Computer Aided Drug Design and Case...

Preview:

Citation preview

Dr. Haifeng Chen

Shanghai Jiaotong University

Computer Aided Drug Design

and Case Study

2013 Nobel Chemistry Prize

The prize was awarded for laying the foundation for the computer models used to understand and predict chemical processes.

Reference Book

Xiaojie Xu and Tingjun Hou. Computer Aided Drug Desing. Chemistry Industry Press, 2004.

Kaixian Chen. Computer Aided Drug Design: Principal, Method and Appication. Shanghai Science and Technology Press, 2000.

Textbook of Drug Design and Discovery. Edited by Povl Krogsgaard-Larsen, Tommy Liljefors, Ulf Madsen,published by Taylor & Francis,2002.

Reference Book

A. R. Leach. Molecular Modelling. Principles and Applications. Addison Wesley Longman, Essex, England, 2001.

Broad introduction to many aspects of molecular modeling and computational chemistry techniques, covering basic concepts, quantum and molecular mechanics models, techniques for energy minimization, molecular dynamics, Monte Carlo sampling, free energy simulations, and drug design applications

SCI Article of Undergraduate Student

1. Z. Li, J. Han, H. F. Chen*. Chem. Biol. Drug Des. 72:350-359, 2008.(2005)(Chinese Academy of Sciences)

2. Z. Li, H. Zhang, Y. Li, J. Zhang, H. F. Chen*. Chem Biol Drug Des 77:63-74, 2011. (2005)

3. F. Qin, Y. Chen, Y. X. Li, H. F. Chen*. J. Chem. Phys. 131: 115103, 2009. (2005)(SJTU)

4. H. Zhang, F. Qin, W. Ye, Z. Li, S. Ma, Y. Xia, Y. Jiang, J. Zhu, Y. Li, J. Zhang, H. F. Chen*. Chem Biol Drug Des 78:427-437, 2011. (2008)(Duke University)

5. G.W. Yan, Y. Chen, Y. Li, H. F. Chen*. Chem Biol Drug Des 79:916-925, 2012. (2011)(University of Michigan State)

6. S.Y. Ma, W. Ye, D. J. Ji, H.F. Chen*. Medicinal Chem. 9: 420

– 433, 2013. (2009) (Purdue University)

Pig Flu/A H1N1

From 1997- now

Avian Flu/H7N9

2002-2003

SARS

Human and animal infectious diseases

Plague (6th century) - Death rate : 30%~100%

Cholera (18th century) - Death rate : 30%~100%

Anthrax (19th century) - Death rate : 20% Ebola virus (1976)

- Death rate : 50%~90% HIV (1980)

- Death rate : 61% Mad cow disease (1985)

- Death rate: 100% Avian flu (1997)

- Death rate : 33.3% Pig flu (2009)……

H7N9 (2013)

10

Drug discovery of post genomics

Function genomics

Target discov

ery

Target evaluat

ion

Lead discovery

Lead optimization

Preclinic test

Clinical trail

Market

Drug development flowchart

New Chemical Entity

Structure optimization

Preclinical test (ADMET)

New drug research application

Clinical trail (I II III)

New drug application

Post market research

Drug design flowchart

Drug Discovery Today 7: 315-323 (2002)

Target Identification

Drug screening, Potential

drug discovery, side effects

Dise

ase

analysis

Clin

ic test

Virus analysis of avian flu

N Engl. J. Med. 2013,20:1888-1897.

Drug target

Success target : 300-400

Receptor Enzyme Ion channel Nucleic acid

Science 2013, 341:84-87.

Success Cases

HIV-1 Protease Inhibitors in the market:

Inverase (Hoffman-LaRoche, 1995)

Norvir (Abbot, 1996)

Crixivan (Merck, 1996)

Viracept (Agouron, 1997)

Drug discovery today 2, 261-272 (1997)

Merck HIV-1 protease drug (Crixivan)

1987

Sequence Function Clone

Crystal

Inhibitor research

Screening L-F35524

Human test

Clinical test for 4000 samples

FDA approve: 42 days

1987-1988

1989

1989

1989-1992

1993

1993-1996

1996.3.

17

Challenge of drug development

New chemical entity: Difficult Time: long (10-15 years) Cost : expensive (800 million US$) Method: do not speed (Combine Chem. & HTS) How to speed?

