AI-Powered Virtual Screening of Large Compound Libraries Leads to the Discovery of Novel Inhibitors of Sirtuin-1.

The discovery of new scaffolds and chemotypes via high-throughput screening is tedious and resource intensive. Yet, there are millions of small molecules commercially available, rendering comprehensive in vitro tests intractable. We show how smart algorithms reduce large screening collections to target-specific sets of just a few hundred small molecules, allowing for a much faster and more cost-effective hit discovery process. We showcase the application of this virtual screening strategy by preselecting 434 compounds for Sirtuin-1 inhibition from a library of 2.6 million compounds, corresponding to 0.02% of the original library. Multistage in vitro validation ultimately confirmed nine chemically novel inhibitors. When compared to a competitive benchmark study for Sirtuin-1, our method shows a 12-fold higher hit rate. The results demonstrate how AI-driven preselection from large screening libraries allows for a massive reduction in the number of small molecules to be tested in vitro while still retaining a large number of hits.

An interaction-based drug discovery screen explains known SARS-CoV-2 inhibitors and predicts new compound scaffolds.

The recent outbreak of the COVID-19 pandemic caused by severe acute respiratory syndrome-Coronavirus-2 (SARS-CoV-2) has shown the necessity for fast and broad drug discovery methods to enable us to react quickly to novel and highly infectious diseases. A well-known SARS-CoV-2 target is the viral main 3-chymotrypsin-like cysteine protease (Mpro), known to control coronavirus replication, which is essential for the viral life cycle. Here, we applied an interaction-based drug repositioning algorithm on all protein-compound complexes available in the protein database (PDB) to identify Mpro inhibitors and potential novel compound scaffolds against SARS-CoV-2. The screen revealed a heterogeneous set of 692 potential Mpro inhibitors containing known ones such as Dasatinib, Amodiaquine, and Flavin mononucleotide, as well as so far untested chemical scaffolds. In a follow-up evaluation, we used publicly available data published almost two years after the screen to validate our results. In total, we are able to validate 17% of the top 100 predictions with publicly available data and can furthermore show that predicted compounds do cover scaffolds that are yet not associated with Mpro. Finally, we detected a potentially important binding pattern consisting of 3 hydrogen bonds with hydrogen donors of an oxyanion hole within the active side of Mpro. Overall, these results give hope that we will be better prepared for future pandemics and that drug development will become more efficient in the upcoming years.

A multiscale model of the regulation of aquaporin 2 recycling.

The response of cells to their environment is driven by a variety of proteins and messenger molecules. In eukaryotes, their distribution and location in the cell are regulated by the vesicular transport system. The transport of aquaporin 2 between membrane and storage region is a crucial part of the water reabsorption in renal principal cells, and its malfunction can lead to Diabetes insipidus. To understand the regulation of this system, we aggregated pathways and mechanisms from literature and derived three models in a hypothesis-driven approach. Furthermore, we combined the models to a single system to gain insight into key regulatory mechanisms of Aquaporin 2 recycling. To achieve this, we developed a multiscale computational framework for the modeling and simulation of cellular systems. The analysis of the system rationalizes that the compartmentalization of cAMP in renal principal cells is a result of the protein kinase A signalosome and can only occur if specific cellular components are observed in conjunction. Endocytotic and exocytotic processes are inherently connected and can be regulated by the same protein kinase A signal.

Analysis of the influence of EDTA-treated reference samples on forensic bloodstain age estimation.

The age estimation of blood traces provides important leads for the chronological assessment of criminal events and their reconstruction. To determine bloodstain age, experimental comparative data from a laboratory environment are used. Under these conditions the utilization of anticoagulants such as EDTA helps to suppress the blood clotting mechanism to allow the examination over a longer time period. This unnatural prevention of blood coagulation is highly questionable when estimating bloodstain age, since the blood's physical and chemical properties are altered. For this reason, the authors determined actual influence of EDTA on blood spectra over time in order to formulate a statement as to whether this effect can be measured. Human and porcine blood samples were aged under controlled conditions. The resulting UV/VIS spectra were separated into their individual components using signal separation techniques, allowing the changes in the ratios of the individual hemoglobin derivatives to be observed over time. The results show a significant influence of EDTA on the conversion of oxyhemoglobin to methemoglobin and a minor influence on the conversion of methemoglobin to hemichrome within the relevant time range of 5-100 h. The use of EDTA thus slows down the aging process of blood spots. To illustrate the great influence of EDTA, spectra of untreated pig blood samples were included as comparison data. These show that the difference between EDTA-treated and untreated blood samples is as great as the difference between human blood and pig blood. As a consequence of our findings experimental comparative data for the age estimation of bloodstains should never result from EDTA-treated blood.

The structural basis of the genetic code: amino acid recognition by aminoacyl-tRNA synthetases.

