The Role of Estrogens and Estrogen Receptors in Normal Prostate Growth and Disease

Estrogens have significant direct and indirect effects on prostate gland development and homeostasis and have been long suspected in playing a role in the etiology of prostatic diseases. Direct effects are mediated through prostatic estrogen receptors alpha (ERα) and beta (ERβ) with expression levels changing over time and with disease progression. The present review examines the evidence for a role of estrogens and specific estrogen receptors in prostate growth, differentiation and disease states including prostatitis, benign prostatic hyperplasia (BPH) and cancer and discusses potential therapeutic strategies for growth regulation via these pathways.

Estrogens in the Male and Effects on the Prostate Gland

While low levels of circulating estrogens are present throughout life in males, there are two time periods, during in utero development and aging, when males are exposed to relatively higher levels of circulating estradiol which have been shown to impact the prostate gland. In addition, estrogens may be produced locally within the prostate via conversion of testosterone to 17β-estradiol by aromatase expressed within the prostate stroma [1, 2], thus estrogen action in the prostate may occur independent of serum levels of this steroid.

During the third trimester of in utero development in humans, rising maternal estradiol and declining fetal androgen production result in an elevated estrogen/testosterone (E/T) ratio. This relative increase in estradiol has been shown to directly stimulate extensive squamous metaplasia within the developing prostatic epithelium which regresses immediately after birth when estrogen levels rapidly decline[3, 4]. Although the natural role for estrogens during prostatic development is unclear, it has been proposed that excessive estrogenization during prostatic development may contribute to the high incidence of BPH and prostatic carcinoma currently observed in the aging male population [5]. African-American men have a two-fold increased risk of prostatic carcinoma as compared to their Caucasian counterparts and it has been suggested that this is related, in part, to elevated levels of maternal estrogens during early gestation in this population [6, 7]. Indicators of pregnancy estrogen levels such as length of gestation, pre-eclampsia and jaundice indicate a significant correlation between higher estrogen levels and prostate cancer risk [8, 9]. In addition, maternal exposure to diethylstilbestrol (DES), a potent synthetic estrogen agonist, during pregnancy was found to result in more extensive prostatic squamous metaplasia in male offspring than observed with maternal estradiol alone [10]. While prostatic metaplasia eventually resolved following DES withdrawal, ectasia and persistent distortion of ductal architecture remained [11]. This has lead to the postulation that men exposed prenatally to DES may be at increased risk for prostatic disease later in life, although this has not been borne out in limited population studies conducted to date [12]. However, extensive studies with rodent models predict marked abnormalities in the adult prostate including chronic inflammatory cell infiltration, prostatic intraepithelial neoplasia and increased susceptibility to adult-onset carcinogenesis as a result of early estrogenic exposures [13-15]. Although DES use during pregnancy is no longer practiced, the recent realization that certain environmental chemicals have potent estrogenic activities [16] has lead to a renewed interest in evaluating the effects and roles of exogenous estrogens during prostatic development [17].

The relative levels of free circulating estrogens increase as men age which can increase estrogenic action in the prostate gland. Bioavailable testosterone levels decline in the aging male due to decreased production by the testis and increased sex hormone binding globulin (SHGB) levels which combine to lower free circulating testosterone [18]. However, circulating levels of free estradiol remain constant in the aging male due to an age-related increase in body weight and adipose cells which express high levels of aromatase and peripherally convert androgens to estrogens [19]. The net result is a significant increase in the E/T ratio allowing the balance between androgen and estrogen regulation of prostate growth to shift towards estrogen dominance. It has been proposed that increased estrogenic stimulation of the prostate in the aging male may lead to reactivation of growth and subsequent neoplastic transformation [20, 21].

