Cdc25 and cancer: molecular modelling approaches for identification of a chemical start point for drug discovery

David Mann 1 & Caroline Low 2 1 Molecular Cell Biology 2 Drug Discovery Centre

Cdc25 Phosphatases

Cdc25 P P Inactive Cdk Cyc P T P Y Cdk ADP Wee1 ATP T Y Cdk Active Cdk Cdc25A Cdc25B 60% identical over catalytic domain Cdc25C

Cdc25 Phosphatases

Cdc25B Cdc25C Cdk 1 Cdk 2 Cdk 2 Cdc25A Cdk 4 ?

Cdc25A Cdc25B Cdc25C 60% identical over catalytic domain

Misregulation in cancer

Nature Reviews Cancer 7 (2007) 495-507

Causal relationship with cancer

Cdc25 phosphatases as potential human oncogenes.

Galaktionov K, Lee AK, Eckstein J, Draetta G, Meckler J, Loda M & Beach D.

Science 269 (1995) 1575-7.

Cdc25A Ras* Cdc25B Ras* Cdc25C Ras*

Causal relationship with cancer

Cancer Res 67 (2007) 6605-11

Cdc25 and cancer

•Over-expressed in many tumour types •Acts as classical ‘co-operating’ oncogene •Reduction inhibits cellular transformation •Alternative to kinases

Where do we start?

Enzyme structure known?

Bioassays available?

Any known ligands?

What do we know about the structure of Cdc25 Structure-based design of Cdc25 inhibitors hampered by • shallow active site region exposed to bulk solvent • nucleophilic reactivity of the thiolate anion of the catalytic cysteine residue.

Swimming pool Catalytic site Cdc25B: 1QB0.pdb

CDK interaction site

Quinones: irreversible Cdc25 inhibitors

Vitamin K3 BN82685

Cdc25B IC50 3.8

m M

IRC-083864/Debio-0931

Cdc25A: 23 nM Cdc25B: 26 nM Cdc25C: 23 nM Quinones arrest cell cycle by •oxidation of Cys in catalytic site •irreversible reaction with Cys

Quinone inhibitors vs standard treatment

Pancreatic Cancer xenografts

No Treatment Vehicle IRC-083864 (i.v.) Gemcitabine (i.p.)

Brezak et al, (2009), Int. J. Cancer, 124, 1449

Initial approaches: modify existing reversible inhibitors

(1) Korean Patent (2) Natural Product (3)Quinones Dysidiolide Cdc25A IC 50 >100

m

M

Small set of reversible inhibitors known

(1)

Assay

Cdc25B

IC50 (

2.0

m

M)

Kim et al WO2006/101307

Assay

Cdc25A, B, C

(3)

PITT-9131

IC50 (

m

M)

5-10

Brisson et al (2004), Mol. Pharm., 824 (2)

Assay

MBP-Cdc25B3

IC50 (

m

M)

13.0 ± 0.5

Montes et al (2008), J. Chem. Inf. Model ,157

Where did they come from?

(2)

Virtual screen

Total compounds docked 310,000 Compounds tested IC 50 < 100 m M Hit rate 1,500 11 0.73%

FRED, Surflex, LigandFit Montes et al (2008), J. Chem. Inf. Model ,157 (3)

Physical screen

Total compounds tested IC 50 < 10 m M Hit rate 10,000 23 0.23%

PRIME collection (ChemBridge) Brisson et al (2004), Mol. Pharm., 824

Where do we start?

Enzyme structure known?

Bioassays available?

Any known ligands?

Modelling with field points De Novo Design Scaffold Hopping Virtual Screen QSAR Library Design • • Ligand based approach to find novel antagonists for GPCRs • • Problem 1 - few known ligands Problem 2 - no X-ray data (until 2007) Collaboration with Andy Vinter at James Black Foundation • 3 clinical candidates developed with this approach • 2002 Cresset founded to exploit virtual screening (www.cresset-group.com)

Thrombin X-rays

PPACK

Proteins don’t see ligands in the same way as chemists

BM14.1248

PDB reference codes PPACK: 1PPB BM14.1248: 1UVT

Why do we need field points?

