Gene-specific actions of transcriptional coregulators facilitate physiological plasticity: Evidence for a physiological coregulator code

Abstract

The actions of transcriptional coregulators are highly gene-specific, i.e. each coregulator is required only for a subset of the genes regulated by a specific transcription factor. These coregulator-specific gene subsets often represent selected physiological responses among multiple pathways targeted by a transcription factor. Regulating the activity of a coregulator via post-translational modifications would thus affect only a subset of the transcription factor’s physiological actions. Using the context of transcriptional regulation by steroid hormone receptors, this review focuses on gene-specific actions of coregulators and evidence linking individual coregulators with specific physiological pathways. Such evidence suggests that there is a “physiological coregulator code”, which represents a fertile area for future research with important clinical implications.

Transcriptional coregulators selectively regulate specific physiological pathways that are dysregulated in human diseases

Transcription factors (TF) (see Glossary) bind specific regulatory DNA sequences, i.e. enhancer and silencer elements of specific target genes. The DNA-bound TF recruits many coregulator proteins which modulate local chromatin conformation and regulate the formation and activation of transcription complexes on the transcription start sites (TSS) of spatially associated genes [1–4]. Thus, TFs choose the genes to regulate, while coregulators carry out the transcriptional activation or repression (Box 1). Coregulators generally do not bind DNA directly but are recruited by TFs to specific genomic regulatory loci.

Box 1.

Each TF and coregulator regulates many genes in a given cell type, either directly or indirectly. Direct gene targets are those where TF or coregulator is physically associated with a regulatory site controlling the target gene. Indirect gene targets do not have regulatory elements bound by the TF and coregulator, but are regulated indirectly by products of direct target genes. While both direct and indirect target genes are relevant to the physiological effects of a TF or a coregulator, this review focuses on direct targets in order to understand the molecular mechanisms of coregulator action. It should be noted, however, that since rigorous assignment of a specific regulatory sequence to a specific gene requires arduous regulatory sequence editing experiments [105], such assignments are usually made by association, relying on correlation of gene regulation by the TF/coregulator with TF/coregulator binding to a genomic site in the vicinity of the gene.

Coregulators are often referred to as “coactivators” or “corepressors” depending on whether they help to activate or repress transcription; but some coregulators were found to act as coactivators on some genes and corepressors on other genes that are targets of the same TF in the same cell, suggesting that at least some coregulators can function in a gene-specific manner [5–7, 48, 106]. In general, depletion of a single coregulator (e.g. by siRNA) results in a partial reduction in transcription efficiency of the target gene, as indicated by a partial reduction in the target gene mRNA level [6, 33]. This suggests functional redundancy and sets up the possibility of modulating the transcriptional efficiency of genes that require a specific coregulator, by regulating its activity, as discussed in this review.

Recent genome-wide studies provided compelling evidence that coregulators generally function in a gene-specific manner. Each coregulator is required for regulation of only a subset of all the genes regulated by a specific TF in any given cell type [5–10]. Based on these findings, this review will focus on the gene-specific actions of coregulators and evidence that: the coregulator-specific gene subsets are enriched for specific physiological pathways; gene specificity of coregulators allows selective regulation of specific physiological pathways among many that are regulated by a given TF, thus selectively modulating the actions of the TF; post-translational modifications (PTMs) and protein-protein interactions regulate coregulator actions; and a “physiological coregulator code” exists, making coregulators potentially good therapeutic targets for pharmacological intervention in human diseases. Specific human diseases, such as cancer and diabetes, usually involve improper regulation of specific physiological pathways. The role of TFs in regulating specific physiological pathways has been well documented, but the more recent discoveries that coregulators also selectively regulate specific pathways opens the possibility of coregulators as therapeutic targets.

Transcriptional regulation is critically important to cellular differentiation and is one of the main mechanisms allowing cells to respond to acute external and internal signals. These signals modulate specific intracellular signaling pathways and alter PTM of TF and coregulators, changing transcriptional efficiency of target genes [11–18]. While functional relevance of PTMs for many TF and a few coregulators is known, essentially all TF and coregulators probably undergo PTMs that have functional consequences. We also suggest herein that coregulators and TFs represent two different layers of transcriptional regulation which are modulated by cellular signaling pathways via PTMs (Figure 1).

Figure 1. Multiple Layers of Regulation for Transcription.

Transcription factors (TF) bind directly to specific DNA sequences that serve as regulatory sites for specific genes. The TFs recruit coregulators (coregs) that remodel chromatin and regulate the assembly and activation of transcription complexes. Signaling pathways triggered by extracellular and intracellular signals stimulate changes in post-translational modifications (PTMs) of the TFs and coregulators that alter their activities.

