目的·通过单细胞转录组分析系统解析黏膜恶性黑色素瘤（mucosal malignant melanoma，MM）与皮肤恶性黑色素瘤（cutaneous malignant melanoma，CM）微环境的免疫组分差异，提示MM中CD8⁺ T细胞耗竭的关键调控机制。方法·收集整合3例初治MM患者手术样本及3例CM单细胞转录组数据，共纳入36 531个细胞进行分析。数据预处理包括批次校正、严格质控（基于线粒体基因比例及基因数筛选）、无监督聚类及细胞类型注释（依据经典标记基因）。进一步采用差异基因分析（Wilcoxon检验）、基因集富集分析［基于基因本体（Gene Ontology，GO）和京都基因与基因组百科全书（Kyoto Encyclopedia of Genes and Genomes，KEGG）数据库］、细胞互作网络分析及癌症基因组图谱（The Cancer Genome Atlas，TCGA）预后关联分析，全面解析MM与CM肿瘤免疫微环境的异质性。结果·与CM相比，MM中CD8⁺ T细胞的细胞毒性相关基因（GZMB、IFNG等）表达降低（P<0.001），而耗竭标志物（PDCD1、LAG3等）则显著上调（P<0.001），同时耗竭性CD8⁺ T细胞亚群的比例增加近6倍［（CM：3.38%） vs （MM：19.26%），P<0.001］，且终末耗竭评分更高。此外，MM微环境中的中性粒细胞浸润水平显著高于CM［（CM：1.6%） vs （MM：34.1%），P=0.009］。单细胞分析进一步将其划分为5个亚群。其中，CD10⁺中性粒细胞亚群（占MM中性粒细胞的23.52%）同时高表达促炎分子（S100A8/A9）和免疫抑制分子（MME/CD10、CD55），且该亚群的髓源性抑制细胞（myeloid-derived suppressor cell，MDSC）特征评分显著高于其他亚群。TCGA及MM队列分析进一步证实，该亚群的高表达基因与患者不良预后独立相关（P=0.049），提示该亚群参与肿瘤免疫抑制作用。机制上，CD10⁺中性粒细胞通过SELPLG-SELL配体-受体对与耗竭性CD8⁺ T细胞互作，驱动T细胞功能失活。结论·对MM临床样本进行单细胞转录组测序，并系统比较MM与CM的单细胞免疫图谱，提示MM中CD10⁺中性粒细胞通过SELPLG-SELL通路诱导CD8⁺ T细胞终末耗竭，解析了MM免疫治疗耐受的分子机制，为靶向中性粒细胞-T细胞互作提供新策略。

Objective ·To compare the immune microenvironment of mucosal malignant melanoma (MM) and cutaneous malignant melanoma (CM) using single-cell transcriptome analysis system, and to elucidate the key regulatory mechanism of CD8⁺ T cell depletion in MM. Methods ·A total of 36 531 cells from three treatment-naïve MM surgical specimens and three CM samples were subjected to single-cell RNA sequencing. Data pre-processing included batch correction, strict quality control (based on mitochondrial gene ratio and gene number screening), unsupervised clustering, and cell-type annotation (using established marker genes). Differential gene analysis (Wilcoxon test), gene set enrichment analysis [based on Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases], cell-cell interaction network analysis, and prognostic association analysis using. The Cancer Genome Atlas (TCGA) were used to comprehensively analyse the heterogeneity of MM and CM tumour immune microenvironment. Results ·Compared with CM, the expression of cytotoxicity-related genes (GZMB, IFNG, etc.) in CD8⁺ T cells in MM was reduced (P<0.001), while the depletion markers (PDCD1, LAG3, etc.) were significantly up-regulated (P<0.001). The proportion of depleted CD8⁺ T cell subsets increased by nearly 6 times (CM: 3.38% vs MM: 19.26%, P<0.010), and the terminal depletion score was higher. In addition, the level of neutrophil infiltration in the MM microenvironment was significantly higher than that in CM (CM: 1.6% vs MM: 34.1%, P=0.009), and single-cell analysis further divided it into five subgroups. Among them, the CD10⁺ neutrophil subset (accounting for 23.52% of MM neutrophils) highly expressed pro-inflammatory molecules (S100A8/A9) and immunosuppressive molecules (MME/CD10, CD55), and the characteristic scores of myeloid-derived suppressor cells (MDSCs) in this subset were significantly higher than those of other subgroups. TCGA and MM cohort analyses further confirmed that the highly expressed genes of this subset were independently associated with poor prognosis (P=0.049), suggesting that this subset was involved in tumour immunosuppressive effects. Mechanistically, CD10⁺ neutrophils interacted with depleted CD8⁺ T cells through the SELPLG-SELL ligand-receptor pair, driving T-cell inactivation. Conclusion ·Single-cell transcriptome sequencing was performed on MM clinical samples, and the single-cell immunomes of MM and CM were systematically compared, suggesting that CD10⁺ neutrophils in MM induce CD8⁺ T-cell terminal depletion through the SELPLG-SELL pathway. The molecular mechanism of MM immunotherapy tolerance was elucidated, providing a new strategy for targeting neutrophil-T cell interaction.