Miroslav Brumovský 28th July 2011 1/12

Miroslav Brumovský

28th July 2011

Methods using polarization for in silico fragment-based drug design

Miroslav Brumovský 28th July 2011 2/12

Outline

• theoretical background• goal of the project• methods used• results

Miroslav Brumovský 28th July 2011 3/12

Drug design

• inventive process of finding new medications• key process is screening libraries of potential drug compounds

(in vivo essays x virtual screening)• this can be done using two approaches

– screening of huge libraries of complete compounds – screening of drug fragments → fragment-based drug design

• fragment-based drug design benefits from better sensitivity and limited need of screened molecules than the classical approach

Miroslav Brumovský 28th July 2011 4/12

Molecular docking

• most popular method for virtual screening• predicts the energy and orientation of ligand molecules with a

receptor in a stable complex (“lock-and-key” problem)• algorithms treating flexible ligand and rigid receptor → best

results• results are sorted by scoring function (relative energy)• scoring function usually calculated using force-field atomic

charges (no polarization)• when used in FBDD, more accurate (polarized) atomic charges

for energy calculations are needed

Miroslav Brumovský 28th July 2011 5/12

Goal of the project

• improvement of standard molecular docking for more accurate prediction of the ligand-receptor interaction using methods based on polarization

• development of a new method applicable in computer-assisted drug design

Miroslav Brumovský 28th July 2011 6/12

Methods

• standard molecular docking (Glide)– standard atomic charges assigned from force-field

• docking with polarized QM/MM ligand charges– ligand atomic charges computed by QM calculations with

standard receptor charges (B3LYP/6-311+G*)

• docking with polarized both receptor (MM charges) and ligand (QM/MM charges)– ligand atomic charges computed by QM (B3LYP/6-311+G*)

with receptor atomic charges (MM) computed by Method of induced charges

Miroslav Brumovský 28th July 2011 7/12

Model systems

1EQG 1FV9 1GWQ 1N1M

1QWC 1WCC 1YZ3

2C90 2OHK 2JJC

Congreve, M et al. (2008) J. Med. Chem, 51, 3661-3680.

1 2

5

3 4

6

7

910 11

121S39

8

2ADU

Miroslav Brumovský 28th July 2011 8/12

Results

Method used

M1 M2 M3

Pos

ition

0

1

2

3

4

5

6average position of the correct pose

Method used

M1 M2 M3

RM

SD

val

ue

0,0

0,5

1,0

1,5

2,0

2,5

average RMSD of the 1st pose average RMSD of the 1st correct pose

Methods: M1 – standard docking, M2 - docked with QM/MM charges, M3 - docked with MM polarized QM/MM ligand charges

RMSD < 2 Å

Miroslav Brumovský 28th July 2011 9/12

Example – docking of 1N1M

A. Standard docking, docking with polarized QM/MM ligand charges

B. Docking with polarized both receptor (MM charges) and ligand (QM/MM charges)

RMSD = 1.4 Å RMSD = 0.9 Å

Miroslav Brumovský 28th July 2011 10/12

Conclusions

• improved methods for docking are more accurate

• effect of receptor polarization is not so relevant comparing to the calculations with lower QM basis set

• this advanced approach can be succesfully used in fragment-based drug design

Miroslav Brumovský 28th July 2011 11/12

Acknowledgement

I would like to say thanks to…my supervisor, Dr. David Řeha, for very educational and friendly leading,University of Essex for providing computational software,organizers of Schola Ludus for the opportunity to participate on the project.

Miroslav Brumovský 28th July 2011

Thank you for your attention