Upload
others
View
33
Download
2
Embed Size (px)
Citation preview
11
Structure-BasedDrug Design
Thomas Funkhouser
Princeton University
CS597A, Fall 2005
Introduction
Drugs• Molecules that can be introduced
to change biological activity
Slide courtesy of Bill Welsh
Introduction
Drug targets• Enzyme - inhibitors• Receptor - agonists or antagonists• Ion channels - blockers• Transporter - update inhibitors• DNA - blockers
Slide courtesy of Bill Welsh
Structure-Based Drug Design
Goal:• Given a protein structure,
and/or its binding site,and/or its active ligand (possibly bound to protein),find a new molecule that changes the protein’s activity
Structure-Based Drug Design
Receptor-based drug design:• Given a protein structure,
and/or its binding site,and/or its active ligand (possibly bound to protein),find a new molecule that changes the protein’s activity
HIV Protease Inhibitor
�������������� ����
Example courtesy of Bill Welsh
Structure-Based Drug Design
Ligand-based drug design:• Given an protein structure,
and/or its binding site,and/or its active ligand (possibly bound to protein),find a new molecule that changes the protein’s activity
Example courtesy of Joe Corkery
22
Challenges
Scoring of chemical models • Activity• Toxicity• “Druggability”• etc.
Search of chemical space• Add/remove/replace chemical groups• Conformations• etc.
Outline
Virtual drug screening• Ligand-based
§ 2D structure matching§ 3D structure matching§ QSAR
De novo drug design• Models
§ Simulation§ Knowledge-based
• Construction algorithms§ Incremental construction§ Fragment-based§ Stochastic optimization
Ligand-Based Drug Screening
HO
OH
17b-Es tradiol Andros tenediol
HO
OH
O
CH3O
Proges terone
HO
OHH3CO
Moxes trol Coumestrol
O OHO
OHO
HO
OH
Diethyls ti lbes trol Genis tein
OHO
OHOOH
Methoxychlor
H3CO
Cl Cl Cl
OCH3 HO OH
Bisphenol A
a- Zearalanol(Zeranol)
O
OOH
HOOH
4-Hydroxytamox ifen
ON
HOO
N
H3CO
Nafox idine
Tamox ifen
ON
Clomifene
Cl
ON
HO
OH
(CH2)10CO N(CH3)C 4H9
ICI 164,384
Ligands for Estrogen Receptor Slide courtesy of Bill Welsh
Ligand-Based Drug Screening
Strategies:• 2D matching• 3D matching• Pharmacore matching• Quantitative Structure Activity Relationships (QSAR)
2D Structure Matching
OHN
N
OH
NH
N NH2
O
N NH
N
NH2
NH
N
NH2
NH
N
N NH
N
NH
OH
Query
Slide courtesy of Bill Welsh
2D Substructure Matching
Query
N
O
O
N NO
O
O
O
N NN
N
NNN
N O
N
N
O
O
NO
O
N
OO
Slide courtesy of Bill Welsh
33
2D Substructure Matching
N-NO-C(-N)-CCH3-Ar-CH3C-N-NN-Ar-Ar-ON-C-ON-Ar-O OH > 1CH3 > 1N > 1NH...
���������������
�
�
�
�
�
�
�
Slide courtesy of Bill Welsh
2D Substructure Matching
Which keys are present?
Slide courtesy of Bill Welsh
2D Substructure Matching
Tanimoto coefficient:
struct A: 00010100010101000101010011110100 13 bitsstruct B: 00000000100101001001000011100000 8 bits
A & B: 00000000000101000001000011100000 6 bits A ∩BA or B: 00010100110101001101010011110100 15 bits A ∪B
BAB)(A
T ∪∩=
)(
T = 6 / 15 = 0.4
Slide courtesy of Bill Welsh
2D Substructure Matching*Activity data from Uehling et al, J. Med.Chem. 1995
0
0.5
1
1.5
2
2.5
1.00 0.95 0.90 0.85
Act
ivity
diff
eren
ce (
log
units
)
Tanimoto similarity
Slide courtesy of Bill Welsh
Outline
Virtual drug screening• Ligand-based
§ 2D structure matchingØ 3D structure matching§ QSAR
De novo drug design• Models
§ Simulation§ Knowledge-based
• Construction algorithms§ Incremental construction§ Fragment-based§ Stochastic optimization
3D Structure Matching
Search database of molecules for ones with similar 3D shape and chemistry
A B
44
3D Structure Matching
Search database of molecules for ones with similar 3D shape and chemistry
Ai Bi
��∈∈
−+−=BB
iAA
i
ii
BABABAd22),(
3D Substructure MatchingO
O N
a
b
c
a = 8.62 0.58 Angstroms
b = 7.08 0.56 Angstroms
c = 3.35 0.65 Angstroms +-
+-
+-
OO
O
O
OO
N
OO
O
N
N
N O
O
O
O
O
O
N
N
N
N
S
OO
O O P O
O
O P O
O PO
O
O
ON
N
N
N
N
OO
O O
O
N
N
N
O
N
O
OO
Slide courtesy of Bill Welsh
3D Substructure Matching
Slide courtesy of Bill Welsh
5.24.2-4.7
6.7
4.85.1-7.1
3D Substructure Matching
Slide courtesy of Bill Welsh
C(u)O(s1)
O(s1)
A
A
[O,S]
O
3.6 - 4.6 Å
3.3 - 4.3 Å
6.8 - 7.8 Å
OCl H
O
Cl
H
3.2Å
