Conformational properties of α- or β-(1→6)-linked oligosaccharides: Hamiltonian replica exchange MD simulations and NMR experiments

Abstract

Conformational sampling for a set of 10 α- or β-(1→6)-linked oligosaccharides has been studied using explicit solvent Hamiltonian replica exchange (HREX) simulations and NMR spectroscopy techniques. Validation of the force field and simulation methodology is done by comparing calculated transglycosidic J coupling constants and proton-proton distances with the corresponding NMR data. Initial calculations showed poor agreement, for example, with >3 Hz deviation of the calculated (3)J(H5,H6R) values from the experimental data, prompting optimization of the ω torsion angle parameters associated with (1→6)-linkages. The resulting force field is in overall good agreement (i.e., within ∼0.5 Hz deviation) from experimental (3)J(H5,H6R) values, although some small limitations are evident. Detailed hydrogen bonding analysis indicates that most of the compounds lack direct intramolecular H-bonds between the two monosaccharides; however, minor sampling of the O6···HO2' hydrogen bond is present in three compounds. The results verify the role of the gauche effect between O5 and O6 atoms in gluco- and manno-configured pyranosides causing the ω torsion angle to sample an equilibrium between the gt and gg rotamers. Conversely, galacto-configured pyranosides sample a population distribution in equilibrium between gt and tg rotamers, while the gg rotamer populations are minor. Water radial distribution functions suggest decreased accessibility to the O6 atom in the (1→6)-linkage as compared to the O6' atom in the nonreducing sugar. The role of bridging water molecules between two sugar moieties on the distributions of ω torsion angles in oligosaccharides is also explored.

Scheme 1. Schematic Representation of the manno-(1–3), gluco-(4–5), and galacto-(6–10)-Configured (1/2→6)-Linked Pyranosides Included in the Current Study

Figure 1

Selected region of 1D ¹H spectra for the site-specifically ¹³C labeled isotopologues of compound 3. Experimental (a) and spin-simulated (b) spectra for [6-¹³C]-3 and experimental (c) and spin-simulated (d) spectra for [1′,6-¹³C₂]-3.

Figure 2

Examples of NMR spectra used in the determination of heteronuclear long-range coupling constants in compound 5: (a) determination of ³J(C1′,H6R) and ³J(C1′,H6S) using the J-HMBC experiment; (b) determination of ³J(C4,H6S) using the HSQC-HECADE experiment. (c) Correlation between the values for ³J(C1′,H6R) (red), ³J(C1′,H6S) (black), and ³J(C6,H1′) in compounds 1–8 labeled according to the stereochemistry at the anomeric carbon of the terminal residue.

Scheme 2. (a) Schematic Representation of Model Compounds; (b) Newman Projection of Ideal Staggered ω Rotamers about the C5–C6 Bond

A–D are model compounds used previously, and E–H represent the new model compounds used for deriving dihedral parameters for the ω torsion angle.

Figure 3

Two-dimensional free energy surfaces for the ϕ (O5′–C1′–O6–C6) vs ψ (C1′–O6–C6–C5) dihedrals for 1–9, given in degrees, calculated from the HREX MD simulations. Free energies are calculated from the natural logarithm of the relative probability and are given in kcal·mol^–1.

Figure 4

Two-dimensional free energy surfaces for the ψ (C1′–O6–C6–C5) vs ω (O6–C6–C5–O5) dihedrals for 1–9, given in degrees, calculated from the HREX MD simulations. Free energies are calculated from the natural logarithm of the relative probability and are given in kcal·mol^–1.

Figure 5

Cross-relaxation measurements in compound 8: (a) 1D ¹H spectrum and (b) 1D ¹H,¹H SPFGSE NOESY spectrum obtained with excitation at H1′ and a 500 ms cross-relaxation delay (t_mix). (c) Normalized peak integrals divided by t_mix for different values of t_mix (crosses) together with the fitted equations (lines). Intra- and inter-residual interactions are shown as dashed and full lines, respectively.

Figure 6

Proton–proton distance distributions for 2, 6, and 8 calculated from the HREX MD simulations. rH1′,H6_pro-R (a), rH1′,H4 (b), rH1′,H5 (c), and rH4,H6_pro-S (d) are given in Å. Solid spikes represent experimental effective distances, and dashed spikes represent effective distances from MD simulations.

Figure 7

ψ vs rH1′–H6_pro-R (a), ψ vs rH1′–H4 (b), ψ vs rH1′–H5 (c), and ω vs rH4–H6_pro-S (d) for 2, 6, and 8 obtained from HREX MD simulations. Proton–proton distances are in Å and dihedral angles in deg. Solid blue lines represent experimental effective proton–proton distances, and dashed blue lines represent calculated effective proton–proton distances.

Figure 8

Distance probability distribution for the O6···HO4 distance in 1–9 obtained from HREX MD simulations.

Figure 9

(a) Radial distribution functions for Ow (water oxygen) and O6. (b) Radial distribution functions for Ow and O6′. Note that O6′ is in the terminal end sugar and O6 participates in the (1→6)-glycosidic linkage.

Figure 10

Representative snapshots from the HREX simulations of 2 and 3 showing bridging water molecules. (a) In 2, the gt conformation at ω allows water to simultaneously form an H-bond to ring oxygen O5 and linkage oxygen O6 atom, (b) in 2, the gt conformation allows a water bridge between the ring oxygen O5′ and the linkage oxygen O6 atom, and (c) in 3, the gg conformation allows a water bridge between the two ring oxygen atoms O5 and O5′.

Figure 11

Molecular model of compound 10 showing H-bonding between O3 in the reducing end glucose residue and HO7′ in the terminal end Neu5Ac residue.