Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 Mar 18;14(1):1516.
doi: 10.1038/s41467-023-36979-y.

The molecular and functional landscape of resistance to immune checkpoint blockade in melanoma

Affiliations

The molecular and functional landscape of resistance to immune checkpoint blockade in melanoma

Su Yin Lim et al. Nat Commun. .

Abstract

Resistance to immune checkpoint inhibitor therapies in melanoma is common and remains an intractable clinical challenge. In this study, we comprehensively profile immune checkpoint inhibitor resistance mechanisms in short-term tumor cell lines and matched tumor samples from melanoma patients progressing on immune checkpoint inhibitors. Combining genome, transcriptome, and high dimensional flow cytometric profiling with functional analysis, we identify three distinct programs of immunotherapy resistance. Here we show that resistance programs include (1) the loss of wild-type antigen expression, resulting from tumor-intrinsic IFNγ signaling and melanoma de-differentiation, (2) the disruption of antigen presentation via multiple independent mechanisms affecting MHC expression, and (3) immune cell exclusion associated with PTEN loss. The dominant role of compromised antigen production and presentation in melanoma resistance to immune checkpoint inhibition highlights the importance of treatment salvage strategies aimed at the restoration of MHC expression, stimulation of innate immunity, and re-expression of wild-type differentiation antigens.

PubMed Disclaimer

Conflict of interest statement

M.S.C. is a consultant advisor to MSD, BMS, Novartis, Roche, Pierre Fabre, Sanofi, Merck Serono, Nektar, Eisia, and Ideaya and received honoraria from MSD, BMS, and Novartis. J.H.L. has received honoraria from AstraZeneca and travel support from BMS and Novartis. R.A.S. has received fees for professional services from F. Hoffmann-La Roche Ltd, Evaxion, Provectus Biopharmaceuticals Australia, Qbiotics, Novartis, Merck Sharp & Dohme, NeraCare, AMGEN Inc., Bristol-Myers Squibb, Myriad Genetics, and GlaxoSmithKline. G.V.L. is a consultant advisor for Aduro Biotech Inc, Agenus Inc, Amgen Inc, Array Biopharma Inc, Boehringer Ingelheim International GmbH, Bristol-Myers Squibb, Evaxion Biotech A/S, Hexel AG, Highlight Therapeutics. S.L., Merck Sharpe & Dohme, Novartis Pharma AG, OncoSec, Pierre Fabre, QBiotics Group Limited, Regeneron Pharmaceuticals Inc, SkylineDX. B.V., Specialised Therapeutics Australia Pty Ltd. A.M.M. has participated in advisory boards for BMS, MSD, Novartis, Roche, and Pierre Fabre. The remaining authors declare no competing interests. R.P.M.S. has received honoraria for advisory board participation from MSD, Novartis, and Qbiotics and speaking honoraria from BMS and Novartis.