4.3Å
Rigid
Flexible
3D Substructure Matching
Slide courtesy of Bill Welsh
DISTANCE CONSTRAINTS(Qualitative Affinity prediction mostly)
DISTANCE CONSTRAINTS(Qualitative Affinity prediction mostly)
BIOPHASE
LIGAND
RECEPTORRECEPTOR
A
B
C
A'
B'
C '
Pharmacore Matching
HO
OH
17b-Es tradiol Andros tenediol
HO
OH
O
CH3O
Proges terone
HO
OHH3CO
Moxes trol Coumestrol
O OHO
OHO
HO
OH
Diethyls ti lbes trol Genis tein
OHO
OHOOH
Methoxychlor
H3CO
Cl Cl Cl
OCH3 HO OH
Bisphenol A
a- Zearalanol(Zeranol)
O
OOH
HOOH
4-Hydroxytamox ifen
ON
HOO
N
H3CO
Nafox idine
Tamox ifen
ON
Clomifene
Cl
ON
HO
OH
(CH2)10CO N(CH3)C 4H9
ICI 164,384
Ligands for Estrogen Receptor Slide courtesy of Bill Welsh
55
Pharmacore Matching
HO
OH
17b-Es tradiol Andros tenediol
HO
OH
O
CH3O
Proges terone
HO
OHH3CO
Moxes trol Coumestrol
O OHO
OHO
HO
OH
Diethyls ti lbes trol Genis tein
OHO
OHOOH
Methoxychlor
H3CO
Cl Cl Cl
OCH3 HO OH
Bisphenol A
a- Zearalanol(Zeranol)
O
OOH
HOOH
4-Hydroxytamox ifen
ON
HOO
N
H3CO
Nafox idine
Tamox ifen
ON
Clomifene
Cl
ON
HO
OH
(CH2)10CO N(CH3)C 4H9
ICI 164,384
Ligands for Estrogen Receptor Slide courtesy of Bill Welsh
HO
OH12 A
Outline
Virtual drug screening• Ligand-based
§ 2D structure matching§ 3D structure matchingØ QSAR
De novo drug design• Models
§ Simulation§ Knowledge-based
• Construction algorithms§ Incremental construction§ Fragment-based§ Stochastic optimization
QSAR
Learn model for activity as function of descriptors (properties) computed from molecules
Compound Activity (pK i) "Y"
Descriptors (Xi) Mol. Vol. (Å3) LogP Dipole Mom (µµµµ)
1 2.34 420 2.8 0.97 2 1.89 332 4.6 2.23 3 0.23 198 -0.3 3.36 4 3.67 467 3.7 0.45 5 2.55 359 -1.5 1.77
etc. etc. etc. etc. etc.
Slide courtesy of Bill Welsh
QSAR
Use model to predict activities for new leads from their descriptors
���������
��� �
��� �������
����� ����� ������������������
���������� ��� �� !"#�$%�&�'� !#(�$��)��&�* !+,�$µ�&
V logP µµµµHO
OH
Slide courtesy of Bill Welsh
Outline
Virtual drug screening• Ligand-based
§ 2D structure matching§ 3D structure matching§ QSAR
De novo drug design• Models
§ Simulation§ Knowledge-based
• Construction algorithms§ Incremental construction§ Fragment-based§ Stochastic optimization
De Novo Drug Design
General Strategy:Ø Given a protein structure• Build model of binding site• Construct molecule that fits model
66
De Novo Drug Design
General Strategy:• Given a protein structureØ Build model of binding site• Construct molecule that fits model
De Novo Drug Design
General Strategy:• Given a protein structure• Build model of binding siteØ Construct molecule that fits model
Outline
Virtual drug screening• Ligand-based
§ 2D structure matching§ 3D structure matching§ QSAR
De novo drug designØ Models
§ Simulation§ Knowledge-based
• Construction algorithms§ Incremental construction§ Fragment-based§ Stochastic optimization
Binding Site Models
Example courtesy of Joe Corkery
Binding Site Models
Simulation• e.g., GRID
Knowledge-based• e.g., X-SITE
C3Dry
N1 OPredicted
Binding SiteModel for
1kp8-1-H-ATP-1-_using GRID
[Goodford85]
Outline
Virtual drug screening• Ligand-based
§ 2D structure matching§ 3D structure matching§ QSAR
De novo drug design• Models
§ Simulation§ Knowledge-based
Ø Construction algorithms§ Incremental construction§ Fragment-based§ Stochastic optimization
77
Incremental Construction
2. Add reactants (substituents)
4. Minimize & Score
O
NNH2
O
1. Docking scaffolds
O
NNH2
O
NH2
O
NNH2
O
O
NNH2
O
3. Find conformation
Slide courtesy of Bill Welsh
Fragment-Based Methods
Slide courtesy of Bill Welsh
Linking
Building
Protein
Fragments
Stochastic Optimization
Monte Carlo search of chemical space:• Start from initial drug• Make random state changes
(add/delete/move chemical group)• Accept up-hill moves with probability
dictated by “temperature”• Reduce temperature after each move• Stop after temperature gets very small
Summary
Virtual drug screening• Ligand-based
§ 2D structure matching§ 3D structure matching§ QSAR
De novo drug design• Models
§ Simulation§ Knowledge-based
• Construction algorithms§ Incremental construction§ Fragment-based§ Stochastic optimization
Discussion
?