18

Challenge of drug development

19

Main methods of CADD

Statistics Math

Statistical mechanics

Quantum mechanics

QM/MM

Molecular mechanics

Molecular Dynamics

Monte Carlo Simulation

Enzyme catalyst

Conformer Search

Classic MD

Ab initio MD Newton Second Law

20

20 20

21

21

Most used technologies

23

Computer aided drug design Method

23

24

Molecule

Structure

Biological activity(Φ)

IC50, Ki …

Cl

Cl

Hasch (1962): Hansch analysis

Richard Cramer III (1987): CoMFA

Gerhard Klebe (1994): CoMSIA

Lowis (1997): HQSAR

Vapnik (1992/2001):SVM

Tin Kam Ho (1995/2003):Random Forest

QSAR,Quantitative Structural-Activity Relationship

25

Molecules Are Not Numbers!

26

Molecular Descriptors

Hansch classic QSAR

HQSAR(Holograph QSAR)

Fragment size

Number of fragments

Atom types

Bond types

Atom hybridization

Stereocenters

PLS analysis result

QSAR Comb. Sci. 23:36-55, 2004.

Principle and application

of 3D-QSAR

Method of 3D-QSAR

Bioactivity

3D-QSAR Model

3D-QSAR

CoMFA (Comparative Molecular Field Analysis )

CoMSIA (Comparative Molecular Similarity Indices Analysis )

The hypothesis condition of CoMFA

All molecules

Have same interaction mechanism with the same kind of receptor (or enzyme, ion channel,etc.)

Have identical binding sites in the same relative geometry.

Create a 3D database

Calculate charges for each of compounds

(Gasteiger-Hückel)

Minimize the structure (Tripos force field)

Calculate the steric and electrostatic field energies (Steric and electrostatic contributions were cutoff

to a value of 30 kcal/mol)

Do regression analyses (partial-least squares (PLS))

Perform using full cross-validation (leave one-out method)

r2 value (q2)

3D-QSAR steps

Contour maps

Alignment

Training set and test set (3:1)

Conformer

search

Conformer Search

Gridsearch

Multisearch

Random search

System search

Alignment Rules

Pharmacophore-based alignment Pharmacophore is a spatial arrangement of

atoms or functional groups which response for bioactivity

Structure-based alignment

MCSS(Maximum Common Substructure) or

skeleton structure

Dock-based alignment Active conformer could align together by the

results of molecular docking

Grid and probe atom

Probe atom

Box must cover the structures.

The type of probe atom

Sp3 C+

Sp2 O-

Sp3 N+

H+

Ca2+ …

Potential function of CoMFA

Partial Least Square

QSAR equation

PLS

Contour Maps

Predictions

QSAR Table = SYBYL MSS

Bio

Construct CoMFA Model

PLS Component

Field contribution

Steric and electrostatic contour plots

Relationship between EA and PA

Interpretation of CoMFA

For drug design, the most powerful tool is to find the relationship

between fields and bioactivity, then design

new lead compounds.

Green: bulk group

Yellow: small group

Red: negative charge

Blue: positive charge

Advantages of CoMFA vs Classical QSAR

Visualization

Higher predictive power

Truly three-dimensional, shape-dependant nature of CoMFA descriptors

CoMFA analyzes the interaction energy of an entire ligand rather than arbitrarily selected substructure of the ligand

CoMFA has been accepted by many as the ultimate solution to the problem of correlating chemical structure and biological activity

Shortcomings of CoMFA

CoMFA parameters do not include hydrophobicity

Need to specify initial “alignment rule” and “active conformation”

Often fail when a few molecules are very dissimilar from all others

The results from one CoMFA analysis are not easily compared with another one

Factors of influence CoMFA

Active conformer

Aligment rules

Probe atom

Lattice size

Orientation of alignment molecular set

Step size

CoMSIA

Steric fields

Electrostatic fields

Hydrophobic fields

Hydrogen bond donor fields

Hydrogen bond acceptor fields

Potential function of CoMSIA

CoMSIA is not sensitive to changes in orientation of

the superimposed molecules in the lattice.

Interpretation of CoMSIA

Yellow: hydrophobic ↑ ; White: Hydrophobic ↓

Cyan: hydrogen bond donor ↑; Purple: hydrogen bond donor ↓

Magenta:hydrogen bond acceptor ↑; Red: hydrogen bond acceptor ↓

Drug Design of HIV-1

Protease Inhibitor

Student : Guanwen Yuan(Shanghai

High School)

Supervisor: Haifeng Chen(SJTU)

Shanghai High school-SJTU Join Program

Content

Backgrounds 1

Methods 2

Results 3

Discussions 4

About AIDS

Infect : 60,000,000

Death : 30,000,000

1,050,000 (2008) 23.3 billion $ China

Up to now, no bacterin

WHO

http://www.cmt.com.cn/xshy/gr/AIDS2010/AIDS2010/201007/t20100714_263494.html

Vacinne design

Envelope trimer

Science 2013,DOI: 10.1126/science.1245627

AIDS of China

2006 2007 2008 2009 20100

2000

4000

6000

8000

10000

12000

14000

Num

ber

of D

eath

Year

HIV-1 Life Cycle

Nature Medicine

5, 740 - 742 (1999).

