This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Smith, C. L. Articles by O’Malley, B. W. Search for Related Content PubMed PubMed Citation Articles by Smith, C. L. Articles by O’Malley, B. W.

Coregulator Function: A Key to Understanding Tissue Specificity of Selective Receptor Modulators

Department of Molecular and Cellular Biology, Baylor College of Medicine, Houston, Texas 77030

Correspondence: Address all correspondence and requests for reprints to: Carolyn L. Smith, Ph.D., Department of Molecular and Cellular Biology, One Baylor Plaza, Houston, Texas 77030. E-mail: carolyns{at}bcm.tmc.edu

	Abstract

Top Abstract I. Introduction II. Biology of Selective... III. Mechanisms of SRM... IV. Molecular Basis of... V. Lessons Learned from... VI. Concluding Remarks References

 
Ligands for the nuclear receptor superfamily control many aspects of biology, including development, reproduction, and homeostasis, through regulation of the transcriptional activity of their cognate receptors. Selective receptor modulators (SRMs) are receptor ligands that exhibit agonistic or antagonistic biocharacter in a cell- and tissue context-dependent manner. The prototypical SRM is tamoxifen, which as a selective estrogen receptor modulator, can activate or inhibit estrogen receptor action. SRM-induced alterations in the conformation of the ligand-binding domains of nuclear receptors influence their abilities to interact with other proteins, such as coactivators and corepressors. It has been postulated, therefore, that the relative balance of coactivator and corepressor expression within a given target cell determines the relative agonist vs. antagonist activity of SRMs. However, recent evidence reveals that the cellular environment also plays a critical role in determining SRM biocharacter. Cellular signaling influences the activity and subcellular localization of coactivators and corepressors as well as nuclear receptors, and this contributes to gene-, cell-, and tissue-specific responses to SRM ligands. Increased understanding of the effect of cellular environment on nuclear receptors and their coregulators has the potential to open the field of SRM discovery and research to many members of the nuclear receptor superfamily.

I. Introduction

II. Biology of Selective Receptor Modulators (SRMs)

A. Selective ER modulators (SERMs)

B. Selective tissue estrogenic activity regulators (STEARs)

C. Selective PR modulators (SPRMs)

D. Selective AR modulators (SARMs)

E. Selective peroxisome proliferator-activated receptor modulators (SPARMs)

F. Other SRMs

III. Mechanisms of SRM Action on Steroid Receptors

A. General steroid hormone action

B. Effect of ligand on receptor structure

C. Coactivators

D. Corepressors

E. SRM hypothesis

IV. Molecular Basis of Cellular Selectivity

A. Receptor-selective recruitment of coactivators

B. Influence of DNA on coregulator interaction

C. Effect of cell signaling on receptor-coregulator interactions

D. Relative coregulator expression

V. Lessons Learned from Coregulator Knockout Mice

VI. Concluding Remarks

	I. Introduction

Top Abstract I. Introduction II. Biology of Selective... III. Mechanisms of SRM... IV. Molecular Basis of... V. Lessons Learned from... VI. Concluding Remarks References

 
NUCLEAR RECEPTORS COMPRISE a large family of eukaryotic transcription factors, and those for whom ligands have been identified are broadly exploited to manipulate various aspects of human biology (1, 2). There is a well-developed pharmacology for many of the nuclear receptors, and the identification of natural and high-affinity synthetic agonistic ligands for these receptors has enabled many studies of the biological effects of these nuclear receptors in vitro and in vivo. The availability of antagonists also has been important. In experimental studies they proved to be useful tools for validating that an effect under consideration is indeed mediated by a nuclear receptor of interest. Clinically, they have been used to block or inhibit undesirable physiological actions of receptors. For instance, tamoxifen, due to its ability to inhibit estrogen receptor (ER) action, is used widely in the treatment and prevention of breast cancer.

Careful examination of the selective biological effects of tamoxifen (e.g., estrogen-like activity in the uterus but antiestrogen-like effects in the breast) led to the emergence of the concept of selective ER modulators or SERMs (see below). The molecular mechanisms through which selective effects are obtained has been the topic of intense investigation with the result that not only do we have at least a basic understanding of how the selective nature of SERM activities is achieved, we have now progressed to a point at which the stage is set to pursue the identification and development of selective receptor modulators (SRMs) for a host of other nuclear receptors. The driving force in this research is the desire to obtain agents that can better separate desired nuclear receptor effects from those that are undesirable. Indeed, the lessons learned from SERMs provide a framework in which to pursue the development of other SRM ligands with high binding affinity to the receptor of interest. The experience with SERMs suggests that compounds with selective activities are likely to be of intense clinical and economic interest, thereby stimulating significant research in this area in both the basic science and pharmacological arenas.

	II. Biology of Selective Receptor Modulators (SRMs)

Top Abstract I. Introduction II. Biology of Selective... III. Mechanisms of SRM... IV. Molecular Basis of... V. Lessons Learned from... VI. Concluding Remarks References

 
A. Selective ER modulators (SERMs)
Estrogens have long been recognized to play critical roles during development and reproduction, as well as in the growth and maintenance of the skeleton. In addition, evidence of the contribution of estrogens to the normal function of the cardiovascular system and central nervous system (CNS), including cognition and potential delayed onset of Alzheimer’s disease, and a variety of other tissues and organs (e.g., colon) indicates that this class of steroids, and by extension its receptors, ER and ERß, plays a significant role in normal biology and pathophysiology. Estrogens are widely used clinically to control reproduction (i.e., oral contraceptives) and for hormone therapy and the management of menopausal symptoms in women. Although beneficial in these contexts, estrogen use also has been implicated as a risk factor in breast and uterine cancer, particularly since the first published report from the Women’s Health Initiative (3), suggesting that a greater measure of flexibility to control unwanted side effects would be desirable. Consequently, the recognition of SERMs as agents able to elicit estrogenic effects in a tissue-specific manner has expanded the potential population that could benefit from ER ligand therapies.

SERMs have been important for their clinical potential as SRMs as well as serving as the focus of a vast body of research defining the molecular mechanisms through which cell and tissue selectivity is achieved. The trans isomer of tamoxifen is the prototypic SERM (4, 5, 6, 7, 8). Although one of its first proposed uses was to regulate fertility, it has been employed primarily as an agent used to treat and, more recently, prevent breast cancer (9, 10, 11, 12). The ability of tamoxifen to inhibit ER action has long been considered integral to its utility in the breast cancer arena, and this is consistent with numerous studies and clinical trials demonstrating an effect of tamoxifen in ER-positive cells or breast tumors and an absence of any significant activity in those lacking ER expression (8, 13). However, the demonstration of tamoxifen’s estrogen-like effects in the rodent uterus and skeleton (7, 14) suggested that this drug may have distinct biological properties depending on the tissue environment. The subsequent observation of the estrogen-like effects of tamoxifen in the human skeleton (15) was important to the conceptualization of SERMs as potential drugs for indications other than breast cancer.

