Computer modeling peptides with

With this level of precision, what remains in accuracy is therefore the underlying atomic model choice (Brooks, Karplus, and Pettitt 1988). Using our -replica exchange, free energy method, one computed model peptide backbone unit transfer free energy to trimethylamine N-oxide (TMAO) solution of -54 cal/mol/monomer (Figure 12.4c) compares quite favorably with -43 cal/mol/monomer determined experimentally (Anton and Bolen 2004). 

Fig. 5. Comparison of ab initio, DFT/BPW91/6-31G -computed IR and VCD spectra over the amide I, II, and III regions for model peptides (of the generic sequence Ac-Alaw-NHCH3). These are designed to reproduce the major structural features of an o -helix (top left, n— 6, in which the center residue is fully H-bonded), a 3i helix (PLP Il-like, top right, n— 4), and an antiparallel /1-sheet (n= 2, 3 strands, central residue fully H-bonded) in planar (bottom left) and twisted (bottom right) conformations. The computations also encompass all the other vibrations in these molecules, but those from the CH3 side chains were shifted by H/D exchange (CH3) to reduce interference with the amide modes.

Another, promising avenue to understand silk protein conformation and assembly is the use of model peptides. Although not recent (Fraser and MacRae, 1973 Lotz et al., 1974), studies of silk-based peptide from chemical synthesis, DNA recombinant technology, and computer simulation (Anderson et al., 1994 Asakura et al., 2003 Fahnestock et al., 2000 Fossey et al., 1991 Heslot, 1998 Kaplan, 1998 Wilson et al., 2000) have shown that selected repeats of silk proteins can be transformable hydrogels, elastomers, or regular thermoplastics and that with a proper design they can function as diverse molecular machines (Altman et al., 2003 Heslot, 1998 Kaplan, 1998 Urry, 1998). 

This approach was later extended to off-lattice models and a more detailed description of the transfer energy of the different amino acid residues [77]. Magainin, melit-tin, and several other amphipathic peptides were simulated. In these simulations, differences in the interaction of the peptides with the lipid phase were observed. For example, magainin only showed adsorption onto the lipid and no crossing of the lipid occurred, whereas melittin crossed the lipid and formed a stable transmembrane helix. These results are in full agreement with later studies reported by other research groups presented below, involving more elaborate simulation protocols and representations of the peptides and the lipid. These examples show the potential of computer simulations even when some simplifications have to be made to make the system computationally tractable. 

An alternative computational strategy to study peptides in solution is therefore to benchmark a fast QM or polarizable MM model [51] with respect to accurate QM data in the gas phase, and to apply this model for the study of peptides in solution. SCC-DFTB has been shown to reasonably reproduce higher level calculations with respect to relative energies, structures [23,34,35] and vibrational properties of different conformers of small polypeptides [44,52], Since the combination of different amino acid residues allows for an overwhelming variety of structures with very different properties, structures with repeated residues like /V-acclyl L-alaninc A-methylamide (Ace-Lala -NME) constitute a good starting point for systematic studies, which will be discussed in the following. 

Studies of proteinase activities comprise some of the most important current research efforts in the field of theoretical enzyme mechanisms. Results from crystallography and kinetics in the 70 s and 80 s paved the way for such theoretical studies, mainly of the serine proteinase family. Such studies are extending nowadays, as more structures of proteinases are solved with high resolution and more detailed kinetic studies are conducted. But, while earlier structural results were available for the native structures alone, recent crystallographic evidence is available for complexes with peptide analogs, with intermediate analogs and with mutant enzymes. When these structural studies are coupled with results of kinetic research, a large database is formed for the theoretician to consider as a basis for construction, simulation and analysis by computer models. 

Based on the experimental data, a computational model of OFQ/N bound to ORLl receptors has been proposed (189). In this model the N-terminal sequence containing the two Phe residues binds in a highly conserved pocket formed by TM3, 5, 6, and 7, which is similar to that proposed for opioid receptors. Residues 5-7 (Thr-Gly-Ala) of OFQ/N are then positioned at the TM-EL2 interface in a largely nonconserved region unfavorable side-chain interactions in this region of the receptor are then used to explain the selectivity of the ORLl receptor for OFQ/N over Dyn A. The positively charged C-terminus of the peptide is proposed to make multiple contacts with the highly acidic EL2. 

---