O-Acetyl Side-Chains in Monosaccharides: Redundant NMR Spin-Couplings and Statistical Models for Acetate Ester Conformational Analysis

Abstract

α- and β-d-glucopyranose monoacetates 1-3 were prepared with selective ¹³C enrichment in the O-acetyl side-chain, and ensembles of ¹³C-¹H and ¹³C-¹³C NMR spin-couplings (J-couplings) were measured involving the labeled carbons. Density functional theory (DFT) was applied to a set of model structures to determine which J-couplings are sensitive to rotation of the ester bond θ. Eight J-couplings (¹J_CC, ²J_CH, ²J_CC, ³J_CH, and ³J_CC) were found to be sensitive to θ, and four equations were parametrized to allow quantitative interpretations of experimental J-values. Inspection of J-coupling ensembles in 1-3 showed that O-acetyl side-chain conformation depends on molecular context, with flanking groups playing a dominant role in determining the properties of θ in solution. To quantify these effects, ensembles of J-couplings containing four values were used to determine the precision and accuracy of several 2-parameter statistical models of rotamer distributions across θ in 1-3. The statistical method used to generate these models has been encoded in a newly developed program, MA'AT, which is available for public use. These models were compared to O-acetyl side-chain behavior observed in a representative sample of crystal structures, and in molecular dynamics (MD) simulations of O-acetylated model structures. While the functional form of the model had little effect on the precision of the calculated mean of θ in 1-3, platykurtic models were found to give more precise estimates of the width of the distribution about the mean (expressed as circular standard deviations). Validation of these 2-parameter models to interpret ensembles of redundant J-couplings using the O-acetyl system as a test case enables future extension of the approach to other flexible elements in saccharides, such as glycosidic linkage conformation.

Figure 1

Calculated ¹J and ²J values in 5 as a function of θ₃. (A) ¹J_C3,H3, (B) ¹J_C2,C3, (C) ¹J_C3,C4, (D) ²J_C2,H3, (E) ²J_C4,H3, and (F) ²J_C2,C4. Red points correspond to the three perfectly staggered rotamers of θ₃. Overall changes in each J-value are shown in green.

Figure 2

Total energy (kJ/mol) as a function of (A) the H3–C3–O3-C1′ torsion angle (θ₃) in 5 (filled symbols) and 6 (open symbols), and (B) the H2–C2–O2-C1′ torsion angle (θ₂) in 7 (open symbols) and 8 (filled symbols).

Figure 3

Calculated ³J_C2′,C3 values in 5 as a function of the H3–C3–O3–C1′ torsion angle θ₃.

Figure 4

(A) Effect of the H3–C3–O3–C1′ torsion angle (θ₃) on calculated values of ²J_C1′,C3 in 5. The dynamic range is shown in red. (B) Summary of data for 5–8 showing the effect of C–O–C bond angle on calculated ²J_C1′,CX, where X = 2 or 3. Filled green, 5; open green, 6; open red, 7; filled red, 8. The inset shows a linear fit of the data.

Figure 5

Calculated ³J_C1′,HX values in 5–9 as a function of θ_X, where x = 2, 3, and 6. Filled green, 5; open green, 6; open red, 7; filled red, 8; filled blue, ³J_C1′,H6R and ³J_C1′,H6S in 9; filled black, data taken from ref . Curves were generated by point-to-point connectivity.

Figure 6

Calculated ³J_C1′,Cx values in 5–9 as a function of the C1′–O_x–C_x–H_x torsion angle θ_x (x = 2, 3 or 6). (A) ³J_C1′,C3 in 5 (filled blue); ³J_C1′,C4 in 6 (open blue); ³J_C1′,C3 in 7 (open red); ³J_C1′,C3 in 8 (filled red); ³J_C1′,C5 in 9 (filled black). (B) ³J_C1′,C4 in 5 (filled blue); ³J_C1′,C2 in 6 (open blue); ³J_C1′,C1 in 7 (open red); ³J_C1′,C1 in 8 (filled red).

Figure 7

Calculated ²J_C1′,CX values in 5–9 as a function of θ_x, where x = 2, 3, or 6. Filled green, ²J_C1′,C3 in 5; open green, ²J_C1′,C3 in 6; open red, ²J_C1′,C2 in 7; filled red, ²J_C1′,C2 in 8; open blue, ²J_C1′,C6 in 9.

Figure 8

Histograms showing the distributions of θ_x in 11 (A), 12 (B), 10 (C), and 13 (D) obtained from 1-μs aqueous molecular dynamics (MD) simulations.

Figure 9

Probability distributions across θ_x in 1, 2α, 2β, and 3 determined from a fit of redundant experimental J-couplings (Table 1) using parametrized eqs 2–9 and a von Mises model. Red, 2α; blue, 2β; green, 1α/β; black, 3.

Figure 10

Different statistical models of θ₆ in 3. Black, raised cosine; red, uniform; green, power of cosine; blue, von Mises; violet, Lorentz. Since platykurtic models approach zero more rapidly than models displaying more kurtosis, rotamers associated with θ_x values near the edge of the distribution are weighted more heavily, rendering the model more sensitive to changes in the CSD.

Figure 11

Parameter space for the uniform model of θ₆ in 3. Since only one minimum falls below an RMS error of 1 Hz, this minimum represents a unique solution for the rotamer distribution.

Scheme 1

Relative Dispositions of Flanking Hydroxyl Groups in Aldohexopyranosyl Rings Bearing an O-Acetyl Side-Chain at C3

Scheme 2

¹³C-Labeled Mono-O-acetylated D-Glucopyranoses 1–3

Scheme 3

Model Structures 5–9 Used in DFT Calculations, Showing Atom Numberings and Definitions of θ_x

Scheme 4

Identification of the H6R Reference Atom Used To Define θ₆ in Structure 9

Scheme 5

Model Structures 10–13 Used in Molecular Dynamics (MD) Simulations

Scheme 6

Structure Queries 14–16 Used by ConQuest in the Cambridge Structural Database Search

Scheme 7

NMR J-Couplings Sensitive to θ₃ and σ₃ in 1β

Scheme 8

Conformational Elements in a β-(1 → 4)-Linked Disaccharide