Coarse-grained (CG) modelling with the Martini force field has come of age. By combining a variety of bead types and sizes with a new mapping approach, the newest version of the model is able to accurately simulate large biomolecular complexes at millisecond timescales. In this perspective, we discuss possible applications of the Martini 3 model in drug discovery and development pipelines and highlight areas for future development. Owing to its high simulation efficiency and extended chemical space, Martini 3 has great potential in the area of drug design and delivery. However, several aspects of the model should be improved before Martini 3 CG simulations can be routinely employed in academic and industrial settings. These include the development of automatic parameterisation protocols for a variety of molecule types, the improvement of backmapping procedures, the description of protein flexibility and the development of methodologies enabling efficient sampling. We illustrate our view with examples on key areas where Martini could give important contributions such as drugs targeting membrane proteins, cryptic pockets and protein–protein interactions and the development of soft drug delivery systems.

Mist is generated by ultrasonic cavitation of water (Fisher Biograde, pH 5.5–6.5) at room temperature (20–25 °C) in open air with nearly constant temperature (22–25 °C) but varying relative humidity (RH; 24–52%) over the course of many months. Water droplets in the mist are initially about 7 μm in diameter at about 50% RH. They are collected, and the concentration of hydrogen peroxide (H2O2) is measured using commercial peroxide test strips and by bromothymol blue oxidation. The quantification method is based on the Fenton chemistry of dye degradation to determine the oxidation capacity of water samples that have been treated by ultrasonication. It is found that the hydrogen peroxide concentration varies nearly linearly with RH over the range studied, reaching a low of 2 parts per million (ppm) at 24% RH and a high of 6 ppm at 52% RH. Some possible public health implications concerning the transmission of respiratory viral infections are suggested for this threefold change in H2O2 concentration with RH.

Peptides mediate up to 40% of protein interactions, their high specificity and ability to bind in places where small molecules cannot make them potential drug candidates. However, predicting peptide–protein complexes remains more challenging than protein–protein or protein–small molecule interactions, in part due to the high flexibility peptides have. In this review, we look at the advances in docking, molecular simulations and machine learning to tackle problems related to peptides such as predicting structures, binding affinities or even kinetics. We specifically focus on explaining the number of docking programmes and force fields used in molecular simulations, so a prospective user can have an educated guess as to why choose one modelling tool or another to address their scientific questions.

The convergence of free-energy calculations based on importance sampling depends heavily on the choice of collective variables (CVs), which in principle, should include the slow degrees of freedom of the biological processes to be investigated. Autoencoders (AEs), as emerging data-driven dimension reduction tools, have been utilised for discovering CVs. AEs, however, are often treated as black boxes, and what AEs actually encode during training, and whether the latent variables from encoders are suitable as CVs for further free-energy calculations remains unknown. In this contribution, we review AEs and their time-series-based variants, including time-lagged AEs (TAEs) and modified TAEs, as well as the closely related model variational approach for Markov processes networks (VAMPnets). We then show through numerical examples that AEs learn the high-variance modes instead of the slow modes. In stark contrast, time series-based models are able to capture the slow modes. Moreover, both modified TAEs with extensions from slow feature analysis and the state-free reversible VAMPnets (SRVs) can yield orthogonal multidimensional CVs. As an illustration, we employ SRVs to discover the CVs of the isomerizations of N-acetyl-N′-methylalanylamide and trialanine by iterative learning with trajectories from biased simulations. Last, through numerical experiments with anisotropic diffusion, we investigate the potential relationship of time-series-based models and committor probabilities.

The chemical potential of water ($ {\mu}_{{\mathrm{H}}_2\mathrm{O}} $) provides an essential thermodynamic characterization of the environment of living organisms, and it is of equal significance as the temperature. For cells, $ {\mu}_{{\mathrm{H}}_2\mathrm{O}} $ is conventionally expressed in terms of the osmotic pressure (πosm). We have previously suggested that the main contribution to the intracellular πosm of the bacterium E. coli is from soluble negatively-charged proteins and their counter-ions. Here, we expand on this analysis by examining how evolutionary divergent cell types cope with the challenge of maintaining πosm within viable values. Complex organisms, like mammals, maintain constant internal πosm ≈ 0.285 osmol, matching that of 0.154 M NaCl. For bacteria it appears that optimal growth conditions are found for similar or slightly higher πosm (0.25-0.4 osmol), despite that they represent a much earlier stage in evolution. We argue that this value reflects a general adaptation for optimising metabolic function under crowded intracellular conditions. Environmental πosm that differ from this optimum require therefore special measures, as exemplified with gram-positive and gram-negative bacteria. To handle such situations, their membrane encapsulations allow for a compensating turgor pressure that can take both positive and negative values, where positive pressures allow increased frequency of metabolic events through increased intracellular protein concentrations. A remarkable exception to the rule of 0.25-0.4 osmol, is found for halophilic archaea with internal πosm ≈ 15 osmol. The internal organization of these archaea differs in that they utilize a repulsive electrostatic mechanism operating only in the ionic-liquid regime to avoid aggregation, and that they stand out from other organisms by having no turgor pressure.

