Interrelationships among human aldo-keto reductase: immunochemical, kinetic and structural properties

Biochimica et Biophysica Acta 840 (1985) 334-343 Elsevier

334

BBA22070

Interrelationships among human aldo-keto reductases: immunochemicai, kinetic and structural properties Satish K. Srivastava a, Ballabh Das a, Gregory A. Hair a, Robert W. Gracy Sanjay Awasthi a, Naseem H. Ansari a and J. Mark Petrash c

b

Department of Human Biological Chemistry and Genetics, The University of Texas Medical Branch, Galveston, TX 77550, b Department of Biochemistry, Texas College of Osteopathic Medicine, Fort Worth, TX 76107, and " Department of Ophthalmology, Emory University, Atlanta, GA 30322 (U.S.A.) (Received December 18th, 1984)

Key words: Aldose reductase; Aldehyde reductase; (Human)

We have proposed earlier a three gene loci model to explain the expression of the aldo-keto reductases in human tissues. According to this model, aldose reductase is a monomer of a subunits, aldehyde reductase I is a dimer of a, fl subunits, and aldehyde reductase II is a monomer of 8 subunits. Using immunoaffinity methods, we have isolated the subunits of aldehyde reductase I (a and fl) and characterized them by immunocompetition studies. It is observed that the two subunits of aldehyde reductase I are weakly held together in the holoenzyme and can be dissociated under high ionic conditions. Aidose reductase (a subunits) was generated from human placenta and liver aldehyde reductase I by ammonium sulfate (80% saturation). The kinetic, structural and immunological properties of the generated aldose reductase are similar to the aldose reduetase obtained from the human erythrocytes and bovine lens. The main characteristic of the generated enzyme is the requirement of Li 2SO4 (0.4 M) for the expression of maximum enzyme activity, and its K m for glucose is less than 50 mM, whereas the parent enzyme, aldehyde reductase I, is completely inhibited by 0.4 M L i 2 S O 4 and its K m for glucose is more than 200 mM. The fl subunits of aldehyde reductase I did not have enzyme activity but cross-reacted with anti-aldehyde reductase I antiserum. The fl subunits hybridized with the a subunits of placenta aldehyde reductase I, and aldose reductase purified from human brain and bovine lens. The hybridized enzyme had the characteristic properties of placenta aldehyde reductase I.

Introduction The aldo-keto reductases of human tissues have been classified into two major groups: (1) aldose reductase, and (2) aldehyde reductases. The aldose reductase (alditol: NADP + 1-oxido reductase, EC 1.1.1.21) was initially thought to be present mainly in the lens, placenta and seminal vesicles. K m of this enzyme for aldo-hexoses is 20-40 fold higher than their physiological levels [1-3]. After the demonstration that aldose reductase can reduce glucose and galactose to sorbitol and galactitol,

respectively [4], and that these polyols play a significant role in the etiology of diabetic complications such as cataractogenesis [5,6], retinopathy [7] and neuropathy [8], the systematic study of aldoketo reductases in human subjects has drawn considerable attention. Our recent studies demonstrate that aldose reductase is present in the lens, brain, aorta, muscle and erythrocytes [9]. Aldehyde reductases have been shown to be present in the brain, liver, kidney and erythrocyte [9,10]. Since there is an overlap in the substrate specificities of aldehyde

0304-4165/85/$03.30 © 1985 Elsevier Science Publishers B.V. (Biomedical Devision)

335 reductases and aldose reductase, different investigators have used different nomenclatures to identify these enzymes based upon their substrate specificities [10]. The identification of these enzymes was further confused by immunological similarities between different aldo-keto reductases present in mammalian tissues [11,12]. Cromlish and Flynn [13] have demonstrated immunological similarities between pig muscle and lens aldose reductase. In addition, several investigators have observed cross-reactivity of anti-lens aldose reductase antiserum with aldo-keto reductases present in various tissues [11-14]. Based upon the chromatographic, electrophoretic, structural and immunological properties we have recently demonstrated that aldose reductase and aldehyde recutase I share a common subunit (a) and aldehyde reductase I has a unique subunit also (fl). Aldehyde reductase II is composed of ~ subunits [9]. In the present studies we have dissociated human placenta and liver aldehyde reductase I into subunits (a and r ) and show that /3 subunit is enzymatically inactive, while a subunit has aldose reductase activity. Aldehyde reductase I can be regenerated by hybridization of a and fl subunits. Material and Methods

Sources of human tissues and chemicals used in this study have been described previously [9,15,16].

