(Received for publication, December 23, 1996)
From the Departments of Medicine and
§ Urology, St. Luke's/Roosevelt Hospital Center, New York,
New York 10019 and the College of Physicians and Surgeons, Columbia
University, New York, New York 10019
These experiments were designed to examine the relationship between the effects of steroid hormones mediated by classic intracellular steroid hormone receptors and those mediated by a signaling system subserved at the plasma membrane by a receptor for sex hormone-binding globulin. It is known that unliganded sex hormone-binding globulin (SHBG) binds to a receptor (RSHBG) on prostate membranes. The RSHBG·SHBG complex is rapidly activated by estradiol to stimulate adenylate cyclase, with a resultant increase in intracellular cAMP. In this paper we examine the effect of this system on a prostate gene product known to be activated by androgens, prostate-specific antigen. In serum-free organ culture of human prostates, dihydrotestosterone caused an increase in prostate specific antigen secretion. This event was blocked by the anti-androgens cyproterone acetate and hydroxyflutamide. In the absence of androgens, estradiol added to prostate tissue, whose RSHBG was occupied by SHBG, reproduced the results seen with dihydrotestosterone. Neither estradiol alone nor SHBG alone duplicated these effects. The estradiol·SHBG-induced increase in prostate-specific antigen was not blocked by anti-estrogens, but was blocked both by anti-androgens and a steroid (2-methoxyestradiol) that prevents the binding of estradiol to SHBG. Furthermore, an inhibitor of protein kinase A prevented the estradiol·SHBG-induced increase in prostate-specific antigen but not that which followed dihydrotestosterone. These data indicate that there is a signaling system that amalgamates steroid-initiated intracellular events with steroid-dependent occurrences generated at the cell membrane and that the latter signaling system proceeds by a pathway that involves protein kinase A.
The standard model of estrogen action posits that hormonal effects are mediated via the intermediacy of intracellular estrogen receptors (ER)1 (1-4). These receptors have nascent transcriptional activity that is unmasked by estrogens, thus allowing the transcription of specific genes. The exclusivity of this model in accounting for estrogen action is disputed by observations that estrogens may act by non-ER-mediated mechanisms (5-10). Furthermore, the plasticity of the model has had to be modified because of observations that the ER may initiate transcription in the absence of estrogens, so-called ligand-independent activation of transcription (2, 11-13). Although the data supporting the existence of ligand-independent activation of transcription is substantive, and also has been demonstrated for the progesterone receptor and vitamin D receptor (11, 14, 15), the physiological role of such activation and/or its relation to ligand-activated transcription remains to be clarified.
We have shown previously that estradiol (E2) participates in a signaling system that originates, not within the cell, but at the plasma membrane (5, 16). Through the intermediacy of the plasma protein, sex hormone-binding globulin (SHBG), it causes the generation of cAMP (17-19). In brief, unliganded SHBG binds to a receptor (RSHBG) on certain cell surfaces and the RSHBG·SHBG complex is rapidly activated by E2 to stimulate adenylate cyclase, with a resultant increase in intracellular cAMP. There is a paucity of information on events subsequent to the generation of cAMP by this system. In this paper we examine the effect of E2·SHBG·RSHBG on an androgen-responsive gene.
The gene for prostate-specific antigen (PSA) contains an androgen response element. After binding its cognate ligand, the androgen receptor (AR) interacts with this response element to initiate PSA mRNA transcription (20, 21) and secretion (22, 23). We show that, in the absence of androgens, E2 in concert with SHBG·RSHBG, acts at the cell membrane to cause secretion of PSA and that this effect is blocked by anti-androgens. This observation provides a first functional link between a classic steroid hormone receptor and a cell membrane-mediated steroidal effect.
Dihydrotestosterone (DHT), E2, and 2-methoxy-E2 were obtained from Steraloids, Wilton, MA. PKI, an inhibitor of protein kinase A (PKA) was purchased from Sigma. All other chemicals were obtained from Sigma or Boehringer Mannheim. Highly purified SHBG was prepared and evaluated for purity as described previously (16). It migrated as an uncontaminated doublet on SDS-polyacrylamide gel electrophoresis. Because SHBG is isolated with an equimolar concentration of DHT, it was stripped of steroids with dextran-coated charcoal before use (16, 24).
