PTEN‐mediated dephosphorylation of 53BP1 confers cellular resistance to DNA damage in cancer cells

Homologous recombination (HR) repair for DNA double‐strand breaks (DSBs) is critical for maintaining genome stability and conferring the resistance of tumor cells to chemotherapy. Nuclear PTEN which contains both phosphatidylinositol 3,4,5‐trisphosphate 3‐phosphatase and protein phosphatase plays a key role in HR repair, but the underlying mechanism remains largely elusive. We find that SUMOylated PTEN promotes HR repair but represses nonhomologous end joining (NHEJ) repair by directly dephosphorylating TP53‐binding protein 1 (53BP1). During DNA damage responses (DDR), tumor suppressor ARF (p14ARF) was phosphorylated and then interacted efficiently with PTEN, thus promoting PTEN SUMOylation as an atypical SUMO E3 ligase. Interestingly, SUMOylated PTEN was subsequently recruited to the chromatin at DSB sites. This was because SUMO1 that was conjugated to PTEN was recognized and bound by the SUMO‐interacting motif (SIM) of breast cancer type 1 susceptibility protein (BRCA1), which has been located to the core of 53BP1 foci on chromatin during S/G2 stage. Furthermore, these chromatin‐loaded PTEN directly and specifically dephosphorylated phosphothreonine‐543 (pT543) of 53BP1, resulting in the dissociation of the 53BP1 complex, which facilitated DNA end resection and ongoing HR repair. SUMOylation‐site‐mutated PTENK254R mice also showed decreased DNA damage repair in vivo. Blocking the PTEN SUMOylation pathway with either a SUMOylation inhibitor or a p14ARF(2‐13) peptide sensitized tumor cells to chemotherapy. Our study therefore provides a new mechanistic understanding of PTEN in HR repair and clinical intervention of chemoresistant tumors.

Homologous recombination (HR) repair for DNA double-strand breaks (DSBs) is critical for maintaining genome stability and conferring the resistance of tumor cells to chemotherapy.Nuclear PTEN which contains both phosphatidylinositol 3,4,5-trisphosphate 3-phosphatase and protein phosphatase plays a key role in HR repair, but the underlying mechanism remains largely elusive.We find that SUMOylated PTEN promotes HR repair but represses nonhomologous end joining (NHEJ) repair by directly dephosphorylating TP53-binding protein 1 (53BP1).During DNA damage responses (DDR), tumor suppressor ARF (p14ARF) was phosphorylated and then interacted efficiently with PTEN, thus promoting PTEN SUMOylation as an atypical SUMO E3 ligase.Interestingly, SUMOylated PTEN was subsequently recruited to the chromatin at DSB sites.This was because SUMO1 that was conjugated to PTEN was recognized and bound by the SUMO-interacting motif (SIM) of breast cancer type 1 susceptibility protein (BRCA1), which has been located to the core of 53BP1 foci on chromatin during S/G2 stage.Furthermore, these chromatin-loaded PTEN directly and specifically dephosphorylated phosphothreonine-543 (pT543) of 53BP1, resulting in the dissociation of the 53BP1 complex, which facilitated DNA end resection and ongoing HR repair.SUMOylation-sitemutated PTEN K254R mice also showed decreased DNA damage repair in vivo.Blocking the PTEN SUMOylation pathway with either a SUMOylation inhibitor or a p14ARF (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13) peptide sensitized tumor cells to chemotherapy.Our study therefore provides a new mechanistic understanding of PTEN in HR repair and clinical intervention of chemoresistant tumors.

Introduction
DNA lesions caused by environmental or endogenous genotoxic insults are major threats to genomic integrity [1,2].DNA double-strand breaks (DSBs) are the most deleterious DNA lesions, which cause gene mutation, cell death, development disorder, and tumor predisposition if not repaired correctly and promptly [3,4].There are two major pathways for DSB repair, homologous recombination (HR), and nonhomologous end joining (NHEJ) repair [5].53BP1, a pro-choice of DSBs, promotes NHEJ repair through inhibiting recruitment of HR repair factors including BRCA1 and CtIP to DSB sites in G1 stage [6].DNA damage-induced phosphorylation at multiple sites in the N terminus of 53BP1 mediates its interaction with downstream factors RIF1 and PTIP [7].RIF1 recruits the Shieldin complex, of which the subunit SHLD2 directly binds ssDNA and blocks DNA end resection, and loss of this complex dramatically increases HR repair [8][9][10].On the other hand, HR repair depends on the exist of sister chromatid, which occurs mainly in S/G2 stage.BRCA1, a critical regulator of HR repair, promotes multiple steps including DNA end resection, RAD51 loading and ssDNA strand pairing [11,12].BRCA1 can recruit a ubiquitin E3 ligase UHRF1 to mediate polyubiquitination of RIF1, resulting in RIF1 dissociation from 53BP1 and thus promoting HR repair in S/G2 stage [13].Moreover, BRCA1 can also facilitate dephosphorylation of 53BP1 during S/G2 stage [14].The region coded by exon11 of BRCA1 is required for the dephosphorylation of 53BP1 and RIF1 release from DNA breaks; however, the underlying molecular mechanism remains largely elusive [15,16].
Here, we provided evidences that SUMOylation of PTEN was increased by SUMO-E3-like p14ARF in DDR.SUMOylated PTEN was recognized and then recruited by the N-terminal SUMO-interacting motif (SIM) of BRCA1 to the chromatin.PTEN located at chromatin directly and specifically dephosphorylated pT543 of 53BP1, leading to RIF1 release and therefore facilitating DNA end resection.PTEN K254R knock-in mice model also showed HR deficiency in DDR in vivo.Notably, inhibiting SUMOylation of PTEN by either a SUMOylation inhibitor or a peptide p14ARF (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13) sensitized tumor cells to DNA damage agents, which might provide a new therapeutic strategy for clinical intervention of chemo-resistant tumor cells.

Antibodies, reagents, plasmids, siRNA, shRNA, and sgRNA
Antibodies used in this study were listed in Table S1.PTEN cDNA was subcloned into pCMV-Flag, pEF-5HA, pEGFP-C1, pCD510B, and pGEX-4T-1 vectors.shRNAs of BRCA1, PTEN and p14ARF were designed and cloned into the vector pLKO.1.RFP-PCNA was a gift from Prof Pumin Zhang [35].Flag-53BP1 and Myc-BRCA1 were gifts form Prof Xingzhi Xu [36].pCBASceI, EJ5-GFP, and DR-GFP were purchased from Addgene (Watertown, MA, USA).p14ARF was cloned into pGEX-4T-1, pEYFP-N1, and pCD513B vectors.PCR-mediated site-directed mutagenesis and truncated proteins were performed according to standard procedures to create the PTEN, p14ARF, 53BP1, and BRCA1 mutants.All clones were sequenced to confirm the desired mutations.siRNAs targeting BRCA1 were synthesized by GenePharma (Shanghai, China).Two sgRNAs were designed to knockout PTEN.In brief, sgRNA was insert into Len-tiCRISPR v2 and delivered into DU145 cells, each single clone was selected and cultured.PTEN knockout clones were identified and used in our study.A list of the sequence information for the shRNAs, siRNAs, and sgRNAs was provided in Table S2.

Immunoblot and denatured immunoprecipitation for SUMOylation detection
Cells were washed once with PBS and lysed in lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM DTT, 1 mM EDTA, 1% NP-40, complete protease inhibitor cocktail (Roche, Basel, Switzerland) and 20 mM N-ethylmaleimide) on ice for 30 min, then lysates were sonicated and centrifuged at 12 000 g for 30 min.Protein concentrations were quantified with BCA kit (Thermo-Fisher, Waltham, MA, USA).Equivalent amounts of protein (1-2 mg) were incubated with 1 lg indicated antibody and 20 lL protein A/G beads (#IP05, Calbiochem, Oakville, Canada) at 4 °C overnight.Beads were collected with centrifugation and washed with lysis buffer for five times, and then boiled with 29 protein loading buffer before analysis by SDS/PAGE.
Denatured immunoprecipitation for SUMOylation detection was carried out as previously described with several modifications [37].Briefly, cells were lysed with SUMO lysis buffer (62.5 mM Tris pH 6.8, 2% SDS), sonicated, and boiled after addition of 1 mM DTT.The lysis was centrifuged at 12 000 g for 15 min.Supernatant was transferred into new EP tube and diluted to a final concentration of 0.1% SDS with lysis buffer.Equivalent amounts of protein were incubated with indicated antibodies (anti-PTEN, anti-HA and anti-Flag) and protein A/G beads overnight at 4 °C.Beads were washed with lysis buffer containing 300 mM NaCl and 0.1% SDS for five times and boiled with 29 protein loading buffer before analysis by SDS/PAGE.All western blot experiments were repeated at least twice.

Ni 2+ -NTA pull down for SUMOylation assay
For the detection of PTEN SUMOylation during DNA damage repair, 293T cells were transfected with His-SUMO1 and indicated plasmids for 48 h.CPT (#S1288, Selleck, Houston, TX, USA) and Zeocin (#60216ES80, Yeasen) were used to induce DNA damage for 1 h.cells were harvested at indicated time, 10% cells were used as input.SUMO-PTEN were pulled down with Ni 2+ -NTA beads (#30210, Qiagen, Germantown, MD, USA) and analyzed with SDS/PAGE as previous described [25].

