The Spatial Landscape of Progression and Immunoediting in Primary Melanoma at Single-Cell Resolution - PubMed (original) (raw)
. 2022 Jun 2;12(6):1518-1541.
doi: 10.1158/2159-8290.CD-21-1357.
Ajit J Nirmal # 1 2 3, Tuulia Vallius # 1 2, Brian Quattrochi 4, Alyce A Chen 1 2, Connor A Jacobson 1 2, Roxanne J Pelletier 1 2, Clarence Yapp 1 2, Raquel Arias-Camison 1 2 4, Yu-An Chen 1 2, Christine G Lian 4, George F Murphy 4, Sandro Santagata # 1 2 4, Peter K Sorger # 1 2 5
Affiliations
- PMID: 35404441
- PMCID: PMC9167783
- DOI: 10.1158/2159-8290.CD-21-1357
The Spatial Landscape of Progression and Immunoediting in Primary Melanoma at Single-Cell Resolution
Ajit J Nirmal et al. Cancer Discov. 2022.
Abstract
Cutaneous melanoma is a highly immunogenic malignancy that is surgically curable at early stages but life-threatening when metastatic. Here we integrate high-plex imaging, 3D high-resolution microscopy, and spatially resolved microregion transcriptomics to study immune evasion and immunoediting in primary melanoma. We find that recurrent cellular neighborhoods involving tumor, immune, and stromal cells change significantly along a progression axis involving precursor states, melanoma in situ, and invasive tumor. Hallmarks of immunosuppression are already detectable in precursor regions. When tumors become locally invasive, a consolidated and spatially restricted suppressive environment forms along the tumor-stromal boundary. This environment is established by cytokine gradients that promote expression of MHC-II and IDO1, and by PD1-PDL1-mediated cell contacts involving macrophages, dendritic cells, and T cells. A few millimeters away, cytotoxic T cells synapse with melanoma cells in fields of tumor regression. Thus, invasion and immunoediting can coexist within a few millimeters of each other in a single specimen.
Significance: The reorganization of the tumor ecosystem in primary melanoma is an excellent setting in which to study immunoediting and immune evasion. Guided by classic histopathology, spatial profiling of proteins and mRNA reveals recurrent morphologic and molecular features of tumor evolution that involve localized paracrine cytokine signaling and direct cell-cell contact. This article is highlighted in the In This Issue feature, p. 1397.
©2022 The Authors; Published by the American Association for Cancer Research.
Figures
Figure 1.
Multimodal profiling of cutaneous melanoma. A, Conceptual framework of sample processing for cyclic immunofluorescence (CyCIF), high-resolution CyCIF, and microregion transcriptomics [GeoMx and PickSeq (mrSEQ)]. Abbreviations for annotated histologies are shown below with color-coding used in subsequent figure panels. B, A 30-plex CyCIF image of a section of specimen MEL1-1 showing selected markers for epidermis (pan-CK: cyan) and tumor cells (SOX10: red), highlighting annotated histologies and microregions (mROIs) that were subjected to mrSEQ (white “+” symbols). This specimen was likely torn during slide processing, and, thus, spatial arrangements in the region marked with a blue dashed boundary are not considered reliable. Other mrSEQ sites are shown in Supplementary Fig. S2A. C, CyCIF image of MEL1-1 corresponding to the MIS and adjacent regions of inflammatory (IR) and terminal regression (TR), respectively (outlined by dashed white lines). Rectangles depict the positions of 110 × 110 μm ROIs in which high-resolution 3D deconvolution microscopy was performed. The region highlighted with orange is magnified in G. D, Uniform manifold approximation and projection (UMAP) of single-cell data derived from CyCIF of patient MEL1, labeled by cell type (top) and the signal intensities of individual markers (bottom). Markers used for cell-type calls are shown in Supplementary Fig. S1C. The UMAP plot was built using 50,000 single cells that were randomly sampled from the full data set (n = 1.1 × 106). APC, antigen-presenting cell; DC, dendritic cell; Macs, macrophages; Treg, regulatory T cell. E, Cell-type assignments (with data points representing the centroids of cells) mapped to their physical locations in a portion of the bTIL region lying just beyond the IM in MEL1-1. F, H&E image of the same region as in E. Regions of tumor and stroma are separated by dashed black lines. G, A 21-plex, high-resolution CyCIF image of a MEL1-1 MIS region (orange square in C) with selected markers shown as a maximum-intensity projection staining for DNA (blue), tumor (SOX10: white), and T cells (CD4: green, CD8: red). The dermal–epidermal junction is denoted with a white dashed line, and all FOXP3+ cells (as determined from other image channels; see Supplementary Fig. S1F) are denoted with an asterisk. Note that all images in G–J are derived from a single multiplex CyCIF 3D image stack. H, Magnified regions from G (outlined with a yellow box) showing