Identification and immunohistochemistry of retinol dehydrogenase from bovine retinal pigment epithelium

Biochimica et Biophysica Acta, 1163 (1993) 201-208 © 1993 Elsevier Science Publishers B.V. All rights reserved 0167-4838/93/$06.00

201

BBAPRO 34489

Identification and immunohistochemistry of retinol dehydrogenase from bovine retinal pigment epithelium Yuko Suzuki, Sei-ichi Ishiguro and Makoto Tamai Department of Ophthalmology, Tohoku University School of Medicine, Sendai (Japan) (Received 26 August 1992) (Revised manuscript received 8 December 1992)

Key words: Retinol dehydrogenase; Pigment epithelium; Monoclonal antibody; 11-c/s-Retinal;Rhodopsin; Visual cycle; (Retina)

We studied the properties of retinol dehydrogenase (11-c/s-specific) from bovine retinal pigment epithelium. Detergents caused a loss of retinol dehydrogenase activity; therefore, we added 3 mM NADH as a stabilizer to solubilize this enzyme and partially purified this enzyme using Sepharose CL-6B and hydroxyapatite column chromatography. The partially-purified sample, which contained two major proteins (66 kDa, 33 kDa), had substrate preference to ll-c/s and 13-c/s-retinal but not to all-trans and 9-c/s isomers. Monoclonal anti-33 kDa protein antibody precipitated oxidation and reduction activity of retinol dehydrogenase, and was confirmed to bind specifically to 33 kDa protein of retinal pigment epithelial crude extract by Western blotting. In addition, we found that monoclonal anti-retinol dehydrogenase antibody bound specifically to retinal pigment epithelium and not to Miiller cells or to rod outer segments by immunohistochemical methods.

Introduction The initial reaction of the visual sense is the isomerization of the ll-c/s-retinal chromophore of rhodopsin to all-trans retinal by light in the photoreceptor cell. In order to regenerate photosensitive visual pigment, alltrans retinal has to be converted into the ll-cis isomer and recombined with opsin. The effective supporting system to supply the chromophore (ll-c/s-retinal) of rhodopsin is called the visual cycle, and the retinal pigment epithelium (RPE) takes a prominent part in the process. Reduction reaction of all-trans retinal to all-trans retinol (vitamin A) by all-trans-specific retinol dehydrogenase occurs in the photoreceptor cells after photobleaching. The resulting all-trans retinol leaves the photoreceptors to RPE. Then, all-trans-retinol, which is delivered though blood flow [1] and rod outer segments [2], is isomerized to ll-cis-retinol [3,4] through esterification [5-7] and de-esterification [8] in RPE. Retinol dehydrogenase present in RPE (ll-cisspecific, EC 1.1.1.105) is reportedly an oxidoreductase

Correspondence to: Y. Suzuki, Department of Ophthalmology, Tohoku University School of Medicine, 1-1 Seiryo-machi, Sendai, Miyagi 980, Japan. Abbreviations: RPE, retinal pigment epithelium; ROS, rod outer segment.

that converts ll-c/s-retinol to ll-c/s-retinal [9] and vice versa, and it is thought to be a microsomal enzyme [10]. The role of this enzyme is to supply l l-cis-retinal for the formation and regeneration of rhodopsin. The product of the enzyme reaction, therefore, is transported to rod outer segments to form rhodopsin [11,12]. Recently, we have reported the purification and partial characterization of another retinol dehydrogenase present in rod outer segments (ROS) [13]. The enzyme had substrate preference to all-trans compounds, and it was present in ROS, as demonstrated by immunohistochemical methods. Retinol dehydrogenase in RPE was so labile that no group has succeeded in purifying it. We examined the optimal conditions for solubilizing labile RPE retinol dehydrogenase in the presence of NADH, partially purified the enzyme, and identified 11-cis-retinol dehydrogenase by using polyclonal and monoclonal antibodies to the enzyme. Moreover, it is well-known that l l-cis-retinol exists in the IRBP. (interphotoreceptor retinoid binding protein) [14] and CRALBP (cellular retinal binding protein) of retinal Miiller cells [15,16]. MiJller cells especially contain ll-c/s-retinal and ll-c/s-retinol and are thought to play a role in the visual cycle. To understand the visual cycle further, we examined with immunohistochemical techniques l l-cis-retinol dehydrogenase to see if it existed in retinal cells other than RPE.

