Despite good safety data in clinical trials, oncolytic adenoviruses have not been efficient enough to make them a viable treatment alternative for cancers. As more potent viruses are being made, transcriptional and transductional targeting to tumor tissues becomes increasingly appealing. To improve antitumor efficacy, oncolytic adenoviruses can be armed with therapeutic transgenes, such as the antiangiogenic soluble vascular endothelial growth factor receptor 1-Ig fusion protein. We hypothesized that an infectivity enhanced, targeted, vascular endothelial growth factor receptor 1-Ig armed oncolytic adenovirus would exhibit improved specificity and antitumor effect in murine kidney cancer models. Two hypoxia inducible factor-sensitive promoters were evaluated for renal cancer specificity using a novel in vivo dual luciferase-imaging system. Earlier data had shown usefulness of the 5/3-serotype chimera capsid modification for kidney cancer. Therefore, we constructed Ad5/3-9HIF-Δ24-VEGFR-1-Ig, which showed good specificity and oncolytic effect on renal cancer cells in vitro and resulted in antitumor efficacy in a subcutaneous in vivo model, in which vascular endothelial growth factor receptor 1-Ig expression and a concurrent antiangiogenic effect were confirmed. In an intraperitoneally disseminated kidney cancer model, significantly enhanced survival was observed when compared with control viruses. These results suggest that a targeted, antiangiogenic, oncolytic adenovirus might be a valuable agent for testing in kidney cancer patients.
Renal cell cancer is diagnosed in >200 000 patients each year.1 Prognosis is poor, especially when metastases are present, as found in one-third of diagnosed patients, who have a 5-year survival of 5–8%.2 Unfortunately, radiotherapy, chemotherapy and hormone therapy have little effect in this disease, whereas immunomodulating agents, such as interferons or cytokines, are active in a small proportion of patients.3 Antiangiogenic molecules including bevacizumab, sorafenib and sunitinib have increased time to progress in patients with metastatic disease, but a survival advantage has not yet been shown.4 Furthermore, these treatments are usually not curative. Thus, novel therapies are needed.
Oncolytic adenoviruses are a promising approach for treatment of various cancers and even early generation agents have shown safety and efficacy in non-randomized and randomized clinical trials.5, 6, 7, 8 However, convincing anticancer effects have only been seen when oncolytic adenoviruses were used in combination with conventional chemotherapy.9, 10 Because of good safety on the one hand and mostly unimpressive single agent efficacy on the other, recent progress has been toward more potent agents. Liver tropism of adenoviruses continues to be concerning and it remains key to restrict viral replication to target tissues.11 Thus, efficacy and specificity have to be improved to make oncolytic adenoviruses a viable treatment option for cancer.
The antitumor effect of oncolytic adenoviruses is achieved through the infection of cancer cells, followed by replication and ultimately cell death, releasing virions for reinfection of surrounding tissue. Replication in non-target normal tissues might, therefore, lead to side effects. To restrict replication to cancer cells, deletions in early genes, such as a 24-bp deletion in constant region 2 of E1A, can be used.12, 13 Consequently, the E1A protein will not be able to bind retinoblastoma for induction of an S-phase-like state, which is required for replication. Hence, replication will be attenuated in normal cells as opposed to cancer cells, in which the p16/retinoblastoma pathway is universally deficient,14 and thus, replication is unhampered.
For further restriction of replication, tumor-specific promoters can be used in conjunction with E1 mutations.15 With regard to renal cell cancer, hypoxia response elements (HREs) might be especially useful16 for transcriptional targeting. Hypoxia inducible factor (HIF), which activates HREs, is upregulated under hypoxic conditions, as found in most tumors.17 Moreover, most kidney cancer cells have a defective von-Hippel–Lindau pathway, which leads to permanently high levels of HIF independent from oxygen supply.18 Therefore, HRE-controlled genes are strongly expressed in kidney cancer tissue.
Various adenoviral capsid modifications have been developed to increase gene delivery to cancer cells.19, 20, 21 With regard to kidney cancer, we have shown earlier that adenoviruses featuring 5/3 serotype chimerism display significantly enhanced gene delivery and antitumor effect compared with viruses with wild-type capsids.22, 23
Renal cell cancer is characterized by high vascularization, due to strong angiogenic activity,24 mediated in part by von-Hippel–Lindau/HIF pathway defects.18 A major player in tumor angiogenesis is vascular endothelial growth factor (VEGF), which is highly expressed in renal cell cancers.25 VEGF binds to fms-like-tyrosine-kinase receptor (flt-1 or VEGFR-1) and kinase domain region receptor or VEGFR-2 with high affinity.26 The soluble VEGF receptor 1-Ig (VEGFR-1-Ig) fusion protein used in this study, as well as the naturally occurring soluble VEGFR-1 (sFlt), also bind VEGF, but do not induce vascular endothelial cells mitogenesis.27, 28 Blocking VEGF with sFlt or other suitable molecules has been shown to inhibit tumor growth29, 30, 31 in various cancer models.
Bioluminescence imaging has become a useful tool in cancer and gene therapy research. It can be used for determination of transduction efficiency, evaluation of promoter activity, in vivo biodistribution and various other applications. Besides firefly luciferase, which is the most commonly used bioluminescence marker gene, there are renilla, click beetle red and click beetle green luciferases, all having different emission spectra. As bioluminescence-imaging systems can be equipped with emission filters, it might theoretically be possible to separate signals from two different luciferases. Thus, a virus expressing luciferase and a cell line expressing a different luciferase could be specifically imaged simultaneously. Here, we sought to develop a dual luciferase-imaging system based on viruses coding for firefly luciferase and a cell line stably transfected with click beetle green luciferase to localize virus gene expression in an in vivo kidney cancer model.
We hypothesized that a transcriptionally and transcomplementationally targeted oncolytic adenovirus, which features a capsid modification and is armed with an antiangiogenic transgene, would exhibit enhanced specificity and increased antitumor effect with regard to renal cell cancer. To this end, we evaluated two different HREs for their specificity for renal cell cancer using a novel dual luciferase-imaging technique. On the basis of these results, and earlier studies on adenoviruses with capsid modification for kidney cancer,22, 23 we created Ad5/3-9HIF-Δ24-VEGFR-1-Ig. This oncolytic adenovirus features a 5/3-serotype chimeric capsid, an HRE (9HIF) driving E1, a 24-bp deletion in E1A and VEGFR-1-Ig in E3. Ad5/3-9HIF-Δ24-VEGFR-1-Ig was evaluated in vitro and in vivo for its specificity and efficacy compared with Ad5/3-9HIF-Δ24-E3, an isogenic, non-armed control virus, and other non-targeted and non-armed viruses.
