Biosynthesis of vitamin B6. Incorporation of glycolaldehyde into pyridoxal

摘要

1622 J.C.S. Perkin IBiosynthesis of Vitamin B6. Incorporation of Glycolaldehyde into Pyri-doxalBy Robert E. Hill, Peter Horsewood, and Ian D. Spenser,rdquo; Department of Chemistry, McMaster University,Hamilton, Ontario L85 4M1, Canada(in part) Yoshiki Tani, Department of Agricultural Chemistry, Kyoto University, Kyoto, JapanRadioactivity from 14Cglycolaldehyde enters pyridoxal phosphate specifically, and is confined to the two-carbonunit C(5)-C(5rsquo;). Glycolaldehyde enters as an intact two-carbon unit, the aldehyde carbon atom supplying C-5,and the carbinol carbon atom C-5lsquo; of the vitamin. These observations are interpreted in terms of the early stagesof the biosynthesis of vitamin B6.WE have demonstrated l-3 that radioactivity fromspecifically labelled D-glUCOSe and glycerol is incorpor-ated non-randomly into pyridoxol, biosynthesized byEscherichia coli B mutant WG2.The data are con-sistent with the hypothesis that the eight-carbonskeleton of pyridoxol is generated from three precursorunits. Two of these, giving rise to the C, fragments ofpyridoxol, C(4rsquo;)-C(4)-C(3) and C(5rsquo;)-C(5)-C(6), arethree-carbon compounds closely related to triose phos-phate, whereas the third, which yields the C, fragment ofpyridoxol, C(2rsquo;)-C(Z), is a two-carbon compoundderivable from glycolytic intermediates or from pyruvate.We inferred1*, on mechanistic grounds that this two-carbon compound was an aldehyde, and suggested thatit might be at the oxidation level of acetaldehyde3 orof gly~olaldehyde.~The notion that glycolaldehyde might be implicatedin the biosynthesis of vitamin B6 was first mooted5 onthe basis of the finding6 that this compound replacedvitamin B, in promoting the growth of two B,-requiringstrains of E.coli. Recently Dempsey characterizedan E. coli B mutant, WG3, with a nutritional require-ment for pyridoxol which was satisfied by glycol-aldehyde,lsquo; and isolated radioactive samples of thevitamin from cultures of the mutant which had beenincubated with 1,2-14C2- and with 2-14C-glycol-aldehyde.* Glycolaldehyde thus apparently served as aprecursor of vitamin B,. Degradation of the labelledsamples of pyridoxol obtained from these experimentsshowed that, contrary to prediction, little, if any,radioactivity was present at the C, fragment, C(2rsquo;)-C(2),of these samples.4 This indicated non-random distri-bution of label, but the location of labelling was notdetermined.We now provide evidence that glycol-aldehyde is incorporated as an intact unit into the C,fragment, C(Srsquo;)-C(5), of pyridoxol and that C-5rsquo; isspecifically derived from the carbinol carbon atom andC-5 from the carbonyl carbon atom of glycolaldehyde.RESULTS AND DISCUSSIONRadioactive samples of pyridoxol hydrochloride,obtained by Tani and Dempsey from pyridoxal phos-phate which they had isolated from cultures of E. coli Bstrain WG3 after incubation with l ,2-14CJglycol-aldehyde (ref. 8, expts. 111-6, 111-9) or with 2-14Cglycol-aldehyde (ref. 8, expts. 111-5, III-S), were diluted withinactive carrier and degraded, by the reactions shownin Scheme 1, to locate the sites of labelling.The specificactivities of the degradation products are listed inTable 1.Within experimental error, all activity of the pyridoxolhydrochloride derived from 2-14Cglycolaldehyde t wasThis sample, prepared by oxidative decarboxylation of recovered in benzoic acid, representing C-5rsquo; of pyridoxol.3-14Cserine, was assumed to be labelled solely at the carbinolcarbon atom on the basis of its mode of preparation, but rigorousevidence proving the position of 1 was not secured.