202 Materials and Methods

Preparation of retinoids. Unless otherwise indicated, experiments with retinoids were carried out in dim red light to prevent light-induced isomerization, and retinoids were preserved at -30°C under argon gas. Retinal isomers were prepared by photoisomerization of all-trans retinal (Sigma, St. Louis, MO, USA) in ethanol and collected on HPLC (Pharmacia LKB Biotechnology, Bromma, Sweden; mobile phase, diethyl ether/hexane, 8:92 (v/v), detection at 365 nm and flow = 1 ml/min, Chemosorb 5Si 4.6 mm x 260 mm) by the method of Tsukida et al. [17]. Retinol isomers were prepared by reducing the retinal isomers with NaBH 4, according to the method of Bridges and AIvarez [18], and were collected on HPLC (ethylacetate/ dichlormethane/hexane, 6.2 : 7.7 : 86.1 (v/v), 365 nm, 1 ml/min, Chemosorb 5Si) [17]. Assay of 11-cis-retinol dehydrogenase activity. Reductase activity of ll-c/s-retinol dehydrogenase was measured by the modification of Lion et al. [9]. Components of substrate solution (50 tzl) were as follows: 10 nmol of each retinal isomer, 12% acetone (v/v), 1.2% Tween-80 (w/v), 1.0 mM NADH and 0.2 M sodiumacetate buffer (pH 5.0). The reaction was started by adding 50/xl substrate solution to 50/zl enzyme solution. After an incubation at 37°C, 0.5 ml of ethanol was added to stop the reaction, and 0.2 ml of thiourea and 0.2 ml of thiobarbituric acid reagent were added to assay the retinal decreased after reduction reaction according to Futterman and Saslaw [19]. After 30 min, samples were centrifuged at 10 000 rpm (Hitachi microcentrifugator CR15B, Tokyo) for 1 min. Color development was measured at 530 nm. To avoid precipitation of salts, 50 /xl of distilled water was added to the supernatant before the measurement was performed. Oxidation reaction was analyzed qualitatively. In brief, we detected the formation of retinal isomers on HPLC after oxidation reaction. Components of substrate solution (50 /zl) were as follows: about 15 nmol of 13-cis and ll-c/s-retinol mixture or 9-c/s and all-trans retinol mixture, 12% acetone (v/v), 1.2% Tween-80 (w/v), 1.0 mM NAD ÷ and 0.2 M phosphate buffer (pH 8.0). The reaction was carried out by adding 50 /zl of enzyme solution (3.4 mg partially-purified material) to 50/~1 of substrate solution. After an incubation at 37°C, the reaction vessels were chilled in solid CO 2. Samples were freeze-dried, and the reaction product was extracted with 200/zl n-hexane. The hexane extract was centrifuged, and 100 /zl of the supernatant was analyzed on HPLC. The protein concentration was measured by the method of Bensadoun and Weinstein [20]. Solubilization of 11-cis-retinol dehydrogenase. Using the methods of Berman and Feeney [21], RPE cells were gathered from bovine eyes obtained from a local slaughterhouse. The pigment epithelial cells were then

homogenized in 5 mM Tris-HCl buffer (pH 7.0), containing 0.25 M sucrose. The RPE homogenate (0.1 mg/ml) was dissolved with 5 mM Tris-HCl buffer (pH 7.0) containing varying amounts of NADH (3" 10 -2, 3" 10 -3, 3" 10 -4, 3-10 -5, 3" 10 -6 M), 20% glycerol (v/v), 0.15 M NaCI and 1% detergents (w/v). The detergents used were Emulphogene BC-720, Triton X-100 and sodium cholate. After 30 min at 4°C, we centrifuged these mixtures at 100 000 x g for 1 h, and added 50 /zl of substrate solution containing 10 nmol of ll-cis-retinal, 12% acetone (v/v), 1.2% Tween-80 (w/v) and 0.2 M sodium acetate buffer (pH 5.0) to 50 /zl of the supernatant. Reduction reaction of ll-c/s-retinol dehydrogenase was carried out for 6 min at 37°C. Solubilized l l-c/s-retinol dehydrogenase activity is shown as percent of RPE homogenate activity.

