Identification of 2-keto-3-deoxy-d-Gluconate Kinase and 2-keto-3-deoxy-d-Phosphogluconate Aldolase in an Alginate-Assimilating Bacterium, Flavobacterium sp. Strain UMI-01

Recently, we identified an alginate-assimilating gene cluster in the genome of Flavobacterium sp. strain UMI-01, a member of Bacteroidetes. Alginate lyase genes and a 4-deoxy-l-erythro-5-hexoseulose uronic acid (DEH) reductase gene in the cluster have already been characterized; however, 2-keto-3-deoxy-d-gluconate (KDG) kinase and 2-keto-3-deoxy-6-phosphogluconate (KDPG) aldolase genes, i.e., flkin and flald, still remained uncharacterized. The amino acid sequences deduced from flkin and flald showed low identities with those of corresponding enzymes of Saccharophagus degradans 2-40T, a member of Proteobacteria (Kim et al., Process Biochem., 2016). This led us to consider that the DEH-assimilating enzymes of Bacteroidetes species are somewhat deviated from those of Proteobacteria species. Thus, in the present study, we first assessed the characteristics in the primary structures of KDG kinase and KDG aldolase of the strain UMI-01, and then investigated the enzymatic properties of recombinant enzymes, recFlKin and recFlAld, expressed by an Escherichia coli expression system. Multiple-sequence alignment among KDG kinases and KDG aldolases from several Proteobacteria and Bacteroidetes species indicated that the strain UMI-01 enzymes showed considerably low sequence identities (15%–25%) with the Proteobacteria enzymes, while they showed relatively high identities (47%–68%) with the Bacteroidetes enzymes. Phylogenetic analyses for these enzymes indicated the distant relationship between the Proteobacteria enzymes and the Bacteroidetes enzymes, i.e., they formed distinct clusters in the phylogenetic tree. recFlKin and recFlAld produced with the genes flkin and flald, respectively, were confirmed to show KDG kinase and KDPG aldolase activities. Namely, recFlKin produced 1.7 mM KDPG in a reaction mixture containing 2.5 mM KDG and 2.5 mM ATP in a 90-min reaction, while recFlAld produced 1.2 mM pyruvate in the reaction mixture containing 5 mM KDPG at the equilibrium state. An in vitro alginate-metabolizing system constructed from recFlKin, recFlAld, and previously reported alginate lyases and DEH reductase of the strain UMI-01 could convert alginate to pyruvate and glyceraldehyde-3-phosphate with an efficiency of 38%.


Introduction
Alginate is an acidic heteropolysaccharide comprising two kinds of uronic acid, β-D-mannuronate and α-L-guluronate [1][2][3]. This polysaccharide exists as a structural material in cell-wall matrices of brown algae and biofilms of certain bacteria. Since alginate solution shows high viscosity study, we first characterized the primary structures of KDG kinase and KDPG aldolase, FlKin and FlAld, of the strain UMI-01 compared with those of other bacterial enzymes. Then, we investigated enzymatic properties of proteins encoded by flkin and flald using recombinant enzymes, recFlKin and recFlAld. Furthermore, we constructed an in vitro alginate-metabolizing system using recFlKin and recFlAld, along with recombinant alginate lyases and DEH reductase of this bacterium to confirm that this enzyme system can produce pyruvate and GAP from alginate in vitro.
Mar. Drugs 2017, 15, 37 3 of 17 and FlAld, of the strain UMI-01 compared with those of other bacterial enzymes. Then, we investigated enzymatic properties of proteins encoded by flkin and flald using recombinant enzymes, recFlKin and recFlAld. Furthermore, we constructed an in vitro alginate-metabolizing system using recFlKin and recFlAld, along with recombinant alginate lyases and DEH reductase of this bacterium to confirm that this enzyme system can produce pyruvate and GAP from alginate in vitro.

Production of recFlKin and recFlAld, and Their Reaction Products
Coding regions of flkin and flald were amplified by PCR with specific primers listed in Table 1, cloned into pCold vector and expressed in E. coli BL21 (DE3). The recombinant enzymes were purified by Ni-NTA affinity chromatography. Molecular masses of recFlKin and recFlAld estimated by SDS-PAGE were 39 kDa and 26 kDa, respectively ( Figure 5). These values were consistent with the calculated molecular masses of these enzymes, i.e., 39,391 Da and 25,808 Da, which include 8 × Gly + 8 × His-tag [26].

