Signaling Cross-talk from Gβ4 Subunit to Elk-1 in the Rapid Action of Androgens*

Androgens act on transcription via intracellular androgen receptors (ARs), but they also have rapid AR-independent effects. We have identified the multistep processes involved in the rapid actions of androgens in male osteoblasts, which also possess the classical AR. Incubating cells with 5α-dihydroxytestosterone (100 pm, DHT) rapidly increased (1 min) the phosphorylation of the transcription factor Elk-1, and this was inhibited by pertussis toxin (PTX). DHT activated ERK1/2, a substrate of Elk-1, within 15 s but had no effect on p38 MAPK or JNK/SAPK. The inhibitors PD98059 (MEK1/2); Gö6976, Gö6983, and chelerythrine (protein kinase C); wortmannin and LY294002 (phosphatidylinositol 3-kinase); PP1 (Src); and PTX all blunted the DHT-stimulated phosphorylation of ERK1/2. DHT increased the phosphorylation of c-Raf-1 within 5 s; this was blocked by conventional protein kinase C and phosphatidylinositol 3-kinase inhibitors. The first activated membrane protein was the PTX-sensitive Gβ4 subunit coupled to phospholipase C-β2, which triggered a rapid (5 s) increase in intracellular calcium and diacylglycerol formation. The androgen antagonist cyproterone acetate did not modify the responses to DHT. Lastly an anti-AR antibody directed against the ligand binding domain recognized a protein at the plasma membrane. The cascade of rapid effects triggered by androgens may involve the classical AR at the plasma membrane or an uncharacterized form of AR that is insensitive to nuclear antagonists.

Androgens play a key role in bone turnover, and the increase in the serum concentration of these steroid hormones at the onset of puberty is important for the production of peak bone mass in males (1). Androgens classically act on gene transcription via intracellular androgen receptors (ARs). 1 The hormonebound ARs act as ligand-dependent transcriptional factors, interacting with androgen response elements to stimulate certain genes in androgen-responsive tissues, and regulate gene transactivation (2)(3)(4)(5). The genomic actions of androgens are relatively slow and sensitive to inhibitors of transcription and translation. Protein kinase C (PKC) negatively regulates androgen receptor-mediated gene transcription by interacting with c-Jun and AR (6), while protein kinase A activates androgen-mediated gene transcription in the absence of androgen (7). Lastly activation of the mitogen-activated protein kinase (MAPK) pathway stimulates androgen receptor-regulated gene expression (8 -10), and calcium regulates the production of AR (11).
Investigators have recognized recently that androgens also have rapid AR-independent actions on several cellular processes. Androgens increase intracellular calcium (12)(13)(14)(15)(16)(17), activate MAPK (17)(18)(19), increase nitric oxide production (18), regulate actin reorganization (20), and increase prolactin secretion (21). We have previously shown that testosterone and testosterone covalently bound to bovine serum albumin activate only the phospholipase C (PLC) pathway in male osteoblasts via a pertussis toxin-sensitive G-protein (12). Activation of PLC leads to the increased formation of inositol 1,4,5-trisphosphate, which regulates the release of intracellular calcium from endoplasmic reticulum, and diacylglycerol (DAG), which regulates the activity of PKC. PLC␤ isoforms are activated by the G␣ subunits of G q family proteins or by G␤␥ dimers (22,23). In many cases, G␤␥ dimers are required for the activation of effectors such as MAPK rather than the G␣ subunits (24). The MAPK modules include the extracellular signal-regulated kinases (ERK1/2), the c-Jun NH 2 -terminal kinases (JNKs), the p38 (␣, ␥, and ␦) MAPKs, and big MAPK (ERK5), whose general compositions and primary sequences are very similar (25)(26)(27). However, MAPK modules must be capable of discriminating between distinct stimuli to elicit the appropriate physiological response. While the events leading to MAPK activation take place in the cytoplasm, MAPK is the first member of the cascade to be translocated to the nucleus where it contacts and phosphorylates key nuclear substrates involved in proliferation/differentiation. Steroid signaling affects the proliferation and differentiation of major endocrine tissues, including bone, via the AR, which can be phosphorylated by protein kinases. Steroid signaling also induces rapid effects that are independent of the classical AR. We have examined the question of how all these aspects can be coordinated in a single experimental model using non-malignant cells that possess classical androgen and estrogen receptors.
