TYRP1 defines a proliferative melanoma cell subpopulation, driving malignant progression and therapy resistance via the GPNMB–Notch1–SOX10/MITF axis Authors: Chengjun Hu, Lingmei Zhu, Pengju Fan, Xin Bu, Chenchen Zuo Published: 2026 ABSTRACT BackgroundTumor cell heterogeneity contributes to melanoma progression, therapeutic resistance, and clinical outcome variability. However, the identity and functional role of specific proliferative subpopulations remain incompletely understood. This study aims to characterize TYRP1-positive melanoma cells and elucidate their role in tumor proliferation, signaling regulation, and treatment response.MethodsWe analyzed single-cell RNA sequencing (scRNA-seq) data from primary and metastatic melanoma samples to identify transcriptionally distinct tumor cell subtypes. Functional validation of TYRP1-positive cells was performed using patient-derived organoids, TYRP1-overexpressing melanoma cell lines (A375, SK-MEL-28), and xenograft mouse models. The downstream molecular mechanisms were investigated through gene expression profiling, siRNA-mediated knockdown, recombinant protein treatment, and pathway inhibition assays. Therapeutic responses were assessed using dabrafenib and pembrolizumab treatments.ResultsTYRP1 marked a transcriptionally distinct melanoma subpopulation associated with poor patient survival. TYRP1-high organoids and cell lines exhibited significantly enhanced proliferation in vitro and accelerated tumor growth in vivo, without increased metastatic capacity. Mechanistically, TYRP1 induced expression of GPNMB, which activated Notch1 signaling and subsequently upregulated SOX10 and MITF. These transcription factors formed a positive feedback loop with TYRP1 that maintained the proliferative phenotype. GPNMB or Notch1 inhibition disrupted this loop and suppressed tumor growth. Importantly, TYRP1-overexpressing tumors demonstrated resistance to immune checkpoint blockade but increased sensitivity to dabrafenib, suggesting distinct therapeutic vulnerabilities.ConclusionsOur findings identify TYRP1 as a marker of a highly proliferative melanoma subpopulation that promotes tumor progression through the GPNMB–Notch1–SOX10/MITF axis. The TYRP1–SOX10–MITF feedback loop represents a key driver of melanoma proliferation and a potential biomarker for stratifying therapeutic response, offering a novel avenue for precision treatment in melanoma.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12967-026-08436-z. Introduction Melanoma is one of the most aggressive forms of skin cancer, characterized by rapid progression, early metastasis, and a high degree of intratumoral heterogeneity [1]. Despite advances in targeted therapies and immune checkpoint inhibitors, many patients develop resistance or relapse, underscoring the need to better understand the molecular determinants that drive melanoma malignancy and treatment response [2–4] Recent studies leveraging single-cell RNA sequencing (scRNA-seq) have revealed profound transcriptional diversity within melanoma tumors, suggesting that distinct cellular subpopulations may differentially contribute to tumor behavior and therapeutic outcome [5, 6]. In line with the increasing understanding regarding the use of plant-sourced or biosynthesized products in regulating cancer cell activity, several researches have indicated that such molecules are capable of regulating not only cell viability but also oxidative stress and selective cytotoxicity in melanoma and similar cancers. For example, biosynthesized selenium nanoparticles from Solanum tuberosum peels demonstrated effective cytotoxic and antimicrobial properties, showing the importance of the use of agricultural waste sources in controlling cell proliferation in sub-populations of tumors in a manner that preserves biocompatibility [7]. Similarly, extracts from prickly pear grown in Turkey revealed high levels of antioxidant and antimicrobial capacities, emphasizing the importance of natural phytochemicals as tools to enhance cell tolerance or selective stress in heterogeneous tumor conditions [8]. Carbon-based nanomaterials have also demonstrated oxidative stress reduction and biological improvement, further emphasizing the importance of nano-based structures in fine tuning the cancer cell response [9]. Finally, Cuscuta spp. extracts [10] showed differential regulation of cell viability between normal fibroblast and colorectal adenocarcinoma cell lines, indicating context-dependent mechanisms, similar to proliferation vs. non-proliferation in TYRP1-positive melanoma subpopulations. TYRP1, namely Tyrosinase-related protein 1, is a melanocyte lineage-restricted gene that has long been recognized for its role in melanin biosynthesis as part of the tyrosinase enzyme family [11]. While traditionally regarded as a pigmentation-related marker, emerging evidence suggests that TYRP1 may exert broader biological functions in melanoma beyond its enzymatic role [12]. Elevated TYRP1 expression has been observed in subsets of primary and metastatic melanomas and has been associated with adverse clinical outcomes in some patient cohorts [13–15]. However, the biological relevance and mechanistic underpinnings of TYRP1 expression in melanoma pathogenesis remain poorly defined. Importantly, TYRP1 is transcriptionally regulated by the melanocyte-lineage transcription factor MITF, which is a master regulator of melanocytic identity, cell survival, and proliferative signaling [16, 17]. In turn, MITF is under the control of upstream regulators such as SOX10, another critical transcription factor that governs neural crest differentiation and melanoma plasticity [18]. These lineage-associated transcription factors not only define melanocyte fate but also contribute to dynamic phenotypic switching between proliferative and invasive states during melanoma progression [19]. Despite this regulatory context, whether TYRP1 functions merely as a downstream target or plays an active role in reinforcing transcriptional networks that promote melanoma growth and therapeutic resistance remains unclear. Moreover, it is unknown whether TYRP1 defines a discrete tumor cell subpopulation with specific phenotypic traits or engages in autocrine or paracrine signaling that shapes the tumor microenvironment. In this study, we employed an integrative approach combining single-cell transcriptomics, patient-derived organoid models, and in vivo functional validation to investigate the role of TYRP1-positive melanoma cells. We discovered that TYRP1 identifies a highly proliferative tumor cell subpopulation that orchestrates a self-reinforcing transcriptional circuit through GPNMB-mediated activation of Notch1 signaling and induction of SOX10 and MITF. This TYRP1–GPNMB–Notch1–SOX10/MITF axis sustains melanoma cell proliferation and modulates therapeutic responses to BRAF inhibition and immunotherapy. Our findings not only uncover a