Production of recFlKin and recFlAld, and Their Reaction Products
Coding regions of flkin and flald were amplified by PCR with specific primers listed in Table 1, cloned into pCold vector and expressed in E. coli BL21 (DE3). The recombinant enzymes were purified by Ni-NTA affinity chromatography. Molecular masses of recFlKin and recFlAld estimated by SDS-PAGE were 39 kDa and 26 kDa, respectively ( Figure 5). These values were consistent with the calculated molecular masses of these enzymes, i.e., 39,391 Da and 25,808 Da, which include 8 × Gly + 8 × His-tag [26].

Primer Name
Nucleotide Sequence The recFlKin was allowed to react with KDG in the presence of ATP. TLC analysis suggested that the reaction product was KDPG ( Figure 6A). Then, the molecular mass of the reaction product was determined by matrix-assisted laser desorption ionization-time of flight mass spectrometer (MALDI-TOF) mass spectrometry ( Figure 7A,B). The 257 m/z peak was considered to be that of KDPG (MW = 258), and the 279 m/z peak was considered to be that of a sodium-salt form of KDPG. These results indicate that the reaction product of recFlKin is KDPG. Thus, we concluded that the protein encoded by flkin is KDG kinase. Here, it should be noted that the peak intensities of KDPG were considerably low. This was ascribable to the low ionization level of KDPG. Therefore, we attempted to improve the signal intensity of KDPG using other matrices, e.g., 2,5-dihydroxybenzoic acid and αcyano-4-hydroxycinnamic acid. Unfortunately, signal intensity of KDPG was not improved much. We still need to investigate the suitable conditions for the detection of KDPG.
Reaction products of recFlAld were also analyzed by TLC ( Figure 6B). recFlAld produced two kinds of reaction products with different mobility on TLC. According to their mobility, they were regarded as pyruvate and GAP. The staining intensity of pyruvate was significantly low compared with that of GAP. This difference was ascribable to the difference in the reactivity between pyruvate and GAP with 2,4-dinitrophenylhydrazine (DNP). Namely, GAP showed much higher reactivity with DNP than pyruvate. Then, the reaction products of recFlAld were subjected to MALDI-TOF mass spectrometry. The 87 m/z and 169 m/z peaks corresponding to pyruvate (MW = 88) and GAP

Primer Name
Nucleotide Sequence The recFlKin was allowed to react with KDG in the presence of ATP. TLC analysis suggested that the reaction product was KDPG ( Figure 6A). Then, the molecular mass of the reaction product was determined by matrix-assisted laser desorption ionization-time of flight mass spectrometer (MALDI-TOF) mass spectrometry ( Figure 7A,B). The 257 m/z peak was considered to be that of KDPG (MW = 258), and the 279 m/z peak was considered to be that of a sodium-salt form of KDPG. These results indicate that the reaction product of recFlKin is KDPG. Thus, we concluded that the protein encoded by flkin is KDG kinase. Here, it should be noted that the peak intensities of KDPG were considerably low. This was ascribable to the low ionization level of KDPG. Therefore, we attempted to improve the signal intensity of KDPG using other matrices, e.g., 2,5-dihydroxybenzoic acid and α-cyano-4-hydroxycinnamic acid. Unfortunately, signal intensity of KDPG was not improved much. We still need to investigate the suitable conditions for the detection of KDPG.
Reaction products of recFlAld were also analyzed by TLC ( Figure 6B). recFlAld produced two kinds of reaction products with different mobility on TLC. According to their mobility, they were regarded as pyruvate and GAP. The staining intensity of pyruvate was significantly low compared with that of GAP. This difference was ascribable to the difference in the reactivity between pyruvate and GAP with 2,4-dinitrophenylhydrazine (DNP). Namely, GAP showed much higher reactivity with DNP than pyruvate. Then, the reaction products of recFlAld were subjected to MALDI-TOF mass spectrometry. The 87 m/z and 169 m/z peaks corresponding to pyruvate (MW = 88) and GAP (MW = 170), respectively, were observed. The peak intensity of GAP was small ( Figure 7C,D). This appeared to be due to the decomposition of GAP during the mass spectrometric analysis. Thus, we may conclude that recFlAld is the KDPG aldolase that splits KDPG to pyruvate and GAP.
Mar. Drugs 2017, 15, 37 7 of 17 (MW = 170), respectively, were observed. The peak intensity of GAP was small ( Figure 7C,D). This appeared to be due to the decomposition of GAP during the mass spectrometric analysis. Thus, we may conclude that recFlAld is the KDPG aldolase that splits KDPG to pyruvate and GAP.  KDG before and after the recFlKin reaction, respectively; (C,D) KDPG before and after the recFlAld reaction, respectively. Reaction products are indicated with red letters along with molecular masses above the peaks.

