Identification and Characterization of d-Hydroxyproline Dehydrogenase and Δ1-Pyrroline-4-hydroxy-2-carboxylate Deaminase Involved in Novel l-Hydroxyproline Metabolism of Bacteria

Background: The bacterial pathway of l-hydroxyproline metabolism has not been identified. Results: Different types of d-hydroxyproline dehydrogenases and unique Δ1-pyrroline-4-hydroxy-2-carboxylate deaminase involved in the bacterial l-hydroxyproline pathway were identified and characterized for the first time. Conclusion: l-Hydroxyproline degradation by bacteria was elucidated at the molecular level. Significance: Our results suggest that d-hydroxyproline dehydrogenases evolved convergently, and we discovered a unique deaminase enzyme likely within the aldolase protein family. l-Hydroxyproline (4-hydroxyproline) mainly exists in collagen, and most bacteria cannot metabolize this hydroxyamino acid. Pseudomonas putida and Pseudomonas aeruginosa convert l-hydroxyproline to α-ketoglutarate via four hypothetical enzymatic steps different from known mammalian pathways, but the molecular background is rather unclear. Here, we identified and characterized for the first time two novel enzymes, d-hydroxyproline dehydrogenase and Δ1-pyrroline-4-hydroxy-2-carboxylate (Pyr4H2C) deaminase, involved in this hypothetical pathway. These genes were clustered together with genes encoding other catalytic enzymes on the bacterial genomes. d-Hydroxyproline dehydrogenases from P. putida and P. aeruginosa were completely different from known bacterial proline dehydrogenases and showed similar high specificity for substrate (d-hydroxyproline) and some artificial electron acceptor(s). On the other hand, the former is a homomeric enzyme only containing FAD as a prosthetic group, whereas the latter is a novel heterododecameric structure consisting of three different subunits (α4β4γ4), and two FADs, FMN, and [2Fe-2S] iron-sulfur cluster were contained in αβγ of the heterotrimeric unit. These results suggested that the l-hydroxyproline pathway clearly evolved convergently in P. putida and P. aeruginosa. Pyr4H2C deaminase is a unique member of the dihydrodipicolinate synthase/N-acetylneuraminate lyase protein family, and its activity was competitively inhibited by pyruvate, a common substrate for other dihydrodipicolinate synthase/N-acetylneuraminate lyase proteins. Furthermore, disruption of Pyr4H2C deaminase genes led to loss of growth on l-hydroxyproline (as well as d-hydroxyproline) but not l- and d-proline, indicating that this pathway is related only to l-hydroxyproline degradation, which is not linked to proline metabolism.

Although genetic and enzymatic characterizations of Lhydroxyproline 2-epimerases from several bacteria, including P. aeruginosa, were recently reported (4), only the genes encoding D-HypDH and Pyr4H2C deaminase have been identified so far. In particular, FAD-dependent amino-acid dehydrogenases, including D-HypDHs from P. aeruginosa (13) and P. striata (15), generally associate with cell and/or organelle membranes and are unstable in solution, which has made purification, preservation, and characterization very difficult. In this study, we identified different types of putative gene clusters related to L-hydroxyproline metabolism of P. putida and P. aeruginosa by bioinformatics analysis. Furthermore, characterization of the recombinant proteins expressed in P. putida or Escherichia coli cells revealed that D-HypDHs from P. putida and P. aeruginosa (referred to as Pp/D-HypDH and Pa/D-HypDH, respectively) had clearly evolved convergently and that Pyr4H2C deaminase belongs to a unique member of the protein family mainly consisting of aldolase proteins.

General Procedures
Basic recombinant DNA techniques were performed as described by Sambrook et al. (23). Bacterial genomic DNA was prepared using a DNeasy Tissue kit (Qiagen). PCR was carried out using GeneAmp PCR System 2700 (Applied Biosystems) for 30 cycles in a 50-l reaction mixture containing 1 unit of KOD FX DNA polymerase (Toyobo), 15 pmol of the appropriate primers, and template DNA under the following conditions: denaturation at 98°C for 10 s, annealing at 50°C for 30 s, and extension at 68°C for time periods calculated as an extension rate of 1 kbp/min. Protein concentrations were determined by the method of Lowry et al. (24) with bovine serum albumin as the standard. SDS-PAGE was performed as described by Laemmli (25). Non-denaturing 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. High pressure liquid chromatography (HPLC) was performed using an Agilent 1120 Compact LC system (Tosoh). NMR spectra were recorded on a Jeol JNM-EC400 NMR spectrometer (Jeol Ltd., Tokyo, Japan) operating at 400 MHz. 2,2-Dimethyl-2-silapentane-5-sulfonate was used as an internal standard.

Bacterial Strain, Culture Conditions, and Preparation of Cell-free Extracts
Pseudomonas strains were cultured aerobically with vigorous shaking at 30°C in LB medium (10 g of tryptone, 5 g of yeast extract, and 5 g of NaCl/liter) or minimal medium (pH 6.8) containing 0.8 g of KH 2 PO 4 /K 2 HPO 4 , 2 g of (NH 4 ) 2 SO 4 , 0.002 g of FeSO 4 , 0.080 g of MgSO 4 ⅐7H 2 O, and 2 g of carbon source/ liter. The grown cells were harvested by centrifugation at 30,000 ϫ g for 20 min, suspended in 50 mM Tris-HCl (pH 8.0) containing 2 mM MgCl 2 , and disrupted by sonication for 20 min at appropriate intervals on ice using an Ultra Sonic Disruptor Model UR-200P (Tomy Seiko Co., Ltd, Tokyo, Japan) and then centrifuged at 108,000 ϫ g for 20 min at 4°C to obtain cell-free extracts. If necessary, 1% (v/v) Tween 20 was added to the buffer for solubilization of D-HypDH.

