Identification of Hot Spots within Druggable Binding Regions by Computational Solvent Mapping of Pro

1
Identification of Hot Spots within Druggable
Binding Regions by Computational Solvent Mapping
of Proteins

• Melissa R. Landon,
• David R. Lancia,
• Jr.,
• Jessamin Yu,
• Spencer C. Thiel,
• Sandor Vajda

2
ProseminarIdentification of Hot Spots within
Druggable Binding Regions by Computational
Solvent Mapping of Proteins

• 1. Introduction
• 2. Computational Methods(Algorithm)
• 3. Results and Discussion
• 4. Conclusion

3
Introduction

• Experimental Hot Spot Identification

Targets in drug design
Protein Binding Sites
Hot Spots
NMR X-Ray crzstallography Other Biophzsical
Methods
Results
4
Introduction

• Computational Solvent Mapping (CS-Map) Algorithm

3 Test Systems (1)Renin aspartic protease
Aliskiren (2) A set of previously
characterized druggable protein (3)E.coli
Ketopantoate reductase KRP
Novartis
CS-Map
Abbott
Results
Rebuild
Ciulli
5
Computational MethodIntroduction

• 5 Steps
• The Regions the highest number of different
  probe clusters overlap regard as the Hot Spots
• Input for the CS-Map X-ray structure of the
  protein without ligands

6
Computational Methods Step 1 Rigid Body Search

7
Computational Method Step1 Rigid Body Search

• a set of 222 initial probe positions using a
  placement algorithm
• Identification of buried and surface points
• Pocket-like surface points as initial probe
  positions
• Place a probe at each of these cluster centers
• Move the probes with a multistart simplex method
  to energetically optimal positions

The number of cluster centers obtained the
remaining number of available probe positions
nonpocket-like surface points are clustered
8
Computational MethodStep1 Rigid Body Search

• Free Energy Score Function for the probe-protein
  complex
• The coulombic component of the electrostatic
  energy
• the charge of each probe atom the
  solvated proteins
• electrostatic field at that position
• The desolvation free energy
• An excluded volume penalty term.
• 0, if the probe does not overlap with
  the protein

CONGEN
The atom contact potential model
9
Computational Method Step2 Free Energy
Refinement and Final Docking

• 6000 conformations of each probe type in the
  probe-protein complex obtained from the step 1
• a revised desolvation term
• a van der Waals energy term
• using the Newton-Raphson method in Charmm
  minimize each probe-protein complex

Charmm version 27
10
Computational Method Step3 Clustering,Scoring,
and Ranking

• Contact Vector

11
Computational Method Step3 Clustering,Scoring,
and Ranking

• Cluster the contact vectors based on their
  D-Values to all other vectors
• Cluster Algorithm Seeding
• threshold of 0.35 for addition to cluster

12
Computational Method Step3 Clustering,Scoring,
and Ranking

• Optimization of the clusters
• Delete the small cluster with lt15 members
• Probability
• For each cluster calculate

13
Computational Method Step4 Creation and
Ranking of Consensus Sites

• Create consensus sites
• 5-10 lowest average free energy clusters of
  each probe type
• Rank consensus sites
• the total number of probe clusters and
• the number of different probe types
• For example
• consesus site with 13 of the 14 different
  probe molecules is ranked higher than only with 7
  of the 14 probe molecules
• if the total number of clusters same

14
Detection of Hot Spots Using CS-Map and HBPLUS

• 10 conformations of aliskiren with GOLD(Genetic
  Optimization for Ligand Docking)
• Calculate nonbonded and H-bonded interactions
• Atom interactions atom interactions for each
  residue / atom interactions for all residues
• gt 4 regarded as being potentially located in a
  hot spot
• Repeat the process above between each protein and
  its bound ligand

15
Results and DiscussionConforamtional analysis
of Aliskiren in the Binding Pocket of Renin

• Conformation of aliskiren wrt a peptidomimetic
• Docked conformation of aliskiren

16
Results and DiscussionIdentification of High
Affinity Subsites in the Binding Pocket of Renin

• Superimposition gaining from the mapping of the
  5 structures(1BIL,1BIM,1HRN,1RNE and 2REN)
• S1, S2, S3 and S3SP contributes for ligand
  affinity

17
Results and DiscussionIdentification of High
Affinity Subsites in the Binding Pocket of Renin

• Both the S1 and S3 subsites of the binding pocket
  for each structure
• no significantly populated consensus site in the
  S4
• only a single, low-ranked consensus site in the
  S2
• Conclusion the S1, S3, S2 and S3SP subsites
  with higher affinity

18

• The Pearson correlation coefficient between the
  CS-Map probes and aliskiren R-Value 0.72
• Between the CS-Map probes and the peptidomimetics
  
• R-Value 0.19

19
Results and DiscussionCharacterization of
Druggable Binding Pockets Using CS-Map

• The training group of 12 proteins including 13
  sites.
• The Binding regions with a low consensus site
  ranking as nondruggable.
• The Consensus sites for each druggable binding
  pocket have top ranking
• Conclusion the CS-Map utilize in prediction of
  the druggability of a binding region with the
  available structural information

20
Results and DiscussionComparison of CS-Map to
Experimental Fragment-Based Approaches for Hot
Spot Identification A Case Study with E.coli KPR

Reduced nicotinamide groups
2-phosphate
ß-phosphate ribose group
21
Results and DiscussionComparison of CS-Map to
Experimental Fragment-Based Approaches for Hot
Spot Identification A Case Study with E.coli KPR

R-Value 0.6
22
Results and DiscussionComparison of CS-Map to
Experimental Fragment-Based Approaches for Hot
Spot Identification A Case Study with E.coli KPR

• Using HBPLUS, normalize the percentage of
  interactions for each residue. The higher values
  represent a higher concentration of ligand
  interaction in that region.
• 4.15 with 2-phosphate
• 4.63 with the reduced nicotinamide
• Conclusion Given spatial information regarding
  the location of hot spots, using CS-Map can
  predict ligand efficiencies of fragments of
  larger molecules

gt ß-phosphate ribose group
23
Conclusion

• The CS-Map can not only be used to predict the
  hot spots for protein targets where only general
  druggability features are currently known, but
  also to validate new protein targets where
  druggability is not known.
• Using the CS-Map can arrive at the same results
  as with the other biophysical methods.
• But for large changes in the conformation of the
  binding site, the results from the CS-Map are not
  so robust.