A novel metal-activated pyridoxal enzyme with a unique primary structure, low specificity D-threonine aldolase from Arthrobacter sp. Strain DK-38. Molecular cloning and cofactor characterization.

The gene encoding low specificity D-threonine aldolase, catalyzing the interconversion of D-threonine/D-allo-threonine and glycine plus acetaldehyde, was cloned from the chromosomal DNA of Arthrobacter sp. strain DK-38. The gene contains an open reading frame consisting of 1,140 nucleotides corresponding to 379 amino acid residues. The enzyme was overproduced in recombinant Escherichia coli cells and purified to homogeneity by ammonium sulfate fractionation and three-column chromatography steps. The recombinant aldolase was identified as a pyridoxal enzyme with the capacity of binding 1 mol of pyridoxal 5'-phosphate per mol of subunit, and Lys59 of the enzyme was determined to be the cofactor binding site by chemical modification with NaBH4. In addition, Mn2+ ion was demonstrated to be an activator of the enzyme, although the purified enzyme contained no detectable metal ions. Equilibrium dialysis and atomic absorption studies revealed that the recombinant enzyme could bind 1 mol of Mn2+ ion per mol of subunit. Remarkably, the predicted amino acid sequence of the enzyme showed no significant similarity to those of the currently known pyridoxal 5'-phosphate-dependent enzymes, indicating that low specificity D-threonine aldolase is a new pyridoxal enzyme with a unique primary structure. Taken together, low specificity D-threonine aldolase from Arthrobacter sp. strain DK-38, with a unique primary structure, is a novel metal-activated pyridoxal enzyme.

The bioorganic synthesis of ␤-hydroxy-␣-amino acids attracts a great deal of attention because of their potential application as chiral building blocks for the synthesis of biologically active molecules (1)(2)(3)(4)(5). A variety of ␤-hydroxy-␣-amino acids is present in complex natural compounds with interesting biological properties. 3,4,5-Trihydroxy-L-aminopentanoic acid is a key component of polyoxins (1). 4-Hydroxy-L-threonine, for example, is a precursor of rizobitoxine, a potent inhibitor of pyridoxal 5Ј-phosphate (PLP) 1 -dependent enzymes (1). The D-isomers are also biologically significant, because they not only exist in mature mammals (6) but are also constituents of a range of antibiotics, for example, Fusaricidin (7) and Viscosin (8).
Threonine aldolase (TA) (EC 4.1.2.5), which catalyzes the reversible interconversion of certain ␤-hydroxy-␣-amino acids and glycine plus aldehydes, has been shown to be useful for the biosynthesis of ␤-hydroxy-␣-amino acids (1)(2)(3)(4)(5). The enzyme appears to fall into two types, L-type and D-type, on the basis of its stereospecificity of the cleavage reaction toward the ␣-carbon of threonine. L-Type TA, acting on L-and/or L-allo-threonine, is further divided into three groups based on its stereospecificity toward the ␤-carbon of threonine as follows: (i) L-allo-TA is specific to L-allo-threonine; (ii) L-TA acts only on L-threonine; and (iii) low specificity L-TA can use both L-threonine and L-allo-threonine as substrates. All of the three L-type enzymes have been found to exist in nature. L-TA was partially purified from Clostridium pasteurianum (9); L-allo-TA was from Aeromonas jandaei (10), and three low specificity L-TAs from Candida humicola (11,12), Saccharomyces cerevisiae (13), and Pseudomonas sp. strain (14) were purified and extensively characterized. The genes encoding for L-allo-TA of A. jandaei and low specificity L-TAs of S. cerevisiae and Pseudomonas sp. strain have been cloned and sequenced (13)(14)(15). Likewise, Dtype TA, acting on D-threonine and/or D-allo-threonine, might include D-allo-TA, D-TA, and low specificity D-TA. However, only low specificity D-TA activity was found from 3 out of 2,000 strains examined (16). Low specificity D-TA was previously purified from Arthrobacter sp. strain DK-38, and the enzyme was shown to have a monomeric structure and to require unusually both PLP and divalent cations for its catalytic activity (16). Due to little available purified enzyme, the identification of the cofactors was not performed.
The present paper describes cloning and sequencing of the dtaAS gene encoding the low specificity D-TA from Arthrobacter sp. strain DK-38, the expression of the gene in Escherichia coli cells, and further characterization of the recombinant enzyme. Our data showed that the low specificity D-TA is a novel metalactivated pyridoxal enzyme. Evidence is also presented that low specificity D-TA has a unique primary structure, probably representing a new family of pyridoxal enzyme. * This work was supported in part by grants-in-aid for scientific research from the Ministry of Education, Science, Sports and Culture of Japan and for research for the future from the Japan Society for the Promotion of Science. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM

