Highly Augmented Cytopathic Effect of a Fiber-mutant E1B-defective Adenovirus for Gene Therapy of Gliomas¹

Gliomas are refractory to conventional therapies, such as chemotherapy and radiotherapy, partly because about 40% of gliomas harbor the p53 mutation that renders the gliomas resistant to this therapeutic approach (1, 2). Therefore, gene therapy targeting gliomas with the p53 mutation is an important therapeutic strategy. In an attempt to restore the p53-mediated apoptotic pathway, p53 genes have been transduced to gliomas through viral vectors (3). Because transduction of the p53 gene induces apoptosis in cultured neurons (4), a method to limit the transduction of p53 genes only to gliomas has yet to be developed. Another promising approach is to use an E1B 55-kDa gene-defective Adv,³ ONYX-015 (5, 6), which replicates in cancers with the p53 mutation, but not in normal tissues. It is of note that controversial results have been reported concerning the relationship between the replication and cytopathic effect of Adv-E1AdB and the constitutive p53 status of the cells infected with Adv-E1AdB (7, 8, 9). On the other hand, Kirn et al. (10) reported that a significant number of p53-mutant head and neck tumors injected with Adv-E1AdB (ONYX-015) had undergone considerable destruction without adjacent normal tissue damage. Although the mechanisms of the therapeutic effect remains to be fully elucidated, the clinical study data justify the therapeutic approach using Adv-E1AdB.

Because the total dose of Adv for the treatment of gliomas is restricted due to the toxicity of Adv to the brain (11), augmentation of transduction efficiency and more specific infection of this Adv (Adv-E1AdB) to gliomas are required. Recently, using an adenoviral vector with a fiber mutant, F/K20, which has a stretch of 20 lysine residues added at the COOH-terminus of the fiber, the transduction efficiency to gliomas was increased by more than 10 times compared with that of the adenoviral vector with the wt fiber (12). In this study, to improve the cytopathic effect of Adv-E1AdB with wt fiber (Adv-E1AdB-F/wt), we constructed Adv-E1AdB with F/K20 (Adv-E1AdB-F/K20) and examined its antitumoral effect.

U-373 MG human glioblastoma (ATCC, Manassas, VA), rat C6 glial tumor cells (ATCC), human embryonic kidney 293 cells (ATCC), and rat 9L gliosarcoma (13) were maintained in DMEM supplemented with 10% fetal bovine serum. The PC-12 rat adrenal pheochromocytoma cells (ATCC) were maintained in DMEM containing 10 ng/ml nerve growth factor, 1% fetal bovine serum, and 2% horse serum for 5 days before the assay of transduction efficiency. Female BALB/cAnNCjr-nu/nu mice, 6 weeks of age, were purchased from Japan Charles River (Yokohama, Japan).

Here we cite the reported nucleotide numbers (nt) in the Ad5 genome (GenBank M73260) to define the position of the restriction enzyme sites. An Ad5 mutant with an E3 deletion, lacking 1879 bp from nt28592 to nt30470 (Ad5dlX; Ref. 14), was used for vector construction. The 3471 bp SacII fragment (nt354 to nt3824) from Ad5dlX was subcloned into the SacII site of pBluescript SKII+ (Stratagene), resulting in pSKII+[S-S]. The 985-bp SacII/XbaI fragment (nt354 to nt1339) from pSKII[S-S] was subcloned into the SacII/XbaI sites of pBluescript SKII+, resulting in pSKII+[S-Xb]. The AflIII site (nt458) of pSKII+[S-Xb] was blunted and ligated with a ClaI linker pdCATCGATG, resulting in pSKd+[A-Xb]c. The 1861-bp PvuII fragment (nt623 to nt2484) from pSKII+[S-S] was subcloned into the SmaI site of pCI (Promega), resulting in pCI+[P-P]. The 1396-bp XbaI/BamHI fragment from pCI+[P-P], which contains the XbaI/PvuII (nt881 to nt2484) from Ad5 and the SV40 late poly(A) addition signal, was subcloned into the XbaI/BamHI sites of pSKd+[A-Xb]c, resulting in pSKd+[A-P]c.

