Generation of Multispecific Antibodies with Immune Cell Modulating Functions

Within the field of biologics, monoclonal antibodies have ruled the market with blockbuster and top selling drugs over the last decades. With the advancements in technological discoveries, lightspeed progress has been achieved to discover novel antibodies and mechanisms of action to address unmet needs and indications. Particularly, display technologies, in vitro systems, and high/ultra-high-throughput technologies have shown monumental progress to ensure the rapid discovery of novel biological entities. In the first study presented herein, the focus laid on the generation of a bidirectional plasmid for recombinant antibody production in mammalian cells to facilitate native antibody folding and post-translational modifications. Conventional approaches for transient antibody production utilise co-transfection of heavy and light chain genes encoded on separate plasmids. Here, a single plasmid under the control of two independent promoters, constructed in a bidirectional fashion, is used. This study assessed promoter combinations resulting in the best antibody yields of two U.S. Food and Drug Administration (FDA)-approved antibodies, durvalumab and avelumab. By comparing promoters with varying strengths (CMV, minCMV, EF-1α and enhanced CMV), gene expression of heavy and light chain genes and subsequent IgG1 yields gave rise to the 2xeCMV combination, consisting of two mirrored eCMV cassettes controlling the expression of the light and heavy chains individually in each direction. This combination effectuated the most promising mRNA synthesis for both chains in two regularly used mammalian cell lines, human embryonic kidney 293 (HEK293) and Chinese hamster ovary (CHO) cells, and the highest yields after IgG quantification, comparable to the conventional co-transfection method. By substituting the co-transfection approach with this bidirectional plasmid, lower plasmid preparation efforts are required and further facilitates the handling of a higher number of mAb candidates simultaneously. In the second study, the described bidirectional plasmid was put into practise by generating a Fab-presenting yeast surface display (YSD) library from immunised OmniRats. After screening of antibody formats via fluorescence-activated cell sorting (FACS), reformation of single candidates into their final IgG format is required, rapidly converting itself into a cumbersome step, and often resulting in the bottleneck to proceed with further characterisation. Within this study, a novel workflow based on Golden Gate Cloning (GGC) was established, allowing the bulk reformatting of antibody candidates after YSD FACS screening. By using an OmniRat-derived Fab library against MerTK, two screening rounds of YSD were performed by FACS. Subsequently, the antibody-encoding genes were transferred into a Mammalian_Destination (MD) vector, which contained a partial hinge-CH2-CH3 sequence, resulting in a full-length heavy chain after GGC with Esp3I. In order to produce the full-length variants, the yeast-specific Gal1,10 promoter was exchanged for the described promoter cassette combinations from the first study, 2xeCMV, by a final GGC step with BbsI. After assembly, the resulting MD vector contained the variable domains from the second sorting round with the respective constant domains required for the production of full-length IgG molecules. Next generation sequencing (NGS) of the screening rounds confirmed that the entire VH family diversity was covered in the resulting clones after bulk reformatting. Ten candidates were subsequently transiently expressed in mammalian cells and characterised for target binding and biophysical properties. This workflow presented a two-pot, two-step, PCR-free method to transition from YSD to a mammalian expression vector, eliminating any unwanted polymerase-introduced mutations and allowing for bulk cloning of yeast display-enriched antibody fragments. By this procedure, heavy and light chain pairing is conserved, contrary to other reformatting approaches, and paves the way to accelerate antibody hit discovery campaigns with YSD. Furthermore, this platform is malleable to other antibody formats and immunisation hosts, such as single chain variable fragments (scFvs) and chickens, and has the potential to be developed for bispecific or multispecific antibodies. Next-generation antibodies, including bi- and multispecific antibodies, have been set under the spotlight for their