Biosynthesis of Riboflavin - Semantic Scholar

4 downloads 0 Views 1MB Size Report
boldt Senior United States Scientist Award (to H. G. F.), by a North ... School (to I. F.), the National Institutes of Health (GM 10448 to D. H. B.) .... that is available from Waverly Press. ... moiety of riboflavin with high and similar efficiency (Table.
Val. 261,No.8, Issue of March 15, pp. 3661-3669,1986 Printed in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1986 by The American Society of Biological Chemists, Inc.

Biosynthesis of Riboflavin ENZYMATIC FORMATION OF 6,7-DIMETHYL-8-RIBITYLLUMAZINE FROM PENTOSEPHOSPHATES* (Received for publication, April 24, 1985)

Peter NielsenS, Gerhard Neubergerg, Isao FujiiP, David H. Bowng, Paul J. Kellers, Heinz G. Floss§, and Adelbert BacherS From the $Lehrstuhl fur Organische Chemie und Bwchemie, Technische Uniuersitat Miinchen, Lichtenbergstr. 4, 0-8046 Garching, Federal Republic of Germany and the §Department of Chemistry, The Ohio State University, Columbus, Ohio43210

The xylene ring of riboflavin originates by dismutation of the precursor, 6,7-dimethyl-8-ribityllumazine. The formation of the latter compound requires a 4-carbon unit as the precursor of carbon atoms 6a, 6, 7, and 7a of the pyrazine ring. The formation of riboflavin from GTP andribose phosphate by cell extract from Candida guilliermondii has been observed by Logvinenko et al. (Logvinenko, E. M., Shavlovsky, G. M., Zakal’sky, A. E., and Zakhodylo, I. V. (1982) Biokhimiya 47, 931-936). We have studied this enzyme reaction in closer detail using carbohydrate phosphates as substrates and synthetic 5-amino-6-ribitylamino-2,4-(lH,3H)-pyrimidinedioneorits 5’-phosphate as cosubstrates. Several pentose phosphates and pentulose phosphates can serve as substrate for the formation of riboflavin with similar efficiency. The reaction requires Mg2+. Various samples of ribulose phosphate labeled with 14C or 13Chave been prepared and used as enzyme substrates. Radioactivity was efficiently incorporated into riboflavin from [ 1-14C]ribulose phosphate, [3,5-14C]ribulosephosphate, and [514C]ribulosephosphate, but not from [4-14C]ribulose phosphate. Label from [ l-13C]ribose 5-phosphate was incorporated into C6 and C8a of riboflavin. [2,3,513C3]Ribose5-phosphate yielded riboflavin containing two contiguously labeled segments of three carbon atoms, namely 5a, 9a, 9 and 8,7,7a. 5-Amino-6-[1’-14C] ribitylamino-2,4( lH,3H)-pyrimidlnedione transferred radioactivity exclusively to the ribityl side chain of riboflavin inthe enzymatic reaction. It follows that the 4-carbonunit used forthe biosynthesis of 6,7-dimethyl-8-ribityllumazine consists of the pentose carbon atoms 1, 2, 3,and 5 in agreement with earlier in vivo studies.

The biosynthesis of riboflavin leads from GTP to5-amino6 - ribitylamino - 2,4( 1H,3H) -pyrimidinedione 5‘ -phosphate (Compound 1, Fig. 1) (3-6)which is converted to 6,7-di*This work was supported by grants from the Deutsche Forschungsgemeinschaft and theFonds der Chemischen Industrie (to A. B.), by National Institutes of Health Grant GM 32910 and a Humboldt Senior United States Scientist Award ( t o H. G. F.), by a North Atlantic Treaty Organization travel grant (toA. B. and P. J. K.), and by postdoctoral fellowships from the Ohio State University Graduate School (to I. F.), the National Institutes of Health (GM 10448 to D. H. B.), and the Alexander von Humboldt Stiftung (to P. J. K.). Preliminary communications on part of this work have been published (1,2). Thecosts 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 18U.S.C. Section 1734 solely to indicate this fact.

methyl-8-ribityllumazine (Compound 2,Fig. 1)by the addition of a 4-carbon unit. Dismutation of 2 yields riboflavin and the pyrimidine 4 (for reviews see Refs. 7 and 8). The structure and origin of the 4-carbon unit required for the formation of 2 remained elusive for several decades in spite of considerable efforts. Early work by Plaut and Broberg (9, 10) showed that both terminal carbon atoms of glucose, i.e. C1 and C6, were incorporated into the xylene ring of riboflavin with similar efficiency (9, 10). Subsequent studies led to the suggestion of acetoin (11, 12), diacetyl (13), pyruvate (14), tetroses (15), pentoses (16, 17), and 5-amino-6-ribitylamino2,4(1H,3H)-pyrimidinedione (4) (18, 19) asthe ultimate source of the 4-carbon moiety. Work byAlworth and coworkers (17) specifically suggested the incorporation of C1 of ribose into the benzenoid moieties of dimethylbenzimidazole and riboflavin. We have previously studied the incorporation of a wide variety of 13C-labeledprecurso,rs into riboflavin by the flavinogenic fungus Ashbya gossypii (20-25). A large number of incorporation experiments with a variety of single and multiple 13C-labeled precursors has been performed and can be summarized as follows. (i) Diacetyl and other symmetrical molecules can be ruled out definitely as precursors. (ii) Carbon atoms 1*-3* are biosynthetically equivalent to Cl-C3 of the pentose pool (see Fig. 1 for explanation of the arbitrary nomenclature used). (iii) C4* of riboflavin is equivalent to C5 of the pentose pool. (iv) C4 of the pentose pool has no equivalent in the 4-carbon moiety; it iseliminated during the biosynthetic process through a rearrangement of the carbon skeleton in which C3 and C5 are reconnected intramolecularly. Consequently, the carbon atoms 3 and 5 of the precursor ribose are contiguous inthe heterocyclic moieties of the products 2 and 3. Several attempts have been made to study the formation of the xylene ring in uitro. The enzymatic formation of riboflavin from 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (4) and acetoin or pyruvate by cell extract of Erernothecium ashbyii has been reported (12, 14). Plaut has suggested that these results may be explained by the enzymatic formation of diacetyl which would subsequently react nonenzymatically with the pyrimidine 4 to yield the lumazine 2 (26). More recently, Hollander et al. (19) reported on the formation of riboflavin from 1or 4 by cell extracts of Escherichia coli. Surprisingly, this reaction required no second substrate. It was, therefore, proposed, in agreement with an earlier hypothesis by Bresler et al. (18), that 4 can donate its ribityl group as a 4-carbon precursor by a dismutation reaction which would be formally analogous to the formation of riboflavin from 2.

3661

Enzymatic Formation

3662

/ 0

of 6,7-.Dimethyl-8-ribityllumazine

H-C-OH

2

H-C-OH ;H,OH

TABLE I Enzymatic formation of riboflavin by cell extract from C. guilliermondii The reaction mixture contained 0.72 mM 1, 38 mM Tris hydrochloride, pH 8.0, 3.8 mMMgC12, 10 mM ditbioerythritol, 0.95 mMof the respective carbohydrate and 0.6 mg of protein; total volume, 400 pl. Incubation was at 37 "C for 1 h. Riboflavin was monitored by bioassay with Lactobacilluscasei. Results are corrected for blank values (approximately 1.6 pM riboflavin). formed Riboflavin Substrate PM

H $HZ H-C-OH H-$-OH

4,R=

H

~ - 6 - 0 ~ H+OH

3

H-?-OH CHZOH

FIG. 1. Biosynthesis of riboflavin. Numbers with asterisks represent an arbitrarynomenclature to illustrate the formation of the 4carbon unit and the regiochemistry of the last reaction step (dismutation of 2).

