Cell Division and Cell Survival in the Absence of Survivin

Cell division and cell survival in the absence of survivin

Dun Yang*, Alana Welm, and J. Michael Bishop

G. W. Hooper Research Foundation and Department of Microbiology and Immunology, University of California, 513 Parnassus Avenue, San Francisco, CA 94143

Contributed by J. Michael Bishop, September 9, 2004 The survivin protein contains structural features of the inhibitor of uncertain which of these effects are primary, and which are apoptosis protein family. Previous studies have suggested that secondary. Because cancer cell lines have been widely used to survivin is essential for cell survival because it counteracts an study the function of survivin in the past, interpretation of the otherwise constitutive propensity to apoptosis during mitosis. In results is complicated by the abnormal ploidy and multiple addition, survivin appears to be a component of the chromosomal genetic defects characteristic of these cells, which may modify the passenger protein complex that participates in multiple facets of outcome of survivin depletion. cell division. Here we report that euploid human cells do not die in We used RNA interference (RNAi) to specifically reduce the absence of survivin. Instead, depletion of survivin caused survivin expression in the IMR-90 strain of human primary lung defects in cell division, followed by an arrest of DNA synthesis due fibroblasts and in a human retinal pigment epithelial (RPE) cell to activation of a checkpoint involving the tumor suppressor line that has normal ploidy and functioning checkpoints for the protein p53. During anaphase mitosis in survivin-deﬁcient cells, cell cycle and mitotic spindle (9). In contrast to studies with sister chromatids disjoined normally, but one or more of the sister cancer cells (3, 7, 10, 11), we report that RPE and IMR-90 cells chromatids frequently lagged behind the main mass of segregating can survive in the absence of survivin but suffer frequent chromosomes, probably because of merotelic kinetochore attach- missegregation of sister chromatids and lack the spindle midzone ments. Survivin-deﬁcient cells initiated but failed to complete and midbody microtubules during late mitosis. These abnormal- cytokinesis, apparently because the spindle midzone and midbody ities result in a previously described failure of cytokinesis and a microtublues were absent during late mitosis. The abnormalities of previously unrecognized p53-dependent arrest of DNA synthe- both chromosome segregation and cytokinesis could be attributed sis. A deficiency of p53 allows DNA endoreduplication with- to a defect in the chromosomal passenger protein complex, with a out completion of cytokinesis in survivin-depleted RPE cells, consequent mislocalization of the kinesin-like motor protein resulting in cell-division defects typically found in survivin- MKLP-1 playing a more immediate role in the microtubule abnor- depleted p53-deficient cancer cells, such as abnormalities of cen- malities. Depletion of another chromosomal passenger protein, trosome number and multipolar spindles (3). We conclude that aurora-B, recapitulated the survivin RNA interference phenotypes. survivin may not be essential for cell survival, but the protein is We conclude that survivin can be essential for the proliferation of required for cellular proliferation, because as a CPP, it has an normal human cells by virtue of its contributions to accurate sister ͞ essential role in both the segregation of sister chromatids and the chromatid segregation and assembly stabilization of microtubules ͞ in late mitosis. However, the protein is not inevitably required for assembly stabilization of microtubules late in mitosis. the survival of normal cells. Materials and Methods Preparation of Small Hairpin RNA (shRNA). apoptosis ͉ cytokinesis ͉ chromosome segregation ͉ chromosomal shRNAs were prepared passenger proteins ͉ mitosis as described in ref. 12. Each RNA had a 24-bp stem that is complementary with a unique sequence within the target gene. All RNAs had an extra GGG at the 5Ј end,aUUatthe3Ј end, he protein survivin has attracted attention because of its and a UUGAGAG loop. The complete sequences of shRNAs Tabundant expression in various human cancers and its po- are available upon request. suv1 and suv2 are complementary to tential as a target in cancer therapy (1). Because survivin sequences 575–598 and 586–609, respectively, within the sur- contains a BIR domain (baculoviral inhibitor of apoptosis vivin mRNA (National Center for Biotechnology Information protein repeats), it was assigned to the inhibitor of apoptosis mRNA record no. NM࿝001168). protein family (2). Numerous studies have shown that survivin overexpression confers cytoprotection against a variety of apo- Western Blot Analysis of Cell Extracts. Immunoblotting was per- ptotic stimuli, whereas loss of survivin expression or function formed as described in ref. 12. Mouse monoclonal antibodies causes spontaneous apoptosis or sensitizes cancer cells to apo- were used to detect p53 and actin, and rabbit polyclonal anti- ptotic stimuli (1). It has been proposed that survivin functions at the interface between cell division and cell survival by counter- bodies were used for all other analysis: p21Cip1, cyclin D1, cyclin acting a constitutive pathway that induces apoptosis during E, cyclin B1, cyclin A, pRB (Ser-780 phosphorylation-specific), mitosis (3, 4). But the existence of that pathway remains and proliferating cell nuclear antigen (PCNA). The antiserum hypothetical, and