A Gain of Function Senescence Bypass Screen Identifies the Homeobox Transcription Factor DLX2 As a Regulator of ATM-P53 Signaling

A Gain of Function Senescence Bypass Screen Identifies the Homeobox Transcription Factor DLX2 as a Regulator of ATM-P53 Signaling

The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters

Citation Wang, Yifan. 2016. A Gain of Function Senescence Bypass Screen Identifies the Homeobox Transcription Factor DLX2 as a Regulator of ATM-P53 Signaling. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.

Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:26718730

Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA

A gain of function senescence bypass screen identifies the Homeobox

transcription factor DLX2 as a regulator of ATM-p53 signaling

A dissertation presented By Yifan Wang

The Division of Medical Sciences In partial fulfillment of the requirements For the degree of Doctor of Philosophy In the subject of Genetics

Dissertation advisor: Dr. Stephen Elledge Yifan Wang

A gain of function senescence bypass screen identifies the Homeobox

transcription factor DLX2 as a regulator of ATM-p53 signaling

Abstract

Senescence stimuli activate multiple tumor suppressor pathways to initiate cycle arrest and a differentiation program characteristic of senescent cells. We performed a two-stage, gain-of function screen to select for the genes whose enhanced expression can bypass replicative senescence. We uncovered multiple genes known to be involved in p53 and Rb regulation, ATM regulation and two components of the CST complex involved in preventing telomere erosion and additional genes such as REST and FOXO4 that have been implicated in aging. Among the new genes now implicated in senescence we identified DLX2, a Homeobox transcription factor that has been shown to be required for tumor growth, metastasis and associates with poor cancer prognosis. Growth analysis showed that DLX2 expression led to increased cellular replicative lifespan. We found that DLX2 expression inhibited p53 activation, and DLX2 reduced the protein level of upstream activator kinases ATM, DNA-PK. Our data suggest that DLX2 expression reduces the protein components of the TTI1/TTI2/TEL2 complex, a key complex required for the proper folding and stabilization of ATM and DNA-PK and other members of the PIKK family. Over-expression of DLX2 exhibit mutual exclusivity with p53 alteration in cancer patients, suggesting DLX2 may attenuate the p53 pathway during tumor formation. Our functional screen identified novel players that may promote tumorigenesis by regulating the ATM-p53 pathway and senescence.

iii

Attributions The genome-scale library described in Chapter 2 was designed by Steve Elledge,

Mamie Li, and Laura Sack and constructed by Mamie Li and Laura Sack. The customized microarrays used for the primary screen were designed by Michael Schlabach and Qikai Xu. The E6 BJ and E7 BJ fibroblasts used for the screens were generated by

Agata Smogorzewska. Qikai Xu and Laura Sack helped with the microarray hybridization of screen samples. The primary screen and subsequent data analysis was performed by Yifan Wang. Chanhee Kang tested the effect of dominant negative p53 on p21.

The barcoded lentiviral destination vector collection was designed by Steve

Elledge and Yifan Wang. Yifan Wang constructed the vectors, as well as the barcoded gene expression vectors for E6 and E7 sublibrary. The sublibrary screens were performed by Yifan Wang. Qikai Xu performed the ORF-Barcode mapping of the sequencing results. Sublibrary data analysis was performed by Yifan Wang.

All other experiments described in chapter 2,3,4 and 5 were designed by Steve

Elledge and Yifan Wang, and performed by Yifan Wang.

Contents Chapter I. Introduction ...... 1 I. An overview ...... 1 II. Triggers of senescence ...... 2 2-1. Telomere attrition ...... 2 2-2. Oncogene induced senescence (OIS) ...... 4 2-3. Oxidative stress and cellular senescence ...... 6 2-4. Metabolic stress ...... 8 2-5. Chromatin perturbation...... 9 2-6. Non-telomeric damage ...... 10 III. Features and pathways of senescence ...... 11 3-1. Growth arrest...... 11 3-2. Activation of the p53-p21 and the p16-Rb pathway...... 12 3-3. Senescence Associated Heterochromatin Foci (SAHF) and other epigenetic changes in senescence ...... 15 3-4. The Senescence Associated Secretory Phenotype (SASP) ...... 17 3-5. Senescence Associated-β-Galactosidase (SA-β-GAL) ...... 19 3-6. The choice between senescence and apoptosis ...... 19 IV. Physiological and pathological senescence ...... 21 4-1. Senescence as a barrier to cancer in vivo...... 21 4-2. Senescence and aging ...... 23 4-3. Senescence and wound healing ...... 26 4-4. Programmed senescence in embryonic development ...... 26 4-5. Senescence: antagonistic pleiotropy and multitasking ...... 27 V. Conclusion ...... 28 Chapter II: A gain of function ORFeome screen to identify senescence regulators ...... 30 Abstract ...... 30 I. Introduction ...... 31 II. Screen design and rationale ...... 32 III. Results...... 36 3-1. Identification of dominant negative p53 from the screen...... 36 3-2. Deconvolution of the screen data ...... 38 Table 2: candidate genes scored from E6 BJ senescence screen ...... 42 3-3. The Identification of known p53 and Rb pathway genes ...... 44 3-4. Genes scoring in both the E6 and E7 branches...... 49

Table 3: candidate genes scored from both E6 and E7 screens ...... 50 3-5. Telomere maintenance: The CST complex...... 50 3-6. Additional factors linked to senescence or aging identified in both screens...... 51 IV. Discussion ...... 52 V. Material and methods ...... 57 5-1. Cell culture and general procedures...... 57 5-2. Growth curve analysis...... 58 5-3. ORFeome based senescence bypass primary screen...... 58 5-4. Primary screen data analysis...... 59 Chapter III: A barcoded platform for re-screening candidates for senescence bypass ...... 60 Abstract ...... 60 I. Introduction ...... 61 II. Results ...... 62 2-1. Selection of genes for the secondary screen...... 62 2-2. Construction of barcoded sublibraries...... 62 2-3. E6 and E7 Sublibrary screening ...... 64 2-4. Improved screen performance in the secondary screen...... 66 2-5. Validation of selected screen candidates...... 69 III. Discussion ...... 72 Table 5: Summary of mis-annotated and mutated ORFs ...... 74 IV. Material and methods ...... 75 4-1. List of selected complementary genes from 60-mer microarray for sublibrary screen...... 75 4-2. E6 and E7 Sublibrary composition ...... 76 4-3. Construction of barcoded lentiviral vectors...... 78 4-4. LR reaction and transformation in 96-well plate...... 78 4-5. Barcoded ORF sublibrary screen ...... 79 Chapter IV. DLX2 expression bypasses senescence by suppressing ATM-p53 signaling ..... 80 Abstract ...... 80 I. Introduction ...... 81 II. Results...... 82 2-1. DLX2 expression delays the onset of replicative senescence...... 82 2-2. DLX2 induced senescence bypass is not due to analtered TGFβ pathway...... 84 2-3. DLX2 expression suppresses p53 activation ...... 85 2-4. DLX2 expression bypasses Ras induced senescence...... 87

2-5. DLX2 suppression of p53 activation requires the DLL and the Homeodomain. .... 88 2-6. DLX2 expression leads to reduced PIKK kinase ATM, DNA-PKcs and mTOR protein levels...... 89 2-8. DLX2 expression leads to reduced Triple T complex component level...... 92 III. Discussion ...... 95 IV. Material and methods ...... 97 4-1. Growth curve and SA-β-galactosidase assays...... 97 4-2. Plasmids and Cloning ...... 98 4-3. Immunoblotting and Antibodies...... 98 4-4. RT-qPCR ...... 98 Chapter V. Conclusions and Perspectives...... 100 I. Genetic screens to identify novel regulators of senescence...... 100 II. DLX2, cellular senescence and tumor suppression...... 101 III. Perspective Mechanism of DLX2 mediated PIKK destabilization...... 105 Reference ...... 108

vii

List of Figures Figure 1. Overview of senescence ...... 1 Figure 2. Scheme of the ORFeome senescence bypass screen...... 35 Figure 3. Identification of a strongly enriched p53 mutant from the E7 screen sample ...... 37 Figure 4. Overview of the E6 and E7 screen results ...... 39 Figure 5. Network analysis identifies multiple p53 related genes from the E7 screen candidates...... 45 Figure 6. Network analysis identifies multiple Rb related genes from the E6 screen candidates...... 46 Figure 7. Disease, Biofunction and GO term analysis of the screen candidates...... 48 Figure 8. Genes scored in both E6 and E7 screens...... 49 Figure 9. Outline of the construction of barcoded ORF expression vectors for sublibrary screen...... 63 Figure 10. Scheme of sublibrary screens...... 65 Figure 11. Secondary screen results...... 66 Figure 12. Validation of selected screen candidates...... 71 Figure 13. DLX2 expression bypasses replicative senescence...... 83 Figure 14. DLX2-induced senescence bypass is not due to an altered TGFbeta pathway .... 85 Figure 15. DLX2 expression suppresses p53 activation...... 86 Figure 16. DLX2 expression bypasses Ras induced senescence...... 87 Figure 17. DLX2 suppression of p53 activation requires the DLL and the Homeodomain. . 88 Figure 18. DLX2 expression leads to reduced PIKK kinase ATM, DNA-PKcs and mTOR protein level...... 90 Figure 19. DLX2 expression reduces PIKK protein levels during senescence...... 92 Figure 20. DLX2 expression leads to reduced TTT complex component level...... 94 Figure 21. DLX2 over-expression exhibits mutual exclusivity with p53 alteration in breast cancer...... 104 Figure 22. Prospective involvement of Hyaluronic Acid Synthase 3 (HAS3) in DLX2 induced senescence bypass and ATM-p53 attenuation...... 106

List of Tables Table 1: candidate genes scored from E7 BJ senescence screen...... 40 Table 2: candidate genes scored from E6 BJ senescence screen ...... 42 Table 3: candidate genes scored from both E6 and E7 screens ...... 50 Table 4: secondary screen results ...... 67 Table 5: Summary of mis-annotated and mutated ORFs ...... 74

viii

Acknowledgements

I am immensely grateful to my advisor Dr. Elledge. Steve is a pioneer in so many fields and I am always stunned by his brilliant creativity and deep insight. I really admire

Steve’s vision, drive of curiosity, his dedication, and the level of focus in the pure pursuit of knowledge. As a student, learning from the way Steve asks and answers questions, interprets experimental data and comes up with new ideas has been truly inspirational. I continue to be amazed by his scientific instinct to steer my project overcoming the challenges and obstacles we encountered in my thesis work. I am grateful for Steve for having me as a graduate student, and the growth I acquired in his lab will be my life fortune.

And I also want to thank the members in our lab for their constant support. First I want to thank Anna Mazzucco, Laura Sack and Qikai Xu for their help with my original screen; Chanhee Kang for the insightful discussion we had about senescence projects.

Jay Guo, the ultimate biochemistry guru from whom I constantly seek help. I thank

Mamie Li, who has always been so kind and shared with me invaluable skills of DNA cloning. I think I’m also good at it now. I want to thank Yumei Leng and Zhuohan Zhang for kindly helping me with the sublibrary construction, we picked as many as one thousand entry clones from the frozen stocks, and you guys sacrificed your own time to help with the task. Thank you to my current and past bay mates, Mo, Shawna, Rupesh,

Lior, Teresa, Mike and Anna for making our bay a friendly and fun space. Lastly, I offer gratitude to Eric Wooten and Melissa Hinley: the work you guys do is essential for the lab business to run smoothly, and I will never forget your kindness whenever I needed help.

I would also like to thank the members of my DAC committee for supporting my work over the course of the last several years. Drs. James Decaprio, Galit Lahav, Wade

Harper: thank you for your invaluable insight, advice and encouragement. Thank you to the members of my dissertation committee, Drs. Karl Munger, Lee Zou and Marc Vidal:

Thank you for volunteering a substantial amount of time for reading this thesis and coming to my defense.

None of my work could have been accomplished without the support of my family. I want to thank my parents for raising me up, teaching me the virtues and doing everything they could to provide me with the opportunity of a better education. Finally, I want to thank my wife, Huilei Xu, for her encouragement and unconditional support.

Chapter I. Introduction

I. An overview

One of the most important aspects of cell biology is the ability to proliferate.

Replicative senescence is a term first used to describe the lack of indefinite proliferative potential in cells maintained in vitro. Numerous studies have confirmed that various triggers are capable of initiating the senescence program, both in vivo and in vitro, leading to a profound physiological impact. Senescence is involved in tumor suppression, organismal aging, embryonic development and wound healing. In this chapter we will discuss the important triggers that cause senescence and the key features and regulatory pathways of senescent cells, as well as the physiological and pathological relevance of senescence (Figure 1).

Figure 1. Overview of senescence

Figure 1. Senescence can be triggered by multiple stimuli, including short telomere due to telomere erosion, and stimuli that result in non-telomeric DNA lesions such as oncogenic mutation, oxidative stress,

(Figure 1 continued from previous page) chemotherapy and ionizing radiation. These factors activate the DNA damage response which turns on the senescence program. Other stressors such as metabolic stress and epigenetic perturbation also contribute to triggering senescence. Features of senescence include cell cycle arrest caused by activated p53-p21 and p16-Rb pathway, the activation of senescence marker SA-β-Gal, the activation of SASP and the formation of SAHF as well as other chromatin alterations. Senescence is involved in many physiological/pathological contexts, including tumor suppression, aging, wound healing, and embryonic development.

II. Triggers of senescence

2-1. Telomere attrition

Advances since the 90s have linked telomere biology to senescence with molecular level mechanistic explanations; hence, telomere attrition is considered the major trigger of replicative senescence. Telomeres are regions at chromosomal ends composed of tandem nucleotide repeats required for protecting genomic integrity [1].

They prevent chromosome ends from being recognized as sites of DNA breaks and prevent telomere ends from fusing with each other. A human telomere is composed of tandem TTAGGG repeats that span several kilo-bases (on average 5-15 kb), and it has a 3’ overhang that folds back to invade the double stranded region, forming a unique structure called the T-loop [2]. The Shelterin complex recognizes the telemeric repeats, binds to the T-loop and helps suppress cellular machineries from recognizing telomeres as DNA damage sites or initiating unwanted DNA repair processes such as homologous recombination (HR) or non-homologous end joining (NHEJ) [3] [4].

During normal DNA replication, the DNA polymerase can only duplicate template sequences in a unidirectional manner. DNA replication on the lagging strand of the replication fork requires coordinated action of a DNA primase for the continued initiation of the Okazaki fragment synthesis, allowing the replication fork to move on.

However, as the replication fork approaches the very end of a chromosome, there is a

region beyond the reach of the lagging strand primer-synthesis machinery, leaving an unattended, un-duplicated region after each cell cycle. This phenomenon is famously known as the end replication problem [5]. From this perspective, the telomeres will become shorter after each round of cell division, and as cells continue to proliferate, they eventually become critically shortened and dysfunctional [6].

In addition to end replication problem, additional factors have been indicated in affecting telomere attrition and length, for the extent of telomere length loss during each replication observed is greater than that anticipated by the end replication problem alone.

For instance, oxidative stress has been shown to accelerate the telomere shortening process [7]. On the other hand, living organisms have evolved telomerase activity to deal with the end replication problem and preserve telomere length. Telomerase is a reverse transcriptase comprised of a RNA component (TERC) and a protein component (TERT).

The TERT protein is recruited to telomere termini and extends the TTAGGG repeat to increase the telomere length. While the RNA component TERC is ubiquitously expressed, the TERT protein expression is absent in most cells with a few exceptions of certain stem cell populations and activated immune cells [8] [9]. Therefore, the option of telomerase is not available for most somatic cells to offset telomere attrition.

As briefly mentioned above, critically shortened telomeres trigger a DNA damage response (DDR) [10] [11] [12]. The DNA damage response is a complex, hierarchical network composed of DNA damage sensors, apical kinases, many downstream kinases and a myriad of effectors. The effect of DDR activation may be dosage and context dependent, yet in general in replicative senescence the DDR program activates multiple downstream tumor-suppressor pathways to halt cell cycle progression. It is not entirely

clear if persistent telomere DDR signaling is required for all aspects of senescence.

However in senescent cell populations, as few as 5 short telomeres co-localizing with

DDR protein foci resembling Double Strand breaks (DSB) have been observed [13], suggesting that human cells may be quite sensitive to the presence of short/dysfunctional telomeres. Malignant cancer cells bypass replicative senescence by activating mutations for telomerase expression, as telomerase activity is detected in most cancers along with other alterations. Furthermore, studies have shown that the ectopic expression of hTERT can immortalize many cell types such as fibroblasts and epithelial cells, suggesting telomere attrition plays a causal role in triggering senescence [14].

2-2. Oncogene induced senescence (OIS)

Serrano and Lowe first observed that the over-expression of oncogenic H-RasV12 mutation halts cell division, rather than promoting cell proliferation as cancer mutations normally do [15]. This growth arrest is due to the activation of downstream tumor suppressors such as p53 and p16. The arrested cells resemble replicative senescent cells in many ways, including the lack of response to growth factor stimulation, extended cellular morphology and positive staining for the classic senescence marker Senescence-

Associated-β-Galactosidase (SA-β-Gal). Further studies confirmed that the overexpression of many other oncogenes triggers senescence response as well, including

Myc, E2F, Raf, cyclin E [16, 17] [18] [19]. Loss of certain tumor suppressors initiates a senescence response as well. For example, the loss of PTEN causes p53 dependent senescence both in vivo and in vitro [20]. Furthermore, senescence marker positive cells have been observed in many premalignant tissues and are absent from advanced stage cancer [21] [19], indicating senescence may pose a barrier that cancer eventually has to

overcome. Most importantly, the genetic inactivation of key senescence regulators has been shown to promote cancer progression, as we will discuss later in this chapter.

Therefore, the oncogenic mutation and other genomic alterations caused growth arrest is widely accepted as an important model of cellular senescence with strong implication in cancer.

How does the oncogene expression turn on senescence? Just like replicative senescence, OIS cells exhibits many hallmark activation of the DDR [22, 23], including elevated level of protein kinase ATM, ATR, CHK1, CHK2 phosphorylation. They also show foci of γH2AX, an important histone phosphorylation marker for DDR. The consequence of the DDR includes the activation of ATM and ATR, and this may provide an initial block on cell cycle progression (the detailed mechanism for p53 pathway activation will be discussed later in this chapter). However, it is still not entirely clear how OIS causes DNA damage. One plausible explanation is that oncogene expression causes replication stress that results in DNA damage [22] [23]. This notion is supported by the observation that oncogene induced senescence growth arrest typically requires several rounds of cell cycle to establish, and the examination of several fragile genomic loci reveals evidence of hyper-duplication of these foci in senescent cells.

The DNA damage response largely determines the outcome of oncogene induced senescence, however, this does not exclude many other parallel pathways that also play a role in senescence. One interesting idea is that oncogenic activation or tumor suppressor loss may trigger several downstream negative feedback mechanisms to counteract the oncogenic activation. Indeed, the loss of tumor suppressor NF-1, or aberrant activation of

Ras or Raf precipitates several downstream negative feedback pathways that antagonize

Ras signaling [24]. The effects include reduced expression of guanosine nucleotide exchange factor SOS, increased expression of RasGAPs, as well as increased level of

MAPK-kinase phosphatases. Therefore it might provide additional perspective how pro- growth alterations result in a cell cycle arrested state of senescence.

Since oncogenic alteration induces senescence and stalls proliferation, why do we still have cancer? If the oncogene induced senescence is not perfectly robust, even an extremely small population that harbors additional mutations undermining senescence growth arrest may stand out over time, given the strong selective pressure and pre- existing genomic alteration profile. Because our understanding of oncogene induced senescence is still preliminary, it would be beneficial to explore if additional genes that affect senescence may exist, in order to understand why tumors form in spite of the senescence barrier.

2-3. Oxidative stress and cellular senescence Oxidative stress is marked by an excessive cellular level of reactive oxygen species (ROS). Elevated levels of ROS impair the redox balance and damage various cellular components including lipids, protein and DNA. As mentioned above, oxidative stress correlates with telomere shortening rate [7] and thus limits human cell replicative lifespan. The telomeric TTAGGG repeat contains a triple G stretch, which is sensitive to oxidative stress, alkylating agents and UV irradiation, and the telomere region is thought to be difficult for the DNA repair machinery to access [25]. This notion suggests that the oxidative stress damage to the telomere can be an important contributing factor to replicative senescence.

One study from the Campisi group examined the effect of different oxygen level to the replicative lifespan in cultured mouse cells. It turns out that murine embryonic fibroblast (MEF) cells slow down proliferation and partially senesce before spontaneous immortalization occurs when cultured under ambient (20%) oxygen conditions. When

MEFs are maintained under physiological oxygen (3%) conditions, these cells can bypass replicative senescence[26]. Because MEFs have much longer telomeres (>20 Kb) than human cells, and are positive for telomerase expression, their sensitivity to oxygen levels suggests oxidative stress can trigger senescence in the presence of long telomeres. It follows that oxidative stress is also involved in the less well known, non-telomeric damage sites observed senescence [27].

In addition to affecting replicative senescence, oxidative stress has also been shown to be involved in OIS. Lee and Finkel examined ROS levels after senescence induction by HRasV12 in human diploid fibroblast cells. HRasV12 expression results in elevated mitochondrial ROS levels, and the senescence growth arrest could be rescued by a hydrogen peroxide (H2O2) scavenger under ambient oxygen conditions. In addition,

HRasV12 fails to induce senescence when grown below 1% oxygen [28]. These findings suggest that oxidative stress may be activated by other senescence triggers such as mutant

Ras allele expression, and yet it plays a pivotal role in mediating the senescence program and/or cause the DNA damage that engages the downstream senescence execution pathways.

2-4. Metabolic stress

Senescent cells have long been thought to be metabolically active. As investigators start to look further beyond this simplistic view point of senescence, an unanticipated picture of metabolic alteration for senescence induction begins to emerge.

As mentioned above, aberrant replication is thought to cause DNA damage during

OIS, however it is unclear if other types of DNA lesion or stress exist in Ras-induced OIS.

In a recent study, Aird et al found that the intracellular dNTP pools decrease during OIS, and supplementing Ras-expressing cells with exogenous nucleosides can bypass the growth arrest associated with OIS [29]. They further substantiated this finding with evidence that RRM2, a key enzyme in dNTP synthesis, is repressed during OIS. More interestingly, they confirmed that this repression is both necessary and sufficient for inducing senescence, providing a novel linkage between nucleotide synthesis deficiency and cellular senescence [29].

In another study with regard to metabolism changes in OIS, Kaplon et al compared the metabolites of normal human diploid fibroblast with those undergoing OIS.

They discovered that OIS cells exhibit increased levels of pyruvate oxidation, which leads to enhanced tricarboxylic acid (TCA) cycle and mitochondria oxidation, along with elevated redox stress caused by excessive ROS production. Mechanistic study revealed that oncogenic senescence triggers such as Raf activation induces the activation of mitochondrial gatekeeper pyruvate dehydrogenase PDH1 [30]. In BRAF-induced senescence, BRAF mutation lead to increased expression of PDP2, an activating phosphatase of PDH1 and an inhibitor of PDK1, which negatively regulates PHD1 to mediate the OIS associated metabolic reprogramming. More importantly, genetic

restoration of the PDK1-PDP2-PDH1 axis bypasses BRAF induced OIS, demonstrating a pivotal role of pyruvate metabolic alteration in oncogene induced senescence [30].

These recent studies provide compelling evidence that metabolic stress plays a role in senescence induction and maintenance. Since diverse processes such as dNTP synthesis and pyruvate metabolism are both involved in senescence, given the complexity of cellular metabolic network, it is highly likely that novel metabolic stress and other pathways involved in senescence will emerge in the future.

2-5. Chromatin perturbation

Our genome is packed in nucleosomes by histone octamers covering both gene- coding regions and regulatory regions. Alterations of gene regulatory networks are intricately linked to chromatin status, making it reasonable to suppose that senescence can be affected by alteration or regulation at the chromatin level. In human fibroblast cells, treatment of HDAC inhibitors (HDACi) has been shown to cause G1 cell cycle arrest and the activation of senescence markers [31]. Later studies confirmed the senescence inducing effect of HDAC inhibitors in other cell types as well [32] [33] [34].

It is as yet unclear how HDAC inhibitors cause senescence. It seems that in human cells the p16-Rb pathway plays a major role in regulating HDACi induced senescence, while in mouse cells the p53 pathway seems to be indispensable [32]. Furthermore, treatment with HDAC inhibitors most likely suppresses heterochromatin formation and promotes euchromatin formation, which is counterintuitive since senescent cells have large, distinct heterochromatin foci formation thought to contribute to the senescence program. Last, but not least, suppression of the p300/CBP acetyltransferase has also been shown to induce senescence in melanocytes [35]. This creates a seeming paradox that both

activation and inhibition of histone acetylation can cause senescence. Perhaps rather than judging if a cell should go to senescence by a global chromatin effect due to a particular treatment, focusing on individual genes or gene sets that tip the balance to drive senescence may provide meaningful understanding. This requires us to develop thorough knowledge of mediators and effectors of the senescence network, and also to understand the role of the diverse chromatin modification mechanisms in regulating senescence.