General Recombinant DNA Technique
Plasmid DNA was purified with a plasmid purification kit from Qiagen, Inc. (Chatsworth, CA). Restriction enzymes and DNA-modifying enzymes were purchased from Takara Shuzo and Toyobo and used according to the manufacturers' protocols. Transformation of E. coli with plasmid DNA by electroporation was performed under standard conditions with a BTX ECM 600 electroporation system (Biotechnologies and Experimental Research, Inc., San Diego, CA). Other general procedures were performed as described by Sambrook et al. (18).

Cloning of the dtaAS Gene
Two oligonucleotide primers were purchased from Amersham Pharmacia Biotech (Tokyo, Japan), each with additional restriction sites (underlined in the following sequences) added to the 5Ј end to facilitate cloning of the amplified product: primer I, 5Ј-CCGAAGCTTATGTC-NCARGARGTNAT-3Ј; and primer II, 5Ј-GCCGAATTCGGGSGTGTCN-ACNCGSGC-3Ј. Degenerate positions are indicated by "S" for C or G, "R" for A or G, and "N" for all bases. Primers I and II were based upon the NH 2 -terminal amino acid sequence of the wild-type low specificity D-TA from Arthrobacter sp. strain DK-38. PCR amplification was performed in a 50-l reaction mixture containing 5 l of Me 2 SO, 10 mM Tris-HCl (pH 8.85), 25 mM KCl, 5 mM (NH 4 ) 2 SO 4 , 2 mM MgSO 4 , 0.1 mM each deoxynucleotide triphosphate, 20 pmol of each primer, 1 g of the genomic DNA, and 0.5 units of PWO DNA polymerase (Boehringer Mannheim, Germany) at 94°C for 1 min, 43°C for 0.5 min, and 72°C for 1 min for a total of 35 cycles. The amplified product was digested with EcoRI and HindIII, separated by agarose gel electrophoresis, and then purified with a GeneClean kit (Bio 101, Vista, CA). The amplified DNA of 90 bp was then cloned into pUC118.
Chromosomal DNA isolated from Arthrobacter sp. strain DK-38 cells was partially digested with Sau3AI and fractionated on a sucrose density gradient (10 -40%) in a Beckman L-70 ultracentrifuge (Beckman Instruments, Inc., Palo Alto, CA) at 100,000 ϫ g for 18 h. Fragments in the molecular size of 1-10 kb were collected and ligated into BamHI-restricted pUC118, and the plasmids were introduced into E. coli XL1-Blue MRFЈ cells to construct a genomic library of Arthrobacter sp. strain DK-38. The genomic library was screened by colony hybridization with the [␥-32 P] dATP-labeled 90-bp DNA fragment as a probe. The clone, pUDTA, carrying an approximately 7.2-kb DNA fragment was selected for further analysis.

Nucleotide Sequence Analysis
pUDTA was used as a sequencing template. The nucleotide sequence was determined by the dideoxy chain termination method with Cy5 AutoRead and Cy5 AutoCycle sequencing kits and an Amersham Pharmacia Biotech ALFred DNA sequencer. A homology search was performed by the sequence similarity searching programs Fasta (19) and Blast (20). The ClustalW method was used to align the sequences (21).

Overexpression of the dtaAS Gene in E. coli Cells
To obtain the entire gene without excessive flanking parts, PCR amplification was carried out in a 50-l reaction mixture containing 5 l of Me 2 SO, 10 mM Tris-HCl (pH 8.85), 25 mM KCl, 5 mM (NH 4 ) 2 SO 4 , 2 mM MgSO 4 , 0.1 mM deoxynucleotide triphosphate, 20 pmol of each primer, 1 g of the genomic DNA, and 0.5 units of PWO DNA polymerase (Boehringer Mannheimn) at 94°C for 1 min, 60°C for 2 min, and 72°C for 3 min for a total of 30 cycles. The 5Ј primer containing a Shine-Dalgarno sequence (lowercase letters) and an ATG initiation codon (bold letters), and the 3Ј primer with the complement of the TGA termination codon (bold letters) had the respective sequences 5Ј-GC-CGAATTCggagCGTCCCGATGTCCCAGG-3Ј and 5Ј-CCGGAATTCT-GAAGACGTCAGCGCGAG-3Ј, which were designed on the basis of the nucleotide sequence of the dtaAS gene; to facilitate the cloning, an additional restriction site (underlined sequence) was incorporated into both primers. The amplified PCR product was digested with EcoRI, separated by agarose gel electrophoresis, and then purified with a GeneClean kit. The amplified DNA of approximately 1.1 kb, which contained the complete coding sequences of the dtaAS gene, was inserted downstream of the tac promoter in pKK223-3 and then used to transform E. coli XL1-Blue MRFЈ cells. The constructed plasmid was designated pKDTA.

