Cell cycle checkpoints: reversible when possible, irreversible when needed

Transcription

1 Cell cycle checkpoints: reversible when possible, irreversible when needed Lenno Krenning

2 ISBN / EAN: Printed by: Gildeprint Copyright 215 Lenno Krenning Cover artwork: Carla Krenning 215 About the cover: The cover shows a schematic representation of the cell cycle. In it a barrier is drawn, which represents the G2 checkpoint. This checkpoint has been the major focus of the work presented in this thesis. The differentially colored circles in the background and at the back of this thesis depict individual cells. Their different colors represent the heterogeneous behaviour that individual cells display upon a similar stimulus (such as DNA damage).

3 Cell cycle checkpoints: reversible when possible, irreversible when needed Celcyclus controlepunten: reversibel waar mogelijk, irreversibel indien nodig (met een samenvatting in het Nederlands) Proefschrift ter verkrijging van de graad doctor aan de Universiteit Utrecht op gezag van de rector magnificus, prof. dr. G.J. van der Zwaan, ingevolge het besluit van het college voor promoties in het openbaar te verdedigen op dinsdag 1 november 215 des middags te 2.3 uur door Lenno Krenning geboren op 2 juni 1984 te Leiden

4 Promotor: prof. dr. R.H. Medema

5 Table of contents Chapter 1 General introduction Thesis outline 7 Chapter 2 The same, only different - DNA damage checkpoints and their reversal throughout the cell cycle (review) 19 Chapter 3 Transient activation of p53 in G2 phase is sufficient to induce senescence 35 Addendum 57 Chapter 4 DNA lesions trigger a cell cycle exit in prophase 67 Chapter 5 The mitotic checkpoint complex protects against DNA damage-induced activation of the APC/C Cdh1 85 Chapter 6 Time-dependent activation of p53 in mitosis controls the mitotic timer checkpoint 97 Chapter 7 General discussion 113 Appendix References Nederlandse samenvatting Curriculum Vitae Publication list Dankwoord Table of contents 5

6

7 Chapter 1 1 General introduction Lenno Krenning and René H. Medema Division of Cell Biology I, The Netherlands Cancer Institute Plesmanlaan 121, 166 CX Amsterdam, The Netherlands General introduction 7

8 1 Summary Faithfull duplication and segregation of the genome is the most fundamental task of the cell cycle. Progression through the cell cycle is driven by the activity of Cyclin dependent kinases (Cdks). Different Cdks partake in each cell cycle transition, and their activity is strictly regulated during the cell cycle. In this chapter, we highlight processes underlying transitions from one cell cycle phase to the other, and discuss how the Cdk activity that drives these transitions is regulated. 8 Chapter 1

9 Introduction to the cell cycle The different phases of the cell cycle The cell division cycle constitutes the basal unit of life. It drives reproduction of unicellular organisms, and, following fertilization, it allows the growth and homeostasis of the entire body of multicellular organisms. In this sense, the most basic task of the cell cycle is the generation of two genetically identical daughter cells. This is achieved through faithful duplication of the genome during the synthesis (S) phase, and equal distribution of the genomic material to two daughter cells in mitosis (Figure 1). Two additional gap phases separate mitosis from S phase, and vice versa (G1 and G2 phase, respectively). In case of undesired conditions, progression through the cell cycle may (temporarily) stop during these gap phases and in mitosis, (Figure 1, red dotted lines). When these conditions are resolved, cells can resume cell cycle progression (1, 2). The transitions from G1 into S phase, and from G2 into mitosis are initiated in response to increasing amounts of Cdk activity (which will be discussed later). In addition, Cdk activity regulates the transcriptional programs required to properly execute each cell cycle phase. Progression through the cell cycle is driven by Cdk4, Cdk6, Cdk2 and Cdk1, themselves subject to multiple layers of regulation. Aberrations in these regulatory mechanisms can promote unscheduled cell growth, and are often, if not always, present in human cancer (3). 1 Cyclin-Cdk activity is heavily regulated Cdk activity is regulated via multiple mechanisms. First, binding of a regulatory subunit, called Cyclin, changes the conformation of the Cdk subunit such that it becomes a functional kinase (Figure 2A). The resulting Cyclin-Cdk complex is, however, inactive, and needs to be phosphorylated by the Cdk-activating kinase (CAK). CAK phosphorylates Cdk within a large loop of the protein, the so called t-loop. This loop blocks substrate binding to Cdks, and this block is alleviated though phosphorylation-dependent repositioning of the t-loop. In addition to t-loop repositioning, CAK-mediated phosphorylation stabilizes the Cyclin-Cdk interaction (Figure 2A). CAK activity is constant throughout the cell cycle. Therefore, in order to restrict the activation of specific Cyclin-Cdk complexes to certain cell cycle stages, the expression of Cyclins is strictly limited to specific moments during the cell cycle. In fact, cyclins received their name due to their intermittent presence during the cell cycle G2 M Figure 1. Schematic representation of the cell cycle The cell cycle is divided into separate phases. In the first gap phase (G1) a cell prepares for DNA replication, which occurs during the synthesis (S) phase. Then, the cell enters another gap phase (G2) during which the cell grows and prepares for division during mitosis. In response to unfavorable conditions cell cycle progression can be halted until necessary in G1, G2 or mitosis. S G1 (Figure 2B). This tight control over Cyclin abundance is achieved through regulated expression as well as regulated degradation of Cyclins. D- and E-type Cyclins are inherently unstable, and their abundance is primarily regulated through transcription. In contrast, both Cyclin A2 and Cyclin B1 are stable throughout most of interphase. Therefore, their abundance must be controlled through regulated expression as well as degradation (4, 5). Additional regulation of Cdk activity comes in the form of phosphorylation of the assembled Cyclin-Cdk complex. Wee1- and Myt1-dependent phosphorylation of Cdk1 (on Y15 and T14, respectively) inhibits the kinase activity of the fully assembled Cyclin-Cdk complex. Activation of these inhibit- General introduction 9

10 1 ed complexes requires dephosphorylation of both residues (Figure 2A). This is done by the Cdc25-family of dual-specificity phosphatases, which can dephosphorylate both Y15 and T14, thereby activating the Cyclin-Cdk complexes. Similarly, phosphorylation of corresponding regions within Cdk2, as well as Cdk4 and Cdk6, control Cdk activity, and are removed by Cdc25 phosphatases (4-7). Finally, in addition to the mechanisms described above, Cdk activity can be inhibited through expression of endogenous Cdk inhibitor proteins (CKI). The two families of CKIs are the Cip/Kip family of proteins, encoding p21 Cip1, p27 Kip1 and p57 Kip2, and the INK4 family, encoding p16 INK4a, p15 INK4b, p18 INK4c and p19 INK4d. CKIs of the Cip/Kip family may bind to all Cyclin-Cdk complexes, which is generally thought to inhibit the kinase activity of the bound Cyclin-Cdk complex. In contrast, members of the INK4 family only bind to Cdk4 or Cdk6, preventing their interaction with the D-type Cyclins. In addition, in complex with INK4 proteins, members of the Cip/Kip family can no longer bind to Cdk4 and Cdk6, and are therefore free to bind and inhibit Cdk2 and Cdk1. This way, INK4 proteins indirectly contribute to the inhibition of Cdk2 (8, 9). Transcriptional regulation underlies the G1 to S transition Entry into a new cell cycle requires the activation of a transcriptional program that is under control of the E2F transcription factors. E2F targets include genes involved in cell cycle control and DNA replication, but also genes involved in cell cycle checkpoints and DNA repair. The activity of E2F transcription factors is under tight control of the pocket protein family (prb, p17 and p13). In the absence of Cdk activity, during early G1, the pocket proteins bind E2Fs and transcription of E2F target genes is inhibited. Cdk-dependent phosphorylation of pocket proteins disrupts the interaction with E2Fs, leading to induction of E2F target genes and progression from G1 into S phase (1). In early G1, mitogenic signaling results in the expression of D-type Cyclins, which preferentially bind and activate Cdk4 and Cdk6 (Figure 3). The resulting Cdk4/6 activity inhibits the pocket proteins, allowing the initiation of E2F-dependent transcription of many genes, of which E-type Cyclins are among the earliest to be expressed. E-type Cyclins preferentially bind and activate Cdk2, further contributing to pocket protein inhibition, E2F target gene induction and entry into S phase (Figures 2B and 3). Activation of this positive feedback loop ensures passage through the so-called restriction point, a point defined as the moment after which cells no longer require continued mitogenic signaling to complete the cell cycle. Once in S phase, levels of E-type Cyclins decline, and from here on E2F-dependent induction of Cyclin A2 maintains Cdk2 activity until mitosis (Figures 2B and 3) (3, 11). Feedback mechanisms regulate the transition from G2 into mitosis Following the completion of DNA replication, the cell enters G2 phase. Progression through G2 phase requires the activity of Cyclin A2-Cdk2, whereas commitment to mitosis requires Cdk1 activity (12). Similar to the Cdk-dependent transcription of E-type Cyclins in G1, the expression of Cyclin B1 in late S and G2 phase is controlled by Cdk activity (Figure 2B). Three transcription factors (FoxM1, B-Myb and NF-Y) contribute to the expression of Cyclin B1, and are activated by Cyclin A2-Cdk2 (Figure 3) (12). Mere accumulation of Cyclin B1 during G2 does not result in activation of Cdk1 (13) and additional mechanisms contribute to the regulation of Cyclin B1-Cdk1 activity and mitotic entry. As described before, Y15 and/or T14 phosphorylation of the assembled Cyclin-Cdk complex inhibits its kinase activity (Figure 2A) (4, 5). Myt1-dependent inhibitory phosphorylation of Cdk1 (on T14) seems to be directly coupled to the activating phosphorylation of Cdk1 (on T161) by CAK, thereby preventing premature activation of the assembled Cyclin B1-Cdk1 complex (14). Subsequent activation of Cyclin B1-Cdk1 at the onset of mitosis is achieved through several positive and negative feedback mechanisms. Most notably, Cdk1 can inactivate both Wee1 and Myt1, while it can stimulate the activity of the Cdc25 phosphatases, resulting in switch-like activation of Cyclin B1- Cdk1 (Figure 3) (12, 15). In addition, Cdk1- and Plk1-dependent degradation of Wee1, as well as Plk1-dependent inhibition of Myt1 and activation of Cdc25 phosphatases contributes to the full activation of Cdk1 (Figures 2A and 3). Thus, once Cdk1 activity reaches a threshold level, it will inhibit its inhibitors and stimulate its activator, leading to switch-like activation of Cdk1 (12). 1 Chapter 1

11 A Protein synthesis Protein degradation 1 Cyclin Dephosphorylation by Cdc25 Cyclin Cyclin Cyclin Cdk Cdk Cdk Cdk Cyclin binding Phosphorylation by CAK Phosphorylation by Wee1 & Myt1 B Cyclin D Cdk4/6 Cyclin E Cdk2 Cyclin A Cdk2 Cyclin B Cdk1 G1 S G2 M Figure 2. Step wise activation of Cyclin dependent kinases (A) Activation of Cyclin-dependent kinases (Cdks) requires binding of the activating Cyclin subunit, and subsequent phosphorylation in the t-loop by Cdk-activiting kinase (CAK). Assembled Cyclin B1-Cdk1 complexes can be inhibited through phosphorylations by Wee1 and Myt1 (on Y15 and T14, respectively), which can be reverted by Cdc25-dependent dephosphorylation. Similarly, Cdk2, Cdk4 and Cdk6 containing complexes can be phosphorylated at corresponding residues, which are removed by Cdc25 phosphatases. (B) Different Cyclins activate different Cdk-counterparts and are expressed at different moments throughout the cell cycle. Growth signals CycD Cdk4/6 CycA Cdk2 FoxM1 B-Myb NF-Y CycB Plk1 AurA Cdc25 Bora prb E2F Cdc25 AurA Bora CycE Cdk2 CycB Cdk1 Myt1 Wee1 Plk1 G1 S G2 M Figure 3. Schematic representation of the regulatory mechanisms underlying cell cycle transitions from G1 into S phase, and from G2 into mitosis Mitogenic stimulation activates a signaling cascade that ultimately results in the activation of Cdk2, which is required for the transition from G1 to S phase. The transition from G2 into mitosis requires Cyclin B1-Cdk1 activity, which is strictly regulated. See main text for more details. General introduction 11

12 1 G2 Prophase Prometaphase Metaphase Anaphase A Anaphase B Telophase Figure 4. Mitosis is subdivided into six different phases The centrosomes separate and chromosomes begin to condense during prophase. As the nuclear envelope is disassembled (not shown) cells enter into prometaphase, during which microtubules can start to interact with chromosomes in order to properly attach all duplicated sister chromatids to the mitotic spindle at metaphase. After proper attachment, chromosome segregation is initiated and sister chromatids are pulled towards opposite sides of the cell during anaphase A. In anaphase B the membrane starts to invaginate, in order to physically separate the two newly formed daughter cells in telophase. Simultaneously, chromosomes decondense and the nuclear envelope reforms. As mentioned before, G2 progression requires Cyclin A2-Cdk2 activity, which contributes to the activation of Cyclin B1-Cdk1 both directly and indirectly (12). In G2, Cyclin A2-dependent regulation of Wee1 contributes to the activation of Cyclin B1-Cdk1 directly (16, 17). In addition, Cyclin A2-Cdk2 stimulates FoxM1-dependent transcription of Plk1 and its activators Aurora A and Bora, the Cdc25 phosphatases, as well as Cyclin B1 itself (Figure 3). The resulting slow increase in Cdk1 activity eventually reaches a threshold at which the positive feed forward loop in the Cdk1-regulatory system overcomes the inhibitory function of Wee1 and Myt1, thus allowing the switch-like activation of Cdk1 at the end of G2 phase which drives entry into mitosis (12). Progression through mitosis and beyond Mitosis is divided into six distinct phases. During prophase chromosomes start to condense and spindle poles are separated to facilitate the formation of a bipolar spindle during prometaphase. At the end of prophase, nuclear envelope breakdown constitutes commitment to mitosis, and the cell enters prometaphase. During prometaphase, microtubules originating from the opposing spindle poles start to interact with individual chromosomes, through their kinetochores, facilitating the alignment of all chromosomes at the cell equator at metaphase. Proper attachment of all sister chromatids to microtubules originating form opposite poles initiates anaphase A, during which the duplicated chromosome pairs are pulled apart toward opposite sides of the cell. In anaphase B, sister chromatid separation continues and invaginations in the membrane start to form in between the separating chromosomes. Then, during telophase, the nuclear envelopes are reformed around the chromosomes, and the cellular membrane between the separated chromosome packs ingresses further in order to physically separate the two new daughter cells (Figure 4) (18). Once a cell has committed to mitosis, it needs to prevent the onset of chromosome segregation until the kinetochores of all sister chromatids are properly attached to the mitotic spindle. This way, cells can accurately segregate the replicated genome during mitosis. Sister chromatid segregation is initiated by the anaphase promoting complex / cyclosome (APC/C) together with its coactivator Cdc2 (APC/C Cdc2 ). Therefore, preventing premature chromosome segregation requires cells to inhibit APC/C Cdc2 activity until all chromosomes have attached to the spindle correctly. This is achieved by the spindle assembly checkpoint (SAC), which senses the microtubule-kinetochore attachment status. In the presence of unattached kinetochores the SAC generates a Cdc2 inhibitory complex known as the mitotic checkpoint complex (MCC). Upon proper attachment of the last chromosome to the mitotic spindle, MCC disassembly releases APC/C Cdc2 activity, which subsequently targets securin and Cyclin B1 for destruction, allowing progression beyond metaphase. Degradation of securin releases separase activity, which cleaves the cohesin rings that hold together the replicated sister chromatids. Thus, cohesin cleavage allows the physical separation of sister chromatids. In parallel, the decrease in Cdk1 activity that results from Cyclin B1 destruction allows the nuclear envelopes to reform around the segregated packs of DNA, and the physical separation of the two newly formed daughter cells during mitotic exit (19, 2). 12 Chapter 1

13 APC/C-dependent proteolysis continues through G1, and is inhibited as cells enter into S phase. Importantly, unscheduled APC/C activity, in interphase or mitosis, has profound implications for the proper execution of the mitotic cell cycle. 1 The anaphase promoting complex/cyclosome functions at various stages of the cell cycle Regulation of APC/C activity throughout the cell cycle The APC/C is a multiprotein subunit E3 ligase that targets many proteins for degradation. In order for the APC/C to ubiquitinate its substrates it requires the association of one of its two coactivator proteins, Cdc2 or Cdh1 (21). During normal cell cycle progression, the APC/C is inhibited from the start of S phase until the moment cells reach metaphase. In metaphase, APC/C Cdc2 promotes degradation of securin and Cyclin B1, to allow progression into anaphase. Once Cyclin B1 is degraded, the second coactivator, Cdh1, can promote APC/C-dependent degradation of many other mitotic regulators. Cdh1 continues to activate the APC/C in late anaphase and throughout G1 until cells enter S phase (Figure 5) (22-24). The two APC/C coactivators Cdc2 and Cdh1 are strictly regulated The sequential activation of the APC/C by its two cofactors is, in part, the result of opposing effects of Cdk activity on the ability of the coactivators to interact with the APC/C. Assembly of APC/C Cdc2 requires the Cdk1-dependent phosphorylation of many APC/C core subunits (25-27), thus limiting Cdc2-dependent APC/C activity to mitosis, when Cdk1 activity is high (Figure 5). How then, is the APC/C Cdc2 inhibited during the early stages of mitosis, to allow for proper chromosome attachment? This is achieved through the MCC, which is generated as long as unattached kinetochores are present. The MCC consists of Mad2, BubR1 and Bub3, and can bind to Cdc2, thereby inhibiting Cdc2-dependent APC/ C activity (28-31). Upon correct attachment of the last chromosome to the mitotic spindle, MCC formation is stalled. Since the MCC is continuously disassembled, this allows rapid activation of the APC/C Cdc2 at Cdh1 Cdc2 Cdh1 Mad2l2 Cdh1 APC/C Cdh1 APC/C Emi1 APC/C Emi1 MCC APC/C Cdc2 APC/C Cdc2 APC/C Cdh1 G1 S G2 Prophase Prometaphase Metaphase Anaphase Cdk4 Emi1 Cdk1 Cdk2 SAC Cdc14 Mad2l2 Figure 5. Periodical activity of the anaphase promoting complex/cyclosome during the cell cycle In G1, with low Cdk activity, Cdh1 can bind and activate the anaphase promoting complex/cyclosome (APC/C). As cells enter S phase, Cdk2-dependent phosphorylation of Cdh1 prevents its interaction with the APC/C. In addition, expression of Emi1 prevents coactivator binding. Finally, the lack of APC/C phosphorylation occludes Cdc2 binding, therefore the APC/C is completely inhibited from S phase onwards. In prophase, Emi1 is degraded, and inactivation of the APC/C depends on Cdk activity. In prometaphase, when core subunits of the APC/C are phosphorylated by Cdk1, Cdc2 binds to the APC/C. However, APC/CCdc2 activity is inhibited due to spindle assembly checkpoint-dependent production of the mitotic checkpoint complex (MCC). Continued Cdk-dependent phosphorylation continues to inactivate Cdh1. At the metaphase-to-anaphase transition, when the MCC dissociates from the APC/C, APC/ CCdc2 drives mitotic progression through the destruction of securin and Cyclin B1. The resulting loss of Cdk activity allows Cdh1 dephosphorylation and activation of the APC/CCdh1 during anaphase and in G1. G1 General introduction 13

14 1 metaphase (Figure 5) (32, 33). Meanwhile, Cdk-dependent phosphorylation of Cdh1 prevents its interaction with the APC/C (22, 24, 25, 34, 35), and therefore APC/C Cdh1 activity is limited to episodes of low Cdk activity (Figure 5). The inhibition of Cdk1 resulting from APC/C Cdc2 -mediated destruction of Cyclin B1, and dephosphorylation of Cdh1 by the Cdc14 phosphatase allow for the APC/C Cdh1 to become active (Figure 5) (35-37). But how then, is Cdh1-dependent APC/C activity inhibited in late G1, and why is this important? Inhibition of APC/C Cdh1 is required for S phase entry (38), and this is achieved through inhibitory phosphorylation of Cdh1, as well as expression of the APC/C Cdh1 -inhibitor protein early mitotic inhibitor (Emi1). Additionally, APC/C Cdh1 inhibition during S and G2 phases is required to couple S phase to mitotic entry. In the absence of Cdh1 inhibition, cells fail to accumulate Cyclins required for mitosis, will undergo rereplication, and may exit the cell cycle (34, 38-41). As mentioned before, D-type Cyclins activate Cdk4/6 upon mitogenic stimulation. In turn, Cdk4-dependent phosphorylation of Cdh1 and prb in G1 stimulates cell cycle entry (42). This may be in part caused by the inability of phosphorylated Cdh1 to activate the APC/C. Additionally, Cdk4/6-dependent expression of Cyclin E activates Cdk2, which may contribute to the phosphorylation and inhibition of Cdh1 in late G1 and S phase (Figure 5) (43). However, full APC/C Cdh1 -inhibition and prevention of re-replication requires the expression of E2F target genes including Cyclin A2 and Emi1 (34, 44-46). In late G1, Emi1 binding inhibits APC/C activity through the prevention of substrate binding and ubiquitin-chain elongation (Figure 5). (47, 48). Therefore, the expression of Emi1 at late G1 helps to stabilize the APC/C Cdh1 target Cyclin A2, which is required for S phase entry (45, 49). In addition, another E2F target gene, the deubiquitinating enzyme USP37 may counteract APC/C Cdh1 -dependent Cyclin A2-ubiquitination, thereby stabilizing Cyclin A2 (5). Interestingly, USP37 itself is activated by Cdk2 activity. Therefore, E2F-dependent expression of Cyclin A2 and USP37 constitute a positive feedback loop that may cooperate with Emi1 to inhibit APC/C Cdh1 -activity in late G1 and drive entry into S phase. Finally, in prophase, Emi1 is targeted for degradation by SCF βtrcp -mediated ubiquitination (41, 51-53). At this moment, binding of Cdh1 to the Mad2-related protein Mad2l2 may prevent Cdh1 association to the APC/C until Mad2l2 is degraded by APC/C Cdc2 following metaphase (Figure 5) (54, 55), allowing Cdh1 to activate the APC/C as cells exit from mitosis into G1. A B ATM ATM R N ATM M ATM ATM Rad17 H2AX H2AX Rad17 R N M H2AX H2AX C ATM ATM D ATM ATM ATM Rad17 R N M γh2ax H2AX ATM ATM ATM R N M ATM Mdc1 Rad17 R N M γh2ax γh2ax Figure 6. DNA double strand break dependent activation of ATM (A) In the presence of intact DNA both the MRN complex and ATM are inactive. (B) Upon induction of a DNA double strand break (DSB) the MRN complex binds to the DNA end, which is facilitated by Rad17 binding to Nbs1. MRN bound to the DNA end can then activate ATM. (C) ATM phosphorylates Rad17, thereby strengthening the MRN- Rad17 interaction. At the same time, ATM phosphorylates H2AX, creating γh2ax. (D) γh2ax is bound by Mdc1, which, in turn, results in the accumulation of more MRN complexes flanking the break site, and additional activation of ATM and spreading of γh2ax flanking the break site. 14 Chapter 1

15 DNA damage and the cell cycle A DNA double strand break (DSB) is generated when the two complementary strands of the DNA double helix are broken simultaneously. A DSB initiates a plethora of reactions, collectively known as the DNA damage response (DDR), including mechanisms contributing to DNA repair and cell cycle arrest (56). As DDR activation requires the detection of a DSB, we will first discuss how a cell detects DNA breaks to initiate DDR signaling. 1 Recognition of DNA damage and activation of the DNA damage response A DSB causes rapid activation of the DNA damage checkpoint kinase Ataxia telangiectasia mutated (ATM) (57). ATM activation requires detection of the DNA break by the Mre11-Rad5-Nbs1 (MRN) complex (58, 59), which rapidly localizes to sites of DSBs in a Rad17-dependent manner (6, 61). There, the MRN complex recruits and activates ATM (58, 59, 62). From here on, ATM is involved in two feedback mechanisms that allow rapid spreading and amplification of the DNA damage signaling cascade. First, ATM-dependent phosphorylation of Rad17 strengthens the MRN-Rad17 interaction, thereby amplifying ATM activation. Second, ATM phosphorylates the histone variant H2AX, creating γh2ax, flanking the DSB (63, 64). MDC1 localizes to γh2ax and, in turn, recruits more MRN complexes as well as ATM to sites flanking a DSB (65, 66). Thus, H2AX phosphorylation results in MDC1-dependent amplification of ATM activation. The initial targeting of the MRN complex to a DSB site by Rad17 does not require MDC1 (61). In agreement with this, H2AX is dispensable for the initial recognition of DNA breaks, but is required for a functional DDR (67, 68). Following ATM activation, the ataxia telangiectasia and Rad3-related (ATR) kinase is activated during the later stages of the response to DSBs. ATR is activated in by RPA-coated single-stranded DNA stretches (69), and is essential for the maintenance of genome integrity (7). Activation of ATR in response to DSBs requires ATM-dependent resection of the broken DNA end, which causes an ATM-to-ATR switch in the DDR (71-75). Following their initial activation, signaling events downstream of ATM and ATR relay the DNA damage signal from the DSB site throughout the rest of the cell. At the crossroads of the cell cycle and the DNA damage response ATM and ATR relay the DNA damage signal to numerous target proteins in order to initiate a variety of processes, including a cell cycle arrest. Of these target proteins, the effector kinases Chk2 and Chk1 (activated by ATM and ATR, respectively) are required to install an ATM/ATR-dependent cell cycle arrest (76). Here, we will briefly focus on two essential mechanisms to halt the cell cycle in G1 and G2, respectively. However, there are many additional signaling events that contribute to cell cycle arrest in response to a DSB (discussed in chapter 2). Both ATM and its downstream target Chk2 can phosphorylate and activate p53 (77-81), which is required for the induction of p21 Cip1/Waf1 and to arrest cells in G1 (82-84). Additionally, p53 and p21 are required for the maintenance, but not initiation, of a G2 arrest (85). In addition to p53 regulation, Chk2, but also Chk1, regulate the Cdc25 phosphatases in response to DSBs, and this is required to efficiently arrest cells in G1 and G2 phases of the cell cycle (Reviewed in 86). Cdc25A is degraded as a result of ATM and Chk2 activation in response to DSBs, which is required for an arrest in G1 and G2 (87, 88). In addition, Cdc25C and Cdc25B are phosphorylated by both Chk2 and Chk1, resulting in binding and inactivation (89-91). More recently it was shown that ATM and ATR activation resulting from telomere deprotection causes degradation of Cdc25A and Cdc25C, which is required to arrest cells with deprotected telomeres in G2 (92). General introduction 15

16 1 Thesis outline Cell cycle checkpoints are reversible in nature, and can prevent progression into the next cell cycle phase if needed. In the case of DNA damage, cells can prevent progression from G1 into S phase, and from G2 into mitosis in the presence of DNA double strand breaks. Following DNA repair, these checkpoints can be silenced, allowing further progression through the cell cycle. In chapter 2 we review recent literature regarding activation and silencing of the DNA damage response. We describe how checkpoint activation and its reversal are tuned to each phase of the cell cycle, despite shared mechanisms of DNA damage detection. We also describe mechanisms that contribute to the maintenance of checkpoint reversibility following DNA damage in G2 phase, which is the main topic of this thesis. Chapter 3 addresses the reversibility of cell cycle checkpoints. We discovered that a DNA damage-induced arrest in G1 remains reversible for much longer compared to an arrest in G2. This finding led us to identify the sequence of events leading to an irreversible cell cycle withdrawal following excessive DNA damage in G2 phase. In addition, in the addendum we show that the decision to withdraw from the cell cycle in G2 phase is not merely determined by the amount of DNA breaks. In chapter 4 we describe how the normal wiring of the cell cycle machinery in prophase results in a loss of DNA damage checkpoint reversibility. As a result, cells are hypersensitive to DNA damage during prophase, causing cells to exit the cell cycle from prophase in response to relatively mild DNA damage. We show that this mechanism helps to limit the proliferation of cells that have undergone DNA replication stress. Chapter 5 describes the identification of a novel mechanism contributing to the maintenance of DNA damage checkpoint reversibility in G2 phase. Finally, in chapter 6, we confirm the existence of a mitotic timer that can limit the proliferation of daughter cells when the completion of mother cell mitosis exceeds a certain time window. We show that time-dependent accumulation of the tumor suppressor p53 in mitosis precedes a cell cycle arrest in G1, and find that mitotic cells lack Mdm2. This suggests that the mitotic timer is controlled through time-dependent accumulation of p53 in mitosis. 16 Chapter 1

17 General introduction 17 1

18

19 Chapter 2 The same only different - DNA damage checkpoints and their reversal throughout the cell cycle 2 Indra A. Shaltiel 1, Lenno Krenning 1, Wytse Bruinsma 2 and René H. Medema 3 Division of Cell Biology I, The Netherlands Cancer Institute Plesmanlaan 121, 166 CX Amsterdam, The Netherlands 1 ) These authors contributed equally to this work 2 ) Present address: Molecular Biology Program, Memoral Sloan-Kettering Cancer Center New York, NY, 165, USA 3 ) Correspondence: r.medema@nki.nl Adapted from Journal of Cell Science. 215 Feb 15; 128(4):67-2 DNA damage checkpoints and their reversal throughout the cell cylce 19

20 Abstract 2 Cell cycle checkpoints activated by DNA double-strand breaks (DSBs) are essential for the maintenance of the genomic integrity of proliferating cells. Following DNA damage, cells must detect the break and either transiently block cell cycle progression, to allow time for repair, or exit the cell cycle. Reversal of a DNA damage-induced checkpoint not only requires the repair of these lesions, but a cell must also prevent permanent exit from the cell cycle and actively terminate checkpoint signalling to allow cell cycle progression to resume. It is becoming increasingly clear that despite the shared mechanisms of DNA damage detection throughout the cell cycle, the checkpoint and its reversal are precisely tuned to each cell cycle phase. Furthermore, recent findings challenge the dogmatic view that complete repair is a precondition for cell cycle resumption. In this Commentary, we highlight cell-cycle-dependent differences in checkpoint signalling and recovery after a DNA DSB, and summarise the molecular mechanisms that underlie the reversal of DNA damage checkpoints, before discussing when and how cell fate decisions after a DSB are made. 2 Chapter 2

21 Introduction The current dogma is that the DNA damage response (DDR) affects proliferation such that the DDR can reversibly arrest cell cycle progression to allow time for DNA repair, and, upon completion of DNA repair, the DDR is turned off to allow cell cycle resumption (Box 1). It is becoming increasingly clear that the DDR not only communicates to the cell cycle machinery, but that the cell cycle machinery is also involved in regulating the DDR, such that a similar DNA break activates different arms of the DDR depending on the cell cycle phase in which the damage occurs. In turn, silencing of the DDR is different in each cell cycle phase. These cell-cycle-dependent changes are intimately intertwined with double-strand break (DSB) end-processing and the usage of non-homologous end-joining (NHEJ) and homologous recombination (HR) repair pathways. In this Commentary, we address differences in the response to DSBs throughout the cell cycle in metazoan cells, highlighting which arms of the DDR function in the different phases. We also discuss how DNA damage-induced checkpoints are silenced in the different cell cycle phases, how cells undergo checkpoint recovery after checkpoint silencing and how cells, when arrested, maintain reversibility of the arrest. 2 DNA damage checkpoints after DSBs throughout the cell cycle Central to the response to a DSB is its recognition by the MRN (Mre11, Rad5 and Nbs1) complex and the subsequent recruitment and activation of the ataxia telangiectasia mutated (ATM) kinase, which is conserved throughout the cell cycle (59, 93). Subsequent activation of various arms of the DDR is precisely tuned to each cell cycle phase as discussed below (Figure 1). G1 ATM phosphorylates S139 on histone variant H2AX (yielding γh2ax) on the DNA adjacent to a DSB, which is recognised by mediator proteins that amplify local ATM activity and recruit repair factors (63, 65, 66). In addition, ATM activates the Chk2 kinase that disseminates from the site of damage and contributes to the DDR (89). In response to a DSB in G1, ATM and Chk2 are required for the stabilisation of p53, which in turn results in the induction of a large variety of transcriptional targets, among which the cyclin-dependent kinase (Cdk) inhibitor protein p21 (also known as CDKN1A) (82, 84). Accumulated p21 binds to and inhibits cyclin-cdk complexes to block cell cycle progression, and loss of p53 or p21 results in complete loss of the G1 checkpoint (94-96). ATM also activates the p38 MAPK family, which Box 1: DNA damage response and the cell cycle DNA damage poses a constant threat to genomic integrity owing to its mutagenic potential. Under normal physiological conditions cells undergo many DNA lesions every day. The vast majority of these lesions are repaired without further consequences, but mutations resulting from incorrect repair can contribute to cellular dysfunction and tumorigenesis. Extensive or prolonged exposure to DNA-damaging agents can cause cell death or cellular senescence. DSBs are a particularly deleterious type of lesion that can cause structural chromosomal rearrangements and chromosome segregation errors when incorrectly repaired (269). In order to cope with DNA damage, cells have evolved elaborate signalling cascades, collectively known as the DDR, that detect DNA damage, coordinate DNA repair and, in proliferating cells, halt cell cycle progression (56). Cell cycle transitions are driven by Cdks that require Cyclin binding for their activity and substrate selectivity. Cyclins are subject to strict transcriptional and post-transcriptional regulation to ensure cell-cycle-specific expression, and Cdks themselves are regulated by activating and inhibitory phosphorylation (5). The DDR curbs the activity of Cyclin-Cdk complexes to elicit a cell cycle arrest in G1 or G2 phase, or to slow down replication in S phase, but the DDR does not directly impede the completion of mitosis (27). Following DNA damage checkpoint silencing, cells can resume the cell cycle, a process known as checkpoint recovery. DNA damage checkpoints and their reversal throughout the cell cylce 21

22 contributes to the G1 checkpoint through stabilisation of p21-encoding mrna (97, 98). This relatively slow transcriptional response is complemented by ATM-, Chk2-, and p38-dependent pathways that rapidly, but transiently, prevent S phase entry by promoting the degradation of Cyclin D and the Cdc25A phosphatase that reverses inhibitory phosphorylation of Cdk2 (at T14 and Y15) (97, 99-12). 2 S phase Once Cdk2 activity initiates DNA replication, DSBs can trigger entirely different signalling responses as some broken ends are not simply religated by NHEJ (For reviews see 13, 14) but instead are repaired by HR, which involves the extensive 5 3 resection of the broken ends to produce a 3 overhang. The process of DNA end-resection commits repair to HR and is tightly controlled such that HR only takes place when a sister template is available in S or G2 phase (15). The single-stranded DNA (ssdna) generated by resection activates the ataxia telangiectasia and rad3-related (ATR) kinase (69), thus promoting a switch from DNA damage signalling exclusively by ATM and Chk2 kinases in G1 to additional contributions of ATR and its effector Chk1 kinase in S phase and G2. This activation of ATR and Chk1 kinases also occurs as a consequence of replication, when the replication fork generates single-strand (ssdna) (16). Cdk2 phosphorylates the key exonucleases CtIP (also known as RBBP8) and Exo1 to enable DNA end-resection and ensure that end-resection occurs only after S phase entry (17-19). In addition, Cdk2-dependent, activating phosphorylations of the ATR-interacting protein ATRIP (at S224) and Chk1 itself (at S286 and S31) further restrict full activation of ATR and Chk1 to S and G2 phase (11, 111). Entry into S phase thus drastically changes the signalling cascades that are elicited by a DSB. Even though ATM, ATR, Chk2 and Chk1 all contribute to the stabilisation of p53 (112), downstream p21 accumulation is prevented during DNA replication by the PCNA-associated CRL4 Cdt2 ubiquitin ligase (a cullin4a RING E3 ubiquitin ligase complex containing Cdt2), which is present on the replication fork ( ). Interfering with p21 degradation in S phase compromises the control over origin licensing and results in detrimental re-replication (116), underscoring the importance of this mode of checkpoint attenuation in S phase. With the continuous degradation of p21, the intra-s checkpoint instead relies on Wee1 kinase, which becomes expressed in S phase and phosphorylates Cdk2 to inhibit its activity (117, 118). Chk1 and Chk2 deposit activating phosphates on Wee1 kinase and target the counteracting phosphatase Cdc25A for degradation to impose an immediate break on further Cdk activation ( ). G2 Wee1-dependent phosphorylation of Cdks remains crucial for checkpoint control in G2. Wee1 inhibitors override the G2 checkpoint in p53 mutant cells (122). In addition, the slower transcriptional induction of p21 that is reinstated in G2 is required for sustained inhibition of Cdk activity and cell cycle arrest (85). In contrast to the checkpoint in response to DSBs in G1, a checkpoint arrest in G2 does not absolutely depend on ATM, p53 and p21. Although ATM and Chk2 control the establishment of an arrest after damage (123, 124), ATR- and Chk1-dependent signals are essential for checkpoint maintenance in G2 ( ). Conversely, ATR and Chk1 are neither necessary nor sufficient for a G1 arrest in response to DSBs (12, 129, 13). This suggests that lesions engaged in HR are determinant for checkpoint maintenance in G2; this hypothesis is supported by a failure to maintain a G2 checkpoint arrest in the absence of PALB2 and BRCA2, which mediate strand invasion (131). In addition, ATM- and ATR-dependent activation of p38 might contribute to the maintenance of a G2 checkpoint by inhibiting both Cdc25A and Cdc25B (97). Mitosis The DDR is potently inhibited during mitosis, suggesting that at least some forms of mitotic DNA repair are undesirable. Even though a DSB sustained or generated in mitosis still activates ATM, downstream activation of Chk2 and recruitment of repair factors does not occur (93, 132). Mitotic attenuation of the DDR is enforced at various levels. The mitotic kinase Plk1 phosphorylates Chk2 on multiple sites to prevent its dimerization and recruitment to the DSB, both of which are required for its activation (132). Similarly, Plk1 abrogates Chk1 activation by inducing SCF βtrcp (a SKP1-cullin-1F-box complex that con- 22 Chapter 2

