PRMT5 control of cGAS/STING and NLRC5 pathways defines melanoma response to antitumor immunity

PRMT5 expression is inversely correlated with an antitumor immune signature

The finding that SHARPIN interaction with PRMT5 is important for PRMT5 methylation of specific substrates (19) led us to assess SHARPIN expression in cohorts of melanoma tumor specimens. Analysis of melanoma patient datasets revealed that low SHARPIN expression in MTAP-low tumors is associated with better survival (fig. S1, A and B). To identify pathways that may be regulated by SHARPIN, we assessed differentially expressed genes (DEGs) in cohorts of metastatic melanoma patient specimens. Relative to MTAP-low tumors with high SHARPIN expression, gene set enrichment analysis (GSEA) revealed that those with low SHARPIN expression exhibited enrichment of genes associated with immune-related pathways [T helper cell 1 (T_H1)/ T_H2, interleukin-2 (IL2)/signal transducers and activators of transcription 5 (Stat5), and tumor necrosis factor–α (TNFα)] (fig. S1, C and D, and table S1), suggesting that SHARPIN may be involved in the control of immune phenotypes in MTAP-low melanoma.

Because both MTAP and SHARPIN augment PRMT5 activity (19), we asked whether high or low PRMT5 expression in melanoma (Fig. 1A) was associated with particular gene sets. GSEA showed that melanomas with low PRMT5 expression exhibited enriched expression of immune-associated genes (Fig. 1, B and C; fig. S1, E and F; and tables S2 and S3), similar to the MTAP-low melanomas with low SHARPIN expression. Analysis of an independent cohort of melanomas (GSE78220, n = 27) (26) confirmed enrichment of an immune gene signature (allograft rejection and IFN-γ response) in tumors with low PRMT5 expression (fig. S2).

• Download high-res image
• Open in new tab
• Download Powerpoint

Among the nine PRMT family members, PRMT5, PRMT1, PRMT2, CARM1, and PRMT7 are expressed at relatively high levels in human melanoma specimens (fig. S3A), and, among these, PRMT5, PRMT1, and CARM1 are coexpressed (fig. S3B). High PRMT1, PRMT5, or CARM1 expression correlated with lower survival (fig. S3C). Low PRMT5 expression exhibited the strongest correlation with expression of immune response genes (fig. S3D). Correspondingly, enrichment of an immune pathway signature was also seen in melanomas harboring low MTAP expression and low PRMT5 activity (fig. S3E), supporting a potential role for PRMT5 in tumor immunity.

PRMT5 inhibition attenuates tumor growth in an immunocompetent murine melanoma model

Inspired by the melanoma tumor results, we assessed PRMT5 function in antitumor immunity in B16F10 (B16) metastatic murine melanoma cells expressing either scrambled (Scr) or PRMT5-specific short hairpin RNA (shRNA). PRMT5 knockdown (KD) in B16 cells reduced PRMT5 abundance and decreased PRMT5 activity (Fig. 2A). PRMT5 KD did not affect the growth of B16 cells in culture (Fig. 2B). However, inoculation of these cells into immunocompetent syngeneic C57BL/6 or into immunocompromised nonobese diabetic–severe combined immunodeficient gamma (NSG) mice revealed important differences. In C57BL/6 mice, PRMT5 KD markedly inhibited growth (37 to 62% reduction in tumor volume and 28 to 54% reduction in tumor weight at day 17; Fig. 2C and fig. S4A), phenotypes not seen in B16 cells that were inoculated in the NSG mice (Fig. 2D and fig. S4B). Despite variation in KD efficiency, most tumors from the PRMT5 KD cells exhibited reduced PRMT activity (fig. S4C). The ability of PRMT5 KD B16 cells to develop tumors in immunocompromised but not immunocompetent mice suggested that PRMT5 inhibition of melanoma growth requires an intact immune system.

