The growth inhibitory effect of p21 adenovirus on androgen-dependent and -independent human prostate cancer cells




To assess the potential of p21 as a gene therapy treatment for prostate cancer, by introducing p21 into both androgen-dependent (AD) and -independent (AI) human prostate cancer cell lines via a recombinant adenoviral vector, Ad5CMV-p21, carrying human p21 cDNA.


The LNCaP, DU145 and PC-3 human prostate cancer cell lines were cultured and infected with Ad5CMV-p21. Cell growth, cell-cycle progression and tumorigenicity were then assessed by thymidine incorporation into cellular DNA, and cell number, flow cytometry, and tumour growth after inoculating the cells into nude mice.


Growth was inhibited in Ad5CMV-p21 viral-infected AD and AI prostate cancer cells. The effects were dose-dependent, regardless of the androgen status of the cell lines. Flow cytometric analysis showed that Ad5CMV-p21 arrested cell-cycle progression at G1/S with no appreciable effect on the levels of apoptotic cells. The tumorigenicity of cancer cells infected with Ad5CMV-p21 was greatly reduced in athymic mice.


These results suggest that Ad5CMV-p21 may be a new therapeutic agent for human prostate cancer gene therapy.


Prostate cancer is the most frequently diagnosed cancer in men and the second leading cause of male cancer death in the USA [1]. Huggins and Hodges [2] first recognized the dependency of prostate cancer growth and progression on testicular androgens, and hormonal ablation, either surgically or chemically, has been the main treatment for advanced prostate carcinoma. However, hormonal therapy only temporarily relieves the clinical symptoms, and no effective therapy is available for treating prostate cancer once the tumour becomes hormone-independent.

The development of novel and effective non-surgical interventions for prostate cancer would represent a significant advantage in managing patients with prostate cancer. As a result of the greater understanding of the molecular and genetic mechanisms involved in the pathogenesis of various malignancies, and the availability of replication-defective viral expression vectors, gene-replacement strategies are being designed and implemented for treating specific malignancies [3]. The feasibility of this approach was shown both in vitro and in vivo[4,5], and in the treatment of patients with lung and head-and-neck cancer, using replication-defective adenoviral vectors expressing wild-type p53 protein [6,7].

One of the downstream effector molecules of p53 suppression is p21, a mammalian cyclin-dependent kinase (CDK) inhibitor [8]. The p21 protein is encoded by a gene termed WAF1 [9], CIP1 [10], SDI1 [11], or PIC1 [12]. Before the identification of p21WAF1/CIP1 and its relationship to p53, there was no clear connection between cell cycle control and genes transcriptionally regulated by p53 [12]. The loss of wild-type p53 is thought to be a key step in cancer progression. p21 protein expression is induced directly by p53 [13]; it inhibits the complexes of cyclins and CDKs and causes cell-cycle arrest [14]. Increased expression of p21 correlated with the induced terminal differentiation of tumour cells, and overexpression of exogenous p21 caused G1/S arrest [15]. As p53 mutations occur frequently in a variety of human tumours, including those arising in the prostate, several laboratories have attempted to suppress tumour growth by delivering p53 via an adenoviral vector to human prostate cancer cells and/or tumours, and by delivering both p53 and p21 to a murine prostate tumour model [4,16].

To expand these findings, we tested the biological activity of p21 as a potential downstream mediator of p53-dependent tumour suppression, and examined the possible tumour suppressive activity of p21 in human prostate cancer cell lines that harbour either p53 mutations or deletion. In the present study, p21 cDNA was cloned into the expression cassette in an adenoviral vector to generate recombinant viruses. This construct was evaluated for its efficiency of infection and expression in human prostate cancer cell lines, assessing growth inhibition in both androgen-dependent (AD) and independent (AI) prostate cancer cells in vitro, and the tumorigenicity of AI prostate cancer cells in nude mice.


LNCaP, DU145 and PC-3 human prostate cancer cell lines were obtained from the American Type Culture Collection (Rockville, MD). All cell lines were maintained in 10% iron-supplemented calf serum in DMEM/F12 (Life Techonologies, Inc., Gaithersburg, MD), supplemented with penicillin (100 units/mL), streptomycin (100 µg/mL) and 5% fetal bovine serum (Sigma Chemical Co., St. Louis, MO). All cells were free of Mycoplasma.

The recombinant Ad5CMV-p21, containing the cytomegalovirus (CMV) promoter, p21 cDNA and SV40 polyadenylation signal, was constructed by inserting the expression cassette into the E1-deleted region of modified Ad5. The recombinant Ad5CMV-β-gal and control adenovirus (Ad5CMV-PA) were constructed in a similar manner.

For the [3H]-thymidine incorporation assay, both AD and AI human prostate cancer cell lines were seeded at a cell density of 4 × 103 cell/well in 96-well tissue culture plate 24 h before viral infection. On the next day, cells were infected by a wide range of viral titres, from 1 to 256 plaque-forming units (PFU), by directly adding the viral stock solutions prepared in medium to each cell well containing 100 µL of the medium. Cells were exposed to fresh medium containing 5% fetal bovine serum after overnight exposure to adenoviral infection. The [3H]-thymidine incorporation assay was described previously [4].