novel regulatory mechanism underlying melanoma heterogeneity and plasticity but also propose TYRP1 as a potential prognostic biomarker and therapeutic target in melanoma. Materials and methods Cell culture and reagents Human melanoma cell lines A375 and SK-MEL-28 were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 1% penicillin/streptomycin (Gibco, USA) at 37 °C in a humidified incubator with 5% CO₂. All cell lines were recently authenticated by STR profiling and tested negative for mycoplasma contamination. Recombinant human GPNMB and SNAI2 were purchased from R&D Systems (USA) and reconstituted according to the manufacturer’s instructions. Brontictuzumab (Notch1 inhibitor) was obtained from MedChemExpress (USA). Dabrafenib and Temozolomide were purchased from Selleck Chemicals (USA). Organoid establishment Fresh melanoma tumor tissues were mechanically and enzymatically dissociated using Collagenase IV, Hyaluronidase, and DNase I (all from Sigma, USA) in Advanced DMEM/F12 (Gibco, USA). Cells were embedded in Matrigel (Corning, USA) and cultured in organoid medium supplemented with B27 (Gibco, USA), N2 (Gibco, USA), EGF, Noggin, and R-spondin1 (all from PeproTech, USA). Medium was changed every 2–3 days. Organoids were passaged every 7–10 days, and TYRP1 expression was assessed via immunofluorescence. Single-cell analysis Single-cell RNA-seq data (GSE139829), including eight primary and three metastatic melanoma samples, were obtained from GEO. Data were processed using Seurat (v4.3.0) in R (v4.1.3). Initial quality control: cells with < 400 UMIs, < 100 or > 8,000 detected genes, or > 10% mitochondrial contents were excluded. To remove potential doublets, we applied DoubletFinder (v2.0.3) with an expected doublet rate of 7.6%, eliminating an additional 1,034 cells. This yielded 58,509 high-quality cells for downstream analysis (median UMIs per cell = 2,845; median genes per cell = 1,532). After LogNormalize normalization, the top 2,000 variable genes were selected. To correct for batch effects across the 11 samples, we applied the Harmony algorithm (v0.1.0) on the PCA embeddings. The robustness of clustering was assessed by testing Louvain resolution parameters from 0.2 to 0.8 (step 0.1). A resolution of 0.4 was chosen because it gave the highest average silhouette width (0.61) and balanced biological resolution with technical noise; key findings (e.g., TYRP1 + subpopulation) were consistent across resolutions 0.3–0.5. PCA was performed, and the top 30 Harmony‑corrected components were used for Louvain clustering (resolution = 0.4) and UMAP visualization. Cell type annotation was based on canonical markers. Cluster-specific markers were identified via Wilcoxon rank-sum test (log2FC > 0.25; gene expression in ≥ 10% of cells). Cell type annotation was based on canonical markers. Cluster-specific markers were identified via Wilcoxon rank-sum test (log2FC > 0.25; gene expression in ≥ 10% of cells). Cell–cell communication was inferred using CellChat (v1.6.1), with significant ligand–receptor interactions defined as p < 0.05. Pseudotime trajectories were reconstructed using Monocle2 (v2.22.0). Functional enrichment analysis was performed using clusterProfiler (v4.4.4) for GO and KEGG terms, with adjusted p < 0.05 indicating significance. Clinical specimens and TCGA analysis Formalin-fixed, paraffin-embedded melanoma tissues (n = 25) were obtained from Xiangya Hospital and experiments were approved by the Ethics Committee of Xiangya Hospital (SLBH-202506100001), including 15 primary and 10 metastatic cases, all histologically confirmed by two independent pathologists according to the 8th AJCC staging system. A 5 years follow-up visit was conducted for confirm the metastasis status again to avoid Omission. Tumor stage distribution included stage II (n = 15), stage III (n = 8), and stage IV (n = 2). Written informed consent was obtained from all participants prior to inclusion in the study, in accordance with the Declaration of Helsinki. Formalin-fixed, paraffin-embedded melanoma tissues (n = 34) were stained using anti–Ki-67 antibody (1:300, ab15580, Abcam). Nuclear staining was quantified in five randomly selected high-power fields (400×), and the Ki-67 labeling index was defined as the percentage of positively stained tumor nuclei among total tumor cells. Cases with Ki-67 labeling index > 10% were designated as ‘Ki-67 high,’ while those with ≤ 10% were designated as ‘Ki-67 low”. This cutoff was used for sample stratification in Figs. 1H, 4N–O and 5H.” Tumor stage distribution included stage II (n = 21), stage III (n = 10), and stage IV (n = 3). TYRP1 expression and survival data from the TCGA-SKCM cohort (n = 567) were obtained from the Sanger database. Kaplan–Meier survival analysis was conducted using the Sanger tool. The expression patterns of TYRP1, GPNMB, SOX10, and MITF in 121 melanoma patients was downloaded from the cBioPortal database. Fig. 1Identification of TYRP1-positive melanoma cell subtype as a predictor of poor prognosis in melanoma. (A) t-SNE plot showing nine distinct cell clusters identified by scRNA-seq analysis of melanoma tissues, including tumor cells, osteoprogenitor cells, plasma cells, B cells, monocytes/macrophages, RPE cells, NK cells, endothelial cells, and photoreceptor cells. (B) t-SNE plot showing 15 melanoma tumor cell subtypes identified by scRNA-seq, including RPS12_Tumor, TYRP1_Tumor, NEAT1_Tumor, FOSB_Tumor, IGLV3−1_Tumor, EEF1B2_Tumor, UCHL1_Tumor, TMSB4X_Tumor, TYMS_Tumor, CD74_Tumor, GDF15_Tumor, IFITM1_Tumor, RPS4Y1_Tumor, MGP_Tumor, and IGFBP5_Tumor. (C) Heatmap showing the expression of five marker genes across the 15 melanoma subtypes. (D) Kaplan–Meier survival curves of 567 melanoma patients from the TCGA database, stratified by high and low expression of the 15 subtype marker genes. (E) Monocle2 analysis depicting the developmental trajectory of tumor cell subtypes. (F) Transcriptomic expression of TYRP1 in para-carcinoma tissues (n = 102) and melanoma tissues (n = 558), and in para-carcinoma (n = 337) versus tumor tissues of all cancer types (n = 132) from TCGA. (G) Melanoma patients were classified into Ki67-negative and Ki67-positive groups based on IHC staining (Ki67-positive cells > 10%), and into SKCM-P and SKCM-M based on imaging. Diagnoses were validated at a 3-year follow-up. A schematic is shown. (H) IHC staining of TYRP1 in melanoma tissues from SKCM-P (n = 15), SKCM-M (n = 10), Ki67-negative (n = 15), and Ki67-positive (n = 20) groups. The representative images of TYRP1 staining (TYRP1 high and low) in melanoma patients were shown Identification of TYRP1-positive melanoma cell subtype as a predictor of poor prognosis in melanoma. (A) t-SNE plot showing nine distinct cell clusters identified by scRNA-seq analysis of melanoma tissues, including tumor cells, osteoprogenitor cells, plasma cells, B cells, monocytes/macrophages, RPE cells, NK cells, endothelial