Pyridoxal Reductase, PdxI, Is Critical for Salvage of Pyridoxal in Escherichia coli

The biosynthetic pathway of pyridoxal 5′-phosphate (PLP) has extensively been studied in Escherichia coli, yet limited information is available about the vitamin B₆ salvage pathway. We show that the protein PdxI (YdbC) is the primary pyridoxal (PL) reductase in E. coli and is involved in the salvage of PL. The orthologs of PdxI occur in a wide range of bacteria and plants, suggesting that PL reductase in the B₆ salvage pathway is more widely distributed than previously expected.

ABSTRACT

Pyridoxal 5′-phosphate (PLP) is the biologically active form of vitamin B₆ and an essential cofactor in all organisms. In Escherichia coli, PLP is synthesized via the deoxyxylulose 5-phosphate (DXP)-dependent pathway that includes seven enzymatic steps and generates pyridoxine 5′-phosphate as an intermediate. Additionally, E. coli is able to salvage pyridoxal, pyridoxine, and pyridoxamine B₆ vitamers to produce PLP using kinases PdxK/PdxY and pyridox(am)ine phosphate oxidase (PdxH). We found that E. coli strains blocked in PLP synthesis prior to the formation of pyridoxine 5′-phosphate (PNP) required significantly less exogenous pyridoxal (PL) than strains lacking pdxH and identified the conversion of PL to pyridoxine (PN) during cultivation to be the cause. Our data showed that PdxI, shown to have PL reductase activity in vitro, was required for the efficient salvage of PL in E. coli. The pdxI⁺ E. coli strains converted exogenous PL to PN during growth, while pdxI mutants did not. In total, the data herein demonstrated that PdxI is a critical enzyme in the salvage of PL by E. coli.

IMPORTANCE The biosynthetic pathway of pyridoxal 5′-phosphate (PLP) has extensively been studied in Escherichia coli, yet limited information is available about the vitamin B₆ salvage pathway. We show that the protein PdxI (YdbC) is the primary pyridoxal (PL) reductase in E. coli and is involved in the salvage of PL. The orthologs of PdxI occur in a wide range of bacteria and plants, suggesting that PL reductase in the B₆ salvage pathway is more widely distributed than previously expected.

INTRODUCTION

Pyridoxal 5′-phosphate (PLP) is the biologically active form of vitamin B₆ that is used as a cofactor in various enzymes. PLP-dependent enzymes catalyze diverse biochemical reactions, including transamination, racemization, decarboxylation, and α,β-elimination or replacement of chemical groups at C_β or C_γ involving amino acid, sugar, and lipid metabolisms (1, 2). Most bacteria harbor several MocR/GabR-type PLP-binding transcriptional regulators to modulate amino acid and/or vitamin B₆ metabolisms (3–6).

Two distinct pathways for PLP biosynthesis, deoxyxylulose 5-phosphate (DXP)-dependent and DXP-independent pathways, have been reported. In Escherichia coli and a few members of the proteobacteria, PLP is synthesized via the DXP-dependent pathway that includes seven reaction steps catalyzed by seven enzymes (GapB, PdxB, SerC, PdxA, DXS, PdxJ, and PdxH) from erythrose 4-phosphate, glyceraldehyde 3-phosphate, and pyruvate (7, 8). In this pathway, the pyridine ring is formed by the action of the pyridoxine 5′-phosphate synthase (PdxJ), which catalyzes the condensation of DXP with 3-hydroxy-1-aminoacetone phosphate to form pyridoxine 5′-phosphate (PNP) (9) (Fig. 1). PNP is oxidized by a flavin mononucleotide (FMN)-dependent enzyme pyridoxine/pyridoxamine 5′-phosphate oxidase (PdxH) to form PLP (10). PdxH can also oxidize pyridoxamine 5′-phosphate (PMP) to produce PLP.

FIG 1.

