Progesterone Receptors, their Isoforms and Progesterone Regulated Transcription

Abstract

This review discusses mechanisms by which progesterone receptors (PR) regulate transcription. We examine available data in different species and tissues regarding: 1) regulation of PR levels; and 2) expression profiling of progestin-regulated genes by total PRs, or their PRA and PRB isoforms. 3) We address current views about the composition of progesterone response elements, and postulate that PR monomers acting through “half-site” elements are common, entailing cooperativity with neighboring DNA-bound transcription factors. 4) We summarize transcription data for multiple progestin-regulated promoters as directed by total PR, or PRA vs. PRB. We conclude that current models and methods used to study PR function are problematical, and recommend that future work employ cells and receptors appropriate to the species, focusing on analyses of the effects of endogenous receptors targeting endogenous genes in native chromatin.

I. INTRODUCTION

Like all nuclear receptors, progesterone receptors (PR) are transcription factors that consist of a DNA binding domain (DBD), sandwiched between an upstream N-terminal region that contains activation (AF) and inhibitory (IF) functions, and a downstream hinge region and C-terminal ligand binding domain (LBD) (Hovland et al, 1998). There are two PR isoforms, PRA and PRB, which differ only in that human PRB contain an additional 164 amino acid far N-terminal region called the “B-upstream segment” (BUS) that confers AF3 activity. BUS is missing in PRA (Sartorius et al, 1994b). Site-specific mutations of amino acids in BUS that are responsible for its AF3 activity destroy PRB-specific gene regulation without however, switching PRB to PRA (Tung et al, 2006). This suggests that global structural differences between PRB and PRA apart from BUS control their unique properties. We review below the distinctive transcriptional activities of the two PR isoforms.

In normal human tissues including the breast, PRA and PRB are generally expressed at similar levels but at least in some breast cancers, their ratio is dysregulated (Graham et al, 1995b; Hopp et al, 2004). Classically, transcription mediated by PR is viewed as follows: 1) The unliganded receptors are cytoplasmic and bound to heat shock proteins in the absence of progestins (P). 2) Liganded receptors are released from heat shock proteins, dimerize and translocate to the nucleus. 3) There they locate specific palindromic DNA binding sites called progesterone response elements (PREs) in promoters of PR regulated genes, after which, 4) transcription is initiated through recruitment of a transcription complex. While this may be the case for some tissues, in some contexts, on some promoters, this simple model cannot explain data showing that: 1) PR are largely localized to the nucleus even in the absence of P; 2) dimerization may not be required for PR mediated transcription; 3) the endogenous structures of PREs have not been defined by unbiased in vivo analyses but are unlikely to be “classical”; and 4) PR also signal through cell membrane and cytoplasmic pathways.

The functions of steroid receptors generally, and of PR specifically, have been studied in a variety of systems. For PR, the most widely used are artificial reporters consisting of tandem palindromic PREs separated by random intervening sequences, linked to luciferase. These double PRE (PRE2) reporters are then transiently or stably transfected into cells that may or may not be relevant to the question under study; conditions and cells are optimized to yield maximum luciferase activity; and conclusions are drawn regarding PR structure, PR function, PR isoform specificity, and coregulators of PR-dependent transcription. In an attempt to address more physiologically relevant conditions, other studies choose bona fide genes regulated by PR, map their proximal promoters (usually less than 5 Kb) by deletion analysis, and search for classical PREs within regions considered to be functionally important. To confirm the validity of any conclusions, candidate sequence(s) are subjected to protein:DNA interaction studies by electrophoretic mobility shift assays, DNA footprinting, or methylation interference; all of which use “naked” DNA. More recently, nuclear receptor binding sites and kinetics of protein:DNA interactions have been analyzed by chromatin immunoprecipitation (ChIP), which has the advantage of searching for PR binding sites in the context of chromatin. However, even ChIP assays are biased by being limited to a search of specified DNA regions or sequences. The least biased methods are ChIP-on-chip, which combines ChIP with microarray technology (chip) and identifies DNA-binding sites on a genome-wide basis; or ChIP-seq, which combines ChIP with massively parallel DNA sequencing. Studies examining PR-regulated promoters have, as of this writing, been limited to proximal promoter regions using ChIP assays. Neither ChIP-on-chip nor ChIP-seq has been reported. Thus many important PR regulatory regions including ones in introns, 3′ untranslated regions (UTR), or tens of thousands of Kb removed from transcriptional start sites remain unexamined, and true endogenous DNA sequences to which PR bind either directly or indirectly remain largely unknown.

In this review we examine available data in a variety of tissues and species addressing: 1) regulation of PR levels by estrogenic and non-estrogenic signaling; 2) analysis of genes regulated by total PR, or by PRA vs. PRB using expression profiling; 3) our current understanding about the composition of PREs, and transcriptional regulation via PR monomers, PRE half-sites and cooperativity with other transcription factors. 4) We review in detail transcriptional regulation of multiple P-regulated promoters by total PR and the two PR isoforms. Although many signaling pathways converge on PR and influence their transcriptional activity (reviewed in (Daniel et al, 2009)), even in the absence of ligands (Jacobsen et al, 2005), and non-genomic effects of PR have also been described (Boonyaratanakornkit et al, 2001), this review is limited to the genomic transcriptional effects of liganded PR.

II. REGULATION OF PR EXPRESSION

To understand PR function, it is important to review factors that regulate PR levels. In T47D human breast cancer cells, which are the major models for human PR, synthesis of the PRA and PRB isoforms is driven by transcription from two promoters (Kastner et al, 1990) located at −711 to +31 and +464 to +1105 of the transcription start-site, respectively. At least 6 transcripts encode PR (Kastner et al, 1990; Wei et al, 1988) containing translational start sites not only for PRA and PRB, but also for at least one N-terminally truncated protein called PRC (Condon et al, 2006; Wei et al, 1990). The majority of normal human P target tissues that have been studied express PRA and PRB in equimolar amounts (Graham & Clarke, 1997) generating 3 theoretical receptor populations consisting of 25% PRA homodimers, 25% PRB homodimers and 50% PRA/PRB heterodimers. Fluorescence resonance energy transfer microscopy, which quantifies the separation distance between two molecules within nanometers, shows the presence of two closely neighboring PR molecules, which could be monomers or dimers, in transcriptionally active foci of P-treated living cells (Arnett-Mansfield et al, 2007; Mote et al, 2007).

Unlike normal human tissues, human malignancies frequently display aberrant PRA:PRB ratios (reviewed in (Mote et al, 2007)). Additionally, non-human but nevertheless important mammalian models used for cancer studies often have isoform ratios that differ from those in human tissues. Typically the ratio of PRA:PRB in normal mature tissues of mice and rats is 3:1 (Ilenchuk & Walters, 1987; Schneider et al, 1991), suggesting that the unknown factors regulating equimolarity in humans differ substantially in rodents. Additionally, this raises questions about the importance or lack thereof of equimolarity, and, if equimolarity is important, about the use of rodent models to study human progesterone physiology and disease processes.

a. E-dependent regulation of PR

The major factors that regulate PR levels in a variety of cells and tissues are 17β-estradiol or related estrogens (E) bound to estrogen receptors (ER). PR are E regulated in the uterus including the endometrium, in the pituitary and brain, and in the mammary gland/breast and other tissues (reviewed in (Graham & Clarke, 1997)). Thus it is rare that normal PR-positive cells do not also express ER. Indeed to our knowledge, discordance between PR and ER expression has only been reported in cancers.