little is known of how survivin might be for phosphorylated pRB was from Cell Signaling Technology involved in the cellular machinery that mediates apoptosis and (Beverly, MA) and the others were from Santa Cruz Biotech- cell division. nology. Horseradish peroxidase-conjugated anti-mouse and an- Survivin has been defined as one of the chromosomal pas- ti-rabbit immunoglobulins were from Amersham Pharmacia. senger proteins (CPPs), which also include the aurora-B kinase, the inner centromere protein, and the telophase disk antigen Abbreviations: CPP, chromosomal passenger protein; CREST, calcinosis, Raynaud’s phenom- (TD-60) (5). These proteins appear to function as part of a enon, esophageal dysfunction, sclerodactyly, and telangiectasia; DAPI, 4Ј,6-diamidino-2- multiprotein complex, referred to as the CPP complex, which phenylindole; RNAi, RNA interference; RPE, retinal pigment epithelial; shRNA, small hair- plays multiple roles during cell division (5). Depletion of survivin pin RNA. in human cells has been reported to cause apoptosis and *To whom correspondence should be addressed. E-mail: [email protected]. pleiotropic defects in cell division (3, 6–8). However, it remains © 2004 by The National Academy of Sciences of the USA

15100–15105 ͉ PNAS ͉ October 19, 2004 ͉ vol. 101 ͉ no. 42 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0406665101 Downloaded by guest on September 29, 2021 Western blots were developed with the ECL detection kit (Amersham Pharmacia).

Fluorescence Microscopy. Immunofluorescence staining was performed as described in ref. 13. Mouse monoclonal antibodies were used to detect p53, survivin, AIM-1, ␤-tubulin, and ␥- tubulin. Rabbit polyclonal antibodies were used for MKLP-1, survivin, and inner centromere protein. Human autoimmune serum from a patient with calcinosis, Raynaud’s phenomenon, esophageal dysfunction, sclerodactyly, and telangiectasia (CREST) was used for staining kinetochores. Antibodies to p53, MKLP-1, and survivin (rabbit) were from Santa Cruz Biotech- nology. Antibodies to ␤- and ␥-tubulin were from Sigma, and monoclonal antibody for survivin was from Cell Signaling Tech- nology. Antibodies to AIM-1 were from BD Transduction Laboratories (Lexington, KY). CREST antibody was a gift from B. R. Brinkley (Baylor College of Medicine, Houston). Anti- bodies to inner centromere protein were from T. Stukenberg (University of Virginia, Charlottesville) and W. C. Earnshaw (University of Edinburgh, Edinburgh). Primary antibodies were detected with Texas red-conjugated or fluorescein isothiocyanate-conjugated secondary antibodies, purchased from Jackson ImmunoResearch. After immunostaining, we mounted cells on Fig. 1. Depletion of survivin elicits arrest of cellular proliferation. (A) Selec- microscope slides with 4Ј,6-diamidino-2-phenylindole (DAPI)- tive silencing of survivin expression by RNAi. Immunoblotting of whole-cell containing Vectashield mounting solution from Vector Labo- extracts from mock-treated cells (lane 1), and cells treated with shRNAs against firefly luciferase (Fluc) (lane 2) or survivin (suv1, suv2, and suv1m) ratories. For fluorescence detection, we used an Axioskop 50 (lanes 3–5) was performed 48 h after RNAi treatment. Actin was used as a immunofluorescence microscope (Zeiss). loading control. (B) Proliferation of RPE cells. Cells were treated with either suv1 or suv1m for 16 h at 70–90% confluence and then passaged into fresh BrdUrd Cellular Proliferation Assay. Cells were incubated for 16–24 medium at 5% confluence. Cells were harvested at daily intervals and counted h in the presence of 10 ␮M BrdUrd and fixed in 75% ethanol. with a hemocytometer. Each data point represents the average of three Nuclear incorporation of BrdUrd was visualized by immuno- independent experiments, and each experiment was done in triplicate. Error staining with a cell proliferation kit (Oncogene Research Prod- bars represent SD. ucts, San Diego). DNA synthesis determined by BrdUrd incorporation assays Histochemistry Staining. Cells were fixed and stained for nuclei and cytoplasm according to the manufacturer’s procedures of diminished in response to survivin RNAi (data not shown and CELL BIOLOGY the Diff-Quik Fixative kit (Dade Behring, Newark, DE). see Fig. 5D). However, apoptosis could not be detected by terminal deoxynucleotidyltransferase-mediated dUTP nick end Cell Culture and shRNA Transfection. Cells were cultured and labeling assays (data not shown). Survivin depletion in IMR-90 treated with shRNA as described in ref. 12. Cells were passaged fibroblasts also caused arrest of proliferation but not cell death after 16 h of incubation with shRNA and then maintained in (data not shown). We conclude that euploid human cells can culture for the indicated time before analysis by Western blot- survive in the absence of survivin but fail to proliferate. ting. For immunostaining and BrdUrd assay experiments, cells were cultured on coverslips after shRNA treatment. The effi- Depletion of Survivin Causes Nuclear Abnormalities. We used flow ciency of transfection was Ͼ95% for RPE cells. cytometry to characterize the proliferative arrest caused by depletion of survivin. The fraction of RPE cells with 4N DNA Results content (N is the haploid DNA content) tripled at the expense Inhibition of Survivin Expression by RNAi. To study the function of of cells with 2N DNA content (Fig. 2A). However, there was no Ͼ Ͻ survivin, we used shRNA to selectively reduce survivin expres- sign of cells with DNA content 4N or 2N, in contrast with sion in RPE cells (12). Because three