Melanoma dedifferentiation induced by IFN-γ epigenetic remodeling in response to anti-PD-1 therapy

Abstract

Melanoma dedifferentiation has been reported to be a state of cellular resistance to targeted therapies and immunotherapies as cancer cells revert to a more primitive cellular phenotype. Here, we show that, counterintuitively, the biopsies of patient tumors that responded to anti-programmed cell death 1 (anti-PD-1) therapy had decreased expression of melanocytic markers and increased neural crest markers, suggesting treatment-induced dedifferentiation. When modeling the effects in vitro, we documented that melanoma cell lines that were originally differentiated underwent a process of neural crest dedifferentiation when continuously exposed to IFN-γ, through global chromatin landscape changes that led to enrichment in specific hyperaccessible chromatin regions. The IFN-γ-induced dedifferentiation signature corresponded with improved outcomes in patients with melanoma, challenging the notion that neural crest dedifferentiation is entirely an adverse phenotype.

Trial registration: ClinicalTrials.gov NCT01621490.

Keywords: Cancer immunotherapy; Cytokines; Oncology.

Conflict of interest statement

Conflict of interest: KL reports receiving speaking fees from Roche Tissue Diagnostics. CS has received grant support from Pfizer, AstraZeneca, BMS, Roche-Ventana, Boehringer-Ingelheim, and Ono. CS has consulted for Pfizer, Novartis, GlaxoSmithKline, MSD Pharmaceuticals, Bristol Meyers Squibb (BMS), Celgene, AstraZeneca, Illumina, Genentech, Roche-Ventana, GRAIL, Medicxi, and the Sarah Cannon Research Institute. CS is a shareholder of Apogen Biotechnologies, Epic Bioscience, and GRAIL, and has stock options in and is co-founder of Achilles Therapeutics. TGG has received an honorarium from Amgen and has consulting and equity agreements with Trethera Corporation. The laboratory of TGG has completed a research agreement with ImmunoActiva. AR has received consulting fees from Amgen, Bristol-Myers Squibb, Chugai, Genentech, Merck, Novartis, Roche, and Sanofi; has been a member of the scientific advisory boards of and holds stock in Advaxis, CytomX, Five Prime Therapeutics, Highlight Therapeutics, Kite-Gilead, and RAPT Therapeutics; is a current member of the scientific advisory boards of and holds stock in Apricity, Arcus Biosciences, Compugen, ImaginAb, IsoPlexis, Lutris Pharma, Merus, PACT Pharma, Rgenix, Synthekine and Tango Therapeutics; and has received research funding from Agilent and Bristol-Myers Squibb through Stand Up to Cancer (SU2C).

Figures

Figure 1. Human melanoma dedifferentiation is associated with a response to anti–PD-1 therapy.

(A) MITF, (B) MLANA, and (C) AXL gene expression levels in pre- and post-treatment biopsies from patients with PD, SD, or CRPR.

Figure 2. Human melanoma dedifferentiation is induced by exposure to IFN-γ.

(A) Flow cytometric data for MART-1 and NGFR in the human melanoma cell lines M262 (baseline differentiated) and M370 (baseline undifferentiated) in response to TNF or IFN-γ exposure. (B) Projection of cytokine-treated cell lines onto melanoma M series differentiation PCA (6). Diff, baseline differentiated; undiff, baseline undifferentiated. (C) Expression of melanoma differentiation genes for 0 hour, IFN-γ, and TNF across cell lines. U, undifferentiated at baseline, U-NC, undifferentiated neural crest–like, NC, neural crest–like, NC-T, neural crest–like transitory, T, transitory, T-M, transitory melanocytic, M, melanocytic. Colors represent z scores. (D) Common melanoma mutations across cell line studies. Nonsense or missense JAK/STAT mutations were not observed.

Figure 3. IFN-γ and TNF stimulation induces the expression of common genes across cell lines to generate comparable MART-1–low/NGFR-high dedifferentiation states.

