200
Carbohydrate
Research
EIxvicr Publishing Company, Amsterdam
Printed in Belgium
THE CHEMICAL STRUCTURE OF KEFIRAN, THE WATER-SOLUBLE POLYSACCHARIDE OF THE KEFIR
GRAIN
P. KOOIMAN Laboratory of General nod Technical Biology,
Technological Uniaersity, Del/t (Tha Netherlands)
(Received June 30th, 1967; in revised form, December Ilth, 1967) ABSTRACT
From kefir grains, a water-soluble polysaccharide, keliran, was isolated, consisting of approximately equal proportions of galactose and glucose residues. A &o-(146) glucanase from Trichoderma viride was able to fragment the polysaccharide, the main products beingequimolar amounts of glucose and a pentasaccharide, kefirose. The results of methylation studies on both kefiran and kefirose, of periodate-oxidation analysis of kefiran, and of a study on the products of partial hydrolysis of kefirose with acid provided substantial evidence for assigning the following structures to kefirose and kefiran, respectively. ~-D-Gp-(1~2)-~-D-Gal~-(l~)~-D-Ga~~-(l~3)-~-D-Gal~-(l~)-D-G i6)-~-D-Gp-[l--t2(6)]-~-D-Galp-(I--t4)-a-D-Galp-(I--t3)-~-D-~alp-(I--t4)-~-~-Gp-(i--t I %a ?
[
i p-D-Gp
(kefirose)
1 (k&ran)
21
The variety of linkage types may account for the rather poor accessibility of kefiran to enzymic attack. The latter property might be important in the ecology of the kefir grain. INTRODUCTIOfJ
The kefir grain is the inoculating material used for the conversion of milk into the Russian beverage kefir. It consists of yeasts and bacteria, preponderantly Saccharomyces delbriickii and LactobaciIhrs brevis. At least 24% of the dry material of the kefir grain consists of the capsular polysaccharide of Lactobacilhcs breois’. The polysaccharide is a water-soluble glucogalactan containing approximately equal amounts of D-glucose and D-galactose residues’. The present study is concerned with the chemical structure of this polysaccharide. RESULTS AND DISCUSSION
A crude preparation of kefiran was obtained by disintegrating kefir grains with a homogenizer and precipitating the polysaccharide from the viscous filtrate with Carboh-vd. Res., 7 (1968) 200-211
STRUCTURE
201
OF KEFIRAN
ethanol. Purification was achieved by freezing an aqueous solution of the crude material, followed by thawing, when the polysaccharide was obtained as a spongy gel releasing most of the liquid by syneresis. The purified kefiran had [& +68” (c I, water) and contained approximately 50% of galactose and 50% of glucose residues. Neither repeated freezing and thawing, nor fractionation with ethanol, indicated inhomogeneity of the polysaccharide. A molecular weight of 20,600 (corresponding to an average of 127 hexose units) was estimated by using the hypoiodite method. On periodate oxidation, the polysaccharide consumed 1.13 moles of periodate per mole of hexosyl residue, yielding 0.30 mole of formic acid. Reduction of the resulting polyaldehyde and hydrolysis of the polyalcohol yielded a mixture of galactose, tetritol, and glycerol in the proportions of 1.00:1.55:3.08. From the tetritol fraction, D-threitol and erythritol were isolated. Kefiran was methylated by the Haworth procedure and subsequently by the method of Wallenfels and Bechtler”. The methylated polysaccharide had [& + 18” (chloroform). After hydrolysis, four products were detected on paper chromatograms. Separation by preparative paper-chromatography yielded 2,3,4,6-tetra-O-methyl-Dglucose (13%), tri-0-methylglucoses (33%), tri-0-methylgalactoses (38%), and 3,Pdi-0-methyl-D-galactose (16%). The tri-0-methylglucoses were separated on a carbon-Celite column by using a gradient of butanone-water, to afford approximately equal amounts of 2,3,4- and 2,3,6-tri-O-methyl-D-glucose. The tri-O-methylgalactoses were partly separated on carbon-Celite, with a gradient of ethanol-water, to give approximately equal amounts of 2,3,6- and 2,4,6-tri-0-methyl-D-galactose. The results of periodate oxidation and methylation indicate that kefiran has a main chain with branch points at O-2 and O-6 of some of the D-galactose residues; the branches are terminated by D-glucose residues. The remaining D-galactose residues are linked through positions 4 and 3, and the remaining D-glucose residues are linked through positions 4 and 6. All of these linkage types are present in approximately equal amounts. Several, crude enzyme-preparations from various sources failed to hydrolyze kefiran; only a crude cellulase from Trichodernra Gride was able to degrade the polysaccharide, during prolonged incubation, to give mainly glucose and a pentasaccharide, in approximately equimolar quantities. The pentasaccharide, kefirose, had [& +90” (water) and contained three galactose and two glucose residues. After reduction with sodium borohydride and hydrolysis, the ratio of galactose to glucose was 3:1, indicating that the reducing end of kefirose is a glucose residue. Methylation of kefirose gave a product having [a]n +35” (chloroform), which, after hydrolysis, gave six products (paper chromatography). Fractionation by preparative paper-chromatography yielded 2,3,4,6-tetra-O-methyl-D-glucose, 2,3,6tri-O-methyl-D-glucose, a mixture of 2,3,6- and 2,4,6-tri-O-methyl-D-gaiactose, 3,4,6-tri-0-methyl-D-galactose, 2,6-di-0-methyl-D-galactose, and 4,6-di-O-methylD-galactose, in molar proportions of 1.O:1.2: 1.9: 1.0:0.6:0.1. The presence of di-0-methylgalactoses in the mixture is attributable to incomcarbohyd.