2-6. Non-telomeric damage

Both replicative senescence and oncogene induced senescence rely on the DDR pathway to initiate senescence cell cycle arrest. Therefore it is logical to assume agents that induce DNA damage, in addition to short telomeres and oncogene mutation, may cause senescence as long as they can create sufficient DNA lesions recognizable to the

DDR pathway. Indeed, DNA damaging agents such as ionizing radiation and UV- irradiation have been shown to cause cell cycle arrest, and can activate SA-β-Gal and other senescence markers. Even in replicative senescence, genomic profiling of DDR protein localization sites reveals that non-telomeric DNA lesions also widely exist in senescent cells [10].

Moreover, after IR treatment and in other senescence contexts, the Campisi group identified DNA lesion structures named DNA-SCARs [36]. DNA-SCARs are lesions co- localizing with PML bodies, and they are marked by co-staining with critical DDR signaling mediators such as 53BP1 [37], indicating their capability to activate the DDR checkpoint. However, DNA-SCARs are depleted of key DNA repair proteins such as

RPA70 and RAD51. It is likely that DNA-SCARs form when cells are challenged with

excessive DNA damage beyond their repair capability, and DNA-SCAR in turn restrains

DNA repair, resulting in the irreversible growth arrest of senescence [36].

Another surprising, yet interesting, discovery of non-telomeric DNA damage was reported by the Adams group. They observed that senescent cells contain a considerable pool of cytoplasmic DNA fragments [38]. It is shown that senescent cells undergo extensive cellular remodeling, including the remodeling of chromatin structures.

Stunningly, the autophagy/lysosome pathway processes the chromatin structures budding out from the nucleus and destroys the histones from the fragments. This process is thought to correlate with the Lamin B1 loss during senescence, indicative of compromised nuclear envelope integrity. It is unclear if the cytoplasmic DNA fragments would remain stable or even actively contribute to the DDR response. However, it suggests the existence of misallocated, permanent damage and severely compromised genomic integrity in cellular senescence, and this may provide an additional explanation for the irreversibility of senescence [38].

III. Features and pathways of senescence

3-1. Growth arrest.

One of the most distinctive features of senescence is growth arrest. Senescent cells are arrested in the cell cycle, despite the presence of sufficient existing mitogenic growth stimuli under experimental culture conditions. Unlike the reversible growth arrest in quiescence, senescence growth arrest is irreversible, in that it persists even after the removal of senescence trigger. Interestingly, in most cases of senescence, the arrested cells display a DNA profile of largely G1 content [39, 40] [31] [15], sometimes with a small fraction of cells with a G2/M content [41] [15], suggesting the importance of G1/S

phase checkpoint in senescence. This G1/S phase cell cycle arrest is thought to be primarily controlled by a few major downstream tumor suppressor pathways, namely the p53-p21 and the p16-Rb pathway.

3-2. Activation of the p53-p21 and the p16-Rb pathway.

Both p53-p21 and p16-Rb pathways contribute to the G1/S cell cycle arrest in senescent cells [15] [42] [43]. Although oftentimes they are viewed as partially redundant pathways, the p53 pathway is considered the one that establishes the initial cell cycle arrest to allow the full installation of senescence [42].

The p53 tumor suppressor encodes a transcription activator, and it is perhaps the most important and well-studied tumor suppressor. In cellular senescence, p53 is believed to be primarily activated by DDR signaling. DDR kinases such as ATM, DNA-PK,

CHK2, actively modify p53 protein by phosphorylating key serine/threonine residues to enhance p53 stability and transcription activation [44]. P21 expression in turn is activated by p53. It encodes a CDK2-Cyclin E/ Cyclin A inhibitor which nullifies CDK2-

CyclinE/A’s ability to phosphorylate Rb family pocket proteins. The hypo- phosphorylated Rb protein binds to E2F transcription factors and inhibits the transcription of many E2F target genes required for cell cycle progression [45] [46].

Numerous studies have shown by inactivating p53-p21 axis using viral factors or dominant negative mutants, normal cells can achieve a greater replicative lifespan and the senescence growth arrest can be at least partially attenuated.

P53 plays a pivotal role in regulating many biological processes. And unsurprisingly p53 itself is subject to critical regulation by multiple pathways. A plain

description of a linear p53-p21 arm perhaps oversimplifies the complex pathways affecting p53 in various contexts. In fact, p53 activates expression of its own negative regulators, namely MDM2 [47, 48] and PPM1D/WIP1 [49]. MDM2 is an E3 ubiquitin ligase that degrades p53 [50], and PPM1D is a protein phosphatase that inactivates p53

[51] and antagonizes upstream kinases ATM and CHK2. Rather than having a stable, persistent p53-p21 activation as a result of DDR activation, p53 activity constantly oscillates at single cell level after activation by IR [52, 53]. There are some interesting proposals that the deregulation of a p53 negative feedback loop/oscillation may contribute to the decision of entering senescence [54].

The p16-Rb pathway is also involved in the G1/S cell cycle arrest [55] [56]. P16 expression is activated in numerous cases of cellular senescence. The p16 gene encodes a

CDK4/6-cyclin D inhibitor that blocks the cell cycle at early S phase. When p16 is over- expressed, the CDK4/6-CyclinD complexes become inhibited, leading to the hypo- phosphorylation of RB protein [45, 46]. As noted above, hypo-phosphorylated RB protein binds to E2F transcription factors and inhibits the transcription of many E2F target genes required for cell cycle progression. Thus, p16 pathway activation is also capable of stalling cell cycle progression in senescent cells.

It seems the p53 pathway and the p16 pathway operate in a partially redundant manner in senescence. Indeed, oftentimes, to fully bypass senescence requires the inactivation/abrogation of both the p53 and the p16-Rb pathway. This is thought to explain why human cells are particularly difficult to transform. However, the p16-Rb pathway remains unique.

First of all, p16 activation temporally differs from p53 activation. In early replicative senescence in fibroblasts, p53 signaling is activated first in response to short telomeres before the detection of elevated p16 levels [43]. And p16 expression level remain low until cells undergo senescence for longer periods, and this delayed p16 pathway activation points to the regulatory machinery that engages p16, perhaps in a manner parallel to p53 activation.

Secondly, unlike the well-studied link between DNA damage and p53 activation, it is not clear how p16 gets activated in senescence. Many isolated reports have been made to provide connections between senescence stimuli and p16 activation, however a comprehensive understanding of p16 activation mechanism is still missing. It is not surprising that in certain conditions p16 pathway is capable of initiating cell cycle arrest while the p53 pathway is not even activated [57] [58]. This is particularly true for certain epithelial cells. For instance the Human Mammary Epithelial cells (HMEC) are very sensitive to the “culture shock” when maintained in vitro [57]; they upregulate p16 as a response and slow down proliferation without p53 activation, until clones that spontaneously silence p16 take over the population.

Thirdly, unlike the p53 pathway, which is regulated by negative feed-back loop and could oscillate at single cell level, p16 activation is thought to be stable and provide a reliable marker for cellular senescence and aging in vivo [59, 60], thus it is imperative to understand how p16 is regulated. In addition, p16 has several close related family members like p15 and p19, which are tumor suppressors known to be co-regulated with p16 in certain senescence conditions. This indicates a common mechanism may exist for defining senescence with regard to p16 related CKIs [61].

So what do we know about how p16 is activated in senescence so far? Since p16 is epigenetically silenced in certain fibroblast cell lines [62] and is activated when triggered by senescence stimuli, unsurprisingly multiple transcription and epigenetic regulation are involved in p16 activation [63-70]. It is believed that basal level of p16 expression is suppressed by the Polycomb group complex (PRC) [63], and it is possible that additional chromatin regulators such as non-coding RNAs [67] and insulators [70] also contribute to maintaining low p16 level in normal cells. In the context of Ras induced senescence, suppressors such as BMI1 and EZH2 dissociate from the p16 promoter [63], and CUL4-DDB1 ubiquitin ligase recruits H3K4 methyltransferase MLL1 to promote p16 transcription [68]. Several parallel pathways have also been reported. For example Ras-Raf-MEK signaling recruits Ets transcription factors to p16 promoters, opposing the silencing effect of ID1 to activate p16 transcription [69]. However, it is not clear how senescence triggers such as DNA damage signaling immediately activate p16 expression, or if p16 expression is a secondary consequence of a senescence-related differentiation-like program.

3-3. Senescence Associated Heterochromatin Foci (SAHF) and other epigenetic changes in senescence

One distinct feature defining the senescence phenotype is the dramatic remodeling and epigenetic changes occur on chromatin. Senescent cells harbor large-scale, punctate

DNA foci made of condensed heterochromatin known as Senescence Associated

Heterochromatin Foci (SAHF) [71]. SAHF contain elevated levels of heterochromatic features such as Histone hypoacetylation, H3K9 trimethylation (H3K9Me3) with increased binding to Heterochromatin Protein 1(HP1) [71]. SAHF are also characterized

by altered nucleosome composition, namely enriched levels of histone variant macroH2A

[72]and HMGA proteins [73] and depletion of histone linker H1 [74]. The p16-Rb pathway is thought to regulate SAHF [71], since Rb protein has been shown to co- localize with SAHF, and the inactivation of p16 pathway blocks SAHF formation. The

Rb protein is thought to recruit several histone chaperone machineries to guide the formation of SAHF on many E2F target genes necessary for cell cycle progression. The formation of SAHF suggests that the senescence cell cycle arrest is maintained not only by the activation of tumor suppressor pathways, such as the p53 and Rb pathway, but also by the large-scale chromatin reorganization to silence cell cycle gene transcription broadly, increasing the depth and stability of senescence growth arrest.

Besides SAHF formation, senescent cells often display irregular-shaped and enlarged nuclei related to LaminB1 loss. Lamin B1 is a major structural component of the nuclear lamina [75], and it was recently discovered that Lamin B1 level is reduced in replicative senescence and OIS in both humans and mice [76] [77] [76]. Lamin B1 loss leads to compromised nuclear lamina integrity. As mentioned above, Lamin B1 loss is thought to trigger DNA budding in senescent cells, causing genetic alterations and lesions.

From an epigenetic perspective, the multiple Lamina-associated domains (LAD) genome regions are characterized by low-gene expression level. Recent examination of the senescent cell epigenome by bisulfide-DNA sequencing and chromatin- immunoprecitpitaion reveal that these LAD regions become DNA-hypomethylated [78] and enriched with active chromatin marks such as H3K27 trimethylation (H3K27Me3)

[79]. These epigenetic features of senescent cells are also shared by many cancer cells

[79] [78], suggesting the senescence epigenetic landscape may foreshadow that of

malignant cells. Epigenetic changes in senescent cells may not guarantee cell cycle arrest but rather, but could facilitate future malignancy. In sum, senescent cells have large scale, complex epigenetic remodeling and chromatin reorganization that may play important roles in senescence physiology and the pathology of cancer malignancies.

3-4. The Senescence Associated Secretory Phenotype (SASP)

Senescent cells secret a plethora of chemokines, cytokines and growth factors as well as extracellular proteases [80]. Early work in microarray analysis of replicative senescent fibroblast indicated senescent cells confer a strong inflammatory response [81].

Numerous subsequent studies further revealed that senescent cells featured dramatically enhanced secretion of inflammatory molecules [82] [80] [83], especially chemokines and cytokines, as well as growth factors and matrix metalloproteinases. This phenomenon is referred to as the Senescence- Associated Secretory Phenotype (SASP) [80] [84] or

Senescence-Messaging-Secretome (SMS) [83]

Functional genomic screens and detailed transcription profiling have revealed multiple important secreted factors/ligands that mediate the SASP phenotype. These factors, such as IL6/IL8 and IL1 [82] [85] [86] [87], reinforce the senescence circuit in an autocrine manner, and genetic alterations have confirmed that the SASP machinery is both necessary and sufficient for turning on senescence. Further studies established the roles of additional important machineries involved in SASP, such as the NFkB pathway

[82] [88] and the autophagy pathway [89] [90]. It is estimated that most of the cytokine induction in SASP production occurs at the level of transcription induction. Numerous cytokine mRNAs show dramatic increases during senescence, and NFkB activation is thought to be the major signaling mechanism. How is SASP connected with known

senescence pathways such as the p53 or p16-Rb pathway? The over-expression of p21 or p16 does not lead to SASP activation [91], therefore it seems SASP can be decoupled from the cell cycle arrest aspect of senescence. Perhaps SASP is activated by more upstream events, i.e. DNA damage signaling [92]. A recent discovery in our lab identified

GATA4, a transcription factor that controls SASP in replicative senescence and sufficient for inducing senescence on its own [93]. GATA4 level increases during senescence and aging, and this increase is thought to be due to an alteration in the activity of selective autophagy. It would be interesting to parse out the potential interplay of autophagy and senescence triggers such as DNA damage to understand how SASP is activated in senescence.

Aside from the autocrine senescence enforcing function of SASP, the secreted factors can also act on neighboring cells in a paracrine manner, leading to various consequences depending on the nature of the trigger, target molecule and cellular context as well as genetic constituents [84]. On one hand, analysis of SASP molecule signaling on premalignant cells such as BRafV600-expressing melanocytes have confirmed that paracrine signaling helps spread senescence to normal neighboring cells, consistent with the role of SASP in tumor suppression. On the other hand, when paracrine signaling involves cells of different origins or background, the effects are often complicated [94-

97]. For example, co-injection of senescent fibroblast cells with malignant cells of epithelial origin into mice indicates that senescence in fibroblasts may confer a tumor- promoting function [96]. Several secreted factors have also been shown to be tumorigenic

[94] [84]. The collective outcomes of the SASP have to be determined by the plethora of factors secreted, the availability of receptor signaling pathways in target cells, as well as

the genetic makeup of target cells [84]. Moreover, recent studies show the effects of the

SASP go beyond fibroblast-fibroblast and epithelia-fibroblast interaction; they are also implicated in wound healing and senescent cell clearance [98, 99]. Much remains to be uncovered with regard to the impact of SASP in physiological context.

3-5. Senescence Associated-β-Galactosidase (SA-β-GAL)

Senescence Associated-β-Galactosidase was described by the Campisi group in

1995 and has become the most widely accepted senescence marker to date. The Campisi group found that SA-β-GAL staining strongly associates with senescence cells cultured in vitro and aged tissue in vivo [100]. The enzymatic activity of SA-β-GAL derives from lysosomal β-galactosidase encoded by the GLB1 gene [101]. In normal, proliferating cells, lysosomal galactosidase activity can be detected at low pH (pH 4) conditions. In senescent cells this activity is strengthened and it can be detected by suboptimal condition of pH 6. This increased activity is thought to reflect an expansion of the lysosomal compartment in senescent cells. While it is generally accepted as a senescence marker, it is unclear if SA-β-GAL activity has functional relevance to the senescence phenotype. Certain growth conditions such as prolonged culture under confluency conditions may also lead to increased SA-β-GAL activity. Therefore, SA-β-GAL should be used in combination with other observations such as lack of proliferation, activation of p53 and/or p16, and the presence of SAHF formation to confirm senescence.

3-6. The choice between senescence and apoptosis

Apoptosis describes a coordinated program that terminates cellular life and ends up with the engulfment and destruction of cellular constituents. Like senescence, apoptosis is an optional response when cells face various stresses such as genotoxic stress,

oxidative stress and toxic chemical treatment. Both senescence and apoptosis require a common master regulator - the p53 tumor suppressor - to execute these actions across various contexts. For example, DNA damage such as ionizing radiation turn on senescence response mediated by the p53 pathway in fibroblast [102], however in thymocytes ionizing radiation treatment leads to p53-mediated cell death [103] [104].

Another example is that senescent fibroblast cells resist cell death by ceramide treatment while endothelial cells do not [105].

Why do different cells choose differentially between senescence and apoptosis?

One explanation is that the cellular response to stress under senescence conditions may involve genes that confer resistance to apoptosis. As mentioned above, p21 is a critical p53 target gene involved in senescence related cell cycle arrest, and beside its role in cell cycle regulation, p21 has also been shown to inhibit apoptosis [106]. The activation of p21 may partially explain the apoptosis resistance in senescence. In fact, fibroblasts can be sensitized for apoptosis following DNA damage treatment by over-expressing the

MYC oncogene, and the underlying mechanism is thought to involve MYC induced suppression of p21 transcription [107]. Microarray studies have also confirmed that senescent cells activate multiple pro-survival, anti-apoptotic pathways [81]. Therefore, the activation of senescence may partially explain the dominance of senescence in certain conditions.

The activation amplitude and mechanism of key regulators such as p53 can clearly impact the outcome following stimuli such as DNA damage. Quantitative analysis of p53 activation suggests that apoptosis requires overcoming a certain signaling threshold to be fully executed [108] [109]. Qualitative evaluation of different p53

modifications may provide additional explanation to the decision making process. For example, phosphorylation on Serine 46 of human p53 has been shown to specifically promote apoptosis [110]. These different modes of p53 activation may lead to selective affinity to subsets of p53 effector genes, resulting in either senescence or apoptosis.

Moreover, binding to certain p53 co-factors could affect the cellular logic of apoptosis in certain contexts. Aside from p53 modifications, it is possible that modifications on other key regulators may also confer selectivity to senescence or apoptosis, and they remain to be uncovered. In sum, the choice between apoptosis and senescence may be attributed to multiple relevant factors, and much remains to be revealed with regard to this question.

IV. Physiological and pathological senescence

4-1. Senescence as a barrier to cancer in vivo

The concept of senescence as an irreversible growth arrest state is consistent with the speculation that it may play a tumor suppressive role in vivo. Replicative senescence is largely due to telomere shortening and deprotection followed by DDR signaling and tumor suppressor pathway activation. Consistently, in vivo studies using mouse models with genetically abolished telomerase activity show reduced incidence of tumorigenesis, demonstrating a positive role of senescence in tumor suppression [111] [112] [113]. The discovery of oncogene-induced senescence (OIS) further expanded the horizon of observation that the cells are vigilant enough to activate senescence when excessive oncogene expression is experienced [15]. Some of the initial criticisms regarding OIS involve the physiological relevance of this type of senescence in vivo, because, for example, turning on Ras in the tissue culture setting induces senescence and requires very high level of oncogene expression which may be supra-physiological. It is unclear if

oncogenic alterations other than Ras can also trigger senescence in cultured cells or disease settings. This issue was addressed in early 2000s, when series of reports confirmed that other than Ras activation, a variety of oncogenic insults, even including tumor suppressor loss, can trigger senescence. Furthermore, evidence of senescence has been found in premalignant samples and but absent in malignant, more advanced cancer, suggesting senescence may act as a barrier against tumorigenesis [19] [20] [21] [114].

Genetic modeling with the activation of a single allele of an oncogene was shown be sufficient to turn on senescence, and in mouse models with genetically compromised senescence pathways, tumor progression is much more aggressive. These findings strongly suggest senescence act as a barrier to tumorigenesis in vivo.

With the discovery of SASP, it is now known that senescent cells secret a plethora of cytokines, chemokines, growth factors and extracellular matrix proteinases, thus raising the possibility of higher level of interaction between senescent cells and other cell types, which may have an impact on tumor suppression. Indeed, early work indicates that the re-activation of the p53 tumor suppressor gene in mouse models leads to tumor regression accompanied by innate immune response mediated clearance of senescent cells [98]. A breakthrough finding in 2011 revealed that senescent cells are subject to active immune-surveillance, as CD4+T cells have been observed co-localizing with premalignant senescent cells [115]. CD4+T cells along with other cells such as monocytes and macrophages can remove pre-malignant senescent cells as a part of an immune surveillance program. Impaired immune-surveillance of senescent cells results in the development of hepatocellular carcinomas in mouse models. This finding reveals an extra layer of protection by senescence against tumorigenesis [115].

How do we take advantage of our knowledge in senescence to fight cancer? Early studies show malignant cancer cells retain senescence capability under certain conditions.

This idea is appealing because many cancers have evolved with impaired apoptotic pathways, and the induction of senescence in those cells may provide a therapeutic window to halt disease progression. Indeed, several studies have exploited the potential of therapy-induced senescence by pharmacological induction with DNA damaging agents or key tumor suppressors (reviewed in [116] ). Interestingly, PTEN inhibition has also been used for inducing senescence in mouse models, consistent with previous reports that

PTEN loss has this capability [117]. This represents a rare case where tumor suppressor inhibition may be exploited for cancer treatment. Another interesting idea, although highly speculative, could be exploited with the recent advances in Chimeric Antigen T cell therapy. By coupling protein moieties/domains that can recognize senescent cell surface antigen to T cell receptor, engineered T cells may boost the immuno-clearance of pre-malignant senescent cells to minimize future cancer risk. To sum up, cellular senescence acts as a physiological anti-tumor barrier in vivo, and exploiting the knowledge of senescence may provide positive prospects to future cancer therapeutics.

4-2. Senescence and aging

Aging can be largely attributed to the impairment of tissue/organ homeostasis and declining functionality over time, as well as the onset of age-related diseases, especially in incidents of degenerative diseases. Senescence has been suggested to serve as the underlying mechanism for tissue and organismal aging since its conceptualization,.

Numerous reports have confirmed that in mice and humans, senescence markers such as

SA-β-GAL staining, p16, DDR activation and SASP can be readily detected in various

old and aged tissues [118] [119] [120] [121]. It has been speculated that if senescence occurs in important cell lineages and stalls their proliferation, it may lead to a deficit of a functional cellular population and a decline in normal biological functions at the tissue and organ level may follow. Recent studies of the key senescence regulator p16 support this hypothesis. Several reports in 2006 showed that p16 is up-regulated in stem cell populations such as pancreatic islet stem cells, neuronal stem cells and hematopoietic stem cells, meaning p16 can constrain regenerative capacity. The genetic ablation of p16 expression can restore the aging-associated decline in tissue regeneration [122] [59] [60].

A more recent study further confirmed that quiescent satellite cells, the muscle stem cell type, can directly be converted to senescent cells during aging, leading to reduced muscle regeneration and wound healing [123].

In addition to the adverse aging-effect associated with proliferation arrest, the senescence program may affect aging via the secretion of SASP factors. Senescent cells may continuously secret cytokines, chemokines, growth factors and matrix metalloproteinases, and the collective effect of these factors may perturb tissue structure and integrity. By cleaving proteins such as cell surface receptors, ligands and extracellular matrix component proteins, senescence alters a cell’s microenvironment. On the other hand, the complex signaling outcome of the SASP repertoire between senescent cells and nearby cells may have profound effects on cell identity and function. One such example is that the SASP factors promote epithelial to mesenchymal transition as shown by the Campisi group [80]. Furthermore, prolonged SASP secretion may result in chronic inflammation, accompanied by cell death and lymphocytes/macrophages infiltration, which may result in various age-related diseases [124]. In sum, the senescence secretory

phenotype may lead to impaired tissue microenvironment and chronic inflammation, providing a plausible link between senescence and aging.

It is worthwhile to mention a recent groundbreaking study by the van Deursen group, which systemically tested the contribution of senescence to the aging phenotype.

Researchers engineered a mouse strain that expresses a toxic gene driven by the p16 promoter. Treatment of a compound AP20187 activates the toxic gene and triggers apoptosis in the p16-expressing cells [125]. In this way they were able to investigate the effect of systemic ablation of senescent genes in aging. It turns out the clearance of senescent cells prevents numerous age-related phenotypes, and even late-life clearance of senescent cells reduces pre-existing aging disorders [125]. These findings provide compelling evidence that senescence directly contributes to aging in certain conditions.

Because the p16 promoter used in the study may represent a subset of senescence population in vivo, it would be interesting to apply the toxic gene tool with the promoters of other senescence marker genes. Since in this study a progeroid BuBR1 mouse aging model was used, it is important to extend the senescence clearance to wild type genetic background for broadening our knowledge. An interesting follow-up study explored the possibility of pharmacologically targeting senescent cells, and by using a combination of two compounds, dasatinib and quercetin, researchers were able to achieve reduced senescence burden in mice as well as improved age-related phenotypes [126]. Hopefully our knowledge in senescence will eventually guide us to achieve improved health span and longevity.

4-3. Senescence and wound healing

Senescent cells have been observed in patients with liver Cirrhosis. Krizhanovsky et al investigated the biology of senescence in damaged liver and revealed an unexpected role of senescence in wound healing [99]. They found that when liver damage in induced, hepatic stellate cells (HSC) become activated and start to differentiate, proliferate, and secrete extracellular matrix proteins. However, the activated cells eventually become senescent, downregulate the expression of ECMs and start to secret SASP factors, including MMPs, which restrain fibrosis by degrading the ECM. Other SASP components, like chemokines, attract NK cells for the immune-clearance of HSCs to facilitate the wound healing process. When the senescence response is inhibited in the

HSCs, the wound healing process is burdened with excessive tissue fibrosis, resulting in prolonged time for the wound to heal [99]. A later similar report also confirmed the positive role of senescence in cutaneous wound healing [127]. These findings suggest additional contributions of senescence in physiological settings, and the secretion of

SASP factors, in particular MMPS and chemokines, plays critical role to benefit proper tissue function.

4-4. Programmed senescence in embryonic development

Cellular senescence has been considered a result of facing harsh, oftentimes unfavorable stresses such as tumorigenic mutation, wound or organismal aging. Until recently, two reports brought up an interesting idea that senescence may be an active, intrinsically programmed process indispensable for normal embryonic development.

