Ramachandran-Type Plots for Glycosidic Linkages: Examples from Molecular Dynamic Simulations Using the Glycam06 Force Field AMANDA M. SALISBURG, ASHLEY L. DELINE, KATRINA W. LEXA, GEORGE C. SHIELDS,* KARL N. KIRSCHNERy Chemistry Department, Center for Molecular Design, Hamilton College Clinton, New York 13323 Received 24 January 2008; Revised 26 May 2008; Accepted 8 July 2008 DOI 10.1002/jcc.21099 Published online 10 September 2008 in Wiley InterScience (www.interscience.wiley.com). Abstract: The goals of this article are to (1) provide further validation of the Glycam06 force field, specifically for its use in implicit solvent molecular dynamic (MD) simulations, and (2) to present the extension of G.N. Ramachandran’s idea of plotting amino acid phi and psi angles to the glycosidic phi, psi, and omega angles formed between carbohydrates. As in traditional Ramachandran plots, these carbohydrate Ramachandran-type (carb-Rama) plots reveal the coupling between the glycosidic angles by displaying the allowed and disallowed conformational space. Considering two-bond glycosidic linkages, there are 18 possible conformational regions that can be defined by (a, /, w) and (b, /, w), whereas for three-bond linkages, there are 54 possible regions that can be defined by (a, /, w, x) and (b, /, w, x). Illustrating these ideas are molecular dynamic simulations on an implicitly hydrated oligosaccharide (700 ns) and its eight constituent disaccharides (50 ns/disaccharide). For each linkage, we compare and contrast the oligosaccharide and respective disaccharide carb-Rama plots, validate the simulations and the Glycam06 force field through comparison to experimental data, and discuss the general trends observed in the plots. q 2008 Wiley Periodicals, Inc. J Comput Chem 30: 910–921, 2009 Key words: Glycam force field; carbohydrate; Ramachandran plot; molecular dynamics; disaccharide Introduction Recently, the Glycam06 force field was published, which was specifically designed to model carbohydrates but whose utility can be extended to other compound classes.1 This force field was validated by comparing gas-phase molecular mechanics results to quantum mechanical gas-phase data, and by comparing explicitly hydrated molecular dynamics simulations to a number of different condensed-phase experimental data. Herein we perform several simulations using an implicit water model and compare the obtained conformers and their distribution to those found by X-ray and NMR methods. To aid our discussion and description of the obtained conformers, we discuss and elaborate on how Ramachandran plots can be extended to carbohydrates. In 1963, G.N. Ramachandran introduced the idea of plotting the phi angle versus the psi angle for amino acid peptide linkages to reveal the coupling between these two angles.2,3 Recognizing that an amino acid’s backbone can define a plane, Ramachandran was able to define the relative orientation of two adjacent amino acids by these two degrees of freedom. This greatly reduced the number of terms required for defining the occupied conformational space of peptide linkages, resulting in the easily interpretable and very popular two-dimensional plots. These plots are well understood, with specific regions of allowed and disallowed conformational areas displayed. Ramachandran’s idea can be extended to di-, poly-, and oligosaccharides. Analogous to amino acids, the representation of a carbohydrate can be reduced to a plane that is formed by four ring atoms (see Fig. 1). In general, a carbohydrate’s ring puckering does not change (e.g., 4C1 for pyranoside), allowing the relative orientation between two adjoined carbohydrates to be defined by the angles between the planes, which can be described by the glycosidic angles phi (/) and psi (w). An addiAdditional Supporting Information may be found in the online version of this article. *Present address: Dean, College of Science and Technology, Armstrong Atlantic State University, Savannah, Georgia 31419 y Present address: Fraunhofer Institute for Algorithms and Scientific Computing, Department of Simulation Engineering, Schloss Birlinghoven 53754 Sankt Augustin, Germany and is a guest at the Max Plank Institute for Molecular Physiology, OttoHahnStr. 11, D44227 Dortmund, Germany. Correspondence to: G.C. Shields; e-mail: George.Shields@armstrong. edu or K. N. Kirschner; e-mail: [email protected] Contract/grant sponsor: National Institutes of Health; contract/grant number: 1RI5CA11552401 Contract/grant sponsor: NSF; contract/grant numbers: CHE0457275, CHE0116435, and CHE0521053 as part of the MERCURY high performance computer consortium (http://mercury.chem.hamilton.edu) q 2008 Wiley Periodicals, Inc. Molecular Dynamic Simulations Using the Glycam06 Force Field Figure 1. Graphical representation of the planes formed by carbohydrates, and the phi (/), psi (w), and omega (x) angles that can be used to provide the relative orientation of the two planes. tional angle, omega (x), is required for 1?6 linkages. The coupling between these angles can be observed by plotting phi versus psi versus omega. As in traditional Ramachandran plots, these carbohydrate Ramachandran-type (carb-Rama) plots display allowed and disallowed conformational space. Unique to carbohydrates are the number of possible linkages that can be formed between carbohydrate residues. Amino acids and DNA form polymeric chains in a single manner regardless of the residue involved; carbohydrates can form several different glycosidic linkages that depend on (a) the position in the ring (e.g., 1?1, 1?3, 1?6), (b) the form of the anomeric center of the nonreducing carbohydrate forming the linkage, axial (a) or equatorial (b), and (c) the glycosidic linkage position, axial or equatorial, that is formed on the reducing carbohydrate.4 The later two factors depend on the individual carbohydrate (e.g., glucose, mannose, galactose). For example, a-D-Glc-(1?2)-b-D-Man will have a set of phi and psi values that may differ from those present in b-D-Glc-(1?2)-b-D-Glc. Finally, the conformations adopted by glycosidic linkages are influenced by the surrounding environment, such as solvent molecules or an attached