Total Synthesis of the Tetrasaccharide Haptens of Vibrio vulnificus MO6-24 and BO62316 and Immunological Evaluation of Their Protein Conjugates.

Vibrio vulnificus is a human pathogen that can cause fatal septicemia and necrotizing infections with a high lethal rate exceeding 50%. V. vulnificus MO6-24 and BO62316 are two predominant virulent strains associated with approximately one-third of the clinical infections. The capsular polysaccharides (CPSs) of V. vulnificus consist of several structurally unique sugars and are excellent targets for developing effective glycoconjugate vaccines. This article describes the first total synthesis of the challenging tetrasaccharide repeating units of V. vulnificus MO6-24 and BO62316 CPSs. A key feature of this synthesis was the assembly of the tetrasaccharide skeleton using a 3,4-branched trisaccharide as the glycosyl donor. A modified TEMPO/BAIB oxidation protocol was developed to directly convert α-d-GalN into α-d-GalAN in not only disaccharides but also tri- and tetrasaccharides. The synthetic haptens were covalently coupled with CRM197 carrier protein via a bifunctional linker. Preliminary immunological studies of the resultant glycoconjugates in mice revealed their high efficacy to induce robust T-cell-dependent immune responses, and the IgG antibodies elicited by each glycoconjugate showed weak cross-reactivity with the other synthetic tetrasaccharide.

Introduction

Vibrio vulnificus is a halophilic Gram-negative bacterium, ubiquitously present in marine environments worldwide. It is a very dangerous human pathogen, especially for patients with chronic diseases and suppressed immunity, causing severe to life-threatening septicemia and necrotizing wound infections.[1] Typically, the infection begins 1–3 days after exposure to the bacterium in seawater and shellfish and progresses rapidly, and the fatality rate for primary sepsis is >50%.[2]V. vulnificus infection is usually treated with antibiotics, whilst infection site management, such as surgical cleaning of damaged tissues, is commonly required. Sometimes, amputation of affected limbs is necessary.[3] Therefore, effective prophylactic vaccines for V. vulnificus are highly sought after.

V. vulnificus is classified into three biotypes, 1–3, based on its biochemical, serological, and genetic properties and host range. Moreover, about 40 clinically virulent V. vulnificus strains have been identified so far,[4] while most human infections are associated with the biotype 1 strains. Like many other pathogenic bacteria, V. vulnificus expresses capsular polysaccharides (CPSs) as the major virulent factors, which can protect the bacterium by conferring resistance to phagocytosis and other bactericidal activities of the host immune system.[1b,5]V. vulnificus CPSs contain a number of rare sugars that are not present in human cell surface glycans, making them excellent haptens for developing novel carbohydrate-based vaccines.[6]

Although many glycoconjugate vaccines, which were prepared via the conjugation of bacterial CPSs with immunogenic carrier proteins, have been licensed and used in the clinic, the conjugate vaccines consisting of synthetic oligosaccharide haptens of CPSs have received increasing attention.[7] These structurally more well-defined glycoconjugates are useful as chemical biology tools to investigate the accurate structure–activity relationships of bacterial CPSs as immunogens for vaccine design and optimization.[8] As the first semisynthetic oligosaccharide-protein conjugate vaccine, Quimi-hib was developed in Cuba to fight Haemophilus influenzae type b (Hib) and its prophylactic efficacy has been verified in several countries.[9]

V. vulnificus biotype 1 strains MO6-24 and BO62316 are the predominant virulent strains that account for approximately one-third of all clinical infections.[6a] Their acidic CPSs were characterized by Morris group[10] nearly three decades ago. Both CPSs are composed of a tetrasaccharide repeating unit (Figure ). The MO6-24 CPS has a repeating trisaccharide backbone of 3-α-l-QuipNAc-(1 → 3)-α-d-GalApNAc-(1 → 3)-α-l-QuipNAc-(1 → , with l-QuipNAc branches α-linked to the GalANAc 4-O-positions.[6c,10a,10b] The repeating unit of BO62316 CPS contains a different trisaccharide main chain, 3-α-d-FucpNAc-(1 → 3)-α-d-GalApNAc-(1 → 3)-α-l-QuipNAc-(1 → , with α-l-RhapNAc as the side chain also attached to the GalANAc 4-O-position.[10c] Morris and co-workers have demonstrated the potential of these two CPSs as useful antigens for the development of anti-V. vulnificus MO6-24 and BO62316 vaccines.[6] It is also worth noting that the structure of V. vulnificus BO62316 CPS is the same as that of the O-antigen of Plesiomonas shigelloides O24:H8 lipopolysaccharide (LPS)[11] and the trisaccharide structure (labeled in blue color) of V. vulnificus MO6-24 CPS also appears in the O-antigen of LPSs from Escherichia coli O:98 and Yersinia enterocolitica O:11,24.[12] In addition, the 3,4-branched α-d-GalANAc motif is also present in many other bacterial glycans, such as that of Acinetobacter baumannii O5,[13]Vibrio cholerae O9,[14] and V. vulnificus 6353[10b] (Figure ).

Figure 1

Structures of V. vulnificus MO6-24 and BO62316 CPSs and some other representative bacterial glycans containing the same 3,4-branched α-d-GalANAc motif.

