Nuclear PTEN Controls DNA Repair and Sensitivity to Genotoxic Stress

See allHide authors and affiliations

Science  26 Jul 2013:
Vol. 341, Issue 6144, pp. 395-399
DOI: 10.1126/science.1236188

This article has a correction. Please see:

PTEN Variations

The product of the tumor suppressor gene phosphate and tensin homolog on chromosome ten (PTEN) is a lipid and protein phosphatase that regulates important cellular processes, including growth, survival, and metabolism (see the Perspective by Leslie and Brunton). Though PTEN is best known for effects on the phosphatidylnositol 3-kinase (PI3K) signaling pathway, the PTEN protein is also found in the nucleus. Bassi et al. (p. 395) found that PTEN's presence in the nucleus was regulated in response to covalent modification of the protein by SUMOylation and phosphorylation. Cells lacking nuclear PTEN showed increased sensitivity to DNA damage and underwent cell death if the PI3K pathway was also inhibited. Hopkins et al. (p. 399, published online 6 June) discovered an alternative translation start site in human PTEN messenger RNA that allowed expression of a protein, PTEN-Long, with about 170 extra amino acids. The unusual enzyme was released from cells and then taken up into other cells. In a mouse tumor model, uptake of the enzyme inhibited the PI3K pathway and inhibited tumor growth.


Loss of function of the phosphatase and tensin homolog deleted on chromosome 10 (PTEN) tumor suppressor gene is associated with many human cancers. In the cytoplasm, PTEN antagonizes the phosphatidylinositol 3-kinase (PI3K) signaling pathway. PTEN also accumulates in the nucleus, where its function remains poorly understood. We demonstrate that SUMOylation (SUMO, small ubiquitin-like modifier) of PTEN controls its nuclear localization. In cells exposed to genotoxic stress, SUMO-PTEN was rapidly excluded from the nucleus dependent on the protein kinase ataxia telangiectasia mutated (ATM). Cells lacking nuclear PTEN were hypersensitive to DNA damage, whereas PTEN-deficient cells were susceptible to killing by a combination of genotoxic stress and a small-molecule PI3K inhibitor both in vitro and in vivo. Our findings may have implications for individualized therapy for patients with PTEN-deficient tumors.

PTEN (phosphatase and tensin homolog on chromosome 10) is encoded by one of the most commonly deleted tumor suppressor genes in human cancer. PTEN acts as a 3′-specific phosphatidylinositol phosphatase and counters the activity of the phosphatidylinositol 3-kinase (PI3K) signaling pathway in the cytoplasm (1). The PTEN signaling network is implicated in the control of cell metabolism, growth, proliferation, survival, and migration; processes invariably aberrant in cancer (26). PTEN may have roles in maintaining genomic stability, mediated at least in part by PI3K-independent mechanisms (710). PTEN is also found in the nuclei of many normal and cancerous cells and tissues, and various molecular mechanisms of PTEN nuclear localization have been described (11).

We noticed that, in addition to the expected PTEN protein of molecular mass ~55 kD, an ~75-kD PTEN protein species could be detected with three PTEN antibodies in various mammalian cell lines (Fig. 1A). The ~75-kD species, which we termed PTEN-H, represented <10% of the total PTEN in the cell lines tested. PTEN-H was absent from mouse embryo fibroblasts (MEFs) and human colon carcinoma HCT116 cells with a targeted disruption of the PTEN gene (Fig. 1A). Transfection of a vector expressing hemagglutinin (HA) epitope–tagged PTEN cDNA gave rise to PTEN and PTEN-H, establishing that PTEN-H arises as a consequence of posttranslational modification rather than alternative initiation or splicing (Fig. 1B). Use of mild detergents largely precluded detection of PTEN-H in cell lysates (fig. S1A). Inclusion of N-ethylmaleamide (NEM), an inhibitor of cysteine-based enzymes including deSUMOylases (SUMO, small ubiquitin-like modifier) and deubiquitinases, during cell lysis increased the detection of PTEN-H (fig. S1B). Mono-SUMOylation could explain an apparent ~20-kD increase in molecular mass. Indeed, PTEN-H, but not “regular” PTEN, reacted with the antibodies to SUMO 2 and SUMO 3 in PTEN immunoprecipitates from human embryonic kidney (HEK) 293 cells (Fig. 1C). Depletion of Ubc9, the sole SUMO-conjugating E2 protein (12); overexpression of the SENP1 or SENP2 deSUMOylases (Fig. 1D and fig. S2); or treatment of PTEN immunoprecipitates with recombinant SENP1 or SENP2 decreased amounts of PTEN-H levels (Fig. 1E), further indicating that PTEN-H is a SUMOylated form of PTEN (SUMO-PTEN).

