Antibody-Neutralized Reovirus Is Effective in Oncolytic Virotherapy

The use of oncolytic virus (OV) therapy (a recognized form of immunotherapy) is progressing in the clinic, with confidence in the field increasing following FDA approval for the first agent in class, talimogene laherparepvec (T-VEC, a herpes simplex virus encoding GM-CSF) to treat melanoma (1). However, OVs are not used as widely as other types of immunotherapy such as checkpoint inhibitors, possibly owing to the perception that systemic administration will be limited by neutralizing antibodies (NAb). NAb may be present at baseline for viruses prevalent in the human population (e.g., herpes simplex virus type 1, and mammalian orthoreovirus type 3, herein referred to as “reovirus”). NAb may also arise following initial doses of OV therapy. Such concerns potentially limit systemic OV therapeutic strategies to a “one shot” approach, whereby patients receive a single high dose of OV (2), or to direct OV injection into tumors. Indeed, FDA approval for T-VEC is for intratumoral (i.t.) delivery only. Although this route ensures viral access to the tumor, it is technically challenging and limits treatment to readily accessible tumors. Systemic delivery is safe, broadly applicable in a clinical setting, and more suitable for targeting visceral or widespread metastatic disease. We and others have previously investigated an approach that circumvents NAb-mediated neutralization by delivering virus within carrier cells (3, 4). This strategy is also clinically challenging but, unexpectedly, developments from this work indicated a potential positive role for NAb in OV therapy.

We showed that i.t. delivery of single-agent reovirus was more effective as an antitumor therapy in mice than systemically administered reovirus (5). However, immune cells (T cells or dendritic cells) loaded with reovirus ex vivo and administered systemically, delivered virus to tumors, even in the presence of antireovirus NAb (3, 6). The results of a translational biological endpoint clinical trial (REO13), in patients with colorectal liver metastases, indicated that free reovirus delivered systemically without cell carriage, could access tumors, and that functional virus was associated with immune cells in the blood but was not found in plasma (7). These data suggest that, although free reovirus is neutralized by NAb in the serum following intravenous (i.v.) delivery, replication-competent virus can be transported to tumors by blood cells. Consistent with this, preconditioning mice with GM-CSF to mobilize the myeloid compartment within the systemic circulation prior to i.v. reovirus treatment resulted in effective therapy, the virus associating predominantly with CD11b⁺ cells in the blood (8). GM-CSF preconditioning was only effective in reovirus-immunized mice with high serum antireoviral NAb, consistent with NAb contributing to therapeutic efficacy.

In the current study, a human in vitro assay is described, in which monocytes are loaded with fully neutralized reovirus in the form of reovirus/neutralizing antibody (reoNAb) complexes and cocultured with tumor cell targets. Antibody-neutralized reovirus was unable to infect and kill tumor cells directly, but when loaded onto human monocytes it was delivered to melanoma cells in a functional/replicative form that resulted in cell lysis. After loading, antibody-neutralized reovirus was internalized by monocytes and processed to release infectious viral particles. The internalization process involved surface Fc receptors (FcR), predominantly FcγRIII expressed on nonclassical monocytes. These data indicate that circulating monocytes may be pivotal to preserving the therapeutic potency of OVs, despite preexisting antiviral immunity.

Cell lines were grown in DMEM or RPMI containing l-glutamine (Sigma) supplemented with 10% (v/v) heat-inactivated fetal calf serum (Life Technologies). Cell lines were monitored routinely using Mycoalert (Lonza) and found to be free of Mycoplasma infection (last test August 2017). Cell lines Mel-624, Sk-Mel-28, PC-3, and SKOV-3 were obtained from the CRUK cell bank in 2003. Mel-624, SK-Mel-28, and SKOV-3 cells were reauthenticated in 2012 using STR profiling and comparison with the DSMZ database; in the absence of a reference profile within the DSMZ database, cell lines were shown to have an original STR profile that was distinct from all other cell lines within the database. PC-3 cells have not been reauthenticated. Vero, L929, and HCT116 cells were obtained from ATCC in 2008, 2012, and 2013, respectively, and have not been reauthenticated. All cell lines were stored in liquid nitrogen. After thawing, cells were routinely passaged twice per week for no more than 20 weeks.