Currently, there has been no reported synthesis of the repeating units of V. vulnificus MO6-24 and BO62316 CPSs, which represent a synthetic challenge. First, all of their monosaccharides are rare and commercially unavailable sugars. Second, they have a number of 1,2-cis glycosidic linkages that are difficult to construct in a sterically controlled manner. Third, construction of the 3,4-branched GalANAc motif is usually problematic because of the generally low reactivity of uronic acids as glycosyl donors or acceptors, especially when the GalANAc 4-O-position is involved. We reported herein the first total synthesis of the repeating units of these CPSs with optimization of the involved chemistry, including the preparation of related rare sugars in the context of oligosaccharide synthesis. Our synthetic targets were 1 and 2 (Figure ), the MO6-24 and BO62316 CPS tetrasaccharide repeating units, respectively, with a free amino group-functionalized linker attached to the glycan reducing end. In addition, both 1 and 2 were covalently coupled with CRM197 carrier protein and bovine serum albumin (BSA) to form glycoconjugates 1a and 2a that were subjected to immunological evaluation and conjugates 1b and 2b that were employed as capture antigens to detect carbohydrate antigen-specific antibodies by enzyme-linked immunosorbent assay (ELISA).

Figure 2

Structures of the synthetic targets 1 and 2, as well as the protein conjugates 1a,b and 2a,b of 1 and 2.

Results and Discussion

Retrosynthetic Analysis of Target Molecule 1

In principle, 1 and 2 may be achieved by a similar synthetic strategy, as they share the same 3,4-branched GalANAc motif. Therefore, herein we only discuss about the retrosynthesis of 1 (Scheme ). On the basis of its uniquely branched structure, we focused on disconnecting the glycosidic linkages around the GalANAc unit. Furthermore, since uronic acids typically have low reactivity and afford poor stereoselectivity during glycosylations,[15] we planned to replace GalANAc with GalN3 during the synthesis and later on convert it into GalANAc via postglycosylation oxidation. This results in tetrasaccharide 3 as the protected precursor of 1. Considering the reported difficulty to glycosylate the Gal or GalN3 3-OH group after their 4-O-glycosylation,[16] there would be only two options left for bond disconnection around GalN3. One option leads to monosaccharide 4 and a linear trisaccharide 5 (Scheme , route A), meaning late-stage installation of the sugar residue to the GalN3 4-O-position. The other option leads to a branched trisaccharide 10 (Scheme , route B), meaning preassembly of the challenging 3,4-branched GalN3 motif. In either way, the 6-OH group of GalN3 needs to be orthogonally protected, e.g., 6′-O-TBS in 5 and 6-O-Bz in 10, to enable its selective deprotection and oxidation to create the GalANAc unit later on. In turn, 5 and 10 can be prepared from monosaccharides 4, 6, 8, 9, and 11, all of which have the nonparticipating azido group at their C2-position serving as a protected amino group to promote 1,2-cis glycosylation. In addition, the azido group should be stable to various reactions, including oxidation, involved in carbohydrate synthesis, whereas it can be readily transformed into the desired acetamido group via selective reduction and N-acetylation. Protections for other positions had to be carefully designed as well to facilitate additional modifications and global deprotection. For example, the amino group in the linker at the glycan reducing end was protected with the benzyloxycarbonyl group (Cbz) to differentiate it from other amino groups.

Scheme 1

Retrosynthesis of Compound 1

Synthesis of the Monosaccharide Building Blocks

Both 1 and 2 are composed of rare sugars only, including d-FucNAc, d-GalANAc, l-QuiNAc, and l-RhaNAc. Thus, an essential task for this research is to tackle the large-scale preparation of these monosaccharides and their derivatives as starting materials and key building blocks. The 2-azido-derivatives of d-GalNAc,[17]d-FucNAc,[18] and l-RhaNAc,[19] including their thioglycoside and phosphate donors, were prepared by reported methods. Several 2-azido-derivatives 4a–c and 6a–c of l-QuiNAc as glycosyl donors were synthesized according to Scheme and used to probe the best methods and conditions for α-selective glycosylation. Dimethyltin dichloride (Me2SnCl2)-mediated regioselective benzoylation[20] of the C3-OH in β-l-thiorhamnoside 12(21) afforded alcohol 13 (H-3 shifted to δ 5.17 ppm) in a 79% yield, which was then converted into the corresponding triflate using triflic anhydride (Tf2O) and pyridine. Substitution of the triflate with an azido group provided orthogonally protected 2-azido-l-QuiN 6a in a 62% yield over two steps. The β-anomeric conformation of 13 is critical for the SN2 reaction to occur, as its α-anomeric counterpart would have unfavorable repulsion between the anomeric thiomethylphenyl group and the incoming nucleophile, known as pyranosidic vicinal axial effect,[22] to inhibit the SN2 reaction and result in thioglycoside decomposition.[19,21,23] Reaction of 6a with dibutyl phosphate under the promotion of N-iodosuccinimide (NIS) and triflic acid (TfOH) gave 6b (88%). Trichloroacetimidate donor 6c was easily derived from 6a upon hydrolyzing the thioglycoside under promotion of N-bromosuccinimide (NBS) in acetone–water and reaction of the resultant hemiacetal with trichloroacetonitrile (Cl3CCN) and 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU). On the other hand, substituting the benzoyl (Bz) group in 6a with a benzyl group (Bn) through conventional transformations gave 3,4-dibenzylated thioglycoside 4a, which was easily converted into 4b,c by the methods described for 6b,c.

