Retargeted and Stealth-Modified Oncolytic Measles Viruses for Systemic Cancer Therapy in Measles Immune Patients

Oncolytic virotherapy using engineered measles viruses (MV) derived from the Edmonston vaccine lineage is a promising experimental approach to the treatment of cancer (1). MV can be engineered to selectively infect, replicate in, and destroy tumor cells with minimal toxicity to normal cells (1, 2). Clinical trials using engineered MVs to treat a variety of human cancers (3) including nervous system (4, 5), hematologic (6–8), gynecologic (9, 10), and urothelial malignancies (11), have shown promising results.

A major barrier to the systemic deployment of MV as an anticancer agent is the high prevalence of anti-measles antibodies in the general population (6, 8, 12). Also, while dose-limiting toxicities have not yet been encountered in MV clinical trials (12), even with systemic MV doses as high as 10 (11) TCID₅₀, the concern remains that the systemic deployment of genetically armed MVs could be toxic to noncancerous cells expressing natural MV receptors, SLAMF1 and NECTIN4. We have, therefore, been working to engineer oncolytic MV whose receptor attachment and entry specificity are targeted to antigens overexpressed on the tumor cell surface, and which have been further “stealthed” to evade neutralization by preexisting anti-measles antibodies.

MV has two surface glycoproteins, the receptor-binding attachment protein, H, and its fusogenic partner, F, that together mediate virus entry and intercellular fusion between infected and uninfected cells expressing the native virus receptors. By a poorly understood mechanism, H receptor binding triggers the depolymerization of the F trimer leading to membrane fusion. We previously developed a versatile system to redirect the specificity of MV attachment and entry by fusing specificity determining polypeptide ligands to the carboxy terminus of H and mutating key residues to disrupt its interactions with the native receptors (13–15).

MVs retargeted in this way were shown to escape neutralization by mAbs targeting the receptor binding sites on H (16, 17), but remained highly susceptible to neutralization by polyclonal anti-measles neutralizing antibodies in measles-immune serum (16). Because all MV neutralizing antibodies target H or F (18, 19), we chose to substitute these surface glycoproteins with the corresponding proteins from a homologous morbillivirus.

Canine distemper virus (CDV) is a morbillivirus endemic in domestic dogs. CDV and MV surface glycoproteins are structurally similar, but serologically distinct. A proof-of-concept study was previously reported for the substitution of MV F and H proteins with homologous CDV glycoproteins to circumvent antibody-mediated neutralization of MV (20–22). However, the previously reported viruses were not specific for clinically relevant tumor-selective receptors (21) and were not blinded for native receptor use, so could still infect nontargeted human cells via the human NECTIN4 receptor (23).

To generate fully retargeted chimeric MVs capable of escaping neutralization by measles-immune human serum, we here substituted the MV F and H genes with homologous CDV F and CDV H genes mutated to destroy native receptor interactions and retargeted via C-terminally fused single-chain antibody fragments (scFv), specific for clinically relevant receptors, EGFR or CD38, that have been previously well-characterized for specificity in the context of retargeted MV (13). These retargeted viruses show the expected reprogramming of their receptor tropisms and are selectively oncolytic for receptor-positive tumors, even in mice passively immunized with measles-immune antiserum. Fully retargeted chimeric MVs that evade measles neutralizing antibodies may be suitable for clinical translation in patients with measles-immune cancer.

The African green monkey kidney cells (Vero), baby hamster kidney cells (BHK), human glioblastoma cell line (U87-MG), human embryonic kidney cells (HEK293), human Burkitt lymphoma Raji and Ramos cell lines, and Chinese hamster ovary (CHO) cells were purchased from ATCC and grown in the ATCC recommended media. ATCC authenticates cell lines using short tandem repeat DNA profiling. The modified cell lines: SKOV3ip.1-FLuc (24), CHO-NECTIN4 (25), CHO-αHIS (13), CHO-EGFR (14), CHO-CD38 (26), CHO-dogSLAMtag (27), and Vero-αHIS (13) were cultured as described previously. The SKOV3ip.1-FLuc was authenticated using short tandem repeat DNA profiling, all other modified cell lines were not further authenticated. All cells were grown at 37°C in a 5% CO₂ humidified incubator. All cell lines were tested 48 hours after thawing for Mycoplasma contamination using a Universal Mycoplasma Detection Kit (ATCC; 30-1012K) and all cell lines routinely tested negative. Cells were not maintained in tissue culture for more than 20 passages.

