Regulation of apoptosis at cell division by p34^cdc2 phosphorylation of survivin

A sequence T³⁴PER matching the consensus phosphorylation site (14) for p34^cdc2 is found in human and mouse survivin, but not in other IAP proteins (6). In in vitro kinase assays, baculovirus-expressed p34^cdc2-cyclin B1 readily phosphorylated histone H1 and wild-type survivin, whereas substitution of Thr³⁴→Ala, i.e., survivin(T34A), abolished phosphorylation by p34^cdc2-cyclin B1 (Fig. 1A). In contrast, baculovirus-expressed Cdk2-cyclin E did not phosphorylate wild-type survivin or survivin(T34A) (Fig. 1A). A rabbit antibody was raised against the survivin peptide L²⁸EGCACT*PERMAEAGFI⁴⁴ containing phosphorylated Thr³⁴ (T*), sequentially affinity purified on nonphosphorylated/phosphorylated peptide-Sepharose and used in immunoblotting. The antibody to phosphorylated Thr³⁴ (α-survivinT34*) recognized wild-type survivin after in vitro phosphorylation by p34^cdc2-cyclin B1, but not unphosphorylated survivin or survivin(T34A) after incubation with p34^cdc2-cyclin B1 (Fig. 1B). In contrast, an antibody to survivin (α-survivin) (12) indistinguishably recognized wild-type survivin or survivin(T34A), irrespective of Thr³⁴ phosphorylation (Fig. 1B). In in vivo experiments, a 16.5-kDa phosphorylated survivin band was immunoprecipitated from orthophosphate-labeled, G₂/M-arrested HeLa cells, whereas no radioactive bands were immunoprecipitated with a control antibody (Fig. 2A). Transfection of HeLa cells with a kinase-dead p34^cdc2 mutant (Asp¹⁴⁶→Asn) (13) nearly completely abolished phosphorylation of endogenous survivin on Thr³⁴ by Western blotting with α-survivinT34* (Fig. 2B). In control experiments, comparable amounts of endogenous survivin were immunoprecipitated from HeLa cells transfected with control pcDNA3 or p34^cdc2(Asp¹⁴⁶→Asn) (Fig. 2B). We next investigated the kinetics of survivin phosphorylation in vivo. The antibody to phosphorylated Thr³⁴ did not recognize survivin immunoprecipitated from synchronized HeLa cells at G₁ or S phase 0, 2, and 4 h after thymidine release, respectively (Fig. 2C). In contrast, phosphorylation of survivin on Thr³⁴ was detected beginning at 8 h after thymidine release, which coincides with entry into mitosis of these cells by DNA content analysis (7), and remained sustained throughout cell division (Fig. 2C). In contrast, α-survivin recognized survivin immunoprecipitated from synchronized interphase and mitotic HeLa cells (Fig. 2C). By immunofluorescence and confocal microscopy, both endogenous survivin and phosphorylation-defective survivin(T34A) localized to midbodies at telophase (Fig. 2D) in agreement with previous observations (7). Although endogenous survivin was phosphorylated on Thr³⁴ on microtubules, no reactivity of α-survivinT34* with survivin(T34A) was demonstrated at midbodies by dual immunofluorescence labeling (Fig. 2D).

We next asked whether survivin and p34^cdc2 physically interacted in vivo. Immunoprecipitates of p34^cdc2 from G₂/M-arrested HeLa cells contained coassociated 16.5-kDa survivin by Western blotting (Fig. 3A). In contrast, no survivin bands were immunoblotted in p34^cdc2 immunoprecipitates from G₁- or S-synchronized cultures (Fig. 3A). Similarly, Cdk2 immunoprecipitates from synchronized HeLa cells did not contain 16.5-kDa coassociated survivin at any cell cycle phase (Fig. 3A). In reciprocal experiments, HA-survivin or HA-survivin(T34A) immunoprecipitated from HeLa cells contained coassociated p34^cdc2 by Western blotting (Fig. 3B), thus demonstrating that a survivin-p34^cdc2 interaction was independent of Thr³⁴. Furthermore, endogenous p34^cdc2 and survivin colocalized on mitotic spindle microtubules at prophase and metaphase, and concentrated at midbodies at telophase, by immunofluorescence and confocal microscopy (Fig. 3C), in agreement with their individual localization (7, 15). This suggests that a phosphorylation-defective survivin(T34A) mutant may act as a dominant negative mutant for its ability to associate with p34^cdc2 on the mitotic apparatus and prevent phosphorylation of endogenous survivin.

