p16/pRb Pathway Alterations Are Required for Bypassing Senescence in Human Prostate Epithelial Cells¹ Free

Normal human cells have a finite proliferative potential (usually 50–100 doublings) in vitro, after which cell cycle arrest occurs (1). This process is termed replicative senescence, and in contrast to apoptotic or programmed cell death, it represents a stable, protracted state in which cells remain viable and metabolically active. The process of immortalization requires overcoming or escaping replicative cellular senescence. Bypassing senescence is thought to be important in the development of human cancer for several reasons (2, 3): (a) immortalization increases cell susceptibility to malignant progression by permitting the extensive cell divisions that are necessary for cells to acquire multiple genetic alterations; (b) tumors, both naturally occurring and experimentally induced, often contain cells that are immortal or have an extended replicative life span (4, 5); and (c) genetic regions associated with the immortalization of cell types in vitro are found to be altered in clinical cancers of the same cell type (6). Therefore, an understanding of the genes controlling normal prostate epithelial cell mortality may provide relevant information on the mechanisms underlying abnormal cell growth and tumorigenesis in prostate epithelial cells.

A series of classical observations have shown that replicative life span can be extended by the expression of DNA tumor virus oncoproteins, including SV40 and HPV16³ E6 and E7 (3, 7). These tumor viruses are known to inactivate p53 and pRb (8), thus implicating these antiproliferative proteins in the process of overcoming senescence. This initial period of extended life span requires overcoming the M1 (mortality) block. However, overcoming M2, which results in an infinite life span, clearly requires additional genetic events (9, 10). Immortalization occurs infrequently, which indicates that the inactivation of several pathways is required. Alterations in other cell cycle regulatory genes, including p21^WAF1 and p16/CDKN2, have been implicated in mediating senescence (11, 12, 13). Notably, all of these genes encode putative tumor suppressor proteins, the functions of which are lost in many human cancers, including prostate cancers (14, 15, 16, 17, 18, 19).

Increases in wild-type p53 induce antiproliferation signals, including the downstream cyclin-dependent kinase p21, that are distinct from apoptosis (20). Levels of p53 have been found to increase at late passage in human fibroblasts, putatively contributing to senescence growth arrest (11, 21). Inhibiting the expression of p53, either by antisense RNA methods (22) or by use of p53 transdominant mutants (23), leads to a delay in senescence. Elevation of p21 levels has also been found in aging fibroblasts (13). However, because levels decrease at senescence, the significance of the role of p21 in senescence growth arrest is unclear. Its role in tumorigenesis is also uncertain: a recent study showed that mice with a homozygous deletion of p21/WAF1 failed to form tumors (24).

pRb and p16 function in a common pathway that appears to be important in the growth arrest of cells after a finite number of population doublings. Phosphorylation regulates pRb function by inducing E2F release and the subsequent expression of E2F-dependent proteins, such as cdc2 and cyclin A. These genes are not expressed in senescent fibroblasts consistent with a block in pRb phosphorylation at senescence (25). Furthermore, pRb is found in a hypophosphorylated state in senescent fibroblasts (26). Overcoming this senescence block is possible by fusion with cells containing the viral oncogene E7, which binds unphosphorylated pRb (26). p16 blocks pRb phosphorylation by binding cdk4 and cdk6 and inhibiting their association with cyclin D (27). This results in a failure of pRb phosphorylation and E2F release and culminates in G₁ cell cycle arrest. Several studies have shown that p16 levels increase in human and rodent fibroblasts as cells are passed to terminal senescence (11, 28). However, only one study to date has demonstrated increased p16 at senescence in a human epithelial cell type, namely urothelial cells (12).

p16 and pRb inactivation are generally not found in the same cancer, consistent with the observation that they function within the same pathway (29, 30). In support of this model, the cell cycle arrest induced by the introduction of p16 occurs only in cells that retain functional pRb (31). The importance of p16 as a tumor suppressor is emphasized by its frequent inactivation in different types of human malignancies (32). We recently reported that p16/CDKN2, located at 9p21, was deleted or hypermethylated in >50% of advanced prostate cancers and that this finding was associated with transcriptional inactivation (19). RB loss of heterozygosity is also frequently found in prostate cancers (16, 17), although whether homozygous alterations occur is not known. The loss of pRb expression does correlate with a poor clinical prognosis (33).