In human breast cancer cells, the proximal promoters controlling PRA and PRB transcription contain estrogen response elements (EREs) recognized by ER, plus binding sites for other transcription factors with which ER interact (Horwitz et al, 1978; Kastner et al, 1990). For example, in MCF7 cells, liganded ER interact with an ERE ½ site within the l Kb PR promoter region (Petz et al, 2004a) that also contains SP1 sites (Schultz et al, 2003), and two different Fos/Jun AP1 binding sites, one of which functions as an activator, one as a repressor of ER regulated transcription (Petz et al, 2002; Petz et al, 2004b). ER-dependent upregulation of PR also likely involves long range transcriptional mechanisms as the region 311 Kb upstream and 4 Kb downstream of the PRB transcription start site contains 8 EREs that bind ERα (Boney-Montoya et al, 2010). These variable regulatory elements suggest possibilities for tissue and PR isoform-specificity. While the PR isoform-specificity controlled by ER via these sites has not been examined, E more strongly induces PRB than PRA in T47D (Graham et al, 1995a; Vienonen et al, 2002) and ZR-75-1 human breast cancer cells (Vienonen et al, 2002), but more strongly induces PRA in MCF7 breast cancer cells (Vienonen et al, 2002). GATA3, often expressed in association with ER, also plays a role in E mediated induction of PR transcripts (Eeckhoute et al, 2007). However, questions have been raised about E regulation of PR in normal breast epithelium, since total PR levels do not vary during the hormone fluctuations associated with the menstrual cycle (Mote et al, 2002). Other signaling pathways regulating PR include retinoic acid receptors (Clarke et al, 1991; Clarke et al, 1990) and P themselves, which promote PR downregulation (Alexander et al, 1989; Ghatge et al, 2005) and block E-dependent PR upregulation (Graham et al, 1995a). Methylation of the PR promoter correlates with silencing of PR expression in MCF7 cells (Xu et al, 2004).

As with human PR, rat PR expression is driven by two functionally distinct promoters, one proximal (+461/+675) controlling PRA, and one distal (−131/+65) controlling PRB (Kraus et al, 1993). Rat and human PR promoters are ~60% homologous (Kraus et al, 1993), and show important differences in their regulation. In contrast to the two human promoters, both of which are E inducible in breast cancer cells (Kastner et al, 1990), only the proximal PRA rat promoter is E inducible in MCF7 cells. When transfected into primary rat uterine cells, the rat proximal promoter is not E regulated unless 1 additional exogenous ERE is introduced, and the distal promoter requires 2 additional exogenous EREs. In mouse pituitary 3T3 cells E induction of the rat promoters requires 2 additional EREs each. Examination of the rat promoters show the presence of 4 imperfect EREs in the 5′ flanking region and within the first exon, and ERE ½ sites are located within 150 bp of 3 of these 4 EREs, implicating ERE ½ sites in conjunction with imperfect EREs in regulating rat PR levels (Kraus et al, 1994).

Expression of mouse PR has been examined in vivo during mammary gland development using the LacZ gene encoding β-galactosidase knocked into the PR locus on one allele. While PR are highly and uniformly expressed in terminal end buds of 5 and 8 week-old intact virgin mice, in 16 week-old sexually mature cycling mice, PR expression is variegated, analogous to the effects of E + P treatment in ovariectomized mice (Ismail et al, 2002). Of note is that only PRB expression was monitored in this study since the LacZ knockin disrupted the translational start site for PRA. Other studies (Aupperlee & Haslam, 2007; Aupperlee et al, 2005; and Kariagina et al, 2007) raise doubt that E regulates PRB in rodents. Rabbits are seemingly even more different than the above species since they express only PRB driven by a single promoter that lacks an ERE in the 5′ flanking region, but contains an ERE that overlaps with the translational start site (Savouret et al, 1991). This ERE also mediates P-dependent PR downregulation (Savouret et al, 1991).

This brief summary highlights the species-specific effects of cell types and promoters in controlling PR expression. One can only conclude that results from most studies are not generalizable, and that at the very least, studies intended to address human biology need to employ human PR in human cells.

b. Non E-dependent regulation of PR

Estrogens do not control PR expression in all tissues. For example, E do not induce PR transcripts in primary rat granulosa cells; rather PR levels are regulated by cAMP and gonadotropins (Park-Sarge & Mayo, 1994). Similarly, PR are not directly regulated by E in rat preovulatory granulosa cells (Clemens et al, 1998) but are instead induced by FSH, LH and forskolin (all of which increase cAMP); an effect mediated in part by SP1/SP3 binding sites in the proximal PR promoter (Sriraman et al, 2003). Even in human breast cancer cells, in which E regulation of PR has been documented for at least 3 decades and forms the basis for clinical assays to predict hormone responsiveness, factors other than ER play a role. For example, in MCF7 cells, activators of protein kinase A (PKA), stimulate cAMP production, which increase PR protein levels (Cho et al, 1994). Interestingly PKA activators increase only PR protein not PR transcripts, while E increase both (Cho et al, 1994) suggesting that the cAMP targets are post-translational. The effects of cAMP are inefficient, raising PRB to 8% and PRA to 51% of the levels raised by E (Kraus et al, 1993). However, these results suggest possible mechanisms for differential PR isoform regulation in mice, rats and human cancer cells.

Many studies addressing regulation of human PR levels use cultured cell lines, and it is therefore of considerable importance that serum factors modify the effects observed. Sera-reduced media increase PR, suggesting that serum growth factors are inhibitory. Indeed, both IGF1 and EGF inhibit PR levels in MCF7, ZR-75-1 and T47D cells through pathways that involve PI3K/AKT/mTOR signaling (Cui et al, 2003b). Similarly in endometrial cancer cells, both IGF1 and IGF2 inhibit PR mRNA expression, while PR levels are increased by the anti-diabetic drug Metformin, which decreases cellular cAMP production, reduces the activities of PKA and MAPK1/3, and inhibits mTOR (Xie et al, 2010). Since in MCF7 cells, liganded PRB in turn upregulate growth factors such as IRS-2, which primes cells for activation by IGF1 (Cui et al, 2003a), this sets up a negative feedback loop in which PRs, through varied signaling pathways, suppress their own biosynthesis.