Immunofluorescence
Cells were seeded on glass coverslips.After treatment with various DNA damage stimuli, cells were washed with PBS and then fixed with 4% (w/v) paraformaldehyde in PBS for 15 min at room temperature.For immunofluorescence after pre-extraction, cells were treated with 1% triton/TBS for 5 min on ice before fixing.When staining RPA32, cells were pre-extracted with cold 0.5% Triton X-100 in PBS for 3 min before fixing.After fixing, cells were permeabilized with 0.5% (v/v) Triton X-100 in PBS for 60 min and blocked with 5% BSA in PBS for 60 min at room temperature.Generally, cells were then incubated with the primary antibody diluted in PBS-BSA overnight at 4 °C.Cells were washed three times with PBST and then incubated with secondary antibodies diluted in PBS-BSA supplemented with 2 lgÁmL À1 of Hoechst 33342 (#62249, Thermo-Fisher) to stain DNA for 1 h at room temperature.Cells were washed three times with PBST and then the coverslips were mounted onto glass slides with Prolong Gold mounting agent (#P36931, Thermo-Fisher).Confocal images were taken with a LSM710 or Zeiss (Oberkochen, Germany) LSM880 laser-scanning confocal microscope.
For the discrimination of cells at S stage, cells were pre-incubated with 10 lM EdU (#C0081S, Beyotime, Shanghai, China) for 30 min before DNA damage induction.Click-It reaction were carried out as manufacture's protocol to staining EdU-positive cells.After this, cells were used for further immunostaining.
Intensity and number of DNA damage induced foci were counted with the FOCO software [38].

HR and NHEJ repair assay
To determine the efficiency of HR-mediated DSB repair in cells expressing different form of PTEN with mutation, we first generated stably PTEN knockdown U2OS-DR-GFP cells with shRNA.Then, PTEN-WT, C124S, G129E, K254R, and K266R were re-expressed in those cells.To increase RFP-I-Sce1 expression efficiency, RFP-I-Sce1 was subcloned into the vector CD510B.Pseudo-lentivirus expressing RFP-I-Sce1 was used to infect U2OS-DR-GFP cells.After 48-72 h, cells were collected for FACS analysis.Percentage of RFP-and GFP-positive cells were quantified.Homologous recombination efficiency was quantified as (GFP + /RFP + )*100%.For NHEJ repair efficiency, we first re-expressed PTEN-WT, C124S, G129E, K254R and K266R in U2OS-shPTEN cells.The NHEJ efficiency was detected with the EJ5-GFP reporter [39].Plasmids EJ5-GFP and ISCE1-RFP were cotransfected into U2OS cells.Forty-eight hours after transfection, cells were collected and fixed for the flow cytometry analysis of RFP and GFP.The NHEJ efficiency was calculated by the ratio of GFP/RFP.

Laser micro-irradiation
Generation of localized DNA damage by laser was done as previously described [35].Briefly, cells were seeded in a live-cell imaging culture dish, transfected with GFP-PTEN and RFP-PCNA and cultured for 48 h. 2 lgÁmL À1 Hoechst 33342 was used to presensitize cells for 10 min before laser micro-irradiation.For micro-irradiation, the cell dish was mounted on the stage of a Leica SP8 microscope at 37 °C.405 nm UVA focused through a 639 1.4NA oil objective was used to induce localized DNA damage.Laser power was set to 50% and iterations were set to 50 times.Time-lapse imaging of recruitment of GFP-or RFP-tagged proteins to DNA damage site was captured every 30 s after micro-irradiation with 488 and 561 nm laser.

GST pull-down
GST-fused protein was expressed in Escherichia coli BL21 and affinity-purified with GST beads (#17-0756-01, GE Healthcare Life Sciences, Marlborough, MA, USA).293T cells overexpressing indicated proteins were treated with DNA damage reagents and lysed.And then cell lysates were incubated with above GSTfused protein beads overnight.Beads were collected and washed five times, followed by western blotting analysis.

In vitro dephosphorylation assay
For measuring dephosphorylation of 53BP1 by PTEN, the GST-fused PTEN WT , PTEN C124S , PTEN G129E , and PTEN G129R , as well as GST protein were purified from E. coli BL21.Full-length Flag-53BP1 and GFP-53BP1 were purified from 293T cells after Zeocin treatment.The dephosphorylation assays were performed in phosphatase assay buffer (20 mmolÁL À1 HEPES, pH 7.2, 100 mmolÁL À1 NaCl, and 3 mmolÁL À1 DTT).The reactions were incubated at 37 °C for 60 min with or without the addition of recombinant GST-fused PTEN WT , PTEN C124S , PTEN G129E , or PTEN G129R , as well as GST protein as negative control, and then were stopped by adding 29 SDS loading buffer for immunoblotting analysis.

Cellular fractionation
Mouse embryonic fibroblasts, DU145, H1299 cells cultured with 90-100% confluence were harvested after treatment with DNA damage reagents and recovery for indicated time.Extraction of cytoplasmic and nuclear proteins was performed using the Nuclear/Cytosol Fractionation Kit (BioVision, Waltham, MA, USA) according to its instruction.
Separation of chromatin-associated proteins was performed as previously described with minor modification [40].Briefly, cell pellets were washed with cold PBS and then incubated with buffer A (10 mM pH 7.9 HEPES, 10 mM KCl, 1.5 mM MgCl 2 , 0.34 M sucrose, 10% glycerol, 1 mM DTT, protease inhibitors, 0.1% Triton-X100) for 10 min on ice.Cell pellets were collected with centrifugation and washed twice with buffer A. Next, cell pellets were gently resuspended in buffer B (3 mM EDTA, 0.2 mM EGTA, 1 mM DTT, protease inhibitors) and incubated for 30 min on ice.Then, cell pellets were collected by centrifugation and lysed in 2% SDS as chromatin-associated proteins.When detecting proteins tightly associated with chromatin, we pipetted cell pellets with buffer B harshly until sticky chromatin pellets were visible after incubation with buffer A. After another 30-min incubation in buffer B, chromatin pellets containing tightly associated proteins was collected by centrifugation and lysed with 2% SDS.

Cell viability and colony formation assay
For cell viability assay, cells were counted and 10 000 cells were seeded into 96-well plates.After 24 h, DNA damage reagents were added and cultured for another 3 days.CCK8-kit was used to detect cell viability and all quantitative results were normalized to nontreatment group.For colony formation assay, 500 or 1000 cells were seeded into a 12-well plate.After 24 h, CPT and Zeocin were added for 48 h and replaced with fresh medium.Cisplatin was added into medium for 72 h and replaced with fresh medium.All culture medium was changed every 3 days until colony was visible.Colonies were washed, fixed and stained with 0.1% crystal violet overnight.Visible colonies were counted and analyzed between groups with IMAGEJ.

Cell cycle profile and EdU incorporation assay
Cells were collected and fixed by 4% PFA when proliferated to 80-90% in culture dish.DNA content and cell cycle profile was determined by flow cytometry after PI staining.The EdU incorporation assay was performed to detect cell proliferation rate.EdU (10 lM) was added into culture dish for 2 h when cell proliferated to 80-90% and then fixed with 4% PFA.Click-It reaction were carried out as manufacture's protocol to staining EdU positive cells.After this, percentage of EdU-positive cells were quantified by flow cytometry.

Mouse model and IHC
All animal studies were conducted with the approval and guidance of Shanghai Jiao Tong University Medical Animal Ethics Committees (Approval NO.A-2019-036).The mice were housed in specific pathogen-free environment, handled with care, and allowed for adaption to the environment before experiments.Pten-K254R and Pten-K266R knockin C57BL/6 mice were generated by BRL medicine company with CRISPR-Cas9 and homozygous mice were verified with PCR sequencing.For IHC, 4-month-old male mice were chosen and subjected to whole body irradiation (IR) with 8 Gy.Mice were sacrificed at Day 4 post IR, small intestines were used for histological analysis.HE staining were used to quantify villi length, and Ki67 staining was used to identify proliferating cells in small intestines.

Statistical analysis
Group data are presented as mean with or without AE SD.The statistical significance between experimental groups was determined by Student's t-test (two tailed and unpaired).P < 0.05 was considered to be significant (n.s., not significant; *0.01 < P < 0.05, **0.001 < P < 0.01 and ***P < 0.001).If not specified, analysis was performed with GRAPHPAD PRISM 8 (Boston, MA, USA).

Ethics approval
All procedures followed were in accordance with the ethical standards of the Animal Care and Use Committee of School of medicine, Shanghai Jiao Tong University.All institutional and national guidelines for the care and use of laboratory animals were followed.

