The MiTF/TFE Family of Transcription Factors: Master Regulators of Organelle Signaling, Metabolism, and Stress Adaptation

MITF is an evolutionarily conserved transcription factor with homologs identified in C. elegans and Drosophila (1). The MITF family encodes four distinct genes; MITF, TFEB, TFE3, and TFEC. Structurally, MITF genes constitute a double helix leucine zipper motif, a transactivating zone, and a domain responsible for DNA contact and binding. The MiT/TFE family of basic helix-loop-helix (bHLH) transcription factors recognizes the transcription initiation or E-box (Ephrussi boxes) sites (CANNTG) in the genome (2). The initial identification of the microphthalmia family of transcription factors revealed that MiT/TFEs must homodimerize or heterodimerize with another member of the MiT/TFE family to activate transcription (3–6). Early studies into the function of these transcription factors identified that mutations in MITF led to Waardenburg syndrome type II, characterized by hypopigmentation and defects in ectodermal development (7), while murine homozygous TFEB knockouts fail to develop due to lack of placental vascularization (8). More recently, MITF, TFEB, and TFE3 were identified as regulators of lysosomal function and metabolism. Numerous lysosomal and autophagy genes with one or more 10 base pair motifs (GTCACGTGAC) termed as Coordinated Lysosomal Expression and Regulation (CLEAR) elements are recognized by the MITF family, which in turn promotes gene transcription (9, 10). Genome-wide chromatin immunoprecipitation sequencing analysis demonstrated direct binding of TFEB to CLEAR elements with concomitant increment in lysosomal proteins (11). Genes that are most associated with TFEB regulation contain clusters of multiple CLEAR sequences. Genome-wide analysis for clustered CLEAR sequences identified 471 direct TFEB targets, which include lysosomal acidification and degradation enzymes along with autophagy, exo-, endo-, and phagocytosis genes. Surprisingly, several gene targets of TFEB also included those executing glucose and lipid metabolism, perhaps underscoring the direct connection between lysosomal function and metabolism (Fig. 1; ref. 11). Subsequent reports have also identified that both TFE3 and MITF are capable of binding CLEAR sequence elements to induce lysosomal biogenesis and autophagy in a comparable manner (12, 13).

The autophagy–lysosome system is a catabolic cellular process for whole organelles, protein aggregates, and other macromolecules (14). During carcinogenesis, autophagy exerts an antitumorigenic effect by degrading and/or recycling damaged cellular organelles, thereby blocking the accumulation of endogenous mutagens, and preventing further genomic alterations. However, following tumor induction, cancer cells coopt autophagy as a cell survival mechanism to promote nutrient reallocation for diverse cellular needs. Therefore, autophagy can suppress cancer development through its cytoprotective properties; however, once cancer has developed, these same properties sustain survival of the tumor (15). Cytoprotective and oncoprotective properties of autophagy include managing oxidative stress, preventing DNA damage, and supporting metabolism under nutrient-depleted conditions (15, 16). Adaptive metabolic reprogramming of cancer cells provides them with the ability to utilize diverse substrates as the building blocks for molecules necessary for proliferation. Indeed, autophagy participates in this program through degradation of lipids and proteins within lysosomes to derive substrates for nucleic acid and membrane biosynthesis (14, 16). The microphthalmia family of transcription factors are regulators of the autophagy–lysosome system (Fig. 1), and increasing evidence suggests that they also directly regulate metabolic and growth signaling pathways, and as such, represent an attractive therapeutic target with wide potential cancer. This review will discuss the prospect of MITF, TFEB, and TFE3 as necessary elements for the viability of several cancer types with specific emphasis on changes in cellular autophagy and metabolism.