Reovirus Type 3 Dearing strain (Reolysin) was supplied by Oncolytics Biotech; Coxsackievirus type A21 (CVA21, CAVATAK) was supplied by Viralytics; and Herpes Simplex virus 1716 (HSV1716, Seprehvir) was supplied by Virttu Biologics. Stock virus concentrations were determined by plaque assay on L929 (reovirus), SK-Mel-28 (CVA21), Vero (HSV1716) cells. UV inactivation of reovirus was by 2-minute UV irradiation of 100 μL aliquots in a 96-well plate, using a Stratalinker UV 1800 (Stratagene), and was confirmed to be nonreplicative by plaque assay.

Serum was obtained from patients enrolled in clinical trials: for reovirus, the REO13-BRAIN trial (ISRCTN70443973); for CVA21, the STORM trial (NCT02043665). All patients gave written informed consent according to good clinical practice guidelines. Protocol, patient information sheet, and consent forms were approved by the United Kingdom Medicines and Healthcare products Regulatory Authority, regional ethics review committee, and institutional review board at St. James's University Hospital. Blood was collected into tubes containing a clotting activator. Samples were centrifuged at 2000 rpm for 10 minutes, the serum fraction was harvested and stored at −70°C. Pleural fluid from patients treated with intrapleural HSV1716 (trial NCT01721018) was a gift from Joe Conner (Virttu Biologics). Where required, serum was heat inactivated by incubation in a water bath at 56°C for 30 minutes.

Untreated or heat-inactivated serum samples were diluted in Gelatin Veronal Buffered (GVB++) Saline (Sigma). Increasing volumes were added to vortexed sheep erythrocytes (Stratech) and GVB++ to a final volume of 1.5 mL according to the manufacturer's protocol (CompTech). Negative and positive controls were included to give background and 100% lysis values, respectively. Tubes were placed in a 37°C water bath for 60 minutes, and cells were resuspended every 10 minutes and then placed on ice and centrifuged for 3 minutes at 800 × g. Supernatants were transferred to a Maxisorp 96-well plate and absorbance at 540 nm was determined using a Multiskan EX plate reader (Thermo).

Percentage lysis = (OD test sample − OD blank)/(OD total lysis − OD blank) × 100.

Halving dilutions of serum or pleural fluid were added to 80% confluent monolayers of susceptible cells (see above) in a 96-well plate. Virus was added to achieve a multiplicity of infection (MOI) 0.05 (reovirus and CVA21) or MOI 1 (HSV1716). Cell survival was assayed at 72 hours by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay.

Peripheral blood mononuclear cells (PBMCs) were isolated from healthy donor leukapheresis cones by density-gradient centrifugation on lymphoprep (Axis Shield). CD14⁺ monocytes were isolated from PBMCs by positive selection with anti-CD14 Microbeads; for electron microscopy (EM), monocytes were negatively selected from PBMCs using the Human Pan Monocyte Isolation kit; CD16⁺ monocytes were selected from PBMCs using the CD16⁺ Monocyte Isolation kit (all kits from Miltenyi).

For in vitro assays, OV was incubated with a predetermined neutralizing volume of patient-derived serum (reovirus, CVA21) or pleural fluid (HSV1716) for 2 to 3 hours at 37°C. For in vivo experiments, the serum was obtained from mice preimmunized i.p. with 2 doses of 2 × 10⁷ pfu reovirus 7 days apart; serum was obtained 7 days after the second immunization. The reovirus-specific monoclonal antibodies used to generate the complexes were obtained from DSHB (Iowa) and were clones G5, 10F6, 8H6, 10G10, and 10C1. Predetermined neutralizing volumes were mixed with the virus and incubated for 2 to 3 hours at 37°C.

OVs, NAb, or OV/NAb complexes were added to isolated monocytes and incubated at 4°C for 2 to 3 hours. Cells were washed 3× in PBS, resuspended in RPMI, and added to target cells either directly or separated by a 1-μm transwell (Greiner Bio-one) at a ratio of 3:1 (monocytes:targets). They were cocultured at 37°C for 72 hours, unless stated otherwise. Cell viability was analyzed by flow cytometry using a LiveDead stain (Thermo) according to the manufacturer's instructions. Where indicated, JAM-A on the target cells was blocked by preincubating with 10 μg/mL anti-JAM-A, clone J10.4 (Santa-Cruz) for 30 minutes at 37°C. FcR on the monocytes were blocked by pretreatment with 100 μg/mL F(ab′)₂ fragment antibodies specific for FcγR (Ancell) or human recombinant anti-FcαR (Miltenyi) at 4°C for 45 minutes.

