Directed natural evolution generates a next-generation oncolytic virus with a high potency and safety profile

Abstract

Oncolytic viruses (OVs) represent a type of encouraging multi-mechanistic drug for the treatment of cancer. However, attenuation of virulence, which is generally required for the development of OVs based on pathogenic viral backbones, is frequently accompanied by a compromised killing effect on tumor cells. By exploiting the property of viruses to evolve and adapt in cancer cells, we perform directed natural evolution on refractory colorectal cancer cell HCT-116 and generate a next-generation oncolytic virus M1 (NGOVM) with an increase in the oncolytic effect of up to 9690-fold. The NGOVM has a broader antitumor spectrum and a more robust oncolytic effect in a range of solid tumors. Mechanistically, two critical mutations are identified in the E2 and nsP3 genes, which accelerate the entry of M1 virus by increasing its binding to the Mxra8 receptor and antagonize antiviral responses by inhibiting the activation of PKR and STAT1 in tumor cells, respectively. Importantly, the NGOVM is well tolerated in both rodents and nonhuman primates. This study implies that directed natural evolution is a generalizable approach for developing next-generation OVs with an expanded scope of application and high safety.

Fig. 1. M1 virus was serially passaged in the refractory HCT-116 cell line to generate more effective OVs.

a Schematic of serial passaging in the HCT-116 cell line and generation of mutated viruses by site-directed mutagenesis. b Phase contrast and GFP fluorescence images of P3, P8 and P10 are shown. Scale bars, 50 μm. Representative images of n  =  3. c Cell viability was evaluated by an MTT assay after cells were infected with serial dilutions of M1-GFP, P3, P8 and P10. EC50 shift was calculated by nonlinear regression and the EC50 values were used for statistical analysis by one-way ANOVA with Tukey’s multiple comparisons test relative to M1-GFP and adjusted P values are indicated. d Schematic of M1-GFP genomic RNA. Viral genomic RNA was isolated for genetic analysis. Nucleotide substitutions are highlighted in red boxes. e HCT-116 cells were infected at an MOI of 0.1 and imaged with a phase contrast microscope and a fluorescence microscope 72 h after infection. Representative images of n  =  3. Scale bars, 50 μm. f HCT-116 cells were infected with M1-GFP and M1-N3E2M (MOI = 1) for 48 hours, and the infection rate was determined by flow cytometry. P < 0.0001 was calculated with Two-tailed unpaired t-test. g After infection of HCT-116 cells with M1-GFP and M1-N3E2M at an MOI of 0.1, the viral titer was tested by the CCID50 method. P = 0.0060 was determined by Two-way ANOVA relative to M1-GFP. h Cell viability was evaluated after cells were infected with serial dilutions of M1-GFP and M1-N3E2M. EC50 shift was calculated by nonlinear regression. Statistical significance was calculated using Two-way ANOVA with Bonferroni’s multiple comparisons test relative to M1-GFP. Adjusted P values are: MOI (Mock), P  >  0.9999; MOI (−3), P = 0.0003; MOI (−2 to 1), P < 0.0001. n.s.: no significance, **P  <  0.01, ***P  <  0.001, ****P  <  0.0001. Data are shown as mean ± SD, for n = 3 biological replicates. Source data are provided as a Source Data file.

Fig. 2. The oncolytic effect of M1-N3E2M was improved in a variety of tumor cells, and it did not cause a CPE in normal cell line.