staining of DNA (blue), CD4 (green), CD8 (red), and TIM3 (white). Four cell types are labeled, including a Treg (green box, shown in J) and two CD8+ CTLs interacting with a tumor cell (shown in I). The dashed line follows the axis of immune synapse polarization and gives rise to the intensity plot in I. The orange box depicts the locations of representative images in I. I, Single optical section images of the immune synapse in H showing staining of tumor (SOX10: white), DNA (blue), and cell membrane (HLA-A: magenta) along with a series of single-channel images of functional T-cell markers. Right, quantified spatial distribution of CD8 and CD3 along the dashed line in H. J, Inset from H (outlined with a green square). Single optical section images of a tumor cell interacting with a Treg. Top: staining for tumor (SOX10: white), cell membrane (HLA-A: magenta), and DNA (blue); bottom: staining for Treg (ICOS: cyan). The two z-sections shown are spaced 2.2 μm apart.
Figure 2.
Recurrent cellular neighborhoods associated with melanoma progression. A, Uniform manifold approximation and projection (UMAP) of single-cell data from 70 ROIs in 12 patients. The plot was generated using 50,000 single cells that were randomly sampled from the full data set of 1.5 × 106 cells. The UMAP is colored based on the phenotype (left), disease progression stage (center), and patient ID (right). B, UMAPs (shown also in A) representing feature plots of expression of selected protein markers. C, The percentage of SOX10+ melanocytes or tumor cells expressing S100A within each stage of progression. **, P < 0.01; ***, P < 0.001. D, Heat map showing the abundance of cell types within the 30 LDA-based cellular neighborhood clusters (numbers to the right of the plot); these were then reduced to the 10 meta-clusters (RCNs) shown to the left of the plot. The bar chart to the right of the heat map depicts the distribution of progression stages within each cluster, and the bar chart to the left of the heat map represents the distribution of patients within each cluster. E, Bar plot depicting the detailed breakdown of cell-type proportions within each RCN (RCN1–10; _x_-axis). Pie charts depicting a simplified breakdown of cell types in each RCN; myeloid (green; dendritic cells, CD11C+ macrophages, macrophages, and Langerhans cells), lymphoid (light orange; CTL, Treg, and T helper), immune-suppressive (dark orange; PDL1+ DCs, PDL1+ Macs, and PD1+ CTL), melanocytes (dark blue), and keratinocytes (yellow). APC, antigen-presenting cell; DC, dendritic cell; LC, Langerhans cell; Macs, macrophages.
Figure 3.
A, Scatter plot (top) showing an FOV of the IM region (specimen MEL1-1). The cells are colored based on recurrent cellular neighborhoods (RCN1–10) to which they belong. The red and blue boxes represent regions that are magnified in the bottom panel (left and right, respectively) depicted as Voronoi diagrams. B, Exemplary CyCIF images highlighting RCNs in the invasive front of specimen MEL1-1. Top, an overall view of the invasive front stained for tumor cells (S100B: blue), macrophages (CD163: cyan), T cells (CD3: red), and dendritic cells (CD11C: green). The inset squares correspond to magnified panels at the bottom. H&E staining of a serial section of the same region is represented in the top right corner. Bottom left (red), RCN9 enriched for dendritic cells (CD11C: green) at the tumor–stroma junction; bottom center (blue), RCN5/8 enriched with PD1+ CTLs (CD8: green; PD1: red); bottom right (yellow), RCN3/4 enriched with myeloid cells (CD163: magenta; CD11C: green). The dashed gray line represents the tumor–stroma boundary. C, Voronoi diagrams of representative FOVs compiled from regions of N, P, and MIS. Each cell is colored based on the recurrent cellular neighborhood (RCN1–10) to which it belongs (as in A). Examples of corresponding CyCIF images from one patient in each case are provided at the bottom. A magnified view is available in Supplementary Fig. S3A. D, Bar plot depicting the proportional distribution of RCNs (RCN1–10) among the disease progression stages (N, P, MIS, IM, and EM). E, Box plots of the distribution of the shortest distance between cells in RCN2 to RCN7 and RCN10 grouped based on progression stages. The t test (*, P < 0.05) was used to depict significant changes in mean distances between the compared stages. The comparison made is described in the upper right corner of each plot (e.g., N vs. P). F, Shift plot shows the distance between melanocytes and CTLs, PDL1+ myeloid cells, and Tregs in normal (top) and precursor (bottom) regions. Significance is calculated for each percentile (10, 20, 30, 40, 50, 60, 70, 80, and 90) using the robust Harrell–Davis quantile estimator. Red indicates a significant difference (P < 0.05) and gray represents nonsignificance for each percentile.
