Tri-specific tribodies targeting 5T4, CD3, and immune checkpoint drive stronger functional T-cell responses than combinations of antibody therapeutics

Abstract

One of the most promising cancer immunotherapies is based on bi-specific T-cell engagers (BiTEs) that simultaneously bind with one arm to a tumor-associated antigen on tumor cells and with the other one to CD3 complex on T cells to form a TCR-MHC independent immune synapse. We previously generated four novel tri-specific tribodies made up of a Fab targeting 5T4, an oncofetal tumor antigen expressed on several types of tumors, a scFv targeting CD3 on T cells, and an additional scFv specific for an immune checkpoint (IC), such as PD-1, PD-L1 or LAG-3. To verify their advantages over the combinations of BiTEs (CD3/TAA) with IC inhibitors, recently used to overcome tumor immunosuppressive environment, here we tested their functional properties in comparison with clinically validated mAbs targeting the same ICs, used alone or in combination with a control bi-specific devoid of immunomodulatory scFvs, called 53 P. We found that the novel tri-specific tribodies activated human peripheral blood mononuclear cells more efficiently than clinically validated mAbs (atezolizumab, pembrolizumab, and relatlimab) either used alone or in combination with 53 P, leading to a stronger tumor cytotoxicity and cytokines release. In particular, 53L10 tribody targeting PD-L1 displayed much more potent effects than the combination of 53 P with all the clinically validated mAbs and led to complete tumor regression in vivo, showing much higher efficacy than the combination of 53 P and atezolizumab. We shed light on the molecular basis of this potentiated anti-tumor activity by evidencing that the insertion of the anti-PD-L1 moiety in 53L10 led not only to stronger binding of the tri-specific to tumor cells but also efficiently blocked the effects of increased PD-L1 on tumor cells, induced by IFNγ secretion also due to T-cell activation. These results are important also for the design of novel T-cell engagers targeting other tumor antigens.

Fig. 1. Schematical representation of 53X tribodies and their binding kinetics, compared to clinically validated mAbs, on their corresponding purified targets via BLI analyses.

The sensorgrams reported in A for 53L10 compared to atezolizumab, B for 53D compared to pembrolizumab, and C for 53 G compared to relatlimab were obtained via BLI analyses. The recombinant PD-L1/Fc, PD-1/Fc, and LAG-3/Fc were used as ligands, whereas the tri-specific tribodies or validated mAbs were used as analytes and tested at increasing concentrations. As a negative control, the tribody 53 P was used as an analyte in a parallel assay. The sensorgrams show the association and dissociation rates of the analytes. The schematical representations of 53X tribodies are reported as insets in each relative binding panel. The constructs contain a Fab domain specific for 5T4 and an anti-CD3 scFv fragment; the additional scFv fragment targets an IC, i.e., PD-L1 (A), PD-1 (B), or LAG-3 (C), respectively. The scFv from Palivizumab was used for 53 P tribody, generated as a control (A). K_D values and association and dissociation constants of the binding of each analyte to its corresponding immobilized ligand on a pro-A biosensor, processed according to the following formula: in which A represents the analyte, and B represents the immobilized ligand. Ka and kd are the association and dissociation rate constants [50]. Analysis was performed using Octet Analysis Studio 13.0 Software (Sartorius, Fremont, CA, USA).

Fig. 2. Binding curves of 53X tribodies on lymphocytes and tumor cells tested in comparison with the clinically approved mAbs.

A The tribodies and the FDA approved mAbs were tested at the indicated concentrations in parallel assays by cell ELISA on human lymphocytes activated with anti-CD3/CD28 beads. As a negative control, the binding of the novel tribodies was tested also on natural killer cells stimulated with SEB (50 ng/ml) for 72 h. B Expression of 5T4 and PD-L1 on a panel of tumor cells. Cell ELISA assays to check the expression level of 5T4 and PD-L1 on lung A-549, breast MDA- MB-231 and liver Li-7 cancer cells were performed with a commercial anti-PD-L1 or anti-5T4 antibody. C Binding curves of 53X tribodies tested at increasing concentrations (0.1–100 nM) on the indicated tumor cell lines. Binding values were reported as the mean of determinations obtained in three independent experiments. Error bars depicted means ± SD. The table reports the EC50 values obtained by ELISA binding curve analyses with Prism (Graphpad) tool according to the following model: Y  =  Bmax*X/(Kd+X)  +  NS*X + Background. EC50 values are obtained by ELISA binding curve analyses with Prism (Graphpad) tool according to the following model: Y  =  Bmax*X/(Kd+ X)  +  NS*X + Background.

Fig. 3. Cytotoxic effects of the novel tribodies in comparison with the combinations of 53 P and the clinically validated mAbs.