Logvinenko and co-workers (27-29) have recently described the formation of 2 from GTP by cell-free extracts of the yeast Candida guilliermondii (27-29). The reaction could proceed with ribose phosphate or, less efficiently, with hexose phosphates as cosubstrate. The question then arises whether this i n uitro reaction follows the same mechanism as the biosynthesis of the vitamin in uiuo. We have used a variety of ribose phosphate and ribulose phosphate samples labeled with I4C or 13Cto characterize the enzymatic reaction in closer detail. The results are in full agreement with those of the i n vivo experiments. The evidence leaves no doubt that the reaction in the cell-free system is equivalent to the truebiosynthetic pathway. EXPERIMENTAL PROCEDURES'

RESULTS

We studied the enzymatic formation of riboflavin from synthetic pyrimidine substrates by cell extracts of C. guilliermondii. The reaction requires the addition of 4 carbon atoms. As shown in Table I, each of two pentose phosphates andtwo pentulose phosphates could serve as donor of the 4-carbon unit. The highest yield of riboflavin was obtained with ribose phosphate as substrate.Unphosphorylated pentoses and hexose phosphates could not serve as precursors. The pyrimidine 4 and its 5"phosphate 1 could both serve as precursors for the in uitro formation of the heterocyclic moiety of the vitamin (Table 11). Yields of riboflavin were similar with both types of pyrimidine substrate. Oxidized and reduced pyrimidine nucleotide cofactors (1 mM) did not increase the formation of riboflavin, nor did

None Arabinose 5-phosphate Fructose Fructose 6-phosphate Glucose Glucose 6-phosphate Ribose Ribose 5-phosphate Ribulose Ribulose 5-phosphate Xylulose 5-phosphate

0 4.5 0.1 0.3 0.1 0.3 0 9.0 0 5.4 8.2

TABLE I1 Enzymatic formation of riboflavin by cell extract from C. guilliermondii Reaction mixtures contained 40 mM Tris hydrochloride, pH 8.0, 4 mM MgC12,1 mM ribose phosphate, pyrimidine substrate as indicated, and protein. Results are corrected for blank values. For other details see Table I. Riboflavin Substrate

formed PM

None 1 (0.63 mM) 4 (0.44 mM)

0 8.1 5.3

TABLEI11 Enzymatic formation of riboflavin by cell extract from C. guilliermondii Reaction mixtures contained 30 mM Tris hydrochloride, pH 8.2, 2 mM MgC12, 1.5 mM 1, 5 mM ribose phosphate, and protein. Data are corrected for blank values. For other details see Table I. formedRiboflavin Addition

7%

None NADH (1mM) NAD (1mM) NADPH (1mM) NADP (1mM) ATP (2 mM) EDTA (10 mM)

100 80 37 83 34 98 4.0

ATP (2 mM) and thiamine pyrophosphate. The reaction was completely inhibited by 10 mM EDTA (Table 111).Cell extract

dialyzed against buffer containing EDTA could be partially reactivated by the addition of an excess of Mg2+. In order to study the fate of individual pentose carbon atoms incloser detail, we prepared enzymatically a variety of ribulose 5-phosphate samples specifically labeled with 14C. Commercial samples of radioactive glucose and glycerol served as precursors. 6-Phosphogluconate has a high retention time "Experimental Procedures" are presented in miniprint at theend on the anion exchanger Dowex 1-X8 and, therefore, can be of this paper. Miniprint is easily read with the aid of a standard easily separated from the other intermediatesoccurring in the magnifying glass. Full size photocopies are available from the Journal reaction sequence. We thus chose to isolate this intermediate of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. in radiochemically pure form. It was subsequently converted Request Document No. 85M-1368, cite the authors, and include a to the desired product, ribulose 5-phosphate. It should be check or money order for $4.80 per set of photocopies. Full size photocopies are also included in themicrofilm edition of the Journal mentioned that radioactive ribulose 5-phosphate samples decomposed rapidly even at thetemperature of liquid nitrogen. that is available from Waverly Press.

Formation Enzymatic

of 6,7-Dimethyl-8-ribityllumazine

A loss of more than 50% usually occurred within several days. Throughout this study, the ribulose 5-phosphate samples were, therefore, prepared immediately prior to further use. The radiolabeled ribulose 5-phosphate samples were used as substrates for the enzymatic formation of riboflavin with the pyrimidine phosphate 1 as cosubstrate. Light riboflavin synthase from B. subtilis was added to the reaction mixtures in order to assure quantitative conversion of the lumazine intermediate 2 to riboflavin. The vitamin was isolated and purified to constant specific radioactivity as described under "Experimental Procedures." It was subsequently converted to lumichrome (Compound 5 , Fig. 2) by illumination with visible light, in order to establish the radioactivity distribution between the ring system and the side chain. Lumichrome was obtained inradiochemically pure form by high pressure liquid chromatography. Radioisotope from [l-14C]ribulose5-phosphate and [5-14C] ribulose 5-phosphate was incorporated into the isoalloxazine moiety of riboflavin with high and similar efficiency (Table IV). Specific activities of riboflavin and of lumichrome obtained by photochemical degradation (Fig. 2) showed no significant difference, thus indicating that no radioactivity was contributed to the riboflavin side chain. [3,5-14C]Ribulose5phosphate contributedlabel to riboflavin with about the same efficiency as [5-14C]ribulose5-phosphate. This indicates that

H-C-OH 5

~ - 6 - 03 ~ H-C-OH CH,OH

H-C-OH H-+-OH H-?-OH CH,OH

6

2

FIG. 2. Photochemical degradation of riboflavin (3) and 6,7-dimethyl-8-ribityllumazine (2).

TABLEIV Enzymatic formationof riboflavin from 14C-and 3H-labeled ribulose 5-phosphate samples For details see "Experimental Procedures." Ribulose 5-phosphate snecific activitv

Riboflavin RA" ~~

dpmlnmol

[ 1-14C]

[3,5-'4C] [4-'4C]

3.3

[5"4C]

a

403 288 429 457 290 142 194 311 296

Lumichrome RA"

~~

%

128 72 110 117 0.7 123 132 159

%

125 NDb ND 109 3.1 ND 113 ND ND

RA, relative specific activity based on ribulose 5-phosphate. ND, not determined.

3663

C3 and C5 of the precursor are both incorporated into the product; ifC5, but not C3, were incorporated, one would expect that the incorporation rate from [3,5-'4C]ribulose 5phosphate should be only half as compared to [5-14C]ribose 5-phosphate. On the other hand, very little radioisotope was transferred to theproduct, riboflavin, from [4-14C]ribulose5phosphate. This finding is in excellent agreement with results of earlier i n vivo studies which had indicated that all pentose carbon atoms except C4 contribute to the xylene ring of riboflavin (20-25). For further confirmation of the correspondence between the i n vitro system and the actual biosynthetic pathway, we decided to prepare two 13C-labeledsamples of ribose 5-phosphate. The positions of the labels were chosen ( a ) to probe the origin of all four carbon atoms of the 4-carbon unit, (b) to probe for intact conversion of the pentose carbon chain by demonstrating intact incorporation of the (2243 bond, i.e. the bond which is cleaved in normal metabolism of the pentose phosphate carbon chain, and ( c ) to demonstrate the occurrence of the rearrangement reaction in which C4 is excised and C3 and C5 are intramolecularly reconnected. Consequently, one sample of the precursor was labeled with 13C at C1 and theother carried three intramolecular 13Clabels, each at 99% enrichment, at C2,C3, and C5. Conversion of the latter precursor by the process demonstrated in vivo should give a sample of riboflavin in which each of the two 4-carbon units is contiguously labeled with 13C at C2*, C3*, and C4* (see Fig. 1 for the arbitrary notation used). Both precursors were synthesized enzymatically on a gram scale from the appropriately labeled glucose 6-phosphate by the combined action of glucose-6-phosphate dehydrogenase and phosphogluconate dehydrogenase. Since ribose 5-phosphate hasbeen found (Table I) to be a more efficient substrate for riboflavin formation than D-ribulose 5-phosphate, the initially formed D-ribulose 5-phosphate was equilibrated i n situ with phosphoriboisomerase, first at room temperature andthen, in viewof the temperature dependence of the equilibrium (47), at lower temperature, to give a mixture of predominantly D-ribose 5-phosphate and asmaller amount of D-ribulose 5-phosphate. This mixture was isolated and used directly, without separation of the two components, in the cell-free synthesis of riboflavin. The D-[2-13C]gl~~ose 6-phosphate for the synthesis of [1-l3C]ribose5-phosphate was prepared by phosphorylation of commercially available D-[Z-~~C] glucose (99% 13C) with ATP and hexokinase. The overall yield, based on glucose, was 85%. The D-[2,3,5-13C3]ribose5phosphate was obtained from D-[1,3,4,6-13C4]glucose6-phosphate, which in turn was prepared from [1,3-13C2]glycerol3phosphate by the action of glycerophosphate dehydrogenase, triosephosphate isomerase, and aldolase, followed by fructose 1,6-diphosphatase and phosphoglucose isomerase. The glycerol phosphate was prepared with ATP and glycerokinase from [1,3-'3C2]glycerolsynthesized chemically (35, 36) from K13CN and [l-13C]aceticacid. It was essential for good yields in the synthesis of glucose 6-phosphate to isolate and purify the glycerol phosphate, in order to remove ATP and ADP. The overall yield of [2,3,5-13Cz]ribose5-phosphate was about 30% based on glycerol. Analysis of both labeled precursors by 13C-NMR spectroscopy confirmed the expected positions of the label. The conversion of the two 13C-labeled precursors into riboflavin with the cell-free system from C. guilliermonclii was carried out underconditions carefully optimized for maximum yields of riboflavin. From incubations with 250 pmol of ~ - [ 1 13C]ribose 5-phosphate and 300 pmol of D-[2,3,5-'3C3]ribo~e 5-phosphate, respectively, 8.24 and 3.99 pmol of pure ribofla-

3664

Enzymatic Formationof 6,7"Dimethyl-8-ribityllumazine

FIG. 3. 13C NMR spectrum of riboflavin formed in vitro from [l13C]ribose5-phosphate.