Storer et al and Munoz-Espin et al observed that cellular senescence can be detected in diverse locations in embryos, including the epical ectodermal ridge, the neural

roof plate, the mesonephros and the endolymphatic sac of the inner ear [128, 129]. Both groups confirmed senescence by the presence of SA-beta gal staining, SAHF, p21, cell cyle-G1 phase arrest and the absence of proliferation marker Ki67 staining. In line with the activation of senescence, enhanced SASP gene expression was also observed. Further analysis revealed that senescence occurred in embryonic development, independent of

DNA damage signaling or p53 activation. The cell cycle regulator p21 is indispensable in this senescence context, as p21 null embryos exhibit a greatly reduced SA-β-Gal positive population [128, 129]. It has been suggested that the embryonic senescence population is eventually resolved by apoptosis and immune-clearance. The loss of senescence caused by p21 ablation can be compensated by apoptosis but still lead to detectable developmental abnormalities [128]. The embryonic senescence program appears to play an indispensable role in development, and it closely resembles the features of conventional senescence pathways: these cells stall proliferation, turn on key tumor suppressor pathways and secret SASP factors [128, 129]. The impact of senescence on embryonic development may be two fold. First, it may limit the proliferation of certain populations, thus it may direct proper tissue/organ cell type balance and architect. Second, the secretion of SASP factors may serve as a means of spatial-temporal signaling, guiding the interaction between specific cell populations. This work has provided an opportunity to broaden our knowledge of senescence, and it would be interesting to explore in depth of the intrinsic similarities and differences between senescence in embryos and in adults.

4-5. Senescence: antagonistic pleiotropy and multitasking

As mentioned above, senescence is a beneficial, potent barrier against cancer in vivo. However, senescence is also involved in aging. The paradox of these opposing

effects is termed antagonistic pleiotropy [130]. From an evolution standpoint, senescence to reduce cancer in early life increases a young organism’s viability and reproductive success. Thus, it provides increased fitness during evolution. The trade-off of aging may be tolerated as long as the overall fitness remains positive for reproductive and survival success for the progeny to secure the continuation of reproduction.

The recent finding of programmed senescence during embryonic development challenges the traditional view of senescence primarily as a stress-response program. The developmental activation of senescence hints that the evolutionary origin of cellular senescence is to coordinate tissue and organ development. The embryonic senescence program shares strikingly similar features with OIS and replicative senescence, suggesting that senescence may acquire additional roles as a stress response mechanism later in evolution[131]. If the commonly known senescence program evolved from an embryonic developmental predecessor as extended function, it would not be so surprising if the incremental, stepwise nature of evolution leads to trade-offs and even seemingly paradoxical consequences like senescence.

V. Conclusion

Cellular senescence has been recognized as multistep process in response to diverse stress stimuli. Short telomeres, oncogenic insults, oxidative stress and metabolic stress, as well as epigenetic perturbation by chemical treatment all lead to cellular senescence. The core senescence program is wired by the activation of key tumor suppressor pathways, namely the p53-p21 pathway and the p16-RB pathway, along with distinct chromatin changes known as SAHF and secretory pathway changes known as

SASP. Senescent cells are resistant to apoptosis, and they actively communicate with

other cell types and shape their microenvironment. These changes collectively determine the profound physio-pathological role that senescence plays in vivo, including tumor suppression, wound healing and aging. Much of our knowledge in senescence can be attributed to the discovery of new molecules, and new genes involved in senescence regulation and physiology, though it is clear that much remains to be uncovered. In the following chapters I will discuss the identification of new genes related to senescence discovered by functional genomics.

Chapter II: A gain of function ORFeome screen to identify senescence regulators

Abstract

Cellular senescence is a pleotropic program that actively participates in tumor suppression, wound healing and organismal aging, as well as embryonic development.

One fundamental feature of senescence is the irreversible cell cycle arrest, and this is thought to have a profound impact on tumor suppression and aging. Understanding the regulatory genes and molecular pathways involved can lead to new insight on senescence.

We performed a large-scale gain-of-function screen to interrogate the human ORFeome to uncover new senescence regulators. We took advantage of sensitized cellular background for senescence to selectively screen for genes affecting the p53 pathway or the p16-Rb pathway, as well as novel regulators that may affecting both pathways. We utilized microarray-based technology to assess the senescence bypass phenotype for

15,000 Open Reading Frame (ORF) simultaneously. From the screen we identified multiple genes known for regulating the p53 and the Rb pathway, as well as novel genes that may affect these pathways. In addition, we also identified genes involved in telomere length maintenance and aging. Our screen results represents a novel gene set that may lead to new insight to senescence.

I. Introduction

Senescence contributes to tumor suppression, wound healing, embryonic development and organismal aging by engaging tumor suppressor networks, halting cell cycle progression, remodeling epigenetic landscape and secreting a plethora of inflammatory factors. Given the profound physiological role of senescence and the various biological processes it is involved in, much remains to be understood. Senescence is known as a tumor suppressor mechanism in vivo, in that senescent cells are arrested in cell cycle, and face surveillance by the immune system. How do premalignant cells overcome this barrier to survive and proliferate? Past studies have pointed out that the inactivation of tumor suppressor p53 impairs the senescence program and p53 is frequently mutated or deleted in cancers. However there are many cases of cancer with intact p53 gene. Perhaps the activation of negative regulators of the p53 pathway, or the inactivation of p53 effector is responsible for bypassing senescence or promoting tumorigenesis. It is very likely that new genes affecting senescence exist in cancer. In addition, senescence contributes to aging at the organismal level, though it is unclear how key senescence pathways are activated during aging and by various stress conditions. For example, while p16 signaling is a master regulator of senescence and widely-used senescence marker in vivo, we still do not fully understand how p16 becomes upregulated by specific stimuli. Moreover, the identification of key genes affecting senescence phenotype has consistently provided keys for deepening our insight in the field. One classic example is the discovery of SASP. Loss-of-function screens for genes involved in cytokine and chemokine signaling lead to the development of the SASP concept of

senescence. Given the above, we decided to utilize functional genetic screen to search for new genes that regulate senescence.

Previous screens in senescence were largely based on loss-of-function phenotypes by RNA interference. While siRNA or siRNA mediated knock-down is a powerful approach for the screening purposes, RNAi screening efficiency can be compromised by limited siRNA knock-down efficiency and widespread off-target effects. Technically loss-of-function senescence bypass screens only identify positive regulators of senescence in a senescence bypass setting. Furthermore, most of the published senescence screens are based on OIS models. While OIS and replicative senescence share a common downstream regulatory network, the nature of specific OIS models may be biased by the specific oncogene applied. For example it takes up to 7 days for Ras induced senescence to arrest the cell cycle. However, in Raf induced senescence the cell cycle arrest can occur within 24 hours, indicative of variation in the senescence execution mechanisms. To broaden our knowledge about senescence, we decided to perform a gain-of-function screen in normal diploid fibroblast cells specifically designed for the bypass of replicative senescence.

II. Screen design and rationale

We chose to perform the replicative senescence bypass screen in BJ foreskin fibroblasts, because this cell line responds well to telomere shortening and telomerase expression. Furthermore, this cell line has been widely used in senescence studies in the past, and it was the first example of human primary cell line immortalized by telomerase expression.

Multiple pathways have been reported to contribute to senescence. For replicative senescence to occur, cells need to be maintained in tissue culture for extended periods of time for telomeres to undergo sufficient shortening, and various downstream effector senescence pathways may cause a great amount of intrinsic heterogeneity of replicative lifespan. One classic experiment done in the 1980s shows that two cell clones derived from one mitosis event may have a significant difference (8 population doublings) in the replicative lifespan [132]. In addition, effects of the abrogation of one senescence pathway may be masked by the effects of the remaining pathways in the background. To increase the screen sensitivity we decided to use viral oncoproteins to create sensitized genetic backgrounds for screening purposes.

We used HPV E7 oncoprotein expressing BJ cells to inactive the Rb branch of the senescence pathway, therefore in the E7 BJ cells the p53 pathway is the major pathway of senescence. Similarly, we created HPV E6 oncoprotein-expressing BJ cell line, with the Rb pathway controlling senescence. These systems allow us to identify senescence regulators of the p53 pathway from E7 BJ cells, and regulators of the Rb pathway from the E6 cells. Since it is possible that novel mechanisms for senescence may exist, we reasoned that regulators affecting parallel pathways independent of p53 or Rb could also be scored in both cell lines regardless of the HPV viral protein expressed.

Some reports suggest the p16-Rb pathway is perhaps not as robust as the p53 pathway in BJ fibroblast, for the p16 promoter is partially methylated in normal BJ cells.

While it is possible that the p16 induction is weakened in BJ cells in senescence due to epigenetic silencing, senescent BJ cells caused by Ras induced senescence still have a robust p16 induction. It follows that depending on the nature and intensity of the specific

senescence trigger, p16 pathway can be engaged in BJ cells. Moreover, the p16 family of

Cyclin-dependent Kinase Inhibitors (CKIs) includes highly-related members such as p15, which is involved in senescence and acts independently of p16 in tumor suppression.

Therefore we believe our screen system could identify relevant regulators of both the p53 and the p16-Rb pathways.

To confirm the selectivity of our sensitized screen background, we performed growth curve assays to determine the replicative lifespan of E6 and E7 BJ cells expressing hairpins targeting p53 or p16. We found that p53 shRNA extends replicative lifespan in E7 cells compared to control hairpin, while in E6 cells Rb hairpin extended replicative lifespan, confirming the sensitivity of our screen background (A. Burrows, unpublished data).

We generated a lentiviral library containing ~15000 Open Reading Frames (ORFs) from the human ORFeome 5.1 (http://horfdb.dfci.harvard.edu/hv5/) [133] and infected mid-aged, proliferating E6 or E7 BJ cells by lentiviral transduction at a low multiplicity of infection (M.O.I) of 0.5. Cells were puromycin-selected and subsequently passaged for a prolonged period after the majority of the population ceased proliferating due to replicative senescence. Cells containing individual ORF that bypassed senescence retain their proliferation potential despite the widespread onset of senescence in the screen cell population, and thus become relatively enriched in the end sample (Figure 2). We quantified the relative enrichment of each ORF in the pooled library with a microarray hybridization protocol. First we recovered the ORFeome DNA from the start and end sample genomic DNA by PCR amplification. The start and end PCR products were then used for generating Cy3 or Cy5 labeled complementary RNA probes for competitive

microarray hybridization that allowed the quantitation of enrichment score measured by the Cy5/Cy3 ratio. We utilized customized microarrays made of multiple oligonucleotide probes for each ORF for quantification and plotted genes based on their average log2

Enrichment score and Z score across screen replicates as a measure of senescence bypass phenotype.

Figure 2. Scheme of the ORFeome senescence bypass screen.

Figure 2. The human ORFeome library 5.1 was introduced to proliferating E6/E7 BJ fibroblasts by lentiviral transduction at low Multiplicity Of Infection (M.O.I.) of 0.5. Transduced cells were pooled together and sampled as the start sample and the remaining cells were then passaged after the population became senescent. Cells that bypassed senescence became enriched in the end pool as indicated by the schematic growth curve. Genomic DNA was extracted from Start and End sample for recovering the ORFeome by PCR. Start and End sample PCR products were used to generate differentially labeled cRNA probes for microarray hybridization.

III. Results.

3-1. Identification of dominant negative p53 from the screen. As a quality control measure, we examined the ORFeome PCR products of screen samples by gel electrophoresis. Surprisingly in the end samples of E7 screen we observed a rather strong band in the PCR product across all three replicates (Figure 3). This indicates a strong ORF is enriched in screen. We cut out these band and subcloned the retrieved DNA into TA-cloning vector, sequenced 30 clones and found these bands all mapped to a p53 ORF with two point mutations (P72R and P278A). Cross examination with the ORFeome library catalog confirmed these mutations are pre-existing in the

ORFeome collection, not acquired from de-novo mutation during the screen. We tested this p53 ORF by expressing it and found it is a dominant negative mutant that its expression suppressed p21 expression (Figure 3). This finding suggested that the E7 screen could successfully identify regulator(s) affecting the p53 pathway in senescence.

We noticed that the dominant negative p53 ORF failed to bypass senescence in the E6 screen despite the strong enrichment in the E7 screen, indicating the selectivity of the sensitized E6 and E7 background for identifying genes related to specific pathways.

Figure 3. Identification of a strongly enriched p53 mutant from the E7 screen sample

Figure 3. Upper panel: ORFeome PCR products of start and end screen samples. The red stars indicate the PCR product corresponding to the strongly enriched mutant p53 ORF in the library. Lower panel: ORFeome library p53 construct was subcloned into a tet-inducible vector. 1ug/mL Doxycyclin (DOX) treatment of 48 hours resulted in induction of p53 protein and reduced the basal level of p21 expression.

3-2. Deconvolution of the screen data To deconvolute the screen results, we tested two customized microarrays made of

25-nucleotide (25-mer) probes or 60-nucleotide (60-mer) probes for each ORF for quantification purposes. While the two platforms produced largely overlapping results, the 25-mer array contained greater than 10 probes per gene compared to the 60-mer array that contained 3 probes per gene on average. Therefore the 25-mer array carries higher fidelity with potentially less bias caused by outlier probes due to signal saturation or cross-hybridization. This was confirmed by a much higher log2 enrichment score of the p53 mutant derived from the 25-mer array hybridization than that from the 60-mer array.

Therefore, we focused on the 25-mer array hybridization data for the primary screen analysis.

We plotted the Z score and log2 enrichment score of each screen obtained from

25-mer microarray hybridization. (Figure 4). We set a Z threshold of 1.65 (P<0.05 one tailed test) for each arm of the screen. Genes above the threshold (log2 enrichment >1.3, n=433 genes for E7 screen, log2 enrichment >1.19, n=340 genes for E6 screen) were considered as hits that increase the replicative lifespan in E6/E7 cells (For gene lists see table I and II).

Figure 4. Overview of the E6 and E7 screen results

Figure 4. Genes were plotted by Z score (left Y-axis) and Log2 Enrichment score (right Y-axis) for E7 senescence screen (left) and E6 senescence screen (right). Unscored genes shown in grey, and genes above Z score of 1.65 shown in red. A mutant p53 ORF was recovered as an outlier from the E7 screen.

Table 1: candidate genes scored from E7 BJ senescence screen.

Symbol Z Score Symbol Z Score Symbol Z Score Symbol Z Score Symbol Z Score TP53 7.029 CYP20A1 2.151 CDH15 1.943 LRRTM4 1.801 FOXD4L6 1.698 HSD17B14 4.032 BDKRB1 2.146 ADD3 1.941 MESP2 1.801 FNDC7 1.696 HOXB6 3.860 PTPN11 2.144 CCDC99 1.940 NIPBL 1.801 XKR9 1.695 CDK6 3.574 FASTK 2.134 FAM160B1 1.938 PLA2G4A 1.799 C1orf2 1.694 MYF6 3.505 PRKD3 2.127 STXBP3 1.937 LY6H 1.799 KCNH7 1.693 DLX2 3.362 CCDC67 2.127 CTNNA2 1.936 DYNC1LI1 1.797 ELP2 1.692 CHRD 3.321 TMEM79 2.116 DBNDD2 1.935 SIAE 1.796 KRT13 1.690 IGKC 3.218 SLC41A3 2.115 WDR34 1.935 SYT14 1.793 ACLY 1.689 LCE1B 3.158 C21orf94 2.113 LSM14B 1.932 SMARCAL1 1.791 PLEKHA9 1.689 IFNA4 3.061 OTOP1 2.111 ZNF540 1.930 SERHL2 1.791 PLEKHA5 1.688 SLC12A8 3.003 PTBP1 2.101 OSMR 1.929 PRICKLE2 1.790 DDX23 1.683 GDPD2 3.002 CMTM6 2.101 ADORA2A 1.928 C16orf86 1.789 LOC284837 1.683 PTK2B 2.997 SAMSN1 2.099 ITGB6 1.925 ST7 1.789 MED24 1.681 KRTAP10-1 2.964 PCDH10 2.098 SH2D2A 1.925 NPFF 1.789 USP22 1.680 KRTAP10-7 2.945 RCBTB2 2.093 ZNF573 1.925 ZNF768 1.789 FASTK 1.680 LOC652276 2.942 ATP6V1B1 2.093 CWF19L1 1.921 WDR60 1.787 PCYOX1 1.679 CDH18 2.904 AQP10 2.091 TEN1 1.920 ADAM21 1.786 CCBE1 1.679 IGKC 2.859 ARHGAP12 2.091 C10orf137 1.920 THAP6 1.781 IPO4 1.679 HOXA1 2.714 FOSL1 2.090 C1QTNF6 1.919 ATM 1.780 TMEM182 1.678 GPBAR1 2.712 PRRT1 2.090 RASGEF1B 1.919 NTSR2 1.780 LPAR5 1.678 SOX7 2.701 KIF5B 2.084 MR1 1.919 FOXO4 1.779 KCNJ3 1.676 LCE2D 2.691 NUMB 2.078 KIAA0528 1.918 XPO5 1.778 ARHGAP10 1.676 SCNN1D 2.688 IGLV4-3 2.077 C3orf63 1.918 NOP2 1.778 SMARCC1 1.675 PSG5 2.683 TAF1C 2.075 RHPN1 1.914 MATN4 1.776 TTC17 1.675 FEM1C 2.669 BFAR 2.075 CD93 1.913 DCI 1.776 ZNF286A 1.675 PRB3 2.649 FAM116B 2.074 GPR52 1.913 DAAM2 1.776 SLC38A3 1.675 TPTE2 2.638 C6orf146 2.074 AP3M2 1.906 KIAA1683 1.775 EPS8L2 1.675 ANKS1B 2.627 NARG1L 2.071 HOOK1 1.905 TAX1BP1 1.774 MGAT1 1.673 KIAA1383 2.613 STXBP1 2.068 AKAP3 1.905 FAM171A1 1.773 TRPM8 1.672 CORO1B 2.607 KLHDC7B 2.067 LPAL2 1.903 OTC 1.773 ZNF658 1.669 PSG3 2.602 CYLC1 2.061 PNPLA4 1.901 FSIP1 1.771 ACE2 1.666 SNRPB2 2.601 C11orf75 2.060 BCL2L15 1.898 PDZD3 1.770 ALG8 1.666 B3GNT5 2.597 XPO1 2.060 EPB41L5 1.895 C4orf29 1.768 C16orf74 1.665 NBN 2.590 FBXW10 2.058 PUS3 1.894 OTUD4 1.767 EIF2B1 1.665 TCL6 2.584 C19orf42 2.054 ZP2 1.894 TRIM49 1.767 POTEB 1.665 GNG11 2.582 MLC1 2.051 CTBP2 1.891 AGTPBP1 1.766 GDA 1.663 RPS16 2.572 SPOCD1 2.049 LARP4 1.889 CAPN7 1.765 RING1 1.661 BRUNOL5 2.531 KLC3 2.048 AKR1C2 1.882 FDFT1 1.764 SPATA7 1.660 ODF3 2.530 IRAK3 2.047 DCP1B 1.879 SLC12A6 1.764 TAS2R42 1.659 TMSL1 2.529 MXRA7 2.045 REV1 1.876 RAPGEF6 1.763 BCL6 1.659 UPP2 2.516 EXOSC10 2.044 ZFP28 1.874 SRGN 1.760 C3orf64 1.658 IL13 2.503 KLF4 2.038 AGFG2 1.874 TMEM223 1.760 FLAD1 1.658 HLA-C 2.493 C15orf21 2.037 CILP 1.874 TCEB3B 1.759 ZNF596 1.657 CTCFL 2.487 C12orf64 2.034 SLC10A3 1.874 ZNF711 1.759 OAS2 1.656 SLC43A2 2.473 KBTBD10 2.033 AAAS 1.873 TJP1 1.759 A4GNT 1.654 ZYX 2.465 ADAMTS1 2.029 RMND5A 1.873 C12orf26 1.757 TRPC4AP 1.654 ZNF614 2.447 SPATA2 2.027 COMP 1.869 ZNF621 1.757 KIF20A 1.654 HLA-C 2.434 COX6A2 2.027 PTCHD3 1.863 TKTL1 1.754 ST3GAL6 1.654 C4orf21 2.431 RICTOR 2.025 TBCKL 1.863 SMEK1 1.752 KPNA6 1.653 SH3D19 2.427 SAMD7 2.024 SPERT 1.862 CD163L1 1.749 CWF19L2 1.653 GRM2 2.418 HSFY1 2.019 MEF2C 1.862 SP110 1.748 UNG 1.652 MCM8 2.412 EEF1D 2.018 LIG3 1.862 KIAA1012 1.748 MORN1 1.652 KIAA1958 2.410 ZFYVE20 2.017 ZHX3 1.861 CCPG1 1.745 PLK4 1.651 UBAP1 2.394 HORMAD1 2.017 PJA2 1.860 GYG1 1.744 CENPL 1.650

(Table 1, continued)

MDM2 2.389 AP2S1 2.017 EFTUD1 1.859 STAU2 1.743 BOLA2 1.650 GCM1 2.388 CNTN2 2.016 POLDIP2 1.853 COL20A1 1.741 C3orf23 1.650 CCNL2 2.370 CLDN14 2.016 ELK3 1.851 C1orf2 1.740 TMEM117 1.650 EEF2 2.366 MTO1 2.010 THPO 1.849 DBR1 1.738 ADAM9 1.650 HSD17B12 2.364 SCUBE3 2.008 PRB1 1.848 HHIPL1 1.738 FAM3C 1.650 RNF145 2.358 LY96 2.007 SOX14 1.846 FLJ12529 1.737 REST 2.356 PDGFRB 2.006 CDC42EP2 1.844 DOCK4 1.737 LOC389458 2.353 DHX29 2.005 ASAP3 1.844 CACNA2D1 1.736 ZNF253 2.347 HLCS 2.005 BPESC1 1.840 CCDC81 1.736 PREPL 2.346 EXO1 2.003 ZNF433 1.839 MEPE 1.734 TFCP2 2.339 FRMD8 2.003 COLEC12 1.838 TMPRSS5 1.733 SLC38A7 2.322 UGT2B4 2.000 SRRM1 1.837 ITPKB 1.731 PSG2 2.321 TBC1D5 1.999 TRPV2 1.834 NLGN3 1.731 RGP1 2.313 TGFBR3 1.997 WDR17 1.834 MCM8 1.730 SEMA3B 2.302 C12orf48 1.997 CCDC46 1.834 SLC39A12 1.729 HMMR 2.300 NULL 1.996 CTC1 1.831 BCKDHB 1.728 ZNF350 2.293 YIPF2 1.996 LGR5 1.827 PDCD6IP 1.728 SPATA12 2.280 QARS 1.989 SLC25A40 1.827 ZFYVE16 1.726 NELL1 2.270 SYNE2 1.989 ALG9 1.825 KLB 1.725 ADAD2 2.268 IFNA10 1.988 ETFDH 1.824 IL1RN 1.722 CC2D2B 2.264 MCM8 1.987 NID2 1.822 DYNC2LI1 1.722 TMEM54 2.262 MARCKSL1 1.987 RASGRP1 1.821 INTS5 1.721 ARHGEF3 2.259 MGC12966 1.985 RNPEPL1 1.820 GPATCH8 1.719 GHITM 2.257 SGK1 1.981 THYN1 1.819 TMSB15B 1.718 LAT 2.253 TYRO3 1.981 ZNF30 1.818 LOC494150 1.718 DKFZp761E198 2.242 PHIP 1.981 OTOA 1.817 VPS41 1.717 ZNF843 2.241 SLC41A3 1.975 LENEP 1.817 CLCN1 1.715 PDE1B 2.240 FADS1 1.975 TRIM49 1.812 NPFFR2 1.715 HAS3 2.234 IFNA13 1.974 FAF2 1.811 RASL10A 1.713 PGLS 2.233 NCF4 1.966 TRDN 1.811 IGHG1 1.711 PPM1D 2.232 C13orf18 1.964 ZNF672 1.811 SYK 1.709 PPM1D 2.232 ITGA5 1.963 S1PR4 1.811 ZNF500 1.709 C20orf114 2.224 RSPH3 1.959 FAM55C 1.810 CLEC14A 1.709 DYRK1B 2.223 NULL 1.956 ARMC3 1.810 PIAS4 1.707 TRMT2B 2.211 AZGP1 1.955 C12orf67 1.809 TEX13A 1.707 KU-MEL-3 2.204 BBX 1.955 KLHL24 1.807 KRTAP4-7 1.707 WHSC2 2.204 ZBTB11 1.953 MLF1IP 1.807 FAM26D 1.705 PTMS 2.201 NCAPH 1.952 MAGT1 1.806 RAD51AP1 1.705 STK3 2.198 ATP6AP2 1.951 BCL11A 1.806 NOP2 1.705 PLOD2 2.195 TMCC2 1.950 MAP3K6 1.805 SLC41A3 1.703 FRMD8 2.179 C22orf39 1.949 NULL 1.805 ZCCHC11 1.702 TPP2 2.167 HYAL1 1.949 HTR5A 1.804 GCNT3 1.701 SPATA12 2.166 IL22RA1 1.948 SETD6 1.803 RBP3 1.701 SOX15 2.164 NUDT10 1.945 PIR 1.803 SCAMP3 1.699 STAT6 2.163 RFC1 1.945 DUOXA1 1.803 HDAC10 1.699 ARHGAP5 2.159 FBXO4 1.944 MUSK 1.802 DENND1C 1.699

