Volume 126, Issue 5 p. 939-948
Review Article
Free Access

The emerging role of mediator complex subunit 12 in tumorigenesis and response to chemotherapeutics

Shengjie Zhang PhD

Shengjie Zhang PhD

McArdle Laboratory for Cancer Research, University of Wisconsin-Madison, Madison, Wisconsin

Institute of Cancer and Basic Medicine (ICBM), Chinese Academy of Sciences, Cancer Hospital of the University of Chinese Academy of Sciences, Zhejiang Cancer Hospital, Hangzhou, China

Search for more papers by this author
Ruth O’Regan MD

Ruth O’Regan MD

Carbone Comprehensive Cancer Center, University of Wisconsin-Madison, Madison, Wisconsin

Search for more papers by this author
Wei Xu PhD

Corresponding Author

Wei Xu PhD

McArdle Laboratory for Cancer Research, University of Wisconsin-Madison, Madison, Wisconsin

Corresponding Author: Wei Xu, PhD, Department of Oncology, University of Wisconsin-Madison, 1111 Highland Avenue, Madison, WI 53705 ([email protected]).

Search for more papers by this author
First published: 23 December 2019
Citations: 26
We thank members of the Xu laboratory for helpful discussions. We also especially thank Kristine Donahue for editing.

Abstract

Transcriptional dysregulation induced by disease-defining genetic alterations of proteins in transcriptional machinery is a key feature of cancers. Mediator complex subunit 12 (MED12) is the central architectural subunit in the kinase module of Mediator, a large transcriptional regulatory complex that controls essential steps of transcription. Emerging evidence links deregulated MED12 to human cancers. MED12 is frequently mutated in benign tumors and cancers. Although the missense mutations of MED12 in benign tumors disrupt the kinase activity of Mediator, MED12 mutations in cancers could eliminate the interaction between Mediator complex and RNA polymerase II, leading to severe transcriptional misregulation. Aberrant expression of MED12 is associated with the prognosis of various types of human cancers. Loss of MED12 function has been associated with the development of resistance to chemotherapeutics. Moreover, MED12 is modified by posttranscriptional regulations. Arginine methylation of MED12 has been shown to regulate MED12-mediated transcriptional regulation and response to chemotherapeutics in human cancer cell lines. In this mini-review, the authors provide an overview of the roles of MED12 in the development of benign and malignant tumors as well as its roles in chemoresistance. The studies of MED12 exemplify that aberrant transcriptional programming is a therapeutic vulnerability for certain types of cancer.

Mediator Composition and Functions

By using genetic and biochemical approaches, the Mediator of RNA polymerase II (RNAPII) was first identified in the 1990s as a central integrator for transcription through interaction with gene-specific transcription factors as well as RNAPII in both yeast and mammalian cells. Mediator is a multisubunit complex comprised of 25 subunits in budding yeast and approximately 30 subunits in humans. Because Mediator interacts with over 3000 transcription factors (TFs), it controls a vast array of the gene transcription in an evolutionarily conserved manner.1 Mediator is assembled in 4 distinct modules, termed “head,” “middle,” “tail,” and “kinase.”2 Structural and biochemical data obtained from yeast and humans revealed that the head and middle modules are functionally essential for transcription regulation, whereas the tail and cyclin-dependent kinase (CDK) kinase modules play regulatory roles.1 Mediator interacts with RNAPII primarily through head module contacts. Gene-specific TFs generally bind Mediator through its tail and kinase domains, then transduce the regulatory information through the middle and head modules of Mediator to RNAPII (Fig. 1). Thus, all Mediator modules are involved in gene-specific transcriptional regulation, and the architecture of Mediator enables long-range gene activation.

Details are in the caption following the image
Mediator is a central integrator and processor of RNA polymerase II transcription. Mediator transduces regulatory information conveyed by signal-activated transcription factors to elicit gene expression changes in diverse biologic processes, including development, differentiation, and homeostasis. Mediator is structurally organized into 4 functional modules: “head,” “middle,” “tail,” and “kinase.” The Mediator kinase module comprises of 4 subunits: mediator complex subunit 13 (MED13), MED12, cyclin C (CycC), and cyclin-dependent kinase 8 (CDK8), which dynamically interact with the core modules during transcription elongation. The schematics show the relative position of Mediator components in transcription complex. Gene-specific transcription factors generally bind Mediator through its tail and kinase domains. Cohesion stabilizes the enhancer-promoter loop to enable the middle and head modules of Mediator to interact with RNA Pol II to conduct signaling dependent transcription. GTFs indicates general transcription factors.