Partial purification of ll-cis-retinol dehydrogenase. All manipulations were performed at 4°C. Microsomes of RPE from 300 bovine eyes were fractionated by the procedure of Zimmerman and coworkers [10]. We partially purified the retinol dehydrogenase of RPE using hydroxyapatite column chromatography at pH 7.0 (Method 1) or pH 6.6 (method 2). Method 1: We solubilized the microsomes with 40/zl of 5 mM phosphate buffer (pH 7.0), containing 0.15 M NaC1, 3 mM NADH, 1% Emulphogene BC-720 (w/v) and 20% glycerol (v/v); centrifuged it at 17000 x g for 30 min; and obtained the supernatant as crude extract. The crude extract was loaded onto a hydroxyapatite column (1.5 x 30 cm). Retinol dehydrogenase was eluted with a linear gradient of 0-40% ammonium sulfate using 5 mM phosphate buffer (pH 7.0), containing 0.15 M NaCI, 3 mM NADH, 0.5% sodium cholate (w/v) and 20% glycerol ((v/v), elution buffer A) at a flow rate of 0.5 ml/min. Total gradient volume was 600 ml. Fraction size was 10 g/tube. Aliquots (50 /zl) of the fractions were used to assay retinol dehydrogenase activity. The oxidation reaction was performed for 60 min at 37°C using 13-cis-retinol as a substrate. The reduction reaction was performed for 6 min at 37°C using ll-cisretinal. The fractions containing 11-cis-retinol dehydrogenase were collected, concentrated with Amicon Diaflo concentration apparatus using PM10 membrane and dialyzed against elution buffer. Method 2: We solubilized the microsomes with 5 ml of 5 mM phosphate buffer (pH 7.0), containing 0.15 M NaCI, 3 mM NADH, 1% Emulphogene BC-720 (w/v) and 20% glycerol (v/v); centrifuged it at 17000 x g for 30 min and obtained the supernatant as crude extract. The crude extract was loaded onto a Sepharose CL-6B column (1.9 cm x 100 cm) and eluted with 5 mM phosphate buffer (pH 7.0), containing 0.15 M NaC1, 3 mM NADH, 0.3% Emulphogene BC-720 (w/v) and 20% glycerol ((v/v), elution buffer B). The flow rate was 0.5 ml/min and the fraction size was 3 g. Fractions containing l l-cis-retinol dehydrogenase were pooled

203 and concentrated. The concentrated sample (about 8 ml) was loaded onto a hydroxyapatite column (0.9 cm × 16 cm), and eluted at the same flow rate and fraction size by the elution buffer B adjusted to pH 6.6. Fractions containing ll-c/s-retinol dehydrogenase were collected, concentrated and used as partially purified sample. Preparation of antibody. Using the method of Laemmli [22], SDS-PAGE of partially-purified samples was performed on 10% acrylamide gel. We collected a 33-kDa protein from the Coomassie brilliant blue R250-stained gels by electrophoresis at 50 V for 12 h. The eluted samples were concentrated by lyophilization. We mixed 10 /zg (100/zl) of the electrophoretically-purified 33 kDa protein with the same volume (100 ~1) of Freund's complete adjuvant and injected the emulsified sample intraperitoneally into mice two times. Incomplete adjuvant and 10 tzg sample were used for the second injection. The third immunization was performed using 50 /zg protein. Plasma was collected 3 days after the final injection and used as polyclonal antibody. Spleen cells of the 33 kDa protein- and rod outer segment retinol dehydrogenaseinjected mice [13] were collected and fused with P6 mouse myeloma ceils using 50% poly(ethylene glycol) 4000. Hybridoma cells were cultured in 96-well tissue culture plates, selected with HAT-selective medium, and screened by the ELISA using electrophoretically purified 33 kDa protein and rod outer segment retinol dehydrogenase. We obtained clones secreting anti-33 kDa protein antibody ( R P E R D H 4 A l l ) and anti-ROS retinol dehydrogenase antibody (ROSRDH4A8).

Absorption of 11-cis-retinol dehydrogenase activity by anti-33-kDa protein antibody. We treated ascites fluids with 50% ammonium sulfate and obtained concentrated monoclonal anti-33-kDa protein antibody (13.6 mg/ml) ( R P E R D H 4 A l l ) and monoclonal anti-rod outer segment retinol dehydrogenase (37 kDa protein) antibody (17.2 mg/mlXROSRDH4A8) as control. We mixed indicated amounts of monoclonal mouse anti-33 kDa protein antibody or anti-ROS retinol dehydrogenase antibody with partially purified l l-c/s-retinol dehydrogenase (Method 2, 1.7 tzg protein). We added 5 /zl of normal mouse serum as a carrier and adjusted the total volume to 40 /zl using varying amounts of normal horse serum. Next, we added 20/xl of 25 mM phosphate buffer (pH 7.0) containing 0.15 M NaC1, 50% glycerol (v/v), 1.5% Emulphogene BC-720 (w/v), 15 mM NADH and 20 mg/ml bovine serum albumin. After 2 h at 4°C, we added 40 tzl of goat anti-mouse IgG serum (Organon Teknika, Westchester, PA, USA) to the mixtures. After 18 h at 4°C, reaction mixtures were centrifuged at 7500 x g for 10 min. We assayed the reductase activities of retinol dehydrogenase in the supernatant for 20 rain at 37°C. Nonenzymatic color development of plasma was subtracted.