We have used male osteoblasts, the cells responsible for osteogenesis, from newborn rats and identified the G subunit, PLC isoform, and PKC isoenzyme involved in the rapid phosphorylation of ERK1/2 by androgens. We have also investigated the effect of androgens on upstream proteins and downstream transcription factors involved in this pathway. Our studies were done using 5␣-dihydrotestosterone (DHT) rather than testosterone as osteoblasts possess an aromatase that can convert testosterone into estradiol.
The DHT was dissolved in ethanol; the final concentration of ethanol in the culture medium or buffer never exceeded 0.01%. This ethanol concentration was without effect on any biological parameter tested in this study (data not shown).
Osteoblast Isolation and Culture-Osteoblasts were isolated from parietal bones of 2-day-old male or female Wistar rats (Charles River/ Iffa Credo, L'Arbresle, France) by sequential enzymatic digestion (28). Cells were grown to confluence on rectangular glass coverslips or in Petri dishes in phenol red-free Dulbecco's modified essential medium supplemented with 10% heat-inactivated FCS. Cells were then incubated for 24 h in phenol red-free medium plus 1% heat-inactivated FCS and transferred to serum-free medium 24 h before use.
Cell Labeling and Lipid Chromatography-DAG formation was assessed by incubating cells for 24 h with 0.25 Ci/ml [ 14 C]arachidonic acid (Amersham Biosciences) in phenol red-free medium with 1% heatinactivated FCS. The labeled cells were incubated for 2 h in fresh serum-free medium; ethanol (0.01%) or 100 pM DHT was added for 5-60 s. The reaction was stopped by removing the medium and adding cold methanol. The lipids were extracted according to Bligh and Dyer (29), and neutral lipids were analyzed by thin layer chromatography as described previously (12).
Cyclic AMP Assay-Cells were seeded at 50,000 cells/well in 6-well plates and cultured until confluent in 10% heat-inactivated FCS medium. Cells were then incubated for 24 h in serum-free medium. The medium was removed, and the cells were incubated for 1-30 min in Dulbecco's modified essential medium containing 1% bovine serum albumin, 0.2 mM isobutylmethylxanthine (a phosphodiesterase inhibitor), and 10 pM-1 M DHT, 1 M forskolin, or ethanol. The cells were placed on ice, and the medium was removed. Cellular cAMP was extracted twice with ethanol. The ethanol extracts were dried and stored at Ϫ20°C. Cyclic AMP was measured by radiocompetition as described by Lust et al. (30). Cell cAMP content is expressed as pmol/mg of protein. All measurements were made six times for each concentration and for each culture.
Calcium Measurement and Experimental Protocol-The confluent osteoblasts grown on glass coverslips were washed with Hanks' HEPES, pH 7.4 (137 mM NaCl, 5.6 mM KCl, 0.441 mM KH 2 PO 4 , 0.442 mM Na 2 HPO 4 , 0.885 mM MgSO 4 ⅐7H 2 O, 27.7 mM glucose, 1.25 mM CaCl 2 , and 25 mM HEPES) and loaded with 1 M Fura-2/AM for 20 min in the same buffer at room temperature. The glass coverslip was inserted into a custom-made holder in a quartz cuvette (12) that, in turn, was placed in a temperature-controlled (37°C) Hitachi F-2500 spectrofluorometer (Sciencetec, Les Ulis, France). Drugs and reagents were added directly to the cuvette with continuous stirring. The Fura-2 fluorescence response to the intracellular calcium concentration ([Ca 2ϩ ] i ) was calibrated as described previously (31). Each measurement on Fura-2loaded cells was followed by a parallel experiment under the same conditions on unloaded cells. We determined the subunit of G-proteins and PLC isoenzyme involved in the DHT effects on [Ca 2ϩ ] i . Osteoblasts were incubated for 5 min with 20 g/ml saponin plus excess anti-Gprotein antibody or anti-PLC antibody as described previously (31,32). The antibodies were used at concentrations of 0.1-10 g/ml. Cells were washed four times with Hanks' HEPES to remove saponin and incubated with anti-G-protein or anti-PLC antibody or nonimmune rabbit serum for 1 h at 37°C (31,32). Fura-2/AM was then added for 20 min.
In some experiments, anti-G-protein antibody or anti-PLC antibody was set up in competition with the antigen against which it was produced or with the antigens for the other anti-G-protein or anti-PLC antibodies at room temperature prior to use.