PTEN promotes HR repair through facilitating DNA end resection
We first validated the role of PTEN in DSB repair.As shown in Fig. S1A-C, knockdown of PTEN in DU145 cells indeed delayed DSB repair after the treatment with Zeocin (a radiomimetic reagent), as measured by ionizing radiation-induced foci (IRIF) of cH2AX and 53BP1 [18,24].Either PTEN knockdown in DU145 and HeLa cells or PTEN knockout in MEFs reduced the numbers of RAD51 (a key HR repair regulator) foci (Fig. S1D-F), which was consistent with previous reports [18,21].In addition to RAD51 filament formation, DNA end resection, a key step prior to RAD51, is also critical for the choice of DSB repair pathway and can be regulated in many ways [41].To test whether PTEN regulates DNA end resection, we detected the phosphorylation level of RPA32(S4/8), a surrogate marker of ssDNA accumulation and DNA end resection [42], to show that knockdown of PTEN downregulated pS4/8-RPA32 in HeLa cells after irradiation (IR) treatment (Fig. 1A), suggesting that PTEN is involved in the regulation of DNA end resection in DSB repair.
To further identify whether both the phosphatase activity and SUMOylation of PTEN are required for its role in DNA end resection, we used the CRISPR/ Cas9 system to knockout PTEN in DU145 cells and then stably re-expressed PTEN WT , PTEN C124S (dual phosphatase dead), PTEN G129E (lipid phosphatase dead, but protein phosphatase still active), PTEN K254R (SUMO-site mutant) and PTEN K266R (SUMO-site mutant) (Fig. S1G).As expectedly, knockout of PTEN decreased IR-induced pS4/8-RPA32 in DU145 cells (Fig. 1B).There were little differences on pS4/8-RPA32 among PTEN WT , PTEN C124S and PTEN G129E after IR treatment; however, pS4/8-RPA32 was decreased in PTEN K254R and PTEN K266R compared to that in

C124S
PTEN WT (Fig. 1C).Similar results of PTEN K254R and PTEN K266R in decreasing pS4/8-RPA32 were also observed by using other DNA damage reagents including Zeocin and Camptothecin (CPT) in DU145 (Fig. S1H,I), and IR in PC3 cells (Fig. S1J).Thus, above results suggest that SUMOylation but not phosphatase activity of PTEN is associated with DNA end resection.
Determination of chromatin-associated proteins has been often applied to monitor DNA damage repair process, for examples, chromatin loading of RPA32 and RAD51 can represent HR efficiency [21,40,43,44].When compared with PTEN WT , mutants PTEN K254R and PTEN K266R inhibited the chromatin loading of RPA32 and RAD51 in DU145 cells induced by Zeocin (Fig. 1D), Etoposide (Fig. S1K) and CPT (Fig. S1L), whereas PTEN C124S and PTEN G129E seemed not affect (Fig. 1D).We also detected chromatin associated p-RPA32 with immunofluorescence after pre-extraction in DU145 cells and the results were consistent with western blot results (Fig. S1N).Furthermore, we also investigated IRIF of RAD51 in S phase (EdU + ) cells to show the similar pattern of results that numbers of RAD51 foci were comparable among PTEN C124S , PTEN G129E and PTEN WT , but decreased in PTEN K254R and PTEN K266R (Fig. 1E).These results suggest that SUMOylation but not phosphatase activity of PTEN affects the chromatin loading of RPA32 and RAD51.
In addition to RAD51 foci as a marker of HR repair, an HR reporter of DR-GFP was employed to detect the overall efficiency of HR repair [45].In accordance with previous reports, HR efficiency was reduced after PTEN knockdown (Fig. S1M).Surprisingly, HR efficiency was also compromised in PTEN C124S , as like PTEN K254R and PTEN K266R , but not in PTEN G129E , suggesting PTEN protein phosphatase was indispensable for it function in HR repair efficiency despite its little influence on chromatin loading of RPA32 and RAD51 (Fig. 1F).We also detected the NHEJ repair efficiency of these mutants with EJ5-GFP reporter [39].Knockdown of PTEN and re-expressing PTEN WT or mutants in U2OS cells did not change the NHEJ repair efficiency (Fig. S1O,P).Given cell cycle can regulate the choice of DNA damage repair pathway, we detected cell cycle profile and EdU incorporation in different human cancer cell lines used in this study.There is no obvious difference in cell cycle among parental, PTEN knockdown, PTEN knockout and PTEN reconstituted cells of HeLa and DU145 (Fig. S1R,S).We only observed a mild decrease of S phase cell population when overexpressing PTEN in PC3 cells, which might be due to downregulation of PI3K/AKT signaling (Fig. S1Q).There was no significant difference in percentage of EdU incorporation in HeLa, PC3, and reconstituted DU145-PTEN À/À cells (Fig. S1T-V).In summary, considering that HR repair mainly occurs in the S/G2 phase, the above results demonstrated that the difference in HR repair efficiency in these cells was not due to the influence of cell cycle or cell proliferation.Taken together, these data provide substantial evidences that PTEN promotes HR repair partially through enhancing DNA end resection, which is dramatically abolished when its SUMO-sites mutated.

DNA damage promotes PTEN chromatin loading by inducing its SUMOylation
As previous reported [29], DNA damage stimuli can strongly induce SUMOylation of proteins involved in different DNA damage repair pathways.To further investigate the induction and turnover of PTEN SUMO modification in DNA damage repair, we overexpressed His-SUMO1 and Flag-PTEN in 293T cells.After Zeocin treatment, cells were collected and lysed at different recovery time as indicated, and His-SUMO1 modified proteins were enriched with Ni 2+ -NTA agarose beads (Fig. 2A) and Co-IP (Fig. S2A) methods under denatured condition.Interestingly, SUMOylated PTEN was significantly increased overtime by Zeocin treatment.Similarly, CPT treatment also enhanced SUMOylation of PTEN (Fig. 2B).These data strongly demonstrated that DNA damage stimuli promoted PTEN SUMOylation.To further determine which SUMO-site is responsible for SUMO1 conjugation induced by DNA damage and whether its phosphatase activity is involved in this process, we overexpressed His-SUMO1 and Flag-tagged PTEN C124S , PTEN G129E , PTEN K254R or PTEN K266R in 293T cells and detected SUMOylated PTEN with the method of Ni 2+ -NTA agarose pull down.In contrast to mutations C124S and G129E, both mutations of K254R and K266R led to significant reduction of PTEN SUMOylation induced by Zeocin, suggesting that these two SUMO-sites were critical for DDR (Fig. 2C).
One previous study reported that K254R mutation prevented PTEN nuclear localization and DNA damage repair function in U87MG glioblastoma cells [18], however our results of immunofluorescence showed no changes in the localization between PTEN WT and PTEN K254R or PTEN K266R in DU145 cells under normal condition (Fig. 2D).We also observed there was little effects on PTEN nuclear localization in DU145 cells after IR treatment (Fig. S2C).Consistently, the detection of the nuclear PTEN level in non-treated and IR-treated DU145 cells with nuclear-cytosol fractionation showed the same result that IR treatment had no effect on PTEN nuclear localization (Fig. S2D).Furthermore, we detected the influence of IR on subcellular localization of PTEN in DU145-shPTEN-PTEN WT/K254R/K266R cells with immunofluorescence.After IR treatment, there was still no difference in PTEN localization between PTEN WT , mutant PTEN K254R , and PTEN K266R (Fig. S2E).Moreover, another two nuclear-cytosol separation results also revealed that the localizations of PTEN were almost not affected in DU145-PTEN À/À re-expressing PTEN-WT and mutants including PTEN K254R , PTEN K266R , PTEN C124S , and PTEN G129E cells even after treatments with Zeocin (Fig. 2E) and CPT (Fig. S2B).More interestingly, the nuclear location of PTEN C124S was markedly higher than that of all others (Fig. 2E and Fig. S2B), the underlying mechanism should be further studied.Although SUMO-site mutations K254R and K266R had little influence on PTEN nuclear localization, we wanted to assess whether SUMOylation of PTEN influences its chromatin loading under DNA damage.First, we observed that PTEN was indeed recruited into the DNA-damage location induced by the laser microirradiation (Fig. S2F).Second, Zeocin (Fig. 2F) and CPT (Fig. S2G) treatments induced PTEN accumulation on the chromatin.The chromatin loading of PTEN K254R was significantly suppressed whereas that of PTEN G129E and PTEN K266R was similar with PTEN WT after Zeocin and CPT treatments.For the case of PTEN K266R , it was unexpectedly and might be a different mechanism from PTEN K254R .In accordance with enhanced nuclear localization of PTEN-C124S, the chromatin loading of PTEN C124S was remarkably increased (Fig. 2F), but how the mutation C124S to increase the nuclear localization and chromatin loading of PTEN was not clear.Third, to further validate whether SUMOylated PTEN can directly accumulate on the chromatin, we separated chromatin associated proteins under harsh condition.After IR treatment, one shifted band with higher molecular weight than normal PTEN was clearly observed in re-expression of PTEN WT but not SUMO-site mutants PTEN K254R and PTEN K266R in DU145-PTEN À/À cells (Fig. 2G).Given that endogenous PTEN gene in DU145 cells was knocked out with CRSPR/Cas9 technique, this shift band from re-expression of normal PTEN WT was most likely to be SUMOylated-PTEN other than variants of PTEN such as PTEN a or b isoform.
Collectively, all above results demonstrate that DNA damage promotes PTEN SUMOylation and Cytosol(C) separation was performed in DU145-PTEN À/À cells stably re-expressing PTEN WT, C124S, G129E, K254R and K266R treated with Zeocin (400 lgÁmL À1 ) for 1 h and recovery for 4 h.Localization of PTEN was detected with immunoblot.(F) Immunoblot of chromatin associated PTEN in DU145 cells treated with Zeocin (400 lgÁmL À1 ) for 1 h and recovery for indicated time.(G) Chromatin protein separation were performed at harsh condition in DU145-PTEN À/À cells stably re-expressing PTEN WT, K254R and K266R after treatment with 10 Gy.PTEN tightly associated with chromatin was detected with immunoblot.All results were shown with one representative image from three independent experiments.Chro., chromatin; CPT, camptothecin; WT, wide type.especially K254-SUMOylation of PTEN is essential for its chromatin loading.