The actions of coregulators will be examined primarily in the context of their function with steroid hormone receptors (receptors for estrogens, androgens, progestins, glucocorticoids, and mineralocorticoids), for which the most intensive study of transcriptional coregulators has, arguably, occurred. Steroid hormone receptors belong to the nuclear receptor protein family, composed of intracellular proteins, many of which function as ligand-regulated transcription factors [19, 20].

Coregulators contribute to transcriptional regulation by diverse mechanisms

The process of gene regulation by TFs and coregulators involves modulation of chromatin conformation, recruitment (or dissociation) of RNA polymerase, and activation (or repression) of transcription initiation (Figure 2). Hundreds of coregulators have been identified [21] with scores involved in regulation of each gene, indicating extensive complexity.

Figure 2. Contribution of coregulators to transcriptional regulation by steroid receptors.

Hormones (black circle) activate a steroid receptor (SR), which binds to specific enhancer elements or hormone response elements (HRE) and modulates transcription by recruiting coregulator proteins, which remodel local chromatin conformation, promote physical proximity of the enhancer element with the promoter and transcription start site (TSS) of the gene, and regulate the assembly and activity of transcription complexes at the TSS. Coregulators include enzymes such as histone methyltransferases (HMT) and histone acetyltransferases (HAT) that add post-translational modifications (PTM) such as methyl groups (Me) and acetyl groups (Ac) on histones, SR, other coregulators, or other proteins in the transcription complex. The SWI/SNF complex and other ATP-hydrolyzing coregulator complexes also contribute to chromatin remodeling. Other coregulators (Coreg A-D) use protein-protein interactions to recruit additional coregulators or components of the basal transcription machinery, as exemplified by the Mediator complex which recruits RNA polymerase II to the transcription start site (solid angled arrow).

Chromatin remodeling.

The proteins responsible for the histone code — enzymes that make and remove histone modifications, along with the proteins that read and act upon these modifications — can be classified as transcriptional coregulators because they do not bind directly to DNA but are instead recruited to specific genomic sites by TFs or by interaction with other chromatin proteins or PTMs of histones [1–3, 22]. The large number of histone modifications along with the writers, readers, and erasers of the code easily begins to explain the large number of coregulators. In addition, there are four large families of ATP-dependent chromatin remodeling complexes, the SWI/SNF, INO80, ISWI, and CHD families, each complex consisting of one ATPase subunit and multiple additional subunits. These complexes change nucleosome positioning on DNA or otherwise alter chromatin conformation [23, 24]. Chromatin conformation and its modification by TF-coregulator complexes play critical roles in the transcriptional regulation process. Additional complexes organize three-dimensional chromatin domains called Topologically Associating Domains (TADs) that divide the linear DNA into contiguous segments of chromatin [4, 25]. CTCF and cohesin proteins are critical for formation of chromatin loops that define the TADs. Most transcriptional regulation is thought to occur by interactions between enhancer/silencer elements and promoter/TSS (Figure 2) that are contained within the same TAD. Binding of a TF-coregulator complex to a regulatory enhancer/silencer element triggers refolding of chromatin within a TAD or between TADs in such a way that the regulatory region is brought into spatial proximity with the promoter and TSS so that the TF-coregulator complex can remodel chromatin to create nucleosome-free regions found at active enhancer elements and TSS or cause chromatin compaction for transcriptional repression [26–29]. Cohesin and the mediator complex are frequently associated with this process [4, 25].

PTMs of non-histone proteins.

PTMs on TFs and coregulators modulate their activities in many ways [9, 11–13, 30]. Defining these PTMs and their functional consequences is a critical step in understanding the physiological code associated with the gene-specific actions of coregulators. For example, estrogen receptor α (ERα) can be acetylated, ubiquitinylated, sumoylated, methylated, palmitoylated, and phosphorylated, influencing breast tumorigenesis by regulating ERα expression, stability, subcellular localization, and sensitivity to hormones [31]. PTMs of coregulators are described in a later section of this review.

Recruitment of other coregulators.

While some coregulators are enzymes that make or remove PTMs or move nucleosomes, many contribute to transcriptional regulation by recruiting other coregulators, basal transcription factors, and RNA polymerase [32, 33]. Most coregulators are large proteins, containing multiple protein-interaction domains [1–3, 34, 35]. p300/CBP, G9a/GLP, and CARM1 contribute to transcriptional regulation through both protein interactions and enzymatic activities [6, 36, 37]. Furthermore, the association of coregulators with specific genomic loci requires that they interact with a TF or another coregulator. Some coregulators also interact with specific histone modifications [38–40].

