Aaron K. Sato, Ph.D. CSO, Head of Antibody Center

Transcription

1 1 Discovery of Potent, Functional Anti-TIGIT Antagonists from Three Different Phage Display Platforms Aaron K. Sato, Ph.D. CSO, Head of Antibody Center

2 Passion for antibody discovery 2

3 Multiple shots on goal for antibody discovery and engineering 3

4 LakePharma Antibody Discovery Phage Display 4 Distributed Bio s SuperHuman 2.0 scfv synthetic phage library Hits against any epitope and fully natural CDR diversity >70 billion antibody diversity 100% germline frameworks and improved thermostability XOMA s 031 human Fab naïve phage library Routinely selected pm affinity antibodies Fully human, very large (>10 11 ), multiple open reading frames Leads in multiple clinical programs Immune custom phage libraries: Mouse, Rat, Rabbit, Chicken, Llama Affinity Maturation High throughput reformatting into IgG isotypes from different species High throughput 96-well transfections and high-yield purifications from proprietary CHO cell lines IgG specificity, KD assessments using Octet and Carterra Array SPR

5 Project: Phage Display Target POC with Multiple Libraries 5 Goal: Highlight the strengths of the Antibody Center Target Single pass transmembrane receptor with recombinant protein available Known receptor ligand interaction that upon blocking results in therapeutic benefit Low human/cyno homology Regarded as difficult (subjective) Libraries Naïve phage display library: Xoma Mouse immune library: Co-immunized with both human/cyno Panning Select only against human antigen to find the very best human antibodies Screening Conduct large HT screen to capture all target binders Assess both human and cyno cross-reactivity Check for affinity, binning, receptor-ligand blocking, polyspecificity, and function

6 TIGIT: MOA Similar to B7/CTLA4/CD28 Pathway 6 Coinhibitory receptors can suppress immune responses by direct signaling in cis, by inducing ligand signaling in trans, and by competition with costimulatory receptors. Competition: TIGIT Impairs CD226 Function by Directly Disrupting CD226 Homodimerization TRANS: Ligation of its ligand PVR (CD155) on dendritic cells, resulting in the conversion of those dendritic cells to a tolerogenic phenotype characterized by increased IL-10 and decreased IL-12 production

7 TPP: TIGIT 7 Receptor Family/Superfamily Expression Profile cell types Known Ligand(s) Human Homology (AA) Specificity Requirement/Epitope Mechanism of Antibody Action Affinity Requirement Monotherapy vs. Combo Therapy? ADCC/CDC Requirement? Recommended Fc? Toxicity associated with anti-target antibody therapy Most advanced competitor antibody & company Ig Superfamily some T cells and NK cells CD155 (PVR) on DCs/macrophages with high affinity, CD112 (PVRL2) with lower affinity 58.6% (mouse), 87.8% (cyno) Block CD155 binding (trans) and inhibition CD226 costimulation signal by preventing dimerization (cis) Antagonist, Block TIGIT/CD155-mediated suppression of T-cell function à promote CD155/CD226 signaling <1 nm Monotherapy and/or Combo therapy with anti-pd1, anti-pdl1, or anti- CTLA4 No, recommend higg4 (PE) or higg2 Not available RG6058, Genentech, Phase I (BENCHMARK)

8 Phage Panning and Screening Workflow 8 4 rounds of selections using XOMA Library Subsequent rounds of selections performed with decreasing Biotin-human TIGIT coupled to Streptavidin-coated magnetic beads Stringency of selections increased with number and duration of washes in successive selection rounds Bacterial periplasmic extracts (PPEs) screened for human and cyno-tigit binding by ELISA (HighRes automation deck) 3 rounds of selections with LakePharma immune library, following human and cyno-tigit mouse immunizations: 3 Rds of selections with Biotin-human TIGIT on Streptavidin-coated ELISA plate wells A. Phage panning selections with XOMA 031 Fab & LakePharma immune libraries amplify phage C. SCREENING PPEs by ELISA Infect E. coli B. ENRICHMENT OF PHAGE using biotinylated human-tigit coupled to streptavidin-coated magnetic beads Wash off unbound phage B SA- SA- Recover bound phage B- Biotin-human-TIGIT B B SA- SA- D. IgG4 reformatting Magnetic beads coated with streptavidin (SA)