Enzymatic Properties of recFlKin and recFlAld
We first investigated the kinetic parameter for recFlAld, since recFlAld was necessary for the KDG kinase assay. In the present study, the kinase activity was assayed by quantifying the pyruvate (MW = 170), respectively, were observed. The peak intensity of GAP was small ( Figure 7C,D). This appeared to be due to the decomposition of GAP during the mass spectrometric analysis. Thus, we may conclude that recFlAld is the KDPG aldolase that splits KDPG to pyruvate and GAP.  KDG before and after the recFlKin reaction, respectively; (C,D) KDPG before and after the recFlAld reaction, respectively. Reaction products are indicated with red letters along with molecular masses above the peaks.

Enzymatic Properties of recFlKin and recFlAld
We first investigated the kinetic parameter for recFlAld, since recFlAld was necessary for the KDG kinase assay. In the present study, the kinase activity was assayed by quantifying the pyruvate Figure 7. Mass spectrometry for reaction products of recFlKin and recFlAld. The reaction products prepared as in Section 4.10 were subjected to matrix-assisted laser desorption ionization-time of flight mass spectrometer (MALDI-TOF) mass spectrometry, and analyzed by negative-ion mode. (A,B) KDG before and after the recFlKin reaction, respectively; (C,D) KDPG before and after the recFlAld reaction, respectively. Reaction products are indicated with red letters along with molecular masses above the peaks.

Enzymatic Properties of recFlKin and recFlAld
We first investigated the kinetic parameter for recFlAld, since recFlAld was necessary for the KDG kinase assay. In the present study, the kinase activity was assayed by quantifying the pyruvate produced from KDPG by the action of recFlAld. KDPG-derived pyruvate was determined by the lactate dehydrogenase (LDH)-NADH system as described in Section 4.6. In the equilibrium state of recFlAld reaction, pyruvate concentration reached 1.2 mM. Since the KDPG concentration was originally 5 mM, that in the equilibrate state was regarded as 3.8 mM. From these values the equilibrium constant (K eq ) and ∆G • were calculated to be 3.8 × 10 −1 M and +0.57 kcal/mol, respectively. This indicated that the equilibrium position of KDPG-aldolase reaction is slightly shifted toward the KDPG side. Next, we determined the reaction rate of recFlAld by the LDH-NADH method. By this method, the specific activity of recFlAld was estimated to be 57 U/mg at pH 7.4 and 30 • C. Coexistence of LDH-NADH in the reaction mixture could extend the aldolase reaction longer time by decreasing pyruvate concentration in the reaction equilibrium.
Next, KDG kinase activity of recFlKin was determined by using recFlAld and LDH-NADH. recFlKin was allowed to react with KDG in the presence of ATP at 30 • C and the reaction was terminated by heating at 100 • C for 3 min at the reaction times 1, 15, and 30 min. The KDPG produced in the reaction mixture was then split to pyruvate and GAP by recFlAld, and the pyruvate was quantified by the LDH-NADH system. At reaction time 90 min, recFlKin was found to produce 1.7 mM KDPG from 2.5 mM KDG at~70% efficiency with the specific activity 0.72 U/mg. recFlKin showed an optimal temperature and pH at around 50 • C and 7.0, respectively, and was stable at 40 • C for 30 min.

Construction of In Vitro Alginate-Metabolizing System Using Recombinant Enzymes
In the present study, we identified flkin and flald in the genome of strain UMI-01 as KDG kinase and KDPG aldolase gene, respectively. Since alginate lyases and DEH reductase in this strain have already been characterized [26][27][28], here we examined if the sequential action of these alginate-degrading and -assimilating enzymes could convert alginate to pyruvate and GAP in vitro. Namely, recombinant alginate lyases (recFlAlyA, recFlAlyB, and recFlAlex) [26,27], DEH reductase (recFlRed) [28], KDG kinase (recFlKin), and KDPG aldolase (recFlAld) were allowed to react alginate in various combinations, and each reaction product was analyzed by TLC ( Figure 8) and quantified by thiobarbituric acid (TBA) and LDH-NADH methods ( Table 2). As shown in Figure 8, alginate was almost completely degraded to DEH by the simultaneous actions of recFlAlyA, recFlAlyB, and recFlAlex. The DEH was also almost completely reduced to KDG by recFlRed. Furthermore, a major part of the KDG was converted to KDPG by recFlKin, and the band of KDPG became faint by the reaction of recFlAld. This indicated the splitting of KDPG to pyruvate and GAP by the action of recFlAld. Accordingly, the sequential action of recombinant enzymes was considered to be capable of converting alginate to pyruvate and GAP in vitro. Then, the yields of intermediates in each reaction step were quantified by TBA and LDH-NADH methods ( Table 2). Concentrations of the unsaturated oligo-alginates, DEH, KDG, KDPG, and pyruvate (and GAP), were determined to be 4.2 mM, 9.8 mM, 9.8 mM, 8.1 mM, and 3.8 mM, respectively. Since the initial concentration of alginate (0.2% (w/v)) corresponds to 10 mM monosaccharide, the yields of DEH and KDG were estimated to be~100%, and the yields of KDPG and pyruvate were estimated to be~80% and~40%, respectively. These results indicated that high-value intermediates such as KDPG could be produced from alginate with fairly high efficiency by the recombinant enzymes of the strain UMI-01 in vitro.

