Targeting SOX10-deficient cells to reduce the dormant-invasive phenotype state in melanoma

Abstract

Cellular plasticity contributes to intra-tumoral heterogeneity and phenotype switching, which enable adaptation to metastatic microenvironments and resistance to therapies. Mechanisms underlying tumor cell plasticity remain poorly understood. SOX10, a neural crest lineage transcription factor, is heterogeneously expressed in melanomas. Loss of SOX10 reduces proliferation, leads to invasive properties, including the expression of mesenchymal genes and extracellular matrix, and promotes tolerance to BRAF and/or MEK inhibitors. We identify the class of cellular inhibitor of apoptosis protein-1/2 (cIAP1/2) inhibitors as inducing cell death selectively in SOX10-deficient cells. Targeted therapy selects for SOX10 knockout cells underscoring their drug tolerant properties. Combining cIAP1/2 inhibitor with BRAF/MEK inhibitors delays the onset of acquired resistance in melanomas in vivo. These data suggest that SOX10 mediates phenotypic switching in cutaneous melanoma to produce a targeted inhibitor tolerant state that is likely a prelude to the acquisition of resistance. Furthermore, we provide a therapeutic strategy to selectively eliminate SOX10-deficient cells.

Fig. 1. Melanoma patient samples show heterogeneous SOX10 expression.

a tSNE plot of scRNA-seq cells colored by SOX10 expression level from Jerby-Arnon, et al. b As in a except cells colored by treatment. c Melanoma patient samples from Thomas Jefferson Hospital stained for SOX10. Scale bars, 200 μm. d TMA samples stained for SOX10. In each sample, the percentage of cancer cells expressing none, low, medium, or high levels of SOX10 was quantified by a pathologist (Dr. Xu). Samples were sorted based on their H-score (low to high). e Immunofluorescence images of A375, cells stained for SOX10 (green), filamentous actin (red), and DAPI (blue). The experiment was performed independently twice, and representative images are shown. Scale bar, 100 μm. f As in e except MeWo cells were used. g As in e except 1205Lu cells were used.

Fig. 2. SOX10 regulates the transcription of genes involved in invasion, proliferation, and cell metabolism pathways.

a MeWo and A375 SOX10 knockout cell lines were generated as described in Materials and Methods. The same number of parental and SOX10 knockout cells were seeded in six-well plates for each cell line. Cells were lysed and lysates western blotted as indicated. The experiment was repeated independently three times with similar results. b A heatmap showing GSEA normalized enrichment scores (NES) for the hallmark gene sets collection for MeWo and A375 SOX10 knockout cells (guide #2 and #4) compared to parental cells. NES values are displayed for enriched gene sets, using a Benjamini-Hochberg False Discovery Rate (BHFDR) cutoff of 0.05. MeWo gRNA#2 includes combined data collected from clones # 2.1, # 2.2, and # 2.8. For all other samples (MeWo gRNA #4, and A375 gRNA#2 and gRNA#4), shown is the mean from three independent replicates generated for each clone. c Enrichment plots of EMT, Hypoxia, MYC targets-1 and MYC targets-2 comparing MeWo and A375 SOX10 knockout cells (guide #2 and #4) with parental cells. d Enrichment plots of proliferative and invasive gene signature (Verfaille, et al.) for MeWo and A375 CRISPR SOX10 knockout (guide #2 and #4) vs parental cells. BHFDR < 0.001.

Fig. 3. SOX10 regulates the expression of genes associated with a slow-cycling and more invasive phenotype.

a The same number of MeWo parental and MeWo CRISPR SOX10 knockout (clone # 2.1; 2.2; 2.8; 4.11) cells were seeded in six-well plates for each cell line. Cells were lysed and lysates western blotted as indicated. The experiment was repeated independently two times with similar results. b MeWo parental, MeWo #2.1, and MeWo #4.11 cells were plated on coverslips coated with 0.2% gelatin. The next day, cells were treated with ascorbic acid (50 µg/ml) for 6 days. Treatment was renewed every 48 h. At the end of the experiment, cells were permeabilized, fixed, and stained for FN1 and collagen IV. The experiment was performed independently twice, and representative images are shown. Scale bars, 50 μm. c Scratch-wound assay comparing MeWo CRISPR SOX10 knockout (clones # 2.1; 2.2; 2.8; 4.11) cells to parental cells. Shown is the mean ± SD from three independent experiments. p-values were calculated using a two-sided t-test and p-values for significant comparisons are shown. d Spheroid in 3D collagen comparing MeWo CRISPR SOX10 knockout (clones # 2.1; 2.2; 2.8; 4.11) cell lines to parental cells. Shown is the mean ± SD from three independent experiments. Scale bar, 25 μm. p-values were calculated using a two-sided one-sample t-test of the null hypothesis and p-values for each comparison are shown.

Fig. 4. Loss of SOX10 induces a dormant/quiescent phenotype in vivo.