A number of studies have determined TFE3 and TFEB as being oncogenes. Chromosomal translocations resulting in gene fusions involving TFE3 or TFEB are implicated in the development of sporadic renal cell carcinomas (RCC) and soft tissue sarcomas. These genetic rearrangements cause overexpression of the TFE proteins (17–19), and in the case of TFEB-MALAT1 fusion, places TFEB under the control of a more active promoter resulting in a 60-fold higher expression (20). Crucially, the resulting protein products from the gene fusion events still have functional basic helix-loop-helix domains and nuclear localization signals, keeping the transcriptional activation function intact (18). It has been reported that lysosomal localization, and thus inhibition, of MiT/TFEs requires the first 30 amino acids, corresponding to exon 1 (21). All reported gene fusions eliminate exon 1 from the resulting protein, indicating that the fusion proteins are unlikely to be able to localize to the lysosome, suggesting a mechanism of constitutive activation (18). Further associating MiT/TFEs in renal neoplasia is a kidney-specific TFEB overexpression mouse model that developed severe kidney enlargement with multiple cysts at 30 days following birth, while Ki-67–positive neoplastic lesions were detected as soon as 12 days after birth (22). Renal-specific TFEB overexpression also resulted in liver metastasis in 23% of mice (22). There are no reports about the activity of TFE fusion proteins; however, a case study has identified strong nuclear staining of the TFE3 fusion protein, and a cell line and xenograft model generated from the patient maintained this nuclear localization (23), a result that has been reproduced in several other IHC screens of TFE3 and TFEB translocation cancers (17, 24, 25).

The role of autophagy in TFE fusion cancers remains controversial. In the aforementioned mouse model of renal TFEB overexpression, LC3 expression was unchanged when compared with control mice, and crossing TFEB overexpression mice with Atg7 knockout mice did not significantly reduce cancer development (22). Conversely, several reports have identified that cathepsin K immunoreactivity and expression is a distinguishing feature of these neoplasms (26–28). Cathepsin K is a lysosomal cysteine protease, which is regulated by MITF in macrophages and osteoclasts (29), however, unlike cathepsins A, B, D, and F, does not contain and upstream CLEAR sequence promoter (11). Despite lacking a CLEAR promoter, the CTSK gene is highly enriched along with other lysosomal genes in pancreatic cancers driven by autophagy (30), and thus likely indicates a probable autophagy gene signature in TFE translocation–driven RCCs. Given that RCCs are characterized by metabolic dysregulation (31, 32), it is tempting to speculate that TFE fusion proteins promote stress response programs and help renal neoplasms to overcome metabolic crisis. There is preliminary support for this idea, as highlighted by increases in mTORC1 activity, the master regulator of growth and metabolism. TFE fusion RCCs display elevated ribosomal S6 phosphorylation, a positive indicator of mTORC1 activity, and therefore linking the MiT/TFE family of proteins with sustaining oncogenic anabolic pathways (33). Molecular analysis of TFE fusion cancers also revealed elevated expression of cell cycle–related proteins Cyclin D1 and D3 along with p21 (CDKN1A; ref. 34), which promotes Cyclin D–CDK4/6 complex formation before becoming inhibitory to cell-cycle progression through CDK4 phosphorylation (35). Interestingly, renal-specific TFEB overexpression in mice also results in elevated cyclin D1 and p21 gene expression (22).