Phosphatidylinositol 4,5-Diphosphate Hydrolysis Assay-Cells were washed with ice-cold phosphate-buffered saline, pH 7.4, and scraped off into ice-cold buffer (20 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, 0.6 mM pepstatin, 0.5 mM benzamidine, 0.1 mM leupeptin, 2 mM phenylmethylsulfonyl fluoride, 0.125 mM aprotinin, and 1 mM dithiothreitol). Cells were sonicated on ice (2 ϫ 20 s) at 40 kHz, and the homogenate was centrifuged for 10 min at 600 ϫ g to remove nuclei. The supernatant was centrifuged for 60 min at 100,000 ϫ g, and the resulting membrane pellet was suspended in extraction buffer. Phospholipid vesicles were prepared according to Hofmann and Majerus (33). In brief, 2. The assays were done essentially as described by Wu et al. (34). Aliquots of the membrane (10 l, 10 g of protein) were added to 40 l of assay buffer (50 mM HEPES, pH 7.0, 100 mM NaCl, 5 mM MgCl 2 , 0.6 mM CaCl 2 , and 2 mM EDTA) plus 10 l of [ 3 H]PIP 2 (10,000 -12,000 cpm) and incubated on ice for 10 min. For the antibody inhibition assay, membranes were incubated for 2 h with 10 l of the antibody at the concentrations indicated prior to adding the reaction mixture. For competition assays with the antigen, 10 l of the antibody was mixed with 2 l of serially diluted antigen before adding the membrane and incubated for 2 h on ice. The reaction was started by adding GTP␥S followed by incubation at 37°C for 15 min. The reaction was stopped by adding 0.5 ml of chloroform/methanol/HCl (40:20:0.5), mixing, and chilling on ice. Soluble inositol phosphates (indicating PIP 2 hydrolysis) were extracted with 150 l of chloroform and 200 l of HCl. Phases were separated by centrifugation, and 200 l of the upper aqueous phase was taken for liquid scintillation counting.
Cell Fractionation and Western Blot Analysis-Cells were scraped off into ice-cold extraction buffer (20 mM Tris-HCl, pH 7.5, 0.5 mM EGTA, 2 mM EDTA, 0.6 mM pepstatin, 0.5 mM benzamidine, 0.1 mM leupeptin, 2 mM phenylmethylsulfonyl fluoride, 0.125 mM aprotinin, and 1 mM dithiothreitol) and sonicated on ice (2 ϫ 20 s) at 40 kHz. An aliquot of the lysate was saved, and the remainder was centrifuged for 10 min at 600 ϫ g to remove nuclei. The supernatant was centrifuged at 100,000 ϫ g for 60 min, and the supernatant was saved. The resulting membrane pellet was suspended in extraction buffer. All the fractions (lysate, cytosol, membrane, and nuclei) were frozen in liquid nitrogen, lyophilized, and stored at Ϫ80°C. Protein was assayed by the method of Bradford (35). Proteins were separated by Tricine/SDS-PAGE for G␥ subunits (36) or SDS-PAGE (resolving gel: 12% for G␤ subunits, 8% for PKC, or 10% for MAPK, AR, c-Raf-1, and Elk-1) in 25 mM Tris base, pH 8.3, 192 mM glycine, 0.1% Tween 20 and transferred to a polyvinylidene difluoride membrane (Millipore, St. Quentin-en-Yvelines, France) in the same buffer with 20% ethanol for 2 h at 100 V. The free sites of the membrane were blocked by immersing it for 1 min in 5 mg/ml polyvinyl alcohol and then in Tris-buffered saline containing 4% skim milk. The membrane was incubated overnight at 4°C with rabbit polyclonal anti-G␤, anti-G␥, anti-PKC, anti-MAPK, anti-c-Raf-1, or anti-Elk1 antibodies. Unbound antibodies were removed by three washes in skim milk/ Tris-buffered saline plus 0.1% Tween 20. The membranes were incubated for 1 h with peroxidase-conjugated goat anti-rabbit IgG (1: 1000 diluted) or peroxidase-conjugated donkey anti-rabbit IgG (1/1000 diluted), washed twice with Tris-buffered saline, 0.5% Tween 20, and washed once with Tris-buffered saline. Immunoreactive proteins were detected by chemiluminescence using the ECL kit. Membranes were exposed to XR films. Signals were digitized with a LAS-1000plus camera (Fujifilm, Raytest France, Courbevoie, France), and the intensities of the signals were quantified with the Aida program (Advanced Image Data Analyzer, Version 3.11, Raytest France).
Statistical Analysis-The data were analyzed by one-way analysis of variance. Treatment pairs were compared by Dunnett's method. The value n represents n different cultures of male or female osteoblasts for a specific experiment.