p14ARF is a novel SUMO E3 ligase to mediate PTEN SUMOylation during DDR
p14ARF, a well-known tumor suppressor, is an atypical SUMO E3 ligase for promoting SUMOylation of its binding proteins such as MDM2, NPM, and EGR1 [37,46,47].Interestingly, we found that PTEN SUMOylation was dramatically enhanced by overexpression of p14ARF and UBC9 (SUMO-conjugating enzyme E2) (Fig. 3A and Fig. S3A) whereas suppressed by knockdown of p14ARF (Fig. 3B).Since p14ARF promotes SUMOylation of target proteins via direct interaction, we assumed p14ARF might interact with PTEN.Co-IP and GST-pull down results showed that p14ARF directly bound to PTEN in cells and in vitro (Fig. 3C,D and Fig. S3B).To identify the interaction region between p14ARF and PTEN, we generated a series of truncates (Fig. S3C) and performed co-IP.Two regions of 2-14aa and 82-101aa in p14ARF are important for substrate interaction, and deletion of these two regions dramatically inhibit SUMO conjugation on substrates mediated by p14ARF [48].Our results also showed that these two regions in p14ARF were required for their interaction.Deletion of either one reduced their interaction and deletion of both completely abolish their interaction (Fig. 3E and Fig. S3D,E).Moreover, the ability of p14ARF to enhance PTEN SUMOylation was indeed weakened when deletion of either one region.Although both regions contributed to PTEN SUMOylation, p14ARF(D2-14) seemed to be more effective than p14ARF(D82-101) in suppression of PTEN SUMOylation (Fig. 3F).In addition, we identified that the C-terminal region 188-403aa of PTEN mediated its interaction with p14ARF (Fig. S3F).
Since DNA damage stimuli can induce PTEN SUMOylation, we wondered whether p14ARF is involved in this process.Knockdown of p14ARF substantially reduced Zeocin-induced PTEN SUMOylation (Fig. 3G).In response to treatments with both Zeocin (Fig. 3H) and CPT (Fig. S3G), the interaction between exogenous PTEN and p14ARF was obviously enhanced.CPT treatment also significantly increased the interaction between endogenous PTEN and p14ARF in DU145 cells (Fig. S3H).Given that phosphorylation signal is critical for DDR and there is exactly one threonine (T8) located in p14ARF(2-14aa), which can be phosphorylated [49], so we detected the phosphorylation of p14ARF with the method of Phostag gel and showed a clear shifted band of p14ARF, which was increased after Zeocin treatment (Fig. 3I), suggesting that DNA damage induced phosphorylation of p14ARF.The mutation T8A at p14ARF significantly inhibited the interaction between PTEN and p14ARF, on the contrary, the mutation T8D enhanced their interaction not only under normal condition but also after treatment with CPT (Fig. 3J).The mutation T8A of p14ARF also attenuated its ability to promote PTEN SUMOylation after Zeocin treatment (Fig. 3K).Consistently, knockdown of p14ARF inhibited PTEN chromatin loading while overexpression of p14ARF increased its loading (Fig. 3L and Fig. S3I-K).Knockdown of p14ARF also sensitized DU145 cell to  Cisplatin (Fig. S3L).All these results suggest that p14ARF is a functional SUMO E3 ligase responsible for promoting SUMOylation of PTEN during DDR.

PTEN relieves HR repair barrier posted by 53BP1 through directly dephosphorylating pT543-53BP1
Given that HR repair efficiency is compromised when PTEN lacking protein phosphatase activity [18] (Fig. 1F), next we tried to identify potential protein substrates targeted by PTEN in DNA damage repair.53BP1, a key negative regulator of HR repair, is phosphorylated at multiple S/TQ sites which are essential for recruitment of downstream effectors [7].Thus, we first examined the phosphorylation levels of 53BP1 in cells after treatment with CPT or Zeocin.The pT543-53BP1, which is necessary for recruitment of RIF1, was dramatically increased in PTEN-knockdown DU145 and HeLa cells (Fig. 4A, left panels) whereas significantly decreased in PTEN overexpression in PC3 cells which are devoid of endogenous PTEN (Fig. 4A, right and upper panel) after treatment with CPT (for 30 min).Moreover, we also detected the pT543-53BP1 level in PTEN-knockout DU145 cells after treatment with higher concentration of CPT for 1 h, to show that the pT543-53BP1 level was higher and lasted longer in PTEN-knockout cells than that of PTEN-WT cells (Fig. 4A, right and lower panel).Consistent with above results, when treated with Zeocin, the pT543-53BP1 level was increased in PTEN-knockdown DU145 and HeLa cells (Fig. S4A), while weakened in PTEN overexpression in PC3 cells (Fig. S4B).In addition, we confirm that the intensity of pT543-53BP1 foci was stronger in PTEN-knockdown DU145 than control cells (Fig. 4B and Fig. S4C) by staining with antibody pT543-53BP1.
Since it has been reported that PTEN loss led to the increase of pS25/29-53BP1 after treatment with etoposide and the mutation S25A enhanced RIF1 recruitment during DNA damage repair [24,50]; thus, we also detected the dynamics of pS25/29-53BP1 in PTENknockdown or -knockout cells after treatment with CPT or Zeocin.Indeed, pS25/29-53BP1 was induced in both PTEN-WT and PTEN-knockout DU145 cells after treatment with CPT.Increased pS25/29-53BP1 level was the same and even a little down whereas as expectedly, increased pT543-53BP1 level was higher in PTEN-knockout DU145 when compared with PTEN-WT DU145 (Fig. 4C).Furthermore, after treatment with Zeocin, the number and intensity of pT543-53BP1 foci were expectedly increased in DU145-shPTEN cells compared to those in DU145-pLKO cells (Fig. 4D); in contrast, there was little difference in the number and intensity of pS25-53BP1 foci between DU145-pLKO and DU145-shPTEN cells (Fig. 4E).These results demonstrate that PTEN regulates the level of pT543-53BP1 other than pS25/29-53BP1 in DDR.
As pT543-53BP1 helps to recruit RIF1 to DNA breaks [7], so we stained RIF1 in DU145-pLKO and shPTEN cells after IR treatment.In accordance with the pT543-53BP1 levels, the number and intensity of RIF1 foci were both increased (Fig. 4F).It has been well-documented that the 53BP1-RIF1-shieldin axis forms a barrier to inhibit HR repair through suppressing DNA end resection and HR repair mainly occurs at S/G2 of cell cycle [9,10,51,52], we wondered that whether PTEN in regulation of pT543-53BP1 is dependent on cell cycle.DU145 cells were firstly synchronized at G1/S with double thymidine block, and then directly released 5 h to enter S phase or synchronized at G1 with lovastatin, respectively.Cyclin A2 was used to show synchronization efficiency (Fig. S4D).The pT543-53BP1 levels were enhanced in both G1 and S (G) Immunoblot analysis of pT543-53BP1 level in DU145-PTEN À/À cells stably re-expressed PTEN-WT, C124S and G129E after 10 Gy treatment and recovery for indicated time.(H) Dot graph of pT543-53BP1 foci number and intensity in DU145-PTEN À/À cells stably reexpressed PTEN-WT, C124S and G129E after 5 Gy treatment and recovery for 2 h.(I) in vitro phosphatase assay.GFP tagged phosphorylated 53BP1(1-600) were purified from 293T cells.GST-PTEN WT , C124S and G129E were purified from E. coli BL21.Unpaired Student's t-test was used (**P < 0.01, ***P < 0.001) and data were shown as mean.n.s., not significant.All results were shown with one representative image from three independent experiments.CPT, camptothecin; GST, glutathione S-transferase tag; NC, negative control; WT, wide type.phases after treatment with Zeocin (Fig. S4E,F).However, the increased pT543-53BP1 levels in S phase were much weaker in PTEN WT cells than those PTEN À/À cells (Fig. S4E).In contrast, the increased pT543-53BP1 levels in G1 phase were slightly weaker in PTEN WT cells when compared with PTEN À/À cells (Fig. S4F).These data suggest that PTEN regulating pT543-53BP1 is a cell-cycle dependent manner and mainly occurs in S phase.
To further verify whether PTEN phosphatase is responsible for dephosphorylating pT543-53BP1, we determined the pT543-53BP1 levels in PTEN À/À , PTEN WT , PTEN C124S and PTEN G129E DU145 cells.After treatment with IR or Zeocin, the pT543-53BP1 levels in PTEN À/À cells were higher than those in PTEN WT cells, as was expected (Fig. 4G and Fig. S4G, top panels).We found that pT543-53BP1 was increased in PTEN C124S but not in PTEN G129E cells when compared with PTEN WT cells (Fig. 4G and Fig. S4G, middle and low panels).As similar to immunoblot results, the number and intensity of pT543-53BP1 foci were also increased in PTEN C124S but not in PTEN G129E cells compared to those in PTEN WT DU145 cells at 2 h after treatment with IR (5 Gy) (Fig. 4H and Fig. S4H).Above results indicated that PTEN protein phosphatase was required for dephosphorylation of pT543-53BP1.Thus, we further speculated whether PTEN directly dephosphorylates pT543-53BP1.
To verify above hypothesis, we first purified phosphorylated full-length Flag-53BP1 from 293T cells for the in vitro reaction with GST-PTEN WT or GST-PTEN G129R (a dual phosphatase deficient mutant).The following western blotting results showed that PTEN-WT but not PTEN-G129R efficiently dephosphorylated pT543-53BP1 (Fig. S4I), suggesting that the PTEN phosphatase activity is required for dephosphorylation of pT543-53BP1.However, we noticed that Flag-53BP1 partly degraded after the in vitro reaction with GST-PTEN WT other than GST-PTEN G129R .Given that 53BP1 consists of 1972 aa which is a relatively high-molecular-weight protein and could be much easier degraded when purified for in vitro reaction, we constructed a short truncated form, GFP-53BP1(1-600), for in vitro phosphatase assay.We found that GFP-53BP1(1-600) was easily detected by antibody anti-pT543-53BP1 and more strongly after CPT treatment; on the contrary, GFP-53BP1(1-300) could not be detected (Fig. S4J), suggesting this antibody could still specifically recognize phosphorylated T543 in GFP-53BP1(1-600).In order to further distinguish protein phosphatase and lipid phosphatase of PTEN in dephosphorylation of pT543-53BP1, similarly, we purified phosphorylated GFP-53BP1(1-600) from 293T cells for the in vitro reaction with GST-PTEN WT , GST-PTEN C124S or GST-PTEN G129E .Consistent with cellular results, PTEN C124S lost the ability to dephosphorylate pT543-53BP1, whereas PTEN G129E could moderately dephosphorylate pT543-53BP1 but was less efficient than PTEN-WT, which might be because the G129E mutation not only abolished lipid phosphatase of PTEN, but also reduced its protein phosphatase by about a half [53] (Fig. 4I).However, we still noticed that GFP-53BP1(1-600) partly degraded after the in vitro reaction with GST-PTEN WT and GST-PTEN G129E , but not GST-PTEN C124S , suggesting that phosphorylation modification might be important to maintain 53BP1 protein stability in vitro.Thus, above results suggest that PTEN directly dephosphorylates pT543-53BP1 in vitro and in DDR.
Taken together, our data demonstrate that PTEN directly dephosphorylates pT543-53BP1 in response to DNA damage, which relieves HR repair barrier posted by 53BP1.

