A gB/CD3 bispecific BiTE antibody construct for targeting Human Cytomegalovirus-infected cells

Abstract

Bispecific T cell engager (BiTE) antibody constructs are successfully used as cancer therapeutics. We hypothesized that this treatment strategy could also be applicable for therapy of human cytomegalovirus (HCMV) infection, since HCMV-encoded proteins are abundantly expressed on the surface of infected cells. Here we show that a BiTE antibody construct directed against HCMV glycoprotein B (gB) and CD3 efficiently triggers T cells to secrete IFN-γ and TNF upon co-culture with fibroblasts infected with HCMV strain AD169, Towne or Toledo. Titration of gB expression levels in non-infected cells confirmed that already low levels of gB are sufficient for efficient triggering of T cells in presence of the BiTE antibody construct. Comparison of redirecting T cells with the bispecific antibody versus a chimeric antigen receptor (CAR) based on the same scFv showed a similar sensitivity for gB expression. Although lysis of infected target cells was absent, the BiTE antibody construct inhibited HCMV replication by triggering cytokine production. Notably, even strongly diluted supernatants of the activated T cells efficiently blocked the replication of HCMV in infected primary fibroblasts. In summary, our data prove the functionality of the first BiTE antibody construct targeting an HCMV-encoded glycoprotein for inhibiting HCMV replication in infected cells.

Figure 1

Titration of the gB-BiTE antibody construct. (a) Flow cytometric analysis of the amounts of BiTE antibody construct bound to T cells, which were previously activated and expanded by anti-CD3/CD28-antibody coated beads. (b) Normalized levels of IFN-γ secreted by activated or resting T cells, respectively, upon co-culture with non-infected or HCMV-infected HFF (AD169, MOI 5, day 4 p.i.) and an increasing amount of the BiTE antibody construct (mean ± standard deviation; n = 3). (c) Normalized levels of TNF secreted by CD3/CD28-activated T cells upon co-culture with non-infected or HCMV-infected HFF (AD169, MOI 5, day 4 p.i.) and an increasing concentration of the BiTE antibody construct (mean ± standard deviation; n = 3). IFN-γ and TNF levels were normalized for each donor to the value obtained with activated T cells co-cultured with HCMV-infected HFF and 1000 ng/ml BiTE antibody construct.

Figure 2

Comparison of the efficiency of redirecting T cells to gB by either the BiTE antibody construct or a CAR. Activated T cells were redirected by either the gB-BiTE antibody construct or a gB-CAR and co-cultured with 293T cells, which were electroporated with increasing amounts of chimeric EpCAM-gB mRNA, as indicated. (a) Expression levels of EpCAM-gB in the target cells 293T 18 hours after electroporation with different amounts of EpCAM-gB mRNA. (b) Expression of the gB-CAR in CD3/CD28-activated primary T cells of one representative experiment. (c) Proportion of degranulating T cells as determined by flow cytometric analysis of cell surface expression of CD107a. (d) Normalized levels of secreted IFN-γ (values obtained with gB-BiTE antibody construct at 10 µg EpCAM-gB mRNA were set to 100%). (c,d) display the mean values ± standard deviation of experiments with three donors.

Figure 3

Recognition of fibroblasts at different time points after infection with different HCMV strains. (a) Expression levels of gB in HFF within 4 days after infection with the HCMV strains AD169, Towne and Toledo (MOI 1). Shown is one representative experiment. (b) Normalized levels of IFN-γ secreted by CD3/28 activated T cells, which were redirected by the gB-BiTE antibody construct and co-cultured with HFF at different time points after infection with the indicated HCMV strains (mean ± standard deviation; n = 3).

Figure 4

Triggering of cytotoxic effector functions by the gB-BiTE antibody construct. (a) Activated T cells were co-cultured with different target cells at an E:T ratio of 25:1 for 4 hours. The target cells 293T were either electroporated with mRNA encoding EpCAM-gB or were mock electroporated. HFF were used for experiments on day 4 after infection (AD169, MOI 5), non-infected cells were used as a control. Data of two representative donors are shown. (b) Proportion of degranulating T cells (characterized by CD107a surface expression) upon co-culture with different target cells as indicated (n = 3).

Figure 5

Inhibition of HCMV infection in HFF by supernatants of T cells redirected by the gB-BiTE antibody construct. (a) Levels of IFN-γ and TNF secreted by activated T cells with or without BiTE antibody construct after co-culture with HFF, non-infected or infected with HCMV (AD169, MOI 5, day 4 p.i.), as indicated (n = 3). For the experiments shown in (b), serial dilutions (1:3, 1:10, 1:30) of the supernatants of the co-cultures analysed in (a) were added to HFF throughout and after the infection with HCMV (AD169, MOI 0.25). The proportion of infected HFF was then analysed 4 days later by flow cytometric detection of GFP expression (n = 3; n = 2 with the supernatants from non-infected HFF without BiTE antibody construct). Additionally, control cultures without added supernatants were analysed for the proportion of infected cells one day after infection. The increase of GFP^pos cells until day 4 beyond these reference levels on day one indicates the fraction of cells, which were infected by the newly generated HCMV particles.