A Mechanistic Evaluation of the Role of Aneuploidy During Oncogenesis by Ruoxi Wendy Wang B.S./M.S. Biology Brandeis University, 2016 SUBMITTED TO THE DEPARTMENT OF BIOLOGY IN PARTIAL FULFUILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN BIOLOGY AT THE MASSACHUSETTS INSTITUTE OF TECHNOLOGY FEBRUARY 2022 © 2022 Ruoxi Wendy Wang. All rights reserved. The author hereby grants to MIT permission to reproduce and to distribute publicly paper and electronic copies of this thesis document in whole or in part in any medium now known or hereafter created. Signature of the Author: __________________________________________________ Department of Biology Nov 8th, 2021 Certified by: ____________________________________________________________ Jaqueline A. Lees Virginia & D.K. Ludwig Professor for Cancer Research Thesis Supervisor Accepted by: ___________________________________________________________ Amy Keating Professor of Biology and biological Engineering Co-Director, Biology Graduate Committee 2 A Mechanistic Evaluation of the Role of Aneuploidy During Oncogenesis by Ruoxi Wendy Wang Submitted to the Department of Biology on November 8th, 2021 in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Biology Abstract Accurate chromosome segregation during cell division is critical to cellular fitness and survival. Errors during the segregation process often lead to aneuploidy, a state where cells harbor whole chromosome gains or losses. In untransformed cells, aneuploidy is highly detrimental to cell physiology, where it elicits multiple stress responses and impairs cell proliferation. Paradoxically, aneuploidy is also a hallmark of cancer and high degrees of aneuploidy in tumors are often associated with aggressive disease progression and poor prognosis. It is thus important to study the molecular mechanisms of aneuploidy during oncogenesis in order to reconcile the different effects of karyotype alteration in untransformed and cancer cells. In this thesis, we first investigate the mechanism by which untransformed cells harboring highly complex karyotypes trigger a natural killer cell-mediated immune response. We find that activation of the NF-κB pathway is responsible for such aneuploidy-associated immune clearance in vitro. We also provide evidence that potential mutations may counteract the NF-κB-mediated cytotoxicity during the cell transformation process. Second, we study the role of frequent chromosome 8 (chr8) gain in Ewing sarcoma, a pediatric bone and soft tissue tumor that is mainly driven by the EWS-FLI1 fusion oncogene. Here, we specifically investigate the molecular mechanism of one chr8 gain-driver gene, RAD21, in mitigating EWS-FLI1-induced replication stress and promoting oncogenesis. We find that the overexpression of RAD21 facilitates the resolution of transcription-replication conflicts in EWS-FLI1 expressing cells. This is achieved partially by RAD21’s recruitment to the stalled replication forks, where RAD21 interacts with DNA repair initiation proteins to promote efficient damage repair and stalled replication fork restart. In summary, our work reveals how the role of karyotype alterations during oncogenesis can be highly context-dependent. Whereas random aneuploidies bring fitness penalty in normal untransformed cells, a tumor-specific aneuploidy can provide benefits for cellular fitness and lead to positive selection of specific karyotypes. Outcomes from our study can implicate the potential therapeutic targets for treatment of aneuploid cancer. Thesis supervisor: Jacqueline A. Lees Title: Professor of Biology; Virginia & D.K. Ludwig Professor for Cancer Research 3 Acknowledgements Foremost, I would like to thank my thesis advisors Angelika and Jackie. Thank you Angelika for your unparalleled enthusiasm in science. Your guidance during the first four years of my graduate school helped me find my internal confidence and scientific perseverance. Thank you Jackie for all your support during the past year. Your patience and courage helped me navigate my way out. I also need to thank my undergraduate research mentor, Jim, for introducing me to the beauty of homologous recombination. Your passion for science set my path to pursue graduate school. I am so grateful to have such caring mentors throughout my training who guided me to grow as a scientist. Thanks to the Amon Lab and the Lees Lab for all your support during the past five years. Thanks to all the members of the Amon Lab for creating a fantastic environment for doing science. I especially need to thank Stefano and Allen for your guidance on my projects and providing mentorship during the past year. Thank you Erica and Monica for taking care of each of us in the Amon lab. You have done so much for the lab! I also want to thank Gabriel, Allegra, Ian, Snow, Tim, and David for being my bay mates during my time in the Amon lab. Thanks for all your great scientific discussions. Thanks to the Lees lab for creating such a welcoming environment and making my lab transition a heartwarming experience. Thank you for spending the time and energy providing valuable feedback on my projects. Thanks to my thesis committee members Mike Hemann, Iain Cheeseman, and Stefani Spranger for providing advice and help on my projects during the past five years. Thank you Alison Taylor for agreeing to be my outside committee member and providing great feedback on my thesis. Thanks to my classmates from Biograd 2016. I feel grateful to have you all as both colleagues and friends. Thank you Irene for being my roommate in graduate school and cooking all the food during quarantine over the pandemic. I would also like to thank my friends outside of graduate school, especially Judy, Michelle, Tianci, Chaoqun, and Hui for all their support. I want to acknowledge my spiritual family MITCEF and CBCGB for being my family in Boston. Thanks for both providing physical support and sending prayers during difficult times. Thank you Tianyi for always being available whenever I needed support. Finally, I would like to thank my incredible family. Thanks to Spencer, Marjorie, Alex, and Catherine for taking care of me during my time in the U.S. Most importantly, thanks to my parents for your never-ending support. Thanks for always listening to my feelings, cheering me up, and encouraging me to never give up. I could not have done this without you! 4 5 Table of contents Abstract ............................................................................................................................... 3 Acknowledgements ............................................................................................................ 4 Chapter 1: Introduction ....................................................................................................... 8 CHROMOSOME SEGREGATION ................................................................................. 9 Regulation mechanism ............................................................................................... 9 The causes of aneuploidy ......................................................................................... 13 Aneuploidy and chromosome instability ................................................................... 14 ANEUPLOIDY IN NORMAL CELLS ............................................................................. 14 Aneuploidy associated cellular stresses................................................................... 14 Aneuploidy compromises cell proliferation ........................................................... 14 Aneuploidy alters transcription .............................................................................. 15 Aneuploidy disrupts protein homeostasis ............................................................. 16 Aneuploidy induces replication stress................................................................... 18 Aneuploidy causes DNA damage ......................................................................... 19 Aneuploidy and immune response ........................................................................... 22 ANEUPLOIDY AND ONCOGENESIS.......................................................................... 22 Aneuploidy is a hallmark of cancer ........................................................................... 22 Aneuploidy as a passenger event during tumorigenesis ......................................... 24 Aneuploidy as a driver for tumorigenesis ................................................................. 25 SUMMARY .................................................................................................................... 26 REFERENCES ............................................................................................................. 27 Chapter 2: Aneuploid senescent cells activate NF-κB to promote their immune clearance by NK cells ....................................................................................................... 33 ABSTRACT ................................................................................................................... 34 INTRODUCTION .......................................................................................................... 35 RESULTS ...................................................................................................................... 37 An assay to assess elimination of ArCK cells by natural killer (NK) cells in vitro ... 37 Prolonged cell cycle arrest associated with features of senescence elicits NK cell- mediated cytotoxicity ................................................................................................. 44 Mechanisms triggering senescence contribute to NK cell recognition in ArCK cells ................................................................................................................................... 47 NF-κB and interferon-mediated pathways are upregulated in ArCK cells ............... 53 Both canonical and non-canonical NF-κB pathways are required for NK cell- mediated killing of ArCK cells ................................................................................... 57 Retrotransposon activation is involved in triggering immune clearance of ArCK cells ................................................................................................................................... 60 NF-κB pathway is upregulated in highly aneuploid cancer cell lines ....................... 65 6 DISCUSSION ................................................................................................................ 68 The NF-κB pathway contributes to the immunogenicity of ArCK cells .................... 68 NF-κB activation in ArCK cells relies on multiple signals ........................................ 69 Immune recognition of aneuploidy in cancer............................................................ 70 MATERIALS AND METHODS ..................................................................................... 71 ACKNOWLEDGEMENTS ............................................................................................. 80 REFERENCES ............................................................................................................. 81 Chapter 3: RAD21 Promotes Repair of Oncogenic Replication Stress-Induced Damage in Ewing Sarcoma ............................................................................................................. 89 ABSTRACT ................................................................................................................... 90 INTRODUCTION .......................................................................................................... 91 RESULTS ...................................................................................................................... 93 EWS-FLI1 induced transcription-replication conflicts cause oncogenic replication stress ......................................................................................................................... 93 RAD21 is enriched at the TRCs and is recruited to stalled replication forks......... 101 Increased RAD21 attenuates TRCs and promotes fork progression under EWS- FLI1 induced stress ................................................................................................. 106 RAD21 interacts with DNA damage initiation proteins upon EWS-FLI1 induction 113 DISCUSSION .............................................................................................................. 120 The sources of EWS-FLI1 induced replication stress ............................................ 120 The role of RAD21 in mitigating oncogene-induced replication stress.................. 121 Interaction of RAD21 with DNA damage repair proteins ....................................... 122 MATERIALS AND METHODS ................................................................................... 125 ACKNOWLEDGEMENTS ........................................................................................... 134 SUPPLEMENTAL TABLE S1 ..................................................................................... 135 SUPPLEMENTAL TABLE S2 ..................................................................................... 137 REFERENCES ........................................................................................................... 138 Chapter 4: Discussion .................................................................................................... 144 THE IMMUNE RESPONSE IN ANEUPLOID CANCER ............................................ 146 DISTINGUISHING ANEUPLOIDY PASSENGER AND DRIVER EVENTS .............. 149 IDENTIFYING DRIVERS FOR TUMOR SPECIFIC ANEUPLOIDY .......................... 150 USING ANEUPLOIDY FOR POTENTIAL THERAPEUTIC IMPLICATIONS ............ 153 SUMMARY .................................................................................................................. 155 REFERENCES ........................................................................................................... 156 APPENDIX .................................................................................................................. 159 7 Chapter 1: Introduction 8 Aneuploidy is defined as a state in which a cell contains a chromosome number that is not a multiple of its haploid complement. In untransformed cells, aneuploidy is highly detrimental to cell physiology. An unbalanced karyotype causes multiple cellular stresses and leads to compromised cell proliferation. Interestingly, aneuploidy is also a hallmark of cancer, a disease characterized by uncontrolled cell proliferation. A high degree of aneuploidy is often associated with aggressive tumor progression and poor patient outcomes. Whether and how does aneuploidy plays a role during oncogenesis is still not fully understood. This introduction provides the background on previous work in the field specifically focusing on this question. First, I will revisit the regulatory mechanism of chromosome segregation and the potential causes of aneuploidy in somatic cells. Next, I will summarize the current understanding of the detrimental effects of aneuploidy on cell physiology. Lastly, I will discuss the current hypothesis on the role of aneuploidy during oncogenesis and how studying aneuploidy in cancer can lead to promising therapeutic implications in certain cancer subtypes. CHROMOSOME SEGREGATION Regulation mechanism Cell division and chromosome segregation occur during the M phase (mitosis) of the cell cycle. A faithful segregation event leads to equal partition of the newly duplicated sister chromatid pairs from one mother cell into two daughter cells. In eukaryotes, the movement and separation of the sister chromatid pairs is mediated by the mitotic spindle, which is a bipolar microtubule array nucleated from centrosomes 9 and constructed around the chromosomes (Sullivan & Morgan, 2007). The mitotic spindle attaches to the sister chromatids at their centromeric regions through a multiprotein complex called the kinetochore (Cheeseman, 2014). The two kinetochores at any sister chromatid pair must be attached to opposite spindle poles, which leads to the bi-orientation of the sister chromatids at the spindle midzone during metaphase (Foley & Kapoor, 2013). This proper attachment also generates tension at the kinetochore-microtubule interface and further stabilizes the properly aligned sister chromatids (Cheeseman, 2014). During anaphase, the two chromosome sets are pulled apart by the mitotic spindle and move to the opposite poles. After the formation of daughter nuclei, cytokinesis leads to the division of the cytoplasmic content, generating two cells harboring equal DNA content (Figure 1a). Figure 1. Chromosome segregation and its regulation mechanisms. a) Accurate chromosome segregation during mitosis. Two euploid daughter cells harboring identical genomic content are generated. The figure is created using Biorender. To prevent premature chromosome segregation, the sister chromatids are held together until the onset of anaphase. The cohesion between the sister chromatids is partially mediated by DNA catenation, where the duplicated DNA is intertwined during replication. In addition, the sister chromatids are held together by a ring-shaped protein 10 complex called cohesin. During anaphase, upon proper alignment of sister chromatids, a protease named separase is activated and cleaves one of the cohesin subunits, Scc1, allowing separation of sister chromatids towards the opposite spindle poles. If proper chromosome attachment or alignment fails, separase’s activity is inhibited by a protein called securin to prevent cohesin cleavage (Nasmyth, 2002; Stemmann et al., 2001). The segregation of sister chromatids is tightly controlled to ensure accurate cell division. This surveillance process is mainly controlled by a regulatory mechanism called the spindle assembly checkpoint (SAC), which acts by monitoring the proper kinetochore-spindle attachment (Figure 2a). In cells with unattached or improperly attached kinetochores, the SAC is immediately activated. This leads to the recruitment of the SAC core components to the kinetochore, including mitotic arrest deficient 1 (MAD1), MAD2, BUB3, BUB1-related 1 (BUBR1) and the checkpoint kinases aurora B and monopolar spindle protein 1 (MPS1). MAD2 switches from an open conformation to a closed conformation upon binding to MAD1 at the kinetochore. This conformational change further catalyzes its binding to another protein, CDC20. CDC20 is a co-activator of the downstream anaphase promoting complex/cyclosome (APC/C), which is a E3 ubiquitin ligase complex essential for mitotic exit. BUBR1 and BUB3 can also form a complex with CDC20.The CDC20-MAD2 complex and the BUBR1-BUB3-CDC20 complex (together called the mitotic check point complex, MCC) work together to block APC/C function, thereby preventing the onset of anaphase (Musacchio & Salmon, 2007). Upon proper kinetochores-microtubule attachment and bi-orientation of the sister chromatids at the spindle midzone, the MCC is disassembled. This allows activation of the APC/CCDC20 complex, which targets securin and the mitotic cyclin B for proteolytic 11 degradation, leading to inactivation of cyclin-dependent kinase 1 (CDK1), loss of sister chromatid cohesin, and proper mitotic exit [Figure 2b,(Shirayama et al., 1999)]. 12 Figure 2. a) Simplified depiction of the core components in spindle assembly checkpoint (SAC). When the SAC is activated, mitotic arrest deficient 1 (MAD1), MAD2, BUB3, and BUB1‑related 1 (BUBR1) are recruited to the unattached kinetochore. The interaction between MAD1 and closed MAD2 catalyzes the formation of MAD2-CDC20 complex. Unattached kinetochores also lead to the formation of CDC20-BUBR1-BUB3 complex. Both complexes inhibit the formation of active APCCDC20 complex. b) Upon proper kinetochore-microtubule attachment, the SAC is turned off and APC/CCDC20 is activated. APC/CCDC20 targets both securin and cyclin B for proteolytic degradation. This inhibits CDK1 function, which further activates separase. Activated separase cleaves cohesin and triggers chromosome segregation. The figures are created using Biorender. The causes of aneuploidy Although surveillance mechanisms exist to ensure the fidelity of chromosome segregation, errors do arise during mitosis. Whereas the rate of spontaneous chromosome errors in vivo remains difficult to assess, studies using mammalian in vitro cell culture system showed that diploid cells on average mis-segregate a chromosome once every hundred cell divisions (Cimini et al., 1999; Thompson & Compton, 2008). Loss of function mutations in the SAC pathway leads to unlicensed anaphase onset before the proper kinetochore-microtubule attachment (Gordon et al., 2012), and thus can increase chromosome mis-segregation frequency. In line with this, inhibiting SAC kinase Mps1 with the chemical inhibitor reversine causes ~50% of mitoses in RPE1- hTERT cells to be aberrant (Santaguida et al., 2017). In addition, gene knockout mouse models have been constructed for many of the known mitotic checkpoint genes, such as Mad2, Bub1, and BubR1. The complete loss of these genes causes early embryonic lethality in mice, potentially due to the high levels of aneuploidy. Heterozygous and hypomorphic mice are viable, but in all cases, display an increased level of aneuploidy and chromosome instability (Holland & Cleveland, 2009). In addition, mutations in genes encoding subunits of the cohesin complex can lead to premature sister chromatid 13 separation and generate aneuploidy. For example, it has been shown that inactivating the function of cohesin subunit STAG2 leads to significant loss of sister chromatid cohesion and causes faulty mitosis. Furthermore, correcting the STAG2 mutation in highly aneuploid cancer cells attenuates chromosome instability (Solomon et al., 2011). Aneuploidy and chromosome instability Aneuploidy and chromosome instability are often co-existing events. By definition, aneuploidy is a cell state with an unbalanced number of chromosomes. Chromosome instability indicates a high frequency of karyotype alteration from one generation to the next (Potapova et al., 2013). It has been shown that aneuploidy leads to increased replication stress and upregulates DNA damage levels (Passerini et al., 2016; Santaguida et al., 2017) . Thus, the initial chromosome mis-segregation event can further potentiate chromosome instability, promoting karyotype evolution and generating cells with highly complex karyotypes. ANEUPLOIDY IN NORMAL CELLS Aneuploidy associated cellular stresses Aneuploidy compromises cell proliferation Aneuploidy is rare in normal somatic tissues (Knouse et al., 2014; Pfau et al., 2016). In untransformed cells, chromosome copy number alteration is highly detrimental to cell physiology. Across all organisms, aneuploidy is accompanied by slow cell proliferation and compromised cell fitness. For example, in haploid budding yeast, the gain of an extra chromosome leads to a significant delay in G1 phase of the cell cycle 14 (Torres et al., 2007). In fruit flies, only the loss of the smallest chromosome can be tolerated, but it causes significant decrease in body size (Zhu et al., 2018). In mammals, an unbalanced karyotype is even more detrimental. All of the constitutional monosomies and most of the trisomies are embryonic lethal in human. The only autosomal aneuploidy that allows survival beyond childhood is trisomy 21 (also known as Down syndrome), but the patients often have physical and mental disabilities as well as reduced life expectancies (Roper & Reeves, 2006). Consistent with these observations, it has been shown in vitro that both trisomic mouse embryonic fibroblasts (MEFs) and human fibroblasts exhibit much slower cell proliferation compared to their euploid counterparts (Sheltzer et al., 2017; Williams et al., 2008). In addition, cells harboring random aneuploidies induced by either genetic or chemical perturbation of the SAC components exhibit significant cell cycle delay or arrest, and are often accompanied with features of cellular senescence (Santaguida et al., 2017). Aneuploidy alters transcription The change of chromosome copy numbers alters the relative gene dosage on the affected chromosome. Notably, RNA sequencing across different trisomic model organisms suggests that the transcription level is proportional to copy numbers in aneuploid cells (Sheltzer et al., 2012). Beside this direct dosage effect due to the alteration of specific chromosomes, meta-analyses of gene expression data across diverse aneuploid organisms also revealed some common signatures that are largely shared across all aneuploidies. In particular, signatures related to cell growth such as DNA replication, mitosis, and chromosome condensation are often down regulated in 15 aneuploid organisms/cells compared to their euploid counterparts. Furthermore, gene ontology (GO) terms related to cellular stress, such as inflammatory response and reactive oxygen species (ROS), are commonly upregulated across all aneuploid cells (Sheltzer et al., 2012). Multiple stimuli can contribute to the upregulated stress response that results from karyotype alteration. First, aneuploidies that have gains of chromosomes require increased energy consumption and altered metabolism to deal with the increased transcription, translation and protein degradation, which overwhelms the stress response system (Zhu et al., 2018). Second, because aneuploid cells often exhibit slower mitotic progression, this prolonged mitosis often triggers activation of the stress kinase p38, which either elicits proinflammatory cytokine secretion or induces apoptosis (Simões-Sousa et al., 2018). Finally, it has been shown that the stress signatures exhibited by aneuploid cells are highly similar to the environmental-stress response (ESR), a gene expression signature that is triggered by exogenous stress stimuli including heat shock and oxidative stress (Gasch et al., 2000). As I describe below, protein aggregations and oxidative stresses can be prominent downstream consequences of chromosome mis-segregation that can contribute to the stress-related gene expression signatures in aneuploid cells. Aneuploidy disrupts protein homeostasis Aneuploidy has been shown to cause proteotoxicity in untransformed cells. Specifically, the change in gene dosage leads to an equivalent change in protein levels, which also disrupts cellular protein composition and stoichiometry (Brennan et al., 16 2019). Cellular protein homeostasis is tightly monitored by the protein quality control system. Chaperones mediate the proper folding of the protein into their functional conformations. In addition, proteases and autophagy cooperate together to eliminate misfolded proteins or improper protein interactions. Under physiological conditions, most the proteins are subunits of multimeric protein complexes and can only be stable and functional when properly assembled. In aneuploid cells, protein subunits encoded by the unbalanced chromosome are generated in excess and thus lack their binding partners. Such protein subunits need to be either bound by chaperones to remain soluble or degraded by the proteases, which significantly increase the burden on the protein quality control system [Figure 3, (Oromendia & Amon, 2014)]. This hypothesis has been confirmed by several studies. For example, yeast disomic strains are more sensitive to both the proteasome inhibitor MG132 and the protein chaperone Hsp90 inhibitor 17-AGG compared to their euploid counterparts (Tang et al., 2011; Torres et al., 2007), indicating an overloaded proteasome in aneuploid cells. Moreover, both trisomic and random aneuploid human cells exhibit upregulation of lysosome mediated degradation and autophagy signatures, and it is evident that protein aggregates accumulate within lysosomes (Santaguida et al., 2015). 17 a. Euploid cells b. Aneuploidy cells Figure 3. Aneuploidy causes proteotoxic stress. a) In euploid cells, quality control (QC) systems maintain protein homeostasis. Protein chaperones promote protein folding and prevent protein aggregation. Misfolded protein subunits are degraded by the proteasome. b) Protein quality control system in aneuploid cells. Alteration of chromosome copy numbers disrupts protein complex stoichiometries. The protein subunits encoded by an unbalanced chromosome do not have their binding partners and rely on chaperones. This leads to an increased burden on the protein quality control system. [Adapted from (Oromendia and Amon, 2014)]. Aneuploidy induces replication stress Replication stress is also a prominent feature observed in most of the aneuploid cells. One possible cause is that the change in gene copy number disrupts protein stoichiometry of critical DNA replication and segregation complexes, thereby yielding to 18 a compromised DNA replication machinery. High replication stress leads to decreased replication fork speed and more frequent replication fork stalling (Primo & Teixeira, 2020). At the stalled replication fork, ssDNA generated by helicase activity is immediately coated and protected by the ssDNA binding protein RPA. The resulting ssDNA-RPA complex prevents formation of secondary structures and also activates downstream replication checkpoints including ATR/CHK1 (Kotsantis et al., 2018), which further stabilizes the fork and inhibits more unscheduled replication origin firing. Despite the mechanism, if replication stress persists, it will eventually induce irreversible replication fork collapse and generate double stranded breaks (DSBs), which activates downstream ATM kinase for DSB mediated DNA damage repair (Liao et al., 2018). Indeed, stalled replication forks and slowed replication fork speed are evident in human RPE1-hTERT cells harboring random aneuploidy as measured by DNA combing assay (Passerini et al., 2016; Santaguida et al., 2017). Furthermore, in both trisomic and random aneuploid human cells, there is an accumulation of unrepaired replication- induced DNA lesions as measured by 53BP1 foci (Ohashi et al., 2015; Santaguida et al., 2017). Thus, replication stress acts to induce genomic instability in aneuploid cells. Aneuploidy causes DNA damage Beside increased replication stress, other factors can also contribute to the elevated DNA damage level in aneuploid cells. For example, lagging chromosomes that remain at the spindle midzone can be trapped in the cleavage furrow and form DSBs during cytokinesis (Janssen et al., 2011). Lagging chromosomes can also be partitioned into membrane bound structures called micronuclei. Because micronuclei are prone to 19 envelope rupture, micronuclear DNA is susceptible to damage by cytosolic components (Hatch et al., 2013). In addition, DNA ultrafine bridges (UFBs), the naked DNA fibers connecting separating sister chromatids, are also frequently observed in aneuploid cell mitosis. UFBs are prone to DNA damage if they are not resolved properly (Levine & Holland, 2018). Thus, multiple sources of DNA damage exist, which eventually lead to p53 activation and cell cycle arrest in the aneuploid cells. In conclusion, chromosome mis-segregation and aneuploidy impose detrimental effect on cell physiology in untransformed cells. Such adverse effects are reflected at the level of both transcription and translation. In addition, aneuploidy causes replication stress and elevates DNA damage, which further induces genomic instability. The cellular toxicity brought by aneuploidy eventually lead to an impaired cell proliferation and compromised cellular fitness (Figure 4). 