Vesicular Stomatitis Virus (VSV) G Glycoprotein Can Be Modified to Create a Her2/Neu-Targeted VSV That Eliminates Large Implanted Mammary Tumors

Abstract

Viral oncolytic immunotherapy is a nascent field that is developing tools to direct the immune system to find and eliminate cancer cells. Safety is improved by using cancer-targeted viruses that infect or grow poorly on normal cells. The recent discovery of the low-density lipoprotein (LDL) receptor as the major vesicular stomatitis virus (VSV) binding site allowed for the creation of a Her2/neu-targeted replicating recombinant VSV (rrVSV-G) by eliminating the LDL receptor binding site in the VSV-G glycoprotein (gp) and adding a sequence coding for a single chain antibody (SCA) to the Her2/neu receptor. The virus was adapted by serial passage on Her2/neu-expressing cancer cells resulting in a virus that yielded a 15- to 25-fold higher titer following in vitro infection of Her2/neu⁺-expressing cell lines than that of Her2/neu-negative cells (~1 × 10⁸/mL versus 4 × 10⁶ to 8 × 10⁶/mL). An essential mutation resulting in a higher titer virus was a threonine-to-arginine change that produced an N-glycosylation site in the SCA. Infection of Her2/neu⁺ subcutaneous tumors yielded >10-fold more virus on days 1 and 2 than Her2/neu^- tumors, and virus production continued for 5 days in Her2/neu⁺ tumors compared with 3 days that of 3 days in Her2/neu^- tumors. rrVSV-G cured 70% of large 5-day peritoneal tumors compared with a 10% cure by a previously targeted rrVSV with a modified Sindbis gp. rrVSV-G also cured 33% of very large 7-day tumors. rrVSV-G is a new targeted oncolytic virus that has potent antitumor capabilities and allows for heterologous combination with other targeted oncolytic viruses. IMPORTANCE A new form of vesicular stomatitis virus (VSV) was created that specifically targets and destroys cancer cells that express the Her2/neu receptor. This receptor is commonly found in human breast cancer and is associated with a poor prognosis. In laboratory tests using mouse models, the virus was highly effective at eliminating implanted tumors and creating a strong immune response against cancer. VSV has many advantages as a cancer treatment, including high levels of safety and efficacy and the ability to be combined with other oncolytic viruses to enhance treatment results or to create an effective cancer vaccine. This new virus can also be easily modified to target other cancer cell surface molecules and to add immune-modifying genes. Overall, this new VSV is a promising candidate for further development as an immune-based cancer therapy.

Keywords: targeted vesicular stomatitis virus; viral oncoimmunotherapy; viral oncolysis.

FIG 1

rrVSV-G genome drawn to scale. The VSV-G gene has been replaced with a mutant G-gene that disables the LDL-R binding site and adds SCA-erbb2 to the amino terminus. The gene for EGFP has been added to the genome as annotated.

FIG 2

Adaptation of rrVSV-G-SCA-erbb2. (Top) Passage number when mutations were first identified, and titers on D2F2/E2, a cell line that expresses Her2/neu, and on D2F2, the parent cell line that does not. (Bottom) Schematic of the G-SCA compound protein and recombinant inserted mutations (K47Q and R354Q) and adapted mutations. Domains I to IV and the transmembrane domain of VSV-G are labeled by colors according to the published crystal structure (41).

FIG 3

Location of rrVSV-G mutations on the crystallographic structure of the SCA and VSV-G. (A) Mutations T91 and T93 were labeled on the SCA using the template from PDB (PDB 6ZQK; HER2-binding scFv-Fab fusion 841). As shown, both mutations are not located in the hypervariable region. (B) The conformation of VSV-G in complex with low-density lipoprotein receptor (LDL-R; blue color) is derived from VSV-G CR2 (PDB 5OYL). Two residues, namely, K47 and R354, essential for the natural binding of VSV-G, were mutated to Q. All six mutations created during adaptation are labeled using colored balls. A purple sphere represents a large insertion of the SCA segment at the beginning of VSV-G.