protein.5,6 911 The structure–activity relationship paradigm is a widely accepted and useful theory in biochemical and medicinal research. Understanding the properties and function of biomolecules relies partially on understanding their three-dimensional geometry. Significant effort has gone into understanding the amino acid code and the nucleic acid code and how they relate to function and geometry, as in the protein-folding problem. Studying the sugar code permits us to recognize the allowed and disallowed three-dimensional geometries of glycosidic linkages. Although there are numerous examples in the literature of carb-Rama plots created for specific linkages, including the formation of potential surface maps (a.k.a. conformational maps) using a particular level of theory,7–18 NMR-NOE based maps,19–22 and MD scatter plots,23–27 there has been little effort at creating generalized carb-Rama plots. In 1999 and 2002, Wormald and coworkers surveyed available NMR and X-ray data of oligosaccharides and glycopeptides. In this work, they presented carb-Rama plots based on experimental data.22,28 Experimental data provide the foundation for the development of allowed and disallowed conformational space of glycosidic linkages. However, much of this data involves large glycans attached to a protein because there is little experimental data for unbound oligosaccharides or disaccharides. Recently, Frank and coworkers have published an online database that provides free energy carbRama maps derived from MD simulations.29,30 These simulations were performed using the MM3 force field modified for Tinker at a temperature of 1000 K and with the carbohydrate rings constrained, and it is unclear if they included solvent in their calculation. This database includes links to experimentally determined (crystallographic and NMR) phi, psi, and omega angle values. A similar online database has been developed by the Centre National de la Recherche Scientifique and is called Glyco3D. Thus, the concept of assigning Ramachandran-type maps to carbohydrate linkages is an issue that is becoming more popular. In this article, we report the results of a 700 ns unrestrained MD simulation performed at a temperature of 300 K on an implicitly hydrated oligosaccharide composed of 12 carbohydrate residues, shown in Figure 2. We have also performed 50 ns MD simulations on the eight constituent disaccharides that compose this oligosaccharide. The objective of this article is threefold: (1) we will present solution-phase carb-Rama plots for each of the eight disaccharides, as well as compare and discuss how they are different from the plots formed by the oligosaccharide; (2) we will validate our modeling, and the Glycam06 force field, by comparing the simulation results to experimental X-ray and NMR data; and (3) we will discuss general trends in the carbRama plots. The simulation results and the carb-Rama plots will be useful for future oligosaccharide studies in determining the Figure 2. The oligosaccharide studied in this article. The disaccharides studied are composed from this oligosaccharide, each having significant X-ray and NMR experimental data available for comparison and validation of simulations. Journal of Computational Chemistry DOI 10.1002/jcc 912 Salisburg et al. • Vol. 30, No. 6 • Journal of Computational Chemistry Figure 3. (a, b) Phi, psi, and omega carb-Rama plots for the disaccharide’s glycosidic linkages. factors that cause specific glycosidic linkage conformations to be adopted (i.e., intrinsic to the linkage type itself or due to external forces arising from the environment). Methods All minimizations and MD calculations were performed using the AMBER 8 and 9 software packages using the GLYCAM06 force field.1 Nonbonded and electrostatic scaling factors were set to unity, consistent with the development of GLYCAM06. Initial geometries were optimized through 5000 cycles of steepest descent, followed by conjugate gradient energy minimization until the convergence criteria of drms ¼ 0.1 was obtained. MD simulations were carried out in implicit solvent using Hawkins, Cramer, and Truhlar’s generalized Born model, with default AMBER radii.31,32 Each system was then heated from 5 to 300 K in 50 ps. Initial velocities were assigned from a Boltz- mann distribution at 5 K. A 2-fs time step was used to integrate the equations of motion. Temperatures were maintained at 300 K using a weak-coupling algorithm.33 A cutoff of 100 Å was used for nonbonded interactions, thus calculating all nonbonded pairs. Bonds containing hydrogen were constrained to their equilibrium lengths using the SHAKE algorithm.34 Subsequently, force evaluations were computed with bond interactions involving the hydrogen atoms omitted. Families were determined through the use and comparison of phi, psi, and omega histograms, which were generated using 58 bins. Results The Ramachandran-type plots resulting from the disaccharides and oligosaccharide MD simulations are shown in Figures 3 and 4. The figures in the supplementary material display the /-, w-, and x-angles as a function of simulation time for all systems Journal of Computational Chemistry DOI 10.1002/jcc Molecular Dynamic Simulations Using the Glycam06 Force Field Figure 4. (a–c) Phi, psi, and omega carb-Rama plots for the oligosaccharide glycosidic linkages. Journal of Computational Chemistry DOI 10.1002/jcc 913 Journal of Computational Chemistry 89.5 121.8 119.8 121.5 90.7 171.7 2179.9 291.5 83.8 2178.9 87.2 274.8 272.3 2123.5 273.9 276.2 272.9 2109.2 276.7 285.9 (b,2sc,þsc|ap) (b,2sc,2sc) (b,þsc,ap) (b,ap,þsc) (b,2sc,þsc|ap) (b,2sc|ap,þsc) 1?6 (a,2sc,ap,gg) (a,2sc,ap,gt) (a,2sc,2sc,gt) (a,2sc,þsc,gg) (a,2sc,ap,tg) (a,2sc,þsc,gt) 124.3 2121.1 52.6 268.7 2107.2 (b,2sc|ap,þsc) (b,þsc,þsc|ap) (b,2sc,2sc) (b,2sc,þsc|ap) (a,ap,2sc) (a,þsc,þsc) 1?4 79.0 2143.7 289.7 50.4 2123.4 53.6 284.7 (b,2sc|ap,ap) (b,þsc,2sc) (b,2sc,þsc) 1?3 (a,þsc,2sc) 283.2 Psi 273.6 Phi (b,2sc,2sc) 1?2 Linkage 270.0 74.8 79.3 277.2 154.4 67.2 Omega Disaccharides DOI 10.1002/jcc 63 19 5 5 4 4 80 20 100 30 \1 69 100 4 1 \1 94 % Psi Omega 287.1 284.9 274.4 2106.3 278.4 278.4 6 \1 94 \1 \1 100/99 % a-L-Fuc-(1?6)-b-D-GlcNAc 169.2 280.5 178.7 71.9 2100.8 87.2 85.6 280.7 2168.6 164.5 94.5 68.8 87 3 4 4 2 \1 59 41 98 1 \1 \1 44/12 2/1 11/3 b-D-Gal-(1?4)-b-D-GlcNac 122.2/124.7 43/84 271.0 90.3 a-D-Man-(1?3)-b-D-Man 2119.6 289.0 54.2 2110.6/ 2136.9 90.4/88.3 55.5/52.2 119.0/123.0 287.4/267.1 264.9/250.5 b-D-GlcNac(1?4)-b-D-GlcNAc 276.3 116.9 264.3 253.1 66.2 135.4 2169.9 60.6 b-D-Man-(1?4)-b-D-GlcNac 270.0 123.4 2135.7 94.3 274.9/271.0 155.9 95.9 79.4 58.1 283.1 b-D-GlcNAc-(1?2)-a-D-Man 293.0/278.3 287.7/286.1 Phi 12-Mera Table 1. Disaccharides Average Phi, Psi, and Omega Angles Obtained from MD Simulations. 