p31^comet Induces Cellular Senescence through p21 Accumulation and Mad2 Disruption Free

During cell division, sister chromatids are separated and distributed to each emerging daughter cell. Microtubules initially attach to generate accurate tension between the bio-oriented kinetochores (1, 2) and pull each chromatid to the respective daughter cell. This process requires reliable orchestration to ensure survival over generations. Consequently, cells are devised with mechanisms to fix and recover errors generated during division. One of these mechanisms is the spindle checkpoint, which is activated during mitosis to monitor attachment of sister chromatids to the spindle. When the spindle checkpoint is on, Mad2 is recruited to kinetochores and released in a modified form that interacts with Cdc20 (3) to form an inhibitory complex regulating APC/C-mediated ubiquitination of securin (4, 5). Following the completion of spindle attachment, Mad2 dissociates from the Cdc20-APC complex to silence the spindle checkpoint (6).

p31^comet was initially identified as a Mad2-interacting protein (7). It is proposed that p31^comet facilitates dissociation of Mad2 by transient interactions with the inhibitory Mad2-Cdc20-containing complexes to allow transition from the metaphase to anaphase during mitotic checkpoint inactivation. Other reports suggest that p31^comet enhances the activity of APC/C, which is otherwise inhibited by Mad2, without disrupting Mad2-Cdc20 binding in the transient Mad2-Cdc20-APC/C complex (8, 9). Both hypotheses concur that p31^comet counteracts Mad2 function and is required for silencing the spindle checkpoint.

Suppression of spindle checkpoint function is invariably lethal because of inadequate chromosomal segregation. For instance, functional inactivation of the BubR1 or Mad2 stimulates apoptotic cell death (10, 11). Mouse embryos deficient in BubR1 or Mad2 fail to survive because of extensive apoptosis (12, 13). Moreover, mutation of Bub1 leads to chromosomal missegregation and failure to block apoptosis in Drosophila (14). These findings collectively illustrate that apoptosis derived from the loss of spindle checkpoint function contributes to protection against the emergence of an aberrant cell population.

Although cellular senescence, together with apoptosis, is a major additional safeguard mechanism, direct evidence that functional attenuation of spindle checkpoint proteins activates the senescence pathway has recently been found by using a mouse model system. The reduced amounts of the BubR1 protein result in the early onset of cellular senescence and accelerate the aging process of mice (15, 16). The combined alteration by Bub3 and Rae1 of the mitotic checkpoint also prominently affects the cellular and organismal processes of senescence and aging (17). Based on alterations in cellular physiology, it has been established that spindle checkpoint proteins are closely connected with cellular senescence. Several anticancer drugs targeting the microtubule, such as Taxol and vincristine, induce senescence-like phenotypes due to abnormal mitosis in a variety of tumor cells (18-20). Senescence of a fibrosarcoma cell line triggered by p21^Waf1/Cip1 is associated with depletion of the cellular pools of mitotic control proteins, such as Mad2, BubR1, cyclin B1, and Cdc2 (21). The induction of senescence-like phenotypes in cancer cells is accompanied by mitotic catastrophe (21). These various plausible interconnections between spindle checkpoint and cellular senescence request to identify their connecting molecules, the elucidation of which can enable to explain the physiologic details.

Based on the interactions between p31^comet and Mad2, we propose that p31^comet functions in inducing cellular senescence. Our data show that p31^comet overexpression is associated with senescence-like morphologic changes and senescence-associated β-galactosidase (SA-β-gal) expression. Induction of senescence by p31^comet is exclusively dependent on disruption of Mad2 and accumulation of p21^Waf1/Cip1. These findings contribute significantly toward establishing the molecular mechanism of senescence derived from abnormal mitosis, which may be exploited to eliminate tumor cells effectively.

To analyze the possible functional effects of p31^comet, we introduced the full-length human gene (7) into a bicistronic retroviral expression system coexpressing enhanced green fluorescent protein (EGFP) or puromycin as a selectable marker (Supplementary Fig. S1A). The resulting constructs, termed p31^comet-IRES-EGFP and p31^comet-IRES-puro, were retrovirally transduced into A549. The infectivity of p31^comet-IRES-EGFP and corresponding control-IRES-EGFP retroviral constructs was >89% as determined by counting cells displaying green fluorescence (Supplementary Fig. S1B).

