THE JOURNAL OF B I O ~ I C A L CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 269, No. 3, Issue of January 21, pp. 2183-2188, 1994 Printed in U.S.A.

Studies on Pig Aldose Reductase IDENTIFICATION OF AN ESSENTIAL ARGININE IN THE PRIMARY AND TERTIARY STRUCTURE OF THE ENZYME*

(Received for publication, May 27, 1993, and in revised form, September 27, 1993)

Terrance J. KubiseskiS, Nancy C. Green$, David W. BorhaniOR and T. Geoffrey FlynnSlI From the +Department of Biochemistry, Queen’s University, Kingston, Ontario, Canada K7L 3N6 and SBioCryst Pharmaceuticals Znc., Birmingham, Alabama 35244

Reaction of pig muscle aldose reductase with phenyl- glyoxal resulted in the chemical modification of 2 argi- nine residues with accompanying loss of catalytic ac- tivity. The amino acid sequences of radioactive peptides resulting from the reaction of aldose reduc- tase with [*4C]phenylglyoxal followed by tryptic diges- tion and high performance liquid chromatography separation allowed identification of the modified argi- nine residues as R268 and R293. In the presence of the coenzyme NADP+, R268 is protected from modification by phenylglyoxal, while R293 becomes hyper-reactive. Phenylglyoxal modification of aldose reductase is slowed 3-fold by the presence of the coenzyme analog ADPRP; however, both arginines are still modified. These chemical modification results are in complete ac- cord with the previously determined crystal structures of human and porcine aldose reductase complexed with NADPH, NADP+, and ADPRP. These structures indicate that R268 is located at the adenosine binding site, salt bridged to the 2”phosphate group of NADP(H) and AD- PRP. Arginine 293 is near the surface of the enzyme and is part of the C-terminal loop. In the apoenzyme or the ADPRP complex, R293 is partially protected by loop 7; upon binding NADP(H), loop 7 folds down over the co- enzyme, thus exposing R293 to solvent. Our modifica- tion studies provide further evidence of the conforma- tional change that occurs during the aldose reductase catalytic cycle.

Aldose reductase (ALR2; EC 1.1.1.21),l a monomeric, pri- marily NADPH-dependent oxidoreductase is a member of the aldo-keto reductase superfamily of enzymes (Carper et al., 1987; Bohren et al . , 1989). Like other members of this family the enzyme exhibits a very broad substrate specificity which makes it difficult to establish its primary physiological role (Flynn, 1986). Unlike other oxidoreductases the enzyme does not have the expected dinucleotide or Rossman binding fold (Rondeau et al . , 1992; Wilson et al . , 1992) and the general structure of the enzyme as a single domain Pla-barrel is more

* This work was supported by a grant from the Medical Research

(to T. J. K.), and by National Institutes of Health Grant DK44789 (to D. Council of Canada (to T. G. F.), a Medical Research Council Studentship

the payment of page charges. This article must therefore be hereby W. BJ. The costs of publication of this article were defrayed in part by

marked ”advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

ll Present address: Southern Research Institute, 2000 Ninth Ave. South, Birmingham, AL 35205.

11 To whom correspondence should be addressed. The abbreviations used are: ALR2, aldose reductase; ADPRP, 2’-

monophosphoadenosine-5’-diphosphoribose; 2’,5’-ADP, adenosine 2’,5’- diphosphate; HPLC, high performance liquid chromatography.

characteristic of triose-phosphate isomerase than that of other oxidoreductases (Rondeau et al . , 1992). Structural studies to date have given considerable insight into the unusual fashion of nucleotide binding, but because ALR2 exhibits ordered ki- netics with compulsory binding of NADPH before substrate (Grimshaw et a l . , 1990a; Kubiseski et al . , 19921, the nature of the substrate binding site can only be arrived at through con- jecture or through the study of unreactive substrate analogs.

There have been many previous studies designed to elucidate amino acids involved in binding and catalysis at the active site of ALR2. Studies from our laboratory have shown K262 to be present at the coenzyme binding site (Morjana et al . , 1989). This residue is part of a motif, IPKS, which occurs in all aldo- keto reductases (Carper et al . , 1987; Bohren et al., 1989). Chemical modification (Morjana et al . , 1989) and site-directed mutagenesis (Bohren et al., 1991) of K262 alters KM and kcat but does not prevent binding of coenzyme. Other studies have implicated arginine (Doughty and Conrad, 1982; Halder et a l . , 1985), histidine (Doughty and Conrad, 1982; Halder et a l . , 1985), and cysteine (Liu et a l . , 1989) in catalysis. Recent work by Tarle et al. (1993) has implied that Y48 acts as an acid-base catalyst in ALR2 and suggests that D43, K77, and HllO are important to the structure and function of the active site. Site- directed mutagenesis of cysteine residues (Petrash et al . , 1992; Bohren and Gabbay, 1993) has explained to a large extent the sensitivity of the enzyme to oxidation (Grimshaw et al., 1989; Vander Jagt et al., 1990) and has suggested that C298 plays a key role in regulating catalytic activity.