Scheme 2

Synthesis of 2-Azido-Derivatives 4a–c and 6a–c of l-QuiNAc as Glycosyl Donors

The linker-equipped monosaccharide 15 as the reducing end unit of 1 and 2 was achieved by glycosylation of 14 with 6. As there are several α-l-QuiNAc residues in 1, stereoselective construction of α-glycosyl linkage of l-QuiN3 is critical, but its reaction has not been examined in detail previously.[24] Therefore, we performed detailed analysis of the glycosylation reactions of 14 using different glycosyl donors 6a–c and reaction conditions. The results are listed in Table . It was revealed that solvents had a great impact on the stereochemical outcomes of the reaction between thioglycoside 6a and 14 with NIS and TfOH as the promoter (Table , entries 1–5), and high yield and α-selectivity (α:β = 8:1, Table , entry 2) were obtained in a mixture of CH2Cl2 and Et2O (1:1, v/v). Using the same solvent system, the reaction under the promotion of p-toluenesulfenyl chloride (TolSCl) and silver triflate (AgOTf)[25] gave even better α-selectivity and yield (α:β = 15:1, 88%, Table , entry 7) at −40 °C, whereas the same reaction at −78 °C afforded a much lower α-selectivity (α:β = 1:1, Table , entry 6), which may be explained by an SN2 mechanism under the conditions. The reaction between 14 and phosphate donor 6b in the presence of trimethylsilyl triflate (TMSOTf) was only moderately α-selective (α:β = 5:1, Table , entry 8) although the yield was good (85%). Finally, the glycosylation reaction of 14 with trichloroimidate donor 6c in the presence of TMSOTf and an α-directing additive thiophene[26] also gave excellent yield and α-selectivity (83%, α:β = 12:1, Table , entry 9). The α-configuration of 15 was verified by its H-1 NMR signal (δ 4.84 ppm, s), the coupling constant of which was too small to measure at the given spectrometer frequency.

Table 1

Optimization of the Conditions for Stereoselective α-Glycosidation of l-QuiN3 Using 14 as a Model Acceptor

entry	donor	solvent	activation method	yielda	α:βb
1	6a	CH2Cl2	NIS, TfOH 0 °C	87	3:1
2	6a	CH2Cl2:Et2O (1:1)	NIS, TfOH 0 °C	85	8:1
3	6a	Tol	NIS, TfOH 0 °C	84	3:1
4	6a	Tol:dioxane (1:1)	NIS, TfOH 0 °C	77	5:1
5	6a	THF	NIS, TfOH 0 °C	78	2:3
6	6a	CH2Cl2:Et2O (1:1)	TolSCl AgOTf −78 °C to rt	87	1:1
7	6a	CH2Cl2:Et2O (1:1)	TolSCl, AgOTf −40 °C to rt	88	15:1
8	6b	CH2Cl2:Et2O (1:1)	TMSOTf 0 °C	85	5:1
9	6c	CH2Cl2	TMSOTf thiophene 0 °C	83	12:1

Isolated yields (%).

α:β anomeric ratios were determined by 1H NMR analysis of reaction mixtures.

Attempts to Synthesize 1 via Route A Outlined in Scheme

Each of the synthetic routes outlined in Scheme should have its own advantages and disadvantages. Thus, we have attempted both. For the synthesis of 1 by route A, as shown in Scheme , we removed the Bz group in 15 with CH3ONa to deliver 8 as a glycosyl acceptor in a 91% yield. Glycosylation of 8 with GalN3 thioglycoside 9a under the promotion of NIS/TfOH in CH2Cl2 afforded α-disaccharide 16 in only a 45% yield. However, the reaction of 8 with phosphate 9b and TMSOTf at 0 °C produced only a trace amount of 16. A plausible explanation for the results was that the low reactivity of the 3-OH group in 8 made its reaction slow to result in competitive side reactions and decomposition of 9a or 9b at relatively high temperatures. To address this issue, we carried out the reaction of 8 and 9b at a lower temperature (−40 °C), and to our delight, a greatly increased yield (80%) was observed this time, although the stereoselectivity was moderate (α:β = 5:1). Again, when the solvent was switched from CH2Cl2 to a 1:1 mixture of CH2Cl2 and Et2O, the same reaction generated α-linked disaccharide 16 exclusively (GalN, H-1, δ 5.63 ppm, JH1′,2′ = 3.0 Hz) in an 84% yield. Thereafter, the acetyl group in 16 was removed, which was followed by glycosylation with 6a by the same protocol used for the synthesis of 15 (Table , entry 7) to afford trisaccharide 17 (JH1″,2″ = 3.0 Hz) in a 73% yield for two steps. We also utilized imidate donor 6c to react with 7 by the protocol listed in Table , entry 9, but no 17 was formed, whereas 6c was decomposed. This suggested that the 3′-OH group in 7 also had poor reactivity. The benzylidene group in 17 was removed with 10% trifluoroacetic acid (TFA) in CH2Cl2, and the primary 6′-OH group in the resultant diol was regioselectively protected with a TBS group to get 5, hence setting up the stage for constructing the 3,4-branched motif. Donors 4a–c instead of 6a–c were utilized to carry out the following glycosylation, so as to differentiate the two 3″-OHs of the QuiN units that were linked to the 3,4-O-positions of GalN3. To our disappointment, however, glycosylation of 5 using glycosyl donors 4a–c and various promoters including TolSCl/AgOTf, NIS/AgOTf, NIS/TfOH, DMTST, and TMSOTf did not give the desired tetrasaccharide as monitored with mass spectrometry. As we thought that this failure might be due to the strong steric hindrance of a bulky 6′-O-TBS group that can further reduce the reactivity of the 4′-OH group in 5, we used a Bz group to replace the TBS group to obtain trisaccharide 18 as an alternative acceptor. Unfortunately, its reaction with glycosyl donor 4a–c was not effective either (results are presented in the Supporting Information), and in the best case, tetrasaccharide 19 was formed in <20% yield. According to Seeberger and co-workers,[27] a sugar unit at the GalN α-anomeric position could cause steric crowdedness, forcing the group at the 6-O-position to adopt a conformation to shield the inherently unreactive 4-OH of GalN to further hinder its glycosylation.[27,28] Therefore, we decided to abandon route A for the synthesis of 1.

