Self-protection Mechanism in D -Cycloserine-producing Streptomyces lavendulae GENE CLONING, CHARACTERIZATION, AND KINETICS OF ITS ALANINE RACEMASE AND D -ALANYL- D -ALANINE LIGASE, WHICH ARE TARGET ENZYMES OF D -CYCLOSERINE*

An antibiotic, D -cycloserine (DCS), inhibits the cata- lytic activities of alanine racemase (ALR) and D -alanyl-D -alanine ligase (DDL), which are necessary for the bio- synthesis of the bacterial cell wall. In this study, we cloned both genes encoding ALR and DDL, designated alrS and ddlS , respectively, from DCS-producing Streptomyces lavendulae ATCC25233. Each gene product was purified to homogeneity and characterized. Escherichia coli , transformed with a pET vector carrying alrS or ddlS , displays higher resistance to DCS than the same host carrying the E. coli ALR- or DDL-encoded gene inserted into the pET vector. Although the S. lavendulae DDL was competitively inhibited by DCS, the K i value (920 (cid:1) M ) was obviously higher (40 (cid:1) 100-fold) than those for E. coli DdlA (9 (cid:1) M ) or DdlB (27 (cid:1) M ). The high K i value of the S. lavendulae DDL suggests that the enzyme may be a self-resistance determinant in the DCS-producing lavendulae— The chromosomal DNA from S. lavendulae ATCC25233, which digested with BamHI, was fractionated on 0.8% agarose gel electrophoresis and transferred to a Hybond-N (cid:3) membrane using the standard protocol (21). Southern hybridization analysis was done using a putative ALR gene (1176-bp) from S. coelicolor (22) as a probe. To obtain the probe DNA, PCR amplification was done using the S. coelicolor genomic DNA as a template together with a sense primer (5 (cid:2) -ATGAGCGAGACAACT-GCTCGGCGGGACGCG-3 (cid:2) and an antisense primer (5 (cid:2) -TCATTCGTT- GACGTAGACGCGCGGGACCCGG-3 (cid:2) PCR was done under the following conditions: cycles min finally, 3-min extension Probe labeling, hybridization, and detection were performed using an AlkPhos direct labeling and detection kit

Since the discovery of streptomycin, tuberculosis, a disease caused by infection of Mycobacterium tuberculosis, has decreased annually; however, currently, it is once again on the rise. The increase in morbidity is likely because of the decline in immunity caused by changes in the environment and diet (1). In addition, the advent of multidrug-resistant M. tuberculosis is also a cause of the return of tuberculosis (2). D-Cycloserine (D-4-amino-3-isoxazolidone (DCS) 1 ), which is a cyclic structural analogue of D-alanine (D-Ala) and is produced by Streptomyces garyphalus and Streptomyces lavendulae, is a clinical medicine for the treatment of tuberculosis. The antibiotic is an effective anti-mycobacterial agent, but it is rarely prescribed and is used only in combined therapies because of its serious side effects (3). The side effects are caused by the binding of DCS to N-methyl-D-aspartate receptors as an agonist. However, application of these adverse effects to treatments for neural diseases (4) such as Alzheimer's (5) and Parkinsonism (6) have been dedicatedly researched. The peptidoglycan layer, which is contained in a bacterial cell wall, is the main component that enables bacteria to be resistant to osmotic pressure. The formation of UDP-N-acetyl muramyl pentapeptide, which is a precursor of peptidoglycan, is followed by a cross-link reaction of the precursors. In the cross-linking process, D-Ala plays an important role as a bridge molecule (7). Because D-amino acids, including D-Ala, are not primarily found in natural resources, bacteria generate D-Ala from L-Ala by the catalysis of Ala racemase (ALR). This enzyme needs a pyridoxal 5Ј-phosphate (PLP) as a cofactor and catalyzes the racemization of both Ala enantiomers. Escherichia coli and Salmonella typhimurium possess two kinds of closely related ALR-encoded genes (alr and dadX in E. coli, and dal and dadB in S. typhimurium) (8 -10). For example, the racemase encoded by dal of S. typhimurium is necessary for peptidoglycan synthesis and displays a 40% identity to a catabolic racemase encoded by dadB (11). D-Ala, generated by ALR, is a substrate to form D-alanyl-Dalanine (D-Ala-D-Ala) (12). The dipeptide is formed by the action of an ATP-dependent enzyme, D-Ala-D-Ala ligase (DDL) and is incorporated into the peptidoglycan precursor by the catalytic activity of the D-Ala-D-Ala-adding enzyme (7). E. coli produces two kinds of DDL, designated DdlA and DdlB, which are encoded by ddlA and ddlB, respectively. S. typhimurium expresses DDL, which has high similarity to the E. coli DdlA (13).