Hypoxia-responsive promoters in vitro
Replication deficient adenoviruses with two different HREs controlling firefly luciferase (Ad5-9HIF-luc and Ad5-OB36-luc) were constructed and their luciferase activity was compared with Ad5luc1 with a highly active cytomegalovirus (CMV) promoter driving luciferase. Under normoxic conditions in the human renal cancer cell line 786-O, Ad5-9HIF-luc showed up to 9% expression compared with Ad5luc1 (Figure 1a), whereas in ACHN and SN12L1 cells, Ad5-OB36-luc had stronger luciferase activity with 1.2 and 1.7% compared with Ad5luc1 (Figures 1b and c). In human fibroblast, which is not expected to express HIF under normoxic conditions, no expression was seen with Ad5-9HIF-luc, whereas Ad5-OB36-luc showed about 6% luciferase activity compared with Ad5luc1 (Figure 1d). Luciferase expression in other renal cancer cell lines under normoxic conditions was similar to Ad5-OB36-luc, usually giving higher activity (Supplementary Figures 1a–c). In Caki-2 cells, Ad5-OB36-luc expressed more luciferase than Ad5luc1 and Ad5-9HIF-luc (Supplementary Figure 1d).
Under hypoxic conditions, Ad5-9HIF-luc showed about 45% luciferase activity compared with Ad5luc1 in 786-O cells (Figure 1e). This corresponds to a 5.5-fold increase compared with normoxic conditions for Ad5-9HIF-luc, whereas the activity of Ad5-OB36-luc only increased 1.9 fold (Supplementary Figure 2a). Also, in ACHN and SN12L1 cells, the activity of Ad5-9HIF-luc markedly increased to 5 and 3.5% compared with Ad5luc1 (25- and 16-fold increase compared with normoxia), whereas the activity of Ad5-OB36-luc went up to 2.5 and 5% (3- and 4-fold increase compared with normoxia) (Figures 1f and g; Supplementary Figures 2b and c). Ad5-OB36-luc displayed 24% luciferase activity (4.5-fold increase compared with normoxia) in fibroblasts.
Hypoxia-responsive promoters in vivo
Ad5-9HIF-luc (Figure 2a, right side tumors) and Ad5-OB36-luc (Figure 2b, right side tumors) showed high luciferase expression after intratumoral injection into subcutaneous 786-O tumors, in comparison with Ad5luc1 (Figures 2a and b, left side tumors). In contrast with in vitro data, quantification of light emission from live animals showed significantly higher luciferase expression from Ad5-9HIF-luc compared with Ad5luc1-injected tumors (Figure 2c, P=0.02). Ad5-OB36-luc-injected tumors also showed higher relative light emission in vivo than in vitro (when compared with Ad5luc1). Tumors were excised and luciferase activity was analyzed ex vivo. Ad5-9HIF-luc-injected tumors had about 10-fold higher luciferase activity compared with Ad5luc1-injected tumors (Figure 2d, P=0.01). As in vitro luciferase expression from Ad5luc1 was >10-fold higher than Ad5-9HIF-luc with this cell line, there was a >100-fold induction of the 9HIF promoter in vivo. Similar results were seen for Ad5-OB36-luc.
Dual luciferase-imaging system
To create 786-O-CBGr, we stably transfected the renal cancer cell line 786-O with the gene for click beetle green luciferase. Both click beetle green luciferase and firefly luciferase convert D-luciferin, but they emit light at different peak wavelengths: 550 and 610 nm, respectively.
To test the system, 786-O and 786-O-CBGr were infected with Ad5luc1, which expresses firefly luciferase under the ubiquitous CMV promoter. Without a filter, light emission was seen in 786-O cells infected with 1000 vp (virus particles) per cell (luciferase produced by the virus) and in all wells with 786-O-CBGr (luciferase produced by virus+cells, Figure 3a). To separate between virus- and cell-produced luciferase, filters were used. When applying the DsRed filter (emission passband 575–650 nm, useful for firefly luciferase present in the virus) only in the wells infected with 1000 vp per cell, light emission could be detected. Using the green fluorescent protein (GFP) filter (emission passband 515–575 nm, useful for CBGr luciferase), light emission from all 786-O-CBGr wells was seen. No light was detected in the 786-O wells with the GFP filter, as expected.
Dual luciferase imaging in vivo
Severe combined immunodeficiency (SCID) mice were injected intraperitoneally with 786-O or 786-O-CBGr cells. When mice were imaged 20 days later without a filter, light emission was detected in 786-O-CBGr-injected mice (Figure 3b). Using the (virus-specific) DsRed filter, basically no light was detected, whereas with the (cell-specific) GFP filter, light emission was comparable to that observed without filters.
Immediately after imaging, mice were injected intraperitoneally with Ad5luc1 and imaged again 2 days later. Without a filter and with the virus-specific DsRed filter, strong light emission from the liver and peritoneum were detected (Figure 3c). When applying the CBGr-specific GFP filter, no light was detected from 786-O injected mice, whereas 786-O-CBGr mice showed similar light emission as before virus injection.
In vivo analysis of promoter specificity with the dual luciferase-imaging system
SCID mice bearing intraperitoneal 786-O-CBGr tumors were injected intraperitoneally with Ad5-9HIF-luc (Figure 4a), Ad5-OB36-luc (animals not shown) or Ad5luc1 (animals not shown). Site of luciferase expression from Ad5-9HIF-luc matched well with the location of tumors as visualized by dual luciferase imaging (Figure 4a). Immediately after imaging, mice were killed and their livers analyzed ex vivo for luciferase expression. Disappointingly, Ad5-OB36-luc was found to express slightly more luciferase than Ad5luc1 (Figure 4b). However, it was promising that Ad5-9HIF-luc showed significantly lower marker gene expression compared with Ad5luc1 and Ad5-OB36-luc in livers of analyzed animals (P<0.0001). The relative tumor to liver expression ratios of 9HIF and OB36 were 62.6 and 2, respectively.