23 G1 S G2 M DNA end resection DNA end resection ATM ATM Cdk2 ATR ATM Cdk2 ATR ATM Cdk2 ATR ATM Plk1 ATR Chk2 p53 p21 Cdc25A Chk2 p53 p21 Wee1 Chk1 Cdc25A Chk2 p53 p21 Chk1 Cdc25A/B/C Wee1 Chk2 p53 p21 Cdk1 Chk1 2 Cdk4/6 Cyclin D prb Cdk2 Cyclin E Cdk2 Cyclin A Cdk1/2 Cyclin A/B Cdk1 Cyclin B APC/C Cdh1 Claspin CRL4 Cdt2 Cdk2 p21 Cdk2 Plk1 Claspin (S3 / S34) Chk2 (S164) 53BP1 (S1618) XRCC4 CtIP (aa ) Exo1(S639 / T732 / S815) ATRIP (S224) Chk1 (S286 / S31) CtIP (aa ) Exo1(S639 / T732 / S815) ATRIP (S224) Chk1 (S286 / S31) Cdk1 RNF8 (T198) 53BP1(T169?) BRCA2 (S3291) XRCC4 (S326) Figure 1. DNA damage signalling after DSBs in the various stages of the cell cycle Interplay between the cell cycle machinery and the DDR results in distinct signalling in response to DSBs in different cell cycle phases. DSBs are recognised by the Mre11-Rad5-Nbs1 (MRN) complex, which subsequently activates the ATM kinase in all phases of the cell cycle. Cdk2 activity in S phase and G2 enables extensive resection of DSB ends (through phosphorylation of CtIP and Exo1) to promote HR, causing the additional activation of ATR kinase. Cdk2-dependent phosphorylation of ATRIP and Chk1 further restrict full activation of ATR and Chk1 to S and G2 phase. Additionally, in G1, APC/C Cdh1 -mediated turnover of claspin disables activation of Chk1. During DNA replication, the PCNA-coupled CRL4 Cdt2 ubiquitin ligase prevents the accumulation of p21. Finally, DNA damage signalling is potently inhibited in mitosis owing to the combined action of Cdk1 and Plk1. Plk1-dependent phosphorylation of claspin targets it for degradation, effectively inhibiting Chk1 activation. In addition, Plk1-dependent phosphorylation of Chk2 prevents Chk2 activation. Both Plk1 and Cdk1 phosphorylate and thereby inhibit the DNA repair proteins 53BP1 and XRCC1, and Cdk1 additionally inhibits RNF8 and BRCA2, resulting in the inhibition of DNA repair during mitosis. The residue subjected to inhibiting or activating phosphorylation is shown in parentheses. tains βtrcp as the F-box protein)-mediated degradation of the scaffold protein claspin that is required to bridge ATR and Chk1 ( ). It is conceivable that checkpoint kinases that inhibit Cdks must be quelled to maintain the full Cdk activity that is necessary for the execution of mitosis. However, the mitotic inhibition of the DDR is not limited to the checkpoint. Cdk1 phosphorylates the ubiquitin ligase RNF8 (at S198) to prevent the formation of K63-linked polyubiquitin at damage sites (136). Additional phosphorylation of residues T169 and S1618 of 53BP1 by Plk1 and an unidentified kinase interferes with its recruitment to damaged DNA (132, 137). Phosphorylation of RNF8 and 53BP1 prevents DNA repair during mitosis (136). Additionally, Plk1 and Cdk1-dependent phosphorylation of XRCC4 contributes to the suppression of NHEJ during mitosis (138). Collectively, these mechanisms eliminate most checkpoint signals and DNA repair in mitosis. Although the picture is far from being complete, it has become clear that a myriad of mechanisms operate to fine-tune the DDR in each phase of the cell cycle. Attenuated checkpoint responses are necessary for the orderly completion of DNA replication in S phase (116), to prevent deleterious fusions of telomeres in mitosis (136), and probably more. DNA damage checkpoints and their reversal throughout the cell cylce 23

24 Dismantling the DNA damage checkpoint 2 Activation of the DDR leads to a highly complex series of events orchestrated by a plethora of posttranslational modifications (PTMs), such as phosphorylation, acetylation, methylation, poly(adp-ribosyl)ation (PARylation), ubiquitylation and small ubiquitin-like modifier (SUMO)ylation. As these modifications trigger checkpoint activation, cell cycle arrest and DNA repair, removal of these modifications by dedicated enzymes or degradation of modified proteins is essential for checkpoint silencing and recovery. Thus far, this emerging field has identified redundancy and complex cross-talk between the modifications as a common theme, as discussed below. Direct reversal of PTMs Phosphorylation is the most widespread and best-studied PTM of the DDR. Many serine/threonine phosphatases are involved in dephosphorylation, and these proteins act in a largely redundant manner (Figure 2) (139, 14). For instance, PP2A (in complex with B55α and B56ε subunits), PP4, PP6 and Wip1 all dephosphorylate γh2ax in vitro and control γh2ax levels in vivo ( ). This redundancy is most likely a reflection of different spatiotemporal regulation of these phosphatases during the DDR. Indeed, Legend DDR PPase DUB DeAc PARG HiCh BET KAP1 PP2A PP1 prb PP2A Wip1 Chk1 Chk2 PP2A PP1 Wip1 PP1 PP4 KAP1 USP2a PP1 Wip1 Mdm2 PP2A p53 Wip1 PP1 SIRT1 Phosphorylation Ubiquitination Acetylation Poly(ADP ribosyl)ation p53 Histone PP5 PP4 Wip1 SIRT1 PP6 PP2A H2AX EYA1/3 FACT ATR ATM PP2A PP1 Wip1 CAF1 HIRA Asf1 53BP1 Ubi Hist USP44 USP3 PP5 PAR Hist PARG Chromatin compaction HDAC BRD4 H2AX H3/H4 H3.3 BRCC3 USP16 POH1 Figure 2. Dismantling the DNA damage checkpoint The DDR results in the generation of many PTMs in order to execute a cell cycle arrest. Thus, recovering from the DNA damage checkpoint requires reversal of phosphorylation by phosphatases (PPase; red), ubiquitylation by deubiquitinases (DUB; green), acetylation by deacetylases (DeAc; purple), poly(adp-ribosyl) (PAR) modification by PAR glycohydrolase (PARG; yellow), as well as the structural rearrangement of the chromatin by histone chaperones (HiCh; blue) and the bromodomain and extra-terminal family of proteins (BET; brown). The presently identified enzymes that act on central components of the DDR are displayed. Aside from PTM reversal, activated checkpoint components can also be removed either by proteasomal degradation, as is the case for p53 and phosphorylated KAP1 (282) (indicated by the red cross), or by exchange, as occurs with γh2ax in a process mediated by the histone chaperone FACT. Chromatin compaction is another important impediment of the DDR. Histone deacetylases (283) and the condensin II chromatin remodelling complex recruiter BRD4 (284) promotes chromatin compaction and restrict spreading of DDR signalling on chromatin. Finally, after repair has taken place the histone chaperones Asf1, Caf1 and HIRA are required for restoring nucleosomes at the site of damage. For further details see: USP2a-dependent deubiquitination of Mdm2 (285); dephosphorylation of ATR (286), Chk1 (178, 287, 288), 53BP1 (289) and prb (29-292); deacetylation of p53 (293, 294) and γh2ax (295); and the removal of (PAR)-groups by PARG (296). 24 Chapter 2

25 PP2A phosphatase activity towards γh2ax-containing nucleosomes is repressed in the first hour, but increases by twofold 5 hours after DSB induction by camptothecin (144). PP1, PP2A, Wip1 and PP6 are regulated through distinct mechanisms by DNA damage, and both PP2A and Wip1 have been shown to accumulate at DSBs (141, 143, 144, ), whereas PP4 appears to be selective for chromatin-dissociated γh2ax in yeast (15), revealing that these phosphatases modulate specific pools of γh2ax at different times. Although less thoroughly assessed for other checkpoint phosphorylation events, redundancy and spatiotemporal regulation are a recurrent theme. For example, both ATM (S1981) and Chk2 (T68) are dephosphorylated by PP1, PP2A and Wip1 (149, ). Recognition of T68-phosphorylated Chk2 by the B56α subunit of PP2A can be prevented by additional ATM-dependent phosphorylation of Chk2 at S33 and/or S35, suggesting that T68 dephosphorylation is facilitated by prior loss of ATM activity and dephosphorylation of S33 and/or S35 by an additional phosphatase (157). Formation of a Chk2-Wip1 complex, by contrast, requires Chk2 kinase activity (158). These examples illustrate that checkpoint phosphorylation events are counteracted by multiple redundant phosphatases that are regulated themselves in time and space, and, in addition, show regulated activity towards specific substrates. Protein phosphorylation also controls the extensive ubiquitylation at sites of damage to create a landing platform for additional signalling and repair proteins. The best-characterised events are the sequential recruitment of the E3 ubiquitin ligases RNF8 and RNF168 that produce K63-linked polyubiquitin chains (159, 16), but many other ubiquitin ligases with largely unknown substrates take part in the DDR (161). Several deubiquitylating enzymes (DUBs) have been implicated in removal or editing of the K63 ubiquitin chains formed at DSBs. USP3 and USP44 overexpression reduces the size and number of DNA damage foci and depletion of POH1, BRCC36 or USP16 results in the defective removal of ubiquitin chains at the site of damage (159, ). Both DUBs POH1 and BRCC36 are recruited to K63-linked polyubiquitin chains to reverse these same RNF8- and RNF168-dependent ubiquitilation events at the chromatin, yet their impact on DNA repair processes can be antagonistic, illustrating the complex regulation of ubiquitin at DSBs (163, 165, 166). Recently, a systematic screen of DUBs identified an additional 29 DUBs that localise to DNA damage when overexpressed as GFP fusions and potentially regulate checkpoint silencing (167). 2 Removal of modified proteins Direct reversal of protein modifications is not the only means to dismantle the DDR. Proteasomal destruction is the end point of many of the modified proteins. The proteasome is recruited to sites of DNA Box 2: Restoring chromatin During the DDR, chromatin is extensively remodelled and nucleosomes are removed at the sites of damage (271). The histone chaperone FACT (for facilitates chromatin transcription ) removes γh2ax- H2B dimers from sites of damage, which is inhibited by PARylation of FACT (272). At the same time, FACT is bound to the E3 ubiquitin ligase RNF2 that mono-ubiquitilates histone H2B (273). Once repair is finished, nucleosomal organisation must be restored to reinstate proper chromatin structure and function (274). Histone chaperones are key players during the restoration of the chromatin. Studies in yeast have shown that reloading of H3K56-acetylated histones into repaired DNA requires histone chaperones Asf1 and CAF-1, which then functions as a repair complete signal (275, 276). Conflicting reports exist as to whether H3K56 acetylation in human cells localises to sites of damage or is removed, although re-establishment of H3K56-acetylated histones is required for recovery from UV-induced DNA lesions ( ). In addition, the histone chaperone HIRA has been implicated in restoring the chromatin after UV-induced lesions, where it promotes transcription restart by promoting the insertion of the H3.3 histone variant (28). The observation that HIRA-dependent incorporation of H3.3 increases after DSBs to promote HR suggests that this is a more general mechanism (281). The contribution of these events to the coupling of ongoing repair with reactivation of the cell cycle machinery remains one of the most crucial issues to be resolved in the field. DNA damage checkpoints and their reversal throughout the cell cylce 25

26 2 damage, possibly by the associated DUBs POH1 or UCHL5 (165, 167, 168). The ubiquitin-ligase RNF4 recognizes SUMOylated MDC1 and RPA1 at DNA repair foci and promotes their proteasomal degradation (169). RNF8 not only facilitates the formation of K63-linked polyubiquitin chains by RNF168, but also targets Ku and Chk2 for proteasomal degradation via distinct K48-linked ubiquitin chains (17). Similarly, the E3-ligase Iduna is recruited to and activated by PAR chains, where it targets the PAR polymerase for proteasomal degradation (171). Alternatively, eviction and replacement of heavily modified nucleosomes might also aid in terminating checkpoint signals (see Box 2). Consequences for recovery As PTMs are central to both the coordination of DNA repair and checkpoint signalling, disruption of enzymes that remove such PTMs can affect cell survival after DNA damage in multiple ways. Disruption of PP2A activity impairs DSB repair, and the observed persistence of γh2ax in PP2A-depleted or inhibited cells might therefore be caused by either persistence of DSBs, lack of checkpoint silencing or both (141, 144). Chemical inhibitors of checkpoint kinases, such as caffeine, present a useful tool to separate defects in DNA repair from defects in terminating the checkpoint downstream of these kinases (172, 173). In contrast to PP2A, loss of Wip1 or PP4 have limited effects on DSB repair, but strongly affect checkpoint silencing, cell cycle restart and survival after DNA damage, even in the presence of checkpoint kinase inhibitors ( ). Crucial functions of these phosphatases in checkpoint reversal are limited to a select number of substrates, possibly owing to the extensive redundancy between phosphatases. G1 Checkpoint Activation Checkpoint Silencing Cdk2 Cyclin E Cdk4/6 Cyclin D Cdc25A ATM Cdk4/6 Cyclin D Cdk2 Cyclin E Cdc25A ATM PP4 Chk2 Kap1 p53 Cdk4/6 Cyclin D p21 Cdk2 Cyclin E Cdc25A ATM PP4 Chk2 Kap1 p53 Cdk4/6 Cyclin D p21 Cdk2 Cyclin E p38 Cdc25A p27 G2 Checkpoint Activation Cdk1 Cyclin B ATM Chk2 Cdk1 Cyclin B ATM Chk2 ATR Chk1 Cdk1 Cyclin B ATR ATM Chk1 Chk2 p53 p21 Cdk1 Cyclin B APC/C Cdh1 Checkpoint Silencing FoxM1 Plk1 Cyclin B Wip1 Cdc25B Plk1 Wip1 Cdc25B Plk1 Wip1 FoxM1 Plk1 Plk1 Cyclin B Undamaged Early DDR Late DDR Figure 3. DSB-induced G1 and G2 phase arrests are functionally distinct A DSB initiates distinct signalling cascades (in red) in G1 and G2 over time (Figure 1), which impinge on distinct factors that promote cell cycle progression (in green) in G1 and G2. In G1, p53 and its transcriptional target p21 impose a reversible cell cycle arrest by preventing the accumulation of cyclin-e-cdk2 activity. In G2, signalling by checkpoint kinases inactivates cyclin-b1-cdk1, reversibly forestalling mitotic progression. As a consequence, recovery from a G1 checkpoint arrest requires PP4-dependent reactivation of KAP1 to turn off p21 transcription, whereas G2 recovery requires Wip1- and Plk1-mediated inhibition of checkpoint kinases and reactivation of cyclin-b1-cdk1. Because p21-mediated inhibition of Cdk complexes in G2 compromises G2 gene expression and recovery competence (Figure 4), induction of p21 marks the first step towards a permanent G2 arrest. In sharp contrast, p21 induction in G1 is central to a reversible cell cycle arrest, resulting in very different time windows of checkpoint reversibility (indicated by green background shading) in G1 and G2. 26 Chapter 2

27 Wip1 dephosphorylates γh2ax, ATM, p53, Mdm2, Chk1 and Chk2 to repress the DDR (143, 152, 154, 178, 179), yet its role in checkpoint recovery is restricted to G2 where it is required to antagonise p53 (Figure 3) (174). Wip1 depletion in p53-deficient cells does not affect recovery from DNA damage in G2, underscoring that Wip1 activity towards these other substrates is dispensable (174). Similarly, PP4 dephosphorylates γh2ax, RPA2, 53BP1, CHD4 and KAP1, but recovery in its absence is mostly affected in G1 cells (Figure 3), and can be effectively overcome by expression of a non-phosphorylatable mutant of KAP1 (S473A) (175, 18). Although several DUBs are required for both DNA repair and cell viability after irradiation (167, ), the role of DUBs in checkpoint silencing and cell cycle restart after DNA damage remains to be addressed. Of the 32 DUBs found to localise to sites of DNA damage, 13 appear to be dispensable for overall DNA repair, as judged by single-cell gel electrophoresis, with depletion of USP4, USP37 or POH1 affecting checkpoint kinase signalling. Their contribution to checkpoint silencing needs to be assessed and mapped to relevant substrates. 2 G2 checkpoint recovery Plk1 couples cell cycle restart to checkpoint deactivation Checkpoint recovery follows largely the same activation pathway as unperturbed mitotic entry (12), but there are clear differences: the DDR inhibits the activities of the pro-mitotic kinases Cdk1, Aurora A and Plk1 ( ), and causes the degradation of several proteins that promote mitotic entry, such as Cdc25A and Bora (87, 187). The resultant rewiring of the cell cycle machinery is most clearly demonstrated by the acquired dependence on Plk1 activity for G2 checkpoint recovery (Figure 3). Plk1 is dispensable for normal mitotic entry and its inhibition in unperturbed cells merely leads to a delay in mitotic entry. However, after a DNA damage-induced G2 arrest, Plk1 and its upstream activators Aurora A and Bora become essential for re-entry into the cell cycle, by contributing to the activation of cyclin-b1-cdk1 (12, 188, 189). In parallel, the degradation of Cdc25A reduces the redundancy of Cdc25 phosphatases normally observed during mitotic entry, and recovering cells become dependent on a specific isoform of Cdc25B, which is stabilised after damage (188, 19). Aside from its roles in re-activation of cyclin-b1-cdk1, Plk1 additionally controls silencing of DDR signals. The same mechanisms that enable Plk1 to prevent checkpoint signalling in mitosis operate during checkpoint recovery in G2 (Figs. 1 and 3). Plk1 activity starts to rise in G2, several hours before mitosis, and at early stages of the DNA damage response Plk1 facilitates strand invasion for HR (191). During checkpoint recovery, Plk1 targets Claspin for degradation, disabling Chk1 activation ( ). Failure to degrade Claspin leads to enhanced and continued Chk1 activity and inability to recover from a G2 arrest. Similarly, Plk1 negatively regulates 53BP1 and Chk2 to promote recovery (132, 192). Plk1 might also contribute to the inactivation of p53 through direct and indirect mechanisms ( ), and proteomic approaches have identified many potential new binding partners and substrates of Plk1, including DNA damage checkpoint and repair proteins (197, 198). Plk1 thus drives transitions from arrest to cell cycle re-entry and from checkpoint signalling to checkpoint silencing and recovery. Adapt now, repair and recover later It is typically thought that cell cycle resumption upon a DSB-induced arrest occurs only following completion of DNA repair. However, cells can terminate the checkpoint and enter mitosis in the presence of DNA damage, a process well characterised in unicellular yeast and known as checkpoint adaptation (Reviewed in 199). The first hints of checkpoint adaptation in human cells came from the discovery that DNA damage in HCT116 cells resulted in aberrant mitoses (2), which was later strengthened by the finding that, as in yeast, checkpoint adaptation in human cells requires Plk1 (21, 22). Plk1 must be degraded following DNA damage in order to achieve a proper G2 arrest (23) and the inability to silence Plk1 activity following DNA damage increases the fraction of cells that is able to enter mitosis DNA damage checkpoints and their reversal throughout the cell cylce 27

28 2 (184). It was subsequently shown that irradiation of cells in G2 results in increased amounts of γh2ax and MDC1-foci in mitotic cells, both markers for DNA damage (24-26), strengthening the notion that maintaining a G2 cell cycle arrest requires a threshold of persistent DNA damage. Using a fluorescence-based probe for Plk1 activity and DNA damage in single cells, Liang and colleagues have recently reported that DNA damage permits a gradual increase of Plk1 activity throughout the arrest (25). Once Plk1 activity increases beyond a certain level, the cell progresses to mitosis, regardless of whether DNA damage persists. G2 checkpoint recovery might thus represent checkpoint adaptation, where DNA damage triggers an arrest whose duration is not necessarily conditioned by DNA repair. In support of this notion, the kinetics of recovery from DSBs in G2 in untransformed cell lines and the mouse intestinal epithelium in vivo indeed show relatively synchronous mitotic entry after a delay of several hours (175, 27). However, it is important to note that the inducible expression of a constitutively active mutant of Plk1 (T21D) fails to override an established DNA damage checkpoint (R.H.M. unpublished results), indicating that Plk1 activity is insufficient to drive mitotic entry of cells with a fully activated DDR. Therefore, other factors that contribute to the silencing of the DNA damage response operate in conjunction with Plk1 to drive checkpoint recovery or adaptation. Checkpoint adaptation is a potentially dangerous process, as the propagation of damaged DNA can contribute to tumorigenesis. Mitotic entry in the presence of severe DNA damage results in cell death through mitotic catastrophe (85, 28, 29). Alternatively, pre-marked breaks present in G2 cells are partially disassembled during mitosis and carried over to daughter cells where the DDR is fully (re)activated in G1 (93, 137). Activation of the DDR in G2 can even prime adapting cells for a cell cycle arrest or delay in G1. It has been recently shown that the amount of residual Cdk2 activity at mitotic exit dictates the quiescence or proliferation decision in daughter cells (21). Although cells normally retain some cyclin- E-Cdk2 activity as they complete mitosis and swiftly progress through G1 without the requirement for mitogenic stimulation, the induction of p21 in late G2 or mitosis gives rise to quiescent daughters with hypophosphorylated retinoblastoma protein (prb) (21). Similarly, PP2A-B56γ-dependent dephosphorylation of prb in G2 results in quiescent daughter cells that lack residual Cdk2 activity (211). Inhibition of PP2A-B56γ in G2 results in increased levels of cyclin E1 in mitotic cells and accelerates Cdk2-dependent S phase entry in daughter cells. As ATM contributes to the activation of PP2A-B56γ (146), ATM might thus synergise with p21 to inhibit Cdk2 activity in checkpoint-adapting cells, which are then programmed to enter a state of quiescence after cell division to safeguard genomic stability. G2 checkpoint recovery maintaining reversibility Balancing pro-mitotic gene expression Expression of G2-specific genes is driven by the transcription factors FoxM1, B-Myb (also known as MYBL2) and nuclear factor Y (NF-Y), and is essential for timely mitotic entry and proper progression through mitosis ( ). FoxM1, NF-Y and B-Myb are all activated by Cdk2, ensuring that they become active only when cells progress through S and G2 phase ( ). Given that Cdk activity is required for the transcription of G2-specific genes, decreased Cdk activity following DNA damage can compromise their expression (218-22). Some residual Cdk activity, possibly supported by specific Cdc25B isoforms (19), must therefore be retained. Temporary elimination of this residual Cdk activity with inhibitor compounds abolishes G2-specific gene expression and compromises the ability of cells to recover after checkpoint inhibition (221). Similarly, small interfering RNA (sirna)-mediated depletion of FoxM1 is detrimental to G2 recovery and can only be rescued by inducible sirna-resistant FoxM1 when it is present throughout the arrest (221). In conjunction with FoxM1, B-Myb contributes to the expression of G2-specific genes during an arrest in G2 and is required for checkpoint recovery in p53-deficient cells (222). Damaged G2 cells must thus carefully suppress the Cdk-driven G2 transcriptional programme during the arrest to prevent premature mitotic entry, but at the same time retain sufficient expression to maintain recovery competence (Fig 4A; for a review on fine-tuning of transcriptional regulation during a DNA damage-induced G2 arrest see 223). 28 Chapter 2

29 p53 counteracts the transcription of many G2-specific genes to contribute to the maintenance of a G2 arrest, and upregulation of p53 is sufficient to permanently arrest cells in G2, even in the absence of DNA damage (85, 222, ). Consequently, cells must also balance p53 activity to maintain the reversibility of a G2 arrest (Figure 4A). This requires negative regulation of p53 by Mdm2 and Wip1, which are both transcriptional targets of p53 (227, 228). Mdm2 is the main E3 ubiquitin ligase that targets p53 for proteasomal degradation and inhibits its function as a transcription factor (for a review on Mdm2 and p53 see 229), whereas Wip1 dephosphorylates both p53 and Mdm2 to promote p53 degradation (178, 179). Recently, TRIM24 was identified as an additional p53-induced ubiquitin ligase that preferentially targets phosphorylated variants of p53 for degradation (23). As a result, p53 levels oscillate in response to DNA damage and altering the oscillatory pattern of p53 results in distinct cell fate outcomes ( ). Wip1 depletion results in excessive p53 signalling after DNA damage and renders cells unable to recover even when the checkpoint is silenced (174, 231). Reintroducing sirna-resistant Wip1 restores recovery competence when it is continuously present during the arrest, but not when it is only induced from the moment of checkpoint silencing, underscoring the importance of negative regulation of p53 in the time frame preceding checkpoint reversal for controlling the outcome of the DNA damage response (174). 2 Degradation of cell cycle proteins regulates recovery competence The anaphase-promoting complex/cyclosome (APC/C) is a multi-protein complex with E3 ubiquitin ligase activity that is normally active from the completion of chromosome alignment in mitosis until the end of G1; it is necessary to direct the degradation of cyclins, securin and several mitotic kinases for the orderly completion of mitosis (21). Two coactivators, Cdc2 and Cdh1, sequentially activate the APC/C. A Cdk2 NF-Y FoxM1 B-Myb Cyc B1 Plk1 Cyc A2 p53 Wip1 Mdm2 Cyc B1 Plk1 Cyc A2 p21 p53 APC/C Cyc B1 Cdh1 Plk1 Cyc A2 B p53 Cdc25 Wee1 Cdk1 Cyclin B1 G2-specific gene expression Mitosis Recovery Competence Threshold p21 Cdc25 Wee1 p21 Cdk1 Cyclin B1 Irreversible arrest G2 Checkpoint ON Checkpoint OFF Figure 4. Regulation of recovery competence in G2 (A) During normal G2 progression, the expression of G2-specific genes increases to allow timely mitotic entry. Expression of these genes is dependent on Cdk2. Following DNA damage, reversibility of a G2 arrest requires the sustained expression of G2-specific genes, which allows mitotic entry once the checkpoint is silenced. Failure to sustain expression of these genes results in an irreversible G2 arrest (bottom panel). Cdk2-dependent activation of FoxM1 during a G2 arrest contributes to the maintenance of recovery competence (top left panel), whereas p53-dependent repression of gene expression negatively regulates recovery competence (top middle panel). Wip1 and Mdm2, both negative regulators of p53, promote recovery competence (top middle panel), whereas p53- and p21-dependent APC/C Cdh1 activation initiates the degradation of pro-mitotic genes, thereby leading to an irreversible arrest (top right panel). (B) Another mechanism regulating recovery competence in G2 is the subcellular relocalisation of cyclin-b1- Cdk1 complexes following DNA damage. After DNA damage, cyclin-b1-cdk1 binds to p21, resulting in its nuclear accumulation. At the same time, Cdc25 phosphatases are excluded from the nucleus following damage, preventing cyclin-b1-cdk1 (re)activation upon silencing of the DNA damage checkpoint and recovery from a G2 arrest. DNA damage checkpoints and their reversal throughout the cell cylce 29

30 2 Cdc2 activates the APC/C following chromosome alignment and is inhibited by the spindle assembly checkpoint. This initiates the destruction of Cyclin B1 and a decrease in Cdk activity, which is required to allow Cdh1-dependent activation of the APC/C during the late stages of mitosis and throughout G1. DNA damage in G2 can activate the APC/C Cdh1, which is essential for a persistent G2 arrest and requires p53- and p21-dependent Cdk inhibition (23, ). Premature activation of the APC/C Cdh1 in G2 by DNA damage targets many proteins required for mitotic entry for destruction, including Cyclin A, Cyclin B1 and Plk1, thus preventing recovery from a G2 arrest (Figure 4A) (12, 226, ). The resulting cells that have a G2 DNA content but lack G2-specific protein expression are committed to senescence (237, 238). In order for a G2 arrest to remain reversible, Cdk activity must thus be maintained at sufficient levels to preserve pro-mitotic gene transcription, but also to prevent premature activation of the APC/C. Subcellular redistribution of proteins as a mechanism to control recovery competence Throughout G2, cyclin B1 continuously shuttles between the cytoplasm and the nucleus, but the rate of export exceeds the rate of import, resulting in a mostly cytoplasmic localisation until the onset of mitosis (239). In response to DNA damage, cyclin B1 is bound by p21 and relocates to the nucleus within 5 hours in normal human fibroblasts (226, 24). The binding of p21 to cyclin-b1-cdk1 complexes in the nucleus most likely prevents their nuclear export. Whereas Cdc25A is degraded upon DNA damage, Cdc25B and Cdc25C are exported out of the nucleus (Reviewed in 241), separating them from cyclin-b1-cdk1 complexes that have translocated to the nucleus in response to damage. These nuclear p21-bound cyclin-b1-cdk1 complexes are inhibited by nuclear Wee1 (226, 24, 242). Moreover, once cyclin-b1-cdk1 complexes have translocated to the nucleus, they become refractory to reactivation by addition of a Wee1 inhibitor, most likely due to the cytoplasmic sequestration of the remaining Cdc25 isoforms. This causes the G2 arrest to be maintained in cells with nuclear Cyclin B1, even if the checkpoint is silenced prior to APC/C activation (Figure 4B) (226, 24, 242). As p53 plays a central role in the nuclear translocation of cyclin B1, APC/C activation and the regulation of repression of pro-mitotic gene transcription, these mechanisms are often attenuated in cancer cells. G1 checkpoint recovery The G1 checkpoint is important in cells that encounter DSBs during G1, as well as those that have adapted from the G2 checkpoint or sustained breaks during mitosis (93). In addition, G2-arrested cells that activate APC/C Cdh1 are likely to be subjected to a G1-type DDR, owing to the lack of Cdk1/2 activity. Reversal of the G1 DNA damage checkpoint appears to be distinct from G2 as, firstly, the critical players that promote checkpoint recovery in G2 are absent (Plk1 and Cdc25B) or do not affect recovery in G1 (Wip1; ref. 175). Secondly, unlike the G2 checkpoint, the G1 checkpoint remains reversible for several days in untransformed cells and the mechanisms that cause an irreversible arrest in G2 (Figure 4) are not relevant in G1 (12, 175, 226). Given the exquisite dependence of the G1 checkpoint on p53 and p21, the oscillatory behaviour of p53 in response to DSBs provides an attractive model of recurrent initiation, in which waves of ATM activity, p53 and p21 come and go until triggers (DSBs) are eliminated (243). The simple take on this model is that activation of the DDR in G1 drives the cell into a quiescent-like state and, as time progresses, the DNA damage can be repaired in order for cells to re-enter G1. In the arrested state, either the DNA damage is repaired in time to drive the cell back to early G1, or the intensity and number of pulses commit the cell to an irreversible arrest. Reality, however, appears to be more complex. First, late signalling events uncouple G1 checkpoint maintenance from ATM activity. This is illustrated by the fact that in the absence of Wip1, ATM activity, gauged by its autophosphorylation, no longer follows a pulsatile pattern, yet p53 oscillations are maintained and G1 checkpoint recovery is unaffected (175, 231). This suggests that recurrent initiation at the level of the DSB-sensing kinase ATM is not required for the reversibility of the G1 checkpoint. Strikingly, although ATM inhibitors can prevent the initiation 3 Chapter 2

31 of a G1 checkpoint arrest, ATM inhibition is insufficient to reverse the checkpoint arrest (175). Instead, G1 checkpoint maintenance relies strongly on the ATM downstream targets Chk2 and p38 MAPK (98, 12, 175, 244). Both kinases can stabilise p53, and p38 further contributes to a G1 arrest by stabilising p21-encoding mrna and p27 (also known as CDKN1B) protein (98, 245). It is currently unknown how p38 is controlled after its initial ATM-dependent activation, although the p53 target Gadd45α has been implicated in maintaining its activity during a persistent arrest (246). ATM thus triggers the arrest in response to DSBs in G1, but Chk2 and p38 take over to sustain the arrest independently from ATM. Chk2 and p38 must therefore be silenced to promote a cell cycle restart. Despite the ability of Wip1 to remove the activating phosphorylations on both Chk2 and p38, spontaneous recovery from a G1 checkpoint arrest excludes a major role for Wip1 in terminating Chk2 and p38 signalling (152, 175, 247). Second, p53 levels do not correlate directly with its transcriptional activity or with the subset of target genes it activates. Target genes are individually regulated by damage signals and other pathways on the levels of DNA, RNA and protein, thereby allowing control beyond p53 dynamics. This is clearly illustrated by p21, whose transcription is quickly terminated by binding of the co-repressor KAP1 to its promoter, allowing p21 mrna levels to closely follow the activity of p53 (233, 248, 249). KAP1 itself is inhibited by ATM- and Chk1- or Chk2-mediated phosphorylation, which is counteracted by the PP4 phosphatase (18, 25, 251). Tampering with the dynamic repression of p21 transcription by depleting either KAP1 itself or PP4 compromises G1 checkpoint reversal, making KAP1 and PP4 essential for G1 checkpoint recovery (175). At the mrna level, p21 transcripts must be stabilised by the RNA-binding protein HuR, which requires HuR to be phosphorylated by Chk2 and/or p38 (98, 252). Thus, whereas p53 remains a prerequisite for p21 induction in response to DNA damage, several pathways converge on its targets to define the reversibility of the arrest. 2 What determines cellular fate? Although the regulation of recovery competence is quite well understood, little is known about factors that determine whether a cell will remain recovery competent or exit the cell cycle upon DNA damage. As many of the processes regulating recovery competence are affected by p53, cell fate decisions will most likely be made at the level of p53. Indeed, a permanent cell cycle exit in G2 requires p53 (85, 226, 237), and cells depleted of p53 are unable to maintain a stable G2 arrest and eventually will progress through mitosis in the presence of DNA damage (85). Recovery competence is much more stringently regulated in G2 compared to G1 phase. Whereas an 8-hour pulse of p53 activation readily induces senescence in G2 cells, it fails to do so in G1 (226) and R.H.M. unpublished results), indicating that the cell cycle position of a DNA-damaged cell influences recovery competence. In addition, the mode of p53 activation greatly impacts upon cell fate decisions (233). One factor that affects p53 activation and cellular outcome is the amount of DNA damage, where increasing doses of DSBs result in a stronger activation of p53 and cause more cells to undergo senescence (233, 253). Furthermore, sustained DSBs cause the continuous oscillation of p53, resulting in an increased induction of p21 and a greater propensity to enter senescence (232, 253, 254). However, recent data show that cell fate decisions in DNA-damaged G2 cells are established within 5 hours, which would correspond to the first wave of p53 (226). This suggests that the strength of the first p53 oscillation, rather than number of oscillations, establishes cellular fate in G2. These observations, albeit contradictory, might very well operate within the same cell. Although DNA-damaged G1 cells retain recovery competence longer than their G2 counterparts, they eventually become incapable of G1 recovery (R.H.M. unpublished results). It is likely that G1 recovery competence is dictated by the number of p53 oscillations, whereas G2 recovery competence is determined by the strength of the first pulse. Another explanation for these contradictory observations might be found in the cell types used for investigation. Much of the knowledge obtained about p53 oscillations comes from transformed cell lines, in which the mechanisms for cell cycle exit are often repressed (242, 255, 256). Clearly, p53 activity is a major determinant in cellular fate decisions; however, how regulation upstream or downstream of p53 affects cellular fate remains largely unresolved. DNA damage checkpoints and their reversal throughout the cell cylce 31

32 2 In addition, the location of DNA damage might greatly influence the outcome of a DNA damaging insult. Up to half of the persistent DNA damage foci in cells undergoing senescence in response to ionising radiation are associated with telomeres (257). This telomeric DNA damage is refractory to repair and as a result causes persistent DNA damage signalling (258, 259). Indeed, p53-oscillations might be different in cells with similar amount of DNA damage foci (253), indicating that the location of DNA damage, its type or repair state may influence p53 dynamics. As not all persistent foci in irradiation-induced senescence are found at telomeres, this raises the question as to whether there are other genomic loci that are refractory to DNA repair, but instead will activate persistent DNA damage signalling and the onset of senescence. Conclusion and future directions DNA damage checkpoint arrests are useful as long as they can be reversed when appropriate. The notion that recovery from a DNA damage-induced arrest is merely a consequence of DNA repair is out-dated: checkpoint recovery involves the timely and spatially controlled action of numerous enzymes that disassemble DNA damage foci as the DNA is repaired and the extensive interplay between the cell cycle machinery and DDR that we are only starting to comprehend. In addition, little is known about the determinants of cell fate following a DNA damaging insult. As the first hints of processes underlying cell fate decisions, such as the mode of p53 activation or DNA damage severity (226, 233), are being reported, many questions are left unanswered. For instance, it remains to be determined whether the location of a DSB, or the repair pathway used for its repair affect cellular fate. Understanding these processes is vital to our understanding of normal cell biology, tumorigenesis and cancer resistance to genotoxic therapy. The accumulation of mutations during tumorigenesis likely requires several events of checkpoint recovery. Consistent with this is the finding that positive regulators of checkpoint recovery in G2 are commonly overexpressed in cancer and associated with poor prognosis (256, ). Conversely, tumorigenesis is impaired in their absence (264, 265), and several regulators of G2 checkpoint recovery are under (pre)clinical evaluation ( ). Understanding the mechanisms underlying cellular fate will allow us to devise new therapeutic strategies to sensitise cancer cells to DNA-damaging agents currently used in the clinic. Funding This work was supported by Koningin Wilhelmina Fonds (KWF) [grant number NKI to R.H.M.]; and Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO) [Cancer Genomics Center, CGC. nl to R.H.M., and grant number to L.K.]. Competing interests The authors declare no competing or financial interests. 32 Chapter 2

