Work flow analysis for the multiplexed immunofluorescence staining.

Description of the MELPREDICT cohort and design of experiments performed in the study.

Background Melanoma shows one of the highest response rates to immune checkpoint inhibitors (ICIs), yet nearly half of patients experience primary or acquired resistance. While immune contexture strongly influences therapeutic efficacy, tumor cell-intrinsic features are increasingly recognized as key regulators of antitumor immunity. In particular, intratumoral heterogeneity driven by melanoma cell plasticity underlies diverse immune escape mechanisms. How this plasticity shapes ICI outcomes in patients remains poorly defined.Methods Tumor cell states and immune contexture were assessed in 57 primary cutaneous melanomas from stage III patients, collected prior to adjuvant anti-programmed cell death protein-1 (anti-PD-1) therapy and stratified according to 2-year relapse status (33 relapse-free, 24 relapsed). Whole slide multiplex immunofluorescence was combined with spatial transcriptomics (Visium, n=4) to investigate the spatial architecture of melanoma cell states, T cells, tumor-associated macrophages (TAMs), dendritic cell subsets, and tertiary lymphoid structures.Results Unsupervised clustering of melanoma cells identified distinct phenotypic states that formed spatially restricted homotypic patches. From these data, we defined a melanoma plasticity ratio (undifferentiated/differentiated tumor patches), which was significantly associated with reduced relapse-free survival. Integrated immune analyses recapitulated prognostically distinct immunotypes, with macrophage subsets displaying striking spatial compartmentalization. Antitumoral macrophages preferentially infiltrated differentiated melanoma regions, while protumoral macrophages localized to undifferentiated patches. Spatial transcriptomics confirmed that melanoma cell states tightly shape the neighboring immune microenvironment, with macrophages emerging as pivotal players. Their polarization was further influenced by tumor-derived signals in addition to microenvironmental cues (interferon-gamma, hypoxia). Entropy-based integration of melanoma cell states, TAMs, and T cell subsets uncovered two dominant spatial ecosystems with opposing associations to ICI efficacy. Ecosystems enriched in differentiated melanoma cells, programmed death-ligand 1 (PD-L1)+ TAMs, and PD-1+CD8+, and CD4+ T cells correlated with favorable outcomes, whereas ecosystems composed of undifferentiated melanoma cells with PD-L1- protumoral TAMs and PD-1-CD8+ T cells were associated with relapse.Conclusions Our study uncovers how cancer cell plasticity shapes spatially organized tumor-immune ecosystems that critically modulate adjuvant ICI efficacy in melanoma. These findings highlight melanoma cell plasticity as a key driver of immune evasion via macrophages reprogramming, through targetable interactions that may represent novel therapeutic avenues to enhance ICI efficacy.

Optimized immunofluorescence staining conditions for the multiplex immunofluorescence panel

Clinical features of MELPREDICT patients at baseline prior immunotherapy onset

Composition of DC aggregates and frequency in R and NR patients.

Cancer cell plasticity plays a key role in tumor progression and treatment resistance in melanoma. While the transcriptional programs enabling adaptative switching between melanocytic and mesenchymal phenotypes are well characterized, unravelling druggable epigenetic regulators that sustain melanoma cell adaptation and resistance remains crucial. Herein, we identified TRIM24, a bromodomain protein frequently upregulated during melanoma metastatic progression, as a crucial regulator of melanoma cell plasticity towards invasive/resistant states. shRNA-mediated knock-down of TRIM24 or degradation using a TRIM24-specific PROTAC decrease the migratory capacities and increase the sensitivity to BRAF inhibitors of melanoma cells. Integration of transcriptomic (RNA-seq) and epigenomic (ATAC-seq, CUT&Tag) analyses reveals that TRIM24 reprograms the epigenome of melanoma cells, promoting mesenchymal and repressing melanocytic transcriptional programs. We further define a TRIM24-specific transcriptional signature, that is consistently enriched in treatment-resistant mesenchymal subpopulations in melanoma single-cell RNA-seq datasets. Accordingly, analysis of TRIM24 protein expression in melanoma patients highlights that high TRIM24 expression correlates with relapse to adjuvant immunotherapy. Finally, TRIM24 knock-down in immunocompetent mouse models synergises with immune checkpoint inhibitors. Overall, our findings spotlight TRIM24 as a major epigenetic regulator driving melanoma cell dedifferentiation and resistance to therapy, representing a promising druggable target to reverse phenotype switching and resensitize to treatment. ### Competing Interest Statement The authors have declared no competing interest.