Table 2: candidate genes scored from E6 BJ senescence screen

Z Symbol Z Score Symbol Z Score Symbol Z Score Symbol Score Symbol Z Score AQP10 4.496 KAT2A 2.210 LRRTM4 1.954 SPATA3 1.799 KLHL2 1.698 FN3KRP 3.979 MGC16075 2.208 SV2B 1.952 ARHGAP25 1.799 ZPBP 1.698 CD200R1 3.976 #N/A 2.205 CYP3A4 1.951 TRIM38 1.799 PDS5B 1.697 RGS16 3.945 #N/A 2.203 COLEC12 1.951 CCDC81 1.798 POLDIP2 1.697 C4orf7 3.676 RBM15B 2.202 ACCS 1.950 PLOD2 1.798 C3orf33 1.697 LOC222699 3.564 SFMBT1 2.199 ATXN1 1.950 NULL 1.795 SPATA2 1.697 HNRNPA0 3.493 GRM2 2.192 GINS2 1.949 ALOXE3 1.795 HBG1 1.696 CTAGE1 3.380 C19orf34 2.187 C20orf114 1.949 HIST1H4I 1.795 ARG1 1.696 NBN 3.375 PLD4 2.182 SYNCRIP 1.946 RHPN1 1.794 STRN3 1.695 C2orf47 3.312 FAM127B 2.178 HBG1 1.941 AKAP3 1.793 TRPM8 1.695 NMNAT2 3.131 SLC43A3 2.175 C14orf159 1.940 PTMS 1.793 TUBA1B 1.694 PSG5 3.113 EXD1 2.174 NRG1 1.939 GCSH 1.792 FLJ12529 1.692 WFDC11 3.101 CC2D2B 2.173 TMEM192 1.937 CACNA2D4 1.789 FLRT3 1.691 CCDC85C 3.083 PGDS 2.164 HOOK1 1.937 CCPG1 1.789 GAB1 1.691 ATG12 3.017 NBR1 2.163 FABP2 1.933 C10orf65 1.788 PASK 1.691 KIAA1383 3.016 TNFRSF10D 2.160 ADAM9 1.933 ZHX1 1.787 TROAP 1.691 NOV 3.005 ANKRD44 2.160 EFCAB4A 1.931 RP2 1.787 ZNF646 1.687 C4orf46 2.981 CCDC94 2.159 LRRC8C 1.929 NRG1 1.784 C7orf31 1.687 ZNF253 2.978 CTHRC1 2.152 LY96 1.929 SAPS3 1.784 C17orf53 1.686 IGKC 2.959 C4orf21 2.152 GATA3 1.928 C12orf48 1.784 TM7SF3 1.685 CD84 2.914 FBXO4 2.151 SSX2IP 1.926 MFSD1 1.781 LOC57228 1.683 LOC339535 2.897 AEN 2.139 ZNF471 1.923 NIPBL 1.780 ZNF695 1.682 PIM2 2.895 KLHDC7B 2.138 CD93 1.922 PTPRS 1.779 G6PC 1.682 HAMP 2.870 CORO1B 2.129 RICTOR 1.915 NUMA1 1.775 ELP2 1.680 HLA-C 2.779 TYRO3 2.126 XKR9 1.915 GYPE 1.774 ZCCHC11 1.680 NULL 2.773 ZNF614 2.118 DCP1B 1.914 KIAA1012 1.773 FGF1 1.679 PSG2 2.768 WDR17 2.116 SCNN1D 1.914 BBS9 1.772 KCND3 1.677 FRMD8 2.760 CMTM6 2.112 IFIH1 1.913 C10orf111 1.770 CDH17 1.676 FGF12 2.738 FRMD8 2.103 PLA2G4A 1.913 GBA 1.770 COL24A1 1.676 PDCD1LG2 2.734 ZFP28 2.099 MXRA7 1.911 RBM41 1.769 RNF166 1.675 KRTAP10-1 2.726 FAM129A 2.096 TLE2 1.911 LPAR3 1.768 SESTD1 1.674 RHOA 2.719 ISG20 2.096 ADORA2A 1.910 C4orf29 1.766 ZNF592 1.673 RHOA 2.719 C9orf116 2.094 GPBAR1 1.909 ZNF518A 1.764 PGK1 1.673 SCMH1 2.690 #N/A 2.093 SRGN 1.907 TEN1 1.762 NID2 1.673 NEGR1 2.662 KIAA1644 2.093 SCUBE3 1.906 ANKS1B 1.761 DENND1C 1.671 SPG7 2.662 ERI1 2.091 FOXO4 1.905 HORMAD1 1.761 P4HA2 1.671 GNG11 2.660 TPP2 2.091 USP45 1.904 TBL2 1.760 CWF19L1 1.670 DDAH2 2.656 TIMP2 2.086 FAF2 1.904 SAMD7 1.760 LSMD1 1.670 E2F2 2.644 SLURP1 2.081 LPAR5 1.901 UQCRC1 1.759 CA11 1.669 KRTAP10-7 2.638 GDPD2 2.080 CD300LG 1.900 MED28 1.758 LOC400713 1.669 KRTAP10-4 2.638 ZNF823 2.080 SLC41A3 1.898 GPR64 1.757 MGC3771 1.668 YIF1B 2.625 FAM160B1 2.072 C19orf42 1.897 CCDC67 1.756 EFCAB7 1.668 KU-MEL-3 2.592 PGM2 2.070 CNTN2 1.897 ORMDL3 1.753 AFAP1L1 1.668 MPST 2.573 CDH18 2.070 NLK 1.897 PTPN11 1.752 HS1BP3 1.668 RRS1 2.551 ZNF433 2.069 TRBV5-4 1.895 C12orf66 1.752 STK10 1.667 CDK2 2.549 ZNF658 2.069 DPY30 1.891 UCN 1.751 AEN 1.667 C11orf75 2.541 CD300E 2.068 LYZL6 1.891 HDAC10 1.749 ABCA9 1.666 PRUNE 2.494 TTLL10 2.067 MESP2 1.890 MYBPC2 1.749 PRRT1 1.665 KRTAP3-1 2.492 COLEC10 2.061 RD3 1.889 REST 1.747 CD3EAP 1.661 PSG3 2.487 CREBZF 2.059 FOSL1 1.889 PNPLA8 1.746 IGHM 1.660 MLC1 2.484 SMEK1 2.059 KCNJ12 1.887 ZNF350 1.745 RASGEF1B 1.660 NULL 2.465 PKP3 2.057 NULL 1.887 KCNN4 1.743 ARS2 1.659 CTAGE5 2.457 MTHFD1L 2.057 STK31 1.884 PRO2012 1.742 SHE 1.655

(Table 2, continued)

CCL11 2.457 DSEL 2.054 HMMR 1.882 AGTPBP1 1.741 SMARCC1 1.654 KIAA1958 2.455 CTC1 2.050 COL8A2 1.881 FHL3 1.741 ITPKB 1.653 NUDT2 2.455 NPFFR2 2.046 NCAPH 1.877 VWA2 1.739 KLHL24 1.653 CNBD1 2.453 ITGB6 2.044 CLCN1 1.876 DGCR14 1.738 PER1 1.652 KRTCAP2 2.447 RAB17 2.043 SEC14L1 1.875 KIAA0195 1.737 HIST1H4E 1.651 IFNA13 2.433 S1PR4 2.039 TRAK1 1.871 39874.000 1.737 LOC653115 1.651 LOC147804 2.432 ATP6AP2 2.034 ZNF846 1.870 RAI1 1.736 LENG1 2.423 NULL 2.034 GNPTAB 1.870 CD163L1 1.736 RAB17 2.418 RPS16 2.033 OCC-1 1.870 CDH2 1.735 CD320 2.409 SEMA3B 2.032 PDK4 1.868 ZBTB11 1.734 LCE1B 2.407 SLC27A6 2.024 ZNF30 1.868 BDKRB1 1.733 TMSL1 2.399 CCL8 2.022 AKT3 1.866 KLC4 1.733 ARL15 2.395 C5orf36 2.022 TSSK4 1.866 ZNF385A 1.732 C7orf55 2.394 CTAGE5 2.021 NULL 1.864 ZNRD1 1.732 KBTBD10 2.382 DSCR4 2.020 #N/A 1.862 ACE2 1.729 F11R 2.374 HYAL1 2.017 CD274 1.862 IGKV1-5 1.729 #N/A 2.362 PCDH10 2.014 USP1 1.859 BUB1B 1.729 LCE2D 2.357 MATN4 2.013 CAPN7 1.857 PUS3 1.727 40061.000 2.349 SPARC 2.011 DISP1 1.856 LENEP 1.726 SLC43A2 2.347 SLC41A3 2.009 CLDN12 1.853 ZNF606 1.724 IGKC 2.346 CYLC1 2.008 TGM3 1.851 ARHGAP5 1.722 TMEM218 2.342 TMSB15B 2.007 ZNF621 1.848 NLRP7 1.722 MDM2 2.336 BCKDHB 2.007 TRPV2 1.848 LRRC2 1.721 ATP6V1B1 2.333 C12orf67 2.007 NULL 1.846 MCM8 1.720 ATG12 2.331 SYT14 2.003 EEF1D 1.835 UNC45A 1.719 CWF19L2 2.321 ZDHHC17 2.000 C21orf121 1.834 GPR52 1.718 KRTCAP2 2.319 ADAD1 1.992 RPS15 1.833 CD68 1.718 RAB10 2.308 CCNA1 1.987 WDR46 1.828 ITFG1 1.718 HLA-C 2.308 TMEM186 1.984 SLC24A1 1.827 C10orf71 1.718 FAM3A 2.307 CCDC46 1.982 C7orf42 1.825 C14orf101 1.717 CIR 2.297 AQP8 1.981 MMRN1 1.823 C22orf28 1.716 MMP12 2.296 ZNF573 1.980 PLEKHA8 1.822 ATXN7L3 1.716 TWIST2 2.290 LOC389458 1.979 HSPC159 1.820 BRF1 1.715 STXBP1 2.278 ZNF672 1.977 RASSF9 1.818 C9orf61 1.715 CGB5 2.278 MPP4 1.976 RFC1 1.818 NULL 1.714 ISG20 2.278 MECP2 1.976 SLC12A6 1.816 NELL1 1.714 GPR65 2.271 CAPRIN1 1.974 RLF 1.816 PHACTR4 1.712 HSD17B12 2.262 PREPL 1.965 MCM8 1.815 SMAD3 1.711 BTC 2.259 LSM14B 1.965 PLEKHA8 1.815 CCDC26 1.709 NARG1L 2.255 EEF2 1.962 SCMH1 1.813 TJP1 1.709 ADRM1 2.248 KBTBD6 1.961 TGFBR3 1.812 GOSR2 1.708 LDHB 2.247 SDS 1.961 KLC3 1.811 RAD23B 1.708 PRKAA2 2.235 OSBP2 1.960 KRT13 1.810 CMTM2 1.708 NULL 2.228 LRRTM2 1.959 LSMD1 1.809 SULT2B1 1.704 C6orf146 2.227 CCDC130 1.958 ZNF599 1.808 KCNJ3 1.704 UBA5 2.219 GPATCH8 1.957 ZFYVE16 1.808 DAPK2 1.702 ARHGAP12 2.219 IQSEC1 1.957 OSMR 1.799 MTHFD1 1.701

3-3. The Identification of known p53 and Rb pathway genes To better understand the screen data, we performed a bioinformatics analysis with

Ingenuity pathway analysis tools (http://www.ingenuity.com/products/ipa). For the E7 screen hits, several networks containing cancer genes were identified. One of the top networks of the E7 candidates was composed of cancer, hematological disease and immunological disease genes (Figure 5). Interestingly this network contains multiple p53 interacting proteins, including MDM2, PPM1D and ATM, indicating the enrichment of the p53 pathway regulators and/or effectors scored from the screen. Note that the ORF we scored as ATM was actually a truncation of the ATM ORF that is also likely to act in a dominant-negative manner. Along these lines NBN, which is required for ATM activation and connects ATM to the MRE11-RAD50 complex also scored. It, too, is likely to act as a dominant negative manner when overproduced as it interacts independently with ATM and MRE11-RAD50. In the E6 screen analysis with the same Z score threshold of 1.65 (log2 enrichment >1.19, n=340 genes), the top network contained genes involved in connective tissue disorders, inflammatory disease, skeletal and muscular disorders (Figure 6). In this network we observed E2F2 and several other RB interacting proteins, including CDK2, Cyclin A and MDM2. MDM2 is known to promote p53 degradation, however it has also been shown to promote RB degradation in a p53- independent manner [134] [135]. Thus, our screen system had identified known p53 and

Rb regulators in the senescence screen.

Figure 5. Network analysis identifies multiple p53 related genes from the E7 screen candidates.

Figure 5. A network composed of connective tissue disorders, inflammatory disease, skeletal and muscular disorders genes with multiple p53 functionally-interacting genes identified from the E7 screen. Genes shown in red are the ones scored from the screen.

Figure 6. Network analysis identifies multiple Rb related genes from the E6 screen candidates.

Figure 6. A network composed of cancer, hematological disease and immunological disease genes containing multiple Rb functionally-interacting genes identified from the E6 screen. Genes shown in red are the ones scored from the screen.

Next we explored the disease and biological functions related to the screen candidates using the IPA suite of analysis tools. Cancer was the top scoring disease category in the disease analysis along with several specific cancer subtypes, suggesting that some of the additional novel candidate genes from the screen might represent new cancer drivers (Figure 7). For the biological function analysis we identified annotations related to cell proliferation, cell cycle progression, DNA repair and replication, suggesting our screen could identify genes with functional relevance to cellular senescence (Figure 7). We also performed Gene Ontology (GO) term analysis search based on gene lists ranked by screen enrichment score with Gorilla tools (http://cbl- gorilla.cs.technion.ac.il/). After applying a FDR-q value threshold of 0.05, we identified several significantly enriched GO terms only from the E7 screen dataset but not the E6 screen dataset, including several GO terms related to cellular senescence (Figure 7). The genes associated with the senescence GO terms on top of the screen list were TP53,

CDK6, ATM and CTC1. The identification of CTC1 was particularly interesting since this gene is involved in telomere maintenance, as we discuss below.

Figure 7. Disease, Biofunction and GO term analysis of screen candidates.

Figure 7. A) Disease and bio-function analysis of E6 screen candidates with Z score>1.65. B) Disease and bio-function analysis of E7 screen candidates with Z score>1.65. C) GO term analysis of E7 screen candidates. The red line indicates a –log (q value) threshold of 1.3.

3-4. Genes scoring in both the E6 and E7 branches. In principle, in addition to genes that affect either the p53 or Rb branches, we should be able to identify a third class of genes that affect both branches. For example we could detect DDR genes that affect sensing of the short telomere signal, genes that reduce telomeric erosion, or genes whose overproduction induce telomerase activity. One example of a gene that was we identified in both branches and is known to affect both the p53 and RB branches is the oncogene MDM2. MDM2 is an E3 ubiquitin ligase that is capable of destroying both p53 and RB proteins. We identified a total of 172 genes that scored in both E6 and E7 screens (Z score>1.65, see Figure 8 and Table 3). Several additional genes of this class of genes are described below.

Figure 8. Genes scored in both E6 and E7 screens.

Figure 8. A) Genes scored in both screens shown in orange. Genes scored only in the E7 screen shown in green. Genes scored only in the E6 screen shown in red. B) The CST complex component CTC1 and TEN1, and aging related genes RFC1, FOXO4, REST were highlighted among the genes scored in both screens.

Table 3: candidate genes scored from both E6 and E7 screens

Symbol Symbol Symbol Symbol Symbol Symbol ACE2 CD93 GRM2 LENEP PSG3 TMSB15B ADAM9 CDH18 HDAC10 LOC389458 PSG5 TMSL1 ADORA2A CLCN1 HLA-C LPAR5 PTMS TPP2 AGTPBP1 CMTM6 HLA-C LRRTM4 PTPN11 TRPV2 AKAP3 CNTN2 HMMR LSM14B PUS3 TYRO3 ANKS1B COLEC12 HOOK1 LY96 RASGEF1B WDR17 AQP10 CORO1B HORMAD1 MATN4 REST XKR9 ARHGAP12 CTC1 HSD17B12 MCM8 RFC1 ZBTB11 ARHGAP5 CWF19L1 HYAL1 MDM2 RHPN1 ZCCHC11 ATP6AP2 CWF19L2 IFNA1 MESP2 RICTOR ZFP28 ATP6V1B1 CYLC1 IGKC MLC1 RPS16 ZFYVE16 BBX DCP1B IGKC MXRA7 S1PR4 ZNF253 BCKDHB DENND1C ITGB6 NARG1L SAMD7 ZNF30 BDKRB1 EEF1D ITPKB NBN SCNN1D ZNF350 C11orf75 EEF2 KBTBD10 NCAPH SCUBE3 ZNF433 C12orf48 ELP2 KCNJ12 NELL1 SEMA3B ZNF573 C12orf67 FAF2 KCNJ3 NID2 SLC12A6 ZNF614 C19orf42 FAM160B1 KIAA1012 NIPBL SLC41A3 ZNF621 C20orf114 FBXO4 KIAA1383 NPFFR2 SLC41A3 ZNF658 C2orf12 FLJ12529 KIAA1958 NULL SLC43A2 ZNF672 C4orf21 FOSL1 KLC3 NULL SMARCC1 C4orf29 FOXO4 KLHDC7B NULL SMEK1 C6orf146 FRMD8 KLHL24 OSMR SPATA2 CAPN7 FRMD8 KPNA6 PCDH10 SRGN CC2D2B GATA3 KRT13 PLA2G4A STXBP1 CCDC46 GDPD2 KRTAP10-1 PLOD2 SYNCRIP CCDC67 GNG11 KRTAP10-7 POLDIP2 SYT14 CCDC81 GPATCH8 KU-MEL-3 PREPL TEN1 CCPG1 GPBAR1 LCE1B PRRT1 TGFBR3 CD163L1 GPR52 LCE2D PSG2 TJP1

3-5. Telomere maintenance: The CST complex.

As telomere shortening initiates replicative senescence, we would expect to identify genes that reduce telomere shortening, enhance its end protection or enhance telomerase activity itself. One candidate that scored strongly in both the E6 and E7 screens was CTC1. CTC1 encodes a protein component of the CST (CTC1-STN1-TEN1) complex. This complex binds to the 3’ telomeric overhang to promote telomere C-strand fill-in synthesis by DNA polymerase , offsetting telomere attrition caused by exonucleases-mediated end processing independent of telomerase activity [136]. If excessive exonuclease resection is not corrected by efficient fill-in resynthesis prior to the

next round of DNA replication, then one sister chromosome’s telomere willbecome substantially shorter and could engage the DDR to induce senescence. Stimulating resynthesis reduces the rate of shortening and therefore the CST complex can play a positive role in reducing the telomere attrition rate and avoiding senescence.

Interestingly, we found another component of the CST complex, TEN1 (ORFeome plate

ID 31007@E10), also scored in both E6 and E7 screens (Figure 8). A third member,

STN1, was also included in the ORFeome library, however it did not score from either screens. The STN1 ORF in the library contains two point mutations (T151A, S248C). It is unclear if these mutations affect the function of this ORF. After all the identification of

2 out of 3 members of the CST complex indicates our screening method can identify genes related to telomere maintenance.

3-6. Additional factors linked to senescence or aging identified in both screens.

We examined the literature for genes on our list that have previous links to senescence or aging in physiological or pathological contexts. We found several such genes (Figure 8). REST, a transcriptional repressor that has been suggested to be both a tumor suppressor and an oncogene in different contexts (reviewed in [137]), was found in both E6 and E7 screens. Interestingly, a recent study showed that REST displays neuroprotective functions and it can ameliorate aging-related neurodegenerative disease

[138]. In addition, RFC1, which encodes a replication factor C subunit involved in DNA replication and repair, also scored in both the E6 and E7 screens. It has been reported that

RFC1 is cleaved and inactivated in the Hutchinson Gilford Progeria Syndrome (HGPS)

[139], and this inactivation might contribute to the replicative cell arrest in HGPS. We also identified FOXO4, a member of the FOXO transcription factor family whose

orthologues have been shown to regulate life span and age related diseases in model organisms [140] [141]. Although it is not clear if FOXO4 can affect age-related physiology in mammals, studies have suggested that FOXO4 is involved in protection against oxidative stress at the cellular level [142] [143], which may link to cellular aging.

Together, our findings strongly suggest our screen data contain genes with functional relevance to telomere maintenance and other pathways involved in pathological/physiological aging.

IV. Discussion

Our screen effort provides a large data set of genes whose ectopic expression lead to increased replicative lifespan in sensitized senescence backgrounds. Our E7 screen identified numerous regulators of the p53 pathway, including p53, ATM, MDM2, WIP1, along with many genes that connect with p53 as regulators or effectors in the IPA network analysis. Our E6 screen also identified multiple genes related to the Rb tumor suppressor regulation. We assume some of our candidate may be oncogenes that help cells bypass senescence and promote oncogenesis in vivo. The pathway analysis confirmed cancer is the top disease category significantly associated with our dataset.

Consistently, biofunctional analysis revealed relevant cellular function such as cell cycle,

DNA damage signaling and repair significantly associate with our screen dataset. This evidence strongly suggests our screen is a largely successful effort.

It is reasonable to conclude our screen data set contains new genes in important senescence regulatory mechanisms in addition to the dozens of genes known to be involved in p53 or Rb regulation. The subset of genes scored in both E6 and E7 screens are particularly interesting, for they may represent novel pathways affecting senescence

independent of p53 or Rb. Particularly, the identification of TEN1 and CTC1 of the CST complex reveals interesting points: the protein level of CST complex components may be a rate-limiting factor in determining telomere shortening rate, and a reduction of telomere attrition rate may delay senescence. Furthermore, we identified a handful of genes with potential linkage to physiological aging by various means. Premature aging syndromes such as HGPS and Werner syndrome are known to exhibit senescence markers, and

RFC1 has been observed to be inactivated in HGPS. However, direct functional assay has been absent to justify its potential role of causality in premature aging. RFC1 is part of the clamp loader Replication Factor C (RFC) complex that is required for PCNA loading and DNA replication. Inactivation of RFC1 may compromise DNA replication, thus contributing to cell cycle arrest in premature aging and possibly senescence. On the other hand, RFC1 shares structural similarity to Rad17, which forms a complex with the components of the RFC complex (RFC2-5) to recruit the Rad1-Rad9-Hus1 (9-1-1) complex and facilitate ATR activation on single strand DNA break[144]. This raises the interesting possibility that RFC1 expression may compete with Rad17 for complex formation with RFC2-5, and thus potentially attenuate the signaling of the ATR DNA damage response pathway. The fact that RFC1 scored in our functional senescence screen indicates that it may represent a new functionally relevant factor in senescence and premature aging. REST has been shown to confer neuroprotective functions during aging, and senescence is thought to also contribute to aging-related diseases. Thus, it would be interesting to investigate if senescence affects the molecular pathways buffered by REST during aging in general. To sum up, we believe that our screen data may lead to important future insights with regard to senescence, tumorigenesis and organismal aging.

Compared with loss-of-function screens, our gain-of-function screen system eliminates off-target effect. A typical shRNA screen usually requires at least 3 to 4 hairpins per gene for phenotype caused by sufficient knock-down to manifest, while in the ORFeome based screen one full-length ORF is enough for testing one transcript. It is worth mentioning that complications such as dominant negative ORF caused by point mutations do arise in ORFeome screens. Because the mutant p53 scored strongly in the screen, precautions must be taken for properly interpreting the screen results. It is also true that the build quality of the ORFeome library may be compromised by truncated or mutated ORFs, and they may lead to false negative results in the screen. Additionally, we also observed ORFs with incorrect annotation in the library. For example, MDM2 and

TEN1 scored in the screen were not named correctly in the ORFeome information database. We have summarized the mis-annotation, ORF mutation and truncation that we encountered in chapter 2 and 3 in table 5. As newer versions of the human ORFeome library with higher build quality become available, the screen performance can be further improved. After all, ORFeome based screens and shRNA based screens are complementary approaches to address the same question, and they can be designed to target different classes of genes involved in a biological process with the same assay.

Although evidence positively justifies our screen effort, there are several concerns we think could be addressed to improve the screen outcome. While the identification of the mutant p53 from our screen justifies our screen scheme for identifying relevant genes of key senescence pathways, the fact that the p53 PCR product is so strong in the pool of amplified ORFeome material in end sample suggests this ORF has an enormous enrichment against other genes in the library. This raises the possibility that it may

overwhelm the population with a bottlenecking effect. In other words, the representation of the bulk of the library may be lost and each replicate in the screen may become divergent due to the selective pressure and bottlenecking effect. To mitigate this problem we chose to use the average Z score of each replicate to set up a threshold, so that if a gene is lost in one replicate, it may still be compensated if it scores well in other replicates. We think that a more stringent threshold based on P value derived from consistent enrichment may limit the range of genes found in the screen. To compensate for the reduced stringency in our analysis, we employed a secondary screening step to select best genes with improved assay performance, which we will discuss in detail in the next chapter.

