A SOD1-dependent mitotic DNA damage checkpoint

In the event of DNA damage, the cell cycle can be slowed or halted to allow for DNA repair. The mechanisms by which this occurs are well-characterised in interphase, although the mechanisms underpinning mitosis slowing in response to damage are unclear. Canonical checkpoints and DNA repair pathways are largely repressed in mitosis, and whilst there is some level of mitotic DNA synthesis and repair, the bulk of DNA damage is processed for post-mitotic repair. How the decision is made between mitotic DNA repair and post-mitotic DNA repair is not known. We have identified the antioxidant enzyme Superoxide Dismutase 1 (SOD1) as an essential factor mediating delayed mitotic progression in response to DNA damage and replication stress. Cells depleted of SOD1 no longer exhibit DNA damage dependent mitotic delay, and display increased levels of damaged centromeres and mitotic defects. Whilst reactive oxygen species (ROS)-inducing agents also lead to SOD1-dependent mitotic delay, intracellular ROS levels do not correlate with mitotic arrest. SOD1 appears to play an important role in DNA repair in interphase and is recruited to the nucleus in response to DNA damage. In addition to control of mitotic progression in response to genotoxic stress, SOD1 also plays a major role in mitotic DNA synthesis. SOD- depleted cells show reduced levels of mitotic EdU incorporation in response to either replication stress or DNA breaks, seemingly in tandem with Rad51 andSOD1-depletion induced mitotic progression in the presence of DNA breaks is Rad52-dependent. We suggest that there are two responses to DNA breaks in mitosis; either arrest and mitotic repair or progression and post-mitotic repair; and these two pathways exist in a fine balance, controlled by a signaling cascade involving SOD1.


Abstract 26
In the event of DNA damage, the cell cycle can be slowed or halted to allow for DNA repair. The 27 mechanisms by which this occurs are well-characterised in interphase, although the mechanisms 28 underpinning mitosis slowing in response to damage are unclear. Canonical checkpoints and DNA 29 repair pathways are largely repressed in mitosis, and whilst there is some level of mitotic DNA 30 synthesis and repair, the bulk of DNA damage is processed for post-mitotic repair. How the decision 31 is made between mitotic DNA repair and post-mitotic DNA repair is not known. 32 We have identified the antioxidant enzyme Superoxide Dismutase 1 (SOD1) as an essential factor 33 mediating delayed mitotic progression in response to DNA damage and replication stress. Cells 34 depleted of SOD1 no longer exhibit DNA damage dependent mitotic delay, and display increased 35 levels of damaged centromeres and mitotic defects. Whilst reactive oxygen species (ROS)-inducing 36 agents also lead to SOD1-dependent mitotic delay, intracellular ROS levels do not correlate with 37 mitotic arrest. SOD1 appears to play an important role in DNA repair in interphase and is recruited to 38 the nucleus in response to DNA damage. In addition to control of mitotic progression in response to 39 genotoxic stress, SOD1 also plays a major role in mitotic DNA synthesis. SOD-depleted cells show 40 reduced levels of mitotic EdU incorporation in response to either replication stress or DNA breaks, 41 seemingly in tandem with Rad51 andSOD1-depletion induced mitotic progression in the presence of 42

Introduction 56