Prostate Tissue ExplantsHuman prostate tissue was obtained and handled as described previously (5). Prostate tissue was obtained at the time of transuretheral or open resection for benign prostatic hyperplasia and immediately brought to the laboratory under sterile conditions. Discolored portions were removed and the remaining tissue was divided into approximately 5-mm cubes. The tissue was placed in 60-mm Primaria culture dishes (Becton Dickinson Labware) in RPMI 1640 (Life Technologies, Inc.) with 5% fetal bovine serum containing 100 units/ml penicillin, 100 µg/ml streptomycin sulfate, and 0.25 µg/ml amphotericin, for 2-3 days. It then was minced into 1-mm3 portions and transferred to 16-mm wells in serum-free medium (0.5 ml of RPMI 1640) for about 18 h before beginning an experiment. All additions, e.g. SHBG, steroids, controls, were made in serum-free RPMI 1640.
Assay for Prostate-specific AntigenAt the conclusion of all experiments, the medium was harvested and assayed for PSA by an immunoradiometric assay, Diagnostic Products Corp., Los Angeles, CA. The sensitivity of the assay is 0.1 ng/ml, and it has an intraassay coefficient of variation of ~4.0%.
Although it is well known that the normal prostate secretes PSA,
the experimental data in vitro are largely based on the
prostate cancer cell line, LNCaP. As is the case for that cell line,
DHT caused an increase in PSA secretion in benign prostatic hyperplasia explants, Fig. 1. In LNCaP cells, E2 also
caused increases in the secretion of PSA (23). That effect was due to
the mutated AR in those cells which permits the binding of estrogens as
well as androgens (25). The normal human prostate contains an ER (26),
but unlike LNCaP cells, does not increase PSA secretion in response to
E2 (Fig. 1). However, in the presence of SHBG, E2 caused a dose-dependent increase in PSA
secretion that was as great as that observed in response to DHT (Fig.
1).
Because E2·SHBG stimulates RSHBG to increase
intracellular cAMP (5, 27), we examined the ability of increases in
intracellular cAMP, brought about either by forskolin or the addition
of cAMP to the medium, to affect PSA secretion. The effects of DHT and E2·SHBG were mimicked both by forskolin (which increases
intracellular cAMP) and a cell-permeant analog of cAMP,
8-(4-chlorophenylthio)-cAMP (Fig. 2). The response to
8-(4-chlorophenylthio)-cAMP was biphasic with a stimulation of PSA
secretion at 1 and 10 nM that diminished at 100 nM and disappeared at 1000 nM. We have observed
a similar biphasic effect of cAMP on the growth of a human prostate
cancer cell line (ALVA-41) (28). cAMP begins its signaling cascade, by
activating a protein kinase (PKA), the catalytic subunit of which then
phosphorylates one or more proteins, which results in a change in their
biological activity. Thus, if E2·SHBG increases PSA
secretion through this pathway and DHT does not, inhibition of the
activation of PKA should prevent E2·SHBG, but not DHT, motivated PSA secretion. That is precisely what occurs when PKA is inhibited with the PKA inhibitor, PKI (Fig. 3).
To evaluate the participation of the ER in the response of PSA to
E2·SHBG, anti-estrogens were added to prostate minces
together with E2 to determine if these agents altered the
effects of E2·SHBG. The increase in PSA caused by
E2·SHBG was blocked neither by tamoxifen nor by the pure
anti-estrogen ICI 164,384 (Fig. 4). However,
2-methoxy-E2, a hormonally inactive metabolite of
E2 that binds tightly to SHBG (29) and blocks
RSHBG-mediated cAMP generation (27), blocked the effect of
E2·SHBG on PSA secretion (Fig. 4).
The promoter of the PSA gene has an androgen response element, and PSA
secretion (22, 23) and the expression of PSA mRNA are
androgen-regulated (20, 30). Hence, we examined the effect of
hydroxyflutamide and cyproterone acetate, both potent anti-androgens, on the E2·SHBG-mediated increase in PSA secretion. As
expected (31, 32), both anti-androgens blocked the effect of DHT on PSA
secretion (Fig. 5). Surprisingly, they also blocked the
effect of E2·SHBG on PSA secretion. Although estrogens
can bind to and activate the mutated LNCaP androgen receptor (25, 33),
there is no evidence that this is the case in nonmalignant prostate cells. Since E2 is not exerting its effect by binding to
the AR, e.g. it is not its cognate ligand, the
E2-induced secretion of PSA observed in this study is
indicative of a ligand-independent activation of the AR.
