The oncoprotein DEK affects the outcome of PARP1/2 inhibition during replication stress

DNA replication stress is a major source of genomic instability and is closely linked to tumor formation and progression. Poly(ADP-ribose)polymerases1/2 (PARP1/2) enzymes are activated in response to replication stress resulting in poly(ADP-ribose) (PAR) synthesis. PARylation plays an important role in the remodelling and repair of impaired replication forks, providing a rationale for targeting highly replicative cancer cells with PARP1/2 inhibitors. The human oncoprotein DEK is a unique, non-histone chromatin architectural protein whose deregulated expression is associated with the development of a wide variety of human cancers. Recently, we showed that DEK is a high-affinity target of PARylation and that it promotes the progression of impaired replication forks. Here, we investigated a potential functional link between PAR and DEK in the context of replication stress. Under conditions of mild replication stress induced either by topoisomerase1 inhibition with camptothecin or nucleotide depletion by hydroxyurea, we found that the effect of acute PARP1/2 inhibition on replication fork progression is dependent on DEK expression. Reducing DEK protein levels also overcomes the restart impairment of stalled forks provoked by blocking PARylation. Non-covalent DEK-PAR interaction via the central PAR-binding domain of DEK is crucial for counteracting PARP1/2 inhibition as shown for the formation of RPA positive foci in hydroxyurea treated cells. Finally, we show by iPOND and super resolved microscopy that DEK is not directly associated with the replisome since it binds to DNA at the stage of chromatin formation. Our report sheds new light on the still enigmatic molecular functions of DEK and suggests that DEK expression levels may influence the sensitivity of cancer cells to PARP1/2 inhibitors.


INTRODUCTION
For the determination of tract length ratios, U2-OS shDEK and control cells were   After 9 min the slides were tilted at 30-45° and the resulting DNA spreads were air-rabbit AlexaFluor-546 (both 1:400; both Thermo Fisher Scientific). For nuclear 194 counterstaining, cells were incubated in 200 ng/μl Hoechst 33342/PBS (Merck). 195 Coverslips were mounted on microscopy slides using Aqua Polymount (Polysciences). 196 Replicating cells were visualized by labelling with 10 μM EdU 10 min prior to replication 197 stress induction. Cells were fixed and immunostainings performed as described above. 198 After incubation with secondary antibodies, EdU was detected using the Klick-it EdU  To test the efficiency of PARP inhibitor ABT-888, cells were left untreated or pre-215 treated with 1 µM ABT-888 for increasing time points, followed by DNA damage 216 induction using 800 µM H2O2 (Merck) for 10 min. Detection of PAR was achieved after 217 fixation with methanol:acetic acid (3:1) using a monoclonal mouse α-PAR antibody (10H, 1:300 in PBSMT (5 % milk powder, 0.05 % Tween 20, PBS) at 4 °C overnight.

219
Confocal microscopy was performed with a Zeiss LSM 510 Meta as described above. 220 PAR nuclear intensities were analysed using Fiji v1.49u.         888. Both drugs were added simultaneously with the IdU-containing medium (Fig 1 A).

372
In line with our previous results, knockdown of DEK expression slowed down 373 replication fork progression per se as indicated by a highly significant reduction of the 374 IdU tract lengths in untreated shDEK cells as compared to controls (Fig 1 C and [43]).  damage response in our experimental setting (Fig 2 A, B). with an iPOND pulse-chase experiment, in which we sought to observe the dynamics 474 of DEK binding to nascent chromatin (Fig 3 D-F). EdU was applied for 10 min to pulse-  To complement this biochemical approach we studied the localization of DEK in 507 replicating cells by superresolution microscopy (Fig 4). To this purpose, we took  (Fig 4 A). This finding was corroborated by stochastic optical reconstruction   In contrast, shDEK cells were completely protected from restart impairment, displaying a slightly higher number of restarting forks as compared to control cells not 552 exposed to ABT-888 (Fig 5 C). These data reflect the same phenotype of DEK 553 downregulation counteracting PARP1/2 inhibition as observed in the context of fork 554 slowing by CPT, and underscore its functional relevance. PARylation inhibition (Fig 6). We applied 2 mM HU for 80 min, as this dose and time 577 of exposure was shown to elicit a maximal HU response in U2-OS cells [56], and 578 measured RPA foci in EdU positive cells (see also S4 Fig for the quantification method).

579
As observed for fork slowing and DSB formation, shDEK cells showed a significantly 580 increased RPA response as compared to control cells. In the latter, additional ABT-888 (Fig 6 B). These data suggest that DEK plays a role in limiting the formation 586 of long ssDNA stretches upon stalling or collapse of replication forks, and that its 587 downregulation compensates for the previously described requirement for PARP1/2 588 activity for RPA binding at a subpopulation of stalled forks. These results further 589 corroborate the existence of a reciprocal functional link between DEK and PARP1/2 in 590 the response to replication stress.   involved too, because the PBD2-Mut2 did not fully restore the level of RPA foci 676 obtained in shDEK cells after HU treatment (Fig 7 F). Taken together, these data 677 strongly suggest that the non-covalent interaction with PAR plays a major role in 678 regulating how DEK affects the response to replication stress providing a first 679 mechanistic insight in the complex molecular interplay of PARP1/2 and DEK.  which is suggestive of a common mechanism.

727
How DEK affects the sensitivity of replication forks towards PARP inhibition, is a matter 728 of speculation so far. Based on our iPOND and superresolution microscopy data it 729 seems unlikely that DEK acts directly at the fork, since we did not find it associated to 730 nascent DNA before the stage of nucleosome formation. Rather, the data presented 731 here lend credit to the hypothesis that DEK is part of a DNA replication regulatory 732 circuitry orchestrated by PAR that affects response to acute PARPi treatment.

733
Deutzmann for initial input in this study, and the FlowKon core facility at the University 749 of Konstanz for cell sorting.