We report the G protein-first mechanism for activation of G protein-coupled receptors (GPCR) for the three closest subtypes of the opioid receptors (OR), μOR, κOR and δOR. We find that they couple to the inactive Gi protein-bound guanosine diphosphate (GDP) prior to agonist binding. The inactive Gi protein forms anchors to the intracellular loops of the inactive apo-μOR, apo-κOR and apo-δOR, inducing opening of the cytoplasmic region to form a pre-activated state that holds Gi protein in place until agonist binds. Then, agonist binds to μOR, κOR and δOR already complexed with Gi protein, to trigger the Gαi to open up the tightly coupled GDP binding site, making GDP accessible for GTP exchange, an essential step for Gi signalling. We show that the agonist alone cannot open the intracellular region of μOR and κOR, requiring Gi protein to open the cytoplasmic region by itself. We consider that this G protein-first mechanism may apply to activation of other Class A GPCRs. However, for δOR, agonist binding can open up the intracellular region to encourage Gi protein recruitment. Thus, activation of Gi protein mediated by δOR favourably may proceed with either ligand-first or G protein-first activation mechanisms.

Chaperones protect other proteins against misfolding and aggregation, a key requirement for maintaining biological function. Experimental observations of changes in solubility of amyloid proteins in the presence of certain chaperones are discussed here in terms of thermodynamic driving forces. We outline how chaperones can enhance amyloid solubility through the formation of heteromolecular aggregates (co-aggregates) based on the second law of thermodynamics and the flux towards equal chemical potential of each compound in all phases of the system. Higher effective solubility of an amyloid peptide in the presence of chaperone implies that the chemical potential of the peptide is higher in the aggregates formed under these conditions compared to peptide-only aggregates. This must be compensated by a larger reduction in chemical potential of the chaperone in the presence of peptide compared to chaperone alone. The driving force thus relies on the chaperone being very unhappy on its own (high chemical potential), thus gaining more free energy than the amyloid peptide loses upon forming the co-aggregate. The formation of heteromolecular aggregates also involves the kinetic suppression of the formation of homomolecular aggregates. The unhappiness of the chaperone can explain the ability of chaperones to favour an increased population of monomeric client protein even in the absence of external energy input, and with broad client specificity. This perspective opens for a new direction of chaperone research and outlines a set of outstanding questions that aim to provide additional cues for therapeutic development in this area.

Membrane transporters including glucose transporters (GLUTs) are involved in cellular energy supplies, cell metabolism and other vital biological activities. They have also been implicated in cancer proliferation and metastasis, thus they represent an important target in combatting cancer. However, membrane transporters are very difficult to study due to their multispan transmembrane properties. The new computational tool, AlphaFold2, offers highly accurate predictions of three-dimensional protein structures. The glutamine, threonine and tyrosine (QTY) code provides a systematic method of rendering hydrophobic sequences into hydrophilic ones. Here, we present computational studies of native integral membrane GLUTs with 12 transmembrane helical segments determined by X-ray crystallography and CryoEM, comparing the AlphaFold2-predicted native structure to their water-soluble QTY variants predicted by AlphaFold2. In the native structures of the transmembrane helices, there are hydrophobic amino acids leucine (L), isoleucine (I), valine (V) and phenylalanine (F). Applying the QTY code, these hydrophobic amino acids are systematically replaced by hydrophilic amino acids, glutamine (Q), threonine (T) and tyrosine (Y) rendering them water-soluble. We present the superposed structures of native GLUTs and their water-soluble QTY variants. The superposed structures show remarkable similar residue mean square distance values between 0.47 and 3.6 Å (most about 1–2 Å) despite >44% transmembrane amino acid differences. We also show the differences of hydrophobicity patches between the native membrane transporters and their QTY variants. We explain the rationale why the membrane protein QTY variants become water-soluble. Our study provides insight into the differences between the hydrophobic helices and hydrophilic helices, and offers confirmation of the QTY method for studying multispan transmembrane proteins and other aggregated proteins through their water-soluble variants.