Functions of MED12 in Mediator Complex and Physiologic Processes

MED12, along with MED13, cyclin C (CycC), and either CDK8 or CDK19, comprise the “kinase” module that reversibly associates with the core Mediator. MED12 activates the kinase activity of CDK8 by bridging the interaction between MED13 and CycC-CDK8 (Fig. 1). The kinase-stimulating activity of MED12 depends on its direct interaction with CycC.3 CycC, a highly conserved cyclin family member, consists of a negatively charged surface groove mediating its CDK8 binding as well as a CycC-specific surface for MED12 binding. MED12 binds to CycC through its N-terminus encoded largely by exons 1 and 2, where the hotspot mutations most frequently reside in hormone-dependent tumors.3 Notably, reciprocal mutation of residues at the interface of MED12 and CycC on either protein uncouples CycC-CDK8 from core Mediator and severely impairs CycC-dependent CDK8 kinase activity.4 Moreover, MED13 interacts with the C terminus of MED12 and plays an allosteric role in regulating the interaction between a mutant form of MED12 and CycC-CDK8/CDK19. Thus, when MED13 is present, mutant MED12 can bind, but is unable to activate, CycC-CDK8/CDK19.5 These studies implicate the MED12-CycC interface as a putative target in CDK8-driven cancers.

Because MED12 regulates essential physiologic processes, such as development and cell fate determination, deregulation of MED12 is often linked to human cancers. MED12 knockout mice elicit full developmental arrest at embryonic day 7.5, demonstrating that MED12 plays critical roles in early development.6 MED12 (also called TRAP230 or HOPA) maps to Xq 13.1, a region highly mutated (approximately 70%) in uterine leiomyomas. Missense mutations of MED12 can cause X-linked mental retardation, notably Opitz-Kaveggia syndrome (also known as FG syndrome) and Lujan-Fryns syndrome.7, 8 The protein encoded by MED12L shares 67% amino acid sequence similarity to MED12 and contains MED12-like proline-rich, glutamine-rich, and leucine-rich (PQL) and glutamine-rich (OPA) domains4; thus, MED12L may compensate MED12 function in a context-dependent manner (Fig. 2). In the current review, we provide an overview of MED12 in human carcinogenesis and its regulation of therapeutic drug response.

Details are in the caption following the image
Somatic mutations of mediator complex subunit 12 (MED12) in human cancers are illustrated. Pathogenic mutations and their approximate locations in MED12 are colored and annotated in the legend. The cyclin C (CycC)/cyclin-dependent kinase 8 (CDK8)-binding interface on MED12, corresponding to its N-terminal 100 amino acids, is disrupted by mutations of exons 1 and 2 (exon 1/2 mutations), leading to loss of mediator-associated CDK activity. In addition to the hotspot of exon 1/2, the distribution of MED12 mutations is dispersed in human cancers. Mutations shown here are drawn based on data from the Catalogue of Somatic Mutations in Cancer (COSMIC) database. FG indicates Opitz-Kaveggia syndrome (also known as FG syndrome); OPA, glutamine-rich domain; PQL, proline-rich, glutamine-rich, and leucine-rich domain;.

MED12 Mutations in Human Cancers

With the advance of next-generation sequencing, the incidence of MED12 missense mutations in human tumors has been continuously growing. MED12 mutations occur at a high frequency (range, 59%-80%) in estrogen-dependent benign tumors, including uterine leiomyomas and fibroadenomas and phyllodes tumors of the breast. These mutations are clustered on highly conserved amino acids residues (L36, Q43, and G44) mapped to exon 2.9-11 As described above, MED12 activates CycC-CDK8 through a direct interaction between MED12 amino acids 1 through 100 (encoded by exons 1 and 2) and a conserved surface groove on CycC. Therefore, the benign tumor-linked mutations at the MED12 CycC-binding interface substantially impair CycC binding and CDK8 activation, leading to global gene expression changes in MED12 mutant–expressing tumors. Because the kinase module is the regulatory unit in the assembly of transcription complex engaging signal-activated TFs, alteration of genes in the transforming growth factor-β (TGF-β), Wnt/β-catenin, and estrogen receptor α (ERα) signaling pathways have been reported that may lead to the initiation of tumorigenesis. Moreover, a uterine leiomyoma-associated MED12 mutation, Med12 c.131G>A, was reported as a “gain-of-function” mutation, which causes genomic instability and drives tumor formation.12