* 4 R. E. ill and 1. D. spenser, Canad. J . ~ i ~ ~ h ~ ~ . , 1973, 51,6 J. G. Morris, J . Gen. Microbial., 1969, 20, 697.6 J. G. Morris and D. D. Woods, J . Gen. Micvobiol., 1969, 20,7 W.B. Dempsey, J . Bucteriol., 1971, 108, 1001.8 Y . Tani and W. B. Dempsey, J . Bucteriol., 1973, 116, 341.1412.R. E. Hill and I. D. Spenser, Science, 1970, 169, 773.R. E. Hill, R. N. Gupta, F. J. Rowell, and I. D. Spenser,8 R. E. Hill, F. J. Rowell, R. N. Gupta, and I. D. Spenser,J . Amer. Chem. Soc., 1971, 93, 618.J . Biol. Chem., 1972,247, 1869.6761975 1623The benzoic acid, i.e. C-5rsquo;, from the pyridoxol derivedfrom 1 ,2-14C,amp;lycolaldehyde,* on the other hand,I 0 I lK+ -N*05 rsquo;H 0 2 C( Y ) 0SCHEME 1 Chemical degradation of pyridoxol, permittingisolation of C-5rsquo; and C-6contained only half the label of the intact vitamin. Theother half must be located at C-5, since the phthaloyl-glycine (i.e. C-5 plus C-5rsquo;) obtained from this sample ofpyridoxol accounted for all its activity.It follows thatglycolaldehyde enters the two-carbon unit, C(5rsquo;)-C(5), ofpyridoxol as an intact unit, the aldehyde carbon atom(i.e. C-1) supplying C-5, and the carbinol carbon atom(i.e. C-2) C-5rsquo; of pyridoxol.?* This was prepared by oxidative decarboxylation of U-14Cserine, and was assumed to have activity equally distributedbetween the carbinol and the carbonyl carbon atoms. Rigorousproof for equal distribution of label was not obtained, however.8 t These results also show that the two carbon atoms of glycol-aldehyde maintain their chemical indivlduality in the course ofmetabolism, i.e. that equilibration of label, by way of enolization(i) (ii) (iii), does not take place.HO-CH,-CH=W-HO-CH=CH-OH~O=CH-CH2-OH(0 (ii) (iii)Growth of E.coli B WG3, a pyridoxol-requiringmutant, can be maintained only if minimal medium,containing either 0.2 glucose 739 or 0.2 glycerol * asgeneral carbon source, is supplemented by pyridoxol, orby gly~olaldehyde.~ It follows that this mutant cannotbiosynthesize glycolaldehyde from glucose or fromglycerol. Since specific incorporation of glycolaldehydeinto pyridoxal phosphate is now demonstrated, and it isshown that a single glycolaldehyde unit enters thepyridoxal molecule, it follows that the samples ofpyridoxal phosphate, produced by E. coli B WG3cultures on incubation with labelled glycolaldehyde,must show a molar specific radioactivity (mCi mmol-l)identical with that of the samples of 14Cglyc~laldehydefrom which they are derived.That is, the specificradiochemical yield (= 100 x molar specific radioactivityof product/molar specific radioactivity of precursor)must be 100. Published data8 tend to bear out thisprediction. $Failure to recognize that in these experimentsmaintenance within the product of the molar specificactivity of the precursor was a consequence of the experi-mental conditions prompted the conclusion in a recentreview lo that whereas lsquo; 14Cglycolaldehyde served as ahighly efficient precursor of labelled pyridoxol,rsquo; theimportance of glycerol and glucose in pyridoxol bio-synthesis was questionable, since in experiments with14C-labelled samples of these substrates lsquo; the total in-corporation $ of radioactivity into pyridoxol wasslight .rsquo; loBefore attempting an interpretation