[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

93

[7] C h e m i c a l S y n t h e s i s o f O l i g o s a c c h a r i d e s B y HAROLD M . FLOWERS

Biological Importance. The most abundant natural material is a polysaccharide--cellulose---and the most abundant saccharides in nature are homopolymers--cellulose, chitin, starch. However, these are structural or storage materials and represent rather an unreactive type of material in vivo. By far the majority of biologically interesting sugar-containing compounds do not have the simple, repeating structure of a single sugar unit fundamental to these homopolymers. There may be larger, more complex repeating units, as in glycosaminoglycans and glycosaminoglycuronoglycans, e.g., mucins, hyaluronic acid, chondroitins; 1 more diverse sugar-building-blocks but still with a repeating pattern appear in the lipopolysaccharides from gram-negative bacteria, where as many as 8 or 9 different sugars may be found in one compound, but some of them--the O-antigen chain form part of a repeating unit. 2 On the other hand, many naturally occurring saccharides, especially those found in animals, are glycolipids or glycoproteins with ramifying structures, t A relatively small number of different sugars--D-glucose, D-galactose, D-mannose, D-glucosamine, D-galactosamine, L-fucose, and sialic acid--are the common constituents, although other sugars, such as L-arabinose and D-xylose, may occur; the sugar units are combined in branched chains. While schemes for the synthesis of homopolymers may be devised, in principle, involving simple polymerization reactions of suitable monomers, it is obvious that the more complex materials cannot be made in this way and that it would be necessary to join the sugar components together in controlled fashion in the requisite order. Sugars usually occur bound glycosidically either to other sugars or to noncarbohydrate materials. C-Glycosides are known, and N-glycosides include the well known nucleic acids and their derivatives; however, it is the O-glycosides that are of greatest interest, especially in the study of enzymes: little is known, for instance, about enzymes effecting the hydrolysis of C- or N-glycosides. I shall therefore limit my discussion to the synthesis of compounds bearing the O-glycosyl linkage. O-Glycosides result, formally, from the condensation of the reducing (anomeric) hydroxyl group of a sugar with the hydroxyl group of another molecule. The latter molecule can be a simple alcohol, a hydroxylated amino acid or another sugar, or even more complex materials containing R. G. Spiro, Adv. Protein Chem. 27, 349 (1973). z O. Liideritz, K. Jann, and R. Wheat, in "Comprehensive Biochemistry" (M. Florkin and E. H. Stotz, eds.), Vol. 26A, p. 105. Elsevier, Amsterdam, 1968.

94

PREPARATIONS

[7]

these constituents. In compounds containing 2 or more sugar units bound together, a variety of positions of linkage are possible depending on the hydroxyl groups involved. In a polysaccharide having only a few different sugars, the theoretical possibilities for different compounds soon become extremely large. Additional isomers can result from the stereochemistry of the glycosyl linkage, and both ~- and fl-anomers occur. In fact, some glycoproteins and glycolipids contain in the same molecule sugar units (such as D-galactose, n-mannose, or n-galactosamine) in both a- and fl-forms. The tremendous number of possibilities is not apparently realized in nature, as certain rules seem to apply about the arrangement of sugar moieties and the types of linkage they exhibit: for example, sialic acids and L-fucose are or-linked and usually at nonreducing termini in glycoproteins and glycolipids; sialic acid is attached to o-galactose or n-galactosamine; generally the attachment of a complex saccharide chain to asparagine in glycoproteins is via a unit of di-N-acetylchitobiose. The stereochemistry (anomeric configuration) of the glycosyl linkage is recognized specifically by glycosidases as well as by other biological molecules (lectins, antibodies, hormones), and glycosidases also show some selectivity with regard to the points of linkage to the sugar rings. The importance and specificity of sugars in the biology of complex molecules is becoming abundantly clear from recent work on such topics as the homing of lymphocytes, 3 the removal of soluble glycoproteins from the circulation, 4 cellular adhesion, ~ immunological specificity, 6 and the binding of many hormones, r Why Synthesize Oh'gosaccharides? Apart from the intellectual challenge and the theoretical interest involved, there are a number of practical reasons for tackling the synthesis of glycosides, disaccharides, and oligosaccharides: 1. Preparation of simple substrates for assaying enzymic activities. 2. Preparation of large quantities of materials that are available in only tiny amounts, after laborious processes of separation, from biological sources. Some postulated intermediates in biological processes, e.g., the glycolipid intermediates participating in glycoprotein synthesis--dolichol or polyisoprenol phosphate sugars--have a very short lifetime in vivo, which makes their isolation extremely difficult. However, they are sufficiently stable in vitro to be synthesized, so that their biochemical fate can be examined. 3 B. M. Gesner and V. Ginsburg, Proc. Natl. Acad. Sci. U.S.A. 52, 750 (1964). 4 G. Ashwell and A. G. Morell, Adv. Enzymol. 41, 99 (1974). S. Roth, Quart. Rev. Biol. 48, 541 (1973). W. M. Watkins, in "Glycoproteins" (A. Gottchalk, ed.), 2nd ed., Part B, p. 830. Elsevier, Amsterdam, 1972. r p. Cuatrecasas, Annu. Rev. Biochem. 43, 169 (1974).

[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

95

3. Establishment of unambiguous structures for simpler fragments, without which further advances in our knowledge of the complex parent molecules would be delayed. Sometimes, syntheses have pointed to errors in accepted structures that were leading to false interpretations of the chemistry and biochemistry of complex molecules (e.g., the wrong structure originally postulated for the bacterial cell wall peptidoglycan, which was corrected after synthetic studies). 8 Polymerization versus Stepwise Synthesis. Some homopolymers have been prepared from single monomers, e.g., by acid-catalyzed condensations of anhydro sugar derivatives or even copolymerization of two different sugar anhydrides. 9These reactions may be compared to the formation of polyamino acids by condensation of amino acid anhydrides. However, as has been pointed out, many of the natural saccharides of greatest interest are heteropolymers containing a variety of sugars in a number of different kinds of linkage, so that methods must be developed for the controlled, stepwise condensation of smaller units. Glycosides and Disaccharides. Good models for oligosaccharide synthesis are glycosides of simple alcohols and disaccharides. Many of the problems involved in the preparation of more complex structures, such as position of linkage and stereochemistry, can be analyzed with these compounds. In addition, in the assay of enzymes, glycosides are the usual substrates, especially those bearing chromophoric or fluorescent aglycons, e.g., nitrophenyl or methylumbelliferyl. Some glycoproteins, in fact, bear relatively simple, disaccharide side chains. As will become obvious from this chapter, the chemical synthesis of complex saccharides is not yet a feasible process and our approaches to this goal are necessarily based on the large amount of information available from studies of the simpler models.

Problems Related to Oligosaccharide Synthesis Recent advances in peptide chemistry have generally solved the problems of yield at each stage and of racemization, thus enabling the synthesis of polypeptides and even proteins both in solution and on insoluble supports. However, special problems exist in carbohydrate chemistry that are absent, or much less acute, in protein syntheses. 1. Multiplicity of reactive groups. The condensing molecules each contain a number of hydroxyl groups that differ only slightly in their chemical reactivity. It is necessary to devise methods for the protection of s N. Sharon, T. Osawa, H. M. Flowers, and R. W. Jeanloz,J. Biol. Chem. 241 223 (1966). C. Schuerch,Acc. Chem. Res. 6, 184 (1973).

96

PREPARATIONS

[7]

certain of them and specifically to debiock them after the condensation reaction without affecting the newly formed glycoside. 2. Stereochemical problems. Since the biological properties of glycosides and oligosaccharides, including enzyme specificity, depend on their anomeric configuration (a or fl), it is essential to be able to direct the condensation to produce the required anomer or, at least, to be able to separate, in good yield, the required anomer from a mixture. 3. Problems of yield. Specific additions and removals of protecting groups rarely involve quantitative reactions, although intense research has led to a number of very good reagents, which are often excellent for certain reactions (but less successful in other cases). Generalizations are not always possible, so that each synthesis may become a separate problem in carbohydrate chemistry. 4. Reactivity of the nucleophile. The reaction involves condensation between an activated sugar derivative acting as an electrophile and the free (or occasionally activated) hydroxyl group of another molecule, acting as a nucleophile. Hydroxyl is less nucleophilic than NH2, so that the reaction will be less favored than that between, e.g., COOH and NH2 to give CONH2 in peptide synthesis. In addition, water competes very strongly and must be removed. Primary hydroxyl groups are much better nucleophiles than secondary, both for electronic and steric reasons, but most of the saccharides of interest contain linkages in which secondary hydroxyl groups are involved. General Approaches to Formation of Glycoside and Disaccharide Transformation Reactions. In some special cases, it has been possible to convert more commonly occurring saccharides to desired products by changing structural units: e.g., hydroxymethyl groups have been oxidized to carboxyl to give uronic acid derivatives; hydroxyl has been replaced by amine via ester intermediates; maltose has been oxidized to 2-O-a-Dglucopyranosyl-D-erythrose, which was converted to 3-O-a-D-glucopyranosyl-D-arabinose and the corresponding disaccharide of D-ribose. In the vast majority of cases, however, synthetic processes are necessary, starting from precursors of lower molecular weight. Koenigs-Knorr Reaction and Adaptations. The first approach to disaccharide synthesis TM was based on reaction of substituted sugars (polyacylated derivatives) having a good leaving group at C(1) with suitably protected sugar derivatives bearing a free hydroxyl group, in the presence of some activating reagent ("acid acceptor") (Fig. 1). The X group most used was halide, especially bromide, for reasons of

10W. Koenigs and E. Knorr, Chem. Ber. 34, 957 (1901).