Large-scale genomic analyses have identified highly recurrent MED12 somatic mutations, albeit at a lower frequency, in hormone-dependent cancers, such as breast cancer, prostate cancer, and ovarian cancer, as well as in other cancers, such as lung cancer, colon cancer, and leukemia. Notably, the spectrum of MED12 mutations and cell types harboring MED12 mutations are different between benign tumors in estrogen-responsive tissues and in cancers. Barbieri et al identified recurrent MED12 missense mutations in 5.4% of prostate cancers using exome sequencing.13 In contrast to MED12 exon 2 mutations found in the stromal stem cells in uterine leiomyomas, the MED12 mutation in prostatic carcinoma resides in exon 26. The prominent mutation found in epithelial cells is the substitution of leucine 1224 by phenylalanine (L1224F). It has been proposed that MED12 mutations in prostate cancer interfere with the androgen signaling pathway and the CDK8-dependent transcriptional regulation of p53.13 However, this notion was challenged by the finding of Kampjarvi and colleagues that the L1224F mutation affects neither the interaction between MED12 and CyC-CDK8/19 in the kinase module nor the Mediator-associated CDK activity. Rather, the L1224F mutation on MED12 affected its binding to other Mediator subunits (MED1, MED13, MED13L, MED14, MED15, MED17, and MED24).14 To validate the prevalence of MED12 p.L1224F mutation, Stoehr et al analyzed a cohort of Caucasian patients with prostate cancer (n = 223) and could not detect the MED12 p.L1224F mutation in any cases.15 Similarly, Yoon and colleagues investigated the mutation sites in exons 2 and 26 of MED12 among 102 Korean patients who underwent radical prostatectomy for prostate cancer and could not identify MED12 mutations.16 Thus, the prevalence of MED12 somatic mutations in prostate cancer and the functional effects of the mutation in prostate carcinogenesis require further study.

The distribution of MED12 mutations in leukemia is diverse. The first identified MED12 mutation (c.97G>T, p.E33X) in T-cell acute lymphoblastic leukemia (T-ALL) is a nonsense mutation.17 The mutation affects the last codon of exon 1 and results in the use of an alternative translation start site. Consequently, an N-terminal–truncated protein was generated that failed to localize to the nucleus because of the absence of a nuclear localization signal.17 The mislocalization of MED12 abrogates its interaction with other Mediator components.17 Subsequent genome sequencing identified additional MED12 mutations, including a frameshift mutation (p.V167fs), a missense mutation (p.R1989H), and a splice site mutation (g.chrX:70339329T>C).18 Interestingly, nearly 60% of MED12 mutations were identified from immature T-ALL cases, suggesting that pediatric T-ALLs harbor genomic complexity. Identifying early disease-driven mutations will help to stratify patients for personalized treatments.18, 19 In chronic lymphocytic leukemia (CLL), N-terminal MED12 mutations were identified with a frequency of 5.2% (37 of 709 patients), 6.9% (12 of 188 patients), and 8.8% (10 110 patients) in 3 independent studies.20, 21 NOTCH signaling is known to regulate differentiation and tumorigenesis in multiple cancers, including CLL.22 Although the occurrence of MED12 and NOTCH1 mutations in CLL are mutually exclusive, NOTCH1 intracellular domain, the active form of NOTCH1, was elevated in CLL samples harboring MED12 mutations. Because NOTCH1 intracellular domain is a substrate of CycC-CDK8 kinase, this finding is congruent with the N-terminal MED12 mutations affecting CDK8 kinase activity and NOTCH1 activation. Therefore, mutating MED12 appears to be an alternative route for the activation of NOTCH signaling in CLL pathogenesis.19, 21