An E1B176R DNA fragment with a CGA to TGA mutation at the 3rd codon of the E1B55K open reading frame was obtained by PCR, using the template pSKII+[S-S] and the following primers, the sequences of which are: 5′CGGAATTCACCATGGAGGCTTGGGAGTGTTTGGAAG3′ and 5′CAGCAGGTACCCCCCGCTCAGATGGGTTTCTTCACTCCATTTATCCTTTA3′. The ≈300-bp PCR product was digested with EcoRI/KpnI and subcloned into the EcoRI/KpnI sites of pBluescript SKII+, resulting in pSKII+E1B176R (TGA55K). The sequence of the fragment was confirmed to be identical with the expected sequence by DNA sequencing. The 278-bp SacI/KpnI fragment from pSKII+E1B176R (TGA-55K) was subcloned into the SacI/KpnI sites (corresponding to nt1770 and nt2048 of Ad5, respectively) of pSKd+[A-P]c, resulting in pSKd+[A-P] (TGA55K). The ClaI fragment from pSKd+[A-P] (TGA55K) containing E1A, E1B19K, and SV40 poly(A) addition signal sequences, which was designated E1AdB, was subcloned into the ClaI site of pAx-cw (14), resulting in pAxE1AdB. In summary, the pAxE1AdB contains a C to T point mutation at nt2025, and the ≈240-bp late poly(A) addition signal from SV40 in place of the 844-bp deletion from the PvuII (nt2484) to the BglII (nt3328) site in the E1B55 region.

Recombinant Adv was constructed essentially according to the procedure described previously (14), using Ad5dlX DNA-TPC and pAx cosmid. Briefly, the Adv AxE1AdB was generated by cotransfection of 293 cells with the pAxE1AdB cosmid DNA together with the Ad5dlX DNA-TPC, which had been digested with NsiI. After plaque purification and restriction enzyme mapping, AxE1AdB adenoviral clones without the artificial ClaI site at the left-side end of the inserted E1A were used. The TGA stop codon at the 3rd codon of E1B55K in these AxE1AdB adenoviral clones was confirmed by DNA sequencing.

The generation of fiber-mutant Adv vectors was carried out essentially as described previously (12). Briefly, the fiber-mutant construct in the plasmid that consisted of the right-side two-thirds of the Ad5 genome (pTR), was cotransfected with Ad5 DNA-TPC, efficiently yielding the recombinant Adv with the fiber mutant that has a linker and a stretch of 20 lysine residues added at the COOH-terminus of the fiber. The DNA-TPC from the mutant Adv was then used to produce a second-step recombinant Adv with deletion of E1B55K to generate Adv-E1AdB-F/K20. Adv-lacZ-F/K20 and Adv-lacZ-F/wt were constructed as described previously (12). The transduction efficiency of Adv with wt fiber and Adv with F/K20 was compared by infection with Adv-lacZ-F/K20 and Adv-lacZ-F/wt. Staining with X-Gal (5-bromo-4-chloro-3-indoyl-b-d-galactopyranoside) was performed as described previously (15).

Twenty thousand U-373 MG cells were infected with Adv-lacZ, Adv-E1AdB-F/wt, or Adv-E1AdB-F/K20 at a MOI of 30. On day 6, the degree of replication of Adv was quantitated by measuring the titer of Adv by a standard plaque formation assay using 293 cells.

Two thousand U-373 MG cells/well in 96-well microtiter plates were infected with Adv-E1AdB-F/K20, Adv-E1AdB-F/wt, or Adv-lacZ-F/K20 at MOI of 0.4, 2, 10, 50, and 250 (n = 8). On day 5, the percentage of surviving cells was measured using a modification of the MTT assay (16). Briefly, 100 μl of 0.5 mg/ml MTT (M-2128; Sigma Chemical Co.) solution was added to each well. The U-373 MG cells were incubated for 1 h at 37°C, and then 100 μl of 100% isopropanol and 0.04N HCl were added. Colorimetry was performed using the microplate reader (Model 3550; Bio-Rad, Hercules, CA). For each MOI, more than two experiments were performed, each representing the average of eight wells. Cell survival was expressed as the mean ± SE of the percentage of surviving cells relative to that in the control groups without infection.

