo1 mutant ES cells and demonstrated that they retained residual MMR activity as measured through heteroduplex plasmid containing a G-T mismatch. In purified systems they subsequently confirmed that a set containing 3 G-T heteroduplex DNA, Mutsα, MutLα, RFC, PCNA, RPA, DNA polymerase δ, and ATP was able to repair the mismatch with high efficiency, suggesting that Exo1 was not 22

34 critical for repair. They propose that this in vitro repair is able to proceed through endonuclease activity of MutLα to generate a DNA nick which recruits DNA polymerase δ to the mismatch to resynthesize the strand and in turn displace the mispaired DNA[81]. Izumchenko et. al (2012) further demonstrated that it is Exo1 catalytic activity that is dispensable for MMR as Exo1 depleted MEFs were shown to display alkylating agent resistance and impaired binding of Msh2 to DNA, while reintroduction of catalytically dead Exo1 restored alkylating agent sensitivity and Msh2/DNA binding[82]. This observation was confirmed in an in vivo setting recently by Schaetzlein et. al (2013) after generating an Exo1null mouse and an Exo1 mutant mouse (E109K) to compared functionality in various cellular processes. In terms of MMR, they found that compared to the Exo1 null mouse, Exo1 E109K retained mismatch repair activity and also displayed normal class switch recombination, another indication towards proficient MMR. Thus they hypothesize that Exo1 enzymatic activity is dispensable for MMR activity while its structural component is necessary for repair, perhaps through recruitment of additional nucleases[80] Exo1 in Homologous Recombination In HR the role of Exo1 has been largely elucidated in yeast and found to be critical to the early DNA end resection step. After introduction of the double strand break, a complex consisting of Mre11-CTIP initiates short tract resection which is followed by a more extensive resection involving either Exo1 or the Blm/Wrn-DNA2 complex [83] [84]. 23

35 It has recently been shown in humans that Exo1 is the predominant nuclease and can complete resection even in the absence of the other nucleases[68]. Additionally multiple groups have demonstrated that Exo1 is necessary for recruitment of Rad51 and RPA to sites of DSBs, and that its loss results in chromosomal instability and hypersensitivity to ionizing radiation [85, 86]. The above referenced Schaetzlein et. al (2013) study also demonstrated both the Exo1 null and Exo1 E109K mouse demonstrated significant hypersensitivity to DSB inducing agents and impaired RPA recruitment, suggesting that while Exo1 nuclease activity may be dispensable for its MMR function it is required for resection activity in HR (Figure 1.4, adapted from [66] and [80])[80]. The growing field of Exo1 studies have demonstrated that it is a critical protein for multiple cellular processes, but the precise role that Exo1 plays in the stem cell populations described above, and mechanistically in both MMR and HR require further study. 24

36 Figure 1.4 Proposed model for Exonuclease 1 catalytic vs. structural role in multiple cellular processes (A) Human Exo1 domains. Asterisks indicate identified mutations that reduce enzymatic activity and diamonds represent mutations that affect Exo1 binding to both Mlh1 and Msh2. Figure adapted from Tran DNA Repair, : (B) Exo1 is implicated in DNA mismatch and double strand break repair but as Schaetzlein et. al (2013) recently reported, the catalytic activity appears dispensable for MMR but critical to HR. Figure adapted from Schaetzlein PNAS, :

37 Figure 1.4 Proposed model for Exonuclease 1 catalytic vs. structural role in multiple cellular processes A Tran DNA Repair, : B Schaetzlein PNAS, :

38 1.5 Statement of Purpose The studies herein aim to address the role of DNA repair pathways, and in particular the enzyme Exonuclease 1, in addressing three major questions. (1) How is the DNA double strand break repair dependence of hematopoietic stem cells affected when HSCs transition from quiescence to active cell cycle, and what implications does this have on human disease? (2) To what extent is the radiation resistance of cancer stem cells dependent on DNA repair, and can inhibition of critical repair enzymes induce stem cell specific sensitivity? (3) What compensatory mechanisms do cells adopt in order to maintain genomic stability after loss of Exo1 enzymatic activity in DNA mismatch repair? Together an understanding of these questions will give insight not only into mechanisms of maintaining genomic stability, but also into the role of DNA repair in adult stem cell maintenance. First we used an Exo1 mutant mouse model (Exo1 mut ) to demonstrate that at steady state HSC function is not compromised by loss of Exo1 function nor are mice hypersensitive to ionizing radiation, confirming exclusive NHEJ reliance for quiescent HSCs. However after activating HSC cell cycle entry via 5-FU or poly-ic we observed that the damage response became reliant on Exo1 mediated HR as Exo1 mut mice displayed IR hypersensitivity and defective HSC function. Thus we propose that DNA damage response is cell cycle dependent in HSCs and that efforts to maintain HSC quiescence may be a promising therapeutic target for certain hematopoietic diseases. 27

39 We next used human lung cancer cell lines to assess the DNA repair reliance of cancer stem cells as identified through the CD133 marker. We demonstrate that CD133+ IR resistance is acquired after prior IR exposure and this resistance correlates with a strong induction of DNA repair genes and an improved DNA repair capacity as measured by ϒ-H2AX foci resolution. We silenced the repair enzymes Exo1 and Rad51 and observed that CD133+ cells demonstrated IR sensitivity after loss of these enzymes, suggesting that these and other DNA repair proteins may serve as CSC specific therapeutic targets. Finally we used Exo1 mut mice to study the effects of loss of Exo1 function on DNA mismatch repair. We demonstrate that Exo1 mut MEFs retain temozolomide sensitivity and are able to repair G-T mismatches. We used gene expression studies after temozolomide treatment to identify the nucleases Artemis, Fan1, and Mre11 as potentially being implicated in Exo1 independent MMR and subsequently silenced these genes using shrna to characterize their role in the compensatory pathway. We found that combined silencing of these genes combined with loss of Exo1 activity mimicked an MMR deficient phenotype, providing insight into the biology of this pathway and providing novel enzymes to further study MMR. In total we believe these studies provide a firm groundwork to continue studying the nuanced relationship between DNA repair, cell cycle, and stem cell populations. In particular we identify the Exo1 mut model as an important research tool because the basal subtlety of phenotype allows for the dissection of additional proteins and 28

40 pathways critical for stem cell function. This can be manipulated to study additional Exo1 mediated processes including antibody diversification and telomere maintenance, and thus may have even broader implications for human disease. 29

41 Chapter 2 1 Exonuclease 1 is a critical mediator of survival during DNA double strand break repair in non-quiescent hematopoietic stem and progenitor cells 2.1 Abstract Hematopoietic stem cell populations require DNA repair pathways to maintain their long term survival and reconstitution capabilities, but mediators of these processes are still being elucidated. Exonuclease1 (Exo1) participates in homologous recombination (HR) and Exo1 loss results in impaired 5 HR end resection. We use cultured Exo1 mut fibroblasts and bone marrow to demonstrate that loss of Exo1 function results in defective HR in cycling cells. Conversely in Exo1 mut mice HR is not required for maintenance of quiescent HSCs at steady state, confirming the steady state HSC reliance on non-homologous end joining (NHEJ). Exo1 mut mice sustained serial repopulation, displayed no defect in competitive repopulation or niche occupancy, and exhibited no increased sensitivity to whole body ionizing radiation. However when Exo1 mut HSCs were pushed into cell cycle in vivo with 5-Fluorouracil or poly IC, the hematopoietic population became hypersensitive to IR, resulting in HSC defects and animal death. We propose Exo1 mediated HR is dispensable for stem cell function in quiescent HSC, whereas it is essential to HSC response to DNA damage processing after cell cycle entry, and its loss is not compensated by intact NHEJ. In HSCs the maintenance of stem cell function after DNA damage is dependent on the DNA repair capacity, segregated by active vs. quiescent points in cell cycle. 1 A version of this chapter was accepted for publication in the journal Stem Cells on September 28 th,

42 2.2 Introduction: Hematopoietic stem cell (HSC) maintenance is essential for sustained longevity and tissue function. The HSC population has lifelong self-renewing capabilities and gives rise to subsets of multipotent progenitor cells, and in turn a progeny of terminally differentiated mature cells consisting of all subtypes of the myeloid and lymphoid lineages. Long term reconstituting HSCs are necessary to replace the differentiated cells after losses caused by normal cell turnover or environmental stress such as infection, radiation, or chemotherapy[87, 88]. Failure to replenish these stores has been linked to a variety of hematopoietic disorders in humans as well as aging phenotypes[89-91]. Under steady state conditions the majority of the HSC population (>95%) is believed to be in a quiescent state with 1-2% turnover a day[22, 92, 93]. HSCs remain in the G0 phase of the cell cycle in order to minimize the potential for genomic instability, and only enter the cell cycle when necessary to replenish the hematopoietic system [94]. Therapeutic intervention and environmental stresses can mobilize HSCs into cycle where they remain until homeostasis is restored in the circulating blood cell populations[22, 95]. Since HSCs are required throughout the lifetime of an organism for blood repopulation, mutation and damage avoidance is crucial, and thus the bridge between quiescence and cycling must be finely monitored [23]. The DNA Damage Response (DDR) system is a series of highly conserved pathways that function to repair damaged DNA in cells. The damage can range from single strand breaks (SSB) and double strand breaks (DSB), mispaired bases, or other 31

43 genomic lesions such as thymine dimers or depurinated bases[96, 97]. Perhaps the most dangerous lesion in cells is the DSB which can be caused by exogenous factors, internal metabolic reaction products, and other sources including mechanical stress on the cell[98]. Two distinct pathways exist to repair DSBs in cells, homologous recombination (HR) and non-homologous end-joining (NHEJ). HR occurs only in cycling cells in the S or G2 phases because it requires a sister chromatid to serve as a homologous template for the damaged strand, resulting in high fidelity repair. NHEJ is a more error prone method to repair DSBs whose basic function is to ligate the two broken ends of the DNA [19, 99]. The interplay between HR and NHEJ is quite complex, with examples of inter-pathways negative regulation to promote proper choice of repair mechanism[100]. In hematopoietic stem cells much remains to be understood about DDR processes after induction of damage. Loss of DSB repair function in DNA-PK (SCID), LigaseIV -/- mice, Ku80 -/- mice, all components of NHEJ, and others has been shown to result in no change in overall stem cell numbers but instead a drastic decrease in stem cell function- characterized by loss of self-renewal capability, competitive repopulation defects, and increases in apoptosis and stem cell exhaustion[38, 41, 101]. These data suggest a fundamental role for DDR in HSC maintenance. However, only recently are the DDR pathways used in HSCs being elucidated. Mohrin et al (2010) used immunofluorescence studies in cultured HSCs to demonstrate that quiescent HSCs use NHEJ for DSB response while proliferating HSCs transition to a dual reliance on NHEJ and HR for repair. Thus HSCs appear to rely on NHEJ as the default repair mechanism under quiescent steady state conditions. However, when and how HSC convert to utilization of 32

44 and perhaps reliance on HR either solely or complemented by NHEJ, has not been evaluated. Here we have investigated the multifaceted enzyme Exonuclease 1 (Exo1) in the context of hematopoietic function. Exo1 has been implicated in the early 5 end resection step of HR and loss of Exo1 has been shown to result in DNA DSB repair defects[83, 85, 86]. We used nuclease-dead Exo1 mut mice[69] and showed at steady state that these mice displayed no defects in HSC function, including IR sensitivity, indicating that intact NHEJ can maintain quiescent HSCs. However when we pushed the HSC into cycle with 5-FU or poly IC we observed that Exo1 mut mice demonstrated IR hypersensitivity which resulted in HSC dysfunction and animal death. Thus when HSCs transition from quiescence to active cell cycle the DNA repair of DSBs transitions from a dependence on NHEJ alone to a requirement for HR. This study suggests that in HSCs the point of greatest reliance on HR for DDR is the point at which HSC enter active cell cycle. 2.3 Materials and Methods: Animals Exo1 mut mice -C57BL/6J background- used in these studies were donated by Dr. Winfried Edelmann[69]. Animals were used along with WT littermates throughout. All mouse studies were approved by the institutional animal care and use committee at Case Western Reserve University. 33

45 Reporter Plasmid Assays DR-GFP: The DR-GFP reporter construct was used as previously described by Pierce et al[102]. The plasmid was stably incorporated into WT and Exo1 mut fibroblasts followed by transfection of the I-Sce1 endonuclease plasmid. Cytometric analysis was performed on these cells 48 hours after I-Sce1 to measure the levels of GFP fluorescence. Student s t-tests were used to determine statistical significance. Pem1-Ad2-GFP: The NHEJ reporter construct Pem1-Ad2-GFP [103] was cut with the I- Sce1 endonuclease and subsequently transfected into WT and Exo1 mut MEFs. 48 hours post transfection cytometric analysis was performed to determine levels of GFP fluorescence. Student s t-tests were used to determine statistical significance. Immunostaining Fibroblasts were grown on coverslips and treated with 4Gy of ionizing radiation. At the indicated time points, cells were fixed, incubated at 1:500 in primary phospho-histone H2AX antibody (Millipore ) and 1:500 for with secondary Alexa Fluor 488. Images were acquired with a fluorescent microscope with a 40X objective and analyzed in AIM Image Browser. Student s t-tests were used to determine statistical significance. CFU Assay Whole bone marrow was isolated from WT and Exo1 mut mice and irradiated at varying doses of IR. Fifty thousand cells were plated in 3 cm 2 plates coated with complete methylcellulose media containing IL3, IL6, SCF, Epo (Stem Cell Technologies M3434) and scored after 14 days. These conditions allow for outgrowth of GCU-GM, CFU-E, BFU-E 34

46 and CFU-GEMM., Summation of differential CFU counts is reported. Student s t-tests were used to determine statistical significance. Blood and Marrow Characterization Studies Peripheral eye blood was extracted from mice, the red blood cells lysed, and cytometric analysis performed examining B220, CD3 and Mac1. To examine the SKL population WT and Exo1 mut mice were sacrificed, bone marrow flushed from femurs and tibias, and cytometric analysis performed using sca1, c-kit, and lineage markers CD3,B220,CD11b,Ter119,Cd4 (BD Biosciences) on a BD LSRII cytometer (BD Biosciences). Data was analyzed on Flowjo Version 8.8 software (Treestar), [39]. Student s t-tests were used to determine statistical significance. Competitive Repopulation Studies One million whole bone marrow cells from 8-12 week WT or Exo1mut mice (CD45.2) were mixed with age matched WT mice (CD45.1) at a 1:1 ratio and injected via tail vein into lethally irradiated CD45.1 recipients. At eight and sixteen weeks post transplant the mouse chimerism was measured via peripheral blood measuring CD45.1/CD45.2 as well as surface markers B220, Mac1, and CD3 on a BD LSRII cytometer[104]. Student s t-tests were used to determine statistical significance. Serial Transplant Studies Recipient BoyJ mice (CD45.1) were lethally irradiated (11Gy) and 16 hours later 2 million WT or Exo1 mut (CD45.2) whole bone marrow cells were injected via tail vein. 16 weeks post transplant these mice were sacrificed, bone marrow flushed and 2 million cells 35

