Multimodel preclinical platform predicts clinical response of melanoma to immunotherapy

Abstract

Although immunotherapy has revolutionized cancer treatment, only a subset of patients demonstrate durable clinical benefit. Definitive predictive biomarkers and targets to overcome resistance remain unidentified, underscoring the urgency to develop reliable immunocompetent models for mechanistic assessment. Here we characterize a panel of syngeneic mouse models, representing a variety of molecular and phenotypic subtypes of human melanomas and exhibiting their diverse range of responses to immune checkpoint blockade (ICB). Comparative analysis of genomic, transcriptomic and tumor-infiltrating immune cell profiles demonstrated alignment with clinical observations and validated the correlation of T cell dysfunction and exclusion programs with resistance. Notably, genome-wide expression analysis uncovered a melanocytic plasticity signature predictive of patient outcome in response to ICB, suggesting that the multipotency and differentiation status of melanoma can determine ICB benefit. Our comparative preclinical platform recapitulates melanoma clinical behavior and can be employed to identify mechanisms and treatment strategies to improve patient care.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Fig. 1: Modeling diverse subtypes of human melanoma in mice.
Fig. 2: Melanoma mouse models recapitulate patient diversity in response to CTLA-4 blockade.
Fig. 3: Antigen presentation is functional in the four melanoma models.
Fig. 4: Intratumoral immune cell correlates of anti-CTLA-4 response in the melanoma models.
Fig. 5: Models resistant to anti-CTLA-4 exhibit T cell dysfunction or exclusion profiles.
Fig. 6: Transcriptomic profiling of the models identifies an MPS that predicts patient outcome in response to ICB.

Data availability

WES raw data of the melanomas and cell lines from the four models and RNA-seq raw data of the cell line-derived allografts from the four models was deposited on Gene Expression Omnibus accession code: GSE144946.

Code availability

The custom code used in this manuscript is available in Online Methods.

References

  1. 1.

    Wei, S. C., Duffy, C. R. & Allison, J. P. Fundamental mechanisms of immune checkpoint blockade therapy. Cancer Discov. 8, 1069–1086 (2018).

  2. 2.

    Zappasodi, R., Merghoub, T. & Wolchok, J. D. Emerging concepts for immune checkpoint blockade-based combination therapies. Cancer Cell 33, 581–598 (2018).

  3. 3.

    Snyder, A. et al. Genetic basis for clinical response to CTLA-4 blockade in melanoma. N. Engl. J. Med. 371, 2189–2199 (2014).

  4. 4.

    Van Allen, E. M. et al. Genomic correlates of response to CTLA-4 blockade in metastatic melanoma. Science 350, 207–211 (2015).

  5. 5.

    Nathanson, T. et al. Somatic mutations and neoepitope homology in melanomas treated with CTLA-4 blockade. Cancer Immunol. Res. 5, 84–91 (2017).

  6. 6.

    Rizvi, N. A. et al. Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science 348, 124–128 (2015).

  7. 7.

    Tumeh, P. C. et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature 515, 568–571 (2014).

  8. 8.

    Chen, P. L. et al. Analysis of immune signatures in longitudinal tumor samples yields insight into biomarkers of response and mechanisms of resistance to immune checkpoint blockade. Cancer Discov. 6, 827–837 (2016).

  9. 9.

    Riaz, N. et al. Tumor and microenvironment evolution during immunotherapy with nivolumab. Cell 171, 934–949 (2017).

  10. 10.

    Zaretsky, J. M. et al. Mutations associated with acquired resistance to PD-1 blockade in melanoma. N. Engl. J. Med. 375, 819–829 (2016).

  11. 11.

    Rodig, S. J. et al. MHC proteins confer differential sensitivity to CTLA-4 and PD-1 blockade in untreated metastatic melanoma. Sci. Transl. Med. 10, eaar3342 (2018).

  12. 12.

    Gao, J. et al. Loss of IFN-γ pathway genes in tumor cells as a mechanism of resistance to anti-CTLA-4 therapy. Cell. 167, 397–404 (2016).

  13. 13.

    Hugo, W. et al. Genomic and transcriptomic features of response to anti-PD-1 therapy in metastatic melanoma. Cell 165, 35–44 (2016).

