Identification of lysyl residues located at the substrate-binding site in UDP-glucose pyrophosphorylase from potato tuber: affinity labeling with uridine di- and triphosphopyridoxals

Uridine di- and triphosphopyridoxals were used to probe the substrate-binding site in potato tuber UDP-glucose pyrophosphorylase (EC 2.7.7.9). The enzyme was rapidly inactivated in time- and dose-dependent manners when incubated with either reagent followed by reduction with sodium borohydride. The inactivations were almost completely retarded by UDP-Glc and UTP but only slightly by alpha-D-glucose 1-phosphate. The complete inactivation corresponded to the incorporation of about 0.9-1.0 mol of either reagent per mole of enzyme monomer. Both reagents appear to bind specifically to the UDP-Glc-(UTP)-binding site. Structural studies of the labeled enzymes revealed that the two reagents modified the identical set of five lysyl residues (Lys-263, Lys-329, Lys-367, Lys-409, and Lys-410), in which Lys-367 was most prominently modified. The ratios of the amounts of labels incorporated into these residues were similar for the two reagents. Furthermore, linear relationships were observed between the residual activities and the amounts of incorporation into each lysyl residue. We conclude that the five lysyl residues are located at or near the UDP-Glc(UTP)-binding site of potato tuber UDP-Glc pyrophosphorylase and that the modification of these residues occurs in a mutually exclusive manner, leading to the inactivation of the enzyme.     

UDP-glucose pyrophosphorylase from potato tuber: cDNA cloning and sequencing

We have isolated a cDNA encoding UDP-glucose pyrophosphorylase from a cDNA library of immature potato tuber using oligonucleotide probes synthesized on the basis of partial amino acid sequences of the enzyme. The cDNA clone contained a 1,758-base-pair insert including the complete message for UDP-glucose pyrophosphorylase with 1,431 base pairs. The amino acid sequence of the enzyme inferred from the nucleotide sequence consists of 477 amino acid residues. All the partial amino acid sequences determined protein-chemically [Nakano et al. (1989) J. Biochem. 106, 528-532] confirmed the primary structure of the enzyme. An N-terminal-blocked peptide was isolated from the proteolytic digest of the enzyme protein, and the blocking group was deduced to be an acetyl group by fast atom bombardment-mass spectrometry. On the basis of the predicted amino acid sequence (477 residues minus the N-terminal Met plus an acetyl group), the molecular weight of the enzyme monomer is calculated to be 51,783, which agrees well with the value determined by polyacrylamide gel electrophoresis. In the cDNA structure, the open-reading frame is preceded by a 125-base-pair noncoding region, which contains a sequence being homologous with the consensus sequence for plant genes, and is followed by a 174-base-pair noncoding sequence including a polyadenylation signal. Amino acid sequence comparisons revealed that the potato UDP-glucose pyrophosphorylase is homologous to the enzyme from slime mold, Dictyostelium discoideum, but not to ADP-glucose pyrophosphorylases from rice seed and Escherichia coli.     

Comparative affinity labeling with reactive UDP-glucose analogues: possible locations of five lysyl residues around the substrate bound to potato tuber UDP-glucose pyrophosphorylase.

By using two reactive analogues of UDP-Glc, uridine di- and triphosphopyridoxals, we have recently probed the substrate-binding site in potato tuber UDP-Glc pyrophosphorylase [EC 2.7.7.9]. In this work, pyridoxal diphospho-alpha-D-glucose was used for the same purpose. This compound is also a reactive UDP-Glc analogue but having its reactive group on the opposite side of the pyrophosphate linkage to those of the above two compounds. The enzyme was rapidly inactivated when incubated with the compound at very low concentrations followed by reduction with sodium borohydride. The inactivation was almost completely prevented by UDP-Glc and UTP. Complete inactivation correspond to the incorporation of 1.0 mol of the reagent per mol of enzyme monomer. The label was found to be distributed in five lysyl residues (Lys-263, Lys-329, Lys-367, Lys-409, and Lys-40. All of these results were similar to those obtained previously with the other compounds, suggesting the presence of a cluster of five lysyl residues at or near the substrate-binding site of this enzyme. However, the incorporations of labels into each lysyl residue differed depending on the compounds used. The substrate retarded the incorporations in different manners. Based on the combined results of the present and previous studies, a hypothetical model is presented for the possible locations of the five lysyl residues around the substrate bound to the enzyme. This model is consistent with the kinetic properties of mutant enzymes in which the five lysyl residues were individually replaced by glutamine via site-directed mutagenesis.     

