Chimeric oncolytic adenovirus evades neutralizing antibodies from human patients and exhibits enhanced anti-glioma efficacy in immunized mice

Abstract

Oncolytic viruses are a promising treatment for patients with high-grade gliomas, but neutralizing antibodies can limit their efficacy in patients with prior virus exposure or upon repeated virus injections. Data from a previous clinical trial using the oncolytic adenovirus Delta-24-RGD showed that generation of anti-viral neutralizing antibodies may affect the long-term survival of glioma patients. Past studies have examined the effects of neutralizing antibodies during systemic virus injections, but largely overlooked their impact during local virus injections into the brain. We found that immunoglobulins colocalized with viral proteins upon local oncolytic virotherapy of brain tumors, warranting a strategy to prevent virus neutralization and maximize oncolysis. Thus, we generated a chimeric virus, Delta-24-RGD-H43m, by replacing the capsid protein HVRs from the serotype 5-based Delta-24-RGD with those from the rare serotype 43. Delta-24-RGD-H43m evaded neutralizing anti-Ad5 antibodies and conferred a higher rate of long-term survival than Delta-24-RGD in glioma-bearing mice. Importantly, Delta-24-RGD-H43m activity was significantly more resistant to neutralizing antibodies present in sera of glioma patients treated with Delta-24-RGD during a phase 1 clinical trial. These findings provide a framework for a novel treatment of glioma patients that have developed immunity against Delta-24-RGD.

Graphical abstract

Fueyo and colleagues observed that anti-viral immune responses may have hindered the efficacy of virotherapy in clinical trials. To overcome this limitation, they generated a chimeric immune-stealth oncolytic adenovirus that evaded neutralizing antibodies from patients previously treated with parental oncolytic viruses and prolonged the survival of glioma-bearing mice.

Introduction

Survival rates for many cancers have improved significantly over the past decade due to progresses in early detection, advances in novel treatments and the advent of precision medicine; however, glioblastoma (GBM) continues to be an exception to this trend with long-term survivorship remaining unacceptably low. Furthermore, immunotherapy with immune checkpoint inhibitors, which has shown efficacy in other tumors, has not conferred survival benefits in most patients with malignant gliomas, suggesting that these tumors are immunologically “cold” and, therefore, resistant to immune checkpoint antibodies.¹ Clinical trials have indicated that oncolytic adenoviruses are a promising new category of anticancer biotherapeutic agents. In fact, the first oncolytic virus (OV) approved for standard clinical practice in 2006 was an adenovirus.² We have developed a platform of oncolytic adenoviruses called Delta-24-RGD that can selectively replicate in the Rb-pathway-deficient cells to treat brain tumors.^3,4 A first-in-human phase 1 clinical trial of Delta-24-RGD was successfully completed in patients with recurrent GBM (NCT00805376). In this trial, we observed durable (>3 years) responses in 20% of patients.⁵ In agreement with these data, similar results were confirmed in two additional clinical trials in patients with diffuse intrinsic pontine glioma (DIPG) and recurrent GBM.^6,7 Results from these trials suggest that Delta-24-RGD treatment induces strong immune responses and durable tumor regression in a subset of patients, but what differentiates responders from non-responders remains unclear.

Interestingly, analyses of data from pediatric patients showed that adenovirus-specific neutralizing antibody (NAb) titers could be used to stratify patients by survival.⁶ Patients who developed higher-than-median NAb titers had a median survival of 12.5 months, whereas those with lower-than-median NAb titers had a median survival of 21.3 months. This discovery led us to examine the role of NAbs in determining the efficacy of OVs. Delta-24-RGD is built on an adenovirus serotype 5 (Ad5) backbone. Ad5 is the most widely used adenoviral vector in cancer and gene therapies, but is also among the most prevalent serotypes.⁸ These factors indicate that the development of NAbs can influence the outcomes of oncolytic virotherapies. Prior studies investigating NAbs have used experimental models involving systemic virus injections or subcutaneous tumors, but largely shied away from using intracranial brain tumor models, possibly due to the notion that the blood-brain barrier (BBB) might prevent entry of large molecules, including antibodies, into the brain parenchyma. As such, conclusions drawn from these studies can be clinically less relevant to locally injected virotherapy of brain tumors.

In this work, we showed that antibodies are found colocalized with virus proteins in the brain tumors of mice treated with Delta-24-RGD. Based on these data, we hypothesized that evasion of NAbs will improve the anti-glioma efficacy of virotherapy. To test this hypothesis, we generated a chimeric oncolytic adenovirus called Delta-24-RGD-H43m. We found that this chimeric OV evades anti-Ad5 NAbs present in sera from mice and yields superior outcomes in orthotopic murine models of glioma. Of clinical interest, we found that intratumoral virus injections led to the development of anti-Ad5 NAbs in patients with malignant gliomas enrolled in a phase 1 clinical trial (NCT00805376),⁵ and demonstrated that Delta-24-RGD-H43m evades NAbs from these patients. These data provide clinically relevant rationale for the development of future clinical trials poised to examine the effect of NAb-resistant virotherapy in patients with gliomas and other solid tumors.