Direct Effects of DHT on Intracellular Calcium and Diacylglycerol Formation-The basal [Ca 2ϩ
] i in confluent male osteoblasts was 112 Ϯ 5 nM, mean Ϯ S.E., n ϭ 6. DHT (100 pM) triggered a transient increase in [Ca 2ϩ ] i that fell rapidly after 15 s but remained above the basal level (plateau phase, 26 Ϯ 4%, mean Ϯ S.E., n (culture) ϭ 4) (Fig. 1A). The dose dependence of the DHT effect on [Ca 2ϩ ] i was bell-shaped between 1 pM and 10 nM (Fig. 1B). The PLC inhibitor U-73122 inhibited the transient peak when added 60 s before DHT but had no effect on the plateau phase (Fig. 1C). U-73343 (0.5-5 M), an inactive analog of U-73122, had no effect (data not shown). The calcium channel inhibitor verapamil (1 M) inhibited part of the transient peak when added 60 s before DHT; it also abolished the plateau phase (Fig. 1C). Incubating cells in medium containing 100 ng/ml pertussis toxin (PTX) for 18 h inhibited the sharp peak Ca 2ϩ response to DHT but had no effect on the plateau phase ( Fig. 1D). Incubation with the androgen antagonist cyproterone acetate (1 M) for 5 min-4 h did not block the calcium response to DHT (Fig. 1E). Flutamide could not be used because of its high fluorescent intensity, which disturbs the calcium measurement with Fura-2/AM. 17␤-Estradiol and 5␤-dihydroxytestosterone (10 pM-10 nM) had no effect on [Ca 2ϩ ] i in male osteoblasts (data not shown). The pure estrogen antagonist ICI 182,780 (1 M) had no effect on the calcium response to DHT when added for 5 min-4 h (Fig. 1F). Lastly DHT (100 pM) produced a rapid increase (within 5 s) in DAG formation ( Fig. 1G) that was blocked by U-73122 and PTX (data not shown). Fig. 2 shows the calcium response to 100 pM DHT in female osteoblasts ( Fig. 2A), and DHT (10 pM-10 nM) had no effect (Fig. 2B).
G-proteins and PLC Isoenzymes Involved in the Effects of Androgens on Intracellular Calcium-We have shown that osteoblasts possess G␣ s , G␣ i , G␣ q/11 , G␤ 1-4 , and G␥ 1 (31,32). We therefore looked for G␤ and G␥ subunits in osteoblast lysates. Whole-cell lysates of male osteoblasts were also found to possess G␤ 1 , G␤ 2 , G␤ 3 , G␤ 4 , G␤ 5 , G␥ 2 , G␥ 3 , G␥ 5 , and G␥ 7 (Western blots not shown). Incubating cells with saponin for 5 min followed by incubation for 60 min in medium without saponin but containing anti-G-protein antibody, anti-PLC antibody, or nonimmune serum did not affect the basal [Ca 2ϩ ] i (data not shown). Incubation with 0.1-10 g/ml anti-G␣ i1-3 , anti-G␣ s , or G␣ q/11 antibodies had no effect (data shown only for anti-G␣ i1-3 , Fig. 3). The DHT-induced transient [Ca 2ϩ ] i peak was blocked by 2 g/ml anti-G␤ 4 antibody, but the plateau phase was not (Fig. 3). The inhibition was abolished when the antibody was incubated with its antigen for 2 h prior to use (Fig. 3). The blockade of the calcium response to DHT by the anti-G␤ 4 antibody was not enhanced when various anti-G␥ antibodies were added to the incubation mixture (Table I). Incubation with anti-G␤ 1 , anti-G␤ 2 , anti-G␤ 3 , or anti-G␤ 5 antibody or different anti-G␥ subunit antibodies (0.1-10 g/ml) did not modify the calcium response to DHT (Fig. 3, data shown only for anti-G␤ 2 antibody and anti-G␥ 2 antibody). Similarly incubation with both anti-G␤ 1 , anti-G␤ 2 , anti-G␤ 3 , or anti-G␤ 5 antibody and different anti-G␥ subunit antibodies (0.1-10 g/ml) did not modify the calcium response to DHT. Table I shows the results of the various combinations of antibody to different G subunits. Incubating the cells with 1 M cyproterone acetate for 4 h did not modify the calcium response to DHT in the presence of anti-G␤ 4 antibody (Fig. 3). The DHT-induced sharp increase in [Ca 2ϩ ] i was inhibited by 1 g/ml anti-PLC␤2 antibody (Ϫ81 Ϯ 3%, mean Ϯ S.E., n ϭ 6), whereas 0.1-10 g/ml antibody to PLC␤1 or PLC␤3 had no effect (Fig. 4, data shown only for 1 g/ml antibody). The residual increase was due to Ca 2ϩ influx because it was blocked by incubating the cells for 30 s with 2 mM EGTA (Fig. 4). Anti-PLC␥1 and anti-PLC␥2 antibodies (0.1-10 g/ml) had no effect (Fig. 3). Anti-PLC␤4 antibody had no effect because rat osteoblasts lack this PLC␤ isoform (30). In some experiments, anti-PLC␤2 antibody was incubated for 2 h with its antigen or other PLC antigens (antibody:antigen, 1:10 or 1:100) before use. The inhibition due to anti-PLC␤2 antibody was lost only when the anti-PLC␤2 antibody was incubated with its own antigen (Fig. 4). In other experiments, the cells were incubated for 4 h with 1 M cyproterone acetate before adding 100 pM DHT. The antagonist did not modify the calcium response to DHT in the presence of anti-PLC␤2 antibody (Fig. 4).