Vitamin B₆ salvage pathway in E. coli. Biosynthesis of PLP via the DXP-dependent pathway and salvage of B₆ vitamers are schematically shown. Pyridoxine 5′-phosphate synthase (PdxJ) catalyzes the condensation of DXP with 3-hydroxy-1-aminoacetone phosphate to form pyridoxine 5′-phosphate (PNP) in de novo synthesis. PLP can be formed from PNP or PMP by pyridoxine/pyridoxamine 5′-phosphate oxidase (PdxH). PdxK can phosphorylate the three B₆ vitamers, pyridoxine (PN), pyridoxamine (PM), and pyridoxal (PL), allowing PLP to be salvaged from the environment by the routes depicted. PdxY contributes to the phosphorylation of PL only. In addition, amino acid transaminases convert PMP to PLP and vice versa during their catalytic cycle. Some phosphatases, including YbhA and PhoA, can dephosphorylate PLP. There is currently little information about the bacterial vitamin B₆ transporter(s). This study demonstrated that PdxI contributes to PL salvage and catalyzes the irreversible conversion of PL to PN under physiological conditions. Known or identified pathways are shown in solid arrows. Broken arrows indicate unidentified/missing pathways.

E. coli can convert various forms of B₆ vitamers to PLP using the salvage pathway (11). Our current understanding of the salvage pathway in E. coli is summarized in Fig. 1. Unlike the biosynthetic pathway, details of the salvage pathway have not been fully elucidated. Two kinases, PdxK and PdxY, can phosphorylate the 5′ position of the B₆ vitamers (12, 13). PdxK is capable of phosphorylating all three B₆ vitamers, pyridoxine (PN), pyridoxamine (PM), and pyridoxal (PL), whereas PdxY phosphorylates only PL (13). Various amino acid transaminases are capable of converting PMP to PLP and vice versa via their reaction mechanism. Some phosphatases, including YbhA and PhoA, are capable of dephosphorylating PLP and/or PNP in vivo and in vitro (14, 15).

During the study of the biosynthesis, homeostasis, and metabolic network of vitamin B₆ in Escherichia coli and Salmonella enterica (16–18), we found that E. coli strains lacking pdxJ had different PL requirements than those lacking pdxH. This result was unexpected given that both lesions abolish PLP biosynthesis. The pdxH mutant required 10 times more PL than the pdxJ mutant to reach an optimal growth yield. This study was initiated to determine the cause of this difference. The data herein show that the protein YdbC (PdxI), reported to catalyze PL reductase activity in vitro (19), was primarily responsible for the differences of PL requirement observed and is a component of PL salvage in E. coli.

RESULTS AND DISCUSSION

Elimination of pdxH increases PL requirement in E. coli.

The ability of exogenous pyridoxal (PL) to allow growth of strains of E. coli blocked in the synthesis of PLP has been attributed to the PL kinase activity of PdxK/PdxY (Fig. 1). In the course of other work, it was noted that pdxJ and pdxH mutants, both of which cannot synthesize PLP de novo, required different concentrations of PL for growth in an M9-glucose medium. The pdxJ-deficient strain grew with a rate and reached a final density similar to those of a wild-type strain (Fig. 2A) when 0.1 μM PL was provided (1-h doubling time) (Fig. 2B). In contrast, the pdxH mutant required 1 μM of PL to achieve an optimal growth yield (1-h doubling time) (Fig. 2C). The final yield, but not the growth rate, of the pdxH mutant was affected by the concentration of PL. This result suggested that PL was limiting and was not simply being poorly utilized. Eliminating PdxJ or PdxH blocked PLP formation, though the pdxH mutant accumulated PNP due to flux through the PdxJ-dependent pathway (Fig. 1). A pdxJ pdxH double mutant exhibited slower growth than either the pdxJ or pdxH single mutant (1.5-h doubling time) but had a growth pattern most similar to that of the pdxH mutant strain, requiring >10-fold more exogenous PL than the pdxJ mutant (Fig. 2D). These data supported the conclusion that flux through the PLP biosynthetic pathway did not contribute to the increased PL requirement of the pdxH mutant.

FIG 2.

PL requirement of E. coli pdxJ, pdxH, or pdxJ pdxH mutants. (A to D) Wild-type (A) and pdxJ (B), pdxH (C), or pdxJ pdxH (D) mutant strains of E. coli were grown in M9-glucose medium in the absence (○) or presence of PL at 0.1 μM (■), 0.25 μM (●), or 1 μM (△). (E and F) pdxJ (E) and pdxJ pdxH (F) mutants of S. enterica were grown in NCE glucose medium in the absence (○) or presence of 0.1 μM (■) or 1 μM (△) PL. The data represent the averages from triplicate experiments. Cell growth was recorded by the OD-Monitor C&T apparatus (A to D) or the ELx808 (E and F) using a 96-well plate.

