Bi-specific T cell engagers (BiTes) are in clinical use as antitumor agents. Dual specificity is achieved through the creation of fusions of a T cell recognition module, typically anti-CD3ε, and an antitumor component. The success of BiTes derives from their ability to direct T cells, regardless of their specificity, to the tumor cell and exploit their effector functions.23,24 Here we extend the concept of bi-specific engagers by exploiting the ability of fusion proteins to target cells relevant to the antitumor response, and harness the effector functions of all Ig classes in the absence of deliberate immunization.
Nanobodies and their derivatives have advantages over the use of conventional antibodies, including: (1) high-yield expression including in prokaryotic expression systems due to their smaller size, superior stability and solubility, as well a lesser dependency on disulfide bond formation and glycosylation; (2) the ability to access epitopes less accessible to conventional antibodies, facilitated by their extended complementarity-determining region 3 (CDR3) loop; (3) low immunogenicity owing to their sequence homology to human immunoglobulin (Ig) V regions and humanization strategies; and (4) their small size and ease of modification, allowing for configurations that are more challenging to achieve with conventional antibodies.20,33,34 However, when used as stand-alone therapeutic agents, nanobodies suffer from a short circulatory half-life (< 30 minutes in mice and 0.5–2 hours in humans).35,36 Nanobodies obviously lack Fc-dependent effector functions. To overcome the latter shortcomings, a widely used strategy is to conjugate the nanobody of interest with an albumin-specific nanobody.37,38 These conjugates extend their half-life by binding to serum albumin in the bloodstream but cannot exert Fc-dependent effector functions. Alternatively, conjugating the nanobody to the Fc fragment can address the lack of Fc-dependent functions, though it compromises some of the nanobody's desirable properties.39 The approach described here achieves both circulatory half-life extension and provides Fc-dependent effector functions, not limited to a single Ig isotype.
Adult mice have the peculiar trait, in that > 95% of all Ig isotypes carry the kappa light chain,40 possibly driven by the large number of V k genes that provide the substrate for selection of B cells of the requisite specificity and a l light chain locus of much reduced complexity. In humans, the percentage of circulating Igs carrying kappa light chains is ~ 60%.41 We do not anticipate this difference to impact the effectiveness of mouse vs. human VHHkappa conjugates significantly. Given the quantities of circulating Igs in humans and mice, as well as the doses of VHHkappa conjugates administered, only a small fraction of the circulating Igs will bind the conjugates. The availability of a nanobody specific for human Ig kappa light chains with picomolar binding affinity suggests the potential for using the VHHkappa conjugate strategy in humans.42
We show improved activity of the H11-VHHkappa conjugate over 9H10 monoclonal antibody in two tumor models. We attribute this improvement to the engagement of all Fc-associated functions, rather than the reliance on a single Fc portion. MC38 is notoriously unresponsive to CTLA-4 blockade,43,44 yet all the H11-VHHkappa conjugate-treated mice successfully rejected tumors. Using H11-Cy5, we confirmed higher levels of CTLA-4 expression on intratumoral Tregs than on splenic Tregs (Fig. 2D) as a possible mechanism for the tumor-specific depletion of Tregs.45 Using FcgR-deficient and complement-deficient mice, we demonstrate that Treg depletion is mediated by FcgR. The fraction of Tregs in the spleen increased upon treatment with the H11-VHHkappa conjugate (Fig. 3C). This was still the case in FcgR -deficient and complement-deficient mice (Fig. 3D), suggesting an Fc-independent mechanism resulting from CTLA-4 signal blockade itself. Our data suggest that polyclonal Ig recruitment and harnessing the diversity of Fc-mediated effects could improve activity of CTLA-4 blockade in human. In MC38 and B16-F10 models, the anti-PD-L1 A12-VHHkappa conjugate outperformed the reference 10F.9G2 monoclonal antibody but failed to confer complete protection. Flow cytometry of established tumors showed that PD-L1 is expressed on both MC38 tumor cells and tumor-infiltrating CD11b+ myeloid cells (Fig. 2E). Because PD-L1 knockout MC38 tumor-bearing mice failed to respond to the A12-VHHkappa conjugate therapy, expression of PD-L1 on tumor cells is required for therapeutic efficacy, supported by the imaging data (Fig. 4).
We hypothesize that depletion of tumor-associated macrophages in the MC38 model is key to initiate tumor rejection. To improve on the suboptimal activity of the A12-VHHkappa conjugate, we incorporated a cytotoxic drug, DM4. Treatment with A12-VHHkappa-DM4 improved tumor rejection in both the MC38 and B16-F10 models. MC38 tumors treated with A12-VHHkappa-DM4 contained few live cells and had fewer CD11b+CD45+ myeloid cells (Fig. 6D), suggesting that A12-VHH kappa-DM4 increased the depletion of tumor-associated macrophages. Delivery of pro-inflammatory signals to the TME promotes tumor clearance.46,47 Site-specific incorporation of a STING agonist in the A12-VHHkappa conjugate improved rejection of MC38 tumors. A12-VHHkappa STING agonist induced activation of CD8+ and CD4 T+ cells (Fig. 7C). Incorporation of the STING agonist in the A12-VHHkappa conjugate delayed tumor growth in the B16-F10 model but failed to improve overall survival.
The greater antitumor activity of the VHHkappa conjugates when compared with the widely used anti-CTLA-4 monoclonal antibody suggests that different FcR+ cells may be responsible for this effect. The contribution of different FcR+ cells in response to deployment of the VHHkappa conjugates requires further exploration. Given the conservation of multiple Ig isotypes across vertebrate species, there may well be an advantage to the engagement of multiple classes of Igs over a single one. We compared the efficacy of monoclonal antibodies specific for CTLA-4 or PD-L1 with the corresponding VHHkappa conjugates and found the latter to be superior. Perhaps other monoclonal antibodies would perform better than those used here, which have been widely used preclinically, but we consider our results as supporting the value of multiple Ig engagement.
The VHHkappa-nanobody approach can be extended to target additional cancer-specific markers. Eradication of CD20-, HER2-, or EGFR-positive cancer cells by therapeutic antibodies is driven by ADCC/CDC as well as ADCP.7 Immunomodulatory reagents can deplete specific immunosuppressive cells to enhance antitumor activities of the remaining T cells. Appealing targets for immunomodulation by VHHkappa conjugates include OX40, CD25, GITR, CD47 and CD73.7,8,48 Our findings support the pursuit of bispecific antibody engagers to target additional cancer-specific and immunomodulatory markers.