p53-Independent Abrogation of a Postmitotic Checkpoint Contributes to Human Papillomavirus E6-Induced Polyploidy Free

The most common manifestation of genomic instability in solid tumors is aneuploidy (1), the state of having an abnormal number of chromosomes. A causal role for aneuploidy in tumorigenesis has not been fully established. Aneuploidy has been shown to originate from a tetraploid intermediate (refs. 2, 3 and references therein) and has been exemplified in the precancerous state of Barrett's esophagus cells (4). Significantly, a recent study with a large number of cervical samples showed that tetraploidy not only occurred as an early event during cervical carcinogenesis but also predisposed cells to aneuploidy (5). Also, a recent report by Fujiwara et al. (6) showed that p53-null mouse tetraploid but not diploid cells are highly competent to induce tumors with numerical and structural chromosome aberrations in nude mice.

Cells with intact spindle checkpoint activity that become arrested in metaphase for prolonged periods of time eventually progress into the G₁ phase of the cell cycle with tetraploid genomes (7, 8). The replication of DNA in these cells is blocked by a proposed p53- and pRb-dependent checkpoint, which has been called the postmitotic checkpoint (7). Abrogation of this postmitotic checkpoint will allow tetraploid cells into the cell cycle. In addition, polyploidy can be generated by other cellular events, including rereplication, a process in which cells replicate their DNA after completion of S phase without entering mitosis; abrogation of the spindle checkpoint followed by failure of cytokinesis; direct interference with cytokinesis; or cell fusion, which can occur during diseases involving viral infection (reviewed in ref. 9).

The postmitotic checkpoint shares some features with the G₁ and G₂ checkpoints: in all cases, cell cycle arrest coincides with high concentrations of p21 and hypophosphorylated pRb (3, 10). p53 seems to play a key role in mediating the postmitotic checkpoint (2, 7, 11), and p21 is responsible for at least part of this p53-mediated postmitotic arrest response (7, 12, 13). It seems that tetraploidy per se is not what induces cell cycle arrest at this checkpoint (14, 15). Activation of the postmitotic checkpoint can also eliminate polyploid cells through apoptosis. Tetraploid cells arising due to mitotic slippage were prone to undergo Bax-dependent mitochondrial membrane permeabilization and subsequent apoptosis that was partially dependent on p53 (16, 17). Interestingly, the mitotic checkpoint kinase BubR1 specifically induces apoptotic cell death of polyploid cells (18). Also, overexpression of the antiapoptotic protein Bcl-xL facilitated polyploidization (19).

The high-risk HPV types, of which type 16 (HPV-16) is the most prevalent, are commonly associated with cervical carcinoma, which is one of the leading causes of cancer death in women worldwide (for review, see ref. 20). The transforming properties of high-risk HPVs primarily reside in the E6 and E7 oncogenes (ref. 20 and references therein). E6 encodes a small protein with ∼150 amino acids and plays an essential role in the HPV life cycle (21). The ability of high-risk HPV E6 protein to promote the degradation of p53 has been suggested as a mechanism by which E6 induces cellular transformation (22). E6 also has functions independent of inactivating p53, including interactions with additional cellular proteins such as the α-helix motif-containing proteins and the PDZ domain–containing proteins (reviewed in ref. 23). However, although E6, along with E7, efficiently immortalizes primary human epithelial cells, it is not sufficient to induce transformation of human cells (23). Instead, it is believed that the genomic instability caused by E6 and E7 enable cells to accumulate additional genomic aberrations necessary to undergo malignant conversion.

It has previously been shown that E6 from high-risk HPV types alleviated cell cycle arrest induced by microtubule disruption and caused polyploidy in primary human foreskin keratinocytes (24–26). E6 in combination with E7 or in the context of the entire HPV genome induces polyploidy more efficiently (25). However, the molecular basis underlying this process is still uncertain. It was thought but not directly shown that E6 causes abrogation of the spindle checkpoint that leads to polyploidy in keratinocytes (24–26), whereas other studies have shown that E6 had no effect on the spindle checkpoint in human fibroblasts or cervical cancer cells (12, 27). To clarify the mechanism by which E6 induces polyploidy in primary human foreskin keratinocytes, we have examined the effect of E6 expression on the spindle checkpoint. We found that E6 expression does not affect the spindle checkpoint but instead abrogates the subsequent G₁ checkpoint after adaptation of the mitotic stress. In addition, previous studies with several E6 mutants showed a correlation between p53 degradation and polyploidy induction by E6 (24, 26). We have tested the ability of several p53 degradation defective E6 mutants to induce polyploidy and showed a p53-independent activity of E6 that can abrogate the postmitotic checkpoint.

