Lineage-Restricted Regulation of SCD and Fatty Acid Saturation by MITF Controls Melanoma Phenotypic Plasticity

Abstract

Phenotypic and metabolic heterogeneity within tumors is a major barrier to effective cancer therapy. How metabolism is implicated in specific phenotypes and whether lineage-restricted mechanisms control key metabolic vulnerabilities remain poorly understood. In melanoma, downregulation of the lineage addiction oncogene microphthalmia-associated transcription factor (MITF) is a hallmark of the proliferative-to-invasive phenotype switch, although how MITF promotes proliferation and suppresses invasion is poorly defined. Here, we show that MITF is a lineage-restricted activator of the key lipogenic enzyme stearoyl-CoA desaturase (SCD) and that SCD is required for MITF^High melanoma cell proliferation. By contrast MITF^Low cells are insensitive to SCD inhibition. Significantly, the MITF-SCD axis suppresses metastasis, inflammatory signaling, and an ATF4-mediated feedback loop that maintains de-differentiation. Our results reveal that MITF is a lineage-specific regulator of metabolic reprogramming, whereby fatty acid composition is a driver of melanoma phenotype switching, and highlight that cell phenotype dictates the response to drugs targeting lipid metabolism.

Keywords: ATF4; MITF; fatty acid saturation; melanoma; metastatic dissemination; phenotype switching; stearoyl CoA desaturase.

Figure 1.. SCD is suppressed by glutamine starvation

(A) SILAC mass spectrometry approach to identifying proteins differentially expressed after 4 h or 24 h glutamine (Q) starvation. Light R0K0, Medium R6K4 and Heavy R10K8 media used to label the cells are described in the STAR Methods. See also Figure S1 (B) Expression ratios of 5885 quantified proteins (all SILAC channels) in both replicates for the 4 h or 24 h glutamine starvation vs un-starved condition. MITF (yellow) and SCD (green) expression are highlighted. The blue area represents 90% of the data density. See also Table S1. (C) Differential expression of MITF and SCD proteins after 4 or 24 h glutamine starvation. (D) Western blot of indicated melanoma cell lines grown in MEM with 2 or 0.1 mM glutamine probed with indicated antibodies. ERK or Tubulin were loading controls. (E) Western blot of IGR37 melanoma cells grown in 0.1 mM glutamine probed with indicated antibodies. ERK was a loading control. (F) RT qPCR for indicated genes grown in DMEM or MEM + 2 mM or 0.1 mM glutamine for 24 or 48 h.** p <0.01; *** p <0.001; **** p <0.0001. Error bars represent mean +/− SEM. (G) The 471 melanomas in the TCGA cohort were ranked by SCD expression (black line). Grey bars indicate average expression of the 103 genes comprising the GSS in each tumor. The blue line represents the moving average of the GSS over each 20 tumors. (H) Western blots of cell lines grown in 0.1 mM glutamine probed with ATF4, MITF and SCD antibodies. GAPDH was a loading control. (I) Western blots of IGR37 melanoma cells treated with 50 ng/mL TNFα probed with indicated antibodies. GAPDH was a loading control. See also Figure S2D

Figure 2.. Translational reprogramming and ATF4 repress SCD expression

(A) Western blot of IGR37 or 501mel cells grown in the absence or presence of 50 μM Salubrinal. ERK was a loading control. (B) RT qPCR for indicated genes from IGR37 cells grown with or without 50 μM Salubrinal. * p <0.05; ** p <0.01. Error bars represent mean +/− SEM. (C) Western blot for indicated genes in 501mel or SKmel28 melanoma cells expressing doxycycline-inducible ATF4 (iATF4). Cells were grown +/− 100 ng/ml doxycycline. Tubulin was a loading control. (D) RT qPCR for indicated genes in Skmel28 iATF4 cells grown in the presence of 100 ng/ml doxycycline for indicated times. * p <0.05; ** p <0.01; *** p <0.001. Error bars represent mean +/− SEM. (E) Immunofluorescence of SKmel28-iATF4 or B16-iATF4 cells grown for 24 h +/− 100 ng/ml doxycycline as indicated. SCD (green) and ATF4-expressing cells (red) are indicated. Arrows indicate cells in which ATF4 was not induced and SCD expression is maintained. Scale bar = 20 μm.