Gene-specific actions of coregulators

Proteomic approaches have confirmed that coregulators do interact with many protein partners [34, 35]. It has generally been found that each coregulator interacts physically and functionally with multiple classes of TF, suggesting at first glance that each coregulator might be part of a universal gene regulation mechanism. This might seem to suggest that coregulators are not involved in physiological regulation, due to lack of pathway-specific actions. However, genome-wide transcriptome analyses by gene knock-out or depletion of specific coregulators clearly indicate that coregulators function in a highly gene-specific manner, as illustrated below.

Each coregulator is required for the regulation of only a subset of the genes that are direct targets of a single TF in a given cell type.

A corollary of this aspect of coregulator gene specificity is that different gene targets of a specific TF in a given cell type require different sets of coregulators for their regulation by the TF (Figure 3). In one genome-wide study of four different coregulators and target genes of the glucocorticoid receptor (GR) in A549 lung adenocarcinoma cells, each coregulator supported the regulation of a distinct set of GR target genes, with surprisingly little overlap between coregulators [8]. In addition, many studies show that individual coregulators are required for the regulation of only a subset of all direct target genes of a specific TF in a given cell type [5–7, 41, 42].

Figure 3. Modulation of coregulator activity by signaling pathways can result in gene-specific effects.

The glucocorticoid receptor (GR), like other transcription factors, binds to many genomic loci that serve as enhancer and silencer elements for specific genes and regulate the transcription of these genes. GR regulates genes involved in many physiological pathways, including anti-inflammatory, metabolic, and tissue remodeling pathways. Different GR target genes require different sets of coregulators (colored rectangles) for glucocorticoid (black circles) regulation of their transcription. Each coregulator can undergo specific post-translational modifications (PTM) controlled by a variety of signal transduction pathways. Different signaling pathways control modifications to different coregulators. The modifications regulate the activities of the coregulators. GR target genes belonging to a specific physiological pathway use similar sets of coregulators, while GR target genes belonging to different physiological pathways require different sets of coregulators. Stimulation of a specific signaling pathway will affect the activities of one or a specific subset of coregulators. As a result, activation of a specific signaling pathway will alter the glucocorticoid regulation of some, but not all of the physiological pathways targeted by GR, thus, in effect, “fine tuning” the physiological responses caused by glucocorticoids by regulating the activities of specific coregulators.

Such diversity in the coregulator requirements of individual genes suggests that the regulatory mechanisms of each gene evolved independently. It is possible that different coregulator requirements of different genes are random accidents of evolution with no physiological significance. However, this increased complexity offers additional opportunities for regulation, and evolution has generally taken advantage of such opportunities. Since each TF generally regulates transcription of genes associated with multiple physiological pathways, the diversity of coregulator requirements represents an opportunity for independent modulation of subsets of genes associated with specific physiological pathways among the multiple pathways regulated by the TF (Figure 3).

Dual gene activation and repression by coregulators

Many, if not most, coregulators can function as a coactivator (helping to activate genes) or a corepressor (helping to repress genes) for a single TF in the same cell type (Box 1). For example, GRIP1 (encoded by Ncoa2) was initially characterized as a coactivator for nuclear receptors (including GR) and other types of transcription factors [32, 43–45], but was subsequently also shown to be critical for repression of proinflammatory genes by GR [5, 46, 47]. Genome-wide gene expression analyses showed that G9a (encoded by Ehmt2) [6], Hic-5 (encoded by Tgfb1i1) [7], and NCoR (encoded by Ncor1) [48] also can contribute to activation or repression of direct target genes of a single steroid receptor. G9a, which is recruited to genomic GR binding sites in a hormone-dependent manner, helps GR to activate some genes and repress other genes, while a third set of GR target genes does not require G9a for their regulation by hormone-activated GR [6, 49]. Hic-5 exerts an even more diverse set of effects, helping GR to activate some genes and repress others, but opposing the hormonal regulation of transcription on other GR target genes, so that Hic-5 depletion actually enhances the hormonal effect on transcription of these genes. Finally, another set of genes are activated or repressed by hormone-activated GR only when Hic-5 has been depleted [7].

Multiple protein interaction domains make coregulators versatile proteins.

