Molecular Basis of Steroid Action in the Prostate

Introduction

Androgens acting via androgen receptor (AR) play essential roles in the prostate development, growth and pathogenesis of benign prostate hyperplasia (BPH) and prostate cancer. Over the last three decades, intensive studies have been carried out to elucidate the molecular basis of androgen action in the prostate. It has been realized that there are two natural potent androgens in the mammal including humans. Although testosterone is the major androgen secreted from the testes, dihydrotestosterone (DHT) is the main androgen in the prostate to mediate the androgen action via the AR. Only one AR has been identified up to date, a member of the steroid/nuclear receptor superfamily, which is a ligand-dependent nuclear transcription factor. When androgens bind to the AR, this results in a conformational change within the AR, leading to the recruitment of co-regulators and transcription factors which mediate androgen-target gene expression. Androgen actions in the prostate can also be modulated by other hormones such as estrogens via estrogen receptors (ER). There are two known isoforms of the ER, ERα and ERβ, which are both co-expressed with AR in the prostate and prostate tumor cells, which provides an anatomical basis for a direct interplay between AR and ER. Although it is well known that androgens are important for prostate development and for the pathogenesis of BPH and prostate cancer, the precise mechanisms as to how androgens control these processes are not yet fully understood. Furthermore, evidence for the direct modulation of androgenAR actions by other hormones within the prostate or prostate tumor cells is emerging, and the clinical implications of these hormones is being explored. In this article, we will assess the importance of androgens, and especially DHT, in prostate physiology and in the pathogenesis of BPH and prostate cancer. We will also review the molecular nature of the interactions between AR and ER and their respective ligands in prostate cells, as well their potential clinical implications.

Molecular Biology And Structure Of the Androgen Receptor

The androgen receptor, a member of the nuclear steroid receptor superfamily, and an associated ligand-dependent nuclear transcription factor were cloned in 1988 (1,2). A polypeptide product of around 910-919 amino acids is encoded by the AR gene, which is located in Xq11-12 (see Figure 1). The AR gene itself is a single-copy gene that spans approximately 90 kilobases of genomic DNA within the long arm of the X chromosome. The encoding region of the AR gene is comprised of 8 exons separated by 7 introns. Like other steroid receptors, the AR is a single polypeptide comprised of relatively distinct domains (Figure 1, 3): an amino-terminal A/B domain, a DNA binding domain (DBD, domain C), a hinge region (domain D), and a ligand-binding domain (LBD, domain E/F).

Figure 1.

Schematic illustration of the location, exon structure and protein domain structure of the AR gene. (Top) The location of AR gene at the q11-12 of X chromosome. (Middle) The AR gene and its mRNA. The AR gene consists of 8 exons (boxes) and 7 introns (line), and the size of each exon and introns is indicated in kbases. (Bottom) The AR protein. The domains of AR are indicated. Relative positions of glutamine (Gln), proline (Pro) and glycine (Gly) repeats within the N-terminal domain are shown by the indicated boxes. The transactivation function domains, AF-1 and AF-2 are located within the N-terminal domain and ligand-binding domain, respectively. Two zinc fingers in the DNA binding domain and a PEST sequence in hinge region are indicated.

The large amino-terminal domain that comprises nearly half of the AR molecule is encoded by exon 1. It is involved in the transcriptional activation of target genes and contains a transactivation domain, known as Activation Function 1 (AF-1). This domain plays an integral role in AR functions via intramolecular and/or intermolecular interaction with other co-regulators (4). The N-terminal transactivation domain also contains three highly polymorphic direct repeats of amino acid residues: each containing glutamine, proline, and glycine residues, respectively. The expansion of the size of the glutamine homo polymeric segment is related to the spinal and bulbar muscular atrophy (Kennedy's disease) (5), whilst the shortening of the lengths of glutamine and/or glycine repeats may be related to prostate cancer incidence (6-8) although it is not conclusive. Several studies have suggested that the change in the size of glutamine/glycine repeats alters the function of AR due to alterations in its binding affinity to its ligand, interaction with co-regulators, responsivity to phosphorylation, or changes in its AR interdomain interactions (8-11). The mechanisms as to how these changes in the homopolymeric segments result in the pathological outcome are unclear, although it has been shown that the expression of long-tract polyglutamine AR in neuronal cells results in aggregate formation (12) and progressive toxicity (13) in these cells.

The DNA binding domain encoded by exon 2 and 3 contains two “zinc finger” motifs that are hallmarks of all nuclear steroid receptor, being the most highly conserved region within this family (14). The formation of the two zinc fingers involves 8 cysteine residues. These 2 zinc-coordinated stem-loop structures are responsible for the specific interaction with the cognate DNA of target genes, and interact with the major groove of the DNA duplex. For the interaction of the AR and the androgen response element (ARE), the first zinc finger (proximal to the N-terminal domain) is associated with the determination of the sequence specificity of DNA binding, whilst the second finger helps to stabilize the DNA-receptor complex when the DNA binding site is an inverted ARE (14). When the ARE is a direct repeat, the second zinc finger and a part of the hinge region directly interact with the target DNA (15). It is worthy to note that despite their exquisite functional specificity in the physiological context, receptors for androgens, glucocorticoids, progesterone, and mineralocorticoids can recognize the same DNA response element both within in vitro binding assays and also within functional analyses using transient cell transfection (14,16). This paradox may be related to the differential recruitment of various co-regulators upon the ligand-receptor interaction, although it remains to be further investigated.

