The regiochemistry and stereochemistry of the biosynthesis of vitamin B6 from triose units.

13C and 2H NMR spectroscopy has been employed to probe the biosynthesis of vitamin B6 in Escherichia coli. The 13C NMR spectrum of a sample of pyridoxol derived biosynthetically from D-[1,2,3,4,5,6-13C6]glucose shows that the bonds, C(2)-C(3) and C(4)-C(5), of the pyridine nucleus are the only two carbon-carbon bonds of pyridoxol which are generated de novo in the course of its biosynthesis from glucose. It follows that the pyridoxol skeleton is generated from two intact triose units and a triose-derived two-carbon unit, all of which are supplied by glucose. From the 2H NMR spectra of samples of pyridoxol derived from (R)-[1,1-2H2]glycerol and (S)-[1,1-2H2]glycerol, respectively, it can be deduced that the rehydroxymethyl group of glycerol enters C-2', C-4', and C-5' of the pyridoxol skeleton. It follows that each of the three fragments is derived from glycerol in stereo-specific fashion. These results answer questions concerning the regiochemistry and the stereochemistry of pyridoxol biosynthesis.

In studies of the biosynthesis of vitamin B6 by tracer experiments with putative precursors labeled with 14C and I3C (2-6), we have employed Escherichia coli B mutant WG2 (7). This mutant, one of the Group IV (pdxH) mutants of E. coli B (8), lacks the enzyme pyridoxol-phosphate oxidase (EC 1.1.1.65 or EC 1.4.3.5) (9). The mutant elaborates the complete ring skeleton of vitamin B6, as exemplified by pyridoxol, and differs from the wild-type strain of E. coli B merely in that there is a block in the final step of vitamin B, biosynthesis, oxidation of pyridoxol (and its 5"phosphate) into pyridoxal (and its 5"phosphate). A study of the biosynthesis of pyridoxol in this mutant thus represents an investigation of the normal biosynthetic process leading from primary precursors into the CsN ring skeleton of the vitamin.
The incorporation pattern within pyridoxol of label from 14C-labeled samples of glycerol and glucose permitted us to put forward a chemically rational hypothesis of the biosynthetic route from triose units into pyridoxol (2, 10). In advancing this hypothesis, we made assumptions concerning the location of newly formed bonds and concerning the identity of the triose intermediates involved in the biosynthetic process.

* This work was supported by grants from the Medical Research
Council of Canada (to R. E. H.) and from the Natural Sciences and Engineering Research Council of Canada (to I. D. S.) and in part by National Institutes of Health Research Resource Grant RR-02231. A preliminary account of part of this work has appeared (1). 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.
We now present evidence which locates the bonds that are newly formed in the process of biosynthesis of pyridoxol from glucose. Furthermore, we report results which establish the stereochemistry of the incorporation of glycerol.
These results answer questions relating to the regiochemistry and stereochemistry of the incorporation of triose precursors into pyridoxol and add support to the model of pyridoxol biosynthesis which we had advanced on the basis of earlier tracer evidence (2, 10).
A small sample was converted into tri-0-acetvlpvridoxol hvdro-  in chloroform (10 ml) was mixed with 80% m-chloroperbenzoic acid (260 mg) in chloroform (5 ml) at room temperature. The mixture was allowed to stand overnight and was then shaken at 0-10 "C with a saturated solution of aqueous sodium bicarbonate (3 x 5 ml), and the chloroform laver was dried (M&O, + NaSO, anhydrous) and evaporated.

Attempted
Exchange of Protons in Tri-0-acetylpyridorol 1 -Oxide (16) A saturated solution of sodium bicarbonate in D20 (2 ml) was added to the reaction mixture containing tri-0-acetylpyridoxol loxide, m-chlorobenzoic acid, and m-chloroperbenzoic acid from the oxidation of tri-O-acetylpyridoxol (see above). The mixture was cooled to 5 "C and stirred for 90 min. Tri-0-acetylpyridoxol l-oxide was then isolated. The 'H NMR, IR, and mass spectra of the isolated sample were identical with those from the sample prepared without
These assignments are based on nuclear Overhauser effect measurements.
It had previously been assumed (26), on the basis of empirical considerations but without direct evidence, that the signal due to the methyl group in the cis position relative to the R group, in a 2,2-dimethyl-4-(R-substituted)-1,3-dioxolane such as 9 or 10, would appear at higher frequency (i.e. downfield) than that due to the trans-methyl group. This assumption turns out to have been correct.
