Biopsy RNA-seq captures TROP-2-linked migration and clonal resistance to forecast aggressiveness in metastatic melanoma

Abstract

Background: Unlike many other solid tumors, melanoma cells possess a remarkable ability to dynamically transition between distinct transcriptional states in response to environmental cues or therapeutic pressure. Among the adaptive mechanisms underlying this plasticity, the epithelial-to-mesenchymal transition (EMT) has garnered increasing attention, as it facilitates the shift from a proliferative, epithelial-like phenotype to a more invasive, mesenchymal-like phenotype frequently associated with cellular dormancy, quiescence and resistance to therapy. Despite growing interest in this phenomenon, the characterization of dormant cellular phenotypes and their clinical significance remains incomplete.

Methods: In this study, we adopted a comprehensive approach integrating patient-derived melanoma cell lines, bulk RNA sequencing from tumor biopsies and analysis of independent bulk and single-cell public datasets. This multi-dimensional strategy enabled the identification of a reproducible dichotomy between “proliferative” and “dormant” phenotypes, characterized by distinct levels of mitotic activity and mesenchymal gene expression profiles. By leveraging an eight-gene transcriptional signature, we constructed a “dormancy score” able to stratify tumors along a dormancy–proliferation axis, revealing strong associations with clinical outcomes such as progression-free survival (PFS), overall survival (OS) and response to immunotherapy.

Results: Within the dormant-associated gene module, the surface glycoprotein TACSTD2 (TROP2) emerged as a central hub gene. TROP2 expression was consistently upregulated in the dormant-like transcriptional state. Supporting these findings, single-cell RNA sequencing data confirmed that TROP2 marks a rare subpopulation of malignant cells that may constitute a quiescent, therapy-resistant niche. Besides, results highlight a predominant intracellular expression of TROP2 in the dormant phenotype.

Conclusions: Together, these findings define a robust dormant phenotype in melanoma with both molecular and clinical significance and evaluate TROP2 as a potential biomarker and therapeutic target for identifying and eradicating dormant and treatment-refractory tumor cells.

Supplementary Information: The online version contains supplementary material available at 10.1186/s13046-026-03646-1.

Keywords: Biomarkers; Cell lines; Dormancy; Immunotherapy; Metastatic melanoma; RNA-seq; TROP2.

Fig. 1

Study Design. A Sample Collection Design. Dark purple represents the steps at which NGS analysis was performed, either at DNA or RNA level; light purple represents the steps at which either IHC or PCR were performed. B Overview of tissue derived samples and corresponding cell lines

Fig. 2

Phenotypic and Morphological Characterization of dormant and proliferative Phenotypes. A-B Phenotypic and morphological characterization of dormant and proliferent samples cells via immunofluorescence and corresponding Phalloidin quantification (C-D) Proliferation characterization of dormant and proliferent samples cells via immunofluorescence and corresponding nuclear Ki67 quantification

Fig. 3

Analytical Workflow for Signature Identification. A Analytical design on the internal discovery cohort and validation on internal and external cohort. In light blue boxes are represented the main outcomes: phenotype description, transcriptomic signature and prognostic biomarkers. B PCA based on individual samples expression levels of DE genes; cutoff selection strategy for gene-signature computation; derived signature distributions per group; AUC obtained using a zero-threshold based classification into predicted dormants or proliferants (C) Gene module positively associated with the dormant phenotype in biopsies-derived samples. Nodes represent genes, nodes size represent central degree; nodes colour represents which cluster each node belongs to. D Functional enrichment of genes in the module represented in (C) (Hallmark DB 2021). E Signature heatmap. Individual gene expression level and signature scores from − 1 to 1 with color-coded classification (dot colour) and ground truth annotation (top bar); right boxplot annotation indicates gene-level expression variability per group

Fig. 4

Prognostic Signature. A Progression Free Survival in DFCI 2019 cohort for patients stratified according to the proposed gene-expression signature (B) Overall Survival in DFCI 2019 cohort for patients stratified according to the proposed gene-expression signature

Fig. 5

Single cell expression in dormancy-associated genes. A Dataset SCP109 expression of TACSTD2, EREG, SOX9 and TMEM27 (CLTRN). Color indicates normalized TPM expression; size indicates % of cells expressing a specific gene within a given sample. B Dataset SCP1493 expression of TACSTD2, EREG, SOX9 and TMEM27 (CLTRN). Color indicates normalized TPM expression; size indicates % of cells expressing a specific gene within a given sample

Fig. 6

Digital IHC and Immunofluorescence Images of Dormant and Proliferant Samples. A Digital 40× whole-slide scans of four FFPE sections from dormant samples stained with hematoxylin–eosin (H&E), as well as immunostained for the melanoma markers S100, TROP2 and SOX9. B Digital 40× whole-slide scans of four FFPE sections from proliferant samples stained with H&E and immunostained for S100, TROP2 and SOX9

Fig. 7

Cellular localization of TROP2. A Western blot of 1 dormant and 1 proliferant samples for intracellular and membrane TROP2 (B) FACS on intracellular and membrane proportions in a dormant sample (C) FACS on intracellular and membrane proportions in a proliferant sample