Overexpression of p31^comet in A549 cells led to enlargement of cells with a flattened shape (Fig. 1A) and multinuclei formation in a cell (Fig. 1B), which are the general characteristics of cellular senescence. Accordingly, we determined whether the p31^comet-mediated morphologic changes are due to cellular senescence. SA-β-gal activity was measured as a biomarker for cellular senescence in cells transduced with p31^comet. As predicted, p31^comet overexpression was associated with increased SA-β-gal activity (Fig. 1B). The p31^comet-mediated SA-β-gal activity was evident 4 to 6 days after transduction and reached a level of 78% after 9 days, whereas control vector transduction failed to induce SA-β-gal activity (Fig. 1C). Among the β-gal-positive cells, 92% contained multiple nuclei (n ≥ 2), whereas only 8% had a single nucleus. The maximal percentage of SA-β-gal-positive cells with multiple nuclei was recorded at 6 days after p31^comet transduction and remained constant thereafter. Live cell imaging showed that p31^comet inhibited cytokinesis during telophase (Supplementary Fig. S2), further supporting the idea that p31^comet activates a pathway that results in accumulation of multinucleate cells. An increase in granularity is generally caused by accumulation of lipofuscin and other lipoprotein granules in the cytoplasm, which is also the characteristic of senescent cells (22). Granularity, presented as a 90° light scatter in flow cytometry analysis, was dramatically increased following p31^comet transduction (Fig. 1D). To exclude the possibility that only extremely high levels of p31^comet can induce senescence, we gradually reduced the level of p31^comet overexpression by adding p31^comet small interfering RNA (siRNA) in a dose-dependent manner to cells receiving p31^comet. This experiment showed that p31^comet could induce senescence even when the overexpression level of the protein was reduced up to 5-fold (Supplementary Fig. S3). Thus, our present results clearly indicate that p31^comet induces senescence-like phenotypes in A549 cells along with multinucleation. Similar p31^comet-mediated acquisition of SA-β-gal activity, flattened shape and multinucleation were evident in human lung cancer cell line Calu-1 (Supplementary Fig. S4) and human osteosarcoma cell line U2OS (Fig. 2A; data not shown). However, large amounts of HeLa and Hep3B cells underwent cell death, including apoptosis and necrosis, on p31^comet overexpression (Fig. 2B). In contrast to plenty of the molecular association between abnormal spindle checkpoints and apoptosis, there is a limited understanding of its molecular link directly to senescence, as it has recently been reported in the cases of BubR1 insufficiency (15, 16) and Bub3/Rae1 double haploinsufficiency (17). In the present study, therefore, we focused on p31^comet-induced senescence using a highly susceptible A549 cell line.

Next, we examined whether p31^comet-mediated senescence influences cellular proliferation. Transduction of p31^comet led to a significant reduction in cellular proliferation, whereas the control vector did not affect the growth rate (Fig. 3A). In a clonal cell growth assay, p31^comet had a profound effect on colony formation. Cells transduced with the p31^comet-IRES-EGFP retroviral vector formed only 25 colonies, none of which were EGFP-positive, indicating that the cells transduced with p31^comet did not form colony. In contrast, control-IRES-EGFP retroviral vector transduction produced 436 colonies, 401 of which were EGFP-positive (Fig. 3B). In view of this finding, we propose that p31^comet elicits complete regression of colony formation. Flattened cells were subcultured for another 30 days to determine whether arrested cells ultimately undergo death. Interestingly, a large proportion of cells (80-90%) were metabolically active over the test period (data not shown). Evidently, p31^comet overexpression hampers cell growth, eventually leading to permanent arrest.