MED12 mutations have also been identified in colorectal cancer (CRC), gastric cancer, and nonanaplastic thyroid cancer by next-generation sequencing. An MED12 exon 2 mutation was observed in 389 patients with colon cancer (0.3%).23 Subsequent reports by Kampjarvi et al and The Cancer Genome Atlas Network identified 2 MED12 exon 2 mutations in 392 (0.5%) and 224 (0.4%) CRC samples, respectively, indicating that mutations of MED12 are rare and may not contribute to CRC tumorigenesis.23-25 The frequency of MED12 mutations is much higher in hereditary diffuse gastric cancer and nonanaplastic thyroid cancer. MED12 mutations occur in 48% (14 of 30) of sporadic gastric cancers (ILe1115Thr) and in 14% (8 of 57) of nonanaplastic thyroid cancers (Gly44Cys).26, 27 However, the numbers of patients in these series were small, and these findings require validation. To date, mutations of MED12 do not appear to correlate with prognosis, and the clinical relevance of MED12 mutations in the pathogenesis of these cancer types remains to be determined. It is unclear whether MED12 mutations are passenger or driver mutations and whether MED12 mutations alter the essential signaling pathways to which cancer cells become addicted.

Collectively, MED12 mutations are frequently found in different human cancer types, meeting the criteria of so-called “cancer driver genes” identified through large-scale genomic analyses.28 Coincidentally, a functionally related MED12L was implicated as a putative cancer driver gene in oral squamous cell carcinomas.29 The spectrum of MED12 mutations in human cancers is diverse and is distinct from mutations noted in benign tumors, in which mutations are largely clustered in exon 2. Table 1 and Figure 2 illustrate the MED12 mutations in different types of cancers in the Catalogue of Somatic Mutations in Cancer (COSMIC) database (Accessed June 29, 2019. http://cancer.sanger.ac.uk). Additional studies are needed to determine mutation prevalence using large cohorts of samples and to interrogate the functional significance of MED12 mutations in human carcinogenesis.

Table 1. Alterations of Mediator Complex Subunit 12 in Human Cancers
Tissue Type Alteration
Point Mutation Copy Number Variation Gene Expression
No. in Sample/Total No. % Gain/Loss No. in Sample/Total No. % Over/Under No. in Sample/Total No. %
Adrenal gland 9/735 1.22       Over 13/79 16.46
      Loss 2/268 0.75 Under 1/79 1.27
Autonomic ganglia 1/1231 0.08            
Biliary tract 13/814 1.6            
Bone 8/796 1.01 Loss 1/80 1.25      
Breast 888/6016 14.76 Gain 1/1544 0.06 Over 64/1104 5.8
      Loss 9/1544 0.58 Under 28/1104 2.54
Cervix 14/386 3.63 Gain 3/313 0.96 Over 45/307 14.66
            Under 4/307 1.3
Central nervous system 36/3066 1.17       Over 27/697 3.87
      Loss 7/1093 0.64 Under 16/697 2.3
Endometrium 67/1097 6.11 Loss 6/598 1 Over 47/602 7.81
Genital tract 1/126 0.79            
Hematopoietic and lymphoid 110/6089 1.81 Loss 1/835 0.12 Over 7/221 3.17
                 
Kidney 23/2668 0.86       Over 17/600 2.83
      Loss 5/1027 0.49 Under 20/600 3.33
Large intestine 180/4609 3.91 Gain 1/773 0.13 Over 107/610 17.54
      Loss 4/773 0.52 Under 12/610 1.97
Liver 38/2280 1.67 Loss 4/692 0.58 Over 20/373 5.36
Lung 135/4153 3.25 Gain 3/1185 0.25 Over 44/1019 4.32
      Loss 11/1185 0.93 Under 25/1019 2.45
Not specified 28/356 7.87            
Esophagus 17/1633 1.04       Over 28/125 22.4
      Loss 3/546 0.55 Under 1/125 0.8
Ovary 10/1431 0.7       Over 30/266 11.28
      Loss 5/729 0.69 Under 5/266 1.88
Pancreas 21/2407 0.87       Over 6/179 3.35
      Loss 1/929 0.11 Under 8/179 4.47
Placenta 1/5 20            
Pleura 2/358 0.56            
Prostate 55/3288 1.67            
Salivary gland 5/359 1.39            
Skin 78/1739 4.49 Loss 2/650 0.31 Over 19/473 4.02
Small intestine 3/111 2.7            
Soft tissue 1875/5900 31.78       Over 13/263 4.94
      Loss 1/286 0.35 Under 1/263 0.38
Stomach 47/1104 3.35 Gain 1/501 0.2 Over 37/285 12.98
      Loss 12/501 2.4 Under 5/285 1.75
Testis 2/398 0.5 Loss 2/152 1.32      
Thyroid 51/1868 2.73       Over 17/513 3.31
      Loss 1/506 0.2 Under 17/513 3.31
Upper aerodigestive tract 23/1674 1.37       Over 52/522 9.96
      Loss 5/563 0.89 Under 18/522 3.45
Urinary tract 41/1114 3.68 Gain 1/419 0.24 Over 63/408 15.44
      Loss 7/419 1.67      

