Oncolytic Activity of p53-Expressing Conditionally Replicative Adenovirus AdΔ24-p53 against Human Malignant Glioma

Malignant gliomas have a poor prognosis, and patients with glioblastoma multiforme have a median survival of <1 year after diagnosis. Current treatment consists of surgical resection, irradiation, and chemotherapy, which are rather ineffective (1). The search for alternative treatment modalities has revived the concept of using oncolytic viruses to treat cancer (2, 3). In this respect, conditionally replicative adenoviruses appear as attractive anticancer agents that are currently being evaluated in clinical trials (4, 5). Conditionally replicative adenoviruses exert intrinsic anticancer activity through selective replication and lysis in cancer cells. In addition, the release of conditionally replicative adenovirus progeny by infected tumor cells provides a potential to amplify the oncolytic effect by lateral spread through solid tumors.

One strategy to engineer conditionally replicative adenoviruses is by introducing deletions in adenovirus genes that abolish viral functions that are dispensable in cancer cells but not in normal cells, such as interaction of viral E1A and E1B proteins with cellular pRb and p53 tumor suppressor proteins, respectively. Dysfunctions of pRb and p53 pathways are common in malignant glioma (6), making conditionally replicative adenoviruses potentially useful to treat this disease. In this respect, the conditionally replicative adenovirus AdΔ24 or dl922–947, which carries a deletion in the CR2 domain of the E1A gene that abrogates E1A binding to pRb, has already been shown to efficiently replicate in glioma cells and kill them, whereas quiescent normal cells were resistant (7, 8). These characteristics make conditionally replicative adenoviruses with pRb-binding deficient E1A particularly appealing for treatment of tumors in the central nervous system, where normal cells are essentially quiescent.

An important determinant for the anticancer potency of conditionally replicative adenoviruses is their efficacy in killing host cells and releasing their progeny. This involves multiple processes, including destruction of the cytokeratin network through cytokeratin-18 cleavage (9), cell death brought about by the E3–11.6kDa adenovirus death protein (10), and induction of p53-dependent and -independent apoptosis (11, 12). Notably, rapid adenovirus-induced cell death appears to require functional p53 (13, 14). However, 30–50% of gliomas carry defective p53 mutants, and others have compromised p53 function, e.g., through mdm2 amplification/overexpression or deletion or methylation of the CDKN2A/p14^ARF tumor suppressor gene (15, 16, 17). Hence, most if not all of the glioma cells have dysfunctional p53, which might delay conditionally replicative adenovirus-induced cell death and, thus, limit conditionally replicative adenovirus efficacy. Indeed, we found recently that exogenous expression of p53 during adenovirus replication in cancer cells accelerated cell death and progeny virus release and that an AdΔ24-derivative conditionally replicative adenovirus expressing p53 exhibited enhanced oncolytic potency in the majority of tested cancer cell lines in vitro (18).

Here we evaluate the anticancer activity of AdΔ24 and its p53-expressing derivative AdΔ24-p53 on different glioma cell lines and primary glioblastoma multiforme specimens in vitro, as well as on primary malignant glioma xenograft models in vivo. Furthermore, we investigate conditionally replicative adenovirus replication in normal human brain tissue.

U87MG and U373MG glioma cell lines were obtained from American Type Culture Collection (Manassas, VA), glioma cell line U251MG was a gift of Dr. Peter Sminia (Department of Radiotherapy, VU University Medical Center, Amsterdam, the Netherlands) and glioma cell line SF763 was provided by Dr. Dolores Dougherty (Neurosurgery Tissue Bank, University of California San Francisco, Brain Tumor Research Center, San Francisco, CA). Tumor suppressor gene status of these cell lines is as follows (19): U87MG has wild-type p53, p14^ARF/p16 deletion, and PTEN mutation; U251MG has mutated p53 (R273H), p14^ARF/p16 deletion, and PTEN mutation; U373MG has mutated p53 (R273H), wild-type p14^ARF/p16, and PTEN mutation; and SF763 has mutated p53 (R158L), p14^ARF/p16 deletion, and wild-type PTEN. All of the cell lines were maintained in F12-supplemented DMEM with 10% fetal calf serum (FCS) and antibiotics (Life Technologies, Inc., Paisley, United Kingdom).

