Involvement of cysteine residues in catalysis and inhibition of human aldose reductase. Site-directed mutagenesis of Cys-80, -298, and -303.

In order to study the potential role of cysteinyl residues in catalysis and inhibition of human aldose reductase, mutants containing cysteine to serine substitution at positions 80 (ALR2:C80S), 298 (ALR2:C298S), and 303 (ALR2:C303S) were constructed. Mutation of Cys298 resulted in the most profound changes, as ALR2:C298S displayed 4- to 5-fold elevation in K'm(NADPH), K'm(DL-glyceraldehyde), and kcat(DL-glyceraldehyde) relative to wild type aldose reductase as well as a 10-fold higher Ki for the aldose reductase inhibitor sorbinil. Wild type and mutant reductases were equally sensitive to tolrestat, a structurally different reductase inhibitor. Carboxymethylation of the wild type enzyme or the C80S and C303S mutants led to a modest decrease in kcat as well as an increase in K'm(DL-glyceraldehyde) and Ki(sorbinil). These parameters were not significantly changed when ALR2:C298S was subjected to carboxymethylation. Lithium sulfate caused activation of ALR2:WT, C80S, and C303S but did not significantly affect the activity of ALR2:C298S. The differential sensitivity of wild type and mutant reductases to inhibition by sorbinil and tolrestat, before and after carboxymethylation, indicates that these inhibitors bind at different sites. These results suggest that Cys-298 is present near the active site and constitutes a regulatory group which controls the catalytic activity and inhibitor sensitivity of the enzyme.

and other aldo-sugars to their corresponding sugar alcohols or polyols (1,2). Sorbitol dehydrogenase completes the pathway by catalyzing the NAD-dependent oxidation of sorbitol to fructose. Although under hyperglycemia the poly01 pathway activity may account for more than 30% of glucose utilized (3), the physiological relevance of this pathway remains unclear. Under hyperglycemic conditions, activation of the pol-yo1 pathway has been suggested to lead to extensive tissue pathogenesis particularly in tissues independent of insulin for glucose uptake (4, 5 ) . The most convincing evidence linking aldose reductase and pathogenesis comes from the study of aldose reductase inhibitors which have been shown to prevent or significantly delay the onset and development of diabetic complications such as cataract, retinopathy, and neuropathy (5)(6)(7). However, most aldose reductase inhibitors are nonspecific and inhibit structurally related enzymes (8, 9) which precludes their long term clinical use. Design of effective and specific aldose reductase inhibitors will depend on an understanding of the mechanism of catalysis of the enzyme and synthesis of mechanism-based inhibitors or transition state analogs.
Little is known about the mechanism of catalysis by aldose reductase or the identity of amino acid residues involved in substrate binding and/or catalysis. One of the significant properties of the enzyme is its sensitivity to oxidation (10)(11)(12). Oxidative modification of the enzyme has been shown to result in its conversion to an "activated" form (12)(13)(14)(15)(16). Oxidation of the enzyme has also been shown to result in generation of multiple isoforms, presumably due to intramolecular disulfide formation, which display marked changes in kinetic behavior and altered susceptibility to inhibition by hydantoin derivatives such as sorbinil (15, 16).
Human placental aldose reductase contains 7 cysteine residues (17, 18), 3 of which are accessible to solvent and react readily with 5,5'-dithiobis(2-nitrobenzoic acid) (18). It has been proposed that oxidation-induced activation of the enzyme and alterations in the kinetic and inhibitor binding properties of the enzyme may be due to modification of cysteine residues (14)(15)(16)19). Although, the specific role of the 3 exposed cysteine residues in enzyme catalysis and inhibitor binding has not been identified, chemical modification studies suggest that Cys-298 may be responsible for oxidationinduced activation of the enzyme (19).
In this study we have utilized an efficient bacterial plasmid for overexpression of human aldose reductase in Escherichia coli. The recombinant and native enzymes appear to have similar structural and kinetic properties. Kinetic characterization of mutant enzyme forms in which Cys-80, -298, or 24833  TABLE I Oligonucleotides used for site-directed mutagenesis C80S 5"GGTACGTGCTCCACAGCTTG-3' C298S 5"CAACAAGGCAGAGACCCTCC-3' C303S 5"TGGGAGGTAGAGCTCAACAA-3' -303 was substituted with serine suggests that none of these residues is essential for catalytic activity of the enzyme. However, replacement of Cys-298 converts the enzyme from a low V,.,/low K, ("unactivated") form to a high V,,/high K,,, (activated) form which displays lowered sensitivity to sorbinil but not to tolrestat. On the basis of these results it is suggested that Cys-298 is present near the active site and regulates the conversion of the enzyme from the unactivated to activated state.