Purifications of aldose and aldehyde reductases from human tissues Aldehyde reductases I and II from human placenta and liver and aldose reductase from human erythrocytes were purified to homogeneity as described previously [15,17,18]. Human and bovine lens aldose reductase was purified to homogeneity by following essentially the same procedures as described previously [17,18]. Homogeneity of each enzyme was established by the appearance of a single protein peak coinciding with the enzyme activity peak on Sephadex gel filtration and appearance of a single protein band on SDS-polyacrylamide gel electrophoresis. Aldose reductase from human muscle was partially purified by DE52 column chromatography as described earlier [9].

Preparation of antisera The antisera against the homogenous preparations of human placenta aldehyde reductases I and II, human lens aldose reductase and aldehyde reductase II, and bovine lens aldose reductase were raised in rabbits as described previously [9].

Preparation of immunoaffinity resins Homogeneous preparations of antigens and partially purified immunoglobulin, IgG fractions, were immobilized on cyanogen bromide-activated Sepharose 4B by following the procedure given in the Pharmacia bulletin and described earlier [9].

Immunochemical studies Titration of antigen with antisera (immunotitrations). For immunotitration studies, different enzyme preparations (antigens, 100 /~1 of the appropriately diluted active enzyme samples) were incubated with 100 ~1 of the various dilutions of different antisera at 4°C for about 20 h. The enzyme activities of aldose reductase and aldehyde reductases I and II were determined as described earlier [9] at zero time and after the incubation period using the 10 000 x g supernatant. Competitive binding studies. Fixed amounts of enzymatically active antigens such as aldose reductase and aldehyde reductases I and II from different tissues were titrated against different antisera. The concentrations of antisera that resulted in the precipitation of about 60% of antigens (enzyme activities) were used for immunocompetition studies in which varying amounts of enzymatically inactive antigens were added. After incubation, the respective enzyme activities were determined in the 10000 x g supernatant. The precipitation of less than 60% of enzyme activity indicates the competition between the enzymatically active protein and the enzymatically inactive protein for the binding sites on the antibody molecules.

Purification of aldehyde reductase I from human placenta by double affinity method and liver by D EA E-cellulose column chromatography The human placenta aldehyde reductase I was purified to homogeneity by DEAE-cellulose (DE52) column chromatography [15] of the enzyme preparations obtained after the second affinity step [19]. Human liver aldehyde reductase I was

336

only partially purified by DE-52 column chromatography as described earlier [17]. Aldehyde reductase I was found to be in the unadsorbed fraction.

Dissociation of aldehyde reductase I into subunits by (NH4):S04 The aldehyde reductase I preparations purified from placenta and liver as described above were divided into two parts; one part was precipitated with (NH4)2SO4, 80% saturation, and allowed to stand overnight at 4°C and the other part served as a control. The a m m o n i u m sulfate-treated fraction was centrifuged at 10000 × g for 40 min, and the resulting pellet was solubilized in minimal volume of buffer A (10 mM potassium phosphate (pH 7.0)/2-mercaptoethanol) and dialysed against the same buffer for 4 h with three changes. Subsequently, the dialyzed material was centrifuged at 10000 × g for 30 min. The control sample was dialysed and centrifuged similarly. The supernatants were passed through a DE-52 column, 0.5 × 10 cm, preequilibrated with buffer A, at a flow rate of 15 m l / h . After washing the column with 50 ml of buffer A, the proteins were eluted with a 100 ml gradient of a m m o n i u m sulfate (0-100 mM). Aldose reductase and aldehyde reductase I activities were determined in all the fractions (2.5 ml each).

