Microplate-Based Measurements of Target Engagement in Live Cells With CETSA - a Reflection on Screen Results and Quantitative Interpretations

Transcription

1 Microplate-Based Measurements of Target Engagement in Live Cells With CETSA - a Reflection on Screen Results and Quantitative Interpretations Thomas Lundbäck Karolinska Institutet ELRIG Drug Discovery 2016

2 Intracellular drug presence and binding Proof of target exposure and engagement at site of action -> clinical trial progression Source: Bunnage et al. (2013) Nature Chem. Biol. 9, 195 (license no: ) See also: Morgan et al. (2012) Drug Disc. Today 17, 419 Cook et al. (2013) Nature Rev. Drug Disc. 13, 419

3 Intracellular probe presence and binding Proof of target exposure and engagement at site of action -> clinical trial progression Source: Arrowsmith et al. (2015) Nature Chem. Biol. 11, 536 (license no: ) Firmly link target engagement to phenotypic effects It is not sufficient do to this for a representative compound Source: Bunnage et al. (2013) Nature Chem. Biol. 9, 195 (license no: ) See also: Morgan et al. (2012) Drug Disc. Today 17, 419 Cook et al. (2013) Nature Rev. Drug Disc. 13, 419 Off-target effects cannot be excluded even when potencies coincide Include full compound series to compare SAR Quantitative comparisons needed

4 Thymidylate synthase (TS) Key enzyme in thymidine synthesis and validated drug target Challenging to develop enzymatic HTS assay Known drugs are inactive or weakly active on the isolated enzyme Cell uptake and metabolism required for activation and potent binding

5 Cell-based screen in 384-well format % stabilization Barry Associates nucleosides Prestwick The screen selected accurately for all tested drugs acting on TS: floxuridine (FdU) 5-fluorouracil (5-FU) methotrexate

6 Time-dependent enzymatic drug conversion Thymidine kinase P FdUMP FdU 10 min 30 min 2h 6h Source: Wikipedia (public domain)

7 Time-dependent enzymatic drug conversion Thymidine kinase P FdUMP FdU 10 min 30 min 2h 6h Source: Wikipedia (public domain)

8 Time-dependent enzymatic drug conversion Thymidine kinase P FdUMP FdU 10 min 30 min 2h 6h Source: Wikipedia (public domain)

9 Decitabine stabilizes TS in cells R a w A lp h a s c r e e n s ig n a l min 30 min 2h 6h Decitabine 5-aza-2 -deoxycytidine lo g [c m p d (M )]

10 Binding requires phosphorylation... C h e m ilu m in e s c e n c e C h e m ilu m in e s c e n c e b Decitabine D r u g c o n c decitabine lo g [c m p d ] (M ) D e c ita b in e + D I-8 2 D e c ita b in e T S -a c tin 200 µm DI-82* DMSO Ctrl * Nomme et al. (2014) J. Med. Chem. 57, 9480 TS -actin Kindly provided by Caius Radu and Raymond Gipson at University of California Deoxycytidine kinase (DCK) DMSO Ctrl P 5-aza-2 -deoxycytidine 5 -monophosphate lo g [c m p d ] (M ) 200 µm DI-82

11 ... and deamination to yield a TS inhibitor % a c tiv ity R e la tiv e F lu o r e s c e n c e (a.u.) decitabine T e m p e r a tu r e (C ) Ctrl Untreated +DCK +DCK/DCTD Deoxycytidylate deaminase (DCTD) P P 4 0 **** aza-2 -deoxycytidine 5 -monophosphate 5-aza-2 -deoxyuridine 5 -monophosphate

12 CETSA data are often overinterpreted The most common application in publications is to prove a causative phenotypic effect by target modulation. But: Confirmation of cellular target engagement of a single lead compound is not sufficient to exclude off-target effects Demonstrating concentration-response data for a few compounds and agreement with those observed in the phenotypic assay is still not sufficient Two solutions: A framework for quantitative interpretation of CETSA data in terms of binding affinities Comparison of the SAR observed with CETSA with that observed in the phenotypic assay over a sufficiently broad concentration range

13 Expectations from equilibrium two-state models T m Scan rate adjusted to ensure equilibrium at all times L t = 1 H U,T e m + C p,u T m T m +T m K b S U,T m + C p,u ln T m+t m T m R T m +T m 1 + P t 2 1 e H U,T m + C p,u T m T m +T m S U,T m + C p,u ln T m+t m T m R T m +T m See e.g. Matulis et al Biochemistry 44, 5258

14 Size of the cooperative unfolding unit 6 kda H U =200 kj mol kda H U =400 kj mol kda H U =800 kj mol -1

15 Influence of drug binding thermodynamics T m ( C) T m ( C) Small protein (200 kj mol -1 ) See Waldron & Murphy 2003 Biochemistry 42, ,0E-09 1,0E-07 1,0E [L] (M) Large protein (800 kj mol -1 ) H b = 20 kj mol -1 H b = -20 kj mol -1 H b = -60 kj mol ,0E+00 5,0E-05 1,0E-04 [L] (M)

16 Relation to CETSA derived data The theoretic simulations are based on a reversible two-state unfolding model, i.e. the only significantly populated states are fully native and denatured proteins Whereas the reality will be more like this: Fully denatured protein Intermediate I Intermediate II Intermediate III Microaggregates Irreversible precipitate

17 Unpublished slide deck

18 Consequences for Thermal Proteome Profiling T a g g re p lic a te 2 XL888 - HSP90s H S P 9 0 B H S P 9 0 H S P 9 0 A ll p e p tid e s Gives a biased view of compound selectivity T a g g re p lic a te 1 See Becher et al Nature Chem. Biol. DOI: /nchembio D Thermal Proteome Profiling represents major improvement 3D multiplexing is needed to include also heating time aspect Significant portion of the proteins will not be shifted by ligand binding

19 Conclusions Microplate CETSA allows for studies of drug binding to native target proteins under a broad range of different experimental conditions Observations for a single model system is broadly in accordance with expectations from equilibrium models Apparent potency is right-shifted with increasing transient heating temperatures Apparent potency is right-shifted with increasing heating time Similar data recently published in Nature Chemical Biology for panobinostat CETSA potentially allows for studies of slow ligand binding kinetics in lysates and cells Frozen equilibria visible when varying both experimental temperature and heating time On-going: Establishing models allowing for quantitative interpretation of ITDRF CETSA data

20 Acknowledgements Chemical Biology Consortium Sweden Research groups of Pär Nordlund Helena Almqvist Rozbeh Jafari Hanna Axelsson Daniel Martinez Molina Martin Haraldsson Dan Chen Thomas Lundbäck Andreas Larsson Brinton Seashore-Ludlow Pär Nordlund Annika Jenmalm Jensen Clinical Proteomics MS Facility Uppsala Drug Optimization and Pharmaceutical Profiling Rozbeh Jafari André Mateus Janne Lehtiö Per Artursson Gianluca Maccalo Funding Nuclear Chemistry, Royal Institute of Technology Swedish Research Council Björn Dahlgren SciLifeLab Mats Jonsson Karolinska Institutet Royal Institute of Technology