Aldose reductase and aldehyde reductase were purified to homogeneity from multiple samples of human kidney cortex and medulla. A single form of aldose reductase is expressed in kidney that is kinetically and immunochemically indistinguishable from aldose reductase expressed in other human tissues. The results support the conclusion that there is a single human aldose reductase, and that aldose reductase is expressed in a reduced form, characterized by high sensitivity to aldose reductase inhibitors and ability to catalyze the reduction of glucose. Aldose reductase is easily oxidized to a form that is insensitive to aldose reductase inhibitors and unable to catalyze the reduction of glucose. This form does not appear to exist in vivo, even in kidney from diabetics. There is wide variation in the level of expression of aldose reductase in kidney, especially in cortex. The immunochemically separate but similar aldehyde reductase is also expressed in kidney as a single enzyme indistinguishable from aldehyde reductase from other human tissues. Aldehyde reductase levels exceed those of aldose reductase, both in cortex and medulla.
Introduction
Aldose reductase (alditol:NADP+oxidoreductase, A L R 2 , EC 1.1.1.21) and a ldehyde r educ ta se (alcohol : N A D P ÷ oxidoreductase, ALR1, EC 1.1.1.2) are members of a monomeric NADPH-dependen t fam- ily of a ldo-ke to reductases that catalyze the reduction of a wide variety of aldehydes and ketones [1-4]. Aldose reductase (ALR2) has been implicated in the etiology of diabetic complications due to its ability to reduce glucose to sorbitol under conditions of hyper- glycemia. High glucose levels may lead to the accumu- lation of sorbitol, especially in tissues with insulin-inde- pendent uptake of glucose, resulting in hyperosmotic stress to the tissue, thereby leading to tissue damage [5-9]. Sorbitol may also interfere with the uptake and metabolism of myo-inositol [10,11]. Recent studies sug- gest that under physiological conditions ALR2 also participates in osmoregulation in renal inner medullary cells by providing sorbitol as an osmolyte [12,13].
There is variability in tissue distribution of ALR2 [14,15]. In addition, in humans the level of expression of ALR2 in a single tissue varies widely [16,17]. The wide variability of ALR2 levels may be important in determining which diabetics develop complications. ALR2 has been shown to exist in two interconvertible forms, a reduced form (ALR2 R) and an oxidized form (ALR2o) . Only ALR2 R catalyzes the reduction of glu- cose [17,18]. This raises the question whether some diabetics may be protected from diabetic complications because their ALR2 is oxidized posttranslationally to ALR2 o, the form that does not catalyze the reduction of glucose. Most studies of human ALR2 have used tissues that are not normally associated with diabetic complications. In the present study, we have evaluated the distribution of ALR2 and of aldehyde reductase (ALR1) in a tissue associated with complications, namely, human kidney, and we have addressed the question of the potential significance in vivo of the two different forms of ALR2.
Materials and Methods
Materials Kidney tissue was obtained from autopsy samples
through the National Disease Research Interchange,
Philadelphia, PA. Tissue was acquired within 12 h post mortem. Tissue samples were stored at -70°C.
NADPH, D,L-glyceraldehyde, D-glucose, p-nitro- benzaldehyde, D-glucuronic acid, pepstatin A, chymo- statin, bestatin, antipain and leupeptin were obtained from Sigma. Red Sepharose CL-6B, PD-10 desalting columns and chromatofocusing resin PBE 94 were pur- chased from Pharmacia LKB Biotechnology. A Bio-Gel hydroxylapatite HPLC column was obtained from Bio- Rad. YM-10 pressure filtration membranes were from Amicon.