10.9 5 4 10 275 6 13.7 286.5 6 10.8 275 6 11.6 271.4 6 296 6 275 6 287 6 274 275 71.5 6 8.8 75 615 80 6 15 80 100 58.3 6 9.4 280.1 6 12.6 292 6 16 280 Phi 7.4 3 4 11 2155.1 6 24.0 110.7 6 19.4 119 6 15.4 132.2 6 154 6 138 6 146 6 138 139 2120.6 6 16.8 2135 6 15 2116 6 25 2130 290 287.2 6 15.2 297.6 6 22.3 283 6 14 290 Psi Omega 256.6 6 12.3 Experimental 24 X-ray22 197 X-ray22 376 X-ray22 28 X-ray22 NMR21,36 130 X-ray22 NMR7,19,35 8 X-ray22 53 X-ray22 NMR35 # 914 Salisburg et al. • Vol. 30, No. 6 • Journal of Computational Chemistry 8 NMR37–40,42–44 (averaged) 14 X-ray22 29 X-ray22 38 X-ray22 29 X-ray22 NMR35 Population percentages are given for each conformation. When two values are present, the first value comes from the 1?6 branch, whereas the second value come from the 1?3 branch. 2125.1 þ 15.5 68.5 6 13.6 10/8 2143.2/2145.8 7 2140.5 39.5 (a,þsc,ap) 45.7/42.6 90/92 a-D-NeuAc-(2?3)-b-D-Gal 2140.5/2143.1 93 2142.5 236.7 (a,2sc,ap) 1 89.4 276.5 96.3 236.0/236.3 2 \1 1 53.3 95.3 269.4 61.0 74.8 99.2 53.3 86.6 97.7 94.4 14 3 273.3 159.0 (a,þsc,þsc,gg) (a,þsc,ap,tg) (a,þsc,þsc,gt) (a,þsc,þsc,gt) (a,þsc,2sc,gt) 2?3 80.7 77.4 95.3 2178.1 81.4 267.0 91 59.4 6 7.5 243.5 6 14.8 94.0 6 17.5 2139.3 6 17.8 68.5 6 12.3 260.3 6 14.0 66.0 6 12.8 60.6 6 20 2178.4 6 10.0 178.5 6 13.7 2170 6 20 5 \1 67 15 177.8 2173.4 78.5 75.0 (a,þsc,ap,gg) (a,þsc,ap,gt) 271.3 72.2 81.2 65.4 a-D-Man-(1?6)-b-D-Man 169.5 269.3 167.6 69.7 64.7 6 10.4 67.0 6 10.5 70 6 20 Omega % Omega Psi % Omega Psi Phi Linkage Table 1. (Continued) Disaccharides Phi 12-Mera Phi Psi Experimental # Molecular Dynamic Simulations Using the Glycam06 Force Field 915 reported and include the histograms for each angle. Table 1 reports the average /-, w-, and x-angles for the families populated about each glycosidic linkage within the disaccharides and the oligosaccharide. We have adopted the following glycosidic angle definitions: / ¼ O50 # #C10 # #Ox# #Cx, w ¼ C10 # #Ox# #Cx# #Cx21, and x ¼ O6# #C6# #C5# #O5 for all 1?n linkages, whereas for the 2?3 linkage / ¼ O60 # #C20 # #O3# #C3 and w ¼ C20 # #O3# #C3# #C2. The a-L-Fuc-(1?6)-b-D-GlcNAc-OMe disaccharide populates six different families, with 63% of the population clustered around the average angle values of 2748 (/), 1728 (w), and 2708 (x), and 19% clustered around 2768 (/), 1808 (w), and 758 (x). The remaining four families have 5% or less population. Experimentally, this linkage possesses average values of 275.78 6 13.78 (/), 2155.18 6 24.08 (w), and 256.68 6 12.38 (x), as determined from 24 crystal structures,*,22 corresponding to the most populated MD family. The same families are populated in the oligosaccharide, with 87% of the population clustered around 2878 (/), 1698, (w) and 2808 (x). The remaining fiveoligosaccharide families have 5% or less population. The b-D-GlcNAc-(1?4)-b-D-GlcNAc-OMe disaccharide populates a single conformation that is clustered around 2758 (/) and 1208 (w). This corresponds very well to the average experimental values of 275 6 11.68(/) and 119 6 15.48(w), obtained from 376 crystal structures that posses this linkage.y,22 The same family is seen in the oligosaccharide, clustered around 2768(/) and 1178(w). The b-D-Man-(1?4)-b-D-GlcNAc-OMe disaccharide populates two families with 80% of the population clustered around 2728(/) and 1228(w), with the remaining 20% clustered around 21248(/) and 918 (w). An average of 197 crystal structures found this linkage to have average values of 286.58 6 10.88(/) and 110.78 6 19.48(w),22 corresponding to the most populated MD family. Both families are seen in the oligosaccharide, with 59% of the population clustered around 2708(/) and 1238 (w) and 41% clustered around 21368(/) and 948(w). The a-D-Man-(1?6)-b-D-Man-OMe disaccharide populates five families, with 67% clustered around 788(/), 1788(w), and 2718(x), 15% clustered around 758(/), 21738(w), and 728(x), 14% clustered around 818(/), 958(w), and 2738(x), and the remaining families have less than 5% occupancy. Crystallographic studies have clearly seen two families populating this linkage. An average of 38 crystal structures found values of 64.7 6 10.48(/), 2178.4 6 10.08(w), and 260.3 6 14.08(x),22 corresponding to the most populated MD family. An average of 29 crystal structures found angles of 67.0 6 10.58(/), 178.5 6 13.78(w), and 66.0 6 12.88(x),22 corresponding closely to the second most populated MD family. NMR spectroscopy on this linkage found values of 70 6 208(/) and 2170 6 208(/) and 60.6 6 208(x), also corresponding to the second most populated MD family.35 The oligosaccharide MD simulation significantly *Note that the omega angle in ref. 22 was defined as O6# #C6# #C5# #O4, and required the addition of 1208 to convert it to the O6# #C6# #C5# #O5 definition used here. y This particular disaccharide underwent a ring flip during the simulation. The system was resimulated with igb ¼ 5, resulting in a stable 4C1 ring. Thus, for the remainder of the article the results computed with igb ¼ 5 will be used. Journal of Computational Chemistry DOI 10.1002/jcc 916 Salisburg et al. • Vol. 30, No. 6 • Journal of Computational Chemistry populates two of these families, with 91% clustered around 818(/), 948(w), and 2678(x), and 5% clustered around 818(/), 1708(w), and 2698(x). The remaining two families are populated by less than 5%. Interestingly, one of these families is not seen in the disaccharides but is seen experimentally, with an average angle (29 crystal structures) of 59.4 6 7.58(/), 94.0 6 17.58(w), and 68.5 6 12.38(x).22 The a-D-Man-(1?3)-b-D-Man-OMe disaccharide