Coregulators use different domains to interact with different TFs that recruit them to specific genomic sites. The steroid receptor coactivators SRC-1/NCoA1, SRC-2/NCoA2/GRIP1, and SRC-3/NCoA3/ACTR (encoded by Ncoa1, Ncoa2, and Ncoa3, respectively) are 160-kDa proteins that interact with the hormone-activated conformation of the ligand binding domains of nuclear receptors through LXXLL motifs found in the central region of the coregulator polypeptide chain [50, 51]. Many other coregulators have similar motifs that interact with nuclear receptors [52]. The SRC proteins also have a large N-terminal bHLH-PAS domain, the most highly conserved domain among the three SRC proteins, which serves as a binding site for a variety of other TFs [32]. Once recruited to specific genomic regulatory sites via these interactions with TFs, coregulators also use a variety of activation domains (AD) to influence the transcription process. SRC proteins use the N-terminal bHLH-PAS domain (sometimes called AD3) to recruit other coregulators, such as CoCoA (encoded by Calcoco1) [32]. GRIP1/SRC-2, but not SRC-1 or SRC-3, also contains a region that is critical for repressing genes but does not contribute to GRIP1 coactivator activity [5]. The conserved C-terminal region of SRC proteins (called AD2) binds the protein arginine methyltransferase CARM1/PRMT4 (encoded by Prmt4), which methylates histones and other coregulators such as p300 and CBP [12, 37, 53, 54]. Adjacent to AD2 is AD1, which binds histone/protein acetyltransferases p300 (encoded by Ep300) and CBP (encoded by Crebbp) [55–57] [38]. CBP and p300 are also very large and have many protein interaction domains. They bind many transcription factors, acetylate histones, and recruit or acetylate other coregulators using various domains [58].

Coregulators use different mechanisms of action on different target genes.

Many coregulators use different domains to exert different mechanisms of action on different target genes of a single TF in a given cell type, as illustrated by the different activation and repression domains of GRIP1 described above [5]. G9a helps many TFs to repress genes by making repressive methylation marks on histone H3 at Lysine 9, using its C-terminal SET domain [59]. G9a also represses genes by recruiting DNA methyltransferase with its ankyrin repeat domain [60]. In contrast, the coactivator activity of G9a involves the N-terminal domain, which contains a GR binding site and an AD [49]. The coactivator mechanism requires self-methylation of a lysine residue in the N-terminal domain, creating a binding site for HP1γ (encoded by Cbx3), which then helps to recruit RNA polymerase II [9].

Hic-5 binds the hinge domain of GR [61] and is recruited to genomic GR binding sites in a hormone-dependent manner [7]. On genes that require Hic-5 for activation by GR, Hic-5 facilitates recruitment of the Mediator complex and RNA polymerase II. When Hic-5 blocks hormonal activation of other genes, it prevents efficient GR association with the relevant genomic binding site [7] by selectively blocking GR interaction with a few of the many chromatin remodeling complexes [62]. Thus, different GR target genes require different remodeling complexes for hormone regulation of their transcription. This explains why Hic-5 selectively blocks GR binding to some DNA binding sites but not to others. The chromatin remodelers are thus another excellent example of differential coregulator requirements by different target genes of a single TF. Likewise, maintenance of DNase I hypersensitive sites associated with transcription regulatory sites involves diverse requirements for chromatin remodeling complexes [26].

What determinants direct the gene-specific requirements for and actions of coregulators?

The fact that different target genes of the same TF require different sets of coregulators presumably is due to inherent properties of their regulatory sequences and chromatin environments. The specific DNA sequence to which a TF binds can influence TF conformation and actions, as demonstrated with GR [63, 64]. Variations in TF conformation caused by subtle differences in DNA binding site sequences can influence the coregulators recruited [65] and could presumably influence the conformation and actions of recruited coregulators. Different synthetic ligands can also alter the conformation of steroid receptors [66], leading to differential recruitment of coregulators [67, 68]. In addition, each target gene is regulated by a unique set of regulatory elements in the DNA, so that the assembled collection of TFs and coregulators provides a unique set of potential protein-protein interactions for each gene, which can influence the requirements for and the actions of any given coregulator. Finally, chromatin conformation, although less well understood, is also presumably locus-specific and may therefore dictate specific coregulator requirements for transitioning to a conformation that is permissive to transcriptional activation or repression. Together, these factors establish a unique regulatory environment for each gene which determines the specific coregulators required for the regulation of transcription. Protein-protein interactions among TF and coregulators at the site, as well as PTM made by enzymatic coregulators that are present, should also influence the specific actions of the TF and coregulators, for example whether they activate or repress transcription.

Are gene-specific actions of coregulators associated with specific physiological pathways?

