Antagonistic cross-regulation between Sox9 and Sox10 controls an anti-tumorigenic program in melanoma

Abstract

Melanoma is the most fatal skin cancer, but the etiology of this devastating disease is still poorly understood. Recently, the transcription factor Sox10 has been shown to promote both melanoma initiation and progression. Reducing SOX10 expression levels in human melanoma cells and in a genetic melanoma mouse model, efficiently abolishes tumorigenesis by inducing cell cycle exit and apoptosis. Here, we show that this anti-tumorigenic effect functionally involves SOX9, a factor related to SOX10 and upregulated in melanoma cells upon loss of SOX10. Unlike SOX10, SOX9 is not required for normal melanocyte stem cell function, the formation of hyperplastic lesions, and melanoma initiation. To the contrary, SOX9 overexpression results in cell cycle arrest, apoptosis, and a gene expression profile shared by melanoma cells with reduced SOX10 expression. Moreover, SOX9 binds to the SOX10 promoter and induces downregulation of SOX10 expression, revealing a feedback loop reinforcing the SOX10 low/SOX9 high ant,m/ii-tumorigenic program. Finally, SOX9 is required in vitro and in vivo for the anti-tumorigenic effect achieved by reducing SOX10 expression. Thus, SOX10 and SOX9 are functionally antagonistic regulators of melanoma development.

Figure 1. Differential expression of SOX10 and SOX9 in human melanocytes, human giant congenital naevi and human melanoma samples.

A, Scheme showing the localization of epidermal melanocytes (in red) in the human skin. B, C, Immunostaining for MITF (green, left panel) and SOX9 (red, right panel) in the human skin demonstrating the lack of SOX9 expression in the epidermal melanocytes. Inserts show higher magnification images of MITF and SOX9 immunostainings. Scale bars, 25 μm. D, Scheme showing the localization of melanocytes (in red) within the hair follicle. E, Immunostaining for MITF (green) and SOX9 (red) in the human skin reveals the expression of SOX9 in the cells of outer root sheath but not in the MITF-positive melanoblasts/melanocytes. Scale bar 100 μm. F, G, High magnification images of immunostaining for MITF and SOX9 in the upper part of human hair follicle (F) and the follicular bulb (G). H, Analysis of SOX9 (red, left panel) and SOX10 (red, right panel) expression in the patients with human giant congenital naevi demonstrates the lack of SOX9 expression in the SOX10-positive giant congenital naevi cells. Inserts show higher magnification. I, Representative examples of immunostaining for SOX9 (green) and SOX10 (red) in a tissue microarray of primary melanoma samples are shown. J-K, Distribution of SOX10 vs. SOX9 expression in human melanoma (based on TCGA database). 334 melanoma patients were divided in two groups, namely SOX10 High/ SOX9 Low and SOX10 Low / SOX9 high based on SOX10 and SOX9 expression levels. DP, dermal papilla; HF, hair follicle; M, melanocytes; ORS, outer root sheath. Scale bars, 25 μm.

Figure 2. Sox9 is not expressed in the mouse melanocytic lineage and is not functionally required for the maintenance of melanocyte stem cells and melanocytes in the postnatal skin.

A, A schematic representation of the experimental strategy used to analyze the expression of Sox9, Sox10, Mitf and Tyr genes in the melanocytic lineage in vivo. B, Results of RNA-seq analysis demonstrating high Sox10 and low Sox9 expression in melanocytic cells at various stages of development. C, A schematic representation of the anatomical location of the melanocyte stem cells, melanoblasts and differentiated melanocytes within the hair follicle in the mouse skin. D, X-Gal staining (blue) combined with Sox9 immunostaining (red) in skin sections of Dct::LacZ mice demonstrating the lack of the Sox9 expression in the melanocyte stem cells located in the bulge region of the hair follicle (upper panels) and in the differentiated melanocytes located in the hair follicular bulb (lower panels). E, Skin sections of Dct::LacZ mice stained for Sox10 (red) in combination with X-Gal staining (blue) reveal the expression of Sox10 in the melanocyte stem cells (upper panels) as well as in the differentiated melanocytes (lower panels). Dashed lines demarcate HFs. Insets show high magnification views. F-I, Experimental strategy used to analyze the effect of the lack of Sox9 (F) and Sox10 (H) expression in the mouse melanocytic lineage. Pictures of two representative mice at 6 months of age lacking Sox9 gene (G) and Sox10 gene (I) demonstrating the effects on hair graying. Bg, bulge; HF, hair follicle; MSCs, melanocyte stem cells; Mo, months; E 15.5, embryonic day 15.5; P0, postnatal day 0; SG, sebaceous gland. Scale bars, 25 μm.