DCS interferes with the activities of both ALR and DDL, which are necessary for the synthesis of peptidoglycan contained in the cell wall of bacteria. Because these enzymes are unique to bacteria, they may become potential targets for the screening of selective anti-bacterial agents (14). ALR and DDL have been considered competitively inhibited because DCS is structurally similar to D-Ala (13,15). However, it was recently reported that DCS inhibits the catalytic activity of ALR in a time-dependent inactivation manner (16). In addition, the an-tibiotic and its enantiomer, L-cycloserine (LCS), inhibit several kinds of PLP-dependent enzymes in the same manner (16 -18).
Antibiotic-producing microorganisms must be protected from the lethal effect of their own products. We recently cloned a 3.5-kb DNA fragment carrying a gene that confers resistance to DCS from DCS-producing S. garyphalus by a "shotgun" cloning technique (19). The hydropathy plot analysis of a protein deduced from the nucleotide sequence of the gene encoding DCS resistance revealed that the protein may carry membraneintegral domains spanning the membrane 10 times, suggesting that the DCS resistance gene product may be a factor associated with DCS transport. Interestingly, an incomplete gene was found to be located upstream of the transmembrane protein gene from S. garyphalus. The incomplete gene consists of 246 bp, and the putative protein has a 52.6% identity with a D-Ala-D-Ala ligase from Pseudomonas aeruginosa (20). On the other hand, although the cloned fragment has a few open reading frames (ORFs), it has no gene, which makes it similar to a gene encoding ALR. Because DDL and ALR are target enzymes of DCS, it is of great interest to know whether these enzymes from the DCS-producing microorganism show resistance to DCS.
In the present study, an effort was made to clone ALR-and DDL-encoding genes from DCS-producing S. lavendulae ATCC25233. Both the S. lavendulae ALR and DDL, which were overproduced in an E. coli host vector system, were purified and characterized biochemically and kinetically. The present study suggests that the Streptomyces ALR and DDL function as self-resistance determinants.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids-E. coli strains TG1, JM109, and DH5␣ and plasmids pUC18 and pUC19 were used for the cloning experiments. E. coli XL1-Blue MRA (P2) was used for the construction of phage libraries. E. coli BL21(DE3)-pLysS and plasmid pET-21a(ϩ) (Novagen) were used for protein expression. E. coli was grown in LB medium (21) at 37 or 28°C. If necessary, ampicillin (100 g/ml) and/or chloramphenicol (34 g/ml) were added to the LB medium. For the cultivation of E. coli XL1-Blue MRA (P2), 0.2% maltose and 10 mM MgSO 4 were added to the LB medium. Streptomyces coelicolor A3 (2) (strain M145), used as the typical strain of the International Streptomyces Genome Project (22), and DCS-producing S. lavendulae ATCC25233 were grown at 28°C in a GMP medium (23) or a YEME medium (24).
DNA Manipulations-Plasmid DNA from E. coli was isolated by the standard method described previously (21). The chromosomal DNA from S. coelicolor A3 (2) and S. lavendulae was isolated from 100 ml of a culture grown at 28°C for 72 h according to a method described earlier (24). Phage DNA from plaque was isolated by the standard method described elsewhere (21).
Analysis of Genes Encoding the DCS Resistance Determinant from S. lavendulae-The chromosomal DNA (500 g) from S. lavendulae was partially digested with BamHI, purified by the phenol/chloroform extraction method, and precipitated by ethanol. The DNA fragments, cleaved within 10 -20 kb, were separated by sucrose gradient (10 -40%) centrifugation and precipitated by ethanol (21). After the 5Ј-dephosphorylation of DNA with bacterial alkaline phosphatase, the resulting DNA fragments were ligated to a BamHI-digested Lambda DASH II vector (Stratagene). In vitro packaging was performed using a Gigapack III Gold Packaging Extract (Stratagene) according to the supplier's instructions. The resulting phages were infected to E. coli XL1-Blue MRA (P2) and plated onto an NZYM medium (21) containing 1.2% agarose to generate plaques.
The plaques generated on the NZYM agarose plate were transferred to a nylon membrane (Hybond-Nϩ, Amersham Biosciences), and the phage DNA was fixed to the membrane by the alkaline treatment (21). Hybridization was performed at 65°C by using a 1.2-kb DNA fragment from pCSPC9 (which contains the DCS resistance gene of S. garyphalus (19)) as a probe DNA. The probe labeling, hybridization, and detection were performed with an AlkPhos direct labeling and detection kit (Amersham Biosciences) according to the supplier's instructions. One positive clone was obtained by plaque hybridization. The phage DNA, isolated from the positive plaque, had a 14-kb DNA insert from S. lavendulae.