Estrogens have significant direct effects on the adult prostate gland as well and have long been suspected in the etiology of prostatic disease [22, 23]. In 1936, Dorothy Price demonstrated that estrogens given to adult rodents leads to hyperplasia, squamous metaplasia and keritinization of the prostate epithelium [24]. Long-term exposure of adult rats to supraphysiologic but non-pharmacologic levels of estradiol and physiologic levels of testosterone leads to prostatic intraepithelial neoplasia (PIN) in the dorsolateral lobe and carcinoma in the periurethral ducts of Noble rats and this is used as a model for estrogen-induced adenocarcinoma of the prostate [25]. Estrogen-induced aberrations in prostate epithelial growth have also been observed in dogs, monkeys and humans with results varying according to species and experimental conditions [26]. In addition to epithelial effects, estrogens induce a preferential stimulation of prostate stromal cell proliferation, consequently, combined administration of estrogens with androgens has been used to experimentally induce BPH in dogs [26, 27]. In humans, the estradiol:DHT ratio increases moderately within normal prostate epithelial and stromal cells upon aging, however, the increased ratio is massive within BPH tissue which directly implicates estradiol in the disease process [28].

Estrogen effects on the prostate gland may also be indirectly mediated through alterations in other serum hormones. Estrogens stimulate the pituitary release of PRL and some, but not all, of estrogenic effects have been attributed to direct PRL action on the prostate [29-31]. Furthermore, estradiol negatively feeds back on the hypothalamic-hypophyseal-testicular axis, blocking lutinizing hormone (LH) secretion and testicular steroidogenesis of androgens (i.e. chemical castration). This feedback regulation was the basis for high-dose estrogen therapy of prostate cancer for several decades [32]. Despite these indirect effects, there is ample evidence through hormone-controlled studies and with in vitro approaches to clearly document that many of the estrogenic effects on the prostate are directly mediated through prostatic expression of estrogen receptors [33-35].

Estrogen Receptors

Estrogen action is mediated through specific estrogen receptors (ER) which are members of a large family of nuclear transcription factors found ubiquitously throughout the animal kingdom. Using a classification system based on the cytochrome P450 family [36] estrogens receptors (ERα and ERβ) are the sole members of the NR3A subgroup, ERα and ERβ being NR3A1 and NR3A2, respectively [37].

The human ESR1 (ERα) cDNA was first cloned in 1985 [38] and has since been isolated in multiple species from humans to fish [39]. The encoded ERα proteins are 595 and 599 amino acids in length in human and mice, respectively, with an approximate molecular weight of 66 kDa (Figure 1A) [38, 40]. Numerous naturally occurring variants of the ESR1 mRNA in normal and neoplastic tissues of several species have been described but the existence of corresponding proteins remains controversial [41, 42].

Figure 1.

A. Modular Structure of Estrogen Receptor. The large gene (>70 Kb) is transcribed into mRNA with 9 exons that encode for a protein with five distinct functional domains referred to as the A/B domain (transactivational), C domain (DNA binding), D domain (hinge region), E domain (ligand binding) and F domain. B. Comparison of the domain structures of ERα and ERβ. The degree of homology for the separate functional domains are indicated at the bottom.

A second ER gene termed ESR2 (ERβ) was discovered in 1996 from rat [43] and human tissues [44] and has also been cloned in multiple species. Unlike the androgen or progesterone receptors, ERα and ERβ are not isoforms, but are encoded by separate genes possibly by gene duplication on different chromosomes, and therefore are distinct receptor forms. Translation of the ESR2 mRNA yields a receptor of 549 amino acids in rodents and 530 amino acids in humans, each with an approximate molecular weight of 60−63 kDa. Therefore, ERβ is slightly smaller than ERα with the majority of this difference in the N-terminus. A number of variant transcripts of the ESR2 gene have been described [45]; however; unlike ERα, it appears that these non-ligand binding isoforms (ERβ2, ERβ4, ERβ5) co-exist with the ligand-binding ERβ1 in certain tissues where they can heterodimerize with ERβ1 and modulate its transcriptional activities.