CDV H from the 5804P wild-type strain (25) or the previously generated SLAM Blind construct (26), was inserted between the PacI/SpeI restriction sites in the pCG vector or the PacI/SfiI restriction sites in the pTN vector bearing the respective scFvs (13) by ligation cloning to generate pCG CDV H, pTN CDV H -CD38, pTN CDV H_RB-CD38, and pTN CDV H_RB-EGFR expression plasmids (Fig. 1A) The pCG CDV F plasmid was generated by excising out the MV F gene in pCG-MV F with HpaI and PacI and ligating an in-frame CDV F DNA coding sequence PCR amplified from a CDV field isolate. Recombinant viruses were generated by replacing MV F and H genes in the MV Moraten vaccine antigenome [ref. 22; with an additional transcription unit after the P gene encoding for enhanced GFP (eGFP) to facilitate visualization of virus during rescue] with the NarI/PacI CDV F gene fragment (from pCG-CDVF) and the PacI/SpeI CDVH_RB EGFR or CDVH_RB-CD38 fragments (from pTN CDVH_RB EGFR and pTN CDVH_RB-CD38, respectively), in compliance with the rule that recombinant MV genomes are hexametric. The sequences of all generated plasmids were verified by Sanger sequencing. Viruses were rescued as described previously (22) and grown on Vero-αHis cells at 32°C.

Approval to generate MV that can evade neutralization was obtained from the Mayo Clinic Institutional Biosafety Committee (Rochester, MN) before initiating this project.

The pooled human immune-measles antibody serum used for all experiments was obtained from Valley Biomedical Inc. (Lot #C80553). This serum has anti-MV IgG of 300 EU/mL and titer determined to be 256 by plaque reduction neutralization titer (PRN; ref. 24). Serum dilutions were made using Opti-MEM or cell culture medium.

Cells (1 × 10⁴) seeded in each well of a 96-well Tissue Culture Plate (Corning) were transfected the next day with 50 ng each of pCG CDV F and the indicated engineered CDV H plasmid using FugeneHD (Promega) according to the manufacturer's instructions. Forty-eight hours after transfection, cells were washed once with PBS, stained with 1% (w/v) crystal violet solution, and imaged with a light microscope at 100× magnification.

To assess sensitivity of intercellular fusion to measles-immune human serum, 1 × 10⁴ CHO-CD38 cells in each well of a 96-well plate were cotransfected the next day with 50 ng each of either the pCG CDV F, pTN CDV H_RB-CD38, and pCDNA3.1-eGFP, or the pCG-MV F, pTN MV HaalsCD38, and pCDNA3.1 GFP using SuperFect (Qiagen) according to the manufacturer's instructions. Two hours after transfection, the transfection medium was replaced with culture medium containing 2-fold serial dilutions of heat-inactivated pooled measles-immune human antibody serum. Photographs were taken 48 hours after transfection under a fluorescence microscope at 100× magnification.

To determine expression of engineered CDV H proteins, HEK293 cells (2 × 10⁶/well in 6-well plate) were transfected with 2 μg of the corresponding CDV H plasmids. Forty-eight hours later, cells were washed with PBS, lysed with RIPA buffer for 30 minutes at 4°C, and centrifuged at 13,000 rpm for 20 minutes at 4°C to clarify the lysate. Lysate (10 μL) was mixed with an equal volume of 2× Laemmli Buffer (Bio-Rad), boiled at 95°C for 5 minutes, fractionated on a 7.5% Tris-HCl gel, and transferred to a polyvinylidene difluoride (PVDF) membrane. Immunoblotting to detect the CDV H was performed as described previously (22) using the following antibody dilutions: rabbit anti-CDV H cyt antibody (MC712, 1:5,000; ref. 21) kindly provided by Dr. Roberto Cattaneo (Mayo Clinic, Rochester, MN), horseradish peroxidase (HRP)-conjugated goat anti-rabbit (Thermo Fisher Scientific; 1:10,000), and HRP-conjugated mouse anti-B-actin (Sigma-Aldrich, A3854; 1:25,000).

