p16 Gene therapy: A potentially efficacious modality for nasopharyngeal carcinoma

Nasopharyngeal carcinoma (NPC) is a malignancy of the head and neck region. It is endemic in many geographical regions, including southern China and Southeast Asia (1), and affects a relatively young population (2). Potential etiological factors found to be associated with NPC include EBV infection, environmental factors, and genetic susceptibility (3). Despite the high rates of local tumor control achieved with complex radiotherapy (RT) techniques, NPC is still associated with a significant risk of distant metastases (4). The overall 5-year survival rate is approximately 65% (5), indicating a need for novel therapeutic strategies that may improve patient outcome.

Allelic loss on chromosome 9 has been frequently observed in NPC (61% of primary tumors) (6), particularly at 9p21–22, where p16 is located. p16 is a regulator of the G₁ phase of the cell cycle; progression through the cell cycle is dependent on the orderly activation of various complexes of cyclins and cyclin-dependent kinases (CDK). It is the CDK4- or CDK6-cyclin D complex that is particularly crucial for the early transition from G₁ to S phase. As a member of the INK4 family (Inhibitor of CDK4), p16 competes with cyclin D for CDK4 or 6 binding. This results in inhibition of CDK activity, leading to G₁ arrest (7). p16 inactivation is a common event in NPC (60–80% of primary tumors), and may thus play an important role in disease progression (6).

In NPC, p16 can be inactivated through multiple mechanisms, including homozygous deletion, promoter hypermethylation, and point mutation (6, 8–12). Therefore, selective replacement of p16 offers a potentially attractive strategy to correct the malignant phenotype. Adenoviral-mediated p16 gene transfer (adv.p16) has been evaluated in several other human cancer models, including head and neck (13), breast (14), lung (15), prostate (16), and pancreas (17). These studies consistently demonstrate that cancers null for p16 status were highly sensitive to p16 gene transfer (18–21), with evidence of growth inhibition, apoptosis, and tumor suppression (21). Together, these results indicate that p16 gene transfer, in combination with conventional therapy, may be particularly effective in treating NPC.

In this study, we examine the contribution of adenoviral-mediated p16 gene transfer to the therapeutic benefit of RT in two NPC cell lines that differ in their endogenous p16 status. We demonstrate that this therapeutic strategy is highly effective against the cell line that lack p16 function.

Two distinct NPC cell lines, CNE-1 (22) and CNE-2Z (23), were obtained from the Cancer Institute/Chinese Academy of Medical Sciences in China. Both cell lines were maintained in αMEM supplemented with 10% v/v fetal bovine serum (FBS). Both cell lines harbor a point mutation in the splice acceptor site of exon 2 of p16 in the second base (A to C) (12). However, our sequencing data suggested that CNE-2Z cells harbor a heterozygous mutation for the p16 gene (data not shown).

The human embryonic kidney (HEK) 293 and cervix HeLa cell lines (American Type Culture Collection) were also used in the study. Both cell lines were maintained in αMEM containing 10% FBS.

Ad5CMV-p16 (adv.p16) is a replication-deficient adenoviral vector containing a cytomegalovirus (CMV) promoter upstream of the human wild-type p16 gene (from Dr. T-J. Liu). A control adenoviral vector, Ad5CMV-β-gal (adv.β-gal) expressing the β-galactosidase reporter gene, was kindly provided by Dr. F. Graham of McMaster University (Hamilton, Canada).

Both adenoviral vectors were propagated in HEK 293 cells, purified by cesium chloride (CsCl) density gradients, and titered using plaque-forming assay (24, 25). Purified adenoviral vectors were also examined for revertant adenoviruses through consecutive rounds of infection using HeLa cells as previously described (25).

All in vitro infections were performed by adding an appropriate volume of the purified and titered adenovirus (adv.p16 or adv.β-gal) to cells, along with a small volume of αMEM containing 2% v/v heat-inactivated FBS. Infections were conducted for 1 h at 37% in 5% CO₂, and were completed by adding αMEM (10% FBS) directly to each flask. Iso-equivalent viral doses were used for CNE-1 and CNE-2Z infections by conducting β-galactosidase assays as previously described (25) to normalize transduction efficiency differences between the two cell lines. Hence, adenoviral doses of 5, 10, or 15 plaque-forming unit (pfu)/cell in CNE-1 cells were iso-equivalent to 17, 34, or 70 pfu/cell for CNE-2Z cells.