a The viability of 57 human cell lines were evaluated by an MTT assay 72 h after infection. Liver cancer is shown in yellow, colon cancer in green, prostate cancer in orange, pancreas cancer in purple, bladder cancer in blue and breast cancer in light blue. b The oncolytic effects of M1-GFP and M1-N3E2M were analyzed by Two-tailed paired t test and P values are indicated. The data are shown as violin plots with the box limits at minima and maxima and center line at median. (colon cancer, n = 8; liver cancer, n = 2; breast cancer, n = 15; pancreas cancer, n = 11; prostate cancer, n = 6; bladder cancer, n = 15). c, e HCT-8 and Huh-7 cells were infected with M1 viruses at an MOI of 0.1. Representative images of n  =  3. Scale bars, 50 μm. d, f The viability of HCT-8 and Huh-7 cells was evaluated by an MTT assay. EC50 shift was calculated by nonlinear regression. Statistical significance was calculated using Two-way ANOVA with Sidak’s multiple comparisons test relative to M1-GFP. Adjusted P values are: d MOI (Mock), P  >  0.9999; MOI (−3), P = 0.9991; MOI (−2), P = 0.0008; MOI (−1 to 1), P < 0.0001; f MOI (Mock), P  >  0.9999; MOI (−3), P = 0.0528; MOI (−2), P = 0.0028; MOI (−1 and 0), P < 0.0001; MOI (1), P = 0.0489. g CCD-18Co cells were infected with M1-GFP and M1-N3E2M at an MOI of 10. Representative images of n  =  3. Scale bars, 50 μm. h The viability of CCD-18Co cells was evaluated by an MTT assay. EC50 shift was calculated by nonlinear regression. n.s.: no significance, *P  <  0.05, **P  <  0.01, ***P  <  0.001, ****P  <  0.0001. Data points represent mean % viability relative to vehicle ± SD, for n  =  3 biological replicates. Source data are provided as a Source Data file.

Fig. 3. The oncolytic effect of M1-N3E2M was potentiated in vivo and ex vivo.

a, b HCT-116 xenografts were treated intravenously with vehicle, M1-GFP or M1-N3E2M for 6 consecutive days. n  =  7 mice per group. c Tumor tissues were analyzed by immunohistochemistry for GFP, cleaved caspase-3, and Ki-67. Representative images of n  =  3. Scale bars, 20 μm. Quantification of GFP, cleaved caspase-3 and Ki-67, n  =  3 biological replicates. Statistical significance was calculated using One-way ANOVA with Sidak’s multiple comparisons test relative to M1-GFP. Adjusted P values are: GFP P < 0.0001, Cl-casps-3 P = 0.0002, Ki-67 P = 0.0376. d Normal tissues from the brain, colon, liver, lungs, heart (Scale bars, 20 μm) and joints (Scale bars, 500 μm). were analyzed by immunohistochemistry for GFP. Representative images of n  =  3. e–h HCT-116 and SW620 xenografts were treated intravenously for 21 consecutive days. In HCT-116 xenograft model, n  =  7 mice per group; in SW620 xenograft model, n  =  10 mice per group. i–l CT26 xenografts in BALB/c mice and HEPA1-6 xenografts in C57BL/6 mice were treated intravenously for 6 consecutive days. In CT26 xenograft model, n  =  5 mice per group; in HEPA1-6 xenograft model: Control n  =  7 mice, M1-GFP n = 9 mice, M1-N3E2M n = 8 mice. m Colorectal tumor tissues from six patients were treated with M1-GFP or M1-N3E2M (1×10⁷ PFUs) for 72 hours, and cell viability was assessed. One graph bar represents the mean cytotoxicity % relative to vehicle of one tumor sample. Statistical significance of tumor volume was calculated using Two-way ANOVA with Tukey’s multiple comparisons test. Adjusted P values are: b M1-GFP vs. control, P = 0.4464; M1-N3E2M vs. M1-GFP, P = 0.0005; f M1-GFP vs. control, P < 0.0001; M1-N3E2M vs. M1-GFP, P < 0.0001; h M1-GFP vs. control, P < 0.0001; M1-N3E2M vs. M1-GFP, P = 0.0006; j M1-GFP vs. control, P = 0.0673; M1-N3E2M vs. M1-GFP, P = 0.0127; l M1-GFP vs. control, P = 0.6654; M1-N3E2M vs. M1-GFP, P = 0.0176. n.s.: no significance, *P  <  0.05, **P  <  0.01, ***P  <  0.001, ****P  <  0.0001. Data are shown as mean ± SD. Source data are provided as a Source Data file.