47 were subsequently transplanted into the next batch of lethally irradiated BoyJ mice for 4 rounds of transplant[36]. Niche Occupancy Assay HSC niche occupancy retention was measured by injecting 5 million whole bone marrow cells from WT 8-12 week old CD45.1 mice into non myeloablated WT or Exo1 mut CD45.2 mice. At 8 and 16 weeks post transplant peripheral eye blood was obtained from recipient mice, stained for CD45.1/CD45.2, B220, Mac1, and CD3 and analyzed on a BD LSR II[39]. Student s t-tests were used to determine statistical significance. 5-FU Mobilization Studies WT and Exo1 mut mice were IP injected with 150 mg/kg 5-FU (Sigma). To measure BrdU incorporation 1mg/mouse BrdU (Sigma-Aldrich) was IP injected into the mice 4 days after 5-FU treatment. 16 hours after 5-FU treatment the mice were sacrificed, marrow was flushed from the femurs and tibias, and stained for SKL and BRDU. The percent of BRDU+ SKL cells was measured using a BD LSRII [105]. Poly(I:C) Studies Mice were treated twice with 10ug/g poly(i:c) (Invivogen) separated by 48 hours. After another 48 hours the mice were treated with 6.5Gy IR and the assays described previously were performed on these treated mice. 2.4 Results: Exo1 mut Fibroblasts Display DNA DSB Repair Defects 36

48 We used Exo1 mut mice[69] to characterize the DSB response. These mice contain nuclease dead Exo1 but possess a slightly truncated form of the full length protein. Exo1 mut mice have been characterized as displaying MMR defects, increased cancer susceptibility, and sterility. Murine embryonic fibroblasts (MEFs) were derived from gestation day 13.5 females and DSB repair proficiency was measured by reporter plasmids. Homologous recombination proficiency was determined using the DR GFP plasmid[102] in which a double strand break is introduced through expression of the I- SCE1 endonuclease with functional gene conversion resulting in GFP expression. Exo1 mut MEFs displayed a marked decrease in HR-reporter activity when compared to WT MEFs indicating an impaired HR pathway (Figure 2.1a, p<0.05), a link previously established by Exo1 depletion in MCF7 cells[85]. Conversely, using the non-homologous end joining reporter plasmid pegfp-pem1-ad2[103] in an assay in which this plasmid is linearized by HINDIII digestion such that upon recircularization via end joining GFP is detected, we found that Exo1 mut MEFs had normal NHEJ activity (Figure 2.1a, p>0.05). This confirmed that the DSB repair defect in Exo1 mut MEFs is HR pathway specific, which additional studies have demonstrated is due to impaired 5 3 resection and initiation of HR[68, 83, 106]. To determine the functional repair defect in Exo1 mut MEFs, we examined the response to ionizing radiation. Using MTT survival assays, we show that the Exo1 mut fibroblasts displayed a ~2-fold increase in radiation sensitivity compared to WT MEFs. This sensitivity was also observed using additional DSB inducing agents etoposide and hydroxyurea (Figure 2.1b). Furthermore, we quantified γ-h2ax immunofluorescence 37

49 staining as a measure of DSBs on WT and Exo1 mut MEFs at various time points following irradiation and found that the Exo1 mut cells contained persistent γ-h2ax foci after 24 hours (Figure 2.1c, p<0.05). Together these data suggest that the loss of Exo1 function in cells undergoing active proliferation results in a survival defect after DSB formation. To compare the fibroblast response to the hematopoietic population, we used a colony forming unit (CFU) assay of unfractionated mouse bone marrow after radiation. Whole bone marrow was collected from WT and Exo1 mut mice, treated with an IR dose range (0.5-5Gy) following which the cells were cultured in methylcellulose in the presence of cytokines, and colony forming unit (CFU) survival enumerated on day 14. This assay induces cell division of the HSC due to cytokine exposure, so the read out is of cytokine responsive proliferating progenitor cells [107]. We examined the ability of WT and Exo1 mut marrow to produce multiple CFU types: CFU-GM, CFU-GEMM, and BFU-E. Exo1 mut marrow had normal numbers of CFU. Since the distribution of CFU among types was not altered, and the impact affected all sublineages, we report the collective data. However, marrow CFU from these mice were more IR sensitive than WT marrow at all doses tested, indicating that when HSC progenitor cells are treated with IR and immediately subject to cytokine proliferative signals, the loss of Exo1 activity results in DNA repair defects and a survival disadvantage (p<.05, Figure 2.1d). Exo1 mut Mice Display Normal Marrow Characteristics To understand the general effect that inactivation of Exo1 had on animal physiology and HSC maintenance, we measured body weight and blood counts of age 38

50 matched WT and Exo1 mut mice and found no change in weight (WT / g vs. Exo mut / g) nor numbers of white blood cells (WT-8.2 +/ K cells/ul vs. Exo mut / K cells/ul), red blood cells (WT / M cells/ul vs Exo mut / M cells/ul), or platelets (WT / K cells/ul vs. Exo mut / K cells/ul). Additionally we observed no differences in total numbers of lineage-, Sca1+, Kit+ (SKL) or the more primitive SKL CD150+,CD48- (SLAM) HSCs in age matched 8-12 week old mice (p>0.05). We also found that levels of T-cells, B-cells and myeloid cells were unchanged in the Exo1 mut mice, suggesting that the loss of Exo1 activity has no effect on steady state HSC progenitor multilineage differentiation (Appendix 1, p>0.05). This apparently normal steady state hematopoietic phenotype appears much milder than FA, Rad50, or BRCA2 mouse models [37, 108, 109], which are associated with loss of HR. Exo1 mut Mice Display no Competitive Repopulation, Serial Transplant or Niche Occupancy Defects To determine whether HSCs from Exo1 mut mice had inherent stem cell defects we performed HSC engraftment and transplant studies. Competitive repopulation studies were performed with Exo1 mut mice (these mice express the CD45.2 variant) to determine whether the loss of Exo1 function would lead to a defect in engraftment and reconstitution when directly competed against WT marrow from mice that express the CD45.1 variant of CD45. One million CD45.1 WT whole bone marrow cells and one million CD45.2 Exo1 mut marrow cells from 8-12 week age matched mice were 39

51 transplanted into lethally irradiated CD45.1 WT recipients. Chimerism was determined by flow cytometry of CD45.1 or CD45.2 marking in blood after 8 and 16 weeks[104]. Exo1 deficient marrow cells displayed no competitive repopulation defect when competed against marrow from WT mice after transplantation (Figure 2.2a, p>0.05). Since this tests the competency of resting and quiescent HSC, this suggested that there is no defect in the Exo1 mut mice. Long term HSC reconstitution and engraftment capabilities were measured via serial transplantation. Two million mouse bone marrow cells were injected into lethally irradiated WT recipients. Marrow from these mice was subsequently injected into a new set of lethally irradiated recipients sequentially at 16 week intervals until the mice can no longer survive typically 4-6 passages for WT mice [110], a biologic marker of the HSC exhaustion. HSCs with intrinsic defects, specifically those with NHEJ or mismatch repair defects, lose the capability to reconstitute recipients after fewer rounds of serial transplantation [36, 38]. Exo1 mut mice were able to repopulate recipient mice equivalent to WT mice after four rounds of transplantation (Figure 2.2b, p>0.05). We also measured HSC hematopoietic niche occupancy of Exo1 mut mice by performing transplantation assays into unconditioned recipient mice. NHEJ deficiency has been reported to lead to long term engraftment of WT donor cells into unconditioned NHEJ deficient mice [39], demonstrating an inherent defect in HSC hematopoietic niche occupancy. As reported, this assay identifies HSC that are not secure in the quiescent hematopoietic stem cell niche, allowing engraftment and long 40

52 term hematopoiesis from the invading cell population to a far greater extent than observed in normal recipients. We found that Exo1 mut mice, like WT mice, displayed no engraftment (< 0.5%) of WT donor marrow in the absence of conditioning (Figure 2.2c, p>0.05). Exo1 mut mice display no radiation sensitivity or competitive repopulation defect after IR treatment Loss of Exo1 expression results in increased IR sensitivity in differentiated dividing cells[86]. The HSC population at steady state is predominantly quiescent and thus reliant on NHEJ. Therefore, we predicted that Exo1 mediated loss of HR would not affect HSC IR sensitivity in otherwise unperturbed Exo1 mut mice. This is distinctly different than the studies observed in Figure 2.1 that evaluated dividing hematopoietic CFU. Exo1 mut and WT mice were treated with either a lethal (8Gy) or sublethal (4-6Gy) dose of IR and animal survival monitored over 6 months. We observed no survival differences between the two groups of mice in the dose range of 4-8Gy (Figure 2.3a). To specifically look at Exo1 mut HSC IR sensitivity, we performed competitive repopulation transplantation of cells into lethally irradiated WT mice as described above. After initial reconstitution, the recipients were given an additional 6 Gy of ionizing radiation at 8 weeks post transplantation to challenge the donor HSCs in order to determine whether loss of Exo1 function impacted DSB repair of HSC after transplantation. At 1, 5, and 8 weeks following the radiation dose, cytometric analysis was performed on peripheral blood. We found that one week post IR treatment, the Exo1 mut cells demonstrated 41

53 decreased chimerism compared to WT (p<0.05) but this recovered by five and eight weeks post transplant to equal chimerism (Figure 2.3b). This suggests that the proliferating short term repopulating Exo1 mut HSCs were IR sensitive whereas the quiescent long term repopulation cells were not dependent on Exo1 after irradiation. Exo1 mut mice Display Increased IR Sensitivity after HSC cell cycle entry The CFU analysis and the competitive repopulation studies (Figures 2.1d and 2.2a) suggest that Exo1 is dispensable for radiation survival at steady state but needed for mediating HR after irradiation induced DSB. Because HSCs normally reside in a quiescent state, it is not surprising that they rely on NHEJ, and that Exo1 and HR function cannot be elucidated. However, we hypothesized that HSCs in active cell cycle would be dependent on Exo1 and intact HR after IR. We wanted to determine whether NHEJ would complement for loss of Exo1 and HR or whether Exo1 and HR were essential after DSB for maintenance of HSC function. If HR is required for DSB repair in cycling HSCs, Exo1 deficiency should result in increased IR sensitivity. To test the role of Exo1 in DSB repair for cycling cells, we induced the HSC population from quiescent to cycling in vivo using the chemotherapeutic agent 5-fluorouracil, and 5 days later, at a point of active cell cycle, treated these mice with whole body radiation. 5-FU treatment forces the quiescent HSC population to enter cell cycle 4-6 days after treatment in order to replenish the depleted hematopoietic pool, leading to reconstitution through selfreplication[ ]. Under these conditions cells without functional HR could become 42

54 susceptible to exogenous IR stress and might rely on NHEJ which, if sufficient, would compensate for loss of HR. To confirm that 5-FU induced a damage mediated proliferative response in HSC in vivo, we treated mice with 5-FU alone and measured proliferation via incorporation of the synthetic thymidine analog BrdU in the SKL (lin-/sca1+/c-kit+) progenitor population. Five days post 150mg/kg 5-FU treatment, bone marrow was harvested from WT and Exo1 mut mice and the percent of BrdU + SKL cells was measured. We found that in both mouse strains the percentage of cycling BrdU + SKL cells increases approximately 2.5 fold (Appendix 2). We also showed that a single dose of 150 mg/kg 5-FU was not more toxic to Exo1 mut mice than to WT (Appendix 2). We asked whether cycling HSCs become dependent on Exo1 for survival following IR. WT and Exo1 mut mice were injected with 150 mg/kg 5-FU IP, and 5 days later were treated with 4 Gy IR. We measured bone marrow cellularity and colony forming unit potential of both groups seven days following the second treatment. Bone marrow cell counts in Exo1 mut mice contained approximately 40% fewer total bone marrow cells than their WT counterparts (Figure 2.4a). Whole bone marrow cell preparations from these mice were plated in methylcellulose containing cytokines and the number of colonies formed was counted at 14 days. A drop in each of the major classes of CFU were observed, so only composite CFU data are reported. Marrow cells from Exo1 mut mice formed ~50% fewer CFU colonies than those from WT mice (Figure 43

55 2.4a). This demonstrated that HSCs from Exo1 mut mice exhibited more severe toxicity compared to WT marrow HSC. We measured the concentration and total numbers of SKL cells in mice 14 days after the dual treatment and showed that while WT mice were able to restore normal levels of SKL cells in their bone marrow, Exo1 mut mice demonstrate a significant decrease in SKL levels 14 days post treatment (p<0.05). Thus, both SKL and CFU recovery after sequential 5-FU and irradiation in Exo1 mut mice indicate a significant loss of HSC at short term time points (Figure 2.4b). To determine if the HSCs in WT or Exo1 mut mice differed in long term repopulation function after sequential treatment, we collected whole bone marrow from WT and Exo1 mut mice 14 days after the 5-FU+ 4 Gy IR, and performed competitive repopulation assays at a 1:1 ratio of dual treated marrow using CD45.1/CD45.2 to distinguish the populations. We observed that Exo1 mut marrow was significantly impaired in repopulating capability 8 weeks post transplant compared to WT marrow (Figure 2.4c). This confirmed the observations showing that the combination of 5-FU+IR was more toxic to Exo1 mut marrow than to WT marrow HSC. In additional experiments, we observed that 13 of 15 Exo1 mut mice died within 6 weeks of the combination 5-FU+IR treatment while none of the WT mice died (Figure 2.4d). Thus both short term and long term HSC function during and after irradiation induced DSBs is impaired with loss of Exo1 activity. Further, intact NHEJ is not able to compensate for loss of HR in cycling HSCs. 44

56 Since 5-FU may induce DNA damage that is impaired by Exo1 loss of function, we confirmed these results using an additional agent that could induce HSC into cell cycle prior to irradiation. WT and Exo1 mut mice were treated with the double stranded RNA mimetic poly IC which has been shown to promote the proliferation of dormant HSCs in vivo [114]. We confirmed this proliferation induction in WT and Exo1 mut mice by treating twice with poly IC and measuring BRDU incorporation 2 days after treatment (Figure 2.5a). We treated mice with poly IC followed by 6.5 Gy IR and observed that Exo1 mut mice displayed significantly reduced bone marrow cellularity, CFU formation, and SKL compared to WT marrow (Figure 2.5b,c, p<0.05). Additionally, all Exo1 mut mice treated in this manner died within 6 weeks of treatment while all WT mice survived the poly IC + IR (Figure 2.5d). These data confirm that HSC proliferation results in an increased reliance on Exo1-mediated HR for repair of DSBs. Further, it indicates that proliferating HSC cannot compensate for loss of HR by using NHEJ alone for DSB repair. 2.5 Discussion: These results indicate that lack of fully functional HR, mediated by Exo1 mut, in HSCs leads to a proliferation-dependent sensitivity to irradiation induced DSB even though such sensitivity is not observed in steady state. The IR hypersensitivity is uncovered by treating mice with 5-FU to induce death of proliferating hematopoietic progenitor cells or by poly IC, both of which induce HSC into cell cycle. In mice containing the Exo1 mutation, these cycling HSCs lack fully competent HR, cannot repair DSBs by that mechanism, are not compensated in function by NHEJ, and lose 45