The phage DNA containing a 14-kb DNA from S. lavendulae was digested with BamHI. The resulting DNA fragments (about 1.2, 2.0, 2.9, and 8.0 kb) were subcloned into pUC18 or pUC19. Using the resulting chimeric plasmids, the DNA sequence was determined with the BigDye terminator cycle sequencing ready reaction kit (Applied Biosystems) and ABI PRIZM 310 genetic analyzer (Applied Biosystems). Of the 14-kb DNA fragment, in the present study, we determined the nucleotide sequence of a 2820-bp DNA fragment including the S. lavendulae DDL gene. Genetic analysis was performed by using GENETYX-Mac software (Software Development, Tokyo, Japan) and the Frame Analysis program (25). The homology search was done with the FASTA program. The DNA sequence determined in this study has been submitted to the DNA Data Bank of Japan (DDBJ accession number AB176675).
Cloning and Analysis of an ALR Gene from S. lavendulae-The chromosomal DNA from S. lavendulae ATCC25233, which was digested with BamHI, was fractionated on 0.8% agarose gel electrophoresis and transferred to a Hybond-Nϩ membrane using the standard protocol (21). Southern hybridization analysis was done using a putative ALR gene (1176-bp) from S. coelicolor (22) as a probe. To obtain the probe DNA, PCR amplification was done using the S. coelicolor genomic DNA as a template together with a sense primer (5Ј-ATGAGCGAGACAACT-GCTCGGCGGGACGCG-3Ј) and an antisense primer (5Ј-TCATTCGTT-GACGTAGACGCGCGGGACCCGG-3Ј). PCR was done under the following conditions: an initial 5 min at 96°C and 3 min at 70°C; then, 24 cycles of 1 min at 96°C and 3 min at 70°C; and, finally, a 3-min extension period at 72°C. Probe labeling, hybridization, and detection were performed using an AlkPhos direct labeling and detection kit according to the manufacturer's instructions.
The probe DNA was hybridized to BamHI-digested genomic DNA that had a size of 3.0 kb. Therefore, BamHI digests of 2.5-3.5 kb were extracted from the agarose gel, purified, ligated to BamHI-digested pUC19, and then introduced into E. coli TG1. The resulting genomic libraries were screened using the colony hybridization technique (21). From ϳ8000 clones, 52 candidates carrying the putative ALR gene were obtained using the putative alr from S. coelicolor as a probe DNA. One of the chimeric plasmids, isolated from these candidate colonies, was hybridized to the probe. Results from the International Streptomyces Genome Project confirmed that a protein deduced from the nucleotide sequence of the S. lavendulae gene, which was inserted into the candidate plasmid, displayed a high similarity with the S. coelicolor putative ALR.
DNA sequencing was performed with the ABI PRIZM 310 genetic analyzer using the BigDye terminator cycle sequencing ready reaction kit according to the manufacturer's protocols. Using a combination of subcloning and chromosome-walking techniques, the entire nucleotide sequence of the cloned DNA fragment containing the S. lavendulae ALR gene, designated alrS, was determined and analyzed for the existence of ORF using GENETYX-Mac software. The ORFs were predicted using a frame analysis program (25). The similarity among proteins was searched using the FASTA program on the website. The sequence data obtained in this study has been submitted to the DDBJ (accession number AB176676).