PTEN chromatin loading is mediated by BRCA1 recruiting SUMOylated PTEN via its Nterminal SIM
BRCA1 can remove HR repair barrier posted by 53BP1 via several molecular mechanisms, including Fig. 5. PTEN chromatin loading is mediated by BRCA1 recruiting SUMOylated PTEN via its N-terminal SIM.(A) Immunoblot of pT543-53BP1 in DU145 cells in which BRCA1 were knockdown with siControl or siBRCA1-1 after Zeocin (400 lgÁmL À1 ) treatment for 1 h and recovery for indicated time.(B) Immunoblot of chromatin loaded PTEN and pS4/8-RPA32 in BRCA1 knockdown DU145 cells after treatment with CPT (20 lM) for 1 h and recovery for indicated time.(C) 293T senp1À/À cells were transfected with Myc-BRCA1, HA-UBC9, His-SUMO1 and Flag-PTEN for 48 h.Co-IP were performed to identify interaction between PTEN and BRCA1.(D) 293T senp1À/À cells were transfected with Myc-BRCA1, HA-UBC9, His-SUMO1 and Flag-PTEN (WT or mutants) for 48 h and treated with Zeocin.Co-IP were performed to identify interaction between PTEN (WT or mutants) and BRCA1.(E) 293T senp1À/À cells were transfected with Myc-BRCA1(WT or SIM mutants), HA-UBC9, His-SUMO1 and Flag-PTEN for 48 h.Co-IP were performed to identify interaction between PTEN and BRCA1 (WT or SIM mutants).(F) Immunoblot of pT543-53BP1 in DU145-PTEN À/À cells stably re-expressed PTEN-WT, K254R and K266R after treatment with Zeocin (400 lgÁmL À1 ) for 1 h and recovery for indicated time.All results were shown with one representative image from three independent experiments.Chro., chromatin; CPT, camptothecin; WCL, whole cell lysis; WT, wide type.
suppression of 53BP1 phosphorylation induced by DNA damage [14], so we wondered whether PTEN is a downstream effector of BRCA1.As expectedly, knockdown of BRCA1 by siRNA strongly increased the pT543-53BP1 level in DU145 after Zeocin treatment (Fig. 5A).Consistent with this, knockdown of BRCA1 by either siRNA or shRNA displayed the same results of enhancing pT543-53BP1 in U2OS cells after treatment with Zeocin and Etoposide (Fig. S5A).For the other hand, we questioned whether BRCA1 is involved in PTEN chromatin loading.Indeed, chromatin separation experiments showed that knockdown of BRCA1 by shRNA or siRNA significantly suppressed PTEN chromatin loading as well as pS4/8-RPA32 (as a positive control) in DU145 cells after treatment with CPT, indicating that BRCA1 was needed for PTEN chromatin loading during DDR (Fig. 5B and Fig. S5B).Therefore, we next tested whether PTEN interacts with BRCA1 and this can be enhanced by SUMOylation.We transfected Flag-PTEN and Myc-BRCA1 with SUMO1 into Senp1 À/À (SUMO Specific Peptidase 1) 293T cells for 48 h, and then treated with or without CPT.Co-IP/western blotting results showed that PTEN interacted with BRCA1, and the interaction was moderately enhanced by CPT treatment.Most strikingly, the interaction was strongest when cotransfected with SUMO1 and UBC9, which was not increased any more even with CPT treatment (Fig. 5C).Furthermore, CPT greatly increased the interaction between endogenous PTEN and BRCA1 in DU145 cells (Fig. S5C).To identify whether the interaction is specifically enhanced by SUMO1 modification, we also detected the ability of other SUMO isoform SUMO2 whose amino acid sequence is a little different from SUMO1.The interaction was much weaker in SUMO2 transfected than that in SUMO1 transfected cells under CPT treatment.Surprisingly, cotransfected with SUMO2 and UBC9 did not enhance the interaction at all (Fig. S5D).These data suggest that the interaction between PTEN and BRCA1 is specifically promoted by SUMO1 modification.
Given that the SUMO-site mutation of PTEN inhibited its chromatin loading induced by DNA damage, we wondered whether the interaction between PTEN and BRCA1 is directly mediated by SUMO-SIM (SUMO interacting motif), which is an important mechanism to mediate the protein-protein interaction.Indeed, the double mutations K254/266R of PTEN remarkably repressed the interaction although the single mutation K254R or K266R did not inhibit, suggesting SUMO1 modification of PTEN was important for its interaction with BRCA1.Additionally, the lack of lipid phosphatase activity of mutants PTEN C124S and PTEN G129E did not affect the interaction (Fig. 5D).
To identify SIMs of BRCA1 which are responsible for the interaction with SUMO1 attached to PTEN, we used two software GPS-SUMO and JASSA [54,55] to predict possible SIMs of BRCA1 (Fig. S5E,F).Since depletion of exon11 of BRCA1 abolishes its suppression on 53BP1 phosphorylation and DNA end resection [15,16], we mainly focused on SIMs located in this region, which are marked in red (Fig. S5E,F).Further to find out which SIM is essential for the interaction, we mutated amino acids of SIM1, 2, 3, 4, into alanine, referred as mSIMn, and deleted SIM5-1 and SIM5-2, referred as MSIM5, respectively (Fig. S5G).Co-IP/western blotting results showed that the interaction was efficiently weakened by mSIM2 ( 412 VLDVL 416 --AAAAA) but not by other mutants of BRCA1 (Fig. 5E).
Since the interaction between PTEN and BRCA1 was dependent on SUMO-SIM, we wondered that the SUMO-site mutant PTEN is also be defective in dephosphorylation of 53BP1.The pT543-53BP1 level was relatively higher in PTEN K254R cells than that in PTEN WT DU145 cells after treatment with Zeocin.However, the pT543-53BP1 level in PTEN K266R cells was comparable with that in PTEN WT cells (Fig. 5F).These suggest that chromatin loading of PTEN mediated by K254-SUMO1 modification is also very important for its role in dephosphorylation of pT543-53BP1.All above results demonstrate that BRCA1 recruits SUMOylated PTEN to chromatin via its N-terminal SIM, thereby dephosphorylating pT543-53BP1 in DDR.