Plasmid Construction for Expression of Recombinant Proteins
Primer sequences used in this study are shown in supplemental Table S1. In this report, the prefixes P. putida (Pp), P. aeruginosa (Pa), and A. brasilense (Ab) have been added to gene symbols or protein designations when required for clarity. PpLhpB (PP_1255) PaLhpB-LhpF-LhpE (PA1266 and PA1267), PpLhpC (PP_1257), and PpLhpG (PP_1256) (GenBank TM accession numbers are given in parentheses) genes were amplified by PCR using primers containing appropriate restriction enzyme sites at the 5Ј-and 3Ј-ends and genomic DNA of P. putida KT2442 or P. aeruginosa PAO1 as a template. In the case of the PaLh-pBFE gene, SacI and HindIII sites were attached to the 5Ј-end of the PaLhpB gene and the 3Ј-end of the PaLhpE gene, respectively. Each amplified DNA fragment was introduced into BamHI-HindIII sites (for all others except the PaLhpBFE gene) or SacI-HindIII (for the PaLhpBFE gene) in pQE-80L (Qiagen), a plasmid vector for conferring an N-terminal His 6 tag on expressed proteins, to obtain pQE/PpLhpB, pQE/PaLhpBFE, pQE/PpLhpC, and pQE/PpLhpG, respectively. pUCP26Km (26) was previously constructed by inserting the Tn5-derived 1.3-kbp BamHI kanamycin resistance (Km r ) cassette of pUC4K (Amersham Biosciences) into the BamHI site in pUCP26, an Escherichia-Pseudomonas shuttle plasmid (27). The promoter and partial open reading frame region (ϳ240 bp) of the ahpC gene from P. putida that encodes a small subunit of alkyl hydroperoxide reductase was amplified and introduced into the SalI site in pUCP26Km to yield pUCP26KmAhpC p . Each DNA fragment of His 6 -PpLhpB-t 0 terminator and His 6 -PaLhpBFE-t 0 terminator was amplified by PCR using pQE/ PpLhpB and pQE/PaLhpBFE as a template, respectively, and introduced into SalI-EcoRI sites in pUCP26KmAhpC p to obtain pUCP/PpLhpB and pUCP/PaLhpBFE, respectively (see Fig. 2F).

Expression and Purification of His 6 -tagged Recombinant Proteins
PpLhpC and PpLhpG genes were expressed in E. coli cells, whereas PpLhpB and PaLhpBFE genes were expressed in P. putida cells. E. coli DH5␣ harboring pQE/PpLhpC or pQE/ PpLhpG was grown at 37°C to a turbidity of 0.6 at 600 nm in Super broth medium (pH 7.0; 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) containing 50 mg/liter ampicillin. After the addition of 1 mM isopropyl ␤-D-thiogalactopyranoside, the culture was further grown for 6 h to induce the expression of His 6tagged protein. On the other hand, P. putida KT2442-oxyR1 (28) harboring pUCP/PpLhpB or pUCP/PaLhpBFE obtained using the the heat-shock transformation method was grown at 30°C overnight in LB medium containing 50 mg/liter rifampicin and 50 mg/liter kanamycin. Cells were harvested and resuspended in Buffer A (50 mM sodium phosphate buffer (pH 8.0) containing 2 mM MgCl 2 , 300 mM NaCl, 1% (v/v) Tween 20, and 10 mM imidazole). The cells were then disrupted by sonication, and the solution was centrifuged. The supernatant was loaded onto a nickel-nitrilotriacetic acid Superflow column (Qiagen), which was equilibrated with Buffer A, linked to the BioAssist eZ system (Tosoh). The column was washed with Buffer B (50 mM sodium phosphate buffer (pH 8.0) containing 2 mM MgCl 2 , 300 mM NaCl, 0.1% (v/v) Tween 20, 10% (v/v) glycerol, and 50 mM imidazole). The enzymes were then eluted with Buffer C (pH 8.0; Buffer B containing 250 mM imidazole instead of 50 mM imidazole). For purification of PpLhpC and PpLhpG, a buffer system without Tween 20 was used. PpLhpB and PpLhpC were stored at 4°C and used within 1 day in further experiments, whereas PaLhpBFE and PpLhpG were concentrated by ultrafiltration with Centriplus YM-30 (Millipore), dialyzed against 50 mM Tris-HCl buffer (pH 9.0) containing 2 mM MgCl 2 and 50% (v/v) glycerol, and stored at Ϫ35°C until use.
The native molecular mass of recombinant proteins was estimated by gel filtration, which was carried out using an HPLC system at a flow rate of 1 ml/min. The purified enzyme was loaded onto a TSKgel G3000SWXL column (Tosoh) equilibrated with 50 mM Tris-HCl buffer (pH 8.0). A high molecular weight gel filtration calibration kit (GE Healthcare) was used as molecular markers.
Pyr4H2C Deaminase Assay-Pyr4H2C deaminase was assayed spectrophotometrically in the coupling system with KGSADH. The recombinant AbLhpG protein (KGSADH-III) was prepared by the method described previously (21). The reaction mixture (1 ml) consisted of 50 mM Tris-HCl buffer (pH 8.0), 10 mM Pyr4H2C, 1.5 mM NADP ϩ , and 1 unit of purified AbLhpG. Pyr4H2C was enzymatically synthesized from D-hydroxyproline by Pa/D-HypDH as described below. No change in absorbance at 340 nm was found at this stage. The reaction was started by the addition of a small amount of protein, and the increasing absorption caused by NADPH produced in the reaction was measured at 340 nm.
L-Hydroxyproline 2-Epimerase Assay-The reaction mixture consisted of 50 mM Tris-HCl (pH 8.0), 0.25 mM INT, 0.06 mM PMS, and 10 mM L-hydroxyproline. After the addition of a small amount of cell-free extract from P. putida or P. aeruginosa, the mixture was incubated for 5 min at 30°C. No change in absorbance at 490 nm was found at this stage. The reaction was started by the addition of 1 unit of Pa/D-HypDH, and the increasing absorption was measured at 490 nm. L-Hydroxyproline 2-epimerase in cell-free extract converts L-hydroxyproline to D-hydroxyproline, which is detectable by the D-HypDH assay system.