Overexpression and Purification of S. lavendulae DDL-A gene encoding DDL from S. lavendulae was amplified by PCR using a sense primer, 5Ј-CACCATATGCGAATCGTGATCTTGTGTGGTGGAGAAG-C-3Ј (NdeI site underlined), and an antisense primer, 5Ј-CACCTCGA-GTCAGCGGGTGGCGAGGGACAC-3Ј (XhoI site underlined). PCR was done under the following conditions: 1 cycle of 5 min at 96°C, 1 min at 55°C, and 2 min at 72°C followed by 29 cycles of 1 min at 96°C, 1 min at 55°C, and 2 min at 72°C. The amplified DNA was digested with NdeI and XhoI and subcloned into the same sites of pET-21a(ϩ) to generate pET-ddlS. E. coli BL21(DE3)-pLysS harboring pET-ddlS was grown at 28°C in 6 liters of LB medium to an A 600 nm ϭ 0.5, whereupon isopropyl-␤-D-thiogalactopyranoside was added to the culture at the final concentration of 1 mM to induce the expression of ddlS. The E. coli cells were grown for 8 h at 28°C. The purification of the S. lavendulae DDL was carried out at 4°C: the E. coli cells were suspended in Buffer I (50 mM sodium phosphate (pH 7.5), 10 mM MgCl 2 , 2 mM dithiothreitol, and 1 mM EDTA⅐Na 2 ) and disrupted by sonication. The cell debris was removed by centrifugation at 17,000 ϫ g for 30 min. Solid ammonium sulfate was gradually added to the supernatant to 20% saturation and centrifuged to obtain the supernatant fluid. Solid ammonium sulfate was added to the supernatant fluid to 50% saturation, and the resulting precipitate was collected by centrifugation. The precipitate was dissolved in a small volume of Buffer I and dialyzed against the same buffer. After the dialysate was applied on a DEAE-Sepharose column (2.5 ϫ 10 cm, Amersham Biosciences) previously equilibrated with Buffer I, the column was washed with the same buffer. Elution was done with a 0 -500 mM KCl linear gradient concentration in Buffer I. The fractions containing the S. lavendulae DDL were pooled and dialyzed against Buffer II (20 mM Tris-HCl (pH 7.5), 2.5 M KCl, 1 mM ATP, 10 mM MgCl 2 , and 1 mM 2-mercaptoethanol) and then subjected to an Octyl-Sepharose column (1.5 ϫ 15 cm, Amersham Biosciences) previously equilibrated with Buffer II. Since no S. lavendulae DDL was bound to the column, the solution that passed through the column was collected. Finally, the solution carrying the DDL activity was dialyzed against Buffer III (50 mM sodium phosphate (pH 7.5), 10 mM MgCl 2 , 2 mM dithiothreitol, 1 mM EDTA⅐Na 2 , and 0.2 mM ATP) and applied on a DEAE-Sepharose column (1 ϫ 15 cm) equilibrated with Buffer III. The S. lavendulae DDL did not bind to the column, possibly because of the presence of ATP. Therefore, the solution that passed through the column was collected. Through these steps, the S. lavendulae DDL was purified to homogeneity.
Overexpression and Purification of S. lavendulae ALR-An ALR gene of S. lavendulae was amplified by PCR using the sense primer, 5Ј-CA-CCATATGAACGAGACACCGACGCGCGTG-3Ј (the underline indicates the NdeI cleavage site), and the antisense primer, 5Ј-TATCTCG-AGGCCGCCGAGGTAGACCCGGG-3Ј (the underline indicates the XhoI cleavage site). The amplified DNA was digested with NdeI and XhoI and then subcloned into the same sites of pET-21a(ϩ) to yield pET-alrS. The pET-alrS plasmid expresses ALR having His 6 tag at the C terminus. E. coli BL21(DE3)-pLysS harboring pET-alrS was grown in 3 liters of LB medium at 28°C. At the exponential phase of growth (A 600 nm ϭ 0.6), isopropyl-␤-D-thiogalactopyranoside was added to the culture at a concentration of 0.1 mM to express the ALR gene. After an additional incubation for 4.5 h, the cells were harvested by centrifugation and washed with a binding buffer (20 mM Tris-HCl (pH 7.9), 500 mM NaCl, and 5 mM imidazole. The washed cells (20 g wet weight) were suspended into 200 ml of the same buffer and disrupted by sonication at 4°C for 60 min. The cell debris was removed by centrifugation at 24,000 ϫ g twice for 20 min. The resulting cell-free extract was brought to 65% saturation with solid ammonium sulfate. The resulting precipitate was collected by centrifugation at 24,000 ϫ g for 20 min, dissolved in a binding buffer, and dialyzed against the same buffer. The dialysate was applied to a Ni(II)-chelated His-bind resin (Novagen) column (1.0 ϫ 30 cm) according to the manufacturer's protocol. The column was washed with a wash buffer (20 mM Tris-HCl (pH 7.9), 500 mM NaCl, and 60 mM imidazole) and eluted with a linear gradient of 60 -350 mM imidazole. The fractions having ALR activity were pooled and concentrated by ultrafiltration (Amicon Ultra, Millipore). The concentrate was loaded onto a Sephadex G-100 super fine (Amersham Biosciences) column (1.5 ϫ 120 cm) equilibrated with a 200 mM Tris-HCl buffer (pH 8.5) supplemented with 50 mM NaCl and eluted with the same buffer. The purified enzyme fractions were collected, concentrated by ultrafiltration, and stored at 4°C until use.
