Background
Adenovirus vectors are commonly utilized in cancer gene therapy experiments. They readily infect numerous tumor cell types and are easily manipulated allowing for transgene expression[
1]. In an effort to improve selectivity of these vectors for malignant cells, replication selective or conditionally replicating adenoviruses were created[
2]. With greater understanding in the molecular aspects of adenovirus replication, these viruses were designed such that replication was predicated on alterations in cell cycle regulation, thus rendering only malignant cells susceptible. These adenovirus systems rely on virus replication as a means of exerting tumoricidal effect. One of the first of these conditionally replicating adenoviruses was dl1520 or ONYX-015[
3]. It has been used extensively in cancer gene therapy clinical trials[
4,
5]. This virus is deleted for the E1B-55 kDA gene[
6].
The E1 region genes of the common serotypes of adenovirus optimize their own replication in target cells and interact with many cell cycle associated gene products. The E1a region genes promote transition of cells into S phase through their interaction with the retinoblastoma protein (pRB)[
7]. pRB in conjunction with p53 can inhibit transition of cells into the S-phase of the cell cycle. Binding of E1a to pRB prevents association of pRB to the E2F transcription factor resulting in activation of E2F[
8]. E2F then functions to transition cells into the S-phase of the cell cycle which is conducive for optimal adenovirus replication. In wild type adenoviruses, the role of the E1B-55 kDa gene product is to neutralize p53[
9,
10]. p53 induction is thought to promote cell cycle arrest resulting in termination of the virus replicative lifecycle in that cell[
11]. Thus it was hypothesized that deletion of the E1B-55 kDa gene product would result in p53 induced termination of virus replication in normal cells while being replication permissive in malignant cells with abnormal p53 expression or regulation[
9]. Early reports with ONYX-015 indicated replication selectivity for tumor cells with abnormal p53 functioning[
12]. However, the exact mechanism of replication selectivity has been brought to question with subsequent reports of ONYX-015 replication independent of p53[
13] and researchers have looked at other factors associated with adenovirus replication.
Various heat shock proteins have been shown to be necessary for efficient adenovirus replication. The avian adenovirus CELO requires the induction of HSP70 and HSP40 for production of viral proteins[
14]. CELO mutants lacking the E1 region genes were replication incompetent in A549 cells. However, heat shock protein induction or expression allowed for permissive replication of these mutants. Similarly in a number of human cell lines, both heat shock and HSP72 expression enhanced the oncolytic effect of E1 containing adenovirus but not for E1 deleted adenovirus[
15].
Adenovirus E1 region genes from serotype 5 and 12 have previously been shown to induce HSP72 expression [
16]. HSP72 is a molecular chaperone protein involved in the repairing denatured proteins. Through this role, it can inhibit apoptosis downstream of caspase activation but prior to loss of mitochondrial membrane potential[
17]. Stably transfected or E1 transformed cell lines show constitutive expression of the inducible HSP72 [
18]. HSP72 levels are correlated with cell-cycle with increases in HSP72 during S-phase with a maximum level in the post-S-phase period [
19]. In 293 cells which are transformed with the adenovirus E1 region gene, E1A mRNA accumulated just prior to HSP72 expression suggesting that E1A is responsible for the cell cycle regulation of HSP72 expression [
18]. Additionally, HSP72 co-localizes with E1A gene products in infected cells [
20]. HSP72 is predominately cytoplasmic but translocates to the nucleus following adenovirus infection. Double labelling experiments with E1A and HSP72 in infected cells showed co-localization of these molecules within the nucleoli and exhibited similar reticular and punctuate nuclear staining patterns depending on cell cycle. HSP72 transport to the nucleus was E1A and virus infection dependent. In 293 (E1 region transformed) cells, HSP72 and E1A co-localization was not observed until after infection with virus [
20]. These data suggest a physical complex is necessary between adenovirus E1A, HSP72 and other adenovirus gene products
One of the difficulties with studying E1 adenovirus deletion mutants is the lack of appropriate animal tumor models which allow for permissive growth of these viruses. Common animal tumor models for glioblastoma are the cell lines 9L and RT2. Previously, it has been shown that replication competent E1 deletion mutants can transduce 9L cells and result in transgene expression but there is a block against virus replication[
21]. In this study we addressed if HSP72 expression could allow for permissive replication of the ONYX-015 adenovirus in these rat glioblastoma cells.
Discussion
ONYX-015 is a promising new cancer gene therapy vector which exerts its tumoricidal effect by selective replication in cancer cells[
6]. Replication selectivity previously was attributed to interactions of E1 gene products with p53[
3]. Subsequent reports have shown that ONYX-015 replicates in cells with wild-type p53 as efficiently as in cells with mutant-p53[
13]. Although the mechanism of cancer cell selectivity is still under investigation, the various hypotheses consistently involve cell-cycle regulatory proteins and their interactions within apoptosis pathways. HSP72 is a potent regulator of apoptosis [
23]. Also, there is evidence that activation of the heat shock response by adenovirus early region genes is necessary for virus replication[
14]. Previously it has been shown that heat shock induced HSP72 expression resulted in increased production of adenovirus proteins[
24]. Additionally, the E1A adenovirus gene products necessary for virus genome replication have previously been shown to co-localize with HSP72[
20].