After adv.p16 infection (at 0, 5, 10, and 25 pfu/cell CNE-1 iso-equivalent doses) and/or 2 Gy of RT administered 24 h later, CNE-1 or CNE-2Z cells were trypsinized and counted. Subsequently, 10²–10⁵ cells were plated onto 100 mm² dishes (Nunc, Denmark), and grown in αMEM with 10% FBS at 37°C/5% CO₂. Colonies were subsequently fixed and stained with methylene blue (in 50% ethanol). Triplicate dishes were set up for each condition, and each experiment was repeated at least twice.

Approximately 1 × 10⁶ CNE-1 or CNE-2Z cells (infected or irradiated) in T25 flasks were rinsed with PBS and harvested in lysis buffer [0.1 m Tris-Cl (pH 8.0), 1% SDS, 10 mm EDTA, and 2 mm DTT]. Samples containing equal amounts of protein were run on SDS-PAGE gels [8% for retinoblastoma protein (pRb), 12% for p16 and β-actin] for 90–120 min at 110 V, and transferred onto nitrocellulose membranes (at 15 V for 30 min). Blots were then probed using the respective primary antibodies in 0.1% PBST containing 5% milk [1.5 μg/ml p16 monoclonal antibody (mAb) (NeoMarkers, Fremont, CA); 1 μg/ml pRb mAb (PharMingen, San Diego, CA); and 1:1000 working dilution of β-actin mAb (Sigma, St. Louis, MO)]. After several washings with 0.1% PBST, the blots were incubated with a horseradish peroxidase-conjugated secondary antibody (Amersham, Piscataway, NJ). Anti-mouse secondary Ab was used against both p16 and pRb mAb, whereas an anti-rabbit secondary Ab was used against the actin mAb. Proteins of interest were visualized using a chemiluminescence reagent (Santa Cruz Biotechnology, Santa Cruz, CA). Y79 is a retinoblastoma cell lysate used as a positive control for p16 protein and a negative control for pRb (kindly provided by Dr. B. Gallie, OCI). A549 is a non-small cell lung cancer cell line (kindly provided by Dr. M. Tsao at OCI) serving as a negative control for p16, and a positive control for pRb.

Twenty-four or 48 h after infection (2 pfu/cell of either adv.p16 or adv.β-gal), or irradiation (2 Gy), 1–3 × 10⁶ CNE-1 or CNE-2Z cells were pelleted and fixed with 4 ml of ice-cold 70% ethanol added dropwise. Ethanol was removed, washed with PBS twice, and then the cell pellets were resuspended in 500 μl of buffer (0.2% Triton X-100, 1 mm EDTA in PBS). The suspension was then treated with 50 μg/ml of RNaseA, and stained with 50 μg/ml of propidium iodide at room temperature for 60 min. Cell cycle analysis was performed on an EPICS analyzer using ModFit LT2.0 software (Beckman Coulter, Fullerton, CA). Each experiment was performed three times. The P-value was obtained using the Student t test by comparing treated samples (adv.β-gal, adv.p16, or irradiated) with control (uninfected) for each phase of the cell cycle.

CNE-1 cells (1.5 × 10⁵) were transferred onto six-well plates and grown for 2 days. Subsequently, they were infected with adv.p16 (5, 10, 15, or 25 pfu/cell) for 24, 32, or 48 h. After those set times, cells were washed twice with PBS and fixed with 2% formaldehyde/0.2% glutaraldehyde for 10 min at 4°C. Cells were then incubated with staining solution (consisting of X-gal solution buffered to pH 6.0 with citric acid/sodium hydrogen phosphate) for 24–48 h at 37°C. Senescent cells with detectable SAB will turn blue, and the number of such cells was counted under a microscope. Infection by adv.β-gal provided negative controls.