Alginate-Metabolizing Enzymes of Flavobacterium sp. Strain UMI-01
In the present study, flkin and flald in the genome of Flavobacterium sp. strain UMI-01 were confirmed to be the enzyme genes encoding KDG kinase and KDPG aldolase. The recombinant enzymes, recFlKin and recFlAld, showed KDG kinase and KDPG aldolase activity although low sequence identities were shown to the corresponding enzymes from other bacteria and archaea (Figures 2-4). Consequently, these genes, along with previously reported alginate lyase and DEH reductase genes were confirmed to be the genes responsible for alginate metabolism of this bacterium. The alginate-metabolizing pathway of this strain is summarized as in Figure 9. The alginate lyases degrade polymer alginate to unsaturated monomer (DEH) in the periplasmic space [24,25]. DEH reductase, KDG kinase and KDPG aldolase convert DEH to pyruvate and GAP in the cytosol. Therefore, DEH produced in the periplasmic space should be incorporated to the cytosol by certain transportation system(s). Such DEH transporters in this strain have not been identified yet; however, sugar permease-like gene sugp and membrane transporter-like genes susc and susd were found in the operons A and B, respectively (see Figure 1). Thus, the putative permease and transporters are also indicated in Figure 9. Another problem is how the expressions of alginatemetabolizing genes are regulated. We recently noticed that expression levels of alginate lyases were significantly low in the absence of alginate but strongly increased by the addition of alginate to the medium. This indicates that the expressions of alginate-metabolic enzymes are up-regulated by alginate. We are now searching regulatory genes for alginate-metabolizing enzyme genes in the UMI-01 strain genome.

Alginate-Metabolizing Enzymes of Flavobacterium sp. Strain UMI-01
In the present study, flkin and flald in the genome of Flavobacterium sp. strain UMI-01 were confirmed to be the enzyme genes encoding KDG kinase and KDPG aldolase. The recombinant enzymes, recFlKin and recFlAld, showed KDG kinase and KDPG aldolase activity although low sequence identities were shown to the corresponding enzymes from other bacteria and archaea (Figures 2-4). Consequently, these genes, along with previously reported alginate lyase and DEH reductase genes were confirmed to be the genes responsible for alginate metabolism of this bacterium. The alginate-metabolizing pathway of this strain is summarized as in Figure 9. The alginate lyases degrade polymer alginate to unsaturated monomer (DEH) in the periplasmic space [24,25]. DEH reductase, KDG kinase and KDPG aldolase convert DEH to pyruvate and GAP in the cytosol. Therefore, DEH produced in the periplasmic space should be incorporated to the cytosol by certain transportation system(s). Such DEH transporters in this strain have not been identified yet; however, sugar permease-like gene sugp and membrane transporter-like genes susc and susd were found in the operons A and B, respectively (see Figure 1). Thus, the putative permease and transporters are also indicated in Figure 9. Another problem is how the expressions of alginate-metabolizing genes are regulated. We recently noticed that expression levels of alginate lyases were significantly low in the absence of alginate but strongly increased by the addition of alginate to the medium. This indicates that the expressions of alginate-metabolic enzymes are up-regulated by alginate. We are now searching regulatory genes for alginate-metabolizing enzyme genes in the UMI-01 strain genome.

