Total synthesis of zwitterionic PS A2 oligosaccharides from Bacteroides fragilis

Abstract

Zwitterionic polysaccharides have unique structures and can elicit T cell–dependent immune responses. To unravel their molecular mode of action and develop potential alternatives for carrier proteins in carbohydrate-based vaccines, we here describe a total synthesis of zwitterionic PS A2 oligosaccharides from Bacteroides fragilis. The pentasaccharide and decasaccharide bearing one or two repeating units were efficiently synthesized via convergent glycosylation strategy. The ingenious protecting groups have enabled the stereoselectivity of glycosylation. A unique synthetic approach was developed to introduce the chiral 3-hydroxybutanoic acid side chain on mannoheptose, which proved very challenging to form in this skeleton.

Zwitterionic PS A2 oligosaccharides synthesis for carbohydrate-based vaccine development.

INTRODUCTION

Bacterial surface poly/oligosaccharides have been shown to be effective antigens for the development of antibacterial vaccines (1–7). Normally, standalone polysaccharides are T cell–independent antigens and can only elicit low-affinity immunoglobulin M (IgM) responses. This limitation of polysaccharide vaccines can be overcome by conjugation of the saccharides to a carrier protein, such as mutant cross-reacting material from diphtheria toxin, diphtheria toxoid, or tetanus toxoid (8, 9). This conjugation can recruit T cell help and induce a T cell–dependent immune response, inducing a long-lasting and boostable IgG antibody repsonse and immunological memory (10, 11). Clinically, glycoconjugate vaccines have been widely used to prevent bacterial infections, such as Streptococcus pneumoniae (PCV7, PCV13, PCV15, and PCV20), Haemophilus influenzae, Neisseria meningitidis (serogroups A, C, and ACWY), and Salmonella typhi (7, 9, 12–16). A considerable challenge in the development of glycoconjugate vaccines is carrier-induced epitope suppression (CIES), which appears reduced immunogenicity when preexisting immunity is present against the same carrier protein it comprises. CIES has been well documented in clinical studies and identified as a substantial risk factor for limited vaccine efficacy. The development of potential alternatives for carrier proteins is a substantial research field (17–19).

Zwitterionic polysaccharides (ZPSs) are the only class of carbohydrates that can induce a T cell–dependent immune response, a mode of action normally restricted to proteins and peptides (20–23). Upon uptake of ZPS into antigen-presenting cells, they can be partially depolymerized to ZPS fragments that can be loaded onto major histocompatibility class II (MHC-II) receptors (HLA-DR). The antigen MHC-II is presented to the cell surface and can bind to CD4⁺ T cells through their αβ T cell receptors, generating an adaptive immune response. The activated CD4⁺ T cells can release cytokines that can activate cytotoxic T lymphocyte cells and B cells to produce high-affinity IgG antibodies and memory B cells that are needed for a long-lasting immune response (Fig. 1A) (23–31). Furthermore, ZPSs have been used as carrier in the development of anticancer vaccines, and ZPSs carrying tumor-associated carbohydrate antigens, such as a Tn-PS A1 conjugate, an sialyl-Tn (STn)-PS A1 conjugate, and a Thomsen-Friedenreich (TF)–PS B conjugate have been generated and these were shown to be capable of eliciting a strong immune response (32–36). Thus, ZPSs may present potential alternatives for carrier proteins in the development of next-generation glycoconjugate vaccines.

Fig. 1. Immunological activity and representative structures of ZPS.

(A) ZPS elicit T cell–dependent immune response. (B) Representative structures of ZPS. (C) This work: Total synthesis of PS A2 oligosaccharide repeating units.

