In Vitro Growth Suppression by Adenoviral Transduction of p21 and p16 in Squamous Cell Carcinoma of the Head and Neck: A Research Model for Combination Gene Therapy

Objectives  To determine the duration of expression of the cell cycle regulators p21 and p16 and the effect of these 2 genes both alone and in combination on the growth of squamous cell carcinoma of the head and neck cell lines in vitro.

Experimental Methods and Design  Cells were transduced via an adenoviral vector with p21 (Ad5CMV-p21), p16 (Ad5CMV-p16), or both. Western blotting was performed to determine duration of expression of the protein products of the transduced p21 and p16 genes. In vitro growth assays and cell cycle analyses were performed on transduced cells.

Results  Transduced gene products were detected up to day 12 after infection. Western blotting showed high levels of p21 and p16 in transduced cells. Growth suppression was observed in squamous cell carcinoma of the head and neck cell lines transduced with Ad5CMV-p21, Ad5CMV-p16, or both, but the combination of p21 and p16 did not achieve significantly greater growth suppression than that seen in cells transduced with Ad5CMV-p16 alone. Cell cycle analysis demonstrated that the percentage of cells arrested at G₁ stage in the cells transduced with Ad5CMV-p16 was similar to that in the cells transduced with both Ad5CMV-p21 and Ad5CMV-p16. No significant in vivo growth suppression was observed in tumor nodules treated with Ad5CMV-p16.

Conclusions  Although p21 and p16 are believed to function as cell cycle regulators of cyclin-dependent kinases, we observed no additive or synergistic effect when using them in combination. The expression of transduced p21 and p16 gene products up to days 9 and 12, respectively, was consistent with the growth suppression and cell cycle arrest observed. This work provides information on the previously uncharacterized duration of p21 and p16 transgene product expression and also lends insight into the interaction of these 2 cell cycle regulators in squamous cell carcinoma of the head and neck.

TECHNOLOGY in the field of molecular biology is developing at a rapid pace, providing investigators with increasing opportunities to attenuate the growth of cancer cells at the molecular level. As of June 1996, there were approximately 232 approved gene therapy clinical trials worldwide, with 175 having been initiated.¹ These trials are evaluating the utility of gene therapy in treating a variety of cancers, including brain, breast, lung, and squamous cell carcinoma of the head and neck (SCCHN). Much interest has focused on the clinical potential of this new modality to treat cancer, especially on the use of tumor suppressor genes. Previous studies of tumor suppressor gene therapy for SCCHN focused largely on the use of p53 .^2-4 The action of p53, as now understood, is to induce cell death (apoptosis) and cause cell cycle arrest at the G₁ checkpoint.⁵

Another potential tumor suppressor gene, p16, has recently been cloned and localized on chromosome 9p21.⁶ This gene has demonstrated many characteristics of a tumor suppressor gene. It is a known inhibitor of cyclin-dependent kinase 4, whose activity is critical to regulation of normal cell cycle progression.⁶ Deletions of p16 have been shown to occur at a high frequency in tumors and have been detected in 51% of squamous cell carcinomas of the esophagus.^7,8

The actions of another tumor suppressor gene, p21, have been demonstrated to be transcriptionally regulated downstream by p53.^9,10 Like p16, p21 has the potential to affect the cell cycle by inhibiting a cyclin-dependent kinase (cyclin E).¹¹ Several features suggest that p21 too is a tumor suppressor gene. Cells overexpressing p21 accumulate in G₁, and p21 has been shown to participate in G₁ checkpoint regulation downstream from p53.^11-13 Others have shown that p21 has the ability to function as a tumor suppressor in both colon and prostate cancer cell lines.^9,14

Since p21 and p16 both regulate the cell cycle as inhibitors of different cyclin-dependent kinases, we sought to determine whether combination gene therapy using both genes would additively or synergistically inhibit the growth of SCCHN cells.

Tumor cells TU-138 and TU-167 were established from human primary SCCHN and have been described elsewhere.¹⁴ TU-138 has a known single-base-pair mutation of p53. TU-167 has a splicing mutation at exon 5 of p53.