The scientific and technological advent of Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) is one of the most exciting developments of the past decade, particularly in the field of gene editing. The technology has two essential components, (1) a guide RNA to match a targeted gene and (2) a CRISPR-associated protein (e.g. Cas 9, Cas12 or Cas13) that acts as an endonuclease to specifically cut DNA. This specificity and reconfigurable nature of CRISPR has also spurred intense academic and commercial interest in the development of CRISPR-based molecular diagnostics. CRISPR Cas12 and Cas13 orthologs are most commonly applied to diagnostics, and these cleave and become activated by DNA and RNA targets, respectively. Despite the intense research interest, the limits of detection (LoDs) and applications of CRISP-based diagnostics remain an open question. A major reason for this is that reports of kinetic rates have been widely inconsistent, and the vast majority of these reports contain gross errors including violations of basic conservation and kinetic rate laws. It is the intent of this Perspective to bring attention to these issues and to identify potential improvements in the manner in which CRISPR kinetic rates and assay LoDs are reported and compared. The CRISPR field would benefit from verifications of self-consistency of data, providing sufficient data for reproduction of experiments, and, in the case of reports of novel assay LoDs, concurrent reporting of the associated kinetic rate constants. The early development of CRISPR-based diagnostics calls for self-reflection and urges us to proceed with caution.

Macrocycles represent an important class of ligands, both in natural products and designed drugs. In drug design, macrocyclizations can impart specific ligand conformations and contribute to passive permeation by encouraging intramolecular H-bonds. AutoDock-GPU and Vina can model macrocyclic ligands flexibly, without requiring the enumeration of macrocyclic conformers before docking. Here, we characterize the performance of the method for handling macrocyclic compounds, which is implemented and the default behaviour for ligand preparation with our ligand preparation pipeline, Meeko. A pseudoatom is used to encode bond geometry and produce an anisotropic closure force for macrocyclic rings. This method is evaluated on a diverse set of small molecule and peptide macrocycles, ranging from 7- to 33-membered rings, showing little accuracy loss compared to rigid redocking of the X-ray macrocycle conformers. This suggests that for conformationally flexible macrocycles with unknown binding modes, this method can be effectively used to predict the macrocycle conformation.

Ankyrin (ANK) repeat proteins are coded by tandem occurrences of patterns with around 33 amino acids. They often mediate protein–protein interactions in a diversity of biological systems. These proteins have an elongated non-globular shape and often display complex folding mechanisms. This work investigates the energy landscape of representative proteins of this class made up of 3, 4 and 6 ANK repeats using the energy-landscape visualisation method (ELViM). By combining biased and unbiased coarse-grained molecular dynamics AWSEM simulations that sample conformations along the folding trajectories with the ELViM structure-based phase space, one finds a three-dimensional representation of the globally funnelled energy surface. In this representation, it is possible to delineate distinct folding pathways. We show that ELViMs can project, in a natural way, the intricacies of the highly dimensional energy landscapes encoded by the highly symmetric ankyrin repeat proteins into useful low-dimensional representations. These projections can discriminate between multiplicities of specific parallel folding mechanisms that otherwise can be hidden in oversimplified depictions.

When biological samples are first exposed to electrons in cryo-electron microcopy (cryo-EM), proteins exhibit a rapid ‘burst’ phase of beam-induced motion that cannot be corrected with software. This lowers the quality of the initial frames, which are the least damaged by the electrons. Hence, they are commonly excluded or down-weighted during data processing, reducing the undamaged signal and the resolution in the reconstruction. By decreasing the cooling rate during sample preparation, either with a cooling-rate gradient or by increasing the freezing temperature, we show that the quality of the initial frames for various protein and virus samples can be recovered. Incorporation of the initial frames in the reconstruction increases the resolution by an amount equivalent to using ~60% more data. Moreover, these frames preserve the high-quality cryo-EM densities of radiation-sensitive residues, which is often damaged or very weak in canonical three-dimensional reconstruction. The improved freezing conditions can be easily achieved using existing devices and enhance the overall quality of cryo-EM structures.

Magnesium ions (Mg2+) are vital for RNA structure and cellular functions. Present efforts in RNA structure determination and understanding of RNA functions are hampered by the inability to accurately locate Mg2+ ions in an RNA. Here we present a machine-learning method, originally developed for computer visual recognition, to predict Mg2+ binding sites in RNA molecules. By incorporating geometrical and electrostatic features of RNA, we capture the key ingredients of Mg2+-RNA interactions, and from deep learning, predict the Mg2+ density distribution. Five-fold cross-validation on a dataset of 177 selected Mg2+-containing structures and comparisons with different methods validate the approach. This new approach predicts Mg2+ binding sites with notably higher accuracy and efficiency. More importantly, saliency analysis for eight different Mg2+ binding motifs indicates that the model can reveal critical coordinating atoms for Mg2+ ions and ion-RNA inner/outer-sphere coordination. Furthermore, implementation of the model uncovers new Mg2+ binding motifs. This new approach may be combined with X-ray crystallography structure determination to pinpoint the metal ion binding sites.