Cell culture. The primary human keratinocytes (PHK) were prepared from neonatal foreskins obtained from the University of Massachusetts Memorial Hospital as described (28). NIKS is a spontaneously immortalized human keratinocyte cell line (29) available from American Type Culture Collection (Manassas, VA). Both PHKs and NIKS cells were maintained on mitomycin C–treated J23T3 feeder cells in F medium composed of three parts Ham's F12 medium and one part DMEM with all supplements as previously described (30). For NIKS cells, cholera enterotoxin was omitted from the F medium. The human telomerase reverse transcriptase–expressing human retinal pigment epithelium cells (RPE1; ref. 14) were maintained in 1:1 DMEM and Ham's F12 medium plus 10% fetal bovine serum. The primary human mammary epithelial cells (HMEC) were maintained in DFCI-1 medium as described (31).

Retroviral infections. PHKs expressing vectors, E6, E7, and both E6 and E7 (E6/E7) were established by retrovirus-mediated successive infections. E6 was in a Babe-puro–based retroviral construct whereas E7 was in an LXSN-based vector. PHKs and NIKS cells expressing E6, the E6 mutant, and the vector control, were established by retroviral-mediated infection using Babe-puro–based vector. After G418 and/or puromycin selections, populations of infected cells were pooled. Experiments were done using cells within eight passages after drug selection. E6 and E7 expression was confirmed by reverse transcription-PCR (RT-PCR) using the following oligos:

• HPV-16 E6, sense, 5′-CTGCAATGTTTCAGGACCCA-3′.

• HPV-16 E6, antisense, 5′-CCTAATTAACAAATC-3′.

• HPV-16 E7, sense, 5′-TCATGCATGGAGATACACCTACATTGCAT-3′.

• HPV-16 E7, antisense, 5′-GTTTCTGAGAACAGATGGGGCACAC-3′.

• Actin, sense, 5′-TGGCATTGCCGACAGGATGCAGAA-3′.

• Actin, antisense, 5′-CTCGTCATACTCCTGCTTGCTGAT-3′.

Flow cytometry. For ploidy analysis, asynchronous cultures of PHKs and NIKS cells expressing HPV oncogenes or vector were treated with DMSO, nocodazole (Sigma, St. Louis, MO; 50 ng/mL), Taxol (ICN Biomedicals, Costa Mesa, CA; 300 nmol/L), or bleomycin (Sigma; 10 μg/mL). Forty-eight hours later or unless otherwise specified, after the feeder cells were removed, keratinocytes were collected, fixed in 70% ethanol overnight, resuspended in PBS, stained in propidium iodide (50 μg/mL; Sigma) staining solution supplemented with 70 μg/mL RNase A (Sigma), and analyzed by flow cytometry. Cell cycle analysis was done using FlowJo software (Tree Star, Inc., Ashland, OR).

Mitotic index. PHKs and NIKS cells expressing vector or E6 were treated with 50 ng/mL nocodazole. Cells under treatment were harvested at various time points, fixed, stained with rat anti–phospho-histone H3 IgG2a (Sigma) and FITC-conjugated anti-Rat IgG2a (BD Biosciences, San Jose, CA), counterstained with propidium iodide, and analyzed by flow cytometry.

Immunoblotting. Protein extraction was done in lysis buffer [10 mmol/L Tris (pH 7.4), 1% SDS, 1.0 mmol/L sodium orthovanadate]. Equal amounts of protein from each cell lysate were separated in a SDS polyacrylamide gel and transferred onto a polyvinylidene difluoride membrane. Filters were blotted with antibodies against cyclin D1 (Santa Cruz Biotechnology, Santa Cruz, CA), cyclin E (BD Biosciences), p53 (Pab 1801, BD Biosciences; DO-1, Santa Cruz Biotechnology), p21 (BD Biosciences), pRb (BD Biosciences), and tubulin (Sigma).