It is vital that cells maintain genomic integrity in order to pass on a faithful copy of their genetic 57 material to the next generation. All the cells in our body are continuously exposed to genotoxic 58 threats, with tens of thousands of DNA-damaging events occurring daily in each cell. The cellular 59 response to DNA damage involves careful coordination of cell cycle control, DNA repair and 60 programmed cell death. 61 In response to DNA damage, the phosphatidylinositol 3-kinase-related kinases Ataxia Telangiectasia 62 Mutated (ATM) and Ataxia Telangiectasia Rad3-Related (ATR) activate cell cycle checkpoints 63 throughout interphase, resulting in cell cycle arrest at the G1/ S and G2/M boundaries, and slowing 64 of DNA replication via Cyclin dependent kinase (CDK) inhibition [1]. Whilst cell cycle control and 65 activation of DNA repair are well characterised in interphase, how the cell cycle responds to DNA 66 breaks in mitotic cells remains unclear. Whilst we and others have reported slowed mitotic transit in 67 response to DNA damage [2][3][4][5], it is generally accepted that there is no mitotic DNA damage-induced 68 cell cycle checkpoint [6]. The reasoning for this is two-fold; Firstly, the interphase checkpoints all act 69 via inhibition of CDKs necessary for cell cycle progression [7], and several mechanisms prevent the 70 initiation of such mechanisms during mitosis, reviewed in [6]. Low levels of transcription in mitosis 71 prevent CDK inhibition via p21 induction, and furthermore the mitotic kinase Plk1 acts to inhibit 72 Wee1 and , both of which are required for the inhibitory phosphorylation of CDK1 at Tyr 15 [6]. 73 Secondly, the canonical DNA damage response (DDR) is largely inhibited in mitosis to avoid the risk 74 of telomere fusion [8], which has led to the hypothesis that there is no requirement for a mitotic 75 DNA damage checkpoint. Whilst the signalling cascade which responds to DNA double strand breaks 76 (DSB) in interphase can be initiated in mitosis, the cascade is thought to be attenuated in mitotic 77 cells, to be reinstated following the completion of mitosis [9]. Studies in fly [10] and human [11] cells 78 suggest that broken chromosomes can be "tethered" together in mitosis to allow for faithful 79 segregation of potential acentric chromosomes. 80 Genomic instability -sometimes resulting in full chromosome rearrangements -is a hallmark of 81 cancer. Many chromosome rearrangements occur at the same sites in chromosomes, known as 82 "common fragile sites" or CFS [12,13]. CFS represent difficult-to-replicate areas of the genome which 83 usually replicate late in S phase and in some cases in early G2 [14] and mitosis [15] in response to 84 replication stress. In mitosis, the endonucleases MUS81/EME1 are recruited to CFS and induce DNA 85 breaks, whereupon the nuclease activity of MUS81 promotes POLD3-dependent Mitotic DNA 86 synthesis (MiDAS) to minimise chromosome mis-segregation in anaphase [15]. Two types of MiDAS 87 have been observed; one which occurs at sites of DNA DSB and is Rad52 dependent: termed DDR-associated MiDAS, and another which occurs in the absence of DNA DSB and is dependent on 89 Rad51: termed non-DDR-associated MiDAS [16]. Mitotic DNA synthesis also occurs at sites of laser-90 induced DNA DSB [17], indicating that this phenomenon is not restricted to sites of DNA under-91 replication. 92 Superoxide Dismutase 1 (SOD1) has a well-established role as a dismutase of toxic superoxide 93 radicals (O2-), which it converts to the more stable and less toxic hydrogen peroxide and dioxygen 94 [18]. More recently, SOD1 has been implicated in the DDR with elevated levels of DNA damage 95 observed in SOD1 mutant ALS cells [19,20] and furthermore over-expression of SOD1 leads to 96 activation of the DDR in SOD1 mutant cells [21]. Loss of SOD1 has also been shown to confer 97 sensitivity to DNA damaging agents and lead to downregulation of the ATM pathway in yeast [22]. In 98 addition, SOD1 has also been implicated in regulation of gene expression. SOD1 is also reported to 99 be activated by the DDR factors ATM and Chk2 [23][24][25] and in turn, to act as a transcription factor to 100 initiate expression of ROS-reducing and DNA damage-related genes [23]. In this manuscript, we 101 further characterise the role of SOD1 in DNA repair and demonstrate a novel role for SOD1 in the 102 control of mitotic progression in response to DNA damage. 103 104

Immunofluorescence Microscopy 129