20 Figure 4. Aneuploidy induces detrimental effects on cell physiology. An aneuploidy- associated gene signature is similar to the environmental-stress response (ESR)-like response where many stress-response genes are upregulated and cell proliferation 21 genes are downregulated. Aneuploidy leads to upregulated proteotoxicity and energy stress and causes slowed proliferation. In addition, chromosome mis-segregation upregulates replication stress and triggers genomic instability, leading to upregulated DNA damage levels, which in turn activates DNA damage kinase ataxia-telangiectasia mutated (ATM) and stress kinase p38. p53 activation is a frequent outcome in aneuploid cells, leading to impaired proliferation or increased apoptosis. AU, arbitrary unit; ROS, reactive oxygen species. Modified from (Santaguida and Amon, 2015). Aneuploidy and immune response The state of aneuploidy elicits cellular stress responses. Such stress responses can include the production of inflammation signals that activate both the innate and adaptive immune system. For example, cells under proteotoxic stress express peptides derived from heat shock proteins, such as Hsp60, which can bind to MHC I complex and activate natural killer (NK) cells for proteolytic killing (Michaëlsson et al., 2002). In addition, cells harboring micronuclei exhibit increased levels of cytosolic DNA due to the frequent envelop rupture, which activates the cGAS-STING pathway and triggers the production of downstream interferons and other proinflammatory cytokines for immune activation (Bakhoum et al., 2018; Dou et al., 2017; Harding et al., 2017; Mackenzie et al., 2017). Studies from our lab have shown that aneuploid cells harboring complex karyotypes exhibit higher NK cell mediated-cytotoxicity (Santaguida et al., 2017). However, the detailed molecular mechanisms responsible for such immune clearance were unclear, and is the focus of Chapter 2 of this thesis. ANEUPLOIDY AND ONCOGENESIS Aneuploidy is a hallmark of cancer Although aneuploidy is highly detrimental to cell physiology and hinders cell proliferation in untransformed cells, it is also a common hallmark of cancer, a disease 22 characterized by uncontrolled cell proliferation (Figure 5). In fact, over a hundred years ago, Theodor Boveri recognized widespread aneuploidy in tumors and hypothesized that abnormal chromosome constitution may be a driving force for oncogenesis (Boveri 1902; Boveri 1914; Weaver and Cleveland 2006). During the past decade, the high prevalence of chromosome number variations across all cancer subtypes has been extensively confirmed with more advanced techniques such as next-generation whole genome sequencing. Here, the concept of aneuploidy in cancer genome field has been expanded from the classical definition of whole-chromosome alterations to include both arm-level gains and losses (Ben-David & Amon, 2020). Pan-cancer analysis revealed that ~90% of solid tumors and ~75% of hematopoietic malignancies are aneuploid (Weaver & Cleveland, 2008). The degree of karyotype alteration is also dramatic; it has been shown that on average ~25% of the genome is altered at either whole- chromosome or chromosome arm levels in a cancer cell (Gordon et al., 2012). Importantly, it has been shown that the degree of aneuploidy in tumors is positively correlated with the disease stage. Specifically, large karyotype alteration is generally associated with higher tumor grade, metastasis and poor prognosis (Ben-David & Amon, 2020). However, whether aneuploidy is actually a cancer driver, and how aneuploidy plays a role during oncogenesis still remains unclear. 23 Figure 5. Chromosome karyotype of a human normal cell (right) and a breast cancer cell (left). The normal cell is euploid and contains 23 pairs of chromosomes. The breast cancer cell is highly aneuploid with multiple chromosome gains and losses, as well as chromosomal translocations and fusions. [Adapted from (Duesberg et al., 2007)]. Aneuploidy as a passenger event during tumorigenesis Given the fact that aneuploidy exhibits deleterious effects on normal physiology, it is entirely possible that the prevalent chromosome copy number alteration seen across all cancer subtypes is merely a consequence, not a cause, of tumorigenity, due to the uncontrolled cell proliferation and high mutational burden during aggressive cancer progression. In such a scenario, additional adaptive mutations could be required to tolerate the detrimental effects of aneuploidy on cellular fitness. For example, the loss of the tumor suppressor gene p53 is observed in over 50% of human cancers (Baugh et al., 2018), and it has been well established that the absence of functional p53 impairs chromosome mis-segregation from triggering cell cycle arrest or apoptosis regardless of the high levels of DNA damage induced by aneuploidy. 24 Aneuploidy as a driver for tumorigenesis Interestingly, in many cases, karyotype alterations in tumors are not purely random, raising the possibility that certain aneuploidy event, might also be drivers for oncogenesis. Sequencing analysis on patient tumor samples revealed strong preferences in gaining or losing specific chromosomes in different cancer lineages, especially in the more aggressive cancer subtypes (Ben-David & Amon, 2020). For example: loss of chromosome arm 3p is observed in over 90% of the clear cell renal cancer patients (Mitchell et al., 2018); chromosome 17p arm loss has a high recurrence in chronic lymphocytic leukemia, and is associated with poor prognosis (Buccheri et al., 2018); and chromosome 22q arm loss is prevalent across different thyroid cancer subtypes (Network et al., 2014). These observations potentiate the hypothesis that certain aneuploidies are selected for during tumorigenesis in specific cancer subtypes, where gaining or losing one or more genes located on a specific chromosome confers fitness benefits and promotes neoplastic growth. In line with this hypothesis, recent work by Su et al specifically investigated the role of highly prevalent chromosome 8 gain in Ewing sarcoma oncogenesis (Su et al, 2021). This study showed that having one extra copy of chromosome 8 significantly promotes cell proliferation in EWS-FLI1 fusion oncogene driven Ewing sarcoma. Computational and evolutionary approaches allowed identification of a group of candidate genes on chromosome 8q arm that significantly promote proliferation in EWS-FLI1 expressing cells when overexpressed (Su et al., 2021). This study provided strong evidence for chromosome 8 gain acting as a driver event in Ewing sarcoma during oncogenesis. Work in Chapter 3 of this thesis builds on this study. 25 SUMMARY In untransformed cells, aneuploidy is detrimental to the cell physiology. An unbalanced karyotype elicits a plethora of cellular stress responses and compromises cell fitness. However, aneuploidy is also a hallmark of cancer, characterized by high proliferation potential. Certain chromosome gains or losses are highly recurrent in specific cancer subtypes. These seemingly paradoxical observations suggest that the role of aneuploidy during oncogenesis is extremely context-dependent. If in cancer cells, aneuploidy is a random passenger event due to the high mutational burden, certain aneuploidy tolerating gene mutations are required to mitigate the deleterious effect of unbalanced karyotypes during cancer evolution. On the other hand, certain chromosome copy number alteration can act as driver events leading to better cellular fitness. Both scenarios are investigated in this thesis. In Chapter 2, I present the study focusing on the adverse effect of aneuploidy in untransformed cells and specifically investigate how unbalanced karyotypes can trigger natural killer (NK) cell- mediated immune response. Then, Chapter 3 provides insight into the beneficial effect of aneuploidy during tumorigenic development, in the context of chromosome 8 gain driving Ewing sarcoma oncogenesis. Specifically, I identify a molecular mechanism by which one driver gene on chromosome 8, RAD21, acts to mitigate oncogene induced replication stress in Ewing sarcoma. Together, my thesis work advanced our understanding of the roles of aneuploidy during tumorigenesis and identified mechanistic insights that might potentially be exploited as a therapeutic target for treatment of aneuploid cancer cells. 26 REFERENCES Bakhoum, S. F., Ngo, B., Laughney, A. M., Cavallo, J.-A., Murphy, C. J., Ly, P., Shah, P., Sriram, R. K., Watkins, T. B. K., Taunk, N. K., Duran, M., Pauli, C., Shaw, C., Chadalavada, K., Rajasekhar, V. K., Genovese, G., Venkatesan, S., Birkbak, N. J., McGranahan, N., … Cantley, L. C. (2018). Chromosomal instability drives metastasis through a cytosolic DNA response. Nature, 553(7689), 467–472. https://doi.org/10.1038/nature25432 Baugh, E. H., Ke, H., Levine, A. J., Bonneau, R. A., & Chan, C. S. (2018). Why are there hotspot mutations in the TP53 gene in human cancers? Cell Death & Differentiation, 25(1), 154–160. https://doi.org/10.1038/cdd.2017.180 Ben-David, U., & Amon, A. (2020). Context is everything: aneuploidy in cancer. Nature Reviews Genetics, 21(1), 44–62. https://doi.org/10.1038/s41576-019-0171-x Brennan, C. M., Vaites, L. P., Wells, J. N., Santaguida, S., Paulo, J. A., Storchova, Z., Harper, J. W., Marsh, J. A., & Amon, A. (2019). Protein aggregation mediates stoichiometry of protein complexes in aneuploid cells. Genes & Development, 33(15– 16), 1031–1047. https://doi.org/10.1101/gad.327494.119 Buccheri, V., Barreto, W. G., Fogliatto, L. M., Capra, M., Marchiani, M., & Rocha, V. (2018). Prognostic and therapeutic stratification in CLL: focus on 17p deletion and p53 mutation. Annals of Hematology, 97(12), 2269–2278. https://doi.org/10.1007/s00277-018-3503-6 Cheeseman, I. M. (2014). The Kinetochore. Cold Spring Harbor Perspectives in Biology, 6(7), a015826. https://doi.org/10.1101/cshperspect.a015826 Cimini, D., Tanzarella, C., & Degrassi, F. (1999). Differences in malsegregation rates obtained by scoring ana-telophases or binucleate cells. Mutagenesis, 14(6), 563– 568. https://doi.org/10.1093/mutage/14.6.563 Dou, Z., Ghosh, K., Vizioli, M. G., Zhu, J., Sen, P., Wangensteen, K. J., Simithy, J., Lan, Y., Lin, Y., Zhou, Z., Capell, B. C., Xu, C., Xu, M., Kieckhaefer, J. E., Jiang, T., Shoshkes-Carmel, M., Tanim, K. M. A. A., Barber, G. N., Seykora, J. T., … Berger, S. L. (2017). Cytoplasmic chromatin triggers inflammation in senescence and cancer. Nature, 550(7676), 402–406. https://doi.org/10.1038/nature24050 Duesberg, P., Li, R., Sachs, R., Fabarius, A., Upender, M. B., & Hehlmann, R. (2007). Cancer drug resistance: The central role of the karyotype. Drug Resistance Updates, 10(1–2), 51–58. https://doi.org/10.1016/j.drup.2007.02.003 27 Foley, E. A., & Kapoor, T. M. (2013). Microtubule attachment and spindle assembly checkpoint signalling at the kinetochore. Nature Reviews Molecular Cell Biology, 14(1), 25–37. https://doi.org/10.1038/nrm3494 Gasch, A. P., Spellman, P. T., Kao, C. M., Carmel-Harel, O., Eisen, M. B., Storz, G., Botstein, D., & Brown, P. O. (2000). Genomic Expression Programs in the Response of Yeast Cells to Environmental Changes. Molecular Biology of the Cell, 11(12), 4241–4257. https://doi.org/10.1091/mbc.11.12.4241 Gordon, D. J., Resio, B., & Pellman, D. (2012). Causes and consequences of aneuploidy in cancer. Nature Reviews Genetics, 13(3), 189–203. https://doi.org/10.1038/nrg3123 Harding, S. M., Benci, J. L., Irianto, J., Discher, D. E., Minn, A. J., & Greenberg, R. A. (2017). Mitotic progression following DNA damage enables pattern recognition within micronuclei. Nature, 548(7668), 466–470. https://doi.org/10.1038/nature23470 Hatch, E. M., Fischer, A. H., Deerinck, T. J., & Hetzer, M. W. (2013). Catastrophic Nuclear Envelope Collapse in Cancer Cell Micronuclei. Cell, 154(1), 47–60. https://doi.org/10.1016/j.cell.2013.06.007 Holland, A. J., & Cleveland, D. W. (2009). Boveri revisited: chromosomal instability, aneuploidy and tumorigenesis. Nature Reviews Molecular Cell Biology, 10(7), 478– 487. https://doi.org/10.1038/nrm2718 Janssen, A., Burg, M. van der, Szuhai, K., Kops, G. J. P. L., & Medema, R. H. (2011). Chromosome Segregation Errors as a Cause of DNA Damage and Structural Chromosome Aberrations. Science, 333(6051), 1895–1898. https://doi.org/10.1126/science.1210214 Knouse, K. A., Wu, J., Whittaker, C. A., & Amon, A. (2014). Single cell sequencing reveals low levels of aneuploidy across mammalian tissues. Proceedings of the National Academy of Sciences, 111(37), 13409–13414. https://doi.org/10.1073/pnas.1415287111 Kotsantis, P., Petermann, E., & Boulton, S. J. (2018). Mechanisms of Oncogene- Induced Replication Stress: Jigsaw Falling into Place. Cancer Discovery, 8(5), 537– 555. https://doi.org/10.1158/2159-8290.cd-17-1461 Levine, M. S., & Holland, A. J. (2018). The impact of mitotic errors on cell proliferation and tumorigenesis. Genes & Development, 32(9–10), 620–638. https://doi.org/10.1101/gad.314351.118 Liao, H., Ji, F., Helleday, T., & Ying, S. (2018). Mechanisms for stalled replication fork stabilization: new targets for synthetic lethality strategies in cancer treatments. EMBO Reports, 19(9). https://doi.org/10.15252/embr.201846263 28 Mackenzie, K. J., Carroll, P., Martin, C.-A., Murina, O., Fluteau, A., Simpson, D. J., Olova, N., Sutcliffe, H., Rainger, J. K., Leitch, A., Osborn, R. T., Wheeler, A. P., Nowotny, M., Gilbert, N., Chandra, T., Reijns, M. A. M., & Jackson, A. P. (2017). cGAS surveillance of micronuclei links genome instability to innate immunity. Nature, 548(7668), 461–465. https://doi.org/10.1038/nature23449 Michaëlsson, J., Matos, C. T. de, Achour, A., Lanier, L. L., Kärre, K., & Söderström, K. (2002). A Signal Peptide Derived from hsp60 Binds HLA-E and Interferes with CD94/NKG2A Recognition. The Journal of Experimental Medicine, 196(11), 1403– 1414. https://doi.org/10.1084/jem.20020797 Mitchell, T. J., Turajlic, S., Rowan, A., Nicol, D., Farmery, J. H. R., O’Brien, T., Martincorena, I., Tarpey, P., Angelopoulos, N., Yates, L. R., Butler, A. P., Raine, K., Stewart, G. D., Challacombe, B., Fernando, A., Lopez, J. I., Hazell, S., Chandra, A., Chowdhury, S., … Consortium, the Tracer. R. (2018). Timing the Landmark Events in the Evolution of Clear Cell Renal Cell Cancer: TRACERx Renal. Cell, 173(3), 611- 623.e17. https://doi.org/10.1016/j.cell.2018.02.020 Musacchio, A., & Salmon, E. D. (2007). The spindle-assembly checkpoint in space and time. Nature Reviews Molecular Cell Biology, 8(5), 379–393. https://doi.org/10.1038/nrm2163 Nasmyth, K. (2002). Segregating Sister Genomes: The Molecular Biology of Chromosome Separation. Science, 297(5581), 559–565. https://doi.org/10.1126/science.1074757 Network, T. C. G. A. R., Agrawal, N., Akbani, R., Aksoy, B. A., Ally, A., Arachchi, H., Asa, S. L., Auman, J. T., Balasundaram, M., Balu, S., Baylin, S. B., Behera, M., Bernard, B., Beroukhim, R., Bishop, J. A., Black, A. D., Bodenheimer, T., Boice, L., Bootwalla, M. S., … Zou, L. (2014). Integrated Genomic Characterization of Papillary Thyroid Carcinoma. Cell, 159(3), 676–690. https://doi.org/10.1016/j.cell.2014.09.050 Ohashi, A., Ohori, M., Iwai, K., Nakayama, Y., Nambu, T., Morishita, D., Kawamoto, T., Miyamoto, M., Hirayama, T., Okaniwa, M., Banno, H., Ishikawa, T., Kandori, H., & Iwata, K. (2015). Aneuploidy generates proteotoxic stress and DNA damage concurrently with p53-mediated post-mitotic apoptosis in SAC-impaired cells. Nature Communications, 6(1), 7668. https://doi.org/10.1038/ncomms8668 Oromendia, A. B., & Amon, A. (2014). Aneuploidy: implications for protein homeostasis and disease. Disease Models & Mechanisms, 7(1), 15–20. https://doi.org/10.1242/dmm.013391 Passerini, V., Ozeri-Galai, E., Pagter, M. S. de, Donnelly, N., Schmalbrock, S., Kloosterman, W. P., Kerem, B., & Storchová, Z. (2016). The presence of extra chromosomes leads to genomic instability. Nature Communications, 7(1), 10754. https://doi.org/10.1038/ncomms10754 29 Pfau, S. J., Silberman, R. E., Knouse, K. A., & Amon, A. (2016). Aneuploidy impairs hematopoietic stem cell fitness and is selected against in regenerating tissues in vivo. Genes & Development, 30(12), 1395–1408. https://doi.org/10.1101/gad.278820.116 Potapova, T. A., Zhu, J., & Li, R. (2013). Aneuploidy and chromosomal instability: a vicious cycle driving cellular evolution and cancer genome chaos. Cancer and Metastasis Reviews, 32(3–4), 377–389. https://doi.org/10.1007/s10555-013-9436-6 Primo, L. M. F., & Teixeira, L. K. (2020). DNA replication stress: oncogenes in the spotlight. Genetics and Molecular Biology, 43(1), e20190138. https://doi.org/10.1590/1678-4685gmb-2019-0138 Roper, R. J., & Reeves, R. H. (2006). Understanding the Basis for Down Syndrome Phenotypes. PLoS Genetics, 2(3), e50. https://doi.org/10.1371/journal.pgen.0020050 Santaguida, S., Richardson, A., Iyer, D. R., M’Saad, O., Zasadil, L., Knouse, K. A., Wong, Y. L., Rhind, N., Desai, A., & Amon, A. (2017). Chromosome Mis-segregation Generates Cell-Cycle-Arrested Cells with Complex Karyotypes that Are Eliminated by the Immune System. Developmental Cell, 41(6), 638-651.e5. https://doi.org/10.1016/j.devcel.2017.05.022 Santaguida, S., Vasile, E., White, E., & Amon, A. (2015). Aneuploidy-induced cellular stresses limit autophagic degradation. Genes & Development, 29(19), 2010–2021. https://doi.org/10.1101/gad.269118.115 Sheltzer, J. M., Ko, J. H., Replogle, J. M., Burgos, N. C. H., Chung, E. S., Meehl, C. M., Sayles, N. M., Passerini, V., Storchova, Z., & Amon, A. (2017). Single-chromosome Gains Commonly Function as Tumor Suppressors. Cancer Cell, 31(2), 240–255. https://doi.org/10.1016/j.ccell.2016.12.004 Sheltzer, J. M., Torres, E. M., Dunham, M. J., & Amon, A. (2012). Transcriptional consequences of aneuploidy. Proceedings of the National Academy of Sciences, 109(31), 12644–12649. https://doi.org/10.1073/pnas.1209227109 Shirayama, M., Tóth, A., Gálová, M., & Nasmyth, K. (1999). APCCdc20 promotes exit from mitosis by destroying the anaphase inhibitor Pds1 and cyclin Clb5. Nature, 402(6758), 203–207. https://doi.org/10.1038/46080 Simões-Sousa, S., Littler, S., Thompson, S. L., Minshall, P., Whalley, H., Bakker, B., Belkot, K., Moralli, D., Bronder, D., Tighe, A., Spierings, D. C. J., Bah, N., Graham, J., Nelson, L., Green, C. M., Foijer, F., Townsend, P. A., & Taylor, S. S. (2018). The p38α Stress Kinase Suppresses Aneuploidy Tolerance by Inhibiting Hif-1α. Cell Reports, 25(3), 749-760.e6. https://doi.org/10.1016/j.celrep.2018.09.060 30 Solomon, D. A., Kim, T., Diaz-Martinez, L. A., Fair, J., Elkahloun, A. G., Harris, B. T., Toretsky, J. A., Rosenberg, S. A., Shukla, N., Ladanyi, M., Samuels, Y., James, C. D., Yu, H., Kim, J.-S., & Waldman, T. (2011). Mutational Inactivation of STAG2 Causes Aneuploidy in Human Cancer. Science, 333(6045), 1039–1043. https://doi.org/10.1126/science.1203619 Stemmann, O., Zou, H., Gerber, S. A., Gygi, S. P., & Kirschner, M. W. (2001). Dual Inhibition of Sister Chromatid Separation at Metaphase. Cell, 107(6), 715–726. https://doi.org/10.1016/s0092-8674(01)00603-1 Su, X. A., Ma, D., Parsons, J. V., Replogle, J. M., Amatruda, J. F., Whittaker, C. A., Stegmaier, K., & Amon, A. (2021). RAD21 is a driver of chromosome 8 gain in Ewing sarcoma to mitigate replication stress. Genes & Development, 35(7–8), 556–572. https://doi.org/10.1101/gad.345454.120 Sullivan, M., & Morgan, D. O. (2007). Finishing mitosis, one step at a time. Nature Reviews Molecular Cell Biology, 8(11), 894–903. https://doi.org/10.1038/nrm2276 Tang, Y.-C., Williams, B. R., Siegel, J. J., & Amon, A. (2011). Identification of Aneuploidy-Selective Antiproliferation Compounds. Cell, 144(4), 499–512. https://doi.org/10.1016/j.cell.2011.01.017 Thompson, S. L., & Compton, D. A. (2008). Examining the link between chromosomal instability and aneuploidy in human cells. The Journal of Cell Biology, 180(4), 665– 672. https://doi.org/10.1083/jcb.200712029 Torres, E. M., Sokolsky, T., Tucker, C. M., Chan, L. Y., Boselli, M., Dunham, M. J., & Amon, A. (2007). Effects of Aneuploidy on Cellular Physiology and Cell Division in Haploid Yeast. Science, 317(5840), 916–924. https://doi.org/10.1126/science.1142210 Weaver, B. A., & Cleveland, D. W. (2008). The Aneuploidy Paradox in Cell Growth and Tumorigenesis. Cancer Cell, 14(6), 431–433. https://doi.org/10.1016/j.ccr.2008.11.011 Williams, B. R., Prabhu, V. R., Hunter, K. E., Glazier, C. M., Whittaker, C. A., Housman, D. E., & Amon, A. (2008). Aneuploidy Affects Proliferation and Spontaneous Immortalization in Mammalian Cells. Science, 322(5902), 703–709. https://doi.org/10.1126/science.1160058 Zhu, J., Tsai, H.-J., Gordon, M. R., & Li, R. (2018). Cellular Stress Associated with Aneuploidy. Developmental Cell, 44(4), 420–431. https://doi.org/10.1016/j.devcel.2018.02.002 31 32 Chapter 2: Aneuploid senescent cells activate NF-κB to promote their immune clearance by NK cells Reproduced from EMBO Reports: Wang, R. W., Viganò, S., Ben‐David, U., Amon, A., & Santaguida, S. (2021). Aneuploid senescent cells activate NF‐κB to promote their immune clearance by NK cells. EMBO Reports. RWW performed all experiments and analyzed the data with the exception of Figures 6D-E, 8A-B and 11A-B. SV performed the experiments and analyzed the data in Figures 6D-E and 8A-B. UBD performed the computation analysis in Figure 11A-B. 33 ABSTRACT The immune system plays a major role in the protection against cancer. Identifying and characterizing the pathways mediating this immune surveillance is thus critical for understanding how cancer cells are recognized and eliminated. Aneuploidy is a hallmark of cancer and we previously found that untransformed cells that had undergone senescence due to highly abnormal karyotypes are eliminated by natural killer (NK) cells in vitro. However, the mechanisms underlying this process remained elusive. Here, using an in vitro NK cell killing system, we show that non-cell autonomous mechanisms in aneuploid cells predominantly mediate their clearance by NK cells. Our data indicate that in untransformed aneuploid cells, NF-κB signaling upregulation is central to elicit this immune response. Inactivating NF-κB abolishes NK- cell mediated clearance of untransformed aneuploid cells. In cancer cell lines, NF-κB upregulation also correlates with the degree of aneuploidy. However, such upregulation in cancer cells is not sufficient to trigger NK cell-mediated clearance, suggesting that additional mechanisms might be at play during cancer evolution to counteract NF-κB mediated immunogenicity. 34 INTRODUCTION Aneuploidy is defined as a state in which the chromosome number is not a multiple of the haploid complement (Pfau & Amon, 2012). In all organisms analyzed to date, an unbalanced karyotype has detrimental effects (Pfau & Amon, 2012; Santaguida & Amon, 2015). In yeast, aneuploidy leads to proliferative defects and proteotoxic stress (Torres et al, 2010). The impact of aneuploidy on higher eukaryotes is even more severe. Most single autosomal gains and all autosomal losses cause embryonic lethality. Aneuploidies that do survive embryonic development cause significant anatomical and physiological abnormalities (Lorke, 1994; Lindsley et al, 1972; Hassold & Hunt, 2001; Roper & Reeves, 2006). The severe impact of aneuploidy on mammalian physiology is also reflected at the cellular level. Trisomic mouse embryonic fibroblasts (MEFs) and aneuploid human cells proliferate more slowly than their euploid counterparts and experience a variety of cellular stresses (Williams et al, 2008; Pfau et al, 2016; Santaguida et al, 2015; Stingele et al, 2012). Among these, aneuploidy- induced replication stress has been extensively studied. Upon chromosome mis- segregation, cells exhibit slow replication fork progression rate and increased replication fork stalling during the following S phase. Replication stress triggers genomic instability and drives the evolution of highly abnormal karyotypes (Ohashi et al, 2015; Lamm et al, 2016; Sheltzer et al, 2012; Santaguida et al, 2017; Passerini et al, 2016). Although aneuploidy is highly detrimental at both the cellular and organismal level in untransformed cells, it is a hallmark of cancer, a disease characterized by uncontrolled cell proliferation (Gordon et al, 2012). About 90% of solid tumors and 75% of hematopoietic malignancies are characterized by whole chromosome gains and 35 losses (Weaver & Cleveland, 2006). A high degree of aneuploidy is often associated with poor prognosis, immune evasion and a reduced response to immunotherapy (Ben- David & Amon, 2020). Given the negative effects of aneuploidy on primary cells, it remains unclear how cells with severe genomic imbalances could gain tumorigenic potential. Furthermore, which aneuploidy-associated molecular features alter immune recognition during tumor evolution remains an active field of research. By inducing high levels of chromosome mis-segregation followed by continuous culturing, we previously generated cells with abnormal complex karyotypes that eventually cease to divide and enter a senescent-like state. We have named such cells Arrested with Complex Karyotypes (ArCK) cells (Wang et al, 2018; Santaguida et al, 2017). Prior work indicated that ArCK cells upregulate gene expression signatures related to an immune response that render them susceptible to elimination by natural killer (NK) cells in vitro (Santaguida et al, 2017). However, the molecular and functional bases for this immune recognition of ArCK cells remained unclear. Several pathways could be involved in this process. Nuclear factor-kappaB (NF-κB) is induced under several stress conditions to elicit a pro-inflammatory response (Hayden & Ghosh, 2012; Liu et al, 2017). In the canonical NF-κB pathway, stress induction causes IκB kinase complex (IKK) to phosphorylate IκB, thereby marking it for proteolytic degradation (Perkins, 2007). As a result of this degradation, RelA-p50 translocates into the nucleus where it activates expression of pro-inflammatory genes. In the non-canonical NF-κB pathway, phosphorylation and cleavage of p100 triggers the nuclear translocation of the RelB-p52 complex to induce a pro-inflammatory response (Perkins, 2007). Recent studies further suggest that cytosolic nucleic acids lead to cGAS/STING activation in 36 senescent cells, which induces an interferon response via JAK-STAT signaling pathway (Glück et al, 2017). Here we investigate which innate immune pathways contribute to NK cell- mediated elimination of aneuploid cells and show that the NF-κB pathway elicits pro- inflammatory signals in ArCK cells. Inactivating both canonical and non-canonical NF- κB pathways in cells with an unbalanced karyotype prevents NK cell-mediated killing in vitro. Furthermore, we find that the NF-κB signature is upregulated in cancer cell lines possessing a higher degree of aneuploidy. However, this activation no longer enhanced NK cell-mediated killing of cancer cells, raising the possibility that aneuploidy-induced immunogenicity might be present only at the early stage of tumorigenesis and aneuploid cancer cells evolve mechanisms to evade immune clearance. RESULTS An assay to assess elimination of ArCK cells by natural killer (NK) cells in vitro To address the molecular basis for immune recognition of ArCK cells, we established a co-culture system to monitor the interactions between NK cells and ArCK cells. In this setup, we utilized human, untransformed RPE1-hTERT cells in which chromosome segregation errors were forced by inhibiting the function of the spindle assembly checkpoint (SAC) (Santaguida et al, 2010). To generate ArCK cells, we synchronized RPE1-hTERT cells at the G1/S boundary and released them into the cell cycle in the presence of the SAC kinase Mps1 inhibitor reversine (Fig 1A). We removed the drug once the cells had undergone one round of aberrant mitosis due to SAC inhibition. 72 hours after inducing chromosome mis-segregation, we exposed cells to 37 the spindle poison nocodazole for 12 hours, which allowed us to remove dividing cells by mitotic shake-off (Wang et al, 2018; Santaguida et al, 2017). We repeated the mitotic shake-off 4 more times to ensure the removal of all cycling cells. Cells that remained on the tissue culture plate by the end of this procedure were highly enriched for the ArCK population [Fig 1A (Wang et al, 2018)]. Importantly, such cell cycle arrest was not due to the prolonged nocodazole treatment since euploid control cells were completely removed after two consecutive rounds of shake-offs (Santaguida et al, 2017). We co- cultured ArCK cells with an immortalized NK cell line activated by constitutive IL2 expression [NK92-MI (Tam et al, 1999)] and monitored their interactions by live cell imaging (Fig 1B). 38 39 Figure 1. ArCK cells are recognized by natural killer (NK) cells in vitro. A. Schematic representation for the generation of ArCK cells. Time 0 is defined by the estimated onset of Mps1 inhibitor-induced chromosome mis-segregation. B. Representative images of euploid control or ArCK cells interacting with NK cells. The NK cell-mediated killing was measured at a 2.5:1 (NK cells: target cells) ratio and was recorded by live cell imaging for 36 hours at a 30-min interval. TO-PRO-3 (1 μM) was added to the medium at the same time of NK cell addition to measure cell membrane integrity. Phase contrast (top) and TO-PRO-3 signal (bottom) from the same field were presented. Arrowheads indicate ArCK cell death. All images were acquired at the same exposure time and light intensity. Scale bar 20 μM. C. Measurement of NK cell-mediated killing of ArCK and euploid control cells (Ctrl) at a 2.5:1 (NK cells: target cells) ratio. 50 randomly chosen target cells were followed for 36 hours by live cell imaging per condition per replicate. The cumulative cell death was calculated. n=3 biological replicates; mean± SEM. The statistical significance was determined using nonparametric Kolmogorov-Smirnov test (KS test) as described in the method section; p< 0.0001. D. Measurement of euploid control (Ctrl) and ArCK cell proliferation without NK cells. Live cell imaging of target cells without NK cells was performed using the same condition as described in the method section. For each condition, 50 cells were randomly chosen at the beginning of the movie as the initial population (indicated by the dashline, ninitial = 50). The cumulative cell number was recorded. Dot plot of individual data points and mean were presented; n=2 biological replicates. E. NK cell-mediated cytotoxicity across various NK cell to target cell ratios. Either euploid control (Ctrl) or ArCK cells were co-cultured with NK cells at the indicated NK cell: target cell ratios. The cumulative killing of target cells was measured. n=3 biological replicates; mean± SEM. p< 0.0001 for all four NK cell: target cell ratios, KS test. To quantify the degree of NK cell killing of target cells, we first defined a killing event as a target cell that was (1) engaged by one or multiple NK cells, and (2) lifted from the tissue culture plate (Fig 1B). We chose these criteria because they coincided with target cell membrane permeabilization as judged by the ability of the nucleic acid dye TO-PRO3 to enter a cell (Fig 1B and 2A). We tracked individual target cells and recorded the time when each of them was killed during the 36-hour live cell imaging. If a target cell divided during the time course, we followed only one of the resulting two cells for the remainder of the assay. We then calculated the cumulative cell death for each condition and generated killing curves at hourly resolution. We found that at a ratio of 40 2.5 NK cells to 1 target cell, ArCK cells were consistently killed twice as effectively as euploid control cells, during a 36-hour co-culture experiment (Fig 1C). ArCK cells hardly divided during the 36-hour time lapse employed in our NK cell killing assay whereas euploid control cells continued to divide (Fig 1D). It was thus possible that the difference in NK cell-mediated cytotoxicity towards euploid and ArCK cells was affected by the fact that NK cells became limiting when co-cultured with euploid cells but not aneuploid cells. To test this possibility, we analyzed the effect of changing the NK cell to target cell ratio. We found that even at high NK cell to target cell ratio (5:1 and 10:1), ArCK cells were still more effectively killed than euploid controls (Fig 1E). We conclude that NK cells were not limited in our assay. We further note that when a cell divided during observation, we followed only one of the two cells after cell division, which corrected for the bias in target cell number. To address the possibility that NK cells became exhausted during the course of the co-culture experiment, we divided the 36-hour assay into two time courses, where the same population of NK cells was consecutively co-cultured with target cells for 18 hours each. NK cells were equally effective in killing the target cells (Fig 2B) in this experimental setup, indicating that NK cell exhaustion did not occur within the time course of the analysis. We propose that the eventual plateauing of the killing curve as the assay proceeds is likely due to NK cells taking longer to find their targets. We next set out to test why euploid control cells are readily killed by NK cells in our in vitro assay. One possible explanation was that RPE1-hTERT cells express human telomerase reverse transcriptase (hTERT) and harbor a KRAS mutation (Nicolantonio et al, 2008), which could generate oncogenic transformation-associated 41 NK cell stimulatory signals (Chiossone et al, 2018; Shimasaki et al, 2020). To test this possibility, we assessed NK cell-mediated cytotoxicity across three different types of early passage euploid primary fibroblasts derived from normal donors (human embryonic lung fibroblast, IMR90, and normal neonatal or adult human dermal fibroblasts, NHDF-Neo or NHDF-Ad). The analysis of these primary cells revealed large variations in both cell proliferation and NK cell-mediated killing (Fig 2C and D). Adult human dermal fibroblasts were not readily eliminated by NK cells whereas both neonatal human dermal fibroblasts and IMR90 cells were highly immunogenic. Thus, it appears that NK cell-mediated killing differs significantly between primary cultured cells. Importantly, we also observed a consistent two-fold increase in killing on the Mps1 inhibitor reversine-induced aneuploid NHDF-Ad cells compared to their euploid controls (Fig 2E), indicating that NK cell-mediated immune clearance of aneuploid cells is not a cell type specific phenotype. We conclude that in the assay we developed here, highly aneuploid RPE1-hTERT cells are more effectively recognized and eliminated by NK92- MI cells in vitro than their euploid counterparts. Since we had developed robust protocols to generate the aneuploid cell population using RPE1-hTERT cells (Wang et al, 2018; Santaguida et al, 2017), we decided to focus on this cell line to investigate the effects of karyotype alterations on NK cell-mediated immune clearance. 