9 Sequences of TIGIT-binding Abs from library selection campaigns 9 XOMA Fab naïve Library: Sequenced 1116 HuTIGIT binders from 12 x 96-well plates of clones 61 sequence-unique binders identified One clone highly enriched Clones with identical VH sequences were coupled to multiple VL sequences 30 unique human-tigit binders, including all VH-unique sequences, selected for human IgG4 reformatting and characterization: All clones with unique VH Clones with most frequent L3CDR All clones were only able to recognize human TIGIT LakePharma scfv Immune Library: Sequenced 1116 HuTIGIT binders from 12 x 96-well plates of clones 67 sequence-unique binders identified 8 clones with highly enriched H3CDR and 9 clones with highly enriched L3CDR 56 sequence-unique human-tigit binders, including all VH unique sequences, selected for human IgG4 reformatting: All clones with unique VH Clones with most frequent L3CDR

10 ELISA binding of TIGIT-specific Fab clones in bacterial PPEs following panning with the XOMA phage library Four rounds of phage selections performed against biotinylated human TIGIT antigen individual clones of bacterial Fab PPEs screened by ELISA against cynomolgus and human TIGIT protein, using LakePharma s HighRes automation deck No cynomolgus TIGIT binders were identified from the XOMA Fab library selections A(490nm) XO_C1_P1 XO_C3_P1 XO_D4_P2 XO_G1_P2 XO_H8_P2 XO_B1_P3 XO_B8_P3 XO_D3_P3 XO_A10_P5 XO_A5_P6 XO_A9_P6 XO_D9_P6 XO_A5_P7 XO_A7_P7 XO_C12_P7 biotin-hutigit Sequencing analysis allowed the selection of 30 unique human TIGIT Fab binders which were subsequently reformatted into human IgG4 for further kinetic and functional characterizations XO_D6_P7 XO_D8_P7 XO_D2_P8 XO_F4_P8 biotin-cytigit XO_F10_P8 XO_A8_P9 XO_C5_P10 XO_C8_P10 XO_E8_P10 XO_C7_P11 XO_D9_P11 XO_D12_P11 XO_F6_P11 XO_C8_P12 XO_D12_P12

11 ELISA binding of TIGIT-specific scfv clones in bacterial PPEs following panning with immune phage library Following co-immunization of mice with human and cynomolgus TIGIT-Fc antigen using LakePharma s optimized hybridoma protocols, a diverse scfv immune phage library was generated with cdna template from murine spleen mrna Three rounds of phage panning selections were performed using biotinylated human TIGIT 1116 individual clones of bacterial scfv PPEs screened by ELISA against cynomolgus and human TIGIT protein, using LakePharma s HighRes automation deck H10_P1_Imm C10_P2_Imm G7_P2_Imm H1_P2_Imm A3_P3_Imm B5_P3_Imm C4_P3_Imm C6_P3_Imm E10_P3_Imm F8_P3_Imm H11_P3_Imm A8_P4_Imm B7_P4_Imm B8_P4_Imm C6_P4_Imm C8_P4_Imm D6_P4_Imm F7_P4_Imm F12_P4_Imm G9_P4_Imm A6_P5_Imm B5_P5_Imm B7_P5_Imm C12_P5_Imm F8_P5_Imm F11_P5_Imm A11_P6_Imm D10_P6_Imm biotin-hutigit biotin-cytigit Sequencing analysis allowed the selection of 56 unique TIGIT scfv binders that were subsequently reformatted into a chimeric IgG4 isotype for further kinetic and functional characterizations E3_P6_Imm E12_P6_Imm F10_P6_Imm A5_P7_Imm A6_P7_Imm E3_P7_Imm F3_P7_Imm F8_P7_Imm P8_A4_P8_Imm P8_A7_P8_Imm P8_C9_P8_Imm P8_F1_P8_Imm B6_P9_Imm D12_P9_Imm A6_P10_Imm A8_P10_Imm A12_P10_Imm B6_P10_Imm D9_P10_Imm D10_P10_Imm D11_P10_Imm H4_P10_Imm H9_P10_Imm A6_P11_Imm B1_P11_Imm D2_P11_Imm E4_P11_Imm H2_P11_Imm A9_P12_Imm C7_P12_Imm D8_P12_Imm 11

12 HT IgG4 Transient Production using LakePharma s TunaCHO TM cell line 12 TunaCHO TM is LakePharma s proprietary cell line (same parental cell line as CHO-GSN) Consistent anti-tigit IgG4 production in 10ml scale transient transfections More cost-effective than other premium CHO transient cell line production systems Anti-TIGIT IgG4 Production Yields 9 8 Yield (mg) XOMA Immune XOMA Immune Number of Antibodies Reformatted Superb expression of 86 IgG4-reformatted anti-tigit Abs from XOMA and immune phage display libraries (average yield ~4.6mg from 10ml transfections)