The recombinant adenovirus Ad5CMV-p21 contains the cytomegalovirus promoter, wild-type p21 complementary DNA, and SV40 polyadenylation signal in a minigene cassette inserted into the E1-deleted region of modified adenovirus type 5 (Ad5). The Ad5CMV-p16 adenoviral vector is similar in construction. Our methods for recombinant viral vector construction have been described elsewhere.²

The SCCHN cell line TU-138 was plated and infected at 30 multiplicity of infection (MOI) of Ad5CMV-p21 or Ad5CMV-p16. Total cell protein lysates were obtained on days 1, 2, 3, 6, 8, 9, 12, and 14 and assayed by Western blot analysis.

Cells infected with the following schema were used for the in vitro growth assays, cell cycle analyses, and corresponding Western blots. Three adenoviruses were used: Ad5CMV-p21, Ad5CMV-p16, and Ad5CMVPA. Ad5CMVPA is an adenovirus that does not carry a therapeutic gene and serves as a vector-only control. We used 4 treatment groups and 1 mock treatment group. All treatment groups received 2 treatments with an adenovirus vector, 48 hours apart, at 50 MOI. In this schema, after 48 hours, all treatment groups had been exposed to two 50-MOI viral infections. The treatment groups were as follows: group 1 received Ad5CMVPA/Ad5CMVPA; group 2 received Ad5CMV-p21/Ad5CMVPA; group 3 received Ad5CMV-p16/Ad5CMVPA; and group 4 received Ad5CMV-p21/Ad5CMV-p16. The mock treatment group received no viral vector infection.

Tumor cells (TU-138 or TU-167) were seeded into 6-well plates at a density of 2.0×10⁴ cells/well. At specified times after adenoviral infection, medium was removed from the wells containing the cells to be counted, and cells were rinsed with phosphate-buffered saline (PBS) and trypsinized (0.25%) with scraping to recover all cells. Cells were immediately added to 5 mL of growth medium containing 10% serum, pelleted by centrifugation, then resuspended in 1 mL of PBS. Growth measurements for a specific day were obtained by counting triplicate wells of cells on an automated cell counter (Coulter model Zf, Coulter Corp, Miami, Fla).

The effect of viral infection on cell growth was analyzed by means of a 1-way analysis of variance test. For the days on which the groups exhibited a significant difference, the individual means were compared with each other also by a 1-way analysis of variance. All statistical analyses were performed by means of SigmaStat software (Jandel Corp, San Rafael, Calif).

Total cell lysates in protein lysis buffer (150-mmol/L sodium chloride, 1.0% NP-40, 0.5% deoxycholate, 0.1% sodium dodecyl sulfate, 50-mmol/L Tris [pH, 8.0], phenylmethylsulfonylfluoride, and aprotinin) were obtained. Protein concentration, transfer, and blocking were performed as described elsewhere.² The membrane was probed with primary antibodies to p16 (rabbit anti–human polyclonal antibody [Santa Cruz Biotechnologies, Santa Cruz, Calif]), p21 (mouse anti–human monoclonal antibody [Oncogene Science Inc, Union Dale, NY]) and β-actin (mouse anti–human monoclonal antibody as a loading control [Amersham Inc, Arlington Heights, Ill]). Secondary antibodies were horseradish peroxidase–conjugated goat antimouse and antirabbit (Amersham Inc). The membranes were processed and developed according to the manufacturer's suggestions.

Cell monolayers were trypsinized, washed with cold PBS, and fixed by storing in 70% ethanol at −20°C. These fixed cells were washed twice with PBS before cell cycle analysis. Approximately 1.0×10⁶ cells were resuspended in 70-µg/mL RNase A and 5-µg/mL propidium iodide in a final volume of 900 µL of PBS. After gentle mixing, cells were incubated in the dark for 30 minutes at room temperature. Samples were analyzed by means of a flow cytometer (EPICS Profile II, Coulter Corp, Miami, Fla). The proliferative fraction was determined as the percentage of cells in S-phase according to the method of Baisch et al.¹⁵

The effect of Ad5CMV-p16 on a microscopic residual disease model of SCCHN was determined in nude mice. Eight mice were used for both cell lines TU-138 and TU-167 and the experiment was performed twice. Experiments were reviewed and approved by institutional committees for both Animal Care and Utilization and the Biosafety Committee for Recombinant DNA Research.