Transcription factors besides ER also regulate PR gene expression. CEBPβ is repressive (Seagroves et al, 2000) while P-occupied PR increase CEBPβ in human breast cancer cells (Richer et al, 2002) and mouse mammary glands (Aupperlee et al, 2009), setting up another possible negative regulatory loop. However, P-dependent regulation of CEBPβ is mouse strain specific (Aupperlee et al, 2009), which makes data interpretation difficult. Another transcription factor involved in regulating the PR gene, apparently specific to the human PRB promoter, is GATA5, which binds adjacent to a polymorphism (+331G/A) in the promoter to upregulate transcript levels of this isoform only (Huggins et al, 2006). This clearly is another mechanism by which PR isoform levels can be dysregulated in malignant cells.

Cyclin D1 enhances PR levels via an enhancer that binds both ER and Cyclin D1 in the 3′-UTR of the PR gene (Yang et al, 2010). However, a single pulse of liganded PR suppresses cyclin levels including those of Cyclin D1 (Groshong et al, 1997) apparently through PR and AP1 regulatory elements in the distal Cyclin D1 promoter (Cicatiello et al, 2004). A second P dose exacerbates this effect; another example of a P-dependent autoinhibitory loop. The fact that multiple examples exist for negative autoregulation of PR suggest that it is physiologically important that expression levels of these receptors be tightly controlled.

Lastly, the above summary details examples of factors that regulate overall PR protein levels. However, activity of the receptor proteins is controlled not only by their levels, but by post-translational modifications including phosphorylation, sumoylation, acetylation and ubiquitination, the last of which targets PR to proteasomes for degradation (Abdel-Hafiz et al, 2002; Daniel et al, 2010; Lange et al, 2000). These modifications exert profound effects on the functional activity of PR. For example, desumoylation increases PR activity ~10-fold (Abdel-Hafiz et al, 2002). Paradoxically, ligand dependent PR downregulation following ubiquitination of the receptors is required for, and coupled to, maximal PR transcriptional activity (Lange et al, 2000; Shen et al, 2001). PR activity is also modified by growth factor signaling (Lange et al, 1998; Richer et al, 1998).

II. GENE REGULATION: EXPRESSION PROFILING

In the last few years, expression profiling studies have greatly expanded our knowledge of human P-regulated genes and the role, if any, of PR homo- and heterodimers. These studies indicate that isoform-specific gene regulation by PR is largely tissue, cell type and promoter specific. As a result, since most studies use cancer cells, conclusions may not apply to normal, physiological states. PR gene regulation has been studied mainly in ER+ T47D human breast cancer cells because, typical of luminal breast malignancies, these cells express high PR levels (Ghatge et al, 2005; Graham et al, 2005; Hopp et al, 2004; Jacobsen et al, 2005; Richer et al, 2002). Additionally their PR are extremely stable and not subject to fluctuation in response to transient shifts in E levels commonly seen in other cells. Therefore P-regulation can be studied without the confounding effects of having to add E in order to induce or maintain PR. Cloning of a PR-negative T47D subline called T47D-Y allowed restoration of either PRA (T47D-YA cells) or PRB (T47D-YB cells) for study of each isoform in isolation (Jacobsen et al, 2002; Sartorius et al, 1994a). These models are important because they were constructed in the background of a naturally ER+PR+ cell line and would be expected to contain appropriate ancillary coregulatory factors needed for faithful PR-dependent gene regulation. ER and PR-negative cells like MDA-MB-231 transfected with PR may lack this competence.

In the T47D models, PRA and PRB regulate some overlapping genes but most are isoform specific. An early study using incomplete gene chips described 94 P-regulated genes: 25 regulated by both PR, 64 by PRB only, 4 by PRA only (Richer et al, 2002), suggesting at least some differences between the two PR isoforms in sequence recognition and/or interactions with other transcription factors. A subsequent study with larger chips defined 337 P-regulated genes: 25 by both PR, 229 by PRB and 83 by PRA (Tung et al, 2006). Thus the trends in both studies were similar with substantially more genes uniquely regulated by liganded PRB than PRA, and relatively few by both isoforms. When both isoforms were regulatory, PRB were usually stronger transactivators than PRA (Jacobsen et al, 2005; Richer et al, 2002). This differs from the unliganded receptors where PRA are the stronger transactivators (Jacobsen et al, 2005). The above studies would seem to demonstrate that both heterodimers and homodimers are capable of gene regulation. A role for heterodimers has not however, been formally proven since in the T47D models homodimers are always in the mix.

Expression profiling of P-regulated genes in ER and PR-negative MDA-MB-231 cells transfected with both PR isoforms has also been reported (Leo et al, 2005), as have P studies in estrogenized MCF7 cells (Leo et al, 2005; Purmonen et al, 2008). Comparisons of genes among these studies are complicated by differing experimental variables, assorted time-points of P treatment, different genomic platforms, and natural vs. synthetic P usage. Synthetic progestins are largely similar to progesterone in terms of genes regulated in breast cancer cells (Bray et al, 2005; Ghatge et al, 2005), the exception being medroxyprogesterone acetate (MPA), which is a potent mixed progestin/androgen, and in cells that contain both PR and androgen receptors (AR), signals through both receptors (Ghatge et al, 2005).

Expression profiling has also been performed in T47D cells in which the PRA:PRB ratio can be altered by allowing PRA to fluctuate from 1:1 to 5:1 (Graham et al, 2005; McGowan & Clarke, 1999). After short P exposure (6 hrs) neither increase of the PRA:PRB ratio, nor the heightened PR levels generated by PRA overexpression, impacted the gene profiles. Interestingly, many of the early genes were transcription factors. At 48 hrs, 6-fold more genes and different functional pathways were regulated. The authors postulate that these late genes are indirect targets. However, the major lesson of this study was that the isoform ratio did not appear to be important at either timepoint. This and other studies also failed to demonstrate the dominant negative effects of PRA on PRB-dependent transcription (a phenomenon previously reported using exogenous reporters) unless PRA exceeded PRB by >15-fold (Graham et al, 2005; McGowan & Clarke, 1999). Such discordant ratios are rarely, if ever, observed in natural cells.

There is little overlap among P-regulated genes in normal human breast explants vs. T47D breast cancer cells grown as organoids (Graham et al, 2009), or in PRA+ mouse mammary organoids vs. PRA+ breast cancer cells (Santos et al, 2009). This emphasizes again how difficult it is to generalize among different models. Other cell types in which P-regulated genes have been analyzed include normal human breast epithelium (Graham et al, 2009) and the normal mouse mammary gland (Fernandez-Valdivia et al, 2008). These genesets have not been compared to each other or to other published datasets. Recall that differences exist in PR isoform ratios between rodent and human tissues. Overall, much remains to be learned about the endogenous genes regulated by progesterone and synthetic progestins in human PR target tissues, and the accuracy of rodents to model human physiology and cancer biology remains to be validated.

IV. COMPOSITION OF PREs

The activity of PR has been extensively studied on 2 exogenous models: the mouse mammary tumor virus-long terminal repeat (MMTV-LTR) and tandem PREs (PRE2) derived from the second PRE of the rat tyrosine amino transferase (TAT) promoter. These sequences, which usually respond to both PR and GR (and often to AR and mineralocorticoid receptors as well), are often referred to as GRE/PREs. It is not surprising that GR and PR recognize the same DNA elements since their core DBDs exhibit 90% amino acid sequence identity (Takimoto et al, 2003).

