PP Chames G Rosas L Rem R Willemsen

PP. Chames G. Rosas L. Rem R. Willemsen R. Bolhuis HR. Hoogenboom Affinity-matured MHC-peptide specific phage-antibodies for targeting T-cell rejection antigens on melanoma cells CD 8 T cell Target cell

Tumors are immunogenic… A large number of tumor antigens have been described, especially on melanomas. A large number of them are peptides presented by MHC complexes. Their expression often limited to tumor cells make them attractive for immunotherapy peptide 8 -9 aa Heavy chain (35 KDa) b 2 m (12 KDa)

The complex HLA-A 1/MAGE-A 1 • MAGE-A 1 expressed in 40% of melanoma tumors and in some other tumors • 25 % of the population is HLA-A 1 positive => the HLA A 1 -MAGE-A 1 complex is expressed in 10% of all melanoma tumors. • Not expressed at all in normal tissues (apart from testis, but no HLA expression) An antibody directed against this complex would be interesting to: • Check the availability of the complex before immunotherapy (in vitro MAGE-A 1 loaded DC) • As a targeting reagent (immunotoxin/cytokine, T cell retargeting)

The peptide-MHC complex is recognized by the TCR Adapted from Garcia KC et al. 1996 Science 274: 209– 19

Production of a recombinant complex Separate cloning of the HLA A 1 and b 2 m in intracellular expression vectors Purification of the inclusion bodies from E. coli Dissolution in 8 M urea buffer Dilution in a reconstitution buffer in the presence of the three partners (reduced and oxidized glutathion)

In vitro Refolding: Results • After reconstitution and concentration, the mix is analyzed by gel filtration and SDS-page HC b 2 m 1 2 3 3 Refolding 1 concentration 2 buffer

The complex is correctly refolded Aggregates b 2 m Complex Strep/biot. Complex 70 CTL 82/30 60 50 TNF (pg/ml) TNF 40 30 20 A 1 MA 1 - CTL 413/13 10 A 2 MA 3 - CTL 82/30 0 50 17 A 1 MA 1 - CTL 82/30 5 2 0. 6 Complexes (µg/ml)

Selection from a large non immunized human Fab fragment repertoire Repertoire: 3. 7 x 1010 Strep. + p. MHC ss b DTT 50 m. M Strep. Round 4 = Streptavidin

Screen for peptide specificity Peptide MAGE-A 1 EADPTGHSY Peptide MAGE-A 3 EVDPIGHLY Out of 14 different binders: 11 were pan-reactive OD 450 nm 2. 5 2 2 had a differential binding (D 2) 1. 5 1 0. 5 0 A 1 D 2 G 8 Tü 155 HK 1 was fully MAGE-A 1 specific (G 8)

G 8 binds the p. MHC complex on cell surface B cells (HLA-A 1 positive or neg. ) are in vitro loaded with MAGE-A 1 or MAGE-A 3 and used in FACS experiment with the phage-antibody G 8 LG 2 -EBV AVL 3 -EBV MZ 2 -EBV HLA-A 1 - HLA-A 1+ fd H 2 fd G 8 Loaded with MAGE-A 1 Loaded with MAGE-A 3

Fab-G 8 can efficiently retarget primary human lymphocytes Transfected human T-cells Fab-G 8 fusion Melanoma cells % Cr release Fusion of G 8 with a signaling element (Fce. R 1 g) and retroviral transfection of human PBL A 1/MAGE 1 Transfected lymphocytes kill MAGE-A 1 loaded cells Transfected lymphocytes kill MAGE-A 1 + melanoma cells

Fab-G 8 can efficiently retarget primary human lymphocytes Production of cytokines by transfected human T cells upon incubation with MAGE-A 1 loaded cells Or with MAGE-A 1+ melanoma cells MAGE-A 1+

Purification of the recombinant Fab fragment + DTT L FT E - DTT L FT E 45 30 SDS PAGE GEL FILTRATION Yield: 1. 2 mg/l (from periplasmic fraction, after metal affinity chromatography and gel filtration)

Surface plasmon resonance measurements RU 400 MAGE-1 (625, 542, 459, 375, 292, 208 n. M) 300 200 100 MAGE-3 (625 n. M) 0 590 604 618 632 646 660 Time (s) kon: 1. 8 x 105 M-1 S-1 koff: 45 x 10 -3 S-1 Kd = koff / kon = 250 n. M

Affinity maturation G 8 has a moderate affinity • 250 n. M may be enough for diagnosis purposes (FACS and tissue staining) • but we need at least a 10 fold better affinity for in vivo targeting The problem • We want to increase the affinity BUT • we don’t want to lose the specificity The new interactions have to be made ONLY with the peptide (around 20% of the interface)

Affinity maturation: The libraries • Chain Shuffling library The G 8 light chain is shuffled with a kappa + lamdba repertoire Diversity 2 x 108 Vl. Cl VHCH 1 G 8 • CDR H 3 spiking VH G 8 EVQLVESGGGLVQPGRSLRLSCAASGFTFDDYAMHWVRQAP GKGLEWVSGISWNSGSIGYADSVKGRFTISRDNAKNSLYLQM NSLRAEDTAVYYCARGGGFHYYYYGMDVWGQGTTVTVSS SPIKING: for each base of the codon, 90% WT and 10% of ACGT -> few mutations per clone, scattered over CDR H 3 Diversity 3. 107

Selection: Results Selection of mutations with additive effect HYB 3 (combination) 3 H 2/ 4 E 3 (direct selections) G 8 Name kon G 8 3 H 2 hyb 3 1. 8 2. 1 3. 8 (105 M-1 s-1) koff 45 16 5. 4 (10 -3 s-1) Kd (n. M) 250 75 14

OD 405 nm Fine specificity analysis 1. 8 1. 6 1. 4 1. 2 1 0. 8 0. 6 0. 4 0. 2 0 TÜ 155 G 8 lac 7 Hyb 3 H 2 MAGE-A 1 MAGE-A 3 M 3 A M 3 T M 3 S INFA INFT INFS E A D P T G H S Y E V D P I G H L Y E A D P I G H L Y E V D P T G H L Y E V D P I G H S Y C T E L K L S D Y C A E L K L S D Y C T E L T L S D Y C T E L K L S S Y Affinity-matured clones did not lose their fine specificity

The selected sequences Vl sequences VH sequences Difficult to localize the crucial mutations (long range effects…)

Fab-G 8 model (swissprot) Lys 30 VH Tyr 107 Gly 100 Val 111 VL

Conclusions Abstract -> We have produced a recombinant version of the tumor-related peptide-MHC complex HLA-A 1/MAGE-1 and used it for phage selection (works as well as HLA tetramer to stain and sort specific T cells) -> We have selected a fully human Fab fragment binding to this epitope in vitro (ELISA, BIAcore) and on cell surface (FACS) -> We have increased the affinity of the antibody by a factor 18 without loss of specificity. -> We have shown that the Fab fragment can be efficiently used as a surface receptor to retarget human T cells against HLA-A 1 / MAGEA 1 positive cells

Perspectives Crystallography -> G 8 and HLA-A 1/ MAGE-A 1 have been produced and purified in high amounts for crystallographic studies ->The p. MHC-Fab complex can be purified by gel filtration -> The high affinity version will also be produced