Predicting Proteolysis in Complex Proteomes Using Deep Learning

Abstract

Both protease- and reactive oxygen species (ROS)-mediated proteolysis are thought to be key effectors of tissue remodeling. We have previously shown that comparison of amino acid composition can predict the differential susceptibilities of proteins to photo-oxidation. However, predicting protein susceptibility to endogenous proteases remains challenging. Here, we aim to develop bioinformatics tools to (i) predict cleavage site locations (and hence putative protein susceptibilities) and (ii) compare the predicted vulnerabilities of skin proteins to protease- and ROS-mediated proteolysis. The first goal of this study was to experimentally evaluate the ability of existing protease cleavage site prediction models (PROSPER and DeepCleave) to identify experimentally determined MMP9 cleavage sites in two purified proteins and in a complex human dermal fibroblast-derived extracellular matrix (ECM) proteome. We subsequently developed deep bidirectional recurrent neural network (BRNN) models to predict cleavage sites for 14 tissue proteases. The predictions of the new models were tested against experimental datasets and combined with amino acid composition analysis (to predict ultraviolet radiation (UVR)/ROS susceptibility) in a new web app: the Manchester proteome susceptibility calculator (MPSC). The BRNN models performed better in predicting cleavage sites in native dermal ECM proteins than existing models (DeepCleave and PROSPER), and application of MPSC to the skin proteome suggests that: compared with the elastic fiber network, fibrillar collagens may be susceptible primarily to protease-mediated proteolysis. We also identify additional putative targets of oxidative damage (dermatopontin, fibulins and defensins) and protease action (laminins and nidogen). MPSC has the potential to identify potential targets of proteolysis in disparate tissues and disease states.

Peptide Location Fingerprinting Reveals Tissue Region-specific Differences in Protein Structures in an Ageing Human Organ

“…The tissue region-conserved, higher peptide yields seen on the C-terminal ends of young COL5A1 and COL2A1 compared to aged (Figure 5a, ii and b, ii) also corresponded to their cleaved C-terminal propeptides, as seen for COL1A2 (Figure 3a, ii). This once again reflects a possible tissue-wide perturbation in collagen maintenance [46] as a consequence of ageing, for collagen II and V fibrils in addition to collagen I. Interestingly, we noticed that the C-terminal propeptides of COL1A2 and COL5A1 are enriched in ROSsensitive residues [56] (Figure . S6), indicating a potential role for these fragments as antioxidants, whose presence is reduced in aged IVD (Figure 3a, ii and Figure 5a, ii). Again in common with COL1A2 (Figure 3a, ii), aged COL2A1 also exhibited an AF region-wide significant increase in peptide yield compared to young, downstream to a prominent MMP cleavage site (Figure 5b, ii, black arrow) [51,52], suggesting that chronic degradation of collagen II by MMPs may be an ageing mechanism at play.…”

Peptide Location Fingerprinting Reveals Tissue Region-Specific Differences in Protein Structures in an Ageing Human Organ

“…The tissue region-conserved, higher peptide yields seen on the C-terminal ends of young COL5A1 and COL2A1 compared to the aged one ( Figure 5 a,b(ii)) also corresponded to their cleaved C-terminal propeptides, as seen for COL1A2 ( Figure 3 a(ii)). This once again reflects a possible tissue-wide perturbation in collagen maintenance [ 46 ] in aged compared to young discs, for collagen II and V fibrils in addition to collagen I. Interestingly, we noticed that the C-terminal propeptides of COL1A2 and COL5A1 are enriched in ROS-sensitive residues [ 56 ] ( Figure S6 ), indicating a potential role for these fragments as antioxidants, whose presence is reduced in aged IVD ( Figure 3 a and Figure 5 a(ii)). Again in common with COL1A2 ( Figure 3 a(ii)), aged COL2A1 also exhibited an AF region-wide significant increase in peptide yield compared to young COL2A1 , downstream to a prominent MMP cleavage site ( Figure 5 b(ii), black arrow) [ 51 , 52 ], possibly suggesting that chronic degradation of collagen II by MMPs may be an ageing mechanism at play.…”

Molecular de-extinction of ancient antimicrobial peptides enabled by machine learning

“…The development of this predictive tool was motivated by the hypothesis that protease-agnostic cleavage site prediction could facilitate biologically inspired prospection for encrypted host defense peptides. Prior cleavage site classifiers are specialized models that predict cleavage activity for only a subset of human proteases (9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). A pan-protease design facilitates proteomewide searches, circumventing the need to hypothesize protease-substrate relationships.…”