57 repopulating function and marrow regenerating capacity. However, this functional HSC defect is not observed under steady state conditions. As a consequence, mice treated under conditions of HSC proliferation die of marrow hypoplasia and the residual marrow is not capable of competitive repopulation. Although others have shown that HSC utilize both HR and NHEJ in response to DSBs after IR, our data indicate that intact NHEJ is insufficient to compensate for loss of HR in DSB repair among proliferating HSCs. There are a number of novel findings in this study. First, we noted normal hematopoietic function of the Exo1 mut mouse under steady state conditions and under conditions of physiologic proliferation during transplantation and competitive repopulation. In the absence of exogenous DNA damage, HSC function was normal. We showed that under quiescent steady state conditions Exo1 mut mice displayed no characteristics of stem cell failure as measured by competitive repopulation, serial transplant, and niche occupancy. Second, Exo1 mut HSCs response to IR induced DSB was normal in vivo. Most likely this indicates that DSB repair in quiescent HSC is mediated by intact NHEJ. Third, our work further clarifies the link between DNA repair dependence and cell cycle status of HSCs. When Exo1 mut hematopoietic progenitors are exposed to irradiation in vitro and then assayed in the presence of cytokines for CFU, a defect is seen. Further, when HSC of these Exo1 mut mice are forced into cell cycle by 5-FU (Poly IC, or likely other sequentially administered chemotherapy), the DSB repair defects results in significant loss of HSC function. What appears to be a normal response to a 46

58 single irradiation dose in HSC without HR is severely compromised when HSC are proliferating. In addition to the HSC population, the hematopoietic failure after cell cycle activation may be caused by damage to multiple cycling populations including myeloid and lymphoid progenitors. Since hypoplastic hematopoiesis leading to animal death is the essence of inadequate marrow reserve, the susceptibility observed in Exo mut mice reflects a combined impact of loss of HR on HSC and progenitor cell death. These results model in vivo the switch between sole NHEJ reliance in HSCs to a preferred reliance on HR during proliferation. Utilizing the HR impaired Exo1 mut mice allowed us to contrast our data with those of the well established NHEJ mouse models (SCID, KU80, KU70) which demonstrate steady state HSC DNA repair defects, hematopoietic stem cell deficiency and irradiation hypersensivitity. The results clarify in greater detail the relationship between cell cycle status and DNA repair dependency, and demonstrate that HR becomes essential for DSB processing after HSC cell cycle entry. This in vivo model of pushing dormant HSCs into cycle is a novel way of clarifying how DNA repair shifts along with cell cycle status in HSCs and could potentially be used to study how other DNA repair pathways are altered by cell cycle as well. Additionally this data may have an impact in the field of leukemic and other cancer stem cells, another population believed to live between states of quiescence and active cell cycle and it may be interesting to determine how cell cycle entry in this population affects DNA repair reliance. 47

59 At steady state, long term HSCs are not completely quiescent but instead have small proportions of cells entering cell cycle over a 24 hour period giving rise to low population cycling rates[22, 24, 26]. Our study examines a point in time radiation exposure and observes a shift in DNA repair reliance from solely NHEJ for the quiescent population to HR dependency when there is a rapid shift of the HSC population from quiescence into cell cycle. This study design simply shifts the HSC population from largely quiescent to one with a sizable proportion in cell cycle, uncovering this rather significant sensitivity to DSB in the absence of functional HR. This is also likely to occur for the population in cell cycle under steady state conditions, albeit at an inconsequential frequency. Alternatively, the DSB stress response of IR, known to be coupled with induction of stimulatory cytokines, may shorten the cell cycle time for individual stem and progenitor cells, increasing their dependency on HR. Schaetzlein et. al have recently described the derivation of an Exo1 null mouse which has a very similar DSB response phenotype to that of the Exo1 mut mouse used in these studies. In that report, the authors propose that the nuclease function of Exo1 is important for its role in end resection of DSBs but appears dispensable for its roles in DNA mismatch repair and class switch recombination [80]. Thus because our studies focus on the DSB response of Exo1 mut HSCs, we would anticipate that the Exo1 null mouse would have a similar IR response phenotype to the Exo1 mut mouse. Other pathways have also linked loss of HR with HSC dysfunction. Studies of mutations in a BRCA2/FANCD1 mutant mouse model have demonstrated defects in both 48

60 HSC repopulation and proliferation capabilities. As is currently understood the core nuclear FA complex responds to DNA damage by monoubiquitinating FANC1 and FANCD2, this subsequent complex interacts with other downstream proteins including BRCA2 and RAD51C[100]. The effect of these interactions is still being elucidated but it is believed that the FA activation results in enhancement of HR. While BRCA2 is a well established HR protein, it also has been implicated in the Fanconi anemia (FA) pathway and the HSC phenotype is similar to those previously described in FA mouse models including Fancc -/- and Fancd2 -/- mice[108, 115]. The Fancc -/- and the Fancd2 -/- mice exhibit a loss of HSC quiescence (different than the more mild phenotype of Exo1 mut mice), with both models displaying significantly higher percentages of HSCs in cell cycle. The BRCA1/Fancd1 mice have been shown to allow engraftment of donor HSCs in the absence of any preconditioning, again suggesting a loss of quiescence [37]. We have shown that loss of quiescence results in a transition to greater HR dependence in the cycling HSCs, similar to that observed in FA mice and humans with FA disorders. As we have shown in the Exo1 mut mouse, increased cycling HSCs with compromised HR results in HSC defects especially in response to exogenous stress. The BRCA2/FANCD1 mice demonstrated a severe IR sensitivity, dying within 5-9 days of 7Gy IR, potentially due to the combined effect of quiescence loss and impaired HR in the cycling HSCs. Thus, the fanconi anemia hematopoietic phenotype can be partially explained by an increase in cycling HSCs that are HR dependent but also HR impaired, and thus unable to repair damage from exogenous stress due to their cell cycle status and DNA DSB reliance. 49

61 Exo1 polymorphisms have been associated with multiple types of human cancers including gastric, breast, and lung[ ]. While conclusive evidence linking Exo1 polymorphisms to disease pathogenesis has yet to be uncovered, our work also suggests that patients containing these polymorphisms may be susceptible to HSC toxicity from chemotherapy and irradiation. For instance sequential doses of IR may result in HSC activation and subsequent hypersensitivity due to loss of Exo1 function. Studies involving these patients and their sensitivity to chemotherapy may yield new evidence of HSC activation and chemosensitivity in humans with Exo1 or other HR DNA repair gene polymorphisms. These observations raise new questions. Does a defect in NHEJ also have differential effects among quiescent vs. proliferating cells? Can counter measures to retain HSC in quiescence reduce or eliminate the DSB repair defect or at least its impact on robust HSC function? Does this cycling dependent defect explain other hypersensitivity genetic defects of DNA repair? Further, are there human defects in these pathways that give rise to DSB hypersensitivity, marrow failure and leukemic transformation? These studies provide a clear teleological rationale for maintaining stem cell populations in quiescence and point to the feedback and complementation mechanisms to protect from forced induction of proliferation, especially after DNA damage. 50

62 2.6 Acknowledgements: This work was supported by the Cytometry & Imaging Microscopy Core Facility of the Case Comprehensive Cancer Center and the Radiation Resources Core Facility of the Case Comprehensive Cancer Center (2P30 CA (Gerson, PI)). It was also supported by National Institutes of health grant 5R42 CA (Gerson, PI) and molecular therapeutics grant 5T32GM

63 Figure 2.1: Exo1 mut fibroblasts and hematopoietic progenitors display DSB repair defects (A) Reporter plasmid quantitation of DR-GFP (HR) and Pem1-Ad2-GFP (NHEJ) in MEFs. Flow cytometry data measured % GFP positive cells. (B) Drug survival studies in MEFs quantified using MTT assay. Error bars represent SEM. (C) Confocal images of γ-h2ax immunostaining. 50 cells per treatment were analyzed using 40x magnification for foci formation and the % positive was plotted. (D) Colony survival analysis of whole bone marrow. Numbers of colonies at each IR dose were counted 14 days post plating in methylcellulose. Error bars represent SEM. P values were calculated by 2-tailed student t test. * p<0.05 vs. WT. 52

64 Figure 2.1: Exo1 mut fibroblasts and hematopoietic progenitors display DSB repair defects 53

65 Figure 2.2: Exo1 mut mice exhibit no defects in competitive repopulation, serial transplant, or niche occupancy (A) Representative FACS profiles. Competitive repopulation assay on WT and Exo1 mut HSCs. 1*10 6 whole bone marrow cells from 8-12 week old Exo1 mut mice (CD45.2) were mixed 1:1 with age-matched WT competitors (CD45.1) and transplanted via tail vein injection into lethally irradiated CD45.1 recipients. (B) Serial transplant results from WT and Exo1 mut mice. Marrow was flushed from age matched CD45.2 mice and 2*10 6 cells were transplanted into lethally irradiated CD45.1 recipients sequentially. (C) Niche occupancy proficiency results 16 weeks post transplant via flow cytometry. P values were calculated by 2-tailed student t test. * p<0.05 vs. WT. 54

66 Figure 2.2: Exo1 mut mice exhibit no defects in competitive repopulation, serial transplant, or niche occupancy 55

67 Figure 2.3: Exo1 mut HSCs at steady state are not more IR sensitive than WT HSCs (A) Age matched (8-12 week) WT and Exo1 mut mice were treated with whole body IR and animal survival was monitored at 6 months post treatment. (B) WT and Exo1 mut whole bone marrow was mixed 1:1 and injected into lethally irradiated CD45.1 recipients. 8 weeks post transplant the recipient mice were treated with 6Gy IR. Cytometric analysis was performed on peripheral blood 1,5, and 8 weeks following the IR treatment to determine percent chimerism. 56

68 Figure 2.3: Exo1 mut HSCs at steady state are not more IR sensitive than WT HSCs 57

69 Figure 2.4: Exo1 mut mice become IR sensitive after 5-FU induced HSC cell cycle entry (A) Age matched WT and Exo1 mut mice were treated with 150 mg/kg 5-FU followed by 4Gy IR 5 days post 5-FU. Seven days post dual treatment 3 mice from each group were sacrificed and bone marrow cellularity and CFU counts were measured. Error bars represent SEM. (B) WT and Exo1 mut mice were treated with the combination of 5-FU and IR and 14 days post treatment 3 mice were sacrificed and flow cytometry was performed on whole bone marrow to determine SKL numbers. Error bars represent SEM. (C) Dual treated WT and Exo1 mut whole marrow were mixed 1:1 and injected into lethally irradiated recipients. Animal chimerism was measured using CD45.1/2 flow cytometry 8 weeks post transplant. (D) Numbers of WT and Exo1 mut mice that survived the 5-FU + IR combination treatment two months post treatment. P values were calculated by 2-tailed student t test. * p<0.05 vs. WT. 58

70 Figure 2.4: Exo1 mut mice become IR sensitive after 5-FU induced HSC cell cycle entry 59

71 Figure 2.5: Poly IC mediated cell cycle entry also results in HSC IR hypersensitivity in Exo1 mut mice only (A) WT and Exo1 mut mice were treated twice with 10 ug/g poly(i:c) separated by 48 hours. 48 hours following the second treatment the percentbrdu+skl cells were recorded via flow cytometry of whole bone marrow. (B) WT and Exo1 mut mice were treated with poly(i:c) followed by 6.5 Gy IR. Fourteen days post treatment 3 mice from each group were sacrificed and bone marrow cellularity and CFU counts were measured.error bars represent SEM. (C) WT and Exo1 mut mice were treated with poly(i:c)+6.5gy and fourteen days post treatment 3 mice were sacrificed and flow cytometry was performed on whole bone marrow to determine SKL numbers. Error bars represent SEM. (D) Numbers of WT and Exo1 mut mice that survived the poly(i:c) + IR combination treatment two months post treatment. P values were calculated by 2- tailed student t test. * p<0.05 vs. WT. 60

72 Figure 2.5: Poly IC mediated cell cycle entry also results in HSC IR hypersensitivity in Exo1 mut mice only 61

73 Chapter 3 2 CD133+ cells contribute to radioresistance via altered regulation of DNA repair genes in human lung cancer cells 3.1 Abstract Acquired radioresistance is implicated in treatment failure and tumor relapse in multiple human cancers. While the mechanisms aren t completely understood, studies have implicated a subset of stem like cells within the heterogeneous tumor population that contribute to resistance. Once such marker of this subset is Prominin 1 (CD133), which has been associated with increased tumorigenic potential and resistance to chemotherapy. Here we study the role of DNA repair genes in the ionizing radiation (IR) response of CD133+ cells in human non small cell lung cancer cell lines A549 and H1299. We show that A549 cells exclusively expand their CD133+ population in a time dependent manner after IR exposure, which corresponds with IR resistance and improved resolution of γ H2AX foci in separated A549 CD133+ cells. To determine the role of DNA repair genes in this observation we performed real time PCR on CD133- and CD133+ cells and found that A549 cells displayed upregulated expression of multiple DNA repair genes, while H1299 cells exhibited no such increase. We found that prior IR exposure and subsequent recovery resulted in acquired DNA repair upregulation in the CD133+ population of both A549 and H1299 which corresponded with increased radioresistance and improved γ-h2ax resolution. Finally we silenced Exo1 and Rad51 in A549 cells and demonstrated that loss of these enzymes abrogated the CD133+ IR 2 A version of this chapter was accepted for publication in the journal Radiotherapy and Oncology on December 9th,

74 expansion phenotype and induced IR sensitivity in sorted CD133+ cells. We believe CD133 identifies a population of cells within specific tumor types containing altered expression of DNA repair genes that are inducible upon exposure to chemotherapy. This altered gene expression contributes to enhanced DSB resolution and the radioresistance phenotype of these cells. We also identify DNA repair genes which may serve as promising therapeutic targets to confer radiosensitivity to CSCs. 3.2 Introduction The emergence of cancer stem cells (CSCs) in recent years as drivers of tumor maintenance and chemoresistance has resulted in new approaches to studying the biology of cancer. CSCs have been implicated in a multitude of human malignancies including tumors of the brain, breast, lung, colon, pancreas, and others[49, 50, ]. While they are identified using cell surface markers, CSCs are phenotypically distinct from the bulk of a tumor by their ability to reconstitute tumors using a small number of cells. CSCs have been associated with an increased metastatic potential and chemoresistance in multiple human cancers and thus may be a significant contributor to tumor relapse and treatment failure [45, 47]. One such marker believed to enrich for a CSC population is Prominin 1 (CD133). CD133 is a transmembrane glycoprotein with still unknown function, although it is found in embryonic stem cells, normal adult stem cells and circulating endothelial progenitors. CD133 has three known isoforms (CD133-1, CD133-2, CD133-3) which are expressed on different tissue types, and loss of CD133 has been correlated with an increase in terminal differentiation[44, 122]. CD133 expression has been associated with 63