HR repair is impaired by PTEN K254R in vivo
To verify whether SUMO site mutated PTEN impairs HR repair in vivo, we generated knock-in mice with a point-mutation of Pten K254R or Pten K266R by using CRISPR-Cas9 technique.Pten K254R mice developed normally and was indistinguishable to Pten WT mice, but interestingly, Pten K266R mice exhibited low birth rate and some of them show abnormal localization of seminal vesicle.Mouse embryonic fibroblasts isolated from Pten WT and Pten K254R mice were verified by DNA sequencing and used to examine DNA damage repair efficiency (Fig. S6A).Compared with Pten WT MEFs, Pten K254R MEFs showed significant increase of cH2AX and 53BP1 foci (Fig. 6A,B and Fig. S6B,C) but decrease of RPA32 and RAD51 foci (Fig. 6C,D and Fig. S6D,E) after treatment with IR, indicating Pten K254R MEFs were insufficient in HR repair.The levels of cH2AX were also higher in Pten K254R MEFs compared to those in Pten WT MEFs in the late stage after CPT or Zeocin (Fig. S6F).Furthermore, the pS4/8-RPA32 levels were much weaker in Pten K254R MEFs than those in Pten WT MEFs after treatment with IR of 20 Gy or 30 Gy (Fig. 6E and Fig. S6G), which proved that PTEN K254R function in regulation of DNA end resection during HR repair was indeed compromised.We also detected pS4/8-RPA32 levels in primary Pten K254R MEFs and Pten WT MEFs after 30 Gy, and the result was similar with immortalized MEFs (Fig. S6H).Consistent with results of tumor cell lines, PTEN chromatin loading induced by IR and CPT was also inhibited in Pten K254R MEFs (Fig. 6F and Fig. S6I).Chromatin-bound RPA32 was also decreased in Pten K254R MEFs in response to IR treatment, indicating reduced DNA end resection efficiency (Fig. 6F).Collectively, these data confirm that PTEN K254R impairs HR repair by decreasing its chromatin loading and DNA end resection.
To validate that K254-SUMOylation is essential for PTEN loading to the chromatin to promote HR repair, we isolated the chromatin for analysis of potential PTEN SUMOylation, showing that a shifted 90-kDa which could be SUMOylated PTEN band, was enhanced in Pten WT but not in Pten K254R MEFs at 4, 8 h after Zeocin treatment (Fig. S6J).One study reported that K254R mutation may affect the subcellular localization of PTEN [18], however our nuclearcytosol separation results showed there were no any differences between Pten WT and Pten K254R MEFs under normal condition, even after treatments with IR, CPT or Zeocin (Fig. 6G and Fig. S6K,L).Moreover, immunofluorescence staining of PTEN also displayed not much differences in subcellular localization between Pten WT and Pten K254R under both normal condition and IR treatment (Fig. 6H).To exclude the impact of cell cycle on DNA damage repair, we also detected the cell cycle profile in primary and immortalized MEFs.There was little differences between Pten WT and Pten K254R in both primary and immortalized MEFs (Fig. S6M).
To compare the protective effect of Pten WT and Pten K254R mice against DNA damage under physiological condition, villus length and proliferation of intestinal crypts cells, which are highly sensitive to IR due to their rapid turnover rate [21,34], were determined after treatment with whole-body IR.Pten WT mice had no significant difference in crypts morphology and villus length between NT (no treatment) and IR (8 Gy) group, which indicated Pten WT mice completely recovered, whereas Pten K254R mice showed significant destructed crypts and length-shortened villi 4 days after treatment with IR (Fig. 6I).Similarly, there was no significant difference in numbers of ki67-positive cells between Pten K254R and Pten WT mice under no treatment, while numbers of ki67-positive cells in intestine crypts of Pten K254R mice were less than those of Pten WT mice after IR treatment, suggesting reduced proliferating cells of intestine crypts might be due to functional deficiency in DDR (Fig. 6J).These results proved K254-SUMOylation was essential for PTEN mediating DNA damage repair and irradiation protection in vivo.Lastly, we tested the sensitivity of Pten WT and Pten K254R MEFs to DNA damage reagents, and found that Pten K254R MEFs were more sensitive to Cisplatin, CPT and Zeocin than Pten WT MEFs in a dose-dependent manner (Fig. 6K).Taken together, above results demonstrate that K254-SUMOylation of PTEN is required for PTEN mediating HR repair in DDR in vivo.

Blocking PTEN SUMOylation pathway sensitizes tumor cells to DNA damage reagents
We first determined appropriate concentrations of DNA damage reagents including Cisplatin, Zeocin and CPT for the treatment on DU145-PTEN WT or DU145-PTEN À/À cells, and found that there was little difference in survival colony numbers between DU145-PTEN À/À and DU145-PTEN WT cells when treated with low doses of Cisplatin (0.5 and 1 lM), Zeocin (2 and 5 lgÁmL À1 ) and CPT (25 and 50 nM), whereas DU145-PTEN À/À cells were much more sensitive to high doses of Cisplatin (1.5 and 2 lM), Zeocin (10 and 15 lgÁmL À1 ) and CPT (150 nM) than PTEN WT (Fig. 7A-C and Fig. S7A-C).
Given that SUMOylation signaling is critical for cell survival during DNA damage and a specific SUMO E3 inhibitor is lacking, we tried to assess the cellular sensitivity to DNA damage agents after blocking SUMOylation pathway with SUMO E1 inhibitor, TAK981 [56], which has been in clinical trial to treat advanced and metastatic solid tumors.In accordance with previous report [57], the total levels of SUMOylation in DU145 cells were inhibited by TAK981 in a dose dependent manner and completely suppressed at the concentration greater than 100 nm (Fig. S7D).We found that TAK981 effectively inhibited the growth/ survival of DU145-PTEN À/À , -PTEN WT , -PTEN K254R and -PTEN K266R cells, although there was no much difference among PTEN WT , PTEN K254R and PTEN K266R cells either with or without TAK981 treatment (Fig. 7D and Fig. S7E).By assessing combination effects of TAK981 with Cisplatin, Zeocin or CPT, we found that both PTEN K254R and PTEN K266R cells were more sensitive compared with PTEN WT cells (Fig. 7E-G and Fig. S7E), which was consistent with our results that PTEN SUMO site mutations of K254R and K266R decreased DNA damage repair efficiency (Fig. 1C,D).Most strikingly, combination treatment with TAK981 resulted in an additive effect on the sensitivity to Cisplatin, Zeocin or CPT (Fig. 7E-G and Fig. S7E).These results indicated that inhibiting the SUMOylation pathway might increase killing efficiency of DNA damage reagents in clinical cancer treatment.Since p14ARF knockdown inhibited chromatin loading of PTEN and sensitized DU145 cells to DNA damage (Fig. 3K and Fig. S3H-K), we further synthesized a small peptide called CPP-p14ARF(2-13), which was a Fitc-labeled cell penetrating peptide (CPP) YGRKKR RQRRR fused by p14ARF(2-13aa)/VRRFLVTLRIRR (Fitc-YGRKKRRQRRRVRRFLVTLRIRR), the main region interacting with and promoting PTEN SUMOylation (Fig. 3F).We assessed whether CPP-p14ARF(2-13) can interfere PTEN SUMOylation and loading to the chromatin under DNA damage.Indeed, CPP-p14ARF(2-13) dramatically inhibited the interaction between PTEN and p14ARF (Fig. S7F) as well as chromatin loading of PTEN (Fig. S7G) under CPT treatment.Next, we tested whether CPP-p14ARF(2-13) enhances the cell sensitivity to DNA damage.As previously reported that a peptide of p14ARF(1-22aa) inhibited cell proliferation [58], so we also detected whether CPP-p14ARF(2-13) influences it.There was no difference in survival colony numbers between PTEN À/À and PTEN WT DU145 cells treated with CPP-p14ARF(2-13) in different doses, suggesting CPP-p14ARF(2-13) did not affect cell proliferation (Fig. 7H and Fig. S7H).Interestingly, the cell sensitivity to Cisplatin, Zeocin or CPT treatment was increased by combination with CPP-p14ARF(2-13) (Fig. 7I-K and Fig. S7H).The low dose of CPP-p14ARF(2-13) at 2 lM was capable to enhance killing efficiency for Cisplatin and Zeocin treatments to DU145-PTEN WT cells (Fig. 7I,J), and the high dose at 10 lM also increased killing efficiency for CPT treatment (Fig. 7K).Furthermore, CPP-p14ARF (2-13) treatment also decreased survival rate of DU145-PTEN À/À cells when combined with these reagents, suggesting CPP-p14ARF(2-13) might block the interaction of p14ARF with other targets besides PTEN.Thus, above results suggest that TAK981 or CPP-p14ARF(2-13) has the potential to enhance the effect of DNA damage reagents in killing tumor cells.