Figure 4.
A, Field of MIS from a whole-slide CyCIF image of MEL1-1. A PDL1+ melanocyte (SOX10: white, PDL1: green) and CTLs (CD8: red) are highlighted with an orange box (left). Right, the polarization of PD1 (red) and PDL1 (green) to the point of contact between the interacting cells. B, Line plot showing the percentage of ROIs that displayed significant (P < 0.05) co-occurrence based on a proximity analysis performed between PDL1+ CD11C+ CD163− dendritic cells and PD1+ CTLs. **C,** Field of IM from a whole-slide CyCIF image of MEL1-1 stained for tumor (SOX10: red), macrophages (CD163: green), and CTLs (CD8: white), with three fields of macrophage–CTL contacts (yellow boxes). Maximum-intensity projections imaged at high-resolution in fields 1 and 2 are stained for DNA (blue), PDL1 (red), and PD1 (green), with cells labeled as myeloid cells (M) and engaged T cells (T); field 3 shows tumor cells (SOX10: red), CTLs (CD8: white), and a macrophage (CD163; green). Inset white boxes in the bottom right show concentration of PD1 (green) and PDL1 (red) to the point of contact, and the long connection between a macrophage (CD163: white) and a CTL is shown in a 3D reconstruction of field 3. **D,** Left shows the same CyCIF FOV as in **C**, stained for DNA (blue), TIM3 (red), and CD8A (green). The white inset box illustrates the staining of one CD163+ CD11C+ TIM3+ myeloid cell next to a CTL (right). **E,** Maximum-intensity projection from the bTIL region (top left) stained for DNA (blue), macrophages (CD163: green), and T cells (CD3D: white). The white inset is magnified and stained for T-cell polarity (CD4: green, CD8A: red), the PD1–PDL1 axis (PD1: green, PDL1: red), and exhaustion markers (TIM3: red, LAG3: green). A Treg in this field is indicated with the label “Tr.” **F,** PDL1 positivity in SOX10+ tumor cells (top) and CD11C+ myeloid cells (bottom). The proportions of PDL1+ tumor cells to all tumor cells (0%–5%, 5%–20%, and >20%) and PDL1+ myeloid cells to all myeloid cells (<1%, 1%–25%, and >25%) are presented in both primary melanoma cohorts (cohort 1: MEL1–13 and cohort 2: 25 primary melanomas). G, Fields of a primary melanoma and a melanoma metastasis from CyCIF images stained for DNA (blue), SOX10 (green), PDL1 (red), and CD11C (white). Top, an example of PDL1+ SOX10+ tumor cells at the deepest invasive region. PDL1+ metastasis is shown in the bottom panel. The tumor–stroma interface is indicated with a white dashed line.
Figure 5.