Amino acid analysis Amino acid analyses were carried out in triplicate by both postcolumn reactions with ophthaldehyde and precolumn reaction followed by high-performance liquid chromatography [20]. Values are not reported for tryptophan or proline. Samples were hydrolyzed in 6 M HC1 for 24 and 72 h at l l 0 ° C in vacuo. Results

Separation and characterization of anti-a and anti-fl antisera from anti-aldehyde reductase 1 antiserum Anti-placenta aldehyde reductase I antiserum which is a mixture of anti-a and anti-/3 antiserum was subjected to aldose reductase immunoaffinity column as described previously [9]. Aldose reductase immobilized on CNBr-Sepharose 4B absorbed all the anti-a antibodies, whereas anti-/3

1 O0

75

g -£

50 ~

i 25

0

100

200

0

Anti a

100

200

Anti-/3 A n t i s e r a D i l u t i o n (fold)

Fig. 1. Immunotitration of aldose reductase and aldehyde reductase I with anti-a and anti-fl antisera. Purified human placenta aldehyde reductase I (El), human muscle aldose reductase (©) and bovine lens aldose reductase (O) were titrated against varying dilutions of the anti-a and anti-fl antisera.

antibodies were recovered in the unabsorbed fraction. The absorbed anti-a was eluted with 4.0 M potassium thiocyanate as described earlier [9]. As shown in Fig. 1, anti-a precipitated muscle and lens aldose reductase and placenta aldehyde reductase I, whereas anti-/~ antiserum precipitated only placenta aldehyde reductase I (a/~). The anti-a and anti-/~ antisera were immobilized on the CNBr-Sepharose 4B and used for the purification of aldehyde reductase I subunits as described below.

Generation of aldose reductase from human placenta and liver aldehyde reductase 1 In view of the immunological similarity between aldehyde reductase I and aldose reductase and immunological identity of one of the subunits of aldehyde reductase I, attempts were made to dissociate aldehyde reductase I and isolate the subunits. As depicted in Fig. 2, a m m o n i u m sulfate (80% saturation) almost completely dissociated the homogeneous preparation of human placenta aldehyde reductase I. The DE-52 elution profile shows that total enzyme activity is recovered as aldose reductase. The control (untreated) shows that aldehyde reductase I is not adsorbed on the DE-52 column under the experimental conditions described in Materials and Methods, whereas after a m m o n i u m sulfate precipitation (80% saturation)

337

Placenta (Affinity Pool) __0.10]

t " " •200

~

I

~ .

"-~ 0.05

ARII

~ I

ARI

,,

/"

l

/

z A

ioo

-~

d

J./" it

e,i "r Z

u,I

160 200 ml through Column 0 03

300

Properties of generated aldose reductase

ARI (NH4)2SO 4 Treated E® AR (Generated)

iA

v:~ 0.02-

/

<

-I00 I I

,'"

E I"

I

0.01"

/

//

O3

'50

1/

df f l Z

i

i

50 100 ml through Column AR I Control

150

c

0.03

-100 , E E

m

// I

0.02 ARI

~ o

s/s

<

~: 0.01

~

/

//

E -50 v0~ "1Z

,"

uJ

5'0

100

the newly generated aldose reductase enzyme is adsorbed on the column and eluted with 16 mM ammonium sulfate. In the case of the partially purified liver aldehyde reductase I, only 45% of the enzyme activity was lost, and the newly generated enzyme adsorbed on the DE-52 and eluted with 20 mM ammonium sulfate (Fig. 3). The generated enzyme (one that adsorbed on DE-52 column) showed negligible activity in the aldehyde reductase I assay system but was active in the aldose reductase assay system.

150

ml through Column Fig. 2. Generation of aldose reductase from human placenta aldehyde reductase I. Human placenta aldo-keto reductases purified by double affinity method were subjected to DE-52 column chromatography as described in the previous paper [15]. The enzymes separated into three distinct peaks (top panel). Fractions containing aldehyde reductase I were pooled and divided into two portions. One portion was brought to 80% ammonium sulfate saturation, allowed to stand for 16 h at 4°C and centrifuged at 10000× g for 30 rain. The precipitate, after reconstituting in minimal volume of buffer A, was thoroughly