Res.,
7 (1968)200-211
202
-P.mKOOIMAN
plete methylation. Apart from the di-O-methylgalactoses, the results demonstrate that kefirose has the same linkage types as kefiran, except for the 6-linked D-glucose residue and the branching point at one of the D-galactose residues, which are not present in kefirose. One non-reducing end-group (a D-glucose residue as in kefiran) is present in kefirose, and the 2,6_disubstituted D-galactose residue in kefiran corresponds with a Zsubstituted D-galactose residue in keflrose. Thus, kefirose is a straight-chain pentasaccharide terminated by D-glucose residues, and having three interior D-galactose residues_ In order to obtain information on the sequence of the linkage types, kefirose was fragmented by mild hydrolysis with acid, and the oligosaccharides produced were separated by paper chromatography. The results are recorded in Table I. Fraction I was 2-O-P-D-glucopyranosyl-D-galactose originating from the non-reducing end of kefirose. Fraction II contained, inter alia, 4-O-P-D-galactopyranosyl-D-glucose (lactose), which could only have originated from the reducing end of kefirose. The non-crystalline part of Fraction II ([a],, t76”) yielded galactose and a small proportion of glucose on hydrolysis; the main component was therefore a galactosylgalactose or a mixture of galactosylgalactoses, and the optical rotation was indicative of at least some CL-Dlinks. From this material3 a small quantity of I-U-/3-D-galactopyranosyI-D-galactose was isolated. TABLE
I
OLIGOSACCHARIDES
Fraciion
OBTAINED
Yield (mg)
BY
PARTIAL,
Componenr
ACID
Mol.
srrgars
HYDROLYSIS
weight
OF KEFIROSE
[x]D
(rCa?er)
63 180
Gal + G Gal + less G
363 -
+ -
III IV v VI VII
105 42 61 32 49
Gal Gal Gal Gal Gal
526 521 702 678 825
+ 42 t139 + 95 + 108 + 8.5
+ less G + less G + G
G)
(degrees)
I II
+ G
(1
32
Fraction Ill was a trisaccharide, probably a glucosylgalactosylgalactose, and thus represented three sugar units from the non-reducing end of kefirose; the specific optical rotation was suggestive of 8-D links. Fraction IV was a trisaccharide consisting of galactose residues, corresponding to the interior of the kefirose molecule remaining after loss of the glucose endgroups. According to the methylation analysis, the following structures were possible: (a) ~-Gal&l +4)-D-Galp-(l-3)-D-Galp and (6) D-Gab+ +3)-D-GaI;I-(l-4)D-Galp. Structure (a) should consume 6 moles and (6) 5 moles of periodate; moreover, (a) would leave no intact D-galactose and (b) would leave one mole of D-galactose. The observed consumption of periodate was at least 5.5 moles per mole of hexosyl residue; only a trace of intact galactose was found in the hydrolyzate of the boroC’arhhyd.