33 DNA damage checkpoints and their reversal throughout the cell cylce 33 2

34

35 Chapter 3 Transient activation of p53 in G2 phase is sufficient to induce senescence 3 Lenno Krenning, Femke M. Feringa, Indra A. Shaltiel, Jeroen van den Berg and René H. Medema 1 Division of Cell Biology I and Cancer Genomics Center, The Netherlands Cancer Institute Plesmanlaan 121, 166 CX Amsterdam, The Netherlands 1 ) Correspondence: r.medema@nki.nl Adapted from Molecular Cell. 214 Jul 3; 55(1):59-72 Transient activation of p53 in G2 phase is sufficient to induce senescence 35

36 Abstract DNA damage can result in a transient cell cycle arrest or lead to permanent cell cycle withdrawal. Here we show that the decision to irreversibly withdraw from the cell cycle is made within a few hours following damage in G2 cells. This permanent arrest is dependent on induction of p53 and p21, resulting in the nuclear retention of Cyclin B1. This rapid response is followed by the activation of the APC/C Cdh1 (the anaphase-promoting complex/cyclosome and its coactivator Cdh1) several hours later. Inhibition of APC/C Cdh1 activity fails to prevent cell cycle withdrawal, whereas preventing nuclear retention of Cyclin B1 does allow cells to remain in cycle. Importantly, transient induction of p53 in G2 cells is sufficient to induce senescence. Taken together, these results indicate that a rapid and transient pulse of p53 in G2 can drive nuclear retention of Cyclin B1 as the first irreversible step in the onset of senescence Chapter 3

37 Introduction The cellular response to DNA double strand breaks involves the activation of the ATM and ATR checkpoint kinases, which promote DNA repair, transient cell cycle arrest, as well as the induction of apoptosis or senescence (297, 298). A central player in the induction and maintenance of cell cycle arrest in response to DNA damage is the tumor suppressor protein p53. p53-dependent upregulation of the Cdk inhibitor p21 is essential to arrest cells in G1 after DNA damage (299, 3). In addition, both p53 and p21 are required to maintain a G2 arrest (85). In part, this is achieved through p53-mediated transcriptional repression of promitotic genes (31, 32). Live-cell microscopy has shown that p53 levels oscillate in response to DNA damage and that the amount of oscillations is dependent on the amount of DNA damage (232). The oscillatory pattern of p53 is dependent on the type of DNA damage present in the cell (33), and changing the oscillatory dynamics of p53 can readily influence cell fate decisions (233). p53 can promote a permanent cell cycle arrest in response to prolonged exposure to DNA damage in G2 (34), and is required for the induction of senescence in response to continuous stimuli such as critically short telomeres or oncogene activation (35). Increasing overall p53 activation results in a failure to sustain sufficient levels of promitotic genes, resulting in a loss of recovery competence in G2 cells (174). This could set the stage for a permanent cell cycle withdrawal. However, it remains unclear how cellular fate is controlled in response to DNA damage in individual cells at different stages of the cell cycle (36). We therefore set out to study the effects of DNA damage on cell fate decisions in single cycling cells. 3 Results Limited reversibility of a G2 checkpoint arrest To visualize cell fate decisions in individual cells following a DNA damaging insult, we took advantage of the Fucci system, which uses fluorescent proteins fused to the degradation motifs of Cdt1 and Geminin, to mark G1 and S/G2 cells, respectively (Figure S1A) (37). As mutations in checkpoint components are common in transformed cells, we made use of untransformed, immortalized retinal pigment epithelial cells (RPE-1), which have a functional G1 and G2 checkpoint in response to DNA damage (Figure S1B). Irradiation (IR) of RPE-Fucci cells with 4 Gy resulted in spontaneous recovery of approximately 5% of the G1 cells in the 7 h following the damage, determined by S phase entry (Figure1A, yellow line). In contrast, only ~25% of the G2 cells spontaneously recovered from this insult (Figure1B, yellow line). The other ~75% of the G2 cells did not recover, but lost expression of Geminin and acquired a G1-like state without proceeding through mitosis (Figure S1C). Interestingly, all of the G2 cells that recovered during the course of the experiment did so within the first 24 h after DNA damage, whereas damaged G1 cells continued to recover throughout the experiment (Figures 1A,B, yellow lines). This implies that cell fate in the G2 cell is determined within 1 day after DNA damage, while the damage-induced cell cycle arrest remains reversible for much longer in G1 cells. Indeed, we found that checkpoint silencing at 24 h after the damaging insult failed to promote cell cycle reentry of G2-arrested cells (Figure 1B, blue line), while it did cause all of the cells arrested in G1 to reenter the cell cycle (Figure 1A, blue line). Addition of checkpoint inhibitors immediately following IR caused all G1 and G2 cells to enter the next cell cycle phase, indicative of efficient checkpoint silencing upon the addition of inhibitors (Figures 1A,B, red lines). Similar results were obtained when damaging RPE-Fucci cells with a lower dose of IR (Figures S1D,E), and in BJ-Tert cells (another untransformed, immortalized cell line) expressing the Fucci-system (BJ-Fucci) (Figures S1F,G). Collectively, these results show that reversibility of the checkpoint response in G2 cells is restricted to a much shorter time period following DNA damage when compared to G1 cells. To investigate if the prolonged cell cycle arrest observed in the G2 cells results in a permanent cell cycle exit, we analyzed the cell cycle distribution of RPE-1 cells up to 1 days post IR. IR resulted in a marked Transient activation of p53 in G2 phase is sufficient to induce senescence 37

38 increase in the 4N fraction, and this fraction remained constant over time (Figure S1H). To investigate if these cells undergo senescence, we irradiated RPE-Fucci cells using 2 or 4 Gy and subsequently isolated the cells in G2 (expressing green fluorescent mag-geminin) by fluorescence-activated cell sorting (FACS). Six days later, we determined the survival of these cells using a colony formation and additionally stained for senescence-associated β-galactosidase (β-gal) (Figure 1C), a widely used marker for cellular senescence (36). IR of G2 cells with 2 or 4 Gy severely reduced cell growth compared to 3 A S-phase entry (% of G1 cells) RPE-Fucci G1 recovery 4 Gy No Checkpoint Silencing 2 Checkpoint Silencing h ad Checkpoint Silencing 24 h ad Time after γ-irradiation (h) B 1 Mitotic entry (% of G2 cells) RPE-Fucci G2 recovery 4 Gy No Checkpoint Silencing Checkpoint Silencing h ad Checkpoint Silencing 24 h ad Time after γ-irradiation (h) C RFP RPE-Fucci / 2 / 4 Gy GFP Sort 6 Days post IR Gy 2 Gy 4 Gy Colony Formation β-gal D 1 RPE-Fucci 2 3 h after 1 Gy γ-irradiation E Cell 1 Cell 2 Cell 3 1 Gy Time after γ-irradiation (h) F G2 SA-β-Gal+ cells (% of G2 cells) Gy 1 Gy G Gy 1 Gy 144 hrs after 1 Gy γ-irradiation Figure 1. Limited reversibility of a G2 checkpoint arrest (A) RPE-Fucci cells were subjected to 4 Gy IR and subsequently imaged every 3 min for 7 h. Checkpoint inhibitors were added at different time-points following IR, and progression from G1 into S phase was monitored. Cells entering S phase within the first 4 h after IR were excluded from analysis, as they have already passed the G1 restriction point at the time of IR. A representative experiment is shown. (B) Same as in (A), except progression from G2 into mitosis was analyzed. Mitotic cells at the moment of IR were excluded. A representative experiment is shown. (C) Irradiated RPE-Fucci cells in G2 were isolated using FACS. The G2 cells were replated, and colony formation and β-gal were analyzed 6 days later. n = 3, representative experiment is shown. (D G) RPE-Fucci cells were subjected to 1 Gy IR, imaged every hour for 6 days, and subsequently stained for β-gal. Numbered cells in (D) were in G2 at the moment of IR. (E) Representative images of cells indicated in (D). (F) Quantification of irradiated G2 cells that arrest for 6 days and become β-gal-positive. Means ±SD, n = 2 (G) Representative images of control and 1 Gy irradiated cells at 6 days after IR stained for β-gal. 38 Chapter 3

39 non-irradiated G2 cells (Figure 1C). Strikingly, irradiated G2 cells show a marked increase in the amount of β-gal-positive cells 6 days after IR (Figure 1C). Using time-lapse analysis, we could confirm that the majority of irradiated G2 cells failed to enter mitosis for up to 6 days and entered senescence from their G2-arrested state (Figures 1D-G). Collectively, these data show that the decision to enter senescence is established within 24 h in a damaged G2 cell. To determine the exact timing of this decision, we damaged asynchronously growing RPE-1 cells by IR or treated cells with the topoisomerase-ii inhibitor etoposide. To determine the fate of cells damaged in G2, BrdU was added to be able to exclude cells damaged in S phase. Subsequently, the checkpoint was silenced at different time points following DNA damage by the addition of caffeine (Figure 2A). G2 recovery was analyzed as the percentage of 4N BrdU-negative cells that were positive for the mitosis-specific phosphorylation of histone H3 on serine 1 (Figure 2B). We observed that, relative to non-irradiated control cells, approximately 15% of the G2 cells recover spontaneously from 5 Gy IR (Figure 2C, no caffeine). Checkpoint silencing at 1 h after IR results in near complete recovery, but the ability to induce checkpoint recovery in G2 cells decreases rapidly over time and is almost completely lost 5 h after IR (Figure 2C). We observed a similar decrease in checkpoint recovery over time in both RPE-1 and BJ-Tert cells after treatment with etoposide (Figure 2D and S2), although the time at which the response becomes irreversible is somewhat delayed in BJ-Tert cells. Taken together, these results show that untransformed cells progressively lose the ability to recover from a DNA damage-induced G2 arrest within the first hours after the damage, causing them to enter a state of senescence. In contrast, G1 cells retain the capacity to reenter the cell cycle after DNA damage for a much longer period of time. 3 Irreversible withdrawal from the cell cycle in G2 depends on p53 p53-mediated repression of mitotic regulators can result in the loss of recovery competence in transformed cell lines (174). In addition, p53 is required for DNA damage-induced senescence (36). Therefore, we investigated whether the establishment of the irreversible G2 arrest observed in the RPE-1 cells is dependent on p53. We observed that recovery competence was fully sustained in the RPE-1 cells after p53 knockdown (kd) (Figure 2E, red bars). In fact, depletion of p53 caused the majority of G2 cells to spontaneously reenter the cell cycle after 5 Gy of IR, even without the addition of caffeine (Figure 2E). Although spontaneous recovery of p53 kd RPE-1 cells was hardly observed following 1 Gy of IR, the addition of caffeine, even at 4 h after IR, was sufficient to drive most of these cells back into the cell cycle (Figure 2E). Similarly, RPE-1 cells depleted of p53 fully retained their ability to reenter the cell cycle when damaged with etoposide (Figure 2F). Taken together, these results show that cell fate decisions after DNA damage in G2 are largely determined by p53. Nuclear translocation and degradation of Cyclin B1 precedes cell cycle exit in G2 The observation that G2 recovery competence is tightly regulated by p53, prompted us to investigate how endogenous Cyclin B1 levels are controlled in response to DNA damage in G2. We therefore made use of RPE-1 cells in which a fluorescent tag was introduced in one allele of Cyclin B1 (RPE CCNB1 YFP ). Introduction of this fluorescent tag does not perturb Cyclin B1 function (175). Similarly to G2 RPE-Fucci cells, IR of G2 RPE CCNB1 YFP cells, isolated by FACS sorting, resulted in impaired outgrowth and the induction of senescence (Figure 3A). Additionally, we find that the RPE CCNB1 YFP cells lose the ability to recover from a G2 arrest over time in a similar manner to wild type RPE-1 cells (Figure S3A). Analysis of Cyclin B1 by time-lapse fluorescence microscopy confirmed that Cyclin B1-YFP accumulates during G2, is degraded at the metaphase-to-anaphase transition, and it is predominantly localized in the cytoplasm until the beginning of mitosis, when it rapidly translocates to the nucleus just prior to nuclear envelope breakdown (38) (Figures S3B, S3C). Interestingly, in response to DNA damage we could discriminate different patterns of Cyclin B1 behavior. In cells that failed to recover from the G2 arrest, we observed a near-complete translocation of Cyclin B1 to the nucleus 2-4 h after IR, followed by a rapid decrease in Cyclin B1 between 5-1 h after IR (Figure 3B, upper panels). In contrast, cells that eventually reentered the cell cycle retained Cyclin B1 in the cytoplasm and maintained Cyclin B1 at high levels, only to trans- Transient activation of p53 in G2 phase is sufficient to induce senescence 39

40 A Irradiation / Etoposide B Asynchronous Cells No Caffeine (2 hrs) Caff 1 h after damage (2 hrs) Caff 2 hrs after damage (2 hrs) BrdU ph3 No drugs BrdU and Nocodazole Caff 3 hrs after damage (2 hrs) Caff 4 hrs after damage (2 hrs) Caff 5 hrs after damage (2 hrs) PI PI C D E 3 Relative Mitotic Entry (% of G2 cells) F Relative Mitotic Entry (% of G2 cells) Undamaged 1 h 2 h 3 h 4 h 5 h No Caffeine Relative Mitotic Entry (% of G2 cells) Undamaged Caffeine 5 Gy γ-irradiation 1 h 4 h No Caff Caff. Etop. Undamaged 1 h 4 h No Caff Caff. Etop Undamaged 1 h 2 h 3 h 4 h No Caffeine Caffeine Etoposide (1 μm) sirna: Luciferase p53 Relative Mitotic Entry (% of G2 cells) Undamaged 1 h 4 h No Caff Caff Undamaged 1 h 4 h No Caff Caff Undamaged 1 h 4 h No Caff 5 Gy 5 Gy 1 Gy 1 Gy sirna: Luciferase p53 Figure 2. Irreversible withdrawal from the cell cycle in G2 depends on p53 (A) G2 recovery assay. Asynchronously growing RPE-1 cells were damaged using IR or 1 h etoposide treatment. Subsequently, BrdU was added to exclude cells damaged in S phase, and nocodazole was added to trap recovering cells in mitosis. Cells were harvested 2 h after the addition of caffeine. (B) FACS analysis of cells treated as in (A), stained for DNA, BrdU, and phosphorylated histone 3, a marker for mitosis. (C) G2 recovery of 5-Gy-irradiated RPE-1 cells ± caffeine, relative to unperturbed mitotic entry, as measured by FACS. Caffeine was added at the indicated times after IR. (D) Same as in (C), except etoposide was used to induce DNA damage. (E) RPE-1 cells were transfected with the indicated sirnas and irradiated with 5 or 1 Gy 24 h later. G2 recovery was measured as described before. (F) Same as in (E), except the cells were damaged using etoposide. Bars represent the means ±SD, N=3. Caff Undamaged 1 h 4 h No Caff Caff sirna: p53 CDK4 Luciferase p53 locate it to the nucleus immediately prior to the onset of mitosis (Figure 3B, lower panels). A minority (<1%) of the cells lost Cyclin B1 rapidly following IR. We next asked if the differential behavior of Cyclin B1 is associated with the distinct cell fate decisions. When using a high dose (5 Gy) of IR, none of the cells reenter the cell cycle, whereas ~2% of the cells reenter the cell cycle when damaged with 2 Gy (Figure 3C, blue and red lines respectively). After 5 Gy of IR, all of the G2 cells translocated Cyclin B1 to the nucleus within 4 h after the damage, while with 2 Gy ~2% of the G2 cells still retained Cyclin B1 in the cytoplasm at this time (Figure 3D, solid blue and red lines, respectively). These latter cells all reentered the cell cycle several hours later (Figure 3B, lower panels, and 3C, red line), while all the cells in which Cyclin B1 was translocated to the nucleus eventually lose Cyclin B1 and remained arrested in G2 (Figure 3B, top panels, and 3D, dotted blue and red lines). The fraction of cells that translocated Cyclin B1 to the nucleus was dependent on the dose of IR (Figure 3D and 3E), which is consistent with a dose-dependent loss of cell cycle reentry (Figure 3C). A similar DNA damage-induced nuclear translocation of Cyclin B1 was observed in BJ-Tert cells (Figure S3D), although the time at which Cyclin B1 accumulates in the nucleus is delayed compared to RPE-1 4 Chapter 3

41 A RFP intensity (log) RPE CCNB1 YFP Sort-Gate 6 days post IR Gy 5 Gy Colony Formation β-gal YFP intensity (log) B Time after IR (2 Gy) RPE CCNB1 YFP (hh:mm): :15 1:15 2:15 3:3 4:45 5:45 6:45 8:15 9:15 9:3 9:45 2 Gy DIC YFP DIC YFP Nuclear Translocation followed by Degradation Mitosis 3 C D Mitotic Entry Cyclin B1 Nuclear Retention and Degradation 1 1 % of G2 Cells Gy 2 Gy 5 Gy Time after γ-irradiation (h) % of G2 Cells Gy 2 Gy 5 Gy Time after γ-irradiation (h) Nucl. Retention Degradation E Cyclin B1 localization (% of G2 cells) RPE (wt) Undamaged 1 Gy 2 h 4 h 2 Gy 2 h 4 h 3 Gy 2 h 4 h 4 Gy 2 h 4 h 5 Gy 2 h 4 h Nucleus Cytoplasm Figure 3. Nuclear retention and subsequent degradation of Cyclin B1 in response to DNA damage (A) RPE CCNB1 YFP cells were treated as in Figure 1C. (B) Time-lapse images of RPE CCNB1 YFP cells following 2 Gy IR. In the majority of cells, Cyclin B1 translocates to the nucleus, which is followed by loss of Cyclin B1 (top panels). A subset of cells initially retains Cyclin B1 in the cytoplasm (bottom panels). In these cells, Cyclin B1 translocates to the nucleus only at the onset of mitosis. (C and D) RPE CCNB1 YFP cells were irradiated with the indicated dose and analyzed by time-lapse microscopy. Mitotic entry of cells irradiated in G2 is shown in (C). Nuclear retention and subsequent degradation of Cyclin B1 (in the same cells) is shown in (D). Bars represent means ±SD, n = 3. (E) RPE-1 cells were irradiated with increasing doses of IR and fixed 2 h or 4 h later. EdU was added directly following IR to exclude cells damaged in S phase. Cyclin B1 localization of G2 cells was visualized using immunofluorescence. Bars represent means ±SD, N=3. cells. Together, these results suggest that DNA damaged-induced nuclear translocation of Cyclin B1, observed previously by others (24, 39), sets the stage for an irreversible cell cycle withdrawal in G2. This is in stark contrast to the nuclear translocation of Cyclin B1 seen at the end of G2 in unperturbed cells, where the nuclear translocation of Cyclin B1 triggers the onset of mitosis. Activation of the APC/C Cdh1 in response to DNA damage in G2 It is well established that p53 and its transcriptional target p21 are both required for the maintenance of a DNA damage-induced G2 arrest (85) and promote activation of APC/C Cdh1 after damage (235, 236). We Transient activation of p53 in G2 phase is sufficient to induce senescence 41

42 A Cyclin B1-YFP degradation (% of G2 cells) sirna: RPE CCNB1 YFP Luc p53 p21 Cdh1 sirna: Cdh1 p53 p21 CDK4 Luciferase p53 p21 Cdh1 B Relative cyclin B1-YFP Fluorescence Intensity 1.5 RPE CCNB1 YFP 1. sirna: Luciferase.5 p53 p21 Cdh Time after 5 Gy γ-irradiation (h) C Time after IR (5 Gy) RPE CCNB1 YFP (hh:mm): :15 2: 4: 7: 1: Luc sirna: p21 p53 Cdh1 3 D Nuclear cyclin B1-YFP (% of G2 cells) sirna: RPE CCNB1 YFP Luc p53 p21 Cdh1 E Cyclin B1 Localization (N/C) (Log2) IR: sirna: Gy Luc 5Gy Luc 5Gy p21 RPE (wt) Nuclear cyclin B1 Cytoplasmic cyclin B p21 fluorescence intensity (a.u.) F p21 fluorescence Intensity (a.u.) RPE (wt) Undamaged 1 h 2 h 3 h 4 h 5 h 6 h 1 h 2 h 3 h 4 h 5 h 6 h 2 Gy 5 Gy G RPE CCNB1 YFP H Nuclear Cyclin B1-YFP (% of G2 cells) DMSO RO336 BI Time after 5 Gy γ-irradiation (hours) Fluorescence Intensity (a.u.) RPE CCNB1 YFP IR : Gy 5 Gy Gy Prophase (n=26) G2 - Nuclear CyclinB1 (n=48) Cyclin B1 (pan) Cyclin B1 ps126 G1 (n=64) I TranslocationProphase G1 DAPI Cyclin B1 Cyclin B1 S126P Merge therefore hypothesized that the loss of Cyclin B1 in response to DNA damage might be caused by APC/ C Cdh1 activation, leading to Cyclin B1 degradation, possibly triggering a permanent cell cycle exit. To test this, RPE CCNB1 YFP cells depleted of p53, p21 or Cdh1 were irradiated with 5 Gy, a dose at which most control cells fail to reenter the cell cycle (see Figures 2C, 3C). Time-lapse analysis showed that nearly all control cells degraded Cyclin B1 in response to 5 Gy IR (Figure 4A). Depletion of p53 and p21 dramatically reduced the percentage of cells that degraded Cyclin B1, while depletion of Cdh1 almost completely prevented degradation of Cyclin B1 (Figure 4A). We observed that Cyclin B1 levels are completely lost at 1 h after IR in RPE CCNB1 YFP cells transfected with control sirna (Figure 4B, yellow line). In contrast, the large fraction of the p53- and p21-depleted cells that failed to degrade Cyclin B1 (Figure 4B) maintained 42 Chapter 3

43 Figure 4. Activation of APC/C Cdh1 and nuclear translocation of Cyclin B1 is p53/p21 dependent (A) RPE CCNB1 YFP cells treated with luciferase, p53, p21 or Cdh1 sirna were irradiated (5 Gy) after which Cyclin B1 YFP translocation and degradation were visualized by time-lapse microscopy. The western blot shows the relative expression levels of the indicated proteins at the moment of IR. Means ±SD, n = 3. (B) Quantification of Cyclin B1 YFP intensities in the large fraction of cells in (A). Means ±SD, N=3. 1 cells quantified per condition per experiment. (C) Representative images of RPE CCNB1 YFP cells, depleted of the indicated proteins and irradiated using 5 Gy. (D) Subcellular localization of Cyclin B1 was quantified for the cells in (C). Means ±SD, n = 3. (E) RPE-1 treated with luciferase or p21 sirnas were subjected to 5 Gy IR and fixed 4 h later. The cells were labeled for EdU, Cyclin B1 and p21. S phase cells were excluded. The ratio of nuclear over cytoplasmic Cyclin B1 is plotted against p21 levels. As most irradiated control cells have retained Cyclin B1 in the nucleus at 4 h following IR, relatively more cells with cytoplasmic Cyclin B1 were included in the analysis of luciferase treated, 5 Gy irradiated cells. (F) RPE-1 cells were irradiated (2 or 5 Gy) and p21 levels in G2 cells were visualized using immunofluorescence. EdU was added directly following IR to exclude cells damaged in S phase. In this experiment, at least 5 cells per time point. Error-bars indicate mean ±SD. (G) RPE CCN- B1 YFP depleted of Cdh1 were irradiated using 5 Gy. Immediately thereafter, BI2536 or RO336 were added to inhibit Plk1 or Cdk1, respectively. The cells were followed for 1 h, and nuclear translocation of Cyclin B1 was monitored. (H and I) Nuclear Cyclin B1 levels and the levels of Cyclin B1 S126 phosphorylation were analyzed for undamaged prophase cells, G1 cells and irradiated G2 cells, 4 h after IR. Cyclin B1 at a constant level throughout the experiment (Figure 4B, red and blue lines respectively). In addition, Cyclin B1 degradation was also dramatically decreased in Cdh1-depleted cells, although we did observe a gradual decrease in Cyclin B1 levels at later times after damage (Figure 4B, green line), most likely due to p53- and prb-mediated transcriptional repression (31). Next, we obtained G2 synchronized RPE CCNB1 YFP cells by FACS sorting of YFP-positive cells and monitored expression of other APC/C Cdh1 substrates after 5 Gy IR. Like Cyclin B1, Plk1 and Aurora A were also degraded in response to DNA damage (Figure S4A) in a Cdh1-dependent manner (Figure S4B). However, degradation of APC/C Cdh1 targets only becomes apparent around 6 h after DNA damage (Figure 4B, S4A), at time at which RPE-1 cells have already lost their ability to revert the DNA damage-induced G2 arrest (see Figures 2C,D). Thus, our data suggest that the decision to permanently withdraw from the cell cycle precedes DNA damage-induced degradation of Cyclin B1. 3 DNA damage-induced nuclear translocation of Cyclin B1 is dependent on p21 Nuclear accumulation of Cyclin B1 in response to DNA damage has been shown to require p53 and p21 (24). In addition, we have shown that overexpression of p21 in U2OS cells causes nuclear translocation of Cyclin B1 and arrests cells in G2 (31). We therefore asked if the DNA damage-induced nuclear translocation of Cyclin B1 in RPE-1 cells is dependent on p21. To examine this, we depleted RPE CCNB1 YFP cells of p53, p21 or Cdh1 and visualized Cyclin B1 localization in response to 5 Gy IR. We observed that depletion of p53 or p21 completely abolished the nuclear accumulation of Cyclin B1, whereas depletion of Cdh1 did not (Figure 4C and 4D). Analysis of p21 protein levels by immunofluorescence showed that the nuclear translocation of Cyclin B1 after damage is limited to the cells that express high levels of p21 (Figure 4E). This indicates that the DNA damage-induced nuclear translocation of Cyclin B1 only occurs in those cells in which p21 has reached sufficiently high levels. This implies that the relative induction of p21 by p53 is decisive for nuclear translocation of Cyclin B1. In agreement with this notion, we find that the kinetics and magnitude of p21 upregulation after DNA damage is dose and time dependent (Figure 4F), similar to what we find for the nuclear accumulation of Cyclin B1 (Figure 3D and 3E). DNA damage-induced nuclear translocation of Cyclin B1 does not require kinase activity of Plk1 and Cdk1 The nuclear translocation of Cyclin B1 that occurs at the onset of mitosis requires the phosphorylation of a cytoplasmic retention signal by Plk1 and Cdk1 (239, ). Addition of a selective kinase inhibitor for Plk1 did not affect nuclear translocation of Cyclin B1 after 5 Gy IR (Figure 4G, blue line). Inhibition of Cdk1 resulted in a somewhat delayed nuclear accumulation of Cyclin B1 in the majority of G2 cells and prevented nuclear accumulation in ~2% of G2 cells when using 5 Gy (Figure 4G, red line). This shows that Plk1 and Cdk1 activity are not essential for the nuclear translocation of Cyclin B1 after DNA damage, but does indicate that Cdk1 somehow promotes the translocation. The latter effect might Transient activation of p53 in G2 phase is sufficient to induce senescence 43

44 3 A Mitotic entry (% of G2 cells) D Mitotic entry (% of G2 cells) G Undamaged 4 sirna: 2 Luciferase Cdh Time after mock irradiation (h) sirna: CDK4 J K Undamaged 4 sirna: Luciferase 2 p53 p Time after mock irradiation (h) Cdh1 p53 p21 Luciferase p53 p21 Cdh1 Relative Mitotic Entry (% of G2 cells) Caff B Mitotic entry (% of G2 cells) Mitotic entry (% of G2 cells) Undamaged 1 h 4 h No Caff sirna Infect Infect Sample Sample 24 h 24 h 24 h 24 h pbabe Empty pbabe Ras G12V H Start puro-selection DAPI Cyclin B1 Merge 48 h post infection 5 Gy L Gy sirna: Luciferase Cdh Time after γ-irradiation (h) E 2Gy 1 sirna: Luciferase 8 p53 p21 6 Nuclear Cyclin B1 (% of G2 cells) Time after γ-irradiation (h) I Undamaged 1 h h Caff 5 Gy No Caff sirna: Cdh1 CDK4 pbabe: E R E R 24 h 48 h Luciferase Cdh1 Luciferase Cdh1 M p21 CDK4 pbabe: CyclinB1-YFP localization (% recovering cells) C Mitotic entry (% of G2 cells) Mitotic entry (% of G2 cells) Gy sirna: Luciferase Cdh Time after γ-irradiation (h) F 5Gy 1 sirna: Luciferase 8 p53 p Time after γ-irradiation (h) Recover Arrest sirna: E 48 h Luc Cdh1 R Cytoplasmic Nuclear CyclinB1-YFP localization (% arrested cells) sirna: Luc Cdh1 Cytoplasmic Nuclear 44 Chapter 3

45 very well be indirect, as Cdk1 inhibition has been shown to result in attenuation of p21 induction upon p53 activation (314). To further strengthen the notion that DNA damage-induced nuclear accumulation of Cyclin B1 does not require Cdk1 activity, we investigated if Cyclin B1 was phosphorylated at serine 126 (S126), a known Cdk1 site associated with nuclear translocation of Cyclin B1 in prophase (313). Indeed, nuclear Cyclin B1 in prophase cells was phosphorylated on S126 (Figure 4H and 4I). Significantly, irradiation-induced, nuclear-translocated Cyclin B1 was not phosphorylated at S126 (Figure 4H and 4I), indicating that the mechanism responsible for the nuclear translocation of Cyclin B1 in damaged cells is different from the mechanism in prophase. Also, these data indicate that there is no Cdk1 kinase activity associated with the Cyclin B1 that translocates to the nucleus in response to DNA damage. Levels of p21 in the nucleus are dramatically increased after IR (Figure S4C), suggesting that p21 may act to sequester Cyclin B1 in the nucleus. To examine this in further detail we irradiated p21-depleted RPE CCNB1 YFP cells with 5 Gy and immediately treated them for 2 h with leptomycin B (LMB), an inhibitor of Crm1-mediated nuclear export. While the depletion of p21 resulted in a failure to accumulate Cyclin B1 in the nucleus after damage, LMB treatment resulted in the rapid nuclear accumulation of Cyclin B1, even in the absence of p21 (Figure S4D). Removal of LMB resulted in redistribution of Cyclin B1 back to the cytoplasm (Figure S4D), indicating that permanent nuclear retention after DNA damage depends on p21. These observations are most consistent with the notion that Cyclin B1 continuously shuttles between the cytoplasm and the nucleus even in damaged cells, but that p21 binds Cyclin B1 (Figure S4E), and thereby prevents nuclear export of Cyclin B1. At high p21 levels, this results in nuclear retention of Cyclin B1 and a failure to reenter the cell cycle. Thus, DNA damage-induced retention of Cyclin B1 occurs by a fundamentally different mechanism than the translocation of Cyclin B1 that is observed at the onset of mitosis. Also, the DNA damage-induced nuclear retention, unlike translocation in an unperturbed cycle, results in high levels of nuclear Cyclin B1 without associated Cdk1 activity. This DNA damage-induced Cyclin B1 regulation mechanism is independent of Plk1 and Cdk1 yet depends on p53 and p21. 3 Nuclear retention of Cyclin B1 is the first irreversible step in a DNA damage-induced G2 checkpoint response To determine if the nuclear retention or the degradation of Cyclin B1 is the decisive event for the permanent cell cycle exit induced by DNA damage in G2 cells, we depleted Cdh1 from RPE CCNB1 YFP cells and asked if this could circumvent the permanent withdrawal from the cell cycle. Cdh1 depletion did not influence mitotic entry in unperturbed cells (Figure 5A and 5G). Additionally, cell cycle reentry was unaltered in Cdh1-depleted cells following 2 or 5 Gy IR (Figure 5B and 5C), indicating that the loss of reversibility does not depend on the degradation of Cyclin B1. Depletion of Cdh1 does not interfere with the nuclear retention of Cyclin B1 (Figure 4C and 4D), and therefore, we asked if the damage-induced nuclear retention of Cyclin B1 could drive cell cycle exit. To examine this, we depleted p53 or p21 (Figure 5G), since both are essential for the nuclear retention of Cyclin B1 (Figure 4C and 4D). Mitotic entry in Figure 5. Nuclear retention of Cyclin B1 is the first irreversible step in a DNA damage-induced G2 checkpoint response (A C) Control RPE CCNB1 YFP cells and cells depleted of Cdh1 were irradiated with, 2 or 5 Gy and subsequently imaged for 24 h. Mitotic entry of undamaged cells (A), 2 Gy- (B), or 5 Gy-irradiated cells (C) is shown. (D F) Same as before, except RPE CCNB1 YFP cells were transfected with sirnas targeting luciferase, p53 or p21. Mitotic entry was scored for undamaged cells (D), 2 Gy- (E) or 5 Gy-irradiated cells (F). (G) Western blot analysis shows the relative expression levels of the indicated proteins in RPE CCNB1 YFP cells used in (A F). (H) Parental RPE-1 cells transfected with the indicated sirnas were treated as in Figure 2A. Inset shows expression levels of the indicated proteins at the moment of IR. (I) RPE CCNB1 YFP cells depleted of the indicated proteins were irradiated using 5 Gy and subsequently imaged for 24 h. Caffeine was added to the medium 4 h after IR. Cyclin B1 localization at the moment of caffeine addition in cells that do recover (left) and do not recover (right) from the G2 arrest is shown. (J) Experimental setup to determine nuclear retention of Cyclin B1 in oncogene-induced senescence. (K) Cells were treated as in (J) and stained for Cyclin B1 and DNA. (L) Quantification of cells treated as in (J). (M) Western blot analysis of p21 expression in cells infected with virus carrying a control vector or a vector containing H-RasV12 at 48 h post infection. Bars represent means ±SD, n = 3. Transient activation of p53 in G2 phase is sufficient to induce senescence 45

46 3 non-irradiated cells was unaltered (Figure 5D). However, depletion of p53 or p21 significantly enhanced the percentage of cells that reentered the cell cycle after 2 Gy of IR (respectively, 7% of p53-depleted cells and 6% of p21 depleted-cells entered mitosis, versus 3% of control cells) (Figure 5E). At 5 Gy of IR, only 7% of the control and Cdh1-depleted cells entered mitosis after an average arrest of 8 h (Figure 5C), whereas up to 5% of the p53- or p21-depleted cells reentered the cell cycle from as early as 4 h after IR (Figure 5F). Thus, preventing the degradation of Cyclin B1 is insufficient to maintain reversibility. However, depletion of p53 or p21 prevents nuclear retention of Cyclin B1 and maintains reversibility of the G2 arrest. This indicates that the nuclear retention of Cyclin B1 after DNA damage coincides with the decision to permanently withdraw from the cell cycle, triggering the onset of a senescent state. Also, depletion of Cdh1 from wild-type RPE-1 cells did not lead to increased recovery upon checkpoint silencing both at early or late time points (Figure 5H, red bars), strengthening the notion that APC/C Cdh1 activation is not the decisive event for the loss of reversibility. Consistent with this, we found that almost all of the control and Cdh1-depleted cells that do recover upon checkpoint silencing 4 h after IR, had Cyclin B1 in the cytoplasm at the moment of caffeine addition (Figure 5I, left graph and S5). Conversely, the control and Cdh1-depleted cells that failed to recover all had Cyclin B1 in the nucleus at the time of caffeine addition (Figure 5I, right graph and S5). This indicates that nuclear retention of Cyclin B1, rather than APC/C Cdh1 activation, constitutes the first step towards an irreversible G2 arrest in response to DNA damage. Nuclear retention of Cyclin B1 during oncogene-induced senescence To investigate whether nuclear retention of Cyclin B1 also occurs under different conditions that induce senescence, we asked if oncogene-induced senescence is associated with nuclear retention of Cyclin B1. Introduction of oncogenic H-Ras is known to induce senescence in untransformed cells, accompa- A 1 Mitosis (% G2 cells) E Undamaged 2 DMSO MK Time after drug addition (h) :15 :15 3: 3:15 6: 3: 3:15 6: B Mitosis (% G2 cells) 1 8 DMSO MK MK1775 8: * 8: 5 Gy Time after γ-irradiation (h) 8:45 * 9: F Cyclin B1 CDK1 Y15P 3 5 Time after 5 Gy IR (h) Input (1%) IP Cyclin B1 p21 RO 336: MK1775: h 1 h 5 h.5 h 1 h 5 h Gy 5 Gy Gy 5 Gy Figure 6. Nuclear Cyclin B1-CDK1 complex is refractory to activation (A) Mitotic entry of undamaged RPE CCNB1 YFP cells treated with DMSO or MK1775. (B) RPE CCNB1 YFP were irradiated with 5 Gy and subsequently imaged for 24 h. DMSO or MK1775 was added to the medium 3 h after IR. (C) Cyclin B1-YFP localization of MK1775 treated cells in (B) was determined before and 3 or 5 h after IR. (D) Mitotic entry of MK1775 treated cells in (B) was analyzed for cells with nuclear or cytoplasmic Cyclin B1 at the moment of drug addition. (E) Representative images of MK1775-treated cells in (B). Asterisk mark cells going through mitosis. Arrows point toward the daughter cells. (F) Western blot analysis of the indicated proteins of RPE-1 cells that were synchronized in G2 by 8 h thymidine release. At this time the cells were irradiated with or 5 Gy, and RO336 and MK1775 were added to the medium at the indicated time following IR. 1.5 h after drug addition, the cells were harvested and processed for immunoprecipitation of Cyclin B1. Bars represent means ±SD, n = 3. C CyclinB1-YFP localization (% of G2 cells) Cyclin B1-YFP nuclear Cyclin B1-YFP cytoplasmic D Recovery according to Cyclin B1-YFP localization (% of G2 cells) MK1775: + + Cyclin B1-YFP nuclear Cyclin B1-YFP cytoplasmic 46 Chapter 3

