5α-reductase type 1 is localized to the outer nuclear membrane

ELSEVIER Molecular and Cellular Endocrinology 110 (1995) 137-147

Sa-Reductase type 1 is localized to the outer nuclear membrane

J.G.A. Savory&b, D. May*, T. Reich kb, E.C. La Casseab, J. Lakinsb, M. Tenniswoodb, Y. Raymondc, R.J.G. HachW, M. Sikorskad, Y.A. Lefebvreabl*

‘Department ofMedicine, University of Ottawa, Loeb Institute for Medical Research, Ottawa Civic Hospital, Ottawa, KI Y 4E9, Canada bDepartment of Biochemistry, University of Ottawa, Ottawa, KIH 8M5, Canada

%stitut du Cancer de MontrCal, Hopital Notre-Dame, Mont&l, H2L 4iU1, Canada dMolecular Biology Group, Institute for Biological Sciences, National Research Council of Canada, Ottawa, KlA OR6 Canada

Received 13 January 1995; accepted 2 March 1995

Abstract

Tbe subcellular distribution of tbe two isozymes of Sa-reductase has been controversial. To resolve this issue which could provide clues about the respective functions of the two isozymes, two antisera were generated, one which was specific for the ‘Qpe 1 5a- reductase and one which recognized both isozymes. In COS cells transfected separately with the ‘ljpe 1 or Qpe 2 cDNA, both isozymes were detected on Western blots at an Mr of 26 000. Subfractionation of the COS cells resulted in tbe partitioning of both isozymes between the crude nuclear and cytosolic fractions, while cytoimmunofluorescence localized both reductases to the nuclear periphery. In rat liver homogenate, the Sa-reduce was also detected at Mr 26 000. The Sa-reductase immunoreactivity was increased after castration of tbe animals with no further effect when castrated animals were treated with androgens. Although the rat liver expresses only the Qpe 1 Sa-reductase. the Sa-reductase was distributed about equally between crude nuclear and cytosolic subfractions; this distribution could be shifted to the cytosolic fractions with harsher homogenization procedures. Further extensive subfractionation and extraction studies identified the rat liver ‘Qpe 1 Sa-reductase as an integral membrane protein present in tbe outer nuclear membrane of the nuclear envelope and in rough endoplasmic reticulum. Thus, tbe subfractionation and cytoimmunofluorescence studies are consistent with the localization of the ‘l)pe 1 Sa-reductase to the outer nuclear membrane of the nuclear envelope which is continuous with and indistinguishable from the endoplasmic reticulum. This study is the first to localize rat liver ‘ljpe 1 Sa-reductase to the nuclear envelope to which the prostatic Sa-reductase activity previously had been localized. We conclude that, contrary to previous tissue distribution studies, but consistent with investigations in transfected cells, both isozymes are similarly localized to the nuclear periphery.

Keywords: Sa-Reductase Qpe 1; Outer nuclear membrane; Isozymes

1. Introduction

Steroid Sa-reductase (3-oxo-5a-steroid A4-reductase, EC 1.3.1.22) catalyzes the conversion of testosterone, the major circulating androgen, to dihydrotestosterone (DHT). The action of DHT in mediating the androgenic effect of testosterone in many tissues, such as the prostate and epididymis, is well established (Anderson and Liao, 1968; Bruchovsky and Wilson, 1968). On the other hand, while it has been known for some time that the liver has

* Corresponding author, Division of Endocrinology and Metabolism, Ottawa Civic Hospital, 1053 Carling Ave., Ottawa, Canada KIY 4E9. Tel.: +l 613 798 5555, Ext. 6249; Fax: +1613 7615358.

high levels of Sa-reductase activity (McGuire and Ton&ins, 1960; Moore and Wilson, 1972), the role that the enzyme plays in hepatic function remains unclear. Our assumption haa heen that better characterization of the liver enzyme’s properties will lead to an elucidation of its role.

Early investigations of Sa-reductase activity had provided several clues suggesting that more than one form of Sa-reductase exists. These included the demonstration of different affinities and rates of conversion of the enzyme for various substrates among tissues and between epithe- lial and stromal cells in the prostate (Bruchovsky et al., 1988; Tunn et al., 1988) and the detection in cultured fibroblasts of separate Sa-reductase activities possessing

0303-7207/95/$09.50 Q 1995 Elsevier Science Ireland Ltd. All rights reserved SSDI 0303-7207(95)03526-D

138 J.G.A. Suvo~ ef al. I Molecuiurand Cellulur Endocrinology 110 (199.5) 137-147

acidic and basic pH optima (Moore et al., 1975), with the primary determinant of genetic disease being subsequently identified as a deficiency in the acidic enzymatic

component (Moore and Wilson, 1976). Early studies also showed that the subcellular distribution varied from tissue

to tissue (McGuire and Tomkins, 1960; Moore and Wil- son, 1972; Golf and Graef, 1978; Scheer and Robaire, 1983; Houston et al., 1985; Enderle-Schmitt et al., 1986; Sargent and Habib, 199 1).

The uncertainty about the existence of more than one

isozyme for the Sa-reductase has been resolved by the

cloning of two human (Andersson and Russell, 1990; Andersson et al., 1991; Jenkins et al., 1992) and two rat

Sa-reductase cDNAs (Andersson et al., 1989; Norming- ton and Russell, 1992). The two isozymes are highly ho-

mologous, share similar substrate preferences and have similar hydropathy profiles and gene structure (Andersson

et al., 1991; Jenkins et al., 1992; Thigpen and Russell, 1992). However, consistent with the earlier findings re-

ported above, differences between the two isozymes were identified: Type 1 isozymes have lower affinities for steroid substrates (Andersson and Russell, 1990; An-

dersson et al., 1991; Thigpen et al., 1993) than the Type 2 isozymes (Thigpen et al., 1993; Faller et al., 1993) and

the Type 1 enzymes have neutral to basic pH optima (Andersson and Russell, 1990; Normington and Russell, 1992) while the Type 2 isozymes share acidic pH optima

(Andersson et al., 1991; Normington and Russell, 1992). Mutations in the Type 2 gene cause the inborn error of Sa-reductase deficiency (Andersson et al., 1991). Fur- thermore, in most male reproductive tissues, mRNAs encoding the Type 2 isozyme are more abundant than Type 1 mRNAs, whereas the Type 1 mRNAs predominate in

peripheral tissues. In fact, in the liver, Type 1 mRNA appears to be exclusively expressed (Berman and Russell,

1993). While the characterization of two isozymes explained

many of the different properties of the enzyme which had been documented earlier, the different subcellular localization reported in various tissues remains unexplained. Earlier subfractionation studies had indicated an endoplasmic reticulum association of the Sa-reductase in the rat liver (Moore and Wilson, 1972). On the other hand, a nuclear association of the prostate Sa-reductase (Moore and Wilson, 1972; Hudson, 1981) and more precisely, a nuclear membrane association, was suggested for the rat prostatic Sa-reductase (Moore and Wilson, 1972) . How- ever, recently, CHO cell lines expressing each of the two human isozymes were established (Thigpen et al., 1993). Surprisingly, the subcellular distribution of the two 5a- reductase isozymes in these cells was identical and both Sa-reductases were concluded to be associated with the endoplasmic reticulum. Thus these later studies in which it was possible to distinguish the two isozymes, did not explain earlier reported differences in subcellular localization of the Type 1 and Type 2 Sa-reductase in tissues.