ERs are modular in both structure and function [39, 46, 47] and are composed of 6 functional modules; an N-terminal (NTD) or A/B domain, the DNA-binding (C) domain, a hinge (D) region, ligand-binding (E) domain and a unique C-terminal F domain of unknown function (Figure 1 B). The NTD (A/B domain) harbors the transcriptional activation function 1 (AF-1) domain and specifies the cell and promoter-specific activity of the receptor as well as co-receptor protein interactions. Phosphorylation of the A/B domain is the most well characterized post-translational modification and occurs via the actions of multiple intracellular signaling pathways such as MAPK and PKA pathways at specific sites [39, 47]. The greatest structural disparity between ERα and ERβ lies within the A/B domain which exhibits only 24% homology between the two receptors and is ∼90 nucleotides shorter in the ERβ as compared to ERβ [48] (Figure 1). This divergence likely accounts for the many functional differences that have been revealed from comparative studies of the two ER forms including coactivator interactions with A/B domain as well as ER-protein interactions, such as AP-1, which modify gene transcription through nonclassical pathways [49-51]. Notably, the AF-1 domain of ERβ is not constitutive and depends on ligand-activated AF-2 action whereas the ERα AF-1 can function independent of AF-2 and ligand. Consequently, certain antagonists (e.g. tamoxifen) that exhibit some agonist-like properties when bound to ERα exhibit no such agonist activity with ERβ [49, 51]. The value in differential responses lies with the potential to selectively modulate estrogen action using SERMs that behave differently for ERα and ERβ in a tissue-specific manner. Overall, ERβ tends to be a less effective transcriptional activator compared to ERα in vitro [44, 49, 52].

The DNA binding domain (DBD) or C domain is the region of the receptor that recognizes and binds to cis-acting enhancer DNA sequences known as estrogen response elements (ERE) located within regulatory regions of target genes. It is the most highly conserved region (>95% homology) of the ERα and ERβ which accounts for the similar affinity for the same EREs [43, 53]. The functionality of the C domain is provided by a motif of two zinc-fingers encoded by separate exons, each composed of four cysteine residues that complex with a single Zn²⁺ ion. The “P-box” (proximal) encodes the first zinc-finger and confers specificity sequences on the ER protein for EREs and forms a “recognition helix” [39]. The “D-box” (distal) encodes the second zinc-finger which is involved in spacer sequence and receptor dimerization [39]. In addition, ERα can induce gene expression when bound as a monomer to an ERE half-site in close proximity to Sp1 binding sites due to cell context-specific interactions of proteins with specific amino acids on Zn fingers 1 and 2 DBD [54].

The hinge region or D domain of the ERs harbors a nuclear localization signal that influences cellular compartmentalization of the receptor. The ligand-binding (LBD) or E domain is a highly structured, multifunctional region that serves to specifically bind estrogens and provide for ligand-dependent transcriptional activity [39]. An Activation Function-2 (AF-2) domain located in the C-terminus of the E domain mediates this latter function and is subject to post-translational receptor modifications [39]. Also harbored within the E domain is a strong receptor dimerization interface. Although there is only ∼ 60% homology in the primary sequence of the LBD of ERα and ERβ, comparative crystal structure studies of liganded and un-liganded LBDs indicate a highly conserved structural arrangement of 12 α-helices which serve as a docking station for agonists and antagonists and a co-activator/repressor recruitment site [43, 55]. The LBD of ERα and ERβ binds the endogenous hormone estradiol, estrone and estriol with similar affinity (ERα = 0.1 nM; ERβ = 0.4 nM) and exhibits equal affinity for DES [56, 57]. However, due to divergence in homology, ERα and ERβ also exhibit measurable differences in their affinity for other endogenous steroids and xenoestrogens [57-59]. For example, ERβ tends to exhibit a stronger affinity for certain phytoestrogens (e.g. genestein and coumestrol) [57, 58]. Recent advances have now permitted the generation of ER-selective non-steroidal ligands [49, 59, 60] that exploit as well as illustrate differences between the LBDs of ERα and ERβ and provide for pharmacological tools to discern the overall function of each ER.

Among the sex steroid receptors, the well-defined F domain is only found in ERs. Studies have indicated a role for the F-domain in receptor stability and co-activator recruitment [61] and in recent findings, in modulating the activational state of AF-2 by affecting the conformational equilibrium of the LBD and selectively modulating its response to ligands [62].

