Androgen-binding Protein PURIFICATION FROM RAT EPIDIDYMIS, CHARACTERIZATION, AND IMMUNOCYTOCHEMICAL LOCALIZATION*

Androgen-binding protein (ABP) was purified from caput epididymis of the rat by sequential chroma- tography on DEAE-Sepharose, hydroxylapatite, dihy-drotestosterone-17,f-hemisuccinyl-1,6-diaminohexane- Sepharose, and Sephadex G-150. The final product migrated as a single band corresponding to a peak of protein-bound [3Hldihydrotestosterone on polyacryl- amide gel electrophoresis. A molecular weight of 100,000 was estimated by sedimentation equilibrium. On sodium dodecyl sulfate-polyacrylamide gel electrophoresis, subunits of Mr = 47,000 and 41,000 were ob- served. Amino acid analysis indicated ABP to be rich in leucine while nonpolar aminoacids totaled only 51%. Its carbohydrate content is 25%. Antibodies to purified ABP were raised in a rabbit and evaluated by immunodiffusion, immunoelectrophoresis, binding inhibition, radioimmunoassay, and immunocytochemistry. Immunoperoxidase staining localized ABP in the basal and adluminal regions of seminiferous tubules of rat testis and in secretory granules ofcultured Sertoli cells. In

100,000 was estimated by sedimentation equilibrium. On sodium dodecyl sulfate-polyacrylamide gel electrophoresis, subunits of Mr = 47,000 and 41,000 were observed. Amino acid analysis indicated ABP to be rich in leucine while nonpolar aminoacids totaled only 51%. Its carbohydrate content is 25%. Antibodies to purified ABP were raised in a rabbit and evaluated by immunodiffusion, immunoelectrophoresis, binding inhibition, radioimmunoassay, and immunocytochemistry. Immunoperoxidase staining localized ABP in the basal and adluminal regions of seminiferous tubules of rat testis and in secretory granules of cultured Sertoli cells. In principal cells of caput epididymis, ABP is concentrated in the supranuclear region known to contain morphological specializations for absorption. These immunocytochemical results confirm that ABP synthesized and secreted by Sertoli cells in the testis is transported to the epididymal duct via testicular fluid and is taken up by epithelial cells of the proximal segments.
The discovery of androgen-binding protein in epididymis and testicular fluid (1-3) and its subsequent identification as a specific Sertoli cell secretory protein (4-6) has provided a valuable marker for studies on the hormonal regulation of Sertoli cell function (7)(8)(9)(10)(11). Until recently, however, ABP1 was identified solely by its androgen-binding activity.
We have purified ABP from rat epididymis. 2 The purified * This work is supported by United States Public Health Service Research Grants HD04466, HD13781, and HD11884. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Present address, Research Institute for Joint Diseases, North General Hospital, 1919 Madison Ave., New York, NY 10035. 1 The abbreviations and trivial names used are: ABP, androgenbinding protein; SDS, sodium dodecyl sulfate; dihydrotestosterone, 17 /3-hydroxy-5 a-androstane-3-one.
2Portions of this paper (including some "Methods," Figs. 1-5, Table   I, and additional references) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda,.MMD 20014. protein has been characterized and antiserum has been raised in rabbits. Immunocytochemical studies show specific staining of Sertoli cells, the site of ABP production, and of epithelial cells in the caput epididymlis, where ABP is absorbed from the testicular fluid.

METHODS
Chemical Analysis-Protein (100 ,g) was hydrolyzed in 6 N HCl in evacuated, sealed tubes for 24 h. Hydrolysate was analyzed in a Bio-Cal BC 200 automatic amino acid analyzer equipped with an Autolab integrator. Cystine was also determined as cysteic acid in an independent run following performic acid oxidation. Tryptophan was not analyzed. Carbohydrate composition was determined by gas-liquid chromatography (12).