Fig. 1 SUMOylation of PTEN at K254 in vivo.

(A) Multiple cell lines express a 75-kD form of PTEN. Whole-cell lysates were immunoblotted with three PTEN antibodies: 6H2.1 mouse monoclonal antibody (mAb) (top), rabbit mAb CST number 9559 (middle), or a rabbit polyclonal antibody (pAb) (bottom) CST number 9552. PTEN-H is indicated (arrow). WT, wild type; IB, immunoblot. (B) Transfection of minimal PTEN cDNA leads to PTEN-H formation. pcDNA3.1-HA-PTEN (WT) or pcDNA3.1-HA (EV) were transfected into HEK293 cells, and HA-immunoprecipitates (IP) immunoblotted for PTEN. (C) SUMOylation of endogenous PTEN. PTEN was immunoprecipitated from HEK293 lysates with protein A beads alone or PTEN antibody (CST 9559) and immunoblotted with the PTEN 6H2.1 antibody (left), then stripped and reprobed with antibodies to Sumo2/3 (right). IgG, immunoglobulin G. (D) Formation of SUMO-PTEN requires Ubc9. HEK293 cells were transfected with scrambled or Ubc9 small interfering RNAs and harvested at 48 (middle lane) or 72 (right lane) hours posttransfection. Lysates were immunoblotted for PTEN, Ubc9, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), as indicated. Arrow indicates SUMO-PTEN. (E) Sensitivity of PTEN-H to deSUMOylases. Immunoprecipitated Flag-PTEN was incubated with GST, GST-SENP1, or GST-SENP2, as indicated, in the presence or absence of 20 mM NEM. Reactions were immunoblotted with antibodies against PTEN (6H2.1) (top) or Sumo2 (bottom). (F) PTEN is SUMOylated on K254 in vivo. HEK293 cells were transfected with FLAG-PTEN-WT or FLAG-PTEN-K254R, together with His-SUMO2 and myc-His-Ubc9, as indicated. Flag immunoprecipitates were immunoblotted for PTEN.

In vitro SUMOylation of a series of PTEN deletion mutations (fig. S3A) identified amino acids 238 to 320 as the minimal SUMOylated PTEN polypeptide (fig. S3, B and C). This PTEN portion contains a strong predicted SUMOylation site at position 254, a mutation of which precluded SUMOylation in vitro (fig. S3C). Nanoflow liquid chromatography–tandem mass spectrometry of SUMOylated PTEN polypeptides combined with SUMmOn pattern recognition software (13) independently identified K254 (14) as a bona fide SUMOylation site (fig. S4, A and B). Consistently, wild-type but not Flag-PTEN K254R (K to R mutation at position 254) was readily SUMOylated in HEK293 cells (Fig. 1F), identifying K254 as the major PTEN SUMOylation site. The PTEN K254R mutant retained the ability to counter PI3K signaling because its expression in PTEN-deficient human U87MG glioblastoma cells (6) resulted in comparable decreases in phosphorylation of protein kinase B (PKB) (also called Akt) and of the PKB target glycogen synthase kinase3β, as did expression of wild-type PTEN (fig. S5A), and in vitro displayed equal phosphatase activity toward phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P3] (fig. S5B).

Unlike wild-type PTEN and an unrelated lysine PTEN K289E mutant, PTEN K254R expressed in HEK293 (Fig. 2A) or U87MG (fig. S6) cells failed to localize to the nucleus, suggesting that SUMOylation might promote nuclear localization of PTEN. In cells treated with leptomycin B, an inhibitor of nuclear export, PTEN-K254R localized in the nucleus, indicating that this mutant can enter the nucleus but is not retained there (Fig. 2B). Treatment of cells with ionizing radiation (IR) led to loss of nuclear PTEN (Fig. 2C). Judging by fractionation of cellular lysates, SUMOylated PTEN was predominantly nuclear and reduced within hours of exposure to DNA damage (Fig. 2D).