cells, and photoreceptor cells. (B) t-SNE plot showing 15 melanoma tumor cell subtypes identified by scRNA-seq, including RPS12_Tumor, TYRP1_Tumor, NEAT1_Tumor, FOSB_Tumor, IGLV3−1_Tumor, EEF1B2_Tumor, UCHL1_Tumor, TMSB4X_Tumor, TYMS_Tumor, CD74_Tumor, GDF15_Tumor, IFITM1_Tumor, RPS4Y1_Tumor, MGP_Tumor, and IGFBP5_Tumor. (C) Heatmap showing the expression of five marker genes across the 15 melanoma subtypes. (D) Kaplan–Meier survival curves of 567 melanoma patients from the TCGA database, stratified by high and low expression of the 15 subtype marker genes. (E) Monocle2 analysis depicting the developmental trajectory of tumor cell subtypes. (F) Transcriptomic expression of TYRP1 in para-carcinoma tissues (n = 102) and melanoma tissues (n = 558), and in para-carcinoma (n = 337) versus tumor tissues of all cancer types (n = 132) from TCGA. (G) Melanoma patients were classified into Ki67-negative and Ki67-positive groups based on IHC staining (Ki67-positive cells > 10%), and into SKCM-P and SKCM-M based on imaging. Diagnoses were validated at a 3-year follow-up. A schematic is shown. (H) IHC staining of TYRP1 in melanoma tissues from SKCM-P (n = 15), SKCM-M (n = 10), Ki67-negative (n = 15), and Ki67-positive (n = 20) groups. The representative images of TYRP1 staining (TYRP1 high and low) in melanoma patients were shown Fig. 2TYRP1-positive tumor cells exhibit enhanced proliferative capacity. (A) Patient-derived melanoma organoids were established, and TYRP1-high and TYRP1-low organoids were identified by immunofluorescence (IF). Representative IF images are shown. (B) Colony formation rate and colony growth comparing the growth of TYRP1-high and TYRP1-low organoids on day 20. The representative images of colonies were shown on day 10. (C) Heatmap of stemness- or melanoma-associated gene expression (SOX10, MITF, SOX2, Oct4, ABCB5, Wnt5A, ALDH1) in TYRP1-high and TYRP1-low organoids. (D) Western blot of TYRP1 in vector control and TYRP1-overexpressing A375 and SK-MEL-28 cells. (E) Cell proliferation assay of vector and TYRP1-overexpressing A375 and SK-MEL-28 cells. (F) Cell migration assay of vector and TYRP1-overexpressing A375 and SK-MEL-28 cells. (G) t-SNE plot identifying nine cell clusters from scRNA-seq data of primary and metastatic melanoma. Frequencies of cell types are shown for each condition. (H) Colony growth assay in vector and TYRP1-overexpressing A375 and SK-MEL-28 cells. (I) Heatmap of stemness-related genes in vector and TYRP1-overexpressing cells. (J) TYRP1-high and TYRP1-low organoids were isolated from 3D Matrigel and implanted into NOD-SCID mice. Tumor volume was monitored until day 25. The tumor cells were isolated on day 25 for Ki-67 positive cell percentage analysis. A schematic is shown TYRP1-positive tumor cells exhibit enhanced proliferative capacity. (A) Patient-derived melanoma organoids were established, and TYRP1-high and TYRP1-low organoids were identified by immunofluorescence (IF). Representative IF images are shown. (B) Colony formation rate and colony growth comparing the growth of TYRP1-high and TYRP1-low organoids on day 20. The representative images of colonies were shown on day 10. (C) Heatmap of stemness- or melanoma-associated gene expression (SOX10, MITF, SOX2, Oct4, ABCB5, Wnt5A, ALDH1) in TYRP1-high and TYRP1-low organoids. (D) Western blot of TYRP1 in vector control and TYRP1-overexpressing A375 and SK-MEL-28 cells. (E) Cell proliferation assay of vector and TYRP1-overexpressing A375 and SK-MEL-28 cells. (F) Cell migration assay of vector and TYRP1-overexpressing A375 and SK-MEL-28 cells. (G) t-SNE plot identifying nine cell clusters from scRNA-seq data of primary and metastatic melanoma. Frequencies of cell types are shown for each condition. (H) Colony growth assay in vector and TYRP1-overexpressing A375 and SK-MEL-28 cells. (I) Heatmap of stemness-related genes in vector and TYRP1-overexpressing cells. (J) TYRP1-high and TYRP1-low organoids were isolated from 3D Matrigel and implanted into NOD-SCID mice. Tumor volume was monitored until day 25. The tumor cells were isolated on day 25 for Ki-67 positive cell percentage analysis. A schematic is shown Fig. 3TYRP1 promotes melanoma malignancy via GPNMB-mediated autocrine signaling. (A) Interaction network illustrating the weight of communication among the nine tumor cell subpopulations. (B) Volcano plot showing differentially expressed genes in TYRP1_Tumor cells versus other subtypes. (C) Heatmap of GPNMB, SEPPINF1, SNAI2, and PAX3 mRNA expression in TYRP1-high and TYRP1-low organoids. (D) Colony growth of TYRP1-low organoids, A375, and SK-MEL-28 cells after treatment with PBS, GPNMB (100 ng/mL), or SNAI2 (100 ng/mL). (E) IF staining of GPNMB in TYRP1-high and TYRP1-low organoids. (F) Western blot of GPNMB in vector and TYRP1-overexpressing A375 and SK-MEL-28 cells. (G) GPNMB mRNA expression in A375 and SK-MEL-28 cells transfected with scramble or GPNMB siRNA. (H) Cell proliferation of vector and TYRP1-overexpressing cells treated with scramble or GPNMB siRNAs. (I) Colony growth of TYRP1-overexpressing cells with scramble or GPNMB siRNA treatment. (J) Heatmap of melanoma/stemness-related genes in these cells treated with scramble or GPNMB siRNA TYRP1 promotes melanoma malignancy via GPNMB-mediated autocrine signaling. (A) Interaction network illustrating the weight of communication among the nine tumor cell subpopulations. (B) Volcano plot showing differentially expressed genes in TYRP1_Tumor cells versus other subtypes. (C) Heatmap of GPNMB, SEPPINF1, SNAI2, and PAX3 mRNA expression in TYRP1-high and TYRP1-low organoids. (D) Colony growth of TYRP1-low organoids, A375, and SK-MEL-28 cells after treatment with PBS, GPNMB (100 ng/mL), or SNAI2 (100 ng/mL). (E) IF staining of GPNMB in TYRP1-high and TYRP1-low organoids. (F) Western blot of GPNMB in vector and TYRP1-overexpressing A375 and SK-MEL-28 cells. (G) GPNMB mRNA expression in A375 and SK-MEL-28 cells transfected with scramble or GPNMB siRNA. (H) Cell proliferation of vector and TYRP1-overexpressing cells treated with scramble or GPNMB siRNAs. (I) Colony growth of TYRP1-overexpressing cells with scramble or GPNMB siRNA treatment. (J) Heatmap of melanoma/stemness-related genes in these cells treated with scramble or GPNMB siRNA Cell proliferation and migration assays Proliferation was assessed using the Cell Counting Kit-8 (Dojindo, Japan). Cells were seeded at 2,000 cells/well in 96-well plates and incubated for 48 h. Absorbance at 450 nm was measured. Migration was assessed using Transwell inserts with 8 μm pores (Corning, USA); cells migrating for 24 h were fixed in 4% paraformaldehyde and stained with 0.5% crystal violet. 