The MMTV-LTR was first demonstrated to be GR regulated and subsequently to be PR regulated. It contains one imperfect palindromic PRE preferentially used by GR, and 3 proximal PRE ½ sites preferentially used by PR (Gowland & Buetti, 1989). Overall, PR occupy a larger DNA footprint (Chalepakis et al, 1988) and interact with different nucleotide residues (von der Ahe et al, 1985), than do GR. The three PRE ½ sites are necessary for PR induction of MMTV-LTR/luciferase in human T47D cells (Cato et al, 1986) and can do so even when isolated from the palindrome (Chalepakis et al, 1988; Jacobsen et al, 2009). PRB are stronger transactivators than PRA on the complete LTR (Kastner et al, 1990) and on the 3 PRE ½ sites (Jacobsen et al, 2009). The MMTV-LTR palindrome in combination with 2 of the 3 proximal ½ sites has been used to map the nucleotides that define a PRE using total PR assessed in T47D cells (Lieberman et al, 1993). Point mutations demonstrate that even “suboptimal” sequences lead to maximum transcription, but the consensus sequence arrived at consists of 2 hexamer ½ sites separated by 3 intervening bases: 5′-RGnACAnrnTGTnCY-3′. The rat TAT promoter was also first shown to be GR regulated and subsequently to be PR regulated. It contains 3 palindromic PREs, but the PREII alone – 5′-TGTACAggaTGTTCT-3′ – is PR responsive and used in many exogenous promoter/reporters (Strahle et al, 1987).

So, what is a GRE/PRE? Depending on which “consensus” sequence is used, the identity varies. Besides the above, two other “consensus” GREs are listed in the TRANSFAC database. One – 5′-GGTACAannTGTYCTk-3′ was derived from 10 sequences, two of which are human: 2 from rat TAT, 2 from rat Tryptophan Oxygenase, 1 from human Metallothionin IIA (MetIIA), 2 from murine Sarcoma Virus, and 1 from human Growth Hormone (Jantzen et al, 1987a; Jantzen et al, 1987b). Among these 10, a 3′ hexamer (TGTTCT) is more highly conserved than the 5′ hexamer (Jantzen et al, 1987a; Jantzen et al, 1987b). The second GRE is experimentally derived from 38 GR binding sequences whose identities are unknown. Its sequence is 5′-nnnnnnCnntnTGTNCTnn-3′. It is likely to be the most accurate “PRE”, reflecting true endogenous variability based on functional analyses, including the possibility that a strong PRE ½ site hexamer is sufficient for regulation by PR.

Spacing of PREs, at least in transient transcription assays using synthetic promoters, is reportedly also important for strong PR-mediated transcription. Optimal cooperativity resulting in transcriptional synergism was observed between the imperfect PRE palindrome of MMTV and several transcription factor binding sites including ones for NF1 and SP1 that were 29bp removed from the PRE (Schule et al, 1988). A second palindromic PRE also synergizes with the first if the two are separated by 29 nucleotides. It has been suggested that this reflects nucleosome spacing periodicity. However, as reviewed below, it is rare if ever the case that two palindromic PREs are spaced 29nt apart in endogenous promoters; indeed even a single palindromic PRE is hard to find. Cooperative protein-protein interactions resulting in transcriptional synergy have only been observed on MMTV or exogenous PRE-containing reporter constructs in transient transfection assays and only with PRB. PRA do not exhibit synergy under the same conditions, and if co-expressed with PRB, ER or GR, the PRAs act as transrepressors. However, this may be due to squelching of transcription factors under transient transfection conditions in which the receptors are overexpressed. PRA are not transrepressors of endogenous genes in T47D cells (Graham et al, 2005). Also with regard to spacing issues, the distance between a GRE/PRE and the transcription start-site appears to be constrained in experimental models (Nordeen et al, 1998), but there is little or no evidence for such rigid position dependence of PREs in natural promoters.

As with MMTV and TAT, many other GR regulated genes are also regulated by liganded PR (Bocquel et al, 1989; Renkawitz et al, 1984; Slater et al, 1988; Vegeto et al, 1993), including the chicken lysozyme, rabbit uteroglobin and human MetIIA genes. They have been used to address differences in binding specificities between GR and PR, but often receptors from different species or only one PR isoform were tested. For example, the chicken lysozyme promoter contains 2 palindromic PREs, binding of rat GR and rabbit PRB are the same on the proximal GRE, but PR protect a larger region than GR on the distal GRE, and the two receptors contact different nucleotide residues (von der Ahe et al, 1985). In MetIIA, GR and PRB apparently contact the same residues (Slater et al, 1988), which, when cloned upstream of the tk promoter confer PR inducibility (Scheidereit et al, 1986). The uteroglobin promoter, contains 3 PR binding sites between −2732 and −2627 that generate 3 separate footprints (Jantzen et al, 1987b), but if they are cloned upstream of a heterologous promoter they fail to confer PR inducibility (Jantzen et al, 1987b). PR also bind a site within the first intron of uteroglobin, perhaps explaining the lack of PR activity of the 3 upstream binding sites (Bailly et al, 1986). These data demonstrate how complex the issues of PR regulation are, and point to the weaknesses of exogenous promoter analyses. While PR and GR regulate some of the same endogenous genes, they also regulate distinct gene subsets (Wan & Nordeen, 2002). Endogenous binding sites for GR have been mapped by ChIP-chip and ChIP-seq (John et al, 2011; So et al, 2007). However, authentic endogenous binding sites for PR have yet to be mapped by such analyses, and the true identity of PREs and whether or how they differ from GREs, is unknown.

a. Do PR dimers exist?

Biochemical studies and mutational analyses indicate that residues involved in PR dimerization are present in the LBD, DBD and N-terminus of PR monomers (Tetel et al, 1997) and it has been assumed that monomers assemble in solution to generate a preformed dimer that binds to DNA with high affinity. Bain et al. (Connaghan-Jones et al, 2008) have carried out a thermodynamic dissection of PR and conclude that dimer binding at a palindrome (whether by successive monomer assembly or via a preformed dimer) is accompanied by an enormous energetic penalty. On the other hand, PR monomers are not only capable of binding DNA but also engage in cooperative interactions among PRE 1/2 -sites of varying distances and orientations, which can account for the majority of the binding energetics on a natural promoter. They conclude that endogenously, PR monomers are the major transcriptionally active species. Our studies, outlined below, indicate that PRE ½-sites may be the major DNA elements through which PR regulate transcription; another argument in favor of a role for monomers.