Previous kinetic studies have shown that during the cata- lytic cycle of ALR2 there is an isomerization of enzyme-coen- zyme complexes. In the forward reaction (aldehyde reduction) this isomerization, which is accompanied by a detectable con- formational change, is rate-limiting (Kubiseski et al., 1992). Recent work on the crystal structure of recombinant ALR2 complexed with NADPH (Wilson et al . , 1992; Borhani et al . , 1992) in comparison to that of the apoenzyme (Rondeau et al., 1992) has clarified the nature of the large conformational change which takes place upon binding of coenzyme. Upon NADPH binding, the large loop between P-strand 7 and a-helix 7 (loop 7) rotates by 51” and folds over NADP(H) locking the coenzyme into place.

In the present paper we provide further evidence that a conformational change occurs. In the absence of coenzyme, chemical modification of ALR2 with phenylglyoxal causes an inactivation of the enzyme with modification of R268 (adeno- sine binding site) and R293 (C-terminal loop). The presence of NADP+ protects R268 from modification, but it makes R293 hyper-reactive. We conclude that this further exposure of R293 to reaction is due to the conformational change that occurs upon coenzyme binding. This work was presented in part at the Enzymology and Molecular Biology of Carbonyl Metabolism Workshop in Dublin, Ireland, July 1992 (Kubiseski et al. 1993)

2183

2 184 Role of Arginines in Aldose Reductase

EXPERIMENTAL PROCEDURES Materials-Aldose reductase was prepared from pig muscle obtained

at the Ross McFedridge Abattoir of Glenburnie, Ontario, by the proce- dure described previously (Cromlish and Flynn, 1983). DL-Glyceralde- hyde, glycyltryptophan, phenylglyoxal, ADPRP, NMN, and trypsin (Grade 111) were purchased from Sigma. Sephadex G-50 was obtained from Pharmacia LKB Biotechnology Inc. [7-14C1Phenylglyoxal was pur- chased from Amersham Corp. A Bradford assay kit was purchased from Bio-Rad. Trichloroacetic acid was purchased from Pierce Chemical Co. HPLC-grade acetonitrile and sodium phosphate were obtained from BDH (Canada) Ltd., Domal, Quebec. NADPH and NADP+ were ob- tained from Boehringer Mannheim Canada Ltd., Laval, Quebec.

Standard Enzyme Steady State Activity Assay-Enzyme assays were performed at 25 "C using a Hewlett-Packard HP8452A diode array spectrophotometer equipped with a Hewlett-Packard 9000-300 (model 98561A) desktop computer. Reaction mixtures consisted of 0.1 M sodium phosphate buffer, pH 7.0,lOO p~ NADPH, and 4 m~ DL-glyceraldehyde. Reactions were started by addition of enzyme, and the activity was determined spectrophotometrically by measuring the change in absor- bance at 340 nm. A 295-nm U V cut-off filter was used to prevent pho- todegradation of the NADPH. Temperature was maintained to within 20.1 "C with a thermostatted circulating water bath and thermospace. Blank rates (DL-glyceraldehyde absent) if any, were subtracted from all reaction rates obtained with the control. ALR2 concentrations were calculated using c = 53 000 M" cm-' at 280 nm (Branlant, 1982).

Stoichiometry of Phenylglyoxal Incorporation into ALR2-ALR2 was incubated at 0.50 mg/ml in a solution containing 75 m~ sodium bicar- bonate, pH 8.0, at 25 "C for 20 min in the presence of varying concen- trations of [7-14Clphenylglyoxal (5.13 x 1013 dpdmol). The sample was then centrifuged through a 1-ml spun column of preswollen Sephadex G-50 as described by Penefsky (Penefsky, 1971). The volumes of the effluents were measured and aliquots were assayed for enzyme activity, protein concentration via the Bradford assay (Bradford, 1976) with purified pig muscle ALR2 as standard, and radioisotope incorporation.

Modification of ALR2 with Phenylglyoxal and Substrate Protection Experiments-Purified ALR2 (0.20 mglml) was incubated with varying concentrations of phenylglyoxal at 25 "C in the dark in 75 m~ sodium bicarbonate buffer, pH 8.0. The reaction was initiated by adding phenyl- glyoxal. Ten-pl aliquots were removed at various times and assayed for enzyme activity. The initial activity ofALR2 was measured in triplicate immediately before adding reagent and the enzyme activity during the time course of reaction was expressed as a fraction of the control. Pro- tection experiments were performed by preincubating the enzyme with various ligands for 20 min before adding phenylglyoxal.

