Prediction of SH3 Domain Binding Motifs

1
Prediction of SH3 Domain Binding Motifs

• Presented by Siba IsmaelSupervised by Mazen
  Ahmad
• University of SaarlandSaarbrücken, 17.10.08

2
Outline

• SH3 motif and proline-rich domains- Motivation
  to find SH3 domains binding sites-Why
  proline-rich domains?
• Binding Free Energy Method What flanking
  sequences govern binding specificity
• Materials and Methods Bioinformatics
• Results of Prediction
• Conclusions and Outlook

3
SH3 DomainsMotivation- Assembly

• Comprise 60 residues
• Play assembly and regulatory roles.
• Assembly role example Grb2
• Cascade Growth factor receptor tyrosine kinase
  Grb2 SOS Ras MAPK - Play roles in
  cell growth and differentiation

4
SH3 DomainsMotivation- Regulation

• Regulation example Src
• Built-in SH2SH3 inactivation (autoinhibition)
• Disruption External SH2 and SH3 domains
  interaction-result in kinase activation
• SH3 interactions week- typical dissociation
  constant- essential for reversible switching
  mechanism.

5
Repetitive Proline-Rich Sequences

• in many cases, thought to function as docking
  sites for signaling modules
• found in the context of larger multidomain
  signaling proteins.
• Binding assembly and targeting of protein
  complexes involved in- cell growth
• - cytoskeletal rearrangements-
  transcription - postsynaptic signaling processes
• play a regulatory role and autoinhibitory
  interactions

6
Repetitive Proline-Rich Sequences

• Why proline in interaction modules?
• Proline unique amino acid in - constraints on
  dihedral angles imposed by cyclic side chain-
  its resulting secondary structural preferences

7
Repetitive Proline-Rich Sequences

• Why proline in interaction modules?
• propensity to form a polyproline type II (PPII)
  helix.- extended left-handed helical structure
  with three residues per turn.
• - useful recognition motif - carbonyls
  point out from the helical axis into solution -
  restricted backbone entropy cost of binding
  reduced
• - twofold rotational pseudosymmetry
• - two binding possibilities
• - orientational switching differing
  domain function

8
Repetitive Proline-Rich Sequences

• Why proline in interaction modules?
• The only naturally occuring N-substituted amino
  acid- sequence-specific recognition without
  high-affinity interaction.- specific and low
  affinity interactions - reversibility -
  intracellular signalling
• Stable cis conformation- high kinetic barrier-
  rate limiting step

9
Proline-Rich Sequences vs. SH3 Interaction

• PxxP motif flanked by different specificity
  elements- K/RxxPxxP and PxxPxK/R classes of
  ligand motif- single recognition surface two N-
  to C-terminal orientations ligand binding
• SH3 fold two antiparallel ß sheets at right
  angles.- in fold RT and n-Src loops flanking
  specificity pockets
• Aromatic SH3 groove PPII helix ridges (a pair
  of residues)

10

• So how to detect the binding affinity to SH3
  domains?Computational Analysis!!!
• Solvation Energy!!
• BIOPHYSICS

11
Binding Free Energy

• mechanical energy to disassemble a whole into
  separate parts scalar
• Binding free energy cycle- in terms of transfer
  free energiesWhy? - from a homogeneous
  dielectric environment (interactions Coulomb's
  law)- to an inhomogeneous dielectric
  environment differing internal and external
  dielectric constants.

12
Binding Free EnergySolvation Energy Contribution

• Solvation energy for the complex and each of its
  parts
• But how to calculate solvation energy?

Remember!! This stands for Coulombic
13
Binding Free EnergySolvation Energy Contribution

• Full solvation energy cycle- Step 1 Total
  Solvation- Step 2 charging of the solute in
  solution inhomogeneous presence of
  mobile ions. -Step 3 attractive
  solute-solvent dispersive interaction - Step 4
  repulsive solute-solvent interaction - Steps 5
  and 6 null steps. - but used to offset
  unwanted energies charging of the solute
  in vacuum homogeneous absence of
  mobile ions.

14
Binding Free EnergySolvation Energy Contribution

• APBS??

ACC??
15
Binding Free EnergyIncluding Coulombic
Contribution

• the sum of pairwise Coulombic interactions- for
  all atoms in the molecule - for a particular
  uniform dielectric
• Coulombs Law
• Potential Dielectric Energy

Coulomb??
16
Binding Free Energy Entropy

• Entropy a measure of the unavailability of a
  systems energy to do work- measure of the
  randomness of molecules in a system - central
  to the second law of thermodynamicsSpontaneous
  changes Entropy (isolated systems)

17
Binding Free Energy van der Waals

• van der Waals force attractive or repulsive
  forces between molecules and per molecule
• not covalent bonds or electrostatic interaction
  of ions, but- permanent dipolepermanent
  dipole forces - permanent dipoleinduced dipole
  forces - instantaneous induced dipole-induced
  dipole