Zymogram Staining Analysis
Cell-free extract or purified D-HypDH was separated by nondenaturing PAGE with a 10% (w/v) gel at 4°C. The gels were then soaked in 10 ml of staining solution consisting of 50 mM Tris-HCl (pH 9.0), 2 mM MgCl 2 , 0.25 mM INT, 0.06 mM PMS, and 10 mM D-hydroxyproline at room temperature for 15 min in the dark. Dehydrogenase activity appeared as a red band.

Internal Peptide Analysis of Recombinant Pa/D-HypDH
As described under "Results," when the PaLhpBFE gene cluster was expressed in P. putida cells, the purified enzyme consisted of three bands with different molecular masses in SDS-PAGE analysis (see Fig. 2G, lane 4). To identify these proteins with molecular masses of 39, 8, and 44 kDa as PaLhpB, PaLhpF and PaLhpE, respectively, each protein band on the 12% (w/v) SDS-PAGE gel was excised, and the gel pieces were incubated with 10 mM dithiothreitol at 56°C for 45 min. Then protein in the gel was incubated with 55 mM iodoacetamide at room temperature for 45 min and subjected to in-gel digestion with trypsin (Promega). The digests were applied to a liquid chromatography (LC)-electrospray ionization-tandem mass spectrometer (LCQ Fleet TM LC-ESI-MSn system, ThermoFisher Scientific) in the positive ion mode. The data were collected in the data-dependent tandem mass spectrometry (MS/MS) scan mode and analyzed using SEQUEST software (ThermoFisher Scientific).

Determination of Flavin Cofactor in D-HypDHs
Purified Pp/D-HypDH and Pa/D-HypDH were dialyzed against H 2 O, and cofactors were released by heat denaturation. After removal of the precipitate formed by centrifugation, the supernatant was used to identify the flavin cofactor(s) by HPLC. HPLC analysis was carried out using a TSKgel ODS-80Tm column (4.6 ϫ 150 mm; Tosoh). An isocratic elution (10 min) with 10 mM potassium phosphate (pH 6.0) followed by a linear gradient (30 min) between 0 and 70% methanol in the same solution was used for the elution. The flow rate was 1.0 ml/min, and Novel Bacterial L-Hydroxyproline Pathway SEPTEMBER 21, 2012 • VOLUME 287 • NUMBER 39 JOURNAL OF BIOLOGICAL CHEMISTRY 32677 elution was monitored by the absorbance at 260 nm. Non-heme iron contents in Pa/D-HypDH were estimated by the method of Fish (29).

Identification of Reaction Product of D-HypDHs
The reaction mixture (200 ml) consisted of 50 mM Tris-HCl buffer (pH 9.0), 2 mM MgCl 2 , 50 mM D-hydroxyproline, and 0.05 mM PMS. After the addition of 50 mg of purified Pp/D-HypDH or Pa/D-HypDH, the mixture was left at 30°C in the dark overnight. Remaining PMS was removed by adding charcoal and filtered (referred to as Solution X). Solution X was applied to a Dowex 1X2 Cl Ϫ form (100 -200 mesh) resin column followed by a 0-1 M gradient of HCl. To identify a putative ⌬ 1 -pyrroline product, 125 l of each fraction was added to 500 l of 5% (w/v) p-dimethylaminobenzaldehyde in n-propanol followed by 1.25 ml of 3 N H 2 SO 4 . The mixtures were shaken thoroughly and incubated at 70°C for 5 min, and the absorbance was measured at 550 nm. Active fractions were combined, adjusted to ϳpH 11 with NaOH, and lyophilized. The 1 H NMR spectrum of D-hydroxyproline (400 MHz, D 2 O) was as follows: 4.41 (1H, m), 4.04 (1H, dd, J 1 ϭ 10 Hz, J 2 ϭ 4 Hz), 3.30 (1H, dt, J 1 ϭ 12 Hz, J 2 ϭ 2 Hz), 3.20 (1H, dd, J 1 ϭ 12 Hz, J 2 ϭ 4 Hz), 2.34 (1H, m), and 2.09 (1H, m). As described under "Results," Pyr4H2C was not purified by this procedure. However, when Solution X was used as the substrate of D-HypDH, no significant activity was found, indicating the complete conversion of D-hydroxyproline to the reaction product. Therefore, this sample was used as 50 mM Pyr4H2C solution for measuring the activity of PpLhpC.