of the mode ofincorporation of glycolaldehyde into vitamin B,, it isnecessary to correct the impression conveyed by thismisleading comparison.Non-random incorporation of radioactivity fromspecifically labelled radiomers of glycerol into pyridoxolwas demonstrated by chemical degradati~n-l-~ It wasshown that approximately one third of the molarspecific activity of pyridoxol derived from 2-14C-glycerol was located at each of C-2 and C-4.Theremaining third was predicted to reside at C-5 ofpyridoxol. This prediction is now confirmed. Whereasbenzoic acid (i.e. C-5rsquo;) obtained from 2-14Cglycerol-derived pyridoxol was devoid of radioactivity, phthaloyl-glycine (i.e.C-5rsquo; plus C-5) accounts for one-third of its$ The apparent discrepancies in the molar specific activities ofglycolaldehyde and of pyridoxal phosphate observed in severalexperiments (Table 2 of ref. 8) (specific radiochemical yieldsreported for four experiments with 2-14Cglycolaldehyde 80, 100,90, 50 and for three experiments with l, 2-14C,glycolaldehyde114, 86, 66) are a measure of the accuracy of the radioactivitydeterminations on which these data are based.The two low values (50 and 66) were obtained in cultures ofWG3 which had undergone partial reversion to wild type.8The reduction in the specific radiochemical yields in these experi-ments indicates that, in wild type cultures, 14Cglycolaldehyde orone of its metabolites encounters a corresponding unlabelled poolprior to incorporation into the product.9 i.e.the percentage incorporation (= 100 x total activityrecovered in product/total activity administered in precursor).9 W. B. Dempsey, J . Bacteriol., 1969, 100, 295.lo G. W. E. Plaut, C. M. Smith, and W. L. Alworth, Ann. Rev.Biochem., 1974, 43, 8991624label, which is therefore located at C-5, as predicted(Table 2).Similarly, it is now shown that C-6 of the pyridoxolderived from 1-14Cglycerol is devoid of activityJ.C.S. Perkin Iand C(5rsquo;)(5)(6) are incorporated intact, The third,giving rise to the C, unit, C(2lsquo;)-C(2), suffers loss of aterminal carbon atom.Since specific incorporation of glycerol into pyridoxolTABLE 1Incorporation of glycolaldehyde into pyridoxolSubstrater 2-1PCGlycolaldehyde 1,2-14C.JGlycolaldehydeL c r t Products C Atoms of Pyridoxol SA RSA SA 0 RSAPyridoxol hydrochloride (I) All 1.18 f 0.04 0 100 amp; 3 1.11 f 0.03d 100 f 2Isopropylidenepyridoxol (11) All 1.20 f 0.04 102 f 4Isopropylideneisopyridoxal (VI) All 1.23 f 0.03 106 f 4 1.07 f 0.03 97 f 4Isopropylidene-6rsquo;-phenylpyridoxol (VII) All 1.16 f 0.04 99 f 4 1.11 f 0.04 looamp; 4Benzoic acid C-6rsquo; 1.11 f 0.02 94f 3 0.68 f 0.01 62 f 2Isopyropylidenep yridoxol (11) AllIsopropylidenephthaloylisopyridoxamine AllPhthaloylglycine (V)Phthaloylglycine minus benzoic acidC-6rsquo;.-6(Iv)C-6 (by difference)0.96 f 0.03 * 100 5 30.91 f 0.06 9 6 k 60.91 f 0.03 97 * 446 f 6Pyridoxol hydrochloride All 0.69 f 0.02f*h 100 f 3 0.76 f 0.03 v p A 100 f 3Isoprop ylidenep yridoxol All 0.70 f 0.02 101 f 4K/R acetate from pyridoxol c-2lsquo;, 2 0.01 amp; 0.003 2 amp; 0.4 0.02 f 0.02 3 f 3K/R acetate from 4lsquo;-deoxypyridoxol C-2rsquo;, -2, -4!, -4 0.001 f 0.01 0.1 f 24rsquo;-Deoxypyridoxol All 0.66 f 0.03 96 amp; 6Specific activity (disint.min-l mmol-1) x b Relative specific activity () (pyridoxol hydrochloride = 100). Obtainedfrom pyridoxd phosphate (expt. 111-8, ref. 8) by conversion into pyridoxol hydrochloride Is and dilution with inactive carrier.d As c, but expt. 111-9, ref. 8. 