Because we used middle-aged E6 or E7 BJ cells as screen platforms, we cannot rule out the possibility that some of the genes scored because of increased proliferation rather than delayed senescence. While it is unknown if any intrinsic ties exist between proliferation and replicative lifespan, and perhaps faster proliferation results in sooner senescence, assuming the “telomere clock” is unbiased, it is possible the interpretation of senescence bypass phenotype will be affected by the proliferation argument. One way to address this issue is to screen with older cells, immediately before the onset of senescence. While this might present some technical challenges such as reduced penetrance for senescence bypass phenotype to fully develop and difficulty for viral infection, it can reduce the complication of proliferation. On the other hand, genes scored in the senescence screen can be individually tested for proliferation effect to address this issue, or we could even perform a separate proliferation screen in young cells to ascertain which genes affect proliferation.

We designed a screen scheme with sensitized senescence background by expressing either HPV E6 or E7. We were able to identify known regulators of the p53 and the Rb pathway, confirming the efficacy of the screen design. Although it is generally accepted the HPV E7 and E6 can abolish the Rb pathway and the p53 pathway respectively in senescence, E7 and E6 have also been shown to be involved in tumorigenesis beyond the E7-Rb and E6-p53 scenario. For example, E6 has been shown to activate hTERT transcription in mammary epithelial cells and keratinocytes [145], and

E7 has been shown to suppress the CDK inhibitor function of p21 in keratinocytes [146].

Therefore, future sensitized screen design may benefit from the use of alternative approaches for inactivating p53 and Rb, such as the short-hairpin RNA mediated RNA interference or CRISPR mediated gene knock-out.

We chose to perform the sensitized screen in BJ fibroblast cells. While we found both p53 and Rb pathway regulators in the screen, we observed more genes with direct links to the p53 pathway than those involved in the Rb pathway by network analysis.

While it is generally true that p53 plays a major role in initiating senescence in many cell lines, we reason that there are fewer known Rb regulators existing in the cellular senescence machinery. How p16 is regulated remains largely unclear, thus we found fewer genes related to Rb than p53. Although we confirmed that our E6 background can be used for screening p16-Rb pathway regulators, it is possible that other cell lines may suit this purpose better. One future direction could be to repeat the screen in cell lines with a strong p16 activation in senescence, such as IMR90 and WI-30 fibroblasts.

We decided to rely primarily on the 25-mer microarray for quantifying screen data, partially because this platform gave a much higher enrichment score and ranking of

the mutant p53 ORF than the 60-mer microarray. While this platform shows p53 is enriched for 30 fold, we think there is still a significant chance the actual p53 enrichment is much higher, given the strong abundance of the p53 product in the PCR product. This raises a concern that a microarray-based quantification approach may have a limited dynamic range than other platforms such as deep sequencing. Because the margin between the top genes was very small, the score genes appear to cluster together. We faced a practical problem of choosing genes for individual validation, so we designed a new, sequencing based approach to re-screen top hits to increase the screening assay resolution and facilitate screen follow-up. The detail of the secondary screen will be discussed in the next chapter.

V. Material and methods

5-1. Cell culture and general procedures.

Human diploid BJ fibroblasts were obtained from the ATCC and were maintained in 3%

O2 conditions. Cells were grown in DMEM-Glutamax (Invigtrogen) with 15% FBS, penicillin-strepomycin, and 0.1 mM nonessential amino acids (Invitrogen). E6/E7 expressing BJs were generated by infecting with retrovirus derived from MSCV-E6 and

MSCV-E7 vectors, which were gifts from the deLange Lab. Retrovirues and

Lentiviruses were packaged in 293T cells transfected with gene expression vectors of interest together with helper plasmids (for retrovirus: VSVg, Gag/pol; for Lentivirus:

VSVg, Gag/pol, TAT and Rev, see ref [147]). Viral infections were carried out in the presence of 4ug/mL Hexadimethrine bromide.

5-2. Growth curve analysis.

Cells were plated at a density between 1-1.5 x 106 per 10 cm tissue culture plate (127-

191*103/cm2) and fed every two to three days. Cells number was monitored using a coulter counter for calculating population doublings. Cells were continuously passaged until the growth curve became flat.

5-3. ORFeome based senescence bypass primary screen.

The general workflow of the senescence primary screen was described in Figure 1.

Approximately 8 x 106 infected cells were cultured in three replicates, with a representation of 500 copies per ORF in the library. Start and end sample cells were harvested with >1 x 107 cells per pellet. Pellets were lysed in 10 mM Tris pH8.0, 10 mM

EDTA, 0.5% SDS, 0.2 mg/ml Proteinase K at 55oC overnight. Genomic DNA was extracted with phaselock tubes with Phenol:Chloroform and Chloroform. The sample was then digested with 25 ug/mL RNaseA for overnight incubation at 37oC and extracted with phaselock tubes again as described above. DNA was precipitated with Ethanol and washed with 75% ethanol three times before being resuspended in H2O. The DNA was

PCR amplified with Takara hot start Taq polymerase (RR006B) with primers flanking the

ORF expression cassette (Primer 1:5’-GATCCCTACCGGTGATATCC-3’;; Primer 2:5’-

TAATAC GACTCACTATAGGGAGAGGCCCTCTAGTCGACCTAGC-3’). Purified

PCR products were used to generate cRNA probes with a T7 RNA polymerase kit

(MEGAscript, Ambion). After purification with Ambion Megaclear kits, the cRNA was further labeled with a ULS labeling kit (Kreatech: start sample- cy3, end sample - cy5).

Differentially labeled cRNA was hybridized to customized 25-mer or 60-mer microarray chips from Agilent. The microarrays were read with an Agilent scanner.

5-4. Primary screen data analysis.

We processed the probes by filtering out the low intensity probes, and an individual gene enrichment score was calculated based on average cy5/cy3 signal ratio across multiple probes of the same gene and normalized to the population median enrichment score. The

60-mer array contains many low intensity probes; after filtering, this platform only provided approximately 60% coverage compared to the 25-mer array data. We ranked genes based on average log2 Enrichment score across three biological replicates, and used the genes scored from the 25mer microarray with an arbitrary cut off of Z score >

1.65 for Ingenuity pathway network analysis.

Chapter III: A barcoded platform for re-screening candidates for senescence bypass Abstract

We identified multiple known regulators of the p53 pathway and the Rb pathway, as well as potential new regulators that affect these pathways and senescence. We also discussed several areas that could be addressed to enhance the performance on our screening effort to uncovering genes that can extend replicative lifespan. To further improve our screen effort, we constructed a collection of barcoded gene expression vectors, and used this collection for rescreening the top genes identified from each arm of the primary screen. 300 genes were reanalyzed for senescence bypass based on barcode sequencing, and this secondary screen out-performed the primary screen. We further validated a handful of genes individually scored from the secondary screen by growth curve analysis, including genes involved in the p53 pathway activation (USP15, KDM4D,

TP53I13) and potential new regulators (DLX2, DLX6).

I. Introduction

Due to the high complexity of the ORFeome library and the extended culturing process required to allow replicative senescence to occur naturally, we were concerned that the screen might be susceptible to the accumulation of random noise. Moreover, the strong enrichment of p53 mutant indicates the screen may be affected by the bottlenecking effect caused by this mutant ORF. We reasoned that in our primary screen result, as we already enriched a number of known regulators of p53 and other pathways controlling senescence, we could just re-examine this collection of scored genes with a more accurate assay. The primary screens were performed in a library with a total of

15,000 genes. We decided if we just took the top 1 per cent of the library for re-screening purposes, the smaller sublibrary would be much less susceptible to random noise and stochastic behavior since the complexity of the library significantly reduced. Furthermore, we could exclude the mutant p53 from the sublibrary construction to reduce the chance of bottlenecking effects.

Another issue we discussed is the limits with the microarray based quantification method. We realized certain arrays have limited dynamic range, and the probes are susceptible to signal saturation. The number of probes and probe quality, as well as cross- hybridization, could compromise the reliability of the microarray problem. The limitations are evident as the p53 mutant scored less strongly with the 60-mer array than the 25-mer array, and in general we observed a narrow distribution of enrichment scores.

One way to address this issue was to detect the enrichment by sequencing the ORFeome

PCR products rather than doing microarray hybridization. Moreover, we decided to add oligonucleotide barcodes to the ORFs to minimize the unpredictable bias caused by the

intrinsic differences of PCR efficiency amongst ORFs in the library. By creating a collection of ORFs labelled by DNA barcodes, we could strongly improve our screen performance, as shown below.

II. Results

2-1. Selection of genes for the secondary screen. We used genes that scored well in the primary screen for re-screening purposes.

We selected genes for constructing one sublibrary for E7 screen hits (log2 >1.75 N=85) and second for the E6 screen hits (log2 >1.75 N=67) with the 25-mer microarray hybridization-derived candidates. The relative high-throughput of the sequencing platform allowed us to include additional genes with log2 enrichment score above an arbitrary threshold (E7 screen: average log2 >1.59 with log2>1.95 in at least 2 replicates,

N=56; E6 screen: log2 >1.45, N=91; see materials and methods) in the sublibraries according to the 60-mer microarray hybridization results. Hence, we selected a total number of 141 genes for the E7 sublibrary screen and 158 genes for the E6 sublibrary.

We also selected a pool of 40 ORFs that did not score in the primary screens as control

ORFs in the secondary screen (for sublibrary and control pool contents, see material and methods).

2-2. Construction of barcoded sublibraries. We synthesized a pool of DNA oligos containing a barcode cassette flanked by common homology regions to the lentiviral vector that we used for the primary screen

(pHAGE-TreX). The barcode DNA pool was integrated into pHAGE-TreX gateway destination vector backbone, and the recombinant DNA was transformed into competent ccdbR E.Coli. The transformants were plated on LB agar plate with selective markers. We

collected individual clones, arrayed them in 96 well plate format and performed Sanger sequencing to identify the exact sequence of the barcode for a given destination vector.

Figure 9. Outline of the construction of barcoded ORF expression vectors for sublibrary screen.

Figure 9. A) A collection of 200 uniquely barcoded lentiviral gateway destination vector was generated by inserting unique barcode sequences (left). Barcoded destination vectors were recombined with entry clones by individual LR reaction to generate barcoded ORF expression vectors (right). B) A map of the barcoded destination vector. Note that the Barcode cassette was inserted upstream of the CMV-TetO promoter.

We generated a total of 200 lentiviral destination vectors, each containing a unique barcode (Figure 9). Next we performed LR reaction to link individual ORFs with a unique barcode. To constructing an E7 sublibrary, we performed 140 LR reaction in 96- well plate on a one-to-one basis, hence a given ORFeome entry clone recombines with one barcoded vector in a single well. Similarly, we used the barcoded destination vector collection to generate a barcoded E6 sublibrary of 158 genes. We barcoded 40 unscored

ORFs as negative controls. In addition, we included 4 barcoded empty vectors as well 4 barcoded GFP-expressing vectors as additional controls.

2-3. E6 and E7 Sublibrary screening

The barcoded sublibraries were introduced into pre-senescent, proliferating E6 or

E7 BJ cells, mixed with control ORF cells and spiked-in GFP+ control cells to repeat the senescence bypass assay. We repeated the screen assay and cultured the screen population until they slowed down proliferation and become senescent (Figure 10). We were able to observe a reduction of GFP+ cells from ~10% to 1% in the end population, suggesting this control was dropping out due to the competition from senescence- bypassing cells. We collected both start and end samples, and quantified the relative abundance of each ORF by barcode-sequencing (For screen result, see table 4).

Figure 10. Scheme of sublibrary screens.

Figure 10. Candidates picked from the primary screen hits were linked to individual barcodes and introduced to young E6 or E7 cells by lentiviral transduction. Puromycin-selected cells were pooled and passaged after they became senescent. Genomic DNA from the Start and End sample was used to amplify the barcodes linked to individual ORF. The relative abundance of each barcode was determined by next- gen sequencing.

Figure 11. Secondary screen results.

Figure 11. A) Box-Whisker plot of Log2 enrichment score of 40 control ORFs, E6 and E7 sublibraries. The E6 and E7 barcoded sublibraries displayed higher log2 enrichment than the control pool of 40 ORFs. The P value was calculated by two-tailed unpaired T-test. B) E7 sublibrary screen results plotted by Z score (left Y axis) and Log2 Enrichment score (right Y axis). MDM2 and PPM1D were strongly enriched in the screen. C) E6 sublibrary screen results plotted by Z score (left Y axis) and Log2 Enrichment score (right Y axis). CDK2 and E2F2 rescored in the sublibrary screen. RHOA, a known regulator of p16 and p21, strongly scored in the screen. Control ORFs shown in grey, ORFs with Z score<2 shown in blue, ORFs with Z score>2 shown in red.

2-4. Improved screen performance in the secondary screen.

Compared to the pool of 40 control ORFs, our candidate genes in the sublibraries achieved significantly higher log2 enrichment scores (Figure 11), indicating our sublibrary content, selected based on the primary screen results, outperformed controls in the assay. Note that we excluded the mutant p53 ORF during the sublibrary construction to avoid potential bottlenecking effects. Nevertheless, we identified multiple strong

negative regulators of p53 with robust enrichment scores. For example, the P53 signaling pathway is regulated by a negative feedback loop mediated by the phosphatase PPM1D and the E3 ubiquitin ligase MDM2. In the primary screen PPM1D and MDM2 both scored modestly. In the secondary screen we identified PPM1D and MDM2 again with greatly improved enrichment scores (Figure 11, PPM1D log2 enrichment >10, Z score =

6.4; MDM2 log2 enrichment >8, Z score = 5.2). Note that two variants of PPM1D ORFs with different barcodes scored with nearly identical enrichment scores, suggesting a high level of reproducibility. In the E6 sublibrary screen while both CDK2 and E2F2 were rescored, and interestingly RHOA, a GTPase gene involved in the p21/p27 and p16/p18/p19 family CDK inhibitor regulatory circuit [148], was identified as one of the highest scoring hits (Figure 11). Together, in our secondary screen, the barcoded sequencing platform allowed us to improve the screen performance for selecting genes affecting senescence.

Table 4: secondary screen results

E7 sublibray screen candidate ( Z score>2) E6 sublibray screen candidate ( Z score>2) log2 Z log2 Z ORF ID symbol Enrichment score ORF ID symbol Enrichment score 11062@C12 PSG3 14.654 9.152 31007@A09 PASK 9.695 9.061 51023@b04 USP15 14.305 8.935 11075@D09 RHOA 9.031 8.440 11082@B07 LOC389458 13.847 8.650 11013@H06 SPHK2 8.181 7.645 11068@F06 GDPD2 12.197 7.626 11052@D10 NMNAT2 7.036 6.575 31024@D10 RTKN 10.542 6.598 31017@H11 ZNF880 6.583 6.152 11088@D08 PPM1D 10.263 6.424 51026@C08 NBN 6.108 5.709 11087@E08 PPM1D 10.147 6.352 31026@D11 AQP10 5.930 5.542 11024@E04 CXorf22 9.490 5.944 11001@B11 MLC1 5.689 5.317 51028@A01 STK39 9.341 5.852 51016@E07 KRTAP3-1 5.414 5.059 31019@D09 MYF6 8.967 5.619 11021@E12 DDAH2 5.231 4.889 51027@A04 PLA2G4F 8.769 5.496 51025@G07 SORCS1 4.759 4.448 51019@F01 FFAR1 8.631 5.411 11028@E05 GINS2 4.737 4.427 51007@B09 GRM2 8.507 5.333 11014@H11 MED24 4.645 4.341

(Table 4, continued)

31024@C07 MDM2 8.355 5.239 11073@C08 NUDT2 4.543 4.246 11006@E03 HAS3 8.348 5.235 31027@H01 CD200R1 4.499 4.204 11029@G11 HOXB6 8.207 5.147 11032@H06 CD300LG 4.261 3.982 31019@E06 IGKC 8.046 5.047 11031@E12 YIF1B 4.188 3.914 51021@A03 KDM4D 7.739 4.856 31029@B09 TCRbeta 4.073 3.807 31023@F08 CDK6 7.470 4.689 11007@H11 PEX10 4.049 3.784 11069@C02 TRAF5 7.467 4.688 11072@C05 CTAGE1 4.024 3.760 31040@E08 DLX2 7.461 4.684 11023@G08 TBC1D17 4.000 3.738 11053@D08 HSD17B12 7.446 4.674 31033@C06 C4orf7 3.975 3.715 51031@B09 SPATA12 7.354 4.618 11062@C12 PSG3 3.953 3.694 31026@E03 DLX6 7.178 4.508 51011@C03 PRDM1 3.925 3.668 51010@B12 KIAA1958 7.092 4.455 11063@H09 ALDH3A1 3.909 3.654 51025@H08 PLEKHA5 6.357 3.998 11052@A02 CDK2 3.791 3.543 51034@B01 IFNA4 6.331 3.982 11079@F04 ARMC3 3.786 3.538 51022@A01 SCNN1D 6.261 3.938 11068@H08 PTCD3 3.624 3.387 51020@B06 GCM1 6.187 3.893 31043@G08 IGHM 3.604 3.368 11053@H05 UPP2 6.091 3.833 11059@B04 CCDC74A 3.576 3.342 31003@C07 RNF145 6.038 3.800 11086@G05 IL12RB1 3.573 3.340 31016@D08 TP53I13 5.740 3.615 11068@B06 GPR114 3.527 3.297

31042@A06 MGC:16614 5.111 3.224 11052@F01 SPARC 3.492 3.263 51008@H03 REST 5.031 3.175 11026@A05 RBM26 3.467 3.240 31041@H07 XPO5 4.958 3.129 11083@C07 CCDC85C 3.456 3.230 11009@F01 RGP1 4.687 2.961 11032@E11 CD84 3.444 3.219 31042@E12 HLA-C 4.659 2.944 51020@D12 ZNF253 3.399 3.176 11070@A11 UBQLN1 4.631 2.926 11039@E03 NEDD1 3.325 3.107

51032@H07 LOC652276 4.552 2.877 51023@A10 XPO6 3.274 3.059 11020@C03 LAT 4.506 2.848 11014@G03 TCIRG1 3.215 3.005 11012@A05 SLC38A7 4.460 2.820 11040@E02 E2F2 3.171 2.963 11025@F09 DIS3L 4.441 2.808 31005@E09 L1H3-like 3.107 2.904 11085@C02 HSD17B14 4.428 2.800 11070@A11 UBQLN1 3.095 2.893 51025@G07 SORCS1 4.255 2.692 31017@D06 NOP14 3.020 2.823 51026@C08 NBN 4.180 2.646 31042@A07 TSPAN17 2.964 2.770 11033@E12 GHITM 4.008 2.539 51001@B02 KU-MEL-3 2.942 2.749 31027@G07 NELL1 3.973 2.518 31009@F07 RDBP 2.926 2.735 51021@F04 DKFZp761E198 3.898 2.471 11009@E12 C12orf41 2.866 2.679 31009@C04 INO80B 3.892 2.467 11008@D10 NOV 2.856 2.669 51012@F06 ZNF583 3.861 2.448 11058@A08 MAPK11 2.787 2.605 51016@G10 LCE1B 3.784 2.400 51020@A02 KRTAP10-4 2.691 2.515 51023@A10 XPO6 3.757 2.383 11022@G10 SLC43A3 2.683 2.507 51017@B05 IL13 3.521 2.236 31015@H10 TGIF2 2.679 2.504 51016@H03 TMSL1 3.510 2.230 11055@G06 PSG5 2.670 2.496 11050@E03 GJB1 3.309 2.105 11082@B07 LOC389458 2.659 2.485

(Table 4, continued)

11034@C10 NUDCD3 3.238 2.061 51032@F08 C4orf46 2.590 2.421 11072@G04 CDH18 3.212 2.045 31011@G07 LOC147804 2.589 2.420 51013@B08 UBAP1 3.169 2.018 11044@F08 PLEKHB1 2.556 2.389 51006@B03 CBLL1 3.164 2.015 31014@H06 PM20D1 2.554 2.386 11053@D05 PIM2 2.511 2.347 31019@E07 IGKC 2.498 2.335 51016@H03 TMSL1 2.472 2.310 31027@F05 CYP3A4 2.454 2.293 51032@H07 LOC652276 2.441 2.281 31017@B10 HAMP 2.324 2.172 51023@A03 KIAA1383 2.311 2.160 51004@E07 PDCD1LG2 2.284 2.135 11012@H04 ZGPAT 2.256 2.108 31003@F07 SCEL 2.238 2.091 11030@H01 CDCA3 2.227 2.081 31035@A02 GTSE1 2.164 2.023 11081@G09 GNG11 2.163 2.021

2-5. Validation of selected screen candidates.

We noticed that the genes scored from the E7 sublibrary screen exhibited higher log2 enrichment score than those scored from the E6 screen, and we identified senescence-related GO terms from the E7 branch of the primary screen. We reasoned this might be due to the fact that the p53 pathway plays a more dominant role in regulating replicative senescence than the Rb pathway in BJ cells. Therefore we decided to focus on the E7 sublibrary hits for validation. We chose a handful of E7 screen genes with varying senescence screen scores to test individually, including a pair of highly related

Homeobox transcription factors (DLX2 and DLX6), a p53 target gene (TP53I13), a secreted protein (PSG3), a ubiquitin peptidase (UPS15), a transcription factor (MYF3) and a p53 interacting histone demethylase (KDM4D). We introduced individual ORFs from the ORFeome library to pre-senescent E7 BJs, repeated the growth curve analysis, and monitored cellular replicative lifespan. DLX2, DLX6, TP53I13, KDM4D

and USP15 each showed a senescence bypass phenotype when tested individually (Figure

12). Previous reports have suggested that some of these genes are involved in the p53 pathway or cell cycle regulation. For example, USP15 was shown to stabilize MDM2 to promote p53 degradation [149], which could explain why it scored. We noticed that the

TP53I13 and the DLX6 ORFs appeared to be truncated in the ORFeome library when compared to the NCBI reference sequence, therefore further analysis of full length ORFs will be required to elucidate the function of these ORFs in senescence.

Figure 12. Validation of selected screen candidates.

Figure 12. A) Summary of screen score, validation status of the ORFs used in the validation assay picked from the human ORFeome. B) Empty vector control, DLX2 and DLX6 ORFs were introduced into pre- senescent E7 BJs and the cells were measured for proliferation over time. DLX2 and DLX6 expression increased the replicative lifespan compared to control vector. C) USP15, TP53I13 and KDM4D expression increased cellular replicative lifespan compared to empty control vector. PPM1D was used as a positive control

III. Discussion

Here we describe a second round of E6 and E7 senescence screens with improved methodology. The candidates picked for secondary screen significantly out-performed control ORFs, and exhibited strongly improved log2 enrichment scores. As mentioned above, this could be attributed the smaller size (140-160 genes) of the sublibraries used subsequently, compared to the whole ORFeome library used in the primary screen. This reduction in library size reduces the complexity and chances for bottlenecking effects, and the average representation of each ORF is much higher given the total number of cells. Another improvement developed with the help of barcode-sequencing based quantification method. The introduction of barcodes eliminated issues associated with

PCR bias, probe-quality, and cross-hybridization. The deep sequencing approach allowed us to achieve a greater detection dynamic range. Our collection of barcoded destination vectors could be applied for future tests with ORFs of interest.

Similar to our primary screen results, we observed multiple genes known to regulate senescence pathways, particularly the p53 pathway. USP15 was among the top gene in the E7 secondary screen list. We also validated this gene individually for increased replicative lifespan. Previous studies have suggested that it plays a role in regulating TGF-beta pathway via regulation of ubiquitination of the SMAD proteins.

However, a recent report suggests that USP15 deubiquitinates MDM2, which is a critical negative regulator for p53 stability [149]. The loss of USP15 results in increased cellular p53 level and stalls proliferation, while over-expression of USP15 leads to reduced p53 level and activation [149]. In our screen, we found USP15 scored from the E7 branch but not the E6 branch, suggesting it may play a role in the p53 pathway. This observation is

well aligned with our sensitized senescence screen scheme. Other key negative p53 regulators scored include PPM1D and MDM2, and these genes showed higher enrichment scores than those in the primary screen. It is of interest to note that the

MDM2 ORF was initially annotated as “cDNA clone IMAGE:2960385” in the ORFeome library. We observed that this gene scored strongly in the sublibrary screen and hence examined its sequence. Surprisingly, the cDNA matches with perfect homology to a predicted transcription variant of MDM2 gene. Therefore we corrected our annotation of this ORF, and this finding serves as a case similar to a “blind test” that we can use to identify functional senescence regulators.

