J. Biochem. 112, 258-265 (1992)

Leucine Dehydrogenase from Bacillus stearotherinophilus: Identification of Active-Site Lysine by Modification with Pyridoxal Phosphate1

Takahiro Matsuyama,*,2 Kenji Soda, ** Toshio Fukui,* and Katsuyuki Tanizawa*,3 *The Institute of Scientific and Industrial Research , Osaka University,Ibaraki, Osaka 567; and **Institute for Chemical Research, Kyoto University, Uji, Kyoto 611

Received for publication, March 16, 1992

We have constructed an efficient expression plasmid for the dehydrogenase gene

previously cloned from Bacillus stearothermophilus. The recombinant was overproduced in Escherichia coli cells to a level of more than 30% of the total soluble protein upon induction with isopropyl ƒÀ-D-thiogalactopyranoside. The enzyme could be readily purified to homogeneity by heat treatment and a single step of ion-exchange chromatography. The purified enzyme was inactivated in a time-dependent manner upon incubation with pyridoxal 5•L-phosphate (PLP) followed by reduction with sodium borohy dride. The inactivation was completely prevented in the copresence of L-leucine and NAD+. Concomitantly with the inactivation, several molecules of PLP were incorporated into each subunit of the hexameric enzyme. Sequence analysis of the fluorescent peptides isolated from a proteolytic digest of the modified protein revealed that Lys80, Lys9l, Lys206, and Lys265 were labeled. Among these residues, Lys8O was predominantly labeled and, in the presence of L-leucine and NAD+, was specifically protected from the labeling. Furthermore, a linear relationship of about 1:1 was observed between the extent of inactivation and the amount of PLP incorporated into Lys80. A slightly active mutant enzyme, in which Lys8O is replaced by Ala, was not inactivated at all by incubation with PLP, showing that the inactivation is correlated with the labeling of only Lys80. Lys8O is conserved in the corresponding regions of all the amino acid dehydrogenase sequences reported to date. These results suggest that Lys8O is located at the and plays an important role in the catalytic function of leucine dehydrogenase.

Leucine dehydrogenase [EC 1.4.1.9] is an NAD+-depen and NH,', ƒ¿-keto-iso-caproate, and NADH are released in dent that catalyzes the reversible deamina that order (1, 4). tion of L-leucine and some other branched-chain L-amino Although the three-dimensional structures of several acids to their keto analogues. The enzyme, occurring mainly NAD(P)+-dependent dehydrogenases have been elucidated in Bacillus species (1), functions catabolically in the in detail, none of those acting on amino acids has been bacterial metabolism of branched-chain L-amino acids (2), analyzed by X-ray crystallography. Further, little is known and has been purified to homogeneity from B. sphaericus about the molecular mechanisms of amino acid dehydro (3) and B. stearothermophilus (4). The gene coding for the genase . In order to shed light on the active-site B. stearothermophilus enzyme has been cloned, sequenced, structure and reaction mechanism of amino acid dehydro and expressed in Escherichia coli (5). The polypeptide genases, we have first constructed an efficient expression deduced from the nucleotide sequence is composed of 429 plasmid for the cloned gene coding for B. stearothermo amino acid residues with a calculated molecular weight of philus leucine dehydrogenase, and then have attempted to about 47,000 and corresponds to a subunit of the hexameric identify an active-site lysine residue by modification with enzyme, whose structure has been revealed by a small pyridoxal 5•L-phosphate (PLP). In this paper, we present angle X-ray scattering study (6). The reaction catalyzed by evidence that Lys80 conserved in the Gly-rich consensus leucine dehydrogenase proceeds according to the ordered sequence is an active-site residue in leucine dehydrogenase, bi-ter mechanism, in which NAD+ and L-leucine are bound presumably being involved in the catalytic function.

1 This study was supported in part by Grants-in-Aid for Scientific EXPERIMENTAL PROCEDURES Research from the Ministry of Education, Science and Culture of Japan, and a grant from the Sapporo Bioscience Foundation. Construction of Overexpression Plasmid-The gene cod 2 Present address: Products Formulation Research Laboratory , Tana ing for leucine dehydrogenase of B. stearothermophilus was be Pharmaceutical Co., Kashima, Yodogawa-ku, Osaka 532.3 previously cloned into the Sail site of pBR322, the con To whom correspondence should be addressed. structed plasmid being designated as pICD2 (5). Since the Abbreviations: HEPES, N-2-hydroxyethylpiperazine-N•L-2-ethanesul expression level of the enzyme in E. coli cells carrying fonic acid; HPLC, high-performance liquid chromatography; IPTG, isopropyl ƒÀ-D-thiogalactopyranoside; kbp, kilobase pair(s); PLP, pICD2 was not high, a new overexpression plasmid was constructed using Bluescript II (Stratagene) as a vector. pyridoxal 5•L-phosphate; PTH, phenylthiohydantoin.

