Cloning, Expression, and Characterization of Bacterial l-Arabinose 1-Dehydrogenase Involved in an Alternative Pathway of l-Arabinose Metabolism*

Azospirillum brasiliense converts l-arabinose to α-ketoglutarate via five hypothetical enzymatic steps. We purified and characterized l-arabinose 1-dehydrogenase (EC 1.1.1.46), catalyzing the conversion of l-arabinose to l-arabino-γ-lactone as an enzyme responsible for the first step of this alternative pathway of l-arabinose metabolism. The purified enzyme preferred NADP+ to NAD+ as a coenzyme. Kinetic analysis revealed that the enzyme had high catalytic efficiency for both l-arabinose and d-galactose. The gene encoding l-arabinose 1-dehydrogenase was cloned using a partial peptide sequence of the purified enzyme and was overexpressed in Escherichia coli as a fully active enzyme. The enzyme consists of 308 amino acids and has a calculated molecular mass of 33,663.92 Da. The deduced amino acid sequence had some similarity to glucose-fructose oxidoreductase, d-xylose 1-dehydrogenase, and d-galactose 1-dehydrogenase. Site-directed mutagenesis revealed that the enzyme possesses unique catalytic amino acid residues. Northern blot analysis showed that this gene was induced by l-arabinose but not by d-galactose. Furthermore, a disruptant of the l-arabinose 1-dehydrogenase gene did not grow on l-arabinose but grew on d-galactose at the same growth rate as the wild-type strain. There was a partial gene for l-arabinose transport in the flanking region of the l-arabinose 1-dehydrogenase gene. These results indicated that the enzyme is involved in the metabolism of l-arabinose but not d-galactose. This is the first identification of a gene involved in an alternative pathway of l-arabinose metabolism in bacterium.

reported that Azospirillum brasiliense, a bacterium used in this study, can grow on L-arabinose as a sole carbon source and has the first alternative pathway of L-arabinose metabolism (13). This is the first enzymological and molecular biological analysis of the proposed pathway of L-arabinose metabolism.

EXPERIMENTAL PROCEDURES
Bacterial Strain, Culture Conditions, and Preparation of Cell-free Extracts-A. brasiliense ATCC29145 was purchased from RIKEN BioResource Center (Saitama, Japan) and cultured aerobically with vigorously shaking at 30°C for 24 h in a minimal medium (13) (pH 6.8), containing 4.0 g of KH 2 PO 4 , 6.0 g of K 2 HPO 4 , 0.2 g of MgSO 4 ⅐H 2 O, 0.1 g of NaCl, 0.026 g of CaSO 4 ⅐2H 2 O, 1.0 g of NH 4 Cl, 0.01 g of FeCl 3 ⅐6H 2 O, 0.002 g of NaMoO 4 ⅐2H 2 O, and 0.0001 g of biotin/liter supplemented with 37 mM L-arabinose. L-Arabinose was sterilized separately by filtration and added to the medium. The grown cells were harvested by centrifugation at 30,000 ϫ g for 20 min, washed with 20 mM potassium phosphate buffer (pH 7.0), containing 2 mM MgCl 2 and 10 mM 2-mercaptoethanol (referred to as Buffer A), and stored at Ϫ35°C until used. The washed cells were suspended in Buffer A, disrupted by sonication for 20 min with appropriate intervals on ice using ASTRASON Ultrasonic Liquid Processor XL2020 (Misonix Inc., New York), and then centrifuged at 108,000 ϫ g for 20 min at 4°C to obtain cell-free extracts.
PAGE-SDS-PAGE was performed as described by Laemmli (29). Nondenaturing PAGE was performed by omitting SDS and 2-mercaptoethanol from the solution used in SDS-PAGE. Proteins on the gel were stained with Coomassie Brilliant Blue R-250 and destained with 7.5% (v/v) acetic acid in 25% methanol.
Enzyme Activity Assay-L-Arabinose 1-dehydrogenase activity was assayed routinely in the direction of L-arabinose oxidation by measuring the reduction of NAD(P) ϩ at 340 nm at 30°C using Jasco spectrophotometer model V-550 (Japan Spectroscopic Co., Ltd., Tokyo, Japan). The standard assay mixture contained 10 mM L-arabinose in 100 mM Tris-HCl (pH 9.0) buffer. The reaction was started by the addition of 10 mM NAD(P) ϩ solution (100 l) with a final reaction volume of 1 ml. The kinetic parameters, K m and k cat values, were calculated by Lineweaver-Burk plot. Protein concentrations were determined by the method of Lowry et al. (30) with bovine serum albumin as a standard.
Zymogram Staining Analysis-The cell-free extracts or purified enzyme were separated on nondenaturing PAGE with 12% gel at 4°C. The gels were then soaked in 10 ml of staining solution (31) consisting of 100 mM Tris-HCl (pH 9.0), 100 mM L-arabinose, 0.25 mM nitro blue tetrazolium, 0.06 mM phenazine methosulfate, and 1 mM NAD(P) ϩ at room temperature for 15 min. The dehydrogenase activity appeared as a dark band.
