RAS/MEK/PI3K pathway inhibition augments response to CD40 agonism by targeting CD11b+ Bregs thereby overcoming melanoma PD1-resistance

Abstract

Development of effective second-line treatment options for patients with BRAF^wtNRAS^wt or BRAF^wtNRAS^mut melanoma resistant to immune checkpoint blockade (ICB) is crucial. While systemic delivery of agonist CD40 (aCD40) plus anti-PD1 (αPD1) showed activity in patients with ICB-resistant melanoma, the objective response rate was modest (15%), in part due to induction of B regulatory cells (Bregs) which suppress CD8⁺ effector T cell responses. We previously reported that RAS/RAF/PI3K-inhibition elevates CD40 expression in melanoma cells and sensitizes tumors to ICB. Here, we show that combined treatment with a RAS/PI3K/AKT-pathway inhibitor rigosertib (RGS), and/or a MEK1/2 inhibitor trametinib (T), plus aCD40, overcomes the ICB resistance of BRAF^wtNRAS^wt and BRAF^wtNRAS^mut melanoma tumors growing in C57BL/6 mice. In addition, overexpression of CD40 in these melanoma cells effectively reverses ICB-resistance and aCD40 + αPD1 treatment induces tumor regression. Mechanistically, RGS + T suppress aCD40-associated CD11b⁺PD-L1⁺ Bregs, promoting CD8⁺ T-cell mediated killing in melanoma. scRNA-Seq analyses confirm CD40-associated CD11b⁺ Bregs across cancer types in patients. Our data demonstrate that addition of RAS/PI3K/AKT and MEK inhibitors to aCD40 resolves the issue of aCD40 induction of CD11b⁺PD-L1⁺ Bregs and provides alternative therapeutic options for ICB-resistant BRAF^wtNRAS^wt or BRAF^wtNRAS^mut metastatic melanoma.

Fig. 1. CD11b⁺ regulatory B cells in patient melanoma tumors.

A Immunofluorescence staining in tumors of patients with metastatic melanoma resistant to αPD1 therapy. DNA (blue), CD8α (green), CD20 (red) and CD11b (yellow). Scale bars indicate 2 mm (left) and 50 μm (right). These results are representative of immunofluorescence staining performed on melanoma tumors from two patients. B–F Single-cell (sc) RNA-Seq analysis identifies CD20⁺CD11b⁺PD-L1⁺ regulatory B cells in patient melanoma tumors. References: (B) Cell 2018, 175, 984–997. doi: j.cell.2018.09.006. C Cell 2018, 175, 998–1013.e1–e20. doi: j.cell.2018.12.034. Samples with total B-cell size <5 were excluded. D Frequency of CD20⁺CD11b⁺PD-L1⁺ regulatory B cells in patient melanoma tumors at baseline and post αPD1 treatment. Data were derived from (C). E, F Patient melanoma scRNA-Seq analysis reveals a high CD40 mRNA expression in CD20⁺CD11b⁺PD-L1⁺ Bregs. E n = 818 B cells from 26 patients, two-sided t-test; and (F) n = 1379 B cells from 30 patients. One-way analysis of variance (ANOVA) with post hoc test.

Fig. 2. Agonist CD40 induces CD11b⁺ regulatory B cells in melanoma.