Results

Virotherapy induces the influx of antibodies into brain tumors

The presence of antibodies within brain tumors carries significant implications for potential virus neutralization and is of critical importance in the context of virotherapy for brain tumors. To determine whether NAbs might be present in gliomas upon OV intratumoral injection, we assessed the presence of immunoglobulins in a murine orthotopic glioma model. We intratumorally injected Delta-24-RGD into GSC005 murine glioma-bearing C57BL/6 mice previously immunized against Ad5. After performing transcardial perfusion with saline to clear blood from the brains, we assessed intratumoral virus replication and antibody presence by immunofluorescence staining of viral hexon and mouse IgG, respectively. As expected, we detected hexon protein in the brains from virus-injected mice but not in those of mock-treated mice (p = 0.007, unpaired t test) (Figures 1A and 1B). Of interest, we detected that 30% of fluorescence signals from intratumoral IgG colocalized with those from viral hexon protein in the Delta-24-RGD-treated specimens (Figures 1C and 1D). These data indicated that intratumoral replication of oncolytic adenoviruses occurred in the presence of IgG. Furthermore, significantly less adenovirus genome copies were detected in the brains of mice pre-immunized with Ad5 before intratumoral injection of Delta-24-RGD, compared with mice that received intratumoral injection of Delta-24-RGD without prior virus immunization (Figure S1). These results, together with data from a previous clinical trial showing the presence of NAbs in DIPG patients treated with Delta-24-RGD, suggested that virotherapy can potentially be limited by antibodies within the tumor microenvironment.

Figure 1.

Immunoglobulins and hexon expression co-localize in Delta-24-RGD-infected murine gliomas

(A) C57BL/6 mice immunized against wildtype Ad5 were intracranially injected with GSC005 cells. After 3 intratumoral injections of PBS or Delta-24-RGD, brains were collected and processed for immunofluorescence imaging. Blue indicates DAPI; green, adenovirus hexon; and magenta, mouse IgG. Scale bar, 50 μm. (B) Quantification of adenovirus hexon fluorescence signals. Data represent mean ± SD from individual mice; the p value was calculated using an unpaired t test. (C) Scatterplots showing hexon and IgG signals. For each pixel in the fluorescence images, signals corresponding to hexon and IgG were quantified and displayed in scatterplots using the Olympus cellSens software. (D) Percentages of IgG-positive fluorescence signals in brain tumors that colocalize with hexon-positive signals. Data represent mean ± SD from individual mice; the p value was calculated using an unpaired t test.

Generation and characterization of the chimeric OV Delta-24-RGD-H43m

To overcome the challenge that NAbs might represent in virotherapy, we designed a chimeric virus called Delta-24-RGD-H43m to treat brain tumors. To this end, using molecular cloning techniques, we swapped the hexon hypervariable regions (HVRs) of Ad5, which constitutes the dominant adenovirus antigens and, therefore, the major targets for NAbs,^9,10 with hexon HVRs from Ad43, a rare serotype to which less than 5% of the human population has been exposed and developed antibodies.¹¹ Delta-24-RGD-H43m has nine replacements in its hexon HVRs¹² (Table S1): a threonine-to-methionine substitution at position 351 (T351M; corresponding with T342M in the wildtype Ad5 hexon) that stabilizes protein folding¹³ and maintain the two mutations present in Delta-24-RGD (24-base pair deletion in E1A and RGD-4C motif insertion in the fiber) (Figure 2A). A homology-based protein folding algorithm predicted that parental and chimeric hexon proteins have distinct surfaces, suggesting antibodies specific to one surface may not recognize the chimeric surface (Figure 2B). Tests of thermostability of the new virus revealed that infectivity of Delta-24-RGD-H43m was not altered compared with Delta-24-RGD when the agent was exposed to several temperatures, such as 32°C, 37°C, and 42°C. Thus, DNA replications of the parental and chimeric adenoviruses were comparable at various times during the adenovirus genome replication cycle (Figure 2C). We next sought to determine whether Delta-24-RGD-H43m infection produced viable and infectious virus progenies. To this end, we observed that infection of human glioma U-87 MG cells, murine glioma GSC005 cells, or human lung adenocarcinoma A549 cells with either Delta-24-RGD-H43m or Delta-24-RGD induced the robust expression of both the early viral protein E1A (indicative of virus infection) and the late viral protein fiber (indicative of virus replication) (Figure 2D). Furthermore, transmission electron microscopy examination 72 h after infection with either Delta-24-RGD-H43m or Delta-24-RGD identified intracellular viral progenies (Figure 2E). In these microscopy images, intranuclear areas of the solid-liquid interface were in proximity to areas of adenovirus assembly, indicating active production of viral proteins and productive assembly of new virions. Together, these results suggested that Delta-24-RGD-H43m was stable and maintained comparable infectious and replicative properties as Delta-24-RGD.

Figure 2.