UDP-glucose pyrophosphorylase from potato tuber: purification and characterization

UDP-glucose pyrophosphorylase from potato tuber was purified 243-fold to a nearly homogeneous state with a recovery of 30%. The purified enzyme utilized UDP-glucose, but not ADP-glucose, as the substrate, and was not activated by 3-phosphoglyceric acid. Product inhibition studies revealed the sequential binding of UDP-glucose and MgPPi and the sequential release of glucose-1-phosphate and MgUTP, in this order. Analyses of the effects of Mg2+ on the enzyme activity suggest that the MgPPi and MgUTP complexes are the actual substrates for the enzyme reaction, and that free UTP acts as an inhibitor. The enzyme exists probably as the monomer of an approximately 50-kDa polypeptide with a blocked amino terminus. For structural comparison, 29 peptides isolated from a tryptic digest of the S-carboxymethylated enzyme were sequenced. The results show that the potato tuber enzyme is homologous to UDP-glucose pyrophosphorylase from slime mold, but not to ADP-glucose pyrophosphorylase from Escherichia coli, and provide structural evidence that UDP-glucose and ADP-glucose pyrophosphorylase are two different protein entities.     

Probing the pyrophosphate-binding site in potato tuber UDP-glucose pyrophosphorylase with pyridoxal diphosphate.

Potato tuber UDP-glucose pyrophosphorylase (EC 2.7.7.9) catalyzes the reversible uridylyl transfer from UDP-glucose to MgPPi forming glucose 1-phosphate and MgUTP, according to an ordered bi-bi mechanism in which UDP-glucose and MgPPi bind in this order. To probe the active site of this enzyme, we have applied pyridoxal 5'-diphosphate, a reactive PPi analogue. The enzyme was rapidly inactivated when incubated with the reagent in the presence of Mg2+ followed by sodium borohydride reduction. The degree of the inactivation was decreased by MgUTP, MgPPi, and glucose 1-phosphate, but enhanced by UDP-glucose. The enhancement was prevented by co-addition of Pi, the competitive inhibitor with respect to PPi. The complete inactivation corresponded to the incorporation of 0.9-1.1 mol of reagent/mol of enzyme monomer. In the presence of UDP-glucose, labels were almost exclusively incorporated into Lys-329. Thus, this residue may be located near the bound MgPPi and its modification is promoted, probably through conformational changes, by the binding of UDP-glucose to the enzyme. The results of the modification by the same reagent of the mutant enzymes in which Lys-329 and Lys-263 are individually replaced by Gln suggest the roles of these lysyl residues in the binding of MgPPi and in the UDP-glucose-induced conformational changes, respectively.     

Identification of lysyl residues at the AMP-binding site of biodegradative threonine deaminase from Escherichia coli.

The biodegradative threonine deaminase from Escherichia coli is activated allosterically by AMP. To identify the residues interacting with the phosphate group of AMP at the binding site, we used the affinity labeling reagent, adenosine diphosphopyridoxal (AP2-PL). In the absence of AMP, the enzyme formed the Schiff base with AP2-PL and Scatchard plot analysis showed a biphasic pattern, the respective Kd values for the high- and low-affinity binding phases being 20 and 110 microM. The former value is comparable to the Kd value of the enzyme for AMP. In the presence of AMP, the Schiff base formation was greatly reduced. Although the maximal activating effect of adenosine diphosphopyridoxine, a non-reactive derivative of AP2-PL, was about 13% of that of AMP, the half-saturation concentration was almost the same. These findings suggest that AP2-PL specifically labeled the lysyl residue(s) at the AMP-binding site of the enzyme. To identify the labeled residue(s), we reduced the modified enzyme with sodium borohydride, then cleaved it with cyanogen bromide and Achromobacter lyticus protease I. Reverse-phase HPLC was used to isolate two labeled peptides from the digest. Their amino acid compositions and sequences showed that Lys-111 and Lys-113 were labeled. We conclude that these two lysyl residues are located around the phosphate group of AMP at the allosteric regulation site of the enzyme.     

