Engineering Proteases to Detect Post-Translational Modifications

Proteases, enzymes that catalyze the hydrolysis of amide bonds in peptides and proteins, are ubiquitous across all forms of life and affect every protein in its natural life cycle. Engineering programmed specificity onto stable protease scaffolds holds promise as a route to a new generation of useful proteases with requisite biotechnological properties. One of the long-standing applications of proteases has been in generating peptides for mass spectrometry (MS) analyses in proteomic experiments. Since post-translational modifications (PTMs) can be labile, proteomic techniques relying on tandem MS can have difficulty in achieving comprehensive coverage of PTM sites within proteins. Proteases which selectively recognize PTMs may improve their identification in complex proteome samples and are easily adaptable to current proteomic methodologies. Chymotrypsin, trypsin and elastase are a family of serine proteases that have superimposable tertiary structures but recognize chemically differing substrates. Trypsin cleaves after basic residues (Arg/Lys) while chymotrypsin cleaves after large hydrophobic residues (Phe/Tyr/Trp), leading biochemists to believe that since the backbone architecture was identical, the origin of the differing substrate recognition capabilities must arise from the chemistry of the actual side-chains that comprise the substrate binding pockets. Unfortunately however, site-directed mutational swapping of selected residues within the binding pocket of trypsin did not endow chymotrypsin like reactivity and required the grafting of extended loops outside the substrate binding pocket to bring about the change in substrate specificity. These and other similar efforts, while noteworthy, serve to highlight the challenges associated with molecular redesign of substrate binding specificity within proteases.

In the current work, we have utilized a combinatorial approach to engineer the substrate binding pocket of chymotrypsin, and isolate variants that can recognize the presence/absence of glycosylation and phosphorylation modifications. This required us to develop a new platform to display mammalian chymotrypsin B on the surface of E.coli by fusing it with C-terminal barrel domain of antigen 43, a native membrane protein. Eleven residues in substrate binding pocket, chosen based on sequence alignment of chymotrypsin-fold proteases were randomized using oligonucleotides with NNS codons, cloned and transformed into E.coli MC1061 to obtain a library of 107 unique variants. High-throughput screening of the library was performed using FACS for simultaneous selection for activity towards Asn, and counter-selection against wild type-like activity, using peptide substrates containing a proteolysis-sensitive linker sandwiched between orthogonal FRET pairs. After six rounds of sorting, we observed sequence convergence from a library of 107 to a ChyB-Asn variant that showed desired fluorescence profile when measured using flow cytometer. Solution phase kinetics was employed to measure the Michaelis-Menten parameters of this variant and mass spectrometry used to confirm proteolysis after Asn containing peptide sequences. We anticipate that that our results can be a stepping stone for engineering differing substrate specificities onto the chymotrypsin scaffold and can help design the rules for molecular recognition within proteases. In parallel, we are also investigating the utility of the Chyb-Asn variant for glycoproteomic studies. Finally, our surface-dsiplay assisted combinatorial engineering methodology can readily be extended to engineering specificity towards other PTMs and we have also identified a chymotrypsin variant showing specific activity towards pTyr peptide substrates