The C-terminus of the AR is the ligand-binding domain, encoded by the 3′-portion of exon 4 and exons 5-8, and is responsible for mediating the specific high-affinity ligand binding. Studies indicate that androgens interact with the ligand-binding domain mostly through hydrophobic interaction, as well as some hydrogen bonding (17,18). Although the structure of ligand and AR LBD interaction is similar to that of ER and the progesterone receptor, the functional consequences are differentiated. These differential functions may be due at least in part to the differential recruitment of co-regulators (4). The carboxyl-terminus also contains subdomains involved in dimerization and transcriptional activation. The second transactivation function (AF-2) domain of AR resides within the ligand-binding domain (19). Upon ligand binding, this AF-2 domain can interact with co-regulators such as coactivators to affect AR function (4,20). A recent study also indicates that the AR-AF2 domain may also function to stabilize the overall structure of the receptor, allowing the amino terminus to interact with appropriate coactivators (21).

Located between the DNA-binding domain and the steroid-binding domain is the hinge region which contains the nuclear translocation signal. This hinge region of all known mammalian AR also contains a PEST (praline, glutamate, serine, and threonine-rich) sequence, which may function in proteasome-mediated androgen receptor turnover (22).

The AR mediates androgen actions via directed changes in target gene expression (Figure 2). The binding of androgen on the AR results in an AR conformational change that promotes the dissociation of chaperone proteins and facilitates receptor dimerization, nuclear transportation, phosphorylation, and DNA binding (23). Upon the recruitment of co-regulators and general transcription factors, the transcription of a target-gene is either induced or inhibited, and leads ultimately to a change in androgen-target gene expression, and altered cellular or biological structures and functions (see Figure 2). Recently, a number of co-regulators have been identified which can interact with AR at various domains to either enhance or reduce androgen-AR action on target gene transcription (4). Furthermore, AR can be activated in the absence of steroid ligand by other hormones such as IGF-1 (24). Although significant progress has been made in the last decade in understanding the nature of androgen-AR action, the detailed mechanisms of the process from androgen binding on AR, to the alteration of target gene transcription is not fully understood.

Figure 2.

Illustrations of the molecular events of androgen-AR action in a target cell, and the 5α-reductase action. When testosterone (T) enters the cell, it can be converted to dihydrotestosterone (DHT) in the cell by 5α-reductases (5αRD) (top panel). Both T and DHT bind to androgen receptor (AR), resulting in a conformational change in AR and translocation of the receptor complex to nucleus. This complex interacts with androgen response element (ARE) on the target gene, and regulates gene expression in concert with co-regulators (CoR), transcription factors (TF) and the general transcription complex. The changes in androgen-target proteins in the cell eventually affect cellular structure and function related to male sexual differentiation, physiology, and pathophysiology. The function of AR can be activated or modified by non-ligand factors such as growth factors, EGF and IGF-1. GTFs – general transcription factors; ARA – androgen receptor associated proteins; P- phosphorylation; TFs – transcription factors; hsp – heat shock protein.

Androgen Insensitivity Syndrome

The importance of androgen-AR actions in human physiology is best illustrated by the study of androgen insensitivity syndrome, an X-linked recessive condition resulting from AR mutations (3,25). Androgen insensitivity syndrome is classified within a clinical spectrum from complete androgen insensitivity syndrome (CAIS) to partial androgen insensitivity syndrome (PAIS). 46,XY subjects with CAIS are born with female genitalia, and an absent or severely hypoplastic Wolffian ductal system in the presence of testes and normal high or increased androgen secretion. The labia are underdeveloped, the clitoris is normal or small, and the vagina ends blindly. Müllerian derived structures such as uterus and cervix are absent or rudimentary due to normal secretion of anti-Müllerian hormone by the testes (3,26). During puberty, there is a normal or augmented breast development due to an elevated estrogen level originated from the testes, the peripheral aromatization of androstenedione and testosterone by aromatase, and the unopposed estrogenic action by androgens. The pubic and axillary hair is scant or absent, and the prostate is absent in adulthood (3).

The testes of patients with CAIS are located usually within the abdomen or inguinal canal. Post pubertal histological study reveals immature tubular development with Sertoli cells, spermatogonia, and no spermatogenesis (27). Leydig cells are hyperplastic due to the elevated levels of plasma luteinizing hormone which results from unresponsiveness to androgens in the areas of the hypothalamus and/or pituitary (28). Plasma levels of testosterone are normal, high or elevated, whilst DHT levels may be decreased due to a secondary 5α-reductase deficiency (29,30). 46 XY subjects with PAIS have a wide-spectrum of clinical phenotypes ranging from near female phenotype, to severely ambiguous genitalia with gynecomastia, to mild hypospadias, gynecomastia alone, or normal male genitalia with infertility to decreased body hair in adulthood (3,25,26).

Genetic analysis has revealed that a variety of AR mutations are responsible for androgen insensitivity syndrome. To date, more than 490 different mutations throughout all 8 exons in the AR gene have been reported (3,25,26) including the mutation for the largest pedigree of CAIS, (31, see also AR gene mutation database at www.mcgill.ca/androgendb). These mutations range from a single point mutation to the deletion of an entire gene, and can result in various AR dysfunction including impaired androgen binding, impaired DNA binding; impaired cofactor interaction, blocked formation of a functional receptor (deletions, nonsense mutations, splice-junction alterations), decreased AR expression, and an unstable androgen-receptor complex. Depending upon the type of dysfunction, various degrees of functional impairment of the AR occur (3,32). A complete loss of AR function due to a mutation in AR gene results in the complete androgen insensitivity syndrome (CAIS). However, a partial loss of AR function causes the partial androgen insensitivity syndrome (PAIS). Generally mutations that delete the entire AR gene or interrupt the AR open-reading frame, thereby blocking the formation of a functional receptor due to premature termination, aberrant splicing, or even the deletion of partial or entire exon segments, are associated with CAIS. This is due to the fact that the AR DNA- and hormone-binding domain, two critical domains for AR function, are located at the carboxyl terminus of the AR protein. As a consequence, defects that truncate the receptor protein at any point during its synthesis will result in the loss of a portion of one or both of these important functional domains, resulting in a complete loss of AR function.