Discussion
Post-translational modifications of PTEN including SUMOylation [18], methylation and phosphorylation [24] are involved in DDR.Although one study reported that PTEN K254R impaired HR repair efficiency, the underlying molecular mechanism remained largely elusive [18].Our data suggested that DNA damage induced PTEN SUMOylation in a timedependent manner other than rapid activation like phosphorylation.Both K254 and K266 of PTEN could be conjugated with SUMO1, but one single mutation was enough to reduce DNA damage-induced PTEN SUMOylation.Interestingly, the mutation K254R, but not K266R, suppressed DNA damagetriggered PTEN chromatin loading.It is an open question how K266-SUMOylation of PTEN to participate in DDR.Given that K266 has also been identified as a ubiquitination site [59], it might exist a crosstalk between SUMOylation and ubiquitination of PTEN during DDR.Moreover, SUMOylation of PTEN on K266 promotes its association with cell membrane and which in turn inhibits PI3K-AKT signaling [25].As PI3K-AKT pathway can be also activated by DNA damage stimuli and regulated function of DDR related proteins [60][61][62][63], K266-SUMOylation of PTEN might also promote DDR through the PI3K-AKT pathway.Additionally, the mutation C124S remarkably enhanced PTEN nuclear localization and chromatin loading.Due to PTEN interacting with RAD51 and RPA32 [19,23], this might explain why PTEN C124S decreased HR repair efficiency but not affected the chromatin loading of RAD51 and RPA32.It remains to be further explored how C124S influences PTEN localization and whether the chromatin trapped PTEN C124S has other additive side effects on DNA damage repair.
Several SUMO E3 ligases including PIAS1, PIAS4, CBX4 and ZNF451 play an important role in regulation of SUMOylation signal transduction in response to DNA damage [32,33,64].Here, we identified an atypical SUMO E3 ligase, p14ARF, responsible for PTEN SUMOylation under DNA damage such as IR, CPT and Zeocin through their interaction, which was enhanced by DNA damage-induced phosphorylation of p14ARF.UV stress disrupts the p14ARF-B23 interaction in the nucleolar, resulting in a transient translocation of p14ARF to the nucleoplasm [65].In addition, ATM, a main DDR kinase, promotes the release of p14ARF from the nucleus and subsequent degradation after doxorubicin treatment [66].So, the sub-nuclear distribution and protein-protein interactions of p14ARF are critical for its function in promoting PTEN SUMOylation after DNA damage.
As for the role of PTEN protein phosphatase in HR repair, the conclusions from two different groups are contradictory [21,24].Our results supported that PTEN protein phosphatase was necessary for HR repair.In addition to cH2AX as a substrate of PTEN during DDR [24], we identified a new target, 53BP1, which was directly dephosphorylated by PTEN in cells and in vitro.More interestingly, PTEN selectively dephosphorylated pT543-53BP1 other than pS25-53BP1 in cells.It has been early reported that phosphorylation at 7S/TQ sites in the N-terminal region of 53BP1 is responsible for its recruitment of RIF1, whereas phosphorylation at the other 8S/TQ sites of 53BP1 is essential for PTIP accumulation at DNAbreak sites [7].And interestingly, as long as one site of 7S/TQ is phosphorylated, it is sufficient for 53BP1 recruiting RIF1 [14].However, most recently one study reported that RIF1 is recruited to IR-induced foci by recognizing three related phosphorylated epitopes on 53BP1 [67].Thus, PTEN might also target phosphorylation at other sites responsible for RIF1 recruitment, besides pT543-53BP1 that is one of 7S/TQ, during DDR.In addition, it has been reported another phosphatase, PP4C, dephosphorylates 53BP1 and promotes HR repair [14].Given that knockout of PP4C or PTEN can upregulate the phosphorylation of 53BP1, both PP4C and PTEN are probably necessary rather than redundant for inhibiting the phosphorylation of 53BP1.However, it is not yet clear whether the function of PP4C or PTEN is specific to cell type or DNA damage.More efforts are needed to further differentiate their roles in this process.
As known that BRCA1 is critical for HR repair by against 53BP1 posted barrier through serval molecular mechanisms, including inhibition of 53BP1 phosphorylation [14].Our results supported that PTEN was a downstream effector of BRCA1 and PTEN SUMOylation was required for their interaction, which was efficiently decreased by double mutation of K254/266R.One SIM located in the N terminus of BRCA1 was essential for recognition of SUMO1 conjugated to PTEN, by which PTEN was subsequently recruited by BRCA1 to DNAbreak sites.This might partially explain why that the deletion of exon11 of BRCA1 resulted in loss of BRCA1 function in inhibition of phosphorylation of 53BP1 [15,16].Moreover, as 53BP1 is pro-choice for DNA breaks and BRCA1 complex can relocate into the core of 53BP1 foci at S/G2 in a time dependent manner [68,69], so we speculate that after entering the core, BRCA1 recruits PTEN to directly dephosphorylate 53BP1, thus releasing downstream effectors of 53BP1 such as RIF1 and PTIP, which further facilitates DNA end resection and ongoing of HR repair.
The in vivo results from Pten K254R mice validated that PTEN SUMOylation promoted HR repair.Significantly, we did not observe the mutation K254R influence PTEN nuclear localization [18], by in vivo and in vitro results, which is consistent with our early report [25].However, our data strongly demonstrated that SUMOylation of PTEN was induced by several DNA damage agents and this sub-pool of SUMOylated PTEN was tightly associated with chromatin via BRCA1.It is worth noting that Bassi et al. [18] claimed that K254-SUMOylated-PTEN was decreased and excluded from the nucleus upon DNA damage, which was totally inconsistent with our observations that SUMO-mutant K254R had no differences in the localization.We have analyzed the possible reasons for these discrepancies as follows.Firstly, in addition to our model, several studies have reported that PTEN nuclear localization or chromatin association is important for DNA damage repair.For examples, Hou et al. [70] showed that Grb2 mediates PTEN nuclear translocation to repair H 2 O 2 -induced DNA damage in HeLa cells.Chen et al. [71] found that the activation of ATM by DNA damage reagents phosphorylates PTEN at Ser113, which promotes PTEN nuclear retention in HeLa and A549 cells.Ma et al. [21] found that FGFR phosphorylates PTEN at Y240 to facilitate PTEN nuclear localization and recruitment onto chromatin after IR treatment.Zhang et al. [24] also presented evidences that the phosphorylation at T398 of PTEN induced by DNA damage is recognized by MCD1 to promote PTEN chromatin loading.All above studies suggest that PTEN nuclear localization or chromatin association is a prerequisite to perform its function during DNA damage repair.However, Bassi et al. showed that re-expressed PTEN in U87MG, which is a PTEN-deficient gliomas cell line, was excluded from the nucleus after IR treatment.This was inconsistent with our observation in DU145 that is a prostate cancer cell line with the expression of endogenous PTEN.We guess the exclusion of exogenous PTEN in U87MG cells after IR might be a cell type-specific phenotype.Second, Bassi et al. also claimed that SUMOylated-PTEN, which was mostly localized in the nucleus in basal conditions, was excluded from the nucleus after DNA damage.They conducted the nuclear-cytosol fractionation to detect endogenous PTEN in HeLa cells after IR treatment and found that the nuclear PTEN was significantly decreased.It was very interesting that two close bands higher than normal PTEN were observed, which they thought as SUMOylated PTEN.However, these two bands are more likely to be the other two isoforms, PTENa and PTENb [72], which are both recently discovered.Third, for the nuclear-cytosol separation assay, nuclear fraction consists of both soluble and precipitated parts.Soluble part contains soluble proteins, while precipitated part is mainly made up of chromatin and its associated proteins.Detailed nuclear-cytosol separation method was not included in their paper.In our results, SUMOylated PTEN was tightly bound with chromatin and mainly in the nuclear precipitates during DNA damage repair.Therefore, another possible explanation is that they might detected only soluble part of nuclear fractions but not whole or precipitated fractions, which resulted in decreased level of SUMOylated PTEN in their observation.In all, the discrepancies between these results needs to be further explored.More efforts should be made to analyze the SUMOylation of PTEN at K254 in different cellular context during DNA damage repair.
Since we proved that activation of SUMOylation pathway was essential for correct DNA damage repair, so the combination of SUMOylation inhibitor with DNA damage reagents might enhance sensitivity of tumor cells to chemotherapy.As expectedly, the SUMOylation inhibitor TAK981 remarkedly increased killing efficiency of DNA damage reagents.Furthermore, the small peptide CPP-p14ARF(2-13) also suppressed DNA damage-induced chromatin loading of PTEN and sensitized tumor cells to chemotherapy.

Conclusions
In summary, our study uncovers a new mechanism that SUMOylated PTEN promotes HR repair through dephosphorylation of 53BP1 (Fig. 7L).In response to DNA damage, p14ARF as a SUMO E3 ligase is phosphorylated to enhance the interaction with PTEN in the nucleus, which subsequently promotes PTEN SUMOylation.Then, SUMOylated PTEN is recognized and recruited to the chromatin near DSB by the N-terminal SIM of BRCA1.This pool of PTEN relieves HR repair barrier posted by 53BP1 through directly dephosphorylating 53BP1, promoting HR repair.Blocking PTEN SUMOylation pathway by TAK981 and CPP-p14ARF(2-13) sensitizes tumor cells to DNA damage reagents.Thus, our study elucidated a new molecular mechanism of the key role of PTEN in HR repair during DDR, which may provide a new strategy for clinical cancer therapy.

Fig. 1 .
Fig. 1.PTEN promotes HR repair through facilitating DNA end resection.(A) Immunoblot of pS4/8-RPA32 of HeLa-pLKO and -shPTEN cells treated with 20 Gy and recovery for indicated time.(B) Immunoblot of pS4/8-RPA32 of DU145-PTEN WT and -PTEN À/À cells treated with 10 Gy and recovered for indicated time.(C) Immunoblot of pS4/8-RPA32 of DU145-PTEN À/À cells stably re-expressed PTEN WT, K254R and K266R cells after treated with 10 Gy. (D) Immunoblot of chromatin associated RPA32 and RAD51 in DU145 cells after treated with Zeocin (200 lgÁmL À1 ) for 1 h and recovered for indicated time.(E) RAD51 foci were quantified in DU145 cells treated with 5 Gy and recovery for 6 h and then presented with dot graph.Representative images of RAD51 foci were shown at right panel.scale bar, 20 lm.(F) HR (homologous recombiantion) efficiency were detected in U2OS-DR-GFP-shPTEN cells stably re-expressing PTEN WT , PTEN C124S , PTEN G129E , PTEN K254R and PTEN K266R .Inset: immunoblot of PTEN in U2OS-DR-GFP-shPTEN cells stably re-expressing indicated PTEN mutants.Left panel: quantification of HR efficiency shown as bar graph.Right panel: representative images of FACS.Unpaired Student's t-test was used (**P < 0.01, ***P < 0.001) and data were shown as mean or mean AE SD.All results were shown with one representative image from three independent experiments.Chro., chromatin; FACS, fluorescence-activated cell sorting; HR, homologous recombination; WT, wide type.