A 1014 cells were injected in C57BL/6 female mice. Treatment of C57BL/6 mice with aCD40 (30 μg/mouse every 3 days, intraperitoneal) started at Day 10 post tumor cell inoculation. Analysis of CD40 levels by flow cytometry of CD45⁺CD19⁺ B cells prepared from tumor samples at day 22 post-treatment (n = 5 per group) using a one-sided Mann–Whitney U test. B Splenocytes were isolated from C57BL/6 mice by magnetic separation, serum-free starved for 4hrs, and stimulated with or without 12.5 μg/ml aCD40. After 4 days of culture, splenic B cells were isolated by magnetic separation and stained for flow cytometric analysis (n = 6 per group). C, D Tumor volume of 1014 melanoma in C57BL/6 female mice. Treatment with 200 μg/mouse αPD1 (every 3 days, intraperitoneal), 100 μg/mouse αCTLA4 (every 3 days, intraperitoneal), 30 μg/mouse agonist CD40 (aCD40, every 3 days, intraperitoneal), or IgG control antibody, starts at day 10 post tumor cell inoculation. E Frequency of CD19⁺CD11b⁺ regulatory B cells in 1014 melanoma tumors or tumor draining lymph node (TDLN) after 22 days of aCD40 treatment (every 3 days, intraperitoneal) TDLN IgG, n = 5; TDLN aCD40, n = 5; Tumor IgG, n = 4; Tumor aCD40, n = 5. F Flow cytometric analysis of CD45⁺CD19⁺ B cells in the tumor microenvironment of 1014 melanoma after 3 weeks treatment of aCD40. Data were replicates from one experiment (n = 5 mice per group). G C57BL/6 mice were intravenously injected with formalin-fixed 2 × 10⁶ metastatic B16F10-Luc2 cells via the tail vein. Splenic CD8⁺ T and total B cells were isolated by magnetic separation after 2 days of tumor antigen priming. Total B cells were stimulated with 12.5 μg/ml aCD40 for 2 days and CD11b⁺ B cells were isolated by magnetic separation. CD8⁺ cytotoxic T cell killing assay were performed via co-culturing CD11b⁺ B cells, CD8⁺ T cells and B16F10-Luc2 cells in the conditioned RPMI complete medium containing 0.5 μg/ml of IL-7, 30U/ml of IL-2, and CD3/CD28 Dynabeads at a bead-to-cell ratio 2:1. Dead cells were removed and the luciferase activity of remaining live tumor cells was quantified to calculate % inhibition of CTL killing capacity. (n = 3 independent experiments). Created in BioRender. Yan, C. (2026) https://BioRender.com/83jrya2. Exact p values are provided as a Source Data file. H Splenic B cells were stimulated with 12.5 μg/ml aCD40 for 4 days. The CD11b^- and CD11b⁺ B cells were isolated by magnetic separation and restimulated with 12.5 μg/ml aCD40 for 30 min. Whole-cell extracts were harvested and immunoblotted. β-actin was used as a loading control for densitometry quantification. n = 3 independent experiments. One-way analysis of variance (ANOVA) with post hoc test. Source data are provided as a Source Data file.

Fig. 3. Anti-PD1 resistant 1014 NRAS^mut melanoma tumors respond to RAS/MEK/PI3K inhibition.

A Tumor volume of 1014 melanoma in C57BL/6 female mice. Treatment with 300 mg/kg rigosertib (RGS, 5 days a week, oral gavage) or vehicle control (H₂O) starts at day 10 post tumor cell inoculation. B Tumor samples were collected from each treatment group at 4 weeks of treatment and immune profiled by FACS analysis. All graphs show Mean ± SEM, n = 5 samples per group. C 1014 tumor cells were treated with 1 μM trametinib (T) and/or 1 μM RGS for 30–60 min. Whole-cell extracts were harvested and immunoblotted. β-actin was used as a loading control for densitometry quantification. n = 3 independent experiments. Source data are provided as a Source Data file. D RGS and T induce CD40 expression in 1014 cells. Cells were treated with RGS + /- T for 24 h. CD40 protein expression on the cell surface was detected by anti-CD40-FITC staining and flow cytometric analysis. n = 3 independent experiments. E Tumor volume of 1014 melanoma in C57BL/6 female mice. Treatment with 200 μg/mouse αPD1 (every 3 days, intraperitoneal), 300 mg/kg rigosertib (RGS, 5 days a week, oral gavage), and/or 1 mg/kg trametinib (T, 5 days a week, oral gavage), starts at day 8 post tumor cell inoculation, n = 7 ~ 10 mice per group. Created in BioRender. Yan, C. (2026) https://BioRender.com/83jrya2. Exact p values are provided as a Source Data file. F Definition of time to resistance to drug: (1) the tumor is > 100 mm³ and (2) has a > 30% increase of tumor volume compared to the previous measurement. Survival curves (resistance-free probabilities or survival probabilities) of treatment groups are estimated using the Kaplan–Meier Plotter and compared using the log-rank test, n = 7 ~ 9 mice per group. Exact p values are provided as a Source Data file. G Mouse body weight recorded for 20 days of treatment. Veh, n = 10; T, n = 8; RGS, n = 7, RGS + T, n = 9 mice. H Tumor and (I) Tumor-draining lymph node (TDLN) samples were collected from each treatment group at day 20 of treatment and immune profiled by FACS analysis, Veh, n = 5; T, n = 4; RGS, n = 4, RGS + T, n = 4 tumors. For tumor samples, we collected 200,000 live singlet cells from each tumor and recovered ~6000 CD45⁺ cells per tumor, regardless of treatment groups. Tc, CD8⁺CD3⁺ cytotoxic T cells. Th, CD4⁺CD3⁺ T helper cells. Exact p values are provided as a Source Data file.