Purification of L-Arabinose 1-Dehydrogenase-All purification steps were performed below 4°C. The cell-free extracts were fractionated between 50 and 60% saturation of (NH 4 ) 2 SO 4 . The precipitate was dissolved in a small volume of Buffer A, and the solution was then dialyzed against a large volume of Buffer A containing 1.3 M (NH 4 ) 2 SO 4 overnight. All chromatography was carried out using an ⌬KTA purifier system (Amersham Biosciences). After insoluble materials were removed by centrifugation, the supernatant was applied to a HiPrep 16/10 Butyl FF column (1.6 ϫ 10 cm, Amersham Biosciences) equilibrated with Buffer A containing 1.3 M (NH 4 ) 2 SO 4 and washed with the same buffer. Proteins were eluted using a reversed linear gradient of 1.3-0 M (NH 4 ) 2 SO 4 in Buffer A (300 ml). The fractions with high enzymatic activity were pooled and dialyzed overnight against a large volume of Buffer A. The enzyme solution was loaded onto a column of HiPrep 16/10 Q FF (1.6 ϫ 10 cm, Amersham Biosciences) equilibrated with Buffer A and washed thoroughly with the same buffer. The column was developed with 300 ml of linear gradient 0 -1 M NaCl in Buffer A. The fractions containing L-arabinose 1-dehydrogenase activity were combined and dialyzed against a large volume of 5 mM potassium phosphate (pH 7.0), containing 10 mM 2-mercaptoethanol (referred to as Buffer B) overnight. The enzyme solution was applied to a column of Ceramic Hydroxyapatite Type I (1.6 ϫ 5 cm, Bio-Rad), equilibrated with Buffer B. The column was washed thoroughly with the same buffer and developed with 200 ml of linear gradient 0.005-0.5 M potassium phosphate in Buffer B. The fractions with high enzymatic activity were combined and concentrated by ultrafiltration with Centriplus YM-30 (Millipore) at 18,000 ϫ g for ϳ2 h. The enzyme solution was loaded onto a column of HiLoad 16/60 Superdex 200 pg column (1.6 ϫ 60 cm, Amersham Biosciences) equilibrated with Buffer A. The active fractions were pooled, concentrated, and re-loaded onto the same column. Proteins in the fractions containing high activity L-arabinose 1-dehydrogenase were analyzed by SDS-PAGE, and the fractions containing a single protein were collected, concentrated, dialyzed against Buffer C (100 mM Tris-HCl (pH 9.0), containing 2 mM MgCl 2 , 10 mM L-arabinose, 1 mM dithiothreitol, and 50% (v/v) glycerol), and stored at Ϫ35°C until used.
The native molecular mass of L-arabinose 1-dehydrogenase was estimated by gel filtration. The purified enzyme was loaded onto a HiLoad 16/60 Superdex 200 pg column equilibrated with 20 mM potassium phosphate (pH 7.0), containing 2 mM MgCl 2 and 1 mM dithiothreitol. High and low molecular weight gel filtration calibration kits (Amersham Biosciences) were used as molecular markers.
Product Identification by HPLC-The product of the dehydrogenation reaction of L-arabinose was identified by HPLC with a Multistation LC-8020 model II system (Tosoh). L-Arabino-␥-lactone was chemically synthesized with boiling potassium arabonate in 0.2 M HCl for 5 min. Potassium arabonate was prepared by the hypoiodite-in-methanol oxidization of L-arabinose (32). A solution containing 100 mM Tris-HCl (pH 9.0), 10 mM L-arabinose, 10 mM NADP ϩ , and the purified enzyme (10 g) was incubated for 30 min at 30°C, and 100 l of this solution was then analyzed. Samples were applied at 30°C into an Aminex HPX-87H Organic Analysis column (300 ϫ 7.8 mm, Bio-Rad) linked to an RID-8020 refractive index detector (Tosoh) and eluted with 5 mM H 2 SO 4 at a flow rate of 0.6 ml/min.
Determination of N-terminal and Internal Amino Acid Sequences-To determine the N-terminal amino acid sequence of L-arabinose 1-dehydrogenase, the purified enzyme was separated by SDS-PAGE with 12% (w/v) gel, and then transferred to Hybond TM -P (Amersham Biosciences) at 3 mA/cm 2 for 0.5 h in a transfer buffer (10 mM CAPS (pH 11) containing 10% (v/v) methanol) with a horizontal electrophoretic blotting system (model AE-7500, Atto). After staining and destaining the protein, an area of the membrane corresponding to the protein band of L-arabinose 1-dehydrogenase was excised and analyzed with a Procise TM 492 HT protein sequencer (Applied Biosystems).
Chemical digestion with cyanogen bromide (CNBr) was carried out to determine internal amino acid sequences (33). The purified L-arabinose 1-dehydrogenase (100 g) was dialyzed overnight against deionized water and lyophilized. The enzyme protein was digested chemically at room temperature in 70% (v/v) formic acid containing 1% (w/v) CNBr (100 l) in the dark and under N 2 overnight. The solution was diluted with 900 l of deionized water, frozen with liquid N 2 , and lyophilized. The sample was dissolved in SDS-PAGE sample buffer (500 mM Tris-HCl (pH 6.8), containing 5% (w/v) SDS, 10% (v/v) glycerol, 0.25% (w/v) bromphenol blue, and 5% (v/v) 2-mercaptoethanol) and separated by SDS-PAGE with 18% (w/v) gel. Peptide fragments on the gel were trans-ferred to a Hybond TM -P membrane as described above. After staining and destaining, areas of the membrane corresponding to the two peptide fragments from the L-arabinose 1-dehydrogenase (see in Fig. 2B) were excised and sequenced.