To gain insight into the role of the rat liver 5a- reductase and to clarify the subcellular localization, we used the deduced amino acid sequence of the rat Type 1

cDNA to design antigenic peptides for the production of rabbit anti-rat Sa-reductase antisera. We report the local-

ization of the Sa-reductase in COS7 cells transiently expressing each of the two Sa-reductases and in rat liver. Our results are consistent with the localization of both reductase isozymes to the endoplasmic reticulum and

outer nuclear membrane. We conclude that this localization at the boundary between the nucleus and cytoplasm explains the hitherto seemingly differential localization of

Sa-reductase in tissues.

2. Materials and methods

2. I. Preparation of synthetic peptides and Sa-reductase antisera production

The open reading frame of the rat Type 1 Sa-reductase encodes a protein of 255 amino acids (Andersson et al., 1989). The deduced amino acid sequence was searched for likely antigenic sites using the University of Wiscon-

sin Genetic Computer Group software package and the Jameson and Wolf algorithm (Jameson and Wolf, 1988). Three peptides representing amino acids 30-48, 165-183, and 236-249 of Type 1 Sa-reductase were synthesized by

the simultaneous multiple peptide synthesis method of Houghten (Houghten, 1985; Houghten et al., 1986). The protected peptide resins were synthesized using a p- methyl-benzyhydrylamine resin (100-200 mesh, 0.4- 0.8 mequiv/g) and N-a-tertiarybutoxycarbonyl amino acids, and the peptides were cleaved off the resin by the conventional hydrogen fluoridelanisole procedure. The

purity of the crude peptides was assessed by chromatography on an HPLC reverse phase PepRPC HR5 column (Pharmacia, Baie d’Urfe, Quebec). The chromatograms were developed with a gradient of 0.1% CFsCOOW

CHsCN. The peptides represented approximately 85% of OD2i4 absorbing material.

New Zealand White rabbits were immunized with a mixture of the three unconjugated peptides to raise antise-

rum 5a-299 and with a mixture of peptides corresponding to amino acids 165-183 and 236-249 conjugated to myoglobin to raise antiserum 5a-310 according to the fol-

lowing schedule (Kwast-Welfeld et al., 1991; Kwast- Welfeld et al., 1993): the first intra-muscular injection of 0.5 mg peptide emulsified with complete Freund’s adju-

vant (1: 1, v/v) was followed by two consecutive subcuta- neous injections of peptides emulsified with incomplete adjuvant at l-week intervals. The rabbits were test bled at 7 and 14 days after the last injection and the titer of their sera was determined by ELISA.

2.2. Monoclonal antibodies The anti-lamin antibody is a mouse monoclonal IgGl

antibody raised against a highly purified preparation of

J.G.A. Savory et al I Molecular and Cellular Endocrinology 110 (1995) 137-147 139

rat liver lamins; it reacts specifically with lamin Bl. The anti-cytochrome P450 antibody is a mouse monoclonal IgG2A antibody obtained from a mouse immunized against purified rat liver nuclear envelopes (Y.R., unpub- lished results). On immunoblots of endoplasmic reticulum, the antibody cross-reacts with the major 50 kDa polypeptide which has been identified previously as cytochrome P450 (Gerace et al., 1982).

2.3. Animals Male Sprague-Dawley rats weighing 200-250 g were

obtained from Charles River Canada Inc. (Montreal, Quebec, Canada) and maintained on a diet of Purina Lab Chow (Ralston-Purina, St. Louis, MO) and tap water ad libitum. Animals were housed and treated according to the guidelines of the Medical Research Council of Can- ada. The rats were sacrificed by decapitation and the livers quickly removed and placed in ice-cold homogenization buffer (0.32 M sucrose, 3 mM MgCl* and 1 mM dithiothreitol in 10 mM Tris-HCl, pH 7.4). For investigations of the effects of hormonal manipulation, castrated male SpragueDawley rats (200-250 g) were purchased from Charles River Canada Inc. (day 0). Castrated animals received sesame oil intra-peritonically each day on days 7-14 after castration. Treated animals received 1 mg testosterone cypionate intra-peritoneally per day on days 7-14 after castration. On day 14 the rats were sacrificed, the livers removed, and placed on ice. For each fractionation, livers from four animals were obtained and tissue pooled.

2.4. Preparation of subcellularfractions Crude nuclear and cytosolic fractions. COS cells were collected, washed and suspended in hypotonic buffer (10 mM HEPES, 1 mM EDTA, pH 7.4) (Dignam et al., 1983). After homogenization by ten strokes in a Dounce A homogenizer, nuclei were pelleted by centrifugation (Jouan MR14- 11 centrifuge, Jouan Inc., Winchester, VA) and the supematant was recovered for the cytosolic fraction. Nuclear extracts for use on Western immunoblots were obtained after dialysis of the supernatant obtained by extraction of the nuclear pellet in high salt buffer (20 mM HEPES, pH 7.9, 25% glycerol, 1.5 mM MgCl*, 1.2mM KCl, 0.2 mM EDTA, 0.2 mM PMSF, 0.5 mM D’IT) for 30 min.

Rat liver tissue on ice was minced with scissors in 2 ml/g homogenization buffer and homogenized with a Polytron (three 10 s bursts at setting 4). The homogenate was forced through a wire screen (900 mesh), and crude nuclear pellets and cytosol supematant fractions were obtained after centrifugation at 3000 X g for 10 min.

Rat liver microsomes. Microsomes were prepared from rat liver according to the method of La Casse et al. (1990) based on the method of Omrani et al. (1983). Further subfractionation of microsomes was achieved by the method

described by LaCasse et al. (1990) from modifications of Courtin et al. (1985).

Pur$ed plasma membranes, mitochondria and nuclei from rat liver. Liver plasma membranes, mitochondria, and nuclei were prepared using the method of Fleischer and Kervina (1974). Briefly, tissue homogenate was centrifuged to generate a pellet consisting of the nuclei, plasma membrane and mitochondria and a supernatant consisting of microsomal membranes, peroxisomes, lysosomes and mitochondria. The pellet was subjected to further purification on a two-layered step gradient of buffered sucrose consisting of the pellet made up to 1.6 M sucrose, overlayered with 0.25 M sucrose. After centrifugation, the mitochondria and plasma membrane were recovered as a band at the interface and were then separated on a second two-step layered gradient of the mitochondria and plasma membrane resuspended in buffered sucrose to a final concentration of 1.45 M sucrose, overlayered with 0.25 M sucrose. Mitochondria were recovered in the pellet and further purified by resuspension in 0.25 M buffered sucrose followed by centrifugation. The plasma membranes were recovered at the 0.25 Ml.45 M sucrose interface and were further purified on a third step gradient consisting of the plasma membrane in buffered sucrose at a final concentration of 1.45 M, overlayered with 0.25 M sucrose. Nuclei were purified from the pellet of the first sucrose gradient by resuspension in 2.2 M buffered sucrose followed by a 2.4-fold dilution with buffer and centrifugation to obtain the purified nuclear pellet.

