Anabolic-androgenic steroid interaction with rat androgen receptor in vivo and in vitro: A comparative study
Boris I. Feldkoren, Stefan Andersson,
Show more
doi:10.1016/j.jsbmb.2004.12.036
Get rights and content
Abstract
Anabolic steroids are synthetic derivatives of testosterone and are characterized by their ability to cause nitrogen retention and positive protein metabolism, thereby leading to increased protein synthesis and muscle mass. There are disagreements in the literature in regards to the interaction of anabolic steroids with the androgen receptor (AR) as revealed by competitive ligand binding assays in vitro using cytosolic preparations from prostate and skeletal muscle. By use of tissue extracts, it has been shown that some anabolic steroids have binding affinities for the AR that are higher than that of the natural androgen testosterone, while others such as stanozolol and methanedienone have significantly lower affinities as compared with testosterone. In this study we show that stanozolol and methanedienone are low affinity ligands of the rat recombinant AR as revealed by a ligand binding assay in vitro, however, based on a cell-based AR-dependent transactivation assay, they are potent activators of the AR. We also show that a single injection of stanozolol and methanedienone causes a rapid cytosolic depletion of AR in rat skeletal muscle. Based on these results, we conclude that anabolic steroids with low affinity to AR in vitro, can in fact in vivo act on the AR to cause biological responses.
Keywords
Androgen receptor; Skeletal muscle; Anabolic steroid; Methanedienone; Stanozolol; Methyltrienolone; 17α-Methyltestosterone; Testosterone
1. Introduction
Anabolic steroids are synthetic derivatives of testosterone and are characterized by their ability to cause nitrogen retention and positive protein metabolism, thereby leading to increased protein synthesis and muscle mass [1], [2], [3] and [4]. The action in vivo of anabolic steroids on skeletal muscle metabolism may also include secondary effects as a result of increased growth hormone secretion and insulin-like growth factor 1 production [5] and [6].
In the past 50 years, over a thousand steroid compounds have been synthesized in an effort to dissociate the protein anabolic properties from the unwanted androgenic effects, which are characterized by growth stimulation of the male internal genitalia (prostate and seminal vesicles) and development of secondary sex characteristics [1] and [7]. It is interesting to note that the synthesis of a compound which demonstrates a complete dissociation between anabolic and androgenic properties has to date not been achieved, hence, this class of substances should more properly be termed anabolic-androgenic steroids (AS).
A general observation is that the potency in vivo of endogenous androgens as well as synthetic AS correlates with their affinity in vitro to the androgen receptor (AR), a nuclear hormone receptor whose gene is located on the X-chromosome [8]. Examples of exceptions from this notion are the AS stanozolol and methanedienone. These AS are very efficacious in eliciting a protein anabolic response in vivo [9], [10] and [11], however, they fail to efficiently compete with radiolabeled ligand in classical binding assays using cytosolic preparations from skeletal muscle and prostate [12] and [13]. Consequently, these observations have led to the proposition that these steroids may exert their biological effects by binding to a nuclear hormone receptor distinct from the classical AR. It should be noted, however, that it is conceivable that artifacts could be caused by contaminating blood or cellular proteins, or membrane lipids that bind these particular steroids and thus quench the unlabeled steroid in a competitive in vitro ligand binding assay using tissue extracts [14].
In an effort to address these differences reported in the literature, we have in the present study compared the properties of select AS by using three different systems: (1) a recombinant AR ligand binding in vitro assay; (2) a cell based AR-dependent transactivation assay; and (3) an in vivo assay based on steroid induced cytosolic AR depletion in skeletal muscle to better determine the affinities of both endogenous and anabolic-androgenic steroids.