Properties of recFlKin and recFlAld
KDG kinase and KDPG aldolase are known to be the enzymes included in Entner-Doudoroff (ED) pathway. This pathway distributes over bacteria and archaea and play important roles in the metabolisms of glucuronate and glucose. In this pathway, KDG kinase phosphorylates KDG to KDPG, and KDPG aldolase split KDPG to pyruvate and GAP. Optimal temperature and pH of recFlKin were 50 °C and ~7.0, which were similar to those of KDG kinase from the bacteria S. marcescens [39]. While thermal stability of recFlKin was considerably low compared with the enzymes from archaea S. tokodaii [47] and S. solfataricus [40], e.g., these enzymes were stable up to 60-70 °C. recFlAld acts only on KDPG unlike archaea aldolases which split both KDG and KDPG [48,49]. Primary structures of bacterial aldolases showed low identity with those of archaea enzymes. The amino acid sequence of FlAld showed only 22%-25% identity with respect to Proteobacteria enzymes, while it showed 61%-65% identity with the Bacteroidetes enzymes. This suggests that somewhat deviated function between the Proteobacteria enzymes and Bacteroidetes enzymes. However, less different properties were found in recFlAld. Reverse reaction of bacterial aldolases was shown to be useful for the production of KDPG from pyruvate and GAP and also various compounds from pyruvate and aldehydes [44]. Our preliminary experiments also indicated that recFlAld could produce KDPG from pyruvate and GAP (data not shown, but see Section 2.3). Thus, recFlAld is also considered to be useful for producing novel compounds from pyruvate and various aldehydes.

Construction of In Vitro Alginate-Metabolizing System
An in vitro alginate-metabolizing system was successfully constructed from the recombinant enzymes, recFlAlyA, recFlAlyB, recFlAlex, recFlRed, recFlKin, and recFlAld. Accordingly, various kinds of intermediates could be produced by this system (Figure 8 and Table 2). Recently, alginateassimilating enzymes of S. degradans 2-40 T were used for the production of KDG, KDPG, GAP and pyruvate [24,25]. However, the reaction efficiency of KDG kinase of S. degradans 2-40 T appeared to be lower than that of our system. Namely, the major part of KDG in the reaction mixture remained to be unphosphorylated in the S. degradans 2-40 T system. On the other hand, recFlKin in our system could convert KDG to KDPG with ~80% efficiency. This difference in the reaction efficiency between S. degradans enzyme and recFlKin may be derived from the origin of this enzyme, namely, from

Properties of recFlKin and recFlAld
KDG kinase and KDPG aldolase are known to be the enzymes included in Entner-Doudoroff (ED) pathway. This pathway distributes over bacteria and archaea and play important roles in the metabolisms of glucuronate and glucose. In this pathway, KDG kinase phosphorylates KDG to KDPG, and KDPG aldolase split KDPG to pyruvate and GAP. Optimal temperature and pH of recFlKin were 50 • C and~7.0, which were similar to those of KDG kinase from the bacteria S. marcescens [39]. While thermal stability of recFlKin was considerably low compared with the enzymes from archaea S. tokodaii [47] and S. solfataricus [40], e.g., these enzymes were stable up to 60-70 • C. recFlAld acts only on KDPG unlike archaea aldolases which split both KDG and KDPG [48,49]. Primary structures of bacterial aldolases showed low identity with those of archaea enzymes. The amino acid sequence of FlAld showed only 22%-25% identity with respect to Proteobacteria enzymes, while it showed 61%-65% identity with the Bacteroidetes enzymes. This suggests that somewhat deviated function between the Proteobacteria enzymes and Bacteroidetes enzymes. However, less different properties were found in recFlAld. Reverse reaction of bacterial aldolases was shown to be useful for the production of KDPG from pyruvate and GAP and also various compounds from pyruvate and aldehydes [44]. Our preliminary experiments also indicated that recFlAld could produce KDPG from pyruvate and GAP (data not shown, but see Section 2.3). Thus, recFlAld is also considered to be useful for producing novel compounds from pyruvate and various aldehydes.

Construction of In Vitro Alginate-Metabolizing System
An in vitro alginate-metabolizing system was successfully constructed from the recombinant enzymes, recFlAlyA, recFlAlyB, recFlAlex, recFlRed, recFlKin, and recFlAld. Accordingly, various kinds of intermediates could be produced by this system (Figure 8 and Table 2). Recently, alginate-assimilating enzymes of S. degradans 2-40 T were used for the production of KDG, KDPG, GAP and pyruvate [24,25]. However, the reaction efficiency of KDG kinase of S. degradans 2-40 T appeared to be lower than that of our system. Namely, the major part of KDG in the reaction mixture remained to be unphosphorylated in the S. degradans 2-40 T system. On the other hand, recFlKin in our system could convert KDG to KDPG with~80% efficiency. This difference in the reaction efficiency between S. degradans enzyme and recFlKin may be derived from the origin of this enzyme, namely, from Proteobacteria species or Bacteroidetes species. To confirm this, we have to directly compare the KDG kinase properties between the enzymes from Proteobacteria and Bacteroidetes in future.

