五种典型力场对气相中不同末端氨基酸的表现.docx

1、五种典型力场对气相中不同末端氨基酸的表现Performances of five representative force fields on gaseous amino acids with different terminiXin Chen1, Zijing Lin1,2*1 Hefei National Laboratory for Physical Sciences at Microscale & CAS Key Laboratory of Strongly-Coupled Quantum Matter Physics, University of Science and Techno

2、logy of China, Hefei 230026, China2 Key Laboratory of Materials Physics, Institute of Solid State Physics, Chinese Academy of Sciences, Hefei 230031, China*Corresponding author: Zijing Lin; Tel +86-551-63600345; Email: zjlinAbstract:There is a growing interest in the study of structures and properti

3、es of biomolecules in gas phase. Applications of force fields are highly desirable for the computational efficiency of the gas phase study. To help the selection of force fields, the performances of five representative force fields for gaseous neutral, protonated, deprotonated and capped amino acids

4、 are systematically examined and compared. The tested properties include relative conformational energies, energy differences between cis and trans structures, the number and strength of predicted hydrogen bonds, and the quality of the optimized structures. The results of BHandHLYP/6-311+G(d,p) are

5、used as the references. GROMOS53A6 and ENCADS are found to perform poorly for gaseous biomolecules, while the performance of AMBER99SB, CHARMM27 and OPLSAA/L are comparable when applicable. Considering the general availability of the force field parameters, CHARMM27 is the most recommended, followed

6、 by OPLSAA/L, for the study of biomolecules in gas phase.Keywords: Conformation, Relative Energies, Correlation Coefficient, Hydrogen Bond, Molecular Mechanics1. IntroductionEmpirical molecular force fields are routinely used in the investigations of structure-property-activity relationships in biol

7、ogical systems as the computational cost of treating these systems quantum mechanically is often unbearably high. The force fields are in general derived by fitting parameters to data from quantum chemical calculations or experiments on model molecules that may mimic the properties of the interested

8、 biomolecules. Fitting the experimental data for the condensed-phase environment is emphasized. This is reasonable as the structures and properties of biomolecules in solution or condensed-phase environment are of the most interest. However, this also means that the accuracy of the force field to pr

9、edict the relevant properties in the gas phase is sacrificed. It also represents a paradox in the fitting philosophy as the quantum chemistry results for the gas phase are indispensible for the force field parameterization 1. The limitation of force fields for the gas phase study is often dismissed

10、as irrelevant. Indeed, impressive progresses have been made in utilizing force fields for the study of biological systems, e.g., simulating the protein dynamics 2-7. However, the contradiction inherent in the fitting philosophy has some serious consequences. As there is no rigorous way to determine

11、the optimal parameter set, many force field variants are proposed based on different emphases of the fitting targets. AMBER 8, CHARMM 9, ENCAD 10, GROMOS 11 and OPLSAA 12 are examples of force fields that are widely in use. There have been numerous articles and reviews discussing about the performan

12、ces of modern force fields for biomolecules in solution or condensed-phase environment 4, 5, 7, 8, 13. These studies show clearly that a proper choice of the force field is dependent on the research subject of interest. Comparative studies of the force fields are crucial for the proper selection of

13、force field, but systematic studies on the performance comparison for objects in gas phase are rare. The limitation of force fields for the gas phase study had been dismissed as irrelevant. However, there is a growing interest and effort in studying the gaseous biomolecules that are free from the co

14、mplication caused by the complex solute-solvent interactions 14-20. Consequently, a systematic comparison of the performances of the force fields for biomolecules in gas phase is becoming increasingly meaningful. The comparison is helpful for the choice of force field that is necessary for many gas

15、phase studies, e.g., MD runs of biomolecule with about 100 or more atoms. Even when the force field is used as a pre-screening tool, the comparative study is also helpful for avoiding using a force field that provides misguided information about the potential energy surface. Moreover, the comparativ

16、e study is the starting point for learning the accuracies of the state-of-the-art force fields for the gas phase study. Such information is also helpful for knowing how much the improvement of force field is required to provide results for the gas phase study with a desirable accuracy. The informati

17、on is useful for guiding the future development of force field. The choice of testing objects in this study is guided by some general considerations. First, protein force field parameters are mainly determined by fitting data for capped alanine, capped glycine and capped proline 13, 14, 21-24. Other