Nuclear envelopes. All procedures were carried out as rapidly as possible at 04°C with ice-cold reagents. The nuclear envelopes were prepared from purified nuclei by a modification of the procedure of Kay et al. (1972), as described previously (Howell and Lefebvre, 1989).

2.5. Immmoblotting Proteins of each subfraction were separated using 20%

sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). The proteins were then transblotted onto PVDF membranes (Immobilon-P, Millipore (Canada) Ltd., Mississauga, Ontario) and probed with antisera 5a-299 or 5a-310 at a 1:lOOO (v/v) dilution. Im- munodetection by enhanced chemiluminescence was carried out according to the instructions in the ECL- Amersham kit using donkey anti-rabbit immunoglobulin (Amersham, Arlington Heights, IL) as the secondary antibody at a dilution of 1:50 000. The blocking solution was 10% skim milk in 0.1% Tween 20 Tris-buffered 0.5 M NaCl (pH 7.5).

2.6. Quantification of crossreactivity on ECL western blots

Crossreactivity to 5a-299 and 5~2-310 was quantified using densitometry according to ECL Western blotting

140 J.G.A. Savory et al. I Molecular and Cellular Endocrinology 110 (1995) 137-147

protocols (Amersham, Arlington Heights, IL). After detection of cross-reactivity on the Western blot by the

techniques described above, the blot was exposed to film. The film was scanned with an LKB Bromma Ultrascan X

L Enhanced Laser Densitometer and the peak area of the cross-reactive band determined as OD units x mm. In preliminary experiments, a standard curve was con-

structed from an immunoblot of several aliquots of rat liver cytosol in increasing volumes from 0.1 to 2OOpg. Several exposure times were compared to determine

lower and upper OD values which resulted in peak areas for the band of interest which correlated linearly with

protein concentrations. In experiments for comparison of total activities among subfractions, peak areas for two

volumes of each subfraction were determined, to ensure that the peak areas corresponded appropriately to the dif-

ferent amounts of protein applied to the blot. The total cross-reactivity in a subfraction was calculated from values obtained for the total amount of protein in that sub-

fraction

2.7. Cell culture and transfections COS7 cells (American Type Culture Collection,

Rockville, MD; CRL 165 1) were grown as monolayers in Dulbecco’s modified Eagle’s medium (Gibco-Bethesda

Research Laboratories, Grand Island, NY) supplemented with 10% fetal bovine serum (G&co-Bethesda Research Laboratories) and maintained at 37°C in a humidified

atmosphere containing 5% COz, as described (Reese and Katzenellenbogen, 1991). Expression vectors for Type 1 (pBSa-REDl, Andersson et al., 1989) and Type 2 (pT801, Normington and Russell, 1992) were constructed from

cDNAs kindly provided by D.W. Russell. A XhoI-XbaI partial cDNA fragment encoding the rat Type I 5a- reductase isozyme and a BarnHI-SalI fragment containing a partial rat cDNA encoding the Type 2 isozyme were subcloned into a mammalian expression vector pSVL (Pharmacia, Piscataway, NJ). The transfection of COS7 cells was accomplished by mixing lipofectin (Gibco- Bethesda Research Laboratories) 1 pg/pl with plasmid DNA in a ratio of 10:3 (w/w). Following lipofection, the cells were incubated for 24 h prior to replating for immunocytochemistry or for 48 h before harvesting for immunoblotting or Sa-reductase assay. Transfection effi- ciency was monitored with a /I-galactosidase expression vector (pRSV 1acZ) (Hall et al., 1983).

2.8. Sa-Reductase assay Tissue homogenate (3OOpg protein) or COS cell ex-

tract (5Opg protein) was incubated in 10 mM KzP04 (at pH 6.6 for Type 1 reductase activity; pH 4.7 for Type 2 reductase activity) (Andersson et al., 1989), 150 mM KCl, 1 mM EDTA, 500pM NADPH, and 1 ,&i [14C]- testosterone (Du Pant Canada Inc. NEN Research Prod- ucts, Mississauga, Ontario) for 1 h at 37*C. Products were separated using thin layer chromatography. Dichloro-

methane and diethyl ether (8:2, v/v) was used as the running solvent. The thin layer chromatography plates were then autoradiographed (Ordman et al., 1991; Normington and Russell, 1992).

2.9. Immunocytochemistry Twenty-four hours after transfection, the COS7 cells

were replated onto cover slips treated with poly-L-lysine. Forty-eight hours later the slides were prepared by fixing

the cells for 30 min with 3% paraformaldehyde in PBS.

The intrinsic fluorescence due to fixation was quenched with 0.1 M glycine/PBS (5 min). The cells were perme- abilized with 0.5% Triton X-100 for 30 min at room tem- perature and subsequently blocked with 5% normal goat

serum (G&co-Bethesda Research Laboratories) for 30 min. Test antisera 5a-299 or 5a-3 10 (1: 100 dilution)

was incubated with the cells overnight at 4”C, and then washed five times in PBS before incubation with FITC- conjugated goat anti-rabbit immunoglobin (10 pg/ml) (Inter Medico, Markham, Ontario) for 1 h at room tem- perature with gentle shaking. The cell preparations were

again washed five times in PBS before mounting in glyc-

erol/PBS (1: 1; v/v) (Pringle, 1991). Photomicrographs were taken on a Zeiss Axiophot using Kodak T-max film

(ASA 3200).

3. Results

3.1. Sa-Reductase isozymes are localized to the nuclear periphery in transfectedfibroblasts

To carry out studies which would enable the subcellu-

lar localization of Sa-reductase, two antisera were obtained from rabbits immunized with Type 1 Sa-reductase peptides; antiserum 5a-299 from a rabbit injected with

peptides representing amino acids 30-48, 165-183, and 236-249 and antiserum 5a-310 from a rabbit injected with peptides corresponding to amino acids 165-183 and

236-249 conjugated to myoglobin. Although both antisera were generated against Type 1 peptides, as the two isozymes of the rat Sa-reductase are highly homologous to each other (Russell and Anderson, 1991), we assessed the specificity of each antiserum for the Type 1 and Type 2 expressed protein. Vectors for the expression of Type 1 or Type 2 Sa-reductase in COS7 cells following transient transfection were constructed. While control COS cells possess some endogenous Sa-reductase activity measured by conversion of 14C-labelled testosterone to i4C-labelled DHT, COS cells transfected with either Type 1 or Type 2 cDNA demonstrated significantly increased Sa-reductase activity measured at the appropriate pH values (Fig. 1A). Immunoblots of crude nuclear and cytosolic subfractions of COS cells after transfection with either Type 1 or Type 2 Sa-reductase were probed with each of the two antisera. The amount of the endogenous Sa-reductase in the COS cells was too low to be detected on immunoblots which had been exposed for periods long enough to demonstrate