As stated above, each coregulator supports regulation of a subset of the target genes of a specific TF. If such a subset of genes represents a particular physiological pathway, this could provide opportunities for cells to alter the regulation of that pathway by the TF, by modulating the activity of the coregulator. Evidence for several coregulators supports the idea of pathway-specific actions [8, 9, 13, 18, 69]. The homologous coregulators SRC-1, SRC-2, and SRC-3 are all widely expressed in mammalian tissues, have many common gene targets, and often behave similarly in in vitro assays. However, the whole-body knockout mice have distinct phenotypes, suggesting that each SRC regulates different physiological pathways. In addition, some distinct functions have been demonstrated [70]. Only GRIP1/SRC-2 participates in glucocorticoid repression of cytokine genes in primary macrophages, which is an important component of the anti-inflammatory actions of glucocorticoids. Macrophage-specific knockout of the Ncoa2 gene encoding GRIP1/SRC-2 results in a broad derepression of lipopolysaccharide-induced genes that are normally repressed by hormone-activated GR [46]. Pathway analysis revealed a high prevalence of terms related to regulation of immune and inflammatory responses, cytokine production, and cell death. Furthermore, mice with macrophage-specific knockout of Ncoa2 were sensitized to systemic inflammatory challenges such as lipopolysaccharide-induced shock.

Similarly, genome-wide analysis of glucocorticoid-regulated genes affected by depletion of G9a/EHMT2 or its homologue GLP/EHMT1 indicated their requirement for glucocorticoid regulation of less than half of all GR target genes in A549 lung adenocarcinoma and Nalm6 B-cell acute lymphoblastic leukemia (B-ALL) cell lines [6, 9, 69]. G9a/GLP-dependent GR target genes were enriched for specific pathways in each cell type. G9a and GLP preferentially regulated GR target genes involved in A549 cell migration, and depletion of G9a or GLP blocked glucocorticoid inhibition of cell migration [9]. In contrast, their depletion in Nalm6 cells preferentially affected glucocorticoid regulation of genes involved in cell proliferation and cell death and desensitized the cells to glucocorticoid-induced cell death [69].

Can coregulator activity be regulated?

If gene-specific coregulator actions are indeed physiologically pathway-specific, then regulating the level of a coregulator (via transcriptional mechanisms) or its activities (through PTM or protein-protein interactions) could essentially fine-tune the actions of a TF in a pathway-specific manner. It would selectively enhance or inhibit TF regulation of some but not all of its targeted pathways (Figure 3). Since this additional layer of gene regulation via coregulators, superimposed on that conferred by TFs (Figure 1), would be a valuable capability for cells and organisms, it seems unlikely that evolution would pass up this opportunity to distinguish between multiple pathways regulated by a specific TF.

Glucocorticoids again offer an excellent example: cortisol, the natural human glucocorticoid, is a homeostatic hormone that regulates a wide variety of physiological pathways in various tissues and are important regulators of immune response and metabolism of glucose, lipids, bone, and muscle [72–76] (Figure 4, Key Figure). Synthetic analogues of cortisol are widely used as anti-inflammatory agents due to their multifaceted immune modulatory activities [77]. Among the many anti-inflammatory actions of glucocorticoids, the ability to trigger apoptosis of immature B and T lymphocytes is also responsible for their wide-spread use in treating many types of leukemia and lymphoma [78–80]. As a homeostatic hormone, circulating levels of cortisol are increased in response to various types of stress [81], such as hunger (low blood glucose levels), cold (low body temperature), fear, and illness (increased inflammation). Appropriate responses to the different types of stress should require different subsets of the many glucocorticoid response pathways, e.g. low blood sugar would require glucose regulation while illness and inflammation would require anti-inflammatory actions of glucocorticoids. There are now a variety of examples where modulation of the amount or activity of a specific coregulator selectively alters actions of steroid hormones or other signaling pathways on selected regulated pathways, as illustrated below.

Figure 4, Key Figure. The physiological coregulator code.

The natural glucocorticoid hormone cortisol (C) maintains homeostasis of many physiological pathways by regulating transcription of specific target genes. Cortisol release by the adrenal cortex is enhanced in response to various types of stress to restore homeostasis. Glucocorticoid target gene groups that regulate different physiological pathways require different sets of coregulators, so that regulation of the amount or activity of a specific coregulator by other signaling pathways will selectively influence specific aspects of the physiological response to glucocorticoids and thus “fine tune” the hormone response.