  14. 14.

    Ayers, M. et al. IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade. J. Clin. Invest. 127, 2930–2940 (2017).

  15. 15.

    Auslander, N. et al. Robust prediction of response to immune checkpoint blockade therapy in metastatic melanoma. Nat. Med. 24, 1545–1549 (2018).

  16. 16.

    Jiang, P. et al. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat. Med. 24, 1550–1558 (2018).

  17. 17.

    Jerby-Arnon, L. et al. A cancer cell program promotes T cell exclusion and resistance to checkpoint blockade. Cell. 175, 984–997 (2018).

  18. 18.

    Dankort, D. et al. Braf(V600E) cooperates with Pten loss to induce metastatic melanoma. Nat. Genet. 41, 544–552 (2009).

  19. 19.

    Chin, L. et al. Cooperative effects of INK4a and Ras in melanoma susceptibility in vivo. Genes Dev. 11, 2822–2834 (1997).

  20. 20.

    Ackermann, J. et al. Metastasizing melanoma formation caused by expression of activated N-RasQ61K on an INK4a-deficient background. Cancer Res. 65, 4005–4011 (2005).

  21. 21.

    Dhomen, N. et al. Oncogenic Braf induces melanocyte senescence and melanoma in mice. Cancer Cell. 15, 294–303 (2009).

  22. 22.

    Damsky, W. E. et al. β-catenin signaling controls metastasis in Braf-activated Pten-deficient melanomas. Cancer Cell 20, 741–754 (2011).

  23. 23.

    Veierod, M. B., Adami, H. O., Lund, E., Armstrong, B. K. & Weiderpass, E. Sun and solarium exposure and melanoma risk: effects of age, pigmentary characteristics and nevi. Cancer Epidemiol. Biomarkers Prev. 19, 111–120 (2010).

  24. 24.

    Day, C. P., Marchalik, R., Merlino, G. & Michael, H. Mouse models of UV-induced melanoma: genetics, pathology, and clinical relevance. Lab. Invest. 97, 698–705 (2017).

  25. 25.

    Noonan, F. P. et al. Neonatal sunburn and melanoma in mice. Nature 413, 271–272 (2001).

  26. 26.

    De Fabo, E. C., Noonan, F. P., Fears, T. & Merlino, G. Ultraviolet B but not ultraviolet A radiation initiates melanoma. Cancer Res. 64, 6372–6376 (2004).

  27. 27.

    Cancer Genome Atlas. Genomic classification of cutaneous melanoma. Cell 161, 1681–1696 (2015).

  28. 28.

    Day, C. P., Merlino, G. & Van Dyke, T. Preclinical mouse cancer models: a maze of opportunities and challenges. Cell 163, 39–53 (2015).

  29. 29.

    Alexandrov, L. B. et al. Signatures of mutational processes in human cancer. Nature 500, 415–421 (2013).

  30. 30.

    Vidwans, S. J. et al. A melanoma molecular disease model. PloS ONE. 6, e18257 (2011).

  31. 31.

    Hayward, N. K. et al. Whole-genome landscapes of major melanoma subtypes. Nature 545, 175–180 (2017).

  32. 32.

    Tsoi, J. et al. Multi-stage differentiation defines melanoma subtypes with differential vulnerability to drug-induced iron-dependent oxidative stress. Cancer Cell. 33, 890–904 e895 (2018).

  33. 33.

    Chen, L. et al. Tumor immunogenicity determines the effect of B7 costimulation on T cell-mediated tumor immunity. J. Exp. Med. 179, 523–532 (1994).

  34. 34.

    Patel, S. J. et al. Identification of essential genes for cancer immunotherapy. Nature 548, 537–542 (2017).

  35. 35.

    Bloom, M. B. et al. Identification of tyrosinase-related protein 2 as a tumor rejection antigen for the B16 melanoma. J. Exp. Med. 185, 453–459 (1997).

  36. 36.

    Wolkers, M. C., Brouwenstijn, N., Bakker, A. H., Toebes, M. & Schumacher, T. N. Antigen bias in T cell cross-priming. Science 304, 1314–1317 (2004).