ADP-glucose pyrophosphorylase from potato tuber: site-directed mutagenesis of homologous aspartic acid residues in the small and large subunits

Asp142 in the homotetrameric ADP-glucose pyrophosphorylase (ADP-Glc PPase) enzyme from Escherichia coli was demonstrated to be involved in catalysis of this enzyme [Frueauf, J.B., Ballicora, M.A. and Preiss J. (2001) J. Biol. Chem., 276, 46319-46325]. The residue is highly conserved throughout the family of ADP-Glc PPases, as well as throughout the super-family of sugar-nucleotide pyrophosphorylases. In the heterotetrameric ADP-Glc PPase from potato (Solanum tuberosum L.) tuber, the homologous residue is present in both the small (Asp145) and the large (Asp160) subunits. It has been proposed that the small subunit of plant ADP-Glc PPases is catalytic, while the large subunit is modulatory; however, no catalytic residues have been identified. To investigate the function of these conserved Asp residues in the ADP-Glc PPase from potato tuber, we used site-directed mutagenesis to introduce either an Asn or a Glu. Kinetic analysis in the direction of synthesis or pyrophosphorolysis of ADP-Glc showed a significant decrease (more than four orders of magnitude) in the specific activity of the SD145NLwt, SD145NLD160N, and SD145NLD160E mutants, while the effect was smaller (approximately two orders of magnitude) with the SD145ELwt, SD145ELD160N, and SD145ELD160E mutants. By contrast, mutation of the large subunit alone did not affect the specific activity but did alter the apparent affinity for the activator 3-phosphoglycerate, showing two types of apparent roles for this residue in the different subunits. These results show that mutation of Asp160 of the large subunit does not affect catalysis, thus the large subunit is not catalytic, and that the negative charge of Asp145 in the small subunit is necessary for enzyme catalysis.     

Expression of rat liver S-adenosylhomocysteinase cDNA in Escherichia coli and mutagenesis at the putative NAD binding site

The cDNA for rat liver S-adenosylhomocysteinase has been cloned, and the nucleic acid sequence has been determined. By comparison of the deduced amino acid sequence for S-adenosylhomocysteinase with that of the dinucleotide binding region for other proteins, the sequence from amino acids 213 to 244 in rat liver S-adenosylhomocysteinase was proposed to be part of the NAD binding site (Ogawa, H., Gomi, T., Mueckler, M. M., Fujioka, M., Backlund, P. S., Jr., Aksamit, R. R., Unson, C. G., and Cantoni, G. L. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 719-723). A vector has been constructed that expresses S-adenosylhomocysteinase in Escherichia coli in the presence of isopropyl beta-D-thiogalactopyranoside by inserting the coding sequence of rat liver S-adenosylhomocysteinase cDNA downstream from the lac promoter of plasmid pUC118. The enzyme that is produced comprises as much as 10% of the soluble cellular proteins. The purified enzyme is a tetramer, contains 4 mol of tightly bound NAD, and has kinetic properties indistinguishable from those of the liver enzyme. Tryptic peptide mapping and NH2-terminal sequence analysis indicate that the recombinant enzyme is structurally identical to the liver enzyme except for the absence of the NH2-terminal blocking group. The rat liver enzyme has a blocked NH2-terminal alanine residue (Ogawa, H., Gomi, T., Mueckler, M. M., Fujioka, M., Backlund, P. S., Jr., Aksamit, R. R., Unson, C. G., and Cantoni, G. L. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 719-723). By oligonucleotide-directed mutagenesis mutant vectors have been generated that express proteins in which each of the glycines in the Gly-Xaa-Gly-Xaa-Xaa-Gly sequence of the putative NAD binding site (residues 219-224) was changed to valine. Immunoblot analysis of extracts of the cells transformed with these vectors reveals the presence of immunoreactive proteins with the subunit molecular weight of S-adenosylhomocysteinase. The mutant proteins have no catalytic activity, contain no bound NAD, and do not form the same quaternary structure as the recombinant S-adenosylhomocysteinase.     

Contributions of the substrate-binding arginine residues to maleate-induced closure of the active site of Escherichia coli aspartate aminotransferase.