3D matrigel colony formation assay Tumor cells (1,000/well) were suspended in 200 µL growth factor-reduced Matrigel (Corning, USA) at 5 mg/kg and seeded into 24-well plates. Colonies were imaged on day 10 using a Nikon Eclipse Ts2 microscope (Japan) and counted using ImageJ 5.0 software (colony diameter ≥ 100 μm). Cell apoptosis assay Apoptosis was analyzed using the Annexin V-FITC/PI Apoptosis Detection Kit (BD Biosciences, Cat#556547, USA). Cells were treated with Dabrafenib (5 nM) or TMZ (200 µM) for 72 h, stained, and analyzed by flow cytometry (BD FACSCanto II, USA). Quantitative real-time PCR Total RNA was extracted using TRIzol (Invitrogen, USA), and cDNA was synthesized using PrimeScript RT Master Mix (Takara, Japan). qPCR was performed using TB Green Premix Ex Taq II (Takara, Japan) on a StepOnePlus system (Applied Biosystems, USA). Expression levels were normalized to GAPDH using the 2^−ΔΔCt method. Primers were designed using PrimerBank. siRNA interference Cells at 60% confluency in 6-well plates were transfected with siRNAs targeting GPNMB, SOX10, or MITF (GenePharma, China) or scrambled control using Lipofectamine 3000 (Invitrogen, USA) in Opti-MEM (Gibco, USA). Final siRNA concentration was 100 nM. Cells were harvested after 48 h for analysis. Sequences were: GPNMB siRNA-1: 5′-GGGCAAUGAAAGACCUUCUTT-3′GPNMB siRNA-2: 5′-GUGGGCUCAAAUAUAACAUTT-3′SOX10 siRNA-1: 5′-GCCACGAGGUA AUGUCCAA-3′SOX10 siRNA-2: 5′-GAGCUACAGUCUUGGAGAA-3′MITF siRNA-1: 5′-GGAAGAAUAUGGAGAGCUA-3′MITF siRNA-2: 5′-CAGAAUGCUUUGAAGACAA-3′ GPNMB siRNA-1: 5′-GGGCAAUGAAAGACCUUCUTT-3′ GPNMB siRNA-1: 5′-GGGCAAUGAAAGACCUUCUTT-3′ GPNMB siRNA-2: 5′-GUGGGCUCAAAUAUAACAUTT-3′ GPNMB siRNA-2: 5′-GUGGGCUCAAAUAUAACAUTT-3′ SOX10 siRNA-1: 5′-GCCACGAGGUA AUGUCCAA-3′ SOX10 siRNA-1: 5′-GCCACGAGGUA AUGUCCAA-3′ SOX10 siRNA-2: 5′-GAGCUACAGUCUUGGAGAA-3′ SOX10 siRNA-2: 5′-GAGCUACAGUCUUGGAGAA-3′ MITF siRNA-1: 5′-GGAAGAAUAUGGAGAGCUA-3′ MITF siRNA-1: 5′-GGAAGAAUAUGGAGAGCUA-3′ MITF siRNA-2: 5′-CAGAAUGCUUUGAAGACAA-3′ MITF siRNA-2: 5′-CAGAAUGCUUUGAAGACAA-3′ TYRP1 and MITF overexpression Full-length TYRP1 (NM_000550) and MITF (NM_198159) cDNAs were cloned into the lentiviral vector pLV[Exp]-CMV> Insert-IRES-Puro (CNSyneros, China). Lentiviruses were packaged in HEK293T cells using psPAX2 and pMD2.G (Addgene, China) and Lipofectamine 3000 (Invitrogen, USA). Viral supernatants were concentrated using PEG-it (System Biosciences, USA). A375 and SK-MEL-28 cells were infected at MOI = 20 with 8 µg/mL polybrene (Sigma, USA) and selected with 2 µg/mL puromycin (Sigma, USA) for 5–7 days. Overexpression was confirmed via Western blot. Cells infected with empty vector served as controls. Western blotting Proteins were extracted using RIPA buffer (Beyotime, Cat#P0013C, China) with protease/phosphatase inhibitors (Roche, Switzerland). 20 µg protein was separated by SDS-PAGE and transferred to PVDF membranes (Millipore, USA). Primary antibodies: TYRP1 (1:1000, PA5-81909, Thermo Fisher, USA), MITF (1:1000, MA5-32554, Thermo Fisher, USA), SOX10 (1:1000, ab227680, Abcam, UK), GPNMB (1:1000, ab227695, Abcam, UK), Notch1 (1:1000, 14-5785-81, Thermo Fisher, USA). HRP-conjugated secondary antibodies were used at 1:5,000 and detected with ECL (Millipore, USA). Immunohistochemistry and immunofluorescence Paraffin sections were deparaffinized, rehydrated, and underwent antigen retrieval in citrate buffer (pH 6.0). Endogenous peroxidase was blocked with 3% H₂O₂. Primary antibodies were applied overnight at 4 °C. IHC staining was visualized using DAB (Dako, Japan) and hematoxylin. For IF, Alexa Fluor 488/594-conjugated secondary antibodies (Invitrogen, USA) were used. Slides were mounted with DAPI-antifade (VectorLabs, USA) and imaged using a Leica TCS SP8 confocal microscope. Primary antibodies included anti- TYRP1 (1:200, PA5-81909, Thermo Fisher, USA), MITF (1:300, MA5-32554, Thermo Fisher, USA), SOX10 (1:500, ab227680, Abcam, UK), GPNMB (1:200, ab227695, Abcam, UK), Notch1 (1:200, 14-5785-81, Thermo Fisher, USA). and p-STAT1 (1:300, ab314033, Abcam, UK). The quantification of protein expression intensity was determined by Image-Pro Plus 6.0 (Media Cybernetics, USA). T-test was used for comparisons. ELISA IFN-γ levels in tumor lysates were measured using a human IFN-γ ELISA kit (R&D Systems, USA) per manufacturer’s protocol. Absorbance was read at 450 nm using a BioTek ELx800 reader (USA). Animal protocol All procedures were approved by the Institutional Animal Care and Use Committee of Xiangya Hospital. Female NOD-SCID and C57BL/6 mice (6–8 weeks; Beijing Vital River) were used (SLBH-202506100001). Animals were excluded if they exhibited severe illness or weight loss > 20% unrelated to treatment. Exclusion criteria were pre-established before the experiments. Mice were randomly assigned to control or treatment groups using a random number generator. Investigators were blinded to group allocation during treatment and outcome assessment. A total of 5 × 105 cells B16 were injected subcutaneously into C57BL/6 mice. Mice with Pembrolizumab (10 mg/kg, intraperitoneal, twice a week, 2 weeks treatment) or PBS (N = 5 in each group). A total of 2 × 10⁶ A375 cells or 50 µL Matrigel/organoid suspension (2 × 10⁶ cells) were injected subcutaneously into NOD-SCID mice. Mice were treated with Dabrafenib: 50 mg/kg/day, oral gavage for 14 days or PBS (N = 5 in each group). The sample size was determined based on previous studies and preliminary data to achieve 80% power at a significance level of 0.05 for detecting a 30% difference in tumor volume between groups. Tumor volumes were measured every 2–3 days and calculated as: Volume = (Length × Width²)/2. On day 25, tumors were harvested; no other organs were collected. After tumor collection, tumor cell apoptosis analysis was determined by the Annexin V-FITC/PI staining. PD-L1 mRNA level was determined by PCR. p-STAT1 (1:300, ab314033, Abcam, UK) immunostaining, and IFN-γ ELISA were performed. Statistical analysis All experiments were performed in triplicate unless otherwise stated. Data are presented as mean ± SD. Data distribution was tested for normality using the Shapiro–Wilk test. For normally distributed data, Student’s t-test or one-way ANOVA was used with Tukey’s post hoc test. Homogeneity of variance among groups was tested using Levene’s test. All groups met the assumption of equal variance, allowing for the use of Student’s t-test or one-way ANOVA. Kaplan–Meier survival curves were analyzed using the log-rank test. A p-value < 0.05 was considered statistically significant. Analyses were performed using GraphPad Prism 9.0 (GraphPad Software, USA). Results Identification of TYRP1-positive melanoma cell subpopulations and association with poor prognosis To elucidate the relationship between cellular heterogeneity and malignant progression in melanoma, we analyzed the single-cell RNA sequencing (scRNA-seq) dataset GSE139829 from the NCBI GEO database, encompassing eight primary and three metastatic melanoma samples. Through t-distributed stochastic neighbor embedding (t-SNE) analysis, we identified nine distinct cell types within the tumor tissues: tumor cells, osteoprogenitor cells, plasma cells, B