258 J. Biochem. Active-Site Lysine in Leucine Dehydrogenase 259

Briefly, the 0.89-kbp Sail fragment encoding the C-termi lysine moiety was measured upon excitation at 330 rim nal half of the enzyme was excised from pICD2 and inserted using known amounts of PLP as a standard, which was into the XhoI site in Bluescript II . The 1.3-kbp XhoI-Ncol modified with excess N-ƒ¿-acetyllysine (Sigma) followed fragment from pICD2 was then inserted into the Sail and by NaBH4 reduction. NcoI sites of the above plasmid to produce the final plasmid Isolation of Labeled Peptides-The enzyme (50 nmol) designated as pBLeuDH (Fig. 1). modified with 2mM PLP for 30min and reduced with Enzyme and Protein Assays-The standard assay mix NaBH, as described above was carboxymethylated, exten ture (1.0ml) contained 0.1M sodium carbonate buffer (pH sively dialyzed against water, and lyophilized. The protein 10.5), 10mM L-leucine, and 1.25mM NAD+ . The reaction was suspended in 0.1ml of 0.1M ammonium bicarbonate was started by the addition of the enzyme , and the increase i buffer (pH 8.0) containing 4M urea, and then diluted with n absorbance at 340 rim was continuously monitored with 0.1ml of 0.1M ammonium bicarbonate. N-Tosyl-L-phe a spectrophotometer at 55°C (5). One unit of the enzyme is nylalanine chloromethylketone-treated trypsin (Millipore) defined as the amount of enzyme that produces 1 p mol of was added to the turbid solution in a 1:50 (mol/mol) ratio NADH per min under the above conditions. Specific activ of protease to . Digestion was performed at 37"C ity is expressed as units per milligram of protein. Protein for 24 h with a second addition of trypsin, similar to the concentration of the purified enzyme was derived from the first, after a period of 12 h. The peptides were separated on absorbance at 280nm. The absorption coefficient (A1mg/mlmg= a Gilson HPLC system equipped with a Cosmosil C,, 0.851) at 280mn determined by amino acid analysis was reverse-phase column using a solvent system of 0.1% used throughout. trifluoroacetic acid (A) and 0.088% trifluoroacetic acid Enzyme Purification-E. coli JM109 cells carrying pB containing 60% acetonitrile (B). A 60-min linear gradient LeuDH were grown at 37•Ž for 12 h in 1 liter of Luria broth from 0 to 80% B was used to elute peptides at a flow rate of (1.0% tryptone, 1.0% NaCl, and 0.5% yeast extract, pH 1.0ml/min. The absorbance at 215 rim and the fluorescence 6.8) containing 50ƒÊg/ml sodium ampicillin and 0.5mM (excitation at 330nm and emission at 395 nm) of effluents IPTG. Cells harvested by centrifugation werer washed with were continuously monitored. The amount of PLP bound to 500ml of 0.85% NaCl. each labeled peptide was calculated from the area of Step 1: The washed cells (about 4 g, wet weight) were fluorescent peaks using known amounts of the reduced suspended in 20ml of 10mM potassium phosphate buffer PLP-N-ƒ¿-acetyllysine as a standard. (pH 7.2) containing 0.01% 2-mercaptoethanol (buffer A) Amino Acid Composition and Sequence Analysis supplemented with 0.1mM phenylmethanesulfonyl fluo - Amino acid composition was determined, after hydrolysis ride. After the cells were disrupted in a French pressure with 6N HCl in an evacuated tube, with a Hitachi 835 amino cell, the resultant solution was centrifuged at 46,000•~g for acid analyzer using o-phthalaldehyde. An authentic N-ƒÃ- 30min. pyridoxyllysine was prepared from N-ƒ¿-acetyllysine by Step 2: The supernatant solution (21 ml) was incubated the procedure reported previously (9). The amino acid at 70°C for 30min, then cooled on ice, and centrifuged as sequence was determined with an Applied Biosystems above to remove the denatured proteins. The supernatant Model 477A protein sequencer linked with an Applied was dialyzed at 4°C for 6 h against 5 liters of buffer A. Biosystems Model 120A PTH analyzer. Step 3: The enzyme solution was applied to a column Determination of Apparent Dissociation Constant-A (2.5x20cm) of DEAE-Toyopearl 650M (Toyo Soda) pre solution (500,u l) containing various concentrations of PLP equilibrated with buffer A. Proteins were eluted with a (0.2-1.0mM) in 50mM HEPES buffer (pH 7.8) and the 1-liter linear gradient of KCl (0.1-0.2M) in the same enzyme (23 ƒÊM) was preincubated at 25°C for 30 min in a buffer. The active fractions were pooled (about 150 ml) and 0.7-m1 cuvette. The same solution without the enzyme was concentrated by ultrafiltration through an Amicon PM-10 placed in a reference cell, and the difference absorption membrane to about 10 ml. The homogeneity of the prepa spectra were measured in the 340-500nm region with a ration was examined by SDS-PAGE with a 12.5% poly recording spectrophotometer. The formation of a Schiff acrylamide gel (7). base between the 4-formyl group of PLP and an E -amino Modification with PLP-The purified enzyme (56.5ƒÊM) group(s) of lysine residue was monitored by following the was incubated at 30°C with various concentrations (0.1-10 increase in absorbance at 434 nm. The apparent dissocia mM) of PLP in 50mM HEPES buffer (pH 7.8) in a final tion constant (Kapp) was calculated by using the following volume of 0.1 ml. At various time intervals (2-30 min), equation (10): 10-ƒÊ1 aliquots of the reaction mixture were withdrawn and ƒ¢ A=LAmax•E[PLP]0/(Kapp+ [PLP]0), mixed with a freshly prepared solution of NaBH4 (final conc., 20mM) to stop the reaction. The solution was kept at where [PLP] o is the initial concentration of PLP, JA is the 30°C for 10 min and an aliquot was used for measurement increase in absorbance at 434nm, and ƒ¢Amax(2.55x103 of the enzyme activity. The remaining solution was filtered M-1) is the maximum increase in absorbance observed three times through Amicon Centricon-30 cartridges with when all the enzyme molecules are presumed to be com addition of 1 ml each of 50mM HEPES buffer (pH 7.8) to plexed with PLP. remove the excess reagents. The protein concentration was NADH Binding Assay-The binding of NADH with the measured by the method of Bradford (8) using the unmodi enzyme was measured by the gel permeation method (11). fied enzyme as a standard, and the amount of PLP incorpo The enzyme (32ƒÊM), unlabeled or labeled with 2mM PLP rated into the enzyme was measured fluorophotometrically for 30 min as above, was incubated with 50ƒÊM NADH and after denaturation of the enzyme protein in 0.1 M HEPES applied to a column (5 x 200 mm) of Sephadex G-50 (fine) buffer (pH 7.8) containing 8 M urea. The fluorescence previously equilibrated with 10mM potassium phosphate emission at 395 rim due to the reduced phosphopyridoxyl buffer (pH 7.8) containing 50ƒÊM NADH. The enzyme was