The above-described data were not anticipated based on the described salvage pathway for B₆ vitamers in E. coli. In contrast, Salmonella enterica displayed a result closer to what was expected based on the described salvage enzymes. Strains lacking pdxJ or pdxH had a growth response to 0.1 μM of PL that was more similar to the full-growth allowed by 1 μM in no-carbon essential (NCE) minimal medium (Fig. 2E and F).

E. coli pdxJ pdxH mutant converts exogenous PL to PN and accumulates PNP.

Based on the kinases implicated in the salvage pathway, PdxK and PdxY, the PLP requirement of a pdxJ mutant should be satisfied with PL, PN, or PM (Fig. 1). This expectation was confirmed by growth analyses with 0.1 μM of each B₆ vitamer (data not shown). In contrast, a pdxH mutant is not expected to utilize exogenous PN or efficiently convert PM to PLP and is likely to accumulate PNP synthesized through the DXP-dependent pathway (Fig. 1). Thus, a difference in the requirement for PL in the pdxJ and pdxH mutants was not predicted a priori. High-performance liquid chromatography (HPLC) analyses of the growth medium showed the disappearance of PL from the medium during the growth of a pdxJ pdxH mutant. As shown in Fig. 3A, the M9 growth medium of the pdxJ pdxH mutant, which contained 5 μM PL at time zero (T = 0), had only PN (∼4.5 μM) after 24 h of cell growth. PL was also absent in the medium after growth of the wild type (Fig. 3B) or the pdxH mutant strain (data not shown) and was seemingly replaced by the PN that accumulated. When cells were removed from the culture medium after 6 h of cultivation (by centrifugation and following filtration with a 0.45-μm filter membrane), no further conversion of PL to PN was observed, indicating that the conversion of PL to PN was mediated by E. coli cells (data not shown). The intracellular pools of B₆ vitamers were analyzed in the pdxJ pdxH double mutant after growth in M9-glucose medium in the presence of 5 μM PL. Despite the block in synthesis and the lack of a known pathway from PL to PNP, a significant amount of PNP accumulated in the cells (Fig. 3C, bottom trace). Consistent with the growth results described above, when an S. enterica pdxJ pdxH double mutant strain was grown under similar conditions, PN did not appear in the medium and PNP did not accumulate in the cells (data not shown). These data suggested that E. coli had a mechanism(s) to convert PL→PN and/or PLP→PNP that is both internal and not present in S. enterica. The pdxH mutation abolishes the utilization of PN and PNP as a PLP source in E. coli (Fig. 1). If PL were partially converted to PN or PNP during growth, it could explain the increased PL requirement observed with the pdxH or pdxJ pdxH mutant of E. coli, and PL availability would be decreased (Fig. 1).

FIG 3.

A pdxI mutation prevents conversion of extracellular PL to PN and intracellular accumulation of PNP. The pdxJ pdxH and pdxI pdxJ pdxH mutant strains of E. coli were grown in M9-glucose medium supplemented with 5 μM PL. The culture supernatant was collected at different time points (0, 6, or 24 h). Internal pool concentrations of B₆ vitamers were determined with cells collected at early stationary phase (OD₆₀₀ of 0.7) as described in Materials and Methods. The pdxJ pdxH double mutant converted all of the extracellular PL to PN during its cultivation (A, left) and accumulated PNP in the cells (C, bottom trace). The pdxI mutation significantly decreased the conversion of exogenous PL to PN (A, right) and intracellular accumulation of PNP (C, top trace). A wild-type E. coli strain showed conversion of extracellular PL to PN during cultivation (B, left), and the pdxI mutation significantly decreased this conversion (B, right). Experiments were performed at least three times with similar results, and representative chromatograms are shown.

PdxI is required for the conversion of PL to PN in E. coli.

The presence of a PL reductase in E. coli would explain these data. A recent global study using a supplemented metabolome extract and purified protein identified a protein that catalyzes the NADPH-dependent reduction of PL to PN (19). The relevant protein, YdbC, contains 286 amino acids, belongs to the aldo-keto reductase (AKR) family (20), and was renamed PdxI based on the activity. The same protein had reductase activity with 5-nitro benzisoxazole→2-cyano-4-nitrophenol with the K_m and k_cat values of 270 μM and 0.11 s⁻¹, respectively (21), and neither activity was attributed a physiological role.