Estrogen Receptor Mechanisms of Action

The ligand/ERE-dependent or “classic” model of estrogen action states that the ER resides in the nucleus but is sequestered in a multi-protein inhibitory complex in the absence of hormone. Upon hormone binding a conformational change occurs in the ER, transforming it to an “activated” state that homodimerizes, shows increased phosphorylation, and binds to an ERE within target gene promoters. The ligand/ERE-bound receptor complex interacts with the general transcription apparatus via co-regulatory proteins to promote transcription of the target gene. This classic estrogen receptor mechanism is dependent on the functions of both AF-1 and AF-2 domains of the receptor which synergize via the recruitment of co-activator proteins, most notably the p160 family members. It is generally believed that the DNA-bound receptor/co-activator complex facilitates chromatin remodeling and formation of a stable transcription pre-initiation complex. When acting via a classic ERE-driven mechanism in vitro, ERα homodimers and ERα:ERβ heterodimers tend to be stronger activators of transcription compared to ERβ1 homodimers [44, 45, 49, 52]. In breast cancer cell lines, ERβ transfection into ERα positive cells inhibited ligand-induced ERα-dependent transcription whereas ERα transfection had little effect on ERβ transactivation [63]. Furthermore, recent findings in prostate cancer cell lines show that non-ligand binding ERβ isoforms (ERβ2, ERβ4, ERβ5) can heterodimerize with ERβ1 under the stimulation of estrogens, but not phytoestrogens, and enhance ERβ1 transactivation in a ligand-dependent manner [45].

There is ample evidence that ERs can be activated in a ligand-independent manner via intracellular second messenger and signaling pathways. This activation through phosphorylation is referred to as receptor crosstalk and it permits induction of ER target genes in the absence of steroid ligand [64, 65]. Polypeptide growth factors phosphorylate ERα-mediated gene expression via the mitogen activated protein kinase (MAPK) pathway and the G-protein/cyclic-adenosine monophosphate (cAMP)/ protein kinase A (PKA) pathway. While growth factors are able to mimic the effects of estradiol in the rodent uterus via this estradiol-independent ERα activation [66, 67], such studies in the prostate are not as yet well described.

Ligand-activated ERα can also stimulate the expression of genes that lack a conspicuous HRE within their promoter [68, 69]. This mechanism of HRE-independent ER activation is postulated to involve a “tethering” of the ligand-activated ER to other transcription factors that are directly bound to DNA via their respective response elements [68, 69]. Therefore, ERα acting in this fashion, may be better defined as a co-regulator rather than a direct acting transcription factor. Estradiol/ERα regulation of several genes, including ovalbumin, collagenase, insulin-like growth factor-1 and cyclin D is believed to occur via a tethering of the receptor to a DNA-bound AP-1 (Fos/Jun) complex with the gene promoter [69]. A similar ERE-independent mechanism of ERα action has been documented for genes that possess a GC-rich region or Sp1 binding site within the promoter where ERα enhances the actions of a bound Sp1 complex [68, 69]. How much of this type of gene regulation is involved in prostate growth and stimulation is unknown at the present time.

Estrogen Receptor Expression in the Prostate Gland

Original studies with ligand binding assays, sucrose density gradients and autoradiography identified specific estrogen binding sites in both epithelial and stromal cell fractions of the prostate gland in different species and these were assumed to be a single ER. With the discovery of ERß in addition to ERα, the localization and relative contributions of each receptor type required reanalysis. Results demonstrated that for the most part, ERα and ERß are expressed in different cellular compartments of the prostate gland with ERα localized primarily to prostatic stromal cells and ERß primarily expressed in prostate epithelium. This differential location as well as differential affinity of the two ERs for ligands, enhancers and co-activators may explain the diverse biological functions of estrogens within the prostate gland and may also be exploited for regulation of prostate disease.

ERα

ERα is localized primarily to the stromal cells of the adult prostate gland in man, dogs, monkeys and rodents [70-74]. Immunohistochemical analysis reveals, however, that ERα expression is heterogeneous in stromal cells, i.e. only a portion of the cells are ERα positive while many remain ERα negative. Studies in rodent prostate glands have shown a relatively high percentage of stromal cells express ERα mRNA and protein during perinatal morphogenesis and this proportion significantly declines thereafter suggesting a specific role for ERα in prostate development [73, 75, 76]. A decline in expression with puberty suggests that androgens may normally suppress ERα expression, a finding that has been borne out in direct studies [77, 78]. In humans, ERα has been consistently observed in stromal cells during fetal development [79]. However, while one report restricts ERα protein to only stromal cells [76], a recent report indicates the presence of ERα in fetal prostatic utricle and periurethral epithelium during mid-to-late gestation [79]. Importantly, squamous metaplasia, observed in all developing human prostates during the third trimester, is directly associated with epithelial ERα in the periurethral ducts and stromal ERα in the peripheral prostatic acini [79].

