Automated Solid-Phase Synthesis of Oligosaccharides

See allHide authors and affiliations

Science  23 Feb 2001:
Vol. 291, Issue 5508, pp. 1523-1527
DOI: 10.1126/science.1057324


Traditionally, access to structurally defined complex carbohydrates has been very laborious. Although recent advancements in solid-phase synthesis have made the construction of complex oligosaccharides less tedious, a high level of technical expertise is still necessary to obtain the desired structures. We describe the automated chemical synthesis of several oligosaccharides on a solid-phase synthesizer. A branched dodecasaccharide was synthesized through the use of glycosyl phosphate building blocks and an octenediol functionalized resin. The target oligosaccharide was readily obtained after cleavage from the solid support. Access to certain complex oligosaccharides now has become feasible in a fashion much like the construction of oligopeptides and oligonucleotides.

The understanding of oligosaccharides and glycoconjugates in nature is still in its infancy (1). Cell surface glycoconjugates are involved in signal transduction pathways and cellular recognition processes and have been implicated in many disease states (2). A major impediment to the rapidly growing field of molecular glycobiology is the lack of pure, structurally defined carbohydrates and glycoconjugates. These biomolecules are often found in low concentrations and in microheterogeneous form in nature, greatly complicating their identification and isolation. The procurement of sufficient quantities of defined oligosaccharides required for detailed biophysical and biochemical studies therefore relies on efficient synthetic methods.

Although much progress has been made in oligosaccharide synthesis (3), the construction of complex carbohydrates remains time consuming and is carried out by a small number of specialized laboratories. The necessary functionalization of all hydroxyl groups present on a monosaccharide is one of the main challenges in oligosaccharide construction. The development of a protecting group scheme that allows for the manipulation of individual hydroxyl groups is pivotal for the success of the synthetic route. Permanent protecting groups, such as benzyl ethers, are installed at positions where a free hydroxyl will be present in the final deprotected molecule. Temporary protecting groups, such as esters, are used to mask hydroxyls that will be exposed at a certain stage of the synthesis. The liberated hydroxyl group then serves as a nucleophile in the reaction with a glycosylating agent.

The stereospecific formation of glycosidic bonds is the central challenge in carbohydrate chemistry (Fig. 1). The chemical formation of a glycosidic linkage involves activation of a glycosyl donor to create a reactive electrophilic species that couples with a nucleophilic acceptor hydroxyl. This coupling reaction can take two possible pathways resulting in formation of either α- or β-anomers. Current methods to control the stereochemistry of the anomeric center rely on the participation of a neighboring functionality, such as an ester-protecting group on the C2 hydroxyl. Formation of a cyclic oxonium ion intermediate shields one face of the molecule, leading exclusively to the formation of trans-glycosidic linkages. Cis-glycosidic bonds are difficult to construct with high specificity because neighboring group participation is not possible.

Figure 1

Stereochemical issues in the synthesis of carbohydrates: Oligosaccharides require the formation of a particular stereoisomer (α or β anomer) during each coupling event. The use of participating groups, such as esters, leads to exclusive formation of trans-glycosidic linkages.

The evolution of a solid-phase paradigm for the construction of oligosaccharides was initiated with Frechet's synthesis of di- and trisaccharides on a polymer support in 1971 (4). Since then, solid-phase oligosaccharide synthesis has seen many advancements. Various glycosyl donors, such as anomeric sulfoxides (5) and anhydrosugars (5), have been applied to the synthesis of carbohydrates on a polymer support. Several linkers serving to connect the growing oligosaccharide chain to the polymer support have been introduced with different reactivities and cleavage procedures (6). Notably, a combinatorial split-and-mix approach on a support resulted in the synthesis of a library of N-acylated di- and trisaccharides (7). Although advancements in solid-phase chemistry have allowed for the construction of complex molecules, the manipulations remain tedious and time consuming.

Ultimately, a general, automated method for oligosaccharide assembly will allow for the rapid preparation of structures of interest. Oligonucleotides (8) and oligopeptides (9) are now routinely prepared in an efficient manner on automated synthesizers with solid-phase strategies. The solid-phase paradigm lends itself particularly well to automation of oligosaccharide synthesis as the repetitive nature of glycosylation and deprotection can easily be framed into a coupling cycle. Excess reagents can be used to drive reactions to completion, and resin washes can remove any soluble impurities. Only a single purification step is necessary after the sugar is liberated from the solid support.

