Combining high selectivity of replication with fiber chimerism for effective adenoviral oncolysis of CAR-negative melanoma cells


Oncolytic adenoviruses constitute a new and promising tool for cancer treatment that has been rapidly translated into clinical trials. However, minimal or absent expression of the adenovirus serotype 5 (Ad5) receptor CAR (coxsackievirus and adenovirus receptor) on cancer cells represents a major limitation for Ad5-based oncolysis. Here, we report on the resistance of CAR-negative primary melanoma cells to cell killing by wild-type Ad5 (Ad5wt) even after high titer infection, thus underlining the need for tropism-modification of oncolytic adenoviruses. We engineered a new generation of oncolytic adenoviruses that exhibit both efficient target cell infection by swapping Ad5 fiber domains with those of Ad serotype 3, which binds to a receptor distinct from CAR, and targeted virus replication. Fiber chimerism resulted in efficient cytopathicity to primary melanoma cells, which was at least 104-fold increased relative to Ad5wt. Since viral infectivity mediated by such modified viral capsids was not cell type-specific, it was pivotal to carefully restrict adenoviral replication to target cells. Towards this end, we replaced both E1A and E4 promoters of fiber chimeric viruses by tyrosinase enhancer/promoter constructs. The resulting viruses showed melanoma-specific expression of E1A and E4 and combined efficient virus replication and cell killing in melanoma cell lines and primary melanoma cells with a remarkable specificity profile that implements strong attenuation in nonmelanoma cells, including normal fibroblasts and keratinocytes.


Viral oncolysis, or virotherapy, is an emerging strategy for the treatment of cancer based on tumor cell killing by virus infection and replication.1 Key features of viral oncolysis are the implementation of a therapeutic mechanism distinct from those of conventional therapies and the amplification of the therapeutic agent in the patient's tumor. Adenoviruses possess several attributes that suggest their exploitation as oncolytic agents.1, 2 These include a lytic replication cycle, high stability, efficient genome transfer, and low pathogenicity. Furthermore, the technology for virus production at high titers is established. Importantly, the adenoviral structure, genome, and replication cycle are well characterized, thus facilitating the engineering of these viruses for therapeutic purposes.

Major requirements for the exploitation of adenoviruses for viral oncolysis are the efficient infection of tumor cells and the restriction of viral replication and toxicity to target tumors, both of which necessitate manipulations to the adenoviral genome. In this respect, tumor selectivity of virus replication is pivotal for viral oncolysis and has been achieved for oncolytic adenoviruses by various strategies: (i) by mutation of adenoviral functions which are required for viral replication in normal cells, but which are dispensable in tumor cells, (ii) by expression of essential viral genes from heterologous, tumor-selective promoters, or (iii) by tumor cell-induced recombination events within the viral genome.1, 3 However, a major obstacle for adenoviral oncolysis, which commonly utilizes adenovirus serotype 5 (Ad5)-derived viruses, is the recent observation that cancer cells in situ or freshly purified from tumor specimen are often poorly transduced by Ad5 gene transfer vectors. This is due to a paucity of expression of the Ad5 receptor CAR (coxsackievirus and adenovirus receptor) in these cells.4, 5, 6 In contrast, established tumor cell lines frequently express CAR and have been previously the main substrate utilized for the evaluation of adenoviral oncolysis. Poor susceptibility of cancer cells to Ad5 infection might also explain the unsatisfactory therapeutic efficacy of oncolytic adenoviruses in previous clinical trials.7 Indeed, we show here that melanoma cells purified from skin metastases lack CAR and were resistant to cell killing by wild-type Ad5 (Ad5wt), even after infection at high titers. In the context of gene therapy, tropism-modification of adenoviruses for CAR-independent and/or tumor-targeted gene transfer has been extensively pursued.8, 9 Bispecific adapter molecules that simultaneously bind to the virus capsid and target cell surface have been developed and resulted in augmented and/or targeted gene transfer. However, this approach is less attractive for applications in virotherapy, because the targeting of progeny viruses is problematic. For this reason, tropism-modification of oncolytic adenoviruses by genetic capsid engineering appears to be the more promising approach in this context. Towards this end, CAR-independent gene transfer has been achieved after genetic incorporation of ligands into the capsid of adenoviral vectors or after swapping of Ad5 fiber domains with those of adenovirus serotypes that bind to distinct receptors (fiber chimerism).8 Cell-specific adenovirus infection, however, has turned out to be rather difficult to achieve with the genetic approach. To address the aforementioned obstacles, we report here on a new strategy for the development of effective oncolytic adenovirus that combines potent (but nonselective) cancer cell infection by fiber chimerism with highly tumor-specific viral replication by transcriptional targeting of the expression of both E1A and E4 genes.


Cytotoxicity of tropism-modified, replication-competent adenoviruses for melanoma cell lines and primary melanoma cells

New therapeutic strategies are critically needed for malignant melanoma, which is characterized by rapidly growing incidence and mortality rates and resistance of metastatic disease to conventional treatment regimens. We have reported previously on a novel oncolytic adenovirus that implements the melanoma-selective replication by expression of E1A from the human tyrosinase enhancer/promoter.10 However, we and others have revealed that melanoma cells in situ or freshly purified from tumor specimen express negligible or no CAR and are thus poorly transduced by Ad5-derived vectors.5, 6, 11 We also found upregulation of CAR expression in purified melanoma cells after prolonged culture (>30 passages, data not shown), indicating that expression of CAR by tumor cells in vitro can be a cell culture artifact.