It is believed that stromal proliferation, a hallmark response to estrogen treatment in most species, may be mediated through stromal ERα. In humans, there is evidence for an increased accumulation of estradiol in nuclei of stromal cells in BPH specimens [80] suggesting that elevated ERα in stromal cells may be involved in the etiology of BPH. Of interest, epithelial cells in prostatic periurethral ducts have also been found to consistently express ERα in both normal and BPH tissue [70]. Since this is the prostatic region which forms BPH, it is possible that epithelial ERα in that specific region are involved. In rodents, periductal mesenchymal and smooth muscle cells, which are in close proximity to epithelial cells, express ERα while interductal fibroblasts are ERα negative [73]. The close proximity of ERα positive stromal cells to epithelial cells allows for paracrine effects of estrogens on prostate epithelium. Indeed, work with ERα knock out mice (αERKO) demonstrated that estradiol-induced squamous metaplasia in adult prostates is mediated through ERα [81]. Similarly, using αERKO and βERKO mice, neonatal estrogenization of the prostate, which includes stromal hyperplasia, epithelial PIN lesions and inflammatory cell infiltration, was shown to be mediated through stromal ERα [82]. It is noteworthy, however, that deletion of ERα in transgenic mice did not produce a marked phenotype in the prostate suggesting that ERα's role in the prostate gland is not necessary for normal growth and function [83, 84].

Recent reports have shown that in prostatic carcinoma, the ERα gene is methylated leading to silencing of this gene, loss of ERα transcription and ERα protein [85, 86]. While comparative data with normal prostate specimens was not available, it was noted that the incidence of ERα gene methylation and silencing increased with progression of prostatic disease from BPH to low grade and to high grade cancer. Thus it was proposed that ERα may have a tumor suppressor role in the prostate gland and loss of its expression may be an early event in prostatic disease. Interestingly, ERα expression has been observed in some prostate cancer cell lines [86] as well as in hormone refractory and metastatic lesions suggesting its re-emergence as cancer progresses [87] although this has not been consistently seen in all studies [88]. It is also noteworthy that prostate cancer risk has been associated with genetic polymorphisms in the ERα gene particularly within Japanese and African American populations implicating a potential causal relationship between ERα mediated estrogenic action and prostate cancer [89, 90].

In summary, there is evidence across multiple species that estrogens acting through stromal ERα may contribute at some level to the etiology of the most prevalent prostatic diseases including chronic prostatitis, BPH, carcinogenesis and cancer progression (Table 1).

Table 1.

Comparison of ERα and ERβ expression and activities in the prostate gland

	ER α	ER β
Localization	Stromal	Epithelial
Proliferation	Epithelial squamous metaplasia	Anti-proliferative
	Stromal proliferation
Differentiation	Epithelial Dysplasia	Pro-differentiation
Immune Response		Anti-inflammatory
		Anti-oxidant
Expression in PCa	Dysregulated in PCa	Dysregulated in PCa
	- Silenced in early cancers	- ↓ organ confined disease
	- Re-emergence with progression	- ↑ in metastatic PCa

ERβ

Since ERβ was originally cloned from a rat prostate cDNA library, it was not surprising to find that the rat prostate expressed this receptor at levels comparable to those found in other high-expressing reproductive organs such as the ovary, endometrium and testis [43, 91]. In the rat and murine prostate, ERβ mRNA and protein are primarily localized to differentiated luminal epithelial cells, which may preclude formation of ERα:ERβ heterodimers in this organ [82, 92, 93]. Expression of ERβ is low at birth, increases as epithelial cells cytodifferentiate and reaches maximal expression with onset of secretory capacity at puberty which suggests a role for ERβ in the differentiated function of the rodent prostate [92]. In the adult, an ERβ expression gradient is observed with low proximal levels and increased expression distally which may contribute to heterogeneity in differentiation and function along the ductal length [92]. In contrast to ERα, androgens up-regulate expression of ERβ in the rodent prostate gland while estrogens to not autoregulate ERβ [92, 94, 95].