U-373 MG tumor xenografts of 2.5 mm in diameter, from mice bearing U-373 MG tumors, were transplanted s.c. into BALB/cAnNCjr-nu/nu mice. Tumors with diameters of 5–10 mm developed 14 days after transplantation. The mice were divided into three groups (n = 6/group) and received intratumoral injection of Adv-E1AdB-F/K20, Adv-E1AdB-F/wt, or Adv-lacZ-F/K20 at 6 × 10⁶ pfu in a 50-μl volume using a 26-gauge needle, on days 0, 1, and 2. The tumor diameter was measured twice a week, and the volume (product of 0.4 × length × width × width) was calculated. Tumor volumes are expressed as the mean ± SE. Statistical analysis was performed using ANOVA, followed by Fisher’s exact test, and P < 0.05 was considered to be statistically significant.

Immunohistochemical analysis for the presence of Adv hexon protein was done by one of two methods. The first method involved the use of acetone/methanol (1:1)-fixed cultured tumor cells on a Lab Tek Chamber slide (Nalge Nunc International, Naperville, IL) and the DAKO LSAB kit (DAKO Japan, Kyoto, Japan), according to the manufacturer’s instructions. Briefly, the primary antibody (MAB805; Chemicon International, Temecule, CA) was applied to the cultured cells at 1:400 dilution, and then a biotinylated goat antimouse secondary antibody was applied, followed by a streptavidin-horseradish peroxidase conjugate. The second method involved the use of formalin-fixed, paraffin-embedded tumor sections. The tumor sections were pretreated with 0.05% Pronase (S2013; DAKO Japan) for 10 min at room temperature after the removal of paraffin. The primary 1:400 diluted antibody (MAB805) was applied overnight at 4°C, followed by application of streptavidin-horseradish peroxidase conjugate. 3–3′ Diaminobenzidine tetrahydrochloride and 3-amino-9-ethyl carbazole were used as the chromogens for the cultured cells and in vivo tumor tissues, respectively. Counterstaining was done with hematoxylin. The control stainings were performed using the isotype-matched mouse monoclonal antibody (IgG1κ), instead of the primary.

We examined the transduction efficiency of Adv-F/K20 for lacZ (Adv-lacZ-F/K20) and Adv for lacZ with wt fiber (Adv-lacZ-F/wt) in U-373 MG cells (Fig. 1 A). The MOI for transduction of 50% of the population (ED₅₀) of Adv-lacZ-F/K20 and Adv-lacZ-F/wt was 17 and 1.8, respectively. Similarly, in other glioma cells (A-172, U251, and T98G), the ED₅₀ of Adv-lacZ-F/K20 was 10 times higher than that of Adv-lacZ-F/wt (12). We also tested the transduction efficiency of Adv-lacZ-F/K20 and Adv-lacZ-F/wt in rat glioma cells (i.e., C6 and 9L), as well as differentiated rat PC-12 cells, which is used as a model of neurons (17). The transduction efficiency of Adv-lacZ-F/K20 in C6 and 9L cells, was three and four times higher than that of Adv-lacZ-F/wt in the respective cell lines, whereas the transduction efficiency of Adv-lacZ-F/K20 in differentiated PC-12 cells was about one-third that of Adv-lacZ-F/wt (data not shown). Although additional studies are needed to clarify the differential transduction efficiency in human neural tissues, these results suggest that the F/K20 mutation would preferentially support Adv infection to gliomas, but not neurons.