Crystallography shows that aspartate aminotransferase binds dicarboxylate substrate analogues by bonds to Arg292 and Arg386, respectively [Jager, J, Moser, M. Sauder, U. & Jansonius, J. N. (1994) J. Mol. Biol., 239, 285-305]. The contribution of each interaction to the conformational change that the enzyme undergoes when it binds ligands via these residues, is assessed by probing mutant forms of the enzyme lacking either or both arginines. The probes used are NaH(3)BCN which reduces the cofactor imine, the reactive substrate analogue, cysteine sulfinate and proteolysis by trypsin. The unreactive substrate analogue, maleate, is used to induce closure. Each single mutant reacted only 2.5-fold more slowly with NaH(3)BCN than the wild-type indicating that charge repulsion by the arginines contributes little to maintaining the open conformation. Maleate lowered the rate of reduction of the wild-type enzyme more than 300-fold but had little effect on the reaction of the mutant enzymes indicating that the ability of this dicarboxylate analogue to bridge the arginines precisely makes the major contribution to closure. The R292L mutant reacted 20 times more rapidly with cysteine sulfinate than R386L but 5 x 10(4) times more slowly than the wild-type enzyme, consistent with the proposal that enzyme's catalytic abilities are not developed unless closure is induced by bridging of the arginines. Proteolysis of the mutants with trypsin showed that, in the wild-type enzyme, the bonds most susceptible to trypsin are those contributed by Arg292 and Arg386. Proteolysis of the next most susceptible bond, at Arg25 in the double mutant, was protected by maleate demonstrating the presence of an additional site on the enzyme for binding dicarboxylates.     

Pyrophosphorylases in potato. V. Allelic polymorphism of UDP-glucose pyrophosphorylase in potato cultivars and its association with tuber resistance to sweetening in the cold.

UDP-glucose pyrophosphorylase (UGPase) was cloned from six American and nine European potato (Solanum tuberosum L.) cultivars. Restriction mapping of the different UGPase-cDNAs with BamHI, HindIII, and EcoRI revealed that at least two mRNA populations were present in most cultivars. Staining for UGPase activity in nondenaturing gels of proteins extracted from developing potato tubers yielded two major isozymes that were highly active and appeared to be dimeric in nature. Following sodium dodecyl sulfate-polyacrylamide gel electrophoresis, all isozymes were disassociated into a single subunit with a molecular mass of 53 kD. Since UGPase has been demonstrated to be a single-copy gene in the haploid genome of potato (A.Y. Borovkov, P.E. McClean, J.R. Sowokinos, S.H. Ruud, G.A. Secor [1995] J Plant Physiol 147: 644-652), there must be allelic differences at the UGPase locus (chromosome 11). The two alleles, designated ugpA and ugpB, were identified by the absence and presence of a BamHI site, respectively. The relative band intensities of the two cDNA populations following polymerase chain reaction amplification and agarose gel electrophoresis were related to a potato cultivar's ability to resist sweetening when exposed to cold temperatures.     

Impact of heterologous expression of Escherichia coli UDP-glucose pyrophosphorylase on trehalose and glycogen synthesis in Corynebacterium glutamicum

Trehalose is a disaccharide with a wide range of applications in the food industry. We recently proposed a strategy for trehalose production based on improved strains of the gram-positive bacterium Corynebacterium glutamicum. This microorganism synthesizes trehalose through two major pathways, OtsBA and TreYZ, by using UDP-glucose and ADP-glucose, respectively, as the glucosyl donors. In this paper we describe improvement of the UDP-glucose supply through heterologous expression in C. glutamicum of the UDP-glucose pyrophosphorylase gene from Escherichia coli, either expressed alone or coexpressed with the E. coli ots genes (galU otsBA synthetic operon). The impact of such expression on trehalose accumulation and excretion, glycogen accumulation, and the growth pattern of new recombinant strains is described. Expression of the galU otsBA synthetic operon resulted in a sixfold increase in the accumulated and excreted trehalose relative to that in a wild-type strain. Surprisingly, single expression of galU also resulted in an increase in the accumulated trehalose. This increase in trehalose synthesis was abolished upon deletion of the TreYZ pathway. These results proved that UDP-glucose has an important role not only in the OtsBA pathway but also in the TreYZ pathway.     