Based on these observations and aiming at improved efficacy of adenoviral oncolysis for treatment of melanoma, we investigated the effect of genetic tropism-modification on adenovirus lytic potency in melanoma and other cells. For this purpose, melanoma cell lines, primary melanoma cells isolated from skin metastases, and nonmelanoma cells were infected with Ad5wt and tropism-modified derivatives thereof, Ad5RGD and Ad5/3, at 100 to 0.01 virus particles (vp)/cell. Ad5RGD contains an integrin-binding RGD peptide incorporated into the fiber HI loop12 and for Ad5/3 the cell-binding fiber knob domain was replaced by the Ad3 knob. Cytotoxicity was revealed by crystal violet staining of surviving cells (Figure 1a). In all melanoma cell lines, cell killing by Ad5/3 was superior to Ad5wt. In addition, Ad5/3 was superior to Ad5RGD in most melanoma cells. These results are intriguing in consideration of strong expression of CAR and integrins by all melanoma cell lines analyzed (for FACS analysis of CAR expression, see Nettelbeck et al6 and Volk et al11 and data not shown).

Figure 1

Cytotoxicity to melanoma cells of wild-type Ad5 (Ad5wt) and derivates thereof, Ad5/3 with a chimeric fiber that contains the Ad3 knob, or Ad5RGD with an RGD peptide inserted into the HI loop of the Ad5 fiber. (a) Melanoma cell lines (SK-MEL-28, Mel888, A375M, and Mel624), primary melanoma cells (patient 1), a nonmelanoma cell line that transcomplements E1 functions and allows for replication of AdCMVLuc (293), and normal nonmelanoma cells (NHF) were infected with Ad5wt, Ad5/3, Ad5RGD, or E1/E3-deleted AdCMVLuc at the indicated titers or were mock-infected (as depicted in the upper right scheme). When cell lysis was observed for Ad5/3 at 0.01 vp/cell (7 to 14 days postinfection, dependent on the cell type), remaining cells were fixed and stained with 1% crystal violet in 70% ethanol to visualize cell killing. NHF, normal human fibroblasts. (b) Analysis of CAR expression by primary melanoma cells isolated from skin metastases of two different patients. Flow cytometry after staining of cells with the anti-CAR monoclonal antibody RmcB (thick line) or isotype control (gray).

Importantly, we did not observe cell killing of CAR-negative (Figure 1b) primary melanoma cell cultures by Ad5wt at titers up to 1000 vp/cell (Figures 1a and 4). Thus, Ad5wt was not able to productively infect these cells in a CAR-independent fashion, for example via binding of the penton base to cellular integrins. In this regard, we previously reported that primary melanoma cells strongly express αvβ3 and αvβ5 integrins.11 Our observation that tumor cells can be resistant to Ad5wt infection is of key relevance as most oncolytic adenoviruses used in current preclinical and clinical studies are derived from Ad5. Importantly, viral lysis of primary melanoma cells was achieved with Ad5RGD or Ad5/3 (Figures 1a and 4). Ad5/3 was substantially more cytotoxic in these cells than Ad5RGD, resulting in cell killing potency that was at least two (Ad5RGD) or four (Ad5/3) orders of magnitude superior to Ad5wt. Ad5RGD and Ad5/3 were more cytopathic than Ad5wt in primary fibroblasts but not in 293 cells or in primary keratinocytes (see Figure 4). We conclude that CAR-negative primary melanoma cells can be resistant to Ad5wt (and thus presumably to conditionally replicative adenoviruses with Ad5 capsid) and that tropism-modification is a critical requirement for efficient adenoviral oncolysis of these cells. In this respect, Ad5/3 fiber chimerism proved more effective than incorporation of an RGD peptide into the Ad5 fiber. These results are in accord with our previous report on capsid-modified adenovirus gene transfer vectors.11

Figure 4

Efficient and melanoma-selective cell killing by fiber-chimeric, tyrosinase promoter-controlled oncolytic adenoviruses. Melanoma cells lines (Mel624, SK-MEL-28), primary melanoma cells (patient 1 and 3), nonmelanoma cell lines (A549, SKOV3.ip1), and normal primary cells (NHF, NHK) were infected with Ad5wt, Ad5/3, Ad2xTyr, Ad5/3.2xTyr, Ad5/3sk.2xTyr (not for patient 3), or AdCMVLuc at the indicated titers as depicted in the upper right scheme. When cell lysis was observed for Ad5/3 at 0.01 vp/cell, remaining cells were fixed and stained with crystal violet. NHF, normal human fibroblasts; NHK, normal human keratinocytes.

Development of fiber-chimeric adenoviruses with melanoma-restricted replication potency

To harness non-cell type-specific infectivity enhancement of adenoviruses by Ad5/3 fiber chimerism for virotherapy of cancer, it is pivotal to effectively restrict replication of the resulting recombinant adenoviruses to tumor cells. Towards this end, we combined within one virus fiber chimerism with a double-pronged strategy for transcriptional targeting of viral replication by expression of both E1A and E4 from a tyrosinase enhancer/promoter with upstream polyA signal (Figure 2). Polyadenylation sequences were included to prevent adverse read-through transcription from the viral ITRs. Our strategy to express E1A and E4 from promoters with similar specificity but different nucleotide sequence, here implemented with the human and murine tyrosinase enhancer/promoters, was chosen to prevent adverse recombination events during virus replication. Importantly, we found that both the human and murine tyrosinase enhancer/promoter mediate efficient and highly melanoma-selective gene expression (Nettelbeck et al13 and data not shown). Recombination events in oncolytic adenovirus genomes would limit therapeutic efficacy and prevent large scale production of corresponding virus therapeutics. Of note, the viruses generated in our study were stable and amenable for laboratory large scale production. In addition to specific promoters, the new generation of oncolytic adenoviruses generated in this study contained chimeric fibers consisting of either the Ad5 tail/shaft and Ad3 knob (Ad5/3.2xTyr), or of the Ad5 tail and Ad3 shaft/knob (Ad5t/3sk.2xTyr). The evaluation of the short-shafted virus Ad5t/3sk.2xTyr in this study was inspired by reports that indicated reduced liver tropism and toxicity of short-shafted adenoviruses.14, 15, 16 The replication and cytotoxicity of Ad5/3.2xTyr and Ad5t/3sk.2xTyr was compared to a matching virus with Ad5wt capsid, Ad2xTyr.