Substrate specificity data given in Table I indicate that the generated aldose reductase has a wide overlapping substrate specificity. Furthermore, the generated enzyme exhibited all the properties which are common between lens aldose reductase [21] and placenta aldehyde reductase I [15]. However, unlike aldehyde reductase 1, generated aldose reductase catalysed the reduction of galactose and glucose (Table I). As shown in Table II, phenobarbital (1 mM) did not significantly inhibit the generated aldose reductase, whereas sorbinil (0.1 mM) significantly inhibited the generated enzyme. The inhibition pattern was similar to bovine lens aldose reductase [17]. Sodium chloride at 400 mM concentration completely inhibited the generated aldose reductase as well as placenta aldehyde reductase I. The major difference between the generated aldose reductase and placenta aldehyde reductase I was the effect of Li2SO4. Aldehyde reductase I is almost completely inhibited by 0.4 M Li2SO4, whereas this concentration of Li2SO4 is essential for the activity of generated aldose reductase. Furthermore, newly generated aldose reductase moved as a single protein band on the SDS/2-mercaptoethanol-polyacrylamide slab gel electrophoresis with a molecular weight around 32 500, whereas aldehyde reductase I is a dimer of 32500 and 39000 [15]. The pH optimum (6.0), the effect of p-hydroxymercury benzoate and Nethymaleimide on the generated aldose reductase dialysed against buffer A and subjected to DE-52 column chromatography (middle panel). Another portion which was not treated with ammonium sulfate served as control (bottom panel). AR, aldose reductase (zx); AR I, aldehyde reductase I (O); AR 11, aldehyde reductase 11 (O).

338 Liver ( 1 0 , 0 0 0 xg S u p e r n a t a n t )

0.15]

~"

ARII 0.10

•2OO

] D

f

I ARI

=E E

8 Wj

0.05

t

lOO ¢s "lzv

I~O

2oo

ml through

Purification of aldehyde reductase I subunits Subunits of aldehyde reductase I were purified

300

Column

ARI (NH4)2SO 4 Treated

EO.10 -

E

-IO0

8

E

AR (Generated)

i t

E

< 0.05

(:~ u~

- 50

£ zv

50

loo

150

ml through Column A R I Control

0.10

E

tt

.~

E

/ -50

"~ =

j ,"

(~ (/) r~ 1" Z

o

50

TOO

(B)

75

:~ so

150

ml through Column

Fig. 3. Generation of aldose reductase from the human liver aldehyde reductase I. The liver 1 0 0 0 0 × g supernatant was subjected to DE-52 column chromatography to fractionate aldehyde reductase I (top panel). The fractions c o n t a i n i n g aldehyde reductase I were pooled and then divided into two portions. One portion was treated with a m m o n i u m sulfate (80% saturation) and another served as control. The precipitate was dissolved, dialysed and subjected to DE-52 column chromatography as described in the legend for Fig. 2. The middle panel represents the ammonium sulfate-treated and the bottom panel represents the control preparations. AR, aldose reductase (zx); ARI, aldehyde reductase I (©); ARII, aldehyde reductase II

(e).

(A)

A

/

< 0.05

by immunoaffinity column chromatography using anti-human lens aldose reductase antiserum, anti-a or anti-/3 antiserum. Homogeneous preparations of human placenta and liver aldehyde reductase I (100 /~g) were applied to the immunoaffinity columns of anti-human lens aldose reductase antiserum. After washing with buffer A, the columns were sequentially washed with 10 mi each of buffer A containing 100 mM ammonium sulfate, 3.0 M urea containing 50 mM 2-mercaptoethanol, 0.5% SDS containing 50 mM 2-mercaptoethanol, and 4.0 M potassium thiocyanate. All the fractions (unabsorbed and the four eluates) were characterized by the immuno-

1 O0

-100

ARI

was similar to placenta aldehyde reductase 1 [15]. As shown in Fig. 4, the generated enzyme from human placenta and liver was precipitated with anti-c~ antiserum only but not with anti-/3 antiserum. This property of the generated aldose reductase is analogous to the aidose reductase of human muscle and bovine lens (Fig. 1). In contrast, the parent enzyme (aldehyde reductase I) from both tissues was precipitated with both anti-a and anti-/3 antisera (Fig. 4).

25 o

10

,6o

2o0

'

Anti-a

1~D Anti-/3

D

200

Antisera Dilution (fold)

Fig. 4. Immunochemical characterization of generated aldose reductase from human liver and placenta aldehyde reductase I. H u m a n placenta aldehyde reductase I (11), human liver aldehyde reductase I (o), generated aldose reductase from h u m a n placenta aldehyde reductase I (D), generated aldose reductase from human liver aldehyde reductase I (O), were titrated against different dilutions of anti-a and anti-fl antisera.