Cell death was evaluated morphologically using acridine orange-ethidium bromide (AO-EB) (Sigma) fluorescence staining. CNE-1 cells were assayed 2 days following infection with adv.β-gal (60 pfu/cell) or adv.p16 (60 pfu/cell), or 24 h after irradiation (2 Gy). They were washed with PBS, pelleted at 1300 rpm, resuspended in a small amount of PBS, and then mixed with 20 μl of AO-EB stock to a final concentration of 2.5 μm. A small volume of stained cells was placed onto glass slides and immediately visualized using fluorescence microscopy (Leica, Switzerland). The assay allows morphological inspection, as well as quantification of cells undergoing either apoptosis or necrosis. For each slide, nine fields of view containing at least 200 cells were counted. The level of apoptosis or necrosis for each treatment condition was determined from three independent experiments. By comparing infected or irradiated cells with untreated samples, P-values were determined using the Student t test.

CNE-1 cells were treated under various conditions: mock infection; adv.p16 (25 pfu/cell); adv.p53 (50 pfu/cell); or adv.β-gal (25 pfu/cell). Genomic DNA was extracted 2 days after infection using the Puregene DNA isolation kit (Gentra Systems, Minneapolis, MN) and then subjected to gel electrophoresis at 100 V in a 1.8% agarose gel. Cells treated with adv.p53 (50 pfu/cell) provided a positive control for apoptosis (25, 26). Cells subjected to four rounds of freezing (−70°C) and thawing (37°C) provided the positive control for necrosis (27).

CNE-1 or CNE-2Z cells were treated with adv.β-gal (25 pfu/cell) or adv.p16 (25 pfu/cell), and 24 h later, were trypsinized and injected into the left gastrocnemius muscle of severe-combined-immunodeficiency (SCID) mice. About 4 × 10⁶ cells resuspended in 100 μl of PBS were injected into each animal. Another set of experiments was conducted, whereby 75% of the injected CNE-1 cells were infected with adv.p16 (3 × 10⁶ infected cells were mixed with 1 × 10⁶ uninfected cells before each injection). Each experiment was repeated twice, and at least two animals were used for each infection condition. Animals were monitored twice a week for tumor formation by measuring the leg plus tumor diameter, which was later converted to tumor weight as previously described (28). Mean tumor weight ± SE was then calculated for each time point. The mice were sacrificed for humane reasons once the tumor burden was judged to be excessive. These experiments are conducted in accordance with guidelines set by the University Health Network Animal Ethics Board.

Before evaluating the therapeutic effect of adv.p16, basal expression levels of p16 and pRb in the two NPC cell lines were evaluated. In CNE-1 cells, p16 expression was virtually undetectable (Fig. 1A), which is in contrast to CNE-2Z cells, where high levels of endogenous p16 expression were observed (Fig. 1B). Both cell lines, however, expressed high levels of pRb.

Western blot analysis demonstrated that recombinant p16 could be detected in CNE-1 cells 4 h following adv.p16 infection (5 pfu/cell) (Fig. 1A). p16 expression continued to increase until 48–52 h postinfection. An inverse relationship was observed between p16 and pRb in that as p16 expression increased, pRb levels decreased correspondingly.

In contrast to CNE-1 cells, p16 kinetics appeared to be delayed in the CNE-2Z cells (Fig. 1B). No further increase in p16 expression was detected until 8–24 h after adv.p16 infection using iso-equivalent viral doses.

Our objective was to combine adv.p16 therapy with RT, hence it was important to examine the effect of RT on recombinant p16 expression. Independent of p16 gene transfer, RT (2 Gy) had no influence on endogenous or recombinant p16 expression in either cell line (Fig. 2). Interestingly, RT appeared to enhance the level of total pRb in CNE-1 cells under both uninfected and infected conditions (Fig. 2A). Similarly, RT alone appeared to increase the expression of total pRb in CNE-2Z cells (Fig. 2B). The increase in p16 expression following adv.p16 treatment might have counteracted the pRb up-regulation caused by RT in both cell lines (Fig. 2).

Clonogenic assays were performed to determine the sensitivity of CNE-1 and CNE-2Z cells to adv.p16 gene transfer, in combination with RT (2 Gy). CNE-1 cells were infected using a range of adv.p16 doses from 0 to 25 pfu/cell. As indicated in Fig. 3, p16 gene transfer alone caused significant cytotoxicity to CNE-1 cells, resulting in a surviving fraction of only 0.1% with adv.p16 at 25 pfu/cell. In contrast, CNE-2Z cells were resistant at iso-equivalent viral doses, displaying a similar response to that induced by the vector control (adv.β-gal) (data not shown). At an iso-equivalent dose of 15 pfu/cell, adv.p16 was approximately 100-fold more cytotoxic to CNE-1 than CNE-2Z cells.