For immunoblot analysis of chimeric viruses, 2 × 10⁵ TCID₅₀ of each virus stock was directly mixed with an equal volume of denaturing loading buffer, boiled at 95°C for 5 minutes, and fractionated on a 7.5% gel. Proteins were transferred to a PVDF membrane and immunoblotted in the standard fashion using primary antibodies: a rabbit antibody against MV N (N12) and rabbit antibodies directed against the cytoplasmic tails of CDV H (MC712; 1:2,000; ref. 21) or MV H (1:2,000; ref. 21; all kind gifts of Dr. Roberto Cattaneo); and an HRP-conjugated goat anti-rabbit (1:10,000) secondary antibody as described previously (22).

Vero-αHis cells (2 × 10⁵) in each well of a 6-well plate were infected with the respective viruses at an multiplicity of infection (MOI) of 0.03 for 2 hours in 1 mL Opti-MEM at 37°C. After which, the infection media were replaced with 2 mL/well of fresh growth media and the cells incubated at 37°C. Cells were harvested at the indicated times by scraping into 1 mL of growth media, subjected to three freeze–thaw cycles, centrifuged at 1,200 rpm for 10 minutes at 4°C, and viral titers in the supernatant determined on Vero-αHis. The experiment was repeated three times.

The surface expression of EGFR and CD38 on human tumor cell lines was performed as described previously (13) using an Alexa Fluor647–conjugated anti-human EGFR (matuzumab) mAb (R&D Systems; FAB10023R) and FITC-conjugated Mouse anti-Human CD38 (BD Biosciences; catalog no., 555459), respectively.

Adherent cells (10⁴) or 10⁵ suspension cells were infected in one well of a 96-well plate at an MOI of 0.5 (titers determined on Vero-αHis cells) in 50 μL Opti-MEM for 3 hours at 37°C before the addition of complete growth medium. Forty-eight hours after infection, cells were examined and photographed under a fluorescence microscope at 100× magnification.

For the neutralization assay, 30 infectious virus units of each virus were incubated with serial dilutions of heat-inactivated pooled antibody human serum (Valley Biomedical, Inc. Lot #C80553) in Opti-MEM at 37°C in quadruplicates for an hour. The mix was then used to infect a 90% confluent monolayer of Vero-αHis cells in a 96-well plate for 2 hours at 37°C. After the infection, fresh culture medium was added, and the cells were cultured at 37°C. Seventy-two hours later, wells were observed for the appearance of viral cytopathic effect (CPE). The reciprocal of the dilution at which there was no viral CPE in all four wells for each virus was reported as the neutralization titer. The experiment was repeated two times.

For in vitro cytotoxicity, 1 × 10⁴ SKOV3ip.1-Fluc cells in each well of 96-well plate were infected with the respective viruses in quadruplicates in Opti-MEM media for 3 hours at an MOI of 10 and cultured at 37°C in culture medium. Seventy-two hours later, cytotoxicity was determined by MTT Assay (Promega) according to the manufacturer's instructions.

All animal experiments were approved by the Mayo Clinic Institutional Animal Care and Use Committee (IACUC) and performed in accordance with IACUC-established guidelines. For all experiments, we used female 6- to 8-week-old athymic nude mice (Taconic).

To assess specificity of infection in vivo, 5 × 10⁶ cells in 200 μL PBS were implanted intraperitoneally, followed 10 days later by six intraperitoneal virus treatments of 2 × 10⁶ TCID₅₀ (200 μL) MV^CDVenv-EGFR or MV^CDVenv-CD38 (n = 4/group) or control treatment (200 μL of Vero-αHis cells lysate; n = 3) every other day. To compare efficacy of MV^EGFR with MV^CDVenvEGFR, 5 × 10⁶ cells in 200 μL PBS were implanted intraperitoneally, followed 10 days later by a single treatment of 200 μL control (Vero-αHis cells lysate) or 2 × 10⁶ TCID₅₀ of MV^EGFR or MV^CDVenvEGFR (n = 5/group). For the immune evasion studies, 2.5 × 10⁶ SKOV3ip.1-Fluc cells were implanted into the peritoneal cavity of athymic nude mice (n = 10/group), followed 6 days later by passive immunization of animals in the serum group as described previously (24). For these experiments, we used 200 μL of pooled human measles-immune serum (Valley Biomedical, Inc; Lot #C80553). Three hours following passive immunization, animals were treated with a single intraperitoneal injection of 200 μL control treatment (Vero-αHis cells lysate) or 2 × 10⁶ TCID₅₀ of MV^EGFR or MV^CDVenvEGFR. Tumor burden was monitored by weekly Luminescence Imaging (IVIS Xenogen; Perkins) and total body luminescence was quantified from these images using the Live Imaging Software (IVIS Xenogen; Perkins) according to the manufacturer's protocol. Animals were euthanized when they developed ascites. For the subcutaneous model, 5 × 10⁶ SKOV3ip.1-Fluc cells in 100 μL PBS were implanted subcutaneously in the right flank and treated with six intratumoral injections of 100 μL control treatment (Vero-αHis cells lysate; n = 3) or 1 × 10⁶ TCID₅₀ of MV^EGFR or MV^CDVenvEGFR (n = 4/group) every other day. Tumor volume was monitored by caliper measurements and animals were sacrificed if they met euthanasia criteria. Tumor volume was calculated by using the formula: 0.5 × length × width × width.