339 TABLE I K I N E T I C C H A R A C T E R I S T I C S OF G E N E R A T E D A L D O S E R E D U C T A S E ( A R G ) A N D H U M A N P L A C E N T A A L D E H Y D E R E D U C T A S E 1 (AR 1) For the determination of K m values for different substrates, N A D P H concentration was kept at 0.1 mM. For the determination of K m for N A D P H and N A D H , DL-glyceraldehyde was kept constant at 10 and 20 m M for generated aldose reductase and aldehyde reductase I, respectively. K m values were determined by the method of Lineweaver and Burke. Best-fit lines were plotted according to linear regression analysis. Vmax is expressed as the maximal velocity relative to that obtained with DL-glyceraldehyde for aldehyde reductase I and generated aldose reductase, n.d., not detectable. Substrate

Glyceraldehyde Glucuronate Propionaldehyde Pyridine-3-aldehyde Glucose Galactose NADPH NADH

ARG

AR I

g m (mM)

relative Vmax

/~m (mM)

relative VmaX

0.09 n.d. 0.35 0.03 47 75 0.04 1.91

100 138 257 54 36 -

0.07 n.d. 0.21 0.02 215 n.d. 0.05 2.22

100 106 606 36 -

competition method as described above. Only the urea-2-mercaptoethanol (urea eluate) and the SDS-2-mercaptoethanol (SDS eluate) eluates were found to be immunologically active. As shown in Fig. 5A, the human placenta al-

dehyde reductase I was precipitated by anti-fl antiserum to the 60% inhibition level, and subjected to immunocompetition by adding varying concentrations of urea and SDS eluates. Only the urea eluate reduced the inhibition. Addition of urea eluate (6 #g protein) to the reaction mixture resulted in the precipitation of about 20% of the

T A B L E I1 E F F E C T O F I N H I B I T O R S / A C T I V A T O R S ON G E N E R A T E D A L D O S E R E D U C T A S E (ARG) A N D H U M A N P L A C E N T A A L D E H Y D E R E D U C T A S E I (AR I)

100 -

(A)

(B)

80

Assay system for aldose reductase contained 0.4 M Li2SO 4 as one of the constituents. In the absence of Li2SO4, no aldose reductase activity was detected. Therefore, the values given for the LiESO4 effect on the generated aldose reductase were obtained in assay systems where the concentration of LiESO4 was maintained as mentioned and not as described in Materials and Methods. The effects of other inhibitors/activators were determined in the respective assay systems as described in the text. Inhibitors/activators

None Phenobarbital (1 mM) Sorbinil (0.1 m M ) T M G (1 raM) NaCl (250 raM) (400 mM) Li2SO 4 (250 raM) (400 raM)

Activity remaining (%) ARG

ARI

100 85 15 39 45 0 58 100

100 57 37 85 50 0 27 4

60 40 20

o

4

;

1'2 o

;

8

1'2 o

;

8

;2

Eluate Added (pg)

Fig. 5. Characterization of the aldehyde reductase ! subunits (.a and fl) by immuno-competition studies. (A) A fixed concentration of purified human placenta aldehyde reductase I and anti-fl antiserum which precipitated 60% of the enzyme activity was titrated against SDS eluate (a) (Q) and urea eluate (fl) (©). (B) A fixe,d concentration of purified human placenta aldehyde reductase I and anti-a antiserum which precipitated 60% of the enzyme activity was titrated against SDS eluate (a) (D) and urea eluate (fl) (©). (C) A fixed concentration of purified bovine lens aldose reductase and anti-a antiserum which precipitated 60% of the enzyme activity was titrated against SDS e]uate (a) (O) and urea eluate (fl) (z,).