Serum was diluted 1:1 in PBS and incubated for 90 minutes at room temperature (RT) with agarose bead-conjugated antibodies specific for the human γ- or α-chain (Sigma). The samples were then centrifuged to remove beads (3,000 × g, 15 seconds), and the supernatant was harvested. Antibody depletion was confirmed by enzyme-linked immunosorbent assay (ELISA) using human IgG/IgA ELISA kits (Mabtech).

Lysates from reovirus-infected (MOI = 1) Mel-624 or L929 cells (20 μg protein per lane) were separated by SDS–PAGE electrophoresis, transferred to nitrocellulose, blocked in 5% milk and probed using patient-derived serum (1:200 dilution) as primary antibody. Blots were washed 3× in PBST and incubated with an HRP-conjugated goat secondary antibody against human IgG, IgA, or IgM (all Thermo), diluted 1:5,000 in 5% milk/PBST. After a further 3 washes, blots were visualized with the chemiluminescent SuperSignal West Pico substrate (Thermo) on a Gel Doc XR system using Image Lab software (Bio-Rad).

Reovirus was added to serum at a 1:5 (v/v) ratio and incubated at 37°C for 3 hours. 1.5 mL Eppendorf tubes were blocked with 3% (w/v) bovine serum albumin for 1 hour at 4°C, prior to the addition of reovirus–antibody samples. Prewashed protein A resin beads (GenScript) in excess were mixed with samples and allowed to bind for 2 hours at 4°C on a rotator. Samples were centrifuged (400 × g, 2 minutes), washed 4× in 0.1% (v/v) Triton-X in PBS, then boiled (95°C, 5 minutes) in loading buffer to dissociate IgG from beads, and centrifuged (13,200 × g, 2 minutes) to yield supernatant for analysis.

Reovirus stock was dropped onto Veco 100-mesh copper grids (Electron Microscopy Sciences) and allowed to attach (RT, 5 minutes). Grids were washed 4× in PBS prior to incubation (90 minutes, RT) with patient-derived serum or control serum, diluted 1:10 in PBS. After 4× washes in PBS, grids were incubated with protein A–conjugated 10 nm gold particles (1:300 in PBS + 1% v/v BSA) for 30 minutes at RT. After washing (4× PBS, 4× ddH₂O), grids were fixed for 1 hour with 1.5% glutaraldehyde in 0.1 mol/L sodium cacodylate. After 4× washes in ddH₂O, grids were negatively stained with 1% phosphotungstic acid for 30 seconds, then blotted and air-dried. Grids were visualized using an FEI Tecnai TWIN microscope at 120 kV (magnification, ×52,000).

Negatively selected monocytes were loaded, with either live reovirus or reoNAb at MOI 50, washed twice with ice-cold PBS and resuspended in 2% (v/v) PFA + 0.2% (v/v) glutaraldehyde in 0.1 mol/L PHEM buffer; they were then pelleted, resuspended in storage buffer (0.5% w/v PFA in 0.1 mol/L PHEM) and kept at 4°C prior to processing. Cells were post-fixed for 1 hour at 4°C with 1% (w/v) osmium tetroxide in 0.1 mol/L sodium cacodylate buffer, rinsed in buffer and resuspended in 2% (w/v) agar. Blocks of 0.5 to 1 mm³ were cut, dehydrated in ethanol followed by propylene oxide, and then infiltrated with ascending ratios of LX-112 Epon resin/propylene oxide (1 hour each) finishing in pure resin. Resin was polymerized at 70°C for 48 hours, and 80-nm sections were cut using an Ultracut S microtome (Leica). TEM sections were viewed using an FEI Tecnai TWIN microscope at 120 kV.

Monocytes were loaded with live reovirus or reoNAb (MOI 10), resuspended in complete RPMI and cultured for 24 hours, then harvested, RNA extracted using the RNeasy mini kit (Qiagen) according to the manufacturer's instructions and treated with DNase I. mRNA libraries were prepared using the NEBNext Ultra Directional RNA library prep kit (New England BioLabs) and sequenced using the HiSeq 2500 system (Illumina). Fastq files were analyzed in R using the DEseq2 package (Bioconductor).