isoforms of survivin previous reports that loss of survivin in cancer cells caused transcripts have been reported (14), we designed two shRNAs polyploidy and͞or apoptosis (3, 15). The accumulation of RPE (suv1 and suv2) that have target sequences common to all three cells with 4N DNA content indicated that the cells remained transcripts. Immunoblotting analysis revealed that survivin pro- capable of DNA replication for one S phase in the absence of tein was almost eliminated after transfection of either shRNA survivin. Thus, the overall decline in DNA synthesis described into RPE cells (Fig. 1A). In contrast, expression of actin was not above in response to survivin RNAi probably arose secondarily affected. Neither an shRNA that has a three-nucleotide mis- from a cessation of the cell division cycle, rather than from a match with survivin mRNA (suv1m) nor an shRNA against direct effect on DNA replication. firefly luciferase (Fluc) reduced survivin expression, further Accumulation of cells with 4N DNA content could be due to demonstrating the specificity with which the survivin shRNAs blockage at the G2–M transition, mitotic arrest, failure of acted. We conclude that expression of survivin was selectively chromosome segregation, or a failure in cytokinesis. To distin- reduced by its cognate shRNAs. guish among these possibilities, we examined the nuclear morphology of survivin-depleted RPE cells by DAPI staining (Fig. Depletion of Survivin in RPE Cells Causes Proliferative Arrest. After 2B). Most cells had nuclei with uncondensed DNA, thus exclud- survivin depletion, RPE cells became abnormally large and ing mitotic arrest (data not shown). However, there was an flattened without a significant increase in cell numbers (Fig. 1B accumulation of cells with either two nuclei (Fig. 2Ba)orone and data not shown). In contrast, RPE cells proliferated nor- bilobed nucleus (Fig. 2Bband c). About 35% of cells had one mally after either mock treatment or treatment with a control or the other of these abnormalities 72 h after introduction of suv1 shRNA (suv1m) (Fig. 1B). Consistent with proliferative arrest, (Fig. 2C). In the same period, the population of cells with 4N

Yang et al. PNAS ͉ October 19, 2004 ͉ vol. 101 ͉ no. 42 ͉ 15101 Downloaded by guest on September 29, 2021 Fig. 3. Chromosome behavior and microtubule assembly during mitosis are abnormal in survivin-depleted cells. RPE cells were fixed 2 days after treatment with either suv1m or suv1. Immunofluorescence was used to detect tubulin and survivin (green and blue, respectively, in the merged images). DNA was Fig. 2. Depletion of survivin elicits nuclear abnormalities. (A) The effect of stained with DAPI (red in the merged images). Cells in metaphase (a, e, aЈ, and survivin RNAi on cell-cycle distribution. RPE cells were either mock-treated or eЈ), anaphase (b, f, bЈ, and fЈ), early telophase (c, g, cЈ, and gЈ), and late treated with shRNAs (suv1 and suv1m). Cell-cycle distribution was analyzed by telophase (d, h, dЈ, and hЈ) are shown. Arrowheads point to misaligned flow cytometry 3 days after shRNA transfection. (B) Nuclear abnormalities in chromosomes, red arrows denote lagging chromosomes or chromatin bridges, cells treated with survivin RNAi. Immunostaining of cells was performed 3 days and white arrows identify visible cortex contraction. after transfection with suv1 (a–e) or suv1m (f). Tubulin was detected with fluorescein isothiocyanate-conjugated mouse anti-␤-tubulin antibody (green), and nuclei were stained with DAPI (red). White arrowheads and of anaphase and telophase cells displayed abnormalities in arrows identify mininuclei and DNA bridges, respectively. (C) Percentage of chromosome segregation not previously attributed to survivin cells with nuclear abnormalities illustrated in B. Each column represents the deficiency in mammalian cells, including lagging chromosomes average of three independent experiments, and each experiment was done in Ј Ј triplicate. More than 500 cells were counted in each measurement. Error bars and DNA bridges (Fig. 3 f and g and data not shown). The DNA represent SD. bridges failed to resolve during telophase, and this failure was presumably a means by which cells with one bilobed nucleus were produced. Mininuclei were evident in 25% of survivin-depleted DNA content increased from 11% to 33%, presumably reflect- cells during late telophase, likely originating from lagging chro- ing the accumulation of tetraploid cells containing two nuclei or mosomes (data not shown and Fig. 4f). one bilobed nucleus (Fig. 2A). This result is in contrast with a Lagging chromosomes during late mitosis could be caused by previous report on survivin depletion in the aneuploid U2OS factors such as chromosome fragmentation, defective chromo- and HeLa cell lines (7). After exiting mitosis, the cells at first some condensation, absence of sister-chromatid cohesion, sister- remained mononucleate, then became multinucleate. The chromatid nondisjunction, or merotelic attachment in which a binucleation and bilobed nuclei reported here with euploid cells were not observed. We also found that 30% of suv1-treated cells had mininuclei, and 5% displayed deformed nuclei (Fig. 2 B and C). The frequencies of the abnormalities were Ն17-fold higher than those of suv1m-treated cells. A second survivin RNAi, suv2, produced proliferative arrest and nuclear abnormalities similar to those caused by suv1 (data not shown). The abnormal number, size, and morphology of nuclei indicate that survivin depletion caused defects in mitosis and͞or cytokinesis.