Effects of Antibodies against PLC Isoforms and G Subunits on Phosphatidylinositol 4,5-Diphosphate Hydrolysis Induced by GTP␥S-The dose-response effect of GTP␥S on PIP 2 hydrolysis showed that GTP␥S had no effect below 1 M (Fig. 5A). Osteoblast membranes were incubated with antibodies prior to simulating with 100 M GTP␥S, the maximal active concentration. Anti-PLC␤1, anti-PLC␤2, and anti-PLC␤3 antibodies inhibited the PIP 2 hydrolysis induced by 100 M GTP␥S, whereas anti-PLC␥1 and anti-PLC␥2 antibodies did not (Fig. 5B). The

TABLE I The effects of DHT on intracellular calcium of male osteoblasts incubated with various combinations of antibodies to different G subunits
Osteoblasts were incubated with the various antibodies alone or in combination and then loaded with 1 M Fura-2/AM. Intracellular Ca 2ϩ concentrations were determined 10 s after adding 100 pM DHT. The increase induced by 100 pM DHT in the presence of nonimmune serum was ⌬͓Ca 2ϩ ͔ i ϭ 295 Ϯ 9 nM. Values are means Ϯ S.E., n (culture) ϭ 4. membranes were incubated with both the anti-PLC antibody and its antigen was the same as that obtained with the preimmune serum (Fig. 5B). Anti-G␣ q/11 , anti-G␤ 1 , anti-G␤ 2 , anti-G␤ 3 , anti-G␤ 4 , and anti-G␤ 1 plus anti-G␥ 1 or -G␥ 2 antibodies inhibited the PIP 2 hydrolysis induced by 100 M GTP␥S, whereas anti-G␣ i , anti-G␣ s , anti-G␥ 1 , and anti-G␥ 2 did not (Fig. 5C). The [ 3 H]inositol 1,4,5-trisphosphate concentration produced when membranes were incubated with both the anti-G␣ q/11 , anti-G␤ 1 , anti-G␤ 2 , anti-G␤ 3 , or anti-G␤ 4 antibody and its antigen was the same as that obtained with the preimmune serum (Fig. 5C).
Conventional Protein Kinases C and Androgens-Since cP-KCs are activated by both Ca 2ϩ and DAG, we studied the effects of DHT on the conventional PKCs, particularly PKC␣ and -␤I. The distribution of cPKC in the cells stimulated by DHT was checked by Western blotting of the membrane and cytosol fractions. PKC␣ and -␤I were translocated from the cytosol to the membrane within 5 s of adding 100 pM DHT (Fig.  6, A and B). The cPKCs at the membrane disappeared after incubating the cells in DHT for 20 s; both cPKCs were then located in the cytosol (Fig. 6, A and B). Incubating cells in medium containing phorbol 12-myristate 13-acetate (100 nM) for 15 min caused the translocation of both cPKC isoforms to the membrane (Fig. 6, A and B).
The effect of DHT on [Ca 2ϩ ] i was checked in the presence of an inhibitor of both PKC␣ and -␤I, Gö6976 (36), to detect any feedback of PKC on PLC activation. The cells were incubated for 20 min with 4.6 nM (the concentration that inhibits only PKC␣) or 12.4 nM (the concentration that inhibits both PKC␣ and -␤I) before adding 100 pM DHT (37). The DHT caused a 2-fold greater increase in [Ca 2ϩ ] i in the presence of 4.6 nM inhibitor and a 3-fold greater increase in the presence of 12.4 nM inhibitor than in its absence (Fig. 7A). Other cells were incubated for 4 h with 1 M cyproterone acetate before adding 100 pM DHT. The antagonist did not modify the blocking of PKC by Gö6976 (data not shown). The calcium concentration of the plateau phase corresponding, in part, to calcium influx was not modified by both concentrations of PKC inhibitor (Fig. 7B).
