fragment-based tailoring of compound libraries for high ...kolblab.org › pdfs ›...

Fragment-based tailoring of compoundlibraries for high-throughput docking and

screening

Peter Kolbpkolb@bioc.uzh.ch

kolb@blur.compbio.ucsf.edu

Caflisch Lab

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.1/28

Acknowledgements[Acknowledgements ]

� Amedeo Caflisch, Marco Cecchini, Andrea Cavalli, Stefan

Jelakovic, Shaheen Ahmed, Francesco Rao, Enrico Guarnera,

Riccardo Pellarin, Steffi Muff, Raffaele Curcio, Dariusz Ekonomiuk,

Pietro Alfarano

� The Dockers

� Catherine Berset Kipouros, Stephan Audétat, Urs

Lüthi, Alcide Barberis

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.2/28

Acknowledgements[Acknowledgements ]

� Amedeo Caflisch, Marco Cecchini, Andrea Cavalli, Stefan

Jelakovic, Shaheen Ahmed, Francesco Rao, Enrico Guarnera,

Riccardo Pellarin, Steffi Muff, Raffaele Curcio, Dariusz Ekonomiuk,

Pietro Alfarano

� The Dockers

� Catherine Berset Kipouros, Stephan Audétat, Urs

Lüthi, Alcide Barberis

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.2/28

Acknowledgements[Acknowledgements ]

� Amedeo Caflisch, Marco Cecchini, Andrea Cavalli, Stefan

Jelakovic, Shaheen Ahmed, Francesco Rao, Enrico Guarnera,

Riccardo Pellarin, Steffi Muff, Raffaele Curcio, Dariusz Ekonomiuk,

Pietro Alfarano

� The Dockers

� Catherine Berset Kipouros, Stephan Audétat, Urs

Lüthi, Alcide Barberis

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.2/28

Outline[Outline ]

� Fragment-based Docking

�DAIM

� SEED

� FFLD

� Anchor-based Library Tailoring

� Concept

� Application to EphB4

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.3/28

Fragments?[Fragments? ]

“A fragment is a fragment is a fragment.”

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.4/28

DAIM–SEED–FFLD[DAIM–SEED–FFLD ]

DAIM

N

H

ON

F

FF

F

FF

O

N

O N

N

H

H

DAIM SEED FFLD

J Med Chem. 2006;49:7384-92 Proteins. 1999;37:88-105 Biol Chem. 2001;382:1365-72

Proteins. 2001;42:256-68 J Comput Chem. 2004;25:412-22

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.5/28

DAIM | Decomposition And Identification ofMolecules

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.6/28

DAIM: decomposition[DAIM: decomposition ]

DAIM identifies fragments in a four-step process:

1. identification of rings.

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.7/28

DAIM: decomposition[DAIM: decomposition ]

DAIM identifies fragments in a four-step process:

1. identification of rings.

2. definition of initial fragments: atoms connected by “unbreakable”

bonds (double, triple, aromatic, amidic, terminal, ring).

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.7/28

DAIM: decomposition (2)[DAIM: decomposition (2) ]

H

F

F

F

F

F F

HN

NH

O

N

ON

+N

O

H

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.8/28

DAIM: decomposition[DAIM: decomposition ]

DAIM identifies fragments in a four-step process:

1. identification of rings.

2. definition of initial fragments: atoms connected by “unbreakable”

bonds (double, triple, aromatic, amidic, terminal, ring).

3. reconnection of functional groups.

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.9/28

DAIM: decomposition (3)[DAIM: decomposition (3) ]

+N

O

HH

F

F

F

F

F F

HN

NH

O

N

ON

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.10/28

DAIM: decomposition[DAIM: decomposition ]

DAIM identifies fragments in a four-step process:

1. identification of rings.

2. definition of initial fragments: atoms connected by “unbreakable”

bonds (double, triple, aromatic, amidic, terminal, ring).

3. reconnection of functional groups.

4. completion of valences.

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.11/28

Possible triplet combinations[Possible triplet combinations ]

CH4

CH4

CH4

FF

F

F FF

O

N

H

H

HN

NO

N

+N

O

H

We need 3 fragments [ ],

but there are 9 to choose from

→

(93

)

= 9!3!(9−3)! = 84

possible combinations

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.12/28

Chemical richness ρχ[Chemical richness ρχ ]

6 1 0 0 0 0 0 0 0 0 2 1 6 7 0.545298614

vectorsDonor

H

H

H

HH

H H

W4/1000amSHalNAtoms

CO P ar Acceptor

N

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.13/28

Chemical richness ρχ[Chemical richness ρχ ]

6 1 0 0 0 0 0 0 0 0 2 1 6 7 0.545298614

vectorsDonor

H

H

H

HH

H H

W4/1000amSHalNAtoms

CO P ar Acceptor

N

DAIM uses the fingerprint to choose fragments based on the chemical

richness ρχ = ∑i fi and to calculate the octanol/water partition coeffi-

cient (CLogP).

