IL-2-armored peptide-major histocompatibility class I bispecific antibodies redirect antiviral effector memory CD8+ T cells to induce potent anti-cancer cytotoxic activity with limited cytokine release

Abstract

T cell engagers (TCEs) are becoming an integral class of biological therapeutic owing to their highly potent ability to eradicate cancer cells. Nevertheless, the widespread utility of classical CD3-targeted TCEs has been limited by narrow therapeutic index (TI) linked to systemic CD4+ T cell activation and aberrant cytokine release. One attractive approach to circumvent the systemic activation of pan CD3+ T cells and reduce the risk of cytokine release syndrome is to redirect specific subsets of T cells. A promising strategy is the use of peptide-major histocompatibility class I bispecific antibodies (pMHC-IgGs), which have emerged as an intriguing modality of TCE, based on their ability to selectively redirect highly reactive viral-specific effector memory cytotoxic CD8+ T cells to eliminate cancer cells. However, the relatively low frequency of these effector memory cells in human peripheral blood mononuclear cells (PBMCs) may hamper their redirection as effector cells for clinical applications. To mitigate this potential limitation, we report here the generation of a pMHC-IgG derivative known as guided-pMHC-staging (GPS) carrying a covalent fusion of a monovalent interleukin-2 (IL-2) mutein (H16A, F42A). Using an anti-epidermal growth factor receptor (EGFR) arm as a proof-of-concept, tumor-associated antigen paired with a single-chain HLA-A *02:01/CMVpp65 pMHC fusion moiety, we demonstrate in vitro that the IL-2-armored GPS modality robustly expands CMVpp65-specific CD8+ effector memory T cells and induces potent cytotoxic activity against target cancer cells. Similar to GPS, IL-2-armored GPS molecules induce modulated T cell activation and reduced cytokine release profile compared to an analogous CD3-targeted TCE. In vivo we show that IL-2-armored GPS, but not the corresponding GPS, effectively expands grafted CMVpp65 CD8+ T cells from unstimulated human PBMCs in an NSG mouse model. Lastly, we demonstrate that the IL-2-armored GPS modality exhibits a favorable developability profile and monoclonal antibody-like pharmacokinetic properties in human neonatal Fc receptor transgenic mice. Overall, IL-2-armored GPS represents an attractive approach for treating cancer with the potential for inducing vaccine-like antiviral T cell expansion, immune cell redirection as a TCE, and significantly widened TI due to reduced cytokine release.

Keywords: Antibody; CRS; MHC; T cell engager; TCE; antibody engineering; bispecific; cancer; cytokine release syndrome; major histocompatibility complex.

Figure 1.

GPS engages virus-specific effector, memory CD8+ T cells to induce potent cytotoxic activity against target tumor cells with reduced risk of cytokine-associated toxicity Conventional TCEs engage CD3 expressed on all T cells, increasing the risk for aberrant cytokine release associated with pan T cell activation. GPS selectively engage with virus-specific effector memory CD8+ T cells toward maintaining potent cytotoxic activity with reduced risk of cytokine release syndrome associated with CD4+ T cell activation.

Figure 2.

GPS induces potent cytotoxic activity in the context of pre-expanded PBMC effector cells with significantly reduced pro-inflammatory cytokine release relative to a conventional TCE. (a). Schematic depiction of GPS and conventional TCE b-d. GPS (green) or conventional TCE (red) treatment groups were evaluated for the ability to induce cytotoxic activity against NCI-H358 cancer cells using pp65/IL-2 expanded PBMCs as effector cells (b-d). Cytotoxicity, T cell activation and cytokine release (b) were determined 24-hours post treatment. (c-d). Pro-inflammatory cytokine release (dashed lines) is depicted on the right y-axes and % cytotoxicity (solid lines) is shown on the left y-axes for comparison of therapeutic indices. Error bars are SEM. Results shown are representative from one of five independent experiments from five different donors.

Figure 3.

GPS and IL-2 GPS modalities exhibit favorable thermal stability, antigen binding. (a) Schematic depiction of GPS armed with IL-2 mutein fused at either the N-terminus of the hole-side heavy chain (Fab IL-2 GPS) or C-terminus of the knob-side heavy chain (Fc IL-2 GPS) (b) Summary of thermostability (Tm) assessed via DSF (c) Summary of octet binding kinetic analyses. Octet analyses are averages of two independent experiments and error values are SEM.