These were carried out at the University of Leeds or the Mayo Clinic, Rochester MN. All in vivo studies were approved by either the Leeds local ethics committee and UK Home Office or the Mayo IACUC. Six- to 8-week-old female C57Bl/6 mice were purchased from Charles River Laboratories or Jackson Laboratories. Mice were challenged subcutaneously with 5 × 10⁵ B16 melanoma cells. One treatment cycle of GM-CSF/reoNAb = 300 ng GM-CSF i.p. on 3 consecutive days followed by 2 × 10⁷ pfu reoNAb complexes i.v. on the following 2 days.

One cycle of treatment was given to mice bearing 7-day established tumors. Tumors were harvested on day 14, weighed, and divided for analysis by plaque assay and qRT-PCR. For plaque assay, the tumor sample was homogenized and subjected to 3 cycles of freeze/thaw and then clarified by centrifugation and viral titer determined by plaque assay on L929 cells. For qRT-PCR, RNA was extracted using the RNeasy mini kit (Qiagen) according to the manufacturer's instructions. cDNA synthesis was carried out using the SuperScript IV first-strand system (Thermo) according to the manufacturer's instructions. Analysis was conducted using the ABI 7500 real-time system (Applied Biosystems), and reovirus S3 copy number was quantified using the ΔΔCT method against GAPDH as comparator.

Mice bearing 3-day established tumors were given one treatment cycle as described above. Tumors were measured 3 times per week, and mice were euthanized when tumors reached 1 cm diameter.

Data were analyzed using GraphPad Prism software. Significance was evaluated using the Student t test (multiple comparisons with Holm–Sidak correction), χ² test, or one-way ANOVA (with Tukey correction) as appropriate, with P < 0.05 considered statistically significant. Survival analysis was carried out using the log-rank test.

Serum was obtained from patients on a biological endpoint clinical trial (REO13-BRAIN) in which they received i.v. reovirus (1 × 10¹⁰ TCID₅₀) as monotherapy prior to surgical resection of brain tumors (primary or metastatic). Blood samples were taken at least 7 days after reovirus treatment to ensure a high titer of antireoviral NAb and the serum was isolated. A standard neutralization assay indicated that serum from all patients was highly neutralizing toward reovirus (Fig. 1A), compared with serum from control donors (Supplementary Fig. S1A). To demonstrate the presence of reovirus-specific antibodies in the patient-derived serum, Western blots of lysates from reovirus-infected cells (L929 cells or Mel-624 cells) were performed using patient-derived serum as the primary detection antibody and secondary antibodies against human IgG, IgA, or IgM. Both IgG and IgA antibodies in the serum recognized a range of reoviral proteins (Fig. 1B); IgM antibodies reactive to reovirus were not found. Depletion of IgG or IgA antibodies from the serum using specific anti-IgG or -IgA agarose beads showed that both isotypes contributed to reovirus neutralization, with IgG antibodies being predominant (Fig. 1C).

It has been suggested that complement plays a role in the neutralization of reovirus (9). We investigated this via heat inactivation of patient-derived serum. Figure 1D shows that heat inactivation did not affect the neutralizing capacity of serum, suggesting that heat-labile factors such as complement do not neutralize reovirus in vitro. Complement activity within patient-derived serum was verified (Supplementary Fig. S1B).

Patients receiving i.v. therapeutic doses of reovirus develop a high reovirus-specific antibody titer, with IgG and IgA antibodies, but not complement, contributing to virus neutralization.

Our preclinical in vivo data led us to propose a model in which, following i.v. infusion, reovirus was bound by NAb to form reoNAb complexes that were delivered to tumors via monocytes (8). Therefore, the formation of the proposed reoNAb complexes was verified using EM. Reovirus was allowed to adhere to EM grids, which were then incubated with patient-derived serum or control serum from normal donors. Protein A gold labeling indicated the association of IgG with reovirus particles, confirming the formation of reoNAb complexes (Fig. 2A). More gold particles were associated with the reovirus following incubation with patient-derived serum (76%) than with control serum (40%; Fig. 2A–C). Some antireoviral NAb in control serum is expected, as most people have had prior exposure to the virus (10–12). Our data are consistent with our previous clinical trial, in which NAb were present at baseline in patients, but increased 100- to 1,000-fold after i.v. reovirus administration (7).