I*

160

5a

I

do

160

ppm

8a 8 7 7a vin were isolated. These two samples were subjected to analysis by13C NMR spectroscopy. The spectrum of riboflavin 2.5 0.9 1.7 from ~-[l-l~C]ribose 5-phosphate is shown in Fig. 3. Absolute 13C enrichments were calculated using the average of the 1* 2* 3* 4* ribityl carbon signals as natural abundance standard (1.1%). 1.4 1.8 The data (Scheme A) show high enrichment of the two 1* 19.6 2.8 positions, i.e. carbon atoms 6and 8a (see Fig. 2 for systematic 6 5a 9a 9 nomenclature of carbon atoms), confirming that C1 of the SCHEME A. "C-Enrichments (a)in riboflavin derived from pentose phosphate gives rise to C1* of the 4-carbon unit as [ l-13C]ribose5-phosphate. demonstrated in vivo (23). A small amount of label seems to be present at C2*, i.e. carbon atoms 5a and 8 of riboflavin the same process which leads to scrambling from C1 to C2 of (see Fig. 1 and Scheme A for numbering). This, too, is con- the pentose. A small amount of labeling of the 1*position sistent with observations made in the in vivo feeding experi- from D-[2,3-13Cz]ribo~e has also been observed in the i n vivo ments, which had shown some scrambling of label from C1 to experiments. In summary, the data confirm that the pentose C2 in themetabolism of ribose (23, 25). carbons 2, 3, and5 correspond to C2*,C3*,C4*of the The 13C NMR spectrum of riboflavin from D-[2,3,5-l3C3] alloxazine moiety, with loss of the pentose C4. ribose 5-phosphate is shown in Fig. 4. The carbons of the two Studies by Hollander et al. (19) had suggested that the 4-carbon moieties show the expected enrichments (Scheme xylene moiety of riboflavin can be formed from the ribityl B). The rather complex 13C-13Ccoupling patterns are most side chain of the pyrimidine 4 by enzymes from E. coli. In easily deduced for carbons which have the fewest enriched order to check whether this reaction occurs in C. guillierone-bond carbon neighbors. Thus, the optimal sites for ex- mondii, we synthesized 5-amin0-6-[1'-'~C]ribitylaminoamination of coupling are 8a (l*), 5a, (2*), 9a ( 3 7 , and 7a 2,4(1H,3H)-pyrimidinedione.Treatment of this compound ( 4 7 (see Scheme C). The other four signals (6, 7, 8, and 9) with cell extract of C. guilliermondii in the presence of unlayield entirely consistent information, but the patterns are beled ribose 5-phosphate afforded riboflavin which was phosomewhat more complicated. Carbon atoms 5a (2*) and 7a tochemically degraded to lumichrome (Compound 5 , Fig. 2). (4*) each show coupling to one other labeled carbon. Notably, The pyrimidine contributed its isotope to riboflavin with high 97% of the totalsignal due to 701 is one-bond carbon coupled. efficiency. However, the lumichrome fragment was devoid of The 3* positions, i.e. carbons 7 and9a, appear predominantly radioactivity (Table V). It follows that radioactivity from the as doubly coupled species. In addition, a small amount of purine precursor was exclusivelyincorporated into theribityl singly coupled species is discernible, reflecting again some side chain of the vitamin. metabolism of ribose 5-phosphate in the system (see below). In a similar experiment, 5-amin0-6-[l'-~~C]ribitylaminoCarbon atoms 8 and 9 show more complex patterns due to a 2,4(1H,3H)-pyrimidinedionewas treated with cell extract of combination of single coupling to the 3* position and statis- C. guilliermondii mutant E6 using unlabeled ribose 5-phostical coupling to each other as a resultof the combination of phate assecond substrate. This mutantis devoid of riboflavin two highly enriched 4-carbon units. A small amount of scram- synthase. Hence, the experiment yields the riboflavin precurbling of label into the 1*position, i.e. C6 and C8a, is also sor, 6,7-dimethyl-8-ribityllumazine(2). Photochemical degevident, with about 48% of the 801 signal showing statistical radation yielded 6,7-dimethyllumazine (Compound 6, Fig. 2). coupling to the 56% enriched C8. This presumably reflects As inthe previous experiment, the radioactive label was

H

H

H

1

3665

Enzymatic Formationof 6,7-Dimethyl-8-ribitylEumazine

7c

7a (4*)

FIG. 4. 13C NMR spectrum of riboflavin formed in vitro from [2,3,5-13C3]ribose 5-phosphate. SC, singly coupled; DC, doubly coupled.

I

135

I

130 PPm

i .

ribityl

carbons

1 lh

8a

I

16 1*

IO

8

7

H 56 H 69 H 78 I 2*

3*

TABLEVI

7a

4*

H 68 H 73 H 75 ]

6 5a 9a 9 SCHEME B. 13C-Enrichments in riboflavin derived from [2,3,5-”C3]ribose 5-phosphate.

Enzymatic formationof 6,7-dimethyl-8-ribityllumazim from 5-

amin0-6-[l’-‘~C]ribitylamino-2,4(1H,3H)-pyrimidinedione For details see “Experimental Procedures.” Specific activity Experiment

6,7-Dimethyl-8ribitvllumazine

6,7-Dimethylluma~ine~

dpmlnmol

1 1370 1180 17 2 1370 1290 21 a 5-Amin0-6-[1‘-’~C]ribitylamino-2,4(1H,3H)-pyrimidinedione. Obtained by photolysis of 6,7-dimethyl-8-ribityllumazine.

transferred to the ribityl side chain without significant dilution, but notto theheterocyclic moiety of 2 (Table VI). DISCUSSION

The incorporation of 14C-labeledpentose phosphates into riboflavin by cell extract of C. guillierrnondii follows accurately the patternobserved in the course of our earlier i n vivo studies with A. gossypii. C1, C2, C3,and C5 of the precursor (ribulose phosphate or ribose phosphate) are efficiently incorporated SCHEME C. 13C-13CCoupling pattern in riboflavin derived into the heterocyclic moiety, whereas C4 of the precursor is eliminated by an intramolecular skeletal rearrangement. from [2,3,5-13C3]-ribose5-phosphate. Generation of the lumazine 2 from ribose 5-phosphate involvestwo distinct processes: (a) rearrangement of a 5carbon unit and ( b ) condensation of the rearrangement prodTABLEV uct with the pyrimidine precursor. The data provide inforEnzymatic formationof riboflavin from 5-amim-6-[1 ‘-l4C1 mation on each process. ribitylamino-2,4(lH,3H)-pyrimidinedione The experiment with [2,3,5-13C3]ribose5-phosphate demFor details see “Experimental Procedures.” onstrates that the rearrangement occurred in 100% of the Specific activity labeled molecules. The experiments with both 13C-labeled Experiment Substrate” Riboflavin Lumichromeb substrates indicate that the condensation with 1 proceeds with the expected regiochemistry. dpmlnmol Cursory examination of the data from the incubation with 1100 30 1 1370 970 2 2 1370 the triple-labeled substrate might lead to theconclusion that the condensation is notentirely regiospecific, as seen by a 5-Amino-6-[l’-’4C]ribitylamino-2,4(1H,3H)-pyrimidinedione. significant 13C-enrichmentof C1*. It appears, however, that * Obtained by photolysis of riboflavin. H-$-OH H-?-OH H-?-OH CHZOH