Fig. 3 .
Fig.3.p14ARF is a novel SUMO E3 ligase to mediate PTEN SUMOylation during DDR.(A) 293T cells were transfected with His-SUMO1, HA-UBC9, Myc-p14ARF and Flag-PTEN for 48 h.His-SUMO1 conjugates were pulled down with Ni 2+ -NTA beads and SUMOylated PTEN were analyzed with immunoblot.(B) 293T-pLKO or shARF-1 cells were transfected with His-SUMO1, HA-UBC9 and Flag-PTEN for 48 h.His-SUMO1 conjugates were pulled down with Ni 2+ -NTA beads and SUMOylated PTEN were analyzed with immunoblot.(C) 293T cells were transfected with Myc-p14ARF and Flag-PTEN for 48 h.Co-IP were used to detect interaction between PTEN and p14ARF.(D) GST pulldown were used to detect interaction between PTEN and p14ARF.Upper panel: GST-PTEN was purified from BL21 and incubated with 293T lysis which transfected with HA-p14ARF for 48 h.Lower panel: GST-p14ARF was purified from BL21 and incubated with 293T lysis which transfected with Flag-PTEN for 48 h.(E) Truncated Myc-p14ARF and Flag-PTEN were transfected into 293T cells, interaction domain between PTEN and p14ARF were identified with Co-IP.(F) Truncated Myc-p14ARF, His-SUMO1, Flag-UBC9 and HA-PTEN were transfected into 293T cells.His-SUMO1 conjugates were pulled down with Ni 2+ -NTA beads and SUMOylated PTEN were analyzed with immunoblot.(G) His-SUMO1 and HA-PTEN were transfected into 293T-pLKO and shARF-1 cells.After 48 h, cells were treated with Zeocin (400 lgÁmL À1 ) for 1 h and recovery for indicated time.His-SUMO1 conjugates were pulled down with Ni 2+ -NTA beads and SUMOylated PTEN were analyzed with immunoblot.(H) Interaction between PTEN and p14ARF post CPT (20 lM) treatment were detected with Co-IP in 293T cells.(I) Phosphorylation of p14ARF after Zeocin (400 lgÁmL À1 ) treatment for 1 h and recovery for indicated time was detected with phos-tag gel.(J) Interaction of PTEN and p14ARF-WT, T8A and T8D were detected under normal condition or CPT (20 lM) treatment with Co-IP in 293T cells.(K) 293T cells were transfected with Flag-SUMO1, Myc-p14ARF-WT or T8A and HA-PTEN for 48 h.Flag-SUMO1 conjugates were immunoprecipitated and SUMOylated PTEN were analyzed with immunoblot.(L) chromatin loading of PTEN were detected with immunoblot in DU145-pLKO and shARF-1 cells after treatment with CPT (20 lM) for 1 h and recovery for indicated time.All results were shown with one representative image from three independent experiments.Chro., chromatin; Co-IP, Co-immunoprecipitation; CPT, camptothecin; GST, glutathione S-transferase tag; WCL, whole cell lysis; WT, wide type.

Fig. 4 .
Fig. 4. PTEN relieves HR barrier posted by 53BP1 through directly dephosphorylating pT543-53BP1.(A) Immunoblot of pT543-53BP1 in DU145 (top left), HeLa (lower left) cells and PC3 (top right) after treatment with CPT (2 lM) for 30 min and recovery for indicated time.pT543-53BP1 was also detected in DU145 (lower right) cells after treated with CPT (20 lM) for 1 h and recovery for indicated time.(B) Dot graph of pT543-53BP1 foci intensity in DU145 cells after Zeocin (100 lgÁmL À1 ) treatment for 30 min and recovery for indicated time.(C) Immunoblot of pT543-53BP1 and pS25/29-53BP1 in DU145 cells after treatment with CPT (20 lM) for 1 h and recovery for indicated time.(D) Dot graph of pT543-53BP1 foci number and intensity after Zeocin (200 lgÁmL À1 ) treatment for 1 h and recovery for 2 h.Representative images were shown at right panel.scale bar, 20 lm.(E) Dot graph of pS25-53BP1 foci number and intensity after Zeocin (200 lgÁmL À1 ) treatment for 1 h and recovery for 2 h.Representative images were shown at right panel.scale bar, 20 lm.(F) Dot graph of RIF1 foci number and intensity after Zeocin (200 lgÁmL À1 ) treatment for 1 h and recovery for 2 h.Representative images were shown at right panel.scale bar, 20 lm.(G) Immunoblot analysis of pT543-53BP1 level in DU145-PTEN À/À cells stably re-expressed PTEN-WT, C124S and G129E after 10 Gy treatment and recovery for indicated time.(H) Dot graph of pT543-53BP1 foci number and intensity in DU145-PTEN À/À cells stably reexpressed PTEN-WT, C124S and G129E after 5 Gy treatment and recovery for 2 h.(I) in vitro phosphatase assay.GFP tagged phosphorylated 53BP1(1-600) were purified from 293T cells.GST-PTEN WT , C124S and G129E were purified from E. coli BL21.Unpaired Student's t-test was used (**P < 0.01, ***P < 0.001) and data were shown as mean.n.s., not significant.All results were shown with one representative image from three independent experiments.CPT, camptothecin; GST, glutathione S-transferase tag; NC, negative control; WT, wide type.

Fig. 6 .
Fig. 6.HR repair is impaired by PTEN K254R in vivo.(A-D) Numbers of 53BP1, cH2AX, RAD51 and RPA32 foci of MEFs post 5 Gy were quantified with FOCO software and presented as dot graph.(E) Immunoblot of pS4/8-RPA32 of MEFs after treatment with 20 Gy. (F) Immunoblot of chromatin loaded PTEN and RPA32 which were separated from MEFs post 20 Gy. (G) Nuclear-cytosol separation of MEFs post 5 Gy and detection of PTEN localization with immunoblot.(H) Immunofluorescence of PTEN in MEFs after 5 Gy treatment and recovery for 4 h.Scale bar, 50 lm.(I) PtenWT and Pten K254R mice were treated with 8 Gy whole-body IR.Villus length were quantified at indicated time and representative image of HE stained sections of small intestine were shown.More than 100 villi were assessed from each mouse.scale bar, 300 lm.(J) Ki-67 positive cells in intestinal crypts were quantified and representative image of Ki-67 stained sections of small intestine were shown.More than 100 crypts were assessed from each mouse.scale bar, 50 lm.(K) Viability of MEFs treated with Cisplatin, Zeocin or CPT were detected with CCK8 and each group were normalized to no treatment group.Unpaired Student's t-test was used (*P < 0.05, **P < 0.01, ***P < 0.001) and data are shown as mean or mean AE SD.All results were shown with one representative image from three independent experiments.Chro., chromatin; CPT, camptothecin; HE, Hematoxylin and Eosin; MEFs, mouse embryonic fibroblasts; NT, nontreatment; WCL, whole cell lysis; WT, wide type.

Fig. 7 .
Fig. 7. Blocking PTEN SUMOylation pathway sensitizes tumor cells to DNA damage reagents.(A-C) DU145-PTEN À/À and PTEN WT cells were treated with or without different doses of Cisplatin for 3 days, Zeocin for 2 days or CPT for 2 days and then cultured for another 7-10 days.Colony number was counted and presented as bar graph.(D) DU145-PTEN À/À , PTEN WT , PTEN K254R and PTEN K266R cells were treated with or without TAK981 (100 nM) for 3 days and cultured for another 7-10 days.Total intensity of stained colony was measured by IMA-GEJ and then normalized to DU145-PTEN À/À group without TAK981 treatment.(E-G) DU145-PTEN À/À , PTEN WT , PTEN K254R and PTEN K266R cells were treated with Cisplatin (1.5 lM) for 3 days, Zeocin (10 lgÁmL À1 ) for 2 days or CPT (100 nM) for 2 days or combination with TAK981 (100 nM) and cultured for another 7-10 days.Colony number was counted and presented as bar graph.(H) DU145-PTEN À/À and PTEN WT cells were treated with different doses of CPP-p14ARF(2-13) for 3 days and cultured for another 7-10 days.(I-K) DU145-PTEN À/À and PTEN WT cells were treated with combination of Cisplatin (1.5 lM for 3 days), Zeocin (10 lgÁmL À1 for 2 days) or CPT (100 nM for 2 days) and different doses of CPP-p14ARF(2-13).Colony number was counted after cultured for another 7-10 days.(L) A schematic model to show the molecular mechanism of PTEN SUMOylation in HR repair.DNA damage induced phosphorylation of p14ARF enhanced its interaction with PTEN and which subsequently promoted SUMOylation of PTEN.SUMOylated PTEN was recognized and recruited into DNA breaks by BRCA1 and then directly dephosphorylated 53BP1 which helped release of its downstream effectors and DNA end resection.500 cells were seeded in 12-well plates for all colony assays except (C) in which 1000 cells were seeded at the beginning.Unpaired Student's t-test was used (*P < 0.05, **P < 0.01, ***P < 0.001) and data are shown as mean AE SD from three biological replicates.All results were shown with one representative image from three independent experiments.CPP-p14ARF(2-13), cell penetrating peptide fused by p14ARF(2-13aa); CPT, camptothecin; WT, wide type.