Target Disruption of L-Hydroxyproline Gene Cluster
The Km r cassette of pUC4K was inserted into the single SalI site in the coding sequence of the PpLhpC gene of pQE/PpLhpC to yield pQE/PpLhpC::Km. To introduce the restriction site for BamHI at the 5Ј-and 3Ј-ends of the DNA fragment containing the Km r gene in the PpLhpC gene, PCR was carried out using pQE/PpLhpC::Km as a template and two primers, P7 and P8 (supplemental Table S1). The 2.3-kbp BamHI DNA fragment was subcloned into the BamHI site of the suicide vector pKNG101 (sacB Sm r ) (30) to yield pKNG/PpLhpC::Km (see Fig.  6C). This plasmid was introduced into parent strain P. putida KT2442, and the resulting transformants were selected for integration of the Km r marker in the chromosome on an LB agar plate containing 50 mg/liter kanamycin. The mutant was analyzed by colony PCR with P15 and P16 primers designed to target the flanking region of the PpLhpC gene to confirm the presence of its disrupted copy of the PpLhpC gene.

Amino Acid Sequence Alignment and Phylogenetic Analysis
Protein sequences were analyzed using Protein BLAST and the ClustalW program distributed by DDBJ (DNA Data Bank of Japan). The phylogenetic tree was produced using the Tree-View 1.6.1. program.

