Transcript Peptide Design - Duke Computer Science

Peptide Design
Kyle Roberts
March 4, 2008
Peptides in Biology





Create peptide antibodies
Used in mass spec to identify proteins
Can probe protein-protein interactions
Function as protein ligands
Antimicrobial Peptides
http://employees.csbsju.edu/hjakubowski/classes/ch331/protstructure/phipsi.gif
Motivation for Peptide Design





Understand how the basic components of proteins
function and interact
Abstract out general rules that can be applied to
understand protein folding
Design useful and novel protein binders and
inhibitors
Utilize the growing pdb structures to refine
structures and build novel ones
Apply knowledge to chemicals similar to peptides to
create novel structures (foldamers)
Peptide Backbone Reconstruction

Reconstruct an all-atom peptide model from a
subset (Cαs or Cβs) of the atomic coordinates
Adcock SA. Peptide backbone reconstruction using dead-end elimination and a
knowledge-based forcefield. J Comput Chem. 2004 Jan 15;25(1):16-27.
Uses of Backbone
Reconstruction


Enhancing low-resolution structures
Conversion of coarse grain structures into allatom models



Ab initio folding
*Comparative modeling techniques*
Normal mode analysis
Current Methods


Use fragment libraries and construct the
backbone with energy, homology or
geometric criteria
Perform de novo construction with geometric
or energy criteria



Statistical positions and frequency tables
Molecular dynamics and Monte Carlo
Maximize peptide dipole alignments (max Hbonds)
Algorithm Overview
Library of amino
acid peptide
sequences
(length 3 )
Cα or Cβ
coordinates
Overlay
peptides on
input coords
Dead End
Elimination
Predicted
Structure
Peptide Backbone Fragments



Selected three-residue
backbone fragments from
1336 random nonredundant
PDB structures
All the fragments were
aligned into a standard frame
Fragments were clustered by
RMSD and duplicates were
discarded
Fragment Overlap


Fragments were overlapped with the input
coordinates by minimizing the sum of squared
distances
Kearsley’s method: state the minimization problem
as an eigenvalue problem with quaternion algebra




not iterative
improper rotations aren’t produced
no special cases
RMSD is easy to calculate
Dead-End Elimination

Minimize the energy function:
N
N
N
Etotal   Esingle( ri )   Epair( ri, sj )
r 1 s 1
ri is residue r with the
backbone conformation i
ri
Energy
r 1
rj
Image: Courtesy of Ivelin
Database-Derived Forcefield
Radial Distribution Function
P(occurance) = e-E/RT
EAB = -RT ln(gAB(r))
gAB(r)  NAB(r)
NA
NB
i 1
j 1
NAB( r )   SAi  (| rAi  rBj | r )
http://www.nyu.edu/classes/tuckerman/stat.mech/lectures/lecture_8/node1.html
Results





Generally structures are obtained at 0.2-0.6Å
RMSD to crystal structure
When compared to a well used server
(MaxSprout) the server was about 20% less
accurate
An alternative algorithm worked “better” but
author claims training set was biased
Phi-Psi angle correlation was on average
0.95 and 0.88 respectively
Computation can be completed in minutes
Input Error
Peptides that Target
Transmembrane Helices



Methods exist for the design or selection of
antibodies for water soluble proteins
Different methods must be developed for
membrane proteins due to our lack knowledge
Idea: Develop a peptide alpha helix that can
insert into membrane and bind target
membrane α-helix
Computational Design of Peptides That Target Transmembrane Helices
Hang Yin, Joanna S. Slusky, Bryan W. Berger, Robin S. Walters, Gaston Vilaire, Rustem
I. Litvinov, James D. Lear, Gregory A. Caputo, Joel S. Bennett, and William F. DeGrado
(30 March 2007) Science 315 (5820), 1817
Transmembrane Proteins




Embedded in lipid bilayer
Difficult to crystallize
Underrepresented in PDB
Allow communication from
outside to inside of cell
INTEGRIN STRUCTURE, ALLOSTERY, AND BIDIRECTIONAL SIGNALING
M.A. Arnaout, B. Mahalingam, J.-P. Xiong
Annual Review of Cell and Developmental Biology 2005 21, 381-410
Design Overview





Choose the target alpha helix sequence
Find matching templates in the pdb database to native
binding structure of target helix
Thread the target sequence onto one of the template
helices
Choose proximal positions to mutate on the other
template helix
Mutate those positions to all hydrophobic residue
rotamers and repack
Find Templates