HeLa cells (5 x 10 4 ) were seeded directly onto coverslips fixed in methanol or paraformaldehyde, 130 permeabilised in 0.2% Triton-X100, blocked in 5% BSA and stained with the indicated antibodies. 131 Alexa-Fluor 488 and AlexaFluor 594 secondary antibodies (Invitrogen) were used. In the final wash, 132 cells were incubated with DAPI (Life technologies) and mounted to slides using Immumount (Thermo 133 Fisher Scientific). 134 Live cell Imaging 135 48 hours post transfection with indicated siRNAs, cells were trypsinised, exposed to ionising 136 radiation (IR) whilst in suspension and reseeded to 24 well plates. Once adhered to the plates (3-4 137 hours post IR), the cells were loaded into imaging system, or following chemical administration as 138 indicated. Live cell images were captured using ZEISS Cell discoverer 7 microscope every 5 minutes 139 for a duration of 20 hours. 140

Intracellular ROS Assay 141
ROS levels were detected using the chloromethyl-2',7'-dichlorodihydrofluorescein diacetate (CM-142 H2DCFDA) general oxidative stress indicator kit (Invitrogen) according to the manufacturer's 143 instructions. Fluorescence was measured at 495/ 527 nm on the SpectraMax M5e multi-mode 144 microplate reader (Molecular Devices) and readings were normalised to the cell-free control. 145 Comet Assay 146 100,000 cells per condition were isolated and the alkaline comet assay protocol was performed 147 (Trevigen) according to the manufacturer's instructions. Slides were imaged using the 10x/0.25 objective lens on the Eclipse TE200 fluorescent microscope (Nikon) using NIS Elements software 149 (Nikon). The FITC channel was used to visualise the cells. Images were converted to the BNP file format 150 in order to process using CometScore software. The average percentage tail DNA was calculated for 151 50 cells per condition and the overall mean determined across the experimental repeats. 152 qRT-PCR 153 RNA extraction was performed using the manufacturer's protocol and reagents from the RNeasy mini 154 kit (Qiagen). The concentration (ng/µL) of the eluted RNA was quantified using the Nanodrop 155 spectrophotometer (ND-1000). The RNA extracted underwent reverse transcription using the high-156 capacity RNA to cDNA kit (Applied Biosystems). qRT-PCR was conducted using Taqman probes 157 according to the manufacturer's instructions and processed on a 7900 RT-PCR machine using the SDS 158 2.4 software 159

Mitotic EdU incorporation assay 160
MiDAS assay was performed 48 hours post transfection as described by Garribba et al [26]. Images 161 were captured using Nikon ECLIPSE Ti2 confocal microscope. 162 163

DNA-damage induced mitotic delay is dependent on SOD1 165
To investigate the mitotic cell cycle response to DNA damage, several cell lines were treated with 166 DNA damaging agents followed by live cell time-lapse microscopy analysis. We observed that the 167 average time spent in mitosis was significantly increased following the introduction of DNA 168 damaging agents ( Figure 1A, Figure S1A). This delay was not restricted to cells inflicted with DNA 169 damage whilst in mitosis, as the delay can still be seen upwards of 16 hours post-irradiation, 170 indicating that cells can escape earlier interphase checkpoints and enter mitosis with damaged DNA 171 ( Figure 1B). The mitotic delay was also expressed as an increase in cells expressing the mitotic 172 marker protein phosphorylated Histone-H3, 16 hours after IR treatment ( Figure 1C), which allowed 173 us to screen for mitotic delay in the absence of selected DNA repair proteins. Unexpectedly, 174 depletion of several DNA repair proteins previously implicated in mitotic progression (including ATM, 175 ATR, Chk1, the MRN complex and MDC1) did not significantly reduce the mitotic population 176 observed following treatment with IR, whereas cells depleted of the antioxidant protein SOD1 no 177 longer exhibited mitotic delay following exposure to IR ( Figure 1D).
