The tumor targeted superantigen ABR-217620 selectively engages TRBV7-9 and exploits TCR-pMHC affinity mimicry in mediating T cell cytotoxicity

Abstract

The T lymphocytes are the most important effector cells in immunotherapy of cancer. The conceptual objective for developing the tumor targeted superantigen (TTS) ABR-217620 (naptumomab estafenatox, 5T4Fab-SEA/E-120), now in phase 3 studies for advanced renal cell cancer, was to selectively coat tumor cells with cytotoxic T lymphocytes (CTL) target structures functionally similar to natural CTL pMHC target molecules. Here we present data showing that the molecular basis for the anti-tumor activity by ABR-217620 resides in the distinct interaction between the T cell receptor β variable (TRBV) 7-9 and the engineered superantigen (Sag) SEA/E-120 in the fusion protein bound to the 5T4 antigen on tumor cells. Multimeric but not monomeric ABR-217620 selectively stains TRBV7-9 expressing T lymphocytes from human peripheral blood similar to antigen specific staining of T cells with pMHC tetramers. SEA/E-120 selectively activates TRBV7-9 expressing T lymphocytes resulting in expansion of the subset. ABR-217620 selectively triggers TRBV7-9 expressing cytotoxic T lymphocytes to kill 5T4 positive tumor cells. Furthermore, ABR-217620 activates TRBV7-9 expressing T cell line cells in the presence of cell- and bead-bound 5T4 tumor antigen. Surface plasmon resonance analysis revealed that ABR-217620 binds to 5T4 with high affinity, to TRBV7-9 with low affinity and to MHC class II with very low affinity. The T lymphocyte engagement by ABR-217620 is constituted by displaying high affinity binding to the tumor cells (KD approximately 1 nM) and with the mimicry of natural productive immune TCR-pMHC contact using affinities of around 1 µM. This difference in kinetics between the two components of the ABR-217620 fusion protein will bias the binding towards the 5T4 target antigen, efficiently activating T-cells via SEA/E-120 only when presented by the tumor cells.

Figure 1. Flow cytometry and TCR-mRNA (IMGT TRB variants) analysis.

A) Flow cytometry analysis of [ABR-217620-Biotin/SA-PE]-complex binding to human PBMC from a healthy donor also stained for CD3, CD4, CD8 and HLA-DR. B) Re-analysis of flow cytometry sorted human PBMC from a healthy donor and C) TCR-mRNA analysis (IMGT TRB-variants) of unsorted and sorted cells binding or not binding the [ABR-217620-biotin/SA-PE]-complex. D) TCR-mRNA analysis (IMGT TRB-variants) of T cells from in vitro cultures activated with SEA, SEA/E-120 or SEE. E) TCR-mRNA analysis (IMGT TRB-variants) of T cells from in vitro cultures activated with different concentrations of SEA/E-120.

Figure 2. Sequences and structure of Sags show putative TCR binding regions on SEA, SEE and SEA/E-120.

A) Sequence alignment of the SEA/E-120, SEA and SEE. Residues in the presumed TCR binding site are colored. Residues believed to be of particular importance for TCRVβ binding are marked with asterisks. B) Putative TCR binding residues are mapped onto the SEA structure (PDB ID: 1lo5) and marked in accordance with the coloring in A.

Figure 3. Flow cytometry, cytotoxicity (SADCC) and NFκB-luciferace reporter gene analysis.

A) Analysis of flow cytometry sorted SEA activated T cells and B) cytotoxicity (SADCC) measured against Caki-2 cells with (filled) or without (open) ABR-217620 using a standard 4 hr ⁵¹Cr-release assay of unsorted (squares) and sorted (triangles) cells binding the [ABR-217620-biotin/SA-PE]-complex and anti-CD3. C) Activation of NFκB-luciferace reporter gene in J.RT3-T3-5 cells expressing TRBV7-9 by Caki-2 cells and different concentrations of ABR-217620 (squares) or SEA/E-120 (triangles). Activation of the NFκB-luciferace reporter without ABR-217620/SEA/E-120 was equal to the defined 0-value in the graph. D) Activation of NFκB-luciferace reporter gene in J.RT3-T3-5 cells expressing TRBV7-9 by 5T4- and anti-CD2-coated (squares) or control-coated (triangles) beads and different concentrations of ABR-217620. Activation of the NFκB-luciferace reporter without ABR-217620 was equal to the defined 0-value in the graph.