Cloning of cDNA for UDP-glucose pyrophosphorylase and the expression of mRNA in rice endosperm

Rice endosperm UDP-glucose pyrophosphorylase (UGPase) cDNA clones were isolated by screening a lambda ZAP II library prepared from poly (A(+)) RNA of japonica rice (cv Sasanishiki) endosperm with a probe of potato UGPase cDNA. One cDNA clone, possessing about 1,700 nucleotides, contained the complete open reading frame of rice UGPase. At the nucleotide-sequence level, the UGPase cDNA of rice endosperm had high homology with the UGPase cDNA of barley endosperm (84%) and potato tuber (71%). The calculated molecular weight (50 kDa) agrees with the value determined by SDS-PAGE (51 kDa). At the amino-acid sequence level, rice UGPase has high homology with the UGPase of barley (92%) and potato (85%). The enzyme contained conserved sequence elements which are thought to be involved in substrate binding and catalytic activity. A Southern-blot analysis indicated that the gene existed as a single copy. Expression of the enzyme in rice endosperm examined by Northern-blot analysis was high at 10-15 days after heading.     

Studies on determination of active site amino acid residues in glyoxylate synthetase from potato tuber chloroplasts

A homogeneous preparation of glyoxylate synthetase from greening potato tubers was used to study the functional role of disulphide groups, lysine and tryptophan residues in enzyme catalysis. The formation of a thioisoindole derivative was demonstrated by spectral analysis of the reduced and o-phthalaldehyde-treated enzymes. o-Phthalaldehyde modification resulted in about a 25 % loss of tryptophan emission at 336 nm and the appearance of a 410-nm emission peak characteristic of a thioisoindole. Ferrous iron was capable of generating thiol groups and addition of substrate resulted in a faster disappearance of these thiols. The optimal time for maximum glyoxylate synthesis by glyoxylate synthetase paralleled the disappearance of these thiols. Involvement of lysine and tryptophan residues in the enzyme reaction was demonstrated by the inhibition of activity by pyridoxal 5′-phosphate and dimethyl(2-hydroxy 5-nitrobenzyl) sulphonium bromide (DMHNB), respectively. Pyridoxal phosphate strongly and reversibly inhibited glyoxylate synthetase, and substrate and metal ion provided significant protection against inhibition. The results suggest that the lysine residue may be at or near the active binding site. The lysyl residue formed a Schiff base with pyridoxal phosphate which was stabilised by NaBH₄. Glyoxylate synthetase was also irreversibly inactivated by a tryptophan selective reagent, DMHNB, while substrate provided substantial protection against inactivation. Kinetic analysis and correlation of the spectral data at 410 nm indicated that complete inactivation by DMHNB resulted from the modification of 5 tryptophan residues/subunit, of which one was essential for activity. The available evidence suggests a possible concerted action of enzyme disulphides, ferrous iron, lysine and aromatic amino acid residues in the synthesis of glyoxylate by this enzyme.     

UDP-glucose 4-epimerase from Saccharomyces fragilis. Presence of an essential arginine residue at the substrate-binding site of the enzyme

UDP-glucose 4-epimerase from Saccharomyces fragilis was inactivated by the arginine-specific reagents phenylglyoxal, 1,2-cyclohexanedione, and 2,3-butanedione following pseudo first order reaction kinetics. The reaction order with respect to phenylglyoxal was 1.8 and that with respect to the other two diones was close to unity. Protection afforded by substrate and competitive inhibitors against inactivation by phenylglyoxal and the reduced interaction of 1-anilinonaphthalene 8-sulfonic acid, a fluorescent probe for the substrate-binding region after phenylglyoxal modification, suggested the presence of an essential arginine residue at the substrate-binding region. Experiments with [7-14C]phenylglyoxal in the presence of UMP, a ligand known to interact at the substrate-binding region, showed that only the arginine residue at the active site could be modified by phenylglyoxal. The characteristic coenzyme fluorescence of the yeast enzyme was found to be enhanced three times in phenylglyoxal-inactivated enzyme suggesting the incorporation of the phenyl ring near the pyridine moiety of NAD.     