Fig. 4. RAS/MEK/PI3K inhibition selectively attenuated aCD40-associated CD11b⁺ regulatory B cells (Bregs) and promoted T effector cell activity.

A, B Splenocytes were isolated from C57BL/6 mice, serum-free starved for 4 h, and stimulated with 12 μl/ml anti-CD3/CD28 Dynabeads plus 12.5 μg/ml agonist CD40 (aCD40). After 5 days of culture, cells were stained for FACS analysis. Exact p values are provided as a Source Data file. C Tumor volume of 1014 melanoma, created in BioRender. Yan, C. (2026) https://BioRender.com/83jrya2, exact p values are provided as a Source Data file; and D Mouse body weight of tumor-bearing C57BL/6 female mice. Treatment with 200 μg/mouse αPD1 (2 lead-in doses, every 3 days, intraperitoneal), 30 μg/mouse aCD40 (every 3 days, intraperitoneal), plus 300 mg/kg rigosertib (RGS, 5 days a week, oral gavage) and/or 1 mg/kg trametinib (T, 5 days a week, oral gavage), starts at day 8 post tumor cell inoculation (n = 10 per group). E Serum level of liver/kidney enzymes alanine aminotransferase (ALT), aspartate aminotransferase (AST) and blood urea nitrogen (BUN) at day 27 post-treatment (n = 5 per group). Live CD45⁺ leukocytes in (F) Tumor and (G) Tumor-draining lymph node (TDLN) samples were concatenated after downsampling to, 25,000 and 10,000 events respectively, for t-SNE analysis through flow cytometry. All graphs show Mean ± SEM, n = 5 per group. H–K Frequency of B cells and effector cells in 1014 tumors and TDLN samples at day 27 post-treatment (n = 5 per group). Exact p values are provided as a Source Data file. L L308 mouse protein array of 1014 tumor lysate samples at day 27 post-treatment. Data were pooled from replicates of one experiment (n = 5 mice per group). Exact p values are provided as a Source Data file.

Fig. 5. CD11b⁺ regulatory B cells reduce CD8⁺ T cells in B16F10 melanoma and promote melanoma lung metastasis and αPD1-resistance.