FIG 4

Titer of rrVSV-G-SCA-erbb2 and wtVSV on cells that do and do not express the Her2/neu receptor. SKBR3 is a human breast cancer cell line and D2F2/E2 is a mouse mammary carcinoma that express erbb2. 143 is a human osteosarcoma cell line and D2F2 is a mouse mammary carcinoma that do not express erbb2. The means of three experiments with SEM bars are shown.

FIG 5

Inhibition by anti-erbb2 monoclonal antibody (MAb) herceptin of infection by VSV-G on D2F2/E2 cells that express Her2/neu and D2F2 cells that do not. Shown are the means of three experiments with SEM bars. The x axis shows a log₂ scale.

FIG 6

Growth assay. Yield of VSV-G-SCA-erbb2 at various times following infection, at an MOI of 0.001, of D2F2/E2 cells that express Her2/neu and D2F2 cells that do not. The titer of the yield was determined on D2F2/E2 cells. Shown are the means of three experiments with SEM bars. AUC 95% confidence intervals are as follows: D2F2E2, 3.42E+8 to 1.14E+9; and D2F2, 0 to 1.26E+8.

FIG 7

Growth assay. Yield of VSV-G-SCA-erbb2 at various times following infection, at an MOI of 10, of D2F2/E2 cells that express Her2/neu and D2F2 cells that do not. The titer of the yield was determined on D2F2/E2 cells. Shown are the means of three experiments with SEM bars.

FIG 8

Comparison of viral stability in vitro at 37°C between the VSV-G-SCA-erbb2, VSV-Sindbis-SCA-erbb2, and wtVSV-EGFP. The rate of decay and half-life (t_1/2) did not differ significantly among the 3 viruses (t_1/2 = 4.7 h for wtVSV, 5.3 h for rrVSV-Sindbis, and 5.87 h for rrVSV-G). The nonlinear regression one phase decay goodness of fit was R² = 0.9681 for wtVSV, 0.9464 for rrVSV-Sindbis, and 0.9747 for rrVSV-G. Data are means of results from 3 experiments with SEM bars. The x axis is shown as a log₂ scale.

FIG 9

In vivo growth assay. Yield of virus at various times following infection of subcutaneous D2F2/E2 or D2F2 tumors. The titer of the yield was determined on D2F2/E2 cells. Area under the curve (AUC) was significantly higher following infection of D2F2/E2 than that of D2F2 (unpaired t test, P = 0.03); n = 5 for each time point except day 3 which had 3. Mean titers with SEM bars are shown.

FIG 10

Histopathological examination of peritoneal tumor nodules at various times after implantation. Tumor nodules are identified by arrows. All size bars, 1,000 μm. Magnification as noted. (A) Black and white photomicrograph of multiple nodules of day 5 tumors. Hematoxylin and eosin (H&E), ×20. (B) Color photomicrograph of tumor in A. H&E, ×100. (C) Multiple nodules of day 7 tumors. H&E, ×20. (D) Multiple nodules of day 10 tumors. H&E, ×20.

FIG 11

Survival curves of D2F2/E2 peritoneal implants treated with rrVSV-G alone, rrVSV-Sindbis alone, or both. (A) Five-day tumors showed better survival following treatment with rrVSV-G than that with rrVSV-Sindbis (n = 10 per group; log rank statistic, P = 0.01). (B) Here, 7- and 10-day tumors showed better survival following treatment with rrVSV-G than that with the control (n = 10 per group; log rank statistic, P < 0.0001, for both 7-day and 10-day tumors).

FIG 12

Antitumor memory CD8 T cells. Cured animals were challenged with i.p. D2F2/E2 cells and peritoneal cells harvested 5 days later. Antitumor memory CD8 T cells were identified by flow cytometry analyses following staining with a tetramer displaying the immunodominant p63 epitope of the Her2/neu protein.