identification of metal-isocitrate binding site of pig heart nadp

THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1993 by The American Society for Biochemistry and Molecular Biology, Inc.

Val. 268, No. 7, Issue of March 5, pp. 5264-5271.1993 Printed in U. S. A.

Identification of Metal-Isocitrate Binding Site of Pig Heart NADP-specific Isocitrate Dehydrogenase by Affinity Cleavage of the Enzyme by Fe2+-Isocitrate”

(Received for publication, October 2, 1992)

Sambanthamurthy Soundar and Roberta F. ColmanS From the Department of Chemistry and Biochemistry, Uniuersity of Delaware, Newark, Delaware 19716

The divalent metal-isocitrate site of pig heart NADP- specific isocitrate dehydrogenase can be located by affinity cleavage of the enzyme by Fez+-isocitrate in the presence of Oz, in analogy to the “chemical nu- clease’’ action of DNA-binding drugs linked to Fe- EDTA. The enzyme is irreversibly inactivated and cleaved by Fez+-isocitrate more rapidly than by Fez+. Mnz+ prevents inactivation and cleavage by Fez+-isocitrate or by Fe2+. Furthermore, other tri- or dicarbox- ylates (such as citrate, tricarballylate, or malate), which are not effective substrates of the enzyme, fail to promote inactivation and cleavage of the enzyme by Fez+. These results indicate that the oxidative inactivation and cleavage reactions are specific. Two pairs of major peptides are generated during Fez+-isocitrate inactivation: 30 + 17 kDa and 35 + 11 kDa, as compared with 46 kDa for the intact enzyme. NHz-terminal sequencing revealed that these peptides arise by a mutually exclusive cleavage at either Asp263-Met2B4 or His30e-Gly310, suggesting Asp2” and HisSoe as coordination sites for Fez+-isocitrate and, by implication, for Mn2+-isocitrate. Fez+ alone produces peptides (32 + 15 kDa) by an alternate specific cleavage between TyrZ7’ and consistent with free metal ion occupying a different site from metal-isocitrate in NADP-dependent isocitrate dehydrogenase. Affinity cleavage may be a generally useful method for locating metal and metal- substrate sites in enzymes.

Isocitrate dehydrogenase (threo-D,-isocitrate:NADP+ oxi- doreductase (decarboxylating), EC 1.1.1.42) from pig heart mitochondria catalyzes the oxidative decarboxylation of isocitrate to a-ketoglutarate. The subunit molecular mass of the enzyme is 46,600 Da, and the complete amino acid sequence of 413 amino acids has recently been determined (1). The enzyme is a dimer of identical subunits under many conditions (2) and is not known to be allosterically regulated. It is thought that the metal-isocitrate complex is the actual substrate for this enzyme (3).

In analogy to the “chemical nucleases” (4 ) in which ferrous- EDTA is tethered to a DNA-binding drug to provide nucleotide sequence specificity for the scission of nucleic acids,

* This work was supported by United States Public Health Service Grant 1 R01 DK17522 and by a grant from the American Heart Association, Southeastern Pennsylvania Affiliate. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertise- ment” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom correspondence should be addressed Dept. of Chem- istry and Biochemistry, University of Delaware, Newark, DE 19716. Tel.: 302-831-2973: Fax: 302-831-6335.

Schepartz and Cuenoud (5) and Hoyer et al. (6) have coined the term “affinity cleavage agents” for reagents that cleave enzymes at their active sites. The prototype is the iron chelate of EDTA covalently linked to the calmodulin antagonist trifluoperazine; in the presence of Fe, 02, and dithiothreitol, the trifluoperazine analogue specifically cleaves calmodulin (5). These reagents have in common a component with specific affinity for the enzyme and a redox-active metal ion capable of oxidative cleavage of the polypeptide backbone (6). Isocitrate dehydrogenase requires a divalent metal ion for activity (3, 7). Several metals have been shown to fulfill this function, including Mn2+, M e , Zn2+, Cd2+, Co2+, and Ni2+ (8-lo), with the highest activity being obtained with Mn2+ (8). Direct binding experiments indicate that isocitrate dehydrogenase binds l mol of Mn2+/mol of enzyme subunit under all conditions (7). Analysis of the kinetics and binding experiments suggests that the metal ion occupies different sites in the absence and presence of the substrate isocitrate (3, 11). It occurred to us that the Fe2+-isocitrate complex might occupy the Mn2+-isocitrate site and satisfy the require- ments for an affinity cleavage agent.

In this paper, we demonstrate the specific cleavage of NADP-dependent isocitrate dehydrogenase by Fe2+-isocitrate in the presence of O2 and identify the substrate binding sites at which cleavage occurs. A preliminary version of this work has been presented (12).

EXPERIMENTAL PROCEDURES

Materials-NADP+ (monosodium salt), DL-isocitrate (trisodium salt), triethanolamine chloride, L-ascorbic acid, CAPS,’ tricarballylate (free acid), DL-malate (free acid), citrate (free acid), and COO- massie Brilliant Blue R were obtained from Sigma. EDTA (disodium salt) and ferrous sulfate were from Fisher Scientific. Acrylamide, bisacrylamide, and glycine were obtained from Bio-Rad. SDS was from Pierce Chemical Co. Dithiothreitol was obtained from Boehrin- ger Mannheim. Immobilon-P transblot membranes were obtained from Millipore, CM-cellulose (CM-52) was from Whatman; Matrex Gel Red A, Blue A, and Centricon-10 microconcentrator tubes were obtained from Amicon Corp.; low molecular weight standard proteins were from Pharmacia LKB Biotechnology Inc.

Determination of Activity of NADP-specific Zsocitrate Dehydrogen- ase-Enzyme assays were performed at 25 “C by monitoring the reduction of NADP+ at 340 nm. The assay solution (1 ml) contained 30 mM triethanolamine chloride buffer (pH 7.41, 0.1 mM NADP’, 4 mM isocitrate, and 2 mM MnS04. One unit of enzyme activity is defined as the amount of enzyme that catalyzes the reduction of 1 rmol of NADP+/min under the assay conditions.

Purification of Pig Heart NADP-specific Isocitrate Dehydrogen- ase-The enzyme has previously been purified in this laboratory from fresh pig hearts by the method of Bacon et al. (13) with certain modifications introduced by Smyth and Colman (14). This procedure involved chromatography on CM-cellulose, followed by chromatog-

The abbreviations used are: CAPS, 3-(cyclohexylamino)pro- panesulfonic acid; HPLC, high performance liquid chromatography.