Bromodeoxyuridine labeling experiment. Asynchronous cultures of NIKS cells expressing HPV oncogenes or vector were treated with DMSO or nocodazole. Bromodeoxyuridine (BrdUrd; final 10 μmol/L) was added to the medium at 34 h after treatment. After an additional 6 h, cells were collected, fixed in 70% ethanol, treated with 2 N HCl-Triton X-100, labeled with monoclonal anti-BrdUrd antibody (BD Biosciences), tagged with FITC-conjugated anti-mouse IgG (Sigma), and counterstained with propidium iodide (50 μg/mL) before flow cytometry.

Statistical analysis. Data were expressed as mean ± SD. Differences between means were assessed by Student's t test. P ≤ 0.05 was considered significant.

HPV oncogenes induce polyploidy in human keratinocytes. PHKs expressing vectors, E6, E7, or E6/E7 were established by retrovirus-mediated successive infections. HPV oncogene expression was confirmed by RT-PCR (Fig. 1A). Western blot analysis shows that p53 and pRb are reduced in PHK-E6 and PHK-E7, respectively (Fig. 1B), indicating that E6 and E7 are functionally active. Notably, the steady-state levels of p53 in PHK-E7 and pRb in PHK-E6 are increased, which is consistent with previous observations (32, 33).

In vitro, spontaneous polyploidization in HPV oncogene–expressing cells is not very efficient until late passages (34), although the HPV genome–containing W12 cervical cells become tetraploid before passage 14 (35). Nonetheless, under our culture conditions, the percentage of the PHKs expressing HPV E6 that are polyploid is consistently higher than that of the vector control cells (P = 0.01; Fig. 2A; Supplementary Fig. S1). To facilitate the polyploidization process, we treated cells with the microtubule poison nocodazole, which disrupts microtubule polymerization. As shown in Fig. 2A, a 5- to 6-fold increase in polyploidy was found in E6- or E7-expressing cells after nocodazole treatment. In the presence of both E6 and E7, the polyploidy generated was close to the sum of both oncogenes alone. These results show that HPV E6 and/or E7 induce polyploidy in PHKs, a result consistent with previous observations (24–26). DNA damage is also known to trigger polyploidization (36) and is probably a more physiologically relevant cellular damage than microtubule disruption. Therefore, we treated PHKs expressing E6 and/or E7 with bleomycin, a radiomimetic agent capable of inducing DNA double-stranded breaks. Significantly, both E6- and E7-induced polyploidy occurred in response to DNA damage (Fig. 2A). E7 expression was more efficient in inducing polyploidy after DNA damage than E6 expression.

We predict that induction of polyploidy is an important mechanism by which E6-expressing cells become aneuploid. We therefore focused our initial efforts to explore the pathway and mechanism of E6-induced polyploidy. Notably, the majority of vector control cells remained in a G₁ stage with 2C DNA content and did not progress into G₂-M phase after nocodazole or bleomycin treatment (comparing the ratio of cells with 4C and 2C DNA content in vector and E6 cells; Fig. 2A), probably a result of cellular senescence. This may result in less polyploid cell formation in the control cells. To alleviate this concern, we expressed HPV-16 E6 in the spontaneously immortalized human keratinocytes (NIKS cells). NIKS cells contain a wild-type (WT) p53 sequence and exhibit many characteristics of early-passage keratinocytes, including the ability to stratify, differentiate, and sustain the HPV life cycle (29, 37). Accordingly, NIKS cells in F medium with J23T3 fibroblast feeders were infected with retroviruses containing vector and HPV-16 E6. Expression of E6 was confirmed by RT-PCR (Supplementary Fig. S2A). Western blot analysis shows that p53 and p21 are reduced in NIKS-E6 cells (Supplementary Fig. S2B), indicating that E6 is functionally active. As shown in Fig. 2B, in response to nocodazole treatment, octoploid (8N) cells increased to 27% in NIKS cell expressing E6 (NIKS-E6 cells) but only to 5% in the vector control cells, although the majority of the control cells have progressed into the G₂-M phase. E6 also induced polyploidy in NIKS cells treated with Taxol, which stabilizes microtubules (Fig. 2B). These studies show that efficient induction of polyploidy by E6 in PHKs is not simply a result of overcoming replicative senescence.