Purification of the Recombinant Low Specificity D-TA
All enzyme purification operations were carried out at 0 -4°C. Unless otherwise noted, 50 mM Tris-HCl (pH 7.4) containing 10 M PLP was used as the buffer throughout the purification procedures.
Step 1, Preparation of Cell Extract-Cells of the E. coli transformant harboring plasmid pKDTA were grown aerobically at 37°C for 16 h in 12 liters of LB medium containing 0.1 mg/ml ampicillin and 0.2 mM IPTG. The cells were harvested and rinsed with buffer. After being suspended in 200 ml of buffer, the cells were disrupted by ultrasonic oscillation at 4°C for 20 min with a model 201M ultrasonic oscillator (Kubota, Tokyo, Japan). The cell debris was removed by centrifugation at 25,000 ϫ g for 30 min.
Step 2, Ammonium Sulfate Fractionation-The supernatant solution was brought to 20% saturation with ammonium sulfate and centrifuged at 25,000 ϫ g for 30 min. Ammonium sulfate was added to the supernatant solution to 50% saturation. The precipitate collected by centrifugation was dissolved in buffer, and the enzyme solution was dialyzed against 1,000 volumes of buffer.
Step 3, Butyl-Toyopearl Column Chromatography-The enzyme solution, brought to 20% saturation with ammonium sulfate, was applied to a Butyl-Toyopearl 650M column (5.0 ϫ 40 cm). Elution was carried out with a 2,400-ml linear gradient of 20 to 0% saturated ammonium sulfate in buffer at a flow rate of 5 ml/min. The fractions with threonine aldolase activity were pooled and concentrated by ultrafiltration with a Centriprep-30 apparatus (Amicon, Inc., Beverly, MA).
Step 4, DEAE-Toyopearl Column Chromatography-The enzyme solution was dialyzed against 1,000 volumes of buffer and applied to a DEAE-Toyopearl 650 M column (2.5 ϫ 20 cm) equilibrated with buffer. After the column was thoroughly washed with the buffer containing 50 mM NaCl, linear gradient elution was performed with a buffer supplemented with NaCl by increasing the concentration from 50 to 200 mM. The flow rate was maintained at 5 ml/min. The fractions with D-TA activity were pooled and concentrated by ultrafiltration with a Centriprep-30 apparatus.
Step 5, Gigapite Column Chromatography-The enzyme solution was dialyzed against 1,000 volumes of 10 mM potassium phosphate buffer (pH 7.4) containing 10 M PLP and applied to a Gigapite column (5.0 by 40 cm) equilibrated with the 10 mM potassium phosphate buffer (pH 7.4). After the column was thoroughly washed with the same buffer, elution was carried out with a 2,400-ml linear gradient of 10 -300 mM potassium phosphate buffer (pH 7.0) at a flow rate of 5 ml/min. The fractions with D-TA activity were pooled and concentrated by ultrafiltration with a Centriprep-30 apparatus. The purified enzyme was concentrated to a protein concentration of about 25 mg/ml and stored at 0 to 4°C for at least 1 month without loss of enzyme activity.