RESULTS
Gene Cluster Related to Bacterial L-Hydroxyproline Metabolism-As described in the Introduction, we previously identified L-hydroxyproline-inducible KGSADH-III (referred to as LhpG in this report) from A. brasilense (18) (Fig. 2A). Protein BLAST analysis revealed that the homologous gene exists as a gene cluster on the genomes of many bacteria, including P. putida and P. aeruginosa, with the ability to grow on L-hydroxyproline. The gene cluster of P. putida consisted of five hypothetical genes, LhpB (FAD-dependent oxidoreductase), LhpG (KGSADH), LhpC (dihydrodipicolinate synthase), LhpA (epimerase), and LhpP (amino-acid permease) among which LhpB was a likely candidate for D-HypDH ( Fig. 2A). Surprisingly, the gene cluster of P. aeruginosa contained several additional genes, LhpB (glycine/D-amino-acid oxidase), LhpE (heterotetrameric sarcosine oxidase ␣-subunit), LhpF (ferredoxin), and LhpD (malate/lactate dehydrogenase), in the order of LhpB-F-E-D (PaLhpF was not identified in the database, but we found an AbLhpF gene homolog between PaLhpB and PaLhpE genes). The 3Ј-part of PaLhpB and PaLhpF was slightly overlapped by the 5Ј-part of PaLhpF and PaLhpE, respectively (Fig. 2F, inset). A similar tendency was also found in heteromeric Tp/L-PDH (8) (Fig. 2B). This suggested the possibility that D-HypDHs of P. putida and P. aeruginosa are largely different and that PaLhpB-F-E encode each subunit in the heteromeric structure.
Expression of Recombinant Proteins in P. putida Cells-Although we first attempted to express PpLhpB, PaLhpB, PaLhpF, and PaLhpE proteins in E. coli cells, all mainly appeared in the fraction of the insoluble form despite efforts involving a survey of E. coli strains and vectors (data not shown). No improvement was found in the co-expression of PaLhpB-F-E. We previously revealed that P. putida KT2442-oxyR1 strain constitutively produces a soluble AhpC protein, a small subunit of alkyl hydroperoxide reductase, with a molecular mass of 24 kDa (Fig.  2G, lanes 1 and 2) (28). Therefore, we expressed His 6 -tagged PpLhpB protein under regulation of the ahpC promoter inserted into the Escherichia-Pseudomonas shuttle plasmid in P. putida KT2442-oxyR1 cells (Fig. 2F) and purified it to homogeneity in a single step using immobilized metal (Ni 2ϩ ) affinity chromatography. SDS-PAGE gave a single band with a molecular mass of ϳ47 kDa, which corresponded to the theoretical value of PpLhpB (46,856.49 Da) (Fig. 2G, lane 3). On the other hand, it was impossible to determine the native molecular mass because of the elution in the void volume of gel filtration column ( Fig. 2H and supplemental Fig. S1). The purified enzyme was yellow, which is typical of flavoproteins, and showed 6.41 units/mg of protein specific activity toward D-hydroxyproline using standard assay conditions in the artificial electron acceptor system (PMS-INT; Table 1), indicating the function of D-hydroxyproline dehydrogenase (Pp/D-HypDH).
PaLhpB-F-E operon in which the His 6 tag sequence was attached at the N terminus of the LhpB gene was also expressed in ahpC promoter in P. putida KT2442-oxyR1 cells (Fig. 2F), and the recombinant protein was successfully purified to homogeneity. Native PAGE gave a single band (see Fig. 3B, inset), and a symmetric peak appeared in gel filtration analysis (supplemental Fig. S1). On SDS-PAGE analysis, two major distinct bands (referred to as ␣and ␤-subunits, respectively) along with one minor band (referred to as ␥-subunit) were observed (Fig. 2G, lane 4). The estimated molecular masses of ␣-, ␤-, and ␥-subunits (48, 42, and 10 kDa, respectively) were similar to the theoretical values of PaLhpE, PaLhpB, and PaLhpF (44,055.06, 41,133.39, and 8,184.55 Da, respectively).
Furthermore, LC-electrospray ionization-MS/MS profiles using tryptic peptides of each protein band revealed that ␣-, ␤-, and ␥-subunits corresponded to PaLhpE, PaLhpB, and PaLhpF, respectively (supplemental Fig. S2   matography suggested that this protein consists of three different subunits. The molar ratio of ␣:␤:␥ was ϳ1:1:1 as judged by image scanning of the SDS-PAGE peak areas using NIH ImageJ Version 1.45s software. The native molecular mass estimated by gel filtration using HPLC was ϳ360 kDa (Fig. 2H), indicating that this protein molecule may be composed of a heterododecameric structure of subunits, i.e. ␣ 4 ␤ 4 ␥ 4 . Under the same assay conditions as PpLhpB, specific activity toward D-hydroxyproline was 21.1 units/mg of protein (Table 1). When compared with PpLhpB, the purified enzyme from P. aeruginosa was orange-brown rather than light yellow, indicating that some chromophores must be bound to protein (see below). These results suggested that PaLhpBFE is a different type of D-HypDH (Pa/D-HypDH). Successful purification of both Pp/D-HypDH and Pa/D-HypDH from P. putida cells was achieved using a buffer system containing Tween 20 (see "Materials and Methods"). Furthermore, when P. putida and P. aeruginosa cells grown on L-hydroxyproline were disrupted using buffer in the absence of Tween 20, no significant D-HypDH activity was found in cellfree extracts (data not shown). These results indicated that both native and recombinant proteins of Pp/D-HypDH and Pa/D-HypDH might be tightly bound to the cytoplasmic membrane as described in preliminary reports (13,15).
Substrate and Electron Acceptor Specificity of D-HypDHs-In addition to D-hydroxyproline, L-hydroxyproline, L-proline, and D-proline were tested as substrates for Pp/D-HypDH and Pa/D-HypDH. Among them, significant activity was observed only with D-proline (see Fig. 3, A and B, insets), and the kinetic parameters of D-hydroxyproline and D-proline determined from the Lineweaver-Burk plot (Fig. 3, A and B) are shown in Table 1 Among the characterized amino-acid dehydrogenases, only D-PDH from a hyperthermophilic archaeon, Pyrobaculum islandicum (31), showed significant activity with D-hydroxyproline and close K m values between D-proline and D-hydroxyproline (4.2 and 9.5 mM, respectively) (supplemental Table S6). This enzyme has broad substrate specificity with many Damino acids in addition to D-hydroxyproline. Furthermore, 2,6-dichloroindophenol and/or ferricyanide is the most preferred electron acceptor(s) for known DL-PDHs, including P. islandicum D-PDH. To our knowledge, this is the first report of amino-acid dehydrogenase(s) with high specificity for D-hydroxyproline.

Effect of pH on Activity of D-HypDHs-pH activity profiles of
Pp/D-HypDH and Pa/D-HypDH were similar: their maximal activities were observed at the range of pH 9 -9.75 (supplemental Fig. S2). Furthermore, the enzymes maintained over 70% of their maximal activity after 2 h of incubation at pH 8 -10.5 (data not shown). Apparent pK a values determined from the acidic limb of the profiles were 8.5-8.9. It is likely that these properties are similar to eukaryotic D-amino-acid oxidases (32) rather than archaeal DL-PDHs (supplemental Fig. S3).
Prosthetic Groups in D-HypDHs-As described above, both Pp/D-HypDH and Pa/D-HypDH were yellow (or orangebrown), and the addition of FAD or FMN in the standard assay mixture had no effect on the activity (data not shown). The UV-visible absorption spectrum of Pp/D-HypDH exhibited two pronounced absorption peaks at 354 -360 and 454 -456 nm, indicating typical flavoproteins (Fig. 3C). On the other hand, that of Pa/D-HypDH revealed absorption maxima at 353, 409, and 449 -453 nm. Because the presence of [2Fe-2S] iron-sulfur cluster motifs was predictable on ␣and ␥-subunits on the primary structure (see below), this spectrum is likely consistent with the sum of the spectra of typical flavoproteins (maxima at about 370 and 450 nm) and an oxidized [2Fe-2S] center (maxima at about 320, 415, and 455 nm): the latter was particularly similar to that of [2Fe-2S] ferredoxin from E. coli (cf. Fig. 5 in Ref. 33).
We attempted to identify the prosthetic cofactors in Pp/D-HypDH and Pa/D-HypDH using HPLC (Fig. 3D). Authentic FAD, FMN, ATP, and ADP were used as standards. As a result,  Pp/D-HypDH as well as most of the other amino-acid dehydrogenases contained only FAD. Although the molar value per mole of enzyme was unclear because the native molecular mass was not determined by gel filtration (supplemental Fig. S1), ϳ0.02 mol of FAD was released from 1 mg of the purified enzyme. Because one FAD-binding motif was predictable on the primary structure (see below), the quaternary structure may be a homodimer. Interestingly, both FAD and FMN were detected within Pa/D-HypDH. Moreover, we determined that there were ϳ8 mol of FAD and ϳ4 mol of FMN/mol of enzyme.