Intra-tumoral heterogeneity in melanoma arises from dynamic cancer cell plasticity and underlies various mechanisms of immune escape. Here, we combined high-plex immunofluorescence imaging with spatially resolved transcriptomics to map the architecture of melanoma cell states and their interactions with the immune microenvironment in primary cutaneous tumours prior to adjuvant anti-PD1 immune checkpoint inhibitor (ICI) treatment. Computational analyses showed that melanoma cells organise into spatially restricted patches, with a preferential organisation of undifferentiated cells associated with poor ICI efficacy. Neighbouring immune cell composition varied according to cancer cell states, with a crucial involvement of specific subsets of tumour-associated macrophages, driven by signalling pathways involving tumour-derived and microenvironmental cues such as IFN-γ and hypoxia. Integrated spatial analyses further revealed tumour-immune ecosystems that stratify patient outcomes, delineating configurations either associated with ICI efficacy or metastatic relapse. These results uncover the spatial landscape of tumour ecosystems and identify signalling pathways as potential targets for improving the efficacy of ICI in melanoma. Highlights ### Competing Interest Statement The authors have declared no competing interest.

ABSTRACT Background Programmed cell death protein‐1 (PD1) antibodies are currently the standard treatment for resected high‐risk melanoma, yet recurrence rate remains high. Objectives This real‐life observational study aimed to describe the outcomes of patients with resected high‐risk melanoma following adjuvant anti‐PD1 immunotherapy and identify factors associated with recurrence risk. Materials and Methods A total of 235 patients with resected stage III/IV melanoma treated with adjuvant nivolumab or pembrolizumab were included. Imaging scans and cerebral imaging were performed every 12 weeks to detect recurrences. Adverse events were collected. Univariate and multivariate analyses were performed to identify predictive factors of recurrence. Overall survival (OS) and recurrence‐free survival (RFS) were estimated. Results Among the 235 patients, 103 experienced at least one recurrence (43%); first recurrences were predominantly locoregional (47%). The predictive factor for recurrence identified by multivariate analysis was ulceration (RR 2,03, 95% CI [1,20; 2,86]). RFS was estimated at 75% [70–81] at 12 months and at 64% [58–71] at 24 months. RFS at 12 months was significantly lower in patients with ulcerations (RFS at 83%) compared to those without ulceration (RFS at 66%), p < 0.01. Overall survival (OS) was estimated at 91% [87%–94%] at 12 months and 84% [79%–89%] at 24 months. The OS after a first recurrence was estimated at 69% [60%–80%] at 12 months and decreased to 43% [32%–57%] at 24 months. After a first locoregional recurrence, surgery with a year of adjuvant immunotherapy (40%) was the favoured therapeutic approach. For distant recurrences, clinical trial enrolment was preferred (21%). Double curative immunotherapy was the preferred strategy for cerebral recurrences (30%). Conclusions In this cohort, nearly half of the patients underwent recurrences and RFS at 24 months was 64%. The RFS and OS data were comparable o those reported in the pivotal study Ulceration was the only significant predictive factor for recurrence, associated with decreased RFS at 24 months.

Dendritic cells (DC) are promising targets for cancer immunotherapies because of their central role in the initiation and control of immune responses. The type 1 conventional DC (cDC1) population is of particular interest because of its ability to cross-present antigens to CD8+ T cells. cDC1s also secrete cytokines that allow Th1 cell polarization and NK cell activation and recruitment. However, the spatial organization and specific functions of cDC1s in response to immunotherapy remain to be clearly characterized in human tumors. In this study, we used a multiplexed immunofluorescence analysis pipeline coupled with computational image analysis to determine the spatial organization of cDC1s in skin lesions from a cohort of patients with advanced melanoma treated with immune checkpoint inhibitors (ICI). For this, we performed a whole-slide image analysis of cDC1 infiltration, distribution, and spatial interaction with key immune partners such as CD8+ T cells and plasmacytoid DCs. We also analyzed LAMP3+ DCs, which correspond to a mature subset of tumor-infiltrating DCs. Distance and cell network analyses demonstrated that cDC1s exhibited a scattered distribution compared with tumor-infiltrating plasmacytoid DCs and LAMP3+ DCs, which were preferentially organized in dense areas with high homotypic connections. The proximity and interactions between CD8+ T cells and cDC1s were positively associated with the response to ICIs. In conclusion, our study unravels the complex spatial organization of cDC1s and their interactions with CD8+ T cells in lesions of patients with melanoma, shedding light on the pivotal role of these cells in shaping the response to ICIs.