Enzyme Assay
Threonine aldolase activity was measured spectrophotometrically at 340 nm by coupling the reduction of acetaldehyde (oxidation of NADH) with yeast alcohol dehydrogenase (Wako, Osaka, Japan). The assay mixture comprised 100 mol of Hepes-NaOH buffer (pH 8.0), 50 mol of D-threonine, 0.05 mol of PLP, 0.1 mol of MnCl 2 , 0.2 mol of NADH, 30 units of yeast alcohol dehydrogenase, and appropriate amounts of the enzyme in a final volume of 1 ml. One unit of the enzyme is the amount of enzyme that catalyzed the formation of 1 mol of acetaldehyde (1 mol of NADH oxidized) per min at 30°C; the molar extinction coefficient of NADH is 6.2 ϫ 10 3 M Ϫ1 cm Ϫ1 . For a qualitative analysis, threonine aldolase activity was also assayed as follows: the reaction mixture comprised 10 mol of D-threonine, 0.01 mol of PLP, 0.02 mol of MnCl 2 , 20 mol of Hepes-NaOH buffer (pH 8.0), and the enzyme in a total volume of 200 l. The reaction was carried out at 30°C for 10 min and was terminated by the addition of 50 l of 30% trichloroacetic acid. The released acetaldehyde was measured spectrophotometrically according to the method of Paz et al. (22). The aldolase activities toward phenylserine, ␤-3,4-dihydroxyphenylserine, and ␤-3,4-methylenedioxyphenylserine were measured spectrophotometrically at 279, 350, and 320 nm, respectively, with molar extinction coefficients of 1.4 ϫ 10 3 M Ϫ1 cm Ϫ1 for benzaldehyde, 8.9 ϫ 10 3 M Ϫ1 cm Ϫ1 for protocatechualdehyde, and 17.6 ϫ 10 3 M Ϫ1 cm Ϫ1 for piperonal.

Determination of Protein
The concentration of the enzyme was determined spectrophotometrically by using a molar extinction coefficient ⑀ M ϭ 29,927 M Ϫ1 cm Ϫ1 (A 1 cm 1 mg/ml ϭ 1.34) at 278 nm and pH 7.4 for the holoenzyme and ⑀ M ϭ 27,580 M Ϫ1 cm Ϫ1 (A 1 cm 1 mg/ml ϭ 1.45) at 278 nm for the apoenzyme, where the ⑀ M values were determined by the method of Edelhoch and coworkers (23,24), and the contributions of tryptophan, tyrosine, and cystine to the ⑀ M values in 6 M guanidine hydrochloride were calculated on the basis of 3 tryptophan, 7 tyrosine, and 8 cystine residues and 5565, 1395, and 140 M Ϫ1 cm Ϫ1 for the three residues, respectively.

Spectrophotometric Measurements
The absorption spectrum of the enzyme was measured at 20°C with 20 mM potassium phosphate buffer (pH 7.4) by a Hitachi model U-3210 spectrophotometer (Hitachi, Tokyo, Japan).

PLP Content
The PLP content of the enzyme was determined according to the method of Wada and Snell (25).

Isolation of Pyridoxyl Peptide
The holoenzyme (10 mg in 200 l of 20 mM potassium phosphate buffer (pH 7.4)) was reduced at 0°C by the addition of about 0.2 mg of solid NaBH 4 . To protect the label against photo-destruction, all the tubes used in the following experiment were covered with aluminum foil. After standing at 0°C for about 30 min, the enzyme protein was precipitated with 2% trichloroacetic acid. The precipitates were dissolved in 100 l of 50 mM Tris-HCl buffer (pH 8.5) containing 6 M guanidine HCl and 0.5 mM dithiothreitol and treated at 37°C for 1 h with 2 mM iodoacetate. The reduced and carboxymethylated protein was extensively dialyzed against 5 mM HCl. The precipitates formed during dialysis were collected by centrifugation, washed twice with water, and suspended in 500 l of 50 mM Tris-HCl buffer (pH 7.4) containing 2.5 nmol of L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Worthington) and subsequently digested at 37°C for 12 h. The resultant peptide fragments were applied to a C 18 column (4.6 ϫ 150 mm; Tosoh, Tokyo, Japan) and eluted with 0.05% trifluoroacetic acid for 10 min, followed by a linear gradient of 0 -80% acetonitrile in 0.05% trifluoroacetic acid over 60 min at a flow rate of 1.0 ml/min. The elution was monitored dually by absorbance at 215 and 330 nm. The peptide, which had strong absorbance at 330 nm, was further purified by high pressure liquid chromatography under chromatographic conditions similar to those previously described, except a directly linear gradient of 5-50% acetonitrile was employed. The fractions which showed absorbance at 330 nm were pooled and subjected to sequence analysis.

Generation of Metal-free Enzyme
Two hundred micromoles of recombinant low specificity D-TA was dialyzed against 50 mM Tris-HCl buffer (pH 7.4) containing 10 M PLP and 10 mM EDTA for 12 h at 4°C. The enzyme solution was further dialyzed twice against the same buffer without EDTA for 12 h at 4°C to remove EDTA.