A working model suggested that PdxI functioned as a PL reductase and converted PL to PN in the growth medium of E. coli strains. An insertion/deletion of pdxI was constructed and transduced into relevant genetic backgrounds. The requirement for B₆ vitamers was determined in the relevant strains. As expected based on the working model, the pdxJ pdxH pdxI triple mutant (DM16865) had a lower requirement for PL in M9-glucose medium, and 0.1 μM PL in the medium allowed full growth (Fig. 4A). The pdxH pdxI double mutant had a similarly low requirement for PL compared to that of the parental pdxH strain (data not shown). Together, these data showed that the increased requirement for PL was dependent on PdxI. Consistently, HPLC analyses of the vitamin B₆ pool showed that pdxI lesion in the pdxJ pdxH mutant background prevented the accumulation of PNP inside the cells (Fig. 3C). In strains with a lesion in pdxI (pdxI pdxH or pdxI pdxJ pdxH mutant), PL was not converted to PN over time in the growth medium (Fig. 3A). Finally, in an otherwise wild-type background, a pdxI mutation significantly decreased the accumulation of PN in medium supplemented with exogenous PL (Fig. 3B). A similar involvement of PL reductase in the accumulation of PN in the growth medium was reported in yeasts (22).

FIG 4.

PL requirement of E. coli pdxI pdxJ pdxH, pdxJ pdxK, or pdxI pdxJ pdxK mutant. The pdxI pdxJ pdxH (A), pdxJ pdxK (B), and pdxI pdxJ pdxK (C) mutant strains of E. coli were grown in M9-glucose medium in the absence (○) or presence of 0.1 μM (■), 0.25 μM (●), or 1 μM (△) PL. The data represent the averages from biological triplicates. Cell growth was recorded by the OD-Monitor C&T apparatus.

The physiological significance of the rapid PdxI-dependent conversion of PL to PN during growth and the location of this conversion remain to be elucidated. After 24 h of cultivation in M9 medium supplemented with PL, the pdxI mutant strain accumulated a small amount of PN in the medium (Fig. 3A and B). This result suggests there might be an additional enzyme(s) capable of reducing PL in E. coli. The genome of E. coli K-12 encodes PdxI and eight other proteins that belong to the AKR superfamily (DkgA [15%], DkgB [19%], YdjG [21%], YgdS [tas] [16%], YdhF [15%], YghZ [17%], YeaE [15%], YajO [20%], and PdxI; the percent values in brackets indicate sequence identities to PdxI). In yeasts and plants, members of the AKR family have PL reductase activity (22–25). Alignment analyses showed that the E. coli PdxI shares weak homology with the eukaryotic PL reductases and is 13.6%, 15.0%, and 15.4% identical to the Schizosaccharomyces pombe PLR1, Saccharomyces cerevisiae PLR1, and Arabidopsis thaliana PLR1, respectively (Fig. 5A). The phylogenic analysis suggested that PdxI is evolutionarily related to PL reductases of yeast, although their amino acid identities are low. In contrast, PdxI is evolutionarily less related to the PL reductase from A. thaliana (Fig. 5B). BLAST analysis using PdxI sequence as a query sequence identified proteins exhibiting greater than 30% identities to PdxI in many bacteria and plants, including Shigella sonnei, Streptomyces coelicolor, Deinococcus radiodurans, Thermotoga maritima, Pseudomonas aeruginosa, Glycine max, and A. thaliana. The BLAST search also showed that most E. coli strains, excluding E. coli B strains such as BL21, have a PdxI ortholog that has >90% identity. No protein-encoding gene that exhibits >30% identity to E. coli PdxI was found in the genome of S. enterica LT2 strain and most of other S. enterica strains, while there were 8 members of the AKR superfamily (YdjG is missing, while YqhE is present in the S. enterica strain) (Fig. 5B), supporting the conclusion that PdxI is the primary enzyme that efficiently converts PL to PN in E. coli.

FIG 5.