Mutations in the FLCN gene result in Birt–Hogg–Dubé (BHD) syndrome characterized by renal and pulmonary cysts, noncancerous tumors of the hair follicles, and an increased risk of RCC (36). FLCN is proposed to act as a tumor suppressor through positive regulation of AMPK (AMP activated kinase; refs. 36, 37), and thus negative regulation of mTOR (38), which is supported by a homozygous knockout mouse model of BHD, which displayed hyperactivation of mTOR (39). The FLCN tumor suppressor's relationship with mTOR is uncertain, given that reports describe FLCN as a GAP (GTPase-activating protein) for Rag C/D (40). Given that GDP-loaded Rag C/D is necessary for amino acid sensing by mTORC1, a GAP for these proteins will activate this pathway (Fig. 2; refs. 41, 42). Indeed, models of BHD in yeast, mammalian cancer cell lines, and mice show that FLCN knockdown or heterozygous knockout results in reduced mTORC1 activity as measured by phosphorylation of S6 or S6K, while still resulting in renal tumorigenesis (43–45). As FLCN seems to have conflicting roles in regulating mTORC1 and AMPK, it seems that there are other mechanisms through which FLCN acts as tumor suppressor and one candidate is through cytoplasmic sequestration of MiT/TFE proteins. A report published in 2010 first highlighted that FLCN and TFE3 have a direct regulatory interaction in RCC (46). FLCN-null cells were shown to have decreased TFE3 phosphorylation, which resulted in increased nuclear localization. FLCN-deficient cells also displayed greater TFE3 M-Box promoter activity and had elevated expression of MiT/TFE target genes, including several related to lysosomal activity, both in vitro and in BHD patients (46). FLCN-null cells also expressed elevated mRNA levels of GPNMB (glycoprotein nmb), a marker of melanoma, glioma, breast cancers, which was also upregulated in a renal-specific TFEB overexpression mouse model (22). Further evidence to support a role for TFEB in FLCN tumor suppression was published in 2013, wherein the authors showed that FLCN loss led to increased nuclear TFEB caused by dysregulated lysosomal signaling. The authors also confirmed that FLCN directly interacts with Rag A/B in the absence of amino acids and promotes GTP loading of Rag A/B, which is a prerequisite for mTOR activation (45). The lysosomal surface is now understood to be a center for nutrient sensing (47, 48), namely amino acids, through a complex of proteins, including Rag GTPases, the Ragulator complex, vATPase, as well as the folliculin complex, containing both FLCN and FNIP1 (42). There is strong evidence to indicate that FLCN is an activator of amino acid signaling to mTORC1 through its GEF activity on Rag A/B and GAP activity on Rag C/D. However, this role conflicts with the conventional wisdom that mTOR activity is required for tumor progression. Indeed, it seems there are few cases where mTORC1 inhibition can promote cell growth, notably in nutrient-deplete conditions. Cell cultures models of oncogenic transformation display increased proliferation in the presence of the mTOR inhibitor Torin1 only when essential amino acids are absent, which is dependent on a functional lysosome, while mouse models of pancreatic cancer have a greater proliferative index in the interior, hypoxic tumor regions after rapamycin treatment (49). Therefore, it is plausible that FLCN loss causes an increased risk of neoplasia as cells acquire the ability to cope with nutrient deprivation following constitutive activation of MiT/TFE proteins. In support of this hypothesis are data that indicate that FLCN-null cells have greater levels of autophagy proteins (50), while suppression of autophagy in these cells results in increased sensitivity to paclitaxel treatment (51).

FLCN loss also promotes significant metabolic remodeling, as indicated by an increase in mitochondrial biogenesis, which is dependent on PGC1α (52–54). Although metabolic changes are thought to be a result of AMPK signaling (52, 53), it is likely that activation of MiT/TFE proteins in FLCN-deficient cells causes an upregulation of PCG1α, which is under the control of a CLEAR promoter, contributing to the phenotype. Like cancers that have lost FLCN, a subset of human melanomas, approximately 10%, are characterized by elevated PGC1α expression, which is driven by overexpression of MITF-M (55). PGC1α is the master regulator of mitochondrial biogenesis, and hence oxidative metabolism, so it is not surprising that PGC1α-elevated tumors exhibit greater respiratory capacity and enhanced reactive oxygen species clearance. PGC1α-elevated tumors require PGC1α for proliferation, and gene knockdown of PGC1α renders the cells susceptible to apoptosis (55). Furthermore, constitutively active BRAF melanomas have suppressed oxidative metabolism caused by downregulation of the MITF–PGC1α axis. Conversely, therapeutic BRAF inhibition results in an increase of MITF–PCG1α axis along with oxidative metabolism. The induction of oxidative metabolism in MITF-overexpressing melanoma cells results in increased sensitivity to the mitochondrial uncoupler 2,4-dintrophenol, revealing a novel biomarker for efficacy of antimetabolism therapeutics (56).