Equilibrium Dialysis Study
Ten nanomoles of the metal-free enzyme was dialyzed twice against 1 liter of 50 mM Tris-HCl buffer (pH 7.4) containing 10 M PLP and various concentrations of MnCl 2 (0 -50 M) for 12 h at 4°C. The metal concentrations, both inside and outside the dialysis bag, were then determined as described below.

Atomic Absorption Measurement
The metal ion concentration was determined with a mode FLA-1000 atomic absorption spectrometer (Nippon Jarrell-Ash, Uji, Japan). Sequence Analysis of the dtaAS Gene-The plasmid, pUDTA, extracted from the positive clone containing an approximately 7.2-kb insert, was directly used as the template for sequencing the dtaAS gene by the gene-walking method; the initial sequencing primer was designed based on the nucleotide sequence of the 90-bp PCR product.

Cloning of the dtaAS
Sequence analysis of the double strand DNA revealed that the ORF consists of 1,140 bp starting with an initiation codon, ATG, and ending with a termination codon, TGA (Fig. 1). A probable ribosome-binding sequence, GGAG, is present eight bases upstream of the putative translational start codon (26). The 379-residue enzyme as deduced from the DNA sequence has a molecular mass of 40,035 Da and composition as follows: Ala 55 -Cys 8 -Asp 25 -Glu 20 -Phe 10 -Gly 34 -His 10 -Ile 17 -Lys 7 -Leu 36 -Met 5 -Asn 9 -Pro 22 -Gln 20 -Arg 22 -Ser 20 -Thr 12 -Val 37 -Trp 3 -Tyr 7 . The NH 2 -terminal amino acid sequence coincided with that of the purified enzyme determined by Edman degradation (Fig. 1).
Sequence Homology with Other Proteins-The predicted amino acid sequence showed no significant similarity to those of the currently known pyridoxal enzymes. However, in a search of protein amino acid and nucleotide sequence data bases (GenBank, EMBL, DDBJ, and Protein Data Bank) by means of the sequence similarity searching programs Blast (19) and Fasta (20), D-serine deaminase (GenBank, U41162) of Burkholderia cepacia and a hypothetical protein (GenBank, U73935) of Shewanella sp. strain SCRC-2738 were found to be significantly similar in primary structure to low specificity D-TA. It should be mentioned that the property of the D-serine deaminase from B. cepacia was not reported, nor had this protein any similarity in primary structure with those of the extensively studied D-serine deaminase from E. coli (27) and a probable D-serine deaminase of Bacillus subtilis (GenBank TM , D84432), although the E. coli and B. subtilis proteins had as much as 55% identity and 69% similarity to each other. The hypothetical protein of Shewanella is encoded by an unnoted ORF, approximately 500 bp downstream of the eicosapentaenoic acid synthesis gene cluster of Shewanella sp. strain SCRC-2738 (28). The amino acid sequence alignment of D-serine deaminase and the hypothetical protein with that of low specificity D-TA is depicted in Fig. 2. D-Serine deaminase of B. cepacia and the hypothetical protein of Shewanella had 24 and 22% identities and 41 and 39% similarities to those of low specificity D-TA, respectively. Remarkably, Lys 59 of low specificity D-TA was completely conserved in the three proteins (Fig. 2).
Overexpression of the dtaAS Gene in E. coli Cells-The whole dtaAS gene amplified by PCR directly from Arthrobacter chromosomal DNA, with a putative Shine-Dalgarno sequence (GGAG) and an initiation codon (ATG), was inserted into the EcoRI site of pKK223-3. The resultant plasmid pKDTA was introduced into E. coli XL1-Blue MRFЈ cells. The nucleotide sequence of the whole amplified gene was further confirmed to have undergone no error matching during the PCR by sequencing of the double strands. The recombinant cells produced a large amount of low specificity D-TA, and the specific activity of the crude extract of E. coli XL1-Blue harboring pKDTA was elevated to 1.8 units/mg, which is about 180-fold over that of  Table I). The protein was only produced in the presence of IPTG (data not shown), indicating that the tac promoter is essential for the overexpression.
Enzyme Purification-The recombinant low specificity D-TA from Arthrobacter sp. strain DK-38 was purified by ammonium sulfate fractionation, Butyl-Toyopearl, DEAE-Toyopearl, and Gigapite chromatography steps (Table I). About 100 mg of purified enzyme was obtained from 51 g wet cells. The purified enzyme showed a single protein band on SDS-polyacrylamide gel electrophoresis with a molecular mass of about 40 kDa (Fig.  3), the same as the value calculated from the deduced amino acid sequence of the enzyme.