Senescence is generally accompanied by the induction of tumor suppressor and cell cycle inhibitor proteins, such as p53 (23, 24), p16^Ink4a (25), p21^Waf1/Cip1 (26), and p27^Kip1 (27). Accordingly, we conducted Western blot analyses to determine whether p31^comet-induced senescence involves the participation of tumor suppressors and cell cycle inhibitors. p31^comet expression, which is very low at the basal level, was visible 24 h after transduction and sustained over the 8-day test period (Fig. 4A). Overexpression of p31^comet led to elevation of the p21^Waf1/Cip1 protein level, which was evident 2 days after the transduction. The p53 tumor suppressor protein, a transcriptional regulator of p21^Waf1/Cip1 (28), was also significantly increased over a similar period. In contrast, we observed no changes in the expression pattern of p27^Kip1 protein, another known cell cycle inhibitor associated with senescence (29). Consistent with previous reports, we could not detect p16^Ink4a protein that is defective in the A549 cell line (30). The present findings indicate that p31^comet-mediated senescence is closely associated with p53 and p21^Waf1/Cip1 accumulation.

In addition to cell cycle inhibitors, cells undergoing senescence produce several common marker proteins, such as osteonectin, PAI-1, and SM22 (31, 32). To evaluate whether these marker proteins are stimulated in p31^comet-mediated senescence, we examined their mRNA levels by reverse transcription-PCR after retroviral transduction. The mRNA levels of these markers were markedly elevated over the monitoring period in p31^comet-transduced cells in contrast to little or no changes in control cells (Fig. 4B). We propose that p31^comet overexpression triggers senescence with typical characteristics.

As mentioned above, both p53 and p21^Waf1/Cip1 were up-regulated in p31^comet-induced senescent A549 cells (Fig. 4A). Because p21^Waf1/Cip1 and p53 are cell cycle inhibitors, we predicted that the up-regulation of one or both of these proteins contributes to senescence. To confirm this theory, we examined whether down-regulation of either protein affects senescence induced by p31^comet. Cells were transfected with p21^Waf1/Cip1 or p53 siRNA at 48 h after transduction with p31^comet. The resulting protein levels were successfully decreased (Fig. 5A). Depletion of p21^Waf1/Cip1 completely blocked p31^comet-induced senescence as revealed by dramatic reversion in the percentage of SA-β-gal-positive cells to the basal level (Fig. 5B). Evidently, induction of p21^Waf1/Cip1 is a critical step in p31^comet-mediated senescence. However, p53 depletion induced a smaller decrease in the percentage of SA-β-gal-positive cells, indicating a limited effect on p31^comet-induced senescence (Fig. 5B). Depletion of p53 additionally led to a decrease in the p21^Waf1/Cip1 protein level to some extent, consistent with the finding that p21^Waf1/Cip1 is a downstream target of p53. Thus, the observed effect of p53 depletion on p31^comet-induced senescence is possibly attributable to a concurrent reduction in levels of p21^Waf1/Cip1.

Next, we performed cell death experiments to determine if p31^comet expression might give rise to a cell population undergoing death. Cell death analysis using trypan blue staining (data not shown) and Annexin V/7-amino-actinomycin D staining analysis by fluorescence-activated cell sorting revealed that, whereas p31^comet induced mostly cellular senescence (Fig. 1C), p31^comet expression also resulted in the death of ∼8% to 15% of cells during the interval of 5 days after transduction (Fig. 2B; data not shown). In the experiments involving p21^Waf1/Cip1 or p53 depletion described above, however, we found a marked phenotypic conversion of adherent senescent cells into floating dead cells. As anticipated from the clear reduction in the senescent population, both p21^Waf1/Cip1 and p53 depletion led to noticeable increases in cell death; in addition, p21^Waf1/Cip1 depletion resulted in more cell death than did p53 depletion (Fig. 5C and D). Specifically, dead cells after p21^Waf1/Cip1 depletion were composed of 21.5% necrotic and 34.5% apoptotic cells 6 days after p31^comet transduction. A malfunction of mitotic checkpoint regulators generates massive chromosomal and nuclear morphologic changes; this process is termed mitotic catastrophe (33, 34). As shown in Fig. 6A, together with multinuclei, micronuclei and anaphase bridges, which are the typical phenomenon of mitotic catastrophe, were seen in response to p31^comet overexpression. The frequency of anaphase bridge formation was ∼8.3% after 1 day and 15.6% after 3 days following p31^comet transduction (Fig. 6B). Moreover, p31^comet caused a centrosome imbalance, in which cells receiving p31^comet had a higher number of centrosomes than cells with empty vectors (Supplementary Fig. S5). These findings indicate that p31^comet-induced senescence is accompanied by mitotic catastrophe with massive nuclear and chromosomal abnormality. The effect of p21^Waf1/Cip1 depletion was negligible on the formation of these nuclear and chromosomal aberrations. Thus, our present observation on the results of p21^Waf1/Cip1 or p53 depletion is that p21 is a major mediator in blocking phenotypic conversion of p31^comet-induced senescence to cell death through mitotic catastrophe. Thus, p21 functions as a negative regulator of cell death through mitotic catastrophe.