Figures

Fig. 1
Fig. 1. IFNγ response in PD1 PROG cell lines.
A ssGSEA scores for the Reactome_Interferon_Gamma_Signaling gene set in 22 PD1 PROG melanoma cell lines treated with BSA control (x-axis) or 1000 U/ml IFNγ (y-axis) for 24 h. The SCC16-0016 PD1 PROG did not respond to IFNγ stimulation and is highlighted. B JAK2 transcript expression in matched BSA control and IFNγ-treated PD1 PROG cell lines (n = 22). C Accumulation of phosphorylated STAT1 (p-STAT1Y727), STAT1, JAK1, and JAK2 in the SCC11-0270 (responded to IFNγ) and SCC16-0016 (did not respond to IFNγ) PD1 PROG cell lines treated with BSA control, IFNα, IFNβ, or IFNγ (all at 1000 U/ml) for 24 h. D Representative histograms showing levels of MHC-I, MHC-II, PD-L1, and PD-L2 expression in SCC16-0016 cell line after BSA control (blue) or 1000 U/ml IFNγ (red) treatment for 72 h. Fluorescence minus one (FMO) controls are shown as shaded histograms. E Representative proliferation curve of SCC16-0016 cells, measured as percent confluence every 4 h, for up to 120 h, after treatment with BSA control, IFNα, IFNβ, or IFNγ (all at 1000 U/ml). Data shown are mean ± sd (six images per treatment per time point). Area under the curve data of biological triplicates were compared using one-way ANOVA with Dunnett correction for multiple comparisons, and adjusted p-values are shown. F Cell surface expression (median fluorescence intensity stained divided by fluorescence minus one control, MFI/FMO) of MHC-I in SCC11-0270 and SCC16-0016 PD1 PROG cell lines treated with BSA control, IFNα, IFNβ, or IFNγ (all at 1000 U/ml) for 72 h. Data shown are mean ± sd (n = 3 biologically independent experiments) and were compared using one-way ANOVA with Dunnett correction for multiple comparisons, and adjusted p-values are shown. Source data are provided as a Source Data file.
Fig. 2
Fig. 2. Intrinsic IFNγ signaling in seven PD1 PROG cell lines.
A ssGSEA score for Reactome_IFNγ_Signaling in BSA control and IFNγ-treated PD1 PROG cell lines with (n = 6; orange) and without (n = 15; blue, not including SCC16-0016) intrinsic IFNγ activity. The dotted line aligns with the lowest ssGSEA score in the IFNγ-treated PD1 PROGs, and solid lines indicate the median ssGSEA scores. B Subset of top-scoring gene sets (GSEA Pre-Ranked; Hallmark gene set collection and melanoma-specific transcriptome signatures) in PD1 PROGs with (n = 6) compared to without (n = 15) intrinsic IFNγ activity (Supplementary Data S1). C Hierarchical clustering of PD1 PROG cell lines (n = 22) with Euclidean distance based on protein expression of Melan-A, MITF, SOX10, and AXL. The transcriptome melanoma clusters defined according to ref. are also shown. PD1 PROG cell lines with intrinsic IFNγ signaling (n = 6) are highlighted in red. D Plots showing IRF1 protein expression (derived from the densitometric normalized protein data after log2 transformation and z score calculation) in PD1 PROGs with (n = 6) and without (n = 5) intrinsic IFNγ activity. The relative cell surface expression (median fluorescence intensity stained divided by fluorescence minus one control, MFI/FMO) of PD-L1, PD-L2, and MHC-I in PD1 PROGs with (n = 6) and without (n = 16) intrinsic IFNγ activity. Data compared using two-sided Mann–Whitney test, p-values shown. E Differentially expressed secreted cytokines (FDR-adjusted p-value < 0.1, dotted line) in control (BSA)-treated PD1 PROGs with (IFNγ-high; n = 6) or without (IFNγ-low; n = 15) intrinsic IFNγ signaling are highlighted in red. The comparison of secreted cytokine expression was performed using log2 transformed fluorescence intensity values (Supplementary Data S3). F Differentially expressed secreted cytokines (FDR-adjusted p-value < 0.1) in control (BSA)-treated PD1 PROGs versus IFNγ-treated PD1 PROG cell lines (n = 21; 1000 U IFNγ/ml for 24) are highlighted in red. The comparison of cytokine expression was performed using log2 transformed fluorescence intensity values (Supplementary Data S4). Source data are provided as a Source Data file.
Fig. 3
Fig. 3. Melanoma and immune cell content in PD1 PROG tumor dissociates.
