György M. Keserű H2020 FRAGNET Network Hungarian Academy of Sciences

Transcription

1 Fragment based lead discovery - introduction György M. Keserű H2020 FRAGET etwork Hungarian Academy of Sciences

2 Hit discovery from screening Druglike library Fragment library Large molecules (MW>300) Small molecules (MW<300) Large library (10 6 compound) Small library (10 3 fragment) Biochemical testing Biophysical testing Less diverse hits Diverse hits High affinity (1-10 mm) Low affinity (<100 mm)

3 What do we call fragments? Simple, small, polar molecules Rule of 3: MW < 300 ( heavy < 22) Log P < 3 umber of H-donors < 3 umber of H-acceptors < 3 Further properties: umber of rot. bonds < 4 Polar surface < 60 Å 2 Good solubility

4 Why fragments are beneficial? They have good physchem profile A smaller library contains more chemical information Their chemical space is smaller: better sampling They bind to protein hot spots They provide rational optimization towards drugs

5 Fragment space is smaller Druglike compounds Estimated number: Compounds in Chemical Abstract Registry: 113x10 6 Compounds commercially available: 10 6 umber of compounds screened: 10 5 Fragments Estimated number: (max 17 C,,, H, S, Cl atom):10 11 Fragments commercially available: 10 5 umber of compounds screened:

6 Sampling is more efficient HSP90 Clinical Candidates Fragment hits from a screen of 719 fragments H H H 2 H Cl H Me Me Me Me H 2 H H H H 2 H 2 H H S CF 3 H 2 H H 2 H H 2 H S H 2 J. Med. Chem. 2011, 54, Key features of all known clinical candidates are represented in hits from a small fragment library

7 Fragments bind to hot spots FRAGMETS FRAGMETS Fragments form limited number of polar interactions within a small region of protein binding sites Keserű JCIM 2012, Vajda PAS 2015, Keserű DDT 2017

8 Rational optimation strategies Linking Growing

9 What do we need for an FBDD program? Fragment library Screening technology Structural information (preferred) ptimization strategy

10 What are the general features of a good fragment? Interactions Diverse polar groups Pharmacophore variety Scaffold variety Physicochemical properties Size, complexity Shape Lipophilicity Solubility Synthetic vectors for growing Reactivity, stability, aggregation Synthetic tractability, cost MW = 278 clogp=1.84 Murray, Rees Angew. Chem. Int. Ed. 2015, 54, 2 7

11 Fragment sources Boyd, Kloe DDT Technologies 2010, 7, e173

12 How to design a library? Level 1 Compound properties Level 2 Sampling Level 3 MedChem Level 4 Experimental Physicochemical Reactivity Aggregator, assay interference Availability Priviledged structure Pharmacophore diversity Scaffold diversity Analogue accessibility Synthetic tractability Stability Purity Solubility Stability Cost Shape diversity Chris Swain (CMC): 1216 fragment hits, 240 publications, 174 molecular targets, 26 detection technologies

13 Physicochemical Fragment size and lipophilicity Libs Hits Libs Hits Chris Swain, CMC

14 Sampling of the fragment space Scaffold diversity Diversity Pharmacophore diversity 2 common bits, 3 bits in both reference and database mols, Tanimoto coeff. = 0.5 Hubbard et al. J Comput Aided Mol Des (2009) 23: common, 2 A-specific, 7 B-specific triangles Tanimoto coeff. = 0.29

15 MedChem and experimental MedChem considerations Multiple synthetically accessible vectors Should be synthesizable in <4 steps Analogues should be available Use racemates Experimental evaluation Purity should be 95% or higher Aqueous solubility (preferably 5 mm in 5% DMS, or other screening co-solvents) Stability (>24 h in solution) MedChem Experimental

16 Screening and structural information Keserű et al. J. Med. Chem. 2016

17 Fragment screening strategy Primary Screening Thermal Shift / SPR / MR Thermal Shift MR Secondary Screening MR Spectroscopy X-Ray ITC Binding Affinity ITC / SPR/ Biochem Biochem SPR Christina Spry

18 ptimization metrics Ligand efficiency indices connect the complexity model to ligandprotein interactions Ligand efficiency concept LE = DG/ hev preferentially >0.3 SILE= DG/( hev ) 0.3 preferentially >0.3 Lipophilic efficiency metrics LLE = ppot logp preferentially >4 LELP = logp / LE preferentially <10 LLE AT (= (LLE/ hev )) preferentially >0.3 LE LLE AT

19 Plexxicon B-Raf inhibitor B-Raf V600E is the most frequently observed oncogenic mutation Screen of 20,000 compounds ( Da) at 200uM against Pim-1, p38, and CSK 238 compounds with 200uM subjected to crystallography: >100 structures solved 7-Azaindole included among hits (IC50 > 200uM) but had different binding modes in the ATP site of the 4 asymetric units of Pim-1 But its derivatives showed conserved binding mode across 3 different kinases! 3-aminophenyl a general motif based on Pim-1 structure 3-benzyl binds similarly to FGFR1 PLX4720 bound to B-Raf 19 PAS, 2008, 105, 3041.

20 B-Raf Hit Progression Based on structures against 17 kinases the azaindole 3,4 and 5 positions were targeted for chemical derivatizations Series & selectivity progression: Binding into a Raf selective pocket Selectivity of the initial fragment hits is not important! 20 PAS, 2008, 105, 3041.

21 Astex HSP90 inhibitor 1,600 fragments were screened with a hit contained motifs similar to those of Radicicol (a natural product) Design for interaction with Lys58 or displacement of Lys58 Compound 3 ew pocket formed with Lys58 moving Compound 4 J. Med. Chem. 2010, 53, 5942 J. Med. Chem. 2010, 53, 5956.

22 Astex HSP90 inhibitor Radicicol like scaffold

23 Abbvie Bcl2 inhibitor 2 nd site 1 st site J. Med. Chem. 2006, 49, J. Med. Chem. 2008, 51, at. Med ,

24 Abbvie Bcl2 inhibitor K i < 0.5 nm MW 973

25 Acknowledgement Dan Erlanson György Ferenczy Mike Hann Chris Murray Stephen Pickett Chris Swain