Vol. 112, No. 2, 1992 260 T . Matsuyama et al.

eluted with the same buffer at a constant flow rate of 0.4 ml/min, and the absorbance at 340 run was continuously RESULTS monitored to determine the amount of NADH bound to the enzyme from the area of the trough. Construction of Overexpression Plasmid and Purification Site-Directed Mutagenesis-The HindIII-KpnI fragment of Leucine Dehydrogenase-The gene coding for leucine (2.0 kbp) from pBLeuDH, which contains the entire coding dehydrogenase from B. stearothermophilus has already region for leucine dehydrogenase, was subcloned into phage been cloned (5). However, E. coli cells transformed with M13 mp19 RF DNA. E. coli BW313 (dut- ung-) cells were the resultant plasmid pICD2 produced the enzyme corre transfected with the M13 phage, and the single-stranded sponding to only about 3% of the total soluble protein in the phage DNA containing uracil was purified from the culture cell extract (5). This was probably because the plasmid had supernatant. An oligonucleotide primer, 5•L-TGACCGTT been derived from pBR322, usually used for the purpose of

GC GCCCCCG-3•L, was designed to be complementary to gene cloning but not for gene expression. The presence of this single-stranded template DNA and to contain appro three Sail sites in pICD2 also appeared unfavorable for future site-directed mutagenesis (see below). Therefore, priate mismatching bases (asterisked) in the codon for using an expression phagemid vector Bluescript II, we have Lys80 (underlined), and was synthesized with an Applied derivatized pICD2 into pBLeuDH containing only a unique Biosystems DNA synthesizer Model 381. Synthesis of Sail site (Fig. 1). mutant DNA and selection were performed by the method A crude extract from E. coli JM109 cells carrying of Kunkel et al. (12), using a commercial kit (Mutan-K; pBLeuDH thus constructed and being grown in the presence Takara Shuzo). The sequence of nucleotides 1-666 corre of IPTG exhibited a high-level expression of leucine sponding to the region from the translational initiation dehydrogenase with a specific activity of about 35 units/mg codon ATG to the Sail site in the mutant gene obtained was protein [cf. 3 units/mg in the extract from E. coli C600/ confirmed by the dideoxy chain termination method (13). pICD2 (5)]. On the basis of the specific activity of the The 1.1-kbp HindIII-SalI fragment containing the mutat purified enzyme (112 units/mg), the amount of the enzyme ed site was excised from the double-stranded M13 mp19 in the cell extract corresponded to more than 30% of that of phage DNA, ligated into the HindIII-SalI site of pBLeuDH, the total soluble protein (see also Fig. 2). The enzyme was and transformed into E. coli JM109 cells. The Ala-for found to be inducibly formed by the addition of IPTG; the - Lys80 mutant enzyme produced was purified to homogene expression level in the absence of IPTG was less than 3% of ity by a similar procedure to that for the wild-type enzyme. the total soluble protein. Thus, the lac promoter in the vector Bluescript II functioned efficiently for expression of the enzyme gene placed 3•L-downstream. The high-level expression of the enzyme in E. coli led to the establishment of a very simple and rapid purification method for the recombinant enzyme, which consisted of heat treatment and a single column chromatographic step as described under "EXPERIMENTAL PROCEDURES." The overall recovery of the enzyme in the purification was about 70%, and the homogeneity of the purified enzyme was >95% as judged by SDS-PAGE (Fig. 2). Direct Edman degradation of the purified enzyme afforded an N-terminal partial sequence of Met Glu-Leu-Phe-Lys-Tyr-Met-Glu - Thr-Tyr-, which coincided with that predicted from the DNA sequence (5). Inactivation by PLP-When incubated with PLP follow