MATERIALS AND METHODS
Isolation and Sequencing of Aldose Reductase cDNA Clones-Complementary DNA (cDNA) clones encoding aldose reductase were isolated from a human placenta library constructed in XZAPII vector (Stratagene, LaJolla, CA). Bacteriophage plaques were screened by hybridization with a bovine aldose reductase cDNA (20) radiolabeled by the random primer method (21). pBluescript I1 SK(-) plasmid containing the cDNA insert from each confirmed positive bacteriophage isolate was recovered by in vivo excision using R408 interference-resistant helper phage as described by the vendor (Stratagene). The sequence of both strands of selected cDNA clones was determined by the dideoxy chain termination method (22) using synthetic oligodeoxyribonucleotide primers designed to anneal at approximately 250-base pair intervals.
Site-directed Mutagenesis-Oligonucleotide-directed mutagenesis was used to construct cDNA sequences encoding cysteine to serine substitutions at positions 80, 298, and 303. Oligonucleotide primers used to construct these mutants are listed in Table I. Primers were synthesized on an Applied Biosystems, Inc. Model 380B and were purified by reverse phase chromatography over a Sep-Pak CIS cartridge (Waters, Milford, MA) using conditions recommended by the manufacturer. Site-directed mutagenesis of aldose reductase sequences carried in a pBLUESCRIPT phagemid vector was carried out as described by Kunkel et al. (23). Clones containing the desired mutation were identified by nucleotide sequence analysis across the mutation site.
Cloning of Aldose Reductase cDNA into Expression Vector-Aldose reductase sequences were excised from pBLUESCRIPT plasmids containing wild type or mutant cDNAs by complete digestion with Hind111 followed by partial digestion with NcoI. A resulting 1354base pair NcoI-Hind111 fragment was isolated from agarose gel slices and was ligated with vector sequences from plasmid pMON5842, a derivative of the vector system described previously (24). The resulting expression plasmids contained the aldose reductase coding sequence fused precisely downstream of the gl0-L ribosome-binding site. Transcription initiates from the tac promoter (25) and terminates at a head-to-head double terminator derived from phage P22 ant gene (26).' Translation is directed by the highly efficient gl0-L ribosomebinding site (27), and a spectinomycin resistance gene provides a selectable marker (28). The nucleotide sequence of both strands of the wild type aldose reductase cDNA contained within the expression plasmid was determined completely. Mutated ALR2 sequences were completely sequenced after cloning into expression plasmids to ensure that no unintended mutations had been introduced.
Expression and Detection of Recombinant Aldose Reductase-Plasmids pMON5997 (wild type human aldose reductase expression construct) and pMON5842 (bacterial chloramphenicol acetyltransferase expression construct) were introduced into E. coli strain JMlOl (29). For small scale expression studies, bacteria were grown overnight at 37 "C in L-broth (30) containing 50 pg/ml spectinomycin. Cultures were then diluted to 30 Klett units (Klett Summerson Colorimeter, 520-580 nm, green filter) into M9 medium (30) supplemented with 1% casamino acids, 5 pg/ml thiamine, and trace metals. The composition of the trace metals mixture was as follows: FeC13.6H20, 5.4 mg/liter; ZnSO,. 7H20, 0.2 mg/liter; CoClZ. 6H20, 0.35 mg/liter; Na2Mo04. 2H20, 0.35 mg/liter; CuSO4 .5H20, 0.4 mg/liter; H3B03, ' S. Rangwala, unpublished data. 0.1 mg/liter; MnSO4.H2O, 0.25 mg/liter. When the cultures had grown to a density of approximately 150 Klett units, IPTG was added to a final concentration of 1 mM and cultures were grown for an additional 2 h. Aliquots (1 ml) of cells for electrophoretic analysis were harvested by centrifugation (7,840 X g, 1 min, 4 "C) immediately before and 2 h following IPTG induction and were resuspended in SDS-PAGE sample buffer. Also, 10-ml aliquots for fractionation by osmotic shock (31) were harvested 2 h post-induction by centrifugation for 5 min at 1,940 X g at 4 "C. For osmotic shock fractionation, harvested cells were suspended in 2 ml of sucrose buffer comprised of 20% sucrose, 30 mM Tris-HC1, pH 7.5, and 1 mM EDTA. After incubation for 15 min at 23 "C, the cells were recovered by centrifugation for 10 min at 1,940 X g at 4 "C. The supernatant (designated as sucrose wash) was reserved, and the pellet was resuspended in 1 ml of ice-cold deionized water and incubated for 10 min on ice. After centrifugation at 12,100 X g for 5 min at 4 "C, the supernatant (water wash) and pellet (resuspended in 1 ml of ice-cold deionized water) were analyzed separately. Each fraction was supplemented to a final concentration of 5 mM Tris-HC1, pH 7.5, and 1.0 mM DTT. All fractions were assayed immediately for aldose reductase activity and total soluble protein, and were subsequently stored at -20 "C.