[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

97

CH20R CH20R RO-~X÷ RIOH ~RO-~ORI÷ HX RO OR RO OR FIG. l. Glycoside formation from acyl~ycosyl halides. R'OH may be an alcohol or a protected sugar.

stability coupled with sufficient reactivity, although it could be acetate and, recently, it has been found possible to prepare extremely unstable glucopyranosyl p-toluenesulfonates in situ and to convert them into disaccharides. 11 OR was acetate or benzoate, or later other ester or ether groups. As HX, which is formed, is acidic, reagents were added to neutralize the acid liberated. In early studies, silver salts, such as silver nitrate, oxide, and carbonate were employed. A complication with these reagents is the water formed when they react with HX, and it was often beneficial to include a desiccating agent in the reaction mixture to eliminate the deleterious effect of moisture on the reaction. The introduction of mercuric salts, especially mercuric cyanide, led to a considerable increase in yield. 12 Glycosides can be prepared from liquid alcohols by using them in large excess to serve as solvent as well as nucleophilic reagent. With solid alcohols or protected sugar reagents, suitable solvents must be employed. It has been found that the most useful solvents are those of either fairly low or intermediate polarity, e.g., ether, chlorinated solvents, benzene, dioxane, nitromethane, acetonitrile, or mixtures of nitromethane and benzene. The polarity of the solvent is often fairly critical; some polarity is essential to afford solubility; e.g., most suitable sugar derivatives are insoluble in petroleum ether, but too polar a solvent, such as dimethyl sulfoxide, leads to unproductive decomposition of the sugar halide. The steric course of the reaction is also often influenced to a marked degree by solvent polarity. Both insoluble silver salts and mercuric cyanide in nitromethane (in which it dissolves) were generally found to produce fl-D-gluco- and galactosides and a-o-mannosides from the commonly occuring o~-D-acylglycopyranosyl halides. However, slight changes in conditions often resulted in unexpected results, thus the use of equimolecular amounts of mercuric cyanide and mercuric bromide in acetonitrile sometimes gave the other anomer, while a-o-gluco- and galacto- and aL-fucosides resulted in some reactions where mercuric cyanide in nitromethane was employed. Mechanistic Considerations. Some analysis of possible mechanisms is necessary in order to be able to understand and, hopefully, to predict the 11 R. Eby and C. Schuerch, Carbohydr. Res. 50, 203 (1976). '~ B. Heiferich and K. Weis, Chem. Bet. 89, 314 (1956).

98

PREPARATIONS

[7]

R I ..O~H (I)

SN2~

--OJ

(slow)

0 OR HBr

3\ ,.o-,.3

30: :yo,

Br

FIG. 2. Mechanism of displacement of glycosyl halides by nucleophile. S~2 processes-bimolecular--are a function of the concentration of halide and nucleophile; in S~I processe s, the reaction rate is independent of nucleophile concentration, and racemization is implied.

steric course of the reaction. Detailed discussion of this question has appeared in a number of reviews 13-15 and will only be dealt with briefly here. It should be pointed out that most investigations were based on reactions of substituted glycopyranosyl halides with large excess of nucleophiles serving also as solvent (solvolytic reactions) so that analogies, by implication, to the usual Koenigs-Knorr conditions of disaccharide synthesis, where approximately equimolar ratios of glycosyl halide and nucleophile are employed, may well be unjustified. Bearing this qualification in mind, certain general conclusions obtained from solvolytic reactions have been applied to disaccharide synthesis with a fair amount of success in predicting the products (Fig. 2). In some cases, notably in the presence of insoluble silver salts in solvents of low polarity, S~2-type processes predominate (1), leading directly to inversion of the configuration of the sugar halide so that a fl-linked disaccharide results from an a-glycopyranosyl halide. Usually, however, it appears that an S~ 1-type process is involved (2). Two corollaries of these mechanistic differences are reflected in the kinetics of the reaction and its steric course: KINETICS, In the SA.2 process (1), the slow, rate-determining step is the one in which halide and nucleophile interact, and the reaction rate will thus be a function of the concentrations of both these species; i.e., it will be influenced by the concentration of the nucleophile. In the SA 1 process (2), the rate-determining step is the one involving the ionization of the halide so that the overall reaction rate is independent of nucleophile concentration. ~3 R. J. Ferrier, Fortschr. Chem. Forsch. 14, 389 (1970). J4 G. Wulff and G. R6hle, Angew. Chem. Int. Ed. Engl. 13, 157 (1974). ~ N. K. Kochetkov and A. F. Bochkov, Methods Carbohydr. Chem. 6, 480 (1972).

[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

99

STEREOCHEMISTRY. The SA,2reaction involves a Walden inversion and. leads to conversion of the halide into a glycoside or disaccharide of opposite anomeric configuration. SA-I-processes, on the other hand, should give racemization, i.e., lead to the production of equal amounts of a and fl anomers in the product, irrespective of the configuration of the starting halide. In fact, while a-D-glycopyranosyl halides of D-glucose and o-galactose often give fl-D-glycosides with high stereospecificity, the corresponding o-mannose halide may give a variety of products but, under suitable conditions, the a-D-glycoside predominates. In order to explain these results and other findings from solvolysis reactions, the following mechanisms were postulated (Fig. 3). In the first case [(1), e.g., D-gluco], the intermediate resonancestabilized carbonium ion is further stabilized by participation of the C(2)acyloxy group to form a cyclic intermediate which is attacked by the nucleophile from the opposite side of the ring, resulting in a fl-D(i.e., trans-2-acyloxy)-glycoside. In the second case [(2), e.g., o-manno], the stereochemistry of the starting halide is such that its ionization may be facilitated by participation of the C(2)-acyloxy group, in preference to the ring oxygen. The resulting reactive cyclic intermediate can then be attacked (at the a side) by the nucleophile. Again, it will be noted, a trans2-acyloxy-glycoside results. Solvents of low polarity, which minimize the ionization of the bromide, may well tend to direct the stereochemistry toward a result to be expected from an S~.2-type reaction; the proximity of R I

(I) O"'C OI = CH3

~ ~0 CH3

~H3

CIH3 /----o c\~

(2) ~/~a, °

0COCH3

0~.C"~ CH3

c

~.

.__~b--°~'~"x°

i.

(/CH=C__R ,--O,'''~~vo

R H

F[(;. 3. Participation of 2-acyloxy group in formation and reaction of 2-acylglycosyl halides. In (1), the oxo-carbonium ion is stabilized by participation, whereas in (2) ionization of the halide is facilitated by participation of this group.

100

[7]

PREPARATIONS

OR ROH 4

OCORI

~/~

ROH 0

~N/IfC\o~

Ib

~00.~ RI \OR

FIG. 4. Reactions of orthoesters to give glycosides or orthoester exchange.

the bromide will interfere with the approach of the nucleophile from the same side of the molecule, so that gluco- and galacto-halides would give fl-D-glycosides, with a similar possibility from manno-halides if participation does not occur (see Fig. 5). Since acetamido groups participate even better than acyloxy groups, acetamido sugar halides may be expected to show an even greater steric control and tendency to produce trans-2-acetamido-glycosides than do the corresponding acyloxy sugars.

Orthoester Method A consideration of the preceding paragraph indicates that prior formation of a stable cyclic 1,2-electrophile to replace the acylglycopyranosyl bromide used in the reaction should increase the possibility of obtaining a good yield of trans-2-acyloxy-glycoside. Orthoesters are 1,2-cyclic compounds of this type and can be conveniently prepared from glycosyl halides by a number of different methods. In the presence of catalytic amounts of Lewis acids (including mercuric bromide), orthoesters react with alcohols. The product is a glycoside if the ratio of acid catalyst to orthoester is strictly controlled, although orthoester interchange may occur instead to give a new orthoester containing the reacting alcohol (Fig. 4). The method was considered stereospecific, yielding trans-2-acyloxy glycosides exclusively. It is possible to adapt it for the synthesis of polymers, and a tricyclic orthoester derivative of L-arabinofuranose condenses with itself to give a linear polymer of several thousand molecular weight. 1.~ However, it has recently been demonstrated that strict steric control is not always maintained, and the anomeric configuration of the product may depend to a considerable extent on the solvent employed in the reaction. 1" ~-Linked Disaccharides. We have seen that there is a tendency toward formation of 2-acyloxy-trans-glycosides from either glycosyl halides or 1,2-cyclic intermediates. In many cases, this leads to isolation of/3-1inked disaccharides. However, there are a large number of examples of a-linked sugars in materials of biological interest. These include antigenic determinants of bacteria and many plant polysaccharides; galacto- and "~P. J. Garegg and I. Kvanstr6m, A c t a Chem. Scand. B30, 655 (1976).