Survivin Depletion Causes Abnormal Behavior of Mitotic Chromo- somes and Centromeres. The presence of mininuclei and bilobed nuclei in survivin-depleted cells suggested that segregation of mitotic chromosomes had been disturbed. To investigate this possibility, we examined the effect of survivin RNAi on chromosome behavior during mitosis (Fig. 3). Depletion of survivin Fig. 4. Mitotic centromere segregation is abnormal in cells depleted of compromised the ability of RPE cells to align metaphase chro- survivin. RPE cells were treated with suv1m (a and b) or suv1 (c–g). Kinetochore mosomes, generating cells that had partial metaphase plates with proteins were detected with the CREST antibody and a ﬂuorescein isothio- Ј cyanate-conjugated secondary antibody (green). DNA was stained with DAPI one or more incorrectly aligned chromosomes (Fig. 3e ), con- (red). Cells in metaphase (g), anaphase (a and c), telophase (b, d, and f), and sistent with recent reports (6, 7). Nevertheless, survivin-depleted interphase (e) are shown. White arrows point to lagging kinetochores, red RPE cells regularly reached anaphase and telophase (Fig. 3 arrows identify paired kinetochores at the metaphase plate, and arrowheads fЈ–hЈ), in contrast with previous observations (6, 7). About 70% denote misaligned paired kinetochores

15102 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0406665101 Yang et al. Downloaded by guest on September 29, 2021 sister kinetochore is connected to microtubules from both poles of the spindle (16). Because we did not see obvious defects in chromosome condensation, we monitored centromeres by staining kinetochore proteins with the human autoimmune serum CREST (Fig. 4 c, d, and f). Survivin-depleted cells displayed typical centromere doublet signals during metaphase, as in mock-treated or suv1m-treated cells (Fig. 4g and data not shown), indicating that sister-chromatid cohesion was not affected by depletion of survivin. We then examined centromere segregation during anaphase. Centromeres were typically grouped into a tight cluster at the leading edge of the segregating chromatids in normal anaphase cells (Fig. 4 a and b). In contrast, one or more centromeres had been left in the middle of separated DNA masses in the majority of survivin-depleted cells (Fig. 4c and data not shown). Thus, the segregation of centromeres was disturbed in the absence of survivin. Because kinetochore proteins could always be detected on missegregated chromosomes by immunostaining with CREST antibody (Fig. 4 c, d, and f), we conclude that missegregated chromosomes were not acentromeric chromosome frag- ments. Furthermore, most lagging chromosomes displayed unpaired centromere signals (Fig. 4 c, d, and f). We suggest that the Fig. 5. The induction of p53 mediates the arrest of DNA synthesis elicited by survivin depletion. (A) The effect of survivin depletion on cell-cycle regulators. lagging chromosomes represent sister chromatids that have been RPE cells were treated with suv1 as described for Fig. 1B, and cell extracts were left behind, rather than the product of sister-chromatid nondis- prepared at indicated time points after shRNA transfection and loaded junction. This conclusion was also supported by the observation equally in each lane. Expression of indicated proteins was assessed by Western that Ͼ95% of mininuclei in interphase cells had unpaired blotting. (B) Induction of nuclear p53 by survivin RNAi. Immunoﬂuorescence centromere signals (Fig. 4e and data not shown). We conclude was performed 2 days after transfection of RPE cells with either suv1 (a and b) that survivin is not required for disjunction of sister chromatids. or suv1m (c and d). The expression of p53 (a and c) was detected with anti-p53 Instead, it is apparently essential for accurate segregation of antibodies and an ﬂuorescein isothiocyanate-conjugated secondary antibody. sister chromatids, in apparent contrast with the previous report Nuclei were stained with DAPI (b and d). (C) Selective depletion of p53 and that survivin-depleted cells failed to segregate their sister chro- survivin proteins by their cognate shRNAs. RPE cells were transfected with (lanes 2–5) or without (lane 1) indicated