The degree of in vitro replication of Adv-E1AdB-F/K20, Adv-E1AdB-F/wt, and Adv-lacZ-F/wt was measured in U-373 MG cells. On day 6 after infection, the viral titer of Adv-E1AdB-F/K20 was 2710 times higher than that of the initially administered Adv, whereas the viral titer of Adv-E1AdB was 254 times higher than that of the initial administration. Thus, the replication activity of Adv-E1AdB-F/K20 was 11 times higher than that of Adv-E1AdB-F/wt. The higher transduction efficiency of Adv-F/K20 led to the higher replication of Adv-E1AdB-F/K20, most likely because the higher number of the Adv-E1AdB-F/K20 viral particles internalized into the initially infected cells induced more efficient viral replication, resulting in the rapid propagation of the infection to the neighboring cells. Immunohistochemical analysis revealed that nearly 100% of the U-373 MG cells infected with Adv-E1AdB-F/K20 at an MOI of 2 were stained with anti-Adv hexon monoclonal antibody 4 days after infection (Fig. 1,B, 3). On the other hand, only 2% of the U-373 MG cells infected with Adv-E1AdB-F/wt at the same MOI of 2, were stained (Fig. 1 B, 2). None of the cells infected with Adv-lacZ-F/K20 was stained with the antihexon antibody (Fig. 1B, 1 ), suggesting that the presence of hexon-positive cells is associated with Adv replication. The results indicate that the percentage of cells that contained replicated Adv by infection with Adv-E1AdB-F/K20 at the end of the 4-day in vitro culture was 50 times higher than that by infection with Adv-E1AdB-F/wt.

U-373 MG cells were infected with Adv-E1AdB-F/K20, Adv-E1AdB-F/wt, and Adv-lacZ-F/K20 at various MOI, and the cytopathic effect was analyzed by MTT assay 5 days after infection. The MOI for cell death of 50% of the population (ED₅₀) of Adv-E1AdB-F/K20 was 0.77, whereas the ED₅₀ of Adv-E1AdB-F/wt was 25 (Fig. 1,C). Thus, the in vitro cytopathic effect of Adv-E1AdB-F/K20 is about 32 times stronger than that of Adv-E1AdB-F/wt. This enhanced cytopathic effect of Adv-E1AdB-F/K20 is most likely due to the enhanced infectivity of Adv-E1AdB-F/K20, which has been demonstrated by the antihexon immunostaining for Adv (Fig. 1,B) and the reporter lacZ gene expression (Fig. 1 A).

The in vivo antitumoral effect of intratumoral injections of Adv-E1AdB-F/K20, Adv-E1AdB-F/wt, and Adv-lacZ-F/K20 was analyzed in nude mice bearing U-373 MG tumors. Intratumoral injection of Adv-E1AdB-F/K20 resulted in significant tumor growth inhibition compared with that in the tumors injected with Adv-E1AdB-F/wt (Fig. 2,A; P = 0.003) and with Adv-lacZ-F/K20 (Fig. 2,A; P = 0.03). Because the total pfu used in this study was as low as 1.8 × 10⁷ pfu/mouse, which is one-thirtieth of the total pfu used in a previous study (6), Adv-E1AdB-F/wt did not inhibit tumor growth. Adv-E1AdB-F/K20 had a drastic antitumoral effect at a MOI at which Adv-E1AdB-F/wt did not show an antitumoral effect. Distinct virus replication was documented in the U-373 MG tumors injected with Adv-E1AdB-F/K20 by immunohistochemical staining for Adv hexon protein (Fig. 2,B, 3 and 4), which was performed as early as on the 8th day after infection. Although <5% of the cells were hexon-positive 8 days after injection of Adv, the growth of tumors injected with Adv-E1AdB-F/K20 was completely inhibited up through 4 weeks after infection, suggesting that extensive in vivo replication of Adv-E1AdB-F/K20 occurred from the 8th day on. In addition, it is important to determine whether the direct cytopathic effect of virus replication or the bystander tumor-infiltrating immune cells mainly contributed to the therapeutic effect. Additional studies are necessary to document the time course of the intratumoral Adv replication and the invasion of immune cells. In the U-373 MG xenografts injected with Adv-E1AdB, hexon-positive cells were only very scarcely found (Fig. 2B, 2 ). Hexon-positive cells were not found in any of the U-373 MG xenografts injected with control Adv-lacZ-F/K20 (Fig. 2 B, 1).