42 Figure 2. Characterization of the NK cell killing assay. A. Side by side comparison analyzing NK cell-mediated killing on euploid control or ArCK cells by phase contrast image (Phase) or TO-PRO3 signal. Cells were cultured as described in Figure 1A. Statistical analyses were performed as in Figure 1C. Individual points and mean were presented; n=2 biological replicates. Ctrl-Phase vs. Ctrl-TOPRO3, p= 0.89, n.s.; ArCK-Phase vs ArCK-TOPRO3, p= 0.89, n.s.; KS test. B. Measurement of NK cell-mediated killing of ArCK cells in two consecutive 18h time lapse experiments. After the first 18 hours of the analysis, the cell suspension was collected and co-cultured with a second set of target cells. NK cell-mediated killing was measured in the first (black and dark red curves) and the second (grey and light red curves) 18h time lapse and plotted on the same graph. The killing assay was performed at a NK cell to target cell ratio of 2.5:1 (left panel) and 5:1 (right panel); n=2 biological replicates. NK:Target= 2.5:1, Ctrl-1st movie vs. Ctrl-2nd movie, p= 1.00, n.s.; ArCK-1st movie vs. ArCK-2nd movie, p= 0.21, n.s.; NK:Target = 5:1, Ctrl- 1st movie vs. Ctrl-2nd movie, p= 0.28, n.s.; ArCK-1st movie vs. ArCK-2nd movie, p= 0.97, n.s.; KS test. C. Cell proliferation measurements in the absence of NK cells. RPE1-hTERT (passage 4), human normal neonatal or adult human dermal fibroblasts (NHDF-Neo, passage 5 or NHDF-Ad, passage 5), and human embryonic lung fibroblast (IMR90, passage 3) were plated side by side in NK cell medium and cell proliferation rate was recorded using live cell imaging as described in Figure 1D. The dashed line indicates the starting cell number (ninitial=50). Dot plot of individual data points and mean are shown; n=2 biological replicates. D. NK cell-mediated cytotoxicity across different cell types. The killing of RPE1-hTERT, human normal neonatal or adult human dermal fibroblasts (NHDF-Neo or NHDF- Ad), and human embryonic lung fibroblast (IMR90) were measured as described in Figure 1 using a NK cell to target cell ratio of 2.5 to 1; n=2 biological replicates. Individual data and mean are shown. 43 E. Human normal adult dermal fibroblasts (NHDF-Ad) were treated with either DMSO or the Mps1 inhibitor reversine (500 nM) for 24 hours. Drugs were washed out and NK cell-mediated killing was compared between DMSO treated (NHDF-Ad Ctrl) and Mps1 inhibitor treated (NHDF-Ad Mps1 inhibitor) cells as described in figure 1C; n=2 biological replicates. NHDF-Ad Ctrl vs. NHDF-Ad Mps1 inhibitor, p= 0.001; KS test. Prolonged cell cycle arrest associated with features of senescence elicits NK cell- mediated cytotoxicity ArCK cells are largely arrested in G1 and exhibit features of senescence (Santaguida et al, 2017; Wang et al, 2018). Permanent cell cycle arrest has been shown to elicit an immune response (Gorgoulis et al, 2019). To determine whether G1 arrest per se is sufficient to cause immune recognition, we assessed NK cell-mediated cytotoxicity towards G1 arrested cells induced by three different methods. We treated RPE1-hTERT cells for 7 days with 1) the topoisomerase II inhibitor, doxorubicin, to induce high levels of DNA damage (Pommier et al, 2010); 2) the cyclin-dependent kinases CDK4/6 inhibitor, palbociclib; or 3) the imidazoline analog, nutlin3, to disrupt the interaction between p53 and its ubiquitin ligase Mdm2 thereby stabilizing p53. All three conditions have been shown to cause features associated with cellular senescence (Sliwinska et al, 2009; Oliveira & Bernards, 2018; Wiley et al, 2018). DNA content analysis by flow cytometry and EdU incorporation showed that after 7 days, all 3 treatments caused the cells to arrest in G1 (Fig 3A-C). With the exception of nutlin3 treatment, these G1 arrests were irreversible: most cells did not resume proliferation following drug washout as judged by cell proliferation assays (Fig 3D). Co- culturing these G1-arrested cells with NK cells revealed that irrespective of the means by which the arrest was induced, NK cells exhibited a two-fold increase in killing on these G1-arrested cells compared to the untreated proliferating control cells (Fig 3E). 44 Inactivation of the TORC1 pathway also causes cell cycle arrest (Sousa-Victor et al, 2015), but cells enter a quiescent state instead of senescence (Kucheryavenko et al, 2019; Sousa-Victor et al, 2015). RPE1-hTERT cells were mostly arrested in cell cycle upon treatment with 1 μM of the mTOR kinase inhibitor torin1 after 7 days (Fig 3F). Yet NK cell recognition and killing was not enhanced in cells treated with torin1 (Fig 3G). We conclude that G1 arrest in target cells contributes to NK cell engagement, but only when accompanied by features of senescence. 45 Figure 3. Prolonged cell cycle arrest associated with features of senescence elicits NK cell-mediated cytotoxicity A. DNA content analysis of various G1 arrests. RPE1-hTERT cells were treated for 7 days with doxorubicin (Doxo; 100 ng/ml), palbociclib (Palbo; 5 μM), or nutlin3 (Nutlin; 10 μM). Total number of cells analyzed is indicated by n in each condition. Results were comparable between 2 biological replicates. B. Schematics of EdU incorporation assay. Drugs were applied to RPE1-hTERT cells 12h after initial cell plating. 6 days later (144h), cells were switched to drug medium containing 5-ethynyl-2’-deoxyuridine (EdU; 10 μM) for 24 hours before fixation and analysis. C. The percentage of EdU positive cells after doxorubicin (Doxo), palbociclib (Palbo), or nutlin3 (Nutlin) treatment. EdU incorporation was performed as described in (B). At least 100 cells were analyzed per condition per replicate. Individual data points and mean are shown; n=2 biological replicates. D. Cell proliferation (in the absence of NK cells) after 7 days of doxorubicin (Doxo), palbociclib (Palbo), or nutlin3 (Nutlin) treatment. The drugs were washed out after 7 days and the cells were re-plated for live cell imaging. Cell proliferation was 46 measured as described in Figure 1D. The dashed line indicates the starting cell number (ninitial = 50). Dot plot of individual data points and mean are shown; n=2 biological replicates. E. NK cell-mediated killing for doxorubicin (Doxo), palbociclib (Palbo), and nutlin3 (Nutlin) treated samples (NK cell: target cell =2.5:1). n=3 biological replicates; mean± SEM. Ctrl vs. Doxo, p< 0.0001; Ctrl vs. Palbo, p< 0.0001; Ctrl vs. Nutlin, p< 0.0001; KS test. F. The percentage of EdU positive cells after 7 days of torin1 treatment was determined as described in (B) and (C). Individual data points and mean are shown; n=2 biological replicates. G. NK cell-mediated cytotoxicity towards torin1-treated cells. Torin1 treated (Torin) cells were generated as described in (F) and the NK cell killing assay was performed as described in Figure 1. n=3 biological replicates; mean± SEM. Ctrl vs. Torin, p= 0.79, not significant (n.s.); KS test. Mechanisms triggering senescence contribute to NK cell recognition in ArCK cells The observation that senescence triggered by multiple mechanisms led to NK cell recognition begged the question of what features in aneuploid cells elicit NK cell- mediated clearance. To address this, we compared a collection of cellular markers contributing to senescence in ArCK cells to those of cells treated with doxorubicin, palbociclib or nutlin3 for 7 days. First, we assessed DNA damage levels across all conditions by measuring nuclear γ-H2AX foci (Fig 4A and 5A). DNA damage can increase the expression of NK cell activating receptor (NKG2D) ligands such as MICA and ULBP2, thereby triggering NK cell-mediated clearance (Raulet & Guerra, 2009). In untreated proliferating control cells, more than 80% of the cells harbored fewer than 10 γ-H2AX foci per nucleus. As expected, doxorubicin caused significantly higher levels of DNA damage in the euploid cells, such that approximately 90% of the cells displayed more than 20 foci and ~50% of this population had 50 foci or more (Fig 4A, panel 2). In contrast, the DNA damage levels in palbociclib and nutlin3 treated cells were 47 comparable to untreated control cells (Fig 4A, panel 3 and 4). About one third of the ArCK cells harbored more than 10 foci (Fig 4A, panel 5), likely caused by replication stress and/or endogenous reactive oxygen species (ROS) associated with aneuploidy (Santaguida et al, 2017; Li et al, 2010; Passerini et al, 2016). Induction of the DNA damage response genes p53 and p21 agreed with the presence of γ-H2AX foci with the obvious exception of nutlin3 treated cells (as nutlin3 inhibits Mdm2 to stabilize p53 but does not cause endogenous DNA damage; Fig 4B). We also assessed senescence-associated beta-galactosidase activity (SA-beta-gal), a biomarker frequently used to assess degree of senescence. Over 80% of beta-gal positive cells were observed in doxorubicin-treated and ArCK cells. Palbociclib- or nutlin3-treated cells exhibited a milder increase in the levels of SA-beta-gal positive cells (~30%), whereas torin1- treated quiescent cells did not show a significant increase in the proportion of SA-beta-gal compared to the untreated control cells (Fig 4C and 5B). We then examined the senescence associated secretory phenotype (SASP), which plays a critical role in immune cell recruitment (Gorgoulis et al, 2019). We found the composition of SASP varied between different G1 arrests (Fig 4D and 5C). Nevertheless, arrested cells that did elicit NK cell-mediated killing all secreted a plethora of chemokines and cytokines including factors contributing to NK cell recognition [e.g., CCL2 (Robertson, 2002)], whereas the secretion in torin1- treated cells remained low. We conclude that ArCK cells attract NK cells, at least in part, by expressing a canonical senescence immune recognition program. 48 49 Figure 4. Mechanisms triggering senescence contribute to NK cell recognition in ArCK cells A. -H2AX foci were analyzed in ArCK and G1 arrested cells (generated as described in Figure 3). At least 50 cells were analyzed per condition per replicate. n=2 biological replicates; individual values and mean are shown. The distribution of the foci number in each treated condition was compared to that of control using Kolmogorov-Smirnov test. Ctrl vs. Doxo, p< 0.0001; Ctrl vs. Palbo, p= 0.07, n.s.; Ctrl vs. Nutlin, p= 0.11, n.s.; Ctrl vs. ArCK, p= 0.0007. B. p53 and p21 levels were determined by western blot analysis. Vinculin was used as loading control. Results were comparable between 2 biological replicates. C. The degree of senescence was measured by senescence-associated - galactosidase (-gal) activity. The graph shows the percentage of -gal positive cells. At least 100 cells were analyzed per condition per replicate. n=3 biological replicates; mean ± SEM. Ctrl vs. Doxo, ***p= 0.0006; Ctrl vs. Palbo, *p= 0.025; Ctrl vs. Nutlin, *p= 0.016; Ctrl vs. Torin, p= 0.806, n.s.; Ctrl vs. ArCK, ***p= 0.0001; unpaired t-test. D. Secreted cytokine and interferon levels were determined in cell supernatants. Media were collected after 36 hours of incubation with cells grown as described in Figure 1 and 2. Cytokine and interferon levels were shown as fold change normalized to euploid control cells. Individual values and mean are shown; n=2 biological replicates. E. NK cell medium was incubated with either euploid control or ArCK cells for 12 hours. At the time of NK cell addition, media were switched between ArCK and euploid control cells (Ctrl). NK cell killing was measured as described in Figure 1C. For reference, NK cell killing of ArCK and euploid control cells (Ctrl) without medium switch were performed side by side and plotted on the graph. Black, euploid control cells without medium switch; red, ArCK cells without medium switch; blue, euploid control cells in ArCK cell condition medium; green, ArCK cells in euploid control cell condition medium. n=3 biological replicates; mean± SEM. Ctrl vs. ArCK, p< 0.0001; Ctrl vs. Ctrl in ArCK med, p< 0.0001; Ctrl vs. ArCK in Ctrl med, p< 0.0001; ArCK vs. ArCK in Ctrl med, p= 0.0002; KS test. To determine the role of secreted factors in NK cell-mediated killing on aneuploid cells, we collected the culture medium from aneuploid cells and applied it to euploid control cells (Fig 4E). Medium previously used to culture ArCK cells for 12 hours increased NK cell-mediated cytotoxicity towards euploid cells by ~1.5 fold. However, this conditioned medium switch did not completely abolish the differences in NK cell- mediated killing between aneuploid cells and their euploid counter parts. NK cells were still more efficient at killing ArCK cells than euploid cells that have been cultured in pre- 50 conditioned medium from aneuploid cells (Fig 4E). This could be due to the fact that secreted factors accumulated to higher levels during the 36-hour live cell imaging, and/or the possibility that ArCK cell surface features enable their elimination by NK cells. We conclude that conditioned medium provides one or more factors that upregulate NK cell-mediated killing and that aneuploid cells could generate both cell autonomous and non-cell autonomous signals that render them susceptible to NK cell- mediated cytotoxicity. 51 Figure 5: Characterization of aneuploid and G1-arrested cells. A. Representative images of -H2AX staining in the indicated samples. -H2AX is in red and DNA in blue. Scale bar 10 μm. B. Representative image of senescence-associated -galactosidase staining in the indicated samples. Scale bar 100 μm. C. Analysis of all cytokines secreted by the indicated cells. Cytokine levels are shown as fold change of euploid control cells; n=2 biological replicates. Individual values and mean are plotted. D. Gene set enrichment analysis (GSEA) for doxorubicin (Doxo), palbociclib (Palbo), nutlin3 (Nutlin), torin1 (Torin)- treated and ArCK cells relative to euploid proliferating control cells. Only the top 10 ranked hallmarks are presented in Doxo, Palbo and ArCK conditions. The normalized enrichment score (NES) are plotted. The numbers 52 on the NES score bar indicate the corresponding p-values for each hallmark (FDR q value≤0.05). NF-κB and interferon-mediated pathways are upregulated in ArCK cells Based on our notion that both secreted factors and cell surface features of aneuploid cells contribute to NK cell-mediated killing, we next aimed to determine how NK cell recognition is induced in aneuploid cells. For this, we profiled the gene expression signature of ArCK cells and compared it to those of doxorubicin, palbociclib, nutlin3, and torin1-treated cells by RNA sequencing. Based on gene set enrichment analysis (GSEA), we found there were common significantly upregulated hallmarks shared by G1 arrests that were associated with features of senescence. For example, the p53 pathway is highly upregulated in ArCK, doxorubicin, palbociclib, and nutlin3- treated cells, but is absent in the torin1-treated quiescent cells (Fig 6A). The hallmark gene set “TNFalpha– via NF-κB signaling” was among the most upregulated pathways in aneuploid and doxorubicin-treated samples (Fig 5D and 6A). ArCK cells also exhibited increased expression of the interferon alpha and gamma response, immune complement, JAK-STAT, and interleukin (IL) related pathways (Fig 6A and B). In palbociclib and nutlin3- treated samples, we did observe a mild upregulation of immune related signatures, but none of them reached significance (FDR q value ≤ 0.05; Fig 5D and 6A). Interestingly, even though NF-κB signaling upregulation was also significant in torin1-treated cells, they were not recognized by NK cells, suggesting NF-κB activation alone in quiescent cells is not sufficient to cause NK cell-mediated cytotoxicity (Fig 5D and 6A). 53 Figure 6. NF-κB pathway is activated in ArCK cells. A. Significantly differentially expressed hallmarks in ArCK cells compared to euploid control cells are shown in the first column in the heatmap (FDR q value≤0.05). Normalized enrichment scores are plotted. The corresponding NES for these hallmarks in doxorubicin (Doxo), palbociclib (Palbo), nutlin3 (Nutlin) and torin1 (Torin) treated cells are also plotted. Hallmarks did not reach statistical significance (FDR q value>0.05) are shown in grey. B. Enrichment plots for “Interferon alpha response”, “TNFalpha signaling via NFKB” and “Interferon gamma response” hallmark signatures in ArCK cells compared to the euploid control cells. C. Significantly differentially expressed RELA and RELB target genes in ArCK cells compared to euploid control cells were identified by ingenuity pathway analysis based on RNA sequencing data (Log2 fold change, p value≤0.05). D. RT-qPCR quantifying NF-κB downstream target gene expression in ArCK and euploid control (Ctrl) cells. n=6 biological replicates; mean± SEM. IL-1b, *p= 0.029; IL-6, *p= 0.032; IL-8, **p= 0.008; unpaired t-test. E. Measurement of NF-κB activity with NF-κB alkaline phosphatase (SEAP) reporter assay. The reporter was expressed in RPE1-hTERT cells by transient transfection. 10 hours after transfection, cells were treated with either DMSO (Ctrl) or reversine (500 nM; Aneuploid senescent) for 96 hours and the secretion of alkaline phosphatase in the culture supernatant was measured. The secretion level was normalized to cell number for each condition. n=3 biological replicates; mean± SEM; unpaired t-test, ****p<0.0001. 54 Given the importance of NF-κB pathway in mediating immune recognition, we further characterized this pathway in aneuploid cells. RNA-seq analysis revealed that both RelA and RelB target genes were significantly upregulated in ArCK cells and this was further confirmed by RT-qPCR analysis (Fig 6C and D). This suggested a possible role of both canonical and non-canonical NF-κB pathway. To substantiate NF-κB activation in aneuploid cells, we employed an NF-κB reporter assay in which the expression and secretion of alkaline phosphatase (AP) is controlled by an NF-κB regulatory element (Signorino et al, 2014). Using this assay, we observed a three-fold increase in AP secretion 96 hours post reversine-induced chromosome mis-segregation (this would allow for the collection of growth medium at a time point reachable from chromosome mis-segregation without cell splitting or the loss of the AP conditioned medium, but also reasonably close to the time of ArCK collection, Fig 6E). Furthermore, we also observed a significant increase in the nuclear translocation of RelA in ArCK cells (Fig 8A-B), in agreement with previous reports showing NF-κB activation following chromosome mis-segregation (Vasudevan et al, 2020). Altogether, these results indicate that NF-κB pathway is indeed upregulated in ArCK cells. 55 Figure 7. Both canonical and non-canonical NF-κB pathway are required for NK cell-mediated killing of ArCK cells A. NK cell-mediated killing of RELA ArCK cells was compared to clones harboring an empty vector. ArCK RELA knock out cells were generated and the NK cell-mediated cytotoxicity was measured as described in the method section. Dot plot of individual data points and mean were presented. n=2 biological replicates; ArCK-c1 vs ArCK RELA-c1, p= 0.70, n.s.; KS test. B. Measurement of RelA protein levels in RELA KO single cell clones generated in RPE1-hTERT cells. C. The effect of inactivating RELB on NK cell-mediated cytotoxicity in ArCK cells. The same experimental methods were used as described in (A). n=3 biological replicates; ArCK-c1 vs ArCK RELB-c1, p= 0.47, n.s.; KS test. D. Measurement of RelB protein levels in RELB KO single cell clones generated in RPE1-hTERT cells. E. The effect of inactivating both RELA and RELB on NK cell-mediated cytotoxicity in ArCK cells. n=2 biological replicates; ArCK-c1 vs ArCK RELA RELB- c1, p= 0.0014; KS test. F. Measurement of RelA and RelB protein levels in RELA RELB double KO single cell clones generated in RPE1-hTERT cells. G. ArCK or euploid proliferating control cells were treated with either DMSO or the NF- κB inhibitor BMS-345541 (5 μM) for 48 hours before assessing NK cell-mediated cytotoxicity. The drug was washed out during the NK cell co-culture assay. n=3 56 biological replicates; mean± SEM. ArCK vs ArCK NF-κB inhibitor, p< 0.0001; KS test. H, I. The effect of inactivating both RELA and RELB on NK cell-mediated cytotoxicity in 7-day doxorubicin (H) and nutlin3-treated (I) cells. n≥2 biological replicates; Doxo- c1 vs. Doxo RELA RELB-c1, p = 0.02. Nutlin-c1 vs. Nutlin RELA RELB-c1, p = 0.44, n.s.; KS test. Both canonical and non-canonical NF-κB pathways are required for NK cell- mediated killing of ArCK cells What is the relevance of the NF-κB pathway activation in NK cell-mediated elimination of aneuploid cells? To address this question, we generated RELA and RELB single and double KO cell lines using CRISPR-Cas9 method. In most of the RELA or RELB single knock out clones, we did not observe a significant decrease in NK cell- mediated cytotoxicity towards ArCK cells (Fig 7A-D and 8C-D). However, when we knocked out both RELA and RELB, NK cell-mediated killing in ArCK cells was significantly reduced to a level comparable to the killing of proliferating euploid controls (Fig 7E-F and 8E). Similar results were observed when ArCK cells were treated with a NF-κB inhibitor BMS-345541 that interferes with both catalytic subunits of IKK [albeit with different binding affinities (Yang et al, 2006; Burke et al, 2003)] to block NF-κB activation (Fig 7G and 8F). Importantly, the NF-κB inhibitor treatment in other high immunogenic primary cells (NHDF-Neo) did not lead to a reduction in NK cell-mediated killing (Fig 8G), suggesting that NF-κB pathway activation in ArCK cells is caused by features associated with aneuploidy induction. We further examined the effect of RELA RELB depletion on NK-cell mediated killing in senescent cells triggered by other mechanisms. Based on GSEA analysis, doxorubicin-treated cells also elicit a NF-κB signature (Fig 6A). Indeed, inactivating NF-κB pathway in doxorubicin treated cells led 57 to a modest but significant decrease in NK cell-mediated killing (Fig 7H and 8H). We conclude that DNA damage downstream of aneuploidy could be involved, at least in part, in eliciting the immune response in ArCK cells. On the other hand, RELA and RELB depletion did not rescue NK cell-mediated killing in nutlin3-treated cells, which did not exhibit NF-κB signature (Fig 6A, 7I and 8I). This suggests that multiple pathways could be involved in eliciting the immune response, most likely depending on the mechanism leading to cellular senescence. Together, we conclude that aneuploidy triggers NF-κB activation in primary untransformed cells, which contributes to their recognition and elimination by NK cells. 58 Figure 8. NF-κB pathway contributes to NK cell-mediated killing in ArCK cells. A-B. Representative images (A) and quantification (B) of RelA nuclear translocation signal in ArCK and euploid control cells. At least 86 cells were analyzed per condition per replicate. n=3 biological replicates; mean±SEM. **p= 0.0063; unpaired t-test. Scale bar 10 μm. C-E. The effect of inactivating RELA (C), RELB (D), and both RELA and RELB (E) on NK cell-mediated cytotoxicity in ArCK cells. The experiment was performed as described in Figure 7, except a different single cell clone was used. ArCK-c1 vs ArCK RELA-c2, p= 0.28, n.s.; ArCK-c1 vs ArCK RELB-c2, p= 0.72, n.s.; ArCK-c1 vs ArCK RELB-c3, p= 0.02; ArCK-c2 vs ArCK RELA RELB-c2, p= 0.0004; KS test. The same controls are plotted in (E) and Figure 7E since all three replicates for both RELA RELB KO clones were performed side by side at the same time. F. RPE1-hTERT cells were treated with either DMSO or the NF-κB inhibitor BMS- 345541 (5 μM) for 48 hours. TNFα (100 ng/mL) was added to cells in both conditions for 1 hour prior to cell fixation. RelA and RelB nuclear translocation was quantified and shown. n=3 biological replicates; mean±SEM. RelA, TNFα vs. TNFα NF-κB inhibitor, ****p<0.0001; RelB, TNFα vs. TNFα NF-κB inhibitor, **p=0.0097; unpaired t-test. G. Human normal neonatal dermal fibroblasts (NHDF-Neo) were treated with either DMSO or the NF-κB inhibitor BMS-345541 (5 μM) for 48 hours. The drug was washed 59 out and NK cell-mediated killing was assessed as described in Figure 1C. Dot plot of individual data points and mean are presented in all of the NK cell killing assays; n=2 biological replicates. NHDF-Neo Ctrl vs NHDF-Neo NF-κB inhibitor, p= 0.28, n.s.; KS test. H-I. The effect of inactivating both RELA and RELB on NK cell-mediated cytotoxicity in 7-day doxorubicin (H) and nutlin3-treated (I) cells. The experiment was performed as described in Figure 7, except a different single cell clone was used. n= 3 biological replicates; Doxo-c1 vs. Doxo RELA RELB-c2, p = 0.04. Nutlin-c1 vs. Nutlin RELA RELB-c2, p = 0.18, n.s.; KS test. Retrotransposon activation is involved in triggering immune clearance of ArCK cells ArCK cells also induced interferon alpha and gamma responses as judged by RNA-seq and RT-qPCR analysis (Fig 9A and B). Alpha and gamma interferon responses are primarily mediated by the JAK-STAT pathway (Villarino et al, 2017). We confirmed that activation of these two gene expression signatures was indeed mediated by the JAK-STAT pathway as inactivation of STAT1 by CRISPR-Cas9 reduced both the interferon alpha and interferon gamma responses in ArCK cells (Fig 9C and D). To determine the biological relevance of the JAK-STAT response, we depleted STAT1 in ArCK cells. Our data indicate that deletion of STAT1 alone is not sufficient to significantly affect NK cell-mediated immune clearance in ArCK cells (Fig 9E). We speculate that JAK-STAT pathway activation is not essential, but perhaps, potentiates aneuploid cells for immune recognition by NK cells. Remarkably, our data suggest that, although multiple pathways might be activated in aneuploid cells, the NF-κB pathway plays a major role in NK-cell mediated immunogenicity of ArCK cells. 60 Figure 9. STAT1-responsive facilitates immune recognition of ArCK cells. A. Significantly differentially expressed STAT1 target genes in ArCK cells compared to euploid control cells were identified by ingenuity pathway analysis based on RNA sequencing data. (Log2 fold change, p value≤0.05). B. Measurement of STAT1 mRNA levels in ArCK cells normalized to euploid control cells. n=3 biological replicates; mean±SEM. **p= 0.0025; unpaired t-test. C. Measurement of STAT1 protein levels in STAT1 KO single cell clones generated in RPE1-hTERT cells. D. GSEA enrichment plot for interferon alpha and interferon gamma hallmarks in ArCK cells generated in either control or STAT1 KO RPE1-hTERT cells. Two single cell cloned control cell lines and three single cell cloned STAT1 KO cell lines (shown in C) were used in the RNA-seq analysis. E. ArCK cells lacking STAT1 were generated as described in the method section. NK cell-mediated cell death in STAT1 ArCK cells was compared to control cells which harbored an empty vector. n=4 biological replicates; mean±SEM. ArCK-c1 vs. ArCK STAT1-c1, p= 0.7, n.s.; ArCK-c1 vs. ArCK STAT1-c2, p= 0.05, n.s.; ArCK-c1 vs. ArCK STAT1-c3, p= 0.14, n.s.; KS test. Chromosome mis-segregation also generates micronuclei (Liu et al, 2018; Crasta 61 et al, 2012; Martin & Santaguida, 2020; Janssen et al, 2011). The nuclear envelope of these micronuclei is unstable, leading to their frequent rupture. This in turn causes DNA to spill into the cytoplasm, which activates the cGAS-STING pathway (Mackenzie et al, 2017; Dou et al, 2017; Harding et al, 2017; Bakhoum et al, 2018). cGAS-STING could lead to the induction of the NF-κB and interferon response (Dunphy et al, 2018). However, in our experimental set up, we did not observe a significant cGAS-STING pathway activation in reversine-induced aneuploid cells as judged by the low degree of IRF3 phosphorylation (Fig 10A). Furthermore, STING knock out did not affect NK cell- mediated killing in ArCK cells (Fig 10B). Together, these data suggest that, at least in our in vitro assay, NF-κB is the predominant pathway by which aneuploidy induces NK cell-mediated cytotoxicity. Senescence is also able to induce retrotransposon activation (Cecco et al, 2019). This could generate dsRNA intermediates to induce both NF-κB and interferon signatures (Alexopoulou et al, 2001; Zamanian-Daryoush et al, 2000). Since ArCK cells are senescent (Wang et al, 2018; Santaguida et al, 2017), we tested whether they might be able to trigger retrotransposon activation. We found that in aneuploid cells, two dsRNA sensors, Rig-I (DDX58) and Mda5 (IFIH1) were upregulated at the mRNA level compared to their euploid control cells (Fig 10C). Furthermore, ORF1p, one of the proteins encoded by the LINE-1 retrotransposon, was mildly induced in ArCK cells (Fig 10D). Thus, up-regulation of retrotransposons could be relevant to NK cell recognition of aneuploid cells. To suppress retrotransposition, we inhibited reverse transcription by treating cells with the cytosine analog 2’,3’-dideoxy-3’-thiacytidine (3TC). We found suppressing retrotransposition led to a partial reduction in NK cell-mediated cytotoxicity 62 towards ArCK cells (Fig 10E). We conclude that retrotransposon activation in aneuploid cells might be involved, at least partially, in NF-κB upregulation for NK cell-mediated immune clearance in ArCK cells. 63 Figure 10. Analysis of cGAS-STING and retrotransposon activity in ArCK cells and NF-κB activation in cancer cells following chromosome mis-segregation A. IRF3 and phospho-IRF3 levels were analyzed by western blot in euploid proliferating cells (Ctrl), cells 60h post reversine treatment (Rev60), or ArCK cells that are either functional for STING (lanes 1-3) or lacking the gene (lanes 4-6). RPE1-hTERT cells treated with cGAMP (10 μg/ml) for 24hrs were used as a positive control (lanes 7 and 8). To confirm the specificity of the phospho IRF3-antibody, protein lysates from cGAMP treated cells were incubated with lambda phosphatase (lane 8). 