6 As c. but expt. 111-9. ref. 8, and followed by conversion to the isopropylidene derivative. f As c,but expt. 111-6, ref. 8. As c, but-expt. 111-6, ref. 8. Cf. ref.4.TABLE 2Degradation t o locate label at C-5 of pyridoxol derived from l-14Cglycerol and from 2-*4CglycerolSubstrateA t *f I -l l-14CGlycerol 2-W GlycerolA AProducts C Atoms of Pyridoxol SA RSA SA 0 RSAPyridoxol hydrochloride (I) All 6.16 f 0.04 0 100 f 1 3.16 f 0.02 100 f 1Isoprop ylidenephthalo ylisop yridoxamine All 6.00 f 0.04 97f 1 3.04 f 0.02 96 f 1C-6rsquo;, -5 1.02 f 0.01 20 f 0.3 1.07 f 0.01 34 f 0.3 Phthaloylglycine (V)Benzoic acid C-6rsquo; 22f 1 6 0.6 f. 0.2Phthaloylglycine minus benzoic acid C-6 (by difference) Inactive 33 f 0.4Isopropylidenepyridoxol (11) All 6.03 f 0.04 98 f 1 3.21 f 0.03 102 amp; 1(IV)As Table 1. Portion of the sample of pyridoxol hydrocyloride isolated in expt. 16, ref. 3. As c, but expt. 16, ref.3.Quoted from Table IV of ref. 3.(Table 2). This is also according to prediction, whichplaced one-fifth of the molar specific activity of pyridoxolderived from l-14Cglycerol at each of C-2lsquo;, -3, -4rsquo;, -5rsquo;,and -6. It is thus established1-3 that one-fifth of themolar specific activity does indeed reside at each ofC-2rsquo;, -4rsquo;, and -5rsquo;. It is also established that C-2, -4,and -5 are free of activity. This leaves two carbonatoms, C-3 and C-6, to accommodate two-fifths of themolar specific activity. It is a plausible assumptionthat each of these two carbon atoms, like the other threelabelled centres, contains one-fifth of the total label.These conclusions are summarized in Scheme 2.The eight carbon atoms of pyridoxol are thusaccounted for by three glycerol units which are incorpor-ated non-randomly into the vitamin.Two of theseglycerol units, giving rise to the C, units C(4lsquo;)-euro;(4)(3)is demonstrated and it is shown that three glycerol units=-one terminal carbon atom) enter the pyridoxolLlsquo;A CH2OH2rsquo;SCHEME 2 Mode of incorporation of glycerol into pyridoxol; sitesof activity derived from l-14Cglycerol (A, A) relative specificactivity ca. 20) and from 2-14Cglycerol 0 ) relative specificactivity ca. 33) shown by degradation (A 0 ) or inferred (A)molecule, and account for all its carbon atoms, it followsthat a sample of pyridoxol synthesized by E . coli B WG1975 1625cultures on incubation with 2-14Cglycerol, in thepresence of glycerol as the sole general carbon source,must show a molar specific radioactivity (mCi mmol-l)three times that of the labelled precursor, i.e.thespecific radiochemical yield must be 300. It followsalso that a sample of pyridoxol produced similarly onincubation with chemically labelled l-14Cglycerol (k.an equimolar mixture of sn-l-14C- and sn-3-14C-glycerol) must show a molar specific radioactivity 2.5times that of the labelled precursor, i.e. a specific radio-chemical yield of 250. Since the distribution patternof label within pyridoxol derived from 14Cglucose 394 isnot as simple as that within glycerol-derived pyridoxol,the specific radiochemical yield cannot be predicted onthe basis of existing evidence. Nor is there a simplerelationship between molar specific activity of precursorand product when labelled precursor ( e g .14Cglycerol)has to compete for incorporation into product withanother, unlabelled substrate (e.g. glucose) which servesas the general carbon source and is present in large excess.Our experiments, in which 1-14C- and 2-14C-glycerolwere tested as a precursor of pyridoxol in cultures ofE. coli B WG2, with glycerol as the sole carbon source(ref. 3, expts. 