We conducted fluorescence-activated cell sorting analyses to monitor changes in the DNA content of cells infected with the p31^comet retrovirus. The proportion of cells (62%) in the G₀-G₁ phase decreased gradually to 58% and 33.5% at 4 and 9 days after p31^comet transduction, respectively (Fig. 7A and B). In contrast, a marked increase in the hyperploid cell population (>4N DNA) was evident (Fig. 7A and B). The nuclear and chromosomal aberration analysis showing the production of multinuclei (Figs. 1B and C and Fig. 6A), anaphase bridges (Fig. 6A), and centrosome imbalance (Supplementary Fig. S5), which are the evidences of ploidy changes, further supports the emergence of the hyperploid population caused by p31^comet. The increased hyperploidy was not observed in cells infected with control retrovirus. The data indicate that multinucleation due to p31^comet overexpression is part of the mechanistic basis for senescence induction. The majority of the hyperploid population underwent cellular senescence with the acquisition of SA-β-gal activity and maintained metabolic activity (data not shown). In addition to increase of hyperploid population, we found a small fraction of sub-G₀-G₁ peaks (Fig. 7A and B), further supporting that apoptotic cell death is also occurring as assessed in Figs. 2B and 5C.

Finally, to verify the endogenous expression of mitotic checkpoint regulators by p31^comet, we analyzed, over time, cells induced with p31^comet. We found that Mad2 and cyclin B1 expression gradually decreased, but there was no change in BubR1, p55Cdc, or Aurora B expression (Fig. 7C). When the expression of p21^Waf1/Cip1 and Mad2 (Figs. 4A and 7C) were compared, we noticed that p21^Waf1/Cip1 up-regulation preceded Mad2 down-regulation in p31^comet-induced senescence. Therefore, we examined whether p31^comet-induced Mad2 down-regulation might be dependent on p21^Waf1/Cip1. As shown in Fig. 7D, p21^Waf1/Cip1 depletion restored Mad2 expression that has been down-regulated by p31^comet, indicating that p21^Waf1/Cip1 is required for Mad2 down-regulation during p31^comet-induced senescence. Additionally, we show that Mad2 depletion itself also led to p21^Waf1/Cip1 up-regulation as caused by excess p31^comet (Supplementary Fig. S6), implying that the endogenous level of Mad2 successfully inhibits the induction of p21^Waf1/Cip1 expression. Thus, excess p31^comet appears to mimic Mad2 depletion via their interaction, thereby inactivating the ability of Mad2 to inhibit p21^Waf1/Cip1 induction. Accordingly, we conclude that p31^comet triggers senescence through p21^Waf1/Cip1 accumulation and downstream events that follow from Mad2 down-regulation.