Fig. 4. The oncolytic effect of M1-N3E2M was synergistically enhanced by the M358L and K4N mutations.

a Cell viability was evaluated by an MTT assay. EC50 shift was calculated by nonlinear regression and the EC50 values were used for statistical analysis by unpaired t-test with Welch’s correction, M1-NS3M vs. M1-GFP, P = 0.0467; M1-E2M vs. M1-GFP, P = 0.0481; M1-N3E2M vs. M1-NS3M, P = 0.0345; M1-N3E2M vs. M1-E2M, P = 0.0237. Data points represent mean % viability relative to vehicle ± SD, for n  =  3 biological replicates. b HCT-116 cells were infected with M1 viruses at an MOI of 0.1 and imaged 48 h after infection. Representative images of n  =  3. Scale bars, 50 μm. c The viral titer was determined by a TCID50 assay. We performed a statistical analysis of the final virus production using unpaired t-test with Welch’s correction, M1-NS3M vs. M1-GFP, P = 0.0240; M1-E2M vs. M1-GFP, P = 0.0327; M1-N3E2M vs. M1-NS3M, P = 0.0451; M1-N3E2M vs. M1-E2M, P = 0.0284. Data points represent mean viral titer ± SD, for n  =  3 biological replicates. d Monolayer HCT-116 cells were infected with M1 viruses at an MOI of 0.1. The medium was replaced with semisolid medium 1 h after infection. Representative images of n  =  3. Scale bars, 500 μm. The scale bars in the magnified images represent 100 μm. e Quantification of the plaque area in (d). Statistical significance was calculated using unpaired t-test with Welch’s correction and P values are indicated. The data are shown as violin plots with the box limits at minima and maxima and center line at median (M1-GFP n = 21, M1-E2M n = 30, M1-NS3M n = 17, M1-N3E2M n = 32). f The plaques in (d) were counted. Statistical significance was calculated using One-way ANOVA with Tukey’s multiple comparisons test and P values are indicated. Graph bars represent mean plaque number per well ± SD, for n  =  3 biological replicates. n.s.: no significance, *P  <  0.05, **P  <  0.01, ***P  <  0.001, ****P  <  0.0001. Source data are provided as a Source Data file.

Fig. 5. The K4N mutation in E2 improved the attachment and entry of M1 virus.

a A modified plaque formation assay was performed and cells were imaged with a fluorescence microscope 48 hours after infection (MOI = 0.5). Scale bars, 500 μm. The scale bars in the magnified images represent 100 μm. b Plaques were counted with a fluorescence microscope 48 hours after infection. Data points represent mean plaque number per well ± SD, for n  =  3 biological replicates. P  <  0.0001 was determined by Two-way ANOVA relative to M1-GFP. c HCT-116 cells were incubated with M1-GFP and M1-E2M at 4 °C for 1 hours. Viral RNA was quantified by qRT-PCR and presented as mean ± SD, for n  =  3 biological replicates. P  <  0.0001 was calculated with Two-tailed unpaired t-test. d Western blot analysis of M1-GFP and M1-E2M incubated with MXRA8-His bound to His-Tag Mouse mAb Sepharose Beads. Precipitated viral particles were detected using an anti-E1 mAb (left). Quantification of E1 expression is shown (right). Graph bars represent mean densitometry of E1 normalized to the densitometry of MXRA8 ± SD, for n  =  3 biological replicates. P  = 0.0060 was calculated with Two-tailed unpaired t-test. e Time course of the binding between M1 viral particles and the MXRA8 protein, as determined via BLI. f, g HeLa, HeLa-Mxra8, Hs 578 T and Hs 578T-ΔMxra8 cells were treated with M1-GFP or M1-E2M. The infection rate was determined by flow cytometry 48 hours after infection. Graph bars represent mean infection rate % ± SD, for n  =  3 biological replicates. Statistical significance was calculated using Two-way ANOVA with Sidak’s multiple comparisons test relative to M1-GFP. Adjusted P values are: f vector P = 0.6199; Mxra8 P = 0.0002; g vector, P = 0.0010; ΔMxra8 P = 0.2713. n.s.: no significance, **P  <  0.01, ***P  <  0.001, ****P  <  0.0001. Source data are provided as a Source Data file.

