Comment

The N-linked glycans of human glycoproteins are characterized by a common pentasaccharide core, Man₃GlcNAc₂, containing three D-mannose (Man) and two N-acetyl-D-glucosamine (GlcNAc) residues, with the terminal -GlcNAc-(14)--GlcNAc portion linked to the L-asparagine side chains of the protein (Taylor & Drickamer, 2003). In contrast, the modes of attachment of O-linked glycans to proteins are more diverse, with N-acetyl-D-galactosamine, D-glucose, D-galactose, D-mannose or D-xylose covalently attached via the side chains of L-serine and L-threonine (Voet & Voet, 2011). For example, in the proteoglycans, chrondroitin sulfate polysaccharide chains are covalently attached to the core protein via a -GlcA-(13)--Gal-(13)--Gal-(14)--Xyl tetrasaccharide (Xyl is xylopyranosyl), (I), with the terminal -Xyl residue linked to L-serine (Nadanaka & Kitagawa, 2008). The residues comprising this linkage tetrasaccharide may be O-sulfated or O-phosphorylated. In this report, the crystal structure of the -Gal-(14)--Xyl substructure of (I) has been determined in the form of its methyl glycoside, namely, methyl -D-galactopyranosyl-(14)--D-xylopyranoside, (II). This new structure complements those of other structurally related -(14)-linked disaccharides reported previously, including methyl -D-galactopyranosyl-(14)--D-glucopyranoside, (III) (Stenutz et al., 1999), methyl -D-galactopyranosyl-(14)--D-glucopyranoside, (IV) (Pan et al., 2005), methyl -L-galactopyranosyl-(14)--D-glucopyranoside, (V) (Pan et al., 2006), methyl -D-galactopyranosyl-(14)--D-mannopyranoside, (VI) (Hu et al., 2010), methyl -D-galactopyranosyl-(14)--D-allopyranoside, (VII) (Zhang et al., 2010), and methyl -D-glucopyranosyl-(14)--D-glucopyranoside, (VIII) (Ham & Williams, 1970).

Methyl -D-galactopyranosyl-(14)--D-xylopyranoside, (II), was prepared by a chemical route (see Supplementary material for synthetic details). After purification by chromatography, (II) was crystallized from water to give microcrystals devoid of water. In this report, the crystal structure of (II) is compared with that of the structurally related disaccharide, methyl -D-galactopyranosyl-(14)--D-glucopyranoside [methyl -lactoside, (III); Stenutz et al., 1999] (Table 2).

The crystal structure of (II) exhibits elements of disorder not observed in those of (III) and other -(14)-linked disaccharides (Pan et al., 2005, 2006; Hu et al., 2010; Zhang et al., 2010; Ham & Williams, 1970). This disorder is located exclusively within the Xyl residue, and modeling of the diffraction data yielded major (70%) and minor (30%) components; atoms in the latter are denoted with the suffix A throughout this article. The observed disorder has been attributed to an oscillation of the Xyl ring perpendicular to the plane of the ring, translating into librational motion across the -(14) linkage. This behavior appears to be a characteristic feature of crystals of (II); data obtained from different samples of crystals yielded the same disorder. Why disorder in the glycosidic linkage appears in crystals of (II) and not in those of other -(14)-linked disaccharides including (III) is unclear, but ring geometry and/or substituent effects in the Xyl ring, or packing considerations, probably play a role. The internal glycosidic linkages in other -(14)-linked disaccharides are constrained by inter-residue hydrogen bonding between atoms O3 and O5'. The lack of an exocyclic -CH₂OH group in the Xyl ring of (II) may allow this hydrogen bonding in a wider range of linkage conformations, thus leading to linkage flexibility/disorder in the solid state. Whether this putative difference plays a functional role in the O-linkages of glycoproteins involving -Xyl remains uncertain.