The siRNA used in the screen was based on pools of 4 individual siRNAs. Deconvoluted, all but one 179 of the 4 SOD1 siRNAs used resulted in bypass of mitotic delay. The remaining siRNA yielded 180 inconsistent data, and western blotting tests for SOD1 protein levels confirmed that it did not 181 effectively reduce SOD1 protein levels ( Figure S1B and C). We selected two siRNAs (5 and 7) for 182 further study, and assessed the duration of mitosis in cells depleted of SOD1 or the Spindle Assembly 183 Checkpoint (SAC) protein BUBR1 following exposure to IR Consistent with previous reports (REF), 184 depletion of BUBR1 reduced mitotic transit time in unperturbed and IR treated cells alike, while 185 depletion of SOD1 had no effect on mitotic progression in the absence of DNA damage ( Figure 1E).

Mitotic delay is ROS independent 195
As the canonical function of SOD1 is the reduction of intracellular ROS, we investigated whether 196 agents that increase ROS also lead to prolonged mitosis. We found that both hydrogen peroxide 197 (H2O2) (Figure 2A) and pyocyanin (a ROS-inducing toxin produced by the bacterium Pseudomonas 198 aeruginosa) ( Figure S2A) induce SOD1-dependent mitotic delay to an extent similar to that observed 199 in the presence of other DNA damaging agents ( Figure 2B). SOD1 is required for the production of 200 H2O2 from superoxide molecules so we initially theorized that the mitotic delay observed could be 201 due to altered levels of cellular H2O2. However, the DNA damaging agents carboplatin and 202 temozolomidealso induce SOD1-dependent mitotic arrest ( Figure 2B), but do not produce 203 significantly increased ROS (Figure 2c). Surprisingly, SOD1 siRNA depletion did not lead to a 204 significant reduction of detectable H2O2 in any condition (Figure 2C), suggesting that altered ROS 205 levels do not underpin SOD1 regulation of mitotic transit 206 Glutathione peroxidase (GPx1) is a ROS-reducing protein which catalyses the reduction of H2O2 to 207 water and oxygen. Selenium supplementation has been shown to upregulate GPx1. We found that 208 supplementing cells with 50nM selenium led to a detectable increase in GPx1 expression ( Figure S2B  209 and C) and prevented SOD-dependent, IR-induced mitotic arrest ( Figure 2D). Surprisingly, selenium 210 supplementation did not reduce cellular ROS ( Figure 2E) further supporting the hypothesis that H2O2 is not the cause of the arrest. Furthermore, selenium supplementation also protected against SOD1-212 dependent mitotic arrest induced by non-ROS-inducing DNA damaging agents ( Figure S2D and 2E), 213 indicating that selenium prevents mitotic arrest via another pathway. Glutathione metabolism has 214 been implicated in regulation of Protein phosphatase 2A (PP2A), which is required for mitotic exit 215 [27]. Glutathione reduces oxidised PP2A, recovering phosphatase activity [28] and selenium has also 216 been shown to potently stimulate PP2A activity directly [29]. We found that overexpression of PP2A 217 was also able to override DNA damage induced arrest (Figure 2F), suggesting one possible 218 mechanism via which selenium overrides the arrest. Taken together, our data suggest that changes 219 in intracellular ROS in the absence of SOD1 do not account for the mitotic delay observed following 220 treatment with DNA damaging agents. 221 222 SOD1 depletion leads to elevated mitotic defects and increased DNA damage but does not impact 223 the SAC 224 DNA damage induces many aberrant mitotic phenotypes such as micronuclei, DNA bridges, lagging 225 chromosomes and failed cytokinesis (Figure S3A), all of which may influence mitotic transit time. We 226 observed all these phenotypes following treatment with IR, however SOD1 depletion induced an 227 additional increase in aberrant mitotic phenotypes (Figure 3A and S3B) indicating that SOD1-induced 228 mitotic delay is not a consequence of these phenotypes. Mikhailov et al suggest that DNA damage 229 induced mitotic arrest is due to damaged centromeres preventing the SAC from being satisfied [3]. 