2. Materials and methods
2.1. Reagents and animals
[17α-Methyl-3H]-methyltrienolone (86 Ci/mmol) and unlabeled methyltrienolone (17β-hydroxy-17α-methyl-4,9,11-estratrien-3-one) were purchased from NEN Life Science Products Inc. (Boston, MA). Testosterone (17β-hydroxy-4-androsten-3-one) and methyltestosterone (17β-hydroxy-17α-methyl-4-androsten-3-one) were obtained from Steraloids Inc. (Wilton, NH). Stanozolol (17α-methyl-5α-androstano[3,2-c]pyrazol-17β-ol), methanedienone (17β-hydroxy-17α-methyl-1,4-androstadien-3-one), dextran and activated charcoal were purchased from Sigma Chemical Co. (St. Louis, MO). All other chemicals were analytical grade products. Male Wistar rats (160–180 g) were maintained with food and water ad libitum, and treated in accordance with the guidelines set forth by the Animal Welfare Information Center. Protocols were approved by the Institutional Animal Care and Use Committee at the University of Texas Southwestern Medical Center at Dallas. Indicated amounts of steroids dissolved in dimethyl sulfoxide were injected intraperitoneally (0.1 ml/100 g body weight). One hour after steroid injection, the animals were euthanized by using an excess of CO2 followed by collection of skeletal muscle (m. quadriceps) for cytosol preparations.
2.2. Cell culture and transient transfection
The HEK-293 and CV-1 cell lines were maintained in Dulbecco's Modified Eagles Medium (DMEM) supplemented with 10% fetal bovine serum, 10 mM Hepes (pH 7.0) and 100 IU penicillin and 100 μg/ml streptomycin. HEK-293 cells were plated into 100 mm dishes and were transfected at ∼60% confluency with 10 μg DNA per dish with a cytomegalovirus promoter based mammalian expression plasmid [15] encoding the full-length rat AR (pCMV-rAR) [16] using a calcium phosphate procedure [17]. For transactivation assays, CV-1 cells were plated in 24-well tissue culture dishes in DMEM supplemented with 5% (v/v) dextran/charcoal treated fetal bovine serum and transfected with 15 ng of pCMV-rAR, 600 ng of a mouse mammary tumor virus luciferase reporter vector (pMMTV-luc) and 60 ng of control β-galactosidase plasmid (pCMX-β-gal) (gifts from David Mangelsdorf) [18].
2.3. Preparation of skeletal muscle cytosol
Two grams of muscle tissue was cleaned free of fat and minced using scissors followed by homogenization in 3.5 ml of buffer A (20 mM Tris–HCl pH 7.4, 25 mM KCl, 1.5 mM EDTA, 10 mM sodium molybdate, 10% (v/v) glycerol and 2 mM β-mercaptoethanol) containing 1% (w/v) dextran-coated charcoal (DCC) by use of an Ultra-Turrax homogenizer. After a 1 h centrifugation at 100,000 × g, the resulting supernatant (25–29 mg protein/ml), referred to as cytosol (S100), was used for determination of specific binding of [3H]-methyltrienolone at 2 nM final concentration which was selected on the basis of a previously published saturation study [19]. Triplicate samples were assayed for total and nonspecific binding by incubation for 16 h at 4 °C, followed by DCC treatment. Specific binding was expressed as fmol/mg cytosolic protein and calculated as described below. Protein concentration was determined by a Coomassie Blue based method (Pierce, Rockford, IL).
2.4. Saturation binding and competitive binding studies of recombinant AR
After transfection for 16 h, the medium was changed and the HEK-293 cells were incubated for an additional 24 h. Cells were washed twice with PBS and scraped off the plate into buffer A. After passing 20 times through a 25 gauge needle, the cell extract was centrifuged for 1 h at 100,000 × g, and the resulting supernatant (0.27–0.30 mg protein/ml) was used for saturation and competitive binding studies.