3666

Formation Enzymatic

of 6,7-Dimethyl-8-ribityllumazine

the I* labeling is due to metabolism of the proffered pentose interchanging C-1 andC-2, perhaps via the pentose phosphate shunt. Thisidea is supported by the observation that thesum of the enrichments of C1* and C2* equals the enrichments of both C3* and C4* and that the amount of 1*-2* coupling is only what is expected on a statisticalbasis. If the C1* labeling were due to inversion of the 4-carbon unit, C1*wouldbe coupled to C2* to the same extent as C4* to C3* (i.e. 97%), which is clearly not observed. The interchange of pentose C1 and C2 also explains the presence of singly coupled species at C3*. This interchinge also occurred approximately 20% of the time with the sample from [l-13C]ribose5-phosphate, as well as in previous in vivo experiments. The specific incorporation rates of carbon from pentose precursors observed in the present experiments present an unresolved problem. Since the 4-carbon unit is duplicated in the last biosynthetic step, 14C-labeledriboflavin should have a relative specific activity of 200% of that of the precursor. The observed values (about 120%) were lower than the expected value. Substantial isotopic dilution has also been observed in the experiment with [1-l3C]ribose5-phosphate, but little dilution occurred in the experiment with [2,3,5-13C3] ribose phosphate. The origin of the isotopic dilution is unknown. The dialyzed cell extracts used in this study contained a small amount of unlabeled riboflavin, but it appears likely that some unlabeled riboflavin was also formed de novo during the enzyme incubation from an unknown precursor present inthe cell extract. We expect thatthis problem willbe OH resolved by further work with purified enzymes instead of crude cell extracts. Other authorshave reported on the formation of riboflavin from 4 by partially purified enzymes from E. coli. This reaction proceeded in the absence of a second substrate, and it has been suggested that the ribityl side chain of 4 was the source of the carbon atoms of the xylene ring (19). In our experiments, the isotope from l'-14C-labeled 4 was not incorporated into the heterocyclic moieties of 2 and 3,whereas it was incorporated into the side chain without significant dilution. Although it is conceivable that different biosynthetic pathways could operate in theyeast C. guilliermondii and the bacterium E. coli, earlier in vivo experiments with Salmonella typhimurium agree with the biosynthetic pattern reported in this paper. Thus, a mutant of S. typhimurium with an absolute requirement for guanosine incorporated label from [1',2',3',4',5'-14C]guanosine into the ribityl side chain, but not into the isoalloxazine moiety of riboflavin (48). I n vivo studies with B. subtilis have indicated the same biosynthetic pattern as in A. gossypii and C. guilliermondii, i.e. formation ofC1* from C1 and extrusion of C4 of the carbohydrate precursor (49). We conclude that a pentose phosphate is the actual precursor of the 4-carbon unit both in bacteria and fungi. Since various carbohydrate phosphates can serve as substrates in the enzyme reaction studied, the structure of the direct riboflavin precursor remains elusive. In particular, we cannot decide whether it is a pentose phosphate or pentulose a phosphate. Apparently, various carbohydrate phosphates are rapidly interconverted by enzymes present in the crude cell extract. We have recently obtained evidence for the formation of an aliphatic compound preliminarily designated as "X" which is produced by the cell extract of a C. guilliermondii mutant H-C-OH be converted from ribose phosphate (50). Compound X can to 2 in the presence of 4 by cell extract from a different C. guilliermondii mutant. The structureof compound X has not yet been determined. However, it was shown that the com-

pound is inactivated by treatment with phosphatase suggesting the presence of a phosphoric acid moiety. It is not known whether compound X contains 5 or 4 carbon atoms. On the basis of the data presented here it is clear that the in vitro system reflects the biosynthetic processes observed i n vivo, as opposed to an artefactual production of riboflavin. Thus, we may confidently bring together information gained from in vivo and in vitro studies to suggest a hypothetical scheme for the biogenesis of lumazine 2 (Figs. 5 and 6). The 4-carbon unit is depicted (Fig. 5) as being generated via a radical mechanism, initiated by abstraction of a hydrogen atom from C3 of ribose 5-phosphate. This latter step has precedent, e.g. in the ribonucleotide reductase reaction (51). However, at this stage an equivalent cationic mechanism, initiated by abstraction of a hydride from C3, is equally likely. Condensation of the proposed 4-carbon precursor with 4 to give 2 (Fig. 6) follows a plausible mechanistic pathway, based on known chemistry of these compounds. The proposed mechanism is not inconflict with any of the information currently available on riboflavin biosynthesis; however, a number of alternative mechanisms cannot be ruled out. It should be noted that theunphosphorylated pyrimidine 4 and its 5"phosphate 1 can both serve as substrates for the

OH

bH

dH

4*

0 H

OH

w

H

3

J HCOOH

OH

I*

,?"h/..

cHpQCH3

on o FIG. 5. Hypothetical mechanism for the formation of a 4carbon moiety from "ribose 5-phosphate. OH

0

0

0

H-C-OH

4

~ - 6 - 0 ~ H-C-OH

H-C-OH

CH,OH

CHzOH

I

-

H-C-OH H-&OH H-C-OH CHPH

H-C-OH

2

H-C-OH CH,OH

H-&-OH H-?-OH H-C-OH CHzOH

FIG. 6. Hypothetical mechanism for the formation of 6,7dimethyl-8-ribityllumazine.

Enzymatic Formationof6,7-Dimethyl-8-ribityZlumazine formation of 2. This could imply that therespective enzyme accepts both pyrimidines. However, it is also possible that the phosphate 1 is enzymatically dephosphorylated by the cell extract prior to incorporation into 2 and riboflavin. Recent studies with a lumazine synthase/riboflavin synthase complex (i.e. “heavy riboflavin synthase”) from B. subtilis suggest that the committed precursor for the lumazine 2 is the unphosphorylated pyrimidine 4.2 The availability of a cell-free system capable of synthesizing riboflavin from 1 and a pentose phosphate opens the way to the mechanistic investigation of the intriguing reactions involved in thisbiosynthesis and to thecharacterization of the intermediates and enzymes involved. The demonstration that the cell-free process shows the same features as the biosynthetic pathway in vivo provides assurance that information gleaned from the in vitro system does relate to the normal biosynthesis of riboflavin in the intactorganism. Acknowledgments-We thank Prof. H. Schirmer, Heidelberg, for helpful discussions. We are indebted to theLos Alamos Stable Isotope Resource, supported under National Institutes of Health Grant RR 02231, for l3C-labe1ed materials, and to the Ohio State University Chemical Instrument Center for mass spectra. We thank Degussa AG for a generous gift of reagents. The secretarial assistance of Astrid Konig and Angelika Kohnle is gratefully acknowledged. REFERENCES 1. Nielsen, P., Neuberger, G., Floss, H. G., and Bacher, A. (1984) Biochem. Biophys. Res. Commun. 118,814-820 2. Bacher, A., Nielsen, P., Neuberger, G., and Floss, H. G. (1984) in

3. 4. 5.

6. 7. 8.

9. 10. 11. 12. 13. 14. 15.

Flavins and Flavoproteins (Bray, R. C., Engel, P. C., and Mayhew, S. G., eds) pp. 799-802, Walter de Gruyter, Berlin Limens, F., Oltmanns,. 0... and Bacher,. A. (1967) . . Z. Naturforsch. 22b, 755-758 Burrows. R. B.. and Brown. G. M. (1978) . , J. Bacterwl. 136.657667 Logvinenko, E. M., Shavlovsky, G.M., Zakal’sky, A.E., and Senyuta, E. Z. (1980) Biokhimiya 4 5 , 1284-1292 Nielsen, P., and Bacher, A. (1981) Biochim. Biophys. Acta 6 6 2 , 312-317 Plaut, G. W. E. (1971) in Comprehensive Biochemistry (Florkin, M., and Stotz, E. H., eds) Vol. 21, pp. 11-45, Elsevier Scientific Publishing Co., Amsterdam Plaut, G. W. E., Smith, C. M., and Alworth, W. L. (1974) Annu. Rev. Bwchem. 43,899-922 Plaut, G. W. E. (1954) J. Biol. Chem. 211,111-116 Plaut, G. W. E., and Broberg, P. L. (1956) J. Biol. Chem. 2 1 9 , 131-138 Goodwin, T. W., and Treble, D. H. (1959) Biochem. J. 70,14P15P Kishi, T., Asai, M., Masuda, T., and Kuwada, S. (1959) Chem. Phurm. Bull. (Tokyo) 7,515-519 Bryn, K., and Stbrmer, F. C. (1976) Biochim. Biophys. Acta 4 2 8 , 257-259 Katagiri, H., Takeda, I., and Imai, K. (1959) J. Vitaminol. (Kyoto) 5,287-297 Alworth, W.L., Baker, H. N., Winkler, M. F., Keenan, A. M., Gokel, G.W., and Wood, F. L. I11 (1970) Biochem. Biophys. Res. Commun. 40,1026-1031 G. Neuberger and A. Bacher, manuscript in preparation.