Single-cell analysis of invasive tumor. A, CyCIF images of MEL1-1 stained for S100A (top), MITF (middle), and S100B (bottom). Boxes represent regions highlighted in B. B, Insets from A of the tumor region (IM) showing gradient expression patterns for MITF (top) and S100B (bottom). Contours describe averaged quantified marker expression. C, Heat map showing median expression of protein markers identified within TCC1–10 TCCs. The bar plot on top of the heat map shows the proportional estimate of the TCCs within histologic annotations (EM, IM, or IB). The heat map at the bottom shows the properties related to the shape of the cells (area, solidity, extent, and eccentricity) derived from the segmentation masks. D, Scatter plot mapping the physical location of the derived tumor cell clusters (TCC1–10: dark blue) in MEL1-1. Each subplot represents the location of cells within a TCC and other cells in gray. E, Scatter plot (left) showing an FOV of the IM region. Cells are colored based on their TCC (TCC1–10). The yellow circle highlights the region in B. Right, a CyCIF image of the same FOV (from specimen MEL1-1) stained for CD163 (green), MITF (yellow), KI67 (red), and MHC-II (HLADPB1: blue). Voronoi diagram (right) generated from an FOV at the apex of the invasive front (inset highlighted in yellow). Cells are colored based on the TCC (TCC1–10) to which they belong. F, Bar plots showing the percentage of S100B-, S100A-, MITF-, KI67-, and MHC-II (HLADPB1)–positive cells within each TCC (TCC1–10).
Figure 6.
Microregional transcript profiling. A, PCA plot of melanoma mrSEQ transcriptomes (GeoMx). Colors indicate regional histopathology: bTIL (pink), IR (brown), MIS (green), invasive front (IB; light green), EM (gray), and center of invasive melanoma tumor (IM; yellow). EM and IM are enriched for tumor cells in this analysis, and IB contains mostly tumor cells with marginal immune infiltration. B, Expression of selected melanoma-related marker genes in mrSEQ data (PickSeq) split into three broad groups based on the PCA of GeoMx data (A). Data are mean ± SEM. ***, P < 0.001; ns, not significant. C, ssGSEA on mrSEQ data (PickSeq). ssGSEA scores highlight enrichment of melanoma-related gene signatures in tumor mROIs (primarily IB, IM, and EM) and immune-related signatures in the immune-rich mROIs (IR, bTIL). D, Fold difference (log2) and significance (log10_P_adj) for expression of 19,500 genes between EM (n = 34) and IM (n = 16) mROIs (PickSeq). Differentially expressed genes above (brown) and below (blue/gray) a significance threshold (_P_adj = 0.05) and above a fold change threshold (log2 fold change = 10) are indicated. E, GSEA for upregulation of the KRAS pathway in IM (n = 16) compared with EM (n = 34) mROIs (PickSeq). FDR < 0.05. F, Expression (log2) of MYC, NFKB1, IGFBP2, IGF1R, and BCL2A1 in IM and EM mROIs (PickSeq). Data, mean ± SEM; *, P < 0.05; **, P < 0.01; ***, P < 0.001. G, Heat map showing expression of genes (listed on the _y_-axis) known to play a role in EMT transition (PickSeq). All genes showed a significant difference between their mean expression in IM versus EM mROIs (P < 0.05). H, CyCIF image showing an FOV in MIS (top) and EM (bottom) regions. The tissue is stained for melanocytes (SOX10: yellow), endothelial cells (CD31: green), keratinocytes (pan-CK: white), and tumor cells (S100B: magenta). Arrows mark examples of melanocytes and tumor cells. I, Correlation network subgraph of genes associated with S100B expression in mrSEQ data (PickSeq). Nodes represent genes, and the edges correspond to the correlation between them. Brown nodes represent the genes that belong to the S100B module. Selected genes are annotated. J, Mean expression of 35 genes identified within the S100B module in mrSEQ data (PickSeq). The _x_-axis represents the mROIs grouped into the histopathologic annotation category from which they were isolated. K, Density plots illustrating the log-scaled protein expression of PMEL and CD63 in MIS and tumor (EM and IM) regions imaged with CyCIF. ***, P < 0.001.
Figure 7.