Fig. 2 PTEN SUMOylation regulates nuclear retention, which is sensitive to genotoxic stress.

(A) Exclusion of the SUMO-deficient mutant PTEN K254R from the nucleus. Flag–fluorescein isothiocyanate (FITC) immunofluorescence images of HEK293 cells transfected as indicated. Insets show DAPI (4′,6-diamidino-2-phenylindole) staining. (B) Nuclear retention of SUMOylated PTEN. Immunofluorescence as in (A). Cells were treated with 10 ng/ml of leptomycin B for 4 hours. (C) Decreased nuclear PTEN localization after genotoxic stress. Immunofluorescence as in (A) of PTEN-FITC U87MG cells transfected as indicated, 4 hours post-IR. Bar graph represents percentage of cells with nuclear PTEN in control or IR-treated cells (P < 0.001, t test, n = 5, bars represent SEM). (D) Decreased nuclear PTEN after genotoxic stress. HeLa cells were treated as indicated, harvested after 4 hours, and separated into cytoplasmic and nuclear fractions followed by immunoblotting for PTEN. Fractionation was monitored by immunoblotting for PARP [poly(ADP-ribose) polymerase, a nuclear protein] and GAPDH (cytoplasmic protein). Arrow indicates SUMO-PTEN.

We investigated the DNA damage response (DDR) of U87MG cells, HCT116 cells, and their variants with the PTEN gene disrupted (HCT116 PTEN−/−) (15)—engineered to express PTEN wild type, K254R, mutants lacking all phosphatase activity (C124S) or only lipid-phosphatase (G129E) activity, or empty vector control (fig. S7A)—by monitoring the formation of p53 binding protein 1 (53BP1) foci (16) (Fig. 3A and figs. S7B and S8A). Four hours after IR exposure, 53BP1 foci were visible in all cell lines, indicating the presence of double-strand breaks (DSBs). However, by 24 hours, 53BP1 foci had largely resolved in wild-type PTEN- and G129E-reconstituted cells but not in PTEN-deficient cells, cells lacking nuclear PTEN (K254R), or cells lacking all PTEN phosphatase activity (C124S) (Fig. 3A and figs. S7B and S8A), indicative of deficient DNA repair. Consistently, PTEN-deficient cells or cells expressing PTEN K254R or PTEN C124S were also defective in resolving phosphorylated histone variant H2AX (γH2AX) and breast cancer gene 1, early onset (BRCA1) foci (Fig. 3, B and C, and figs. S7C and S8B). Indicative of deficiency in homologous recombination (HR)–based repair, PTEN-null cells, as well as their K254R- and C124S-expressing counterparts, failed to recruit RAD51 to the sites of DNA damage (Fig. 3D and figs. S8C and S7D) without affecting RAD51 mRNA or protein abundance (fig. S9, A to C). Moreover, U87MG cells and cells derived from a mouse mammary tumor cell with a conditional PTEN gene disruption (WAP-Cre PTEN−/− MMTCs) expressing PTEN-K254R or PTEN-C124S were deficient in repair of a stably integrated, HR-mediated DSB repair reporter (17) (Fig. 3E and fig. S10). To distinguish the importance of nuclear localization versus SUMOylation for PTEN function in the response to DSBs, we fused a nuclear localization sequence from SV40 large T antigen (18) to the non-SUMOylatable PTEN mutant (NLS-PTEN K254R). Despite constitutive localization to the nucleus (fig. S11A), NLS-PTEN K254R (fig. S11B) did not restore the impaired response of U87MG cells to IR (fig. S11, C and D), indicating that SUMOylation is required for PTEN’s function in DDR. Reexpression of wild-type PTEN or various PTEN mutants did not yield changes in cell-cycle distribution (fig. S12A) or the engagement of cell-cycle checkpoints after IR (fig. S12B), indicating that DNA repair deficiency was not secondary to the potential effects of PTEN on the cell cycle. We monitored the effects of PTEN on radiosensitivity of cells by scoring the surviving fraction of U87MG cells and PTEN−/− MMTCs expressing PTEN mutants 5 days after exposure to IR. Although PTEN K254R-expressing cells exhibited a decreased surviving fraction after IR exposure (Fig. 3F and fig. S13), the survival of PTEN-null cells was indistinguishable from that of wild-type PTEN-expressing cells (Fig. 3F and fig. S13), possibly reflecting the activation of PI3K-mediated cell survival signaling in cells lacking PTEN.