A, B Tumor growth, created in BioRender. Yan, C. (2026) https://BioRender.com/83jrya2, exact p values are provided as a Source Data file, and (C) Mouse body weight of B16F10 NRAS^wtBRAF^wt melanoma in C57BL/6 female mice, n = 8 mice per group. Treatment with 200 μg/mouse αPD1 (lead-in 2 doses, every 3 days, intraperitoneal), followed by 30 μg/mouse aCD40 (every 3 days, intraperitoneal), with or without 300 mg/kg rigosertib (RGS, 5 days a week, oral gavage) and 1 mg/kg trametinib (T, 5 days a week, oral gavage), starts at day 8 post tumor cell inoculation. D Tumor samples were collected from each treatment group at day 20 of treatment and immune profiled by FACS analysis. No αPD1-prime groups, Veh + IgG, n = 6, RGS + T + IgG, n = 8, Veh + aCD40, n = 8, RGS + T + aCD40, n = 7 tumors; with αPD1-prime groups, Veh + IgG, n = 5, RGS + T + IgG, n = 5, Veh + aCD40, n = 8, RGS + T + aCD40, n = 8 tumors. Exact p values are provided as a Source Data file. E The baseline frequency of CD11b⁺ B cells and CD8⁺ Tc cells in melanoma tumors at 1-month post tumor cell inoculation (1014 tumors, n = 5; B16F10 tumors, n = 3). Exact p values are provided as a Source Data file. F The lung colonization model was established by intravenously injecting 0.5 × 10⁶ metastatic B16F10-Luc2 cells via the tail vein. Splenic B cells were treated with 12.5 μg/ml aCD40 for 2 days then CD11b⁺ B cells were isolated by magnetic separation for adoptive transfer. Starting at day 2 post tumor cell inoculation, ~2 × 10⁶ CD11b⁺ B cells were intravenously injected via the tail vein (once a week) in 200 μl HBSS buffer. Treatment of 200 μg/mouse αPD1 (every 3 days, intraperitoneal) starts at day 2 post tumor cell inoculation (n = 5 mice per group). Twenty-one days after tumor implantation, tumor lung was weighed and subtracted from tumor-free lungs in mice. Created in BioRender. Yan, C. (2026) https://BioRender.com/83jrya2. Exact p values are provided as a Source Data file. G Flow cytometry, exact p values are provided as a Source Data file, and (H, I) Immunohistochemistry (IHC) staining of B16F10 melanoma lung metastatic tumors after 21 days of treatment. The percentage of positive red IHC staining cells of Ki67 and CD8 was quantified from 20 fields of each FFPE slide sample per group (n = 5 tumors per group). Exact p values are provided as a Source Data file. Tc, CD8⁺CD3⁺ cytotoxic T cells. Th, CD4⁺CD3⁺ T helper cells.

Fig. 6. CD40 overexpression in melanoma cells reduced CD11b⁺ regulatory B cells in tumors, improved CD8⁺ T cell activity, and shifted αPD1-non-responding melanomas into responders.

Treatment with 200 μg/mouse αPD1 (every 3 days, intraperitoneal), 30 μg/mouse aCD40 (every 3 days, intraperitoneal), or 300 mg/kg rigosertib (RGS, 5 days a week, oral gavage), starts at day 8 post tumor cell inoculation, n = 5 mice per group. A Tumor volume, created in BioRender. Yan, C. (2026) https://BioRender.com/83jrya2, exact p values are provided as a Source Data file, and (B) Treatment effect on immunotherapy-resistant NRAS^mut 1014 melanoma growth in C57BL/6 female mice. Exact p values are provided as a Source Data file. C, D Flow cytometric analysis of the 1014 tumors at day 22 post-treatment. Exact p values are provided as a Source Data file. E Tumor volume, created in BioRender. Yan, C. (2026) https://BioRender.com/83jrya2, exact p values are provided as a Source Data file, and F Treatment effect on NRAS^wtBRAF^wt B16F10 melanoma growth in C57BL/6 mice, n = 5 mice per group. Exact p values are provided as a Source Data file. G Flow cytometric analysis of the B16F10 tumors at day 18 post-treatment. n = 5 mice per group. Tc, CD8⁺CD3⁺ cytotoxic T cells. Exact p values are provided as a Source Data file. H Correlation analysis between CD8⁺CD3⁺ Tc cell frequency in total CD45⁺ cells in B16F10 tumors and endpoint tumor weight at day 18 post-treatment.

Fig. 7. Treatment-responding NRAS^mut 1014 tumors exhibited an inflammatory tumor microenvironment with elevated CD8⁺ T cell responses and expansion of selective T cell clonotypes.

