Precision Targeting of Tumor antigens by glycan Integrated Design Antibody Libraries

Targeted cancer therapy with monoclonal antibodies has been successful for many cancers. However, their efficacy is limited by the availability of the optimal antibody targets. Some of the most targetable class of antigens include cell surface proteins that are frequently overexpressed in tumors. However, their presence both in normal and malignant cells are major limitations for novel immunotherapeutic strategies like chimeric antigen receptor (CAR) T cells as severe toxicities have been reported in clinical trials. Therefore, the ideal tumor target must be present exclusively on tumor cells and not on healthy cells to prevent on-target/off-tumor toxicities.

In that context, protein glycosylation can offer novel targets for immunotherapy. Aberrant glycosylation is a hallmark of cancer that results in short truncated O-glycans, such as the Tn antigen and its sialylated form- STn antigen- that are exclusively present on tumor cells. Numerous efforts have taken place the last decades to develop anti-glycan antibodies mainly through hybridoma technology. Hybridoma strategies resulted in the development of hapten antibodies, antibodies against peptide backbones but the development of antibodies against glycopeptide epitopes like 5E5, are more scarce.

However, faster antibody development can be achieved by utilizing phage display libraries designed to specifically isolate antibodies against glycopeptide targets, the so called combotopes. To achieve this goal, the variable heavy chain of G2D11, a previously identified single-chain variable fragment (scFv) that binds two adjacent GalNac residues, was successfully shuffled with naïve murine light chains to create a Tn-templated library. In addition, a mutant variant of G2D11 that specifically binds two adjacent sialylated GalNac residues was utilized to create a second library, the STn-templated library.

As a proof of concept, MUC1 was used to identify novel binders. After three selection rounds and the subsequent soluble expression, specific bisTn-MUC1 scFvs were isolated. Their therapeutic potentials were evaluated on breast adenocarcinoma cells lines while next generation sequencing with nanopore technology, revealed that some of the isolated binders shared the same CDR3 sequence with 5E5. CD43 was the first target that Tn-templated library was panned against. CD43 specific scFvs were identified, that were able to recognize CD43 on Jurkat cells with unique CDR1 and CDR3 sequences as sequencing results revealed. Finally, scFvs against STn-MUC1 were identified while the possibility to isolate scFvs against glycopeptides with one glycosylation site was tested and indeed STn-MUC1 specific scFvs were identified. However, the STn-MUC1 scFvs have yet to be further validated.

The major advantage of the constructed libraries is the glycoform specificity which can be pre-selected and the carrier protein specificity will be determined by the variable light chain contribution. In fact, all the identified scFvs showed higher binding affinity compared to G2D11 as it was determined by the affinity studies. In addition, employing nanopore sequencing reveals the diversity of the VL sequences that are enriched in every round as well as the expected sequences of specific target binders.

To conclude, the two generated libraries hold great potentials to generate novel antibodies against any aberrant O-glycosylated glycoprotein. In this way, we can generate specific antibodies against combotopes in a fast and efficient way. The ultimate goal is to further evaluate the developed antibodies as novel therapeutic agents.