cells, monocytes/macrophages, retinal pigment epithelial (RPE) cells, natural killer (NK) cells, endothelial cells, and photoreceptor cells in those 11 samples (Fig. 1A). Focusing on tumor cells, unsupervised clustering delineated 15 transcriptionally distinct tumor cell subtypes: RPS12_Tumor, TYRP1_Tumor, NEAT1_Tumor, FOSB_Tumor, IGLV3−1_Tumor, EEF1B2_Tumor, UCHL1_Tumor, TMSB4X_Tumor, TYMS_Tumor, CD74_Tumor, GDF15_Tumor, IFITM1_Tumor, RPS4Y1_Tumor, MGP_Tumor, and IGFBP5_Tumor. A heatmap illustrated the characteristic gene expression patterns of these subtypes (Fig. 1C). Survival analysis using The Cancer Genome Atlas (TCGA) melanoma cohort (n = 567) demonstrated that high expression of TYRP1, a marker gene of the TYRP1_Tumor subtype, was significantly associated with poorer overall survival (Fig. 1D), while other genes, such as FOSB and TMSB4X, show no association or an inverse relationship with unfavorable outcomes. Single-cell trajectory analysis with Monocle2 indicated that the TYRP1-positive subpopulation was not at the terminal state of the differentiation trajectory, and its distribution did not overlap with that of invasive tumor cells (Fig. 1E), suggesting that TYRP1-positive melanoma cells adopt a dedifferentiated phenotype, which appears to be independent of tumor metastatic potential. Multivariate analysis in patients from TCGA database further confirmed TYRP1 as an independent prognostic factor (HR = 1.06, p = 0.004; Figure S1). Further analysis of the TCGA database confirmed that TYRP1 expression was significantly higher in melanoma tissues compared to adjacent normal tissues, while its expression was not broadly upregulated in other tumor types, suggesting a specific role for TYRP1 in melanoma (Fig. 1F). To validate the clinical relevance of TYRP1, we collected paraffin-embedded melanoma tissue samples from patients and categorized them based on metastasis status (metastatic vs. non-metastatic) and Ki-67 expression levels (Ki-67-positive vs. Ki-67-negative) (Fig. 1G). Immunohistochemical analysis revealed that TYRP1 expression was elevated in the high Ki-67 expression group (Fig. 1J), but not significantly increased in metastatic melanoma tissues (Fig. 1H), suggesting that TYRP1 may be involved in tumor cell proliferation rather than metastasis. This observation aligns with the single-cell trajectory analysis results (Fig. 1E). TYRP1-positive melanoma cells exhibit enhanced proliferative capacity To further elucidate the functional role of TYRP1 in melanoma progression, we established patient-derived organoid models and stratified them into TYRP1-high and TYRP1-low groups based on TYRP1 expression levels (Fig. 2A). Organoid formation assays demonstrated that TYRP1-high organoids developed larger structures by day 20, although the cloning efficiency did not significantly differ, indicating that high TYRP1 expression confers enhanced proliferative capacity (Fig. 2B). We assessed the expression of key melanoma stemness and malignancy-associated genes, including SOX10, MITF, SOX2, Oct4, ABCB5, Wnt5A, and ALDH1. SOX10 and MITF were notably upregulated in TYRP1-high organoids, while the expression of other genes remained unchanged (Fig. 2C). Given the established association of SOX10 and MITF with melanoma cell proliferation [20], these findings corroborate our previous results, although TYRP1 did not significantly enhance melanoma cell stemness. In vitro, we generated TYRP1-overexpressing A375 and SK-MEL-28 melanoma cell lines, with successful overexpression confirmed by Western blot analysis (Fig. 2D). Cell proliferation and apoptosis assays were conducted by using empty vector–transfected cells to control for potential effects of transfection procedures (Figure S2A). Functional assays revealed that TYRP1 overexpression significantly increased cell proliferation (Fig. 2E) without affecting migratory capacity in vitro and in vivo (Fig. 2F and S2B). Single-cell analysis indicated no significant difference in the distribution of TYRP1-positive tumor cells between metastatic and non-metastatic melanoma tissues (Fig. 2G). Moreover, TYRP1 overexpression enhanced the clonogenic growth of A375 and SK-MEL-28 cells in 3D culture systems (Fig. 2H) and upregulated SOX10 and MITF expression (Fig. 2I). In vivo, mice implanted with TYRP1-high organoids developed significantly larger tumors by week 4 compared to the TYRP1-low group (Fig. 2J), with a higher proportion of Ki-67-positive cells observed in the tumor tissues (Fig. 2K). Collectively, these results indicate that TYRP1 plays a critical role in promoting melanoma cell proliferation. Fig. 4The SOX10/MITF axis drives TYRP1-associated melanoma progression. (A) Top 50 upregulated genes in TYRP1_Tumor cells were subjected to survival analysis, identifying SOX10 and MITF as key regulators. (B) IF staining of SOX10 and MITF in TYRP1-high and TYRP1-low organoids. (C) Western blotting of SOX10 and MITF in vector and TYRP1-overexpressing cells treated with scramble or GPNMB siRNAs. (D–E) SOX10 (D) and MITF (E) mRNA levels following siRNA-mediated knockdown in A375 and SK-MEL-28 cells. (F) MITF protein level in TYRP1-overexpressing cells treated with scramble or SOX10 siRNAs. (G–H) Proliferation of TYRP1-overexpressing cells treated with SOX10 or MITF siRNAs. (I-J) Colony growth assay in TYRP1-overexpressing cells after SOX10 or MITF knockdown. (K) Western blotting of TYRP1 in vector and MITF overexpressed A375 and SK-MEL-28 cells. (L-M) Kaplan–Meier survival analysis of 567 melanoma patients stratified by SOX10 and MITF expression (TCGA). (N-O) IHC staining of SOX10 and MITF in SKCM-P (n=15), SKCM-M (n=10), Ki67-negative (n=15), and Ki67-positive (n=20) melanoma tissues. The representative images of SOX10 and MITF staining (SOX10/MITF high and low) in melanoma patients were shown The SOX10/MITF axis drives TYRP1-associated melanoma progression. (A) Top 50 upregulated genes in TYRP1_Tumor cells were subjected to survival analysis, identifying SOX10 and MITF as key regulators. (B) IF staining of SOX10 and MITF in TYRP1-high and TYRP1-low organoids. (C) Western blotting of SOX10 and MITF in vector and TYRP1-overexpressing cells treated with scramble or GPNMB siRNAs. (D–E) SOX10 (D) and MITF (E) mRNA levels following siRNA-mediated knockdown in A375 and SK-MEL-28 cells. (F) MITF protein level in TYRP1-overexpressing cells treated with scramble or SOX10 siRNAs. (G–H) Proliferation of TYRP1-overexpressing cells treated with SOX10 or MITF siRNAs. (I-J) Colony growth assay in TYRP1-overexpressing cells after SOX10 or MITF knockdown. (K) Western blotting of TYRP1 in vector and MITF overexpressed A375 and SK-MEL-28 cells. (L-M) Kaplan–Meier survival analysis of 567 melanoma patients stratified by SOX10 and MITF expression (TCGA). (N-O) IHC staining of SOX10 and MITF in SKCM-P (n=15), SKCM-M (n=10), Ki67-negative (n=15), and Ki67-positive (n=20) melanoma tissues. The representative