Sequence alignment of PL reductases. (A) Primary sequences of PL reductase of E. coli (PdxI), S. pombe (SpPLR1), S. cerevisiae (ScPLR1), and A. thaliana (AtPLR1) were aligned using the ClustalW program (https://www.genome.jp/tools-bin/clustalw). E. coli PdxI exhibits weak similarity to the three eukaryotic PL reductases. The PdxI is 13.6%, 15.0%, and 15.4% identical to the S. pombe PLR1, S. cerevisiae PLR1, and A. thaliana PLR1, respectively. (B) A phylogenetic tree of PdxI, SpPLR1, AtPLR1, and other proteins belonging to the AKR superfamily of E. coli and S. enterica based on the ClustalW algorithm is shown.

Kinetic parameters of PdxI.

Our growth data suggested that in E. coli pdxH and pdxJ pdxH mutant strains, the flux to PLP from direct salvage (PL to PLP) via kinases PdxK and PdxY (12, 13) is lower than flux via a salvage pathway requiring reduction of PL by PdxI (PL→PN→PNP→PLP). PL reduction catalyzed by PdxI was kinetically characterized using the purified enzyme. PL showed Michaelis-Menten kinetics in the substrate concentration range examined (0 to 0.5 mM). The apparent K_m and k_cat values of the PL reductase activity in the presence of 0.3 mM NADPH were determined to be 31 ± 3 μM and 84 ± 2 s⁻¹, respectively. PdxI had no detectable reductase activity with either PLP or PM as the substrates under the assay conditions used. Pyridoxal reductases of S. cerevisiae and S. pombe catalyze the reverse reaction (conversion of PN to PL in the presence of NADP⁺), although the efficiencies are low (22, 24). This activity was not detected for PdxI.

The K_m and k_cat values for the kinase activity of PL catalyzed by PdxK are reported to be 60 μM and 4 s⁻¹, respectively (26). PdxY exhibits only 1% of the kinase activity compared to that of PdxK (27, 28). The expression levels of PdxI, PdxK, and PdxY were not expected to be significantly different (29). In total, these data support the hypothesis that the conversion of PL to PN (by PdxI) is favored over the conversion of PL to PLP (by PdxK and/or PdxY) in E. coli cells. The efficient conversion of PL to PN would result in a rapid shortage of PL in the medium, causing the increased requirement for PL in the pdxH mutants, since these strains are unable to use the PN or PNP that would result from PdxI and PdxK activity.

Contribution of PdxI in the B₆ salvage pathway.

In S. pombe, the deletion of the pyridoxal reductase gene (plr1) resulted in a decrease in total vitamin B₆ and PMP contents (22). The transfer DNA (T-DNA) insertion mutant in the pyridoxal reductase in A. thaliana exhibits significantly lower levels of total B₆ vitamer, PL, PLP, PM, and PMP (25). We investigated the influence of disruption of pdxI on the homeostasis of B₆ vitamers. The wild type and the pdxI mutant were grown in M9-glucose medium, and their intracellular B₆ pools were analyzed. Importantly, the growth of the pdxI mutant was not significantly different from that of the wild-type strain when they were cultivated in M9-glucose medium. PL did not accumulate in either of the strains. The result showed that lack of pdxI did not significantly impact the intracellular B₆ pools in E. coli under the condition examined (Table 1).

TABLE 1.

Effect of pdxI deletion on intracellular B₆ pool

Genotypea	Amt (pmol/mg [wet wt] cells) ofb :
	PLP	PNP	PMP
WT	52 ± 7	ND	30 ± 4
pdxI	48 ± 2	ND	25 ± 3

^aWild-type (WT) and pdxI mutant were grown in M9-glucose medium.

^bThe intracellular B₆ pool was determined as described in Materials and Methods. The pdxI mutation did not affect the levels of intracellular B₆ vitamers. ND, no detectable amount was observed. The data represent the averages and standard deviations from triplicate experiments.

Conclusions.

Here, we report that E. coli possesses an efficient pathway to convert PL to PN. PdxI, which belongs to the AKR superfamily, exhibits weak homology to the known PL reductases. Our data showed that PdxI is also responsible for the accumulation of PN in the medium over time when E. coli is grown in the presence of PL. There are proteins homologous to PdxI in a wide range of bacteria and plants, suggesting that PL reductase in the B₆ salvage pathway is widely distributed in nature, despite being absent in a close relative of E. coli, S. enterica.