5264

Affinity Cleavage of NADP-specific Isocitrate Dehydrogenase 5265

raphy on Matrex Gel Red A and gel filtration on Sephadex G-150- 120. Some of the preparations obtained by this method were found to contain minor amounts of mitochondrial malate dehydrogenase as well as an unidentified protein of 43,000 Da, which is very close to the subunit molecular weight of the NADP-dependent isocitrate dehydrogenase. Since the work described here involves cleavage of enzyme into fragments, it is essential for the intact enzyme to be free from any detectable impurity. Accordingly an improved purification scheme has been developed; the CM-cellulose step is conducted as described by Bacon et al. (13), followed by the Matrex Gel Red A column as reported by Smyth and Colman (14), after which an additional chromatography step on Matrex Gel Blue A was substi- tuted for the gel filtration step.

After NH4S04 (80%) precipitation, a crude extract containing 22,000 enzyme units with a specific activity of 1 unit/mg was applied to a CM-cellulose column, as described (13). The pooled fractions from the CM-cellulose column, containing 9,500 units with a specific activity of 13 units/mg, were then applied to a Matrex Gel Red A column. Fractions eluted from the procedure previously reported (14), which exhibited a specific activity of 34 units/mg (3,000 units), were pooled and concentrated in an Amicon ultrafiltration cell, fitted with a PM-10 membrane, to a protein concentration of 3 mg/ml.

The concentrated enzyme was dialyzed against 18 mM triethanolamine chloride, pH 7.1, containing 10% glycerol and 20 mM Na2S04 (Buffer A) and applied to a Matrex Gel Blue A column (2 X 33 cm) which had been previously equilibrated with Buffer A. (Before use, the resin was treated with 8 M urea in 0.5 M NaOH and washed to neutrality with distilled water.) The impure proteins were separated by elution at a flow rate of 30 ml/h with a linear gradient formed by Buffer A (100 ml) and 100 ml of 18 mM triethanolamine chloride, pH 7.4, containing 10% glycerol and 0.13 M Na2S04. When the enzyme began to emerge from the column, the eluent was changed to 18 mM triethanolamine chloride, pH 8.0, 10% glycerol, and 0.4 M Na2S04. Isocitrate dehydrogenase was eluted as a single peak, and aliquots from fractions were subjected to SDS-gel electrophoresis as described below. Fractions yielding a single protein band, without any traces of impurity, were pooled and stored frozen at -80 "C. The final enzyme preparation exhibited a specific activity of 44 units/mg, with a recov- ery greater than 75% of the protein applied. The use of the Matrex Blue A resin removes all the minor impurities which were sometimes detected in the previous preparations (13, 14). The protein concentration was determined from = 10.8 (15), and a subunit M, of 46,600 (1) was used to calculate the concentration of enzyme subunits.

Inactivation and Cleavage of NADP-specific Isocitrate Dehydrogen- ase by Fez' or Fez+ and Isocitrate-Isocitrate dehydrogenase (22 p ~ ) was incubated with 20 p~ ferrous sulfate and 20 mM ascorbate with or without the addition of 4 mM isocitrate in a 0.1 M triethanolamine chloride buffer, pH 7.7, containing 10% glycerol and 0.3 M Na2S04 at 25 "C for time periods up to 3 h. Aliquots from the reaction mixture were assayed for residual enzyme activity at various times. Aliquots were screened by electrophoresis in polyacrylamide gels containing SDS to detect the presence of fragments generated by cleavage of the enzyme. EDTA (4 mM) was added directly to the incubation mixture in order to quench the inactivation of enzyme caused by Fez+. For determination of the time-dependent cleavage of enzyme, EDTA (4 mM) was added to aliquots of the reaction mixture prior to gel electrophoresis. The concentration of reactants and conditions for inactivation and cleavage of enzyme were maintained the same in all experiments, unless specified otherwise.

Electrophoresis in Polyacrylamide Gels Containing SDS-The cleaved fragments from enzyme were analyzed in a 15% polyacrylamide slab gel (Laemmli system) containing 0.1% SDS in a discon- tinuous pH electrophoresis system (16). A 5% stacking gel was cast from a solution which contained 0.13 M Tris, pH 6.8, buffer and a resolving gel in a 0.38 M Tris, pH 8.8, buffer. The electrophoresis was carried out on a Mini-Protean unit (Bio-Rad) in a buffer containing Tris (6 g), glycine (28.8 g), and 5 ml of 20% SDS per liter, pH 8.4, at 4 "C and a constant voltage (100 V) for 2 h.

Proteins in the gel were stained with 0.2% Coomassie Brilliant Blue after fixing the proteins in 10% methanol and 15% acetic acid overnight and were destained in 12% ethanol and 7% acetic acid. For determination of the molecular weight and relative distribution of protein produced by cleavage of enzyme, the samples were subjected to electrophoresis in tube gels under conditions similar to those described for slab gel electrophoresis. The protein bands were scanned at 590 nm in a Gilford 240 spectrophotometer using a Gilford model 2410s linear transport accessory. The distribution of proteins was determined from the relative peak areas given by fragments and

control enzyme. The molecular weight of fragments was determined from the relative migration of protein bands as compared to standard proteins.

Isolation of Protein Fragments-Inactivated enzyme (1 mg/ml) was concentrated in Amicon microconcentrator tubes to 0.3 ml and was maintained for 5 min at 100 "C after the addition of 30 pl of 5% SDS. An equal volume of sample buffer containing 60 mM Tris buffer, pH 6.8, 100 mM dithiothreitol, and 30% glycerol was added and the proteins were separated on a 13% polyacrylamide slab gel as described above.

The slab gel was soaked in "transfer buffer" (10 mM CAPS, pH 11) containing 10% methanol for 5 min. The gel and a presoaked Immobilon-P membrane (polyvinylidene difluoride) were inserted between filter papers. Before use, the membranes were rinsed in 100% methanol for 3 min and equilibrated by soaking in transfer buffer for 5 min. The electroblotting was conducted in a blotting cassette (Bio- Rad) at a constant 150-mA current for 18 h at 4 "C with protein transfer from the gel estimated at greater than 95%. After transfer, the membrane was rinsed several times in HPLC-grade water and air-dried. A strip of the transblot was cut, and the protein bands were visualized by staining with 0.1% Coomassie Brilliant Blue in 50% methanol and destained with 50% methanol and 10% acetic acid. The sections containing the unstained protein bands in the membranes were separated by comparison with the stained bands. The membranes containing the transblotted protein were stored at -80 "C.