Owing to their unique immune activity and potential applications, many ZPS oligosaccharides have been synthesized over the years (Fig. 1B). In 2007, the first synthetic study toward the PSA1 tetrasaccharide repeating unit was reported by van der Marel and co-workers, but the final deprotection was not achieved because of the low reactivity of C4-OH of galactose fragment and could not obtain enough fully protected tetrasacchride (37). Four years later, Seeberger and co-workers reported the first synthesis of the PS A1 tetrasaccharide repeating unit after switching the glycosylation strategy (38); subsequently, the groups of Andreana and Kulkarni also disclosed the synthesis of PS A1 tetrasaccharides in 2018 and 2021 (39, 40); Codée and co-workers demonstrated the first total synthesis of a PS A1 dodecasaccharide, representing three repeating units in 2023 (41). Bundle and co-workers were the first to complete the assembly of Sp1 oligosaccharides, comprising one or two repeating units in 2010 (42, 43). One year later, Codée and co-workers accomplished the assembly of the three possible frame-shifted trisaccharide Sp1-repeating units (44). Seeberger and co-workers reported the synthesis of conjugation-ready Sp1 trisaccharide in 2014 (45), and the corresponding conjugate vaccine candidates were obtained in 2018 (46). Kulkarni and co-workers described the synthesis of an Sp1 trisaccharide using a late-stage oxidation in 2021 (47). In 2019, Codée and co-workers reported on a set of synthetic Sp1 fragments, ranging from the trisaccharide to the dodecasaccharide. The three-dimensional (3D) structure and antibody binding capacity were also reported in (48). In 2022, Codée and co-workers reported the first total synthesis of Sp1 oligosaccharides, carrying the labile O-acetyl esters (49). Synthetic reports on other ZPS oligosaccharides of other bacteria, such as Bacteroides fragilis PS B (50), the lipoteichoic acid of S. pneumoniae (PnC) (51), Morganella morganii (52), Shigella sonnei (53), and Photorhabdus temperata ZPS (54, 55) have also been disclosed. Despite the substantial advancements in ZPSs synthesis, there is no synthetic report for the synthesis of PS A2 owing to its complicated structure and synthetic challenges. Herein, we describe a total synthesis of PS A2 oligosaccharides encompassing one or two repeating units. The resulting oligosaccharides with appropriate amino spacers at the reducing ends are ready for microarray or surface plasmon resonance to enable biophysical interaction studies with MHC-II (Fig. 1C) (56, 57).

RESULTS

The repeating unit of PS A2 is a branched pentasaccharide with a [→3)-α-d-FucN4NAc(1 → 2)-α-d-glycero-d-mannoHep(1 → 3)-α-d-ManNAc(1 → 4)-β-d-Qui3NAc(1→] backbone, carrying an α-l- fucose at the C2-OH of the Qui3NAc residue. The C6-OH of mannoheptose linked with a chiral 3-hydroxybutanoic acid through an ether linkage (58). The unique structure of PS A2 presents many synthetic challenges: First, the repeating unit features the two rare sugars FucN4NAc and Qui3NAc, which are difficult to obtain (59). Second, the higher carbon sugar mannoheptose is substituted with a 3-hydroxy butanoic acid via an ether linkage (60–63). Third, the pentasaccharide repeating unit presents two demanding 1,2-cis-glycosidic linkages (64–71). Furthermore, the structure is highly functionalized, and it includes free amine, carboxylate and acetamido groups. To achieve an efficient synthesis, a convergent glycosylation strategy and orthogonal protecting groups were designed (Fig. 2). The targeted PS A2 fragment 1, comprising two repeating units, could be obtained from the fully protected decasaccharide 3 via desilylation and hydrogenation steps. We planned to synthesize decasaccharide 3 through a [8 + (2 × 1)] glycosylation of octasaccharide diol 4 and l-fucose monosaccharide donor 5, using the intrinsic nature of fucose to control the 1,2-cis-glycosidic bonds. The fully protected octasaccharide 6 could be constructed through a convergent [4 + 4] glycosylation using tetrasaccharide acceptor 7 and donor 9.

Fig. 2. The retrosynthesis analysis for PS A2 oligosaccharides 1 and 2.

