Introduction

The structural diversity of carbohydrates and glycoconjugates is very large and alterations of the oligosaccharide part in molecules presented on cell surfaces are often the result of changes in the metabolism of the cell [1]. Moreover, the anti-inflammatory activity of the monomeric immunoglobulin G is highly dependent of precise structure of the carbohydrate portion that is linked to the antibody, thereby stressing the importance of structure–function relationships [2, 3]. Ten different monosaccharides are used to build glycan structures in humans [4, 5], whereas in bacteria additional monosaccharides are present [6] and the number of different monosaccharides described is considerably larger [7]. Many of these monomers are aminosugars, of which most are N-acetylated. In hexopyranoses the amino group is often linked to C2, but also to C3 or C4, e.g., in 2-amino-2-deoxy-d-glucose, 3-amino-3,6-dideoxy-d-galactose or 4-amino-4,6-dideoxy-d-mannose.

The crystal structure of methyl 3-O-α-d-glucopyranosyl 2-acetamido-2-deoxy-α-d-galactopyranoside monohydrate (I) is presented herein (Fig. 1) and represents a model for part of the microbial cell wall teichoic acid polymer from Micrococcus sp. A1, which contains the disaccharide entity α-d-Glcp-(1 → 3)-α-d-GalpNAc connected by phosphodiester linkages at O6 of the glucose residue and O1 of the galactosamine residue [8].

Fig. 1
figure 1

Schematic representation of 3-O-α-d-glucopyranosyl 2-acetamido-2-deoxy-α-d-galactopyranoside hydrate (I)

Experimental

Synthesis and Crystallization

The synthesis of methyl 3-O-α-d-glucopyranosyl 2-acetamido-2-deoxy-α-d-galactopyranoside was described by Weinz [9], in which the constituent monosaccharides have the d absolute configuration and was subsequently characterized by 1H and 13C NMR spectroscopy [10]. The compound was crystallized by slow evaporation of a mixture of water and ethanol (1:1) at a temperature of 20–25 °C.

Single Crystal Diffraction

The tiny crystals of the title compound where mounted with epoxy glue on a thin glass fiber and mounted on a MAR platform equipped with a 165 mm MARCCD detector at beam line I911/5 in Maxlab-II, Lund, Sweden. The crystal to detector distance was 35 mm and the rotation range for each frame was 1°; 200 frames were collected in a single φ scan. Data reduction was done with the Crysalis software package [11].

The structure of 3-O-α-d-glucopyranosyl 2-acetamido-2-deoxy-α-d-galactopyranoside hydrate (I) was solved by direct methods with SHELXS [12]. Most of the non-hydrogen atoms were located in the initial electron density map and the rest of them in subsequent difference Fourier maps. All hydrogen atoms except the two on the hydrate water molecules were geometrically placed and allowed to ride on the carbon or oxygen atom to which they were connected; those in the water molecule were located from difference-density maps and restricted with a bond distance restraint as well as a bond distance restraint between the two hydrogen atoms in the water molecule, in order to maintain the known geometry of the water molecule. The model was refined by full-matrix least-square calculations with SHELXL [13]. The Flack parameter [14] was inconclusive but the absolute configuration could be assigned by reference to an unchanging chiral center. Anomalous dispersion corrections for the used wavelength were calculated with “CROMER for Windows” based on the Cromer-Liberman method using a Kissel-Pratt correction [15]. Anisotropic displacement parameters were used for all atoms in combination with an ISOR-restraint because of the small amount of significant data. A summary of the crystallographic data is found in Table 1.

Table 1 Crystal data for 3-O-α-d-glucopyranosyl 2-acetamido-2-deoxy-α-d-galactopyranoside hydrate (I)

Even with the high intensity radiation from the Max-II synchrotron the diffraction data above 1 Å resolution show very low signal-to-noise ratios (Fig. 2); thus, the R-values including all data were quite high. However, the R-values decreased considerably when less data were included in the refinements, essentially without any effect on the geometrical parameters. As a result of the small crystal size (50 × 20 × 10 μm3) and possibly poor crystallinity, the final figure of merit at different resolutions is rather high (cf. Table 2). The reconstruction of the 0kl plane shown in Fig. 2 shows the limited resolution of the data set. Up to 1 Å resolution the data are quite significant but further out in diffraction space the reflection data is much less significant, thus mainly contributing to the large R-values but not much to change the structure model.

Fig. 2
figure 2

Reconstructed 0kl projection resolution limits at 0.82 Å and 1.00 Å are marked by red circles. At resolutions higher than 1.00 Å the reflections became less significant and mainly contribute to larger R-values

Table 2 The least square refinements with reflection data of (I) cut at different resolution limits show the influence of a large number of insignificant reflections at higher resolutions

Results and Discussion

The crystal structure of 3-O-α-d-glucopyranosyl 2-acetamido-2-deoxy-α-d-galactopyranoside hydrate (I) has the two hexopyranosides residues arranged in a syn conformation at the central glycosidic linkage with the hydrogen atoms H1′ and H3 on the same side of the plane perpendicular to the C1′–H1′ vector (Fig. 3). The two hexopyranosides both reside in the 4C1 conformation as θ values are close to zero degrees; the Cremer & Pople puckering parameters [16], calculated by PLATON [17] are given by: Q = 0.52(2) Å, θ = 3(2)° and φ = 71(23)° for the ring O5 → C5 and Q = 0.57(2) Å, θ = 2(2)° and φ = 198(41)° for the O5′ → C5′ ring.