Deregulation of MED12 Expression in Human Cancers

In addition to mutations, MED12 is often overexpressed in human cancers (Table 1). Nuclear accumulation of MED12 can be detected in 40% of castrate-resistant prostate cancers and in 20% of locally recurrent prostate cancers compared with <11% in androgen-dependent prostate cancers and nondetectable levels in benign prostatic tissues.30 Because knockdown of MED12 decreases proliferation, reduces G1-phase to S-phase transition, and increases cell-cycle–inhibitor p27 levels, MED12 expression levels have been associated with a high proliferative index. Positive feedback regulation between TGF-β signaling and MED12 transcriptional function has been reported. The activation of TGF-β signaling stimulates the nuclear accumulation of MED12 in cell lines, whereas knocking down MED12 reduces the expression of TGF-β target genes, such as vimentin. Moreover, genomic and epigenomic dysregulation of the MED12/MED12L axis also occurs frequently in castrate-resistant prostate cancer. These studies indicate that MED12 may promote the proliferation of prostate cancer cells and contribute to antiandrogen resistance.30

In contrast to prostate cancer, MED12 expression shows the opposite prognostic association in patients with breast cancer. Higher MED12 expression is associated with longer relapse-free survival in patients with breast cancer who receive chemotherapy.31, 32 Expression levels of CDK8 positively correlate with MYC, as well as CDK19, CycC, and MED13 in, breast tumors but are inversely correlated with MED12. Moreover, p53-mutant–expressing breast tumors typically have higher expression of CDK8, CDK19, and CycC but lower expression of MED12, indicating that MED12 may have distinct roles in regulating the CDK8 kinase module or eliciting kinase module-independent function in breast cancers.33

The level of MED12 expression is important for hematopoietic stem cell (HSC) homeostasis because in vivo knockout of MED12 results in rapid bone marrow aplasia, leading to acute lethality. Aranda-Orgilles and colleagues demonstrated that MED12 maintains HSC function in a cell-autonomous manner and is independent of CycC/CDK8 because the deletion of Mediator kinase module subunits does not affect HSC survival. MED12, together with p300, preserves the enhancers in an active state. Loss of MED12 causes depletion of H3K27Ac, an active histone modification marked by p300, at the enhancers of essential HSC genes and abrogates hematopoietic-specific transcriptional programs. This MED12-dependent enhancer regulation may be essential for maintaining the normal physiologic functions of HSC, and, consequently, aberrant regulation leads to malignant hematopoiesis.34 MED12 expression is also strongly associated with the sensitivity of leukemia to chemotherapeutics. In Jurkat leukemia cells, loss of MED12 prevented cells from entering apoptosis after chemotherapy.18

The oncogenic function of MED12 may involve crosstalk with oncogenic signaling pathways in ovarian and lung cancer. Loss of MED12 decreases the expression of EGFR in epithelial ovarian cancer. This coincides with the positive correlation of MED12 with EGFR expression in patients with epithelial ovarian cancer. MED12 knockout in ovarian cancer cell lines decreases EGFR protein levels. Thus, MED12 may regulate the dormancy of epithelial ovarian cancer through regulation of EGFR.35 In addition to the function of MED12 in genomic signaling, MED12 reportedly activated TGF-β receptor 2 (TGF-βR2) in the cytoplasm, constituting a nongenomic signaling pathway. MED12 has been detected in both the nucleus and the cytoplasm in lung cancer cells. It is highly plausible that MED12 in different cellular compartments elicits different functions: nuclear MED12 may be involved in modulating the function of the CDK8 kinase module to regulate transcription, whereas the cytoplasmic MED12 might interact with cell-surface growth factors like TGF-βR2, both of which may contribute to therapeutic drug resistance. How MED12 cellular localization is regulated and the contribution of nuclear versus cytoplasmic MED12 to the development of cancers are questions that remain to be elucidated.