Generation and characterization of the chimeric oncolytic adenovirus Delta-24-RGD-H43m

(A) Genomic structure of Delta-24-RGD-H43m showing the mutations in the adenoviral genome, including the Ad5 and Ad43 components. (B) The predicted folded structures of the wildtype Ad5 (red) and chimeric (blue) hexon trimers were generated using the SWISS-MODEL algorithm. Hexon hypervariable regions are highlighted in colors. (C) Adenovirus genome copies were quantified using qPCR analyses of DNA extracted from A549 cells infected with Delta-24-RGD (red) or Delta-24-RGD-H43m (blue) and incubated at 32°C, 37°C, or 42°C. (D) Expression of viral proteins in U-87 MG, GSC005, and A549 cells was assessed 48 h after exposure to PBS (M) or infection with Delta-24-RGD (RGD) or Delta-24-RGD-H43m (H43m). Bar graphs show the quantification of the Western blots as mean ± SD from two to three independent experiments; M, PBS; R, Delta-24-RGD; H, Delta-24-RGD-H43m. ∗p ≤ 0.05, two-way ANOVA. (E) Transmission electron microscopy was used to visualize adenovirus particles in GSC005 and A549 cells 72 h after infection with Delta-24-RGD or Delta-24-RGD-H43m. Black or gray dots of adenovirus particles (arrowhead) and areas of solid-liquid interface near adenovirus assembly (arrow) are displayed. Scale bar, 500 nm.

Delta-24-RGD-H43m induces potent anti-glioma effect in vitro

We wanted to ascertain that the chimeric oncolytic virus retains the ability to infect and lyse human and murine glioma cells in vitro. Therefore, we plated human glioma U-87 MG, murine glioma GSC005, and human lung adenocarcinoma A549 and infected the cultures with Delta-24-RGD-H43m or Delta-24-RGD at doses ranging from 0 to 100 infectious units per cell (multiplicity of infection [MOI]). Daily observation of the cells under light microscopy showed profound morphological changes and dramatic signs of cytopathic effects at doses of 50 MOIs and higher three days after the infection (Figure S2). At a dose of 100 MOI, most cells infected with each of the viruses were lysed 72 h after infection. Therefore, Delta-24-RGD-H43m infection strongly inhibited cancer cell viability in a dose-dependent manner (Figures 3A and S3A). These data clearly demonstrated that the genetic modifications in Delta-24-RGD-H43m did not impair its capacity to effectively lyse cancer cells, reaffirming its potential as an oncolytic therapeutic for gliomas.

Figure 3.

Chimeric oncolytic adenovirus Delta-24-RGD-H43m exerts potent anti-cancer effect

(A) Viabilities of U-87 MG, GSC005, and A549 cells were measured 96 h after infection with Delta-24-RGD-H43m at the indicated MOIs. Data represent mean ± SD from three independent experiments. (B and C) J:NU outbred nude mice received intracranial (ic) injections of U-87 MG human glioma cells. PBS or Delta-24-RGD-H43m was injected intratumorally (it) on days 3, 5, and 7 after tumor implantation. (n = 9–10 per group) (B and D) C57BL/6 mice received intracranial injections of GSC005 murine glioma cells. PBS or Delta-24-RGD-H43m was injected intratumorally on days 3, 6, and 8 after tumor implantation. (n = 8–10 per group) (C and D) Survival was monitored for 100 days; p values were calculated using a RMST test. Brains were collected and subjected to histopathological analyses. Shown are representative images. Scale bar, 2 mm. IC50, half-maximal inhibitory concentration.

Treatment with Delta-24-RGD-H43m showed robust anti-tumor efficacy in glioma-bearing mice

To determine whether Delta-24-RGD-H43m treatment leads to tumor regression, we used orthotopic models of U-87 MG human gliomas in immunodeficient J:NU nude mice and GSC005 murine gliomas in immunocompetent C57BL/6 mice (Figures 3B–3D). Compared with their mock-treated counterparts, the Delta-24-RGD-H43m treatment of U-87 MG glioma-bearing mice prolonged the median survival and resulted in long-term survivors (35.5 vs. 14 days; 40% vs. 0%; p = 0.0003, restricted mean survival time [RMST] test) (Figure 3C). Next, we tested the effect of Delta-24-RGD-H43m in a syngeneic glioma model. Similarly, Delta-24-RGD-H43m treatment of GSC005 glioma-bearing mice resulted in higher median survival and more long-term survivors compared with mock treatment (undefined vs. 45.5 days; 62.5% vs. 0%; p < 0.0001, RMST test) (Figure 3D). Post-mortem examination of brains of PBS-treated mice revealed huge tumor masses at the time of their deaths. However, examination of brains in Delta-24-RGD-H43m-treated mice at the time of experiment termination (censored mice at 100 day of the experiment) showed complete regression of tumor masses (Figures 3C and 3D). Similar results were observed in mice bearing more established intracranial GSC005 tumors, tested by initiating the treatment with intratumoral Delta-24-RGD-H43m on day 7 after tumor implantation. In this experiment, the virus-treated cohort had prolonged median survival and more long-term survivors than those that received mock treatment (61 days vs. 42 days; 11% vs. 0%; p = 0.006, RMST test) (Figure S3B).