Overexpression and Purification of E. coli ALR-A gene encoding ALR from E. coli K12 W3110 was amplified by PCR using the primers 5Ј-CACCATATGCAAGCGGCAACTGTTGTGATT-3Ј (the underline indicates the NdeI cleavage site) and 5Ј-TATCTCGAGATCCACGTATT-TCATCGCGAC-3Ј (the underline indicates the XhoI cleavage site) according to the genome information (26). The amplified DNA was inserted into pET-21a(ϩ) to generate pET-K12alr. The cell-free extract of E. coli BL21(DE3)-pLysS harboring pET-K12alr was applied to a Ni(II)-chelated His-bind resin column (1.0 ϫ 30 cm) and washed with the binding buffer. Elution was done with the same buffer containing 500 mM imidazole. The enzyme fraction was concentrated by 65% ammonium sulfate precipitation, and the resulting precipitate was dissolved in a 50 mM ammonium phosphate buffer (pH 8.2) containing 50 mM NaCl followed by dialysis against the same buffer. The dialyzed solution was subjected to a DEAE-Sepharose column (1.0 ϫ 30 cm) equilibrated with the same buffer. The fraction that passed through the column containing ALR from E. coli K12 was concentrated by ultrafiltration.
Enzyme Assay of DDL-The DDL activity was monitored by the continuous ADP release-coupled assay method as described previously (27). This assay monitors the absorbance at 340 nm. The DDL activity at different pH values was compared with buffers prepared as follows. A solution containing 100 mM Tris, 100 mM glycine, and 100 mM MES was adjusted to pH 6.0 and 10.5. By mixing the two solutions, buffers with pH values of 6.0, 7.0, 8.0, 9.0, 9.5, and 10.0 were prepared. All reactions were carried out at 37°C. The protein concentrations were determined using a Bio-Rad Protein Assay (Bio-Rad), which is based on the method of Bradford (28).
Enzyme Kinetic Study of DDL-Kinetic assays for the purified S. lavendulae DDL were carried out by the continuous ADP release-coupled assay method (27) 4), in which the y intercept (K 2 /V max ) and slope (K 1 K 2 /V max ) provide the two K m values.
Because the K 1 value in the above equations is very small (13), the value can be ignored when the concentration of the substrate ([S]) is high; therefore, Equations 2 and 3 can be represented as Equations 5 and 6, respectively. We determined the K 2 value alone using these equations (Equations 5 and 6).
Enzyme Assay and Kinetic Study of ALR-The enzyme assay and kinetic study of ALR using circular dichroism (CD) spectrometry were performed by a novel method. 2 The CD signals of samples were measured using a spectropolarimeter (JU-720 type, JASCO, Japan).
To understand the inhibition mode of DCS to the S. lavendulae ALR, the assay was repeated using a reaction mixture incubated with DCS at various concentrations (0.05-0.5 mM) for more than 10 min before the reaction. [D-Ala], [L-Ala], and v were also calculated as described, 2 and numerical analysis was performed to maximally fit to Equations 7 and 8, which are equations applied for competitive and noncompetitive inhibition, respectively. ATCATATGGAAAAACTGCGGGTAGGAATC-3Ј (the underline indicates the NdeI cleavage site) and 5Ј-CCCAAGCTTTTACATTGT-GGT-TTTCAATGC-3Ј (the underline indicates the HindIII cleavage site) for ddlA and 5Ј-CACCATATGACTGATAAAATCGCGGTCCTG-3Ј (the underline indicates the NdeI cleavage site) and 5Ј-CACAAGCTTTTAGT-CCGCCAGTTCCAGAAT-3Ј (the underline indicates the HindIII cleavage site) for ddlB, and the amplified DNA was inserted into pET-21a(ϩ) to generate pET-ddlA and pET-ddlB, respectively. The ddlA and ddlB fragments including the T7 promoter and terminator region were then amplified by PCR using the primers 5Ј-CACGCATGCGAAATTA-ATACGACTCAC-3Ј and 5Ј-TATGCATGCCAAAAAACCCCTCAAGA-C-3Ј (the underline indicates the SphI cleavage site), and the amplified DNA was inserted into the SphI-digested pET-K12alr to generate pET-K12alr-ddlA and pET-K12alr-ddlB, respectively. On the other hand, after pET-ddlS was double-digested with XhoI and BglII, a 1.1-kb DNA fragment carrying ddlS was blunted and inserted into pET-alrS, which was digested with SphI and blunted to generate pET-alrS-ddlS.

RESULTS AND DISCUSSION
Cloning of Genes Encoding the DCS Resistance Determinant from DCS-producing S. lavendulae-We recently cloned a 3.5-kb DNA fragment from DCS-producing S. garyphalus, which includes a DCS resistance gene, designated orfB (19). We suggest that the orfB gene product, which may carry membrane-integral domains spanning the membrane 10 times, may be a transporter for the efflux of DCS to the outside cells. To determine whether orfB is conserved in another DCS-producing microorganism, we examined whether a gene homologous to orfB is located on the chromosome from DCS-producing S. lavendulae ATCC25233.
