Transcript PDP Nano TEM - George Mason University

Engineering Proteins for
Enhanced Activity
Steven Fairchild
George Mason University
November 4, 2008
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
The MITRE Corporation
• MITRE is a private, independent, not-for-profit organization,
chartered to work in the public interest
• Founded in 1958 to provide engineering and technical
services to the U.S. Air Force
• Currently manages three Federally Funded Research and
Development Centers – for the Department of Defense, the
Federal Aviation Administration, and the Internal Revenue
Service
• Supports a broad and diverse set of sponsors within the U.S.
government as well as internationally
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Biotechnology Work at MITRE
•
•
•
•
•
•
•
•
Biosensors
Bioforensics
Data mining
Disease surveillance
Epidemiological modeling
Genetic database management
Protein engineering
Synthetic biology
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Cellular Engineering
• Create cells that can
detect and respond to
certain chemicals
• Modify binding sites on
receptor proteins
• Integrate into signaling
pathways
• Couple to desired
response genes
• Example:
Bioremediation
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Protein Engineering
• Goal: build a protein with specific functionality
– Bind a particular molecule
– Catalyze a chemical reaction
• Start with a protein that has similar function
• Identify atoms required for activity
– Computational analysis
– Experimental studies
• Determine what modifications will create the
desired functionality
• Experimental verification
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Unique Protein Structures
High Free
Energy
Unique Structure
at Physiological
Temperature
Low Free
Energy
Phys. Biol. 2 (2005), S44-S55
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Determining a
Protein’s Structure
• Protein engineering requires an initial model
for the protein’s tertiary structure
• Experimental procedures (X-ray and NMR)
are most accurate but are time consuming
• Homology modeling is a good alternative
– Accurate structures even with low sequence
identity (down to ~30%)
• Ab initio methods are improving, but are not
always accurate
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Binding Site Characterization
• Protein engineering requires identifying key
residues in the active/binding site
• Computational Methods
– Static receptor docking
(Autodock, Dock, etc)
– Simulated binding
– Solvent mapping
• Experimental Methods
– Co-crystalization
– Mutation studies
PEDS 17 (2004) 809-819
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Protein Sequence
Optimization
• Changing N amino acids gives 20N possible variants
– 16 modification sites implies 1020 possibilities
– 1014 years to examine all possibilities
• Intelligent search procedure is required
• Genetic algorithm (GA)
– Mimics survival of the fittest in nature
– Initially random population
– Individuals with greater fitness have higher
selection probability
– Crossover and mutation operators
– Population evolves to have increased fitness
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Genetic Algorithm
Generation 0
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Genetic Algorithm
Generation 0
Generation 1
X
X
X
X
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Genetic Algorithm
Generation 0
Generation 1
X
X
X
X
X
X
X
X
X
X
X
X
X
X
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Multi Objective Function GA
• MITRE has developed a unique procedure to
measure fitness over multiple objective functions
– Enables simultaneous optimization of multiple metrics
on varying measurement scales (e.g. binding energy,
internal energy, and ligand position)
• Robust across a range of parameters (mutation
rate, crossover rate, etc)
• Rapid detection of optimal sequences
– For 3 residues (8k possibilities), GA finds the global
optima at 8% the cost of a full factorial search
– For > 3 residues, GA finds greatly improved proteins
after 30 generations (1000 individuals per generation)
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Protein Design Pipeline (PDP)
• Software developed at MITRE for
engineering proteins
• Combines multiple functionalities
into a single, integrated system
–
–
–
–
–
–
–
Sidechain replacement
Protonation
Charge modeling
Energy minimization
Ligand docking
Internal energy
Scoring
Molecule position
Functions
• Incorporates a genetic algorithm for
high-dimensional optimization
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
PDP Website
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Protein Engineering Process
Wildtype
Protein
Structure