Functions of MED12 Methylation in Human Carcinogenesis

Transcriptional abnormalities caused by aberrant epigenetic events are an emerging theme of the cancer epigenome.36 TFs, cofactors, and epigenetic enzymes are often subjected to posttranscriptional modifications that regulate their respective activities. Protein arginine methylation catalyzed by protein arginine methyltransferases in mammalian cells shares many attributes with other covalent modifications.37 We have identified MED12 as a substrate for coactivator-associated arginine methyltransferase 1 (CARM1), which asymmetrically dimethylates protein substrates on the arginine residues.38 We have demonstrated that the expression levels of CARM1 and MED12 in cancers are positively correlated. In addition, their high expression often predicts a better prognosis in patients with breast cancer who receive chemotherapy. We identified 2 MED12 methylation sites: arginine 1862 (R1862) and arginine 1912 (R1912).38 By searching the COSMIC database for mutations on R1862 and R1912, we found that R1862 was mutated in a case of lung carcinoma, and a somatic, homozygous mutation at R1912 was identified in a melanoma from a patient who developed resistance to combined treatment with RAF and mitogen-activated protein kinase (MEK) inhibitors. Most recently, Mark Bedford's group reported that a third CARM1-catalyzed MED12 methylation site, arginine 1899 (R1899), is involved in recruitment of the Tudor domain-containing effector molecule (TDRD3), and potentiates the interaction of MED12 with activating noncoding RNAs.39 These studies suggest that MED12 is modulated by methylation and possibly by other posttranscriptional modifications, and distinct methylation sites may engage different functions of MED12 in cancer cells.

MED12 in Drug Resistance

The finding of and mechanism by which MED12 is involved in mediating therapeutic drug response have been described in 2 studies. It is known that mutations of oncogenes could be acquired during the course of treatment and lead to failure of response to therapeutic drugs. Lung cancers harboring activating EGFR mutations develop resistance to EGFR inhibitors. Similarly, melanomas with activating BRAF mutations develop resistance to BRAF inhibitors. By using a large-scale RNAi screen, Huang and his colleagues identified MED12 as a predictor of response to ALK and EGFR inhibitors in non-small–cell lung cancer (NSCLC) cells.40 Those authors noted that a portion of MED12 is localized in cytoplasm, where it physically interacts with TGF-βR2 and negatively regulates TGFβ-R2 signaling. MED12 suppression is accompanied by the activation of TGF-βR signaling through the RAS-RAF-MEK-ERK pathway, causing resistance to EGFR, ALK, and BRAF inhibitors. In MED12-deficient cells, the inhibition of TGF-βR signaling could restore drug responsiveness, indicating that TGF-βR signaling is downstream of MED12, and the ablation of both induces synthetic lethality in MED12-deficient, drug-resistant tumors.40 The findings by Huang and colleagues highlight the causal effects of MED12 deficiency in tumor resistance to tyrosine kinase inhibitor treatment (Fig. 3).40, 41 Reduced MED12 function was later confirmed by Rosell and colleagues as a hallmark of resistance to tyrosine kinase inhibitors in EGFR-mutant NSCLC.42 In contrast to NSCLC, MED12 function seems to be intact in small-cell lung cancer (SCLC), in which no association has been found between the expression of MED12 and TGF-βR2 or with clinical variables such as overall survival. Thus, the functional link of MED12 and TGF-βR2 only exists in the NSCLC subtype of lung cancer, and MED12 and TGF-βR2 should not be considered as universal biomarkers for all types of lung cancer.43

Details are in the caption following the image
The association between mediator complex subunit 12 (MED12) and cancer drug resistance is illustrated. In the canonical transforming growth factor-β (TGF-β) pathway, TGF-β receptor 2 (TGF-βR2) activates TGF-βR1, which, in turn, phosphorylates Smad2 and Smad3 (p). Then, Smad2 and Smad3 translocate to the nucleus and regulate the expression of TGF-β target genes. TGF-β signaling induces epithelial-to-mesenchymal transition (EMT) through the phosphorylation of β-catenin (p), leading to the up-regulation of Snail, vimentin, and N-cadherin. Loss of MED12 causes drug resistance by increasing the level of TGF-βR2, which activates the ERK and SMAD pathways. This leads to cell proliferation and the features of EMT. In addition, the methylation of MED12 by CARM1 (me) renders cancer cells sensitive to chemotherapy drugs through suppression of p21 transcription.