The inhibition of transactivation of the estrogen and progesterone receptors by appropriate antihormones, in the absence of their cognate ligands, has been part of the evidence adduced to support the concept of ligand-independent activation of these receptors (reviewed in Ref. 15). Our demonstration of the blockade of an AR-mediated effect by an anti-androgen, in the absence of androgens, is consonant with observations first made by Culig et al. (34). They used prostate cancer cell lines, cotransfected with an AR expression vector and an androgen-responsive reporter, to show that, in the absence of androgens, AR-mediated gene transcription could be driven by keratinocyte growth factor, epidermal growth factor, or insulin-like growth factor I. The transactivation induced by each of these was inhibited by anti-androgens. More recently, Nazareth and Weigel (35) showed that forskolin transactivated the human AR cotransfected into CV-1 cells with an appropriate reporter construct. Like the response to androgens, the response to forskolin was blocked by anti-androgens. Furthermore, PKI (an inhibitor of PKA) markedly diminished the response not only to forskolin but also to androgens. Thus, these data not only confirm that AR can undergo ligand-independent activation, but suggest a role for the PKA pathway in androgen initiated activation of the AR in this model. Although the data in that communication are generally consonant with our observations, the inhibition of androgen-induced reporter activity by PKI is at odds with the inability of PKI to inhibit DHT-induced PSA secretion (Fig. 3). It is reasonable to hypothesize that this important variance may reflect the difference between observations made on prostate explants and those made on transfected, non-prostatic cell lines. Ikonen et al. (36) elicited synergism between androgens and activators of PKA (in cells cotransfected with an AR expression plasmid and a reporter containing two androgen response elements); however, they could not demonstrate ligand-independent activation of the AR. The reasons for these discrepancies are undoubtedly several but probably include the fact that they transfected cells with an expression plasmid containing the rat AR, whereas the other studies were all human AR-based.
Although the existence of the RSHBG system has been apparent for a number of years, it is only recently that studies dealing with its biology have appeared. DHT, but not E2, caused an increase in the rate of growth of ALVA-41 cells (a human prostate cancer cell line) in serum-containing media (37) but failed to do so in serum free media (28). However, the introduction of SHBG enabled both DHT and E2 to enhance the growth of these cells in serum free media. Furthermore, the increase in growth was as great as that seen with DHT in serum-containing media (28). Working with a breast cancer cell line (MCF-7), Fortunati et al. (38) demonstrated that the RSHBG system antagonized the E2-induced growth of these cells. The data supported their conclusion that this effect was not caused by the sequestration of E2 by SHBG. Growth is a complex phenomenon, and the mechanisms underlying these observations on ALVA-41 and MCF-7 cells will undoubtedly be difficult to sort out. The observations in this communication deal with a simpler system and should be more amenable to a dissection of mechanisms.
There are a number of general ways in which E2·SHBG·RSHBG-promoted increases in PSA secretion might be accomplished. PSA is synthesized and secreted by the prostate epithelial, but not the stromal, cell. However, although E2·SHBG·RSHBG-mediated increases in cAMP occur in the epithelial cell, both normal (5) and cancerous (16, 38), the increases are most robust in the prostate stromal cell. Thus, although all the elements (RSHBG, AR, and PSA) needed to participate in E2-induced increases in PSA are contained in the epithelial cell, the stromal cell may be involved. That the stromal cell products, keratinocyte growth factor, epidermal growth factor, and insulin-like growth factor I all have been shown to activate the AR (34), is consistent with such a mechanism. Furthermore, stromal cell conditioned medium enhanced DHT-induced PSA secretion by LNCaP cells (39). Taken together, these observations raise the possibility of stromal cell participation in the E2-induced increase in PSA secretion.
In summary, we have shown that E2 can activate a typical AR-mediated event, PSA synthesis and secretion. It does so by activating SHBG·RSHBG and makes clear that there is cross-talk between a classic intracellular steroid hormone receptor and a steroid hormone-engendered event at the cell membrane.
We thank Diagnostic Products Inc. for providing kits with which to measure PSA.