Kein Folientitel

1
(No Transcript)
2
Homology Modelling

• Goal Generation of a 3D protein model from
  sequence only
• Steps
• Find homologues proteins with known structures,
  e.g. from PDB (www.rcsb.org/pdb)
• Align sequences of structurally known proteins
• Find structurally conserved regions
• Align with target sequence (structurally unknown
  protein)
• Construct structurally conserved regions (SCRs)
• Construct structurally variable regions (SVRs)
• Run Molecular Dynamics simulation to refine model

3
Homology Modelling Sequence Alignment
4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
Homology Modelling

• Summary
• Works well for sequences with high homology to
  structurally known protein (gt30 identity) and
  when several members of the family are known
• Structurally variable regions ("loops") cause
  problems when loops have different lengths
• Correct prediction of amino acid sidechains can
  be very difficult

8
Structure-based Drug Design
9
Drug Design Techniques
TARGET structure

• "Small molecule" Modelling
• Superimposition of known ligands
• Conformational analysis
• Pharmacophore hypotheses
• 3D-Database searches

• Structure-based Design
• Search for "hot spots"
• Pharmacophore hypotheses
• 3D-Database searches
• Docking
• De-novo Design

10
Structure-based Drug Design
11
Analysis of Protein Pockets - the Search for
Hot Spots

• Where are the best regions for protein-ligand
  interactions?
• Which functional groups are best suited to form
  strong interactions with the protein?
• Two standard programs
• GRID (P. Goodford, Oxford)
• IsoStar/SuperStar (Cambridge Crystallographic
  Data Centre)

12
Analysis of Protein Pockets - the Search for
Hot Spots

• GRID
• A regular 3D grid is placed into a protein pocket
• At each grid point, the interaction energy of a
  probe with the protein is calculated (force
  field calculation)
• Probes are taken from a probe library of typical
  functional groups (e.g. hydroxyl, carboxylate,
  amide, ammonium, aliphatic carbon etc.)
• Hot spots are visualized by contouring grid
  points of identical interaction energy
  (isocontour surfaces)

13
Analysis of Protein Pockets - the Search for
Hot Spots
14
Analysis of Protein Pockets - the Search for
Hot Spots

• IsoStar/SuperStar
• Interaction geometries of functional groups are
  derived from empirical (statistical) analyses of
  small-molecule crystal structures (Program
  IsoStar)
• For each amino acid of the protein pocket,
  interaction sites of functional groups can be
  constructed and visualized(Program SuperStar)

15
Analysis of Protein Pockets - the Search for
Hot Spots
Positions of OH-groups in small-molecule crystal
structures, determined by IsoStar
16
Analysis of Protein Pockets - the Search for
Hot Spots
Positions of Hbond donors and acceptors,
calculated by SuperStar
in Chymotrypsin
in Thrombin
17
Two methods to place ligands into protein pockets

• Docking
• Tries to find the best positions of known ligands
  stored in 3D databases
• De novo-Design
• Constructs completely new ligands from small
  fragments step by step
• Both methods require scoring functions to rank
  the ligands

18
Docking
(Adapted from Verlinde Hol, Structure 1994, 2,
577-587)
19
De Novo Design Seed Grow Strategy
(Adapted from Verlinde Hol, Structure 1994, 2,
577-587)
20
De Novo Design Place Link Strategy
(Adapted from Verlinde Hol, Structure 1994, 2,
577-587)
21
Steps in De Novo Design
Step 3 Scoring
Step 4 Decision making(manual or automatic)
22

De Novo Design Methods
Method Year of first
publication ADAPT 2001 BUILDER 1995 CAVEAT
1994 CLIX 1992 CONCERTS 1996 DLD 1995 GENSTAR
1993 GROUPBUILD 1993 GROW 1991 GROWMOL 1994 LEAPF
ROG 1993 LEGEND 1991 LUDI 1992 MCDNLG 1991 MCSS/HO
OK 1991 MOLMAKER 1996 NEWLEAD 1993 PRO-LIGAND 1995
PRO-SELECT 1997 SKELGEN 1997 SME 1998 (COMBI)SMOG
1996 SPLICE 1993 SPROUT 1993 TOPAS 2000 ......
23
Hot Topics in Drug Design
24
Example of a De Novo Design Session