Identification of Reaction Product by D-HypDH-
The reaction products of D-hydroxyproline dehydrogenation catalyzed by Pp/D-HypDH and Pa/D-HypDH were purified using anion exchange chromatography and identified by NMR. Although the reaction product was postulated to be Pyr4H2C or Pyr3H5C, the NMR spectra were very similar to those of pyrrole-2-carboxylate (Fig. 3E). Furthermore, when the reaction product and authentic pyrrole-2-carboxylate were co-solved in Asterisks indicate peaks derived from an internal standard. In this analysis, pyrrole-2-carboxylate was identified as the reaction product (see in text). Similar results were obtained when Pp/D-HypDH was used as an alternative (data not shown). mAU, milliabsorbance units. D 2 O, the NMR spectra overlapped completely (data not shown). It is known that in rats L-hydroxyproline is converted to Pyr4H2C by L-amino-acid oxidase (dehydrogenase) followed by pyrrole-2-carboxylate upon acidification of the reaction mixture via the unstable oxidation intermediate, probably Pyr4H2C (34). In this study, we observed that a pyrroline-related compound detected using p-dimethylaminobenzaldehyde is eluted from a column resin in the range of 0.7-1 M HCl (see "Materials and Methods"). Furthermore, when the reaction product (under alkaline conditions) was used directly for the PpLhpC-PpLhpG coupling assay, the final reaction product was identified as ␣-ketoglutarate (see below). These results suggested that the reaction product from D-hydroxyproline by Pp/D-HypDH and Pa/D-HypDH is Pyr4H2C, not Pyr3H5C. Although only P. islandicum D-PDH shows a similar conversion from D-proline to ⌬ 1 -pyrroline-2-carboxylate (but not ⌬ 1 -pyrroline-5-carboxylate) (31), its metabolic fate is unclear.
Amino Acid Sequence Analysis of D-HypDHs-Pp/D-HypDH showed only ϳ20% sequence similarity to ␤-subunit of Pa/D-HypDH (Pa/D-HypDH␤) and other homomeric D-amino-acid dehydrogenases (31,35) (Fig. 4A). On the other hand, the N-terminal 30 amino acid residues among these proteins had somewhat higher homology (ϳ30%) than other parts of the proteins. This region contains a highly conserved typical ADPbinding motif with a ␤␣␤-fold (Gly-X-Gly-X 2 -Gly) (shaded in red), described by Wierenga et al. (36), confirming that Pp/D-HypDH contains only FAD as a prosthetic group (probably one FAD per subunit).
Relatively high homology with Pa/D-HypDH (30 -40%) was found in the HcnABC complex of hydrogen cyanide synthase, NoxABC complex of D-nopaline oxidase, and OoxABC complex of D-octopine oxidase from bacteria (Fig. 4, A, B, and C, PfHcn, AtNox, and AtOox, respectively; Refs. [37][38][39]. "HcnB, NoxA, and OoxA," "HcnC, NoxB, and OoxB," and "HcnA, NoxC, and OoxC" correspond to ␣-, ␤-, and ␥-subunits, respectively, of Pa/D-HypDH. Furthermore, ␣and ␤-subunits of Ph/L-PDH1 and Tp/L-PDH were also homologous with those of Pa/D-HypDH. Two ADP-binding motifs as described above were found at the N terminus of ␣and ␤-subunits of Pa/D-HypDH (shaded in orange and red, respectively), indicating that the binding sites of two FADs in the Pa/D-HypDH␣␤␥ complex may be here. In the case of Tp/L-PDH␣, another ADP-binding motif is located in the middle region (shaded in pink) that may be related to the unique NADH dehydrogenation activity (8). In fact, Pa/D-HypDH showed no NADH dehydrogenation activity (data not shown). Interestingly, the homologous motifs of Ph/L-PDH1␣ and Tp/L-PDH␣ are located at over 120 amino acid residues in the internal region from the N terminus (see next section). On the other hand, two different types of [2Fe-2S]binding motifs were found in Pa/D-HypDH␣ and Pa/D-HypDH␥ (shaded in cyan and yellow, respectively). The former consisted of Cys-X-Cys-X 15-20 -Cys-X 4 -Cys near the C terminus, whereas the latter consisted of Cys-X 4 -Cys-X 2 -Cys-X 11-12 -Cys and matched the iron-sulfur binding signature of [2Fe-2S] ferredoxins. Therefore, the either of the two may be the [2Fe-2S]-binding site in the Pa/D-HypDH␣␤␥ complex.
Functional Characterization of PpLhpC as Pyr4H2C Deaminase-The LhpC gene is commonly contained within L-hydroxyproline operons of P. putida and P. aeruginosa (sequence identity of ϳ30%) (Fig. 2A). To estimate enzyme function, PpL-hpC was overexpressed in E. coli cells by induction with isopropyl ␤-D-thiogalactopyranoside as a His 6 -tagged protein and purified to homogeneity with a nickel-chelating affinity column (Fig. 2G, lane 5). The apparent molecular mass estimated by SDS-PAGE was 35 kDa, coinciding with the theoretical value of 35,341.67 Da. The native molecular mass estimated by gel filtration was ϳ275 kDa (Fig. 2H and supplemental Fig. S1), indicating a homooctameric structure.
