Stereocontrolled Synthesis of the Equatorial Glycosides of 3-Deoxy-D-manno-oct-2-ulosonic Acid: Role of Side Chain Conformation

Abstract

The pseudosymmetric relationship of the bacterial sialic acid, pseudaminic acid, and 3-Deoxy-D-manno-oct-2-ulosonic acid (KDO) affords the hypothesis that suitably protected KDO donors will adopt the trans,gauche conformation of their side chain and consequently be highly equatorially selective in their coupling reactions conducted at low temperature. This hypothesis is borne out by the synthesis, conformational analysis, and excellent equatorial selectivity seen on coupling of per-O-acetyl or benzyl-protected KDO donors in dichloromethane at −78 °C. Mechanistic understanding of glycosylation reactions is advancing to a stage at which predictions of selectivity can be made. In this instance, predictions of selectivity provide the first highly selective entry into KDO equatorial glycosides such as are found in the capsular polysaccharides of numerous pathogenic bacteria.

Graphical Abstract

Knowledge of the factors influencing the anomeric reactivity and selectivity of glycosyl donors is paramount for the development of efficient and stereocontrolled glycosylation methods.¹ Among the various factors, side chain configuration and conformation have been revealed to play significant roles influencing the reactivity and selectivity of glycosyl donors as first understood in the context of the 4,6-O-benzylidene acetal effect in β-mannopyranosylation and related reactions.^2-4

Moving from rigid bicyclic donors and the benzylidene effect to monocyclic donors, we demonstrated the influence of side chain configuration and conformation on the anomeric reactivity in the neuraminic acid series. Thus, we showed that change in configuration of a single stereogenic center in the side chain of a N-acetyl neuraminic acid donor 1 to the 7-epi-isomer 2 (Figure 1) resulted in a change in predominant conformation of the exocyclic bond from gauche,gauche (g,g) in 1 to gauche,trans (gt) in 2 (Figure 1), with concomitant changes in reactivity and selectivity, under our standard conditions of activation with N-iodosuccinimide and trifluoromethane sulfonic acid in dichloromethane/acetonitrile at −78 °C.⁵ We subsequently showed that inversion of configuration at the 5-position in the pyranoside ring of 1 to give the 5-epi-donor 3 also resulted in a change of the side chain conformation to predominantly gt coupled with a change in reactivity and selectivity.⁶ So-informed, we predicted that a pseudaminic acid donor 4, epimeric at the 5,7,and 8-positions relative to 1 would predominantly populate the least reactive and most equatorially selective trans,gauche (t,g) conformation, as was borne out in practice.⁷

Figure 1.

Diastereomeric neuraminic acid derivatives and their predominant conformation about the exocyclic C6-C7 bond

Recognizing the pseudo-enantiomeric relationship between pseudaminic 5 and 3-deoxy-D-manno-oct-2-ulosonic acid (KDO) 6 (Figure 2), we predicted that KDO donors would have the tg conformation about the exocyclic 6,7-bond and would display excellent equatorial selectivity under our standard conditions. Indeed, the early literature⁸ reveals the ³J_{H6, H7} coupling constant isn the ¹H NMR spectrum of KDO itself to be 9.0 Hz and suggests the tg conformation is predominant. More recently, computational work by Kiessling and co-workers also pointed KDO adopting the tg side chain conformation.⁹

Figure 2.

Pseudo-enantiomeric relationship of pseudaminic acid 5 and KDO 6

The ability to synthesize equatorial glycosides of KDO is important in view of their widespread occurrence in the capsular polysaccharides of numerous pathogenic bacteria.¹⁰ Unfortunately, while much effort has been directed at the synthesis of the more common axial glycosides^11–14 of KDO as found in lipopolysaccharides, little work has been reported on the synthesis of the equatorial glycosides,^15–22 and none with consistently high levels of control. We report here on the reduction of our hypothesis to practice and demonstrate excellent equatorial selectivity in the synthesis of bacterial-like KDO glycosides as a function of side chain conformation under our standard conditions.

We began with the synthesis of KDO thioglycosides 9, 11, 12 and 15 (Scheme 1). The intermediate 8 was obtained from 7 in 91% yield by the literature protocol.²³ It was converted into the thioglycoside 9 in 84% yield with 1-adamantanethiol in the presence of BF₃·Et₂O in dichloromethane. Removal of the acetonide groups with aqueous trifluoracetic acid then gave the pivotal tetraol 10 in 94% yield. Acetylation of 10 cleanly gave donor 11, while benzylation similarly gave 12 in excellent yield. Donor 15 was obtained from 10 via selective silylation of the primary alcohol to give 13, followed by installation of the di-tert-butylsilylene acetal in 14, and eventually benzoylation, all in excellent yield. The ³J_H6,H7 coupling constants of 8.3, 8.6 and 10.0 Hz in the ¹H NMR spectra of donors 9, 11, and 12, respectively, were consistent with that of 9.8 Hz in the pseudaminic acid donor 4⁷ and confirm the predominant tg conformation about the exocyclic bond in these molecules. On the other hand donor 15 in which the 6,7-bond is locked in the gg conformation has ³J_H6,H7 = 0.7 Hz consistent with our expectations.

Scheme 1.

Synthesis of KDO thioglycosides 9, 11, 12 and 15

Turning to glycosidic bond formation, a series of coupling were conducted with acceptors 16–21 (Figure 3), which were selected as model for the various bacterial KDO equatorial glycosides and prepared according to the literature methods.^24–28

Figure 3.