In summary, an alteration in the p16/pRb pathway is commonly seen in prostate cancers. Several lines of evidence discussed above suggest that these alterations may play a role in bypassing the tumor suppressor checkpoint of cellular senescence. Here, we test this hypothesis using an HPEC in vitro transformation system that uses HPV16 E6 and E7 to inactivate p53 and pRb, respectively. We report here for the first time with HPECs that p16 levels are reproducibly elevated at senescence and that bypassing senescence in HPV16 E6- or E7- transformed cells is always accompanied by an alteration in the p16/pRb pathway. Furthermore, we identify additional genetic alterations, including +5q, +8q, and +20, that are nonrandomly associated with HPEC immortalization. We conclude that alteration of the p16/pRb pathway is necessary but not sufficient for overcoming senescence in HPECs.

Biopsies of histologically confirmed normal prostate tissue from men (ages 47–65 years) without diagnosed prostate cancer were obtained from cystoprostatectomy specimens or from cadavers. The culture system used is based on the work of Peehl et al. (34, 35) and Reznikoff et al. (36). To establish HPECs, we minced fresh prostate tissues and plated them on 100-mm collagen dishes in F-12+ medium with penicillin/streptomycin (9). F-12+ is a supplemented Ham’s medium (Life Technologies, Inc., Gaithersburg, MD; intermediate Ca²⁺, 0.3 mm) that also contains 1% FBS. Bovine pituitary extract (25 μg/ml; Life Technologies, Inc.) and cholera toxin (100 ng/ml; Life Technologies, Inc.) were added to this medium. HPECs were passaged routinely when confluent after dispersion with trypsin-EDTA (Life Technologies, Inc.). Prostate cancer cell lines were grown in RPMI with 10% FBS.

Immunohistochemistry was performed on early-passage HPECs and HPEC E6- and E7- immortalized cell lines after harvesting and 10% formalin fixation (37). Primary antibodies and dilutions included cytokeratins 5 (1:500; Enzo) and 18 (1:100; Sigma Chemical Co., St. Louis, MO) and prostate-specific antigen (1: 100; Sigma). Biotinylated antimouse or rabbit IgG was used as a secondary antibody (Sigma). Staining was accomplished with the ABC reagent (Vector Laboratories, Burlingame, CA) and the substrate diaminobenzidine. Immunohistochemistry was duplicated on each of three independent HPEC cultures. Controls included tissues with no primary antibody and uncultured normal prostate tissues.

HPECs were transferred to chambered slides (Nunc, Naperville, IL) and grown for 2 days. Cells were then rinsed and fixed in a β-galactosidase stain solution as described (38). Positive β-galactosidase staining, as defined by blue cytoplasmic and nuclear staining, was a marker for senescence and was not present in differentiated or apoptotic cells (38).

Transformation of HPECs with HPV16 E6 and/or E7 and selection for immortal clones was performed as described for human urothelial cells (9). Briefly, retroviruses carrying either the HPV16 E6 and/or E7 gene(s) (received from Dr. D. Galloway, Seattle, WA) were prepared (39). Subconfluent proliferating HPECs were infected with 10³–10⁵ infectious viral units at early-passage (∼5 × 10⁵ cells per 100-mm dish) in 3 ml of 1% FBS-F-12+ containing 4 μg/ml polybrene (Sigma). The virus was then removed after 6 h, and selection with 50 μg/ml G418 (Life Technologies, Inc.) was performed for a minimum of 7 days. Two uninfected control dishes were monitored for senescence. Clones infected with E6, E7, or E6/E7 were checked for expression of E6 and E7 transcripts using RT-PCR as described previously (9). Southern blot analysis was also performed as described previously (9) using a 0.8-kb BamHI-HindIII fragment from p1321 that contains the HPV16 E6/E7 genes. Fifteen μg of DNA were loaded in each lane.