Tumor cells can evade antitumor immune response by expressing the PD-L1 ligand, leading to the inhibition of PD-1-expressing T lymphocytes. The mechanisms that regulate PD-L1 expression in cancer cells are imperfectly characterized. The transcription factor ZEB1, a major regulator of phenotype switching in melanoma cells, was shown to promote immune escape in melanoma by repressing T cell infiltration. Using inducible models of phenotype switching and ZEB1 gain/loss-of-function melanoma, we show that ZEB1 binds to the CD274 (PD-L1) promoter, directly enhancing PD-L1 mRNA transcription and its expression at the cell membrane. Furthermore, using single-cell spatial analyses on human primary melanoma samples, we demonstrate the correlation of ZEB1 and PD-L1 expression in tumor cells. Overall, these data identify ZEB1-mediated regulation of PD-L1 tumor expression as a mechanism that could contribute to immune escape in melanoma.

Background Dendritic cells (DCs) are promising targets for cancer immunotherapies owing to their central role in the initiation and the control of immune responses. Their functions encompass a wide range of mechanisms mediated by different DC subsets. Several studies have identified human tumor- associated DC (TA-DC) populations through limited marker-based technologies, such as immunostaining or flow cytometry. However, tumor infiltration, spatial organization and specific functions in response to immunotherapy of each DC subset remain to be defined.Methods Here, we implemented a multiplexed immunofluorescence analysis pipeline coupled with bio-informatic analyses to decipher the tumor DC landscape and its spatial organization within melanoma patients’ lesions, and its association with patients’ response to immune checkpoint inhibitors (ICI). For this aim, we analyze a cohort of 41 advanced melanoma patients treated with anti- PD1 alone or associated with anti-CTLA4. Distance and cell network analyses were performed to gain further insight into the spatial organization of tumor-associated DCs. A Digital Spatial Profiling analysis further characterized ecosystem of tumor-infiltrating DCs.Results Plasmacytoid DCs (pDCs) were the most abundant DC population, followed by conventional cDC1 and mature DCs, present in equal proportions. In contrast to CD8+ T cell frequency, and despite varying densities, all DC subsets were associated with a favorable response to ICI. Distance and cell network analyses demonstrated that tumor-infiltrating DCs were largely organized in dense areas with high homotypic connections, except for cDC1 that exhibited a more scattered distribution. We identified four patterns of ecosystems with distinct preferential interactions between DC subsets. Significantly, the proximity and interactions between CD8+ T cells and cDC1 were positively associated with patients’ response to ICI.Conclusions Our study unravels the complex spatial organization of DC subsets and their interactions in melanoma patient lesions, shedding light on their pivotal role in shaping the response to ICI. Our discoveries regarding the spatial arrangement of cDC1, especially with CD8+ T cells, provide valuable clues for improving immunotherapeutic strategies in melanoma patients.What is already known on this topic Dendritic cells (DCs) are promising targets for cancer immunotherapies owing to their central role in the initiation and the control of immune responses. Although conventional type 1 dendritic cells (cDC1) were proposed to contribute to immunotherapy response, their precise functions and interactions with other immune populations in human cancers are largely unknown.What this study adds This study provides a precise characterization of the spatial distribution and organization of tumor- infiltrating DCs in a large cohort of advanced melanoma patients, and in correlation with response to immunotherapy. While DCs are organized in dense areas with high homotypic connections, cDC1 exhibit a more scattered distribution and form heterotypic aggregates with other DC subsets. More importantly, a close connection between cDC1 and CD8 T cell is uniquely correlated with the patients’ response to immunotherapy.How this study might affect research, practice or policy This study improves our understanding of CD8-DC spatial organization within the tumor microenvironment and will have a broad spectrum of implications in the design of anti-tumor immune-activating compounds and the design of biomarkers of response to immunotherapy for melanoma patients.### Competing Interest StatementThe authors have declared no competing interest.

The need for reliable biomarkers to predict clinical benefit from anti-PD1 treatment in metastatic melanoma (MM) patients remains unmet. Several parameters have been considered in the tumor environment or the blood, but none has yet achieved sufficient accuracy for routine clinical practice. Whole blood samples from MM patients receiving second-line anti-PD1 treatment (NCT02626065), collected longitudinally, were analyzed by flow cytometry to assess the immune cell subsets absolute numbers, the expression of immune checkpoints or ligands on T cells and the functionality of innate immune cells and T cells. Clinical response was assessed according to Progression-Free Survival (PFS) status at one-year following initiation of anti-PD1 (responders: PFS > 1 year; non-responders: PFS ≤ 1 year). At baseline, several phenotypic and functional alterations in blood immune cells were observed in MM patients compared to healthy donors, but only the proportion of polyfunctional memory CD4+ T cells was associated with response to anti-PD1. Under treatment, a decreased frequency of HVEM on CD4+ and CD8+ T cells after 3 months of treatment identified responding patients, whereas its receptor BTLA was not modulated. Both reduced proportion of CD69-expressing CD4+ and CD8+ T cells and increased number of polyfunctional blood memory T cells after 3 months of treatment were associated with response to anti-PD1. Of upmost importance, the combination of changes of all these markers accurately discriminated between responding and non-responding patients. These results suggest that drugs targeting HVEM/BTLA pathway may be of interest to improve anti-PD1 efficacy.