The majority of AR mutations related to androgen insensitivity, some 67%, are located in AR-LBD. Most of these mutations are single nucleotide substitutions resulting in defects in androgen binding. Approximately 15% of AR mutations are located within the DNA-binding domain, affecting the interaction of androgen-AR complex with the ARE of the androgen-target genes. Some AR mutations affect AR expression, resulting in decreased AR protein levels and decreased androgen binding capacity in the genital skin fibroblasts of affected individuals. These mutant ARs may have subtle differences in their function upon selected target genes when analyzed by in vitro transfection. However, the levels of AR in these individuals are significantly decreased due to altered expression levels of the AR gene arising from AR gene mutation, due to altered translation, post-translational processes or stability of the gene product (33).

Recently, a novel defect has been reported at the post-receptor level in an individual with CAIS as diagnosed based upon clinical and biochemical features (34). The AR gene in this patient is normal. However, the transmission of the activation signal from the AF-1 domain of the AR is disrupted, probably due to a defect in an AR coregulator or coactivator in this patient, indicating that defects in AR signal transmission to target gene expression can also result in androgen insensitivity despite the absence of an AR mutation. The functional significance of androgen-AR pathway in females is poorly understood since the AR gene is located within the X chromosome (35). Recently, Chang's group (36) has generated a cre-lox conditional AR knock-out in female mice, and demonstrated that androgen-AR signaling may also play a role in follicular maturation, female fertility and /or ovulation (8).

The Importance Of DHT In The Prostate

There are two potent natural androgens, testosterone and DHT, which are present in all mammals, including humans (30). Although testosterone is the primary androgen synthesized and secreted from the testes, it functions as a prohormone in certain tissues including the prostate, where it is converted to DHT, a more potent androgen, by 5α-reductase isozymes (30). Although testosterone and DHT bind to the same intracellular androgen receptor (Figure 2), they produce some distinct biological responses. The molecular mechanism for the differential activities and actions of testosterone and DHT is unclear, although differences in receptor binding (37) and DNA interaction (38) between testosterone and DHT have been reported.

5α-reductase isozymes

Steroid 5α-reductase isozymes are microsomal NADPH-dependent proteins that reduce the double bond at the 4-5 position of a variety of C19 and C21 steroids including testosterone (Figure 2). Two 5α-reductase isozymes, type 1 and type 2, encoded by two distinct genes SRD5A1 and SRD5A2 have been identified in the mammal (39-42). The characteristics of two human 5α-reducases are listed in Table 1. Both human 5α-reductase-2 and 5α-reductase-1 genes have five exons and four introns, and encode a highly hydrophobic 254 and 259 amino acid protein with a molecular weight of approximately 28.4 and 29.5 kDa, respectively. 5α-reductase-2 is mapped to the short arm of chromosome 2p23, and 5α-reductase-1 to chromosome 5p15. There is approximately a 50% homology in amino acid compositions between human type-1 and type-2 isozymes. The type-2 isozyme has a much higher affinity than type 1 isozyme for substrates such as testosterone, and is sensitive to finasteride, a 5α-reductase-2 inhibitor, while the type 1 isoform has a low binding affinity and a high capacity for substrate, and is less sensitive to finasteride. The apparent Km (3-10 μM) for the NADPH cofactor is similar for both isozymes. The type 2 isozyme has an acidic pH optimum in the enzymatic assays, while the type 1 has a broad alkaline pH optimum (40,42). However, studies with transfected Chinese hamster ovary cells suggest that the type 2 isozyme may actually have a neutral pH optimum in its native state (43). The affinity of the type 2 isozyme for steroid substrates is higher at a neutral pH than an acidic pH (pH 5.0), suggesting that this isozyme acts primary at neutral pH in the cell (43,44).

Table 1.

Comparison of human 5α-reductase isozymes

	Type 1	Type 2
Gene structure	5 exons, 4 introns	5 exons, 4 introns
Gene, chromosome location	SRD5A1, 5p15	SRD5A2, 2p23
Size	259 amino acids, Mr=29,462	254 amino acids, Mr=28,398
Tissue distribution	Liver, nongenital skin, prostate, brain	Prostate, epididymis, seminal vesicle, genital skin, uterus, liver, breast, hair follicle, placenta, brain
pH optima	Neutral to basic	Acidic or neutral
Prostate level	Low	High
Substrate (T) affinity	Km = 1-5 μM	Km = 0.004-1 μM*
Activity in 5α-reductase deficiency	Normal	Mutated
Finasteride inhibition	Ki > 300 nM	Ki > 3-5 nM

^*Note: - The Km of 5 α-reductase-2 for testosterone (T) is dependent on the assay condition (43).

The functional domains of 5α-reductase-2 have been deduced from in vitro mutagenesis-transfection analysis of natural mutations of 5α-reductase-2 gene in cultured mammalian cells (42,45,46), and mutagenesis analysis of 5α-reductase-1 isozyme (47). Mutations affecting NADPH binding map to the C-terminal half of type-2 isozyme, suggesting that the C-terminal of the isozyme appears to be a cofactor-binding domain, even though consensus adenine dinucleotide-binding sequences have not yet been identified. In contrast, 5α-reductase-2 mutations that affect substrate (testosterone) binding appear to be located at both ends of the protein. However, the amino acid determinants of the substrate binding domain appear to be mainly located at the amino terminal of the protein (42). 5α-reductase-2 is expressed in external genital tissues early in gestation (48). In adulthood, its expression in prostate, genital skin, epididymis, seminal vesicle and liver is relatively high, while it is quite low in other tissues. This isozyme also appears to be expressed in the ovary and hair follicles (49,50).