PLP Requirement-The recombinant enzyme exhibited absorption maxima at 278 and 417 nm, with an A 278 /A 417 ratio of 4.8 (curve 1, Fig. 4). The solution of the pure enzyme was distinctly yellow. As has been demonstrated with other PLPcontaining enzymes, the absorption peak around 417 nm is characteristic of an azomethine linkage between the coenzyme and a protein amino group. Reduction of the enzyme with sodium borohydride by the method of Matsuo and Greenberg (29) resulted in a loss of the enzyme activity, with a disappearance of the absorption maximum at 417 nm and a concomitant increase in the A 330 (data not shown). The reduced enzyme was catalytically inert, and the addition of PLP did not restore the enzyme activity. This result suggests that sodium borohydride reduces the aldimine linkage of the internal Schiff base. The holoenzyme was converted to the apoenzyme (curve 3, Fig. 4) by treatment with 1 mM hydroxylamine at 4°C for 30 min and then dialyzed against 20 mM potassium phosphate buffer (pH 7.4). The constructed apoenzyme did not show D-TA activity. A putative Shine-Dalgarno sequence is indicated as SD with a double line. The thinly underlined amino acid sequence is identical to that determined for the purified enzyme by Edman degradation, and the boldly underlined amino acid sequence is identical to that determined for the NaBH 4 modified pryidoxyl peptide, except for Lys 59 (see text for details).
However, the activity was restored to 78% that of the native enzyme with a corresponding recovery of the A 417 (curve 2, Fig.  4) on dialysis against 20 mM potassium phosphate buffer (pH 7.4) supplemented with 10 M PLP. In contrast, the two analogs of PLP, pyridoxal and pyridoxamine 5Ј-phosphate, neither restored the enzyme activity nor the A 417 (data not shown). Resolution of low specificity D-TA was also carried out by treatment of the enzyme with cysteine. As shown in Fig. 4, inset, cysteine caused the disappearance of the 417-nm peak with a concomitant appearance of a peak at 330 nm (curves 2-5, Fig. 4, inset). The new absorbance peak at 330 nm disappeared on subsequent dialysis. This result indicates that cysteine resolved the enzyme by combining with the enzyme-bound PLP (417-nm peak) to form the more stable thiazolidine compound (330-nm peak) (30). All of these results show that PLP forms a Schiff base with a lysine residue of the low specificity D-TA from Arthrobacter sp. strain DK-38.
To determine the bound PLP content of low specificity D-TA, 5 mg of the enzyme was extensively dialyzed against 20 mM potassium phosphate buffer (pH 7.4) containing 10 M PLP; the PLP concentrations inside and outside the dialysis bag were subsequently determined according to the method of Wada and Snell (25), and the difference between the inside and outside was taken as the PLP content of the enzyme. The PLP content of the enzyme was determined in triplicate to be 0.85, 0.88, and 0.94 mol/mol of 40-kDa subunits, respectively, suggesting that the enzyme has the capacity to bind 1 mol of PLP as a cofactor per mol of 40-kDa subunits.
Identification of the PLP-binding Site-The enzyme was treated with NaBH 4 as described under "Experimental Procedures," and the isolation of the modified pyridoxyl peptide is depicted in Fig. 5. The amino acid sequence of the isolated peptide, which showed the absorbance peak at 330 nm, was determined by the Edman degradation procedure with a model PPSQ-10 protein sequencer. The 13 steps of degradation, except the 10th step, gave an identical amino acid sequence 50 HDVALRPHAKAHK 62 of the amino acid sequence deduced from the dtaAS gene (Fig. 1). The 10th step, which did not show an identifiable peak on the sequencer, was presumably the  cofactor-binding lysine residue. Remarkably, this lysine residue is completely conserved in the three proteins aligned as previously mentioned (Fig. 2).
Metal Requirement for the Aldolase Activity-To qualitatively analyze the bound metal ions of the purified recombinant low specificity D-TA from Arthrobacter sp. strain DK-38, gel filtration (HiLoad Superdex 200, Amersham Pharmacia Biotech, Uppsala, Sweden) was employed to remove free metal ions attached to the enzyme. The bound metal ions of the enzyme were subsequently analyzed with a mode ICPS-1000III sequential plasma spectrometer (Shimadzu, Kyoto, Japan); the enzyme so treated was determined to contain no detectable divalent cations, such as Mn 2ϩ , Mg 2ϩ , Co 2ϩ , Ni 2ϩ , Fe 2ϩ , and Ca 2ϩ , which were previously shown to be activators of the wild-type low specificity D-TA from Arthrobacter sp. strain