A Immune cell content of PD1 PROG tumor samples profiled by flow cytometry, the indicated percentage of CD45+ cells in each dissociate (left panel) and immune cell subtypes are shown as a percentage of CD45+ cells (right panel). PD1 PROG tumor dissociates with persistent baseline IFNγ signaling (in the matched PD1 PROG cells) are highlighted in red. The dotted vertical line indicates the median percentage of CD45+ cells (12.2% of total cells). TCR, T cell receptor, TAMs, tumor-associated macrophages, G-MDSCs, granulocytic myeloid-derived suppressor cells, DC, dendritic cell, NK, natural killer cell, Treg, regulatory T cells. See Supplementary Table S5 for subset identification. B Scatterplots showing correlation of PD1++Tbet+GzB+Ki67+, 4-1BB+, and CTLA4+LAG3+ CD8+ T cell fractions in tumor biopsies with the ssGSEA scores for Reactome_IFNγ_Signaling derived from matching PD1 PROGs. Correlation was calculated using a two-sided Spearman’s rank correlation coefficient, exact p-values shown. Samples highlighted in red indicate PD1 PROGs with intrinsic IFNγ signaling. C Scatterplots showing correlation of PD1++Tbet+GzB+Ki67+ CD8+ T cell fractions in tumor biopsies with regulatory T (Treg, FOXP3+) cells in the immune (CD45+) subset from matching PD1 PROGs. Correlation was calculated using a two-sided Spearman’s rank correlation coefficient, exact p-values shown. Samples highlighted in red indicate PD1 PROGs with intrinsic IFNγ signaling. D Representative flow cytometric profile of SCC16-0016 tumor dissociate showing low MHC-I and PD-L1 expression in the melanoma (CD45SOX10+) subset and presence of B (CD19+), NK (CD244+/CD16+/), CD8+, CD4+ conventional (Tconv, FOXP3) and regulatory T (Treg, FOXP3+) cells in the immune (CD45+) subset. Blue boxes show frequencies of gated cells (% CD45+), black boxes show the frequency of T cells expressing the activating receptors as defined by Boolean gating. Plots display flow cytometry data for the percentage of CD8+ T, Treg, and NK cells in the CD45+ population and the ratio of CD8/Treg in tumor dissociates (n = 19). Source data are provided as a Source Data file.
Fig. 4
Fig. 4. Melanoma de-differentiation in immune checkpoint resistant melanoma.
A Plots showing the percentage of NGFR+ melanoma cells (n = 20), proliferating Ki67+ melanoma cells (n = 18) in the melanoma (CD45-SOX10+) subset, and PD1++Tbet+ CD8+ T (n = 17) cells in the CD8+ T cell subset in tumor dissociates from matched PD1 PROG cell lines. The samples are organized according to the differentiation phenotype of the matched PD1 PROG cell line. Horizontal bars indicate the median. Data compared using two-sided Mann–Whitney test, p-values shown. B Representative scatterplots showing reactivity of CD8+ T cells (CD107+/IFNγ+) to the indicated PD1 PROG cell lines after pulsing with DMSO control or 10 µg/ml Melan-A peptide (AAGIGILTV) 2 h before co-culture with HLA-A02:01+ CD8+ T cells derived from the WMD-084 tumor dissociate. Bar graph shows the mean ± sd difference (Melan-A peptide pulsed minus control pulsed) in percentage of CD107+/IFNγ+ CD8+ T cells (n = 3 for SMU14-0301 and WMD-084#1 and n = 4 for WMD-084#2 biologically independent experiments). The change in CD8+ T cell reactivity in the patient-matched WMD-084 #1 and #2 PD1 PROG cells was compared using an unpaired, two-tailed t-test, exact p-values shown. C Cell surface expression (median fluorescence intensity stained divided by fluorescence minus one control, MFI/FMO) of MHC-I in the indicated cell lines at 72 h after treatment with BSA control or IFNγ (1000 U/ml). Mean ± sd average of three biological replicates shown for each cell line. Source data are provided as a Source Data file.
Fig. 5
Fig. 5. MHC-I expression in PD1 PROG cell lines.
A Cell surface expression (median fluorescence intensity stained divided by fluorescence minus one control, MFI/FMO) of MHC-I in PD1 PROG cell lines (n = 22) at 72 h after treatment with BSA control or IFNγ (1000 U/ml). Average of at least three biological replicates shown for each cell line. B B2M transcript expression at 24 h after treatment of PD1 PROG cell lines (n = 22) with BSA control or IFNγ (1000 U/ml). C Cell surface expression (median fluorescence intensity stained divided by fluorescence minus one control, MFI/FMO) of B2M on the SCC16-0016 (n = 4 biological replicates), SMU-092 (n = 3 biological replicates), SCC13-0156 (n = 3 biological replicates), and SMU13-0183M3 (n = 4 biological replicates) PD1 PROG cell lines at baseline. Individual values and mean ± sd of biological replicates are shown. D Accumulation of B2M protein in 5 PD1 PROG cell lines 24 h after treatment with BSA control (−) or IFNγ (1000 U/ml). Experiment repeated independently three times. E Representative histograms showing melanoma MHC-I expression in tumor dissociates of B2M-null SCC13-0156 and SMU-092 (unshaded red histograms) compared to autologous TILs (blue). MHC-I expression on B2M-wild-type SCC15-0111 and WMD-084 is shown for comparison (shaded red histograms). F Representative scatterplots showing reactivity of autologous CD8+ T cells (CD107+/IFNγ+) after pre-treating the SMU17-0132 (unrelated melanoma cell line) and SMU17-0263 (PD1 PROG) melanoma cells for 1 h with IgG2a isotype control or 10 µg/ml MHC-I blocking antibody. The expression of CD107 and IFNγ in T cell mono-cultures (left panels) was used to establish the gating strategy for these experiments. Bar graph shows percentage of CD107+ ± IFNγ+ CD8+ T cells (mean ± sd, (SMU-0132, n = 3; SMU17-0263, n = 4 biological replicates)) after treatment with isotype control (−) or MHC-I blocking antibody (+). The percentage CD107+ ± IFNγ+ CD8+ T cells was compared using a paired, two-tailed t-test, exact p-values shown. Source data are provided as a Source Data file.