Figure 2

Engineering of oncolytic adenoviruses that combine fiber chimerism with transcriptional targeting of the expression of two essential adenoviral genes, E1A and E4. Schematic structures of the oncolytic adenovirus genomes generated in this study. Numbers refer to nucleotide positions of the adenoviral genome or of the tyrosinase genes. L/RITR, left or right inverted terminal repeat; E1Aenh, E1A enhancer; ψ, packaging signal; polyA, synthetic polyadenylation sequence; h/mTyrE, human or murine tyrosinase enhancer; h/mTyrP, human or murine tyrosinase promoter; E1AΔ24, E1A mutant with 24 nucleotide deletion in the conserved region 2; Ad5, adenovirus serotype 5; Ad3, adenovirus serotype 3.

Selectivity of expression of both E1A and E4 mRNA by Ad2xTyr, Ad5/3.2xTyr, and Ad5t/3sk.2xTyr (2xTyr viruses) was determined by real-time PCR (Figure 3a). Our results show two or more than one order of magnitude reduced mRNA copy numbers for E1A or E4, respectively, in nonmelanoma cells versus melanoma cells. Interestingly, lowest E1A and E4 mRNA copy numbers in nonmelanoma cells were observed for Ad5/3.2xTyr, whereas slightly higher E1A and E4 expression was observed for Ad5/3 versus Ad5wt. Compared with Ad5wt, E4 mRNA copy numbers were markedly reduced for Ad2xTyr, Ad5/3.2xTyr, and Ad5t/3sk.2xTyr in Mel624 melanoma cells and to a lesser extent in SK-MEL-28 cells. In contrast, reduction in mRNA copy numbers was less pronounced for E1A. Therefore, we next determined E4 protein expression by Western blot (Figure 3b). Our results show melanoma-specific expression of E4ORF3 protein by 2xTyr viruses. E4 protein expression by these viruses in melanoma cells was lower than for Ad5wt and Ad5/3, but clearly detectable. Attenuation of E4 expression by 2xTyr viruses seemed less pronounced at the protein level than at the mRNA level. Importantly, no E4ORF3 signal was seen for the 2xTyr viruses in SKOV3.ip1 cells, even after infection with high titers and with long exposure of the film.

Figure 3

Specific expression of E1A and E4 from tyrosinase enhancer/promoters by oncolytic adenoviruses. (a) Specific expression of E1A and E4 mRNA by tyrosinase enhancer/promoter-controlled oncolytic adenoviruses. Messenger RNA copy numbers were quantified by real-time PCR after infection of melanoma (Mel624, SK-MEL-28) and nonmelanoma (SKOV3.ip1) cells as described in Materials and methods. Results are shown as mRNA copy numbers per ng total cellular RNA (as detected with GAPDH primers) and after standardization with mean values for Ad5wt. All experiments were performed in triplicates; bars show mean values and error bars show standard deviations. (b) Specific expression of E4 protein by tyrosinase enhancer/promoter controlled oncolytic adenoviruses after infection of melanoma (Mel624) and nonmelanoma (SKOV3.ip1) cells. E4ORF3 protein was detected by immunoblot as described in Materials and methods.

In aggregate, our data confirm that the specificity of both the human and murine tyrosinase enhancer/promoter with upstream polyA is conserved after incorporation into the adenoviral E1 and E4 loci, respectively, and during viral replication. However, E4 expression from the murine tyrosinase enhancer/promoter is reduced compared with the endogenous E4 promoter in melanoma cells.

Analysis of cell killing potency of fiber-chimeric, tyrosinase promoter-controlled oncolytic adenoviruses

Having shown specific expression of E1A and E4, we next evaluated the cytotoxicity profile of the tropism-modified 2xTyr adenoviruses. Melanoma cell lines, primary melanoma cells, nonmelanoma tumor cells, and normal human fibroblasts (NHF) and keratinocytes (NHK) were infected with Ad2xTyr, Ad5/3.2xTyr, Ad5t/3sk.2xTyr, Ad5wt, Ad5/3, AdCMVLuc, or were mock-infected and cell survival was revealed by crystal violet staining (Figure 4). Ad2xTyr and Ad5/3.2xTyr resulted in efficient cell killing of the melanoma cell lines Mel624 and SK-MEL-28. However, in comparison to Ad5wt or Ad5/3, respectively, 10-fold higher titers were required for similar lytic potency. Of note, both Ad5/3 and Ad5/3.2xTyr, but not the matching viruses with Ad5 fiber, effectively lysed primary melanoma cells. In nonmelanoma cell lines and normal cells, Ad2xTyr and Ad5/3.2xTyr were severely attenuated relative to the matching E1 and E4 wild-type viruses. More than three to five orders of magnitude higher viral titers were required for similar cytotoxic effects. No cell killing was observed at any titer in NHF and SKOV3.ip1 cells. These results clearly demonstrate that the differential of E1A and E4 expression by 2xTyr and Ad5/3.2xTyr in melanoma versus nonmelanoma cells translates into melanoma-specific cell killing. Of note, the marked attenuation of E4 mRNA expression by these viruses in melanoma cells did not translate into a similarly severe attenuation of cell killing potency (or virus replication, see below).