Aliquots of total E. coli lysates, sucrose wash, water wash, and cell pellets were diluted into SDS-PAGE sample buffer and subjected to electrophoresis through 12% polyacrylamide gels containing SDS (32). For Western blot analysis, electrophoretically separated proteins were visualized by staining with Coomassie G-250 and were then transferred to activated diazobenzyloxymethyl-cellulose paper (Schleicher and Schuell) as described previously (33). Membranes were incubated with antibodies to bovine lens aldose reductase (34) which had been preadsorbed with an extract of E. coli JM101. Antigen-antibody complexes were detected using lZ5I-labeled protein A and autoradiography.
High Cell Density Fermentation-Ten-liter fermentations were run in a 15-liter B. Braun Biostat E fermenter using M9 medium supplemented with 2% casamino acids, 2.5 pg/ml thiamine, and trace metals (the concentration of trace metals was twice that was used in small scale inductions with the exception of FeC13.6H20, which was 7 X concentrated). Glucose was added to maintain a final concentration of approximately 1.0 g/liter. The fermentation temperature was maintained at 37 "C and the pH was controlled at 7.0 with ammonium hydroxide. Air flow was fixed at 15 liters/min. Culture growth was monitored by measuring A550 nm with Gilson Stasar I1 spectrophotometer. IPTG was added to a final concentration of 1 mM at &,o = 15 to induce the tac promoter. Cells were harvested 4 h after the addition of IPTG.
Large Scale Osmotic Shock-Cells from 1 liter of cell suspension from a fermentation run (AS60 = 70) were collected by centrifugation (7480 X g, 7 min, 4 "C) and were suspended in 3.6 liters of 20% sucrose, 30 mM Tris, pH 7.5, 1 mM EDTA. After suspension, cells were kept at 23 "C for 15 min. Cells were then collected by centrifugation (8670 X g, 10 min, 4 "C) and suspended in 1.8 liters of ice-cold water containing protease inhibitors (1 pg/ml each of chymostatin, bestatin, antipain, leupeptin, and pepstatin A) and kept on ice for 10 min. After centrifugation (9800 X g, 15 min, 4 "C), the supernatant (designated water wash) containing aldose reductase was collected and supplemented with DTT and Tris-HC1, pH 7.5, to 0.1 and 5 mM, respectively.
Purification of Recombinant Human Aldose Reductase-All purification steps were carried out at 4 "C in buffers containing 1.0 mM DTT. The following method was used to purify human recombinant aldose reductase from the osmotic shock extract (water wash) obtained from 1 liter of JMlOl(pMON5997) fermentation culture. TO concentrate the proteins present in the initial water wash, an ammonium sulfate fractionation step was carried out. Material precipitating in the presence of 50-80% ammonium sulfate was collected by centrifugation at 10,000 X g for 20 min, and resuspended in 100-200 ml of 25 mM imidazole-HC1, pH 7.4. From this point, the crude enzyme preparation was split into three aliquots, and each was processed separately through the following purification steps. The crude lysate was applied at 30 ml/h to a 1.6 X 27-cm chromatography column packed with PBE 94 chromatofocusing resin (Pharmacia LKB Biotechnology Inc.) which had been previously equilibrated with 25 mM imidazole buffer, pH 7.4. Proteins were eluted from the column with 1:8-diluted Polybuffer 74, pH 5.0. The column eluate was continuously monitored by measuring A280 nm. Fractions of 5.3 ml were collected and were measured for pH, aldose reductase activity, and protein content by the dye-binding method of Bradford (35). Fractions containing aldose reductase activity were pooled and di-alyzed against 120 volumes of 10 mM potassium phosphate, pH 7.0, containing 0.5 mM EDTA. The dialyzed sample was then applied a t 60 ml/h to a 2.5 X 30-cm hydroxylapatite column (Bio-Rad) equilibrated with 10 mM potassium phosphate buffer, pH 7.0, containing and 0.5 mM EDTA. After washing the column with 250 ml of this buffer, the enzyme was eluted with a linear gradient (10-300 mM) of potassium phosphate, pH 7.0. Fractions containing aldose reductase activity were pooled, concentrated to approximately 1-2 mg/ml, and stored at 4 "C.