E6 does not abrogate the spindle checkpoint in human keratinocytes. It was thought that HPV E6 is able to abrogate the mitotic spindle checkpoint and through this mechanism cause polyploidy in PHKs (24–26). To determine to what extent E6 abrogates the spindle checkpoint, we compared the mitotic indices of E6 and vector control PHKs. The mitotic index was assessed by measuring phospho–histone H3, a specific marker of mitosis (ref. 38 and Supplementary Fig. S3A), at several time points after nocodazole application. As shown in Fig. 3, the E6 cells entered and exited mitosis with kinetics similar to that of vector control cells. Both cell types arrested at mitosis in response to nocodazole. They reached a maximal mitotic index at 16 h. By 48 h, few cells remained in mitosis. Similar results were obtained in NIKS cells expressing E6 (Supplementary Fig. S3B). Our data show that E6 does not induce polyploidy by abrogating the spindle checkpoint in response to microtubule disruption.

E6 is capable of inducing polyploidy by abrogating the postmitotic checkpoint. We also considered the likelihood that polyploidization is a manifestation of rereplication in E6-expressing cells in which cellular DNA undergo another round of replication without entering mitosis. However, the kinetics of polyploid cell formation in E6-expressing NIKS cells suggest that polyploidization does not occur until 30 h after nocodazole treatment, a point later than E6-expressing cells exit from mitosis (peaked at 16 h after treatment; Supplementary Figs. S3B and S4). Nevertheless, because not all E6-expressing cells entered mitosis, we could not completely rule out the possibility that some polyploid cells are formed before cells enter mitosis (i.e., by rereplication).

There is also a possibility that E6 promotes cell fusion to induce polyploidy. However, we consider this is unlikely because cells with 8C DNA content were generated with intermediate DNA content of >4C <8C, especially at earlier time points such as 30 h (Supplementary Fig. S4).

To test the possibility that polyploidy formation in E6-expressing cells is through abrogation of the postmitotic checkpoint and not rereplication, we attempted to separate interphase cells from mitotic cells and examine the two cell populations for polyploidization independently. Because neither NIKS cells nor PHKs efficiently detach from the culture dish during mitosis, we used RPE1 cells, which can be readily synchronized and shaken-off at mitosis. Accordingly, RPE1 cells were infected with amphotropic retroviruses containing HPV-16 E6 or vector control. Expression of E6 was confirmed by RT-PCR (Supplementary Fig. S5A). Western blot analysis showed that the steady-state levels of p53 and p21 were lower in RPE1 cells expressing E6 (RPE1-E6 cells) than in RPE1-vector cells (Supplementary Fig. S5B). Upon nocodazole treatment, the p53 and p21 levels increased significantly in RPE1-vector but not in RPE1-E6 cells (Supplementary Fig. S5B).

Subsequently, RPE1-E6 cells were treated with nocodazole, and polyploidy was assessed by flow cytometric analysis of the DNA content. Similar to what was observed in keratinocytes, significantly more RPE1-E6 cells become polyploid after nocodazole treatment (Fig. 4C), and they also become polyploid after treatment with monastrol, a mitotic kinesin Eg5 inhibitor (not shown). RPE1-E6 and RPE1-vector cells exhibited almost identical mitotic indices (Supplementary Fig. S5C), indicating that similar to what was observed in the keratinocytes, E6 does not have a major effect on the spindle checkpoint in RPE1 cells. Upon nocodazole treatment, at least 80% of RPE1-E6 cells entered mitosis (Fig. 4B) and after 72 h, ∼35% of the E6 cells became polyploid (Fig. 4C), whereas <4% polyploid cells were formed from the vector control cells (Fig. 4C , inset). Furthermore, up to 87% of E6 cells can be shaken-off over a period of 10 h after nocodazole treatment (12–22 h), indicating that they have rounded up and entered mitosis. The shaken-off cells are believed to be at “points of no return” in mitosis (39). Because the number of cells that became polyploid was greater than those that did not progress into mitosis, these results indicate that at least a subset of polyploid cells was formed after cells have entered mitosis.

After being shaken-off, mitotic RPE1-E6 cells were obtained (Fig. 4D) and cultured in medium containing nocodazole. A significant percentage (∼25%) of RPE1-E6 mitotic cells became polyploid (Fig. 4E). The reason that mitotic shaken-off cells showed less polyploidy than those not subjected to shake-off (Fig. 4C) is likely due to mechanical damage to the cells during shaking. These results suggest that E6 abrogates the postmitotic checkpoint to induce polyploidy upon microtubule disruption, although it does not completely rule out the possibility that rereplication contributed to polyploidization to a lesser extent.