Construction of a targeted and armed oncolytic adenovirus
On account of high in vivo tumor expression and low liver expression, 9HIF was chosen as a promoter to control E1 for the new oncolytic adenoviruses Ad5/3-9HIF-Δ24-E3 and Ad5/3-9HIF-Δ24-VEGFR-1-Ig (Figure 5a). In parallel to Ad5/3-Δ24, these viruses also feature a 24-bp deletion in E1A and the 5/3-serotype chimerism capsid modification. Ad5/3-9HIF-Δ24-VEGFR-1-Ig additionally carries the gene for VEGFR-1-Ig in the E3 region driven by the endogenous E3 promoter.
Western blot for VEGFR-1-Ig was done with supernatant from infected 786-O cells (Figure 5b, lanes 1–3) and 769-P cells (lanes 4–6). The expected band at circa 180 kDa was detected in samples infected with Ad5/3-9HIF-Δ24-VEGFR-1-Ig (lanes 1 and 4), whereas Ad5/3-9HIF-Δ24-E3 (lanes 2 and 5) and mock (lanes 3 and 6)-infected samples did not show any bands.
To assess whether insertion of the VEGFR-1-Ig affects virus replication, a progressive infectivity assay was performed with two kidney cancer cell lines (Figure 5c). Ad5/3-9HIF-Δ24-VEGFR-1-Ig showed the same replication kinetics as Ad5/3-9HIF-Δ24-E3, which is the isogenic control virus without the VEGFR-1-Ig transgene, suggesting that insertion of the VEGFR-1-Ig gene does not affect virus replication.
In vitro cytotoxicity of Ad5/3-9HIF-Δ24-VEGFR-1-Ig
Renal cancer cell lines 786-O and SN12L1-luc were infected and cell viability was measured after 9 days. Ad5/3-Δ24 killed all cells at the highest concentration of 10 vp per cell, whereas with Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad5/3-9HIF-Δ24-E3, around 50–60% of the cells remained alive (Figures 6a and b). Ad300wt, a wild-type adenovirus type 5, and Ad5/3luc1, a replication deficient control virus, did not have any effect on cell viability. Similar results were seen with ACHN and Caki-2 cells (Supplementary Figure 3). Ad300wt had a stronger oncolytic effect on 769-P cells, whereas Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad5/3-9HIF-Δ24-E3 seemed to be less active. On Sv7tert cells, Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad300wt seemed to have the same oncolytic effect, whereas Ad5/3-9HIF-Δ24-E3 was basically inactive (Supplementary Figure 3). Therefore, the rather low in vitro (as opposed to in vivo) activity of 9HIF seemed nevertheless sufficient for killing of renal cancer cell lines.
Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad5/3-9HIF-Δ24-E3 did not kill human fibroblasts (FHS173WE), whereas all Ad5/3-Δ24-infected cells were dead and also Ad300wt had some oncolytic effect (Figure 6c), despite low levels of the Ad5 receptor in these cells.15 On human umbilical vein cells (HUVEC), Ad5/3-Δ24 and Ad300wt were equally effective in killing all cells, whereas Ad5/3-9HIF-Δ24-E3 was completely inactive (Figure 6d). Rapidly replicating normal cells phosphorylate retinoblastoma,32 rendering Ad5/3-Δ24 replication competent, which might explain the emphatic oncolysis seen in HUVEC and FHS173WE cells. When Ad5/3-9HIF-Δ24-VEGFR-1-Ig replicates, it produces VEGFR-1-Ig, which is toxic to endothelial cells, as they need VEGF for survival.33 However, only few HUVEC cells were killed by Ad5/3-9HIF-Δ24-VEGFR-1-Ig, suggesting tight control over virus replication by the constant region 2 deletion (Δ24) combined with the 9HIF promoter.
In vivo efficacy of Ad5/3-9HIF-Δ24-VEGFR-1-Ig
In a subcutaneous kidney cancer model induced with 786-O cells, wild-type adenovirus (Ad300wt) was able to slow down tumor growth to some extent, whereas Ad5/3-9HIF-Δ24-VEGFR-1-Ig could significantly reduce tumor size (Figure 7a, P<0.001 compared with mock). Ad5/3-9HIF-Δ24-E3 and Ad5/3-Δ24 had the strongest antitumor effects, reducing average tumor sizes to 33 and 22% of the initial size. In both groups, three out of eight-injected tumors were completely eradicated. ELISA with serum from Ad5/3-9HIF-Δ24-VEGFR-1-Ig-injected mice confirmed VEGFR-1-Ig expression and secretion, which was significantly higher compared with the serum levels of the Ad5/3-9HIF-Δ24-E3-treated group (P<0.001 for all time points, Figure 7b). Tumor sections stained for blood vessels showed markedly decreased vasculature in Ad5/3-9HIF-Δ24-VEGFR-1-Ig compared with Ad5/3-9HIF-Δ24-E3-injected tumors (Figure 7c). Moreover, quantification of blood vessel density yielded significantly lower values for tumors treated with Ad5/3-9HIF-Δ24-VEGFR-1-Ig compared with Ad5/3-9HIF-Δ24-E3 (Figure 7d, P=0.02). Hexon staining of tumor sections showed that more virus was present in the Ad5/3-9HIF-Δ24-VEGFR-1-Ig-treated tumors compared with the Ad5/3-9HIF-Δ24-E3 group (Figure 7e).
Subcutaneous tumors may not be particularly representative of advanced kidney cancer found in human. Therefore, we used an orthotopic intraperitoneal model induced with SN12L1-luc cells. In this model, all oncolytic viruses prolonged survival (Figure 8a). Ad5/3-9HIF-Δ24-E3 showed statistically significant survival improvement compared with mock (P=0.008), whereas Ad5/3-9HIF-Δ24-VEGFR-1-Ig injection resulted in statistically longer survival compared with mock, Ad5/3-Δ24 and Ad5/3-9HIF-Δ24-E3 injections (P<0.0001, P=0.006 and P=0.028).
Taking advantage of luciferase production by SN12L1-luc cells, we could estimate the number of tumor cells remaining in live mice at multiple time points. Luciferase light emission from mock mice increased rapidly during the experiment, whereas signals from Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad5/3-9HIF-Δ24-E3 initially decreased and then increased modestly (Figure 8b; Supplementary Figure 4). Luciferase light emission from Ad5/3-Δ24-injected mice decreased slightly more initially, but was eventually higher than signals of Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad5/3-9HIF-Δ24-E3-injected mice.