5α-reductase-1 is detected at birth in the liver and non-genital skin, and is present throughout life. Its expression in embryonic tissues, however, is quite low. In adulthood, it is expressed in non-genital skin, liver and certain brain regions; whereas its presence in the prostate, genital skin, epididymis, seminal vesicle, testis, adrenal and kidney is low. The function of 5α-reductase-1 in human physiology remains to be defined. In the human prostate, both 5α-reductase isozymes are present in epithelial cells and stromal cells, while 5α-reductase-2 is the predominant isozyme expressed within the stromal cells (42,48,51). Both isozymes are expressed in BPH and prostate cancer tissues, and in prostate tumor cells (52-56).

5α-Reductase-2 deficiency syndrome

5α-reductase-2 deficiency syndrome is an inherited autosomal recessive disease resulting from mutations in the 5α-reductase-2 gene (40). To date, over thirty-three mutations in the 5α-reductase-2 gene [(30,57-60) and our unpublished data] have been identified, including mutations in the 3 largest kindreds of male pseudohermaphrodites with 5α-reductase-2 deficiency in the world - the Dominican, New Guinean and Turkish kindred. Mutations have been identified in all five exons of the 5α-reductase-2 gene, and range from a single point defect to a deletion of the entire gene (30). These mutations result in various enzymatic dysfunction including a complete loss of enzymatic activity; impaired binding of substrate and cofactor to the isozyme; blocked formation of a functional isozyme (deletion, nonsense mutation, splice-junction alterations); and an unstable isozyme (42,45,46).

Male pseudohermaphrodites due to 5α-reductase-2 deficiency provide a natural human model to elucidate the importance of DHT in human physiology and pathophysiology. The clinical syndrome of 5α-reductase-2 deficiency was first described in a large Dominican kindred (61), and in two siblings from Dallas (62). Subsequently large cohorts in New Guinea (63) and Turkey have been described (46,64,65) as well as many other cases worldwide (30). Due to 5α-reductase deficiency and a decreased conversion of testosterone to DHT, the affected subjects have high normal to elevated levels of plasma testosterone, and low normal to decreased levels of plasma DHT, resulting in an increased testosterone to DHT ratio at baseline and/or following hCG stimulation (30,66). The levels of plasma and urinary 3α-androstanediol glucuronide, a major metabolite of DHT, are decreased. These subjects also have a global deficit in 5α-reduction as demonstrated by both decreased urinary 5α-reduced metabolites with increased 5β/5α urinary metabolite ratios of both C21 steroids and C19 steroids other than testosterone, i.e. cortisol, corticosterone, 11β-hydroxy-androstenedione and androstenedione (67,68).

Most affected 46,XY subjects with 5α-reductase-2 deficiency have ambiguous external genitalia with a clitoral-like phallus, severely bifid scrotum, pseudovaginal perineoscrotal hypospadias and a rudimentary prostate at birth (58,61,69). Wolffian duct differentiation in affected males is normal, and no Müllerian structures are present. Cryptorchidism is frequently described though it is not invariably present with testes usually found in the inguinal canal or scrotum and occasionally located in the abdomen. With the onset of puberty, the affected males have increased muscle mass and deepening of the voice. The genitalia enlarge with growth of the phallus as well as rugation and hyperpigmentation of the scrotum. The affected adult males have intact libido and are capable of erections (66). Although most subjects studied are generally oligo- or azoospermic due to undescended testes, normal sperm concentrations have been reported in subjects with descended testes (70-72), and the affected men can father children (72,73). These findings indicate that sufficient DHT activity is required for the development of male external genitalia and the prostate, and the descending of testes.

DHT action on prostate development and growth

The prostate is a ductal-acinar gland whose growth and development initiates in fetal life and completes at sexual maturity (74). Development of prostate begins when prostatic buds emerge from the urogenital sinus that derived from endoderm at tenth week in human fetuses, on day 17 in embryonic mice and on day 19 in embryonic rats. Normal prostate development requires many coordinated cellular processes and involves multiple genes and hormonal actions (75). Androgens, mainly DHT, play an essential role in the prostate development and growth as 46,XY subjects with CAIS do not have a prostate, and DHT deficient subjects with 5α-reductase deficiency have a rudimentary prostate. Studies in rabbit, rat (76), and human (77) fetuses have shown that 5α-reductase activity is present in the urogenital sinus and external genital anlage prior to prostate and external genital differentiation. However, 5α-reductase activity is not present in the Wolffian duct, at the time of epididymal, vas deferens, and seminal vesicle differentiation. Thus testosterone, and its 5α-reduced metabolite DHT have selective roles in male sexual differentiation during embryogenesis. Testosterone mediates Wolffian ductal differentiation, while DHT mediates male external genital and prostate differentiation. This hypothesis is confirmed by the study of 5α-reductase-2 deficiency syndrome.

Patients with 5α-reductase-2 deficiency syndrome have decreased circulating and prostatic DHT concentration due to decreased conversion of testosterone to DHT. In the affected male adults, the prostate is nonpalpable on rectal examination (61,78) and is found to be rudimentary on transrectal ultrasound and MRI visualization (69). Prostatic volumes are approximately a tenth of those of age-matched controls (see Figure 3). Histological analysis of prostate biopsy from these patients reveals fibrous connective tissue, smooth muscle, and no identifiable epithelial tissue, suggesting atrophic epithelium or lack of epithelial differentiation (69). Plasma PSA is low or undetectable in these patients. Administration of DHT results in enlargement of the prostate (Figure 3), and an elevation of plasma PSA levels (66,79,80). These findings provide clinical evidence that prostate differentiation and growth as well as circulating PSA level is mediated largely by DHT. However, the mere presence of a prostate in these individuals supports the notion that other growth factors are also involved in its organogenesis. Further supportive evidence is provided by animal studies using 5α-reductase-2 inhibitors and gene knockout. Administration of a 5α-reductase-2 inhibitor, finasteride, in rats (81,82) and monkeys (83) impairs male sexual differentiation and prostate development. The prostate in mice with genetic disruption of either 5α-reductase-2 or 5α-reductase-2 plus 5α-reductase-1 gene is small, but it is puzzling that these knockout animals have normal external genitalia in male offspring (84).