Preparation of Antiserum-Lyophilized antigen (100 gg) was dissolved in 2 ml of saline (0.9% NaCl solution) and emulsified with an equal volume of Freund's complete adjuvant and 10 mg of mycobacterium butyricum. A New Zealand white rabbit was immunized as described by Vaitukaitis et al. (14). Booster injections of antigen (100 gg) were given 2 and 4 months following the primary immunization, and the rabbit was bled 10 days following each booster.
Evaluation ofAntiserum-ABP antibodies were detected by double immunodiffusion (15) against purified ABP. Specificity was further examined by crossed immunoelectrophoresis according to Laurell (16) and by inhibition of ABP binding of [3H]dihydrotestosterone on polyacrylamide gel electrophoresis (17). RESULTS ABP was purified by column chromatography on DEAE-Sepharose, hydroxylapatite, [3H]dihydrotestosterone hemisuccinate coupled to 1,6-diaminohexane Sepharose, and Sephadex G-150 (Fig. 1). Purified ABP formed a single band (RF = 0.5 relative to bromphenol blue) on polyacrylamide gel electrophoresis and corresponded to a peak of bound radio-Request Document No. 80M-1811, cite author(s), and include a check or money order for $4.00 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press.

5170
This is an Open Access article under the CC BY license. activity (Fig. 2). Equilibrium centrifugation of intact ABP indicated a single component system. Observed data fit well into the Lamm equation (Fig. 3), yielding a molecular weight estimate of 100,000 using a partial specific volume of 0.71 cm 4/ g as determined from the amino acid composition. The homogeneity of ABP was also examined by electrophoresis in a 10%e-SDS-polyacrvlamide gel (Fig. 4). ABP migrated as two major bands corresponding to molecular weights of 41,000 and 47,000 in a ratio of approximately 1:3, respectively.
The chemical composition of ABP is shown in Table I and some of its physicochemical properties are listed in Table II. ABP is exceptionally rich in leucine, even though the content of nonpolar amino acids (513) is not extraordinary. Its hydrophobicitv is average whereas fractional charge is relatively high. The carbohydrate content of ABP is unique in that it is high in glucose. The amount of N-acetvlglucosamine recovered in the amino acid analvzer (12.2 nmol) was in acceptable agreement with the amount obtained by carbohydrate analysis (14.9 nmol).
Immunization of a rabbit with ABP raised precipitating Data from equilibrium centrifugation of purified ABP plotted according to the Lamm equation (19), where c is the concentration of protein and r is the distance from center of rotation.  (18). Electrophoresis was performed in Tris/glvcine buffer, pH 8.9. at 2 to 4°C, 2 mA/tube, and continued until the tracking dve, bromphenol blue, reached the end of the gel. The gel was sliced in half lengthwise and corresponding segments of the two halves were marked with wires. One half was sliced into 2.3-mm crosssectional segments which were eluted and counted in 5 ml of toluenebased scintillator. antibodies, the highest titer being reached following the second booster. Antiserum was judged to be monospecific by immunodiffusion (not shown) and crossed immunoelectrophoresis (Fig. 5) against purified ABP. Single precipitin lines were observed with both methods. No precipitation was observed with preimmunization serum from the same rabbit.
Further evidence that the immune serum contained antibodies against ABP was obtained by quantitation of ['H]dihydrotestosterone binding during polyacrylamide gel electrophoresis ( Fig. 6). Incubation of epididymis cytosol with 5171 j   where is partial specific volume, p is densitv is rotation rate, c is concentration, and r is the distance from center of rotation. l)etermined bv SDS-polvacrvlamide gel electrophoresis. Calculated from the amino acid compositionl.