Modulation of coregulator amount

1. PGC-1α protein levels increase in response to thermogenic and nutritional challenges [82–84]. In the latter case, PGC-1α is strongly upregulated in mouse liver by fasting and helps GR and HNF-4α to upregulate gluconeogenic genes. Thus, stimulation of increased glucose production by glucocorticoids is enhanced by PGC-1α upregulation, while glucocorticoid regulation of other pathways involving PGC-1α-independent GR target genes presumably is not enhanced.

2. Estrogen stimulates C-terminal domain methylation of SRC-3 by CARM1, inducing dissociation of SRC-3 from CBP and CARM1 and subsequent reduction of SRC-3 stability [85, 86]. In contrast, C-terminal SRC-3 phosphorylation by atypical protein kinase C stabilizes SRC-3 by inhibition of its interaction with proteasomes. This enhances induction of specific ERα target genes in breast cancer cells and enhances estrogen-dependent proliferation [87].

3. Coregulators can also be expressed in a tissue-specific manner that can cause different responses to the same external signal in different cell types. For example, SRC-1 is expressed at higher levels in uterine cells than in breast cells. Experimental manipulation of SRC-1 levels showed that the higher SRC-1 level in uterine cells is responsible for the growth-promoting activity of the anti-estrogen compound tamoxifen, which inhibits cell proliferation in breast cells unless SRC-1 levels are experimentally increased [88].

Modulation of coregulator activity

PTMs regulate coregulator interactions with TFs or other coregulators, subcellular location, conformation, or enzymatic activity [11–13, 30, 37]. The signaling pathways that control PTMs of a specific coregulator provide an input for modulating TF regulation of the subset of target genes that require this coregulator (Figure 3). While many coregulator PTMs have been identified, as illustrated below, it is anticipated that the great majority of PTMs to coregulators have yet to be discovered and functionally characterized.

1. Extensive studies on a multitude of PTMs of the three SRC proteins illustrate many of the functional alterations mentioned above [14]. For example, phosphorylation, methylation, and ubiquitylation of SRC-3, stimulated by various signaling pathways, modify its cellular localization, molecular interactions, and stability [30, 85–87, 89]. Phosphorylation at six sites on SRC-3 made with differential specificity by several serine-threonine protein kinases is induced differently by estrogens and androgens versus TNFα, and distinct sets of these 6 phosphorylation events were required for interactions with steroid receptors, NFκB, and coregulators CBP and CARM1. Different phosphorylation patterns were required for stimulation of IL-6 expression by TNFα versus RasV12-induced oncogenic transformation of cells. Thus, distinct physiological responses are controlled by different signaling pathways by inducing distinct phosphorylation patterns in SRC-3 [89].

2. GRIP1/SRC-2 is phosphorylated at several sites by CK2 and CDK9 after binding to hormone-activated GR, but not after binding to hormone-activated TRβ1. GRIP1 phosphorylation is required for glucocorticoid regulation of a subset of GR target genes [90]. In macrophages N-terminal GRIP1 phosphorylation by CDK9 is required for glucocorticoid induction of several anti-inflammatory gene targets and occurs selectively at GR binding sites where CDK9 is recruited with GRIP1. Phosphorylation potentiates GRIP1 coactivator activity but not its corepressor properties. Phosphorylated GRIP1 isoforms and CDK9 are not detected at GR binding sites associated with pro-inflammatory genes that are repressed by hormone-activated GR [91]. Thus, through differential recruitment of CDK9 and resulting phosphorylation of GRIP1, GR selectively enhances actions of GRIP1 on a subset of GR target genes encoding anti-inflammatory proteins.

3. Activity of the corepressor NCoR is regulated by phosphorylation induced by insulin signaling upon feeding after fasting. This induces NCoR binding to and inhibition of nuclear receptors PPARα and ERRα, while reducing its association with and inhibition of LXRα. In the liver this derepresses lipid synthesis induced by LXRα but reduces oxidative metabolism mediated by PPARα and ERRα in response to refeeding [92]. Similarly, phosphorylation and methylation of RIP140 regulate its corepressor activity [93–95].