230 However, whilst we observe damaged centromeres in mitotic cells (as represented by ɣH2AX) 231 4Dfollowing treatment with IR, these are not only present but are increased in the absence of SOD1 232 ( Figure 3B), indicating that they are not directly responsible for the observed arrest. 233 In lower eukaryotes, DNA damage induces mitotic arrest dependent on the spindle assembly 234 checkpoint (SAC) [30]. To determine whether SOD1 as both a catalytic enzyme and a transcription 235 factor to control the SAC, we first studied the protein and mRNA levels of the mitotic checkpoint 236 complex (MCC) which are required for SAC arrest. qPCR revealed no significant transcriptional 237 changes of the SAC proteins following SOD1 siRNA depletion ( Figure S4A) and this also translated to 238 the protein level ( Figure 3C). To confirm that SOD1 does not affect the SAC, we treated cells with the 239 mitotic inhibitor nocodazole, which leads to microtubule destabilisation and permanent SAC 240 activation. When treated with nocodazole, cells arrest in mitosis at the SAC and do not undergo 241 cytokinesis. BUBR1 (an essential protein in the SAC) depletion abrogates nocodazole-induced arrest 242 and the majority of cells escape mitosis without dividing (a process known as "mitotic slippage"). 243 However, following SOD1 depletion, we observed no significant change in the proportion of arrest, death or slippage when compared with the control sample, indicating that SOD1 is not required for 245 initiation or maintenance of the SAC (Figure 3D). 246 247 SOD1 is involved in the DDR 248 SOD1 mutations lead to the accumulation of DNA damage [20] indicating a role for SOD1 in the DDR. 249 SOD1 accumulates in the nucleus in response to H2O2 [23,25]. We show that this is also the case in 250 response to IR (Figure 4A,B) We also observed higher levels of ɣH2AX foci, a DNA damage marker, in 251 SOD1-depleted cells, than in cell transfected with a non-targeting control siRNA cells following 252 treatment with H2O2 or exposure to IR Figure 4C). Increased DNA damage in SOD1-depleted cells was 253 further confirmed via COMET assay (Figure 4D and E). However, despite higher or equal levels of 254 damage ( Figure 4F), SOD1 depletion resulted in a reduction in observable RAD51 foci 4 hrs post-IR, 255 indicating reduced DNA repair capacity ( Figure 4G). 256

SOD1 depletion leads to reduced Mitotic EdU incorporation 258
Limited DNA replication and repair have been observed in mitosis via incorporation of EdU in 259 response to replication stress [15,31,32] and more recently, DNA damage [17]. We observed that 260 SOD1 promotes efficient mitotic EdU incorporation in response to both replication stress ( Figure  261 5A,B) and DNA damage ( Figure 5C). Rad51 [16] and Rad52 [31] have both been demonstrated to play 262 a role in mitotic EdU incorporation. Our data demonstrates that SOD1 depletion reduces mitotic EdU 263 incorporation to an extent similar to the RAD51 inhibitor (BO2) and this reduction was not 264 synergistic in combination with BO2 treatment (Figure 5d). RAD52 inhibition did not reduce the 265 levels of mitotic EdU incorporation to the same extent as SOD1 depletion or RAD51 inhibition but 266 displayed an additive decrease in combination with SOD1 inhibition (Figure 5d) Here we demonstrate a novel role for Superoxide Dismutase 1 in control of the cell cycle and DNA 274 repair following induction of DNA damage. We have uncovered a SOD1-dependent mechanism that 275 drives mitotic slowing in the presence of DNA damage. In the absence of this delay, we observe 276 elevated levels of DNA damage, increased mitotic defects and reduced mitotic DNA synthesis. 