Mindful of the advantages of solid-support synthesis, we considered several key issues for the development of an automated oligosaccharide synthesizer: (i) an instrument capable of performing repetitive chemical manipulations at variable temperatures, (ii) the design of an overall synthetic strategy with either the reducing or the nonreducing end of the growing carbohydrate chain attached to the support (10), (iii) selection of a polymer and linker that are inert to all reaction conditions during the synthesis but cleaved efficiently when desired, (iv) protecting group strategies consistent with the complexity of the target oligosaccharide, and (v) stereospecific and high yielding glycosylation reactions.

Rather than designing a new machine, we opted to reengineer an existing apparatus used in automated peptide synthesis (11). Several adaptations were necessary before the peptide synthesizer could be used for carbohydrate synthesis (12). Using this modified peptide synthesizer, we undertook a systematic investigation of the variables involved in automated solid-phase oligosaccharide synthesis. We chose the “acceptor bound” strategy for solid-phase oligosaccharide synthesis (13). In this method, the reactive glycosylating agent is delivered in solution while the nucleophilic acceptor hydroxyl group is exposed on the solid support. Productive coupling events result in support-bound oligosaccharides that are purified by simply washing the soluble side products through a filter. Removal of a temporary protecting group on the newly formed saccharide unit reveals another hydroxyl group, thereby continuing the coupling cycle.

To explore the compatibility of glycosylating agents and protecting groups with a solid-phase linker and a polymer support, we investigated the construction of polymannosides. The synthesis of α-mannosides has been the focus of substantial research because of their occurrence in biological structures such as glycolipids andN-linked glycoproteins (14, 15). A series of α-(1→2) mannosides (Scheme 1,3 to 5) served as our initial targets for automation because structures of this type have been synthesized previously in solution and on the solid support (16). Trichloroacetimidate donor 2 was chosen as the donor building block because it can be prepared on a multigram scale, is activated at room temperature, and bears a C2-ester functionality to control the anomeric configuration of the polymer (17). Activation of 2 was carried out under acidic conditions with the Lewis acid trimethylsilyl trifluoromethanesulfonate (TMSOTf). Removal of the acetyl ester-protecting group was accomplished under basic conditions with sodium methoxide.

To allow for the use of acidic and basic reaction conditions in the coupling cycle, we investigated a polymer support and linker for compatibility (18). A variety of commercially available polymer supports were examined. Merrifield's resin (1% cross-linked polystyrene) and polystyrene-based Argopore displayed excellent properties throughout the coupling cycle. Our previous work demonstrated that olefinic linker 1 was stable to the coupling cycle conditions while readily cleaved from the solid support at the end of the synthesis by olefin cross metathesis. By varying the concentration and quantity of reagents as well as the reaction times, we arrived at the cycle shown in Table 1. Applying the conditions in Table 1 with octenediol functionalized 1% cross-linked polystyrene, the synthesis of pentamannoside 3was carried out in 14 hours (Scheme 1).

Table 1

Cycles used with trichloroacetimidate and phosphate donors.

View this table:

Analysis of the resin-bound oligosaccharide is essential for the successful development of a solid-phase synthesis method (19). Two-dimensional nuclear magnetic resonance (2D-NMR) analysis of the resin-bound pentamer 6 was performed with high-resolution magic angle spinning (HR-MAS) NMR techniques (Fig. 2). Analysis of the NMR spectra revealed characteristic anomeric signals between 97 and 103 parts per million. Further homonuclear total correlation spectroscopy (TOCSY) HR-MAS analysis confirmed the presence of five unique anomeric protons (20). In accordance with previous experiments, minor line broadening was observed in the spectra of the resin-bound sample (21). Overall, the HR-MAS NMR data of the polymer-bound pentamer corresponded unequivocally with an authentic pentamer sample prepared in solution. The remarkable purity of the resin-bound pentasaccharide 6, after nine synthetic steps without any purification, encouraged us to explore the synthesis of larger structures. Heptamer 4 and decamer 5 were prepared in average yields of 90 to 95% per step. The short reaction times, 3 hours per monomer unit, allowed for the synthesis of4 in 20 hours and 42% overall yield. As a comparison, we manually synthesized heptamannoside 4 on the solid support in 14 days and 9% overall yield (18). These results demonstrate that the synthesis of a linear oligosaccharide is fast and high yielding when constructed on an automated synthesizer.