4. G9a and GLP are required for glucocorticoid regulation of subsets of GR target genes associated with a subset of the many physiological glucocorticoid responses [9, 69]. The N-terminal coactivator domain of G9a and GLP binds directly to ER and GR and contains an AD [6, 9, 49, 96] with a self-methylated lysine followed immediately by a threonine targeted by Aurora kinase B (AURKB). Methylation promotes coactivator activity by creating a binding site for HP1γ, which cooperates with G9a and GLP for glucocorticoid activation of a subset of GR target genes. Phosphorylation blocks binding of HP1γ and prevents the coactivator function of G9a and GLP. Glucocorticoid activation of G9a/GLP/HP1γ-dependent GR target genes is enhanced by AURKB inhibition or by increasing N-terminal G9a/GLP methylation via inhibition of specific lysine demethylases. In contrast, glucocorticoid activation of G9a/GLP/HP1γ-independent GR target genes is not affected by these inhibitors [9, 69, 97]. In A549 lung adenocarcinoma cells, G9a/GLP-dependent GR target genes are enriched for the cell migration pathway, including CDH1 (encoding E-cadherin), which is upregulated by glucocorticoids, causing inhibition of cell migration. Over-expression of G9a with a methylation site mutation prevents glucocorticoid repression of cell migration, demonstrating a requirement for G9a methylation [9]. In B-ALL cell line Nalm6 G9a/GLP/HP1γ-dependent genes that are activated by glucocorticoids are enriched for the cell death pathway. Inhibitors of AURKB or lysine demethylases enhance glucocorticoid-induced cell death by enhancing glucocorticoid-induced expression of G9a/GLP-dependent target genes, while having no effect on glucocorticoid induced expression of G9a/GLP-independent GR target genes [69, 97]. Standard-of-care treatment for pediatric B-ALL involves a combination of synthetic glucocorticoids with other chemotherapeutic drugs. Patients who relapse after initial treatment generally have glucocorticoid-resistant disease [78–80]. In addition to B-ALL cell lines, the Aurora kinase B inhibitors enhanced glucocorticoid-induced death of primary B-ALL cell samples from patients with relapsed disease, even though the cells were relatively resistant to glucocorticoids alone [69]. Such results merit further testing in patient-derived xenograft models to determine whether the combination of the inhibitors with glucocorticoids may represent a potential treatment for B-ALL patients with relapsed disease.

5. Splicing variations in genes Ncor1 and Ncor2 produce variants of NCoR and SMRT coregulators with divergent functions. Altered splicing patterns of NCoR contribute to adipogenesis, and splice-variant-specific knock-outs have strikingly different phenotypes than the complete Ncor1 gene knockout [98]. Furthermore, glucocorticoids, which help to promote preadipocyte differentiation to mature adipocytes in vitro, promote the appropriate Ncor1 transcript splicing variation that is required for adipocyte differentiation. Dietary changes in mice can also alter the splicing pattern of Ncor1 transcripts [99]. Thus, generation of coregulator variants can be regulated by hormonal and metabolic signals.

Concluding Remarks

The evidence reviewed here suggests that there is a physiological coregulator code, whereby cellular regulation of the activity of a coregulator, especially but not exclusively by controlling specific regulatory PTMs, preferentially modulates the regulation of genes associated with one or more specific physiological pathways among multiple pathways that are regulated by a TF. Within such a coregulator code, the gene specific actions and their regulation of specific physiological pathways establish coregulators, along with TF, as potentially valuable targets for intervention in diseases. One potential difficulty with targeting coregulators clinically is that each coregulator generally, if not universally, facilitates transcriptional regulation by multiple (probably many) transcription factors. However, we can once again invoke evolution in assuming that the target genes of two or more different TF that regulate a common physiological pathway might be served by the same coregulator (Box 2).

Box 2.

PGC-1α provides an excellent example of a coregulator that controls a specific physiological pathway by facilitating the actions of multiple TF. Its regulation of energy metabolism in response to stress involves PPARα, PPARγ, TRβ, ERRα, GR, MEF2C, NRF-1, and HNF-4α [82–84, 107].

Modulation of the activity of a coregulator could have a more pathway-specific effect than modulating TF activity, could modulate expression of pathway-specific genes targeted by multiple TFs, and could target the intended pathway with fewer unwanted side effects, due to the gene- and pathway-specific actions of coregulators. For example, in cancer some over-expressed coregulators (SRC-3 in breast cancer and AURKB in B-ALL) contribute to the cancer phenotype and thus would potentially be more important for survival of cancer cells than normal cells [69, 87]. Down-regulation of expression of a particular coregulator (e.g. via virus-delivered shRNA) is one potential therapeutic approach [100], but small molecule inhibitors of coregulator enzymatic activity or enzymes responsible for PTMs that regulate the activity of a specific coregulator may provide a therapeutically more attractive approach. Inhibitors of coregulators that serve as writers, erasers, or readers of the histone code are already being tested in the clinic. Examples are inhibitors of histone methyltransferase EZH2 (encoded by Ezh2) [101] and histone deacetylases [102, 103], and small molecules that inhibit binding of bromodomain proteins (histone code readers) such as BRD4 to acetylated histones [104]. Coregulators that do not have enzymatic activities pose more challenging targets, but targeting PTMs that control specific coregulator activities represents a potential solution, as illustrated above by lysine demethylases inhibitors or Aurora kinase B inhibitors which enhance G9a/GLP coactivator activity and thereby preferentially promote glucocorticoid-induced leukemia cell death over other glucocorticoid-regulated pathways [69, 97].

