The application of cancer gene therapy has heretofore been restricted to local, or locoregional, neoplastic disease contexts. This is owing to the lack of gene transfer vectors, which embody the requisite target cell selectivity in vivo required for metastatic disease applications. To this end, we have explored novel vector engineering paradigms to adapt adenovirus for this purpose. Our novel strategy exploits three distinct targeting modalities that operate in functional synergy. Transcriptional targeting is achieved via the hROBO4 promoter, which restricts transgene expression to proliferative vascular endothelium. Viral binding is modified by incorporation of an RGD4C peptide in the HI loop of the fiber knob for recognition of cellular integrins. Liver sequestration is mitigated by ablation of factor X binding to the major capsid protein hexon by a serotype swap approach. The combination of these technologies into the context of a single-vector agent represents a highly original approach. Studies in a murine model of disseminated cancer validated the in vivo target cell selectivity of our vector agent. Of note, clear gains in therapeutic index accrued these vector modifications. Whereas there is universal recognition of the value of vector targeting, very few reports have validated its direct utility in the context of cancer gene therapy. In this regard, our article validates the direct gains that may accrue these methods in the stringent delivery context of disseminated neoplastic disease. Efforts to improve vector targeting thus represent a critical direction to fully realize the promise of cancer gene therapy.
A wide range of strategies have been developed to apply gene therapy to the context of neoplastic disease (1–4). The central goal embodied in these molecular interventions is the achievement of an improved therapeutic index compared with conventional cancer therapies. Heretofore, such in vivo cancer gene therapies have been applied for local, or locoregional, neoplastic disease. This is owing to the fact that currently available gene transfer vectors lack the in vivo target cell selectivity mandated for the clinical context of disseminated disease.
In this regard, there has been a field-wide recognition of the need for gene transfer vectors, which embody the capacity for target cell selectivity (5). Indeed, an NIH report on gene therapy highlighted this goal as the highest mandate for the field. For cancer, such a vector-targeting capacity thus represents the sine qua non for practical advancement of these promising strategies to the problematic clinical setting of metastatic disease. Given this consideration, the paucity of reports of the successful application of vector targeting for cancer gene therapy is noteworthy.
To this end, we have endeavored modification of adenoviral vectors (Ad) to address this key gene delivery mandate. On the basis of the unique molecular promiscuity of the parent virus, we hypothesized that targeting might be achieved by exploiting multiple biologic axes. Furthermore, we sought to combine such distinct targeting strategies to realize functional synergy vis-à-vis the achievement of target cell selectivity. On this basis, the requirement for in vivo selectivity, in the context of disseminated neoplastic disease, might be achieved.
Materials and Methods
The replication incompetent E1-deleted Ad5 vectors used for the study were prepared using a two-plasmid cloning method. Untargeted or triple-targeted Ad5 encoding the GFP reporter gene or the HSVtk therapeutic gene were produced in accordance with the standard techniques (6). Briefly, adenoviral genome–including plasmids were digested with PacI for releasing the recombinant viral genomes, and transfected into HEK293 cells. Rescued viruses were serially amplified, and then purified by centrifugation on CsCl gradients according to the standard protocols. For in vitro and in vivo study, viruses were dialyzed against PBS containing 10% glycerol, and stored at −80°C. The titers of physical viral particles were determined by methods described by Maizel and colleagues (7).
In vitro validation of adenovirus
HUVEC (human primary endothelial), bEnd-3 (mouse primary endothelial), and NIH/3T3 (mouse embryonic fibroblast) cells were obtained from the ATCC and maintained for assays according to the manufacturer's instructions. Total expression levels of HSVtk proteins in whole-cell lysate were determined with anti-tk antibody (kindly provided by Dr. William Summers, Yale University, New Haven, CT) by Western blot analysis in accordance with the standard protocols. For HSVtk/ganciclovir-killing activity assay, cells were infected with virus encoding either GFP or HSVtk gene, ganciclovir (Selleckchem) prodrug was administered to cells via serial diluted drug concentration. To measure the cellular ATP contents (Promega) as a marker for cell viability, assay plates were read in a microplate luminometer (Berthold detection system) and cell viability was analyzed. Dose–response analysis curves were plotted by Graph Pad Prism v7.0c software.
Murine xenograft models
Triple immunodeficient NOD/SCID/IL2Rγ (NSG) mice were injected subcutaneously with 1 × 106 786-0 renal carcinoma cells (mCherry expressed cell line). Two weeks later, the mice were intravenously injected with 1 × 1011 viral particles of untargeted Ad5.CMV.GFP or triple-targeted Ad-GFP viruses. To perform histopathologic analysis in tumor or organs, mice were sacrificed under anesthesia (Avertin, Sigma-Aldrich) at 3 days post-virus injection. The tumor-bearing tissues were harvested, followed by postfixed in 4% paraformaldehyde for 2 hours at room temperature, cryopreserved in 30% sucrose for 16 hours at 4°C, and cryo-embedded in NEG50 (Thermo Fisher Scientific) over 2-methylbutane/liquid nitrogen. Histopathologic images were captured by epifluorescence microscopy. To perform histopathologic analysis in SPC6 subcutaneous tumor, all assay procedures were conducted in accordance with the same protocols.