The presence of disorder in crystals of (II) complicates a quantitative analysis of molecular parameters such as bond lengths, angles and torsions due to averaging effects on the electron densities (and corresponding displacement ellipsoids) used to determine the structure. This fact is evident, for example, when comparing the C4-C5 bond length in (II) with that in (III), where a difference of 0.04 Å is observed (Table 2). In contrast, inspection of the X-ray structure of methyl -xylopyranoside, (IX) (Takagi & Jeffrey, 1977), shows a C4-C5 bond of 1.519 Å, in better agreement with the corresponding value in (III) than in (II).

For the major component of (II), the C1-O6-C6 bond angle is 118.3 (9)°, statistically larger than the value of 113.7 (2)° found in (III), with the latter value similar to the corresponding value found in methyl -xylopyranoside, (IX) (113.0°). In contrast, the internal C1'-O1'-C4 bond angle in (II) is 113.6 (7)°, similar to the corresponding value found in (III) [116.2 (2)°; Table 2]. This internal glycosidic bond angle is typically larger than that associated with the terminal methyl glycosides, presumably due to greater steric strain present in the internal linkage. The normal value of this angle in (II) suggests that the internal glycosidic torsion angles in (II) (i.e. ' and ') are probably minimally affected by the presence of disorder in the Xyl residue. This conclusion is supported by the similar bond lengths observed in the Gal residues of (II) and (III) (Table 2).

Cremer-Pople (CP) puckering parameters for the pyranosyl ring constituents of (II) and (III) are shown in Table 3 (Cremer & Pople, 1975; Boeyens, 1978; Spek, 2009). The -Gal ring in both structures adopts a chair conformation with similar and values, indicating similar degrees and directions for the slight distortion towards the ^C3TB_C1 conformation (TB = twist-boat). The -Xyl ring of (II) and -Glc ring of (III) show the same degree of distortion, with the former slightly skewed towards ^C3,O5B (B = boat) and the latter towards ^O5TB_C2. In contrast, the crystal structure of methyl -D-galactopyranoside, (X) (Takagi & Jeffrey, 1979), has = 5.89° and = 346.6°, indicating a direction of distortion (^O5TB_C2) different from that in the Gal residues of (II) and (III). The CP parameters for methyl -D-xylopyranoside, (IX), and methyl -D-glucopyranoside, (XI) (Jeffrey & Takagi, 1977), are very similar ( = 7-8° and = 36-38°), indicating a direction of distortion in (IX) similar to that in the Xyl reside of (II) but a direction of distortion in (XI) different from that found in the Glc residue of (III). However, it must be noted that the distortion in (II) is highly dependent on the model used to treat the disorder. Examination of the electron-density map of the disorder shows no evidence for discrete regions of electron density corresponding to the separate components, so it is highly probable that the -Xyl ring of (II) is dynamic, making the distortion difficult to quantify reliably.

The exocyclic hydroxymethyl conformations in the Gal residues of (II) and (III) are similar, with O5'-C5'-C6'-O6' torsion angles near 60°, corresponding to the gauche-trans (gt) conformation. This behavior is similar to that found in methyl -D-galactopyranoside, (X).

The internal glycosidic linkage conformation in (II) differs from that in (III), with the difference associated more with ' than with '; the ' values differ by 3°, whereas the ' values differ by 16°. Table 4 summarizes the ' and ' values observed in a series of -(14)-linked disaccharides for which crystal structures have been reported. In this comparison, torsion angles involving heavy atoms were used to define ' and ', namely, C2'-C1'-O1'-C4 for ' and C1'-O1'-C4-C3 for '. It is noteworthy that considerable variability in both ' and ' is observed, with that for ' (37.7°) larger than that for ' (28.4°). These findings, although confined to a relatively small data set, show that the stereoelectronic (exo-anomeric) effect does not severely constrain the ' torsion angle in these linkages; presumably, crystal packing forces are strong enough to rotate the C1'-O1' bond in order to minimize the packing energy. The comparatively greater variability in ' is presumably caused by different nonbonded (steric) effects that are structure-dependent, although packing forces may also contribute. Disaccharide (V) was excluded from this comparison since it contains an L-Gal residue and thus its internal glycosidic linkage is structurally distinct from the others.