18
Poisson-Boltzmann Equation

• Differential equation describes electrostatic
  interactions between molecules in ionic
  solutions
• models implicit solvation (continuum solvation )

19
Methods

• APBS Package Adaptive PoissonBoltzmann
  Solver- numerical solution for the
  Poisson-Boltzmann equation - modeling
  biomolecular solvation In my work apbs
  electrostatic potential and polar solvation
  acc SASA calculation solvent accessible
  surface area nonpolar solvation
  coulomb coulombic interactions in vacuum
• Pdb2pqr Package platform-independent utility -
  converts protein files in PDB format to PQR
  format

20
Methods

• PQR file PDB file temperature and occupancy
  columns replaced by the per-atom charge (Q) and
  radius (R)
• Jackal package for protein structure modeling
  scap protein side-chain program
• predicts side-chain conformations and side chains
  of a whole protein and in
• mutates specified residues in a protein
• R language Package Statistical Language
  environment

21
Methods

• To predict a binding motif of length 10- chose
  the crystal structure of the peptide APSYSPPPPP
  complexed with the Abl SH3 domain - mutate it
  to other sequences
• Try predicton of 10 very good out of the 600
  candidates, and 15 of the nonbinders
  almost all have a PxxP domain!

Fix P at P0 and P3
22
Methods

• From literature Binding Free Energy Difference
  to the base sequence with the following mutations

23

• Results

24
Correlation?!
Correlation 0.9534504
Correlation 0.722554

• Correlation
• 0.4530898

25
Reproducibility?!

• Without vdW or entropycorrelation0.5435262

For both Correlation 0.3357690Second compared
to base sequence
Why not much good?
26
Peptide Binding-Solvation Polar
Easier barrier to break for binders
27
Coulombic Interactions
Mean Coulombic Energy is less for binders!
28
Nonpolar Solvation Contribution
Neglicted effect!
29
van der Waals Contribution
Major contribution to binding specificity

30
Entropy Contribution
Most non-Binders Lost more Entropy upon Binding
than did Binders!
31
Binding Free Energy
Less Binding Free Energy for Binders!
Easier barrier to break
32
Separation of Binders from non-bindersPrediction

• Linear Discriminant Analysis!

33
From Literature
Nonbinders
Binders
Sequence ??Gpred
SKKEMQPTHP 19.6
ASQKMEPRAP 43.3
WELSSQPTIP 26.3
LAPASTPTSP 13.6
ASTPTSPSSP 11.4
SSPGLSPVPP 13.8
RGVLIEPVYP 38.9
DEPNLEPSWP 26.4
RLVGARPLLP 24.6
RTESEVPPRP 26.6
LASRPLPLLP 20.1
ISQRALPPLP 30.8
ITMRPLPALP 17.3
RSGRPLPPIP 32.7
KWDSLLPALP 17.4
YWDMPLPRLP 4.2
YYQRPLPPLP 9.1
YFSRALPGLP 8.8
SLWDPLPPIP 15.2
DPYDALPETP 28.6
34
Results concerning Prediction!

• Proline preferece in the binding motif-
  Available experimental measurements at positions
  P3, P0, P-3, and P-5 - Particularly important
  for the peptide binding - conserved Pro
  residues at P3 and P0 strong binding affinity
  (PxxP- work here) - residues at P-3, and
  P-5 the binding specificity (the other work)
• Other residues, especially hydrophobic (Phe, Leu,
  Met, Val, and Trp), also favored

35
Conclusion and Outlook!

• Binding free energy - nice method predictiong
  binding preferences- easy to deal with data
• Can be used in prediction of different sets of
  protein-ligand interaction prediction
• High throughput results in the fields of
  medicine, pharmacy, and biology

36
References

• Tingjun Hou, Ken Chen, William A McLaughlin,
  Benzhuo Lu, and Wei Wang. Computational Analysis
  and Prediction of the Binding Motif and Protein
  Interacting Partners of the Abl SH3 Domain
• Wikipedia
• T.Geyer, Dynamic Cell Simulation
• Jackal supported by National Science Foundation
  and National Institute of Health developed in
  Honig Lab
• Baker NA, Sept D, Joseph S, Holst MJ, McCammon
  JA. APBS Electrostatics of nanosystems
  application to microtubules and the ribosome.
  Proc. Natl. Acad. Sci. USA 98, 10037-10041 2001.
• Dolinsky TJ, Nielsen JE, McCammon JA, Baker NA.
  PDB2PQR an automated pipeline for the setup,
  execution, and analysis of Poisson-Boltzmann
  electrostatics calculations. Nucleic Acids
  Research, 32, W665-W667 (2004).
• R Regulatory Compliance and Validation Issues A
  Guidance Document for the Use of R in Regulated
  Clinical Trial Environments
• Google Machine Search

37
Thanks for Attention
Questions?!