We updated the vitamin B₆ salvage pathway of E. coli taking account into the data described above and some results obtained with mutant strains that have single or multiple mutations in the enzyme(s) of the B₆ salvage pathway. The results are summarized in Table 2 and Fig. 1. As described, E. coli synthesizes PLP from exogenous PL using a detoured pathway (PL→PN→PNP→PLP) involving PdxI, PdxK, and PdxH rather than the direct pathway involving PdxK/PdxY kinases (PL→PLP). This mechanism was further supported by results showing the pdxJ pdxK mutant had poor growth in the presence of low levels of exogenous PL (0.1 μM), but the pdxJ pdxK pdxI triple mutant did not (Fig. 4B and C) (Table 2).

TABLE 2.

Growth of E. coli strains that have single or multiple mutations in B₆ salvage enzymes

Genotype	Growth witha :
	PL	PN	PM
pdxJ	+	+	+
pdxH	±	−	+
pdxJ pdxH	±	−	+
pdxJ pdxK	±	−	−
pdxJ pdxH pdxI	+	−	+
pdxJ pdxK pdxI	+	−	−

^a+, grew well; ±, poor growth (final OD was less than 0.2 after 16 h of cultivation); −, no growth.

MATERIALS AND METHODS

Bacterial strains and media.

All strains used in this study are derivatives of Escherichia coli BW25113 and are listed with their genotypes in Table 3. M9 medium supplemented with 0.2% glucose as the sole carbon source was used as the minimal medium for E. coli strains. Luria-Bertani (LB) broth was used for preculturing. Strains of S. enterica were grown in NCE medium. Agar (1.5%) was added for solid medium. When necessary, pyridoxal, pyridoxine, or pyridoxamine was added at designated concentrations. Antibiotics were added to the medium in a rich and minimal medium at the following final concentrations: ampicillin, 100 and 20 μg/ml; kanamycin, 50 and 10 μg/ml. Growth analyses in liquid medium were performed in a glass test tube using Taitec OD monitor C&T apparatus (for E. coli strains) or in 96-well microtiter plates in a BioTek ELx808 plate reader (for S. enterica strains). Media were incubated at 37°C with shaking, and growth was monitored every 30 min.

TABLE 3.

Strains, plasmids, and primers

Strain, plasmid, or primer	Description or sequence (5′→3′)	Source or reference
E. coli strains
BW25113	Wild type	Laboratory collection
DM16769	pdxI::Cm	This study
DM16696	pdxH::Kan	This study
DM16698	pdxH	This study
DM16027	pdxJ::kan (Keio collection, JW2548-KC)	33
DM16037	pdxJ	This study
DM16713	pdxJ pdxH::Kan	This study
DM16847	pdxH::Kan pdxI::Cm	This study
DM16865	pdxH::kan pdxI pdxJ	This study
AG1/pCA24N-pdxI	ASKA clone, JW1403-AM	32
Plasmid pCA24N-pdxI	N-terminal His-tagged PdxI expression (ASKA JW1403-AM)	32
Primers
pdxH-H1	ATGTCTGATAACGACGAATTGCAGCAAATCGCGCATCTG
	CGCCGTGAATGTGTAGGCTGGAGCTGCTTCG
pdxH-H2	TCAGGGTGCAAGACGATCAATCTTCCACGCATCATTTTC
	ACGCTGGTCATATGAATATCCTCCTTAG
pdxI-H1	ATGAGCAGCAATACATTTACTCTCGGTACAAAATCCGTT
	AACCGTCTTGTGTAGGCTGGAGCTGCTTCG
pdxI-H2	TTATTCTCGCGAAATACCATCCAACGTAGA
	CAACACTTCCTCAGAAAGATCATATGAATATCCTCCTTAG
pdxH-check-fw	CGCATCGTCTTGAATAACTGTCAG
pdxH-check-rv	CACCTTTGCCGGTACACGACTTTTC
pdxI-check-fw	GCAACTCATCCAGTAATCTTGTTTACACC
pdxI-check-rv	GTAAACGTATCCAGCCGCAATTCC