shRNA(s) and collected at either 48 h matids (6, 7). (lanes 1–3 and 5) or 72 h (lane 4) after transfection. Western blotting was performed with antiserum against p53, survivin, or actin. (D) Percentage of Survivin Regulates Assembly and͞or Stability of Microtubule Struc- BrdUrd-positive cells. RPE cells were labeled with BrdUrd 72 h after transfec- ture During Late Mitosis. Accumulation of binucleate cells after tion with shRNA(s). Incorporated BrdUrd was visualized by immunostaining. treatment with survivin RNAi indicates that mitosis, but not CELL BIOLOGY cytokinesis, occurs in the absence of survivin. The mitotic spindle plays an essential role in chromosome segregation during those obtained by depleting survivin, including similar nuclear early mitosis, whereas microtubules of the spindle midzone and abnormalities, failure of cytokinesis, and arrest of DNA synthesis midbody assembled during late mitosis are crucial for cytokinesis (data not shown). We conclude that the phenotype elicited by (17). To understand the defects in cytokinesis that arise in the depletion of survivin is probably due to perturbed function of the absence of survivin, we examined microtubule structure in CPP complex. survivin-depleted cells. Centrosomes were well nucleated with astral microtubules Induction of p53 and p21 by Depletion of Survivin. Because DNA (data not shown), and the bipolar mitotic spindle was normal synthesis was arrested during the period that nuclear anomalies during early mitosis in cells depleted of survivin (Fig. 3 a and aЈ), were detected, we speculated that a checkpoint had been acti- consistent with previous reports (6, 7). Midzone microtubules vated to prevent DNA endoreduplication subsequent to abnor- were also normal during early anaphase (data not shown). This mal mitoses. To test this hypothesis, we examined the effect of finding was expected because the majority of chromosomes had survivin depletion on some known regulators of the cell cycle. a bipolar orientation at the metaphase plate and moved to the We found that suv1 elicited a sustained increase of both the spindle poles during anaphase (Figs. 3 and 4 and data not tumor suppressor protein p53 and its downstream target p21, a shown). In contrast, the midzone microtubules during late cell-cycle inhibitor (Fig. 5A). Analysis with immunofluorescence anaphase were either disorganized or absent in survivin- staining revealed p53 accumulation in survivin-depleted cells depleted cells, indicating that spindle elongation was impaired irrespective of their nuclear number or morphology (Fig. 5Ba). (Fig. 3 b and bЈ). The midbody microtubules were undetectable The induction of p53 was specific to survivin depletion, because in telophase cells (Fig. 3 c, d, cЈ, and dЈ). We suggest that cells neither suv1m nor Fluc gave rise to p53 expression (Fig. 5Bc and lacking the midbody microtubules correspond to those under- going failure in cytokinesis, because the microtubules are essen- data not shown). tial for completion of cytokinesis (17). In contrast with previous p21 apparently acts on the cell cycle by inhibiting the phos- reports (6, 7), we frequently observed considerable cortex phorylation of pRB by cyclin-cdks and by suppressing expression contraction bisecting between poles of a bipolar spindle (Fig. 3 of mitotic cyclins A and B1 (19). We found that phosphorylation gЈ and hЈ), indicating that the initiation of cytokinesis was not of pRB on Ser-780, an event critical for the G1–S transition, affected. We conclude that the involvement of survivin in diminished after suv1 treatment (Fig. 5A). Moreover, the ex- assembly͞stabilization of microtubules during late mitosis makes pression of mitotic cyclins A and B1 was undetectable in the protein essential for completion of cytokinesis by mamma- survivin-depleted cells, whereas the abundance of G1-type cyc- lian cells, as reported previously for nematodes (18). lins D1 and E remained constant. We conclude that the prolif- We also used RNAi to deplete RPE cells of aurora-B, another erative arrest caused by survivin depletion had characteristics of component of the CPP complex. The results closely resembled the activation of a p53-dependent checkpoint response.