populates a single family clustered around 798(/) and 21078(w) for the entire simulation. This corresponds well with the experimental average values of 71.5 6 8.88(/) and 2120.6 6 16.88(w) obtained from 130 crystal structures,22 as well as NMR values of 75 6 158(/) and 2135 6 158(w), 808(/) and 21308(w), 1008(/) and 2908(w), and 80 6 15(/) and 2116 6 25(w).35,7,19 Nearly the same conformation is populated in the oligosaccharide simulation, with 94% clustered around 798(/) and 21208(w), and a new family that is clustered around 1568(/) and 2718(w) with 6% population. The b-D-GlcNAc-(1?2)-a-D-Man-OMe disaccharide populates four families during the MD simulation, with 94% clustered around 2748(/) and 2838(w), and the remaining three families having less than 5% occupancy. The major family corresponds well to the experimentally averaged values of 280.1 6 12.68(/) and 297.6 6 22.38(w), obtained from 53 crystal structures.22 NMR experiments have determined /- and w-values of 2808(/) and 2908(w), and 292 6 168(/) and 283 6 148(w),35 both corresponding to the most populated MD family. A second experimental family is seen, with average angles of 58.3 6 9.48(/) and 287.2 6 15.28(w) as determined over eight crystal structures.22 This family is observed in the disaccharide simulation with a population of 1%, whose angles have an average value of 548(/) and 2908(w). The b-D-GlcNAc-(1 ? 2)-aD-Man linkage is present twice in the oligosaccharide, each clustered around 2938(/) and 2888(w) (1?6 branch) and 2788(/) and 2868(w) (1?3 branch) for 100% of the simulation. The b-D-Gal-(1?4)-b-D-GlcNAc-OMe disaccharide populates three families during the MD simulation, with 69% clustered around 2698(/) and 1248(w), 30% clustered around 21218 (/) and 908(w), and less than 1% clustered around 538(/) and 1228(w). An average of 28 crystal structures found average angles of 271.4 6 10.98(/) and 132.2 6 7.48(w),22 corresponding to the most populated MD family. In agreement with these findings are five NMR whose //w values include 296 6 5/154 6 3, 275 6 4/138 6 4, 287 6 10/146 6 11, 274/138, and 275/139.21,36 The two most populated families are also observed in both branches of the oligosaccharide. The 1?6 branch possesses 43% of the population clustered around 2758(/) and 1228(w), whereas the 1?3 branch possesses 84% of the population clustered around 2718(/) and 1258(w). For the second family the 1?6 branch possesses 44% of the population clustered around 21118(/) and 908(w), whereas the 1?3 branch possesses 12% of the population clustered around 21378(/) and 888(w). A third family is populated in the oligosaccharide, one that is not seen in the disaccharides, with a population of 11% clustered around 2878(/) and 2658(w) in the 1?6 branch and 3% clustered around 2678(/) and 2508(w) in the 1?3 branch. Finally, the a-D-NeuAc-(2?3)-b-D-Gal-OMe disaccharide populates two families during the MD simulation, with 93% clus- tered around 2378(/) and 21428(w), whereas the remaining family is clustered around 408(/) and 21408(w). In terms of a-DNeuAc-(2?n) linkages, the 2?3 linkage has the most available NMR6,21,37–43 and X-ray experimental data.22 The NMR studies are coupled with computations to provide values for the /- and w-angles that are consistent with the NMR spectra. Eight NMR values yielded a /- and w-values of 243.58 6 14.88 and 2139.38 6 17.88,37–40,42–44 corresponding to the most populated MD conformation. One NMR study yielded a values of 358 and 21088,43 whose /-value corresponds to the less populated MD family. An average of 14 crystal structures found angles of 68.58 6 13.68 (/) and 2 125.18 6 15.58 (w),22 which is questionably populated by the disaccharide MD simulation at 7%. The oligosaccharide simulation populates the same two families seen in the disaccharide. The most abundant family has a population of 90% (1?6 branch) and 92% (1?3 branch) clustered around 2368(/) and 21428(w). The second family has 10% and 8% clustered around 448(/) and 21448(w) in the 1?6 and 1?3 branches, respectively. Discussion Conformational Space of Carb-Rama Plots for Ideal Linkage Angles Recently, da Silva and coworkers presented a nice discussion of the anomeric and exo-anomeric effect in carbohydrates.45 The anomeric effect refers to the thermodynamic preference for the axial (a) position of electronegative groups (e.g., methoxy or carbohydrate) over the beta (b) position at the anomeric C1 atom. Quantifying this is the angle C50 # #O50 # #C10 # #Ox (h), where a has a h-angle of 608 and b has a y-angle of 1808.{ The exo-anomeric effect refers to the preference for conformations about the /-angle. The /-angle may adopt conformations centered around 2608 (2sc, synclinal or gauche), 608 (þsc), and 1808 (ap, antiperiplaner or trans).45,46 Considering only the hand /-angles for two-bond glycosidic linkages (e.g., 1?2, 1?3, 2?3. . .), the relative order in increasing stability, based on QM calculations on 2-methoxytetrahydropyran, is (a, þsc), (b, 2sc), (a, ap), (b, þsc), (b, ap), and (a, 2sc).45,46 Including the wangle results in 18 possible conformations defined by (a, /, w) and (b, /, w).§ Including the x-angle for the 1?6 glycosidic linkage results in 54 possible conformations defined by (a, /, w, x) and (b, /, w, x), whose ideal angles are presented in Table 2. The common nomenclature for the x-angle is gauche–trans (gt), trans–gauche (tg), and gauche–gauche (gg), referring to the x-angle and O6# #C6# #C5# #C4, respectively. The gt, tg, and gg conformations have x-angle values of approximately 608, 1808, and 2608, respectively. As a nomenclature example, (a, þsc, 2sc, gt) indicates a conformation about an a 1?6 linkage that possesses a positive gauche /-angle, a negative gauche w-angle and a gt x-angle. Based on these ideal angles, the carb-Rama plot for any two-bond a-linkages, will have nine regions of { The exception to this is a-D-NeuAc-, where the a conformation possesses a y-angle of 1808 as defined by C60 # #O60 # #C2# #Ox. § A similar type of conformational analysis uses a diamond-lattice to describe the conformational space of glycosidic linkages, and interested readers are referred to refs. 47 and 48 for further