4 5 6

5c+299 + peptides

123456 kDa

C 5a-310

123456

5a-310 + peptides

Fig. 1. Detection of cloned rat steroid Sa-reductase Types 1 and 2 transiently expressed in COS7 cells. A pSVL expression plasmid containing either the rat type 1 or Type 2 cDNA was introduced into COS7 cells as described in Section 2. Cells were then processed for measure- ment of 5a-reductase activity and for cross-reactivity to rabbit antiserum 5a-299 or 5a-310. (A) 5a-Reductase activity was assayed after incubation of [‘4C]testosterone with 5Opg cell extracts for 1 h at 37’C in 5OOpM NADPH. Steroid products were separated by thin layer chromatography and autoradiographed. Lane 1, testosterone; lane 2, dihydrotestosterone; lane 3, untransfected COS7 cells; lane 4, COS7 cells co-transfected with Type 1 cDNA and &gal cDNA; lane 5, COS7 cells co-transfected with Type 2 cDNA and/?-gal cDNA; lane 6, COS-7 ceils transfected only with the /Lgalactosidase construct. (R),(C) Im- munoblots of crude cytosolic (5Opg protein per lane) and nuclear fractions (5Opg protein per lane) with rabbit antiserum 5a-299 (R) or 5u- 310 (C) (1:lOOO dilution) were obtained from COS ceils transfected with the j?-galactosidase vector (Control), or COS7 cells transfected with Type 1 or Type 2 Sa-reductase cDNA and the /J-gal vector. Identi- cal blots obtained in the presence of an equimolar mixture of the pep tides used to generate each antiserum (final concentration 10 @ml) are also shown. Lane 1, cytosolic fraction from control cells; lane 2. nuclear fraction from control cells; lane 3. cytosolic traction from cells transfected with Type 1 mductase; lane 4, nuclear fraction from cells transfected with Type 1 mductase; lane 5, cytosolic fraction from cells transfected with Type 2 reductase; lane 6, nuclear fraction from cells transfected with Type 2 reductase.. Identical results were obtained in three separate experiments.

transiently expressed Sa-reductase. Antiserum 5a-299 detected a cross-reactive component at M, 26 000 in subfractions from cells transfected with either Type 1 or Type 2 Sa-reductase (Fig. 1B). On the other hand, 5a-310 detected a component at M, 26 000 only in cells which had been transfected with the Type 1 reductase cDNA (Fig. 1C). Thus 5a-310 was judged to be specific for the Type 1 Sa-reductase, while 5a-299 detected both 5a- reductase isozymes.

On control immunoblots, synthetic peptides which had been used to generate each antiserum competed for binding of the antiserum to the 26 000 M, component (Fig. lB,C). Furthermore, pre-immune serum showed no cross- reactivity with the COS cells transfected with either Type 1 or Type 2 Sa-reductase (results not shown).

The subcellular localization of the transiently expressed proteins was then determined by cytoimmunofluorescence. Fig. 2 shows that 5a-299 cross-reacted with antigen localized to the nuclear periphery and extending into the cytoplasm in cells transfected with either the Type 1 or the Type 2 Sa-reductase cDNA. The immunofluorescence pattern obtained with 5a-3 10 in cells transfected with the Type 1 Sa-reductase was identical to that obtained with 5a-299 but, consistent with results shown

Sa-299 5a-310

Fig. 2. Subcellular localization of Sa-reductase in COS7 cells transfected with Type 1 or Type 2 cDNAs detected with 5a-299 and 5a-310. Immunofluomscence micrographs of COS7 cells, after co-transfection with either Type. 1 or Type 2 Sa-reductase expression vectors and a p- galactosidase expression vector, obtained after incubation of the cells with either antiserum 5a-299 and 5a-310. Similar distributions were observed in cells from two separate experiments.

142 J.G.A. Savory et al. I Molecular and CeNulur Endocrinology II0 (1995) 137-147

A

B

5a-299 5c.P310

I cv CT I CV CT

5a-299 5a-310 C N C N

46-

30-

21.5-

kDa

Fig. 3. Detection of anti-5a-reductase crossreactivity in rat liver cellular fractions. Liver fractions (5Opg protein per lane) were subjected to electrophoresis on a 20% SDS-polyacrylamide gel. The proteins were transblotted and probed with 5~~299 or Sa-310 (dilution 1:looO). Im- munodetection was by enhanced chemiluminescence. (A) Rat liver homogenate obtained from normal intact animals (I), castrated animals treated with vehicle (CV), castrated animals treated with testosterone (CT). (B) Crude nuclear (N) or crude cytosolic (C) subfractions of rat liver from normal intact animals. This experiment was repeated three times and results are from a representative immunoblot.

in Fig. 1, only background fluorescence was observed in cells transfected with the Type 2 plasmid .

3.2. Rat liver Sa-reductase is distributed between cytosolic and nuclear subfractions

We then investigated the cross-reactivity of the antisera to rat liver homogenate. Both antisera detected 5a- reductase at M, 26 000 in homogenates obtained from livers of normal male rats, rats which had been castrated 14 days previously and animals who had been castrated for 7 days and then treated with testosterone for 7 days (Fig. 3A). As described in Section 2 we determined that it was valid to obtain an approximation of the relative amount of Sa-reductase immunoreactivity in rat liver fractions on immunoblots by quantification of the cross- reactivity using densitometric determinations of immunoreactive material at il4, 26 000. The peak areas varied linearly with increasing homogenate protein if exposure times were chosen to ensure that ODs were between 0.2 and 2.0 OD units. Detection with either antisera indicated an increase of approximately 2-fold in the M, 26 000 component’upon castration of the animal and a reduction to normal values in the fractions from rats treated with hormone. This finding concurs with previ-

ously reported results for the response of Sa-reductase activity to androgenic manipulation in the rat liver (Yates et al., 1958) and together with the detection of immunoreactivity reported above in COS7 cells is convincing evi- dence that the crossreactive component was indeed the Sa-reductase. Furthermore, since the reductase was detected by 5~~310, our studies confirm that the Type 1 reductase is present in the rat liver. However, ss neither of the antiserum was specific for the Type 2 isozyme, we cannot rule out the existence of the Type 2 isozyme, based on our studies.

Localization of the Sa-reductase was then determined in crude nuclear and cytosolic subfractions obtained from livers of intact animals (Fig. 3B). Both antisera detected a component at M, 26 000 in crude cytosolic and nuclear subfractions. We calculated total relative units for cross- reactivity in crude nuclear and crude cytosolic subfractions derived from 10 g of liver tissue from densitometric analysis, and found that approximately 60% of the 5a- reductase immunoreactivity to 5a-299 or 5a-310 was recovered in the crude nuclear fraction and 40% in the crude cytosolic fraction. Importantly, we were also able to demonstrate a shift in this distribution toward the cytoplasm if we used harsher means of homogenization; when homogenization was achieved with a Dounce apparatus, increasing the numbers of strokes with the pestle from 10 to 20 to 30 resulted in an increase in the proportion of 5a- reductase activity recovered in the cytoplasm from approx. 30% to 95% while the total immunoreactivity measured remained constant (Table 1).