Storage and directed transfer of information is the key requirement for the development of life. Yet any information stored on our genes is useless without its correct interpretation. The genetic code defines the rule set to decode this information. Aminoacyl-tRNA synthetases are at the heart of this process. We extensively characterize how these enzymes distinguish all natural amino acids based on the computational analysis of crystallographic structure data. The results of this meta-analysis show that the correct read-out of genetic information is a delicate interplay between the composition of the binding site, non-covalent interactions, error correction mechanisms, and steric effects.

Structural change in GadD2 of Listeria monocytogenes field isolates supports nisin resistance.

The lantibiotic nisin is used as a food additive to effectively inactivate a broad spectrum of Gram-positive bacteria such as Listeria monocytogenes. In total, 282 L. monocytogenes field isolates from German ready-to-eat food products, food-processing environments and patient samples and 39 Listeria reference strains were evaluated for their susceptibility to nisin. The MIC90 value was <1500 IU ml-1. Whole genome sequences (WGS) of four nisin susceptible (NS; growth <200 IU ml-1) and two nisin resistant L. monocytogenes field isolates (NR; growth >1500 IU ml-1) of serotype IIa were analyzed for DNA sequence variants (DSVs) in genes putatively associated with NR and its regulation. WGS of NR differed from NS in the gadD2 gene encoding for the glutamate decarboxylase system (GAD). Moreover, homology modeling predicted a protein structure of GadD2 in NR that promoted a less pH dependent GAD activity and may therefore be beneficial for nisin resistance. Likewise NR had a significant faster growth rate compared to NS in presence of nisin at pH 7. In conclusion, results contributed to ongoing debate that a genetic shift in GAD supports NR state.

Application of an interpretable classification model on Early Folding Residues during protein folding.

BACKGROUND: Machine learning strategies are prominent tools for data analysis. Especially in life sciences, they have become increasingly important to handle the growing datasets collected by the scientific community. Meanwhile, algorithms improve in performance, but also gain complexity, and tend to neglect interpretability and comprehensiveness of the resulting models. RESULTS: Generalized Matrix Learning Vector Quantization (GMLVQ) is a supervised, prototype-based machine learning method and provides comprehensive visualization capabilities not present in other classifiers which allow for a fine-grained interpretation of the data. In contrast to commonly used machine learning strategies, GMLVQ is well-suited for imbalanced classification problems which are frequent in life sciences. We present a Weka plug-in implementing GMLVQ. The feasibility of GMLVQ is demonstrated on a dataset of Early Folding Residues (EFR) that have been shown to initiate and guide the protein folding process. Using 27 features, an area under the receiver operating characteristic of 76.6% was achieved which is comparable to other state-of-the-art classifiers. The obtained model is accessible at https://biosciences.hs-mittweida.de/efpred/. CONCLUSIONS: The application on EFR prediction demonstrates how an easy interpretation of classification models can promote the comprehension of biological mechanisms. The results shed light on the special features of EFR which were reported as most influential for the classification: EFR are embedded in ordered secondary structure elements and they participate in networks of hydrophobic residues. Visualization capabilities of GMLVQ are presented as we demonstrate how to interpret the results.

Backbone Brackets and Arginine Tweezers delineate Class I and Class II aminoacyl tRNA synthetases.

The origin of the machinery that realizes protein biosynthesis in all organisms is still unclear. One key component of this machinery are aminoacyl tRNA synthetases (aaRS), which ligate tRNAs to amino acids while consuming ATP. Sequence analyses revealed that these enzymes can be divided into two complementary classes. Both classes differ significantly on a sequence and structural level, feature different reaction mechanisms, and occur in diverse oligomerization states. The one unifying aspect of both classes is their function of binding ATP. We identified Backbone Brackets and Arginine Tweezers as most compact ATP binding motifs characteristic for each Class. Geometric analysis shows a structural rearrangement of the Backbone Brackets upon ATP binding, indicating a general mechanism of all Class I structures. Regarding the origin of aaRS, the Rodin-Ohno hypothesis states that the peculiar nature of the two aaRS classes is the result of their primordial forms, called Protozymes, being encoded on opposite strands of the same gene. Backbone Brackets and Arginine Tweezers were traced back to the proposed Protozymes and their more efficient successors, the Urzymes. Both structural motifs can be observed as pairs of residues in contemporary structures and it seems that the time of their addition, indicated by their placement in the ancient aaRS, coincides with the evolutionary trace of Proto- and Urzymes.

Simulation of diffusion using a modular cell dynamic simulation system.

A variety of mathematical models is used to describe and simulate the multitude of natural processes examined in life sciences. In this paper we present a scalable and adjustable foundation for the simulation of natural systems. Based on neighborhood relations in graphs and the complex interactions in cellular automata, the model uses recurrence relations to simulate changes on a mesoscopic scale. This implicit definition allows for the manipulation of every aspect of the model even during simulation. The definition of value rules ω facilitates the accumulation of change during time steps. Those changes may result from different physical, chemical or biological phenomena. Value rules can be combined into modules, which in turn can be used to create baseline models. Exemplarily, a value rule for the diffusion of chemical substances was designed and its applicability is demonstrated. Finally, the stability and accuracy of the solutions is analyzed.