3667

16. Ali, S. N., and al-Khalidi, U. A. S. (1966) Biochem. J. 98, 182188 17. Alworth, W. L., Dove, M. F., and Baker, H. N. (1977) Biochemistry 16,526-531 18. Bresler, S. E., Perumov, D. A., Chernik, T. P., and Glazunov, E. A. (1976) Genetika 12,83-91 19. Hollander, I. J., Braman, J. C., and Brown, G. M. (1980) Biochem. Biophys. Res. Commun. 94, 515-521 20. Bacher, A., Le Van, Q., Biihler, M., Keller, P. J., Eimicke, V., and Floss, H.G. (1982) J. Am. Chem. SOC.104,3754-3755 21. Keller, P. J., Le Van, Q., Bacher, A., Kozlowski, J. F., and Floss, H. G. (1983) J. Am. Chem. Soc. 105,2505-2507 22. Keller, P. J., Le Van, Q., Bacher, A., and Floss, H. G. (1983) Tetrahedron 39,3471-3481 23. Bacher, A., Le Van, Q., Keller, P. J., and Floss, H. G. (1983) J. Biol. Chem. 2 5 8 , 13431-13437 24. Floss, H. G., Le Van, Q., Keller, P. J., and Bacher, A. (1983) J. Am. Chem. SOC. 105,2493-2494 25. Bacher, A., Le Van, Q., Keller, P. J., and Floss, H. G. (1985) J. Am. Chem. SOC.107,6380-6385 26. Plaut, G. W. E. (1961) Annu. Rev. Biochem. 3 0 , 409-446 27. Logvinenko, E. M., Shavlovsky, G.M., Zakal’sky, A. E., and Zakhodylo, I. V. (1982) Biokhimiya 47,931-936 28. Logvinenko, E. M., Shavlovsky, G.M., and Tsarenko, N. Y. (1984) Biokhimiya 49,43-50 29. Logvinenko, E. M., Shavlovsky, G.M., and Tsarenko, N. Y. (1985) Biokhimiya 5 0 , 744-748 30. Plaut, G. W. E., and Harvey, R. A. (1971) Methods Enzymol. 1 8 , 515-538 31. Nielsen, P., and Bacher, A. (1983) in Chemistry and Siology of Pteridines (Blair, J. A., ed) pp. 705-709, Walter de Gruyter, Berlin 32. Bacher, A. (1984) Z. Naturforsch. 39b, 252-258 33. Cresswell, R. M., and Wood, H.C. S. (1960) J. Chem. SOC.47684775 34. Kuhn, R., and Cook, A. H. (1937) Chem. Ber. 70,761-768 35. Ott, D. G. (1981) Synthesis with Stable Isotopes of Carbon, Nitrogen and Oxygen, pp. 33-35,57, Wiley-Interscience, New York 36. Murray, A., and Williams, D.L. (1958) Organic Synthesis with Isotopes, pp. 931-932, Interscience Publishers, Inc., New York 37. Bacher, A., Baur, R., Eggers, U., Harders, H.-D., Otto, M. K., and Schnepple, H. (1980) J. Biol. Chem. 255,632-637 38. Klein, G., and Bacher, A. (1980) 2. Naturforsch. 35b, 482-484 39. Oltmanns, O., Bacher, A., and Lingens, F. (1968) 2. Naturforsch. 23b, 1556 40. Shavlovsky, G. M., Sibirny, A. A., Kshanovskaya, B. V., Koltun, L. V., and Logvinenko, E. M.. (1979) Genetika 15,1561-1568 41. Longenecker, J. P., and Williams, J. F. (1980) J. Labelled Compd. Radiopharm. 18,309-317 42. Jermyn, M. A. (1975) Anal. Biochem. 6 8 , 332-335 43. Bergmeyer, H. (1974) Methods of Enzymatic Analysis, p. 1343, Academic Press, New York 44. Rozutal, M., and Tomaszewski, L. (1974) Clin. Chim. Acta 5 0 , 311-317 45. Chen, P. S., Toribata, T. Y., and Warner, H. (1956) Anal. Chem. 28,1756-1758 46. Bergmeyer, H. (1974) Methods of Enzymatic Analysis, p. 1238, Academic Press, New York 47. Axelrod, B. (1955) Methods Enzymol. 1,363-366 48. Mailiinder, B., and Bacher, A. (1976) J. Biol. Chem. 251, 36233628 49. Le Van, Q., Keller, P. J., Bown, D. H., Floss, H. G., and Bacher, A. (1985) J. Bacteriol. 162, 1280-1284 50. Neuberger, G., and Bacher, A. (1985) Biochem. Biophys. Res. Commun. 127,175-181 51. Stubbe, J. A. (1983) Mol. Cell. Biochem. 105, 7428-7435

Continued on next page.

Enzymatic Formationof 6,7-Dimethyl-8-ribityllumazine SUPPLEMENTARY MATERIALTO Biosynthesis of Riboflavin. Enzymatic zine from Pentose Phosphates

Foormatian of 6.7-Dimethyl-8-ribityllu~~

Peter Nielsen, Gerhard Neuberger, Isao Fujii, David H. Bown, Heinr G. Floss and Adelbert BaCher EXPERIMENTAL Chemicals

PaulJ. Keller

PROCEDURES

-

5-Nitroso-6-ribitylamino-2.4(1H.3H)-pyrimidi"~dio"~ 1301, 5-nitro 5"phosphate 131). 5-nitro-6 ribitylamino-2,4(1H.3H)-pyrimidinedione 5"phosphate (32), 5-nitro-6-chloro 2,4(1H,3Hl-pyrimidinedione 133). 6.7-dimethyl-8-ribityllumazine (30). and 6.7 dimethyllumazine (34) were prepared as described. [2-14ClGlucose, [6-14Clglu c o s e , [l-14Clglycerol, [2-14c]glycerol and [l-14Clribose were purchased fro % l3C1 was obtained fro Amersham/Buchler. Braunschweig. D-[2-'3C1Glucose (99 Omicron Biochemicals. Ithaca, N.Y., and 11,3-13C21glycerol (99 % 13C) wa synthesized from [l-13Clacetic acid and K13CN, kindly provrded by the La ~ l a m o sStable Isotope Resource, by standard literature procedures (35. 36) Ribose 5-phosphate and Plorisil were Obtained from Sigma. so-6-ribitylamino-Z,4(lH,3H)-pyrlmidinedione

removed by centrifugation and the supernatant was adjusted topH 7.0 with 50 % (w/v) NaOH. Barium acetate 124.4 mmol) was then added, a small amoilnt of was added to precipitate of Ba-ADP was removed by centrifugation, and ethanol the supernatant to a final concentration of 80 % ( v / v ) . After standing overnight at 0 OC, the precipitated Ba Salt of glucose 6-phosphatewas collected by centrifugation and washed with 8 0 % l V / V ) ethanol. The salt was suspended was dried in a vacnum in ethanol. The ethanol was evaporated and the residue to give 4.0 9 of crude product. This material w a s dissolved in 40 m l Of H20 with the aid of some HC1. 6.66 mmol Na2S04 was added and. after neutralization with NaOH, the BaS04 was removed by centrifugation. The Supernatant was applied to a column (4x20 c m ) Of Dowex 1x8 (formate form, 100-200 mesh), which w a s then washed with H20 and eluted with a gradient of ammonium formate buffer (1 1 Of H20 V s 1 1 of 0.4 M HCOOH/0.12 M ammonium formate). The fractionscontaining glucose 6-phosphate, a s determined by the anthrone/sulfuric acid reaction (42). were pooled and the product was precipitated as the barium salt. The resulting bariumglucose 6-phosphate (2.53 g) was essentially 100 % (431. pure, based on enzymatic analyslswith glucose 6-phosphate dehydrogenase Yield. 88.5 %, based on glucose.