Cross-reactive components at M, 52 000 were also sometimes observed (see immunoblot with antiserum 5a- 310, Fig. 3B). This was of interest in light of previous reports of a rat liver protein of M, 50 000 that demonstrated Sa-reductase activity (Liang et al., 1985) and a rat ventral prostate protein identified by photoaffinity label- ling with a specific Sa-reductase inhibitor (Enderle Schmitt et al., 1989a). The cross-reactive M, 52 000 species displayed some specificity as no component at this M, was detected with pre-immune serum, or on control

Table 1

Effect of method of homogenization on distribution of Sa-reductase immunoreactivity between crude nuclear and crude cytosolic fractions of rat liver

Homogenization Protein Scr-Reductase immunoreactivity procedure (no. of strokes 5a-310 So-299 with Dounce) (96 Total) (96 Total) (% Total) pestle -. -

N C N C ,d C

1. 10 18 82 70 30 66 34 2. 20 18 82 37 63 40 60 4. 30 22 78 6 94 2 98

N, crude nuclear fraction; C, crude cytosolic fraction.


A 6

30 -

21.5 m

46-

30-

kDa 21.5 - s M

C 2.0- z

F 0

g lS-

2 - z 1.

s

P2 Pl

9 Y 0.

z

kE 1

25 50

H Pl P2 P3

P3

I

I I

75 100

% MICROSOMAL PROTEIN

Fig. 4. (A) Immunoblot of the purified high speed supematant (S) and microsomes (M) (SOpg protein) probed with 5a-299. (B) Immunoblots of subfractions obtained after further fractionation of the microsomal pellet; H, microsomal pellet; Pl, P2 and P3 am the collected fractions (5Opg protein). (C) Quantification of crossreactivity at Mr 26 000. The band detected by immunoblotting at Mr 26 000 was quantified by densitometry as desCribed in Section 2. The horizontal axis represents the relative protein content of the individual fractions as a proportion of total microsomal protein. tir bars am SD from the results of three experiments.

blots in which primary antiserum had been omitted (results not shown). Furthermore, excess synthetic peptide was able to compete for binding to the antisera. However, excess peptide was always a less effective competitor for binding of the antibody to this component than to the 26 000 M, component and, as the appearance of the cross- reactive entities at M, 52,000 was inconsistent, we are unable to draw firm conclusions about their physiological relevance.

3.3. Rat liver Sa-reductase is localized to the outer nuclear membrane/rough endoplasmic reticulum

As our investigations in COS cells and rat liver indicated a dual localization of the Sa-reductase, i.e. to both nuclear and cytosolic subfractions, we extended studies to further localize the Sa-reductase by subfractionating liver microsomes and liver nuclei. In all experiments, immunoblots were obtained with both 5a-299 and 5~~310 with identical results. However, since 5a-299 was more

sensitive than 5a-310 with less background interference (see Fig. 3), all subsequent immunoblots of rat liver subfractions shown were obtained with 5a-299.

Fig. 4A is an immunoblot of the crude cytosolic fraction after separation into a microsomal pellet and supernatant. A prominent component in the pellet was detected at Mr 26,000. However, in the supernatant, we observed a lower M, entity at 23 000. Quantification by densitometry of the M, 26 000 entity in the microsomal pellet and of the M, 23 000 entity in the microsomal supematant, indicated that the majority of the microsomal cross-reactivity (83%) was recovered in the microsomal pellet at M, 26 000 although the microsomal pellet contained 44% of the total microsomal protein. Furthermore, when we assayed for Type 1 Sa-reductase activity, we were unable to detect any activity in 1Opg of the supernatant, whereas activity of an equal concentration of the microsomal pellet was easily detectable on chromatograms. We conclude that the entity at 23 000 M, is a minor component of the rat liver Sa-reductase activity, but further work is needed to establish whether it has been generated as a conse- quence of proteolysis of the 26 000 M, Sa-reductase.

We then further subfractionated the rat liver microso- ma1 pellet on a discontinuous sucrose density gradient by a procedure we have previously optimized and fully described (La Casse et al., 1990). This subfractionation procedure yields three major fractions: particulate material obtained at the 1.03/1.14 g/ml (Pl), 1.14/1.18 g/ml (P2), and 1.18/1.23 (P3) g/ml interfaces. Using 5’-nucleotidase as a marker for plasma membrane and glucose 6-phosphatase as a marker for endoplasmic reticulum we established previously that this procedure yielded plasma membrane enriched fractions (Pl) and an endoplasmic reticulum enriched fraction (P2), although endoplasmic reticulum contaminates the Pl fraction to some extent. P3 also contains endoplasmic reticulum, and we would ex- pect enrichment of rough endoplasmic reticulum membranes in this heavier fraction. The M, 26 000 component was detected in each fraction (Fig. 4B), as well as a higher molecular weight species which appeared to be identical to that observed in Fig. 3. Although a relationship between the amount of the higher molecular weight entity at Mr 52 000 with that of the M, 26 000 component was sought in several subfractionations, there was no consistent correlation between the amount of these two polypeptides. As noted earlier, we remain unsure of the identity of the 52 000 M, entity. Fig. 4C shows that P2 and P3 which contained the greatest amount of glucose 6- phosphatase possessed approx. 75% of the total microso- ma1 cross-reactivity to 5a-299. In fact, most (64%) of the microsomal activity is localized to P3 which is expected to be predominantly rough endoplasmic reticulum (La- Casse et al., 1990).

Having determined the subcytosolic localization of the rat liver Sa-reductase to the endoplasmic reticulum, we then determined the localization of the Sa-reductase in

144 J.G.A. Savory et al. I Molecular and Cellular Endocrinology II0 (1995) 137-147

A

46-

30-

21.5 -

kDa PM N MT

5a-299

30-‘_

1234567

119D5-Fl

1234567

Fig. 5. Immunodetection of 5a-reductase in purified subcellular frac-

tions from crude rat liver nuclei. Subcellular fractions from the crude

rat liver nuclei were isolated and immunodetection with 5a-299 was

performed as described in Section 2. (A) Immunodetection of compo-

nents in isolated plasma membranes (PM), nuclei (N) and mitochondria

(MT) (5Opg protein). (B) Effect of detergent treatment of isolated

nuclei on the association of So-reductase. Before being recovered by

centrifugation, nuclei (IOOpg) in 0.25 M sucrose containing 0.1 mM

MgC12 were incubated on ice for 30 min with vortexing in the presence

of lane 2, buffer; lane 3, 0.1% Triton X-100; lane 4, 0.3% Triton X-

100; lane 5, 0.5% Triton X-100; lane 6, 0.75% Triton X-160; lane 7,

1.00% Triton X-106. Lane 1 is a control in which the nuclei have not

been incubated on ice before being recovered by centrifugation. Recov-

ered pellets were resuspended in 0.25 M sucrose containing 0.1 mM

MgCI.2 and solubilized in SDS-sample buffer before running on SDS-

PAGE, tnnsblotted and probed with anti-5a-reductase or anti-hunin B 1

antibodies. Each of these results has been obtained in at least two sepa-

rate experiments.