NMR Spectra of %'-and 'H-Labeled Samples of Pyridonol-The samples of pyridoxol hydrochloride which were isolated from E. coli B WG2 which had been incubated with D-[1,2,3,4,5,6-i3Cs]g1ucose (Experiment 1) and (R)-and (S)-[l,l-'H,]glycerol plus [2-'%]glycerol (Experiments 2 and 3) were analyzed by NMR spectroscopy. The 13C NMR spectrum of pyridoxol hydrochloride from Experiment 1 is shown in Fig. 3. The 'H NMR spectra of the samples from Experiments 2 and 3 are shown in Fig. 4 (B and C), which also shows a 'H NMR spectrum of pyridoxol hydrochloride (A). An interpretation of the spectra is offered in the discussion which follows.
Degradation of "C,3H-Labeled Pyridoxol Derived from [.% I%', 1 -3H]Glycerol-The sample of pyridoxol was degraded by the reaction sequence shown in Scheme 2. The 'H/Y! ratio of the final product, tri-O-acetyl-6-chloropyridoxol (17), was identical with that of the initial pyridoxol sample. The results of the degradation are summarized in Table II.

DISCUSSION
In our earlier tracer work (2-6), we demonstrated that in E. coli B mutant WG2, the entire carbon skeleton of pyridoxol is derived from glycerol in a very specific manner: five of the eight carbon atoms of pyridoxol (C-2', C-3, C-4', C-5', and C-6) are derived from the primary carbon atoms of glycerol, and the other three (C-2, C-4, and C-5) from its secondary carbon atom. On the basis of these results we concluded that the CR skeleton of pyridoxol was derived from three glycerol units. We surmised that two glycerol units were incorporated intact, supplying the two C, fragments, C-3/C-4/C-4', and C- 5'/C-5/C-6 of the vitamin; whereas the third glycerol unit loses one of its primary carbon atoms on route to supplying the C, unit, C-2'/C-2, of the vitamin. Experiments with specifically I4C-labeled samples of pyruvate and D-glucose supported this view. Label from [3-I4C] pyruvate entered C-2' and no other site, and all activity from [2-14C]pyruvate was confined to C-2 of pyridoxol. An experiment with intermolecularly doubly labeled [ 1,3-14C2]pymvate showed that the carboxyl group of pyruvate does not enter pyridoxol (2). Furthermore, addition of unlabeled pyruvate to a culture containing [2-14C]glycerol spared the incorporation of label into C-2 of pyridoxol, but did not affect its entry into C-4 or C-5 (6). Thus, the C, unit, C-2'/C-2, of pyridoxol originates from a Cp unit derived from pyruvate by decarboxylation.
Since glycerol itself is not sufficiently reactive, chemically or biochemically, to undergo either the carbon-carbon bondbreaking reaction that leads to the Cp unit serving as the precursor of C-2'/C-2 or the two carbon-carbon bond-making reactions that are required to produce the vitamin skeleton from two C3 fragments and one C p fragment, we surmised further that it was much more likely that the reactive species were the triose intermediates of glycolysis, 3-phosphoglyceraldehyde and dihydroxyacetone l-phosphate, which are generated from glycerol (2).
The normal source of these two triose phosphates is glucose. Glycolytic breakdown of D-[l-'"Cl-and D-[6-'4C]glucose yields triose phosphates labeled at one site only, namely at the terminal carbon atom carrying the phosphate ester. Incorporation of these triose phosphates into pyridoxol would be expected to deliver label into only three sites: C-2', the carbon atom derivable from the methyl group of pyruvate; and two other sites, one in each of the C3 segments, either C-3 or C-4', but not both as was the case with [1-YJ]glycerol, and either C-5' or C-6, but not both. Indeed, activity of pyridoxol derived from D-[l-'4C]ghCOSe (2, 3) or from D-[6-"C]glucose (3) was located at C-2', C-4', and C-5' and at no other site.
This, in summary, is the evidence on the basis of which we proposed that the CB units of pyridoxol, C-3/C-4/C-4' and C-5'/C-5/C-6, are derived from intact triose phosphate generated from glucose by the normal glycolytic sequence and that the Cp unit, C-2'/C-2, of pyridoxol is generated from one such triose phosphate by loss of a terminal carbon (2, 4).