Scheme 3

Failed Attempts to Synthesize 1 via Route A

Synthesis of 1 via Route B Outlined in Scheme

According to the plan (Scheme ), 11 was glycosylated with 6a under the promotion of TolSCl/AgOTf to generate α-linked 20 in a 76% yield and good stereoselectivity (α:β = 7:1). The stereochemistry of the newly formed α-glycosidic bond was confirmed by the H-1 NMR signal of QuiN3 (δ 5.22 ppm, s). Next, the benzylidene group in 20 was removed under acidic conditions and the exposed 6-OH group was selectively benzoylated to afford 21. Compared to sugar residues, the PMP group at the anomeric position of GalN3 was smaller and designed to have β-configuration so as to avoid potential steric influences on the C6-position. As expected, glycosylation of 21 with 4a under the above-established glycosylation conditions proceeded smoothly to give the target trisaccharide 10 in excellent yield (81% for the pure α-anomer; QuiN3′, H-1, δ 5.47 ppm, s) and with excellent α-selectivity (α:β = 14:1). Subsequently, 10 was transformed into trichloroacetimidate 22 in two steps including removal of the anomeric PMP group with diammonium cerium(IV) nitrate (CAN) and reaction of the resultant hemiacetal with CCl3CN and DBU in a 71% yield for 2 steps. Glycosylation of 8 with trisaccharide 22 in the presence of TMSOTf at −40 °C in CH2Cl2-Et2O was unexpectedly but delightfully smooth to provide α-linked tetrasaccharide 19 exclusively (GalN, H-1, δ 5.72 ppm, s) in an 82% yield. Removal of the two Bz groups in 19 under saponification conditions afforded diol 3 in a 92% yield, which was ready for oxidation to convert GalN3 to GalAN3.

Scheme 4

Assembly of Tetrasaccharide 3 via Route B

TEMPO can selectively oxidize primary hydroxyl groups in the presence of secondary hydroxyl groups and consequently is frequently used to prepare sugar uronic acid.[26c,29] Unfortunately, oxidation of the 6′-OH group in 3 using TEMPO (0.2 equiv.) and BAIB (2.0 equiv.) in CH2Cl2 and H2O[30] turned out to be problematic. MS analysis of the reaction mixture revealed the formation of mono- and disaccharides but not the desired 23. Evidently, at this stage, we needed to establish a robust protocol to transform sugar alcohol to uronic acid.

Selective oxidative of polyols in complex oligosaccharides is challenging, which is the major reason for some failed carbohydrate syntheses.[31] In particular, although many novel oxidation methods are available to transform alcohol into carboxylic acid, they have been rarely used in the synthesis of oligosaccharides containing GalANAc residues.[15,32] Codée group[15] found that the TEMPO/BAIB protocol was successful in the preparation of 2-azido derivatives of α-GalANAc as monosaccharides, but it did not deliver the same uronic acid in the context of a disaccharide. Thus, they developed a two-step TEMPO/BAIB-Pinnick oxidation protocol to address the problem. Besides, a two-step Dess–Martin–Pinnick oxidation protocol was used by the Lay group[32a] to achieve Salmonella typhi Vi oligosaccharides containing α-GalANAc. Although these two-step oxidation strategies might be effective for our synthesis herein, we planned to modify the TEMPO/BAIB oxidation protocol with an aim to probe a more facile one-step oxidation methodology for the preparation of oligosaccharides containing the GalAN residue.