Western blots were performed as described previously (12) Briefly, cells were lysed in buffer containing protease inhibitors and resolved (50 μg/lane) on 12.5% SDS-polyacrylamide gels. After transfer to a nylon blot (Immobilon P; Millipore, Bedford, MA), primary antibodies were applied. These included specific antibodies to p53 (AB2; Oncogene Science), phosphorylated and unphosphorylated forms of pRb (14001A; PharMingen), p16 (C-20, Santa Cruz Biotechnology, Santa Cruz, CA), and p21(Oncogene Science). Immunoreactive proteins were visualized using enhanced chemiluminescence. Urothelial cells immortalized with HPV16 E7 (12) were used as positive controls for p16, p53, and p21. Immunoblotting was performed in duplicate.

Sample DNA was amplified by PCR using primers spanning exons 1 and 2 of the p16/CDKN2 gene (40, 41). The amplified products were cloned into pCR 2.1-TOPO using the TOPO TA cloning kit from Invitrogen according to manufacturer’s instructions. The cloned fragments were sequenced using an automated DNA sequencer. Methylation analysis was performed as described previously using a 340-bp p16/CDKN2 exon 1 probe (19). Serial concentrations of HPEC E6-15 DNA was used in Southern blotting to confirm deletion of p16/CDKN2.

Hybridization of differentially labeled immortalized cell lines to metaphase chromosomes from normal peripheral blood was performed exactly as described previously (42). We have described how we define changes in the relative copy number of DNA sequences (i.e., gains and losses) in an earlier manuscript (43). CGH loss and gain data were analyzed using a simple statistical model, as described previously (44). Briefly, the null hypothesis asserts that changes are sporadic and thus present randomly according to a background rate that is constant among chromosome arms. The alternative hypothesis allows that some arms exhibit an elevated rate of change. The model also allows the possibility that changes on the p and q arms are linked. Gain data exhibit a pattern significantly different from one would expect under the null hypothesis, with P = 0.001, whereas loss data are consistent with sporadic change (P = 0.31).

The growth of HPEC explants on collagen-coated dishes in F-12+ medium results in cultures of prostate epithelial cells that are epithelial in morphology (i.e., tightly adherent and polygonal) and apparently uncontaminated with other cell types (Fig. 1,A). After four to five passages, HPEC proliferation slows, and cells enter a nonproliferative state. These HPECs adopt a morphological phenotype characteristic of senescence (Fig. 1,B) and acquire SA β-galactosidase activity (Ref. 38; Fig. 1 C).

To assess the phenotype of cultured HPECs using our conditions, we performed immunohistochemical staining for cytokeratin markers of luminal or basal prostate epithelial cells (45, 46) in both uncultured tissue sections and cultured HPECs. Cytokeratin 5, a basal cell marker, and cytokeratin 18, a luminal protein, are both expressed (data not shown). Another marker of differentiated luminal epithelial cells, prostate-specific antigen, is not found at detectable levels in HPECs using immunohistochemistry. It is, however, detectable using RT-PCR (data not shown). These results demonstrate that HPECs coexpress both basal and luminal keratins, thus representing the amplifying or stem cell population that is hypothesized to be the cellular precursor to prostate cancer (46). Additionally, expression of these prostate epithelial markers essentially excludes contamination by cells of stromal origin.

To determine whether alterations of pRb, p53, p21, and p16 might play a role in mediating prostate epithelial senescence, we compared protein levels in proliferating (presenescent) and senescent HPECs. In contrast to previous studies in fibroblasts (11), results showed no alterations in p53 or p21 levels at senescence in HPECs. However, HPECs at senescence reproducibly show at least a 10-fold elevation of p16 (Fig. 2), when compared to early-passage proliferating HPECs. We also note a consistently decreased level of phosphorylated pRb in senescent cells consistent with a G₁ cell cycle block. pRb is detectable on longer exposure. The antibody used for pRb detection recognizes primarily phosphorylated forms of pRb.