As described above, LhpC belongs to the dihydrodipicolinate synthase/N-acetylneuraminate lyase family (22). Because most members of this protein family produce pyruvate and several aldehydes in the aldol cleavage reaction (see Fig. 5C), potential aldolation activity with Pyr4H2C in PpLhpC was first measured using lactic dehydrogenase as a coupling enzyme but was not detected (data not shown). Therefore, we alternatively used the purified recombinant AbLhpG (KGSADH), which shows high specificity with ␣KGSA, as a coupling enzyme (21) (also see next section). Under this assay system, PpLhpC showed specific activity of 2.41 units/mg of protein in the presence of 10 mM Pyr4H2C (and 1.5 mM NADP ϩ ). Furthermore, when NADP ϩdependent L-glutamate dehydrogenase co-existed in the reaction mixture, the increase in absorbance at 340 nm was completely eliminated (data not shown) as a result of cooperative reactions by PpLhpC (Pyr4H2C 3 ␣KGSA ϩ NH 3 ), KGSADH (␣KGSA ϩ NADP ϩ 3 ␣-ketoglutarate ϩ NADPH) and L-glutamate dehydrogenase (␣-ketoglutarate ϩ NH 3 ϩ NADPH 3 L-glutamate ϩ NADP ϩ ). These results suggested that PpLhpC plays a role as a Pyr4H2C deaminase. The K m and k cat values with Pyr4H2C were estimated to be 0.693 Ϯ 0.031 mM and 71.5 Ϯ 2.1 min Ϫ1 , respectively. Pyruvate is a common substrate for dihydrodipicolinate synthase/N-acetylneuraminate lyase proteins as described above. On the other hand, L-2-keto-3deoxyarabinonate (L-KDA) has a hydroxyl group at the C5 position of Pyr4H2C (4-hydroxy-2-oxo-5-aminovalerate) instead of an amino group. Because L-KDA dehydratase also belongs to the dihydrodipicolinate synthase/N-acetylneuraminate lyase family (see "Discussion"), one possibility was that L-KDA is an alternative substrate for potential dehydration of Pyr4H2C deaminase. However, both pyruvate and L-KDA inhibited the enzyme competitively, and the K i values were 0.492 and 0.592 mM, respectively (Fig. 5A).
Functional Characterization of PpLhpG as KGSADH-In addition to PpLhpC, His 6 -tagged PpLhpG was overexpressed in E. coli cells and purified to homogeneity with a nickel-chelating affinity column (Fig. 2G, lane 6). The apparent molecular mass of His 6 -tagged PpLhpC was estimated by SDS-PAGE analysis as 55 kDa, coinciding with the theoretical value of 56,244.13 Da. The native molecular mass estimated by gel filtration was ϳ113 kDa (Fig. 2H and supplemental Fig. S1), indicating a homodimeric structure. As expected from the significant sequence similarity to AbLhpG (66% identity), the recombinant PpLhpG protein showed specific activities for ␣KGSA of 55.5 and 25.2  Table S3. E, structural formulas for D-hydroxyproline, glycine, D-nopaline, and D-octopine and schematic reactions by D-HypDH, Hcn, Nox, and Oox. The common moiety with D-hydroxyproline is shown in bold.
units/mg of protein in the presence of NADP ϩ and NAD ϩ , respectively ( Table 2). The k cat /K m with NADP ϩ (83,500 min Ϫ1 ⅐mM Ϫ1 ) was 134-fold higher than that with NAD ϩ , mainly caused by the lower K m values (0.0230 and 6.15 mM Ϫ1 , respectively; Table 3). Under the same assay conditions, AbL-hpG showed similar k cat /K m values to ␣KGSA in the presence of NAD ϩ or NADP ϩ , whereas k cat /K m with NADP ϩ was ϳ22-fold higher than with NAD ϩ (21). Various aldehydes, including ␣KGSA, were tested as substrates for dehydrogenation by PpL-hpG in the presence of NADP ϩ . ␣KGSA was clearly the best substrate, whereas activities with a number of aliphatic aldehydes from C 3 to C 8 were only 10% less than those with ␣KGSA ( Table 2). These results indicated that PpLhpG possesses NADP ϩ preference and high specificity toward ␣KGSA similar to AbLhpG.
Gene Regulation and Disruptant Analysis-When P. putida and P. aeruginosa were grown on L-hydroxyproline or D-hydroxyproline, almost all four metabolic enzymes were induced in cell-free extracts (Fig. 6A). The induction of D-HypDH was also detected in zymogram staining analysis (Fig. 6B). Although (partially) incomplete induction of KGSADH activity was found in P. aeruginosa, it has been reported that this bacterium possesses a constitutively expressed KGSADH isozyme(s) and/or other aldehyde dehydrogenase(s) with activity for ␣KGSA (14). To estimate the physiological role(s) of the Lhp gene cluster in L-hydroxyproline metabolism, we carried out gene disruption experiments by introducing a Km r gene at the center of the PpLhpC gene (encoding Pyr4H2C deaminase) (Fig. 6, C and D). The obtained LhpC Ϫ mutant strain was distinct from the wild-type strain (KT2442) in that neither L-hydroxyproline nor D-hydroxyproline as a sole carbon source supported growth (Fig. 6E). On the other hand, there was no difference in growth on D-proline (an alternative substrate for D-HypDH), glucose, and L-proline between the two strains. Indeed, only L-hydroxyproline 2-epimerase activity was significantly found in the cell-free extract prepared from LhpC Ϫ mutant cells grown on LB medium supplemented with L-hydroxyproline; D-HypDH, Pyr4H2C deaminase, and KGSADH activities were not found (data not shown). These results indicated that LhpC disruption also led to the inactivation of LhpB and LhpG genes within the Lhp operon ( Fig. 2A). Overall, these results clearly suggested that the Lhp gene cluster is involved in L-hydroxyproline (as well as D-hydroxyproline) but not D-proline metabolism.