• Download high-res image
• Open in new tab
• Download Powerpoint

We transferred B16 tumors from NSG into C57BL/6 mice and observed growth inhibition of PRMT5 KD but not control (Scr) tumors (Fig. 2F and fig. S4D), indicating that the tumors from the NSG mice could be controlled in an immunocompetent environment. Both the abundance and activity of PRMT5 were reduced in PRMT5 KD cells (shPRMT5 pools 1 and 2) and tumors (shPRMT5) (Fig. 2, E and G). Consistent with the observations in B16 cells, inducible PRMT5 KD in YUMM1.7 cells, which are derived from a genetically engineered murine melanoma model (Braf^V600E/Pten^−/−/Cdkn2a^−/−), inhibited tumor growth (52% by day 24; fig. S4E), and had a limited effect on growth of these cells in culture (fig. S4F). Effectiveness of KD was confirmed at the levels of PRMT5 abundance and activity (fig. S4G).

To substantiate the phenotypes seen upon KD of PRMT5, we performed gain-of-function studies using YUMMER1.7 cells, which are derived from ultraviolet type B (UVB)–irradiated YUMM1.7 cells and exhibit increased mutation burden and antigenicity (27). Growth of tumors derived from YUMMER1.7 cells is inhibited in C57BL/6 but not Rag^−/− mice (27). We hypothesized that gain of PRMT5 function in YUMMER1.7 cells reduced immunogenicity and increased tumor growth in immunocompetent mice. To explore this hypothesis, we generated YUMMER1.7 cells overexpressing PRMT5 or its adaptor WDR77, which is essential for PRMT5 methyltransferase activity (28), or a combination of both (Fig. 2H). YUMMER1.7 cells expressing both PRMT5 and WDR77 had increased PRMT5 protein and corresponding activity (Fig. 2H). Those cells also showed a moderate, albeit statistically significant (P = 0.0025), growth advantage in culture (relative to cells with empty vector) (Fig. 2I). The impact of PRMT5 and WDR77 coexpression on tumor growth in immunocompetent mice was variable (Fig. 2J), likely due to variability in the abundance of the proteins and PRMT5 activity (fig. S4H). Nevertheless, coexpression of both PRMT5 and WDR77 in YUMMER1.7 cells increased tumor growth in immunocompetent mice, relative to mice with control YUMMER1.7 tumors (Fig. 2J). Such an increase in tumor growth was restricted to immune-competent animals and was not seen in immunocompromised (NSG) mice (Fig. 2, K and L), consistent with PRMT5 activity supporting tumor growth by limiting antitumor immune responses through a tumor-intrinsic mechanism.

PRMT5 controls melanoma infiltration of immune cells

Immunocompetent mice harboring tumors derived from PRMT5 KD B16 cells exhibited reduced tumor growth (70 to 75% lower tumor weight at day 17) relative to mice with tumors from Scr cells (fig. S4, I and J). To assess whether the difference in tumor growth related to differences in the infiltrating immune cells, we profiled the immune cells in tumors isolated at day 17 by flow cytometry (fig. S4K). In two independent experiments (Fig. 3A, left and right), we observed significantly (P < 0.05) higher numbers of tumor-infiltrating immune cells in PRMT5 KD tumors, relative to Scr tumors. In the first experiment (Fig. 3A, left), we evaluated T lymphocytes [CD45⁺, CD3⁺, CD4⁺, and CD8⁺ cells, natural killer (NK) cells (NK1.1⁺), dendritic cells (DCs) (CD11c⁺MHCII⁺), and macrophages (CD11b⁺F4/ 80⁺)] (Fig. 3A, left graph). In the second experiment (Fig. 3A, right), we evaluated CD45⁺ T cells and two immune suppressor cell types, myeloid-derived suppressor cells (MDSCs; CD11b⁺GR1⁺) and regulatory T cells (T_regs; CD4⁺FOXP3⁺). To ascertain whether the PRMT5 KD promoted an antitumor immunity phenotype, we calculated the relative abundance of active CD8⁺ T cells (CD44^hiCD8⁺) to that of MDSC or T_regs. In PRMT5 KD tumors, the ratio of activated CD8⁺ T (CD44^hi CD8⁺) to MDSCs (CD11b⁺GR1⁺) or T_regs (CD4⁺FOXP3⁺) was significantly higher (P = 0.0246 for MDSC and P = 0.0383 for T_regs) (Fig. 3B). Consistent with these observations, immunohistochemical analysis of tumors collected at day 12 confirmed increased infiltration of CD4⁺ and CD8⁺ lymphocytes in PRMT5 KD tumors (Fig. 3C).