It is noteworthy that the developmental pattern for ERβ in the human prostate differs markedly from the rodent. As early as fetal week 7, ERβ is expressed throughout the urogenital sinus epithelium and stroma and this strong expression is maintained in most epithelial and stromal cells throughout gestation suggesting the involvement of ERβ and estrogens in morphogenesis and differentiation [79]. While this pattern is maintained postnatally for several months, ERβ expression declines thereafter with a noticeable decrease in adluminal cells at puberty [76]. In the adult human prostate, ERβ mRNA expression is low relative to testicular expression, again showing a divergence from the rodent prostate gland [44, 91]. Reports vary on ERβ localization in the human prostate which may be a function of antibodies used in immunohistochemical assays. While some have shown that ERβ is expressed by basal epithelial cells with lower stromal cell expression [74, 88], others have shown high expression of ERβ in both basal and luminal epithelial cells of the adult human prostate [96, 97]. In normal and tumorigenic human prostate epithelial cell lines, ERβ is expressed at high levels while ERα is typically absent [86, 98]. In response to in vitro estrogen exposure, estrogen regulated genes (progesterone receptor, pS2) are activated again pointing towards a role for ERß in the differentiated function of the prostatic cell. Together, these findings indicate that, in the prostate epithelium, ERβ may be the key mediator of estrogen-induced events. The prostate gland also expresses ERβ isoform variants which have been shown to act as either constitutive activators, transcription enhancers or dominant negative regulators of estrogen action which further complicates estrogenic action within this gland [45, 99, 100].

Putative roles of ERβ

While ERβ is the predominant estrogen receptor expressed in the adult prostate gland, its role has not yet been clearly established. As stated above, indirect evidence exists for a role of ERβ in the differentiated state of the prostate epithelium. A recent study using βERKO mice showed a shift in basal, intermediate and luminal epithelial cell markers in the prostate towards a less differentiated gland which supports this purported role [101]. As a counterpart to the hypothesis that ERß plays a role in epithelial differentiation, it has also been suggested that ERß has an antiproliferative role in the prostate and participates as a brake for androgenic stimulation of prostate growth [102]. Indirect evidence for this role was suggested by the hyperplastic and dysplastic adult prostate epithelium with reduced ERβ expression following neonatal estrogen exposure [92]. Direct studies for a role of ERβ in prostate proliferation using βERKO mice have yielded conflicting results. While some studies show epithelial hyperplasia with increased BrdU labeling in the βERKO prostates [103], this was not supported in subsequent studies using the same mice [82, 104] or different βERKO models [105]. More recently, direct evidence has been presented for an antiproliferative role of ERβ. Prostate epithelium in aromatase knock out mice (ArKO) becomes hyperplastic with age due to hormonal imbalance. Using tissue recombinants of ArKO neonatal seminal vesicle mesenchyme and adult ArKO anterior prostate epithelium grafted under the renal capsule, the hyperplastic epithelium was reverted to the normal state with the administration of an ERβ specific agonist [106].

Studies in our laboratories suggest that ERβ may also play an immunomodulatory role in the prostate gland. βERKO (n=18) and wild-type (n=18) mice were aged to one year of age and blinded histologic analysis by two independent investigators was performed for the entire glandular complex using serial sections (>100 sections/prostate complex). Aging associated changes were noted at similar incidence in the ventral, dorsal and lateral lobes of both genotypes which consisted of reduced secretions, flattened epithelium and degenerated apoptotic cells within ductal lumens as is typical in the aged murine prostate. Proliferative and apoptotic scores were not different for the two genotypes. However, a feature observed in βERKO but not in wild-type prostates was abundant-to-massive lymphoid aggregates which were at times associated with reactive (proliferative) epithelium consistent with injury-repair cycles (Figure 2). T-cell infiltration throughout the prostatic complex was blindly scored as 0 (absent), 1 (rare), 2 (focal), 3 (abundant) and 4 (massive) for the two genotypes. As shown in Figure 2, the percent of wild-type mice in each category revealed that rare or focal T-cell infiltration were present in 50% of animals while the remaining 50% of wild-type mice had no inflammatory cells present. In marked contrast, T-cell infiltration scores for the βERKO mice showed that the vast majority (89%) of mice possessed prostatic inflammatory cells with 44% of cases presenting with abundant or massive immune cell infiltration (p< 0.005; Chi-square test). Thus we propose that ERß may normally play an immunoprotective role in the prostate gland perhaps limiting tissue damage or modulating expression of stimulus for immune cell infiltration. Estrogens are widely known to affect the development and regulation of the immune system and have been shown to exert potent anti-inflammatory effects [107, 108]. The present data suggests that the anti-inflammatory effects at the level of the prostate gland may be mediated through ERß. This hypothesis is supported by a study which demonstrated that an ERβ-selective ligand was able to prevent inflammatory bowel disease in a rat model [109]. Such an approach may hold promise for treatment of prostatitis which is the most prevalent of prostatic diseases. Furthermore, there is increasing evidence for a link between chronic prostatic inflammation and prostate cancer etiology [110] suggesting a potential protective role for prostatic ERβ in this regard.

