Biosynthesis of Vitamin B 6 in Rhizobium *

The biosynthetic pathway of pyridoxol (vitamin B 6 ) in Rhizobium was clarified by studies on the incorporation of 13 C- or 15 N-labeled precursors into pyridoxol or its biosynthetic intermediates. Pyridoxol was formed by ring closure of two compounds, 1-deoxy- D -xylulose and 4-hydroxy- L -threonine. The former was formed from D glyceraldehyde and pyruvate through decarboxylation of pyruvate, and the latter from glycine and glycolaldehyde. Extensive studies have been carried out on the biosynthetic pathway of vitamin B 6 in Escherichia coli . Spenser and col leagues (1, 2) proposed that pyridoxol is synthesized from two compounds, 1-deoxy- D -xylulose and 4-hydroxy- L -threonine, which serve as the C 5 unit, C-2 9 ,2,3,4, and 4 9 and the C 3 N unit, N-1, C-6,5, and 5 9 , respectively, of pyridoxol (Structure 1). (FAB)-MS 1 2 shifts external to 3-(trimethylsilyl)-propionic acid- d 4 sodium salt (0 ppm). FAB-MS car- ried out with a Jeol SX-102/102 mass spectrometer equipped with Apollo Series 400 data system. m -Nitrobenzyl alcohol was used as the liquid matrix for the FAB-MS experiments.

Extensive studies have been carried out on the biosynthetic pathway of vitamin B 6 in Escherichia coli. Spenser and colleagues (1,2) proposed that pyridoxol is synthesized from two compounds, 1-deoxy-D-xylulose and 4-hydroxy-L-threonine, which serve as the C 5 unit, C-2Ј,2,3,4, and 4Ј and the C 3 N unit, N-1, C-6,5, and 5Ј, respectively, of pyridoxol (Structure 1). 1-Deoxy-D-xylulose (or its 5-phosphate) has been identified as a biosynthetic precursor of isopentenyl diphosphate in a nonmevalonate pathway (3), the thiazole moiety of thiamin diphosphate (4), and pyridoxol in E. coli. Its formation from pyruvate and D-glyceraldehyde (or its 3-phosphate) has recently been demonstrated by an enzyme system using 1-deoxy-D-xylulose-5-phosphate synthase (5). For the formation of 4-hydroxy-Lthreonine in E. coli, two biosynthetic routes have been proposed. One is from D-erythrose 4-phosphate by a four-step reaction (6), and the other is from glycine and glycolaldehyde (7)(8)(9)(10). When the former pathway was blocked in E. coli, however, the microorganism became vitamin B 6 -deficient. This finding suggested that only the former pathway is active in E. coli.
During our screening search for vitamin B 6 overproducers, we found that Rhizobium meliloti IFO 14782 produced higher amounts of vitamin B 6 in the culture (11). In this study, we elucidated the biosynthetic pathway of pyridoxol in R. meliloti with regard to the following points: 1) formation of pyridoxol from 1-deoxy-D-xylulose and 4-hydroxy-L-threonine, 2) formation of 1-deoxy-D-xylulose from pyruvate and D-glyceraldehyde, and 3) formation of 4-hydroxy-L-threonine from glycine and glycolaldehyde.

EXPERIMENTAL PROCEDURES
Microorganism-The organism used in this study were R. meliloti IFO 14782, E. coli IFO 13168, Pseudomonas andropogonis ICMP 2809, and Saccharomyces carlsbergensis ATCC 9080 for quantitative determination of vitamin B 6 .
Formation of Vitamin B 6 -Cells of R. meliloti IFO 14782 were harvested from 3-day culture broth by centrifugation, washed twice with sterile 0.85% saline, and suspended in a small amount of sterile water. In a tube, 10 ml of the following mixture (2 mg each of 1-deoxy-Dxylulose and 4-hydroxy-L-threonine, 0.425% NaCl, and washed cells (final A 600 ϭ 20)) was prepared and incubated on a reciprocal shaker (285 rpm) at 28°C. After shaking for 24 h, the reaction mixture was centrifuged at 10,000 ϫ g for 10 min, and then vitamin B 6 in the supernatant was assayed with S. carlsbergensis ATCC 9080.