2, 15, IS), were designed solely with aview to establishing the distribution of radioactivitywithin the newly synthesized pyridoxol. To achievethis purpose it was not essential to know the amount ofpyridoxol formed or its specific radioactivity, and thesevalues were not determined. Precise data to confirmthe above predictions that glycerol-derived pyridoxolshould have a molar specific activity 3 times or 2.5 timesthat of the precursor, are therefore lacking.However, from data which are available it can becalculated,* that the predicted specific radiochemicalyield of 300 in the experiment with 2-14Cglycerol(expt.16) and of 250 in those with l-14Cglycerol(expts. 2 and 15) would obtain if 60, 75, and 72 pg,respectively, of pyridoxol had been synthesized de novoin the course of these experiments. These values areindeed within the expected order of magnitude, 60-180 pg 1-1 of bacterial culture.11J2It is evident from the foregoing discussion that acomparison of specific radiochemical yields obtained inexperiments with glycolaldehyde and with glycerol doesnot serve as a fruitful basis on which conclusions regard-ing their position in the biosynthetic pathway leadingto pyridoxol can be drawn.It remains to examine whether a comparison ofpercent incorporation of radioactivity in the variousexperiments leads to useful conclusions.For nineexperiments with labelled glycerol ,, glucose ,33 and* E.g. for expt. 16, specific activity (nominal) of 2-14C-glycerol : 0.091 mCi mmol-l; weight of pyridoxol added ascarrier 1000 mg; specific activity of pyridoxol isolated aftercarrier dilution 1.09 x lo" counts min-l mmol-1; efficiency ofcounting system ca. 30; 1 mCi = 2.22 x lo9 disint. min-l;weight (mg) of newly synthesized pyridoxol. of specific activity3 x 0.091 mCi mmol-l:= 0.06.1-09 x lo4 : 1000 x - 100 I3 x 0.091 30 2.22 x 109pyruvate the percent incorporation (100 x totalactivity recovered within pyridoxol ?/total activity inprecursor added to the medium) ranged from 2 xto 8 x 10".For six experiments with labelled glycolaldehydethe percent incorporation ranged from 3.5 x lo9 to9.5 x lo-,.One experiment gave an exceptionally highyield of product and thus an exceptionally high value of1.6 xThe recovery of radioactivity within the products ofthe two series of experiments is thus of the same loworder of magnitude. This is hardly surprising. Ineach case the molar amount of product formed isseveral orders of magnitude lower than the molaramount of labelled substrate which is available. In theexperiments with E.coli B WG2 ca. 0.3 pmol ofpyridoxol per litre was formed whereas ca. 5 mmol ofsubstrate per litre (glucose, expt. 14), ca. 11 mmol ofsubstrate per litre (glycerol, expts. 15, 16), and ca. 22mmol of substrate per litre (glycerol, expt. 2), wereavailable. In the experiments with E. coli B WG3,8ca. 0.03 pmol of pyridoxal per litre was formed whereasca. 