At present, p31^comet is the recognized binding and regulatory partner of Mad2 that activates the mitotic checkpoint (7). Thus, we determined whether inactivation of Mad2 by complex formation with p31^comet is involved in senescence. Mad2 expression was initially depleted by siRNA transfection to analyze whether its down-regulation mimics conditions of p31^comet overexpression in inducing cellular senescence. To avoid the activation of the IFN system by siRNA (35), three different Mad2 siRNAs, which can deplete >80% of the basal level (Fig. 8A; data not shown), were applied and SA-β-gal activity was measured. As with p31^comet overexpression, the introduction of all three Mad2 siRNAs led to acquisition of SA-β-gal activity in A549 cells (Fig. 8B and D). In experiments using M2-2, a Mad2 siRNA, cell growth inhibition (Fig. 8E) and p21^Waf1/Cip1 accumulation (Supplementary Fig. S6) were clearly observed. The similar consequences of p31^comet overexpression and Mad2 down-regulation support our hypothesis that inhibition of Mad2 function is predominantly involved in the mechanism of p31^comet-induced senescence. To further verify these results, p31^comet mutants defective in interactions with Mad2 were employed. We designed a retroviral construct for a p31^comet mutant lacking 11 amino acids (residues 55-65) of the Mad2 binding site, denoted p31ΔM2, and confirmed expression at the appropriate level and size (Fig. 8C). In contrast to wild-type p31^comet, p31ΔM2 failed to induce senescence phenotypes, such as SA-β-gal activity (Fig. 8D), cell growth retardation (Fig. 8E), and flattened shape (data not shown). In addition, this mutant did not effectively down-regulate Mad2 expression compared with p31^comet wild-type (Fig. 8F), indicating that interaction between p31^comet and Mad2 is required for Mad2 down-regulation. To further delineate whether the observed senescence might be derived from the fluctuation of p31^comet content in a cell, we additionally examined the effect of p31^comet depletion. For this purpose, three different p31^comet siRNAs were introduced into A549 cells (Supplementary Fig. S7A), and changes in SA-β-gal activity were monitored. On p31^comet depletion, there was no evidence of a phenotype displaying characteristics of senescence, such as SA-β-gal activity (Supplementary Fig. S7B) or cell growth retardation (Supplementary Fig. S7C). Finally, to confirm whether senescence by p31^comet results from Mad2 inhibition, we coinduced Mad2 with p31^comet. We found that ∼2% of cells induced with Mad2 and p31^comet were SA-β-gal-positive, which was in contrast to 78% of control cells induced with p31^comet alone (Fig. 8D). These findings further indicate that Mad2 is a key mediator of p31^comet-induced senescence. Therefore, we suggest that p31^comet may induce senescence by mediating Mad2 inhibition dependent on a direct interaction of p31^comet and Mad2.

As shown in Fig. 5, p21^Waf1/Cip1 depletion led to cell death instead of senescence. Accordingly, we analyzed the effect of Mad2 on this phenotypic change to determine whether the observed reduction in senescent cells as shown in Fig. 8D might be due to a concomitant increase in dead cells. Unlike the marked increase in dead cell population affected by p21^Waf1/Cip1 depletion, Mad2/p31^comet coexpression, Mad2 depletion alone, or deletion of the Mad2 binding site in p31^comet did not cause significant changes in the levels of cell death (data not shown). These findings together with data on the effects of p21^Waf1/Cip1 depletion in p31^comet-induced senescence suggest that phenotypic conversion of p31^comet-induced senescence to cell death is primarily due to p21^Waf1/Cip1 and not Mad2.

Accelerated senescence is an anticancer mechanism that inhibits unlimited cell proliferation, eventually leading to the irreversible arrest of tumor cell growth (24, 36). Senescence triggered by drugs, radiation, or oncogenes is occasionally accompanied by mitotic defects (21), but the mechanism underlying this relationship remains to be established. In this study, we show that overexpression of p31^comet leads to senescence-like phenotypes, such as increase in multinucleation and SA-β-gal activity (Fig. 1), permanent growth arrest (Fig. 3), and accumulation of senescence-associated marker proteins (Fig. 4) in the human cancer cell lines, A549 and Calu-1. Among the multinucleated cells (Fig. 1B; Supplementary Fig. S4B), binucleated cells were dominant, suggesting that mitotic defects leading to binucleated cells do not allow cells to divide and reenter the cycle. In addition to multinucleation, p31^comet generated massive chromosomal and nuclear abnormalities, including the formation of micronuclei and anaphase bridges, indicating that p31^comet-induced senescence was accompanied by mitotic catastrophe. In contrast to induction of senescence, p31^comet predominantly induced cell death, including apoptosis and necrosis, in other tumor cell lines such as HeLa and Hep3B (Fig. 2B). The cells undergoing death still exhibited mitotic catastrophe with multinucleation and micronuclei (data not shown). At present, the p31^comet-related mechanisms controlling cell susceptibility to senescence or apoptosis are unclear. Several DNA-damaging agents stimulate distinct responses of cancer cells depending on the dose, specifically apoptosis at high doses and senescence-like phenotype at low doses (32, 37). In view of the variations in susceptibility of specific cancer cells to drugs, low doses are sufficient to cause senescence through mitotic catastrophe but not to activate the apoptotic pathway. Similar to our present findings, recent studies on the molecular mechanisms of these phenotypic changes also indicate that p21^Waf1/Cip1 is a positive regulator of senescence and a negative regulator of mitotic catastrophe (38).