[7]

C H E M I C A L SYNTHESIS OF OLIGOSACCHARIDES

101

N-acetylgalacto-a-o-pyranosyl units are B and A blood group determinants, respectively, and L-fucose residues are always found a-linked whenever they occur in glycolipids and glycoproteins. D-Mannose occurs frequently in both configurations in the same molecule; in this case, it is the E-configuration that is the c/s-1,2-arrangement and it forms an important branching point in the structures of many glycoproteins. The mechanistic analysis reviewed previously showed the possible effect of "participating" groups vicinal to the glycosidic center on the steric course of the faction. An obvious approach to a-glycosides would then be to eliminate this effect, thus affording the possibility of ensuring a l : 1 a: fl ratio at least. In general, c/s-2-acyloxyglycosyl halides are more stable than the trans anomers. In cases where the trans anomer can be prepared, it would seem that if an SA,2-type reaction could be ensured, formation of the required product would follow. However, since the preparation and use of such halides is often not feasible and SN 1-processes are usually involved in glycoside formation, it is necessary to guarantee additional factors if any approximation to stereospecifcity is to be attained. The extreme interest manifested in this problem is shown by the variety of reagents applied in attempts to prepare such a-linked compounds. Unfortunately, no universal panacea has emerged, so that each synthesis is almost a separate research topic in itself. In most cases, however, the C(2)-acyloxy group usually appearing in synthesis of trans1,2-1inked disaccharides has been replaced by a substituent with a different electronic arrangement which does not favor participation with the C(1)-carbonium ion. Other factors can then intervene to determine the steric outcome of the reaction. In some examples, the C(2) substituent is potentially convertible to either NH2 or OH, thus enabling preparation of disacchrides of 2-amino sugars as well as the usual C(2)-hydroxylated compounds. A number of three examples will now be discussed in more detail. Inversion ofTrans-Glycosyl Halides. The stability of glycosyl halides is affected greatly by the nature of the substituent at C(2). A number of crystalline trans-halides have been prepared, stabilized by strongly electronattracting C(2) substituents, such as nitrate and trichloroacetate. Unfortunately, stabilization of the halide also lends to deactivation, so that, although successful reactions of these compounds occurs with primary alcohols, yields with secondary alcohols, particularly in disaccharide synthesis, are much lower. In this way, an a(1-->6)-disaccharide and trisaccharide of D-glucose (isomaltose and panose, respectively) were synthesized 1r'18 by way of a 2-nitrate ester, disaccharides of glucose and galacx~ M. L. Wolfrom, A. O. Pittet, and I. C. Gillam, Proc. Natl. Acad. Sci. U.S.A., 47, 700 (1961). ~ M. L. Wolfrom and K. Koizumi, J. Org. Chem. 32, 656 (1967).

102

PREPARATIONS

[7]

tose were synthesized19 from 2-O-trichloroacetates, a-D-xylosides "° and a-L-fucosides 21 from O-chlorosulfates, and isomaltose and panose were even prepared successfully from a halide with an unprotected C(2)hydroxyl group. 22 In most of these reactions, insoluble silver salts were present in excess, so that the steroselectivity obtained may have resulted from involvement of an Ss2-type process. However, since no mechanistic studies were pursued, this point was not clarified. Other examples of reactions with halides bearing nonparticipating C(2) substituents, such as benzyl ethers 23 or a variety of amino-protecting groups (e.g., Lloyd et a1.,24), gave a :/3 mixtures, thus being more indicative of the commoner Sx2-type processes. Anomerization o f the Halide in Situ. Consideration of the mechanism of reaction of glycosyl halides reveals that even Sxl-processes can be utilized under suitable conditions to produce c/s-l,2-glycosides, irrespective of the starting configuration of the halide. It is thus possible to use the common c/s-halides for the preparation of c/s-glycosides. Since a free carbonium ion is an extremely reactive species, it is considered that it may be immediately stabilized by a number of alternative routes (Fig. 5, a-f). The first stage of the Sxl-process (a) may formally be regarded as incomplete separation of the counter-ion (halide) from the sugar, to produce an ion pair, followed by complete ionization (b); the ion-pair intermediate structure will be favored in solvents of low polarity. Reaction of such an ion pair with an alcohol (g) should yield a glycoside of opposite configuration (i.e., trans- 1,1). However, the presence of excess halide ions (either during the later stages of the synthesis with soluble mercuric salts or through the deliberate addition of well ionizing halides), may convert the " a " ion pair to a "/3" ion pair of higher reactivity, which will then react with alcohols to produce a-glycosides (f). In more polar solvents, complete separation of the halide from the sugar will occur; the resulting carbonium ion will react with nucleophile to give both c/s- and transglycosides (c). In strongly nucleophilic solvents (ether, nitromethane, etc.) the carbonium ion will be solvated (e) and the fate of the resulting solvate will determine the sterochemistry of the final product. On the other hand, participation of the C(2)-substituent (pathway d) will lead to trans- 1,2-glycosides. ,9 B. Helferich, W. M. MiJller, and S. K a r b o c h , Justus Liebigs Ann. Chem. 1514 (1974). 2o H. J. Jennings, Can. J. Chem. 46, 2791 (1968). 21 H.-R. P o u g n y and P. S i n a i , Carbohydr. Res. 34, 351 (1974). 22 B. Helferich and W. M. Miiller, Chem. Ber. 106, 2508 (1973). 23 p. W. Austin, F. E. Hardy, J. G. B u c h a n a n , and J. Baddiley, J. Chem. Soc. p. 2128 (1964); p. 1419 (1965). .24 p. F. Lloyd, B. Evans, and R. J. Fielder, Carbohydr. Res. 22, 111 (1972).

[7]

103

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES - - 0 OR

--o.OR

-4

S

-- 0

+

OCOR'

¢ T(ROH)

cT OR

--0 •

~

w.

•

(ROH) OCORI

OCOR'

--

0 COR '

:

q.

OCORI

--0

OCORf

--0

40.

OCOR'

OCOR'

OR'

--0

OCOR'

OR

-4

OCOR'

Fro. 5. Possl'ble reactions of C(1)-carbonium ion, to produce: (a), (f), ion pairs; (c) glycosides with racemization; (d) glycosides with inversion; in (e), solvent (S) is involved.

Pathway f suggests a simple approach to cis-glycosides which has afforded considerable success in recent years. In solvents of rather low polarity (e.g., methylene chloride), c/s-l,2-disaccharides were obtained from perbenzyl c/s-glycosyl halides of D-glucose. 2~ D-galactose, 2~''6 and L-fucose 2~'2r in the presence of tetraethylammonium bromide and either ethyldiisopropylamine or molecular sieves. Perbenzylated glycosyl halides in the presence of silver perchlorate, also afforded a number of u-linked disaccharides, "~presumably via a reactive fl-glycosyl perchlorate intermediate. Effect of Substituents in the Glycosyl Halide. The steric control exerted by C(2) substituents has already been discussed. An interesting observation that was made was that ester groups at C(3)-C(6) also play some role in direction toward c/s-l,2-disaccharide formation, but in the opposite direction to C(2) substituents. Thus, the presence of acyl groups (acetyl or p-nitrobenzoyl) in 2-O-benzyl glycosyl bromides of D-glucose "9':~° and L-fucose:~' led to formation of a-disaccharides. Apparently, the C(4) and '-'~R. U. Lemieux, B. Hendricks, R. V. Stick, and K. James, J. Am. Chem. Soc. 97, 4056 (1975). "-'"P. A. Gent and R. Gigg, J. Chem. Soc. Perkins Trans. 1, 361 (1975). ~7J.-C. Jacquinet and P. Sinai', Carbohydr. Res. 42, 251 (1975). '-'~K. Igarishi, J. Irisawa, and T. Honma, Carbohydr. Res. 39, 213; 341 (1975), ~" H. M. Flowers, Carbohydr. Res. 18, 211 (1971). 30 j. M. Berry and G. G. S. Dutton, Can. J. Chem. 52, 681 (1974). :" M. Dejter-Juszynski and H. M. Flowers, Carbohydr. Res. 23, 41 (1972); 41,308 (1975).