Tamoxifen’s activity profile has long been thought to be a reflection of its partial agonist/antagonist activity. For instance, in the absence of endogenous estrogens, tamoxifen frequently exhibits weak estrogenic activity, such as modest stimulation of uterine wet weight and bone density in ovariectomized rats, whereas in the presence of estradiol, it can serve as an antiestrogen, inhibiting responses to a level corresponding to the comparatively modest agonist activity of tamoxifen itself (7, 16). Much effort has been devoted toward understanding the molecular mechanisms through which selective ER actions are achieved, and this is the focus of much of this review. It is appropriate, however, to recognize that the success of tamoxifen as a SERM has been a driving force in the search for new SERMs as well as selective modulators for other nuclear/steroid receptors. Raloxifene-like tamoxifen exhibits antiestrogen activity in the breast and estrogen activity in the skeleton. However, raloxifene lacks the significant uterotropic activity associated with tamoxifen and therefore represents an improved agonist/antagonist profile (6, 17, 18, 19). However, because neither tamoxifen nor raloxifene possesses significant estrogen-like activity in the CNS, there is clearly a market for other SERMs to fill this niche. Indeed, a number of other compounds, including lasofoxifene, arzoxifene, and bazedoxifene, are under development, which may one day be of clinical use for chemoprevention of breast cancer or treatment and prevention of osteoporosis (20, 21, 22). It is also noteworthy that SERMs likely exist in nature. For instance, the estrone metabolite, ^{8, 9}-dehydroestrone sulfate suppresses hot flushes in postmenopausal women, an estrogenic action. However, it is unable to significantly affect certain other parameters associated with estrogenic responses such as total cholesterol, low-density lipoprotein cholesterol, and high-density lipoprotein cholesterol (23).

B. Selective tissue estrogenic activity regulators (STEARs)
Although distinct from SRMs, a new classification of compounds with tissue-selective biological activity, called selective tissue estrogenic activity regulators (STEARs), has arisen. This class of compounds does not interact directly with receptors such as the ERs, progesterone receptor (PR), or androgen receptor (AR), thus distinguishing them from SRMs. Instead, a precursor (prohormone) compound relies on tissue-selective metabolism to generate hormonal metabolites that have a tissue-specific functional profile. An example of such a drug is tibolone, which currently is the leader in worldwide sales for hormone therapy outside of the United States (24). The oral form of this steroid prohormone is inactive; once in the body, however, it is metabolized to 3- and 3ß-hydroxy-derivative forms, which are estrogenic, and a ⁴-isomer, which has weaker androgenic and progestational activities (25). Moreover, tibolone is a sulfatase inhibitor (e.g., blocks conversion of estrone sulfate into estrone) and can also stimulate local sulfotransferase activity (26). The resultant activity profile in humans shows estrogenic activity in bone, as well as CNS vasomotor suppression of hot flushes, but no significant uterotropic or mammotropic activities (27). With the exception of the notable CNS effects, tibolone has a SERM profile, but achieves this without directly modulating the estrogen, androgen, or PRs as it is the various metabolites of tibolone that exert their effects directly on these receptors (24, 26).

C. Selective PR modulators (SPRMs)
PRs control a number of processes critical for reproduction (28). In addition to the natural ligand progesterone, there presently are good synthetic PR agonists and antagonists available for use in humans, and these are used to control reproduction and the function of various reproductive organs such as the uterus. Although the need for a SPRM may be less obvious than for SERMs and selective AR modulators (SARMs) (see below), it has been suggested that a SPRM may be useful for treatment of endometriosis. Current therapies for endometriosis include the use of GnRH antagonists, PR agonists, or androgens. However, these options are not without side effects. GnRH antagonists, by virtue of their ability to induce a hypoestrogenic state, induce hot flushes and urogenital symptoms and may increase the risk of osteoporosis. Progestins may produce breakthrough bleeding, bloating, breast tenderness, mood changes, and breast epithelial proliferation, and androgen use is associated with seborrhea, acne, hirsutism, negative changes in lipid profiles, and, potentially, virilization. As an alternative, an agent that can more specifically target the endometrium may be an effective treatment for endometriosis, while minimizing side effects. A SPRM in combination with an ER ligand also may be useful in menopausal hormone therapy, particularly with respect to hormonal effects on the breast.

Several potential SPRMs have been identified [e.g., dexamethasone-oxetanone and J867 (29, 30)]. The biological properties of some of these agents have been shown to exert partial PR agonist activity in that they can stimulate the proliferation and differentiation of epithelial endometrial cells in juvenile, estrogen-primed rabbit uterus, although not as effectively as progesterone (30). In contrast, in the presence of progesterone these agents partially inhibit the activity of the PR and thus also possess antagonist properties. Various other measures of progestin/antiprogestin activity, such as induction of cervical ripening and parturition in pregnant guinea pigs, and the ability to inhibit mammary gland proliferation, indicate that the activity of these agents falls between that of antiprogestins such as onapristone (ZK 98,229) and mifepristone (RU486) and progestins such as progesterone and R5020. Differences in the action of SPRMs in comparison with antiprogestins also have been observed at the gene level; whereas medroxyprogesterone acetate (a progestin) and ZK137,316 (an antiprogestin) inhibit vascular endothelial growth factor expression in endometrial fibroblasts, the SPRM J867 is unable to do so (31). Moreover, at least one agent, J1042, is able to induce a marked reduction in endometrial thickness in cynomolgus macaques, suggesting it or a related compound may have potential as an endometriosis therapy (32). Recently, it has been shown that the SPRM asopril can inhibit estrogen-dependent uterine growth but is devoid of the breast stimulatory effects of progesterone. Such a compound is being used to treat endometriosis and uterine fibroids, while sparing the breast of significant stimulation (33).

D. Selective AR modulators (SARMs)
AR expression is widespread throughout the body, and androgens play a desirable role in promoting and maintaining bone strength, increasing muscle mass, decreasing fat tissue, and enhancing libido (34). Although androgen therapies are currently available, they are primarily based on delivery of testosterone or its derivatives by injections or skin patches (35). Neither approach is optimal because injections result in undesirable fluctuations in serum testosterone levels, and skin patches are associated with irritation and rashes. Oral preparations of currently available androgens are not recommended because of their relatively low efficacy and potential hepatic toxicity. There is, therefore, a desire to develop a form of androgen therapy that is easily administered orally and that will avoid the considerable fluctuations of serum androgens observed for injectables. More importantly, the goal exists to obtain androgenic therapies that do not exert undesirable side effects such as alterations in lipid profiles (e.g., high- and low-density lipoproteins), fluid retention, liver toxicity, prostatic hypertrophy, and gynecomastia. The more severe side effects associated with supraphysiological doses of androgens taken by body builders and athletes, such as increased aggression, decreased testicular size, and azoospermia, are unacceptable under all conditions. Simply stated, the goal of preservation of positive androgen effects in some tissues, while minimizing negative side effects in other tissues, has stimulated a search for SARMs.

Although the use of estrogens and SERMs is widespread, there has not been an equivalent trend observed for androgen therapies. This is not due to a lack of indication for androgen treatment, as this type of therapy would be of benefit for treatment of men with primary or secondary hypogonadism, osteopenia and osteoporosis, HIV wasting and cancer-related cachexia, anemias, various muscle dystrophies, and, potentially, male contraception. Although there is a paucity of well-designed studies supporting an indication for androgen therapy in women, androgen therapy has been advocated for improvement of bone strength, libido and other sexual parameters, as well as a sense of well-being in postmenopausal women (36).