Internal (inter-residue) hydrogen bonding occurs in (II) between atom O3 of the Xyl residue and atom O5' of the Gal residue. The O3O5' internuclear distances of 2.729 (5) and 2.803 (13) Å (major and minor components, respectively) involving the two O3 (O3/O3A) positions show that two hydrogen-bonding geometries are possible; this observation suggests some plasticity in the overall conformation in accommodating this type of intramolecular hydrogen bond. It is noteworthy that all of the disaccharides in Table 4 contain this hydrogen bond except (V) and (VII). The major-component hydroxy atom O3 also has a long inter-residue contact with hydroxymethyl atom O6'.

All hydroxy H atoms in the Gal moiety of (II) are involved as donors in intermolecular hydrogen bonds. In the Xyl residue, both atoms O2 and O3 are hydrogen-bond donors, whereas atom O2A is not well positioned to form a hydrogen bond with nearby acceptors. Atom O3 is involved as a donor in an inter-residue hydrogen bond with atom O5'. Atoms O1 and O2 of the Xyl residue are not involved as acceptors, whereas atoms O3 and O5 serve as mono-acceptors. Atoms O1', O2', O4' and O5' of the Gal residue are not involved as acceptors in intermolecular hydrogen bonds, while atom O3' serves as a mono-acceptor.

The overall packing motif of (II) is a three-dimensional network of hydrogen-bonded molecules (Fig. 2). The interaction of atom O2' with ring atom O5ⁱⁱ forms chains related by the screw axis parallel to the c axis [symmetry code: (ii) -x + , -y + 2, z + ]. These chains are linked to other chains related by screw axes parallel to the a axis through the hydrogen bonds from atoms O4' to O3^iv, O2 to O4'ⁱ and O6' to O3'ⁱ [symmetry codes: (i) x - , -y + , -z + 2; (iv) x + , -y + , -z + 2]. These screw axes are translated along the a axis with respect to the others. Lastly, atom O3' forms a hydrogen bond with atom O6'ⁱⁱⁱ related by the screw axis parallel to the b axis [symmetry code: (iii) x + , -y + , -z + 1]. The disordered hydroxymethyl atom C6 and the minor component hydroxy atom O2A are oriented towards a void space within the lattice.

Figure 1 The molecular structure of (II), with atom numbering and showing both disordered components of the molecule. Displacement ellipsoids are depicted at the 50% probability level.

Figure 2 The hydrogen-bonding network for (II), viewed along the c axis. The minor disorder component has been omitted for clarity. Dashed lines represent hydrogen bonds.

Experimental

The crystal structure of (II) was determined using a sample prepared chemically by the eight-step synthesis described in the Supplementary materials ; the relevant literature references are: Schmidt & Michel (1985); Wu & Serianni (1991); Podlasek et al. (1995); Tropper et al. (1992); Gruzman et al. (2008); Ning et al. (2003). Disaccharide (II) was crystallized from water to give colorless needle-like microcrystals. Due to the growth of the compound as microcrystals, conservatively estimated at 10 µm in thickness, and the presence of only light atoms within the sample, synchrotron radiation was a necessity for the determination of the structure. Standard laboratory instruments only yielded data suitable for a low-quality preliminary structure.