In addition, the target pentasaccharide 2 can be obtained from tetrasaccharide 8 via delevulinoylation, [4 + 1] glycosylation and deprotection. The glycosylation of mannoheptosyl donor has shown its capability of good α selectivity without using any of neighboring group participation (61, 62). Therefore, we planned to generate the tetrasaccharide building blocks 8 and 9 using disaccharide donor 10, in combination with disaccharide acceptors 11 and 12. The required disaccharides 10 to 12 can be generated from monosaccharide building blocks 13 to 17. The acyl group on C4 position and bulky triisopropylsilyl (TIPS) group on C3 position of donor 13 are used to control 1,2-cis selectivity of glycosylation, while groups capable of neighboring group participation are placed on the C2 position of the donors to achieve the required 1,2-trans selectivity (72, 73).

Our synthesis began with the preparation of monosaccharide building blocks 13 and 15 to 17 (Fig. 3). The rare sugar FucN4NAc building block 13 was assembled using a slight modification of Kulkarni’s method, starting from 3-O-triisopropylsilyl protected β-d-rhamnose building block 18 (Fig. 3A) (41, 48, 59). Triflation of both the C2 and C4 hydroxy groups in 18, subsequent inversion of the C2-triflate with tetrabutylammonium azide (TBAN₃), and substitution of the less reactive C4-OTf with ammonia, delivered the C4-NH₂ that was protected with a trichloroacetyl group (TCA), to give the core FucNNAc sugar structure 19 in 44% yield over four steps. The thioglycoside 19 was then smoothly transformed into the corresponding N-phenyl trifluoroacetimidate donor 13 in 62% yield over two steps (74). For the synthesis of the d-mannosamine donor, we started from the known d-glucose 2,3-diol 20 (Fig. 3B) (75). Regioselective triflation of C2-OH was followed by azide substitution using TMSN₃/KF to afford mannosazide 21 in 74% yield. The remaining hydroxyl group was masked as a 2-naphthylmethyl (NAP) ether, and subsequent removal of the benzylidene acetal and benzylation gave compound 22 in 94% yield. Reduction of the azides and protection of the resulting free amine were protected with a trichloroacetyl (TCA) group that generated compound 23. The anomeric methyl group in 23 could be transformed into an acetyl group by use of Ac₂O and H₂SO₄ (75). Under the strongly acidic conditions, the C6-benzyl ether was concommitantly transformed into an acetyl ester. Last, the anomeric acetyl in 24 was selectively removed by hydrazine hydrate, and the liberated lactol was reacted with the imidoyl chloride to furnish monosaccharide donor 15′ in 68% yield over two steps.

Fig. 3. Synthesis of monosaccharide building blocks 13, 15′, 16, and 17.

(A) Synthesis of 13; (B) Synthesis of 15’; (C) Synthesis of 16–17.

The rare sugar Qui3NAc building block 16 bearing an anomeric linker was synthesized from commercially available 1,2:5,6-di-O-isopropylidene-α-d-allofuranose 25 (Fig. 3C). The C3 azido group was introduced smoothly using a previous method. Removal of the isopropylidene groups and acetylation then gave compound 26 in 48% over four steps. Introduction of a temporary 4-methoxyl phenyl group (MP) on the anomeric position afforded 3-azido-α-d-glucopyranose and followed by deacetylation gave 27 in 64%. Next, protection of the C4,6-diol with a benzylidene acetal, installation of the C-2 levulinoyl ester, and removal of the benzylidene group gave building block 28 in 68% over three steps. Regioselective tosylation of the primary alcohol furnished 29 in 76% yield. The remaining secondary hydroxyl group was masked as a tert-butyldimethylsilyl (TBS) ether using TBSOTf and 2,6-lutidine to obtain compound 30 in 74% yield. The tosyl group was then replaced for an iodide, after which the iodide and azide were reduced using tri-n-butyltin hydride/azodiisobutyronitrile at 85°C (76). Protection of the so generated free amine with trichloroacetyl chloride delivered 31 in 45% yield over three steps. Oxidative cleavage of the anomeric MP group liberated the hemiacetal, which could be transformed into N-phenyl trifluoroacetimidate donor 32 in 64% yield. The glycosylation of donor 32 and the 5-azidopentanol linker catalyzed by TBSOTf and subsequent removal of the TBS group with HF/pyridine afforded the desired Qui3NAc building block 16 in 66% yield. In addition, monosaccharide acceptor 17 was obtained from 31 in 86% yield via standard desilylation chemistry.