b. PRE ½ sites and PR monomers

Our in silico analysis of PR-regulated promoters for the presence of PREs shows an abundance of imperfect palindromes, direct repeats and PRE ½ sites. On the other hand, palindromic PREs are rare, and are also detected in promoters of non-PR regulated genes, so that their presence is insufficient to define a PR-regulated promoter (Jacobsen et al, 2009). Several lines of evidence suggest that PRE ½ sites may be important PR-regulatory sites: 1) PRE ½ sites are abundant in P-regulated promoters while consensus or perfect palindromes are rare (Jacobsen et al, 2009); 2) transcription can be stimulated by P via 2 or 3 PRE ½ sites (Chalepakis et al, 1988; Jacobsen et al, 2009); 3) PR monomers retain the ability to regulate endogenous genes; indeed mutated PR that are unable to dimerize are better transactivators than wild-type PR (Jacobsen et al, 2009); 4) functional studies of endogenous promoters illustrate the recurring theme that liganded PR act via PRE ½ sites (Brayman et al, 2006; Buser et al, 2007; Hewetson & Chilton, 2003; Salama et al, 2007); 5) PR bind to PREs as monomers, and monomer binding to the MMTV-LTR ½ sites is cooperative (Connaghan-Jones & Bain, 2009; Connaghan-Jones et al, 2008); 6) At molecular protein concentrations found in cells, PR exist as monomers (Connaghan-Jones & Bain, 2009); 7) mobility shift assays using DNA containing PRE ½ sites show PR binding as monomers (Roemer et al, 2006). 8) The transcriptional activity of monomeric PR is higher than of dimeric PR on tandem PRE2, possibly due to cooperativity between monomers (Jacobsen et al, 2009). Conceptually, these findings may also apply to GR monomers, since mice expressing mutations in the dimerization interface of GR are viable, demonstrating that GR monomers retain the capacity to regulate essential genes (Reichardt et al, 1998).

c. Cooperativity, Coregulators and co-Response Elements (coRE)

There is ample evidence that promoter structure confers transcriptional specificity (Maston et al, 2006) and the structure and location of PREs relative to adjacent or overlapping DNA binding sites for other transcription factors undoubtedly do so as well. This is the case for the majority of ER binding sites (Carroll et al, 2005). Whether PRs, like ERs, require a “pioneer factor” (Carroll et al, 2005) that enables their interaction with PREs is unknown. Many sequence-specific DNA binding transcription factors regulated by liganded PR are themselves coregulators of PR-dependent transcription. FOXO1 (Foxo1a) is P-regulated (Ghatge et al, 2005) and a coactivator of PRA-dependent transcription (Rudd et al, 2007; Takano et al, 2007). Other examples include STAT5a (Richer et al, 1998), c-Fos (Musgrove et al, 1991), c-Jun (Alkhalaf & Murphy, 1992), and CEBPβ (Richer et al, 2002), all of which are both P-regulated and capable of interacting with PRE-bound PR. In some cases PREs are not required. Instead, PR are tethered indirectly to DNA-bound factors including AP1 (Bamberger et al, 1996; Owen et al, 1998) and JDP2 (Hill et al, 2009). Occasionally, as is the case for AP1, these are in turn P-regulated (Dai et al, 2003). The fact that gene regulation can be preserved with liganded PRs whose DNA binding domain has been mutated supports a tethering model in some cases (Jacobsen and Horwitz, unpublished). With regard to the PR isoforms, our in silico analyses show no differences in the PRE composition of PRA vs. PRB-regulated promoter (Jacobsen and Horwitz, unpublished) suggesting that the key to PR isoform specificity may lie in the transcription factors that flank PR binding sites, and to differences in binding affinities between these factors and each receptor isoform. While PRA have a higher affinity than PRB for corepressors on an exogenous PRE reporter (Giangrande et al, 2000), whether this is the case on endogenous PR binding sites is unknown. These issues will remain unresolved until the structures of the full-length PR isoforms have been solved, and authentic in vivo PR binding sites have been identified.

Because we found that imperfect inverted repeats, direct repeats, and PRE ½ sites are widespread in the genome, not only in PR-regulated but also in non-PR-regulated and random promoters, we speculated that PREs may be necessary but not sufficient to control endogenous PR-dependent transcription. A search for PRE partners identified a highly conserved 234-nt sequence invariably located within 1–2 Kb of the transcription start sites of P-regulated genes (Jacobsen et al, 2009). It resembles ALU repeats and contains binding sites for 11 transcription factors. Interactions between PR and many of these factors have been documented (Kang et al, 2001; Maccarrone et al, 2003; Owen et al, 1998). We cloned the 234-nt sequence of the PR-regulated 8-oxoguanine DNA glycosylase promoter in the forward or reverse orientation in front of 0, 1, or 2 inverted repeat PREs and one or tandem PRE ½-sites driving luciferase. Under these conditions the 234-nt sequence functions as a co-response element (coRE): from the PREs or tandem ½ -sites the reverse coRE is a strong activator of PR and GR-dependent transcription. The forward coRE is a powerful repressor (Jacobsen et al, 2009). As discussed above, the prevalence of PRE ½ -sites in natural promoters suggested that PR monomers regulate transcription. Indeed, dimerization-domain mutant PR monomers are stronger transactivators than wild-type PR on PREs or tandem ½ -sites. This is repressed by the forward coRE. We propose that in natural promoters the coRE functions as a composite response element adjacent to imperfect PREs and PRE ½ -sites that presents variable, orientation-dependent transcription factors for cooperative interaction with nearby PR. As discussed below, PR rarely if ever regulate endogenous promoters without the cooperativity provided by coregulatory partners.

V. DETAILED PROMOTER ANALYSES AND PR ISOFORMS

The majority of studies addressing transcriptional mechanisms of PR focus on the proximal promoter (usually 5Kb or less from the 5′-UTR) of the regulated gene, cloned upstream of a reporter, and transfected into cells that contain endogenous or transfected PR. Deletion and/or mutation analysis of the targeted promoter mark PR functional sites, which are then analyzed for presence of “consensus” PRE palindromes, and/or for DNA:PR protein interactions. Often these analyses point to non-consensus PREs or PRE ½ sites, or to other transcription factor binding sites. While these studies are informative, several caveats need to be considered (Table 1): 1) Most studies use chimeric experimental models in which the species of the promoter, the PR, and/or the transfected cells, are heterogenous. Since PRE sequences and sequences for other transcription factors, PR amino acid composition, and cellular factors, differ from species to species, the problems with data interpretation are obvious. 2) Transiently transfected proximal promoters lack the complexity of higher order chromatin structure, and generally lack important regulatory regions found in introns, the 3′ UTR, or thousands of Kb removed from the transcription start-site. All of these sites, distant from proximal promoters, are involved in regulation by ER, AR and GR (Carroll et al, 2005; John et al, 2011; Lin et al, 2009) and are likely to be involved in PR-dependent transcription as well. 3) The origin of PR is important. Endogenous PR activate transcription driven by the MMTV-LTR in the context of chromatin, while transiently expressed PR fail to do so (Botos et al, 2004). 4) Many studies use the pGL3/luc vector for promoter construction. This vector is studded with transcription factor binding sites including 46 GRE ½-sites, as well as a palindromic GRE (Dougherty & Sanders, 2005) that would be expected to be PR responsive irrespective of the promoter cloned therein (see below).