HPV16 E6 selectively abrogates p53, whereas HPV16 E7 binds and inactivates pRb. These viral genes were used as molecular tools to test, in HPECs, the functional significance of losing pRb and p53 in the process of overcoming senescence. Results showed that immortalization of human epithelial cell lines by HPV16 genes occurs in two stages (M1 and M2), as was initially described in mammary epithelial cells by Shay et al. (47). After infection of HPECs at passage 2 (P2) and selection with G418, all HPV16-transformed cultures (E6/E7, E6, and E7) showed an extended life span of two to three additional passages (eight versus five passages) before undergoing senescence. E6- and/or E7-transformed HPECs then entered an extended crisis period lasting 2–3 months in which no net gain in cell numbers was evident. After 2–3 months, small, single colonies of HPV16-transformed HPECs began to proliferate and could eventually be dispersed and passaged. These colonies represent the emergence of immortal clones from the so-called M2 stage, and they occurred at a low frequency of ∼1 × 10⁻⁵. E6/E7 immortal colonies typically emerge sooner than E6 or E7 colonies (∼6 weeks versus 12 weeks). Using this approach, we established three sets of HPV16 E6, E7, and E6/E7 HPEC immortal cell lines. Each set of cell lines was generated from a single initial prostate epithelial culture taken from one of three different individuals, thus the lines can be defined as isogeneic in origin.

Phenotypic differences are noted between the E6- and E7-transformed prostate epithelial cell lines. E6 HPEC and E6/E7 HPEC lines are characterized by a pleomorphic cell size, irregular shape, and loose adhesion (Fig. 3,A). In contrast, E7 HPEC cells have flat, tightly adherent cells more typical of normal epithelial morphology (Fig. 3 B). All HPV16-transformed HPEC lines are now at P40 or more (400 doublings), consistent with the immortal phenotype. Cytokeratin staining for 5 and 18 confirmed their epithelial origin.

Southern blot analysis for HPV16 insertion was performed on the eight HPV cell lines generated at an early-passage (P18→P24). Lines were not considered immortal until approximately P15. Results show a single integration site in 5 cell lines, as indicated by a single band on Southern blot (Fig. 4,A). Two bands are demonstrated in HPEC lines E6-9, E6/E7-9, and E7–14. Given their apparent clonal origin (see above) and equal band intensity at different passages, these two bands probably represent two integration sites in each of these cell lines. RT-PCR confirms the correct expression of E6 and/or E7 RNA within the E6, E7, or E6/E7 HPEC lines (Fig. 4 B).

The sufficiency of pRb or p53 inactivation in overcoming senescence during immortalization was evaluated in HPECs. Consistent with a functioning HPV16 E6, Western blot analysis of HPEC E6 and E6/E7 immortal lines (P20) shows that p53 levels are undetectable (Fig. 5). In contrast, HPEC E7 cell lines show slightly elevated p53 levels compared to normal HPECs (Fig. 2). p21 expression was positively regulated by p53, and consistent with this, p21 was strongly expressed in HPEC E7. In contrast, even with prolonged Western exposure times, p21 protein was not detectable in p53 negative HPEC E6 lines. Therefore, p21 levels correlate with p53 expression and p53 function in E7 cell lines was confirmed.

pRb is nondetectable in the E7 cell lines and present at low levels in E6/E7 HPEC when compared to proliferating normal HPECs. The low level of pRb detected in HPEC E6/E7 lines may represent incomplete ubiquitin-induced degradation of pRb (48). pRb was strongly detected in HPEC E6 lines, as expected. These levels of pRb protein were comparable to those seen in normal prostate epithelium. However, in all three HPEC E6 lines, immortalization is accompanied by p16 loss. In contrast, p16 is abundantly expressed in the HPEC E7 and E6/E7 immortal lines (that have lost pRb function). These elevated levels of p16 are comparable to the levels seen in HPECs at senescence (Fig. 2). High levels of p16 do not inhibit proliferation in E7-transformed HPECs because of the inactivation of downstream regulatory protein pRb. Therefore, all HPV16 E6- and/or E7-transformed HPECs showed either p16 or pRb inactivation.