Accordingly, a simple disaccharide 24, which was easily prepared by removal of the benzylidene group in 16 (see the Supporting Information), was employed as a model to optimize the protocol for TEMPO/BAIB oxidation of GalN3 (Scheme ). It was reported that oxidation of primary hydroxyl groups with a catalytic amount of TEMPO and 1.0–1.5 equiv. of BAIB under anhydrous conditions could result in aldehydes, whereas in the presence of H2O, aldehydes were further oxidized to carboxylic acids.[29a,32b,33,34] We examined a similar protocol using model compound 24 and found that its reaction with of 0.2 equiv. TEMPO and 1.5 equiv. BAIB in dry CH2Cl2 afforded the aldehyde intermediate almost quantitatively in 4 h. At this stage, if excessive H2O was added to form a heterogeneous solvent system, e.g., CH2Cl2:H2O = 4:1, the aldehyde was decomposed quickly to give complex results. However, our careful analysis revealed that by adding water-saturated CH2Cl2 and another batch of BAIB (3.0 equiv.) into the reaction mixture containing the aldehyde intermediate, the reaction (0 °C to rt for 24 h) was clean to afford the desired uronic acid in good yields. To facilitate product purification, we transformed uronic acid in situ to less polar benzyl uronate upon benzylation using BnBr. Eventually, 25 was isolated in an 85% yield. Encouraged by this result, we probed the modified oxidation protocol further using a more complex trisaccharide 26 (see the Supporting Information) containing 3 free hydroxyl groups and an α-linked GalN3 as well. The reaction gave excellent results. The application of this protocol to disaccharide 28 (see the Supporting Information) that contained a β-linked GalN3 was also very successful in providing the desired uronate 29 in an 84% yield. However, the modified protocol failed for regioselective oxidation of triols 30 and 31 (see the Supporting Information). A comparison of the structures of 24, 26 (which is also a triol), 30, and 31 as well as their reaction results suggested that this failure was probably due to the presence of a cis-diol (3,4-OHs) motif in 30 and 31. A solution to the problem is to protect the cis-diol motif before oxidation reaction as proved in the synthesis of 2. The Kulkarni and Oscarson groups[31,35] had similar observations with both cis- and trans-diols in oligosaccharides. Nonetheless, through rational design and analysis, a practical TEMPO/BAIB oxidation protocol was established for the synthesis of GalAN in complex oligosaccharides.

Scheme 5

Investigation of TEMPO/BAIB Oxidation Protocol

After the proper TEMPO/BAIB oxidation protocol to convert GalN3 into GalAN3 was established, it was then applied to 3 to continue the synthesis of 1 (Scheme ). Delightfully, the reaction went very smoothly to provide the desired carboxylate 32 in an 80% overall yield over two steps. Thereafter, the four azido groups in 32 were transformed into acetamido groups upon reduction with 1,3-propanedithiol and then regioselective N-acetylation using acetic anhydride in CH3OH. The product 33 was easily purified by flash silica gel column purification. It is worth noting that AcSH, a commonly used reagent for one-step reductive acetylation of the azido group,[36] was initially employed for the direct conversion of 32 to 33; however, only a very low yield (<10%) of 33 was obtained. Finally, all of the Bn and Cbz groups in 33 were concurrently removed via Pd(OH)2-catalyzed hydrogenolysis to give the synthetic target 1 in a 79% isolated yield, after purification on a G15 column using H2O as the eluent. The structure of 1 was rigorously characterized with 1D and 2D NMR and HR MS data.

Scheme 6

Endgame to Finish the Synthesis of 1

Synthesis of 2

On the basis of the success in synthesizing 1 according to route B (Schemes and 6), we designed to synthesize 2 by a similar [3 + 1] strategy (Scheme ). Glycosylation of 11 with tricholoroimidate of 2-azido modified d-FucN 34(18) under the promotion of TMSOTf at −40 °C using the CH2Cl2-Et2O (1:1) solvent system provided α-linked disaccharide 35 (JH1′,2′ = 3.0 Hz) only in an 88% yield. Additional exploration of the reaction condition and related results are presented in the Supporting Information. The exclusive α-selectivity was possibly a result of the combination of several α-directing factors, such as solvent effects and the relatively low nucleophilicity of 3-OH in 11, which resulted from the electron-withdrawing effect of the azido group.[16c,37] Additionally, the remote participation of the axial 4-OAc group could play a role by blocking the β-face, thereby facilitating favorable attack of the approaching nucleophile from the α-face in the preferred 4H3 half-chair conformation.[38] Besides, the 3-OAc group was revealed to more readily participate in glycosyl cation stabilization and thus favor generation of the α-glycosidic linkage (see insertion in Scheme ).[38b,38f] Next, selective debenzylidenation of 35 followed by regioselective benzoylation of the 6-OH group in the resulting diol as described above produced 36 in a 74% overall yield. Thereafter, the 3,4-branched motif around GalN3 was constructed by glycosylation of 36 by utilizing l-RhaN3 donor 37b (see the Supporting Information) under the promotion of TMSOTf to furnish the desired trisaccharide 38 in an 86% yield with exclusive α-selectivity. The stereochemistry of the newly created glycosyl bond of RhaN3 was verified by the finding that no Nuclear Overhauser Effect (NOE) was observed between H-1 and H-3 or between H-1 and H-5, whereas strong NOE was observed between its H-1 and H-2. To the best of our knowledge, this represented the first example for stereoselective construction of an α-l-RhaN glycosyl linkage in oligosaccharides. It is worth noting that the reaction between 36 and tricholoroimidate 37a (see the Supporting Information) under the same conditions also afforded 38 in an α-specific manner, but the reaction yield was moderate (55%), which was probably controlled by the weak nucleophilicity of the GalN3 4-OH group in 36. After 38 was obtained, it was converted into imidate 39 in two steps and then coupled with acceptor 8, following the same procedures used to prepare 19. Again, the glycosylation reaction between 39 and 8 was α-specific to give tetrasaccharide 40 (GalN, H-1, δ 5.62 ppm, s). Removal of the acyl groups in 40 with CH3ONa resulted in triol 41, ready for the conversion of GalN3 into GalAN3.