Amino Acid Sequence Analysis of Recombinant Aldose Redwtase-Aliquots of recombinant aldose reductase taken from selected chromatofocusing fractions (greater than 90% pure) were subjected to automated Edman degradation using an Applied Biosystems, Inc. Model 470A gas-phase sequenator (36). The respective phenylthiohydantoin derivatives were identified by reverse phase high performance liquid chromatography analysis in an on-line fashion using an Applied Biosystems, Inc. Model 120A phenylthiohydantoin analyzer fitted with a Brownlee 2.1-mm inner diameter phenylthiohydantoin-Cla column.
Purification of Aldose Reductase from Human Pkzcenta-Aldose reductase was purified to apparent homogeneity from human placenta as described previously (37), except that ammonium sulfate fractionation was used as the first step. During purification, the enzyme solution was maintained in 10 mM imidazole-HC1, pH 7.0, containing 5 mM @-mercaptoethanol. Homogeneity of the purified enzyme was established by the movement of a single protein band on reducing SDS-PAGE a t pH 8.6 and by the appearance of a single protein peak coincident with aldose reductase following Sephadex G-100 gel filtration. The purified enzyme was stored at 4 "C in 50 mM imidazole-HCI, pH 7.0, containing 5 mM @-mercaptoethanol.
Enzyme Assays-Aldose reductase activity in purification fractions was measured at 23 "C in 10 mM sodium phosphate, pH 6.2, containing 5 mM @-mercaptoethanol, 10 mM DL-glyceraldehyde, and 0.15 mM NADPH in a total volume of 1 ml as described previously (38). The reaction was initiated by addition of NADPH and enzyme activity determined by measuring the rate of disappearance of NADPH by monitoring absorbance of the reaction mixture a t 340 nm using a Beckman DU64 spectrophotometer. One unit of enzyme oxidized 1 rmol of NADPH/min. Reaction mixtures without enzyme and/or aldehyde substrate were used as controls. For kinetic studies, enzyme activity was determined in 50 mM potassium phosphate, pH 7.0, 0.1 mM NADPH, aldehyde substrate, 0.4 M ammonium sulfate or lithium sulfate if indicated, and appropriate amounts of enzyme in a total reaction volume of 1 ml.
Data Analysis-Individual saturation curves used to obtain Vma= and K ' , were fitted to a general Michaelis-Menton equation using a nonlinear iterative fitting program (14) where v is the enzyme velocity and A is the substrate concentration. Data conforming to linear uncompetitive, and noncompetitive or competitive inhibition were fitted to Equations 2-4, respectively, where I is the inhibitor concentration and Kii and K, are the intercept and slope inhibition constants. In all cases, best fit to the data was chosen on the basis of the standard error of the fitted parameter and the lowest value of u, which is defined as the sum of the squares of the residuals divided by the degrees of freedom (number of observations minus the number of parameters calculated). The concentration of the substrate/inhibitor used to estimate kinetic parameters was in most cases varied from 4 X Ki or K ' , to one-fourth the value of Ki and K',,,. Data are expressed as means f S.E. Statistical significance was determined using unpaired Student's t test. Data were considered statistically significant when the p value was <0.001.  (Fig. 1, lanes 2-6). The abundance of this polypeptide increased significantly in cells cultured for 2 h following the addition of 1 mM IPTG (compare Fig. 1, lanes  2 and 3). No corresponding immunoreactive material was observed in extracts of pMON5842, a control plasmid directing the expression of chloramphenicol acetyltransferase (Fig.  1, lanes 7-1 1 ). Aldose reductase activity was abundant in the water wash extract fraction from cells containing pMON5997 but was undetectable in similarly treated extracts from a culture containing pMON5842 (data not shown). Comparison of whole cell extracts and material from osmotic shock fractions from cultures containing pMON5997 revealed that most of the aldose reductase expressed in E. coli can be released by osmotic shock. This confirms a previous report that recombinant human aldose reductase expressed in E. coli can be released by osmotic shock despite the absence of a recognizable secretion signal in the protein structure (39). Recovery of recombinant aldose reductase in the water wash fraction provides obvious advantages as it affords substantial partitioning of the enzyme from intracellular proteins and hence easier purification. Table I1 summarizes the results of a typical purification of aldose reductase extracted from E. coli JMlOl fermentation cultures containing the pMON5997 expression construct. As a final step, aldose reductase was purified by hydroxylapatite chromatography where it eluted as a single coincident peak of enzyme activity and protein (Fig. 2). The final enzyme preparation (specific activity, 3.34 units/mg protein) was homogeneous as evidenced by a single polypeptide band on SDS-PAGE (Fig. 2, inset). Purified recombinant and native human aldose reductase comigrated on SDS-PAGE and reacted with antibodies against human placental aldose reductase (Fig. 2,  inset). This purification method resulted in a 40% recovery of starting enzyme units and greater than 15-fold purification with a total yield of 56 mg of recombinant aldose reductase derived from 1 liter of fermentation culture (Table 11). Sequential Edman degradation of the purified human recombinant aldose reductase revealed the following sequence: NHZ-A-S-R-L-L-L-N-N-G-A-K-M-P-, thus confirming the identity of the recombinant material as aldose reductase. Purified recombinant aldose reductase was stable for at least 6 months when stored at 4 "C. Approximately 50% activity was lost when the enzyme solution was stored at -20 or -70 "C in the presence of 3% polyethylene glycol (molecular weight 7000-9000). The enzyme was completely inactivated when stored at -20 or -70 "C in the absence of polyethylene glycol.

Expression
Expression of mutant aldose reductase cDNA sequences was accomplished by replacing the wild type aldose reductase cDNA sequence in pMON5997 with restriction fragments containing cDNA sequences encoding cysteine to serine substitutions at positions 80 (ALR2:C80S), 298 (ALR2:C298S), and 303 (ALR2:C303S). Conditions identical to those described above were used to express, extract, and purify wild type and mutant reductases from shaker flask cultures. Purified wild type and mutant reductases comigrated on SDS-PAGE and cross-reacted with antibodies to native human aldose reductase (Fig. 3).
Structural and Kinetic Comparison of Native and Recombinant Human Aldose Reductases-Analytical chromatofocusing of native and recombinant reductases revealed a slight difference in their apparent isoelectric pH values. When applied separately to different chromatofocusing columns, recombinant and native aldose reductases eluted at positions corresponding to pH 5.83 and 5.67, respectively (Fig. 4A). This minor difference in the apparent isoelectric pH was confirmed when the enzymes were mixed and subjected to chromatofocusing simultaneously (Fig. 4B). It is probable that the difference in acetylation of the amino-terminal alanine residue is responsible for this apparent difference in isoelectric pH (40). Apart from potential differences at the amino terminus, it is unlikely that the primary structure of the recombinant aldose reductase differs from the native human aldose reductase because the coding sequence in pMON5997 was identical to all aldose reductase cDNA clones isolated from the human placental cDNA library (data not shown).
Kinetic Constants for Wild Type and Mutant Aldose Reductases-Kinetic constants of the wild type and mutant reductases determined with a series of hydrophilic and hydrophobic substrates are shown in Table 111. All of the enzyme forms preferentially utilized hydrophobic aldehyde substrates as the catalytic efficiencies (kJK',) measured with benzaldehyde and p-nitrobenzaldehyde were 1 to 4 orders of magnitude greater than that measured with the hexose and pentose substrates glucose and xylose, respectively. Among this series of substrates, the turnover number (kat) measured with the C80S and C303S mutants deviated <2-fold from wild type, with the exception of glucose. In this case, the bat values were either increased (>4.l-fold for C80S) or decreased (>2.3-fold for C303S) relative to wild type. No consistent trend was evident to associate mutation of Cys-80 and Cys-303 and pattern of substrate selectivity. For example, divergent changes in catalytic efficiency for the reduction of xylose and glucose were measured with the C80S mutant. This mutant also displayed enhanced utilization of benzaldehyde (kCat/K, increased >4-fold relative to wild type) although its utilization of para-nitro-substituted benzaldehyde (p-nitrobenzaldehyde) was decreased almost 3-fold. The catalytic efficiency of the C303S mutant was also decreased with both hydrophilic and hydrophobic substrates. In contrast, the $,, values measured with the C298S mutant were consistently elevated (24fold) with all substrates tested even though the catalytic efficiency for these substrates was lower than wild type enzyme. Such behavior seems characteristic of ALR2:C298S because reduction in catalytic efficiency of the C80S and C303S mutants was usually not accompanied by an increase in the turnover number. Increased turnover number for the ELUTION VOLUME reduction of DL-glyceraldehyde was also evident with ALR2:C298S but not with the other mutants (Table IV). Importantly, there was no statistically significant difference in K l r n (~m p~) observed among the wild type and mutants (Table IV).