Cloning of L-Arabinose 1-Dehydrogenase Gene-The N-terminal and internal peptide sequences were used to design PCR primers for amplification of a partial DNA fragment of the L-arabinose 1-dehydrogenase gene. Eight upstream primers (U1-U8, 26-mer) were designed from (M)SDQVSLGV, the N-terminal amino acid sequence, as follows: 5Ј-ATG(TCN/AGY)GAYCARGTN(TCN/AGY) (CTN/TTR)GGN-GT-3Ј. Two downstream primers (D1 and D2, 26-mer) were designed from the internal amino acid sequence (M)LEKPPGAT, as follows: 5Ј-GTNGCNCCNGGNGGYTTYTC(NAG/YAA)CAT-3Ј. A. brasiliense genomic DNA was prepared using a DNeasy TM tissue kit (Qiagen). PCR was carried out using a PCR Thermal Cycler-Personal (Takara) for 30 cycles in a 50-l reaction mixture containing 10 pmol of primers, 1.25 units of Ex Taq DNA polymerase (Takara), and 300 ng of A. brasiliense genomic DNA under the following conditions: denaturation at 98°C for 10 s, annealing at 50°C for 30 s, and extension at 72°C for 30 s, each for 30 cycles. Based on the results of genomic PCR using each set of primers, U6 and D1 were chosen for cloning. The sequences of U6 and D1 were 5Ј-ATGAGYGAYCARGTNTCNTTRGGNGT-3Ј and 5Ј-GTNGCNCCNGGNGGYTTYTCNAGCAT-3Ј, respectively. A single PCR product with a length of ϳ300 bp was purified, cloned into a pGEM-T vector (Promega) (referred to as pGEM1), and sequenced using a Dual CyDye TM terminator sequencing kit (Veritas) and appropriate primers with Long-Read Tower, UBC DNA sequencer (Amersham Biosciences). The inserted fragment was amplified with U6 and D1 primers and with pGEM1 as a template DNA, and the PCR product was purified and utilized as a probe for Southern and Northern blot analysis and colony hybridization (34).
For Southern blot analysis, ϳ1.8 g of A. brasiliense genomic DNA was digested with six restriction enzymes, EcoRI, HindIII, NotI, PstI, SalI, and XbaI, separated on 1% (w/v) agarose gel and blotted to Hybond TM -N (Amersham Biosciences) by capillary transfer using 10ϫ SSC as a transfer buffer (1ϫ SSC is 15 mM sodium citrate (pH 7.0), and 0.15 M NaCl). The blotted filter was cross-linked in an UV cross-linker CX-2000 (Ultra-Violet Products, Ltd.). A double-stranded probe DNA was labeled with digoxigenin-11-dUTP and hybridized using a DIG-High Prime DNA labeling and detection starter kit (Roche Applied Science). Membrane was visualized using a nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate reagent detection system (Roche Applied Science).
A. brasiliense partial genomic library was prepared with genomic DNA with NotI based on the results of Southern blot analysis. The DNA fragments corresponding to a positive band in size (ϳ2.0 kbp of the length) were ligated to a plasmid pBluescript SK(Ϫ) (Stratagene). Colony hybridization was carried out under the same conditions as Southern blot analysis except for the use of nylon membranes for colony and plaque hybridization (Roche Applied Science). The plasmid from a positive clone (referred to as pBS1) was purified, and the inserted A. brasiliense genome fragment was sequenced.
Northern Blot Analysis-A. brasiliense cells were cultured at 30°C to the mid-log phase (A 600 ϭ 0.6 -0.8) in minimal medium supplemented with 37 mM appropriate sugar (D-glucose, L-arabinose, D-galactose, or D-xylose) or nutrient medium (10 g of peptone, 10 g of meat extract, and 5.0 g of NaCl (pH 7.0 -7.2)) and harvested by centrifugation. Total RNAs from A. brasiliense were prepared with an RNeasy mini kit (Qiagen) and subsequently treated with RNase-free DNase I. The isolated RNA (4 g) was subjected to electrophoresis on 1.2% (w/v) agarose gel containing 0.66 M formaldehyde. The subsequent steps were performed using the same methods as for Southern blot analysis.
Cloning of the L-Arabinose 1-Dehydrogenase Gene into Expression Plasmid Vector-To introduce the restriction site for BglII at the 5Ј-end and PstI at the 3Ј-end of the L-arabinose 1-dehydrogenase gene, PCR was carried out using pBS1 as a template and the following two primers (lowercase letters indicate additional bases for introducing digestion sites of BglII and PstI (underlined letters)): 5Ј-caccatagaTCTGATCAG-GTTTCGCTGGGTG-3Ј (HIS BglII ) and 5Ј-gcttggctgcagTCAGCGGC-CGAACGCGTCG-3Ј (HIS PstI ). The amplified DNA fragment was introduced into BamHI-PstI sites in pQE-80L (Qiagen), a plasmid vector for conferring N-terminal His 6 tag on the expressed proteins, to obtain pHIS WT .