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.13/28

Docking performance: DAIM↔ random[Docking performance: DAIM↔ random ]

84565656120358484205684351020561204204165165441201203520202084102010101056

1hte1b6l

1ets1ppc

1uvs1dbm1dwc1hwr

1ett1bmn1etr2dbl

1ejn7upj

3tmn1atl

1aoe1pph

2tsc1ela

1fkg1c83

1c841mcj

1mnc1hgh

1nnb1nsc

1nsd5tln

1xig4hmg

1tpp2acs2xis

1hgj

0 0.2 0.4 0.6 0.8 1relative number of cases with RMSD below 2Å

0 0.2 0.4 0.6 0.8 1relative rank of the DAIM triplet

tripletsn

below 2Å above 2Å

DAIM rank

n

Using the DAIM triplet re-sults in at least one poseclose to the X-ray structure in20/36 cases. The probabilityto be below 2 Å at least 20times through random selec-tion of triplets is < 0.28%.

J Med Chem. 2006;49:7384-92

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.14/28

SEED | Solvation Energy for ExhaustiveDocking

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.15/28

SEED[SEED ]

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.16/28

SEED (II)[SEED (II) ]

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.17/28

SEED (III)[SEED (III) ]

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.18/28

ALTA | Anchor-based Library Tailoring

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.19/28

ALTA[ALTA ]

12

5

Selection basedon target structure

3

6

Flexible ligand docking

4

Pool of fragmentsPool of compounds

Decomposition

Fragment docking

AIMDAIMD

top−ranking Screening with

fragments

AIMD

Subset of compounds

Candidate fragments

Fragment ranking

Final ranking (LIECE)

AIMDAIMD

AIMD

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.20/28

Application: EphB4

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.21/28

Introduction[ Introduction ]

EphB4

� belongs to the largest family of receptor tyrosine kinases.

� corresponding transmembrane ligand: ephrinB2.

� probably involved in the control of interaction of endothelial cells

→ role in angiogenesis and tumorangiogenesis.

� elevated expression levels in prostate cancer.

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.22/28

EphB4 structure[EphB4 structure ]

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.23/28

Statistics[Statistics ]

ρχ1353335513

52351205

30

7

21418

8849

9960

docked a second time

fragments selected

forming 2HBs

fragments

molecules docked

1 or 2 HB

tested

show some inhibitoryactivity on EphB4

728202total library size

more rings, min1 acce, 1 dono, 1 or

#1: IC50 1.36 µM∗

#2: IC50 76 µM

#3: 20% @ 4.6 µM

#4-7:

15-40% @ 125 µM∗has been developed into a series of in-

hibitors with IC50s between 1 and 50 µM

Proteins. 2008;73:11-18

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.24/28

The hits[The hits ]

enzymatic assays cell-based assay

compound structure MW IC50 IC50 IC50 CHO predicted ∆G

[Da] [µM] [µM] [µM] % [kcal·mol−1]

1

H2N

N

N

O

O

OS

H

H

349 76 n.d. n.d. n.d. −8.8

2-8OO

O

N N

NN

R

HN

R=

2 –(CH2)4OH 353 1.4, 1.8 6.8 5.6, 7.9 neg. −11.0

3 –(CH2)3CH3 337 1.5, 1.5 8.3 n.d. neg. −10.7

4 –(CH2)2CH3 323 1.9 n.d. n.d. neg.

5 ortho-methoxy-phenyl 387 27 n.d. n.d. 14% @ 20 µM

6 –(CH2)3OCH3 353 50 n.d. n.d. 28% @ 20 µM

7 –(CH2)2Ph 385 30% @ 14 µMe n.d. n.d. 34% @ 20 µM

8 –(CH2)2-meta-methylphenyl 399 29% @ 5 µMe n.d. n.d. 14% @ 20 µM

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.25/28

Fragment- vs. compound pose[Fragment- vs. compound pose ]

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.26/28

CPU time requirements[CPU time requirements ]

ALTA hours

DAIM 2

SEED (fragments) 2200

substructure search 1000

docking (DSF) 3300

Σ 6500

Docking only hours

DAIM 2

SEED 10000

FFLD 119000

Σ 129000

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.27/28

CPU time requirements[CPU time requirements ]

ALTA hours

DAIM 2

SEED (fragments) 2200

substructure search 1000

docking (DSF) 3300

Σ 6500

Docking only hours

DAIM 2

SEED 10000

FFLD 119000

Σ 129000

factor of 20!

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.27/28

Conclusions[Conclusions ]

�DAIM can generate chemically sensible molecular fragments and

rank them according to their suitability to serve as anchors in

docking

� ALTA is a technique to tailor docking libraries on the fly by

selecting molecules based on the suitability of their fragments for

a given binding site. This reduces computation time while

preserving diversity.

236th ACS National Meeting, Philadelphia, August 18, 2008 – p.28/28