Indeed, the current lack of SARM-based therapies results from a lack of a suitable agent. However, the greater understanding of the molecular events through which SERM actions are achieved established a rational basis for identifying and characterizing SARMs. Ideally, such an agent would be orally active, and, as a treatment for hypogonadism, should be capable of stimulating muscle mass and strength, bone strength, libido, and virilization but with minimal hypertrophic effects on the prostate. For osteopenia or osteoporosis indications, a SARM with anabolic activity in bone and possibly muscle, but with relatively little activity on sex-accessory tissues, would be desirable. SARMs for females might target libido and other sexuality parameters while avoiding virilization. Progress has been made in this area, and several compounds that possess a mixture of agonist- and antagonist-like activities in transient transfect assays measuring AR trans-activation of a target gene in cells have been identified (37, 38). Assessments of the in vivo SARM activity of these compounds is underway in animals as well as in humans, and they show a promising tissue-selective activity profile. Animal experiments with one such SARM, LGD2226, revealed that it prevented loss of bone mineral density associated with orchidectomy in rats and exerted anabolic activity in the levator ani muscle; in contrast, LGD2226 did not stimulate prostate weights above those observed for intact rats (39).

E. Selective peroxisome proliferator-activated receptor modulators (SPARMs)
Peroxisome proliferator-activated receptors (PPARs) are nuclear receptors that bind to DNA as heterodimers with the retinoid X receptors. The PPARs have been shown to play an important role in regulating genes involved in lipid metabolism and may well be exploited in the management of various other aspects of health and disease including inflammation and cancer (see Refs. 40, 41, 42, 43 for review). The synthetic ligands for the PPAR receptor, thiazolidinediones, have been exploited for their antidiabetic activity, but due to their side effects (e.g., weight gain and increase in LDL-cholesterol), the search for additional PPAR agonists has continued (40). Several potential compounds have been identified. One of these, N-9-fluorenylmethyloxycarbonyl (FMOC)-L-leucine improves insulin sensitivity without greatly activating PPAR adipogenic pathways (44). A second compound, the SR-202 phosphonophosphate, inhibits thiazolidinedione-induced PPAR activity as measured by trans-activation assays and 3T3-L1 cell differentiation to adipocytes, yet is still able to improve insulin sensitivity in vivo (45). Most recently, a novel non-thiazolidinedione-selective PPAR modulator, nTZDpa, which activates a unique profile of PPAR target genes in comparison with full agonists in white adipose tissue, was shown to ameliorate hyperglycemia and hyperinsulinemia in mice fed a high-fat diet while reducing body weight gain and adipose depot size (46). Moreover, nTZDpa did not cause cardiac hypertrophy, an effect associated with several other PPAR agonists (47). Although it is not yet clear the extent to which the effects of either of these two compounds are mediated by PPAR in vivo, these agents represent an advancement in the separation of insulin sensitization and adipocyte differentiation properties associated with thiazolidinediones, and they or agents modeled on their structures have potential as a future source of antidiabetic drugs.

F. Other SRMs
With our increased understanding of the mechanisms through which SRM activities are achieved (see following sections), it is now possible to extend the pharmacological concept of executable modulation to other members of the nuclear receptor superfamily. For example, the identification of a selective glucocorticoid receptor (GR) modulator (SGRM) with good antiinflammatory activity and reduced side effects (e.g., diabetogenic, osteoporotic) would be highly beneficial. The recent identification of the nonsteroidal GR ligand, AL-438, indicates that selective modulation of GR activity is obtainable (48). This compound retains potent antiinflammatory activity, as measured by a carrageenan-induced paw edema assay, yet has reduced the severity of side effects associated with glucocorticoid use. In rats, AL-438 did not induce hyperglycemia, nor did it exert the same negative effects on bone associated with use of the potent glucocorticoid, prednisolone. These differences are accompanied by differential regulation of GR target genes (i.e., osteocalcin and aromatase) by prednisolone in comparison with AL-438.

The appearance of tissue selectivity may also be achieved through the use of ligands that distinguish between receptor isoforms and, consequently, regulate biological processes in an apparently tissue-specific manner. For example, therapeutic use of thyroid hormone (T₃) will reduce serum cholesterol, but this is accompanied by various cardiac side effects, such as tachycardia and increased cardiac output. Studies from thyroid hormone receptor- (TR) and TRß knockout mice reveal that the effects of T₃ on the heart are primarily TR regulated, whereas the effects on cholesterol are mediated via TRß (49, 50). The development of thyromimetics that selectively target TRß offers the opportunity to achieve cholesterol-lowering effects with a reduced risk of negative effects on the heart (51). One such agent, KB-141, caused a significant reduction in cholesterol, lipoprotein(a), and body weight with no effect on heart rate in a short-term treatment of primates and suggests that selective TRß activation may have utility in the treatment of obesity and hypercholesterolemia (52).

	III. Mechanisms of SRM Action on Steroid Receptors

Top Abstract I. Introduction II. Biology of Selective... III. Mechanisms of SRM... IV. Molecular Basis of... V. Lessons Learned from... VI. Concluding Remarks References

 
A. General steroid hormone action
The basic molecular mechanisms of actions of the ligands for nuclear receptors such as ER, ERß, AR, PR, PPAR, and others have been revealed through research conducted over the last 15⁺ years. In general, the effects of these ligands are mediated via their cognate nuclear receptors, which are members of a superfamily of transcription factors. For clarity and because SERMs are the founding members of this class of compound, the basic model of steroid receptor action will be presented relative to the ERs. There are two ERs, ER and ERß (53, 54). It has long been established that 17ß-estradiol (E2) binding to ER induces a conformational change in the receptor’s hormone-binding domain and enhances receptor dimerization and the ability of receptors to bind to estrogen response elements (EREs) generally located in the promoter region of target genes (1, 55). Studies of the effect of ligands on the structure of the ERß hormone-binding domain reveal ligand-induced conformational alterations (56, 57, 58), and it is clear that ERß also binds with high affinity to EREs (59, 60). Two distinct regions within ER, apart from the centrally located DNA-binding domain, specifically contribute to transcriptional activity: the constitutively active, activation function-1 (AF-1), which is located in the amino terminus, and the ligand-regulatable AF-2 found within the hormone-binding domain. The ligand-binding domain (LBD) of ERß exhibits a moderately high degree of homology to the corresponding region of ER, and like all other steroid receptors, it possesses an AF-2 domain the activity of which is sensitive to ligand. However, its AF-1 domain is poorly conserved with that of ER and is less active (61); certain data suggest that the ERß A/B domain possesses a repressive function (61, 62, 63). Once at the promoter, the "activated," ligand-bound receptor, either ER or ERß, interacts with coactivator proteins to form a multiprotein complex that activates the general transcriptional machinery and increases the expression of target genes through processes involving chromatin remodeling, formation of stable preinitiation complexes, and enhanced rates of RNA polymerase II reinitiation (2, 64, 65, 66, 67, 68, 69).