277 Previous reports have proposed that damage to kinetochores causes delayed mitotic progression 278 though persistent SAC activation [3,16]. However, we found that whilst cells treated with IR exhibit 279 high levels of damaged centromeres, this was elevated further in cells depleted of SOD1. Since 280 SOD1-depleted cells do not exhibit DNA damage-induced mitotic arrest but are capable of 281 nocodazole-induced mitotic arrest we propose that DNA damage leads to prolonged mitotic 282 progression through a pathway distinct from the canonical SAC. 283 Previously studies suggest that RAD51 inhibition leads to reduced mitotic EdU incorporation and 284 prolonged mitosis and it was hypothesized that mitotic exit cannot occur until mitotic DNA synthesis 285 is complete [16]. Surprisingly, we observed that whilst SOD1 depletion has the opposite effect to 286 RAD51 inhibition in that it promotes mitotic exit, SOD1 depletion phenocopied Rad51 inhibition and 287 reduced mitotic EdU incorporation. RAD52 has been shown to be important for mitotic EdU 288 incorporation [31] though to a lesser extent than RAD51 [16] and our data reflect this. However, 289 whilst SOD1 depletion had no effect on mitotic EdU incorporation in the absence of RAD51, SOD1 290 depletion led to a further reduction in mitotic EdU incorporation in the absence of RAD52. This can 291 potentially be explained by the two types of mitotic EdU incorporation described by Wassing et al 292 [16]; RAD51-dependent non-DDR-associated MiDAS which occurs in the absence of breaks and 293 RAD52-dependent DDR-associated MiDAS which occurs in the presence of breaks. We propose that 294 SOD1 and Rad51 are required for both pathways, but Rad52 is required only for DDR-associated 295

MiDAS. 296
Interestingly, we found that mitotic progression in the absence of SOD1 is dependent on RAD52. We 297 propose that there are two potential fates for mitotic DNA breaks; mitotic delay and repair through 298 MiDAS or mitotic progression, resulting in mitotic defects, followed by repair in G1. Our model 299 ( Figure 5F) suggests the existence of a fine balance between these two fates, controlled in part by 300 SOD1 and RAD52. We propose that in response to DNA breaks, SOD1 guides cells down a pathway 301 whereby mitosis is slowed and mitotic DNA synthesis and repair is instigated, however in the 302 absence of SOD1, this pathway can no longer function, meaning cells prepare for repair in the 303 subsequent G1. We propose a "point of no return" in this pathway, after SOD1 but before RAD51. If 304 cells are guided down the repair in mitosis pathway by a functional SOD1 but RAD51 is not present, a 305 prolonged mitosis results but repair cannot be completed resulting in mitotic cell death. 306 In this study, we demonstrate the existence of a signaling cascade leading to prolonged mitosis and 307 mitotic DNA synthesis following genotoxic stress induced by DNA damage or replication defects.
Further work is required to elucidate the mechanism controlling mitotic progression in the presence 309 of DNA damage and whether this involves SAC components -however, our data indicate that the 310 canonical SAC is active in the absence of SOD1. These data suggest the existence of a previously 311 unknown DNA damage checkpoint in mitosis which detects DNA damage and prolongs mitosis to 312 allow for mitotic DNA repair. This checkpoint appears to be independent of the interphase DNA 313 damage checkpoints and could be a potential druggable target for future therapeutics design.  48 hours prior to treatment with 5Gy IR, followed by G2 arrest with 9μM RO3306 1 hour later for 5 392 hours. Cells were released into mitosis and then pulsed for 30 minutes with EdU. Number of EdU foci 393 per mitotic cell was scored. Error bars represent SEM of 3 individual repeats. N=150. Results were 394 analysed using a one-way ANOVA with Dunnett's multiple comparison test. *p=0.05, **p=0.01, 395 ***p=0.001, ****p=0.0001 D. As in A. In the case of chemical inhibition, cells were pulsed for 30 396 minutes with EdU in the presence of either 20μM B02 or 20μM AICAR. Results were analysed using a 397 one-way ANOVA with Dunnett's multiple comparison test. *p=0.05, **p=0.01, ***p=0.001, 398 ****p=0.0001 E. HeLa cells were incubated with the indicated siRNA for 48 hours prior to treatment 399 with 5Gy IR and Rad52 inhibitor. Cells were anlysed by time lapse microscopy for 20 hours. Error 400 bars represent mean and SEM of 3 independent biological replicates. N=150. Results were analysed 401 using a one-way ANOVA with Dunnett's multiple comparison test. *p=0.05, **p=0.01, ***p=0.001, 402 ****p=0.0001 F. Proposed model. 403