Treatment of Inactivated Protein with CNBr-Transblot membranes containing the fragments F2 and F, were cut and rinsed in 95% acetone for 5 min. For each milligram of protein, 1 ml of 7% CNBr in 70% formic acid was added, and this mixture was incubated at room temperature in the dark for 24 h with intermittent mixing. The CNBr-treated proteins were eluted from the transblot paper with 0.1% trifluoroacetic acid in 40% CH&N at 50 "C for 30 min; this elution procedure was repeated 3-4 times. The eluate from the F2 fragment was lyophilized and dissolved in 1 ml of 0.1% trifluoroacetic acid containing 30% acetonitrile and separated by HPLC on a C4 column (Vydac) using a gradient elution from 30 to 60% CH&N in 0.1% trifluoroacetic acid for 3 h at a flow rate of 1 ml/min, with the eluate monitored at 220 nm. The separated CNBr peptides were subjected to sequential analysis by Edman degradation. In the case of fragment Fs, the whole CNBr digest was subjected to amino- terminal sequential analysis.

NH2-terminal Sequential Analysis-Automated sequence analysis was performed on an Applied Biosystems gas phase protein sequenator (model 470A), equipped with an on-line phenylthiohydantoin analyzer (model 120) and computer (model 900A). The transblot papers containing 3-6 nmol of protein fragments were directly sequenced by Edman degradation.

RESULTS

Inactivation of NADP-specific Isocitrate Dehydrogenase by Fez+ and Fe'+-Isocitrate-Addition of Fez+ (20 PM) in the presence of the reducing agent ascorbate (20 mM) causes an irreversible inactivation of NADP-dependent isocitrate dehydrogenase (22 PM) upon incubation at 25 "c in 0.1 M triethanolamine chloride buffer, pH 7.7, containing 10% glycerol and 0.3 M NaZSO4. Fig. 1 (line a) shows the time dependence of this loss of activity. The rate of inactivation by Fez+ is much more rapid when isocitrate is included in the incubation mixture (Fig. 1, line b), consistent with a distinct mode of binding by the enzyme of free metal ion as compared with metal-isocitrate complex.

Table I demonstrates that ascorbate is required for inactivation by added Fez+ when isocitrate is either absent (compare lines 3 and 2), or present (compare lines 7 and 6). An electron donor (such as ascorbate) has previously been shown to be required, along with oxygen and Fe2+/Fe3+ to promote metal- catalyzed oxidation of proteins (17, 18). Fenton chemistry is involved, in which reactive oxygen species are generated lo- cally at the metal binding site as the metal ion is oxidized and a reductant such as ascorbate is used to recycle the oxidized metal ion (5, 19). Inclusion of Mn2+ totally prevents inactivation of isocitrate dehydrogenase by Fez+ (compare lines 3 and 5) or by Fe2+-isocitrate (compare lines 7 and 9). Among

5266 Affinity Cleavage of NADP-specific Isocitrate Dehydrogenase

- 80

t L= 5 60 0 4 -I 2 40 E

a 20

r n W

0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

TIME (h)

FIG. 1. Time dependence of inactivation of NADP-dependent isocitrate dehydrogenase by Fez+ or Fez+-isocitrate. Incu- bation of NADP-isocitrate dehydrogenase (22 p M ) with ascorbate (20 mM) at 25 “C in 0.1 M triethanolamine chloride, pH 7.7, buffer containing 10% glycerol and 0.3 M NazSOl in the presence of 20 pM FeS04 (a) or (20 p ~ ) FeS04 + (4 mM) isocitrate (b). A t the times indicated, aliquots were removed, diluted about 5,000-fold, and assayed for activity under the standard conditions described under “Experimental Procedures.”

TABLE I Effect of addition of ascorbate and Mn2+ on inactivation of isocitrate

dehydrogenase by Fez+ or Fe2+-isocitrate Isocitrate dehydrogenase (22 p ~ ) was incubated at 25 “C for 3 h in

0.1 M triethanolamine chloride buffer, pH 7.7, containing 10% glycerol and 0.3 M NazS04, with additions as indicated. After 3 h, aliquots were removed, diluted about 5,000-fold, and assayed as described under “Experimental Procedures.”

Addition Residual activity

1. None 2. Fez+ (20 pM) 3. Fez+ (20 p ~ ) + ascorbate (20 mM) 4. Mn2+ (2 mM) + ascorbate (20 mM) 5. Fez+ (20 p ~ ) + Mn2+ (2 mM) + ascorbate

6. Isocitrate (4 mM) + Fez+ (20 p M ) 7. Isocitrate (4 mM) + Fe2+ (20 p M ) + as-

8. Isocitrate (4 mM) + Mn2+ (2 mM) + as-

9. Isocitrate (4 mM) + Fez+ (20 pM) + Mn2+

(20 mM)

corbate (20 mM)

corbate (20 mM)

(2 mM) + ascorbate (20 mM)

% 100 103 61 98

101

97 2

98

104

the divalent metal ions tested, Mn2+ has been reported to yield the highest rate in the overall reaction catalyzed by isocitrate dehydrogenase (8). The results of Table I thus suggest that inactivation by Fez+ involves specific interaction at the natural metal site of the enzyme.

Fez+ has not previously been tested for its ability to activate the enzyme-catalyzed oxidative decarboxylation of isocitrate dehydrogenase. We have now measured the enzymatic rates under the standard conditions described under “Experimental Procedures” except for the inclusion of 0.05 mM EDTA (to chelate any adventitious metal ion), of 1.2 mM ascorbate (to maintain ferrous ion in its reduced state) and of either 1 mM MnS04 or 1 mM FeS04. When the assay was started by the addition of enzyme, the initial velocities (within the first 30 s) were comparable with Mn2+ or Fez+; thereafter, the velocity in the presence of Fez+ decreased, presumably due to the time- dependent metal-catalyzed oxidation of isocitrate dehydrogenase. We conclude that Fez+ can fulfill the metal require-

ment for the conversion of isocitrate to a-ketoglutarate, which is normally catalyzed by the enzyme.