Delta-24-RGD-H43m shows serotype-specific resistance to murine neutralizing antibodies

We next investigated the extent to which the chimeric virus exhibits resistance to anti-Ad5 NAbs generated in mice by immunization (Figure 4A). To test the potency of the sera to inactivate adenoviruses, we identified serum dilution that inhibited more than 50% of virus activity. As expected, sera from treatment-naïve mice did not inactivate the activity of either virus (Figure 4B). Of interest, sera from mice immunized against Delta-24-RGD significantly inhibited Delta-24-RGD but not Delta-24-RGD-H43m (p = 0.004, paired t test) (Figure 4C). Sera from mice immunized against wild-type Ad5 showed similar results (data not shown). In contrast, sera from mice immunized against Delta-24-RGD-H43m showed the opposite effect and neutralized Delta-24-RGD-H43m but not Delta-24-RGD (p = 0.03, paired t test) (Figures 4D and S4). These results suggested that serotype-specific antibodies inhibited viral activities and confirmed that Delta-24-RGD-H43m is more resistant to anti-Ad5 antibody-mediated neutralization compared with Delta-24-RGD.

Figure 4.

Delta-24-RGD-H43m evades anti-Ad5 NAbs from Delta-24-RGD-immunized murine model

(A) C57BL/6 mice received intramuscular injections of PBS (n = 6), Delta-24-RGD (n = 10), or Delta-24-RGD-H43m (n = 4). Boost immunizations were performed 4 weeks later. Three weeks after boost immunization, sera were collected. (B–D) A549 cells were infected with Delta-24-RGD or Delta-24-RGD-H43m in the presence of sera collected from mice in experiment A. Cell viability was measured 48 h later to identify the serum dilutions that inhibit 50% of cell death. Each dot represents serum from each immunized mice; p values were calculated using paired t tests. (E and F). C57BL/6 mice bearing GSC005 tumors received Delta-24-RGD-H43m, Delta-24-RGD, or PBS alone injections on (E) days 3, 6, and 8 or (F) days 7 and 11 after tumor implantation. Mice also received intraperitoneal (ip) transfusions of Ad5-immunized serum on (E) day 2 or (F) days 6 and 10. Survival was monitored for 100 days; n = 9–11 per group; p values were calculated using RMST tests.

Delta-24-RGD-H43m has enhanced anti-tumor efficacy in glioma-bearing mice with anti-Ad5 immunity

To compare the therapeutic efficacies of Delta-24-RGD-H43m and Delta-24-RGD in the presence of anti-Ad5 immunity, we intracranially implanted GSC005 gliomas into C57BL/6 mice. To isolate the role of humoral immunity in the response to OVs, we induced a passive immunization by intraperitoneal injection of Ad5-immunized sera 2 days after tumor implantation, followed by intratumoral administration of Delta-24-RGD-H43m, Delta-24-RGD, or PBS alone (Figure 4E). Corroborating our data from in vitro experiments, mice treated with Delta-24-RGD-H43m had a longer median survival and more long-term survivors than those treated with Delta-24-RGD (undefined vs. 63 days; 80% vs. 45%; p = 0.05, RMST test) (Figure 4E). Similarly, in mice bearing more established intracranial GSC005 tumors, Delta-24-RGD-H43m treatment resulted in prolonged median survival and a higher percentage of long-term survivors than those treated with Delta-24-RGD (48.5 days vs. 43 days; 10% vs. 0%; p = 0.03, RMST test) (Figure 4F). These data suggest that therapeutic outcomes achieved with Delta-24-RGD-H43m are substantially better than those achieved with Delta-24-RGD in animals with immunity against Ad5.

Intratumoral Delta-24-RGD injection leads to the development of NAbs in patients with malignant gliomas

After analyzing the role of murine NAbs in neutralizing Delta-24-RGD, and since antibodies from humans and mice may target different viral epitopes,¹⁴ we sought to test the clinical relevance of these findings. We analyzed sera from a completed phase 1 clinical trial (NCT00805376) involving 37 patients with recurrent malignant gliomas treated with Delta-24-RGD.⁵ In this trial, 37 patients were divided into two groups: 25 patients, in group A, received a single stereotactic injection of Delta-24-RGD into tumors and were followed for toxicity and clinical outcomes; and 12 patients, in group B, received an intratumoral injection of Delta-24-RGD, followed 2 weeks after by en bloc tumor resection and multiple injections of Delta-24-RGD in the surgical cavity.⁵ Sera were collected from these patients at various time points, including before the initial virus injection, 1 month and 4 months after virus injections (Figure 5A). Before the initial treatment, two patients in group A had detectable amounts of anti-adenovirus NAbs, whereas no NAbs were detected from the patients in group B. We observed that approximately 40% of the patients from both group A and group B harbored NAbs one month after virus injection (Figure 5B). The percentage of patients in group A with NAbs did not change at 4 months after treatment, but 82% of the patients from group B developed NAbs at this time point (p = 0.05, paired t test) (Figures 5B and S4). These data demonstrated that administration of the OV resulted in the production of NAbs against the virus, and that multiple virus injections received by patients enrolled in group B resulted in a higher percentage of patients developing NAbs against the virus.