Methods Enzyme purification. Both ALR2 and ALR1 from
human kidney were purified as reported previously for the purification of ALR2 and ALR1 from human pla- centa and muscle [16-18]. The kidney samples were dissected in order to separate cortex and medulla. Approx. 40 g samples of either cortex or medulla were homogenized in 200 ml of cold phosphate-buffered saline containing 1 /zg/ml of antipain, chymostatin, bestatin, leupeptin and pepstatin A. The homogenate was centrifuged for 30 min at 100000 x g. ALR2 and ALR1 were rapidly extracted by placing the 100 000 × g supernatant onto a Red Sepharose column; ALR2 and ALR1 were removed with NADP +. Following concen- tration, ALR2 and ALR1 were separated by chromato- focusing on Pharmacia PBE 94. Final purification of ALR2 and ALR1 to homogeneity was accomplished by chromatography on a Bio-Gel hydroxylapatite HPLC column. Protein concentrations were measured by the dye binding procedure of Bradford [19]. Isolation of ALR2 and ALR1 was completed in two days.
Enzyme assays. ALR2 and ALR1 activity was mea- sured routinely in 1 ml volumes of 0.1 M sodium phosphate (pH 7), containing 10 mM glyceraldehyde and 0.1 mM NADPH. Reactions were monitored using a Perkin-Elmer Lambda 6 spectrophotometer, at 340 nm and 25°C. Kinetic studies of substrate and inhibitor binding properties of both enzymes were carried out in
261
the same buffer. K m and kca t values were obtained from nonlinear regression analysis of the initial rate data using the Enzfitter Data Analysis program (Else- vier-Biosoft). Binding constants for aldose reductase inhibitors (ARI) were determined from Dixon plots of the initial rate data.
Immunochemistry. Rabbit antisera to human- placenta ALR2 and human-liver ALR1 [20] were used to compare human kidney ALR2 and ALR1 (both cortex and medulla) with ALR2 and ALR1 from other human tissues, by an ELISA procedure previously re- ported [16,20].
Results
Purification of ALR2 and ALR1 from human kidney ALR2 and ALR1 were purified to homogeneity
from multiple human kidney samples (n = 5). Cortex was separated from medulla by careful dissection, and ALR2 and ALR1 were purified from each tissue sec- tion. Table I shows a representative purification scheme for the isolation of ALR2 and ALR1 from kidney cortex. The chromatofocusing profiles for ALR2 and ALR1 from cortex are shown in Fig. 1. The smaller peak represents the ALR2 fraction which focused at an apparent isoelectric point of pH 5.9. The predominant peak represents ALR1 which focused at pH 5.5. Each peak of enzyme activity was pooled individually and purified further by hydroxylapatite HPLC (Figs. 2 and 3). Chromatography on hydroxylapatite HPLC yielded a single peak of ALR2 activity (Fig. 2) and two peaks of ALR1 activity (Fig. 3). However, these two apparent forms of ALR1 were indistinguishable by size, immunochemical properties and kinetic analysis (see below). In addition, isolated peaks when rechromato- graphed did not always show the same retention times. The number of peaks and their sizes depended upon the amount of NADP+present in the buffer. We con- clude that there is only one form of ALR1 present in kidney cortex. The purification scheme described in
TABLE I
Purification of aldose and aldehyde reductases from kidney cortex
Step Volume Protein (ml) (mg/ml)
Units Specific (/~mol/min) activity
(/xmol/min per mg)
Yield
Supernatant 108 5.3 Red Sepharose 300 0.2 Chromatofocusing
ALR2 4.8 0.5 ALR1 8.8 0.5
Hydroxylapatite ALR2 7.7 0.03 ALR1 28 0.02
13.2 0.02 100 12.6 0.2 96
0.8 0.3 6 7.3 1.7 56
0.4 1.8 3 5.3 10.0 41
262
10
• 9
v 8 m
E 7
_c 6 E
5 a
o 4 ..¢::] ~-- 3
~ 2
0
0
A
o°oo
10 20 30 4-0 50
7 A
6 v 5 -1-
0 .
o
1.2 E
0.8 E
z 0.4 ~
0.0 ~ 6 0 70 80 n
FRACTION NUMBER Fig. 1. Separation of kidney cortex aldose reductase from aldehyde reductase by chromatofocusing on Pharmacia PBE 94. The material from the Red Sepharose step (Table I) was concentrated and then desalted on a Pharmacia PD 10 desalting column before being added to the chromatofocusing column, as described previously for the purification of aldehyde and aldose reductases from other human
tissues [16,17].