Crystal data

C12H22O10 Mr = 326.30 Orthorhombic, P 21 21 21 a = 13.7878 (14) Å b = 22.892 (2) Å c = 4.6367 (5) Å V = 1463.5 (3) Å3 Z = 4 Synchrotron radiation = 1.23990 Å = 0.52 mm-1 T = 150 K 0.08 × 0.01 × 0.01 mm

Data collection

Bruker APEXII diffractometer Absorption correction: empirical (using intensity measurements) (SADABS; Sheldrick, 2008) Tmin = 0.566, Tmax = 0.749 11197 measured reflections 2300 independent reflections 1900 reflections with I > 2(I) Rint = 0.073 max = 45.3°

Refinement

R[F2 > 2(F2)] = 0.067 wR(F2) = 0.158 S = 1.16 2300 reflections 298 parameters 299 restraints H-atom parameters constrained max = 0.35 e Å-3 min = -0.27 e Å-3 Flack parameter: 0.6 (9), 920 Friedel pairs

Table 1 Hydrogen-bond geometry (Å, °) D-HA D-H HA DA D-HA O2-H2O4'i 0.84 2.27 2.997 (7) 146 O3-H3O5' 0.84 2.09 2.729 (5) 133 O3-H3O6' 0.84 2.48 2.978 (6) 119 O3A-H3AAO5' 0.84 2.09 2.803 (13) 143 O2'-H2'O5ii 0.84 1.88 2.714 (9) 169 O2'-H2'O5Aii 0.84 2.03 2.86 (2) 169 O3'-H3'O6'iii 0.84 1.90 2.653 (5) 148 O4'-H4'O3iv 0.84 1.91 2.720 (6) 163 O4'-H4'O3Aiv 0.84 2.01 2.795 (13) 154 O6'-H6'O3'i 0.84 1.85 2.678 (5) 169 Symmetry codes: (i) ; (ii) ; (iii) , ; (iv) .

Table 2 Comparison of structural parameters in (II) and (III) gt stands for gauche-trans. Bond distances and internuclear contacts (Å) (II)# (III) C1-C2 1.495 (9) 1.516 (3) C2-C3 1.519 (8) 1.519 (3) C3-C4 1.518 (9) 1.531 (3) C4-C5 1.489 (9) 1.530 (3) C5-C6   1.508 (3) C1'-C2' 1.530 (6) 1.527 (3) C2'-C3' 1.529 (6) 1.531 (3) C3'-C4' 1.535 (6) 1.521 (3) C4'-C5' 1.520 (6) 1.521 (3) C5'-C6' 1.516 (7) 1.511 (3) C1-O1 1.373 (9) 1.384 (3) C1-O5 1.444 (8) 1.413 (3) C2-O2 1.413 (8) 1.418 (3) C3-O3 1.424 (8) 1.421 (3) C5-O5 1.443 (8)   C6-O6   1.424 (3) C1'-O1' 1.406 (5) 1.387 (3) C1'-O5' 1.441 (5) 1.425 (3) C2'-O2' 1.429 (6) 1.414 (3) C3'-O3' 1.434 (5) 1.422 (3) C4'-O4' 1.417 (6) 1.423 (3) C5'-O5' 1.452 (6) 1.432 (3) C6'-O6' 1.439 (5) 1.426 (3) C4-O1' 1.443 (17) 1.437 (3) O3O5' 2.729 (5) 2.764 (2) O3O6' 2.978 (6)         Bond angles (°)     C1'-O1'-C4 113.6 (7) 116.2 (2) C1-O1-C6 118.3 (9) 113.7 (2)       Torsion angles (°)     C2-C1-O1-C6 () 164.5 (9) 164.2 (2) O5-C1-O1-C6 () -81.4 (11) -77.4 (3) C2'-C1'-O1'-C4 (') 156.4 (5) 153.8 (2) O5'-C1'-O1'-C4 (') -85.7 (6) -88.4 (2) C1'-O1'-C4-C3 (') 94.0 (11) 78.4 (2) C1'-O1'-C4-C5 (') -141.6 (8) -161.3 (2) H1'A-C1'-O1'-C4 (') 34.3 31.9 C1'-O1'-C4-H4A (') -25.2 -43.7 O5'-C5'-C6'-O6' (') 60.7 (5) (gt) 57.4 (2) (gt) #Only parameters pertaining to the major component are reported.