Statistical analyses and calculations were performed with the GraphPad Prism Software (GraphPad). We used the Kaplan–Maier method to graph the survival of treated animals, and the log-rank test to compare the differences in the survival curves. Differences in subcutaneous tumor volumes and total body luminescence were assessed by two-way ANOVA with Turkey multiple comparison test to correct for multiple comparisons. Unpaired Student t test was used to compare differences between two groups. A P < 0.05 was considered to be statistically significant for all analyses.

Because the vaccine strains of CDV have evolved to use other uncharacterized receptor(s) in addition to the known CDV receptors (28), we focused our engineering efforts on a wild-type CDV H. The CDV H coding sequence of the wild-type strain (5804P; ref. 29) or the previously engineered receptor blind variant (CDV H _RB; ref. 30) were cloned as shown in Fig. 1A into an expression vector with or without C-terminally fused scFvs targeting EGFR or CD38 to generate expression plasmids for CDV H, CDV H-CD38, CDV H_RB-CD38, and CDV H_RB-EGFR. These plasmids were transiently transfected into human embryonic kidney (HEK293) cells and the lysates were assessed for protein expression by Western blot analysis using an antibody directed against the cytoplasmic tail of CDV H protein (21). The immunoblot in Fig. 1B shows that all proteins were equally expressed and the scFv-tagged proteins (CDV H-CD38, CDV H_RB-CD38, and CDV H_RB-EGFR) had the expected approximately 25 kDa increase in molecular weight compared with the unmodified CDV H protein.

Next, to assess receptor specificity of the engineered CDV H proteins, we cotransfected a panel of CHO cells expressing the relevant receptors with each of the CDV H expression plasmids in combination with a CDV F expression plasmid. The readout for these experiments was syncytia formation resulting from intercellular fusion. As expected, we observed syncytia formation with the wild-type CDV H protein only in CHO cells expressing native CDV receptors canine SLAM or NECTIN4, and not in the parental CHO, CHO-EGFR, or CHO-CD38 cell lines (Fig. 1C). In contrast, the six mutations introduced into the CDV H receptor binding sites selectively eliminated native receptor–dependent fusion without impairing retargeted fusion via displayed ligands (compare the CDV H-CD38 row with the CDV H_RB-CD38 or CDV H_RB-EGFR rows). After confirming full retargeting of CDV H, we next evaluated the impact of measles neutralizing antibodies on the intercellular fusion mediated by the engineered CDV envelope. As shown in Fig. 1E, intercellular fusion mediated by the retargeted CDV H-CD38 was resistant to neutralization by pooled measles-immune human serum even with serum dilutions as low as 1:10. In contrast, intercellular fusion mediated by MV H-CD38 cotransfected with the MV F protein in CHO-CD38 cells was strongly inhibited even at high serum dilutions.

Taken together, these data demonstrate that wild-type CDV H can be fully retargeted and that intercellular fusion mediated by retargeted CDV H in concert with CDV F is resistant to neutralization by antibodies present in measles-immune human serum.