Figure 2

2D-NMR comparison of (A) solution-phase and (B) resin-bound pentamer.

Given the success with the automated α-mannoside construction, the fully protected phytoalexin elicitor (PE) β-glucan 7 was selected as a more complex target structure (Fig. 3) (22). The presence of a fungal β-glucan oligosaccharide triggers the soybean plant to release antibiotic phytoalexins. The response initiated by the PE β-glucans in the host soybean plant is the most studied defense mechanism in plants. These oligosaccharides have been synthesized previously in solution (23) and on the solid support (24) and were expected to serve well as a benchmark in our automation endeavor.

Figure 3

Dodecamer phytoalexin elicitor β-glucan.

For the synthesis of the branched β-(1→3)/β-(1→6) PE structure, we envisioned the use of two different glycosyl phosphate donors, 8 and 9. Recently, we introduced glycosyl phosphates as glycosylating agents that are readily prepared from glycal precursors and that performed well in solution and on a solid support (25). Strategic protecting group considerations prompted us to use the levulinoyl ester as a 6-O temporary protecting group and the 2-O-pivaloyl group to ensure complete β-selectivity in the glycosylation reaction. Deprotection of the levulinoyl ester was accomplished with a hydrazine solution in pyridine/acetic acid, whereas the phosphate building block was activated with TMSOTf.

Unlike peptide and nucleic acid synthesis, many of the manipulations involved in oligosaccharide chemistry are not carried out at room temperature. Drawing from solution-phase studies, we were cognizant that the use of glycosyl phosphates, like many donors (26), would require low temperature for optimal results. To address this need, we designed a temperature-controlled reaction vessel. The vessel is enclosed by a cooling jacket that is easily attached to a commercial cooling apparatus. Model reactions with phosphate donor 8 demonstrated the ease of incorporating a temperature variable in the automation cycle.

The coupling and deprotection conditions were adjusted for the use of glycosyl phosphates and levulinoyl esters, resulting in the cycle shown in Table 1 (phosphate cycle). The activation of phosphate donor8 at −15°C required shorter reaction times than were needed for trichloroacetimidate 2, as determined in preliminary solution-phase studies. As anticipated from previous studies (27), the levulinate ester could be rapidly removed at +15°C (15 min), compared with longer reaction times necessary for acetyl ester cleavage. The levulinoyl group deprotection cycle incorporates a different washing cycle (pyridine/acetic acid) than the acetyl group to ensure removal of any excess hydrazine. As in the synthesis of polymannosides 3 to5, double glycosylations and double deprotections were used. Incorporation of these modifications to the automated cycle resulted in an excellent yield [92% by high-pressure liquid chromatography (HPLC) analysis] of a model β-(1→6) trisaccharide.

The automated cycle was then applied to the synthesis of more complex PE oligosaccharides with alternating phosphate building blocks (Scheme 2). Branched hexasaccharide 10 was constructed in 10 hours in >80% yield as judged by HPLC analysis. Also, we prepared dodecasaccharide 7 (Fig. 3) in 17 hours and >50% yield using the same cycle. Notably, the solution-phase synthesis of only two phosphate building blocks was necessary, which greatly reduces the manual labor usually required to assemble a structure of this size. The expedient generation of material through automation represents a major improvement over conventional methods for polysaccharide synthesis.

Scheme 1

Automated oligosaccharide synthesis with trichloroacetimidates. Glycosylation conditions: 25-μmol scale: 25 μmol of resin (83 mg, 0.30 mmol/g loading); 10 equiv. donor2 (160 mg); 0.5 equiv. TMSOTf (1 ml, 0.0125 M TMSOTf in CH2Cl2) repeated two times for 30 min each. Deprotection conditions: 25-μmol scale: 10 equiv. NaOMe (0.5 ml, 0.5 M NaOMe in MeOH) in 5 ml of CH2Cl2repeated two times for 30 min each.

Scheme 2

Automated oligosaccharide synthesis with glycosyl phosphates. Glycosylation conditions: 25 μmol scale: 25 μmol resin (83 mg, 0.30 mmol/g loading); 5 equiv. donor 8 or9 (90 and 170 mg, respectively); 5 equiv. TMSOTf (1 ml, 0.125 M TMSOTf in CH2Cl2) repeated two times for 15 min each at –15°C. Deprotection conditions: 25 μmol scale: 4 ml, 0.25 M N2H4 in pyridine:acetic acid (3:2) repeated two times for 15 min each at 15°C.

