Novel TLR2xTLR4 Bispecific Antibody Inhibits Bacterial Sepsis

Generation and screening of anti-TLR2 and anti-TLR4 antibodies

To generate anti-TLR2 and anti-TLR4 antibodies, VelocImmune mice (Regeneron, NY, USA) were immunized with recombinant human TLR2 (R&D Systems, MN, USA) or recombinant human TLR4 (R&D Systems), which were emulsified with TiterMax Gold (Titer Max, GA, USA) and injected into the hind footpad. Inguinal lymph nodes were collected from the immunized mice and fused to a myeloma cell line, and a conventional method was used to generate hybridomas.⁽⁸⁾ Supernatants from the hybridomas were screened for neutralizing activity (see Neutralization assay) against TLR2 (#31) or TLR4 (#48). Generation and characterization of #31 was described earlier.⁽⁸⁾ LPS-induced interleukin (IL)-6 secretion in U937 cells was completely neutralized by #48.

To increase the physicochemical stability of these antibodies, they were subjected to complementarity-determining region (CDR)-grafting according to a published method.⁽⁹⁾ CDR sequences were determined⁽⁸⁾ and human heavy chain variable region (VH) subgroup III and light chain variable region (VL)-kappa subgroup I consensus sequences, which were used for CDR-grafting,^(10,11) were selected as stable frameworks on which to graft CDRs.⁽¹²⁾ After CDR-grafting, #31 and #48 were designated monoclonal antibody (mAb)-2 and mAb-4. There were only, respectively, 3 and 25 amino acid changes between the VH and VL frameworks of #31 and mAb-2, and 7 and 3 amino acid changes between the VH and VL frameworks of #48 and mAb-4.

The Institutional Animal Care and Use Committee of Astellas Pharma, Inc., approved all experimental procedures using animals. Further, Astellas Pharma, Inc., Tsukuba Research Center has been awarded Accreditation Status by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) International. All efforts were made to minimize the number of animals used and their suffering.

Linkage of mAb-2 to the single chain Fv of mAb-4

The single chain Fv (scFv) of mAb-4 was designed and constructed as follows. The VH was combined with the VL through the flexible linker (GGGGS)₃. The scFv sequence was modified to introduce Cys residues at positions VH44 and VL100 to stabilize the structure through a disulfide bond linking VH and VL.⁽¹³⁾ The scFv moiety of mAb-4 was genetically fused to the carboxyl terminus of the immunoglobulin γ1 chain of mAb-2 through a (GGGGS)₈ peptide linker. The heavy chain sequence was synthesized and inserted into the glutamine synthetase (GS) expression vector pEE6.4 (Lonza, Basel, Switzerland) (HindIII–EcoRI sites). The light chain of mAb-2 was inserted into the GS vector pEE12.4, as previously described.⁽⁸⁾ A dual gene vector encoding heavy and light chains was constructed by ligation of the PvuI–NotI fragment of pEE6.4 to the pEE12.4 vector.

Expression and purification of antibodies

Electroporation was used to introduce the dual gene vector into CHOK1SV^® (Lonza) cells. After culture for ∼1 week, the medium was collected and centrifuged to prepare cell-free supernatants. The antibody in the supernatant was purified by using protein-A affinity chromatography.

Size-exclusion chromatography

Size-exclusion chromatography (SEC) analysis was performed by using an LC1100 (Agilent, CA, USA) equipped with a TSK gel Super SW3000 SEC column (Tosoh, Tokyo, Japan). The diluted samples were loaded onto the column, and separation was conducted at 0.075 mL/min at 30°C.

Neutralization assay using U937 cells

The human monocyte-like cell line U937 (ATCC CRL1593.2) was maintained in RPMI 1640 medium (Thermo Fisher Scientific, MA, USA) containing 10% fetal bovine serum. The cells (7.5 × 10⁴) were added to the wells of 96-well cell culture plates (AGC Techno Glass, Tokyo, Japan) containing 100 nM PMA (Sigma–Aldrich, St. Louis, USA). The next day, purified mAbs and TLR agonists were added to the plate. E. coli Serotype O55:B5 S-form LPS (Enzo Life Sciences, NY, USA) (final concentrations, 10–100 ng/mL) was used as a TLR4 ligand, and Pam2CSK4 (final concentration, 10 ng/mL) (Invivogen, CA, USA) was used as a TLR2 ligand. After incubation overnight at 37°C, the levels of IL-6 secreted into the cell culture supernatant were measured by using an AlphaLISA human IL-6 Immunoassay Kit (PerkinElmer, MA, USA), and an EnVision 2105 Multimode Plate Reader (PerkinElmer). The percentage inhibition represents the reduction of IL-6 levels in the presence of antibodies compared with that of untreated wells.

Peripheral blood mononuclear cell assay

Human peripheral blood mononuclear cells (1.875 × 10⁴) (Lonza) were added to the wells of a 384-well cell culture plate containing RPMI 1640 medium with 10% human serum (Lonza). Purified antibodies were added to the plate, and the cells were incubated with 1 × 10⁶ CFU/mL of heat-killed P. aeruginosa PAO-1 (provided by Dr. Naomasa Goto, Kyoto Pharmaceutical University) or 1 × 10⁶ CFU/mL of heat-killed E. coli 21006 (provided by BML, Tokyo, Japan). Heat-killed E. coli and P. aeruginosa were prepared by placing the bacteria in boiling water for 10 min. After overnight incubation at 37°C, the levels of secreted tumor necrosis factor alpha (TNFα) in the cell culture supernatant were measured by using an AlphaLISA human TNFα Immunoassay Kit (PerkinElmer) and an EnVision 2105 Multimode Plate Reader (PerkinElmer). The percentage inhibition represents the reduction of TNFα in the presence of antibodies compared with that of untreated wells.