To generate fully retargeted chimeric MV bearing retargeted CDV envelopes, we substituted the MV F and H genes into the MV Moraten vaccine (vac2) genome (with an eGFP cistron inserted upstream of the P gene; ref. 22) with the respective CDV F and CDV H_RB-EGFR or CDV H_RB-CD38 protein-encoding genes. The resulting viruses were named MV^CDVenv-EGFR and MV^CDVenv-CD38, respectively. As a control, we also generated MV^EGFR, in which the MV H gene was replaced by a previously engineered sequence encoding a native, receptor blind MV H protein displaying a C-terminal scFv against EGFR (Haals-EGFR; Fig. 2A). All three viruses were rescued and propagated using the six-histidine tag rescue (STAR) system (13) and the incorporation of chimeric envelopes was confirmed by immunoblot analysis (Fig. 2B). All the retargeted viruses exhibited slower growth kinetics compared with unmodified MV-Moraten, but compared with one another had similar growth kinetics on the Vero-αHis cell line (Fig. 2C) and yielded stocks with similar maximum titers of 3 × 10⁸ TCID₅₀/mL for MV^EGFR, 2 × 10⁸ TCID₅₀/mL for MV^CDVenv-EGFR, and 6 × 10⁷ TCID₅₀/mL for MV^CDVenv-CD38. In vitro neutralization assays showed that the viruses bearing the retargeted CDV envelopes maintained their infectivity even in the presence of a high concentration (<1:10 dilution) of pooled measles-immune human serum. In contrast, the virus bearing the retargeted MV envelope was efficiently neutralized at serum dilutions as high as 1:320 (Fig. 2D). Collectively, these data demonstrate that chimeric MVs with the retargeted CDV envelope can be generated and propagated with the STAR pseudoreceptor system and that these viruses are resistant to neutralization in vitro by measles-immune serum.

To assess the receptor specificity of the chimeric-retargeted viruses, we infected the panel of CHO cells expressing the relevant receptors and monitored for GFP-expressing syncytia 48 hours after virus exposure as readout for infection. As shown in Fig 3A, all chimeric viruses exhibited the expected receptor specificities. None infected the parental CHO cells, whereas all viruses infected the CHO-αHis cell line via the interaction between their displayed His tag and the αHis pseudoreceptor. MV^EGFR and MV^CDVenv-EGFR specifically infected CHO-EGFR cells with no discernable infection in CHO-CD38 cells, while MV^CDVenv-CD38 specifically infected the CHO-CD38 cells without infecting the CHO-EGFR cells. None of the viruses infected CHO cells expressing native CDV receptors, SLAM or NECTIN4.

Entry specificity was further tested in a panel of four human tumor cell lines known to express either EGFR or CD38: the SKOV3ip.1 ovarian cancer and U87-MG brain tumor cell lines are positive for EGFR and negative for CD38, whereas the Raji and Ramos Burkitt lymphoma cell lines are positive for CD38 and negative for EGFR, as confirmed by flow cytometric staining (Fig. 3B). As expected, MV^EGFR and MV^CDVenv-EGFR infected only the EGFR-positive tumor cell lines, while MV^CDVenv-CD38 infected only the CD38-positive tumor cell lines (Fig. 3C). Receptor-dependent killing of SKOV3ip.1 tumor cells by the retargeted viruses was evaluated using an MTT assay as readout for cell viability 72 hours after infection. Consistent with the cell infectivity data, only the EGFR-retargeted viruses significantly diminished cell viability (Fig. 3D), indicating that EGFR-dependent entry and intercellular fusion translates to receptor-dependent cell killing.

To evaluate the in vivo specificity and oncolytic activity of the retargeted chimeric viruses, we utilized the EGFR-overexpressing SKOV3ip.1-FLuc orthotopic ovarian cancer xenograft tumor model (24, 31, 32). SKOV3ip.1-Fluc cells express luciferase, which allows for noninvasive monitoring of tumor burden by Bioluminescence (IVIS Xenogen) Imaging (BLI). To assess for specificity of infection and oncolytic efficacy in vivo, SKOV3ip.1Fluc cells were implanted into the peritoneal cavity in athymic mice to develop a model of disseminated carcinomatosis peritonei, followed 10 days later by six intraperitoneal treatments of 2 × 10⁶ TCID₅₀ of MV^CDVenv-EGFR, MV^CDVenv-CD38, or an equivalent volume of control cell lysate (from the producer Vero-αHis cell line), given every other day. Figure 4A shows that only the MV^CDVenv-EGFR virus was able to control tumor growth, as demonstrated by a marked reduction in the tumor luminescence signal in this treatment group compared with the control treatment groups. Control of tumor growth was associated with significant prolongation in survival of animals in the MV^CDVenv-EGFR treatment group compared with the MV^CDVenv-CD38 or control groups (median survival of 54 vs. 30 days; P = 0.01; Fig. 4B), indicating that the specificity of infection observed in vitro for the MV^CDVenv-EGFR translates to specific oncolytic efficacy in vivo.