Production of a High-Value Product KDPG from Alginate
KDPG is a valuable leading compound for novel drugs and medicines. Synthesis of KDPG has been attempted by several methods [44,48,50]. For example, KDPG was first produced from gluconate with archaea enzymes [48]. However, this method required high-temperature reaction since the archaea enzymes are thermophilic. Reverse reaction of KDPG aldolase was also used for the production of KDPG from pyruvate and GAP [44,50]. However, this method required GAP, a significantly expensive raw material. On the other hand, we could produce KDPG from a much cheaper material, alginate, using the enzymes from the strain UMI-01. High recovery of KDPG from alginate (~80%) also indicated the practical potentiality of this enzyme. Thus, Flavobacterium sp. strain UMI-01 was considered to be a useful enzyme source for the production of value-added materials from alginate.

Materials
Sodium alginate (Macrocystis pyrifera origin) was purchased from Sigma-Aldrich (St. Louis, MO, USA). Alginate-assimilating bacteria, Flavobacterium sp. strain UMI-01, was cultivated at 25 • C in a mineral salt (MS) medium including 1% (w/v) sodium alginate as described in our previous report [26]. Cell lysate (crude enzyme) of this strain was extracted from cell pellets by freeze and thaw followed by sonication as described previously [28]. DEH was prepared by the digestion of sodium alginate with the crude enzyme and purified by SuperQ-650S (Tosoh, Tokyo, Japan) anion-exchange chromatography [28]. Standard KDG, KDPG, pyruvate, and GAP were purchased from Sigma-Aldrich. pCold I expression vector was purchased from TaKaRa (Shiga, Japan) and modified to the form that can add 8 × Gly + 8 × His-tag to the C-terminus of the expressed proteins [26]. E. coli DH5α and BL21 (DE3) were purchased from TaKaRa. Ni-NTA resin was purchased from Qiagen (Hilden, Germany). A TLC silica gel 60 plate was purchased from Merk KGaA (Darmstadt, Germany). TSKgel DEAE-2SW (4.6 × 250 mm) and Superdex peptide 10/300 GL were purchased from Tosoh Bioscience LLC (King of Prussia, PA, USA) and GE Healthcare (Little Chalfont, Buckinghamshire, UK), respectively. Lactate dehydrogenase (LDH; porcine heart origin) and NADH were purchased from Oriental Yeast Co., LTD. (Tokyo, Japan). ATP and 9-aminoacridine were purchased from Sigma-Aldrich. Other chemicals were purchased from Wako Pure Chemical Industries Ltd. (Osaka, Japan).

Phylogenetic Analysis for KDG Kinases and KDPG Aldolases
Phylogenetic analysis was carried out using the amino acid sequences of KDG kinases or KDPG aldolases from Proteobacteria, Bacteroidetes and Archaea currently available. Bacteroidetes enzymes used are from Gramella forsetii KT0803, Lacinutrix sp. 5H-3-7-4, and Dokdonia sp. MED134, which were reported to be located in the alginolytic gene cluster of each species [51]. These amino acid sequences were first aligned with the sequences of FlKin or FlAld by the ClustalW program, then aligned sequences were trimmed with GBlocks. Phylogenetic trees were generated by the maximum likelihood algorithm on the basis of the LG model implemented in the Molecular Evolutionary Genetics Analysis version 6.0 (MEGA 6) software. The bootstrap values were calculated from 1000 replicates.

Cloning, Expression, and Purification of Recombinant FlKin and FlAld
Genomic DNA of strain UMI-01 was prepared with ISOHAIR DNA extraction kit (Nippon Gene, Tokyo, Japan). Coding regions of flkin and flald, 1023 bp and 669 bp, respectively, were amplified by PCR using specific primers including restriction sites, NcoI and BamHI, in the 5 -terminal regions ( Table 1). Genomic PCR was performed in a medium containing 10 ng of genomic DNA, 0.2 µM each primer, and Phusion DNA polymerase (New England Biolabs, Ipswich, MA, USA). The reaction medium was preincubated at 95 • C for 2 min, and a reaction cycle of 95 • C for 10 s, 55 • C for 20 s, and 72 • C for 60 s was repeated 30 times. The PCR product was ligated to pCold I vector pre-digested by NcoI and BamHI using In-Fusion cloning system (Clontech Laboratories, Mountain View, CA, USA). Insertion of the genes in the vector was confirmed by nucleotide sequencing with DNA sequencer 3130xl (Applied Biosystems, Foster, CA, USA). Recombinant enzymes, recFlKin and recFlAld, were expressed with the pCold I-E. coli BL21 (DE3) system. The transformed BL21 (DE3) was inoculated to 500 mL of 2× YT medium and cultivated at 37 • C for 16 h. Then, the temperature was lowered to 15 • C and isopropyl β-D-1-thiogalactopyranoside was added to make the final concentration of 0.1 mM. After 24-h induction, bacterial cells were harvested by centrifugation at 5000× g for 5 min and suspended in a buffer containing 10 mM imidazole-HCl (pH 8.

