## All Dissertations

8-2007

Dissertation

#### Degree Name

Doctor of Philosophy (PhD)

Bioengineering

Latour, Robert A

Boland , Thomas

Bruce , David

Stuart , Steven

#### Committee Member

Vertegel , Alexey

#### Abstract

%Abstract
The objectives of this project are 1) the setup and refinement of peptide-SAM surface model systems with explicit water using CHARMM and 2) the development of up-to-date simulation protocols for the accurate calculation of adsorption free energy by incorporating the recent development of non-Boltzmann sampling methods in molecular dynamics and applying it for the calculation of adsorption free energy of short peptides onto well-characterized self-assembled monolayers, which is important for understanding protein/surface interactions.
A software package called the Simulation Template Engine for Peptides at Surfaces (STEPS) was developed for the fulfillment of the first objective. It facilitates the automatic setup of a ternary model system consisting of a nine residue peptide, a well-defined SAM surface and a water box with explicit solvent buffered with $Na^+$ and $Cl^-$ ions at about 133mM concentration to represent a physiological environment. Peptides can be modeled with three sequence types TGTG-X-GTGT, GGGG-X-GGGG, or GSGS-X-SGSG, with two capping schemes including non-capped zwitterions and patching by acetylation and amination, and with four different types of secondary structure conformations including {$\alpha$}-helix, {$\beta$}-sheet,random coil and $3_{10}$-helix. SAMs can be modeled with nine types of surface functional groups on alkanethiol chains, two configurations (double-ended or single-ended), three geometrical configurations (hexagonal, rhomboid or square). The water box can be modeled with four geometrical shapes including square, rhomboid, hexagonal and spherical to any tailored dimension. STEPS has the potential to generate thousands of different ternary model systems for peptide/SAM surface interaction simulations, then greatly facilitating molecular simulation studies with this type of molecular systems.
A software package called the Toolkit for Advanced Sampling with Non-Boltzmann (TASNoB) methods was also designed to facilitate the second objective. TASNoB includes one replica exchange molecular dynamics (REMD) facility, which offers both regular REMD and replica exchange with solvent tempering (REST), together with all the post-processing utilities for data analysis; one adaptive umbrella sampling (AUS) facility for doing AUS over surface separation distance (SSD); and one windowed umbrella sampling (WUS) facility for doing conventional windowed umbrella sampling by restraining the system at different SSD positions.
A model system consisting of a short peptide (GGGGKGGGG) and a COOH/COO${^-}$ SAM surface with TIP3P explicit water box buffered with $Na^+$ and $Cl^-$ to 133 mM with square cross-section and variable height can be easily generated with STEPS and conventional molecular dynamics, WUS, regular REMD, or biased REMD can be performed with TASNoB to determine the adsorption free energy of the above-described systems by generating the potential of mean force (PMF) profile of the peptide over the surface as a function of surface separation distance (SSD). Sampling efficiency was compared between different simulation protocols. A conventional MD simulation was performed at temperature 310K for 5 nanoseconds. This simulation showed the sampling problem both along the most interesting degree of freedom, which was SSD, and the configuration space of the ${\phi-\psi}$ dihedral angles of the middle residue of the peptide. WUS was performed with 26 windows based on SSD position with interval of 1.0 \AA\ spanning from 4.0 \AA\ to 29.0 \AA. Each window was simulated for 1.0 nanosecond and a PMF profile was obtained by weighted histogram analysis method (WHAM). WUS overcomes the sampling problem along the SSD, but not the peptide conformation. Regular REMD was then performed with 24 replicas spanning temperature range from 300 K to 400 K with exchanges attempted every 1 picosecond, which provided a 20\% exchange acceptance ratio during a 5.0 nanosecond simulation. Regular REMD overcomes the sampling problem of the peptide configuration space, but the improvement of sampling along SSD is relatively limited. To Address this limitation, biased REMD simulation was performed with a biasing potential obtained from WUS and the REMD was configured in the same way as the regular REMD above. With this simulation, the peptide was replaced with single sodium ion to make the SSD more accurately defined, thus enabling the resulting PMF profile to be compared with the PMF vs. SSD profile obtained from WUS. The results show that the novel combination of WUS with biased REMD provides significant improvement of the sampling efficiency for this type of molecular system both over SSD and the conformations of peptide, compared to WUS, regular REMD, or conventional MD alone.
This work serves to develop and validate a method for the simulation of protein-surface interactions that enables adsorption free energy to be properly calculated for comparison with experimental results. These types of comparisons can now be used to evaluate, modify, and validate an empirical force field for protein surface interactions.

pyscript.tar.gz (305 kB)

COinS