A genomic library of S. lavendulae, prepared in a phagederived vector, was screened with the S. garyphalus orfB as a probe DNA. A 2.8-kb DNA portion of the 14-kb DNA fragment inserted in the phage vector, which was hybridized with orfB, was analyzed for the nucleotide sequence. Frame analysis (25) suggested that three ORFs, designated orf I, II, and III, are present in the 2.8-kb region (Fig. 1). The central gene, designated orf II, consists of 903 bp and encodes a protein with a molecular weight of 30,930. The protein exhibits a 98.7% identity (300-aa overlap) to the protein encoded by orfB from S. garyphalus (19). That is, orf II shown in Fig. 1 should be called orfB, which suggests that the gene is conserved in both DCS-producing Streptomyces strains. A gene located downstream of orf II, designated orf III (456 bp), encodes a protein consisting of 151 aa with a molecular weight of 15,882. The orf III-encoded protein displays the highest similarity to an unknown protein from S. coelicolor (42.4% identity, 151-aa overlap) (22). The orf III gene product has an 89% identity to a protein encoded by a gene designated orfC, which is located in the 3Ј-adjacent region of the DCS resistance gene in S. garyphalus (19). Interestingly, a protein encoded by a gene, designated orf I (1,038 bp), exhibits a significant similarity (42.0% identity) to a DDL from P. aeruginosa (20). In a previous study (19), an incomplete gene from S. garyphalus, which is predicted to be the 3Ј-portion of the gene, was found to be present just upstream of orfB. The incomplete gene product from S. garyphalus is completely identical to a protein encoded by orf I from S. lavendulae. In addition, the order and transcriptional direction between orf I and orfB in S. lavendulae are the same as those in S. garyphalus. The predicted molecular weight and pI of the orf I-encoded protein (345 aa) are 35,987 and 4.81, respectively. As described below, the protein was confirmed to exhibit DDL activity using the gene product, which was purified to homogeneity. Therefore, orf I and the gene product are referred to hereafter as ddlS and DDL, respectively. Fig. 2 shows a comparison of the aa sequence of DDL from S. lavendulae with those from various bacteria. The amino acids that interact with ATP and D-Ala (30) are conserved except for Leu 320 , which corresponds to Leu 282 of the E. coli DdlB. In some cases, the Leu residue is replaced by Met (Fig.  2). Although the consensus sequence of the -loop in these DDLs is Ser (or Ala or Thr)-Lys-Tyr-Ile (or Met or Ser) (31), the loop in the S. lavendulae DDL is Ala-Lys-Tyr-Gln. The Gln residue, present in the -loop, is characteristic of D-Ala-D-Ser ligases, which belong to VanC, found in vancomycin-resistant bacteria (31,32).
Cloning and Sequence Analysis of a Gene Encoding ALR from S. lavendulae-We found that an ORF that is contained in the 2.8-kb DNA fragment cloned from S. lavendulae is homologous to the putative ALR from S. coelicolor A3 (2) M145 but is not complete. Therefore, we newly cloned an additional 500-bp DNA fragment, which is adjacent to the 2.8-kb DNA fragment, by conducting a chromosome-walking experiment. The nucleotide sequence analysis of the 3,296-bp DNA fragment suggests that it contains a gene encoding a complete ALR protein from S. lavendulae. As shown in Fig. 3, frame analysis (25) of the 3,296-bp DNA fragment suggests the presence of three complete ORFs, designated orf1, orf2, and orf3. One of these, orf1, consists of 1,134 bp, and a protein deduced from the nucleotide sequence has 378 aa, with a molecular mass of 39.9 kDa. The protein shows a 74.9% identity to a putative ALR from S. coelicolor (22). This orf1 is referred to as alrS hereafter. The nucleotide sequence of alrS was deposited in the DDBJ (accession no. AB176676). The aa sequence of the putative S. lavendulae ALR also shows a significant homology to ALRs from mycobacteria (33,34). A Lys residue in the S. lavendulae ALR, which is the putative binding site for PLP (35,36), is present (Lys 38 ), similar to other ALRs expressed by a few microbial sources. A Tyr residue, which plays an essential role in the racemization of alanine (37), is also conserved in the Streptomyces ALR (Tyr 270 ).
E. coli and S. typhimurium have been known to possess two kinds of closely related ALR-encoded genes (alr and dadX in E. coli and dal and dadB in S. typhimurium) (8 -10). A genomic Southern analysis performed using alrS as a probe indicated that the alrS-related gene, which is hybridized to the probe, was the only one in the S. lavendulae chromosome (data not shown).