Table I and Figs. 1 -3 afforded 44% yield of total reductase activity, 41% of which was ALR1 and 3% was ALR2.
The results for the isolation of A L R 2 and ALR1 from kidney medul la are slightly different from that of cortex, the major difference residing in the level of ALR2. Fig. 4 shows a representa t ive chromatofocusing profile for A L R 2 and ALR1 from medulla . Fu r the r
purif icat ion of each peak of activity by hydroxylapati te H P L C provided a single peak of A L R 2 (Fig. 5), and, similar to the cortex, two peaks of ALR1 activity (Fig.
1 . 2 5 @
1 .00 - E
.E 0 . 7 5 - E
E3 o 0 . 5 0 -
>" 0 . 2 5
< 0 . 0 0 , : : : A 0
s
s
i
s
, s
, s
,J
_ _ _ 7 _ . . j
10 20 30
F r a c t i o n N u m b e r
0 .3 I i v
v 0.2
"10 8
0.1 ~
O3 o
0 .0 ~- 13..
Fig. 3. Purification of kidney cortex aldehyde reductase on an hy- droxylapatite HPLC column. Conditions were as listed in Fig. 2.
3.0
i 2.5
2.0
~.~ 1.5
"~ 1.0
.~ 0.5
5 < 0.0
"&&z~,~A.A 7 A 6 "-, 5 ~
1.2 o
0.8
0.4 --- Z
0.0 ~
10 20 30 40 50 60 70 80 n
FRACTION NUMBER Fig. 4. Separation of kidney medulla aldose reductase from aldehyde
reductase by chromatofocusing. Conditions were as listed in Fig. 1.
• 0 . 3 I
v i . .'" ~ 0.4
c 0.15 -'" 0.2 ~ - ."
"" m 0.3 ~-~ , ' " 0 .2 .~
s S o.1o ,'"'"'" ." 0.1 ~ ~ 0 .2 ~.
J s # ~ s I S S t ~ 0.05 =o ~ 0 . 1
"'~ ~ ~ 0.1 < o . o 0 . . . . . . . . o .~
0 1 0 20 30 < 0 .0 0 .0 n
F r a c t i o n N u m b e r 0 10 20 30 Fig. 2. Purification of kidney cortex aldose reductase on an hydroxyl- apatite HPLC column. Material from the chromatofocusing step F r a c t i o n N u m b e r (Table I) was added to an hydroxylapatite HPLC column. Conditions Fig. 5. Purification of kidney medulla aldose reductase by hydroxyl-
were as described previously [16,17]. apatite HPLC; conditions as listed in Fig. 2.
v
0.20
E
= 0.15
Q o 0.10
0.05
< 0 .00 0
S
J S
S S
S
J
S
Ii • , ~ . . e . 9 e e e . . . . . , v v . . . . .
10 20 30
Frac t ion Number
0.3 I
I
v
"5 v
0.2 -~
0.1 ~
n
f f l
o 0.0 .c
Fig. 6. Purification of kidney medulla aldehyde reductase by hydroxylapatite HPLC; conditions as listed in Fig. 2.
6). Table II summarizes the scheme for purification of ALR2 and ALR1 from kidney medulla. The overall yield of total reductase activity was 27%, 15% of which was ALR1 and 12% was ALR2.
