SOX10 Regulates Melanoma Immunogenicity through an IRF4-IRF1 Axis

Abstract

Loss-of-function mutations of JAK1/2 impair cancer cell responsiveness to IFNγ and immunogenicity. Therefore, an understanding of compensatory pathways to activate IFNγ signaling in cancer cells is clinically important for the success of immunotherapy. Here we demonstrate that the transcription factor SOX10 hinders immunogenicity of melanoma cells through the IRF4-IRF1 axis. Genetic and pharmacologic approaches revealed that SOX10 repressed IRF1 transcription via direct induction of a negative regulator, IRF4. The SOX10-IRF4-IRF1 axis regulated PD-L1 expression independently of JAK-STAT pathway activity, and suppression of SOX10 increased the efficacy of combination therapy with an anti-PD-1 antibody and histone deacetylase inhibitor against a clinically relevant melanoma model. Thus, the SOX10-IRF4-IRF1 axis serves as a potential target that can bypass JAK-STAT signaling to immunologically warm up melanoma with a "cold" tumor immune microenvironment. SIGNIFICANCE: This study identifies a novel SOX10/IRF4 pathway that regulates noncanonical induction of IRF1 independent of the JAK-STAT pathway and can be targeted to improve the efficacy of anti-PD-1 therapy in melanoma.

Figure 1.. SOX10 is negatively correlated with PD-L1 expression in melanoma.

A, Expression analysis of SOX10 mRNA and all other mRNAs in melanoma cell lines shows an inverse Pearson correlation relative to PD-L1 mRNA (GSE7127). B, SOX10 mRNA shows a negative correlation with PD-L1 mRNA by real-time RT-PCR. Pearson correlation value (r) were shown. C, The expressions of SOX10 and MITF were determined in melanoma cells with high SOX10 levels (UACC257 and A2058) or with low SOX10 levels (RPMI7951 and Hs.940.T). PD-L1 expressions on the cell surface were also determined by flow cytometry. D, Kaplan-Meier curves for SOX10low (red) and SOX10high (blue) groups of TCGA melanoma patients. Log-rank p-value = 2.3×10⁻², age-adjusted Cox-model p-value = 4.1×10⁻². The number of patients in each group is indicated in parentheses.

Figure 2.. SOX10 negatively regulates PD-L1 expression in melanoma.

A, Melanoma cells with high SOX10 expression (UACC257, A2058, and A375) or medium SOX10 expression (MeWo, SK-MEL-28, and Malme-3M) transfected with siCNTL or siSOX10 (#9 or #10) were subjected to flow cytometry. Representative flow cytometry graphs were shown. The mean fluorescent intensities are presented as the mean ± SD of at least three independent experiments. *p <0.01 vs. mean fluorescent intensity in each siCNTL-transfected cells by one-way ANOVA followed by the Bonferroni post test. B, UACC257, A2058, SK-MEL-28, and Malme-3M transfected with siCNTL or siSOX10 (#10) were treated with IFN-γ and then subjected to flow cytometry. Representative flow cytometry graphs were shown. The mean fluorescent intensities are presented as the mean ± SD of at least three independent experiments. *p <0.01 vs. mean fluorescent intensity in siCNTL/IFN-γ-treated cells compared with siSOX10/IFN-g -treated cells by one-way ANOVA followed by the Bonferroni post test. Other conditions were similar to those in Fig. 2A. C, A2058 (–) and A2058/SOX10 cells were treated with IFN-γ and then subjected to flow cytometry. Right panels shows the overexpression of SOX10 determined by Western blotting. Representative flow cytometry graphs were shown. The mean fluorescent intensities are presented as the mean ± SD of at least three independent experiments. *p <0.01 vs. mean fluorescent intensity in IFN-γ-treated A2058 cells compared with IFN-g-treated A2058/SOX10 cells by two-way ANOVA followed by the Bonferroni post test. Other conditions were similar to those in Fig. 2A. D, UACC257 cells were transfected with siCNTL and siSOX10 #10, and treated with 0.5 μM Baricitinib. After 72-hour transfection, the culture medium was refreshed with 0.5 μM Baricitinib. The cells were subjected to flow cytometry. Representative flow cytometry graphs were shown. The mean fluorescent intensities are presented as the mean ± SD of at least three independent experiments. Other conditions were similar to those in Fig. 2A.

Figure 3.. SOX10 regulates PD-L1 expression by repressing IRF1.

A, UACC257 cells were transfected with the indicated siRNAs for 96 hours. The chromatin immunoprecipitation assay was performed using antibody against the polymerase II phosphorylated serine 2 site (α-pol II S2). Immunoprecipitated DNA was quantified by real-time PCR using primers specific to the gene region of PD-L1 or IRF1. Results are normalized to the Pol-II S2 occupancy in siCNTL-transfected cells. Data are presented as the mean ± SD of three independent experiments. *p <0.01 vs. α-pol II S2-precipitated DNA in siCNTL-transfected cells by two-way ANOVA followed by the Bonferroni post test. B, UACC257 cells transfected with siCNTL or siSOX10 (#9 or #10) were subjected to qRT-PCR. Data are presented as the mean ± SD of three independent experiments.*p <0.01 vs. mRNA in siCNTL-transfected cells by two-way ANOVA followed by the Bonferroni post test. C, Melanoma cells transfected with the indicated siRNAs were subjected to Western blotting. The band intensities were measured by ImageJ, normalized to that of each Histone H3. The relative expression values were normalized to that of siCNTL-transfected cells, and shown below each panel. D, UACC257 cells transfected with siCNTL or siSOX10 (#9 or #10) were subjected to flow cytometry after staining PD-L2 or HLA-A/B/C. Representative flow cytometry graphs were shown. The mean fluorescent intensities are presented as the mean ± SD of at least three independent experiments. *p <0.01 vs. mean fluorescent intensity in each siCNTL-transfected cells by one-way ANOVA followed by the Bonferroni post test. E, UACC257 cells transfected with siCNTL, siSOX10 #10, or siIRF1 #3 were subjected to flow cytometry and Western blotting. Representative flow cytometry graphs were shown. The mean fluorescent intensities are presented as the mean ± SD of at least three independent experiments. *p <0.01 vs. mean fluorescent intensity in siSOX10-transfected UACC257 cells compared with siSOX10/siIRF1-transfected UACC257 cells by two-way ANOVA followed by the Bonferroni post test. Other conditions were similar to those in Fig. 3D.