64 B. ArCK STING KO cells were generated and NK cell-mediated cytotoxicity was compared to control cell lines that harbor an empty vector as described in Figure 1C; n=2 biological replicates. ArCK vs. STING ArCK, p= 0.91, n.s.; KS test. C. Measurement of DDX58 and IFIH1 mRNA levels in ArCK cells by RT-qPCR shown as fold change compared to euploid control cells. RPE1-hTERT cells treated with PolyIC (10 μg/mL) for 24h were used as a positive control. n=4 biological replicates; mean±SEM. DDX58, Ctrl vs. ArCK, ***p= 0.0001; Ctrl vs. PolyIC, **p = 0.0023; IFIH1, Ctrl vs. ArCK, *p=0.029; Ctrl vs. PolyIC, ***p= 0.0003; unpaired t-test. D. ArCK cells were grown as described in Figure 1A to determine ORF1 protein levels. RPE1-hTERT cells treated with azacitidine (Aza, 5 μM) for 5 days (lane 4) was used as a positive control. ORF1p levels under both long and short exposure were presented. Results were comparable between 2 biological replicates. E. The effect of inhibiting reverse transcriptase activity on NK cell-mediated cytotoxicity in aneuploid cells. ArCK cells were generated as described in Figure 1A and were continuously treated with reverse transcriptase inhibitor 3TC (7.5 μM) following chromosome mis-segregation. Control RPE1-hTERT cells were treated with 3TC for 3 days. 3TC was washed out during the NK cell co-culture assay. n=3 biological replicates; mean±SEM. ArCK vs. ArCK-3TC, p< 0.0001; KS test. F. HCT116 and DLD1 cells were treated with either DMSO (Ctrl) or the Mps1 inhibitor reversine (500 nM) for 48 hours (Aneuploid). The percentage of cells with nuclear RelA (left) and nuclear RelB (right) signals were quantified. n=3 biological replicates; mean±SEM. RelA, HCT116 Ctrl vs. HCT116 Aneuploid, **** p< 0.0001; DLD1 Ctrl vs. DLD1 Aneuploid, *p= 0.017. RelB, HCT116 Ctrl vs. HCT116 Aneuploid, **p= 0.004; DLD1 Ctrl vs. DLD1 Aneuploid, **p= 0.006; unpaired t-test. NF-κB pathway is upregulated in highly aneuploid cancer cell lines Our data indicate that in untransformed cells, aneuploidy induction upregulates NF-κB pathway, which contributes to NK cell-mediated immune clearance in vitro. Interestingly, in tumor cells high levels of aneuploidy correlate with immune evasion (Davoli et al, 2017; Taylor et al, 2018). It is thus possible that the transformed state of cancer cells suppressed aneuploidy-induced NF-κB signaling. To test this hypothesis, we interrogated the association between NF-κB activation and the degree of aneuploidy in full-blown tumors using the cancer cell line encyclopedia [CCLE; (Barretina et al, 2012; Ghandi et al, 2019)]. The degree of aneuploidy was scored in almost 1000 cell lines in the CCLE as recently described (Cohen-Sharir et al, 2021). We then created 65 two groups of cell lines, a highly-aneuploid and a near-euploid group, defined as the top and bottom quartiles of the number of arm-level chromosome gains and losses, respectively. To assess NF-κB activity in these two cell line groups, we created a ssGSEA signature score (Subramanian et al, 2005) for the Hallmark_TNFA_signaling_via_NFKB gene set and computed the association between this signature and the degree of aneuploidy by linear regression analysis (see Methods). We found that highly aneuploid cancer cell lines exhibit significantly higher transcriptional signature of NF-κB activity compared to the near-diploid lines (Fig 11A and B). This suggests that aneuploidy could also contribute to NF-κB upregulation in transformed cells. To test whether NF-κB activation in cancer cell lines also contribute to NK cell-mediated killing, we induced aneuploidy in the pseudo-diploid colon cancer cell lines HCT116 and DLD1 using the Mps1 inhibitor reversine. In agreement with previous reports, we observed NF-κB pathway upregulation following chromosome mis- segregation in both cancer cell lines indicated by a significant increase in both RelA and RelB nuclear translocation frequency [Fig 10F, (Vasudevan et al., 2020)]. However, we did not see an increase in NK cell-mediated killing in reversine-induced aneuploid HCT116 and DLD1 cancer cells (Fig 11C). Our data suggest that although an aneuploidy associated NF-κB response may still be evident in transformed cell lines, it is not sufficient to enhance the NK cell-mediated immune response. 66 Figure 11. NF-κB is active in highly aneuploid cancer cell lines A. Comparison of the Hallmark gene expression signature “TNFA_signaling via NFKB” in near-diploid (low aneuploidy) and highly-aneuploid (high aneuploidy) human cancer cell lines from the CCLE. *, p= 5e-08, empirical Bayes-moderated t-statistics. The y-axis represents the ssGSEA expression score. The grey line marks an ssGSEA score of 0, indicating the genes in the hallmark “TNFA_signaling via NFKB” geneset is not differentially regulated. B. Correlation plot between ssGSEA expression score of “TNFA_signaling via NFKB” and the aneuploid score for human cancer cell lines from the CCLE. The trend line for Spearman’s correlation plot is indicated in red. Spearman’s ρ= 0.181, p-value = 2.2e-08. C. HCT116 (left) or DLD1 cells (right) were treated with either DMSO or the Mps1 inhibitor reversine (500 nM) for 48 hours before assessing NK cell-mediated cytotoxicity. The drug was washed out during the NK cell co-culture assay. n=3 biological replicates; mean±SEM. HCT116 vs. HCT116 Mps1 inhibitor, p= 0.44, n.s.; DLD1 vs. DLD1 Mps1 inhibitor, p= 0.44, n.s.; KS test. 67 DISCUSSION We previously found that untransformed cells that underwent senescence due to highly abnormal karyotypes are recognized by NK cells in vitro. Here we investigated the molecular mechanism contributing to NK cell-mediated immune clearance and identified NF-κB signaling to be central to the interaction between aneuploid cells and NK cells. The NF-κB pathway contributes to the immunogenicity of ArCK cells Multiple studies have shown that senescent cells are recognized and eliminated by NK cells in vitro (Soriani et al, 2009, 2014; Iannello et al, 2013; Sagiv et al, 2016). In this study, we investigated how aneuploidy-induced senescence causes NK cell recognition. We first tested the hypothesis that G1 arrest elicits an NK cell response by comparing NK cell killing kinetics between various G1 arrests. Our analysis revealed that G1 arrest per se is not sufficient to cause recognition by NK cells because mTOR inhibition, which caused a quiescence-like G1 arrest, did not elicit an NK cell-mediated killing. Instead, in accordance with other reported NK cell–senescent cell interactions (Iannello et al., 2013), we found that the senescence-associated secretory program is primarily responsible for the NK cell-mediated killing of ArCK cells. Medium harvested from aneuploid cell cultures increased the ability of NK cells to kill euploid cells, suggesting that aneuploid cells can establish a pro-inflammatory environment where immune clearance takes place. Our data revealed that NF-κB pathway upregulation could be one of the major causes of NK cell- mediated immunogenicity of aneuploid cells in vitro. We observed 68 upregulation of both the canonical and non-canonical NF-κB pathways in ArCK cells, and inactivation of both pathways, but not of either one of these pathways alone, is sufficient to protect them from NK cell-mediated killing in vitro. Could other immune-response inducing pathways contribute to aneuploid cell recognition by NK cells? Chromosome mis-segregation has been linked to induction of an interferon response (Vasudevan et al, 2020; Mackenzie et al, 2017). Although we observed an up-regulation of the alpha and gamma interferon response in ArCK cells, our data indicate that inactivation of STAT1, the major transcription factor mediating the interferon response, did not significantly affect the elimination of ArCK cells by NK cells. Interestingly, STAT1 activation has been observed in untransformed and cancer cell lines where aneuploidy was induced by continuous exposure to reversine (F. Foijer, personal communication). We speculate that in such experimental set up, persistent DNA damage, which accompanies CIN rather than aneuploidy per se is likely to be the primary cause for JAK/STAT1 pathway activation. NF-κB activation in ArCK cells relies on multiple signals A key question arising from our findings is what causes NF-κB activation in aneuploid cells. Micronuclei – a well-known byproduct of chromosome segregation errors and aneuploidy (Crasta et al, 2012) – do not appear to be a major source of immune pathway activation in ArCK cells. On the other hand, while increased retrotransposition could contribute to NF-κB activation in aneuploid cells, the aneuploidy-associated stresses are likely to be the major activators of this immune- response. Proteotoxic, oxidative and genotoxic stress are defining features associated 69 with the aneuploid state [reviewed in (Santaguida & Amon, 2015)]. We previously found that the surface molecules MICA, MICB, CD155, CD112, ULBP1 and ULBP2- that mediate NK cell recognition- are subtly (about 2-fold) upregulated in ArCK cells (Santaguida et al, 2017). MICA and MICB are activated in response to proteotoxic stress, CD112 (also known as Nectin-2) and CD155 (also known as PVR), are expressed in response to DNA damage, and ULBP1 and ULBP2 are the product of both cellular stresses and DNA damage (Raulet & Guerra, 2009). This suggested that multiple features of the aneuploid state contribute to the upregulated immunogenicity. We propose that instead of a unique NK cell activating feature, the upregulated NK cell- mediated clearance of aneuploid cells is likely mediated by a combination of stresses elicited by the aneuploid state. Immune recognition of aneuploidy in cancer Aneuploidy is a hallmark of cancer that correlates with aggressive disease and immune evasion. Yet in primary cells there is ample evidence that chromosome instability and aneuploidy both elicit a variety of immune responses. For example, micronuclei associated with chromosome mis-segregation activate the interferon response via the cGAS/STING pathway (Mackenzie et al, 2017; Harding et al, 2017). Chromosome instability upregulates stress-activated protein kinase (SAPK) and c-Jun N-terminal kinase (JNK) pathways, which contribute to inflammatory response (Clemente-Ruiz et al, 2016; Benhra et al, 2018). MHC complex and antigen processing gene signatures have also been shown to be associated with aneuploidy (Dürrbaum et al, 2014). Hence, a crucial question in the field is how malignant transformation 70 dampens aneuploidy and CIN-induced immunogenicity. We investigated whether the aneuploidy induced NF-κB signature could be down-regulated in highly aneuploid cancer cell lines. Our data argue that this was not the case; indeed, the more aneuploid the cancer cell lines, the higher the NF-κB signaling levels, suggesting that aneuploidy also contributes to NF-κB upregulation in transformed cells. However, we found that the induction of aneuploidy in pseudo-diploid cancer cells did not further enhance the NK cell- mediated killing. We speculate that in transformed cells, other events occurring during tumorigenesis likely serve to counteract the NF-κB mediated immunogenicity and render the cancer cells insensitive for NK cell-mediated killing. Interestingly, in a tumor environment, the degree of NF-κB activation inversely correlates with the degree of aneuploidy (Taylor et al, 2018), raising the possibility that silencing of the aneuploidy- induced immunogenicity could be a non-cell autonomous event in cancer, perhaps induced by cells in the tumor microenvironment. Understanding which aspects of aneuploidy activate NF-κB signaling and how the activity of the pathway is modulated during tumor evolution will be the critical next steps in understanding the role of aneuploidy during tumorigenesis. MATERIALS AND METHODS Cell culture RPE1-hTERT cells (ATCC Cat# CRL-4000), HCT116 cells (ATCC Cat# CCL-247), and HeLa cells (ATCC Cat# CCL-2) were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen) supplied with 10% FBS (Atlanta Biologicals of South America origin) and penicillin/streptomycin (100 U/ml) and L-Glutamine (2 mM). Human primary 71 IMR90 (ATCC Cat# CCL-186), adult and neonatal normal human dermal fibroblasts (NHDF-Ad and NHDF-Neo; Lonza Cat# CC-2511 and Cat# CC-2509, respectively) were cultured in Eagle’s Minimum Essential Medium (EMEM, ATCC) supplied with 10% FBS and penicillin/streptomycin (100 U/ml) and L-Glutamine (2 mM). DLD-1 cells (ATCC Cat# CCL-221) were cultured in RPMI-1640 (Invitrogen) supplied with 10% FBS and penicillin/streptomycin (100 U/ml) and L-Glutamine (2 mM). NK92-MI cells (ATCC Cat# CRL-2408) and were cultured in MyeloCult H5100 medium (STEMCELL Technologies). All cells were grown at 37 °C with 5% CO2 in a humidified environment. Generation of cells arrested with complex karyotypes (ArCK) To generate ArCK cells, 2.5 x 105 RPE1-hTERT cells were plated on a 10cm culture dish and synchronized with thymidine (5mM) for 24 hours. Cells were then released into complete medium. 6 hours after thymidine release, cells were switched into medium containing reversine (500 nM). Reversine were washed out 18 hours later. 60 hours after drug washout, nocodazole (100 ng/ml, Sigma Aldrich) was added to the culture. 12 hours after nocodazole addition, mitotic cells were eliminated from the cell population by shake off. The shake-off process was performed five times in total at a 12-hour interval. The cells left on the plate after five shake-offs were called the ArCK population. Generation of cell-cycle arrested cells To generate G1 arrested samples, 5 x 105 RPE1-hTERT cells were plated on a 10cm culture dish and treated with the following drugs: doxorubicin (100 ng/ml, Sigma 72 Aldrich), palbociclib (5 μM, LC Laboratories), nutlin3 (10 μM, Cayman Chemical) and torin1 (1 μM, LC Laboratories). Cells were collected after 7 days of drug treatment. Video Microscopy All live cell imaging was performed using a spinning disk microscope (10x objective) with the environmental chamber maintained at 37 °C and 5% CO2 level. Target cells were plated onto 12-well glass bottom plates in complete normal growth medium at a density 4~6x104 cells/well overnight to allow attachment. Target cells were switched into NK cell growth medium MyeloCult H5100 and incubated for 10 hours before starting the live cell imaging. To assess NK cell-mediated cytotoxicity, NK92-MI cells were re- suspended into target cell condition medium at the indicated NK cell to target cell ratio immediately before the start of the NK cell killing assay. The cell mixture was filmed for 36 hours at a 30min time interval. To assess target cell growth without NK cells, target cells were filmed using the same imaging setting and time scale except no NK cells were added. Cell cycle analysis by flow cytometry Cells were trypsinized and resuspended into 10ml complete growth medium. After washing twice in cold phosphate-buffered saline (PBS), cells were fixed and permeablized in 70% ethanol at -20 °C overnight. Cells were then pelleted and incubated with RNAse (100 μg/mL, Thermo Fisher) and DAPI (1 μg/ml, Thermo Fisher) for 1h before flow cytometry analysis. 73 Immunofluorescence RPE1-hTERT cells were seeded onto fibronectin (10 μg/ml, Sigma Aldrich) coated coverslips at 50-70% confluency and allowed to attached overnight. For RelA and anti- Phospho-histone H2A.X staining, cells were fixed at room temperature with 4% paraformaldehyde in PBS for 15mins and permeabilized with 0.1% triton-X100 in PBS for 10mins. Cells were then blocked in 3% bovine serum albumin (BSA) in PBS for 40mins. Cells were incubated with primary antibodies for 90 mins at room temperature. The RelB staining protocol was adapted from (Vasudevan et al., 2020). The following primary antibodies were used: anti-phospho-histone H2A.X (Cell Signaling Technology #9718, 1:500), anti-RelA (Santa Cruz Biotechnology sc-8008 or Abcam ab16502, 1:50), anti-RelB (Abcam ab180127 or ab33907, 1:1000). The following secondary antibodies were used: Donkey anti-mouse IgG Cy3 (Euroclone, 1:10000), Alexa Fluor 488 goat anti-Rabbit IgG (Thermo Fisher, 1:1000). Either Hoechest or DAPI was used to stain DNA. Images were acquired using a DeltaVision (60x) or Leica SP8 Confocal (AOBS, 63X oil objective) microscope. Acquired images were analysed with Fiji software. Western Blot analysis To prepare protein samples, protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktail (Roche) were added to RIPA lysis buffer (Thermo Fisher Scientific) immediately before use. Cells were lysed in cold RIPA buffer and the lysate concentration was measured by Bradford assay. The lysate was then diluted with loading buffer and heated at 98°C for 5 min. Proteins were resolved on NuPAGE 4-12% Bis-Tris gels (Thermo Fisher Scientific) based on the manufacturer’s instructions and 74 transferred onto 0.2 μm PVDF membranes. Blots were blocked for 1 hour at room temperature in OneBlock blocking buffer (Genesee Scientific). Primary antibodies were incubated over night at 4°C. The following primary antibodies were used: anti-GAPDH (Santa Cruz sc-365062, 1:1000), anti-Vinculin (Sigma-Aldrich V9131, 1:5000), anti-p53 (Santa Cruz sc-126, 1:200), anti-p21 (Cell Signaling Technology #2947, 1:1000), anti- p65/RelA (Cell Signaling Technology #8242, 1:1000), anti-RelB (Abcam #180127, 1:1000), anti-Stat1 (Cell Signaling Technology #9175, 1:500), anti-IRF3 (Cell Signaling Technology, #11904, 1:1000), anti-Phospho-IRF3 (Cell Signaling Technology, #4947, 1:1000), anti-ORF1p (Cell Signaling Technology #88701, 1:1000). Beta galactosidase staining ArCK, doxorubicin, palbociclib, nutlin3, torin1 and untreated euploid proliferating RPE1- hTERT cells were plated with normal complete growth medium into 6-well plates at a density of 4x105 cells/well. 24 hours later, cells were fixed and stained for the β- Galactosidase activity using a senescence β-Galactosidase staining kit (Cell signaling technology #9860) following manufacturer’s instructions. Cytokine measurement ArCK, doxorubicin, palbociclib, nutlin3, torin1 and untreated euploid proliferating RPE1- hTERT cells were generated as described above. 8 ml of complete normal growth medium was placed onto cells and incubated for 36 hours. The conditioned medium was harvested and cell debris was eliminated by centrifugation. To determine the levels of secreted cytokines, condition medium was incubated with proteome profiler human 75 XL cytokine array (R&D Systems, ARY022B) and cytokine levels were measured based on manufacturer’s instructions. The levels of interferon alpha and interferon beta in the condition medium were determined using IFN alpha and IFN beta human ELISA kit (Thermo Fisher Scientific, 411001 and 414101 respectively). The total cell number from each sample was measured using a cellometer (Nexcelom). All cytokines, IFN alpha, and IFN beta readings were normalized to cell number. RNA sequencing and data analysis Total RNA was purified using RNeasy Mini Kits (QIAGEN) and sequenced on Illumina HiSeq2000. The RNAseq data were aligned to a transcriptome derived from the human hg38 primary assembly and an ensembl version 89 annotation with STAR version 2.5.3a (Dobin et al, 2013). Gene expression was summarized using RSEM version 1.3.0 (Li & Dewey, 2011) and samtools/1.3 (Li et al, 2009). An integer count table for differential expression analysis and log2 transcripts per million (TPM) with a plus 1 offset for data visualization was prepared with Tibco Spotfire Analyst (version 7.11.1). Differential expression analysis was done with DESeq2 [version 1.24.0 or version 1.30.0, (Love et al, 2014)] running under R (version 3.6.0 or version 4.0.3). Pre-ranked Gene Set Enrichment Analysis [versions 2-3.0_beta2, 4.0.3 or 4.1.0, (Subramanian et al, 2005)] was run using the DESeq2 Wald statistic as a ranking metric and gene set collections from ΜsigDB [versions 6.2, 7.0 or 7.2, (Liberzon et al, 2015)]. Three biological replicates were included per condition in the RNA sequencing. The treated samples were compared to the euploid control cells processed and sequenced from the same experiment to avoid batch effect. 76 Generation of knock-out cell lines using the CRISPR-Cas9 system Lentiviral CRISPR-Cas9 plasmid LentiCRISPR_v2-Puro (Brett Stringer’s Lab) cloned with guide RNA (gRNA) designed by Feng Zhang’s lab from the Broad Institute targeting exons of human RELA (AGCGCCCCTCGCACTTGTAG), RELB (TCGCCGCGTCGCCAGACCGC), and STAT1 (ATTGATCATCCAGCTGTGAC) were purchased from GenScript. CRISPRv2 constructs along with packaging plasmids pMD2.G (Addgene 12259) and psPAX2 (Addgene 12260) were transfected into 293FT cells (Thermo Fisher, Cat# R70007) using TransIT-LT1 transfection reagent (Mirus). Virus was collected and the lentiviral titer was estimated by Lenti-X GoStix Plus (TaKaRa). RPE1-hTERT cells were plated at ~60% confluency and infected at a MOI of 1 for RELA, RELB, or STAT1 single knockouts, and at a MOI of 2 for RELA RELB double knockout. Virus was washed out 20 hours post infection and the non-infected cells were selected against by puromycin treatment. LentiCRISPR_v2-Puro vector without gRNAs was integrated into RPE1-hTERT cells to generate the control cell line. 2 days after puromycin selection, single cells were sorted into 96 wells and expanded as individual clones. The successful knockout of RELA, RELB or STAT1 was verified by western blot analysis. All experiments described above were performed in at least two single cell knock-out clones. RT-qPCR analysis Total RNA was purified using an RNeasy Mini Kit (QIAGEN). RNA concentration was determined using a nanodrop. 750ng of RNA was used for reverse transcription reactions using SuperScript III First-Strand Synthesis SuperMix (Invitrogen). The mRNA 77 levels were then quantified by qPCR using SYBR Premix Ex Taq (TaKaRa) on Roche Light Cycler. Primer sequences: STAT1-Forward GATGTTTCATTTGCCACCATCCGTTTTC. STAT1-Reverse GGCGTTTTCCAGAATTTTCCTTTCTTCC. GAPDH-Forward CCATGTTCGTCATGGGTGTGAACCATG. GAPDH-Reverse CCACAGCCTTGGCAGCGCCAGTAGAGG. DDX58_Forward CGGCACAGAAGTGTATATTGGATGCATTC. DDX58_Reverse GGAAGCACTTGCTACCTCTTGCTCTTC. IFIH1_Forward CTGGGACTAACAGCTTCACCTGGTGTTG. IFIH1_Reverse GCATCTGCAATGGCAAACTTCTTGCATG. To measure NF-κB target expressions, 500ng of RNA were retro-transcribed using OneScript Plus cDNA Syntheiss Kit (abm, G236) and the following TaqMan assays were used (ThermoFisher Scientific): IL-1b, hs00174097_m1; IL-6, Hs00985639_m1; IL-8, hs00174103_m1. NF-κB Secreted Alkaline Phosphatase (SEAP) Reporter Assay A NF-κB Secreted Alkaline Phosphatase Reporter Assay Kit (Novus biologicals, NBP2- 25286) was used to measure the secretion of the SEAP protein under the control of the NF-κB promoter. To generate NF-κB secreted alkaline phosphatase reporter cell line, RPE1-hTERT cells were plated in a six-well plate in 2 mL DMEM, containing 10% FBS, L-Glutamine and nonessential amino acids. 12 hours later, cells were transfected with 1 g pNF-κB /SEAP plasmid using 6 L Lipofectamine 3000 (Invitrogen, L3000008) per well. 10 hours after transfection, cells were replaced with medium containing either reversine (500 nM) or DMSO for 96 hours. The supernatant was then collected to 78 measure the levels of SEAP according to the manufacturer’s instructions. The SEAP assay standard curve used to calculate the sample SEAP concentration was generated by loading a serial dilution of SEAP standard on the same plate. The absorbance was measured after 1 hour of incubation with the PNPP substrate using a PHERAStar FSX Microplate Reader (BMG LABTECH). SEAP secretion levels were then normalized to cell number for each condition. CCLE data analysis Gene expression data for the CCLE lines were obtained from DepMap 19Q1 DepMap release; www.DepMap.org). A ssGSEA signature (Subramanian et al, 2005) score was calculated for the Hallmark_TNFA_signaling_via_NFKB gene set. Aneuploidy scores were obtained from (Cohen-Sharir et al.). The association between the signature score and the aneuploidy groups was assessed by linear regression analysis, using the R package limma (Ritchie et al, 2015). Significance was calculated by empirical-Bayes moderated t-statistics. Statistical analysis To test statistical significance on the killing assay, the raw data corresponding to the time points when a target cell was killed by NK cells were pooled from individual biological replicates for each condition. The cells that were not killed at the end of the 36h killing assay were assigned to 36h as the killing time. The distribution of the killing time in each condition was compared and the significance was determined using nonparametric Kolmogorov-Smirnov test (KS test). Significance was called at p < 0.05. 79 All statistical analysis was performed using GraphPad Prism or R software. Details of the statistical tests on data besides the killing assay were stated in the associated figure legends. Data availability The RNA-seq data sets generated for this study can be accessed at Gene Expression Omnibus (GEO) database with the accession number GEO: GSE154919. ACKNOWLEDGEMENTS We thank Floris Foijer for insightful discussions and for sharing data prior to publication. We thank Jacqueline Lees, Iain Cheeseman, Federica Facciotti, and members of the Amon and Santaguida labs for their helpful comments and discussions regarding this project and manuscript. We thank Charlie Whittaker, Dikshant Pradhan, and the Swanson Biotechnology Center for help with the gene expression analysis. This work was supported by grants to S. S. from the Italian Association for Cancer Research (MFAG 2018 - ID. 21665 project), Fondazione Cariplo, Ricerca Finalizzata (GR-2018- 12367077), the Rita-Levi Montalcini program from MIUR and by the Italian Ministry of Health with Ricerca Corrente and 5x1000 funds and by NIH grant CA206157 to A.A., who was an investigator of the Howard Hughes Medical Institute, the Paul F. Glenn Center for Biology of Aging Research at MIT and the Ludwig Center at MIT. Work in the Ben-David lab is supported by the Azrieli Foundation, the Richard Eimert Research Fund on Solid Tumors, the Tel-Aviv University Cancer Biology Research Center, and the Israel Cancer Association. This work is dedicated to the memory of Angelika Amon. 80 REFERENCES Alexopoulou L, Holt AC, Medzhitov R & Flavell RA (2001) Recognition of double- stranded RNA and activation of NF-κB by Toll-like receptor 3. Nature 413: 732–738 Bakhoum SF, Ngo B, Laughney AM, Cavallo J-A, Murphy CJ, Ly P, Shah P, Sriram RK, Watkins TBK, Taunk NK, et al (2018) Chromosomal instability drives metastasis through a cytosolic DNA response. Nature 553: 467–472 Barretina J, Caponigro G, Stransky N, Venkatesan K, Margolin AA, Kim S, Wilson CJ, Lehár J, Kryukov GV, Sonkin D, et al (2012) The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 483: 603–607 Ben-David U & Amon A (2020) Context is everything: aneuploidy in cancer. Nat Rev Genet 21: 44–62 Benhra N, Barrio L, Muzzopappa M & Milán M (2018) Chromosomal Instability Induces Cellular Invasion in Epithelial Tissues. Dev Cell 47: 161-174.e4 Burke JR, Pattoli MA, Gregor KR, Brassil PJ, MacMaster JF, McIntyre KW, Yang X, Iotzova VS, Clarke W, Strnad J, et al (2003) BMS-345541 Is a Highly Selective Inhibitor of IκB Kinase That Binds at an Allosteric Site of the Enzyme and Blocks NF- κB-dependent Transcription in Mice*. J Biol Chem 278: 1450–1456 Cecco MD, Ito T, Petrashen AP, Elias AE, Skvir NJ, Criscione SW, Caligiana A, Brocculi G, Adney EM, Boeke JD, et al (2019) L1 drives IFN in senescent cells and promotes age-associated inflammation. Nature 566: 73–78 Chiossone L, Dumas P-Y, Vienne M & Vivier E (2018) Natural killer cells and other innate lymphoid cells in cancer. Nat Rev Immunol 18: 671–688 Clemente-Ruiz M, Murillo-Maldonado JM, Benhra N, Barrio L, Pérez L, Quiroga G, Nebreda AR & Milán M (2016) Gene Dosage Imbalance Contributes to Chromosomal Instability-Induced Tumorigenesis. Dev Cell 36: 290–302 Cohen-Sharir Y, McFarland JM, Abdusamad M, Marquis C, Bernhard SV, Kazachkova M, Tang H, Ippolito MR, Laue K, Zerbib J, et al (2021) Aneuploidy renders cancer cells vulnerable to mitotic checkpoint inhibition. Crasta K, Ganem NJ, Dagher R, Lantermann AB, Ivanova EV, Pan Y, Nezi L, Protopopov A, Chowdhury D & Pellman D (2012) DNA breaks and chromosome pulverization from errors in mitosis. Nature 482: 53–58 81 Davoli T, Uno H, Wooten EC & Elledge SJ (2017) Tumor aneuploidy correlates with markers of immune evasion and with reduced response to immunotherapy. Science 355: eaaf8399 Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M & Gingeras TR (2013) STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29: 15–21 Dou Z, Ghosh K, Vizioli MG, Zhu J, Sen P, Wangensteen KJ, Simithy J, Lan Y, Lin Y, Zhou Z, et al (2017) Cytoplasmic chromatin triggers inflammation in senescence and cancer. Nature 550: 402–406 Dunphy G, Flannery SM, Almine JF, Connolly DJ, Paulus C, Jønsson KL, Jakobsen MR, Nevels MM, Bowie AG & Unterholzner L (2018) Non-canonical Activation of the DNA Sensing Adaptor STING by ATM and IFI16 Mediates NF-κB Signaling after Nuclear DNA Damage. Mol Cell 71: 745-760.e5 Dürrbaum M, Kuznetsova AY, Passerini V, Stingele S, Stoehr G & Storchová Z (2014) Unique features of the transcriptional response to model aneuploidy in human cells. Bmc Genomics 15: 139 Ghandi M, Huang FW, Jané-Valbuena J, Kryukov GV, Lo CC, McDonald ER, Barretina J, Gelfand ET, Bielski CM, Li H, et al (2019) Next-generation characterization of the Cancer Cell Line Encyclopedia. Nature 569: 503–508 Glück S, Guey B, Gulen MF, Wolter K, Kang T-W, Schmacke NA, Bridgeman A, Rehwinkel J, Zender L & Ablasser A (2017) Innate immune sensing of cytosolic chromatin fragments through cGAS promotes senescence. Nat Cell Biol 19: 1061– 1070 Gordon DJ, Resio B & Pellman D (2012) Causes and consequences of aneuploidy in cancer. Nat Rev Genet 13: 189–203 Gorgoulis V, Adams PD, Alimonti A, Bennett DC, Bischof O, Bishop C, Campisi J, Collado M, Evangelou K, Ferbeyre G, et al (2019) Cellular Senescence: Defining a Path Forward. Cell 179: 813–827 Harding SM, Benci JL, Irianto J, Discher DE, Minn AJ & Greenberg RA (2017) Mitotic progression following DNA damage enables pattern recognition within micronuclei. Nature 548: 466–470 Hassold T & Hunt P (2001) To err (meiotically) is human: the genesis of human aneuploidy. Nat Rev Genet 2: 280–291 Hayden MS & Ghosh S (2012) NF-κB, the first quarter-century: remarkable progress and outstanding questions. Gene Dev 26: 203–234 82 Iannello A, Thompson TW, Ardolino M, Lowe SW & Raulet DH (2013) p53-dependent chemokine production by senescent tumor cells supports NKG2D-dependent tumor elimination by natural killer cellsSenescence promotes tumor elimination by NK cells. J Exp Medicine 210: 2057–2069 Janssen A, Burg M van der, Szuhai K, Kops GJPL & Medema RH (2011) Chromosome Segregation Errors as a Cause of DNA Damage and Structural Chromosome Aberrations. Science 333: 1895–1898 Kucheryavenko O, Nelson G, Zglinicki T von, Korolchuk VI & Carroll B (2019) The mTORC1-autophagy pathway is a target for senescent cell elimination. Biogerontology 20: 331–335 Lamm N, Ben-David U, Golan-Lev T, Storchová Z, Benvenisty N & Kerem B (2016) Genomic Instability in Human Pluripotent Stem Cells Arises from Replicative Stress and Chromosome Condensation Defects. Cell Stem Cell 18: 253–261 Li B & Dewey CN (2011) RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. Bmc Bioinformatics 12: 323 Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R & Subgroup 1000 Genome Project Data Processing (2009) The Sequence Alignment/Map format and SAMtools. Bioinformatics 25: 2078–2079 Li M, Fang X, Baker DJ, Guo L, Gao X, Wei Z, Han S, Deursen JM van & Zhang P (2010) The ATM–p53 pathway suppresses aneuploidy-induced tumorigenesis. Proc National Acad Sci 107: 14188–14193 Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP & Tamayo P (2015) The Molecular Signatures Database Hallmark Gene Set Collection. Cell Syst 1: 417–425 Lindsley DL, Sandler L, Baker BS, Carpenter ATC, Denell RE, Hall JC, Jacobs PA, Miklos GLG, Davis BK, Gethmann RC, et al (1972) SEGMENTAL ANEUPLOIDY AND THE GENETIC GROSS STRUCTURE OF THE DROSOPHILA GENOME. Genetics 71: 157–184 Liu S, Kwon M, Mannino M, Yang N, Renda F, Khodjakov A & Pellman D (2018) Nuclear envelope assembly defects link mitotic errors to chromothripsis. Nature 561: 551–555 Liu T, Zhang L, Joo D & Sun S-C (2017) NF-κB signaling in inflammation. Signal Transduct Target Ther 2: 17023 Lorke DE (1994) Developmental Characteristics of Trisomy 19 Mice. Cells Tissues Organs 150: 159–169 83 Love MI, Huber W & Anders S (2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15: 550 Mackenzie KJ, Carroll P, Martin C-A, Murina O, Fluteau A, Simpson DJ, Olova N, Sutcliffe H, Rainger JK, Leitch A, et al (2017) cGAS surveillance of micronuclei links genome instability to innate immunity. Nature 548: 461–465 Martin S & Santaguida S (2020) Understanding Complexity of Cancer Genomes: Lessons from Errors. Dev Cell 53: 500–502 Nicolantonio FD, Arena S, Gallicchio M, Zecchin D, Martini M, Flonta SE, Stella GM, Lamba S, Cancelliere C, Russo M, et al (2008) Replacement of normal with mutant alleles in the