Interaction of Leuconostoc mesenteroides glucose-6-phosphate dehydrogenase with pyridoxal 5'-diphospho-5'-adenosine. Affinity labeling of Lys-21 and Lys-343.

Pyridoxal 5'-diphospho-5'-adenosine (PLP-AMP) inhibits glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroides competitively with respect to glucose 6-phosphate and noncompetitively with respect to NAD+ or NADP+, with Ki = 40 microM in the NADP-linked and 34 microM in the NAD-linked reaction. Incubation of glucose-6-phosphate dehydrogenase with [3H]PLP-AMP followed by borohydride reduction shows that incorporation of 0.85 mol of PLP-AMP per mol of enzyme subunit is required for complete inactivation. Both glucose 6-phosphate and NAD+ protect against this covalent modification. The proteolysis of the modified enzyme and isolation and sequencing of the labeled peptides revealed that Lys-21 and Lys-343 are the sites of PLP-AMP interaction and that glucose 6-phosphate and NAD+ protect both lysyl residues against modification. Pyridoxal 5'-phosphate (PLP) also modifies Lys-21 and probably Lys-343. Lys-21 is part of a highly conserved region that is present in all glucose-6-phosphate dehydrogenases that have been sequenced. Lys-343 corresponds to an arginyl residue in other glucose-6-phosphate dehydrogenases and is in a region that is less homologous with those enzymes. PLP-AMP and PLP are believed to interact with L. mesenteroides glucose-6-phosphate dehydrogenase at the glucose 6-phosphate binding site. Simultaneous binding of NAD+ induces conformational changes (Kurlandsky, S. B., Hilburger, A. C., and Levy, H. R. (1988) Arch. Biochem. Biophys. 264, 93-102) that are postulated to interfere with Schiff's-base formation with PLP or PLP-AMP. One or both of the lysyl residues covalently modified by PLP or PLP-AMP may be located in regions of the enzyme undergoing the NAD(+)-induced conformational changes.

Pyridoxal 5'-diphospho-5'-adenosine (PLP-AMP) inhibits glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroides competitively with respect to glucose 6-phosphate and noncompetitively with respect to NAD' or NADP+, with Ki = 40 pM in the NADP-linked and 34 p~ in the NAD-linked reaction. Incubation of glucose-6-phosphate dehydrogenase with [3H]PLP-AMP followed by borohydride reduction shows that incorporation of 0.85 mol of PLP-AMP per mol of enzyme subunit is required for complete inactivation. Both glucose 6-phosphate and NAD+ protect against this covalent modification. The proteolysis of the modified enzyme and isolation and sequencing of the labeled peptides revealed that Lys-21 and Lys-343 are the sites of PLP-AMP interaction and that glucose 6-phosphate and NAD+ protect both lysyl residues against modification. Pyridoxal 5"phosphate (PLP) also modifies Lys-21 and probably Lys-343. Lys-21 is part of a highly conserved region that is present in all glucose-6-phosphate dehydrogenases that have been sequenced. Lys-343 corresponds to an arginyl residue in other glucose-6-phosphate dehydrogenases and is in a region that is less homologous with those enzymes. PLP-AMP and PLP are believed to interact with L. mesenteroides glucose-6-phosphate dehydrogenase at the glucose 6-phosphate binding site. Simultaneous binding of NAD' induces conformational changes (Kurlandsky, S. B., Hilburger, A. C., and Levy, H. R. (1988) Arch. Biochem. Biophys. 264,[93][94][95][96][97][98][99][100][101][102] that are postulated to interfere with Schiff's-base formation with PLP or PLP-AMP. One or both of the lysyl residues covalently modified by PLP or PLP-AMP may be located in regions of the enzyme undergoing the NAD+-induced conformational changes. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroides can utilize either NAD' or NADP+ as coenzyme (1,2). The enzyme is a dimer consisting of identical subunits with a molecular weight of 54,800 (3). Earlier studies showed that PLP' acts as an affinity probe for the Glc-6-P site (4,5 ) . PLP-AMP is an analog of PLP that was designed to have enhanced specificity for the nucleotide or coenzyme binding sites of kinases and dehydrogenases (6)(7)(8). We have labeled L. mesenteroides Glc-6-P-DH with PLP-AMP to ascertain whether this reagent could modify the coenzyme binding site.