We next investigated a potential role of survivin phosphorylation by p34^cdc2 in apoptosis control and/or cell cycle progression (8). In the absence of exogenous apoptotic stimuli, transfection of HeLa cells with survivin(T34A) fused to a GFP caused apoptotic morphology of chromatin condensation and DNA fragmentation in GFP-expressing cells (Fig. 4A). This was associated with generation of hypodiploid cells by DNA content analysis and flow cytometry, in a reaction reversed by the caspase inhibitor Z-VAD-fmk (Fig. 4B). Similar results were obtained after transfection of GFP-caspase-9 (Met¹-Asp³³⁰), whereas expression of GFP vector or wild-type survivin did not affect nuclear morphology in HeLa cells (Fig. 4 A and B). Next, stable lines of survivin-positive YUSAC2 melanoma cells (12) were generated to express survivin(T34A) under the control of a tetracycline (Tet)-regulated promoter (Tet-off system) (16), and one clone (YUSAC2/T34A-C4) was selected for further investigation. On Tet withdrawal, YUSAC2/T34A-C4 cells exhibited loss of the mitotic (G₂/M) fraction, appearance of hypodiploid cells, and labeling for internucleosomal DNA fragmentation by TUNEL (Fig. 4C), thus distinguishing this pathway from mitotic catastrophe at G₂/M (17). In contrast, Tet⁺ YUSAC2/T34A-C4 cells had normal viability and DNA content, and did not label by TUNEL (Fig. 4C). To determine the kinetics of apoptosis induced by survivin(T34A) with respect to cell cycle progression, YUSAC2/T34A-C4 cells were thymidine-synchronized in the presence or absence of Tet, released and analyzed for DNA content at increasing 3-h intervals. Tet⁺ or Tet⁻ YUSAC2/T34A-C4 cells exhibited comparable kinetics of cell cycle progression, approaching the first mitosis 9 h after thymidine release, completing cell division by 15–18 h, and reentering G₁ after 21 h (Fig. 4D). However, coinciding with entry into mitosis, Tet⁻ YUSAC2/T34A-C4 cells began to accumulate within the hypodiploid apoptotic fraction, which increased steadily throughout mitosis and in the postmitotic phase (Fig. 4D). In contrast, no changes in the hypodiploid fraction were observed in Tet⁺ YUSAC2/T34A-C4 cells at the various cell cycle phases (Fig. 4D). Coinciding with entry into mitosis, Tet⁻ YUSAC2/T34A-C4 cells also exhibited progressive proteolytic cleavage of ≈46 kDa proform caspase-9 to active subunits of ≈35 kDa and ≈37 kDa, which peaked in the postmitotic phase with nearly complete disappearance of the ≈46-kDa proform band (Fig. 4E). In contrast, no proteolytic processing of caspase-9 was observed in Tet⁺ YUSAC2/T34A-C4 cells at the various cell cycle phases (Fig. 4E).