ABP antiserum prior to electrophoresis decreased ABP binding of [;'H]dihydrotestosterone. This method of detecting antibodies was described previously (17). We have not determined whether the decrease in binding resulted from antibody inactivation of the binding site on ABP or from failure of the antigen-antibody complex to enter the gel. Antibody was evaluated further bv immunocytochemical staining using the peroxidase technique (20). Specific staining for ABP was observed in cultured rat Sertoli cells, seminiferous tubules of rat testis, and in the epithelium of caput epididymis (Fig. 7). "Method specificity" (21) was tested by staining with increasing dilutions of anti-ABP serum. Optimal staining was obtained with dilutions from 1:1,000 to 1:10,000. No staining was detectable at dilutions of 1:100,000 or higher, indicating that staining depended on the primary antiserum and not on subsequent reagents or on endogenous peroxidase. As an indication of antibody specificity, staining was significantly reduced when anti-ABP serum was preabsorbed in the testis and totally eliminated in the epididymis with purified ABP or epididymis cytosol. Specific staining was localized to secretory granules of cultured Sertoli cells (Fig. 7A). In seminiferous tubules, immunoreactive ABP staining was observed in elusters of small granules at the basal and adluminal regions of Sertoli cell cytoplasm (Fig. 7B). Different segments of the same tubule varied greatly in their content of immunoreactive ABP. In epididymis (Fig. 70), ABP was localized largely in the caput, where it was found to coat the microvilli and to be concentrated within the Golgi region of principal cells. These cells are known to contain numerous coated vesicles, secondarv lysosomes, and multivesicular bodies which constitute morphological evidence for the function of absorption (23). DISCUSSION Purif'ication of ABP to homogeneity from rat epididymis cytosol is made possible by affinity chromatography coupled with conventional methods of protein separation. Similar procedures have been used to purify steroid-binding proteins from plasma (24,25). In preparing the affinity column, the 17-/3-hemisuccinate ester of dihydrotestosterone is readily synthesized. We found the mixed anhydride method of coupling the steroid ester to 1,6-diaminohexane Sepharose more efficient and reproducible than the conventional carbodiimide method. Following the affinity chromatography step, albumin is the major contaminant and can be separated from ABP by gel filtration chromatography. ABP  Electrophoresis on SDS-polvacrylamide gels yielded two bands corresponding to molecular weights of 41,000 and 47,000 in a ratio of approximately 1:3. These two dissimilar subunits are identical in size and mass ratio to those reported by Musto et al. (28). ABP isolated from the rat testis has recently been shown to consist of the same subunits which, bv peptide mapping, appear to be identical with the subunits of epididvmal ABP (29). Moreover, it has been shown bv photoaffinitv labeling that ABP isolated from the medium of cultured Sertoli cells is composed of the same subunits (30). ABP purified from rabbit epididymis is reported to have a similar subunit structure (31). These findings suggest that the different subunits may result from posttranslational processing of ABP in Sertoli cells. Differences in the carbohydrate content of ABP subunits might account for their unequal size; howevxer, this remains to be demonstrated. Our finding that ABP is a glvcoprotein is in contrast to a report in which sugars could not be demonstrated by periodic acid-Schiff staining or amino acid analysis of purified ABP (28). Others, however, have recently reported that purified ABP does stain with periodic acid-Schiff stain and binds to coneanavalin A (32). This propertv of rat ABP is shared bv testosterone-binding presumptive Sertoli cell cxtoplasm. Note lack of immunostaining in spermatogonia (sg) and spermatocvtes (se) (X 850). C, Cross-section of the epididvmal duct from the region of the caput, showing immunoreactive ABP (dark stain) in the lumen and all epithelial cells lining the duct. In all epithelial cells, ABP staining is confined to the supranuclear (Golgi) region (arroawheads) of the cxtoplasm. Conmplete arrouws point to nuclei in the basal region of epithelial cells. Note lack of staining in the basal region of epithelial cells and in the surrounding smooth muscle and connective tissue (X 138). globulin (24), a serum protein not present in the rat, but found in numerous other mammalian species (33). Indeed, it appears likely that ABP and testosterone-binding globulin are very similar proteins when they coexist in the same species (34).