The xenografts IGRG88 and IGRG121 derived from primary malignant gliomas diagnosed as malignant oligodendroglioma and glioblastoma multiforme, respectively, have been described before (20). IGRG88 expresses high levels of mutant p53 (F113I) lacking transactivation function and has deleted p14^ARF; IGRG121 carries wild-type p53 and p14^ARF (20). Tumors were passed as s.c. flank xenografts in nude mice, and for in vitro experiments cells were dissected from tumors and kept in short-term cultures in DMEM supplemented with 10% FCS and antibiotics (Life Technologies, Inc.).

Fresh tumor material was collected during brain tumor surgery at the Departments of Neurosurgery of the VU University Medical Center and the Academic Medical Center (Amsterdam, the Netherlands) and processed within 3 h after surgical resection. Pathological confirmation of diagnosis was made on tumor material that was processed for cell culture. All of the samples included in this study were diagnosed as glioblastoma multiforme. Primary glioblastoma multiforme cells were obtained after mechanical dissociation of tumor resection material and cultured in DMEM supplemented with 10% FCS and antibiotics. Conditionally replicative adenovirus replication and p53 function experiments were done before passage 10.

Normal brain tissue was obtained as waste material from the corticotomy tract during surgery for low-grade astrocytoma ∼2 cm away from the tumor (specimen 1) or by resection of neocortical tissue for treatment of epilepsy (specimens 2 and 3) performed at the Department of Neurosurgery of the VU University Medical Center. The tissue was cut into small pieces of ∼2 mm using 21-gauge needles and used immediately for experiments.

Regarding the use of human tissues in this study, ethics, informed consent, and approval comply with the principles stated in the Declaration of Helsinki of the World Medical Association (1989).

All of the recombinant adenoviruses have been described before. AdGFP is a replication-deficient adenovirus vector expressing cytomegalovirus immediate early promoter-driven enhanced green fluorescence protein (21). AdSVEp53 is a replication deficient adenovirus vector carrying a SV40 early promoter-driven human p53 expression cassette (22). The conditionally replicative adenovirus AdΔ24 (18) carries a 24-bp deletion corresponding to amino acids 122–129 in the CR2 domain of E1A necessary for binding to pRb (7), and AdΔ24-p53 was derived from AdΔ24 by inserting the SVEp53 expression cassette from AdSVEp53 (18). Wild-type adenovirus serotype 5 was kindly provided by Dr. Rob C. Hoeben (LUMC, Leiden, the Netherlands).

Viruses were plaque purified, propagated on 293 cells (American Type Culture Collection) for replication-deficient vectors or A549 cells (American Type Culture Collection) for (conditionally) replicating viruses, and CsCl gradient purified according to standard techniques. The E1Δ24 mutation and p53 expression cassette insertion were confirmed by PCR on the final products. Functional plaque-forming unit (pfu) titers were determined by limiting dilution plaque titration on 293 cells according to standard techniques. In all of the experiments, infections were normalized on the basis of pfu titers.

Functional p53 status was determined as described previously (18). Briefly, cells were transfected either with the reporter plasmid PG13-Luc, carrying the firefly luciferase gene driven by a p53-dependent promoter, or with the negative control construct MG15-Luc that has mutated p53-binding elements (23). After 48 h, luciferase expression was measured. The relative luciferase expression in PG13-Luc-transfected cells compared with MG15-Luc-transfected cells was used as a measure for functional p53 expression. Ratios between 0.5 and 2.0 were considered to represent a p53-deficient status, ratios of >2–10 an impaired or heterogeneous p53 activity, and ratios >10 were scored as representing a functional p53 status.