Thus, the reovirus-specific antibodies present in patient-derived serum can bind reovirus in vitro producing reoNAb complexes. ReoNAb complexes would be formed in vivo following systemic reovirus therapy, and concentrations would increase upon repeat reovirus administration.

To determine whether monocytes might be capable of delivering reoNAb to tumors in patients, we assessed the association of reovirus with human monocytes in the presence of neutralizing serum. Whole blood from normal donors was mixed with patient-derived serum and reovirus was then added. In the presence of NAb, virus was loaded onto CD14⁺ cells more efficiently than other immune cells (Supplementary Fig. S2). Next, we designed a human in vitro assay (Fig. 3A) in which human monocytes were loaded with either live nonneutralized reovirus or preformed reoNAb complexes. The ability of these monocytes to induce tumor cell death was examined. ReoNAb complexes were generated by incubating reovirus with a predetermined volume of neutralizing patient-derived serum at 37°C for 3 hours. The complexes or nonneutralized reovirus were loaded onto isolated human monocytes that were then cocultured with melanoma target cells. Melanoma targets were also treated with reovirus or reoNAb complexes in the absence of monocytes. After 72 hours, the cells were harvested and melanoma cell viability was determined by flow cytometry. Mel-624 cells treated only with reoNAb complexes showed no loss of viability compared with controls; however, when tumor cells were cultured with monocytes carrying reoNAb complexes, significant cell death was observed (Fig. 3B and C). Monocytes loaded with nonneutralized reovirus induced more Mel-624 death than those loaded with reoNAb complexes (monocytes loaded with reovirus induced mean 96.5% ± 0.40% cell death, those loaded with reoNAb induced mean 81% ± 2.74% cell death). Mel-624 cells cultured with monocytes alone showed no loss of viability or reduction in growth rate (Fig. 3B and C).

These results show that reovirus was fully neutralized within the reoNAb complexes but following loading onto monocytes, the complexes induced tumor cell death.

The observed tumor cell death could be mediated either by the monocytes themselves, following their activation by reoNAb complexes, or by release or transfer of infectious reovirus from the monocytes. Therefore, reoNAb complexes were generated using either live or UV-inactivated reovirus; both activate monocytes (Supplementary Fig. S3) but UV-inactivated reovirus is unable to infect and kill tumor cells directly (13). Monocytes loaded with UV-reoNAb complexes abrogated melanoma cell death following coculture (Fig. 4A), suggesting that tumor cell death was due to reovirus infection and replication, rather than monocyte cytotoxicity. In support of this, reovirus titer within monocyte-reoNAb and tumor cell cocultures increased over time (Fig. 4B), indicative of an ongoing productive infection. Furthermore, blocking JAM-A (the known reovirus entry receptor) on the target melanoma cells inhibited cell death (Fig. 4C), indicating that reovirus infection occurred via the normal entry route. However, separation of monocytes and tumor targets with a transwell abrogated cell death (Supplementary Fig. S4), suggesting that initial hand-over from the monocytes was contact dependent and that JAM-A was required for later viral spread between tumor cells. Reovirus replication occurred predominantly within tumor cells rather than monocytes, as reovirus titer did not increase over time in monocytes loaded with reoNAb complexes (Fig. 4D). This is in contrast to our previous observations in myeloid-derived human dendritic cells, which do support reovirus replication (3).

These data indicate that antibody-neutralized reovirus can be loaded onto monocytes and delivered to tumor cells in a functional form, resulting in infection and oncolysis.

Previously, we showed that live reovirus could be internalized by dendritic cells for delivery to tumor cells (6). Here, we investigated the fate of reoNAb complexes following their loading onto monocytes. EM demonstrated that reoNAb complexes were internalized by monocytes (Fig. 5A). Free reovirus was also internalized by monocytes but this appeared less efficient than reoNAb internalization, as some noncomplexed virus particles remained on the monocyte surface following loading with noncomplexed reovirus, whereas no reoNAb complexes were visible on the surface (Fig. 5B).