  37. 37.

    Godfrey, D. I., Le Nours, J., Andrews, D. M., Uldrich, A. P. & Rossjohn, J. Unconventional T cell targets for cancer immunotherapy. Immunity 48, 453–473 (2018).

  38. 38.

    Guo, X. et al. Global characterization of T cells in non-small cell lung cancer by single-cell sequencing. Nat. Med. 24, 978–985 (2018).

  39. 39.

    Philip, M. et al. Chromatin states define tumour-specific T cell dysfunction and reprogramming. Nature 545, 452–456 (2017).

  40. 40.

    De Palma, M. & Lewis, C. E. Macrophage regulation of tumor responses to anticancer therapies. Cancer Cell 23, 277–286 (2013).

  41. 41.

    Newman, A. M. et al. Robust enumeration of cell subsets from tissue expression profiles. Nat. Methods 12, 453–457 (2015).

  42. 42.

    Chen, Z. et al. Inference of immune cell composition on the expression profiles of mouse tissue. Sci. Rep. 7, 40508 (2017).

  43. 43.

    Bottcher, J. P. et al. NK cells stimulate recruitment of cDC1 into the tumor microenvironment promoting cancer immune control. Cell. 172, 1022–1037 (2018).

  44. 44.

    Bottcher, J. P. & Reis, E. S. C. The role of type 1 conventional dendritic cells in cancer immunity. Trends Cancer 4, 784–792 (2018).

  45. 45.

    Collin, M. & Bigley, V. Human dendritic cell subsets: an update. Immunology 154, 3–20 (2018).

  46. 46.

    Barry, K. C. et al. A natural killer-dendritic cell axis defines checkpoint therapy-responsive tumor microenvironments. Nat. Med. 24, 1178–1191 (2018).

  47. 47.

    Binnewies, M. et al. Unleashing type-2 dendritic cells to drive protective antitumor CD4(+) T cell immunity. Cell 177, 556–571 (2019).

  48. 48.

    Vandamme, N. & Berx, G. From neural crest cells to melanocytes: cellular plasticity during development and beyond. Cell Mol. Life Sci. 76, 1919–1934 (2019).

  49. 49.

    Marie, K. L. et al. Melanoblast transcriptome analysis reveals pathways promoting melanoma metastasis. Nat. Commun. 11, 333 (2020).

  50. 50.

    Joshi, S. S. et al. CD34 defines melanocyte stem cell subpopulations with distinct regenerative properties. PLoS Genet. 15, e1008034 (2019).

  51. 51.

    Chan, T. A. et al. Development of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic. Ann. Oncol. 30, 44–56 (2019).

  52. 52.

    Landsberg, J. et al. Melanomas resist T cell therapy through inflammation-induced reversible dedifferentiation. Nature 490, 412–416 (2012).

  53. 53.

    Mehta, A. et al. Immunotherapy resistance by inflammation-induced dedifferentiation. Cancer Discov. 8, 935–943 (2018).

  54. 54.

    Aguirre, A. J. et al. Activated Kras and Ink4a/Arf deficiency cooperate to produce metastatic pancreatic ductal adenocarcinoma. Genes Dev. 17, 3112–3126 (2003).

  55. 55.

    Bald, T. et al. Ultraviolet-radiation-induced inflammation promotes angiotropism and metastasis in melanoma. Nature 507, 109–113 (2014).

  56. 56.

    Bald, T. et al. Immune cell-poor melanomas benefit from PD-1 blockade after targeted type I IFN activation. Cancer Discov. 4, 674–687 (2014).

  57. 57.

    Vasioukhin, V., Degenstein, L., Wise, B. & Fuchs, E. The magical touch: genome targeting in epidermal stem cells induced by tamoxifen application to mouse skin. Proc. Natl Acad. Sci. USA 96, 8551–8556 (1999).

  58. 58.

    Virador, V. et al. Production of melanocyte-specific antibodies to human melanosomal proteins: expression patterns in normal human skin and in cutaneous pigmented lesions. Pigment Cell Res. 14, 289–297 (2001).

  59. 59.

    Iida, N. et al. Commensal bacteria control cancer response to therapy by modulating the tumor microenvironment. Science 342, 967–970 (2013).