Res., 7 (1968) 200-211
203
STRUCTURFi OF KEFIRAN
hydride-reduced product. Fraction IV therefore had structure (a>. The rather high optical rotation was indicative of CL-D linkages. Since Fraction III, containing three residues from the non-reducing part of the kefirose molecule, probably has exclusively P-D linkages, it is likely that the (l-3)-bond is of the a-D type, all other bonds of the kefirose molecule being of the /I-D type. Fraction V was a glucosylgalactotriose, while Fraction VI was a galactotriosylglucose. Both Fractions V and VI had rather high optical rotations, indicating the presence of an CL-D linkage in addition to the B-D linkages. This was in harmony with the structures of the other oligosaccharides. Fraction VII was unchanged kefirose. From the yields of the oligosaccbarides, it may be concluded that, during acid hydrolysis, the weakest bond in the kefirose mo!ecule is the a-D-(1 33) linkage. The evidence obtained permits the assignment of the following structure for kefirose:
The release of kefirose and glucose from kefiran by the Trichoderma enzyme implies that kefiran contains hexasaccharide repeating-units, consisting of kefirose to which a D-glucosyl residue is linked at position 6 of the 24inked D-gaIactosy1 residues. The difference in specific optical rotation between kefiran and kefirose indicates that the linkages split by the enzyme are of the B-D type. It follows that there are two possibilities for the structure of kefiran:
1 L
or (b)
J 21
P-D-Gp
I -A)-x-D-Galp-(1
‘k-6)-fi-D-Gp-(1+6)-P-D-Galp-(
-z-3)-/3-D-Galp-(1+4)-QD-Gp-(I-+
I
T
/3-D-Gp
21
Each of these structures would consume 1.17 moles of periodate per mole of hexosyl residue (found 1.13) and produce 0.33 mole of formic acid (found 0.30). After periodate oxidation, reduction of the pdlyaldehyde, and hydrolysis, galactose, erythritol, D-threitol, and glycerol would be keeleasedin ratios of 1: 1: 1:3; the observed ratios of galactose, tetritols, and glycerol were 1.00: 1.55:3.08. The linkages which would be split by the Trichoderma enzyme in either (a) or (b) are indicated by arrows. Obviously, the enzyme responsible for the hydrolysis is a /I-( 1+6)-glucanase Ip-( 1+6)-glucan 6-glucanohydrolase), which is known to Carbohyd. Res., 7 (1968)
200-211
P. KOOIMAN
204
be present, although in small proportions, in Trichoderma species, including T. uiride’. Considerably greater amounts of this enzyme have been found in many Penicillium species, in several Aspergillus species, and in other fungi2. However, kefiran was not attacked by several crude fungal and bacterial enzyme-preparations; one of these was derived from Aspergillus niger, of which species at least one strain produces a /I-D-(1+6)-g1ucanase2. Attempts to find micro-organisms which were able to hydrolyze the polysaccharide led to the isolation of bacteria identified as belonging to the genus Arfhrobacter; they attacked the polysaccharide slowIy3. The resistance of kefiran to enzymic attack is possibly due to the succession of different linkage types and the presence of some uncommon linkage types, and explains the capacity of kefir grain to survive microbial contamination. EXPERIMENTAL Melting points are corrected. Specific optical rotations were equilibrium rotations unless otherwise specified. Evaporations were done below 30”. Paper chromatograms were made by the descending method on either Whatman No. 1 or 3 MM papers, with the following solvents (v/v): (A) butyl alcohol-pyridine-water (6:4:3); (B) butyl alcohol-ethanol-water (40: 11: 19); (C) butanone-water (2: 1) ; (0) butyl alcohol saturated with water. For detection of sugar spots, anisidine phosphate4 was used; sugar alcohols were detected with a silver nitrate reagent’. In addition to the former reagent, triphenyltetrazolium chloride6, aniline-diphenylamine’, and periodate-benzidine’ were also used. Preparation of kefiran and determination of rhe sugar composition. - The polysaccharide was isolated and purified by repeated precipitation with ethanol, followed by two freezing and thawing treatments as described earlier’. During the two latter treatments, ca. 15% of the material was lost. Third and fourth freezing and thawing treatments gave losses of 0.5 and O.l%, respectively. The resulting preparation was dissolved in water to form a 1% solution. Fractional precipitation with ethanol yielded four fractions (Table II). AII subsequent experiments were carried out with the preparation obtained after two freezing and thawing treatments. TABLE II FRACTIONATION
OF
Precipitate front
20% Ethanol 30% Ethanol 40% Ethanol
Dissolved residue
KEFIRAN
WITH
ETHANOL
Yield (%)
25 57 15 3
Glacose in hydrolysate (%)
49.7 50.5 50.8 -
[z]!]D(degrees)
+ 65.8 + 66.6 + 65.9 f72.6
The polysaccharide was hydrolyzed, and the sugar composition of the hydrolyzate was determined by a modified hypoiodite methodg, before and after removal of glucose with glucose oxidase (D-glucose aerodehydrogenase, Boehringer). Determination of molecular weight. - The hypoiodite method was used’. Carbohyd. Res., 7 (1968) 200-211