Recent trials with oncolytic adenoviruses suggest safety and preliminary evidence of efficacy is also available.9, 10 However, treatment with virus alone has only rarely led to significant responses in patients with advanced cancers. Thus, the efficacy of oncolytic adenoviruses has to be improved to make them a viable treatment option for cancer. Increasing the effectiveness of a replicating virus might in turn lead to more severe side effects. The most likely organ at risk is the liver and, therefore, it is key to reduce hepatic viral replication. To this end, we constructed Ad5/3-9HIF-Δ24-VEGFR-1-Ig, a targeted and armed oncolytic adenovirus, and evaluated its specificity and efficacy.
We evaluated two tissue-specific promoters (9HIF and OB36) and found that both were active in various renal cancer cell lines (Figures 1a–c; Supplementary Figure 1), although generally rather low activity compared with CMV promoter was seen. However, it has been earlier suggested that because the activity of the CMV promoter is very high, even a few percent of it can be sufficient for expressing E1 at levels sufficient for virus replication.34, 35 Importantly, only 9HIF showed completely abrogated activity in fibroblasts (Figure 1d), suggesting that it is more tumor specific than OB36. Both promoter elements showed increased activity under hypoxic conditions (Figure 1e–h), with 9HIF having the stronger induction in all cell lines (Supplementary Figure 2).
Surprisingly, in vivo both elements showed higher activity than the CMV promoter (Figures 2c and d), whereas in vitro with the same cell line <10% of the activity of the CMV promoter was seen (Figure 1a). This underlines the induction of HIF expression that may occur in the tumor environment.17 Together with von-Hippel–Lindau mutations found in kidney cancers,18 hypoxic conditions typical of in vivo tumors seem to make HIF-responsive promoters, such as 9HIF appealing for restriction of gene expression or virus replication to renal cancer tissue.
The luciferase most commonly used in bioluminescence imaging is from the firefly. It converts D-luciferin, thereby emitting light with a peak wavelength of 610 nm. Click beetle green luciferase also converts D-luciferin, but with a peak wavelength of 550 nm of the emitted light. Thus, both genes can be specifically imaged simultaneously by applying suitable emission filters. We set up a dual luciferase-imaging system with viruses expressing firefly luciferase and a cell line expressing click beetle green luciferase and were able to simultaneously localize viral marker gene expression and cells after a single injection of D-luciferin (Figure 3).
The specificity of the 9HIF and OB36 HRE elements in comparison to the CMV promoter was evaluated in an intraperitoneal kidney cancer model using the dual luciferase-imaging system. Localization of tumors and 9HIF-driven viral gene expression were well matched, suggesting high specificity of the HRE promoter for renal cancer tissue (Figure 4a). Additionally, livers from mice were excised and analyzed ex vivo for luciferase expression. Here, we saw significantly reduced activity of 9HIF compared with OB36 and CMV promoter (Figure 4b). This suggested that placing adenoviral replication regulating genes, such as E1A, under the control of 9HIF might yield a tumor-specific virus.
On the basis of the promoter evaluation presented in this paper, and earlier studies on adenovirus capsid modification for renal cell cancer22, we constructed Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad5/3-9HIF-Δ24-E3, an isogenic control virus (Figure 5a). These viruses are targeted with 9HIF driving the E1 genes, which are the key regulators of adenoviral replication. Additional tumor cell targeting is achieved through a 24-bp deletion in E1A, making replication preferential for p16/retinoblastoma pathway deficient cells, which may include all kidney cancer cells.14 Furthermore, infectivity of these viruses was enhanced by inserting a serotype 3 fiber knob into an otherwise serotype 5-based virus. This retargets the viruses to serotype 3 receptor, which is still disputed, but highly expressed on kidney cancer cells.22, 23
Additionally, Ad5/3-9HIF-Δ24-VEGFR-1-Ig is armed with the gene for the soluble VEGFR-1-Ig fusion protein in the E3 region under the control of the endogenous E3 promoter, coupling the expression to virus replication.36 VEGFR-1-Ig binds VEGF28 without triggering angiogenic signal transduction, as VEGFR-1-Ig consists only of the extracellular immunoglobulin homology domains of VEGFR-1. Thus, VEGFR-1-Ig efficiently blocks VEGF-mediated angiogenesis for inhibition of tumor growth.29, 37 For example, sFlt, a physiologically occurring soluble VEGFR-1 splice variant, can yield an antitumor effect when overexpressed by cancer cells.30, 31
We hypothesized that Ad5/3-9HIF-Δ24-VEGFR-1-Ig would exhibit enhanced specificity and improved antitumor effect with regard to renal cell cancer. In vitro, oncolytic effect of Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad5/3-9HIF-Δ24-E3 was weaker than of the non-targeted, non-armed control Ad5/3-Δ24, but usually stronger than with wild-type virus Ad300wt (Figures 6a and b; Supplementary Figure 2). This was an expected result, as 9HIF activity was shown to be low in vitro (Figure 1), reducing replication of Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad5/3-9HIF-Δ24-E3. Also, VEGFR-1-Ig is not expected to add utility in vitro.
Practically, no effect of Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad5/3-9HIF-Δ24-E3 was seen on fibroblasts, which underscores the high specificity for cancer cells of these viruses (Figure 6c). The non-targeted viruses killed HUVECs very efficiently, whereas Ad5/3-9HIF-Δ24-E3 had no effect (Figure 6d), as HIF is not active in these cells under normoxic conditions.38 However, Ad5/3-9HIF-Δ24-VEGFR-1-Ig caused about a 30% reduction in cell viability, suggesting that low levels of VEGFR-1-Ig are expressed from this virus even in the absence of oncolysis.