340

enzyme activity as opposed to 60% precipitation without the eluate. The addition of the SDS eluate, however, did not affect the enzyme activity. This indicated the presence of fl subunits in the urea eluate. When human placenta aldehyde reductase I was precipitated by anti-a antiserum (Fig. 5B) to the 60% inhibition level, only the SDS eluate demonstrated the ability to compete with the enzyme for anti-a binding sites. This indicated the presence of a subunits in the SDS eluate. The urea eluate, however, had no effect on these immunocompetition studies. Further confirmation of the existence of the ,~ and fl subunits in the SDS and urea eluates, respectively was obtained by immunoprecipitation of bovine lens aldose reductase with anti-a antiserum and immunocompetition studies (Fig. 5C). As expected, only the SDS eluate (a subunit) competed for anti-a binding sites while the urea eluate (fl subunit) had no effect on the enzyme activity. Although, by this procedure we were able to

completely separate a and fl subunits of aldehyde reductase I, neither of the subunits had enzyme activity. Since, ammonium sulfate dissociates aldehyde reductase I into enzymatically active a subunits (aldose reductase), a second method for the separation of a and fl subunits was used. After precipitation of placenta and liver aldehyde reductase I with ammonium sulfate, as described above, the precipitate was dissolved in buffer A and divided into two portions. One portion was applied to anti-a column and another portion was applied to anti-fl column. The anti-a column absorbed all the a subunits and the fl subunits were not absorbed. Similarly anti-fl absorbed all the fl subunits and the ct subunits were recovered in the filtrate. Both the a and fl fractions were analysed by immunoprecipitation for cross contamination and found to be homogenous. The anti-fl unabsorbed fraction (a subunits) expressed aldose reductase activity, whereas anti-a unabsorbed fraction (fl subunits) had none of the aldo-keto reductase activity.

T A B L E III A M I N O A C I D C O M P O S I T I O N S OF T H E a, fl A N D 8 S U B U N I T S Values are expressed as residues per mole based on molecular weights of 32500 for the a and 8 subunits and 39000 for the fl subunit. Erythrocyte a, erythrocyte aldose reductase; generated a, aldose reductase (a) generated from placenta aldehyde reductase I (a, fl) by a m m o n i u m sulfate (Fig. 2); placenta a and fl, subunits of placenta aldehyde reductase I (a, fl) separated using immunoaffinity columns; placenta 8, placenta aldehyde reductase II (8 subunit). A m i n o acid

Erythrocyte a

Generated a

Placenta

Asx Thr Ser Glx Gly Ala Val Met lle Leu Tyr Phe His Lys Arg Cys

21.8 12.6 18.4 29.8 28.9 23.4 24.4 3.1 10.0 21.9 5.5 6.8 9.4 20.0 13.1 2.1

19.3 14.1 23.0 30.7 31.7 23.0 26.3 3.2 10.1 20.3 6.0 6.8 8.2 20.7 12.7 1.8

19.9 12.5 18.1 29.5 26.8 25.1 26.8 2.5 9.6 20.7 4.5 5.9 8.0 19.4 15.2 2.0

29.7 17.1 21.8 33.9 25.9 26.9 30.1 3.0 13.8 26.2 9.2 11.4 7.8 18.9 12.3 1.9

28.3 12.0 18.5 36.2 20.2 28.1 22.7 1.9 12.4 33.4 7.9 7.0 8.6 19.3 14.1 2.3

341

Regeneration of aldehyde reductase I by reconstitution of a and fl subunits Equal amounts of enzymatically active a and inactive fl subunits were incubated with 10 ml of buffer A and 3 M guanidinium chloride at 25°C for 1 h. The incubation mixture was diluted 9-fold with buffer A containing 0.1 mM EDTA and 25% (v/v) glycerol. This mixture was again incubated for 1 h at 25°C. Subsequently, the mixture was dialysed against buffer A (3 × 50 vol.) at 4°C for 24 h. The enzyme activity determination in the dialysed material showed a significant decrease in the activity of aldose reductase and presence of aldehyde reductase I activity. Both these enzyme activities were separated by DE-52 column chromatography as described above. On the basis of enzyme activities and immunological studies, the ratio of aldose reductase and aldehyde reductase I was found to be approx. 3 : 2. Similar results were obtained when enzymatically inactive fl subunits from placenta were hybridized with aldose reductase purified from human brain and bovine lens. In all the cases generated aldehyde reductase I did not efficiently utilize glucose as substrate ( g m > 200 mM), was inhibited by 0.4 M Li2SO4, showed a molecular weight approx. 74 000 by Sephadex G-100 chromatography and was precipiated by anti-a as well as anti-fl antiserum. Amino acid composition The amino acid compositions of the a and fl subunits of the placenta aldehyde reductase I are significantly different from each other (Table III). The amino acid compositions of the generated aldose reductase was similar to the a subunits of placenta aldehyde reductase I, and erythrocyte aldose reductase (Table III). The amino acid compositions of the aldehyde reductase II (8 subunits) was found to be significantly different from erythrocyte aldose reductase but is somewhat similar to the fl subunits of placenta aldehyde reductase I (Table III). The amino acid composition of placenta aldehyde reductase II was, however, similar to the aldehyde reductase II from liver [17] and erythrocytes [18]. Discussion