• Download high-res image
• Open in new tab
• Download Powerpoint

Conversely, PRMT5- and WDR77-overexpressing YUMMER1.7 tumors from immunocompetent mice collected at day 12 showed decreased immune cell infiltration. The number of T cells (CD45⁺), activated T cells (CD44^hiCD4⁺ and CD44^hiCD8⁺), NK cells, DCs (CD11c⁺MHCII⁺), and macrophages (CD11b⁺F4/80⁺) were all significantly (P < 0.05) decreased in PRMT5- and WDR77-overexpressing YUMMER1.7 tumors, compared to tumors from cells with empty vectors (Fig. 3D).

To substantiate the possible contribution of key immune infiltrating cell types on the degree of tumor growth, we monitored the size of PRMT5 KD B16 tumors after depletion of either NK cells or CD8⁺ T cells (fig. S4, L and M). PRMT5 KD tumors were bigger in mice injected with neutralizing antibodies to NK (anti-NK1.1) compared with PRMT5 KD tumors grown in mice injected with immunoglobulin G (IgG) (Fig. 3E). Depletion of NK cells also abolished the survival benefit gained by PRMT5 KD (Fig. 3F). Likewise, growth of PRMT5 KD tumors was higher in mice injected with neutralizing antibodies to CD8⁺ T cells (anti-CD8), compared with tumors from these cells in mice injected with IgG (Fig. 3G). Depletion of CD8⁺ T cells also reduced the survival of mice with PRMT5 KD tumors (Fig. 3H). These observations indicated that PRMT5 has a tumor-intrinsic effect that limits immune cell infiltration and that impairing PRMT5 activity confers a survival benefit that depends on immune cell infiltration into the melanoma tumor.

PRMT5 methylates IFI16, an intracellular DNA-sensing protein that interacts with SHARPIN

We explored the expression of a particular PRMT5 adaptor–encoding genes and antitumor immunity. We found that metastatic melanoma specimens with low variation in PRMT5 expression (fig. S5A) could be divided into distinct groups based on high or low expression of the adaptor genes for SHARPIN, WDR77, coordinator of PRMT5 and differentiation stimulator (COPRS), chloride nucleotide-sensitive channel 1A (CLNS1A), RIO kinase 1 (RIOK1), and multiple endocrine neoplasia I (MEN1) (fig. S5B). With GSEA and excluding PRMT5 expression as a variable, we evaluated the association between these adaptor-encoding genes and immune-associated genes (Fig. 4, A and B). Low SHARPIN expression correlated with an enriched immune gene signature (Fig. 4B).