genome of normal human cells unveils mutation-specific drug responses. Proc National Acad Sci 105: 20864–20869 Ohashi A, Ohori M, Iwai K, Nakayama Y, Nambu T, Morishita D, Kawamoto T, Miyamoto M, Hirayama T, Okaniwa M, et al (2015) Aneuploidy generates proteotoxic stress and DNA damage concurrently with p53-mediated post-mitotic apoptosis in SAC-impaired cells. Nat Commun 6: 7668 Oliveira RL de & Bernards R (2018) Anti‐cancer therapy: senescence is the new black. Embo J 37 Passerini V, Ozeri-Galai E, Pagter MS de, Donnelly N, Schmalbrock S, Kloosterman WP, Kerem B & Storchová Z (2016) The presence of extra chromosomes leads to genomic instability. Nat Commun 7: 10754 Perkins ND (2007) Integrating cell-signalling pathways with NF-κB and IKK function. Nat Rev Mol Cell Bio 8: 49–62 Pfau SJ & Amon A (2012) Chromosomal instability and aneuploidy in cancer: from yeast to man. Embo Rep 13: 515–527 Pfau SJ, Silberman RE, Knouse KA & Amon A (2016) Aneuploidy impairs hematopoietic stem cell fitness and is selected against in regenerating tissues in vivo. Gene Dev 30: 1395–1408 Pommier Y, Leo E, Zhang H & Marchand C (2010) DNA Topoisomerases and Their Poisoning by Anticancer and Antibacterial Drugs. Chem Biol 17: 421–433 Raulet DH & Guerra N (2009) Oncogenic stress sensed by the immune system: role of natural killer cell receptors. Nat Rev Immunol 9: 568–580 Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W & Smyth GK (2015) limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43: e47–e47 84 Robertson MJ (2002) Role of chemokines in the biology of natural killer cells. J Leukocyte Biol 71: 173–83 Roper RJ & Reeves RH (2006) Understanding the Basis for Down Syndrome Phenotypes. Plos Genet 2: e50 Sagiv A, Burton DGA, Moshayev Z, Vadai E, Wensveen F, Ben-Dor S, Golani O, Polic B & Krizhanovsky V (2016) NKG2D ligands mediate immunosurveillance of senescent cells. Aging Albany Ny 8: 328–344 Santaguida S & Amon A (2015) Short- and long-term effects of chromosome mis- segregation and aneuploidy. Nat Rev Mol Cell Bio 16: 473–485 Santaguida S, Richardson A, Iyer DR, M’Saad O, Zasadil L, Knouse KA, Wong YL, Rhind N, Desai A & Amon A (2017) Chromosome Mis-segregation Generates Cell- Cycle-Arrested Cells with Complex Karyotypes that Are Eliminated by the Immune System. Dev Cell 41: 638-651.e5 Santaguida S, Tighe A, D’Alise AM, Taylor SS & Musacchio A (2010) Dissecting the role of MPS1 in chromosome biorientation and the spindle checkpoint through the small molecule inhibitor reversine. J Cell Biology 190: 73–87 Santaguida S, Vasile E, White E & Amon A (2015) Aneuploidy-induced cellular stresses limit autophagic degradation. Gene Dev 29: 2010–2021 Sheltzer JM, Torres EM, Dunham MJ & Amon A (2012) Transcriptional consequences of aneuploidy. Proc National Acad Sci 109: 12644–12649 Shimasaki N, Jain A & Campana D (2020) NK cells for cancer immunotherapy. Nat Rev Drug Discov 19: 200–218 Signorino G, Mohammadi N, Patanè F, Buscetta M, Venza M, Venza I, Mancuso G, Midiri A, Alexopoulou L, Teti G, et al (2014) Role of Toll-Like Receptor 13 in Innate Immune Recognition of Group B Streptococci. Infect Immun 82: 5013–5022 Sliwinska MA, Mosieniak G, Wolanin K, Babik A, Piwocka K, Magalska A, Szczepanowska J, Fronk J & Sikora E (2009) Induction of senescence with doxorubicin leads to increased genomic instability of HCT116 cells. Mech Ageing Dev 130: 24–32 Soriani A, Iannitto ML, Ricci B, Fionda C, Malgarini G, Morrone S, Peruzzi G, Ricciardi MR, Petrucci MT, Cippitelli M, et al (2014) Reactive Oxygen Species– and DNA Damage Response–Dependent NK Cell Activating Ligand Upregulation Occurs at Transcriptional Levels and Requires the Transcriptional Factor E2F1. J Immunol 193: 950–960 85 Soriani A, Zingoni A, Cerboni C, Iannitto ML, Ricciardi MR, Gialleonardo VD, Cippitelli M, Fionda C, Petrucci MT, Guarini A, et al (2009) ATM-ATR–dependent up- regulation of DNAM-1 and NKG2D ligands on multiple myeloma cells by therapeutic agents results in enhanced NK-cell susceptibility and is associated with a senescent phenotype. Blood 113: 3503–3511 Sousa-Victor P, García-Prat L & Muñoz-Cánoves P (2015) Dual mTORC1/C2 inhibitors: gerosuppressors with potential anti-aging effect. Oncotarget 6: 23052–23054 Stingele S, Stoehr G, Peplowska K, Cox J, Mann M & Storchova Z (2012) Global analysis of genome, transcriptome and proteome reveals the response to aneuploidy in human cells. Mol Syst Biol 8: 608 Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES, et al (2005) Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. P Natl Acad Sci Usa 102: 15545–15550 Tam YK, Maki G, Miyagawa B, Hennemann B, Tonn T & Klingemann H-G (1999) Characterization of Genetically Altered, Interleukin 2-Independent Natural Killer Cell Lines Suitable for Adoptive Cellular Immunotherapy. Hum Gene Ther 10: 1359–1373 Taylor AM, Shih J, Ha G, Gao GF, Zhang X, Berger AC, Schumacher SE, Wang C, Hu H, Liu J, et al (2018) Genomic and Functional Approaches to Understanding Cancer Aneuploidy. Cancer Cell 33: 676-689.e3 Torres EM, Dephoure N, Panneerselvam A, Tucker CM, Whittaker CA, Gygi SP, Dunham MJ & Amon A (2010) Identification of Aneuploidy-Tolerating Mutations. Cell 143: 71–83 Vasudevan A, Baruah PS, Smith JC, Wang Z, Sayles NM, Andrews P, Kendall J, Leu J, Chunduri NK, Levy D, et al (2020) Single-Chromosomal Gains Can Function as Metastasis Suppressors and Promoters in Colon Cancer. Dev Cell 52: 413-428.e6 Villarino AV, Kanno Y & O’Shea JJ (2017) Mechanisms and consequences of Jak– STAT signaling in the immune system. Nat Immunol 18: 374–384 Wang RW, MacDuffie E & Santaguida S (2018) Generation and Isolation of Cell Cycle- arrested Cells with Complex Karyotypes. J Vis Exp Jove: 57215 Weaver BA & Cleveland DW (2006) Does aneuploidy cause cancer? Curr Opin Cell Biol 18: 658–667 Wiley CD, Schaum N, Alimirah F, Lopez-Dominguez JA, Orjalo AV, Scott G, Desprez P- Y, Benz C, Davalos AR & Campisi J (2018) Small-molecule MDM2 antagonists attenuate the senescence-associated secretory phenotype. Sci Rep-uk 8: 2410 86 Williams BR, Prabhu VR, Hunter KE, Glazier CM, Whittaker CA, Housman DE & Amon A (2008) Aneuploidy Affects Proliferation and Spontaneous Immortalization in Mammalian Cells. Science 322: 703–709 Yang J, Amiri KI, Burke JR, Schmid JA & Richmond A (2006) BMS-345541 Targets Inhibitor of κB Kinase and Induces Apoptosis in Melanoma: Involvement of Nuclear Factor κB and Mitochondria Pathways. Clin Cancer Res 12: 950–960 Zamanian-Daryoush M, Mogensen TH, DiDonato JA & Williams BRG (2000) NF-κB Activation by Double-Stranded-RNA-Activated Protein Kinase (PKR) Is Mediated through NF-κB-Inducing Kinase and IκB Kinase. Mol Cell Biol 20: 1278–1290 87 88 Chapter 3: RAD21 Promotes Repair of Oncogenic Replication Stress-Induced Damage in Ewing Sarcoma Manuscript in preparation Wang, R. W., Panday, A., Ma, D; Lees, J.A., Scully, R., Amon, A., Su, X.A. RAD21 Promotes Repair of Oncogenic Replication Stress-Induced Damage in Ewing Sarcoma. RWW and XAS performed experiments and analyzed the data in Figures 1H-J, 2G, 5D- G, and 6A-C. XAS performed experiments and analyzed the data in Figures 1A, 1D-F, 5A-C, and 5H- I. AP performed and analyzed the data in Figure 3C-D. RWW performed the experiments and analyzed the data in the rest of the figures. DM analyzed all the CUT&RUN data. 89 ABSTRACT EWS-FLI1, the fusion oncogene that drives aggressive pediatric Ewing sarcoma, is a strong inducer of replication stress. It accelerates G1/S phase transition, increases R-loop formation and elevates DNA damage levels, thus leading to growth defects and senescence in primary cells. We previously demonstrated that gaining of a copy of the cohesin subunit gene, RAD21, significantly mitigates EWS-FLI1-induced replication stress, promotes oncogenesis, and partially drives high recurrence of chromosome 8 gain in Ewing sarcoma. Here, we report that EWS-FLI1 expression in both normal euploid fibroblasts and Ewing sarcoma cancer cells induces transcription-replication conflicts (TRCs). Importantly, we found RAD21 is enriched at the TRC regions and is recruited to the stalled replication forks. Overexpression of RAD21 significantly reduced the TRCs caused by EWS-FLI1 expression in the primary euploid cells, and reduction of RAD21 levels in a trisomy 8 cancer cell line increases TRCs. Using a BioID approach, we discovered that RAD21 exhibited increased interactions with several DNA damage repair initiation factors in the primary cells experiencing oncogene-induced replication stress. These findings provide mechanistic evidence on how RAD21 promotes repair of oncogenic stress-induced DNA damage, which further suggest the potential of targeting cohesin for cancer treatment. 90 INTRODUCTION The accurate and complete replication of DNA is a key step for ensuring genomic integrity and faithful cell division. Although the DNA replication process is tightly regulated, it is vulnerable to many endogenous and exogenous threats. For example, cellular metabolites, ultraviolet (UV) and ionizing radiation can all cause damage to the template DNA. Such damage compromises the replication machinery and leads to slowed DNA synthesis and/or frequent replication fork stalling, known as replication stress (Zeman & Cimprich, 2014). Replication stress is also a frequent hallmark of cancer (Gaillard et al., 2015). Many oncogenes drive cell proliferation in a manner that promotes S-phase entry, induces unlicensed replication origin firing, and/or disrupts nucleotide metabolism, thereby deregulating the replication process (Kotsantis et al., 2018). Persistent replication stress can lead to genomic instability, which is highly toxic to cell physiology. If a stalled replication fork is not properly stabilized and repaired, it results in irreversible replication fork collapse. Such events cause high levels of DNA damage and ultimately lead to cellular senescence or apoptosis (Gaillard et al., 2015; Kotsantis et al., 2018). Consequently, tumor maintenance and progression can be enabled by genetic alterations that relieve oncogene-induced replication stress. We have previously investigated oncogene-induced replication stress in Ewing sarcoma. Ewing sarcoma is the second most common aggressive pediatric bone and soft tissue tumor driven by a fusion oncogene EWS-FLI1 (Grünewald et al., 2018). EWS-FLI1 is generated by the reciprocal t(11;22)(q24;q12) chromosomal translocation, which brings together the RNA binding transcriptional activator EWSR1 and the DNA binding domain of the ETS family transcription factor FLI1(Anderson et al., 2018). 91 Recent studies have shown that EWS-FLI1 expression accelerates S-phase entry and globally upregulates transcription levels, which can increase the formation of R-loops (Gorthi et al., 2018; Su et al., 2021). R-loops are RNA:DNA hybrids formed outside of the transcription bubble or the Okazaki fragment RNA priming region. These structures are highly prone for mutation and DNA damage (Aguilera & García-Muse, 2012). We hypothesized that the EWS-FLI1 induced R-loops might result from frequent collisions between transcription and replication machineries, called transcription-replication conflicts (TRCs), since these are known to be a frequent consequence of upregulated transcription and elevated replication stress (Aguilera & García-Muse, 2012; Kotsantis et al., 2018). We previously identified RAD21, a subunit in the cohesin complex, as one of the key players in mitigating replication stress caused by EWS-FLI1 in Ewing sarcoma (Su et al., 2021). The present of one extra copy of RAD21 is sufficient to rescue both the cell proliferation defects and high levels of DNA damage induced by expression of EWS-FLI1 in primary human fibroblasts (Su et al., 2021). RAD21 is the rate limiting subunit for cohesin loading onto the sister chromatids, and its structure and function are both evolutionarily conserved. In addition, RAD21 facilitates DNA damage repair in multiple organisms. In yeast, the RAD21 homolog, SCC1, is recruited to the double stranded break (DSB) sites and stalled replication forks, where it promotes efficient damage repair by homologous recombination [HR; (Ünal et al., 2004; Ünal et al., 2008)]. In mammalian cells, RAD21 is necessary for post-replicative damage repair, where its function at the break site is enabled by the SUMOlyation process (Wu et al., 2012). Despite these findings, the mechanism(s) by which RAD21 acts to mitigate oncogene 92 induced replication stress and decrease DNA damage in cancer remains unknown. Here, we elucidated this mechanism in the context of EWS-FLI1 induced replication stress in Ewing sarcoma. Our data show that RAD21 is recruited to the stalled replication fork and acts to facilitate the resolution of transcription-replication conflicts. Furthermore, we found that RAD21 associates with a group of DNA damage repair initiation proteins and such interactions could stabilize the stalled fork, facilitate efficient DNA damage repair, and promote fork restart in a coordinated manner. These findings provide potential therapeutic implications for targeting RAD21 in cancer treatment. RESULTS EWS-FLI1 induced transcription-replication conflicts cause oncogenic replication stress EWS-FLI1 expression globally upregulates transcription level, increases R-loop formation (Gorthi et al. 2018), and accelerates S phase entry, which can cause unlicensed early origin firing (Bertoli et al., 2013; Su et al., 2021). These observations led us to hypothesize that EWS-FLI1 expression elevates transcription-replication conflicts (TRCs) in Ewing sarcoma. To test this notion, we assessed the convergence of transcription and replication events in the established Ewing sarcoma cell line TC32, which expresses EWS-FLI1 (Fig 1A, left panel). We used the cleavage under target & release using nuclease (CUT&RUN) assay (Skene & Henikoff, 2017) to map the genome-wide transcription and replication signals by detecting chromatin-bound RNA polymerase II (RNAPII) and DNA clamp protein proliferating cell nuclear antigen (PCNA) respectively. To determine the contribution of EWS-FLI1 on TRCs, we used a 93 TC32 variant (shEWS-FLI1) carrying a stable short hairpin RNA (shRNA) that reduced the levels of EWS-FLI1 protein to 50% (Fig 1A, left panel), and compared its RNAPII and PCNA signals to those of the TC32 shMock control. To analyze the global RNAPII and PCNA enrichment signals, we used the Sparse Enrichment Analysis for CUT&RUN (SEACR) algorithm, which was specifically designed for highly selective and confident peak calling for CUT&RUN analysis (Meers et al., 2019). We also determined the DNA pulldown signal intensity in each condition by counting the total sequencing reads within the peaks of interest. 94 Figure 1. EWS-FLI1-induced transcription-replication conflicts cause oncogenic replication stress. A. EWS-FLI1 protein levels in TC32 cells and human primary neonatal fibroblasts (NHDF-Neo). GAPDH was used as a loading control. n=2, representative pictures are shown. shMock: mock shRNA; shEF: EWS-FLI1 shRNA. Vector: overexpression vector control plasmid; EF: induction of EWS-FLI1 for 48 hours. B. A representation of RNAPII and PCNA CUT&RUN overlapping peak region in the TC32 cells harboring a mock shRNA vector control (shMock; top 2 tracks) or an EWS-FLI1 shRNA (shEWS-FLI1). All pictures were within the same coordinates. Peak height represents the sum of read counts from two independent biological replicates. The height of each track is 250 read counts. C. Signal intensity for RNAPII or PCNA CUT&RUN total read counts in RNAP-PCNA overlapping region in TC32 cells harboring either a mock shRNA vector control (shMock) or an EWS-FLI1 shRNA (shEF) within the same genomic coordinates. The 95 read counts were ranked from top to bottom in descending order in each condition. Total read counts are as indicated by color gradients. D. Representative images for RNAPII-PCNA proximity ligation assay (PLA) foci in NHDF-Neo and TC32 cells. PLA foci (in red) and DNA (in blue) are shown. Scale bar, 10 μm. E-F. Quantification of RNAPII-PCNA PLA foci in NHDF-Neo (E) and TC32 cells (F). At least 40 cells were scored in each condition. ****, p < 0.0001, two-tailed nonparametric two-group Mann–Whitney U-test. Vector: overexpression vector control plasmid; EWS-FLI1: induction of EWS-FLI1 for 48 hours. shMock: mock shRNA; shEF: EWS-FLI1 shRNA. G. Illustration of EWS-FLI1 induced transcription-replication conflicts (TRCs) in a head- on (top) or co-directional (bottom) conformation. Minichromosome maintenance complex (MCM) unwinds the double-stranded DNA helix. DNA polymerase ε (Pol ε) and DNA polymerase δ (Pol δ) elongate the leading and lagging strands respectively. RNA polymerase II (RNAPII) and DNA clamp protein proliferating cell nuclear antigen (PCNA) are also depicted. A star represents a stalled replication fork caused by a collision event. H. Experimental scheme and representative images of DNA combing assay. IdU labeled track shown in green, CldU labeled track shown in red, and DNA fiber shown in blue. I-J. Measurement of replication fork speed (I) and the distribution of IdU:CldU ratio (J) in the NHDF-Neo. Fork speed was calculated as the IdU-CldU labeled fiber length/labeling time (40mins). ****, p < 0.0001, Two-tailed nonparametric two-group Mann–Whitney U-test. The distribution of IdU:CldU ratio was plotted and the Gaussian normal distribution fit curve is shown. The variances between the two conditions were compared by using F-test: p < 0.001. At least 100 intact DNA fibers were analyzed in each condition. EF: EWS-FLI1,V: Vector control. We mapped 14755 RNAPII peaks in the shMock TC32 line (Fig 2A). Notably, these overlapped significantly (72.4%) with RNAPII peaks previously identified in the wild type TC32 cells using traditional chromatin immunoprecipitation with sequencing (ChIP-seq) and the MACS2 peak calling algorithm [Fig 2A and B (Gothi et al 2018)]. In the shEWS- FLI1 cells, we detected a 46% decrease of RNAPII peaks (from 14755 to 8001) and a visible reduction of total sequencing reads (Fig 2A and C). This confirmed that EWS- FLI1 promotes transcription, as previously reported (Gorthi et al., 2018). For PCNA, the SEACR algorithm identified 11645 regions as peaks in the shMock TC32 Ewing sarcoma cells (Fig 2A). This enrichment of PCNA at certain genomic zones suggests 96 the presence of replication forks that are slowed or stalled, but have not collapsed. This is entirely consistent with the prominent replication stress that exists in these EWS-FLI1 expressing cells (Su et al., 2021). EWS-FLI1 knockdown did not significantly alter either the fraction of cells that were in S-phase, or the levels of PCNA protein, as assessed by EdU (5-ethynyl-2'-deoxyuridine) labeling and western blotting respectively (Fig 1A, left panel, 2D and E). However, EWS-FLI1 knockdown almost completely abolished the PCNA peaks (Fig 2A), arguing that this largely restored the processivity of the replication forks and thereby prevented enrichment at specific locations. Notably, the loss of PCNA peaks was observed in both of the independent biological replicates analyzed, and also supported by total read intensity analysis, for example within the 11645 peak regions (Fig 2C). We also examined H3K4Me3, as a control for CUT&RUN, and found that the H3K4Me3 peaks were detected at similar levels in shEWS-FLI1 versus shMock cells (Fig 2A). Thus, we conclude that the EWS-FLI1 oncogene yields high levels of slowed or stalled replication forks in the Ewing sarcoma cells, and a 50% reduction in the level of EWS-FLI1 greatly suppresses these defects, while also greatly reducing the level of transcription. The presence of higher transcriptional levels and fork slowing/stalling in the shMock TC32 cells, was suggestive of TRCs. To address this hypothesis, we performed a peak overlapping analysis between the PCNA and RNAPII signals (Fig 2F). A fraction of the PCNA peaks did not overlap with the RNAPII peaks (Fig 2F). This could reflect the fact that RNAPII is present, but below the threshold of the SEACR peak-calling algorithm, or that some replication fork slowing/stalling events occur through transcription independent mechanisms. Importantly, our analysis did identify 2282 97 RNAPII-PCNA overlapping regions in the wild type TC32 cells (Fig 1B and S1A). These overlaps, which we named "RNAPII-PCNA hotspots", support the existence of TRCs. Since knocking down the EWS-FLI1 expression essentially abolished the PCNA peaks, we observed no RNAPII-PCNA overlapping peaks for the shEWS-FLI1 cells (Fig 2A). Thus, to more directly assess the extent to which EWS-FLI1 knockdown might mitigate the identified TRCs, we quantified the total number of RNAPII or PCNA pulldown reads (signal intensity) within RNAPII-PCNA hotspots. In the shEWS-FLI1 cells, we found a modest decrease in the RNAPII signal intensity, compared to the shMock cells (Fig 1B and C). In contrast, the PCNA signal was dramatically decreased (Fig 1B and C). Collectively, these data argue that EWS-FLI1 expression causes TRCs in these Ewing sarcoma cells. To validate these EWS-FLI1 induced TRCs events at the cellular level, we utilized the proximity ligation assay (PLA), which yields foci when two proteins of interest are within 40nm range (Söderberg et al., 2006). For this experiment, we induced EWS-FLI1 expression in the euploid normal human primary neonatal fibroblasts (NHDF-Neo), as previously described in (Su et al., 2021), yielding a level comparable to that of the Ewing sarcoma cell line (Fig 1A). We then probed for PCNA and RNAPII PLA foci under this acute oncogene-induced condition. The single PCNA or RNAPII antibodies yielded similar low background staining in the NHDF-Neo cells without or with EWS-FLI1 (Fig 1E). When PCNA and RNAPII antibodies were included together, the average PLA foci number was still minimal in the parental NHDF-Neo cells (vector) but showed a highly significant increase in the EWS-FLI1 expressing cells (Fig 1D and E). Consistent with this finding, knockdown of EWS-FLI1 in the TC32 Ewing sarcoma 98 cells reduced the level of RNAPII-PCNA PLA foci by ~2 fold. (Fig 1D and F). These data show that EWS-FLI1 expression is sufficient to induce TRCs in otherwise normal diploid cells, and reduction of its levels reduces TRC levels in the Ewing sarcoma cells. TRCs can occur in both head-on and co-directional orientations. Although R- loops are mainly caused by head-on TRCs, both head-on and co-directional TRCs impair replication fork progression and are prone to causing DNA damage (Fig 1G; (Dutta et al., 2011; Hamperl et al., 2017; Lang et al., 2017; Merrikh et al., 2011). To determine the direct effect of EWS-FLI1 on replication forks, we utilized a DNA combing assay to assess fork progression in the primary NHDF-Neo with or without acute induction of EWS-FLI1. Specifically, we performed a pause-chase labeling with the thymidine analogs 5-Iodo-2′-deoxyuridine (IdU) and subsequently 5-Chloro-2′- deoxyuridine (CldU; Fig 1H). We calculated the replication fork speed of the unidirectional bicolored forks (Fig 1H) and found that this was significantly reduced upon induction of EWS-FLI1 compared to the vector control (Fig 1I), indicating that EWS-FLI1 slows down replication. To probe for replication fork stalling, we measured the ratio of IdU versus CldU fiber length. Undisturbed replication fork progression leads to roughly equivalent IdU and CldU lengths, whereas frequent fork stalling results in unequal labeling (Fig 1H). We found that the cells with EWS-FLI1 induction exhibit a greater variation in the relative IdU and CldU track lengths, as indicated by the large deviation from 1 for the IdU:CldU ratio (Fig 1J and 2G). We conclude that the expression of EWS- FLI1 fusion oncoprotein significantly slows down replication fork progression and increases fork stalling. This results from EWS-FLI1 induced replication stress, which TRCs clearly contribute to. 99 Figure 2. Validation of transcription-replication conflicts in EWS-FLI1 expressing cells. A. Summary of the number of RNAPII, PCNA, and RNAPII-PCNA overlapping CUT&RUN peaks identified by SEACR algorithm in TC32 cells harboring shMock or shEWS-FLI constructs. Peaks from H3K4Me3 pulldown samples were used as a positive experimental control in both conditions. B. Comparison between peaks of RNAPII associated region in TC32 cells identified by either CUT&RUN followed by SEACR analysis or ChIP-seq followed by MACS2 analysis (Gorthi et al. 2018). The number and percentage overlapping peaks are shown. C. Genome-wide heat maps summarizing total number of read counts +/-5kb around RNAPII or PCNA CUT&RUN peak summit in TC32 cells harboring a mock shRNA control (shMock) or an EWS-FLI1 shRNA (shEF). Both types are within the same genomic coordinates. The heat maps were centered by the peak summit of the indicated proteins, with +/-5kb plotted region around each summit. The read counts were ranked from top to bottom in a descending order in each condition. D. Percentage of proliferating cells after 1 hour of EdU pulse labeling in shMock or shEWS-FLI (shEF) TC32 cells. At least 100 cells were measured in each condition. 100 mean ± standard error of the mean (SEM); n=3, independent biological replicates. n.s., no significance, Two-tailed unpaired t-test. E. PCNA and RAD21 protein levels in the TC32 cells. GAPDH was used as a loading control. n=3, representative pictures are shown. shMock: mock shRNA; shEF: EWS- FLI1 shRNA. Numbers on top indicate the degree of expression relative to the shMock control. F. Venn diagram showing RNAPII and PCNA overlapping CUT&RUN peaks in shMock TC32 cells. G. The raw values of IdU:CldU ratio measurements in Figure 1J. The grey line indicates a IdU:CldU ratio of 1 (i.e., IdU tract length = CldU tract length). At least 100 intact bidirectional fibers were analyzed. V: Vector, EF: EWS-FLI1. RAD21 is enriched at the TRCs and is recruited to stalled replication forks We previously demonstrated that increased levels of RAD21 mitigate EWS-FLI1 induced replication stress to promote oncogenic growth in Ewing sarcoma (Su et al., 2021). Given this, we wondered whether RAD21 might be recruited to TRC regions where stalled replication forks are accumulated. To investigate the possible association of RAD21 with EWS-FLI1 induced TRCs, we used the TC32 Ewing sarcoma cell line. TC32 is trisomic for chromosome 8 and harbors an additional copy of RAD21 (Fig 4A). Initially, we used CUT&RUN to determine whether RAD21 was enriched at the TRC regions. RAD21 functions in the cohesin complex to tether the sister chromatids together until the onset of anaphase (Peters et al., 2008). However, induction of DNA damage leads to RAD21 enrichment to DNA damage sites (Ünal et al., 2004; Ünal et al., 2008; Wu et al., 2012). Thus, we would expect its CUT&RUN signal to be spread evenly throughout the genome under normal conditions and to coincide with the PCNA peaks under DNA damage conditions caused by stalled replication forks, and specifically to the RNAPII-PCNA hotspots if it is recruited to TRCs. Our data fit these predictions. In the shMock TC32 cells, we detected 10933 RAD21 CUT&RUN peaks, 42% (4578/10933) of which significantly overlapped with the PCNA peaks (Fig 4B). This 101 co-enrichment of RAD21 and PCNA at specific genomic sites is entirely consistent with the existence of EWS-FLI1 induced replication stress and DNA damage. Notably, the knockdown of EWS-FLI1 did not significantly alter RAD21 protein levels, as measured by western blot (Fig 2E) but, in concert with the loss of PCNA peaks (Fig 2A), RAD21 was no longer present in peaks but instead showed relatively even distribution across the genome in the shEWS-FLI1 TC32 cells (Fig 4B). These results argue that EWS- FLI1 causes RAD21 to accumulate at genomic sites of oncogenic DNA damage. To address whether RAD21 is present at TRCs in the shMock TC32 cells, we considered the RNAPII-PCNA hotspots. Peak overlapping analysis showed that about 49% (1123/2282) of these contained RAD21 peaks (Fig 2A and 4B), and thus we hereby refer to these as RNAPII-PCNA-RAD21 hotspots. This suggested a clear association of RAD21 to the TRC events (Fig 4B). In contrast, in the shEWS-FLI1 TC32 cells, the SEACR algorithm found no RNAPII-PCNA-RAD21 hotspots due to the abolishment of both PCNA and RAD21 peaks (Fig 4B). We also quantified the total number of reads for RAD21 within the RNAPII-PCNA hotspots in the shEWS-FLI1 cells, and found that these were greatly attenuated compared to the control shMock cells (Fig 4C). Similar results were obtained when we plotted the total read counts centered at peak summits within the RNAPII-PCNA-RAD21 hotspot coordinates, confirming a reduction in both PCNA and RAD21 total read signal intensity in the shEWS-FLI1 versus the shMock TC32 cells (Fig 3B). Altogether, these observations argue that RAD21 is enriched at the TRCs regions that are induced by EWS-FLI1. 102 Figure 3. RAD21 enriched at the TRCs is recruited to stalled replication forks. A. A representation of RNAPII-PCNA-RAD21 overlapping peak region in the TC32 cells harboring a mock shRNA control (shMock) or an EWS-FLI1 shRNA (shEWS- FLI1). All pictures were within the same coordinates. Peak height represents the sum of read counts from two independent biological replicates. The height of each track is 35 read counts. B. Heat maps summarizing total number of read counts for RNAPII, PCNA, and RAD21 signals within the RNAPII-PCNA-RAD21 overlapping region in the shMock or the shEWS-FLI1 TC32 cells. All types are within the same genomic coordinates. The heat maps were centered by the peak summit of the indicated proteins, with +/- 5kb plotted region around each summit. The read counts were ranked from top to bottom in a descending order in each condition. C. Illustration of the Tue/Ter stalled replication fork system. The red half-arrows indicate qPCR primer pairs for ChIP enrichment detection. D. ChIP analysis of RAD21 at both upstream (-128bp) and downstream (+109bp) of the Tus/Ter locus (See methods). mean ± SEM; n= 3 independent biological replicates. ***, p < 0.001, Unpaired t test. TRCs induce stalled replication forks, and in yeast, the RAD21 homolog, Scc1, is physically recruited to stalled replication forks (Tittel-Elmer et al., 2012). Whether mammalian RAD21 is similarly recruited to the TRC-induced stalled forks remains unknown. To address this question, we used a mouse embryonic stem (ES) cell line 103 containing the inducible Escherichia coli Tus/Ter stalled replication fork system in which an inducible Tus (Tus-F140A, a Tus mutant with higher binding affinity) targets the Ter site and causes a site-specific replication fork stalling [Fig 3C; (Willis et al., 2014, 2018)]. Using chromatin immunoprecipitation (ChIP) and subsequent qPCR analysis (Panday et al., 2021) with two distinct sets of primers, we found a dramatic and significant increase in RAD21 association at the integrated Ter sites in the presence of induced Tus expression and thus replication fork stalling, compared to the un-induced state (Fig 3D). This provided strong evidence that RAD21 is recruited and highly enriched at the stalled replication forks, which can be caused by TRCs and/or other genotoxic processes. Moreover, since this RAD21’s role is observed in otherwise normal ES cells, it appears to be broadly relevant, rather than unique to EWS-FLI1. 