47 A Mitosis (% of G2) RPE CCNB1 YFP 4 DMSO 2 5uM Nutlin 1uM Nutlin Time after Nutlin addition (h) B Nuclear Retention of cyclin B1 (% of G2) RPE CCNB1 YFP DMSO 5uM Nutlin 1uM Nutlin Time after Nutlin addition (h) Figure 7. Transient activation of p53 is sufficient for senescence induction in G2 (A and B) Asynchronously growing RPE CCNB1 YFP cells were imaged for 1 h after the addition of Nutlin. Mitotic entry (A) and nuclear retention of Cyclin B1 (B) in response to Nutlin addition were scored. (C) RPE-1 cells were treated with Nutlin for the indicted durations. After Nutlin removal BrdU and nocodazole were added, as described before. Mitotic entry is plotted as the percentage of the cells that were in G2 at the time of Nutlin removal. (D) RPE-Fucci cells were treated for 8 h with 1 mm Nutlin, and G2 cells were isolated by FACS, based on green fluorescence. Six days after replating, the cells were stained for β-gal. Representative images are shown. Bars represent means ±SD, n = 3. C Mitotic entry (% of G2 cells) RPE (wt) Nutlin3a: - 2h 4h 8h 5uM Nutlin 1uM Nutlin D DMSO Nutlin RPE-Fucci 6 days post treatment 3 nied by expression of high levels of p53 and p21 (315, 316). We depleted Cdh1 to prevent Cyclin B1 degradation and infected with virus carrying a control vector or a vector containing H-RasV12 (Figure 5J). At day 1 after infection Cyclin B1 was in the cytoplasm, but on day 2 after infection we could observe a marked increase in the fraction of H-RasV12-infected cells that retained Cyclin B1 in the nucleus (Figure 5K and 5L), and this coincided with increased expression of p21 (Figure 5M). This indicates that p21-dependent nuclear retention of Cyclin B1 also sets the stage for an irreversible cell cycle exit induced by oncogene activation. Nuclear Cyclin B1-Cdk1 complex is refractory to activation During G2, the majority of Cdk1 is inactivated by Wee1 and Myt1, which phosphorylate Cdk1 on Y15 and T14, respectively. At the onset of mitosis, Cdk1 is dephosphorylated at these residues, resulting in its activation (4, 5). It has recently been shown that inhibition of Wee1 alone can force Cdk1 activation in interphase cells and stimulates Cdk1 activity in DNA-damaged G2 cells (317). We find that checkpoint inhibition in cells with nuclear Cyclin B1 fails to induce cell cycle reentry (Figure 5I), and therefore, we wondered if nuclear Cyclin B1-Cdk1 complexes become refractory to activation. To this end, we treated cells with a dose of IR that causes nuclear retention in all of the cells but forced activation of Cyclin B1/ Cdk1 at a time when not all cells had completed the nuclear retention process. As expected, inhibition of Wee1 by the small molecule inhibitor MK1775 accelerated mitotic entry of undamaged RPE CCNB1 Y- FP cells (Figure 6A, dotted versus solid lines). Interestingly, addition of MK1775 at 3 h following a high dose IR caused ~3% of the cell to enter mitosis, compared to ~7% in the controls (Fig 6B, dotted lines versus solid lines). Approximately 55% of the cells had translocated Cyclin B1-YFP to the nucleus at this time (Figure 6C), and of these, very few entered mitosis after addition of MK1775 (Figure 6D and 6E). In contrast, the majority of cells in which Cyclin B1 was still in the cytoplasm entered mitosis after addition of MK1775 (Figure 6D and 6E). Thus, nuclear Cyclin B1-Cdk1 complexes appear refractory to activation. To address this more directly, we treated RPE-1 cells with MK1775 in G2, or at 1 or 5 h following 5 Gy IR, and added RO 336 to prevent cells from entering mitosis. Addition of MK1775 in unperturbed G2 cells or at 1hr post IR resulted in Y15-dephosphorylation of Cdk1, both in the total pool, as well as in the pool of Cdk1 associated to Cyclin B1 (Figure 6F). Addition of MK1775 at 5 h post IR, when virtually all of the Cyclin B1 had translocated to the nucleus (Figure 6C), failed to induce Y15-dephosphorylation in the pool of Cdk1 associated to Cyclin B1 (Figure 6F). Interestingly, non-cyclinb1-bound Cdk1 was dephosphorylated in this setting (Figure 6F), indicating that the nuclear Cyclin B1-bound pool is refractory to activation. In accordance with this, we find that this pool is bound to p21 (Figure 6F). Transient activation of p53 in G2 phase is sufficient to induce senescence 47

48 3 Transient activation of p53 is sufficient for senescence induction in G2 Our observation that p53- and p21-dependent nuclear accumulation of Cyclin B1 occurs within 4 h of the damaging insult and constitutes the first irreversible step in a DNA damage-induced G2 arrest indicates that a transient induction of p53 in G2 might be sufficient to trigger the onset of senescence. To address this question, we treated asynchronously growing RPE CCNB1 YFP cells with the p53-activating drug Nutlin (318) and asked if this affected Cyclin B1 localization and mitotic entry. All of the G2 cells treated with DMSO entered mitosis within 8 hr (Figure 7A, yellow line, and S6). Nutlin did not have a large effect on mitotic entry for the first 3 h, but between 3 and 4 h after the addition of Nutlin, further mitotic entry was completely blocked (Figure 7A, red and blue lines, and S6). Thus, a 4 h induction of p53 is sufficient to block mitotic entry in the RPE CCNB1 YFP cells. Importantly, addition of Nutlin caused the nuclear retention and subsequent degradation of Cyclin B1 in all of the cells that failed to enter mitosis (Figure 7B, red and blue lines, and S6). Next, we wanted to test if a transient activation of p53 in G2 is sufficient to induce a permanent withdrawal from the cell cycle. For this purpose, we treated asynchronously growing RPE-1 cells with Nutlin for different amounts of time and asked if the arrest was reversible. To track the G2 cells treated with Nutlin, we added BrdU at the moment of Nutlin washout to discriminate S phase cells from G2 cells. We found that mitotic entry was reduced to approximately 4% after 4 h of treatment with Nutlin, while treatment for 8 h reduced mitotic entry of G2 cells to approximately 15% (Figure 7C). Finally, we treated RPE-Fucci cells with Nutlin for 8 h and isolated G2 cells by FACS sorting. After replating, the cells were allowed to grow for 6 days before they were subjected to β-gal staining. Very few β-gal-positive cells were observed in the control G2 cells, whereas a significant amount of the G2 cells treated with Nutlin for only 8 h became β-gal positive (Figure 7D). Since a subset of the G2 cells treated with Nutlin for 8 h still enters mitosis (Figure 7C), the actual percentage of G2 cells that become senescent after 8 h of Nutlin treatment is therefore much higher than the relative fraction that is detected after 6 days of culture. Collectively, these data show that in untransformed cells a transient induction of p53 in G2, in the absence of a DNA-damaging insult, can promote a permanent cell cycle exit and establish cellular senescence. Discussion The outcome of a DNA-damaging insult can be variable, but our understanding of the mechanisms underlying damage-induced cell fate decisions is very limited. Here, we have identified a decision point in the DNA damage response in G2 cells, associated with permanent cell cycle withdrawal and the onset of senescence. We find that the decision to exit the cell cycle is rapidly established in a G2 cell, resulting in the onset of senescence. In contrast, a G1 cell retains the ability to revert the G1 arrest for much longer, similar to a state of quiescence, where cells can remain arrested in a G state for long periods of time before reentering the cell cycle (319). Nonetheless, the induction of senescence after DNA damage is not strictly limited to G2 cells (data not shown), which has also been observed during replicative senescence (32). Complete telomere deprotection results in a marked increase of cyclin B1-negative 4N cells, some of which are binucleated or contain internuclear chromatin bridges. (321). This indicates that at least a fraction of these cells did not exit the cell cycle in G2, but withdrew from the cell cycle after passage through mitosis. It should be noted that a permanent withdrawal from the mitotic cell cycle in G2 is not a novel phenomenon. Developmentally programmed polyploidy occurs by loss of Cdk1 activity and an exit from the mitotic cell cycle in G2 (Reviewed in 322). This leads to a switch to endoreduplication, causing the cells to differentiate into polyploid cells. The subsequent endocycles critically depend on periodicity of Cyclin E-Cdk2- and APC/C Cdh1 -activity (322). While we do see that a minority of the G2-arrested cells can enter a subsequent S phase, the majority of cells fail to do so (Figure 1F and S1H). This difference is most likely due to continued DDR signaling in the majority of cells (323), leading to persistent inhibition of Cdk2 and the continued activation of APC/C Cdh1, making it impossible to license origins and initiate a 48 Chapter 3

49 new round of DNA replication. This is similar to control of the proliferation-quiescence decision in undamaged cells in G1. Whereas cells that enter G1 with high Cdk2 activity commit to the next cell cycle, cells entering G1 with low Cdk2 activity enter a transient state of quiescence (21). We find that high levels of p21 in G2 result in nuclear retention of Cyclin B1. Importantly, we show that nuclear retention of Cyclin B1 constitutes the first irreversible step in a cell cycle exit induced by DNA damage in G2, and that Cyclin B1 degradation is not absolutely required to block cell cycle reentry. In line with our observation that nuclear retention of Cyclin B1 upon DNA damage in G2 constitutes a decisive step in cell cycle withdrawal is the finding that nuclear accumulation of Cyclin B1 is associated with IR-induced apoptosis in hematopoietic cells (39). While we find that nuclear retention, rather than degradation of Cyclin B1 is the decisive event, it is clear that degradation can act to enforce the permanent cell cycle exit. Activation of the DDR results in inhibition of Cyclin B/Cdk1, alleviating its inhibitory action towards APC/C Cdh1. APC/C Cdh1 -mediated degradation of Cyclin B1 sets off a negative feedback loop that could promote a switch-like transition from recovery-competent to permanent arrest. Interestingly, Cyclin B1 displays a switch-like localization pattern in response to induction of p21 (Figure 4E), indicating that this also coincides with an all-or-nothing transition point in the cell cycle. Transient induction of p53 in untransformed G2 cells, in the absence of a DNA-damaging insult, is sufficient to trigger the onset of senescence (Figure 7D), while depletion of p53 results in a failure to induce a cell cycle exit in G2 after DNA damage (Figure 2E and 2F). This implies that loss of p53 in an otherwise untransformed cell presents a serious liability for genomic integrity; such a cell will be unable to arrest in G1 and, as we now show, will be unable to exit the cell cycle in G2 after damage. Interestingly, transformed cells such as U2OS and HCT116 (both p53 wild-type) are able to recover from a G2 arrest as late as 18 h after the DNA-damaging insult upon silencing of the checkpoint (174, 189). Additionally, U2OS cells do not display nuclear retention of Cyclin B1 in response to DNA damage (data not shown). Therefore, it is not unlikely that in these transformed cell lines the prolonged reversible G2 arrest merely results from the inability to efficiently induce p21 (255), and thereby, these cells fail to establish a cell cycle exit. Indeed, increasing p53-signaling in these cells results in a loss of G2 recovery competence (174), and overexpression of p21 in U2OS has been shown to result in nuclear retention of Cyclin B1 in G2 (31). This indicates that the cellular machinery required to promote a cell cycle exit in G2 is still present in these transformed cells, yet failure to fully activate the p53 pathway allows them to remain in cycle. Understanding the mechanisms that suppress the p53 pathway in these cell types may help to device novel therapeutic strategies to sensitize tumor cells to DNA-damaging treatments used in the clinic, such as IR or the chemotherapeutic agent etoposide. Although we show that p21 levels need to reach a specific threshold after DNA damage in G2 to result in cell cycle exit, what causes the heterogeneity in the response to DNA damage from cell to cell remains largely unexplored. One possibility is that the DNA repair pathway chosen to repair the various lesions within a cell influences cell fate decisions. In G2, the preferred repair pathway is non-homologous end joining (NHEJ) (324). However, not all DNA lesions induced by γ-irradiation can be repaired by NHEJ, and some are subjected to slower repair trough homologous recombination (HR) (325). Lesions that occur in regions of heterochromatin are preferentially repaired via HR (325, 326), while DNA damage in telomeres has been shown to result in persistent activation of the DDR (258) and induction of senescence (259). Thus, choice of repair pathway and location of the damage can affect the DNA damage load over time, possibly inducing different cell fates. How these aspects of the DDR play a role in cell fate decisions in response to DNA damage will be of interest for future research. 3 Transient activation of p53 in G2 phase is sufficient to induce senescence 49

50 Experimental Procedures 3 Time Lapse Microscopy and Immunofluorescence For live-cell imaging, cells were grown in Lab-Tek II chambered coverglass (Thermo Scientific) in Dulbecco s modified Eagle Medium (DMEM)/F12. Directly before imaging, the medium was replaced by Leibovitz s L-15 (GIBCO) CO 2 -independent medium. Images were obtained using a DeltaVision Elite (Applied Precision) maintained at 37 C equipped with a 1x.4 NA or 2x.75 NA lens (Olympus) and cooled CoolSnap CCD camera. Image analysis was done using ImageJ software. Immunofluorescence was performed as previously described (189). FACS, G2 recovery, Checkpoint inhibition and irradiation FACS analysis was performed as described previously (188), except phospho-histone H3 positivity of 4N BrdU-negative cells was analyzed using flow cytometry of at least 1 4N BrdU-negative cells. G2 recovery is determined as the fraction of 4N, BrdU negative cells that are positive for the mitosis specific marker phospho-histone H3. Chk2 and p38 inhibitors (Chk2 inhibitor II and SB2219, respectively) were used to inhibit the G1 DNA damage checkpoint. For inhibition of the G2 DNA damage checkpoint, caffeine was used. Cells were γ-irradiated using a Gammacell Exactor (Best Theratronics) with a 137 Cssource. FACS-Sort, Colony Formation and Senescence-Associated β-gal assay For sorting of G2 cells, cells were trypsinized and resuspended in ice-cold Leibovitz s L-15 medium. Cell sorts were performed using a FACSAria (BD). Sorted cells were collected in DMEM/F-12 and subsequently replated. For colony formation, 1, cells were plated and allowed to grow for 6 days. Cells were fixed in ice-cold methanol for 1 min and subsequently stained with.1% crystal violet for 5 min. After crystal violet staining, cells were washed extensively with water. β-gal stainings were performed as described previously (327). sirna s, transfections and infections ON-TARGETplus SMARTpool sirnas targeting Luciferase, p53, p21 and Cdh1/FZR1 were from Thermo Scientific and were transfected using RNAiMAX (Life Technologies) according to the manufacturer s protocol in a final concentration of 2 nm. All sirna transfections were performed 24 hours before the start of the experiment. For virus production, amphotropic Phoenix cells were transfected with the indicated plasmids using X-tremeGENE (Roche) according to the manufacturer s protocol. After 2 days, virus containing medium, supplemented with hexadimethrine bromide, was added to target cells for 24 hours. Cell Culture, cell line construction, sirna s and transfections RPE-1 and BJ-Tert cells were maintained in Dulbecco s Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12, Gibco) supplemented with antibiotics and 1% fetal calf serum. RPE-Fucci and BJ-Fucci cell lines were constructed as described by (37). RPE CCNB1 YFP cells have been described before (175). Antibodies, western blots, immunoprecipitation, chemicals and drugs Antibodies against p53 (DO-1), p21 (C-19 and F-5), Cyclin B1 (GNS1), Plk1 (F8) and Cdk4 (C22, goat) were from Santa Cruz. Cyclin B1 ps126 (EPR367) and BrdU (ICR1) were from Abcam. Cdh1 (DH1) was from Neo Marker, Aurora A (AIK) from Cell Signaling and Histone 3 ps1 (6-57) was from Upstate. The following secondary antibodies were used: peroxidase-conjugated goat anti-rabbit, goat anti-mouse, rabbit anti-goat antibodies were from DAKO. Goat anti-rabbit/alexa 488, Goat anti-mouse/alexa 568, goat anti-rabbit/alexa 568, goat anti-mouse/alexa 647, goat anti-rat/alexa 647 and goat anti-rabbit/ Alexa 647 were from Molecular Probes. Western blots and immunoprecipitations were performed as described previously (216). Thymidine, nocodazole, caffeine, leptomycin B, Chk2 inhibitor II, BrdU and hexadimethrine were purchased from Sigma, and used at 2.5 mm, 25 ng/ml, 5 mm, 4 μm, 1 μm and 1 μm and 4μg/ml respectively. RO 336 and SB2219 were from Calbiochem, and both used at 1 ¼M. Nutlin3a (Cayman), BI Chapter 3

51 (Axom Medchem), EdU (Life Technologies) were used at 5 ¼M or 1 ¼M, 1 nm and 1 ¼M respectively. Acknowledgments We thank members of the Medema lab for helpful discussions and the flow cytometry and digital microscopy facilities for technical support. This work was supported by research grants from the NWO Graduate Program (22.1.3), the NWO Gravitation Program (Cancer Genomics Center, CGC.nl), and the Dutch Cancer Society (KWF) (NKI ). 3 Transient activation of p53 in G2 phase is sufficient to induce senescence 51

52 Supplementary Figures A B Undamaged 4Gy γ-ir 3 mag-hgem D 1 S-phase entry (% of G1 cells) M G2 S G1 mko2-hcdt1 C Undamaged 4 Gy IR RPE-Fucci G1 recovery 2 Gy No Checkpoint Silencing Checkpoint Silencing h ad Checkpoint Silencing 24 hrs ad Time after γ-irradiation (h) Counts PI E 1 Mitotic entry (% of G2 cells) 8 6 Counts :15 2:3 4:15 4:3 5:15 :15 6:15 8:45 1:45 14: 3: 5: 7: PI 12:3 2: 25: RPE-Fucci G2 recovery 2 Gy 4 No Checkpoint Silencing 2 Checkpoint Silencing h ad Checkpoint Silencing 24 hrs ad Time after γ-irradiation (h) F 1 S-phase entry (% of G1 cells) 8 BJ-Fucci G1 recovery 4Gy No Checkpoint Silencing Checkpoint Silencing h ad Checkpoint Silencing 8 hrs ad Time after γ-irradiation (h) G BJ-Fucci G2 recovery 4Gy 1 Mitotic entry (% of G2 cells) Time after γ-irradiation (h) No Checkpoint Silencing Checkpoint Silencing h ad Checkpoint Silencing 8 hrs ad H 1 Gy 2 Gy Days post (Mock) IR Figure S1. Limited reversibility of a G2 checkpoint arrest (A) Schematic representation of the Fucci system for visualization of cell cycle progression (37). (B) Asynchronously growing RPE-1 cells (left), or RPE-1 cells irradiated with 4 Gy were fixed 24 h after IR and stained for DNA content using propidium iodide (PI). DNA content was analyzed by FACS. (C) Representative images of cells in Figure 1B. (D) RPE-Fucci cells were irradiated using 2 Gy and subsequently imaged for 7 h. Checkpoint inhibitors were added at different time-points following irradiation to silence the G1 checkpoint, and progression from G1 into S phase was monitored after IR. Cells entering S phase within the first 4 h after irradiation were excluded from analysis, as they have 52 Chapter 3

53 Figure S1 (continued). Limited reversibility of a G2 checkpoint arrest already passed the G1 restriction point at the time of IR. A representative experiment is shown. (E) Same as in (D), except progression from G2 into mitosis was analyzed. Mitotic cells at the moment of irradiation were excluded. A representative experiment is shown. (F) BJ-Tert cells were irradiated with 4 Gy and treated as in (D). A representative experiment is shown. (G) BJ-Tert cells were irradiated with 4 Gy and treated as in (E). A representative experiment is shown. (H) Asynchronously growing RPE cells were irradiated with or 1 Gy, fixed at the indicated days after IR and stained for DNA content using PI. DNA content was analyzed by FACS. Mitotic Index (% of G2 cells) A BJ-Tert (wt) Undamaged 2 h 4 h 8 h 16 h Caffeine No Caffeine Etoposide (1 μm) RPE CCNB1 YFP Figure S2. BJ-Tert cells show a decrease in checkpoint recovery over time Mitotic entry of wild type BJ-Tert cells damaged with etoposide during 1 h in the presence or absence of caffeine, relative to unperturbed mitotic entry, as measured by FACS. Caffeine was added at the indicated times after etoposide treatment. Figure S3. Characterization of RPE CCNB1 YFP cells and BJ-Tert cells in response to DNA damage (A) Mitotic entry of 5 Gy irradiated RPE CCNB1 YFP cells in the presence or absence of caffeine, relative to unperturbed mitotic entry, as measured by FACS. Caffeine was added at the indicated times after irradiation. Bars indicate means ±SD, N=3. (B) Time-lapse images of asynchronously growing RPE CCNB1 YFP cells. Cyclin B1 levels increase in the cytoplasm during G2. At the onset of mitosis, Cyclin B1 translocates to the nucleus, and the cells rounds up within 15 minutes thereafter. (C) Cyclin B1 YFP levels were measured for cells progressing through mitosis and in time aligned to metaphase. (Average of 18 cells, ±SD). (D) BJ-Tert cells were damaged with etoposide for 1 h, and fixed at the indicated time points after etoposide treatment. EdU was added directly following etoposide removal to exclude cells damaged in S phase. Cyclin B localization of G2 cells was visualized using immunofluorescence. At least 5 cells per condition per experiment were quantified. B Time Relative to Metaphase (hh:mm): RPE CCNB1 YFP - 4:3-3:3-2:3-1:3 - :3 - :15 : :15 3 Relative Mitotic Entry (% of G2 cells) Undamaged 1 h 2 h 3 h 4 h 5 h No Caffeine Caffeine 5 Gy γ-irradiation C CyclinB1-YFP Fluoresence intensity (a.u.) YFP DIC RPE CCNB1 YFP Time to Metaphase (h) D Cyclin B1 localiation (% of G2 cells) BJ-Tert (wt) Undamaged 2 h 5 Gy 4 h 6 h Nuclear Cytoplasmic Transient activation of p53 in G2 phase is sufficient to induce senescence 53

54 3 A Undamaged B RPE CCNB1 YFP 5 Gy γ-irradiation sirna: Cyclin B1-YFP Cyclin B1 Plk1 Plk1 Aurora A Aurora A CDK4 CDK4 2 h 4 h 6 h 8 h 1 h C DAPI EdU p21 Merge 5 Gy 4 hr Undamaged E RPE (wt) Input (1%) IP Cyclin B1 Cyclin B1 p21 (HE) p21 (SE) Luc D Nuclear Cyclin B1-YFP (% of G2 cells) G2 1 RPE (wt) Cdh Gy 16 hrs Luc DMSO / LMB Cdh1 RPE CCNB1 YFP p21 KD DMSO Leptomycin B Time after 5 Gy γ-irradiation (hours) G2 2 h 4 h Etoposide G2 2 h 4 h Etoposide Figure S4. Activation of APC/C Cdh1 and nuclear translocation of Cyclin B1 (A) RPE CCNB1 YFP cells synchronized in G2 by FACS were irradiated using 5 Gy. Samples were taken every 2 h following IR. Western blot analysis revealed the degradation of APC/C Cdh1 substrates in response to DNA damage. (B) Western blot analysis of APC/C Cdh1 substrates of RPE-1 cells depleted of luciferase or Cdh1 that were synchronized at the G1/S border by the addition of thymidine for 16 h, and subsequently released for 8 hrh In G2, cells were irradiated using 1 Gy of irradiation and lysed 16 h later. (C) Wild type RPE-1 cells were irradiated with 5 Gy and fixed 4 h later. Directly following IR EdU was added to the medium to label S phase cells. Cells were labeled for DNA content, EdU and p21. (D) RPE CCNB1 YFP depleted of p21 were irradiated with 5 Gy and imaged during 15 h. During the first 2 h, leptomycin B was added to inhibit nuclear export. Nuclear translocation of Cyclin B1 was analyzed. (E) RPE cells synchronized in G2 by thymidine block and release were collected for immunoprecipitation of Cyclin B1. Alternatively, G2 cells were DNA damaged using etoposide and collected 2 h or 4 h after etoposide treatment. 54 Chapter 3

55 + Caffeine si Luciferase :15 2: 4: 4:15 6: 1:45 :15 2: 4: 4:15 6: 8:15 11:15 12:15 12:3 12:45 * * si Cdh1 :15 2: 4: 4:15 6: 8:15 1:45 11:15 11:3 :15 2: 4: 4:15 6: 8:15 * 11:15 * 14:45 18:3 5 Gy γ-irradiation Figure S5. Checkpoint silencing fails to induce mitosis in cells with nuclear Cyclin B1 Representative images of RPE CCNB1 YFP cells transfected with the indicated sirnas and irradiated using 5 Gy 24 h later. Thereafter the cells were followed by time-lapse microscopy for 24 h. Caffeine was added to the medium 4 h after irradiation. Asterisk mark cells going through mitosis. Arrows point towards the newly formed daughter cells. 3 :15 1:15 1:3 1:45 2: 2:15 Nutlin 1uM Nutlin 5uM DMSO :15 :15 :15 :15 :15 1:15 1:15 1:15 1:15 1:15 3: 3:15 3:3 3:45 1:3 1:45 2: 3:3 5:15 6: 7:45 1:3 1:45 2: 2:15 2:3 * * * * * * * * 2:15 5: 7:3 1: Cyclin B1 High Cyclin B1 Low Cyclin B1 High Cyclin B1 Low Cyclin B1 High Cyclin B1 Low Figure S6. Nutlin causes nuclear translocation and subsequent degradation of Cyclin B1 Representative images of RPE CCNB1 YFP cells expressing high or low Cyclin B1 levels at the moment of drug addition. Thereafter the cells were followed by time-lapse microscopy for 1 h. Asterisk mark cells going through mitosis. Arrows point towards the newly formed daughter cells. Transient activation of p53 in G2 phase is sufficient to induce senescence 55

56

57 Chapter 3, addendum Ionizing radiation-induced cell cycle withdrawal from G2 requires ATM and is not solely determined by the amount of DNA damage 3 & Lenno Krenning, and René H. Medema Division of Cell Biology I, The Netherlands Cancer Institute Plesmanlaan 121, 166 CX Amsterdam, The Netherlands Addendum 57

58 Summary The cellular response to DNA double strand breaks (DSBs) is heterogeneous, and ranges from transient cell cycle arrest to permanent withdrawal from the cell cycle. We have previously shown that increasing amounts of DNA damage will cause increasing amounts of cells to permanently withdraw from the cell cycle in G2. Thus, at intermediate levels of DNA damage a fraction of cells continues to proliferate and the other fraction ceases proliferation. In the present study we set out to identify the signaling events responsible for cell cycle withdrawal from G2, and identified ATM as the main driver of cell cycle exit in G2. Importantly, using a live marker for DNA damage, we show that the decision to exit the cell cycle is not exclusively dose-dependent. We find that G2 cells that remain in cycle initially have similar amounts of DNA damage compared to G2 cells that cease to proliferate, and discuss potential causes that may account for this heterogeneous response of cells to similar amounts of DNA breaks. 3 & 58 Chapter 3

59 Introduction Permanent withdrawal from the proliferative cycle, a cellular state often referred to as senescence, is thought to be a tumorsuppressive barrier (35, 328, 329). Recent work has shown that DNA damage may cause G2 cells to exit the cell cycle and become senescent (Chapter 3) (237, 238, 242, 34). It is generally thought that DNA damage-induced senescence occurs in response to excessive DNA damage (36, 33). Indeed, cells are more likely to exit from the cell cycle in response to higher levels of DNA damage, with all cells undergoing cell cycle exit in response to excessive DNA damage (Chapter 3) (242). However, at intermediate levels of DNA damage, the cellular response is heterogeneous such that some arrested cells eventually reenter into the cell cycle, whereas others fail to do so and undergo (Chapter 3) (242). It is currently unclear what underlies this differential cellular behavior; therefore, we set out to identify the cause of cellular heterogeneity in response to DNA damage. Using time-lapse microscopy of damaged cells, we show that cell cycle exit in G2 occurs in response to ATM activation. Strikingly, cell cycle exit in irradiated cells did not correlate to the initial number of DNA damage-induced foci. However, G2 checkpoint recovery was limited to cells that rapidly repaired the DNA damage, whereas cells that undergo cell cycle exit failed to do so. These observations show that cell cycle exit in G2 is not only dictated by the amount of DNA damage, but also by other factors. 3 & Results Cell cycle exit from G2 occurs in response to DNA double strand breaks In order to investigate cell fate decisions following irradiation (IR) we made use of a previously described untransformed cell line in which eyfp was introduced into the Cyclin B1 gene (RPE CCNB1 YFP ) (175). Using this cell line, we have previously shown that IR in G2 results in p53- and p21-dependent nuclear retention of Cyclin B1 as the first irreversible step towards cellular senescence (Chapter 3). The initial activation of p53 following IR mainly requires signaling through ATM, whereas the maintenance of p53 activation may require ATR (78, 331). Hence, we first addressed whether cell cycle exit from G2 was observed in response to DNA damaging agents that result in ATM- or ATR activation. To this end, we treated RPE CCNB1 YFP cells with DNA damaging agents that lead primarily to ATM activation (etoposide or camptothecin) or with ultraviolet light, which causes activation of ATR (56), and imaged them during the subsequent 15 h. When analyzing the fate of G2 cells, identified as Cyclin B1-positivity cells at the start of the movie, we observed that the majority of non-irradiated cells entered mitosis within this time frame (Figure 1A). In stark contrast, none of the etoposide-treated cells and only ~15% of the camptothecin-treated G2 cells entered mitosis (Figure 1A). Instead, in the vast majority of etoposideor camptothecin-treated cells Cyclin B1 translocated to the nucleus and was subsequently degraded (~8% and ~7%, respectively). In addition, etoposide treatment caused ~2% of the cells to degrade Cyclin B1 directly (without preceding nuclear translocation), whereas ~5% of the cells degraded Cyclin B1 directly in response to camptothecin, and ~15% eventually entered into mitosis (Figure 1A). In contrast to the decreased mitotic entry after etoposide or camptothecin treatment, the majority of UV-irradiated cells entered mitosis, with only a subset of ~2% that remained arrested in G2, and in another ~5% Cyclin B1 translocated to the nucleus and was subsequently degraded (Figure 1A). Since treatment with etoposide or camptothecin results in activation of ATM, and UV irradiation causes ATR activation, these results indicate that the DNA damage-induced cell cycle exit in G2, as evidenced by the nuclear translocation and degradation of Cyclin B1, occurs mainly in response to ATM activation. ATM activity drives cell cycle exit in response to physiologically relevant levels of γ-irradiation To address this more directly, we irradiated RPE CCNB1 YFP cells using 2 Gy, a dose we have previously shown to cause nuclear translocation and degradation of Cyclin B1 in a subset of cells (Chapter 3). Inhibition of ATM in non-irradiated G2 cells had no effect on mitotic entry (Figure 1B). Following 2 Gy Addendum 59

60 IR ~3% of the G2 cells entered mitosis, an additional ~3% degraded Cyclin B1 directly, whereas ~4% of the G2 cells translocated Cyclin B1 to the nucleus before degrading it (Figure 1B). Inhibition of ATM in cells irradiated with 2 Gy of IR completely prevented nuclear retention and subsequent Cyclin B1 degradation, and increased mitotic entry to ~7%. However, inhibition of ATM alone did not decrease the fraction of cells that degrade Cyclin B1 directly in response to 2 Gy (Figure 1B). Collectively, these results show that the cell cycle exit in G2, through the nuclear translocation and degradation of Cyclin B1, requires ATM activation. ATM and ATR collaborate to limit the proliferation of cells in response to high doses of γ-irradiation Since ATM inhibition alone did not completely restore mitotic entry after 2 Gy of IR, we wanted to investigate a potential role for ATR in arresting these cells, as ionizing radiation is known to not only 3 & A B C Phenotype (% of Cyclin B1-YFP positive cells) DMSO Etop Cpt UV Mitotic entry Translocate Degrade Direct degradation Stay G2 Phenotype (% of Cyclin B1-YFP positive cells) ATMi: Gy 2 Gy Mitotic entry Translocate Degrade Figure 1. Cell cycle exit in response to low levels of DNA double strand breaks is ATM dependent (A) DNA double strand break inducing agents cause p53-dependent nuclear translocation and degradation of Cyclin B1. Asynchronously growing RPE CCNB1 YFP cells were imaged during 15 h. during the first hour of imaging cells were treated with the indicated inhibitors. UV-irradiation was performed immediately prior to the start of time-lapse imaging. The indicated phenotypes were scored for cells that were in G2 at the start of the movie. (B) ATM inhibition prevents the nuclear translocation and subsequent degradation observed after 2 Gy IR. Asynchronously growing RPE CCNB1 YFP cells were irradiated or not, subsequently imaged during 15 h and the indicated phenotypes were scored for cells that were in G2 at the start of the movie. Where indicated, ATM-inhibitor was added just prior to irradiation. (C) ATM is required for p53-dependent nuclear translocation of Cyclin B1. Same as in (B), except cells were irradiated using 5 Gy IR, and ATM or ATR inhibitors (or both) were added just prior to irradiation. Bars represent means ± SD of 3 independent experiments. At least 5 cells were counted per condition per experiment. Direct degradation Stay G2 Phenotype (% of Cyclin B1-YFP positive cells) ATRi: ATMi: Gy 5 Gy Mitotic entry Translocate Degrade Direct degradation Stay G2 No. of 53BP1-mCherry foci 2 Translocate Degrade (n=14) Mitotic entry (n=9) Time after 2 Gy IR (h) Figure 2. Cell cycle exit is not necessarily dictated by the amount of DNA damage Similar maximal amounts of DNA damage foci are found in cells that eventually continue proliferation and those who cease to proliferate. RPE CCNB1 YFP /53BP1 mcherry -expressing cells were irradiated with 2 Gy IR and subsequently imaged for 15 h. 53BP1 mcherry -foci were traced in individual cells and grouped according to the indicated phenotype. Pooled data of n cells from 2 independent experiments. 6 Chapter 3

61 induce DSBs and can also result in ATR signaling (56). To this end, we inhibited ATM, ATR or both, and subsequently subjected cells to 5 Gy of IR. We observed that single or combined inhibition of ATM and ATR had little effect on mitotic entry in non-irradiated cells (Figure 1C). Following 5 Gy of IR, nuclear translocation and subsequent degradation of Cyclin B1 was observed in ~7% of G2 cells (Figure 1C). An additional ~15% directly degraded Cyclin B1 and another ~5% entered mitosis (Figure 1C). Inhibition of ATR did not influence the fraction of cells that directly degrade Cyclin B1 in response to 5 Gy IR (Figure 1C). Instead, a slight decrease in the fraction of cells that translocated and degraded Cyclin B1, and a concomitant increase of mitotic entry, was observed in ATR-inhibited, irradiated G2 cells. In comparison, ATM inhibition almost completely prevented nuclear translocation and subsequent degradation of Cyclin B1 (Figure 1C). However, in contrast to what was observed after 2 Gy of IR, ATM inhibition alone failed to significantly increase mitotic entry of G2 cells irradiated with 5 Gy of IR (Figure 1B and 1C). Instead, these cells all degrade Cyclin B1 without preceding nuclear translocation (Figure 1C). Combined inhibition of ATM and ATR caused the majority of G2 cells irradiated with 5 Gy of IR to enter mitosis (Figure 1C). These results show that in response to high doses of IR both ATM- and ATR-dependent mechanisms may prevent mitotic entry, but ATM-dependent nuclear translocation and degradation of Cyclin B1 seems to operate already at lower levels of DNA damage. DNA damage-induced cell cycle exit in G2 is not solely dependent on the amount of DNA breaks The results described above do not explain why, in response to intermediate levels of DNA damage, the response to DNA damage is heterogeneous. The amount of DNA damage-induced γh2ax and 53BP1-foci, widely used markers for DNA double strand breaks (DSBs), can vary within a population of cells irradiated with a single dose of homogeneously distributed IR (332, 333). Therefore, we reasoned that, in cells that undergo a cell cycle exit, IR might simply have induced more breaks compared to a neighboring cell that eventually resumes cell cycle progression. To address this, we generated RPE CCN- B1 YFP -expressing cells that also stably express a 53BP1 mcherry fusion previously shown to localize to DSBs (334, 335). We followed RPE CCNB1 YFP /53BP1 mcherry -expressing cells following 2 Gy of IR and counted the amount of 53BP1-foci in cells that entered mitosis or exited the cell cycle (based on Cyclin B1 nuclear translocation and degradation). To our surprise, we observed that cells that exit the cell cycle did not exhibit increased amounts of 53BP1-foci early during the response, compared to cells that eventually enter mitosis (Figure 1C). Instead, a reduction in foci number, indicative of DNA repair, is observed approximately 2 h earlier in the cells that eventually enter mitosis. In addition, cells that exit the cell cycle fail to successfully complete DNA repair within 15 h, in contrast to cells that enter mitosis (Figure 1C). It should be noted here that the latter cells typically had 1 or 2 remaining 53BP1-foci, which disappeared approximately 3 min before they entered into mitosis (not shown). Based on these data we conclude that the cell cycle exit in G2 requires ATM activation, but that heterogeneity observed between cells is not necessarily caused by an increased amount of DNA breaks. Thus, other factors such as the physiological state of the cell at the moment of IR, or the ease at which the breaks can be repaired, possibly influenced by the nature or location of the DSBs induced by IR may influence cell fate decisions. 3 & Discussion We have previously shown that the induction of DSBs causes p53- and p21-dependent nuclear translocation and degradation of Cyclin B1, eventually resulting in senescence (Chapter 3). In agreement with the finding that p53 stabilization in response to IR relies mainly on ATM (297), we find that DNA damaging agents that cause activation of ATM result in the induction of the p53-dependent cell cycle exit from G2 (Figure 1A). In addition, we show that inhibition of ATM largely prevents the cell cycle exit in G2 cells that occurs in response to low amounts of DSBs, whereas ATM inhibition alone fails to prevent the cell cycle exit in G2 in response to high amounts of DSBs (Figure 1B). These results show that activation of ATM is the main driver of the cell cycle exit in G2. However, inhibition of ATM did not fully rescue mitotic Addendum 61