• Download high-res image
• Open in new tab
• Download Powerpoint

Because SHARPIN serves as an adaptor for PRMT5 and its substrates, we used liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis to identify SHARPIN-interacting proteins in the human melanoma WM115 cell line, which has homozygous MTAP deletion and is sensitive to SHARPIN KD (19). In addition to identifying known SHARPIN-interacting proteins, PRMT5 and components of the linear ubiquitin assembly complex [RNF31 (ring finger protein 31) and RBCK1 (RANBP2-type and C3H4-type zinc finger containing 1)], we identified IFN-γ–inducible protein 16 (IFI16) (table S4). IFI16 contains a DNA binding hematopoietic IFN-inducible nuclear protein (HIN) domain (29, 30) and is implicated in controlling p53 transcriptional activity (31), in cell cycle regulation by binding to the retinoblastoma protein (32), in antimicrobial immunity by sensing cytosolic DNA (29), and in inflammasome assembly through its interaction with cGAS and STING (23, 33). Interaction of IFI16, or its murine homolog IFI204, with SHARPIN was confirmed by immunoprecipitation in human WM115 melanoma cells, human A375 melanoma cells, mouse B16 melanoma cells, and transfected human embryonic kidney (HEK) 293T cells (Fig. 4, C and D, and fig. S6, A to C). WM115 are MTAP-deleted and have lower basal PRMT5 activity than A375 cells, which express MTAP (19). Inhibition of PRMT5 with EPZ015666 appeared to facilitate the interaction between IFI16 and SHARPIN in A375 melanoma cells (fig. S6C). In both A375 cells and WM115 melanoma cells, inhibition of PRMT5 reduced IFI16 methylation (Fig. 4E and fig. S6D). Likewise, IFI204 exhibited reduced methylation in EPZ015666-treated B16 cells (Fig. 4F).

A search for consensus RG motifs with arginine residues that could be methylated (19, 34, 35) identified Arg¹² in the N-terminal PYRIN (protein-protein interaction) domain and Arg⁵³⁸ in the C-terminal HIN (DNA binding) domain in the murine IFI204 protein (fig. S6E). When expressed in B16 cells or immunoprecipitated from HEK293T cells and tested in vitro, the IFI204 mutants were less methylated than those with wild-type IFI204 (Fig. 4, G and H, and fig. S6F).

IFI16 methylation attenuates dsDNA–induced TBK1–IRF3 activation and IFN and chemokine production

IFI16 binding to intracellular double-stranded DNA (dsDNA) induces the expression of IFNB1 and the chemokines CCL5 and CXCL10 (23, 29, 33). Because we detected PRMT5-mediated methylation of IFI16 and IFI204, we explored the effect of this modification on IFI204-dependent IFN-β and chemokine induction. Attenuating PRMT5 activity (by KD or EPZ015666) in B16 cells increased the expression of Ifnb1, Ccl5, and Cxcl10 after stimulation of the cells with 70–base pair dsDNA [referred to as V70mer (29)] (Fig. 5, A and B). Conversely, PRMT5 overexpression decreased dsDNA-stimulated expression of all three genes in both B16 and YUMMER1.7 cells (Fig. 5C and fig. S7A). These observations suggested that PRMT5-dependent methylation of IFI204 limits dsDNA stimulation of IFN-β and chemokines.

• Download high-res image
• Open in new tab
• Download Powerpoint

Sensing of dsDNA involves an interaction between IFI16/IFI204 and the cGAS-STING pathway to activate TANK Binding Kinase 1 (TBK1)–IFN regulatory factor 3 (IRF3) signaling and IFN production (23, 33). Therefore, we assessed whether PRMT5 methylation of IFI204 affected cGAS-STING signaling. Overexpression of SHARPIN in B16 cells attenuated intracellular dsDNA-mediated activation of TBK1-IRF3 signaling and subsequent Ifnb1, Ccl5, and Cxcl10 expression (fig. S7, B and C). Conversely, PRMT5 KD in B16 cells augmented dsDNA-induced activation of TBK1-IRF3 as reflected by increased TBK1, IRF3, and STING phosphorylation (Fig. 5D), increased STING dimerization, detected with seminative polyacrylamide gel electrophoresis (PAGE) (Fig. 5E), and greater STING polymerization, detected with blue native PAGE (Fig. 5F) (33, 36). Similarly, pharmacological inhibition of PRMT5 in B16 cells appeared to enhance TBK1-IRF3 signaling (Fig. 5G). Consistent with these findings, ectopic expression of PRMT5 + WDR77 in B16 or YUMMER1.7 cells appeared to decrease TBK1-IRF3 activation (Fig. 5H and fig. S7D).

