Inhibitors of ROCK kinases induce multiple mitotic defects and synthetic lethality in BRCA2-deficient cells

BRCA2-deficient cells are highly sensitive to poly-ADP-ribose polymerase inhibitors (PARPi) due to their impaired homologous recombination repair. This increased cytotoxicity is triggered by DNA replication stress induced by PARP trapping on DNA. Thus, it is broadly assumed that DNA damage is a prerequisite for BRCA2 synthetic lethality (SL). Here we show that inhibiting ROCK kinases in BRCA2 deficient cells, triggers SL independently from acute replication stress. In contrast, such SL is preceded by enhanced M-phase defects such as anaphase bridges, and abnormal mitotic figures, which were associated with multipolar spindles, supernumerary centrosomes and multinucleation. SL was also triggered by inhibiting Citron Rho-interacting kinase, another enzyme which, similarly to ROCK kinases, regulates cytokinesis. Together, these observations suggest cytokinesis failure as trigger of mitotic abnormalities and SL in BRCA2 cells. Furthermore, preventing mitotic entry by Early mitotic inhibitor 1 (EMI1) depletion promoted survival of BRCA2 deficient cells treated with inhibitors of ROCK kinases, thus reinforcing the association between M-phase and the cell death in BRCA2 deficient cells. This novel mechanism of SL induction is in contrast to the one triggered by PARPi and uncovers mitosis as an Achilles heel of BRCA2 deficient cells.

One immediate consequence of cytokinesis failure is that the resulting cell contains two 241 centrosomes instead of one (Ganem, Storchova, & Pellman, 2007). Normal cells harbor one 242 centrosome which duplicates only once during S phase. Duplicated centrosomes form a bipolar 243 mitotic spindle during a normal mitosis ensuring equal chromosome distribution in daughter 244 cells. (Nigg, 2007). In contrast, multiple centrosomes can lead to multipolar mitosis and cell 245 death (Ganem, Godinho, & Pellman, 2009). We stained cells for gamma-tubulin and alpha-246 tubulin, central components of centrosomes and microtubules, respectively (Brinkley, 1997; 247 silencing of CITK and ROCK1/2 was not additive/synergistic ( Figure 7B), thus suggesting that 267 ROCK and CITK depletion activate the same synthetic lethal mechanism in BRCA2-deficient 268 cells. Together, these findings indicate that cytokinesis failure by multiple sources could induce 269 death in BRCA2-deficient cells. 270 If aberrant transit through mitosis is the origin of the cell death triggered by ROCKi, then the 271 bypass of mitosis should protect those cells from cell death. To this end, we downregulated 272 Early mitotic inhibitor-1 (EMI1), an anaphase promoting complex (APC) inhibitor that has a key 273 role in the accumulation of mitosis activators including B-type cyclins (Reimann et al., 2001). 274 When transfecting siEMI1, we observed a 65% reduction in EMI1 expression ( Figure 7D Remarkably, cell death by ROCK inhibition or depletion was recapitulated by the inhibition of 292 another enzyme that facilitates cytokinesis, CITK, hence suggesting that binucleation precedes 293 multinucleation and SL (see model in Figure 7G). In fact, robust evidence in the literature 294 indicates that highly abnormal metaphases/anaphases, such as the ones we observed, are 295 incompatible with cell viability (Ganem et al., 2009) and are therefore the most plausible cause  We therefore postulate that the cytokinesis failure of a cell with 4N or more DNA content is the 302 major driver for BRCA2 cell death following ROCK inhibition. As such, targeting mitosis alone in 303 the absence of increased replication stress may be sufficient to kill BRCA2 cells. Future 304 research on the mitotic functions of BRCA will certainly provide valuable information on 305 synthetic lethal alternatives for cancers whose hallmark is the loss of this tumor suppressor 306 gene. 307 308 BRCA2-deficient cells can also be killed by mild replication defects which do not cause H2A.X 314 accumulation in S phase (Adam et al., 2021). This is dependent on the transmission of under-315 replicated DNA from S to M phase triggered by BRCA 1 or BRCA2 deficiency and the lack of 316 CIP2A-TOPBP1 complex formation in M phase. In the absence of the later complex, under-317 replicated DNA is aberrantly processed into acentric chromosomes and micronuclei which are 318 the source of SL (Adam et al., 2021). Our present work reveals yet another weakness of 319 BRCA2-deficient, but not BRCA1-deficient, cells: cytokinesis. Strikingly, such SL is not 320 preceded by the accumulation of broken chromosomes or micronuclei and is independent from 321 canonical players of the DDR, as it is observed after ROCK or CITK inhibition. It should also be mentioned that our experimental analysis does not rule out that background 337 levels of replication stress or increased levels of under-replicated DNA induced by BRCA2 338 deficiency could be promoting cell death by ROCK inhibition. As suggested by (Adam et al., 339 2021), it is possible that BRCA2-deficient cells rely more on M phase due to their defects in of 340 the completion of DNA synthesis, making them more susceptible to suboptimal M phase (e.g.: 341 triggered by ROCKi). However, if the source of SL was simply associated with DNA synthesis 342 events, then it would also be likely present in BRCA1-deficient backgrounds, which we did not 343 observe. Importantly, BRCA1 backgrounds are also vulnerable during M phase, as we 344 previously observed SL between BRCA1 and PLK1 inhibitors (Carbajosa et al., 2019). This 345