Next, we asked whether survivin phosphorylation on Thr³⁴ was required to modulate a potential interaction with effector molecules of apoptosis, and we focused on the intrinsic initiator caspase-9 (18) for its association with survivin in affinity purification experiments (unpublished observations). Immunoprecipitates of HA-survivin from mitotic HeLa cells collected by mitotic shake-off revealed the presence of coassociated ≈35-kDa caspase-9, by immunoblotting (Fig. 5A). In contrast, HA-survivin(T34A) immmunoprecipitated from mitotic HeLa cells did not contain coassociated caspase-9 (Fig. 5A). Similar results were obtained in drug-synchronized cultures. HA-survivin immunoprecipitates from mitotic HeLa cells 12 h after thymidine release contained coassociated caspase-9 of ≈35 kDa, whereas no caspase-9 bands were coprecipitated with HA-survivin(T34A) from mitotic HeLa cells (Fig. 5B). By immunofluorescence and confocal microscopy, HA-survivin and HA-survivin(T34A) transfected in HeLa cells bound to mitotic spindle microtubules and accumulated at midbodies at telophase, indistinguishably from endogenous survivin (Fig. 5C), and in agreement with the data presented above. In HA-survivin transfectants, simultaneous labeling for caspase-9 revealed a diffuse cytoplasmic reactivity, and a prominent colocalization between survivin and caspase-9 at midbodies (Fig. 5C). In contrast, and consistent with the immunoprecipitation experiments described above, caspase-9 did not colocalize with HA-survivin(T34A) at midbodies (Fig. 5C). In control experiments, a survivin Leu⁶⁴→Ala mutant did not cause apoptosis (7), and colocalized with caspase-9 at midbodies (Fig. 5C). Next, we asked whether cell death induced by survivin(T34A) was mediated by mislocalized, active caspase-9. Transfection of HeLa cells with a caspase-9(C287A) dominant negative mutant (19) inhibited nuclear fragmentation and chromatin condensation induced by the anti-cancer drug etoposide, and reversed HeLa cell apoptosis induced by survivin(T34A) (Fig. 5D). In contrast, cotransfection of HeLa cells with a caspase-8(C360S) dominant negative mutant did not affect apoptosis induced by survivin(T34A), whereas it completely inhibited cell death induced by TNFα + cycloheximide (20) (Fig. 5D).

In summary, these data identify survivin as a mitotic substrate of p34^cdc2-cyclin B1 (21), and suggest that survivin phosphorylation on Thr³⁴ may regulate apoptosis at cell division via an interaction with caspase-9 (2). Although a role of unscheduled p34^cdc2 activity in apoptosis has been debated (22, 23), the data presented here are consistent with a cytoprotective role of p34^cdc2, previously suggested from genetic inactivation (24), or pharmacologic inhibition of the kinase (25). Similar to other mitotic substrates (26, 27), phosphorylation by p34^cdc2 may increase the affinity of survivin for active caspase-9, or, alternatively, stabilize survivin or a survivin-caspase-9 anti-apoptotic complex at midbodies during cell division. Based on the survivin crystal structure, Thr³⁴ is ideally positioned to influence survivin function(s) potentially mediated by the baculovirus IAP repeat (BIR) (28). Conversely, loss of survivin phosphorylation on Thr³⁴ resulted in dissociation of a survivin-caspase-9 complex at midbodies and caspase-9-depedent apoptosis at cell division. Although survivin clearly plays an evolutionary conserved role (9, 10) in regulation of mitotic progression and cytokinesis (8), cells expressing survivin(T34A) progressed through the cell cycle and did not exhibit a G₂/M arrest, as typically seen after interference with regulators of cytokinesis (29, 30). Taken together with the identification of a Drosophila survivin homolog, Deterin, which acts as a genuine apoptosis inhibitor (31), these data suggest that survivin may have integrated a primordial role in cell division (9, 10) with a more recent function in apoptosis control at G₂/M. Because of the over-expression of survivin in cancer but not in normal tissues (5, 11), selective antagonists of survivin phosphorylation by p34^cdc2 may trigger apoptosis of transformed cells and facilitate their elimination at mitosis.

This paper was submitted directly (Track II) to the PNAS office.

Article published online before print: Proc. Natl. Acad. Sci. USA, 10.1073/pnas.240390697.

Article and publication date are at www.pnas.org/cgi/doi/10.1073/pnas.240390697

We thank Drs. J. C. Reed for discussion, D. Schatz for pTA-Neo plasmid, V. Dixit for caspase-3, -9, and caspase-8(C360A) cDNAs, and K. Weber and M. Osborn for mAb 20C6. This work was supported by National Institutes of Health Grants CA78810, HL54131 (to D.C.A.) and CA72878 (to H.Z.), and Associazione Italiana per la Ricerca sul Cancro and Telehon (to P.C.M.). D.G. was supported by National Institutes of Health Dermatology Training Grant 5T32AR07016 and by a Dermatologist Investigator Research Fellowship from the Dermatology Foundation.