On the basis of the parallel slopes of the survival curves (with or without RT), RT at 2 Gy appeared to contribute to the cytotoxicity of adv.p16 in an additive manner for both cell lines, causing a further 10-fold reduction of survival in CNE-1 cells.

In light of the impressive cytotoxic effects observed in the CNE-1 cells after adv.p16 gene transfer, we proceeded to examine the possible modes of death in these cells.

Adenoviral-mediated p16 gene transfer has been demonstrated to block cell proliferation in various cancer models (14, 21, 29, 30). To determine the effect of p16 overexpression on cell cycle distribution, CNE-1 and CNE-2Z cells were treated with either adv.p16 infection (2 pfu/cell) or RT (2 Gy), and then examined by propidium iodide flow cytometry 24 or 48 h later. A 10% increase in the proportion of cells in the G₀-G₁ phase relative to untreated or adv.β-gal-infected cells was observed after adv.p16 therapy (from 62% to 72%) (Fig. 4A). G₁ arrest occurred concurrently with a reduction in the percentage of S-phase cells. RT alone perturbed the cell cycling slightly with a higher proportion in the G₂-M phase.

We could not detect any effect of adv.p16 infection on cell cycle at 48 h because by this time, the cells have become confluent, with >85% of control cells already in the G₀-G₁ phase (data not shown). CNE-2Z cells were also examined for possible changes in cell cycling as a result of adv.p16 treatment. As can be observed in Fig. 4B, no changes were detected in these cells as a result of either adv.p16 or adv.β-gal infection, although RT had a similar effect as with the CNE-1 cells.

Numerous studies have demonstrated that p16 overexpression can facilitate premature senescence in both transformed and non-transformed cells (31–35). One striking effect of adv.p16 therapy on the CNE-1 cells is the change in morphology characterized by increased cell size and flattening, reminiscent of cellular senescence. One biochemical feature of senescence is the expression of SAB. Twenty-four hours following adv.p16 infection (15 pfu/cell) of CNE-1 cells, SAB activity was detected, suggesting that p16 can induce senescence in NPC cells (Fig. 5, A and B). There was evidence of both a time- and dose-dependent increase in the absolute number of SAB-expressing cells after treatment with adv.p16 (Fig. 5C). However, the proportion of SAB-expressing cells remained at less than 10%, suggesting that senescence may not play a major role in adv.p16-induced cytotoxicity of CNE-1 cells.

Many investigators have reported that p16 overexpression can induce apoptosis and necrosis in several human cancer models (16, 17, 20, 30, 36–38). To determine whether these processes contribute to p16 cytotoxicity in CNE-1 cells, cells were stained with AO-EB 48 h after adv.p16 infection, and/or 24 h after RT (2 Gy). Cell viability was determined by the differential uptake of the two fluorescent DNA-binding dyes: AO (green) and EB (orange). Treated cells were analyzed and counted, and the data are summarized in Table 1.

Twenty-four hours after CNE-1 cells were treated with RT, morphological changes associated with apoptosis were observed (Fig. 6A). Treatment of CNE-1 cells with adv.p16 (60 pfu/cell) resulted in significant induction of necrosis (20%) (Fig. 6B and Table 1).

DNA fragmentation was performed as an alternate assay to corroborate the modes of cell death determined by AO-EB staining. Two days following treatment of CNE-1 cells with adv.p16 (25 pfu/cell), the DNA agarose gel displayed a combination of both smearing and nuclear fragmentation, with smearing apparently being more prominent (Fig. 7). While DNA smearing is a feature of necrosis (DNA is extensively degraded) (27), DNA laddering is an indication that cells are undergoing apoptosis. Together, these data demonstrate that adv.p16-infected CNE-1 cells undergo both necrosis and apoptosis, with the former perhaps playing a more important role than the latter.