A protein encoded by orf2 (1,176 bp) consists of 392 aa and exhibits the highest identity (71.8%) to a putative lipase from S. coelicolor (22). A protein encoded by orf3 (462 bp) consists of 154 aa and displays the highest identity (79.0%) to a putative ATP/GTP-binding protein from S. coelicolor (22). The organization of these ORFs in S. lavendulae ATCC25233 is identical to that in S. coelicolor A3 (2).
Overproduction and Purification of the S. lavendulae DDL-Because DDL is a target enzyme of DCS, it would be significant to determine whether the DDL of DCS-producing S. lavendulae exhibits resistance to DCS. We overproduced the S. lavendulae DDL using an E. coli host vector system and purified it to homogeneity (Fig. 4). The DDL shows a molecular mass of about 38 kDa on SDS-PAGE, which is almost the same as that calculated from the deduced aa sequence. The molecular mass, measured by gel filtration chromatography performed on a Sephacryl S-300 HR column (1.5 ϫ 120 cm, Amersham Bio-sciences), is about 67 kDa, suggesting that the S. lavendulae DDL is a dimeric protein, like DDL from E. coli, designated DdlB (13).
Purification of ALRs from S. lavendulae and E. coli-Each ALR from S. lavendulae and E. coli K12 W3110 was overproduced as a protein with the C-terminal His 6 tag in E. coli and purified to homogeneity (Fig. 5). As shown in Fig. 5, the molecular masses of the purified S. lavendulae and E. coli ALRs, as estimated by SDS-PAGE, are 42 and 40 kDa, respectively. The bacterial ALRs have been classified into two types of subunit structures, a monomer and a homodimer structure (38). Gel filtration chromatography with Sephacryl S-200HR (1.5 ϫ 120 cm, Amersham Biosciences) revealed that the S. lavendulae ALR has a molecular mass of about 80 kDa, suggesting that it may have a homodimer structure.
Enzyme Properties of the S. lavendulae DDL-DdlB from E. coli was observed to display a higher DDL activity at pH 9.2 than at pH 6.0 -7.5 (39). Therefore, the catalytic activity of the S. lavendulae DDL was measured by varying the pH in the reaction mixture. We observed that the Streptomyces DDL activity is 15 M⅐min Ϫ1 at pH 7.0 and 57 M⅐min Ϫ1 at pH 10.0, respectively, suggesting that the enzyme exhibits higher activity as the pH values increase. The -loop in the S. lavendulae DDL has an Ala-Lys-Tyr-Gln sequence, raising the question of whether the enzyme displays D-Ala-D-Ser ligase activity. A TLC assay (40) confirmed that the S. lavendulae DDL did not display D-Ala-D-Ser ligase or D-Ala-D-Lac activities (data not shown). The latter observation is consistent with the fact that D-Ala-D-Lac ligases, such as VanA (41,42) and VanB (43,44) from vancomycin-resistant bacteria, possess the consensus -loop sequence of Pro-Glu-Lys-Gly (31). The -loop consensus in D-Ala-D-Lac ligases from lactic acid bacteria, including Lactobacillus confusus, Lactobacillus salivarius, and Lactobacillus plantarum, has the Asn-(Lys/Met)-Phe-Val sequence (31).
The kinetic parameters of the S. lavendulae DDL were measured using an ADP release-coupled assay method (27) and compared with those for the S. typhimurium DdlA, the E. coli DdlA and DdlB, and the Enterococcus faecium VanA (13,42). The turnover number (k cat ) of the S. lavendulae DDL was 4 -10-fold lower than those of the E. coli DdlA and DdlB and the Salmonella DdlA (Table I). In addition, the K m value of the Streptomyces DDL for the second D-Ala substrate (K 2 ) was 4and 2-fold higher than those of DdlA and DdlB, respectively. The k cat value of the S. lavendulae DDL was much closer to  (42). c D-Ala-D-Ala ligase from DCS-producing S. lavendulae.  (45), whereas the E. coli DdlA and DdlB and the Salmonella DdlA were noncompetitively inhibited (13). The reason that there is a difference between the inhibition modes in these enzymes is currently unclear. DCS inhibits DdlAs and DdlB competitively, with K i values in the range of 9 -27 M (Table I). Although the S. lavendulae DDL was competitively inhibited by DCS, the K i value of DCS for the protein (920 M) was obviously higher (40ϳ100-fold) than those for DdlAs and DdlB. This value was close to that of the VanA ligase. The high K i value suggests that the S. lavendulae DDL may be involved in the self-resistance mechanism in DCS-producing S. lavendulae. The kinetic properties of the S. lavendulae DDL were similar to those of DdlAs and DdlB (K 2 for the second D-Ala and K i for D-Ala-D-Ala) and, in part, to those of VanA (k cat , K m for ATP, and K i for DCS), suggesting that the structure of the substrate-binding sites of the S. lavendulae DDL might be different from those of the enzymes. Therefore, crystallization experiments are in progress to determine the three-dimensional structure of the S. lavendulae DDL. 3 Kinetic Studies of Both ALRs-The kinetic parameters of the S. lavendulae ALR and the E. coli K12 W3110 ALR were determined using a CD assay that we developed. 2 The K m values of both ALRs were not significantly different from each other, whereas the k cat value of the S. lavendulae ALR was twice as large as that of the E. coli ALR.

