Combination of two novel blocking antibodies, anti-PD-1 antibody ezabenlimab (BI 754091) and anti-LAG-3 antibody BI 754111, leads to increased immune cell responses

Abstract

Upregulation of inhibitory receptors, such as lymphocyte activation gene-3 (LAG-3), may limit the antitumor activity of therapeutic antibodies targeting the programmed cell death protein-1 (PD-1) pathway. We describe the binding properties of ezabenlimab, an anti-human PD-1 antibody, and BI 754111, an anti-human LAG-3 antibody, and assess their activity alone and in combination. Ezabenlimab bound with high affinity to human PD-1 (KD = 6 nM) and blocked the interaction of PD-1 with PD-L1 and PD-L2. Ezabenlimab dose-dependently increased interferon-γ secretion in human T cells expressing PD-1 in co-culture with PD-L1-expressing dendritic cells. Administration of ezabenlimab to human PD-1 knock-in mice dose-dependently inhibited growth of MC38 tumors. To reduce immunogenicity, ezabenlimab was reformatted from a human IgG4 to a chimeric variant with a mouse IgG1 backbone (BI 905725) for further in vivo studies. Combining BI 905725 with anti-mouse LAG-3 antibodies improved antitumor activity versus BI 905725 monotherapy in the MC38 tumor model. We generated BI 754111, which bound with high affinity to human LAG-3 and prevented LAG-3 interaction with its ligand, major histocompatibility complex class II. In an in vitro model of antigen-experienced memory T cells expressing PD-1 and LAG-3, interferon-γ secretion increased by an average 1.8-fold versus isotype control (p = 0.027) with BI 754111 monotherapy, 6.9-fold (p < 0.0001) with ezabenlimab monotherapy and 13.2-fold (p < 0.0001) with BI 754111 plus ezabenlimab. Overall, ezabenlimab and BI 754111 bound to their respective targets with high affinity and prevented ligand binding. Combining ezabenlimab with BI 754111 enhanced in vitro activity versus monotherapy, supporting clinical investigation of this combination (NCT03156114; NCT03433898).

Keywords: BI 754091; BI 754111; LAG-3; PD-1; ezabenlimab.

Conflict of interest statement

No potential conflict of interest was reported by the authors.

© 2022 Boehringer Ingelheim RCV GmbH & Co KG. Published with license by Taylor & Francis Group, LLC.

Figures

Figure 1.

Blockade of inhibitory PD-1/PD-L1 signaling by ezabenlimab leads to T-cell activation. Beads coated with PD-L1 protein and agonistic CD3 and CD28 antibodies were used to stimulate and simultaneously block activation of Jurkat cells expressing human PD-1. Dose titration with ezabenlimab blocked the inhibitory interaction leading to NFAT-mediated luciferase activity. Left panel (a) shows graphical description of the assay. Right panel (b) shows the assay results from one study (EC90 was calculated from three studies). Luciferase activity was calculated: 0% activation represents the luminescence value of cells incubated in the presence of anti-CD3/anti-CD28/PD-L1 beads without antibodies, and 100% activation represents the signal obtained with maximum luciferase levels in the presence of anti-CD3/anti-CD28/mouse IgG1 loaded control beads without antibodies present. In red, ezabenlimab mediated a dose-dependent increase in luciferase activity. In gray, titration of control IgG4 antibody did not affect luciferase activity.

Figure 2.

Ezabenlimab treatment leads to tumor regression in the MC38 mouse tumor model. MC38 colon adenocarcinoma were subcutaneously grown in hPD-1 knock-in mice. Isotype control (a) or ezabenlimab (b–f) were administered by intraperitoneal injection once every 3 weeks (starting on d 6 after tumor cell injection). Tumor diameters were measured three times per week. TGI was calculated on d 25 (as indicated by the vertical red line) and the responding tumors were counted on d 36. Response was defined by a tumor size that was the same or smaller than the measurement at the start of treatment. Graphs show individual tumor volumes over time per treatment group and the table provides a summary of TGI and response for each dosage schedule.TGI, tumor growth inhibition; q21d, every 21 days.

Figure 3.