Cleavage of NADP-specific Isocitrate Dehydrogenase by Fez+ and Fe2+-Isocitrate-In addition to inactivation, incubation of isocitrate dehydrogenase with Fe2+ or Fez+-isocitrate causes cleavage of the enzyme subunit. Fig. 2 illustrates the fragments generated by limited cleavage of the enzyme subsequent to inactivation, as visualized by electrophoresis on polyacrylamide gels containing SDS. After incubation of enzyme for 3 h in the presence of Fez+ alone (lane Z ) , one pair of predominant fragments is observed in addition to the intact enzyme subunit: F2 + Fs. In contrast, after incubation of enzyme for 3 h in the presence of Fez+-isocitrate (lane 1 ), two distinguishable pairs of major fragments are seen: F1 + Fs, and FJ + F4; smaller amounts of a protein doublet close to the position of Fz, as well as a band close to the position of Fs, are also noted, which may be due to some cleavage by the residual free Fez+.

The molecular weights and the relative distribution of protein resulting from enzyme cleavage are summarized in Table 11. About 60-70% of the enzyme inactivated by Fez+ or Fe2+-isocitrate is cleaved into smaller fragments. The molecular weights of the two predominant fragments produced in the presence of free Fez+ indicates that they derive from a single cleavage of the intact enzyme subunit: Fa (32,000) + FS (15,000) z 46,000. Smaller polypeptides are stained less well by Coomassie Blue than are larger polypeptides, probably accounting for the quantitative inequality in the staining of Fz and Fs. In contrast, the molecular weights of the four major bands observed after inactivation by Fe2+-isocitrate are best

5 f 94,000 1 + 67,000

Enzyme ”*

Fl ”*

F3 -@

F4 ”*

F6 -@

t- 43,000

I- 30,000

0- 20,100

I- 14,400

FIG. 2. SDS-polyacrylamide gel electrophoresis of isocitrate dehydrogenase after inactivation by Fez+ or Fez+-isocitrate. Enzyme (22 p ~ ) was incubated with ascorbate (20 mM) at 25 “c for 3 h in the presence of 20 p M FeS04 + 4 mM isocitrate (lane 1 ) or 20 p~ FeS04 (lane 2). Proteins of standard molecular weight are shown in lane 3 phosphorylase b (94,000), albumin (67,000), ovalbu- min (43,000), carbonic anhydrase (30,000), trypsin inhibitor (20,100), and n-lactalbumin (14,400). Electrophoresis was conducted as described under “Experimental Procedures.”


TABLE I1 Molecular weight and relative amount of fragments produced upon

cleavage of isocitrate dehydrogenase by Fe2+-isocitrate or Fez+ The molecular weights and relative amounts of protein fragments

were measured from scans of tube gels stained with Coomassie Blue as described under "Experimental Procedures." The protein samples used were similar to those used in Fie. 2.

Fez+-isocitrate Fez+ Band designation Mr Relative amount

of protein Mr Relative amount

of protein % %

Enzyme 46,000 30 46,000 40 F1 35,000 23 Fz 32,000 25 F3 30,000 12 F4 17,000 6 F5 15,000 16 Fs 11,000 15

100 3

i 0 " " " " " " ' ~ 0.0 0.5 1.0 1.5 2.0 2.5 3.0

TIME (h)

FIG. 3. Quenching by EDTA of inactivation of isocitrate dehydrogenase by Fez+ or Fe2+-isocitrate. Enzyme was incubated with either Fez+ (a) or Fez+-isocitrate ( b ) as described in Fig. 1. At the time indicated by the arrows, 4 mM EDTA was added to the incubation mixtures and measurements of the enzymatic activity of aliquots were continued as a function of time.

explained by two alternative (i.e. mutually exclusive) cleav- ages of the intact subunit: F1 (35,000) + F6 (11,000) E 46,000, or F3 (30,000) + F4 (17,000) 2 46,000. Cleavage to produce the F1 + F6 fragments occurs more frequently.

Time-dependent Cleavage of Isocitrate Dehydrogenase by Fez+-Isocitrate-In order to examine the cleavage of the enzyme as a function of time of incubation'with Fez+ or Fez+- isocitrate, it is necessary to stop the inactivation reaction at a defined time. Fig. 3 demonstrates that the addition of EDTA to an incubation mixture of enzyme with either Fez+ alone (line a) or Fez+-isocitrate (line b ) rapidly quenches the metal- catalyzed oxidative inactivation.

The time-dependent cleavage of enzyme by Fez+-isocitrate is shown in Fig. 4. (EDTA (4 mM) was added to the reaction mixture in order to stop the inactivation and cleavage at each time indicated.) It appears that the cleavage reaction occurs subsequent to the inactivation, since the intact enzyme subunit (Fig. 4, top protein band) is lost more slowly than is the enzymatic activity (Fig. lb ) .

Evaluation of Specificity of Oxidative Cleavage of Isocitrate Dehydrogenase-A critical test that must be met by an affinity reagent is competition by a natural ligand. In the case of affinity cleavage, this criterion would be met by the observa- tion of protection against cleavage being provided by a natural metal ion which does not cause oxidative inactivation. As

shown in Table I, Mn2+ fails to promote oxidative inactivation of isocitrate dehydrogenase and it protects the enzyme against the inactivation produced by Fe2+ or Fez+-isocitrate. Fig. 5 shows that ascorbate is required for cleavage of the enzyme by Fe2+ alone (compare lanes 3 and 5 ) or by Fe2+-isocitrate (compare lanes 7 and 9). However, even in the presence of ascorbate, Mn'+ (lane 2 ) or Mn'+-isocitrate (lane 6) does not cleave the enzyme. Furthermore, the addition of Mn2+ also prevents cleavage by Fez+ alone (compare lanes 3 and 4 ) or by Fez+-isocitrate (compare lanes 7 and 8). These results indicate that the enzyme is specifically cleaved at its metal binding site.