Ad5wt and Ad2xTyr showed minimal or no cell killing of primary melanoma cells, even at 1000 vp/cell (Figure 4). Fiber chimerism had no influence on the cytotoxicity of oncolytic adenoviruses in Mel624 cells, but in SK-MEL-28 cells lytic potency of tropism-modified viruses was two orders of magnitude increased compared with Ad5 capsid viruses. Finally, Ad5/3 viruses were more potent than Ad5 viruses in SKOV3.ip1 cells and in normal fibroblasts, whereas no difference in cell killing potency between these capsid variants was observed in A549 cells or in normal keratinocytes.

In aggregate, our data revealed a remarkable specificity profile for Ad5/3.2xTyr. For example, Ad5/3.2xTyr showed a lytic potency that was more than four orders of magnitude stronger compared with Ad5wt in primary melanoma cells, but more than four orders of magnitude weaker in normal keratinocytes. Of note, these results were obtained in primary tumor cells and normal cells that represent the most stringent cell culture system and clinically relevant substrate for analysis of oncolytic adenoviruses.

To our surprise, the in vitro cell killing profile of Ad5t/3sk.2xTyr was considerably different from Ad5/3.2xTyr. For instance, cell killing was weaker than for Ad5/3.2xTyr in some cells (SK-MEL-28, primary melanoma, NHK), but stronger in others (Mel624, A549, SKOV3.ip1). Thus, a different cell binding profile mediated by the short shafted fiber results in a different cell killing profile of Ad5t/3sk.2xTyr compared with Ad5/3.2xTyr in vitro.

Analysis of genome replication of fiber-chimeric, tyrosinase promoter-controlled oncolytic adenoviruses

For characterization of oncolytic adenovirus replication, primary melanoma cells (patient 1), melanoma cell lines (SK-MEL-28 and Mel624), and SKOV3.ip1 cells were infected with Ad2xTyr, Ad5/3.2xTyr, Ad5t/3sk.2xTyr, Ad5wt, or Ad5/3, and viral genomes were quantified at 2 or 8 days postinfection (Figure 5). Overall, the results corroborated the cytotoxicity data. In primary melanoma cells, viral genome copy numbers were substantially higher for fiber chimeric viruses compared with Ad5 fiber viruses at 2 and 8 days postinfection. Furthermore, viral genome copy numbers increased dramatically from 2 to 8 days postinfection for the tropism-modified viruses but less for Ad5 fiber viruses. All viruses generated high DNA copy numbers in melanoma cell lines 8 days postinfection. In SK-MEL-28 cells, Ad5/3 fiber viruses generated higher genome copy numbers than Ad5 fiber viruses at 8 days postinfection. Tyrosinase viruses yielded somewhat lower titers than E1A and E4 wild-type viruses at that time point. These data are in accord with the results of the cytotoxicity assay for that cell line. In Mel624, all viruses yielded similar genome copy numbers at 8 days after virus infection. Increased cytotoxicity but similar genome replication kinetics of E1A and E4 wild-type viruses versus 2xTyr viruses in Mel624 cells might result from stronger expression of toxic E4 proteins. Ad2xTyr and Ad5/3.2xTyr viruses showed only minimal DNA replication in SKOV3.ip1 cells. In contrast, E1A and E4 wild-type viruses generated genome copy numbers similar to or higher than those obtained in melanoma cells, indicating efficient DNA replication of E1A/E4 wild-type viruses and strong attenuation of 2xTyr viruses.

Figure 5

Efficient and melanoma-selective replication of fiber-chimeric, tyrosinase-promoter-controlled oncolytic adenoviruses. Melanoma cells lines (Mel624, SK-MEL-28), primary melanoma cells (patient 1), and the nonmelanoma cell line SKOV3.ip1 were infected with Ad5wt, Ad5/3, Ad2xTyr, Ad5/3.2xTyr, or Ad5/3sk.2xTyr at 1 vp/cell. Cells were harvested at 2 or 8 days postinfection, DNA was purified and adenoviral genome copy numbers were determined by real-time PCR as described in Materials and methods. Data are presented as Ad genome copy numbers/ng cellular genomic DNA. All experiments were performed in triplicates; bars show mean values and error bars show standard deviations.

Ad5t/3sk.2xTyr showed reduced genome copy numbers relative to Ad5/3.2xTyr in SK-MEL-28 cells, but increased genome copy numbers in SKOV3.ip1 cells. These data mirror the results of the cytotoxicity experiments. However, for primary melanoma cells and Mel624, the differences in genome copy numbers between these viruses were minimal.


Fiber- or knob-pseudotyping, the swapping of the fiber protein or of the fiber knob domain between different adenovirus serotypes, has been exploited for the analysis of adenovirus cell binding, definition of adenovirus receptors, and to improve gene transfer vectors in gene therapy (reviewed in Barnett et al8). The latter strategy allowed for CAR-independent infection of different target cell types, such as hematopoietic cells or cancer cells. Recently, fiber chimerism has been applied to replication-competent adenoviruses resulting in enhanced virus infection, replication, and spread.17 Furthermore, the potency of adenoviruses with restricted replication capacity based on E1A mutations has been augmented by fiber chimerism or genetic incorporation of cell binding ligands into the fiber.18, 19 We extended this strategy by introduction of an advanced transcriptional targeting strategy into fiber-chimeric adenoviruses and showed a marked attenuation of the resulting viruses in nontarget cells. The combination of transductional and transcriptional targeting for the specific expression of therapeutic genes in the context of gene transfer has been reported before.20, 21 However, the transcriptional targeting of adenovirus replication demands promoter attributes and virus analyses distinct from those required for gene transfer vectors. This is due to the necessity to precisely regulate viral gene expression in order to achieve targeted adenovirus replication. In addition, the potential influence of virus replication, including viral transcription factors that are expressed during the adenoviral replication cycle, on promoter activity needs to be considered. Oncolytic adenoviruses with an integrin-binding RGD peptide incorporated into the Ad5 fiber and the E1A gene expressed from the cyclooxygenase-2 promoter have shown feasibility for virotherapy of pancreatic and ovarian cancer in recent studies.12, 22 For targeted melanoma treatment, however, we clearly show (i) that Ad5/3 fiber chimerism is superior to integrin-targeted RGD viruses with respect to both gene transfer efficacy and cell killing potency (Volk et al11 and Figure 1) and (ii) that targeted expression of both E1A and E4 improves cell killing specificity compared with single regulation of E1A.23 The strong attenuation of Ad2xTyr and Ad5/3.2xTyr in nonmelanoma cells can be explained by distinct blocks to virus replication implemented by lacking E1A or E4.23 Hence, targeted expression of both E1A and E4 within one virus synergizes to efficiently restrict virus replication to target cells.