Next, we set out to synthesize the mannoheptose building block 14 as shown in Fig. 4. The properly protected mannoheptose 37 was obtained from d-mannose in 11 steps, and the newly formed chiral central at C6 was confirmed by x-ray diffraction (Fig. 4A) (61). The construction of the ether linkage between 3-hydroxy butanoic acid and the C6 alcohol of the mannoheptose proved to be challenging, and several approaches were tried. First, the canonical Williamson ether formation was investigated (Fig. 4B). Many reaction conditions for the connection of mannoheptose 37 and electrophilic reactants 38a-d were explored, but the desired product could not be obtained. Next, the base-mediated Oxa-Michael addition was probed, which gave the desired product 42 as an R/S mixture in poor yield (Fig. 4C) (77, 78). We then turned to a Pd(II)-catalyzed C(sp3)-H activation alkoxylation, which has been shown to afford ether linkages through a Pd(IV) intermediate (79, 80). Compounds 43a-b were used in combination with different bidentate ligands, but the desired products could hardly be detected. We also tried to use a two-step method, involving coupling of 43c-d in the first Pd(II)-catalyzed C(sp3)-H activation alkoxylation step, and subsequent methylation, but this approach failed as well (81, 82). Therefore, an indirect synthetic method was investigated using a three-step reaction sequence involving esterification, enol ether formation, and hydrogenation (Fig. 4E) (83, 84). The esterification of the C6-OH in 37 proceeded smoothly to give 46 in 94% yield (85), and the so-formed ester was successfully converted into the enol ether 47 using dicyclopentadienyldimethyl titanium (Petasis reagent) in 85% yield (84). Hydrogenation using Pd(OH)₂/C with triethyl amine as additive then gave the desired ether as an inseparable mixture of diastereomers. Removal of the primary silyl group afforded the diastereomeric primary alcohols, which proved to be separable (48R:48S = 3:1). Oxidation of the primary hydroxyl group in 48R to the corresponding carboxylic acid using 2,2,6,6-tetramethyl-1-piperidinyloxyl/[bis(acetoxy)iodo]benzene (TEMPO/BAIB), and subsequent treatment with BnBr and KHCO₃ afforded benzyl ester 42R in 81% yield. Removal of the isopropylidene group using 80% AcOH and following tin-mediated regioselective benzylation of the C3-OH using Bu₂SnO and BnBr afforded the desired mannoheptose building block 14 in 73% yield over three steps. The newly generated chiral center in the 3-hydroxybutanoic acid residue in compound 48 was confirmed by the formation of products 51a and 51b, which showed characteristic nuclear Overhauser effect (NOE) signals in nuclear magnetic resonance (NMR) (Fig. 4F).

Fig. 4. Synthesis of mannoheptose building block 14.

(A) Synthesis of heptose intermediate 37; (B) Intermolecule nucleophilic substitution; (C) Oxa-Michael addition; (D) Pd(II)-catalyzed C(sp3)-H activation alkoxylation; (E) Successful synthetic route for installation side chain ether; (F) Determination of the configuration of C3’ on compound 39.