Figure 3.

Comparison of prostate sizes between age-matched normal adult males and 5α-reductase-2 deficient patients before (5α-RD) and after (5α-RD/DHT) DHT treatment for 3 to 6 months. Panels A and B show representative sonograms of prostate in a 5α-reductase-2 deficient patient pre-DHT treatment (A) and post 2% DHT cream (B) applied to the genital area for approximately 3 months. Note the crosses at the outer edges of the prostate. Panel C shows the prostate volume as determined by sonogram in 5α-reductase-2 deficient patients and age-matched normal male controls.

DHT action on BPH

BPH is responsible for considerable morbidity due to urethral obstruction. Histological evidence of BPH is found in 50% of males by the age of 50, and 90% of males by the age of 80 (85). The development of BPH is exclusively dependent on androgens, and BPH does not occur in men castrated prior to puberty (86). Although testosterone is the major androgen from the testes, DHT is known to be the major intracellular androgen to mediate androgen actions in the prostate cells (56,87,88). Androgen withdrawal by castration leads to atrophy of prostate gland due to prostatic cell apoptosis (89). In a classic study by White in 1895, 111 men with bladder outlet obstruction were castrated (90). Nearly 87% had rapid atrophy of their enlarged prostate and almost 58% had relief of symptoms. Administration of either DHT or testosterone to castrated dogs results in an increase in intraprostatic DHT, and in BPH (91). However, the concomitant administration of testosterone with a 5α-reductase inhibitor results in a decreased DHT formation and a prevention of BPH (92,93). Administration of finasteride, a specific 5α-reductase-2 inhibitor causes a selective decrease in both circulating and intraprostatic DHT, a significant decrease in prostate size due to prostatic cell apoptosis (94,95), and an improvement of clinical symptoms in patients with BPH (96). In addition, finasteride has been shown to reduce blood flow in both the ventral and dorsal lobes of the rat prostate, which may be mediated by decreasing vascular-derived endothelial growth factor gene expression (97).

The specific 5α-reductase-2 inhibitor, finasteride has now been used for the treatment of BPH. This novel therapeutic strategy evolved from the clinical observation that adult male pseudohermaphrotides with 5α-reductase-2 deficiency and a deficiency in DHT production have rudimentary prostate (61,66,69). Recently, a specific 5α-reductase-1 inhibitor and a dual 5α-reductase-1 and 2 inhibitor have been developed. Pre-clinical and preliminary clinical studies have shown that inhibition of both 5α-reductase isozymes resulted in greater and more consistent suppression of circulating DHT than that observed with the selective inhibitor of 5α-reductase-2, finasteride (98). However, whether inhibition of both 5α-reductase isozymes is more efficacious in the treatment of BPH than 5α-reductase-2 inhibition alone remains to be defined.

DHT action on prostate cancer

Prostate cancer is the most commonly diagnosed and the second leading cause of cancer death in Western males (99,100). The importance of DHT in prostate development and growth, and the dependence of prostate cancer upon androgens indicate that DHT may play a role directly or indirectly in the onset, maintenance, or progression of prostate cancer. Epidemiological studies indicate that although the autopsy prevalence of latent non-infiltrating prostate cancer appears to be similar in many different racial groups, there is a 50-fold difference between Asian males and American black males in the incidence of clinically overt disease (101,102). Analyses of the 5α-reductase activity in these ethnic groups demonstrate that 5α-reductase activity in Asian males is lower than African-Americans and American Whites (103-106), and is increased in Asian males who live in North American compared to those in Asia (106). These data correlate with the progressive increase in clinical prostate cancer incidence in immigrants who have moved from Asia to United States, in the second-generation Asian-American males when compared to the first-generation immigrants (107,108). Furthermore, allelic variants in the 5α-reductase-2 gene may be associated with prostate cancer risk and prostate cancer progression: a variant with higher enzymatic activity is associated with a higher incidence of prostate cancer in population studies (109-111). These data suggest that increased 5α-reductase activity may relate to the pathogenesis and progress of prostate cancer.

Clinical studies of male pseudohermaphrodites with an inherited DHT deficiency due to 5α-reductase-2 gene mutations provide additional evidence to support the concept that DHT plays an important role in prostate cancer development. As stated above, patients with 5α-reductase-2 gene mutations have an underdeveloped prostate and undetectable plasma PSA levels. After following such cases for many years, neither BPH nor prostate cancer has been observed in these patients (66). Further evidence comes from studies with specific 5α-reductase inhibitors. Administration of a 5α-reductase inhibitor prevents the development of spontaneous prostate cancer in both in animals (112) and in humans (113). In a large randomized, placebo controlled chemoprevention trail (113), prophylactic use of finasteride in males over the age of fifty decreases the incidence of prostate cancer by 25% compared to those in the placebo group. However, the tumor malignancy is increased in the finasteride-treated group, a dilemma when using finasteride for the prevention of prostate cancer.

Although both type 1 and 2 5α-reductases are expressed in prostate tumor cells (56), and the administration of 5α-reductase inhibitor alone or in combination with androgen antagonist inhibits prostate tumor cell growth in culture (114,115), the efficacy of finasteride in the treatment of prostate cancer is disappointing (116). Since 5α-reductase-1 activity is 3 to 4 times greater in malignant than in benign prostate tissues, while the 5α-reductase-2 activity is similar in both forms (117), it follows that 5α-reductase-1 activity is implicated in malignancy. Whether inhibition of both isozymes by combinatorial treatment with both specific type 1 and type 2 inhibitors will be more effective in the prevention and therapy of prostate cancer is an interesting strategy, and remains to be evaluated.