Figure 4. Surface plasmon resonance analysis of ABR-217620 binding to 5T4.

ABR-217620 was injected for 3 min at a flow rate of 20 µL/min over 5T4Fc (density 770 RU). Regeneration of the surface was made by injecting 10 µL 10 mM glycine-HCl, pH 1.5, during 30 s. Sensorgrams from bottom to top represent sample buffer and ABR-217620 in the concentration range 1.56-50 nM. An affinity of 2.1 × 10^-10 M was calculated after kinetic analysis of binding data by fit of sensorgrams to a 1:1 model with on- and off-rates of 3.6 × 10⁵ (1/Ms) and 7.5 × 10^-5 (1/s).

Figure 5. Surface plasmon resonance analysis of TRBV7-9 interaction with ABR-217620 or 5T4Fab-SEA.

25 nM ABR-217620 or 5T4Fab-SEA was captured (5 min at 20 µL/min) on immobilized 5T4Fc (density ~ 990 RU) prior to injection (5 min at 20 µL/min) of 0.156-5 µM TRBV7-9. The surface was regenerated with a short 10 µL pulse of 10 mM glycine-HCl, pH 1.5. A) Sensorgrams obtained after injection of 25 nM ABR-217620 (1) followed by 0-5 µM TRBV7-9 (3) with buffer pumped over the surface at (2) and (4). B) Injection of 0.156-5 µM TRBV7-9 over ABR-217620 and 5T4Fab-SEA after subtraction of sensorgram with sample buffer without TRBV7-9. C) Responses at early association phase (t ~ 530 s) plotted versus TRBV7-9 concentration for binding to captured 5T4Fab-SEA (squares) and ABR-217620 (triangles). Curves were fit to a one-site hyperbola model in GraphPad Prism for calculation of maximal response and apparent affinity.

Figure 6. Surface plasmon resonance analysis of TRBV7-9 binding to fusion proteins containing SEA, SEE and SEA/E-120.

A) Sensorgrams obtained after injection (5 min at 20 µL/min) of 25 nM C215Fab-SEA, -SEE and –SEA/E-120 over amine coupled EpCAM/Fc (density ~ 870 RU). Capture levels of 109, 134 and 133 were calculated for C215Fab-SEA, –SEE and –SEA/E-120 respectively. No binding was observed when these samples were injected over CD28/Fc (density ~ 680 RU) or 5T4Fc (density ~ 990 RU). B) Sensorgrams showing binding of 0.156 to 5 µM TRBV7-9 to EpCAM captured C215Fab-SEA, C215-SEE or C215–SEA/E-120 after subtraction of sensorgram with sample buffer. C) Response levels at t ~ 530 s plotted against TRBV7-9 concentration for binding to SEA (squares), SEE (circles) and SEA/E-120 (triangles) fusion proteins. Curves were fit to a one-site hyperbola model in GraphPad Prism for calculation of maximal response and apparent affinity.

Figure 7. Surface plasmon resonance analysis of HLA-DR1 HA peptide complex binding to antigen captured SEA, SEE and SEA/E-120.

50 nM fusion proteins containing SEA, SEE and SEA/E-120 were captured (5 min at 20 µL/min) on either immobilized 5T4Fc (ABR-217620 and 5T4Fab-SEA) or EpCAMFc (C215Fab-SEE) prior to injection (2 min at 20 µL/min) of 0.625-5 µM HLA-DR1/HA (mol wt ~ 44.7 K). Sensorgrams show binding of HLA-DR1/HA to captured 5T4Fab-SEA, C215Fab-SEE and ABR-217620 after subtraction of SPR signal obtained when sample buffer (HBS-P containing 10 µM Zn²⁺) was injected over captured Fab fusion proteins. For ABR-217620 only binding of the HLA-DR1/HA complex at 5 µM is shown as the signal at the lower concentrations was too low for accurate kinetic evaluation.