Figure 5.

Delta-24-RGD-H43m evades NAbs developed by patients with recurrent GBM after Delta-24-RGD treatment

(A) Patients with recurrent high-grade gliomas who were enrolled in a phase I trial (NCT00805376) received intratumoral injections of Delta-24-RGD (1 × 10⁷–3 × 10¹⁰ viral particles). In group A (n = 25), the patients received a single intratumoral virus injection. In group B (n = 12), the patients received an intratumoral virus injection, followed by en bloc tumor resection 2 weeks later and several doses of virus injection into the wall of the surgical cavities. Sera were collected at various time points. (B) Table showing percentage of NAb-positive sera from each group at baseline, 1 month and 4 months after virus injections. (C and D) A549 cells were infected with Delta-24-RGD or Delta-24-RGD-H43m in the presence of patient sera. Cell viability was measured 48 h later to identify the serum dilutions that inhibit 50% of cell death. Each dot represents serum from each patient; the p value was calculated using a paired t test. Numbers of patients are indicated above the x axes.

Delta-24-RGD-H43m evades NAbs present in sera from patients treated with Delta-24-RGD

To demonstrate that Delta-24-RGD-H43m escapes the inhibitory effects of sera from patients with malignant gliomas treated with Delta-24-RGD,⁵ we infected A549 cells with Delta-24-RGD or Delta-24-RGD-H43m mixed with the obtained patient sera and quantified cell viability 48 h later. In accordance with the results of our experiments using murine sera, the capability of Delta-24-RGD-H43m to evade the neutralization by sera collected 1 and 4 months after virus injection was significantly more potent than that of Delta-24-RGD (p = 0.01 and p = 0.0007, paired t test) (Figure 5D). Evaluation of virus activity in response to individual patient sera accentuated the higher resilience of Delta-24-RGD-H43m to neutralizing sera compared with Delta-24-RGD (Figure 5D). These data strongly suggested that Delta-24-RGD-H43m would be more efficacious than Delta-24-RGD in patients with immunity against Ad5.

Discussion

Our results highlight that oncolytic adenovirus capable of evading neutralizing antibodies against Ad5 should have anti-glioma efficacy in a greater percentage of patients. We also showed that modifications of the vector backbone by replacing the adenovirus hexon immunodominant epitopes are a feasible approach to circumvent adenovirus epitope immunodominance.

Humoral immunity to adenovirus is widely recognized as a hindrance to the widespread use of these biological agents as cancer therapies.¹⁵ Notably, two recent phase 1 clinical trials using oncolytic viruses to treat pediatric patients with gliomas revealed that virus-specific antibody titers could serve as potential survival biomarkers after virotherapy.^6,16 For instance, the median survival of pediatric patients with DIPG after Delta-24-RGD treatment could be stratified according to their virus-specific NAb titers.⁶ These findings align with results from another trial involving intratumoral injection of the oncolytic virus HSV-1 G207 in pediatric patients with HGGs. In this trial, patients with baseline HSV-1 IgG antibodies had a shorter median survival of 5.1 months, compared with 18.3 months for those without antibodies at the time of treatment.¹⁶ However, despite these correlations, there is currently a lack of direct clinical or preclinical evidence to confirm whether antibodies within brain tumors impede viral activities.

Arguably, evidence gathered from previous models to test the relevance of antibodies during virotherapy are clinically less relevant for gliomas, because they primarily focused on the treatment of subcutaneous tumors, either through systemic or local intratumoral virus injections.^17,18,19 There are exceptionally limited investigations regarding antibody-mediated neutralization during virotherapy for high-grade gliomas, possibly due to the prevailing notion that the BBB restricts antibody access to brains.²⁰ In this study, we used an orthotopic, immune-competent animal model to reveal the presence of antibodies in brain tumors and suggest their potential interference with oncolytic virotherapy.

We found that anti-Ad5 antibodies pose a significant challenge to the effectiveness of oncolytic adenoviruses. Thus, sera from either wildtype Ad5-or Delta-24-RGD-immunized mice inhibited the cytotoxicity of Delta-24-RGD in vitro (Figure 4C) and limited its anti-glioma efficacy in vivo (Figures 4F and S3). These findings are in agreement with prior studies that demonstrated the inhibitory impact of NAbs on adenoviral vector activity.¹⁵ Of further relevance, a patient undergoing intra-arterial adenoviral vector gene therapy suffered lethal systemic inflammation caused by the formation of adenovirus-antibody complexes due to pre-existing humoral immunity.²¹ These studies raise safety- and efficacy-related concerns that provided the rationale for the development of several strategies to protect therapeutic viruses from NAbs, which have included the use of polymer coatings and cellular vehicles.^22,23 Although these strategies can be useful for replication-deficient viral vectors that only require the initial delivery of transgenes into target cells, they could have limited efficacy for replication-competent viruses whose replicated progenies spread to nearby cells and would be exposed to the immunity of the host. In the case of oncolytic viruses, genetic modifications of the virus are arguably the most feasible strategy. The genetically modified, chimeric oncolytic adenovirus Delta-24-RGD-H43m effectively evaded anti-Ad5 NAbs, and consequently provided a significant survival benefit over Delta-24-RGD in glioma-bearing mice immunized against Ad5.