Yang et al. PNAS ͉ October 19, 2004 ͉ vol. 101 ͉ no. 42 ͉ 15103 Downloaded by guest on September 29, 2021 survivin or p53 alone in RPE cells had no effect on either centrosome number or structure of the mitotic spindle (Fig. 6C a–c and data not shown), in contrast with previous reports (3, 10). However, combined deficiencies gave rise to a Ն10-fold increase of mitotic cells having supernumerary centrosomes and multipolar spindles (Fig. 6Cdand e and data not shown). We attribute the effect of the combined deficiencies to restoration of DNA replication by the p53 deficiency. Absence of the p53–p21 pathway can exacerbate the cell-division defects caused by survivin depletion by allowing DNA endoreduplication without completion of cytokinesis. Discussion Survivin Is Required for Equal Segregation of Sister Chromatids. We have used RNAi in euploid human cells to obtain evidence that survivin is essential for accurate sister-chromatid segregation but not for sister-chromatid disjunction. Our finding is in apparent contrast with the report that survivin-depleted HeLa cells and U2OS cells entirely lacked sister-chromatid segregation (6, 7). We suggest that the more severe defect in chromosome behavior reported by others may be attributable to abnormal ploidy and͞or mutations in the tumor cells used in the experiments. We attribute the anaphase-lagging chromosomes observed in our experiments to merotelic kinetochore attachments, because lagging chromosomes represented sister chromatids that were Fig. 6. Survivin depletion causes multinucleation in cells depleted of p53. (A) often left near the spindle equator in anaphase. The aurora-B Multinucleation elicited by simultaneous elimination of p53 and survivin. RPE kinase, like its homolog IpI in yeast, has been reported to cells were either mock-transfected (a) or transfected with the indicated shRNA promote bipolar kinetochore attachments by destabilizing mero- against p53 (b), survivin (c), or both p53 and survivin (d). Five days later, cells telic kinetochore attachments (20, 21). Because the kinetochore were ﬁxed, stained for nuclei and cytoplasm, and examined by phase contrast localizations of aurora-B and survivin are interdependent (refs. ϫ microscopy (400 ). Arrows in c and d identify binucleation and multinucle- 6 and 7 and Fig. 7, which is published as supporting information ation respectively. (B) Quantiﬁcation of nuclear abnormalities. The histogram on the PNAS web site), we propose that survivin ensures the shows the percentage of cells with more than two nuclei after the shRNA treatment described for A. Each column represents the average of three bipolar attachment of kinetochores by being essential for as- independent experiments (four measurements each). Three hundred cells sembly of the CPP complex at centromeres in early mitosis. were counted in each measurement. (C) Abnormalities of the centrosome number and mitotic spindle. Cells were treated with RNAi as follows: a, mock; Survivin Is Required for Assembly͞Stabilization of the Central Spindle b, p53sh1; c, suv1; d and e, p53sh1 and suv1. Cells were stained for survivin in in Late Mitosis. Survivin has been implicated in regulating the red (a–e), ␥-tubulin (a–d) and ␤-tubulin (e) in green, and DNA (a–e) in blue. assembly of microtubules, because survivin-null mouse embryos One metaphase cell is shown in each panel. lacked the mitotic spindle and spindle midzone microtubules (22). Injection of survivin antibodies into HeLa cells has been reported to cause shortened mitotic spindles and abnormal Arrest of DNA Synthesis in Response to Depletion of Survivin Is multipolar spindles (11). We found that survivin depletion in Mediated by p53. To determine whether the p53-p21 pathway was RPE cells had no apparent effect on the mitotic spindle in early responsible for the DNA synthesis arrest induced by survivin mitosis, in accord with two recent reports for HeLa cells (6, 7). depletion, RPE cells were subjected to RNAi against both p53 Instead, the spindle midzone and the midbody microtubules were and survivin (Fig. 5C). The p53 levels induced by depletion of absent during the late mitosis of survivin-deficient RPE cells, an survivin were suppressed by p53 RNAi. We monitored DNA abnormality not previously associated with survivin deficiency in synthesis by BrdUrd incorporation in RNAi-treated cells (Fig. cultured mammalian cells. 5D). More than 90% of p53-depleted cells were BrdUrd-positive, Survivin depletion in RPE cells disrupted association of the Ͻ whereas 10% of survivin-depleted cells had BrdUrd incorpo- chromosomal passenger proteins inner centromere protein and ration. However, depletion of p53 in the absence of survivin aurora-B and the spindle midzone motor protein MKLP-1 with largely restored DNA synthesis (Fig. 5D) but failed to restore the spindle midzone, the equatorial cortex, and the midbody proliferation (data not shown). Thus, p53 was required for (Fig. 7 and Supporting Text, which is published as supporting inhibition of DNA synthesis caused by depletion of survivin, yet information on the PNAS web site). These findings suggest that survivin itself was not required for DNA synthesis. the CPP complex is required for the recruitment of MKLP-1 to organize microtubules of the spindle midzone