Tissue harvest and immunofluorescence staining
Mice were administrated with 1 × 1011 viral particles of triple-targeted adenoviral vectors via tail vein injection. Three days post-virus infection, mice were anesthetized and tissues (liver, lung, pancreas, spleen, kidney, small bowel, heart, muscle, and brain) harvested for immunofluorescence staining. For frozen sections, organ slices were cryo-preserved in 30% sucrose in PBS at 4°C overnight, embedded in NEG50 Mounting Medium (Thermo Fisher Scientific), and then frozen in a liquid nitrogen prechilled 2-methylbutane containing bucket. Sectioning of frozen organs was carried out using the CryoJane Taping System (Leica Biosystems Inc). All frozen section slides were subject to immunofluorescence staining analysis in accordance with the standard techniques (8). Primary antibodies used in this study included hamster anti-CD31 (EMD Millipore), rat anti-endomucin (eBioscience), rat anti-PDGFRβ (eBioscience), rabbit anti-HSVtk (Dr. Summers's laboratory), and Alexa Fluor 488 or 594-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories).
Animal studies for therapeutic index gains
Triple-immunodeficient NSG mice allow for modeling of all metastatic sites as published previously (8, 9). For systemic metastases, mice were injected intracardially with 5 × 105 786-0 renal carcinoma cells (Luciferase gene expressed). Tumor growth was monitored and quantified using bioluminescence imaging (BLI) analysis. Three weeks after tumor implantation, mice were intravenously injected via tail vein with 1 × 1011 viral particles of each virus. During the following 7 days, mice received daily intraperitoneal injection of ganciclovir (50 mg/kg/7 days) according to the different treatment groups [(n = 8 mice/group): untargeted Ad5-GFP, untargeted Ad5-HSVtk, and triple-targeted Ad-HSVtk plus ganciclovir]. Mice were weighed daily to gauge health and survival during the assays. For study of tumor progression, final ROI values were monitored and compared (increased fold = last ROl value/initial ROl value) in the survival mice. Survival data were plotted on a Kaplan–Meier curve in all animal groups by Graph Pad Prism v7.0c software. The studied was terminated at 8 weeks (56 days posttumor implantation) according to the approved protocols and pursuant to NIH guidelines for the care and use of laboratory animals’ standards. All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee of the Washington University in St. Louis, School of Medicine (protocol #.20160292, St. Louis, MO).
Generated triple-targeting vector for gene therapy
As the main limit to the systemic employment of adenoviral vectors is liver sequestration (10), we first sought to address this issue. Our studies, and those of others, had linked this phenomenon to vector capsid association with serum factor X (11, 12). On this basis, we employed a strategy of capsid chimerism whereby the major capsid protein hexon of the serotype 5 vector base was substituted with a cognate from the alternate human adenovirus serotype 3. We had previously shown that this modification substantially mitigated sequestration in the reticuloendothelial system in the context of systemic vector administration (12, 13). This liver “untargeting” method was then combined with modifications designed to direct the expression of delivered transgenes to tumor vascular endothelium. Specific vector particle binding was enhanced for the target cells by the incorporation into the adenoviral vectors capsid fiber knob of the RGD4C peptide (14, 15). This ligand exhibits preferentiality for integrins of αvβ3 and αVβ3 class overexpressed in angiogenic endothelium. This approach was also combined with a transcriptional targeting strategy to restrict transgene expression target cells. In this case, the promoter of the roundabout guidance receptor 4 (ROBO4) gene, which is selectively inductive for proliferative endothelium, was employed. We hypothesized that the combination of the liver untargeting and vector-targeting methods (Fig. 1) would provide functional synergy with respect to the achievement of the in vivo selectivity mandated for disseminated neoplastic disease cancer gene therapy.
Evaluated vector for cancer gene therapy
Such a triple-targeted adenoviral vectors was thus constructed to achieve expression of the HSVtk therapeutic gene selectively within proliferative tumor endothelium in vitro. This gene has been applied historically for molecular chemotherapy cancer gene therapy approaches (16, 17). Its employment here thus served to reconcile our findings with the extensive literature relating to the use of HSVtk for cancer gene therapy. After rescue and upscale, the virion was subjected to genomic analysis (Fig. 2A). This study confirmed that the control untargeted adenovirus, and the triple-targeted adenovirus, both contained the cassette with the HSVtk therapeutic gene. Infection of human endothelial cells (HUVEC), with follow-on Western blot analysis, confirmed expression of HSVtk in cells infected via both vectors (Fig. 2B). Next, analysis of the specificity of expression was carried out utilizing an in vitro assay whereby HSVtk expression sensitized cells to the cytotoxic effects of ganciclovir. Target cells were human endothelial (HUVEC) or murine endothelial cells (bEnd-3). In this analysis, it could be seen that the target cells were sensitive to vector-mediated cytotoxicity exclusively after infection with the triple-targeted adenoviral vectors encoding HSVtk (Fig. 2C), under ganciclovir treatment. In addition, regarding to the transduction efficacy, we confirmed with NIH/3TC cells (mouse embryonic fibroblast) including demonstration of the target cells specificity (Fig. 2D). These studies thus validated the endothelial selectivity of our vector targeting schema.