information. Journal of Computational Chemistry DOI 10.1002/jcc Molecular Dynamic Simulations Using the Glycam06 Force Field 917 a Table 2. Nomenclature and ideal torsion angles for the 18 and 54 possible conformations for two-and three-bond glycosidic linkages, respectively. y y / w Two-bond glycosidic linkage (e.g. 1?2,1?3, 1?4) (a,þsc,þsc) 60 60 60 (a,ap,þsc) (a,þsc,2sc) 60 60 260 (a,ap,2sc) (a,þsc, ap) 60 60 180 (a,ap,ap) (a,2sc,þsc) 60 260 60 (b,þsc,þsc) (a,2sc,2sc) 60 260 260 (b,þsc,2sc) (a,2sc,ap) 60 260 180 (b,þsc,ap) 60 60 60 180 180 180 180 180 180 60 60 60 60 260 180 60 260 180 Three-bond glycosidic linkage (e.g. 1?6) (a,þsc,þsc,gt) 60 60 60 (a,þsc,þsc,gg) 60 60 60 (a,þsc,þsc,tg) 60 60 60 (a,þsc,2sc,gt) 60 60 260 (a,þsc,2sc,gg) 60 60 260 (a,þsc,2sc,tg) 60 60 260 (a,þsc, ap,gt) 60 60 180 (a,þsc,ap,gg) 60 60 180 (a,þsc,ap,tg) 60 60 180 (a,2sc,þsc,gt) 60 260 60 (a,2sc,þsc,gg) 60 260 60 (a,2sc,þsc,tg) 60 260 60 (a,2sc,2sc,gt) 60 260 260 (a,2sc,2sc,gg) 60 260 260 (a,2sc,sc,tg) 60 260 260 (a,2sc,ap,gt) 60 260 180 (a,2sc,ap,gg) 60 260 180 (a,2sc,ap,tg) 60 260 180 60 60 60 60 60 60 60 60 60 180 180 180 180 180 180 180 180 180 180 180 180 180 180 180 180 180 180 60 60 60 60 60 60 60 60 60 60 60 60 260 260 260 180 180 180 60 60 60 260 260 260 180 180 180 / w x 60 260 180 60 260 180 60 260 180 60 260 180 60 260 180 60 260 180 (a,ap,þsc,gt) (a,ap,þsc,gg) (a,ap,þsc,tg) (a,ap,2sc,gt) (a,ap,2sc,gg) (a,ap,2sc,tg) (a,ap,ap,gt) (a,ap,ap,gg) (a,ap,ap,tg) (b,þscþsc,gt) (b,þsc,þsc,gg) (b,þsc,þsc,tg) (b,þsc,2sc,gt) (b,þsc,2sc,gg) (b.þsc,2sc,tg) (b,þsc,ap,gt) (b,þsc,ap,gg) (b,þsc,ap,tg) y / w (b,2sc,þsc) (b,2sc,2sc) (b,2sc,ap) (b,ap,þsc) (b,ap,2sc) (b,ap,ap) 180 180 180 180 180 180 260 260 260 180 180 180 60 260 180 60 260 180 (b,2sc,þsc,gt) (b,2sc,þsc,gg) (b,2sc,þsc,tg) (b,2sc,2sc,gt) (b,2sc,2sc,gg) (b,2sc,2sc,tg) (b,2sc,ap,gt) (b,2sc,ap,gg) (b,2sc,ap,tg) (b,ap,þsc,gt) (b,ap,þsc,gg) (b,ap,þsc,tg) (b,ap,2sc,gt) (b,ap,2sc,gg) (b,ap,2sc,tg) (b,ap,ap,gt) (b, ap,ap,gg) (b,ap,ap,tg) 180 180 180 180 180 180 180 180 180 180 180 180 180 180 180 180 180 180 260 260 260 260 260 260 260 260 260 180 180 180 180 180 180 180 180 180 60 60 60 260 260 260 180 180 180 60 60 60 260 260 260 180 180 180 x 60 260 180 60 260 180 60 260 180 60 260 180 60 260 180 60 260 180 x 60 260 180 60 260 180 60 260 180 60 260 180 60 260 180 60 260 180 a Two-bond glycosidic linkages are defined by (a, /, w) and (b, /, w), and three-bond glycosidic linkages are defined #O05# #C01# #Ox; / ¼ O05# #C01# #Ox# #Cx; w ¼ by (a, /,w,x) and (b, /,w,x). Torsion definitions: a/b ¼ C05# 0 C# #Ox# #Cx# #Cx-1; x ¼ O6# # C6# #C5# #O5. The abbreviations 2sc,þsc and ap refer to an angle adopting a value that is 2608 (negative synclinal or gauche), þ608 (positive synclinal or gauche) and 1808 (antiperiplaner or trans), respectively. The abbreviations gt, tg and gg refer to gauche-trans, trans-gauche, and gauche-gauche in reference in the x angle and O6# #C6# #C5# #C4, respectively. possible conformational space, with each region enclosed by a 1208 3 1208 (w 3 /) area as shown in Supplementary Figure 1. The same will be true for the carb-Rama plots for any two-bond b-linkages. The three-dimensional carb-Rama plots for any three-bond a- or b-linkage (e.g., 1?6, 2?6) will have 27 regions of possible conformational space, each enclosed by a 1208 3 1208 3 1208 (w 3 / 3 x) volume. However, not all of the conformational regions may be accessible due to energetic restraints, while other regions may be preferentially populated. Finally, in the situations where angles adopt a value that borders two regions, we will indicate this by a ‘‘|’’ in the conformation nomenclature. For example if an MD simulation samples conformations that possess /-angles in the þsc and ap regions, but all of the conformations belong to a single / family, the family would be indicated by (b, þsc, þsc|ap). The angles that fall into this category have angle values that are 1208 6 58 or 21208 6 58. Linkage Conformations from Simulations and Experiment The carb-Rama plots for b-D-GlcNAc-(1?2)-a-D-Man show that the major family populated by the disaccharide and the oligosac- charide is (b, 2sc, 2sc). In the oligosaccharide this linkage behaves similarly in the 1?3 and 1?6 branches. The 1?6 branch linkage has the average /-angle shifted by approximately 2178, whereas the w-angle resembles that of the disaccharide and the 1?3 branch. This shift can be explained by the 1?6 linkage forming multiple contacts with the inner carbohydrates, exemplified in Figure 5a, which induces the shift in its average angle adopted. The disaccharide also has a (b, 2sc|ap, ap) conformation that is populated for 4% of the simulation time. The remaining two conformational families are (b, þsc, 2sc) and (b, 2sc, þsc), both having very low populations (i.e., $1%). The carb-Rama plots for the b-D-GlcNAc-(1?4)-b-D-GlcNAc linkage are very similar for both the disaccharide and the oligosaccharide, adopting essentially 100% of the (b, 2sc, þsc|ap) conformation. Replacing the terminal b-D-GlcNac with b-D-Man, forming b-D-Man-(1?4)-b-D-GlcNAc, opens a nearby (b, 2sc|ap, þsc) conformational space, while retaining (b, 2sc, þsc|ap) as the most abundant family in the disaccharide and oligosaccharide simulations. Interestingly, the carb-Rama plots for the b-D-Man-(1?4)-b-D-GlcNAc linkage shows a 21% population shift between the disaccharide and oligosaccharide simulations. The most abundant family is populated 80% of the time in Journal of Computational Chemistry DOI 10.1002/jcc 918 Salisburg et al. • Vol. 30, No. 6 • Journal of Computational Chemistry Figure 5. The representations of three oligosaccharide conformations that were sampled during the MD simulations, labeled A, B, and C. The blue residues are the carbohydrates belonging to the 1 ? 6 branch, the red represent the carbohydrates on the 1 ? 