The specificity for the substrate in the affinity cleavage of isocitrate dehydrogenase was evaluated by comparing the effect of the addition of isocitrate on the inactivation and cleavage caused by Fe2+ to that of the addition of other di- and tricarboxylic acids. The results illustrated in Fig. 6 indicate that only in the presence of isocitrate (lane 8) is enzyme activity almost completely lost during the 3-h incubation period. Moreover, cleavage by Fez+ with isocitrate (lane 8) generates the unique fragments (F1, F3, F4 and F6), not observed with other di- and tricarboxylates, such as citrate (lane 5 ) , tricarballylate (lane 6), and malate (lane 7), which are not effective substrates of the pig heart NADP-isocitrate dehydrogenase.' The enzyme fragments observed in lunes 5-7 appear to be similar to those produced by Fe2+ alone as shown in Fig. 6 (lane 3 ) . (The appearance in small amounts of fragments in lane 2 is likely due to cleavage of enzyme by trace amounts of iron contaminants; this cleavage is completely prevented by the addition of EDTA (lane 4 ) . ) These results indicate that the enzyme is specifically cleaved at the metal-isocitrate binding site.

Identification of Major Fragments Cleaved from Isocitrate Dehydrogenase by Fez+-Isocitrate-After separation by electrophoresis in polyacrylamide gels containing SDS of the fragments obtained from cleavage by Fe2+ (as in Fig. 2), the protein bands from the gel were transferred electrophoreti- cally onto Immobilon membranes. The amino acid sequence of each fragment was determined by applying the Immobilon strips directly to the gas phase sequenator. Table I11 shows the amino-terminal sequences obtained by automated Edman degradation of the major fragments, F1, FS, F4, and F6. The sequences of F1 and F3 are identical with the amino-terminal sequence of intact pig heart NADP-dependent isocitrate dehydrogenase (14). Fragment F4 has the sequence identified as Met254 to Gln277, whereas Fragment Fs exhibits the sequence

The molecular weights of F1 and F6 estimated by electrophoresis (Table 11) together with these sequences (Table 111) indicate that Fl consists of Ala' through His3", with a calculated molecular weight of 35,013, whereas F6 consists of Gly3'' through Gln413, with a calculated molecular weight of 11,634. The results suggest that F1 and F6 arise from a single cleavage of intact enzyme between His3'' and Gly310.

The molecular weights of F3 and F4 estimated by electrophoresis (Table 11) together with their sequences shown in Table I11 indicate that F3 consists of Ala' to with a calculated molecular weight of 29,215; and F4 consists of

A pig heart isocitrate dehydrogenase solution, which catalyzed the formation of 103 pmol of NADPH/min/ml of enzyme when measured under standard assay conditions with 4 mM isocitrate as substrate, yielded no detectable NADPH when either 4 or 40 mM malate was tested as an alternate substrate. Although malate has been described as a substrate of the E. coli NADP-dependent isocitrate dehydrogenase, the V,,, for isocitrate was reported as more than 10' times that for malate, and the K,,, for isocitrate was given as less than that of malate (20).

~ 1 ~ ~ 1 0 to pros2' (1).


0 2 4 6 8 10 20 60 120 180min

FIG. 4. Time-dependent cleavage of isocitrate dehydrogenase by Fez+- isocitrate as shown by SDS-polyacrylamide gel electrophoresis. En- zyme (22 pM) was incubated with ascorbate (20 mM) at 25 "C in the presence of Fez+ (20 p M ) and isocitrate (4 mM). EDTA (4 mM) was added to the samples (prior to electrophoresis) at the times indicated above the lanes.

1 2 3 4 5 6 7 8 9 10 kDa

-94 -67

c43

FIG. 5. Effect of Mna+ on cleavage of isacitrate dehydrogenase by Fez+ and Fe**-isocitrate as shown by SDS-polyacrylamide gel electrophoresis. Enzyme (22 PM) was incubated at 25 "C for 3 h with the following additions: lane I , none; lane 2, Mn2+ + ascorbate; lane 3, Fe2+ + ascorbate; lane 4, Fe2+ + ascorbate + Mn2+; lane 5, Fe2+; lane 6, Mn2+ + isocitrate + ascorbate; lane 7, Fez+ + isocitrate + ascorbate; lane 8, Fez+ + isocitrate + ascorbate + Mn2+; lane 9, Fe2+ + isocitrate. Standard proteins are shown in lane 10. The concentrations of additions were as follows: FeS04, 20 pM; MnS04, 2 mM; isocitrate, 4 mM; ascorbate, 20 mM.

Met254 to Gln4I3, with a calculated molecular weight of 17,432. These results suggest that F3 and F4 arise from a single cleavage of intact enzyme between Asp253 and Met254.

Identification of Major Fragments Cleaved from Isocitrate Dehydrogenase by Fe2+ Alone-The major fragments generated by Fez+ cleavage, Fz, and F5, were also isolated by elec- trophoretic transfer to Immobilon membranes. Upon sequential analysis by Edman degradation, Fz exhibited clearly the sequence of the amino-terminal part of the intact protein (Table IV, column 1). F, yielded no phenylthiohydantoin derivatives, indicating that the amino terminus had appar- ently become blocked during the oxidative cleavage or the subsequent isolation.

In an alternate approach to identify the cleavage site of isocitrate dehydrogenase by Fez+, Fz and F5 were each treated with CNBr to chemically fragment the enzyme a t its 9 me- thionine residues. Met254, Met291, and Met372 are the 3 methi- onines closest to the COOH-terminal end of the protein (1).

The cyanogen bromide digest of F5 yielded two CNBr peptides containing the amino-terminal sequences ThrZg2 to Va1312 and Thr373 to as shown in Table IV, columns 3 and 4. Thus, fragment F5 contains CNBr peptides 9 and 10 of the intact enzyme (which constitute the sequences Thgg2 to Met372 and Thr373 to Gln413, respectively), indicating that it is the COOH- terminal product of the cleavage of the enzyme by Fe2+. It was concluded that the site of cleavage by Fe2+ was within the region spanned by CNBr peptide 8, i.e. residues 255-291.

In order to further localize the cleavage site, the CNBr peptides derived from the F2 fragment were separated by high performance liquid chromatography on a C4 column using 0.1% trifluoroacetic acid as solvent with an acetonitrile gradient. The peptide closest to the COOH terminus of the F:! CNBr digest was a portion of CNBr peptide 8 Val255 to Tyr272, as shown in Table IV, column 2. Detection of phenylthiohydantoin derivatives did not continue after Tyr272, indicating an incomplete peptide as compared to the original CNBr


1 2 3 4 5 6 7 8 9 kDa

'4-94 1+67

14-30

Activity (%) 100 82 56 99 84 43 69 3

FIG. 6. Effect of addition of 4 mM di- and tricarboxylic acids on inactivation and cleavage of enzyme by Fez+ (20 pM). Incubation mixtures all contained 22 p~ enzyme. The samples shown in lanes 2-8 also contained 20 mM ascorbate. The samples were maintained at 25 "C for 3 h with the addition of: lane 3, Fez+; lane 4, Fez+ + 4 mM EDTA; lane 5, Fez+ + citrate; lane 6, Fez+ + tricarballylate; lane 7, Fez+ + malate; lane 8, Fez+ + isocitrate. Standard proteins are shown in lane 9. The residual enzymatic activity shown below the gels is that measured after 3 h of incubation, immediately prior to electrophoresis.