We have demonstrated that both glycosyl phosphates and trichloroacetimidates are useful donors in the automated synthesis of oligosaccharides, as well as the utility of acetate and levulinate esters as temporary protecting groups. To illustrate the generality of this method, we synthesized trisaccharide 13, incorporating all aspects of our automated chemistry. This trisaccharide motif occurs in complex type N-linked glycoprotein structures and contains two challenging linkages. Glycosylations of the C2 position of mannose with glucosamine donors and the C4 hydroxyl of glucosamine with galactose donors are notoriously difficult reactions, often leading to the formation of unwanted side products (28).

The automated synthesis of trisaccharide 13 required the preparation monosaccharide building blocks (2,11, and 12). Donor 2 was chosen on the basis of the glycosylation and deprotection protocols developed for the synthesis of α-(1→2) mannosides. Glucosamine donor 11was designed with the phthalimide amine-protecting group to confer β-selectivity during glycosylation. Incorporation of a levulinate ester at the C4 position of 11 allows for rapid deprotection with hydrazine as demonstrated in the synthesis of the PE β-glucans. Galactosyl phosphate 12 was chosen to fashion the terminal glycosidic linkage on the basis of the high reactivity of this donor with unreactive substrates (25).

The automated synthesis of trisaccharide 13 was carried out with the cycle described in Scheme 3(29). HPLC analysis of the crude reaction mixture after cleavage from the support indicated a 60% overall yield. Subsequent removal of all protecting groups was carried out to demonstrate that oligosaccharides prepared in an automated fashion are efficiently deprotected. Cleavage of the phthalimide group with hydrazine (30) and N-acetylation with acetic anhydride were followed by removal of the pivaloyl group with LiOH (5). Global debenzylation proceeded smoothly with concomitant reduction of the olefin functionality to afford n-pentyl glycoside14 in 62% yield from the fully protected trimer13. The cleavage conditions described here for the removal of phthaloyl, pivaloyl, and benzyl groups are routine procedures (31) that can be applied to a wide variety of substrates.

Scheme 3

Automated synthesis of trisaccharide 13. 100 μmol scale: 100 μmol resin (333 mg, 0.30 mmol/g loading); a, 4 equiv. donor 2, 0.4 equiv. TMSOTf repeated two times for 30 min each. b, 5 equiv. NaOMe in CH2Cl2 repeated two times for 30 min each. c, 4 equiv. donor 11, 0.4 equiv. TMSOTf repeated two times for 15 min each at –15°C. d, 4 ml, 0.25 M N2H4 in pyridine:acetic acid (3:2) repeated two times for 15 min each at 15°C. e, 4 equiv. donor12, 4 equiv. TMSOTf repeated two times for 15 min each at –15°C. f, 16.4 mg catalyst, 1 atm CH2CH2, CH2Cl2. g, N2H4, EtOH, 90°C. h, Ac2O, MeOH/CH2Cl2. i, LiOH, THF/CH3OH. j, H2, Pd/C, EtOH.

In summary, these examples illustrate major improvements in time and yield as compared with manual syntheses of oligosaccharides. A glycosylation/deprotection cycle was developed and applied to the synthesis of a decamer of an α-(1→2) mannoside. Two phytoalexin elicitor β-glucans, hexasaccharide 10 and dodecasaccharide7, were constructed in rapid fashion with glycosyl phosphates. Trisaccharide 13 was assembled with both glycosyl phosphate and trichloroacetimidate donors and a protecting group strategy based on acetate and levulinate esters. The automated method described above represents an important advance toward streamlining the synthesis of oligosaccharides.

Although several limitations still exist in oligosaccharide construction, it is now possible to transfer many solution and solid-phase chemistries to an automated synthesizer. A number of challenges such as the synthesis of sialic acid containing oligosaccharides and heparinlike glycosaminoglycans still remain. Additional research in these areas is needed to further advance the field of oligosaccharide synthesis. The ease of acquiring defined structures from a machine will impact the field of glycobiology such that we may one day be able to fully appreciate the importance of oligosaccharides and glycoconjugates in nature.

  • * To whom correspondence should be addressed. E-mail: seeberg{at}


View Abstract

Stay Connected to Science

Navigate This Article