Synthesized steroidal androgens, due to their ability to mimic the actions of their endogenous counterparts, have been used clinically as valuable therapeutic agents to target a variety of male and female disorders resulting from androgen deficiency. The principle clinical indication of androgens is as replacement therapy for hypogonadal men (Conway et al., 1988; Wu, 1992). Other documented clinical uses of androgens include delayed puberty in boys, anemias, primary osteoporosis, hereditary angioneurotic edema, endometriosis, estrogen receptor-positive breast cancer, and muscular diseases (Wu, 1992; Bagatell and Bremner, 1996; Nieschlag, 1996; Bhasin and Tenover, 1997). Also, androgens have been investigated as hormone replacement therapy for aging men and for regulation of male fertility (Wu, 1992; Tenover, 1997).

Since the discovery of the therapeutic benefits of testosterone in the 1930s, a variety of androgen preparations have been introduced and tested clinically. Unfortunately, virtually all currently available androgen preparations have severe limitations (Wu, 1992; Bhasin and Bremner, 1997). Unmodified testosterone is impractical for oral administration due to its low systematic bioavailability (Handelsman et al., 1990). Testosterone esters (e.g., testosterone propionate and testosterone enanthate) are presently the most widely used testosterone preparations, usually administered by intramuscular injection in oil vehicles (Snyder and Lawrence, 1980; Velazquez and Bellabarba Arata, 1998). A prolonged duration of action is achievable with these esters. However, they produce highly variable testosterone levels. 17α-Alkylated testosterones (e.g., methyltestosterone and oxandrolone) can be given orally. Nevertheless, they often cause unacceptable hepatotoxicity and are less efficacious; hence, they are not recommended for long-term androgen therapy (Heywood et al., 1977; Ishak and Zimmerman, 1987; Velazquez and Bellabarba Arata, 1998). Another common concern about steroidal androgens is the undesirable effects resulting from the cross-reactivity of the androgens or their in vivo metabolites with steroid receptors other than the androgen receptor (AR) (Wilson et al., 1980; Bhasin and Bremner, 1997).

During studies with affinity ligands for the AR, our group discovered a group of nonsteroidal androgens that are structural derivatives of bicalutamide and hydroxyflutamide, two known antiandrogens (Dalton et al., 1998; Mukherjee et al., 1999). Other laboratories have also reported the identification of nonsteroidal compounds that possess androgen activity (Dalton et al., 1998; Hamann et al., 1999; Negro-Vilar, 1999). The discovery of these nonsteroidal androgens offers an opportunity for the development of a new generation of selective androgen receptor modulators (SARMs) superior to current steroidal androgens. Theoretically, SARMs are advantageous over their steroidal counterparts in that they can obtain better receptor selectivity and allow greater flexibility in structural modification. Thus, SARMs can potentially avoid the undesirable effects caused by receptor cross-reactivity and achieve superior pharmacokinetic properties.

Subsequent to our initial discovery of several nonsteroidal androgens, our laboratories designed and synthesized multiple series of nonsteroidal compounds, and explored the structure-activity relationships for androgenic and anabolic activities, both in vitro and in vivo (He et al., 2002; Yin et al., 2003a,b). According to results from these structure-activity relationship studies, we designed a group of novel nonsteroidal compounds (Fig. 1) that were structurally optimized. We report herein the results of our studies to examine the in vitro AR binding affinity and the androgenic and anabolic activities of these new compounds in an animal model. Two potent and tissue-selective SARMs were identified from these structurally similar compounds, and they are members of a promising new class of drug candidates for further development.

In accordance with literature reports (Saksena and Chaudhury, 1970; Teutsch et al., 1994; Battmann et al., 1998), we observed significant decreases in the weights of prostate, seminal vesicles, and levator ani muscle in castrated, vehicle-treated rats (Figs. 2-​-5).5). The weights of the prostate, seminal vesicles, and levator ani muscle in castrated rats were 6.2, 8.1, and 40.9%, respectively, of those in intact animals. The reduction in masses of these androgen-targeted organs in castrated animals is the result of ablation of endogenous androgen production (Saksena and Chaudhury, 1970). Exogenous administration of TP, an androgenic and anabolic steroid, increased weights of the prostate, seminal vesicles, and levator ani muscle in castrated rats (Fig. 2). The increases in organ weights induced by TP were dose rate-dependent.