Fig. 2. SDS-PAGE of the recombinant leucine dehydrogenase. The superna tants from cell lysates of E. coli clones (pICD2 and pBLeuDH) and the purified [Wild-type and the Ala-for Fig. 1. Construction scheme of overexpression plasmid for - Lys80 mutant (K80A) ] were analyzed in leucine dehydrogenase. Only relevant restriction sites are indicat a 12.5% polyacrylamide gel. Standard ed, and lengths of DNA are shown arbitrarily. Arrows corresponding proteins (Std) are phosphorylase b (Mr to the coding region for leucine dehydrogenase are shown at approxi 94,000), bovine serum albumin (Mr mate positions. The restriction fragments are shown by bars and 67,000), egg albumin (Mr 43,000), car marked differently for an easier understanding of the cloning proce bonic anhydrase (Mr 30,000), and trypsin dure; lines are the regions from vector DNA. The details are inhibitor (Mr 20,000) (from top to bot described under "EXPERIMENTAL PROCEDURES." tom).

J. Biochern. Active-Site Lysine in Leucine Dehydrogenase 261

Fig. 4. Difference absorption spectrum between the enzyme plus PLP and PLP alone. (a) The difference absorption spectrum of the solution containing 0.2mM PLP and 23,u M enzyme in 50mM HEPES buffer (pH 7.8) against the solution without the enzyme was measured in the wavelength region from 340 to 500 nm. (b) The increases in absorbance at 434 nm in the difference spectra obtained with various concentrations of PLP were plotted against the PLP concentration. The curve is the least-squares best fit with Kapp=0.92

Fig. 3. Inactivation of leucine dehydrogenase by PLP. (a) Time mM. course. The wild-type enzyme (57ƒÊM) (•œ) or the Ala-for-Lys80 mutant enzyme (57ƒÊM) (•œ) was incubated at 30°C with 2mM PLP in 50mM HEPES buffer (pH 7.8) in a final volume of 0.1ml. At each indicated time, an aliquot was withdrawn and reduced with 20mM NaBH4, and the remaining activity was assayed. (b) Effect of sub strates and coenzymes. The above inactivation mixtures were sup plemented with 10mM L-leucine plus 2mM NAD+(•£), 2mM NADH (h), 10mM L-leucine (•¡), 2mM NAD+ or 10mM ƒ¿-keto-iso caproate (• ), or 10MM L-glutamate plus 2mM NAD+ (•›). The control reaction without these additives (•œ) was the same as in (a). ed by reduction with sodium borohydride, the enzyme was inactivated in a time-dependent manner, reaching a plateau after about 20 min (Fig. 3). The extent of inactivation depended on the concentration of PLP; about 90% of the original activity was lost by incubation with 4mM PLP. In addition, the inactivation by PLP was also dependent on pH, the enzyme being inactivated most effectively around pH 8.5. The optimum pH for inactivation may reflect the pKaof a reactive lysine residue(s) that is essential for activity. PLP is known to react with a -amino groups of lysine residues in proteins to form a Schiff base that can be Fig. 5. Binding assay for NADH by gel filtration. The enzyme reduced by sodium borohydride (14). Actually, the differ unmodified (a) or modified (b) with PLP was mixed with 50 p M ence spectrum between the enzyme plus PLP and PLP NADH and applied to a Sephadex G-50 column equilibrated with 50ƒÊ M NADH as described under "EXPERIMENTAL PROCEDURES." alone exhibited an absorption maximum at 434 nm due to Absorbance at 340 nm was continuously monitored. The area of the the formation of the Schiff base (Fig. 4a). The apparent trough from the baseline corresponds to the decrease of free NADH dissociation constant for PLP of the Schiff base was and is equal to the amount bound to the enzyme. The higher determined to be 0.92mM from the increases in absorb absorption peak of the enzyme protein in (b) as compared to that in (a) ance at 434 nm in the difference spectra with various is due to the additional absorption of bound PLP at 340 nm in the concentrations of PLP as described under "EXPERIMENTAL modified enzyme. PROCEDURES" (Fig. 4b). Protection by Coenzymes and Substrates-The protective effect by substrates and coenzymes on the enzyme inactiva The enzyme was completely protected from inactivation by tion by PLP was investigated to elucidate where PLP was the concomitant presence of both L-leucine and NAD+. bound. The enzyme was incubated with 2mM PLP in the NADH also considerably protected the enzyme, whereas presence of 10mM substrate or 2mM coenzyme (Fig. 3b). L-leucine alone protected it only moderately and ƒ¿-keto