To isolate vitamin B 6 synthesized from labeled substrates, 20 tubes (total volume: 200 ml) containing the reaction mixture with 2 mg each of [1, 2-13 C 2 ]1-deoxy-D-xylulose and [ 15 N]4-hydroxy-L-threonine, 0.425% NaCl, and washed cells were shaken on a reciprocal shaker at 28°C for 24 h. The reaction mixture containing vitamin B 6 of 10.2 g/ml was centrifuged, and the vitamin produced was purified from the supernatant through column chromatography with Amberlite CG-120 (H ϩ ) * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Tel.: 81-467-47-2224; Fax: 81-467-45-6812; E-mail: masaaki.tazoe@roche.com. 1 The abbreviations used are: DABS, 4-(dimethylamino)azobenzene-cation exchange resin (1.2 cm in diameter and 15 cm in length) developed by 5% ammonium solution. Fractions containing a vitamin B 6 peak were pooled, concentrated under reduced pressure, and subjected to 13 C NMR structural analysis. 13  French press homogenizer. After centrifugation at 13,000 ϫ g for 60 min, the supernatant was used as the cell-free extract for the following enzyme reaction. The reaction mixture containing 100 mM sodium pyruvate, 100 mM D-glyceraldehyde, 80 mM Tris-HCl buffer (pH 8.0), 5 mM MgCl 2 , 0.5 mM thiamin diphosphate, 2 mM EDTA, and the cell-free extract (5.6 mg of protein) in a total volume of 1 ml was incubated at 37°C. After incubation for 1, 2, 3, or 4 h, the reaction mixture was heated in a boiling bath for 3 min and then centrifuged at 10,000 ϫ g for 10 min. One microliter of the supernatant was loaded onto a TLC plate of Silica gel 60 (Merck, ethyl acetate/pyridine/H 2 O ϭ 90/5/3) and developed. The 1-deoxy-D-xylulose formed was detected by staining the plate with alkaline-triphenyltetrazolium chloride.
To isolate the 1-deoxy-D-xylulose synthesized from labeled substrate, the enzyme reaction was carried out by incubating a tube containing the following reaction mixture (7.5 ml) at 37°C: 100 mM [2, 3-13 C 2 ] pyruvate, 100 mM D-glyceraldehyde, 80 mM Tris-HCl buffer (pH 8.0), 5 mM MgCl 2 , 0.5 mM thiamin diphosphate, 2 mM EDTA, and the cell-free extract of R. meliloti IFO 14782 (42 mg of protein). After incubation for 4 h, a 5-fold volume of methanol was added to the mixture, and then it was centrifuged at 10,000 ϫ g for 10 min. The supernatant was concentrated under reduced pressure, dissolved in a small amount of methanol, and then chromatographed on a column (2.3 cm in diameter and 25 cm in length) of Silica gel 60 with ethyl acetate/pyridine/H 2 O ϭ 90/5/3. 1-Deoxy-D-xylulose was followed by TLC on Silica gel 60 plates, and the fractions containing 1-deoxy-D-xylulose were collected and concentrated under reduced pressure. The residue was dissolved in a small amount of water and further purified by high pressure liquid chromatography (HPLC) under analytical conditions as follows: column, Daisopack SP-120 -5-ODS-BP (Daiso Co., Japan); mobile phase, H 2 O; flow rate, 0.5 ml/min; detector, RI. Fractions containing 1-deoxy-D-xylulose were collected and concentrated under reduced pressure. Physicochemical properties of the obtained 1-deoxy-D-xylulose were as follows.
