PSA is a useful biomarker for screening prostate cancer and for evaluating prostate cancer progression (Kantoff and Talcott, 1994
). The expression of PSA gene is induced by androgens via AR, and multiple AREs have been identified in the PSA gene promoter (Riegman et al, 1991
; Cleutjens et al, 1996
; Schuur et al, 1996
). As a result of its tissue specificity and androgen inducibility, human PSA gene has been used as a model system in studying androgen action in prostate gene expression and in prostate cancer research.
Using DHT-induced PSA transcription in cotransfection assays as a model system, the effects of ER-isoforms and various ER-ligands on modulation of androgen action were analyzed. The observation that E2 inhibited DHT-induced PSA transcription activity and MMTV-CAT activity via ERα is in agreement with the results of previous studies (Kumar et al, 1994
; Panet-Raymond et al, 2000
). More importantly, we demonstrate the ER-isoform and ER-ligand specificity in this androgen-estrogen interaction. Based on these results, ER-ligands are classified into 4 different categories related to their effects on androgen action (the Table). This classification is different from the pharmacological classification based on estrogenic activity. E2, αE2, and ICI all inhibit DHT action via ERα, although they are pharmacologically classified as an estrogen agonist, a weaker agonist, and a pure antagonist, respectively.
The observed ER-isoform–specific actions are consistent with the demonstration that ERα and ERβ possess distinctive action in the regulation of gene expression. Paech et al (1997)
have shown that ERα and ERβ signal in opposite ways in the presence of E2 from an AP1 site; with ERα, E2 activates transcription, whereas with ERβ, E2 inhibits transcription. Panet-Raymond et al (2000)
observed that E2 inhibits androgen action via ERα, but not via ERβ, a result that is in agreement with our current results. Taken together, these data indicate that ERα and ERβ possess differential actions in mediating ligand-specific activity, although they share high homology and common functional modalities.
An ER-ligand–specific modulation of DHT action is observed via either ERα or ERβ. This ligand specificity is consistent with the concept that different ER-ligands cause differential conformational changes in the ERs (Egner et al, 2001
; Nilsson et al, 2001
; Margeat et al, 2003
). The differential changes in ER conformation by various ligands may result in a recruitment of various transcriptional factors or coregulators (Shang and Brown, 2002
; Margeat et al, 2003
) and may process different downstream actions. This explains the differential ER-ligand effects, but it does not explain why various ER-ligands, including agonist, partial agonist, and pure antagonist, all lead to the inhibition of DHT action via ERα. The observation that both E2 and ICI inhibit DHT action is consistent with previous demonstrations that both E2 and ICI induce the same target gene transcription (Kim et al, 2003
) and cause proteasome-dependent ERα degradation (Preisler-Mashek et al, 2002
A stereo-specificity of αE2 and E2 in modulating DHT action is observed. Like E2, αE2 binds to ERs and processes estrogenic activity, although in a much weaker fashion (Edwards and McGuire, 1980
). Unlike E2, which inhibits DHT-induced PSA transcription via ERα only in cotransfection assays, αE2 inhibits DHT action via either ERα or ERβ.
The importance of ERα functional domains in this estrogen-androgen interaction is analyzed using ERα mutants. The E2-inhibition of DHT action via ERα mainly involves the DBD. However, gel shift analysis shows that neither ERα nor ERβ directly binds to any of the 3 functional AREs in the human PSA promoter (our unpublished data), indicating that direct competition of ARE-binding between ligand-AR and ligand-ER is unlikely. On the other hand, αE2 inhibition of DHT action via ERα is completely eliminated when the ERα F-domain is deleted. The F-domain is highly variable in the nuclear receptor superfamily, and there is little homology between ERα and ERβ. The ERα F-domain potentially contains helix-13 and β-strand motifs (Kumar et al, 1987
; Kim et al, 2003
). The functional significance of the ERα F-domain is just emerging. It is required for the agonist/antagonist action of tamoxifen (Schwartz et al, 2002
) and the E2 activation of ERα/SP1 pathway (Kim et al, 2003
The biological significance of this androgen-estrogen interaction is exemplified by determining the effects of αE2 and E2 on the regulation of DHT induction of PSA gene expression and cell growth in LAPC-4 prostate tumor cells. Previous studies show that LAPC-4 cells express a wild-type AR and are sensitive to androgen stimulation (Klein et al, 1997
). We here demonstrate that LAPC-4 cells also express ERs with predominantly ERβ and low level of ERα, which is consistent with previous demonstration in other prostate tumor cells (Lau et al, 2000
). As expected, treatment with DHT in LAPC-4 cells increases PSA expression and cell growth (see ). These DHT effects are significantly inhibited by coadministration of E2, or αE2 in a dose-dependent manner. The inhibitory effects of these estrogen analogs are as potent as cyproterone acetate, an AR antagonist, although they act via different mechanisms.
The estrogen inhibition of DHT-induced PSA gene expression and cell growth in LAPC-4 cells may be mediated via ERs, as both ERα and ERβ are expressed in these cells. If this is the case, then the data obtained in LAPC-4 cells are consistent with the cotransfection studies, in which both E2 and αE2 inhibit DHT-induced PSA transcription activity when both ERα and ERβ are cotransfected in the cells. This hypothesis is further supported by our recent studies showing that the estrogen inhibition of AR action is mainly mediated via ERβ using RNA interference analysis (unpublished data). However, it should be noted that the receptor-isoform– and ligand-specific modulation of DHT action by estrogens needs to be further evaluated in prostate cells, and there is no direct linkage between DHT-induced PSA gene expression and prostate tumor cell growth. This notion is supported by comparing the dose-response data between estrogen inhibition of DHT-induced PSA expression and estrogen inhibition of DHT-induced cell growth in LAPC-4 cells (see ), in which differential dose-response curves are observed. Although PSA has been shown to degrade insulin-like growth factor binding protein 3, thereby leading to a potentiation of insulin-like growth factor action on cell growth by in vitro analysis (Cohen et al, 1994
), such PSA action has not been demonstrated in vivo, and the significance of PSA in prostate pathogenesis remains to be determined.
In vivo studies have shown that estrogens can either inhibit or enhance androgen effects, although the mechanisms of these apparent opposing estrogen actions are not explored. The present analyses using in vitro cotransfection assays provide a reasonable, potential explanation of these seemingly opposing estrogen actions. Since the outcome of androgen-estrogen interaction is cell-type, ER-isoform– and ligand-specific, the final result can be either inhibitory or stimulatory depending on the ER-isoforms expressed in the cells, as well as the ligands used in the experiment. Further study using ER knockout animals will provide valuable information for androgen-estrogen interaction in the prostate.
In summary, estrogens have been used in hormonal therapy of prostate cancer, presumably by inhibiting testosterone biosynthesis via the negative feedback of the hypothalamus-pituitary-gonadal axis (Huggins and Hodges, 1941
). Our current study provides an additional mechanism for estrogens acting directly through ERs to antagonize androgen actions within the cells. The demonstration of ER-isoform and ligand specificity mediated via differential mechanisms in androgen-estrogen interaction explains the seemingly opposing estrogen action in the prostate and provides important information for the development of new estrogen analogs in the prevention and control of prostate cancer.