  60. 60.

    Araya, R. E. & Goldszmid, R. S. Characterization of the tumor immune infiltrate by multiparametric flow cytometry and unbiased high-dimensional data analysis. Methods Enzymol. 632, 309–337 (2020).

  61. 61.

    Chen, H. et al. Cytofkit: A bioconductor package for an integrated mass cytometry data analysis pipeline. PLoS Comput. Biol. 12, e1005112 (2016).

  62. 62.

    Schrock, E. et al. Multicolor spectral karyotyping of human chromosomes. Science 273, 494–497 (1996).

  63. 63.

    Nesbitt, M. N. & Francke, U. A system of nomenclature for band patterns of mouse chromosomes. Chromosoma 41, 145–158 (1973).

  64. 64.

    Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).

  65. 65.

    Li, B. & Dewey, C. N. RSEM: accurate transcript quantification from RNA-seq data with or without a reference genome. BMC Bioinformatics 12, 323 (2011).

  66. 66.

    Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).

  67. 67.

    Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005).

  68. 68.

    Azizi, E. et al. Single-cell map of diverse immune phenotypes in the breast tumor microenvironment. Cell 174, 1293–1308 (2018).

  69. 69.

    Brown, C. C. et al. Transcriptional basis of mouse and human dendritic cell heterogeneity. Cell. 179, 846–863 (2019).

Download references

Acknowledgements

We thank T.A. Chan (Memorial Sloan-Kettering Cancer Center) for suggestions and comments; T. Tüting (University of Magdeburg), M. McMahon (University of Utah) and M. Bosenberg (Yale School of Medicine) for mouse reagents; E. Van Allen (Dana-Farber Cancer Institute) for sharing clinical data; S. Burkett (Molecular Cytogenetics Core Facility, MCGP, NCI-Frederick) for SKY analysis; C. Redon (NCI) for assistance with gamma irradiation; Y. Boumber (Fox Chase Cancer Center) and C. Alewine (NCI) for manuscript revision and editing. This research was supported in part by funds from the NIH intramural research program and a FLEX Synergy Award from the NCI Center for Cancer Research. An NCI Director’s Innovation Award to E.P.-G. helped support this project. R.S.L. received founding from the Ressler Family Foundation and from Merck and Bristol-Myers Squibb. W.H. received a Daneen & Charles Stiefel Investigative Scientist Award from American Skin Association and a Young Investigator Award from Melanoma Research Alliance.

Author information

Affiliations

Authors

Contributions

The contributions of the authors as as follows: conceptualization, E.P.-G., C.-P.D. and G.M.; methodology, R.E.M., H.T.M., S.K.V. and C.-P.D.; software, H.H.Y. and M.P.L.; validation, E.P.-G., H.H.Y., M.P.L. and C.-P.D.; formal analysis, E.P.-G., H.H.Y., R.E.M., H.T.M., S.K.V., K.L.M., A.M.M., M.P.L. and C.-P.D.; investigation, E.P.-G., R.E.A., R.E.M., S.K.V., K.L.M., C.S., S.C., K.C.L., A.T., A.F., A.J.I., A.K., W.H., R.S.L. and C.-P.D.; data curation, H.H.Y., A.M.M. and M.P.L.; writing (original draft), E.P.-G., C.-P.D. and G.M.; writing (review and editing), E.P.-G., H.H.Y., R.E.M., H.T.M., S.K.V., N.P.R., T.V.D., S.K.S., R.S.G., Z.W.O., C.-P.D. and G.M.; visualization, E.P.-G., H.H.Y., R.E.A., M.P.L. and C.-P.D.; supervision, S.K.S., T.V.D., R.S.G., Z.W.O., M.P.L., C.-P.D. and G.M.; project administration, E.P.-G., R.E.M. and C.-P.D.; and funding acquisition, G.M.

Corresponding authors

Correspondence to Maxwell P. Lee or Chi-Ping Day or Glenn Merlino.

Ethics declarations

Competing interests

R.S.L. received funding from Merck and Bristol-Myers Squibb. The rest of the authors declare no competing interests.