TABLE 111 N-terminal sequences of fragments obtained from oxidative cleavage

of isocitrate dehydrogenase by Fez+-isocitrate

Cycle Amino acid (amount in pmol)

Frament F, Frament F, Frament F. Frament Fa 1 Ala (152) Ala (64) Met (79) Gly (66) 2 Asp (67) Asp (28) Val (62) Thr (32) 3 Gln (64) Gln (33) Ala (71) Val (34) 4 Arg (22) Arg (11) Gln (48) Thr (33) 5 Ile (116) Ile (54) Val (59) Arg (27) 6 Lys (79) Lys (42) Leu (74) His" 7 Val (119) Val (50) Lys (38) Tyr (24) 8 Ala (131) Ala (56) Ser (16) Arg (38) 9 Lys (87) Lys (46) Ser (13) Glu (26)

10 Pro (71) Pro (34) Gly (38) His" 11 Val (111) Val (44) Gly (40) Gln (21) 12 Val (118) Val (52) Phe (34) Lys (26) 13 Glu (54) Glu (21) Val (37) Gly (46) 14 Met (70) Met (30) Trp" Arg (55) 15 Asp (67) Asp (23) Ala (44) Pro (20) 16 Gly (79) Gly (34) Cysb Thr (29) 17 Asp (66) Asp (26) Lys (21) Ser (10) 18 Glu (44) Glu' (16) Asn (23) Thr (30) 19 Met' (48) Tyr (18) Asn (24) 20 Asp (19) Pro' (11) 21 Gly (29) 22 ASP (22) 23 Val (20) 24 Gln' (17)

a Expected from the known sequence of the enzyme (1,14) but not determined conclusively because of the low yields of this amino acid.

Expected from the known sequence of the enzyme (1,14) but not detected because neither the enzyme nor the fragments were modified to block free cysteines (e.g. with iodoacetate or N-ethylmaleimide) prior to sequencing.

The peptide was not sequenced beyond this point.

peptide 8. These results, together with the estimated molecular weights from polyacrylamide gel electrophoresis (Table 11), indicate that F2 consists of Ala' to Tyr2I2, with a calculated molecular weight of 31,285, whereas Fs consists of Asp273 to Gln413, with a calculated molecular weight of 15,362. It thus appears that F2 and Fs arise from affinity cleavage of isocitrate dehydrogenase at the Tyr2I2 to Asp2I3 peptide bond.

TABLE IV N-terminal sequences of fragment F2 and CNBr peptides isolated

from fragments Fz and Fs obtained from oxidative cleavage of isocitrate dehvdrogenase bv Fez+

Amino acid (amount in pmol)

Cycle CNBr peptides isolated from:

number Fragment FP Fragment FP Fragment F5 8 9 10

1 Ala (577) Val (60) Thr (210) Thr (210) 2 Asp (244) Ala (70) Ser (85) Lys (418) 3 Gln (318) Gln (72) Val (293) Asp' 4 Arg (60) Val (37) Leu (833) Leu (833) 5 Ile (377) Leu (49) Val (281) Ala (434) 6 Lys (386) Lys (69) Cysb Gly (329) 7 Val (436) Ser (45) Pro (156) Cysb 8 Ala (502) Ser (56) Asp (130) Ile (338) 9 Lys (393) Gly (61) Gly (248) His (66)

10 Pro (277) Gly (42) Lys (185) Gly (345) 11 Val (343) Phe (37) Thr (73) Leu (337) 12 Val (438) Val (48) Ile (123) Ser (55) 13 Glu (182) Trp" Glu (141) Asn (203) 14 Met (273) Ala (27) Ala (155) Val (172) 15 Asp (200) Cysb Glu (131) Lys (172) 16 Gly (194) Lys (29) Ala (165) Leu (269) 17 Aspd (210) Asn (10) Ala (208) Asn (162) 18 T y f (30) His (36) Glu (137) 19 Gly (94) His (48) 20 Thr (42) Phe (135) 21 Vald (67) Leud (174)

a Expected from the known sequence of the enzyme (1,14) but not determined conclusively because of the low yields of this amino acid.

Expected from the known sequence of the enzyme (1,14) but not detected because neither the enzyme nor the fragments were modified to block free cysteines (e.g. with iodoacetate or N-ethylmaleimide) prior to sequencing.

Interference prevented determination of this residue. The peptide was not sequenced beyond this point.

e No amino acids were detected beyond the tyrosine.

DISCUSSION

The pig heart NADP-dependent isocitrate dehydrogenase has been found to bind 1 mol of manganous ion and 1 mol of isocitrate/mol of enzyme subunit, under all conditions (3, l l ) . The dissociation constant for Mn2+ is considerably decreased


in the presence of isocitrate, but no more than 1 mol of metal ion/enzyme site is occupied (7). On the basis of marked differences in the pH dependence of binding of free manganous ion, free isocitrate, and manganous-isocitrate complex, it was postulated that metal ion binds to distinguishable but mutually exclusive sites in the absence and presence of isocitrate (11). A distinction between the location of the binding sites of free metal ion and metal-isocitrate was also indicated by resonance energy transfer studies involving the effect of the divalent metal activators NiZ+ and Co2+ on isocitrate dehydrogenase covalently labeled by a fluorescent nucleotide analogue (9). Only in the presence of isocitrate did Ni2+ or Co2+ quench the fluorescence, yielding different estimates of the distance between the fluorescent probe and metal-isocitrate or free metal ion (9). Although the entire amino acid sequence of the pig heart NADP-dependent isocitrate dehydrogenase is now known (l) , these metal sites have not previously been identified.