75 chemoresistance and increased metastatic potential in multiple human cancers although the mechanistic relationship between Prominin 1 and these phenotypes is still unknown. Many factors are thought to contribute to the chemoresistance observed in cancer stem cells including an upregulated capacity for DNA repair. Bao et. al demonstrated that human glioma stem cells become enriched after radiation therapy in primary tumors. Sorted CD133+ cells were shown to exhibit an activated capacity for DNA repair, including basal phosphorylation of the checkpoint proteins Chk1 and Chk2. Inhibition of Chk1 with a selective inhibitor resulted in subsequent IR sensitivity suggesting that DNA repair proteins may provide promising therapeutic targets for CSCs[62]. While groups have demonstrated an increased DNA repair capacity in CD133+ cells including in medulloblastoma cells[123] and prostate cancer cell lines [124],[124] others have shown that CD133+ cells do not equate to increased radiation resistance nor to an enhanced DNA repair capacity[63, 125] and thus the impact of CD133 on radioresistance and enhanced DNA repair may be tumor type specific. In human lung cancer cells, CD133 has also been shown to enrich for a cancer stem cell like population[121, 126]. Indeed Shien et. al (2012) examined the expression of CD133 in 50 patients with non-small cell lung cancer and found that those patients with CD133-positive samples had significantly worse 5-year survival rates than those that were CD133-negative[59]. Iida et. al have proposed that hypoxic conditions promote CD133 expression in lung cancer cell lines via Oct4 and Sox2 activity[127] and 64

76 groups including Bertolini et. al have shown that in lung cancer cell lines CD133+ cells are enriched in response to cisplatin therapy[128]. Liu et. al have recently proposed that this resistance is due to activation of the notch pathway and that sensitivity can be induced via notch inhibition[129]. However the role that DNA repair plays in contributing to IR resistance in CD133+ lung cancer cells has yet to be fully elucidated. In this manuscript we demonstrate the role that DNA repair capacity plays in the ionizing radiation (IR) response of CD133 populations in two human lung cancer cells lines. We compare CD133+ IR resistance in A549 cells against H1299 cells and show that CD133 contributes to radiation resistance in only A549 cells in part due to increased capacity to resolve DNA double strand breaks (DSBs). We additionally examined the DNA repair capacity and mrna gene expression levels of multiple DNA repair genes in CD133+ vs. CD133- cells and found that only in A549 cells did the CD133+ population exhibit a higher basal expression of these genes. We treated both cell lines with multiple exposures of ionizing radiation and after a recovery period found that CD133+ cells demonstrated an acquired IR resistance in both cell lines, and that in H1299 cells this corresponded with an induction of DNA repair expression. Finally we used lentiviral mediated shrna silencing of two consistently upregulated CD133+ repair genes (Rad51 and Exo1) in A549 cells and showed that this inhibition abrogated CD133 expansion post IR and also conferred radiation sensitivity in CD133+ cells. This work demonstrates that CD133 identifies a population of cells with altered expression of DNA repair genes that are induced by exposure to radiation therapy. This altered expression appears to 65

77 contribute to radiation resistance of these cells, thus identification of critical repair elements in CD133+ cells may yield promising therapeutic targets. 3.3 Materials and Methods Cell lines: Human non small cell lung cancer cell lines A549 and H1299 acquired from ATCC were used in these studies. H1299 cells contain a homozygous partial deletion of p53 which results in loss of p53 protein while A549 cells are WT for p53 but contain a KRAS mutation. Flow cytometry: For CD133 IR expansion studies A549 and H1299 cells were treated with 4Gy IR and at multiple time points cells were trypsinized, incubated with CD133 antibody (293C3- Miltenyi Biotec), and flow cytometry performed on a BD LSRII instrument to measure the percent of PE+ cells. CD133 Separation: Separation of CD133+ cells from CD133- cells was performed using the CD133 microbead kit (Miltenyi Biotec) and MACs separation columns (MS columns- Miltenyi). Separation was confirmed via flow cytometry using the CD133 antibody on sorted cells. Clonogenic Survival Assays: After CD133 separation, 100 cells were plated in 3 cm 2 plates and treated with an IR dose range days post treatment the plates were stained with crystal violet for 66

78 colony counting. Only colonies containing >50 cells were counted. Student s t-tests were performed to determine significance. Immunostaining: Cells were grown on coverslips and 24 hours post plating were treated with 4Gy of ionizing radiation. At the indicated time points, cells were fixed in 4% paraformaldehyde, permeabilized in 0.1% Triton X-100, blocked with 5% NGS, and incubated at 1:500 for one hour in primary phospho-histone H2AX antibody (Millipore ). Cells were washed three times in PBS and incubated at 1:500 for 45 minutes with Alexa Fluor 488. Coverslips were washed and mounted onto coverslips with Dapi fluoromount G (southern biotech). Images were acquired with a fluorescent microscope (Nikon) with a 40X objective and analyzed in AIM Image Browser. Real-time PCR: Separated CD133 cells were assessed for gene expression via realtime PCR. RNA was extracted using the Trizol method from CD133- and CD133+ cells and cdna synthesis was performed (Superscript III First Strand Kit- Invitrogen). Real-time PCR was performed using validated primers (Applied Biosystems). Student s t-tests were performed to determine significance. Lentiviral Gene Silencing: Rad51 and Exo1 were silenced via shrna transduction with validated clones (Sigma). Lentiviral particles were synthesized via HEK293 cells and target cells were infected, selected for with puromycin, and clones were assessed for verification of gene silencing. 67

79 3.4 Results CD133+ cell enrichment post IR treatment Lung cancer cells have previously been shown to expand their CD133 population following chemotherapy[128]. Here we wanted to assess the effect of ionizing radiation on this CD133 expansion phenotype. A549 and H1299 cells were treated with 4Gy IR and the percent of CD133+ cells in culture at multiple timepoints (24, 48, 72 hrs post treatment) was assessed via flow cytometry. We show in A549 cells by 48 hours the percentage of CD133+ cells had more than doubled compared to untreated, and expanded approximately five-fold by 72 hours. In H1299 cells the expansion was not significant at any time point measured (Appendix 3). Radiosensitivity and DNA repair capacity of separated CD133+ cells To study whether CD133 contributes to radiation resistance in these cell lines we performed magnetic separation of CD133+ from CD133- cells (Miltenyi) and subjected these sorted cells to an IR dose range of 2Gy- 8Gy. The purity of magnetically sorted cells was confirmed via flow cytometry (Appendix 4) Cell viability was assessed days post treatment via clonogenic survival. In A549 cells separated CD133+ cells conferred approximately a 2-fold increase in radiation resistance, while H1299 cells demonstrated an insignificant difference in IR sensitivity (Figure 3.1a). To assess the role of DNA repair proficiency in the radiation response data we quantified ϒ-H2AX immunofluorescence staining as a measure of DNA double strand breaks on CD133- and CD133+ cells from A549 and H1299 cells at multiple time points following irradiation. We found that in A549 cells exclusively the CD133+ cells were able to completely 68

80 resolve DSBs by 24 hours post treatment while the CD133- cells retained persistent H2AX foci. In H1299 cells both CD133- and CD133+ populations retained some persistent DSBs by 24 hours (Figure 3.1b). This data corroborates previous suggestions that CD133 serves as a marker for radioresistance only in specific cell and tumor types. DNA repair gene expression analysis in CD133+ cells To confirm that the increased ability of A549 CD133+ cells to resolve H2AX foci was due to an enhanced capacity for DNA repair we performed real time PCR analysis on separated CD133+ cells in A549 and H1299 cells. We hypothesized that the level of IR resistance observed in Figure 3.1 would correlate with basal gene expression levels of DNA repair genes. We examined the expression of genes implicated in multiple DNA repair pathways, but a particular focus was given to DNA double strand break (DSB) repair pathways due to the dominant type of IR induced DNA lesions. Double strand breaks are repaired by two conserved pathways in mammalian cells, homologous recombination (HR) and non-homologous end joining (NHEJ). HR has higher fidelity but requires a homologous sister chromatid for repair thus can only be utilized in the S and G2 phase of the cell cycle[19, 99]. In addition to DSB repair genes BRCA1, Exo1, Mre11, Rad51, Ku70 and DNA PKC we included representative genes critical for the DNA checkpoint response (Chk1), crosslink repair (Fan1), and mismatch repair (Msh2). Gene expression analysis demonstrated that A549 cells contained basal upregulation of all genes examined in the 69

81 CD133+ population. On the other hand in H1299 cells we observed no significant increase in any of the genes analyzed (Figure 3.2), similar to Figure 3.1 in which only A549 CD133+ cells demonstrated IR resistance and an increased capacity for DSB repair. This expression data was confirmed using western blotting (Appendix 5). Contribution of IR exposure to DNA repair induction in CD133+ cells While DNA repair capacity did not appear to be critical for the CD133+ IR sensitivity of H1299 cells, we asked whether prior exposure to IR would result in an activation of CD133+ cells and enhance their capacity for repair. This would help define whether IR in physiological settings contribute to an acquired CD133+ radioresistance. We irradiated A549 and H1299 cells with 4Gy IR and after a 1 week recovery period, exposed them again to 4Gy IR. After another 2 week recovery period these cells were harvested and CD133+ cells separated. IR sensitivity, persistence of ϒ-H2AX, and DNA repair gene expression was measured in CD133+ vs. CD133- cells. In A549 cells we again found that CD133+ cells displayed increased radioresistance over CD133- cells at each dose measured, however we also observed that H1299 CD133+ cells exhibited a significant increase in IR resistance at the 2Gy, 4Gy, and 6Gy doses, with a trend toward increased resistance at 8Gy (Figure 3.3a). To determine whether an increase in DNA repair capacity contributed to this IR resistance we again performed ϒ-H2AX staining on sorted CD133+ and CD133- cells and found that both A549 and H1299 CD133+ cells were able to resolve DSBs by 24 hours post treatment (Figure 3.3b). Gene expression analysis confirmed this data as H1299 cells demonstrated a moderate increase in expression of all genes analyzed (Figure 3.4). The DNA repair expression and IR 70

82 sensitivity changes of CD133+ H1299 cells after IR recovery suggests that prior IR exposure has a repair gene-inducing effect on CD133+ cells and results in a promotion of radiation resistance. Exo1 and Rad51 silencing in A549 cells and effects on CD133+ cells After determining that DNA repair contributes to CD133+ radioresistance in A549 cells, we were interested in studying whether silencing critical repair proteins would confer radiation sensitivity on these cells. For this study we silenced Rad51 and Exonuclease1 (Exo1) due to their observed upregulation in CD133+ cells and the fact that homologous recombination has been proposed to be vital to CSCs in their response to chemotherapy. We sought to determine the subsequent effect on CD133+ IR induced expansion as well as CD133+ IR sensitivity. Rad51 is a critical HR protein that controls strand invasion during homologous mediated repair. Rad51 foci form rapidly after DSB induction and loss of Rad51 is associated with severe HR defects[130, 131]. Additionally Rad51 has been proposed to be an effective therapeutic target in multiple cancer models[132, 133]. Exo1 is another critical HR protein although its precise role in the pathway is still being elucidated. Exo1 has been implicated in the early end resection step of HR and its loss is associated not only with defective HR but also with impaired checkpoint activation[68, 84]. We silenced these genes using lentiviral shrna (Appendix 6) and examined IR expansion and CD133+ IR sensitivity. While A549 cells transduced with a scrambled shrna expanded significantly 72 hours post IR, silencing of both Exo1 and Rad51 resulted in complete loss of the expansion. In addition separated CD133+ cells from 71

83 both A549 shexo1 and shrad51 display increased radiation sensitivity compared to the control cells at each dose tested, while CD133- cells were only more sensitive than control at the 8Gy dose (Figure 3.5). This data suggests that the upregulation of DNA repair observed in A549 cells does contribute directly to the IR resistance, and thus inhibition of appropriate repair proteins may be an effective way to target CD133+ cells. 3.5 Discussion This work provides insight into the role of DNA repair in promoting radiation resistance in CD133+ lung cancer cells. We showed that CD133 selected for a radioresistant population in A549 but not H1299 cells and that this correlated with their altered basal expression of important DNA repair genes as well as enhanced DSB repair capacity. Additionally we showed that CD133+ cells surviving multiple rounds of IR treatment induce genes associated with an increased capacity for DNA repair. In H1299 cells the previously sensitive CD133+ population acquired radiation resistance which was linked to enhanced DNA repair capacity in these cells. Finally we demonstrated that silencing critical DNA repair genes abrogated the expansion of A549 cells after IR treatment and also induced IR sensitivity in CD133+ cells. Our work begins to characterize an IR induced enrichment and expansion of CD133+ cells caused in part by upregulation of DNA repair capacity, and also suggests that inhibition of DNA repair proteins may provide a promising therapeutic direction. Since CSCs appear in part responsible to post treatment rebound and relapse of many solid tumors, and since we 72

84 have observed that CSCs rely on DNA repair, targeting these pathways may impact tumor regrowth. The case for targeted approaches to CSC therapy is becoming stronger. As discussed earlier Bao et. al and others have demonstrated in glioblastomas that CD133+ cells are more radioresistant than CD133- and that Chk1 specific inhibitors can specifically target these cells. Work in other cancer models has suggested that this may also be true for a plethora of diseases. Mihatsch et. al (2011) used ALDH1 as a marker for cancer stem cells in human lung cancer (A549) and breast cancer (SK-BR-3) and found that ALDH1+ cells displayed increased radioresistance and increased DNA repair capacity, and that PI3K inhibition induced significant radiation sensitivity in these cells[64]. Other groups including the previously mentioned Liu et. al demonstrated that CD133+ cells in H460 and H661 lung adenocarcinomas were also resistant to cisplatin and that pre-treatment with ϒ-secretase inhibitors or Notch1 shrna resulted in increased drug sensitivity. Mizugaki et. al (2013) analyzed primary samples of human patients with non-small cell lung cancers and found that high CD133 expression correlated with late pathological stages of the disease and that high expression was correlated with an unfavorable prognosis[134], suggesting that the in vitro data correlates with what is observed in the human disease. Similar to Roppolo et. al (2009) and Dittfield et. al (2009) we show that CD133+ resistance to therapy may be tumor type specific, as A549 but not H1299 CD133+ cells correlated with increased radiation resistance, but our work goes in detail to demonstrate that increased DNA repair 73