Interaction Of Androgen And Estrogen In The Prostate

In vivo, the effect of a hormone is the outcome of the integration of multiple signaling events within its micro- and macro-environments (118). Interactions between androgens and estrogens typically have antagonistic effects. These interactions are documented (119) in various targets including breast (120), ovary (121), uterus (122), testis (123), and brain (124). However, androgen-estrogen interactions within the prostate are complex, and may result in either potentiating or attenuating actions (125-130).

Estrogens and estrogen receptors (ER) in the prostate

Like androgens, estrogens also act through their corresponding ERs. Both AR and ERs are members of the nuclear receptor superfamily and are ligand-dependent nuclear transcription factors (16). ERs mediate the genomic actions of estrogens via the interaction of the hormone-receptor complex with the cognate Estrogen DNA Response Element (ERE) of the target genes. Like the AR, ERs also feature ligand binding, DNA binding, dimerization and transactivation domains (Figure 4, 19). Both agonists and antagonists bind to the ligand-binding domain but produce differential conformational changes (131,132). Similar to the AR, ERs can also interact with co-regulators which modulate estrogen actions (20,133).

Figure 4.

A structural comparison of human ERα and ERβ.

A schematic structural comparison of human ERα and ERβ. Receptor domains are illustrated with different colored boxes, and the approximate size of each domain is indicated.

Recently, it has been found there are at least two ER isoforms (Figure 4) encoded by two distinct genes ERα (134) and ERβ (135-137) in mammals, including humans. These two ER isoforms exhibit high amino acid homology, notably within the DNA binding domain (∼96% homology) and in the ligand binding domain (∼60% homology, 137). The N-terminal A/B domains of these two ER isoforms are more diversified with only 30% homology. ERβ, like ERα, contains a weak activation domain, AF-1, and also a repressor domain (138). In addition, both receptors contain an AF-2 element within the ligand-binding domain. Although ERβ shares many functional characteristics with ERα, it has distinct properties both in the regulation of estrogen-responsive gene expression and in its interaction with other signals (138-141). The partial agonistic activity of 4-hydroxytamoxifen upon ERα mediated changes in gene expression is not observed with ERβ (142). In co-transfection assays, SRC-1 can upregulate ERβ transactivation in the absence of estrogen, whereas SRC-1 only potentiates ERα transactivation in the presence of estrogen (142,143). Paech et al have shown that ERα and ERβ mediate opposing signals in the presence of 17β-estradiol from an AP1 site: through ERα, 17β-estradiol activates transcription, whereas via ERβ, 17β-estradiol inhibits transcription (140). Recently, we demonstrate that estrogen modulation of DHT action in prostate gene expression and prostate tumor cell growth via ER may be also ER-isoform specific (144).

ERα and ERβ can bind to ERE as homodimers which possesses transcription activity within their target genes (142,145,146). ERα and ERβ can also form a heterodimeric complex which also retains DNA-binding capacity and specificity (142,146,147). Thus the genomic actions of estrogens may be transduced through three differential pathways in vivo: as ERα homodimers, as ERβ homodimers and as ERα/ERβ heterodimers (Figure 5).

Figure 5.

A graphic illustration of ligand-AR and ER cross-talk in the prostate cells.

A graphic illustration of ligand-AR and ER cross-talk in the prostate cells. T – testosterone, E2 – estrogens, and 5αRD - 5α-reductase.

Both ERα and ERβ are expressed in the human and animal prostate along with the AR (148-152). In prostate epithelial cells, the prominent ER-isoform expressed is ERβ, while ERα may be expressed primarily within the stromal cells. Thus, it appears that in the prostate stromal and epithelial cells, two receptors (AR and ERα, or AR and ERβ), or three receptors (AR, ERα and ERβ) may be co-expressed. In prostate cancer cells, either one or two ERs isoforms are expressed. Using RT-PCR, Lau et al have shown that among 3 prostatic cancer cell lines tested, LNCaP and DU145 cells only expressed ERβ mRNA, while PC-3 cells expressed both ERα and ERβ (149). In LAPC-4 and MDA Pca-2b prostatic tumor cells, ERβ is the major ER expressed while the level of ERα is quite low (144). ERα and ERβ are also differentially expressed in human prostate cancer tissues (153-155). These data suggest that prostate cancer cells may differentially express ER isoforms, providing a potential anatomical basis for differential direct estrogen actions in the prostate.

The physiological role of estrogens in the prostate is still obscure. Analyses by gene disruption of ERα, or ERβ, or ERα plus ERβ indicate that estrogens appear to have only minimal effect on prostate differentiation and development (156-158). However, notable changes are present in ER knockout animals (156,158,159). In ERα knockout animals the prostate is enlarged upon aging, although it remains histologically indistinguishable from that of its wild-type littermates. In ERβ knockout mice, prostate epithelial hyperplasia is observed although there appears to be some inconsistency. In ERα/ERβ double knockout mice, the prostate weight and morphology are not different from those of their wild-type littermates up to 8 months of age. These data suggest that either ERα and ERβ do not have significant effects in the prostate, and the observed prostate hyperplasia in ERβ knockout mice is due to some other unknown action; or that ERα and ERβ may possess opposing actions (140,159), resulting in a balance of ER actions when both are either present or absent.

The significance of estrogens in prostate physiology has been further analyzed in aromatase knockout (ArKO) animals. In both young and aged mice, prostate enlargement and hyperplasia, but not malignancy, are demonstrated, which may result directly from elevated androgen levels (158,160). The discrepancy of the prostate changes between ArKO and double ERα/β knockout animals suggests that other pathways in addition to ERα/β may mediate estrogen actions in the prostate. In transgenic male mice overexpressing aromatase, the prostate is rudimentary with squamous epithelial metaplasia (158,161). Taken together, these genetic analyses indicate that estrogens acting via ERs or other pathways play a role in prostate physiology and pathophysiology, which is consistent with classic studies using estrogen analogs in animal studies (128). One of the mechanisms of estrogen action in the prostate could be mediated via the modulation of androgen actions.