The role of MiT/TFEs in FLCN tumor progression also highlights their role in lysosomal signaling and nutrient sensing. Indeed, it is now understood that MITF and TFEB must be recruited to lysosomes to undergo inactivation by mTOR phosphorylation, a process accomplished through GTP-loaded Rag A/B, which have a direct interaction with MiT/TFEs (21). Recent studies have identified the major players in TFEB regulation as mTORC1 (mTOR complex 1) and the ERK, which are central regulators of anabolism and proliferation (57, 58). Phosphorylation of TFEB by mTOR at serine 211 creates a 14-3-3 binding site, which results in cytosolic retention of the transcription factor (57). In the absence of mTOR repression, TFEB is no longer bound by 14-3-3 and is free to enter the nucleus where it can enhance transcription of target genes (57). Subsequent studies have concluded that both TFE3 and MITF are controlled through mTOR-mediated phosphorylation (Fig. 2; refs. 12, 30). This gives rise to a negative feedback loop, where activated MiT/TFE promotes lysosomal biogenesis, and increases autophagy, which in turn induces mTORC1 activation through increasing lysosomal amino acids, and transcriptional upregulation of mTOR signaling proteins, such as FNIP2, RagC/D, and vATPase (11). Therefore, loss of inhibitory feedback by mTORC1 on MiT/TFEs could promote oncogenic transformation through constitutive mTOR activation and signaling (Fig. 2). Work published in 2015 supports this hypothesis, where the authors confirmed that one of TFE3, MITF, or TFEB was overexpressed in most human pancreatic ductal adenocarcinoma (PDAC) cells and patient samples and showed constitutive nuclear localization (30). Localization and activation of MiT/TFEs in PDAC cells was not dependent on mTOR activation or nutrient status, indicating a loss of inhibitory feedback. The constitutive activation and expression of MiT/TFEs resulted in elevated levels of autophagy–lysosome genes and increased autophagic flux. Interestingly, mTOR activity remained constant in PDAC cells even after 60 minutes of amino acid starvation, while siRNA knockdown of the overexpressed MiT/TFE rendered mTOR amino acid sensitive. A metabolomics approach confirmed that levels of free amino acids were most affected by MiT/TFE knockdown, while overexpression of MITF in a noncancerous pancreatic duct epithelial cells supported growth in amino acid–deficient media. Further evidence for a MiT/TFE feed-forward mechanism in cancer was provided in a report published in 2017. The authors found that overexpression of MiT/TFE genes in PDAC, RCC, and melanoma directly resulted in the overexpression of RagD, which rendered mTORC1 insensitive to nutrient starvation, fueling cell proliferation and oncogenesis in an mTORC1-dependent manner (59). It is clear that MiT/TFEs and autophagy can fuel cancer metabolism; however, there remains several questions about the mechanisms that regulate their expression and activity in cancer cells. It is currently unknown as to why only one of the transcription factors (TFEB, TFE3, or MITF) is overexpressed in a particular cancer, and whether this has any relevance to the phenotype. Likewise, the role of MITF in cancers beyond melanoma is understudied; however, more recently, it has been shown in that MITF, but not the splice variant MITF-M, is overexpressed in some pancreatic cancers and thus should be researched for a role in tumorigenesis of other tissue types (30). Furthermore, questions remain about how MiT/TFEs escape regulatory control by mTOR, given that mTOR is commonly activated in many cancers. One proposed mechanism is through overexpression of nuclear importin 8 (IPO8), which was identified as a common binding partner of MiT/TFEs in PDAC cells, and knockdown of IPO8 decreased nuclear localization of the transcription factors. It remains to be discovered exactly how IPO8 prevents mTOR inhibition; however, there may be alternate mechanisms through which MiT/TFEs become constitutively activated. Alterations in hetero- or homodimerization frequency and spontaneity with other MiT/TFE family members, as well as incorrect spatial regulation, that is, failing to recruit MiT/TFEs to the lysosome where mTOR resides could also account for constitutive activity.