To further test the association between inactivation of pRb expression or p16 in bypassing senescence in HPECs, we evaluated a series of immortal prostate cancer cell lines (Fig. 5). These lines derive from independent biopsies of metastatic prostate cancer. Two cell lines, Du145 and LnCaP, showed no pRb expression. The Du145 prostate cancer cell line contains a known RB mutation (49). In both of these lines, p16 was present and detectable by Western analysis. All other metastatic immortal prostate cancer lines expressed apparently wild-type pRb, but no p16 protein was detectable. These results demonstrate a loss of p16 or pRb in all immortalized prostate cancer cells.

Abnormal p53 expression was detected in the majority of metastatic cell lines, including PC3, TSU-PR1, DuPro, and PPC-1. Inactivating mutations within the coding exons (exons 5–8) have been described previously in PC3 and TSU-PR1 (50). Du145 contains a stabilized mutant p53. The mutant p53 was undetectable in PC3 and TSU-PR1. In LnCaP, p53 is wild type, and low levels were detectable only by immunoprecipitation (50). Therefore, p53 inactivation was frequent but not necessary for overcoming senescence and acquisition of the metastatic phenotype in prostate cancer.

To identify mechanisms underlying the loss of p16 expression in immortal HPEC E6 lines and in representative metastatic prostate cancer lines described above, we performed methylation analysis using Southern blot (Fig. 6). Loss of p16 expression in two of three HPEC E6 lines (E6-9 and E6-14) was due to biallelic DNA hypermethylation, as evidenced by the presence of a 6-kb fragment after restriction with the methylation-sensitive enzyme SmaI. Methylation of this region has been previously correlated with loss of transcription (40). The radiolabeled probe used for Southern analysis demonstrates a reproducibly decreased signal (>50%) in cell line HPEC E6-15, indicating a heterozygous deletion of p16/CDKN2. This is consistent with CGH results (Table 1). Sequencing of exons 1 and 2 of p16/CDKN2 was performed in all HPEC E6 lines. E6-15 contains a mutation in exon 1 in codon 33 (GAG→TAG) that generates a stop codon and has been described previously (51). A second mutation was detected downstream in codon 34 (GCG→ACG; Glu→Thr). A complete loss of p16 expression from E6-15 resulted from a deletion of the other allele of CDKN2/p16 (Fig. 6).

An analysis of the p16/CDKN2 locus in prostate cancer cell lines by Southern demonstrated no methylation in LnCaP and Du145 (19), both of which express p16 by Western blot. However, a missense mutation in exon 2 on codon 84 (GAC→TAC; Arg→Leu) is present in Du145. This was identified in all sequencing reactions performed and confirmed an earlier report of mutation in this line (19). It did not alter expression of the protein. Hypermethylation was detected in PC3, TSU-PR1 (data not shown), DuPro, and PPC-1. Therefore, the loss of p16 expression in HPEC E6 and metastatic prostate cancer lines can, in all cases, be explained by hypermethylation, mutation, or deletion.

CGH on early-passage HPV16 immortalized HPECs was used to identify genetic regions that were gained or lost in association with immortalization (Table 1). A gain of chromosome 20 was present in seven of eight HPEC-immortalized lines. 8q was gained in four cell lines. Gain of 5q was seen in all HPEC E7. Loss of 8p, a commonly deleted region (∼80%) in clinical prostate cancer (52), was present in one HPEC E6. Notably, CGH is relatively insensitive for the detection of genomic losses smaller than 10 Mb. Further mapping may detect smaller deletions. Nevertheless, these findings suggest that genetic alterations, including +20, +8q, and +5q may contribute to overcoming senescence.

To test this, we analyzed CGH loss and gain data using a simple statistical model as described earlier (5, 44). The null hypothesis asserts that changes are sporadic and thus present randomly according to a background rate that is constant among chromosome arms. The alternative hypothesis allows that some arms exhibit an elevated rate of change. The model allows the possibility that changes on the p and q arms are linked. Gain data exhibited a pattern that was significantly different from that which one would expect under the null hypothesis with P = 0.001, whereas loss data were consistent with sporadic change (P = 0.31). Among gains, 5q, 8q, and 20 were significant in the sense that the probability exceeded 0.95 (in each case) that gain of each arm exhibited an elevated rate. No significant losses were detected in our sample group.