The affinity cleavage experiments presented in this paper localize the metal sites and provide structural evidence that the enzyme ligands of free metal ion and metal-isocitrate are different. Ferrous ion has been used as a probe of the divalent metal sites of this enzyme. Since Fez+ can function as an activator of the oxidative decarboxylation of isocitrate catalyzed by isocitrate dehydrogenase, and Mn2+ protects against both the inactivation and cleavage reactions promoted by Fez+, it is reasonable to conclude that Fez+ binds to the same sites as Mn2+. The specificity of the reaction of Fez+ with isocitrate dehydrogenase is demonstrated by the discrete fragments observed upon SDS-polyacrylamide gel electrophoresis and by the ability of Mn2+ to prevent the fragmentation and the loss of activity. Moreover, the substrate isocitrate has a special effect in enhancing the Fez+ inactivation and changing the fragmentation pattern; other di- and tricarboxylates (such As citrate, tricarballylate, or malate), which can form metal chelates but are not effective substrates, do not promote the inactivation, and the enzyme fragments generated in their presence are similar to those produced by Fez+ alone.

The distinctive fragmentation patterns produced by affinity cleavage of isocitrate dehydrogenase by Fez+-isocitrate and by Fez+ alone are shown schematically in Fig. 7, as deduced from the molecular weight and amino acid sequence data. As illustrated in Fig. 7a, the results of cleavage by Fez+ in the presence of isocitrate are best understood in terms of a major cleavage

(b) F& c- F, ( 3 2 , O O O ) y

Ala’UAsp” MetzYulyr”’ Asp”’ ~et’’$+Leu’~’ 413 1“ F, (15,000) --, HZN - bCOOH

y8 in?+ MeIN$Val’‘2 ‘lo

FIG. 7. Scheme of affinity cleavage of pig heart NADP- dependent isocitrate dehydrogenase by Fez+-isocitrate (a) and Fez+ (6) . The sequences which have been determined are shaded. In a, the determined sequences are those shown in Table 111. In b, the determined sequences are those shown in Table IV.

of the H i ~ ~ ” - G l y ~ ~ ~ peptide bond to yield F1 + F6, with a minor cleavage at the peptide bond to produce F3 + Fq. Histidine has been shown previously to be a target of metal-catalyzed oxidation (17-19, 21, 22), and it has been proposed that modification of histidine may lead to scission of the histidyl peptide bond (19, 22). Aspartic acid has not been reported to be particularly susceptible to metal-catalyzed cleavage. However, Hoyer et al. (6) suggested that oxidative cleavage of any peptide bond may involve initial oxidation of the a-carbon to the a-hydroxylatedproduct. The only require- ment would be that the reactive oxygen species must react with a target close to the metal binding site where it is generated. For isocitrate dehydrogenase, the amino acids ad- jacent to the cleavage sites of F1-Fs and F3-F4 are therefore considered to be at or near the metal-isocitrate binding site. It is notable that is reasonably close in sequence to CysZ6’, whose modification (which can be prevented by Mn2+- isocitrate) is the major cause of inactivation by N-ethylmaleimide (14); CysZ6’ is postulated to be located at or near the metal-isocitrate binding site.

The results of the affinity cleavage experiments suggest that Asp253 and His309 may be coordination sites for metal ion in Mn2+-isocitrate, as pictured in Fig. 7a. The NADP-dependent isocitrate dehydrogenase is a dimer of identical subunits. Each metal ion may either be shared between two subunits with one ligand contributed by each subunit (as seems to be the situation for the Escherichia coli isocitrate dehydrogenase; Ref. 23) or both coordination sites may be located within a single subunit. In either case, after an oxidative cleavage occurs in the vicinity of one of the ligand sites of Fez+- isocitrate, binding of metal-isocitrate to that subunit will be disrupted and no further cleavage will take place. Thus, cleavage at or His309-Gly310 would be expected to be alternative reactions within a given subunit.

Analysis of electron paramagnetic resonance spectra indicates that isocitrate is directly coordinated to enzyme-bound manganese (24). Relaxation measurements of water protons indicate that displacement of one of the two water ligands occurs when the enzyme-Mn2+-isocitrate complex is formed from the enzyme-Mn2+ complex (25). In the ternary enzyme- metal-isocitrate complex, the metal ion is likely to be coordinated to the oxygen of one water molecule, and to 1-2 oxygens of isocitrate (probably those of the @-carboxyl group since the 13C resonance of this group is markedly perturbed by the binding of magnesium in the enzyme-Mg2+-isocitrate complex; Ref. 26). The results of the present paper indicate that Asp253 and His3” supply additional ligands to the metal ion in the ternary complex. A hexacoordinate complex is typical of enzymes which use Mn2+ or Mg+. Measurement of the ‘13Cd NMR spectrum of the isocitrate dehydrogenase-Cd2+- isocitrate complex yielded a peak at 9 ppm (27) which was considered to be characteristic of cadmium bound to six oxygen-containing ligands (28). However, the chemical shift of enzyme-bound cadmium represented a 32-ppm downfield shift from the -23 ppm measured for Cd2+-isocitrate in the absence of enzyme. Thus, the observed l13Cd chemical shifts (27) may also be consistent with replacement of one of the six oxygen ligands of the Cd2+-isocitrate complex by one nitrogen ligand in the isocitrate dehydrogenase-Cdz+-isocitrate complex, in analogy to the change reported in the complex of phosphoglucomutase-Cd2+ upon binding substrate (29).

Fig. 7b indicates that affinity cleavage of isocitrate dehydrogenase by Fez+ alone produces the discrete polypeptide products Fa and F5 that result from the cleavage of the Tyr272-

peptide bond. Metal-catalyzed oxidative modification of tyrosine has been reported (19); however, the cleavage may


Pig heart D 2 9 8 - G - K - T - I - E - A - E - A - A - H 3 0 9 - G - T - V - T - R 3 1 4 Yeast D 3 0 0 - G - K - T - F - E - S - E - A - A - H 3 1 0 - G - T - V - T - R 3 1 s E. coli E33' -C-A-L- F - E - A - - T - H 3 3 9 - G - T - A - p - K 3 4 4

occur subsequent to the oxidation of the a-carbon of either TyrZ7' or (6). The enzyme ligand to the metal ion in the binary enzyme-metal complex is likely to be the carbox- ylate of Asp273, although coordination to the -OH of TyrZ7' cannot be excluded. Clearly, the enzyme binding site for free metal ion is close to, but distinguishable from, that for metal- isocitrate.