Figure 3.. SCD expression is controlled by MITF

(A) The 471 TCGA melanomas were ranked by expression of SCD (black line). Grey bars indicate average expression of MITF in each tumor, and brown line represents the moving average of MITF expression over each 20 tumors. (B) Heatmap of relative expression of indicated genes from triplicate 3’RNA-seq across indicated cell lines. (C) Western blot of indicated melanoma cell lines. Tubulin was a loading control. (D) RT qPCR for indicated genes and western blot for MITF and SCD in melanoma cell lines 72 h after transfection with control or MITF-specific siRNA. ** p <0.01. Error bars represent mean +/− SEM. See also Figure S3B. (E) Biological replicate MITF ChIP-seq showing MITF occupancy at the SCD gene. (F) RT qPCR for ATF4, MITF and SCD mRNA in response to exogenous ATF4 or MITF overexpression. ** p <0.01; *** p <0.001; **** p <0.0001. Error bars represent mean +/− SEM. See also Figure S3C. (G) GS-MS analysis showing ratio of saturated and monounsaturated fatty acids in IGR37 and IGR39 cells. ** p <0.01. Error bars represent mean +/− SEM. See also Figure S3D–F. (H) GS-MS analysis showing ration of saturated to monounsaturated fatty acids in 501mel cells following transfection with control of MITF-specific siRNA. * p<0.05. Error bars are +/− SEM. See also Figure S3D–F.

Figure 4.. Identification of an ATF4-SCD-MITF feedback loop

(A) GS-MS analysis showing total fatty acid content of IGR37 cells +/− treatment with 1 μM A939572 SCD inhibitor for 48 h. * p < 0.05. Error bars represent mean +/− SEM. See also Figure S4A. (B) RT PCR for XBP1 splice isoforms in IGR37 cells treated with SCD specific siRNA or 1μM A939572 SCD inhibitor. H: hybrid from unspliced and spliced XBP1 isoforms; u: unspliced XBP1; s: spliced XBP1. See also Figure S4B–E. (C) Western blot of IGR37 or 501mel cells showing ATF4 and MITF expression after SCD inhibition using 1 μM A939572 or 100 nM CAY10566, or transfection with SCD-specific siRNAs for 48 h. GAPDH or ERK were loading control.s See also Figure S4F. (D) Western blot of IGR37 or 501mel cells showing phospho-eIF2α and eIF2α expression after SCD inhibition using 1 μM A939572, +/− of 100 μM oleic acid (18C:1). GAPDH was a loading control. (E) Coomassie staining or ³⁵S-methionine incorporation into IGR37 or 501mel cells treated with 1 μM A939572 or depleted of SCD using different siRNAs.

Figure 5.. Phenotype-specific dependency on SCD

(A) 471 melanomas in the TCGA cohort were ranked by SCD expression (black line). Grey bars indicate average expression in each tumor of genes comprising the Verfaillie Proliferative signature. Brown line represents the moving average of the Proliferative signature over each 20 tumors. (B) Indicated cell lines were grown for 120 h in a 12 well plate +/− 1 μM A939572 or 100 μM of oleic acid as indicated before crystal violet staining. See also Figure S5A,B. (C) Proliferation of indicated cell lines grown +/− 1 μM A939572 or 100 μM of oleic acid as indicated. Quantification obtained by 0.1% crystal violet staining of three independent experiments. Data presented as mean ± SEM. Statistics show differences between different treatments at each specific time point. * p < 0.05; ** p <0.01 (D) Apoptosis assays using cleaved anti-caspase 3/7 antibody to identify the apoptotic population by flow cytometry 72 h after treatment with A939572 1 μM in the MITF^High melanoma cells IGR37 and 501mel. Data presented as mean ± SEM of 5 biological replicates. **** p-value<0.0001 (E) Western blot for PARP and PARPc of indicated cell lines treated with 1 μM A939572. GAPDH was a loading control. (F) Western blot of IGR37 cells showing MITF and E2F1 expression after treatment with 1 μM A939572, +/− 100 μM oleic acid (18C:1). TUBULIN was a loading control. See also Figure S5C. (G) Western blot for MITF and E2F1 in melanoma cell lines transfected for 72 h with control or MITF-specific siRNAs 72 h. GAPDH was used as loading control. (H) Biological replicate MITF ChIP-seq showing MITF occupancy at the E2F1 gene. (I) GS-MS analysis showing ratios of saturated and monounsaturated fatty acids in IGR37 and IGR39 melanoma cells treated with 1 μM A939572 for 48 h. Data presented as mean ± SEM. ** p <0.01; *** p <0.001 (J) Western blot for LC3 of cells grown in the absence or presence of 1 μM A939572 or 100 nM CAY10566 for 72 h. TUBULIN was a loading control.