As summarized and exemplified above, the evidence for gene-specific requirements for coregulators is now quite extensive. Similarly, the association of many coregulators with specific physiological pathways is well documented. There are also compelling examples where PTMs of coregulators control their functions. However, in the context of the several hundred known coregulators, knowledge of the physiological pathways associated with each coregulator in specific cell types and the regulatory PTMs for each coregulator is extremely limited. Future research in these areas will be critical for defining the physiological coregulator code and realizing the potential of coregulators as therapeutic targets (See Outstanding Questions).

Outstanding Questions.

• The ability to target coregulators and their PTMs for therapeutic applications will require further advances in understanding coregulator function. In terms of elucidating the physiological coregulator code, we have only exposed the tip of the iceberg so far. In the great majority of cases, the specific coregulators that are required for regulating genes involved in specific physiological pathways have yet to be determined. Which coregulators and transcription factors are involved in regulating specific physiological pathways in specific tissues?

• Furthermore, PTMs that regulate the activities of specific coregulators have only been identified in a small number of cases. What signals and PTMs regulate the cellular levels and activities of each coregulator?

• Thus, a huge amount of exploratory and mechanistic basic research is still needed to understand fully the important roles of coregulators in physiological regulation and to realize their potential as therapeutic targets. What are the signaling pathways and enzymes responsible for PTMs that regulate coregulators, and what inhibitors would be suitable for clinical applications?

Highlights.

• Transcription factors (TFs) and coregulators represent two different layers of transcriptional regulation controlled by cellular signaling pathways via post-translational modificaitons (PTMs)

• Each coregulator functions in a gene-specific manner, supporting regulation of only a subset of target genes of a TF; the gene subset is often enriched for a specific physiological pathway among many regulated by the TF

• Cellular signaling pathways can regulate coregulator levels or modulate coregulator activity via PTMs, thus enhancing or inhibiting selected physiological responses among those regulated by the TF

• Human diseases involve misregulation of specific physiological pathways; because of gene- and pathway-specific actions, coregulators may represent more selective therapeutic targets than TF to correct the misregulation and inhibitors of the signaling pathways that control coregulator activity may thus limit side effects of treatment

Glossary

Activation domain (AD)

a domain of a transcription factor or coregulator that activates transcription after the transcription factor or coregulator is bound to a specific DNA regulatory locus; the AD exerts its action by recruiting other transcriptional regulators such as other coregulators, basal transcription factors, or RNA polymerase

AURKB

Aurora kinase B, an enzyme that phosphorylates specific proteins to promote cell division and also phosphorylates G9a and GLP to inhibit their coactivator function

B-ALL

B cell acute lymphoblastic leukemia

Coregulator

protein, or other macromolecule such as RNA, which is recruited to specific genomic regulatory loci by TF and helps TF to enhance or inhibit transcription by altering chromatin conformation and the formation of an active transcription complex

Enhancer element

DNA sequence serving as a binding site for transcription factors that activate transcription of associated genes

Glucocorticoid

steroid hormone made in the adrenal cortex which regulates many physiological pathways involved in homeostasis

Gluconeogenic genes

genes responsible for glucose synthesis in the liver

Glucocorticoid receptor (GR)

a steroid hormone-regulated transcription factor

LXXLL

an amino acid sequence, where L is leucine and X is any other amino acid, that is frequently found in coregulators and serves as a binding site for steroid receptors and other nuclear receptors

Post-translational modification (PTM)

covalent addition of a chemical moeity (such as a phosphate, methyl, or acetyl group, or the protein ubiquitin) to a protein

Silencer element

DNA sequence serving as a binding site for transcription factors that repress transcription of associated genes

Tamoxifen

a synthetic compound used to treat estrogen receptor-positive breast cancer because it binds to the estrogen receptor and thereby blocks estrogen stimulation of breast cancer cell growth; however, it does stimulate uterine tissue growth due to differential expression of coregulators in breast and uterine tissue

Transcription factor (TF)

binds specific DNA sequences that serve as enhancer and silencer elements for specific genes

Topologically Associating Domains (TADs)

relatively tissue-invariant, self-associating domains of chromatin that define chromatin domains on contiguous segments of DNA

Transcription start site (TSS)

the specific nucleotide in DNA where transcription of a gene begins