A Correlation heatmap of top 25% differentially altered genes summarized across all treatment groups (15,903 genes plotted). Heatmap3 was used for cluster analysis and visualization. Significantly differential expressed genes with absolute fold change ≥ 2 and FDR adjusted p value ≤ 0.05 were detected by DESeq2. B CIBERSORT analysis of a leukocyte signature matrix (LM22) to deconvolution of 1014 tumors. Exact p values are provided as a Source Data file. C, D Hallmark Gene Set Enrichment Analysis (GSEA) of the transcriptomic profile of 1014 tumors at baseline or with agonist CD40 (aCD40) plus αPD1 treatment at 22 days using GSEA package. E–K TCR-β chain sequencing results were evaluated using the Archer Immunoverse analyzer. CDR3 sequences and frequency tables were extracted from the manufacturers’ analysis platform and imported for analysis using the Immunarch package (https://immunarch.com) in R. Comparisons were made within specific tumor types and across different tumor types (CD40-Ctrl vs CD40-OE) to evaluate drug effects, with the analysis conducted on a natural logarithmic scale. Number of (E) total and (F) unique TCR clonotypes. Exact p values are provided as a Source Data file. G Treatment effect (ratio of TCR clonotypes among treatment to clonotypes among Veh + IgG control) comparisons between CD40-Ctrl and CD40-OE tumors were performed using multiple comparisons test for generalized linear hypothesis test based on the negative Binomial generalized linear model. All graphs show Mean ± SD. H Estimated distribution of clonotype abundances. I Table summary of FDR (False Discovery Rate) adjustment. J Estimated relative abundance (also known as clonal space homeostasis), which characterizes the proportion of repertoire occupied by clonal groups with specific abundances indicated. K The top 20 most abundant TCRs of small- and medium-expanded clonotypes in TRBV26 are shown. Flow between samples indicates shared TCRs. Boxed CDR3s indicate most clonal TCRs in TRBV26. The Holm correction was used to adjust the p value for multiple comparisons. A–K Pooled values of n = 3 tumors per group.

Fig. 8. RAS/MEK/PI3K inhibition plus immunotherapy attenuated the growth of patient-derived organoids (PDO), while systemic CD40 activation is associated with protumor CD20⁺CD11b⁺PD-L1⁺ B cells in patient melanoma tumors.

A PDO cultures were developed using previously frozen patient melanoma tissue that was thawed and immediately subjected to fine needle aspiration. IL-2 was added to all cultures to ensure the survival and proliferation of T cells. PDOs were grown in organoid media containing 5% Matrigel. After allowing PDOs to establish for 3–4 days, treatments were initiated as follows: RGS, CD40 agonist (aCD40), αPD1, RGS + aCD40, RGS + αPD1, or RGS + aCD40 + αPD1. Cultures were allowed to grow for 10 days, and the organoid area was assessed and quantified at day 10, PDO-2624, n = 18 per group; PDO-3453A, n = 18 per group; PDO-3529 primary, n = 12 per group; PDO-3529LN, n = 12 per group; PDO-3101, n = 12 per group. All graphs show Min to Max with all points +/− SD. Exact p values are provided as a Source Data file. B, C, F Melanoma scRNA-Seq dataset. Analysis of malignant and immune cell gene expression of 6173 scRNA-seq profiles from 32 human melanoma tumors. D, E, G Melanoma scRNA-Seq dataset. Transcriptome analysis of 16,291 individual immune cells from 49 tumor samples of melanoma patients treated with checkpoint inhibitors. C, E Functional enrichment of Geneset and KEGG pathway over-representation analysis were performed on differentially expressed genes of CD20⁺CD11b⁺PD-L1⁺ Bregs in patient melanoma tumors using WebGestaltR package (NULL). F, G Patient melanoma scRNA-Seq analysis reveals a distinct transcriptome profile of CD20⁺CD11b⁺PD-L1⁺ Bregs. All graphs show Median ± SD. H The Breg gene signature (Sig. Bregs) is associated with poorer patient OS in multiple cancer types. Kaplan-Meier plots show OS for patients in the TCGA pan-cancer datasets. Patients were split into high and low expression of the Breg gene curated signature. Median OS survival is represented on the graph when available. P-values were calculated by log-rank test. GBM, glioblastoma; LGG, low-grade glioma; LUSC, lung squamous cell carcinoma; STAD, stomach adenocarcinoma; MESO, mesothelioma; OV, ovarian carcinoma.