Various forms of aldo-keto reductases present in human tissues have wide overlapping substrate

specificities and some forms have immunological cross-reactivity [9,10]. This has resulted in different nomenclatures for these enzymes. Based upon the kinetic properties, effects of inhibitors and activators, structural properties and immunological characterizations, we have recently shown the presence of three forms of aldo-keto reductases in human tissues, namely aldose reductase, aldehyde reductase I and aldehyde reductase II [9]. On the basis of our investigations, we have proposed a three gene loci model to explain the expression of these enzymes in human tissues [9]. According to the model, aldose reductase is a monomer of subunits, aldehyde reductase I is a dimer of aft subunits, and aldehyde reductase II is a monomer of 8 subunits. Indeed, we have found that the amino acid compositions of the a, fl, and 8 subunits are significantly different from each other. The amino acid compositions of human erythrocyte aldose reductase, a subunits immunologically purified from placenta aldehyde reductase I, and aldose reductase generated from placenta aldehyde reductase I were similar. This would indicate that aldose reductase and aldehyde reductase I have a common subunit. The amino acid composition of fl subunits of aldehyde reductase I is significantly different from aldose reductase, indicating the genetic identity of this subunit. Aldehyde reductase II appears to be a different protein because the amino acid compositions of this enzyme, aldose reductase, and the a subunits of aldehyde reductase I are significantly different. Of all the aldo-keto reductases, aldose reductase has been implicated to play a significant role in diabetic complications such as diabetic cataractogenesis, retinopathy, neuropathy and nephropathy [5-8]. This enzyme catalyses the reduction of glucose to sorbitol which being relatively impermeable through biomembranes may accumulate in the cells and cause osmotic imbalance [4]. As a result, cells may imbibe water and lead to membrane stretch and dysfunction. This enzyme was first demonstrated in placenta and seminal vesicles be Hers [22,23]. Since then a number of investigators have studied this enzyme in human, sheep and bovine placenta [2,19,23-25]. Recently this enzyme has been purified to homogeneity from human placenta by using (NH4)2SO 4 precipitation and double affinity chromatography methods [19].

342 Without ( N H 4 ) 2 S O 4 precipitation we found mainly aldehyde reductases I and II in human placenta [15]. By following the exact procedure of K a d o r et al. [19], i.e., precipitation of human placenta 10000 x g supernatant with ( N H 4 ) 2 8 0 4 , AH-Sepharose 4B, and matrex orange gel A affinity chromatographies, we found the final preparation contained aldose reductase, aldehyde reductases I and II. These enzymes were separated upon DE-52 column chromatography and immunologically characterized. U p o n comparison of the purification procedures used by us [15] and other investigators [2,19,24], it appeared that high salt concentration may be responsible for dissociating aldehyde reductase I (aft) into subunits. The a subunits thus generated may be responsible for the aldose reductase activity. The dissociation of aldehyde reductase I into ct a n d / 3 subunits was confirmed by precipitating the homogenous preparation of human placenta aldehyde reductase I with (NH4)2SO 4 (80% saturation). The DE-52 elution profile indicated that all the placenta aldehyde reductase I activity was lost and aldose reductase was generated. A similar result was obtained with the (NH4)2SO 4 precipitation of the partially purified aldehyde reductase I from human liver. However, the conversion efficiency of the partially purified aldehyde reductase I preparation into aldose reductase was significantly lower (45%). This could be attributed to the protein-protein interaction in the high protein medium, preventing thereby the complete dissociation of the aldehyde reductase I subunits. The a and /3 subunits of aldehyde reductase I after dissociation with ( N H 4)2SO4 did not reassociate by dialysis of the salt. Partial denaturing conditions such as guanidinium-HCl were necessary for reassociation of the subunits. Subsequent dialysis to remove the hybridization mixture resulted in ascertainable aldehyde reductase I activity. The a and fl subunits of aldehyde reductase I after dissociation with (NH4)2SO 4 were also purified by immunoaffinity column chromatography. The a subunits had all the kinetic, structural and immunological properties of aldose reductase. This enzyme utilized glucose with K m < 50 mM and required LizSO 4, 0.4 M, for the expression of the m a x i m u m activity, whereas aldehyde reductase I was completely inhibited by 0.4 M LizSO 4 and did