In a subcutaneous in vivo experiment, Ad5/3-9HIF-Δ24-E3 and Ad5/3-Δ24 showed the best antitumor effects with complete eradication of 38% of the tumors in both groups (Figure 7a). This confirmed the high induction of HREs in vivo compared with in vitro. Although Ad5/3-9HIF-Δ24-VEGFR-1-Ig had antitumor efficacy, it was not more potent than Ad5/3-9HIF-Δ24-E3 and Ad5/3-Δ24, despite confirmed VEGFR-1-Ig expression (Figure 7b) and significant reduction of tumor vasculature (Figures 7c and d). A reason for Ad5/3-9HIF-Δ24-VEGFR-1-Ig being less effective than Ad5/3-9HIF-Δ24-E3 in this model might be collapse of vasculature due to VEGFR-1-Ig compromising intratumoral dissemination and/or vascular reinfection, as suggested before.22, 39 In fact, we found more virus present in Ad5/3-9HIF-Δ24-VEGFR-1-Ig compared qwith Ad5/3-9HIF-Δ24-E3-treated tumors (Figure 7e), supporting this hypothesis. Compromised viral dissemination resulting in virus sequestration within the tumor might subsequently increase local VEGFR-1-Ig expression levels.
As subcutaneous tumors may not be a clinicinally meaningful representation of metastatic renal cell cancer, we set up an orthotopic model with intraperitoneally metastatic SN12L1-luc cells. Ad5/3-9HIF-Δ24-VEGFR-1-Ig-treated mice survived significantly longer than any other group, suggesting that in this model the antiangiogenic effect was useful (Figure 8). One reason might be that in the intraperitoneal scenario, vascular dissemination and/or reinfection is not as crucial as in subcutaneous xenografts.
Our data suggest that triple targeting, including dual-level transcriptional targeting, capsid modification and arming can improve the selectivity and efficacy of oncolytic adenoviruses. Thus, as Ad5/3-9HIF-Δ24-VEGFR-1-Ig seems to be a promising oncolytic virus for treatment of metastatic kidney cancer, a thorough preclinical toxicity evaluation should follow. For example, it is known that high-level expression of sFlt may have potential for toxicity.40 With regard to safety, it may, therefore, be useful that the VEGFR-1-Ig expression system used in Ad5/3-9HIF-Δ24-VEGFR-1-Ig is tightly linked to replication,41, 42 and thus, VEGFR-1-Ig expression occurs preferentially at tumor sites.
Importantly, there is promising human safety and efficacy data for aflibercept (VEGF-trap),43 a related antiangiogenic molecule that consists of selected domains of VEGFR-1 and -2 fused to an Fc tail from IgG. Furthermore, as the biodistribution of viruses is determined by their capsid, it is promising that murine data suggest that Ad5/3-modified viruses, thus including Ad5/3-9HIF-Δ24-VEGFR-1-Ig, have similar biodistribution as Ad5,21, 44, 45 which has been safe in >15 000 reported cancer patients.5, 7 In summary, if formal preclinical toxicity analyses provide acceptable data, Ad5/3-9HIF-Δ24-VEGFR-1-Ig might be feasible for clinical testing in patients with renal cell cancer refractory to all other available treatments.
Materials and methods
Renal cancer cell lines 786-O, ACHN, 769-P, Sv7tert and Caki-2 were obtained from American Type Culture Collection (ATCC, Manassas, VA, USA) and maintained in recommended conditions. The 786-O-CBGr, which stably expresses click beetle green luciferase, was generated by transfection of 786-O cells with a plasmid carrying the puromycin resistance gene and the click beetle green luciferase gene and subsequent antibiotic selection of surviving cell clones. Renal cancer cell lines SN12C, SN12L1 and SN12L1-luc, which expresses firefly luciferase, were a kind gift from Dr Christoph Peter (Department of Cardiovascular Medicine, Stanford University, CA, USA). Human umbilical vein endothelial cells (Lonza, Basel, Switzerland) and the human fibroblast cell line FHS173WE (obtained from ATCC) were cultured under recommended conditions. The 293, 911 and A549 cells, which were used for generation of adenoviruses, were also obtained from ATCC and cultured under recommended conditions.
The HREs 9HIF and OB36 have been described before.46, 47 For construction of non-replicating adenoviruses, expression cassettes with either 9HIF or OB36 HREs controlling firefly luciferase gene were inserted into the multiple cloning site of pShuttle (Stratagene, La Jolla, CA, USA). Shuttle plasmids were recombined with pAdeasy-1 plasmid (Stratagene), which carries the whole adenovirus genome, and resulting rescue plasmids were transfected to 293 cells to generate Ad5-9HIF-luc and Ad5-OB36-luc.48
For construction of replicating adenoviruses, the gene for VEGFR-1-Ig (first five domains of VEGF receptor 1 fused to Fc tail of human IgG antibody, kindly provided by Dr Kari Alitalo, University of Helsinki, Finland) was cloned into pTHSN plasmid that contains the E3 region of the adenoviral genome replacing the 6.7K/gp19K genes.49 The resulting plasmid was recombined with pAdeasy-1.5/3-Δ24,49 an adenovirus rescue plasmid containing the serotype 3 knob and a 24-bp deletion in E1A, resulting in pAdeasy-1.5/3-Δ24-VEGFR-1-Ig. The 9HIF was inserted into pSEΔ24,15 a shuttle plasmid containing the E1 region and a 24-bp deletion in E1A, to construct pSEΔ24-9HIF. This shuttle plasmid was then recombined with pAdeasy-1.5/3-Δ24-VEGFR-1-Ig and pAdeasy-1.5/3-Δ24 resulting in pAdeasy-1.5/3-9HIF-Δ24-VEGFR-1-Ig and pAdeasy-1.5/3-9HIF-Δ24-E3, which were transfected to 911 cells for generation of Ad5/3-9HIF-Δ24-VEGFR-1-Ig and Ad5/3-9HIF-Δ24-E3.
Viruses were amplified on 293 (non-replicating viruses) or A549 (replicating viruses) cells and purified on double cesium chloride gradients. Presence of inserted genes and absence of wild-type virus was confirmed by PCR and sequencing. The vp to plaque-forming units ratios for Ad5-9HIF-luc, Ad5-OB36-luc, Ad5/3-9HIF-Δ24-VEGFR-1-Ig, Ad5/3-9HIF-Δ24-E3, Ad5luc1, Ad5/3luc1, Ad5/3-Δ24 and Ad300wt were 184, 119, 19, 60, 4, 13, 10 and 19, respectively.
Marker gene transfer assays
Cells were infected with replication deficient, luciferase-expressing viruses for 30 min, and then washed and incubated with complete growth medium at 37 °C under normoxic or hypoxic conditions (Invivo2 400 Hypoxic Workstation, Ruskinn Technology, Pencoed, UK). After 24 h, luciferase assay (Luciferase Assay System, Promega, Madison, WI, USA) was performed according to the manufacturer's manual.