The 2xTyr viruses contained the Δ24 mutation, a 24 bp deletion within the conserved region 2 (CR2) of the E1A gene, in addition to the transcriptional targeting mechanism. This mutant was designed to inhibit virus replication in cells with functional pRb repressor, for example in quiescent skin melanocytes, based on the function of the E1A-CR2 domain to bind and inactivate pRb as required for virus replication.24, 25 In contrast, pRb is frequently inactivated in melanoma26 and other tumor cells. Indeed, we reported previously that the Δ24 mutation does not attenuate virus replication and cell killing potency in melanoma cells or other cancer cells including SKOV3.ip1 and A549.10, 23 Thus, the specificity of the 2xTyr viruses as depicted in Figure 4 cannot be attributed to the Δ24 mutation but must be a consequence of the transcriptional targeting of E1A and E4 expression. Cultured normal cells, such as fibroblasts, keratinocytes, or melanocytes, proliferate and thus inactivate pRb during cell cycle progression. These attributes do not reflect the in vivo situation properly, where cells differentiate and are mostly quiescent. The Δ24 mutation-dependent attenuation of virus replication in normal cells is therefore difficult to demonstrate in monolayer cultures. To address this caveat, we currently develop a three-dimensional organotypic model for analysis of oncolytic adenoviruses in differentiating epithelium23 that might also allow for incorporation of fibroblasts and melanocytes.

The reason for the differences in the cell killing potency of Ad5t/3sk.2xTyr versus Ad5/3.2xTyr remains to be investigated. A different cell binding profile of the Ad5t/3sk capsid versus the Ad5/3 capsid might result in either a different and cell type-dependent transduction efficacy or in distinct signaling events that influence virus replication or tyrosinase promoter activity. In this regard, it has been reported that the length of the fiber shaft critically determines adenovirus transduction mediated by the knob domains of serotypes 5 and 9, but not by the serotype 35 fiber knob.27 Differences in cell binding might result from interference of Ad3 knob-binding to its receptor with the penton–integrin interaction or vice versa. Indeed, reduced integrin binding by short-shafted fiber adenoviruses has been described.14 Also, binding to different coreceptors along with distinct receptor expression profiles of different cell types might be responsible for the observed differences. After all, it will be interesting to determine systemic activity and toxicity of Ad5t/3sk.2xTyr and Ad5/3.2xTyr, and to evaluate the effect of fiber shaft length on virus biodistribution in adequate animal models.

Future studies need to address the efficacy, biodistribution, and toxicity of the fiber chimeric 2xTyr viruses in vivo in preparation of clinical applications of these agents. Such studies have previously been hampered by the observations that adenovirus replication is species specific and that rodent cells might not express the adenovirus type 3 receptor.11, 14 The recent description of CD46 as a receptor for B-type adenoviruses16, 28, 29 has initiated endeavors to better understand the molecular events of cell binding and entry by these viruses. However, Ad3 cell binding and infection is not yet fully understood. In this regard, the involvement of CD46 in Ad3 cell binding needs to be further explored.16, 29 Furthermore, both CD80 and CD86 have been reported as Ad3 receptors, recently.30 Clearly, a better understanding of cell binding by Ad3 will open new avenues for the evaluation of the biodistribution and toxicity of Ad5/3 chimeric oncolytic adenoviruses. In combination with the analysis of virus efficacy after systemic application to immunodeficient mice that bear human tumor xenografts, such strategies will also help to determine the potency of Ad5/3.2xTyr and Ad5t/3sk.2xTyr in vivo.

In conclusion, our results demonstrate that freshly purified melanoma cells, and thus probably melanoma cells in situ, can be resistant to cell killing by Ad5wt and by oncolytic adenoviruses with Ad5 capsid. These findings underline (i) the importance of primary tumor material for the analysis of oncolytic viruses and (ii) the requirement of tropism-modification of oncolytic adenoviruses for efficient virotherapy. Importantly, we established a strategy to overcome this caveat by combining fiber chimerism that implements efficient but nonselective cancer cell infectivity with stringent transcriptional targeting of adenoviral replication. The latter was achieved by expression of both E1A and E4 from the human or mouse tyrosinase enhancer/promoter, respectively.

We suggest the resulting virus Ad5/3.2xTyr as a candidate agent for clinical applications in melanoma therapy that features (i) efficient target cell infection resulting from genetic capsid modification, (ii) a high level of target cell specificity due to multiple transcriptional targeting, and (iii) the insertion of different sequence/same specificity profile promoters as a safety feature to avoid adverse recombination events.