Advanced vector targeting to tumor in vivo
We next sought to validate the in vivo selectivity of our vector methods in a systemic delivery context. For this analysis, we derived versions of the control and triple-targeted adenovirus, which encoded the GFP reporter gene. Immunodeficient NSG mice were xenografted with subcutaneous nodules of the human renal carcinoma cell line 786-0 (8). Animals were then treated with 1 × 1011 particles of the control and targeted adenoviral vectors with analysis of harvested tumor and organs at 3 days later. As the major site for ectopic adenoviral vectors localization is the liver, analysis was limited to this site and target tumor. In these studies it could be seen that untargeted adenoviral vectors was largely sequestered in the liver, with no detectable reporter noted within vascular areas of the tumor (Fig. 3A). In marked contrast, the triple-targeted adenoviral vectors avoided ectopic localization within the liver and spleen. Of note, high levels of reporter gene were seen within the harvested subcutaneous tumor nodules via secondary staining analysis that confirmed the identity of gene modified cells as tumor endothelium (Fig 3B). Our findings confirm that the hexon modification of our triple-targeted adenoviral vectors achieves the desired goal of liver untargeting, without the accrual of major new sites of ectopic gene delivery (Fig. 3C). Furthermore, as an even more stringent model (xenografted with subcutaneous nodules of the Syrian golden hamster pancreatic carcinoma cell line, SHPC6), the targeting specificity of our triple-targeted adenovirus in the context of other tumors could be seen (Fig. 3D). Thus, our triple-targeting strategy achieves highly specific in vivo selectivity.
Improved vector to achieve the therapeutic index gains in vivo
Finally, we endeavored evaluation of the therapeutic index gains, which accrue synergistic targeting. The therapy experiment with targeted adenovirus (vs. nontargeted adenovirus) was employed in a murine model of metastatic disease. The triple-targeting specificity mitigated the vector-associated toxicity known to be associated with the systemic application of untargeted adenoviral vectors encoding HSVtk (ref. 18; Fig. 4A). In parallel with that, of note, it could be seen that extended survival time accrued to the treatment group that received only the triple-targeted adenoviral vectors encoding HSVtk with significant differences (**, P < 0.005; Fig. 4C). Our demonstration of antitumor therapy was accomplished in a context more stringent than that required only for local control; however, it could be seen as the potential for improvement of the therapeutic effect (Fig. 4B). Also, it could be seen the persistence of the vector in the metastatic tumor as well. These findings demonstrated directly the improved therapeutic index, which accrued the employment of vector targeting for this established cancer gene therapy approach.
Our study highlights the direct benefits that may derive from the application of vector-targeting methods to the context of cancer gene therapy. In this regard, in vivo cancer gene therapies have been restricted to local diseases contexts owing to the limits of currently available gene transfer vectors. Thus, the problematic clinical context of metastatic cancer has not been addressable to this point via gene therapy methods. Further in this regard, the universal recognition of the potential value of vector targeting has not been translated to the context of cancer gene therapy studies demonstrating therapeutic gains. The advent of our novel vector-targeting technology now provides adequate selectivity in vivo to test this hypothesis of field-wide significance. Our early finding reported here thus provide the rationale for future studies designed to apply cancer gene therapy to this problematic clinical context most warranties novel interventions.
Disclosure of Potential Conflicts of Interest
D.T. Curiel is a paid consultant for Samyang Biopharmaceuticals and is an advisory board member (unpaid consultant) for Precision Virologics, Inc., Unleash Immuno Oncolytics, Inc., DNAtri, Inc., and Altimmune, Inc. No potential conflicts of interest were disclosed by the other authors.
Conception and design: M. Lee, Z.H. Lu, I.P. Dmitriev, D.T. Curiel
Development of methodology: Z.H. Lu, E.A. Kashentseva, I.P. Dmitriev, S.A. Mendonca
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Lee, Z.H. Lu, E.A. Kashentseva
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Lee, J. Li
Writing, review, and/or revision of the manuscript: M. Lee, Z.H. Lu, I.P. Dmitriev, D.T. Curiel
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Lee, E.A. Kashentseva, I.P. Dmitriev
Study supervision: Z.H. Lu, I.P. Dmitriev
We thank Dr. Arbeit for providing critical precedent research. We also appreciate Dr. Summers's laboratory for providing HSVtk antibody essential to this study. We also thank Dr. Toth's laboratory for providing SHPC6 (Syrian Hamster Pancreatic Carcinoma) cells essential to this study. We especially thank Amanda Baker Wilmsmeyer for extensive advice on the article and the scientific graphics. This study was funded by the NIH (RO1's CA211096, to principle investigator D.T. Curiel) research grants.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.