We have demonstrated that higher levels of MED12 and CARM1 predict a better response to chemotherapeutics in patients with breast cancer.31 Methylation of MED12 by CARM1 renders cells sensitive to 5-fluorouracil, but not to the receptor tyrosine kinase inhibitors, in vitro and in vivo, suggesting that the mechanism of mediating drug response in breast cancer cells is different from TGF-βR2-mediated pathways in NSCLC. Moreover, p21/WAF1 levels are up-regulated in MED12 methylation–defective cells. In breast cancer, high levels of p21/WAF1 are associated with a poor prognosis in patients who receive chemotherapy. Thus, MED12 methylation may serve as a predictive biomarker for patients with breast cancer who receive chemotherapy (Fig. 3).31 This is in keeping with the study by Huang et al in which loss of MED12 not only resulted in tyrosine kinase inhibitor resistance but also rendered lung cancer cells resistant to 5-fluorouracil and cisplatin.40 Thus, these findings demonstrate that alterations of MED12, both transcriptionally and posttranslationally, modulate chemotherapeutic response, suggesting that interruption of the MED12 signaling pathway could be a new strategy for treating drug-resistant cancers.

The Roles of MED12 in the Regulation of Super-Enhancer–Associated Genes

In addition to amplification, high levels of oncogene expression in cancer cells are often attributed to transcriptional regulation by highly active enhancers: so-called super-enhancers. Inhibiting cancer-acquired, super-enhancer–addicted transcription has emerged as a new means of targeting historically nondruggable oncogenic proteins, such as c-Myc, in highly proliferative cancer cells. Because transcriptional regulators, such as Mediator and BRD4, are enriched in super-enhancers, super-enhancer–associated genes are often highly sensitive to the inhibition of BRD4. Bromodomain and extraterminal motif inhibitors have been extensively investigated for treating hematopoietic cancers.44 Tumors can develop resistance to bromodomain and extraterminal motif inhibitors with prolonged treatment. Thus, there is sustained interest in identifying combinatory therapies to selectively decrease the expression of super-enhancer–associated oncogenes such as MYC. A functional shRNA screen conducted in acute myeloid leukemia (AML) cells identified the MED12, MED13, MED23, and MED24 subunits sharing functional similarity with BRD4, ie, restraining myeloid maturation.45 These findings indicate that BRD4 and Mediator functionally coordinate in gene-specific transcriptional activation that is essential for AML maintenance. Therefore, targeting Mediator subunits MED12, MED13, MED23, and MED24 potentially can inhibit cell proliferation in AML cells.45

Recently, Kuuluvainen and colleagues demonstrated that depletion of MED12 or MED13/MED13L drastically reduced the expression of super-enhancer–addicted oncogenes (eg, MYC) in colon cancer cells, leading to growth inhibition. The transcription factor involved is β-catenin, a known TF that interacts with MED12. β-Catenin depletion causes reduction of MED12 binding to the super-enhancer of MYC. Although both Mediator components and BRD4 regulate the expression of oncogenes in a super-enhancer–dependent fashion in cancer cells, the genes effected by MED12 or MED13/MED13L depletion and inhibition of BRD4 do not completely overlap, suggesting that more efficient inhibition may be achieved by co-targeting MED12 and BRD4. At least in colon cancers, the oncogenic gene expression shows dependency of 3 Mediator subunits, MED12 and MED13/MED13L. Therefore, targeting the Mediator subunits, either alone or in combination with BRD4 inhibition, may provide a new treatment strategy for colon cancer.46