the crude nuclear fraction from male rat liver tissue. Crude nuclei were first purified on a two-layered step gradient of sucrose (Fleischer and Kervina, 1974). On this gradient, the nuclei were recovered in the pellet whereas mitochondria and plasma membranes formed a band at the interface. The protein loaded on the SDS-gel represented 1.25% of the total plasma membrane protein, 0.4% of the total nuclear protein and 0.1% of the total mito- chondrial protein. Thus Fig. 5A confirms that most of the Sa-reductase in the crude nuclear fraction is associated with the nuclei and not with mitochondria or with plasma membrane, consistent with the findings from the subfractionation of the microsomes. This result coupled with the observation in COS cells of a localization to the nuclear periphery prompted us to investigate the potential nuclear envelope association of the Sa-reductase. The nuclei were further purified by washes in sucrose. We have used this method of purification extensively as the starting material for the preparation of nuclear envelopes which have been shown to have only low levels of glucose 6phosphatase, the marker for endoplasmic reticulum (Howe11 and Le- febvre, 1989). We routinely verified the purity of the nuclei obtained in this preparation by the absence of cyto-

plasmic tags and cellular debris as determined by phase contrast microscopy. Incubation of nuclei with the detergent, Triton, strips the outer nuclear membrane from isolated nuclei (Kline et al., 1981). Indeed, when purified isolated nuclei were incubated with increasing concentra-

tions of Triton (Fig. 5B), the Sa-reductase was removed from nuclei, indicating an outer nuclear envelope local-

ization of the Sa-reductase. The intactness of the residual nuclear structure was confirmed by the continued pres-

ence of the lamin Bl, a protein of the fibrous lamina, a proteinaceous layer apposed to the inner aspect of the inner nuclear envelope, in the detergent-treated nuclei (Gerace et al., 1984).

We have previously reported the isolation procedure and full biochemical and enzymatic characterization of male rat liver nuclear envelopes (Howe11 and Lefebvre, 1989). The nuclear envelope preparation contains less than 3% of the total nuclear DNA and has very low con-

tamination by lysosomes, mitochondria, plasma membranes and endoplasmic reticulum as assessed by deter-

mination of acid phosphatase, succinic dehydrogenase, 5’-nucleotidase and glucose 6-phosphatase activities, respectively. To further define the interaction of the protein

A 5a-299 119DCF1 6A5

46- 97.4 -

30-

69-

21.5 -

14.3 - 46-

kDa I II M NE M NE 1111 NE

0 46-

66-

21.5.

123456

Fig. 6. Association of Sa-reductase with the nuclear envelope. (A)

lmmunodetection of Sa-reductase, hunin Bl and cytochrome P450 in

microsomal and nuclear envelope preparations. Microsomal (M) and

nuclear envelope (NE) preparations were run on SDS-PAGE

(lOO&lane), transblotted and probed with anti-5a-reductase (5a-299).

anti-lamin Bl (119D5-Fl), or anti-cytochrome P450 (6A5) antisera. (B)

Effect of NaCl and detergent extraction on Sa-reductase association

with isolated rat liver nuclear envelopes. Rat liver nuclear envelope

(2OOpg protein) was aliquoted into tubes and immediately extracted on

ice, with vortexing for 30 min with either buffer (lane 2) or 0.1 M (lane

3). 0.6 M (lane 4), or 1 .O M NaCl (lane 5) or 1% Triton X-106 (lane 6).

The pellets were centrifuged for 1 min in a microfuge, the supematant

was removed, and the pellets were washed twice more in buffer. The

pellets were then immediately solubilized in SDS-sample buffer, run on

a 20% gel, transferred to a PVDF membrane by electroblotting and

processed for immunodetection of Sa-reductase as described in Section

2. Untreated nuclear envelopes which had not been subjected to incu-

bation for 30 min are also shown as a control (lane I). Identical results

were obtained in two separate experiments.


with the nuclear envelope, in a final series of experiments we investigated the association of the Sa-reductase with the nuclear envelope. In the first experiment we compared the presence in microsomal and nuclear envelope fractions of lamin Bl, a protein restricted to the nuclear envelope, and cytochrome P450, an enzyme of the endoplasmic reticulum and outer nuclear membrane with that of the Sa-reductase. Fig. 6A shows that whereas there is no lamin Bl detected in the microsomal fraction, 5a- reductase and cytochrome P450 are detected in both subfractions.

Finally, to determine the nature of the association of the Sa-reductase with the nuclear envelope, extractability of the nuclear envelope Sa-reductase in NaCl and Triton was examined (Fig. 6B). Extraction of the nuclear envelope preparation with increasing concentrations of NaCl to 1 M did not succeed in removal of any of the antigenic component, indicating that the enzyme is not a peripher- ally associated protein. Furthermore incubation of the isolated nuclear envelope in 1% Triton X-100 removed the Sa-reductase from the membrane indicating that the 26 000 M, entity in the isolated nuclear envelope fraction is an integral membrane protein.

4. Discussion

We have localized the Sa-reductase in the rat liver to the outer nuclear membrane which is continuous with and difficult to distinguish from the endoplasmic reticulum. Our study based on extensive subcellular fractionation identifies for the first time the association of rat liver reductase with the nuclear envelope. Furthermore, our investigations of Sa-reductase distribution in rat liver and in COS7 cells transiently expressing either of the reductases allow us to conclude that both Type 1 and Type 2 5a- reductases have an identical subcellular localization at the nuclear periphery.

Antiserum 5a-310 recognized only the Type 1 isozyme in cellular subfractions by Western blotting and in cells by immunofluorescence while antiserum 5a-299 recognized both isozymes. Since the only difference in the mixture of Type 1 peptides used to generate the two antisera was that an additional peptide corresponding to amino acids 30-48 was included for 5a-299, one would suspect that the amino acid sequence from 30-48 is homologous in the two isozymes. However this is not the case; in fact, the only region of the rat Type 2 enzyme displaying some homology to the human Type 2, but which is distinct from the amino acid sequence of the Type 1 reductase is from position 8 to 47 (Andersson et al., 1989). Further, the peptides corresponding to amino acids 165-183 and 236-249 of the rat Type 1 reductase are highly homologous to those of the Type 2 reductase (Russell and Wilson, 1994). Thus the cause of the specificity of 5a-310 for the Type 1 enzyme remains un- known.