1)
Here we report the sequences of two peptides which contain most of the incorporated label. We have also repeated the P L P labeling studies, showing that both PLP and PLP-AMP appear to interact with the same lysyl residues of the enzyme. Protection against covalent modification by PLP-AMP and P L P was afforded by both Glc-6-P and NAD+ but inhibition was competitive only with respect to Glc-6-P. We conclude that both PLP-AMP and PLP bind at the Glc-6-P site.

EXPERIMENTAL PROCEDURES
Materials-L. mesenteroides Glc-6-P-DH was obtained from Worthington Biochemicals. PLP-AMP and ["HIPLP-AMP were synthesized according to the method of Tamura et al. (7). ["HIPLP was synthesized by the method of Johansson et al. (9) as modified by Tamura et al. (7). NAD', NADP+, and endoproteinase Lys-C came from Boehringer Mannheim; PLP and Glc-6-P were from Sigma. Aqueous counting scintillants were Liquiscint from National Diagnostics or BioSafe I1 from Research Products International Corp.
Reversed-phase C4 analytical HPLC columns were purchased from J. T. Baker.
Assays-Glc-6-P-DH was routinely assayed at 25 "C by measuring the reduction of NADP+ a t 340 nm. Assay solutions contained 57 p M NADP+, 0.81 mM Glc-6-P, and 33 mM Tris-HCI, pH 7.8, and reactions were initiated by the addition of enzyme. Protein concentration was assayed using a modified Lowry procedure (10) with a standard curve constructed from the colorimetric response of known weights of defatted bovine serum albumin. Covalent incorporation of ['HIPLP-AMP was measured by the following procedure. An aliquot of column effluent containing a known quantity of enzyme was added to a 7-ml vial containing 1 ml of water, followed by addition of 100 pl of 0.15% deoxycholate. After a 10-min incubation at room temperature, protein was precipitated by addition of 100 p1 of 72% trichloroacetic acid. Quantitative precipitation of the enzyme was verified experimentally, in agreement with observations with other proteins (10). Vials were centrifuged and the supernatant solutions were carefully aspirated. The pellets were dissolved in aqueous counting scintillant and counted in a Beckman LS 1800 scintillation counter.
Kinetics-The kinetics of PLP-AMP inhibition were determined by incubating either 0.25 or 0.75 nM Glc-6-P-DH at 25 "C in the dark in 25 mM potassium phosphate, pH 7.6, either without PLP-AMP or with 10.5 or 31.6 p~ PLP-AMP. Incubation mixtures included the varied substrate (Glc-6-P or coenzyme) at one of five concentrations, ranging from 0.2 to 5.0 times its K, value (11). After a 60-min preincubation, the nonvaried substrate was added to a final concentration of 20 times its K, and the reaction rate was measured. All assays were conducted in duplicate. Glc-6-P and NAD+ solutions were adjusted to pH 7.6 prior to their addition to the assay solution. Data were analyzed using double-reciprocal plots drawn with the aid of the COMP program described by Cleland (12).
Specificity Studies-Covalent modification of Glc-6-P-DH was achieved by incubating the enzyme in the dark with 422 p~ [3H] PLP-AMP (1.58 X 1013 cpm/mol) for various times, ranging from 1 to 45 min. After a brief incubation (approximately 15 s) with a large molar excess of sodium borohydride from a freshly prepared solution kept on ice, unbound ligand was removed from the medium by centrifugation through a 1-ml column of Sephadex G-50-80 (13) equilibrated in a solution containing 35 mM potassium phosphate, 1.0 mM EDTA, 0.2 M NaC1, pH 7.2 ("storage buffer"). The volume of the effluent was measured, and aliquots were used to determine catalytic activity, radioisotope incorporation, and protein concentration as described above.
Preparatiue Labeling and Peptide Isolation-Glc-6-P-DH was incubated at 9.6 p~ for 10 min in storage buffer at room temperature, either without ligands or with 12 mM Glc-6-P or 72 mM NAD'; these concentrations are 10 times the respective KO values (14).