[␣] D 26.5°(c ϭ 1, H 2 O), 13  Isolation of 4-hydroxy-L-threonine from labeled substrate was done as follows. Ten tubes each containing 10 ml of the following mixture: 40 mM glycolaldehyde, 32 mM [1-15 N, 2-13 C]glycine, 100 mM Tris-HCl buffer (pH 7.6), and the cells of R. meliloti IFO 14782, were shaken on a reciprocal shaker at 28°C for 24 h. The total reaction mixture (100 ml) was passed through a Dowex 1X4 column by the same method as described in the previous paragraph and derivatized with DABS. The DABS derivative of 4-hydroxy-L-threonine was purified by reversedphase TLC with C 8 silica plates. A band having the same R F value as 4-hydroxy-L-threonine was extracted by using a solvent mixture of

RESULTS AND DISCUSSION
Vitamin B 6 formation from 1-deoxy-D-xylulose and 4-hydroxy-L-threonine as substrates was examined by an intact cell system of R. meliloti IFO 14782. As shown in Table I, vitamin B 6 was formed only when both of the substrates were present. The vitamin B 6 formed in the reaction mixture was analyzed by TLC of silica gel 60, and identified as pyridoxol. Further, an incorporation test of the stable-labeled substrates, [1,2-13 C 2 ]1deoxy-D-xylulose and [ 15 N]4-hydroxy-L-threonine, into a pyridoxol molecule was done and analyzed by 13 C NMR spectrometry. The 13 C NMR spectrum of authentic pyridoxol showed eight signals: 155.5, 145.5, 143.3, 139.5, 132.5, 60.9, 59.7, and 17.2 ppm, which were assignable to C-3, -2, -4, -5, -6, -4Ј, -5Ј, and -2Ј of the skeleton, respectively (Fig. 1A). On the other hand, the 13 C NMR spectrum of the 13 C-and 15 N-isotopically enriched pyridoxol showed only two signals, 146.6 and 18.0 ppm, which were assigned to be C-2 and -2Ј of the skeleton, respectively (Fig. 1B). The former signal appeared as a double doublet (J 1 ϭ 48.9 Hz, J 2 ϭ 12.2 Hz), due to 13 C, 15 N enrichment in contiguous two carbon and one nitrogen atoms, C2Ј-C2-N1 of pyridoxol skeleton (Fig. 1B-1) and the latter as a doublet (J 1 ϭ 48.9 Hz), due to 13 C enrichment in adjacent two carbon atoms, C2Ј-C2 of the skeleton (Fig. 1B-2). This result indicates that the double-labeled carbon bond, 13 C1-13 C2 of 1-deoxy-D-xylulose, and 15 N of 4-hydroxy-L-threonine were incorporated into the C2Ј-C2-N1 bond of the pyridoxol skeleton. This suggests that the C 5 unit of 1-deoxy-D-xylulose and the NC 3 unit of 4-hydroxy-L-threonine enter the C 5 (C-2Ј, -2, -3, -4, and -4Ј) and NC 3 (N-1, C-6, C-5, and C-5Ј) units, respectively, of the pyridoxol skeleton. Formation of 1-deoxy-D-xylulose was examined for the enzyme system of R. meliloti IFO 14782 by using pyruvate and D-glyceraldehyde as substrates. The time course of the reaction was analyzed by TLC on a Silica gel 60 plate, and a purple spot having the same R F value as 1-deoxy-D-xylulose was observed in 1, 2, 3, or 4 h of incubation by staining with alkalinetetrazolium chloride (Fig. 2). Further, incorporation of stablelabeled pyruvate into 1-deoxy-D-xylulose was elucidated by 13 C NMR spectroscopy. The bulk production of 1-deoxy-D-xylulose was carried out by using [2,3-13 C 2 ]pyruvate and D-glyceraldehyde as substrates. After the reaction, 1-deoxy-D-xylulose was purified by column chromatography in Silica gel 60, followed by HPLC, and then it was subjected to 13 C NMR structural analysis. The 13 C NMR spectrum of authentic 1-deoxy-D-xylulose showed five signals of the open form, 215.8, 80.0, 74.2, 65.0, and 28.5 ppm, which were assignable to C-2, -3, -4, -5, and -1, respectively, of the skeleton (Fig. 3A). On the other hand, the 13 C NMR spectrum of the 13 C-isotopically enriched 1-deoxy-Dxylulose showed only two signals of the open form, 215.8 and 28.5 ppm, which were attributed to C-2 and -1, respectively, of the 1-deoxy-D-xylulose skeleton (Fig. 3B). In expanded spectra of 215.4 -216.5 and 28 -30 ppm spectral regions (Fig. 3, B-1 and  2), two signals were observed as doublet peaks having the same coupling constant (J ϭ 41.2 Hz) due to 13 C enrichment in the contiguous carbon atoms, C1-C2, of the 1-deoxy-D-xylulose skeleton. This result indicates that the double-labeled carbon bond, 13 C2-13 C3, of pyruvate was incorporated into the C-1 and -2 carbons of the 1-deoxy-D-xylulose skeleton. Furthermore, 1-deoxy-D-xylulose isolated from the reaction mixture using [1,2,3-13 C 3 ]pyruvate showed a similar