Site-directed Mutagenesis-The following sense and antisense primers were designed to introduce the mutations into the L-arabinose 1-dehydrogenase gene (the mutated regions are underlined): to substitute Ala for Asp 168 (D168A), 5Ј-CGGCGTGTTCGCGCCGGGCATC-3Ј (D168A S ) and 5Ј-GATGCCCGGCGCGAACACGCCG-3Ј (D168A AS ); to substitute Ala for Asn 172 (N172A), 5Ј-CCCGGGCATCGCG-GCGCTGTCG-3Ј (N172A S ) and 5Ј-CGACAGCGCCGCGATGC-CCGGG-3Ј (N172A AS ). The mutations were introduced by sequential steps of PCR (35) with small modifications. In the first round, two reactions, I and II, were performed with the appropriate primers and pHIS WT as a template: reaction I, HIS BglII and one of the antisense primers containing the mutations; and reaction II, one of the sense primers containing the mutations and HIS PstI . In the final amplification step, purified overlapping PCR products were used as templates and HIS BglII and HIS PstI as primers. The final PCR products were cloned into pQE-80L to obtain plasmids pHIS D168A and pHIS N172A , respectively. The coding region of the mutated genes was confirmed by subsequent sequencing in both directions.
Functional Expression and Purification of His 6 -tagged L-Arabinose 1-Dehydrogenase-E. coli DH5␣ harboring the expression plasmid for the His 6 -tagged wild-type and mutated enzymes was grown at 37°C to a turbidity of 0.6 at 600 nm in Super Broth medium (12 g of tryptone, 24 g of yeast extract, 5 ml of glycerol, 3.81 g of KH 2 PO 4 , and 12.5 g of K 2 HPO 4 /liter (pH 7.0)) containing 50 mg/liter ampicillin. After the addition of 1 mM of isopropyl-␤-D-thiogalactopyranoside, the culture was further grown for 6 h to induce the expression of His 6 -tagged L-arabinose 1-dehydrogenase protein. Cells were harvested and resuspended in Buffer D (50 mM sodium phosphate containing 2 mM MgCl 2 , 300 mM NaCl, 1 mM L-arabinose, 10 mM 2-mercaptoethanol, and 10 mM imidazole (pH 8.0)). The cells were then disrupted by sonication, and the solution was centrifuged. The supernatant was loaded onto a nickelnitrilotriacetic acid spin column (Qiagen) equilibrated with Buffer D. The column was washed three times with Buffer E (Buffer D containing 10% (v/v) glycerol and 50 mM imidazole instead of 10 mM imidazole (pH 8.0)). The enzymes were then eluted with Buffer F (Buffer E containing 250 mM imidazole instead of 50 mM imidazole (pH 8.0)). The elutant was dialyzed against Buffer C and stored at Ϫ35°C until use.
Western Blot Analysis of His 6 -tagged L-Arabinose 1-Dehydrogenase-For Western blot analysis, the purified L-arabinose 1-dehydrogenase from A. brasiliense and/or recombinant His 6 -tagged L-arabinose 1-dehydrogenase from E. coli was separated by SDS-PAGE, and the proteins on the gels were transferred onto a nitrocellulose membrane (Hybond TM -ECL; Amersham Biosciences). Western blot analysis was carried out using the ECL TM Western blotting analysis system (Amersham Biosciences) and RGS⅐His horseradish peroxidase antibody, a horseradish peroxidase-fused mouse monoclonal anti-body against Arg-Gly-Ser-His 6 in the N-terminal additional peptide of the expressed recombinant proteins (Qiagen).
Disruptant Construction-The overall scheme of the plasmid construction for disruption of the L-arabinose 1-dehydrogenase gene is shown in Fig. 9A. The Tn5-derived 1.3-kb BamHI kanamycin resistance (Km r ) cassette of pUC4K (Amersham Biosciences) was inserted into the single BamHI site in the coding sequence of the L-arabinose 1-dehydrogenase gene of pHIS WT to yield pHIS WT::Km . To introduce the restriction site for MfeI at the 5Ј-and 3Ј-end of the DNA fragment containing the Km r gene in the L-arabinose 1-dehydrogenase gene, PCR was carried out using pHIS WT::Km as a template and the following two primers (lowercase letters indicate additional bases for introducing digestion sites of MfeI (underlined letters)): 5Ј-caccatcaattgGATCAGGTTTCGCTGG-GTGTCGTCGGCATCG-3Ј (MfeI-up) and 5Ј-gcttggcaattgTCAGCG-GCCGAACGCGTCGGTCTGCACGCGC-3Ј (MfeI-down). The 2.3-kbp MfeI DNA fragment was subcloned into the EcoRI site in the chloramphenicol resistance (Cm r ) cassette of the suicide vector pSUP202 (36) to yield pSUP WT::Km .