Glioma cells were seeded 5 × 10⁴ cells/well in 24-well plates for cell lines or, depending on cell size, at 2 to 5 × 10⁵ cells/well in six-well plates for short-term glioblastoma multiforme cultures to prepare subconfluent monolayers. Cells were infected with AdΔ24 or AdΔ24-p53 at 1 pfu/cell for 1 h at 37°C. After culture at 37°C for up to 25 days, cell survival was analyzed by WST-1 conversion assay as described (24). WST-1 conversion per hour was expressed either as a percentage of the conversion by uninfected control cells at the start of the culture (Fig. 1) or as a percentage of the conversion by uninfected control cells cultured in parallel (Fig. 2) after subtraction of background values of WST-1 incubated in the absence of cells. Analysis of statistical difference between susceptibility of cells to AdΔ24 and AdΔ24-p53 was done by Wilcoxon signed rank test using InStat software (GraphPad Software, San Diego, CA).

Short-term cultured xenograft-derived cells were seeded 10⁴ cells per well in a 96-well plate and the next day infected with AdΔ24 or AdΔ24-p53 at a 10-fold dilution titration ranging from 1 to 0.001 pfu/cell. After 19 days of culture, remaining viable cells were fixed and stained with crystal violet as described (18) and photographed at 25-times original magnification.

Approximately 2-mm small pieces of normal brain tissue were subjected to recombinant adenovirus infection individually at 10⁸ pfu/piece in 100 μl of DMEM-F12 with 2% FCS and antibiotics for 1 h at 37°C. Subsequently, pieces were washed once with culture medium, transferred to separate wells in a 96-well plate containing 200 μl of DMEM-F12 with 10% FCS and antibiotics, and cultured for 7 days at 37°C. Successful infection was confirmed 2 days later by inspection of AdGFP-infected tissue pieces under a fluorescence microscope. On day 7, pieces were washed once in PBS and subjected to three freeze-thaw cycles in PBS to release adenovirus. The lysates were cleared by centrifugation and used to determine protein content by BCA Protein Assay (Pierce, Rockford, IL) and adenovirus titer by end point dilution titration on 293 cells. To discriminate between input virus that survived 7 days of culture without replication and progeny released from infected cells, virus output in pfu per microgram protein was compared with the output of pieces infected with replication-deficient AdGFP. Analysis of statistical differences between groups was done by two-tailed nonparametric Kruskal-Wallis test using InStat software (GraphPad Software).

AdΔ24 and AdΔ24-p53 antitumor activities were evaluated against advanced-stage human malignant glioma IGRG88 and IGRG121 xenografts derived from primary tumors, as described previously (20). All of the animal experiments were carried out under the conditions established by the European Community (Directive 86/609/CCE). Briefly, female SPF-Swiss athymic nude mice bearing s.c. tumors of 100–300 mm³ were randomly assigned to treatment groups, and intratumoral injections were performed with AdSVEp53, AdΔ24, or AdΔ24-p53 at 10⁷ pfu, 10⁸ pfu, or 10⁹ pfu in 50 μl PBS (Life Technologies, Inc.) or with PBS only for 5 consecutive days. Different sites of the tumor were chosen for each injection. Tumors were measured two to three times weekly and tumor volume calculated according to the equation: V (mm³) = width² (mm²) × length (mm)/2. The experiment was stopped after 120 days when there were tumor-free survivors. Statistical significance in tumor growth rate (time to reach five times initial tumor volume) between treatment groups and controls was assessed by the two-tailed nonparametric Kruskal-Wallis test. Tumor regression was described using standard terminology including complete tumor regression (i.e., total tumor regression or tumor volume < 15 mm³ in at least two consecutive measurements) and partial tumor regression (i.e., ≥50% decrease in tumor volume in at least two consecutive measurements; Ref. 25).