Fig. 3 Requirement of nuclear SUMO-PTEN for homologous recombination repair of DNA DSBs.

U87MG cells were reconstituted with the indicated proteins. Cells treated with IR (5 Gy) were immunostained at the indicated times with antibodies to γH2AX, 53BP1, BRCA1, or RAD51. Cells containing more than five foci were scored as positive. Bars represent SEM. ctr, control. PTEN SUMOylation is required for the resolution of 53BP1 foci (P = 0.0014, t test, n = 3) (A), γH2AX foci (P = 0.007, t test, n = 3) (B), and BRCA1 foci (P = 0.0136, t test, n = 3) (C). (D) PTEN SUMOylation is required for RAD51 focus formation (P = 0.0016, t test, n = 3). (E) Impaired homologous recombination-based DNA repair in PTEN- and SUMO-PTEN–deficient cells. U87MG cells stably expressing a DR–green fluorescent protein reporter were reconstituted with the indicated PTEN proteins, and their HR-mediated repair assessed as previously described (17). HR efficiency is expressed relative to that of WT (P = 0.012, t test, n = 2). (F) Lack of SUMO-PTEN increases radiosensitivity. Surviving fraction was determined by sulforhodamine B staining 6 days after exposure to 3 Gy of IR (P = 0.02, t test, n = 3).

IR led to a gradual reduction in the amounts of SUMO-PTEN beginning 1 hour after IR exposure with the steady-state amounts returning 8 hours later (Fig. 4A). Other forms of genotoxic stress, such as treatment with cisplatin or doxorubicin, also led to depletion of SUMO-PTEN, with the timing consistent with appearance of DNA damage elicited by these agents (fig. S14). Protein kinases ataxia telangiectasia mutated (ATM) and ATM and Rad3–related phosphorylate multiple targets after DNA damage (16). Inhibition of ATM impaired SUMO-PTEN turnover in response to IR (Fig. 4B), whereas ATM immunoprecipitated from γ-irradiated cells phosphorylated both human and mouse glutathione S-transferase (GST)–PTEN to a similar extent as it did p53, a known ATM substrate (Fig. 4C). PTEN contains a putative ATM phosphorylation site (19) at position 398. Mutation of this residue to alanine (T398A in human, S398A in mouse) decreased PTEN phosphorylation by ATM (Fig. 4C). PTEN was also phosphorylated at this site in vivo after IR in an ATM-dependent manner (Fig. 4D) establishing this residue as a likely ATM phosphorylation site within PTEN. Unlike wild-type SUMO-PTEN, SUMO-PTEN S/T398A was resistant to IR-induced turnover (Fig. 4E) and was not excluded from the nucleus in cells exposed to IR (Fig. 4F).

Fig. 4 PTEN is part of a ATM-regulated signaling cascade governing sensitivity to genotoxic agents.

