Predicting Proteolysis in Complex Proteomes Using Deep Learning.

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.

Circadian rhythms in skin and other elastic tissues.

Circadian rhythms are daily oscillations that, in mammals, are driven by both a master clock, located in the brain, and peripheral clocks in cells and tissues. Approximately 10% of the transcriptome, including extracellular matrix components, is estimated to be under circadian control. Whilst it has been established that certain collagens and extracellular matrix proteases are diurnally regulated (for example in tendon, cartilage and intervertebral disc) the role played by circadian rhythms in mediating elastic fiber homeostasis is poorly understood. Skin, arteries and lungs are dynamic, resilient, elastic fiber-rich organs and tissues. In skin, circadian rhythms influence cell migration and proliferation, wound healing and susceptibility of the tissues to damage (from protease activity, oxidative stress and ultraviolet radiation). In the cardiovascular system, blood pressure and heart rate also follow age-dependent circadian rhythms whilst the lungs exhibit diurnal variations in immune response. In order to better understand these processes it will be necessary to characterise diurnal changes in extracellular matrix biology. In particular, given the sensitivity of peripheral clocks to external factors, the timed delivery of interventions (chronotherapy) has the potential to significantly improve the efficacy of treatments designed to repair and regenerate damaged cutaneous, vascular and pulmonary tissues.

Selective proteolysis by matrix metalloproteinases of photo-oxidised dermal extracellular matrix proteins.

Photodamage in chronically sun-exposed skin manifests clinically as deep wrinkles and histologically as extensive remodelling of the dermal extracellular matrix (ECM) and in particular, the elastic fibre system. We have shown previously that loss of fibrillin microfibrils, a key elastic fibre component, is a hallmark of early photodamage and that these ECM assemblies are susceptible in vitro to physiologically attainable doses of ultraviolet radiation (UVR). Here, we test the hypotheses that UVR-mediated photo-oxidation is the primary driver of fibrillin microfibril and fibronectin degradation and that prior UVR exposure will enhance the subsequent proteolytic activity of UVR-upregulated matrix metalloproteinases (MMPs). We confirmed that UVB (280-315 nm) irradiation in vitro induced structural changes to both fibrillin microfibrils and fibronectin and these changes were largely reactive oxygen species (ROS)-driven, with increased ROS lifetime (D2O) enhancing protein damage and depleted O2 conditions abrogating it. Furthermore, we show that although exposure to UVR alone increased microfibril structural heterogeneity, exposure to purified MMPs (1, -3, -7 and - 9) alone had minimal effect on microfibril bead-to-bead periodicity; however, microfibril suspensions exposed to UVR and then MMPs were more structurally homogenous. In contrast, the susceptibly of fibronectin to proteases was unaffected by prior UVR exposure. These observations suggest that both direct photon absorption and indirect production of ROS are important mediators of ECM remodelling in photodamage. We also show that fibrillin microfibrils are relatively resistant to proteolysis by MMPs -1, -3, -7 and - 9 but that these MMPs may selectively remove damaged microfibril assemblies. These latter observations have implications for predicting the mechanisms of tissue remodelling and targeted repair.

Defining tissue proteomes by systematic literature review.

Defining protein composition is a key step in understanding the function of both healthy and diseased biological systems. There is currently little consensus between existing published proteomes in tissues such as the aorta, cartilage and organs such as skin. Lack of agreement as to both the number and identity of proteins may be due to issues in protein extraction, sensitivity/specificity of detection and the use of disparate tissue/cell sources. Here, we developed a method combining bioinformatics and systematic review to screen >32M articles from the Web of Science for evidence of proteins in healthy human skin. The resulting Manchester Proteome ( www.manchesterproteome.manchester.ac.uk ) collates existing evidence which characterises 2,948 skin proteins, 437 unique to our database and 2011 evidenced by both mass spectrometry and immune-based techniques. This approach circumvents the limitations of individual proteomics studies and can be applied to other species, organs, cells or disease-states. Accurate tissue proteomes will aid development of engineered constructs and offer insight into disease treatments by highlighting differences in proteomic composition.

A potential role for endogenous proteins as sacrificial sunscreens and antioxidants in human tissues.

Excessive ultraviolet radiation (UVR) exposure of the skin is associated with adverse clinical outcomes. Although both exogenous sunscreens and endogenous tissue components (including melanins and tryptophan-derived compounds) reduce UVR penetration, the role of endogenous proteins in absorbing environmental UV wavelengths is poorly defined. Having previously demonstrated that proteins which are rich in UVR-absorbing amino acid residues are readily degraded by broadband UVB-radiation (containing UVA, UVB and UVC wavelengths) here we hypothesised that UV chromophore (Cys, Trp and Tyr) content can predict the susceptibility of structural proteins in skin and the eye to damage by physiologically relevant doses (up to 15.4 J/cm(2)) of solar UVR (95% UVA, 5% UVB). We show that: i) purified suspensions of UV-chromophore-rich fibronectin dimers, fibrillin microfibrils and ß- and γ-lens crystallins undergo solar simulated radiation (SSR)-induced aggregation and/or decomposition and ii) exposure to identical doses of SSR has minimal effect on the size or ultrastructure of UV chromophore-poor tropoelastin, collagen I, collagen VI microfibrils and α-crystallin. If UV chromophore content is a factor in determining protein stability in vivo, we would expect that the tissue distribution of Cys, Trp and Tyr-rich proteins would correlate with regional UVR exposure. From bioinformatic analysis of 244 key structural proteins we identified several biochemically distinct, yet UV chromophore-rich, protein families. The majority of these putative UV-absorbing proteins (including the late cornified envelope proteins, keratin associated proteins, elastic fibre-associated components and ß- and γ-crystallins) are localised and/or particularly abundant in tissues that are exposed to the highest doses of environmental UVR, specifically the stratum corneum, hair, papillary dermis and lens. We therefore propose that UV chromophore-rich proteins are localised in regions of high UVR exposure as a consequence of an evolutionary pressure to express sacrificial protein sunscreens which reduce UVR penetration and hence mitigate tissue damage.

The FRIENDS Emotional Health Programme: Initial Findings from a School-Based Project.

Emotional disorders in children are common and although effective interventions are available comparatively few receive specialist help. School nurses were trained to deliver an evidence-based emotional health cognitive behaviour therapy programme, FRIENDS, to 106 non-referred children aged 9-10 attending three schools. Levels of anxiety and self-esteem were stable in the 6-month period before FRIENDS. Three months after completing FRIENDS, anxiety had significantly decreased and self-esteem increased. Children with the most severe emotional problems benefited from the programme. The value of delivering standardised evidence based programmes in schools by school nurses is discussed and the need for further research highlighted.