Figure 3 shows that compound S-1 had no significant effect on prostate, seminal vesicles, and levator ani muscle in castrated animals at 0.1 and 0.3 mg/day, but significantly stimulated the growth of these organs at higher doses. The weights of prostate, seminal vesicles, and levator ani muscle were maximally restored by S-1 to 14.9, 13.4, and 74.3%, respectively, of those in intact animals. The ED₅₀ values of S-1 in prostate, seminal vesicle, and levator ani muscle, as obtained by nonlinear regression analysis of dose-response relationships, were 0.42 ± 0.04, 0.38 ± 0.26, and 0.44 ± 0.01 mg/day, respectively (Fig. 3B; Table 2), corresponding to 1.63, 1.47, and 1.70 mg/kg, respectively, based on the mean body weight of S-1-treated animals at the end of the study. The elevations in organ weights by S-1 demonstrated its androgenic and anabolic activities in animals. In comparison to TP, corresponding dose rates of S-1 induced significantly smaller increases in the weight of the prostate and seminal vesicles but a similar degree of increase in levator ani muscle weight (compare Fig. 2A with ​with3A).3A). This result denoted the tissue selective androgenic and anabolic activity of S-1 in rats. The selectivity was also demonstrated by its relative efficacy compared with TP (Table 2). The relative efficacy in maintaining levator ani muscle weight was 0.72, much higher than the relative efficacies in maintaining prostate and seminal vesicle weights, which were less than 0.20.

Despite their high AR binding affinity, compounds S-2 and S-3 failed to exert any significant effect on the weights of prostate, seminal vesicles, and levator ani muscle in castrated animals, with dose rates up to 1 mg/day (Fig. 4). This suggests that rapid metabolism or clearance of these compounds led to lower plasma concentrations of these drugs, and thus no pharmacological activity. Likewise, compound R-1 (the stereoisomer of S-1), at 1 mg/day, produced no apparent effect on the weights of prostate, seminal vesicles, and levator ani muscle in castrated animals, demonstrating the stereoselective pharmacological action of these compounds.

Compound S-4 (Fig. 5) caused dose-dependent stimulation of growth in prostate, seminal vesicles, and levator ani muscle, with their weights in castrated animals being maximally promoted to 33.8, 28.2, and 101% of intact controls, respectively. Nonlinear regression analysis of dose-response relationships showed that the ED₅₀ values of S-4 were 0.43 ± 0.01, 0.55 ± 0.02, and 0.14 ± 0.01 mg/day in prostate, seminal vesicles, and levator ani muscle, respectively (Fig. 5B; Table 2), corresponding to 1.62, 2.07, and 0.53 mg/kg, respectively, based on the mean body weight of S-4-treated animals at the end of the study. These results clearly revealed the androgenic and anabolic activities of S-4 in animals. In particular, S-4 exhibited potent and efficacious anabolic activity, as indicated by its ability to fully maintain the levator ani muscle weight in castrated animals at the same level as intact controls, at a dose rate as low as 0.3 mg/day (Fig. 5A). The relative potency and efficacy of S-4 in androgenic tissues were less than 0.3 and 0.4, respectively, compared with 1 for TP, whereas its relative potency and efficacy in the levator ani muscle was 1.07 and 0.97, respectively, compared with 1 for TP (Table 2).

Table 2 compares the androgenic and anabolic activities of S-1 and S-4, two compounds that exhibited in vivo functional activity in the present study, with those of TP. The efficacy for androgenic activity of S-4 (as indicated by relative efficacies in prostate and seminal vesicle) was about twice that of S-1, but the potency for androgenic activity (as indicated by relative potencies in prostate and seminal vesicle) was similar between these two compounds. As to anabolic activity, S-4 displayed much higher efficacy (as indicated by relative efficacy in levator ani muscle) and 2-fold greater potency (as indicated by relative potency in levator ani muscle) than S-1. These results suggest the greater selectivity of S-4 toward the anabolic target organ.

We also determined the serum levels of LH and FSH in animals that received S-1 and S-4, and compared them with the levels of these hormones observed in the intact, castrated, or TP-treated animals. As shown in Table 3, castration led to a significant elevation in FSH and LH levels, compared with intact animals. TP showed no dose-dependent effect on castration-induced change in FSH, but partially inhibited the castration-induced increase in LH levels at higher doses. The activities of S-4 on LH and FSH were similar to those produced by TP. S-1 and S-4 partially suppressed LH production at dose rates of 0.5 mg/day or higher. However, it is important to note that S-1 and S-4 did not suppress LH production at the dose levels needed to produce the desired pharmacological effects in the levator ani muscle or prostate. Interestingly, S-1 also partially suppressed FSH production at dose rates of 0.5 mg/day or higher. The FSH suppression noted at higher doses of S-1 suggested that this compound might interact with other steroid receptors, most probably progesterone receptors, in addition to the AR. Although statistically significant differences were noted in some instances, GH, AST-SGOT, ALT-SGPT, and serum lipids (including cholesterol, high-density liprotein, and triglyceride) were all within normal ranges for drug-treated animals. No drug- or dose-related changes in these indices were observed.