62 entry following 2 Gy of IR, and failed to stimulate mitotic entry G2 cells irradiated with 5 Gy of IR (Figure 1B and 1C). This may be an effect of partial ATM inhibition, or it may indicate that signaling through other kinases can result in the degradation of Cyclin B1 following IR. In this respect, it is important to note that DNAdamage induced by IR is not limited to DSBs. Besides DSBs, IR causes single-stranded DNA breaks, which could potentially activate ATR directly (56). In agreement with this, combined inhibition of ATM and ATR causes G2 cells irradiated with 5 Gy of IR to enter mitosis (Figure 1C). 3 & Even though the initial accumulation of 53BP1-foci is similar in cells that remain in cycle compared to those that do not, we observe clear differences regarding the dissociation of 53BP1-foci (Figure 2), indicative of differences in DNA repair. DNA repair through non-homologous end joining (NHEJ) typically occurs much faster than homologous recombination (HR) (325), therefore, the observation that 53BP1 focus dissociation occurs much slower in cells that undergo a cell cycle exit may indicate that the repair mechanisms involved may be different in cells that remain in cycle compared to cells that withdraw from the cell cycle. The initiation of DNA resection, required for homology-directed DNA repair, causes an ATM-to-ATR signaling switch (71-73, 75, 336, 337). As DNA resection and ATR activation occur already within the first hours following IR, the additional signaling through ATR might contribute to the efficient induction of p53. However, the finding that ATR inhibition only slightly decreases p53-dependent nuclear translocation and degradation of Cyclin B1 in response to 5 Gy (Figure 1C), and the finding that inhibition of ATM alone allows mitotic entry of most G2 cells irradiated with 2 Gy of IR (Figure 1B), suggest that ATR signaling hardly affects p53-dependent cell fate decisions following low levels of DSBs (Figure 2C). Therefore, other properties of the DNA damage itself, rather than DNA resection and ATR activation, likely influence cell fate decisions. This is not to say, however, that ATR is incapable of eliminating cells from the cell cycle, as ATM-inhibited cells irradiated with 5 Gy of IR fail to proliferate. In addition, forced activation of ATR in the absence of DNA damage can drive senescence (338). Specific properties of the DNA damage-induced break may be dependent on the location of the break. It is becoming increasingly clear that heterochromatin can influence the signaling response elicited by a DSB (339). For instance, heterochromatin may pose a barrier to γh2ax-spreading (34). Recruitment of Mdc1 requires γh2ax (66), and Mdc1 itself is required for the maintenance of an IR-induced G2 arrest (127), potentially by maintaining ATM-dependent activation of Chk2. This would suggest that DNA damage in heterochromatin might result in less efficient ATM activation compared to breaks in euchromatin. Therefore, a cell may be more likely to exit the cell cycle in the presence of DSBs in euchromatin, an hypothesis which is supported by the finding that forced chromatin relaxation results in increased γh2ax expansion, and hypersensitivity of the G2 checkpoint (341). However, DNA damage at heterochromatin is repaired with slower kinetics compared to DNA damage in euchromatin (325). Our results indicate that the damage that occurs in cells that will exit the cell cycle is repaired with slower kinetics than cells that remain in cycle (Figure 2), arguing against a model in which a DSB located in heterochromatin would be less likely to induce exit from the cell cycle compared to a DSB in euchromatin. However, DSBs at some genomic locations, such as telomeres, are refractory to repair and instead result in the continuous activation of the DNA damage response (257, 258). In addition, IR-induced foci that persist over long periods of time following IR continue to activate the DDR (323, 342), and this may contribute to cell fate decisions (259). Therefore, the induction of DSBs at sites inherently refractory to DNA repair might cause cells to exit the cell cycle. In support of this notion are the observations that more cells exit the cell cycle after higher doses of IR (Chapter 3), paralleled by an increase in the amount of irreparable DNA damage foci (259). Finally, the observation that not all persistent DNA damage foci induced by IR associate with telomeres suggests that other genomic loci might be inherently refractory to DNA repair (257, 258). Besides properties of the DNA damage induced by IR, the physiological condition of the damaged cell might contribute to the outcome of a DNA damaging insult. We and others have previously shown that the cell cycle exit in G2 cells is preceded by p53- and p21-dependent nuclear translocation and subse- 62 Chapter 3

63 quent degradation of Cyclin B1 (Chapter 3) (24, 242, 34). In non-stressed cells, both p53 as well as p21 are expressed, and this basal expression differs from cell to cell (21, 254, 343). It is not hard to imagine that such differences in the basal conditions of a cell might impact the efficiency of IR-induced p53 activation and p21-induction, and therefore may contribute to cell fate decisions. In conclusion, many different DNA-damage-related or non-related cellular properties might contribute to cell fate decisions. Thus, to further our understanding of processes underlying heterogeneity in the DNA damage response, we should combine live-cell markers for different DNA repair pathways or DDR signaling strength with markers for cell cycle progression, many of which have been developed over the past years (175, 189, 21, ). 3 & Addendum 63

64 Materials and methods Tissue culture and irradiation RPE CCNB1 YFP cells and derivatives were maintained in Dulbecco s modified Eagle Medium (DMEM)/F12, supplemented with 1% FCS and antibiotics. For live cell imaging, the appropriate amount of cells were seeded in a Lab- Tek II chambered cover glass (Thermo Scientific) on the day before imaging. Directly before imaging, the medium was replaced by Leibovitz s L-15 (GIBCO) CO 2 -independent medium (in the presence or absence of inhibitors). Then, cells were γ-irradiated using a Gammacell Exactor (Best Theratronics) equipped with a 137 Cs source. Alternatively, cells were UV-irradiated (1 J/m 2 ). For UV-irradiation, cells were washed twice with PBS, which was subsequently removed from the cells. Then, cells were irradiated with a UV-lamp (delivering ~3 J/m 2 /second) for 3.5 seconds and L-15 medium was added to the cells before imaging. 3 & Live cell imaging and image analysis Images were obtained using a Deltavision Elite (Applied Precision) maintained at 37 C, equipped with a 1x or 2x air objective (Olympus, NA.4 or.75, respectively) and cooled coolsnap CCD camera. Image analysis was done using Image J software. In order to count 53BP1-foci, single cells were segmented from time-lapse movies after which automated image analysis was performed to count foci, as described in chapter 4. Constructs, stable cell line construction and chemicals The retroviral construct for expression of 53BP1-mCherry was generated by exchanging GFP for mcherry in a previously described construct (132). Virus was produced as described before (226). Following infection of RPE CCNB1 YFP cells, cells expressing similar levels of 53BP1-mCherry were isolated using a FACSAria (BD) as was described before (226). Chemicals used in this study are Etoposide (Sigma, used at 1 μm), Camptothecin (Sigma, used at 1 μm), the ATM inhibitor KU (Calbiochem, used at 1 μm) and the ATR inhibitor ATR-45 (Ohio State University, used at 5 μm). Acknowledgements We would like to thank members of the Medema lab for encouragement and helpful discussions. 64 Chapter 3

65 Addendum 65 3 &

66

67 Chapter 4 DNA damage triggers a cell cycle exit in prophase Femke M. Feringa 1, Lenno Krenning 1, André Koch, Bram van den Broek, Kees Jalink and René H. Medema 2 Division of Cell Biology I and Cancer Genomics Center, The Netherlands Cancer Institute Plesmanlaan 121, 166 CX Amsterdam, The Netherlands 1 )equal contribution 4 2 ) Correspondence: r.medema@nki.nl Under review DNA lesions trigger a cell cycle exit in prophase 67

68 Abstract Activation of the DNA damage response can lead to the induction of an arrest at various stages in the cell cycle. These arrests are reversible in nature, unless the damage is too excessive. Here, we show that checkpoint reversibility in G2 phase critically depends on Emi1, which is lost in prophase cells. DNA damage in prophase leads to rapid activation of APC/C Cdh1 and the induction of senescence, even at low levels of damage. Importantly, while suppression of Emi1 renders the checkpoint irreversible throughout G2, restoration of checkpoint reversibility in prophase promotes cell division with residual DNA damage. Thus, the loss of checkpoint reversibility in prophase provides a final option for a cell cycle exit in cells containing residual DNA lesions Chapter 4

69 Introduction To protect their genome, cells depend on the action of DNA damage checkpoints that ensure detection and repair of DNA damage, coupled to the induction of a cell cycle arrest (347, 348). These checkpoints can induce a reversible arrest at different stages of the cell cycle that is required for repair to take place before the cell divides (1, 349). Functionality of these checkpoints requires accurate coordination between repair, checkpoint signalling and cell cycle progression, particularly in G2 phase, since mitotic entry with broken chromosomes poses a direct threat to proper chromosome segregation and genome stability (85, 2). Emi1 is a well-established inhibitor of the anaphase promoting complex/cyclosome (APC/C) (52, ) and has previously been implicated in the response to severe DNA damage in G2 (23, 226, ). Emi1 acts to inhibit the APC/C in G2 and is degraded in prophase (41, 51, 351). Here we show that the scheduled loss of Emi1 in prophase results in a previously undescribed response to DNA damage. Results and discussion To investigate the fate of cells that encountered DNA damage at distinct stages in G2 phase, we performed time-lapse microscopy of untransformed RPE-1 cells with endogenously tagged Cyclin B1 YFP (175). Cyclin B1 expression rises as cells progress through G2 into M, and the absolute level of fluorescence in these cells can be used to derive temporal information regarding the cell cycle position of the individual cell (353). Using increasing doses of ionizing radiation (IR) we find that the subset of cells that can recover from the damage and enter mitosis decreases with increasing dose (Figures 1A and 1B). As the dose increases, the recovering fraction is replaced by cells in which Cyclin B1 translocates to the nucleus (Figures 1A and 1B), a process we and others have previously shown to lead to the induction of senescence (226, 24, 242). Interestingly, we find that a subset of Cyclin B1 YFP -positive cells displays a distinct behaviour. This subset directly loses Cyclin B1 expression after DNA damage (Figures 1A and 1B), without the prior translocation of Cyclin B1 to the nucleus. The fraction of cells that directly loses Cyclin B1 is already maximal after low doses of irradiation (Figure 1B, red bars), in sharp contrast to the dose-dependent nuclear Cyclin B1 retention (Figure 1B, blue bars). Remarkably, the cells that directly loose Cyclin B1 have significantly higher levels of Cyclin B1 YFP at the moment of irradiation (Figures 1C, 1D and 1Ee). In contrast, cells that recover from the damaging event, as well as the cells that translocate Cyclin B1 to the nucleus, express lower levels of Cyclin B1 YFP at the moment of irradiation, suggesting that these cells are in the earlier stages of G2 phase (Figures 1C, 1D and 1E). To further define the cells that directly loose Cyclin B1 we analysed if in this population centrosomes had separated at the moment of irradiation. Strikingly, the great majority of cells within this population had already started to separate their centrosomes at the time of irradiation, coincident with phosphorylation of histone H3 (Figures 1F, 1G and S1A). Centrosome separation and phosphorylation of H3 are both initiated in prophase, and therefore these results imply that direct loss of Cyclin B1 upon irradiation is restricted to prophase cells. Subsequently, we tested the consequence of this unique response for the fate of a cell exposed to low dose irradiation. We used time-lapse imaging to track Cyclin B1 YFP -low and Cyclin B1 YFP -high cells from the moment of irradiation onwards. We found a clear difference in the fraction of Cyclin B1 YFP -low cells that managed to recover when compared to the Cyclin B1 YFP -high cells (Figure 1H), indicating that the capacity to recover is compromised in late G2/prophase. We confirmed that the unique response of prophase cells is not due to a difference in overall damage or repair signalling, as DNA damage foci are formed and resolved at similar kinetics in G2 cells that translocate Cyclin B1 to the nucleus, compared to prophase cells that loose Cyclin B1 directly upon irradiation (Figure S1B). Time-lapse imaging of human dermal microvascular endothelium- (HMEC-1), mammary gland epithelial- (MCF-1a) and human osteosarcoma (U2OS) cells with endogenously tagged Cyclin B1 YFP revealed that the response observed in prophase is conserved in various cell types (Figure S1C). We therefore conclude that cell fate after DNA 4 DNA lesions trigger a cell cycle exit in prophase 69

70 A Arrest and Recover :15 1: 2: 2:15 2:3 3:15 4:15 5:15 5:3 5:45 2 Gy IR Translocate Degrade :15 1: 2: 2:15 2:3 3:15 4:15 5:15 6:15 7:15 Degrade :15 1: 2: 2:15 2:3 4 B Phenotype (% of G2 cells) F Centrosome distance (μm) Gy 1 Gy 2 Gy 5 Gy 1 Gy Mitotic entry Translocate Degrade Direct degradation Stay G2 Unperturbed (n=2) C Cyclin B1-YFP intensity (a.u.) Time to Cyclin B1 nuclear import (h) G ** 4 T. D. (n=17) Degrade (n=59) Unperturbed (n=15) Time to metaphase (h) Centrosome distance (μm) H 6 Mitotic entry (% of G2 cells) Figure 1. Prophase cells show a unique response to DNA damage (A) Time-lapse images represent distinct responses of RPE CCNB1 YFP cells to ionizing radiation (IR). (B) Quantification of the frequency with which the responses in (a) are observed in Cyclin B1 YFP -positive (G2) cells within 16 h after IR. Mean ± s.d. of three independent experiments. (C) Cyclin B1 YFP intensity during unperturbed G2/M progression in individual cells, and in silico aligned at metaphase. Mean ± s.d., n=15 RPE CCNB1 YFP cells from one experiment. (D) Cyclin B1 YFP intensity measured 15 min after 2 Gy IR in cells undergoing the indicated responses. Dots represent individual cells (n), mean ± s.d., results are representative of three independent experiments. **P<.1 (unpaired t-test) (E) Cyclin B1 YFP levels measured in individual cells that either recovered from 2 Gy IR and entered mitosis or that lost Cyclin B1 completely. Mean ± s.d. n=13 RPE CCNB1 YFP cells for each response. (F) Centrosome separation measured in unperturbed RPE CCNB1 YFP cells, and in silico aligned to nuclear import of Cyclin B1. Mean ± s.d. n=2 RPE CCNB1 YFP cells from one experiment. (G) Centrosome distance measured 15 min after 5 Gy IR in cells undergoing the indicated responses. Dots represent individual cells (n) pooled from three independent experiments, mean ± s.d. is shown. **P<.1 (unpaired t-test) (H) Spontaneous recovery after 1 Gy or 2 Gy IR in Cyclin B1 low - (G2) vs Cyclin B1 high - (prophase) cells. Prophase and G2 cells were selected based on top 25% Cyclin B1 YFP expressing cells vs bottom 75% Cyclin B1 YFP expressing cells, measured 15 min after IR. Mean ± s.e.m. of at least three independent experiments. 4 2 D 2 Gy IR n.s. ** ** Mitosis (n=24) T. D. (n=34) Degrade (n=24) 1 Gy 2 Gy CycB1 YFP -Low CycB1 YFP -High Cyclin B1-YFP intensity (a.u.) E Cyclin B1-YFP intenstity (a.u.) Gy IR Cyc B1 degradation Recovery Time after 2 Gy IR 7 Chapter 4

71 A Cyclin B1 degradation (% of CycB1 YFP -High cells) 5 Gy IR - Prophase sirna: Luc Cdh1 Cdc2 B Relative Cdk2 activity Gy IR - Prophase Cdk2 activity. Cyclin B1 YFP level Time after irradiation (h) Relative Cyclin B1 YFP level C Cyclin B1 degradation (% of cells) Cdk inhibition Cdk1/2i: CycB1 YFP -Low cells CycB1 YFP -High cells D Cdk inhibition 1 Mitotic entry (% of cells) Cdk1/2i: CycB1 YFP -Low cells CycB1 YFP -High cells E IR - Prophase 1 Cyclin B1 degradation (% of CycB1 YFP -High cells) Gy IR: DMSO MK F IR - Prophase 1 Mitotic entry (% of CycB1 YFP -High cells) Gy IR: DMSO MK1775 Figure 2. DNA damage-induced Cdk inhibition results in rapid APC/C Cdh1 activation in prophase (A) Direct degradation of Cyclin B1 in prophase cells (selected as in Figure 1H) depleted for luciferase, Cdh1 or Cdc2 that were irradiated with 5 Gy. Mean ± s.e.m. of three independent experiments. (B) Relative Cdk2 activity and Cyclin B1 YFP intensity were measured in individual prophase cells that degraded Cyclin B1 after 1 Gy IR. All time-points were normalized to Cdk2 activity and Cyclin B1 level at the first frame, which were set to one. Mean ± s.e.m. of three independent experiments. (C and D) Cyclin B1 degradation (C) and mitotic entry (D) scored in undamaged G2 and prophase cells (selected as in Figure 1H) within 1 h after washout of Cdk1 (RO-336) and Cdk2 (Rosco- vitine) inhibitors that had been present for 5 h. Mean ± s.e.m. of 3 independent experiments. (E and F) Cyclin B1 degradation (E) and mitotic entry (F) scored in prophase cells (selected as in Figure 1H) within 1 h after IR 1 Gy. Wee1 inhibitor (MK 1775) or DMSO were added immediately after IR. Mean ± s.d. of three independent experiments. 4 damage is regulated in a unique way in prophase cells, which is intrinsically different from the known G2 response. Importantly, this response causes prophase cells to be highly sensitive to DNA damage. Excessive DNA damage results in activation of APC/C Cdh1 in G2 phase, to promote degradation of multiple G2/M targets, including Cyclin B1 (23, 226, , 242). This activation of APC/C Cdh1 normally occurs several hours after the damage, much later than the onset of Cyclin B1 degradation that we observe in prophase cells. Nevertheless, we set out to test if the direct loss of Cyclin B1 observed after DNA damage in prophase was also caused by APC/C Cdh1 -dependent degradation. Indeed, we could effectively prevent Cyclin B1 degradation when cells were depleted of Cdh1 (Figure 2A). This effect was not seen after depletion of Cdc2, the other co-activator of the APC/C (Figures 2A and S2A). In addition, we find that loss of Cyclin B1 is prevented when irradiated prophase cells are treated with the proteasome inhibitor MG-132 (Figure S2B). While direct degradation of Cyclin B1 in prophase cells is completely prevented after depletion of Cdh1 (Figure 2A), we find that after high dose irradiation Cyclin B1 translocates to the nucleus, similar to what we observe in the general G2 population (Figures S2C and S2D). Immunofluorescent staining of the APC/C targets Aurora A and Cyclin A2 shows that these are also lost in cells that degraded Cyclin B1 (Figures S2E and S2F). Collectively, these results show that the loss of Cyclin B1 following DNA damage in prophase results from activation of the APC/C Cdh1. Next, we aimed to find out how APC/C Cdh1 can be activated specifically in prophase cells in response to low dose irradiation. Activation of APC/C Cdh1 in undamaged cells normally occurs in anaphase, following the loss of Cdk activity (354). Therefore, we investigated whether Cdk inhibition induced by DNA damage precedes the onset of APC/C Cdh1 activation in prophase cells. Using a previously described live-cell sensor for Cdk2 activity (21), we measured Cdk2 activity and Cyclin B1 levels in single cells after irradiation (Figure S2G). A clear drop in Cdk2 activity precedes Cyclin B1 degradation in irradiated prophase cells (Figure 2B). Next, we tested whether inhibition of Cdk1 and/or Cdk2 activity by itself would be sufficient to cause APC/C Cdh1 activation in prophase. Live-cell imaging of prophase cells treated with Cdk1 and/or Cdk2 inhibitors revealed that only dual inhibition effectively induced degradation of Cyclin B1, implying that Cdk1 or Cdk2 activity alone is sufficient to keep APC/C Cdh1 inactive in prophase (Figure DNA lesions trigger a cell cycle exit in prophase 71

72 A Time post s-phase sort (h): Cyclin B1 EMI1 Cdk4 Time post s-phase sort (h): Cyclin A2 EMI1 HSP9 B Cyclin B1 degradation (% of CycB1 YFP -Low cells) sirna: IR - G2 Luc Emi1 Luc Emi1 Gy 1 Gy C Mitotic entry (% of CycB1 YFP -Low cells) sirna: Luc IR - G2 Emi1 Luc Emi1 Gy 1 Gy 4 D Cyclin B1 degradation (% of CycB1 YFP -Low cells) Rosco: RO-336: Cdk inh. - G Emi1 sirna E Cyclin B1 degradation (% of CycB1 YFP -High cells) Doxycyclin: (Emi1 o.e.) 5 IR - Prophase Figure 3. Emi1 acts to maintain recovery competence in G2 cells (A) Western blot showing EMI1, Cyclin A2 and Cyclin B1 protein levels at the indicated time following early S phase sort. Representative blots of two independent experiments are shown. (B and C) Cyclin B1 degradation (B) and mitotic entry (C) within 1 h after 1 Gy IR was analysed in RPE CCNB1 YFP cells partially depleted for EMI1. Mean ± s.d. of three independent experiments. (D) Cdk1 (RO-336), Cdk2 (Roscovitine) or both were inhibited in undamaged RPE CCNB1 YFP cells partially depleted of EMI1. Direct Cyclin B1 degradation of cells in G2 at the moment of Cdk inhibition was analysed. Mean ± s.d. of three independent experiments. (E) EMI1 Turq overexpression was induced in RPE CCNB1 YFP 4 h before IR using doxycycline. Direct Cyclin B1 degradation was scored in prophase cells (selected as in fig 1h). Mean ± s.d. of three independent experiments. S2H). More importantly, temporary inhibition of Cdk1 and Cdk2 activity was enough to induce Cyclin B1 degradation in undamaged prophase cells, but did not affect G2 cells in the same way. Instead, G2 cells halted progression to mitosis, but as expected, the majority continued cell cycle progression after they were released from Cdk1/2 inhibition (Figure 2D). In contrast, the prophase cells degraded Cyclin B1 and were not able to enter mitosis after wash out of both inhibitors (Figures 2C, 2D and S2I). This shows that mere inhibition of Cdk activity in prophase cells is sufficient to activate APC/C Cdh1. Conversely, inhibition of Wee1, the kinase responsible for inhibitory phosphorylation of Cdk-subunits (4), almost completely prevented the DNA damage-induced degradation of Cyclin B1 in prophase and promoted mitotic entry (Figures 2E, 2F and S2J). Thus, abrupt Cdk inhibition induced by activation of the DNA damage checkpoint in prophase can cause premature APC/C Cdh1 activation, resulting in Cyclin B1 degradation and cell cycle exit. While our data clearly show that loss of Cdk activity in prophase cells causes APC/C Cdh1 activation, treatment with Cdk1/2 inhibitors does not activate APC/C Cdh1 in G2 cells. This implies that G2 cells are protected from a rapid cell cycle exit upon stress-induced Cdk inhibition. A well-known antagonist of APC/C Cdh1 activity in S and G2 phase is early mitotic inhibitor-1 (Emi1) (52, 35) Therefore we wondered if Emi1 is needed to protect G2 cells from APC/C Cdh1 activation by DNA damage. Indeed, Plk1- and Cdk1-dependent phosphorylation of Emi1 results in its destruction in prophase (41, 51, 53, 351, 355). Also, excessive DNA damage in G2 cells is known to cause p21-dependent down-regulation of Emi1 resulting in APC/C Cdh1 activation and degradation of its targets (235, 236). However, this latter response is limited to cells that contain high levels of damage, and follows only after p21-dependent nuclear retention of Cyclin B1-Cdk complexes (226, 242), not matching the fast response we observe in prophase cells. Using drug-free synchronyzed RPE-1 Fucci cells (Figures S3A, S3B and S3C), we set out to determine the timing of Emi1 degradation in RPE-1 cells. Staining for Cyclin A2, Cyclin B1 and Emi1 on western blot revealed that Emi1 is indeed degraded prior to the Cyclins (Figure 3A). Since Cyclin A degradation occurs directly after nuclear envelope break down (NEBD) ( ) this is most consistent 72 Chapter 4

73 A 1 Gy IR Checkpoint recovery and daughter cell proliferation :15 2: 2:3 19: 2: 21: * * Cyclin B1 degradation and cell cycle exit :15 2: 2:3 24: 48: 72: 53BP1 Cyclin B1 53BP1 Cyclin B1 B Progression into new cell cycle (%) Cell fate after 1 Gy: 1 Gy IR Cyclin B1 degradation Mitotic Entry C 1 Gy IR 1: 3: 144: SA-β-Gal 53BP1 Cyclin B1 D Mitotic entry (% of CycB1 YFP -High cells) Dox: IR - Prophase Gy 5 Gy E % of G1 daughter cells Gy IR - Prophase No doxycyclin (n = 16) Doxycyclin (n = 29) (Emi1 o.e.) Nr. of 53BP1-foci per G1 daughter cell after prophase DNA damage F Mitotic cells with broken chromosomes (%) sirna: 1 Gy IR n = 19 n = 81 n = 122 n = 16 Luc Cdh1 Luc Cdh1 Gy 1 Gy 4 G γh2ax Merge Merge γh2ax CREST DAPI DAPI CREST Figure 4. Hypersensitivity to DNA damage in prophase is needed to protect genomic stability (A and B) Representative images from time-lapse movies of RPE CCNB YFP 53BP1 mcherry cells following 1 Gy IR. Stills are representative of three independent experiments. Star indicates tracked daughter cell after mitosis. Progression into a subsequent cell cycle scored in RPE CCNB1 YFP cells that had degraded Cyclin B1 after 1 Gy and cells that had recovered and entered mitosis after 1 Gy IR. Rebuilding of Cyclin B1 expression within 72 h from IR was used to score entry into a next cell cycle. Mean ± s.d. of three independent experiments. (C) As in a, except cells were followed for 144 h and then stained for SA-β-gal to identify senescent cells. Stills are representative of n=16 cells pooled from two experiments. (D) Mitotic entry of prophase cells (selected as in Figure 1H) after.5 or 5 Gy IR. EMI1 Turq overexpression was induced by doxycycline addition 2 h before IR. Mean ± s.d. of 3 independent experiments. (E) Quantification of DNA damage foci present in G1 daughter cells after mitosis in the presence or absence of Emi1 overexpression, induced as in Figure 4D. DNA damage foci were counted based on 53BP1 mcherry signal 3 h after mitosis. n, number of cells analysed, pooled from three independent experiments. (F) Quantification of mitotic cells with broken chromosomes in Luc or Cdh1 depleted cells that entered mitosis within 4 h after IR. Average of 2 independent experiments is shown. n, number of cells analysed in two independent experiments. (G) Representative image of broken (left) or intact (right) chromosomes from a cell quantified in Figure 4C. DNA lesions trigger a cell cycle exit in prophase 73

74 4 with degradation of Emi1 in prophase, similar to what was observed previously (41, 51, 53, 351, 355). Thus, DNA damage-induced activation of APC/C Cdh1 in prophase may be a consequence of limited Cdk activity in cells that have already lost Emi1, and are therefore unable to prevent APC/C Cdh1 activation. To corroborate this notion, we asked if reduction of Emi1 expression could render the checkpoint irreversible throughout G2. Depletion of Emi1 leads to dramatic phenotypes like rereplication, due to the increased activity of the APC/C Cdh1 (4, 46). We therefore established conditions for partial depletion of Emi1. Interestingly, partial Emi1 depletion, which hardly affects cell cycle progression in undamaged cells (Figures 3D and 3E), leads to a very clear hypersensitivity to DNA damage in G2 (Figures 3B, 3C and S3F). The majority of control G2 cells are able to recover from 1 Gy of irradiation, whereas >8% of the Emi1-reduced cells degrade Cyclin B1 at this dose of irradiation, preventing their recovery (Figures 3B, 3C and S3F). Consistent with this, we find that Emi1-reduced RPE-1 cells with low Cyclin B1 levels (matching cells in early G2 phase), show a response that is normally restricted to prophase cells (Figure 3G). Also, reduction of Emi1 in undamaged cells allows direct Cyclin B1 degradation in all G2 cells when Cdk activity is chemically inhibited (Figures 3D and S3H). Conversely, overexpression of Emi1 completely prevents the direct DNA damage-induced degradation of Cyclin B1 in prophase cells (Figures 3E, S3I and S3J). Thus, our data shows that Emi1 acts to sustain checkpoint reversibility in G2, and its degradation in prophase results in an irreversible DNA damage response that ensures a rapid cell cycle exit, even at low levels of DNA damage. Our observation that prophase cells directly degrade Cyclin B1 after low doses of irradiation indicates that prophase cells withdraw from the cell cycle in the presence of low levels of DNA damage. Indeed, none of the prophase cells that degraded Cyclin B1 after exposure to 1 Gy were able to proliferate within 72 h after damage (Figures 4A and 4B). In contrast, 75% of the cells that recovered from this dose also progressed into subsequent cell divisions within 72 h (Figures 4A and 4B). Moreover, time-lapse analysis of individual cells that were followed for 5 days and then stained for senescence-associated β-galactosidase activity (SA-β-gal), confirmed that prophase cells enter a senescent state in response to low dose irradiation (Figure 4C). While high doses of irradiation lead to extensive checkpoint activation, including long-lasting Cdk inhibition and high levels of p53 and p21, low doses of irradiation induce a much milder checkpoint response only temporarily inhibiting Cdk activity (226). We therefore hypothesized that rapid cell cycle exit of cells damaged in prophase would be especially important after low levels of damage, since prophase cells will likely progress into mitosis as soon as Cdk activity is restored. In such cases, the little time available for repair could pose a serious threat to genomic integrity, and this could be compensated by a rapid cell cycle exit to prevent cell division with broken chromosomes. Since we could prevent the DNA damage induced Cyclin B1 degradation in prophase cells by overexpression of Emi1, we asked if this could promote mitotic entry. Indeed, we find that mitotic entry is restored in cells expressing Emi1 in prophase, and this restoration is limited to low doses of irradiation (Figures 4D and S4A). Importantly, the increased mitotic entry of prophase cells is associated with increased reappearance of 53BP1 foci in G1 daughter cells (Figure 4E), indicating that loss of the prophase-specific response to low levels of DNA damage results in carry-over of damaged DNA to the daughter cells. Similarly, the number of mitotic cells with broken chromosomes observed in Cdh1-depleted cells after 1 Gy irradiation was twice as much as seen in control cells (Figures 4F and 4G). This difference is not a consequence of altered damage signalling caused by Cdh1 depletion since the number of breaks was similar in luciferase- and Cdh1- depleted cells that were pushed into mitosis by the addition of caffeine (Figure S4B). In conclusion, this previously unidentified cell cycle exit mechanism in prophase is important to prevent cell division with broken chromosomes. Our data thus far explains why cells in prophase are hypersensitive to low levels of acute DNA damage. Interestingly, we occasionally observed untreated cells that spontaneously degraded Cyclin B1 in prophase (Figure 1B). It is well known that the genome contains regions that are difficult to replicate during S phase, such as common fragile sites (CFS) (36, 361). Replication stress can create DNA lesions, particularly at fragile sites, but progression from S to G2 phase is not prevented (36, 362). We therefore wondered whether the cell cycle exit we observed in prophase cells treated with damaging agents 74 Chapter 4

75 Mitosis Cyclin B1 degradation A :15 1:3 2:45 4: 5:15 6:3 7:45 9: 1:15 11:3 12:45 14: 15:15 16:3 :15 1:3 2:45 4: 4:15 4:3 53BP1 Cyc B1 53BP1 Cyc B1 G B Degradation (% of G2 cells) Aphidicolin: (.4 µm) G2 at t = h G2 at t = 15 h C Mitotic entry (% G2 cells) Aphidicolin: (.4 µm) G2 at t = h G2 at t = 15 h D No. of 53BP1-mCherry foci per cell Aphidicolin (.2 µm) G2 at t = 15 h ** Mitotic entry Cyclin B1 degradation E Mitotic entry (% G2 cells at t = 15 h) ATR inh: - + Aphidicolin (.2 µm) F No. of 53BP1-mCherry foci per G1 daughter cell Aphidicolin (.2 µm) G2 at t = 15 h ATR inh: ** - + Mitotic entry 4 G2 Prophase NEBD Metaphase Anaphase G2 Prophase G2 Senescence Emi1 protein levels APC/C Cdh1 Cdk 1/2 activity APC/C Cdh1 APC/C Cdc2 APC/C Cdh1 Emi1 protein levels APC/C Cdh1 Cdk 1/2 activity APC/C Cdh1 Unperturbed cell cycle Following DNA damage Figure 5. Prophase exit eliminates cells that experienced replication stress (A) Representative stills from RPE CCNB1 YFP 53BP1 mcherry cells treated with.4 μm Aphidicolin. Stills are representative of 3 independent experiments. (B and C) Quantification of degradation (B) and mitotic entry (C) in RPE CCN- B1 YFP positive cells h after.4 μm Aphidicolin addition or in cells that are RPE CCNB1 YFP positive 15 h after.4 μm Aphidicolin addition. Mean ± s.e.m. of 3 independent experiments. (D) Quantification of 53BP1 foci in RPE CCNB1 YFP positive cells that enter mitosis or that degrade Cyclin B1 following 15 h treatment with.2 μm Aphidicolin. Foci were quantified using automated foci analysis 15 min before onset of Cyclin B1 degradation or 9 min before metaphase. Dots represent individual cells. Mean ± s.d., results are representative of three independent experiments. **P<.1 (unpaired t-test) (E) Quantification of mitotic entry in RPE CCNB1 YFP positive cells treated with.2 μm Aphidicolin for 15 h followed by addition of ATR inhibitor. Mean ± s.d. of 2 independent experiments. (F) Quantification of 53BP1 foci 3 h after metaphase in cells treated with.2 μm Aphidicolin for 15 h followed by addition of ATR inhibitor where indicated. Dots represent individual cells. Mean ± s.d., results are representative of three independent experiments. **P<.1 (unpaired t-test) (G) Model for irreversible cell cycle exit in the presence of residual DNA damage in prophase cells. Limited Cdk1 and Cdk2 activity leads to rapid activation of APC/C Cdh1 in prophase, but not G2 cells due to the presence of EMI1. APC/C Cdh1 activation in prophase results in degradation of pro-mitotic Cyclins, reverting prophase cells back to a G2 state followed by the induction of senescence. DNA lesions trigger a cell cycle exit in prophase 75

76 4 could also serve to prevent division of cells that carry such lesions, which are not forced out of the cell cycle by other checkpoint mechanisms. Treatment with low doses of aphidicolin (,2-,4μM) induces replication stress, but does not prevent S to G2 phase progression or activate an ATM- or P53-dependent DNA damage response (363, 364). We therefore used a low dose of aphidicolin to induce mild replication stress and tracked the fate of RPE Cyclin B1 YFP -expressing cells. Remarkably, 5% of the cells that went through S phase in the presence of aphidicolin did not enter mitosis, but instead degraded Cyclin B1 (Figures 5A, 5B and 5C). In contrast, cells that already past S phase at the moment of aphidicolin addition were not affected (Figures 5B and 5C). Cyclin B1 degradation after replication stress was preceded by a prolonged period of Cyclin B1 accumulation and clear centrosome separation, indicating that these cells first entered prophase before Cyclin B1 was degraded (Figure 5A). These data imply that the cell cycle exit we observe in prophase is a much more general mechanism used to eliminate cells with low amounts of DNA damage from the cell cycle, and that this response is not limited to cells carrying double-stranded breaks. To confirm that the aphidicolin treated cells that degrade Cyclin B1 in prophase indeed carry residual sites of damage, we monitored 53BP1 foci after replication stress. Indeed, whereas the majority of cells that progressed in mitosis had no foci left in prophase, cells that degrade Cyclin B1 in prophase in response to aphidicolin contained on average three 53BP1 foci (Figure 5D). The remaining foci do generate a checkpoint signal, since inhibition of ATR caused all cells to enter mitosis and resulted in propagation of DNA lesions to the following G1 phase (Figures 5E and 5F). Thus, low levels of DNA breaks or replication stress that fail to prevent the onset of prophase, can trigger Cyclin B1 degradation and cell cycle exit to prevent division of damaged cells (Figure 5G). Our data show that prophase cells have a very unique, irreversible response to DNA damage, in sharp contrast to the reversible cell cycle arrests that act in other stages of the cell cycle (349). We demonstrate an essential role for Emi1 in the DNA damage checkpoint in G2 in that it acts to maintain checkpoint reversibility, thereby reducing sensitivity of a cell to DNA damage by allowing time for DNA repair and subsequent checkpoint recovery. As such, Emi1 could be particularly important for post-replication repair, needed to repair lesions that are created during replication in S phase (365, 366). The mechanism of cell cycle exit in prophase that we uncovered here may therefore have arisen as a mechanism to remove cells that have failed at post-replication repair from the proliferative cycle. Interestingly, long ago, prophase cells were reported to revert back into an interphase-like state upon DNA damage, but assumed to re-enter the cell cycle afterwards ( ). We now show that these cells irreversibly exit the cell cycle, already when facing low amounts of damage. Moreover, our results reveal the underlying mechanism, and emphasize the importance of this response in protecting genomic integrity. We find that reinstalling checkpoint reversibility in prophase cells results in an increased carry over of DNA damage from mother to daughter cells. Thus, removal of damaged prophase cells from the cell cycle is an important function of APC/C Cdh1, and impairing this function could promote genomic instability. In this respect it is of interest to note that both loss of Cdh1 and overexpression of Emi1 have been reported in several tumour types (23, 35, 37, 371). Similarly, ATR is well-known for its role in protecting cells from the hazardous effects of replication stress (362, 363, 372). We now show that ATR, in addition to slowing S phase progression in the presence of unreplicated DNA, may fail to inhibit progression into prophase when residual lesions cannot be resolved, but instead drives cells out of the cell cycle once this occurs. 76 Chapter 4