With all the monosaccharide building blocks in hand, the assembly of PS A2 pentasaccharide repeating unit 2 was undertaken using a convergent [2 + 2 + 1] glycosylation synthetic strategy as shown in Fig. 5. The glycosylation between the diaminofucose donor 13 and mannoheptose acceptor 14 using TBS triflate (TBSOTf) as a promotor furnished disaccharide 52 as single anomer in 83% yield. The stereoselectivity was benefited from the acyl group on C4 position and the bulky TIPS group on C3 hydroxyl of the donor 13. Oxidative cleavage of the anomeric MP group in 52 gave the hemiacetal that could be transformed into the corresponding N-phenyl trifluoroacetimidate donor 10 in 66% yield. In parallel, mannosamine donor 15′ was glycosylated with 3-aminoquinovose acceptor 16 to generate the desired disaccharide 53 in 74% yield. Removal of the Nap group in 53 using 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) gave the disaccharide acceptor 11′ in 75% yield, which was coupled with disaccharide donor 10 to furnish tetrasaccharide 8′ in 65% yield with excellent stereoselectivity. Selectively removal of the levulinoyl group (Lev) in 8′ using hydrazine hydrate delivered the tetrasaccharide acceptor 54 in 82% yield. Last, the [4 + 1] glycosylation of this tetrasaccharide and fucose donor 5 using TBSOTf as a promotor gave fully protected pentasaccharide 55 in 84% yield as single anomer. Deprotection of the pentasaccharide was achieved by the removal of the silyl group by HF/pyridine, after which the Pd(OH)₂/C catalyzed hydrogenolysis of all benzyl ethers, the benzyl ester, and the concomitant transformation of the TCA into the corresponding acetamide and final hydrolysis of the acetate ester, to afford the target PS A2 pentasaccharide repeating unit 2 in 70% yield, over three steps.

Fig. 5. Synthesis of target PS A2 repeating unit pentasaccharide 2.

Encouraged by the successful synthesis of the PS A2 pentasaccharide, we undertook the synthesis of the more challenging decasaccharide 1 containing two repeating units (Fig. 6). We started with the synthesis of tetrasaccharide donor 9′ (Fig. 6A). The glycosylation between mannosamine donor 15′ and 3-aminoquinovose acceptor 17 in the presence of TBSOTf furnished disaccharide 56 as single anomer in 90% yield. Oxidative cleavage of the Nap group using DDQ gave disaccharide acceptor 12′ in 78% yield, which was coupled with disaccharide donor 10 catalyzed by TBSOTf to furnish tetrasaccharide 57 as single anomer in 69% yield. Oxidative cleavage of the anomeric MP group in 57 gave the hemiacetal that could be transformed into the corresponding N-phenyl trifluoroacetimidate donor 9′ in 74% yield. The tetrasaccharide donor 9′ was glycosylated with tetrasaccharide acceptor 7′ obtained from tetrasaccharide 8′ to generate the desired octasaccharide 6′ in 65% yield. The octasaccharide diol acceptor 4′ was obtained by selective removal of the Lev groups in 6′ using hydrazine hydrate in 89% yield. Double fucosylation of 4′ adopting a [8 + (2 × 1)] glycosylation gave fully protected decasaccharide 3′ in 67% yield with pure selectivity. Last, deprotection through desilylation, Pd(OH)₂/C catalyzed hydrogenation, and subsequent basic acetate hydrolysis, afforded the target PS A2 decasaccharide 1 encompassing two repeating units in 62% yield, over three steps.

Fig. 6. Synthesis of target decasaccharide 1 containing two repeating units of PS A2.

(A) Synthesis of tetrasaccharide donor 9’; (B) Synthesis of decasaccharide 1.