Cyclic AMP and Androgens-The basal cAMP content of male osteoblasts was 30.1 Ϯ 1.7 pmol/mg of proteins, n (culture) ϭ 4, and this was not affected by 10 pM-1 M DHT whatever the incubation time (1-30 min). However, incubation with 1 M forskolin for 5 min increased the cell cAMP content to 121 Ϯ 12 pmol/mg of proteins (n (culture) ϭ 4 and p Ͻ 0.001).

Activation of MAPK Upstream Proteins and Transcription
Androgen Receptors-We used two different anti-androgen receptor antibodies, one directed against the carboxyl terminus (C-19) and the other directed against the amino terminus (N-20). Membranes, cytosols, and nuclei were immunoblotted with both antibodies (0.5 g/ml). Both antibodies detected most of the AR in the cytoplasm with lesser amounts detected in the nucleus (Fig. 11, A and B). ARs were detected at the plasma membrane only with the anti-AR antibody directed against an epitope in the carboxyl terminus of the AR (Fig. 11A). DISCUSSION We believe that this is the first demonstration that the rapid (15 s) activation of the ERK1/2 pathway by a low dose (100 pM) of 5␣-dihydrotestosterone requires the involvement of a pertussis toxin-sensitive G␤ 4 whose intracellular transducer is phospholipase C-␤2 in male osteoblasts. This involves the rapid activation of a cascade of intracellular proteins, such as PI3K, PKC, c-Src, c-Raf-1, and MEK1/2, and leads to the activation of the transcription factor Elk-1. These effects are independent of the activation of the classical AR since the nuclear antagonist cyproterone acetate did not block any of the parameters tested.
A variety of extracellular stimuli cause the sequential phosphorylation of intracellular proteins leading to the activation of MAPKs, which are divided into three major classes: MAPK/ ERK, JNK/SAPK, and p38 MAPK (25,26,38,39). DHT (100 pM) triggered only the phosphorylation of ERK1/2 in male osteoblasts. This increase started at 15 s with a maximum at 5 min and a return to base line by 15 min. This finding is in agreement with previous reports on the effects of androgens in skeletal muscle cells (17), prostate cancer cell lines (19), and osteocytic MLO-Y4 cells (18). But the androgen-induced phosphorylation of ERK1/2 started later (1-5 min) in these cell types and occurred at higher androgen concentrations (10 -100 nM).
While the tyrosine kinase growth factor receptors transmit signals to ERKs in a well defined multistep process, the stimulation of ERK activity by G-protein-coupled receptors may be mediated by different classes of G-protein, including G s , G i , G␣ q/11 , ␤␥ complexes, or G i/o (24,39,40). As expected, the anti-G␣ q/11 antibody did not block the Ca 2ϩ response to DHT since pertussis toxin blocked the DHT-induced increase in calcium mobilization from the endoplasmic reticulum. Similarly this response to DHT was not blocked by incubating cells with anti-G␣ i1-3 and anti-G␣ s antibodies, indicating that the rapid effects of DHT on male osteoblasts are not mediated via the subunits of the G-protein linked to the adenylyl cyclase pathway (41). This is in agreement with the absence of a direct effect of DHT on cAMP formation. It also confirms our previous finding that testosterone and testosterone covalently linked to bovine serum albumin do not increase cAMP in male osteoblasts (12). Since the ␤␥ dimers also play an important role in signal transduction (24,39,40) and are involved in the pertussis toxin-sensitive pathway, we identified the G␤␥ subunits involved in the Ca 2ϩ mobilization by DHT. Only anti-G␤ 4 antibody blocks the DHT-induced increase in [Ca 2ϩ ] i ; combining anti-G␤ 4 antibody with various other anti-G␥ antibodies did not enhance the blockade of the response to DHT. Other anti-G␤ antibodies alone or combined with various other G␥ antibodies had no effect, although they are active on the hydrolysis of PIP 2 . This corroborates our previous findings on the rapid action of testosterone and testosterone covalently bound to bovine serum albumin on [Ca 2ϩ ] i in male osteoblasts involving a pertussis toxin-sensitive G-protein (12). PLC␤ isoforms are important transducers of heterotrimeric G-protein signaling upon agonist activation. The four isoforms, ␤1, ␤2, ␤3, and ␤4, of the PLC␤ family activate the hydrolysis of phosphatidylinositol 4,5-diphosphate (22,23,42). We have previously shown that rat osteoblasts possess PLC␤1, PLC␤2, and PLC␤3 but not PLC␤4 (31). Only the anti-PLC␤2 antibody blocked the DHTinduced increase in [Ca 2ϩ ] i that was due to Ca 2ϩ mobilization from the endoplasmic reticulum. PLC␥1 and PLC␥2 take no part in the DHT effects because the PLC␥ isoforms are effectors of receptors that are protein-tyrosine kinases or linked to cellular tyrosine kinases (42,43) and are generally activated by G i -protein (44). This is the first demonstration that G␤ 4 whose intracellular transducer is PLC␤2 is the first membrane protein target involved in the rapid effects of androgens. The effect of DHT is specific since 1,25-dihydroxyvitamin D 3 activates a PTX-insensitive G ␣/q11 -protein coupled to PLC␤1 (31, 32) and progesterone activates a PTX-insensitive G ␣/q11 -protein coupled to both PLC␤1 and PLC␤3 in male osteoblasts (45). Neither the androgen antagonist cyproterone acetate nor the estrogen antagonist ICI 182,780 blocked the effects of DHT. Lastly the rapid effects of DHT were gender-specific; 17␤-estradiol had no rapid effect in male osteoblasts, although male osteoblasts bear classical estrogen receptors, and DHT, even at high concentrations, did not act on female osteoblasts. The action of 5␣-dihydrotestosterone was also stereospecific since 5␤-dihydroxytestosterone had no effect. The effect of DHT was identical in amplitude to that of testosterone and mimicked the effect of testosterone covalently bound to bovine serum albumin, which did not enter the cells, in male osteoblasts (12).
We then identified the multiple steps from the activation of G␤ 4 to the phosphorylation of ERK1/2. It is well established that the activation of conventional PKCs, which include cPKC␣, cPKC␤I, and cPKC␤II, depends on the concentration of intracellular calcium and diacylglycerol (46), which were increased by DHT in male osteoblasts. DHT caused the translo-cation of PKC␣ and -␤I to the membrane within 5 s, although it took 15 min for phorbol 12-myristate 13-acetate to do so. Gö6976, an inhibitor of the ATP-binding site in the catalytic domain of cPKC, can discriminate between cPKC and novel PKC (37) and block the activity of PKC␣ or PKC␤I, depending on its concentration. The DHT-induced calcium mobilization was 2-or 3-fold higher in osteoblasts incubated with 4.6 or 12.4 nM Gö6976 than that in controls. Since calcium influx was not altered by DHT or by DHT plus Gö6976, this suggests that both the cPKCs activated by DHT have a negative feedback action on PLC␤2 activity. The early (15 s) ERK1/2 phosphorylation by DHT was inhibited in osteoblasts incubated for 20 min with Gö6976. Gö6983, an inhibitor of cPK␤II, PKC␥, PKC␦, and PKC, and chelerythrine, an inhibitor of all types of PKCs (38,46), blocked only the late (5 min) ERK phosphorylation triggered by DHT. This suggests that temporal differences in the activation of different PKCs contribute to ERK1/2 phosphorylation by DHT as has been described in endothelial cells under mechanical strain (47). Incubating male osteoblasts with wortmannin or LY294002, two inhibitors of PI3K activity, decreased only the late (5 min) DHT-stimulated phosphorylation of ERK1/2. DHT may therefore activate the ERK1/2 route via two pathways involving cPKC and PI3K, one responsible for the early phosphorylation of ERK1/2 and the other responsible for its sustained phosphorylation. Lastly the induction of ERK phosphorylation by DHT was blocked by the specific inhibitor PD98059 showing that the immediate upstream activator of ERK1/2 is MEK1/2.
Signaling from certain G-protein-coupled receptors via G␤␥ subunits to MAPK requires the activation of Src or Src-like kinases (48). Moreover ARs, estrogen receptors, and progesterone receptors interact with and activate the intracellular tyrosine kinase c-Src (17,49,50). Incubating the male osteoblasts with PP1, a potent inhibitor of Src tyrosine kinases, diminished only the late (5 min) DHT-stimulated phosphorylation of ERK1/2. This suggests that AR interacts with the Src homology 3 domain of c-Src in response to 100 pM DHT as was found in the AR-positive LNCaP prostate cancer line (19) with 10 nM R1881 (a synthetic androgen), in the AR-positive osteocytic cell line MLO-Y4 with 10 nM DHT (18), and in prostate cancer cells (50).