Fig. 6
Fig. 6. MHC-II expression is silenced in PD1 PROG cell lines.
A Expression (median fluorescence intensity of at least three biological replicates for each cell line divided by fluorescence minus one control, MFI/FMO) of MHC-II on PD1 PROGs (n = 22), 72 h after treatment with BSA control or IFNγ (1000 U/ml). SCC16-0016 and PD1 PROGs with low MHC-II expression are highlighted in orange. B Representative histograms showing melanoma MHC-II expression in tumor dissociates of B2M-mutant SCC13-0156 and SMU-092 (unshaded red histograms) compared to autologous TILs (blue). MHC-II expression on B2M-wild-type SCC15-0111 and WMD-084 shown for comparison (shaded red histograms). C Accumulation of CIITA protein in 5 PD1 PROGs, 24 h after treatment with BSA control (−) or IFNγ (1000 U/ml). Experiment repeated independently at least three times. D Scatterplot showing correlation of CIITA transcript expression pre and post IFNγ treatment in 22 PD1 PROG cell lines. PD1 PROGs with low CIITA expression (n = 5) are highlighted in orange. Correlation calculated using two-sided Spearman’s rank correlation coefficient, exact p-values shown. E Scatterplot showing correlation of MHC-II expression (MFI/FMO, MFI post IFNγ treatment) with expression of CIITA transcript (post IFNγ treatment) in PD1 PROGs (n = 22). PD1 PROGs with low MHC-II expression are highlighted in orange. Correlation calculated using two-sided Spearman’s rank correlation coefficient, exact p-values shown. F Accumulation of CIITA protein in the SCC11-0270 and SMU17-0263 PD1 PROGs, 72 h after treatment with BSA control, panobinostat (HDACi, 25 nM) in the presence or absence of IFNγ (1000 U/ml). Experiment repeated independently at least three times. G Representative histograms showing melanoma MHC-II expression in MHC-IIlow SCC17-0263 and SCC11-0270 PD1 PROGs treated with 1000 U/ml IFNγ (shaded red histograms) or IFNγ with panobinostat (HDACi; shaded blue histograms). H Bar graphs show relative MHC-I and MHC-II expression (IFNγ/control-treated) in SCC17-0263 and SCC11-0270 PD1 PROGs treated with 1000 U/ml IFNγ or IFNγ with panobinostat (HDACi). Individual values and mean ± sd of three biological replicates are shown and paired, two-tailed t-test was used to compare the data, exact p-values shown. Source data are provided as a Source Data file.
Fig. 7
Fig. 7. Loss of MHC antigen presentation molecules.
A Representative histograms showing expression of HLA-A02:01 in the WMD17-0012 PD1 PROG melanoma cells (HLA-A02:01 LOH, Melan-Anull). Untransfected cells are compared to cells transfected with HLA-A02:01 pre and post flow sorting (left panel). Reactivity of allogenic HLA-A02:01 matched CD8+ T cells (CD107+/IFNγ+) to untransfected or HLA-A02:01-transfected WMD17-0112 cells pulsed with 10 µg/ml Melan-A peptide (AAGIGILTV) or vehicle control, 1.5 h before co-culture (middle panel). Bar graph shows percentage (individual values and mean ± sd) of reactive CD107IFNγ+ CD8+ T cells co-cultured with untransfected/Melan-A pulsed (+) WMD-0112 cells (n = 6 biological replicates) and HLA-A02:01-transfected Melan-A pulsed (+) or HLA-A02:01 vehicle treated (−) WMD17-0112 cells (n = 4 biological experiments). Data compared using one-way ANOVA with Tukey’s multiple comparison test, adjusted p-values are shown. B Scatterplot showing correlation of MHC-II and MHC-I expression (expression score relative to TILs) in PD1 PROG tumor dissociates (n = 19). Correlation calculated using two-sided Spearman’s rank correlation coefficient, p-value is shown. Tumor samples with established alterations in JAK2 (SCC16-0016), MHC-I/II (WMD17-0112, SMU-059), B2M/CIITA (SMU-092, SCC13-0156) are highlighted and color-coded. Three tumors with low MHC-I/-II expression on melanoma without causal mechanisms are circled. Dotted lines indicate median MHC-I and MHC-II scores. C Percentage of CD45+ cells and frequency of CD8+, CD4+ and regulatory T cells (Treg), macrophages (Mø), and TCRαβ cells (as a percentage of CD45+ cells) in PD1 PROG tumor dissociates with high MHC-I and/or MHC-II score (above median; n = 6) vs low MHC-I and/or MHC-II score (below median; n = 13). Data compared using one-way ANOVA corrected for multiple comparisons by controlling the False Discovery Rate (q < 0.1); adjusted p-values are shown.