104

PREPARATIONS

[7]

C(6) substituents play a special role in the steric direction of the reaction, as was shown by the low stereoselectivity exhibited in their absence, :~'-'-'~ and the presence o f a nucleophilic solvent (pyridine) was found efficacious in directing toward a-disaccharide formation. '-'4':~z For example, it was found 24 that the halide (I) gave mainly/3-glycosides with silver carbonate in nitromethane when R was acetoxy, but a-glycosides in pyridineCH2 OR

Ac 0"-'-~ Br

NO 2

(I)

chloroform. However, replacement of the 6-acetoxy group by methoxy resulted in fl-glycosides. It was argued that the 6-acetoxy group assisted in displacement of the halide, and the resulting oxonium ion was stabilized by pyridine, thus controlling the stereochemistry. In this case, since the cyclic intermediate is formed at the upper side (/3) of the sugar ring, the alcohol group enters at the lower side (a). This sort of reasoning is similar to that put forward by other workers 3''3a to explain the cis-directing effects of C(4)- and C(6)-acyl substituents. CH2OAc

Brigl's Anhydride. A cyclic derivative that reacts with nucleophiles, albeit sluggishly, to produce disaccharides is "Brigl's anhydride" [tri-Oacetyl- 1,2-epoxy-a-D-glucopyranose (II)]. This 1,2-epoxide was used in the first synthesis of sucrose 36 and in a recent synthesis of O-ot-D-glucopyranosyl-(1--~4)-2-acetamido-2-deoxy-oglucose37 (part of the repeating unit ofheparin). Inspection of the structure (II) would indicate that /3-D-glucosides should be expected from such as M. Dejter-Juszynski and H. M. Flowers, Carbohydr. Res. 28, 61 (1973). 33 j. M. Frechet and C. Schuerch, J. Am. Chem. Soc. 94, 604 (1972). 34 K. Eklind, P. J. Garegg, and B. Gotthammer, Acta Chem. Scand. B30, 300 (1976). .~5M. Petitou and P. Sinai, Carbohydr. Res. 40, 13 (1975). '~ R. U. Lemieux and G. Huber, J. Am. Chem. Soc. 78, 4117 (1956). •~7 p. C. Wyss, J. Kiss, and W. Arnold, Heir. Chim. Acta 58, 1847 (1975).

[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

105

reactions. However, only the a-anomers are isolated, and it was suggested that the steric course of the reaction could be explained a6 as being due to participation of the 6-acetoxy group in (II). More recent proposals have been based on a prior opening of the 1,2-epoxide under the extreme reaction conditions (high temperatures for prolonged times) leading to a glycosyl carbonium ion lacking a C(2) participating group and thus enabling cis- 1,2-disaccharide formation. Cis-l,2-Glycosides from Peracylglycosyl Halides. The complexity of factors involved in the steric control of the Koenigs-Knorr reaction is illustrated in the products of a number of reactions of peracylglycopyranosyl bromides. Thus, there are examples of derivatives of D-glucose, z8 o-galactose, 3a'4° L-fucose, aL42 and D-mannose 4z giving either cis-l,2disaccharides almost exclusively or cis,trans mixtures. In the majority of these cases, (1-~2)-disaccharides were synthesized and the stereochemistry might have been influenced by the proximity in the sugar nucleophile of the bulky glycopyranosyl group substituted at C(2) by the reaction to its own anomeric group at C(1). However, other examples (e.g., see references cited in footnotes 40 and 43) do not involve this effect. In some of the above reactions, mercuric bromide was used in addition to mercuric cyanide, and acetonitrile was the solvent, so that ion-pair and solvent effects may also have been sufficiently strong to overcome the trans-([3)-directing effect to be expected from participation of the C(2)-acyloxy substituent. However, it was shown that many years ago 44 that the use of mercuric acetate with tetra-O-acetylglucopyranosyl bromide in the Koenigs-Knorr reaction could lead to either o~- or /3-glycosides, depending on the alcohol and the reaction conditions employed. Amino Sugar Derivatives. Some special problems arise in the synthesis of glycosyl derivatives of 2-amino sugars owing to the greater nucleophilicity of nitrogen over oxygen. This is also reflected in the instability of such sugars, which, for storage, must be kept either, in a form in which there is no free amino group (e.g., acetamido) or in which its nucleophilicity has been reduced by protonation (hydrochloride). Some crystalline glycopyranosyl chlorides of 2-acetamido sugars have been prepared, but the bromides cannot be isolated owing to their extreme ease of conversion into other products by way of cyclic intermediates. The relatively low 38 B. Helferich and J. Zirner, Chem. Ber. 95, 2604 (1962). 39 j. Lehmann and D. Beck, Justus Liebigs Ann. Chem. 630, 56 (1960). ao M. E. Chacofi-Fuertes and M. Martfn-Lomas, Carbohydr. Res. 43, 51 (1975). 41 H. M. Flowers, A. Levy, and N. Sharon, Carbohydr. Res. 4, 189 (1967). 42 A. Levy, H. M. Flowers, and N. Sharon, Carbohydr. Res. 4, 305 (1967). 43 M. Shaban and R. W. Jeanloz, Carbohydr. Res. 17, 193 (1971). 44 G. Zemplen and A. Gerecs, Chem. Ber. 63, 2720 (1930).

106

PREPARATIONS --0

[7] - - 0 OR'

__~0 N...~C

+ R'OH

"

__~ NHCOR

I

R F1G. 6. Reactions of oxazolines with nucleophiles to give trans-2-glycosides of amino sugars.

reactivity of the chlorides toward alcohols and the extreme instability of the bromides lead to low yields in glycoside-synthesis from these reagents and the extreme difficulty in obtaining any reasonable amounts of disaccharides from them. A large variety of other N-protecting groups have been utilized and have led to either cis- or trans-2-amidoglycosides, depending presumably upon their tendency to participation. These have included N-benzoyl,45 N_2,4_dinitrophenyl,46 N-dichloroacetyl,4r N-trifluoroacetyl,48 N-diphenoxyphosphoryl,49 N_phthalimid050 and even the 2-amino bromide hydrobromide, 51 among others. In some of these examples, the reaction was facilitated by activation of the nucleophilic reagent (i.e., the protected sugar to be condensed with the amino sugar halide), and such reactions will be discussed later. A considerable advance in the chemistry of synthetic trans-2-aminodisaccharides came with the preparation of cyclic 1,2-derivatives of amino sugars. 52 In analogy with the orthoesters described previously, such cyclic compounds should also exert a strong steric control over condensation reactions at C(1). Furthermore, the increased nucleophilicity of N over O should facilitate their formation and increase their stability. In fact, these compounds (oxazolines) have been utilized increasingly during the last few years for disaccharide synthesis 53'~4(Fig. 6), and the corresponding oxazoline of a disaccharide has also been described. 53 A different approach that has recently offered considerable success has been to prepare a reactive intermediate which, while not itself a 2-amino sugar, could be readily converted into one either directly or after glycosylation. The first class of compounds of this type to be used were 2-nitroso 45 F. Micheel and H. K~chling, Chem. Ber. 93, 2377 (1960). 46 K. Heyns, K. Propp, R. Harrison, and H. Paulsen, Chem. Ber. 100, 2655 (1967). 47 D. Shapiro, Y. Rabinsohn, and A. Diver-Haber, Biochem. Biophys. Res. Commun. 37, 28 (1969). 4s W. Meyer zu Reckendorf and N. Wassiliandou-Michaeli, Chem. Bet. 103, 1792 (1970). 49 C. Merser and P. Sina% Tetrahedron Lett. p. 1029 (1973). ~0T. Osawa, Chem. Pharm. Bull. (Tokyo) 8, 597 (1960). ~' J. C. Irvine, D. McNicoll, and A. Hynd, J. Chem. Soc. 99, 250 (1911). n2 F. Micheel and H. KSchling, Chem. Ber. 90, 1597 0957). ~'~S. E. Zurabyan, T. S. Antonenko, and A. Y. Khorlin, Carbohydr. Res. 15, 21 (1970). 54 R. Kaifu, T. Osawa, and R. W. Jeanloz, Carbohydr. Res. 40, 111 (1975).

[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

107

sugars. Addition of nitrosyl chloride to glycals produces crystalline dimers of 2-nitrosoglycosyl chlorides. These derivatives react quite well with a variety of protected sugars 55and even give fairly reasonable yields of disaccharides from secondary hydroxyl groups. Reduction of the addition-product, an oxime, produces the desired amino sugar disaccharide which is mainly cis-linked. However, in contrast to the earlier findings of apparently almost complete stereospecificity of addition, in some later syntheses considerable trans product resulted. 5~'57 A further advantage of this approach is that the intermediate oxime can alternatively be converted to a keto sugar by deoximination and the C==O group reduced to CH(OH), thus affording a good route to some nonnitrogenous cis-l,2-disaccharides, s~ It should be noted that the intermediate oxime has lost its tetrahedral arrangement at C(2) so that on reduction (or reduction of the ketone), two epimers can result. The control of stereochemistry here depends to a considerable extent on the reagent employed, but it is not specific and in some cases almost equal amounts of the two epimers are obtained. More recently, 2-azido sugars have also served as good intermediates in the synthesis of c/s-2-amino disaccharides. ~9'~° In this case, the cis-2azido bromide reacts with alcohols to produce trans-2-azido glycosides, and an unstable trans-2-azido chloride can be prepared that reacts with formation of c/s-2-azido disaccharides in good yield (even at secondary hydroxyl groups) under extremely mild conditions. Silver perchlorate together with silver carbonate or collidine were used in these reactions, and it was believed that SA2-type processes were involved, with completely stereospecific inversion of configuration of the reacting trans-2-azido chloride. However, owing to the extreme reactivity of this material, partial anomerization to the more stable c/s-chloride occurred during the preparation and condensation, leading to contamination of the disaccharide with varying, usually small, proportions of the trans anomer. Disaccharides from diamino sugars and trisaccharides have also been synthesized by this method. Reduction of the 2-azido group to amine is quite simple, and, in contrast to the previous method, only a single epimer results. On the other hand, the azido group cannot be easily replaced by OH, so that the method is rather more limited than that starting from 2-nitroso sugars. Furthermore, the intermediate azides are prepared by an ~ R. U. Lemieux, K. James, and T. L. Nagabhushan, Can. J. Chem. 51, 48 (1973). ~6 K. Miyai and R. W. Jeanloz, Carbohydr. Res. 21, 45 0972). ~7 R. U. Lemieux, Y. Ito, K. James, and T. L. Nagabhushan, Can. J. Chem. 51, 7 0973). ~8 R. U. Lemieux, K. James, and T. L. Nagabhushan, Can. J. Chem. 51, 42 0973). ~ H. Paulsen and W. Stenzel, Angew. Chem. Int. Ed. Engl. 14, 558 (1975). 60 H. Paulsen, O. Lockhoff, B. Schr6der, B. Sumfleth, and W. Stenzel, Tetrahedron Lett. 2301 (1976).