The impact of adenoviral-mediated p16 gene transfer on tumor formation was assessed for both CNE-1 and CNE-2Z cells in vivo. Approximately 4 × 10⁶ cells (untreated, infected with adv.β-gal or adv.p16) were injected into the gastrocnemius muscle of SCID mice 24 h after in vitro infection. Mice given injections of mock or adv.β-gal-infected CNE-2Z cells were sacrificed at 60–70 days due to tumor burden. In addition, tumor formation was consistently detected for the CNE-2Z cells infected with adv.p16 (Fig. 8A).

In contrast, for the mice given injections of CNE-1 cells treated with 100% adv.p16 (25 pfu/cell), no tumors ever formed, even after monitoring for over 100 days (Fig. 8B). Interestingly, for the mice given injections of 75% of the CNE-1 cells having been infected with adv.p16 (25 pfu/cell), with the other 25% uninfected, tumors were still detectable by 40 days postinjection. This result indicates that close to 100% tumor transduction in vivo is necessary, to achieve complete tumor cure.

The results presented in this study provide important evidence that adenoviral-mediated p16 gene transfer induces significant cytotoxicity in NPC cells lacking p16 function. Sensitivity is strongly correlated with level of endogenous p16, being most effective in cells with undetectable p16 expression under basal conditions.

On the basis of our previous extensive experience with CNE-1 and CNE-2Z cells, these two NPC cell lines share identical p53 mutations and display similar morphological features. In addition, they exhibited comparable sensitivity to various treatment modalities, including RT, adv.p53 gene therapy, and hyperthermia (25, 26, 39). Therefore, their disparate p16 kinetics and sensitivity to adv.p16 treatment is almost certainly related to their distinct endogenous p16 expression. This correlation is corroborated by other reports in the literature (14, 19–21) concerning adv.p16 efficacy in other human cancer models.

The differential sensitivity of cancer cells to gene transfer being associated with the endogenous level of the targeted gene has been previously documented. This is particularly evident with adv.p53 therapy, wherein cells that are null or mutant for p53 were far more sensitive to exogenous p53 therapy than cells harboring the wild-type gene (40, 41). Although the precise mechanism behind this finding remains unclear, it might reflect a homeostatic balance between cancer cell survival and apoptosis (40–43). Specifically, if the proliferation or survival signal of a cancer cell were dependent on the mutational defect of a specific gene, it is conceivable that introduction of the wild-type gene into the cell will perturb the fine balance of pro- and anti-proliferative signals, sensitizing the cancer cell toward death. In contrast, introducing a therapeutic gene that was normally expressed or regulated in a cancer cell would be expected to cause minimal effect on cell survival (40, 41).

However, because these two NPC cell lines are non-isogenic, a definitive conclusion relating to endogenous p16 status cannot be drawn. There may be a myriad of other genetic differences that influence sensitivity to adv.p16 therapy. Nevertheless, until such factors have been elucidated, endogenous p16 status appears to closely correlate with sensitivity to adv.p16-mediated cytotoxicity.

The inverse relationship between p16 and pRb expression in both NPC cell lines (Fig. 1) is consistent with previous evidence demonstrating that p16 is capable of down-regulating pRb expression at the transcriptional level (16, 44). Therefore, the decline in pRb protein levels by 24 h post-adv.p16 infection (Fig. 1) likely reflects an earlier decline in pRb mRNA levels and normal turnover of the pRb protein (half-life > 8 h) (45). However, the possibility that p16 also affects the half-life of pRb cannot be formally excluded without further experimentation. It is conceivable that p16 and pRb could counterbalance each other, to maintain cell cycle regulation. It has been reported that if pRb expression were unresponsive to p16 regulation, p16 would lose its ability to inhibit cell cycle progression (46).

The combination of adv.p16 and RT resulted in an additive cytotoxic interaction in both NPC cell lines (Fig. 3). Because other groups have reported that p16-induced radiosensitization is a p53-dependent event (47, 48), and that both cell lines are heterozygous mutants for p53, it is predicted that RT-mediated cytotoxicity for our NPC model would be merely additive. Introduction of wild-type p53 to our NPC cell lines before adv.p16 infection would allow us to address the issue of p16/p53-mediated radiosensitization, which could be a subject for future studies.