The structure of synthetic pentasaccharide 2 and decasaccharide 1 were next investigated using NMR data analysis. The ¹J_C1,H1 coupling constants of synthetic oligosaccharides 1 and 2 are consistent with those established for native PS A2: pentasaccharide 2 sugar A (172 Hz), sugar B (170 Hz), sugar C (174 Hz), sugar D (172 Hz), and sugar E (162 Hz); decasaccharide 1 sugar A/A′ (173 Hz), sugar B/B′ (170 Hz), sugar C/C′ (172 Hz), sugar D/D′ (173 Hz), and sugar E/E′ (162 Hz). The NMR data of decasaccharide 1 are closer to the reported data for the native PS A2 than that of pentasaccharide 2 (fig. S1). For example, the ¹³C chemical shifts difference (¹³CΔδ) for C1 of sugar E of pentasaccharide 2 is 2.1 parts per million (ppm), while the ¹³C chemical shift difference for C1 of sugar E′1 of decasaccharide 1 is close to 0. However, both the ¹³C chemical shift for C1 of sugar A (FucN4NAc) and the ¹H/¹³C chemical shift for H1/C1 of sugar C (mannoHep) do not match well with the reported shifts for the native PS A2. The difference in chemical shift for the reducing end 3-aminoquinovose residue may be accounted for by the fact that the synthetic fragments carry a linker. The other chemical shift discrepancies may be explained by the influence of the oligosaccharide length and frame sequence. The synthetic pentasaccharide 2 and decasaccharide 1 with this frame sequence are not long enough to form a stable 3D structure and take up a conformation that resembles the structure of the natural polysaccharide. The stereochemistry of the 3-hydroxybutanoic acid may also have influence. As we have also generated the 3-(S)-hydroxybutanoic acid functionalized mannoheptose building block 48S, our synthetic approach will allow for the assembly of the oligomers with the alternative stereochemistry of the hydroxybutanoic acid appendage.

DISCUSSION

We have described a total synthesis of zwitterionic PS A2 oligosaccharides, comprising one and two repeating units. To enhance the synthetic efficiency of decasaccharide 1, the octasaccharide backbone was assembled through a convergent [4 + 4] glycosylation, while the branching fucosyl residues were introduced in a double fucosylation event. The tetrasaccharide intermediates were obtained via a convergent [2 + 2] glycosylation. Stereoselectivity in the 1,2-cis-glycosylation reactions of the D-FucN4NAc donor was achieved by placing the acyl group on C4 position and a bulky TIPS group on the C3-hydroxyl of the donor. The formation of the ether linkage between the 3-hydroxybutanoic acid residue and the C-6 hydroxy of the mannoseheptose proved very challenging. Eventually, a three-step reaction sequence, involving ester formation, enol ether formation, and hydrogenation, proved effective to introduce the hydroxy butanoic acid side chain on the mannoheptose. The generated oligosaccharides carry amino spacers at their reducing end for the construction of microarrays or attachment to surface plasmon resonance chips to enable biophysical interaction studies. The synthesis of longer PS A2 oligosaccharides and their analogs, their 3D structural analysis, and immunological activity studies will be reported in due course.

MATERIALS AND METHODS

All reagents were of commercial grade and used as received. All moisture sensitive reactions were performed under an argon/nitrogen atmosphere. The solvent (dichloromethane, toluene, ethyl ether, and CH₃CN) used in the glycosylation reactions was distilled over CaH₂ and stored on activated 4-Å molecular sieves before being used. Reactions were monitored by thin-layer chromatography analysis with detection by ultraviolet (254 nm) and where applicable by spraying with 20% sulfuric acid in EtOH or with a solution of (NH₄)₆Mo₇O₂₄·4H₂O (25 g/liter) and (NH₄)₄Ce(SO₄)₄·2H₂O (10 g/liter) in 10% sulfuric acid (aq.) followed by charring at ~150°C. Flash column chromatography was performed on silica gel (300 to 400 mesh). ¹H and ¹³C spectra were recorded on a Bruker AV 400, Bruker AV 600 in CDCl₃, CD₃OD, CD₃COCD₃, or D₂O. Chemical shifts (δ) are given in parts per million relative to tetramethylsilane as internal standard (¹H NMR in CDCl₃) or the residual signal of the deuterated solvent. Coupling constants (J) are given in hertz. All ¹³C spectra are proton decoupled. Where applicable, correlation spectroscopy, heteronuclear single-quantum coherence, NOE spectroscopy, heteronuclear multiple-bond correlation, and distortionless enhancement by polarization transfer (DEPT)135 experiments were used to further elucidate the structure. The anomeric product ratios were analyzed through integration of ¹H NMR signals or seperation. For the details of synthetic process, please see the Supplementary Materials.

Supplementary Materials

This PDF file includes:

Electronic Supplementary Information (ESI) available: details of experimental procedure and NMR spectra

Fig. S1

Tables S1 and S2