13 Octet analysis (sensogram example shown) demonstrates CD155 (PVR) blocking of TIGIT-binding antibodies 13 XOMA blocking IgG4 Immune blocking IgG4 XO_A5_P6 XO_A5_P7 XO_A7_P7 XO_A8_P9 XO_A9_P6 XO_B1_P3 XO_B8_P3 XO_C1_P1 XO_C5_P10 XO_C8_P12 XO_D3_P3 XO_D4_P2 XO_D9_P11 XO_D9_P6 XO_F10_P8 XO_F6_P11 Imm_A12P10 Imm_A3P3 Imm_A6P10 Imm_A6P5 Imm_A7P8 Imm_B1P11 Imm_B5P3 Imm_D9P10 Imm_E10P3 Imm_E3P6 Imm_E3P7 Imm_E4P11 Imm_E9P12 Imm_F10P6 Capturing biotin-tigit Binding Anti-TIGIT Abs CD155 Binding XO_C12_P7 XO_C3_P1 XO_G1_P2 Imm_B6P10 Imm_B7P5 Imm_B8P4 Imm_C4P3 Imm_F12P4 Imm_F3P7 Imm_F8P7 Imm_H11P3 Streptavidin Octet biosensor loaded with Biotin-human TIGIT (3 min) Imm_C6P3 Imm_D10P10 Imm_H4P10 Imm_H9P10 TIGIT-specific IgG4 Abs added at saturating concentration (8 min) Human CD155 ligand allowed to bind (1 min) Imm_D2P11 Imm_D6P4 Imm_D8P12 CD155-blocking antibodies were discovered following XOMA and in-house immune phage display library selections

14 AlphaLISA ligand CD155 blockade assay for Immune and XOMA library reformatted clones 14 TIGIT:CD155 Homogeneous Assay Kit in AlphaLISA format (BPS Bioscience) using biotinhutigit and human CD155 ligand Anti-TIGIT IgGs ( µg/ml) incubated with human TIGIT before adding CD155 Acceptor and donor beads added before monitoring Alpha Counts Alpha Counts Immune Library IgGs TIGIT benchmark antibody Antibody free control Imm_B7P4 Imm_F8P7 Imm_A6P7 Imm_F9P7 Imm_C9P8 Imm_F11P5 Imm_H11P3 Imm_G9P4 Imm_B6P9 Imm_D9P10 Imm_E9P12 Imm_B8P4 Imm_C6P3 Imm_F7P4 Imm_H2P11 Imm_B7P5 Imm_A5P7 Imm_C6P4 Imm_D8P12 Imm_H1P2 Imm_C7P12 Imm_E3P7 Imm_H4P10 Imm_C4P3 Imm_B5P3 Imm_A4P8 Imm_A7P8 Imm_C10P2 Imm_C8P4 Imm_F10P6 Imm_C12P5 Imm_D6P4 Imm_H10P1 Imm_A8P10 Imm_B6P10 Imm_D10P10 Imm_F1P8 Imm_A12P10 Imm_A3P3 Imm_F8P5 Imm_D12P9 Imm_H9P10 Imm_F3P7 Imm_F12P4 Imm_E12P6 Imm_A6P10 Imm_A8P4 Imm_D11P10 Imm_D2P11 Imm_F8P3 Imm_A6P5 Imm_E4P11 Imm_B1P11 Imm_E10P3 Imm_E3P6 Imm_G7P2 Lower Alpha Count values indicate CD155 blockade Full titrations conducted with functional leads (see later) Alpha Counts TIGIT benchmark antibody Antibody free control XO_G1_P2 XO_D2_P8 XO_F4_P8 XO_D9_P6 XO_D8_P7 XO_F10_P8 XO_C5_P10 XO_C8_P12 XO_A10_P5 XO_D12_P11 XO_A5_P6 XO_A8_P9 XO_B8_P3 XO_C12_P7 XO_A7_P7 XO_D12_P12 XO_C7_P11 XO_E8_P10 XO_A5_P7 XO_A9_P6 XO_D6_P7 XO_D9_P11 XO_D3_P3 XO_H8_P2 XO_C1_P1 XO_B1_P3 XO_F6_P11 XOMA library IgGs XO_C8_P10 XO_C3_P1 XO_D4_P2