Preparation of KDG
KDG was prepared from alginate using the crude enzyme of the strain UMI-01 as follows; 0.5% (w/v) sodium alginate (50 mL) was digested at 30 • C with 1 mg/mL of the crude enzyme, which contains alginate lyases and other metabolic enzymes. NADH was added to the mixture to make the final concentration 10 mM to reduce DEH with DEH reductase contained in the crude enzyme. After 12 h, four volumes of −20 • C 2-propanol were added to terminate the reaction and the proteins and NADH precipitated were removed by centrifugation at 10,000× g for 10 min. The supernatant containing KDG was dried up in a rotary evaporator at 35 • C. The dried powder was dissolved in 50 mL of distilled water and subjected to a TOYOPEARL SuperQ-650S column (2.4 × 22 cm) equilibrated with distilled water. The absorbed KDG and trace amount of unsaturated disaccharide were separately eluted by a linear gradient of 0-0.2 M NaCl in distilled water (total 400 mL). Elution of KDG and unsaturated disaccharide was detected by TBA reaction. In this chromatography, KDG was eluted at around 80 mM NaCl, while disaccharides were eluted at around 120 mM. Approximately 90 mg of KDG was obtained from 0.25 g of sodium alginate.

Preparation of KDPG
KDPG was prepared from the KDG by using recFlKin. Namely, recFlKin was (final concentration 10 µg/mL) was added to the reaction mixture (10 mL) containing 2.5 mM KDG, 2.5 mM ATP, 5 mM MgCl 2 , 20 mM Tris-HCl (pH 7.4), 100 mM KCl, and 1 mM dithiothreitol, and incubated at 40 • C for 3 h. The mixture was lyophilized, dissolved in 500 µL of distilled water and the supernatant was subjected to a Superdex peptide 10/300 GL column equilibrated with 0.1 M CH 3 COONH 4 . KDPG and KDG, which eluted together in this chromatography, were lyophilized, dissolved in 1 mL of distilled water, and subjected to HPLC (Shimadzu Prominence LC-6AD, Tokyo, Japan) equipped by TSKgel DEAE-2SW (Tosoh). KDG and KDPG were separately eluted at around 150 mM and 320 mM CH 3 COONH 4 by the linear gradient of 0-0.4 M CH 3 COONH 4 . The amount of KDPG was quantified by the system comprising recFlAld and LDH-NADH using authentic KDPG as a standard. By the above procedure, 1.2 mg of KDPG was obtained from 4.5 mg of KDG.

Assay for KDPG Aldolase Activity
KDPG aldolase activity of recFlAld was assayed by the determination of pyruvate using a lactate dehydrogenase (LDH)-NADH coupling system [50]. Namely, the aldolase reaction was conducted at 30 • C in a reaction mixture containing 5 mM KDPG, 20 mM Tris-HCl (pH 7.4), 100 mM KCl, 1 mM DTT, and 1 µg/mL recFlAld in the presence of 0.2 mM NADH and 1 unit/mL LDH. The reaction rate was estimated from the decrease in the Abs 340 nm due to the oxidation of NADH accompanied by the reduction of pyruvate. One unit (U) of KDPG aldolase activity was defined as the amount of enzyme that produced 1 µmol of pyruvate per min.