Prevailing evidence suggests that tumor suppressor and cell cycle inhibitor proteins are related to senescence induction and maintenance (24, 39). Overexpression of oncogenic ras leads to permanent cell cycle arrest in normal fibroblasts displaying distinct phenotypes from senescence (40). Our results show that permanent growth arrest by p31^comet differs mechanistically from that caused by oncogenic ras. Unlike oncogenic ras, p31^comet readily induces senescence independently of functional p53 (Figs. 4A and 5). Senescence by p31^comet, but not ras, is accompanied by obvious mitotic aberrations. The first direct link of abnormal mitotic checkpoints to senescence clearly defined tumor suppressors as key mediators of senescence induced by the attenuated function of mitotic checkpoint proteins. This observation rested on the surprising finding that the reduction of the BubR1 protein in natural aging is linked to senescence (15). In this context, when using mice insufficient for BubR1, p16^Ink4a acts as an effector and p19^ARF acts as an attenuator of senescence and aging (16), because the senescence- and aging-associated phenotype appears earlier through p16 depletion but later through p19 depletion. Our present study showed that, instead of accelerating or delaying senescence, p21 depletion reversed the p31^comet-induced phenotype from senescence to cell death without affecting mitotic catastrophe. This indicates that p21 is a major determinant of p31^comet-induced senescence through mitotic catastrophe. However, the elevation of p21 level is common in two other examples showing senescence through abnormal expression: BubR1 (15) and Bub3/Rae1 (17). Although these two events highlight the relatively early arrival of replicative senescence compared with that in wild-type cells, our present report emphasizes the immediate progression of senescence in tumor cells within one or two rounds of cell division after receiving p31^comet. In our analysis using untransformed human fibroblast cells, p31^comet overexpression led to immediate cell death rather than senescence.⁹

Mad2 plays a central role in regulation of the mitotic checkpoint via interactions with Cdc20, an essential accessory subunit of APC/C (4, 41-43). The importance of Mad2 in mitotic checkpoints is supported by reports that treatment with siRNA for Mad2 causes chromosome decondensation and premature sister chromatid separation, resulting in cell death (10, 44). It is plausible that p31^comet induces senescence through interactions with Mad2. In our present study, several lines of evidence clearly showed that p31^comet-induced senescence starts via Mad2 inhibition by forming complexes between them. First, the mutant lacking Mad2 binding region completely lost its ability to induce senescence. Second, both p31^comet overexpression and Mad2 depletion induced senescence in a similar manner. A prominent observation from the analysis of the relationship between p21^Waf1/Cip1 and Mad2 in p31^comet-induced senescence was that p21^Waf1/Cip1 is a key mediator of p31^comet-induced down-regulation of Mad2, an observation based on the fact that p21^Waf1/Cip1 depletion restored the level of Mad2 expression. This finding, together with additional result that Mad2 depletion alone also elicited p21^Waf1/Cip1 up-regulation as caused by excess p31^comet (Supplementary Fig. S6), suggests that the increased p21^Waf1/Cip1 levels via this new modulation cycle between p31^comet and Mad2 can in turn lead to Mad2 down-regulation. Therefore, p31^comet-mediated down-regulation of Mad2 makes it possible to sustain p21^Waf1/Cip1 up-regulation.

On microtubule attachment, the p31^comet protein specifically interacts with closed Mad2 but not open Mad2 (7, 8). Interactions between p31^comet and the Mad2-Cdc20-APC complex lead to inhibition of Mad2 activity and subsequent activation of APC by release from the inactive state, resulting in degradation of securin and cyclin B (7, 8). Closed Mad2-p31^comet interactions inhibit conformational changes by hindering further binding of open Mad2 to the Mad1-Mad2 core (45). Complete depletion of Mad2 results in mitotic defects by rendering the APC-Cdc20 complex free of Mad2 and dissociating Mad1 from the kinetochore followed by premature sister chromatid separation. On p31^comet expression, most Mad1-bound Mad2 molecules interact with p31^comet during early mitosis or G₁-G₂. This, in turn, prevents binding of Mad2 to Cdc20, a conveyable mitotic checkpoint signal, and induces continuous activation of APC/Cdc20, similar to that observed with complete Mad2 depletion. Thus, both p31^comet expression and complete Mad2 depletion produce similar outcomes. Specifically, these include p21^Waf1/Cip1 accumulation, mitotic defects, and chromosome decondensation followed by senescence.