Kinetic differences between ALR2:C298S and other forms of the enzyme were even more marked after carboxymethylation. Carboxymethylation of the wild type, C80S and C303S mutants resulted in an increase in K'm(DL-glymmldeh&+ There was no statistically significant difference between the kinetic constants of the carboxymethylated wild type or carboxymethylated C80S or C303S mutants. Moreover, carboxymethylation had no significant effect on the kinetic parameters of C298S mutants. These results suggest that the functionally significant site of carboxymethylation of aldose reductase is To further investigate the role of these cysteine residues, the inhibitor sensitivity of wild type and mutant reductases was determined. Substitution of Cys-80 caused a small but statistically significant increase in Ki(mrbini1) ( Table v). How-CYS-298. Fraction "&W ever, the Ki(aorbinil) was not affected on substitution of Cys-303 with serine. While the sorbinil inhibition constants for the wild type and C80S and C303S mutants were similar, the C298S mutant was approximately 10-fold less sensitive to this inhibitor than the other enzyme forms tested (Table V). Sensitivity to tolrestat was largely unaffected by serine substitutions a t positions 80,303, or 298. Although Kictolrestet) was 2-3-fold higher for the ALR2C303S mutant, the difference in the estimated Ki values were slight compared to a 10-fold increase in Ki(mrbinil) observed with ALR2:C298S. The Kictolrrstat) for ALR2:C298S was also similar to the Ki value of the wild type enzyme, indicating that the substitution of Cys-298 does not affect tolrestat binding but significantly attenuates the sensitivity of the enzyme to sorbinil. Carboxymethylation dramatically elevated the Ki(mhinil) for ALR2:WT, C80S, and C303S. In contrast, this treatment had virtually no effect on the Ki(mrbini1) observed with ALR2:C298S (Table V). Carboxymethylation also had no significant effect on the Ki(tolrestat) observed for the wild type and C298S and C303S reductase mutants. However, carboxymethylation dramatically elevated the Ki(tolmstat) for ALR2C80S mutant.
Modulation of Enzyme Activity by Lithium Sulfate-To investigate the effect of sulfate ions on the activity of the enzymes, the enzyme activities were determined with and without 0.4 M Li2S04 in the standard assay buffer (n = 5). Addition of Li2S04 to the assay buffer increased the activity of the wild type enzyme 2.36 0.22-fold. The activities of the C80S and C303S mutants were similarly affected (1.89 +. 0.04 and 2.27 f 0.15-fold increases, respectively). However, the activity of the C298S mutant assayed in the presence of Li2S04 was increased only 1.16 f 0.04-fold. Compared to the wild type enzyme, the difference in fold activation was statistically significant only for ALR2C298S ( p < 0.001).

DISCUSSION
In spite of its proposed role in the etiology of secondary complications of diabetes (1-5), little is known about the active site and the reaction mechanism of aldose reductase. Difficulties in purifying adequate quantities of human aldose reductase and its marked sensitivity to oxidation have hampered efforts toward elucidation of the kinetic mechanism (14)(15)(16). In this article we describe methods to overexpress wild type and mutant forms of human aldose reductase using E. coli host cells. Use of an efficient bacterial plasmid in conjunction with a simple three-step purification method has provided yields similar or favorable to other bacterial and eucaryotic systems used for overexpression of human aldose reductase (39-44). The similarity in kinetic behavior observed with native and recombinant human aldose reductases in this and other studies (44,45) appears to validate the use of aldose reductase produced in procaryotic cells as a reliable source of enzyme for structural and functional studies. We and others (44, 45) have detected a minor difference in isoelectric pH between native and recombinant reductases produced in procaryotic expression systems, most likely a reflection of differences in amino-terminal acetylation (40). Although the potential contribution of the acetylated amino terminus to the gross conformation of the enzyme or its mechanism of catalysis was not directly studied, we consider it unlikely that acetylated and nonacetylated enzymes are functionally different as their kinetic properties appear to be virtually identical.