Distribution of ALR2 and ALR1 in human kidney The distribution of ALR2 and ALR1 activities be-
tween cortex and medulla for the five kidney samples is summarized in Table III. The specific activity values listed represent the amount of enzyme in the 100 000 x g supernatant fraction. These values were obtained by calculating the percentage of each enzyme activity after purification and multiplying the total specific activity of the 100 000 x g supernatant by the fraction of ALR2 and ALR1. The level of ALR2 activity present in the cortex varies greatly, from undetectable to 4 nmol/min per mg. The latter value represents a level of ALR2 activity comparable to that observed in other human tissues [16,17]. The amount of ALR2 activity in kidney medulla does not show the variability seen in cortex. ALR1 accounts for the majority of reductase activity in both cortex and medulla.
TABLE III
Distribution of aldose and aldehyde reductases in human kidney
263
Sample Specific activity (cortex) Specific activity (medulla) (nmol/min per mg) (nmol/min per mg)
Immunochemical comparison of ALR2 and ALR1 Both ALR2 and ALR1 from cortex and medulla
were homogenous on SDS-PAGE with silver staining and showed a molecular mass of 36.7 kDa for ALR2 and 38.9 kDa for ALR1. ALR1 from cortex and medulla (both peaks of activity, Figs. 3 and 6) was compared with ALR1 from liver by an ELISA analysis in which antisera to human-liver ALR1 was titered against the various purified reductases (data not shown). Both peaks of ALR1 activity from the hydroxylapatite HPLC steps in the purification of ALR1 from kidney cortex and medulla are immunochemically equivalent to liver ALR1 and do not crossreact with kidney ALR2. Like- wise, ALR2 from cortex and medulla are immuno- chemically equivalent to each other as well as to ALR2 from human placenta and muscle. Kidney ALR2 does not crossreact with ALR1.
Kinetic analysis of kidney ALR2 and ALR1 The substrate properties of both ALR2 and ALR1
were analyzed with the same set of substrates previ- ously used to analyze ALR2 from human placenta and human muscle and to analyze ALR1 from human pla- centa [16,17]. The kinetic properties of kidney ALR2 are summarized in Table IV. ALR2 from cortex and medulla were analyzed separately with each of the substrates and were found to be indistinguishable. The K m values listed in Table IV are similar to values
TABLE II
Purification of aldose and aldehyde reductases from kidney medulla
Step Volume Protein (ml) (mg/ml)
Units Specific (/zmol/min) activity
(/zmol/min per mg)
Yield
Supernatant 146 3.5 Red Sepharose 196 0.2 Chromatofocusing
ALR2 6.8 0.2 ALR1 9.0 0.1
Hydroxylapatite ALR2 10.8 0.02 ALR1 36 0.003
6.0 0.01 100 5.7 0.2 95
1.4 1.3 24 1.7 1.4 29
0.7 2.6 12 0.9 8.2 15
264
TABLE IV
Substrate specificity of kidney aMose reductase
Substrate kca t Km kcat/Km (min -1 ) (mM) (M- lmin - l )
previously reported for human placenta and muscle ALR2 [16,17]. Based on the kinetic data, it appears that ALR2 isolated from both kidney cortex and medulla is the same as that isolated from placenta and muscle. More importantly, the kinetic constants in Table IV correspond to the values reported previously for the reduced form of human ALR2, ALR2 R [17]. There was no evidence for the presence of the oxidized form, ALR2 o. Thiols, such as DTI', were not present during the purification of ALR2 and ALR1. It was shown previously that DTT will reduce ALR2o to ALR2 R [17]. Therefore, the lack of evidence for the presence of ALR2 o in the five samples of kidney used in this study argues against the presence of ALR2 o in vivo.