Computational
Design
Laboratory
Production
Define Key
Residues
Produce
DNA
Sequence
Optimize
Protein
through PDP
Overexpress
&
Purify
Pick
Candidates
Test
Functionality
Modified
Protein
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Engineered RBP
• Started with ribose binding protein (RBP)
• Varied 16 binding site residues
• Utilized PDP and GA to optimize binding
pocket for 2-diisopropylaminoethanol
– Population size of 1000
– Tournament sizes: Binding=10, Internal=2
• Rapid convergence to optimized sequence
• Significant improvements in binding
energy and internal energy
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Binding Energy Improvement
-3.5
Avg
Min
-4
Binding Energy
-4.5
-5
-5.5
-6
-6.5
-7
• Average binding
energy drops by
2.96 kcal/mol
• Best binding
energy is -8.06
kcal/mol (WT
values is -1.67
kcal/mol)
• Convergence
after ~30
generations
-7.5
-8
-8.5
0
10
20
30
40
Generation
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Internal Energy Improvement
-1355
• Average internal
energy drops by
35.8 kcal/mol in
40 generations
• Final average
internal energy is
-1396.4 kcal/mol
(WT is -1344.9
kcal/mol)
• Large increase in
stability despite
low emphasis on
internal energy
-1360
Internal Energy
-1365
-1370
-1375
-1380
-1385
-1390
-1395
-1400
0
10
20
30
40
Generation
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Engineered RBP
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
MD Simulations
• Molecular dynamics (MD) simulations
describe protein motions at the atomic level
• Discrete integration of Netwon’s equations
of motion
• Femtosecond resolution
• Ability to analyze various system properties
– Binding mechanisms
– Thermodynamic motion
– Structural stability
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Example MD Simulation
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Protein Stability Analysis
• Residues with unfavorable local interactions
could prevent formation of expected tertiary
structure (inactive protein)
• Molecular dynamics simulations help
indicate protein stability
– Simulate variant with mutations mapped back
to original protein structure
– Compute RMSD values over time and compare
to the wild type protein
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Protein Stability Analysis
Stable Protein
Unstable Protein
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Protein Stability Analysis
Average RMSDs to Initial Structure
7
6
RMSD
5
4
WT-14
WT-13
WT-12
WT-11
WT-10
WT-9
WT-8
WT-7
WT-6
WT-5
WT-4
WT
3
2
1
0
20
40
60
80
100
120
140
160
180
200
Time (picoseconds)
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Protein Design Overview
•
•
•
•
•
Determination of 3D structure
Characterization of interaction site
Optimization through PDP and GA
Stability analysis via MD
Engineered proteins
– Signaling responses to novel molecules
– Enhanced enzymatic capabilities
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Enzymatic Engineering
• Enzymes work by stabilizing a reaction’s
high energy transition state
• Enzymes can be
engineered:
– Characterize the
transition state
– Find residues
that stabilize the
transition state
Nelson & Cox, Lehninger Principles of Biochemistry, 3rd ed., 2000
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
QM/MM Simulations
• Simulating enzymatic processes requires
explicit modeling of electrons
– Too costly for large systems
• Mixed quantum mechanical / molecular
mechanical (QM/MM) energy functions
– Most atoms modeled by molecular mechanics
– Key area modeled by quantum mechanics
• Ability to simulate covalent bond creation
and destruction in enzymatic processes
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Summary
• Proteins can be engineered to have unique
binding or enzymatic functionality
• MITRE’s Protein Design Pipeline enables
computational optimization of a protein for
interacting with a small molecule
• PDP shown to rapidly optimized RBP for binding
2-diisopropylaminoethanol
• MD simulations permit analysis of variant stability
• Simulations can be extended to the quantum
mechanical level for modeling enzymatic reactions
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Next Steps
• Engineering enzymes
• Additional objective functions
– Alternate binding energy calculations
– Ligand localization scores
– Modified binding procedures
• In-depth binding energy analysis (MMPBSA, SIE)
• Multi-function proteins
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616
Acknowledgements
•
•
•
•
•
Eileen Chang
Matt Peterson
David Bauer
Brian Wickham
Brandon Higgs
• John Dileo
• Russell Graef
• Haley Smith
• MITRE Innovation
Program
©2008 The MITRE Corporation, All Rights Reserved
Approved for Public Release, #08-1616