The activities of the ERs, when assessed on EREs, can be inhibited by binding to antagonistic ligands, of which there are two types. Class I antiestrogens, such as 4-hydroxytamoxifen (4HT) and raloxifene, are referred to as partial or mixed agonists/antagonists, whereas type II antiestrogens, such as ICI 182,780, are called "pure" antiestrogens. These compounds bind to the ERs with high affinity (70, 71, 72); raloxifene and 4HT block the ligand-activated AF-2 domain and particularly in the case of 4HT, leave AF-1 able to initiate gene expression (73, 74). Depending on the cell type and promoter examined, AF-1 and AF-2 can mediate E2-induced transcription independently or synergistically (75, 76); their relative abilities to stimulate gene expression vary in a promoter- and cell type-specific manner (73, 77). As a result, 4HT and raloxifene stimulate gene expression in some, but not all, contexts, and these antiestrogens are therefore classified as SERMs. Importantly, the ability of 4HT or raloxifene to either activate or inhibit gene expression in a context-specific manner indicates that intrinsic cellular differences in processes or factors, such as cell-signaling pathways, accessory transcription factors, and/or transcription factor modulatory proteins, may account for the distinct interpretations of SERM biocharacter (i.e., agonist vs. antagonist activity). For ERß, 4HT and raloxifene also inhibit its AF-2 domain, and due to the relatively poor AF-1 activity of the receptor, these ligands generally block ERß transcriptional activity measured on EREs (61, 78, 79). In contrast to 4HT and raloxifene, the pure antiestrogens, ICI 164,384 and ICI 182,780, inhibit ER and ERß transcriptional activity in a context-indiscriminate manner. It should be noted that 4HT and raloxifene also exert effects on the transcriptional activity of both ER and ERß tethered to DNA indirectly through interaction with other transcription factors such as activator protein 1 (AP-1) and Sp1 (80, 81). In this context, the agonist activities of these ligands also is apparent in a cell-specific manner.

B. Effect of ligand on receptor structure
The first results indicating that ligands affect the structure of steroid receptors were obtained from antibody epitope mapping and partial proteolysis experiments (64, 77, 82, 83). For the latter, receptors were incubated with various ligands, subjected to limited digestion with enzymes such as chymotrypsin or trypsin, and, depending on whether ERs were bound to estradiol or antiestrogens, protected polypeptide fragments of different sizes were obtained. This was taken as an indication that these different classes of ligands induced distinct conformational changes in the C-terminal LBD of the receptor, later termed helix 12 (58, 64, 77); similar approaches have revealed ligand-induced conformational changes for AR (84), PR (83), GR (85), and PPAR (86).

Subsequently, the crystal structures of the LBDs of a number of nuclear receptors, including ER and ERß complexed with the agonistic ligands 17ß-estradiol or diethylstilbestrol (87, 88), were solved, and, like other members of the steroid receptor superfamily, these regions were found to be composed of 12 -helices. Surrounding a tightly packed central core composed of helices 5, 6, 9, and 10 are helices 2, 3, 4, 7, 8, and 11; helix 1 is not part of the conical nuclear receptor LBD "sandwich" motif (helices 2–11), and the position of helix 12 is variable. The region on the surface of the LBD to which coactivators bind via their NR boxes (see Section III.C for more details) is referred to as the coactivator-binding groove and is composed of residues from helices 3, 4, 5, and 12. The bottom and sides of this groove are nonpolar, but the ends are charged (87). When agonists occupy the LBD, helix 12 packs against helices 3, 5/6, and 11, thus forming part of the coactivator-binding groove. However, relative to the agonist-bound structures of ER and ERß, the position of the 12th helix in relation to the remainder of the LBD differs when mixed ER agonists/antagonists, such as 4HT or raloxifene, occupy the ligand-binding pocket (56, 87, 88). In these structures helix 12 is reoriented to partially occlude the coactivator-binding groove, therefore enabling it to block certain AF-2-dependent interactions with coactivators (56, 87). Thus, crystallography substantiates that ER ligands are determinants of the conformation of the LBD of the receptor. It is important to note that structures for the A/B domain or full-length ERs, including information on how ligands affect amino- and carboxy-terminal interactions, are not available. These will undoubtedly be important for fully understanding SERM action.

The conformation of ERß LBD bound to the pure antiestrogen ICI 164,384 is distinct as revealed by crystallography; helix 12 is disordered and therefore does not appear in the structure (89). However, the bulky side chain of ICI extends out of the ligand-binding pocket and makes contact with a portion of the coactivator-binding groove, therefore likely precluding productive LBD interaction with coactivators. As noted above, ICI antiestrogens, unlike SERMS such as tamoxifen and raloxifene, do not possess partial agonist activity. This has been suggested to result from the ability of ICI 164,384 to inhibit ER dimerization (90). However, recent data suggest that this is not the case; the ERß-ICI 164,384 crystal structure reveals a dimer (89), and fluorescence resonance energy transfer experiments demonstrate that both ICI 164,384 and ICI 182,780 induce ER LBD dimerization (91). The ICI compounds also have been shown to promote a modest nuclear to cytoplasmic shuttling of ER and induce the degradation of this receptor (92, 93, 94). The ICI 182,780 antagonist also has been shown to significantly retard the intranuclear mobility of ER and to render the receptor resistant to extraction, both suggesting a tight association with a subnuclear compartment (95, 96). The potent antiestrogen activity of the ICI compounds therefore resides in their ability to efficiently block coactivator interactions as well as other aspects of receptor function and expression required for transcriptional activity.

A second approach used recently to characterize the effect of ligands on the conformation of steroid receptors utilizes affinity selection of phage-displayed peptides (reviewed in Ref. 97). Full-length recombinant ER or ERß immobilized on tissue culture plates incubated with phage in the absence or presence of various ER ligands revealed that ligand-induced ER-phage interactions varied depending on the ligand and type of receptor (57, 98, 99). These patterns of interactions are consistent with the ability of each ligand to induce a receptor conformation that exposes a unique peptide-binding surface. These results are notable not only because they are consistent with the above mentioned crystallography studies, but also because they reveal structural differences in the context of full-length ERs instead of only the LBDs, as is the case for the crystallographic analyses. In addition, they clearly reveal differences in the structures of ERs bound to either 4HT or raloxifene, which are difficult to discern in the crystallographic analyses. This is important for understanding the molecular basis for differences in the biological activities of these two SERMs. Peptide-based approaches also have been used to evaluate the conformation of AR (100).