A, CyCIF image of specimen MEL1-1 showing a protruding edge of the invasive tumor (SOX10: violet, S100B: pink) into the dermis; outside the tumor boundary marked by a white line (dashed) is the brisk TIL region, which contains activated/exhausted T cells (CD3: green, PD1: red) and myeloid cells (CD11C: blue). B, Expression of CXCL10, CXCL11, IDO1, MIF, and CD74 among histologic sites (PickSeq data). Values represent mean ± SEM; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. C, CyCIF FOV of MEL1-1 highlighting the spatial arrangement of MHC-II+ tumor cells at the invasive front. Tumor cells were stained with SOX10 (cyan), MHC-I (HLA-A: green), and MHC-II (HLADPB1: red). Magnified regions outlined in magenta and yellow squares illustrate MHC-II+ and MHC-II− staining of tumor cell membranes. D, CyCIF of specimen MEL1-1. Left, zoomed-out view of invasive front stained for melanocytes (SOX10: blue), myeloid cells (CD11C: red), and interferon signaling (IRF1: green). Top right, zoomed-in view of invasive front apex stained for melanocytes (SOX10: blue), myeloid cells (CD11C: red), and interferon signaling (IRF5: yellow). Bottom right, zoomed-in view of invasive front apex stained for melanocytes (MART1: green), myeloid cells (CD11C: blue), and interferon signaling (IRF1: red). E, Line plot showing scaled fluorescence intensity of SOX10 (blue) and IRF1 (pink) within (tumor; left of the dashed blue line) and outside (stroma; right of the dashed blue line) the invasive tumor front seen in D. F, Stacked bar graph showing the proportions of lymphoid and myeloid cells between the histologic regions (IR, MIS, and bTIL) in specimen MEL1–1. APC, antigen-presenting cell; Macs, macrophages. G, CyCIF maximum-intensity projection images of MEL1-1 of the region of inflammatory regression (shown in C). Fields are stained for DNA (blue), PD1 (green), and MHC-II (HLA-DPB1: magenta). The dermal–epidermal junction is indicated with a dashed white line. The bar plot shows the proportions of all cell types in the epidermis (top), with lymphocyte and myeloid subset further highlighted, and in the dermis (bottom); color code is as in F. H, Heat map showing expression of genes related to immune checkpoints and T-cell activation between histologic mROIs in patient MEL1 (GeoMx). Significant upregulation in comparison with the EM region (P < 0.05) is highlighted in red and nonsignificance in gray. I, Schematics of remodeling of the TME with disease progression; see text for details. J, Summary of mechanisms of immune suppression detected in sample MEL1-1.
Comment in
- Cancer Discov. 12:1397.
- Cancer Discov. 12:1397.
Similar articles
- Transcriptional profiling of macrophages in situ in metastatic melanoma reveals localization-dependent phenotypes and function.
Martinek J, Lin J, Kim KI, Wang VG, Wu TC, Chiorazzi M, Boruchov H, Gulati A, Seeniraj S, Sun L, Marches F, Robson P, Rongvaux A, Flavell RA, George J, Chuang JH, Banchereau J, Palucka K. Martinek J, et al. Cell Rep Med. 2022 May 17;3(5):100621. doi: 10.1016/j.xcrm.2022.100621. Epub 2022 Apr 27. Cell Rep Med. 2022. PMID: 35584631 Free PMC article. - Spatially Resolved Transcriptomics Enables Dissection of Genetic Heterogeneity in Stage III Cutaneous Malignant Melanoma.
Thrane K, Eriksson H, Maaskola J, Hansson J, Lundeberg J. Thrane K, et al. Cancer Res. 2018 Oct 15;78(20):5970-5979. doi: 10.1158/0008-5472.CAN-18-0747. Epub 2018 Aug 28. Cancer Res. 2018. PMID: 30154148 - Stromal reaction in cutaneous melanoma.
Labrousse AL, Ntayi C, Hornebeck W, Bernard P. Labrousse AL, et al. Crit Rev Oncol Hematol. 2004 Mar;49(3):269-75. doi: 10.1016/j.critrevonc.2003.10.007. Crit Rev Oncol Hematol. 2004. PMID: 15036266 Review. - Lessons from cancer immunoediting in cutaneous melanoma.
Aris M, Barrio MM, Mordoh J. Aris M, et al. Clin Dev Immunol. 2012;2012:192719. doi: 10.1155/2012/192719. Epub 2012 Aug 14. Clin Dev Immunol. 2012. PMID: 22924051 Free PMC article. Review. - Tumor CD155 Expression Is Associated with Resistance to Anti-PD1 Immunotherapy in Metastatic Melanoma.