Cell plasticity sustains intra-tumor heterogeneity and treatment resistance in melanoma. Deciphering the transcriptional mechanisms governing reversible phenotypic transitions between proliferative/differentiated and invasive/stem-like states is required. Expression of the ZEB1 transcription factor is frequently activated in melanoma, where it fosters adaptive resistance to targeted therapies. Here, we performed a genome-wide characterization of ZEB1 transcriptional targets, by combining ChIP-sequencing and RNA-sequencing, upon phenotype switching in melanoma models. We identified and validated ZEB1 binding peaks in the promoter of key lineage-specific genes crucial for melanoma cell identity. Mechanistically, ZEB1 negatively regulates SOX10-MITF dependent proliferative/melanocytic programs and positively regulates AP-1 driven invasive and stem-like programs. Comparative analyses with breast carcinoma cells revealed lineage-specific ZEB1 binding, leading to the design of a more reliable melanoma-specific ZEB1 regulon. We then developed single-cell spatial multiplexed analyses to characterize melanoma cell states intra-tumoral heterogeneity in human melanoma samples. Combined with scRNA-Seq analyses, our findings confirmed increased ZEB1 expression in Neural-Crest-like cells and mesenchymal cells, underscoring its significance in vivo in both populations. Overall, our results define ZEB1 as a major transcriptional regulator of cell states transitions and provide a better understanding of lineage-specific transcriptional programs sustaining intra-tumor heterogeneity in melanoma.

Natural Killer (NK) cell subsets differ to ensure complementary and crucial roles in tumor immunosurveillance. Their biology is critically regulated by cytokines. Here, we show that IL-33 synergizes with IL-12 to strongly activate a subset of CD56dim NK cells acquiring ST2 expression. Transcriptomic and biological analysis of human ST2+ CD56dim NK cells revealed a distinct intermediate differentiation state between canonical CD56bright and CD56dim NK cells, combining high proliferative properties, cytokines/chemokines production, and cytotoxicity. NK cells expressing ST2 protein or exhibiting a ST2-linked transcriptional signature were identified in human and mouse tumors. Accordingly, IL-12 unleashes human breast tumor ST2+ NK cell potential to produce IFN-γ in response to IL-33 and IL-33/IL-12 co-injection resulted in a NK-dependent IFN-γ secretion and anti-tumor effects in murine mammary tumors. An IL33hi-NKhi score in solid tumors correlated with increased progression-free patient survival. Our findings thus identify polyfunctional ST2+ NK cells which effector functions can be harnessed by IL-33 to boost anti-tumor immunity. One sentence summary The IL-33/IL-33R(ST2)/NK cell axis is a key determinant of cancer immunity and immunotherapy.

Melanoma is the deadliest form of skin cancer due to its propensity to metastasize. It arises from melanocytes, which are attached to keratinocytes within the basal epidermis. Here, we hypothesize that, in addition to melanocyte-intrinsic modifications, dysregulation of keratinocyte functions could initiate early-stage melanoma cell invasion. We identified the lysolipid sphingosine 1-phosphate (S1P) as a tumor paracrine signal from melanoma cells that modifies the keratinocyte transcriptome and reduces their adhesive properties, leading to tumor invasion. Mechanistically, tumor cell-derived S1P reduced E-cadherin expression in keratinocytes via S1P receptor dependent Snail and Slug activation. All of these effects were blocked by S1P2/3 antagonists. Importantly, we showed that epidermal E-cadherin expression was inversely correlated with the expression of the S1P-producing enzyme in neighboring tumors and the Breslow thickness in patients with early-stage melanoma. These findings support the notion that E-cadherin loss in the epidermis initiates the metastatic cascade in melanoma.