Tumor cells (A-549, MDA-MB-231, and Li-7) were co-cultured with hPBMCs (3:1 effector/target ratio) and treated for 48 h or 72 h with 53 P (■), 53L10 (A), 53D (B), and 53 G (C) tribodies (all ●). The clinically validated immunomodulatory mAbs atezolizumab, pembrolizumab, and relatlimab (all □), or their combinations with 53 P (○) were tested in parallel at the indicated concentrations (0.05–10 nM). Co-cultures untreated or treated with an unrelated IgG (*) were used as negative controls. Cell lysis was measured by detecting LDH release. Error bars depicted means ± SD. P values were calculated by comparing each treatment with the tri-specific tribody to that of the corresponding combinatorial treatment (53 P + mAb targeting the same IC), and the P values reported are: ***P ≤ 0,001; **P < 0.01; *P < 0.05, by student’s t test (two variables).

Fig. 4. IFNγ and IL-2 release in co-cultures of tumor cells with hPBMCs treated with the novel tribodies.

Tumor cells (A-549 and MDA-MB-231) were co-cultured with hPBMCs (3:1 effector/target ratio) and treated for 48 h or 72 h with 53 P (■), 53L10, 53D and 53 G tribodies (●). The parental immunomodulatory mAbs atezolizumab, pembrolizumab, and relatlimab (□), or their combinations with 53 P (○) were tested in parallel assays, for comparison, at the indicated concentrations (0.05–10 nM). The levels of IFNγ (A) and IL-2 (B, C) released in the supernatant of co-cultured cells were measured by cytokine secretion kit (R&D systems), following the manufacturer’s recommendations. Co-cultures untreated or treated with an unrelated IgG (*) were used as negative controls. Error bars depicted means ± SD. P values were calculated by comparing each treatment with the tri-specific tribody to that of the corresponding combinatorial treatment (53 P + mAb), and the P values reported are: ***P ≤ 0.001; **P < 0.01; *P < 0.05, by student’s t test (two variables).

Fig. 5. T-cell activation by 53L10 tribody in comparison with the combination of 53 P and atezolizumab.

Tumor cells (A-549 and MDA-MB-231) were co-cultured with commercially available hPBMCs (10:1 effector: target ratio) and treated with 53 P (light gray), 53L10 (striped), atezolizumab (dark gray) and 53 P plus atezolizumab (black) at the indicated concentrations for 24 h and 48 h. As negative controls, both untreated cells or treated with an unrelated IgG (empty bars) were used. T-cell activation was directly measured by percentage of CD69-positive cell population in CD3-positive cells at 24 h and percentage of the CD25-positive cell population in CD3-positive cells at 48 h by flow cytometry. A, C Percentage of CD69⁺ population in CD3⁺ hPBMCs co-cultured with A-549 (A) or MDA-MB-231 cells (C). B, D Percentage of CD25⁺ population in CD3⁺ hPBMCs co-cultured with A-549 (B) or MDA-MB-231 cells (D). The data were expressed as mean ± SD. Statistical significance was calculated by student’s t test (two variables). P values reported are: ***P ≤ 0,001; **P < 0.01.

Fig. 6. Comparison of cytotoxic effects of 53X tribodies on tumor cells with the combinations of bi-specifics and 53 P.

Tumor cells were co-cultured with hPBMCs (3:1 effector/target ratio) and treated for 48 h with 53 P (■), 53L10 (A), 53D (B) and 53 G (C) tribodies (●). The bi-specific immunomodulatory mAbs TR0304 and TR0506 (□), or their combinations with 53 P (○) were tested in parallel assays at the indicated concentrations (0.1–10 nM). Co-cultures untreated or treated with an unrelated IgG (*) were used as negative controls. Cell lysis was measured by detecting LDH release. Error bars depicted means ± SD. P values were calculated by comparing each treatment with the tri-specific tribody to that of the combination of 53 P with the bi-specific tribodies containing the same targeting domain, and the P values reported are: **P < 0.01; *P < 0.05, by student’s t test (two variables).

Fig. 7. In vivo anti-tumor efficacy of the novel tribody 53L10 compared with atezolizumab, 53 P and their combination.

Tumor growth curves of A-549 subcutaneous tumors treated with atezolizumab or the novel tribody 53L10 in the presence of human PBMCs. In particular, 5 × 10⁶ A-549 cells mixed with the same number of activated human PBMCs were subcutaneously transplanted into the right flank of NOD/SCID mice. hPBMCs were activated by stimulation with Dynabeads Human T-Activator CD3/CD28 (Veritas) for 4 days before transplantation and then injected. Intravenous administrations of vehicle (PBS; ●), 53 P (△), or 53L10 (○) on Days 0, 2, 4, 6, 8, and 10 (6 times in total) after cell transplantation were performed at the doses indicated in the figure. Intravenous administrations of atezolizumab (□) at the dose of 200 μg were performed on Days 0, 4, and 8 (3 times in total as indicated by the arrows). The combined treatment of 53 P with atezolizumab (◇) was carried out by administering 53 P on Days 0, 2, 4, 6, 8, 10 and atezolizumab on Days 0, 4, 8. Tumor volumes are expressed as means ± standard deviation (SD). Tukey’s multiple test (for all the combinations) was performed as a significance test. *P < 0.05 vs vehicle (PBS) group, **P < 0.05 (53 P + atezolizumab vs 53L10). The table reports the tumor volume and TGI in each treatment group at Day 42 (final day of the study). Three tumor volume and TGI obtained in each treatment group on Day 42 (final day of the in vivo study).