The only isocitrate dehydrogenase for which the crystal structure has been analyzed is the E. coli NADP-specific enzyme; structures of the enzyme-Me-isocitrate and enzyme-NADP complexes have been described (23). The active site Mg2' in the ternary complex of the E. coli enzyme is visualized as coordinated to six oxygen ligands, two of which are supplied by the side chains of Aspm7 and Asp283' (from the second subunit). The amino acid sequence of the pig heart NADP-dependent isocitrate dehydrogenase exhibits a low degree of homology with the E. coli enzyme. However, the sequences of the pig heart, yeast and E. coli NADP-dependent isocitrate dehydrogenases have been aligned using the CLUS- TAL program of PC/GENE to maximize relatedness (1). This alignment, a section of which is shown in Table V, places Asp253 of the pig heart enzyme (one of the sites of affinity cleavage by Fez+-isocitrate) a t a position equivalent to Aspz7' of the E. coli enzyme, which is close to but not identical with the Aspza3 found to be coordinated to the Mg2+ in the crystalline ternary complex of the E. coli enzyme (23). Table V shows that of the pig heart enzyme (close to Asp273, which is a cleavage site of Fez+ alone) can be aligned with the E. coli enzyme's Asp307, the second group found to be coordinated to the Mg2+ in the E. coli enzyme ternary complex (23). It is possible that the metal-isocitrate binding sites are located in the same general region of these two isocitrate dehydrogenases, despite their low overall amino acid sequence similarity. His3'', the third residue of the pig heart enzyme which we have designated (based on cleavage by Fez+-isocitrate) as close to the metal-substrate site, aligns well with His339 of the E. coli (see Table V). In the crystalline NADP-enzyme complex of the E. coli enzyme, His339 has been located near the adenine of the coenzyme (23). In contrast, there is no evidence impli- cating His3'' in the coenzyme site of the pig heart enzyme and measurements of the pH dependence of the binding of NADPH in the absence and presence of Mn2+ indicate that

the nucleotide binds in its unprotonated, metal-free form (30). Sequence alignments between the pig heart and E. coli enzymes are poorest in the region of the bacterial enzyme which has been implicated in coenzyme binding. Detailed compari- sons between the corresponding binding sites of the pig heart and E. coli NADP-dependent isocitrate dehydrogenases must await analysis of the crystal structures of the mammalian enzyme and its complexes. However, the results of the present investigation of the affinity cleavage of the pig heart isocitrate dehydrogenase place the Asp253 and His309 at the metal-isocitrate site and Asp273 at the nearby, but distinct binding site of the free metal ion.

Acknowledgments-We thank Dr. Yu-Chu Huang for the peptide sequencing and Dr. Robert S. Ehrlich for helpful discussions.

REFERENCES 1. Haselbeck, R. J., Colman, R. F., and McAlister-Henn, L. (1992) Biochem.

2. Kelly, J. H., and Plaut, G. W. E. (1981) J. Biol. Chem. 256 , 330-334

4. Si man, D. S., and Chen, C. B. (1990) Annu. Reu. Biochem. 59,207-236 3. Colman, R. F. (1983) Pept. Protein Reu. 1 , 41-69

5. Sc%epartz, A,, and Cuenoud, B. (1990) J. Am. Chem. Soc. 112,3247-3249 6. Hoyer, D., Cho, J., and Schultz, P. G. (1990) J. Am. Chem. SOC. 112,3249-

31,6219-6223

wsn

8. Colman, R. F. (1972) J. Biol. Chem. 247,215-223 7. Vii&anca, J. J., and Colman, R. F. (1972) J. Biol. Chem. 2 4 7 , 209-214

10. NortK;op, D':B., and Clela'nd, W. W. (1970) Fed. Proc. Fed. Am. Soc. Exp. 9. Bade J M and Colman R. F. (1987) Emhemutry 26,4893-4900

11. Ehrlich, R. S., and Colman R. F. (1976) Biochemistry 15,4034-4041 12. Soundar, S., and Colman, k. F. (1992) Sixth Symposium of The Protein

13. Bacon, C! R., Bednar, R. A., and Colman, R. F. (1981) J. Bid. Chem. 256 ,

14. Smyth G. E. and Colman, R. F. (1991) J. Biol. Chem. 2 6 6 , 14918-14925 15. Johanion, R.'A., and Colman, R. F. (1981) Arch. Biochem. Biophys. 207 ,

Biol. 49, 408

Societ , San Diego, CA, July 25-29, p. 144

6593-6599

9-20 16. Blackshear P. J. (1984) Methods Enzymol. 104,237-255 17. Levine, R. L. (1983) J. Biol. Chem. 2 6 8 , 11823-11827 18. Amici, A., Levine, R. L., Tsai, L., and Stadtman, E. R. (1989) J. Biol. Chem.

19. Kim, K., Rhee, S. G., and Stadtman, E. R. (1985) J. Biol. Chem. 260 , 264,3341-3346 I & ~ Q A _ I K ? Q ~

20. Dean A. M., and Koshland D. B. (1990) Science 2 4 9 , 1044-1046 21. LeviAe, R. L. (1983) J. Biol.' Chem. 2 5 8 , 11828-11833 22. Creeth, J. M., Cooper, B., Donald, A. S. R., and Clamp, J. R. (1983)

A"""" l""" I

23. Hurley, J. H., Dean, A. M., Koshland, D. E., Jr., and Stroud, R. M. (1991) Biochem. J. 21 1,323-332

Rinrhomrrtrv Wh RG71-RG7R 24. Levy R. S., and Villafranca J. J. (1977) Biochemistry 16,3293-3301 25. Villahanca J. J., and Colmin, R. F. (1984) Biochemlstry 13 , 1152-1160 26. Ehrlich. R.'S.. and Colman. R. F. (1987) Biochemistrv 26. 3461-3466

_".._I_.II ",_ "._ "._

27. Ehrlich; R. S.; and Colman; R. F. (1989) Biochemist6 26; 2058-2065 28. Balley, D. B., Ellis, P. D., and Fee, J. A. (1980) Biochemistry 19 , 591-596 29. Rhyu, G. I., Ray, W. J., and Markley, J. L. (1985) Blochemutry 24,2536-

'?.&dl 30. Mi;". T., and Colman, R. F. (1984) Biochemistry 2 3 , 1675-1683

identification of metal-isocitrate binding site of pig heart nadp

Documents