TABLE II Kinetic parameters of ALR computed as a competitive or noncompetitive inhibition model
The resultant parameters of S. lavendulae ALR computed as a competitive or noncompetitive inhibition model are shown in Table II. The equilibrium constants (K eq ) (46) in each analysis are 1.27 (for competitive) and 1.12 (for noncompetitive), which are almost the same as the theoretical value (1.0). However, the value of the correlation coefficient (C c ) in each case is equal (0.977 and 0.978), and the K m values are largely different from the results without DCS. 2 Therefore, neither competitive nor noncompetitive inhibition is applied to the inhibition mode of DCS to the S. lavendulae ALR.
Time-dependent Inactivation by DCS of ALRs-Because it is difficult to apply the inhibition mode of DCS to each mechanism based on steady-state equilibrium (Equations 7 and 8), an attempt was made to apply the inhibition mode of DCS based on the time-dependent inactivation manner (16). This manner originates from the fact that DCS reacts with PLP bound to the enzyme (E) and forms a complex of a PLP-unbound enzyme (EЈ) and a 3-hydroxyisoxazole pyridoxamine 5Ј-phosphate derivative (X) (16, 17) (Scheme 1).
To investigate the effect of PLP degeneration on the remaining activity of ALR, the ALR activities after incubation with DCS at given times were analyzed by observing the CD signal at 205 nm (Fig. 6, A and B). The slope of the regression line was defined as the ALR activity (v) at each incubation interval, and FIG. 6. Effect of PLP degeneration on the remaining activity of ALR after incubation with DCS. After ALR from S. lavendulae or E. coli was incubated with 2.0 mM DCS for the given times (0, 2, and 8 min), it was added to a solution (3 ml) consisting of a 30 mM sodium phosphate buffer (pH 8.2) and 4 mM D-Ala. The CD signal at 205 nm in the reaction mixture is shown as dots (A and B). C, the double reciprocal plots for each enzyme. The ALRs from S. lavendulae and E. coli are shown as closed circles and closed squares, respectively. the decrease of the remaining activity was evaluated from Equation 9. The value of k app is an apparent rate constant. At the beginning of the reaction, [EЈ⅐X] is regarded as zero, and the reverse reaction (k Ϫ2 ) is ignored in Scheme 1. This equation means that k app can be regarded as k 2 ⅐[I]/(K I ϩ [I]) at the initial phase of the reaction. Using these equations, K I and k 2 are determined from double reciprocal plots ( Fig. 6C and Table III). As shown in Table III, the K I values of DCS for both ALRs are similar, but the k 2 value of S. lavendulae ALR is smaller than that of E. coli ALR. This kinetic experiment for the S. lavendulae ALR suggests that the time-dependent inactivation rate of the enzyme by DCS is absolutely slower than that of the E. coli ALR. It may be concluded that ALR from DCS-producing S. lavendulae is also one of the self-resistance determinants.
Comparison of the Inhibitory Effect of DCS with That of LCS on the S. lavendulae ALR Activity-Proteins which carry PLP as a cofactor, such as aminotransferases, are inhibited by LCS as they are by DCS. In fact, because the catalytic activity of the Bacillus stearothermophilus ALR is inhibited by LCS (16), we also examined the inhibitory effects of LCS on the S. lavendulae and E. coli ALR activities. Table III lists the kinetic parameters of both enzymes, which were determined by a CD spectrometric assay. The k 2 /K I value (4.5 ϫ 10 Ϫ3 s Ϫ1 ) of DCS for the S. lavendulae ALR is smaller than that (9.2 ϫ 10 Ϫ3 s Ϫ1 ) for the E. coli ALR, suggesting that the former enzyme displays resistance to DCS when compared with the latter. However, the k 2 /K I value (0.48 ϫ 10 Ϫ3 s Ϫ1 ) of LCS for the E. coli ALR is lower than that of DCS. The time-dependent inactivation of the   a ND, not determined because the decrease of ALR activity was not observed in the given time range.