Hao Fu, Mengyao Wang, Haibo Zhu, Weihua Li, Xiaopei Shen, Haidan Yan, Jun He
Resistance to BRAF inhibitors (BRAFi), alone or with MEK inhibitors (MEKi), limits durable responses in BRAF-mutant melanoma. To characterize resistance-associated cell-state evolution, we analyzed 674 melanoma cells from six mice bearing tumors from a single patient-derived BRAF V600E-mutant melanoma xenograft (PDX) lineage before treatment, during initial regression, at minimal residual disease, and at resistant regrowth. Unsupervised clustering based on a BRAF-centered network feature set comprising 2506 candidate genes identified six transcriptional states, which were characterized using transcriptomic analyses. Cluster 4 was detected exclusively at resistant regrowth, a phase marked by MAPK pathway reactivation, and exhibited enhanced JAK-STAT/interferon signaling and increased STAT1, STAT2, IRF7, IRF9, and RELB regulon activities. Network inference predicted STAT1-BRAF and IRF7-MET regulatory links, suggesting candidate routes to MAPK reactivation through BRAF overexpression and MET-mediated bypass signaling. External analyses partially recapitulated the resistance-associated transcriptional program in independent melanoma cell-line datasets and yielded limited, inconclusive evidence for the predicted STAT1-BRAF association in public perturbation datasets. Cluster 2 represented a pre-existing proliferative state whose signature was associated with shorter progression-free survival in pretreatment clinical cohorts. Together, these findings distinguish a therapy-associated acquired-resistance state from a pre-existing proliferative resistance-associated state and nominate the predicted STAT1-BRAF and IRF7-MET links for functional validation.