104 Figure 4. RAD21 enrichment at TRCs assessed by CUT&RUN assay. A. Karyotype of wild-type TC32 Ewing sarcoma cells by low-coverage DNA whole genome sequencing. B. Summary of RAD21, PCNA-RAD21, and RNAPII-PCNA-RAD21 overlapping CUT&RUN peaks called by the SEACR algorithm. C. Heat maps for RAD21 total number of read counts within the RNAP-PCNA-RAD21 overlapping region in the shMock or shEWS-FLI1 (shEF) TC32 cells. Both types are 105 within the same genomic coordinates. Total read counts are as indicated by color gradient. Increased RAD21 attenuates TRCs and promotes fork progression under EWS- FLI1 induced stress The experiments above show that RAD21 is enriched at sites of DNA replication damage, and specifically at TRCs and TRC-induced stalled replication forks. We have previously found that over-expression of RAD21, like trisomy of chromosome 8, acts to suppress the replication stress caused by EWS-FLI1 in Ewing sarcoma. Given this, we next asked whether increased RAD21 is able to promote resolution of EWS-FLI1 induced TRCs and whether this reflects enhancement of replication fork progression. For this, we introduced RAD21, or a vector control, into the euploid human primary neonatal fibroblasts expressing EWS-FLI1 (NHDF-Neo-EWS-FLI1; Fig 5A). Notably, the RAD21 levels in the over-expressing cells are 140% of those of the vector control (Fig 5A), and thus appropriately models the increase in RAD21 levels caused by trisomy of chromosome 8. First, we used these cells to assess the RNAPII-PCNA interaction with the PLA assay. We found that the increase in RAD21 significantly decreased the levels of RNAPII-PCNA PLA foci, suggesting a reduction of TRCs (Fig 5B and C). Second, we used the DNA combing assay to probe for replication events. RAD21 overexpression significantly increased the replication fork speed in the EWS-FLI1 expressing cells, such that this reached the replication fork speed in the vectors-only control cells (Fig 5D). This establishes RAD21's role in enhancing fork progression and its ability to mitigate the slowing of replication forks that is triggered by EWS-FLI1. Notably, RAD21 also increased replication fork speed, compared to the empty vector control, when 106 expressed in NHDF-Neo cells that lack EWS-FLI1 (Fig 5D). We believe that this reflects mitigation of the basal level of replication stress in these primary cells. Our analysis of the DNA combing assay data further showed increased RAD21 levels significantly decreased the deviation in the IdU to CldU ratio in the EWS-FLI1 expressing NHDF-Neo cells (Fig 5E and 6A). Thus, RAD21 also acts to suppress the frequency and/or persistence of stalled replication forks. 107 Figure 5. Increased RAD21 attenuates TRCs and promotes fork progression under EWS-FLI1-induced stress. A. RAD21 and EWS-FLI1 protein levels in NHDF-Neo. Vinculin was used as a loading control. Numbers on top indicate the degree of over-expression relative to the shMock control. n=3, representative pictures are shown. 108 B-C. Representative images (B) and quantification (C) of RNAPII-PCNA PLA foci in the EWS-FLI1 induced NHDF-Neo. PLA foci (in red) and DNA (in blue) are shown. Scale bar, 10 μm. At least 60 cells were scored in each condition. ****, p < 0.0001, two-tailed nonparametric two-group Mann–Whitney U-test. D-E. Measurement of replication fork speed (D) and the distribution of IdU:CldU ratio (E) in the NHDF-Neo harboring the indicated constructs. Fork speed was calculated as the IdU-CldU labeled fiber length/labeling time (40mins). V,V’- Vector +Vector’; EF,V’- EWS-FLI1+Vector’; V,RAD21- Vector + RAD21; EF,RAD21 -EWS- FLI1+RAD21. **, p < 0.01; *, p < 0.05, one-Way ANOVA test. The distribution of IdU:CldU ratio was plotted and the Gaussian normal distribution fit curve is shown. The variances between the indicated two conditions were compared using F-test: V +V’ - EF+V’,Vector +Vector’ vs EWS-FLI1+Vector’, p < 0.001; EF+V’ vs EF+RAD21, p < 0.001; V+ V’- EF+RAD21, n.s. At least 100 intact DNA fibers were analyzed in each condition. F-G. Measurement of replication fork speed (F) and the distribution of IdU: CldU ratio (G) in the shMock or shRAD21 TC32 cells. Same methods and analysis were used as indicated in (D) and (E). Fork speed: ****, p < 0.0001, two-tailed nonparametric two-group Mann–Whitney U-test. Variance of IdU:CldU distribution were compared using F-test: ***p < 0.001. At least 100 intact DNA fibers were analyzed in each condition. H. Illustration of the direct-repeat GFP (DR-GFP) system to assess I-SceI -induced homologous repair (HR; Nakanishi et al., 2011; Pierce et al., 1999). I-SceI induced double stranded break by HR can lead to gene conversion and generate a functional GFP gene. I. Quantification of HR frequency in the shMock or shRAD21 U2OS cells. Cells harboring a catalytically dead endonuclease (I-SceI-D44A) were used as a negative control. mean ± SEM; n= 3 biological replicates. **, p < 0.01, unpaired t test. As a complementary approach to RAD21 overexpression, we also examined the consequences of RAD21 knockdown in the TC32 Ewing sarcoma cell line. In this context, we found a significant decrease in replication fork speed as well as a significantly larger deviation from 1 for the CldU to IdU ratio (Fig 5F, G and 6B), indicating that replication forks are slowed, and stall more frequently, when RAD21 protein levels are reduced. In conclusion, RAD21 is present at TRCs and increases or decreases in its levels inversely impact replication fork progression in response to replication stress. 109 Persistent fork stalling can cause irreversible replication fork collapse and generate one-ended double-stranded breaks (DSBs) at the fork lesion. The DSBs generated during S phase can be repaired through homologous recombination (HR) using sister chromatids as templates (Scully et al., 2019). RAD21 (and its yeast homolog Scc1) has been shown to enable HR repair of DSB and to prevent end-joining (EJ) repair (Cortés‐Ledesma & Aguilera, 2006; Gelot et al., 2016; Kim et al., 2002; Potts et al., 2006; Sjögren & Nasmyth, 2001; Ström et al., 2004; Ünal et al., 2004). Given this, we hypothesized that the recruitment of RAD21 to TRCs, and sites of replication damage more broadly, might enable repair of DNA damage by promoting HR. To address this, we employed an established system to measure HR-mediated repair frequency (Nakanishi et al., 2011; Pierce et al., 1999). This involves a direct-repeat GFP (DR-GFP) construct, integrated into the U2OS cells, that contains an SceI cleavage site and thus allows assessment of HR-mediated gene conversion in response to induction of the I-SceI endonuclease and DSBs through quantification of cells that generate a functional GFP (Fig 5H). We also included a catalytically inactive I-SceI endounuclease (I-SceI-D44A), to ensure the precision of our assay. Whole genome sequencing karyotype analysis established that these U2OS cells are tetraploid for the RAD21 gene (sequencing data in SRA). Therefore, we used an shRNA to decrease the RAD21 expression in these U2OS cells. This led to an evident reduction of RAD21 protein, to ~40% of the control cell level, without significantly altering the cell cycle profile or percentage of S phase cells within the experimental timeframe (Fig 6C-E). Consistent with our hypothesis, RAD21 knockdown significantly reduced the percentage of GFP positive cells arising from DSB induction (Fig 5I), indicating compromised HR efficiency. 110 In contrast, the GFP signal remained at similarly low in the shRAD21 and shMock cells with the catalytically dead I-SceI mutant (Fig 5I). This observation suggests that RAD21 mitigates replication stress in a dose-dependent manner by promoting HR repair of DNA breaks at the impaired replication forks. 111 Figure 6. RAD21 reduces replication fork stalling and promotes homologous recombination. A-B. The raw values of IdU:CldU ratio measurements in Figure 5E and G in the indicated cells. The grey line indicated a IdU:CldU ratio of 1 (i.e. IdU tract length = CldU tract length). At least 100 intact bidirectional fibers were analyzed. V,V’- Vector +Vector’; EF,V’- EWS-FLI1+Vector’; V,RAD21- Vector + RAD21; EF,RAD21 -EWS- FLI1+RAD21. C. RAD21 protein levels in the U2OS cells harboring indicated constructs. GAPDH was used as a loading control. n=2, a representative picture is shown. D-E. Exponentially growing U2OS cells harboring indicated constructs were pulse labeled with EdU for 30minuts. The cells were then fixed and permeabilized for staining for EdU and DNA (by DAPI). The cell cycle profile was analyzed by flow cytometry (D) and determined by the quantification of the percentage of cells in each cell cycle phase with 3 independent biological replicates in each condition (E). At least 9000 cells were analyzed per condition per replicate. Representative flow 112 cytometry plot images were shown. No significance was called between the S-phase cells (red bars) between all four conditions using one-Way ANOVA test. RAD21 interacts with DNA damage initiation proteins upon EWS-FLI1 induction Although the importance of RAD21 in repair of DNA damage has been established (Ünal et al., 2004; Ünal et al., 2008; Wu et al., 2012), the molecular mechanism by which RAD21 mediates this remains unclear. In particular, the in vivo interaction between RAD21 and proteins other than its known cohesin complex partners remains unexplored. We hypothesized that RAD21 promotes DNA repair by interacting with one, or more, proteins involved in the DNA damage repair pathway. To test this possibility, we used an in vivo protein-proximity labeling system, called TurboID, which ligates biotin to proteins located within the 10nm range, while more distal proteins remain unmodified (Cho et al., 2020; Fig 7A). We generated a lentiviral construct that fused TurboID to the C-terminus of RAD21, and then introduced multiple copies of RAD21-TurboID via lentiviral transduction into the euploid NHDF-Neo cells carrying either the EWS-FLI1 or the control empty vector. We used serum starvation to synchronize the cells in G1 phase, induced cell cycle re-entry via the re-addition of serum, while also adding biotin into the culture medium, and then harvested 24 hours later (Fig 7B). Flow cytometry analysis showed that this experimental scheme greatly increased the percentage of cells that had passed through S phase during the biotinalyation process, compared to unsynchronized cells (Fig 7C and D), thereby enriching for S phase-specific RAD21-interacting proteins. Interestingly, the presence of RAD21-TurboID shortened the time needed for the EWS-FLI1 expressing cells to reach G2/M (Fig 7C and D; compare the percentage of S and G2/M phase cells at 18h and 113 24h post serum re-addition). This is entirely consistent with our finding that RAD21 overexpression mitigates the EWS-FLI1-induced replication stress, and argued that RAD21’s activity is maintained in the present of the TurboID moiety. Accordingly, immunofluorescence staining confirmed that the TurboID tag did not alter RAD21’s nuclear localization (Fig 7E). Most importantly, addition of biotin for 24 hours yielded numerous biotinylated protein species in both EWS-FLI1 and vector control cells that were dependent upon the presence of the RAD21-TurboID protein (Fig 7F). 114 Figure 7. Validation of the RAD21-TurboID system. A. An illustration of RAD21-TurboID proximity-based biotinylation system to identify RAD21 interacting proteins. B. The scheme depicting timeline for RAD21-TurboID experiment in the NHDF-Neo to identify the RAD21 interacting partners. C-D. Cells were pulse labeled with EdU for 30mins at the indicated time points after serum starvation (SS) release. The cell cycle profile was analyzed by flow cytometry (C) and the quantification of percentage of S/G2/M phase cells was presented in (D, mean of 2 biological replicates). At least 9000 cells were analyzed per condition per replicate. Representative flow cytometry plots are shown. E. Representative Immunofluorescence images for the cells with or without Rad21- TurboID. Cells were fixed at the sample collection time point as indicated in (B). Cells without Rad21-TurboID (uninfected) were used as a negative control. DNA (DAPI, blue); Rad21 (anti-RAD21, green); Rad21–TurboID (anti-V5, red); NeutrAvidin marked biotinylated proteins (magenta). Scale bars, 10 μm. F. Measurement of biotinalytion activity in the Rad21-TurboID cells. Cells without Rad21-TurboID (uninfected) were used as a negative control. Cells were generated using procedure in (B) and protein was collected at the sample collection point. Equal amount of whole-cell protein lysates was analyzed by probing with 115 streptavidin–IRD800. Anti-V5 was used to probe for Rad21–TurboID. GAPDH was used as a loading control. N=2, representative pictures are shown. G. Biological replicates of IP-MS results using the same criteria as indicated in Figure 8A and D. Rep: independent biological replicate. Grey dotted line denotes the threshold used for identifying hits (1.5-fold enrichment compared to un-infected no RAD21-TurboID control cells. H. Western blot probing for c-Myc protein level in the NHDF-Neo harboring a Rad21- TurboID and a doxycycline inducible MYC or a vector control. Cells were induced for MYC expression for 48 hours before subjected to protein analysis. Results were consistent between replicates and representative pictures were shown (n=2). Having validated the proper function of RAD21-TurboID in our system, we used streptavidin to recover the biotinaylated proteins from three independent biological samples of RAD21-TurboID expressing euploid NHDF-Neo cells harboring either EWS- FLI1 or the vector control. We then used mass spectrometry (MS) to identify candidate RAD21-interacting proteins. We used uninfected cells (i.e. without RAD21-TurboID expression) to set the background for non-specific protein pulldown, and 1.5-fold enrichment as the threshold to identify candidate RAD21 interacting proteins. Gratifyingly, for both the EWS-FLI1 and vector control cells, the results were highly consistent across the biological replicates (Fig 7G and Supplemental table 1) and we further narrowed the list of hits to proteins that were detected in at least two of the three replicates. Using these criteria, we identified 91 unique protein hits in the vector control and 143 hits in the EWS-FLI1 expressing cells (Supplemental table 1). Remarkably, 89 hits were represented in both conditions. Furthermore, more proteins interacting with RAD21 in the EWS-FLI1 cells than in the cells harboring vector controls (Fig 7G and Supplemental table 1), consistent with the notion that RAD21 has an expanded role in the presence of this onco-protein, targeting DNA damaging sites and enabling HR. Importantly, the identity of the associated proteins directly supports this notion. First, we 116 found that cohesin complex subunits, and also associated proteins known to interact with RAD21, showed similar significant enrichment in both vector control and EWS-FLI1 expressing cells (Fig 8A and B). This included the cohesin subunits SMC1 and SMC3, the cohesin binding factor PDS5A, as well as the cohesin loader NIBPL and the cohesin remover WAPL (Fig 8A and B). To validate these mass spectrometry results, we immunoprecipitated streptavidin from fresh cell samples, and then screened for SMC1 and SMC3 by western blotting (Fig 8C). SMC1 and SMC3 were recovered at similar strong levels in both the EWS-FLI1 and vector control NHDF-Neo cells expressing RAD21-TurboID, but not the non-expressing cells, even though SMC1 and SMC3 were present at comparable levels in the whole cell lysates (Fig 8C). This validates the specificity and efficiency of RAD21-TurboID in identifying known targets, and is consistent with the notion that the RAD21-cohesin complex plays a similar role in the presence or absence of EWS-FLI1. We then turned our attention to RAD21-associated proteins identified by mass spectrometry as being significant enriched in the EWS-FLI1 and not the vector control cells (Supplemental table 1). Amongst this list were proteins known to be responsible for initiating DNA damage repair (Fig 8D and E). This includes two well established regulators of HR repair (Liao et al., 2018): MRE11, the nuclease responsible for resection upon DSB induction, and poly (ADP-ribose) polymerase 1 (PARP1), which protects the DNA breaks and recruits repair protein to the site of damage (Fig 8D and E). Importantly, we confirmed these interactions by streptavidin IP- western blotting, showing 2-fold enrichment for MRE11 and 1.4-fold enrichment for PARP1 in cells expressing EWS-FLI1 compared to the vector control (Fig 8F). This is consistent with the notion that RAD21 interaction with MRE11 and PARP1 could enable 117 repair of DSBs caused by EWS-FLI1 induced replication stress. Interestingly, previous studies have shown that PARP1 promotes the recruitment of MRE11 to the stalled replication fork for efficient processing of DNA break ends (Bryant et al., 2009). Given the increased interactions between RAD21 and MRE11 or PARP1 in the EWS-FLI1 expressing cells, we hypothesize that RAD21 acts to coordinately stabilize the stalled replication fork and recruit proteins involved in the early initiation stage of HR-mediated repair. This would enable both efficient DNA damage repair and/or fork restart. To investigate the generality of RAD21’s role in interacting with DNA repair initiation factors in stress conditions, we induced expression of another oncogene c-Myc in the NHDF-Neo cells (Fig 7H), and then followed the RAD21-TurboID introduction and cell synchronization experimental scheme (Fig 7B). In a similar manner to EWS-FLI1, the streptavidin IP-western blotting experiments revealed an evident increase in RAD21-MRE11 and RAD21-PARP1 interactions in the oncogene-expressing cells compared to the vector controls (Fig 8G). This experiment reinforces the notion that the role of RAD21 in associating with sites of DNA damage, including TRCs, and promoting DNA damage repair and/or replication fork restart is not limited to EWS-FLI1 induction, but a general mechanism for mitigating replication stress caused by stress conditions including those induced by many other oncogenes. 118 Figure 8. RAD21 interacts with DNA damage initiation proteins upon EWS-FLI1 induction. A. Proteins identified by streptavidin IP-MS in NHDF-Neo carrying indicated constructs (Vector control, blue; EWS-FLI1, orange) Proteins in the cells harboring RAD21- TurboID construct (y-axis) were plotted against those in the cells without RAD21- TurboID (x-axis). Grey dotted line denotes the threshold used for identifying hits (1.5-fold enrichment compared to un-infected (no RAD21-TurboID) control). The known RAD21 cohesin-interacting partners were marked in dark blue triangles. The results were consistent between three replicates (the results for the two other replicates in Fig 7G). B. Summary of proximity-based biotinylation with RAD21-TurboID for known RAD21 interacting proteins. Individual peptide reads in all three biological replicates are shown. All 0 peptide detections were denoted as 1 for normalization purpose. C. Streptavidin IP-Western blot experiment probing for enrichment of cohesin subunits SMC1A and SMC3 in the NHDF-Neo. Equal amount of input proteins was used in the IP experiment. GAPDH and Tubulin were used as negative controls. n=3, representative images were shown. V: vector control; EF: EWS-FLI1. 119 D. Same streptavidin IP-MS plots were presented as in (A) with identification of PARP1 and MRE11 shown as red triangles. E. Summary of proximity-based biotinylation with RAD21-TurboID for DNA damage repair proteins. Individual peptide reads from all 3 biological replicates are shown. All 0 peptide detections were denoted as 1 for normalization purpose. F-G. Streptavidin IP-Western experiment probing for enrichment of PARP1 and MRE11 in the NHDF-Neo upon EWS-FLI1 (F) or MYC (G) induction. Equal amount of input proteins was used in IP experiment. N=3, representative images were shown. V: vector control; EF: EWS-FLI1. DISCUSSION The sources of EWS-FLI1 induced replication stress EWS-FLI1 is a known oncogene causing DNA damage and replication stress (Gorthi et al., 2018; Koppenhafer et al., 2020; Lessnick et al., 2002; Nieto-Soler et al., 2016; Su et al., 2021). Here, we are the first to show physical evidence of EWS-FLI1 induced transcription-replication conflicts (TRCs), which serve to impair replication fork progression and cause fork stalling, and thus create replication stress. We identified 2282 RNAPII-PCNA hotspots, marking the regions of TRCs, that are widely spread throughout the genome of the TC32 Ewing sarcoma cells. Partial knockdown of EWS- FLI1 almost completely abolishes TRCs, reduces the total read signals within the RNAPII-PCNA hotspots and largely eliminates all of the PCNA peaks identified by the CUT&RUN and SEACR analysis. This establishes the key role of EWS-FLI1 in inducing TRCs. Using DNA combing assays, we further showed that replication fork slowing and stalling are amongst the significant outcomes upon EWS-FLI1 expression. We noted that the induced TRCs, as defined by RNAPII-PCNA hotspots, only account for 20% of total PCNA SEACR peaks. This suggests that EWS-FLI1 induced DNA damage events can occur through additional mechanisms. Previous studies showed that EWS-FLI1 instigates R-loop formation and impaired BRCA-mediated HR 120 for DNA damage repair (Gothi et al 2018). This suggests that the PCNA-only peaks could result from other types of EWS-FLI1 associated replication errors. Moreover, PCNA has also been indicated in bypassing DNA damage at the stalled replication forks, which includes translesion synthesis and error-damage bypass (Boehm et al., 2016). The role of the DNA damage bypass mechanism in the cells with EWS-FLI1 induced replication stress needs further investigation, but this could also induce accumulation of PCNA at stalled replication forks independent of BRCA-mediated repair. In this situation, PCNA’s accumulation and enrichment at the stalled forks could be TRC independent. The role of RAD21 in mitigating oncogene-induced replication stress We previously showed that gain of additional RAD21 drives trisomy 8 in Ewing sarcoma and promotes oncogenic growth by mitigating replication stress (Su et al., 2021). Here, we have specifically focused on investigating RAD21’s function in the context of EWS-FLI1 induced replication stress. This showed that RAD21 is enriched at the TRCs sites in the Ewing sarcoma cells, and is physically bound to the stalled replication fork in mammalian cells. Moreover, our data show that RAD21 influences fork restart and repair in a dose-dependent manner. Specifically, RAD21 overexpression significantly suppressed TRC events, rescued replication fork slowing and reduced fork stalling caused by EWS-FLI1, while RAD21 knockdown has the opposing effect. TRC can occur in two confirmations: a head-on TRC accumulates R-loops and triggers ATR checkpoint response for stalled replication fork whereas a co-directional 121 TRC resolves R-loop but causes ATM checkpoint for DSBs (Hamperl et al., 2017). Both cases can cause damage to the replication forks. It is possible that increased RAD21 helps reduce TRC by directly promoting R-loop resolution, and this needs to be tested. Whether this occurs, or not, our data suggest that RAD21 helps to resolve TRCs and rescue fork progression by promoting HR-mediated fork lesion repair (Fig 5). Our CUT&RUN analyses in EWS-FLI1 expressing cells argue that RAD21 mitigates other types of replication stress beyond TRC-induced problems. Specifically, this identified enrichment of chromatin bound RAD21 at specific genomic regions, approximately 40% of which overlap with PCNA signals and 10% overlap with TRCs, as marked by the RNAPII-PCNA hotspots. Indeed, many of the RAD21 peaks are located at intergenic region where transcription is rare. The cause of the RAD21 SEACR peaks that do not overlap with PCNA remains to be elucidated. However, it is entirely possible that these reflect collapsed forks, where PCNA has already disassociated. Interaction of RAD21 with DNA damage repair proteins Using the TurboID system, our study also yielded insights into the RAD21 interactome. This confirmed cohesin complex complexes as RAD21-associated and showed that these interactions occurred in both control and EWS-FLI1 expressing cells, consistent with their housekeeping role. It also revealed interactions between RAD21 and important DNA damage repair proteins, including MRE11 and PARP1, which revealed RAD21's role in promoting HR. Importantly, these interactions were specifically enriched in the EWS-FLI1 expressing cells, versus vector controls. These interactions connect RAD21 to the early stage of the HR repair process because MRE11 and 122 PARP1 are both key players in upstream sensing and recognition of break ends (Chaudhuri & Nussenzweig, 2017). Previous studies have shown that RAD21’s loading at the damage site depends on the signaling of MRE11 (Kim et al., 2002). We speculate that EWS-FLI1 expression increases the frequency of stalled replication forks and/or the DSBs, which signal DNA damage repair pathway, which recruits the RAD21/cohesin complex to the sites of damage to stabilize the stalled replication fork and facilitate efficient downstream repair through recruitment of repair proteins like MRE11 and PARP1. We propose that the loading of RAD21 at the stalled replication fork is a transient and dynamic process. First, previous studies under normal growth condition have not revealed the physical interaction of RAD21 with MRE11 or PARP1 (Panigrahi & Pati, 2012). Here, we used a more sensitive TurboID system which permanently biotinylated any protein on a proximity basis. This would enable detection of transient interactions between RAD21 and MRE11 or PARP1 resulting from EWS-FLI1 induction. Second, cohesin antagonists PDS5 and WAPL are known to be required for repair of damage fork. This created a conundrum of how cohesin loading and removal can both be required at the stalled forks for damage repair (Benedict et al., 2020; Carvajal- Maldonado et al., 2018; Frattini et al., 2017; Morales et al., 2020). The fact that our RAD21-TurboID experiments did not detect any interaction between RAD21 and late- stage DNA damage repair proteins offers a simple explanation to reconcile this existing conundrum; it suggests RAD21 is required only for the initiation of DNA repair but then needs to be unloaded from the break site to ensure the recruitment of downstream 123 repair protein, such as RAD51 and BRCA1/2, and to allow efficient homology search and strand invasion. Although we primarily focused on the role of RAD21 in mediating HR-mediated DSB repair, the TurboID experiment also identified increased interactions between RAD21 and repair initiation proteins associated with the nonhomologous end joining (NHEJ) pathway, including XRCC5 and XRCC6. Given the previously reported role of cohesin in preventing NHEJ process (Gelot et al., 2016), we speculated that these interactions may inhibit the function of NHEJ repair proteins and thus block damage repair by the NHEJ pathway. In this context, RAD21 could serve to limit potential genomic rearrangement by preventing error-prone damage repair in S phase cells experiencing replication stress. How RAD21 mediates such process could be an important next step for dissecting its role in promoting oncogenesis. Finally, while our study largely focused on the role of RAD21 in EWS-FLI1 expressing cells, our data argue that it plays a much broader role. First, we find that RAD21 also associates with the early repair proteins, MRE11 and PARP1, in response to expression of the c-Myc oncogene. Second, RAD21 accelerates replication fork speed in otherwise normal diploid fibroblasts. Third, RAD21 associates with DSB induced by the SceI endonuclease in the ES cells lacking any oncogene. Given its importance in mediating post-replicative repair of DSBs, we speculate that RAD21 could also play important roles in replication stress induced by other stimuli, such as reactive oxygen species (ROS), altered nucleotide metabolism, or unlicensed origin firing (Kotsantis et al., 2018). The details of RAD21 in mitigating these various replication 124 stresses need to be further determined to fully understand RAD21’s role in mitigating DNA damage and facilitating repair. MATERIALS AND METHODS Cell culture Neonatal normal human dermal fibroblasts (NHDF-Neo; Lonza; Cat# CC-2509) were cultured in Eagle’s minimum essential medium (EMEM, ATCC) supplied with 15% FBS. TC32 Ewing sarcoma cell line (a gift from Dr. Kimberly Stegmaier) was cultured in Minimum Essential Medium α (MEM-α, Invitrogen) supplied with 10% Hyclone FBS (Cytiva). U2OS DR-GFP cells were regularly maintained and cultured in Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen) supplied with 10% FBS. All culture media were supplied with penicillin/streptomycin (100 U/ml), and L-Glutamine (2 mM). Cells were grown at 37°C with 5% CO2 in a humidified environment. NHDF-Neo cells carrying either EWS-FLI1 or Vector control were induced for EWS-FLI1 expression for 48 hours with culture medium containing 1µg/mL of doxycycline prior to CUT&RUN analysis, combing assay or Western blot. Karyotypes of NHDF-Neo, TC32 and U2OS DR-GFP cells were confirmed by low-coverage whole genome sequencing methods described in (Su et al., 2021). All cell lines in this study were determined mycoplasma- negative using MycoAlert Mycoplasma Detection Kit (Lonza) by the MIT Koch Institute HTS Facility. 125 Lentiviral transduction and transfection Lentiviral constructs (Supplemental table 2) along with packaging plasmids pMD2.G (Addgene 12259) and psPAX2 (Addgene 12260) were transfected into 293FT cells (Thermo Fisher, Cat# R70007) using TransIT-LT1 transfection reagent (Mirus). The virus titer was estimated by Lenti-X GoStix Plus (TaKaRa) and target cells were infected at an MOI 3-5. CUT&RUN assay and sequencing TC32 cells were plated at ~50% confluency and grown overnight. Cells were then trypsinized and counted. 5x105 cells were carried into CUT&RUN analysis for each condition using the CUT&RUN kit (Cell Signaling; #86652) based on manufacturer’s instructions. The eluted DNA samples were purified with CUT&RUN DNA purification kit (Cell Signaling #14209). The concentration of purified DNA products were measured using an Agilent FemtoPulse prior to library preparation and were prepared into Illumina libraries using NEB UltraII chemistry (New England Biolabs) amplifying for 16 cycles. Final libraries were quality controlled using an Agilent Fragment Analyzer and the concentrations were confirmed by qPCR prior to sequencing on a high output NextSeq500 flowcell with 40bp paired-end reads (RTA version 2.4.11). CUT&RUN analysis Sequencing reads from CUT&RUN were mapped using bwa/0.7.12. backtrack algorithm with a hg38 referencing genome. Bam files from replication samples were merged using samtools/1.5 merge command. Bam files were sorted and indexed by 126 samtools/1.5 sort and index commands. The bam files were either visualized using Integrative Genomics Viewer (IGV, Broad Institute) or converted to bedgraph files using MIT IGB/BMC in house tools. The bedgraph files were further used for normalization and peak calling analyses by SEACR /1.3 (Meers et al., 2019). The bam files were also later visualized again by IGV for further quality control (QC) and confirmation. Peak overlapping was carried out by bedtools/2.29.2. Heatmaps were plotted using TIBCO Spotfire/10.10.3. TIBCO Spotfire input files were prepared by MIT IGB/BMC in house tools. Proximity ligation assay Cells grown on a coverslip were fixed with 4% paraformaldehyde (PFA) in PBS for 15 minutes at room temperature (RT). Then, the fixed cells were washed with PBS + 3% BSA (Sigma-Aldrich) once before incubated with 0.5% Triton X-100 (Sigma-Aldrich) in PBS for 15 minutes at RT. PLA assay was then performed using the Duolink In Situ Red Starter Kit (DUO92101) based on manufacture’s instruction. Samples were incubated for overnight at 4oC with primary antibodies at 1:400 dilution (each) in the Duolink Antibody Diluent. PLA assay was then performed using the Duolink In Situ Red Starter Kit (DUO92101) based on manufacture’s instruction. The following antibodies were used: anti-PCNA (Cell Signaling, #13110); anti-RNAPII (CTD4H8, Santa Cruz, Sc- 47701). DAPI was used to stain DNA. Images were taken using a Nikon Elipse 90i Fluorescent Microscope with a Plan Apo 20X/0.3 objective, ORCA-ER camera and NIS- Elements software. 127 DNA combing assay Cells were grown to 70% confluence on a 10cm dished under desired conditions. Freshly-made prewarmed medium with 5-Iodo-2'-deoxyuridine (IdU, 25μM; Sigma- Aldrich) was added to the cells and incubated for 20 minutes at 37oC with 5% CO2. Then IdU-containing medium was replaced with freshly-made prewarmed medium with 5-Chloro-2'-deoxyuridine (CldU, 250μM; Sigma-Aldrich) and incubated for another 20 minutes at 37oC with 5% CO2. This IdU/CldU pulse labeling procedure was also depicted in (Fig 1H). Then, cells were washed once with ice-cold PBS before harvesting by trypsinization. Cell suspension was kept in PBS on ice and were counted prior to agarose plug preparation. 