Deletion mutants of pdxH (DM16696) and pdxI (DM16769) were constructed with lambda Red recombineering as described previously using E. coli BW25113 as the parental strain (30). A kanamycin resistance cassette with a 40-bp homology region of the pdxH sequence was amplified by PCR with primers pdxH-H1 and pdxH-H2 and plasmid pKD4. A chloramphenicol resistance cassette flanked by 40-bp pdxI sequences was amplified by PCR with primers pdxI-H1 and pdxI-H2 and plasmid pKD3 as a template. These PCR products were purified from agarose gels and electroporated into the E. coli cells. Transformants were selected on an LB plate containing kanamycin (50 μg/ml) or chloramphenicol (30 μg/ml). Insertion of the antibiotic resistance gene at the pdxH or pdxI locus was confirmed by PCR using primer pair pdxH-check-fw/pdxH-check-rv or pdxI-check-fw/pdxI-check-rv. The antibiotic cassette was removed with pCP20 plasmid. When required, the pdxJ::Km, pdxK::Km, pdxH::Km, or pdxI::Cm mutations were transferred into the desired strain by P1 transduction as described previously (31).

Purification of PdxI.

An E. coli AG1/pCA24N-PdxI strain (JW1403-AM [32]) was grown overnight in LB medium supplemented with chloramphenicol (30 μg/ml). Two milliliters of the preculture was inoculated into 200 ml of LB medium containing chloramphenicol and grown at 37°C with shaking to an optical density at 650 nm (OD₆₅₀) of 0.5. Isopropyl-β-d-thiogalactopyranoside (IPTG; 0.1 mM) was added to the medium, and cells were incubated for 4 h prior to harvesting by centrifugation (6,000 × g, 5 min, room temperature [RT]). The cell pellet was resuspended in a binding buffer containing 50 mM sodium phosphate (NaPB), 500 mM NaCl, 20 mM imidazole, pH 7.4, with DNase (0.025 mg/ml), lysozyme (1 mg/ml), and phenylmethylsulfonyl fluoride (0.1 mg/ml) and disrupted with a high-pressure cell disruption device, Constant Systems Limited One Shot (United Kingdom) at 20,000 lb/in², and cell lysate was cleared by centrifugation at 48,000 × g for 30 min at 4°C. The cell extract was passed through a 0.45-μm filter membrane and applied to a HisTrap HP Ni-Sepharose column (GE Healthcare) preequilibrated with the binding buffer. The column was washed with 5 column volumes of the binding buffer and an elution buffer (20 mM NaPB, 500 mM NaCl, 500 mM imidazole, pH 7.4). Fractions containing PdxI, where the purified PdxI existed as a single band of ∼31 kDa by SDS-PAGE, were combined and concentrated with a centrifugal filter device. The buffer was replaced with a buffer consisting of 20 mM NaPB and 10% glycerol (pH 7.4) with a PD10 desalting column (GE Healthcare), flash-frozen in liquid nitrogen, and stored at −80°C until use.

Enzyme assay.

PL reductase activity catalyzed by PdxI was determined as described previously with slight modification (22, 24). Briefly, purified His-tagged PdxI (0.26 μg) was incubated in a reaction mixture containing 50 mM NaPB, 0.3 mM NADPH, and various concentrations (0, 0.025, 0.05, 0.1, 0.25, and 0.5 mM) of PL at 37°C. The reaction was started by the addition of purified PdxI. The decrease of NADPH absorption at 340 nm in a 1.0-ml reaction mixture was monitored over 5 min at 37°C. Control experiments were also performed in the absence of enzyme, PL, or NADPH.

B₆ vitamer analysis.

B₆ vitamers were extracted from the cells with 10 volumes (vol/wt) of 0.9 M HClO₄ containing 50 μM deoxypyridoxine as an internal standard (100 μl of the HClO₄ solution for 10 mg [wet weight] E. coli cells). The suspension was vortex mixed and incubated on ice for 15 min, and 5 volumes (vol/wt) of 0.9 M K₂CO₃ solution (50 μl for 10 mg E. coli cells) was added. The mixture was centrifuged, and the resultant supernatant was diluted three times with water and used for the HPLC analysis (25 μl). The culture medium was deproteinized with HClO₄ (a final concentration of 0.9 M), neutralized by K₂CO₃, and clarified by centrifugation. The B₆ vitamers were separated with an octadecylsilyl (ODS) column (Cosmosil AR-II; Nacalai Tesque) (250 mm by 4.6 mm, 5-μm particle size) using a gradient program as described previously (17, 18). The flow rate was 1.0 ml per min, and the excitation and emission wavelengths were 328 nm and 393 nm, respectively.