Direct interactions of androgens and estrogens in the prostate

In the prostate, the interactions of estrogens and androgens appear very complicated [see recent reviews (127,128)], partially due to the presence of both direct and indirect estrogen actions (Figure 6). Estrogens may either potentiate or inhibit androgen-induced biological responses and gene expression. Thus, estrogens may have either beneficial or deleterious actions with regard to BPH and prostate cancer.

Figure 6.

Illustration of the direct and indirect actions of estrogens in the modulation of androgen actions in the prostate cell.

Estrogens have been used as anti-androgens within androgen ablation therapy of prostate cancer (127,162), presumably acting by inhibiting testosterone biosynthesis via the negative feedback of the hypothalamus-pituitary-gonadal axis (see Figure 6). Large phase III trials in multiple clinic centers have shown a beneficial effect of the estrogen treatment of prostate cancer (127,163,164). A double-blind, randomized study of stage D2 prostate cancer patients showed that patients treated with an estrogen analog, diethylstilbestrol, had a significantly longer time to treatment failure (26.4 vs 9.7 months) and longer survival (43.2 vs 28.5 months) than flutamide (an androgen antagonist) treated patients, but with more cardiovascular complications (165). Recently, clinical studies have also shown that estrogen analogs may be effective as a second, or third-line drugs for the treatment of androgen-independent prostate cancer in the improvement of symptoms and reduction of plasma PSA levels (129,166,167). Although the mechanisms of this estrogen effect in androgen-independent prostate cancer are unknown, it is unrelated to the inhibition of androgen biosynthesis since the prostate cancer is in an androgen-independent phase. Actually, direct modulation of androgen actions in the prostate cells by estrogens has been demonstrated. Jarred et al have shown in neonatal rat prostate tissue culture that estrogen inhibits androgen-induced prostate growth and differentiation (168). Recently, we have demonstrated both within in vitro cotransfection assays and within prostate tumor cells that estrogens produce a receptor and ligand specific inhibition of DHT-induced prostate specific antigen (PSA) gene expression and tumor cell growth (144). Direct estrogen modulation of gene expression, cell growth and androgen actions in prostate tumor cells has attracted broad attention (169-171).

On the other hand, synergistic effects of estrogen and androgen interaction have been reported (126,130,151). Some epidemiological studies suggest that high plasma levels of estradiol may be a risk factor for BPH and prostate cancer (128). Animal studies have shown that estrogens can potentiate androgen-induced prostate growth and DNA biosynthesis in the dog (130) and in castrated rats (126). Administration of testosterone alone in the rat results in a low incidence of prostate cancer, but the addition of estrogens produces a significantly higher incidence of prostate cancer (128,172,173). Estrogens also potentiate the androgen-promoting effect on carcinogen-induced prostate cancer (128). Although potentiating androgen action may be a mechanism of carcinogenesis in the prostate, other estrogen effects, including DNA damage via catechol metabolites, also play a significant role (128,174).

In summary, estrogens exert complex and differential actions within the prostate, which may be dose-dependent (175), species-dependent (130), cell-specific, and receptor-isoform and ligand differentiated (144,158). Furthermore, the estrogen modulation of androgen actions in the prostate may be either indirect through the hypothalamus-pituitary gonadal axis to affect testosterone biosynthesis, or exerted directly via ERs in the prostate cells (Figure 6).

Molecular basis of direct ligand AR and ligand ER interactions in the prostate

Growing evidence supports the belief that estrogens acting via ERs directly modulate androgen actions via the AR, both in the prostate and within prostate tumor cells (144,168-170). Since both AR and ER belong to the nuclear receptor superfamily and are ligand-dependent nuclear factors, such a ligand ARER interaction could occur at multiple levels (see Figure 5) as with other nuclear receptor interactions (118,176,177). Previous studies of hormonal interactions in the nuclear receptor system indicate that ligand nuclear receptors can either interact at the DNA level via protein-DNA interaction, or at the protein level via protein-protein interaction (23,118). At the DNA level, ligand receptors may compete for the same DNA binding site, or indirectly affect protein-DNA interactions (pathway 1 of Figure 5). For example the ligand TRα1 can directly compete with ligand ERα on the DNA binding site of oxytocin gene to inhibit estrogen-induced oxytocin gene expression (178,179). At the protein level, ligand receptors may interact directly (pathway 2 of Figure 5, 170,180), or display a squelching effect to compete with common transcriptional factors or co-regulators (pathway 3 of Figure 5, 141,181), resulting in a mutual inhibition of their actions.

Using PSA and a MMTV-promoter directed reporter gene construct, the mechanisms of ligand-AR and ligand-ER interaction have been analyzed. Human PSA gene expression is induced by androgens, and multiple functional AREs have been identified in the proximal and distal promoter region of the human PSA gene (182-184). A consensus ARE is identified in the MMTV promoter (14). It is known that the consensus DNA binding elements of AR and ERs are distinct, and that ERs do not bind to a consensus ARE and vice versa (14,19). However, it is not clear whether ER can bind to the functional AREs of an androgen-target gene such as PSA. Using gel mobility shift assay, we observed that both ERα and ERβ do not directly bind to the AREs in the human PSA gene promoter (our unpublished data), suggesting that direct competition of ARE binding between ligand AR and ER in the PSA promoter is unlikely. However, this does not exclude other interactions at the DNA level. For instance, ligand ER could bind at nearby DNA sequences of a functional ARE to interfere with ligand AR binding due to spatial hindrance; or else ligand ER could affect AR binding via another intermediate factor such as AP-1, SP-1, or NFκB (23,185). Actually, it has been shown that the inhibition of DHT-induced PSA transcription by 17β-estradiol via ERα is dependent on the ERα DNA binding domain (144), indicating that the ligand ERα has a potential to interact at the PSA promoter. However, how this interaction occurs remains to be elucidated.