(A) Decreased SUMO-PTEN after DNA damage. Lysates from HeLa cells treated as indicated were immunoblotted for PTEN. A ratio of SUMO-PTEN to PTEN immunoreactivity relative to the nonirradiated control is indicated. (B) ATM regulation of SUMO-PTEN abundance by ATM. Lysates from HeLa cells treated as indicated were immunoblotted for PTEN and quantified as in (A). (C) ATM phosphorylates S/T 398 of PTEN in vitro. The indicated GST-fusion proteins were incubated with active ATM in the presence of 32P-ATP. Incorporation of 32P was quantified by PhosphorImager (Molecular Dynamics, Incorporated) and normalized to protein loading. (D) Phosphorylation of PTEN S398 in vivo by ATM. HA immunoprecipitates from HEK293 cells transfected and treated as indicated were immunoblotted with antibody against phospho-ATM substrates (CST 9607). (E) SUMO-PTEN-T398A is not sensitive to IR. Flag immunoprecipitates from HEK293 cells transfected and treated as indicated were immunoblotted for PTEN. (F) PTEN-T398A remains in the nucleus after IR. Cells were treated and imaged as in Fig. 2C. (G and H) PTEN loss sensitizes cells to combination treatment with IR and a pan-PI3K inhibitor. Bars represent SEM. (G) U87MG cells reconstituted with empty vector or PTEN-WT were treated with either 2 Gy of IR, 500 nM BKM120, or both, and apoptosis was measured by annexin V+/7-AAD+ staining (P = 0.0019, t test, n = 3). (H) HCT116 PTEN−/− and parental PTEN-WT cells were treated with either IR (2 Gy) or 1 mM cisplatin alone or in combination with 500 nM BKM120, as indicated. Apoptosis was measured as in (G) (IR+BKM120 P = 0.0025, t test, n = 3) (Cis+BKM120 P = 0.0475, t test, n = 3). (I) PTEN-deficient cells are sensitive to the cisplatin and PI3K inhibitor combination in vivo. HCT116 parental or HCT116 PTEN−/− cells were injected subcutaneously into nonobese diabeticor severe combined immunodeficient mice (n = 10). Mice were treated with cisplatin, BKM120, or both. Data points represent mean tumor volume ±  SEM. Tumor growth difference was measured between day 1 and 16, and one-way analysis of variance was performed (P = 0.0321 for HCT116 parental and P < 0.0001 for HCT116 PTEN−/−) followed by Dunnet’s multiple comparison test (HCT116 PTEN−/− BKM120 + cisplatin versus Ctr, P < 0.0001).

We compared the sensitivity of U87MG cells reconstituted with either empty vector or wild-type PTEN and the parental HCT116 and HCT116 PTEN−/− cells to BKM120, a small molecule pan-PI3K inhibitor; IR (2 Gy; 1 Gy = 100 rads); or a combination thereof at doses that produced minimal toxicity when administered alone (Fig. 4, G and H, and fig. S15). PTEN-deficient U87MG cells displayed increased sensitivity to a combination of IR and BKM120, whereas a combination of BKM120 with the genotoxic agent cisplatin had an enhanced effect in HCT116 PTEN−/− cells (Fig. 4H). In vivo, immunocompromised mice carrying HCT116 and HCT116 PTEN−/− xenografts were treated with a single dose of cisplatin, daily BKM120 for 15 days, or a combination thereof (Fig. 4I). Although either drug alone had limited effect [<30% tumor growth inhibition (TGI) over the course of the treatment], the combination reduced the growth of HCT116 PTEN−/− xenografts (>90% TGI) but not their PTEN-proficient counterparts (Fig. 4I). Such synthetic sensitivity of PTEN-deficient cells to the combined action of DNA-damaging agents and PI3K pathway inhibitors might be useful in treating PTEN-deficient tumors (fig S16). By contrast, our results suggest that administering genotoxic agents alone to such tumors could accelerate the acquisition of additional mutations (fig S16). The ongoing clinical development of numerous agents countering activated PI3K signaling in cancer (20, 21) and next-generation genotoxic agents (22) should facilitate testing of these concepts in the clinic.

Supplementary Materials

Materials and Methods

Figs. S1 to S16

References (2325)

References and Notes

  1. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; E, Glu; G, Gly; K, Lys; R, Arg; S, Ser; and T, Thr.
  2. Acknowledgments: We thank P. Fraser for reagents, D. Durocher for advice on DNA repair assays, R. Hakem for critical reading of the manuscript, A. Wakeham for advice on xenograft experiments, and Novartis for BKM120. Supported by grants from the Canadian Cancer Society to V.S. (2011-700891) and from NIH to B.G.N. (R37 CA49152) and partially supported by a grant from the Ontario Ministry of Health and Long Term Care and the Princess Margaret Hospital Foundation. B.G.N. and T.W.M. hold Canada Research Chairs (Tier I), and B.R. holds Canada Research Chair (Tier II). C.B. was supported by the Excellence in Radiation Research for the 21st century (EIRR21st) fellowship, and R.J.O.D. was supported by a Banting Postdoctoral Fellowship from the Canadian Institutes of Health Research. C.B. and V.S. designed research; C.B., J.H., T.S., R.D., C.G., and S.J.M. performed research; C.B., T.S., B.G.N., B.R., and V.S. analyzed data; T.M., B.G.N., B.R., and V.S. supervised research; and C.B., B.G.N., and V.S. wrote the paper.
View Abstract

Navigate This Article