References

    1. Robert C, et al. Pembrolizumab versus ipilimumab in advanced melanoma (KEYNOTE-006): post-hoc 5-year results from an open-label, multicentre, randomised, controlled, phase 3 study. Lancet Oncol. 2019;20:1239–1251. doi: 10.1016/S1470-2045(19)30388-2. - DOI - PubMed
    1. Larkin J, et al. Five-year survival with combined nivolumab and ipilimumab in advanced melanoma. N. Engl. J. Med. 2019;381:1535–1546. doi: 10.1056/NEJMoa1910836. - DOI - PubMed
    1. Wolchok JD, et al. Long-term outcomes with nivolumab plus ipilimumab or nivolumab alone versus ipilimumab in patients with advanced melanoma. J. Clin. Oncol. 2022;40:127–137. doi: 10.1200/JCO.21.02229. - DOI - PMC - PubMed
    1. Carlino MS, Larkin J, Long GV. Immune checkpoint inhibitors in melanoma. Lancet. 2021;398:1002–1014. doi: 10.1016/S0140-6736(21)01206-X. - DOI - PubMed
    1. Ribas A, et al. Association of pembrolizumab with tumor response and survival among patients with advanced melanoma. J. Am. Med. Assoc. 2016;315:1600–1609. doi: 10.1001/jama.2016.4059. - DOI - PubMed

Publication types

Substances