108

PREPARATIONS

[7]

involved process starting from 1,6-anhydro sugars, and so are not readily available. This approach should be of great promise in the case of rare amino sugars as it includes the synthesis of the amino sugar from a more common precursor. The sialic acids are a rather special case of amino sugars. The location of the amino group makes them different in their properties from 2-amino sugars. Their extreme lability, owing to the deoxy arrangement vicinal to the anomeric center and the fact that they are keto sugars rather than aldoses of the type we have discussed so far, pose acute problems which have hindered progress in synthesis of glycosides and disaccharides of this type, but some success has been attained using acetamido halides. ~'"2 Reactivity of Hydroxyl Groups and Activated Nucleophiles. There are some differences in the reactivities of different hydroxyl groups in sugars, depending on steric and electronic effects, so that it is possible in favorable cases to prepare disaccharides without having to protect all those groups that are not intended to react; e.g., primary hydroxyl groups are more reactive than secondary,"3 and there is some gradation of reactivity of the secondary hydroxyl groups depending on their location. In hexopyranosides, the C(4)-hydroxyl group seems to be especially deactivated and difficult to condense. "4 Since a considerable number of natural saccharides, including the common homopolymers cellulose, starch, and chitin, contain the 1---~4 linkage, there is great interest in activating this position. There have been two major approaches to this goal: (1) to prepare intermediates with a different conformation in which the C(4)hydroxyl might be expected to be more reactive; (2) to prepare activated derivatives of this hydroxyl group, e.g., activated ethers, which react with liberation of the substituent and formation of the disaccharide linkage. Conformational Changes. It is believed that the proximity of the bulky C(5) substituent plays a role in the deactivation of the C(4)-hydroxyl group in hexopyranosides. Two alternative methods have been devised to remove this effect. The first is to prepare open-chain derivatives to serve as nucleophiles in the Koenigs-Knorr reaction, and the second depends on the altered reactivity of hydroxyl groups in the fused ring system of 1,6-anhydro sugars. Good examples of the use of open-chain sugars are (a) the synthesis of lactose ~5 from (III) and (b) the synthesis of di-N-acetylchitobiose and its a-analog 4~ from (IV). 61 p. Meindl and H. Tuppy, Monatsh. Chem. 96,802 (1965). ,2 A. Y. Khorlin, I. M. Privalova, and I. B. Byotrova, Carbohydr. Res. 19, 272 (1971). ,:1R. R. King and C. T. Bishop, Can. J. Chem. 53, 1970 (1975). ~4 D. Beith-Halahmi, H. M. Flowers, and D. Shapiro, Carbohydr. Res. 5, 25 (1967). ,i.~ E. J. C. Curtis and J. K. N. Jones, Can. J. Chem. 37, 358 (1959).

[7]

109

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES CH(OMe)2 0

CH (OEt) 2

CMe20~

0~

C• 0

o\

CH201CMe2

CH20

(m)

-

(i~)

The use of 1,6-anhydro sugars has enabled the synthesis of a number of disaccharides of biological interest, including lactose 47 and the "TaySachs" trisaccharide ~ (V) from 1,6-anhydro-fl-o-glucose, and a noteworthy preparation 4'~of the repeating unit of the bacterial cell wall peptidoglycan (VI) starting from 1,6-anhydro-/3-D-glucosamine.

CH20H CH~OH CH.jOH

CH20H

C.H20H

OH

OH H

NHAc

OH

OH

NHAc

I

NHA¢

c.3c.coz.

(£)

Activated Nucleophiles. The possible activation of hydroxyl groups by tritylation has been known for some time and used successfully in a number of disaccharide syntheses.6r In the synthesis of panose, a trisaccharide, it was demonstrated that although condensation occurred directly with the free, C(6) hydroxyl group, the trityl ether reacted better. TM These condensation were all adaptations of the Koenigs-Knorr procedure using a protected glycosyl halide and silver perchlorate as catalyst. A variation in this procedure "s was to employ the corresponding/3-acetate with ailyl bromide and silver perchlorate (Fig. 7). CHzOAc

C.20rr

oo)-oo, o ,Co IQpAc-/~ +A¢O~'-~ 't'--l" i(OA¢ ~OBn OAc

NHAc

AgC,04

•

C"20H

C" 2

HO.,I-o o

J-o

~

HO~/~O H OH

NHA¢

FIG. 7. Activation of C(6) hydroxyl by tritylation. Trityl perchlorate is readily formed in situ, enabling facile reaction with the C(l)-O-acetate. Note that reaction may also be facilitated by participation of the C(2)-O-Ac in formation of the glycosyl carbonium ion.

"" D. Shapiro, A. J. Acher, Y. Rabinsohn, and A. Diver-Haber, J. Org. Chem. 36, 832 (1971). ~r H. Bredereck, A. Wagner, D. Geissel, P. Gross, U. Hutten, and H. Ott, Chem. Bet. 95, 3056 (1962). "sV. A. Nesmeyanov, S. E. Zurabyan, and A. Y. Khorlin, Tetrahedron Lett. 3213 (1973).

110

PREPARATIONS

[7]

It was suggested that an allyl cation was formed; this attacked the glycosyl acetate to produce a glycosyl carbonium ion, which reacted with the nucleophilic oxygen of the trityl ether. A limitation in this method is the difficulty in preparation o f trityl ethers from many secondary hydroxyl groups. Recently, however, other ethers have been found to extend the range of this approach/;:' T._, Although the m e t h o d is generally more complicated, as the etherifying reagents may not always be so simple to handle and additional steps are often involved, the increased reactivity attained may enable the preparation o f hitherto inaccessible oligosaccharides and lead to its popularization. Oligosaccharide Synthesis The problems of yield, stereochemistry, and protection arising in disaccharide synthesis become c o m p o u n d e d as the degree o f polymerization (D.P.) o f the saccharide rises. Stepwise processes of condensation involve reaction of such activated derivatives as C(1)-glycosyl halides with nucleophiles of increased D.P. or, alternatively, glycosyl halides o f increasing D.P. with nucleophiles of low molecular weight. In the former case, it will usually be necessary to employ a number o f different protecting groups, since each condensation step will be preceded by a selection deblocking reaction to free only the hydroxyl group required for further reaction. Use o f glycosyl halides with the desired hydroxyl group(s) free offers a possibility 7'~'74which is limited owing to complications of side reactions, but there have been some recent notable successes with suitably protected derivatives, including the synthesis o f pentasaccharides ~ of D-glucose and some blood group active trisaccharides. 7~ The preparation of oligosaccharide sugar C(I) halides on the other hand, becomes increasingly more difficult as D.P. increases, owing to a lower rate of conversion of the intermediate C(I) esters to halide and lability of the interglycosidic bond to the reaction conditions employed in this conversion. Furthermore, isolation of the resulting halides from unreacted esters and degradation products is a more complex problem with higher-molecular-weight materials. The condensation reaction at each stage is, in principle, the same as that employed in disaccharide synthesis, so that the same rules of control of stereochemistry and relative reactivity of hydroxyl groups apply. But ~:' A. Klemer, K. Gaupp, and E. Buhe, Tetrahedron Lett. 4585 (1969). 7. T. Ogawa and M. Matsui, Carbohydr. Res. 51, C13 (1976). 7~N. K. Kochethov, V. A. Derevitskaya, and E. M. Klimov, Tetrahedron Lett. 4769(1969). re A. Y. Khorlin, V. A. Nesmeyanov, and S. E. Zurabyan, Carbohydr. Res.. 43, 69 (1975). 73S. Haq and W. J. Whelan, J. Chem. Soc., p. 4513 (1956). 74H. M. Flowers, Carbohydr. Res. 2, 188 (1966). r.',R. U. Lemieux, and H. Driguez, J. Am. Chem. Soc. 97, 4063; 4069 (1975).