Glycosyl acceptors employed

Glycosylation reactions were conducted under our standard conditions^6,7 for ease of comparison. Thus, 2/1 dichloromethane/acetonitrile solutions of the donor-acceptor pairs were activated at −78 °C by NIS and TfOH in the presence of acid-washed 4 Å molecular sieves. Reaction mixtures were quenched by the addition of triethylamine at −78 °C before warming to room temperature and workup. Excellent yields were obtained for each system studied (Figure 4).

Figure 4.

Synthesis of KDO glycosides

In parallel with the neuraminic acid series, the assignment of anomeric configuration of KDO glycosides is best achieved by measurement of the of the ³J_C-1/H-3ax heteronuclear coupling constant, provided the standard ⁵C₂ chair conformation is retained by the KDO ring.^8,29 In this method a ³J_{C‑1/H‑3ax} value of 5.0−7.0 Hz indicates an equatorial glycoside, while a value of ≤1.0 Hz characterizes an axial glycoside. To a first approximation based on analysis of ³J_H,H coupling constants around the pyranoside ring all donors employed in this study adopted the ⁵C₂ chair conformation, albeit in the 4,5;7,8-di-O-acetonide 9 and the ensuing glycosides some minor distortion was revealed by the ²J_H3eq,H3a geminal coupling constant of 15.6 Hz, in contrast to that of ~12.0 Hz seen in all other donors and glycosides.

Coupling of the peracetylated donor 11 to methyl 2,3,4-tri-O-benzyl-α-D-glucopyranoside 16 yielded glycoside 22 in 89% yield, as a single equatorial anomer with ³J_{C‑1/H‑3ax} = 6.8 Hz (Figure 4). Similarly, coupling of 11 with acceptors 17, 18, 19, 20, and 21 afforded the desired equatorial glycoside 23, 24, 25, 26, and 27, each as single equatorial anomers in excellent isolated yield with ³J_{C‑1/H‑3ax} coupling constants ranging from 6.7–7.0 Hz of 7.0 Hz. It is therefore established that readily available donor 11 provides an excellent, selective entry into the type of equatorial glycosides of KDO that are found in bacterial capsular polysaccharides. We note that the elimination product 31 was formed in minor amounts in each of these reactions as determined by mass spectrometry of crude reaction mixtures: the pooling of fractions from multiple reaction mixtures enabled the isolation and purification of a sample for characterization (SI). We also note that the cleavage of the primary silyl ether from 24 occurred during work-up and not in the course of the glycosylation; competing glycosylation at the primary position of the acceptor was not observed. Glycosylation of acceptor 21 with the perbenzyl donor 12 gave the desired equatorial glycoside 29 in 83% isolated yield, which was characterized by a ³J_{C‑1/H‑3ax} value of 6.9 Hz. We did not find evidence for formation of the axial anomer of 29, suggesting that 12 is also a highly selective donor. Thus, consistent with our expectations donors 11 and 12 furnish outstanding equatorial selectivity.

Coupling of 4,5;7,8-di-O-isopropylidene-protected donor 9 to the mannosyl acceptor 17 also proceeded in high yield produced and the equatorial glycoside 28β in 86% isolated yield albeit accompanied by the formation of its anomer 28α in 10% yield. These assignments are confirmed by ³J_{C‑1/H‑3ax} coupling constants of 6.3 Hz and <1.0 Hz for 28β and 28α, respectively. The reduction in selectivity observed with 9, as compared to that seen with donors 11 and 12, despite all three systems adopting the tg side chain conformation, must be related to the distortion of the pyranoside ring imposed by the presence of the cis-fused 4,5-O-acetonide group.

Finally, we turned to donor 15 as it has been reported in the literature by Yang and co-workers that closely related 5,7‐O-silylene‐protected KDO thioglycosides are excellent donors for the formation of axial glycosides.¹¹ Consistent with the literature, coupling of 15 with the acceptor 21 gave the axial glycoside 30, with a ³J_{C‑1/H‑3ax} value of <1 Hz, in 87% isolated yield. We suggest that in addition to the steric shielding of the β-face advanced by Yang and coworkers, the α-selectivity of 15 derives from the gg conformation of the side chain imposed by the presence of the 5,7-O-silylene acetal. As with donor 11, trace amounts of the elimination products were observed mass spectrometrically in the reactions of donors 9, 12, and 15, albeit they were not isolated and quantified. Finally, as with 24, the primary silyl ether was lost from 30 in the course of the work-up and not during the glycosylation reaction.

Overall, we suggest that the selectivity observed in this series of KDO donors is consistent with β-selective attack on either a covalent axial glycosyl triflate or its functional equivalent the contact ion pair 31, or on a glycosyl nitrilium ion 32, when the protecting group array favors the most electron-withdrawing tg conformation of the side chain (Scheme 2)s. On the other hand, when the protecting group array imposes the gg conformation on the side chain a looser solvent separated ion pair 33 is populated to a greater extent, which results in inverted selectivity.

Scheme 2.

Mechanistic hypothesis outlining the importance of side chain conformation on reactivity and selectivity

In conclusion, the understanding gained from an extensive series of studies on the role of configuration and conformation on the reactivity and selectivity of neuraminic acid glycosyl donors, led to the prediction that KDO donors, due to their pseudosymmetry with pseudaminic donors, would be highly equatorially selective in glycosylation reactions conducted under our typical conditions provided that the protecting group array supports the inherent tg conformation of the KDO side chain. These predictions are borne out by the excellent equatorial selectivity observed with the peracetyl and perbenzyl-protected donors 11 and 12, and to a lesser extent by the diacetonide 9. Also consistent with this analysis is the reversal of seen with donor 15, whose side chain is forced to adopt the gg conformation. The successful demonstration of this principle enables the ready and highly selective synthesis of KDO equatorial glycosides such as are found in numerous pathogenic bacteria.