Subcutaneous flaps were seeded with tumor cells in the dorsal flanks of the animals in a manner described elsewhere.⁴ Forty-eight hours after tumor cell delivery, the flaps were infected with Ad5CMV-p16, Ad5CMVPA, or PBS. The mice were observed daily for tumor development and were killed immediately in cases of excessive tumor burden or after 12 weeks of observation.

To investigate the duration of expression of p21 and p16 transgenes in TU-138 and TU-167, after viral infection of each of these cell lines, cells were harvested and protein extracts prepared at different times after infection. Western blot analyses were performed to detect protein levels at different times. Western blotting showed high levels of transduced gene products up to day 9 to 12 after infection. No significant expression was seen by day 14 (Figure 1).

Forty-eight hours after the first infection, protein extracts were obtained for all 5 groups represented in the in vitro growth assays. Western blotting demonstrated high levels of protein expression of p16 and p21. We did not detect any significant basal levels of either p16 or p21 in mock-infected cells (Figure 2). In addition, sequencing of p16 in our laboratory from both TU-138 and TU-167 disclosed no mutations (data not shown). Therefore, the cell cycle arrest observations in these experiments can be attributed to exogenously supplied p16 and p21, which are not in competition with abnormal levels of either wild-type or mutant p16. Expression of p21 or p16 was similar whether the genes were delivered alone with control vector or in combination. Actin controls confirmed equal protein loading (Figure 2).

The effects of transduced p21, p16, and both p21 and p16 on the growth rates of SCCHN were assessed. The vector Ad5CMVPA served as a control for growth inhibition observed in cells treated with viral vector alone. With the use of the infection schema described above, cell counts were obtained for 1 week after the first infection. Marked growth suppression was seen in cells transduced with Ad5CMV-p21/Ad5CMV-p16. However, this effect was not significantly different from that of the group Ad5CMVPA/Ad5CMV-p16. On days 5 to 7 (Figure 3), the mock-infected cells (no viral infection) or the cells infected with vector only exhibited statistically significant differences (P<.001) from those cells infected with either p16 alone or the combination of p16 and p21. No significant growth suppression was seen in the Ad5CMV-p21/Ad5CMVPA group or the 2 control groups, mock or vector-only control (Ad5CMVPA/Ad5CMVPA), for TU-167 (Figure 3, right). Significant growth suppression (P<.001) was observed for TU-138 (Figure 3, left) in the Ad5CMV-p21/Ad5CMVPA group compared with mock-infected or vector-only infected groups.

To measure the effect of the p16 and p21 transgenes on cell cycle progression in both TU-138 and TU-167, we performed flow cytometric analyses of infected cells. Flow cytometry demonstrated that cells transduced with either p21, p16, or both became arrested at the G₁ stage of the cell cycle up to 96 hours after infection (Table 1). At 96 hours, TU-138 cells transduced with p21 alone were 88% in G₁ and cells transduced with p16 alone were 70% in G₁, while those cells transduced with both p16 and p21 were 90% in G₁. By comparison, at 96 hours, TU-167 cells transduced with p21 alone were 86% in G₁, cells transduced with p16 alone were 87% in G₁, and cells transduced with both p16 and p21 were 92% in G₁.

On the basis of the significant in vitro growth suppression attributable to p16 gene alone, we used a nude mouse model for microscopic residual disease to determine the biological effect of p16 in vivo. No significant growth suppression was observed in these studies (data not shown).