Additional information

Peer review information Javier Carmona was the primary editor on this article and managed its editorial process and peer review in collaboration with the rest of the editorial team.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Study design and sample processing.

Melanomas were induced in four genetically engineered mice (GEM) harboring the indicated genetic modifications by ultraviolet (UV) radiation or 7,12-Dimethylbenz[a]anthracene (DMBA) topical administration at postnatal day 3 and activation of CreERT2 at day 7. Fragments from each melanoma were expanded in C57BL/6 syngeneic mice and viably archived at low passage to generate a GEM-derived allograft (GDA) biobank. A cell line (CL) from each model was isolated. GDA and cell line-derived allografts (CLDA) were implanted in C57BL/6 mice and treated with CTLA-4 blocking antibody (αCTLA-4) or isotype control. GDAs, CLDAs and CLs from each model were processed in triplicates for whole exome sequencing (WES), RNA sequencing and/or histopathology.

Extended Data Fig. 2 Generation of four mouse models representative of diverse human melanomas.

a, Representative images showing chromosomal duplications and translocations analyzed by spectral karyotyping (SKY) of the four melanoma cell lines. N = 12 cells for M1, N = 15 cell for M2, N = 15 for M3 and N = 10 cells for M4 from one experiment. See also Supplementary Table 2. b, Distribution of COSMIC mutation signatures29 found in the four models (GDAs and cell lines). c, GDAs and cell lines derived from the four mouse models were clustered accordingly to their mutation profiles with TCGA patient samples from different mutation categories (i.e., BRAF, NRAS or NF1 mutants, or triple-wildtype27,31). d, Immunoblot showing MAPK and PI3K activation and expression of the indicated proteins in the four melanoma cell lines. Representative cropped images of three independent experiments are shown (see full scans in Source Data Extended Data Fig. 2). Source data

Extended Data Fig. 3 Tumor immunogenicity correlates with αCTLA-4 response in the melanoma models.

a, Schematic of the in vivo study design. 1.0×106 γ-irradiated melanoma cells from each model were injected subcutaneously into C57BL/6 mice as vaccinated groups. After 4 weeks, the vaccinated and non-vaccinated control groups were challenged with the same number of viable melanoma cells from paired models. b-e, Tumor growth curves (left panels) and tumor onset (right panels) of vaccinated (magenta) and control (light blue) groups from M1 (b, N = 9 mice in control and N = 8 mice in vaccinated group), M2 (c, N = 10 mice), M3 (d, N = 5 mice) and M4 (e, N = 5 mice). Asterisk highlights the mice that did not develop tumors (d, 2/5 from vaccinated group). The time of tumor onset was considered as the first day after implantation when tumors were measurable. Data is depicted as the mean and error bars represent S.E.M. Mann-Whitney test two-tailed P-values are indicated (b-e, right panels). Source data

Extended Data Fig. 4 Increased tumor-infiltrating lymphocytes are associated with αCTLA-4 response in the melanoma models.

a, Representative images of CD3 immunostaining (red staining) of the four melanoma models (untreated). N = 6 melanomas (one section per tumor) from two independent experiments. Arrows point to the area magnified and bars represent 200μm. b, Automated quantification of CD3 positive area obtained from (a). Whole-tumor sections from melanomas of each model were quantified using Aperio software from one experiment. N = 6 melanomas (one section per tumor) from two independent experiments. c, Tumor growth curves of M4 melanomas treated with αCTLA-4 (green and red lines) or isotype antibody (blue dashed line). Red arrow indicates the time of the first dose of the treatment. Tumors were considered non-responders (NR, green) when their size was >150mm3 and increasing for more than 2 consecutive measurements and responders (R, red) when their size was <120mm3 and decreasing for more than 2 consecutive measurements. d, Percentage of CD3 positive area in whole-tumor sections from M4 treated with αCTLA (NR, green bar and R, red bar) or isotype antibody (blue bar). Data are depicted as the mean and error bars represent S.E.M (N = 5 isotype control treated tumors, N = 6 NR tumors and N = 9 R tumors). Kruskal-Wallis test P-values adjusted by Dunn’s test for multiple comparisons (b) and Mann-Whitney test two-tailed P-values (d) are indicated. Source data

