Chimeric oncolytic Ad5/3 virus replicates and lyses ovarian cancer cells through desmoglein‐2 cell entry receptor

Abstract Despite new therapies, the estimated 229 875 women living with ovarian cancer have a 5‐year survival rate of 47.6%. This cavity‐localized cancer lends itself to local administration of modalities, such as the oncolytic adenovirus (Ad) Ad5/3‐D24‐granulocyte‐macrophage colony‐stimulating factor virus (ONCOS‐102). Its repeated administration to a patient with chemotherapy‐refractory ovarian cancer induced CD8+ antitumor immune responses with the overall survival reaching 40 months. Here we probe the dominant receptor used by ONCOS‐102 in four established epithelial ovarian cancer cell lines. Ad3 can use the desmoglein‐2 (DSG2) and CD46 receptors on susceptible cells. DSG2 was nearly absent in A2780 cells but was expressed in more than 90% of OAW42, OVCAR3, and OV‐90 cells. After 96 hours, ONCOS‐102 treatment showed significant oncolytic activity (≧50%) in OAW42, OVCAR3, and OV‐90 cells, but minimal activity in A2780 cells, suggesting DSG2 as the dominant receptor for ONCOS‐102. Furthermore, retrospective analyses of phase I clinical trial of ONCOS‐102 treatment of 12 patients with varied tumors indicated a correlation between viral genomes in blood and DSG2 RNA expression. These data support the role of DSG2 expression on cancer cells in virus infectivity and the continued development of ONCOS‐102 for ovarian cancer treatment.


| INTRODUCTION
Ovarian cancer is diagnosed at an advanced stage in most patients (59%), 1 3 This cavity-localized cancer lends itself to local administration of modalities, such as oncolytic adenoviruses, that is, adenoviruses that replicate in and lyse tumor cells but not normal cells. [4][5][6][7][8][9][10][11][12][13][14] For example, nine administrations of an oncolytic adenovirus, ONCOS-102, into a patient with chemotherapy-refractory ovarian cancer in a phase I trial induced and progressively enhanced CD8+ infiltration in the treated tumor lesions. 15 Systemic CD8+ T cell responses against several tumor antigens were detected from 8 to 113 days after treatment initiation. 15 Herein, we investigate the suscept- ranging from 30% to 99% of cells, 16 many groups have pursued the development of alternate serotypes of adenoviral vectors such as Ad3, or they have engineered chimeric Ad5 fiber knob proteins to bind to an alternate receptor. 5,17,18 CD46 is considered a receptor for Ad3, Ad7, Ad11, and Ad35 19 ; it functions as a complement regulatory protein.
Most epithelial ovarian cancer (EOC) samples from a primary laparotomy and secondary cytoreduction procedures stained positive for CD46 (60% and 70%, respectively). 20 CD46 was highly expressed in 100% of primary EOC cancer lines (5 of 5). 21 Desmoglein-2 (DSG2) is also a receptor for Ad3, Ad7, and Ad11. 22 DSG2 is overexpressed in many types of ovarian cancers. 23 We and others have engineered Ad5 with chimeric Ad5/3 fiber knobs to target epithelial cancers 18,24 that overexpress the DSG2 and/or the CD46 receptors.
ONCOS-102 has three modifications that can contribute to its safety and its efficacy against ovarian cancer. 18 Its chimeric Ad5/3 fiber knob changes the binding specificity of the virus: instead of binding to the CAR, this chimeric Ad5/3 adenovirus targets the frequently overexpressed membrane proteins DSG2 and CD46. The replication of ONCOS-102 is restricted to tumor cells with an altered Rb pathway by its 24 bp deletion in the E1A gene. Its expression of granulocytemacrophage colony-stimulating factor (GM-CSF) can augment the immunostimulatory milieu in the infected tumor. 5,15,[25][26][27] Here, we investigated the oncolytic activity of ONCOS-102 in vitro in four ovarian adenocarcinoma cell lines that differ in expression of DSG2 but have a similar expression of CD46 or CAR to explore the prominent receptor for transduction. To assess the role of expression of the DSG2 and CD46 receptors on ONCOS-102 treatment of patients with solid tumors from a previously reported phase I trial, 28 (Table S1). Cells were passed at 80% of the confluence with trypsin-EDTA and the doubling times ranged from 1 day to 7 days: A2780, 1 day; OAW42, 2 days; OVCAR3, 3 days; OV-90, 7 days.