Materials and methods

Cell culture

Human tumor cell lines SK-MEL-28 (melanoma, American Type Culture Collection (ATCC), Manassas, VA, USA), and A549 (lung adenocarcinoma, ATCC) were cultivated in DMEM (Mediatech, Herndon, VA, USA). Human melanoma cell lines Mel888, Mel624 (both kindly provided by Dr J Schlom, Bethesda, MD, USA) and A375M (kindly provided by Dr IJ Fidler, Houston, TX, USA) were cultivated in RPMI1640 (Mediatech). The human ovarian adenocarcinoma cell line SKOV3.ip1 (kindly provided by Dr J Price, Houston, TX, USA) and 293 cells (purchased from Microbix, Toronto, Canada) were grown in DMEM/F12 (50:50; Mediatech). Foreskin-derived primary NHF (kindly provided by L Rivera, Birmingham, AL, USA) were cultivated in EMEM (Mediatech). All media were supplemented with 10% fetal bovine serum (HyClone, Logan, UT, USA), 2 mM L-glutamine, 100 IU/ml penicillin, and 100 μg/ml streptomycin (all Mediatech). Foreskin-derived primary normal human fibroblasts (NHF, kindly provided by Dr F Noya, Birmingham, AL, USA) were grown in serum-free keratinocyte medium (Invitrogen, Carlsbad, CA, USA). Primary melanoma cells were obtained from surgically removed skin metastasis as approved by the local ethics committee. Tissue samples were dissected free of fat and epidermis and incubated in antibiotic/antimycotic solution (Life Technologies, Eggenstein, Germany) for 20 min at room temperature. Afterwards, the tissue was washed with PBS and cut into small pieces with sterile scalpels. These pieces were incubated overnight with 1 mg/ml collagenase/dispase (Roche, Mannheim, Germany) at 4°C. On the next day, the tissue was washed with PBS and incubated for 60 min at 37°C in trypsin/EDTA (Life Technologies). Tissue pieces were further minced by an inverted syringe and passed through a 40 μm cell strainer (Falcon, Heidelberg, Germany). Purified cells were analyzed for the expression of melanoma markers and for the absence of fibroblast contamination. Cells were then either stored frozen in FCS 10% DMSO or cultivated for further experiments in RPMI containing 10% FCS (BioWhittaker, Walkersville, MD, USA), 20 μg/ml gentamycin (Sigma, Deisenhofen, Germany), and 2 mM glutamine (BioWhittaker). Cells were grown at 37°C in a humidified atmosphere of 5% CO2.

Plasmids and recombinant adenoviruses

Generation of recombinant adenoviruses Ad5/3 and Ad5RGD is described elsewhere (Davydova et al, submitted; and Yamamoto et al12). Ad2xTyr is derived from Ad5 by replacing both the E1A promoter with the human tyrosinase enhancer/promoter and upstream polyA and the E4 promoter with the murine tyrosinase enhancer/promoter and upstream polyA.23 The fiber shuttle plasmid pNEB.PK.F5t.3sk was cloned by inserting sequences encoding the Ad3 fiber shaft and knob into pNEB.PK.SnaBI.31 Oligonucleotides Age/5T/3Sfor (5′-IndexTermGGA AAC CGG TCC TCC AAC TGT GCC TTT TCT TAC TCC TCC CTT TGT ATC CCC CAA TGG GTT TCA AGA GAG TCC CCC TGG GGT ATT AAG TCT TAA ATG TGT TAA TCC AC)and MfeAd5/3Krev (5′-IndexTermTCT GCA ATT GAA AAA TAA ACA CGT TGA AAC ATA ACA CAA ACG ATT CTT TAT TCT TGT TAG TCA TCT TCT CTA ATA TAG GAA AAG G) were used for PCR cloning with plasmid pBR.Ad3Fib (generously provided by J Chrobozcek, Grenoble, France) as template. Plasmids pNEB.PK.F5t/3sk or pNEB.PK.F5/332 were linearized by EagI and KpnI digestion and used for homologous recombination with SwaI linearized plasmid pVK500, generating pAdbackF5t/3sk, or pAdback5/3, respectively. Subsequent homologous recombination with shuttle plasmid pShTyrE1AD24mTyrE4 that contained the E1A and E4 genes downstream of the tyrosinase promoters23 resulted in plasmids pAd5t/3sk.2xTyr or pAd5/3.2xTyr, respectively, which contained the recombinant adenovirus genomes. Plasmids were validated for promoter insertion and fiber gene modification by PCR and restriction digest. Adenovirus particles were produced by transfection of PacI-digested pAd plasmids into SK-MEL-28 cells using Lipofectamine (Life Technologies, Rockville, MD, USA) following the manufacturer's protocol. Viruses were amplified in SK-MEL-28 cells and purified by two rounds of CsCl equilibrium density gradient ultracentrifugation. Replication-deficient, E1-deleted AdCMVLuc,20 and Ad5wt were amplified in 293 cells and purified similarly. Verification of viral genomes and exclusion of wild-type contamination was performed by PCR and restriction digest. Physical particle concentration (vp/ml) was determined by OD260 reading.

Cytotoxicity assay

For determination of virus-mediated cytotoxicity, 1.5 × 104 (cell lines, 24-well plate), 3 × 104 (primary cultures, 24-well plate), or 5 × 103 (primary cultures, 96-well plates) cells were seeded and infected with adenoviruses at indicated titers or were mock-infected. To visualize cell killing, cells were fixed and stained with 1% crystal violet in 70% ethanol for 20 min followed by washing with tap water to remove excess dye. The plates were dried and images were captured with a Kodak DC260 digital camera (Eastman Kodak, Rochester, NY, USA).