Binding of ADPRP to ALR2-The interaction of ADPRP to ALRS was examined through intrinsic enzyme fluorescence measured on a Perkin Elmer LS50 luminescence spectrometer. A solution consisting of 2.5 ml of 0.1 p~ ALR2 and varying amounts of ADPRP in 100 m~ sodium phosphate, pH 7.0, was excited at 280 nm (slit width of 2.5 nm), and fluorescence was measured between 300 and 540 nm. A cut-off filter (290 nm) was in place for the emission beam to prevent light scattering from being detected. The correction for the internal filter effect on the excitation beam was determined with glycyltryptophan and NADP+; all fluorescence values were corrected for dilution.

Enzymatic Digest of Modified AL.R2-ALR2 (0.17 mg) was incubated with 5 m~ [7-14Clphenylglyoxal in 75 m~ sodium bicarbonate buffer, pH 8.0 (in a total volume of 100 pl), at 25 "C in the dark for 60 min. Next the samples were desalted using the rapid microcentrifuge desalting techniques described by Penefsky (Penefsky, 1971). The samples were dried and digested with trypsin by the method of Stone et al. (1989) with a few modifications. To the dry protein was added 50 pl of 8 M urea, 0.4 M ammonium bicarbonate, 4 m~ calcium chloride buffer, and 10 pl of 45 r m dithiothreitol. This was then incubated at 50 "C for 10 min and, after cooling to room temperature, 10 pl of 120 m~ iodoacetamide was added, and the sample was allowed to incubate at room temperature for 10 min. The denatured enzyme was then digested by adding 140 pl of H20 and 5.0 pl of trypsin (1.6 mglml). After incubation at 37 "C for 8 h, the reaction was stopped by acidifying the digest with 0.1% trifluoro- acetic acid and injected directly into an HPLC.

Peptide Purification and Isolation-The enzyme digests were sub- jected to HPLC (Beckman) using a reverse-phase column (0.46 X 25-cm Vydac C18) pre-equilibrated with 0.1% aqueous trifluoroacetic acid. The peptide peaks were eluted with a linear gradient of 040% acetonitrile containing 0.1% trifluoroacetic acid at a flow rate of 1 d m i n over a period of 80 min. The peptides were monitored at 217 nm and the radioactivity of individual fractions was determined by scintillation counting using Beckman Ready Cap solvent-free scintillation medium.

0.0

-0.5

-1 .o

- -1.5 W 0

2 -2.0 W

c -2.5

-3.0

-3.5

-4.0 0 10 20 30 40 50 60

Time (min)

FIG. 1. Inactivation of pig muscle ALR2 by phenylglyoxal. The enzyme (5 p ~ ) was incubated with the following concentrations of phenylglyoxal: 0 m~ (O), 0.25 m~ (01, 0.50 m~ (V), 1.0 m~ ( W , 2.5 m~ ( 0 , and 5.0 m~ (W) in 75 m~ sodium bicarbonate buffer, pH 8.0, at 25 "C. The natural logarithm of fractional residual activity is plotted versus time of reaction.

The [7-14Clphenylglyoxal-labeled peptides were purified by reapplying them to the HPLC column followed by elution with a shallower linear gradient of acetonitrile at a flow rate of 1 mumin. Fractions containing labeled, purified peptides were dried down to 50 pl, and kept at -20 "C for subsequent identification by amino acid sequencing.

RESULTS

Reaction of M R 2 with Phenylglyoxal-Reaction of ALR2 with different concentrations of phenylglyoxal resulted in a time-dependent inactivation of the enzyme activity, as shown in Fig. 1. When the enzyme (5 p ~ ) was incubated with 5 II~M phenylglyoxal for 40 min, no activity remained, suggesting that an arginine residue important for aldose reductase activity was being modified. The inactivation of the enzyme followed pseudo-first order kinetics over a concentration range of 0.25- 5.0 mM phenylglyoxal. The plot of the logarithm of residual activity versus the time of reaction with phenylglyoxal is linear as predicted by Equation 1:

-In (EIE,) = k,,,t (Eq. 1)

where E and Eo are the activity at time t and zero time, re- spectively, and kobs is the observed first-order rate constant for the loss of enzymatic activity.

Fig. 2A illustrates the observed hyperbolic dependence be- tween the value of hobs and the concentration of phenylglyoxal. This indicates that the reagent forms a reversible complex with the enzyme (E.1) prior to the covalent attachment of phenylg- lyoxal to ALR2. This type of kinetic behavior can be described bY

kl k,

k - I E + I e E . I + E - I (Eq. 2)

where E is the enzyme, Z is phenylglyoxal, and E-Z is the cova- lently-modified ALR2. A steady state treatment of this mecha- nism is described by Equation 3:

l/koba = K,/k, [ I ] + l/k, (Eq. 3)

where KI = (kl + k Z ) / k l = [EI[Zl/[EZ]. A double-reciprocal plot of

Role of Arginines in Aldose Reductase 2185

12

10

“ 8 ”