77 Methods Tissue culture and cell line construction htert-immortalized retinal pigment epithelium (RPE-1) cells were maintained in Dulbecco s Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12, Gibco) supplemented with ultraglutamine, antibiotics and 1% fetal calf serum. RPE-1 cells in which a fluorescent tag was introduced in one allele of Cyclin B1 (RPE CCNB1 YFP ) have been described before (175). HMEC-1 cells and MCF1a cells were maintained in Dulbecco s Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12, Gibco) supplemented with ultraglutamine, antibiotics, EGF (2ng/ml), Hydrocortisone (5 ng/ml) and Insulin (1 μg/ml). U2OS cells were maintained in Dulbecco s Modified Eagle Medium (DMEM, Gibco) supplemented with ultraglutamine, antibiotics and 6% fetal calf serum. A fluorescent tag was introduced in one allele of Cyclin B1 (RPE CCNB1 YFP ) in HMEC-1, MCF1a and U2OS cells as described before (175). HEK293 cells were transfected with Fucci constructs which have been described before (37) using X-tremeGENE (Roche) according to manufacturer s protocol. RPE-1 cells expressing ecotropic receptor were infected after 2 days for 24 h and double positive Fucci cells were sorted by FACS two weeks later. The venus tag in the previously described CSII-EF-DHB-Venus contruct (21) was exchanged for mcherry to generate RPE CCNB1 YFP DHB mcherry expressing cells. HEK293 cells were transfected with CSII-EF-DHB-mCherry using X-tremeGENE (Roche) according to manufacturer s protocol. RPE CCNB1 YFP cells were infected after 2 days for 24 h and mcherry positive cells were sorted out by FACS two weeks later. RPE CCNB1 YFP Turq Emi1 cells were generated as described for the DHB mcherry expressing cells above, except that individual grown clones were selected to generate a monoclonal cell line. The venus tag in the previously described pt7-venus-emi1 construct (41) (purchased from addgene, #39854) was exchanged for mturquoise. Subsequently, Cas9 was exchanged for mturq-emi1 in the all in one dox-inducible lentiviral pcw-cas9 construct (purchased from addgene, #5661) using Nhe1 and BamH1 restriction sites. RPE CCNB1 YFP 53BP1 mcherry and RPE CCNB1 YFP - Turq-Emi1-53BP1 mcherry cells were generated using the 53BP1 mcherry construct described before (373). Amphotropic Phoenix cells were transfected with 53BP1 mcherry using X-tremeGENE (Roche) and virus was used to generate 53BP1 mcherry positive cells as described above. Positive cells were sorted out by FACS two weeks later. All cell lines described above were tested negative for mycoplasma contamination. 4 Time lapse microscopy and irradiation Cells were grown in Lab-Tek II chambered coverglass (Thermo Scientific) in DMEM/F12, which was replaced by Leibovitz s L-15 (Gibco) CO 2 -independent medium just before imaging. Images were obtained using a DeltaVision Elite (applied precision) maintained at 37 C equipped with a 1X.4 NA or 2X.75 NA or 4X 1.35 NA lens (lympus) and cooled CoolSnap CCD camera. Only for time lapse imaging of the RPE-Fucci cells, cells were grown in 96-wells plate in DMEM/F12 during filming. Images were obtained using a CCD microscope (Zeiss AxioObserver.Z1 gemot.) maintained at 37 C and 5% CO2 equipped with a 1 x /.25 Achroplan Ph1 lens and cooled Hamamatsu ORCA R2 Black and White CCD-camera. Image analysis was done using ImageJ software. Cells were γ-irradiated using a Gammacell Exactor (Best Theratronics) with a 137 Cs-source. FACS-sort and senescence-associated β-galactosidase assay Cells were trypsinized and re-suspended in Leibovitz s L-15 medium for sorting using a Becton Dickinson FacsAria Sorter or a Beckman Coulter Moflo Astrios. G2 cells were sorted based on Cyclin B1-YFP signal and re-plated for filming. RPE-Fucci S phase cells were sorted based on Azami-Green (AG) and Kusabira-Orange (KO) double positive signal and replated for filming, FACS and Western Blot samples at indicated time-points after the sort. For FACS analysis of PI profiles after the double positive Fucci sort, cells were fixed in ice-cold ethanol at indicated time-points after the sort. Cells were washed with 1xPBS before they were resuspended in 1xPBS + RNAse and PI (1mM, Sigma). PI profiles were analysed using a Becton Dickinson FACSCalibur analyser. FACS experiments to determine spontaneous recovery, suppl. Fig3d, were performed as described before (226). β-gal stainings were performed as described DNA lesions trigger a cell cycle exit in prophase 77

78 previously (226). Chromosome spreads RPE-1 cells were mock irradiated or irradiated with 1 Gy followed by nocodazole addition 1 h after IR to retain cells in mitosis, but exclude cells that were damaged in mitosis. Caffeine was added to control samples to push G2 cells into mitosis with DSBs as a positive control. Mitotic RPE-1 cells were collected by shake-off 4 h after IR, washed in 1xPBS and treated for 25 minutes with 75 mm KCl at 37 C. Cells were spun on coverslips at 18 rpm for 5 minutes in a Cytospin 4 (Thermo Scientific). Cells were permeabilized for 1 minute with PEM buffer (1 mm PIPES, 2 mm EGTA, 1 mm MgSO 4 ; ph 6.8) containing.25% Triton X-1 and then fixed for 1 minutes in 4% paraformaldehyde containing.1% Triton X-1. Fixed cells were washed three times in 1xPBS containing.1% Tween-2 and then blocked for 3 minutes in PEM/3% BSA/.1% Tween-2. Antibody incubations (ACA and γh2ax) were performed overnight at 4 C. Cells were washed three times in 1xPBS containing.1% Tween-2, and then incubated with secondary antibodies and.1 µg/ml DAPI in PEM/3% BSA/.1% Tween-2 for 2 h at room temperature. After three washing steps the coverslips were mounted on microscopic slides with Prolong Gold (Invitrogen) and stored at 4 C. DNA breaks were quantified based on DAPI and γh2ax signal in chromosome spreads. 4 Antibodies, chemicals and oligonucleotides The following antibodies were used in this study: anti-phospho-h3, anti-γh2ax (ser139p) (Upstate); anti-emi1 (Novex); anti-cdh1 (DH1) Neo; anti-cyclin B1 (GNS1), anti-cdk4 (C-22), HSP9 (Santa Cruz); anti-cyclin A2 (H432) (Tebu); anti-tubulin gamma (GTU-88) (Abcam); anti-crest serum (Cortex Biochem); anti-iak1 (Aur A) (BD Transduction labs); anti-gfp (homemade, gift from Geert Kops). The following secondary antibodies were used: peroxidase-conjugated goat anti-rabbit, goat anti-mouse, rabbit anti-goat antibodies were from DAKO. Goat anti-rabbit/alexa 488, Goat anti-mouse/alexa 568, were from Molecular Probes and combined with DAPI for DNA stain. Western blots were performed as described previously (Alvarez-Fernández et al., 21). Immunofluorescence was performed as described previously (Krenning et al., 214). Immunofluorescent staining of Cyclin A and Aurora A was performed after fixation of cells that were tracked by live cell imaging within the Lab-Tek II chambered coverglass (Thermo Scientific). Chemicals used in this study: RO-336 (used at 5 μm), Roscovotine (used at 25 μm) (Calbiochem). The ATR inhibitor ATR-45 was synthesized by the Medicinal Chemistry Shared Resource (Ohio State University) (used at 1 μm). Nocodazole (used at 25 μm), caffeine (used at 5 mm), Doxocycline (used at 1mM), Wee1 inhibitor MK-1775 (used at 3 μm), MG-132 (used at 5 μm) and Aphidicolin (used at.2 or.4 μm) were purchased at Sigma. ON-TARGETplus SMARTpool sirnas targeting Luciferase (GL2 duplex), Cdh1/FZR1, Cdc2 and EMI1/FBXO5 were from Thermo Scientific and were transfected using RNAiMAX (Life Technologies) according to the manufacturer s protocol. All transfections were performed 24 h before experiments. ImageJ macros used for segmenting individual cells and quantifying DNA damage foci Monitoring DNA damage foci requires following individual cells over time. However, faithful automatic tracking of the highly motile RPE cells in densely covered samples proved to be unfeasible. Therefore, a hybrid approach was taken, where single cells were first manually isolated using an in-house developed cell tracking macro in ImageJ (NIH), after which the DNA damage response was fully automatically quantified with a second ImageJ macro. User-assisted tracking and segmentation of single cell (nuclei) is facilitated as follows: Z-stacks are converted to 2D using a maximum intensity projection. For every frame, a square region of defined size around the x,y position of the mouse cursor is copied from the original 3D/4D image stack into a new image stack. The size of the cropped square has to be chosen large enough to fully encompass the cell (nucleus) of interest, which is now centered in the newly generated movie. When holding down the mouse button the time series advances at a desired speed, allowing accurate manual tracking. Single cell (nuclei) are isolated from such tracked-cell movies in the following manner: 3D timelapse movies are projected to 2D via one of several user-defined methods: maximum 78 Chapter 4

79 intensity projection, automatically select sharpest slice, manually select a slice, or via a extended depth of field algorithm. ROIs of candidate nuclei are automatically obtained throughout the image stack by auto-thresholding an outlier-removed median-filtered (.7 µm radius) z-projection of the nuclei channel, followed by a watershed command to separate touching nuclei, and particle analyzer run with size (>4 & <4 µm 2 and circularity (>.25) constraints. In each frame, the distances of all detected ROIs to the x,y center of the image are calculated, after which all except the closest ROI are removed. This procedure thus yields a movie with a single ROI per frame, tightly surrounding the nucleus followed with the mouse in the manual tracking macro. In the detection of DNA damage foci, the foci threshold level is defined by the signal-to-noise ratio (SNR): a (user-set) factor times the standard deviation of the background fluorescence intensity of the nucleus. The latter property is approximated by first crudely removing signal outliers (the foci) and then taking the median and standard deviation of the lower ~8% pixel values in the ROI, respectively. The background intensity is subtracted using a Difference of Gaussians filter. Foci are then identified as regions of adjacent pixels with gray values exceeding the SNR threshold and area larger than a certain minimum. In the procedure, the SNR is the only user-defined parameter, and is iteratively optimized by comparing the detected foci with the original signal in an overlay image. The evolution of the DNA damage foci is quantified by reporting the number of foci, foci intensity, foci area, and the total signal above threshold for each time frame. Sample sizes For all experiments where phenotypic outcome was quantified at least 5 cells per condition in each independent biological replicate were scored, n 5. Exceptions are fig. 1h. n=19-6, Suppl. fig. 1c. n=9-19, fig. 2a. n=25-57, fig. 2c. n=6-15, Suppl. fig. 2f. n>35, fig. 3e. n=13-35, fig 4b. n=18-22, fig 4d. n=19-49, fig 5b,c. n>32. Acknowledgements 4 The authors would like to thank members of the Medema lab for helpful discussions, and the NKI cytometry facility for technical assistance and cell sorting. DNA lesions trigger a cell cycle exit in prophase 79

80 Supplementary figures 4 A Centrosome distance (μm) Unperturbed (n=1) centrosome 2 centrosomes ph3 Ser1 fluorescence intensity (a.u.) B No. of 53BP1-mCherry foci per cell Gy IR 8 Degrade 4 Translocate Degrade Time after irradiation (h) Figure S1. Prophase cells show a unique response to DNA damage (A) Centrosome distance plotted against the intensity of ph3 Ser1 staining. Dots represent 1 individual unperturbed RPE cells from one experiment. (B) DNA damage foci were tracked in time-lapse movies of individual RPE CCNB YFP cells with 53BP1 mcherry expression using automated foci analysis (see methods). Number of foci per cell at indicated time-points after 2 Gy IR is shown. Mean ± s.e.m. of 3 independent experiments. (C) Quantification of Cyclin B1 degradation in prophase cells (selected as in Figure 1H) of indicated cell type. Mean ± s.e.m. of 3 independent experiments. C Cyclin B1 degradation (% of CycB1 YFP -High cells) 5 Gy IR - Prophase RPE U2OS MCF1α HMEC Figure S2. DNA damage-induced Cdk inhibition results in rapid APC/C Cdh1 activation in prophase (A) Cdh1 and Cdc2 knockdown in RPE CCNB YFP cells. (B) Quantification of Cyclin B1 degradation in prophase cells (selected as in Figure 1H) after 5 Gy IR in the presence or absence of proteasome inhibitor MG132. Mean ± s.e.m. of three independent experiments. (C) Time-lapse images representing loss of Cyclin B1 degradation in Cdh1 depleted RPE CCNB YFP cells after 5 Gy IR. All Cdh1-depleted cells translocate Cyclin B1 to the nucleus after 5 Gy IR. (D) Quantification of Cyclin B1 YFP intensity 15 min after 5 Gy IR and color-coded according to subsequent phenotype in Luc- or Cdh1-depleted RPE cells. Mean ± s.d., results are representative of 3 independent experiments. *P<.5 (unpaired t-teest) (E and F) Quantification of Cyclin A (e) and Aurora A (f) fluorescence intensity in cells that either degraded Cyclin B1 YFP, still have Cyclin B1 YFP or are in G1 phase. Mean ± s.d., results are representative of 3 independent experiments. **P<.1 (unpaired t-test) (G) Time-lapse images of RPE CCNB YFP DHB mcherry cells show the higher cytoplasmic signal before mitosis and the increased nuclear signal after mitosis in unperturbed growing cells. Relative Cdk2 activity is determined by dividing the mean Cytoplasmic intensity over the mean nuclear intensity in every frame. (H) Quantification of Cyclin B1 degradation in prophase cells (selected as in Figure 1H) after addition of indicated inhibitors. Mean ± s.d. of 3 independent experiments. (I) Line graphs of the data presented in Figure 2 c, d showing cumulative mitotic entry or cumulative Cyclin B1 degradation. Mean ± s.d. of 3 independent experiments. (J) Line graphs of the data presented in Figure 2 e, f showing cumulative Cyclin B1 degradation. In addition cumulative mitotic entry is shown. Mean ± s.d. of 3 independent experiments. 8 Chapter 4

81 A sirna: Cdc2 Cdh1 Cdk4 D 5 Gy IR 3 * Cyclin B1 YFP intensity (a.u.) 2 1 sirna: Luc Cdh1 Cdc2 Luc Luc Cdh1 B Cyclin B1 degr. (% of CycB1 YFP -High cells) Degrade Translocate Degrade 5 Gy IR - Prophase 8 MG132: - + Translocate E Cyclin A intensity (a.u.) C Luciferase Degrade Cdh1 Translocate :15 1: 2: 2:3 3: :15 1: 2: 2:3 3: 4: 8: 12: 5 Gy IR F 5 Gy IR ns ns ** 5 ** Aurora A intensity (a.u.) Cyc B1-YFP degraded Cyc B1-YFP present G1 cell G Gy 1 Gy :15 :3 :45 1: 1:15 1:3 1:45 2: 2:15 NEBD :15 :3 :45 1: 1:15 1:3 1:45 2: 2:15 CyclinB1-YFP DHB-mCherry (Cdk2 activity) Cyclin B1-YFP DHB-mCherry (Cdk2 activity) H Cyclin B1 degradation (% of CycB1 YFP -High cells) Cdk inh. - Prophase Rosc: RO-336: I Mitotic entry (% of CycB1 YFP -High cells) J Mitotic entry (% of CycB1 YFP -High cells) Cdk inhibition - Prophase Cdk Inh. DMSO RO + Rosc Time after drug addition (h) Unperturbed - Prophase DMSO MK Time after mock IR (h) Cyclin B1 degradation (% of CycB1 YFP -High cells) Cyclin B1 degradation (% of CycB1 YFP -High cells) Cdk inhibition - Prophase 1 DMSO 8 RO + Rosc Time after drug addition (h) Cdk Inh. Unperturbed - Prophase Time after mock IR (h) Mitotic entry (% of CycB1 YFP -Low cells) Mitotic entry (% of CycB1 YFP -High cells) Cdk inhibition - G2 Cdk Inh. DMSO RO + Rosc Time after drug addition (h) 1 Gy IR - Prophase DMSO MK Time after 1 Gy IR (h) Cyclin B1 degradation (% of CycB1 YFP -Low cells) Cyclin B1 degradation (% of CycB1 YFP -High cells) Cdk inhibition - G2 Cdk Inh. DMSO RO + Rosc Time after drug addition (h) 1 Gy IR - Prophase Time after 1 Gy IR (h) DNA lesions trigger a cell cycle exit in prophase 81

82 4 A mag-hgem(1/11) D Mitosis (% of G2 cells) F Mitotic entry (% of CycB1 YFP -High cells) H sirna: Mitotic entry (% of CycB1 YFP -High cells) Luc M G2 S Unperturbed Emi1 G1 Unperturbed - G2 mko2-hcdh1(3/12) E sirna: Luciferase Emi Time after mock IR (h) Cdk inhibition - Prophase DMSO 6 RO-336 Roscovitine 4 RO + Rosc Time after drug addition (h) Cyclin B1 degradation (% of CycB1 YFP -High cells) mko2 sirna: HSP9 (α/β) Emi1 (short) Emi1 (long) Cyclin B1 degradation (% of CycB1 YFP -High cells) G1 Luc mag Sort Emi1 S G2 & M Unperturbed - G Time after mock IR (h) Cdk inhibition - Prophase Time after drug addition (h) Cyclin B1 YFP Intensity (a.u.) Mitotic entry (% of CycB1 YFP -Low cells) Mitotic entry (% of CycB1 YFP -Low cells) B 2N 4N G sirna: Luc Luc Emi1 γ-ir: 5Gy 5Gy 1Gy IR ** 1 Gy IR - G2 sirna: Luciferase Emi1 Nuclear Retention Direct Degradation Time after 1 Gy IR (h) Cdk inhibition - G2 DMSO RO-336 Roscovitine RO + Rosc Time after drug addition (h) 3 1 As. Time post s-phase sort (h) C Mitosis (% S phase sorted cells) Cyclin B1 degradation (% of CycB1 YFP -Low cells) Cyclin B1 degradation (% of CycB1 YFP -Low cells) Unperturbed (n=51) 1 Gy IR - G2 Mean: 1,5 h SD: 1,7 h 1 2 Time after S phase sort (h) Time after 1 Gy IR (h) Cdk inhibition - G Time after drug addition (h) I Cyclin B1 degradation (% of High CycB1 YFP cells) Gy IR - Prophase No doxycyclin Doxycyclin (Emi1 o.e.) Time after 5 Gy IR (h) J Doxycyclin (h): Emi1 RNAi: GFP Emi1 CDK4 RPE CCNB1 YFP RPE CCNB1 YFP mturq-emi1 mturq-emi1 endogenous 82 Chapter 4

83 Figure S3. Emi1 acts to maintain recovery competence in G2 cells (A) Scheme representing how early S phase cells were obtained by FACS sorting of double positive RPE-Fucci cells (Azami-Green (mag) and Kusabira-Orange (mko2)). (B) FACS plots showing the cell cycle profiles at indicated timepoints after the early S phase sort based on PI staining. n 1 cells measured per time-point from one experiment. (C) Cumulative mitotic entry of unperturbed RPE-Fucci cells sorted in early S phase and followed by time-lapse imaging after re-plating shows the highly synchronous progression through S/G2 phase. n=51 cells from one experiment. (D and E) A partial knock down of Emi1 was obtained by titration of siemi1 in RPE CCNB YFP cells. Mitotic entry of undamaged G2 cells upon Luc- and Emi1-depletion was determined as described previously (226). Representative graph and western blot of two independent experiments. (F) Line graphs of the data presented in Figure 3b and c showing cumulative mitotic entry or cumulative Cyclin B1 degradation. Mean ± s.d. of 3 independent experiments. (G) Cyclin B1 YFP intensity measured at the first frame after IR and color-coded according to subsequent phenotype in Luc- or partially Emi1-depleted RPE cells. Dots represent individual cells. Mean ± s.d., results are representative of three independent experiments. **P<.1 (unpaired t-test) (H) Line graphs of the data presented in Figure 3 d showing cumulative Cyclin B1 degradation. Mean ± s.d. of 3 independent experiments. (I) Line graph of the data presented in Figure 3f. Mean ± s.d. of 3 independent experiments. (J) Turq EMI overexpression induced by doxycycline addition. Representative western blot of two independent experiments. 4 A Mitotic entry (% of CycB1 YFP -High cells) Gy IR - Prophase No doxycyclin Doxycyclin (Emi1 o.e.) Time after irradiation (h) B 1 Gy IR + Caffeine sirna: Luc Cdh1 Figure S4. Hypersensitivity to DNA damage in prophase is needed to protect genomic stability (A) Line graph of the data presented in Figure 4d. Mean ± s.d. of 3 independent experiments. (B) Quantification of mitotic cells with broken chromosomes as in Figure 4c. Caffeine was added 3 min before IR to override DNA damage checkpoint signalling. Average of 2 independent experiments is shown. Mitotic cells with broken chromosomes (%) n = 99 n = 1 DNA lesions trigger a cell cycle exit in prophase 83

84

85 Chapter 5 The mitotic checkpoint complex protects against DNA damage-induced activation of the APC/C Cdh1 Lenno Krenning, Marmendia R. Meester, Mar Soto Ruiz de la Torre, Jonne A. Raaijmakers and René H. Medema Division of Cell Biology I, The Netherlands Cancer Institute Plesmanlaan 121, 166 CX Amsterdam, The Netherlands 5 The MCC protects against DNA damage-induced activation of the APC/C Cdh1 85

86 Abstract In mitosis, the mitotic checkpoint complex (MCC) inhibits the anaphase promoting complex/cyclosome (APC/C) to allow time for chromosome alignment. It was previously described that MCC assembly already takes place during interphase. However, the function of the MCC that is assembled during interphase is currently unclear. Here, we show that upon DNA damage in G2, the MCC proteins Mad2 and BubR1 prevent premature activation of the APC/C Cdh1, contributing to the reversibility of a DNA damage-induced G2 arrest. Challenging cells depleted of either Mad2 or BuBR1 with a low dose of γ-irradiation causes Cyclin B1 degradation, which is completely dependent on APC/C Cdh1. We find that interphase cells lacking Mad2 express low levels of Emi1, while overexpression of Emi1 levels in Mad2-depleted cells restores G2 checkpoint reversibility. These findings reveal a novel role for the MCC in interphase. In addition, they show that the MCC can control Cdh1-dependent APC/C activity, in addition to its widely known function as an APC/C Cdc2 inhibitor Chapter 5

87 Introduction In order to cope with lesions to the DNA, cells have evolved an elaborate signaling response, called the DNA damage response (DDR). The DDR can detect DNA damage and halt cell cycle progression to allow time for DNA repair (56). Following DNA repair, the DDR is silenced and cells may resume the cell cycle. Alternatively, in the face of severe or irreparable DNA damage, cells may exit the cell cycle and undergo senescence (Reviewed in chapter 2). As (proliferating) cells undergo numerous DNA damaging lesions every day (56, 374), reversibility of a DNA damage-induced cell cycle arrest is essential to allow continuous proliferation of cells. Previous work has shown that the DDR and the reversibility of a cell cycle arrest differ throughout the cell cycle (Reviewed in chapter 2). For instance, whereas a G1 arrest remains reversible up to several days following DNA damage, reversibility of a G2 arrest is lost within hours (Chapter 3). In addition, it was shown that permanent withdrawal from the cell cycle in response to DNA damage occurs mainly in G2 (237). Such a cell cycle exit from G2 is accompanied by activation of the anaphase promoting complex/cyclosome (APC/C) in complex with its cofactor Cdh1, resulting in the degradation of pro-mitotic proteins and loss of the G2-identity of the cell (Chapter 3) (23, ). The APC/C is a multiprotein E3 ubiquitin ligase targeting many cell cycle proteins for degradation (354). Timely activation of the APC/C ensures proper progression through the cell cycle and is achieved through regulation of the two APC/C coactivators, Cdc2 and Cdh1. Early mitotic inhibitor1 (Emi1) inhibits the APC/C from the start of S phase until prophase, when Emi1 is targeted for degradation by the SCF βtrcp ubiquitin ligase complex (41, 46, 51-53, 375). At this time, the presence of kinetochores that are unattached to microtubules activate the spindle assembly checkpoint (SAC) (2). When active, the SAC promotes the assembly of the mitotic checkpoint complex (MCC), which binds to Cdc2 and inhibits Cdc2-dependent APC/C activity, thereby stabilizing APC/C Cdc2 substrates (28, 376, 377). Simultaneously, Cdk-dependent phosphorylation of Cdh1 blocks APC/C Cdh1 activity (25, 34, 35, 44). Upon bipolar attachment of all chromosomes to the mitotic spindle, the MCC is disassembled and the production of new MCC is prohibited, which results in the activation of the APC/C Cdc2. In turn, active APC/C Cdc2 causes the destruction of securin and Cyclin B1, which induces chromosome segregation (19, 354). In addition, the loss of Cdk1 activity resulting from Cyclin B1 destruction allows for APC/C Cdh1 activation during the later stages of mitosis and during G1 (35, 36, 378). Although the primary function of MCC-Cdc2 binding is to prevent premature chromosome segregation during mitosis, by inhibiting APC/C Cdc2, the MCC already binds to Cdc2 in interphase (379). Indeed, it was shown that interphase assembly of MCC-Cdc2 complexes contributes to inhibition of the APC/C during the early stages of mitosis (38, 381). Since APC/C activity also needs to be inhibited in DNA damaged cells, in order to maintain the ability of revert a DNA damage-induced cell cycle arrest in G2, we investigated a potential role for the MCC in APC/C regulation following DNA damage. Here, we identify an essential role for the MCC in maintaining reversibility of a DNA damage-induced G2 arrest. More specifically, we find that Mad2 and BubR1 act to prevent DNA damage-induced APC/C Cdh1 activation in G2. Cells lacking Mad2 exhibit decreased Emi1 expression, and forced Emi1 overexpression partially alleviates the hypersensitivity to DNA damage observed after Mad2 depletion. Collectively, these data identify an important role for the MCC during a DNA damage-induced G2 cell cycle arrest, and, for the first time, link the MCC to regulation of Cdh1-dependent APC/C activity. 5 Results The mitotic checkpoint complex proteins Mad2 and BubR1 prevent Cyclin B1 degradation in response to DNA damage DNA damage in G2 can lead to premature APC/C Cdh1 activation and cell cycle exit (Chapter 3) ( , 242). The finding that assembly of functional MCC complexes already occurs during interphase led us to investigate a potential role for the MCC in preventing premature APC/C activation upon DNA dam- The MCC protects against DNA damage-induced activation of the APC/C Cdh1 87

88 A B C D % of G2 cells Mitosis Degrade Translocate Degrade Time in mitosis (h) % of G2 cells Mitosis Degrade Translocate Degrade Time in mitosis (h) sirna: Ctrl Mad2 Ctrl Mad2 sirna: Ctrl Mad2 Gy 1 Gy STLC E F 1 1 sirna: Ctrl BubR1 Ctrl BubR1 Gy 1 Gy sirna: Luc BubR1 STLC 5 Cyclin B1-YFP fluorescence intensity (a.u.) sirna: Ctrl Mad2 Cyclin B1-YFP fluorescence intensity (a.u.) sirna: Ctrl BubR1 Figure 1. The mitotic checkpoint proteins Mad2 and BubR1 prevent DNA damage-induced APC/C activation (A) Quantification of the indicated phenotypes of RPE CCN- B1 YFP cells. Cells were transfected with control or Mad2-targeting sirnas and irradiated 24 h later. Subsequently, the cells were followed using time-lapse microscopy for 15 h. STLC was added to the non-irradiated cells to allow analysis of SAC-dependent mitotic arrest (shown in B). Means ± SD of 3 independent experiments. (B) Quantification of the duration of SAC-dependent mitotic arrest of the non-irradiated cells in (A). Means ± SD of 3 independent experiments. (C) Same as in (A), except sirna oligos targeting BubR1 were used. Means ± SD of 3 independent experiments. (D) Quantification of SAC-dependent mitotic arrest in cells used in (C). Means ± SD of 3 independent experiments. (E) Cyclin B1-YFP fluorescence intensities were measured at the last time point before mitosis, in non-irradiated cells used in (A). A representative of 3 independent experiments is shown. (F) Cyclin B1-YFP fluorescence intensities were measured at the last time point before mitosis, in non-irradiated cells used in (C). A representative of 3 independent experiments is shown. A 1 % of G2 cells si Mad2: si Cdh1: Gy 1 Gy + + Mitosis Degrade Translocate Degrade Translocate (no degradation) B si Mad2: si Cdh1: Time in mitosis (h) STLC + + Figure 2. The mitotic checkpoint protein Mad2 prevents IR-induced activation of the APC/C Cdh1 (A and B) Quantification of the indicated phenotypes of RPE CCNB1 YFP cells. Cells were transfected with the indicated sirnas and irradiated 24 h later. Subsequently, the cells were followed using time-lapse microscopy for 15 h (A). STLC was added to the non-irradiated cells in order to allow analysis of SAC-dependent mitotic arrest (shown in B). Means ± SD of 3 independent experiments. 88 Chapter 5

89 age. In order to visualize cell fate decisions after DNA damage, we performed time-lapse microscopy of irradiated RPE-1 cells in which eyfp was introduced in the Cyclin B1 gene (RPE CCNB1 YFP ) (175). Using this cell line, we have previously shown that the cell cycle exit that occurs in G2 following DNA damage is preceded by nuclear retention of Cyclin B1 and subsequent APC/C Cdh1 activation (Chapter 3), whereas damaged prophase cells exit the cell cycle by immediate APC/C Cdh1 activation (Chapter 4). Since Mad2 is required for chromosome segregation ( ), and chromosome missegregation can result in activation of ATM and p53 (386, 387), we performed our analysis of Mad2-depleted cells 24 h after sirna transfection, when Mad2-depleted cells are unlikely to have undergone a cell division with a non-functional mitotic checkpoint. This is supported by the finding that, at 24 h after sirna transfection, Mad2-depleted cells entered mitosis at a similar rate and to a similar extent as control cells (Figure S1A). To address the efficiency of Mad2 depletion, we analyzed mitotic duration of RPE CCNB1 YFP cells that were arrested in mitosis in a SAC-dependent manner. Exit from mitosis occurred after an average arrest of 4 h in cells transfected with sirnas targeting Mad2, compared to an average arrest of >8 h in control cells, indicating that Mad2-function was at least partially impaired (Figure 1B). In response to 1 Gy γ-irradiation (IR), we observed that ~65% of Cyclin B1-YFP-positive control cells entered mitosis (Figure 1A). In addition, ~25% degraded Cyclin B1 directly, indicative of a cell cycle exit from prophase (Chapter 4), and ~1% underwent nuclear retention and subsequent degradation of Cyclin B1 (Figure 1A), suggesting that these cells exit the cell cycle from G2 as a result of p53 activation (Chapter 3). Mad2 depletion hardly increased this fraction of cells (Figure 1A), suggesting that p53 activation after DNA damage is largely unaltered in Mad2 depleted cells. In sharp contrast, Mad2 depletion markedly increased the fraction of cells that directly degrade Cyclin B1 in response to IR, which was accompanied by decreased mitotic entry (Figure 1A). Similar results were obtained when depleting Mad2 using two single sirnas (Figure S1C and S1D). These results show that Mad2 is required to prevent Cyclin B1 degradation in response to IR. To investigate whether this function is unique to Mad2, or a function of the MCC, we analyzed the effect of IR on BubR1-depleted RPE CCNB1 YFP cells. We observed that BubR1 depletion caused ~2% of non-irradiated cells to degrade Cyclin B1, compared to ~5% of control cells. Following IR, ~75% of control cells entered mitosis, whereas only ~15% of BubR1-depleted cells managed to enter mitosis (Figure 1C). Instead, similar to Mad2 depletion, depletion of BubR1 increased the percentage of cells degrading Cyclin B1 directly (Figure 1C). In addition, we also observed a small increase in the fraction of cells that showed nuclear accumulation and subsequent degradation of Cyclin B1 after IR. This, together with the increase in spontaneous Cyclin B1 degradation in non-irradiated, BubR1-depleted cells indicates that BubR1 depletion by itself already causes some stress, which may result in more efficient p53 induction in response to 1 Gy IR. The efficiency of BubR1 depletion was addressed by analyzing the duration of a SAC-dependent mitotic arrest of cells transfected with scrambled- or BubR1-targeting sirnas. Mitotic exit occurred after an average arrest of ~2 h in cells transfected with BubR1 sirnas, compared to an average arrest of ~12 h in control cells, indicating a partial BubR1 depletion (Figure 1D). Collectively, these data show that Mad2 and BubR1 are both involved in the regulation of IR-induced Cyclin B1 degradation. We therefore conclude that this is a function of the MCC. 5 Mad2 and BubR1 inhibit APC/C activity specifically following γ-irradiation Several reports have shown that the MCC is involved in the inhibition of APC/C in interphase (38, 381). More specifically, these studies have shown that Mad1- or BubR1-knockout cells progress into mitosis with lower levels of Cyclin B1, which the authors argued to be a result of APC/C Cdc2 -dependent degradation of Cyclin B1 in (late) G2 (38, 381). To test whether Mad2 and BubR1 are required to inhibit APC/C activity during unperturbed cell cycle progression in RPE1 cells, we measured Cyclin B1-YFP levels immediately before mitotic entry (Figure 1B and 1D, respectively). Compared to control cells, we observed no difference in Cyclin B1 levels just prior to mitotic entry in both Mad2- or BubR1-depleted RPE CCNB1 YFP cells (Figure 1E and 1F, respectively). These results show that, in our setting, Mad2 and BubR1 contribute to the inhibition of the APC/C after IR, but not during unperturbed G2 progression. The MCC protects against DNA damage-induced activation of the APC/C Cdh1 89

90 A B C 1 14 sirna: Emi1 Mad2 Ponceau Ctrl Mad2 % of G2 cells Mitosis Degrade Translocate Degrade Translocate (no degradation) Time in mitosis (h) si Mad2: Emi1 o.e.: Gy 1 Gy + + Emi1 o.e.: si Mad2: Figure 3. Mad2 depletion causes decreased expression of Emi1 and hypersensitivity to DNA damage (A) Emi1 protein levels analyzed by western blot. Cells were transfected with the indicated sirnas, and subsequently synchronized in G2 via thymidine block and release. In G2, cells were irradiated with 1 Gy and collected after 1 h. A representative western blot is shown. (B and C) Quantification of the indicated phenotypes of RPE CCNB1 YFP cells. Cells were transfected with the indicated sirnas and irradiated 24 h later. Where indicated Emi1 overexpression was induced 3 h before IR. Subsequently, the cells were followed using time-lapse microscopy for 15 h. STLC was added to the non-irradiated cells in order to allow analysis of SAC-dependent mitotic arrest (shown in C). Means ± SD of 3 independent experiments. 5 Mps1 activity is not required to prevent APC/C activation upon DNA damage Formation of the Cdc2-bound MCC in interphase occurs at the nuclear pores and was shown to require the kinase activity of Mps1 (379, 381). Therefore, we wondered whether interphase Mps1-activity is required to prevent premature APC/C activation upon DNA damage. To address this, we treated RPE CCNB1 YFP cells with the Mps1-inhibitor NMS-5 8 h prior to irradiation. This way, cells will have progressed through G2 in the absence of Mps1-activity at the moment of irradiation, and therefore Cdc2 should not be bound to the MCC (379, 381). Inhibition of Mps1, as evidenced by a complete loss of SAC-dependent mitotic delay (Figure S1F), hardly affected mitotic entry of both non-irradiated cells and cells irradiated with 1 Gy IR (Figure S1G). However, it should be noted that in these experiments, the amount of IR-induced Cyclin B1 degradation as well as Cyclin B1 nuclear translocation and degradation was relatively high, potentially masking any Mps1-dependent effects. Nonetheless, these results suggest that Mps1 activity is not required to prevent Cyclin B1 degradation in response to IR, which suggests that inhibition of Cdc2 by the MCC does not play a role in regulating Cyclin B1 degradation following IR. This notion, combined with previous observations that IR results in activation of Cdh1-dependent APC/C activity and Cyclin B1 degradation (Chapters 3 and 4) ( , 242), led us to hypothesize that the MCC might be involved in regulating IR-induced APC/C Cdh1 activation. Mad2 helps to prevent premature activation of APC/C Cdh1 after DNA damage We and others have shown that DNA damage can cause premature activation of APC/C Cdh1 in G2 cells (Chapters 3 and 4) ( , 242). However, at present there is no evidence for APC/C Cdh1 regulation by the MCC. Hence, we set out to test whether the MCC specifically prevents APC/C Cdh1 activation in response to DNA damage, rather than APC/C Cdc2. Therefore, we transfected RPE CCNB1 YFP cells with sirnas targeting Mad2, Cdh1 or both, and asked if Cdh1 depletion could prevent Cyclin B1 degradation caused by IR in Mad2-depleted cells. Again, Mad2 depletion hardly affected mitotic entry of non-irradiated cells (Figure 2A). Cdh1 depletion prevented spontaneous Cyclin B1 degradation in non-irradiated control- and Mad2-depleted cells (Figure 2A). Following 1 Gy IR, ~3% of control cells degraded Cyclin B1, ~5% translocated and degraded Cyclin B1, and ~65% entered mitosis (Figure 2A). Also, in line with our previous observations, Mad2 depletion increased the fraction of cells that directly degraded Cyclin B1 upon irradiation, resulting in decreased mitotic entry (Figure 2A). In agreement with previous find- 9 Chapter 5