(A) Varimax-rotated PLSR on IFN-γ–exposed samples compared with 0-hour (untreated) samples. (B) Left: Genes contributing to a common IFN-γ response across samples. Right: K-means clustering of top 300 gene loadings. Left column shows untreated and right column shows after IFN-γ exposure for each cell line. (C) Varimax-rotated PLSR on TNF compared with untreated samples. (D) Left: Genes contributing to a TNF response across samples. Right: K-means clustering of top 300 gene loadings. Left column shows untreated and right column shows after TNF exposure for each cell line. (E) Overlap of IFN-γ– and TNF-induced gene expression by ranked loadings. (F) Concordant GO term overlap (NESs) between IFN-γ– and TNF-induced gene expression. The number shown at the top of E and F is the maximum –log10(P value) of the RRHO heatmap. (G and H) Enrichment of gene sets involving pigmentation, mitosis, transcription, IFN signaling, and cytokines following IFN-γ or TNF exposure. *P < 0.1, **P < 0.01, and ***P < 0.001, by signed KS test.

Figure 4. IFN-γ, compared with TNF, alters the chromatin landscape in a stimulus-specific manner.

(A) Examples of hyperaccessible peaks upon cytokine stimulation. (B) Total number of hyper- and hypoaccessible peaks called for each listed comparison. U, undifferentiated at baseline; D, differentiated at baseline. (C) PCA of peaks differentially hyperaccessible from baseline after cytokine treatment. (D) K-means clustered heatmap of induced ATAC-Seq peaks across any stimulation condition for differentiated and undifferentiated melanomas (subcolumns are in the order 0 hour, IFN-γ, and TNF for each cell line). (E) Motif enrichment of IFN-γ– compared with TNF-induced genes. bZIP, basic leucine zipper domain; RHD, rel homology domain. (F) Top divergent GO terms of nearby genes for IFN-γ– versus TNF-specific peaks.

Figure 5. The basal chromatin landscapes of differentiated and undifferentiated lines result in distinct epigenomic responses upon cytokine stimulation.

(A) Overlap of induced IFN-γ and TNF ATAC-Seq peaks. (B) Overlap of peaks separated by cell line baseline state. (C) Heatmap of differentially IFN-γ–inducible peaks for baseline differentiated and undifferentiated cell lines, with the top motif for each cluster listed (subcolumns are in the order 0 hour, IFN-γ, and TNF for each cell line), and heatmap of differentially TNF-inducible peaks for baseline differentiated and undifferentiated lines, with the top motif for each cluster listed. (D) Motif enrichment of IFN-γ– compared with TNF-inducible peaks for baseline differentiated and undifferentiated cell lines separately. Colors represent q values.

Figure 6. Differentiated and undifferentiated cell lines respond to cytokine stimulation with differences in inferred activity of both signal-dependent and lineage-determining transcription factors.

(A and B) VIPER analysis showing inferred transcription factor activity for baseline differentiated versus undifferentiated cell lines following (A) TNF or (B) IFN-γ exposure. Regulators such as PRDM1, HMGA1, and SOX9 had high inferred activity only in the baseline differentiated group.

Figure 7. Gene expression differences between differentiated and undifferentiated cell lines may be attributed to lipid, ribosomal, mitochondrial, and adhesion processes.

(A and B) Enrichment of gene set groups (C5: GO gene sets), based on ranked lists of differentially expressed genes, for TNF and IFN-γ.*P < 0.1, **P < 0.01, and ***P < 0.001, by signed KS test.

Figure 8. Enrichment of IFN-γ–induced dedifferentiation gene signatures in melanomas correlates with the response to anti–PD-1 and better overall survival.

(A) Enrichment of the dedifferentiation signature in the paired pre- and post-treatment biopsies (n = 68) from responders and nonresponders in the CheckMate 038 biopsy cohort. (B) Enrichment of the dedifferentiation signature in the baseline biopsies from the CheckMate 038 biopsy cohort, including the paired and unpaired biopsies (n = 101), from responders and nonresponders. (C) Correlation of baseline enrichment of the dedifferentiation signature (Sig) with overall survival in TCGA melanoma data set. med, medium.

Figure 9. Survival plots.

Survival plots based on the analysis of (A) MITF, (B) MLANA, and (C) AXL. Analysis of the baseline biopsy samples (n = 84) from patients treated with nivolumab with or without ipilimumab in the CheckMate 038 trial. The graphs separate the upper and lower quartiles of the log2 FPKM of each gene.