Figure 4.

IL-2 GPS promote robust expansion of cmv-specific effector, memory CD8+ T cells. (a) PBMCs were treated with 6 nM test article (GPS, Fab IL-2 GPS, or Fc IL-2 GPS) or positive control (1 ug/mL pp65, 30 units/mL IL-2) in 24 well GREX plates and evaluated for antigen-specific T cell expansion at days 0, 7, and 14 via tetramer staining and flow cytometry. FACS plots are representative from one of four independent experiments with four different donors. (b) Expanded PBMC samples from a were employed as effector cells in cytotoxicity assays against NCI-H358 cells by adding serial-diluted, treatment-group matching test articles. Cytotoxicity and T cell activation was evaluated 48 hours post treatment.

Figure 5.

IL-2 GPS expand CMV-specific effector, memory CD8+ T cells in the presence of NCI-H358 cells leading to potent tumor cell killing (a) Schematic depiction of Fab IL-2 GPS, irrelevant pMHC control (Fab IL-2 GPS PRAME), TAA control (Fab IL-2 GPS NIP228), and GPS. The active molecular domains are highlighted in blue, green, yellow and pink, respectively. (b and c) Test articles were evaluated in the one-pot expansion and killing assay using non-expanded PBMCs as effector cells. (b) Following 14 days of treatment, CMV+ antiviral cell expansion was assessed by tetramer staining as shown for the 6.25 nM treatment group. (c) % cytotoxicity was determined by xCelligence as shown at the 14 day post-treatment condition. Error bars are SEM (n = 2). Expansion and cytotoxicity data correspond with the test article in each of the four columns, with cytotoxicity curve fits colored as blue (Fab IL-2 GPS), green (Fab IL-2 GPS PRAME), orange (Fab IL-2 GPS NIP228), and pink (GPS), respectively. Data shown are representative from one of three independent experiments.

Figure 6.

IL-2-armored GPS molecules selectively expand CMVpp65-specific CD8+ T cells. Whole PBMCs from HLA-A*02:01+ healthy CMV+ donors were cultured in the presence of various concentrations of GPS molecules for a period of 10-11 days for a period of 10-11 days. (a) The absolute number and fold expansion (mean ± SEM) of CMVpp65-specific CD8+ T cells on day 10-11 of culture with GPS and IL-2-armored GPS molecules. As a positive control, PBMCs were stimulated with CMVpp65 peptide + wildtype IL-2. Data shown here are from six individual donors from two independent experiments. (b) The absolute number (mean± SEM) of CMVpp65-specific and nonspecific CD8+ T cells, NK cells, CD4 tregs and total CD4+ T cells on day 10-11 of PBMC culture with Fc IL-2 GPS or Fab IL-2 GPS molecules. Data shown are from six individual donors from two independent experiments.

Figure 7.

GPS and IL-2 GPS modalities exhibit mAb-like pharmacokinetics in Tg32 mice. PK analysis of GPS, Fab IL-2 GPS and Fc IL-2 GPS administered via a single intravenous dose of 5 mg/kg in female Tg32 mice (n = 3 per group) with human IgG serum content determined by ELISA. The error bars are SEM. ‘ns’ refers to no significant difference for the area under the curve based on one-way ANOVA with Tukey multiple comparisons test.

Figure 8.

IL-2-armored GPS expands CMV-reactive effector, memory CD8+ T cells in vivo. (a) Summary of in vivo assay design: female NSG mice (NOD.Cg-Prkdc^scidIL-2Rg^tm1Wjl/SzJ) aged 6 to 8 weeks (n = 10 per group) received adoptive transfer of unstimulated PBMCs from a HLA-A*02:01+, CMV+ donor (2e6 per mouse) on day 0. Mice were dosed with Fab IL-2 GPS, Fab IL-2 GPS PRAME and GPS (1 mg/kg) on days 0, 3, 7, 10, and 14. On day 24 (b) and day 31 (c), blood samples were collected and analyzed for expansion via tetramer staining and flow cytometry analysis. The untreated group received adoptive transfer of pre-expanded PBMCs from a HLA-A*02:01+, CMV+ donor (2e6 per mouse) on day 0. For statistical analyses, ** indicates p < 0.0021 as assessed by Tukey multiple comparison test. Error bars are SEM. Outliers were calculated and removed using the ROUT method with Q set to 1%.