Figure 4.. SOX10 represses IRF1 through direct binding to the IRF4 enhancer region.

A, Melanoma cells transfected with indicated siRNAs were subjected to Western blotting. The same image of Histone H3 was used as in Fig. 3C. The band intensities were measured by ImageJ, normalized to that of each Histone H3. The relative expression values were normalized to that of siCNTL-transfected cells, and shown below each panel. B, The chromatin immunoprecipitation assay was performed using antibody against Pol-II S2. Immunoprecipitated DNA was quantified using the specific primers for IRF4 gene region. Results are normalized to the input DNA in each siRNA-transfected cells. Relative pol-II occupancies are presented as the mean ± SD of three independent experiments. *p <0.01 vs. pol-II occupancy in siCNTL-transfected cells by one-way ANOVA followed by the Bonferroni post test. Other conditions were similar to those in Fig. 2A. C, The chromatin immunoprecipitation assay was performed using antibody against SOX10 (α-SOX10). Immunoprecipitated DNA was quantified by real-time PCR using primers specific to the enhancer region of IRF1 and MIA, known as a SOX10 binding site. Results are normalized to the input DNA in each region. Data are presented as the mean ± SD of three independent experiments. *p <0.01 vs. α-SOX10-precipitated DNA in the ACTB region by two-way ANOVA followed by the Bonferroni post test. D, UACC257 cells transfected with the indicated siRNAs were treated with IFN-γ for 24 hours. Total RNA was subjected to qRT-PCR. Results are normalized to PD-L1 expression in siCNTL-transfected cells without IFN-γ. Data are presented as the mean ± SD of three independent experiments. *p <0.01 vs. siCNTL-transfected cells with IFN-γ by two-way ANOVA followed by the Bonferroni post test.

Figure 5.. Involvement of HDAC1/3 in controlling PD-L1 expression through SOX10

A, Melanoma cells were treated with 2 μM vorinostat for 24 hours, and then subjected to flow cytometry or Western blotting. Representative flow cytometry graphs were shown. The mean fluorescent intensities are presented as the mean ± SD of at least three independent experiments. *p <0.01 vs. mean fluorescent intensity in DMSO-treated cells by student t-test. The band intensities were measured by ImageJ, normalized to that of each Histone H3. The relative expression values were normalized to that of DMSO-treated cells, and shown below each panel. B, A375 and A375/SOX10 cells were treated with 2 μM vorinostat for 24 hours, and then subjected to flow cytometry. Results are normalized to PD-L1 expression in non-treated A375 (–) cells. Representative flow cytometry graphs were shown. The mean fluorescent intensities are presented as the mean ± SD of at least three independent experiments. *p <0.01 vs. mean fluorescent intensity in DMSO-treated cells by two-way ANOVA followed by the Bonferroni post test. Right panels shows the overexpression of SOX10 determined by Western blotting.

Figure 6.. SOX10 suppression underlies the synergistic effect of HDAC inhibitor and PD-1 blockade.

A, D4M melanoma cells were treated with 2 μM vorinostat for 24 hours, and then subjected to flow cytometry. Representative flow cytometry graphs were shown. The mean fluorescent intensities are presented as the mean ± SD of three independent experiments. Other conditions were similar to those in Fig. 5A. B, D4M melanoma cells were engrafted subcutaneously in C57BL/6J mice. Mice were randomly divided into 4 groups (n=6 mice/group), and treated with vehicle (–) as a control, vorinostat, anti-PD-1 antibody (αPD-1), or a combination of vorinostat and αPD-1 (vorinostat +αPD-1). Tumor volumes were measured three times per week by Vernier calipers and the values were plotted as means ± SEM. Mice were sacrificed if the tumor was > 1,000 mm³. *P <0.01 by two-way ANOVA with Bonferroni correction compared with the vehicle-treated group. C, Kaplan-Meier survival curve displaying percent survival among treatment groups after treatment (n = 6 mice/group). Significance was determined by the Log-rank (Mantel-Cox) test. D, Percent of CD3+ T cells infiltrating melanoma tissues was measured on Day 7. E, SOX10 mRNA expression on tumors at an endopoint was determined by qRT-PCR. F, PD-L1 expression on tumor cells on Day 7 are shown. Results are normalized to the average of PD-L1 expression in tumor cells prepared from melanoma tissues with non-treated mice on Day 7. Data are presented as the mean ± SEM of 8 mice. G, Median fluorescent intensity of 4–1BB on CD8⁺ T cells is shown on Day 11. H, D4M cells overexpressing SOX10 (D4M/SOX10) were engrafted subcutaneously in C57BL/6J mice. Mice were randomly divided into 2 groups (n=6 mice/group), and treated with vehicle (–) as a control or a combination of vorinostat and αPD-1 (vorinostat +αPD-1). Other conditions are similar to those in Fig. 5B. Right panels shows the overexpression of SOX10 determined by Western blotting.