[3H]PLP-AMP or ['HIPLP was added to a final concentration of 384 or 375 FM, respectively, and each sample was incubated for 10 min prior to reduction with a large molar excess of NaBH4. Unbound ligand was removed by passage through a 3-ml centrifuge column containing Sephadex G-50-80 equilibrated in storage buffer. Aliquots were withdrawn for assay of activity, protein, and bound 3H as described above. Remaining protein was precipitated by addition of trichloroacetic acid to 14.4%, and pellets were washed with ice-cold acetone and then dissolved in 5 M urea, 50 mM HEPES-KOH, pH 8.0, 5 mM methylamine. Glc-6-P-DH was digested for 24 h with endoproteinase Lys-C at a final protease-to-Glc-6-P-DH ratio of 1:50 (w:w). Equivalent weights of the proteolytic digests were separated by reversed-phase HPLC.
Analytical HPLC was done using linear gradients of 0.1% ammonium trifluoroacetate, pH 6.5 (solvent E), to 0.1% ammonium trifluoroacetate in 90% acetonitrile, pH 6.5 (solvent F) (15). The gradient was developed from 0 to 15% solvent F at a rate of 3.00% F/min, from 15 to 60% solvent F at 1.00% F/min, and from 60 to 100% solvent F at 4.00% F/min. Fractions of 1.5 ml were collected. Each fraction was mixed with aqueous counting scintillant and counted by liquid scintillation. Yields of total radioactivity recovered from the column ranged from 75 to 90%.
Semi-preparative HPLC for peptide purification was done in two steps.
Step 1 was exactly the same as the analytical HPLC procedure described above. In step 2, the major radioactive fractions from step 1 were further purified using a linear gradient from 0.1% trifluoroacetic acid (solvent A) to 0.1% trifluoroacetic acid in 90% acetonitrile (solvent B). The rate of gradient development was the same as in step 1. Eluting peptides were collected manually in separate tubes. Purified peptides were sequenced by automated Edman degradation using an Applied Biosystems 470A gas-phase peptide micro-sequencer. Affinity labeling, peptide purifications, and sequence analyses were all performed in duplicate.

RESULTS
Reversibility of Inhibition of PLP-AMP-Glc-6-P-DH was incubated at 114 nM at room temperature in 25 mM potassium phosphate, pH 7.6, with 10.2, 31.3, and 100 p~ PLP-AMP. Activities were measured at various times and, by 1 h, reached equilibrium values of 82, 58, and 33%, respectively, of the original activity. The enzyme solutions were then dialyzed for 72 h against storage buffer, pH 7.2. The activities of the dialyzed enzyme solutions returned to 97, 94, and 99%, respectively, of the original activity.
Kinetics of Reversible PLP-AMP Inhibition-Inhibition of Glc-6-P-DH by PLP-AMP was measured (without borohydride reduction) using 10.5 and 31.6 p~ PLP-AMP, varying either coenzyme or Glc-6-P. Inhibition was noncompetitive with respect to either NAD' or NADP+ and competitive with respect to Glc-6-P using either coenzyme (data not shown). I(, values for PLP-AMP were determined to be 40 PM in the NADP-linked and 34 PM in the NAD-linked reaction, respectively.
Stoichiometry of PLP-AMP Incorporation-Glc-6-P-DH was incubated at 9.6 PM with 422 p M [3H]PLP-AMP. Aliquots were removed at various times, reduced with excess sodium borohydride, followed by centrifugation through a column of Sephadex G-50-80 to remove PLP-AMP. Fig. 1 shows that loss of activity is linearly related to the amount of PLP-AMP incorporated. Extrapolation of the plot yields a value of 0.85 mol of PLP-AMP incorporated per mol of enzyme subunit at 0% residual activity.
Inactivation by PLP-AMP and Protection by Coenzymes and Substrute-Glc-6-P-DH was incubated with f3H]PLP-AMP either in the absence of substrates, or in the presence of Glc-6-P or NAD', each at 10 times its respective KD value (14). Following borohydride reduction and column centrifugation, enzyme activities and radioisotope incorporation were measured. Both Glc-6-P and NAD' protected against PLP-AMP inactivation and modification ( Table I).