Extended Data Fig. 5 High parametric flow cytometry analysis of the intratumoral immune cell populations from the four melanoma models.

a, Total number of leukocytes (CD45+ cells) infiltrating the untreated cell line-derived allografts of the four melanoma models. b-f, Percentage of intratumoral eosinophils (CD11b+Ly6GLy6CintSiglecF+F4/80+CD64CD24+) (b), neutrophils (CD11b+Ly6G+Ly6CintSiglecFCD24+) (c), macrophages (CD11b+CD68hiCD64+F480+Ly6CLy6GSiglecFCD24) (d), dendritic cells (Ly6GSiglecFCD11c+MHC-II+CD64CD24int/hiCD135+) (e) and Treg (CD3+TCRβ+CD4+CD25+FoxP3+) (f). Data from a representative of two experiments is depicted as the mean (N = 5 untreated tumors per model) and error bars represent S.E.M. Kruskal-Wallis test P-values adjusted by Dunn’s test for multiple comparisons are indicated (a-f). See also Supplementary Table 6 for the population definition. Source data

Extended Data Fig. 6 High expression of T-cell exhaustion markers in αCTLA-4-resistant M1 melanomas.

a, Expression of the indicated markers per intratumoral CD3+ T-cell from the four models analyzed by t-stochastic neighbor embedding (t-SNE). The t-SNE plots show a representative of two independent experiments (N = 5 tumors per model). The samples from each model were color coded as indicated (first left upper plot). b-d, Expression of PD-1 (b), TIM3 (c) and LAG3 (d) in CD8+ T-cells as the mean fluorescence intensity (MFI) from flow cytometry. e-f, Expression of PD-L1 in macrophages (e) and dendritic cells (DCs) (f) as MFI from flow cytometry. Data from a representative of two experiments is depicted as the mean (N = 5 untreated tumors per model) and error bars represent S.E.M. Kruskal-Wallis test P-values adjusted by Dunn’s test for multiple comparisons are indicated (b-f). See also Supplementary Table 6 for the population definition. g, Expression of T-cell exhaustion and dysfunction markers38,39 in untreated tumors from the four models. Data is shown as FPKM from RNA sequencing analysis (N = 4 tumors). Source data

Extended Data Fig. 7 Specific myeloid and lymphoid populations correlate with the response to αCTLA-4.

a, Percentage of CD206+ macrophages infiltrating untreated cell line-derived allografts from the four models. b-c, Expression of CD206 (b) and MHC-II (c) in the intratumoral macrophages measured by flow cytometry (MFI: mean fluorescence intensity). Data from a representative of two experiments is depicted as the mean (N = 5 tumors per model) and error bars represent S.E.M. (a-c). Kruskal-Wallis test P-values adjusted by Dunn’s test for multiple comparisons are indicated (a-c). See also Supplementary Table 6 for the population definition. d-e, Fraction of pro-tumor macrophages obtained by CIBERSORT41,42 analysis of the transcriptomes of the four models (d) and metastatic melanoma patients treated with Ipilimumab (αCTLA-4, Van Allen data set4) (e). Bar center represent the mean (N = 4 untreated tumors) and error bars represent S.E.M. (d). Boxes represent the median, upper and lower quartiles and the whiskers the minimum to maximum range (N = 9 non-responder and N = 7 responder patients). Unpaired t-test two-tailed P-value is indicated (e). f, Percentage of intratumoral NK cells (CD3NK1.1+) in untreated melanomas from the four models by flow cytometry (N = 5 tumors). g, Ccl5 and Xcl1 expression from RNA sequencing analysis of the four models (N = 4 untreated tumors). Data is depicted as the mean and error bars represent S.E.M (f,g). h, CCL5, XCL1 and XCL2 expression in ipilimumab-treated melanoma patients (αCTLA-4, Van Allen data set) with High (N = 16 patients, orange) and Low (N = 17, blue) NK and cDC1 signature levels. Boxes represent the median, upper and lower quartiles and the whiskers the minimum to maximum range. i, Expression of CD44 and PD-1 as the mean fluorescence intensity (MFI) from flow cytometry in the conventional CD4+ T-cells (CD4+ Tconv) infiltrating the four models (N = 5 untreated tumors). P-values from Kruskal-Wallis test adjusted by Dunn’s test for multiple comparisons (f,g) and two-tailed P-values from Mann-Whitney test (h) are indicated. Source data