Based on the data presented here and in previous reports, we hypothesize that p31^comet inhibits Mad2 function by direct interaction and accumulate p21^Waf1/Cip1. Abnormal APC activity and the resulting increased mitotic defects accumulate p53 and p21^Waf1/Cip1 proteins, and it finally resulted in senescence. Because the senescence pathway can be targeted for cancer treatment, p31^comet serves as an effective target molecule that can be exploited to induce senescence in tumor cells. Screening the chemical library for p31^comet up-regulation or enhanced interactions between p31^comet and Mad2 is a feasible approach.

Human cancer cell lines employed were cultured in Ham's F-12 (A549 cells), McCoy's medium (U2OS and Calu-1 cells), and MEM (Hep3B cells), respectively, supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin (Life Technologies) at 37°C in a humidified 5% CO₂ incubator.

p31^comet cDNA was amplified by PCR with sense (5′-ataaccATGGCGGCGCCGGAGGCG-3′) and antisense (5′-ccaggatccTCACTCGCGGAAGCCTTT-3′) primers using high-fidelity Taq polymerase (Takara). The capital letters represent nucleotides encoding p31^comet. The amplified product was digested with NcoI and BamHI and ligated into the corresponding sites in MFG-IRES-EGFP and MFG-IRES-puro retroviral vectors. The Mad2 binding domain was deleted to generate p31ΔM2 using the QuikChange Site-Directed Mutagenesis Kit (Stratagene) using the manufacturer's protocol. The following oligonucleotides were used for site-directed mutagenesis: forward primer 5′-CAGGAAGGCTGCTGTCAGTTTACT-3′ and reverse primer 5′-TCTTGGGCAAAAGGCCTCCGAAGC-3′. p31ΔM2 was additionally subcloned into MFG-IRES-EGFP and MFG-IRES-puro retroviral vectors.

The p31^comet antibody was generated in our laboratory by injecting purified GST-p31^comet into rabbit. Antibodies for p21^Waf1/Cip1 (SC-397), p27^Kip1 (SC-1641), p16^Ink4a (SC-759), p53 (SC-6243), and β-actin (SC-1616) were purchased from Santa Cruz Biotechnology, and Mad2 antibody (A300-300A) was obtained from Bethyl Laboratories.

Clonogenic cell survival was evaluated with a colony-forming assay. At 1 day after virus infection, cells were plated in 100 mm dishes and incubated at 37°C. For colony formation, cells were seeded at a density of 1.5 × 10³ per 10 mm dish. After 7 to 11 days of incubation, cells were fixed and stained with crystal violet. We counted total and EGFP-positive colonies containing >200 cells.

To generate a retrovirus-producing cell line, pMFG-p31^comet was introduced into the H29D retrovirus packaging culture by transient transfection using Lipofectamine (Invitrogen). After 72 h, supernatant fractions were harvested, and polybrene (Sigma) was added to a concentration of 6 μg/mL. Unwanted cells were removed by filtration through 0.4 μm pores. Virus titers, measured in a NIH3T3 cell line by counting EGFP-positive colony formation, were between 10⁵ and 5 × 10⁵ mL^-1 (retrovirus-IRES-EGFP).

Cells were stained for β-gal activity as described earlier (22). In brief, 5 × 10⁵ cells were seeded on a 60 mm plate 2 days after virus infection. After the indicated days, cells were washed twice with PBS, fixed in 2% formaldehyde and 0.2% glutaraldehyde for 5 min in PBS, and washed twice with PBS. This was followed by staining for 12 to 16 h in X-gal staining solution [1 mg/mL X-gal, 40 mmol/L citric acid/sodium phosphate (pH 6.0), 5 mmol/L potassium ferricyanide, 5 mmol/L potassium ferrocyanide, 150 mmol/L NaCl, 2 mmol/L MgCl₂]. After washing twice with PBS, cells were additionally stained with 4′,6-diamidino-2-phenylindole.