15 K D distribution of IgG4-reformatted TIGIT-binding antibodies by Carterra 15 Binding assays performed using Array SPR Cyno and Human TIGIT were injected as monovalent analytes as a 3-fold dilution series ( nM) over IgGs captured onto discrete spots to create a 192-spotted array via anti-human IgG Fc-coated CMD50M chip Kinetic rate constants were calculated using a monovalent (1:1) binding model No cynomolgus TIGIT IgGs from XOMA library Human Cyno L o g [K D ] d is trib u tio n fo r tw o lib ra rie s a g a in s t h ult oig g IT [K-H D ] isd is trib u tio n fo r tw o lib ra rie s a g a in s t c y T IG IT -H is Spot 135, Imm_B7P5, K D =563pM L o g [K D ] -8 L o g [K D ] X O M A lib r a r y Im m u n e lib r a r y C o n tr o l A b X O M A lib r a r y Im m u n e lib r a r y C o n tr o l A b

16 Human TIGIT (affinity ranked from high to low): Representative Data from 96 spots within a 192 spot array 16

17 Ligand blockade K D (nm) htigit K D (nm) cytigit 17 Bin

18 Color by ligand blockade Color by bin 18

19 Cell-Based Immunoblockade Assay (Promega) for Immune (A) and XOMA (B) library reformatted clones 19 CD226-induced luminescence is inhibited when Jurkat effector T cells expressing human TIGIT with luciferase reporter are cocultured with CHO-K1 cells expressing CD155 Anti-TIGIT IgGs ( µg/ml) block interaction of TIGIT/CD155, resulting in CD226-activated luminescence (Higher RLUs indicate CD155 blockade) Full titrations conducted with functional leads (see later) RLUs RLUs TIGIT benchmark antibody Antibody free control XO_G1_P2 Immune Library IgGs TIGIT benchmark antibody Antibody free control Imm_B7P4 Imm_F8P7 Imm_A6P7 Imm_F9P7 Imm_C9P8 Imm_F11P5 Imm_H11P3 Imm_G9P4 Imm_B6P9 Imm_D9P10 Imm_E9P12 Imm_B8P4 Imm_C6P3 Imm_F7P4 Imm_H2P11 Imm_B7P5 Imm_A5P7 Imm_C6P4 Imm_D8P12 Imm_H1P2 Imm_C7P12 Imm_E3P7 Imm_H4P10 Imm_C4P3 Imm_B5P3 Imm_A4P8 Imm_A7P8 Imm_C10P2 Imm_C8P4 Imm_F10P6 Imm_C12P5 Imm_D6P4 Imm_H10P1 Imm_A8P10 Imm_B6P10 Imm_D10P10 Imm_F1P8 Imm_A12P10 Imm_A3P3 Imm_F8P5 Imm_D12P9 Imm_H9P10 Imm_F3P7 Imm_F12P4 Imm_E12P6 Imm_A6P10 Imm_A8P4 Imm_D11P10 Imm_D2P11 Imm_F8P3 Imm_A6P5 Imm_E4P11 Imm_B1P11 Imm_E10P3 Imm_E3P6 Imm_G7P2 XOMA library IgGs XO_D2_P8 XO_F4_P8 XO_D9_P6 XO_D8_P7 XO_F10_P8 XO_C5_P10 XO_C8_P12 XO_A10_P5 XO_D12_P11 XO_A5_P6 XO_A8_P9 XO_B8_P3 XO_C12_P7 XO_A7_P7 XO_D12_P12 XO_C7_P11 XO_E8_P10 XO_A5_P7 XO_A9_P6 XO_D6_P7 XO_D9_P11 XO_D3_P3 XO_H8_P2 XO_C1_P1 XO_B1_P3 XO_F6_P11 XO_C8_P10 XO_C3_P1 XO_D4_P2

