Transcript Bioorganic & Medicinal Chemistry Letters 14 (2004) 1447

Bioorganic & Medicinal Chemistry Letters 14 (2004) 1447-1454
HIV-1 integrase pharmacophore
model derived from diverse
classes of inhibitors
Gabriela Iurcu Mustata,a Alessandro Brigob and James M. Briggsa,*
-Presentation by Kavitha Bharatham
Introduction to HIV-1 Integrase
Life cycle of HIV-1
A movie2,3
Structure of HIV-1 Integrase
Catalytic core (A),
N-terminus (B),
C-terminus (C)
The Integration of HIV-1 DNA
Integration4 occurs in three consecutive steps
3' processing
Removal of two nucleotide at 3 ends
Exposure of CA dinucleotide
Strand transfer
Joining of the previously processed 3'
ends to the 5' ends of strands of target
DNA at the site of integration
Disintegration
Repair of the small gaps in the target
sequence flanking the viral genome and
joining of the 5`-ends of viral DNA to
host DNA
Introduction to Pharmacophore
• One of the most analogue based drug design techniques is the generation
of Pharmacophore models
아날로그에 기초를 둔 약 디자인 기술의 하나는 Pharmacophore 모형
의 형성이다
• A Pharmacophore is a simplified 3-D description of the key structural
features of a set of known compounds.
Pharmacophore는 한 세트이 알려진 성분들의 중요한 구조상 특징의
간단한 3-D 묘사이다
• The structure of a chemical influences its properties and biological activity.
“Similar compounds behave similarly”
화학제품의 구조는 그것의 특징 및 생물학적 활동을 좌우한다
“동이한 화합물은 동이하게 행동한다”
• Catalyst1 generates a pharmacophore hypothesis that represents the
structure-activity relationships from a set of compounds.
Catalyst1은 화합물의 세트에서 구조 활동 관계를 대표하는
pharmacophore 가설을 생성한다.
Pharmacophore Generation Methodology
431 molecules with HIV-1 integrase activity data
Selected based on criteria
26 structures as training set were sketched
Conformational search
High active
IC50<=10 uM
Moderately active
100 uM>IC50>10 uM)
Pharmacophore Generation
Inactive
IC50>=100 uM
Selection of Training Set
• Set of molecules taken to generate a pharmacophore is a Training set
• Molecules must have same assay
Ex: 50% inhibitory activity against HIV-1 integrase
• Training set must covers 4 orders of magnitude of biological activity data
(IC50 ranges from 0.04-1000).
Ex: 0.01-0.1; 0.1-0,0-10,10-100 represents each magnitude
• Each order of magnitude is represented by at least three compounds
•
If two compounds had similar
structures, they had to differ in
activity by one order of magnitude
to be included in our dataset.
S
If two compounds were found to have
similar activities, they had to be
structurally distinct in order to be
included.
+
N
N
N
O
IC50 =705 uM
IC50 =57 uM
H
•
O
S
Cl
S
N
H
O
Cl
N
IC50 =110 uM
O
IC50 =115 uM
H
Training Set Molecules
Pharmacophore Generation
n-Butane
H
H
H
H
H
H
H
H
H
H
H
H H
H
H
H
H
H H
H
H
H
H
H
After Conformational search (maximum 256), pharmacophore is generated based
on the following phases
Constructive Phase:
• Identifies most active compounds
• Identifies hypothesis that are common among active compounds
Subtractive phase:
• Identifies most inactive compounds
• Removes hypothesis that are common among inactive compounds
Optimization Phase
• Attempts to improve the initial hypothesis
Pharmacophore Hypothesis Generation
O
O
높게 활동
O
O
O
O
IC50 = 0.059
H
Hydrogen bond Donor (HD)
Hydrogen bond Acceptor (HA)
N
IC50 = 0.04
N
O
O
O
H
O
O
O
Negative Charge (NC)
H
H
Hydrophobic (HY)
IC50 = 325
Pharmacophore Hypothesis
HY
12.05 Å
10.29 Å
5.32 Å
HA2
6.98 Å
3.6 Å
HA1
3.0 Å
HD
Criteria for a good hypothesis
The HypoGen module in catalyst performs
• A fixed cost calculation, which represents the
simple model that fits all data perfectly.
Perfect
Pharmacophore
Fixed cost is Minimum
Ex:72 bits
• A null cost calculation, which presumes that there
Very Bad
is no relationship in the dataset and that the
Pharmacophore
experimental activities are normally distributed
Null cost is Maximum
around their average value.
Ex:149 bits
A meaningful pharmacophore hypothesis may result when the difference
between these two values is large
[Very Bad Pharmacophore (null cost)]-[Perfect Pharmacophore (Fixed
cost)] > 40 bits5
149-72 = 77 bits
Good hypothesis
The total cost of any pharmacophore hypothesis should be close to the fixed
cost to provide any useful models.
Results and Discussion
Ten pharmacophore
hypothesis are generated
Null Cost = 182 bits
Fixed Cost = 106 bits
Null Cost – Fixed Cost = 72 bits
Our Hypothesis is good!
Total cost = cost taken for
generated hypothesis
Total Cost must be near Fixed
cost
Results and Discussion
Validation of the pharmacophore model
•
•
•
•
Molecules used for validation are called Test Set
14 highly active molecules were selected
They were built, minimized and conformers were generated
They were mapped onto the pharmacophore hypothesis and activities were
predicted
Validation of the pharmacophore model
Conclusions
• Our pharmacophore was able to accurately predict known inhibitors,
including the recently published azido-containing HIV-1 IN inhibitors6
• The mapping information based on the pharmacophore model we
developed is now being taken advantage of in the identification of novel
lead compounds with improved inhibitory activity through 3-D database
searches.
• These hypotheses thereby save valuable time in the laboratory
References
1)
2)
3)
4)
5)
CATALYST 4.6; Accelrys, Inc., San Diego, CA, 2001, http://www.accelrys.com
Asante-Appiah, E.; Skalka, A. M. Antiviral Res. 1997, 36, 139.
Hindmarsh, P.; Leis, J. Microbiol. Mol. Biol. Rev. 1999, 63, 836.
Brown, P. O.; Co.n, J. M.; Hughes, S. H.; Varmus, H. E. Retroviruses; Cold
Spring Harbor Lab. Press: Plainview, New York, 1997; 161.
Guner, O. F. In Pharmacophore Perception, Development, and Use in Drug
Design; Ed.; International University Line: La Jolla, California, 2000, pp
173.188