Also, we analyzed the oxidase activities of the supernatant (50 /zl) on HPLC. Components of substrate solution (50 ~1) were as follows: about 30 nmol of 13-c/s and ll-c/s-retinol mixture or 9-c/s and all-tram retinol mixture, 12% acetone (v/v), 1.2% Tween 80 (w/v), 30 mM NAD and 0.2 M phosphate buffer (pH 8.0). The oxidation reaction was carried out for 20 min at 37°C. Western blotting and immunostaining. Crude extract (5 ~g protein) was applied to SDS-PAGE. We transferred proteins to nitrocellulose membranes according to the method of Towbin et al. [23]. The nitrocellulose membrane was treated with 3% gelatin for 1 h at 37°C. The first antibody was 1:100 diluted-RPERDH4All mouse anti-33-kDa protein antibody (culture medium), and the second antibody was 1 : 5000 diluted-goat antimouse IgG antibody conjugated with alkaline phosphatase (Organon Teknika). The antibodies were diluted with phosphate buffered saline (0.14 M NaCI and 10 mM phosphate buffer (pH 7.4)) containing 1% bovine serum albumin and 0.05% Tween-20. Incubation time was 1 h at 37°C. Washing was performed after each step with phosphate buffered saline containing 0.05% Tween-20. The membrane was washed with 50 mM Tris-HCl buffer (pH 9.5) containing 150 mM NaC1. Color development was carried out with 5 ml of dye solution (100 mM NaCI, 5 mM MgC1 z and 100 mM Tris-HCl buffer (pH 9.5)) containing 33/zl NBT solution (50 mg/ml nitroblue tetrazolium in 70% dimethylformamide) and 16.5 tzl BCIP solution (50 mg/ml 5-bromo-4-chloro-3-indolyl-phosphate, p-toluidine salt in dimethylformamide). ELISA of 33 kDa protein. We performed ELISA by the modified method of Ishiguro et al. [24]. Each fraction of hydroxyapatite column chromatography (Method 1) was diluted 1:100 with carbonate-bicarbonate buffer (pH 9.6). 50/zl of this diluted solution was added to Nunc-Immuno Plate I and allowed to stand for 1 h. The incubation procedure was carried out at room temperature. 33 kDa protein bound to the wells was detected with anti-33-kDa protein antibody (1:50 diluted-RPERDH4All culture medium, 1 h) and peroxidase-conjugated goat anti-mouse IgG (1 : 500 dilution, 1 h). Antibodies were dissolved in phosphate buffered saline (PBS; 0.14 M NaCI-10 mM phosphate buffer (pH 7.4)) containing 1% bovine serum albumin and 0.05% Tween-20. The microplates were washed with PBS containing 0.05% Tween-20 between incubations. Color was developed by incubation with 50/zl of 3.7 mM o-phenylenediamine and 0.006% H 2 0 2 in citrate-phosphate buffer (pH 5.0). After stopping the reaction with 2.4 M H2504, absorbance at 492 nm was read. Immunohistochemistry. Immunohistochemical techniques were carried out according to the method of Ishiguro et al. [24], except that we used 50 /zl of

204 8O

:3. 0 . 8 o

a

,m > °m

t~

"300

p~ i

0.6

200

060 i

"1-

0.4

11:40

0,2

o!

0.0

E

vE

_¢ lOO

o o.

0

'-20

20

40

Fraction

i

0,003

i

0.03

!

i

i

0.3

3

30

60

number

b 10

20

30

NADH (mM)

40

50

Fig. 1. Effect of NADH concentration on the stability of l l-c~ retinol dehydrogenase in detergents. After the RPE homogenate was dissolved in 1% detergents containing the indicated concentration of NADH, its activity of supernatant was measured (see Materials and Methods), compared with RPE homogenate and plotted. Emulphogene BC-720, (B); Triton X-100, (o); sodium cholate, (<3).

10000-times-diluted monoclonal mouse anti-33-kDa protein antibody (1.36 g g / m l ) containing 0.06% SDS as first antibody and 200-times diluted peroxidase-conjugated goat anti-mouse IgG antibody as second antibody. We prepared control sections treated with anti-33 kDa protein antibody absorbed with excess antigen (2.5 /zg) containing 0.06% SDS in 50/xl of solution. Results We measured the stability of retinol dehydrogenase of RPE at various pH values using PRE homogenate. The enzyme activity remained about 50% from pH 3.5 to 5.5 and about 80% from pH 6.0 to 9.5 after 70 h at 4°C. The enzyme activity was most preserved from pH 6.5 to 7.5. Retinol dehydrogenase activity was seriously

1,2"

C

A g 1 . 0.8' 0,6" 0,4'

0

'

~

• •. . . . 0

20

40

.