Metastatic melanoma patients carrying a BRAF V600 mutation can be treated with BRAF inhibitors (BRAFi), in combination with MEK inhibitors (MEKi), but innate and acquired resistance invariably occurs. Resistance can involve transcriptional- and epigenetic-based phenotypic adaptations, as yet unpredictable. Predicting patient response to targeted therapies is crucial to guide clinical decision. We describe here the development of a highly efficient patient-derived xenograft model adapted to patient melanoma biopsies, using the avian embryo as a host (AVI-PDX ™ ). In this in vivo paradigm, we depict a fast and reproducible tumor engraftment of patient samples within the embryonic skin, preserving key molecular and phenotypic features. We show that sensitivity and resistance to BRAFi/MEKi targeted therapies can be reliably modeled in these AVI-PDX ™ , as well as synergies with other drugs, such as HDACi. We further provide proof-of-concept that the AVI-PDX ™ models the diversity of responses of melanoma patients to BRAFi/MEKi, within days, hence positioning it as a valuable tool for the design of personalized medicine assays and for the evaluation of novel combination strategies.

Torque Teno Virus (TTV) is a small, non-enveloped, single-stranded and circular DNA virus that infects the majority of the population worldwide. Increased levels of plasma TTV viral load have been observed in various situations of immune deficiency or dysregulation, and several studies have suggested that TTV levels may be inversely correlated with immune competence. The measurement of TTV viremia by qPCR has been proposed as a potential biomarker for the follow-up of functional immune competence in immunosuppressed individuals, particularly hematopoietic stem cell transplant recipients. We hypothesized that TTV viral load could be used as a prognostic marker of immune checkpoint inhibitor (ICI) efficacy, and therefore investigated the TTV viral load in melanoma patients treated with nivolumab or pembrolizumab before and after 6 months of treatment. In the present study, TTV viral load was not different in melanoma patients before anti-PD-1 introduction compared to healthy volunteers, was not modified by ICI treatment and did not allowed to distinguish patients with treatment-sensitive tumor from patients with treatment-resistant tumor.

Interleukin (IL)-33 is an alarmin belonging to the IL-1 family. Through its receptor ST2, IL-33 promotes both type 1 and type 2 immune responses, depending on the type of responding cells and the microenvironment in damaged tissues. Consistently, paradoxical roles of IL-33 have been reported in the context of cancer. These observations prompted us to investigate i) the expression of IL-33/ST2 in breast tumors and ii) the role of IL-33/ST2 in NK cell activation in physiologic and tumor contexts. Using immunohistochemistry (IHC) and transcriptomic analyses, we observed that IL-33 is expressed at higher levels in ductal carcinoma in situ (DCIS) compared to invasive breast cancer (IBC), especially in the stroma and in luminal BC subtype. In situ, IL-33 was mainly detected in endothelial cells and at lower extent in scattered cells within the stroma of breast tumors. Furthermore, we unraveled a new pathway for NK cells’ activation where IL-12 upregulates ST2 on human NK cells, which in turn become responsive to IL-33 by secreting high levels of IFN-γ and increasing their cytotoxic activity. This effect was specific to a subset of human CD56dim NK cells (20%) and was dependent on STAT-4 phosphorylation. In mice, we also observed a strong activation of spleen NK cell by IL-33 in combination with IL-12 in vitro. Finally, using IL-33ko mice, preliminary results show a contribution of endogenous IL-33 in the prevention of experimental lung metastasis development following B16 iv injection in therapeutic settings using exogenous administration of IL-12. Following up on these results, our aim is now to i) better characterize IL-33-responsive NK cells subpopulation in blood, ii) understand why human CD56bright NK cells do not upregulate ST2 despite response to IL-12 by phosphorylating STAT-4, iii) characterize NK cells’ response to IL-33 in human breast tumors, iv) evaluate ST2 expression by immune cells infiltrating breast and ovarian tumors, and v) confirm the activating role of IL-33 on NK cells’ biology in tumor models in vivo. All together, our observations are in favor of an NK-mediated antitumor role of IL-33 that we are currently pursuing as a potential novel therapeutic strategy in cancer. Citation Format: Anais Eberhardt, Elena Blanc, Emilie Lardenois, Sarah Renaudineau, Jennifer Herbulot, Benoit Dumont, Isabelle Durand, Emilie Charrier, Pauline Schmitt, Jean-Philippe Girard, Antoine Marcais, Thierry Walzer, Christophe Caux, Nathalie Bendriss-Vermare. Decipher the role of IL-33 as an activator of NK cells’ antitumor activity [abstract]. In: Proceedings of the AACR Special Conference on Tumor Immunology and Immunotherapy; 2018 Nov 27-30; Miami Beach, FL. Philadelphia (PA): AACR; Cancer Immunol Res 2020;8(4 Suppl):Abstract nr B84.