0.4 mmol of glycolaldehyde per litre was available.Neither percent incorporation nor specific radio-chemical yield thus serves as a guide to precursor statusin this instance. The only reliable guide is distributionof label within the product.We have reported the pattern of labelling withinpyridoxol generated from a number of 14C-labelledsubstrates.l+ This established the distribution andaccounted for all the radioactivity within samples ofpyridoxol derived from 2-14C- and 3-14C-pyruvateand from 1-14C- and 6-14C-glucose. Our publishedwork, together with results reported here, similarlyclarify the quantitative distribution of label withinpyridoxol derived from 2-14Cglycerol and, with aplausible assumption (see above), from l-14Cglycerol.These results are entirely consistent with the hypo-thesis which is referred to in the Introduction.Thenon-random and specific incorporation of glycolaldehydeinto pyridoxal, now demonstrated, must be reconcilablewith this hypothesis. Any attempt to interpret therole of glycolaldehyde in pyridoxol biosynthesis mustbe tentative, because of the dearth of knowledge con-cerning the metabolism of this simple two-carboncompound.In attempting to account for the position of glycol-aldehyde in pyridoxal biosynthesis (Scheme 3), relevantavailable evidence must be reviewed.In E .coli mutant WG2 the eight carbon atoms ofpyridoxol are derived from two C, precursors and one C,precursor. All three are generated from glycolyticintermediates, presumably trioses, related to glucosefor the percent incorporation.t Cf. footnote t and ref. 3, Table 1: total activity recoveredwithin pyridoxol = total activity (counts min-l) recovered per100 1 0.1 mCi of substrate x - xl1 W. B. Dempsey, J . Bacteriol., 1966, 92, 333.l2 W. B. Dempsey, J . Bacteriol., 1967, 95, 1179.30 2.22 x 1091626 J.C.S. Perkin Iand glycerol.That the three precursors are closelyrelated to each other is shown by the observation that,when derived from 2-14Cglycerol, each of the fragmentswithin pyridoxol, corresponding to the three precursors,contained the same molar specific activity. That thethree precursors are all different is shown by the facts(i) that pyruvate supplies the C, unit, C(2rsquo;)-C(2) ofpyridoxol, but neither of the C, units, C(4rsquo;)(4)-C(3)and C(5rsquo;)-C(5)-C(6), and (ii) that these two C, unitsshow different molar specific activities when 14Cglucoseis the labelled substrate. Finally, the C, precursor is-5, then leads to the inference that, even though glycol-aldehyde metabolism excludes its conversion intoglycolytic intermediates, it leads to a C, unit whichserves as the precursor of the pyridoxal unit C(5rsquo;)-C(5)-C(6).This chain extension might involve a one-carbonunit, but this is unlikely since one-carbon donors do notparticipate in pyridoxol biosynthesis in E. coli B WGZ3A possible pathway for such a chain elongation, re-presented in general terms as glycolaldehyde + C, +C, + Cn-l, might proceed by a three-step sequence,involving a ketolase, a kinase, and an aldolase reaction.t r i o s eC3 precursor pyruvate C3 precursorof u n i t o f u n i tC ( 4 lsquo; 1- C ( 4 1 -C(3) C(5rsquo;)-C(5)-C(61C lC 2 Profc(24ec u rsorlsquo;)-C( 2 1u n i t ______9*JJtlsquo; Cn-1 CnC H 2 O HICHOSCHEME 3 Route of precursor fragments into pyridoxol; A and B denote possible sites of the genetic glock in E.coli B mutant WG3the arrows showing the paths from the C , and C , precursors are intended to indicate the sites of entry of these units; it is notimplied that these units are the immediate precursors of pyridoxol (cf. ref. 3)more closely related to the precursor of the C, unitC(4rsquo;)-C(4)-C(3), than to that of the C,unit C(Srsquo;)-C(5-C(6)since, in two samples of glucose-derived pyridoxol themolar specific activity of the C, unit was identical withthat of the C, unit C(4rsquo;)-C(4)-C(3) but different fromthat of C(5rsquo;)-C(5)-C(6).It is two carbon atoms of the precursor of this