Hitting the target in phenotypic drug discovery: Advances in receptor deconvolution Jim Freeth, Retrogenix, UK Jim.Freeth@retrogenix.com ELRIG Research and Innovation Meeting March 11th 2014
The right drug targets are critical for success Target selection may be one of the most important determinants of attrition and overall R&D productivity (analysis by Paul et al, Nat Rev Drug Disc, 2010) All linked to target selection (From Kola, Nat Rev Drug Disc, 3, 711, 2004)
Phenotype-led vs. Target-led Terstappen, Nat Rev Drug Disc, 2007 Target-led drug discovery Phenotype-led drug discovery
Percentage (number) Number of First-in-class NMEs A phenotype-led approach is very successful Phenotypic screening Target-based screening Modified natural substances Phenotypic screening Target-based screening 40 56% (28) 30 35 30 25 20 34% (17) 25 20 15 15 10 5 10% (5) 10 5 No apparent lag 0 First in First-in-class small molecule drug 0 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Year First-in-Class NMEs approved between 1999 and 2008, categorised by drug discovery approach (adapted from Swinney, Nat Rev Drug Disc, 2011)
Hypothesis-free target and lead discovery Library source Disease-relevant Phenotypic screen(s) PHENOTYPIC or FUNCTIONAL MOLECULES Target deconvolution NOVEL TARGETS & LEADS Target identity required for: Compound optimisation Understand safety/tox implications Novel IP Help regulatory package
Target deconvolution approaches Affinity-based 3D structures + PTMs maintained Molecule needs to be tethered Hard for low abundance targets Hard for low affinity interactions Lengthy Not ideal for membrane targets Expression cloning TARGET DECONVOLUTION Suppressionbased e.g. Y3H, M3H, Phage display Low abundance targets artificially increased Proteins may not be full-length Molecule needs to be tethered Not ideal for membrane targets Protein Arrays e.g. Suppression with cdna or biochemical fractions No chemical mods required Good for low affinity Hit may not be the actual target Biochemical approach poor for membrane targets All proteins screened equally Possible steric hindrance No cellular context Not ideal for membrane targets
Solving the membrane target deconvolution bottleneck: Cell Microarray technology Full-length human plasma membrane proteins expressed in natural human cell environment
Retrogenix has developed a unique, vast human plasma membrane (PM) protein screening set >3500 full-length, un-fused human plasma membrane proteins expressed 2600 unique PM genes 60% of total possible Largest set available Retrogenix s PM protein set by major classes Percentage of total possible by key sub-classes Unclassified 24% Transporters 21% Enzymes 7% Misc 12% Receptors 36% 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%
Development of a robust technology Reverse transfection of human HEK293 cells with arrayed GFP vector Optimised transfection/expression conditions Optimised spot density and size Tight, spatially-separated transfected cell clusters Intra- and inter-experimental variability (CVs) <10%
Specific interaction between 3 H-naloxone and its primary target, m-opioid receptor Plasma membrane proteins 1 to 384 Plasma membrane proteins 385 to 768 Duplicate 1 Duplicate 2 Duplicate 1 Duplicate 2 Fluorescence** Phosphorescence m-opioid receptor [NB:**Variation in fluorescence expected as ZsGreen1 follows 1 st gene-ires sequence]
Retrogenix s target deconvolution process INPUT: Test molecules Workup detection system if required Confirm low background binding of test molecule(s) to humanhek293 cells Pre-screen Full screen Screen test molecule (or pool of molecules) at single dose against 3500 expressed membrane proteins Vectors sequenced to double-check their identity All protein hits are rearrayed and probed with each test molecule and negative controls Confirmation screen OUTPUT: Confirmed, specific interactors delivered in formal data package
384 of 2500 expressed PM proteins Uncovering an Unknown Target: A case study Test antibody vs. 2500 human PM proteins (Primary screen) 7 hits are sequenced, re-expressed and and re-probed (Confirmation screen) Test antibody Negative control Ab ZsGreen1 ZsGreen1 Antibody detection Antibody detection 1 of 7 duplicate hits A confirmed, specific antibody target is uncovered Antibody name Format Rep. Av TE hit 1 hit 2 hit 3 hit 4 hit 5 hit 6 hit 7 EGFR -ve FCGR1A undisclosed FCGR2A undisclosed undisclosed undisclosed undisclosed Sequence verified (Y/N)--> Y Y Y Y Y Y Y Y Test sample murine IgG Fc fusion 1 3.4 strong strong strong weak Test sample murine IgG Fc fusion 2 4.6 strong strong strong weak Negative control murine IgG Fc fusion 1 4.8 weak weak weak weak Negative control murine IgG Fc fusion 2 4.3 weak weak weak
Case Study BioInvent s FIRST TM platform Conversion to full IgGs n-decoder TM phage library Differential biopanning using patient vs normal cells Functional testing In patient cells Target Id and target confirmation In vivo Testing Patient CLL cells versus normal PBMCs PCD and ADCC in patient materials i.e. combines antibody biology and target biology Targets/epitopes upregulated in disease Target and antibody biology combined from start Novel targets
High-throughput target deconvolution Primary screen: Pool of 20 mabs versus 2500 human PM proteins Hit 1 Confirmation/Matching Screen: Each mab profiled individually against all hits Spotting pattern (ZsGreen1) Protein set 1 of 7 Hit 2 Hit 3 (FCGR1A) Protein set 2 of 7 Hit 4 Hit 1 FCGR1A mab1 mab2 mab3....etc. Hit 7 FCGR1A FCGR1A Hit 5 Hit 7 Protein set 3 of 7, etc Hit 6 (FCGR2A) Hit 7 Hit 5 IGHG3 IGHG3 IGHG3 Novel, specific, cell surface target identified for each antibody
BioInvent Targets identified
Cell microarray is a powerful technology for deconvoluting receptor targets 60% hit rate for identification of targets for phenotypic antibodies High success rate due to human cell context Sensitive technology: Can identify um interactions Scalable
Scope of activities. ANTIBODIES Ab FRAGMENTS Fc-FUSION PROTEINS (Turner et al., June 2013) PROTEIN/PEPTIDE LIGANDS His-tagged V5-tagged Flag-tagged Biotinylated Directly fluorescently conjugated Protein-specific secondaries Polyclonal antisera (under review) Primary and/or secondary target Id PLASMA MEMBRANE TARGET Primary and/or secondary target Id LIVE VIRUSES SMALL MOLECULES ( 3 H-labelled)
Retrogenix technology identified the receptor associated with severe childhood malaria: Nature, June 2013
Summary Phenotypic drug discovery is a powerful approach to identify novel, disease-relevant targets Antibody-based PDD is gaining pace High throughput disease-relevant screens now possible Target deconvolution has traditionally been a bottleneck Cell microarray technology now provides a powerful solution for deconvolution of cell surface targets