C. Coactivators
1. Coactivator interactions with ERs.
The hypothesized ability of coactivators to bind to steroid receptors in an agonist-dependent manner was exploited in the initial predictions of coactivators and corepressors (101, 102, 103). Upon cloning of the first authentic steroid receptor coactivator (SRC)-1, its interaction with PR or ER was demonstrated to be promoted by agonist and inhibited by antagonist (104). SRC-1 (NCoA-1) was the progenitor molecule for the SRC-1/p160 family of coactivators, which includes SRC-2 [transcriptional intermediary factor 2 (TIF2)/GR-interacting protein 1 (GRIP-1)] and SRC-3 [(ACTR/pCIP/receptor associated coactivator (RAC3)/TRAM-1/amplified in breast cancer 1 (AIB1) (104, 105, 106, 107, 108, 109, 110, 111, 112)]. This slightly confusing nomenclature will vary in this review, depending on the laboratory source for the data. Subsequent analyses have defined the NR box motifs (LxxLL, where L = leucine and x is any amino acid) found within many coactivators including the p160s as critical for their ability to bind to steroid receptors via their coactivator-binding groove within the LBD (113, 114). The structure of the co-crystal complex of the ER agonist, diethylstilbestrol, with the ER LBD and a NR box-containing portion of the GRIP1 coactivator reveals that residues within helix 12 as well as within helices 3 and 5 are important for mediating interactions between ERs and coactivators (87, 88, 115). As discussed above, helix 12, by virtue of the ability of ligands to alter its position relative to the remainder of the LBD, plays a critical role in regulating coactivator interactions with this region of the receptor (87, 116, 117). Thus, agonists promote coactivator binding to ERs by inducing a LBD structure favorable for this interaction. However, it is important to note that steroid receptors and coactivators can utilize other regions within their structures to bind to one another. For example, SRC family coactivators as well as cAMP-response element binding protein (CREB)-binding protein (CBP) and p300 also bind to the A/B domain of ER and ERß in a hormone-independent fashion (118, 119, 120); in the case of at least GRIP1 this interaction with the A/B domain is not dependent on the coactivator’s LxxLL motifs.

2. Coactivators functional roles.
Experiments performed in yeast provided some of the first indications that there may be competition among transcription factors (squelching) for binding to a limiting pool of accessory factors necessary for gene expression (121). Work in transient transfection systems utilizing cotransfection of PRs and ERs extended this concept to the nuclear receptor superfamily and strongly suggested that, in order to activate gene expression, receptors had to interact with some unknown factors in the cell (122). After initial biochemical identification of several ER-interacting proteins (103, 123), molecular biological approaches, chief among them yeast two-hybrid assays, resulted in the cloning of more than 50 coactivators in a relatively short period of time (66, 68, 69). In general, coactivator proteins (note: there is one RNA coactivator; see Ref. 124) do not bind to DNA, but interact indirectly through association with other DNA-binding proteins (e.g., nuclear receptors). Once recruited to the promoter, coactivators enhance transcriptional activity through a combination of mechanisms, including efficient recruitment of basal transcription factors such as template-activating factors and TATA-binding protein. In addition, nuclear receptor-interacting coactivators possess themselves, or recruit other nuclear proteins that possess, enzymatic activities crucial for efficient gene expression including the ATP-coupled chromatin-remodeling SWI-SNF complex, a number of acetyltransferase proteins (e.g., CBP/p300, pCAF, and p160s), methyltransferases [e.g., coactivator-associated arginine (R) methyltransferase-1 (CARM1) and PRMT-1/2] and ubiquitin ligases (e.g., E6-AP and Rsp5) (125, 126, 127, 128). A general model of coregulator interactions with ER is presented in Fig. 1 (see accompanying legend for more details).

View larger version (34K): [in this window] [in a new window]   FIG. 1. Model of nuclear receptor-dependent gene expression. This represents a hypothetical schematic of the exchange of coregulators involved in activation of a gene by a steroid hormone receptor, such as ER. Coactivators and corepressors exist in complexes in the cell and do not appear to bind to receptor as monomers. A, In the presence of antiestrogens, such as tamoxifen (T), the receptor interacts with a complex of corepressor proteins, including SMRT and/or NCoR, that maintains the gene in an inactive state. B, In the unliganded state, ER may bind to either corepressor or coactivator complexes. Intracellular signaling can influence the extent of interaction with these complexes and therefore the relative magnitude of basal receptor activity: less activity when bound to corepressor complexes and more activity when the equilibrium is shifted to coactivator complex interaction. C–E, When estrogen (E) activates the receptor, a series of coactivator complexes bind and exchange in a programmed sequence to deliver functions needed to activate the gene (see series of reactions, panels C–E). This arguably involves the sequence of histone acetylation (or other modifications) carried out by histone acetylases (CBP/p300 and SRCs), followed by a complex containing BRG-1/BAF57, which unwinds DNA and remodels the chromatin, followed by a complex involved in initiation of transcription. These early complexes all may include SRC-1 or one of the other members of the SRC-1 family. After initiation, reinitiation/maintenance of transcription is carried out by TRAP220 and the TRAP/DRIP complex of proteins, which, in turn, interact with RNA polymerase II itself. F, Finally, coactivator complexes and the receptor itself are turned over at the promoter by proteasome-dependent processes. The presence of protein complexes containing ubiquitin ligases, such as E6-AP and MDM2, which polyubiquitinate proteins and target them for degradation by the 26S proteasome, have been noted. The turnover leads to down-regulation of receptor/coactivator levels, but this turnover also is required for efficient continued transcription of the gene. DBD, DNA-binding domain; HDAC, histone deacetylase; Ub, ubiquitin.

Several recent reports primarily employing chromatin immunoprecipitation (ChIP) assays suggest that receptor and coregulator association in gene promoters is temporally regulated. For instance, ER appears to cycle off and on the pS2 promoter in certain cells (e.g., MCF-7) in response to continuous E2 stimulation (132). In vitro transcriptional assays indicate that SRC-1 must be recruited to the promoter before p300 for efficient gene expression (133). In addition, a recent report has shown that SRC-1 and the DRIP205/TRAP220 coactivator cycle off and on the pS2 promoter after E2 treatment; notably when SRC-1 is bound, TRAP220 is absent and vice versa (134). Although the rapid kinetics of receptor and coactivators are evident in cellular imaging experiments (95, 135), it is unknown what controls the dynamic association of nuclear receptors and coregulators with target genes, or whether these processes can be regulated in a cell-specific fashion. Nonetheless, the ability of steroid receptors to activate transcription is a product of the ability of the receptor to interact with coactivators and other proteins required for gene expression, and the effect of various enzymatic activities on the formation, function, and disassembly of the receptor-coactivator complex.

D. Corepressors
Although there are far fewer nuclear receptor corepressors, these molecules serve important roles in negatively regulating receptor-dependent gene expression. Unliganded retinoic acid receptor (RAR) and thyroid-hormone receptor (TR) repress basal transcription in the absence of their cognate ligands, and this function is mediated, at least in part, by two, large (270 kDa) nuclear proteins, silencing mediator of retinoic acid and thyroid hormone receptor (SMRT) and nuclear receptor corepressor (NCoR). These corepressors bind to the unliganded receptors via CoRNR boxes, which consist of LxxxI/HIxxxI/L motifs (136, 137, 138, 139). Upon hormone binding, these corepressors dissociate from receptor and enable TR and RAR to associate with coactivator(s) and stimulate gene expression (138). Accordingly, the occupancy of the LBD, and therefore its conformation (140), dictates whether these receptors interact with coactivators or corepressors and activate or repress transcription (141). Although these corepressors do not appear to possess intrinsic repressive activity, they, like coactivators, also function as part of larger protein complexes that include histone deacetylases, which enhance tight nucleosome-DNA interactions and inhibit transcription factor recruitment and gene expression.