85 capacity, and in particular a reliance on specific DNA repair genes may be a major contributing factor for observed radiation resistance. Our work identified Rad51 as a gene of interest contributing to CD133 radiation resistance. Rad51 has been proposed as a suitable therapeutic target in multiple cancer types due to its critical role in the strand invasion step of HR. Ko et. al (2008) silenced Rad51 in human lung cancer lines A549 and H1650 and found that it increased sensitivity to both cisplatin and mitomycin C[133]. Others such as Tsai et. al (2010) also examined Rad51 silencing in human lung cancer cells and found they were able to induce sensitivity of previously drug resistant cells to gemcitabine[135]. Quiao et. al (2005) additionally demonstrate in human non-small cell lung cancer patient samples that high Rad51 expression correlated with a worse patient prognosis. Our data shows that Rad51 silencing in human lung cancer cells induces radiation sensitivity due to the reliance of CD133+ cells on this gene for DSB repair, and thus may be a suitable target for all CSCs demonstrating increased DNA reliance. Exo1 on the other hand is a very interesting potential therapeutic target. Exo1 has been implicated in the end resection step of HR and thus is essential for activation of not only homologous recombination but also cell cycle checkpoints. While it has yet to be targeted for cancer therapy studies we believe that its ability to confer IR sensitivity to CD133+ cells, as well as its possible disruption of the checkpoint activation also makes it a very desirable target for cancer therapy. We identified additional DNA repair factors, notably BRCA1 and DNA-PKcs which were significantly upregulated 74

86 basally in A549 CD133+ cells and could also serve as promising CSC specific therapeutic targets. While their targeting has been proposed in multiple cancer models, their efficacy in inducing radiosensitivity in CSCs has largely yet to be explored [17, 136, 137]. Multiple factors are believed to contribute to the radioresistance observed in cancer stem cells. In addition to the DNA repair aspect that we have discussed in this manuscript, CSCs are believed to rely on a hypoxic niche environment which has been shown to result in therapy resistance and metastatic potential. Groups attempting to disrupt this niche via inhibition of the hypoxia inducible factors have had success in conferring sensitivity in glioblastoma models[138]. A point of interest for our work is that groups have demonstrated that hypoxic conditions can result in decreased expression of homologous recombination proteins including Rad51. Chan et. al (2008) demonstrate this in H1299 cells cultured chronically at 0.2% O 2 [139], and Bindra et. al (2007) use MCF-7 cells cultured at 0.5% 0 2 to show that Rad51 is repressed under hypoxic conditions[140]. How these studies relate to CSCs is not currently well understood. Chan et. al demonstrate that the hypoxic cells with decreased HR capacity were also hypersensitive to MMC and cisplatin, which contradicts what has been shown in other CSC models. Thus the role of hypoxia on HR functionality in CD133+ and other CSC cells needs to be further examined, as it will be interesting to determine the hypoxic response in CD133+ vs. CD133- cells. CSCs are also believed to maintain an environment with very low levels of reactive oxygen species (ROS) via upregulation of ROS scavengers to minimize potential 75

87 genomic damage. Efforts to inhibit scavengers such as glutathione are also underway[141]. Other hypotheses such as activation of signaling pathways and drug efflux pumps have also been proposed[142, 143]. Thus effective CSC specific therapies will likely involve some combination of drugs targeting these multiple mechanisms, but our work begins to demonstrate the importance of DNA repair as a mediator of IR resistance in human lung cancer cells and establishes potential targets to combat this resistance. 3.6 Acknowledgments This work was supported by the Cytometry & Imaging Microscopy Core Facility of the Case Comprehensive Cancer Center (P30CA43703) and the Radiation Resources Core Facility of the Case Comprehensive Cancer Center (P30 CA43703). It was also supported by National Institutes of health grants R01AG and R01CA and molecular therapeutics grant 5T32GM

88 Figure 3.1: The CD133 marker promotes IR resistance in A549 cells but not H1229 cells. (a) CD133+ cells were magnetically separated from CD133- cells and IR sensitivity was measured via clonogenic survival on the two populations. Cells were treated with an IR dose range of 2-8Gy and permitted to grow for days prior to survival analysis. Unpaired t-test performed with two-tailed p value shown, error bars = SEM. (b) Representative images of γ-h2ax immunostaining. 50 cells per treatment were analyzed for foci formation and the % positive was plotted. 77

89 Figure 3.1: The CD133 marker promotes IR resistance in A549 cells but not H1229 cells. 78

90 Figure 3.2: CD133+ cells display basally upregulated DNA repair genes in A549 cells only RNA was isolated from magnetically separated CD133- and CD133+ cells in A549 (A) and H1229 (B) cell lines and real time PCR analysis was performed examining multiple DNA repair genes. Error bars= SEM. 79

91 Figure 3.2: CD133+ cells display basally upregulated DNA repair genes in A549 cells only 80

92 Figure 3.3: CD133 contributes to radioresistance in cells surviving multiple IR doses in both A549 and H1229 cells via upregulation of DNA repair genes (a) A549 and H1229 cells were treated as diagrammed. Sorted cells were treated with an IR dose range and RNA was extracted for real time PCR analysis. Error bars = SEM. (b) Representative images of γ-h2ax immunostaining. 25 cells per treatment were analyzed. 81

93 Figure 3.3: CD133 contributes to radioresistance in cells surviving multiple IR doses in both A549 and H1229 cells via upregulation of DNA repair genes 82

94 Figure 3.4: CD133+ cells display upregulated DNA repair genes in both A549 and H1299 following IR exposure RNA was isolated from magnetically separated CD133- and CD133+ cells in A549 (A) and H1229 (B) cell lines and real time PCR analysis was performed examining multiple DNA repair genes. Error bars= SEM. 83

95 Figure 3.4: CD133+ cells display upregulated DNA repair genes in both A549 and H1299 following IR exposure 84

96 Figure 3.5: Rad51 and Exo1 silencing in A549 cells abrogates the IR expansion phenotype and induces CD133 IR sensitivity (A) Exo1 and Rad51 were silenced via lentiviral transduction in A549 cells. Mixed populations were irradiated with 4Gy IR and the percent CD133+ cells 72 hours post IR were measured. (B) CD133+ cells were magnetically separated from CD133- cells and IR sensitivity was measured via clonogenic survival. Cells were treated with either 4Gy or 8Gy and permitted to grow for days prior to survival analysis. 85

97 Figure 3.5: Rad51 and Exo1 silencing in A549 cells abrogates the IR expansion phenotype and induces CD133 IR sensitivity 86

98 Chapter 4 3 Exo1 Independent DNA Mismatch Repair Involves Multiple Compensatory Nucleases 4.1 Abstract Functional DNA Mismatch repair (MMR) is essential for maintaining the fidelity of DNA replication and genetic stability. In hematopoiesis, loss of MMR results in methylating agent resistance and a hematopoietic stem cell (HSC) repopulation defect. Additionally MMR failure is associated with a variety of human malignancies, notably Lynch syndrome. We focus on the 5 3 exonuclease Exo1, the primary enzyme excising the nicked strand during MMR, preceding polymerase synthesis. We found that nuclease dead Exo1 mutant cells are sensitive to the O6-methylguanine alkylating agent temozolomide when given with the MGMT inactivator, O6benzylguanine (BG). Additionally we used an MMR reporter plasmid to verify that Exo1 mut MEFs were able to repair G:T base mismatches in vitro. We showed that unlike other MMR deficient mouse models, Exo1 mut mouse HSC did not gain a competitive survival advantage post temozolomide/bg treatment in vivo. To determine potential nucleases implicated in MMR in the absence of Exo1 nuclease activity, but in the presence of the inactive protein, we performed gene expression analyses of several mammalian nucleases in WT and Exo1 mut MEFs before and after temozolomide treatment and identified upregulation of Artemis, Fan1, and Mre11. Partial shrna mediated silencing of each of these in Exo1 mut cells resulted in decreased MMR capacity and increased resistance to temozolomide/bg. We propose that nuclease function is required for fully functional 3 A version of this chapter was submitted to the journal DNA Repair 87

99 MMR, but a portfolio of nucleases is able to compensate for loss of Exo1 nuclease activity to maintain proficiency. 4.2 Introduction The DNA Mismatch Repair pathway (MMR) is a fundamental process in cells that functions to repair mispaired bases and insertion/deletion loops caused by errors in replication or recombination[76, 144]. The pathway has been elucidated largely through studies in Escherichia coli and Saccharomyces cerevisiae although many of the required enzymes are conserved in higher organisms[145]. In eukaryotic cells the pathway is similar to that found in prokaryotes but the nature of the enzymes involved can vary based on the nature of the mismatch[11]. Several studies have shown that failure in MMR can lead to the accumulation of mutations and carcinogenesis, notably Lynch syndrome [10, 12, 77]. As is currently understood, MMR is initiated when a mispaired base or insertion/deletion loop is recognized and bound by either heterodimer of MSH2-MSH6 (MutSα) or MSH2-MSH3 (MutSβ). This binding triggers an ATP dependent reaction which recruits an additional heterodimer consisting of either MLH1-PMS2 (MutLα) or MLH1-PMS1 (MutLβ). The Mutlα complex has been characterized as containing latent endonuclease activity which serves as a discrimination signal for the nascent strand[146]. This results in the recruitment of a variety of other factors including the proliferating cell nuclear antigen (PCNA) which is thought to interact with the MutSα complex to enhance repair, Exonuclease 1 (Exo1) which functions to excise the mispaired base to allow for subsequent resynthesis, and replication protein A (RPA) 88

100 which functions to stabilize the single stranded intermediate formed by Exo1 activity. DNA polymerase δ has been implicated in the resynthesis of the mismatched strand with DNA ligase functioning to reseal the break[74, ]. Exo1 is a 5 3 exonuclease member of the Rad2 family of nucleases. It has been implicated in a variety of cellular pathways including homologous recombination, meiotic crossing over, telomere maintenance, and mismatch repair[66, 70, 73, 150, 151]. It is currently the only nuclease identified in MMR where the enzyme is recruited to the complex by the Mutlα discrimination signal. While it was previously unclear how the 5 ->3 exonuclease could excise a 3 heteroduplex, the Mutlα endonuclease activity is proposed to be directed to the distal side of the mismatch, thus resulting in incision 5 to the mispair and allowing for Exo1 recruitment [152]. At that point it excises the mismatched strand of DNA for as long as 1,000 nucleotides, thus generating a single stranded DNA gap which is bound by RPA to provide stability. Recently Liberti et. al (2013) have described a more precise role in which Exo1 processes replication errors by demonstrating it is specific to lagging strand errors, where it is recruited to the DNA by the 5 ends of Okazaki fragments[75]. While MMR mutations in Mlh1, PMS2 and Msh2 have been implicated in the pathogenesis of Lynch syndrome (LS) [153, 154], suggesting that a dysfunctional MMR contributes to cancer pathogenesis, mutations in Exo1 are not conclusively linked to patients with LS[78, 79]. Additionally studies in yeast have shown that the mutator phenotype in Exo1 deficient S. pombe is less significant than that of Msh2 deficient strains. Exo1 mut ES cells display an increase in mutation rates at the Hprt locus, although they were 5-fold lower than those observed in MSH2 -/- cells. 89

101 Microsatellite instability (MSI) studies in Exo1 mut genomic DNA demonstrated increased MSI at mononucleotide markers but levels comparable to WT at dinucleotide markers[69]. These data suggest that loss of Exo1 function does not completely ablate MMR as occurs with loss of other MMR proteins, and that Exo1 nuclease independent MMR exists which at least partially maintains the efficiency of the pathway. This complementation of nucleases has been characterized in E. coli, where at least 4 exonucleases contribute to MMR[155]. In this manuscript we examined the proficiency of the mismatch repair pathway in mice containing an inactivating mutation in the nuclease domain of Exo1. Of note, unlike the Exo1 null mouse recently described by Schaetzlein et. al (2013), this nucleasedead protein is stable and might contribute to the DNA repair processes, perhaps as a scaffold protein. We show that unlike MMR deficient models, cells derived from these mice were sensitive to the alkylating agent temozolomide and the MGMT inhibitor BG, and that MEFs displayed DNA repair capacity similar to WT mice. Additionally we show that unlike the MMR deficient Msh2 -/- mice Exo1 mut mice did not gain a hematopoietic competitive survival advantage in vivo post temozolomide/bg treatment[36]. Using a heteroduplex egfp plasmid containing a G:T mismatch we show that the Exo1 mutant MEFs were able to repair G:T mismatches at a level comparable to WT MEFs, indicating a proficient mismatch repair pathway. We used gene expression studies after temozolomide/bg treatment in WT and Exo1 mut MEFs and identified 3 upregulated nucleases: Artemis, Fan1, and Mre11. shrna knockdown of these nucleases resulted in impaired repair of G:T mismatches and an increased resistance to temozolomide/bg, 90

102 suggesting that complementary nucleases are able to partially maintain MMR capacity in the absence of Exo1. We have identified a potential compensatory mechanism cells utilize to ensure replicative fidelity and mutation avoidance in the absence of a functional Exo Materials and Methods Animals Exo1 mut mice used in these studies were donated by Dr. Winfried Edelmann from the Albert Einstein College of Medicine. Their generation was described by Wei et al (2003) [69]. Mice were used along with their WT littermates throughout. All mouse studies were approved by the institutional animal care and use committee at Case Western Reserve University. Temozolomide Survival Assay: Temozolomide (Ochem Inc) was prepared by dissolving in DMSO and diluting with serum free DMEM. Final DMSO levels were always < 2%. Cells were plated at 5,000 cells per well in 6-well tissue culture plates and treated with 10uM 06-benzylguanine for 1 hour to inactivate MGMT[156]. Cells were treated with a temozolomide dose range for 3 hours after which media was replaced. 10 um 06-benzyguanine was added every 24 hours for 3 days after which MTT assay was performed to assess cell viability[157]. All experiments were performed in triplicate with identical controls for each replicate and student s t-tests were performed at each treatment dose for all cell types. 91

103 Heteroduplex egfp Plasmid: The heteroduplex egfp plasmid was donated by Dr. Luzhe Sun from the University of Texas Health Science Center at San Antonio. The plasmid was prepared as outlined[158]. Cells were transfected using Lipofectamine 2000 (Invitrogen) with 1 ug of the reporter plasmid, and 24 hours post transfection the levels of GFP in cells were measured using cytometric analysis on a BD LSRII instrument[159]. All experiments were performed in triplicate with identical controls for each replicate and student s t-tests were used to measure statistical significance. Competitive Repopulation Assay: 10 Boy J mice were lethally irradiated with 1100 Rad Cs mice were injected with 1:1 ratio of C57/B6:Boy J whole bone marrow and 5 mice were injected with 1:1 ratio of Exo1 mut :BoyJ whole bone marrow. After 4 weeks the mice were treated with 40mg/kg 06-benzylguanine followed 1 hour later by 80 mg/kg temozolomide for 3 consecutive days. 8 weeks post temozolomide treatment cytometric analysis was performed on peripheral eye blood of transplanted recipients to measure percent chimerism via the CD45.1 and CD45.2 cell surface markers. Student s t-tests were used to measure statistical significance. Gene Expression Studies: WT and Exo1 mut MEFs were treated with 250ug/mL temozolomide. 24 hours post treatment RNA was extracted using the Trizol method and cdna synthesized (Superscript III First Strand Kit- Invitrogen). Gene expression was measured using validated primers (Applied Biosystems) and quantitative real-time PCR. 92