Flow cytometry analysis

Cells were detached from cell culture dishes with 0.02% EDTA. After washing twice with PBS/1% FBS, cells were incubated with antibodies diluted in PBS/1%FBS for 45 min at 4°C. The monoclonal CAR-binding antibody RmcB (Bergelson et al33, hybridoma cell line purchased from ATCC) or mouse IgG (Sigma) as a control were employed at a final concentration of 2 μg/ml and detected with Alexa488-labeled anti-mouse antibody (Molecular Probes, Eugene, OR, USA) diluted 1:200. Cells were washed twice with PBS, resuspended in PBS and analyzed by flow cytometry.

Quantification of E1A and E4 mRNA

For quantification of E1A and E4 mRNA expression, 1.5 × 105 cells were seeded per well in a six-well plate. The next day, cells were infected with the indicated viruses at 100 vp/cell, or mock infected. Cells were harvested at the indicated time points after infection and RNA was purified from the cell lysate with the RNeasy kit including DNase digest (Qiagen, Valencia, CA, USA) following the manufacturer's instructions. Quantification of mRNA copy numbers was performed by real-time PCR as previously described.34 For the assay, known amount of template DNA of pTG3602 (108, 106, 104, and 102 copies/μl) was amplified to generate a standard curve for quantification of the copy numbers of unknown samples. Human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as house keeping gene for internal control. Known amount of human total RNA (200, 20, 2, and 0.2 ng/μl) was amplified to generate a standard curve for determination of the RNA concentration of samples. Data were analyzed with LightCycler software and plotted as mRNA copy numbers per ng of RNA after standardization with mean values for Ad5wt. All experiments were performed in triplicates; error bars show standard deviations.

Western Blot

To determine adenoviral E4 expression, 1 × 105 Mel624 or SKOV3.ip1 cells were seeded in 24-well plates. Cells were infected with the indicated viruses at 5000 vp/cell for melanoma cells or 10 000 vp/cell for SKOV3.ip1 cells or were mock-infected. High virus titers were required for the detection of E4ORF3 protein. Cells were lysed in SDS sample buffer (containing 10 mM β-mercaptoethanol) 48 h after infection. Boiled samples were separated by SDS-polyacrylamide gel electrophoresis in 15% gels and transferred to a PVDF membrane (BioRad, Hercules, CA, USA). The membranes were probed with monoclonal antibody specific for E4ORF3 protein (kindly provided by Dr Gary Ketner, Baltimore, MD, USA). Bound anti-E4ORF3 antibody was detected with a secondary HRP-conjugated antibody (Sigma, St Louis, MO, USA) and enhanced chemiluminescence (NEN Life Science Products, Boston, MA, USA).

Quantification of viral genomes

To quantify intracellular adenoviral genomes, 1.5 × 104 cells were seeded in 24-well plates and infected at 1 vp/cell. Cells were harvested at 2 or 8 days postinfection and DNA was purified from the samples with the Qiagen DNA Blood kit according to the manufacturer's protocol (Qiagen, Valencia, CA, USA). Adenoviral genome copy numbers were quantified by real-time PCR as described previously.23 As an internal control, cellular genomic DNA was quantified using the PCR primers β-actin. Data were analyzed with LightCycler™ software and are presented as Ad genome copy numbers/ng cellular genomic DNA. All experiments were performed in triplicates; error bars show standard deviations.


  1. 1

    Alemany R, Balague C, Curiel DT . Replicative adenoviruses for cancer therapy. Nat Biotechnol 2000; 18: 723–727.

  2. 2

    Zhang WW . Development and application of adenoviral vectors for gene therapy of cancer. Cancer Gene Ther 1999; 6: 113–138.

  3. 3

    Bernt K et al. A new type of adenovirus vector that utilizes homologous recombination to achieve tumor-specific replication. J Virol 2002; 76: 10994–11002.

  4. 4

    Miller CR et al. Differential susceptibility of primary and established human glioma cells to adenovirus infection: targeting via the epidermal growth factor receptor achieves fiber receptor-independent gene transfer. Cancer Res 1998; 58: 5738–5748.

  5. 5

    Hemmi S et al. The presence of human coxsackievirus and adenovirus receptor is associated with efficient adenovirus-mediated transgene expression in human melanoma cell cultures. Hum Gene Ther 1998; 9: 2363–2373.

  6. 6

    Nettelbeck DM et al. Retargeting of adenoviral infection to melanoma: combining genetic ablation of native tropism with a recombinant bispecific single-chain diabody (scDb) adapter that binds to fiber knob and HMWMAA. Int J Cancer 2004; 108: 136–145.

  7. 7

    Reid T, Warren R, Kirn D . Intravascular adenoviral agents in cancer patients: lessons from clinical trials. Cancer Gene Ther 2002; 9: 979–986.

  8. 8

    Barnett BG, Crews CJ, Douglas JT . Targeted adenoviral vectors. Biochim Biophys Acta 2002; 1575: 1–14.

  9. 9

    Krasnykh VN, Douglas JT, van Beusechem VW . Genetic targeting of adenoviral vectors. Mol Ther 2000; 1: 391–405.

  10. 10

    Nettelbeck DM et al. Novel oncolytic adenoviruses targeted to melanoma: specific viral replication and cytolysis by expression of E1A mutants from the tyrosinase enhancer/promoter. Cancer Res 2002; 62: 4663–4670.

  11. 11

    Volk AL et al. Enhanced adenovirus infection of melanoma cells by fiber-modification: incorporation of RGD peptide or Ad5/3 chimerism. Cancer Biol Ther 2003; 2: 511–515.

  12. 12

    Yamamoto M et al. Infectivity enhanced, cyclooxygenase-2 promoter-based conditionally replicative adenovirus for pancreatic cancer. Gastroenterology 2003; 125: 1203–1218.