3 branch, and the traditionally colored residues are the inner core of the oligosaccharides. the disaccharide, which reduces to 59% in the oligosaccharide. Concurrently, the new and second most abundant family goes from a 20% population in the disaccharide to 41% population in the oligosaccharide. The increase in the oligosaccharide’s (b, 2sc|ap, þsc) family is due to the 1?6 branch interacting with the inner core carbohydrates, such that the 1?6 branch lies behind the fucose residue Figure 5b. The other 1?4 linkage, b-D-Gal-(1?4)-b-D-GlcNAc, displays similar traits to those mentioned above, with (b, 2sc, þsc|ap) being the most abundant family and (b, 2sc|ap, þsc) being the second most abundant family. As in the b-D-Man(1?4)-b-D-GlcNAc linkage, there is a shift that occurs between these families. The b-D-Gal-(1?4)-b-D-GlcNAc linkage present in the oligosaccharide’s 1?6 branch accesses a new conformation, (b, 2sc, 2sc), for 11% of the simulation time, which is in addition to the equally populated (b, 2sc, þsc|ap) and (b, 2sc|ap, þsc) families at %44%. This is not the case in the 1?3 branch, where the (b, 2sc,þsc|ap) family increases in population by 15% relative to the disaccharide. The populated (b, 2sc|ap, þsc) and (b, 2sc, 2sc) families have significantly reduced occupancy of 12% and 3%, respectively. The a-D-Man-(1?3)-b-D-Man linkage has the (a, þsc, 2sc) family populated for 100% of the disaccharide simulation, and 94% of the oligosaccharide simulation. The oligosaccharide possesses an additional conformation, (a, ap, 2sc), with a 6% population. This new conformation occurs when the 1?3 branch wraps back to interact with the inner core carbohydrates, forming a stacked topology as exemplified in Figure 5c. The carb-Rama plots for a-D-NeuAc-(2?3)-b-D-Gal indicate that the (a, 2sc, ap) conformation is the most dominate family in both simulations, with 93% population in the disaccharide, and 90% and 92% in the 1?6 and 1?3 oligosaccharide branches, respectively. This family is seen in nearly all NMR studies. The less populated MD family is (a, þsc, ap), with 10% or less occupancy, which is not seen experimentally. However, the closely related family, (a, þsc, 2sc|ap), is seen in X-ray experiments. Of the residues studied here, a-D-NeuAcproved the most challenging to parameterize due to its potential to be ionized in solution and the number of mixed functionalities that govern the /-angle torsion.1 These results suggest that a revisit of the a-D-NeuAc parameterization might be warranted. In all simulations involving the 1?6 linkage, the x-angle predominately samples the gt and gg conformations, with a sampling of less than 4% for the tg conformation. The lack of the tg conformation is understood based on the ‘‘repulsive’’ xangle curve, which will be qualitatively similar to the repulsive curve presented for a-D-glucopyranoside that we presented previously.49 This curve is characterized by minima at gt and gg, whereas none exist around tg due to a repulsive interaction between the O6 and the O4 atoms. As explained previously,49 electrostatic and steric repulsions are the underlying factors in determining the x-angle’s gt and gg preferences when the carbohydrate’s hydrogen bonding is occupied by solvent interactions. For the disaccharides and oligosaccharides, these forces dominate the behavior of x-angle even more because the reducing carbohydrate’s O6 atom is unable to be a hydrogen bond donor, which would marginally stabilize the tg conformation. The carb-Rama plots for the a-L-Fuc-(1?6)-b-D-GlcNAc linkage are similar between the disaccharides and the oligosaccharide, both primarily populating an (a, 2sc, þsc, gg) conformation. Five other families are populated, which include (a, 2sc, ap, gt), (a, 2sc, 2sc, gt), (a, 2sc, þsc, gg), (a, 2sc, ap, tg), and (a, 2sc, þsc, gt). This linkage adopts a single /-angle conformation that is 2sc in both simulations, consistent with the exo-anomeric effect. The (a, 2sc, ap, gg) population occurs with a /-angle near 21078, an approximately 2308 distortion from the other families, and is directly coupled to the adoption of a w-angle and an x-angle of 808 and 2808, respectively. In this conformation, the /-angle distorts to avoid steric hinderence between the # #NHC(¼ ¼O)H and the methyl group attached to fucose’s C5 atom. The (a, 2sc, ap, tg) conformation (4%) occurs when the x-angle is 1608 (tg), violating the gauche effect, and occurs only when / and w adopt values around 2758 and 21708, respectively. The largest a-L-Fuc-(1?6)-b-D-GlcNAc family, (a, 2sc, ap, gg), shows a 24% population increase in the oligosaccharide at the expense of (a, 2sc, ap, gt), the second largest disaccharide family. The change in family preference is due to the x-angle adopting a conformation that allows for more favorable contacts with the 1?6 branch. Journal of Computational Chemistry DOI 10.1002/jcc Molecular Dynamic Simulations Using the Glycam06 Force Field The carb-Rama plots for a-D-Man-(1?6)-b-D-Man show a significant shift in populations when comparing the disaccharides with the oligosaccharide. The largest shift occurs with the (a, þsc, ap, gg) conformation, which is the dominant conformation in the disaccharide (67%) to the second most dominant conformation (5%) in the oligosaccharide. This corresponds to an increase in the third most populated disaccharide conformation, (a, þsc, þsc, gg) (14%), which becomes the most dominant conformation (91%) in the oligosaccharide. Interestingly, the xangle preferentially adopts a gg conformation that is long lived in the oligosaccharide, which is contrary to the frequent sampling observed in the disaccharide simulations. Thus, the addition of the 10 other carbohydrate residues stiffens the x-angle. Comparison of MD Simulations to Experiment The average /-, w-, and x-angles of major families in the MD simulations correspond well with available experimental data. The most populated family in each linkage is experimentally observed in X-ray or NMR spectroscopy. By comparing the average MD value (Table 1) with the average X-ray values determined by Wormald and coworkers,22 the /-, w-, and x-angles are modeled very well, with an average absolute difference from the X-ray of 10.38, 12.68, and 14.08, respectively. The largest /-angle difference occurs in the oligosaccharide for b-D-Man-(1?4)-b-DGlcNAc linkage’s (b, 2sc, þsc|ap) conformation and for a-DMan-(1?6)-b-D-Man linkage’s (b, þsc, ap, gg) conformation, with an absolute difference of 16.58 and 