Besides the genes mentioned above, we also validated two other genes, KDM4D and TP53I13, which have been implicated in p53 regulation. KDM4D encodes a histone demethylase, and previous reports have suggested the KDM4D protein directly interacts with p53. It is reported that the loss of KDM4D halts proliferation; however, KDM4D has also been found to stimulate p53 transcription target p21. It is unclear how KDM4D affects p53 activation in the context of cellular senescence. The fact that KDM4D binds to p53 and bypasses senescence in E7 cells implies KDM4D could regulate senescence via the p53 pathway. Still, extensive further analysis will be required to reveal the nature of the regulation. TP53I13 is expressed ubiquitously in adult tissues, and is induced by p53 activation and genotoxic stress. Over-expression of TP53I13 stalls cancer cells in vitro. We found the TP53I13 ORF in the ORFeome library is truncated by sanger sequencing, lacking its N terminal compared to wild type ORF (see table 5). It is possible that the TP53I13 ORF can act through a dominant negative manner. An examination of the loss-of-function phenotype of TP53I13 on senescence will be critical to assess the

nature of this gene. Albeit the complication of dominant-negative effects, our screen efforts stand as a robust approach to reveal relevant genes in cellular senescence.

Interestingly, we identified a pair of Homeobox transcription factors, DLX2 and

DLX6, which both scored strongly in the senescence screen and validated individually.

DLX2 and DLX6 belong to the distal-less family transcription factors, which contains 6 members, DLX1-DLX6. Both DLX2 and DLX6 have been shown to be involved in brain and bone development and have not been linked to senescence. DLX6 has been shown to be focally amplified in cancer, while DLX2 has been observed to be over-expressed in several cancer types and correlates with poor prognosis. We examined the ORFeome sequence of both DLX2 and DLX6, and it turns out the ORFeome version of DLX6 represents an open reading frame from an alternative translation start site and renders it truncated compared to the NCBI reference sequence. To avoid any confusion with regard to sequence annotation, we decided to focus on the study of DLX2 to address how its expression leads to the bypass of senescence, which will be discussed in detail in next chapter.

Table 5: Summary of mis-annotated and mutated ORFs

ORFeome ID Discrepancy Annotation TP53 11060@B02 Point mutation P72R and P278A STN1 11008@G02 Point mutation T151A, S248C ATM 31028@E12 Truncated Protein ORF corresponds to N terminal TAN domain TP53I13 31016@D08 Truncated Protein N terminal truncation identified by Sanger sequencing MDM2 31024@C07 mis-annotation Annotated as "chimeric protein" TEN1 31007@e10 mis-annotation Annotated as MGC:54300 IMAGE:6500495

IV. Material and methods 4-1. List of selected complementary genes from 60-mer microarray for sublibrary screen. Complementary genes from 60-mer array hybe for E6 sublibrary

ACAD10 CYP3A4 MAPK11 PTCD3 TSPAN17 ADCK2 CYP4F22 MED24 RBM26 UBQLN1 ALDH3A1 DBNL MMP12 RDBP UCKL1 ANXA11 ERCC2 NDOR1 RIMS2 XPO6 ARHGEF7 GINS2 NEDD1 SCEL ZGPAT ARMC3 GINS2 NME4 SLC24A6 ZNF414 BEND2 GMDS NOP14 SLC25A45 ZNF430 C11orf72 GPR114 NULL SLC43A3 ZNF462 C12orf41 GTSE1 NULL SMEK1 ZNF473 C16orf13 GTSE1 NULL SORCS1 ZNF583 CA3 HEPH OSM SPARC ZP2 CALB2 HTR1D PASK SPHK2 CBLL1 IGHM PDE11A STK39 CCDC101 IL10RB PEX10 TBC1D17 CCDC41 IL12RB1 PEX10 TCIRG1 CCDC64B KAT2A PKP3 TFPT CCDC74A KLHL28 PLCB1 TGIF2 CD300LG LCP2 PLEKHB1 TIMP2 CDCA3 LOC389458 PM20D1 TMEM174 CPA3 LOC400713 PRDM1 TOM1 Complementary genes from 60-mer array hybe for E7 sublibrary A4GALT KLHL28 SLMAP APOBEC3H LYPD5 SMEK1 BEND2 MAP3K14 SORCS1 BMX MED19 STK39 CBLL1 MXD3 SUPT7L CD3EAP MYBL2 TP53I13 CXorf22 MYL6 TRAF5 DIS3L NUDCD3 UBQLN1 DLX6 NUFIP2 UBQLN1 DOCK4 NULL UCKL1 ELOVL4 NULL UNC5C ENHO PABPC3 USP15 ERI1 PDE11A XPO5

(List 4-1 Complementary genes from 60-mer array for sublibrary screen: continued)

FFAR1 PLA2G4F XPO6 GGT7 PLCB1 ZNF335 GJB1 PLEKHA5 ZNF583 HIF3A PLEKHG4 INO80B RTKN JMJD2D RTKN KLC3 SLC39A10

4-2. E6 and E7 Sublibrary composition E7 sublibrary content symbol symbol symbol symbol symbol HOXA1 PDE1B TP53I13 MCM8 HMMR TMEM54 GDPD2 UCKL1 LCE2D SLMAP HAS3 TRAF5 MYF6 ODF3 KIAA1383 A4GALT ADAD2 IGKC CBLL1 XPO6 RGP1 UBQLN1 IGKC GRM2 USP15 SLC38A7 BMX PSG2 SLC12A8 CHRD MXD3 CDH18 TCL6 PLEKHG4 PDE11A LAT BEND2 CDK6 REST DOCK4 CXorf22 TFCP2 MDM2 EEF2 ZNF335 MAP3K14 GNG11 UBQLN1 SLC39A10 SORCS1 DIS3L LOC389458 RTKN SH3D19 PLEKHA5 HOXB6 HSD17B14 DLX6 KIAA1958 PREPL SNRPB2 RTKN NELL1 CTCFL NBN GHITM ZYX CC2D2B ZNF583 PLA2G4F NUDCD3 PPM1D RPS16 CD3EAP NUFIP2 ZNF350 PPM1D SUPT7L UBAP1 GGT7 MYBL2 PABPC3 APOBEC3H ARHGEF3 STK39 MYL6 FEM1C SOX7 LCE1B TPTE2 GJB1 SEMA3B SLC43A2 TMSL1 SPATA12 HSD17B12 BRUNOL5 HIF3A IL13 NULL ELOVL4 RNF145 ANKS1B KRTAP10-1 IFNA4 UPP2 C4orf21 PLCB1 KRTAP10-7 GPBAR1 LYPD5 CCNL2 FFAR1 PSG5 NULL DLX2 PRB3 ZNF843 KLHL28 MED19 GCM1 ERI1 PTK2B PGLS ZNF253 HLA-C UNC5C XPO5 JMJD2D B3GNT5 INO80B NULL DKFZp761E198 PSG3 ENHO HLA-C ZNF614

(List 4-2 E6 and E7 sublibrary composition: continued)

CORO1B KLC3 SMEK1 SCNN1D

E6 sublibrary content symbol symbol symbol symbol symbol symbol MLC1 NEDD1 ARMC3 CCL11 IL10RB ACAD10 PEX10 NULL PKP3 LOC400713 KAT2A PDE11A PEX10 E2F2 NULL IGKC SMEK1 SORCS1 NOV NME4 FRMD8 ARL15 HEPH NBN C12orf41 PLEKHB1 GNG11 PSG2 GTSE1 STK39 ERCC2 RGS16 LOC389458 C2orf47 CD320 CCDC64B ZGPAT TMEM174 CCDC85C KRTCAP2 KU-MEL-3 C4orf46 SPHK2 CA3 TFPT TIMP2 LOC222699 NULL TCIRG1 CCDC101 ZNF414 IFNA13 LOC339535 MED24 CDK2 CALB2 AQP10 PDCD1LG2 C7orf55 NMNAT2 C11orf75 CYP3A4 CNBD1 OSM SPARC SPG7 CYP4F22 CBLL1 DDAH2 HNRNPA0 IL12RB1 CD200R1 KIAA1958 ADCK2 PIM2 LCP2 ARHGEF7 PRDM1 SLC43A3 PSG5 SCEL PRUNE SLC24A6 TBC1D17 MAPK11 RIMS2 NULL ZNF583 RBM26 HTR1D SLC25A45 CTAGE5 ZNF430 ATG12 CCDC74A NULL C4orf7 MMP12 GINS2 PSG3 KLHL28 WFDC11 C11orf72 GINS2 ALDH3A1 ZNF462 GTSE1 KRTAP3-1 RHOA GPR114 PASK PLCB1 LCE1B NEGR1 PTCD3 RAB17 LENG1 TMSL1 CDCA3 UBQLN1 RDBP NDOR1 KRTAP10-1 YIF1B SCMH1 TOM1 C16orf13 KRTAP10-7 CD84 CTAGE1 LOC147804 TSPAN17 KRTAP10-4 CD300LG BEND2 PM20D1 HLA-C ZNF253 RRS1 NUDT2 TGIF2 MPST CCDC41 GMDS FGF12 UCKL1 FN3KRP ZP2 DBNL RHOA HAMP IGHM KIAA1383 ANXA11 ZNF473 NOP14 CPA3 XPO6

Control ORF pool content symbol symbol symbol symbol CTNND1 BUD13 SLC5A6 GTSE1 FLNA MVP TTC17 CPSF1

(List 4-2 E6 and E7 sublibrary composition: continued)

SLC22A4 TMSB15B IP6K2 TRIM73 CNOT6 POLR2H FLJ42875 CENPM C13orf18 CCBL1 RASSF9 NUP93 RPA1 XPNPEP1 MAN1B1 TBL3 ARHGEF1 RNF31 HEPH RPL39 SMEK1 GSTM3 LOC440173 SPINK6 CYorf15B RPL24 TMEM217 CDC20B DDO C16orf89 DIP2A INTS7

4-3. Construction of barcoded lentiviral vectors. A synthetic 2 X2 5mer barcode pool from IDT was PCR amplified with universal

primer containing homology to the pHAGE-TREX vector. (Primer 1: 5’-AGA GAT CCA

GTT TGG ACT AGG CTA GCC GGT GTC GGT CTC GTA GTT-3’. Primer 2:5’- AGT

TAT GTA ACG ACA TGC ATG CTA GCG TCG TCC AGC TGC GAA CGA-3’). The

PCR products were gel purified and mixed with Nhe-I linearized pHAGE-TREX vector

following the SLIC (Sequence and Ligation Independent Cloning) cloning protocol [150].

The cloning reaction mix was incubated at 37oC for 30 min and transformed in to ccdB

resistant E. coli. Single colonies were picked after transformation, and the plasmid DNA

was used for determining the exact barcode sequence by Sanger sequencing. Then 200

barcoded pHAGE-TREX destination vectors with available barcode information were

arrayed on 96-well plates.

4-4. LR reaction and transformation in 96-well plate. LR reaction was carried out in a 96-well plate with individual ORF entry clones

and barcoded pHAGE-TREX destination vectors. In a total volume of 5uL, 5-10ug of

pDNOR223 entry clone with candidate ORF were mixed with 10-30ug of pre-linearized

barcoded destination vectors in the presence of 2uL LR clonase mix. LR reaction was carried out at room temperature for 16 hours and diluted to 20uL.

5uL of individual LR reaction product was used to transform competent E. coli cells in 96-well plates. The transformation was performed following standard procedure.

After transformation the E. coli from 200 wells were pooled together to prepare the sublibrary DNA for virus packing purposes.

4-5. Barcoded ORF sublibrary screen The workflow of the barcoded sublibrary screen was described in Figure 10.

Specifically, 1 x 106 cells were infected with the E6 or E7 barcoded sublibrary with low

M.O.I. Same number of cells were infected with the barcoded control ORF and GFP vector pool and mixed with the sublibrary cells to form the Start cell population. The cells were passaged after reaching senescence, and both start and end cells were harvested for extracting genomic DNA. Genomic DNA was extracted from >4M cells with the ChargeSwitch genomic DNA mini tissue kit (Thermos Fisher, Catalog

#cs11204). The genomic DNA was PCR amplified with following primers: primer 1: 5’-

CGC CGA ATT CAC AAA TGG CAG T-3’;; Primer 2: 5’-

GGACATCCGAGCTCGATATCATCG-3’. The PCR products were subsequently amplified with indexing primers, pooled together and sequenced with a customized sequencing primer (5’- CTA GGC TAG CCG GTG TCG GTC TCG TAG TT-3’). Each gene’s enrichment score was calculated using the relative barcode abundance in the start and end sample and normalized to the average control ORF enrichment score.

Chapter IV. DLX2 expression bypasses senescence by suppressing ATM-p53 signaling

Abstract

DLX2 is a transcription factor involved in embryonic development. Recently

DLX2 was reported to mediate TGF beta signaling, and DLX2 has also been implicated in cancer. DLX2 expression correlates with increased proliferation, metastasis and confers poor prognosis in several types of cancer [151, 152] [153]. We identified DLX2 as a gene whose expression leads to a strong replicative lifespan extension. We explored the mechanism by which DLX2 expression affect senescence. And we found that the expression of DLX2 both reduces the expression of p21 and suppresses p53 activation.

Analysis of key DDR genes upstream of p53 upon DLX2 induction suggested that DLX2 reduced the protein level of PIKK family members, including ATM, DNA-PK and mTOR. Further analysis revealed that DLX2 expression impairs the function of the triple

T complex by reducing its protein components TTI-1 and TEL-2 through a post- transcriptional mechanism. Our results suggest a novel mechanism by which DLX2 regulates the DDR response, which may provide new insight on how cancer cells bypass senescence.

I. Introduction

The DNA damage response has been shown to play a critical role in initiating the senescence program. Apical DDR kinases such as ATM, ATR and DNA-PK are directly recruited to DNA lesions such as single strand breaks or double strand breaks (DSB) and become rapidly activated. These activated kinases readily phosphorylate a myriad of downstream mediators and effectors, trigger cell cycle arrest, cell death and facilitate

DNA damage repair. In senescence, it is thought that ATM activation is the major signaling trigger. Dampening DDR response by blocking ATM signaling effectively bypasses or delays senescence growth arrest. Individuals with mutated DNR kinase genes are known to harbor strongly increased cancer risk.

To date our understanding of ATM activation focuses on posttranslational modifications, especially its phosphorylation. Only a handful of genes are thought to affect ATM functionality other than affecting its phosphorylation. One such mechanism is the Triple T complex mediated stabilization of PIKK family kinases. ATM, ATR and

DNA-PK belong to the PIKK protein kinase family, including three extra kinases mTOR,

TRRAP, SMG1. The PIKK family members are high molecular kinases containing multiple structural and functional domains, and they require proper protein folding to be stabilized. Tel2, TTI-1 and TTI-2 form a complex called the Triple T (TTT) complex [154]

[155] [156], and it plays a critical role in bringing in the heat shock protein chaperons for client protein folding via interaction with another protein complex called the R2TP complex [154]. Reduced Tel2, TTI-1 and TTI-2 protein level is known to result in impaired PIKK stabilization, leading to compromised DDR kinase pool and causing checkpoint defects.

We identified DLX2 as a strong hit from our senescence screen. Originally identified as a gene involved in development, DLX2 has been shown to be involved in cancer. DLX2 expression is thought to promote proliferation, metastasis and survival, and expression of DLX2 predicts poor prognosis in several cancers. However it is yet unclear how DLX2 may contribute to senescence and cancer progression. Previous reports have suggested that DLX2 can negate TGFβ signaling by suppressing TGFβ receptor expression [157]. This observation may be in line with our senescence screen results because p21, a common target of p53 and TGFβ activation, is critical for BJ cell senescence, However as we examined the relationship between TGFβ signaling and

DLX2 in our cell line, it turns out that DLX2 does not affect TGFβ signaling in our E7 BJ cell context. Interestingly, we found that DLX2 expression dampens the DDR response by reducing the Triple T complex protein levels, thus reducing levels of ATM and DNA-

PK, which are important DDR kinases. The findings will be discussed in detail below.

II. Results.

2-1. DLX2 expression delays the onset of replicative senescence. DLX2 scored strongly in both the primary and secondary E7 senescence screens.

We were able to validate the DLX2 senescence bypass phenotype by infecting relatively old, pre-senescent E7 BJ cells (Figure 12). Because multiple reports indicate that DLX2 may affect several genes involved in cell proliferation, including Cyclin D and EGFR, we cannot rule out the possibility that the extended replicative lifespan we observed reflects an increased proliferation status in a pre-senescent subpopulation in the old cells rather than directly bypassed senescence.

Figure 13. DLX2 expression bypasses replicative senescence.

Figure 13. A) Growth curve of E7 cells infected with a DLX2-expressing vector or a control vector. B) Quantitation of SA-β-Gal staining in senescing EV control cells (PD=52) and age-matched DLX2 cells. Error bars represent the standard deviation across three technical replicates. C). Annexing V staining of old EV control cell (PD=56, 9.4% apoptotic population) and age matched DLX2-expressing cells (7.87% apoptotic population).

To address this issue we introduced DLX2 to mid-aged, non-senescent cells and monitored proliferation and population doubling (Figure 13). We found that DLX2 expression did not alter the growth rate until the cells became aged, before the onset of senescence, suggesting the senescence bypass phenotype was not due to the cumulative

effect of proliferation alteration in a non-senescent population. Both cell division and cell death can affect growth curve analysis because they affect cell number. We examined if

DLX2 expression leads to altered percentage of the apoptotic population in old cells, and found that DLX2 expression does not significantly influence cell death (Figure 13). Next we examined the staining of Senescence-Associated-β-Galactosidase (SA-β-Gal), a classic senescence marker, and confirmed that when compared to senescing control cells, age-matched DLX2-expressing cells show reduced SA-β-Gal staining (Figure 13).

2-2. DLX2 induced senescence bypass is not due to analtered TGFβ pathway.

Yilmaz et al reported that DLX2 expression leads to reduced TGFβ receptor II

(TGFβRII) expression, causing suppressed p21 induction by TGFβ [157]. P21, as a critical regulator of senescence, is known to be subject to several regulatory pathways in addition to the p53 pathway, including the TGFβ pathway, retinoic acid pathway, interferon pathway and vitamin D signaling. Therefore it is possible that DLX2 regulates p21 via TGFβ signaling, in parallel to the p53-p21 senescence pathway in E7 cells.

We examined the effect of TGFβ treatment on E7 cells for p21 induction.

Unexpectedly we did not observe any induction of p21 by TGFβ treatment, suggesting the TGFβ mediated induction is not active in the context of E7 BJ cells (Figure 14).

Furthermore, we were not able to observe a decrease of TGFβRII expression in cells over-expressing DLX2; therefore it is unlikely that p21 is regulated by DLX2 via

TGFβRII suppression (Figure 14). We conclude that the DLX2 caused senescence bypass is not due to altered TGFβ pathway.

Figure 14. DLX2-induced senescence bypass is not due to an altered TGFbeta pathway

Figure 14. A) E7 BJ cells were treated with indicated concentration of recombinant TGF- β protein for 8 days. TGFβ effector SMAD2 is phosphorylated by TGFβ treatment, however p21 is not induced by TGFβ treatment. B) E7 BJ cells expressing DLX2 exhibit similar level of TGFβ RII expression.

2-3. DLX2 expression suppresses p53 activation

Since the p53-p21 pathway plays a pivotal role in senescence regulation in response to DNA damage, and we have identified DLX2 together with several other known p53 regulators in the E7 senescence screen, we examined the status of p53 activation phosphorylation mark on serine 15 and p21 expression level in both young and senescent cells (Figure 15). We found that DLX2 expression led to reduced p53 serine 15

phosphorylation and p21 expression. As critically shortened telomeres are considered the main source of DNA damage in replicative senescence, we looked at the telomerase activity in DLX2 expressing cells. We did not find any change in DLX2-expressing cells compared to control cells (data not shown), suggesting DLX2 probably acts downstream of short telomeres during replicative senescence.

Next we tested if DLX2 also suppresses p53 activation in response to non- telomeric DNA damage. We treated proliferating E7 BJ cells with Ionizing Radiation (IR) to generate DNA double-strand breaks (DSB), and we found DLX2 expression also suppresses p53-p21 activation and γ-H2AX formation in IR treated cells (Figure 15), suggesting DLX2 expression suppress the DDR-p53 pathway in a general manner that is not telomere-specific.

Figure 15. DLX2 expression suppresses p53 activation.

Figure 15. A) DLX2 expression reduced p53 and p21 activation in old (PD=52) E7 cells. B) DLX2 expression reduced p53, p21 and H2AX activation in young, non-senescent cells (PD<45). C) DLX2 expression reduced p53 and H2AX activation after IR induced DNA damage. Cells were either treated with 10Gy IR or mock treated, and harvested 24 hours post treatment. Whole cell lysate was used for western blot.

2-4. DLX2 expression bypasses Ras induced senescence. We next asked if DLX2 expression can bypass Ras induced senescence.

Consistently with our previous observation, we found DLX2 expression in hTert- immortalized BJ cells lead to reduced senescent cells and the growth arrest caused by Ras induced senescence. Analysis of p53 and γ-H2AX phosphorylation also confirmed that

DLX2 expression reduces DDR and p53 signaling in Ras induced senescence (Figure 16).

Figure 16. DLX2 expression bypasses Ras induced senescence.

Figure 16. A) SA-β-Gal staning of hTERT-BJ cells expressing either DLX2 or empty vector (EV) 14 days post pWZL-HRasV12 virus infection. B). EV and DLX2 cells were infected with control and Ras virus to trigger OIS, 14 days later 1X10e5 cells were plated in 6-well plate. Population doubling was determined after 5 days later. DLX2 expression suppresses Ras induced growth arrest compared to EV control. C). DLX2 expression reduces p53 and γ-H2AX phosphorylation by Ras induced senescence.

2-5. DLX2 suppression of p53 activation requires the DLL and the Homeodomain.

Figure 17. DLX2 suppression of p53 activation requires the DLL and the Homeodomain.

Figure 17. Left: Scheme of DLX2 protein domain deletion and point mutation clones. From top to bottom: WT-DLX2, DLX2- ΔDLL (lacking AA 51-132), DLX2- ΔHomeo Domain (lacking AA 153-211), DLX2- Homeo Domain mutant (R203A, R204A, K206A, K208A, K209A). Right: Both DLL and the Homeodomain are required for p53 and H2AX activation suppression by DLX2 induction. E7 BJ Cells containing the tet-inducible WT or mutant DLX2 were treated with 1ug/mL Doxycycline for 4 days and harvested for western blot.

DLX2 encodes a distal-less family transcription factor, which contains a

Homeodomain at the C-terminus (aa153-211) and a DLL domain in the N-terminus

(aa51-132). It is unclear if the suppression of DDR-p53 activation is due to the transcription factor activity of DLX2 or by some other functions. Therefore we generated depletion and mutation constructs to determine which domain(s) is required for DLX2 function. We generated DLX2 mutants with deletions in either the DLL domain(Δ51-132) or Homeodomain(Δ153-211), as well as a mutant ORF that has mutated five Arginine and Lysine residues in the Homeodomain critical for DNA binding(R203A, R204A,

K206A, K208A, K209A) (Figure 17). Using a tetracycline-inducible vector, we found

that both the DLL domain and the Homeodomain were required for p53 and DDR suppression by DLX2 induction. In addition, the ORF with five mutated basic amino acid also failed to suppress p53 activation, suggesting the DNA-binding residues are necessary for DLX2 function. In sum, DLX2 suppression of p53 activation is due to its transcription factor activity.

2-6. DLX2 expression leads to reduced PIKK kinase ATM, DNA-PKcs and mTOR protein levels.

Short telomeres and DSBs activate the DDR kinase ATM, which activates p53 transcriptional activity by phosphorylation of p53, Mdm2 and Chk2. To further explore the mechanism by which DLX2 expression attenuates p53 and bypasses senescence, we examined the activation of ATM on serine 1981. We found that after 4 days of doxycycline induction of DLX2 expression, E7 expressing fibroblasts exhibit reduced phosphorylated and total protein levels of ATM. This finding suggests DLX2 may suppress senescence by its regulation of ATM protein levels (Figure 18). We reasoned that reduced ATM protein level may lead to compromised chromatin recruitment of various downstream DDR mediators and effectors. Consequently, we examined the protein level of mediators such as 53BP1 and MDC1, as well as KAP-1, a protein recruited to chromatin when phosphorylated by ATM. Consistently, DLX2 expression reduced the level of these proteins on chromatin, suggesting the ATM level decrease reduced the overall DDR signaling activation (Figure 18).

Similar to ATM, DNA-PK also responds to DNA damage by phosphorylating downstream factors including p53. With this in mind, we examined if DLX2 induction also affects the protein level of the DNA-PK catalytic subunit (DNA-PKcs) (Figure 18).

Interestingly, DLX2 induction profoundly reduces DNA-PKcs protein levels. We quantified the mRNA level of ATM and DNA-PKcs by RT-qPCR, and found that DLX2 induction did not affect ATM and DNA-PKcs mRNA levels (Figure 18), suggesting the

Figure 18. DLX2 expression leads to reduced PIKK kinase ATM, DNA-PKcs and mTOR protein level.

Figure 18. A) DLX2 induction reduced phosphorylated ATM (Ser1981) and total ATM protein levels. B) DLX2 expression reduced chromatin-bound fraction of MDC1, 53BP1, as well as phosphorylated ATM substrate KAP-1 S824.C) DLX2 induction reduced ATM, DNA-PKcs and mTOR protein level. D) DLX2 induction did not alter mRNA levels of ATM and DNA-PKcs. Error bars represent the standard deviation across three technical replicates. E7 BJ cells containing the tet-inducible DLX2 vector were induced with 1ug/mL Doxycycline for 4 days. involvement of post-transcriptional mechanisms. Both ATM and DNA-PKcs belong to the Phosphatidylinositol 3-kinase-related Kinase (PIKK) family, which contains four additional members including mTOR, ATR, SMG1 and TRAPP. We examined mTOR

protein amounts by western blotting, and found that mTOR protein levels were also reduced after DLX2 induction (Figure 18), indicating DLX2 affects multiple PIKK family proteins.

