MITF, TFEB, and TFE3 drive distinct adaptive gene expression programs and immune infiltration in melanoma

Abstract

Cells can contain multiple related transcription factors targeting the same sequences, leading to potential regulatory cooperativity, redundancy, competition, or temporally regulated factor exchange. Yet, the differential biological functions of co-targeting transcription factors are poorly understood. In melanoma, three highly related transcription factors are co-expressed: the mammalian target of rapamycin complex 1 (mTORC1)-regulated TFEB and TFE3 (both key effectors of a wide range of metabolic and microenvironmental cues assumed to perform similar functions) and the microphthalmia-associated transcription factor (MITF), which controls melanoma phenotypic identity. Here, we reveal the functional specialization of MITF, TFE3, and TFEB and their impact on melanoma progression. Notably, although all bind the same sequences, each regulates different and frequently opposing gene expression programs to coordinate differentiation, metabolism, and protein synthesis and qualitatively and quantitatively impacts tumor immune infiltration. The results uncover a hierarchical cascade whereby microenvironmental stresses, including glucose limitation, lead MITF, TFEB, and TFE3 to drive distinct biologically important transcription programs that underpin phenotypic transitions in cancer.

Keywords: CP: cancer; CP: genomics; MITF; TFE3; TFEB; melanoma; melanoma gene regulation; tumor immune infiltration.

Figure 1.. Control of mTORC1 by MITF

(A) Left, western blot of IGR37 melanoma cell line transfected with control or MITF-specific small interfering RNA (siRNA). Right, RT-qPCR for PTEN mRNA from corresponding IGR37 cells. (B and C) Heatmaps showing relative mRNA expression in CCLE melanoma or Tsoi et al. melanoma cell lines. (B) Right, Pearson correlation of gene expression. (D) TCGA melanomas ranked by MITF expression (MITF; black line). Gray bars indicate expression of RRAGD or FNIP2 in each melanoma, with the moving average of each per 20 melanoma window indicated by colored lines. (E) MITF ChIP-seq showing binding to the RRAGD or FNIP2 genes. (F) Western blot of melanoma cell lines. (G and H) Box and whisker plots showing relative expression of MITF, RRAGD, and FNIP2 based on triplicate RNA-seq of IGR37 cells expressing doxycycline-inducible β-catenin (iCTNNB1) (G) or in MeWo cells after short hairpin RNA (shRNA)-mediated depletion of MITF (H). See also Figure S1.

Figure 2.. Differential expression of MITF, TFEB, and TFE3 on glucose limitation

(A–C) TCGA melanomas ranked by expression of MITF (A and B) or TFEB (C) (black lines). Gray bars indicate expression of TFE3 (left and right) or TFEB (middle) in each melanoma, with the moving average of each per 20 melanoma windows indicated by colored lines. (D and E) Heatmaps showing relative gene expression in the Tsoi et al. melanoma cell line panel (D) or an in-house panel (E). (F) Western blot of a panel of melanoma cell lines. (G and H) Western blot of 501mel melanoma (G) or HT29 colorectal cancer cells (H) grown in low-glucose medium. (I) Flow cytometry of IGR37 cells transfected with control siRNA or a siTFE3 pool for 48 h before being placed in glucose-free DMEM for 6 h. Error bars indicate SD. n = 3. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns, not significant, Student’s t test.

Figure 3.. MITF family members bind a similar repertoire of genes

(A) Sequences of MITF family member basic regions. (B) Western blot showing expression of doxycycline-inducible HA-tagged TFEB in 501mel cells. (C) Jitter plots showing peak scores of HA-TFEB ChIP-seq in 501mel cells ± 250 nM Torin for 6 h and either 0 or 5 ng doxycycline. Black lines indicate the mean. Boxes indicate the interquartile range and median. (D) Western blot showing expression of doxycycline-inducible HA-tagged TFE3 in 501mel cells. (E and F) Venn diagrams showing the number of statistically significant ChIP peaks or genes bound by MITF family members. HA-TFEB and HA-TFE3 ChIPs were performed in the presence of 250 nM Torin. No doxycycline was used for the HA-TFE3 ChIP, and 5 ng was used for the TFEB ChIP. The MITF data were taken from Louphrasitthiphol et al. (G) Relative peak scores at sites bound by each factor alone or indicated combinations. Boxes indicate interquartile range and median. ****p < 0.001. One-way ANOVA. (H) Motifs identified beneath ChIP peaks for MITF, TFEB, and TFE3. (I) Genome ontology of binding sites identified for each factor. (J) Color-coded ChIP profiles for input control and HA-TFEB, HA-TFE3, and HA-MITF. Only one replicate is shown. Similar results were obtained for replicate 2. See also Figure S2.