[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

111

the rapid fall in reactivity to be expected as D.P. rises poses a problem that may require novel approaches to overcome. The highly reactive p-toluenesulfonates recently described" offer some promise for cis-linked oligosaccharides, but, so far, only such derivatives of D-glucose are known, and special precautions are necessary in their preparation in situ and their reaction, reducing the possibility of their general application. There have been a number of polymerizations of reactive intermediates (e.g., 1,6-anhydro sugars, and orthoesters) to give homopolymers or mixed copolymers, 9 but, as I pointed out earlier, this approach is also limited. A highly successful development in synthetic polypeptide chemistry in recent years has been the introduction of insoluble polymer supports. This method implies a number of possible advantages over the usual reactions in solution: (1) it is technically easier to isolate insoluble products from excess of soluble reagents and decomposition products; (2) it may be possible to attain high local concentrations of reactive species increasing the rate of reaction; (3) in principle, it may be possible to prepare a polymer that directs the orientation of the reagent and thus strictly controls the stereospecificity of the reaction. In this way it would be possible to prepare o~-, or/3-, glycosides at will from the same reagents using the requisite supports. This type of steric control would be analogous to biosynthetic processes, especially those occurring in (on) insoluble membranes. In fact, however, this possibility has not been attained, and investigations have been performed on soluble systems in a number of laboratories interested in developing the method of insoluble supports in order to prepare good reagents for controlling the stereochemistry of the reaction. In other words, soluble systems are taken as the models for the "insoluble" syntheses and no steric advantage accrues from the use of the insoluble supports. One of the earliest studies in this field was that employing a polystyrene support. TM Reaction with free hydroxyl groups on the support was attained by employing a 6-acyl-perbenzylated glycopyranosyi halide. Deacylation then permitted disaccharide formation between the deblocked 6-hydroxyl function and another molecule of protected sugar halide. The nonparticipating benzyl groups directed the stereochemistry so that mainly c/s-l,2-disaccharide resulted. However, separation of the product from the resin in the form of a free disaccharide was not readily achieved. Another laboratory prepared styrene-divinylbenzene copolymers 77 76j. M. Frechet and C. Schuerch, J. Am. Chem. Soc. 93, 492 (1971). 77 R. D. Guthrie, A. D. Jenkins, and G. A. F. Roberts, J. Chem. Soc. Perkin Trans. 1, 2414 (1973).

112

PREPARATIONS

[7]

with a number of different glucosyl substituents. These copolymers, in distinction to the polymers described previously, were soluble. The authors reasoned that such soluble polymers would increase the possibility of achieving a high yield at each step and facilitate following the reaction course. However, the ease of separating insoluble intermediates from complex mixtures was thereby lost and any possibility of ultimately developing an automated process was nullified. In the previous method a polymer-bound nucleophile reacts with an unbound glycosyl halide, whereas, in this second approach, the sugar is converted to a polymerbound halide. This halide is reacted further to produce a 1,2-orthoester which gives trans-l,2-disaccharides with protected sugars acting as nucleophiles. In this way, a gentiobiose derivative was synthesized and readily removed from the polymer in reasonable yield. An amino sugar-disaccharide was synthesized TM from a 4,6-0benzylidene acetal attached to an insoluble styrene-divinylbenzene resin. The product was debenzylidenated, and, of the resulting two free OH groups, that at C(6)--primary--reacted selectively with 2-acetamido-triO-acetyl-2-deoxy-a-D-glucopyranosyl chloride to give a derivative of fl-l,6-di-N-acetyl-o-glucosamine, which was readily removed from the resin and could be easily de-O-acetylated and converted to the free disaccharide (Fig. 8). Some trisaccharide, presumably following reaction at both C(4) and C(6), also resulted. Close similarity was demonstrated between the steric course of reactions on insoluble supports and those in solution, both in the synthesis of amino sugar disaccharides TM and oligosaccharides of D-glucose TM using a variety of reagents. o CHz

Ph&)--o Xo( NHAc B

CH20H ~' H~O~OBn NHAc

CH20H ~

CH2

HO~--t HO'---f NHAc NHAc

( R = polymer - CO)

NHAc FIG. 8. Disaccharide preparation on an insoluble polymer support. The nucleophile is bound to the resin by ester linkages. Ts G. Excoffier, D. Gagnaire, J.-P. Utille, and M. Vignon, Tetrahedron 31, 549 (1975). z9 G. Excoffier, D. Gagnaire, and M. R. Vignon, Carbohydr. Res. 46, 201,215 (1976).

[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

113

A rather different series of reactions involving thio sugars was developed in solution for disaccharide synthesis and adapted to both soluble s° and insoluble polymer supports, s' In this case, derivatives of 1-thio-D-glucose were bound through the S atom to the polymers. Selective deblocking freed C(4)- or C(6)-OH, which was allowed to react with 6-O-acetyltri-O-benzyl-a-D-glucopyranosyl bromide. The resulting disaccharide (almost completely a--maltose or isomaltose derivatives-with very little of the corresponding fl-linked disaccharides) was removed from the polymer in good yield as benzyl glycoside by cleavage with methyl iodide and benzyl alcohol in refluxing benzene without splitting the disaccharide linkage. Saccharide Peptides. Portions of the saccharide-amino acid linkage region found in many glycoproteins and analogous, model, structures, consisting of di- and trisaccharides of mannose and N-acetylglucosamine coupled through N-acetylglucosamine to asparagine, have been synthesized in recent years. 8z A new type of reagent has been described (2 amino-2-methoxyethylthioglycosides) for attachment of sugars to proteins. 83 The derivatives thus obtained are not found naturally, but their properties are of interest, and proteins, such as a-amylase and lysozyme, which have been modified by them retain their full enzymic activity. Such derivatives of o-galactose showed markedly enhanced binding to liver membranes in a specific manner not exhibited by those of o-mannose or N-acetylglucosamine--an obvious analogy to the in vivo binding of asialoglycoproteins with D-galactose nonreducing termini to such membranes. Methods The present survey has given some indications of the diversity of approaches employed in the preparation of di- and oligosaccharides. There is obviously no universal method, and a great deal of research is often necessary before a desired derivative can be successfully synthesized. The author considers that it might be of interest to discuss in more detail a number of these preparations and some of the problems and pitfalls involved. Examples of the use of mercuric salts with peracetylglycopyranosyl halides in the Koenigs-Knorr reaction were recently described in full 84 and will not therefore be repeated here: I shall limit myself to a description ~0 S. A. Holick, S.-H. L. Chiu, and L. Anderson, Carbohydr. Res. 50, 215 (1976). 8, S.-H. Chiu and L. Anderson, Carbohydr. Res. 50, 227 (1976). az M. A. E. Shaban and R. W. Jeanloz, Carbohydr. Res. 26, 315 (1973). ~:~M. Krantz, N. A. Holtzmann, C. P. Stowell, and Y. C. Lee, Biochemistry 15, 3963 (1976). 84 H. M. Flowers, Methods Carbohydr. Chem. 6, 474 (1972).

114

PREPARATIONS r0

OH

r0

[7]

Br

{ Villi

NHAc

( ix1 FIc. 9. Nonparticipating C(2)-O-benzyl group in cis-(1,2)-disacchafide synthesis.

of the use of perbenzylated halides for the synthesis of l,2-cisdisaccharides and a number of approaches with amino sugars.

2-Acetamido-2-deoxy-3-O-oeL-Fucopyranosyl-D-Glucose (IX) Disaccharide (IX) forms part of the saccharide chains of A, B, H and Le blood-group substances s~ and its /3-e-anomer was synthesized from tri-O-acetyl-a-L-fucopyranosyl bromide, 86 stereochemistry presumably being controlled by the participating acetoxy group at C(2) in the fucosyl portion. The use of a halide with a nonparticipating group at C(2) enabled preparation of the required a-L-disaccharide. 87 The two reactants needed for the condensation are 2-O-benzyl3,4-di-O-p-nitrobenzoyl-a-L-fucopyranosyl bromide (VII) and benzyl 2-acetamido-4,6-O-benzylidene-2-deoxy-a-t)-glucopyranosideS, (VIII) and are prepared as follows: Compound (VII). The known methyl 3,4-O-isopropylidine-a-e-fucopyranoside 89 is benzylated by stirring it for 2-3 hr at 100° with benzyl alcohol and powdered potassium hydroxide in toluene. After processing, an oil is obtained that is purified by chromatography on silica gel. The resulting colorless oil is hydrolyzed by stirring for 2 hr at 100° with 1.5 M sulfuric acid. After neutralization of the acid with barium carbonate, the aqueous solution is evaporated and the residue is chromatographed on silica gel. The product, 2-O-benzyl-e-fucose, crystallizes from water. xz K. O. Lloyd and E. A. Kabat, Proc. Natl. Acad. Sci. U.S.A. 61, 1470 (19681. "" E. S. Rachaman and R. W. Jeanloz, Carbohydr. Res. 10, 429 (19691. ~7 M. Dejter-Juszynski and H. M. Flowers, Carbohydr. Res. 30, 287 (1973). x~ R. Kuhn, H. H. Baer, and A. Seeliger, Justus Liebigs Ann. Chem. 611,286 0958). 8~ E. E. Percival and E. G. V. Percival, J. Chem. Soc. p. 690 (1950).

