Biosynthesis of Branched Chain Deoxysugars

A new cytidine nucleotide was found in Azotobacter uinelandii strain 0. On the basis of chemical analyses, mass spectra, infrared spectra, nuclear magnetic resonance spectra, periodate oxidation studies, and a series of degradation studies followed by comparison with authentic samples, it is proposed that the new compound is cytidine-S’-diphospho-6-deoxy-3-C-methyl-2-O-methyl-4-O-(O-methylglycolyl)-L-aldohexopyranose. The compound is therefore an 0-methylglycolic acid ester of cytldine-5’-diphospho-vinelose previously found in the same microorganism.

On the basis of chemical analyses, mass spectra, infrared spectra, nuclear magnetic resonance spectra, periodate oxidation studies, and a series of degradation studies followed by comparison with authentic samples, it is proposed that the new compound is cytidine-S'-diphospho-6-deoxy-3-C-methyl-2-O-methyl-4-O-(O-methylglycolyl)-L-aldohexopyranose.
The compound is therefore an 0-methylglycolic acid ester of cytldine-5'-diphospho-vinelose previously found in the same microorganism.
Two nucleotides present in estracts of Azobbacter trinekandii strain 0 have been shown to be derivatives of cytidine 5'.diphosphate (1,3). In one of these (CDP-X) the pyrophosphate group is attached to an 0-methylated, branched chain sugar, designated vinelose, and in the other (CDP-Y) it is attached to a carboxylic acid ester of 0-methylated, branched chain sugar. The structure of these nucleotidyl sugars is of interest, particularly in view of the extensive investigations (see the recent review by Grisebach (4)) on biosynthesis of branched chain sugars in connection with the participation of nucleotides in the branch formation.
Interest in these compounds also arises from earlier observations (for a review of the literature, see References 4 and 5) that most of branched chain sugars as well as their carboxylic acid ester and methyl ether derivatives were found in antibiotic glycosides produced by microorganisms. It seems possible that these nurlmtidcs would function in either transformations or transfer proceases involving the sugar residues at some stage in the biosynthetic route leading to this type of bacterial glycoside. * This work was supported ill part by researc*h grants from the hlinistry of Education, .Japan. The previous papers in this series (l-3) were not numbered.
The yield was 146 mg (185 pmoles, if calculated from ultraviolet light absorption measurements with the use of 13.0 as a millimolar extinction coefficient); [ali S12.5' (c, 1.55, in HzO). The absorption spectrum in the ultraviolet region from 220 to 310 mp was identical with that of highly purified 5'-CMP.
The millimolnr extinction coefficient in 0.01 K HCl at 280 mc( was 13.0, based on absorbance per 2 moles of total phosphntc. This value is also identical with that of 5'-CMI'. The ratio of total phosphate to acid-labile phosphate to pentose (as ribose) t.o mcthoxyl group was 2.00: 1.01:0.92:1.90. The occurrence of two methoxyl groups distinguishes CDP-Y from CDP-vinelose. Formalion of CDP-vinelose from CDP-Y by -Alkaline Hydrolysis -The presence of an ester linkage in CDP-Y was first indicated by its positive reaction with t.he hydroxylamine-FeC13 reagent (14) for carboxylic acid rster.
In support of t.he result of the color reaction, it was further shown t.hat CDP-Y is susceptible to hydrolysis with weak alkali.
A complctc degradation of CDP-Y under the mildest possible condition was found to occur on treating in 1 K ammonia at 35" for 2 hours. Under this condition, an ult.rnviolet light absorbing product was obtained and tentatively identified as CDP-vinelosc by paper chromatography in Solvents A, 13, G, and H and by paper electrophoresis in Buffers A and 13 (Table I).
To obtain a substantial amount of the nucleotide fragment (CDP-dcacyl-Y; Compound II, Fig. 1) for further characterization, 58 mg of CDP-Y, dissolved in 3 ml of 1 K ammonia, were placed in a water bath at 35" for 2 hours.
The nucleotide product after hydrolysis was recovered from the mixture by the charcoal met.hod (15). The resulting eluate (84 ml) from charcoal was reduced to 0.4 ml on a rotating evaporator, from which the nucleotide was precipitated as white powder with acetone.
With this preparation, :I series of identification procedurrs was undertaken.
;ill of the evidence so far obtained indicates that CDP-deacyl-Y is, in fact, CDP-vinelose. The evidence may be outlined as follows.