ABP is secreted by Sertoli cells into the seminiferous tubular fluid and is carried through the efferent ducts into the caput epididymis where most of its binding activity is lost (1). Immunocvtochemical studies indicate that ABP is largely absorbed by epithelial cells of the caput epididymis, where it has been shown previously that most of its binding activity is destroyed (1). A smaller portion of secreted ABP enters the blood stream (35): however, its route of entry has not yet been established. The striking accumulation of ABP in the basal portions of Sertoli cells suggests that its entry into blood max be bv way of the interstitial fluid of the testis. Immunoperoxidase staining of cultured Sertoli cells indicates that ABP is packaged in secretory granules of variable diameter. Secretion of ABP containing granules occurs in response to folliclestimulating hormone and is associated with striking changes in shape from flat to stellate or elongate morphology (22,36,37). In tLio ABP production is regulated by follicle-stimulating and androgenic hormones (7, 10); however, a variable content of immunoreactive ABP has been found in different segments along the same seminiferous tubule (38). Recent studies on isolated segments of seminiferous tubules have shown that Sertoli cell secretion of ABP varies in association with different stages of the spermatogenic cycle, being highest during the late stages of spermatid maturation (38). These observations suggest that hormonal stimulation of spermatogenesis is modulated by variation in functional activities of Sertoli cells (39). Control mechanisms intrinsic to the seminiferous epithelium may influence Sertoli cell responses to %.am* A l follicle-stimulating hormone and androgens. The biological function of ABP remains to be determined; however, it could have local effects in the germinal epithelium associated with androgen action on spermatogenesis and more distal effects on epithelial cells in the caput epididymis. It has been suggested that ABP taken up by epithelial cells of the caput epididymis might serve to transport androgen from the testicular fluid to the cytoplasmic androgen receptor protein (1,40). Rad Labs. 5 a-dihydrotestosterone (DHT) and succinic anhydride were from Sigma Chemical Co. as were bovine serum albumin, ovalbumin, and chymotrypsinogen A which were used as protein markers Tri-n-butylamine and isobutylchlorocarbonate were purchased from Aldrich Chemical Co. Toluene, ether, dioxane (spectral grade), pyridine, and N, N]-dimethylfornamide (99 Mol' pure) were from Fi sher Synthesis of 3H] DHT-173-hemisuccinate DHT (1450 mg). [3iH] DHT (25~tC 120 C/mole) and 6 miole succinic anhydride were dissolved in 10 ml toluene and refluxed overnioht. The solution was partitioned between 250 ml each of ether and 0.01 N HC1 (pH 2). The water phase was removed, and the ether re-equilibrated with an equal volume of 0.01 HCl. The ether phase containing DHT-hemisuccinate was equilibrated with S00 ml of 0.8 M NaHCOP DHT-hemisuccinate is extracted into the NaHCO3 solution, whereas free DHT remains in the ether phase. The aqueous phase was collected, and the ether phase equilibrated with NaHCO3 in equal volume. The NaHCP3 extracts were pooled and acidified with concentrated HC1 to a pH of~2. This suspension was equilibrated with 250 ml of ether, the ether extract was collected and the ether evaporated.
The residue was checked for purity against a DHT-hemisuccinate standard bv thin layer chronatography on silica-gel F-254 in benzene; ethyl acetate (1:1) containing 1 acetic acid. Product yield was about 76O.