E. coli S17-1 (36) was transformed with pSUP WT::Km , and then the transformant was further mobilized to A. brasiliense by biparental mating. The transconjugants were selected on a minimal medium agar plate supplemented with 5 g of sodium malate and 25 g of kanamycin/liter using Km r (the presence of Km r cassette) and Tc S (loss of pSUP202) phenotypes. The construction was confirmed by genomic PCR and Southern hybridization on total DNA digested with NotI. One of the resulting disruptants of A. brasiliense was named ⌬ARA5034 and was used in this study.
Amino Acid Sequence Alignment and Phylogenetic Analysis-Protein sequence of L-arabinose 1-dehydrogenase from A. brasiliense was analyzed using the Protein-BLAST and ClustalW program distributed by DDBJ (www.ddbj.nig.ac.jp). The phylogenetic tree was produced using the TreeView 1.6.1. program.

RESULTS
Purification of L-Arabinose 1-Dehydrogenase from A. brasiliense-NAD ϩ -and NADP ϩ -dependent enzymatic oxidization of L-arabinose was found in cell-free extracts prepared from A. brasiliense cells grown on L-arabinose as a sole carbon source (Fig. 2C). The L-arabinose 1-dehydrogenase was purified by ammonium fractionation and five chromatographic steps. A typical result of purification is summarized in Table 1. During the purification procedure, the ratio of NADP ϩ -and NAD ϩ -linked activity remained almost constant, suggesting the presence of only one protein as L-arabinose 1-dehydrogenase. The purified enzyme was electrophoretically homogeneous (Fig. 2, A and C) and showed NAD ϩ -and NADP ϩ -dependent specific activity of 25 and 44 units/mg protein, respectively (Table 1). SDS-PAGE revealed only one subunit with an apparent value of 39.5 Ϯ 0.7 kDa ( Fig. 2A). To estimate native molecular mass, the enzyme was loaded onto a HiLoad 16/60 Superdex 200 pg column. The value from the calibration curve with marker proteins was ϳ46.4 Ϯ 1.9 kDa (data not shown), indicating that the L-arabinose 1-dehydrogenase is monomeric. There was no significant increase in activity in the presence of MgCl 2 , MnCl 2 , ZnCl 2 , CoCl 2 , NiCl 2 , or CaCl 2 at final concentrations of 1 mM (data not shown). Zymogram staining analysis showed a major active band in the cell-free  extracts. The position corresponded to that of the purified enzyme (Fig.  2C). A minor staining band with NAD ϩ as a coenzyme may be derived from the concomitant activity of other NAD ϩ -dependent sugar dehydrogenases. Substrate Specificity and Kinetic Analysis-In addition to L-arabinose, various sugars were tested as substrates for dehydrogenase. In the first approach, the sugar concentrations were fixed at 10 mM. Activity was observed with L-arabinose, D-galactose, D-xylose, and D-talose but not with D-arabinose, D-glucose, D-ribose, L-xylose, L-mannose, L-lyxose, and D-fructose (less than 1% of the activity with L-arabinose). The enzyme was subjected to further kinetic analysis with the active substrates, and the determined parameters are listed in Table 2. The catalytic efficiency (k cat /K m ) values with L-arabinose, D-galactose, and D-xylose in the presence of NADP ϩ were significantly higher than those in the presence of NAD ϩ because of the lower values of K m and the higher values of k cat . Furthermore, when L-arabinose was used as a substrate, the enzyme showed 5.6-fold higher affinity with NADP ϩ (K m ϭ 0.0095 Ϯ 0.0083 mM) than NAD ϩ (K m ϭ 0.053 Ϯ 0.020 mM). The kinetic parameters of D-galactose were very similar to those of L-arabinose in the presence of either NAD ϩ or NADP ϩ , indicating that the enzyme functions not only as L-arabinose 1-dehydrogenase but also "D-galactose 1-dehydrogenase." The enzyme showed 51-and 79-fold lower values of k cat /K m with D-talose in the presence of NAD ϩ and NADP ϩ , respectively, compared with those with L-arabinose. k cat /K m values with D-xylose were lower than those with D-talose by 2 orders of magnitude in the presence of either NAD ϩ or NADP ϩ because of remarkably high K m values and low k cat values.
Enzyme activities with these sugars, and the stereoconfiguration of these sugars are shown in Fig. 3. Active sugars for L-arabinose 1-dehydrogenase belong not only to pentose(s) but also hexose(s), indicating    Table 2 with NAD ϩ (light gray bar) and NADP ϩ (dark gray bar). Asterisks indicate the inactive substrate. Gray-shaded configuration is identical to that of L-arabinose.

A Novel Bacterial L-Arabinose 1-Dehydrogenase
FEBRUARY 3, 2006 • VOLUME 281 • NUMBER 5 that the activity is not dependent on the C-5 and C-6 configuration. D-Galactose has the same configuration as L-arabinose at C-2, C-3, and C-4. The activity was also found with D-talose, C-1 epimer of D-galactose, but not with L-mannose and D-glucose, C-3 and C-4 epimer, respectively. Similarly, the activity with D-xylose, the C-4 epimer of L-arabinose, decreased significantly, and no activity was observed with L-lyxose, the C-3 epimer of L-arabinose. It therefore appeared that the enzyme prefers the L-arabinose-specific configuration at C-3 and C-4. It has been reported that the product of the enzyme reaction in the cell-free extract system is L-arabino-␥-lactone by paper chromatography (10,13,14). We reinvestigated the product of the purified enzyme by HPLC as shown in Fig. 4. The retention time of L-arabino-␥-lactone was slightly earlier than that of L-arabinose, and the main product from L-arabinose was L-arabino-␥-lactone. The minor concomitant L-arabonate is probably due to spontaneous lactone hydrolysis (see Fig. 4B). These results indicated that the enzyme possessed only dehydrogenase activity with L-arabinose but not lactonase with L-arabino-␥-lactone.