Vol. 112, No. 2, 1992 262 T. Matsuyama et al.

Fig. 6. Stoichiometry of inactivation and labeling. The enzyme was incubated with various concentrations of PLP for 30 min or with 2mM PLP for various times and reduced with NaBH,. The remaining activities were plotted against the amounts of PLP incorporated into the enzyme subunit (•œ) or Lys8O (•›) determined from the areas of the Fig. 7. HPLC elution profiles of tryptic digest of PLP-labeled fluorescence peak of T, in HPLC of tryptic digests of the labeled leucine dehydrogenase. The tryptic digests of the enzyme labeled enzyme (see Fig. 7). Further details are given in the text. with PLP in the absence (a) or presence (b) of 10 MM L-leucine and 2 mM NAD+ were chromatographed on a Cosmosil C18 column. The peptides were detected by continuous monitoring of the absorbance at 215 nm (not shown) and the fluorescence (excitation at 330 nm and iso-caproate or NAD+ alone offered little protective effect. emission at 395 nm). The peaks were designated as T,-T, in order of These results can be explained by the ordered sequential elution. mechanism, in which L-leucine is bound to the enzyme following NAD+ in the oxidative deamination, and NADH is bound to the enzyme before the others in the reductive described under "EXPERIMENTAL PROCEDURES." Fluores amination (1, 4). L-Glutamate, a non-substrate amino acid, cent peptides were eluted as one major peak accompanied gave no protection against inactivation by PLP. by three minor peaks (designated T1-T4 in order of elution, The marked protection of the enzyme from inactivation Fig. 7a). In contrast, the modification by PLP in the by PLP in the presence of NAD+ plus L-leucine or NADH copresence of 10 MM L-leucine and 2mM NAD+ yielded a has suggested that the amino group(s) whose modification labeling pattern, in which only the major peak (T,) in the with PLP leads to the enzyme inactivation is located at the digest of the unprotected enzyme was dramatically de for substrate and/or coenzyme. To examine creased (Fig. 7b). The elution profiles were reproducible, whether the inactivated enzyme still binds NADH, the showing that the multiple peaks were not due to incomplete amounts of NADH bound to the enzyme protein were digestion. After the labeled peptides containing the fluores measured by the gel permeation method (11) with a cent phosphopyridoxyl moiety were further purified by Sephadex G-50 column (Fig. 5). The amount of NADH re-chromatography on a reverse-phase column, their amino bound to the modified enzyme (having 28% of its original acid compositions were analyzed. All four peptides (T1-T4) activity) was 1.08 mol/mol of enzyme subunit, which was were found to contain an amino acid that does not corre almost identical with that (1.09 mol/mol of enzyme sub spond to any natural amino acid in the automated amino unit) bound to the unmodified enzyme. This indicates that acid analysis, but corresponds to the authentic N-ƒÃ- the modification with PLP does not diminish the binding pyridoxyllysine eluted just after histidine and much before ability of the enzyme for NADH. arginine, although its recovery in the hydrolysates was Stoichiometry of Inactivation and Labeling-The en rather low (5-30%) as compared to other amino acids. zyme was incubated with various concentrations (0.2-5 Amino acid sequences of the four peptides were determined mM) of PLP for 30 min and reduced, then the excess with a protein sequencer, and the results are summarized in reagents were removed by centrifugal ultrafiltration. The Table I. The PTH derivative of N-ƒÃ-pyridoxyllysine is not fluorescence derived from the pyridoxyllysine moiety, identifiable with a protein sequencer (15). However, when residual activities, and protein concentrations were mea the partial sequences were compared with the complete sured. The residual activities were plotted against the sequence of the enzyme from B. stearothermophilus (5), all amounts of PLP incorporated (Fig. 6). The line deviated the unidentified residues corresponded to particular lysine greatly from linearity at high degrees of inactivation and residues. These structures were also consistent with the the molar ratio of bound PLP to the enzyme subunit was far amino acid compositions. Therefore, it was concluded that above unity. This result shows that PLP is incorporated Lys265, Lys9l, Lys206, and Lys80 were the lysine resi into several lysine residues other than that (or those) dues labeled with PLP in T1-T4, respectively. Lys80 was related to the activity. predominantly labeled, and its labeling was preferentially Identification of Labeled Residues-To identify the prevented in the copresence of L-leucine and NAD+, sug lysine residue(s) labeled by PLP, the enzyme was modified gesting that Lys80 is located at the active site of the with PLP followed by carboxymethylation and then digest enzyme. ed with trypsin. The digest was separated by HPLC as Stoichiometry of Inactivation by PLP and Labeling of