Crystallographic studies of porcine lens aldose reductase complexed with 2'-monophosphoadenosine-5'-diphosphoribose revealed that the enzyme is a single domain 8-stranded parallel b/a-barrel similar to triose phosphate isomerase (46). From this structure, it may be inferred that the nucleotide cofactor binding domain, and presumably the active site, should be located at the COOH-terminal portion of the barrel. The structural model for porcine aldose reductase predicts three solvent-exposed cysteine residues, namely those corresponding to human Cys-80, Cys-298, and Cys-303, all of which are located in the carboxyl-terminal region of the @/a-barrel (46). Given the high degree of sequence identity between human and porcine enzymes (>88%), it is likely that their structures will be similar.
An accumulating body of evidence suggests that oxidation of cysteine residues alters the catalytic properties and inhibitor sensitivity of aldose reductase (12)(13)(14)(15)(16). The oxidized (activated) enzyme is a high K,,,/high Vmax form of the reduced or unactivated enzyme (13-16). The activated enzyme form, which is relatively insensitive to sorbinil but not to tolrestat (15), is thought to arise from reduced aldose reductase by modification of 1 or more exposed cysteine residues (15, 16,19).
To study the potential involvement of cysteine residues in the catalytic mechanism and interaction with inhibitors, we constructed aldose reductase mutants with structurally conservative cystein-serine substitutions at positions 80, 298, and 303. From the data summarized in Table 111, it seems clear that none of the cysteine residues is essential for catalytic activity, as all three mutants express activity with both hydrophobic aromatic aldehydes as well as hydrophilic sugar substrates. Overall, it appears that mutation of these cysteines causes a decrease in catalytic efficiency when measured with these substrates. However, the C80S mutant was exceptional when assayed with benzaldehyde and glucose, as the kJK',,, was increased >4-fold for both substrates as compared to control. However, when assayed with p-nitrobenzaldehyde and xylose, the kat/K',,, was lower than wild type (1.2-and 2.7-fold, respectively). In terms of altered kinetic parameters, the most striking changes involved ALR2:C298S. This mutant displayed markedly decreased catalytic efficiency with the substrates examined ( K m increased 5-50-fold) even though the kat was increased 24-fold. The susceptibility of ALR2:C298S to inhibition by sorbinil was markedly reduced (approximately 10-fold) while its sensitivity to tolrestat was virtually unaffected.
The potential role of cysteine residues in catalysis and inhibition of human aldose reductase was also investigated by chemical modification studies. Carboxymethylation of ALR2:WT, C80S, and C303S resulted in a 4-7-fold increase in K'm(oL.glyceraldehyde) and a marked reduction in sensitivity to sorbinil. These kinetic parameters were unchanged after carboxymethylation of ALR2:C298S. On the basis of these observations, we suggest that the reported changes in the kinetic properties of the enzyme upon carboxymethylation are predominantly due to selective modification of Cys-298. However, aldose reductase:C298S is not equivalent in its kinetic properties to the carboxymethylated ALR2:WT or C80S and (2303s mutants; carboxymethylation of these enzymes caused a decrease in the kat, whereas the kat of ALR2:C298S was 3-4-fold higher than that of the wild type under identical conditions. Moreover, the Ki(&,inil) observed with ALR2:WT increased more than 100-fold upon carboxymethylation, in contrast to a 10-fold increase observed on substitution of Cys-298 with serine. The resistance of ALR2:C298S to reduction in sensitivity to sorbinil as a result of carboxymethylation further suggests that Cys-298 is the predominant cysteine residue subject to modification and consequent kinetics changes. A recent structural model of human aldose reductase complexe! with NADPH has shown that Cys-298 is located within 4.13 A of the C-4 of the nicotinamide portion of the coenzyme and thus may be considered in the active site (47). The possibility that Cys-298 could be the proton donor in the catalytic reaction has been rejected on the argument that it is not conserved among members of the aldo-keto oxidoreductase enzyme family (47). Our data provide support for the idea that Cys-298 cannot be the proton donor, as it appears that substitution of the thiol group with an alcohol has only a nominal effect on catalytic activity, unlike that expected if the mutation replaced a proton donating group.