[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

1 15

Treatment of this product with p-nitrobenzoyl chloride a° in dry pyridine gives crystalline 2-O-benzyl-l,3,4-tri-O-p-nitrobenzoyl-ot-L-fucopyranose, m.p. 202°-204°; [a]o 26 - 2 8 5 ° (c 1.30, chloroform). Treatment of a solution of the p-nitrobenzoate in dry dichloromethane with dry hydrogen bromide leads to replacement of the C(1)-pnitrobenzoate by Br. p-Nitrobenzoic acid precipitates and is removed by filtration, while the filtrate contains the required bromide (VII), which is isolated as an oil and can be purified, if necessary, by chromatography on silica gel. It is not very stable, especially in the light and in the presence of moisture, and so is usually employed directly for condensation with the nucleophile; [a]o 27 - 272° (c 1.0, chloroform). Compound (VIII). A stirred mixture of N-acetylglucosamine and excess benzyl alcohol containing 0.5% HCI, TM is kept for 30 min at boiling point. The acid is removed by neutralization, and the product is precipitated by the additior~ of excess diisopropyl ether. Crystallization from alcohol gives benzyl 2-acetamido-2-deoxy-a-o-glucopyranoside contaminated with some /3-0 anomer, which can be removed completely, with considerable losses, by several recrystallizations from alcohol. However, it is unnecessary to employ pure material in the next stage, and the 4,6acetal obtained is readily purified in good yield. The acetal is prepared by treatment of the above product with excess benzaldehyde and zinc chloride overnight at room temperature or for 30 minutes at 60°, followed by precipitation as a white solid by the addition of water and petroleum ether. Crystallization from pyridine-water affords a beautifully crystalline 4,6-benzylidene compound that is readily freed of any residual /3-anomer by a second crystallization; m.p. 260°-261 °, [Ot]D 23 "4- 114° (c 1.1 pyridine).

Benzyl 2-Acetamido-3-O-(2-O-benzyl-a-L-fucopyranosyl)-2-deoxy-a-Dglycopyranoside. A solution of (VIII) (2.0 g) in nitromethane-benzene (100 ml, equal volumes) is concentrated to about 80 ml by azeotropic distillation and cooled to 40 °. Mercuric cyanide (1.3 g) and compound (VII) (3.2 g) are added, and the mixture is stirred with exclusion of moisture for a total of 48 hr, a further portion of compound (VII) (1.6 g) being added after 24 hr. The mixture is diluted with benzene, and the ,o It is important in this esterification reaction to employ pure p-nitrobenzoyl chloride. Most commercial samples are contaminated with considerable quantities of free acid: these pose problems in the reaction owing to formation of p-nitrobenzoic anhydride and sodium p-nitrobenzoate in the work-up, which are not easily removed. The chloride can be readily purified by extraction of the commercial sample with excess hot, dry benzene, concentration of the filtered extract, and precipitation by addition of dry petroleum ether. 9~ The HCI can be conveniently replaced by Dowex 50 (H +) resin (l g for each i0 ml of alcohol), facilitating processing.

116

PREPARATIONS

[7]

organic layer is washed several times with sodium hydrogen carbonate solution and then with water, dried, and evaporated. Mild acid hydrolysis (by stirring a suspension of the product in a mixture of 100 ml o f p - d i o x a n e and 40 ml o f aqueous 0.5 M H2SO4 for 1 hr at 100°) removes the benzylidene group selectively, and the product is extracted into chloroform and purified by chromatography on silica gel. A syrup results [overall yield from compound (VIII) approx. 50%], 92 which is deacylated to the title compound in almost quantitative yield by adding a catalytic a m o u n t of sodium methoxide to a solution in methanol. It crystallizes from alcohol, m.p. 224°-226°; [O~]D23 +37° (C 1.16 methanol). 2-Acetatnido-2-deoxy-3-O-a-L-fucopyranosyl-a-o-glucose (IX). A solution of the above product (0.50 g) in 90% ethanol (100 ml), containing a drop of acetic acid, 93 is hydrogenolyzed in the presence of 10% palladium-on-charcoal (50 mg) at 3.5 atm for 48 hr at room temperature. The product often contains a little unhydrogenolyzed material and, if necessary, is purified by chromatography on silica gel. 94 Crystallization from ethanol-methanol-water gives a crystalline disaccharide, m.p. 218°-220 ° (dec); [a]D 23 --60 ° ---> --74 ° (C 0.83, water). Reduction of the free disaccharide by sodium borohydride in the presence of excess boric acid to the corresponding glycitol, followed by trimethylsilylation, gives a product showing only one peak on gas-liquid chromatography, which is readily distinguished from that obtained from the corresponding fl-linked disaccharide. Since none of the "fl" peak appears even when glycitol prepared from disaccharide isolated without purification of intermediates is injected, it is established that the reaction o f (VII) and (VIII) is completely stereospecific.

2-Acetamido-2-deoxy-O- fl- D-galactopyranosyl-( l --> 4)-0- [3-D-galactopyranosyl-( l --->4)- D-glucose a~ (XVII) Trisaccharide (XVII) is the carbohydrate involved in t h e glycolipid accumulating in the brain of patients with amaurotic familial idiocy, viz., asialo-Tay-Sachs ganglioside. Its synthesis illustrates an approach to glycosides of amino sugars and utilization of 1,6-anhydro derivatives to increase the reactivity of C(4) h y d r o x y l groups. A mixture of 2-dichloroacetamido-2-deoxy-3,4,6-tri-O-benzoyl-a-ogalactopyranosyl bromide (X) (15 mmol), 2-O-acetyl-l,6-anhydro-fl-D~'~Use of the purified bromide (VII) in the reaction raises the yield by 10-20%, but its purification involves considerable losses owing to its lability. 93The free disaccharide is extremely labile to alkali but less so to weak acid, hence it is desirable to ensure a pH of less than 7 during hydrogenolysis and processing. ~4Crystallization is hindered by the presence of inorganic contaminants. ~':'D. Shapiro, A. J. Acher, and Y. Rabinsohn, Chem. Phys. Lipids 10, 28 (1973).

[7]

CHEMICAL SYNTHESIS

C

2

~2o ~

.zo~o (x)

OF O L I G O S A C C H A R I D E S

OR20

)~~o~

~

c.2-o .o)'-~.I OR3

. c..~o~. ~-o ~0)--~o .o, )--:ol NHR

~ c.zo~3 R30)-rO,0, }--q

OR

NHR3

(x'm)

(x I'V )

CHEOH GH20H

.oj--o

.o.,Lo.,o,

NHR3 .0~--0

CH-O

z ~.-~o,l

(xlT)

(X])

(x~) - - - ( ~ )

117

OH

CH20H

)-o. OH

(xvIT)

(x~.) FIG. 10. Condensation of the C(1) halide of a protected amino sugar with a derivative of 1,6-anhydro-/3-D-galactose. The C(1) halide of the resulting disaccharide is c o n d e n s e d with a derivative of 1,6-anhydro-/3-D-glucose. (R 1 = COCHClz; R 2 = Bz; R a = Ac)

galactopyranose (XI) (17 mmol), and mercuric cyanide (15 mmol) in dry ethylene chloride (100 ml) is stirred at 40° for 2 days with exclusion of light and moisture. It is then washed with sodium bicarbonate solution and water, and the solvent is removed by evaporation. Chromatography over a silica gel column separates (XIII) from its 1 ~ 3 isomer (XII), and it can be crystallized from isopropanol. The rather higher reactivity of the C(4)-hydroxyl (equatorial in the stable conformation of 1,6-anhydro-/3-Dgalactose) over that of C(3), which is axial, is shown by the ratio obtained of (XIII) : (XII) = 3 : 2. Careful O-deacylation of (XIII) with barium methylate in the cold can be performed; alternatively, overnight treatment at ambient temperature results in O- and N-deacylation. The product is completely (O and N) acetylated with acetic anhydride-pyridine, and the resulting peracetate is acetolyzed by keeping a solution of it (0.8 g) in a mixture of acetic anhydride (14 ml), acetic acid (6 ml), and sulfuric acid (0.14 ml) at 40 ° for 4 hr. After processing, the product (XIV) is purified by column chromatog-

118

PREPARATIONS

[7]

raphy on silica gel G and converted to the bromide (XV) by treatment with HBr-acetic acid at 4 ° for several hours, following the reaction by thin-layer chromatography (TLC). The 1,6-anhydroglucose derivative (XVI) is prepared from phenyl 2,3,6-tri-O-acetyl-fl-D-glucopyranoside by way of its 4-O-t-butyl ether. 9~ Condensation of XV and XVI is effected similarly to that described for (XIII), and the product is isolated in 60% yield and can be crystallized from methanol-isopropyl ether. After acetolysis as described previously, followed by careful O-deacetylation (barium methylate in methanol at 2°), the desired trisaccharide (XVII) is precipitated from methanol solution by the addition of ether as a white, hygroscopic powder; m.p. 185°-188 °, [O£]D23 30.3 ° (c 0.8, water).