Cloning of the Gene Encoding L-Arabinose 1-Dehydrogenase and Its Functional Expression in E. coli-The N-terminal amino acid sequence up to 15 amino acid residues of L-arabinose 1-dehydrogenase was as follows: SDQVSLGVVGIGKIA. To determine the internal amino acid sequence, the purified enzyme was digested chemically by CNBr, which digests specifically at the C termini of methionyl residues in polypeptides. As shown in Fig. 2B, two peptide bands with ϳ26.4 and 0.92 kDa of molecular mass (referred to as fragments I and II, respectively) were found on SDS-polyacrylamide gel. The 15-amino acid sequence of Fragment II was completely identical to the N-terminal sequence. The amino acid sequence up to 15 amino acid residues of Fragment I was as follows: (M)LEKPPGATLGEVAVL, suggesting that Fragment II is located upstream of Fragment I in the protein sequences. Using synthetic DNA primers designed from N-terminal and internal amino acid sequences, an ϳ300-bp nucleotide fragment of the L-arabinose 1-dehydrogenase gene was obtained by genomic PCR. Southern blot analysis with this DNA fragment as a probe showed a single band on each restriction enzyme digestion of A. brasiliense genomic DNA with EcoRI, HindIII, NotI, PstI, SalI, and XbaI of ϳ3.9, 15, 2.0, 4.5, 2.2, and 17 kbp (data not shown), revealing only a single copy of the gene on the genome. Based on the results, we isolated a 1,805-bp NotI fragment containing the full L-arabinose 1-dehydrogenase gene (Fig. 5). The G ϩ C content of the entire sequence was significantly high (ϳ69%), consistent with that of chromosomal DNA from A. brasiliense (70%) (37). The L-arabinose 1-dehydrogenase was 927 bp long, and a putative ribosomebinding site (Shine-Dalgarno sequence, AGGAG) was found 9 -13 bases upstream of the ATG codon. The deduced amino acid sequence of the protein (polypeptide of 309 amino acids with a predicted mass of 33,795.12 Da) agreed with the determined N-terminal amino acid sequences from the corresponding purified protein, except for removed formylmethionine. The internal amino acid sequences determined by the digestion of purified protein with CNBr were found at positions 89 -103 of the protein sequences. The nucleotide sequence was submitted to GenBank TM with accession number AB211983. Two partial ORFs were found upstream and downstream of the L-arabinose 1-dehydrogenase gene (referred to as ORF 1 and ORF 2 , respectively) (Fig. 5). The deduced ORF 1 and ORF 2 proteins showed significant identity with dihydrodipicolinate synthase/N-acetylneuraminate lyase and periplasmic L-arabinose-binding proteins (involving in the ABC-type transport system) from many organisms, respectively. L-Arabinose 1-dehydrogenase was overexpressed in E. coli cells by induction with isopropyl-␤-D-thiogalactopyranoside as a His 6 -tagged enzyme and purified homogeneously with a nickel-chelating affinity column (Fig. 6A). The mobility of the recombinant enzyme in SDS-PAGE and zymogram-staining analysis (Fig. 6, A and C) was slightly  later than that of the native enzyme purified from A. brasiliense cells because of the additional 13 amino acid residues, including a His 6 tag at the N terminus. Western blot analysis with anti-His 6 tag antibody confirmed the His 6 tag in the enzyme expressed in E. coli (Fig. 6B). The enzyme showed similar NADP ϩ preference and kinetic parameters for L-arabinose in the presence of NAD(P) ϩ to those of the native enzyme (Tables 2 and 3), confirming that the isolated gene encodes L-arabinose 1-dehydrogenase.
Identification of Catalytic Amino Acid Residues-It is known that Asp 265 and Tyr 269 in GFOR from Zymomonas mobilis (ZmGFOR) (20) and Asp 176 and Tyr 180 in dimeric DD (43,44) are important for the catalytic function. It was almost impossible to align the amino acid sequence of L-arabinose 1-dehydrogenase and ZmGFOR in the C-terminal domains because of weak sequential homology. Therefore, we carried out multiple alignments with over 500 amino acid sequences of Gfo/Idh/MocA family enzymes, revealing that Asp 168 -Asn 172 in L-arabinose 1-dehydrogenase corresponds to Asp 265 -Tyr 269 in ZmGFOR and Asp 176 -Tyr 180 in dimeric DD (Fig. 7, A and C). To obtain insight into the catalytic mechanism of L-arabinose 1-dehydrogenase, Asp 168 and Asn 172 were substituted with alanine residues by site-directed mutagenesis, as described under "Experimental Procedures," to construct D168A and N172A mutants, respectively. They were overexpressed in E. coli cells as a His 6 -tagged enzyme and purified with the same procedures for the wild-type enzyme (Fig. 6A). No activity of these mutants was found in zymogram-staining analysis (Fig. 6B). Kinetic analysis showed that the N172A mutant decreased by 4 orders of magnitude in k cat /K m values, compared with the wild-type enzyme ( Table 3). The D168A mutant showed no activity under standard assay conditions. These results suggested that both Asp 168 and Asn 172 were important for catalytic function in L-arabinose 1-dehydrogenase.