To support the importance of IFI204 methylation, we monitored changes in TBK1-IRF3 signaling after dsDNA stimulation in cells expressing wild-type IFI204 or methylation-defective mutants. Compared to cells that expressed empty vector, ectopically expressed IFI204 in B16 cells enhanced TBK1-IRF3 signaling (fig. S7E) and increased the expression of Ifnb1 and Ccl5 (fig. S7F) after dsDNA treatment. Compared to cells overexpressing wild-type IFI204, cells expressing methylation-defective R12A IFI204 (IFI204Mt1), but not R538A (IFI204Mt2), had greater TBK1-IRF3–mediated expression of downstream genes (Fig. 5I). Consistent with these observations, IFI204Mt1 expression, but not that of IFI204Mt2, increased STING dimerization and polymerization after dsDNA stimuli (Fig. 5J and fig. S7G), suggesting a critical role of IFI204 methylation on Arg¹² in regulating the activity of the dsDNA-stimulated STING pathway. In agreement, small interfering RNA (siRNA)–mediated STING KD reduced activation of Ifnb1, Ccl5, and Cxcl10 seen after PRMT5 KD (Fig. 5K and fig. S7H). Changes in PRMT5 abundance did not alter cGAS or STING abundance (fig. S7I), suggesting that PRMT5 limits the activation but not the amounts of cGAS/STING pathway components.

After dsRNA stimuli, Toll-like receptor 3 (TLR3) and retinoic acid-inducible gene-I (RIG-I) also stimulate a type I IFN response (37). To determine whether PRMT5 regulated this nucleic acid–sensing pathway, we treated B16 melanoma cells with the TLR3/RIG-I agonist polyinocinic and polycytidylic acid [poly(I:C)] and monitored Ifnb1 and chemokine expression. PRMT5 KD significantly (P < 0.05) enhanced the expression of Ifnb1, Ccl5, and Cxcl10 in response to either high–molecular weight (HMW) or low–molecular weight (LMW) poly(I:C) (fig. S7J). Consistent with cGAS/STING pathway observations, B16 melanoma cells overexpressing IFI204Mt1, but not IFI204Mt2, exhibited increased induction of Ifnb1 and chemokine expression by poly(I:C), relative to cells overexpressing wild-type IFI204 (fig. S7K). Together, these observations revealed an unexpected role for PRMT5/IFI204 in dsDNA-induced STING-dependent and dsRNA-induced activation of the type I IFN response.

PRMT5 regulates antigen presentation by controlling NLRC5 expression

To search for PRMT5-regulated genes that may affect tumor immune responses, we surveyed both The Cancer Genome Atlas (TCGA) metastatic melanoma dataset (n = 368) and the Cancer Cell Line Encyclopedia (CCLE) (n = 58) (38). Of coregulated genes (155 genes from TCGA and 135 from CCLE) (fig. S8A), 9 were common to both datasets. One of these nine, only the one encoding transcriptional activator NLRC5 (nucleotide-binding oligomerization domain-like receptor family caspase recruitment domain containing 5), had been implicated in regulating MHCI gene expression (39, 40). NLRC5, along with B2M, HLA-A, HLA-B, HLA-C, and proteasome 20S subunit beta 9 (PSMB9), are implicated in antigen presentation, as predicted by ingenuity pathway analysis (IPA) [log (P value) = −13.9; Fig. 1B and fig. S8A]. NLRC5 expression was inversely correlated with PRMT5 expression in both CCLE [correlation coefficient (r) = −0.516, P < 0.0001] and TCGA (r = −0.3158, P < 0.0001) datasets (Fig. 6, A to C, and fig. S8B). Increased NLRC5 expression is also reported in lung cancer cells subjected to PRMT5 KD (14), consistent with our observations (fig. S8C).