p-Nitrophenyl-O-( 2-acetamido-2-deoxy- [~D-glucopyranosyl)-(1---~3)-O(2-acetamido-2-deoxy-[3- D-glucopyranosyl)-(1--~6)-2-acetamido-2deoxy-fl-D-glucopyranoside (XXIII) 53 Oxazolines such as (XVIII) and (XXI) are prepared in a general way by treatment of the C(1) halide (obtained from the amino sugar hydrochloride or disaccharide with HCI in acetic anhydride) with soluble silver salts in collidine; silver perchlorate is especially effective. When a mixture of (XVIII) (1.0 g), (XIX) 97 (0.85 g), and p-toluenesulfonic acid (10 mg) in dry nitromethane (50 ml) is heated for 30 min at 110° and then cooled, a product precipitates that is collected, washed with nitromethane and ether, and recrystallized from p-dioxane. The yield is excellent (1.25 g, 81%). After debenzylidenation with hot aqueous acetic acid, catalytic deacetylation, and removal of the benzyl group by catalytic hydrogenolysis, disaccharide (XX) is isolated as a crystalline product (from methanol), and is readily converted into the reactive oxazoline (XXI). A mixture of (XXI), p-nitrophenyl 3,4-di-O-acetyl-2-acetamido-2deoxy-/3-D-glucopyranoside (XXII), 98 and p-toluenesulfonic acid in 1 : 1 nitromethane-toluene are treated as described for (XX), using a ratio of solvent:XXI = 10:1 (v/w). Again, the product of condensation precipitates in high yield and can be recrystallized from methanol. A suspension of this product (480 mg) in dry methanol (60 ml) containing 1 M sodium methylate in methanol (0.6 ml) is heated at 400-45 ° until dissolved (a few minutes). The solution is kept for 16 hr at 5°, and a 9~ D. Shapiro, Y. Rabinsohn, A. J. Acher, and A. Diver-Haber,.J. Org. Chem. 35, 1464 (1970). .~7p. H. Gross and R. W. Jeanloz, J. Org. Chem. 32, 2759 (1967). '~ T. Osawa, Carbohydr. Res. 1,435 (1966).

[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

CH,O~

pC.z

AoO,~, "o

o,~,

N,~ CMe (xvm)

C~O.

CH:,OH

.o~--~

NHAc

.o,~o.

NHAc

(xlx)

NHAc (xx)

CH2OAc

AcO~L-.-( CH2OH

CH2OH OCH2

119

CH2OAc

AC6"~L~/ 9

NHAc

/

~.=~,_ /

N ~CMe

(xxI)

/ HO'~N i~Ac HO~'~NHAHO~NHAc (xxm)

c~

oojO-¢,~'"°2 A¢O'---I' NHAc (xxE)

FIG. 11. Use of oxazolines in synthesis of oligosaccharides of amino sugars.

precipitate forms, which is collected and washed with methanol. A further crop of crystals can be removed from the mother liquors by desalting, evaporation to a small volume, and cooling. After recrystallization from aqueous ethanol, trisaccharide (XXIII) is obtained pure; m.p. 209°-210 ° (decomp.), [O~]D20 "~-8° (C 0.18, aq. methanol). This synthesis shows the high reactivity of the oxazolines and the excellent yields obtained on condensation with either primary or secondary hydroxyl groups. Stereospecificity in the reaction is apparently complete.

Use of 2-Nitrosoglycopyranosyl Chlorides in the Synthesis o f cis-l,2Linked Disaccharides of Hexoses and 2-Amino Sugars ~.~r.~8 Condensation Reaction. A mixture of (XXIV) 99 (13 g) and (XXV) (8 g) in N,N-dirnethylformamide (210 ml) is kept at ambient temperature for 36 hr. The resulting solution is evaporated to a yellow oil, which is dissolved in 3 : 2 ether-Skellysolve B and passed through a column of silicic acid. The major fraction is collected and crystallized from ether-Skellysolve B; yield, 10.2 g of (XXVI) (59%), m.p. 160°-161 °, ~'~R. U. Lemieux and T. L. Nagabhushan, Methods Carbohydr. Chem. 6, 487 (1972).

120

PREPARATIONS

[7]

] CHzOAc|

I O-- C,"2 u 0 0

CHzOA¢ M._p~O'cH2

Me 2C...

~q)-o

~ o~o_

-

NOR

IXXW)

IXXV)

CH2OH

XX VI; R,H; X X V I I , R , A c

Me2 c"O"CH2

CH20H

HO,L-o

°

~--(b - - - - - - - - - - - L 4 ~ NHA¢ (xxva)

CH20H

)--O

" ~.i~0./'~/~OH NHAc Ot4

O-CMe2

(xxv~)

I.B 2 H 6

O~CMe 2

(XXX)

2.Ac20

CH2OH .o,_o -

Me2C~% H2

CH2OH HO O

I~am~.. 0 -

. O-CMe 2

CH2OH O OH OH

( XXIX)

(XXX]')

CHzOH

HO)--O

) -

CHz,OH q

OHio~ O H OH

I'TiCI3 I (XXVZ) 2"N°BH4 [

( XxXll )

CH2OH H0)--O~ (XXX m')

OH

CH20H f ~ OH OH

FIG. 12. S y n t h e s i s of cis-l,2-disaccharides from 2-nitroso chlorides. T h e intermediate o x i m e s can be r e d u c e d to amines or c o n v e r t e d to ketones that give hydroxyl groups on reduction, s t e r e o c h e m i s t r y depending to a considerable extent on the conditions of reduction.

[OL]D27 +66 ° (c 1:1, chloroform). Acetylation of (XXVI) with acetic

anhydride-pyridine at 0 ° for 12 hr gives an oil that crystallizes from ether-Skellysolve B (compound XXVII). Reduction of (XXVII). A solution of (XXVII) (I0 mmol) in tetrahy-

[7]

CHEMICAL SYNTHESIS OF OLIGOSACCHARIDES

121

drofuran (THF) is cooled to - 5 ° and kept under an atmosphere of N2. A molar solution of diborane in T H F (45 ml) is added at such a rate that the temperature remains at - 5 ° . The solution is kept at room temperature for 3 hr, then methanol (100 ml) is added dropwise and the solution is evaporated to a solid residue, which is redissolved in methanol, deionized, and concentrated to a foam which is a mixture (TLC) of galacto and talo epimers. N-Acetylation. The above product is dissolved in 50% aqueous methanol (20 ml), and acetic anhydride (10 ml) is added. After 2 hr at room temperature, the solution is evaporated in vacuo and the residue is chromatographed on Dowex 1 (OH). The first fraction collected (35%) is the galacto epimer, and then the talo epimer (47%) emerges from the column. Both compounds can be crystallized, the first from ethyl acetate and the second from ethyl acetate-Skellysolve B. De-O-Acetylation. A solution of the compound (10 mmol) in 50% aqueous methanol (50 ml) containing triethylamine (2 ml) is kept at 4 ° for 12 hr and then evaporated to dryness. De-O-Isopropylidenation. A solution of the compound (10 mmol) in 90% trifluoroacetic acid (50 ml) is kept at room temperature for 15 min. The acid is removed by evaporation in vacuo, and the resulting oil is triturated with ether to give an amorphous powder: (XXX) or (XXXI). 3-O-~D-Galactopyranosyl-D-glucose (XXXII) and 3-O-a-D-Talopyranosyl-D-glucose (XXXIlI). A 20% aqueous solution of titanium trichloride (30 g) is added dropwise to a mixture of compound (XXVI) (6.5 g), ammonium acetate (12 g), and dioxane (60 ml) stirred under an atmosphere of N2. Stirring is continued for an additional 3 hr, then further portions of ammonium acetate (12 g) and dioxane (60 ml) are added, followed by TiCI3 (30 g, dropwise). The solution is stirred for 2 hr longer, then extracted with methylene chloride (500 ml); the organic layer is washed with saturated aqueous NaHCO3 and water and evaporated to an oil {15.9 g). Part of the oil (5.7 g) is reduced with sodium borohydride (0.6 g) in 60% aqueous dioxane (45 ml) for 3 hr, and the solution is deionized and concentrated to an oil that is deacetylated with triethylamine in methanol. The resulting product (5 g) contains two components (TLC) and is passed through a column of Dowex 1 (OH). The galacto epimer (1.14 g) emerges first, followed by the talo epimer (1.39 g). Both compounds can be crystallized from ethyl acetate-Skellysolve B. After de-O-isopropylidenation with 90% trifluoroacetic acid, the free disaccharides are isolated as amorphous powders (galacto epimer (XXXII), [a]D22 +159°; talo epimer (XXXIII), [O~]Dz2 +115°).