Murine LAG-3 (mLAG-3) tool antibodies potently block the interaction of mLAG-3 and murine MHC-II. (a) 3A9 mouse T cell line expressing mLAG-3 (3A9-mLAG-3) was incubated with 20 µg/mL isotype control or 20 µg/mL C9B7W rat IgG1 anti-mLAG-3 antibody. Bound antibodies were detected by PE-labeled secondary antibody and the signal was detected by flow cytometry. The graph illustrates that a specific signal was seen only when cells were incubated with C9B7W confirming that it binds to mLAG-3 in this particular test. (b) 3A9-mLAG-3 expressing cells were incubated with 20 µg/mL isotype control or 20 µg/mL C9B7W in the presence of PE-labeled OVA323-339 MHC-II-IAb tetramer for 60 min, washed and fluorescent signals were detected by flow cytometry. Bound tetramers were observed only in the isotype treated control (left graph); presence of C9B7W blocked the interaction of these mouse Class II tetramers to mLAG-3 (right graph). (c) C9B7W (rat IgG1) and EX 80658 (C9B7W variant on mouse IgG1 D265A backbone) were tested for their binding properties to mouse LAG-3 expressed on a 3A9 mouse T cell line. Bound primary antibody was detected using either anti-mouse or anti-rat fluorescently labeled secondary antibody. After washing, the remaining signal was detected by flow cytometry and EC50 values were calculated. Left graph shows dose-dependent binding of antibodies to mLAG-3 as indicated by increasing MFI. The table provides an overview of the calculated EC50 values including the 95% CI for the two antibodies. CI, confidence interval; MHC-II, major histocompatibility complex class II; MFI, median fluorescent intensity.

Figure 4.

In vivo combination of BI 905725 with anti-mLAG-3 treatment is more efficacious than BI 905725 therapy only. MC38 colon adenocarcinoma were subcutaneously grown in hPD-1 knock-in mice. Starting on d 6, animals were treated by intraperitoneal injection with isotype control (a), C9B7W (rat IgG1 anti-mLAG-3) (b), EX 80658 (mouse IgG1 D265A anti-mLAG-3) (c), BI 905725 (mouse IgG1 D265A anti-hPD-1) (d) or the combination of BI 905725 with either C9B7W (e) or EX 80658 (f). All antibodies were administered at a dose of 10 mg/kg twice weekly. Tumor diameters were measured three times per week. Responding tumors were counted on d 47. Graphs show individual tumor volumes over time per treatment group and the table provides a summary of response per treated group. Response was defined by a tumor size that was the same or smaller than the measurement at the start of treatment. TGI values were determined on d 21 because control tumors are required for calculation; long-term efficacy results are best described in this study by response and the individual tumor growth curves. hPD-1, human programmed cell death protein-1; mLAG-3, mouse lymphocyte activation gene-3, q3or4d, every 3 or 4 d.

Figure 5.

BI 754111 potently blocks binding of soluble hLAG-3 to MHC class II expressing Raji cells. Recombinant human LAG-3 Fc chimera was incubated with serially diluted BI 754111 or isotype control. Raji cells expressing MHC class II were added. After incubation and washing, bound human LAG-3 was detected using fluorescently labeled anti-human IgG antibody using flow cytometry. For dose response curves and IC90 calculation, data were normalized to 0% inhibition (defined as the fluorescence value of cells incubated with the soluble hLAG-3 and no antibodies present) and 100% inhibition (defined as the signal obtained with cells without soluble hLAG-3 or antibodies). The graph shows BI 754111 dose-dependent inhibition of binding of soluble LAG-3 to Raji cells expressing MHC class II (one example from four donors). hLAG-3, human lymphocyte activation gene-3; Ig, immunoglobulin; MHC, major histocompatibility complex.

Figure 6.

Blockade of inhibitory LAG-3/MHC class II signaling by BI 754111 leads to T-cell activation. (a) Detached LK35.2 cells were pre-incubated with HEL48-62 peptide. In parallel, a 3A9 mouse T cell line expressing human LAG-3 was incubated with serially diluted BI 754111 or isotype control. A co-culture of 3A9-hLAG-3: LK35.2 cells (ratio of 2:1) was set up and incubated for 24 hours. Thereafter the supernatant was harvested and mouse IL-2 concentrations were analyzed by ELISA. EC50 was calculated by data transformation; 0% activation was defined as IL-2 secretion of cells incubated with assay medium only and 100% activation was defined as IL-2 secretion of cells incubated with the highest concentration of the reference tool anti-human LAG-3 antibody. (b) PBMCs from healthy human donors who had recently been vaccinated against tetanus were cultivated in the presence of the antigen and T cells were expanded for 13 d. In parallel, monocyte derived DCs were generated and loaded with tetanus antigen. T cell culture and antigen-loaded DCs were co-cultured for one day and on the next day freshly antigen-loaded DCs were added a second time together with antibodies. After 4 d of treatment, supernatants were analyzed for IFN-γ secretion by ELISA (see the scheme in Supplementary Fig. S1). Left graph shows results from one donor: a dose-dependent increase of IFN-γ secretion following treatment with ezabenlimab (blue curve), BI 754111 (orange curve) or ezabenlimab in combination with BI 754111 (green curve). Right graph shows the data from four donors treated with ezabenlimab, BI 754111 or their combination. APC, antigen-presenting cell; DCs, dendritic cells; IFN, interferon; IL, interleukin; MHC-II, major histocompatibility complex class II; LAG-3, lymphocyte activation gene-3; PBMCs, peripheral blood mononuclear cells; SD, standard deviation.