a-Ketoglutarate 115.8 mmol), NH4Cl 120.1 mmoll, dithioerythritol (219 mmol), and MgC12-6 H20 12.08 mmol) were dissolved in 150 ml of0.2 M Tris HC1 buffer, pH 1.9. The pH w a s adjusted to 7.8 with 1 N NaOH and 0,401 m m o l of NADP+ and 2.53 g of barlum [2-13C1glucose 6-phosphate. dissolved in 50 ml water,were added. The reaction w a s started by the addition of 294 U of glucose 6-phosphate dehydrogenase. 214 u of 6-phosphogluconate dehydrogenase and 209 U of glutamate dehydrogenase. Afterone h incubation at 27 OC, 1020U Of phosphori~icroarganisms Lactobacillus & ATCC 7469 was a gift of Dr. Bernhard bolsomerase was added and the incubation w a s continued for one h at 27 OC and Maillnder, Pfizer Inc., Karlsruhe. -guillrermonaiiATCC 9058 was obthen over night at 4 OC. Protein was removed by trichloroacetic acidI30 ml 50 tained from the American Type Culture Collection andWas grown aS described % sol. w/v) precipitation, the supernatant was neutralized with 50% NaOH, and earlier (38). Frozen cell pastewas stored at -20 OC. A mutant E6 was derived th; barium salts were preclpltated by addltion of 10 mmol of barium acetate from the C.guilliermondii wild type strain after treatment with ethyl meand ethanol to 80 8 v/v. The crude mixture of barium salts 15.19 g) was purified by chromatography on a D O W ~ X1 column as descrlbed above for barium thanesulfonate (39, 401. The mutant accumulated6.7-dimethyl-8-ribityllumazine glucose 6-phosphate. The fractions, a s determined by the orcinol reaction and required the addition of riboflavin I200 m g / l ) to the growth medium. 144). w e r e combined, and the sugar phosphates were again precipitated a5 the Preparation of cell extractFrozen 5 guilliermondii cellswere disrupted by harium salts. The product, 2.19 g, w a s found by enzymatic assay 143) to contain 4.1 mmol of ribose 5-phosphate and 0.59 mmol of ribulose 5-phosphate, passage throughan X-press. The frozen plugw a s thawed and the suspensionwas correspondrng to an overall yield of84.8 8 based on [2-13Clglucose. centrifuged. The supernatant was dialyzed against 0.1 M Tris hydrochloride pH 7.5 containing 10mM MgC12. D-[2,3,5-13C31ribose 5-pho~phate/D-[2,3.5-~~C,lribulose 5- hoS hate - I1.3T3C21Glycerol 11.44 91, ATP I23 mmol), dithiothreitol 10.6 : m o l : , and MgC12 Enzyme Substrates 5-Amino-6-ribitylamino-Z.4(lH,3H)-pyrimidi"~dion~ ( 4 ) was I23 m m o l l were dissolved in 600 ml of H20and the pH w a s adjusted to 9.6. prepared by catalytic hydrogenation of an aqueous solution of 5-nitroso-6Glycerokinase ( 6 0 0 u ) was then added and the mixture was incubated at room ribitylamino-2,4(1H,3H)-pyrimidinedione over palladium on charcoal at room temperature while the pHwas maintained at 9-9.5 by titration with 1 N NaOH. temperature and atmospheric pressure. The Catalyst w a s removed by filtration After COnSUmptlOn of 15.5 m l of 1 N NaOH, the pH change Stopped and the through a Millipore fllter. The solutionwas used mmediately. mixture w a s incubated for one more h. The pHw a s then adjusted to 1.1 and 50 mmol o f BaCl2 was added. The precipitated barium salt of ADP W a s removed by 5-Amino-6-~ibitylamino-Z,4llH,3H)-pyrimidi"=di~~~ 5"phosphate (1) was precentrifugation and the Supernatant loaded O n a column of D O W ~ X1x8 (4x20 cm, pared by catalytic hydrogenation of 5-nitroso-6-ribitylamino-2,4llH,3HI-pyrimidinedione 5"phosphate or 5-nitro-6-ribitylamino-2.4(lH,3H)-pyrimidi~~dion~ formate form). a-Glycerophosphate appeared in the "on-adsorbed fraction and in the first fractions eluted with ammonium formateb u f f e r (gradient of 1 1 of 5"phosphate as described above. The solution was used immediately. H20 vs 1 1 of 1.3 M HCOOH/ammonium formate, pH 3). Precipitation of the barium Preparation of isotopically carbohydrate phosphates salt with acetone w a s attempted, but w a s unsuccessful. The purifLcatlon on D O W ~ X1x8 w a s repeated and the fractions containing o-glycerophosphate, a s Radioactive phos ho luconate from lucose - Reaction mixtures contamed 0.25 M determined by phosphate assay I451 were pooled. Barium acetate I50 mmol) was l 5 m M NADP+, 15 mM ATP, 5.6 " m o l I5 added, and ethanol w a s subsequently added to a final concentration of 8 0 % Tris hydrochlarBdegpH 7.8, 1 K M h uci) of glucose appropriately labeled with l4C, 50 U of hexokinase, and 50 u (v/vl. The precipitated harium salt(6.04 gl contained 12.3 mmol of 0-glyceroof glucose phosphate dehydrogenase ina total volume of 400 "1. The reaction phosphate based on enzymatic assay 146). This corresponds to an isolation yield of 79.3 % based on the amount of glycerol reacted (15.5 mmole by titramixture was incubated at 30 OC for 30min. The mixture was dilutedto a v o l u m e tion) and a total yield ofat least 59.3 % based on startlng glycerol. of 10 ml with deionized water and placed on a c o l u m n of D O W ~ X1x8 (formate f o r m , 200-400 mesh, 0.9 x 15 cm). The column was washed with 25 ml of deioniThe [ 1,3-13C21-a-glycerophosphate was converted into glucose 6-phosphate in zed water and subsequently developed with an eluent containing 0.1 M formic two batches. The barium Salt(7.25 m m o l ) w a s dissolved in water with the aid acid and 0.1 M ammonium formate. The flow rate was 0.6 ml/min. Fractions were collected and monitored by liquid scintillation counting of aliquots. 