Having previously demonstrated that Fc receptors (FcR) were involved in the delivery of reovirus to tumors via monocytes in mice (8), their role in the delivery of reoNAb by human monocytes was examined. Expression of FcγRIII (CD16), FcγRII (CD32), FcγRI (CD64), and FcαR (CD89) was confirmed by flow cytometry (Supplementary Fig. S5) and the receptors were blocked prior to reoNAb loading. Blocking FcγRI or FcγRII had little effect on the amount of reovirus loaded onto the monocytes or the delivery of reoNAb to tumor cells. By contrast, blocking FcγRIII significantly reduced reovirus loading onto monocytes (Supplementary Fig. S6) and also melanoma cell death following coculture (Fig. 5C). Nonclassical CD16⁺ monocytes represent only a small fraction (approximately 10%) of the monocytic population. To analyze the involvement of FcγRIII, CD16⁺ and CD16⁻ monocytes were separated and their ability to deliver reoNAb to melanoma cells was compared. We confirmed that nonclassical CD16⁺ monocytes were more efficient in delivering reoNAb to induce melanoma cell death, whereas both classical and nonclassical monocytes were able to deliver free, noncomplexed reovirus efficiently (Fig. 5D). Furthermore, RNA-seq data showed that FcγR mRNA was upregulated in monocytes loaded with reoNAb complexes, FcγRIII mRNA showing the greatest increase (Supplementary Fig. S7). FcαR may also mediate uptake of reoNAb by monocytes but the effect was not as marked as for FcγRIII (Fig. 5C).

These data show that FcR, particularly FcγRIII, are involved in the uptake of reoNAb complexes by monocytes, though they may not be the only mechanism of uptake because tumor cell death was not abrogated by blocking these receptors.

In order to show that the delivery of reoNAb complexes to tumor cells via monocytes was not melanoma-specific, tumor cells of other histologic types were tested. ReoNAb complexes loaded onto monocytes were delivered to colorectal, prostate, and ovarian tumor cells, resulting in significant cell death (Fig. 6A). In addition, we have previously shown that preconditioning with GM-CSF, followed by systemic reovirus treatment, enhances survival in reovirus-immunized mice bearing TC2 (prostate) tumors (8) or intracranial GL-261 (glioma) tumors (14). Thus, the therapeutic efficacy of antireoviral NAb is likely to be applicable over a range of tumor types.

Various OVs are candidates as therapeutic agents; therefore, we asked whether antibodies to other OVs could contribute to therapy. Serum or pleural fluid was obtained from patients undergoing clinical trials with coxsackievirus (CVA-21) or herpes simplex virus (HSV1716) and used to generate CVA/NAb and HSV/NAb complexes. Both of these OV/NAb complexes were ineffective when cultured directly with melanoma targets, indicating complete neutralization of the viruses. Following loading onto monocytes, CVA/NAb were comparable with reoNAb in mediating tumor cell death, whereas HSV/NAb complexes were ineffective (Fig. 6B). Thus, anti-OV NAb may be useful in some but not all oncolytic virotherapies.

Although we have demonstrated the importance of reovirus-specific antibodies in the therapeutic response to i.v. reovirus therapy following GM-CSF preconditioning in mice (8), we have not shown that preformed reoNAb complexes can mediate delivery of functional reovirus to tumor-bearing mice. Therefore, we used serum from mice that had been preimmunized against reovirus—high antireoviral NAb (Supplementary Fig. S8)—to generate reoNAb complexes. These were injected i.v., with or without prior GM-CSF conditioning, into tumor-bearing mice. After 3 days, the tumors were harvested and examined for functional reovirus by plaque assay. Functional reovirus was detectable within the tumors of all of the mice that had received GM-CSF preconditioning, but in only 2 of 4 mice that did not receive GM-CSF (Fig. 7A). This indicates that in spite of antibody neutralization, functional reovirus can access tumors in vivo. These results are consistent with our previous data showing that i.v. administration of reovirus was not therapeutic in tumor-bearing mice unless the mice had been preconditioned with GM-CSF (8). Furthermore, administration of GM-CSF followed by reoNAb complexes delayed tumor growth and increased survival in tumor-bearing, reovirus-naïve mice (Fig. 7B and C). This therapeutic effect was less than we had previously seen following GM-CSF/reovirus treatment in reovirus-immunized mice (8) and suggests that systemic antireoviral NAb have an additional role in mediating therapy.

ReoNAb complexes formed using a reovirus-specific monoclonal antibody were delivered as efficiently as those generated using serum from reovirus-immunized mice (Fig. 7A), further supporting our hypothesis that this is an antibody-mediated process rather than being dependent on other serum factors. Moreover, delivery was enhanced by using a combination of monoclonal antibodies rather than a single one. This has implications for therapy as it suggests the possibility of improving therapeutic outcome by manipulation of the antibodies coating the reovirus.