75000 cells were embedded into 90μL of agarose plug per sample. Protein digestion and plug washes were then performed by using the Plug Preparation Kit from GenomicVision according to the manufacturer’s instructions. The combing experiment was done by Genomic Vision Molecular Combing service. DNA fibers were analyzed and the image of the fibers were stored using EasyScan software by FiberStudio (Genomic Vision). Western blot Cells were lysed in the RIPA buffer (Thermo Fisher Scientific) containing protease inhibitor and phosphatase inhibitor (PhosStop) cocktails (Roche). Protein extracts were quantified by Bradford assay (Bio-Rad) and equal amounts of denatured protein were loaded on NuPAGE 4-12% Bis-Tris gels (Thermo Fisher Scientific) based on the manufacturer’s instructions. Proteins were transferred to a PVDF or a nitrocellulose membrane. Blots were blocked for 1 hour at room temperature in OneBlock blocking 128 buffer (Genesee Scientific). Primary antibodies were incubated over night at 4°C. Fluorophore-conjugated secondary antibodies were blotted for 1 hour at room temperature. The biotinylated proteins in the TurboID experiment were detected using IRDye 800CW Streptavidin (1/1000, LI-COR BioScience) together with secondary antibody incubation. Protein signals were detected by ChemiDoc MP imaging system (Bio-Rad). Signals were quantified using the Fiji-ImageJ gel analysis software. The primary and secondary antibodies used in this study were listed in the Supplemental table 2. DR-GFP reporter repair assay The general DR-GFP assay was carried out as described in (Nakanishi et al., 2011; Pierce et al., 1999). The DR-GFP U2OS cells (ATCC, CLR-3455) were grown to 60% confluence on a 10cm cell culture dish. 15 μg of Plasmid containing either ISceI (Addgene 26477) or ISceI-D44A mutant (Addgene 59424) was transfected into the U2OS cells with TransIT-LT1 transfection reagent according to the manufacturer’s instructions. 48 hours after transfection, the medium containing transfection reagents was washed out and fresh medium was used to culture the cells for an additional 24 hours. The GFP-positive cells were assessed by using flow cytometry 72 hours post transfection. Flow cytometry for cell cycle analysis Cells were switched to medium containing 5-ethynyl-2’-deoxyuridine (EdU; 10 μM) 30 minutes prior to trypsinization. Trypsinized cells were fixed with 4% paraformaldehyde in 129 PBS for 15 minutes and permeabilized with 1% BSA, 0.5% TritonX-100 in PBS for 10 minutes. Cells were then incubated in the dark for 30 minutes with the labeling solution: CuSO4 (1mM), Alexa Fluor 647 Azide (1μM), and ascorbic acid (100mM) in PBS. Cells were then washed and resuspended in PBS containing 1% BSA, 0.5% TritonX-100, DAPI (1μg/mL) and RNAse A (100ug/ml). Stained cells were analyzed using a FACS LSR II analyzer and measured by BD FACSDiva software. Chromatin immunoprecipitation (ChIP) of RAD21 in Tus/Ter system The ChIP assay is performed as described in (Panday et al., 2021; Willis et al., 2018). The recombinant anti-RAD21 antibody (ChIP-grade; abcam, ab217678) was used for the ChIP pulldown of RAD21 protein associated with chromatin. Three independent replicate experiments were performed and RAD21 enrichment was analyzed as described in (Panday et al., 2021; Schmittgen & Livak, 2008) using 2- ΔΔCT methods. TurboID and streptavidin immunoprecipitation assays TurboID biotinylation-base labeling and sample preparation To identify RAD21 interacting partners, NHDF-Neo cells grown at ~20-30% confluency were transduced with lentiviral TurboID tagged RAD21. 24 hours after lentiviral transduction, cells were washed with pre-warmed (37oC) PBS for 2 times and serum free culture medium containing doxycycline (1µg/mL) was applied to cells for serum starvation. 48 hours after serum starvation, cells were released into pre-warmed (37oC) normal culture medium (10%FBS) containing doxycycline (1µg/mL) and biotin (100 μM). Cells were allowed to progress through the cell cycle in biotin-containing medium for 24 130 hours. At the time of sample collection, excessive biotin was washed out with ice-cold PBS for 5 times and cells were scrapped off and pelleted from the tissue culture plate. Cells were lysed with RIPA lysis buffer supplemented with 1× protease inhibitor cocktail (Roche) and PMSF (1mM). The general scheme of this protocol was outlined in Figure 4B. Immunoprecipitation (IP)-Western blot For each sample, 300 μl of MyOne streptavidin C1 dynabeads (Thermo Fisher Scientific) were used and these streptavidin-conjugated beads were washed twice with 1mL of RIPA buffer. 350 mg of whole protein lysate were then incubated with the streptavidin beads at 4°C overnight. After the protein enrichment on the beads, the beads were pelleted using a magnetic rack. The beads were washed twice with 1mL of RIPA lysis buffer (2 minutes at RT), once with 1 mL of KCl (1M, 2 minutes at RT), once with 1mL of Na2CO3 (0.1 M, ~10 seconds), once with 1mL of 2 M urea in 10 mM Tris- HCl (pH 8.0, ~10 seconds), and twice with 1mL of RIPA lysis buffer (2 minutes at RT). After the final wash, the beads were transferred to fresh tubes and small aliquots of protein-bound beads were taken for IP-Western blot experiments. For IP-Western analysis, the enriched protein was eluted from the beads by boiling each sample in 30 μL of 5×protein loading buffer supplemented with biotin (2mM) and DTT (20 mM) at 95 °C for 10 minutes. The corresponding inputs and eluted samples were resolved on NuPAGE 4-12% Bis-Tris gels (Thermo Fisher Scientific) based on the manufacturer’s instructions. The Western blot procedure was described as above. 131 TurboID IP-liquid chromatography/mass specrometry (LC/MS) The rest of the protein bound streptavidin beads were carried into mass spectrometry analysis. The protein-bound streptavidin beads were further washed with 1mL of 50 mM Tris-HCl (pH 7.5) followed by two washes with 200 μL of 2M urea in 50 mM Tris (pH 7.5) buffer. On-bead reduction, alkylation, and digestion were performed. Proteins were reduced with dithiothreitol (10mM, Sigma) for 1h at 56oC and then alkylated with iodoacetamide (20mM, Sigma) for 1h at 25oC in the dark. Proteins were then digested with modified trypsin (Promega) in ammonium bicarbonate (100mM; pH 8) at 25oC overnight. Trypsin activity was halted by addition of formic acid (99.9%, Sigma) to a final concentration of 5%. Peptides were desalted using C18 Spin columns (Pierce) then vacuum centrifuged. Right before injection on LC-MS the samples were re- suspended in 10 µl of 0.2% formic acid. 1 µl was injected on the LC-MS. Peptides were separated by reverse phase HPLC (Thermo Ultimate 3000) using a Thermo Scientific Easy-Spray HLPC column (50 cm of 2 µm C18) over a 60-minute gradient before nanoelectrospray using a Orbitrap Exploris 480 mass spectrometer (Thermo). Solvent A was 0.1% formic acid and solvent B was 80% MeCN/0.1% formic acid. The mass spectrometer was operated in a data-dependent mode. The parameters for the full scan MS were: resolution of 100,000 across 375-1600 m/z and maximum IT 25 ms. The full MS scan was followed by MS/MS. The mass spec acquired as many dependent scans as possible in a 2 second window with a NCE of 28 and dynamic exclusion of 20 s. Raw mass spectral data files (.raw) were searched using Proteome Discoverer 2.5 (Thermo) and Sequest. Sequest search parameters were: 10 ppm mass tolerance for precursor ions; 0.02 Da for fragment ion mass tolerance; 2 missed cleavages of trypsin; fixed 132 modification was carbamidomethylation of cysteine; variable modifications were methionine oxidation, acetyl on protein N-term, Met-loss on protein N-term, and Met- loss+acetyl on the protein N-term. EdU staining and analysis Cells were plated on coverslips and grown overnight. Cells were labeled with 5-ethynyl- 2’-deoxyuridine (EdU; 10 μM) for 1 hour before fixation. Cells were then permeabilized for EdU staining using Click-iT EdU Imaging Kit (Invitrogen) based on manufacturer’s instructions. Images were taken using a Nikon Elipse 90i Fluorescent Microscope with a Plan Apo 20X/0.3 objective, ORCA-ER camera and NIS-Elements software. Exposures times were determined by using the auto-exposure function of the NIS-Element software. Images were analyzed by using FIJI-ImageJ software. Immunofluorescent staining and microscopy Cells were plated onto fibronectin (10 μg/ml, SigmaAldrich)-coated coverslips at 50-70% confluency overnight and fixed at room temperature with 4% paraformaldehyde in PBS for 15 minutes. Cells were then permeabilized with 0.1% Triton X-100 in PBS for 10 minutes and blocked in 3% bovine serum albumin (BSA) in PBS for 40 minutes. Cells were incubated with primary antibodies for 90 minutes and with secondary antibodies for 45 minutes at room temperature in the dark. The biotinylated proteins were detected using homemade NeutrAvidin–Alexa Fluor 647 following the protocol indicated in (Cho et al., 2020) at a 1/1000 dilution together with secondary antibody incubation. The primary and secondary antibodies utilized in this study were shown in Supplemental 133 table 2. Hoechest was used to stain DNA. Images were acquired using a DeltaVision Ultra (60×) microscope and analyzed with Fiji-ImageJ software. ACKNOWLEDGEMENTS We thank the Swanson Biotechnology Center for help with the gene expression analysis. This work was supported by NIH grant CA206157 and GM118066 to A.A , who was an investigator of the Howard Hughes Medical Institute, the Paul F. Glenn Center for the Biology of Aging Research at MIT and the Ludwig Center at MIT’s Koch Institute for Integrative Cancer Research. XAS was supported through funding from the Virginia and D.K. Ludwig Fund for Cancer Research, and a Jane Coffin Childs Memorial Fellowship. RWW was supported through Bridge Project and the Department of Biology. 134 SUPPLEMENTAL TABLE S1: LIST OF RAD21-TURBOID TARGET PROTEINS Molecular Neo_Ctrl pLVX Vector pLVX_EF Fold change V/Ctrl Fold change EF/Ctrl protein ID Weight 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 MYH9 227 kDa 34 14 41 442 2219 2797 652 2555 711 13 158.5 68.2195122 19.1764706 182.5 17.3414634AHNAK 629 kDa 35 9 9 287 242 286 348 415 627 8.2 26.8888889 31.7777778 9.94285714 46.1111111 69.6666667 NUMA1 238 kDa 116 141 189 217 221 214 272 261 323 1.87068966 1.56737589 1.13227513 2.34482759 1.85106383 1.70899471 ACTB 42 kDa 33 38 43 112 268 431 124 351 92 3.39393939 7.05263158 10.0232558 3.75757576 9.23684211 2.13953488 IFI16 88 kDa 18 13 14 126 71 91 127 115 114 7 5.46153846 6.5 7.05555556 8.84615385 8.14285714 HNRNPM 78 kDa 91 24 39 124 46 61 138 73 64 1.36263736 1.91666667 1.56410256 1.51648352 3.04166667 1.64102564 SMC1A 143 kDa 3 6 3 81 84 113 115 118 134 27 14 37.6666667 38.3333333 19.6666667 44.6666667 MYL6 17 kDa 9 6 6 31 119 189 36 127 31 3.44444444 19.8333333 31.5 4 21.1666667 5.16666667 ZNF619 151 kDa 42 21 32 45 21 23 77 61 76 1.07142857 1 0.71875 1.83333333 2.9047619 2.375 SMC3 142 kDa 1 1 1 41 28 38 84 66 82 41 28 38 84 66 82 PDS5B 165 kDa 11 23 32 38 25 37 41 45 52 3.45454545 1.08695652 1.15625 3.72727273 1.95652174 1.625 NIPBL 316 kDa 1 1 1 42 34 48 44 44 51 42 34 48 44 44 51 GSN 86 kDa 1 1 1 27 51 71 36 59 27 27 51 71 36 59 27 MTA1 81 kDa 23 14 24 31 16 21 39 31 36 1.34782609 1.14285714 0.875 1.69565217 2.21428571 1.5RBM25 111 kDa 13 8 17 26 11 22 36 27 33 2 1.375 1.29411765 2.76923077 3.375 1.94117647 ADNP 124 kDa 14 9 14 21 21 22 31 28 27 1.5 2.33333333 1.57142857 2.21428571 3.11111111 1.92857143 YAP1 54 kDa 1 1 1 27 31 28 28 26 38 27 31 28 28 26 38 NCL 77 kDa 7 1 1 21 24 22 34 28 33 3 24 22 4.85714286 28 33 HNRNPU 91 kDa 15 12 16 23 14 14 23 21 25 1.53333333 1.16666667 0.875 1.53333333 1.75 1.5625 XPC 116 kDa 15 11 14 22 11 21 27 32 21 1.46666667 1 1.5 1.8 2.90909091 1.5 RAD21 72 kDa 1 1 1 13 21 23 31 42 47 13 21 23 31 42 47 XRCC6 71 kDa 16 1 8 15 3 21 24 27 19 0.9375 3 2.625 1.5 27 2.375 LARS1 134 kDa 11 7 12 19 8 21 19 19 19 1.72727273 1.14285714 1.75 1.72727273 2.71428571 1.58333333 RPS26 13 kDa 7 6 12 13 11 12 16 17 18 1.85714286 1.83333333 1 2.28571429 2.83333333 1.5 EP411 343 kDa 1 1 1 17 21 14 21 18 39 17 21 14 21 18 39 MAP4 121 kDa 6 9 5 15 7 9 26 22 32 2.5 0.77777778 1.8 4.33333333 2.44444444 6.4 WAPL 133 kDa 1 1 1 19 11 17 26 18 28 19 11 17 26 18 28 ANXA1 39 kDa 1 1 1 25 14 11 22 29 25 25 14 11 22 29 25 NUP214 214 kDa 7 7 8 12 15 11 21 12 26 1.71428571 2.14285714 1.375 3 1.71428571 3.25 NRG1 71 kDa 5 8 9 2 2 4 21 21 21 0.4 0.25 0.44444444 4.2 2.625 2.33333333 HCFC1 219 kDa 1 1 1 21 18 16 21 18 24 21 18 16 21 18 24 PRPF3 78 kDa 1 1 1 16 11 13 25 21 24 16 11 13 25 21 24 DDX23 96 kDa 4 9 13 4 4 11 16 19 25 1 0.44444444 0.84615385 4 2.11111111 1.92307692 NPM1 33 kDa 5 5 1 11 9 11 17 11 11 2.2 1.8 11 3.4 2.2 11 DDX46 117 kDa 1 1 1 7 11 15 26 24 26 7 11 15 26 24 26 CHAMP1 89 kDa 1 1 1 14 17 15 15 15 24 14 17 15 15 15 24 NCOR1 271 kDa 1 1 1 8 14 11 15 15 31 8 14 11 15 15 31 JUNB 36 kDa 1 1 1 21 18 15 17 8 14 21 18 15 17 8 14 PCNP 19 kDa 1 1 1 12 9 14 18 16 25 12 9 14 18 16 25 PARP1 113 kDa 6 7 6 9 6 11 16 17 13 1.5 0.85714286 1.83333333 2.66666667 2.42857143 2.16666667 PAK2 58 kDa 8 1 3 11 11 11 18 12 26 1.375 11 3.66666667 2.25 12 8.66666667 SART1 91 kDa 1 1 1 12 8 9 15 13 26 12 8 9 15 13 26 CUX1 164 kDa 1 1 1 6 11 11 15 15 36 6 11 11 15 15 36 SUGP2 121 kDa 1 1 1 6 13 14 13 8 33 6 13 14 13 8 33 NCOR2 274 kDa 1 1 1 14 12 13 16 12 26 14 12 13 16 12 26 TOP2B 183 kDa 7 8 4 13 3 1 19 15 19 1.85714286 0.375 0.25 2.71428571 1.875 4.75 HIRA 112 kDa 2 4 7 11 8 11 13 6 6 5.5 2 1.57142857 6.5 1.5 0.85714286 SMARCD2 59 kDa 7 7 6 11 7 9 12 13 12 1.57142857 1 1.5 1.71428571 1.85714286 2 GPATCH1 113 kDa 1 1 1 6 11 13 12 11 16 6 11 13 12 11 16 MYL12B 21 kDa 1 1 1 3 19 21 3 18 8 3 19 21 3 18 8 SF3B1 146 kDa 3 5 5 8 6 5 11 9 25 2.66666667 1.2 1 3.66666667 1.8 5 SRSF11 54 kDa 8 7 7 11 4 8 14 11 11 1.375 0.57142857 1.14285714 1.75 1.57142857 1.57142857 SLIT2 171 kDa 1 5 9 1 1 2 5 17 25 1 0.2 0.22222222 5 3.4 2.77777778 CCT8 61 kDa 1 1 1 4 5 6 18 21 26 4 5 6 18 21 26 HNRNPA1 39 kDa 1 1 1 8 9 11 18 14 16 8 9 11 18 14 16 ZNF281 97 kDa 1 1 1 13 9 15 11 8 13 13 9 15 11 8 13 PDS5A 151 kDa 1 1 1 7 7 6 13 14 18 7 7 6 13 14 18 RANBP2 358 kDa 1 1 1 1 3 1 6 11 36 1 3 1 6 11 36 XRCC5 83 kDa 4 1 1 4 1 7 8 11 13 1 1 7 2 11 13 SART3 111 kDa 1 1 1 3 1 4 14 9 13 3 1 4 14 9 13 HNRNPK 51 kDa 7 3 1 11 6 5 15 9 12 1.57142857 2 5 2.14285714 3 12 TMOD3 41 kDa 1 1 1 5 14 21 9 16 4 5 14 21 9 16 4TLN1 271 kDa 4 3 1 11 3 8 11 8 13 2.75 1 8 2.75 2.66666667 13 SMARCD3 55 kDa 1 2 1 11 5 8 8 8 17 11 2.5 8 8 4 17 BRD4 152 kDa 1 1 1 7 4 9 14 12 12 7 4 9 14 12 12 EXOSC11 111 kDa 1 1 1 8 4 7 15 12 18 8 4 7 15 12 18 ZFP91 63 kDa 1 1 1 5 4 8 13 11 11 5 4 8 13 11 11 PHF6 41 kDa 1 3 1 2 1 1 9 5 7 2 0.33333333 1 9 1.66666667 7 CSNK2A1 45 kDa 1 1 5 3 4 8 8 12 9 3 4 1.6 8 12 1.8 MRPS31 45 kDa 1 1 1 4 3 9 8 11 14 4 3 9 8 11 14 CBX3 21 kDa 3 1 1 11 1 6 13 5 13 3.66666667 1 6 4.33333333 5 13 GTF2I 112 kDa 1 1 1 7 4 8 9 11 19 7 4 8 9 11 19 TRIM28 89 kDa 1 1 1 4 8 11 8 12 16 4 8 11 8 12 16 SAFB 113 kDa 1 1 1 5 4 3 11 7 15 5 4 3 11 7 15 BRIX1 41 kDa 3 1 1 11 1 8 6 7 3 3.66666667 1 8 2 7 3CTTN 62 kDa 1 1 1 1 7 11 3 4 8 1 7 11 3 4 8 LYAR 44 kDa 2 1 1 5 2 1 7 5 5 2.5 2 1 3.5 5 5 PTCD3 79 kDa 1 1 1 2 5 7 7 13 17 2 5 7 7 13 17 ZNF148 89 kDa 1 1 1 4 7 11 8 9 12 4 7 11 8 9 12 EMSY 141 kDa 1 1 1 1 2 6 4 4 18 1 2 6 4 4 18 KDM3B 192 kDa 1 1 1 4 4 5 8 8 14 4 4 5 8 8 14 THRAP3 119 kDa 1 1 1 3 3 3 9 6 14 3 3 3 9 6 14 RBM17 45 kDa 1 1 1 4 4 4 11 8 13 4 4 4 11 8 13 DDX42 113 kDa 1 1 1 4 8 6 7 12 11 4 8 6 7 12 11 SMCHD1 226 kDa 1 1 1 2 1 1 13 13 18 2 1 1 13 13 18 MRPS9 46 kDa 1 1 1 1 3 6 4 12 19 1 3 6 4 12 19 SUMO1 12 kDa 5 1 1 9 2 2 9 3 3 1.8 2 2 1.8 3 3 SYMPK 141 kDa 1 1 1 4 4 6 7 9 5 4 4 6 7 9 5 TPR 267 kDa 1 1 1 1 1 1 4 4 19 1 1 1 4 4 19 DIDO1 244 kDa 1 1 1 5 1 5 11 3 13 5 1 5 11 3 13 MIDEAS 115 kDa 1 1 1 6 3 6 8 6 11 6 3 6 8 6 11 CSNK2A2 41 kDa 1 1 3 1 1 7 6 8 5 1 1 2.33333333 6 8 1.66666667 NOP11 8 kDa 3 3 1 5 1 6 8 7 4 1.66666667 0.33333333 6 2.66666667 2.33333333 4 PRR12 211 kDa 1 1 1 1 5 5 11 3 11 1 5 5 11 3 11 RAI14 111 kDa 1 1 2 3 3 11 4 4 5 3 3 5.5 4 4 2.5 ATP1A1 113 kDa 3 1 3 2 1 1 7 6 11 0.66666667 1 0.33333333 2.33333333 6 3.66666667 TXN 12 kDa 1 1 1 3 1 5 6 7 8 3 1 5 6 7 8 ARHGAP11 89 kDa 1 1 1 1 1 2 4 9 3 1 1 2 4 9 3 CCAR2 113 kDa 1 1 1 3 3 5 7 7 11 3 3 5 7 7 11 CAND1 136 kDa 1 1 1 5 4 3 4 4 5 5 4 3 4 4 5 FLT1 151 kDa 1 1 3 1 1 1 1935 7 11 1 1 0.33333333 9 7 3.66666667 ARPC1B 41 kDa 1 1 1 4 6 9 4 8 4 4 6 9 4 8 4 RAB5C 23 kDa 1 1 1 1 1 1 5 5 12 1 1 1 5 5 12 KHSRP 73 kDa 1 1 1 1 1 1 6 7 11 1 1 1 6 7 11 SBNO1 154 kDa 1 1 1 1 4 5 3 4 6 1 4 5 3 4 6 FIP1L1 67 kDa 1 1 1 4 1 2 5 7 11 4 1 2 5 7 11 ILKAP 43 kDa 1 1 1 3 1 1 6 8 6 3 1 1 6 8 6 HMGXB4 66 kDa 1 1 4 2 1 4 6 6 6 2 1 1 6 6 1.5 ACTR3 47 kDa 1 1 1 4 6 8 2 8 2 4 6 8 2 8 2 SLC25A3 41 kDa 3 1 2 2 4 2 6 4 5 0.66666667 4 1 2 4 2.5 MRE11 81 kDa 1 1 1 1 1 1 6 7 13 1 1 1 6 7 13 PRPF38B 64 kDa 1 1 1 3 1 2 6 6 4 3 1 2 6 6 4 NDUFS1 79 kDa 4 1 1 1 4 1 6 3 5 0.25 4 1 1.5 3 5 TMPO 75 kDa 1 1 1 3 2 2 4 8 8 3 2 2 4 8 8 ACIN1 152 kDa 1 1 1 1 1 3 6 5 9 1 1 3 6 5 9 TAGLN2 22 kDa 1 1 1 4 1 1 8 8 6 4 1 1 8 8 6 BUD13 71 kDa 1 1 1 3 4 3 1 4 5 3 4 3 1 4 5 ZNF638 221 kDa 1 1 1 1 1 3 5 4 6 1 1 3 5 4 6 CSTF2 61 kDa 1 1 1 1 3 1 5 5 9 1 3 1 5 5 9 DPF2 44 kDa 1 1 1 5 4 4 1 1 1 5 4 4 1 1 1 MATR3 95 kDa 3 1 1 1 1 1 9 3 3 0.33333333 1 1 3 3 3 NFIC 56 kDa 1 1 1 4 2 5 3 1 9 4 2 5 3 1 9 SMAP 21 kDa 1 1 1 1 1 1 5 5 9 1 1 1 5 5 9 CFDP1 34 kDa 1 1 1 1 3 1 5 2 7 1 3 1 5 2 7 SNW1 61 kDa 1 1 1 1 1 1 5 5 13 1 1 1 5 5 13 HIP1R 119 kDa 1 1 1 1 1 1 7 7 4 1 1 1 7 7 4 RAB5B 24 kDa 1 1 1 3 1 1 3 5 3 3 1 1 3 5 3 KAT5 59 kDa 1 1 1 3 1 3 4 3 6 3 1 3 4 3 6 EHD3 61 kDa 1 1 3 1 1 1 6 4 7 1 1 0.33333333 6 4 2.33333333 MBOAT7 53 kDa 1 1 1 1 1 1 3 4 4 1 1 1 3 4 4 PSME3IP1 29 kDa 1 1 1 1 1 3 5 5 5 1 1 3 5 5 5 RPRD2 156 kDa 1 1 1 2 1 1 3 2 6 2 1 1 3 2 6 ICE1 248 kDa 1 1 1 1 1 4 3 3 4 1 1 4 3 3 4 RAB1B 22 kDa 1 1 1 1 1 1 2 4 3 1 1 1 2 4 3 TAGLN 23 kDa 1 1 1 5 1 1 5 3 3 5 1 1 5 3 3 PHF3 229 kDa 1 1 1 1 1 1 3 4 5 1 1 1 3 4 5 CUL4B 114 kDa 1 1 1 1 1 1 3 4 4 1 1 1 3 4 4 HSPE1 11 kDa 1 1 1 1 1 1 4 4 11 1 1 1 4 4 11 FLI1 51 kDa 1 1 1 1 1 1 7 4 6 1 1 1 7 4 6 REV3L 353 kDa 1 1 1 2 2 2 1 1 1 2 2 2 1 1 1 PSMC5 46 kDa 1 1 1 1 1 1 3 4 7 1 1 1 3 4 7 ZMYM4 173 kDa 1 1 1 1 1 1 2 3 3 1 1 1 2 3 3 TSPYL2 79 kDa 1 1 1 1 1 1 2 5 5 1 1 1 2 5 5 ESS2 53 kDa 1 1 1 1 1 1 2 4 6 1 1 1 2 4 6 PSMC3 49 kDa 1 1 1 1 1 1 2 3 4 1 1 1 2 3 4 Molecular Neo_Ctrl pLVX Vector pLVX_EF Fold change V/Ctrl Fold change EF/Ctrl protein ID Weight 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 MYH9 227 kDa 34 14 41 442 2219 2797 652 2555 711 13 158.5 68.2195122 19.1764706 182.5 17.3414634 AHNAK 629 kDa 35 9 9 287 242 286 348 415 627 8.2 26.8888889 31.7777778 9.94285714 46.1111111 69.6666667 NUMA1 238 kDa 116 141 189 217 221 214 272 261 323 1.87068966 1.56737589 1.13227513 2.34482759 1.85106383 1.70899471 ACTB 42 kDa 33 38 43 112 268 431 124 351 92 3.39393939 7.05263158 10.0232558 3.75757576 9.23684211 2.13953488 IFI16 88 kDa 18 13 14 126 71 91 127 115 114 7 5.46153846 6.5 7.05555556 8.84615385 8.14285714 HNRNPM 78 kDa 91 24 39 124 46 61 138 73 64 1.36263736 1.91666667 1.56410256 1.51648352 3.04166667 1.64102564 SMC1A 143 kDa 3 6 3 81 84 113 115 118 134 27 14 37.6666667 38.3333333 19.6666667 44.6666667 MYL6 17 kDa 9 6 6 31 119 189 36 127 31 3.44444444 19.8333333 31.5 4 21.1666667 5.16666667 ZNF619 151 kDa 42 21 32 45 21 23 77 61 76 1.07142857 1 0.71875 1.83333333 2.9047619 2.375 SMC3 142 kDa 1 1 1 41 28 38 84 66 82 41 28 38 84 66 82 PDS5B 165 kDa 11 23 32 38 25 37 41 45 52 3.45454545 1.08695652 1.15625 3.72727273 1.95652174 1.625 NIPBL 316 kDa 1 1 1 42 34 48 44 44 51 42 34 48 44 44 51 GSN 86 kDa 1 1 1 27 51 71 36 59 27 27 51 71 36 59 27 MTA1 81 kDa 23 14 24 31 16 21 39 31 36 1.34782609 1.14285714 0.875 1.69565217 2.21428571 1.5 RBM25 111 kDa 13 8 17 26 11 22 36 27 33 2 1.375 1.29411765 2.76923077 3.375 1.94117647 ADNP 124 kDa 14 9 14 21 21 22 31 28 27 1.5 2.33333333 1.57142857 2.21428571 3.11111111 1.92857143 YAP1 54 kDa 1 1 1 27 31 28 28 26 38 27 31 28 28 26 38 NCL 77 kDa 7 1 1 21 24 22 34 28 33 3 24 22 4.85714286 28 33 HNRNPU 91 kDa 15 12 16 23 14 14 23 21 25 1.53333333 1.16666667 0.875 1.53333333 1.75 1.5625 XPC 116 kDa 15 11 14 22 11 21 27 32 21 1.46666667 1 1.5 1.8 2.90909091 1.5 RAD21 72 kDa 1 1 1 13 21 23 31 42 47 13 21 23 31 42 47 XRCC6 71 kDa 16 1 8 15 3 21 24 27 19 0.9375 3 2.625 1.5 27 2.375 LARS1 134 kDa 11 7 12 19 8 21 19 19 19 1.72727273 1.14285714 1.75 1.72727273 2.71428571 1.58333333 RPS26 13 kDa 7 6 12 13 11 12 16 17 18 1.85714286 1.83333333 1 2.28571429 2.83333333 1.5 EP411 343 kDa 1 1 1 17 21 14 21 18 39 17 21 14 21 18 39 MAP4 121 kDa 6 9 5 15 7 9 26 22 32 2.5 0.77777778 1.8 4.33333333 2.44444444 6.4 WAPL 133 kDa 1 1 1 19 11 17 26 18 28 19 11 17 26 18 28 ANXA1 39 kDa 1 1 1 25 14 11 22 29 25 25 14 11 22 29 25 NUP214 214 kDa 7 7 8 12 15 11 21 12 26 1.71428571 2.14285714 1.375 3 1.71428571 3.25 NRG1 71 kDa 5 8 9 2 2 4 21 21 21 0.4 0.25 0.44444444 4.2 2.625 2.33333333 HCFC1 219 kDa 1 1 1 21 18 16 21 18 24 21 18 16 21 18 24 PRPF3 78 kDa 1 1 1 16 11 13 25 21 24 16 11 13 25 21 24 DDX23 96 kDa 4 9 13 4 4 11 16 19 25 1 0.44444444 0.84615385 4 2.11111111 1.92307692 NPM1 33 kDa 5 5 1 11 9 11 17 11 11 2.2 1.8 11 3.4 2.2 11 DDX46 117 kDa 1 1 1 7 11 15 26 24 26 7 11 15 26 24 26 CHAMP1 89 kDa 1 1 1 14 17 15 15 15 24 14 17 15 15 15 24 NCOR1 271 kDa 1 1 1 8 14 11 15 15 31 8 14 11 15 15 31 JUNB 36 kDa 1 1 1 21 18 15 17 8 14 21 18 15 17 8 14 PCNP 19 kDa 1 1 1 12 9 14 18 16 25 12 9 14 18 16 25 PARP1 113 kDa 6 7 6 9 6 11 16 17 13 1.5 0.85714286 1.83333333 2.66666667 2.42857143 2.16666667 PAK2 58 kDa 8 1 3 11 11 11 18 12 26 1.375 11 3.66666667 2.25 12 8.66666667 SART1 91 kDa 1 1 1 12 8 9 15 13 26 12 8 9 15 13 26 CUX1 164 kDa 1 1 1 6 11 11 15 15 36 6 11 11 15 15 36 SUGP2 121 kDa 1 1 1 6 13 14 13 8 33 6 13 14 13 8 33 NCOR2 274 kDa 1 1 1 14 12 13 16 12 26 14 12 13 16 12 26 TOP2B 183 kDa 7 8 4 13 3 1 19 15 19 1.85714286 0.375 0.25 2.71428571 1.875 4.75 HIRA 112 kDa 2 4 7 11 8 11 13 6 6 5.5 2 1.57142857 6.5 1.5 0.85714286 SMARCD2 59 kDa 7 7 6 11 7 9 12 13 12 1.57142857 1 1.5 1.71428571 1.85714286 2 GPATCH1 113 kDa 1 1 1 6 11 13 12 11 16 6 11 13 12 11 16 MYL12B 21 kDa 1 1 1 3 19 21 3 18 8 3 19 21 3 18 8 SF3B1 146 kDa 3 5 5 8 6 5 11 9 25 2.66666667 1.2 1 3.66666667 1.8 5 SRSF11 54 kDa 8 7 7 11 4 8 14 11 11 1.375 0.57142857 1.14285714 1.75 1.57142857 1.57142857 SLIT2 171 kDa 1 5 9 1 1 2 5 17 25 1 0.2 0.22222222 5 3.4 2.77777778 CCT8 61 kDa 1 1 1 4 5 6 18 21 26 4 5 6 18 21 26 HNRNPA1 39 kDa 1 1 1 8 9 11 18 14 16 8 9 11 18 14 16 ZNF281 97 kDa 1 1 1 13 9 15 11 8 13 13 9 15 11 8 13 PDS5A 151 kDa 1 1 1 7 7 6 13 14 18 7 7 6 13 14 18 RANBP2 358 kDa 1 1 1 1 3 1 6 11 36 1 3 1 6 11 36 XRCC5 83 kDa 4 1 1 4 1 7 8 11 13 1 1 7 2 11 13 SART3 111 kDa 1 1 1 3 1 4 14 9 13 3 1 4 14 9 13 HNRNPK 51 kDa 7 3 1 11 6 5 15 9 12 1.57142857 2 5 2.14285714 3 12 TMOD3 41 kDa 1 1 1 5 14 21 9 16 4 5 14 21 9 16 4 TLN1 271 kDa 4 3 1 11 3 8 11 8 13 2.75 1 8 2.75 2.66666667 13 SMARCD3 55 kDa 1 2 1 11 5 8 8 8 17 11 2.5 8 8 4 17 BRD4 152 kDa 1 1 1 7 4 9 14 12 12 7 4 9 14 12 12 EXOSC11 111 kDa 1 1 1 8 4 7 15 12 18 8 4 7 15 12 18 ZFP91 63 kDa 1 1 1 5 4 8 13 11 11 5 4 8 13 11 11 PHF6 41 kDa 1 3 1 2 1 1 9 5 7 2 0.33333333 1 9 1.66666667 7 CSNK2A1 45 kDa 1 1 5 3 4 8 8 12 9 3 4 1.6 8 12 1.8 MRPS31 45 kDa 1 1 1 4 3 9 8 11 14 4 3 9 8 11 14 CBX3 21 kDa 3 1 1 11 1 6 13 5 13 3.66666667 1 6 4.33333333 5 13 GTF2I 112 kDa 1 1 1 7 4 8 9 11 19 7 4 8 9 11 19 TRIM28 89 kDa 1 1 1 4 8 11 8 12 16 4 8 11 8 12 16 SAFB 113 kDa 1 1 1 5 4 3 11 7 15 5 4 3 11 7 15 BRIX1 41 kDa 3 1 1 11 1 8 6 7 3 3.66666667 1 8 2 7 3 CTTN 62 kDa 1 1 1 1 7 11 3 4 8 1 7 11 3 4 8 LYAR 44 kDa 2 1 1 5 2 1 7 5 5 2.5 2 1 3.5 5 5 PTCD3 79 kDa 1 1 1 2 5 7 7 13 17 2 5 7 7 13 17 ZNF148 89 kDa 1 1 1 4 7 11 8 9 12 4 7 11 8 9 12 EMSY 141 kDa 1 1 1 1 2 6 4 4 18 1 2 6 4 4 18 KDM3B 192 kDa 1 1 1 4 4 5 8 8 14 4 4 5 8 8 14 THRAP3 119 kDa 1 1 1 3 3 3 9 6 14 3 3 3 9 6 14 RBM17 45 kDa 1 1 1 4 4 4 11 8 13 4 4 4 11 8 13 DDX42 113 kDa 1 1 1 4 8 6 7 12 11 4 8 6 7 12 11 SMCHD1 226 kDa 1 1 1 2 1 1 13 13 18 2 1 1 13 13 18 MRPS9 46 kDa 1 1 1 1 3 6 4 12 19 1 3 6 4 12 19 SUMO1 12 kDa 5 1 1 9 2 2 9 3 3 1.8 2 2 1.8 3 3 SYMP K 141 kDa 1 1 1 4 4 6 7 9 5 4 4 6 7 9 5 TPR 267 kDa 1 1 1 1 1 1 4 4 19 1 1 1 4 4 19 DIDO1 244 kDa 1 1 1 5 1 5 11 3 13 5 1 5 11 3 13 MIDEAS 115 kDa 1 1 1 6 3 6 8 6 11 6 3 6 8 6 11 CSNK2 A2 41 kDa 1 1 3 1 1 7 6 8 5 1 1 2.33333333 6 8 1.66666667 NOP11 8 kDa 3 3 1 5 1 6 8 7 4 1.66666667 0.33333333 6 2.66666667 2.33333333 4 PRR12 211 kDa 1 1 1 1 5 5 11 3 11 1 5 5 11 3 11 RAI14 111 kDa 1 1 2 3 3 11 4 4 5 3 3 5.5 4 4 2.5 ATP1A1 113 kDa 3 1 3 2 1 1 7 6 11 0.66666667 1 0.33333333 2.33333333 6 3.66666667 TXN 12 kDa 1 1 1 3 1 5 6 7 8 3 1 5 6 7 8 ARHGAP11 89 kDa 1 1 1 1 1 2 4 9 3 1 1 2 4 9 3 CCAR2 113 kDa 1 1 1 3 3 5 7 7 11 3 3 5 7 7 11 CAND1 136 kDa 1 1 1 5 4 3 4 4 5 5 4 3 4 4 5 FLT1 151 kDa 1 1 3 1 1 1 9 7 11 1 1 0.33333333 9 7 3.66666667 ARPC1B 41 kDa 1 1 1 4 6 9 4 8 4 4 6 9 4 8 4 RAB5C 23 kDa 1 1 1 1 1 1 5 5 12 1 1 1 5 5 12 KHSRP 73 kDa 1 1 1 1 1 1 6 7 11 1 1 1 6 7 11 SBNO 1 154 kDa 1 1 1 1 4 5 3 4 6 1 4 5 3 4 6FIP1L1 67 kDa 1 1 1 4 1 2 5 7 11 4 1 2 5 7 11 ILKAP 43 kDa 1 1 1 3 1 1 6 8 6 3 1 1 6 8 6 HMGXB4 66 kDa 1 1 4 2 1 4 6 6 6 2 1 1 6 6 1.5 ACTR3 47 kDa 1 1 1 4 6 8 2 8 2 4 6 8 2 8 2 SLC25 A3 41 kDa 3 1 2 2 4 2 6 4 5 0.66666667 4 1 2 4 2.5 MRE11 81 kDa 1 1 1 1 1 1 6 7 13 1 1 1 6 7 13 PRPF38B 64 kDa 1 1 1 3 1 2 6 6 4 3 1 2 6 6 4 NDUFS1 79 kDa 4 1 1 1 4 1 6 3 5 0.25 4 1 1.5 3 5 TMPO 75 kDa 1 1 1 3 2 2 4 8 8 3 2 2 4 8 8 ACIN1 152 kDa 1 1 1 1 1 3 6 5 9 1 1 3 6 5 9 TAGLN2 22 kDa 1 1 1 4 1 1 8 8 6 4 1 1 8 8 6 BUD13 71 kDa 1 1 1 3 4 3 1 4 5 3 4 3 1 4 5 ZNF63 8 221 kDa 1 1 1 1 1 3 5 4 6 1 1 3 5 4 6CSTF2 61 kDa 1 1 1 1 3 1 5 5 9 1 3 1 5 5 9 DPF2 44 kDa 1 1 1 5 4 4 1 1 1 5 4 4 1 1 1 MATR3 95 kDa 3 1 1 1 1 1 9 3 3 0.33333333 1 1 3 3 3 NFIC 56 kDa 1 1 1 4 2 5 3 1 9 4 2 5 3 1 9 SMAP 21 kDa 1 1 1 1 1 1 5 5 9 1 1 1 5 5 9 CFDP1 34 kDa 1 1 1 1 3 1 5 2 7 1 3 1 5 2 7 SNW1 61 kDa 1 1 1 1 1 1 5 5 13 1 1 1 5 5 13 HIP1R 119 kDa 1 1 1 1 1 1 7 7 4 1 1 1 7 7 4 RAB5B 24 kDa 1 1 1 3 1 1 3 5 3 3 1 1 3 5 3 KAT5 59 kDa 1 1 1 3 1 3 4 3 6 3 1 3 4 3 6 EHD3 61 kDa 1 1 3 1 1 1 6 4 7 1 1 0.33333333 6 4 2.33333333 MBOAT7 53 kDa 1 1 1 1 1 1 3 4 4 1 1 1 3 4 4 PSME3IP1 29 kDa 1 1 1 1 1 3 5 5 5 1 1 3 5 5 5 RPRD2 156 kDa 1 1 1 2 1 1 3 2 6 2 1 1 3 2 6 ICE1 248 kDa 1 1 1 1 1 4 3 3 4 1 1 4 3 3 4 RAB1B 22 kDa 1 1 1 1 1 1 2 4 3 1 1 1 2 4 3 TAGLN 23 kDa 1 1 1 5 1 1 5 3 3 5 1 1 5 3 3 PHF3 229 kDa 1 1 1 1 1 1 3 4 5 1 1 1 3 4 5 CUL4B 114 kDa 1 1 1 1 1 1 3 4 4 1 1 1 3 4 4 HSPE1 11 kDa 1 1 1 1 1 1 4 4 11 1 1 1 4 4 11 FLI1 51 kDa 1 1 1 1 1 1 7 4 6 1 1 1 7 4 6 REV3L 353 kDa 1 1 1 2 2 2 1 1 1 2 2 2 1 1 1 PSMC5 46 kDa 1 1 1 1 1 1 3 4 7 1 1 1 3 4 7 ZMYM4 173 kDa 1 1 1 1 1 1 2 3 3 1 1 1 2 3 3 TSPYL2 79 kDa 1 1 1 1 1 1 2 5 5 1 1 1 2 5 5 ESS2 53 kDa 1 1 1 1 1 1 2 4 6 1 1 1 2 4 6PSMC3 49 kDa 1 1 1 1 1 1 2 3 4 1 1 1 2 3 4 Note: Cells highlighted in red marks hits that are greater or equal to 1.5 fold enrichment compared to no turbo-ID control. 136 SUPPLEMENTAL TABLE S2 137 Supplemental table S2: List of antibodies used in this study Antibody Purchase # Vendor Dilution Experiment Type Anti-FLI1 Antibody - ChIP Grade ab15289 abcam 1/500 WB Primary Anti-RAD21 Antibody 05-908 Millipore Sigma 1/500 WB Primary 1 ul per reaction for Anti-RAD21 Antibody-ChIP Grade ab217678 abcam ChIP and CUT&RUN Primary CUT&RUN 1/1000 for WB, 1 ul per Anti-PCNA Antiboty (D3H8P) 13110 Cell Signaling Technology CUT&RUN and WB Primary reaction for CUT&RUN 1 ul per reaction for Anti-RNAPII Antibody (CTD4H8) sc-47701 Santa Cruz Biotechnology CUT&RUN Primary CUT&RUN Anti-Tri-Methyl-Histone H3 Antibody 1 ul per reaction for Cell Signaling Technology Secondary (Lys4) (C42D8) 9751 CUT&RUN 5 ul per reaction for Anti-Rabbit IgG (DA1E) 66362 Cell Signaling Technology CUT&RUN Primary CUT&RUN Anti-SMC3 Antibody-ChIP Grade ab9263 abcam 1/1000 WB Primary Anti-SMC1 Antibody ab21583 abcam 1/1000 WB Primary c-Myc Antibody ab32072 abcam 1/1000 WB Primary alpha Tubulin Monoclonal Antibody MA1-80017 Thermo Fisher Scientific 1/1000 WB Primary (YL1/2) GAPDH Antibody (G-9) sc-365062 Santa Cruz Biotechnology 1/1000 WB Primary Donkey anti-Mouse IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, A-21202 Thermo Fisher Scientific 1/500 IF Secondary Alexa Fluor 488 Goat anti-Rabbit IgG (H+L) Cross- Adsorbed Secondary Antibody, Alexa A-11010 Thermo Fisher Scientific 1/500 IF Secondary Fluor 546 Anti-mouse IgG (H+L) (DyLight™ 680 5470 Cell Signaling Technology 1/10000 WB Secondary Conjugate) Anti-rabbit IgG (H+L) (DyLight™ 800 5151 Cell Signaling Technology 1/10000 WB Secondary 4X PEG Conjugate) REFERENCES Aguilera, A., & García-Muse, T. (2012). R Loops: From Transcription Byproducts to Threats to Genome Stability. Molecular Cell, 46(2), 115–124. https://doi.org/10.1016/j.molcel.2012.04.009 Anderson, N. D., Borja, R. de, Young, M. D., Fuligni, F., Rosic, A., Roberts, N. D., Hajjar, S., Layeghifard, M., Novokmet, A., Kowalski, P. E., Anaka, M., Davidson, S., Zarrei, M., Said, B. I., Schreiner, L. C., Marchand, R., Sitter, J., Gokgoz, N., Brunga, L., … Shlien, A. (2018). Rearrangement bursts generate canonical gene fusions in bone and soft tissue tumors. Science, 361(6405), eaam8419. https://doi.org/10.1126/science.aam8419 Benedict, B., Schie, J. J. M. van, Oostra, A. B., Balk, J. A., Wolthuis, R. M. F., Riele, H. te, & Lange, J. de. (2020). WAPL-Dependent Repair of Damaged DNA Replication Forks Underlies Oncogene-Induced Loss of Sister Chromatid Cohesion. Developmental Cell, 52(6), 683-698.e7. https://doi.org/10.1016/j.devcel.2020.01.024 Bertoli, C., Skotheim, J. M., & Bruin, R. A. M. de. (2013). Control of cell cycle transcription during G1 and S phases. Nature Reviews Molecular Cell Biology, 14(8), 518–528. https://doi.org/10.1038/nrm3629 Boehm, E. M., Gildenberg, M. S., & Washington, M. T. (2016). Chapter Seven The Many Roles of PCNA in Eukaryotic DNA Replication. The Enzymes, 39, 231–254. https://doi.org/10.1016/bs.enz.2016.03.003 Bryant, H. E., Petermann, E., Schultz, N., Jemth, A., Loseva, O., Issaeva, N., Johansson, F., Fernandez, S., McGlynn, P., & Helleday, T. (2009). PARP is activated at stalled forks to mediate Mre11‐dependent replication restart and recombination. The EMBO Journal, 28(17), 2601–2615. https://doi.org/10.1038/emboj.2009.206 Carvajal-Maldonado, D., Byrum, A. K., Jackson, J., Wessel, S., Lemaçon, D., Guitton- Sert, L., Quinet, A., Tirman, S., Graziano, S., Masson, J.-Y., Cortez, D., Gonzalo, S., Mosammaparast, N., & Vindigni, A. (2018). Perturbing cohesin dynamics drives MRE11 nuclease-dependent replication fork slowing. Nucleic Acids Research, 47(3), gky519-. https://doi.org/10.1093/nar/gky519 Chaudhuri, A. R., & Nussenzweig, A. (2017). The multifaceted roles of PARP1 in DNA repair and chromatin remodelling. Nature Reviews Molecular Cell Biology, 18(10), 610–621. https://doi.org/10.1038/nrm.2017.53 Cortés‐Ledesma, F., & Aguilera, A. (2006). Double‐strand breaks arising by replication through a nick are repaired by cohesin‐dependent sister‐chromatid exchange. EMBO Reports, 7(9), 919–926. https://doi.org/10.1038/sj.embor.7400774 138 Dutta, D., Shatalin, K., Epshtein, V., Gottesman, M. E., & Nudler, E. (2011). Linking RNA Polymerase Backtracking to Genome Instability in E. coli. Cell, 146(4), 533– 543. https://doi.org/10.1016/j.cell.2011.07.034 Frattini, C., Villa-Hernández, S., Pellicanò, G., Jossen, R., Katou, Y., Shirahige, K., & Bermejo, R. (2017). Cohesin Ubiquitylation and Mobilization Facilitate Stalled Replication Fork Dynamics. Molecular Cell, 68(4), 758-772.e4. https://doi.org/10.1016/j.molcel.2017.10.012 Gaillard, H., García-Muse, T., & Aguilera, A. (2015). Replication stress and cancer. Nature Reviews Cancer, 15(5), 276–289. https://doi.org/10.1038/nrc3916 Gelot, C., Guirouilh-Barbat, J., & Lopez, B. S. (2016). The cohesin complex prevents the end-joining of distant DNA double-strand ends in S phase: consequences on genome stability maintenance. Nucleus, 7(4), 00–00. https://doi.org/10.1080/19491034.2016.1194159 Gorthi, A., Romero, J. C., Loranc, E., Cao, L., Lawrence, L. A., Goodale, E., Iniguez, A. B., Bernard, X., Masamsetti, V. P., Roston, S., Lawlor, E. R., Toretsky, J. A., Stegmaier, K., Lessnick, S. L., Chen, Y., & Bishop, A. J. R. (2018). EWS–FLI1 increases transcription to cause R-loops and block BRCA1 repair in Ewing sarcoma. Nature, 555(7696), 387–391. https://doi.org/10.1038/nature25748 Grünewald, T. G. P., Cidre-Aranaz, F., Surdez, D., Tomazou, E. M., Álava, E. de, Kovar, H., Sorensen, P. H., Delattre, O., & Dirksen, U. (2018). Ewing sarcoma. Nature Reviews Disease Primers, 4(1), 5. https://doi.org/10.1038/s41572-018-0003-x Hamperl, S., Bocek, M. J., Saldivar, J. C., Swigut, T., & Cimprich, K. A. (2017). Transcription-Replication Conflict Orientation Modulates R-Loop Levels and Activates Distinct DNA Damage Responses. Cell, 170(4), 774-786.e19. https://doi.org/10.1016/j.cell.2017.07.043 Kim, J.