In liver which contains exclusively Type 1 5a- reductase activity (Moore and Wilson, 1973), both antisera detected a component in crude nuclear and cytosolic rat liver subfractions of M, 26 000. Both antisera also detected a component in COS cell extracts after transfection with the Type 1 cDNA at M, 26 000 and 5a-299 detected a component of M, 26 000 in COS cells expressing the Type 2 reductase. The predicted molecular weights of the Type 1 and 2 reductases based on the open reading frames of their cDNA are 29 340 (Andersson et al., 1989) and 28 750 (Normington and Russell, 1992), respectively. However, on SDS PAGE gels the Type 1 has been reported to migrate with an apparent M, of about 26 000 after expression in an in vitro translation system (Andersson et al., 1989). Thus the migration of the liver Sa-reductase Type 1 is not indicative of a post-trans- lational modification. Using as antigens a trp E-5a- reductase fusion protein ‘expressed in bacteria or a synthetic oligopeptide corresponding to the C-terminus of rat Sa-reductase Type 1, Hiipakka et al. (1993) produced rabbit polyclonal antibodies. Both antibodies reacted with a single component of rat liver microsomes of M, 26 Ooo on Western blots. Immunoprecipitated proteins from CHO cells stably transfected with the Sa-reductase cDNAs have been characterized with polypeptide antisera directed against amino acid residues 232-256 of the Type 1 isozyme and amino acid residues 227-251 of the Type 2 isozyme (Thigpen et al., 1993). The anti-Type 1 serum identified a polypeptide with a M, of 23 000 while the anti-Type 2 serum identified a Sa-reductase Type 2 protein with a M, of 21 000. The antibodies generated against the fusion protein described above (Hiipakka et al., 1993) also recognized a M, 23 000 component on Western blots of detergent extracts of rat ventral prostate nuclei. In only one instance is a lower molecular weight entity recognized by 5a-299 or 5a-310; in the liver high speed microsomal supematant a Mr 23 000 component was detected suggesting the possibility of proteolytic cleavage of the 26 000 component. Since this was calculated to be a mi- nor component of the microsomal Sa-reductase, the soluble component may have been generated during the fractionation procedure. It will be important to determine whether the differences in reported molecular weights of reductases from various tissue and cell source6 are the result of tissue-specific modifications.

Results from cytoimmunofluorescence and subcellular fractionation led us to the consideration of an outer nuclear membrane/endoplasmic reticulum localization of the Sa-reductase. First, immunofluorescence of COS cells transfected with Type 1 or Type 2 demonstrated a perinuclear localization of the reductases. Thigpen et al. (1993) had concluded an endoplasmic reticulum association of both isozymes expressed in CHO cells based on similar findings of a perinuclear localization, but with cross-reactivity extending into the cytoplasm and subcellular fractionation. They did not, however, test the pos-

146 J.G.A. Savory et al. I Molecular and Cellular Endocrinology 110 (1995) 137-147

sibility of nuclear envelope association. Second, we iden-

tified immunoreactive Sa-reductase in both crude nuclear and crude cytosolic fractions. Earlier studies had also

identified a dual localization in the liver although over 80% of the total homogenate Sa-reductase activity was

measured in the cytosol (Moore and Wilson, 1973) whereas in this study we detected by the approximation afforded by densitometric analysis of immunoblots ap-

proximately 40% of the total immunoreactivity in asso-

ciation with the cytosolic fraction. Our demonstration that

harsher means of homogenization shifts more of the 5a- reductase to the cytosolic fraction provides a convincing explanation for the disparity between the present findings

and those reported earlier in the rat liver. Extensive subfractionation of both microsomes and nuclei verified the

localization of the Type 1 Sa-reductase in the rat liver. The majority of the microsomal immunoreactivity was in the rough endoplasmic reticulum, and the nuclear activity was restricted to the outer nuclear membrane of purified nuclei. It is important to note that purified nuclei used for

Triton extraction and as starting material for nuclear membranes were free of cytoplasmic tags. Nonetheless outer nuclear membrane associated Sa-reductase was

identified. The difficulty in assigning the true localization of

proteins which are localized to the nuclear periphery of the cell by subcellular fractionation was brought to light during the course of these investigations. Structural conti-

nuities between the outer nuclear membrane and the rough endoplasmic reticulum have been described in a number of tissues, and biochemical studies have been interpreted as demonstrating a very close relationship

between the two membrane systems (Richardson and Maddy, 1980). This close relationship along with the fact

that the two membrane systems are at the interface be-

tween the cytoplasmic and nuclear compartments of the cell and thus can be partitioned in either compartment depending on conditions used for subcellular fractiona-

tion explain the seemingly dual localization of the Type I enzyme in the rat liver. Further, our studies suggest an explanation for the observation that exogenously expressed Type 1 and Type 2 reductases in cell lines appear to have identical subcellular distributions and yet the localization of the reductases, as identified by subcellular fractionation, in the liver and prostate is different. It has been suggested that the hormonal milieu and phospho lipid environment affect the distribution of the 5a- reductase observed after subcellular fractionation @cheer and Robaire, 1983; Bruchovsky et al., 1988). The finding of nuclear envelope-associated Type 1 Sa-reductase of- fers the possibility that a higher affinity interaction of both Sa-reductase isozymes with the outer nuclear membrane in the prostate is responsible for the shift to a more nuclear localization of the prostatic reductase compared to the liver. Alternatively, it has been suggested that the different tissue subcellular localization may reflect a dif-

ference in the proliferation of the endoplasmic reticulum

(Russell and Wilson, 1994). The original findings of a nuclear membrane associa-

tion of the prostatic Sa-reductase, which is mostly com- prised of the Type 2 activity responsible for the genera- tion of biologically active androgen in several tissues, allowed the speculation that the Type 2 enzyme was stra-

tegically positioned on the nuclear envelope close to the androgen receptor. Kaufman et al. have recently proposed

a model that enables the Sa-reductase enzyme to influ- ence the dissociation of testosterone receptor complexes by engaging in coupling with the androgen receptor (Kaufman et al., 1993). A different distribution of the

Type 1 reductase, whose role is not known but whose affinity for testosterone is such that one can assume that it

is not to biologically provide androgen to the androgen receptor, would have been supportive of this model. However, our results suggest that the cellular localization of the Type 1 and Type 2 Sa-reductases is identical and

therefore our work provides no support for a functionally related localization of the Sa-reductase.

Acknowledgments

We are grateful to Dr. D.W. Russell for his gift of the

Sa-reductase plasmids. We thank Dr. Joanna Kwast- Welfeld for helpful discussions and Drs. Peter Walker and

Marc Ekker for comments on the manuscript. This work was supported by an operating grant from the Medical Research Council of Canada to Y. Lefebvre. R.J.G. Hacht is a Scholar of the Medical Research Council of Canada and the Cancer Research Society Inc. E.C. LaCasse was the recipient of a Studentship from the Medical Research

Council of Canada.

References

Anderson, K.M. and Liao, S. (1968) Nature 219,277-279.

Andersson, S. and Russell, D.W. (1990) Proc. Natl. Acad. Sci. USA 87,

3640-3644. Andersson, S., Bishop, R.W. and Russell, D.W. (1989) J. Biol. Chem.