. 60

• 80

fraction n u m b e r

Fig. 2. Results of hydroxyapatite column chromatography at pH 7.0 (Method 1), blotting and ELISA of 33 kDa protein. (a), Aiiquots (50 /.d) were used to assay ll-c/s-retinol dehydrogenase activity ((o), reduction of ll-c/s-retinal, A530, 6 min, 50 /zl; (o), oxidation of 13-c/s-retinol, A530, 60 min, 50 /zl (see Materials and Methods)). Arrow indicates start of a linear gradient of 0-40% ammonium sulfate in elution buffer A. Protein concentration (zx) was determined by the method of Bensadoun and Weinstein [20]. (b), We performed Western blotting using 1.0 /.d of these fractions and stained with RPERDH4All anti-33-kDa protein antibody (see Materials and Methods). (c), ELISA of 33 kDa protein was carried out using same fractions (details are described in Materials and Methods). Peaks of ELISA are consistent with the minor and major peaks of retinol dehydrogenase activity.

TABLE I

Results of column chromatography

Method 1 Crude extract Hydroxyapatite (pH 7.0) Method 2 Crude extract Sepharose CI-6B Hydroxyapatite (pH 6.6)

Volume (ml)

Total protein (mg)

Total activity (nmol/min)

Specific activity (nmol/min per mg)

Purification (-fold)

Yield (%)

40

64.4

18 400

286

i

100

1902

294

1.03

7.45

60

fraction numbm

6.48

10.3

5.6 7.7

24 9.4

20000 5 100

830 540

1 0.65

100 26

5.0

3.4

2 300

680

0.82

12

205 affected when dissolved with 1% detergents (w/v) (cetyltrimethylammonium bromide, dodecyltrimethylammonium bromide, sodium cholate, SDS, sodium deoxycholate, Triton X-100 and Nonidet P-40); it disappeared within 30 min. A large amount of NADH was required to stabilize the enzyme (Fig. 1). During column chromatography, ll-c/s-retinol dehydrogenase was labile, even when the enzyme was stabilized with 3 mM NADH (Table I). Con-A Sepharose and DEAE A-25 column chromatographies were not helpful in the purification process, and advanced chromatographies reduced specific activity and yield. When ll-c/s-retinol dehydrogenase was absorbed to the hydroxyapatite column at pH 7.0 and was eluted with a linear gradient of ammonium sulfate (Method 1, Fig. 2), most microsomal proteins were not removed, but specific activity increased slightly (Table I) and a prominent increase of 33 kDa protein and a decrease of 66 kDa protein was observed (Fig. 4c). On Sepharose CL-6B column chromatography, active fractions were widely distributed in the area of high molecular mass standards (data not shown). The gel filtration column chromatography removed some proteins other than 33 kDa and 66 kDa proteins, but the column chromatography was not very effective in the purification step. As the next step, we used a small hydroxyapatite column and adjusted the elution buffer (see Materials and Methods) to pH 6.6, so that the enzyme could pass through the column (Method 2). The ll-c/s-retinol dehydrogenase was effectively collected (Fig. 3). The majority of the bands were reduced after the column chromatography, but our purification folds were poor because the enzyme was labile even in the presence of excess NADH. The partially purified sample obtained by Method 2 consisted of two major proteins (Fig. 4g). The molecular mass of these protein

0.5"

_•

0.4

0.4

A

g ~,o.3 e~E a: ~ o . 2 ._= =~

e-

0.2~ D.

o 10

20

30

Fraction number Fig. 3. Results of hydroxyapatite column chromatography at pH 6.6 (Method 2). Aliquots (50 /zl) were used to assay the reductase activity of ll-c/s-retinol dehydrogenase activity using ll-c/s retinal as a substrate (o). Protein concentration (e) was determined by the same methods as described in Fig. 2. At pH 6.6, ll-c/s-retinol dehydrogenase did not bind to the hydroxyapatite column.

4

0

a

b

c

d

e

f

g

h

i

Fig. 4. SDS-PAGE summary of partial purification of ll-c/s retinol dehydrogenase from bovine RPE (a-h) and characterization of anti33 kDa protein antibodies by Western blotting and immunostaining techniques (i). Proteins were stained with Coomassie brilliant blue R-250 ((a-h), see Materials and Methods). (a), molecular mass standards: phosphorylase b (97 kDa); bovine serum alubumin (66 kDa); ovaibumin (43 kDa); carbonic anhydrase (31 kDa); soybean trypsin inhibitor (22 kDa) and lysosome (14 kDa); (b), crude extract (10 /~g protein); (c), partially-purified ll-c/s retinol dehydrogenase by hydroxyapatite column chromatography at pH 7.0 (Method 1, 10 /~g protein); (d), molecular mass standards; (e), crude extract (10 ~g protein); (0, partially-purified retinol dehydrogenase by Sepharose CL-6B (10/~g protein); (g), partially-purified ll-c/s-retinol dehydrogenase by hydroxyapatite column chromatography at pH 6.6 (Method 2, 10 /zg protein); (h), isolated 33 kDa protein (2 mg protein); (i), Western blot of crude extract (5 ~g protein) reacted with monoclonal anti-33-kDa protein antibody (see Materials and Methods). Arrows indicate 33 kDa protein. Monoclonal antibodies that precipitated ll-c/s-retinol dehydrogenase activity bound specifically to 33 kDa protein in crude extract. The faint band of the 33 kDa protein dimer is seen.