J. Biochem. Active-Site Lysine in Leucine Dehydrogenase 263

TABLE I. Structures of labeled peptides. One letter abbreviations are used. X represents an unidentified residue that was concluded to be the labeled lysine residue. Details of the sequence determinations are given in the text. Positions are in the complete sequence of leucine dehydrogenase from B. stearothermophilus (5).

Fig. 8. Comparison of the leucine dehydrogenase par tial sequence containing Lys8O with corresponding sequences of other amino acid dehydrogenases. Lys80 of leu cine dehydrogenase is shown by an asterisk, and conserved res idues are boxed. Abbreviations: LeuDH, leucine dehydrogenase; PheDH, phenylalanine dehydro genase; G1uDH, glutamate de hydrogenase; and AlaDH, ala nine dehydrogenase.

Lys80-The enzyme was modified with PLP to various extents by changing the incubation time, and the amount of DISCUSSION PLP incorporated into Lys80 was determined after diges tion with trypsin and separation by HPLC of the fluorescent PLP has been used to modify reactive lysine residues in peptide (T4) containing Lys80. The amounts calculated various proteins including several NAD(P)+-dependent from the areas of fluorescence due to peak T, and corrected dehydrogenases (16-22), though the modifications are not per mol of enzyme subunit were plotted against the residual necessarily active-site-directed (23). The inactivation by activities after the modification. As shown in Fig. 6, a PLP of leucine dehydrogenase from B. sphaericus has also straight line was obtained, extrapolation of which to 0% been reported, but the modified lysine residue(s) has not remaining activity gave a value of approximately 0.9 for been identified (24). The present study has shown that the the molar amount of PLP incorporated into Lys80 per mol recombinant enzyme from B. stearothermophilus is effec of enzyme subunit. This result indicates that the modifi tively inactivated by PLP with concomitant modification of cation with PLP of Lys80 is directly related to the loss of four different lysine residues. Lys80 was predominantly enzyme activity and suggests that Lys80 is an active-site labeled and was selectively protected from the modification residue, presumably playing an essential role in the cataly in the copresence of substrate and coenzyme. Furthermore, sis. The other labeled lysine residues (Lys9l, Lys206, and the binding to Lys80 of about 1 mol of PLP per mol of Lys265) may be located at the surface of the enzyme enzyme subunit corresponded to the complete loss of molecule and are functionally unimportant. activity. These results lead to the conclusion that only Inactivation of the Ala-for-Lys8O Mutant Enzyme by Lys80 is located at or near the active site of the enzyme and PLP-To confirm that the inactivation of the enzyme by the other three lysine residues may be located at the PLP is correlated with the covalent modification of Lys80, surface of the enzyme molecule without having direct roles the lysine residue was replaced with Ala by site-directed in the enzymic functions. mutagenesis, and the mutant enzyme was treated with By comparison of the sequence of B. stearothermophilus PLP. The Ala-for-Lys80 mutant enzyme purified to homo leucine dehydrogenase with those of proteins registered in geneity (Fig. 2) had a specific activity of 1.7 units/mg the National Biomedical Research Foundation (NBRF) protein, which is markedly lower than that (112 units/mg) protein sequence data bank (25), we have found that Lys80 of the wild-type enzyme but is detectable under the in leucine dehydrogenase is conserved in the corresponding standard assay conditions with a large amount of the regions of all other amino acid dehydrogenases sequenced enzyme. When incubated with 2mM PLP followed by to date (Fig. 8); glutamate dehydrogenases from bovine reduction with sodium borohydride, the Ala-for-Lys80 liver (26), chicken liver (27), human liver (28), E. coli (29, mutant enzyme showed no loss of activity at all, whereas 30), Neurospora (31, 32), and yeast (33), phenylalanine the wild-type enzyme was inactivated to 40% of the initial dehydrogenases from B. sphaericus (34), Sporosarcina activity (Fig. 3a). The result clearly demonstrates that the ureae (35), and Thermoactinomyces intermedius (36), and enzyme inactivation by PLP is solely due to the modifi alanine dehydrogenases from B. sphaericus and B. stear cation of Lys80. othermophilus (37). It is noteworthy that this lysine residue occurs in a Gly-rich tetrapeptide sequence, (G or H)-G-(G or A or S)-K (38), the flanking regions of which are also highly conservative (Fig. 8). This finding suggests that