-S., Krasieva, T. B., LaMorte, V., Taylor, A. M. R., & Yokomori, K. (2002). Specific Recruitment of Human Cohesin to Laser-induced DNA Damage. Journal of Biological Chemistry, 277(47), 45149–45153. https://doi.org/10.1074/jbc.m209123200 Koppenhafer, S. L., Goss, K. L., Terry, W. W., & Gordon, D. J. (2020). Inhibition of the ATR–CHK1 Pathway in Ewing Sarcoma Cells Causes DNA Damage and Apoptosis via the CDK2-Mediated Degradation of RRM2. Molecular Cancer Research, 18(1), 91–104. https://doi.org/10.1158/1541-7786.mcr-19-0585 Kotsantis, P., Petermann, E., & Boulton, S. J. (2018). Mechanisms of Oncogene- Induced Replication Stress: Jigsaw Falling into Place. Cancer Discovery, 8(5), 537– 555. https://doi.org/10.1158/2159-8290.cd-17-1461 139 Lang, K. S., Hall, A. N., Merrikh, C. N., Ragheb, M., Tabakh, H., Pollock, A. J., Woodward, J. J., Dreifus, J. E., & Merrikh, H. (2017). Replication-Transcription Conflicts Generate R-Loops that Orchestrate Bacterial Stress Survival and Pathogenesis. Cell, 170(4), 787-799.e18. https://doi.org/10.1016/j.cell.2017.07.044 Lessnick, S. L., Dacwag, C. S., & Golub, T. R. (2002). The Ewing’s sarcoma oncoprotein EWS/FLI induces a p53-dependent growth arrest in primary human fibroblasts. Cancer Cell, 1(4), 393–401. https://doi.org/10.1016/s1535- 6108(02)00056-9 Liao, H., Ji, F., Helleday, T., & Ying, S. (2018). Mechanisms for stalled replication fork stabilization: new targets for synthetic lethality strategies in cancer treatments. EMBO Reports, 19(9). https://doi.org/10.15252/embr.201846263 Meers, M. P., Tenenbaum, D., & Henikoff, S. (2019). Peak calling by Sparse Enrichment Analysis for CUT&RUN chromatin profiling. Epigenetics & Chromatin, 12(1), 42. https://doi.org/10.1186/s13072-019-0287-4 Merrikh, H., Machón, C., Grainger, W. H., Grossman, A. D., & Soultanas, P. (2011). Co- directional replication–transcription conflicts lead to replication restart. Nature, 470(7335), 554–557. https://doi.org/10.1038/nature09758 Morales, C., Ruiz-Torres, M., Rodríguez-Acebes, S., Lafarga, V., Rodríguez-Corsino, M., Megías, D., Cisneros, D. A., Peters, J.-M., Méndez, J., & Losada, A. (2020). PDS5 proteins are required for proper cohesin dynamics and participate in replication fork protection. Journal of Biological Chemistry, 295(1), 146–157. https://doi.org/10.1074/jbc.ra119.011099 Nieto-Soler, M., Morgado-Palacin, I., Lafarga, V., Lecona, E., Murga, M., Callen, E., Azorin, D., Alonso, J., Lopez-Contreras, A. J., Nussenzweig, A., & Fernandez- Capetillo, O. (2016). Efficacy of ATR inhibitors as single agents in Ewing sarcoma. Oncotarget, 7(37), 58759–58767. https://doi.org/10.18632/oncotarget.11643 Panday, A., Willis, N. A., Elango, R., Menghi, F., Duffey, E. E., Liu, E. T., & Scully, R. (2021). FANCM regulates repair pathway choice at stalled replication forks. Molecular Cell, 81(11), 2428-2444.e6. https://doi.org/10.1016/j.molcel.2021.03.044 Panigrahi, A. K., & Pati, D. (2012). Higher-order orchestration of hematopoiesis: Is cohesin a new player? Experimental Hematology, 40(12), 967–973. https://doi.org/10.1016/j.exphem.2012.09.010 Potts, P. R., Porteus, M. H., & Yu, H. (2006). Human SMC5/6 complex promotes sister chromatid homologous recombination by recruiting the SMC1/3 cohesin complex to double‐strand breaks. The EMBO Journal, 25(14), 3377–3388. https://doi.org/10.1038/sj.emboj.7601218 140 Scully, R., Panday, A., Elango, R., & Willis, N. A. (2019). DNA double-strand break repair-pathway choice in somatic mammalian cells. Nature Reviews Molecular Cell Biology, 20(11), 698–714. https://doi.org/10.1038/s41580-019-0152-0 Sjögren, C., & Nasmyth, K. (2001). Sister chromatid cohesion is required for postreplicative double-strand break repair in Saccharomyces cerevisiae. Current Biology, 11(12), 991–995. https://doi.org/10.1016/s0960-9822(01)00271-8 Skene, P. J., & Henikoff, S. (2017). An efficient targeted nuclease strategy for high- resolution mapping of DNA binding sites. ELife, 6, e21856. https://doi.org/10.7554/elife.21856 Söderberg, O., Gullberg, M., Jarvius, M., Ridderstråle, K., Leuchowius, K.-J., Jarvius, J., Wester, K., Hydbring, P., Bahram, F., Larsson, L.-G., & Landegren, U. (2006). Direct observation of individual endogenous protein complexes in situ by proximity ligation. Nature Methods, 3(12), 995–1000. https://doi.org/10.1038/nmeth947 Ström, L., Lindroos, H. B., Shirahige, K., & Sjögren, C. (2004). Postreplicative Recruitment of Cohesin to Double-Strand Breaks Is Required for DNA Repair. Molecular Cell, 16(6), 1003–1015. https://doi.org/10.1016/j.molcel.2004.11.026 Su, X. A., Ma, D., Parsons, J. V., Replogle, J. M., Amatruda, J. F., Whittaker, C. A., Stegmaier, K., & Amon, A. (2021). RAD21 is a driver of chromosome 8 gain in Ewing sarcoma to mitigate replication stress. Genes & Development, 35(7–8), 556–572. https://doi.org/10.1101/gad.345454.120 Tittel-Elmer, M., Lengronne, A., Davidson, M. B., Bacal, J., François, P., Hohl, M., Petrini, J. H. J., Pasero, P., & Cobb, J. A. (2012). Cohesin Association to Replication Sites Depends on Rad50 and Promotes Fork Restart. Molecular Cell, 48(1), 98–108. https://doi.org/10.1016/j.molcel.2012.07.004 Ünal, E., Arbel-Eden, A., Sattler, U., Shroff, R., Lichten, M., Haber, J. E., & Koshland, D. (2004). DNA Damage Response Pathway Uses Histone Modification to Assemble a Double-Strand Break-Specific Cohesin Domain. Molecular Cell, 16(6), 991–1002. https://doi.org/10.1016/j.molcel.2004.11.027 Ünal, E., Heidinger-Pauli, J. M., Kim, W., Guacci, V., Onn, I., Gygi, S. P., & Koshland, D. E. (2008). A Molecular Determinant for the Establishment of Sister Chromatid Cohesion. Science, 321(5888), 566–569. https://doi.org/10.1126/science.1157880 Willis, N. A., Chandramouly, G., Huang, B., Kwok, A., Follonier, C., Deng, C., & Scully, R. (2014). BRCA1 controls homologous recombination at Tus/Ter-stalled mammalian replication forks. Nature, 510(7506), 556–559. https://doi.org/10.1038/nature13295 Willis, N. A., Panday, A., Duffey, E. E., & Scully, R. (2018). Rad51 recruitment and exclusion of non-homologous end joining during homologous recombination at a 141 Tus/Ter mammalian replication fork barrier. PLOS Genetics, 14(7), e1007486. https://doi.org/10.1371/journal.pgen.1007486 Wu, N., Kong, X., Ji, Z., Zeng, W., Potts, P. R., Yokomori, K., & Yu, H. (2012). Scc1 sumoylation by Mms21 promotes sister chromatid recombination through counteracting Wapl. Genes & Development, 26(13), 1473–1485. https://doi.org/10.1101/gad.193615.112 Zeman, M. K., & Cimprich, K. A. (2014). Causes and consequences of replication stress. Nature Cell Biology, 16(1), 2–9. https://doi.org/10.1038/ncb2897 142 143 Chapter 4: Discussion 144 Given the high prevalence of aneuploidy across all cancer subtypes, it is important to understand the role of karyotype alterations in cancer. In this thesis, I have evaluated, in two contexts, the molecular mechanisms of how aneuploidy influences oncogenesis under two specific situations. In Chapter 2, I focused on the consequences of random aneuploidy in untransformed cells. I found that cells harboring complex karyotypes are eventually arrested in the cell cycle and displayed features of senescence. Furthermore, these highly aneuploid cells upregulate NF-κB pathway to elicit a natural killer (NK) cell-mediated immune clearance in vitro. Interestingly, by assessing gene expression signatures from around a thousand cancer cell lines across different cancer subtypes using the CCLE database, we found that aneuploidy-mediated NF-κB pathway activation is also prominent in transformed cells. However, such activation may not be sufficient to enhance NK cell-mediated immune responses. In Chapter 3, I investigated the role of frequent chromosome 8 gain in Ewing sarcoma tumorigenesis. Specifically, I focused on dissecting the molecular mechanism by which the chromosome 8 gene, RAD21, acts to mitigate replication stress and promote Ewing sarcoma oncogenesis caused by the EWS-FLI1 oncogenic gene fusion. I found that RAD21 associates with EWS-FLI1-induced transcription-replication conflicts (TRCs) sites and is recruited to the stalled replication forks. Moreover, I found that RAD21 interacts with DNA damage repair initiation proteins under replication stress, strongly suggesting that it plays a role in stabilizing the stalled replication forks to promote efficient DNA damage repair and facilitate replication fork restart. This study revealed a specific mechanism by which aneuploidy can promote cancer growth. Collectively, these two studies argue that the role of karyotype alteration is context- 145 dependent during oncogenesis. Whereas random aneuploidy is detrimental to the cell, such as through influencing immune clearance, tumor-specific aneuploidy can alleviate oncogenic stress, enabling cell with specific karyotypes to proliferate. However, both studies raise additional questions that needed to be answered to better understand how aneuploidy influences cancer development. THE IMMUNE RESPONSE IN ANEUPLOID CANCER I have shown that untransformed cells harboring complex karyotypes elicit NK cell-mediated immune clearance. Consistent with this finding, recent studies have demonstrated that micronuclei caused by chromosome mis-segregation triggers cGAS- STING pathway activation and upregulates the innate immune response (Harding et al., 2017; Mackenzie et al., 2017). Interestingly, the relationship between karyotype alteration and immune response is opposite in tumors: high degrees of tumor aneuploidy is often correlated with increased markers of immune evasion and insensitivity to immunotherapy (Buccitelli et al., 2017; Davoli et al., 2017; Taylor et al., 2018). Computational analysis utilizing the TCGA database revealed that gene expression signatures associated with both the adaptive immunity and the CD8+ T cell- and NK cell- mediated cytotoxicity are significantly decreased in tumors harboring high degrees of aneuploidy across 11 cancer subtypes (Davoli et al., 2017). In addition, it has been shown that cGAS-STING signaling is associated with high chromosome instability (CIN) in cancer, and can promote tumor growth and metastasis, which are indicators of immune evasion (Bakhoum et al., 2018; Liu et al., 2018). 146 It remains unclear how highly aneuploid tumors escape immune clearance during cancer evolution. One possibility is that the aneuploidy triggers an immune response before the cell is transformed or in an early stage of tumorigenesis. In this situation, as the cancer progresses, the aneuploid tumor cells could develop mechanisms to gradually evade immune recognition. For example, aneuploidy-induced DNA damage response can upregulate secretion of proinflammatory cytokines and expression of cell surface antigen that enable immune recognition. Additionally, increased genomic instability or karyotype alterations can also lead to more frequent mutations in the DNA damage repair pathway, which could render the cells unresponsive to high levels of damage and influence downstream effectors. Such scenarios may contribute to immune evasion in aneuploid tumors. Consistent with this notion, it is possible that genomic instability and karyotype evolution triggered by initial chromosome mis-segregation can select for specific chromosome gains or losses which favor immune tolerance of aneuploidy during tumorigenesis. To test this hypothesis, correlation analysis between specific copy number variation and expression levels of the genes related to immune response on altered chromosomes should be performed. For example, chromosomes containing genes that are responsible for cytotoxic function such as granzymes or T cells receptors might show a higher frequency of loss, which could lead to a reduced immune recognition on aneuploidy cells. In fact, a recent study suggested that chromosome 9p arm loss in the head and neck squamous-cell carcinoma (HNSC) plays an important role in immune evasion and the size of chromosome 9 deletion progressively increases during cancer progression (William et al., 2021). Notably the chromosome 9p arm contains many immune response genes including JAK2, 147 interferon-alpha gene cluster, as well as PD-L1 and PD-L2. Accordingly, loss of the 9p arm not only leads to depletion of cytotoxic T cell-mediated immune clearance but also confers PD-1 inhibitor resistance in the neck squamous-cell carcinoma (William et al., 2021). The hypothesis of aneuploidy as a driver for immune evasion should be further validated experimentally. In our NK cell-aneuploid cell in vitro co-culture experimental system, the NK cells (NK92-MI) were already immortalized and activated by constitutive IL2 expression. This led to an aggressive killing of target cells such that most of the aneuploid cells were recognized and killed by NK cells within the 36 hours. At the end of the co-culture assay, the remaining cells were too low of quality to enable karyotyping. With this experimental setting, we lacked the resolution to determine if there were aneuploid cells with specific karyotypes that were recognized and eliminated much slower than others by NK cells. However, experiments to address this question could be conducted in the in vivo settings. For example, mouse cancer cell lines harboring random aneuploidies could be injected into syngeneic mice and the karyotype evolution closely monitored by harvesting sequencing samples at different stages of tumor progression. If specific chromosome gains or losses that lead to alteration of immune related genes are frequently selected for, this would suggest that karyotype alterations could be one way for aneuploidy cancer cells to evade the immune system during oncogenesis. Such analysis could also elucidate potential therapeutic targets in specific cancer subtypes to re-sensitize highly aneuploid tumors for immune surveillance. 148 DISTINGUISHING ANEUPLOIDY PASSENGER AND DRIVER EVENTS As previously discussed in this thesis, the high degree of aneuploidy in aggressive cancer subtypes could be passenger events, or by-product of the high mutational burden and chromosomal instability of proliferating cancer cells. On the other hand, tumor specific aneuploidy events could be drivers of cancer because gains or losses of one, or a group of, genes located on the affected chromosome can yield growth benefits and enhance cell fitness. To distinguish between the aneuploidy driver versus passenger models requires dissecting the molecular mechanisms of oncogenesis in specific cancer subtypes, which may enable identification of potential drug targets. To date, the notion that aneuploidy act as a cancer driver has been demonstrated only in cancer subtypes with rather low mutation background. Ewing sarcoma is one such clear example; 85-90% of these tumors are driven by the EWS- FLI1 fusion oncogene, and patients generally harbor a low mutational burden (Brohl et al. 2014; Crompton et al. 2014). Furthermore, beside the prominent gain of the long arm (8q) or all of chromosome 8, other chromosomal aberrations are low. In thyroid cancers, BRAF or NRAS mutations are the main oncogenic drivers for tumorigenesis. Whereas most of early-stage thyroid cancers are euploid, the loss of chromosome 22q arm occurs as disease progresses. TCGA sequencing data revealed that while only 18% of early stage tumors harbor the loss of chromosome 22q arm, over 50% of the late stage disease show this specific karyotype change (Network et al., 2014). Additionally, chromosome 22q arm loss is generally associated with fast disease progression and poor prognosis (Lan et al., 2020). Ewing sarcoma and thyroid cancers could serve as 149 good experimental models to investigate the role of aneuploidy as a driver of oncogenesis, as they display few mutations, other than hallmark karyotype changes. On the other hand, in more prevalent cancer type, such as breast and lung cancers, aggressive disease progression is often driven by multiple genetic alterations or other oncogenic mechanisms, and these tumors typically harbor highly complex karyotypes rather than single chromosome gains or losses (Taylor et al., 2018). In addition, tumor aneuploidy and chromosome instability are often associated with high intratumoral heterogeneity (Andor et al., 2016). All these factors impose great challenges in studying the role of specific chromosome alterations during oncogenesis and identifying potential aneuploidy driver events during cancer evolution. To better monitor karyotype evolution and identify driver aneuploidies in specific cancer subtypes, we could employ relevant mouse tumor models that harbor clean driver mutations. For example, for pancreatic ductal adenocarcinoma (PDAC), we could use the recent developed K-rasLSL.G12D/+; Pdx-1-Cre (KC) or K-rasLSL.G12D/+; Trp53R172H/+; Pdx-1-Cre (KPC) mouse models (Guerra & Barbacid, 2013) to monitor the karyotype evolution along tumorigenesis by DNA sequencing. If there are specific chromosome gains or losses exhibiting increasing occurrence through different stages of the cancer development, this would be a strong indication for the existence of one, or more driver events for PDAC oncogenesis and/or progression. IDENTIFYING DRIVERS FOR TUMOR SPECIFIC ANEUPLOIDY The mechanism driving specific karyotypes is mostly the cumulative effects of oncogenes and/or tumor suppressor genes that are located on the aberrant 150 chromosomes (Davoli et al., 2013; Sack et al., 2018). Thus, it is of great importance to identify such driver genes in relevant cancer subtypes. This often involves utilizing an unbiased screening approach to look for chromosome-specific gene candidates where the gains or losses of such genes can promote tumor progression upon certain oncogene expression. To avoid confounding factors and provide physiologically relevant growth environment, such screens need to be conducted in either 2D primary cells or 3D organoid primary cultures in the presence of the acutely induced oncogene driver. This may lead to difficulties for the screen, because cell proliferation might be limited to a certain number of passages for gene editing and downstream in vitro analysis. In addition, the oncogene expression level needs to be carefully titrated both to prevent the cells from becoming senescent upon oncogene induction and arresting, and also to appropriately model the oncogene levels in the tumors. In our previous study to identify potential gene targets that drive chromosome 8 gain in Ewing sarcoma, we narrowed down the target gene list in our refined screening library from the 2372 genes encoded on chromosome 8 by interrogating both Ewing sarcoma patient RNA sequencing data and mouse Ewing sarcoma model karyotype data. We used synteny analysis to identify human chromosome 8 genes that are both highly upregulated in the human RNA sequencing co-upregulation network analysis and present in the gain region of the mouse chromosome (Chr15). Using this approach, we curated a list of 26 gene candidates for overexpression screening (Su et al., 2021). The Ewing sarcoma study suggested that rather than using a generic library screening, such chromosome specific screening approach to identify potential aneuploidy gene drivers 151 may require sophisticated analysis to optimize the library before conducting the screening experiment. Another challenge lies in the screening method to specifically look at gene targets on the lost chromosomes. Unlike chromosome gains, where over-expression ORF libraries are effective, loss-of-function screens need to be performed in the case of chromosome loss. The advancement of CRISPR/Cas9 methodology has greatly improved both the efficiency and accuracy of target identification, yet it may not be suitable to identify driver genes for specific chromosome loss, because CRISPR knockout often leads to a complete loss of gene function. Thus, a gene target identified in such manner does not necessarily imply that losing a single gene copy, which is the state enabled by chromosome loss (in the absence of a mutation in the remaining gene copy), is sufficient to promote cancer progression. As an alternative, RNA interference or the inducible CRISPR interference (CRISPRi) screening systems could be used to achieve partial repression of potential targets. In addition, chromosome loss during cancer progression could be a consequence of loss of heterozygosity. If one allele of a tumor suppressor is mutated, the chromosome carrying the normal allele may become more prone to be lost during cancer progression. To test this hypothesis, evolution experiments could be conducted to test whether the frequency of a specific chromosome loss is increased upon inactivation of one copy of the tumor suppressor gene. 152 USING ANEUPLOIDY FOR POTENTIAL THERAPEUTIC IMPLICATIONS The ultimate goal of studying the role of aneuploidy during oncogenesis is to take advantage of aneuploidies for potential therapeutic interventions. First, the degree of aneuploidy can often be used as a prognostic tool to predict cancer aggressiveness and chemo- or immunotherapy treatment outcomes (Ben-David & Amon, 2020). Secondly, in cancers harboring random aneuploidies, targeting certain aneuploidy tolerating genes could be effective in hindering cancer cell proliferation. For example, given the prevalent proteotoxic stress in aneuploid cells, inhibiting key players in the protein quality control system, such as the gene encoding ubiquitin carboxyl- terminal hydrolase 10 (USP10), sensitized the cells for either cell cycle arrest or apoptosis (Dodgson et al., 2016; Donnelly et al., 2014; Tang et al., 2011). Past studies have also focused on exploiting chromosome instability (CIN) for therapeutic implications (Thompson et al., 2017). On one hand, inhibitors that reduces CIN, such as the anaphase-promoting complex/cyclosome (APC/C) inhibitor Tosyl-L-arginine methyl ester (TAME), could be used to prevent rapid karyotype evolution and reduce intratumoural heterogeneity (Sansregret et al., 2017), which in turn slows down cancer evolution and potential acquisition of drug resistance. On the other hand, drugs that further upregulate CIN, such as SAC kinase inhibitors or microtubule stabilizers/destabilizers, could be used to induce extremely high aneuploidy, which would lead to cell death (Thompson et al., 2017). The effort in identifying cancer subtype-specific aneuploidy drivers can provide more potential gene targets located on the specific recurrent gained or lost chromosomes. In the case of Ewing sarcoma, by an ORF overexpression screening 153 experiment for chromosome 8 gain, we have identified four genes ATAD2, RAD21, MTBP, and E2F5 as candidates in which their overexpression significantly reduced EWS-FLI1 induced DNA damage levels and promoted cell proliferation (Su et al., 2021). These genes could be promising targets for hindering oncogenesis in Ewing sarcoma. Whereas the molecular mechanisms of how ATAD2, MTBP, and E2F5 in promoting Ewing sarcoma oncogenesis remain to be elucidated, the analysis of RAD21 in this thesis has shed light on its druggability in cancer treatment. RAD21’s interaction with DNA damage repair initiation proteins such as MRE11 and PARP1 in stabilizing the replication fork and promoting downstream DNA damage repair led us to speculate the potential synthetic lethality between RAD21 and DNA damage repair initiation proteins. In recent years, PARP inhibitors have been extensively studied for its usage beyond BRCA1/2-deficient patients (Dias et al, 2021). Based on the evident RAD21-PARP1 interaction upon EWS-FLI induction, we could further investigate whether a combinatorial therapy for targeting both PARP and RAD21 could cause increased drug toxicity and thus yield a more effective cancer treatment, especially for patients who exhibit high levels of RAD21 in their tumors. In fact, preliminary data indeed suggests that RAD21 overexpression in the Ewing sarcoma cancer cell line TC32 (wild-type for BRCA1/2) led to a significant increase in PARP inhibitor resistance (see appendix). This result supports the potential synergistic effect for targeting PARP and RAD21 especially in RAD21-high cancers subtypes. However, it could be challenging to target RAD21 for cancer treatment. Specifically, knockdown of RAD21 to a third of its level does not have a significant effect on the normal cell physiology (Haarhuis et al., 2014), but complete RAD21 inhibition is lethal. A potential solution for this is to use CRISPRi or RNA 154 interference to fine-tune its gene repression and provide a tight therapeutic window for its proper function. SUMMARY The role of aneuploidy during oncogenesis is highly context-dependent. In this thesis, I have provided mechanistic evaluations on both the detrimental and beneficial effects of aneuploidy on cell physiology. First, I have showed that cells harboring random aneuploidies secret proinflammatory cytokines and elicit NK cell-mediated immune clearance. Secondly, I have showed that chromosome 8 gain promotes Ewing sarcoma tumorigenesis, and more specifically, I demonstrated the mechanism of gain of the chromosome 8 gene RAD21 in mitigating oncogene-induced replication stress by promoting DNA damage repair. These studies help us to better understand how aneuploidy influences cancer development and provide important insights on using aneuploidy as a therapeutic target for cancer treatment. 155 REFERENCES Andor, N., Graham, T. A., Jansen, M., Xia, L. C., Aktipis, C. A., Petritsch, C., Ji, H. P., & Maley, C. C. (2016). Pan-cancer analysis of the extent and consequences of intratumor heterogeneity. Nature Medicine, 22(1), 105–113. https://doi.org/10.1038/nm.3984 Bakhoum, S. F., Ngo, B., Laughney, A. M., Cavallo, J.-A., Murphy, C. J., Ly, P., Shah, P., Sriram, R. K., Watkins, T. B. K., Taunk, N. K., Duran, M., Pauli, C., Shaw, C., Chadalavada, K., Rajasekhar, V. K., Genovese, G., Venkatesan, S., Birkbak, N. J., McGranahan, N., … Cantley, L. C. (2018). Chromosomal instability drives metastasis through a cytosolic DNA response. Nature, 553(7689), 467–472. https://doi.org/10.1038/nature25432 Ben-David, U., & Amon, A. (2020). Context is everything: aneuploidy in cancer. Nature Reviews Genetics, 21(1), 44–62. https://doi.org/10.1038/s41576-019-0171-x Buccitelli, C., Salgueiro, L., Rowald, K., Sotillo, R., Mardin, B. R., & Korbel, J. O. (2017). Pan-cancer analysis distinguishes transcriptional changes of aneuploidy from proliferation. Genome Research, 27(4), 501–511. https://doi.org/10.1101/gr.212225.116 Davoli, T., Uno, H., Wooten, E. C., & Elledge, S. J. (2017). Tumor aneuploidy correlates with markers of immune evasion and with reduced response to immunotherapy. Science, 355(6322), eaaf8399. https://doi.org/10.1126/science.aaf8399 Davoli, T., Xu, A. W., Mengwasser, K. E., Sack, L. M., Yoon, J. C., Park, P. J., & Elledge, S. J. (2013). Cumulative Haploinsufficiency and Triplosensitivity Drive Aneuploidy Patterns and Shape the Cancer Genome. Cell, 155(4), 948–962. https://doi.org/10.1016/j.cell.2013.10.011 Dodgson, S. E., Santaguida, S., Kim, S., Sheltzer, J., & Amon, A. (2016). The pleiotropic deubiquitinase Ubp3 confers aneuploidy tolerance. Genes & Development, 30(20), 2259–2271. https://doi.org/10.1101/gad.287474.116 Donnelly, N., Passerini, V., Dürrbaum, M., Stingele, S., & Storchová, Z. (2014). HSF1 deficiency and impaired HSP90‐dependent protein folding are hallmarks of aneuploid human cells. The EMBO Journal, 33(20), 2374–2387. https://doi.org/10.15252/embj.201488648 Guerra, C., & Barbacid, M. (2013). Genetically engineered mouse models of pancreatic adenocarcinoma. Molecular Oncology, 7(2), 232–247. https://doi.org/10.1016/j.molonc.2013.02.002 156 Haarhuis, J. H. I., Elbatsh, A. M. O., & Rowland, B. D. (2014). Cohesin and Its Regulation: On the Logic of X-Shaped Chromosomes. Developmental Cell, 31(1), 7– 18. https://doi.org/10.1016/j.devcel.2014.09.010 Harding, S. M., Benci, J. L., Irianto, J., Discher, D. E., Minn, A. J., & Greenberg, R. A. (2017). Mitotic progression following DNA damage enables pattern recognition within micronuclei. Nature, 548(7668), 466–470. https://doi.org/10.1038/nature23470 Lan, X., Bao, H., Ge, X., Cao, J., Fan, X., Zhang, Q., Liu, K., Zhang, X., Tan, Z., Zheng, C., Wang, A., Chen, C., Zhu, X., Wang, J., Xu, J., Zhu, X., Wu, X., Wang, X., Shao, Y., & Ge, M. (2020). Genomic landscape of metastatic papillary thyroid carcinoma and novel biomarkers for predicting distant metastasis. Cancer Science, 111(6), 2163–2173. https://doi.org/10.1111/cas.14389 Liu, H., Zhang, H., Wu, X., Ma, D., Wu, J., Wang, L., Jiang, Y., Fei, Y., Zhu, C., Tan, R., Jungblut, P., Pei, G., Dorhoi, A., Yan, Q., Zhang, F., Zheng, R., Liu, S., Liang, H., Liu, Z., … Ge, B. (2018). Nuclear cGAS suppresses DNA repair and promotes tumorigenesis. Nature, 563(7729), 131–136. https://doi.org/10.1038/s41586-018- 0629-6 Mackenzie, K. J., Carroll, P., Martin, C.-A., Murina, O., Fluteau, A., Simpson, D. J., Olova, N., Sutcliffe, H., Rainger, J. K., Leitch, A., Osborn, R. T., Wheeler, A. P., Nowotny, M., Gilbert, N., Chandra, T., Reijns, M. A. M., & Jackson, A. P. (2017). cGAS surveillance of micronuclei links genome instability to innate immunity. Nature, 548(7668), 461–465. https://doi.org/10.1038/nature23449 Network, T. C. G. A. R., Agrawal, N., Akbani, R., Aksoy, B. A., Ally, A., Arachchi, H., Asa, S. L., Auman, J. T., Balasundaram, M., Balu, S., Baylin, S. B., Behera, M., Bernard, B., Beroukhim, R., Bishop, J. A., Black, A. D., Bodenheimer, T., Boice, L., Bootwalla, M. S., … Zou, L. (2014). Integrated Genomic Characterization of Papillary Thyroid Carcinoma. Cell, 159(3), 676–690. https://doi.org/10.1016/j.cell.2014.09.050 Sack, L. M., Davoli, T., Li, M. Z., Li, Y., Xu, Q., Naxerova, K., Wooten, E. C., Bernardi, R. J., Martin, T. D., Chen, T., Leng, Y., Liang, A. C., Scorsone, K. A., Westbrook, T. F., Wong, K.-K., & Elledge, S. J. (2018). Profound Tissue Specificity in Proliferation Control Underlies Cancer Drivers and Aneuploidy Patterns. Cell, 173(2), 499- 514.e23. https://doi.org/10.1016/j.cell.2018.02.037 Sansregret, L., Patterson, J. O., Dewhurst, S., López-García, C., Koch, A., McGranahan, N., Chao, W. C. H., Barry, D. J., Rowan, A., Instrell, R., Horswell, S., Way, M., Howell, M., Singleton, M. R., Medema, R. H., Nurse, P., Petronczki, M., & Swanton, C. (2017). APC/C Dysfunction Limits Excessive Cancer Chromosomal Instability. Cancer Discovery, 7(2), 218–233. https://doi.org/10.1158/2159-8290.cd- 16-0645 157 Su, X. A., Ma, D., Parsons, J. V., Replogle, J. M., Amatruda, J. F., Whittaker, C. A., Stegmaier, K., & Amon, A. (2021). RAD21 is a driver of chromosome 8 gain in Ewing sarcoma to mitigate replication stress. Genes & Development, 35(7–8), 556–572. https://doi.org/10.1101/gad.345454.120 Tang, Y.-C., Williams, B. R., Siegel, J. J., & Amon, A. (2011). Identification of Aneuploidy-Selective Antiproliferation Compounds. Cell, 144(4), 499–512. https://doi.org/10.1016/j.cell.2011.01.017 Taylor, A. M., Shih, J., Ha, G., Gao, G. F., Zhang, X., Berger, A. C., Schumacher, S. E., Wang, C., Hu, H., Liu, J., Lazar, A. J., Network, T. C. G. A. R., Caesar-Johnson, S. J., Demchok, J. A., Felau, I., Kasapi, M., Ferguson, M. L., Hutter, C. M., Sofia, H. J., … Meyerson, M. (2018). Genomic and Functional Approaches to Understanding Cancer Aneuploidy. Cancer Cell, 33(4), 676-689.e3. https://doi.org/10.1016/j.ccell.2018.03.007 Thompson, L. L., Jeusset, L. M.-P., Lepage, C. C., & McManus, K. J. (2017). Evolving Therapeutic Strategies to Exploit Chromosome Instability in Cancer. Cancers, 9(11), 151. https://doi.org/10.3390/cancers9110151 William, W. N., Zhao, X., Bianchi, J. J., Lin, H. Y., Cheng, P., Lee, J. J., Carter, H., Alexandrov, L. B., Abraham, J. P., Spetzler, D. B., Dubinett, S. M., Cleveland, D. W., Cavenee, W., Davoli, T., & Lippman, S. M. (2021). Immune evasion in HPV− head and neck precancer–cancer transition is driven by an aneuploid switch involving chromosome 9p loss. Proceedings of the National Academy of Sciences, 118(19), e2022655118. https://doi.org/10.1073/pnas.2022655118 158 APPENDIX The dosage response to the indicated PARP inhibitors in TC32 Ewing sarcoma cancer cell line with or without RAD21 overexpression. Three technical replicates were included in each experiment. 159