1. The, nucleotide exhibited an ultraviolet light absorption spectrum typical of a cytidine derivat,ivc.
The ratio of cytidinc to total phosphate to labile phosphate to pentose (as ribose) to mcthoxyl group was 1.00: 1.99 :0.96 :0.91: 1.05. The reducing value, determined by ferricyunide reduction (16) after hydrolysis with 0.01 N IICI for 5 min at 100" and expressed relative to a glucose standard, was 0.26. With this method, the reducing value of authentic vinelose is 0.27 relative to glucose, 2. Enzymatic hydrolysis of the nucleotide with snake venom nucleotide pyrophosphntase yielded t.wo products, which were separated from each other by ion exchange chromatography on Dowex l-chloride.
The separation method was essentially the same as t,hat used previously for the separation of pyrophosphatase digest of CDP-vinelose (3). One of the products, which was elutcd first from the column, was recovered and concentrated by t.he charcoal met.hod and precipitated as the barium salt with ethanol from water; 25 mg of crystallinc~ material were obtained from 46 mg of CDPdeacyl-Y.
Its ultraviolet. absorption spectrum, infrnred absorption spectrum, and opt.ical rotatory dispersion spectrum were TANLE 1 Paper chrondography and paper elecfrophoresis of GDP-Y and its degradation products ~. I This n~icleotidc was obtained from CI>P-Y by hydrolysis with ammonia.
The nucleotide product of the hydroxylaminct treatment behaved as this compound.
c This nuclcotide was obtained from CDP-Y by hydrolysis with nucleotide pyrophosphatase.
The nucleotide product from CDPdeacyl-Y behaved as this compound.
all identical w&h the corresponding spectra of 5'-CUP. Inorganic phosphate was liberated from the &sample when it was exposed to the action with 5'-nucleotidase.
It is concluded therefore that the first product is 5'-CMP.
The second product was likewise recovered and concentrated to amorphous solids. The yield from 46 mg (65 /Imoles as cytidine) of CDP-deacyl-Y was 55 pmoles, as estimated from the phosphate content; [ill]i4 +4540" (in H20).
The substance was shown by paper chromatography and paper electrophoresis (Table I) as well as by the measurement of nuclear magnetic resonance spect.rum ( Fig. 2; Table II) and optical rotat.ory dispersion (300 to 650 mp) to be indistinguishable from vinelose-1-P. The phosphate ester from CDP-deacyl-Y is referred to as descyl-Y-l-P (Compound III, Fig. 1). 3. Dcacyl-Y-l-P is susceptible to degradation by B. co/i alkaline phosphatase.
One of the cleavage products was shown to be inorganic phosphate with the Lowry-Lopez reagent (17) and the other (den&Y; Compound IV, Fig. 1) to be vinelose by means of paper chromatography (Table III), infrared absorption spectroscopy (Fig. 3), and mass spectrometry.
Since the mass spectrum of deacyl-Y is essentially identical with the mass spectrum of vinelose presented in the previous paper (3), it is not shown hcrc.
All of the results described above support the view that CDPdeacyl-Y is identical with CDP-vinelose. It follows that CDP-Y is a derivative of CDP-vinelose bearing a carboxylic acid residue through nn ester link.
Identijcation of Carboxylic .4cid Residue-Information regarding the molecular weight of the carboxylic acid residue was obtained through comparison of the mass spectrum of sugar Y with that of vinelose.
The sample of sugar Y (Compound VI, Fig. 1  The yield was 24 mg. During these proceses, the nucleoside monophosphate fragment was obt.aincd from t.hc pyrophosphntasc digest as the crystalline barium salt and identified as ?-CLIP essentially as described above for the 5'.CMP fragment obtained from CDPdeacyl-Y. The sugar phosphate fragment (Y-l-P; Compound V, Fig. 1) was also obtained from the pyrophosphxtase digest as the ammonium salt; [U]i4 $3500" (in H20). The yield was 98 pmoles, as est.imated from the phosphate content.
This sample was used for the studies with nuclear magnetic resonance spcctroscopy (see below).