Identification of Amino Acids Modified by PLP-AMP-Labeled samples for peptide purification were prepared as described in Table I. The enzymatic activities remaining after PLP-AMP modification and the stoichiometries of PLP-AMP incorporation are shown in Table I. Proteolysis of labeled enzyme was performed with endoproteinase Lys-C, previously shown not to cleave at PLP-or PLP-AMP-modified lysyl residues (16). Equivalent weights of the initial digests were separated by HPLC, as described under "Experimental Procedures." Column eluates were monitored for absorbance at 214 (peptides), 260 (adenine, with interference from Tyr and Trp), and 325 nm (reduced pyridoxamine derivative) and for radioactivity.  Inactivation and modification of glucose-6-phosphate dehydrogenase by 13H]PLP-AMP Glc-6-P-DH (9.64 PM) was incubated a t room temperature for 5 min either without substrate or with 12 mM Glc-6-P or 72 mM NAD' in 35 mM potassium phosphate, 1.0 mM EDTA, 0.2 M NaCI, pH 7.2.
[3H]PLP-AMP was added to a final concentration of 384 PM and incubations continued for 10-20 min in the dark. Reduction by NaBH4, gel filtration, and measurements of volume, enzyme activity, protein concentration, and radioisotope incorporation were performed as described in the legend to Fig. 1. In the sample labeled in the absence of protecting ligand (Fig. 2, panel A: solid bars and panel B ) , there were two principal peaks eluting at 22 and 29% F that accounted for the bulk of the radioactivity and the absorption at 325 nm. The amount of radioactivity eluting at fraction 11 was variable in different experiments and substrate protection was not evident. A broad band of radioactivity, mostly eluting after the two major peaks, may in part reflect the co-elution of the 22% F and 29% F peptides with other peptides. This was supported by the finding that when these fractions were rechromatographed, some of the label eluted at 29% F. The two principal peaks were significantly diminished in incubations that included Glc-6-P (panel A: open bars) or NAD+ (data not shown).
The peptides corresponding to the two principal peaks were   Table  11. The peptide isolated as the 29% F peak (Peptide 1) was labeled at Lys-21, whereas the 22% F peptide (Peptide 2) was labeled at Lys-343. Peptide isolation and sequencing were reproduced in two independent experiments.
Inactivation by PLP-Experiments were performed in which [3H]PLP incorporation was measured and peptides labeled with [3H]PLP were isolated. As found with PLP-AMP, P L P incorporation is substantially protected against by Glc-6-P and NAD+ ( Table 111). Samples of Glc-6-P-DH covalently modified with PLP were digested with endoproteinase Lys-C, and labeled peptides were separated by HPLC, using the same procedure as that described for PLP-AMPlabeled enzyme. Two separate experiments produced elution patterns that were indistinguishable from those seen with the peptides generated from PLP-AMP-modified Glc-6-P-DH. The peptide eluting at the same position as peptide 1 from PLP-AMP-modified enzyme was subjected to amino acid sequencing (11 cycles) and its sequence proved to be identical to that of peptide 1.

DISCUSSION
PLP-AMP was designed and synthesized independently in two different laboratories as an affinity label for nucleotide or coenzyme binding sites on kinases and dehydrogenases (6)(7)(8). P L P acts as an affinity label for lysyl residues that interact with phosphates, particularly sugar phosphates (4, 17). The attachment of an AMP moiety to PLP should direct this affinity label to lysyi residues of proteins, such as kinases and dehydrogenases, that have a strong affinity for adenine nucleotides. It was anticipated, therefore, that enzymes that bind nucleotides or nicotinamide coenzymes would bind PLP-AMP with greater affinity and selectivity than PLP. This is known to be the case for hexokinase, 3-phosphoglycerate kinase, adenylate kinase, and alcohol dehydrogenase (7). Experiments described in this communication were undertaken to test this possibility with Glc-6-P-DH from L. mesenteroides.