not efficiently utilize glucose as substrate ( K m glucose> 200 mM). This would indicate that a subunits when not associated with the fl subunits express aldose reductase activity. The fl subunit, on the other hand, is enzymatically inactive but when associated with the a subunits, as in aldehyde reductase I, it modulates the kinetic and immunological properties and the susceptibility of the enzyme towards activators and inhibitors. This was confirmed by the hybridization studies using a a n d / 3 subunits of aldehyde reductase I. In view of the high specificity of aldose reductase for glucose as compared to aldehyde reductase I, association and dissociation of the a/3 subunits, giving rise to aldehyde reductase I and aldose reductase enzyme activities, respectively, may, if occurring in vivo also, have a significant physiological role, especially in diabetic complications as mentioned above.

Acknowledgement This investigation was supported in part by the PHS grant EY 01677 awarded by the National Eye Institute, D H H S .

References 1 Gabbay, K.H. (1975) Annu. Rev. Med. 26, 521-536 2 Clements, R.S. and Winegrad, A.I. (1972) Biochem. Biophys. Res. Commun. 47, 1473-1479 3 Srivastava, S.K., Ansari, N.H., Brown, J.H. and Petrash, J.M. (1982) Biochim. Biophys. Acta 717, 210-214 4 Kinoshita, J.H. (1974) Invest. Ophthalmol. 13, 713-324 5 Gabbay, K.H. (1973) New Engl. J. Med. 288, 831-836 6 Pirie, A. and Van Heyningen, R. (1964) Exp. Eye Res. 3, 124-131 7 Buzney,S.M., Frank, R.N., Varma, S.D., Tanishima, T. and Gabbay, K.H. (1977) Invest. Ophthalmol. 16, 392-396 8 Gabbay, K.H., Merola, L.O. and Field, R.A. (1966) Science 151,209-210 9 Srivastava, S.K., Ansari, N.H., Hair, G.A. and Das, B. (1984) Biochim. Biophys. Acta. 800, 220-227 10 Flynn, T.G. (1982) Biochem. Pharmacol. 31, 2705-2712 11 Dons, R.F. and Doughty, C.C. (1976) Biochim. Biophys. Acta 452, 1-12 12 Cromlish, J.A. and Flynn, T.G. (1983) J. Biol. Chem. 258, 3416-3424 13 Cromlish, J.A. and Flynn, T.G. (1983) J. Biol. Chem. 258. 3583-3586 14 Herrmann, R.K., Kador, P.F. and Kinoshita, J.H. (1983) Exp. Eye Res. 37, 467-474 15 Das, B. and Srivastava,S.K. (1985) Biochim. Biophys. Acta 840, 324-333

343 16 Srivastava, S.K., Goldblum, R.M., Hair, G.A. and Das, B. (1985) Lens Res. in the press 17 Petrash, J.M. and Srivastava, S.K. (1982) Biochim. Biophys. Acta 717, 105-114 18 Das, B. and Srivastava, S.K. (1985) Arch. Biochem. Biophys., in the press. 19 Kador, P.F., Carper, D. and Kinoshita, J.H. (1981) Anal. Biochem. 114, 53-58 20 Oray, B., Lu, H.S. and Gracy, R.W. (1983) J. Chromatog. 270, 253-266

21 Hayman, S. and Kinoshita, J.H. (1964) J. Biol. Chem. 240, 877-882 22 Hers, H.G. (1956) Biochim. Biophys. Acta 22, 202-203 23 Hers, H.G. (1960) Biochim. Biophys. Acta 37, 120-138 24 Hastein, T. and Velle, W. (1969) Biochim. Biophys. Acta 178, 1-10 25 Velle, W. and Engel, L.L. (1964) Endocrinology 74, 429-439