spectrum of 13 C NMR to that isolated from the reaction mixture using [2,3-13 C 3 ]pyruvate (the spectrum not shown). This indicates that C-2 and -3 of pyruvate were incorporated into C-1 and -2 of the 1-deoxy-D-xylulose skeleton without the incorporation of C-1 of pyruvate. These results suggest that the C 2 (C-1 and -2) and C 3 (C-3, -4, and -5) units of the 1-deoxy-D-xylulose skeleton are derived from C-3 and -2 of pyruvate and C-1, -2, and -3 of D-glyceraldehyde, respectively. This result is consistent with the proposal of Yokota and Sasajima (17) that 1-deoxy-D-xylulose is formed from pyruvate and D-glyceraldehyde through decarboxylation of pyruvate. Formation of 4-hydroxy-L-threonine was studied in the intact cell system of R. meliloti IFO 14782 with glycolaldehyde and glycine as substrates in Tris-HCl buffer. The reaction product was identified by comparing it with a DABS derivative of authentic 4-hydroxy-L-threonine. With this derivatization, 4-hydroxy-L-threonine can be separated from the glycine used as one of the substrates. The reaction mixture was analyzed by reversed-phase TLC on a C 8 silica plate, and an orange spot having the same R F value as 4-hydroxy-L-threonine was observed on the plate as shown in Fig. 4. Further, incorporation of labeled glycine into 4-hydroxy-L-threonine was elucidated by mass spectrometry. Then, the formation of 4-hydroxy-L-threonine in the intact cell system was carried out by using [1-15 N, 2-13 C]glycine instead of glycine as the substrate. After incubation for 24 h, the reaction mixture was passed through a Dowex 1X4 column, derivatized with DABS, and purified by reversedphase TLC on C 8 silica plates. The mass spectrum of the DABS derivative of 4-hydroxy-L-threonine enriched with labeled glycine and glycolaldehyde had a 423 m/z corresponding to the molecular ion minus 1 (Fig. 5B, lower spectrum), whereas the DABS derivative of 4-hydroxy-L-threonine had a 421 m/z corresponding to the molecular ion minus 1 (Fig. 5A, upper spectrum). These results indicate that a nitrogen and a carbon label of [1-15 N, 2-13 C]glycine were incorporated into the 4-hydroxy-L-threonine molecule, and that the 4-hydroxy-L-threonine might be constructed from the NC unit of glycine and the C 2 unit of glycolaldehyde.
Recently, Lam and Winkler (6) have proposed that 4-hy-droxy-L-threonine would be formed from D-erythrose 4-phosphate by four stepwise reactions as the major pathway in E. coli, and that activity of the first enzyme, D-erythrose-4-phosphate dehydrogenase, on the pathway was found in the cell-free extract of an E. coli strain (Fig. 6). Nevertheless, when we attempted to determine the presence of this enzyme activity in the cell-free extract of R. meliloti IFO 14782 according to the method they reported, we were unable to detect this enzyme activity (data not shown). In other experiments, we have isolated a vitamin B 6 -requiring mutant derived from R. meliloti IFO 14782 that is defective in formation of 4-hydroxy-L-threonine from glycine and glycolaldehyde (data not shown). These results support that R. meliloti IFO 14782 synthesizes 4-hydroxy-L-threonine from glycine and glycolaldehyde, but not from D-erythrose 4-phosphate. Accordingly, we conclude that the biosynthetic pathway of 4-hydroxy-L-threonine in R. me-liloti is different from that in E. coli.
Incorporation of glycine into the pyridoxol molecule was studied in an intact cell system of R. meliloti IFO 14782 with [1,2-13 C 2 ]1-deoxy-D-xylulose, [1-15 N]glycine, and glycolaldehyde as substrates. Formed pyridoxol was purified from the supernatant of the reaction mixture by cation exchange column chromatography and then analyzed with the 13 C NMR spectrometer. The spectrum was similar to that of pyridoxol formed from [1,2-13 C 2 ]1-deoxy-D-xylulose and [ 15 N]4-hydroxy-L-threo-  nine. The result indicates that the labeled nitrogen of glycine enters the N-1 position of pyridoxol, and that the NC (N-1 and C-6) unit of pyridoxol would be derived from the NC unit of glycine.
In conclusion, the biosynthetic pathway of vitamin B 6 in R. meliloti is summarized as follows. Pyridoxol is synthesized from 1-deoxy-D-xylulose and 4-hydroxy-L-threonine; the former is from pyruvate and D-glyceraldehyde through decarboxylation of pyruvate, and the latter is from glycolaldehyde and glycine.