Replicative senescence is a mechanism of tumor suppression and represents a safeguard against the development of neoplasia (2, 4). Therefore, the identification of genes that function in normal replicative senescence in prostate epithelial cells is important for our understanding of tumorigenesis in the prostate. In this study, we report for the first time that replicative senescence in normal prostate epithelial cell cultures is reproducibly associated with an elevation of p16. We also demonstrate for the first time using prostate epithelial cells that bypassing senescence requires an alteration in the p16/pRb pathway. However, this latter alteration is apparently insufficient for bypassing senescence, as nonrandom gains of 5q, 8q, and 20q also accompany immortalization.

Our data showing that p16 is elevated at senescence in prostate epithelial cells supports a model for replicative senescence that has been previously proposed for human fibroblast and urothelial cells (11, 12). p16 specifically binds to and inhibits CDK4 blocking progression of the cell cycle beyond G₁. Thus, in this model, elevated levels of p16 play a critical role in senescence G₁ growth arrest. The mechanisms underlying the dramatic increase in p16 levels at terminal senescence have not been defined but may result from an accumulation of p16, possibly due to an increased stabilization of p16 mRNA and/or a loss of p16 repression by decreased levels of pRb (53). We find a consistent down-regulation of phosphorylated forms of pRb in senescent human prostate epithelial cells, which may, in turn, contribute to a failure of cell cycle progression. Functional pRb is required for normal G₁ growth arrest, as studies using transfected and expressed p16 have shown (31, 54). Our findings, therefore, implicate p16 as important in the G₁ cell cycle arrest characteristic of senescence in normal prostate epithelial cells.

Notably, increases in expression of p21 and p53 do not occur during replicative cellular senescence in prostate epithelial cells. This result is similar to findings in cultured human urothelial cells (12) but differs from findings in human and rodent fibroblasts (11, 28). Some studies suggest that p21, which functions to arrest cells in cycle by binding to cyclins D, A, and E, is overexpressed in and is a mediator of replicative senescence in fibroblasts (13). However, other data question the finding that p21 is elevated at terminal senescence in fibroblasts (11). It is also uncertain whether p53 up-regulation is critical to replicative senescence. For example, p53-null fibroblasts from patients with Li-Fraumeni syndrome undergo normal senescence in the absence of expression changes in either p53 or p21 (55). Therefore, p21 and p53 elevation at senescence may represent a characteristic of selected fibroblast cell strains, but it does not appear to be important for senescence G₁ cell cycle arrest in human prostate (or urothelial) epithelial cells.

To test our hypothesis that p16 elevation plays a critical role in replicative senescence, we examined the status of p16 and pRb in three isogeneic sets of HPV16 E6 and/or E7 immortal prostate epithelial cells that bypassed senescence. Each cell set arises from an initial single epithelial culture generated from one of three independent normal prostate specimens. Our second important finding is that overcoming the cell cycle block associated with senescence requires an alteration in the p16/pRb pathway. In the cells lines containing E7, pRb function is lost by the binding of E7 oncoprotein to underphosphorylated pRb and by an enhancement of ubiquitin-induced degradation of pRb (48). HPV16 E7 immortal lines show elevated levels of p16, similar to those seen at senescence. This finding supports the observation that the cell cycle arrest imposed by p16 is only apparent in cells that retain functional pRb (31, 54). The high levels of p16 in E7 pRb deficient cells provide further evidence for a feedback regulatory loop involving pRb and p16. We also tested spontaneously immortalized cell lines derived from biopsies of metastatic prostate cancer and demonstrate that pRb expression is lost and p16 is elevated in two of six lines.