Glioma xenografts were injected with AdΔ24 or AdΔ24-p53 as described above for 5 consecutive days: IGRG88 at 10⁸ pfu/injection and IGRG121 at 10⁷ pfu/injection. Tumors were excised at days 10 and 7, respectively, or at day 120 when long-term survivors carried a residual tumor, fixed in Glyo-fixx (Shandon, Pittsburgh, PA), paraffin-embedded, and cut into 4-μm-thick sections. Hematoxylin-eosin-safranin staining was performed on all of the xenografts for analysis of morphology. Masson’s trichrome staining for collagen type I using aniline blue (Reactifs RAL, Martillac, France) was done according to the manufacturer’s instructions. The polyclonal antiadenovirus hexon protein antibody AB 1056 (Chemicon International, Temecula, CA), diluted 1:300, was detected by a biotinylated rabbit antigoat immunoglobulin antibody streptavidin-horseradish peroxidase conjugate (DAKO), and the chromogen diaminobenzidine, as described before (20). Slides were counterstained with hematoxylin. Terminal deoxynucleotidyl transferase-mediated nick end labeling (TUNEL) was performed using the In Situ Cell Death Detection kit, AP (Roche Diagnostics GmbH, Mannheim, Germany), according to the manufacturer’s instructions and as described before (26). TUNEL-positive cells, displaying compaction or segregation of the nuclear chromatin or breaking up of the nucleus into discrete fragments, were counted per view at 100-times magnification. Five representative fields were chosen for counting; necrotic fields were excluded.

We first studied the oncolytic potencies of AdΔ24 and AdΔ24-p53 on four human glioma cell lines with different p53 status, i.e., U87MG cells with wild-type p53 sequence and confirmed expression of functional p53 in a p53 transactivation reporter assay (PG13:MG15 ratio 13.7); and U251MG, U373MG, and SF763 cells that carry p53 genes with missense mutations and are p53-deficient (PG13:MG15 ratios 2.0, 0.8, and 1.6, respectively). Cells were infected with AdΔ24 or AdΔ24-p53 at low multiplicity of infection and cultured for 7, 12, or 17 days, when cell viability was determined by WST-1 assay (Fig. 1). AdΔ24-p53 was more effective in killing glioma cells than AdΔ24 with the most significant efficacy improvement on the most resistant cell line, U373MG. The four human glioma cell lines were eradicated by AdΔ24-p53 with a similar rapid course comparable with AdΔ24-mediated killing of the most sensitive cell line, SF763. Hence, exogenous p53 expression overcame resistance to AdΔ24-induced oncolysis in p53 wild-type and p53-deficient glioma cells.

A panel of eight short-term cultures was established from primary human brain tumor specimens with histologically confirmed diagnosis of glioblastoma multiforme. Reporter assay for p53 transactivation function revealed that one of the samples was p53 positive, four were p53 deficient, and three had low but detectable p53 activity. The latter could indicate presence of functional p53 protein impaired in its activity by, e.g., enhanced degradation. Alternatively, specimens with low p53 activity could be composed of a mixture of p53 wild-type and p53-deficient cells. Short-term cultures were infected with AdΔ24 or AdΔ24-p53 at low multiplicity of infection and cultured until cytopathic effects became evident or for a maximum of 25 days, when cell survival was measured by WST-1 conversion assay (Fig. 2). Duration of culture was dependent on adenovirus receptor coxsackie adenovirus receptor expression level, with coxsackie adenovirus receptor-deficient samples cultured longer than coxsackie adenovirus receptor-positive samples (24). AdΔ24 killed only specimen VU-78 effectively, with >1 log cell kill in 12 days (and >2 logs in 14 days; data not shown). The other seven samples were much more refractory to this conditionally replicative adenovirus, with viability of four specimens not even affected after prolonged culture up to 25 days. In contrast, AdΔ24-p53 exerted detectable cytotoxicity on five of these resistant samples, causing 2–12-fold more cell kill than AdΔ24. The efficacy of AdΔ24-p53 against the panel of primary glioblastoma multiforme specimens was significantly enhanced compared with its parent control AdΔ24 (P < 0.01). Oncolysis enhancement did not correlate with p53 status of the cells.

We also compared the oncolytic potencies of AdΔ24 and AdΔ24-p53 in vitro on short-term cultures of the human malignant glioma xenografts IGRG88 (p53 mutant) and IGRG121 (p53 wild-type). Cells were infected at a range of multiplicity of infection (1 to 0.001 pfu/cell) and cultured for 19 days to allow multiple conditionally replicative adenovirus replication cycles. After culture, remaining viable cells were stained and photographed. As can be seen in Fig. 3, in both xenograft-derived cell cultures AdΔ24-p53 caused extensive cell death at an estimated 10–100 times lower viral dose than AdΔ24, indicating that AdΔ24-p53 replicated faster than its parental control lacking p53.