104 Lentiviral Gene Silencing: Artemis, Fan1, and Mre11 were silenced via shrna transduction with validated clones (Sigma-Aldrich). Artemis clone IDs (NM_ s1c1 and NM_ s1c1). Fan1 clone IDs (XM_ s21c1 and XM_ s21c1). Mre11 clone IDs (NM_ s1c1 and NM_ s1c1). Lentiviral particles were synthesized via HEK293 cells and target cells were infected, selected for with puromycin, and clones were assessed for verification of gene silencing. 4.4 Results Exo1 mut MEFs demonstrate temozolomide sensitivity and repair G/T mismatches in vitro To determine whether Exo1 mut MEFs would display the same methylating agent resistance observed in the MMR deficient Msh2 -/- models[36, 160] we treated WT, Exo1 mut and Msh2 -/- MEFs with a temozolomide dose range and measured cell viability via MTT assay three days post treatment. MGMT was inactivated by 06-benzylguanine (BG). Temozolomide is an alkylating agent that forms O6-methylguanine which base pairs with thymidine (T) invoking G:T mismatch recognition by MMR and induces cytotoxicity in cells via a cycle of futile repair[161, 162]. We found that WT and Exo1 mut MEFs were sensitive to the drug in a dose dependent fashion. Exo1 mut MEFs were slightly more resistant to temozolomide/bg than WT at each dose, suggesting an MMR defect, but the difference was statistically significant (p<0.05) only at the 50ug/mL dose. 93

105 Msh2 -/- MEFs demonstrated a complete dose independent resistance relative to WT (p<0.05), consistent with loss of MMR function (Figure 4.1a). We used a heteroduplex egfp plasmid described by Zhou et. al[158] to show that Exo1 mut MEFs are capable of repairing base mismatches that would be processed by functional MMR. The plasmid contains a G:T mismatch and a nick on the template strand (to serve as a strand discrimination signal) and repair of the G:T to the proper G:C pairing eliminates a premature stop codon in the egfp gene. Those cells able to convert the G:T mispair display GFP fluorescence. Upon transfecting WT, Exo1 mut and Msh2 -/- MEFs with the reporter plasmid and measuring GFP fluorescence via flow cytometry we found that Exo1 mut MEFs demonstrated a modestly decreased MMR capacity compared to WT MEFs that was not quite statistically significant (p= ). The Exo1 mut MMR capacity was significantly greater than the Msh2 -/- MEFs, which were severely defective in their MMR response compared to WT (p< 0.05, Figure 4.1b). This data suggests that Exo1 mut cells are able to process and repair mismatches via a functional, albeit somewhat impaired MMR. Exo1 mut bone marrow does not exhibit a competitive repopulation defect nor does it gain a competitive survival advantage post temozolomide treatment: We performed a hematopoietic competitive repopulation assay to measure hematopoietic stem cell (HSC) proficiency of Exo1 mut bone marrow. Competitive repopulation measures HSC proficiency in engrafting into the HSC niche and performing long term multi-lineage reconstitution. Previous studies have shown that HSC from 94

106 MMR deficient Msh2 -/- mice display defects in competitive repopulation and additionally gain a competitive survival advantage over WT marrow when the chimeric mice are treated with temozolomide[36]. We used cytometric analysis to show that at both 8 and 16 weeks post marrow transplantation, the Exo1 mut marrow remained an approximately 1:1 ratio with WT marrow suggesting proficient hematopoietic function. We treated 1:1 chimeric mice with 40mg/kg BG to inactivate MGMT followed by 80mg/kg temozolomide (x3), measured the percent chimerism 8 weeks post treatment and found that the ratio remained approximately 1:1 (Figure 4.2). This data demonstrates in an in vivo setting that HSC in Exo1 mut mice are sensitive to methylating agents due to proficiency in the MMR pathway. In contrast, we previously observed a profound selection advantage in favor of MSH2 -/- HSC after temozolomide in similar hematopoietic competitive repopulation studies[36]. Gene expression changes of multiple nucleases following temozolomide treatment: To elucidate the mechanism of MMR in the absence of Exo1, we performed gene expression analysis of multiple nucleases in WT and Exo1 mut MEFs. We examined five mammalian nucleases also containing 5 3 enzymatic activity (Artemis, Fan1, Fen1, Mre11, XPF) and studied changes in expression after temozolomide/bg treatment. In WT MEFS we found that temozolomide induced strong expression of Artemis, Fan1, and Mre11 24 hours after treatment. XPF and Fen1 levels were unchanged as was expression of Exo1, suggesting that basal levels of Exo1 are sufficient for MMR activity in normal settings (Figure 4.3a). Transcript induction was followed by increases in protein 95

107 concentrations detected by western blot (Appendix 7). The upregulation of the three nucleases in WT MEFs suggested that these enzymes may complement each other in normal DNA damage response pathways, including mismatch repair. In contrast to the pattern observed in WT MEFs, Exo1 mut MEFs demonstrated increased basal expression of Artemis, Fan1, and Mre11 but no induction of any nucleases was observed after temozolomide treatment (Figure 4.3b and 4.3c). While the mechanism of post DNA damage induction is not yet elucidated, one interpretation is that levels of WT Exo1 are sufficient for MMR whereas the other three enzymes are upregulated in response to damage, and constitutively upregulated in the absence of Exo1 nuclease activity in Exo1 mut MEFs, perhaps in a compensatory manner. Partial silencing of Artemis, Fan1, and Mre11 results in increased temozolomide resistance and decreased MMR proficiency in Exo1 mut MEFs: After identifying Artemis, Fan1, and Mre11 as potentially being involved in MMR proficiency after loss of Exo1 nuclease activity, we partially silenced these genes via lentiviral shrna knockdown in WT and Exo1 mut MEFs to determine whether the loss of each nuclease singly and collectively would result in an impaired MMR phenotype. We additionally included a scrambled shrna sequence in WT and Exo1 mut MEFs to demonstrate that the transduction had no effect on temozolomide sensitivity (Appendix 8). As shown in Figure 4.4a only moderate levels of gene silencing were achieved in surviving cells. A second shrna vector to each gene was also used (Appendix 9). In addition it should be noted that the level of Artemis, Fan1, and Mre11 nuclease 96

108 knockdown in Exo1 mut MEFs, which, as noted, displayed a higher basal level of each transcript than WT MEFs (Figure 4.3c), results in remaining transcript levels that are similar to those of WT MEFs. We treated each cell type (WT, WT+shArtemis, WT+shFan1, WT+shMre11, Exo1 mut, Exo1 mut +shartemis, Exo1 mut +shfan1, Exo1 mut +shmre11) with BG and a temozolomide dose range and measured cell survival after three days via MTT assay. We showed that partial knockdown of each nuclease in both WT and Exo1 mut MEFs resulted only in a modest increase in temozolomide resistance. For shartemis, only the 50ug/mL dose in Exo1 mut shartemis cells was significantly different from WT MEFs (p<0.05). WTshFan1 cells demonstrated a significant difference at the 50ug/mL dose, while Exo1 mut shfan1 cells were significant at both the 50ug/mL and 125ug/mL doses (p<0.05). This was the same trend observed in shmre11 cells. When directly comparing the WT silenced vs. Exo1 mut silenced cells we found that the Exo1 mut curves were significantly more resistant than WT for shartemis and shfan1 (p<0.05). We used the MMR reporter assay to determine whether the increased temozolomide resistance correlated with a decreased MMR capacity and found that only Exo1 mut shmre11 cells were statistically reduced compared to WT (p<0.05), while Exo1 mut shfan1 demonstrated a not quite significant reduction in GFP levels compared to WT (p=0.0572, Figure 4.4). Interestingly, partial knockdown of any single nuclease in WT MEFs did not result in statistically significant differences in MMR capacity. These results suggested that loss of a single nuclease, even in combination with Exo1 mut, was not enough to yield a defective MMR phenotype. To elucidate whether 97

109 multiple nucleases were responsible for Exo1 nuclease independent MMR, we silenced a combination of each nuclease (Artemis/Fan1, Artemis/Mre11, Fan1/Mre11) in both WT MEFs- which contain functional Exo1, and Exo1 mut MEFs. This would help clarify whether functional Exo1 would be sufficient to maintain proficient MMR in the absence of the additional nucleases. It would also elucidate the necessity of these nucleases in maintaining MMR in the absence of Exo1. Similar to what we observed with single gene silencing, only modest levels of silencing of nuclease pairs was achieved (Figure 4.5a). However, even with the moderate silencing, cells with loss of these nuclease pairs demonstrated proliferation defects and spontaneous apoptosis. This suggests that these genes play important roles in multiple cellular pathways and are important for cell survival. We examined temozolomide sensitivity of each of the double knockdown cells. WTshArt/Fan cells were more resistant than WT MEFs at 50ug/mL, while Exo1shArt/Fan cells were significantly different at both 50ug/mL and 125 ug/ml (p<0.05). WTshArt/Mre cells were also only more resistant than WT MEFs at the 50ug/mL dose, while Exo1 mut shart/mre cells were more resistant at all three drug doses (p<0.05). Finally the WTshFan/Mre MEFs were significantly different from WT MEFs at both the 50ug/mL and 125ug/mL doses while the Exo1 mut shfan/mre cells were resistant at all three doses (p<0.05). All three sets of Exo1 mut dual knockdown cells were more temozolomide resistant than their WT counterparts, suggesting a higher degree of loss of MMR as indicated by tolerance to DNA methylation from temozolomide (p<0.05). The MMR capacity in these combination knockdown cells was measured using the heteroduplex plasmid. Compared to WT MEFs, partial combination silencing in WT 98

110 MEFs yielded reduced MMR capacity only in the combination of WTshFan/Mre11, while WTshArt/Fan (p=0.051) and WTshArt/Mre11 (p=0.11) cells were not statistically reduced. Knockdown in Exo1 mut MEFs however displayed reduced MMR capacity with each combination (p<0.05). While neither complete temozolomide resistance nor loss of MMR proficiency was observed in the Exo1 mut nuclease KD combination silenced cells, the data strongly suggests that significant MMR deficiency develops under conditions of loss of Exo1 nuclease activity combined with loss of the other nucleases. While these nucleases appear complementary, the data from WT MEFs suggests that Exo1 alone is sufficient to retain a functional level of MMR capacity. Finally to determine whether complete MMR deficiency would be observed after loss of all three identified nucleases, we attempted to silence the combination of Artemis/Fan1/Mre11 in both WT and Exo1 mut MEFs. As with previous experiments the level of gene silencing was modest in both cell types and was accompanied by significant proliferation defects and spontaneous apoptosis. Surviving cells were assessed for gene knockdown level and treated with the temozolomide dose range and MMR reporter plasmid. Temozolomide resistance data demonstrated that both WTshArtemis/Fan1/Mre11 cells and Exo1 mut shartemis/fan1/mre11 cells were significantly more resistant than WT MEFs at all doses (p<0.05), but that the Exo1 mut triple knockdown cells were also more resistant than WT triple knockdown cells. The MMR reporter assay was performed on these cells and both the WT and Exo1 mut knockdown cells demonstrated significantly reduced activity compared to WT 99

111 MEFs (p<0.05), and were comparable to the GFP levels of Msh2 -/- cells. (Figure 4.6b). Given this level of MMR dysfunction compared to WT, the difference between the WT and Exo1 mut triple knockdown cells was not quite statistically significant (p=0.08). These data demonstrated that the combination loss of these three nucleases resulted in increased temozolomide resistance and significant loss of MMR capacity, closely resembling the phenotype of Msh2 -/- cells. Thus, these complementary nucleases contribute to Exo1 nuclease independent MMR function. 4.5 Discussion Our work demonstrates that while Exo1 nuclease activity is important for normal MMR function, its loss can be compensated for by multiple complementary nucleases. While these nucleases are induced in response to temozolomide in WT cells, perhaps implicating them in the normal repair of methylating agent induced DNA damage, they are basally upregulated in Exo1 mut MEFs which suggests that they serve as primary nucleases in the absence of Exo1 to maintain partial MMR function. This differs from MMR proteins Mlh1 and Msh2 for which there are no redundant proteins. We showed that Exo1 mut MEFs were relatively sensitive to the O6-methylguanine alkylating agent temozolomide, that they were able to repair G:T mismatches in vitro, that HSC from Exo1 mut mice did not demonstrate a hematopoietic competitive advantage in vivo after temozolomide treatment, that Exo1 mut MEFs displayed altered expression of nucleases potentially involved in MMR, and that partial silencing of these nucleases in Exo1 mut MEFs resulted in increased temozolomide resistance and decreased MMR function. 100

112 The nucleases Artemis, Fan1, and Mre11, which we identified as being upregulated after temozolomide treatment in Exo1 mut MEFs are the likely lead candidates in this proposed pathway. Mre11 for instance has been shown to be frequently mutated in MMR deficient cancers[163], its physical interactions with Mlh1 have been characterized[164], and its loss has been shown to result in increases in microsatellite instability (a marker for loss of MMR function) and impaired MMR proficiency[165]. Thus our work and these previous findings strongly suggest that Mre11 functions in MMR. Additionally the Fanconi Anemia enzyme Fan1, which also contains 5 3 exonuclease activity and endonuclease activity, was identified by a genetic screen through its interactions with MMR proteins Mlh1 and PMS2[166]. Fan1, while possibly not critical to MMR in normal settings, may become actively involved in MMR after loss of Exo1. Finally the nuclease Artemis, which has been implicated in the DNA double strand break repair pathway non-homologous end joining, also has single strand specific 5 3 exonuclease activity although it has not been found to interact with critical MMR proteins. However, Katsube et. al (2011) have described Artemis deficient MEFs to demonstrate increased resistance to MMS and propose that it may play a role in multiple DNA repair pathways[167]. The three candidate genes however all demonstrate similar 5 3 exonuclease function as Exo1, with Mre11 also containing 3 5 activity, and will require further exploration to determine their precise role in MMR. An additional factor to consider is that while our studies demonstrated even modest knockdown of these enzymes resulted in changes in both temozolomide 101

113 sensitivity and MMR activity, the effects could be more striking when working with models where these enzymes are completely inactivated. Of interest is the fact that basal Exo1 mut MEFs contained significantly higher transcript levels of all three genes than WT MEFs, thus the partial knockdown achieved resulted in remaining Exo1 mut levels that were similar to WT MEFs. This observation suggests that the upregulation of these genes is essential for Exo1 independent MMR. Partial knockdown assumes that functional protein is still present for repair, and thus it would be assumed that complete gene inactivation could produce a phenotype that mimics complete MMR failure. Because these genes were upregulated basally in the Exo1 mut MEFs, the enzyme present after partial knockdown may still have been sufficient for the partial MMR functionality that we observed. Additionally the spontaneous apoptosis and proliferation defects we observed after combination silencing of these genes confirm that the nucleases play crucial roles in DNA repair processes beyond MMR, and it will be interesting to study how the loss of those additional pathways (such as potential defects in double strand break repair) would affect MMR capacity. Previous works have demonstrated that loss of Exo1 function has a more modest phenotype in terms of cancer predisposition and mutability when compared to loss of critical MMR proteins Msh2 and Mlh1. However unpublished work from our group and recently Schaetzlein et. al (2013) have demonstrated its activity is crucial for DNA double strand break (DSB) processing[80]. Our studies have shown that loss of Exo1 nuclease function results in increased sensitivity to DSB inducing agents due to impaired end processing. Thus its enzymatic function is critical to DNA end resection of 102