Integrase Inhibitors

HIV-1 drug target

CCR5 1IKY 2013 HIVRT 1HNI 1995 HIVIN 3L3V 2010 HIVPT 1HXW 1995

HIV RT inhibitor

HIV Protease inhibitor

HIV Integrase inhibitor

Cocktail therapeutics

AIDS

Cocktail Types

Drug resistance

New anti-HIV inhibitor

Therapeutic method

Combination of drugs

Dolutegravir(整合酶抑制剂)+ Abacavir (阿巴卡韦)(非核苷HIVRT inhibitor)+ Lamivudine(拉米夫定)(核苷类似物)(DTG-ABC-3TC)

EFV (非核苷HIVRT inhibitor)(泰诺福韦)-TDF (核苷类HIVRT inhibitor)(替诺福韦酯)-FTC(核苷类HIVRT inhibitor)(恩曲他滨)

N Engl J Med. 7, 2013; DOI: 10.1056/NEJMoa1215541

Background

HIV Protease (HIVPR)

HIV-1 codes p55 and p60

HIVPR can break pre protein and activate

protein. Specific enzyme of virus.

HIVPR is the key enzyme of mature for

HIV-1 virus.

Drug target

Binding mode between inhibitor and HIVPR

作用机理:

抑制剂与酶结合→使酶失去催化活性→阻断HIV在体内的复制→

抗AIDS的药效

PNAS 109:20449–20454, 2012.

Simulation open and close of HIVPR

Motion with PCA

Research Methods

Computer Aided Drug Design

Molecular Dynamics (MD) Simulation

3D-Quantitative Structure-Activity

Relationship (3D-QSAR)

Data Set

Bioorg Med Chem Lett 2003, 13, 3601-3605.

Molecular dynamics simulation

Molecular dock: M17-HIVPR, M35-HIVPR

AMBER8.0 & Parm99SB force field

5000ps simulation - 298K

Result analysis

Hydrogen bond

Hydrophobic interaction

Binding free energy

Molecular Dock

Calculating the binding free energy

Finding the molecular mechanism between ligand and receptor

Finding the relationship between bioactivity and binding free energy

Using with other methods(Find active conformer to build CoMFA model)

Virtual screening

The AutoDock Software

Developed by AJ Olson’s group in 1990.

AutoDock uses free energy of the docking molecules using 3D potential-grids

Uses heuristic search to minimize the energy.

Search Algorithms used: Simulated Annealing

Genetic Algorithm

Lamarckian GA (GA+LS hybrid)

Docking complex

Ligand

Receptor

Docking preparing - ligand

Assign charges

Align with tempale molecule

Define rotatable bonds

Merge non-polar hydrogens

Write .pdbq ligand file

Receptor preparing

Delete water and ligand in complex

Add polar hydrogen

Load charge (Kollman_all)

Minimize hydrogen atom

Add solvation parameters

Write .pdbqs protein file

Autodock result

Correlation between Binding Free Energy and bioactivity

-16 -15 -14 -13 -12 -11 -10 -9 -8 -7

4

5

6

7

8

9

Lo

g(1

/EC

50

)

Binding free energy (kcal/mol)

pIC50=0.759 - 0.503*△G (n=76, r=0.739, F1,75=89.217, SD=0.861)

3D-QSAR Study

Construct structure-bioactivity model, predict drug bioactivity and virtual screen

Molecular alignment, then calculate force parameter

PLS parameter represents the quality of model. R2 ,accuracy

Test set can evaluate model

Contour plot

3D-QSAR Study

SYBYL7.0 program package

Training set & Test set (3:1)

(Random class)

Molecules Alignment

CoMFA & CoMSIA models

Results

Molecular Dynamics Simulation

The most important hydrophobic and

hydrogen bonding interactions

Action Mechanism of Inhibitors

Stability of complex

0 25 50 75 100 125 150 1750.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

RM

SF

)

Residues

M17-HIV PR

M35-HIV PR

M35 < M17

Hydrogen bond

0 1000 2000 3000 4000 50002345678345678

Time (ps)

Asp30-F3(M35)

Asp25-O1(M35)B

Dis

tance (

Å)

Ala28-O6(M17)

Asp25-O6(M17)

A

Bioactivity:M35>M17 M35/HIVPT: Two strong hydrogen bonds M17/HIVPT: Two weak hydrogen bonds