Figure 2.

A. Incidence of wild-type and βERKO mice presenting with T-cell infiltration within the prostatic complex of at one year of age. T-cell infiltration throughout the prostatic complex was blindly scored as 0 (absent), 1 (rare), 2 (focal), 3 (abundant) and 4 (massive) for the two genotypes. Overall incidences between the genotypes was shown to be significant at P < 0.005 using a Chi-square test for trend. B. Inflammatory cell infiltration in the ventral prostate lobes of wild-type (with 0 and 2 Score) and βERKO (with 3 and 4 Score) mice at 1 year of age. Wild-type prostates were either free of lymphoid infiltration (0) or contained rare or focal (2) T-cell infiltration. In contrast, focal, abundant (3) and massive (4) stromal lymphocytic infiltrate were routinely observed in the βERKO prostates. As infiltration became abundant, evidence for diapadesis in the epithelium was observed (arrows). In several instances, lymphoid aggregates were associated with reactive, proliferative epithelium in the immediate vicinity (arrowheads). Magnification = 40×.

ERβ has also been proposed to have an anti-oxidant function. ERβ can bind the electrophile/antioxidant response element (EpRE) and is a more potent activator at the EpRE element than ERα [111]. Thus ERß is capable of inducing genes that encode chemoprotective detoxification enzymes (quinone reductase, glutathione S-transferase) and may play an active role in protecting prostate epithelial cells from carcinogens by detoxifying electrophiles. This is particularly relevant since GST-pi has been shown to be critical in protecting against prostate cancer through genome damage initiated by inflammatory cells and carcinogens [112].

ERβ in prostate cancer

Dynamic changes in ERβ expression have been observed during the progression of prostate cancer which suggests that estrogen action through ERβ may play an important role in prostate carcinogenesis, metastasis and perhaps, androgen independence. Most reports on ERβ expression concur that levels decline in localized prostate cancer with increasing grade from PIN through low to high Gleason scores [88, 97, 113, 114]. This loss of ERβ expression in organ confined prostate cancer has been shown to be epigenetically regulated by progressive hypermethylation of ERβ promoter CpG islands causing transcriptional silencing [115]. A recent study suggests that this promoter hypermethylation may be preceded by loss of AP-2 expression in prostate cancer cells which permits methylation at a critical AP-2 binding site in the ERβ promoter [116]. This expression pattern fits with the proposed antiproliferative and pro-differentiation function of prostatic ERβ with its loss permitting unregulated growth and de-differentiation of prostatic epithelium. In support of ERβ as a putative tumor suppressor, a recent study using adenoviral vectors found that ERβ expression in prostate cancer cell lines inhibited growth and invasiveness suggesting that loss of ERβ in higher grade tumors permits proliferation and eventual metastasis [117].