To directly compare the oncolytic activities of the MV^CDVenv-EGFR and MV^EGFR viruses, we used both subcutaneous and intraperitoneal SKOV3ip.1-Fluc tumor models. In the subcutaneous model, we assessed the effect of six intratumoral treatments of 1 × 10⁶ TCID₅₀ of virus versus control cell lysate administered every other day starting 10 days after tumor implantation. The tumor volumes of subcutaneous tumors assessed by caliper measurements, shown in Fig. 5A, demonstrate that MV^CDVenv-EGFR and MV^EGFR were equally effective in this model, significantly impacting tumor volumes compared with control (P = 0.0019 for MV^EGFR and P < 0.0001 compared with control treatment). Tumor volumes (mean ± SD) were 27.5 ± 9.1 mm³, 23.4 ± 11. 8 mm³, and 20.9 ± 2.5 mm³ before virus injection and 1,346.7 ± 527.6 mm³, 67.3 ± 76.6 mm³, and 9.8 ± 19.3 mm³, 30 days after the first virus treatment for control, MV^EGFR, and MV^CDVenv-EGFR, respectively. There was no statistical significance in tumor growth curves between the two virus treatments (P = 0.2234). Next, in the orthotopic SKO3ip.1-Fluc model, we compared the oncolytic activity of a single intraperitoneal virus treatment (2 × 10⁶ TCID₅₀) versus control therapy 10 days after tumor cell implantation. Tumor burden was monitored weekly by luciferase imaging. As shown in Fig. 5B, the luciferase signal (i.e., tumor burden) increased more rapidly in the animals in the control group versus those in either virus treatment group. There was a statistically significant decrease in the whole-body luminescence of animals in the treatment groups compared with control-treated animals on day 14 (P = 0.0473 for MV^EGFR and P = 0.0324 for MV^CDVenv-EGFR compared with control), with no statistically significant difference in signal intensities between the two treatment groups on day 14 (P > 0.9999; Fig. 5C). This decrease in luminescence correlated with a statistically significant increase in median survival from 18 days for the control treatment to 45 days for the MV^EGFR and 53 days for MV^CDVenv-EGFR treatment groups (P = 0.0063 and P = 0.0015, respectively), with no difference in survival between the two treatment groups (P = 0.6217; Fig. 5D). Overall, these data show that MV^CDVenv-EGFR is equipotent compared with the previously reported MV^EGFR in an EGFR-overexpressing tumor model.

To assess the impact of measles neutralizing antibodies on the oncolytic activity of MV^EGFR and MV^CDVenv-EGFR, athymic mice with established intraperitoneal SKOV3ip.1-Fluc tumor xenografts were passively immunized with measles-immune human antibody serum (60 EU/mouse) and treated 3 hours later with a single intraperitoneal virus dose of 2 × 10⁶ TCID₅₀ (Fig. 6A). Tumor growth was monitored by serial BLI as shown in Fig. 6B with the corresponding quantification of total body luminescence graphed in Fig. 6C. As expected, both MV^EGFR and MV^CDVenv-EGFR reduced tumor burden (P < 0.0001) and prolonged the survival (P < 0.0001) of treated animals compared with control treatment in the absence of measles neutralizing antibodies. However, as anticipated, the efficacy of the MV^EGFR was severely diminished in the presence of measles neutralizing antibodies, resulting in a statistically significant increase in the luminescence signal (P < 0.0001) and corresponding decrease in survival (P < 0.0018) for passively immunized mice compared with naïve mice treated with MV^EGFR. In contrast, the oncolytic activity of MV^CDVenv-EGFR was maintained in the presence of neutralizing measles antibodies, with no statistically significant difference in tumor luminescence (P = 0.7083) or overall survival (P = 0.3381) for passively immunized mice compared with nonimmunized control mice treated with MV^CDVenv-EGFR. Together, these results demonstrate that MV^CDVenv-EGFR retains full antitumor activity in vivo even in the presence of a high body fluid concentration of MV neutralizing antibodies.