Fig. 3
figure 3

Crystal structure of 3-O-α-d-glucopyranosyl 2-acetamido-2-deoxy-α-d-galactopyranoside hydrate (I) with atomic labels. Anisotropic displacement ellipsoids are drawn at 30% probability level [25]

Torsion angles related to major degrees of freedom are compiled in Table 3 from which it can be observed that the ϕ torsion angles have the exo-anomeric conformation, most commonly observed for glycans and the ψ torsion angle has an eclipsed conformation with a value close to zero degrees. The ω torsion angle has a gt conformation for the sugar with galacto-configuration whereas the ω′ torsion angle has a gg conformation for the sugar with gluco-configuration, both of which being one of the two preferred conformations of hydroxymethyl groups in hexopyranoses [18]. The torsion at the amide bond of the N-acetyl group has a trans conformation, which is almost exclusively observed in saccharides in solution [19, 20]. Interestingly, the N-acetyl group is tilted as the torsion angle τH = 146° and deviates from an idealized antiperiplanar orientation. It is quite similar to the corresponding torsion in d-GlcNAc as a constituent of the pentasaccharide LNF-1 then having an average value for τH = 157° based on molecular dynamics (MD) simulations [21] and to N-acetyl groups in LNF-1 with τ = 60° (± 20) and in the hexasaccharide LND-1 with τ = 90° (± 30) deduced from 1H,1H-NOESY NMR experiments [22] as compared to τ = 85° in disaccharide I. The recent investigation of conformational preferences of the N-acetyl groups in e.g. methyl N,N′-diacetyl-chitobioside based on NMR scalar coupling constants revealed only small deviations from an antiperiplanar orientation of H2 and HN, with τ = 115° and τ′ = 110°, whereas in α-d-GlcNAc-OMe it differed somewhat more with τ = 106°, but that good agreement with MD simulation in aqueous solution was observed [20]. Thus, it is surmised that the additional tilting in disaccharide I, as well as that for the milk oligosaccharides, is due to the fact that the N-acetyl containing sugar residue has a glycosyl group at O3 vicinal to its N-acetyl group.

Table 3 Selected torsion angles in disaccharide I

The N-acetyl group at position 2 of the d-galactosamine residue makes the pyranose sugar bulkier as the substituent protrudes compared to d-galactopyranose. The structural entity α-d-Hexp-(1 → 3)-α-d-Galp where Hexp corresponds to a hexose sugar residue having the pyranose ring form is, like the title compound, present in a B blood group trisaccharide, specifically in α-l-Fucp-(1 → 2)[α-d-Galp-(1 → 3)]-β-d-Galp-OMe, for which the torsion angles at the (1 → 3)-linkage are ϕ =  − 63° and ψ =  − 55° in the crystal structure (Cambridge Structural Database code LOKDIY) [23]. In the trisaccharide the fucosyl residue is linked to position 2 of the disubstituted d-galactopyranoside, which in effect results in a change of, in particular, the ψ torsion angle relative to the sterically less crowded disaccharide that upon potential energy minimization using the HSEA (hard sphere exo anomeric) force field showed ϕ =  − 49° and ψ =  − 32° [23]. Geometry optimization of 3-O-α-d-glucopyranosyl 2-acetamido-2-deoxy-α-d-galactopyranoside by density functional theory (DFT) calculations at the theory level B3LYP/6-31G* using NWChem [24] resulted in ϕ =  − 55° and ψ =  − 44°. Thus, the presence of the N-acetyl group may have an influence such that the ψ torsion angle in the disaccharide on its own (in vacuo) becomes similar to that of the trisaccharide carrying a fucosyl residue in the corresponding position. Moreover, the computational results of the title disaccharide as a single molecule whose ψ torsion angle differs by ~ 45° to that in the solid state indicates that intermolecular forces (crystal packing) affect the conformation of the molecule significantly in the crystal.

Packing of disaccharides and water molecules in the bc-plane is depicted in Fig. 4. Extensive N–HO(C) hydrogen bonding takes place along the a-direction and the hydrate water molecules mediate additional hydrogen bonding along the a-direction O–HO3′ as well as between molecules via O–HO(C) along the b-direction (Table 4 and Fig. 5).

Fig. 4
figure 4

Packing of the disaccharide molecules in the bc-plane and hydrogen bonding scheme; the water molecule has been illustrated as an oxygen atom with larger radii and light blue color [25]

Table 4 Hydrogen bonds in the structure of disaccharide I as a monohydrate
Fig. 5
figure 5

Part of the hydrogen bonding scheme along the a-direction. The left structure is the model from the X-ray data and the right one is from the DFT model [24]

The structure model was further investigated by solid state DFT calculations with plane waves and pseudopotentials using NWChem [24], describing both the heavy atom framework as well as the hydrogen bond geometry very well. The differences between the X-ray model and the model derived from DFT calculations are minor, with slightly longer C–H, N–H and O–H distances in the DFT model, thus supporting the structure of the X-ray model even though the high angle diffraction data were quite insignificant. A visual comparison of hydrogen bond geometries from X-ray and DFT models is shown in Fig. 5, as well as a comparison between selected torsion angles from the X-ray and DFT models given in Table 3. Note that there are four generated individual values in the DFT model for each torsion angle defined by the X-ray model, due to the fact that the DFT optimizations were done in space group P1.