Our report is based on the hypothesis that anti-adenovirus NAbs negatively affect the oncolytic treatment efficacy with Delta-24-RGD. This hypothesis is borne out of the interpretation of the results from a previously published clinical trial⁶ and based on our data. An alternative explanation for these data could be that the patients that have a less favorable response and more tumor-cell survival, producing adenoviruses for a longer time or at higher levels, which could result in a higher titer of NAbs. In this alternate scenario, the higher NAb titers are not the cause of the decreased survival, but an indirect consequence.

Less than 5% of the human population has prior exposure to Ad43, which makes Ad43 an ideal serotype to build an antibody-evading chimera virus.¹¹ Ad43 may also have a different binding affinity to coagulation factor X (FX). A previous study has revealed that the γ-carboxyglutamic acid domain residue K10 of FX forms electrostatic interactions with Ad5 hexon residues E424 and T425.²⁴ Detection of Ad5-bound FX initiates an innate immune response in cells.²⁴ In contrast, FX coating protects Ad5 from IgM and the classical complement pathway.²⁵ While FX binding has not been directly for Delta-24-RGD-H43m, threonine and glutamic acid residues in Ad5 HVR5 and HVR7 responsible for interaction with FX are absent in Ad43 HVRs, disrupting FX binding to Ad43.²⁶ These factors have not been investigated in this study, but a comparison of Delta-24-RGD and Delta-24-RGD-H43m in glioma-bearing mice without Ad5-immunity showed comparable efficacies of the two viruses (Figure S3B). Interestingly, the fiber protein of Ad43 is short and predicted to be rigid, which may result in different infectivity compared with Ad5.²⁶ These factors make Delta-24-RGD-H43m that has HVRs of serotype 43 and fiber of serotype 5 an interesting oncolytic agent and calls for further investigation.

Chimeric viruses like Delta-24-RGD-H43m, which are focused on the ablation of the adenovirus dominant antigen, also warrant careful consideration, as other epitopes may become targets for immunity. In this regard, other genetic modifications can be added to further diminish immunodominance. For instance, some investigators have proposed the deletion of early adenoviral genes such as E4²⁷ or the replacement of adenoviral promoters²⁸ to reduce Ad-specific immune responses.²⁹ Although these strategies may enhance antibody evasion, they may, in turn, decrease the replication ability of the therapeutic virus.³⁰ Combination of hexon and fiber modifications may also increase immune evasion.³¹ These combined strategies should be examined in detail for immune-resistant oncolytic adenoviruses and from the mechanistic angle of adenovirus antigen immunodominance. Another strategy to evade the immune response against adenoviruses would be to sequentially inject different oncolytic viruses in the same patient instead of treatment with multiple injections of the same virus. Future studies should provide further insights into the intricate interactions between different viruses and various immune factors. Nonetheless, the data from our current study, coupled with previous research, strongly support our tenet that endowing oncolytic adenoviruses with the ability to partially evade antibody-mediated neutralization significantly enhances their anti-tumor efficacy.

Data from of a phase 1 clinical trial (NCT00805376)⁵ showed that multiple dosing of Delta-24-RGD led to the development of NAbs in 82% of the patients, compared with 41% of the patients who received a single dose of the virus (Figures 5B and 5C). These data will be relevant to the design of future clinical trials that aim to use repeated administrations of OVs, especially given the approval of multiple injections of the oncolytic HSV G47Δ for the treatment of malignant gliomas by the Japanese Ministry of Health, Labor and Welfare.³² Surprisingly, only 2 of 37 patients with recurrent malignant gliomas in this trial had basal levels of NAbs against the virus. This is in contrast with 6 of 12 patients with newly diagnosed DIPGs who had basal levels of NAbs against the virus.⁶ One possible explanation for the low prevalence of basal NAbs against the adenovirus may be attributed to the immunosuppressed state of the patients due to prior treatments.

Our chimeric oncolytic adenovirus evaded sera containing NAbs from these patients. These data, together with results from mouse models, showed that Delta-24-RGD-H43m exhibited a robust resilience to neutralization and significantly improved the anti-cancer effect of the parental virus, Delta-24-RGD. Therefore, these pre-clinical and clinical data warrant further studies to propel a future clinical trial to explore the anti-cancer effect of genetically modified oncolytic adenoviruses designed to mitigate anti-adenovirus immunity in patients with glioma.