The interaction of L. mesenteroides Glc-6-P-DH with PLP-AMP, however, appears to be identical to its interaction with PLP. Kinetic studies showed that the mechanism of reversible inhibition by PLP-AMP is competitive with respect to Glc-6-P when either NAD+ or NADP+ is the coenzyme and noncompetitive' with respect to both NAD+ and NADP'. This is the same as the mechanism of P L P inhibition of L. mesenteroides Glc-6-P-DH (4). The IC, values for PLP-AMP ' Noncompetitive, rather than uncompetitive, inhibition occurs because the kinetic mechanism of the NAD-linked reaction is random, whereas it is ordered, with coenzyme binding first, for the NADPlinked reaction. The enzyme-Glc-6-P complex is utilized in the NADlinked, hut not the NADP-linked reaction (18).  I1 PLP-AMP-labeled peptides of glucose-6-phosphate dehydrogenase Peptide 1 is the "29% F"peak (Fig. 2); its amino acid sequence corresponds t,o residues 20-31." Peptide 2 is the "22% F" peak (Fig. 2); its amino acid sequence corresponds to residues 339-352." At the position indicated (Lys) there was a blank in the sequence; the presence of a Lys residue at this location was inferred from its presence in the complete amino acid ~e q u e n c e .~ Previous experience with radioactive phenylthiohydantoin-derivatives of PLP-Lys or PLP-AMP-Lys (16) has shown that the bulk of such derivatives do not elute from the glass-fiber disk of the gas-phase sequencer. The small amount that does elute is found superimposed with the injection artefact of the HPLC profile and is neither identifiable nor quantifiable. Peptides were sequenced in an Applied Biosystems model 475 gas-phase sequencer with on-line HPLC (Applied Biosystems model 120) phenylthiohydantoin-derivative analysis. For Peptide 1, 62.5 pmol was applied to the sequencer; the repetitive yield was 93.5%. For Peptide 2, 86 pmol was applied to the sequencer; the repetitive yield was 93%. The sequence analysis for each peptide was repeated once.  Inactivation and modification of glucose-6-phosphate dehydrogenase by ["HIPLP Glc-6-P-DH (9.64 p~) was incubated at room temperature for 5 min either without substrate or with 12 mM Glc-6-P, or 70 mM NAD' in 35 mM potassium phosphate, 1.0 mM EDTA, 0.2 M NaCl, pH 7.2.
["HIPLP was added to a final concentration of 375 pM and incubations continued for 11 min in the dark. Reduction by NaBH4, gel filtration, and measurements of volume, enzyme activity, protein concentration, and radioisotope incorporation were performed as described in the legend to Fig. 1 are virtually identical to those for P L P (4). As with PLP, complete inactivation with PLP-AMP leads to the incorporation of approximately 1 mol of affinity probe/mol of subunit (4). Finally, Lys-21 is modified by both PLP and PLP-AMP. In addition, P L P is incorporated into a peptide eluting at the same position on HPLC as a peptide labeled by PLP-AMP and shown, upon sequencing, to be labeled on Lys-343. Although this PLP-labeled peptide was not sequenced, the evidence suggests that it too is modified at Lys-343. In contrast, pyridoxal and pyridoxamine 5"phosphate are only weakly inhibitory toward L. mesenteroides Glc-6-P-DH (4). This suggested to us that the phosphate and aldehyde moieties are essential for binding or inhibition, but the results with PLP-AMP indicate that esterification of the phosphate to an AMP moiety has no perceptible influence on the binding or inhibition characteristics of the probe. It seems likely, therefore, that both PLP and PLP-AMP are functioning as affinity probes for the sugar phosphate site on L. mesenteroides Glc-6-P-DH rather than at the coenzyme site.