Cells were infected with 10 vp per cell, medium was changed after 1 h and cells were incubated for 72 h. Western blot was done with cell culture supernatant using anti-human-IgG antibody (GE Healthcare, Barrington, IL, USA) for detection of VEGFR-1-Ig protein.
Progressive infectivity assay
Cell were infected with 0.0001–1000 vp in 10 replicates and cytopathic effect was microscopically followed over 10 days. The plaque-forming units per milliliter values were calculated based on the standard TCID50 method.42
In vitro cytotoxicity assays
Cells were infected with viruses, and after 1 h, infection medium was replaced with medium containing 5% FCS, which was changed every other day. After 8–11 days (at the optimal time point for each cell line), cell viability was analyzed with MTS assay (Cell Titer 96 AQueous One Solution Cell Proliferation Assay, Promega). All viruses were analyzed simultaneously for each cell line.
All animal experiments were approved by the experimental animal committee of the University of Helsinki and the provincial government of Southern Finland. Female nude or SCID mice were purchased from Taconic (Ejby, Denmark) at the age of 4 to 5 weeks and quarantined for 1 week. Mice were frequently monitored for their health status and euthanized as soon as any sign of pain or distress was observed.
For luciferase expression experiments, two nude mice per group were injected subcutaneously with 5 × 106 786-O cells inducing four tumors per mouse. When diameters of tumors were approximately 5 mm, 3 × 108 vp was injected intratumorally. Two days later, mice were imaged using IVIS imaging system series 100 (Xenogen, Alameda, CA, USA) and photon emission values were calculated with Living Image v2.5 software (Xenogen). Mice were killed, tumors were excised, ground and resuspended in lysis buffer and analyzed for luciferase expression as described above.
For the intraperitoneal models, tumors were induced with 107 786-O or 786-O-CBGr cells in three SCID mice per group. After 20 days, mice were imaged as described above and 108 vp was administered intraperitoneally. Two days later, mice were imaged using GFP, DsRed or no emission filter (IVIS imaging system series 100, Xenogen). Mice were then killed, livers were excised and prepared and analyzed for luciferase expression as described above.
In the oncolytic efficacy experiments, four nude mice per group were injected with 5 × 106 786-O cells to induce two subcutaneous tumors per mouse. When diameters of tumors were approximately 5 mm, 108 vp was injected intratumorally. Tumor size was measured with calipers and blood samples were taken on day 7, 11 and 15 after virus injection. VEGFR-1-Ig concentration in the mouse serum was determined with a human IgG ELISA kit (Immunology Consultants Laboratory, Newberg, OR, USA). At the end of the experiment on day 17, mice were killed and tumors were excised and frozen for immunostainings.
For the survival experiment, 10 SCID mice per group were injected intraperitoneally with 107 SN12L1-luc cells. A single intraperitoneal virus injection of 5 × 108 vp was performed on day 10 after cell injection. Mice were monitored for survival and imaged as described above on day 9, 18, 25 and 32 after cell injection.
Cryosections of 4–5 μm thickness of frozen tumors embedded in Tissue Tek OCT. (Sakura, Torrance, CA, USA) were prepared and fixed either in 4% paraformaldehyde 15 min for immunohistochemistry or in acetone for 10 min at −20 °C for immunofluorescence staining.
For adenovirus hexon staining, sections were boiled for 7 min in 0.01 M sodium citrate and incubated in 0.3% H2O2 in PBS for 10 min. The slides were incubated with the primary adenovirus hexon antibody (MA1-82982, ABR-Affinity BioReagents, Burlingame, CA, USA) for 1 h at room temperature. Slides were washed in PBS and then incubated with the kit PowerVision Poly-HRP-anti-mouse/rabbit/rat (ImmunoVision Technologies Co., Brisbane, CA 94005, USA). Bound antibodies were visualized using 0.3 μg/μl 3,3′-diaminobenzidine (DAB, Sigma, St Louis, MO, USA) in PBS, to which 0.03% H2O2 was added. Sections were counterstained with Mayer hematoxylin. Representative pictures were captured at × 20 magnification using an Axioplan 2 microscope equipped with Axiocam (Zeiss, Oberkochen, Germany).
For immunofluorescence staining, sections were incubated with normal donkey serum for 15 min, and then reacted with primary polyclonal rabbit anti-Von Willebrand Factor (1:200 dilution, DakoCytomation, Glostrup, Denmark) overnight. After washing with PBS, sections were incubated with Alexa Fluor 594-labeled secondary antibody (1:250 dilution, Molecular Probes, Invitrogen) for 30 min. Sections were fixed in 4% paraformaldehyde and mounted with Vectashield mounting medium (Vector Laboratories, Burlingame, CA, USA). Representative pictures of areas of the tumors with the highest microvessel density were captured at × 20 magnification.
For determination of the vascular density, stained blood vessels were computationally segmented from the images by the Otsu's method.51 The ratio of vessel coverage was determined as the ratio of the area of the vessels and the total area of the images.
To compare differences between groups, two-tailed student's t-test was used and a P-value of <0.05 was considered significant. P-values of the in vivo subcutaneous experiment were calculated by Mann–Whitney test (SPSS 13.0). Data of survival experiments was plotted as Kaplan–Meier graphs and a log-rank t-test (SPSS 13.0) was used for pairwise comparison of groups.
Conflict of interest
The authors declare no conflict of interest.
Parkin DM, Bray F, Ferlay J, Pisani P . Global cancer statistics, 2002. CA Cancer J Clin 2005; 55: 74–108.
Vogelzang NJ, Stadler WM . Kidney cancer. Lancet 1998; 352: 1691–1696.
Coppin C, Porzsolt F, Awa A, Kumpf J, Coldman A, Wilt T . Immunotherapy for advanced renal cell cancer. Cochrane Database Syst Rev 2005; 25: CD001425.
Yang JC, Haworth L, Sherry RM, Hwu P, Schwartzentruber DJ, Topalian SL et al. A randomized trial of bevacizumab, an anti-vascular endothelial growth factor antibody, for metastatic renal cancer. N Engl J Med 2003; 349: 427–434.