“ 6 E .-

n 0 n

x 4

2

0 0

0.8

- 0 .6 C .- E v

n 0.4

0

Y \ 7

0.2

A

I I I I I I

1 2 3 4 5 6

[Phenylglyoxal] (mM)

B

t

I I I I

0 1 2 3 4

l/[Phenylglyoxal] (mM)”

FIG. 2. Dependence of the pseudo-firat order rate constants of inactivation on phenylglyoxal concentration. The enzyme was in- cubated with various concentrations of phenylglyoxal as described in the legend to Fig. 1. A, the rate constants of inactivation are plotted

data in A. uersus the phenylglyoxal concentration. B, double reciprocal plot of the

kobs uersus phenylglyoxal concentration gave a straight line with a positive intercept on the y axis (Fig. 2 B ) . From the graph, k2 and KI were calculated to be 0.13 min-l and 2.4 m ~ , respectively.

Stoichiometry of Reaction of ALR2 with Phenylglyoxal- ALR2 was incubated with varying concentrations of [7-’4Clphenylglyoxal for 20 min at 25 “C. The samples were centrifuged through a column of Sephadex G-50 to remove free [7-14Clphenylglyoxal. Fig. 3 shows that loss of activity is lin- early related to the amount of phenylglyoxal incorporation. Extrapolation of the plot yields a value of 3.45 2 0.36 mol of phenylglyoxal incorporated per mol of enzyme subunit at 0% residual activity. Since phenylglyoxal reacts in a 2:l ratio with respect to arginine (Takahashi, 1968), this indicates that 1.73 arginine residues are being modified under these conditions.

Protection Studies-The effect of substrates and competitive inhibitors on the course of modification of ALR2 was examined in order to determine whether the inactivation was due to direct modification of an essential arginine residue at the en- zyme’s active site, or due to modification of an arginine residue remote from the active site but which, nevertheless, resulted in an inactive enzyme form. Fig. 4 shows the effect of additives on the modification. In the absence of additives, ALR2 was inac-

1 .o

0.8

0 W

0.6

0.4

0.2

0.0 0 1 2 3 4

Moles Phenylglyoxal/Mole ALR2 FIG. 3. Stoichiometry of phenylglyoxal incorporation inALFt.2.

Incorporation of [7-14C]phenylglyoxal into pig muscle ALR2 was carried out as described under “Experimental Procedures.” Fractional residual enzyme activity is plotted uersus the number of equivalents of phenyl- glyoxal incorporated into ALR2.

0.0

-0.5

- 1 .o n 0

2 W -1.5

W

-2.0 c -

-2.5

-3.0

-3.5 0 10 20 30 40 50 60

Time (min) FIG. 4. Effect of ligands on the inactivation of- by phenyl-

glyoxal. The enzyme (5 p ~ ) was incubated in 75 m~ sodium bicarbon- ate buffer, pH 8.0, in the absence of phenylglyoxal (O), and in the presence of 3.5 m~ phenylglyoxal with no ligand (O), 4 m~ DL-glyceral- dehyde (V), or 2 m~ ADPRP (V) added. Aliquots of the reaction mixture were periodically withdrawn for the assay of enzymatic activity. The natural logarithm of fractional residual activity is plotted uersus time of reaction.

tivated in about 30 min. No significant change was noted when DL-glyceraldehyde or NMN (the nicotinamide moiety of NADP; data not shown) were present during the modification. How- ever, ADPRP protected the enzyme somewhat against phenyl- glyoxal modification, reducing the rate of inactivation by a factor of 3. In contrast, when NADP+ was included in the re- action mixture, a biphasic reaction occurred (Fig. 5). A fast initial rate of inactivation, which resulted in a 50% loss in the enzyme activity, was followed by a much slower rate of further inactivation. The rate of the slow phase in the presence of NADP’ was about one-seventh that observed in the absence of NADP+.

2186 Role of Arginines in Aldose Reductase

0.0

-0.5

n

W 0 -1.0

27 W

-1.5 -

-2.0

-2.5 I I I I I I 1

0 10 20 30 40 50 60

Time (rnin)

FIG. 5. Effect of NADP+ on the inactivation of ALR.2 by phenyl- glyoxal. The enzyme (5 p ~ ) was incubated with 0 (0) or 1 (0) m~ phenylglyoxal or with 1 m~ phenylglyoxal and 100 NADP+ (V) in 75 m~ sodium bicarbonate buffer, pH 8.0. Aliquots of the reaction mixture were periodically withdrawn for assay of enzymatic activity. The natu- ral logarithm of fractional residual activity is plotted uersus time of reaction.