Incubation of L. mesenteroides Glc-6-P-DH with PLP-AMP followed by NaBH4 reduction leads to the covalent modification of Lys-21 and Lys-343. The modification of 2 lysyl residues with a stoichiometry of approximately one modified per subunit suggests that the t-NH, groups of both Lys-21 and Lys-343 are sufficiently close to the aldehyde group of PLP-AMP so that either one can react. Both lysyl residues are protected approximately equally well against this modification by either Glc-6-P or NAD', when these ligands are present at 10 times their respective KI, concentrations. Similar results are found with P L P modification. The fact that both Glc-6-P and NAD+ protect the enzyme against covalent modification by PLP-AMP and PLP, whereas inhibition is competitive with respect to Glc-6-P, but not with respect to NAD', suggests the following interpretation. P L P and the PLP moiety of PLP-AMP bind at the Glc-6-P site. Simultaneous binding of Glc-6-P and either probe is not possible, explaining the fact that the probes inhibit competitively with respect to Glc-6-P and that Glc-6-P protects the enzyme against modification and inactivation. In contrast, NAD' can still bind when P L P or PLP-AMP is bound at the Glc-6-P site, but the bound coenzyme either interferes sterically with binding of the probes or causes a conformational change that makes it difficult to form the Schiffs-base complex that is required for inactivation and modification following borohydride reduction. This interpretation is consistent with the finding that the probes inhibit noncompetitively with respect to NAD', yet NAD' provides some protection against modification and inactivation. Steric interference between NAD' and probe is unlikely in view of the fact that NAD' and Glc-6-P promote each other's binding during catalysis (11) and that covalent modification of Glc-6-P-DH with P L P increases its affinity for NAD+ nearly 10-fold (5). Evidence which favors interference by NAD+ with Schiffsbase formation includes extensive documentation for a major conformational change upon NAD' binding (5,14,19).
Previously, a pyridoxyllysine-containingpeptide from PLPlabeled L. mesenteroides Glc-6-P-DH was isolated and sequenced before the entire amino acid sequence of the enzyme was known (3). Nonradioactive P L P was used in that study and the modified peptide was identified from its absorbance a t 315 nm and its fluorescence a t 390 nm upon excitation at 325 nm. The sequence of the PLP-labeled peptide determined earlier (3) does not correspond to any portion of the recently determined L. mesenteroides Glc-6-P-DH sequence.:i We are unable to provide a satisfactory explanation of this finding but assume that a contaminating peptide was co-purified, obscuring the previous sequence analysis.
inhibition (26). The lysyl residue modified is part of a sequence of 13 amino acids that is identical in Glc-6-P-DHs from human erythrocytes, L. mesenteroides (27, 28), rat liver (20,23), and Drosophila (25). Homologous peptides consisting of 11 amino acids have also been isolated from Glc-6-P-DHs from Pichia jadinii (29) and Saccharomyces cerevisiae (30). In the Glc-6-P-DHs from yeast (31) and P. jadinii (29) the lysyl residue in this conserved sequence is selectively modified by acetylsalicylic acid. Again, Glc-6-P, but not NADP+, protects effectively against modification (31). This conserved lysinecontaining sequence is, therefore, likely to form part of the active site, and the evidence with the Glc-6-P-DHs from human erythrocytes, S. cereuisiae, and P. jadinii suggests that it may be involved in Glc-6-P binding. This conserved sequence includes Lys-182 in L. mesenteroides Glc-6-P-DH; and our results show that this lysyl residue is not covalently modified with PLP-AMP or PLP. L. mesenteroides Glc-6-P-DH is unique, among the Glc-6-P-DHs for which amino acid sequence information is available, in its ability to utilize either NAD' or NADP+ as its coenzyme (1,2), and this is expected to lead to some differences in its interactions with Glc-6-P and/or coenzymes. Its interactions with NAD+ and Glc-6-P has been shown to involve major conformational changes (5, 14,19). One or both of the lysyl residues covalently modified by PLP and PLP-AMP may be located in regions of the enzyme undergoing these ligand-induced conformational changes. It has been shown recently that PLP-AMP-labeled lysyl residues in hexokinase (32) and 3-phosphoglycerate kinase (33) must traverse considerable distances during the ligand-induced conformational closure of the central binding cleft on these enzymes. Support for similar conformational changes involving those regions of L. mesenteroides Glc-6-P-DH containing Lys-21 and/or Lys-343 must await the results of detailed x-ray structural studies of this enzyme, currently in progress (34).