114 double strand breaks, while in MMR its exonuclease activity appears somewhat dispensable due to compensatory nucleases. However a mechanism for MMR in the absence of Exo1 has yet to be fully elucidated. Kadyrov et. al (2009) described a possible mechanism for Exo1 independent MMR by utilizing a purified system containing MutSα, MutLα, replication factor C, PCNA, RPA, and DNA polymerase δ and observing MMR function in vitro dependent on the endonuclease activity of MutLα and strand displacement via polymerase δ[81]. While our work suggests that additional nucleases may also be involved, the fact that total MMR function is not completely ablated when silencing these enzymes suggests that additional mechanisms such as the strand displacement may also be occurring. Additionally multiple groups have recently published that the structural function of Exo1 protein may be completely distinct from its catalytic one, very notably in MMR. Izumchenko et. al (2011) showed in vitro and most recently Schaetzlein et. al (2013) showed in vivo that the catalytic nuclease function of Exo1 may not be necessary for its role in MMR[80, 82]. In fact it has been proposed that the structural component of Exo1 may serve as a docking site or play a scaffolding role in the formation of protein complexes including additional nucleases in MMR. Our work with the Exo1 mut mouse, which contains a slightly truncated Exo1 protein, may corroborate these studies which would explain the modest phenotype observed after loss of Exo1 function, because the structural component could still participate to dock the additional nucleases. 103

115 Linking four nucleases to MMR suggests that cells prioritize genomic stability. We might next ask how this compensatory pathway affects the response to other types of DNA damage which requires these identified enzymes. In human disease this story raises some interesting questions such as whether mutations in Artemis, Fan1, or Mre11 are found in Lynch syndrome, spontaneous malignancies with loss of MMR from promoter methylation, or other diseases associated with dysfunctional MMR. The complementary nature of these nucleases also provides data linking seemingly independent DNA repair pathways. It appears cells may use multiple compensatory pathways to maintain MMR function and avoid genomic instability. This work further characterizes the multifaceted nature of mismatch repair and the implications that its loss has on genomic stability and other DNA repair pathways. 4.6 Acknowledgements This work was supported by the Case Comprehensive Cancer Center (2P30 CA (Gerson, PI)). It was also supported by National Institutes of health grant 5R42 CA (Gerson, PI) and molecular therapeutics grant 5T32GM We thank Dr. Luzhe Sun of the UT Health Science Center San Antonio for donating the heteroduplex egfp plasmid. We also thank Mojibade Hassan of Washington University in St. Louis for her assistance with experiments. 104

116 Figure 4.1: Exo1 mut MEFs demonstrate temozolomide sensitivity and repair G/T mismatches in vitro A) WT, Exo1 mut, and Msh2-/- MEFs were treated with a temozolomide dose range and survival was monitored three days post treatment via MTT assay. Error bars indicate SEM from 3 independent experiments. B)The heteroduplex egfp plasmid was used on WT, Exo1 mut, and Msh2-/- MEFs and GFP fluorescence was measured 24 hours post transfection via cytometric analysis. Error bars indicate SEM from 3 independent experiments. 105

117 Figure 4.1: Exo1 mut MEFs demonstrate temozolomide sensitivity and repair G/T mismatches in vitro 106

118 Figure 4.2: Exo1 mut mice do not gain a hematopoietic competitive advantage after temozolomide treatment in vivo WT and Exo1 mut whole bone marrow was mixed 1:1 and transplanted into lethally irradiated WT recipients. 4 weeks post transplants chimeric mice were treated with 40mg/kg BG followed by 80 mg/kg temozolomide (x3 days) and 8 weeks post treatment chimerism was measured via cytometric analysis of peripheral eye blood. 107

119 Figure 4.2: Exo1 mut mice do not gain a hematopoietic competitive advantage after temozolomide treatment in vivo 108

120 Figure 4.3: Exo1 mut MEFs demonstrate upregulated gene expression of multiple nucleases after temozolomide treatment WT and Exo1 mut MEFs were treated with 250 ug/ml temozolomide and 24 hours later RNA was extracted from cells and cdna synthesized for real-time PCR analysis. Data shown is the fold change of expression of each nuclease compared to WT or Exo1 mut untreated MEFs. 109

121 Figure 4.3: Exo1 mut MEFs demonstrate upregulated gene expression of multiple nucleases after temozolomide treatment 110

122 Figure 4.4: shrna mediated silencing of Artemis, Fan1, and Mre11 in Exo1 mut MEFs results in mild temozolomide resistance and decreased MMR capacity A) Artemis, Fan1 and Mre11 were silenced via lentiviral shrna in WT and Exo1 mut MEFs. Silenced cells were subsequently treated with a temozolomide dose range and survival measured three days post treatment via MTT assay. Error bars indicate SEM from 3 independent experiments. B) MMR reporter heteroduplex egfp plasmid was used on each silenced cell type and GFP fluorescence was measured 24 hours post transfection via cytometric analysis. Error bars indicate SEM from 3 independent experiments. 111

123 Figure 4.4: shrna mediated silencing of Artemis, Fan1, and Mre11 in Exo1 mut MEFs results in mild temozolomide resistance and decreased MMR capacity 112

124 Figure 4.5: Combination silencing of Artemis/Fan1, Artemis/Mre11 and Fan1/Mre11 in Exo1 mut MEFs demonstrate increased temozolomide resistance and decreased MMR capacity A) Artemis, Fan1 and Mre11 were silenced in combination via lentiviral shrna in WT MEFs. Silenced cells were subsequently treated with a temozolomide dose range and survival measured three days post treatment via MTT assay. Error bars indicate SEM from 3 independent experiments. B) MMR reporter heteroduplex egfp plasmid was used on each silenced cell type and GFP fluorescence was measured 24 hours post transfection via cytometric analysis. Error bars indicate SEM from 3 independent experiments. 113

125 Figure 4.5: Combination silencing of Artemis/Fan1, Artemis/Mre11 and Fan1/Mre11 in Exo1 mut MEFs demonstrate increased temozolomide resistance and decreased MMR capacity 114

126 Figure 4.6: Triple silencing of Artemis/Fan1/Mre11 demonstrates significant MMR loss in both WT and Exo1 MEFs A) Artemis, Fan1 and Mre11 were silenced via lentiviral shrna in WT and Exo1 MEFs. Silenced cells were subsequently treated with a temozolomide dose range and survival measured three days post treatment via MTT assay. Error bars indicate SEM from 3 independent experiments. B) MMR reporter heteroduplex egfp plasmid was used on each silenced cell type and GFP fluorescence was measured 24 hours post transfection via cytometric analysis. Error bars indicate SEM from 3 independent experiments. 115

127 Figure 4.6: Triple silencing of Artemis/Fan1/Mre11 demonstrates significant MMR loss in both WT and Exo1 MEFs. 116

128 Chapter 5: Discussion and Future Directions Together our studies described herein have paved the way for a clear path to further understand the role of DNA repair in normal and malignant stem cell populations. Through three distinct studies we have shown that Exo1 is an important mediator of DNA double strand break repair in cycling hematopoietic stem cells, is a potential therapeutic target to combat radiation resistance in cancer stem cells, and may serve as part of a portfolio of exonucleases responsible for proficient MMR. Thus Exo1 is a promising target to not only model the effects of loss of homologous recombination on DNA repair capacity, but also its role in MMR may lead to the identification of a novel compensatory pathway that cells adopt to maintain genomic stability. While the methods described in these chapters will be useful for studying additional properties of stem cells, such as whether cell cycle activation induces other alterations in DNA repair reliance, whether cancer stem cells in additional tumor types are activated in response to prior therapy, and whether additional compensatory pathways exist after loss of DNA repair genes, these studies also provide the framework to examine potential impacts on human disease. These questions and the future directions outlined below point to a critical need to further understand the complex relationship between DNA repair, cell cycle, and stem cell populations. 5.1 The Relationship between DNA Repair and Cell Cycle in HSCs Hematopoietic stem cell function is required for lifelong blood reconstitution and immune maintenance, thus understanding mediators of HSC function is critical for both normal and diseased states. In Chapter 2 we discuss our findings in the Exo1 mut 117

129 mouse on the effects of loss of Exo1 mediated HR on HSC function both at steady state and in active cell cycle. We show that steady state HSCs maintain complete function in the Exo1 mut mouse, as demonstrated by normal SKL levels, normal lineage committed cell development, and normal competitive repopulation, serial repopulation, and niche occupancy, suggesting that intact NHEJ is sufficient to maintain normal hematopoiesis at steady state. However after HSC cell cycle entry using 5-FU and poly-ic NHEJ could not compensate for loss of HR in activated cells as mice demonstrated hematopoietic defects and died due to hematological failure. While NHEJ alone was not sufficient for animal survival in cycling Exo1 mut HSCs after 4Gy IR, we evaluated whether these mice could tolerate lower doses of IR after 5-FU treatment. Preliminary studies suggested that at lower doses such as 1Gy the loss of Exo1 mediated HR could be compensated for by NHEJ as measured by animal survival. Thus there may be a damage threshold in cycling HSCs after which both HR and NHEJ are required. We have illustrated a proposed model for the Exo1 mut HSC response to 5-FU + IR in Figure

130 Figure 5.1 Proposed model for Exo1 mut mouse HSC Response to 5-FU + IR We demonstrated in Chapter 2 that Exo1 mut mice demonstrated hematopoietic failure and animal death after combined treatment of 5-FU + IR. We model the proposed HSC effects of this treatment in the Exo1 mut mouse. 5-FU treatment has been demonstrated to kill cycling HSCs and push the quiescent population into active cell cycle, where our work suggests they transition from NHEJ reliance exclusively to combined reliance on NHEJ and HR. Subsequent IR treatment in the HR impaired Exo1 mut HSCs results in HSC death followed by animal death. 119

131 Figure Proposed model for Exo1 mut mouse HSC Response to 5-FU + IR 120

132 5.1.2 Proposed role of HR and HSC Cell Cycle status in Fanconi Anemia In Chapter 2.5 we introduce the concept that these findings may play an important role in such hematological diseases as Fanconi anemia (FA), in which mutations in critical FA proteins have been found to result in loss of HSC quiescence and a potential defect in HR function. Fanconi anemia is a rare autosomal recessive disease characterized by progressive bone marrow failure and increased cancer predisposition. FA patients have an average lifespan of years with bone marrow failure, leukemia, or solid tumors responsible for the most common causes of death. The disease manifests due to germline mutations in genes responsible for the repair of a specific form of DNA damage termed interstrand crosslinks (ICLs) [168]. ICLs covalently link two DNA strands and are cytotoxic because they block essential DNA metabolic processes. They are caused endogenously by events such as lipid peroxidation or exogenously by agents including formaldehyde. Generation of ICLs are exploited in cancer therapy by agents such as mitomycin C (MMC) but in normal settings ICL repair is essential for maintenance of genomic stability, and thus preservation of the Fanconi anemia pathway is essential for human health[169]. The FA pathway functions to repair ICLs through coordination of at least 15 genes. The FANCM subunit initiates the pathway by forming a heterdimer with FAAP24 to recognize the ICL, stabilize the stalled replication fork, activate ATR mediated checkpoint activation, and recruit the FA core complex. The FA core complex consists of eight FA proteins (FANCA, B, C, E, F, G, L, M) and functions as a multisubunit ubiquitin E3 121

133 ligase complex. This complex monoubiquitinates FANCD2 and FANCI, resulting in the formation of the FANCD2/FANCI heterodimer which relocalizes to the ICL. At the site of the ICL FANCD2 serves as a dock for recruiting multiple nucleases including FAN1 which identify specifically the ubiquitin moiety on FANCD2. These nucleases convert the stalled replication fork to a double strand break (DSB) which allows for DNA translesion sysnthesis to restore a nascent strand. The DSB is subsequently repaired by homologous recombination (HR) which is activated by the FA proteins FANCD1, J, N, O. FANCD2 is finally deubiquitinated by USP1 after the pathway is completed and the ICL has been resolved[170, 171]. Several mouse models deficient in the FA pathway have been synthesized and while most demonstrate the sensitivity to MMC expected with an FA defect, very few demonstrate hematopoietic defects necessary to mimic the FA human phenotype. Included in those models with no hematopoietic phenotype are FANCA, FANCC, FANCD2, FANCG, FANCL, and FANCM[172, 173]. However a mouse model containing a hypomorphic mutation in the FANCD1 gene (BRCA2) demonstrates not only hypersensitivity to MMC but also significant hematopoietic abnormalities, suggesting that FANCD1 may most closely resemble the human disease. FANCD1 mice demonstrated reduced numbers of hematopoietic CFCs, their bone marrow contained high levels of spontaneous chromosomal instability, the mouse HSCs displayed a competitive repopulation defect, a niche occupancy defect, a proliferation defect, and the mice were hypersensitive to ionizing radiation[37]. 122

134 We thus propose that FANCD1 is a suitable mouse model with which to study the effects of quiescence maintenance on the HSC damage response. While the issue is a bit more complicated due to the lack of basal defect in Exo1 mut mice vs. a fundamental defect in FANCD1 mice we hypothesize that an HR defect in FANCD1 mice combined with the observed loss of HSC quiescence contributes to HSC dysfunction and IR hypersensitivity, both of which may be ameliorated via a reversion to HSC quiescence. We propose to test this hypothesis using the following specific aims: (1) Characterize the DSB repair capacity of FANCD1 mutant mice compared to mice defective in NHEJ (SCID) and HR (Exo1 mut ). We will assess DSB sensitivity and damage persistence in cycling cells in vitro and compare to the phenotype of mice at both steady state and after 5-FU or Poly IC HSC mobilization. (2) Determine whether ex vivo treatment of FANCD1 mutant bone marrow will result in a reversion to HSC quiescence and a subsequent improvement in hematopoietic function and DSB repair capacity following marrow transplantation. These findings would substantiate our hypothesis that quiescence restoration can supersede defects in DNA repair and may have clinical implications in the treatment of FA patients. Our studies have demonstrated a fundamental role for HSC cell cycle status in determining DNA repair capacity and can be applied to other repair genes and pathways. For instance our observation that loss of Exo1 mediated HR had no effect on steady state HSC function was surprising, and because it differed from such HR mouse models as BRCA2 and Rad50, suggests mutations in other HR genes may also yield unique HSC phenotypes. Also because Exo1 is critical to DNA end resection and the IR 123