Materials and methods

Cell lines and culture conditions

Human lung adenocarcinoma A549 cells (ATCC, Manassas, VA) were cultured in DMEM/F12 supplemented with 10% fetal bovine serum (FBS) and antibiotics (100 μg/mL penicillin and 100 μg/mL streptomycin; Corning, Fort Worth, TX). Human glioma U-87 MG cells (ATCC) were cultured in minimum essential medium containing non-essential amino acid solution, 10% FBS, and antibiotics. Murine glioma GSC005 cells (kindly provided by I.M. Verma, The Salk Institute for Biological Studies, La Jolla, CA)³³ were cultured in DMEM/F12 supplemented with N2 (Invitrogen, Carlsbad, CA), murine fibroblast growth factor 2 (20 ng/mL; PeproTech, Cranbury, NJ), murine epidermal growth factor (20 ng/mL; Promega, Madison, WI), and heparin (50 μg/mL; Sigma, St. Louis, MO). HEK293 cells (ATCC) were cultured in DMEM supplemented with 10% FBS and antibiotics.

Generation of oncolytic adenoviruses

The construction of Delta-24-RGD was previously described.³⁴ For the generation of Delta-24-RGD-H43m, a synthetic gene containing the hexon HVRs from Ad43 (replacing those in Ad5)^12,13 was inserted into SfiI sites of the pMA-T backbone (Invitrogen). Then, a homologous DNA recombination was performed between the SfiI-cut synthetic gene and the AsiSI- and NdeI-digested pVK500C-Δ24 in Escherichia coli BJ5183 (Agilent Technologies, Santa Clara, CA) to produce pVK500C-Δ24-AdH43m. An additional homologous recombination between the SwaI-digested pVK500C-Δ24-AdH43m and the XbaI- and KpnI-digested pXK-F-RGD resulted in a plasmid, pAd5-D24-RGD-H43m, that contained the full oncolytic adenoviral genome with a chimeric hexon. This plasmid was digested with PacI and transfected into HEK293 cells using X-tremeGENE HP DNA transfection reagent (Roche, Basel, Switzerland) to produce infectious viral particles. Sfil, AsiSI, NdeI, Swal, Xbal, Kpnl, and Pacl were purchased from New England BioLabs (Ipswich, MA). PCR and subsequent sequencing were used to test the integrity of the viral modifications. Oncolytic adenoviruses were amplified in A549 cells and purified using a two-step CsCl gradient centrifugation (O.D. 260 Inc, Boise, ID). Virus titers were calculated by infecting A549 cells with serially diluted viruses and counting the hexon-expressing cells as described previously.³⁵ The wildtype Ad5 was purchased from the American Type Culture Collection (ATCC).

Immunofluorescence imaging

Nine- to 12-week-old male and female C57BL/6 mice received intramuscular injections of 2.5 × 10⁸ infectious units of wild-type Ad5 diluted in 50 μL of PBS in their hind limbs. Four weeks later, 50,000 GSC005 cells in 5 μL of media were intracranially implanted into the mice. Mice received intratumoral injections of PBS alone or 5 × 10⁷ infectious units of Delta-24-RGD diluted in 5 μL of PBS on days 3, 6, and 8 after tumor implantation. On day 8, the mice were humanely euthanized, transcardial perfusion with PBS was performed, and their brains were collected and fixed with 10% neutral buffered formalin. Brain slices were embedded in paraffin, and immunofluorescence staining was performed as described previously.³⁶ Antibodies and their concentrations are given in Table S2.

Folding prediction modeling

DNA sequences of Ad5 hexon and chimeric hexon were used to generate prediction models using the protein structure homology-modeling server SWISS-MODEL.^37,38

Quantification of viral genome copies

A549 cells were seeded onto six-well plates (300,000 cells/2 mL media/well), and Delta-24-RGD or Delta-24-RGD-H43m was added at a concentration of 10 MOI. Cultures were incubated at 32°C, 37°C, and 42°C and cells were collected at 2, 24, and 48 h after infection. Genomic DNA was extracted using DNeasy Blood & Tissue kit (Qiagen, Hilden, Germany). Quantitative PCR was performed on the 7500 Fast Real-Time PCR System (Applied Biosystems) using SYBR Green PCR Master Mix (Applied Biosystems), as previously described.³⁶ Primers were bought from Sigma Aldrich (Ad5F: CAGCGTAGCCCCGATGTAA; Ad5R: TTTTTGAGCAGCACCTTGCA). Plasmid of known length and concentration was used as standards ranging from 10⁴ to 10⁸ genome copies using the following formula: Number of copies = (DNA concentration (ng/μL) × [6.022 × 1,023])/(length of template (bp) × [1 × 10⁹] × 650).

