Prediction and design of protease enzyme specificity using a structure-aware graph convolutional network.

Site-specific proteolysis by the enzymatic cleavage of small linear sequence motifs is a key posttranslational modification involved in physiology and disease. The ability to robustly and rapidly predict protease-substrate specificity would also enable targeted proteolytic cleavage by designed proteases. Current methods for predicting protease specificity are limited to sequence pattern recognition in experimentally derived cleavage data obtained for libraries of potential substrates and generated separately for each protease variant. We reasoned that a more semantically rich and robust model of protease specificity could be developed by incorporating the energetics of molecular interactions between protease and substrates into machine learning workflows. We present Protein Graph Convolutional Network (PGCN), which develops a physically grounded, structure-based molecular interaction graph representation that describes molecular topology and interaction energetics to predict enzyme specificity. We show that PGCN accurately predicts the specificity landscapes of several variants of two model proteases. Node and edge ablation tests identified key graph elements for specificity prediction, some of which are consistent with known biochemical constraints for protease:substrate recognition. We used a pretrained PGCN model to guide the design of protease libraries for cleaving two noncanonical substrates, and found good agreement with experimental cleavage results. Importantly, the model can accurately assess designs featuring diversity at positions not present in the training data. The described methodology should enable the structure-based prediction of specificity landscapes of a wide variety of proteases and the construction of tailor-made protease editors for site-selectively and irreversibly modifying chosen target proteins.

Prediction and Design of Protease Enzyme Specificity Using a Structure-Aware Graph Convolutional Network.

Site-specific proteolysis by the enzymatic cleavage of small linear sequence motifs is a key post-translational modification involved in physiology and disease. The ability to robustly and rapidly predict protease substrate specificity would also enable targeted proteolytic cleavage - editing - of a target protein by designed proteases. Current methods for predicting protease specificity are limited to sequence pattern recognition in experimentally-derived cleavage data obtained for libraries of potential substrates and generated separately for each protease variant. We reasoned that a more semantically rich and robust model of protease specificity could be developed by incorporating the three-dimensional structure and energetics of molecular interactions between protease and substrates into machine learning workflows. We present Protein Graph Convolutional Network (PGCN), which develops a physically-grounded, structure-based molecular interaction graph representation that describes molecular topology and interaction energetics to predict enzyme specificity. We show that PGCN accurately predicts the specificity landscapes of several variants of two model proteases: the NS3/4 protease from the Hepatitis C virus (HCV) and the Tobacco Etch Virus (TEV) proteases. Node and edge ablation tests identified key graph elements for specificity prediction, some of which are consistent with known biochemical constraints for protease:substrate recognition. We used a pre-trained PGCN model to guide the design of TEV protease libraries for cleaving two non-canonical substrates, and found good agreement with experimental cleavage results. Importantly, the model can accurately assess designs featuring diversity at positions not present in the training data. The described methodology should enable the structure-based prediction of specificity landscapes of a wide variety of proteases and the construction of tailor-made protease editors for site-selectively and irreversibly modifying chosen target proteins.

Age-Dependent Changes in the Plasma and Brain Pharmacokinetics of Amyloid-ß Peptides and Insulin.

BACKGROUND: Age is the most common risk factor for Alzheimer's disease (AD), a neurodegenerative disorder characterized by the hallmarks of toxic amyloid-ß (Aß) plaques and hyperphosphorylated tau tangles. Moreover, sub-physiological brain insulin levels have emerged as a pathological manifestation of AD. OBJECTIVE: Identify age-related changes in the plasma disposition and blood-brain barrier (BBB) trafficking of Aß peptides and insulin in mice. METHODS: Upon systemic injection of 125I-Aß40, 125I-Aß42, or 125I-insulin, the plasma pharmacokinetics and brain influx were assessed in wild-type (WT) or AD transgenic (APP/PS1) mice at various ages. Additionally, publicly available single-cell RNA-Seq data [GSE129788] was employed to investigate pathways regulating BBB transport in WT mice at different ages. RESULTS: The brain influx of 125I-Aß40, estimated as the permeability-surface area product, decreased with age, accompanied by an increase in plasma AUC. In contrast, the brain influx of 125I-Aß42 increased with age, accompanied by a decrease in plasma AUC. The age-dependent changes observed in WT mice were accelerated in APP/PS1 mice. As seen with 125I-Aß40, the brain influx of 125I-insulin decreased with age in WT mice, accompanied by an increase in plasma AUC. This finding was further supported by dynamic single-photon emission computed tomography (SPECT/CT) imaging studies. RAGE and PI3K/AKT signaling pathways at the BBB, which are implicated in Aß and insulin transcytosis, respectively, were upregulated with age in WT mice, indicating BBB insulin resistance. CONCLUSION: Aging differentially affects the plasma pharmacokinetics and brain influx of Aß isoforms and insulin in a manner that could potentially augment AD risk.

The Relationship Between Dopamine Neurotransmitter Dynamics and the Blood-Oxygen-Level-Dependent (BOLD) Signal: A Review of Pharmacological Functional Magnetic Resonance Imaging.

Functional magnetic resonance imaging (fMRI) is widely used in investigations of normal cognition and brain disease and in various clinical applications. Pharmacological fMRI (pharma-fMRI) is a relatively new application, which is being used to elucidate the effects and mechanisms of pharmacological modulation of brain activity. Characterizing the effects of neuropharmacological agents on regional brain activity using fMRI is challenging because drugs modulate neuronal function in a wide variety of ways, including through receptor agonist, antagonist, and neurotransmitter reuptake blocker events. Here we review current knowledge on neurotransmitter-mediated blood-oxygen-level dependent (BOLD) fMRI mechanisms as well as recently updated methodologies aimed at more fully describing the effects of neuropharmacologic agents on the BOLD signal. We limit our discussion to dopaminergic signaling as a useful lens through which to analyze and interpret neurochemical-mediated changes in the hemodynamic BOLD response. We also discuss the need for future studies that use multi-modal approaches to expand the understanding and application of pharma-fMRI.