STRUCTURE
205
OF KEFEAN
Periodate oxidation. Periodate consumption and formic acid production were determined by the method of Chanda et al.lo . In Table III, the results are given in moles per mole of hexosyl residue. The data were extrapolated to zero time to TABLE PERIODATE
III CONSUhfPlTON
hfOLE
OF HEXOSYL
The
(days)
AND
FORMIC
ACID
PRODUCTION
DURING
OJUDATION
OF
REFIRAN
(MOLES
PER
RESIDUE)
Periodate consumption Formic acid production
I
3
6
IO
14
I7
21
1.13 0.277
1.17 0.310
1.27 0.323
1.22 0.349
1.27 0.360
1.32 0.376
1.36 0.408
obtain net values for periodate consumption and formic acid production. Kefiran (100 mg) was oxidized with 0.02~ periodate (100 ml) for 8 days at room temperature. Barium chloride hydrate (490 mg) was added to precipitate periodate and iodate, and, after several days, the salts were removed by filtration. Potassium borohydride (300 mg) was added to reduce the polyaldehyde, and, after a day, the solution was acidified with acetic acid. The solution was passed through a mixed bed of ionexchange resins Amberlite IR-120 and IR-45, and the eluate was concentrated to dryness. Boric acid was removed by repeated evaporation of methanol from the residue. The residue (88 mg) was hydrolyzed in 0.1~ sulphuric acid (10 ml) for 4 h at 90”, and neutralized with barium carbonate, and the filtered solution was concentrated to a small volume. The mixture was fractionated by paper chromatography with solvent (A). The following components were eluted from the paper strips. D-Galactose, m-p. 163-165”, [a]n + 77.9” (water); lit.“*“, m.p. 167”, [cx]u+ 81.1”. Erythritol(5 mg) crystallized spontaneously when the tetritol fraction (12 mg) was dissolved in ethanol-ethyl ether. After recrystallization, the m-p. and mixed m.p. was 118” (lit. l2 118.5”). The mother liquor, on treatment with benzaldehyde in the presence of hydrochloric acidi3, gave di-O-benzylidene-D-threitol, m-p. and mixed m-p. 221”. Glycerol was identified as its tris(p-nitrobenzoate)“, m-p. and mixed m.p. 192-194”. &fethyZationof kefian. - Kefiran (10 g) was methylated in the usual way by the Haworth procedure. After extraction of the partly methylated product with chloroform from the neutralized soiution, the solvent was removed, and the product was dissolved in methyl sulphoxide (50 ml). Methylation was completed by two consecutive treatments with methyl iodide and barium oxide”. The methylated polysaccharide was extracted with chloroform from the semi-solid mixture, and the extract was shaken with sodium thiosulphate solution to remove iodine. The chloroform solution was concentrated, the syrupy residue was dissolved in benzene (25 ml), and an equal volume of light petroleum was added. A small amount of impurity was removed by centrifugation, and light petroleum (200 ml) was added to precipitate the bulk of the methylated product. The material was washed with 10% chloroform in hexane and redissolved in chloroform-benzene (2:3,25 ml). The solution was poured Carbobyd. Res., 7 (1968) 200-211
206
P. KOOIMAN
into light petroleum (200 ml) with stirring. The precipitate was washed with chloroform-hexane and pentanc, and finally dried, The product (10.4 g) did not show hydroxyl absorption in the infrared. (Found: OMe, 41.8%). Hydrolysis of metlrylated kefran. Methylated kefiran (2 g) was dissolved in 70% sulphuric acid (10 ml) and then diluted to 100 ml. The solution was heated for 4 h at iOO”, neutralized with barium carbonate, filtered, and concentrated. The methyl sugars were separated by paper chromatography in solvent (B); the following fractions were obtained. Fraction I (175 mg) was 2,3,4,6-tetra-O-methyl-D-glucose. After recrystallization from pentane, it had m-p. 95-96”, [c& +81.5” (c 1.6, water); lit.r6, m.p. 96”, [LY],,+84”. The “anilide” had m-p. 136-137”; lit16, m-p. 137-138”. Fraction 2, on demethylation”, yielded glucose with a trace of gaIactose. On paper chromatograms in solvents (B) and (C), this fraction had the same mobility as 2,3,6-tri-O-methyl-D-glucose, although the colour developed with anisidine rcagent was not the same. The results of periodate oxidation of kefiran indicated the presence of both (1+4) and (I-6) linkages; therefore, both 2,3,4- and 2,3,6-t&O-methylhexoses could be expected in the hydrolyzate of the methylated polysaccharide. Fraction 2 was therefore expected to contain 2,3,4- as well as 2,3,6-tri-O-methylD-glucose. The product (600 mg) was subjected to fractionation on a carbon-Celite column with a gradient of butanone-water”. The subfractions obtained were tested by applying drops on filter paper and developing with anisidine reagent; subfractions yielding similar colours were combined to give four fractions. The first of these (157 mg) contained 2,3,6-tri-O-methyl-D-glucose contaminated with tri-O-methylgalactoses. The second preparation (107 mg) contained exclusively 2,3,6-tri-o-methylD-glucose, m-p. 111-l 12”, [a],, +7O”(c 1.6, water); lit.