and midbody in Depletion of p53 Exacerbates the Effect of Survivin Depletion on mammalian cells, as in nematodes (23, 24), and that failure of this Ploidy. Persistence of DNA synthesis in the presence of combined recruitment is at least partially responsible for the defects in deficiencies of survivin and p53 raises the possibility of increased cytokinesis displayed by survivin-depleted RPE cells. ploidy. Consistent with that prediction, we found that Ն10% of cells had more than two nuclei 96 h after introduction of shRNAs Survivin Is Required for Cellular Proliferation but Not Cell Survival. against both p53 and survivin (Fig. 6 A and B). In contrast, Survivin has been proposed to be an inhibitor of apoptosis survivin RNAi alone caused binucleation, whereas depletion of protein that is required to protect mitotic cells against a default p53 alone had no effect. The size of cells and of each nucleus in apoptotic program (3, 4). Our finding that RPE and IMR-90 cells multinucleated cells was increased in comparison with that after can survive despite abnormal mitoses caused by survivin deple- treatment with RNAi for either survivin or p53 alone. tion supports neither the existence of such a default apoptotic Abnormal mitoses may arise from abnormalities of centro- pathway nor an essential role for survivin to protect cells somes or mitotic spindles. We found that depletion of either suffering abnormal mitoses. Two recent studies have failed to

15104 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0406665101 Yang et al. Downloaded by guest on September 29, 2021 reproduce the substantial apoptosis of HeLa cells described played multinucleate giant cells, and that survivin loss in mouse previously as a consequence of survivin depletion (6, 7), but the thymocytes triggered a p53-dependent accumulation of cells possibility of incomplete depletion of survivin has been raised as with 2N DNA content (22, 28). Thus, it seems likely that the role an explanation for the absence of apoptosis (10). Because the of survivin varies with cell type and context, even in normal loss of survivin in many cancer cell lines has been associated with settings. apoptosis, further studies are needed to understand the cellular In conclusion, we have shown that in the absence of survivin, context that makes survivin essential for cell survival. certain euploid human cells suffer missegregation of chromo- Our findings indicate that, in at least some euploid human somes, abortive assembly of microtubules late in mitosis, failure cells, survivin is essential for cellular proliferation but not for cell of cytokinesis, and arrest of DNA synthesis. Cessation of DNA survival. Survivin-deficient cells were arrested in G1 in a p53- synthesis appears to be dependent on a postmitotic checkpoint dependent manner immediately after a defective cell division. A pathway involving p53. We attribute the defects in mitosis and p53-dependent G1 arrest after failure of cytokinesis has been cytokinesis to disruption of the CPP complex. We found no documented in cells treated with either spindle toxins or inhib- evidence that survivin regulates centrosome duplication, initial itors of actin assembly (25, 26). It has been proposed that assembly of the mitotic spindle, or cortex contraction. These tetraploidy triggers p53-dependent arrest in these contexts (26). findings are in contrast with several previous reports about the However, a recent study indicates that p53-dependent arrest is role of survivin in cell survival, cytokinesis, segregation of sister not triggered by binucleation, polyploidy, multiple centrosome, chromatids, centrosome duplication, and the mitotic spindle (3, or failure of cytokinesis (27). Our finding that p53 was induced 6, 7, 10, 11). We attribute the discrepancies to the use of different in cells with both normal and abnormal number nuclei after cell lines that vary in their ploidy and genotype. Despite the depletion of survivin is consistent with the latter report. Because discrepancies, it appears that survivin is essential to both chro- survivin is not required for DNA synthesis, we speculate that mosome segregation and cytokinesis in various mammalian cells. disruption of the microtubule cytoskeleton during cytokinesis by It seems reasonable to suspect that abnormal expression or depletion of survivin may generate a signal to activate p53 and function of survivin might contribute to tumorigenesis. cause G1 arrest. Activation of a p53-dependent mitotic checkpoint after de- We thank Ron Vale and David Morgan for advice on the manuscript; pletion of survivin has been reported in ref. 10. That study was Francoise Chanut, Sue Kim, Kevin Hill, and Jian Qu for help in preparation of the manuscript; and B. R. Brinkley, T. Stukenberg, and not designed to distinguish a G2–M arrest from a postmitotic W. C. Earnshaw for antibodies. This work was supported by National arrest, so it is possible that p53 was mediating a postmitotic Institutes of Health Grant CA44338 (to J.M.B.) and funds from the checkpoint like that observed here. Our studies are in apparent G. W. Hooper Research Foundation. D.Y. is supported by a postdoctoral contrast with reports that survivin-null mouse embryos dis- fellowship from the Susan G. Komen Breast Cancer Fund.