Location of lysyl residues at the allosteric site of fructose 1,6-bisphosphatase

Various flavin analogs were used as alternate substrates or competitive inhibitors to characterize the FMN binding sites of the NADH- and NADPH-specific FMN oxidoreductases from Beneckea harveyi. Several polyhydroxyl compounds were found to be poor competitive inhibitors for the FMN sites of these enzymes. The FMN binding sites of the two enzymes were found to be quite similar. The NADH:FMN oxidoreductase binds FMN exclusively through the isoalloxazine ring. The methyl groups at positions 7 and 8 contribute significantly to this binding. Utilizing lumichrome as a competitive inhibitor of the FMN binding site and AMP as a competitive inhibitor of the NADH binding site, we were able to determine that the NADH:FMN oxidoreductase forms an active ternary complex with NADH binding first in an ordered mechanism. The NADPH oxidoreductase also binds FMN primarily through the isoalloxazine ring. Unlike their participation in reaction with the NADH-specfic enzyme, the methyl groups at positions 7 and 8 are not involved in binding. There was no significant binding of the ribityl phosphate moiety with either enzyme. Both enzymes have lower K_m values for lumiflavin than FMN.     

UDP-glucose:sterol glucosyltransferase: cloning and functional expression in Escherichia coli/

Steryl glucosides are characteristic lipids of plant membranes. The biosynthesis of these lipids is catalyzed by the membrane-bound UDP-glucose:sterol glucosyltransferase (EC 2.4.1.173). The purified enzyme (Warnecke and Heinz, Plant Physiol 105 (1994): 1067–1073) has been used for the cloning of a corresponding cDNA from oat (Avena sativa L.). Amino acid sequences derived from the amino terminus of the purified protein and from peptides of a trypsin digestion were used to construct oligonucleotide primers for polymerase chain reaction experiments. Screening of oat and Arabidopsis cDNA libraries with amplified labeled DNA fragments resulted in the isolation of sterol glucosyltransferase-specific cDNAs with insert lengths of ca. 2.3 kb for both plants. These cDNAs encode polypeptides of 608 (oat) and 637 (Arabidopsis) amino acid residues with molecular masses of 66 kDa and 69 kDa, respectively. The first amino acid of the purified oat protein corresponds to the amino acid 133 of the deduced polypeptide. The absence of these N-terminal amino acids reduces the molecular mass to 52 kDa, which is similar to the apparent molecular mass of 56 kDa determined for the purified protein. Different fragments of these cDNAs were expressed in Escherichia coli. Enzyme assays with homogenates of the transformed cells exhibited sterol glucosyltransferase activity.     

Expression of human angiotensinogen cDNA in Escherichia coli

Human angiotensinogen cDNA clones were isolated from a human liver library. Nucleotide sequence analysis of these cDNA clones revealed that position 1075 in the messenger RNA, which is part of a PstI recognition sequence, is different from the published sequence (Kageyama, R., Ohkubo, H., and Nakanishi, S. (1984) Biochemistry 23, 3603-3609). This change results in an altered amino acid at this position in the corresponding protein sequence and suggests possible restriction fragment length polymorphism. The full length human angiotensinogen cDNA was constructed from partial cDNA clones and ligated into an isopropyl-1-thio-beta-D-galactopyranoside inducible bacterial expression vector pUC9 to develop expression plasmid pUCHAG27. This plasmid permitted the synthesis of human angiotensinogen in Escherichia coli. The recombinant bacteria overproduced a 53-kDa protein which was recognized by anti-human angiotensinogen antibodies. The synthesis of this protein was greatly increased upon induction with isopropyl-1-thio-beta-D-galactopyranoside. The chimeric protein, almost identical to human angiotensinogen, was partially purified by ammonium sulfate fractionation and gel filtration on Sephadex G-100. Human kidney renin was shown to enzymatically cleave this recombinant protein to produce des-(angiotensin I)-angiotensinogen and a small polypeptide. Thus, we provide evidence that recombinant human angiotensinogen synthesized through E. coli is biologically active and serves as a substrate for human renin.     

ADPglucose pyrophosphorylase: Evidence for a lysine residue at the activator site of the Escherichia coli B enzyme

Pyridoxal-P can be covalently linked to $\text{E. coli}$ B ADPglucose pyrophosphorylase by reduction with sodium borohydride. The modified enzyme is almost fully active when less than 1 mole of pyridoxal-P is incorporated per mole of enzyme subunit and is no longer dependent on the presence of allosteric activators in reaction mixtures for high activity. The allosteric activators, fructose-P₂ or hexanediol 1,6 bisphosphate, decrease the incorporation of pyridoxal-P into enzyme suggesting that the pyridoxal-P is linked at or near the allosteric activator binding site. Acid hydrolysis of the modified enzyme yields pyridoxyllysine suggesting that the epsilon amino group of lysine is functional in the binding of the allosteric activators of the enzyme.