Novel induction of CD40 expression by tumor cells with RAS/RAF/PI3K pathway inhibition augments response to checkpoint blockade

Abstract

Background: While immune checkpoint blockade (ICB) is the current first-line treatment for metastatic melanoma, it is effective for ~ 52% of patients and has dangerous side effects. The objective here was to identify the feasibility and mechanism of RAS/RAF/PI3K pathway inhibition in melanoma to sensitize tumors to ICB therapy.

Methods: Rigosertib (RGS) is a non-ATP-competitive small molecule RAS mimetic. RGS monotherapy or in combination therapy with ICB were investigated using immunocompetent mouse models of BRAFwt and BRAFmut melanoma and analyzed in reference to patient data.

Results: RGS treatment (300 mg/kg) was well tolerated in mice and resulted in ~ 50% inhibition of tumor growth as monotherapy and ~ 70% inhibition in combination with αPD1 + αCTLA4. RGS-induced tumor growth inhibition depends on CD40 upregulation in melanoma cells followed by immunogenic cell death, leading to enriched dendritic cells and activated T cells in the tumor microenvironment. The RGS-initiated tumor suppression was partially reversed by either knockdown of CD40 expression in melanoma cells or depletion of CD8+ cytotoxic T cells. Treatment with either dabrafenib and trametinib or with RGS, increased CD40+SOX10+ melanoma cells in the tumors of melanoma patients and patient-derived xenografts. High CD40 expression level correlates with beneficial T-cell responses and better survival in a TCGA dataset from melanoma patients. Expression of CD40 by melanoma cells is associated with therapeutic response to RAF/MEK inhibition and ICB.

Conclusions: Our data support the therapeutic use of RGS + αPD1 + αCTLA4 in RAS/RAF/PI3K pathway-activated melanomas and point to the need for clinical trials of RGS + ICB for melanoma patients who do not respond to ICB alone.

Trial registration: NCT01205815 (Sept 17, 2010).

Keywords: CD40; Immune checkpoint blockade; Immunogenic cell death; Melanoma; RAS/RAF/PI3K.

Conflict of interest statement

EPR is a member of the Board of Directors of Onconova, receives grant funding from Onconova, and is a paid consultant for Onconova. All other co-authors have declared that no additional conflict of interest exists.

Figures

Fig. 1

Rigosertib induces melanoma cell death and inhibits melanoma tumor growth in vivo. a IC50 determined by CellTilter Blue assay at 72 h post RGS treatment. The melanoma and normal cells were seeded in 96-well plate with 3000 cells per well. b Dot plots and quantified results of the Annexin V+ and 7-AAD+ cell percentage in YUMM3.3 culture with or without RGS treatment. c Whole-cell extracts were harvested and immunoblotted. HSP90 was used as a loading control for densitometry quantification (red numbers). d, e Tumor volume and weight of YUMM3.3 melanomas in C57BL/6 female mice. Daily RGS treatment starts at day 10 post tumor cell inoculation. CR, complete regression. f Representative images and quantitative results of cleaved-caspase3 protein levels observed from day 17 post RGS treatment (300 mg/kg) of YUMM3.3 tumors by immunohistochemistry. g Gene ontology enrichment analysis of gene targets downregulated in YUMM3.3 tumors post 17 days of RGS treatment (300 mg/kg). Results for Fold Enrichment > 3 and FDR p < 0.05 are displayed. h Heatmap summarizes 21 p53-associated targets from a total of 1495 genes screened that were significantly downregulated due to RGS treatment. a-f pooled data obtained from at least two different experiments (n = 4 ~ 15) are shown. g, h were triplicates

Fig. 2

Rigosertib increases the frequency of dendritic cells in tumors and tumor-draining lymph nodes (TDLN). YUMM3.3 tumors and TDLN were collected 17 days post-treatment. a Total number and density of CD45+ leukocytes in tumors were obtained by flow cytometry. b Gating strategy for dendritic cells in tumors and TDLN. c-e Frequency and mean fluorescent intensity (MFI) of dendritic cell subsets and their correlation with tumor burden. Pooled data obtained from at least two different experiments (n = 5 ~ 10) are shown

Fig. 3

Rigosertib promotes T cell and NK cell responses but attenuates tumor-associated M2 macrophages in the tumor microenvironment. YUMM3.3 tumors were collected 17 days post-treatment. a, h Live CD45+ leukocytes were concatenated after downsampling to 20,000 events for subsequent high-dimensional data analysis to normalize the contribution among samples under different treatments. Samples were then analyzed in parallel by t-SNE and manually gated leukocyte populations were overlaid onto the total t-SNE map using FlowJo 10.5.3. b-e, g, i Flow cytometric and f IHC analysis of YUMM3.3 tumors at day 17 post-treatment. j Flow cytometric tSNE analysis of PBMC samples from YUMM3.3 tumor-bearing mice treated with RGS before and after CD4 and/or CD8 antibody depletion treatments. (K) Tumor weight (day 16 post RGS treatment) of YUMM3.3 tumors in C57BL/6 mice with and without CD4 and/or CD8 depletion. a-i pooled data obtained from at least two different experiments (n = 5 ~ 10) are shown. j, k were replicates (n = 10 per group)