Figure 3. Mouse giant congenital naevi and melanoma reveal no expression of Sox9.

A-D, Immunostaining for Sox9 (A, C) and Sox10 (B, D) in the skin sections of Tyr::Nras^Q61K and Tyr::Nras^Q61KINK4a^−/− mice. E-H, Experimental strategy used to abrogate the expression of Sox9 (E) and Sox10 (G) in the mouse melanocytic lineage. Pictures of two representative mice 1 year after tamoxifen injections reveal no reduction in the skin hyperpigmentation in Tyr::Nras^Q61KSox9^fl/+Tyr-CreERT2 mice as compared to their Tyr::Nras^Q61K littermates (F) in contrast to a pronounced skin whitening observed upon Sox10 loss (H). BF, bright field; HF, hair follicle; mo, months; P0, postnatal day 0. Scale bars, 25 μm.

Figure 4. SOX10 knockdown results in elevated SOX9 expression in mouse and human melanocytes.

A, Experimental design used to investigate the level of SOX9 and SOX10 expression in vitro. Cultured human keratinocytes, melanocytes, cells derived from biopsies of patients with giant congential naevi and melanoma cells (M010817 cell line) were subjected to RNA isolation and subsequent Q-RT-PCR analysis. Keratinocytes were used as a control. B, C, Quantitative real-time PCR analysis showing the decline of SOX9 expression (C) and increase of SOX10 expression (B) that correlate with the acquisition of malignant state by human NRAS^Q61K-mutated cells. Data are presented as the mean fold change and are normalized over levels found in melanocytes. D, E, SOX10 and SOX9 expression in a large set of proliferative and invasive cell lines analysed by gene expression using microarrays (D) and Western blot (E) techniques. F, Experimental design used to deregulate SOX10 expression in human cells derived from the biopsy of a patient with NRAS^Q61K-mutated giant congenital naevus. G, H, Quantitative real-time PCR analysis of SOX10 (G) and SOX9 (H) expression after the knockdown of SOX10. I, Experimental design used to analyze the expression of Sox9 in the melanocytic lineage from Tyr::Nras^Q61K and Tyr::Nras^Q61K Sox10^LacZ/+ mice. K, L, Cells were isolated from the trunk skin of Tyr::Nras^Q61K and Tyr::Nras^Q61K Sox10^LacZ/+ mice and stained for Melan-a and c-Kit antibodies. FACS-sorted cells were subsequently used for the RNA isolation and quantitative real-time PCR with primers specific for the coding regions of Sox9 gene. Data are presented as the mean fold change and are normalized to the control. Kerat, keratinocytes; M, melanocytes; Nev, naevus cells; Mel, melanoma cells; KD, knock down.

Figure 5. Experimental suppression of SOX9 expression rescues the effects of SOX10 deregulation in human melanoma cells.

A, SOX9 overexpression in human melanoma cells closely resembles the gene expression signature of SOX10 knockdown as revealed by unsupervised hierarchical clustering of control M010817 melanoma cells, SOX9 overexpressing M010817 cells and SOX10 knock down M010817 cells. Microarray gene expression accession number: GSE37059. B, Western blot analysis showing that SOX10 expression is downregulated upon overexpression of SOX9 in two independent human melanoma cell lines (A375 and M010817). C, Chromatin immunoprecipitation assay demonstrating the binding of SOX9 to the promoter of SOX10 in human melanoma M010817 cells. D, E, Quantitative real-time PCR analysis of SOX10 (E) and SOX9 (F) expression after the knockdown of SOX10 and after the double knockdown of SOX10 and SOX9 in M010817 cell line. F, Quantification of number of Annexin V-positive cells based on the FACS analysis in the melanoma M010817 cells upon SOX9 KD, SOX10 KD or double SOX9/SOX10 KD. OE, overexpression; KD, knock down; ChIP, chromatin immunoprecipitation; prom, promoter.

Figure 6. Homozygous deletion of Sox9 rescues the effects of Sox10 loss in Tyr::Nras^Q61K mice and restores hyperpigmentation in vivo.

A, B, Representative pictures of back skin, paws and snouts from mice of the indicated genotypes. C, D, Histological evaluation of the hyperpigmentation phenotype in the skin. Haematoxylin and eosin staining of back skin (C) was followed by the quantification (D) of the percentage of hair follicles associated with the hyperpigmentation. HF, hair follicle; H&E, haematoxylin and eosin.