If this interpretation is correct, then only two carboncarbon bonds within pyridoxol, C(2)-C(3) and C(4)-C(5), are generated de nouo in the course of biosynthesis.
All other carbon-carbon bonds of pyridoxol, C(2')-C(2), C(3)-C(4), C(4)-C(4'), C(5')-C(5), and C(5)-C(6), represent bonds already preformed within glucose serving as the precursor. The first objective of the present work was to submit this proposition to critical experimental examination. Furthermore, the sequence from glucose to pyridoxol via triose phosphates does not assign a function to glycerol, a compound which has been shown to serve as a source of all eight carbon atoms of pyridoxol(2,5).
In our biogenetic model we had assumed that glycerol acts as a precursor of pyridoxol because it serves as an alternative source of triose phosphate. The second objective of the present investigation was to test this assumption experimentally.
13C NMR Spectroscopy as a Probe in the Detection of Intact Bonds Transferred from Precursors into Product: Incorporation of [1,2,3,4,5,6-'3CJGlucose into Pyridoxol-The use, in biosynthetic studies, of intramolecularly multiply 13C-labeled substrates which are 13C-enriched at contiguous carbon atoms and the detection, by means of i3C NMR spectroscopy, of incorporation into biosynthetic products of intact fragments of such "bond-labeled" precursors was pioneered by  who were the first to employ intramolecularly doubly i3C-enriched acetate, i.e. sodium [1,2-'3C,]acetate, in the study of polyketide biosynthesis. The 13C NMR spectrum of the product into which an intact precursorderived 13C-13C unit is incorporated exhibits characteristic signals due to the enriched carbon sites. These signals appear as doublets (or higher multiplets if a 13C-13C-'3C unit had entered the product) as a consequence of coupling of contiguous 13C atoms (spin l/z).
We have examined the 13C NMR spectrum of a sample of pyridoxol isolated from cultures of E. coli B mutant WG2 which had been incubated with a mixture consisting of 20% [ 1,2,3,4,5,6-"c~s]glUCOSe (99% l3C per carbon atom) and 80% unenriched glucose (1.1% I3C per carbon atom) as the sole carbon source. Pyridoxol derived from this substrate will yield a "C NMR spectrum in which only those signals will show multiplicity which are due to carbon atoms derived from glucose as part of an intact multicarbon unit.
Only such carbon atoms will give rise to a l3C-I3C coupling pattern. Carbon-carbon bonds formed de novo during the biosynthetic process, from fragments generated by cleavage of a 13C-13C bond within the precursor, will not maintain this coupling pattern since such new bonds will be formed largely by union of two 12C atoms (64%) or one 12C and one "C atom (2 X 16%) and only to an extent of 4% by union of two 13C atoms.
In the "C NMR spectrum of the newly biosynthesized pyridoxol, the 13C-"C spin-coupling pattern associated with each carbon atom will thus indicate whether the carbon originated from glucose as part of an intact multicarbon unit or whether the bond to its direct neighbor had been newly formed during the biosynthetic process. Furthermore, the 13C-13C spin-coupling pattern will indicate the size of each intact multicarbon unit and therefore the number of intact units that had been used to form the pyridoxol skeleton from glucose. Thus, the signals corresponding to carbon atoms that entered as part of an intact two-carbon unit should appear as triplets, composed of a central line (due to natural abundance I3C and, if present, due to enriched 13C attached to 12C) straddled by a doublet caused by I3C-l3C coupling between the enriched carbon atoms of the 13C2 unit. In the same way, carbon atoms that entered pyridoxol as the primary carbons of an intact I3C3 unit are bonded to a single neighbor, i.e. to the central carbon of the unit, and their signals will also appear as triplets. The central carbon of such a unit, being bonded to two primary carbons, will yield a quintet, composed of a doublet on each side of a central line. In turn, more complex but equally predictable signals would be associated with carbon atoms that are derived from interior carbons of larger intact 13C units.
The sample of pyridoxol, whose 13C NMR spectrum was determined (Fig. 3), consisted of a mixture of newly biosynthesized pyridoxol (-80 pg/l-liter culture), derived from E. coli B WG2 which had been incubated in the presence of a 1:4 mixture of [1,2,3,4,5,6-"C&1UCOSe (99% ' ' C per carbon atom) and unenriched glucose (1.1% I3C per carbon atom), and natural abundance pyridoxol (2.5 mg) which had to be added as carrier in order to facilitate isolation and purification of the enriched sample.