Although corepressors bind very well to some nuclear receptors in the absence of their cognate ligands, this is less the case for steroid receptors such as ER, PR, GR, and AR. Rather, corepressors bind to these receptors in the presence of their respective antagonists, 4HT, RU486 (for PR and GR), and cyproterone acetate (119, 142, 143, 144, 145). Both 4HT and raloxifene have been shown to recruit NCoR and SMRT to certain ER target gene promoters (132, 146). The molecular basis of the interactions between steroid receptors and corepressors is not well defined, but a CoRNR box-containing peptide can bind to ERs in the presence of tamoxifen, and mutations within helices 3 and 5 inhibit this interaction (147). Transient transfection assays first demonstrated that both NCoR and SMRT can selectively repress the agonist activity of 4HT and RU486 on ER and PR, respectively (142, 143). Subsequently, it was demonstrated that injecting inhibitory antibodies to NCoR or SMRT promoted the agonist activity of 4HT (119) and that 4HT was a relatively potent ER agonist in fibroblasts derived from NCoR null mice (148). Taken together, the data indicate that SMRT and NCoR are weak ER corepressors, but that these proteins inhibit the agonist potential of antiestrogen-liganded ER activity. Evidence for coregulator inhibition of agonist-bound steroid receptor by the DEAD box RNA helicase DP97 and a novel corepressor, LCoR, also has been obtained (149, 150). Whether corepressor binding represses the ligand-independent activity of these receptors is less clear, but is possible (151). The existence of a cellular equilibrium of coactivators and corepressors that can be shifted toward corepressor preference by antagonist is most likely. Interaction of NCoR and/or SMRT with AR, GR, and PR also has been demonstrated (142, 144, 145, 152, 153).

In addition to NCoR and SMRT, several other molecules have been associated with negative regulation of steroid receptor activity. SAP30 has been shown to be important for NCoR-mediated repression of antagonist bound ER whereas Sharp may play a role in repressing estrogen-induced ER activity indirectly via its effects on the SRA coactivator (154, 155, 156, 157). Corepressor activity also has been shown for MTA (158) receptor-interacting protein (RIP140)/Nrip140 (159, 160), REA (161, 162), RTA (163), and DAX (164).

E. SRM hypothesis
With the identification of coactivators and corepressors, and the biochemical demonstrations that ligands regulate the interactions of receptors with coregulator proteins, it has become possible to more fully consider the role of coregulators in regulating receptor function. Early work demonstrated that overexpression of SRC-1 significantly enhanced 4HT-stimulated ER activity in a cell environment (e.g., HepG2 cells) where 4HT exhibits agonist activity (143). However, exogenous SRC-1 does not robustly increase 4HT agonist activity in all cells, suggesting some component of cell specificity (143). Nonetheless, the ability of SRC-1 to modulate ER-4HT activity in HepG2 cells suggests that productive ER-SRC-1 interactions do occur in some cells in the presence of 4HT. It is possible that in vitro interactions between 4HT-occupied receptor and SRC-1 are generally not observed because studies often fail to consider the contribution of the AF-1 or DNA-binding domains (103). When the corepressors NCoR or SMRT were ectopically expressed in cells, the agonist activity of 4HT was reduced (142, 143, 148), and this is consistent with antiestrogens promoting interactions between ER and corepressors. Collectively, these data suggest that perturbing the expression of coactivators and corepressors within a cell affects the relative agonist and antagonist activity of the SERM, 4HT (see model in Fig. 2). Many studies have followed this line of reasoning, and the activity of a potential corepressor is typically assessed by determining the ability of the candidate molecule to reduce the agonist activity of the SERM, 4HT.

View larger version (24K): [in this window] [in a new window]   FIG. 2. Model of the contribution of coactivators and corepressors to relative SRM agonist/antagonist activity. In the presence of agonist, nuclear receptors in the active conformation interact well with coactivators and are transcriptionally active. In the presence of antagonist, receptors adopt an inactive conformation and preferentially interact with corepressors, resulting in loss of transcriptional activity. In the presence of SRMs, nuclear receptors adopt a conformation intermediate between the active and inactive states and therefore have the potential to interact with either coactivators or corepressors and exert partial activity. The activity of SRM-occupied receptors depends on the relative expression of coactivators and corepressors in a given cell environment and the effect of cell signaling on coregulator subcellular localization and/or activity.

	IV. Molecular Basis of Cellular Selectivity

Top Abstract I. Introduction II. Biology of Selective... III. Mechanisms of SRM... IV. Molecular Basis of... V. Lessons Learned from... VI. Concluding Remarks References

 
A. Receptor-selective recruitment of coactivators
With the possible exception of SRM-induced action emanating from the plasma membrane and/or cytoplasm (e.g., see Refs. 168 and 169), the ability of SRMs to regulate gene expression is dependent on their binding to an appropriate receptor. In the case of SERMs, SPRMs, and SGRMs, there are, however, two distinct receptors to which ligands can bind. In the case of estrogens, the identification of a second receptor gene brought about the realization that estrogen effects are mediated through either ER or ERß. Work from many investigators has characterized functional similarities and differences between these two receptors (170). Although both receptors bind estradiol and SERMS with similar affinity and interact with the same DNA response element, the transcriptional activity of these receptors is distinct. For instance, estradiol generally stimulates greater transcriptional activity via ER than through ERß (61, 171, 172). More pronounced differences are observed in the case of SERM-bound receptors. For ERE-dependent gene expression, tamoxifen is a partial agonist of ER but is generally unable to stimulate ERß transcriptional activity (61, 171, 173). Conversely, when assessing ER activity on AP-1 containing reporter genes, tamoxifen will stimulate ER and ERß transcriptional activity (174).

It has been postulated that differences in the activities between the respective forms of each of these receptors is due to differences in the abilities of the receptors to interact with coregulatory proteins. In support of this, a number of differences in the ability of ER and ERß to interact with coactivators have been noted. For estrogen-bound receptors, ERß, but not ER, binds well to the receptor-interacting component of the mammalian mediator complex, TRAP220 (175). In contrast, the PPAR coactivator-1 (PGC-1)-related coactivator, PERC, coactivates ER, but not ERß, transcriptional activity (176). There are also several AF-1-specific coactivators, such as p68/p72 and MMS19; in the case of the former, they enhance ER, but not ERß, transcriptional activity (177, 178, 179).

There are differences between the relative affinities of ER and ERß for members of the p160 coactivator family. The ER-specific agonists, propyl pyrazole triol and R,R-tetrahydrochrysenes bind to both ER and ERß, but induce interaction of p160 coactivators only with ER, not ERß (180). Other differences in binding are more subtle. Using an approach in which the NR box regions of the SRC family are assessed for their ability to bind to ERs in the presence of estradiol, ER, compared with ERß, was shown to bind with greater affinity to each of these coactivators (181). Real-time interactions between ERs and NR box domains of each of the p160s assessed by BIAcore technology substantiate that result (182). However, this approach also revealed that differences between ERs and full-length SRC family members were more similar with a relative strength of RAC3/SRC-3 greater than SRC-1 greater than TIF2/SRC-2. Differences in binding also can be observed in analyses of the ability of ER and ERß to interact with peptides encompassing each of the individual NR boxes of the p160 coactivator family (183, 184). For instance, NR box IV of SRC-1 binds much better to ERß than to ER, regardless of ligand (e.g., estradiol, diethylstilbestrol, or genistein). In addition, the nature of the ligand influences the relative affinity of the receptor for particular NR boxes; interaction of the estradiol-bound ERß is approximately two times greater than genistein-bound ERß to SRC-2 NR box I, whereas genistein-bound ERß binds approximately five times better to SRC-2 NR box III than the same receptor bound to estradiol. This is also consistent with genistein-bound ERß binding better to p160 coactivators than genistein-bound ER (185).