Figure 4.. Differential gene expression in TFEB, TFE3, and TFEB/TFE3 KO human melanoma cells

(A) Triplicate RNA-seq showing differential gene expression of control 501mel cells or cells grown in HBSS for 12 h. (B) Volcano plot showing numbers of significantly (FC ≥ 2, p < 0.05) differentially expressed genes comparing control and HBSS-treated cells. (C) GSEA plots showing significantly differentially enriched gene sets comparing control and HBSS-treated 501mel cells. (D) GSEA plots comparing control versus HBSS-treated 501mel cells. (E) Western blot using parental and TFEB and/or TFE3 501mel KO cell lines. (F) Triplicate RNA-seq showing differential gene expression (FC ≥ 2, p < 0.05) of parental or TFEB and/or TFE3 KO 501mel cells. (G) Volcano plots showing numbers of significantly (FC ≥ 2, p < 0.05) differentially expressed genes comparing parental cells and indicated 501mel KOs. (H) GSEA plots comparing parental or indicated TFEB and/or TFE3 501mel KO cells. See also Figure S3 and Table S1.

Figure 5.. TFE3 and TFEB are not equivalent in gene regulation

(A) Western blot using B16-F10 parental and functional MITF KO cell lines. (B) Western blot using parental and indicated B16-F10 KO cell lines. (C) Triplicate RNA-seq showing differential gene expression between parental and indicated B16-F10 KO cell lines. (D–K) GSEA plots showing significantly differentially enriched gene sets comparing parental or indicated MITF, TFEB, and TFE3 B16-F10 KO cell lines. See also Figures S4 and S5 and Table S2.

Figure 6.. Differential regulation of metabolism and differentiation by MITF, TFEB, and TFE3

(A) Seahorse assays measuring ECAR and OCR in B16-F10 WT or KO cell lines. Error bars: SD; n = 3. (B and C) Seahorse assays summarizing differential basal ECAR (B) and OCR (C) between B16-F10 WT or KO cell lines. Error bars: SD; n = 3; ns, non-significant, *p < 0.05, and **p < 0.01 (black, comparison with parental cells; red, comparison with TKO); Student’s t test. (D) ATP production from glycolysis or oxidative phosphorylation by the indicated cell lines. Numbers indicate the percentage derived from each source. Error bars: SD; n = 3; ns, non-significant, *p < 0.05, and **p < 0.01; Student’s t test comparing mutants to B16-F10 WT cells. (E–G) Triplicate RNA-seq showing relative gene expression in parental and B16-F10 KO cell lines of a subset from the HALLMARK OXIDATIVE PHOSPHORYLATION gene set (E) or indicated genes (F) or in 501mel cells in which MITF has been induced using 100 ng doxycycline (G). (H) Western blot of 501mel cells expressing ectopic HA-tagged MITF induced with 100 ng doxycycline for 16 h. (I) Triplicate RNA-seq showing relative gene expression in parental and B16-F10 KO cell lines. See also Figure S6.

Figure 7.. Reduced tumor growth of B16-F10 MITF/TFE KO cells in vivo

(A) Bioluminescent imaging of primary tumors 28 days after subcutaneous injection of luciferase-expressing B16-F10 parental or KO cells (n = 5). Firefly luciferin (120 mg/kg) was injected intraperitoneally. The scale bar of the bioluminescent image radiance (photons/s/cm²/sr) color scale was set to Min = 1.00e5 and Max = 1.00e6 for all pictures. (B) Quantification of bioluminescence in primary tumors performed using the automatic ROI tool (n = 5), showing average and standard deviation of radiance. *p < 0.05 and **p < 0.01 when compared to WT cells using a two-tailed Student’s t test. (C) Tumor growth in mice after subcutaneous inoculation of parental or KO B16-F10 cells. n = 5 mice/group. Tumor volume was measured using a caliper. *p < 0.05, **p < 0.01, and ns, not significant when compared to WT cells on the last day. (D) Lung colonization of mice imaged 5 weeks after tail vein injection of 2e5 luciferase-expressing parental or KO B16-F10 cells. n = 5 per group. (E) Quantification of bioluminescence in primary tumors performed using the automatic ROI tool (n = 5), showing average and standard deviation of radiance. **p < 0.001 and ns, not significant using a two-tailed Student’s t test. (F) Immune cell infiltration of B16-F10 tumors determined using immunohistochemistry of paraffin-embedded sections stained with antibodies to detect T cells, macrophages, and NK cells. Arrowheads indicate examples of positive cells. Scale bars: 50 μm. (G) Pathologist assessment of immune infiltration in B16-F10 WT and KO tumor sections. Scoring was done using 5 fields from 3 tumor sections. (H) Model showing sequential changes in MITF, TFEB, and TFE3 expression following glucose limitation over time.