To evaluate the selectivity of AdΔ24 and AdΔ24-p53 in the context of malignant glioma treatment, we investigated their replication in normal brain tissue samples in vitro. To this end, individual small pieces dissected from three fresh brain explants were infected with AdΔ24 or AdΔ24-p53 at high multiplicity of infection (10⁸ pfu per tissue piece). Control samples were infected with wild-type adenovirus serotype 5 or with replication-deficient AdGFP. After 7 days of culture, the content of infectious adenovirus was determined in each tissue piece by titration on 293 cells and normalized by protein content. As can be seen in Fig. 4, significantly more virus was recovered from pieces infected with wild-type adenovirus serotype 5 and the two conditionally replicative adenoviruses than from AdGFP infected controls, confirming that wild-type adenovirus serotype 5 and the conditionally replicative adenoviruses replicated in normal human brain tissue ex vivo. However, the amounts of progeny virus produced were very low and highly variable. Median offspring recoveries per microgram tissue were below 10⁴ pfu, which corresponds to the amount of wild-type adenovirus serotype 5 produced by a single permissive primary cell (27).

We additionally investigated the antitumor activity of AdΔ24 and AdΔ24-p53 on the malignant human glioma xenografts IGRG88 and IGRG121 at an advanced tumor stage in vivo. Animals bearing s.c. IGRG88 xenografts on their flank were treated with intratumoral injections of AdΔ24 or AdΔ24-p53. Control animals were injected with PBS and, in one experiment, with the replication-deficient p53-expressing adenovirus vector AdSVEp53. As can be seen in Table 1, the two conditionally replicative adenoviruses AdΔ24 and AdΔ24-p53 caused tumor regressions and significant tumor growth delay compared with PBS controls. In contrast, AdSVEp53 did not result in any tumor regression or significant tumor growth delay. Although in general AdΔ24-p53 treatment tended to achieve longer, more significant tumor growth delay and more frequent tumor regression than treatment with AdΔ24, IGRG88 escaped treatment in most cases, and the differences between the two conditionally replicative adenoviruses were not significant. In contrast, intratumoral treatment of IGRG121 xenografts with AdΔ24 was highly effective, with five injections of 10⁸ pfu inducing 90% complete tumor regression and tumor-free survival (Fig. 5,A). To study the effect of p53 expression, the comparison between AdΔ24 and AdΔ24-p53 on IGRG121 was conducted at a reduced dose of 10⁷ pfu daily injections. At this dosage, AdΔ24 induced 30% tumor regression (2 of 10 complete tumor regression and 1 of 10 partial tumor regression) and significant median tumor growth delay of 24 days (P < 0.01; Fig. 5,B). Two animals survived tumor free at day 120; a third one had a residual tumor without viable tumor cells that showed signs of postnecrotic changes, avascular fibrosis, calcium deposits, fatty acid crystals, and macrophages, some of which were multinucleated, characteristic of foreign body reactions, as determined by histology (Fig. 5,D, left). AdΔ24-p53 (Fig. 5,C) induced more tumor regression (70%; 3 of 10 complete tumor regression and 4 of 10 partial tumor regression) and yielded a more significant median tumor growth delay of >113 days (P < 0.001). Three of 10 AdΔ24-p53-injected animals survived tumor-free 120 days after treatment; 3 additional animals had tumor residuals at day 120, 2 of which were without viable tumor cells (data not shown) and the third one with a minimal remnant of tumor cells with atypic nuclei encapsulated in stromal tissue (Fig. 5 D, right). Thus, 60% of AdΔ24-p53-treated animals survived at least 120 days with no or a minimal tumor residual. In conclusion, exogenous p53 expression by AdΔ24-p53 augmented AdΔ24 antitumor efficacy against IGRG121 xenografts in vivo.