Scheme 7

Synthesis of Tetrasaccharide 2

On the basis of the lessons learned from the oxidation of 30 and 31 (Scheme ), to play it safe, we chose to protect the 3″,4″-cis-diol in 41 with an isopropylidene group before the oxidation of its 6′-OH group, and this reaction was very easy and high yielding (88%). Our modified TEMPO/BAIB oxidation protocol was then applied to 42 to smoothly convert it into the desired uronate 43 in a high yield (81%). Thereafter, the isopropylidene group was removed under acid conditions, which was followed by the transformation of all four azido groups into acetamido groups, as described for 33, to afford 44. The final global deprotection of 44 by Pd(OH)2-catalyzed hydrogenolysis to remove all of the Bn and Cbz groups was straightforward. After purification on a G15 column and lyophilization, the target tetrasaccharide 2 was obtained in a 58% yield over four steps. Its structure was fully characterized with 1D and 2D NMR and HR MS data.

Conjugation of 1 and 2 with Carrier Proteins

Free oligosaccharides 1 and 2 were conjugated with carrier proteins CRM197 and BSA through the bifunctional glutaryl group.[39] The conjugation was accomplished in two steps (Scheme ). First, tetrasaccharides 1 and 2 were reacted with excessive disuccinimidyl glutarate (15 equiv., DSG, 45), which was a commercially available and dually activated ester, to generate the monoamide-monoester products 46 and 47, respectively. The excess of DSG was removed by washing with ethyl acetate. Subsequently, activated esters 46 and 47 were individually conjugated with CRM197 and BSA proteins (oligosaccharide/protein, 2:1 mass ratio) in 0.1 M phosphate-buffered saline (PBS, pH 7.4) upon reaction with the free amino groups on the carrier protein surface to afford the target glycoproteins 1a, 1b, 2a, and 2b. The oligosaccharide/protein molar ratio of these conjugates were determined by MALDI-TOF MS analysis, and an average of 3.0–3.8 molecules of synthetic tetrasaccharide were loaded onto one molecule of carrier protein (Table ).

Scheme 8

Preparation of Oligosaccharide–Protein Conjugates 1a, b and 2a, b

Table 2

Carbohydrate Loadings of Glycoconjugates 1a, b and 2a, b

	CRM197 conjugates		BSA conjugates
sample	1a	2a	1b	2b
glycan/protein molar ratio	3.0	3.8	3.2	3.7

Immunological evaluation of glycoconjugates 1a and 2a

These studies were carried out with 6 to 8 weeks old female Balb/c mice. Conjugates 1a and 2a were administrated as emulsions with Freund’s complete or incomplete adjuvant (FCA/FIA).[39,40] For the initial immunization, the FCA emulsion of each conjugate (100 μL, containing 3.0 μg of carbohydrate) was subcutaneously (s.c.) injected to a group of six mice on day 1. Thereafter, the FIA emulsion of the same conjugate (the same dose) was injected s.c. to each mice on days 15, 22, and 29, respectively, to boost immunization. Blood samples were collected from each mouse on day 0 before the initial immunization, which was used as the blank, and on day 36 after boosting immunizations. Next, the blood samples were treated by standard protocols[41] to produce sera that were evaluated by ELISA to detect the carbohydrate hapten-specific IgM, IgG1, IgG2a, IgG2b, and IgG3 antibodies using BSA conjugates 1b and 2b as the capture antigens and goat antimouse IgM, IgG1, IgG2a, IgG2b, and IgG3 secondary antibodies. Antibody titers were calculated from linear regression analysis of curves obtained from the detected optical density (OD) value (after subtracting blanks) at a 405 nm wavelength against serum dilution numbers and were defined as the serum dilution numbers yielding an OD value of 0.2.

As depicted in Figure , both glycoproteins 1a and 2a provoked high titers of oligosaccharide-specific IgM, IgG1, IgG2a, IgG2b, and IgG3 antibodies. In addition, the titers of IgG1 and IgG2b antibodies induced by 2a were higher than that of 1a. These results clearly revealed the high efficacy of conjugates 1a and 2a to elicit T-cell-mediated immune responses. Moreover, the elicitation of high titers of IgG antibodies is usually associated with antibody class switch, antibody affinity maturation, as well as long-term immunological memory, etc.,[6a,6b,42] which are desirable properties for effective vaccines.

Figure 3

ELISA results of oligosaccharide-specific IgM, IgG1, IgG2a, IgG2b, and IgG3 antibody titers of day 36 antisera of mice immunized with CRM197 glycoconjugates (A) 1a and (B) 2a. Antibody titer was shown as the mean value of three parallel experiments for each serum. The error bar represented the standard error of mean (SEM).