• Download high-res image
• Open in new tab
• Download Powerpoint

Down-regulation of proteins functioning in MHCI-mediated antigen presentation can underlie immune evasion by cancer cells. Thus, we asked whether PRMT5 activity inhibited NLRC5 expression, the regulation of NLRC5 target genes implicated in immune evasion (39), or both. Reducing PRMT5 (either by KD or pharmacological inhibition with MTA) in B16 cells increased basal Nlrc5 expression and that of Nlrc5 target genes implicated in antigen processing and presentation (Fig. 6, D and E). In contrast, ectopic expression of PRMT5 + WDR77 in B16 or YUMMER1.7 melanoma cells decreased the expression of NLRC5 and its target genes (Fig. 6F and fig. S8D). In agreement with earlier reports (38, 40), overexpression of NLRC5 or stimulation with IFN-γ, which induces NLRC5 expression, increased the abundance of PSMB9 and of the amount of MHCI (H-2K^b) at the surface of B16 cells (fig. S8, E to G). Likewise, PRMT5 KD in B16 cells increased NLRC5 and PSMB9 abundance in cells stimulated with IFN-γ (Fig. 6G) and increased MHCI abundance at the cell surface (Fig. 6H). PRMT5 KD did not alter the abundance of the IFN-γ receptor (fig. S8H), suggesting that PRMT5 was not affecting this receptor (39). Overall, these findings suggest that PRMT5 controls MHCI abundance and antigen presentation through an effect on NLRC5 transcription.

IFI204 and NLRC5 expression inhibits in vivo growth of mouse melanoma tumors

Because our data indicated that PRMT5 impaired the function of IFI204 in stimulating type I IFN activity and reduced NLRC5 expression, we asked whether the expression of a methylation mutant form of IFI204 or the overexpression of NLRC5 would phenocopy PRMT5 KD. We established B16 cells expressing methylation defective IFI204 (IFI204Mt1), murine NLRC5, or both (Fig. 7A). Ectopic expression of NLRC5 alone, but not of IFI204Mt1 alone, significantly (P = 0.009) inhibited tumor growth in mice (Fig. 7B). Tumor growth suppression was enhanced in melanoma expressing both NLRC5 and IFI204Mt1, indicating that both pathways mediate antitumor immunity and could contribute to the effects of PRMT5 inhibition. When cells were grown in culture, however, we did not observe growth suppression of lines ectopically expressing IFI204Mt1 and/or NLRC5, consistent with lack of changes observed for PRMT5 inhibition and consistent with the notion that in vivo these factors function in immune recognition of the tumor (Fig. 7C).

• Download high-res image
• Open in new tab
• Download Powerpoint

As we observed for tumors from PRMT5 KD cells in immune competent mice, immunohistochemical analysis of tumors (collected at day 12) showed increased infiltration of CD4⁺ and CD8⁺ T cells in tumors expressing both NLRC5 and IFI204Mt1 (fig. S9A). Moreover, the expression of PRMT1, PRMT5, and PRMT7 decreased in cells expressing both IFI204Mt1 and NLRC5 (fig. S9B), suggesting a possible feed-forward mechanism limiting expression and, concomitantly, activity of other PRMTs.

Expression of IFI16 and NLRC5 is associated with prolonged patient survival

To determine whether our findings were relevant for human melanoma, we analyzed melanoma specimens in the TCGA dataset (n = 368 metastatic melanomas). We performed survival analysis on 100 samples, each with high or low IFI16 or NLRC5 expression (Fig. 7, D and E). Tumors with higher expression of either IFI16 (P = 0.0257) or NLRC5 (P < 0.0001) were associated with a significantly (P = 0.0064 for IFI16 and P < 0.0001 for NLRC5) prolonged survival of patients with melanoma (Fig. 7, D and E). Correspondingly, higher expression of IFI16 or NLRC5 correlated with enrichment of an immune gene signature (fig. S9, C and D), supporting the notion that IFI16 and NLRC5 are important for the antitumor immune response.