Determination of equilibrium binding parameters (Kd and Bmax) for the recombinant rat AR was performed at [3H]-methyltrienolone concentrations ranging from 0.06 to 4.0 nM in a 16 h incubation at 4 °C. Non-specific binding was assessed by using a 1000-fold molar excess of unlabeled methyltrienolone in the assay. Unbound [3H]-methyltrienolone was separated from bound compound by adsorption to dextran-coated charcoal (DCC) in final concentration of 0.6% (w/v) and specific binding was calculated as a difference between total and nonspecific binding. The competitive effect of respective steroid on specific binding of [3H]-methyltrienolone was studied at a concentration of 2 nM [3H]-methyltrienolone in the presence of 1–1000-fold molar excess of respective unlabeled steroid.
2.5. AR activation assay
After CV-1 cells were transfected for 16 h, the medium was changed to DMEM without serum and steroids in ethanol (1 μl/well) were added to a final concentration of 0.2 pM to 5 nM. After a 24 h incubation, the cells were washed with PBS and harvested in 100 μl lysis buffer (Promega, Madison, WI). Ten microlitres of cell lysate from each well was used for luciferase (Promega) and β-galactosidase (Tropix, Bedford, MA) activity assays by using a Monolight 2010 luminometer (Analytical Luminescence Laboratory, Ann Arbor, MI). Relative luciferase activity was calculated by dividing luciferase activity by β-galactosidase activity. All values are expressed as mean ± S.D. of measurements derived from triplicate wells.
2.6. Data analysis
The curve fittings were performed using the SigmaPlot 4.0 computer program. The binding parameters of expressed rat AR were calculated by computer fitting the specific binding (B) data to equation: B = Bmax × F/(F + Kd), where F was the concentration of free [3H]-methyltrienolone in the incubate, or using Scatchard transformation B/F = (Bmax/Kd) − (B/Kd). Competitive binding data were fitted to equation B = B0(1 − C/(IC50 + C)), where B0 was specific binding [3H]-methyltrienolone in the absence of unlabeled ligand, and C was ligand concentration to obtain an IC50 value, i.e., the ligand concentration that reduced specific binding by 50%. The equilibrium dissociation constant (Ki) for ligands was calculated using equation Ki = IC50/{1 + T[(B0/F0) + 2]/[2Kd((B0/F0) + 1)] + (B0/F0)} + Kd(B0/F0)/[(B0/F0) + 2], where T and F0 is total and free concentrations of [3H]-methyltrienolone, respectively [20]. Transcription activation data were fitted to a five parameter logistic function that gives maximal values of regression coefficients and EC50 for each steroid was calculated, as the concentration having half maximal effect of activation.
3. Results
The different AS used in the present study are depicted in Fig. 1. In an effort to compare their affinity for the AR in vitro, the full-length recombinant rat AR was over-expressed in HEK-293 cells and then in vitro ligand binding assays were performed using cytosolic protein extracts from these HEK-293 cells using 3H-methyltrienolone as radiolabeled ligand. Fig. 2 shows the presence of saturable high affinity binding sites for 3H-methyltrienolone, with a calculated equilibrium dissociation constant (Kd) of 0.25 nM and maximal number of binding sites (Bmax) of 157 fmol/mg protein. As shown in Fig. 3, the specific binding of 3H-methyltrienolone to the recombinant AR decreased when different unlabeled steroids were added to the ligand binding assay. From the inhibition curves depicted in Fig. 3 the respective binding affinities (Ki) for the different steroids were calculated ( Fig. 4). Methyltrienolone possessed the highest affinity with a Ki of 0.20 nM, in agreement with its apparent Kd of 0.25 nM; the Ki for testosterone and 17α-methyltestosterone were 0.80 and 0.90 nM, respectively. And interestingly, both stanozolol and methanedienone were the least effective competitors with a Ki for stanozolol of 4.5 nM and methanedienone of 5.0 nM.
Full-size image (12 K)
Fig. 1.
Steroid structures.
Figure options
Full-size image (6 K)
Fig. 2.
Saturation analysis of 3H-metyltrienolone binding to recombinant rat AR. The determinations were performed with cytosol from HEK-293 cells over-expressing the recombinant rat AR. The data points representing specific binding were calculated by subtracting non-specific binding (<20% of total binding) from total binding as described in Section 2. The inserted figure shows transformation data represented as a Scatchard plot.