The Raf-1 kinase is a key protein in the signaling network that controls cell proliferation, neoplastic transformation, differentiation, and apoptosis (51)(52)(53). Many of its effects are transmitted via the MAPK/ERK pathway. In this three-tiered kinase cascade, Raf-1 phosphorylates and activates MEK, which then phosphorylates ERK (54). Raf-1 responds to a wide range of extracellular stimuli, and its regulation is complicated and still incompletely understood (55). Raf may be activated via two distinct pathways, one involving the PLC␤/PKC pathway and the other involving the PI3K/c-Src/Shc/Ras route (51,52). DHT increases the phosphorylation c-Raf-1 within 5 s, and this is blocked by inhibitors of PKC and PI3K. The pathways seem to act in synergy with the action of cPKC being prolonged by that of PI3K.
One of the best characterized ERK substrates is Elk-1, a member of a large family of transcription factors containing an Ets domain. Elk-1 contains an amino-terminal DNA binding domain, a region that interacts with serum response factor, a carboxyl-terminal transactivation domain that contains phosphorylation sites regulating transcription, and a region that binds ERK2, which then phosphorylates and activates Elk-1 (56). Moreover the phosphorylation of ERK1/2 potentiates Elk-1-mediated ternary complex formation and transactivation (57). DHT increased the phosphorylation of Elk-1 within 1 min in male osteoblasts; this phosphorylation was blocked by pertussis toxin. This confirms that the first signal responsible for the cascade of phosphorylations of intracellular proteins is the activation of a G-protein.
Lastly the cellular distribution of androgen receptors was checked with two AR antibodies, one directed against the carboxyl terminus (C-19) and the other directed against the amino terminus (N-20). Only the anti-AR antibody C-19 directed against the ligand binding domain recognized a protein located at the membrane. This is in agreement with the results of Kousteny et al. (18) who demonstrated that the rapid (10 min) antiapoptotic activity of AR lies within the ligand binding domain of AR and requires the protein to be outside the nucleus. But the androgen receptor antagonist flutamide blocks androgen-induced apoptosis under their experimental conditions. This finding does not fit with all published data, showing that nuclear antagonists have no influence on the rapid effects of androgens (58). But the sex hormone-binding globulin allows certain steroids to act without entering the cell (59, 60) by acting via specific receptors at the plasma membrane (61) in addition to regulating the free concentration of a number of steroid hormones. This receptor is thought to be a G-proteincoupled receptor or functionally linked to the G␣ s subunit of the G-protein complex modulating adenylyl cyclase and cAMP production (61) in the presence of sex hormone-binding globulin. This sex hormone-binding globulin receptor cannot be involved in the rapid action of DHT in our system since male osteoblasts are incubated in serum-free medium and DHT had no direct effect on cAMP.
In conclusion, physiological (100 pM) concentrations of DHT can rapidly activate the ERK1/2 pathway from G␤ 4 via two routes, one involving PLC␤2/cPKC/c-Raf-1/MEK1/2 and the other involving PI3K/Src/c-Raf-1/MEK1/2, and the two act in synergy (Fig. 12). This leads to the rapid activation of the transcription factor Elk-1 coupled to early genes involved in proliferation and differentiation. However, it is possible that the non-classical action of androgens influences the transcriptional activity of AR since MAPK can directly phosphorylate AR and AR co-activators (8 -10), although the action of FIG. 12. Cross-talk between classical and non-classical actions of androgens in male osteoblasts. Androgens (DHT) stimulate second messenger cascades by activating G␤ 4 coupled to PLC␤2. DHT stimulates the ERK1/2 pathway by activating c-Raf-1 by two routes, one involving cPKC and the other involving PI3K leading to activation of c-Src. The activation of c-Raf-1 leads to the activation of MEK1/2, which phosphorylates ERK1/2, which in turn activates the transcription factor Ekl-1. All these effects are very rapid and vanish after 5 min. This may occur for the physiological concentration (100 pM) via the classical AR located at the membrane. The activation of ERK1/2 and PKC may directly influence the transcriptional activity of the nuclear AR (gray arrows). HBD, hormone-binding domain.
DHT is too rapid (Ͻ10 s) (Fig. 12). We need to know more about the physiological role of AR at the membrane, which is linked to a G-protein, before we can evaluate the processes by which cells respond to androgens under normal and pathological conditions.