PRMT5 inhibition enhances ICT in a murine melanoma model

On the basis of our results, we propose a model for PRMT5 as a suppressor of the antitumor immune response, which is achieved by limiting infiltration or activation (or both) of immune cells and tumor cell recognition by immune cells (Fig. 8A). Consistent with this model, PRMT5 KD tumors had higher expression of Ifnb1, Ccl5, and Cxcl10 (Fig. 8B) and of Pd-l1(Cd274), an immune checkpoint ligand (Fig. 8C). Because enhancing the immune response to so-called “cold” tumors could augment ICT effectiveness (41–43) and considering our finding that PRMT5 KD increased immune cell infiltration into melanoma tumors in immunocompetent mice, we asked whether PRMT5 inhibition enhanced ICT effectiveness. PRMT5 KD (shPRMT5 + IgG) significantly (P = 0.0406, Fisher’s exact test) attenuated B16 tumor growth in six of eight mice compared with one of eight in the control group (Scr + IgG) (Fig. 8D, top).

• Download high-res image
• Open in new tab
• Download Powerpoint

When combined with anti–PD-1 (programmed cell death protein 1) therapy, PRMT5 KD (shPRMT5 + anti–PD-1) significantly (P = 0.0002, Fisher’s exact test) suppressed tumor growth (eight responders of eight), compared with anti–PD-1 treatment alone (Scr + anti–PD-1) (zero responders of eight) (Fig. 8D, top). Although the difference in tumor growth suppression at early time points was not significantly (P = 0.4667, Fisher’s exact test) different between the combination of anti–PD-1 therapy plus PRMT5 KD and PRMT5 KD alone (Fig. 8D, top, and fig. S10A), the long-term survival of mice was significantly better (P = 0.0001) when treated with combination of shPRMT5 (PRMT5 KD) + anti–PD-1 therapy (median survival, 27 days) relative to mice treated with anti–PD-1 alone (median survival, 17.5 days) or PRMT5 KD alone (median survival, 20 days) (Fig. 8D, bottom). These findings indicated improved therapeutic efficacy by the combination of PRMT5 KD with anti–PD-1 therapy.

We next subjected B16 and YUMM1.7 melanoma models to combined therapy, consisting of anti–PD-1 antibodies and the PRMT5i GSK3326595. The dose of the PRMT5i (40 mg/kg) was based on an earlier report (22) and confirmed in the B16 tumor model (fig. S10B). Compared to anti–PD-1 therapy alone, treatment with GSK3326595 plus anti–PD-1 antibodies augmented the antitumor response, reflected in reduced tumor size and survival in both the B16 (Fig. 8E and fig. S10, C and D) and YUMM1.7 (Fig. 8F) tumor models. Note that tumor growth inhibition seen in both B16 or YUMM1.7 models after treatment with anti–PD-1 antibody or PRMT5i alone was limited (one to two responders; Fig. 8, E and F), consistent with the expected response of cold tumors. Changes in tumor burden, which was monitored at different time points in the course of tumor development, revealed a greater response rate (number of mice responding with a <50% reduction in tumor burden, compared to the control group) to the combination therapy (57.1 to 87.5%), than that seen in mice undergoing either monotherapy [anti–PD-1 antibody (12.5 to 33.3%) or PRMT5i (14.3 to 66.7%)] (Fig. 8, E and F, fig. S10C, and table S5). Consistent with the limited response observed after PRMT5i monotherapy, notable changes were not observed in immune cell infiltration or activation (fig. S10, E to G). The effective inhibition in tumor growth observed after combination of PRMT5i with anti–PD-1 antibody was abolished upon the administration of neutralizing antibodies to deplete CD8⁺ cells (Fig. 8G and fig. S10H). These observations confirmed that CD8⁺ cells mediate antitumor immunity elicited by the combination therapy (PRMT5 inhibition with anti–PD-1 therapy). Unlike anti–PD-1 antibody therapy, administration of anti–cytotoxic T lymphocyte associated protein 4 (CTLA-4) antibody did not augment the effect of PRMT5i, compared to control or either treatment alone (fig. S10, I and J), indicating that a subset of immune checkpoint components are regulated by PRMT5.