Here, we substituted the MV F and H genes with CDV F and CDV H retargeted to EGFR or CD38, and thereby generated fully retargeted oncolytic MVs capable of escaping neutralization by preexisting measles antibodies. MV^CDVenv-EGFR was potently oncolytic in athymic mice bearing aggressive, intraperitoneally disseminated, EGFR-overexpressing, SKOV3ip.1 human ovarian cancer tumors, even after the mice had been passively immunized against MV using pooled human serum.

Ovarian cancer, the second most common cancer of the female genitourinary tract, is a lethal disease accounting for an estimated 13,980 number of deaths in the United States in 2019 (33). Intraperitoneal EGFR-targeted MV virotherapy is an interesting possibility for the treatment of this disease. Ovarian cancer often remains localized to the peritoneal cavity even in patients with relapsed and treatment refractory disease, and we, therefore, have been developing MV as a potential intraperitoneal therapy for this malignancy (9, 10, 24, 31). A recently completed phase I clinical trial of intraperitoneally administered MV doubled the median overall survival of heavily pretreated, platinum-resistant ovarian cancer patients compared with historical controls (10), despite the fact that the enrolled patients had high titers of measles neutralizing antibodies in their blood and peritoneal fluid (10). Our preclinical studies have underscored the importance of circumventing antibody neutralization of MV to maximize the antitumor potency of the approach and, with this goal in mind, we are clinically evaluating the intraperitoneal administration of ex vivo–infected autologous mesenchymal stem cell carriers as an antibody-resistant virus delivery system (24).

In this study, we have pursued an alternative approach to circumvent antibody neutralization, using envelope glycoprotein exchange in conjunction with tropism engineering via surface display of single-chain antibodies (scFv). This study builds on previous reports of chimeric MV bearing CDV envelope glycoproteins (20–22). However, in contrast to previous studies, our current approach has clinical translational relevance because we have targeted the virus to EGFR, a clinically relevant ovarian cancer marker (34) also expressed at high levels in several other cancers (35, 36), and to CD38, a clinically relevant target in multiple myeloma (37). The scFvs we displayed in this study are derived from mAbs, EGFR (matuzumab) and CD38 (daratumumab), that have already been well-characterized for specificity and toxicity in humans, as well as in the context of retargeted MV (13). We elected here to substitute measles H with a wild-type CDV H (5804P) because of its narrower range of receptor use compared with the H proteins from CDV vaccine strains (28). While the mutations (D526A, I527S, S528A, R529A, Y547A, and T548A) we introduced into the CDV H protein were initially reported to ablate only SLAM-dependent fusion (30), we have confirmed here (Fig. 1C), as others have shown (38), that these mutations also ablate NECTIN4-dependent fusion making our vector completely “blind” to native pathogenicity-determining CDV receptors. Given the similarity of the CDV and MV envelopes, we anticipate that there will be little limitation to the diversity of tumor-associated surface antigens that can be targeted using these chimeric viruses (15).

MV does not replicate efficiently in murine cells (39), because of, as yet, undefined intracellular barriers to the viral replication cycle, independent of receptor expression. Thus, we did not test our viruses in syngeneic immunocompetent tumor models. Instead, we used the extensively characterized intraperitoneal SKOV3ip.1-Fluc tumor model in athymic mice, in which we could adoptively transfer measles-immune human serum. Therefore, we did not assess the potential contributions of the other compartments of the immune system to the antitumor efficacy of this new vector. However, data previously generated in the SKOV3ip.1 model were deemed sufficiently informative to justify initial clinical testing of MV in ovarian cancer, and the measles antibody titers achieved by passive immunization in our efficacy studies are comparable with the ascites concentrations present in patients with ovarian cancer (24). Clinical testing of the viruses described in this article may therefore be warranted.

It is worth noting that despite the high doses of virus used in our ovarian cancer studies, neither MV^CDVenv-EGFR or MV^EGFR therapy was curative. Treated animals eventually succumbed to progressive intraperitoneal disease and were typically euthanized because of ascites formation. Examination of explanted peritoneal tumors from virus-treated animals under fluorescence light showed strong expression of GFP indicating that these tumors were persistently infected with the virus. The presence of an intact T-cell immune response might facilitate clearance of these virus-infected cells, and this will be tested in future adoptive T-cell transfer experiments, which may also facilitate the evaluation of immunotherapy combination approaches such as MV virotherapy plus checkpoint inhibition to achieve more durable cures (40). While this study focused exclusively on the SKOV3ip.1 ovarian cancer model, we do plan to investigate the systemic administration of MV^CDVenv-EGFR in other EGFR-overexpressing tumors such as lung cancers (35) and glioblastoma (36). Furthermore, MV^CDVenv-CD38 will be evaluated in CD38-overexpressing models of multiple myeloma (26)