Assay for KDG Kinase Activity
KDG kinase activity was assayed as follows. The reaction mixture containing 2.5 mM KDG, 2.5 mM ATP, 5 mM MgCl 2 , 20 mM Tris-HCl (pH 7.4), 100 mM KCl, 1 mM DTT, and 10 µg/mL recFlKin was incubated at 30 • C. At reaction times, 1, 15, and 30 min, an aliquot (160 µL) of the reaction mixture was taken out and heated at 100 • C for 3 min to terminate the reaction. To the mixture, 240 µL of a buffer containing 84 mM Tris-HCl (pH 7.4), 167 mM KCl, 0.67 mM NADH, 2.5 µg/mL recFlAld, and 1 unit of LDH was added and the pyruvate released was determined by the LDH-NADH system. One unit (U) of KDG kinase activity was defined as the amount of enzyme that produced 1 µmol of KDPG per min. Temperature dependence of recFlKin was determined at 10-60 • C. Thermal stability of recFlKin was assessed by measuring the activity remaining after the incubation at 10-50 • C for 30 min. pH dependence of recFlKin was determined with reaction mixtures adjusted to pH 4.5-5.3 with 20 mM CH 3 COONa buffer, pH 5.6-7.3 with 20 mM PIPES-NaOH buffer, pH 7.1-8.8 with 20 mM Tris-HCl buffer, and pH 9.1-9.7 with 20 mM glycine-NaOH buffer. The activity assay was conducted three times and the mean value was shown with standard deviation in each figure.

Determination of Unsaturated Sugars
Unsaturated sugars were determined by the TBA method [52]. The sample containing unsaturated sugars (150 µL) was mixed with 150 µL of 20 mM NaIO 4 -0.125 M H 2 SO 4 and allowed to react for 1 h on ice. Then, 100 µL of NaAsO 2 -0.5 N HCl was added to the mixture and incubated for 10 min at room temperature. To the mixture, 600 µL of 0.6% (w/v) TBA was added and heated for 10 min at 100 • C. The unsaturated sugars were determined by measuring Abs 548 nm, adopting the absorption coefficient for DEH and KDG, ε = 41 × 10 3 M −1 ·cm −1 , which we determined in the present study using KDG and DEH standards.

Thin-Layer Chromatography
TLC silica gel 60 plate was used for the analysis of the reaction products produced by recFlKin and recFlAld. The reaction product of recFlKin was prepared with a reaction mixture containing 2.5 mM KDG and 2.5 mM ATP and 200 µg/mL recFlKin. The reaction was carried out at 30 • C for 0-15 min and terminated by heating at 100 • C for 2 min. Four microliters of each reaction mixture was applied to a TLC plate. The reaction product was developed with 1-butanol:acetic acid:water = 2:1:1 (v:v:v) and detected by heating at 130 • C for 10 min after spraying 10% (w/v) sulfuric acid-90% (w/v) ethanol. The reaction product of recFlAld was prepared with a reaction mixture containing 5 mM KDPG and 1 µg/mL recFlAld. After the reaction at 30 • C for 0-15 min, six microliters of the reaction mixture were applied to TLC plate and developed with the same solvent as described above. The reaction product on the plate was detected with 0.5% (w/v) 2,5-dinitrophenylhydrazine (DNP)-20% (v/v) sulfuric acid-60% (v/v) ethanol. In case of unsaturated sugars, they were visualized with 4.5% (w/v) TBA after the periodic acid treatment.

Mass Spectrometry
Phosphorylation of KDG by recFlKin was detected by mass spectrometry. The KDG phosphorylated by recFlKin in the conditions described in Section 4.10 was mixed with 6.7 mg/mL 9-aminoacridine-methanol at 1:3 (v:v). One microliter of the mixture was applied to a sample plate and air-dried at room temperature. The sample was subjected to a matrix-assisted laser desorption ionization-time of flight mass spectrometer (MALDI-TOF-MS) (Proteomics Analyzer 4700, Applied Biosystems, Foster City, CA, USA) and analyzed in a negative-ion mode.

Determination of Protein Concentration
Protein concentration was determined by the method of Lowry [54] using bovine serum albumin fraction V as a standard.

Conclusions
Enzymes responsible for the metabolism of alginate-derived DEH had not been well characterized in alginolytic bacteria. In the present study, KDG kinase-like gene flkin and KDPG aldolase-like gene flald in the genome of Flavobacterium sp. strain UMI-01 were investigated and the activities of the proteins encoded by these genes were assessed by using recombinant enzymes recFlKin and recFlAld. Analyses for reaction product of recFlKin and recFlAld indicated that these enzymes were KDG kinase and KDPG aldolase, respectively. Thus, the alginate metabolism of Flavobacterium sp. strain UMI-01 was considered to be achieved by the actions of FlKin and FlAld along with alginate lyases FlAlyA, FlAlyB and FlAlex, and DEH reductase FlRed. An in vitro alginate-metabolizing system was successfully constructed from the above enzymes. This system could convert alginate to pyruvate and GAP with 38% efficiency. This result indicates that the UMI-01 enzymes are available for the production of high-value materials like KDPG from alginate.