Because oligosaccharides 1 and 2 shared an identical disaccharide motif, we next investigated whether their antisera could cross-react with the other hapten or not. Accordingly, the oligosaccharide-specific total IgG antibody titers of each antiserum to both haptens were examined by ELISA using plates coated with BSA conjugates 1b and 2b, respectively. The results, as outlined in Figure , revealed that each antiserum could indeed bind both oligosaccharides but in significantly different levels. Notably, the reactivity of the antiserum of 2a with hapten 2 was about 3.5 fold higher than that with 1. These results suggested that antibodies provoked by conjugates 1a and 2a might recognize the shared disaccharide motif to some degree but the main recognition motifs for the antibodies were probably the entire structure of haptens 1 and 2. The above findings were essentially in accordance with the reported results that the V. vulnificus MO6-24 and BO62316 strains showed weak serological cross-reaction and the antisera elicited by the MO6-24 CPS-TT vaccine did not confer protection against lethal challenge with the BO62316 strain.[6a]

Figure 4

Cross-reactivity of the total IgG antibodies in the pooled day 36 antisera of conjugates 1a and 2a with two synthetic haptens detected by ELISA using BSA conjugates 1b and 2b as coating antigens. The mean antibody titers of three parallel experiments are presented for each sample, and the error bar shows the SEM, where *** means an extremely significant difference (P < 0.001).

Conclusion

The first total synthesis of the repeating units of V. vulnificus MO6-24 and BO62316 CPSs was achieved with preassembled GalN3-based 3,4-branched trisaccharides as the key intermediates. These oligosaccharides have several important features, such as composition of all rare amino sugars and uronic acid and the presence of multiple difficult 1,2-cis glycosidic linkages and especially the 3,4-branched motif around the notoriously unreactive GalANAc residue, which have significantly elevated the challenge of their total synthesis. Our initial attempt to assemble the branch architecture based on a linear trisaccharide substrate was unsuccessful possibly because of the poor reactivity and high steric hindrance of the trisaccharide acceptor, a proof of the synthetic challenges. Through careful designing of protection–deprotection tactics and evaluation and comparison of various glycosylation methods/conditions and glycan assembly strategies, we were able to solve all of the problems and synthesize the target molecules 1 and 2 by an efficient [3 + 1] strategy. All of the glycosyl linkages, including difficult 1,2-cis glycosidic linkages of amino sugars, were generated in high yields and excellent to exclusive α-selectivity. Furthermore, a modified TEMPO/BAIB oxidation protocol was developed for efficient conversion of GalN to GalAN within complex oligosaccharides. This would enable the design of highly convergent and efficient synthesis of various glycans containing GalANAc or other sugar uronic acids. The overall synthetic strategy for 1 and 2 can be useful for other similar oligosaccharides composed of rare amino sugars and uronic acids. In addition, some of the involved synthetic intermediates, such as 32, 33, and 44 (Schemes and 7), are ready for further elongation to access more complex oligosaccharides, e.g., dimers of these tetrasaccharides.

Synthetic oligosaccharides 1 and 2 carried a free amino group at the glycan downstream end, facilitating their regioselective coupling with carrier proteins. Immunological studies revealed that the resultant CRM197 glycoconjugates 1a and 2a provoked robust antigen-specific T-cell-dependent total IgG and IgG1 antibody responses. Moreover, it was found that the IgG antibodies elicited by each conjugate exhibited weak mutual recognition or cross-reactivity with the other oligosaccharide hapten, which was essentially in accordance with the reported results. More detailed studies on the glycoconjugates, such as the binding affinity of elicited antisera with bacterial cells and the efficacy of induced immune responses to protect animals against bacterial infections, as well as further optimization of the oligosaccharide haptens and carrier proteins, are currently underway in our laboratory.

Methods

General Procedures of the Modified TEMPO/BAIB Oxidation and Benzylation to Synthesize Compounds 25, 27, 29, 32, and 43

To a stirred solution of each oligosaccharide 24, 26, 28, 3, or 42 (0.20 mmol, 1.0 equiv.) in anhydrous CH2Cl2 (4.0 mL, 0.05 M) were added TEMPO (6.2 mg, 0.04 mmol, 0.2 equiv.) and BAIB (96.6 mg, 0.30 mmol, 1.5 equiv.) at 0 °C. The reaction mixture was warmed up to room temperature and afforded the aldehyde intermediate almost quantitatively in 4 h, at which time TLC analysis indicated complete conversion of the starting material (the polarity of aldehyde intermediate is similar to the starting alcohol). Subsequently, the water (40 μL, 1% in CH2Cl2) and BIAB (193.2 mg, 0.60 mmol, 3.0 equiv.) were added at 0 °C, and the mixture was stirred vigorously at room temperature for about 20 h. TLC analysis (CH2Cl2/MeOH, 20:1 v/v) indicated complete conversion of the aldehyde intermediate. The reaction mixture was quenched with dropwise addition of saturated Na2S2O3 solution and diluted with CH2Cl2. The organic layer was washed with saturated NaHCO3 solution and brine, dried over Na2SO4, and concentrated under a vacuum to give a residue.