“, m.p. 121-123”, [a]b +70.5”; its bis(y-nitrobenzoate) had m-p. 191-192” (litzO, m-p. 189-1900). The third preparation (96 mg) was a mixture, and the fourth one (196 mg) was 2,3,4-tri-O-methyl-D-glucose, [aID +72” (c 3.4, acetone): lit.16, +70.5”. The “anilide” had m-p. 147-149”; lit.2’*22, m-p. 145-146” and 150”. Fraction 3 yielded exclusively galactose on demethylation. In solvents (B) and (C), the material behaved as a single compound, although the results of periodate oxidation and the identity of fraction 4 indicated the presence of 2,3,6- and 2,4,6-triO-methyl-D-galactoses. Attempts at separating the tri-0-methylgalactoses (800 mg) on a carbon-Celite column with a gradient of ethanol” resulted in a partial separation. The first fraction (107 mg) contained material eluted before the peak attained its maximum, and was 2,4,6-tri-0-methyl-D-galactose, m-p. 102-104”, [aJD+86” (c 1.4, m.p. water); lit.*j, m.p. lO2-105”, [LX],, t89” (water). It gave an “anihde”, 171-172” (Iit.24, m.p. 170-179”). The corresponding amide, prepared in the usual m.p. 167”). The way, had m-p. 166” (lit.25, 2,4,6-tri-0-methyl-D-galactonamide, second fraction (117 mg) was the eluate around the maximum of the peak; part of it (47 mg) crystallized and was 2,4,6-tri-O-methyl-D-galactose. The third fraction contained the material (318 mg) trailing behind the maximum of the peak; it was oxidized with bromine water in the presence of sodium carbonate to give 2,3,6-tri-OCurbohyd. Re.s., 7 (1968) 200-211
STRUCTURE
207
OF KERRAN
methyl-D-galactono-1,5-lactone, m-p. 99” (from ether-pentane), [a],, - 38” (c 3.1, water); lit.26*27,m-p. lOO”, [a]n -37” (water)_ Fraction Q was a methyl ether of galactose as indicated by demethylation, and its mobility on paper chromatograms suggested that it was a di- 0-methylgalactose. The sugar crystallized from ethanol-acetone-ethyl acetate and had m-p. 109-116”. Recrystallization from 95% ethanol-ether yielded a modtication having m.p. 16917i”, [a]n +92.6(10 min)+ + 115.6”(24 h) (~3.6, water); 3,4-di-O-methyl-D-galactose28 has m.p. 164-166”, [a]n +95+ 117” (water). There was no depression of m.p. on admixture
with
authentic
material.
Samples of Fractions 2 (2,3,4- and 2,3,6-tri-O-methyl-D-glucose) and 3 (2,3,6- and 2,4,6-t&0methyl-D-galactose) (50 mg) were dissolved in water (1 ml) and reduced with potassium borohydride (75 mg in 2 ml of water) overnight. The solutions were then acidified with acetic acid, deionized with Amberlite IR-120 (H+), and concentrated to dryness. Methanol was evaporated several times from the residue to remove boric acid. Reducing end-group estimation by the hypoiodite method showed that in both preparations 96% of the reducing groups had been reduced by the borohydride. The tri-0-methylalditol mixtures were dissolved in water (20 ml), and to a sample (0.8 ml) was added 0.1~ sodium metaperiodate (0.4 ml). After 30 min at room temperature, periodate consumption” and formaldehyde production2’ were determined. The tri-0-methylgalactitol mixture consumed 0.49 mole of periodate per mole, with the production of a trace of formaldehyde. This means that half of the quantity of tri-0-methylgalactitol was resistant to periodate, the other half consuming one mole per mole. Since 2,4,6-tri-0-methyl-D-galactitol is expected to be resistant towards periodate, whereas the 2,3,6-isomer should consume one mole, it is concluded that the two tri-0-methylgalactoses were present in equimolar amounts. The tri-0-methylglucitol mixture consumed 0.71 mole of periodate per mole while 0.35 mole of formaldehyde was produced_ On complete oxidation, a periodate consumption of 1.OOmole was expected_ Pure 2,3,6-tri-O-methyl-D-glucitol consumed 0.81 mole of periodate per mole. When the consumption by this compound in the mixture (0.71-0.35 = 0.36 mole) is corrected for incomplete oxidation, a value of 0.44 is obtained; this means that 44% of.the mixture was 2,3,6-tri-O-methyl-Dglucitol. Pure 2,3,4-tri-O-methyl-D-glucitol consumed 0.93 mole of periodate per mole and gave rise to 0.89 mole of formaldehyde. Correction of the formaldehyde produced from the tri-0-methylglucitol mixture indicated that 39% of the mixture consisted of 2,3,4-tri-0-methyl-D-glucitol. Enzymic hydrolysis of kejiran. The following, crude enzyme-preparations were tested for their ability to hydrolyze the polysaccharide: Myrothecium cellulase3’; Pektolyt (De Betuwe, Tiel), a crude commercial preparation derived from Aspergiflus niger; Luizym (Luitpoldwerke, Munich), derived from Aspergillus oryzae; Hydralase D (Jacques Wolf, Passaic, N.Y.), from Aspergillus oryzae; Rapidase, an a-amylase from Baciifus subtilis; Rhozyme, an a-amylase probably from BaciIhts stearothermophihcs; Takadiastase (Parke, Davis and Co., London), from Aspergillrrs oryzae; Attempted estimation of the amounts of tri-O-methyl sugars. -
Carbohj~L Res., 7 (1968) 200-211
208
P. KOOIhfAN