Kinetic analysis of ALR2 modified in the presence of NADP showed no significant changes in the KM values of either the NADPH or DL-glyceraldehyde. The KM of NADPH in the ab- sence and presence of modification was 0.87 * 0.3 and 0.54 * 0.18 p, respectively; the KM for DL- glyceraldehyde in the ab- sence and presence of modification was 41.2 * 4.2 and 43.2 * 4.5 p ~ , respectively. The effect on the enzymatic activity was due to a decrease in the Kcat of the reaction (Kcat = 0.45 0.02 s-l before modification and 0.25 * 0.01 s-l after modification). The stoichiometry of reagent incorporation in the presence of NADP' using radioactive phenylglyoxal was determined to be 2.03 at a residual activity of 54%, indicating that 1 arginine was modified under these conditions.

Fluorometric Titration of ALR2 by ADPRP-The lack of a biphasic curve of inactivation when ALR2 is modified by phenylglyoxal in the presence of ADPRP suggests that the con- formational change that occurs upon NADP' binding either does not occur or does not occur to the same extent upon ADPRP binding. 'Ib confirm this, the intrinsic fluorescence of ALR2 upon ADPRP binding was measured and compared to previous results obtained with NADP' (Grimshaw et al., 1990b; Kubiseski et al., 1992). ADPRP only quenches the emission of ALR2 intrinsic fluorescence by about 12% (data not shown), i.e. about one-third the extent of NADP' (32%).

Identification of Phenylglyoxnl-modified Peptides-A tryptic digest of [7-14C]phenylglyoxal-modifiedALR2 was separated by HPLC. The elution profile (Fig. 6 A ) of the digest showed two major radioactive peaks (peaks I and 11) between fractions 20 and 35. These two peaks, which routinely appeared in all mock digestions carried out in the absence of trypsin, appear to be artifactual, and were not further characterized. The peptide corresponding to radioactive peak IV was further purified by HPLC (Cis). Amino acid sequence analysis showed that this peptide spans residues 263-269 of the pig ALR2 sequence (263SVTPER1269) (Kubiseski et al., 1993) in which no phenyl- thiohydantoin-derivative for R268 was obtained. The peptide corresponding to a second major radioactive peak (peak 111, Fig. 6 A ) was also purified. In some digests, the radioactivity present in this peak approached 80% of that present in peak IV. Se- quence analysis of the peak I11 peptide indicated that it com-

\

I

II

A

0 20 40 60 80 1 0 0 120 1 4 0

Fraction Number

Ill B

"

I I , 1 1

0 20 40 60 80 100 120 140 Fraction Number

FIG. 6. Radioactivity elution profile following HF'LC of the tryptic digest of [7-14C]phenylglyoI-modified ALR2. A, elution profile of phenylglyoxal modified pig muscle ALR2. Peaks I, 11,111, and IV are described in the text. E , elution profile of ALR2 modified by phenylglyoxal in the presence of NADP'.

prised residues 290-296 (290SYNFtNWR296) in which R293 showed no phenylthiohydantoin-derivative.

A tryptic digest of ALR2 modified by [7-14Clphenylglyoxal in the presence of NADP' was performed to determine which amino acid was modified during the rapid phase of the partial inactivation shown in Fig. 5. As shown in Fig. 6B, radioactive peak IV was no longer present. Purification and sequence anal- ysis of the peak I11 peptide indicated that it contained residues 290-296, in which no phenylthiohydantoin-derivative was ob- tained for R293.

DISCUSSION

The solution-phase chemical modification results of aldose reductase (ALR2) described here are in complete accord with previous crystallographic results obtained for the ALRB apoen- zyme and the ADPRP complex (Rondeau et al., 19921, the

Role of Arginines in Aldose Reductase 2187

FIG. 7. Stem view of loop 7 and the NADP(H) binding site of ALR2.NADPE showing the arginine residues modified by phenyl- glyoxal. This view of human recombinant ALFBNADPH (C298S mutank Borhani et &., 1992) is from the top outside of the enzyme looking down on the coenzyme and the active site. Loop 7 is on the left, the C-terminal loop is on the far left, and the active site is in the back. The coenzyme is in the center, with adenosine at the bottom and the nicotinamide ring at the top. The ALR2 model and NADPH are color-coded by atom type: yellow (carbon), blue (nitrogen), red (oxygen), green (phosphorus). The Ca trace of porcine ALR2 apoenzyme (Rondeau et al., 1992) is shown in purple (dashed lines indicate disordered residues). The loop 7 residues of human ALR2 (through D216) fold over the adenosine portion of NADPH; those of porcine ALR2 apoenzyme (212LGSPARPW19) are shown in orange. On the right side, K262 and R268 coordinate the 2'-PO, of NADPH. On the left, R293 is protected by the C-terminal loop residue W295, and by the loop 7 residues (when loop 7 is open). The loop closes from left to right as exemplified by the relative positions of P215 and D216 in the open (orange) and closed (yellow, blue, and red) positions.