We finally evaluated morphological changes, intratumoral viral replication, and induction of apoptotic cell death in vivo in IGRG88 (Fig. 6,A) and IGRG121 (Fig. 6 B) xenografts injected for 5 consecutive days with 10⁸ and 10⁷ pfu recombinant adenovirus, respectively. Tumors were excised shortly after treatment to study efficacy of conditionally replicative adenovirus delivery and immediate antitumor effects. As can be seen in the Masson’s trichrome and TUNEL panels, PBS-injected IGRG88 and IGRG121 xenografts were highly cellular, rather monotonous tumors with high mitotic activity, vascular proliferation, and some necrosis at baseline. The trichrome staining, furthermore, shows that IGRG88 tumors contained more collagen type I protein than did IGRG121 tumors. Conditionally replicative adenovirus delivery to IGRG88 tumors was very inefficient; hardly any cells in these tumors were infected. Consequently, histological changes were minimal. IGRG88 tumors remained dense without evidence of stromal reaction, although some fibrotic fascicles and infiltration of lymphocytes and macrophages were detected. In contrast, IGRG121 showed much more profound post-treatment changes, with scarce tumor cell nuclei within an edematous tissue; tumors were dissociated, showed lymphatic infiltrations, and extended foci of necrosis, most markedly in AdΔ24-p53-treated tumors. Immunohistochemistry for adenoviral capsid protein confirmed effective conditionally replicative adenovirus delivery by showing well-dispersed infection throughout IGRG121 xenografts injected with either conditionally replicative adenovirus. Replication was mainly detected in tumor borders and around areas of necrosis or fibrosis. TUNEL staining determined in control xenografts a low apoptotic rate in viable tumor areas of only 0.2% in IGRG88 and 0.4% in IGRG121. Consistent with the low infection efficiency, adenovirus treatment caused apoptosis in only few IGRG88 cells, with 0.4%, 0.9%, and 1.7% TUNEL-positive IGRG88 cells after injection of AdSVEp53, AdΔ24, or AdΔ24-p53, respectively. In contrast, in IGRG121, AdΔ24 and AdΔ24-p53 treatment induced much more significant apoptotic cell death with 7.1% and 8.1% TUNEL-positive tumor cells, respectively. Thus, induction of apoptosis correlated with Conditionally Replicative Adenovirus replication and appeared an important determinant for adenovirus mediated cell kill and tumor growth inhibition in malignant glioma. Because AdΔ24-p53 injected tumors contained more lysed cells and larger areas of necrosis, conditionally replicative adenovirus replication and apoptosis induction were probably even underestimated in these tumors.

Conditionally replicative adenoviruses were introduced recently as new agents for cancer therapy. Several preclinical evaluations already showed the potential of these replication-selective adenoviruses for the treatment of malignant brain tumors, such as high-grade glioma (20, 28, 29); and the conditionally replicative adenovirus ONYX-015 has entered into Phase I clinical trials for malignant glioma.

The efficacy of conditionally replicative adenoviruses against cancer including glioma is, however, limited by several factors, including reduced adenoviral entry and adenovirus-induced oncolysis. The former process can be augmented by redirecting infection via alternative molecules expressed on glioma cells (24, 28, 29). The latter process appears to involve the adenoviral E1B55k protein and the p53 tumor suppressor protein (14). The lack of functional p53 in most gliomas might, therefore, reduce sensitivity to conditionally replicative adenovirus-induced cytolysis. To enhance the oncolytic potency of AdΔ24-type conditionally replicative adenoviruses (7, 8), we recently constructed the new conditionally replicative adenovirus AdΔ24-p53 that expresses functional p53 during viral replication in cancer cells (18). Here, we evaluated the efficacy of AdΔ24 and AdΔ24-p53 against human malignant glioma in vitro and in vivo. Consistent with our previous findings on cancer cell lines of various tissue origins, we found that exogenous expression of p53 by AdΔ24-p53 led to enhanced oncolytic potency compared with its parent AdΔ24 on most glioma cell lines and, importantly, also on primary glioblastoma multiforme cultures tested in vitro and on at least one of two glioma xenograft models tested in vivo. The superior efficacy of AdΔ24-p53 appeared independent of the cellular p53 genetic background.