16.38, respectively. The largest w-angle difference, with respect to the average crystal values, occurs in the disaccharide and oligosaccharide a-L-Fuc(1?6)-b-D-GlcNAc linkage (a, 2sc, ap, gg) conformation, with absolute values of 32.98 and 35.98, respectively. The third largest w-angle difference occurs at an absolute value of 14.68 in the disaccharide b-D-GlcNAc-(1?2)-a-D-Man’s linkage. Additionally, the a-D-NeuAc-(2?3)-b-D-Gal’s (a, 2sc, ap), when compared with the average NMR values, has an absolute differences of less than 78. The largest x-angle difference, with respect to the average crystal values, occurs in the disaccharide and oligosaccharide a-L-Fuc-(1?6)-b-D-GlcNAc linkage’s (a, 2sc, ap, gg) conformation, with absolute values of 13.48 and 23.48, respectively. Likewise, the average MD values compare very well with available NMR data,7,19,21,35,36 with an average absolute difference for /- and w-angles of 8.68 and 15.28, respectively. Only one NMR determined x-angle is known, a-D-Man-(1?6)-b-D-Man’s (b,þsc,ap,gg) linkage, and the disaccharide and oligosaccharide simulations are within 11.48 and 9.48, respectively. General Trends b-D-(1?4)-equatorial: In this study, there are three different carbohydrate combinations that belong to this linkage type: b-DGlcNAc-(1?4)-b-D-GlcNAc, b-D-Man-(1?4)-b-D-GlcNAc, and b-D-Gal-(1?4)-b-D-GlcNAc. The two significant conformations that these b-(1?4)-equatorial linkages occupy in the simulations are (b,2sc,þsc|ap) and (b,2sc|ap,þsc). The average of 376 crystal structures that possess a b-D-GlcNAc-(1?4)-D-GlcNAc linkage yields average /- and w-angles of 275.98 6 11.68 and 119.08 6 15.48,22 respectively, placing them in the (b,2sc,þsc|ap) 919 region. The average of 28 crystal structures that possess a b-DGal-(1?4)-D-GlcNAc linkage yields average /- and w-angles of 2718 6 108 and 132.28 6 7.48,22 respectively, placing them in the (b,2sc,ap) region. The average of 197 crystal structures that possess a b- D-Man-(1 ? 4)-D-GlcNAc linkage yields average /- and w-angles of 286.58 6 11.68 and 110.78 6 19.48,22 respectively, placing them in the (b,2sc,þsc) region. NMR spectroscopy on Lactose (b-D-Gal-(1?4)-b-D-Glc) and Cellobiose (b-DGlc-(1?4)-b-D-Glc) yields a /-angle of approximately 2878 and a w-angle of approximately 998, placing these two linkages in the (b,2sc,þsc) region.50,51 The average of two crystal structures that possess b-D-GlcNAc-(1?4)-D-Man linkages provide average /and w-angles of 2170.08 6 10.78 and 94.78 6 6.18,22 respectively, placing them in the (b,ap,sc) region. This data taken together suggest that b-D-(1?4)-equatorial linkages preferentially dominate the (b,2sc,þsc|ap) region, whereas the (b, sc|ap,þsc), (b,2sc,ap), (b,2sc,2sc), and (b,ap,þsc) regions are accessible to a lesser degree. a-D-(1?3)-equatorial: The a-D-Man-(1?3)-b-D-Man linkage preferentially populates the (a,þsc,2sc) region of the carb-Rama plot. There are 130 crystal structures that possess an a-D-Man(1?3)-D-Man linkage whose average /- and w-angles are 71.58 6 8.88 and 2120.68 6 16.88,22 respectively, placing an average crystal structure at (a,þsc,2sc|ap). In addition to this specific linkage, there is NMR data available for Nigerose (a-D-Gal-(1?3)-b-D-Glc), whose conformation is in the (a,þsc,ap) region with a /-angle of 818 and a w-angle of 21438.50 These values suggest a strong a-D(1?3)-equatorial linkage preference for the (a,þsc,2sc) and (a,þsc,ap) regions. a-D-(1?6)-equatorial: This linkage is the most difficult type for which to provide general trends. The a-D-Man-(1?6)-b-DMan linkage populates three significant areas in the carb-Rama plots, which are (a,þsc,ap,gg), (a,þsc,ap,gt), and (a,þsc,þsc,gg). There are 96 crystal structures that possess an a-D-Man-(1?6)-DMan linkage occupying the regions (a,þsc,ap,gg), (a,þsc,ap,gt) and (a,þsc,þsc,gt) whose average /-, w-, and x-angles in region one are 64.78 6 10.48, 2178.48 6 10.08, and 260.38 6 14.08, in region two are 67.08 6 10.58, 178.58 6 13.78, and 66.08 6 12.88, and in region three are 59.48 6 7.58, 94.08 6 17.58, and 68.58 6 12.38,22 respectively. An NMR study on Melibiose (a-D-Man(1?6)-b-D-Man) found /- and w-angles of 76.88 and 21358.50 Based on the simulation results and the known experimental results, the a-D-(1?6)-equatorial linkage preferentially populates the regions (a,þsc,ap,gg), (a,þsc,ap,gt), (a,þsc,þsc,gg), and (a,þsc,þsc,gþ). The MD simulations show that the a-L-Fuc-(1?6)-b-DGlcNAc linkage significantly populates the (a,2sc,ap,gg) and (a,2sc,ap,gt) regions. Fucose posses an L configuration, which causes the preferred /-angle conformation of þ608, as found in the other a-D-(1?6)-equatorial linkages, to become 2608 for this linkage. The three dimensional geometry of these regions is similar for both the a-D-(1?6)-equatorial and a-L-(1?6)-equatorial, with a gauche conformation in the /-angle and a trans conformation in the C20 # #C10 # #O6# #C6 torsion angle. Conclusion In this article, we present the extension of G.N. Ramachandran’s idea of plotting the amino acid phi and psi angles to the glyco- Journal of Computational Chemistry DOI 10.1002/jcc 920 Salisburg et al. • Vol. 30, No. 6 • Journal of Computational Chemistry sidic phi, psi and omega angles formed between carbohydrates. Considering two-bond glycosidic linkages, there are 18 possible conformational regions that can be defined by (a, /, w) and (b, /, w), whereas for three-bond linkages there are 54 possible regions that can be defined by (a, /, w, x) and (b, /, w, x). We have reported the results of an implicitly hydrated molecular dynamics simulation on an oligosaccharide composed of 12 carbohydrate residues, and on the eight constituent disaccharides that compose this oligosaccharide. The Glycam06 force field reproduces known experimental data well, where the average /-, w-, and x-angles of major MD conformational families correspond well with available experimental data. The most populated family for each linkage is experimentally observed in X-ray or NMR spectroscopy. By comparing the average MD values to the average X-ray values determined by Wormald and coworkers,22 the /-, w- and xangles are modeled very well, with an average