| ONCOS-102 preparation and treatment
Adenovirus ONCOS-102 is a class II genetically modified microorganism. The engineering of ONCOS-102 has been described previously. 18 ONCOS-102 was produced and stored at −80°C, as previously described. 29,30 Briefly, the concentration of total viral particles (VP) was assessed by measurements with UV/Vis spectrophotometry at 260 and 280 nm. The VP was calculated with the formula: OD 260 reading × dilution factor × 1.1 × 10 12 particles = number of particles per mL of sample.

| Cell viability
Oncolytic efficacy was determined with the MTS cell viability assay

| Retrospective analysis of samples from ONCOS-102 treated patients (NCT01598129)
Retrospective analyses were performed with previously obtained data from patients who had enrolled and participated in the escalating dose phase I clinical trial, NCT01598129. 28 Briefly, 12 patients with various types of solid tumors had been injected intratumorly and intravenously with ONCOS-102 on days 1, 4, 8, 15, 29, 57, 85, 113, and 141. 28 The low, medium, and high dose groups had received 3 × 10 10 , 1 × 10 11 , and 3 × 10 11 VP/injection, respectively, at each time point. 28 The tumor samples had been harvested at baseline, 1 month, and 2 months postinitiation of ONCOS-102 treatment. 28 Blood samples had been collected before each treatment, 6 and 24 hours after each treatment, then processed, and archived. 28 DSG2 and CD46 RNA expression levels of the tumor samples had been obtained by microarray but had not been previously reported. 28 The quantity of ONCOS-102 viral genomes in the blood samples had been measured by real-time polymerase chain reaction and previously reported. The different leukocyte populations in the tumor biopsies, called TILs, had been determined by immunohistochemistry performed on formalin-fixed and paraffinembedded tissues as previously described and reported. 28

| Data analysis
All variables were analyzed by using GraphPad Prism (v8) software. The correlation was calculated using the nonparametric Spearman test (twotailed, 95% confidence interval). Statistical significance for in vitro studies was performed by two-way analysis of variance.

| Receptor expression
Both A2780 and OAW42 ovarian carcinoma cell lines are categorized as nonserous adenocarcinoma cell lines and OVCAR3 as high-grade serous by genomic profiles. 31 OV-90 is classified as undifferentiated adenocarcinoma. These four ovarian carcinoma cell lines highly expressed CD46, a major receptor for the Ad3 serotype ( Figure 1A). DSG2 was expressed in 1.8% of A2780 cells, whereas other cell lines had a significantly higher expression of 97%, 96%, and 96% for OAW42, OVCAR3, and OV-90, respectively. Presence of CAR adenovirus receptor on the cell surface was observed in 53% ± 19% of A2780 cells, 58% ± 18% of OAW42, 76% ± 11% of OVCAR3, and 56% ± 8% of OV-90 cells.    Table 1). In contrast, ONCOS-102 treatment did not reduce the viability of the A2780 cells at 72 hours ( Figure 1B and Table 1).
Interestingly, DSG2 RNA expression levels in the tumor tissue from the 12 patients were significantly negatively correlated with the fold change in the peak number of CD8+ TILs at 1 or 2 months (Spearman's rank correlation, R = −.6270; P = .03; Figure 3C). DSG2 expression levels appeared to be negatively associated with fold change in intratumoral CD8+ cells. A positive correlation (not significant) was observed between CD46 RNA expression levels in the tumor tissue from the 12 patients and the number of CD8+ TILs (Spearman's rank correlation, R = .4085; P = .1874; Figure 3D).
As an update to the survival data from the phase I trial, 28 28 RNA expression had previously been determined via microarray but was not reported. 28 39 and malignant melanoma (MM). 40 DSG2 overexpression in HCC is an independent risk factor for reduced overall survival. 39 In MM, DSG2 overexpression may promote vasculogenic mimicry via cell-cell interactions and adhesion, but not viability, motility, and proliferation. 40 Therefore, cancer cells expressing DSG2 seem to be a good target for treatment with chimeric oncolytic adenoviruses such as ONCOS-102.

| Limitations
First, although the oncolytic activity of ONCOS-102 was not detected in A2780 cells, A2780 is susceptible to the oncolytic activity of replication-selective Ad5 agents. 41

LK and A-SWM are employees and/or shareholders in Targovax Oy in
Finland and Targovax ASA in Norway.

AUTHOR CONTRIBUTIONS
LK and A-SWM conceptualized the study. LK and A-SWM gave the methodolgy. LK provided software, worked on validation, and conducted formal analysis. Data LK and A-SWM did data curation.
LK and A-SWM wrote the orginal draft and reviewed and edited it.
LK visualized. LK and AS supervised the study. LK and A-SWM were responsible for project administration. LK acquired funding.