135 sensitivity is due at least in part to defective resection, it will be interesting to examine whether Exo1 mut combined with overexpression of other resection enzymes including CtIP would be able to partially restore the HSC failure we observed after dual treatment. These findings are not specific to DSB repair and IR, for instance what effect does HSC activation have on mouse models of MMR deficiency such as the previously described Msh2 -/- mouse? Would prior HSC activation alter the alkylating agent response of these mice perhaps through a transition in repair reliance? In addition there are still unanswered questions about the post damage response of HSCs such as factors necessary to return to quiescence and whether complete damage repair is necessary before returning to quiescence. Thus while our studies begin to answer some of these remaining questions, it will require a combination of the Exo1 mut mouse and other repair deficient models to further elucidate the HSC response to multiple forms of DNA damage. 5.2 DNA Repair and Cancer Stem Cell Specific Therapies Cancer stem cells are believed to be a malignant subset within a bulk tumor that contribute to tumor relapse and therapy resistance. In Chapter 3 we report our findings regarding the role of DNA repair pathways involved in IR repair in cancer stem cell populations. Using two lung cancer cell lines we show that CD133+ radiation resistance is cell-type specific and that only in basal A549 non-small cell lung carcinoma cells did DNA repair appear to contribute to this IR resistance. However in both cell types CD133+ cells demonstrated enhanced DSB repair capacity following prior IR exposure 124

136 and that following an IR recovery period, both A549 and H1299 cells acquired IR resistance corresponding with an increase in gene expression of multiple DNA repair genes and improved DNA repair capacity. We also silenced Exo1 and Rad51 in A549 cells and found that this abrogated CD133+ expansion following IR and induced IR hypersensitivity. Together these findings provide insight into the role that DNA repair upregulation plays in non-small cell lung cancer stem cells and a clear rationale for DNA repair genes as therapeutic targets to combat IR resistance DNA Repair Protein Targets in Clinical Trials DNA repair proteins have long been exploited for their potential in many human cancers, perhaps none better studied than the synthetic lethality observed using poly A ADP ribose polymerase (PARP) inhibitors in breast and ovarian cancers containing mutations in BRCA1 or BRCA2. This treatment manipulates the fact that due to loss of BRCA function these cancers are DSB repair defective and the inhibition of PARP creates large numbers of single strand breaks that are converted to DSBs that cannot be repaired. This accumulation results in cell death[174]. Other factors including the NHEJ protein DNA-PK have been studied for their potential as cancer therapy, with the inhibitor NU7026 and others currently being explored in clinical trials for their ability to induce hypersensitivity to IR and other DNA damaging agents[175]. MGMT expression has been shown critical to cancer cell survival following alkylating agent treatment via its removal of alkyl groups from DNA, thus inhibitors including Lomeguatrib and 06- benzylguanine have entered clinical trials with the goal of improving traditional 125

137 alkylating agent therapy[176, 177]. Drugs that target proteins from all DNA repair and signaling pathways have entered clinical trials including the Chk1 inhibitor UNC-01 being studied in humans to block the G2/M checkpoint, BER inhibitors such as TRC102 which has demonstrated promise in synthetic lethal settings, and as described in Chapter 3.5 Rad51 inhibitors to inactive homologous recombination and induce IR hypersensitivity[18]. These trials and others have provided substantial evidence for the importance of repair enzymes in many cancers, yet further studies specifically on the CSC populations will be even more effective in determining their therapeutic value. Because CSCs are believed to rely on several pathways that contribute to chemoresistance, including generation of a hypoxic niche, activation of pro-survival pathways, and upregulation of DNA repair, it will be interesting to determine which enzymes are most effective in targeting more than one of these pathways. For instance inhibition of a DSB repair enzyme such as Rad51 combined with IR may impair the DSB response of CSCs and also generate enough ROS to disrupt the hypoxic niche. Thus a deeper understanding of the reliance that individual tumor types have to specific repair pathways and the role that this may play in maintaining a conserved cancer cell phenotype, especially as it relates to the cancer stem cell niche will potentially lead to more effective and long lasting therapies. In addition to identifying and developing DNA repair proteins as potential CSC specific therapies, we believe that our work described in Chapter 2 may have relevance 126

138 to the CSC population as well. We described how the HSC transition from quiescence to active cell cycle resulted in a shift in DNA repair reliance, and in the case of the Exo1 mut mouse, subsequent hypersensitivity to IR. Cancer stem cells are believed to be another populations that reside in a largely quiescent state as well, becoming active in response to stress or tumor repopulation. It is this quiescent state that is also believed to contribute to therapy resistance via maintenance of a conserved niche [92]. It is thus very likely that cell cycle activation of quiescent CSCs would result in sensitization to traditional chemotherapy via combined disruption of the CSC niche and the fact that actively cycling cells display higher drug sensitivity to DNA damaging agents[178, 179]. Thus combination therapy similar to what we previously utilized, such as 5-FU + IR, may have a significant effect on cell sensitization. In the context of cancers already containing impaired DNA repair capacity, such as BRCA mutant breast or ovarian cancer, this concept may be even more beneficial because the tumor cells would already be hypersensitive to IR. While this treatment strategy has been used in pancreatic cancer for years, this potential mechanism of cell cycle activation contributing to cell death has not been evaluated. Thus the therapeutic potential of this strategy may be expanded upon further study. We propose to test this hypothesis by studying the following specific aims, which are also modeled in Figure 5.2. (1) Measure the cell cycle status of multiple cancer stem cell markers in established xenografts of lung cancer cell lines. We propose to perform pyronin Y and propidium iodide staining in the CD133+, CD166+, and CD34+ populations to determine which marker selects for the most quiescent stem cell population in vivo. 127

139 (2) Identify treatment strategies to mobilize the quiescent population into active cell cycle. We will treat xenograft mice with IR, 5-FU, and poly IC and after multiple time points (1 d, 2 d, 5 d) measure cell cycle status of the cancer stem cell markers listed above. This will provide insight into the treatment regimen that will push quiescent cancer stem cells into cycle where they can subsequently be treated with a second agent to induce cytotoxicity. (3) Measure in vivo tumor progression of mice treated with sequential therapies. After identifying the agent and time required to CSC cell cycle entry, we will treat mice with a second DSB inducing therapy of IR, etoposide, or doxorubicin. We will measure tumor progression over time and sacrifice animals at multiple timepoints to determine CSC frequencies after this dual therapy. Thus we believe that our work points to the necessity of studying DNA repair capacity of CSCs vs. the differentiated bulk tumor to potentially identify stem cell specific targets. Using NSCLC cell lines we identified that this reliance can be tumor specific but also may be activated upon therapy exposure. Further studies using additional chemotherapy agents and multiple tumor types will determine whether CSCs globally become DNA repair reliant after therapy or whether certain forms of damage will induce this repair upregulation. Further exploration of DNA repair proteins critical to CSCs, and disruption of the CSC quiescent niche via cell cycle activation may be important tools in treating therapy resistant cancers. 128

140 Figure 5.2 Proposed model for studying CSC cell cycle activation and combination treatment Cell cycle quiescence has been a hypothesized mechanism for therapy resistance in multiple tumor types. We propose to study the effects of activating CSCs in tumors followed by treatment with traditional chemotherapy to determine whether disruption of the quiescent niche would yield tumor hypersensitivity to DSB inducing agents. 129

141 Figure 5.2 Proposed model for studying CSC cell cycle activation and combination treatment 130

142 5.3 Maintenance of genomic stability via compensatory DNA Repair pathways Genomic instability is characterized by high frequencies of mutations in cells and is a driving force towards tumorigenesis. DNA repair pathways are essential for resolving these mutations and defective repair genes are highly correlated with increased instability. In Chapter 4 we discussed our findings of MMR capacity in Exo1 mut MEFs. We showed that MMR remained functional after loss of Exo1 enzymatic activity as measured through temozolomide sensitivity and G:T mismatched base repair. We performed gene expression analysis of multiple nucleases following temozolomide treatment and identified the 5 3 nucleases Artemis, Fan1, and Mre11 as potentially being involved in Exo1 independent MMR. After partially silencing each of these enzymes alone or in combination in WT and Exo1 mut cells we measured temozolomide sensitivity and G:T repair capacity and concluded that these nucleases are involved in a compensatory pathway cells utilize to avoid mutation accumulation. Our work along with the cited studies in Chapter 4.5 suggests that the structural component of Exo1 may be required to recruit additional factors in this compensatory pathway. However the domains required, the binding partners involved, or the mechanism of recruitment involving Exo1 not been explored. Mre11 and Fan1 have both been found to interact with MutLα, as has Exo1, thus its structural role may be to serve as a scaffolding protein to compile larger protein complexes. To study this would require testing the following aims: (1) Identify the Exo1 independent temozolomide induced MMR protein complex in vitro. This would involve a co-immunoprecipiation study of Mlh1 binding proteins after temozolomide treatment in WT and Exo1 mut MEFs. 131

143 Subsequent western blotting of Artemis, Fan1, Mre11, and other nucleases in the eluted proteins would determine whether any of these enzymes is pulled down with the MMR complex. It would also help determine other non-nuclease proteins that may be involved in Exo1 independent MMR, as the pulldown products would also be run on an SDS-PAGE gel, stained with coomassie blue, and differences between the WT and Exo1 mut temozolomide treated complexes compared for differences. Bands appearing singly in either complex would be analyzed for mass-spectroscopy and potential lead proteins would be silenced to assess MMR capacity after its loss (Figure 5.3). (2) Determine the structural domains of Exo1 necessary for protein recruitment in MMR. This could be studied by synthesizing truncated mutants of full length Exo1. These mutants would be introduced into Exo1 silenced cells and the effects on MMR capacity measured, as indicated by temozolomide sensitivity and reporter plasmid. Those mutants that demonstrated an MMR defective phenotype, and conversely those mutants that retained MMR capacity, would provide specific insight into amino acids required for Exo1 in MMR. These mutants would then also be assessed to determine if the same components of Exo1 that are required for MMR are also the ones necessary for somatic hypermutation, class-switch recombination, and meiosis. Thus there are several future studies required to truly distinguish the structural vs. catalytic roles of Exo1 in many cellular processes. This compensation of nucleases to maintain pathway fidelity is interesting for a variety of reasons. In MMR it remains to be fully determined whether these enzymes together form a complex of nucleases necessary for proper repair, if any of these is 132

144 interchangeable at steady state, or if they are activated to compensate for the loss of a critical nuclease as our data suggests in the Exo1 mut model. Because nucleases are required for all DNA repair pathways and many cellular processes, and complete loss of nuclease activity would result in cell death, it becomes important to ask whether this redundancy exists in other pathways. For instance in the end resection step of HR it has recently been proposed that while Mre11 activity is dispensable for resection with compensation through Exo1, the loss of Exo1 cannot be compensated for. In this study we used temozolomide treatment followed by gene expression analysis to identify potentially implicated nucleases and believe that this strategy may also be useful for other pathways. For instance in nucleotide excision repair the activity of XPG and XPF is responsible for excising damaged nucleotides. Exposing cells to a crosslinking agent such as UV radiation followed by expression analysis of multiple nucleases may also yield potentially implicated components, with further silencing studies as described in Chapter 4 being used to characterize their roles. Continuing to study the mechanism of MMR in the absence of Exo1 as described above, and beginning to study whether such enzymatic compensation occurs in other cellular processes will glean important insight into previously uncharacterized tactics cells use to conserve genomic stability. Whether these findings would have direct implications on human health and disease remains to be discovered. In total our work described in these projects presents a multifaceted view on the importance of DNA repair in normal and malignant stem cell populations and may have implications on understanding human disease. Continuing to closely study the 133

145 relationship between DNA repair pathways and multiple cellular processes may uncover previously unknown factors or pathways that cells utilize to maintain genomic stability and thus provide potential therapeutic targets in cancers and other malignancies. 134

146 Figure 5.3 Proposed studies to identify additional components involved in Exo1 independent MMR Co-immunoprecipitation studies in WT and Exo1 mut MEFs treated with temozolomide may provide insight into differences between the MMR protein complexes in each setting. After treating cells with temozolomide, the nuclear fraction would be isolated and Mlh1 and its associated proteins pulled down. This eluted complex would be probed for the identified nucleases Artemis, Fan1, and Mre11, and would also be stained with coomassie blue to distinguish additional proteins that may be involved exclusively in either complex. 135

147 Figure 5.3 Proposed studies to identify additional components involved in Exo1 independent MMR 136

148 APPENDIX 137

149 Appendix 1: Exo1 mut mice demonstrate normal bone marrow characteristics at steady state (A) Animal weight of age matched (8-12 week) mice. Error bars represent SEM. (B) Peripheral eye blood counts were recorded from five age matched mice. (C) Representative FACs profiles of whole bone marrow pregated for lineage negative cells. The Sca1+,C-Kit+ (SKL) cells and the SKL CD48-,CD150+ (SLAM) cells are represented. (D) Flow cytometry of peripheral eye blood stained for CD3,B220 and Mac1 cells from B6 WT and Exo1mut mice. 138

150 Appendix 2: Measurement of 5-FU Induced SKL Cell Cycle Entry and SKL Measurement after 5-FU and IR in WT and Exo1 mut Mice (A) WT and Exo1 mut mice were treated with 150 mg/kg 5-FU and the percent BRDU+ SKL cells were recorded 5 days after treatment via flow cytometry of whole bone marrow. (B) SKL levels of WT and Exo1 mut mice 14 days following either 150 mg/kg 5-FU or 4Gy IR only. Number of animals alive 3 months post either treatment is displayed. 139

151 Appendix 3: CD133+ cells expand following ionizing radiation in human lung cancer cell lines A549 and H1299 human lung cancer cell lines were treated with 4Gy ionizing radiation and the percent positive CD133 cells were measured using cytometric analysis every 24 hours post treatment. 140

152 Appendix 4: Purity of magnetically sorted CD133 cells Cells were sorted via magnetic separation (Miltenyi). Sorted cells were stained with PE tagged CD133 antibody (293C- Miltenyi) and analyzed via flow cytometry. 141

153 Appendix 5: Confirmation of gene expression data using western blotting CD133+ cells were magnetically separated from A549 and H1299 cells. Western blot analysis was performed on cell lysates from each population probing for DNA repair proteins Exo1, Ku70, Rad51 and Msh2 to demonstrate basal protein levels in CD133- and CD133+ cells. 142

154 Appendix 6: Confirmation of lentiviral knockdown of Exo1 and Rad51 in A549 cells A549 cells were transduced with lentiviral shrna specific to Exo1, Rad51, or scrambled control (Sigma). After puromycin selection cells were probed for Exo1 and Rad51 using western blot analysis to determine the levels of protein knockdown. 143

155 Appendix 7: Confirmation of real-time PCR data using western blotting WT and Exo1mut MEFs were treated with 250ug/mL temozolomide and protein lysate recovered 24 hours post treatment. Western blotting was performed using antibodies to Artemis (Abcam- ab83309), Mre11 (Cell Signaling-4895), Fen1 (Santa Cruz- SC13051), XPF (Santa Cruz- SC10164), and a tubulin control (Calbiochem- CP06). 144