The mass spectrum of sugar Y is presented in Fig. 4. As can be seen, the peak of maximum m/e value is sit.uated at 246. Since, in most instances, " .\I -ILO" fragments, but not. molccular ions, are recognizable in the mass spectra of free carbohydrates  Fig. l), prepared from CDL'-Y by the treatment with hydroxylamine, was indistinguishable from the hydroxamate of 0-methylglycolic acid in its behavior on paper chromatography (Table IV). The nuclear magnetic resonance spectrum of Y-l-l' (Fig. 5; Table V), which gives rise to signals at 3.57 and 4.43 ppm corresponding to the 5 protons of O-methylglycolyl group (i.e. the protons of -OCH, and --CH2-), is a further indication of the structure of the carboxylic acid.
Properties of Sugar Y-It may be appropriate here to make a few comments on the properties of the new sugar, Y.
The Rp values of sugar Y in the solvents indicated in Table  III arc all higher than the respective values of vinelose.
Xs expected from the proposed struct.ure, reagent.s of broad spcci6cit.y for carbosylic acid esters, such as hydrosylamine-Fe& (24), also reveal sugar Y on the chrornatograms.
The infrared spectrum of sugar Y is shown Fig. 3, together with the spectra of vinelose (or deacyl-Y) and 6-dcoxy-3-C-methyl-2-0-methyl-n-allose, one of the 16 possible isomers (i.e. eight disstoreoisomers and their optical enantiomorphs) for vinelose. The bands at 1740 and 1200 cm', which are not seen in the other two spectra, are a strong indication of the presence of au ester linkage in sugar Y. -41~0 to be noted is the fact that the spectrum of vinelosc (or deacyl-Y) is distimt from that of 6deoxy-3-C-methyl-2-0-methyl-n-allose.
Position of 0-Methylglycolyl These results then suggest that the Omcthylglycolyl group is located at position 4 of the vinelose residue since with a &substituted vinelose uptake of 1 mole of periodate would occur at C, to Cs.
Evidence that supports the shove interpretation was also provided by the studies with nuclear magnetic resonance spectroscopy.
When the spectrum of Y-l-1' (Fig. 5; Table V) is compared with that of deacyl-Y-l-P ( Fig. 2; Table II), the Cd-H signal indicates a tendency toward shifting to a lower position (paramugnetic shift).
Since it is known that signals from protons attached as in acyl-0-CH normally lie in a lower position than those attached as in HO-CH, the shift on the Cd--II signal in the Y-l-P spectrum is consistent with t.he assignment of the position for 0-methylglycolyl group.
Conj~gurattin of I'inelose-It was reported in our previous paper (3) Fig. 1). The intensities of the peaks are expressed in relation to the m/e 45 peak, which is assigned an arbitrary value of 100%.  The measurement was performed in 0.4 ml of 1)2O with 97.5 pm&s of sample (ammonium salt). TM&', tet.ramethylsilane (external standard).  Since this acid serves as a substrate for L-lactic acid dehydrogenase, an L configuration was suggested for the vinelose in CDP-vinelose.
The use of 5 N HCl at lOO", however, resulted in significant breakdown of lactic acid, and it was difficult to increase the yield of lactic acid over 48%.
To overcome this difficulty, a modified m&hod has been dcveloped, in which the oxidation product is dephosphorylated by alkaline phosphatasc and subjected to determination of D-and L-lactic acid with the use of Leuconostoc D-lactic acid dehydrogenase and bovine heart L-lactic acid dehydrogenase. The method has been applied to vinelose-1-P (and de:@-Y-l-P) as follows: the sugar phosphate (1 pmole) was mixed with 0.04 M acetate buffer, pH 5.0 (500 pl), and 0.  (6) 60 hours. Excess bromine was then removed by aeration and the pH of the solution was adjusted to 8.5 with ammonia.
Alkaline phosphatase (1.5 units) was added to this solution and incubated at 37". After 5 hours, the tube was placed in a boiling water bath for 3 min and then centrifuged to remove the denatured prot.eins. With the supernatant solution, enzymatic analyses of D-and Llactic acid were carried out according to the method of Hiyama, Koga,and Fukui (6) and the method of Hohorst (25), respectively.
The lactic acid (Compound VIII, Fig. 1) formed in this way was shown to have t.he L configuration.
The yields of L lactic acid were 0.95 pmole from vinelose-1-P and 0.97 pmole from deacyI-Y-l-P, in close agreement with the amount. of starting material.