Hemicellulase (Mann Research Laboratories, New York) of fungal origin; Hydrolase (Calbiochem, Los Angeles) from Aspergillrrs oryzae; Helisol (Swiss Ferment Camp., Basle, Switzerland); Cryprochiton stelleri enzyme, lyophilized juice from the sugar gland31n32*Trichoderma uiride cellulase (Meiji Seika Kaisha Ltd., Tokyo), a crude preparatidn. Solutions (1 ml, 3%) of kefiran were incubated with the enzymes (10 mg; in the case of Myrotheciwn cellulase, 0.1 ml) and incubated for several weeks at 37”; a drop of toluene was added to prevent microbial contamination. The viscous solution was liquefied by the Trichoderma enzyme, whereas none of the other enzymes showed any activity. On inspection by paper chromatography of the hydrolyzate of kefiran with Trichoderma cellulase, glucose, a trace of a disaccharide, and a slow-moving oligosaccharide were present. A blank experiment (cellulase incubated without substrate) demonstrated that no sugars were produced from the enzyme preparation. Preparation of kefrose. Trichoderma cellulase (1 g) was dissolved in water (10 ml) and dialyzed for several hours at 0”. The solution was brought to 20% saturation with ammonium sulphate, and the precipitated material was removed. The filtrate was adjusted to 60% saturation, and the precipitate was dissolved in water (5 ml) and added to a solution of kefiran (5 g) in water (100 ml). The solution was incubated at 35” (with addition of toluene), and samples were taken periodically for determination of reducing end-groups by the hypoiodite method. After 190 days, the increase in reducing power was negligible; by that time, the percentage of hydrolysis was 42%, whereas determination (hypoiodite method in conjunction with glucose oxidase) of the amount of glucose released was 19.1%. The hydrolyzate was fractionated on a carbon-C!elite column by elution with ethanol-water mixtures33. Kefirose (3.4g) was eluted with 20-40% ethanol. Endgroup determination with hypoiodite yielded a value for the molecular weight of 814 (talc. for a pentasaccharide, 828). The sugar composition was determined in the hydrolyzate, before and after reduction with borohydride, as described for kefiran. Methylation of kefiose and hydrolysis of methylated kefrose. The oligosaccharide was methylated by the Haworth procedure followed by the Purdie method. The product contained cu. 47% of methoxyl group (talc., 49.5%), as estimated from the hydroxyl absorption in the infrared by comparison with 2,3,4,6-tetra-O-methyl-Dglucose. Repeated methylation by the Purdie method did not result in a decrease of the hydroxyl peak near 3500 cm-‘. After hydrolysis of the product (100 mg) in N sulphuric acid (5 ml) for 4 h at lOO”, paper-chromatographic inspection of the neutralized (BaCO,) hydrolyzate in solvents (B) and (C) showed the presence of six sugar spots. The components were separated on sheets of filter paper with solvent (B), the following fractions being obtained. Fraction (a) (17 mg) was 2,3,4,6-tetra-O-methyl-D-glucose, m.p. 85-86” (from pentane); lit.16 96”. The “anilide” had m.p. 137-138”; lit.r6 137-138”. Fraction (6) (19 mg) crystallized from ethanol-ethyl ether-light petroleum and had m.p. 110-111”; ht.” for 2,3,6-tri-O-methyl-D-glucose, 121-123”. The bis(pnitrobenzoate) had m-p. 190-191”; lit.” 189-190”. C’nrbohyd. Res.,
7 (1968)
200-211
STRUCTURE
OF KEHRAN
209
Fraction (c) (31 mg) was a mixture of 2,3,6- and 2,4,6-tri-O-methyl-D-galactose. It was partly resolved into its components, as described for the mixture of t&Omethylgalactoses from kefiran. The first part of the eluate contained 2,4,6-tri-Umethyl-D-galactose, m.p. 102-104” (lit.23 102-105”), which yielded an “anilide”, m.p. 170” (lit.24 170-I 79”) . From the last part of the eluate, the corresponding lactone was prepared, m.p. 99” (lit.26 for 2,3,6-tri-O-methyl-D-galactono-l,Hactone, 100°). Fraction (d) (16 mg), on demethylation, yielded galactose, di-U-methylgalactose, and mono-0-methylgalactoses, as disclosed by paper chromatography in solvent (B). The methyl sugar gave a bright spot on paper chromatograms sprayed with triphenyltetrazolium chloride, indicating that the C-2 hydroxyl group was unsubstituted. It crystallized from ethanol-ethyl acetate-pentane and, after recrystallization, had m.p. 87-89”, falD i-144 (10 min)-, f 110” (c 1.3, water) (lit.34 for 3,4,6tri-O-methyl-D-galactose, m.p. 88-89”, [a&, -I- 154-+ + 1 IO”). Fraction (e) (9 mg), on demethylation, yielded galactose. The substance crystallized from ethanol-ethyl acetate-pentane and, after recrystallization, had m.p. 126”. It reacted weakly with triphenyltetrazolium chloride, which indicated that the C-2 hydroxyl group was substituted. On paper chromatograms (solvent A), the sugar migrated at the same rate as authentic 2,6-di-O-methyl-D-galactose, and the colours developed with the anisidine reagent were similar. On paper elcctrophoresis in 0.05~ sodium borate, both substances had MG 0.29. The sugar was thus identified as 2,6di-O-methyl-D-galactose (lit.35, m.p. 128-130”). Fraction
P.