264, 16249-16255. Andersson, S., Berman, D.M.. Jenkins, E.P. and Russell, D.W. (1991)

Nature 354, 159-161. Berman, D.M. and Russell, D.W. (1993) Proc. Nat]. Acad. Sci. USA

90.9359-9363. Bruchovsky, N. and Wilson, J.D. (1968) J. Biol. Chem. 243, 2012-

2021.

Bruchovsky, N., Rennie, P.S., Batzold, F.H., Goldenberg. S.L.,

Fletcher, T., Courtin, F., Pelletier, G. and Walker, P. (1985) Endo-

crinology 117.2527-2533.

Dignam, J.D., Lebovitz, R.M. and Roeder, R.G. (1983) Nucleic Acids

Res. 11,1475-1489. Enderle-Schmitt, U., Volck-Badouin, E., Schmitt, J. and Aumuller, G.

(1986) Biochemistry 25.209-217.

En&de-Schmitt, U., Neuhaus, C. and Aumuller, G. (1989) Biochim.

Biophys. Acta 987,21-28. Enderle-Schmitt, U., Seitz. J. and Aumuller, G. (1989a) J. Steroid Bio-

them. 33.379-387.


Failer, B., Farley, D. and Nick, H. (1993) Biochemistry 32,5705-5710. Reischer, S. and Kervina, M. (1974) Methods Enzymol. 31.641. Gerace. L., Comeau, C. and Benson, C. (1984) J. Cell Sci. (Suppl. 1)

137-160. Gerace, L., Ottaviano, Y. and Condor-Koch, C. (1982) J. Cell Biol. 95,

826-837. Golf, S.W. and Graef, V. (1978) J. Steroid B&hem. 9, 1087-1092. Griffin, J.E. and Wilson, J.D. (1989) in The Metabolic Basis of Inher-

ited Disease (Striver C.R., Beaudet, A.L.. Sly, W.S. and Valle, D., eds.), pp. 1919-1944, McGraw-Hill, New York.

Gustafsson, J.-A., Mode, A., Norstedt, G. and Skett, P. (1983) Annu. Rev. Physiol. 45.51-60.

Hall, C.V., Jacob, P.E., Reingold. G.M. and Lee, F. (1983) J. Mol. Appl. Genet. 2, 101-109.

Hiipakka, R.A., Wang, M., Bloss, T., Ito, K. and Liao. S. (1993) J. Steroid Biochem. Mol. Biol. 45.539-548.

Houghten. R.A. (1985) Proc. Natl. Acad. Sci USA 82,5131-5135. Houghten, R.A., Bray, M.K., Degraw, J.T. and Kirby, C.J. (1986) Int. J.

Peptide Res. 27,673-678. Houston, B., Chisholm, G.D. and Habib, F.K. (1985) J. Steroid Bio-

them. 22.461467. Howell, G.M. and Lefebvre, Y.A. (1989) J. Steroid Biochem. 33, 977-

986. Hudson, R.W. (1981) J. Steroid B&hem. 14,579-584. Imperato-McGinley, J. and Gamier, T. (1986) Trends Genet. 2, 130-

133. Jameson, B.A. and Wolf, H. (1988) Comp. Appl. Biosci. 4, 181-186. Jenkins, E.P., Andersson, S., Imperato-McGinley, J., Wilson, J.D. and

Russell, D.W. (1992) J. Clin. Invest. 89.293-300. Kaufman, M.L., Pinsky, L.. Trifiro, M., Lumbroso, M.. Sabbaghian, N.

and Gottlieb, B. (1993) J. Steroid Biochem. Mol. Biol. 45, 467- 476.

Kay, R.R., Fraser, D. and Johnston, J.R. (1972) Eur. J. Biochem. 30, 145-154.

Kline, L.D., Lefebvm, F.-A.T. and Lefebvre, Y.A. (1981) 1. Steroid Biochem. 14.855-860.

Kwast-Welfeld, J., de Belle, I., Walker, P.R., Whitfield, J.F. and Sikor- ska, M. (1991) Cancer Res. 51.528-535.

Kwast-Welfeld, J., de Belle, I., Walker, P.R., Whitfield, J.F. and Sikor- ska, M. (1994) J. Biol. Chem. 268.19581-19585.

La&se, EC., Howell, G.M. and Lefebvre, Y.A. (1990) I. Steroid Biochem. 35.47-54.

Lefebvre, Y.A. and Novosad, Z. (1980) B&hem. J. 186,641-647. Lefebvre, Y.A., Golsteyn, E.J. and Michiel, T.L. (1985) J. Steroid Bio-

them. 23, 107-l 11. Liang, T., Cheung, A.H., Reynolds, G.F. and Rasmusson, G.H. (1985)

J. Biol. Chem. 260,4890-4895. McGuim, J.S. and Tomkins, G.M. (1960) J. Biol. Chem. 235, 1634-

1638. McLaughlin, M.G. (1988) J. Clin. Endocrinol. Metab. 67, 806-816. Moore, R.J. and Wilson, J.D. (1972) J. Biol. Chem. 247.958-967. Moore, R.J. and Wilson, J.D. (1973) Endocrinology 93.581-592. Moore, R.J. and Wilson, J.D. (1976) J. Biol. Chem. 251.5895-5900. Moore, R.J., Griffin, J.E. and Wilson, J.D. (1975) J. Biol. Chem. 250,

7168-7172. Normington, K. and Russell, D.W. (1992) I. Biol. Chem. 267, 19548-

19554. Omrani, G.R., Furudawa, H., Sherwood, J.A. and Loeb, J.N. (1983)

Endocrinology 112, 178-186. Grdman, A.B., Farley, D., Meyhack, B. and Hansputer, N. (1991) J.

Steroid Biochem. Mol. Biol. 39.487-492. Pringle, J.R. (1991) Methods Enzymol. 194,565-607. Reese, J.C. and Katzenellenbogen, B. (1991) Nucleic Acid Res. 19,

6595-6602. Richardson, J.C.W. and Maddy, A.H. (1980) J. Cell Sci. 43.269-277. Richardson, J.C.W. and Maddy, A.H. (1980) J. Cell Sci. 43.269-277. Russell, D.W. and Wilson, J.D. (1994) Annu. Rev. Biochem. 63, 25-

61. Sargent, N.S.E. and Habib, F.K. (1991) J. Steroid Biochem. Mol. Biol.

38.73-77. Scheer. H. and Robaire, B. (1983) B&hem. J. 211.65-74. Thigpen, A.E. and Russell, D.W. (1992) J. Biol. Chem. 267, 8577-

8583. Thigpen, A.E., Cala, K.M. and Russell, D.W. (1993) J. Biol. Chem.

268, 17404-17412. Tunn, S., Hoochstrate, H., Habenicht, U.-F. and Krieg, M. (1988)

Prostate 12.243-253. Yates, F.E., Herbst, A.L. and Uquhart, J. (1959) Endocrinology 63,

887-902.

5α-reductase type 1 is localized to the outer nuclear membrane

Documents