In the gene therapy of cancers, gene transduction at high efficiency and with high specificity for cancers is required (18). Polylysine has been reported to interact with heparan (19) and polyanionic molecules (20), including chondroitin sulfate and mucins. Thus, the Adv with the mutant fiber containing polylysine would show increased transduction in various cell types (21). However, because the F/K20 mutation shows an increased transduction efficiency preferentially in gliomas (12), it might be possible that a putative specific receptor molecule for F/K20 is abundantly expressed in gliomas. Additional investigations are required to evaluate the mechanisms accounting for the greater efficacy of F/K20 in gliomas.

The high transduction efficiency obtained by the F/K20 mutant Adv is beneficial to the therapy of gliomas in various respects. First, an increase of transduction efficiency will augment the expression of transduced genes. In U-373 MG cells, the transduction efficiency of Adv-lacZ-F/K20 was nine times higher than that of Adv-lacZ-F/wt, and the cytopathic effect of Adv-E1AdB-F/K20 was 32 times stronger than that of Adv-E1AdB-F/wt. The high transduction efficiency of Adv-E1AdB-F/K20 led to high replication activity, which resulted in a remarkably augmented cytopathic effect of Adv-E1AdB-F/K20. It is to be noted that U-373 MG cells are deleted for p53, and it is possible that Adv-E1AdB-F/K20 may be more effective for killing glioma cells with loss of wild-type p53 function. Therefore, it would be interesting to evaluate the cytopathic effect of Adv-E1AdB-F/K20 in the gliomas with or without wild-type p53 function.

To obtain stronger cytotoxicity, a combination of proapoptotic gene expression with the E1AdB vector may be rewarding. Recently, we have shown that infection of Advs that encode proapoptotic genes, such as Fas ligand, induced a remarkable cytopathic effect in gliomas (22). The coexpression of proapoptotic genes with Adv-E1AdB-F/K20 could lead to an enhanced antitumoral effect by working together with viral gene products to induce apoptosis of the host cells.

Another advantage is that enhancement of transduction efficiency reduces the total dose of Adv to obtain the same level of cytopathic effect. It has been reported that injection of Adv doses over 10¹⁰ pfu into the human brain is toxic (11). Even if the damage of Adv infection is limited to only a small area of the brain, it might be fatal or severely disabling because the brain contains the important integral area of neural activity. Therefore, it is important to minimize the total dose of Adv administration as much as possible. It is highly advantageous to use Adv with F/K20 to reduce the toxicity of Adv infection because of its high transduction efficiency to gliomas. Promoters such as E2F (23) and myelin basic protein (24) have been reported to induce gene expression specifically in gliomas. Production of Adv-E1AdB-F/K20, in which the E1A gene is driven by one of these glioma-specific promoters, would be a promising approach to further restrict undesired viral replication. F/K20, the Adv fiber with the stretch of lysine residues, itself did not show any cytotoxic effect in fibroblasts such as WI38 cells (12) and nerve growth factor-treated PC-12 cells (data not shown), and infection of Adv-EIAdB did not damage normal adjacent tissues in head and neck cancers (10), suggesting that the infection of Adv-E1AdB-F/K20 would be relatively safe to normal tissues. However, to evaluate the undesired effects of Adv-E1AdB-F/K20 for normal human brains, careful basic and clinical studies using this vector is further required.

Adv induces an immune response to the vector by neutralizing antibodies, which mainly consist of antifiber and antihexon antibodies (25, 26). Construction of an Adv with a different type of fiber such as F/K20, desirably in combination with a mutant hexon, might be useful to evade the immune response. Additional studies on the immune response to recombinant Adv capsid proteins are required to develop a vector system with which we can repeatedly administer the therapeutic Adv.

In summary, Adv-E1AdB-F/K20 has a significantly stronger cytopathic effect than Adv-E1AdB-F/wt. Although it remains to be determined whether Adv-E1AdB-F/K20 exerts any toxicity to normal brain tissues, this therapeutic approach would be highly promising for the treatment of gliomas.

We thank S. Satoh for technical assistance with the animal study, R. Sato for technical assistance with the cell culture, and Dr. H. Shinoura for assistance in the preparation of this manuscript.