images of SOX10 and MITF staining (SOX10/MITF high and low) in melanoma patients were shown TYRP1 promotes melanoma proliferation and malignancy via autocrine GPNMB signaling To explore the mechanisms by which TYRP1 drives phenotypic changes in melanoma cells, we analyzed intercellular communication among different tumor cell subpopulations (Fig. 3A). The TYRP1_Tumor subtype exhibited significant signal output capacity, suggesting that it may regulate the tumor microenvironment or its own function through secreted factors. Differential gene expression analysis between the TYRP1_Tumor subtype and other tumor cell subtypes identified upregulated genes such as GPNMB, SNAI2, SEPPINF1, and PAX3, which are potentially involved in melanoma progression (Fig. 3B). Quantitative PCR analysis revealed that GPNMB and SNAI2 expression levels were elevated in TYRP1-high organoids (Fig. 3C). To determine whether TYRP1-mediated melanoma progression is driven by GPNMB or SNAI2, we treated TYRP1-low organoids with recombinant GPNMB or SNAI2 proteins. GPNMB treatment enhanced organoid clonogenic growth, whereas SNAI2 treatment had no significant effect (Fig. 3D), suggesting that TYRP1-positive tumor cells may promote proliferation through GPNMB secretion. Immunostaining demonstrated higher GPNMB protein expression in TYRP1-high organoids (Fig. 3E), and TYRP1-overexpressing A375 and SK-MEL-28 cells also exhibited increased GPNMB expression (Fig. 3F and S2C). Additionally, GPNMB promoted proliferation in a dose‑dependent manner in A375 and SK‑MEL‑23 cells (Figure S2D). To further validate the role of GPNMB in TYRP1-mediated melanoma progression, we performed siRNA-mediated knockdown of GPNMB in A375 and SK-MEL-28 cells (Fig. 3G). Cell proliferation and apoptosis assays were conducted by using scramble siRNAs–transfected cells to control for potential effects of transfection procedures (Figure S2E). GPNMB silencing significantly inhibited proliferation and clonogenic growth in TYRP1-overexpressing cells, with minimal impact on control cells (Fig. 3H and I). Additionally, GPNMB knockdown reduced SOX10 and MITF expression levels (Fig. 3J). These findings indicate that TYRP1 promotes melanoma proliferation and malignancy through upregulation of GPNMB-mediated autocrine signaling. Fig. 5GPNMB upregulates Notch1 signaling to enhance SOX10 expression. (A) KEGG pathway enrichment of genes upregulated or downregulated in TYRP1_Tumor cells. (B) IF staining of Notch1 in TYRP1-high and TYRP1-low organoids. (C) Western blot of Notch1 in TYRP1-overexpressing cells treated with scramble or GPNMB siRNAs. (D) Western blot of SOX10 and MITF in TYRP1-overexpressing cells treated with PBS or Notch1 inhibitor (Brontictuzumab, 20 nM). (E–F) Proliferation and colony growth of TYRP1-overexpressing cells with or without Notch1 inhibition. (G) Kaplan–Meier survival analysis of melanoma patients stratified by Notch1 expression (TCGA). (H) IHC staining of Notch1 in SKCM-P (n=15), SKCM-M (n=10), Ki67-negative (n=15), and Ki67-positive (n=20) melanoma tissues. The representative images of Notch1 staining (Notch1 high and low) in melanoma patients were shown GPNMB upregulates Notch1 signaling to enhance SOX10 expression. (A) KEGG pathway enrichment of genes upregulated or downregulated in TYRP1_Tumor cells. (B) IF staining of Notch1 in TYRP1-high and TYRP1-low organoids. (C) Western blot of Notch1 in TYRP1-overexpressing cells treated with scramble or GPNMB siRNAs. (D) Western blot of SOX10 and MITF in TYRP1-overexpressing cells treated with PBS or Notch1 inhibitor (Brontictuzumab, 20 nM). (E–F) Proliferation and colony growth of TYRP1-overexpressing cells with or without Notch1 inhibition. (G) Kaplan–Meier survival analysis of melanoma patients stratified by Notch1 expression (TCGA). (H) IHC staining of Notch1 in SKCM-P (n=15), SKCM-M (n=10), Ki67-negative (n=15), and Ki67-positive (n=20) melanoma tissues. The representative images of Notch1 staining (Notch1 high and low) in melanoma patients were shown The SOX10/MITF transcriptional axis drives the maintenance of the TYRP1-positive phenotype We further investigated the core transcription factors involved in TYRP1-mediated tumor progression. Among the top 50 upregulated genes in the TYRP1_Tumor subtype, SOX10 and MITF were identified as transcription factors, which might be associated with poor prognosis in melanoma patients (Fig. 4A). Immunofluorescence and Western blot analyses confirmed that SOX10 and MITF expression levels were significantly elevated in TYRP1-high organoids (Fig. 4B) and TYRP1-overexpressing A375 and SK-MEL-28 cells, but decreased upon GPNMB knockdown (Fig. 4C). To validate the roles of SOX10 and MITF, we employed siRNA to knock down their expression in A375 and SK-MEL-28 cells (Figs. 4D, E and S3A, B). Cell proliferation and apoptosis assays were conducted using scramble siRNAs–transfected cells to control for potential effects of transfection procedures (Figure S3C and D). SOX10 suppression led to decreased MITF expression in TYRP1-overexpressing cells (Fig. 4F), indicating that SOX10 acts upstream of MITF, consistent with previous reports [19, 21]. Knockdown of either SOX10 or MITF inhibited cell proliferation (Fig. 4G and H) and clonogenic growth (Fig. 4I and J). Interestingly, prior studies have shown that MITF can act upstream of TYRP1 to enhance its expression [22]. Consistent with these findings, our study demonstrated that MITF overexpression not only activated downstream pathways but also upregulated TYRP1 expression (Fig. 4K). ChIP‑qPCR confirmed that MITF also directly binds to the TYRP1 promoter in MITF‑overexpressing A375 cells (Figure S3E). Additionally, in TYRP1-high organoids, inhibition of TYRP1 suppressed MITF1 expression (Figure S3F), and inhibition of MITF suppressed TYRP1 expression (Figure S3G). Inhibition of MITF or TYRP1 suppressed the expression of SOX10 and GPNMB (Figure S3H), suggesting a positive feedback loop between MITF and TYRP1. We further analyzed the expression patterns of TYRP1, GPNMB, SOX10, and MITF in 121 melanoma patients using the cBioPortal public database. The results revealed a positive correlation between TYRP1 and GPNMB (Figure S3I), as well as between TYRP1 and MITF (Figure S3J). In addition, SOX10 and MITF showed no potential correlation (Figure S3K). Survival analysis further confirmed that high SOX10 and MITF expression levels are significantly associated with reduced patient survival (Fig. 4L and M). Immunohistochemical analysis supported the upregulation of SOX10 and MITF in tissues with high Ki-67 expression (Fig. 4N). Collectively, these results highlight the central regulatory role of the SOX10/MITF axis in maintaining the TYRP1-positive phenotype. GPNMB promotes SOX10 expression by activating Notch1 signaling To elucidate the mechanism by which GPNMB upregulates SOX10 expression, we conducted KEGG pathway enrichment analysis on the