Table 3 Cremer-Pople puckering parameters in (II), (III) and (IX)-(XI)# Compound (°) (°) Q (Å) q2 (Å) q3 (Å) (II), Galp 7.3 (5) 14 (4) 0.596 (5) 0.078 (5) 0.591 (5) (II), Xylp 13.9 (10) 6(5) 0.551 (11) 0.131 (10) 0.535 (11) (III), Galp 4.84 (19) 28.0 (3) 0.595 (2) 0.049 (2) 0.593 (2) (III), Glcp 11.9 (2) 341.3 (13) 0.558 (2) 0.116 (2) 0.546 (2) (IX), Xylp 8.17 36.4 0.5795 0.0824 0.5737 (X), Galp 5.89 346.7 0.5824 0.0597 0.5793 (XI), Glcp 6.91 37.9 0.5972 0.0718 0.5928 #No s.u. values were provided in the original reports for (III), (IX), (X) and (XI).

Table 4 Comparison of ' and ' glycosidic torsion angles in several -(14)-linked disaccharides See Comment for literature references to individual disaccharide X-ray reports. Compound C2'-C1'-O1'-C4, ' (°) C1'-O1'-C4-C3, ' (°) Gal(14)XylOCH3, (II) 156.4 (5) 94.0 (11) Gal(14)GlcOCH3, (III) 153.8 (2) 78.4 (2) Gal(14)GlcOCH3, (IV) 148.1 (1) 93.5 (1) -L-Gal(14)GlcOCH3, (V) -146.19 (12) 111.14 (13) Gal(14)ManOCH3, (VI) 173.1 (2) 115.2 (2) Gal(14)AllOCH3, (VII) 144.74 (10) 77.55 (13) Glc(14)GlcOCH3, (VIII)# 152.0 80.3 #No s.u. values were given in the original report.

Examination of the xylose moiety showed positional disorder in the peripheral atoms and close inspection of the displacement ellipsoids demonstrated that the entire ring was affected. The positions of the atoms in the major and minor components were determined initially by location of the major component and subsequent refinement of these sites at less than full occupancy, enhancing the difference Fourier map which displayed the location of the minor component atoms. The occupancies of the major and minor components were refined and summed to unity, yielding a 0.692 (9):0.308 (9) ratio. The minor component was restrained to have bond distances and angles similar to those of the major component to within a small error (s.u. = 0.02 Å or 0.02°). The major and minor components were both refined with anistropic displacement parameters, with displacement parameters of adjacent atoms restrained to have similar U^ij values in the two disorder components.

H atoms were initially located from a difference Fourier map and subsequently included as riding atoms in geometrically idealized positions, with C-H = 0.98 (methyl), 0.99 (methylene) or 1.00 Å (methine) and O-H = 0.84 Å. For all H atoms, U_iso(H) = kU_eq(parent), where k = 1.5 for methyl groups and 1.2 for all other H atoms. Hydroxy H atoms were permitted to rotate but not tilt.

The assignment of the absolute configuration was based on the known configuration of the disaccharide from the synthesis. Refinement of the Flack x parameter [x = 0.6 (9); Flack, 1983] and Bayesian analysis of Bijvoet pairs of reflections [y = 0.4 (4); Hooft et al., 2008] did not yield a conclusive analysis of the correct absolute configuration.

Data collection: APEX2 (Bruker, 2008); cell refinement: SAINT (Bruker, 2008); data reduction: SAINT; program(s) used to solve structure: SHELXS97 (Sheldrick, 2008); program(s) used to refine structure: SHELXL97 (Sheldrick, 2008); molecular graphics: XP (Sheldrick, 2008), POV-RAY (Cason, 2003) and DIAMOND (Brandenburg, 2009); software used to prepare material for publication: XCIF (Sheldrick, 2008) and publCIF (Westrip, 2010).

Supplementary data for this paper are available from the IUCr electronic archives (Reference: TP3005 ). Services for accessing these data are described at the back of the journal.