BRCA2 deficient cells can be killed in a manner that is independent from the induction of acute
indicates that HR impairment is not the only possible trigger of SL in BRCA1 and BRCA2 346 backgrounds that could be therapeutically exploited. In the future, M phase may provide a 347 window of opportunity for novel treatments in patients that do not respond to PARPi therapy.  with Mowiol (Sigma-Aldrich). Slides were analyzed with 40x or 100x objectives using an Axio 518 Observer microscope (Zeiss). 519 520 FACS 521 Cells were seeded, treated and harvested at different time points (24 hours-6 days). Cells were 522 trypsinized, fixed with ice-cold ethanol overnight, and stained with a solution of 100 μg/ml 523 RNase (#10109142001, Roche) and 50 μg/ml propidium iodide (#P4170, Sigma-Aldrich). A total 524 of 10,000 events were recorded using a FACSCalibur (BD Biosciences). Cell cycle distribution 525 was analyzed with the Cytomation Summit software (Dako version 4.3). To assess cell death, 526 cells were treated as above but following trypsinization they were stained with SYTOX Green 527 staining following manufacturer's instructions (#S34860, Invitrogen). A total of 10,000 events 528 were recorded and analyzed using a FACSAria (BD Biosciences). 529 530

Quantitative real-time PCR 531
Total RNA was extracted with TRIzol reagent (Invitrogen), following manufacturer's instructions. 532 A total of 2 ug of RNA was used as a template for cDNA synthesis using M-MLV reverse 533 transcriptase (#28025, Invitrogen) and oligo-dT as primer. Quantitative real-time PCR was 534 performed in a LightCycler 480 II (Roche) using the 5X HOT FIREPol EvaGreen q PCR Mix 535 Plus (#08-24-00001, Solis BioDyne). 536 To calculate relative expression levels, samples were normalized to GAPDH expression. 537

Statistical analysis 553
GraphPad Prism 5.0 was used for all statistical analyses. Regular two-way ANOVA, followed by 554 a Bonferroni post-test or Student's t-tests were used as appropriate. BrdU intensity was 555 analyzed with a Kruskal-Wallis non-parametric test followed by a Dunn's multiple comparison 556 test. Statistical significance was set at p<0.05. 557 558 20

Conflict of interest 560
The authors declare that they have no conflicts of interest. 561

Authors María F. Pansa, Israel Gloger, Gerard Drewes and Kevin P. Madauss are affiliated with 562
GlaxoSmithKline and have no other competing interests to declare. 563 564 Acknowledgements 565 We would like to thank Dr. Fernanda Ledda for providing critical reagents for this work. We 566 would also like to thank all members of the Gottifredi and Soria Laboratories for insightful 567 comments and discussions. We thank Pamela Rodriguez, Esteban Miglietta, Andrés Hugo 568 Rossi and Carla Pascuale for technical support with tissue culture, microscopy and flow 569 cytometry. We also thank the flow cytometry, microscopy, and cell culture facilities of CIBICI-570 CONICET for technical support. were visualized by staining centrosomes (γ-tubulin) and microtubules (α-tubulin) and DNA was 987 stained with DAPI. Cells were classified as having 3 spindles, 4 spindles or 5 spindles or more.