2-7. DLX2 expression reduces PIKK protein levels during senescence

DLX2 expression in proliferating cells resulted in reduced PIKK protein levels.

However, we wanted to examine PIKK levels under the physiologically relevant conditions of senescence. Therefore, we asked if DLX2 expression also reduces PIKK protein levels during the onset of replicative senescence. We found that senescing E7 cells containing vector alone express substantial levels of ATM and mTOR, while age- matched DLX2 expressing cells have markedly reduced levels of ATM and mTOR

(Figure 19). We also blotted for DNA-PKcs, however DNA-PK levels in replicative senescent cells could not be detected (data not shown). We hypothesized that this might be a characteristic of replicative senescence-mediated differentiation. Consistent with our results with replicative senescence, DLX2 expression also led to reduced levels of ATM and DNA-PKcs protein (Figure 19), indicating a conserved mechanism of the DLX2-

ATM regulation across senescence caused by different stimuli.

Figure 19. DLX2 expression reduces PIKK protein levels during senescence.

Figure 19. A) E7 BJ cells constitutively expressing DLX2 exhibited reduced level of ATM, mTOR compared to senescent empty vector(EV) control cells. C) DLX2 expression reduced ATM, DNA-PKcs and mTOR in Ras induced senescence in hTERT-immortalized BJ cells. Cells were infected with retroviral H-RasV12 virus and harvested 14 days post infection.

2-8. DLX2 expression leads to reduced Triple T complex component level.

The fact that DLX2 expression affects the levels of multiple PIKK family members prompted us to explore if DLX2 was involved in a general post-transcriptional regulatory mechanism shared by the whole PIKK family. One such mechanism is PIKK protein stabilization by the Triple T complex [155] [156]. As mentioned, the Triple T complex contains three components: TEL2, TTI1 and TTI2. This complex interacts with

PIKKs and another complex named the R2TP complex to recruit the HSP70/90 chaperone machinery for the proper folding and stabilization of the PIKKs [158]. Genetic

ablation and siRNA mediated knock-down of Triple T genes have been shown to reduce

PIKK protein levels [156].

To ascertain whether DLX2 expression impacts the Triple T complex we examined the protein levels of TEL2 and TTI-1 after DLX2 induction by western blot.

We found that DLX2 induction indeed lowered the protein levels of TEL2 and TTI-1

(Figure 20), however DLX2 expression did not affect TEL2 and TTI-1 mRNA levels

(Figure 20), suggesting the Triple T components are regulated by DLX2 by post- transcriptional mechanism. Since it is known that impaired PIKK kinase folding by the

Triple T complex leads to lysosome mediated protein degradation [159], we tested if lysosomal protease inhibition would restore ATM and DNA-PKcs protein levels in the presence of DLX2 expression. We observed that Leupeptin, a lysosomal protease inhibitor restored ATM and DNA-PK levels after DLX2 induction, indicating that DLX2 expression results in impaired protein folding and degradation of PIKKs mediated by the

Triple T complex (Figure 20)

Figure 20. DLX2 expression leads to reduced TTT complex component level.

Figure 20. A) DLX2 induction led to reduced TTI-1 protein level in E7 BJ and IMR90 cells. B) DLX2 induction led to reduced TEL2 protein level in IMR90 cells. C) DLX2 induction did not alter the mRNA level of TTI1, TTI2 and TEL2. D) Leupeptin treatment restored DLX2 induced ATM protein reduction. E7 BJ cells with inducible DLX2 were induced with 1ug/mL Doxycycline for 5 days and then treated with DMSO or 100uM Leupeptin 24 hours prior to harvesting

III. Discussion

DLX2 and its family members were originally discovered as a class of genes involved in neuronal cell specification and development. Later studies showed that DLX2 is involved craniofacial development as well. Recent reports indicate DLX2 also functions out of the developmental context, as DLX2 has been shown to be involved in cell proliferation, survival, metastasis, and TGFβ signaling. Our finding indicates that

DLX2 expression also affects cellular senescence by manipulating the most upstream

DDR kinases ATM and DNA-PK. We consider the reduction of the PIKK kinases may be due to reduced Triple T complex level. While we think our findings provide new knowledge with regard to senescence and DDR regulation, several issues remain to be explore in the future, as discussed below.

First, the physiological relevance of DLX2 regulation on senescence remains to be elucidated. Because DLX2 was originally defined as a developmental gene, and in fact its expression is extremely low in most normal tissues, therefore it is challenging to examine the loss-of-function phenotype of DLX2 in normal tissues. Another way to examine the physiological role of DLX2 would be to perform loss-of-function studies in cancer cells to see if its ablation compromises tumorigenesis. Because PIKK family kinases participate in many critical biological processes, a total reduction or increase of all PIKK proteins as a result of DLX2 alteration may promote proliferation in one way but suppress it in another way. A simple knock down of DLX2 may have confounding consequences depending on the cell context. Therefore, a careful examination with multiple cell lines expressing DLX2 will be required to draw a comprehensive conclusion.

We recently compared the DLX2 over-expression status and p53 status in several large

TCGA studies, and found that DLX2 over-expression has mutually exclusive relationship with p53 alteration, suggesting DLX2 may play a role in cancer to attenuate p53 pathway,

We will discuss this in the next chapter.

Second, it is unclear how DLX2 reduces the PIKK protein level via the Triple T complex. We observed protein level changes in TEL2, TTI-1 and TTI-2 in the absence of mRNA changes. Because the Triple T complex proteins are inter-dependent on each other, that the loss of one component reduces other components as well. In addition, the triple T proteins appear to be co-regulated by a common regulatory mechanism. For example, inositol-pyrophosphate has been shown to promote CK2 phosphorylation on both TTI-1 and TEL2, and those phosphorylation sites have been indicated in regulating the stability of TTI-1 and TEL2. Therefore, it is unclear if the reduction of the proteins is a secondary effect due to the loss of one particular protein, or if DLX2 causes a reduction of multiple TTT proteins simultaneously. It is difficult to perform a rescue experiment without the knowledge of direct target of DLX2 mediating this effect. To partially address this issue, we attempted to perform a rescue experiment by abrogating the lysosomal protease activity to restore PIKK protein level. Results suggest at least the

DLX2 mediated reduction of PIKK proteins is due to the lysosomal degradation of misfolded protein commonly known downstream of The TTT complex. An alternative approach should be to perform a rescue experiment with both single Triple T gene expressing construct, as well as combinations of the constructs, to restore PIKK levels upon DLX2 expression.

Third, as mentioned above, we still lack the knowledge how DLX2 reduces individual TTT protein level. Previously, some reports suggested the IP6K2-CK2

phosphorylation pathway affects TTT protein levels [159]. We tested this pathway with

IP6K2 null cells, and also examined CK2 activity in DLX2 expressing cells. It turns out

IP6K2 is not required for DLX2 mediated regulation on PIKKs, and the CK2 phosphorylation on Triple T proteins is not affected by DLX2. Therefore, it remains an unresolved issue that requires further efforts.

Last but not least, we identified two ORFs mapped to DLX2 and DLX6, two members of the Distal-less transcription factor family. Actually we observed the over- expression of the DLX2 ORF in the ORFeome library leads to reduced p21 expression and p53 activation. Given the structural similarity between DLX2 and DLX6, we also looked at other Homeobox and DLX genes in the library. It turns out all other DLX family proteins (DLX1, DLX3, DLX4,DLX5 ) did not score from either branch of the senescence screen. Among the 20 Homeobox (HOX) genes in the ORFeome library, only

2 scored in the E7 screen, indicating an ability to bypass senescence through attenuated

ATM-p53 signaling is a DLX2-specific attribute not shared by other related proteins.

IV. Material and methods

4-1. Growth curve and SA-β-galactosidase assays. Cells were plated at a density between 1-1.5 x 106 per 10 cm tissue culture plate

(127-191*103/cm2) and fed every two to three days. Cells number was monitored using a coulter counter for calculating population doublings. Cells were continuously passaged until the growth curve became flat. The SA-β-galactosidase assay was performed using the Senescent Cell Staining Kit (Sigma) following the manufacturer’s instructions.

4-2. Plasmids and Cloning pWZL-hygro-HRasV12 and pWZL-hygro control vectors for the Ras induced senescence experiments were obtained from Addgene. pHAGE-TREx lentiviral vector was used for constitutive DLX2 expression in E7 BJ and IMR90 cells and this vector was previously described (Yoon et al. 2014). pHAGE-CMV-N-FLAG-HA lentiviral vector was used for DLX2 expression in hTERT-immortalized BJ cells.

The DLX2 mutant ORFs used for domain analysis were generated by overlap extension PCR, and subsequently cloned into pDONR-223 vector with BP cloning followed by LR reaction into pHAGE-inducer-20(Meerbrey et al. 2011).

4-3. Immunoblotting and Antibodies Cells were lysed in 1X NuPAGE LDS sample buffer supplemented with 1X

Bond-Breaker TCEP (Life Technologies) and 1X Halt protease and phosphatase cocktail

(life technologies) on ice. The whole cell lysates were sonicated briefly to break down the chromatin. Samples were boiled at 95oC for 5 minutes and then loaded to 4-12% Bris-

Tris SDS protein gel or 3-8% Tris Acetate SDS protein gel (life technologies).

Antibodies used were as follows: p53(DO-1, Calbiochem), phospho-p53s15

(#9284, Cell signaling), p21(OP64, Calbiochem), DLX2(ab30339, Abcam), γH2AX(05-

636, Millipore), Vinculin(V9131, Sigma). Phospho-ATMs1981(ab81292, Abcam),

ATM(GTX70103, Genetex), DNA-PKcs(A300-517A, Bethyl), mTOR(#4517, Cell

Signaling), TTI1(A303-451A, Bethyl), TEL2(15975-1-AP, ProteinTech).

4-4. RT-qPCR Total RNA was isolated using the RNAeasy Mini Kit (Qiagen) and cDNA was synthesized using Superscript VILO cDNA synthesis kit (Invitrogen) according to the

manufacturer’s instructions. Quantitative RT-PCR was performed in triplicate or quadruplicate using the TaqMan Gene Expression Master Mix (Invitrogen) with Taqman gene expression assay (Life Technologies) on an Applied Biosystems Fast 7500 machine, using GAPDH as the endogenous normalization control. Data were normalized to empty vector control cells. The IDs for the Taqman assays used were as follows: ATM

(Hs01112314_m1), DNA-PKcs (Hs04195439_s1), TTI-1 (Hs00384863_m1), TTI-2

(Hs00228357_m1), TEL2 (Hs00211542_m1), GAPDH (Hs02758991_g1) and p21

(Hs99999142_m1).

Chapter V. Conclusions and Perspectives. I. Genetic screens to identify novel regulators of senescence.

Forward genetic screens have led to countless meaningful discoveries in various biological processes and model organisms. In cellular senescence, unbiased screens revealed many senescence regulators. For example, some of the early SASP factors were originally identified through loss-of-function screens [82], and such screens helped add to the detailed, molecular illustration of how known senescence regulators work. It is widely known that the p53 tumor suppressor gene plays a pivotal role in regulating the expression of the key senescence gene p21. However, not all tumor cells bypassing senescence inactivate their p53 locus by deletion or mutation. Loss-of-function screens performed in our lab and others found that a Bromodomain protein BRD7 binds to p53, and is required for p53 activation during senescence. Many cancer lines that have intact p53 gene harbor BRD7 alterations instead, since BRD7 inactivation is sufficient to shut down the senescence program in both replicative senescence and oncogene induced senescence [160, 161]. Therefore, genetic screens have a proven record of identifying relevant players in senescence.

Unlike the loss-of-function screen approaches, we took aim at senescence by using another set of tools: the human ORFeome library. The ORFeome library provides a substantial coverage of the transcriptome, and it also provides nearly uniform representation of ORFs available, facilitating the universal interrogation of most of the available genes. Moreover, our over-expression-based screen platform eliminates the concern from potential off-target effects commonly observed in RNAi screens and possibly CRISPR based screens. However, it is noteworthy that concerns may exist with

100

regard to the physiological relevance of the genes identified with our screen. Since the screen assay itself does not take into consideration of the loss-of-function phenotype, it is important to put any discovery from the screen in a physiologically relevant settings such as aging or tumor suppression. One future direction is to systemically examine the loss- of-function phenotype of genes scored in our gain-of-function screen to identify genes that are both necessary and sufficient for senescence. Nevertheless, we believe our screen results provide an example that justifies our approach for the identification of meaningful genes in senescence. We identified multiple genes that functionally interact with known pathways, including dnp53, PPM1D, MDM2, USP15, RHOA, as well as new ones such as DLX2. In addition, we were able to find genes representing novel upstream or parallel pathways for senescence, including the identification of members of the CST complex, which may provide insight on how telomere attrition regulation can be harnessed to delay senescence. As mentioned previously, senescence is controlled by multiple unique and yet somewhat redundant pathways, and the senescence program makeup may be cell type- and context-dependent. Therefore it would be interesting to extend our screening effort into additional cell types and/or engineered cell types that are known to rely on less-understood senescence pathways such as the p16 pathway, the SASP pathway and/or the telomere pathway.

II. DLX2, cellular senescence and tumor suppression.

One critical question that remains to be answered is on the physiological involvement of DLX2 in senescence, tissue and organismal aging, and tumor suppression.

One main drawback we face in mechanistic study is that the basal level of DLX2 expression in proliferating cells is extremely low, hence it is hard to investigate the

101

cellular phenotype when DLX2 is abolished in fibroblasts. For future attempts, to elucidate the role of DLX2 in regulating senescence, we can characterize the knock-down phenotype of DLX2 in tissues known to express DLX2, e.g. the embryonic brain.

Additionally, the fact that fibroblast cells responding to DLX2 expression by bypassing senescence, though DLX2 is not normally expressed outside of embryonic development, suggests certain element(s) that enables cells to reduce the triple T complex and PIKK protein level in normal cells. It would be important to identify such element(s) as well as the molecular mechanisms involved, for these factors may play a relevant role such as preventing premature senescence in fibroblasts.

Another area worth investigating is the role of DLX2 in tumor suppression. Since we found that DLX2 expression attenuates ATR-p53 signaling and senescence, it is possible that DLX2 is involved in inactivating senescence and/or apoptosis in cancer cells.

Therefore, we tried to knock down DLX2 expression by shRNA in cancer cell lines that express high DLX2, hoping to observe an activation of DDR and increased ATM protein level. However, we were unable to observe such effects in the handful of cell lines we examined (data not shown). We reasoned that the effect on ATM and other PIKK protein levels may be context-dependent; cancer cells may once require DLX2 to extend their replicative lifespan (e.g. to allow for the activation of telomere maintenance mechanisms), but later may acquire changes to restore PIKK protein levels since the PIKKs play profound roles in DNA repair, cell proliferation, autophagy, etc. On the other hand,

DLX2 as a transcription regulator has also been implicated in regulating cell proliferating genes such as Cyclin D and EGFR, as well as TGFβ receptor II. It is possible that DLX2 expression contributes to cancer progress by affecting these additional pathways.

102

Therefore a thorough, large-scale examination of these candidate pathways across many cancer cell lines will be required to address this question. However, this approach can be labor intensive and costly depending on the number of cell lines examined.

To provide additional insight, we examined published cancer sequencing and gene expression data for the relationship of DLX2 over-expression status and p53 mutation status. Because p53 is perhaps the single most frequently mutated tumor suppressor, and our study suggests DLX2 expression attenuates p53 signaling, we reason if DLX2-expression indeed abolishes p53 signaling in vivo, then cancer cell lines with

DLX2 over-expression would have a reduced need for p53 alteration, including p53 mutation, copy number variation, or expression changes. We utilized the C-bio portal bioinformatics tool [162] [163] and examined two large scale breast cancer patient cohort studies with available sequencing and transcriptome data [164] [165]. Interestingly, we found that in breast cancer patients, DLX2 overexpression (Z score >2) exhibited a mutually exclusive relationship with p53 alterations (Figure 21). Given the high frequency of p53 and related genes’ alteration in cancer, it is very likely that DLX2 plays a role in reducing p53 signaling during tumor formation.

103

Figure 21. DLX2 over-expression exhibits mutual exclusivity with p53 alteration in breast cancer.

Figure 21. C-bioportal analsysis of mutual exclusivity and co-occurrence between p53 alteration and DLX2 over-expression. A) DLX2 over-expression is mutually exclusive with p53 alteration (mutation/copy number variation/mRNA up/down regulation. P value= 0.049. DLX2 over expression is determined based on a Z score >2. Over expression is determined by Z >2 between normal and cancer tissue from microarray analysis of 463 patients [164]. B) DLX2 over-expression is mutually exclusive with p53 alteration (mutation /mRNA up/down regulation. P value= 0.006. DLX2 over expression is determined based on Z score >2. Over expression is determined by Z >2 between normal and cancer tissue from RNA-sequencing analysis of 971 patients [165].

104

III. Perspective Mechanism of DLX2 mediated PIKK destabilization.

In our study we found that DLX2 expression reduces TTT complex protein levels, such as TEL2 and TTI-1, and leads to destabilized PIKK protein levels and a bypassing of senescence. Although our findings suggest a new cellular strategy of bypassing senescence by destabilizing the PIKKs, it is not entirely clear how DLX2 affects TTT protein levels.

Relatively little is known about the protein level and functional regulation of TTT complex components. The TTT components are interdependent, in that depleting one component of the complex not only affects the TTT complex formation but also will readily reduce other protein levels, resulting in profound reduction of all PIKK proteins.

It has been suggested that the protein levels of TTT components are regulated by Casein

Kinase II (CK2) [166]. CK2 kinase activity has been shown to phosphorylate TEL2 and

TTI-1 on multiple sites to promote TTT ubiquitination and degradation. The binding of

Inositol Pyrophosphate (IP7) to CK2 stimulates this phosphorylation [159]. However,

CK2 mediated phosphorylation also promotes TTT binding to PIKK proteins to facilitate

PIKK folding and stabilization. This paradox suggests CK2 may affect both TTT and

PIKK in both positive and negative ways, depending on the degree of phosphorylation and cellular context. CK2 activity is affected by its upstream activator IP6K2, which has also been shown to affect PIKK protein levels, adding to the complexity of CK2-TTT regulation.

We tested if DLX2 expression could affect TTT protein levels when the IP6K2-

CK2 axis is inactivated by either genetic depletion of IP6K2 gene or chemical inhibition of CK2 kinase activity, and found that DLX2 can still reduce TTT protein levels in the

105

absence of IP6K2 or in the presence of CK2 inhibitors (data not shown). Therefore,

DLX2 mediated TTT protein level reduction may be carried out through novel mechanism(s) independent of CK2 function.

Figure 22. Prospective involvement of Hyaluronic Acid Synthase 3 (HAS3) in DLX2 induced senescence bypass and ATM-p53 attenuation.

Figure 22. A) Upper panel: Hyaluronic Acid (HA) signaling pathway genes were scored from the E7 senescence primary screen. Both HAS3 and HA receptor HMMR were scored above the Z score threshold of 1.65. Lower panel: HAS3 was included in the sublibrary and re-scored in the secondary screen. Z score=5.23. Log2 Enrichment=8.35. B) Upper panel: HAS3 expression is induced by DLX2. HAS3, together with positive control MSX2, were induced by DLX2. P value<0.05. Lower panel: DLX2 induction of HAS3 was confirmed by RT-qPCR. Cells with tet-inducible DLX2 were treated for 4 days with 1ug/lm Doxycycline, and RNA was extracted for transcription analysis of HAS3. Error bar represents S.D. of technical replicates.

106

To find additional clues for how DLX2 mediates TTT protein level reduction, we performed a transcriptome analysis following induced DLX2 expression. Among the genes identified from the transcriptome profiling, we found MSX2, which is a positive control gene known to be transcriptionally activated by DLX2 [167], confirming the validity of our transcription profiling analysis (Figure 22). In addition, MSX2 has also been shown to physically interact with DLX2 to block its function, suggesting DLX2 is possibly subject to negative feedback regulation [167]. Interestingly, we found the transcript level of HAS3 (hyaluronic acid synthase 3) was strongly induced DLX2 expression. Hyaluronic acid (HA) is a ubiquitous components in the extracellular matrix, and certain HA species have been implicated involved in tumor suppression and longevity in naked mole rat, a long-lived rodent species. We found that a HAS3 ORF and its receptor HMMR both scored in the primary senescence screen along with DLX2, and

HAS3 gene was examined and scored again in the secondary screen (Figure 22). These preliminary findings raise the interesting possibility the senescence bypass phenotype of

DLX2 may be mediated by HA signaling, which may provide more hints to explain how

DLX2 expression affects TTT and PIKK protein level.

To sum up, our work provides a well-curated candidate list for potential senescence regulators, and it will be useful to explore the underlying mechanism of the candidate genes, including DLX2, to gain insight of how senescence is regulated physiologically and pathologically. It is clear that our knowledge is far from saturating the field and opportunities remain open to future exploration.

107

Reference 1. Greider, C.W., Telomere length regulation. Annu Rev Biochem, 1996. 65: p. 337- 65. 2. Greider, C.W., Telomeres do D-loop-T-loop. Cell, 1999. 97(4): p. 419-22. 3. de Lange, T., Shelterin: the protein complex that shapes and safeguards human telomeres. Genes Dev, 2005. 19(18): p. 2100-10. 4. de Lange, T., How telomeres solve the end-protection problem. Science, 2009. 326(5955): p. 948-52. 5. Levy, M.Z., et al., Telomere end-replication problem and cell aging. J Mol Biol, 1992. 225(4): p. 951-60. 6. Campisi, J., et al., Cellular senescence, cancer and aging: the telomere connection. Exp Gerontol, 2001. 36(10): p. 1619-37. 7. Richter, T. and T. von Zglinicki, A continuous correlation between oxidative stress and telomere shortening in fibroblasts. Exp Gerontol, 2007. 42(11): p. 1039-42. 8. Shay, J.W. and W.E. Wright, Telomeres and telomerase in normal and cancer stem cells. FEBS Lett, 2010. 584(17): p. 3819-25. 9. Bodnar, A.G., et al., Mechanism of telomerase induction during T cell activation. Exp Cell Res, 1996. 228(1): p. 58-64. 10. d'Adda di Fagagna, F., et al., A DNA damage checkpoint response in telomere- initiated senescence. Nature, 2003. 426(6963): p. 194-8. 11. Zou, Y., et al., Does a sentinel or a subset of short telomeres determine replicative senescence? Mol Biol Cell, 2004. 15(8): p. 3709-18. 12. Takai, H., A. Smogorzewska, and T. de Lange, DNA damage foci at dysfunctional telomeres. Curr Biol, 2003. 13(17): p. 1549-56. 13. Kaul, Z., et al., Five dysfunctional telomeres predict onset of senescence in human cells. EMBO Rep, 2012. 13(1): p. 52-9. 14. Bodnar, A.G., et al., Extension of life-span by introduction of telomerase into normal human cells. Science, 1998. 279(5349): p. 349-52. 15. Serrano, M., et al., Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a. Cell, 1997. 88(5): p. 593-602. 16. Bartkova, J., et al., DNA damage response as a candidate anti-cancer barrier in early human tumorigenesis. Nature, 2005. 434(7035): p. 864-70. 17. Vafa, O., et al., c-Myc can induce DNA damage, increase reactive oxygen species, and mitigate p53 function: a mechanism for oncogene-induced genetic instability. Mol Cell, 2002. 9(5): p. 1031-44.

108

18. Zhu, J., et al., Senescence of human fibroblasts induced by oncogenic Raf. Genes Dev, 1998. 12(19): p. 2997-3007. 19. Michaloglou, C., et al., BRAFE600-associated senescence-like cell cycle arrest of human naevi. Nature, 2005. 436(7051): p. 720-4. 20. Chen, Z., et al., Crucial role of p53-dependent cellular senescence in suppression of Pten-deficient tumorigenesis. Nature, 2005. 436(7051): p. 725-30. 21. Collado, M., et al., Tumour biology: senescence in premalignant tumours. Nature, 2005. 436(7051): p. 642. 22. Bartkova, J., et al., Oncogene-induced senescence is part of the tumorigenesis barrier imposed by DNA damage checkpoints. Nature, 2006. 444(7119): p. 633-7. 23. Di Micco, R., et al., Oncogene-induced senescence is a DNA damage response triggered by DNA hyper-replication. Nature, 2006. 444(7119): p. 638-42. 24. Courtois-Cox, S., et al., A negative feedback signaling network underlies oncogene-induced senescence. Cancer Cell, 2006. 10(6): p. 459-72. 25. von Zglinicki, T., Oxidative stress shortens telomeres. Trends Biochem Sci, 2002. 27(7): p. 339-44. 26. Parrinello, S., et al., Oxygen sensitivity severely limits the replicative lifespan of murine fibroblasts. Nat Cell Biol, 2003. 5(8): p. 741-7. 27. Sedelnikova, O.A., et al., Senescing human cells and ageing mice accumulate DNA lesions with unrepairable double-strand breaks. Nat Cell Biol, 2004. 6(2): p. 168-70. 28. Lee, A.C., et al., Ras proteins induce senescence by altering the intracellular levels of reactive oxygen species. J Biol Chem, 1999. 274(12): p. 7936-40. 29. Aird, K.M., et al., Suppression of nucleotide metabolism underlies the establishment and maintenance of oncogene-induced senescence. Cell Rep, 2013. 3(4): p. 1252-65. 30. Kaplon, J., et al., A key role for mitochondrial gatekeeper pyruvate dehydrogenase in oncogene-induced senescence. Nature, 2013. 498(7452): p. 109-12. 31. Ogryzko, V.V., et al., Human fibroblast commitment to a senescence-like state in response to histone deacetylase inhibitors is cell cycle dependent. Mol Cell Biol, 1996. 16(9): p. 5210-8. 32. Munro, J., et al., Histone deacetylase inhibitors induce a senescence-like state in human cells by a p16-dependent mechanism that is independent of a mitotic clock. Exp Cell Res, 2004. 295(2): p. 525-38.