Western blot analysis

Cells were seeded onto six-well plates (300,000 cells/2 mL media/well), and Delta-24-RGD or Delta-24-RGD-H43m was added at a concentration of 10 MOI for U-87 MG and A549 cells or 100 MOI for GSC005 cells. Higher MOIs were used to infect murine cells, since Delta-24-RGD was generated on the backbone of human species C Ad5. Cultures were incubated at 37°C for 48 h, and then cells lysates were prepared using RIPA lysis buffer as described previously.³⁹ Total proteins (8 μg/sample) were loaded onto 4%–20% Novex Tris-Glycine gels (Invitrogen), transferred onto a polyvinylidene fluoride membrane, and detected with antibodies as described previously.³⁹ Antibodies and their concentrations are provided in Table S2. Bands were visualized using an enhanced chemiluminescence Western blot detection system (Amersham Pharmacia Biotech, Amersham, UK).

Cell viability assay

Cancer cells were plated in 96-well plates (10,000 cells/50 μL media/well); 2 h later, serially diluted Delta-24-RGD or Delta-24-RGD-H43m in 50 μL of media was added to each well. The cell-virus mixtures were incubated for 96 h at 37°C. On day 2, 100 μL additional media was added to the wells. On day 4, 30 μL CellTiter-Blue (Promega) was added to each well, and fluorescence was measured according to the manufacturer’s protocols.

Transmission electron microscopy

A549 and GSC005 cells were seeded onto six-well plates (300,000 cells/2 mL media/well), and Delta-24-RGD or Delta-24-RGD-H43m were added at a concentration of 2 MOI for A549 cells or 20 MOI for GSC005 cells. Cells were incubated at 37°C for 72 h and then fixed as described previously.⁴⁰ The samples were polymerized in a 60°C oven for approximately 3 days. Ultrathin sections were cut in a Leica Ultracut microtome (Leica, Deerfield, IL), stained with uranyl acetate and lead citrate in a Leica EM Stainer, and examined in a JEM 1010 transmission electron microscope (JEOL, USA, Inc., Peabody, MA) at an accelerating voltage of 80 kV. Digital images were obtained using an AMT Imaging System (Advanced Microscopy Techniques Corp, Danvers, MA).

Generation of NAb-containing mouse sera

Nine- to 12-week-old male and female C57BL/6 mice received intramuscular injections of 2.5 × 10⁸ infectious units of wild-type Ad5 diluted in 50 μL PBS into their hind limbs. The mice received boost immunization 4 weeks later. Three weeks after boost immunization, blood was collected through cardiac puncture, allowed to solidify at room temperature for 30 min, and centrifuged at 2,000×g for 10 min to obtain sera.

Antibody-mediated neutralization of OVs

Sera from immunized mice or patients treated with Delta-24-RGD were heat-inactivated at 56°C for 1 h, centrifuged at 2,000×g for 10 min, serially diluted in media, and placed in 96-well plates (25 μL/well). Then, we added Delta-24-RGD or Delta-24-RGD-H43 m at a concentration of 20 MOI in 25 μL per well. A549 cells (10,000/50 μL media/well) were added, and the mixture was incubated at 37°C for 48 h. On day 2, 30 μL CellTiter-Blue (Promega) was added to each well, and fluorescence was measured according to the manufacturer’s protocols.

Survival studies

U-87 MG (500,000 cells/5 μL media/mouse) were implanted into the caudate nucleus of 9- to 14-week-old male J:Nu immunodeficient nude mice (Jackson Laboratory, Bar Harbor, ME) using a guide-screw system as described previously.⁴¹ Animals were housed in groups of five per cage. GSC005 cells (50,000 cells/5 μL media/mouse) were implanted into the caudate nucleus of 9- to 14-week-old male or female C57BL/6 mice (Jackson Laboratory) using the same method. On indicated days, the mice were randomly assigned to the indicated control and treatment groups. The number of mice per group treatment or control is indicated for each experiment in the corresponding figure legends, decided by previous experimentation studies.³ Mice received intratumoral injections of 5 μL PBS alone or with 5 × 10⁷ infectious units of Delta-24-RGD or Delta-24-RGD-H43m. Personnel involved in intratumoral treatments were not informed of the expected results and did not prepare the treatments or cells. For intraperitoneal serum transfusion, mice received 100 μL immunized sera 18 h before the initial virus injection. Mice were periodically monitored and euthanized when they showed signs of local or general disease or at 100 days after tumor implantation. No mice were excluded from the analyses. Brains were collected for histopathological analyses. All experimental procedures involving the use of mice were done in MD Anderson vivarium facilities, following recommended analgesia and anesthesia procedures, in accordance with protocols approved by MD Anderson’s Animal Care and Use Committee and in accordance with National Institutes of Health and U.S. Department of Agriculture guidelines.

Statistical analysis

GraphPad Prism 9 was used to perform statistical analyses and generate graphs for in vitro and in vivo experiments. Two-tailed Student t tests were used to determine statistical differences between 2 groups, and one-way or two-way ANOVA with multiple comparisons was used to determine statistical differences among three or more groups. Animal survival curves were plotted according to the Kaplan-Meier method. To compare the survival rates of different treatment groups, the RMSTs were calculated using the R package survRM2 version 1.0-4. Default values of the truncation time tau were used for the analyses.

Data and code availability

The data generated in this study are available within the article and its supplementary data files.