Despite structural similarities, this series of ether-carrying compounds exhibited diverse in vitro and in vivo activity profiles. The S-isomers of compounds 1, 2, 3, and 4 displayed moderate to high binding affinity for the AR, whereas the R-isomer of compound 1 had poor receptor binding. This finding was consistent with our previous observation regarding the stereoselective AR binding of nonsteroidal ligands (Mukherjee et al., 1996, 1999). Compounds S-1, S-3, and S-4 were further characterized as AR agonists with the in vitro cotransfection assay. The failure of S-2, a moderate AR binder, to stimulate AR-mediated gene transcription confirmed our previous finding that high receptor binding affinity is a prerequisite for agonist activity (Yin et al., 2003a). When tested in the castrated rat model, S-1 and S-4 demonstrated potent in vivo functional activity, and compounds S-2 and S-3 were inactive. Specifically, S-4 produced the greatest androgenic and anabolic activity in animals, with anabolic activity greater than that of TP. S-1 had a similar degree of anabolic activity as TP, but had much less androgenic activity. Interestingly, the ED₅₀ values for S-1 in prostate, seminal vesicle, and levator ani muscle were approximately the same (i.e., about 0.4 mg/day), whereas S-4 demonstrated more than 2-fold greater potency in levator ani muscle compared with prostate and seminal vesicle, as indicated by the ED₅₀ values (Table 2). The distinction in functional activities in vivo among the four structurally related compounds could be caused by difference in any of numerous factors, including intrinsic activity, in vivo disposition and metabolism, or intracellular signaling pathway. Further studies to explore the physicochemical, physiological, and cellular/molecular determinants for nonsteroidal androgenic and anabolic activity will lead to insights into the mechanism of action of these nonsteroidal agents, and thereby provide a basis for future structural optimization.

The in vitro cotransfection assay is generally regarded as a valuable tool for screening of nonsteroidal AR ligands. With this assay, compounds S-1, S-3, and S-4 were successfully identified as potential AR agonists. However, as demonstrated in the animal study, S-3 did not show any measurable in vivo functional activity. Thus, the observation of in vitro agonist activity in the cotransfection assay can be but is not always predictive of in vivo activity. The pharmacological activity in vivo is determined not only by the ability of the compound to interact with the receptor, but also limited by complicated factors governing the accessibility of the compound to the effect site, such as disposition and metabolism. To fully predict the in vivo behavior and understand the structure-activity relationships, it is necessary to perform further studies examining the pharmacokinetics and metabolism of the compound. As a result of these and other studies, we abandoned use of the in vitro cotransfection assay in favor of in vivo pharmacologic assessment for discovery of SARMs.

The tissue-selective anabolic activity exhibited by S-1 and S-4 validated the feasibility of developing SARMs as a new generation of androgens. The possible mechanisms underlying the tissue-selectivity of these agents could be tissue-specific recruitment of cofactors/corepressors during the AR signaling pathway, or very likely for our nonsteroidal ligands, their distinct in vivo disposition from testosterone and its ester derivatives. The effects of testosterone in certain tissues, including most accessory reproductive organs and skin, are amplified through local conversion to DHT, the more potent bioactive form, by 5α-reductase (Mooradian et al., 1987). Nevertheless, testosterone exerts direct effects in the testis, skeletal muscles, and bone (Mukherjee et al., 1996). For nonsteroidal ligands, their actions in accessory reproductive organs such as prostate would not be amplified as they are for testosterone; therefore, such a nonsteroidal androgen with equivalent activity to testosterone on bone and muscle would likely have less activity on prostate or other accessory reproductive organs than testosterone.

Compounds S-1 and S-4 are the first nonsteroidal androgens with in vivo functional activity among our series of compounds. More significantly, the discovery of these two in vivo functional drug candidates represents a major progress toward the development of therapeutically useful SARMs. SARMs, like the clinically available selective estrogen receptor modulators, would offer unique therapeutic advantages over their steroidal counterparts. The tissue selectivity of these agents offers an exciting opportunity to differentially regulate the androgen effects in various target tissues, thus minimizing the interference to normal physiological processes while targeting desirable therapeutic goals. For example, SARMs with potent anabolic activity but minimal androgenic activity would be ideal for the treatment of patients who bear muscular diseases (such as sarcopenia or trauma-induced muscle wasting) but are contraindicated for androgenic stimuli (such as for aging population or prostate cancer patients). In perspective, not only could SARMs be used as superior alternatives to current steroidal androgens in therapy of male hypogonadism but also they could expand the scope of androgen therapy to include wasting syndromes, aging-related disorders due to declined androgen levels, male fertility regulation, and other androgen deficiency-related diseases.