Figure 6.. SCD promotes cell proliferation and suppresses inflammatory signaling and invasion

(A) Heatmaps and GSEA analyses derived from 3’ RNA-seq of IGR37 or 501mel cells depleted for SCD using a specific siRNA (siSCD#1) for 48 h. See also Tables S2–S5. See also Figure S6A–C. (B) 3D migration collagen assays performed using 501mel and IGR37 melanoma cells. Results presented as mean ± SEM of 5 or 6 biological replicates. * p < 0.05; ** p <0.01. See also Figure S6D. (C) Western blot for FN1 of 501mel or IGR37 cells treated with 1 μM A939572 or 100 nM CAY10566 for 5 days. GAPDH was a loading control. (D) Western blot using anti-AXL antibody of IGR37 and 501mel cells treated with 1 μM A939572 or 100 nM CAY10566 for 5 days. TUBULIN was used as a loading control. (E) Luciferase assay and western blot of IGR37 or 501mel cells transfected with SCD-specific siRNAs for 48 h together with an NFκB-dependent promoter-luciferase reporter. * p < 0.05; ** p <0.01, *** p <0.001; **** p <0.0001 (F) Western blots showing IκBa levels after SCD inhibition or siRNA-mediated knockdown. (G) RT qPCR for indicated genes from 501mel or IGR37 cells treated with 1 μM A939572 for 48 h or 72 h as indicated. Data represent mean ± SEM. * p < 0.05; ** p <0.01, *** p <0.001; **** p <0.0001. See also Figure S6E, F. (H) IL-8, IL-6, and CXCL1 levels in media of 501mel and IGR37 cells treated with 1 μM A939572 at indicated times. Data represent mean ± SEM of 5 to 8 biological replicates. * p < 0.05; ** p<0.01, *** p <0.001; **** p <0.0001. (I) Western blot for phospho-eIF2α, eIF2α, ATF4 and MITF of IGR37 is 501mel cells treated as indicated with A939572 and transfected with control or p65 (NFκB)-specific siRNA. GAPDH was is a loading control.

Figure 7.. SCD inhibition in vivo induces inflammation dedifferentiation and metastasis

(A) GSEA analysis of B16 cell primary tumor 3’RNA-seq from mice treated either with vehicle or A939572 SCD inhibitor (dose: 25 mg/kg). (B) Gene expression values for the fatty acid importers SLC27A1, SLC27A3, SLC27A4, SLC27A6 and CD36 obtained from the 3’RNA-seq of B16 cells primary tumors from mice treated either with vehicle or A939572 SCD inhibitor. (C) Western blot for phospho-eIF2α, eIF2α, ATF4, MITF and SCD of SW1 mice cells treated in vitro with 1 μM A939572. TUBULIN is a loading control. (D) Schematic showing the SW1 mouse experiments. 3 groups were established (n=10 mice/group): GROUP 1 (treated with vehicle); GROUP 2 (treated for 9 days, twice per day with A939572); GROUP 3 (treated for 16 days, twice per day with A939572). The red arrows indicate the days when A939572 was administrated (IP, Dose: 25 mg/kg). D=day. All samples were collected at D 20. (E) SW1 primary tumor volumes from mice treated with either vehicle, SCD inhibitor during 9 days or SCD inhibitor for 16 days. N=10 mice/group. Error bars represent mean +/− S.E.M. (F) Images of lung sections SW1 primary tumors treated or not with A939572 as indicated time. The table shows the number of metastases observed in numbers of lungs. (G) Western blot for MITF of SW1 primary tumors from mice treated or not with A939572 for the indicated time. GAPDH is a loading control. (H) Schematic showing the mouse experiment using HCmel3 cells treated with cytotoxic T-cells targeting the melanoma differentiation antigen Pmel/Gp100. GROUP 1 (Non treated (NT); n=8); GROUP 2 (early during treatment (EDT); n=13); GROUP 3 (Relapse (R); n=5). Primary tumors were collected and analyzed by RNA-seq. (I) Box plots showing the relative expression of the indicated genes from microarrays of untreated (UT), early during treatment (EDT) or relapsed (R) mouse melanomas after T-cell immunotherapy. Box plot horizontal lines and whiskers indicate quartiles. * p < 0.05; ** p < 0.01; *** p < 0.001, **** p < 0.0001 unpaired two-sided t-test with Benjamini and Hochberg correction. (J) The MITF-SCD axis in melanoma.