Existing computational methods for estimating pK_a values in proteins rely on theoretical approximations and lengthy computations. In this work, we use a data set of 6 million theoretically determined pK_a shifts to train deep learning models, which are shown to rival the physics-based predictors. These neural networks managed to infer the electrostatic contributions of different chemical groups and learned the importance of solvent exposure and close interactions, including hydrogen bonds. Although trained only using theoretical data, our pKAI+ model displayed the best accuracy in a test set of ∼750 experimental values. Inference times allow speedups of more than 1000× compared to physics-based methods. By combining speed, accuracy, and a reasonable understanding of the underlying physics, our models provide a game-changing solution for fast estimations of macroscopic pK_a values from ensembles of microscopic values as well as for many downstream applications such as molecular docking and constant-pH molecular dynamics simulations.

Centaurium erythraea is recommended for the treatment of gastrointestinal disorders and to reduce hypercholesterolemia in ethno-medicinal practice. To perform a top-down study that could give some insight into the molecular basis of these bioactivities, decoctions from C. erythraea leaves were prepared and the compounds were identified by liquid chromatography-high resolution tandem mass spectrometry (LC–MS/MS). Secoiridoids glycosides, like gentiopicroside and sweroside, and several xanthones, such as di-hydroxy-dimethoxyxanthone, were identified. Following some of the bioactivities previously ascribed to C. erythraea, we have studied its antioxidant capacity and the ability to inhibit acetylcholinesterase (AChE) and 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGR). Significant antioxidant activities were observed, following three assays: free radical 2,2-diphenyl-1-picrylhydrazyl (DPPH) reduction; lipoperoxidation; and NO radical scavenging capacity. The AChE and HMGR inhibitory activities for the decoction were also measured (56% at 500 μg/mL and 48% at 10 μg/mL, respectively). Molecular docking studies indicated that xanthones are better AChE inhibitors than gentiopicroside, while this compound exhibits a better shape complementarity with the HMGR active site than xanthones. To the extent of our knowledge, this is the first report on AChE and HMGR activities by C. erythraea decoctions, in a top-down analysis, complemented with in silico molecular docking, which aims to understand, at the molecular level, some of the biological effects ascribed to infusions from this plant.

Electrostatic interactions play a pivotal role in the structure and mechanism of action of most biomolecules. There are several conceptually different methods to deal with electrostatics in molecular dynamics simulations. Ionic strength effects are usually introduced using such methodologies and can have a significant impact on the quality of the final conformation space obtained. We have previously shown that full system neutralization can lead to wrong lipidic phases in the 25% PA/PC bilayer (J. Chem. Theory Comput. 2014, 10, 5483–5492). In this work, we investigate how two limit approaches to the ionic strength treatment (implicitly with GRF or using full system neutralization with either GRF or PME) can influence the conformational space of the second-generation PAMAM dendrimer. Constant-pH MD simulations were used to map PAMAM’s conformational space at its full pH range (from 2.5 to 12.5). Our simulations clearly captured the coupling between protonation and conformation in PAMAM. Interestingly, the dendrimer conformational distribution was almost independent of the ionic strength treatment methods, which is in contrast to what we have observed in charged lipid bilayers. Overall, our results confirm that both GRF with implicit ionic strength and a fully neutralized system with PME are valid approaches to model charged globular systems, using the GROMOS 54A7 force field.

PK_a values of ionizable residues and isoelectric points of proteins provide valuable local and global insights about their structure and function. These properties can be estimated with reasonably good accuracy using Poisson–Boltzmann and Monte Carlo calculations at a considerable computational cost (from some minutes to several hours). pKPDB is a database of over 12 M theoretical pK_a values calculated over 120k protein structures deposited in the Protein Data Bank. By providing precomputed pK_a and pI values, users can retrieve results instantaneously for their protein(s) of interest while also saving countless hours and resources that would be spent on repeated calculations. Furthermore, there is an ever-growing imbalance between experimental pK_a and pI values and the number of resolved structures. This database will complement the experimental and computational data already available and can also provide crucial information regarding buried residues that are under-represented in experimental measurements. Gzipped csv files containing p K_a and isoelectric point values can be downloaded from https://pypka.org/pKPDB. To query a single PDB code please use the PypKa free server at https://pypka.org. The pKPDB source code can be found at https://github.com/mms-fcul/pKPDB. Supplementary data are available at Bioinformatics online.