Table 1.

Summary of promoters analyzed for regulation by PRA or PRB

Gene Symbol	PRA	PRB	Pattern of regulation PRA/PRB	Species (PR)	Species (promoter)	Species (cells)	PRE^	Reference
Regulated by Total PR
FAAH	endog	endog	up (total)	human	Human	human; T lymphocytes	None; Ikaros binding site	(Maccarrone et al, 2003)
*ITGA6	endog	endog	up (total)	human	Human	human; MCF7	Imperfect palindrome PRE	(Nishida et al, 1997)
*FKBP5	endog	endog	up (total)	human	Human	human; T47D	PRE-like sequences	(Hubler & Scammell, 2004)
LYZ	endog	endog	up (total)	chicken	Chicken	chicken; primary oviduct	hexanucleotide (PRE 1/2 site)	(Renkawitz et al, 1984)
dPRL	exog	exog	up, up PRB > PRA	human	Human	human; primary endometrial stroma	PRE 1/2 site close to C/EBPβ	(Christian et al, 2002)
PRLR	endog	endog	up (total)	human	Mouse	human; T47D	None; AP1, C/EBPβ	(Goldhar et al, 2011)
HSD11B2	exog, endog	exog, endog	up, up PRB > PRA	human	Human	human; T47D- YV	Putative HRE and S5RE	(Subtil-Rodriguez et al, 2008)
*BCL2L1 (Bclxl)	endog	endog	up (total)	human	Mouse	human; T47D	2 imperfect HREs	(Viegas et al, 2004)
MYC	endog	endog	up (total)	human	Human	human; T47D	PRE-like sequence	(Moore et al, 1997)
OGG1	exog	exog	up (total)	human	Human	human; HeLa, T47D	None; coRE binds 11 transcription factors	(Jacobsen et al, 2009)
CDKN1A (p21)	exog	exog	up (total)	human	Human	human; HeLa (for PRA) T47D-YB	None; SP1 binding site	(Owen et al, 1998)
F3	endog	endog	up (total)	human	Human	human; primary endometrial stroma	None; 3 SP1 binding sites	(Krikun et al, 2000)
ADAMTS1	exog	exog	up (total)	human	Mouse	rat; granulosa	None; C/EBPβ, NF1 and 3 SP1 sites	(Doyle et al, 2004)
PAEP (glycodelin)	endog	endog	up (total)	human	Human	human; HEC- 1B	None; SP1 sites	(Gao et al, 2001)
HLA-G	endog	endog	up (total)	human	Human	human; JEG3	Novel PRE	(Yie et al, 2006)
*COMT	endog	endog	down (total)	human	Human	human; T47D and MCF7	3 PRE 1/2 sites	(Salama et al, 2007)
Regulated by PRB only or PRB > PRA
UF (uteroferrin)	exog	exog	nc, up; PRA inhibits PRB	human	Porcine	human and monkey; HEC-1A and COS1	Non-consensus PRE and 2 PRE 1/2 sites	(Lamian et al, 1993; Zhang et al, 2003)
PTGER2 (prostaglandin E2 receptor)	exog	exog	up, up; PRB > PRA	mouse	mouse	human; HeLa	Novel PR binding sequence (GG/ACCGGA) and SP1	(Tsuchiya et al, 2003)
BCRP	exog	exog	up, up; PRA inhibits PRB	human	human	human; BeWo	Novel PRE	(Wang et al, 2008)
MUC1	exog	exog	up, up; PRA inhibits PRB	human	human	human; HEC-1A	2 PRE 1/2 sites	(Brayman et al, 2006; Gowland & Buetti, 1989)
MMTV	exog	exog	up, up; PRB > PRA	human	virus	human; HeLa, T47D	3 PRE 1/2 sites, 1 imperfect PRE	(Cato et al, 1986; Gowland & Buetti, 1989; Kastner et al, 1990)
ESR	exog	exog	nc, down	human	human	human; MCF7, ZR75, HeLa	PRE 1/2 site	(De Amicis et al, 2009)
FSHB	nd	exog	nd, up	human	rat	rat; primary pituitary	3 PRE-like sequences	(O’Conner et al, 1997)
RUSH/SMARC3	nd	endog	nd, up	rabbit	rabbit	rabbit; SV40 transformed uterine epithelial	PRE 1/2 site	(Hewetson & Chilton, 2003)
GNRH1	nd	exog	nd, down	human	rat	mouse; GT1-7 pituitary	Non-consensus PRE	(Kepa et al, 1996)
*ITGA6	exog	exog	nc, up	human	human	human; HeLa	Unknown	(Richer et al, 2002)
*FKBP5	nd	exog	nd, up	chicken	mouse	mouse; 3017.1 cells and human; T47D, HeLa	composite PRE/GATA2 binding site	(Magklara & Smith, 2009)
Regulated by PRA only or PRA > PRB
FN1	exog	exog	up, up; PRA>PRB	human	human	human; decidual fibroblasts	none; AP1/CRE and SP1 sites	(Tseng et al, 2003)
*BCL2L1 (Bclxl)	exog	exog	up, down	human	human	human; HeLa	unknown	(Richer et al, 2002)
SERPINB9 (Ovalbumin)	exog	exog	up, nc	human	chicken	chicken; embryonic fibroblasts	unknown	(Kastner et al, 1990)
SERPINB9 (Ovalbumin)	exog	exog	up, up; PRA>PRB	chicken	chicken	chicken; embryonic ibroblasts	unknown	(Tora et al, 1988)
IGFBP5	exog	nd	up, nd	human	human	human; U2 osteosarcoma	2 tandem caccc boxes	(Boonyaratanakornkit et al, 1999)
TAT	exog	exog	up, PRA>PRB	human	rat	human; HeLa	3 GREs as reported in (Jantzen et al, 1987a)	(Vegeto et al, 1993)
PGDH	exog	exog	up, up PRA > PRB; +cAMP PRB > PRA	human	human	human; myometrial smooth muscle	unknown; contains 8 PRE 1/2 sites	(Greenland et al, 2000)
Abcb1b (mdr1b)	endog, exog	endog, exog	up, nc	human	mouse	human; T47D	unknown	(Piekarz et al, 1993)
CSN2 (Beta casein)	exog	exog	PRL up, PRA>PRB down	human	rat	monkey; COS1	2 PRE 1/2 sites; 5′ site > 3′ site	(Buser et al, 2007)
IGFBP1	exog	exog	up, up PRA>PRB	human	human	human; endometrial stromal	2 PREs	(Gao et al, 2000)
Genes oppositely regulated by PRA and PRB
*COMT	exog	exog	up, down	human	human	human; Ishikawa endometrial cancer	2 PRE 1/2 sites	(Salih et al, 2008)
CRH	exog	exog	down, up	human	human	human; primary placental	cAMP regulatory element	(Ni et al, 2004)
GNRHR	endog, exog	endog, exog	down, up	human	human	JEG3 human choriocarcinoma placental	Putative PRE	(Cheng et al, 2001)
PR:DNA interactions, no functional data
ELF5	endog	endog	mRNA up	human	mouse	ChIP in T47D	unknown; 4th Intron	(Hilton et al)
ID4	nd	exog	nd, mRNA up mouse PR	human	mouse	mouse; normal mammary epithelial	2 PRE 1/2 sites	(Fernandez-Valdivia et al, 2008)
Uteroglobin	nd	endog	nd, mRNA up in endometrium	rabbit	rabbit	in vitro	3 PREs	(Jantzen et al, 1987b)
MET IIA	nd	endog	nd, up (heterologous promoter)	rabbit	human	T47D (mRNA)	HRE	(Slater et al, 1988)