Thrombin inhibitors

cLogP H-bond donors H-bond acceptors 2D similarity

BM14.1248

PPACK

D-Phe-Pro-Arg-CH 2 Cl

0.24

3.10

2 5 0.17 (Tanimoto) 5 5

The 3D Field Overlay Principle

Add field points to each structure

Negative Positive Surface Shape

The 3D Field Overlay Principle

Compare individual sets of field points

The 3D Field Overlay Principle

The 3D Field Overlay Principle

rms fit to crystal structure 0.76

T.Cheeseright et al (2006),J. Chem. Inf. Mod., 665

Create new class of reversible Cdc25 inhibitor using field points •Create single model from 3 different ligands •Dissect out field point pattern for one compound Model Virtual Screen •Use as pharmacophore probe for virtual screen •Hunt for compounds with similar field point patterns Test •Purchase commercial compounds suggested •Test compounds in enzyme bioassay

Pairwise comparisons can pull out the common features of all three molecules 1

200 conformations Energy cut-off 6 kcal/mol

2

111 conformations 18

3

conformations

Summarise common biology with field points

1 (conf 81) 2 (conf 5) 3 (conf 2) Field point template (A)

•

Two other solutions identified

Defining virtual screening input

(1) (2) (3)

Template

A B C

Compound 1

81 81 81

Compound 2

5 8 8

Compound 3

2 4 16

High throughput virtual screening to identify novel series

1

~100,000,000

Fieldscreen Database List of commercially available compounds

Fieldscreen results • First screen gave trivial analogues of seed • Top 200 were analogues of Compound 1 • 989/1000 were pyrazoles • So ran screen again WITHOUT pyrazoles in Fieldscreen database • This time chose top 100 hits …….

Processing the 2 nd hitlist

100 compounds

Including 3 from 1 st list

No structural similarity to any known actives.

MW range 250-350

40 available for purchase 35 arrived & tested 7 active (20 50

m

M)

20% hit rate

Initial thiazole hits from virtual screen (

1

) MW 484

T5896241

MW 337 Cdc25B IC 50 2.3 m M • Selective against related phosphatases • PTP1B, MKP-1 & 3 and alkaline phosphatases • Cellular target confirmed (n=1) • predicted increase in phosphorylated CDK2 • Later compounds amongst most potent reversible Cdc25 inhibitors described Cdc25A IC 50 Cdc25B IC 50 Cdc25C IC 50 35.5 ± 0.1 m M 17.2 ± 0.1 m M 47.3 ± 0.1 m M

Summary of project to date 1.

Created single model from three different chemotypes with FieldTemplater 2.

Identified bioactive conformations 3.

4.

Used one field point pattern as probe for virtual screen (FieldScreen) Found compounds active in vitro at m M concentrations 5.

Identified new chemotype for Cdc25 inhibitors 6.

Series under development • • • • Composition of matter patent filed Synthesis of analogues underway to explore SAR In vitro enzyme assay in place Cell proliferation assays in place

31

Thanks to James Collins Michelle Heathcote Hayley Cordingley Cathy Tralau-Stewart Albert Jaxa-Chamiec Alan Armstrong Katie Chapman Kate Judd Kathy Scott Pascale Hazel Funding from: Andy Vinter Mark Mackey Tim Cheeseright www.cresset-group.com

Figure 5. From (1) Brezak et al, (2009), Int. J. Cancer, 124, 1449-1456) Growth inhibition of xenografted tumors in nude mice treated with IRC-083864. (a) Cells of the human pancreatic carcinoma cell line MIA PaCa-2 were injected subcutaneously into the flank of female athymic mice. Tumors were allowed to reach a volume of 100 mm3. Once tumors were established, treatment was started by intravenous route as 10 mg/kg once a week for 4 weeks (qwk Paclitaxel (20 mg/kg, qodx5, iv) was used as current standard care.

× 4). Gemcitabine was used as current standard treatment. (b) Cells of the human prostate carcinoma cell line LNCaP were injected subcutaneously into the flank of female athymic mice. Tumors were allowed to reach a volume of 150 mm3. Once tumors were established, treatment was started by the oral route at 70 mg/kg for 2 days on /5 days off/ 2 on / 5 off /1 on.

Solubility is a problem with some initial hits Cdc25B IC 50 No detergent 2.3 m M (

1

) MW 484

T5896241

MW 337 A B C

Cdc25 isoform IC 50 No (uM) detergent N=3

2.4 ± 0.3

8.9 ± 0.5

10.2 ± 0.3

IC 50 (uM) With detergent N=4-7

35.5 ± 0.1

17.2 ± 0.1

47.3 ± 0.1