Expression of L-Arabinose 1-Dehydrogenase in A. brasiliense and Mutant Analysis-L-Arabinose 1-dehydrogenase has been shown to have dehydrogenase activities with other sugars such as D-galactose and D-xylose in vitro (Table 2). To estimate the physiological role of these dehydrogenase activities, Northern blot analysis was carried out with RNAs isolated from A. brasiliense cells grown on different carbon sources as follows: L-arabinose, D-galactose, and D-xylose (active substrates in vitro) and D-glucose (inactive substrate in vitro) (Fig. 8A). The significant expression of the L-arabinose 1-dehydrogenase gene was only found in cells grown on L-arabinose as a sole carbon source. Changes in L-arabinose 1-dehydrogenase activity in the cell-free extracts prepared from cells grown on different carbon sources (Fig. 8, B and C) were analogous to those observed at the level of transcription.

TABLE 3 Kinetic parameters of recombinant L-arabinose 1-dehydrogenase
Stocked L-arabinose 1-dehydrogenase was dialyzed against 100 mM Tris-HCl (pH 9.0) containing 2 mM MgCl 2 , overnight at 4°C. Values are the means Ϯ S.D., n ϭ 3.  in a minimal medium supplemented with L-arabinose as a sole carbon source, no substantial growth was found. On the other hand, no difference in the growth rate on D-glucose, D-galactose, and D-xylose was found between the wild-type and ⌬ARA5034 (Fig. 9B). Northern blot and disruptant analysis revealed that enzymatic activity with L-arabinose in vitro has a physiological meaning, but activity with D-galactose does not, and that the L-arabinose 1-dehydrogenase gene is clearly involved in L-arabinose metabolism of A. brasiliense.

DISCUSSION
In this study, we reported a novel L-arabinose 1-dehydrogenase involved in an alternative pathway of L-arabinose metabolism. The purified enzyme showed significant NADP ϩ -preferring activity not only with L-arabinose but also D-galactose. The gene expression was only induced by L-arabinose, suggesting that the physiological role is limited in L-arabinose metabolism. L-Arabinose 1-dehydrogenase was a potential member of the Gfo/Idh/MocA protein family and contained some unique catalytic amino acid residues.
Catalytic Insight into L-Arabinose 1-Dehydrogenase-Generally, subunit structures of the Gfo/Idh/MocA family are comprised of two domains, the N-terminal and C-terminal domains. The N-terminal domain (ϳ120 amino acid residues) has a characteristic GXGXX(G/A) fingerprint motif (45) in the classical ␤␣␤ dinucleotide binding pocket (46) and a recently postulated fingerprint motif for a novel class of dehydrogenases with a consensus AGKHVXCEKP (where X is any amino acid) (19). These motifs are also conserved in L-arabinose 1-dehydrogenase with little deviation (Fig. 7A, motifs I and II). The C-terminal domain of Gfo/Idh/MocA family enzymes (ϳ220 amino acid residues) shows significant sequential divergence and contains some essential amino acid residues for enzyme catalysis and/or substrate binding (Fig. 7A, open and closed circles, respectively), probably because of inherent substrate specificity in each enzyme. The structural features and catalytic mechanisms of ZmGFOR have been extensively studied by enzymatic and crystallographic analysis (18 -20). The enzyme contains tightly bound coenzyme NADP ϩ , and the strict NADP ϩ specificity is rationalized by Ser 116 , Lys 121 , and Tyr 139 (Fig. 7A, shaded circle) (47). In particular, the serine residue is important to distinguish between NAD ϩ and NADP ϩ ; the S116D mutant of ZmGFOR shows significant activity with NAD ϩ , and NAD ϩ -dependent myo-inositol 2-dehydrogenase pos-sesses aspartate residue at the corresponding position (19). On the other hand, NAD ϩ -preferring D-galactose 1-dehydrogenase from P. fluorescens (22,41) possesses threonine residue but not aspartate at the corresponding position to Ser 116 . No L-arabinose/D-galactose 1-dehydrogenase has amino acid residues equivalent to Lys 121 and Tyr 139 . These comparisons indicate that there are several modifications for determining the coenzyme specificity between L-arabinose/D-galactose 1-dehydrogenases and other Gfo/Idh/MocA family enzymes. Furthermore, compared with ZmGFOR, almost all putative amino acid residues around the substrate binding regions are substituted for those with functionally different characteristics in L-arabinose 1-dehydrogenase (Fig. 7A, up arrow), clearly creating a different environment at the active center.
