Our aim is to construct a transducer system that is able to efficiently recognise free ligands inside the blood to activate downstream reporter in real time. The SynNotch and CAR systems have all been proved potential for cell therapy, while both of them are limited to cell-surface antigens. That’s why we finally chose another new synthetic system, namely the Modular Extracellular Sensors Architecture, as our transducer framework, and the main part of our STEP system.

This cell membrane receptor consists of two chains, one chain of transcription factor (TC) and one of protease (PC). The two chains are in a free state under normal circumstances, but will dimerize in the presence of the ligands. Once these two chains get close the protease on the PC chain can cleave the peptide linker between the transcription factor and the transmembrane domain of the TC chain, then the transcription factor can be released into the cytoplasm to activate the reporter (Figure 1).
It is obvious that the extracellular domain can literally be any kinds of protein that can bind to the tumour markers, including natural receptors or binding domains as well as antibodies and scFvs. Two different binding domains can be used to recognise different sites of the ligand, while identical domains can also be used for dimers. Considering the accessibility of ligands and sequences, we decided to use the two dimers, VEGF and D-Dimer, as our targets for the demonstration of our system, one for broad-spectrum cancer targeting while the other for NSCLC. And as a consequence we are allowed to use the same binding domain for both TC and PC chains.

For VEGF recognition, we selected one of its single-chain variable fragment (scFv) as the extracellular domain of our receptors. The only VEGF-scFv in iGEM parts collection (BBa_K1694003) has no known structure, so we found ourselves another one with experimental structure (PDB ID: 2FJF & 2FJG, Figure 2) (BBa_K2886002). We then add a signal sequence to it as well as a transmembrane linker (BBa_K2886000) to help it appropriately locate on the cell membrane. In addition to that, we want to test another receptor, which is the binding domains of a natural VEGF receptor, the kinase insert domain receptor (KDR). We cloned the two IG-like domains of it (BBa_K2886003) to the extracellular domain to test its efficiency. By making use of fragment from endogenous receptors, we hope to achieve the goal to reduce the possible immunological rejection while keep the orthogonality of STEP system at the same time.
For D-Dimer, we were only able to find a MA-15C5 scFv of it. So we simply attach it to the TC and PC chains and test the whole system.

To see if the STEP system works and to optimise it, we have to fully test it in vivo. Considering the practicability, we decide to use human HEK293T and HeLa cell line as our chassis. These two cell lines have the advantage of handleability and effectiveness and with them we are able to carry out our experiments with ease and reliability. The first thing we need to confirm is that our extracellular domain works, or more specifically, that the two chains can correctly locate on the cell membrane and that they can bind when target ligand is added. To that end we design a bimolecular fluorescence complementation (BiFC) test. We remove the intracellular domains of both chains completely and instead attached the N- and C-terminal of a split ECFP (BBa_K2886004 & BBa_K2886005) respectively. It means that if the receptors are expressed on the membrane and can efficiently bind to the ligand, we’ll observe the fluorescent signal after induction. We took the ECFP from a previous part (BBa_E0020) and split it at the A155 site and use the addition of leucine zipper (BBa_K2886006) as a positive control, and test it in both E. coli BL21 and our two cell lines (Figure 3).
After the BiFC test is done we are able to test the whole STEP system. To guarantee a reliable reporter expression we first built a stable cell line of mCherry reporter with pTight promotor of both HEK293T and HeLa cells, and used transient transfection of transcription factor tTA to select a best monoclonal cell strain. After that we transfect it with a single chain or both chains of VEGF-STEP, adding VEGF to the solution and see if it makes a difference. We also test different transfection quantity and ratio of two chains, and build a model to study the best transfection condition (Figure 4).
Another work we decide to do is to raise the binding affinity of the scFvs and ligands. An improved binding affinity means a stronger attraction between the two chains when signal protein exists, thus increasing the dynamic range of our system. A little change of sequence on the interface of the extracellular domain can affect the structure of that interface to a large extent, and the binding affinity might be greatly increased or decreased. As it should be, we can certainly try every permutation of residues on the interface by mutating residues again and again and screen out the best sequence through wet experiment. A better strategy is to look before you leap. Instead of enumerating those possibilities of the sequence in wet lab, we can first test it virtually in dry lab. Here, we use Rosetta, a software for computational modeling and analyzing of protein structures, to accelerate the process of designing and redesigning. Through Rosetta we can achieve structure predicting, macromolecule interacting and sequence designing for the ligand binding domain of these two chains. After acquiring a certain mutation with the lowest energy, we can express the mutant protein, compare its binding energy with the original protein through ITC test and in turn model our work or optimise the algorithm we use (Figure 5).
In conclusion of our project design, we test different binding domains of VEGF and D-Dimer by both BiFC and overall experiment. Also, we attempt to carry out in silico protein design to increase the binding affinity and prove its results in wet lab. Through all these experiments we are able to construct, test and optimise the STEP system, and make it ready for application in cancer cell therapy.