A further mechanism through which MiT/TFEs can become implicated in cancer is through interplay between other known oncogene networks, namely the Wnt/β-catenin pathway. The Wnt signaling pathway functions through promoting nuclear localization of β-catenin, an oncogenic transcription factor, as a result of degrading the destruction complex, which includes GSK3β, AXIN, and APC among other proteins. β-Catenin is found to be constitutively activated in multiple cancer types as well as in mesenchymal and stem cell–like cancer cells (60–63). Two reports from 2015 indicate that MiT/TFEs are under the direct regulation of Wnt signaling pathway member GSK3β as a result of three conserved serine residues in the C-terminus region. Ploper and colleagues (13) showed that Wnt treatment of melanoma cells results in increased MITF stability and nuclear localization, while mutation of the putative GSK3β phosphorylation sites produced the same phenotype. With regards to TFEB, another group (64) similarly noted that GSK3β inhibition caused increased TFEB nuclear localization, as well as lysosomal biogenesis and autophagy; however, the stability of the protein was unstudied. Interestingly, recent reports have also highlighted the role of MiT/TFEs in promoting Wnt signaling through sequestration and degradation of the destruction complex in autolysosomes. A tetracycline-inducible MITF melanoma cell line displayed greater Wnt reporter gene activity following MITF induction, in a manner dependent on crucial endosome trafficking protein Vps27 (13). Furthermore, MITF induction in C32 melanoma cells caused colocalization of Axin1, the scaffold for the β-catenin destruction complex, with vesicular structures indicating that MITF induced sequestration of the destruction complex as a mediator of Wnt signaling (13). Modulation of Wnt signaling through destruction complex sequestration is not limited to MITF, as chronic TFEB inhibition in AMPK double knockout (DKO) mouse embryos led to impaired endoderm differentiation due to increased β-catenin phosphorylation resulting in decreased gene expression of β-catenin targets. Wnt signaling was partially rescued in AMPK DKO mouse embryos through expression of constitutively active TFEB. Interestingly, TFEB and MITF appear to mediate Wnt signaling through similar mechanisms, as wild-type but not AMPK DKO mouse embryos displayed extensive colocalization between lysosomes and GSK3β (65). Wnt signaling and gene expression is also upregulated in TFEB overexpression renal cancer mouse models, while treating these mice with Wnt inhibitors successfully reduces tumor growth (22). Although the evidence is clear that MiT/TFEs feed into Wnt signaling and vice versa (Fig. 2), the requirement of autophagy in the process is arguable. Indeed, autophagy can negatively regulate β-catenin signaling upon nutrient deprivation through binding of LC3 to β-catenin and subsequent degradation (66). Likewise, autophagy can inhibit Wnt signal transduction through degradation of dishevelled (DVL), the cytoplasmic effector of Wnt receptor: frizzled (67). Although the association between the MiT/TFE family and Wnt signaling has only recently become apparent, MITF has been linked to the Wnt pathway in melanoma far earlier. β-Catenin–induced melanoma growth requires functional MITF, and nuclear accumulation of β-catenin was correlated with increased MITF expression (68). Likewise, the melanocyte-specific isoform, MITF-M, is directly regulated by downstream transcription factor of Wnt, LEF1, through binding of the upstream M promoter (69, 70). Conversely, MITF is regulated by noncanonical Wnt family members, namely WNT5A, which causes downregulation of MITF and is characteristic of a distinct class of melanomas that are resistant to BRAF inhibition and immunotherapy (71, 72).

In conclusion, the microphthalmia family of transcription factors is emerging as important players in the development and sustainment of cancer. These transcription factors are activated or overexpressed in a diverse array of cancers where they are involved in sustaining proliferation, driving metabolism, and overcoming stress. As a result, they represent an attractive therapeutic target alone and in combination with other chemotherapeutic agents that induce cell stress. However, important questions still need to be answered before this research can be translated to improved patient outcomes. A better understanding of mechanisms surrounding activation in the cancer cell will help design therapies, especially as MiT/TFE inhibitors tend to be oncogenic drivers. Furthermore, an understanding of the systemic consequence of MiT/TFE inhibition must be studied, given that these transcription factors are considered essential for preventing neurodegeneration and cardiovascular disease, and with respect to cancer, impacts on immune system function merit further investigation.

No potential conflicts of interest were disclosed.

This work was funded by grants from the Natural Sciences and Engineering Research Council of Canada (RGPIN-2014-03687) and Beatrice Hunter Cancer Research Institute seed funding grant (to T. Pulinilkunnil). L. Slade is supported by the Cancer Research Training Program of the Beatrice Hunter Cancer Research Institute, with funds provided by the Canadian Breast Cancer Foundation—Atlantic Region and the New Brunswick Health Research Foundation.