Comparisons between Pa/D-HypDH and Pp/D-HypDH-Al-
though Pa/D-HypDH and Pp/D-HypDH show very high specificity for D-hydroxyproline (as a substrate) and PMS-INT or PMS-NBT (as an electron acceptor), the former heteromeric enzyme contains FAD, FMN, and a [2Fe-2S] iron-sulfur cluster as prosthetic groups, which is different from the latter homomeric enzyme (only FAD), strongly indicating that D-HypDHs appeared convergently in P. putida and P. aeruginosa during an evolutionary stage. Only P. islandicum D-PDH shows significant activity for D-hydroxyproline in addition to D-proline, and the manner of dehydrogenation reaction is homologous to that of D-HypDH (31). However, it is likely that their substrate specificities are also acquired independently of each other because their sequence similarities are not high.
In this study, we first revealed that Pa/D-HypDH is phylogenetically related to hydrogen cyanide synthase (Hcn), D-nopaline oxidase (Nox), and D-octopine oxidase (Oox) of bacteria (37)(38)(39). Hcn, Nox, and Oox complexes are encoded by hcnA-B-C, noxB-C-A, and ooxB-C-A gene clusters, respectively, on the bacterial genome ( Fig. 2, C, D, and E). Furthermore, each substrate of Pa/D-HypDH, Hcn, Nox, and Oox possesses structural formulas similar to the secondary amine (Fig. 4E), clearly indicating that they evolved from a common ancestor. Although it is likely that Hcn, Nox, and Oox are membraneassociated flavoproteins as are most amino-acid dehydrogenases (31,35), the detailed features have not been reported. Pa/D-HypDH is the first of this type of protein to be experimentally characterized at the molecular level.
Comparisons with DL-PDHs from Archaea-D-HypDHs, in particular Pa/D-HypDH, show high sequence similarities to several dehydrogenases for L-and D-proline from archaea  a Under standard assay conditions as described under "Materials and Methods." All aldehyde concentrations were 1 mM. b KGSADH-III from A. brasilense (21). c Values relative to ␣KGSA in the presence of NADP ϩ (100%). d Values relative to ␣KGSA in the presence of NAD ϩ (100%). respectively, correspond to the first ϳ100 and last ϳ70 amino acid residues, respectively, of Ph/L-PDH1␣ (Fig. 4A). Therefore, it is likely that that these heteromeric proteins evolved from a common ancestral flavoenzyme (Type 1), such as homomeric Pp/D-HypDH, by genetic duplication, division, and/or fusion and that they eventually acquired a new function(s). Pyr4H2C Deaminase within Aldolase Protein Family-It is likely that Pyr4H2C deaminase belongs to the dihydrodipicolinate synthase/N-acetylneuraminate lyase family based on the phylogenetic analysis. Generally, members of this protein family possess a common (␤/␣) 8 -barrel fold and share a common reaction step in their reactions, namely the formation of a "Schiff base" intermediate between a strictly conserved lysine residue and C2 carbon of the common ␣-keto acid moiety of the substrate (22) (Fig. 5B). The phenolic hydroxyl group of the tightly conserved tyrosine residue forms double hydrogen bonds to carboxyl and hydroxyl groups at position 4 of the substrate. Interestingly, 4-hydroxy-2-oxoglutarate aldolase, which is involved in mammalian L-hydroxyproline metabolism (Fig. 1A), is also a member of this protein family (4). Although lysine (Lys 164 ) and tyrosine (Tyr 136 ) align well with these residues in the sequence of PpLhpC (Fig. 5B), there is a poor phylogenetic relationship to any subfamilies of the other members (Fig. 5D), which is probably the reason for the unique enzyme catalysis (Fig. 5C). Site-directed mutagenesis and/or crystallographic analysis of Pyr4H2C deaminase would provide clearer evidence for assignment to the dihydrodipicolinate synthase/ N-acetylneuraminate lyase protein family and help to elucidate novel catalytic mechanism in this protein family.