6-Pho5- Of HC1, and barium ions were removed by addition of Na2S04 and centrifugation. phogluconate was eluted from 8 0 to 120ml. Fractions were combined and lyophi- The supernatant w a s added to 1 1 of 0.2 M Tris HC1 buffer, pH 7.9, containing 9.1 mmol of NAD', 73 mmol of sodium pyruvate, and 1 mmol Of dithiothreitol. alized. The radiochemical yield was 60-83 %. Glycerophosphate dehydrogenase I500 U1 and L-lactate dehydrogenase I 5 0 0 0 Ul were then added, and the formation of dihydroxyacetone phosphatewas followed Phosphogluconate w a s prepared by modi114ClPhosphogluconate from glycerol by assay for alkali-labile phosphate. When about 40 % of the o-glycerophosfication of a published procedure141). Reaction mixtures contained0.1 M Tris phate had been convertedto dihydroxyacetone phosphate, 15,000 U of triosehydrochloride pH 7.8, 20 mM MgC12, 13 mM ATP, 27 mM NAD+, 20 mM fructose 6phoshate, 27 mM sodium pyruvate, 1.3 "mol I10 uci) of 14C-labeled glycerol, 34 phosphate isomerase and 500 u of aldolase were added and the formation of fructose 1.6-diphosphate was followed by the anthrone/sulfuric acid reaction U of glycerokinase, 1.6 U of glycerol 3-phosphate dehydrogenase, 350 U of triose phosphate isomerase,1.5 u of transaldolase, and5.5 U of lactate dehy- (42). When fructose diphosphate formation had reached its maximum, fructose1,6-diphosphatase (100U ) and MgC12 (toa concentration of1 mM) was added and drogenase in a total volume of 300ul. 14C-Labeled glycerol w a s purchased as an ethanol solution.The ethanol w a s removed by a Stream of nitrogen prior tothe pH w a s adjusted to 9. Then phosphoglucose isomerase I880 U) was added and the r e a c t ~ o nmixture was incubated over night. The pH was adjusted to 7.5. use. The reaction mixturew a s incubated at 30 OC for 2 h. barium acetate (294 mmol) w a s added andyellow colored impuritieswere removed by treatment with activated charcoal (55 91, which w a s filtered off and washed Phosphoglucose isomerase I10 ul was added and the mixturewas again incubated with 0.01 M NH40H. Ethanol w a s added to thecombined filtrate and washings to for 30mi" at37 Oc in order to produce glucose 6-phosphate from fructose 6a final concentration of 80 8 v/v. The precipitated crude barium Salt was phosphate which had been formed in the first reaction Step. purified on a D O W ~ X1x8 column (4x20 cm. formate form) as described above. The fractions containing the ~ r o d u c twere pooled and the barium Salt W a S again Glucose 6-phosphate was further converted to 6-phosphogluconateby the addition of 6.1 mg (7 umoll of NADP+,50 u of glucose phosphate dehydrogenase, andprepared. The product ( 8 3 5 mg) w a s found by enzymatic assay to contain 1.28 250 ul of 0.25 M Tris hydrochloride pH 7.9 followed by incubation at 30 OC for mmol of glucose 6-phosphate and 0.16 mmol of fructose 6-phosphate. In another batch, 5.0 mmol of a-glycerophosphate was converted to 749mg of barium salt 40 mi". The reaction mixture w a s subjected to ion exchange chromatography as containing 1.09 mmol of glucose +phosphate and 0.22 mmol of fructose 6described above. The radiochemical yield was 27-46 %. phosphate. The combined yield from o-glycerophosphate w a s 44.7 %. Radioactive ribulose 5-phosphate from 6-phosphogluconate The reaction mixThe mixture of [1,3,4,6-13C4]g1ucose 6-phosphate and [l,3.4,6-13C41fru~tose6ture contained 90 mM Tris hydrochloride pH 7.8, 7 mM NADP*, 0.3-2 UM 6phosphogluconate from the previous step, and5 u of 6-phosphogluconate dehy- phosphate w a s converted to a mixture of [2,3.5-l3C ribose 5-phosphate and [2,3,5-13C3]~ibulo~e5-phosphate as described for [~-'3Clglucose 6-phosphate. drogenase in a total volume of 0.9 ml. The mixture w a s incubated at 37 OC for except that 1000 u of phosphoglucose isomerasewere included in the reaction 90 mi". diluted to 10 ml with deionized water, and placed on a column of Dowex we obtained 1.03 g of mixture. F r o m 1.75 mmol of the hexose phosphate mixture, 1x8 (200-400 mesh,formate farm, 0.9 x 15 cm). The column w a s developed with a purified pentose phosphate barium salt, which according to enzymatic assay Solntion containing 0.1 M formic acid and 3 0 mM ammonium formate. The flow contained 2.07 m m o l of ribose 5-phosphate and 0.27 m m o l of ribulose 5-phosrate was 0.6 ml per mi". FractlOnS were collected and monitored by liquid phate. This correspondsto a yield of 85.1 8 based on hexose phosphateor 30.2 was eluted from 35-65 scintillation countingof aliquots. Ribulose 5-phosphate % overall yleld basedon consumed glycerol. ml. Fractions were pooled and lyophlllzed. The radiochemical yleld based on 6phosphogluconate Was 20-45%. as evidenced The samples underwentsome decomposition upon prolonged storage, by reexaminingthe "C-NMR Spectrum of an aliquotafter 6 months at-20 OC. D-[l-13Clribase 5 - p h 0 ~ p h a t e / D - [ l - ~ ~ C l r i b u l o s e5-phosphate ATP 16.15 mmol) and MgC12.6 H20 (6.11 mmol) were dissolved in 240 ml H20 and the pH W a s Preparation of 5-nitro-6-[l'-14C]ribitylamino-2,4(1H.3H)-py~imidi"~d~o"~ adjusted to 6.75 Wlth 1 N NaOH. Dlthiothreitol I 1 6 8 u m o l l , D-[2-13C191ucase (1.0 g ) and hexokinase 1602 IO), dissolved in 10 m l H20. w a s added. and the [l-14C]Ribose oxime ~n aqueous solution of [1-14Clribose (8mg. 50 U C ~ Iw a s mixture was incubated at 29.5 OC. while the pH was maintained at 6.6 by lyophilized in a 1.5 ml plastic cup. Subsequent to the addition of 0.15 ml of titration with 1 N NaOH. When the pH change stopped. after about 30 min, a 1.3 M hydroxylamine solutionin absolute methanol (30). the mixture was kept incubation was continued far another h, the mixture w a s then cooled in ice and C . Crystals of for 12 h at room temperature and Subsequently for 24 h at 4 ' 50 ml of 50 % (w/vl trichloroacetic acid was added. Precipitated protein w a s