E6 is capable of inducing DNA replication after the postmitotic checkpoint arrest. After determining that neither rereplication nor abrogation of the spindle checkpoint is a major mechanism of E6-induced polyploidy after spindle disruption, we wanted to directly show that E6 could abrogate the postmitotic checkpoint. To confirm at the molecular level that NIKS cells with 4C DNA content treated with nocodazole are in a G₁ stage instead of other cell cycle stages, we measured the steady-state levels of G₁ cyclins E and D1, and pRb, which is hypophosphorylated when cells are under G₁ arrest. Our data showed that cyclins D and E were increased (2- and 1.4-fold, respectively) and pRb became hypophosphorylated after nocodazole treatment in NIKS-vector cells (Fig. 5A). A comparison of increased levels of cyclins D and E along with hypophosphorylated pRb in nocodazole-treated cells with proliferating cells indicate a G₁ cell cycle stage. In contrast, the steady-state level of cyclin E remains unchanged and pRb remains phosphorylated in E6-NIKS cells after nocodazole treatment (Fig. 5A). These results are consistent with the ability of E6 to abrogate the postmitotic G₁ checkpoint and drive cells into S phase. Therefore, our data indicate that NIKS-vector cells have adapted to the spindle checkpoint after prolonged microtubule disruption and have arrested in G₁, whereas NIKS-E6 cells progressed to S-phase.

Next, we wanted to test if NIKS-E6 cells are capable of replicating DNA when similarly treated. We incubated nocodazole-treated NIKS-vector and NIKS-E6 cells with BrdUrd and examined the ability of cells to incorporate BrdUrd after they have adapted to the spindle checkpoint and exited mitosis. Up to 18% of NIKS-E6 cells with DNA content >4C <8C incorporated BrdUrd (Fig. 5B), whereas few vector control cells with >4C DNA content incorporated BrdUrd (4%). These results suggest a direct effect of E6 on the postmitotic checkpoint.

E6-induced polyploidy can occur in the absence of p53 degradation. We next started to examine the molecular basis of E6-induced polyploidy. Because E6 degrades p53 and the latter seems to play an important role in the postmitotic checkpoint (2, 7, 11), it is reasonable to speculate that E6 induces polyploidy through inactivation of p53. In addition, previous studies with several E6 mutants showed a correlation between p53 degradation and polyploidy induction by E6 (24, 26). However, most of the previously used E6 mutants that are defective for p53 degradation are also defective in other biological activities such as immortalization. Our large-scale mutational analysis of E6 identified an E6 mutant (F2V, Phe-2 to Val mutation) that is defective for p53 degradation but competent for immortalization of HMECs (40). We therefore introduced this E6 mutant into PHKs and NIKS cells by retroviral-mediated infection. The expression of F2V has been confirmed by RT-PCR (Supplementary Figs. S2A and S6A) and the mutation has been verified by sequencing of the PCR product in both cell types. The steady-state protein level of F2V in PHKs remains to be examined, but it was similar to WT E6 expression in HMECs (40). The keratinocytes expressing F2V were treated with nocodazole and examined for DNA content. Interestingly, after a 48-h nocodazole treatment, F2V-expressing PHKs (Fig. 6A) or NIKS cells (Supplementary Fig. S7) exhibited ∼20% polyploidy, which is significantly higher than that found in the vector control cells. Like the WT E6, F2V-expressing NIKS cells also showed increased polyploidy upon Taxol treatment (not shown). Nevertheless, F2V-induced polyploidy was consistently lower than WT E6 (P = 0.005), suggesting that p53 degradation by E6 plays a role but is not absolutely required for polyploidization. Increased polyploidy was also observed in primary HMECs expressing F2V upon nocodazole treatment (not shown). Notably, after nocodazole treatment, PHKs expressing F2V showed a significant difference in apoptosis compared with vector or WT E6 cells. Specifically, the extent of apoptosis in PHK-F2V (20%) after nocodazole treatment was significantly more than that of PHK-WT E6 (9%; P = 0.03) but less than PHK vector (24%; P = 0.017; Fig. 6A). The percentage of cells undergoing apoptosis is inversely correlated with polyploidy, suggesting that polyploid cells were subjected to apoptotic elimination and E6 can inhibit polyploidy-associated apoptosis. As the only known functional difference between F2V and WT E6 is the ability to target p53 for degradation, decreased apoptosis in PHK-WT E6 compared with that in PHK-F2V is likely a result of p53 degradation. However, the results regarding antiapoptotic effect of E6 in polyploid cells should be interpreted cautiously, as such observations were not made in NIKS cells. Specifically, more apoptosis were found in NIKS-E6 cells than NIKS-F2V and NIKS-vector cells after nocodazole treatment, although these numbers are not statistically significant (Supplementary Fig. S7).

