Quantitative And Spatial Optimization Of Therapeutic Fusion Proteins Jeffrey Way, Ph.D. Senior Staff Scientist, Wyss Institute, HMS CEO, General Biologics, Inc. 1
Chimeric Activators Drug design and targeted delivery platform Reduced side effects Greater efficacy with safer enhanced dosing Predictable pharmacokinetics and biodistribution First-generation protein drugs Second-generation protein drugs Chimeric activators: Third-generation proteins Target cell Some drug molecules bind correctly Target cell Increased binding?? Target cell Rescued binding frequency resulting in efficacy Others do not, resulting in side effects Non-target cell (side effects) But still with side effects Non-target cell (side effects) With far fewer side effects Non-target cell (side effects) 2
Key elements to achieve targeting Activity element with mutation Targeting element (e.g. scfv) Linker to allow simultaneous binding Examples with activity in animals Anemia chimeric activator Chimeric Activator a 4-part protein Activity Element Mutation Linker Targeting Element EPO Receptor EPO Glioblastoma multiforme chimeric activator IFNa (mut) Linker EPO Receptor scfv Glycophorin A EGFR Mutation reduces on-rate and equilibrium binding Does not affect signaling when bound IFNa Receptor Linker scfv 3
Modular product generation presents vast opportunities Activity Element Mutation Linker Targeting Element Hormones Erythropoietin Growth hormone Cytokines Interferons Interleukins Immune suppressors Proteases Amyloid degradation Blocking antibodies Stimulating antibodies Neuropeptides Stem cell factors Proteins Antibody V regions (scfv s) Whole antibodies Ligands Cells Tumor cells Immune cell subclasses Red blood cell precursors Stem cells Extracellular matrix Neurons Heart, lung, Gut, pancreas, etc. 4
Implications for similar product classes Bispecific antibodies BiTEs Antitumor Anti-CD3, activating Essentially a cytokine (what is desired binding/activity?) Whole antibodies Antigens on the same cell surface Geometry of angles Antigens on different cells Which cell to bind to first? CAR-T cells Cell surface fusion protein Tumor cell Engagement is driven by T cells finding tumor cells Need to calculate cell cell onrates, off-rates CAR-T cell Want high off-rate for nontarget cells with low abundance of tumor antigen 5
Software-driven drug design platform Computational technology applies to all engineered fusion protein drugs Allows for software-driven optimization vs current trial & error methods Linker optimization example: (Gly 4 Ser) n T A R T R A T = Targeting element; A = Activity element; R T = Targeting element receptor; R A = Activity element receptor Simulation Clip Shapesifter - http://shapesifter.org/ 6
Differential equation modeling predicts macroscopic behavior from molecular features Candidate protein design 3D representation and simulation Ordinary differential equations PK/PD predictions Cell-level state models Compartment models RBCs RBC precursors Plasma Tissues Kidney EPO-R + cells Cell and tissue behavior can be predicted from engineered binding constants Our approach builds on intensive basic research of others 7
Mechanism of the quantitative enhancement Illustrative calculations to predict specificity enhancement 35 amino acids o 140 A = 14 nm = 0.014 mm [ ] = 1 nm in tissue extracellular space K D = 100 nm [ K D = GlyA] [ = ]*[GlyA] 1 nm [GlyA][ 100 GlyA] nm 1% occupancy 50,000/cell = 1% GlyA occupancy Volume around cell = 1000 mm 2 * 0.014 mm = 14 mm 3 Concentration of fusion protein on cell surface: 50,000 * 1% = 500 molecules 14*10-15 L = 35 x 10 15 molecules/l = 60 nm EPO Rec. Late RBC precursor (Erythroblast) 8
Anemia Research Program 9
Worldwide total sales ($Billions) Erythropoietin market disappearance & unmet needs 14 12 10 8 6 4 2 Total = J&J + Amgen + Roche Procrit/Eprex Epogen Aranesp Neorecormon Mircera Total 0 2000 2002 2004 2006 2008 2010 2012 2014 2016 Black Box Warning: Cancer patients Not to be used in curable cancer patients Result of safety issues Cancer market greatly reduced (>50%) Reduced use at major dialysis centers Biosimilars will have the same black box safety issues 10
The actual need for a safer EPO 11
Anemia drug: Targeted EPO Increase in target specificity while lowering off-target binding EPO Antibody scfv R150A Glycophorin Platelet Precursors Protein K D EPO 5.4 nm EPO(R150A) 81 nm 10F7 100 nm Capillary Cells Red Blood Cell Precursor Mature Red Blood Cells 12
Demonstration of targeting: Experimental design EPO-based drug Therapeutic effect: Red blood cells ( % Reticulocytes) Human glycophorin transgenic mouse Side effect: Thrombosis/ (Heart attack & stroke) Platelets ( % Reticulated platelets) Antibody DOES NOT bind mouse glycophorin Human EPO DOES bind mouse EPO-R Compare commercial molecule and structural variants 13
Structure-function analysis of a targeted protein: Targeted EPO yields red blood cells - Not platelets Targeted EPO Burrill et al. Targeted erythropoietin selectively stimulates red blood cell expansion in vivo PNAS 2016.
Targeted EPO: Extended pharmacokinetics Targeted EPO Half-life in mice 28 hours Half-life in rats 2.5 hours 6.9 hours In transgenic animals with RBC binding In non-transgenic animals 15
Current status Targeted erythropoietin Antibody element in original product was inflammatory RBCs can be pro- or anti-inflammatory (Not just a bag of hemoglobin who knew?) Choosing other V regions Other targeted products in earlier stages of development Cancer Viral infections Stem cell therapy enhancement Hormone delivery Partnerships in protein design 16
Special Thanks To: Boston Biomedical Innovation Center B-BIC: an NIH grant with sub-grants The Wyss Institute Harvard Blavatnik Innovation Fund DARPA Pamela Silver, Devin Burrill, Avi Robinson-Mosher, Larry Levan (Other founders of General Biologics, Inc.)