The Relevance of MED12 to the Estrogen Pathway

Although, in breast cancer cell line models, MED12 is in the activator complex regulating ER transcriptional activity, the mechanism by which MED12 affects ER signaling remains unclear. One model is that Mediator and cohesin cooperate to enforce the long-range chromosomal interactions (through looping) that are essential for enhancer-driven RNAPII transcription (Fig. 1). Mediator and cohesin co-occupy different promoters to generate cell-type–specific DNA loops to control the gene expression program.47 Recently, it was found that MED12 colocalizes on the ERα (ESR1) gene with the cohesion subunit SMC3. The occupancy of either subunit across the ESR1 gene depends on the other, and both are required for RNAPII–mediated transcriptional initiation, suggesting that high-order chromatin architecture controlled by MED12 and cohesin may be exploited for the regulation of ER expression and the treatment of estrogen-dependent breast cancer.48 However, control of ER expression and ER-regulated transcription in ER-driven malignancies is far more complex and has been elegantly reviewed by others.49, 50 Liu and colleagues recently reported that MED12 interacts with the JmjC-domain–containing protein (JMJD6) and recruits JMJD6 to the active enhancers bound by ERα, releasing paused RNA polymerase on cognate estrogen target genes. These studies identified JMJD6 as a critical regulator for the productive transcription of estrogen/ERα–bound enhancer coding genes. JMJD6 is a multifunctional protein: it is an iron (Fe2+) and 2-oxoglutarate (2-OG)–dependent dioxygenase as well as a protein arginine demethylase. Because JMJD6 plays an essential role in ER transcription and can hydroxylate lysines on both histones and nonhistone proteins,51, 52 it may be exploited as a molecular target for developing epigenetic therapies for ERα-positive breast cancers.53 Although MED12 functionally regulates ERα signaling, mutations of MED12 were more common in triple-negative breast cancer than in ERα-positive breast cancer, indicating that MED12 might play different roles in triple-negative-dependent and estrogen-dependent breast cancers,54 as in the cases of lung cancers.

Conclusions

The reason for the dispersed mutation spectrum of MED12 in human cancers is intriguing. MED12 mutations are frequently found in estrogen-dependent benign tumors and in various cancer types, suggesting that mutation of MED12 could be a tumor-initiating event. However, whether MED12 mutations in benign tumors, in conjunction with other genetic mutations, drive cancer progression awaits further investigation. The expression levels of MED12 are frequently altered in various types of human cancers, yet the prognostic association may be cancer type-specific. Attention also should be drawn to the subcellular localization of MED12 and the posttranscriptional regulation of MED12, because both genomic and nongenomic functions of MED12 have been associated with therapeutic drug resistance. Although the exact functions of MED12 overexpression or mutations in the context of Mediator complex remain to be elucidated, MED12 is a therapeutic vulnerability in broad types of human cancers, and synthetic lethal screens should be exploited for the treatment of MED12-deregulated cancers in the future.

Despite the strong implication of MED12 in human carcinogenesis, many questions remain to be addressed. For example, why is the frequency of MED12 mutations in malignant tumors much lower than that in benign tumors? How is MED12 protein nuclear localization controlled? Are posttranscriptional modifications of MED12 involved in the regulation of genomic and nongenomic functions of MED12? Although our search of the COSMIC database reveals copy-number variations of MED12 in various human cancers (Table 1), alterations in copy number changes related to human carcinogenesis have not been reported. It is worth noting that MED12 is located on the X-chromosome; therefore, future mutation analyses should take into consideration whether mutation of MED12 occurs on the activated or inactivated X-chromosome in gynecologic cancers. Finally, more effort is needed to delineate the cell-specific and tissue-type–specific functions of MED12 and its mechanism of action in normal cells and cancer cells.

Funding Support

This project is supported by National Natural Science Foundation of China (81502603), the Natural Science Foundation of Zhejiang Province (LGF18H160016), and the Medicine and Health Science Fund of Zhejiang Province (2018RC021) to Shengjie Zhang, and by grants from the National Institutes of Health (R01 CA213293 and R01 CA236356) to Wei Xu.

Conflict of Interest Disclosures

Ruth O'Regan reports grants from Pfizer and Cascadian; grants and personal fees from Novartis and PUMA; and personal fees from Genomic Health, Genentech, Lilly, and Macrogenics outside the submitted work. Shengjie Zhang and Wei Xu made no disclosures,