20 Polyspecificity ELISA of selected TIGIT-blocking antibodies 20 Polyspecificity ELISA experimental design [1] : Coating with Baculovirus particle (BVP) Loading anti-tigit Abs (150, 50, 16.7 µg/ml) Detection with HRP-conjugated anti-human Fc (2 nd Ab) Positive control (PC) and negative control (NC) Abs RG6058 benchmark anti-tigit IgG4 Background BVP wells coated only with 2 nd Ab Cut-off: 5x coated background signal BVP Score Calculation [1] : BVP score is determined by normalizing absorbance on control wells with 2 nd Ab only BVP score = OD 450 average of antibody (150 µg/ml) /OD 450 average of 2nd Ab only B V P S c o r e P C N C R G n d A b O n ly X O _ G 1 _ P 2 X O _ B 8 _ P 3 X O _ C 1 _ P 1 X O _ F 1 0 _ P 8 X O _ C 1 2 _ P 7 X O _ D 9 _ P 1 1 X O _ B 1 _ P 3 X O _ C 3 _ P 1 X O _ C 5 _ P 1 0 XOMA blocking Abs X O _ A 5 _ P 6 X O _ A 7 _ P 7 X O _ A 5 _ P 7 X O _ D 3 _ P 3 X O _ F 6 _ P 1 1 X O _ D 4 _ P 2 X O _ D 9 _ P 6 X O _ C 8 _ P 1 2 X O _ A 8 _ P 9 X O _ A 9 _ P 6 B V P S c o r e Im m _ B 7 P 5 Im m _ C 4 P 3 Im m _ A 1 2 P 1 0 Im m _ F 1 2 P 4 Im m _ A 6 P 5 Im m _ B 5 P 3 Im m _ D 6 P 4 Im m _ A 3 P 3 Im m _ E 4 P 1 1 Im m _ F 8 P 7 Im m _ D 9 P 1 0 Im m _ A 6 P 1 0 Im m _ E 9 P 1 2 Im m _ D 8 P 1 2 Im m _ A 7 P 8 Immune Library blocking Abs Im m _ B 1 P 1 1 Im m _ B 8 P 4 Im m _ B 6 P 1 0 Im m _ H 9 P 1 0 Im m _ E 1 0 P 3 Im m _ C 6 P 3 Im m _ D 1 0 P 1 0 Im m _ D 2 P 1 1 Im m _ E 3 P 6 Im m _ E 3 P 7 Im m _ F 1 0 P 6 Im m _ F 3 P 7 Im m _ H 1 1 P 3 Im m _ H 4 P 1 0 [1] Jain et al. Biophysical properties of the clinical-stage antibody landscape. PNAS, 2017, 114(5),

21 Lead human TIGIT-blocking reformatted antibodies from immune library phage library selection campaigns (3 best candidates highlighted in green color) 21 Immune Library clones Blocking hutigit Abs (Octet) Yield in KD (human TIGIT) CHOTuna TM (mg) (nm) KD (cyno TIGIT) (nm) Blocking hutigit Abs AlphaLISA (<3000 counts) AlphaLISA (IC50) IO assay (>500 RLUs) IO Assay (IC50) Polyspecificity (BVP score < 5) Imm_A6P ü 5.46E-09 ü 3.85E-07 ü Imm_A6P ü 3.93E-09 ü 1.46E-07 ü Imm_A7P ü 2.94E-09 ü 3.02E-08 ü Imm_A12P ü 6.11E-09 ü 4.48E-08 ü Imm_B5P ü 2.99E-09 ü 1.72E-08 ü Imm_C4P ü 2.75E-09 ü 5.01E-09 ü Imm_C6P ü 3.60E-09 ü 4.88E-08 ü Imm_D2P ü 4.36E-09 ü 1.52E-01 ü Imm_D8P ü 6.07E-09 ü 3.94E-09 ü Imm_E3P ü 6.45E-09 ü 1.20E-08 ü Imm_E3P ü 3.07E-09 ü 5.70E-07 ü Imm_F3P ü 3.61E-09 ü 3.90E-09 ü Benchmark ü 4.12E-09 ü 6.12E-09 ü Multiple TIGIT-blocking antibodies from our immune library phage display selections (12) were able to meet most ligand blocking, affinity, productivity and polyspecificity criteria

22 CONCLUSIONS 22 We were able to discover multiple TIGIT-specific antibodies following selections with the XOMA 031 Fab and LakePharma scfv immune libraries There was a very high sequence diversity of TIGIT binders derived from the XOMA and LakePharma scfv phage libraries A high number of cynomolgus and human TIGIT cross-reactive binders was discovered following panning with the LakePharma immune libraries Co-immunization of mice with human and cynomolgus TIGIT resulted in higher affinity CD155 blocking antibodies, as shown by both AlphaLISA and cell-based immuno oncology assays Most TIGIT Abs originating from the LakePharma immune library did not bind non-specifically to baculovirus particles (low BVP score), possibly predicting fewer developability issues