Coupling of [3H] DHT-hemisuccinate to 1,6-diaminohexane-Sepharose (AH-SeDharose 4B) The mixed anhydride procedure (1,2) was adapted (Fig. 1). LPi DTT-hemi sccinate 1430 mo, 16,000 cam/mg was dissolved in dioxane, 10 ml, and tri-n-butylamine 0.24 ml was added. The solution was cooled to 11t, and 0.130 ml isobutylchlorocarbonate was added. The reaction was allowed to proceed for at least 20 minutes and mixed with 15 g AH-Sepharose in a volume of 100 ml of 1:1 dioxane/water plus 2 ml 1N NaOH (rinsultinq pH 9.5). After 15 min, the pH fell and an additional 0.7 ml of 1N NaOH was added to bring the pH back to 8.5. Small amounts of 1N laOH were added (0.2 ml, after a 30 min interval, and 0.05 m] after another 45 min) to maintain an alkaline pP. The reaction was allowed to proceed overnight and the AH-Sepharose was washed with dioxane on a sintered glass filter, until no radioactivity was detected in the eluate (at least 1 liter of dioxane). This was followed by washinq with 800 methanol (1 liter) and finally distilled water (500 ml). The extent of steroid coupling was estimated by hydrolvainq 0.4 ml aliquots of the washed gel with 0.4 ml of N NaOH at 37°C overnight followed bv addition of 0.0 ml of 1N HCl. The gel was allowed to settle, and aliquots of supernatant counted in a liquid scintillation spectrophotometer. Knowing the specific activity of the steroid, coles of steroid/ml AH-Sepharose is easily calculated. Theoretical capacity for AH-Senharose is 7-10 umoles/ml. and Function (Steinberger, A., and Steinberger, E., eds) pp.
Purification of ABP Purification steps were carried out at 0-40C. Ut absorbance from column effluent was monitored at 280 nm using an ISCO UA-5 absorbance monitor equipged with a 5 mm flow cell.  (Fig. 2). In this and subsequent purification steps the presence of ABP in radioactive peak fractions was confirmed by polyacrylamide gel electrophoresis (4).  (Fig. 3). However, elution of ABP varied somewhat with different lots of hydroxylapatite. ABP sometimes adsorbed to the column along with bulk proteins but could be eluted in low concentrations of phosphate buffer (5-10 mM) ahead of the major protein peak. This elution pattern was more often obtained when the sample was dialysed to remove salt prior to the hydroxylapatite step.

FRACTION NUMBER
Protein eluting from the affinity column in TGKD buffer containing DHT was desalted on a Sephadex G-25 column (35 x 2.6 cm) equilibrated with TG buffer. Fall-through protein was concentrated on a small DEAE-Sepharose column (3 x 0.9 cm) equilibrated with TG buffer. Adsorbed protein was step eluted with a small volumne 4 ml) TG buffer containing 1M KCl.
Gel filtration chromatography was performnd on Sephadex G-150. The concentrated protein from DEAE-Sepharose was incubated with 10 nM[ H] DHT for 1 h at O'C and chromatoqraphed on a Sephadex G-150 column (95 x 1.6 cm) equilibrated with TG buffer. Fractions of 4.5 ml were monitored for protein and radioac[ivity (Fig. 5). ABP eluted in a symmetrical peak corresponding to a Stokes radius of 48 0. The void peak contained aggreqated ABP as judqed by SDSpolyacrylamide gel electrophoresis. It was found that the amount of aggregated ABP was increased by lyophilization of the sample prior to gel filtration chromatography. The small peak eluting behind ABP had an electrophoretic mobility identical to albumin on SDS-polyacrylamide gels.
In a purification carried out subsequent to obtaining ABP antiserum, immunoreactive ABP was measured by radioimmunoassay (6). Recovery of irnmunoreactive ABP is shown in Table 1   Affinity chromatography was performed on an AH-Sepharose 4B column containing approximately 4 jmoles/ml of covalently linked [3H] DHT-17B-hemisuccinate. The column was equilibrated in TG buffer containing 1M KC]. Pooled ABP fractions (750 ml) from the hydroxylapatite column were made to contain 1M KC] and applied at a flow rate of 75 ml/h following which the column was washed with TG buffer containing 1M KCl and 10% dimethylformamide (TGKD) (5) until a constant zero baseline level of absorbance was reached with UV monitor setting of 0.05 absorbance unit full scale (~20 column volumes of buffer). ABP was eluted in 2-4 volumes of TGKD buffer containing 100 jM DHT (Fig. 4). During chromatography of the sample, there was no loss of covalently linked [3H] DHT-17B-hemisuccinate suggesting that any esterase activity capable of cleaving the hemisuccinate linkage (5)