D-Lactic acid was not detected in the reaction mixtures more than trace amounts. D-Lactic acid, added to the test materials after bromine oxidation, gave about 98% of the theoretical recovery. It leaves little doubt, therefore, that the vinelose in the nucleotides has the L-configuration. Information on the configuration of C3 and Cr has been obtained from the kinetic studies of periodate oxidation.
Since it is known that the rate of periodate oxidation is dependent principally on the stereochemistry of the a-glycol group, the above data suggest the presence of a cis-a-glycol group in vinelose-1-P as well as in deacyl-Y-1-7'.
The oxidation was carried out at 20" in 1. The availability of a synthetic sample of 6-deoxy-3-C-methyl-2-O-methyl-D-allose has shown that the sugar differs in position ou paper chromatograms from vinelose (Table III).' The infrared spectra of both samples are also distinct (SIX Fig. 3). Since pairs of optical enantiomorphs would not be readily dist.inguished by t.hesc means, it seems unlikely that vinrlose is in a relation of optical enantiomorph w&h 6-deosy-3-C-methyl-2-0methyl-n-allose.
Thus, it is tentatively concluded that vinelose has the configuration relating to one of the three diastereoisomers, L-altrosc, Lgalactose, and L-talose, although a definite identification must wait until all of the diustereoisomers are synthesized for comparison.

DISCUSSlOE;
In view of the suggest.ion that vinelose has the configuration relating to Laltrose, L-galactose, or L-tnlose, the possible chair conformations of P-tvinelose-1-P are illustrated in Fig. 7. Since nearly all, if not all, of the nucleotidyl sugars isolated from natural sources have been shown to have either a-n or P-L configuration, it is reasonable to assume that the anomeric carbon of t.he trinelose residue has the p configuration. In P-6-deoxy-3-C-methyl-2-0-methyl-caltrose l-phosphate and its L-galactose isomer the bulky groups on Cl, Ca, and Ca must be axial in the Cl conformations (Conformations A and C, Fig. 7), whereas no such 1,3diaxial interactions of bulky substituents are present in the 1C conformations (Conformations B and D, Fig. 7). It follows that 1C is the more likely possible form for these sugars. In Conformation R, Cd-H is in a 1 ,2-diaxial relation with C&H and, in ('onformation D, Cl-11 is in a 1 ,2-diaxial relation with 1 Dr. G. B. Howarth kindly informed IIS that he also compared vinclose with his synthetic sugar by paper chromatography and gas-liquid chromatography of their 0-acetyl derivatives and follnd them to be different. G-H. The absence of such 1,2-diaxial arrangements in vinelose-1-P is indicated by the fact that, in the nuclear magnetic resonance spectrum of this compound, the signals for Cl--H, G-H, C&--H, and Co-H at 6 = 5.29, 3.39, 3.31, and 3.96, respectively, showed a small 5,,2 or J1,5 constant (about 2 cps, see Fig. 2 and Table II).
i\ccordingly, either of the chair conformations of the talose isomer (Conformations E or F, Fig. 7), in which a 1,2-diaxial arrangement of hydrogens is absent, is considered as being the most preferred conformation for /3-r,-vinelose-1-P (the isomers B and D would give larger values of J1,5 and J1.2, respectively).
Experiments on the biochemical significance of CDP-vinelose and its 0-methylglycolic acid derivative are in progress, hut preliminary results support our previous postulate (3) tha.t CDP-vinelose is derived from CDP-glucose via a CDP-6-deoxy-4-keto-hexose.
It was shown previously (2) that a pyrophosphorylase is present in A. tinelan& which synthesizes CDPglucose from CYI'P and a-D-glucose-l-l'.
We have found more recent.ly that an enzyme preparation from A. t&elan&i is able to convert CDP-glucose into a CDP-6-deo>y4-keto-hexose. The product is similar in its chemical properties to CDP-6deoxy-4-keto glucose, an intermediate in the synthesis of CDP-3, 6-didcoxyhcxoses (27, 25), hut it can bc converted by another enzyme fract.ion from .-l. tinelundii into at least two methylated compounds in the presence of S-adenosylmethioninc.* The mechanism whereby such transmethylation occulg has not yet been estnblished experimentally, but it seems likely that the nucleotidyl keto sugar may undergo rearrangement to constitute the site (i.e. 3,4-enediol) to which the methyl group is transferred. Either the enediol nuclrotido or its C-methylated product may 2 hl. Tnkngi, K. Kimata, S. Okuda, and S. Suzuki, unpublished ohservations.