DK-38 (16). Kinetic analysis further showed that the gel-filtrated enzyme had the same V max and K m values toward DLthreo-and DL-erythro-phenylserine as those of the metal-free ones treated with EDTA, supporting the finding that the purified enzyme contained no bound metal ions, at least no activating divalent cations. Our previous work already demonstrated that the Mn 2ϩ ion stimulated the enzyme to give the highest aldolase activity among all cations examined (16). Mn 2ϩ ion was thus selected as a target for the present study. Kinetic constants of the metal-free enzyme toward various compounds were determined in the presence or absence of Mn 2ϩ ion, and the results are comparatively given in Table II. The K m values of the enzyme toward all substrates examined were independent of the presence or absence of Mn 2ϩ ion, suggesting that the metal ion may not be involved in the substrate binding. In contrast, the V max values of the enzyme were significantly increased in the presence of Mn 2ϩ toward all examined compounds.
Furthermore, experiments were carried out to correlate the enzyme activity and amount of bound Mn 2ϩ ions of the recombinant low specificity D-TA. The metal-free enzyme was dialyzed twice against 1,000 volumes of 50 mM Tris-HCl buffer (pH 7.4) containing 10 M PLP and 100 M MnCl 2 for 12 h. Following gel filtration to remove free Mn 2ϩ ions, the amount of bound Mn 2ϩ ions of the enzyme was determined by atomic absorption spectrometry to be less than 0.01 mol/mol of subunit. As a consequence of the low affinity of the aldolase for the Mn 2ϩ ion, the stoichiometry of Mn 2ϩ ion binding was thus performed by equilibrium dialysis study. As shown in Fig. 6, the maximum number of bound Mn 2ϩ ions was determined to be 0.92 mol/mol of subunit, suggesting that the recombinant enzyme could bind 1 mol of Mn 2ϩ ion per mol of subunit. The maximal enzyme activity was restored on the binding of approximately 1 mol eq of Mn 2ϩ (Fig. 6). DISCUSSION We have previously reported the occurrence, isolation, and catalytic properties of a novel enzyme, D-threonine aldolase, that catalyzes the interconversion of D-threonine/D-allo-threonine and glycine plus acetaldehyde from Arthrobacter sp. strain DK-38 (16). To study the structural and functional relationships of the enzyme, we here cloned the dtaAS gene encoding the enzyme from the genomic DNA of Arthrobacter sp. strain DK-38 and expressed the gene in E. coli cells. Our data showed that the recombinant low specificity D-TA from Arthrobacter sp. strain DK-38, with a unique primary structure, is a novel metal-activated pyridoxal enzyme.
The recombinant low specificity D-TA from Arthrobacter sp. strain DK-38 was concluded to be a metal-activated enzyme, because of the following: 1) the purified enzyme contained no detectable metal ions; 2) the enzyme activity was stimulated by exogenous metal ions (Fig. 6, Table II, and Ref. 16); and 3) stoichiometric analysis revealed that the enzyme could bind 1 mol of Mn 2ϩ /mol of subunit and the saturation of the metalbinding showed the maximal aldolase activity (Fig. 6). We now have insufficient data to illustrate exactly the binding mode of the divalent cation to the enzyme. However, according to the three general coordination schemes for the binding of enzyme, metal, and substrate, Elution of pyridoxyl peptide (downward) by absorbance at 330 nm and that of other peptides (upward) was by absorbance at 215 nm. The collected fraction showing an absorbance peak at 330 nm was further purified by repeated high pressure liquid chromatography and subjected to a protein sequencer for the amino acid sequence analysis. strate (Fig. 6), which excluded the model of substrate bridge complex, and 2) the fact that the K m value of the enzyme was almost the same in the presence or absence of the metal ion (Table II) highlighted that metal bridge complex was not the case.