The crude residue was dissolved in DMF (2.0 mL) and mixed with BnBr (35.4 μL, 0.30 mmol, 1.5 equiv.) and K2CO3 (55.2 mg, 0.40 mmol, 2.0 equiv.) at room temperature. The reaction mixture was stirred under N2 protection for 30 min, quenched with ice cold water, and extracted with ethyl acetate (3 × 20 mL). The combined organic layer was washed with brine (2 × 20 mL), dried over Na2SO4, and concentrated under a vacuum. The residue was purified by silica gel column chromatography with EtOAc and toluene as the eluent to give the desired uronate 25, 27, 29, 32, or 43.

Preparation of Activated Esters 46 and 47

Oligosaccharide 1 or 2 (10.0 mg, 1.0 equiv.) and DSG (58.3 mg, 15.0 equiv.) were added to a mixture of DMF and 0.1 M PBS buffer (v/v 4:1, 0.5 mL). The mixture was stirred at room temperature for 6 h and most of the solvents were removed under vacuum. The activated esters 46 and 47 were separated from the excessive DSG upon precipitation with 9 volumes of ethyl acetate, which was followed by further purification by washing with ethyl acetate 10 times and then drying under a high vacuum for 2 h. Compounds 46 and 47 were then used for preparation of the glycoproteins.

Preparation of Glycoproteins 1a, 1b, 2a, and 2b

The above-obtained activated esters 46 (6.0 mg) and 47 (6.0 mg) were individually mixed with CRM197 (3.0 mg, Absin 03058) or BSA (3.0 mg, Absin 9157) with a 2:1 oligosaccharide/protein mass ratio in 0.1 M PBS buffer (pH 7.4, 0.5 mL). The mixture was gently stirred at room temperature for 3 days and was then applied to a Biogel A 0.5 column with 0.1 M PBS buffer as the eluent for purification. Fractions containing glycoproteins, which were detected by the MALDI-TOF MS analysis, were combined and dialyzed against distilled water for 1 day. Finally, the residual solutions were lyophilized to afford conjugates 1a, 1b, 2a, and 2b, respectively, as white fluffy solids. The average molecular weight of these glycoconjugates were characterized by MALDI-TOF mass spectrometry using CRM197 or BSA as standard. An average of 3.0–3.8 molecules of oligosaccharide were loaded onto one molecule of carrier protein (CRM197 or BSA).

Immunization of Mice

Immunological studies were performed with female Balb/c mice (6–8 weeks of age). Each CRM197 conjugate (1.60 mg of 1a, 1.30 mg of 2a, containing 75 μg oligosaccharide) was dissolved in 1.5 mL of 2 × PBS buffer. They were completely mixed with 1.5 mL of FCA (Sigma F5881) or FIA (Sigma F5506) according to manufacturer’s instructions to generate emulsions. Each group of six mice was initially immunized (day 1) of FCA emulsion (0.1 mL/mouse, containing ca. 3 μg of carbohydrate antigen) by s.c. injection. Following the initial immunization, mice were boosted three times on days 15, 22, and 29 via s.c. injection of FIA emulsions (0.1 mL/mouse). In the meantime, two control groups of mice were immunized with free oligosaccharides 1 and 2 together with FCA/FIA by the same protocol, respectively. Blood samples were collected through the tail vein of each mouse on day 0 before the initial immunization and on day 36 after the boost immunizations. Finally, antisera were obtained from the clotted blood samples according to the standard protocols and stored at −80 °C before immunological analysis.

ELISA Protocol

ELISA plates were treated with a solution of BSA conjugate 1b or 2b (100 μL/well, 2 μg/mL) dissolved in coating buffer (0.1 M aqueous bicarbonate, pH 9.6) at 4 °C for overnight and then incubated at 37 °C for 1 h. The plates were washed with PBST (PBS buffer containing 0.05% Tween-20) three times and then incubated with the blocking buffer (1% BSA in PBST) at room temperature for 1 h followed by washing with PBST three times. Each mouse serum with serial dilutions from 1:300 to 1:218700 in PBS (100 μL/well) was added to the coated plates, which were followed by incubation at 37 °C for 2 h. After being washed with PBST three times, the plates were incubated at room temperature for 1 h with a 1:1000 diluted solution of alkaline phosphatase-linked goat antimouse IgM, IgG, IgG1, IgG2a, IgG2b, or IgG3 (Abcam 98672, 98710, 98690, 98695, 98700, or 98705) antibody. The plates were washed with PBST three times and then incubated with 100 μL of p-nitrophenyl phosphate (PNPP) solution (1.67 mg/mL in PBS buffer) at room temperature for 30 min. Finally, the optical density (OD) values of the ELISA plates were examined using a microplate reader at 405 nm wavelength. After deducting the background optical density values obtained with day 0 sera, the OD values were plotted against serum dilution values and the equation of the best-fit line was obtained. The equation of the line was taken to calculate the dilution value at which an OD value of 0.2 was achieved, and the antibody titer was calculated at the inverse of the dilution value. ELISA experiments to measure the cross-reactivity between groups were performed by the same protocol using each BSA conjugate as the capture antigen to detect antibodies in both antisera.