The kinetic constants for the same set of substrates were determined using ALR1; results are summarized Table V. The two apparent forms of ALR1 that were isolated during purification on hydroxylapatite HPLC were analyzed separately with the various substrates. From kinetic analysis it was determined that there is no difference between the two forms. Also, there is no kinetic difference between ALR1 isolated from kidney cortex and ALR1 isolated from medulla. The values obtained for kidney ALR1 (Table V) are similar to those reported for placenta ALR1 [16]. Another simi- larity that exists between the kidney form of the en- zyme and the placenta form is the fact that both enzymes are inhibited at high substrate concentrations. Substrate inhibition is a general characteristic observed with ALR1 but not ALR2. A repesentative example of substrate inhibition of kidney ALR1 is shown in Fig. 7 where p-nitrobenzaldehyde was used as substrate. The kinetic parameters listed in Table V were determined by nonlinear regression analysis of the initial rate data
TABLE V
Substrate specificity of kidney aldehyde reductase
1/[p-Nitr0benzaldehyde] (mM) Fig. 7. Kinetic profile of kidney aldehyde reductase with p-nitro- benzaldehyde as substrate shows the substrate inhibition character-
istic of human aldehyde reductase.
at lower substrate concentrations where substrate inhi- bition was not significant.
Inhibition of kidney ALR2 by aldose reductase inhibitors Sorbinil, statil and tolrestat, compounds that have
been or are being tested clinically, were evaluated as inhibitors of kidney ALR2 and ALR1. Results are summarized in Table 6 for ALR2. There was no differ- ence observed between cortex and medulla forms of ALR2. Tolrestat shows the tightest binding (K~ = 0.034 jxM). Kidney ALR2 shows the same sensitivity to these inhibitors that was previously reported in studies of
TABLE VI
Inhibition of kidney aldose and aldehyde reductases by aldose reductase inhibitors
Enzyme K i (p.M)
sorbinil tolrestat statil
ALR2 0.18 0.034 0.053 ALR1 3.5 2.0 4.3
human placenta and muscle AR [16,17], where the reduced form ALR2 R was used.
Constants for the inhibition of kidney ALR1 are listed in Table 6. Kidney ALR1 is not as sensitive as ALR2 to inhibition by these compounds. The values in Table 6 are comparable to those previously reported for ALR1 from human placenta [16]. All of the kinetic and immunochemical data are consistent with the con- clusion that a single form of ALR2 is expressed in human kidney that is identical to ALR2 R described previously in other human tissue. We also conclude that a single form of ALR1 is expressed in kidney that is identical to ALR1 from other human tissues.
Expression of ALR2 in kidney from diabetics ALR2 R is easily oxidized to ALR2 o simply by dialy-
sis in the absence of NADP+or thiols [17]. Diabetics show evidence of oxidative stress [21-25]. Therefore, the possibility that some diabetics might express ALR2 o rather than ALR2 R was examined by isolation of ALR2 from the cortex and medulla of kidney tissue obtained from diabetics, n = 4. For these studies, small samples of cortex or medulla were homogenized, and the 100000 × g supernatant fraction was chromato- graphed directly on an hydroxylapatite HPLC column to separate ALR2 from ALR1. The kinetic properties of ALR2 were evaluated using the partially purified sample by examining ALR2 for sensitivity to sorbinil. In all cases, ALR2 from diabetic tissue showed high sensitivity to sorbinil, leading to the conclusion that even in diabetics ALR2 exists mainly as ALR2 R.
Discussion
Numerous kinetic studies of human ALR2 suggest that there are multiple isoenzyme forms with different kinetic properties. For example, reported dissociation constants for the inhibition of human ALR2 by sorbinil are 0.7, 10 and 110 /zM [26-28]. These three studies utilized ALR2 from placenta, erythrocyte and muscle, respectively, suggesting the possible existence of tissue specific isoenzymes. The existence of a multigene fam- ily of ALR2 was suggested recently based upon the complex pattern observed in Southern blots of human DNA with a placenta cDNA probe for the ALR2 gene [29]. In contrast with these observations, the reported cDNA sequences for ALR2 from several human tissues are nearly identical [29-32] with no evidence for tissue specific isoenzymes.
