Synthesis and Molecular Structure of the 5-Methoxycarbonylpentyl α-Glycoside of the Upstream, Terminal Moiety of the O-Specific Polysaccharide of Vibrio cholerae O1, Serotype Inaba

The trimethylsilyl trifluoromethanesulfonate (TMSOTf)-catalyzed reaction of methyl 6-hydroxyhexanoate with 3-O-benzyl-4-(2,4-di-O-acetyl-3-deoxy-l-glycero-tetronamido)-4,6-dideoxy-2-O-levulinoyl-α-d-mannopyranosyl trichloroacetimidate followed by a two-step deprotection (hydrogenolysis over Pd/C catalyst and Zemplén deacylation, to simultaneously remove the acetyl and levulinoyl groups) gave 5-(methoxycarbonyl)pentyl 4-(3-deoxy-l-glycero-tetronamido)-4,6-dideoxy-α-d-mannopyranoside. The structure of the latter, for which crystals were obtained in the analytically pure state for the first time, followed from its NMR and high-resolution mass spectra and was confirmed by X-ray crystallography. The molecule has two approximately linear components; a line through the aglycon intersects a line through the mannosyl and tetronylamido groups at 120°. The crystal packing separates the aglycon groups from the tetronylamido and mannosyl groups, with only C-H…O hydrogen bonding among the aglycon groups and N-H…O, O-H…O and C-H…O links among the tetronylamido and mannosyl groups. A carbonyl oxygen atom accepts the strongest O-H…O hydrogen bond and two strong C-H…O hydrogen bonds. The geometric properties were compared with those of related molecules.


Introduction
O-specific polysaccharides (O-SP, O-antigens) are essential virulence factors and protective antigens of many pathogenic bacteria [1]. In Gram-negative bacteria, the same class of polysaccharides is responsible for the serological specificity of these pathogens. The O-SP of the two main strains of Vibrio cholerae O1, Inaba and Ogawa, consists of less [2] than 20 (1→2)-linked 4-amino-4,6-dideoxy-α-D-mannopyranosyl (perosaminyl) residues, the amino groups of which are acylated with 3-deoxy-L-glycero-tetronic acid. The two strains differ in that the terminal perosamine residue in the O-SP of the Ogawa strain is methylated at O-2 [3]. Following the pioneering work by Kenne et al. [4] on the synthesis of the methyl α-glycoside of the terminal, monosaccharide determinant of the O-SP of Vibrio cholerae O1, serotype Inaba, we have reported [5] an improved synthesis and the crystalline nature of the same compound. The presence of the methoxycarbonyl group in the title, spacer-equipped Compound 3 described here makes it amenable to conversions to an array of derivatives suitable for conjugation to proteins through different chemical processes. Thus, it will be useful, within Vibrio cholerae O1 strains, for making tools for immunological/immunogenicity studies towards elucidating the molecular basis for serotype specificity, which often require glycoconjugates. We have synthesized analogous substances from related oligosaccharides and converted them to conjugates [6] within our work towards a conjugate vaccine for cholera. The crystal structure of the complex from murine Fab S-20-4 (from a protective anti-cholera Ab specific for the lipopolysaccharide antigen of the Ogawa serotype) with synthetic mono-and di-saccharide fragments of the Ogawa O-SP has already been described [7]. The crystal structure of 3, whose synthesis (Scheme 1) and isolation in the crystalline state and full characterization is described here for the first time, will aid in the interpretation of data resulting from a similar study in the Inaba series. Scheme 1. Synthetic route of Compound 3.