An alternate pathway for reentry into the cell cycle from senescence in the presence of wild-type pRb involves a loss of p16 expression. Because pRb is a downstream component of the p16 pathway, loss of p16 function would, in theory, function equivalently to pRb loss. This scenario is found in all HPEC E6-immortalized cell lines that have lost functional p53 and retained pRb. In the E6 cell lines tested, p16 loss is mediated most commonly by DNA hypermethylation. We also find that one line, E6-15, contains an allelic deletion and a mutation on the remaining allele. Therefore, several common mechanisms for inactivating p16 are demonstrated in our experimental model using normal prostate epithelial cells. The requirement for a p16/pRb pathway alteration is also met in the spontaneously immortalized metastatic prostate cell lines. Loss of p16 expression due to hypermethylation of p16/CDKN2, along with wild-type pRb, is found in four of six of these cell lines. Hypermethylation of p16/CDKN2 is a selective mechanism for inactivating p16 expression, and does not appear to alter the expression of the alternative reading frame splice variant p14/ARF in bladder cancer cell lines (56). The finding of methylation inactivation of p16/CDKN2 in prostate cancer and HPEC E6-immortalized lines (six of nine) emphasizes a unique epigenetic feature of the tumorigenic process in prostate cells. Mutations of p16/CDKN2 occur rarely in prostate cancer (19) in contrast to other tumors (57, 58). The Du145 cell line, which contains a pRb mutation, we have found on sequencing to contain a p16/CDKN2 missense mutation (19, 59). However, wild-type p16 is apparently encoded from the second allele in this pRb-negative line.

The results above document that an alteration in the p16/pRb pathway is critical for immortalization of HPEC. However, we have identified a number of additional genetic alterations that are nonrandomly associated with overcoming the senescence block in HPECs. Both HPEC E7- and E6-infected cells undergo a crisis period of low to undetectable proliferation (M2 block) for several months before proliferative clones that give rise to immortal lines are detected. The most significant genetic change, identified in seven of eight immortal lines, is a gain of chromosome 20. Gain of chromosome 20 has been seen in many human cancer types, including bladder, breast, ovarian, and prostate (9, 43, 60). By CGH, centromeric regions of 20 are amplified in almost half of in vivo prostate cancer metastases (15). This amplification is infrequently found in primary prostate tumors. Finally, gain of chromosome 20 has been identified in human urothelial cells transformed in vitro by HPV16 (9). These observations indicate the presence on chromosome 20 of one or several oncogenes. Several candidate genes have been identified, including ZNF217, NABC1, and CAS (61). Other regions of gain in HPEC E6 and E7 lines include 5q and 8q. 8q gains are noted infrequently in primary prostate cancers, but they occur commonly in metastatic and recurrent (80%) prostate cancers (15). c-Myc is located at 8q24, and Myc protein levels are increased in E6 cells by an undefined posttranscriptional mechanism. Recently, it was found that telomerase induction is required for E6 epithelial immortalization, in addition to alterations in the p16/pRb pathway (10). Notably, we have demonstrated telomerase activity in cultured normal prostate epithelium, as well as normal bladder, ureter, and mammary epithelium (62). It has also been noted that Myc protein increases telomerase mRNA through an undefined mechanism (63).

In the current model of prostate carcinogenesis, inactivation of the p16-pRb pathway appears to play an important role in overcoming the cell cycle block imposed by p16 at senescence. RB, on 13q14, is a region of intermediate frequency (∼30%) deletion in prostate cancer samples (16, 64) We have previously demonstrated that p16/CDKN2 alterations occur frequently (∼50%) in prostate cancer (19). Given the present data supporting the mutual exclusion of p16 and pRb alterations within the same cell line, inactivation of this pathway may represent an extremely common alteration in prostate cancer. Although this correlation has not been tested in prostate cancers in vivo, it has been demonstrated in several other tumors (29, 65). Inactivation of the p16/pRb pathway is necessary but not sufficient for immortalization of HPEC. Thus, our in vitro model using E6 and E7 contains genetic gains and losses also seen in in vivo tumors. Therefore, it may be useful in identifying genes altered in prostate cancer and defining pathways of cancer progression via different combinations of genetic and epigenetic alterations.

Our special thanks to Robert Huffman (Department of Surgery) and Dr. David Uehling (Division of Urology) of the University of Wisconsin for their support.