Figure options
Full-size image (7 K)
Fig. 3.
In vitro inhibition of 3H-metyltrienolone binding to recombinant rat AR. Duplicate aliquots of 293 cell cytosol were incubated with 2 nM 3H-methyltrienolone in the presence of the indicated amounts of steroids. Non-specific binding (assessed by 2 nM 3H-methyltrienolone binding in the presence 1000-fold molar excess of unlabeled methyltrienolone) was subtracted in all instances. Results are expressed as % of control (4200 dpm, average of triplicate). (○) Testosterone; (●) methyltrienolone; (▴) 17α-methyltestosterone; (■ ) stanozolol; (♦) methanedienone.
Figure options
Full-size image (17 K)
Fig. 4.
Binding affinity of various steroids to the recombinant rat AR. The data presented in Fig. 3 were analyzed to calculate the equilibrium dissociation constant Ki for each steroid as described in Section 2. T, testosterone; MT, methyltrienolone; 17α-MeT, 17α-methyltestosterone; S, stanozolol; MA, methanedienone.
Figure options
To assess further whether these steroids were capable of activating the AR, a cell-based transactivation system was employed consisting of the full-length recombinant rat AR and a reporter plasmid which contains an androgen response elemement. Methyltrienolone was found to be the most effective transcriptional activator with an EC50 of 5 pM, in agreement with its high affinity in vitro (Fig. 5). The calculated EC50 values for the other steroids were 44 pM for 17α-methyltestosterone, 52 pM for stanozolol, and 79 pM for methanedienone; all four steroids induced the same level of maximum transactivation.
Full-size image (6 K)
Fig. 5.
Effect of various steroids on the recombinant rat AR-mediated transcriptional activation of the MMTV-luciferase reporter gene. The data points represent the average value of four independent experiments. Luciferase activity was normalized to β-galactosidase activity in individual wells to correct for transfection efficiency. Values were obtained are expressed as % of maximal relative luciferase activity as described in Section 2. Symbols assigned to steroids are the same as in Fig. 3.
Figure options
Testosterone was not tested because it is known to be metabolized to the inactive androgen androstenedione by an oxidative 17β-hydroxysteroid dehydrogenase present in CV-1 cells [21]. Hence, in intact cells, both stanozolol and methanedienone are potent activators of the AR. It should be mentioned that similar EC50 values were obtained when the transactivation experiments were performed with HEK-293 cells, as well as when the human AR was utilized. Control experiments in which 5 nM steroid was incubated with cells only transfected with the reporter plasmid resulted in low relative luciferase activities that were comparable to those obtained with cells containing reporter plasmid and recombinant AR in the absence of steroid. Therefore, the increased transactivation in response to added steroid was mediated by the presence of recombinant AR.
To investigate whether the selected steroids had the ability to activate the AR in vivo, we employed a cytosolic AR depletion assay developed by Feldkoren and coworkers [19]. The steroid dose-range for the endogenous androgen testosterone and the synthetic AS stanozolol was assessed 1 hour after treatment. This time-point was chosen based on a previous study which described the effect of exogenous testosterone on AR levels in skeletal muscle cytosol prepared from intact male rats. The data in Fig. 6 shows that a low amount of testosterone (0.1 mg/kg) had no statistically significant effect on the numbers of cytosolic AR, whereas, higher amounts of testosterone and stanozolol (1 mg/kg and 2.5 mg/kg) induced maximal depletion of cytosolic AR. Based on these data, a 0.3 mg/kg body weight dose was selected to compare the ability of the five steroids to deplete cytosolic AR measured one hour after treatment. Table 1 shows that methyltrienolone resulted in a 67% reduction of androgen binding sites in muscle cytosol (0.7 fmol/mg protein versus 1.3 fmol/mg protein for vehicle alone). Interestingly, stanozolol possessed nearly the same activity (44% depletion) as testosterone, followed by methanedienone, which caused a 33% reduction in binding sites. 17α-Methyltestosterone demonstrated the lowest degree of cytosolic AR depletion (11%) of all of the AS.