Interestingly, the two glioma xenograft models included in our study, which were both sensitive to AdΔ24 treatment and were killed more effectively by AdΔ24-p53 in vitro, responded quite differently to conditionally replicative adenovirus treatment at an advanced tumor stage in vivo. Conditionally replicative adenovirus injection into IGRG88 tumors resulted in very inefficient infection that allowed these tumors to escape treatment. In contrast, conditionally replicative adenovirus injection into IGRG121 xenografts resulted in dispersed infection of cells throughout the tumor. This allowed both conditionally replicative adenoviruses to cure animals from IGRG121 xenografts, with AdΔ24-p53 being more effective than AdΔ24. Immunohistochemistry confirmed that the adenovirus receptor coxsackie adenovirus receptor was expressed on IGRG88 tumor cells (data not shown), excluding low receptor expression as explanation for the poor infection of these cells. Perhaps the tumor composition played a role, because IGRG88 tumors contained more collagen than did IGRG121 tumors. Clearly, efficient and dispersed delivery of injected adenovirus was important for effective treatment. Because poor tissue penetration of injected adenovirus is recognized as an important limitation in treating patients with malignant glioma (30), our findings underscore the importance of evaluating conditionally replicative adenovirus efficacy in established tumors in vivo.

The clear conditionally replicative adenovirus replication and apoptosis induction observed after conditionally replicative adenovirus injection into sensitive IGRG121 xenografts suggested a mechanistic link between oncolysis and apoptosis induction. However, although IGRG121 xenografts were treated more effectively by AdΔ24-p53 we did not detect significant difference in apoptosis induction between AdΔ24- and AdΔ24-p53-injected IGRG121 tumors. We envision three alternative explanations for this. First, the extensive post-treatment changes in IGRG121 cells may have underestimated the apoptosis score in AdΔ24-p53-treated tumors, because TUNEL positive cells can only be determined in viable tumor cell areas. Second, the TUNEL analysis was performed only a few days after conditionally replicative adenovirus injection, when antitumor effects were apparent in both AdΔ24- and AdΔ24-p53-treated tumors, whereas differences in tumor growth only became evident after ∼1 month (see Fig. 5), which is consistent with successful treatment requiring multiple cycles of conditionally replicative adenovirus replication and lateral spread. Third, it cannot be excluded that the mechanism for conditionally replicative adenovirus–induced oncolysis enhancement by p53 expression might be distinct from classical p53-dependent apoptosis. The latter would be in agreement with the proposed importance of the p53/E1B55k complex in this process (14).

Finally, we wished to obtain preliminary information on the selectivity of AdΔ24-type conditionally replicative adenoviruses in normal brain tissue. We used a small set of fresh normal human brain specimens cultured in vitro and found very low but significant replication of AdΔ24 and of its derivative AdΔ24-p53 in these explants. Although this was somewhat surprising and disappointing, it was not in complete disagreement with earlier findings. Whereas replication of conditionally replicative adenoviruses with pRb-binding deficient E1A was reported attenuated in comparison to wild-type adenovirus serotype 5 in quiescent normal cells including brain astrocytes (7, 8, 28, 31), this was not found for proliferating normal cells (8). In the ex vivo experimental setting used herein, the proliferation status of the cells in the cultured brain explants was not assessed. Brain injury can induce proliferation of astrocytes (32), and this might also have happened in the explants. Notably, conditionally replicative adenoviruses with pRb-binding deficient E1A have already been made more selective by combining E1A-CR2 deletion with a second E1A mutation or with the E2F-1 promoter driving early gene expression (31, 33). Because the concept of oncolysis enhancement by p53 expression should also apply to such conditionally replicative adenoviruses with additionally restricted pRb-selectivity, the combination of these favorable attributes to design safe and efficacious conditionally replicative adenoviruses for glioma treatment is warranted. In addition, such conditionally replicative adenoviruses should incorporate the clearly established improvement of glioma cell infection enhancement (24, 28, 29).

In conclusion, we have shown that expression of functional p53 enhances the oncolytic potency of a conditionally replicative adenovirus in malignant glioma cell lines, primary cultures, and xenografts. Hence, enhancement of conditionally replicative adenovirus potency through expression of functional p53 may have utility for more effective treatment of malignant glioma.

We thank Elisabeth Connault for excellent technical assistance with histology and Patrice Ardouin and the staff of the animal facilities of the Institut Gustave Roussy.