In the case of progestins and glucocorticoids, there are two hormone-binding receptors for each class of ligand. However, these do not arise from separate genes, but rather from variations in transcriptional start sites and translation initiation (186, 187, 188, 189). The A form of PR and the B form of GR (PR_A and GR_B) represent N-terminal deletions of the larger proteins, PR_B and GR_A, respectively. Regardless of the route in which the different receptors are derived, they have distinct biological activity. In both cases, the activity of the B forms of each receptor is greater than the A receptor isoform (186, 190). Moreover, PR_A can also repress the transcription of other transcription factors, and this extends to other steroid receptors such as ER (190, 191, 192). In the case of PRs, an inhibitor domain has been mapped within the amino terminus of PR_A (193) that is required for the ability of PR_A to bind with greater affinity than PR_B to the corepressor SMRT (194). In contrast, PR_B binds better to NR box peptides of SRC-1 and GRIP1, and this undoubtedly contributes to this receptor’s greater transcriptional activity.

Another example of the specificity of nuclear receptor interactions with coactivators comes from a study examining the requirement of specific p160 coactivators for activation of an integrated chromosomal reporter gene, MMTV-CAT, by GR and PRs and their respective ligands (195). ChIP assays revealed that both receptors recruited SRC-3 to the target gene. In addition, PR recruited SRC-1 and CBP, and this was associated with acetylation of histone H4 on Lys⁵. In contrast, ligand activation of GR led to recruitment of SRC-2 and pCAF followed by Ser¹⁰ phosphorylation and Lys¹⁴ acetylation of histone H3. Thus, even for identical cell and promoter contexts, closely related receptors can utilize different complements of coactivators in the process of activating gene expression. Differences in the interactions of PPAR with p160s also have been observed (44). In the presence of rosiglitazone, TIF2 interaction with this receptor is greater than for SRC-1, whereas in the presence of the selective PPAR modulator, FMOC-L-leucine, which activates insulin-sensitizing but not adipogenesis pathways, PPAR preferentially interacts with SRC-1. Taken together, these results suggest that p160s may substitute for one another to the extent that they promote overall target gene expression, but that differences in downstream events (e.g., histone modification or biological responses) are associated with different coactivator usage.

B. Influence of DNA on coregulator interaction
Although consensus DNA response element sequences have been defined for members of the steroid receptor superfamily, it is clear that not all target genes contain the ideal sequence required to mediate receptor-DNA interactions. For instance, for the 38 estrogen-responsive genes reviewed by Klinge (196), most of the functional EREs located within the promoters or 3'-untranslated regions are not the traditional consensus sequence. Thus, many target genes contain response elements that bear little similarity to consensus EREs. It has been demonstrated that the sequence of the response element affects the affinity that a given receptor has for binding DNA. As might be expected, ER binds with the greatest affinity to the consensus ERE sequence found within the vitellogenin A2 gene and less well to the imperfect EREs found within the vitellogenin B1, pS2, and oxytocin genes (197). This explains, at least in part, how the sequence of the response element can be one important determinant of the extent to which ERs can activate gene expression (197, 198, 199, 200).

However, the conformation of transcription factors can be altered through binding to DNA (reviewed in Ref. 201). The specific sequence of the receptor response elements for ERs and TR can exert distinct allosteric effects on the conformation of ER, ERß, and TR. This has been shown by experiments in which DNA-bound ERs or TR are subjected to limited proteolysis (197, 202) as well as experiments in which LXXLL-containing peptides sensitive to the conformation of the ER or ERß vary in their relative ability to bind to receptor depending on the nature of the ERE (203). Just as ligand-induced changes in receptor conformation influence receptor interactions with coactivators, consensus and imperfect EREs also influence the relative ability of ERs to bind to coactivators. Using an approach based on the relative ability of ERs bound to various response elements to bind to coactivators in HeLa cell nuclear extracts, it has been shown that TIF2 interacts better with ER bound to EREs from the vitellogenin A2, pS2, or oxytocin genes than from the vitellogenin B1 gene (198). In contrast, the response element sequence did not affect ER interaction with AIB1 (198). A similar theme of ERE sequence affecting ERß interaction with coactivators in U2OS nuclear extracts also has been documented (199). However, in this case, interaction between ERß bound to EREs from either the pS2 or vitellogenin B1 genes bound to TIF2 and AIB1 less well than ERß bound to the vitellogenin A2 gene (199). This also correlated with the inability of overexpressed AIB1 to enhance ERß-mediated transcription of synthetic target genes containing either a pS2 or vitellogenin B1 ERE.

In addition to the nature of the steroid response element itself, the context in which the response element resides also determines the ability of steroid receptors to interact with coactivators. The estrogen responsiveness of the pS2 gene is dependent upon both an ERE and an AP-1 site located adjacent to one another; mutation of either one of these sites significantly compromises, but does not block, induction of target gene expression by estrogens (204, 205). It is known that SRC-1 and TIF2 both can coactivate pS2 gene expression (204, 206). However, SRC-1 coactivation of estrogen-stimulated pS2 expression is more dependent on a functional AP-1 site than is coactivation by TIF2 (204). The difference in the relative ability of SRC-1 and TIF2 to coactivate pS2 expression is dependent on the sequence of the ERE; substitution of the pS2 ERE with a consensus ERE enables SRC-1 to coactivate gene expression regardless of whether or not the AP-1 site is mutated.

It should be noted that many estrogen-responsive genes do not appear to contain functional ERE sequences, and the ability of ERs to regulate gene expression is achieved via indirect tethering of the receptor to DNA via other transcription factors, such as AP-1 and Sp1 (reviewed in Refs. 80 and 81). ChIP experiments have demonstrated distinct patterns of tamoxifen-induced association of the promotors of the c-Myc and cathepsin D genes with coactivators and corepressors (146). In Ishikawa cells, tamoxifen stimulates the expression of c-Myc but not cathepsin D mRNA; this result correlates with the recruitment of coactivators to the c-Myc promoter and corepressors to the cathepsin D promoter. Intriguingly, ER regulation of c-Myc expression is dependent upon a discrete non-ERE-containing site (207), whereas ER regulation of cathepsin D is directly mediated via an ERE (208). Although not formally proven, these data raise the interesting hypothesis that tamoxifen cell specificity may be influenced by the mechanism by which ER is tethered to the promoters of target genes, and therefore the ability of the receptor to recruit coactivators or corepressors. It is interesting to note in this study that raloxifene recruited corepressors to both target gene promoters, thus clearly distinguishing itself from tamoxifen.

C. Effect of cell signaling on receptor-coregulator interactions
A number of cellular signaling pathways influence the ability of coregulators to exert their effects on nuclear receptor-dependent gene expression. A summary of enzymatic, protein-protein, and regulatory effects for a selected group of coactivators is located in Table 1. More details are presented in the following sections.