Full-size image (20 K)
Fig. 6.
Effect of testosterone and stanozolol on AR levels in rat skeletal muscle cytosol. Rats were injected intraperitoneally with 0.1 mg/kg (four animals), 1 mg/kg (four animals), 2.5 mg/kg (three animals) of testosterone (T), 1 mg/kg (four animals), 2.5 mg/kg (three animals) of stanozolol (S) or with vehicle (DMSO) (six animals). AR levels were measured in skeletal muscle cytosol 1 h post-treatment by a 3H-methyltrienolone binding assay as descried in Section 2.
Figure options
Table 1.
AR levels in rat skeletal muscle cytosol after intraperitoneal injection of AS
Steroid AR binding (fmol/mg protein) Depletion (% of control)
Vehicle (6)a 1.3 ± 0.1 0
Testosterone (4)a 0.8 ± 0.2 55
17α-Methyltestosterone (4)a 1.2 ± 0.1 11
Methanedienone (4)a 1.0 ± 0.1 33
Stanozolol (6)a 0.9 ± 0.1 44
Methyltrienolone (3)a 0.7 ± 0.1 67
a
Number of animals. Cytosol was prepared from skeletal muscle 1 h after an intraperitoneal injection of 0.3 mg steroid/kg body weight. AR levels were measured by a ligand binding assay using saturating amounts of 3H-methyltrienolone (2 nM) as described in Section 2.
Table options
4. Discussion
Many aspects of the physiological effects of AS such as nitrogen retention have been studied extensively, however, their action at the molecular level remain to a large extent unexplained [3], [4] and [6]. It is well established that the endogenous androgens testosterone and 5α-dihydrotestosterone as well as the AS methyltrienolone exert their mechanism of action by direct binding to the ligand-binding domain of the AR, thereby, initiating a series of events including recruitment of co-activators culminating in the activation of target genes [21], [22] and [23]. However, there are disagreements in the literature in regards to the interaction of AS with the AR in vitro as revealed by competitive ligand binding assays using cytosolic preparations from prostate, seminal vesicles and skeletal muscle [12], [13] and [24]. By using tissue extracts, it has been shown that some AS have binding affinities for the AR that are higher than that of the natural androgen testosterone, while others such as stanozolol and methanedienone have significantly lower affinities as compared with testosterone [12] and [13]. In an effort to address this conundrum, we compared the properties of select AS by using three different systems: (1) a recombinant AR ligand binding in vitro assay; (2) a cell based AR-dependent transactivation assay; and (3) an in vivo assay based on steroid induced cytosolic AR depletion in skeletal muscle.
To reassess the rat skeletal muscle data, we performed competitive ligand binding assays with cytosol preparations from HEK-293 cells containing over-expressed recombinant rat AR. The data clearly show the presence of saturable high affinity binding sites for 3H-methyltrienolone in cytosol and the Kd value for 3H-methyltrienolone using the recombinant rat AR. These data are in agreement with corresponding values obtained with cytosolic preparations from androgen target tissues of male rats [12] and [25]. These results imply that the AR and not other binding proteins for methyltrienolone are present in the skeletal muscle cytosol. The recombinant AR demonstrated binding parameters for all steroids examined with a relative binding affinity for methyltrienolone that was higher than that for testosterone, whereas stanozolol and methanedionone demonstrated significantly lower affinity for the androgen receptor than testosterone and 17α-methyltestosterone. Hence, our data with the recombinant AR are in agreement with previously published studies on the AR in cytosolic preparations from target tissues. This also suggests that eventual post-translational modifications such as phosphorylation of the receptor in target tissues are most likely not affecting its ligand binding properties, at least for the steroids used in the present study. Alternatively, HEK 293 cells may modify the recombinant receptor in a similar fashion as the target cell post-translationally modifies the native receptor.