210
KOOIMAN
borohydride and hydrolysis, only glucose was present. The glucosylgalactose had m.p. 171-172”, [& + 32” (c 1.7, water); lit. 37 for 2-O-#LD-glucopyranosyl-D-galactose, m.p. 171-172”, [a]n +42.6”. The sugar did not react with the triphenyhetrazolium chloride reagent, and its reducing power towards Somogyi reagent was 2.5% as compared to glucose; this behaviour was expected for a (l-+2)-linked disaccharide38*3g. Fraction II was a mixture of disaccharides, part of which (90 mg) crystallized. The crystalline material yielded glucose and galactose on hydroIysis, and galactose after reduction and hydrolysis. The recrystallized galactosylglucose had m-p. 220-221 O, [oL]~f53” (c 2.0, water); lit.4o for 4-O-/I-D-galactopyranosyl-D-glucose, m.p. 223”, [t~]n +55.3”. The mother liquor contained 90 mg of disaccharides which had [& t76” (c 2, water). Part of this material (23 mg) crystallized and yielded exclusively galactose on hydrolysis. After recrystallization, the sugar had m.p. 209-210”, [a&, +82 (10 min)+ f63” (c 2.3, water); lit.41 4-0-/3-D-galactopyranosyl-D-galactose, m-p. 210-212”, [a]n +67”. Fraction III was a trisaccharide containing glucose and (more) galactose residues. After borohydride reduction and hydrolysis, both sugars were present in approximately equal amounts, as judged from the spots on paper chromatograms. Probably, the sugar was a glucosylgalactosylgaIactose and not a galactosylglucosylgalactose, since such a trisaccharide could not be formed from kefirose. The oligosaccharide of fraction IV was a galactose trisaccharide. The sugar (10 mg) was dissolved in water (2 ml) and reduced with potassium borohydride (30 mg) overnight. By paper-chromatographic examination along with known quantities of the oligosaccharide, it could be deduced that not more than 2% of the reducing groups were present after reduction. After recovery of the reduced compound, it was dissolved in water (5 ml) and 0.1~ sodium metaperiodate (3 ml) was added. Periodate consumption was 5.6 moles per mole of trisaccharide after 1.5 h; 6.4 moles after 4 h, and 6.9 moles after 6.5 h. Extrapolation of these values to zero time yielded a minimum value of 5.5 moles of periodate per mole of trisaccharide. The remainder of the solution was reduced with borohydride (30 mg) overnight; the solution was treated as usual, and the deionized product was hydrolyzed. The neutralized hydrolyzate was examined by means of paper chromatography (solvent A). Fractions V and VI were tetrasaccharides of glucose and (more) gaiactose. The hydrolyzate of reduced V still contained more galactose than glucose, whereas the hydrolyzate of VI contained exclusively galactose. ACKNOWLEDGMENTS
Expert technical assistance by Miss D. C. Reuvers is gratefully acknowledged. Gifts of 3,4-di-0-methyl-D-galactose by Dr. D. J. Bell (Edinburgh), of 2,6- and 4,6-di-0-methyl-D-galactoses by Dr. D. A. Rees (Edinburgh), of Trichoderrna cellulase by Dr. Nobuo Toyama (Miyazaki), and of Cryptochiton enzyme by Prof. B. J. D. Meeuse (Seattle, Washington) are highly appreciated. The infrared spectra Cmbohyd. Res.. 7 (1968)