From the composition of this mixture, the relative signal areas of the central line, due to the natural abundance component of each signal, and of the multiplet straddling the central line, due to the coupled 13C-enriched component, can be calculated. The calculated value' for the ratio, area of the central line/total area of the outer doublet(s), is 64:36. Thus, the central line of each carbon signal contributes 64%, whereas the outer lines together contribute 36% of the total signal area. For example, each line of the outer doublet associated with the signal from a carbon atom, which is attached to a single neighbor, will contribute 18% to the total signal area; whereas each line of the two outer doublets, which are associated with the signal from a carbon that enters pyridoxol as the central carbon of an intact C3 unit, will constitute 9% of the total signal area.

Biosynthesis of Vitamin Bg from Triose
Units 7469 relative area of 64, within experimental error, straddled by a doublet, total relative area 36, within experimental error (see Table 111). Thus, each of these six carbon atoms shows direct coupling to only one other carbon. The two carbon atoms, C-4 and C-5, on the other hand, must each be the central carbon atom of a C3 unit: the signals due to these two carbon atoms (Fig. 3) consist of a central line, relative area 82, straddled by a doublet, total relative area 18 (Table 111). This indicates that the signal is in fact degenerate, consisting of a central natural abundance line, relative area 64, straddled by a pair of doublets, total relative area 36. The inner lines of the pair of doublets are not resolved from the central line.
Comparison of the measured separation of the outer lines with calculated values confirms this interpretation. The separation of the outer lines, 109.6 Hz for C-4 and 112.8 Hz for C-5 (Table HI) The relatively simple appearance of the 13C NMR spectrum indicates that no carbon fragment larger than C3 is implicated in the biosynthesis of pyridoxol from glucose. From the coupling constants and relative areas it can be inferred that there is biosynthetic connectivity between C-2' and C-2, but not between C-2 and C-3. Thus, the bond C(2)-C(3) is newly formed in the course of biosynthesis. Similarly, each of C-3 and C-4' and each of C-6 and C-5' is shown to have biosynthetic connectivity to a neighbor, C-4 and C-5, respectively. However, there is no measurable coupling between C-4 and C-5 (predicted -65 Hz, cf. 1Jc.3,c.4 and 'Jc.,,c. Table 111) or between C-3 and C-6 (predicted -14 Hz, c.f. pyridine 3Jc.2,c.5 = 14 Hz (30)). Thus, there is no connectivity between C-4 and C-5, and the bond C(4)-C(5) is also newly formed in the course of biosynthesis. These results establish that, in E. coli B, pyridoxol is derived by the combination of a Cz unit, which supplies C-2'/C-2 of the vitamin, and two C3 units, which supply C-3/C-4/C-4' and C-5'/C-5/C-6. These three units are derived intact from glucose. Thus, the inferences that were drawn from earlier tracer experiments (2-6) are confirmed. On the basis of this tracer work with 14C-labeled substrates, we had inferred that two bonds of pyridoxol, C(2)-C(3) and C(4)-C(5), were newly formed in the course of the biosynthesis of pyridoxol from  glucose. The present experiment provides direct experimental support for the earlier inference and shows that C(2)-C(3) and C(4)-C(5) are, indeed, the only carbon-carbon bonds of the pyridoxol skeleton which are formed de nouo in the course of its biosynthesis from glucose.

ality: Incorporation of (R)-and (S)-[l,l-'HziGlycerol into Pyr-
idorol-It has been shown by means of tracer experiments with [1-14C]-, [1,3-13C,]-, and [2-14C]glycerol (2, 5) that all eight carbon atoms of pyridoxol are derived from the carbon atoms of glycerol in nonrandom fashion. Glycerol is a prochiral molecule. As a consequence of this prochirality, there are several modes whereby glycerol can yield the two CB units and the Cz unit which combine to yield the carbon skeleton of pyridoxol. These three units are derived from glucose in such a manner that label from D-[l-14C]glucose and from ~-[6-'~C]ghcose enters C-4' of the C3 unit, C-4'/C-4/C-3, C-5' of the C3 unit, C-5'/C-5/C-6, and C-2' of the C, unit, C-2'/C-2 (2, 3). This mode of incorporation of label, together with the now proven maintenance of the integrity of bonds present within glucose on the route into pyridoxol, is consistent with the intermediacy between glucose and pyridoxol of the normal glycolytic triose phosphate intermediates, 3-phosphoglyceraldehyde and dihydroxyacetone 1-phosphate, whose phosphorylated primary carbon atoms then give rise to C-4' and (2-5' and, via the methyl group of pyruvate, to C-2' of the C, unit, C-2'/C-2. If glycerol entered pyridoxol by way of the same intermediates, its prochiral primary carbon atoms must be incorporated into pyridoxol in a specific and predictable manner. This supposition can be tested by means of incorporation experimentswith (S)-[l,l-'H2]glycero1 and (R)-[l,l-2H2]glycerol. Samples of these compounds were prepared by standard methods (see "Materials and Methods").