Since the main goal of the present study was to compare AR binding properties of AS with low affinity in vitro to their ability to induce biological effects in vivo, we performed cell-based AR-dependent transactivation assays and cytosolic AR depletion assays. Using the transcriptional activation assay, the tested AS stimulated transcription of the MMTV-reporter gene in a concentration-dependent manner, and interestingly, we found that all steroids induced the same level of maximum transactivation. These results are in consonance with a previous report on testosterone and 5α-dihydrotestosterone activation of AR using a MMTV-promoter based reporter plasmid [21]. The data are, however, in contrast to the study by Holterhus et al. [26], in which ligands including testosterone, 5α-dihydrotestosterone, stanozolol and methyltrienolone demonstrated different levels of maximum transactivation in experiments using the MMTV-promoter in CHO cells. One explanation for the difference in results between studies could be that in the current study, we used serum-free assay conditions and CV-1 cells, whereas Holterhus et al. used CHO cells in culture medium with charcoal-stripped serum.
Taken together, in view of the recent data on the crystal structures of the human AR ligand-binding domain in complex with the 3-keto, Δ4-steroids methyltrienolone [27] and 5α-dihydrotestosterone [28], it is reasonable to propose that N-2 in the pyrazole ring of stanozolol is forming a hydrogen bond with arginine 752, thus, mimicking the O-3 and arginine 752 hydrogen bond interaction by methyltrienolone and 5α-dihydrotestosterone in the ligand binding pocket (Fig. 1). Hence, our data show that AS with low affinity to AR in vitro, can in fact in vivo act on the AR to cause biological responses.
Even with these results, we can still not explain the difference in the anabolic-androgenic potency ratio between the endogenous androgen testosterone and the AS. It is conceivable that these different ligands may affect AR conformation and co-activator recruitment, resulting in differential gene activation that may be cell-type specific [29]. Nevertheless, involvement of other AR-like or G-protein coupled receptors or target proteins/enzymes cannot be excluded [30], [31], [32], [33], [34] and [35]. In the past 3 years, a series of studies have reported on non-genomic actions of androgens via the MAP kinase pathways of signal transduction. Skeletal muscle cells have been shown to respond to androgens within minutes via a membrane receptor that is coupled to a pertussis toxin-sensitive G-protein; activation of phospholipase C is followed by increased levels of IP3 which leads to an increase in intracellular Ca2+; Ca2+ then stimulates protein kinase C followed by activation of the Ras/MEK/ERK signaling pathway [32]. It is conceivable that activated ERK may phosphorylate the classical nuclear AR, thus influencing its transcriptional activity. Interestingly, an inhibitor of the classical AR cyproterone acetate does not inhibit the rapid response to either testosterone or the AS nandrolone in this system; this was also observed in osteoblasts in which the ERK signaling pathway was activated by dihydrotestosterone [33], suggesting that the receptor protein at the plasma membrane is different from the classical nuclear AR. Non-genomic effects of androgens involving the Src/Shc/ERK [31], MAP kinase/cyclin-dependent kinase 1 [34], and MAP kinase/CREB [35] signaling pathways have also been demonstrated in cultured cells. Therefore, future investigations require the study of the effects of various AS on these signaling pathways.
We conclude that AS with low affinity to AR in vitro, can in fact in vivo act on AR to cause biological responses involving its classical transcriptional mechanism of action. Our study shows that even though stanozolol and methanedienone possessed low affinity to recombinant AR assessed by ligand binding assay, they are in fact able to efficiently activate the recombinant androgen receptor in intact cells, which is in agreement with efficacy data such as nitrogen retention. Thus, this difference in potency of methanedienone and stanozolol relative to 17α-methyltestosterone and methyltrienolone in the in vitro binding assay versus the transactivation assay may be explained by unique solubility properties of the stanozolol and methanedienone molecules. Whether the action of AS also entails the recently discovered non-genomic signaling pathways are still to be evaluated.