Vol. 112, No. 2, 1992 264 T . Matsuyama et al. the region constitutes a part of the catalytic site in amino ities when measured under the standard assay conditions. acid dehydrogenases. This apparently conflicts with the suggestion in the present Before undertaking the chemical modification with PLP , study that Lys80 is an indispensable residue at the active we tested the affinity labeling reagents, adenosine poly site. However, we have found that the reaction rate by phosphopyridoxals, developed in this laboratory (see Ref. these mutant enzymes is logarithmically enhanced with the 39 for a recent review), for specific labeling of the NAD} increase of solvent pH (proportionally to [OH-]) and is also binding site in the leucine dehydrogenase (unpublished markedly enhanced by the addition of various primary results). Such reagents have been successfully used for amines with different pKavalues. These results supporting identification of the nucleotide-binding sites in various the acid-base mechanism for the leucine dehydrogenase proteins, including rabbit muscle adenylate kinase (40, 41), reaction will be reported shortly. E. coli glycogen synthase (42), and E. coli H+-ATPase (43).

However, for inactivation of leucine dehydrogenase , millimolar concentrations of the affinity labeling reagents REFERENCES were needed to attain the same level of inactivation as with 1. Ohshima,T., Misono,H., & Soda,K. (1978)J. Biol. Chem.253, PLP. Such concentrations are much higher than those 5719-5725 generally employed for affinity labeling (micromolar 2. Zink,M.W. & Sanwal,B.D. (1962)Arch. Biochem. Biophys. 99, order). In addition, the reagents showed no inhibition in the 72-77 assay containing an eight-times-higher concentration of the 3. Soda,K., Misono,H., Mori,K., & Sakato,H. (1971) Biochem. reagent than that of NAD+. These preliminary results Biophys.Res. Commun. 44, 931-935 suggest that the NAD+-binding site of leucine dehydrogen 4. Ohshima,T., Nagata,S., &Soda, K. (1985)Arch. Microbiol. 141, 407-411 ase has little affinity for adenosine polyphosphopyridoxals. 5. Nagata,S., Tanizawa,K., Esaki,N., Sakamoto,Y., Ohshima,T., A recent study using adenosine diphosphopyridoxal for Tanaka,H., & Soda,K. (1988)Biochemistry 27, 9056-9062 affinity labeling of glucose 6-phosphate dehydrogenase 6. Hiragi,Y., Soda,K., & Ohshima,T. (1982)Macromol. Chem. from Leuconostoc mesenteroides showed that the lysine 183,745-751 residues modified by the reagent are probably located at the 7. Laemmli,U.K. (1970)Nature 227, 680-685 substrate-binding site rather than at the NAD+-binding site 8. Bradford,M.M. (1976) Anal. Biochem.72, 248-254 9. Dempsey,W.B. & Christensen,H.N. (1962)J. Biol.Chem. 237, (44). 1113-1120 The binding of NADH by the enzyme was not influenced 10. Ohnishi,M., Yamashita,T., & Hiromi,K. (1977)J. Biochem.81, by modification with PLP, suggesting that the active-site 99-105 Lys80 is not located near the -binding site and that 11. Hummel,J.P. &Dreyer, W.J. (1962)Biochim. Biophys. Acta 63, the modification with the bulky phosphopyridoxyl moiety 530-532 does not affect the local conformation around the cofactor 12. Kunkel, T.A., Roberts,J.D., & Zakour,R.A. (1987)Methods Enzymol.154, 367-382 binding region. NADH, however, considerably protected 13. Sanger,F., Nicklen,S., &Coulson, A.R. (1977) Proc. Natl. Acad. the enzyme from inactivation by PLP. The substrate Sci. U.S.A.74, 5463-5467 L-leucine or the cofactor NAD+ alone offered little protec 14. Shapiro,S., Enser,M., Pugh,E., & Horecker,B.L. (1968)Arch. tive effect, but the enzyme was protected almost complete Biochem.Biophys. 128, 554-562 ly from inactivation by their copresence (i.e., in the react 15. Hountondji,C., Schmitter,J.-M., Fukui, T., Tagaya, M., & ing ternary complex). These results can be interpreted on Blanquet,S. (1990)Biochemistry 29, 11266-11273 the basis of the ordered sequential mechanism of the 16. Chen,S.-S. & Engel,P.C. (1975)Biochem. J. 149, 627-635 17. Chen,S.-S. & Engel,P.C. (1975)Biochem. J. 151, 447-449 leucine dehydrogenase reaction, in which NAD+ is bound 18. Wimmer,M.J., Mo,T., Sawyers,D.L., & Harrison,J.H. (1975) first to the free enzyme followed by L-leucine binding to the J. Biol. Chem.250, 710-715 enzyme having undergone a conformational change. The 19. Blaner,W.S. & Churchich,J. (1979)J. Biol. Chem.254, 1794 e -amino group of Lys80 is probably unreactive with PLP in 1798 this ternary complex or in the binary complex with NADH, 20. Gould,K.G. & Engel,P.C. (1980)Biochem. J. 191, 365-371 both of which would have conformations different from that 21. Ogawa,H. & Fujioka,M. (1980)J. Biol.Chem. 255, 7420-7425 22. Burger,E. & Gorisch,H. (1981)Eur. J. Biochem.118,125-130 of the free enzyme. It is therefore assumed that Lys80 is 23. Feeney, R.E., Blandenhorn,G., & Dixon, H.B. (1975) Adv. located at or near the substrate-binding site and its ProteinChem. 29, 135-203 modification with PLP results in the loss of enzymic 24. Ohshima,T. & Soda,K. (1984)Agric. Biol. Chem. 48, 349-354 activity. 25. George,D.G., Barker, W.C., &Hunt, L.T. (1986)Nucleic Acids On the basis of chemical modification with PLP, Smith et Res.14,11-15 al. (45) proposed that an active-site lysine residue in 26. Moon,K. & Smith,E.L. (1973)J. Biol. Chem.248, 3082-3088 27. Moon,K., Piszkiewicz,D., & Smith, E.L. (1973)J. Biol. Chem. (corresponding to Lys80 of leu 248,3093-3107 cine dehydrogenase) functions in catalysis by forming a 28. Julliard,J.H. & Smith, E.L. (1979)J. Biol. Chem. 254, 3427 Schiff base adduct with its substrate ƒ¿-ketoglutarate. 3438 However, an alternative mechanism for the glutamate 29. McPherson,M.J. &Wootton, J.C. (1983)Nucleic Acids Res. 11, dehydrogenase reaction, in which an active-site lysine 5257-5266 residue with an unusually low pKavalue probably partici 30. Valle,F., Becerril,B., Chen,E., Seeburg,P., Heyneker, H., & Bolivar,F. (1984) Gene27, 193-199 pates in the reaction as a general acid-base catalyst without 31. Blumenthal,K.M., Moon,K., & Smith, E.L. (1975) J. Biol. forming the Schiff base, has also been proposed (46). Chem.250, 3644-3654 To elucidate the catalytic role of Lys80 in leucine 32. Haberland,M.E. & Smith,E.L. (1980)J. Biol. Chem.255 ,7984 dehydrogenase, we have replaced it not only with Ala but 7992 also with some other amino acid residues. All the Lys80 33. Moye,W.S., Amuro,N., Rao, J.K.M., & Zalkin, H. (1985)J. mutant enzymes showed minuscule but detectable activ Biol. Chem.260, 8502-8508