differentially expressed genes within the TYRP1_Tumor subpopulation. This analysis revealed a significant enrichment of the Notch signaling pathway among the upregulated gene sets (Fig. 5A). Consistently, immunofluorescence staining showed elevated Notch1 expression in TYRP1-high organoids (Fig. 5B), which was further corroborated by Western blot analysis in A375 and SK-MEL-28 melanoma cell lines (Fig. 5C). Notably, inhibition of GPNMB abrogated the TYRP1-induced activation of Notch1 signaling, indicating that TYRP1 enhances Notch1 activity in a GPNMB-dependent manner. To investigate the functional consequences of Notch1 activation, TYRP1-overexpressing cells were treated with a Notch1-specific inhibitor Brontictuzumab. Consistently, TYRP1 overexpression increased HES1 mRNA (the downstream signaling of Notch1), which was reversed by Brontictuzumab (Figure S4A). This intervention resulted in a marked reduction in SOX10 and MITF expression levels (Fig. 5D), accompanied by impaired cell proliferation (Fig. 5E) and diminished colony formation ability (Fig. 5F). These findings suggest that TYRP1 promotes melanoma cell proliferation through the activation of the SOX10/MITF transcriptional axis via Notch1 signaling. Prognostic analysis using clinical melanoma samples from TCGA database demonstrated that high Notch1 expression was significantly correlated with poorer overall survival (Fig. 5G). Moreover, immunohistochemical staining revealed elevated Notch1 levels in tumor tissues with high Ki-67 expression (Fig. 5H), further supporting its association with aggressive tumor phenotypes. Collectively, these results indicate that GPNMB contributes to melanoma progression by activating Notch1 signaling, thereby enhancing the transcriptional activity of the SOX10/MITF axis. The MITF–TYRP1–SOX10 positive feedback loop regulates resistance to immunotherapy and targeted therapy To determine whether the MITF–TYRP1–SOX10 positive feedback loop contributes to therapeutic resistance, we treated TYRP1-high and TYRP1-low organoids and cell lines with clinically relevant first-line therapeutic agents, including Dabrafenib and Temozolomide (TMZ). Apoptosis assays revealed that TYRP1-high organoids and TYRP1-overexpressing cells exhibited significantly higher apoptotic rates upon Dabrafenib treatment, indicating increased drug sensitivity (Fig. 6A–B). Additionally, TYRP1-OE cells showed enhanced sensitivity to dabrafenib with a leftward shift in dose-response curves (Figure S4B). In contrast, TYRP1 expression had no significant effect on cellular sensitivity to TMZ (Fig. 6C–D), possibly due to the enhanced epithelial-like phenotype and proliferative potential conferred by the SOX10/MITF axis. We next explored the impact of TYRP1 expression on the efficacy of immune checkpoint blockade therapy. Using a B16-C57 melanoma mouse model treated with pembrolizumab, we observed that tumors derived from TYRP1-overexpressing B16 cells grew more rapidly compared to those in the control group. Furthermore, compared to vector controls, TYRP1-overexpressing tumors displayed reduced responsiveness to pembrolizumab, as evidenced by diminished tumor volume suppression (Fig. 6E) and lower levels of tumor cell apoptosis (Fig. 6F). Immunohistochemistry and Western blot analyses further demonstrated decreased expression of PD-L1, phosphorylated STAT1 (p-STAT1), and interferon-gamma (IFN-γ) in TYRP1-overexpressing tumors (Fig. 6G–I), suggesting a compromised anti-tumor immune response. Given these findings, we hypothesized that targeted therapy with Dabrafenib may be particularly effective in tumors with high TYRP1 expression. To validate this, we established xenograft models using A375 cells stably transduced with either vector control or TYRP1. As anticipated, TYRP1-overexpressing tumors exhibited a more pronounced apoptotic response to Dabrafenib treatment (Fig. 6J–L). In summary, while the MITF–TYRP1–SOX10 positive feedback loop promotes melanoma progression, it concurrently enhances the tumor’s sensitivity to targeted therapy, thereby representing a potential therapeutic vulnerability in melanoma. Fig. 6The MITF–TYRP1–SOX10 feedback loop as a potential therapeutic target in melanoma. (A–B) Apoptosis of TYRP1-high/low organoids and TYRP1-overexpressing/vector cells after Dabrafenib (5 nM, 72 h) or not. (C–D) Apoptosis of organoids and cells treated with TMZ (200 µM, 72 h) or not. (E) Tumor volume of B16-bearing mice overexpressing TYRP1 or vector control treated with Pembrolizumab (10 mg/kg, intraperitoneal, twice a week, 2 weeks treatment) or PBS (N = 5 in each group). (F) Cell apoptosis analysis of tumor cells from E on day 25 (N = 5 in each group). (G-I) PD-L1 mRNA, p-STAT1 immunostaining, and IFN-γ ELISA results from tumor tissues of TYRP1-overexpressing or control B16-bearing mice from E (N = 5 in each group). (J-K) Tumor volume of A375-bearing mice treated with Dabrafenib (50 mg/kg, oral gavage, daily, 2 weeks treatment) or PBS. (L) Cell apoptosis analysis of tumor cells from K-L on day 25 (N = 5 in each group) The MITF–TYRP1–SOX10 feedback loop as a potential therapeutic target in melanoma. (A–B) Apoptosis of TYRP1-high/low organoids and TYRP1-overexpressing/vector cells after Dabrafenib (5 nM, 72 h) or not. (C–D) Apoptosis of organoids and cells treated with TMZ (200 µM, 72 h) or not. (E) Tumor volume of B16-bearing mice overexpressing TYRP1 or vector control treated with Pembrolizumab (10 mg/kg, intraperitoneal, twice a week, 2 weeks treatment) or PBS (N = 5 in each group). (F) Cell apoptosis analysis of tumor cells from E on day 25 (N = 5 in each group). (G-I) PD-L1 mRNA, p-STAT1 immunostaining, and IFN-γ ELISA results from tumor tissues of TYRP1-overexpressing or control B16-bearing mice from E (N = 5 in each group). (J-K) Tumor volume of A375-bearing mice treated with Dabrafenib (50 mg/kg, oral gavage, daily, 2 weeks treatment) or PBS. (L) Cell apoptosis analysis of tumor cells from K-L on day 25 (N = 5 in each group) Discussion In this study, we identified TYRP1 as a marker of a transcriptionally distinct melanoma cell subpopulation characterized by heightened proliferative capacity and poor clinical prognosis. Through integrative analysis of scRNA-seq data, patient-derived organoids, in vitro functional assays, and in vivo tumor models, we demonstrate that TYRP1-positive melanoma cells actively drive tumor growth through a novel autocrine signaling axis. Mechanistically, TYRP1 upregulates GPNMB, which activates Notch1 signaling and subsequently enhances SOX10 and MITF expression. These lineage-determining transcription factors, in turn, reinforce TYRP1 expression, establishing a self-sustaining feedback loop that maintains the proliferative phenotype. Disruption of this GPNMB–Notch1–SOX10/MITF circuit impairs proliferation and clonogenic potential, highlighting its essential role in sustaining melanoma cell fitness (Fig. 7). Fig. 7Proposed working