Counter to this concept, ERβ expression remerges as prostate cancer metastasizes to distance sites with 100% of osseous and non-osseous metastatic cells expressing ERβ to varying degrees [96, 118, 119]. ERβ promoter analysis revealed complete hypomethylation of the three 5’ CpG islands in the ERβ 5’ flanking region in metastatic prostate cancer which permits high ERβ gene expression at these distant sites. Thus if ERβ is considered antiproliferative, it is unclear how high ERβ expression in metastatic disease permits uncontrolled proliferation with metastatic spread. Since one study reported that localized prostate cancers which retain ERβ in the primary tumor were associated with a higher rate of recurrence [114], it is possible that ERβ expression in metastatic cells may be a function of a selective advantage of a subset of ERβ-retaining cells in the primary tumor to metastasize. However, this possibility must be reconciled with evidence that ERβ can inhibit migration and invasion in prostate cancer cells [117, 120]. Thus it is currently unclear whether ERβ functions has an antiprolferative role in prostate cancer or whether it promotes metastasis and growth at distant sites. Whatever the case, the important feature is the strong ERβ expression in metastatic, androgen-independent prostate cancer. This suggests that the metastatic cells are targets of estrogen action and thus may be potential targets for therapeutic interventions with antiestrogenic agents or more effective ERβ selective antagonists.

In summary, evidence has been shown for a variety of roles for prostatic epithelial ERβ which are pro-differentiation, anti-proliferative, anti-inflammatory and as an inducer of anti-oxidant genes. While loss of ERβ may contribute to prostate cancer progression in organ confined disease, strong re-emergence of ERβ at metastatic sites implicates a potential role in androgen-independent progression (Table 1).

Estrogen agonists, antagonists, modulators as therapeutic agents for prostate cancer

Due to the above considerations concerning the effects of estrogens on prostate growth and carcinogenesis and the expression of ERs in prostate stromal and epithelial cells, the use of estrogen/antiestrogen therapy may have efficacy for the treatment of prostate cancer. Thus many estrogens including phytoestrogens, DES and 2-methoxyestradiol as well as selective estrogen receptor modulators (SERMs) including raloxifene, ICI 182,780, trioxifene and torimifene have been shown to affect prostate tumor growth through a variety of mechanisms [20, 120-126]. While initial clinical trials using tamoxifene and toremifene proved to be unremarkable for the treatment of prostate disease [127, 128], more recent Phase II studies indicate that the toremifene may in fact effectively block development and progression of clinical prostate cancer and further clinical trials are underway [126, 129]. In is unclear at present whether these agents act primarily through antagonism of ERα or ERβ or whether they differentially modify the actions of both receptors. Future studies with ERα and ERβ specific antagonists and agonists [109] as well as the development of third generation SERMs may provide insight into the specific prostatic ER which is the most effective target for therapeutic use.

Another possible site for treatment of prostate cancer is through inhibition of estradiol production using aromatase inhibitors [130, 131]. Recent work has shown high expression of aromatase in prostate cancer with alternate promotor utilization which suggests that intraprostatic production of estradiol may contribute to progression of this disease [132]. Despite this, clinical trials with aromatase inhibitors failed to show efficacy of global estrogen reduction as an effective prostate cancer therapy [133, 134].

Therapeutic basis for natural estrogenic products as chemopreventive agents

In addition to pharmaceutical approaches, there is considerable evidence that phytoestrogens may modulate prostate growth which has been the basis for herbal supplements and dietary modulation for the treatment of abnormal prostate growth. Genestein [135, 136], resveratrol [137] and soy [138] have all been shown to have beneficial effects and consumption of these products has been inversely correlated with prostate cancer risk [138-140]. High prostatic expression of ERβ may explain why phytoestrogens (genestein, apigenin, coumestrol) are beneficial to prostate health since these compounds have been found to bind to ERß with an affinity up to 10 times higher than for ERα [57, 141]. Thus if ERβ is in fact anti-proliferative in early stage prostate cancer, specific activation of this receptor may be the basis for the beneficial effects of these natural products. However, since clinical trials evaluating the usefulness of these dietary approaches have been limited, caution in their usage must be issued until their safety and effectiveness has been clearly demonstrated.

Conclusions

A considerable amount of evidence now exists that estrogens play an important role in prostate development and homeostasis with different actions being mediated by the stromal ERα and epithelial ERβ. Evidence also indicates that estrogen action mediated through the separate receptors may contribute to the etiology and progression of multiple prostate diseased states. These findings provide new avenues and alternative approaches for the treatment of prostate diseases including prostate cancer with novel therapies directed at estrogen receptors or estrogen metabolism. Since the two types of estrogen receptors may play distinct and perhaps opposing roles in many diseases of the prostate including cancer progression, it is possible that receptor-specific agonists and antagonists may provide the most beneficial therapeutic strategies in future clinical trials.