109

33. Di Bernardo, G., et al., Histone deacetylase inhibitors promote apoptosis and senescence in human mesenchymal stem cells. Stem Cells Dev, 2009. 18(4): p. 573-81. 34. Okamoto, H., et al., Trichostatin A, an inhibitor of histone deacetylase, inhibits smooth muscle cell proliferation via induction of p21(WAF1). J Atheroscler Thromb, 2006. 13(4): p. 183-91. 35. Bandyopadhyay, D., et al., Down-regulation of p300/CBP histone acetyltransferase activates a senescence checkpoint in human melanocytes. Cancer Res, 2002. 62(21): p. 6231-9. 36. Rodier, F., et al., DNA-SCARS: distinct nuclear structures that sustain damage- induced senescence growth arrest and inflammatory cytokine secretion. J Cell Sci, 2011. 124(Pt 1): p. 68-81. 37. Wang, B., et al., 53BP1, a mediator of the DNA damage checkpoint. Science, 2002. 298(5597): p. 1435-8. 38. Ivanov, A., et al., Lysosome-mediated processing of chromatin in senescence. J Cell Biol, 2013. 202(1): p. 129-43. 39. Di Leonardo, A., et al., DNA damage triggers a prolonged p53-dependent G1 arrest and long-term induction of Cip1 in normal human fibroblasts. Genes Dev, 1994. 8(21): p. 2540-51. 40. Herbig, U., et al., Telomere shortening triggers senescence of human cells through a pathway involving ATM, p53, and p21(CIP1), but not p16(INK4a). Mol Cell, 2004. 14(4): p. 501-13. 41. Fang, L., et al., p21Waf1/Cip1/Sdi1 induces permanent growth arrest with markers of replicative senescence in human tumor cells lacking functional p53. Oncogene, 1999. 18(18): p. 2789-97. 42. Beausejour, C.M., et al., Reversal of human cellular senescence: roles of the p53 and p16 pathways. Embo j, 2003. 22(16): p. 4212-22. 43. Stein, G.H., et al., Differential roles for cyclin-dependent kinase inhibitors p21 and p16 in the mechanisms of senescence and differentiation in human fibroblasts. Mol Cell Biol, 1999. 19(3): p. 2109-17. 44. Bode, A.M. and Z. Dong, Post-translational modification of p53 in tumorigenesis. Nat Rev Cancer, 2004. 4(10): p. 793-805. 45. Harbour, J.W. and D.C. Dean, The Rb/E2F pathway: expanding roles and emerging paradigms. Genes Dev, 2000. 14(19): p. 2393-409. 46. Giacinti, C. and A. Giordano, RB and cell cycle progression. Oncogene, 2006. 25(38): p. 5220-7. 47. Momand, J., et al., The mdm-2 oncogene product forms a complex with the p53 protein and inhibits p53-mediated transactivation. Cell, 1992. 69(7): p. 1237-45.

110

48. Barak, Y., et al., mdm2 expression is induced by wild type p53 activity. Embo j, 1993. 12(2): p. 461-8. 49. Fiscella, M., et al., Wip1, a novel human protein phosphatase that is induced in response to ionizing radiation in a p53-dependent manner. Proc Natl Acad Sci U S A, 1997. 94(12): p. 6048-53. 50. Honda, R., H. Tanaka, and H. Yasuda, Oncoprotein MDM2 is a ubiquitin ligase E3 for tumor suppressor p53. FEBS Lett, 1997. 420(1): p. 25-7. 51. Lu, X., B. Nannenga, and L.A. Donehower, PPM1D dephosphorylates Chk1 and p53 and abrogates cell cycle checkpoints. Genes Dev, 2005. 19(10): p. 1162-74. 52. Lahav, G., Oscillations by the p53-Mdm2 feedback loop. Adv Exp Med Biol, 2008. 641: p. 28-38. 53. Loewer, A., et al., Basal dynamics of p53 reveal transcriptionally attenuated pulses in cycling cells. Cell, 2010. 142(1): p. 89-100. 54. Purvis, J.E., et al., p53 dynamics control cell fate. Science, 2012. 336(6087): p. 1440-4. 55. Alcorta, D.A., et al., Involvement of the cyclin-dependent kinase inhibitor p16 (INK4a) in replicative senescence of normal human fibroblasts. Proc Natl Acad Sci U S A, 1996. 93(24): p. 13742-7. 56. Hara, E., et al., Regulation of p16CDKN2 expression and its implications for cell immortalization and senescence. Mol Cell Biol, 1996. 16(3): p. 859-67. 57. Brenner, A.J., M.R. Stampfer, and C.M. Aldaz, Increased p16 expression with first senescence arrest in human mammary epithelial cells and extended growth capacity with p16 inactivation. Oncogene, 1998. 17(2): p. 199-205. 58. Ramirez, R.D., et al., Putative telomere-independent mechanisms of replicative aging reflect inadequate growth conditions. Genes Dev, 2001. 15(4): p. 398-403. 59. Krishnamurthy, J., et al., p16INK4a induces an age-dependent decline in islet regenerative potential. Nature, 2006. 443(7110): p. 453-7. 60. Molofsky, A.V., et al., Increasing p16INK4a expression decreases forebrain progenitors and neurogenesis during ageing. Nature, 2006. 443(7110): p. 448-52. 61. Yap, K.L., et al., Molecular interplay of the noncoding RNA ANRIL and methylated histone H3 lysine 27 by polycomb CBX7 in transcriptional silencing of INK4a. Mol Cell, 2010. 38(5): p. 662-74. 62. Holst, C.R., et al., Methylation of p16(INK4a) promoters occurs in vivo in histologically normal human mammary epithelia. Cancer Res, 2003. 63(7): p. 1596-601.

111

63. Bracken, A.P., et al., The Polycomb group proteins bind throughout the INK4A- ARF locus and are disassociated in senescent cells. Genes Dev, 2007. 21(5): p. 525-30. 64. Gonzalez, S., et al., Oncogenic activity of Cdc6 through repression of the INK4/ARF locus. Nature, 2006. 440(7084): p. 702-6. 65. Itahana, K., et al., Control of the replicative life span of human fibroblasts by p16 and the polycomb protein Bmi-1. Mol Cell Biol, 2003. 23(1): p. 389-401. 66. Jacobs, J.J., et al., The oncogene and Polycomb-group gene bmi-1 regulates cell proliferation and senescence through the ink4a locus. Nature, 1999. 397(6715): p. 164-8. 67. Kotake, Y., et al., Long non-coding RNA ANRIL is required for the PRC2 recruitment to and silencing of p15(INK4B) tumor suppressor gene. Oncogene, 2011. 30(16): p. 1956-62. 68. Kotake, Y., Y. Zeng, and Y. Xiong, DDB1-CUL4 and MLL1 mediate oncogene- induced p16INK4a activation. Cancer Res, 2009. 69(5): p. 1809-14. 69. Ohtani, N., et al., Opposing effects of Ets and Id proteins on p16INK4a expression during cellular senescence. Nature, 2001. 409(6823): p. 1067-70. 70. Witcher, M. and B.M. Emerson, Epigenetic silencing of the p16(INK4a) tumor suppressor is associated with loss of CTCF binding and a chromatin boundary. Mol Cell, 2009. 34(3): p. 271-84. 71. Narita, M., et al., Rb-mediated heterochromatin formation and silencing of E2F target genes during cellular senescence. Cell, 2003. 113(6): p. 703-16. 72. Zhang, R., et al., Formation of MacroH2A-containing senescence-associated heterochromatin foci and senescence driven by ASF1a and HIRA. Dev Cell, 2005. 8(1): p. 19-30. 73. Narita, M., et al., A novel role for high-mobility group a proteins in cellular senescence and heterochromatin formation. Cell, 2006. 126(3): p. 503-14. 74. Funayama, R., et al., Loss of linker histone H1 in cellular senescence. J Cell Biol, 2006. 175(6): p. 869-80. 75. Vergnes, L., et al., Lamin B1 is required for mouse development and nuclear integrity. Proc Natl Acad Sci U S A, 2004. 101(28): p. 10428-33. 76. Freund, A., et al., Lamin B1 loss is a senescence-associated biomarker. Mol Biol Cell, 2012. 23(11): p. 2066-75. 77. Shimi, T., et al., The role of nuclear lamin B1 in cell proliferation and senescence. Genes Dev, 2011. 25(24): p. 2579-93. 78. Cruickshanks, H.A., et al., Senescent cells harbour features of the cancer epigenome. Nat Cell Biol, 2013. 15(12): p. 1495-506.

112

79. Shah, P.P., et al., Lamin B1 depletion in senescent cells triggers large-scale changes in gene expression and the chromatin landscape. Genes Dev, 2013. 27(16): p. 1787-99. 80. Coppe, J.P., et al., Senescence-associated secretory phenotypes reveal cell- nonautonomous functions of oncogenic RAS and the p53 tumor suppressor. PLoS Biol, 2008. 6(12): p. 2853-68. 81. Shelton, D.N., et al., Microarray analysis of replicative senescence. Curr Biol, 1999. 9(17): p. 939-45. 82. Acosta, J.C., et al., Chemokine signaling via the CXCR2 receptor reinforces senescence. Cell, 2008. 133(6): p. 1006-18. 83. Kuilman, T. and D.S. Peeper, Senescence-messaging secretome: SMS-ing cellular stress. Nat Rev Cancer, 2009. 9(2): p. 81-94. 84. Coppe, J.P., et al., The senescence-associated secretory phenotype: the dark side of tumor suppression. Annu Rev Pathol, 2010. 5: p. 99-118. 85. Kuilman, T., et al., Oncogene-induced senescence relayed by an interleukin- dependent inflammatory network. Cell, 2008. 133(6): p. 1019-31. 86. Orjalo, A.V., et al., Cell surface-bound IL-1alpha is an upstream regulator of the senescence-associated IL-6/IL-8 cytokine network. Proc Natl Acad Sci U S A, 2009. 106(40): p. 17031-6. 87. Acosta, J.C., et al., A complex secretory program orchestrated by the inflammasome controls paracrine senescence. Nat Cell Biol, 2013. 15(8): p. 978- 90. 88. Chien, Y., et al., Control of the senescence-associated secretory phenotype by NF-kappaB promotes senescence and enhances chemosensitivity. Genes Dev, 2011. 25(20): p. 2125-36. 89. Young, A.R., et al., Autophagy mediates the mitotic senescence transition. Genes Dev, 2009. 23(7): p. 798-803. 90. Narita, M., et al., Spatial coupling of mTOR and autophagy augments secretory phenotypes. Science, 2011. 332(6032): p. 966-70. 91. Coppe, J.P., et al., Tumor suppressor and aging biomarker p16(INK4a) induces cellular senescence without the associated inflammatory secretory phenotype. J Biol Chem, 2011. 286(42): p. 36396-403. 92. Rodier, F., et al., Persistent DNA damage signalling triggers senescence- associated inflammatory cytokine secretion. Nat Cell Biol, 2009. 11(8): p. 973-9. 93. Kang, C., et al., The DNA damage response induces inflammation and senescence by inhibiting autophagy of GATA4. Science, 2015. 349(6255): p. aaa5612.

113

94. Yang, G., et al., The chemokine growth-regulated oncogene 1 (Gro-1) links RAS signaling to the senescence of stromal fibroblasts and ovarian tumorigenesis. Proc Natl Acad Sci U S A, 2006. 103(44): p. 16472-7. 95. Parrinello, S., et al., Stromal-epithelial interactions in aging and cancer: senescent fibroblasts alter epithelial cell differentiation. J Cell Sci, 2005. 118(Pt 3): p. 485-96. 96. Krtolica, A., et al., Senescent fibroblasts promote epithelial cell growth and tumorigenesis: a link between cancer and aging. Proc Natl Acad Sci U S A, 2001. 98(21): p. 12072-7. 97. Dilley, T.K., G.T. Bowden, and Q.M. Chen, Novel mechanisms of sublethal oxidant toxicity: induction of premature senescence in human fibroblasts confers tumor promoter activity. Exp Cell Res, 2003. 290(1): p. 38-48. 98. Xue, W., et al., Senescence and tumour clearance is triggered by p53 restoration in murine liver carcinomas. Nature, 2007. 445(7128): p. 656-60. 99. Krizhanovsky, V., et al., Senescence of activated stellate cells limits liver fibrosis. Cell, 2008. 134(4): p. 657-67. 100. Dimri, G.P., et al., A biomarker that identifies senescent human cells in culture and in aging skin in vivo. Proc Natl Acad Sci U S A, 1995. 92(20): p. 9363-7. 101. Lee, B.Y., et al., Senescence-associated beta-galactosidase is lysosomal beta- galactosidase. Aging Cell, 2006. 5(2): p. 187-95. 102. Kuerbitz, S.J., et al., Wild-type p53 is a cell cycle checkpoint determinant following irradiation. Proc Natl Acad Sci U S A, 1992. 89(16): p. 7491-5. 103. Lowe, S.W., et al., p53 is required for radiation-induced apoptosis in mouse thymocytes. Nature, 1993. 362(6423): p. 847-9. 104. Clarke, A.R., et al., Thymocyte apoptosis induced by p53-dependent and independent pathways. Nature, 1993. 362(6423): p. 849-52. 105. Hampel, B., et al., Differential regulation of apoptotic cell death in senescent human cells. Exp Gerontol, 2004. 39(11-12): p. 1713-21. 106. Gartel, A.L. and A.L. Tyner, The role of the cyclin-dependent kinase inhibitor p21 in apoptosis. Mol Cancer Ther, 2002. 1(8): p. 639-49. 107. Seoane, J., H.V. Le, and J. Massague, Myc suppression of the p21(Cip1) Cdk inhibitor influences the outcome of the p53 response to DNA damage. Nature, 2002. 419(6908): p. 729-34. 108. Chen, X., et al., p53 levels, functional domains, and DNA damage determine the extent of the apoptotic response of tumor cells. Genes Dev, 1996. 10(19): p. 2438- 51.

114

109. Zhao, R., et al., Analysis of p53-regulated gene expression patterns using oligonucleotide arrays. Genes Dev, 2000. 14(8): p. 981-93. 110. Oda, K., et al., p53AIP1, a potential mediator of p53-dependent apoptosis, and its regulation by Ser-46-phosphorylated p53. Cell, 2000. 102(6): p. 849-62. 111. Greenberg, R.A., et al., Short dysfunctional telomeres impair tumorigenesis in the INK4a(delta2/3) cancer-prone mouse. Cell, 1999. 97(4): p. 515-25. 112. Gonzalez-Suarez, E., et al., Telomerase-deficient mice with short telomeres are resistant to skin tumorigenesis. Nat Genet, 2000. 26(1): p. 114-7. 113. Qi, L., et al., Short telomeres and ataxia-telangiectasia mutated deficiency cooperatively increase telomere dysfunction and suppress tumorigenesis. Cancer Res, 2003. 63(23): p. 8188-96. 114. Braig, M., et al., Oncogene-induced senescence as an initial barrier in lymphoma development. Nature, 2005. 436(7051): p. 660-5. 115. Kang, T.W., et al., Senescence surveillance of pre-malignant hepatocytes limits liver cancer development. Nature, 2011. 479(7374): p. 547-51. 116. Perez-Mancera, P.A., A.R. Young, and M. Narita, Inside and out: the activities of senescence in cancer. Nat Rev Cancer, 2014. 14(8): p. 547-58. 117. Alimonti, A., et al., A novel type of cellular senescence that can be enhanced in mouse models and human tumor xenografts to suppress prostate tumorigenesis. J Clin Invest, 2010. 120(3): p. 681-93. 118. Herbig, U., et al., Cellular senescence in aging primates. Science, 2006. 311(5765): p. 1257. 119. Wang, C., et al., DNA damage response and cellular senescence in tissues of aging mice. Aging Cell, 2009. 8(3): p. 311-23. 120. van Deursen, J.M., The role of senescent cells in ageing. Nature, 2014. 509(7501): p. 439-46. 121. Lawless, C., et al., Quantitative assessment of markers for cell senescence. Exp Gerontol, 2010. 45(10): p. 772-8. 122. Janzen, V., et al., Stem-cell ageing modified by the cyclin-dependent kinase inhibitor p16INK4a. Nature, 2006. 443(7110): p. 421-6. 123. Sousa-Victor, P., et al., Geriatric muscle stem cells switch reversible quiescence into senescence. Nature, 2014. 506(7488): p. 316-21. 124. Freund, A., et al., Inflammatory networks during cellular senescence: causes and consequences. Trends Mol Med, 2010. 16(5): p. 238-46. 125. Baker, D.J., et al., Clearance of p16Ink4a-positive senescent cells delays ageing- associated disorders. Nature, 2011. 479(7372): p. 232-6.

115

126. Zhu, Y., et al., The Achilles' heel of senescent cells: from transcriptome to senolytic drugs. Aging Cell, 2015. 14(4): p. 644-58. 127. Jun, J.I. and L.F. Lau, The matricellular protein CCN1 induces fibroblast senescence and restricts fibrosis in cutaneous wound healing. Nat Cell Biol, 2010. 12(7): p. 676-85. 128. Munoz-Espin, D., et al., Programmed cell senescence during mammalian embryonic development. Cell, 2013. 155(5): p. 1104-18. 129. Storer, M., et al., Senescence is a developmental mechanism that contributes to embryonic growth and patterning. Cell, 2013. 155(5): p. 1119-30. 130. Campisi, J., Senescent cells, tumor suppression, and organismal aging: good citizens, bad neighbors. Cell, 2005. 120(4): p. 513-22. 131. Banito, A. and S.W. Lowe, A new development in senescence. Cell, 2013. 155(5): p. 977-8. 132. Smith, J.R. and R.G. Whitney, Intraclonal variation in proliferative potential of human diploid fibroblasts: stochastic mechanism for cellular aging. Science, 1980. 207(4426): p. 82-4. 133. Rual, J.F., et al., Human ORFeome version 1.1: a platform for reverse proteomics. Genome Res, 2004. 14(10B): p. 2128-35. 134. Sdek, P., et al., MDM2 promotes proteasome-dependent ubiquitin-independent degradation of retinoblastoma protein. Mol Cell, 2005. 20(5): p. 699-708. 135. Uchida, C., et al., Enhanced Mdm2 activity inhibits pRB function via ubiquitin- dependent degradation. Embo j, 2005. 24(1): p. 160-9. 136. Wu, P., H. Takai, and T. de Lange, Telomeric 3' overhangs derive from resection by Exo1 and Apollo and fill-in by POT1b-associated CST. Cell, 2012. 150(1): p. 39-52. 137. Negrini, S., et al., REST: an oncogene or a tumor suppressor? Trends Cell Biol, 2013. 23(6): p. 289-95. 138. Lu, T., et al., REST and stress resistance in ageing and Alzheimer's disease. Nature, 2014. 507(7493): p. 448-54. 139. Tang, H., et al., Replication factor C1, the large subunit of replication factor C, is proteolytically truncated in Hutchinson-Gilford progeria syndrome. Aging Cell, 2012. 11(2): p. 363-5. 140. Kenyon, C.J., The genetics of ageing. Nature, 2010. 464(7288): p. 504-12. 141. Eijkelenboom, A. and B.M. Burgering, FOXOs: signalling integrators for homeostasis maintenance. Nat Rev Mol Cell Biol, 2013. 14(2): p. 83-97. 142. Essers, M.A., et al., FOXO transcription factor activation by oxidative stress mediated by the small GTPase Ral and JNK. EMBO J, 2004. 23(24): p. 4802-12.

116

143. van der Horst, A., et al., FOXO4 transcriptional activity is regulated by monoubiquitination and USP7/HAUSP. Nat Cell Biol, 2006. 8(10): p. 1064-73. 144. Zou, L., D. Cortez, and S.J. Elledge, Regulation of ATR substrate selection by Rad17-dependent loading of Rad9 complexes onto chromatin. Genes Dev, 2002. 16(2): p. 198-208. 145. Klingelhutz, A.J., S.A. Foster, and J.K. McDougall, Telomerase activation by the E6 gene product of human papillomavirus type 16. Nature, 1996. 380(6569): p. 79-82. 146. Jones, D.L., R.M. Alani, and K. Munger, The human papillomavirus E7 oncoprotein can uncouple cellular differentiation and proliferation in human keratinocytes by abrogating p21Cip1-mediated inhibition of cdk2. Genes Dev, 1997. 11(16): p. 2101-11. 147. Luo, J., et al., A genome-wide RNAi screen identifies multiple synthetic lethal interactions with the Ras oncogene. Cell, 2009. 137(5): p. 835-48. 148. Zhang, S., et al., RhoA regulates G1-S progression of gastric cancer cells by modulation of multiple INK4 family tumor suppressors. Mol Cancer Res, 2009. 7(4): p. 570-80. 149. Zou, Q., et al., USP15 stabilizes MDM2 to mediate cancer-cell survival and inhibit antitumor T cell responses. Nat Immunol, 2014. 15(6): p. 562-70. 150. Li, M.Z. and S.J. Elledge, Harnessing homologous recombination in vitro to generate recombinant DNA via SLIC. Nat Methods, 2007. 4(3): p. 251-6. 151. Tang, P., et al., Increased expression of DLX2 correlates with advanced stage of gastric adenocarcinoma. World J Gastroenterol, 2013. 19(17): p. 2697-703. 152. Yan, Z.H., et al., Upregulation of DLX2 confers a poor prognosis in glioblastoma patients by inducing a proliferative phenotype. Curr Mol Med, 2013. 13(3): p. 438-45. 153. Morini, M., et al., Mutually exclusive expression of DLX2 and DLX5/6 is associated with the metastatic potential of the human breast cancer cell line MDA-MB-231. BMC Cancer, 2010. 10: p. 649. 154. Takai, H., et al., Tel2 structure and function in the Hsp90-dependent maturation of mTOR and ATR complexes. Genes Dev, 2010. 24(18): p. 2019-30. 155. Takai, H., et al., Tel2 regulates the stability of PI3K-related protein kinases. Cell, 2007. 131(7): p. 1248-59. 156. Hurov, K.E., C. Cotta-Ramusino, and S.J. Elledge, A genetic screen identifies the Triple T complex required for DNA damage signaling and ATM and ATR stability. Genes Dev, 2010. 24(17): p. 1939-50. 157. Yilmaz, M., et al., Transcription factor Dlx2 protects from TGFbeta-induced cell- cycle arrest and apoptosis. Embo j, 2011. 30(21): p. 4489-99.

117

158. Horejsi, Z., et al., CK2 phospho-dependent binding of R2TP complex to TEL2 is essential for mTOR and SMG1 stability. Mol Cell, 2010. 39(6): p. 839-50. 159. Rao, F., et al., Inositol pyrophosphates mediate the DNA-PK/ATM-p53 cell death pathway by regulating CK2 phosphorylation of Tti1/Tel2. Mol Cell, 2014. 54(1): p. 119-32. 160. Burrows, A.E., A. Smogorzewska, and S.J. Elledge, Polybromo-associated BRG1-associated factor components BRD7 and BAF180 are critical regulators of p53 required for induction of replicative senescence. Proc Natl Acad Sci U S A, 2010. 107(32): p. 14280-5. 161. Drost, J., et al., BRD7 is a candidate tumour suppressor gene required for p53 function. Nat Cell Biol, 2010. 12(4): p. 380-9. 162. Cerami, E., et al., The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov, 2012. 2(5): p. 401-4. 163. Gao, J., et al., Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal, 2013. 6(269): p. pl1. 164. Comprehensive molecular portraits of human breast tumours. Nature, 2012. 490(7418): p. 61-70. 165. Ciriello, G., et al., Comprehensive Molecular Portraits of Invasive Lobular Breast Cancer. Cell, 2015. 163(2): p. 506-19. 166. Fernandez-Saiz, V., et al., SCFFbxo9 and CK2 direct the cellular response to growth factor withdrawal via Tel2/Tti1 degradation and promote survival in multiple myeloma. Nat Cell Biol, 2013. 15(1): p. 72-81. 167. Diamond, E., et al., Functional interactions between Dlx2 and lymphoid enhancer factor regulate Msx2. Nucleic Acids Res, 2006. 34(20): p. 5951-65.

118