To elucidate the biological impact of exogenous p16 on CNE-1 cells, various modes of cytotoxicity were examined, including cell cycle arrest, senescence, apoptosis, and necrosis. Our results indicate that adv.p16-mediated effect is likely multimodal in its contribution to cytotoxicity and tumor suppression. The increase in G₀-G₁ arrest in CNE-1 cells following p16 gene transfer clearly validates p16 as an important G₁ checkpoint regulator. Similar results have already been reported, whereby stable transfection of p16 into CNE-1 cells led to significant G₁ arrest (49). In contrast to adv.p16, RT induced G₂-M arrest in both CNE-1 and CNE-2Z cells. A number of molecular changes might have occurred in these cells, such as reduced cyclin B expression and altered phosphorylation of CDK1, which could prevent proper formation of nuclear cyclin B-CDK1 complexes, resulting in G₂ block (50–52).

Senescence is identified by finite number of doublings that primary cells can undergo in culture, along with expression of certain biomarkers, such as SAB (53). Little is understood about the origin and function of SAB, which is thought to be an alternatively spliced isoform of lysosomal β-galactosidase, with increased activity in senescent cells (54). The finding of increased SAB activity in CNE-1 cells after adv.p16 infection indicates that a subset of the treated population has entered senescence (Fig. 5). This is consistent with previous studies showing that adenoviral-mediated p16 overexpression can induce premature senescence (30, 34).

Although apoptosis was not a major feature of p16-mediated cytotoxicity for the CNE-1 cells, others have reported that exogenous p16 can induce significant tumor cell apoptosis (55). In a recent study of several human cancer cell lines, stable transfection of p16 was shown to restore anoikis (56), the induction of apoptosis due to loss of anchorage to surrounding tissues. The observation that exogenous p16 expression can also down-regulate the anti-apoptotic protein bcl-2 (19) further strengthens the association between p16 and apoptosis.

We observed that adv.p16 (alone or in combination with RT) led to significant necrosis (∼20%), consistent with other reports for prostate (16) and ovarian cancer models (36). Although p16-induced necrosis has been described in many studies, the mechanism remains to be elucidated. In the final analysis, we have examined several potential modes of cytotoxicity, including apoptosis, necrosis, arrest, and senescence. The arithmetic sum of all of these data will still not fully account for the 99.9% clonogenic death of CNE-1 cells after infection with adv.p16 (25 pfu/cell in Fig. 3). This suggests at least two possibilities. The first is that there are additional modes of cellular damage, which result in failure of reproductive potential in cancer cells. A second consideration would be that arithmetic summation of these assays is an oversimplification of the clonogenic data, particularly when the former assays are short-term assessments (within 48 h of infection), whereas the clonogenic data are derived after 10–14 days of incubation.

Several studies have documented loss of tumorigenic potential for various human cancers in vivo following p16 gene transfer (20, 21, 29). In particular, Wang et al. (21) demonstrated significant growth suppression by stably transfecting p16 into a NPC cell line HK-1, which harbors mutant p16 (12). Their in vivo study also indicated that p16-expressing NPC transfectants failed to form tumors in athymic nude mice (21). In our study, tumorigenicity experiments revealed that adv.p16-treated CNE-1 cells failed to establish tumors, in contrast to the CNE-2Z cells. However, for this treatment to completely prevent tumor formation, 100% of the cells need to be infected. Even if 75% of the CNE-1 cells were infected, tumors will still form (Fig. 8B). This underscores the current challenge of cancer gene therapy whereby the vast majority of cancer cells need to express the therapeutic gene before complete tumor eradication can be achieved.

In summary, this study demonstrated that adv.p16 therapy is significantly cytotoxic to NPC cells that harbor low levels of endogenous p16 protein. The mechanism of p16 cytotoxicity appears to be multimodal, with necrosis, G₁ arrest, and senescence being most significant. RT provided an additive therapeutic benefit to adv.p16 for both cell lines by inducing G₂ arrest and apoptosis. Overall, given the exquisite sensitivity of CNE-1 cells to adv.p16 gene transfer, and the absence of p16 expression in the majority of primary NPC (9, 57), the therapeutic potential of this strategy warrants further detailed investigations.