The differences between ALR2:C298S and the carboxymethylated wild type enzyme may be due to different perturbations introduced into the active site environment by carboxymethylation and cysteine to serine substitution, respectively. Based on the close proximity of the Cys-298 thiol to the nicotinamide ring, it seems plausible that substitution of the thiol with an alcohol could alter the interaction of this side chain with the nicotinamide ring either directly or indirectly by changing the solvation properties of the active site pocket. A structural model of ALR2:C298S complexed to NADPH solved to 2.75-A resolution revealed that the interactions between the enzyme and NADPH appeared to be identical to those observed in the wild type holoenzyme structural model (see accompanying paper, Ref. 48). However, a higher resolution structure of the C298S mutant, as well as construction and analysis of additional Cys-298 mutants containing side chains with different physical properties, will be required to more fully evaluate the potential interaction of Cys-298 with the nicotinamide ring. It seems likely that introduction of a large charged group at Cys-298 by carboxymethylation would also be expected to alter the active site milieu. By virtue of its proximity to the active site, it is plausible that the increase in K'm(DL.glycereldehyde) observed after carboxymethylation of Cys-298 in the wild type and C80S and C303S mutants may reflect altered access of the substrate to the active site due to steric hindrance provided by the carboxymethyl group.
Recent studies have suggested that isomerization of the ALR2. NADP+ binary complex is the rate-limiting step in the reaction catalyzed by unactivated aldose reductase (15, 49). Upon activation, the isomerization rate of the enzyme-coenzyme binary complex is increased, which facilitates NADP+ release and increases the turnover rate of the catalytic cycle (14,15). Therefore, if activation is due to oxidation of one or more critical cysteine residues, it seems plausible that modification of these residues should lead to an increase in kcat.
The evidence presented above demonstrates that substitution of Cys-298 with serine results in an increase in kcat. Therefore, we suggest that modification of Cys-298 may be responsible for oxidation-induced activation of aldose reductase.
Stimulation by sulfate is a well known property of aldose reductase (38). Although the mechanism is not known, the inability of sulfate to stimulate ALR2:C298S suggests that this process involves the participation of Cys-298. However, unlike oxidation-induced activation, stimulation by sulfate does not alter the Ki(sorbinil),3 although sulfate increases lation by sulfate and activation by oxidation are not identical processes, although both seem to involve Cys-298. Substitution of Lys-262, a residue thought to participate in NADPH binding (50), is reported to attenuate the ability of sulfate to stimulate aldose reductase activity (44). Further investigations are necessary to delineate the role of Cys-298 and Lys-262 in the catalytic reaction, but it appears likely that alterations to residues involved in binding (and release) of NADP(H) may give rise to kinetic behaviors consistent with the activated form of the enzyme. The observation that carboxymethylation or substitution of Cys-298 alters Ki(sorbini1) but not &,,Irestat) suggests that the two inhibitors do not bind to the same site. These results are consistent with our earlier observations that double inhibition plots of sorbinil and tolrestat are nonparallel (9). It has been suggested that all aldose reductase inhibitors bind at a unique site distinct from the substrate-binding site (51). However, mutations or structural modifications giving rise to concomitant changes in substrate binding and sorbinil binding suggest that the sorbinil-binding site may be at or close to the active site of the enzyme. Cys-298 may form a part of the sorbinil binding site, or the "S" site. The tolrestat-binding site, or "T" site, which is insensitive to carboxymethylation or substitution of Cys-298, is likely to be located at a different site possibly farther removed from the active site of the enzyme.
These results indicate that among the 3 solvent-exposed cysteine residues located in the COOH-terminal portion of the aldose reductase BIa-barrel, cysteine 298 may play a key role in the conversion of aldose reductase from an unactivated to an activated state (19). The chemical and structural details of this conversion have practical importance in the design of aldose reductase inhibitors, as the activated form of aldose reductase displays markedly reduced sensitivity to some enzyme inhibitors. In view of these findings, it is possible that the in uiuo oxidation state of the cell may regulate the proportion of activated and unactivated forms of aldose reductase.