bands was 33 kDa and 66 kDa. The results of two different hydroxyapatite column chromatographies showed that the 33 kDa protein is a leading candidate for 11-c/s-retinol dehydrogenase of RPE. The partially-purified sample (by Method 2) reacted with 11-c/s and 13-c/s isomers but not with 9-c/s and all-tram isomers (Figs. 5 and 7). The same results were obtained with RPE homogenate and crude extract (data not shown). The stereospecificity was preserved during the processes of solubilization and partial purification. Then, we tried to make mouse anti-33-kDa protein antibody and anti-66-kDa protein antibody. We could not obtain anti-66-kDa protein antibody. Polyclonal anti-33-kDa protein antibody precipitated more than 80% reductase activity of retinol dehydrogenase (data not shown). As it was possible that the 33 kDa protein shown in Fig. 4 did not necessarily consist of a single protein, we made monoclonal anti-33-kDa protein antibody. Monoclonal anti-33 kDa protein antibody also precipitated almost 90% reductase activity (Fig. 6). The antibody also absorbed the oxidation activity (Fig. 7). As shown in Fig. 4, monoclonal (Fig. 4i) and polyclonal antibodies (data not shown) reacted only with 33 kDa protein. The 66 kDa protein did not appear to be

206

a

10

10

i

bi

~

6

2

0

•

"

Incubation time

"

(min)

Fig. 5. Stereospecificityof partially purified ll-c/s-retinol dehydrogenase (reduction reaction). Panel a, homogenate of retinal pigment epithelial cells (5 /zg/reaction mixture). Panel b, partially-purified ll-c/s-retinol dehydrogenase (Method 2, 1.3 /zg/reaction mixture). Reduction reaction of ll-c/s-retinol dehydrogenase was measured (see Materials and Methods). Substrate used are all-trans ([]), 9-cis (-), ll-c/s (e) and 13-c~ (o) retinal. RPE homogenate and the partially-purified ll-c/s-retinol dehydrogenase had substrate preference toward ll-c/s and 13-c/s compounds.

and immunostaining using the fractions of the hydroxyapatite column chromatography by Method 1 (Fig. 2b). We found that the antibody stained only 33 kDa protein in the major peak (fractions 26-60), whereas two bands existed in the minor peak (fractions 10 and 12). We showed that our R P E R D H 4 A l l antibody stained only 33 kDa protein in the crude extract (Fig. 4). An upper band, however, appeared in these two fractions after the column chromatography and increased if the sample stand at 4°C. These results show that minor heterogeneity may exist in the retinol dehydrogenase of RPE. We performed ELISA of 33 kDa protein using P R E R D H 4 A l l to know the relative amount of the antigen in the fractions of Method 1. As shown in Fig. 2c, peaks of ELISA was consistent with the major and minor peaks of retinol dehydrogenase activity. Immunoreactivity was present in R P E cells (Fig. 8). We found no immunoreactivity in Miiller cells and photoreceptor cells. In addition, no immunostaining was observed when we used absorbed-anti-RPE retinol dehydrogenase antibody with excess purified R P E retinol dehydrogenase and when we omitted the first antibody reaction. Discussion

bovine serum albumin, because rabbit anti-bovine serum albumin did not bind to it (data not shown). To confirm the specificity of anti-33-kDa protein antibody further, we carried out SDS-PAGE, blotting

100

100

o ~

80

80

•o = ~

60

60

~, -, ~' ~.

40

•

20

40 1 20

~M

"Of"

0

0

,

,

5

10

0

5

0

,

,

5

10

15

Antibody (1~1) Fig. 6. Precipitation of ll-c/s-retinol dehydrogenaseactivityby monoclonal mouse anti-33-kDa protein antibody (reduction reaction). Panel a, substrate used was ll-c/s-retinal. Panel b, substrate used was 13-c/s-retinal. ll-c/s-Retinol dehydrogenase partially purified by Method 2 was reacted with indicated amounts of monoclonal mouse anti-33-kDa protein antibody (o) or with monoelonal mouse antiROS retinol dehydrogenase antibody (o). Almost 90% of ll-c/s-retinol dehydrogenase activitywas precipitated by monoclonal anti-33kDa protein antibody, indicating that the 33 kDa protein is ll-cisretinol dehydrogenase.

