G. Haran et al., MOLECULAR-DYNAMICS SIMULATIONS OF SIMPLE PEPTIDE MODELS - SOLVENT EFFECTS AND COMPARISON WITH EXPERIMENT, Journal of physical chemistry, 98(40), 1994, pp. 10294-10302
The dynamics of simple models for peptide chains were studied by molec
ular dynamics simulations, Three models were used: a polyglycine-like
model, with a united atom representation of both hydrogen atoms and ca
rbonyl oxygens (PG), a polyglycine-like model, with an explicit repres
entation of carbonyl oxygens (PGO), and a polyalanine-like model (PA).
The peptide chains were simulated in a periodic box filled with a sof
t-sphere solvent. A series of simulations with varying solvent densiti
es were conducted with a PG five-residue chain. The viscosity dependen
ce of dynamics was studied by calculating bond vector and end-to-end d
istance (EED) autocorrelation functions. These functions showed a nonm
onotonic dependence on solvent viscosity, reminiscent of Kramers-type
dynamics. The EED autocorrelation function at low viscosity contained
oscillations, interpreted as the signature of an underdamped vibration
in the polymer chain. At higher solvent viscosities, this vibration w
as overdamped. The chain length dependence of structure and dynamics w
as determined from simulations with varying peptide lengths, carried o
ut with all three models. According to EED distribution functions obta
ined from the simulations, the pc chains were more expanded than the o
thers; this was attributed to the strong, unrealistic repulsive intera
ctions between neighboring united carbonyl atoms in the PG model. The
experimental results of Haas et al. (Biopolymers 1978, 17, 11-31) were
compared with these results. The rms EEDs obtained from the experimen
tal results were somewhat larger than the simulated distributions. Thi
s could be explained on the basis of the structural difference between
the peptides used in the experiment and the simulated chains. The EED
dynamics were shown to be nonexponential in the case of PGO and PA ch
ains. Bond-bond cross-correlation functions were used to deduce an app
roximate speed for the propagation of conformational changes along the
chain. This speed was significantly smaller in a PA chain than in a P
G chain, a consequence of the larger inertia of the PA chain as compar
ed to the pc chain, which lacks side chains.