NADPH complex (Wilson et al., 1992; Borhani et al., 1992), and the NADP+ complex.2 Furthermore, these new results shed additional light on the d y n k c s of the conformational change ALR2 undergoes upon binding coenzyme. This discussion will focus on the interpretation of our current modification results, as well as previous results on the phenylglyoxal modification of aldehyde reductase (ALR1 (Davidson and Flynn, 1979)) and the pyridoxal phosphate modification of ALR2 (Morjana et al., 19891, in the context of the crystal structures of ALRS. The relevant portion of the three-dimensional structure of ALR2 is shown in Fig. 7.

Chemical modification of pig muscle ALR2 by phenylglyoxal results in a loss of enzymatic activity. This inactivation exhibits pseudo-first order kinetics, but the rate constants of inactiva- tion are not linearly dependent upon the reagent concentration. The reaction does follow saturation kinetics, and such behavior is interpreted to demonstrate that phenylglyoxal binds to the enzyme reversibly, prior to covalent modification (Chang and Huang, 1981).

The stoichiometry of the reaction of ALR2 and phenylglyoxal revealed that the modification of 2 arginines results in the complete loss of enzymatic activity. Radioactive phenylglyoxal- containing peptide fragments were isolated; the modified resi- dues were identified as R268 and R293. Despite 2 residues being modified, first-order kinetics were observed for the inac- tivation of the enzyme by phenylglyoxal. While sometimes this is interpreted that modification of a single residue is respon- sible for inactivating the enzyme, it is not always possible to detect deviation from first-order kinetics even when the reac- tion rates for each group are quite different (Dixon and Webb, 1979).

The substrate DL-glyceraldehyde and the coenzyme analog NMN did not protectALR2 from modification by phenylglyoxal. However, the coenzyme analog ADPRP did afford partial pro- tection, in that the rate of inactivation slowed by a factor 3. The coenzyme NADP+ protects R268 h m modification, while at the same time making R293 hyper-reactive to phenylglyoxal.

D. W. Borhani, unpublished results.

It is not surprising that R268 of the ALR2 apoenzyme should be easily modified. This residue is solvent-exposed and acces- sible in the apoenzyme (Rondeau et al., 1992). When ALR2 binds NADP(H), R268 moves closer to and forms a salt bridge with the 2'-P04 of NADP(H) (Wilson et al., 1992; Borhani et al., 199212 This interaction would be expected to protect the resi- due from modification by phenylglyoxal, as we observe here.

At first consideration, the fact that modification of R268 should entirely eliminate enzymatic activity seems reasonable; after all, R268 is one of 2 residues which make salt bridges to the 2'-P04 of the coenzyme, the other being K262. Presumably, NADP(H) can no longer bind productively to the modified en- zyme. However, covalent modification of K262 with pyridoxal phosphate affects activity only slightly; KM (NADPH) rises 14- fold and kat rises 3.5-fold (Morjana et al., 1989). Furthermore, the crystal structures suggest that R268 would be able to tol- erate modification more easily than K262, in that it is more solvent-exposed and more to one end of the adenosine phos- phate binding site. Apparently, however, either phenylglyoxal- modified arginine is substantially more sterically demanding than is pyridoxal phosphate-modified lysine, or, less likely, R268 enjoys a signiscantly stronger interaction with the 2'- PO4 than does K262.

The inactivation of ALR2 caused by phenylglyoxal could be due to the fact that 2 residues are modified, namely R268 and R293. However, in the presence of NADP+ only R293 is modi- fied by phenylglyoxal and this singly modified enzyme retains 50% of its initial activity. Furthermore, the closely related en- zyme ALRl(51% identical, 65% similarity) lacks an arginine at position 293 (it is, rather, lysine), and yet is nonetheless modi- fied at 1 arginine by phenylglyoxal (presumably R268), and thereby inactivated (Davidson and Flynn, 1979; Branlant et al., 1981). Thus, we favor the interpretation that R268 is the pri- mary residue whose modification destroys the enzymatic activ- ity of ALR2.

Since the active site of the enzyme is over 13 A from R268, it comes as little surprise that NMN does not protect ALR2 from modification by phenylglyoxal. Also, since ALR2 has been shown to follow an ordered reaction mechanism with coenzyme

2188 Role of Arginines in Aldose Reductase

binding first to the enzyme (Grimshaw et al., 1990a; Kubiseski et al., 1992), the substrate DL-glyceraldehyde is not expected to protect ALRB from modification. The moderately protective ef- fect of ADPRP, however, reinforces the notion (Borhani et al., 1992) that this coenzyme analog sterically prevents the confor- mational change which occurs upon binding NADP(H), namely the rotation and clamping down of loop 7 over the coenzyme (Wilson et al., 1992; Borhani et al., 1992).

