Single-cell RNA Sequencing of Melanoma Circulating Tumor Cells

Primary CTC isolation and generation of the CTC single cell RNA sequencing (scRNAseq) dataset was previously published by Hong et al.^{12 (link)}. using the SMART-seq2 protocol^{31 (link)}. Briefly, after microfluidic enrichment, CTCs were identified by size (>10 μm) and lack of CD45 staining, and they were collected by micromanipulation. The identity of melanoma CTCs was then validated by RNA sequencing by their expression of melanoma CTC markers and separation from white blood cells in hierarchical clustering analysis^{12 (link)}. Putative CTCs that clustered with white blood cells, or did not express known melanoma markers, were discarded. Additionally, ‘RNA-SeQC’ analysis was used for further quality control of melanoma CTCs. Specifically, any CTCs which had percent MT > 25%, rRNA rate >10%, exonic rate <10%, exon CV MAD > 2.5, or median exon CV > 2 were excluded from further analysis. 46 CTCs from 15 patients receiving immune checkpoint inhibitors targeting PD1 or both PD1 and CTLA4 were selected for differential sequencing analysis. One patient had previously progressed on anti-PD1 therapy (pembrolizumab) prior to CTC isolation and was receiving anti-CTLA4 therapy (ipilimumab) at the time of CTC isolation. The best overall response after three months was used to separate CTCs by “Complete Response”, “Partial Response”, and “Progressive Disease”. Differential gene expression analysis was performed using ‘DESeq2’ comparing CTCs from patients with “Complete Response” compared to “Progressive Disease” (see Supplementary Table 1 for full differential sequencing results). Statistical comparisons of individual genes were made using Wilcoxon rank sum test for significance on log2 transformation of normalized counts (transcripts per million+1).