Crystallography
The details of the crystallographic determination are shown in Table 1. The molecule is shown in Figure 1 with atomic numbering for the heavy atoms, confirming the chemical and NMR analyses of the structure. The molecule has a pronounced bend; lines that connect C17 to C1 and C1 to C13 intersect with an angle of 120 °C due to the axial α-glycosidic bond and the exo-anomeric effect.
The molecule is amphiphilic, and the crystal is organized by both conventional O-H…O and N-H…O hydrogen bonds, as well as by van der Waals and C-H…O interactions. The hydrophilic portion of the molecule is formed by the O-2 and O-3 side of the perosaminyl residue, and liberal criteria for hydrogen bonds (as per the PLATON crystal analysis software) [11] yield seven conventional H-bonds. Six C-H…O bonds were also identified by PLATON, and two others were identified visually with lengths just slightly past the PLATON criterion. Such long bonds are feasible; in a recent atoms-in-molecules analysis of cellulose, a C-H…O bond as long as 2.83 Å had an electron density at its bond critical point of 0.004 e/au [12]. Support for stabilization from interactions having small O-H…O angles was found in studies of 1,2-dihydroxycyclohexane [13]. In those vacuum calculations for rotations of one of the hydroxyl groups, stabilizations of about 2 kcal/mol occurred despite an O-H…O angle of about 105° and a H…O length of 2.4 Å. This was also despite the absence of a confirmatory bond critical point. All proposed hydrogen bonds are shown in Table 2. Figure 2 shows the conventional hydrogen bonding that consists of a ring and an infinite chain. All hydroxyl groups are both donors and acceptors. As shown in Table 2, the N4-H…O2 and O2-H…O3 links are of marginal quality (long H…O distances and small O-H…O angles) and were not reported by the ShelXL program used to refine the crystal structure. The double acceptor O10 permits the reversal of the nominal polarity of the hydrogen bonding (a fully cooperative network would have a "head-to-tail" donor-acceptor-donor-acceptor arrangement).    [14] with the default criteria as such, although both found a close H9b…O6 short contact. Mercury's criteria were adjusted to include carbon donors and a minimum D-H…O angle of 100° to prepare the drawings of Figure 3.  Table 2 for the geometric values). These hydrogen bonds are located near the two-fold screw axes that perpendicularly intersect the a-c plane. The C-H…O hydrogen bonds are more evenly distributed, as shown in Figure 3. The carbonyl oxygen O8 not only is the acceptor for the shortest O-H…O bond, but also is the acceptor for two short C-H…O hydrogen bonds, from H4 on the carbohydrate ring and from H16A on another tetronylamido residue ( Table 2). The aglycon participates only in C-H…O bonds.
Cremer-Pople puckering parameters for 3 and three other molecules of this series of compounds are in Table 3. All are within the ranges observed for rings described as 4 C1. Another measure of ring geometry is the distance across the ring, shown as the O1-N4 distance. The analogous O1-O4 distance for α-D-glucose determines (in a model-building sense) or is determined by (in an experimental sense) the location of substituents in the 1-and 4-positions (e.g., glucose residues in starch). The O1-N4 values are about 4.6 Å for this limited set of compounds, all near the upper end of the range (3.9 to 4.8 Å) observed for α-D-glucose [15].  [16] 0.550 (3) 3.5 (3) 166 (4) 4.520 TEDJOB a [16] 0.527(10) 9.2(12) 259 (7) 4.678 a The six-letter codes are the "reference codes" used in the Cambridge Crystal Structure Database (CSD) [17]. The   Table 5 provides geometric details for the acetamido groups of the four related compounds, as well as a survey of the CSD. The first TEDJOB C-C value is very short, probably because of the disorder of the attached fluorines and high R factor. In 3, the length of the carbonyl carbon to nitrogen bond (1.3309(15) Å) is much shorter than the nominally similar N-C4 bond (1.4648(14) Å). This difference is in excellent agreement with values from a search of the CSD. The endocyclic C4-C3 and C4-C5 bond lengths are normal, apparently not affected by the presence of the adjacent nitrogen. Both the nitrogen and the adjacent carbonyl carbon atom have nominal sp 2 hybridization that places them and their attached atoms in a common plane. The means of the absolute deviations from this plane are shown in Table 6. As reported in [5], there is a small deviation for SUNFEM, but as shown in Table 6, there is less deviation for 3. There is even less deviation for the other structures or the mean deviations from the search of 4,086 structures in the CSD. Finally, the six atoms of the three related rings were fit to the ring in 3, and the rings had a high degree of similarity, despite the variations in puckering and bond lengths. Table 6. Miscellaneous.

General Information
Optical rotation was measured at ambient temperature with a digital Jasco automatic polarimeter, Model P-2000 (Easton, MD, USA). The melting point was measured on a Kofler hot stage. All reactions were monitored by thin-layer chromatography (TLC) on silica gel 60-coated glass slides. Column chromatography was performed by elution from prepacked columns of silica gel (Varian, Inc., Palo Alto, CA, USA) with the Isolera Flash Chromatograph (Biotage) connected to the external Evaporative Light Scattering Detector, Model 380-LC (Varian, Inc.). Nuclear Magnetic Resonance (NMR) spectra were measured at 600 MHz for 1 H and 150 MHz for 13 C, with Bruker Avance spectrometers (Billerica, MA, USA). Solvent peaks were used as the internal reference relative to tetramethylsilane (0 ppm). Assignments of NMR signals were made by homonuclear and heteronuclear two-dimensional correlation spectroscopy, run with the software supplied with the spectrometers. When reporting assignments of NMR signals, nuclei associated with the Tetronic side chain are denoted with a prime, and those associated with the spacer are denoted with a double prime. Liquid chromatography-electron spray-ionization mass spectrometry (ESI-MS) was performed with a Hewlett-Packard 1100 MSD spectrometer (Palo Alto, CA, USA).
Single crystal X-ray diffraction intensities were collected using a Bruker Kappa APEX II 4K CCD (Madison, WI, USA) 4-circle automated diffractometer and MoKα radiation. During data collection, the sample was cooled to 200(2) K using a stream of cold N2 gas generated with an Oxford Cryosystems 700 low-temperature system. The crystal structure was solved [18] using SHELXS-97 and refined using SHELXL-97 [19]. The absolute configuration of the structure was established from the known configuration of the α-D-mannopyranoside ring, and the absolute structure (Flack) parameter, although not definitive, due to the weak anomalous scattering contributions with MoKα radiation, is consistent with the assignment. Hydrogen atoms attached to O and N atoms were located in a difference Fourier map and refined with isotropic temperature factors. The positions of H atoms attached to C atoms were calculated using idealized sp 3 geometry and included as riding atoms in the least-squares refinement. For the methyl hydrogens, the torsion angle about the C-Me bond was optimized during the refinement.