At the protein level, direct AR-ERα interaction has been reported using yeast and mammalian two-hybrid systems (170), resulting in a mutual inhibition of the ligand AR and ligand ERα action in regulating gene expression. However, such mutual inhibition of ligand AR and ligand ERα action has not been demonstrated in other studies although significant inhibition of androgen action by ligand ERα is present [(169,186) and our unpublished data]. The lack of mutual inhibition of ligand AR and ER actions suggests that direct receptor interaction, or squelching of common transcriptional factors or co-regulators may be not the major or the sole mechanism for estrogen modulation of androgen action in ligand AR-ER interaction. The reason for the discrepancy of ligand AR-ER interaction is unclear. One of the notable factors is the different androgen used in the experiments. Kumar et al (169) and our group induced the androgen-target gene expression by using DHT, a natural potent androgen, while Panet-Raymond et al (170) used mibolerone, a synthetic potent androgen, to stimulate target gene expression. Although both of them bind AR and stimulate target gene expression, differences are observed in AR induced conformational changes when binding to these ligands. The AR helix 12 is split into two shorter helical segments when AR binds to mibolerone (18), while the structure of AR bound to DHT shows a continuous helix 12 (17). Whether the androgenestrogen interaction via their corresponding receptors is AR-ligand specific is an interesting question in of itself.

It is interesting to note that the modulation of androgen action by estrogens is both ER-isoform and ligand specific (144). For instance, 17β-estradiol inhibits androgen induced target gene transcription via ERα, but not ERβ (144,170), while 17α-estradiol inhibits androgen action via either ERα or ERβ (144). 17α-estradiol is a stereoisomer of 17β-estradiol, which also binds to ERα and ERβ (139), and processes estrogenic activity although much weaker (187). This ER-ligand specificity may be due to the differential conformational changes in the ERs upon binding ER-ligands (185,188,189), which could result in a recruitment of various transcriptional factors or co-regulators (189,190), which would then lead to different down-stream actions as stated above. Furthermore, even though both 17β and 17α-estradiol inhibit DHT-induced PSA transcription via ERα in cotransfection assays, the mechanisms of their actions may be different (144). The inhibition of DHT-induced PSA transcription by 17β-estradiol via ERα mainly involves actions at the level of the ERα DNA binding domain, whilst 17α-estradiol action via ERα is mainly mediated through the C-terminal F domain of ERα. The F domain is highly variable within the nuclear receptor superfamily, and there is little homology between ERα and ERβ. The F domain of ERα potentially contains helix 13 and β-strand motifs based on secondary structure calculations (191,192). The F domain of ERα is required for the agonist/antagonist action of tamoxifen (193), and the estrogen activation of ERα/SP1 pathway (192), although its functional significance remains to be further elucidated.

In summary, ligand AR and ER may interact at multiple levels to regulate target gene expression and consequently cell function. Although some efforts have been made to understand the mechanisms of their interaction, further studies are necessary to elucidate in detail the nature of the ligand AR-ER interaction. It is still unclear how ligand receptors interact with their co-regulators; the functional significance of each receptor domain; and most importantly, the biological significance of this AR-ER interaction in prostate physiology and pathophysiology.

Conclusions And Perspectives

Basic and clinical studies, especially in naturally occuring human genetic models with AR and 5α-reductase-2 gene defects, have established the importance of androgen-AR signaling in prostate physiology and pathophysiology. The androgen-AR signaling pathway in the prostate cells is modulated by estrogens via estrogen receptors. It has been reported both in vivo and in vitro that estrogens can either inhibit or potentiate androgen effects within the prostate. Although the mechanisms of these apparent opposing estrogen actions are unknown, it may be related, at least in part, to distinctive receptor isoforms and the ligand specificity of their interplay (144). Since the outcome of androgen-estrogen interaction is species (130), cell-type, coregulator, ER-isoform and ligand specific, the final result can be either inhibitory or stimulatory dependent upon the ER-isoforms and co-regulators expressed in the cells, as well as the ligands used in the experiment.

Estrogens have been used for a long time in hormonal therapy of metastatic prostate cancer, presumably by inhibiting testosterone biosynthesis via the negative feedback of the hypothalamus-pituitary-gonadal axis (see Figure 6) (127,162). The identification of the ERβ isoform, which is expressed within both prostate epithelial cells and prostate tumor cells, and the demonstration of the direct action of estrogens and the direct interaction of ligand AR and ligand ER in prostate cells, provides important information for designing new strategies in the development of new estrogen analogs for prostate cancer prevention and therapy.

Treatment of estrogen analogs in prostate tumor cells has been shown to produce a dose-dependent inhibition of androgen-induced tumor cell growth (144), and to directly inhibit prostate tumor cell growth via ER (149,171). This suggests that ligand ERs within prostate cells have biological significance and clinical implications via their regulation of the androgen-AR signaling pathway and other mechanisms. As the major side-effects of estrogen therapy in prostate cancer are cardiovascular side-effects and gynecomastia, which are related to hepatic first-pass metabolites and ERα actions, the development of ERβ specific ligands with few or no side-effects may provide a new generation of effective regimens for prostate cancer prevention and therapy.

Androgen-independent prostate cancer growth and metastasis is the major cause of prostate cancer mortality. The mechanisms of androgen-independency in prostate cancer cells can be subcategorized into two subtypes: AR-dependent and AR-independent (194). Most of the androgen-independent prostate tumor cells are AR-positive, and their growth may be accounted for the aberrant AR activation by other hormones due to AR amplification, mutation, or abnormal coregulator expression, etc. The direct inhibition of ligand-AR signaling by ligand-ER in prostate cells will have the potential not only to block the effects of AR signaling by androgens originating from the testes and adrenal, but also the effects of aberrant AR activation by other hormones (see Figure 6), providing an effective therapy for patients with androgen-independent prostate cancer.