Intravenous delivery of an OV represents not only an optimal means of accessing disseminated neoplastic tissue, but also a practical way of stimulating a systemic response from the immune system. However, this route of infusion is often eschewed in favor of more local methods given the many hurdles to viral persistence present in the vasculature, for example the presence of NAb. As seroprevalence for reovirus is common, in most individuals any i.v.-administered virus will encounter some NAb. A number of early-phase clinical trials have involved the administration of OV as a large i.v. bolus. Seen in the context of a preexisting immunity to the virus, these therapeutic infusions represent a reexposure to abundant viral antigens and result in a large-scale anamnestic response. This is characterized by the generation of virus-specific antibodies in circulation at high titer (15, 16), which is considered to preclude further therapeutic i.v. doses.

Our previous work, which focused on potentiating the delivery of reovirus to tumors by evasion of the antireoviral NAb response, uncovered a role for these antibodies in the therapeutic response (8). Here, we have further investigated the therapeutic potential of NAb, specifically in the form of reoNAb complexes where the virus is fully neutralized and unable to infect susceptible tumor cells. The source of antireoviral NAb was serum from patients on the REO13-BRAIN clinical trial. All patients had high antireovirus NAb, comprising IgG and IgA isotypes, both contributing to reovirus neutralization. There was no evidence for involvement of the complement system. This contradicts a recent study in which an inhibitor of the complement C3 molecule precluded reovirus neutralization in plasma (9). The basis for this disparity is unclear. We used a different strategy of disabling complement (HI vs. inhibitor) and output method (MTT assay vs. plaque assay), and used serum rather than anticoagulant-treated plasma, all of which could contribute to the difference in outcome.

We generated reoNAb complexes by incubating reovirus with a neutralizing volume of serum and confirmed their formation by EM. Reovirus neutralization was confirmed by incubating the reoNAb complexes with susceptible melanoma target cells; no cell death was observed, indicating that the virus was fully neutralized and unable to infect the cells. However, following loading onto isolated human monocytes, the reovirus within the complexes could be transferred to melanoma targets to induce target cell death. The mechanism by which the reoNAb complexes are processed by monocytes and transferred to tumor cells is currently the subject of further investigation in our laboratory but we have shown that it involves their internalization by the monocytes, this being partly dependent on FcγRIII. Nonclassical monocytes expressing FcγRIII form the minor subset of peripheral blood monocytes but we have demonstrated that, within a mixed population, their contribution to reoNAb transport is proportionally larger than that of classical monocytes. Nevertheless, there appears to be some contribution by classical monocytes, which may depend on an alternative mechanism of uptake. In contrast to human myeloid-derived dendritic cells, which support some viral replication, reovirus does not appear to replicate within freshly isolated human monocytes, indicating that viral amplification does not occur following internalization. The role of FcR in reoNAb transport suggests that NK cells and neutrophils, which express FcγRIII, may also play a role in reoNAb transport.

We examined the delivery of reoNAb complexes to other tumor cell lines and found that it is not restricted to melanoma, suggesting the applicability of our findings in influencing treatment design for patients with cancer in general. Furthermore, we have demonstrated that the phenomenon of reoNAb delivery is not reovirus-specific because CVA/NAb complexes are delivered to tumor cells by monocytes in a similar manner, although HSV/NAb are not, suggesting that specific aspects of virus physiology may determine applicability. It is unclear which aspects govern the delivery of OVs via NAb complexes but given our observations with reovirus, CVA and HSV1716, one possibility is the presence or absence of a viral envelope. However, a preexisting immune response improves the therapeutic efficacy of Newcastle Disease virus (Jacob Ricca, abstract O15, SITC 2016) and Maraba virus (17), suggesting a possible role for OV/NAb delivery via monocytes for both of these enveloped viruses and therefore we postulate that delivery of OV/NAb complexes might be restricted to small RNA viruses rather than those with DNA genomes.