Binding affinity analysis

The binding affinities of the antibodies were analyzed by using a Biacore T200 (GE Healthcare UK Ltd, England). An anti-human IgG antibody was immobilized to a CM5 biosensor chip by using a Human Antibody Capture Kit (GE Healthcare). To measure equilibrium dissociation constants (KDs), the antibodies were diluted in HEB-EP+ buffer to 0.25 or 0.5 μg/mL and captured in the flow cells at 5 μL/min. Recombinant TLR2 or TLR4 was serially diluted in HBS-EP+ buffer (GE Healthcare) for kinetic analysis. The diluted recombinant proteins were injected at 50 μL/min. The kinetic analysis was performed by using Biacore T200 software (GE Healthcare) with the 1:1 binding model (KD = Kd/Ka).

Characterization of anti-TLR2 and anti-TLR4 antibodies and generation of a tetravalent bispecific antibody

The activities of mAb-2 and mAb-4 were evaluated by using a neutralization assay employing the U937 cell line. Both mAbs inhibited IL-6 secretion from U937 cells, which were incubated with the synthetic TLR2-specific ligand, Pam2CSK4, or the TLR4-specific ligand, LPS (Fig. 1A, B).

OPEN IN VIEWER

FIG. 1.

ICU-1, comprising mAb-2 and mAb-4, neutralizes agonist signaling through TLRs. (A) The neutralizing activity of mAb-2 was determined by using U937 cells. The cells were treated with Pam2CSK4, which is an agonist of TLR2, in the presence of mAb-2. Data (mean ± SD) represent two independent experiments performed in triplicate. (B) The neutralizing activity of mAb-4 was determined by using U937 cells treated with LPS. Data (mean ± SD) represent two independent experiments performed in triplicate. (C) Design of the ICU-1 bispecific antibody. mAb-4 scFv was fused to the C-terminus of mAb-2. mAb-4 scFv had an engineered disulfide bond. (D) Aggregates of ICU-1 antibody were analyzed by using SEC. The ratio of monomer to multimers was maintained at approximately the same level after incubation at 5°C for 13 weeks. ICU-1; LPS, lipopolysaccharide; mAb, monoclonal antibody; scFv, single chain Fv; SD, standard deviation; SEC, size-exclusion chromatography; TLR, toll-like receptor.

For dual blockade of TLR2 and TLR4, we generated the tetravalent bispecific antibody ICU-1 by genetically fusing the scFv domain of mAb-4 to the carboxyl terminus of the heavy chain of mAb-2 (Fig. 1C). The physicochemical stability of ICU-1 was assessed to evaluate its potential as a therapeutic antibody. The amount of remaining monomer content and aggregation was quantified by using SEC analysis. ICU-1 remained in the monomeric state (98%) after storage at 5° for 13 weeks (Fig. 1D).

Characterization of ICU-1

To determine whether ICU-1 retained the comparable binding activity to that of parental antibodies, surface plasmon resonance and neutralization assays were performed. Both mAb-2 and ICU-1 bound to TLR2 with similar affinities, whereas the KD of ICU-1 was approximately threefold higher than that of mAb-4 (Table 1). Although the binding affinity of TLR4 was slightly decreased, the inhibitory activities of ICU-1 were comparable with those of the parental antibodies in the neutralization assay (Fig. 2A, B).

OPEN IN VIEWER

FIG. 2.

Characterization of ICU-1. (A) Neutralizing activities of ICU-1 and combination of mAb-2 and mAb-4 against TLR2 were evaluated by using U937 cells treated with Pam2CSK4. (B) The neutralizing activity of ICU-1 and combination of mAb-2 and mAb-4 against TLR4 was evaluated by using U937 cells incubated with LPS. (C) Human PBMCs were treated with clinically relevant Pseudomonas aeruginosa in the presence of ICU-1 or (D) either 1 × 10⁵ pM mAb-2 or mAb-4. (E) Human PBMCs were treated with clinically relevant Escherichia coli in the presence of ICU-1 or (F) either 5 × 10⁵ pM mAb-2 or mAb-4. Data (mean ± SD) represent two independent experiments performed in triplicate. PBMC, peripheral blood mononuclear cell.

In patients with sepsis, TLR2 and TLR4 may be simultaneously activated by combined infection of different species with bacterial pathogens that cause sepsis. An in vitro assay was performed by using heat-killed bacteria to trigger signaling through TLR2 and TLR4. We measured the neutralizing activities of ICU-1 against P. aeruginosa and E. coli, which are frequently isolated from patients with sepsis.^(4,5) ICU-1 completely inhibited stimulation by both species, and its inhibitory activity was more potent than those of mAb-2 or mAb-4 at the highest concentration tested (1 × 10⁵ or 5 × 10⁵ pM) (Fig. 2C–F). These data indicate that ICU-1 suppressed cytokine production evoked by multiple species of sepsis-causing bacteria to a greater extent than mAb-2 or mAb-4.

A limitation of this study was that we were unable to test ICU-1 in an animal model of sepsis, because it did not react with the rodent homologs of TLR2 and TLR4. Previous studies found that the blockade of TLR2 and TLR4 improves the survival rate of mice with sepsis.^(7,14–16) Accordingly, ICU-1 would be rationally expected to have therapeutic potential for sepsis.

In conclusion, a novel bispecific antibody against TLR2 and TLR4 may serve as a treatment option for sepsis caused by bacterial pathogens.