a Average tumor volume ±SEM for MeWo parental and SOX10 knockout clones #2.1 and #4.11. Number of mice per cohort as indicated in the figure. b Average Luciferase signal (Avg Radiance [p/s/cm²/sr]) ±SEM for MeWo parental and SOX10 knockout clones #2.1 and #4.11. Number of mice per cohort as indicated in the figure. c Luciferase signal in MeWo #2.1 (2 mice) and #4.11 (2 mice) 90 days after injection. Mice injected with parental MeWo cells did not survive 90 days; hence, their exclusion from the 90 days imaging. d IHC comparing p21^Cip1 and Ki67 expression in MeWo parental versus SOX10 knockout (clones #2.1 and #4.11) tumors collected at the end of the experiment (day 35). Shown is the mean ± SD from three independent tumors generated either from parental or SOX10 knockout MeWo cells. Scale bar, 200 μm. p-values were calculate using two-sided t-test. e Representative images showing second-harmonic generation signatures from three independent tumors from MeWo parental or SOX10 knockout xenografts. Images shown correspond to reconstituted monochromatic images pseudocolored according to “intensity heat-maps” of total second-harmonic generation signal. Warmer tones indicate higher second-harmonic generation signals (color tone bar is provided). Scale bars, 50 μm. f Quantitative analysis of polymerized collagen signatures from images in e. Three independent tumors generated either from parental or SOX10 knockout MeWo cells were analyzed. In total, 12 images from parental and 7 images from SOX10 knockout cells tumors were assessed. Values presented are normalized to MeWo parental xenograft. Data are expressed as mean ± SD. p = 0.025. p-values were calculated using the Wilcoxon two-sample test.

Fig. 5. SOX10 loss induces tolerance to MAPK targeting agents.

a Cells were treated with 50 or 100 nM trametinib. Treatment was renewed three times per week. Shown is the mean ± SD from three independent experiments. p-values were calculated using two-sided model-based t-test tests and adjusted for multiple testing using the Hochberg method and represent statistical analysis comparing growth inhibition of MeWo parental vs MeWo #2.1 and MeWo #4.11 from day 0 to day 10. p-values are shown. b mCherry-MeWo and GFP-MeWo SOX10 knockout #2.1 or #4.11 cells were co-mixed at the ratio of 2:1. Cells were co-cultured for 90 days in the presence or absence of 50 nM trametinib. Co-cultures were collected and analyzed by FACS for mCherry and GFP positivity. c Cells were treated with 1 µM PLX4720 plus 35 nM PD325901 for 24 h. Cells were lysed and western blotted as indicated. The experiment was repeated independently three times with similar results. d Cells were treated with BRAFi (500 nM PLX8394). Cells were lysed and western blotted as indicated. The experiment was repeated independently three times with similar results. e Venn diagrams showing the commonality of enriched gene sets (BHFDR < 0.05) from the Hallmark gene set collection (n = 50) for comparisons between A375 gSOX10 or CRTs in the presence of BRAFi+MEKi and parental control groups. f A scatter plot showing differential gene expression for gSOX10 and CRT groups over parental A375. Wald’s test statistic was averaged across the gSOX10 #2 and gSOX10 #4, and the CRT34 and CRT35, conditions. Red dots and gene names indicate the top 10 most significantly up and downregulated genes for the CRT and gSOX10 groups, and genes that are common between CRT and gSOX10are labeled in red. g Cells were treated with 1 µM PLX4720 + 35 nM PD0325901. Treatment was renewed every 48 h. Cells were stained for FN1 and collagen IV. The experiment was performed independently twice, and representative images are shown. Scale bars, 50 μm. h Enrichment plot of EMT (BHFDR < 0.001) comparing patient samples before and after MAPK targeting treatment from the Sun. et al. dataset. i Proliferative/invasive signature (Verfaille, et al.) for patients post vs pretreatment, controlled for patient (Sun, et al.).

Fig. 6. Synthetic lethality of IAP inhibitors towards SOX10-deficient population.

a Synthetic lethality of IAP1/2-XIAP inhibitors toward the SOX10-deficient population expressed as inhibition of cell viability. b The same number of cells were seeded for each cell line. Cells were treated with increasing concentrations of birinapant. Treatment was renewed every 48 h. Shown is the mean ± SD from three independent experiments. p-values were calculated using two-sided model-based t-test tests and adjusted for multiple testing using the Hochberg method and represent statistical analysis comparing birinapant-induced growth inhibition in MeWo parental vs MeWo #2.1 and MeWo #4.11. p-values are shown. c Cell lysates were western blotted, as indicated. The experiment was repeated independently three times with similar results. d scRNA-seq cell line data for SOX10 and cIAP2/BIRC3. p < 0.001 for zero-inflation model and p = 0.002 for mean model. e Cells were treated with PLX4720 (1 µM) + PD0325901 (35 nM) for 24 h, lysed and western blotted. The experiment was repeated independently three times with similar results. f The same number of cells were seeded for each cell line. Cells were treated with increasing concentrations of birinapant. Treatment was renewed every 48 h. Shown is the mean ± SD from three independent experiments. p-values were calculated and adjusted as in b. and represent statistical analysis comparing birinapant-induced growth inhibition in A375 parental vs CRT35 or CRT34 cells. p-values are shown. g A375 xenograft tumor growth, day 0 corresponding to the first day of treatment. Mice were treated 200 PPM PLX4720 plus 7 PPM PLX2695 and/or injected with 100 μL (for female mice) or 150 μL (for male mice) birinapant solution (3 mg/ml). AIN-76A diet was used as vehicle. Number of mice per cohort as indicated. Statistical significance was calculated as the time to tumor regrowth (tumor volume > 100 mm³) and corresponding median survival times were estimated using the Kaplan–Meier method. The two-sided log-rank test was used to compare the time to regrowth between treatment groups. p-value is shown. h Mouse survival curve for the in vivo experiment shown in g. p-values were calculated using the two-sided log-rank tests and were adjusted for multiple testing to control for the False Discovery Rate (FDR) using the method of Benjamini and Hochberg. p-values are shown.

Fig. 7. Schematic showing the effects of SOX10 loss in melanoma.

SOX10 loss leads to an invasive but dormant-like/quiescent phenotype associated with ECM remodeling and tolerance to MAPK targeting agents. SOX10-deficient cells show upregulation of cIAP2 that is associated with sensitivity to cIAP1/2 inhibitors.