1. Insights into All-Atom Protein Structure
Prediction via in silico Simulations
2013 Sigma Xi Student Research Showcase
Daniel Wang
2. RESEARCH GOALS
To utilize in silico methods to perform de
novo simulations of protein folding
pathways and predict the functional
structures of proteins.
3. BACKGROUND
Proteins – Biological Workhorses
The central dogma of biology A map of 3200 protein interactions between 1700 proteins
(Image from http://www.nyu.edu/ (Image from http://www.mdc-berlin.de/en/news/archive/2008/20080910-erwin_schr_dinger_prize_2008_goes_to_resea/index.html)
projects/vogel/Pics/centraldogma_2
Proteins serve a plethora of vital functions: growth and repair, cell-to-cell signaling,
defense against pathogens, movement, catalyzing reactions
~130,000 binary protein-protein interactions in a human cell at any given time
Protein function is determined by specific 3-dimensional structure
4. Protein Folding Problem
Random Coil 3-dimensional
Structure (Image from http://www.ks.uiuc.edu/villin-folding-process) native structure
Proteins gain specific functions through folding, a poorly understood process in which
a chain of amino acids assembles into a specific three-dimensional structure
No current method exists to predict the functional structure of a protein from its
amino acid sequence
The protein folding process has remained a mystery to biochemists for several
decades. Understanding this process would allow for:
Greater insight into protein function
Clues into how proteins may misfold and aggregate to cause a range of
diseases, such as Alzheimer’s and Parkinson’s
5. Folding Funnel Model
Modern folding model = energy
of a protein with respect to
systemic changes in geometry and
is represented by funnel-shaped
energy landscapes
Protein chain must negotiate
multiple folding pathways with
valley traps and mountain barriers
Conformational entropy that is
lost during the folding process is
compensated by an increase in
free energy as the global
minimum is approached
Thermodynamic protein folding funnel
(Image from http://www.learner.org/courses/physics/visual/visual.html?shortname=funnel)
6. Molecular Dynamics Simulations
Schematic of molecular dynamics steps Implicit molecular dynamics environment
(Image from http://www.yasara.org/benchmarks.htm)
MD simulations calculate the physical movements of atoms in a system over a
period of time, known as a trajectory.
Timesteps in the femtosecond (10-15 of a second) scale, MD simulations offer
insight into intra- and inter-molecular interactions at an atomistic level
7. Replica Exchange MD
Schematic of replica exchange molecular dynamics
Allows for larger conformational searches by utilizing independent realizations of a
system, known as replicas.
Each replica is coupled to a different thermostat temperature. Replicas are exchanged
at regular time intervals, effectively allowing conformations to escape low temperature
kinetic traps by “jumping” to alternate minima being sampled at higher temperatures
8. METHODOLOGY
Selection of Prototypical Peptides
Rationale: Model protein systems were selected to represent a
variety of structural motifs ubiquitous to all proteins, including
• α-helices,
• β-pleated sheets
• globular shape
10. Construction of Peptide Sequences
• Atomic coordinates for peptides
obtained from Protein Data Bank
as nuclear magnetic resonance or
X-ray crystallography data
• Structures evaluated by MolProbity
structure evaluation program
to perform steric adjustments
11. Equilibration Protocol
Successive rounds of energy minimization and molecular dynamics
with decreasing restraints allows the system to transition from the
experimental environment to the simulation environment
1 | Energy 3 | Energy
minimization 2 | Langevin 4 | Langevin
Dynamics Minimization
System is restrained Dynamics
System is slowly Backbone is restrained
but atoms added by System is slowly
heated to appropriate but side chains are free
Molprobity and LEaP heated to appropriate
temperature to move
are allowed to move. temperature
7 | Production MD Run 6 | Langevin 5 | Langevin
Unrestrained all-atoms molecular dynamics Dynamics Dynamics
to assess system stability in simulation Lower restraints Lower restraints
environment
12. All-Atom Molecular Dynamics
Simulations and analyses were carried out using the Assisted
Model Building with Energy Refinement (AMBER)
o prmtop: input molecular topology,
1. tleap force field parameters
Preparation o inpcrd: input coordinate file and
module velocities
o mdin: input control data for the
minimization/simulation run
o GB8 implicit solvent model
13. o Determine the number of replicas
2. AMBER and temperatures for each replica
Simulation
Engine
using tslop3 REM temperature
program
o Submit job script to NICS Kraken
o Energy and coordinate trajectory
file is written over simulation time
(minimum 100ns for all systems)
14. o Extract temperature replica
3. ptraj
Post- trajectories
trajectory o Backbone root mean square
analysis deviation (RMSD) analysis: a
quantitative comparison
between a representative
structure of a folded cluster and
the native structure
15.
16. RESULTS
Equilibration produces stable structures
trp-cage K19
Backbone RMSD generally increases via a
chignolin “plateau” profile, alternating between periods
of stabilization and spikes from imposed
positional restraints
RMSD <1.5Å during unrestrained molecular
dynamics indicates that all three structures are
stable in the simulation environment
Peptides are stable in the simulation
environment
ff12SB force field and solvent parameters are
ideal for