In order to confirm that experiments with these chiral samples of deuterium-labeled glycerol would yield unequivocal results, it was necessary to determine which of the hydrogen atoms of pyridoxol are derived directly from glycerol. It is clear that only seven of the eight hydrogens, at C-2', C-4', C-5', and C-6 of the vitamin, can be so derived since the eighth, one of the three methyl protons at C-2', must have entered in the course of the conversion of a CH20H group of glycerol into the CH3 group of pyruvate. Furthermore, the results of a tracer experiment with [2-14C, 1,3-3H]glycerol as the precursor (6) showed that one of these seven protons is lost in the course of the biosynthesis of the vitamin. The sample of [2-14C, 1,3-3H]glycerol that was administered to the E. coli B WG2 culture had a 3H/14C ratio of 15.4 & 0.4. Retention within pyridoxol of seven tritium atoms/three glycerol units, the theoretical maximum, should have yielded a sample of pyridoxol, 3H/14C = 9.0. The observed ratio, 3H/14C = 7.7, indicated that one of these seven tritium atoms had not been incorporated into the product (6).
Degradation of the pyridoxol sample ('H/14C = 7.7) by the reaction sequence shown in Scheme 2 established the site of the missing tritium. 6-Chlorotri-0-acetylpyridoxol (17) was obtained from the pyridoxol sample by way of tri-0-acetylpyridoxol 1-oxide (16). 3,4'-O-Isopropylidenepyridoxol (14) was also prepared (2). These three derivatives showed the same 3H specific activity, the same I4C specific activity, and therefore the same 'H/14C ratio as the pyridoxol hydrochloride (13) from which they were obtained (Table 11). Since 6chlorotri-0-acetylpyridoxol (17) showed the same 3H/14C ratio as the other three compounds, even though it lacks the proton at C-6 that is present within the other compounds, C-6 of the pyridoxol generated from [2-14C, 1,3-3H]glycerol must have been the site into which tritium of the precursor had not been incorporated in the course of biosynthesis. Another degradation, Kuhn-Roth oxidation (2), that localized the tritium label at C-2' (Scheme 2) further substantiated this conclusion. These results, which are summarized in Table II, prepared the ground for the experiments with (R)-[l,l-*Hz]and (S)-[l,l-*HJglycerol.
In two separate experiments, cultures of E. cob B WG2 were incubated in the presence of (R)-[ l,l-2H2]glycerol (1 g/ liter) (Experiment 2) and (S)-[l,l-'H2]glycerol (1 g/liter) (Experiment 3). Tracer amounts of [2-%]glycerol (250 &i/ liter) were also added to each incubation. Pyridoxol hydrochloride was isolated from each culture after carrier dilution and a derivative, isopropylidenepyridoxol, was prepared in each case. The samples were radioactive. The samples from the experiment with (R)-glycerol (Experiment 2) had specific activities of 7.5 X lo5 dpm/mmol (pyridoxol hydrochloride) and 7.4 x lo" dpm/mmol (isopropylidenepyridoxol hydrochloride), respectively. The corresponding samples from (S)-glycerol (Experiment 3) had specific activities of 3.0 x lo5 and 3.1 x lo5 dpm/mmol, respectively. Each of the two samples of pyridoxol hydrochloride was dissolved in methanol (-50 ~1) and examined for deuterium content by 'H NMR spectroscopy. The spectra of the two samples (Fig. 4, B and C) were different.