Extended Data Fig. 8 Association of the predictive signature with melanocytic lineage differentiation.

a, Top 10 IPA Disease and Bio function categories enriched in the 45 genes of the signature. X axis represent Fisher’s exact test P-values. The number of MPS genes in each category (N) are indicated. b, c, Expression of the Melanocytic Plasticity Signature (MPS) genes and MPS scores of mouse melanoblasts (days E15.5 and E17.5) vs. melanocytes (P1 and P7)49 (N = 1 sample per stage) (b) and multipotent (CD34+) vs. melanocytic committed (CD34) melanocyte stem cells from the hair follicles of P56 mice50 (N = 3 samples per population) (c). Data is represented as the z-scores from RNA sequencing (b, c).

Extended Data Fig. 9 The Melanocytic Plasticity Signature (MPS) predicts patient outcome in response to αPD-1.

a, Melanocytic Plasticity Signature (MPS) in ipilimumab-naïve melanoma patients treated with αPD-1 (Hugo and Riaz data sets9,13). The box-plot shows the MPS scores of baseline samples from non-responder (NR, green, N = 25) and responder (R, violet, N = 25) patients. Boxes represent the median, upper and lower quartiles and the whiskers the minimum to maximum range. Mann-Whitney test two-tailed P-values are indicated. b, Kaplan-Meier curves of the overall survival of the patients in (a) accordingly to their MPS scores. c, Area under the ROC curve (AUC) values comparing the prediction performance of Tumor mutation burden (TMB), PD-L1 expression, TIDE16 and MPS scores in Van Allen4 (left, N = 42 patients) and Hugo-Riaz9,13 (right, N = 50 patients) data sets. d, Kaplan-Meier curves of the overall survival of the patients in (a) accordingly to their TIDE scores. N for each group and two-tailed P-values from the Log-rank (Mantel-Cox) test are indicated (b, d). Source data

Extended Data Fig. 10 The response of the four melanoma models to αPD-1 is similar to αCTLA-4.

a, Schematic of the in vivo study design. 1.0 ×106 melanoma cells from each model were implanted subcutaneously into C57BL/6 syngeneic mice. When the tumors reached 50mm3, mice were randomized and treated with αPD-1 or isotype antibody as control every 3 days for a total of 3 doses. b, Tumor growth of the melanomas from the indicated models upon treatment with αPD-1 (red lines) or isotype control (blue lines). N = 10 per group. c, Kaplan-Meier survival curves from (b). A representative from two experiments is shown. Two-tailed P-value from Gehan-Breslow-Wilcoxon test is indicated (c). Source data

Supplementary information

Supplementary Information

Supplementary Figs. 1 and 2 and Supplementary Tables 2, 3 and 6–8.

Reporting Summary

Supplementary Table 1

Supplementary Table 1

Supplementary Table 4

Supplementary Table 4

Supplementary Table 5

Supplementary Table 5

Source data

Source Data Fig. 1

Statistical source data.

Source Data Fig. 2

Statistical source data.

Source Data Fig. 4

Statistical source data.

Source Data Fig. 5

Statistical source data.

Source Data Fig. 6

Statistical source data.

Source Data Extended Data Fig. 2

Unprocessed western blots.

Source Data Extended Data Fig. 3

Statistical source data.

Source Data Extended Data Fig. 4

Statistical source data.

Source Data Extended Data Fig. 5

Statistical source data.

Source Data Extended Data Fig. 6

Statistical source data.

Source Data Extended Data Fig. 7

Statistical source data.

Source Data Extended Data Fig. 9

Statistical source data.

Source Data Extended Data Fig. 10

Statistical source data.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Pérez-Guijarro, E., Yang, H.H., Araya, R.E. et al. Multimodel preclinical platform predicts clinical response of melanoma to immunotherapy. Nat Med 26, 781–791 (2020). https://doi.org/10.1038/s41591-020-0818-3

Download citation