Combination gene therapy, which may selectively regulate tumor cell growth, induce apoptosis, or regulate sensitivity to radiation therapy or chemotherapy, may have promise as our understanding of molecular mechanisms regulating malignant neoplasms progresses. Clinically, one could hypothesize the theoretical advantage of treating tumors with 2 genes that in combination have a greater tumor-suppressive potential than either gene alone. While other authors have examined the effects of either p21 or p16 in vitro, no one has looked at the potential of these 2 genes to act synergistically.^9,14,16 The combination of p16 and p21 was chosen because the products of both of these genes are known to function through similar mechanisms of cell cycle regulation by inhibition of cyclin-dependent kinases.

We have described elsewhere that the adenoviral vector is an effective vehicle for the delivery of exogenous genes.² To better characterize the efficacy of the p21 and p16 genes in suppressing tumor growth, Western blotting was performed to confirm the duration of overexpression of these transgene products. To our knowledge, this had not been previously described and was critical to our understanding of the duration of action of these cell cycle regulators. Infections were performed at only 30 MOI because further increases in MOI resulted in such marked growth suppression that extracts could not be obtained for Ad5CMV-p16 beyond day 6 (unpublished data). Western blotting of protein extracts from SCCHN cell lines up to 14 days after transduction with either p21 or p16 confirmed that high levels of exogenous protein were detectable up to 12 days after infection.

Western blotting of the protein extracts corresponding to the 5 experimental treatment groups used in the in vitro growth assay confirmed also that high levels of p21 and p16 could be detected in cells transduced with either p21 or p16 adenoviruses, or a combination of both. However, no expression of these gene products was detected in the mock infection or vector-only control groups.

Moderate growth suppression was seen in cells transduced by Ad5CMV-p21. However, more marked growth suppression was observed in cell lines transduced with either Ad5CMV-p16 or the combination Ad5CMV-p21/Ad5CMV-p16. The growth suppression resulting from infection with Ad5CMV-p21/Ad5CMV-p16 was similar to that seen with the group Ad5CMVPA/Ad5CMV-p16. These observations suggest that the growth suppression seen with combination p21/p16 was not the result of an additive or synergistic interaction of these 2 genes, but rather was attributable principally to the effects of the p16 gene alone.

No significant in vivo growth inhibition by Ad5CMV-p16 was observed in xenograft nude mouse animal experiments. This is understandable, since the growth-suppressive transgene product persisted for only about 12 days, which would have allowed at most only a slight delay in tumor formation. In contrast, our previous studies have shown that transient expression of p53 inhibits tumor formation by induction of apoptosis, not by cell cycle arrest.³ This suggests that cell cycle regulators may have little utility as a gene intervention strategy alone. However, cell cycle regulators may sensitize tumors to other interventions, such as chemotherapy or radiation therapy. Further basic investigations of gene therapy strategies, alone or in combination, that may selectively destroy malignant neoplasms but spare normal cells, are needed.

Accepted for publication July 24, 1997.

This study was supported in part by a Career Development Award (grant 93-9) from the American Cancer Society, Atlanta, Ga; grant 1-P50-DE11906 from the National Institute of Dental Research, Bethesda, Md; First Investigator Award R29 DE11689-01A1 from the National Institutes of Health, Bethesda; Training of the Academic Head and Neck Surgical Oncologist grant T32 CA60374-03 (GLC) and cancer center core grant NIH-NCI-CA-6672 from the National Cancer Institute, National Institutes of Health, Bethesda; and a gift to the Division of Surgery and Anesthesia from Tenneco Inc, Greenwich, Conn, for its core laboratory facility.

Presented at the Fourth International Conference on Head and Neck Cancer, Toronto, Ontario, July 28, 1996.

We thank Mary Wang, MS, for technical support, Jose C. Juarez for his assistance in computer graphics, Dianna B. Roberts, PhD, for her assistance with statistical analyses, and Peggy Tinkey, DVM, for veterinary assistance.

Reprints: Gary L. Clayman, DDS, MD, The University of Texas M. D. Anderson Cancer Center, Department of Head and Neck Surgery, 1515 Holcombe Blvd, Box 69, Houston, TX 77030 (e-mail: gclayman@notes.mdacc.tmc.edu).