To study the role of HSP72 in augmenting ONYX-015 replication, we chose two glioma cell lines with different derivations. 9L is a chemically induced gliosarcoma while RT2 is a virally transformed glioblastoma. Compared to many human tumor cell lines, these cell lines show relative resistance to the cytotoxic affects of ONYX-015. Resistance in part was secondary to low adenovirus transduction efficiency. At the multiplicities of infection utilized in this study, we never achieved 100% cell transduction (table
1). In most cases, when less than 30% of permissive HSP72-transfected cells were transduced, tumor cell growth was rapid and outpaced ONYX-015 virus replication resulting in no discernable cytotoxic or cytostatic effect after 1 week in culture (data not presented). Therefore, replicative adenovirus cytotoxicity is likely a balance between virus transduction efficiency, replication efficiency and cell growth characteristics. HSP72 transfection had no bearing on cell doubling time of either 9L or RT2 cells (data not presented) RT2 cell-doubling time is approximately 8 hours while 9L cell-doubling time is approximately 14 hours. Both cell lines after HSP72 transfection showed equal susceptibility to ONYX-015. Further evidence for the balance between transduction efficiency, replication efficiency and cell doubling time is evident in that ONYX-015 infection resulted in a predominantly cytostatic affect resulting in decreased cell division before evidence of cytotoxicity.
The most dramatic affect of the role HSP72 in promotion of virus replication is evident in figure
1. At 96 hours, the GFP transduced cells showed very little cytotoxic effect of ONYX-015 infection. In contrast, there is a distinct cytopathic effect at the same doses of ONYX-015 infection in the HSP72 transduced cells. These cells were rendered susceptible to ONYX-015 toxicity. The mechanism of cell death is related to increased virus replication. Significantly more ONYX-015 was isolated from HSP72 transduced cells than the GFP-transfected controls.
The E1a gene products of adenovirus are responsible for activation of the HSP72 promoter and may affect HSP72 levels during the cell cycle[
18]. Conversely, there is no evidence that HSP72 interacts with adenovirus promoters to stimulate viral transcription. There is evidence of HSP72 interactions with adenovirus structural proteins such as hexon[
25] and fiber[
26]. Adenovirus assembly is known to be inefficient with only a small percentage of total structural proteins eventually utilized in the production of infectious virus [
26].
In conclusion, our studies demonstrated that HSP72 expression in rodent glioma tumor cells potentiated ONYX-015 replication and oncolysis of tumor cells. Further studies on the role of HSP72 in tumor types with wild-type and mutant p53 tumor suppressor genes is warranted to understand the molecular interactions of this chaperone protein in promoting virus replication. Additionally, addressing the role of HSP72 as an inhibitor of apoptosis in virus replication would further help characterize the interaction of cell-cycle regulatory pathways with virus replication. Furthermore, we intend to evaluate the role of HSP72 associated permissive replication in other animal cell lines to establish animal models for the study of ONYX-015 and similar replication competent adenoviruses.
Conclusion
For clinical applicability, these findings have a number of implications. Currently, clinical trials are ongoing with ONYX-015. Due to the previous association of ONYX-015 replication with p53 status, the current tumor types chosen for clinical trials are those exhibiting abnormal p53 expression in a majority of cells. Our study indicates, that tumor types with high HSP72 may also be viable candidates to evaluate ONYX-015 efficacy, as they are likely to demonstrate enhanced susceptibility to this replication competent virus. As HSP72 expression in tumors is associated with a more aggressive tumor phenotype, ONYX-015 may be ideal as adjunctive therapy for advanced disease. Additionally, one of the greatest limitations to adenovirus gene therapy is patient safety at high doses of adenovirus administration. Since, HSP72 expression resulted in oncolysis at much lower MOIs, a lower dose of virus administered to a HSP72 positive tumor may have the same benefit as a higher dose, allowing one to use lower doses to achieve the same effect. In addition, for some tumor types such as breast cancer, hyperthermia has been shown to significantly improve response rates[
27]. HSP72 is readily induced with hyperthermia. Therefore, HSP72 induction by hyperthermia may also serve as an effective strategy to augment ONYX-015 oncolytic activity. We are currently in the process of studying this hypothesis. Lastly, the role of HSP72 in cell cycle regulation suggests alternate pathways for the mechanism of E1b deleted adenovirus replication in tumorigenic cells. HSP72 overexpression in non-transformed fibroblasts did not result in ONYX-015 oncolysis (data not presented). Understanding the molecular interactions of adenovirus proteins with molecular chaperones would help determine cell selectivity to replication competent adenoviruses influencing the design of more potent viruses.
Authors' contributions
JM performed all of the experiments outlined in this study. JAK independently replicated results for validity. JAK and MRS contributed equally to the molecular cloning of the plasmids and adenovirus propagation and purification. All authors participated in maintenance of cells in culture. The hypothesis, study design and statistical analysis were all performed by MRS. MRS drafted the entire manuscript with editing and approval by all of the authors.