Hydrophobic interaction

1 2 3 4 5 6 70

20

40

60

80

100

Popula

tion (

%)

Native contact

M17

1 2 3 4 5 6 7 8 910

M35

Bioactivity:M35>M17 M35/HIVPT: 10 M17/HIVPT: 7

Binding free energy

Binding mode and key residue

Common residues: Ile50(A) Ile50(B) Asp25(A) Ile83(A) Ala28(A)

Conclusion of MD simulation

Similar action mechanism both systems have hydrogen bond with catalytic

residue of Asp25 of HIVPR

Key residues: Ile50(A), Ile50(B), Asp25(A), Ile83(A),

and Ala28(A)

Hydrogen bond offered by the OH

Strong hydrophobic interaction offered by the benzene ring

Result of 3D-QSAR

Force Combination of CoMSIA

406080

SEHDASEDASEHDSEHASEASEDSEH

F

Force fieldSE

6

8

1/S

EE

0.50.60.7

R2

0.8

1.0Q

2

Best CoMSIA model:SEHA

Prediction ability

1.5 2.0 2.5 3.0 3.51.5

2.0

2.5

3.0

3.5

4.0

2.0

2.5

3.0

3.5

Ca

lcu

late

d a

ctivity

Experimental activity

CoMSIA

Training set

Test set

CoMFA

r2 of test set : 0.939 (CoMFA) 0.825 (CoMSIA)

CoMFA & CoMSIA

X: bulk & positive

charge

Y: small volume &

positive charge

Contour plot analysis

Field + —

Steric green yellow

Electro

static

blue red

Contour plot of CoMSIA

X: hydrophilic substitute

Y: hydrophobic & hydrogen bond donor

Field + -

Hydrophob

ic

Orange white

Hydrogen

bond

acceptor

cyan purple

Conclusion of 3D-QSAR

X: bulk and positive groups

M30(4-COOCH3) > M28(4-CH3) > M29(4-CN) > M31(4-COOH); M1(4-H) > M21(4-F)

Y: small and positive charge groups

M14(4-CH3) > M15(4-CN) > M13(4-CF3) > M16(4-COOCH3) > M18(4-CH2OH) > M20(4-CONH2) > M17(4-COOH).

X:hydrophilic group

M32(4-CH2OH)>M30(4-COOCH3); M32>M31(4-COOH); M32>M28(4-CH3); M32 >M29(4-CN)

Y: hydrophobic and hydrogen bond donor group

M16(4-COOCH3)>M17(4-COOH); M14(4-CH3)> M19(4-CH2NH2)

OO

O

OO

O

X

X

Y

Y

MD vs 3D-QSAR

MD: two hydrogen bonds between M35(F3/O1)and Asp25/Asp30

3D-QSAR: hydrogen bond favour regions F3/O1

MD: M35 has hydrophobic interactions with Ile50(A), Ile50(B), Ile83(A), and Ala28(A).

3D-QSAR: Benzene is covered by hydrophobic favour regions.

The result of MD is consistent with that of 3D-QSAR.

Comparison with previous work

1.5 2.0 2.5 3.0 3.5 4.01.5

2.0

2.5

3.0

3.5

2.0 2.5 3.0 3.5 4.0

Pre

dic

ted

activity

Experimental activity

This work

r2 = 0.970

Previous work

r2 = 0.867

997.2219.0189.0

461.0116.0246.0151.0191.0

5156.0252.0148.01198.0246.0

____

_______

_____

OFYFdiY

pdoHbondYpYOXFYFX

pYpYpYpXpX

II

IINNN

BmrBPA

Quality: same Quantity:better

Conclusion

MD simulation suggests interaction mechanism, key hydrogen bond and hydrophobic interactions.

3D-QSAR method constructs robust prediction model.

The results of MD are agreement with those of 3D-QSAR.

Better than previous works.

Pulications

Insight into the Binding Mode of HIV-1 Protease Inhibitor Using Molecular Dynamics Simulation and

3D-QSAR. Chem Biol Drug Des 2012. (IF=2.46)

Insight into the Stability of Cross-β Amyloid Fibril from Molecular Dynamics Simulation. Biopolymers 93: 578-586, 2010. (IF=2.82)

Conformational Selection or Induced Fit for Brinker and DNA Recognition. Physical Chemistry Chemical Physics. 2011 (IF =4.06)

Forecoming research of HIV-1 protease

Complex of HIV-1 PR and nelfinavir

Mutant residues: V32I I50V/L I54M/V I84V L90M A71V

Drug resistant?

Thank You!

Recommended