absolute difference from the X-ray values of 10.38, 12.68, and 14.08, respectively. Three general trends can be seen in the carb-Rama plots and experimental data. First, the b-(1?4)-equatorial linkages preferentially dominate the (b,2sc,þsc|ap) region, whereas the (b,2sc|ap,þsc), (b,2sc,ap), (b,2sc,2sc), and (b,ap,þsc) regions are accessible to a lesser degree. Second, a-(1?3)-equatorial linkages show a strong preference for the (a,þsc,2sc) and (a,þsc,ap) regions. Third, the a-D-(1?6)-equatorial linkages preferentially populate the regions (a,þsc,ap,gg), (a,þsc,ap,gt), and (a,þsc,þsc,gg). The reported simulation results and the carb-Rama plots will be useful for future oligosaccharide studies in determining the factors that cause specific glycosidic linkage conformations to be adopted. We hope that this presentation of carb-Rama plots will aid others in unraveling the complex structure of di-, poly-, and oligosaccharides. Acknowledgments Acknowledgment is made to NSF, NIH and to Hamilton College for support of this work. This project was supported in part by NIH grant, NSF grant, and by NSF Grants as part of the MERCURY highperformance computer consortium (http://mercury. chem.hamilton.edu). References 1. Kirschner, K. N.; Yongye, A. B.; Tschampel, S. M.; GonzálezOuteiriño, J.; Daniels, C. R.; Foley, B. L.; Woods, R. J. J Comput Chem 2008, 29, 622. 2. Ramachandran, G. N.; Ramachandran, C.; Sasisekharan, V. J Mol Biol 1963, 7, 95. 3. Ramachandran, G. N.; Sasisekharan, V. Adv Protein Chem 1968, 23, 283. 4. Pérez, S.; Gautier, C.; Imberty, A. In Oligosaccharides in Chemistry and Biology: A Comprehensive Handbook; Ernst, B.; Hart, G.; Sinay, P., Eds.; Wiley/VCH: Weinheim, Germany, 2000; p. 969. 5. Moothoo, D. N.; Naismith, J. H. Glycobiology 1998, 8, 173. 6. Cooke, R. M.; Hale, R. S.; Lister, S. G.; Shah, G.; Weir, M. P. Biochemistry 1994, 33, 10591. 7. Homans, S. W.; Pastore, A.; Dwek, R. A.; Rademacher, T. W. Biochemistry 1987, 26, 6649. 8. Kuttel, M. M.; Naidoo, K. J. J Phys Chem B 2005, 109, 7468. 9. French, A. D.; Kelterer, A. M.; Johnson, G. P.; Dowd, M. K.; Cramer, C. J. J Mol Graph 2000, 18, 95. 10. Schnupf, U.; Willett, J. L.; Bosma, W. B.; Mornany, F. A. Carbohydr Res 2007, 342, 2270. 11. Stortz, C. A. Carbohydr Res 2006, 341, 2531. 12. da Silva, C. O.; Nascimento, M. A. C. Theor Chem Acc 2004, 112, 342. 13. da Silva, C. O.; Nascimento, M. A. C. Carbohydr Res 2004, 339, 113. 14. Stortz, C. A.; Cerezo, A. S. J Carbohydr Chem 2003, 22 (3–4), 217. 15. Stortz, C. A.; Cerezo, A. S. Carbohydr Res 2002, 337, 1861. 16. Ha, S. N.; Madsen, L. J.; Brady, J. W. Biopolymers 1988, 27, 1927. 17. Campen, R. K.; Verde, A. V.; Kubicki, J. D. J Phys Chem B 2007, 111, 13775. 18. da Silva, C. O. Theor Chem Acc 2006, 116 (1–3), 137. 19. Brisson, J. R.; Carver, J. P. Biochemistry 1983, 22, 3671. 20. Brisson, J. R.; Carver, J. P. Biochemistry 1983, 22, 1362. 21. Poppe, L.; Brown, G. S.; Philo, J. S.; Nikrad, P. V.; Shah, B. H. J Am Chem Soc 1997, 119, 1727. 22. Wormald, M. R.; Petrescu, A. J.; Pao, Y. L.; Glithero, A.; Elliott, T.; Dwek, R. A. Chem Rev 2002, 102, 371. 23. Höög, C.; Widmalm, G. Arch Biochem Biophys 2000, 377, 163. 24. Landersjö, C.; Stevensson, B.; Eklund, R.; Östervall, J.; Söderman, P.; Widmalm, G.; Maliniak, A. J Biomol NMR 2006, 35, 89. 25. Lins, R. D.; Hünenberger, P. H. J Comput Chem 2005, 26, 1400. 26. Mikkelsen, L. M.; Hernáiz, M. J.; Martı́n-Pastor, M.; Skrydstrup, T.; Jiménez-Barbero, J. J Am Chem Soc 2002, 124, 14940. 27. Almond, A. Carbohydr Res 2005, 340, 907. 28. Petrescu, A. J.; Petrescu, S. M.; Dwek, R. A.; Wormald, M. R. Glycobiology 1999, 9, 343. 29. Frank, M.; Lütteke, T.; von der Lieth, C. W. Nucleic Acids Res 2007, 35, D287. 30. Lütteke, T.; Bohne-Lang, A.; Loss, A.; Goetz, T.; Frank, M.; von der Lieth, C. W. Glycobiology 2006, 16 (5), 71R–81R. 31. Feig, M.; Onufriev, A.; Lee, M. S.; Im, W.; Case, D. A.; Brooks, C. L. J Comput Chem 2004, 25, 265. 32. Hawkins, G. D.; Cramer, C. J.; Truhlar, D. G. J Phys Chem 1996, 100, 19824. 33. Berendsen, H. J. C.; Postma, J. P. M.; Vangunsteren, W. F.; Dinola, A.; Haak, J. R. J Chem Phys 1984, 81, 3684. 34. Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. J Comput Phys 1977, 23, 327. 35. Homans, S. W. In Biomolecular NMR Spectroscopy; Evans, J. N. S., Ed.; Oxford University Press: Oxford, UK, 1995. 36. Harris, R.; Kiddle, G. R.; Field, R. A.; Milton, M. J.; Ernst, B.; Magnani, J. L.; Homans, S. W. J Am Chem Soc 1999, 121, 2546. 37. Miyazaki, T.; Sato, H.; Sakakibara, T.; Kajihara, Y. J Am Chem Soc 2000, 122, 5678. 38. Wu, W. G.; Pasternack, L.; Huang, D. H.; Koeller, K. M.; Lin, C. C.; Seitz, O.; Wong, C. H. J Am Chem Soc 1999, 121, 2409. 39. Ichikawa, Y.; Lin, Y. C.; Dumas, D. P.; Shen, G. J.; Garciajunceda, E.; Williams, M. A.; Bayer, R.; Ketcham, C.; Walker, L. E.; Paulson, J. C.; Wong, C. H. J Am Chem Soc 1992, 114, 9283. 40. Sabesan, S.; Duus, J. Ø.; Fukunaga, T.; Bock, K.; Ludvigsen, S. J Am Chem Soc 1991, 113, 3236. Journal of Computational Chemistry DOI 10.1002/jcc Molecular Dynamic Simulations Using the Glycam06 Force Field 41. Scarsdale, J. N.; Prestegard, J. H.; Yu, R. K. Biochemistry 1990, 29, 9843. 42. Breg, J.; Kroonbatenburg, L. M. J.; Strecker, G.; Montreuil, J.; Vliegenthart, J. F. G. Eur J Biochem 1989, 178, 727. 43. Poppe, L.; Dabrowski, J.; Vonderlieth, C. W.; Numata, M.; Ogawa, T. Eur J Biochem 1989, 180, 337. 44. Acquotti, D.; Poppe, L.; Dabrowski, J.; Vonderlieth, C. W.; Sonnino, S.; Tettamanti, G. J Am Chem Soc 1990, 112, 7772. 45. Bitzer, R. S.; Barbosa, A. G. H.; da Silva, C. O.; Nascimento, M. A. C. Carbohydr Res 2005, 340, 2171. 921 46. Tvaroska, I.; Carver, J. P. J Phys Chem 1994, 98, 9477. 47. Gagnaire, D.; Pérez, S.; Tran, V. Carbohydr Res 1980, 78, 89. 48. Goekjian, P. G.; Wei, A.; Kishi, Y. In Carbohydrate-based Drug Discovery; Wong, C. H., Ed.; Wiley/VCH: Weinheim, Germany, 2003; p. 305. 49. Kirschner, K. N.; Woods, R. J. Proc Natl Acad Sci USA 2001, 98, 10541. 50. Cheetham, N. W. H.; Dasgupta, P.; Ball, G. E. Carbohydr Res 2003, 338, 955. 51. Hayes, M. L.; Serianni, A. S.; Barker, R. J Am Chem Soc 1982, 119, 1336. Journal of Computational Chemistry DOI 10.1002/jcc
© Copyright 2026 Paperzz