A Novel Catalyst Type in L-Arabinose 1-Dehydrogenase-Glucose-6phosphate dehydrogenase (G6PDH) is a structurally homologous enzyme with GFOR and has a very strong functional relationship with GFOR. In Leuconostoc mesenteroides G6PDH (LmG6PDH) (48,49), the catalytic base (His 240 ) removes a proton from the C-1 OH group of Glc-6-P, and NAD(P) ϩ abstracts a hydride ion (H Ϫ ) from the C-1 atom. Asp 235 forms a hydrogen bond with C-2 OH of Glc-6-P. In ZmGFOR, the structurally homologous Asp 265 -Tyr 269 residues play equivalent roles (Fig. 7, A and C) (20). In enzymes of the Gfo/Idh/MocA family, the pair of Asp-(Tyr/His) is conserved completely (Fig. 7C). Multiple alignments revealed that Asp 168 -Asn 172 in L-arabinose 1-dehydrogenase corresponds to Asp 265 -Tyr 269 in ZmGFOR and Asp 235 -His 240 in LmG6PDH (Fig. 7, A and C). It is very interesting to identify the involvement of Asn 172 in the catalytic function because the LmG6PDH H240N mutant shows a decreased k cat value by 4 orders of magnitude (48) and a 10-fold higher K m value for Glc-6-P. Therefore, we constructed two mutants of L-arabinose 1-dehydrogenase, D168A and N172A, by sitedirected mutagenesis. When Asp 168 was replaced with Ala, a D168A mutant of L-arabinose 1-dehydrogenase lacked sufficient activity for detection in our standard assay conditions and zymogram staining analysis (Table 3 and Fig. 6C). On the other hand, the N172A mutant retained some enzymatic activity (less than 1% of k cat /K m of the wildtype enzyme) (Table 3), whereas the activity was too weak to detect in zymogram staining analysis (Fig. 6C). These results suggest that Asn 172 has a very important role in catalysis but is not absolutely necessary. This site-directed mutagenesis indicates that Asp 168 and Asn 172 are surely involved in the catalytic function of L-arabinose 1-dehydrogenase, raising doubt as to whether Asn 172 is a catalytic base. In the crystal structure of wild-type LmG6PDH, the distance between Asp 235 and C-1 OH of Glc-6-P is 4.8 Å, which is only slightly distant for a hydrogen bond (49). Considering that Asp 168 in L-arabinose 1-dehydrogenase corresponds to Asp 235 in LmG6PDH, we speculate that Asp 168 functions as a catalytic base in L-arabinose 1-dehydrogenase. This novel mechanism evolved early from an ancestral Gfo/Idh/MocA enzyme possessing a pair of Asp-(Tyr/His), such as ZmGFOR, because the asparagine residue equivalent to Asn 172 in L-arabinose 1-dehydrogenase is only found in the enzymes of subfamily I (Fig. 7, B and C).
Physiological Insight of L-Arabinose Metabolism in A. brasiliense-We could find many L-arabinose 1-dehydrogenase gene homologs on the bacterial genome sequence (Fig. 7B, Subfamily I). It is noteworthy that all members are from bacteria associating pathologically and/or parasitically with plants. In particular, the nucleotide sequences of the L-arabinose 1-dehydrogenase gene and the flanking region of A. brasiliense show high similarity (ϳ91%) to part of the genomic sequence of Burkholderia cepacia strain R18149 (Fig. 5), which was isolated from forest soil. On the other hand, the comparison of rRNA sequences revealed their far phylogenic relationship (ϳ77% of identity) (data not shown). These results suggest that wide range Fifty g of protein was applied. After electrophoresis, the gel was soaked into staining solution in the presence of 100 mM L-arabinose and 1 mM NADP ϩ . N is a purified native L-arabinose 1-dehydrogenase (1 g). C, enzyme activity in cell-free extracts. L-Arabinose 1-dehydrogenase activity was measured under standard assay conditions using NADP ϩ as a coenzyme. Values are the means Ϯ S.D., n ϭ 3.
genome transportation among plant-associated bacteria occurred at a very early evolutional stage.
The analysis of substrate specificity revealed that the purified enzyme functions not only as L-arabinose 1-dehydrogenase but also D-galactose 1-dehydrogenase. However, gene induction was only observed in A.
brasiliense cells grown on L-arabinose (Fig. 8), and ⌬ARA5034, a disruptant of the L-arabinose 1-dehydrogenase gene, showed the same growth rate in D-galactose as the wild-type (Fig. 9B). A deduced ORF 2 protein, located downstream of the L-arabinose 1-dehydrogenase gene (Fig. 5), has a high degree of sequential similarity to bacterial periplas- FIGURE 9. A, schematic diagram of the plasmid construction for disruption of the L-arabinose 1-dehydrogenase gene. Gray-colored region indicates the region of the L-arabinose 1-dehydrogenase gene. B, E, and M indicate BamHI, EcoRI, and MfeI restriction enzyme sites, respectively. Km r , Tc r , Amp r , and Cm r are kanamycin, tetracycline, ampicillin, and chloramphenicol resistance cassettes, respectively. B, growth of wild-type and ⌬ARA5034 strain of A. brasiliense on D-glucose (E), L-arabinose (Ⅺ), D-galactose (⌬), and D-xylose (224) as a sole carbon source. Each sugar was supplemented at the concentration of 37 mM in a minimal medium. For ⌬ARA5034, 25 g of kanamycin/liter was added. Cell growth was monitored by measuring absorbance at 600 nm.