Lion et al. [9] demonstrated two kinds of retinol dehydrogenase in ROS and R P E independently that were not liver alcohol dehydrogenase. Instead, these enzymes displayed a new scheme in the visual cycle and an important role for R P E in the formation of rhodopsin. We have shown recently that ROS retinol dehydrogenase is present specifically in the bovine retinal ROS by purification and immunohistochemical studies of the enzyme, and have concluded that the reduction reaction of all-trans retinal to all-trans retinol appears to be a specific reaction in the ROS in the eye [13]. In the present study, we p e r f o r m e d the identification and immunohistochemistry of R P E retinol dehydrogenase to find the localization of oxidation reaction of ll-c/s-retinol to ll-c/s-retinal in the retina and to understand the possible role of Miiller cells in the visual cycle. First, we attempted to solubilize R P E retinol dehydrogenase with Triton X-100 or Emulphogene BC-720 by the method of Lion et al. [9]. We, however, could not solubilize this enzyme stably, and we needed a large amount of N A D H to solubilize the enzyme while preserving its activity. The purification achieved with the 33 kDa polypeptide was quite good but a loss of its enzymatic activity occurred, presumably by denaturation or separation of a co-factor in the procedure employed. We succeeded in solubilizing this enzyme without affecting its stereospecificity and found that

207 the partially-purified enzyme converted not only l 1c/s-retinal but also 13-c/s-retinal into corresponding isomers and vice versa. The results emphasized the difference between the retinol dehydrogenase identified in the present study and that found in ROS. Through the use of chromatography, crude extract was purified partially containing two major proteins and a few other minor proteins. After partial purification, we established which of these two major proteins was the retinol dehydrogenase molecule of RPE by

a

L.I,

~ I i !

"B.2 Io

,

using specific antibody to the enzyme. Monoclonal anti-33-kDa protein antibody precipitated almost 90% of ll-c/s-retinol dehydrogenase activity, whereas monoclonal anti-rod outer segment retinol dehydrogenase antibody did not precipitate the enzyme activity under such conditions (Fig. 6). Results of partial purification by Method 1 (Fig. 4c) showed a marked increase of 33 kDa protein and supports these immunochemical results. We performed Western blotting using the crude

f

0.0005

f "7-rrl b

[o.oo?

I! J1. !

....

•- - ~

C ~

co~5

~~

"r-.r..I

I L.

r i!4!: :t .~ :i :i

.

t

g

o.00015

I:..

h

C

[;i d [ O~_~jO~O0 1 .................

e

0 15 20 r e t e n t i o n time (min)

,

o.oot A

j

0.ooo15

--

lb retention

1"5 2"o t i m e (rain)

Fig. 7. Stereospecificity ( a - e ) and precipitation (f-j) of ll-c/s-retinol dehydrogenase activity by monocional mouse anti-33 kDa protein antibody (oxidation reaction). The oxidation reaction was carried out qualitatively (see Materials and Methods). The position of standard retinal isomers on HPLC is shown in a and f. Substrate used was a mixture of 13-c/s and ll-c/s-retinol in b, d, e, g, h, i and j. Enzyme, partially purified by Method 2, was added to a mixture of 13-c/s and ll-c#-retinol and incubated (b). 13-c~ and ll-c/s-retinai peaks formed from the oxidation reaction were evident. However, no peak was present in a reaction mixture of 9-ct~ and ail-trans retinol (c). Also, no reaction product was detected when the enzyme was inactivated by boiling it for 3 rain at 80"C (d), or when the reaction mixture was not incubated (e). When the enzyme was preincubated with normal horse serum (g) or monoclonal anti-ROS retinol dehydrogenase antibody (i), the enzyme activity was not precipitated and the same retinoid peaks (13-c/s and ll-cL~) were evident. On the other hand, no reaction product was detected when the enzyme activity was precipitated by monoclonal anti-33-kDa protein antibody (h), or when no enzyme was added (i). As RPE retinol dehydrogenase did not react with 9-cL~ and all-trans retinol as shown in panel c, a small amount of ail-trans retinal that appeared in panels g and i was thought t o b e a non-specifically isomerized product from 13-c/s and ll-c/s-retinal.

208

converted rapidly into 11-c/s-retinal by RPE retinol dehydrogenase in the RPE. Acknowledgements Mr. Yuzo Abe performed the photography for this study and Miss Maxine A. Gere reviewed the article. This work was supported by Grant-in-Aid No. 63480390 for Scientific Research of The Ministry of Education, Science and Culture, Japan.