J. Biochem. Active-Site Lysine in Leucine Dehydrogenase 265

34. Okazaki, N., Hibino, Y., Asano, Y., Ohmori, M., Numao, N., & 8257-8261 Kondo, K. (1988) Gene 63, 337-341 41. Yagami, T., Tagaya, M., & Fukui, T. (1988) FEBS Lett. 229, 35. Hibino, Y., Asano, Y., Okazaki, N., & Numao, N. (1988) 261-264 Japanese Patent 63-157986, 553-573 42. Furukawa, K., Tagaya, M., Inouye, M., Preiss, J., & Fukui, T. 36. Takada, H., Yoshimura, T., Ohshima, T., Esaki, N., & Soda, K. (1990) J. Biol. Chem. 265, 2086-2090 (1991) J. Biochem. 109, 371-376 43. Noumi, T., Tagaya, M., Miki-Takeda, K., Maeda, M., Fukui, T., 37. Kuroda, S., Tanizawa, K., Sakamoto, Y., Tanaka, H., & Soda, K. & Futai, M. (1987) J. Biol. Chem. 262, 7686-7692 (1990) Biochemistry 29, 1009-1015 44. LaDine, J.R., Carlow, D., Lee, W.T., Cross, R.L., Flynn, T.G., & 38. Piszkiewicz, D., Landon, M., & Smith, E.L. (1970) J. Biol. Chem. Levy, H.R. (1991) J. Biol. Chem. 266, 5558-5562 245,2622-2626 45. Smith, E.L., Austen, B.M., & Nyc, J.F. (1975) The Enzymes 39. Colman, R.F. (1990) The Enzymes (3rd ed.) 19, 283-321 (3rd ed.) 11, 293-367 40. Tagaya, M., Yagami, T., & Fukui, T. (1987) J. Biol. Chem. 262, 46. Rife, J.E. & Cleland, W.W. (1980) Biochemistry 19, 2328-2333

Vol. 112, No. 2, 1992