Total RNA was extracted using the RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol. RNA samples were reverse transcribed with random hexamers and the SuperScript First Strand Synthesis System (Life Technologies). cDNA was amplified using Takara Taq polymerase (Takara). The primer sets for osteonectin, SM22, and PAI-1 and PCR conditions are described elsewhere (32). The housekeeping gene β-actin (sense 5′-atcatgtttgagaccttcaacacccc-3′ and antisense 5′-catctcttgctcgaagtccagggcga-3′; products size: 317 bp) was used as an internal control for RNA loading.

Cells were washed twice with ice-cold PBS and lysed in TNN buffer [50 mmol/L Tris-HCl (pH 7.7), 150 mmol/L NaCl, 0.5% NP-40] containing protease inhibitors (10 mmol/L sodium fluoride, 2 mmol/L sodium orthovanadate, 1 mmol/L phenylmethylsulfonyl fluoride, 0.2 mmol/L EDTA, 1 mmol/L DTT, 10 μg/mL aprotinin). Lysates were cleared by centrifugation at 13,000 rpm for 10 min at 4°C. Protein concentrations were determined with the Bio-Rad protein assay kit. A 30 μg aliquot of total cell protein was subjected to 10% SDS-PAGE and transferred to Protran nitrocellulose transfer membrane (Schleicher & Schuell). After transfer, membranes were blocked in 5% milk in TBS-Tween 20 [10 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 0.05% Tween 20] for 30 min and incubated with the appropriate primary antibody in 5% milk in TBS-Tween 20 for 2 h at room temperature. Next, membranes were washed three times with TBS-Tween 20 and incubated for 1 h at room temperature in TBS-Tween 20 containing horseradish peroxidase-linked anti-immunoglobulin. After three washes in TBS-Tween 20, immunoreactive products were detected with an enhanced chemiluminescence system (Amersham).

At the indicated days after infection with retrovirus for p31^comet expression or control, floating and trypsinized cells were combined and then used for cell cycle analysis. The progression of cell cycle profile was analyzed by fluorescence-activated cell sorting using a Becton Dickinson FACSort flow cytometer. A cell cycle analysis program (CELLQuest; Becton Dickinson) was used to determine the percentage of cells at different stages of the cycle.

Cell death was determined by acridine orange-ethidium bromide double staining (46). Briefly, A549 cells transduced with p31^comet were costained with acridine orange and ethidium bromide and then examined under ×200 magnification using a fluorescence microscope. Counts were done by a reader blind to the experimental protocol. Cells were classified as viable, apoptotic, or necrotic population. To further investigate cell death, we used 7-amino-actinomycin D and Annexin V-PE double staining (47). A549 and HeLa cells transduced with p31^comet were costained with 7-amino-actinomycin D and Annexin V-PE, and dead cells were counted by flow cytometry.

p21^Waf1/Cip1 and p53 siRNAs were purchased from Santa Cruz Biotechnology. Two days after virus infection, 5 × 10⁵ cells were seeded in 60 mm plates containing 2.5 mL medium. Mad2 and p31^comet siRNAs were generated by Invitrogen. The siRNA sequences employed were as follows: 5-AATACGGACTCACCTTGCTTG-3 (M2-1), 5-AAGTGGTGAGGTCCTGGAAAG-3 (M2-2), and 5-AAAGTGGTGAGGTCCTGGAAA-3 (M2-3) for Mad2 and 5-AAGAGACTGCATGGTACCAGT-3 (p31-1), 5-AAGCTCTACGCAGGAACCTCTCA-3 (p31-2), and 5-AAGTCGAGTTCATAGAACTC C-3 (p31-3) for p31^comet. Transfection was done using Lipofectamine 2000 according to the manufacturer's instructions and repeated every 72 h for a maximum of two consecutive transfections. At the indicated time points, siRNA-treated cells were collected and used for various assays.

No potential conflicts of interest were disclosed.