The first challenge in the construction of a new protease-based signaling cascade was the selection of appropriate proteases. The candidate proteases should recognize defined target cleavage sequences, preferably as long as possible; they should be active in mammalian cells, but not toxic to them and inducible, ideally through the reconstitution of split protein fragments. Most importantly, a large number of proteases with similar properties but with different specificities should be available to allow for modular construction of signaling pathways and logic functions and these proteases should be orthogonal to each other, meaning they should have specific cleavage sites not recognized by the other proteases in the system.

We found that the tobacco etch virus protease (TEVp) was the only protease described in the literature to match our criteria.

To overcome this lack of inducible orthogonal proteases, we looked for the characterized TEVp mutants and naturally occurring proteases closely related to TEVp that might also be used to function as split proteases.

TEVp variants

Based on the sequence alterations described by Yi et al. Yi2013 our team decided to test the two variants of TEVp - TEVpE and TEVpH.

To test these two proteases we used a cleavable firefly luciferase (fLuc) reporter with an appropriate cleavage sequence inserted in a permissible site. We observed a significant decrease in the fLuc activity upon coexpression of the reporters with their corresponding proteases, whereas the coexpression of reporters with an orthogonal protease resulted in a much lower decrease of fLuc activity (1). These results were additionally confirmed by results from western blot where the cleaved luciferase was detected only in cells cotransfected with a reporter and its corresponding protease but not with other reporter-protease combinations (2).

No data has previously been reported on TEVpE and TEVpH toxicity. Therefore we performed a viability test for expression of all three TEVp variants in HEK293T cells. Even after transfection with a high amount of the plasmid for each respective protease, the cells showed high viability, with practically no difference when compared to control transfections (3).

TEVp homologs

Introduction of two new TEVp variants expanded our repertoire of tools, demonstrating that we can use the results of the mutational screening of protease variants, but a larger number of strictly orthogonal proteases would be required for modular design of logic gates. We therefore decided to investigate activity of de novo created split proteases from the potyviridae family.

We searched for sequences of different potyviruses available on UniProt, paying particular attention to any evidence of orthogonality among their target substrates. We decided to test the plum pox virus protease (PPVp), the soybean mosaic virus protease (SbMVp), and the sunflower mild mosaic virus protease (SuMMVp).

All selected potyviral proteases were designed as synthetic genes and tested in mammalian cells for the activity using the cyclic luciferase reporters, which results in the luciferase activity only upon cleavage. We detected an increase of luciferase activity only in the corresponding protease-reporter pairs, confirming exquisite orthogonality of the selected proteases and their activity in the human cell chassis (4).