The spectrum in methanol of the sample of pyridoxol hydrochloride obtained from the incubation with (R)-deuterioglycerol ( Fig. 4B ) showed, after only 600 scans, three signals in addition to the natural abundance signals of the solvent (3.30 ppm, CDH,; 4.95 ppm, OD) (Fig. 40). The three signals, of equal intensity, at 2.54, 4.65, and 5.00 ppm, are assigned to deuterium at C-2', C-5', and C-4' of the vitamin, respectively. These carbon atoms are therefore derived from the re-hydroxymethyl carbon atom of glycerol, which is incorporated into three sites of the vitamin with retention of its proton pair.
Even after 28,000 scans, the spectrum (Fig. 4C) of the pyridoxol obtained from the incubation with (S)-deuterioglycerol did not differ from that of the solvent (Fig. 40). Two signals, at 3.30 and 4.95 ppm, i.e. the natural abundance signals of the solvent, were apparent and were the only signals in both spectra. Thus, no deuterium, above natural abundance, was present within the pyridoxol that had been isolated from the incubation with (S)-[l,l-2H~]glycerol. Since this sample was radioactive (3.0 X lo5 dpm/mmol, see above), biosynthesis of pyridoxol had occurred in this incubation. Since no deuterium was present, it follows that C-2', C-4', and C-5', the three pyridoxol carbon atoms derived from the re-hydroxymethyl carbon of glycerol, are not derived from the si-hydroxymethyl carbon of glycerol. Thus, glycerol enters the pyridoxol skeleton in a stereospecific manner. C-3 and C-6 of the vitamin, the two other carbon atoms which are supplied by a primary carbon of glycerol, must be supplied by the sihydroxymethyl carbon atom of glycerol. Glycerol-derived deuterium does not enter these two sites because one of them (C-3) does not carry hydrogen, and the other (C-6) was shown, by means of the experiment with [a-%, 1,3-3H]glycerol (see above), not to retain glycerol-derived hydrogen in the course of biosynthesis. CONCLUSION On the basis of the results here reported, the mode of incorporation of C3 units, derived from glucose and from glycerol, into the two C, segments, C-3/C-4/C-4' and C-5'/ C-S/C-S, of pyridoxol becomes clear.
The experiment with D-[ 1,2,3,4,5,6-13CG]glucose (Experiment 1) shows that each of the two C, segments is derived intact from three contiguous carbon atoms of glucose. The results of the earlier incorporation experiments with D-[l-"C]glucose (2) and D-[6-'4C]glucose (3), together with the present result, show that these contiguous carbon atoms in fact represent C-l/C-2/C-3 and C-6/C-5/C-4 of glucose, i.e. the two glycolytic fragments, 3-phosphoglyceraldehyde and dihydroxyacetone l-phosphate, and furthermore, that it is Cl and C-6 of glucose, i.e. the carbon atoms corresponding to the phosphorylated hydroxymethyl group in each of the two trioses, which enter the peripheral carbons, C-4' and C-5', within the two C3 segments of the vitamin.
These two carbon atoms, C-4' and C-5', are supplied by the re-hydroxymethyl group of glycerol when prochirally deuterated glycerol serves as the substrate, whereas the other termini of the two C3 segments, C-3 and C-6, are supplied by the si-hydroxymethyl carbon (Experiments 2 and 3).
It is known from the classical studies of Karnovsky et al. (31) that it is the re-hydroxymethyl group of glycerol which is phosphorylated in the course of metabolism and that subsequent biochemical elaboration of the resulting L-a-glycerol phosphate yields dihydroxyacetone l-phosphate and 3-phosphoglyceraldehyde in which the phosphorylated hydroxymethyl group is in each case derived stereospecifically from the re-hydroxymethyl group of glycerol, whereas the other terminal carbon atom originates stereospecifically from the si-hydroxymethyl group of glycerol.
The observed stereochemistry in the stereospecific incorporation of label from prochirally labeled samples of glycerol into the C3 segments of pyridoxol is thus entirely consistent with the observed regiochemistry in the regiospecific incorporation of label from specifically labeled samples of glucose.
Similar consistency is observed with respect to the origin of the C, segment, C-2'/C-2, of pyridoxol. This moiety arises from a C, unit derived from pyruvate by decarboxylation. The methyl group of pyruvate is known to arise either from C-l and C-6 of glucose or from the re-hydroxymethyl group of glycerol. This is the incorporation pattern into C-2' of pyridoxol that is indeed observed.