18、 amino acid residues play only minor role in the parameter determination. The approach seems to work well for proteins in condensed-phase environment 2, 5, 7, 14, 18, partly due to the fact that most attention is paid to the backbone structures. For short peptides, the influence of the side chain on

19、 the backbone is expected to be substantial and how well the force fields work for other amino acid residues requires detailed examinations. Second, the overall influence of the amino and carboxyl terminal groups is relatively small for large molecules such as proteins, but the corresponding influen

20、ce for short peptides can be significant. Short peptides are a class of biomolecules with important biological and physiological roles. They are also the starting point for the peptide-based drug design study 25. The amino and carboxyl groups are important in the force field performance study. Third

21、, proton transfer is critically important for numerous biological processes. It is therefore highly meaningful to include the protonated amino group and deprotonated carboxyl group in the testing object set. Therefore, the force field performances are tested here for a large number of gaseous amino

22、acids with natural, protonated, deprotonated and capped termini. Force fields are meant to reproduce the structural and energetic information obtained by experiments, preferably, or quantum chemistry calculations. As the experimental data are very limited, comparing the force fields with the quantum

23、 chemistry calculations are more convenient and, possibly, statistically more meaningful. In fact, the best one may hope for is that the results by a force field are close to that by quantum chemistry calculations. Therefore, the quantum chemistry results may be used as the references to test the pe

24、rformances of different force fields. Notice that there are a number of quantum chemistry methods and the density functional theory (DFT) based approach is favored for its accuracy and computational efficiency. Among the DFT variants tested for all neutral, protonated and deprotonated amino acids, t

25、he BHandHLYP/6-311+G(d,p) method has been shown to produce the best energetic results statistically 26. When benchmarked with the CCSD/6-311+G(d,p) results, the statistical quality of the BHandHLYP results is even slightly better than that of the MP2 computations 26. Consequently, the force fields a

26、re benchmarked with the BHandHLYP results. 2. MethodThe abilities of five force fields, AMBER, CHARMM, ENCAD, GROMOS and OPLSAA, to mimic the quantum chemistry, BHandHLYP/6-311+G(d,p) energetic and geometric results are tested with four representative properties: 1) relative conformational energies,

27、 2) energy differences between cis and trans structures, 3) number of predicted hydrogen bonds (H-bonds), 4) structural similarity as measured by the root-mean-square-difference (RMSD). The five force fields have various versions of parameterizations and their relatively new parameterization sets, A

28、MBER99SB 27, CHARMM2721, GROMOS53A6 28, ENCADS 29, OPLSAA/L 30, are used in the test. 19 naturally occurring amino acids including glycine (Gly or G), alanine (Ala or A), valine (Val or V), leucine (Leu or L), isoleucine (Ile or I), asparagine (Asn or N), glutamine (Gln or Q), cysteine (Cys or C), m

29、ethionine (Met or M), serine (Ser or S), threonine (Thr or T), proline (Pro or P), tyrosine (Tyr or Y), tryptophan (Trp or W), phenylalanine (Phe or F), histidine (His or H), lysine (Lys or K), aspartic acid (Asp or D) and glutamic acid (Glu or E) are considered in the testing set of the current stu

30、dy. Four terminus forms of amino acids, natural, protonated, deprotonated and capped, are used in the testing. In caped amino acids, the N- and C-termini are capped with the acetyl and N-methylamine groups, respectively. To illustrate, the structures of natural, protonated, deprotonated and capped a

31、lanine are sketched in Figure 1.To be of high statistical significance, the conformations of these amino acids are obtained though systematic searches by considering all combinations of bond rotational degrees of freedom in the trial structure generations, as described in literatures 31. The QM stru

32、ctures are determined at the BHandHLYP/6-31G* level. The numbers of unique conformations thus obtained for these molecules are shown in parentheses as the followings. For natural, protonated, deprotonated and capped amino acids, they are respectively Ala (8, 3, 2, 6), Asn (62, 9, 11, 71), Asp (106, 26, 24, 20), Cys (78, 5, 12, 49), Gln (59, 8, 14, 138), Glu (340, 10, 16, 72), Gly (16, 3, 1, 4), His (57, 12, 8, 56), Ile (85, 33, 19, 62), Leu (96, 12, 23, 68), Lys (186, 13, 51, 427), Met (172, 15, 31, 190), Phe (31, 10, 8, 32), Pro (20, 6, 7, 10), Ser (52, 12, 19, 49), Thr (71, 12, 7, 52), Trp