91 ings (Chapters 3 and 4), Cdh1 depletion prevented IR-induced Cyclin B1 degradation. Instead, these cells retained Cyclin B1 in the nucleus and remain arrested in G2 (Figure 2A). In contrast, in Mad2/Cdh1 co-depleted cells, mitotic entry was rescued in a large fraction of cells following irradiation, while a similar fraction of cells retain nuclear Cyclin B1 upon irradiation compared to Cdh1-depleted control cells (Figure 2A). Mad2 knockdown was confirmed by measuring the time of SAC-dependent mitotic arrest (Figure 2B). In addition, Cdh1 depletion was confirmed due to a lack of Cyclin B1 degradation (Figure 2B and Chapter 4). Collectively, this shows that Mad2 functions to restrict premature activation of the APC/C Cdh1 after IR. Mad2 depletion leads to decreased expression of Emi1 Emi1, which is expressed from the start of S phase and is degraded during prophase, contributes to the inhibition of the APC/C Cdh1 during interphase (45, 51). Previous work from our lab has shown that decreased Emi1 levels severely sensitize G2 cells to DNA damage (Chapter 4). In fact, reduced Emi1 levels cause G2 cells to activate the APC/C Cdh1 in response to low dose IR (Chapter 4), a phenotype highly similar to that observed after IR of Mad2- or BubR1-depleted cells (Figures 1A and 1C). We therefore set out to investigate whether Emi1 protein levels are affected in Mad2-depleted cells. To this end, we analyzed Emi1 protein levels in 1 Gy irradiated G2 cells, transfected with scrambled or Mad2-targeting sirnas. We observed that Emi1 protein levels were markedly reduced in Mad2-depleted cells compared to control cells (Figure 3A). This shows that Mad2, and possibly the MCC, is involved in the regulation of Emi1 protein levels. Since we have previously shown that decreased Emi1 protein levels causes hypersensitivity to DNA damage (Chapter 4), we hypothesized that the decrease in Emi1 levels following Mad2 depletion causes these cells to become hypersensitive to DNA damage. To test this hypothesis, we made use of RPE CCNB1 YFP cells in which we can induce mturq Emi1 overexpression (Chapter 4). We observed that Emi1 overexpression increased mitotic entry of Mad2-depleted cells (Figure 3B). Whereas ~85% of control RPE CCNB1 YFP mturq Emi1 (non-induced) cells entered mitosis and ~15% spontaneously degraded Cyclin B1, all of the cells entered mitosis upon Emi1 overexpression (Figure 3B). Similarly, ~8% of the Mad2-depleted RPE CCNB1 YFP mturq Emi1 cells entered mitosis and ~2% spontaneously degraded Cyclin B1 (Figure 3B). Again, spontaneous degradation was completely prevented by Emi1 overexpression, causing all cells to enter mitosis (Figure 3B). Irradiation of RPE CCNB1 YFP mturq Emi1 cells with 1 Gy IR caused ~5% of the non-induced control cells to degrade Cyclin B1 (Figure 3B). In addition, ~1% showed nuclear translocation and degradation of Cyclin B1, and another ~4% entered mitosis. Overexpression of Emi1 greatly increased the fraction of cells entering mitosis (~9%), and prevented the degradation of Cyclin B1 in the ~1% of cells that translocated Cyclin B1 to the nucleus following 1 Gy of IR (Figure 3B). The vast majority of Mad2-depleted RPE CCNB1 YFP mturq Emi1 cells (~85%) degraded Cyclin B1 in response to 1 Gy of IR (Figure 3B). In addition, ~5% entered mitosis and another ~1% translocated Cyclin B1 to the nucleus followed by its degradation. In contrast, Emi1 overexpression in irradiated, Mad2-depleted RPE CCNB1 YFP mturq Emi1 cells increased mitotic entry (~3% with Emi1 overexpression compared to ~5% without Emi1 overexpression). The remaining ~7% of cells all translocated Cyclin B1 to the nucleus (Figure 3B). Degradation was completely prevented however, due to Emi1-dependent APC/C-inhibition (Figure 3B). As before, the extent of Mad2 depletion was analyzed by measuring the duration of a SAC-dependent mitotic arrest. This showed that the SAC-dependent mitotic arrest was ablated in cells transfected with Mad2-targeting sirnas (Figure3C), indicative of successful Mad2 depletion. This effect was not influenced by Emi1 overexpression (Figure 3C). Mad2 depletion caused only a mild decrease of unperturbed mitotic entry in RPE CCNB1 YFP mturq Emi1 cells, whereas it severely reduced mitotic entry following 1 Gy IR. Simultaneously, Emi1 overexpression in Mad2-depleted cells resulted in mitotic entry of a similar fraction of cells compared to control cells following 1 Gy of IR. We therefore conclude that the hypersensitivity observed in Mad2-depleted cells is caused by decreased expression of the APC/C Chd1 -inhibitor Emi1. 5 The MCC protects against DNA damage-induced activation of the APC/C Cdh1 91

92 Discussion 5 It is widely accepted that the MCC functions to inhibit APC/C Cdc2 function during the early stages of mitosis (2, 18, 19, 354). In mitosis, the MCC is required to prevent the APC/C Cdc2 -dependent onset of anaphase until all chromosomes have attached to microtubules emanating from opposite spindle poles. Here, we show that the MCC-proteins Mad2 and BubR1 also perform a function outside of mitosis. Specifically, Mad2 and BubR1 are required to prevent premature APC/C Cdh1 activation following DNA damage in G2, and thereby contribute to reversibility of a DNA damage-induced G2 arrest. This, for the first time, shows that the MCC proteins Mad2 and BubR1 are involved in the regulation of Cdh1-dependent APC/C activity, suggesting that the MCC has an additional function outside of mitosis. APC/C Cdh1 regulation by the MCC seems to be independent of MCC binding to Cdc2. Inhibition of Mps1 during G2, which is known to cause dissociation of MCC-proteins from Cdc2 (379, 381), does not result in increased APC/C activation upon IR (Figure S1G). Therefore, we propose that the MCC performs an additional function that somehow contributes to APC/C Cdh1 -inhibition following DNA damage. Indeed, we find reduced levels of Emi1 in cells depleted of Mad2 (Figure 3A). Emi1 is a potent inhibitor of APC/C Cdh1 (388), and we have previously shown that Emi1 is essential to maintain the reversibility of a G2 arrest after IR (Chapter 4). We now show that lowered Emi1 expression caused by Mad2 depletion sensitizes G2 cells to DNA damage, as Emi1 overexpression (partially) restores the reversibility of a G2 arrest (Figure 3B). It remains to be determined whether depletion of BubR1, similarly to Mad2 depletion, results in a decrease of Emi1 protein levels, and whether the reduced mitotic entry following depletion of BubR1 can be rescued by overexpression of Emi1. Throughout the cell cycle, Emi1 levels are regulated through E2F-dependent transcription as well as SCF βtrcp -dependent degradation (45, 51, 389). Therefore, the reduction in Emi1 expression that we observe in Mad2-depleted cells (Figure 3A), and the resulting hypersensitivity to DNA damage (Figure 3B), may be due to alterations in either of these two processes. In G2 and prophase, turnover of Emi1 is regulated by the SCF βtrcp -complex (51, 389). During G2, however, Emi1 degradation is prevented through binding to Evi5 or Pin1, which both can prevent SCF βtrcp -dependent degradation of Emi1 (389, 39). The interaction of Emi1 with Pin1 and Evi5 is lost as cells progress into mitosis (389, 39). Therefore, Pin1 and Evi5 promote the expression of Emi1 during G2, and their inhibition sets the stage for degradation of Emi1 as cells progress into prophase (51). Thus, it will be interesting to investigate whether the MCC may, similar to Evi5 or Pin1, prevent SCF βtrcp -dependent turnover of Emi1 or, alternatively, might regulate the turnover of the Emi1-stabilizing proteins Evi5 and Pin1 themselves. Alternatively, the MCC may promote the transcription or translation of Emi1. This, however, is unlikely, as there is no clear evidence linking Mad2 or BubR1 to the regulation of transcription. In conclusion, the data presented here suggest that the MCC acts as a bi-functional regulator of cell cycle progression, allowing reversibility of a cell cycle arrest in G2 upon mild DNA damage, and preventing anaphase onset in the presence of unattached kinetochores. 92 Chapter 5

93 Materials and methods Tissue culture, cell lines, oligonucleotides and transfections RPE CCNB1 YFP cells were described before (175) and RPE CCNB1 YFP cells inducibly overexpressing mturq Emi1 were described in chapter 4. Cells were maintained in Dulbecco s modified Eagle Medium (DMEM)/F12 supplemented with 1% fetal calf serum and antibiotics. On-TARGETplus SMARTpool Non-Targeting sirnas as well as sirnas targeting Mad2, BubR1, Cdh1/FZR1 were from Dharmacon. All sirna transfections were performed 24 h before the start of the experiments, using RNAiMAX (Life Technologies) according to manufacturers protocol. Time-lapse microscopy, irradiation and chemicals Live cell imaging and irradiation were performed as described before (226). Chemicals used in for this study were doxycycline and STLC (used at 1 μm and 1.5 μm, respectively) and were purchased from Sigma. Acknowledgements The authors would like to thank members of the Medema lab for useful discussions and suggestions. This work was supported by grants from the NWO Graduate Program (22.1.3) to L.K., and the NWO Gravitation Program (Cancer Genomics Center, CGC.nl), the Dutch Cancer Society (KWF) (NKI ) and the European union seventh research framework grant 213 (67722) to R.M. 5 The MCC protects against DNA damage-induced activation of the APC/C Cdh1 93

94 Supplementary figure A 1 Ctrl sirna - Unirradiated B 1 Mad2 sirna - Unirradiated % of G2 cells Mitosis Degrade Translocate Degrade % of G2 cells Mitosis Degrade Translocate Degrade Time (h) Time (h) C 1 D 1 E 1 F 1 % of G2 cells Mitosis Degrade Translocate Degrade Time in mitosis (h) Time in mitosis (h) % of G2 cells Mitosis Degrade Translocate Degrade 5 1 Gy IR: - + sirna: Mad2 # Mad2 #14 sirna: - Mad2 #13 Mad2 #14 STLC Mps1i: - + STLC Mps1i: Gy 1 Gy Figure S1. The mitotic checkpoint proteins Mad2 and BubR1 prevent DNA damage-induced APC/C activation (A and B) Quantification of the indicated phenotypes of non-irradiated RPE CCNB1 YFP cells. The cells were transfected with the indicated sirnas and 24 h thereafter followed using time-lapse microscopy for 15 h. Means ± SD of 3 independent experiments. (C) Quantification of the indicated phenotypes of RPE CCNB1 YFP cells. Cells were transfected with control or Mad2-targeting sirnas and irradiated 24 h later. Subsequently, the cells were followed using timelapse microscopy for 15 h. STLC was added to the non-irradiated cells to allow analysis of SAC-dependent mitotic arrest (shown in D). (D) Quantification of the duration of SAC-dependent mitotic arrest of the non-irradiated cells in (C). (E) Quantification of the duration of SAC-dependent mitotic arrest of the non-irradiated cells in (F). Means ± SD of 3 independent experiments. (F) RPE CCNB1 YFP cells were treated with a specific inhibitor for Mps1. 8 h later, the cells were irradiated with the indicated doses and followed for the subsequent 15 h using time-lapse microscopy. Means ± SD of 3 independent experiments. 94 Chapter 5

95 The MCC protects against DNA damage-induced activation of the APC/C Cdh1 95 5

96

97 Chapter 6 Time-dependent activation of p53 in mitosis controls the mitotic timer checkpoint Lenno Krenning and René H. Medema Division of Cell Biology I and Cancer Genomics Center, The Netherlands Cancer Institute Plesmanlaan 121, 166 CX Amsterdam, The Netherlands Manuscript in preparation 6 Time-dependent activation of p53 in mitosis controls the mitotic timer checkpoint 97

98 Abstract The cell cycle culminates in the equal distribution of the replicated genome to the two daughter cells during mitosis. To achieve this, the duplicated chromosomes need to attach to the mitotic spindle in a bipolar fashion, in order for the cell to be able to properly segregate sister chromatids to the respective daughter cells. Attachment to the spindle is error-prone, therefore cells are equipped with a checkpoint that delays anaphase onset until all chromosomes have stably attached to the spindle. Interestingly, recent work has shown that the duration of mitosis determines whether or not the daughter cells will continue to divide. Thus, problems to achieve correct attachments can lead to a cession of cell proliferation, but the exact molecular mechanism underlying this mitotic timer remains unknown. Here, we show that p53 levels rise as a function of the time spent in mitosis. We find that the negative regulator of p53, Mdm2, is absent in mitotic cells. Indeed, mitotic cells are insensitive to the Mdm2-inhibitor Nutlin-3a, and forced expression of Mdm2 leads to continued proliferation following a prolonged mitosis. Finally, we show that a G1 arrest following prolonged mitosis is independent of DNA damage signaling, and results in the induction of senescence. Taken together, these data show that p53 accumulates over time in mitosis, likely due to the absence of Mdm2 in mitotic cells. This mitotic timer could play an important role in the prevention of proliferation of daughter cells generated from a problematic cell division Chapter 6

99 Introduction Accurate segregation of the genome is essential for the maintenance of genomic integrity. During mitosis, the replicated genome is equally distributed across the two daughter cells. This requires the formation of a bipolar spindle, consisting of microtubules emanating from opposing spindle poles, which attach to the DNA through structures called kinetochores, allowing sister chromatids to be pulled to opposite sides of the cell (391). Attachment occurs at random such that many incorrect microtubule-kinetochore attachments can occur during the early stages of mitosis (392). These incorrect attachments are resolved through destabilization of the kinetochore-microtubule interaction, yielding an unattached kinetochore that can engage in a new microtubule interaction. If left uncorrected, these incorrect attachments would lead to the improper segregation of sister chromatids, causing aneuploidy in mammalian tissue culture cells (2, 393, 394). The spindle assembly checkpoint (SAC) senses unattached kinetochores and generates a wait-anaphase signal, which allows time for kinetochores to attach to microtubules. At metaphase, once all kinetochores are stably attached to microtubules, the SAC will be silenced. This allows activation of the anaphase promoting complex/cyclosome (APC/C) and causes progression into anaphase (352, 354, 395). Nonetheless, some incorrect kinetochore-microtubule attachments do not prevent SAC silencing, resulting in unequal sister chromatid segregation (2, 396). Mitosis is typically completed within one hour (397-4), however, depending on the amount of monooriented chromosomes at the start of mitosis, it may take up to three hours (399). In addition, chemical attenuation of microtubule dynamics can arrest cells in mitosis in a SAC dependent manner for more than 2 hours depending on the cell type and species (397, 4, 41). Cells in which the SAC cannot be satisfied will eventually slip out of mitosis (42, 43) and arrest in the following G1 phase as tetraploid cells (397, 44) or undergo apoptosis during or after mitosis (45-47). The G1 arrest of tetraploid cells following mitotic slippage was shown to be dependent on the duration of mitosis (48). More recently, evidence was presented for the existence of a checkpoint (the mitotic timer) that could restrict the proliferative capacity of cells that are derived from a potentially problematic mitosis, even if that division generated two (near-) diploid daughter cells (49). The progeny proliferates readily if a cell completes mitosis within ~9 min, whereas extending mitosis over this threshold generated daughter cells that arrested in G1 phase, despite normal completion of mitosis (49). Much like the tetraploidy checkpoint (48), the mitotic-timer-induced arrest was shown to depend on p53 (49). p53 is a tumor suppressor that is activated in response to a wide range of cellular stresses (41). In unstressed conditions p53 levels and activity are regulated through Mdm2-dependent ubiquitination, targeting p53 for degradation (229). Upon stress, Mdm2-mediated turnover of p53 is inhibited, resulting in p53 stabilization and activation. This results in the expression of a large array of p53 target genes, amongst which the Cyclin-dependent kinase (Cdk)-inhibitor p21, which can promote a cell cycle arrest (82, 83, 3, 411). Here we set out to investigate the molecular mechanism underlying the mitotic timer, and find that a stable G1 arrest following prolonged mitosis requires both p53 and p21, but is independent of DNA damage signaling. We show that Mdm2 expression is lost as cells enter mitosis, resulting in a gradual, time-dependent increase in p53 protein levels during mitosis. p53 accumulation continues in cells that progress into G1 following a prolonged mitosis, and these cells stably arrest and enter a state of senescence. 6 Results Prolonged mitosis leads to a G1 arrest To validate the existence of a mitotic timer (49), we made use of asynchronously growing RPE-1 cells expressing the Fucci system (37) (Figure 1A). In order to delay mitotic progression, cells were transiently treated with the microtubule-destabilizing drug nocodazole, which results in a SAC-dependent prometaphase arrest (41). Transient exposure of RPE-Fucci cells to nocodazole resulted in mitotic du- Time-dependent activation of p53 in mitosis controls the mitotic timer checkpoint 99

100 rations ranging from 1 h to 5 h (Figure 1B). In line with previous observations (49), we noticed that the duration of mother cell mitosis dictates the proliferative capacity of the respective daughter cells; daughters originating from mothers that spend less than ~1,5 h in mitosis enter a subsequent S phase within 24 h (Figure 1C, upper panel, and 1B, green bars), whereas daughters from mothers that spend over ~1,5 h arrest in G1 for at least 24 h (Figure 1C, lower panel, and 1B, red bars). We occasionally observed cell divisions that generate dissimilar daughter cells of which one arrests in G1, while the other enters S phase (Figure 1B, orange bars). These results confirm the existence of a mitotic timer, which restricts the proliferative capacity of daughter cells when the preceding mitosis is completed in an untimely manner. G1 daughter cell arrest in response to prolonged mitosis is p53- and p21-dependent Next, we wanted to verify whether the arrest caused by prolonged mitosis is p53- and p21-dependent. Therefore, we transfected RPE-Fucci cells with sirnas targeting luciferase, p53 or p21, and monitored cell cycle progression following transient nocodazole treatment. Concurrently, to confirm the efficiency of p53 and p21 depletion, we treated cells with the DNA damaging agent etoposide, which results in a p53- and p21-dependent G1 arrest (224, 299, 3). The majority of control cells arrested in G1 in response to etoposide, whereas cells transfected with sirnas targeting p53 or p21 continued to proliferate (Figure S1), indicating efficient depletion of p53 or p21 in these cells. Analysis of the proliferative capacity of daughter cells showed that p53- or p21-depleted daughter cells proliferated regardless of the duration of the preceding mitosis (Figures 2B and 2C, respectively), confirming that prolonged mitosis causes a p53- and p21-dependent G1 arrest (49). Collectively, these results confirm the existence of a time-sensing mechanism in mitosis that can restrict the proliferative capacity of cells following a problematic mitosis. p38 is not required for the mitotic timer The stress-responsive kinase p38 is active in mitosis, may stimulate p53 activity in response to DNA damage and is required for the maintenance of a DNA damage-induced G1 arrest (98, 175, 412, 413). In addition, p38 was shown to regulate the G1 arrest following a prolonged mitosis, such that inhibition of p38 in G1 allowed proliferation of daughter cells regardless of the duration of the preceding mitosis A C M G2 G1 Noco (4 h) : S 5:45 6:15 B Duration of mother cell mitosis (h) Both daughters proliferate Dissimilar daughter cells Both daughters arrest 2 Noco (4 h) : 4: 6:15 8: 12:15 14:3 19: 8: 11:3 15:15 19:45 1:3 15:3 22: 28: 1:3 15:3 22: 28: 7: 8: 1 Daughter cell pairs (n=1) Figure 1. Cells arrest in G1 following delayed completion of mitosis (A) Schematic overview of the Fucci system for visualization of the cell cycle. (B and C) Asynchronously growing RPE-Fucci cells were treated with nocodazole for 4 h and subsequently released into fresh medium. Images were taken every 15 minutes, during nocodazole treatment and 24 h thereafter. Quantification of proliferation of daughter cells pairs plotted against mother cell mitotic duration is shown in (B). Green bars indicate daughter cells that both proliferate, red bars indicate cell cycle arrest in both daughter cells and orange bars indicate dissimilar daughter cell behavior. Representative images of cells transiently treated with nocodazole are shown in (C). Upper panels show a cell that spends 3 minutes in mitosis; lower panels show a cell that spends 2 hours and 15 minutes in mitosis. Arrows indicate both daughter cells that were followed after division. Pooled data of n cells from 3 independent experiments. 1 Chapter 6

101 (49). In order to verify this finding we treated cells with a selective p38 inhibitor (SB2358 (414)) following transient a nocodazole treatment, and monitored the proliferation of the respective daughter cells. Strikingly, we observed that p38 inhibition could not overcome the G1 arrest that followed a prolonged mitosis (Figures 2D and 2E), suggesting that the G1 arrest induced by a prolonged mitosis functions independent of p38 signaling. ATM signaling is not required for G1 daughter cell arrest Mitotic spindle dysfunction, caspase activation or telomere uncapping during prolonged mitosis can give rise to DNA damage ( ), which can result in the activation of ATM, limiting the proliferative capacity of daughter cells in G1 (93, 416, 417). To address this possibility, we investigated if the G1 arrest that follows a prolonged mitosis is due to activation of ATM. To this end, we treated RPE-Fucci cells with the ATM inhibitor KU55933 (418) following a transient nocodazole treatment. ATM inhibition in G1 did not prevent G1 cells from arresting following a prolonged mitosis (Figure 2F). Collectively, our results suggest that a prolonged mitosis results in a p38- and ATM-independent G1 arrest. p53 accumulates in mitosis in a time-dependent manner A Duration of mother cell mitosis (h) Luciferase sirna Both daughters proliferate Dissimilar daughter cells Both daughters arrest Daughter cell pairs (n=1) D Duration of mother cell mitosis (h) DMSO Both daughters proliferate Dissimilar daughter cells Both daughters arrest Daughter cell pairs (n=1) B C Duration of mother cell mitosis (h) Duration of mother cell mitosis (h) p53 sirna Both daughters proliferate Dissimilar daughter cells Both daughters arrest Daughter cell pairs (n=1) p21 sirna Both daughters proliferate Dissimilar daughter cells Both daughters arrest Daughter cell pairs (n=1) E Duration of mother cell mitosis (h) F Duration of mother cell mitosis (h) p38i (SB 2358) Both daughters proliferate Dissimilar daughter cells Both daughters arrest Daughter cell pairs (n=1) ATMi (KU ) Both daughters proliferate Dissimilar daughter cells Both daughters arrest Daughter cell pairs (n=1) 6 Figure 2. G1 arrest after prolonged mitosis is p53- and p21-dependent, but p38- and ATM-independent (A C) Quantification of daughter cell proliferation as in Figure 1B of cells that were transfected with the indicated sirnas 24 hours prior to imaging. (D F) Daughter cell proliferation quantified as described before, except the indicated inhibitors were added following removal of nocodazole. Pooled data of n cells from 3 independent experiments. Time-dependent activation of p53 in mitosis controls the mitotic timer checkpoint 11

102 6 In order to investigate the molecular basis of the mitotic timer, and to confirm our results in wild-type RPE-1 cells, we setup a protocol that allows synchronous mitotic entry, as well as progression through and exit from mitosis (Figure 3A). Inhibition of Cdk1 using the specific inhibitor RO-336 causes cells to arrest at the G2/M boundary (419). 1 h release from Cdk1-inhibition into nocodazole, followed by mitotic shake-off, resulted in a population of cells highly enriched for mitotic cells (Figure 3B and S2A). Subsequently, half of the population was released from the nocodazole-induced mitotic arrest, while the other half was held in mitosis for an additional 3 h before being released (Figure 3A). Following the release, cells were incubated in BrdU-containing medium, to label proliferating cells as they enter S phase. After 2 h, cells were fixed and stained for phosphorylated histone H3 and BrdU (to label mitotic and S phase cells, respectively) (Figure 3A, S2A). Release from a nocodazole-induced mitotic arrest after 1 h allowed ~4% of daughter cells to enter a subsequent S phase (Figure 3C), indicating that this fraction managed to complete mitosis within ~9 minutes. The other ~6% failed to enter S phase within 2 h following release from 1 h nocodazole (Figure 3C), indicating that these cells may have completed mitosis in >9 minutes. Alternatively, stress induced by the mitotic shake-off procedure might have caused these cells to become arrested. However, while ~4% of cells that were released from a 1 h mitotic arrest entered S phase, nearly all cells released from a 4 h mitotic arrest failed to enter a subsequent S phase (Figure 3C). Cells that were blocked in nocodazole for up to 4 h did complete mitosis as indicated by the 2N DNA content of daughter cells. These results show that our protocol for synchronous progression and release from a mitotic arrest results in a G1 arrest depending on mitotic duration. To corroborate our finding that p38 is not required for the G1 arrest triggered by the prolonged duration of mitosis, we continuously treated cells with SB2358 from the moment cells progressed into mitosis, and during the subsequent G1. p38 inhibition failed to prevent the G1 arrest that follows a prolonged mitosis (Figure S2B). However, p38 inhibition slightly enhanced the proliferation of cells released from both a 1 h and 4 h nocodazole arrest, which likely resulted from a higher fraction of interphase cells present of more interphase cells in the p38-inhibited mitotic shake-offs (Figure S2C). We therefore conclude that the mitotic timer acts independent of p38. We next wanted to investigate the activation of p53 by the mitotic timer in more detail. As the G1 arrest induced by a prolonged mitosis is p53-dependent, and microtubule poisons have been shown to cause p53 stabilization and activation (397, 48, ), we first monitored p53 expression as cells were spending increasing amounts of time in mitosis. Using the same protocol as described above, we synchronized cells in G2, released and blocked them in mitosis for up to 1 h. Following mitotic shakeoff, cells were either collected immediately, or after an additional 3 or 6 h in mitosis. Interestingly, we observed a marked increase in p53 protein levels as cells spend more time in mitosis (Figure 3D). To investigate whether the increase of p53 was caused by exposure to nocodazole, or due to the prolonged time in mitosis, we repeated the assay using Taxol, STLC, or protame. These drugs prevent progression through mitosis, albeit through very different modes of action. Taxol stabulizes microtubules (423), thereby preventing the proper generation of tension by kinetochore microtubiles and arresting cells in mitosis in a SAC-dependent manner in prometaphase (41). STLC is an inhibitor of the microtubule motor protein Eg5 (424), which is required for the formation of a bipolar spindle (425, 426). Therefore, STLC-dependent Eg5 inhibition results in a SAC-dependent arrest in prometaphase. Finally, TAME, the functional metabolite of the prodrug protame, inhibits APC/C-dependent ubiquitination of substrates such as Cyclin B1, thereby preventing progression in to anaphase (427, 428). Unlike the microtubule poisons nocodazole and taxol, or the Eg5 inhibitor STLC, TAME does not prevent the formation of stable kinetochore-microtubule interactions or a bipolar spindle. Therefore, cells treated with protame arrest at metaphase (427). We observed that arresting RPE-1 cells in mitosis using either of these mitotic drugs results in an increase of p53 during mitosis (Figure S2D-S2G), indicating that the time-dependent increase in p53 levels in mitotic cells is a result of the time spent in mitosis, rather than the exposure to mitotic drugs. To address whether the p53 accumulation in response to these drugs is indeed restricted to mitotic cells, we treated asynchronously growing RPE-1 cells with nocodazole or STLC for 7 h, separated mitotic from interphase cells and analyzed p53 levels in drug-treated interphase and mitotic cells 12 Chapter 6

103 A Treatment: Cell cycle phase: RO 336 (18h) G2 Shake-off Noco (3h) Noco (1h) release into BrdU or EdU (2 h) M G1 S(?) release into BrdU or EdU (2 h) G1 S(?) B Mitotic cells (% ph3+) Asynchornous Shake-off C S-phase entry (% BrdU+) h 4h Noco D p53 α-tub G2 (RO 336) Nocodazole 1 h 4 h 7 h E p53 CDK4 Noco Noco 1 h 4h 1 h 4h G1 Daughters Figure 3. p53 accumulates during, and in G1 following a prolonged mitosis (A) Schematic overview of the experimental setup used to obtain cells synchronously entering into, and exiting from mitosis. In brief: cells are arrested at the G2/M border using the Cdk1-inhibitor RO 336. Cells are released from Cdk1-inhibition in the presence of nocodazole (to block progression through mitosis). After 1 hour, mitotic cells are collected through shake-off. Thereafter, cells are released immediately or 3 h later into medium containing BrdU or EdU, to label daughter cell S phase entry. (B and C) FACS analysis of asynchronous RPE-1 cells or RPE-1 cells treated as in (A) stained for DNA, phosphorylated histone 3 (H3 ps1), a marker for mitosis and BrdU, to label cells that have entered S phase. Means ± SD of 3 independent experiments. (D) Western blot showing mitotic increase of p53 expression. Cells were treated as in (A), except following shake-off mitotic cells were either lysed immediately, or following an additional 3 or 6 h in mitosis. A representative blot of 3 experiments is shown. (E) p53 accumulation continues in G1 following a prolonged mitosis. Cells were blocked in mitosis for 1 or 4 h (lane 1 and 2) and subsequently allowed to progress into G1 for 6 h (lane 3 and 4). compared to untreated cells. Both nocodazole- and STLC-treated mitotic cells displayed increased p53 levels, whereas their interphase counterparts did not (Figure S2H), showing that the p53 accumulation observed during mitosis is a consequence of mitotic duration, rather than exposure of cells to mitotic drugs. Finally, we investigated if p53 is stabilized in daughter cells originating from a short versus long mitosis. Therefore, we trapped cells in mitosis for 1 h or 4 h, as described before, and subsequently allowed them to progress into G1 for 6 h. As expected, we observed a time-dependent increase in p53 expression in mitosis (Figure 3E). Interestingly, p53 levels continued to increase during 6 h following release from a prolonged mitosis, whereas release from a short mitosis did not result in continued p53 accumulation (Figure 3E). Thus the enhanced expression of p53 that is initiated during the mitotic delay is somehow maintained in the following G1. This is indicative of potential positive feedback acting on p53 that is activated following a prolonged mitosis. We conclude that p53 accumulates both during mitosis and in G1 following a prolonged mitosis. 6 Mdm2 is rapidly lost in mitosis In normal conditions, p53 protein levels are kept low mainly due to Mdm2-mediated degradation (229, 41). Loss of Mdm2 or inhibition of the Mdm2-p53 interaction results in the accumulation and activation of p53 (429, 43). Also, chemical inhibition of RNA transcription results in loss of Mdm2 expression and p53 accumulation (431, 432). Since RNA transcription is known to be inhibited during mitosis (433), we hypothesized that Mdm2 might be progressively lost during mitosis. To test this, we monitored Mdm2 levels as cells synchronously progressed from G2 into mitosis. We observed that Mdm2 is lost within 1 h after mitotic entry, and the loss of Mdm2 is accompanied by an increase of p53 protein levels Time-dependent activation of p53 in mitosis controls the mitotic timer checkpoint 13

104 (Figure 4A). In addition, p21 levels did not increase during mitosis, despite accumulating p53 levels, confirming the mitotic inhibition of transcription. We confirmed the loss of Mdm2 in cells that were trapped in mitosis using three other mitotic drugs (Figure S3A), indicating that the loss of Mdm2 is due to mitotic entry as opposed to exposure to mitotic drugs. Finally, to investigate whether Mdm2 is synthesized during mitosis, we treated interphase or mitotic cells with the proteasome inhibitor MG132 for 3 h. Whereas interphase cells rapidly accumulate Mdm2 in response to MG132, mitotic cells fail to do so (Figure 4B), showing that Mdm2 synthesis is absent in mitosis. As the half-life of Mdm2 protein is approximately 2 minutes (434), the absence of Mdm2 synthesis may well explain the loss of Mdm2 within 1 h of mitosis. However, it is important to note that the half-life of MDM2 mrna exceeds 1 h ( ). This suggests that, in mitosis, Mdm2 expression is regulated by additional mechanisms that may affect translation efficiency or mrna stability, and thereby decrease Mdm2 protein synthesis in mitosis. 6 Loss of Mdm2 in mitosis likely causes mitotic p53 accumulation Since Mdm2 is absent in mitosis, we next wanted to investigate whether Mdm2-dependent p53 turnover is altered in mitosis. Therefore, we allowed cells to synchronously progress into mitosis, as described before. After 1 h, mitotic cells were collected and cells were maintained in mitosis for an additional 3 hours in the absence or presence of drugs inhibiting Mdm2 (Nutlin-3a), the proteasome (MG132), translation (cyclohexamide, CHX) or transcription (5,6-Dichloro-1-β-D-ribofuranosylbenzimidazole, DRB). Strikingly, Mdm2 or proteasome inhibition in mitotic cells did not enhance the time-dependent increase of p53 protein levels (Figure 4C). However, a 3 h treatment with either drug resulted in a marked increase in p53 protein levels in interphase cells (Figure 4D), showing that Mdm2-dependent turnover of p53 is lost in mitosis. This suggests that the continued synthesis of p53 during mitosis, when degradation of p53 is absent, causes the time-dependent increase in p53 expression. This notion is supported by the finding that translation inhibition (CHX) prevented the mitotic accumulation of p53, whereas inhibition of transcription (DRB) failed to do so (Figure 4C and 4D). Interestingly, p53-levels remained constant between 1 h mitosis and 4 h mitosis in the presence of CHX, whereas p53 is completely lost from interphase cells treated with CHX for 3 h (Figure 4C and 4D), which supports the loss of p53 degradation in mitotic cells. Collectively, these results show that, during mitosis, p53 is continuously translated but not degraded, resulting in p53 accumulation during mitosis. We observed that p53 levels continue to increase in daughter cells originating from a prolonged mitosis (Figure 3E), which prompted us to investigate whether Mdm2 expression in G1 would be different in daughter cells originating from a short or long mitosis. Release of cells that were blocked in mitosis for either 1 h or 4 h into G1 revealed that re-establishment of Mdm2 in G1 occurs with the same dynamics following a short or long mitosis (Figure 4E). These data suggest that loss of Mdm2 in mitosis results in a gradual increase of p53 during mitosis, and that this stabilization is somehow sustained in G1 following a prolonged mitosis. The finding that Mdm2 expression is not affected by the duration of the preceding mitosis, yet p53 continues to increase in G1 following a prolonged mitosis suggest that Mdm2 might somehow be kept inactive in G1 cells originating from a prolonged mitosis. Mitotic accumulation of p53 likely causes daughter cell G1 arrest Since Mdm2 expression is lost in mitosis (Figures 4A, 4B and S3A), we wanted to test whether forced expression of Mdm2 could prevent the G1 arrest caused by a prolonged mitosis. Therefore, we generated RPE-1 cells stably overexpressing Mdm2. We first tested whether these cells produced Mdm2 protein in mitotic cells. To this end, we collected interphase and mitotic cells, which were treated (or not) with MG132 for 3 h. We observed that overexpression of Mdm2 results in the presence of Mdm2 in mitotic cells (Figure S3B), albeit at lower levels compared to their interphase counterparts. In addition, inhibition of the proteasome resulted in the stabilization of Mdm2 both in interphase as well as mitotic cells (Figure S3B). This supports the notion that Mdm2 turnover, in combination with reduced protein synthesis in mitosis, may account for the loss of Mdm2 in mitosis. To test whether overexpression of Mdm2 could prevent daughter cells from arresting in G1 after a long 14 Chapter 6

105 A Mdm2 p53 G2 (RO 336) STLC 1 h 4 h 7 h B Interph. Mitosis MG132 (3 h): Mdm2 (long exp.) Mdm2 (short exp.) C p53 Asynchr. G2 (RO 336) 1 hr noco DMSO Noco (4 h) MG132 Nutlin-3 CHX DRB CDK4 Cdk4 CDK4 p21 F 8 D p53 CDK4 DMSO MG132 Nutlin-3 CHX DRB E Mdm2 CDK4 Noco (1h) 2 h 4 h Release Mdm2 o.e.: - + Figure 4. Loss of Mdm2 mitosis potentially causes p53 accumulation and subsequent daughter cell G1 arrest (A) Mdm2 expression is lost in mitosis. Cells were treated as in Figure 3D. A representative blot of 3 experiments is shown. (B) Synthesis, but not degradation, of p53 is continued during mitosis. Cells were treated as described before, except the indicated inhibitors were added to cells during the last 3 h of the 4 h mitotic arrest. A representative blot of 2 experiments is shown. (C) Western blot revealing continuous synthesis and degradation of p53 in interphase cells. Asynchronously growing RPE-1 cells were treated with the indicated inhibitors for 3 h. A representative blot of 3 experiments is shown. (D) Mdm2 expression is resumed at the same time in G1 after a short or long mitosis. RPE-1 cells were blocked in mitosis for 1 or 4 h as described before, and subsequently allowed to progress into G1 for the indicated amount if time. A representative blot of 3 experiments is shown. (E) Mdm2 protein synthesis is absent in mitosis. Interphase cells or nocodazole treated, mitotic cells were treated with the proteasome inhibitor MG132 for 3 h. (F) Mdm2 overexpression prevents G1 daughter cell arrest following prolonged mitosis. Wild type RPE-1 cells, or RPE-1 cells infected with a construct expressing Mdm2 were treated as in Figure 3A, stained for DNA and EdU, and the fraction of EdU positive cells was counted. Means ± SD of 3 independent experiments. Noco (4h) Release 6 h 8 h 2 h 4 h 6 h 8 h EdU Incorporatio (% of cells) A B Colony formation C ß-Gal D 16 Asynchronous 1 Mitotic shake-off 1 hr nocodazole hr nocodazole 6 Cell number Days post release 1 hr nocodazole 4 hr nocodazoe 6 days post release Figure 5. Prolonged mitosis results in stable cell cycle exit of daughter cells (A and B) Daughter cells originating from a prolonged mitosis fail to grow for up to 6 days. Cells were blocked in mitosis for 1 or 4 h, as described before (Figure 3A), released and counted every other day for 6 days. Simultaneously, cells collected through drug-free mitotic shake-off and asynchronous cells were plated and counted (A). In addition, a colony formation assay was performed with 5 cells released from 1 or 4 h mitosis block (B). Means ± SD of 3 independent experiments. A representative colony formation is shown. (C and D) Prolonged mitosis leads to the induction of senescence in daughter cells. Cells released from 1 or h 4 mitosis block were seeded and fixed and stained for senescence-associated β-galacotisdase (β-gal) activity 6 days later (C). A quantification of the amount of β-gal-positive cells is shown in (D). A representative image of a β-gal assay is shown. Means ± SD of 3 independent experiments. 1 hr nocodazole 4 hr nocodazoe 6 days post release % of cells hr noco 4 hr noco SA-ß-Gal - SA-ß-Gal + 6 Time-dependent activation of p53 in mitosis controls the mitotic timer checkpoint 15