The results presented in this paper confirmed that low specificity D-TA is a PLP-dependent enzyme, and Lys 59 was identified to be the PLP-binding site of the enzyme by chemical modification with NaBH 4 (Fig. 5). To our knowledge, this is the first report dealing with the identification of a divalent cationactivated pyridoxal enzyme, although some monovalent cations have been shown to be the activators of other PLP-dependent enzymes, such as 2,2-dialkylglycine decarboxylase, tryptophanase, and tyrosine phenol-lyase (32). The roles of Na ϩ and K ϩ in PLP-dependent enzyme catalysis were recently reviewed by Woehl and Dunn (32). On the basis of the x-ray structural information of the metal sites for three PLP-dependent enzymes and advances stemming from solution spectroscopic and mechanistic studies, Woehl and Dunn proposed two types of mechanisms to explain the effects of monovalent metal ions on catalytic activity as follows: (i) mechanisms involving the metal ion in a static structural role wherein binding activates the enzyme by simply stabilizing the catalytically active conformation of the protein and (ii) mechanisms where the metal ion plays a dynamic role in which binding selectively assists one or more of the protein conformational transitions essential for complementarity between enzyme site and the structure of an activated complex (32). It would be interesting to reveal whether divalent cations play the similar role in the low specificity D-TA catalysis.
PLP-dependent enzymes catalyze manifold reactions in the metabolism of amino acids. On the basis of a computer analysis, Alexander et al. (33) classified most of the known PLP-dependent enzymes into ␣, ␤, and ␥ families correlating in most cases with their regio-specificity. The ␣ enzymes, with a few exceptions, catalyze the transformation of amino acids in which the covalency changes are limited to the ␣-carbon atom that carries the amino group forming the aldimine linkage to the coenzyme, such as serine hydroxymethyltransferase, 5-aminolevulinate synthase, and 8-amino-7-oxononanoate synthase. The ␤ and ␥ enzymes catalyze ␤-replacement or ␤-elimination and ␥-replacement or ␥-elimination reactions, respectively. We have recently cloned and determined the primary structures of several L-type threonine aldolases, L-allo-TA from A. jandaei DK-39 (15) and three low specificity L-TAs from S. cerevisiae S288C (13), Pseudomonas sp. strain NCIMB 10558 (14), and E. coli GS245. 2 These four L-type TAs with a significant amino acid sequence identity to one another showed no structural similarity to other PLP-dependent enzymes (14), although they belong reaction-specifically to the ␣ family. In a search of protein sequence data bases (GenBank TM , EMBL, DDBJ, and PDB) using either the total or partial sequence containing Lys 59 as a central amino acid as a probe, low specificity D-TA from Arthrobacter sp. strain DK-38 showed neither similarity in primary structure to the members belonging to the ␣, ␤, and ␥ families nor to the four L-type TAs, suggesting that the low specificity D-TA probably represents another new family of pyridoxal enzyme.
In summary, our findings reported here showed that low specificity D-threonine aldolase from Arthrobacter sp. strain DK-38, with a unique primary structure, is a novel metalactivated pyridoxal enzyme, and Lys 59 of the enzyme was determined to be the pyridoxal binding site. To study further the The unresolved DL-threo-␤-phenylserine, DL-erythro-␤-phenylserine, DL-threo-␤-(3,4-methylenedioxyphenyl)serine, DL-erythro-␤-(3,4-methylenedioxyphenyl)serine, and DL-threo-␤-(3,4-dihydroxyphenyl)serine were used as substrates, and the D-form stereoisomer was shown to be specifically cleaved as follows: the reaction mixture comprising 20 mol of the DL-form amino acid, 100 mol of Hepes-NaOH buffer (pH 8.0), 0.05 mol of PLP, 0.1 mol of MnCl 2 , and 1.2 units of the enzyme, in a final volume of 1 ml, was kept at 30°C for 30 min. High performance liquid chromatography analysis (see "Experimental Procedures") confirmed that the D-form stereoisomers disappeared, whereas the L-form stereoisomers remained unchanged. Low specificity L-TA from Pseudomonas sp. NCIMB 10558 (14) was used as control to show that the L-form stereoisomers were stereospecifically cleaved, whereas D-form stereoisomers remained unchanged.
FIG. 6. Correlation of the bound Mn 2؉ ion content and catalytic activity of the recombinant low specificity D-TA. The enzyme was dialyzed twice against 50 mM Tris-HCl buffer (pH 7.4) containing 0 -50 M MnCl 2 . The bound Mn 2ϩ ion content, expressed in moles of Mn 2ϩ ion per mol of subunit (E), was calculated from the difference of the inside and outside of the dialysis bag. Mn 2ϩ ion concentrations were measured with an atomic absorption spectrometer. The D-threonine aldolase activity (‚) was determined following the standard assay method, except the Mn 2ϩ ion concentration was adjusted to the amount outside the dialysis bag. Correlation of enzyme activity and the bound Mn 2ϩ ion content was replotted (q). role of Lys 59 and the activation mechanism of the divalent cation, site-directed mutagenesis experiments and spectroscopic studies are on the way.