C,unit, C(5rsquo;)-C(5)-C(6), which in E. coli B mutant WG3are generated from glycolaldehyde. Since glycerolserved as the general carbon source in these experimentsit must have supplied the other six carbon atoms ofpyridoxal.The mode of entry of glycerol carbon intothese sites has not yet been investigated. For purposesof the present discussion we will assume, in the absenceof evidence to the contrary, that the origin of the C,unit C(Brsquo;)-C(2) and the C,-unit C(4rsquo;)-C(4)-C(3) of thepyridoxal generated in mutant WG3 follows the patterndemonstrated for the origin of pyridoxol in mutant WG2.The mode of incorporation of glycolaldehyde intopyridoxal, exclusively into two carbon atoms, C-5lsquo; andMulder l3 has reported the production of erythrulose incultures of E. coli mutant B 166 growing on a glycol-aldehyde-enriched medium.The entry of glycolaldehyde into the precursor of theC, unit C(5rsquo;)-C(5)-C(6) may be dictated by the natureof the genetic block in mutant WG3.If the directroute to this precursor from glycerol and the trioseintermediates of glycolysis were blocked in WG3(Scheme 3, block A), an irreversible route via glycol-aldehyde may be induced, and in the absence of adequatede novo biosynthesis of glycolaldehyde, this compoundmust be supplied to support pyridoxal biosynthesis andgrowth of the mutant.Alternatively, glycolaldehyde may be a mandatoryintermediate on the route to the precursor of the C,unit C(5rsquo;)-C(5)-C(6) whose formation from glycolyticintermediates, in an irreversible step, has been blockedin the mutant (Scheme 3, block B).l3 C. Mulder, D.Phi1. Thesis, University of Oxford, 1959;personal communication1975 1627It may be significant in this context that reduction ofglycolic acid to glycolaldehyde, a process which occursin wild type E.coli, is blocked in mutant WG3.14These possibilities are open to independent experi-mental test. Thus, if glycolaldehyde were a mandatoryintermediate, its addition to the culture medium ofE. coZi B mutant WG2 should lead to a change in thedistribution of label in pyridoxol derived from 2-14C-glycerol. Incorporation of label into C-5 should besuppressed, leading, in the limit, to a sample of pyridoxolcontaining 50 of its molar specific activity at each ofC-2 and C-4, and 0 at C-5. If incorporation ofglycolaldehyde into pyridoxal in mutant WG3 ismutation-induced, addition of carrier glycolaldehyde tothe culture medium of WG2 would leave the distributionof label in pyridoxol, derived from 2-14Cglycerol,unchanged, with 33 of the molar specific activity ateach of C-2, C-4, and C-5.This, and other experiments designed to clarify thefunction of glycolaldehyde in pyridoxal biosynthesis areunder way.EXPERIMENTALThe conditions for bacterial growth, and the proceduresfor the isolation of pyridoxol hydrochloride from E.coli Bstrain WG3 after incubation with 14Cglycolaldehyde 8* l5and from E. coli B strain WG2 after incubation withglycerol have been described, as have the methodsemployed in the systematic degradation of l4C-labelledsamples of pyridoxol, permitting assay of radioactivity a tindividual carbon atoms (C-2', C-2, C-4, C-4', C-5').3 Anew method, permitting determination of label a t C-5 ofpyridoxol, is now reported.Radioactivity was assayed byliquid scintillation counting (Mark 1 liquid scintillationcomputer, model 6860 Nuclear Chicago). Samples weredissolved in water or methanol and dispersed in Aquasol(New England Nuclear). Triplicate samples of eachcompound were counted under comparable conditions ofquenching. Confidence limits shown in the Tables arestandard deviations of the mean.Isolation of Carbon Atoms 5 and 5' of Pyridoxol asPhthaloylglycine (Scheme 1) .