Table refers to functional studies except where noted

Only genes whose promoter studied listed in table

nd= not done

nc=no change

endog=endogenous

exog= exogenous

^*=gene present in more than one category

^{^}=as defined by authors

We outline below examples of genes believed to be PR-regulated in an isoform-specific manner (Table 1). The mechanism(s) of PR isoform-specific gene regulation remain largely a mystery, however. By expression profiling, some genes appear to be differentially regulated by PRA and others by PRB. Since the two isoforms probably differ structurally (Bain et al, 2000; Bain et al, 2001) mechanisms for these differences could include unequal DNA sequence binding recognition or affinities, and disparate interactions with coregulatory transcription factors. In silico analysis of putative PREs within promoters of genes uniquely regulated by PRA vs. PRB do not differ substantially, however (Jacobsen and Horwitz, unpublished).

a. Promoters regulated by both PR (or PRA versus PRB not examined)

Promoter regulation by PR invariably involves interactions with coregulators that differ from gene to gene and sometimes species to species. Undoubtedly these differences are responsible in part, not only for regulatory differences between the PR isoforms, but also between PR and the other steroid receptors that also require “PREs”, and between the same genes in different species or cells.

Endogenous FK506 binding protein 5 (FKBP5) is regulated by liganded PR in T47D cells (Hubler et al, 2003) through sequences contained within intron E (Hubler & Scammell, 2004). PRB are stronger transactivators than PRA (Richer et al, 2002). In mice, FKBP5 regulation requires concomitant binding of PR and GATA-2 also at an intronic sequence (Magklara & Smith, 2009) but it is not known if the two proteins interact or if GATA-2 is required for the human promoter. It is thought that PR bound to intronic DNA causes it to loop back to interact with the proximal promoter.

The human BCL2-like1 (Bclxl) gene is more strongly regulated by PRA than by PRB in T47D cells (Richer et al, 2002). The mouse Bclxl P4 promoter (one of multiple promoters in mice) is also regulated by PR when transfected into T47D cells, but the isoform specificity, if any, is unknown. However, the same promoter is PRB-regulated in HC11 mouse mammary epithelial cells and in COS7 monkey kidney cells. It is postulated, but has not been demonstrated that PR regulation of the mouse promoter involves 2 imperfect PREs that resemble PRE ½ sites (Viegas et al, 2004) with GATA3 as a coregulator (Proietti et al, 2011).

In T47D cells, cyclin-dependent kinase inhibitor 1A (p21) is regulated by both PRA and PRB (Richer et al, 2002), with coregulation by SP1 (Owen et al, 1998), STAT5a (Richer et al, 1998) and STAT3 (Proietti et al, 2011). The v-myc myelocytomatosis viral oncogene homolog (avian) (MYC) gene which is rapidly induced by P, contains an imperfect PRE in its promoter that binds PRA and PRB (Moore et al, 1997). Interestingly, overexpression of the breast cancer susceptibility BRCA1 protein blocks PR binding to this PRE and increases corepressor recruitment, while knockdown of BRCA1 enhances coactivator recruitment (Katiyar et al, 2009). The human NAD+-dependent 15-hydroxyprostaglandin dehydrogenase (PGDH) promoter is regulated by P together with Ets, AP1 and TF. A 2.3 Kb promoter fragment contains 6 PRE ½ sites. Both PR isoforms are functional but PRA regulation is stronger than PRB (Greenland et al, 2000) unless PRB effects are enhanced by cAMP signaling, in which case PRB effects exceed PRA (Greenland et al, 2000; Sartorius et al, 1994a). The ADAM metallopeptidase with thrombospondin type 1 motif (ADAMTS-1) lacks a PRE but is nevertheless PRA and PRB-regulated. It is postulated that regulation occurs through multiple tethering factors including CEBPβ, NF1- like factor, and 3 SP1 sites (Doyle et al, 2004). PR also require SP1 to activate the human glycodelin promoter (Gao et al, 2001) but SP1 does not work on the baboon glycodelin promoter (Jaffe et al, 2003).

b. Promoters regulated more strongly or only by PRB

There are many examples of genes and promoters more strongly regulated by PRB. The breast cancer resistance protein (BCRP) promoter is P-induced by PRB but not PRA in human placental choriocarcinoma cells (Wang et al, 2008), and PRA repress PRB when the two are co-expressed implicating the heterodimer. The proximal 100 nt BCRP promoter contains a “novel PRE” to which both PRA and PRB bind by mobility shift assay. This suggests a dominant negative effect of PRA within the heterodimer as previously reported (Mohamed et al, 1994). Similarly, the mucin 1, cell surface associated (MUC1) promoter is stimulated by PRB in human uterine epithelial cells, which is antagonized by PRA (Brayman et al, 2006). Two PRE ½ sites partially mediate this response, but interactions with other transcription factors are also likely since mutation of the PRE ½ sites does not completely abolish P effects. There is a STAT5 binding site in the promoter that is not involved in regulation of MUC1 in HEC-1A endometrial cancer cells (Brayman et al, 2006). Interestingly, PRA represses an ERα-mediated increase of MUC1.

Tissue factor (F3) is strongly regulated by PRB in breast cancer cells (Richer et al, 2002) and by total PR in HESC human endometrial stromal cells (Krikun et al, 2000) via 3 SP1 sites in the proximal promoter. Similarly, the mouse prostaglandin E2 receptor subtype EP2 promoter is more strongly activated by PRB than PRA through 3 regions that include a novel PR binding sequence and two SP1 binding sites (Tsuchiya et al, 2003). The SWI/SNF-related matrix-associated actin-dependent regulator of chromatin-3 (RUSH/SMARC3) promoter is regulated by PRB in uterine epithelial cell lines through a PRE ½ site immediately adjacent to a Y-box. PRB mediated regulation is conferred by a chromatin looping mechanism similar to that observed for FKBP5 (Hewetson & Chilton, 2003). The follicle stimulating hormone, beta polypeptide (FSHB) promoter is regulated by PRB through 3 PRE-like sequences resembling ½-sites that bind PRA and PRB (O’Conner et al, 1997). PRB downregulate the gonadotropin releasing hormone (GnRH) promoter via non-consensus PREs that also resemble a series of ½ sites (Kepa et al, 1996). PRA function was not examined in the latter two studies.