Our previous studies of ALR2 isolated from multi- ple samples of human placenta, skeletal muscle and cardiac muscle suggested that these human tissues express a single form of ALR2 that is characterized by high sensitivity to aldose reductase inhibitors and abil- ity to catalyze the reduction of glucose [16-18]. How- ever, ALR2 is easily oxidized during purification and
265
the oxidized form has markedly different kinetic prop- erties than the reduced form. This may explain some of the variation in kinetic constants that have been re- ported [26-28]. In addition, other recent studies have reported that there are several pseudo genes of ALR2, which may explain the complex patterns observed by Southern analysis [33]. Collectively, these studies sup- port the conclusion that there is a single gene coding for ALR2 in human tissues. However, these studies did not address the question whether the oxidized form of ALR2 may be formed in some tissue, such as tissue from diabetics who often are under oxidative stress, as a posttranslational event. The oxidized form of ALR2, ALR2 o, does not utilize glucose as substrate, which raises the question whether some diabetics might be protected from developing complications because they modify their ALR2 through oxidation [17]. These pre- vious studies [16-18, 26-33] also did not include tis- sues that are associated with diabetic complications.
In the present study, we examined the distribution and the kinetic properties of ALR2 and of the im- munochemically distinct but similar enzyme ALR1 in human kidney cortex and medulla. All of the results are consistent with the conclusion that single forms of ALR2 and ALR1 are expressed in human kidney, and that these forms are the same as those expressed in other human tissues. We conclude that there is likely a single ALR2 gene and a single ALR1 gene. Further- more, the studies that utilized samples of kidney tissue from diabetics lead to the conclusion that ALR2R, the form of ALR2 that is able to catalyze the reduction of glucose, is the only significant form expressed in vivo, even in diabetics. This conclusion is tentative since only a small number of tissue samples were included.
The distribution of ALR2 in kidney shows consider- able variation in cortex and medulla (Table III). In previous studies of ALR2 from human placenta and muscle, we observed a wide variation in the level of expression [16,17]. This also is observed with kidney cortex where ALR2 levels ranged from undetectable to 4 nmol/min per mg. By comparison, ALR2 levels in muscle range from 3 to 14 nmol/min per mg [17]. The range of ALR2 in kidney medulla is smaller than in cortex (Table III). In all samples of kidney, ALR1 levels exceed ALR2 levels. The wide variation of ALR2 in cortex is particularly interesting. Diabetic kidney disease involves the development of structural damage to the glomerulus. In view of the variation in ALR2 levels in kidney cortex and other human tissues, the question arises whether diabetic complications are re- lated to the level of expression of ALR2. This question is currently being addressed.
Acknowledgement
This work was supported by NIH grant DK43238.
266
References
1 Wermuth, B. (1985) in Enzymology of Carbonyl Metabolism, 2: Aldehyde Dehydrogenase, Aldo-keto Reductase, and Alcohol Dehydrogenase (Flynn, T.G. and Weiner, H., eds.), pp. 209-230, Alan R. Liss, New York.
2 Morjana, N.A. and Flynn, T.G. (1989) J. Biol. Chem. 264, 2906- 2911.
3 Wermuth, B. and von Wartburg, J.-P. (1982) Methods Enzymol. 89, 181-186.
4 Von Wartburg, J.-P. and Wermuth, B. (1982) Methods Enzymol. 89, 506-513.
5 Kinoshita, J.H. (1974) Invest. Ophthalmol. 13, 713-724. 6 Gabbay, K.H. (1975) Am. Rev. Med. 26, 521-536. 7 Kador, P.F. (1988) Med. Res. Rev. 8, 325-352. 8 Kador, P.F. and Kinoshita, J.H. (1985) Am. J. Med. 79, 8-12. 9 Kinoshita, J.H. and Nishimura, C. (1988) Diabetes Metab. Rev. 4,
323-337. 10 Finegold, D., Lattimer, S.A., Nolle, S., Bernstein, M. and Greene,
D.A. (1983) Diabetes 32, 988-992. 11 Greene, D.A., Lattimer, S.A. and Sima, A.A.F. (1987) N. Engl. J.