References

Fig. 8. Immunohistochemical demonstration of RPE retinol dehydrogenase in bovine retina. Marked immunoreactivity was present in the retinal pigment epithelium (arrow in panel b), whereas no immunoreactivity was observed in the control sections, for which we used absorbed-anti-RPE retinol dehydrogenase antibody with excess antigen (2.5 /~g, panel a). OS, outer segment; IS, inner segment; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. Bar indicates 50 mm.

extract of RPE microsomes. Polyclonal and monoclonal anti-33-kDa protein antibody reacted only with the 33 kDa molecule of the microsomal proteins (Fig. 4i). It appeared unlikely that 66 kDa protein has similar construction to 33 kDa protein or that the 33 kDa protein is degradation product of other proteins of RPE crude extract. As shown in Fig. 7, the monoclonal anti-33-kDa protein antibody also precipitated the oxidation activity of RPE retinol dehydrogenase. Thus, we confirmed that the 33 kDa protein is RPE retinol dehydrogenase and that RPE retinol dehydrogenase has oxidoreductase activity. CRALBP (cellular retinal binding protein), which reportedly serves as the substrate carrier protein for retinol dehydrogenase [25], exists in RPE and Miiller cells [15,16]. As Miiller cells may play some role in the visual cycle, it is of interest to know whether RPE retinol dehydrogenase is present in Miiller cells or not. We found that RPE retinol dehydrogenase immunoreactivity is present in RPE, but not in Miiller cells. Therefore, the role of Miiller cells in the visual cycle remains unknown. Our results, however, can explain why CRALBP carries only ll-c/s-retinal in RPE, whereas CRALBP from Miiller cells has both ll-c/sretinal and ll-c/s-retinol [15,16]. ll-c/s-retinol may be

1 Bok, D. and Heller, J. (1976) Exp. Eye Res. 22, 395-402. 2 Dowling, ,I.E. (1960) Nature 188, 114-118. 3 Bemstein, P.S., Law, W.C. and Rando, R.R. (1987) Proc. Natl. Acad. Sci. USA 84, 1849-1953. 4 Bridges, C.D. and Alvarez, R.A. (1987) Science 236, 1678-1680. 5 Andrews, ,I.S. and Futterman, S. (1964) ,I. Biol. Chem. 239, 4073-4076. 6 Saari, ,I.C. and Bredberg, D.L. (1988) J. Biol. Chem. 263, 80848090. 7 Saari, J.C. and Bredberg, D.L. (1989) J. Biol. Chem. 264, 86368640. 8 Deigner, P.S., Law, W.C., Canada, F.J. and Rando, R.R. (1989) Science 244, 968-971. 9 Lion, F., Rotmans, ,I.P., Daemen, FJ.M. and Bonting, S.L. (1975) Bioehim. Biophys. Acta 384, 283-292. 10 Zimmerman, W.F., Lion, F., Daemen, F.J.M. and Bonting, S.L. (1975) Exp. Eye Res. 21, 325-332. 11 Liou, G.I., Bridges, C.D.B., Fong, S.-L, Alvarez, R.A. and Gonzalez-Fernandez, F. (1982) Vision Res. 22, 1457-1467. 12 Lai, Y.-L., Wiggert, B., Liu, Y.-P. and Chader, G.,I. (1982) Nature 298, 848-849. 13 Ishiguro, S-I., Suzuki, Y., Tamai, M. and Mizuno, K. (1991) J. Biol. Chem. 266, 15520-15524. 14 Lin, Z.-S., Fong, S.-L. and Bridges, C.D.B. (1989) Vision Res. 29, 1699-1709. 15 Saari, J.C., Bredberg, L. and Garwin, G.G. (1982) J. Biol. Chem. 257, 13329-13333. 16 Bunt-Milam, A.H. and Saari, J.C. (1983) J. Cell Biol. 97, 703-712. 17 Tsukida, K., Kodama, A., Ito, M., Kawamoto, M. and Takahashi, K. (1977) J. Nutr. Sci. Vitaminol. 23, 263-264. 18 Bridges, C.D.B. and Alvarez, R.A. (1982) Methods Enzymol. 81, 463 -485. 19 Futterman, S. and Saslaw, L.D. (1961) J. Biol. Chem. 236, 16521657. 20 Bensadoun, A. and Weinstein, D. (1976) Anal. Biochem. 70, 241-250. 21 Berman, E.R. and Feeney, L. (1976) Invest. Ophthalmol. 15, 238-240. 22 Laemmli, U.K. (1970) Nature 227, 680-685. 23 Towbin, H., Staehelin, T. and Gordon, J. (1979) Proc. Natl. Acad. Sci. USA 76, 4350-4354. 24 Ishiguro, S.-I., Fukuda, K., Kanno, C.-I. and Mizuno, K. (1987) Cell Struct. Funct. 12, 141-155. 25 Saari, J.C. and Bredborg, L. (1982) Biochim. Biophys. Acta 716, 266-272.