Fig. 4

Rigosertib improves T-cell responses to αPD-1/αCTLA-4 therapy. a Schematic of ICB non-responsive B16F10 melanoma model. b-d Survival of C57BL/6 female mice, day 11 tumor weight post treatment and tumor volume of B16F10 tumors. Mice reached end point and sacrificed when tumors in the experiment exceeded 15 mm in diameter or became perforated. The experiment was terminated when no mice survived in groups other than the RGS + ICB group. e Analysis of synergy. Interaction plot for tumor growth rate with 95% confidence interval by treatment. f Live CD45+ leukocytes were concatenated after downsampling to ~ 12,500 events for t-SNE analysis through flow cytometry and T-cell frequency was shown. g Tumor volume of YUMM3.3 tumors in C57BL/6 female mice. Treatment starts at day 10 post tumor cell inoculation. h Spider plots of tumor volume changes overtime for individual YUMM3.3 tumors. i IHC and j-l Flow cytometric analysis of YUMM3.3 tumors at day 17 post-treatment. Data were replicates from one experiment (n = 10 per group)

Fig. 5

Rigosertib induces CD40 on melanoma cells and promotes the anti-tumor immunity. a, b Cytokine array of YUMM3.3 tumor lysate samples at day 17 post-treatment. c Live cells were concatenated after downsampling to ~ 20,000 events for t-SNE analysis through flow cytometry. d Flow cytometric analysis of YUMM3.3 tumors in ROSA reporter mice at day 14 post RGS treatment. e Mean fluorescence intensity (MFI) on the CD45−CD40+ cells isolated from day 17 YUMM3.3 tumors in C57BL/6 mice. f, g Melanoma cells were treated with indicated drugs for 48 h and CD40 expression was detected by flow cytometry. h, i Viability of YUMM3.3 and B16F10 cells was examined by flow cytometry. j Flow cytometric analysis of CD40 expression in response to treatment with IFNγ (500 U/ml, 48 h) in different clones of YUMM3.3 cells where shRNAs targeted coding sequence (CDS), 3″ untranslated region (UTR), or random sequences (Ctrl). k Tumor volume of CD40 knockdown clones in C57BL/6 mice was determined (n ≥ 4 mice per group). l Tumor weight and tumor volume of YUMM3.3 cells growing in C57BL/6-CD40 knockout male and female mice treated with RGS (300 mg/kg). a-e data were replicates from one experiment (n = 5 ~ 10 per group). f-l pooled data obtained from at least two different experiments (n = 3 ~ 6) are shown

Fig. 6

RGS and BRAF/MEK inhibitors induce CD40 expression on melanoma cells and tumors of patients responsive to RAS/RAF/MEK pathway inhibition. a NRASmut melanoma PDX 1179 and PDX1214 tumor growth over time in NSG mice with IHC staining of cleaved-caspase3, CD40 and SOX10 in from mice treated with the vehicle or 300 mg/kg rigosertib (RGS). 20X and 40X images and the scale marker are shown. b IHC staining of CD40 and SOX10 in BRAFmut melanoma PDX tumors from mice treated with the vehicle or Dabrafenib + Trametinib (D + T). c Correlation analysis of T. Ratio and CD40 fold-change post treatment in PDXs shown in b. The T. Ratio, obtained from the statistical analysis of treatment difference comparisons of the tumor growth rate based on the tumor volume, for each PDX treatment comparison. d H&E and multiplex sequential IHC analysis of hematoxylin, CD40, CD80, CD11c, CD8 and SOX10 of patient melanoma sections. Cell number of SOX10+CD40+ cells was identified of 11 paired tumor sections pre- and post-BRAF inhibitor treatment (paired t-test). Pooled data (n = 4 ~ 11) from one experiment are shown

Fig. 7

Correlations between CD40 expression in human melanoma and response to RAF inhibitors and ICB. a Correlations of CD40 gene expression with CD80 and ICOS-L expression from melanoma tumors in the TCGA dataset (n = 480) in cBioPortal. b Overall survival in reference to CD40, CD80 and ICOSL copy number and mutation burden in melanoma patients in TCGA-Melanoma dataset (n = 358). c Correlation matrix with Spearman’s rank correlation coefficients after data transformation analysis of CD40, CD4, CD8, IFNγ, GzmB, IL-2, CXCL9, IL-5 and IL-8 (y = (log(x-min(x) + 1))). d Drug sensitivity analysis of Barretina CellLine 2 (DNA) and Garnett CellLine (mRNA) human cancer cell line datasets in Oncomine database. e, f Analysis of gene expression levels and their correlation to ICB resistance. Gene expression in defined melanoma cells (n = 1883) were downloaded and analyzed based on single cell RNA-Seq (https://singlecell.broadinstitute.org/)