Kanerva A, Hemminki A . Adenoviruses for treatment of cancer. Ann Med 2005; 37: 33–43.
Ribacka C, Hemminki A . Virotherapy as an approach against cancer stem cells. Curr Gene Ther 2008; 8: 88–96.
Liu TC, Kirn D . Gene therapy progress and prospects cancer: oncolytic viruses. Gene Therapy 2008; 15: 877–884.
Alemany R . Cancer selective adenoviruses. Mol Aspects Med 2007; 28: 42–58.
Yu W, Fang H . Clinical trials with oncolytic adenovirus in China. Curr Cancer Drug Targets 2007; 7: 141–148.
Khuri FR, Nemunaitis J, Ganly I, Arseneau J, Tannock IF, Romel L et al. A controlled trial of intratumoral ONYX-015, a selectively-replicating adenovirus, in combination with cisplatin and 5-fluorouracil in patients with recurrent head and neck cancer. Nat Med 2000; 6: 879–885.
Alemany R, Curiel DT . CAR-binding ablation does not change biodistribution and toxicity of adenoviral vectors. Gene Therapy 2001; 8: 1347–1353.
Heise C, Hermiston T, Johnson L, Brooks G, Sampson-Johannes A, Williams A et al. An adenovirus E1A mutant that demonstrates potent and selective systemic anti-tumoral efficacy. Nat Med 2000; 6: 1134–1139.
Fueyo J, Gomez-Manzano C, Alemany R, Lee PS, McDonnell TJ, Mitlianga P et al. A mutant oncolytic adenovirus targeting the Rb pathway produces anti-glioma effect in vivo. Oncogene 2000; 19: 2–12.
Sherr CJ . Cancer cell cycles. Science 1996; 274: 1672–1677.
Bauerschmitz GJ, Guse K, Kanerva A, Menzel A, Herrmann I, Desmond RA et al. Triple-targeted oncolytic adenoviruses featuring the cox2 promoter, E1A transcomplementation, and serotype chimerism for enhanced selectivity for ovarian cancer cells. Mol Ther 2006; 14: 164–174.
Cuevas Y, Hernandez-Alcoceba R, Aragones J, Naranjo-Suarez S, Castellanos MC, Esteban MA et al. Specific oncolytic effect of a new hypoxia-inducible factor-dependent replicative adenovirus on von Hippel–Lindau-defective renal cell carcinomas. Cancer Res 2003; 63: 6877–6884.
Vaupel P, Kallinowski F, Okunieff P . Blood flow, oxygen and nutrient supply, and metabolic microenvironment of human tumors: a review. Cancer Res 1989; 49: 6449–6465.
Kim WY, Kaelin WG . Role of VHL gene mutation in human cancer. J Clin Oncol 2004; 22: 4991–5004.
Kanerva A, Mikheeva GV, Krasnykh V, Coolidge CJ, Lam JT, Mahasreshti PJ et al. Targeting adenovirus to the serotype 3 receptor increases gene transfer efficiency to ovarian cancer cells. Clin Cancer Res 2002; 8: 275–280.
Dmitriev I, Krasnykh V, Miller CR, Wang M, Kashentseva E, Mikheeva G et al. An adenovirus vector with genetically modified fibers demonstrates expanded tropism via utilization of a coxsackievirus and adenovirus receptor-independent cell entry mechanism. J Virol 1998; 72: 9706–9713.
Ranki T, Kanerva A, Ristimaki A, Hakkarainen T, Sarkioja M, Kangasniemi L et al. A heparan sulfate-targeted conditionally replicative adenovirus, Ad5.pk7-Delta24, for the treatment of advanced breast cancer. Gene Therapy 2007; 14: 58–67.
Guse K, Ranki T, Ala-Opas M, Bono P, Sarkioja M, Rajecki M et al. Treatment of metastatic renal cancer with capsid-modified oncolytic adenoviruses. Mol Cancer Ther 2007; 6: 2728–2736.
Haviv YS, Blackwell JL, Kanerva A, Nagi P, Krasnykh V, Dmitriev I et al. Adenoviral gene therapy for renal cancer requires retargeting to alternative cellular receptors. Cancer Res 2002; 62: 4273–4281.
Fukata S, Inoue K, Kamada M, Kawada C, Furihata M, Ohtsuki Y et al. Levels of angiogenesis and expression of angiogenesis-related genes are prognostic for organ-specific metastasis of renal cell carcinoma. Cancer 2005; 103: 931–942.
Nicol D, Hii SI, Walsh M, Teh B, Thompson L, Kennett C et al. Vascular endothelial growth factor expression is increased in renal cell carcinoma. J Urol 1997; 157: 1482–1486.
Ferrara N . Role of vascular endothelial growth factor in the regulation of angiogenesis. Kidney Int 1999; 56: 794–814.
Kendall RL, Wang G, Thomas KA . Identification of a natural soluble form of the vascular endothelial growth factor receptor, FLT-1, and its heterodimerization with KDR. Biochem Biophys Res Commun 1996; 226: 324–328.
Olofsson B, Korpelainen E, Pepper MS, Mandriota SJ, Aase K, Kumar V et al. Vascular endothelial growth factor B (VEGF-B) binds to VEGF receptor-1 and regulates plasminogen activator activity in endothelial cells. Proc Natl Acad Sci USA 1998; 95: 11709–11714.
Kim KJ, Li B, Winer J, Armanini M, Gillett N, Phillips HS et al. Inhibition of vascular endothelial growth factor-induced angiogenesis suppresses tumour growth in vivo. Nature 1993; 362: 841–844.
Goldman CK, Kendall RL, Cabrera G, Soroceanu L, Heike Y, Gillespie GY et al. Paracrine expression of a native soluble vascular endothelial growth factor receptor inhibits tumor growth, metastasis, and mortality rate. Proc Natl Acad Sci USA 1998; 95: 8795–8800.
Mahasreshti PJ, Navarro JG, Kataram M, Wang MH, Carey D, Siegal GP et al. Adenovirus-mediated soluble FLT-1 gene therapy for ovarian carcinoma. Clin Cancer Res 2001; 7: 2057–2066.
Yang B, Cao DJ, Sainz I, Colman RW, Guo YL . Different roles of ERK and p38 MAP kinases during tube formation from endothelial cells cultured in 3-dimensional collagen matrices. J Cell Physiol 2004; 200: 360–369.