For the in vitro cell growth assay, cells were plated at a density of 1 × 104 cells/mL in six-well plates 24 h before viral infection. The cells were infected with either Ad5CMV-p21 or Ad5CMV-PA (as a control) at 20 PFU/cell. Cells were harvested every 2 days and counted; their viability was determined by trypan-blue exclusion.

For flow cytometry analysis, cells were inoculated at a density of 5 × 10 cells per 60-mm culture dish 24 h before infection with Ad5CMV-p21 or Ad5CMV-PA at 20 PFU/cell. At 24 h after infection, cells were treated with 1% Tween-20 and stained with propidium iodide, and analysed by flow cytometry to determine the status of cell-cycle progression, using a FACScan (Becton Dickinson, San Jose, CA) equipped with an air-cooled 15-mW, 488 nm argon laser. Culture medium alone was used as mock infection control.

For the tumorigenicity assays, the AI cell lines DU145 and PC-3 were infected with Ad5CMV-p21 or Ad5CMV-PA at a dose of 20 PFU/cell. An equal number of cells was treated with medium and served as the mock infection. At 24 h after infection the treated cells were harvested and rinsed twice with PBS. For each treatment, 1 × 106 cells in 0.1 mL were injected subcutaneously into Balb/c nu/nu athymic male nude mice (Harlan Co., Indianapolis, IN) and tumour formation evaluated after 5 weeks. A pair of slide callipers was used to assess tumour size, measuring the largest and smallest diameters of the tumour every week, to give the estimated weight as (largest × smallest2) × 0.5 [17].


The overexpression of wild-type p21 protein markedly inhibited the rate of [3H]-thymidine incorporation into cellular DNA of LNCaP (Fig. 1a), DU145 (Fig. 1b) and PC-3 (Fig. 1c) cells in vitro. There was dose-dependent growth inhibition in all cell lines tested and the level of inhibition correlated with the efficiency of viral infectivity, i.e. LNCaP > DU145 > PC-3 (data not shown). LNCaP appeared to be more sensitive than DU145 and PC-3 cells to virus-induced growth inhibition, even by the control Ad5CMV-PA virus. The differential sensitivity of various human cell lines to virus-induced therapeutic response or toxicity was reported previously [4].

Figure 1.

Inhibition of [3H]-thymidine incorporation into human prostate cancer cellular DNA. LNCaP (a), DU145 (b) and PC-3 (c) cells were infected with recombinant Ad5CMV-p21 (red square) or Ad5CMV-PA (green circle) at doses of 1–256 PFU/cell.

Cell numbers were also determined after adenoviral infection; because control Ad5CMV-PA virus inhibited growth at higher doses, we examined the effect of 20 PFU Ad5CMV-p21 on the numbers of AD and AI cells in vitro. The growth of LNCaP (Fig. 2a), DU145 (Fig. 2b) and PC-3 cells (Fig. 2c) was inhibited by Ad5CMV-p21 but not by the control Ad5CMV-PA infection. In uninfected or control virus-infected LNCaP cells, growth peaked at 3 days possibly because of the extreme sensitivity of LNCaP cells to the metabolites accumulated in the medium. However, the growth of DU145 and PC-3 cells, similarly infected, continued even 5 days after infection, supporting the notion that LNCaP cells are more sensitive to adenoviral infection than DU145 and PC-3 cells. A single dose of Ad5CMV-p21 infection appeared to consistently impair the growth of all prostate cancer cell lines at 3–5 days after infection.

Figure 2.

Inhibition of prostate cancer cell growth in vitro by infecting cells with Ad5CMV-p21. Growth curves of Ad5CMV-p21-infected cells (red squares), Ad5CMV-PA-infected cell (green open circles), and mock-infected cells (light green closed circles); Cells were exposed at 0 days to 20 PFU of either Ad5CMV-p21 or Ad5CMV-PA virus. (a) LNCaP, (b) DU145, (c) PC-3. Data are the mean (sem) of three replicates.

To examine the mechanism by which Ad5CMV-p21 may retard prostate cancer growth in vitro, the cell cycle was analysed using flow cytometry to assess cell-cycle kinetics in the three cell lines 24 h after infection. As shown in Table 1, overexpression of p21 protein after Ad5CMV-p21 infection significantly increased the number of cells in G1 and decreased the number in the S and G2+M phases in each of the cell lines tested. These results are consistent with the interpretation that the overexpression of p21 protein in prostate cancer cells may induce cell-cycle arrest at the G1/S boundary.