1. Altieri, D. C. (2003) Nat. Rev. Cancer 3, 46–54. 14. Mahotka, C., Wenzel, M., Springer, E., Gabbert, H. E. & Gerharz, C. D. (1999) 2. Ambrosini, G., Adida, C. & Altieri, D. C. (1997) Nat. Med. 3, 917–921. Cancer Res. 59, 6097–6102. 3. Li, F., Ackermann, E. J., Bennett, C. F., Rothermel, A. L., Plescia, J., Tognin, 15. O’Connor, D. S., Grossman, D., Plescia, J., Li, F., Zhang, H., Villa, A., Tognin, 1, 97,

S., Villa, A., Marchisio, P. C. & Altieri, D. C. (1999) Nat. Cell Biol. S., Marchisio, P. C. & Altieri, D. C. (2000) Proc. Natl. Acad. Sci. USA CELL BIOLOGY 461–466. 13103–13107. 4. Li, F., Ambrosini, G., Chu, E. Y., Plescia, J., Tognin, S., Marchisio, P. C. & 16. Tanaka, T. U. (2002) Curr. Opin. Cell Biol. 14, 365–371. Altieri, D. C. (1998) Nature 396, 580–584. 17. Straight, A. F. & Field, C. M. (2000) Curr. Biol. 10, R760–R770. 5. Adams, R. R., Carmena, M. & Earnshaw, W. C. (2001) Trends Cell Biol. 11, 18. Speliotes, E. K., Uren, A., Vaux, D. & Horvitz, H. R. (2000) Mol. Cell. 6, 49–54. 211–223. 6. Carvalho, A., Carmena, M., Sambade, C., Earnshaw, W. C. & Wheatley, S. P. 19. Taylor, W. R. & Stark, G. R. (2001) Oncogene 20, 1803–1815. (2003) J. Cell. Sci. 116, 2987–2998. 20. Lampson, M. A., Renduchitala, K., Khodjakov, A. & Kapoor, T. M. (2004) Nat. 7. Lens, S. M., Wolthuis, R. M., Klompmaker, R., Kauw, J., Agami, R., Brum- Cell Biol. 6, 232–237. melkamp, T., Kops, G. & Medema, R. H. (2003) EMBO J. 22, 2934–2947. 21. Tanaka, T. U., Rachidi, N., Janke, C., Pereira, G., Galova, M., Schiebel, E., 8. Fortugno, P., Wall, N. R., Giodini, A., O’Connor, D. S., Plescia, J., Padgett, Stark, M. J. & Nasmyth, K. (2002) Cell 108, 317–329. K. M., Tognin, S., Marchisio, P. C. & Altieri, D. C. (2002) J. Cell Sci. 115, 22. Uren, A. G., Wong, L., Pakusch, M., Fowler, K. J., Burrows, F. J., Vaux, D. L. 575–585. & Choo, K. H. (2000) Curr. Biol. 10, 1319–1328. 9. Jiang, X. R., Jimenez, G., Chang, E., Frolkis, M., Kusler, B., Sage, M., Beeche, 23. Kaitna, S., Mendoza, M., Jantsch-Plunger, V. & Glotzer, M. (2000) Curr. Biol. M., Bodnar, A. G., Wahl, G. M., Tlsty, T. D. & Chiu, C. P. (1999) Nat. Genet. 10, 1172–1181. 21, 111–114. 24. Severson, A. F., Hamill, D. R., Carter, J. C., Schumacher, J. & Bowerman, B. 10. Beltrami, E., Plescia, J., Wilkinson, J. C., Duckett, C. S. & Altieri, D. C. (2004) (2000) Curr. Biol. 10, 1162–1171. J. Biol. Chem. 279, 2077–2084. 25. Lanni, J. S. & Jacks, T. (1998) Mol. Cell. Biol. 18, 1055–1064. 11. Giodini, A., Kallio, M. J., Wall, N. R., Gorbsky, G. J., Tognin, S., Marchisio, 26. Andreassen, P. R., Lohez, O. D., Lacroix, F. B. & Margolis, R. L. (2001) Mol. P. C., Symons, M. & Altieri, D. C. (2002) Cancer Res. 62, 2462–2467. Biol. Cell 12, 1315–1328. 12. Yang, D., Buchholz, F., Huang, Z., Goga, A., Chen, C.-Y., Brodsky, F. M. & 27. Uetake, Y. & Sluder, G. (2004) J. Cell Biol. 165, 609–615. Bishop, J. M. (2002) Proc. Natl. Acad. Sci. USA 99, 9942–9947. 28. Okada, H., Bakal, C., Shahinian, A., Elia, A., Wakeham, A., Suh, W. K., 13. Kalman, D., Weiner, O. D., Goosney, D. L., Sedat, J. W., Finlay, B. B., Abo, Duncan, G. S., Ciofani, M., Rottapel, R., Zuniga-Pflucker, J. C. & Mak, T. W. A. & Bishop, J. M. (1999) Nat. Cell Biol. 1, 389–391. (2004) J. Exp. Med. 199, 399–410.

Yang et al. PNAS ͉ October 19, 2004 ͉ vol. 101 ͉ no. 42 ͉ 15105 Downloaded by guest on September 29, 2021