-

~ n m z e5 Light riboflavin synthase of Bacillus subtilis was purified to a c i & activity of 18,000 " m o l m g - l h-l as described earlier (37). Triose phosphate isomerase w a s purchased from Boehringer. Other enzymes used were obtained from Sigma.

-_

-

-

-

-

-

"~

-

-

3669

Enzymatic Formationof 6,7-Dimethyl-8-ribityllumazine

HC1 were edded, and the incubation at 30 OC was continued for 7.5 h. The g of trichloroacetic acid. Protein reaction was stopped by the addition of 24 was removed by centrifugation at 32,000 9 for 20 nin. The supernatant was brought to pH 7 by the addition of 30 g of sodium acetate. Water w a s added to a f i n a l volume of 500 ml. The Solution w a s applied to a column of rlorisil (0.8 x 11.5 om). The column w-5 Washed with 500 m l of deionized water. Pluorescent materialwas eluted witha mixture of acetone/2 M N H ~ O H(50 ml,1:1, [1-14c]~ibit lamine A suspension of 1.1 mg of PtGZ in 2 m l of acetic acid v/vl. The eluate was evaporated to dryness under reducedpressure. The residue was bydrogenaied-fF20 mi". [1-14C1Ribose oxime (2.8 mgl was added and hydrowas dissolved in 30m l o f water. l'he solutlon was placed on a column of AG 50 genated for 12 h at room temperature and atmospheric pressure. Platinum was WX8 I H' form, 1.0 x 20 om). The column was deveLoped with 0.1 M KC1. Fractions removed by filtration, and the s o l u t i o n was evaporated to dryness under redu- containing riboflavin were combined and concentrated to dryness under reduced ced pressure. The oily residue was dissolved in l o m l of water and placed OD a pressure. Yield, 3.1 mg. column of AG 50 W X 8 (H' form. 200-400 mesh, 0 . 8 X 12.5 C m l . The Column W a s washed with 100 m l of deionized water. Ribitylamine w a s subsequently eluted Riboflavin Exom [2,3,5-13C3]ribose 5-phosphate - The reaction mixture aontaiwith 50 m l of 0.2 M NX~OH. The effluent was evaporated to dryness under ned 0.1 M P r i s HC1 pH 7.5, 5 mM MgClZ. 2 m M CaCl2, 10ml< dithioerythritol, 1.1 reduced pressure. Deionized waterwas added and evaporation was repeated three mM [2,3,5-13C,Iribose 5-phosphatel 0.65 m M 1, and dialyzedcrude cell extract times. from C.guill;ermtn ( 1 . 2 g of protein) in a total volume of 136 ml. After incubation at 3 0 OC for 2 h, 4.0 m l of 135 mM 1. 10 ml of 14.7 mM [2.3,55-Nitro-6-[l'-14Clribitylami"o-2;4(lH,3HJ-pyrimidi~edio~e A solution contai- 13C31ribose 5-ph06phater and 75,000 U of light riboflavin synthase dissolved n i y 6.3 mg of 5-nitro-C-chloro-2,4(lH.3~I-pyrimidinedionc in 350 ul of ethai n 13 ml of 0.1 M TriS WCl were added, and the incubation was continued for no1 was added to the ribitylamine. Phosphate buffer 8pH .0 (0.1 M, 1.5 nil1 WaS 9.5 h. RiboElavin was purified 2s described above. Yield,1.5 mg. added brd the Solution was kept at 4 0 OC for 24 h. The solventwas evaporated under reduced pressure. and the residue was dissolved i n 25 ml deionized of Riboflavin from 5-amino-6-I.l'-14C1ribityiamino-Z,4l1H,3H)-pyrimidineaione - A water. The solutionWas passed through a preparative HPLC Column ILichrosorb solution of 5-nitro-6-[1'-*4C]ribitylamino-2.4l1H.3H~-pyrimidinadione (1.0 mg, RP18, 16 x 250 mm; eluent. 2 0 mM ammonium formatepH 3 . 8 ; flow rate, 6-12 3.2 u m o l ) in 1. ml O f deionized waterw a s hydrogenated over palladium on Charml/minJ. Fractions were collected and lyophilized. Yield, 11.0 umol. Speoific coal. The suspension was passed through a nitrocellulose filter to remove the activity , 1320 dpm/nmol. Radiochemical yield based on Il-14Clribose, 13.2 %. catalyst. The filter was Wasbed With 200 u l of Water. A n aliquot of the filtrate i0.46 m i containing 1.4 uno1 of the pyrimidinedionelw a s mixed with 1.14 The second crop of [l-14C]ribose oxime obtaioed at the end of step 1 m l of a solution containing0.14 M Tris hydrochloride pH 7.5, 14 mM MgCl2, 14 was separately oarried through the subsequent steps yielding an m M dithioerythrit.ol,1.4 mM ribose 5-phosphate, and cell extract fromC. guiladditional amount of 8 umol of 5-nitro-6-[l~-14Clribitylamino-2,411H,3Kl-pyriliermondii ATCC 9058 (12 mg of protein). The mixtureWas incubated for 2 h at midinedione witha specific activityof 1370 dpm/nmol. The total radiochemical 37 OC. L'rght riboflavin synthase from 0. subtilis 12200 0) wa§ added and the yield basedan ribose was 23 Both batches appeared pur@ as judged by irnaly- incubation was continued for another 1.5 h. Riboflavin was isolated and puritical HPLC and thin layer chromatography (cellulose, 1-butanol/acetic acid/wafied as described above. ter 50:15:35, v/v; Rf = 0.62). They showed the expected absorption spectrum ~ = 1.95 ~at p H 1.01. ~ / c ~ (xmax 227 nm, 323 nm, ~ 6 .~7 - D i m e t h~y l - 8 - r i b i t y l l u m a z i n a from 5 - a m i n o - 6 - [ 1 ' - 1 4 C ] ~ i b i t y l a . i n o - 2 , 4 ( l H , 3 H l pyrimidinedione Assay mixtures contained 0.53 mM 5-amino-6-Il'-1'Clribitylamino-2,4(1H.3Hl-pyrinidineiiione [freshly prepared as described above), 0.1 M Enzyme Studies Tris hydrochloride pH7.5, l o m M MgC12, 10 mM dithioerythritol, 1 InM ribose 5phosphate, and cell extract from 5 guilliermre mutant E6 (15 m g of proAll experiments Were performed i n dim light to minimize photolysis of proteinl in a total volume of 1.6 nl. After incubation for 2 h at 37 OC the ducts. The pyrimidiile substrates used are very oxygen sensitive. They were mixture was passed througha column of Florisil (0.8 x 10 cm). The column was expofreshly preparedfor each experiment,and efforts were made to minimize washed with 10 m l of deionized Water. Fluorescent material was eluted with sure to molecular oxygen. The eluate was brought t o dryness under acetone/2 MNH40H (15 m l , 1:l. "/VI. Oolumn reduced pressare. The residuewas passed througha reversed phase HPLC Riboflavinfrom14C-labeledribulose5-phosphate Assaymixtures con(Nucleosil 10 CIS, 0.4 x 25 om; flow rate, 1 ml/min; eluent, 20 mM ammonium formate buffer pH 3.8). The effluent was monitored photometrically (254 and tained 0.1 M Tris hydrochloride pH 7.5, lo mH dithioerythritol, 10 m~ ~gel,, 1 mM rreshly prepared5-amino-6-ribitylamimo-2.4(1W,3H)-p~~imidi~~d~o"~ 5'-phos405 nm). The fraction containing 6 , 7 - d i m e t h y l - 8 - r i b i t y l l u ~ a ~ i(retention ~e phate, 1 mM radiolabeled ribulose phosphate. 1000u of iight riboflavin syn- time, 32 min)was collected and lyophilized. thase from B. silbtilis, and dialyzed cell extract from & gullliermondii RTCC 9058 (3 mg of protein) in a total volume of 2 ml. The mixture w a s incubated Degradations for 1 h at 30 OC. An aliquot (100ul] w a s retrieved for bioassayof riboflavin using b. casei. Riboflavin (25nmol dissolved in 400 ul of water) was added as Fhoeocheinical degradation of riboflavin Riboflavin was dissolved in buffer carrier to the assay mixture, and the solution was passed through acolumn of containing 0.25 M formic acid and 0.25 M ammonium formate pH 3.8. The eolutian Florisil (0.5 X 7 em). The c o l u m n was washed with 10 m l of deronized water. w a s illuminated far 2 h bya 500 N lamp (distance, 0.1 m). The solution was Fiuorescent material was eluted with 5 m l of acetone/2 M N H ~ O H(~:1, The passed through a HPLC column (Nucleoail 1 0 C18. 0.4 x 25 ~ m flow , rate, 2 eluate w a s evaporated to dryness Under reduced pressure. The residue m l / m i n ; eluent. 34 % methanol in 0.1 M ammonium formate pi3 6.31. Absorbance dissolved in 0.4 m l of deianized water and freedfrom so,lids by centrifugation was monitozed at 254 and 365 nm. The fraction containing lumichrome( 5 ) (rein a Beckmann Airfuge at 100,000 g for 10 m i " . The sLpernatant was passed tention time,1 7 minl was collected and lyophilized. through a reversed phase HPLC column (Nuclaosil lo cis, 0.4 x 25 cm, flow rate 2 ml/min; eluent. 20 % methanol in 0.1 M ammonium formate). m e e l u a t e w a s Shotochemical degradation 6.l-dimethyl-8-ribityllu~n+ A Solution of monitored photometrically (254 nm and 405 nm). The fraction containing 6,7-dimethyl-8-ribityllumazine in 0.1 M a m m o n L u m formate buffer pH 3.5 w a s flavin (retention time, 30 ni.n) w a s collected and evaporatedto dryness undei illuminated f a 1 h as described above. The solution was passed througha HPLC reduced pressure.. The residuewas dissolved in 0.3 m l of deionized water, and column (Nucleosil 10 C18, 0.4 x 2 5 cm; eluent, 10% methanol in 0.1 M ammonium cowaliqilots were used for riboflavin bioassay and for liquid scintillation formate pH 6.3; flow rate, 1 ml/minl. Absorbance was monitored at 254 and 340 ting. The remaining material was converted to lumichrome. nm. The fraction Containing 6.7-dimethyllumazine (61 (retention time, 25 min) wa5 collected and Lyophilized.

ribose o x i m e were collected on a fritted glass disk by Suction and washed three eimes with70 u l of ethanol yielding 2.8 mg of white solid. The filtrate was to dryness and yielded approximately 2 a9 of ribose oxime Which was separately converted to 5-nitro-6-ll'-14ClribitylaminO-2,~(lH~3H)pyrimidinediono.

-

-

-

"

%.

F

-

"

~-

____-

-

-

-

Riboflavin from [1-I3CIribose 5-phos hate The reaction mixture contained 0.1 M Tris XCl pH 7.5, 5 mlrl MgCIZ, 10 m z dithioerythritol, 1 m M [1-13C]ribose 5phosphate, 0.9 m M freshly prepared 5-amino-6-ribitylamino-2,4[lX,3H)-pyrisidinedione S'-phosphate (1). &Rd dialyzed crude cell extractf r ~ m guilliermondii (1.3 g of protein1 i n a total volume of 132 ml. After incubation at 30 OC for 2 h,5.0 ml of 18.6 m M 1. 16 m l of 7.7 m M [1-13Clribose 5-phosphate, and a solution of100,000 U of light riboflavin Synthase in 6 ml of 0.1 M Tris

c-

NMR

SpectroscOpy

'H-decoupled 13C-NMR spectra Were obtained a t 7.0 T on a Bruker W M - 3 0 0 N M R spectrometar. Samples of r i b o f l a v i n were dissolved in approximately 0.b m l of D@X30-~6. The Spectral acquisition parameters were: Pulse angle 60 O, repetition tile = 2.0 sec, Spectral width = 15.15 KHz. 32 K data sets, 1.0 HZ line broadening, Chemical shifts are referenced totetramethylsilane,

-