model of TYRP1-mediated melanoma progression. TYRP1 defines a transcriptionally distinct melanoma cell subpopulation and promotes tumor proliferation through a GPNMB–Notch1–SOX10/MITF signaling axis. TYRP1 upregulates the secretion of GPNMB, which activates Notch1 signaling and enhances the transcription of SOX10 and MITF. MITF, in turn, promotes TYRP1 expression, forming a self-reinforcing positive feedback loop. This circuit enhances melanoma cell proliferation but does not induce stemness. Moreover, MITF activation suppresses PD-L1, p-STAT1, and IFN-γ expression, contributing to immune evasion and reduced response to immune checkpoint blockade (e.g., pembrolizumab). However, TYRP1-high tumors remain sensitive to BRAF-targeted therapy such as dabrafenib Proposed working model of TYRP1-mediated melanoma progression. TYRP1 defines a transcriptionally distinct melanoma cell subpopulation and promotes tumor proliferation through a GPNMB–Notch1–SOX10/MITF signaling axis. TYRP1 upregulates the secretion of GPNMB, which activates Notch1 signaling and enhances the transcription of SOX10 and MITF. MITF, in turn, promotes TYRP1 expression, forming a self-reinforcing positive feedback loop. This circuit enhances melanoma cell proliferation but does not induce stemness. Moreover, MITF activation suppresses PD-L1, p-STAT1, and IFN-γ expression, contributing to immune evasion and reduced response to immune checkpoint blockade (e.g., pembrolizumab). However, TYRP1-high tumors remain sensitive to BRAF-targeted therapy such as dabrafenib Our findings reposition TYRP1 from a passive melanocytic marker to an active modulator of oncogenic transcriptional programs in melanoma. Although MITF and SOX10 are well-known regulators of melanocyte lineage specification [23, 24], our data reveal that TYRP1 not only lies downstream of these transcription factors but also contributes to a reciprocal regulatory network that locks melanoma cells into a proliferative state. Importantly, this proliferative state is not accompanied by the acquisition of stem-like features. While TYRP1-high cells displayed robust growth in vitro and in vivo, the expression of established melanoma stemness genes—including SOX2 [25], OCT4 [26], ABCB5 [27], and ALDH1 [28]—remained unchanged. This observation is consistent with previous reports that MITF activation promotes a more epithelial-like, differentiated state that supports rapid cell cycle progression but lacks tumor-initiating potential [29–31], and the pro-growth role of TYRP1 in melanoma [32]. Meanwhile, TYRP1 has been reported to exhibit dual or context-dependent functions in relation to tumor patient prognosis [12]. Our findings provide a possible explanation for this phenomenon: TYRP1 promotes a differentiated and proliferative state in melanoma cells through MITF-mediated regulation, which consequently leads to a partial loss of stemness. This phenotypic shift may underlie the “two functions for a single gene” behavior of TYRP1 observed in melanoma patients. Furthermore, the MITF high phenotype has been associated with suppression of immune-related genes [33–35], including PD-L1, which may explain the reduced expression of PD-L1, p-STAT1, and IFN-γ in TYRP1-overexpressing tumors and their diminished response to immune checkpoint blockade. MITF, acting downstream of Notch1 activation, is known to transcriptionally repress multiple interferon-stimulated genes, including those involved in antigen presentation and cytokine response [34]. This repression attenuates STAT1 phosphorylation and IFNγ-mediated signaling, leading to a concomitant reduction in PD-L1 expression. Consistent with this, TYRP1-overexpressing tumors in our study exhibited decreased levels of p-STAT1, IFNγ, and PD-L1, indicating that the MITF-driven proliferative state is coupled with an “immune-cold” microenvironment. These findings align with previous reports showing that MITF-high melanoma cells display low immunogenicity and limited T-cell infiltration [16, 20]. Therefore, the proposed feedback loop does not directly regulate PD-L1 or IFNγ at the post-translational level, but rather reshapes the transcriptional landscape toward a proliferative yet immune-silent phenotype, explaining the reduced interferon responsiveness observed in TYRP1-driven melanoma cells. From a translational perspective, these findings offer mechanistic insight into why TYRP1-high melanoma tumors exhibit poor response to immunotherapy despite their proliferative advantage. The MITF high transcriptional program, induced by TYRP1 via the GPNMB–Notch1 pathway [36], appears to promote immune evasion by suppressing interferon signaling and antigen presentation pathways. In contrast, this phenotype remains susceptible to BRAF inhibition, likely due to its dependence on MAPK-driven proliferation. Indeed, TYRP1-overexpressing tumors responded more effectively to dabrafenib treatment, suggesting that TYRP1 may serve as a biomarker to guide treatment stratification. Clinically, this has substantial implications: patients with TYRP1-high, MITF-high tumors may benefit from up-front targeted therapy rather than immune checkpoint inhibitors, which are less effective in this phenotypic context. Nevertheless, several limitations should be acknowledged. Although we used both uveal and cutaneous melanoma models, further studies are needed to evaluate the prevalence and stability of the TYRP1 high state across melanoma subtypes and stages. Additionally, the upstream triggers that initiate TYRP1 overexpression remain to be elucidated, and the extent to which TYRP1 affects stromal and immune cell interactions via paracrine mechanisms warrants deeper investigation. Future studies incorporating spatial transcriptomics, epigenetic profiling, and immune deconvolution may provide further insights into the tumor ecosystem shaped by TYRP1-driven programs. In mice model, we recognize that mouse Tyrp1 shares only partial homology with human TYRP1, which may limit the translational relevance of pembrolizumab response data. To address this, future studies will employ humanized mouse models (e.g., NSG mice reconstituted with human CD34⁺ hematopoietic stem cells) bearing patient-derived xenografts (PDXs) from TYRP1-high versus TYRP1-low melanoma patients. Conclusion In conclusion, our study uncovers a previously unrecognized role of TYRP1 as a key regulator of melanoma cell proliferation through the GPNMB–Notch1–SOX10/MITF signaling axis. This transcriptional circuitry induces a MITF high, epithelial-like, yet non-stem-like tumor cell phenotype that fosters tumor growth but compromises immunotherapy responsiveness. These findings provide a rationale for using TYRP1 as a dual prognostic and predictive biomarker and support the development of combination strategies that target proliferative melanoma states while overcoming immunotherapy resistance. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3