2. Coregulator phosphorylation.
Although there is a long history of steroid receptor phosphorylation and the effect of this posttranslation modification on receptor function (209, 210), it has been recognized only recently that both coactivators and corepressors are also substrates for kinases (Fig. 3). For instance, SRC-1 is a phosphoprotein in which seven phosphorylation sites have been identified (216), two of which (Thr¹¹⁷⁹ and Ser¹¹⁸⁵) can be phosphorylated in vitro by the MAPK, Erk2 (216). Although treatment of cells with a steroid receptor ligand, progesterone, does not increase SRC-1 phosphorylation, cAMP treatment of cells does; however, this probably occurs as a result of cAMP stimulating the activity of the Erk1/2 kinases (217). SRC-1 phosphorylation contributes to its ability to coactivate cAMP- or progesterone-induced PR transcriptional activity, as evidenced by reduced coactivation by a SRC-1 phosphorylation mutant and the ability of the MAPK kinase (MEK)1/2 inhibitor U0126 to reduce SRC-1 coactivation (217). This is not due to any effect on the intrinsic transcriptional activity of SRC-1, or the ability of this coactivator to bind to PR or CBP, but rather a reduction in SRC-1 interaction with the pCAF coactivator, and a loss of SRC-1 functional cooperation with CBP has been observed (217). Phosphorylation of SRC-1 is also important for its coactivation of AR ligand independently activated via IL-6 signaling (218). In contrast, mutation of all seven known SRC-1 phosphorylation sites does not specifically inhibit coactivation of ER function stimulated by either E2- or cAMP-signaling pathway; rather this results in a general reduction in coactivation function (219).

View larger version (20K): [in this window] [in a new window]   FIG. 3. Effect of phosphorylation events on coregulators. Signaling pathways activated by steroids through nongenomic signaling pathways (e.g., src kinase or novel G protein-coupled receptors). TNF or growth factors via their receptors, or elevation of intracellular cAMP (via stimulation of cells with neurotransmitters or pharmacological agents) can communicate with coactivators and corepressors, resulting in the phosphorylation of coregulators in the cytoplasm or nucleus. In the case of the p160 coactivator, SRC-3 (schematic of structural and functional domains is given, see color key), phosphorylation (P) takes place in the cytoplasm and is associated with the translocation of SRC-3 from the cytoplasm to the nucleus. bHLH, Basic helix-loop-helix; HAT, histone acetyltransferase; PAS, Per/Arnt/Sim domain.

Phosphorylation of coactivators also may affect their ability to interact with steroid receptors. For instance, TRBP/ASC-2/RAP250/AIB3/PRIP is a coactivator that in addition to binding to nuclear receptors via a single LXXLL motif (224) associates with CBP/p300, the DRIP130 component of DRIP/TRAP complexes, the RNA recognition motif-containing coactivator CoAA, DNA-dependent protein kinase, and poly(ADP-ribose) polymerase complexes (224, 225, 226, 227). Although wild-type TRBP binds well to both ER and ERß, mutational analyses of the region surrounding the LXXLL motif within TRBP revealed that the amino acid located at the -3 position relative to the NR box could influence the relative binding of TRBP mutants to ER vs. ERß (228). Moreover, in vitro experiments demonstrate that MAPK phosphorylation of the wild-type -3 position amino acid, Ser⁸⁸⁴, inhibits the interaction between a TRBP peptide and both ER and ERß, thereby suggesting that phosphorylation of TRBP may be a negative regulator of its coactivator function (228). Likewise, phosphorylation of the AR coactivator, ARA55, by the proline-rich tyrosine kinase 2 (Pyk2) reduces ARA55 coactivation of AR-dependent gene expression (229). This appears to be related, at least in part, to a reduction in interaction between AR and ARA55.

Finally, phosphorylation of CBP/p300 may also affect the activity of these cointegrators of nuclear receptor function. For example, phosphorylation of p300 on Ser⁸⁹ by PKC reduces its coactivation ability as well as its histone acetyltransferase activity measured on histones in vitro (230). Moreover, phosphorylation of this same serine residue by the AMP-activated protein kinase reduces interaction between p300 and the nuclear receptors PPAR, TR, RAR, and retinoid X receptor (231). Thus, phosphorylation of Ser⁸⁹ by either of these two kinases is a negative regulatory event that may become important during fluctuations in cellular metabolism reflected by elevated AMP (for AMP kinase) or during differentiation or apoptosis (for PKC). Conversely, phosphorylation of CBP or p300 by members of the MAPK signaling pathway, p42/p44 and MAPK kinase kinase 1, positively regulates the activity of these coactivators (232, 233, 234). In addition, cAMP can also increase the intrinsic transcriptional activity of CBP (233).

Just as phosphorylation can regulate the ability of coactivators to affect nuclear receptor activity, by either affecting coactivator function or the ability to interact with receptor, so too can phosphorylation influence the interaction of corepressors with nuclear receptors. The SMRT corepressor may be phosphorylated by the MEK1 kinase, MAPK kinase kinase-1 (as well as MEK-1) resulting in loss of repression due to reduction in the interaction between SMRT and TR, which is accompanied by nuclear export of SMRT to the perinuclear or cytoplasmic compartments (235). Phosphorylation by the p38 MAPK also increases the activity of the PGC-1 coactivator, and this appears to be the result of a phosphorylation event that impairs the functional interaction between PGC-1 and a putative repressor of this coactivator’s activity; estrogen-related receptor- has been shown to possess this activity (236, 237, 238). A nuclear to cytoplasmic shift in the localization of the corepressor RIP140 by the 14–3-3 protein and by cytoplasmic sequestration of SMRT and NCoR by p65 and IB also may affect gene expression positively by reducing levels of corepressors in their functional compartments of interest (239, 240).

3. Regulation of coregulator function by ubiquitinylation or sumolyation.
Steroid receptors, such as ER, PR, PPAR, and GR can be down-regulated by their cognate ligands in cells. The ability of ligands to induce polyubiquitination of these steroid receptors and the ability of inhibitors of the 26S proteasome such as MG132 and lactacystin to block ligand-dependent degradation of these steroid receptors argues that ligand-dependent degradation of these receptors occurs via the 26S proteasome (131, 241, 242, 243, 244). The 26S proteasome is a multiprotein entity, which possesses protease activity that ultimately leads to the cleavage and degradation of polyubiquitinated target proteins (see Refs. 245 and 246 for review). In seeming contradiction to the ability of proteasome inhibitors to increase receptor levels, they also block the transcriptional activity of ER and a number of other members of the nuclear receptor superfamily including progesterone, T₃, ARs, and RARs (131, 247, 248). There is not a simple correlation between the ability of ligands to down-regulate nuclear receptor expression and the requirement of proteasome activity for receptor-dependent gene expression. Notably, the transcriptional activity of the human GR, which is a ligand-dependent target of the 26S proteasome, is not compromised by proteasome inhibitors such as MG132 or lactacystin (243, 249, 250, 251). In contrast, the transcriptional activity of the AR is blocked by proteasome inhibitors, even though ligand stabilizes this molecule instead of inducing its down-regulation (248). Thus, the mechanisms by which proteasome inhibitors block transcription are not defined. This inhibition may signify that receptor turnover is required for its efficient transcriptional activity and/or that proteasome activity is required for other aspects of the transcription process. In this regard, MG132 blocks association of the phosphorylated RNA polymerase II with the pS2 gene promoter in MCF-7 cells, and also reduces the frequency with which ER