Gerber HP, Dixit V, Ferrara N . Vascular endothelial growth factor induces expression of the antiapoptotic proteins Bcl-2 and A1 in vascular endothelial cells. J Biol Chem 1998; 273: 13313–13316.
Casado E, Gomez-Navarro J, Yamamoto M, Adachi Y, Coolidge CJ, Arafat WO et al. Strategies to accomplish targeted expression of transgenes in ovarian cancer for molecular therapeutic applications. Clin Cancer Res 2001; 7: 2496–2504.
Barker SD, Coolidge CJ, Kanerva A, Hakkarainen T, Yamamoto M, Liu B et al. The secretory leukoprotease inhibitor (SLPI) promoter for ovarian cancer gene therapy. J Gene Med 2003; 5: 300–310.
Hawkins LK, Johnson L, Bauzon M, Nye JA, Castro D, Kitzes GA et al. Gene delivery from the E3 region of replicating human adenovirus: evaluation of the 6.7 K/gp19 K region. Gene Ther 2001; 8: 1123–1131.
Millauer B, Shawver LK, Plate KH, Risau W, Ullrich A . Glioblastoma growth inhibited in vivo by a dominant-negative Flk-1 mutant. Nature 1994; 367: 576–579.
Calvani M, Rapisarda A, Uranchimeg B, Shoemaker RH, Melillo G . Hypoxic induction of an HIF-1alpha-dependent bFGF autocrine loop drives angiogenesis in human endothelial cells. Blood 2006; 107: 2705–2712.
Heldin CH, Rubin K, Pietras K, Ostman A . High interstitial fluid pressure—an obstacle in cancer therapy. Nat Rev Cancer 2004; 4: 806–813.
Mahasreshti PJ, Kataram M, Wang MH, Stockard CR, Grizzle WE, Carey D et al. Intravenous delivery of adenovirus-mediated soluble FLT-1 results in liver toxicity. Clin Cancer Res 2003; 9: 2701–2710.
Kanerva A, Bauerschmitz GJ, Yamamoto M, Lam JT, Alvarez RD, Siegal GP et al. A cyclooxygenase-2 promoter-based conditionally replicating adenovirus with enhanced infectivity for treatment of ovarian adenocarcinoma. Gene Therapy 2004; 11: 552–559.
Raki M, Hakkarainen T, Bauerschmitz GJ, Sarkioja M, Desmond RA, Kanerva A et al. Utility of TK/GCV in the context of highly effective oncolysis mediated by a serotype 3 receptor targeted oncolytic adenovirus. Gene Therapy 2007; 14: 1380–1388.
Riely GJ, Miller VA . Vascular endothelial growth factor trap in non small cell lung cancer. Clin Cancer Res 2007; 13: s4623–s4627.
Ranki T, Sarkioja M, Hakkarainen T, von Smitten K, Kanerva A, Hemminki A . Systemic efficacy of oncolytic adenoviruses in imagable orthotopic models of hormone refractory metastatic breast cancer. Int J Cancer 2007; 121: 165–174.
Kangasniemi L, Kiviluoto T, Kanerva A, Raki M, Ranki T, Sarkioja M et al. Infectivity-enhanced adenoviruses deliver efficacy in clinical samples and orthotopic models of disseminated gastric cancer. Clin Cancer Res 2006; 12: 3137–3144.
Aragones J, Jones DR, Martin S, San Juan MA, Alfranca A, Vidal F et al. Evidence for the involvement of diacylglycerol kinase in the activation of hypoxia-inducible transcription factor 1 by low oxygen tension. J Biol Chem 2001; 276: 10548–10555.
Boast K, Binley K, Iqball S, Price T, Spearman H, Kingsman S et al. Characterization of physiologically regulated vectors for the treatment of ischemic disease. Hum Gene Ther 1999; 10: 2197–2208.
Hemminki A, Dmitriev I, Liu B, Desmond RA, Alemany R, Curiel DT . Targeting oncolytic adenoviral agents to the epidermal growth factor pathway with a secretory fusion molecule. Cancer Res 2001; 61: 6377–6381.
Kanerva A, Zinn KR, Peng KW, Ranki T, Kangasniemi L, Chaudhuri TR et al. Noninvasive dual modality in vivo monitoring of the persistence and potency of a tumor targeted conditionally replicating adenovirus. Gene Therapy 2005; 12: 87–94.
Kanerva A, Zinn KR, Chaudhuri TR, Lam JT, Suzuki K, Uil TG et al. Enhanced therapeutic efficacy for ovarian cancer with a serotype 3 receptor-targeted oncolytic adenovirus. Mol Ther 2003; 8: 449–458.
Ostu N . A threshold selection method from gray-level histograms. IEEE Trans Sys, Man, Cyber 1979; 9: 62–66.
This work was supported by European Research Council, EU FP6 APOTHERAPY and THERADPOX, HUCH Research Funds (EVO), Finnish Cancer Society, Sigrid Juselius Foundation, Academy of Finland, Biocentrum Helsinki, University of Helsinki, Helsinki Graduate School in Biotechnology and Molecular Biology, Helsinki Biomedical Graduate School, K Albin Johansson Foundation, Orion-Farmos research foundation and Finnish Cultural Foundation. We acknowledge Eerika Karli, Aila Karioja-Kallio, Ville Rantanen and Roxana Ola for technical assistance.
About this article
Cite this article
Guse, K., Diaconu, I., Rajecki, M. et al. Ad5/3-9HIF-Δ24-VEGFR-1-Ig, an infectivity enhanced, dual-targeted and antiangiogenic oncolytic adenovirus for kidney cancer treatment. Gene Ther 16, 1009–1020 (2009). https://doi.org/10.1038/gt.2009.56
- oncolytic adenovirus
- renal cell cancer
- antiangiogenic gene therapy
- dual luciferase imaging
Expert Opinion on Biological Therapy (2019)
Oncolytic adenovirus and doxorubicin-based chemotherapy results in synergistic antitumor activity against soft-tissue sarcoma
International Journal of Cancer (2015)
Combinatorial treatment with oncolytic adenovirus and helper-dependent adenovirus augments adenoviral cancer gene therapy
Molecular Therapy - Oncolytics (2014)
Expert Opinion on Biological Therapy (2012)
Molecular Pharmaceutics (2011)