Fig. 6. The M358L mutation in nsP3 inhibited the activation of PKR and STAT1, further inhibiting IFN-mediated antiviral responses.

a HCT-116 cells were infected with M1-GFP and M1-NS3M (MOI = 10), and the infection rate was determined by flow cytometry. b The expression of GFP in infected cells in (a) was detected. MFI, mean fluorescence intensity. c A modified plaque formation assay was performed and plaques were counted with a fluorescence microscope 48 hours after infection. d Schematic of the process for identifying interactions with nsP3 WT and nsP3 M358L. e This Venn diagram shows the host proteins that interact with nsP3 WT and nsP3 M358L. f, g Bar graph of enriched terms across the 59 nsP3 WT interactors and 22 nsP3 M358L interactors, colored by p values. The top 10 enriched pathways are shown. See also Table S1 and S2. h Coimmunoprecipitation was conducted with an anti-Flag antibody or isotype control IgG prior to immunoblot analysis with anti-nsP3 and anti-PKR antibodies. Representative images of n  =  3. i, j HCT-116 cells were treated with control, M1-GFP or M1-NS3M for 4 and 24 hours (MOI = 10), and the levels of proteins in the cell lysates was examined by Western blotting (left). Quantification of p-PKR and p-STAT1 (right). k Quantification of PKR and STAT1 expression in (i, j). l The transcript levels of ISGs and viral RNA were quantified by qRT‒PCR after 24 h infection with control, M1-GFP or M1-NS3M (MOI = 1). (m-o) qRT-PCR (m) and western blotting (n) were used to evaluate the shRNAs knockdown efficiency of PKR. PKR knockdown cells were infected with M1-GFP and M1-NS3M (MOI = 1) and the infection rate was determined by flow cytometry (o). p, q Proteins were detected by Western blotting in CCD-18Co cells after infection with M1-GFP or M1-NS3M (MOI = 1). Quantification of p-PKR and p-STAT1. Statistical significance was calculated using Two-way ANOVA with Sidak’s multiple comparisons test (a, b, i, j, o, q) or One-way ANOVA with Tukey’s multiple comparisons test (k–l), and adjusted P values are indicated. Data are shown as mean ± SD from three biological replicates. Source data are provided as a Source Data file.

Fig. 7. M1-N3E2M is well tolerated in nonhuman primates.

a Timeline of the administration schedule. iv, intravenously. b Animals were weighed weekly after treatment initiation. c Animal body temperature, electrocardiographic and blood pressure measurements were conducted on D-3, D1, D5, D20, D43 and D71. MBP, mean blood pressure. d Blood was collected from a subcutaneous vein of the hind limbs of animals on D-3, D1, D6, D21, D44 and D72 for analysis of serum biochemical parameters. ALT, alanine aminotransferase. AST, aspartate aminotransferase. Alb, albumin. Cre, creatinine. e Blood was collected from a subcutaneous vein of the hind limbs of animals on D-3, D1, D6, D21, D44 and D72 for analysis of hematological parameters. WBC, white blood cell. f Analysis of blood coagulation function on D-3, D6, D21, D44 and D72. APTT, activated partial thromboplastin time. g Cytokine and serum C-reactive protein detection. h Complement detection on D-3, D1, D6, D21, D44 and D72. See also Table S3-S6. Source data are provided as a Source Data file.