The integrin alpha helices that were
chosen as targets contain a small-X3small motif and a right handed crossing
angle
Membrane proteins in the pdb were
searched for a helix-helix dimer with
this motif and crossing angle
Note: Among the few crystallized
membrane helix-helix pairs they seem
to fall into a few well defined motifs
Threading and Allowable
Mutations



Change the amino acid identities
of one of the alpha helices to that
of the target sequence (αIIB)
Align the small-X3-small motif
Allow mutations (for design of
“anti” helix) at positions close to
helix-helix interface (pink)
Repacking




“Anti” peptide designed with Monte
Carlo simulated annealing
At each step one residue identity is
changed, and then the rotamers are
optimized with DEE
The new energy is then calculated
with a linearly damped LennardJones potential and membrane
depth-dependent knowledge based
potential
Accept structure based on a
Boltzman coin flip
Testing the Design



Target membrane was integrin αIIb alpha helix
Integrins are inactive when the α-subunit helix is
bound to the β-subunit helix and active when not
bound
αIIb causes the aggregation of platelets through
binding with fibrinogen
Platelet inhibitor through
signal transduction
ADP scavenger (ADP
stimulates plate
aggregation)
Inhibits binding
to fibrinogen
Extensions

Currently this method is restricted to dimers
and helices that are non-polar




Could include motifs with polar side chains
Design for multispan bundles rather than dimers
Use negative design to avoid amyloid formation
or binding to undesired targets
Improve scoring function to account for more
interaction types
http://sb.web.psi.ch/images/amtb_in_membrane.png
Membrane Targeting Helices



Probe TM helix binding and function by
targeting different membrane helices
Characterize folding of membrane proteins by
blocking alpha helices as they form
Requires novel testing methods in order to
determine whether helix is actually binding
and affecting function
Moving Past Peptides:
Foldamers



Proteins and RNA are unique in that
they adopt specific compact, stable
conformations
Biology has been fairly constrained
so there should be much potential
for other compactly folded polymers
Foldamer: “any polymer with a
strong tendency to adopt a specific
compact conformation”
http://www.geneticengineering.org
/chemis/ChemisNucleicAcid/Graphics/tRNA.jpg
Gellman, SH. Foldamers: A manifesto. Acc. Chem. Res.1998, 31, 173-180
Creating Foldamers



Find new backbone units with suitable folding
propensities
Give the created foldamer interesting
chemical functions
Be able to produce foldamers efficiently
Foldamer Uses

Test our understanding of protein function


Since all our analysis has been on only α-amino
acids, have we “overfit” our understanding
Develop new building blocks and molecular
frameworks for the design of
pharmaceuticals, diagnostic agents,
nanostructures, and catalysts
Foldamers as versatile frameworks for the design and evolution of function
Catherine M Goodman, Sungwook Choi, Scott Shandler & William F DeGrado
Nature Chemical Biology 3, 252-262 (2007)
Monomer Framework Selection

Aliphatic

Aromatic
Foldamer Secondary Structure
C10(310)
C12
C13(α)
C14
Predictability of Secondary
Structure



Adding salt bridges spaced one turn apart
introduce stability
Charged groups at helix ends stabilize
according to their polarity
α-amino acid knowledge can be transferred
about stabilization by disulfides, covalent
bridges, and binding of metal ions
Aromatic Oligomers

Size of monomer and substitution of aromatic
ring provide reliable determination of helical
radius
Jiang, H., Leger, J.M. & Huc, I. Aromatic -peptides. J. Am. Chem. Soc. 125, 3448–3449 (2003).
Designing Foldamer Function





Foldamers can interrupt Tat/TAR binding
Penetrate bacterial cells in a passive process
Have antimicrobial properties dependent on
the length and hydrophobicity
Mimics to interrupt protein-protein
interactions with Ki up to 0.8 uM and 7.1nm
By using an α/β sequence a ten-fold higher
affinity was found than the native peptide
ligand
Foldamer Tertiary Structure


A zinc finger-like motif
was recently built
consisting of β-peptides
with a β hairpin and 14helix
An octomer consisting
of β-peptides was
created with only noncovalent interactions
Benefits of Foldamers



Foldamers are more resistant to enzymatic
attack then peptides
Fewer monomeric units are needed to adopt
a well-defined secondary structure
Can be used as a strategic method to
downsize peptides to small molecules
Natural Peptide
14-helix β-peptide
Arylamide foldamer Phenylalkylnyl
Summary

Peptide design can be used in a variety of
ways




Backbone reconstruction
Antibodies for membrane proteins
Foldamers
All of these methods help us understand how
proteins fold and the underlying rules, which
will allow better models and hopefully better
functional designs
Questions?
“It is not clear to the author why LYS-59 is reported as such, because
the crystal structure contains a valine at position 59.”