In the crystal structures, when loop 7 is "open" (porcine apo- enzyme; Rondeau et al., 1992), R293 is shielded from the sol- vent by its interactions with W295 and the loop 7 residues R217, P218, and W219. Upon bindingADPRP, loop 7 closes very slightly (Rondeau et al., 19921, but R293 is still sequestered. Only the complete coenzyme NADP(H) allows loop 7 to close down completely, thus revealing R293 (Wilson et al., 1992; Borhani et al., 1992L2 In solution, however, loop 7 is clearly dynamic, when it closes transiently in the apoenzyme, R293 becomes accessible to modification by phenylglyoxal, and when it opens transiently R268 is modified. The consistently greater modification of R268 over R293 in ALRB apoenzyme indicates that loop 7 is more open than closed in the apoenzyme. In the presence of ADPRP, the proportion of time that loop 7 is closed is reduced, or alternatively it cannot close to the same extent possible for the apoenzyme or the holoenzyme. Hence, by per- turbing the closure of loop 7, ADPRP slightly protects R293 from modification. Presumably, ADPRP also slightly protects R268 from modification by engaging that arginine in a salt bridge with the 2'-P04.

Why does ADPRP perturb the ability of loop 7 to close? We believe that it is energetically costly for NADPH to adopt the unusual (Borhani et al., 1992) ALR2-bound conformation. AD- PRP, which perforce lacks the four tight interactions found between ALR2 and the nicotinamide ring (Wilson et al., 1992; Borhani et al., 1992) cannot adopt the bound conformation found for NADP(H). The energetic cost of binding, which is not met by the interactions between ALRB and the terminal ribose ring, is only repaid when the enzyme-nicotinamide interactions are present as well. Since the conformation of the adenosine phosphate portion of NADP(H) bound to ALR2 is unremarkable (Borhani et al., 1992), 2',5'-ADP, which is unencumbered by the terminal ribose, is able to bind better to ALR2 than ADPRP. Indeed, although ADPRP is a good competitive inhibitor with respect to NADPH (K, ranges from 10 to 44 p ~ ) (Hyndman, 1986; Rondeau et al., 19921, 2',5'-ADP is an even better inhibi- tor (K, ranges from 0.5 to 5 p ~ ) (Hyndman, 1986; Petrash et al., 1993). Using the familiar relationship between the binding con- stant and standard free energy change of a reaction it thus appears that the terminal ribose of ADPRP destabilizes the binding ofADPRP, relative to 2',5'-ADP, by about 1.5 kcallmol.

When NADP+ is bound to ALR2, the enzyme and coenzyme conformations2 are nearly indistinguishable from those ob- served for the ALR2.NADPH complex (Wilson et al., 1992; Borhani et al., 1992). Naturally, therefore, NADP+ protects R268 from phenylglyoxal modification by engaging R268 in a salt bridge with the 2'-P04. Furthermore, since loop 7 is now in the closed position, locked over the coenzyme, R293 has lost the protection afforded by the loop 7 residues, and hence becomes hyper-reactive to phenylglyoxal.

It is noteworthy that modification of R293 by phenylglyoxal results in only a 50% loss of enzymatic activity. Again using the crystal structures as a reference point, we can easily envision that loop 7 might encounter steric resistance to opening, prior to the release of NADP+ (the rate-determining step in the for- ward reaction), if R293 is modified. This was clearly indicated by the kinetic analysis of R293 modified ALR2 in which the kcat of the reaction decreased without any significant changes in the KM values for the substrates NADPH and DL-glyceraldehyde.

In summary, we have shown that 2 arginine residues of al- dose reductase, R268 and R293, are reactive to phenylglyoxal. The coenzyme analog ADPRP affords some protection to both of these arginines, in part by sterically blocking the closure of loop 7. NADP+ protects R268, and yet it accentuates the reactivity of R293, precisely because loop 7 is now closed down over the coenzyme, which causes R293 to become solvent-exposed. Modification of R268 by phenylglyoxal, unlike modification of K262 by pyridoxal phosphate, inactivates ALR2, even though both residues are salt-bridged to the 2'-P04 of NADPH. Pre- sumably, NADPH cannot bind productively to the phenylgly- oxal-modified enzyme, due to steric hindrance. Modification of R293, which is distant from both the active site and the coen- zyme binding site, reduces activity by just 50%. The postulated mechanism for this reduction in activity, namely steric hin- drance of loop 7 motion, causing in turn even slower release of NADP', may be directly relevant to the mechanism of action of numerous potent, yet udnoncompetitive aldose reductase in- hibitors.

Acknowledgments-D. W. B. thanks Dr. Jean-Michel Rondeau (Bio- structure S.A.) and Dr. Albert0 Podjarny and Prof. Dino Moras (Univer- sity of Strasbourg) for providing the porcine aldose reductase apoen- zyme and ADPRP complex coordinates to BioCryst Pharmaceticals, Inc. We are grateful to Dianne Hyde-Kelcey for preparation of the manu- script.