  13. 13

    Nettelbeck DM, Jerome V, Müller R . A dual specificity promoter system combining cell cycle-regulated and tissue-specific transcriptional control. Gene Therapy 1999; 6: 1276–1281.

  14. 14

    Vigne E et al. Genetic manipulations of adenovirus type 5 fiber resulting in liver tropism attenuation. Gene Therapy 2003; 10: 153–162.

  15. 15

    Bernt KM et al. The effect of sequestration by nontarget tissues on anti-tumor efficacy of systemically applied, conditionally replicating adenovirus vectors. Mol Ther 2003; 8: 746–755.

  16. 16

    Gaggar A, Shayakhmetov DM, Lieber A . CD46 is a cellular receptor for group B adenoviruses. Nat Med 2003; 9: 1408–1412; Epub 2003 Oct 1419.

  17. 17

    Kawakami Y et al. Substitution of the adenovirus serotype 5 knob with a serotype 3 knob enhances multiple steps in virus replication. Cancer Res 2003; 63: 1262–1269.

  18. 18

    Suzuki K et al. A conditionally replicative adenovirus with enhanced infectivity shows improved oncolytic potency. Clin Cancer Res 2001; 7: 120–126.

  19. 19

    Kanerva A et al. Enhanced therapeutic efficacy for ovarian cancer with a serotype 3 receptor-targeted oncolytic adenovirus. Mol Ther 2003; 8: 449–458.

  20. 20

    Reynolds PN et al. Combined transductional and transcriptional targeting improves the specificity of transgene expression in vivo. Nat Biotechnol 2001; 19: 838–842.

  21. 21

    Barnett BG, Tillman BW, Curiel DT, Douglas JT . Dual targeting of adenoviral vectors at the levels of transduction and transcription enhances the specificity of gene expression in cancer cells. Mol Ther 2002; 6: 377–385.

  22. 22

    Kanerva A et al. A cyclooxygenase-2 promoter-based conditionally replicating adenovirus with enhanced infectivity for treatment of ovarian adenocarcinoma. Gene Therapy 2004; 11: 552–559.

  23. 23

    Banerjee NS et al. Analyses of melanoma-targeted oncolytic adenoviruses with tyrosinase enhancer/promoter-driven E1A, E4, or both in submerged cells and organotypic cultures. Mol Cancer Ther 2004; 3: 437–449.

  24. 24

    Fueyo J et al. A mutant oncolytic adenovirus targeting the Rb pathway produces anti-glioma effect in vivo. Oncogene 2000; 19: 2–12.

  25. 25

    Heise C et al. An adenovirus E1A mutant that demonstrates potent and selective systemic anti-tumoral efficacy. Nat Med 2000; 6: 1134–1139.

  26. 26

    Piepkorn M . Melanoma genetics: an update with focus on the CDKN2A(p16)/ARF tumor suppressors. J Am Acad Dermatol 2000; 42: 705–722; quiz 723–726.

  27. 27

    Shayakhmetov DM, Lieber A . Dependence of adenovirus infectivity on length of the fiber shaft domain. J Virol 2000; 74: 10274–10286.

  28. 28

    Segerman A et al. Adenovirus type 11 uses CD46 as a cellular receptor. J Virol 2003; 77: 9183–9191.

  29. 29

    Sirena D et al. The human membrane cofactor CD46 is a receptor for species B adenovirus serotype 3. J Virol 2004; 78: 4454–4462.

  30. 30

    Short JJ et al. Adenovirus serotype 3 utilizes CD80 (B7.1) and CD86 (B7.2) as cellular attachment receptors. Virology 2004; 322: 349–359.

  31. 31

    Wu H et al. Double modification of adenovirus fiber with RGD and polylysine motifs improves coxsackievirus-adenovirus receptor-independent gene transfer efficiency. Hum Gene Ther 2002; 13: 1647–1653.

  32. 32

    Krasnykh VN, Mikheeva GV, Douglas JT, Curiel DT . Generation of recombinant adenovirus vectors with modified fibers for altering viral tropism. J Virol 1996; 70: 6839–6846.

  33. 33

    Bergelson JM et al. Isolation of a common receptor for coxsackie B viruses and adenoviruses 2 and 5. Science 1997; 275: 1320–1323.

  34. 34

    Rivera AA et al. Mode of transgene expression after fusion to early or late viral genes of a conditionally replicating adenovirus via an optimized internal ribosome entry site in vitro and in vivo. Virology 2004; 320: 121–134.

Download references


This work was supported by the Deutsche Forschungsgemeinschaft (grant NE832/1 to DMN, and Graduiertenkolleg 592 to SS), the Deutsche Krebshilfe (grant 10-2186-Ne 1 to DMN) and the National Cancer Institute (grants R01 CA83821, P50 CA83591, R01 CA93796, and R01 CA94084). We are grateful to Dr D Dieckmann, Dr IJ Fidler, Dr F Noya, Dr J Price, L Rivera, Dr J Schlom, and Dr T Strong for cell lines and primary cells, to Dr J Chrobozcek for providing plasmid pBR.Ad3Fib, Dr Ruben Hernandez-Alcoceba for plasmid pUC19-E4P-, Dr Gary Ketner for the E4 antibody, and to T Uil for plasmid pNEB.PK.SnaBI.

Author information

Correspondence to DM Nettelbeck.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Rivera, A., Davydova, J., Schierer, S. et al. Combining high selectivity of replication with fiber chimerism for effective adenoviral oncolysis of CAR-negative melanoma cells. Gene Ther 11, 1694–1702 (2004).

Download citation


  • tyrosinase enhancer/promoter
  • conditionally replicative adenovirus
  • fiber chimerism
  • viral oncolysis
  • melanoma

Further reading