-5-Chloromethyl-2,2,8-trimethyl-4H-m-dioxino 4,5-cpyridine hydrochloride (111) Thionylchloride (0.055 ml) was added to a cold (0 "C) stirredsuspension of 3,4'-O-isopropylidenepyridoxol hydrochloride(11) (70 mg) in chloroform (1.0 ml).A clear solutionwas obtained after stirring for ca. 5 min a t room tem-perature. Stirring was continued for a further 1 h. Whenl4 Y. Tani, H. Morita, and K. Ogata, Agric. and Biol. Chem.(Japan), 1974, 38, 2057.l5 W. B. Dempsey, Biochim. Biophys. Acta, 1972, 284, 344.the solution was evaporated under reduced pressure, thehydrochloride (111) was obtained as a white crystallinesolid (73 mg, 98), m.p. 186-189" (1it.,l6 191-192"). Toobtain the corresponding free base the hydrochloride,dissolved in water, was basified with solid sodium hydrogencarbonate and extracted with ether (3 x 8 ml). When theextract was dried (Na,SO,) and evaporated the productwas obtained as a pale yellow oil.2,2,8- Trimethyl- 5-phthalimidomethyl-4H-m-dioxin0 4,5-c -pyridine (isopropylidenephthaloylisopyridoxamine) (IV) .The free base obtained from (111) (73 mg) was dissolved indry dimethylformamide (1 ml) .N-Potassiophthalimide(60 mg) was added and the solution was heated withstirring a t 130-140 "C for 2 h and then allowed to cool toroom temperature. When water (10 ml) was added, andthe solution stirred a t room temperature, the phthalimido-derivative (IV) crystallized (75 mg, 80) ; m.p. 140-142".Recrystallization from 95 ethanol gave a sample of m.p.147-148" (Found: C, 67.7; H, 5.35; N, 8.25. C,,H,,N,O,requires C , 67.45; H, 5.35; N, 8.3); vmx. 1777 and1715 cm-l; 6 (CDC1,) 1.55 (6 H, s), 2.37 (3 H, s), 4.67(2 H, s), 5.08 (2 H, s), 7.75 (4 H, m), and 8.11 (1 H, s).Phthaloylglycine (V) by Oxidation of the Phthalimido-derivative (IV). The phthalimido-derivative (IV) (65 mg)was dissolved in dilute sulphuric acid (3 ml; 1 ~ ) ; thesolution was heated on a steam-bath for 20 min and thenallowed to cool to room temperature. Potassium per-manganate (150 mg) was added in small portions withstirring over 1 h, and stirring was then continued for afurther 1 h. Precipitated manganese dioxide was dissolvedby addition of sodium hydrogen sulphite and the resultingturbid solution was extracted with ethyl acetate (3 x 8 ml).The extract was dried (Na,SO,) and evaporated underreduced pressure to give a cream-coloured residue (40 mg).This was dissolved in saturated aqueous sodium hydrogencarbonate (1 ml), the solution was washed with chloroform(2 x 1 ml), and the aqueous layer was acidified with a fewdrops of 6~-hydrochloric acid. The phthaloylglycine(21 mg, 53) which crystallized was sublimed at lo-, mm Hgand 120 "C; m.p. 193-194" (lit.,l7 191-192"). The i.r.spectrum was identical with that of an authentic sampleobtained by fusion of glycine with phthalic anhydride.17We are indebted to Dr. W. B. Dempsey for makingavailable to us labelled samples of pyridoxol from experi-ments with 14Cglyc~laldehyde. This work was supportedby grants from the Ministry of Health, Province of Ontario(to R. E. H.) and the National Research Council of Canada(to I. D. S.).5/104 Received, 17th January, 19751l6 I. Tomita, H. G. Brooks, and D. E. Metzler, J . Heterocyclicl7 J. H. Billman and W. F. Harting, J . Amer. Chem. SOC., 1948,Chem., 1966, 3, 178.70, 1473