106 mitosis, we subjected wild-type or Mdm2-overexpressing RPE-1 cells or cells to a 4 h nocodazole arrest and subsequently released cells for 2 h in medium containing EdU, to label S phase entry. As expected, the majority of control cells failed to proliferate following 4 h nocodazole treatment. In contrast, overexpression of Mdm2 caused the majority of cells to enter the subsequent S phase, despite prolonged mitotic duration (Figure 4F). It is important to note however that Mdm2 overexpression can also prevent a DNA damage-induced G1 arrest (Figure S4B), indicating that Mdm2 overexpression generally abolishes the G1 checkpoint. Therefore, additional experiments are required to definitively resolve whether sustained expression of Mdm2 in mitosis overrides the G1 arrest following a prolonged mitosis. Prolonged mitotic duration results in the induction of senescence In order to investigate the long-term fate of daughter cells originating from a prolonged mitosis, we followed cells released from a short or long mitotic arrest for several days. Population doublings of cells treated with nocodazole for 1 h were indistinguishable from asynchronous cells or mitotic cells that were collected by shake-off using cells that were not treated with any drugs (Figure 5A). In stark contrast, cells released from a 4 h nocodazole arrest hardly divided during a 6-day period (Figure 5A), which was confirmed in colony formation assays (Figure 5B). In addition, β-gal-staining revealed that the majority of cells released from a 4 h nocodazole treatment entered into senescence, whereas cells released from a 1 h nocodazole treatment do not (Figure 5C and 5D). These results show that the G1 arrest triggered by prolonged mitotic duration eventually causes cells to undergo senescence, thereby permanently preventing daughter cells originating from a potentially problematic mitosis from proliferating. Discussion 6 Progression from one cell cycle phase into the next is carefully regulated. Next to extensive regulatory mechanisms that act within one cell cycle phase, recent studies have identified mechanisms acting in cells that are carried over to their progeny. For instance, it was shown that the pre-mitotic abundance of p21 greatly influences the proliferation/quiescence decision (21), such that p21-dependent inhibition of Cdk2 during mitotic exit causes cells to enter into quiescence. In addition, it was found that deregulation of PP2A and the resulting increase in Cyclin E transcription and Cdk2 activity in G2 speeds up progression through the subsequent G1 (211). Here, we have identified the molecular mechanism responsible for the mitotic timer that can arrest daughter cells originating from a prolonged mitosis. With this we have uncovered yet another mechanism of cell cycle regulation that is carried over from mother to daughter cells. The completion of mitosis typically occurs within one hour, may take up to three hours (399). This duration is heavily influenced by mitotic spindle dysfunction and impaired DNA decatenation, both causes for decreased chromosome segregation fidelity ( ). The mitotic timer may operate as a back-up mechanism to prevent the proliferation of cells that have originated from such a problematic mitosis. In agreement with this hypothesis is the finding that loss of centrioles, a known cause for chromosomal instability (443), increases mitotic duration and results in p53-dependent apoptosis in mouse embryos (444). Unstressed G1 cells typically exhibit a transcriptionally attenuated p53 pulse (254), suggesting that cells set the stage for rapid activation of p53-dependent responses upon stress. Our results imply that loss of Mdm2 results in time-dependent accumulation of p53 during mitosis (Figure 3D and S2D-S2H), which is continued in G1 following a prolonged mitosis (Figure 3E), despite the re-accumulation of Mdm2 (Figure 4E). This suggests that Mdm2 is inhibited in G1 cells originating from a prolonged mitosis. There are several positive feedback-loops that may account for sustained p53 activation. First, ARF is a potent inhibitor of Mdm2, and expression of ARF can result in a p53-dependent G1 arrest (445, 446). ARF is stabilized in response to various cellular stresses (Reviewed in 328) and it will be interesting to see 16 Chapter 6

107 whether prolonged mitosis may result in the accumulation of ARF in the subsequent G1. Second, the Lats2 tumor suppressor has been shown to stabilize p53 following mitotic slippage, by inhibiting Mdm2. In turn, p53 can selectively up-regulate Lats2 expression (447), thereby re-enforcing its own stabilization. It remains to be tested if either of these feedback mechanisms is involved in arresting cells in G1 following a prolonged mitosis. Mdm2 is down-regulated in response to transcription inhibition, and transcription itself is inhibited during mitosis (431, 433). These findings led to the hypothesis that mitotic inhibition of transcription results in the absence of Mdm2 in mitosis, allowing p53 accumulation as a sensor for mitotic timing (448, 449). In line with this hypothesis, we show that p53 accumulates throughout mitosis as a consequence of continuous translation and impaired proteolysis of p53 in mitosis (Figure 4C). Consequently, once a cell has spent a certain time in mitosis, p53 levels reach a level that is high enough to prevent the daughter cells from dividing again. The mechanism(s) contributing to loss of Mdm2 in mitotic remain to be elucidated. As Mdm2 is an unstable protein (434), maintenance of Mdm2 requires continuous transcription and translation, both of which seem to be inhibited in mitotic cells (Figure 4B). Lack of Mdm2 synthesis in mitosis may be due to low mrna stability, so that inhibition of transcription can rapidly deplete mitotic cells from Mdm2 mrna and thus prevent protein synthesis. However, the overall halflife of Mdm2 mrna exceeds one hour ( ). This indicates that Mdm2 mrna might be destabilized in mitosis, but it is also possible that Mdm2 proteostasis in mitosis is not merely regulated through the inhibition of transcription. In fact, increased protein degradation or inhibition of translation may also contribute to the loss of Mdm2 in mitotic cells. In support of this notion is the finding that SCF βtrcp1 -dependent ubiquitination controls the stability of Mdm2 during the cell cycle (45). Therefore, it will be interesting to investigate if SCF βtrcp1 -dependent regulation of Mdm2 contributes to p53 accumulation during mitosis. In addition to proteolysis of Mdm2, we observed continued synthesis of exogenous Mdm2 (Figure S3B), but not endogenous Mdm2 in mitosis (Figure 4B). These results suggest that an additional layer of regulation at the level of translation may exist for Mdm2 in mitosis. This, however, needs further validation. 6 Time-dependent activation of p53 in mitosis controls the mitotic timer checkpoint 17

108 Materials and methods Cell lines, cell culture and cell clines and transfections and infections RPE-1 cells were from ATCC, and RPE-Fucci cells have been described before (175). RPE-1 cells overexpressing Mdm2 were generated by lentiviral infection with phr-mcherry-p2a-strepii-mdm2. For virus production, Hek293 cells were transfected with packaging plasmids and the Mdm2-encoding plasmid using X-tremeGENE (Roche) according to manufacurers protocol. Two days later the virus containing medium was applied to RPE-1 cells for 24 h. ON-TARGETplus SMARTpool sirnas targeting Luciferase (D ), GAPDH (LU ), p53 (LU ), p21 (LU ) and Mdm2 (LU ) were purchased from Dharmacon and transfected using RNAiMAX (Life Technologies) using the manufacturers protocol. All sirna transfections were performed 24 h before the start of the experiment. Antibodies and chemicals Antibodies used in this study are: anti-p53 (DO-I), anti-mdm2 (smp14), anti-p21 (C-19), anti-cdk4 (C- 22) from Santa Cruz Biotechnologies, anti-α-tubulin (t5168) from Sigma, anti-brdu (BU1/75 ICR1) from Abcam, anti-histone 3 phospho serine 1 (6-57) from Upstate and anti-actin (I-19) were from Tebu-bio. Chemicals used in this study are: EdU and fluorescent azide (Alexa Fluor 488) purchased from Life Technologies. BrdU, KU55933, SB2358, nocodazole, taxol, STLC, MG132, cyclohexamide and DRB were purchased from Sigma-Aldrich. RO336 was purchased from Calbiochem, protame from Boston Biochem and Nutlin3a from Cayman Chemical. Live cell and fluorescence imaging For live cell imaging, RPE-Fucci cells were plated in Lab-Tek II chambered cover glass (Thermo Scientific) in DMEM/F12, which was replaced by Leibovitz s L15 (Gibco) CO 2 -independent medium directly before imaging. Images were acquired using a DeltaVision Elite (Applied Precision), maintained at 37 C equipped with a 1x.4 NA or 2x.75 NA lens (Olympus) and cooled CoolSnap CCD Camera. Image analysis wad performed using ImageJ software. EdU labeling was performed as described before (451). FACS analysis was performed as described previously (226). Western blot analysis was performed as described before (216). 6 Growth curve, colony formation and β-galactosidase assay For growth curves, 2 cells were plated per well. In order to count, cells were collected through tripsinization, and counted using a TC2 automated cell counter (Biorad). For colony formations, 5 cells were plated and allowed to grow for 6 days. Thereafter, cells were fixed using ice-cold methanol and subsequently stained with.1% crystal violet followed by extensive washing. β-gal stainings were performed as described elsewhere (327). Acknowledgements We would like to thank members of the Medema lab, as well as members of the S.M.A. Lens lab and G.J.P.L. Kops lab for fruitful discussions. We would like to thank M.E. Tanenbaum for sharing reagents. This work was supported by KWF grant to R.M. and NWO grant to L.K. 18 Chapter 6

109 s-phase entry (% of G1 cells) Etoposide: sirna: Luc p53 p21 Figure S1. Time-lapse microscopy of undamaged and DNA-damaged RPE-Fucci cells Cells were transfected with sirnas targeting lucicerase, p53 or p21, and imaged 24 h later. S phase entry was monitored for cells treated with solvent or the DNA-damaging agent etoposide (5 μm) for the duration of the movie. Bars represent average ± SD, of 3 independent experiments. At least 5 cells were counted per condition per experiment. Figure S2. Prolonged mitosis causes G1 daughter cell arrest and accumulation of p53 (A) Staining of RPE-1 cells for DNA content (propidium iodide), mitotic cells (phosphorylated histone 3, serine 1) and S phase entry (BrdU), of cells that were treated as in Figure 3A. Representative images are shown. (B) p38-inhibition does not prevent G1 daughter cell arrest caused by prolonged mother cell mitosis. Cells were treated as in Figure 3A, except where indicated the p38-inhibitor SB2358 (1 μm) was added to the cells during mitotic arrest and thereafter. Means ± SD of 3 independent experiments. (C) Quantification of the amount of mitotic cells following shake-off in the cells used in (B). Means ± SD of 3 independent experiments. (D - G) p53 accumulates during mitotic arrest. Cells were treated as in Figure 3D, except progression though mitosis was inhibited using the indicated drugs. A representative blot is shown, n=3 (H) p53 accumulation during mitotic arrest is caused by mitosis, rather than mitotic drugs. Asynchronously growing RPE-1 cells were treated with the indicated drug for 7 h. Thereafter, interphase and mitotic cells were separated using shake-off and p53-levels were visualized using western blot. A representative blot is shown, n=2 A BrdU phospho-h3 Counts BrdU (2 h) Shake-off 1 h noco 4 h noco Propidium Iodide B S-phase entry (% EdU+) SB2358: D Asynchronous G2 (RO 336) 1h 4h Nocodazole Noco 1 h 4 h 7 h p53 E C SB2358: - + Mitotic cells (% PH3+) Asynchronous G2 (RO 336) Shake-off protame 1 h 4 h 7 h p53 6 Actin Actin F G H Asynchronous G2 (RO 336) STLC 1 h 4 h 7 h p53 Asynchronous G2 (RO 336) 1 h Taxol 4 h 7 h p53 Asynchronous Noco Adherent Shake-off STLC Adherent Shake-off p53 Actin Actin CDK4 Time-dependent activation of p53 in mitosis controls the mitotic timer checkpoint 19

110 A Mdm2 CDK4 Asynchr. G2 (RO336) Noco STLC 1 h Taxol protame B Interph. Mitosis MG132 (3 h): Mdm2 (long exp.) Mdm2 (short exp.) Cdk4 C EdU Incorporation (% of cells) 5uM Etoposide Mdm2 o.e.: - + Figure S3. Endogenous Mdm2 is lost in mitosis and overexpression of Mdm2 overrides the G1 checkpoint (A) Mdm2 expression is lost in cells blocked in mitosis using various drugs. RPE-1 cells were treated as in Figure 4A, except mitotic progression was blocked using the indicated drugs. A representative blot is shown, n=3 (B) Exogenous Mdm2 is synthesized in mitosis. RPE-1 cells infected with a construct expressing StrepII-Mdm2 were treated as in Figure 4E. A representative blot is shown, n=2 (C) Mdm2 overexpression prevents DNA damage-induced G1 arrest. Wild type RPE-1, or RPE-1 cells overexpressing Mdm2 were grown in the presence of EdU and etoposide (5 μm) for 2 h. After fixation, cells were stained for DNA and EdU. S phase entry was determined as the percentage of EdU-positive cells. Means ± SD of 3 independent experiments, at least 5 cells were counted per condition per experiment 6 11 Chapter 6

111 Time-dependent activation of p53 in mitosis controls the mitotic timer checkpoint 111 6

112

113 Chapter 7 Thesis summary and general discussion Lenno Krenning and René H. Medema Division of Cell Biology I, The Netherlands Cancer Institute Plesmanlaan 121, 166 CX Amsterdam, The Netherlands 7 General discussion 113

114 Thesis summary DNA mutations and aneuploidy are thought to drive tumorigenesis. Therefore, in order to limit cellular transformation, proliferating cells need to evenly distribute their genomes during mitosis and properly repair their DNA in case it is damaged. The most cytotoxic kind of DNA damage is the double strand break (DSB), in which the two complementary stands of the DNA double helix are broken simultaneously. This sets in motion a complex signaling response culminating in cell cycle arrest and DNA repair. Following DNA repair, cells may resume progression through the cell cycle. Alternatively, in the face of excessive or irreparable DNA damage, cells may permanently withdraw from the proliferative cycle. Work in this thesis has demonstrated that the reversibility of such a DSB-induced cell cycle arrest varies as a function of cell cycle position. In chapter 3, we used live-cell microscopy to study cell cycle progression in response to DNA damage. We discovered that reversibility of a G1 arrest is maintained for several days, whereas a G2 arrest loses its reversibility within several hours (Chapter 3). This prompted us to investigate the mechanisms leading up to an irreversible cell cycle arrest in G2. We found that p53-dependent induction of p21 results in the nuclear sequestration of inactive Cyclin B1-Cdk1 complexes. These p21-bound complexes are refractory to activation by Cdc25-phosphatases, which is thought to cause irreversibility of the G2 arrest. The arrested cells subsequently lose their G2-identity after several hours, due to activation of the anaphase promoting complex/cyclosome (APC/C), in complex with its coactivator Cdh1, resulting in the degradation of many (if not all) pro-mitotic proteins. We find that the likelihood of undergoing cell cycle exit increases as a function of DNA damage load. However, in the addendum of chapter 3, we show that DNA damage load does not per se determine whether a cell will cease proliferation or not, and that the speed or extent of DNA repair also contributes to this decision. In chapter 4 we show that prophase cells are more sensitive to DNA damage compared to G2 cells. In response to DNA damage, prophase cells directly activate the APC/C Cdh1, which occurs as a result of the scheduled degradation of Emi1 in prophase. We show that this prophase cell cycle exit mechanism operates to prevent cell division of cells that have residual DNA lesions that arise from DSBs or replication stress. In chapter 5 we investigated a potential role for the mitotic checkpoint complex (MCC) in regulation of the APC/C Cdh1 after DNA damage. We identified an essential role for the MCC proteins Mad2 and BubR1 in maintaining reversibility of the DNA damage-induced G2 arrest, and find that Mad2 regulates Emi1 expression. Finally, in chapter 6, we elucidate the mechanism of a mitotic timer that likely functions to prevent further cell divisions of cells that have originated from a problematic mitosis. We find that time-dependent accumulation of p53 in mitosis, likely caused by the absence of Mdm2 in mitosis, limits the replicative capacity of the respective daughter cells Chapter 7

115 General discussion 1. Control of G2 checkpoint reversibility Proliferating cells encounter numerous DNA lesions every day. Thus, in order to maintain tissue homeostasis, a DNA damage-induced cell cycle arrest is only useful as long as it can be reversed when appropriate. Therefore, cells have to carefully regulate recovery competence, which is achieved through various mechanisms. 1.1 p21-dependent inhibition of Cyclin B1-Cdk1 complexes In chapter 3 we have shown that the p53-dependent induction of p21 results in activation of the anaphase promoting complex/cyclosome (APC/C), in complex with its coactivator Cdh1, and the onset of senescence. Importantly, we show that p21-dependent inhibition of Cdk1 irreversibly prevents checkpoint recovery, independently of subsequent APC/C Cdh1 activation. We find that nuclear, p21- bound Cyclin B1-Cdk1 complexes are refractory to activation upon addition of a Wee1 inhibitor. This may be caused by the inability of Cdc25 phosphatases to remove inhibitory phosphorylations of Cdk1 within this complex. Classically, p21 was thought to be a relatively poor inhibitor of Cyclin B1-Cdk1 (82, 84). However, these conclusions were based on studies using purified proteins, and therefore do not incorporate the effect that p21 may have additional Cdk-regulatory mechanisms. Indeed, it was shown that p21-bound, nuclear Cyclin B1-Cdk1 complexes lack in vitro kinase activity (24), which could not be stimulated by Cdc25 phosphatases. In addition, using antibodies to interrogate the phosphorylation status of Cdk1, we have now shown that these nuclear, p21-bound Cyclin B1-Cdk1 complexes fail to be dephosphorylated on Y15 upon inhibition of Wee1 (Chapter 3). Collectively, these data suggest that p21 may be a poor inhibitor of active Cyclin B1-Cdk1 complexes, but instead prevents the Cdc25-dependent activation of Cdk1. As the majority of Cyclin B1-Cdk1 complexes in G2 are phosphorylated on Y15, and thus inactive, p21-expression will potently prevent Cdk1-activity and thus block progression beyond G2 (Chapter 3) (411). This is further supported by the finding that induced expression of a constitutively active Cyclin B1-Cdk1 fusion protein, which cannot be phosphorylated on T14 and Y15, promotes mitotic entry of cells arrested in G2 in a p21-dependent manner (237). 1.2 Premature activation of the APC/C Cdh1 causes a loss of recovery competence Inhibition of the APC/C Cdh1 from S phase until anaphase is required to couple DNA replication to chromosome segregation (34, 38, 39). This is achieved through the combined action of Emi1 and Cdk-activity. Emi1-binding to the APC/C Cdh1 inhibits its activity (47, 48). In addition, Cdk-dependent phosphorylation of Cdh1 prevents its interaction with the APC/C (22, 24, 25, 34, 35). In an unperturbed cell cycle, Emi1 is degraded in prophase (51, 53), during which a cell critically depends on Cdk-dependent inhibition of the APC/C Cdh1 (Chapter 4). Failure to do so, through DNA-damage induced inhibition of Cdk-activity, results in cell cycle withdrawal (Chapter 4). In addition, the presence of Emi1 allows cells to reversibly arrest cell cycle progression in response to DNA damage in (Chapter 4) (235, 236). Interestingly, Emi1 turnover in interphase is regulated by Cdk-activity (51, 389). At the same time, Cdk-dependent phosphorylation of Cdh1 directly prevents APC/C Cdh1 activity (22, 24, 25, 34, 35). Therefore, one could imagine that the DNA damage-induced inhibition of Cdk activity removes one mode of APC/C Cdh1 inhibition, which is counteracted by stabilization of Emi1 as a result of Cdk inhibition. This way, a cell could maintain a G2 arrest for as long as transcription can sustain sufficient expression of Emi1 (235, 236). Indeed, in p53-impaired cells, the maintenance of a reversible G2 arrest exceeds the half-life of Emi1 protein in conditions where Cdk-activity is uninhibited (389, 39). In addition to Emi1, in chapter 5 we show that the mitotic checkpoint complex (MCC)-proteins Mad2 and BubR1 are required to prevent premature activation of the APC/C Cdh1 and the maintenance of G2 checkpoint reversibility. This, for the first time, links the MCC to the regulation of Cdh1-dependent AP- C/C activity. Intriguingly, we find that, following DNA damage, Emi1 expression is altered in cells lacking Mad2. This suggests that Mad2 somehow influences the stability or expression of Emi1. Thus, it will be 7 General discussion 115

116 interesting to investigate whether Mad2, and possibly other MCC proteins, functionally interact with known regulators of Emi1, such as Pin1, Evi5, SCF βtrcp or Plk1- and Cdk-activity (51, 53, 355, 389, 39). With respect to Cdk-activity it is important to note that Mad2-depletion, in our hands, did not alter the expression of Cyclin B1 levels at the onset of mitosis (Chapter 5). This, however, does not mean that overall Cdk-activity is unaltered, which is something that remains to be determined. 2. Factors determining the outcome of a DNA damage-induced cell cycle arrest For decades, a lot of research has been directed to a better understanding of the effects of DNA damage on the proliferative cycle of cells. This has created a wealth of knowledge regarding the types of DNA damage that can occur, how this DNA damage is detected, and how the DNA damage signals to the cell cycle and DNA repair machineries (56). In addition, it has become increasingly clear that the cell cycle machinery itself is also directly involved in the regulation of DNA damage signaling and DNA repair (Reviewed in chapter 2). However, the effect of the cell cycle on the outcome of a DNA damage-induced cell cycle arrest remained largely unexplored. Most DNA damage-induced cell cycle exit occurs from G2 (Chapter 3) (237), and therefore we will now focus this discussion on factors that may influence cell fate decisions in G2 phase of the cell cycle. The reversibility of a DNA damage-induced G2 arrest is largely controlled by p53 (Reviewed in chapter 2). In p53-proficient cells, the capability of a cell to recover may be lost within several hours following DNA damage (Chapter 3) (175, 242). In contrast, p53-impared cells can retain reversibility of the arrest for much longer (174, 221, 242). Cells devoid of any p53-activity eventually fail to sustain a G2 arrest and merely display a delay in mitotic entry (85) Cell cycle position influences DNA damage checkpoint reversibility It is widely accepted that DSBs are the most cytotoxic lesions induced by IR (335, 452, 453). In response to DSBs, cells may arrest in the G1 or G2 phase of the cell cycle. Early work investigating differences in the reversibility of a cell cycle arrest induced by DNA damage in various stages of the cell cycle showed that Chinese hamster ovary cells were more likely to progress into mitosis following DNA damage in G1 compared to G2 phase of the cell cycle (454), indicative of potential differences in DNA damage sensitivity according to cell cycle position (455). The work presented in chapters 3 and 4 of this thesis has demonstrated that cell cycle position greatly influences the capability of cells to resume proliferation upon the induction of DNA double strand breaks (DSBs). In agreement with initial findings that G2 cells are more sensitive to DNA damage, in chapter 3 we show that a larger fraction of G1 cells compared to G2 cells are able to recover from a similar DNA damaging insult. In addition, it was found that cell cycle withdrawal in response to DNA damage preferentially occurs in cells with a 4N DNA content (237). The finding that DNA damaged cells first need to progress into G2 phase before being able to undergo cell cycle exit (242), and the finding that prophase cells are more sensitive to DNA damage compared to G2 cells (Chapter 4) further solidify the hypothesis that DNA damage sensitivity is dissimilar at various stages of the cell cycle. These differences may be, in part, caused by the fact that similar activation of p53 causes a fully reversible cell cycle arrest in G1 but induces senescence in G2 (Chapter 3) (175). In addition, the differential requirements for checkpoint inactivation at various stages of the cell cycle may contribute to differences in DNA damage sensitivity throughout the cell cycle (175). Finally, in chapter 4 we show that chronic replication stress preferentially eliminates cells from the cell cycle at prophase, again highlighting the notion that similar DDR signaling events lead to dissimilar outcomes depending on the position in the cell cycle. 2.2 DNA damage location may influence cell fate The initial activation of p53 upon DSBs mainly requires ATM-dependent signaling (331, 456), and cell cycle exit in response to physiological levels of DNA damage is largely regulated by ATM (Chapter 3, 116 Chapter 7

117 addendum). Therefore, it is tempting to speculate that factors that influence the extent of ATM activation upon DNA damage will contribute to cell fate decisions. In line with this notion is the finding that p53- and p21-induced cell cycle exit is dose dependent (Chapter 3), and that p53 target genes, including p21, are expressed in a dose- and ATM-dependent manner (457). In addition to dose-dependent activation of ATM, we also observed that cells that initially have similar amounts of DNA damage foci exhibited different cell fate outcomes (Chapter 3, addendum), indicating that factors other than DNA damage dose can influence cell fate decisions. The excessive spreading of H2AX phosphorylation, yielding γh2ax, is limited if a DSB is located in heterochromatin (34). However, γh2ax is required for MDC1 localization to sites of DNA damage (66), which acts as an amplification step in ATM signaling and is required for G2 checkpoint maintenance (65, 127). Thus, a similar amount of DNA damage may cause dissimilar activation of ATM signaling, depending on its localization. Indeed, forced chromatin relaxation results in increased spreading of γh2ax and ATM hyperactivation (341). In addition, depending on their genomic location, DSBs may be irreparable (257, 258). Such irreparable DSBs continue to activate ATM, which may drive cells into senescence (258, 259, 323, 342). Similar to IR-induced APC/C Cdh1 activation and cell cycle exit in G2 (Chapter 3) (237), persistent telomere damage has shown to result in APC/C Cdh1 activation in G2 cells and a bypass of mitosis (458). Thus, generation of a DSB at an irreparable locus may result in higher overall ATM activation, compared to the generation of a DSB at a repairable locus. In agreement with this hypothesis, we observe that cells undergoing cell cycle exit fail to complete DNA repair for at least 15 h (Chapter 3, addendum). 2.3 DNA repair pathway choice may influence checkpoint reversibility In addition to the location of a DSB being a determinant of cell fate outcomes, the DNA repair pathway usage may influence cell fate outcomes. DSBs induced by IR can be classified into simple and complex breaks, which are repaired with different kinetics (459, 46). The initial fast repair represents DNA damage that is repaired through non-homologous end joining (NHEJ), whereas homologous recombination (HR) typically occurs with slower kinetics (325, 326). DNA end-resection, which is required for homology-directed repair (336, 337), results in the activation of ATR sometime during the first hour following the induction of a DSB (73, 461). In p53-impaired cells, the DNA damage checkpoint is maintained through the activity of ATR and Chk1 (126, 462). In addition, preventing DNA resection leads to the premature inactivation of ATR- and Chk1, resulting in a failure to maintain the DNA damage checkpoint in p53-impaired cells (131, 461). This is to illustrate that at later times following DSB-induction, in G2 cells, DDR-signaling is predominantly sustained through ATR and Chk1. It is therefore not unthinkable that ATR-dependent signaling contributes to cell fate decisions at later time points following DNA damage by sustaining p53 activation (331, 456). However, the finding that ATR inhibition only slightly increases the fraction of cells that enters into mitosis following a high dose of IR (Chapter 3, addendum) suggests that the contribution of ATR signaling to cell fate decisions following DSBs in p53-proficient G2 cells is limited. 2.4 Basal cellular conditions may influence cell fate In addition to the factors described above, the physiological state of a cell at the moment it encounters a DSB may affect the outcome of the DNA damaging insult. For instance, although p21 expression is strongly induced by DNA damage, it is expressed at low levels in unperturbed cells (Chapter 3) (83). These basal p21 levels have been shown to govern the proliferation/quiescence decision made in G1 phase, through regulation of Cdk2-activity (21, 343). In G2, the induction of p21 beyond a permissive threshold seems to irreversibly arrest cell cycle progression (Chapter 3) (411). Therefore, a 5-fold increase in p21 expression after DNA damage may cause some cells to irreversibly arrest, depending on the basal expression of p21 at the moment they encountered DNA damage. Much the same, p53 oscillates during unperturbed cell cycles, but this p53 is transcriptionally attenuated (254). This, however, does imply that cells may have different basal levels of p53 at the moment they encounter DNA damage, which may affect the efficiency of p53 target gene induction. Indeed, early investigations into the p53 response have illustrated heterogeneous responses in both p53-stabilization and p53 target 7 General discussion 117

118 induction in clonal cell lines following a similar DNA damaging insult (232, 463). Clearly, our understanding of the determinants underlying cell fate decisions is far from being complete. However, the use of live cell microscopy offers promising opportunities for the investigation of such factors (discussed later). 3. Maintaining a G2 arrest In cells with impaired p53-function, activation of the APC/C Cdh1 is required for the maintenance of a DNA-damage induced G2 arrest (23, ). In contrast, in p53-proficient cells APC/C Cdh1 -activity is dispensable for a stable G2 arrest (Chapter 3) (237). p53-proficient cells devoid of Cdh1-dependent APC/C-activity eventually lose the expression of mitotic regulators (237), which is likely caused by the cessation of E2F-dependent transcription. Indeed, cell cycle withdrawal from G2 phase in response to DNA damage requires transcriptional inhibition of E2F target genes, which is mediated by the pocket proteins (237, 31). Once cellular senescence has been established, it is maintained through the combined action of p53 and p16 (237, 464, 465), and prb-dependent formation of heterochromatin that prevents the reactivation of E2F gene expression (466). Therefore, we propose that the p21-dependent inhibition of Cdk1 acts to prevent mitotic entry until the expression of E2F target genes is lost and cells commit to senescence. It is important to note that prolonged Cdk1-inhibition can result in APC/ C Cdh1 activation and Cdk2-dependent genome reduplication (467, 468). Therefore, cells that undergo G2 cell cycle exit need to inhibit Cdk2 in order to prevent genome reduplication. Besides inhibiting Cdk1, p21-dependent inhibition of Cdk2 can prevent genome reduplication (469). Importantly, p21 expression is maintained for several days following a DNA damaging insult (237). This might be achieved through the continued signaling from sites of persistent DNA damage (259, 323). As APC/C Cdh1 activation in G2 reverts the cell into a G1-like status, it will be interesting to investigate whether the prevention of genome reduplication following cell cycle exit in G2 requires continued activation of Chk2 and p38, as these kinases maintain a cell cycle arrest following DNA damage in G1 (175). Alternatively, positive feedback mechanisms may be activated following the p53-dependent initiation of cell cycle exit in G2, as transient stabilization of p53 by the Mdm2-inhibitor Nutlin3 in G2 cells may cause senescence (Chapter 3). Indeed, Nutlin 3 treatment was shown to result in sustained p21-expression in p53-proficient tumor cells arrested in G2 (47). 7 Finally, in chapters 4 and 5 we have shown that (partial) depletion of Emi1, Mad2 or BubR1 increases the fraction of cells that activate the APC/C Cdh1 following IR. If the continued signaling from sites of DNA damage is indeed required to prevent genome reduplication, one might speculate that these cells, that would have normally continued to proliferate, might instead undergo genome reduplication. Indeed, genome reduplication is observed in a minor fraction of cells following high levels of DNA damage (Chapter 3). With this in mind, it is important to note that the forced induction of genome reduplication results in DNA damage and senescence (4). Therefore, senescence induced by the genome reduplication of a cell that failed to terminally exit the cell cycle in G2, might act as a failsafe mechanism that prevents the proliferation of a cell that initially escaped the induction of senescence. 4. Cells may also exit the cell cycle in G1 Even though cells in G2 and prophase are more likely to undergo a cell cycle exit in response to DNA damage compared to cells in G1 (Chapters 3 and 4), cell cycle exit is not limited to G2 phase or prophase. In fact, early investigations reported that IR-induced DNA damage in G1 results in a prolonged p53-dependent cell cycle arrest in G1 (471). Unlike a cell cycle arrest in G2 phase, which initially is p53- and p21-independent, in G1 phase, even a reversible cell cycle arrest is critically dependent on p53 and p21 (85, 94-96, 175, 455). Consequently, IR-induced cell cycle exit from G1 phase requires p53 activity 118 Chapter 7

119 (471, 472). But if p53 regulates both a reversible G1 arrest and cell cycle exit from G1, what determines the outcome of p53 activation in G1? The answer may be found in the discovery that p53, in some cell types, is required to maintain senescence (464, 465). This implies that p53 must be continuously activated in order to remain arrested. In response to DNA damage, p53 levels oscillate, and the number of oscillations has been shown to increase with increasing amounts of DSBs (232, 253). In addition, continuous p53 oscillations caused by sustained DSBs lead to an increased susceptibility to become senescent (233, 254). These data suggest that the decision to withdraw from the cell cycle in G1 may depend on the number of p53 oscillations a cell undergoes. However, as transient induction of p53 alone is insufficient to induce senescence in G1 cells (237), additional DDR signaling likely contributes to the decision to withdraw from the cell cycle in G1. Indeed, p53 continues to be activated in response to sustained DDR signaling (254, 323), suggesting that factors determining how long a DNA break will continue to signal, such as the inability to repair a DSB ( , 323, 342), may contribute to decision to exit the cell cycle following DNA damage in G1. 5. Harnessing the power of single cell imaging to investigate potential causes for cell cycle exit Throughout this thesis, we have studied the effects of DNA damage on cell cycle progression using live cell microscopy. Visualizing cell cycle progression using live cell microscopy has proven to be a powerful tool to study the effect of DNA damage on the cell cycle. Over the past years, many tools to visualize cell cycle position (175, 37, 473, 474), to measure kinase activity (13, 189, 344), to investigate regulatory feedback mechanisms initiated by DNA damage (232, 463) and to visualize DNA repair processes (6, 334, 345, 475, 476) have been developed. This has yielded much insight into the regulation of any of these individual processes. However, much less is known about the functional interplay between these processes. Combining live cell markers for various processes allows us to investigate reciprocal regulation between these processes. For example, by combining markers for different DNA repair pathways with a cell cycle marker, it was shown that homology-directed repair is maximally used in mid-s phase (324). In yet another example, the combined use of a Plk1 activity probe and a live cell marker for sites of DNA damage, it was shown that, during a G2 arrest, Plk1 activity rises despite the presence of unrepaired DNA damage, which can eventually drive cells into mitosis in the presence of DNA damage (25). Finally, using fluorescently tagged Cyclin B1 and 53BP1, we have shown that cell cycle exit is not per se dictated by the amount of DNA damage (Chapter 3, addendum). Thus, the combined use of such probes generates a unique opportunity to investigate the reciprocal effects of DNA damage repair, or the kinases involved in signaling a DSB on the cell cycle and the reversibility of a DNA damage-induced cell cycle arrest. Finally, the recent development of localization based kinase activity probes allows the multiplexing of probes (21, 477), as these probes require only one fluorophore to measure the activity of a given kinase, as opposed to the classically used FRET-based sensors. 6. Cell cycle exit caused by prolonged mitosis For several decades, researchers have investigated the effects of binucleation of cells. Initial findings showed that the prevention of cytokinesis, which results in binucleated G1 cells, causes a proliferative block (478-48). In addition, mitotic slippage, which is caused by the gradual loss of Cyclin B1 in cells arrested in mitosis (42, 43), generates tetraploid G1 cells that arrest in a p53-dependent manner (397, 42, 44). These findings led to the proposal that a checkpoint exists that can restrict the proliferation of tetraploid cells. However, later findings challenged the existence of such a checkpoint. For instance, it was shown that the cytokinesis inhibitor used to generate binucleated G1 cells also induces a proliferative block in mononucleated G1 cells (481, 482). In addition, it was shown that the cession of proliferation following mitotic slippage depends on a prolonged mitotic arrest, rather than the tet- 7 General discussion 119

120 raploid state of the resulting G1 cell (48). Subsequent work showed that a prolonged mitosis causes a G1 arrest upon successful completion of mitosis, indicative of a mitotic timer (49). In chapter 6, we show that time-dependent accumulation of p53 controls this mitotic timer. Our data suggests that entry into G1 with elevated levels of p53 results in p21 induction, arresting the cell in G1. In support of this hypothesis is the recent finding that tetraploidization does not necessarily arrest a cell in G1, rather it initiates a gradually increasing p53 response, which results in a G1 arrest once p21 expression has reached a critical threshold (483). Also, it was recently shown that p21 levels at mitotic exit control the proliferation/quiescence decision (21). Although p21 expression itself is not increased during a prolonged mitotis (Chapter 6), likely due to the mitotic inhibition of transcription (433), it may be rapidly induced following entry into G1 with elevated levels of p53. Indeed, a G1 arrest following prolonged mitosis is p21-dependent (Chapter 6) (49). Loss of centrioles causes delayed mitotic progression as well as chromosome misalignment (443, 444, 484, 485). In the developing mouse embryo, Sas4 knockdown causes centriole loss, which prolongs mitosis (444). This prolonged mitosis is followed by p53-dependent apoptosis in the subsequent G1. Strikingly, no detectable chromosomal abnormalities were observed in Sas4 knockout embryos, supporting the hypothesis that the mitotic timer functions to prevent further proliferation of cells that have originated from a prolonged and potentially problematic cell division. Concluding remarks Faithfull duplication and segregation of the genome is the most fundamental task of the cell cycle. This is achieved through mechanisms that prevent the propagation of damaged DNA, as well as mechanisms that increase the fidelity of DNA segregation. For multicellular organisms, it is beneficial to exclude cells from further proliferation in case a cell fails to correct errors that are potentially tumorigenic. Thus, limiting the reversibility of cell cycle checkpoints is likely of great importance to the survival of the organism. The induction of DNA damage is routinely used in the treatment of cancer, and relies on DNA damage-induced cell death or withdrawal from the cell cycle. However, over time resistance may occur, as not all cells initially respond to such treatments. Chapters 3 and 4 of this thesis unravel DNA damage-induced cell cycle exit mechanisms that limit the proliferation of damaged cells. Yet, in many tumor cells harboring wild-type p53, such mechanisms are repressed (174, 242, 255, 486). Nevertheless, restoration of the cell cycle exit machinery is shown to restore the sensitivity of tumor cells to DNA damaging therapies (174, 255). Thus, the development of pharmaceutical compounds that re-enable such cell cycle exit mechanisms (266, 487) opens promising therapeutic opportunities to improve therapies. The work presented in this thesis has contributed to our understanding of mechanisms that regulate the proliferative capacity of untransformed cells. Knowing how a process should work is a prerequisite for understanding which processes may be deregulated in cancer. Therefore, future work should focus on elucidating how cell cycle exit mechanisms are deregulated in tumor cells, which, hopefully, allows the development of novel therapeutic strategies in the future Chapter 7

121 General discussion 121 7

122