Finally, we have demonstrated that following i.v. delivery of reoNAb complexes to tumor-bearing mice, functional reovirus can be retrieved from the tumors (Fig. 7), supporting our hypothesis that i.v. reovirus therapy in preimmunized mice results in the formation of reoNAb complexes in vivo which are then delivered to tumors via monocytes (8). Although we know that following i.v. delivery, reovirus is neutralized by NAb, this cannot be instantaneous and it is possible that transport of nonneutralized reovirus by monocytes was responsible for viral delivery to the tumors. Although we have not ruled out this possibility, we have demonstrated that neutralized reovirus in the form of reoNAb can be delivered in a functional form in vivo. Furthermore, tumor-bearing mice treated with GM-CSF followed by preformed reoNAb have delayed tumor growth and prolonged survival compared with controls, indicating that reoNAb have therapeutic potential. The therapeutic effect of reoNAb following GM-CSF preconditioning in naïve mice was less than we had previously shown using noncomplexed reovirus in reovirus-immunized mice. Thus, the enhanced therapeutic effect of a preexisting antireoviral immune response (8) can only partly be mediated by reoNAb complexes formed after i.v. reovirus treatment, and other immune mechanisms (e.g., antibody-dependent cell-mediated cytotoxicity or reovirus-specific CTL) must be involved. The data also suggest that reovirus therapy could be enhanced by manipulation of the antibodies bound to the virus. We found that although a single neutralizing monoclonal antibody was as effective as antireoviral serum in mediating delivery of functional virus to tumors, a combination of monoclonal antibodies was significantly more effective. This suggests the possibility of preformed reoNAb complexes as a therapeutic in which the antibodies are selected to provide the most efficient viral delivery to tumors.

Although this reactivation and release of antibody-neutralized virus by human monocytes may appear counterintuitive, there is some related evidence supporting our observations. First, dendritic cells release macropinocytosed antigen in a native unprocessed form from late endocytic compartments with stimulate B cells (18), indicating that not all internalized antigen is necessarily degraded by myeloid cells. With regard to FcR involvement, Ab-neutralized adenovirus has been found to mediate gene transfer via an FcR-dependent mechanism (19), though there was no viral release from the cells. The reports most closely related to our findings are of antibody-dependent enhancement (ADE) of infection. This occurs during infection with flaviviruses including dengue virus, whereby patients previously exposed to another dengue virus serotype form nonneutralizing-Ab-virus complexes which are taken up by FcR-expressing cells (including monocytes) resulting in enhanced virus infection (20, 21). ADE has also been reported for measles virus (22), another OV currently undergoing clinical trials. However, in contrast to our observations, ADE depends on the cross-reactivity of nonneutralizing antibodies, whereas our research highlights a hitherto unidentified role for NAb in mediating viral dissemination.

In conclusion, we have demonstrated that antibody-neutralized reovirus is internalized and processed by monocytes resulting in transfer of infectious virions that are able to infect and destroy tumor cells. Taken together with our previous data indicating the positive involvement of antiviral NAb (8), we suggest that this provides a rationale for exploiting antiviral NAb in OV therapy. Our results show that this approach is not specific to reovirus. Further research is needed to identify the factors that determine which OVs can be delivered in this manner.

M. Coffey is chief executive officer at and has ownership interest in Oncolytics Biotech Inc. R.C. Hoeben reports receiving a commercial research grant from Crucell. A. Melcher reports receiving a commercial research grant from Oncolytics Biotech Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R.A. Berkeley, A. Melcher, E.J. Ilett

Development of methodology: R.A. Berkeley, R.C. Hoeben, E.J. Ilett

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.A. Berkeley, L.P. Steele, A.A. Mulder, T.J. Kottke, J. Thompson, R.C. Hoeben, E.J. Ilett

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R.A. Berkeley, L.P. Steele, M. Coffey, E.J. Ilett

Writing, review, and/or revision of the manuscript: R.A. Berkeley, L.P. Steele, A.A. Mulder, M. Coffey, R.C. Hoeben, R.G. Vile, A. Melcher, E.J. Ilett

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.J.M. van den Wollenberg

Study supervision: E.J. Ilett

Other (contributed to data concerning the EM experiments, done in the facility at the LUMC in Leiden): D.J.M. van den Wollenberg

This work was supported by a University of Leeds PhD scholarship award (to R.A. Berkeley). Publication costs were covered by Oncolytics Biotech.

We thank Nik Matthews, Ritika Chauhan, and James Campbell for assistance with transcriptomic profiling and analysis.

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.