REFERENCES

Bohren, K. M. & Gabbay, K. H. (1993) in Enzymology and Molecular Biology of Carbonyl Metabolism 4 (Weiner, H., Crabb, D. W., and Flynn, T. G., eds) Vol. 328, pp. 267-277, Plenum Press, New York

Bohren, K. M., Bullock, B., Wermuth, B. & Gabbay, K. H. (1989) J. Biol. Chem. 264, 9547-9551

Bohren, K. M., Page, J. L., Shankar, R., Henry, S. P. & Gabbay, K. H. (1991) J. Biol.

Borhani, D. W., Harter, T. M. & Petrash, J. M. (1992) J. Biol. Chem. 267,24841- Chem. 266,24031-24037

24847 Bradford, M. M. (1976) Anal. Biochem. 72,248-255 Branlant, G. (1982) Eur. J. Biochem. 129,99-104 Branlant, G., Tritsch, D. & Biellmann, J. F. (1981) Eur. J . Biochem. 116,505-515 Carper, D., Nishimura, C., Shinohara, T., Dietzchold. B., Wistow, G., Craft, C.,

Chang, G . G . & Huang, T:M. (1981) Biochim. Biophys. Acta 660,341-347

Davidson, W. S. & Flynn, T. G. (1979) J. Biol. Chem. 261,3724-3729 Cromlish, J. A. & Flynn, T. G. (1983) J. Biol. Chem. 268,3583-3586

Dixon, M. & Webb, E. C. (1979) Enzymes, Third Ed., pp. 374-377, Academic Press,

Doughty, C. C. & Conrad, S. M. (1982) Biochim. Biophys. Acta 708,358364 Flynn, T. G. (1986) Metab. Clin. Exp. 35, Suppl. 1, 105-108 Grimshaw, C. E., Shahbaz, M., Jahangiri, G., Putney, C. G., McKercher, S . R. &

Grimshaw, C. E., Shahbaz, M. & Putney, C. G. (199Oa)Biochemistry 29,9947-9955 Grimshaw, C. E., Shahbaz, M. & Putney, C. G. (1990b)Biochemistry 29,9936-9946

Hyndman, D. J. (1986). Chemical Modification of Pig Muscle Aldehyde Reductase Halder, A. B., James, M. & Crabbe, C. (1985) Ophthalmic Res. 17,165188

by Pyridoxnl5'-Phosphate and Other Lysine Modifiing Reagents. Master's The-

Kubiseski. T. J., Hyndman, D. J., Morjana, N. A. & Flynn, T. G. (1992) J. Biol. sis, Queen's University, Kingston, Ontario, Canada

Kador, P. & Kinoshita, J. H. (1987) FEES Lett. 220,209-213

New York

Mathur, E. J. (1989) Biochemistry 28,5343-5353

Chem. 267,651643517 Kubiseski, T. J., Green, N. C. & Flynn, T. G. (1993) in Enzymology and Molecular

Biology of Carbonyl Metabolism 4 (Weiner, H., Crabb, D. W. and F l w , T. G., eds) Vol. 328. DO. 259-265. Plenum Press, New York , ~~ ~

Liu, S., Bhatnagar, A., Das, B. & Srivastava,'S. K (1989)Arch. Biochem. Biophys.

Morjana, N. A,, Lyons, C. & Flynn, T. G. (1989) J. Biol. Chem. 264,2912-2919 Penefsky, S. H. (1971) Methods Enzymol. 56,527-529 Petrash, J. M., Harter, T. M., Devine, C. S., Olins, P. O., Bbatnagar, A., Liu, S. &

Srivastava, S. K. (1992) J. Biol. Chem. 267,24833-24840 Petrash, J. M., Harter, T., Tarle, I. & Borhani, D. W. (1993) in Enzymology and

Molecular Biology of Carbonyl Metabolism 4 (Weiner, H., Crabb, D. W., and

Rondeau, J.-M., TGte-Favier, F., Podjamy, A,, Reymann, J.-M., Barth, P., Biell- Flynn, T. G., eds) Vol. 328, pp. 289400, Plenum Press, New York

Stone, K. L., LoPresti, M. B., Crawford, J. M., DeAngelis, R. & Williams, K. R. mann, J.-F. & Moras, D. (1992) Nature 365,469-472

quencing (Matsudaira, P. T., ed) pp. 3147 , Academic Press Inc., Toronto (1989) in A Practical Guide to Protein and Peptide Purification for Microse-

I I.

275, 112.121

Takahashi, K. (1968) J. Biol. Chem. 243,61714179 Tarle, I., Borhani, D. W., Wilson, D. K, Quiocho, F. A. & Petrash, J. M. (1993) J.

Vander Jar&, D. L., Hunsaker, L. A., Robinson, B., Stangebye, L. A. & Deck, L. M. Biol. Chem. 268,25687-25693

Wilson, D. K., Bohren, K. M., Gabbay, K. H. & Quiocho, F. A. (1992) Science 257, (1990) j. Bioi. Chem. 266, 10912-10918

81-84