Table 1. Flow cytometry analysis of AD and AI human prostatecancer cell lines exposed to either Ad5CMV-p21 or Ad5CMV-PA ( as control) in vitro
Cell type and viruses% of cells
+ mock2872
+ AdCMV-PA3070
+ AdCMV-p214456
+ mock3070
+ AdCMV-PA3367
+ AdCMV-p215050
+ mock2179
+ AdCMV-PA2575
+ AdCMV-p213664

In the assessment of prostate tumour growth, the tumours, first palpable at 14 days after injection, were formed only from the mock- or control virus-infected cells; mice that received Ad5CMV-p21-treated cells did not develop tumours during a 5-week observation period (Table 2). The mean tumour volumes for PC-3 and DU145 at the end of the 5-week period were 256–1350 mm3, respectively.

Table 2. Tumour growth and progression in AI human prostate cancer cell lines overexpressing p21
No. of mice developing tumours/total inoculated
Ad5CMV-PA  8/124/4
Ad5CMV-p21  0/120/4


We assessed the use of p21WAF1/CIP1, an inhibitor of cell-cycle regulators, as a repressor for the growth of both AD and AI prostate cancer cells in vitro and tumour growth in vivo. The loss of p53 tumour suppressor function, either through mutation or deletion, occurs frequently in human tumours, including primary and metastatic prostate cancer [18]. By delivering p53 tumour suppressor to cells we and others have reported growth suppression and a reduction of the metastatic potential of prostate tumour cells overexpressing p53 protein [4,19]. As mutation and deletion of p53 occur in prostate tumours, the present study was designed to test whether other inhibitors of cell-cycle regulators downstream from p53 might also block the growth of both AD and AI prostate cancer cells in vitro and in vivo.

The WAF1/CIP1 gene encodes a 21 kDa protein (p21WAF1/CIP1) which is capable of blocking the activity of cyclin-CDK complexes and inhibiting cell-cycle progression [9,10]. Recent studies indicate that p21 protein is also capable of inhibiting DNA replication in vitro in what appears to be a cyclin- CDK-independent pathway [20]. p21 is transcriptionally activated by wild-type, but not mutated, p53 protein and induces cell-cycle arrest, although p21 induction can be stimulated independently of p53 [21]. It is possible that p21 overexpression in prostate cancer cells is compensating for other defective pathways of the cell cycle. Alternatively, p21 over-expression might be caused by its binding to another molecule that sequesters it and keeps it in an inactive form.

The identification of p53-dependent and -independent pathways of WAF1/CIP1 induction has several important implications: (i) WAF1/CIP1 inducibility by DNA damaging agents that rely on p53-dependent pathways may provide a molecular indicator to assess the functional link between p53 tumour suppressor and cell cycle arrest or apoptosis induced by these agents; (ii) alternative strategies, in addition to adenoviral-mediated p21 gene transfer, might be developed to induce WAF1/CIP1 gene expression in tumours, particularly in p53-defective tumour cells by pharmacological agents which could enhance tumour cell death [22]. Such strategies are likely to improve the clinical utility of tumour suppressors as inhibitors of tumour growth and progression.

In the present study we showed that Ad5CMV-p21 mediated the growth inhibition of both AD and AI cells in vitro, and the growth of AI tumours in vivo. On the basis of β-galactosidase reporter expression in Ad5CMV-β-gal transfected cells, the adenoviruses are very infectious, reaching 95% and 65% in LNCaP and PC-3 cells. Growth inhibition induced by adenoviral infection appears to be determined by both the dose and efficacy of adenoviral infection. Because the control viruses, Ad5CMV-PA and Ad5CMV-β-gal, caused some growth inhibition when applied at higher concentrations (> 100 PFU/cell), prostate cancer cell lines were infected with Ad5CMV-p21 at a dose of < 20 PFU/cell in the hope of expressing p21 protein in target cells without inducing nonspecific toxicity. The high level of p21 expression in prostate cancer cells as a result of adenoviral infection is presumably augmented by the strong CMV-promoter which functions under the conditions of the removal of E1 enhancer/silencer residues within the viral construct [23]. The primary mechanism mediating p21 inhibition of the growth of AD and AI cancer cells is the ability of p21 to inhibit CDK2 activity, thereby blocking the proliferating cells from entering from G1 to S phase. This suggestion is supported by the data showing that there was an accumulation of prostate cancer cells at the G1/S interphase after p21 adenoviral infection. These results agree with the hypothesis that p21 can block cell-cycle progression in both p53 wild-type and -mutated or deletion cells. In the present study, despite the differences in p53 background in the various prostate cancer cell lines (LNCaP cells contains a silent mutation p53, DU145 cells contain two p53 mutations whereas PC-3 cells have p53 deletion), p21 blocked cell-cycle progression at the G1/S phase in all of the cell lines tested. Ad5CMV-p21 (20 PFU/cell), when introduced ex vivo, completely blocked prostate tumour growth in vivo. The tumoricidal effect was probably increased by the direct activity of p21 on the proliferation of both AD and AI prostate cancer cells, as adenoviral infection caused a transient overexpression of p21 and the transduced viral genome existed as an episome that was never integrated into the host genome [16]. The high efficiency of p21 in blocking tumour progression after in vivo infection poses the interesting possibility that p21 may be used as a powerful tumour suppressor gene for treating both AD and AI human prostate cancer.


cyclin-dependent kinase




plaque-forming units.