200-211
211
STRUCTURE OF KE&RAN
were obtained by Dr. A. van Veen (Delft). Thanks aregiven to Dr. J. W. M. la Rivi&re (Delft) for providing kefiran and for valuable discussions on the subject. REFERENCES 1 J. W. M. LA RIVI~RE, P. KOOIMAN, AND K. SCHMIDT, Arch. Mikrobiol., 59 (1967) 269. 2 E. T. REESE, F. W. PARRISH.AND M. MANDELS, Con. J. Microbial., 8 (1962) 327. 3 J. W. M. LA RIV&RE, in preparation. 4 S. MUKHERJEEAND H. C. SRIVASTAVA, Natare, 169 (1952) 330. 5 W. E. TREVELYAN, D. P. PROCTER,AND J. S. HARRISON, Natare, 166 (1950) 444. 6 K. WALLENFELS,Naturroissenschaften, 37 (1950) 491. 7 J. L. BUCHAN AND R. I. SAVAGE, Analyst, 77 (1952) 401. 8 J. A. CIFONELLIAND F. SMITH, Anal. Chem., 26 (1954) 1132. 9 R. WILLSTX~R AND G. SCH~~DEL,Ber., 51 (1918) 780. 10 S. K. CHANDA, J. K. N. JONES, AND E. G. V. PERCIVAL, J. Cilem. Sot., (1950) 1289. 11 C. N. RIIBER, Ber., 56 (1923) 2185. 12 A. S. PERLINAND C. DRICE, Can. J. Gem., 33 (1955) 1216. 13 R. C. HOCKEY, J. Am. Chem. SOL, 57 (1935) 2260. 14 A. M. UNRAU, Con. J. Cheat., 42 (1964) 916. 15 K. WALLENFELSAND G. BECHTLER,Attgew. Chem., 75 (1963) 1014. 16 J. C. IRVINE AND J. W. H. OLDHAM, J. C/rem. Sot., 119 (1921) 1744. 17 L. HOUGH, J. K. N. JONES, AND W. H. WADMAN, J. Clrem. Sot., (1950) 1702. 18 B. LINDBERGAND B. WICKBERG, Acta Cltem. Sca&_, 8 (1954) 569. 19 J. C. IRVINE AND E. L. HIR~T, J. Cilem. Sot., 121 (1922) 1213. 20 P. A. REBERS AND F. SMITH, J. Ant. Chem. SW., 76 (1954) 6097. 21 S. PEAT, E. SCHL~CHTERER, AND M. STACEY. 1. Chem. Sot., (1939) 581. 22 5. D. GEERDES. B. A. LEWIS, AND F. ShrITw, J. Am. Chem. SOL, 79 (1957) 4209. 23 D. J. BELL AND S. WILLIAMSON, J. Chem. Sot., (1938) 1196. 24 F. BROWN, E. L. HIR~T, AND J. K. N. JONES, J. Cltem. Sot., (1949) 1757. 25 E. G. V. PERCIVALAND J. C. SohIEwILLE, J. Chem. Sor., (1937) 1615. 26 L. HOUGH AND D. B. POWELL, J. Chem. Sot.. (1960) 16. 27 J. K. N. JONES AND M. B. PERRY, J. Am. Gem. Sot., 79 (1957) 2787. 28 J. S. D. BACON AND D. J. BELL, J. Chem. Sot.. (1939) 1869. 29 M. LAMBERTAND A. C. NEISH, Can. J. Research. 28B (1950) 83, 30 P. KOOIMAN. P. A. ROELOFSEN, AND S. SWEERIS, Eazymologia, 16 (1953) 237. 31 B. J. D. MEEUSEAND W. FLUEGEL, Natare, 181 (1958) 699. 32 B. J. D. MEEUSEAND W. FLUEGEL. Arch. Neerf. ZooI., 13 (1958) Suppl. 1, 301. 33 R. L. WHISTLER AND D. F. DURSO, J. Ant. Chem. Sot., 72 (1950) 677. 34 R. KUHN AND H. H. BAER, Ber., 88 (1955) 1537. 35 J. W. H. OLDHAM AND D. J. BELL, J. Am. Cltem. Sot., 60 (1938) 323. 36 J. S. D. BACON, D. J. BELL, AND J. LORBER, f. Chem. Sot., (1940) 1147. 37 A. M. GAKHOKIDZE AND N. D. KUTIDZE, J. Cell. Chem. USSR (Engl. TramI-), 22 (1952) 167. 38 S. HAQ AND W. J. WHELAN, Natare, 178 (1956) 1222. 39 D. J. BELL, in Modern Methods of PIant Analvsis, K. PEACH AND M. V. TRACEY (Eds.), Vol. 2, Springer, Berlin, 1955, p. 9. 40 J. R. CLAMP, L. HOUGH, J. L. HICKSON. AND R. L. WHISTLER, Aduan. Carbohyd. Chem., 16 (1961) 159. 41 H. 0. BOWENG Ahm H. MEIER, Acto Chem. Stand.. 13 (1959) 1884.
Carboltyd.
Res., 7 (1968) 200-211