The results from studies with the mutant F2V suggest that E6 possesses p53-independent activities to alleviate cell cycle arrest induced by microtubule disruption. However, although the E6 mutant F2V is defective for promoting p53 degradation and failed to abrogate p53-mediated G₁ arrest upon actinomycin D–induced DNA damage (40), one study suggested that it may inactivate p53 through association with ADA3 (41). Another study showed that an E6 mutant unable to target p53 for degradation overcame the G₁ arrest by down-regulation of p21 in primary human fibroblasts (42). A more recent study reported that E6 repressed p53-dependent gene activation via inhibition of protein acetylation independently of its degradation (43). To measure the extent to which F2V may inactivate p53, we examined the levels of p53 and its targets p21 in PHKs and NIKS cells. p53 and p21 levels in vector control cells were similar to those in F2V-expressing PHKs (Fig. 6B and Supplementary S6B) and NIKS cells (Supplementary Fig. S2B), suggesting that p53 is functionally active in F2V-expressing keratinocytes.

To strengthen the notion that E6 possesses p53-independent function in inducing polyploidy, we have examined two additional E6 mutants (L37S and W132R) that are defective in degrading p53 at early passages in HMECs (40). Our studies confirmed that these E6 mutants are indeed defective for inactivating p53 (Supplementary Fig. S8A). What is more important, both mutants induced polyploidy (Supplementary Fig. S8B). These studies provide further support that HPV E6 is able to induce polyploidy via p53-independent mechanism.

Our present studies have shown a novel mechanism of E6-induced polyploidy in human keratinocytes. Our data indicate that E6 does not affect the spindle checkpoint but instead abrogates the postmitotic checkpoint. Our results also indicate that rereplication contributes little to polyploid formation in response to microtubule disruption. It was observed that when subjected to stresses, including exposure to microtubule poisons, some cells in prophase may decondense their chromosomes and return to interphase (39), where they could undergo rereplication. Although this does not happen efficiently and is an unlikely mechanism for E6 to induce polyploidy in response to spindle disruption, our results do not completely rule out this possibility.

p53 seems to play a key role in mediating the postmitotic checkpoint, and because E6 targets p53 for degradation, it is reasonable to propose that E6 induces polyploidy through inactivation of p53 (24, 26). However, we have identified an E6 mutant that is capable of inducing polyploidy in the absence of p53 degradation or inactivation. The demonstration of a p53-independent function of E6 in inducing polyploidy is important, as E6-mediated p53 degradation is not complete (Figs. 1B and 6B) and p53 may still be functioning in cervical cancer cells (44). The mechanism underlying E6-induced polyploidy in the absence of p53 degradation remains to be determined.

In this study, we have focused our effort on exploring the mechanisms of E6-induced polyploidy upon microtubule disruption. We also showed that HPV E6 induces polyploidy in response to DNA damage. A recent study by Incassati et al. (45) observed increased polyploidy in HPV E6/E7 PHKs in response to Adriamycin treatment. However, Adriamycin not only induces DNA damage but also inhibits DNA topoisomerase, and it was not known whether this activity can be attributed to E6 or E7. Polyploidy formed in response to DNA damage may be more relevant because under physiologic conditions, cells may experience far more DNA damage than microtubule disruption. In fact, expression of both E6 and E7 in PHKs are associated with DNA damage (46, 47). Future studies should explore the mechanism by which HPV oncogenes induce polyploidy upon DNA damage.

Grant support: NIH grant R21 CA108437 (J.J. Chen).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Drs. Elliot Androphy, Tim Kowalik, Joanna Parish, Nick Rhind, Alonzo Ross, and Wei Jiang for advice; Dr. Zhi-Guo Liu for help with establishing and characterizing the initial NIKS cells; and Ning Zhao for a pRb Western blot.