There is significant crosstalk between prolactin (PRL) and P in the mouse mammary gland, with many of the same genes regulated by both hormones, possibly coregulated by STAT5 (Fernandez-Valdivia et al, 2008; Goldhar et al, 2011; Hilton et al). In human breast cancer cells both liganded and unliganded PRs upregulate PRL receptors (PRLR) (Jacobsen et al, 2002; Tseng & Zhu, 1998). The PRLR promoter lacks a consensus PRE but contains multiple transcription factor binding sites including ones for CEBPβ, SP1 and AP1 (Hu et al, 1998) all of which interact with PR (Goldhar et al, 2011). In normal mouse epithelial cells, P regulate inhibitor of differentiation 4 (ID4) via PRB. Its promoter contains 8 putative PRE ½ sites but ChIP studies show PR recruitment to only two of these. Upon PRL treatment, STAT5a also binds this promoter at STAT5 responsive element (S5RE) (Fernandez-Valdivia et al, 2008) and co-treatment with PRL and P bring both STAT5a and PR to the promoter. The human hydroxysteroid (11-beta) dehydrogenase 2 (HSD11B2) gene is more strongly regulated by PRB than PRA (Richer et al, 2002); an effect that again involves STAT5a (Subtil-Rodriguez et al, 2008). Two mechanisms operate: 1) direct binding of PR to the proximal HSD11B2 promoter; and 2) recruitment of PR to the distal promoter via STAT5a. Indeed, STAT5a is an important coregulator of both PRA and PRB activity on many genes. This can be shown by Janus Kinase inhibition, which prevents STAT5 phosphorylation and in turn inhibits P-induction of genes including DUSP1, IL6st, Jun, HMGB3 and STAT5a itself (Subtil-Rodriguez et al, 2008).

E74-like factor 5 (ets domain transcription factor) (ELF5), is regulated by recruitment of PR to a binding site within the 4th intron. A putative second PR binding site has been identified 30Kb upstream of the transcription start-site (Hilton et al). Liganded PR downregulate expression of ERα in breast cancer cells (Ghatge et al, 2005) via a PRE ½ site in the ERα promoter (De Amicis et al, 2009). ERα downregulation appears to be PRB-specific.

c. Genes regulated more strongly or only by PRA

The β–casein promoter is induced by PRL and suppressed more strongly by PRA than PRB (Buser et al, 2007). Two PRE ½ sites flank a STAT5a response element in the promoter and P ligands recruit both PR and STAT5 to these sites. Although PRs bind both PRE ½ sites, their affinity and function are stronger on the 5′-site. Both human and mouse PR were tested in this model, with human PRA stronger inhibitors than mouse PR. The human insulin-like growth factor binding protein 1 (IGFBP1) proximal promoter contains two PREs that are more strongly upregulated by PRA than PRB in endometrial stromal cells. Co-expression of PRB suppresses PRA (Gao et al, 2000). Interestingly, HOX proteins suppress (Gao et al, 2002) and Foxoa1 increase PR-dependent induction of IGFBP1 via either PR isoform (Kim et al, 2005).

Liganded PRA regulate the mouse multidrug resistance 1b (MDR1b) promoter (Piekarz et al, 1993) in T47D cells. The P-responsive region is in the first untranslated exon of the gene, which contains a partial PRE. However PR do not bind there; instead regulation by PR involves CEBPβ and NF-Y (Mallick & Horwitz, 1997). Similarly, the fibronectin promoter, which is regulated more strongly by PRA than PRB, lacks a PRE; instead PRA function via AP1 and SP1 sites (Tseng et al, 2003). Liganded human PRA more strongly regulate the chicken ovalbumin promoter than PRB in chicken embryonic fibroblasts (Kastner et al, 1990). Similarly, the rat tyrosine amino transferase (TAT) promoter is activated more strongly by PRA than PRB in HeLa human cervicocarcinoma cells, but PRB are not dominant negative inhibitors (Vegeto et al, 1993). However, PRs do not stimulate the TAT promoter in monkey kidney CV-1 cells; and both PRs stimulate it equally well in human hepatocellular carcinoma HepG2 cells. This emphasizes the importance of controlling the tissue and species specificity of reagents in such studies.

d. PRA and PRB have opposite effects

Kruppel like factor-9 (KLF9) and PR interact to regulate MMTV, cyclin D1 and uteroferrin expression. The uteroferrin promoter contains two PRE-like ½ sites (Lamian et al, 1993; Zhang et al, 2003) at which KLF9 interacts with liganded PRB to superactivate transcription. KLF9 does not modify PRA-dependent transcription, but it enhances repressive effects of PRA without physically interacting with the receptors (Zhang et al, 2003). Both PRA and PRB activate the decidual prolactin (dPRL) promoter in human endometrial stromal cells. The promoter lacks a palindromic PRE but contains a PRE ½ site and a CEBPβ binding site. Both PR isoforms interact with the inhibitor (LIP) and activator (LAP) isoforms of CEBPβ in vitro but only PRA activates dPRL transcription in vivo (Christian et al, 2002). The catechol-O-methyltransferase (COMT) gene has two promoters; one regulates the soluble form of the protein, the other regulates the plasma membrane form. Transcription of COMT is upregulated by PRA and downregulated by PRB, mediated by 3 PRE ½ sites (Salama et al, 2007). In human choriocarcinoma placental cells, the human gonadotropin releasing hormone receptor (GnRHR) gene is downregulated by PRA and upregulated by PRB, which bind to a nonconsensus PRE (Cheng et al, 2001). Similarly, corticotrophin-releasing hormone (CRH) promoter expression is decreased by PRA and increased by PRB; in this case mediated by a cAMP regulatory element (Ni et al, 2004).

VI. SUMMARY

Progesterone is a key physiologic hormone of women and its receptors play a major role in the diagnosis and treatment of breast cancers. Nevertheless its fundamental mechanisms of action remain in doubt. The receptor proteins have not been purified, crystallographic structural data are unavailable, virtually nothing is known about authentic DNA binding sites on promoters of PR regulated genes, and very little about the true nature of PR interactions with native chromatin. As we review here, in all likelihood, the classical view of PR as dimeric transcription factors that bind a palindromic hormone response element is likely to be either oversimplified or incorrect, given strong evidence that PR function as monomers and that the promoters of PR responsive genes are studded with “half-site” elements. Cumulative data point to cooperative interactions between PR monomers and neighboring coregulatory proteins to bestow progestin specificity to transcriptional responses. A great deal of work has focused on the two PR subunits. Expression profiling data indicate that PRA and PRB regulate different gene subsets suggesting that the two receptors subserve different functions. But, as we point out here, these conclusions are at best preliminary because the models used to perform most studies are problematical. The objective of this review is not meant to highlight these deficiencies and gaps in our knowledge. Rather, we hope to inspire investigators to study these interesting and important receptors using contemporary molecular tools.