ability to combine multiple modes of action simultaneously and result in higher efficacy, where monoclonal antibodies are lacking. A special class of such are immune cell engagers which target immune cells and tumour-associated antigens (TAAs) to create an immune synapse. Depending on the effector cell being targeted, specialised killing mechanisms are triggered to efficiently kill the targeted cells. Macrophage engagers are aimed at forcing targeted phagocytosis of the engaged cell type and have typically targeted the CD47/SIRPα axis up to date, known as the “do not eat me” signal. Nevertheless, targeting CD47 lacks specificity due to its ubiquitous expression pattern. On the other hand, T-cell engagers (TCEs) result in very specialised signals by targeting CD3 on T cells and additional TAAs. The hyperactivation of T cells results in a feedback loop through the activation of macrophages and the over-release of cytokines, resulting in cytokine storms or cytokine release syndrome (CRS). If left untreated, these can provoke life-threatening conditions. Thus, macrophage engagers and TCEs require novel cell-specific targets and widening of their therapeutic windows for restored patient alleviation. In the third study within this cumulative thesis, the first bispecific macrophage engager targeting the receptor tyrosine kinase MerTK and epidermal growth factor receptor (EGFR) is presented. From the 10 antibody candidates derived in the second study, one candidate displayed agonistic properties, detected by the dose-dependent activation of the downstream signalling molecule phospho AKT (pAKT). MerTK’s overexpression on macrophages and tumour-associated macrophages within the tumour microenvironment (TME) lays the foundation to generate macrophage-engaging bispecific antibodies for targeted phagocytosis of tumour cells. Therefore, tandem biparatopic EGFR-binding VHH molecules (termed 7D9G) were combined in different architectures to generate bispecific molecules. By using the Knob-into-Hole technology, a bispecific with a MerTK-binding Fab arm and an EGFR-binding tandem VHH arm were generated, abolishing the agonistic properties of the parental MerTK mAb. On the other hand, a tetravalent bispecific antibody was generated by fusing the tandem VHHs to the C-terminus of the CH3 domain, resulting in intact MerTK-binding Fabs. The bispecific antibodies were able to bind both targets simultaneously in their soluble form and engage macrophages with EGFR+ tumour cells. Furthermore, they were able to compete with the binding site of EGF and therefore inhibit EGF-mediated signalling transduction by inhibiting pAKT. EGFR domain mapping of 7D9G by YSD resulted in binding to domain III of the extracellular EGFR domain, confirming its ligand-inhibiting abilities. Moreover, the bispecific antibodies resulted in targeted phagocytosis of EGFR+ tumour cells by macrophage-like THP-1 cells. This work represents the first bispecific macrophage-engager targeting MerTK for immuno-oncology applications by harnessing its expression and role in the tumour microenvironment to selectively phagocytise tumour cells. In the last study presented here, a trispecific T-cell engager and cytokine release modulating antibody (TriTECM) was generated. In brief, a tetravalent, bispecific two-in-one antibody binding EGFR and PD-L1 simultaneously with a single Fab arm was combined with anti-CD3 and anti-IL-6R scFvs, derived from foralumab or sarilumab, respectively. By testing two TriTECM architectures varying mainly in the anti-CD3 scFv positioning and valency of IL-6R binding, tetraspecific molecules were generated with multiple mechanisms of action. Firstly, increased tumour specificity was ensured by targeting EGFR and PD-L1 with a low nanomolar on-cell affinity. Checkpoint inhibition by blockage of the PD-1/PD-L1 axis was mediated by binding to PD-L1. T-cell engagement and subsequent T-cell-mediated cytotoxicity was attenuated, resulting in reduced pro-inflammatory cytokine release. And lastly, inhibition of the IL-6/IL-6R pathway can modulate cytokine storms after T-cell activation. The attenuation of CD3 binding could allow existing CD3-binders to be used, that were previously shown to result in cytotoxicity. With cytokine release still putting obstacles in the way of novel immune cell engagers, TriTECM designs represent a novel class of therapeutics with the potential to inertly modulate over-activated immune responses and widen the therapeutic index of T-cell-engaging therapeutics