A critical parameter requiring further optimization for the effective clinical translation of the stealthed and retargeted MVs described in this article is to improve the virus manufacturing yield. While we were eventually able to generate viral stocks with titers as high as 10⁸ TCID₅₀/mL of the fully retargeted chimeric viruses for our in vivo animal experiments, retargeted MVs generally grow to much lower titers compared with MVs with native envelope glycoproteins, and are more challenging to propagate. Following their initial rescue, the viruses propagate very slowly, until they have been through several tissue culture passages. This could be related to the use of an artificial receptor (the αHis pseudoreceptor) on the cells that are used for virus amplification, or to altered kinetics of virus maturation in infected cells. Significant further optimization of virus yield may be achievable using alternative cell substrates (41), alternative receptors for virus propagation, or by virus engineering, perhaps through cytoplasmic tail truncations of the surface glycoproteins (42) or substitution of the M gene (20, 43). An alternative approach to bypass the MV manufacturing difficulty might be to incorporate the targeted CDV envelope glycoprotein complexes into other viral platforms such as adenovirus (14) or VSV (44, 45) that are easier to manufacture at high titer.

By design, we used the Edmonston lineage measles vaccine backbone with a remarkable safety record both as a vaccine and oncolytic agent in clinical trials to construct our viruses. The genome of this backbone has remained stable, with no reversions to pathogenicity reported to date, in the past 50+ years since routine childhood measles vaccination was introduced. Because there are more than 40 different attenuating mutations dispersed all over the genome, of which mutations in the innate immune antagonizing P/V/C gene are the most important (46), we retained all of the core MV proteins, to ensure that our viruses would be safe and not cause disease in humans. Moreover, these core proteins harbor T-cell epitopes that could enhance the antitumor effect of MV by eliminating residual virus-infected tumor cells (47). While wild-type CDV can cause severe disease in dogs (48) and has been reported capable of pathogenicity in other species including nonhuman primates (49), the viruses described in this article do not use any of the natural virus receptors responsible for CDV pathogenicity (48) and are fully retargeted to receptors that are not compatible with normal virus biology (50). Moreover, most pet dogs are vaccinated against CDV. Of course, additional extensive formal toxicology studies in consultation with the FDA will be needed before any human clinical trials can be planned.

In summary, envelope glycoprotein exchange between MV and CDV allowed us to generate chimeric MVs retargeted to EGFR or CD38 that were resistant to neutralization by anti-measles antibodies in serum from measles immune human subjects. As such, these viruses may warrant clinical advancement for the treatment of measles immune patients with disseminated malignancies overexpressing EGFR or CD38. Furthermore, the platform might be suitable for generating additional “stealthed” and retargeted MV tailored for the treatment of a variety of human cancers.

K.W. Peng reports being a cofounder and officer of Vyriad, outside the submitted work, as well as technology licensed to Vyriad. M.Á. Muñoz-Alía reports lecture fees paid by a commercial entity (honoraria). S.J. Russell reports being a cofounder and officer of Vyriad, and Mayo Clinic holds equity in Vyriad. No potential conflicts of interest were disclosed by the other authors.

E.S. Bah: Conceptualization, data curation, formal analysis, investigation, methodology, writing-original draft, writing-review and editing. R.A. Nace: Resources, investigation, methodology, writing-review and editing. K.W. Peng: Resources, writing-review and editing. M.Á. Muñoz-Alía: Conceptualization, data curation, supervision, investigation, methodology, writing-review and editing. S.J.Russell: Conceptualization, resources, supervision, funding acquisition, investigation, methodology, writing-original draft, project administration, writing-review and editing.

We are grateful to Dr. Roberto Cattaneo (Mayo Clinic) for the anti-CDVH cyt, anti-MV H cyt, and Anti MV-N antibodies used for Western blot analysis. We also thank the Mayo Clinic Microscopy and Cell Analysis Core for experimental and technical support. This work was funded by the Mayo Foundation. E.S. Bah was supported by the Mayo Clinic Medical Scientist Training Program grant (T32GM065841) and Initiative for maximizing student diversity grant (R25GM055252) from the NIH.

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.