Lepletier A, Madore J, O'Donnell JS, Johnston RL, Li XY, McDonald E, Ahern E, Kuchel A, Eastgate M, Pearson SA, Mallardo D, Ascierto PA, Massi D, Merelli B, Mandala M, Wilmott JS, Menzies AM, Leduc C, Stagg J, Routy B, Long GV, Scolyer RA, Bald T, Waddell N, Dougall WC, Teng MWL, Smyth MJ. Lepletier A, et al. Clin Cancer Res. 2020 Jul 15;26(14):3671-3681. doi: 10.1158/1078-0432.CCR-19-3925. Epub 2020 Apr 28. Clin Cancer Res. 2020. PMID: 32345648
Cited by
- Defining precancer: a grand challenge for the cancer community.
Faupel-Badger J, Kohaar I, Bahl M, Chan AT, Campbell JD, Ding L, De Marzo AM, Maitra A, Merrick DT, Hawk ET, Wistuba II, Ghobrial IM, Lippman SM, Lu KH, Lawler M, Kay NE, Tlsty TD, Rebbeck TR, Srivastava S; Precancer Think Tank Team. Faupel-Badger J, et al. Nat Rev Cancer. 2024 Oct 1. doi: 10.1038/s41568-024-00744-0. Online ahead of print. Nat Rev Cancer. 2024. PMID: 39354069 Review. - Single-cell and spatial omics unravel the spatiotemporal biology of tumour border invasion and haematogenous metastasis.
Cheng X, Cao Y, Liu X, Li Y, Li Q, Gao D, Yu Q. Cheng X, et al. Clin Transl Med. 2024 Oct;14(10):e70036. doi: 10.1002/ctm2.70036. Clin Transl Med. 2024. PMID: 39350478 Free PMC article. Review. - Computational immune synapse analysis reveals T-cell interactions in distinct tumor microenvironments.
Wang VG, Liu Z, Martinek J, Foroughi Pour A, Zhou J, Boruchov H, Ray K, Palucka K, Chuang JH. Wang VG, et al. Commun Biol. 2024 Sep 28;7(1):1201. doi: 10.1038/s42003-024-06902-2. Commun Biol. 2024. PMID: 39341903 Free PMC article. - Spatial Transcriptomics Identifies Cellular and Molecular Characteristics of Scleroderma Skin Lesions: Pilot Study in Juvenile Scleroderma.
Liu T, Esencan D, Salgado CM, Zhao C, Lai YJ, Hutchins T, Sanyal A, Chen W, Torok KS. Liu T, et al. Int J Mol Sci. 2024 Aug 23;25(17):9182. doi: 10.3390/ijms25179182. Int J Mol Sci. 2024. PMID: 39273131 Free PMC article. - MDM2 inhibitors in cancer immunotherapy: Current status and perspective.
Zeng Q, Zeng S, Dai X, Ding Y, Huang C, Ruan R, Xiong J, Tang X, Deng J. Zeng Q, et al. Genes Dis. 2024 Mar 28;11(6):101279. doi: 10.1016/j.gendis.2024.101279. eCollection 2024 Nov. Genes Dis. 2024. PMID: 39263534 Free PMC article. Review.
References
- O'Donnell JS, Teng MWL, Smyth MJ. Cancer immunoediting and resistance to T cell-based immunotherapy. Nat Rev Clin Oncol 2019;16:151–67. - PubMed
- Shain AH, Yeh I, Kovalyshyn I, Sriharan A, Talevich E, Gagnon A, et al. . The genetic evolution of melanoma from precursor lesions. N Engl J Med 2015;373:1926–36. - PubMed
- Lian CG, Murphy GF. The genetic evolution of melanoma. N Engl J Med 2016;374:994–5. - PubMed
Publication types
MeSH terms
Substances
Grants and funding
- P30 NS072030/NS/NINDS NIH HHS/United States
- U2C CA233262/CA/NCI NIH HHS/United States
- K99 CA256497/CA/NCI NIH HHS/United States
- R50 CA252138/CA/NCI NIH HHS/United States
- P30 CA006516/CA/NCI NIH HHS/United States
LinkOut - more resources
Full Text Sources
Other Literature Sources
Medical
Research Materials