Everything AlphaFold tells us about protein knots.

Recent advances in Machine Learning methods in structural biology opened up new perspectives for protein analysis. Utilizing these methods allows us to go beyond the limitations of empirical research, and take advantage of the vast amount of generated data. We use a complete set of potentially knotted protein models identified in all high-quality predictions from the AlphaFold Database to search for any common trends that describe them. We show that the vast majority of knotted proteins have 31 knot and that the presence of knots is preferred in neither Bacteria, Eukaryota, or Archaea domains. On the contrary, the percentage of knotted proteins in any given proteome is around 0.4%, regardless of the taxonomical group. We also verified that the organism's living conditions do not impact the number of knotted proteins in its proteome, as previously expected. We did not encounter an organism without a single knotted protein. What is more, we found four universally present families of knotted proteins in Bacteria, consisting of SAM synthase, and TrmD, TrmH, and RsmE methyltransferases.

Knotted artifacts in predicted 3D RNA structures.

Unlike proteins, RNAs deposited in the Protein Data Bank do not contain topological knots. Recently, admittedly, the first trefoil knot and some lasso-type conformations have been found in experimental RNA structures, but these are still exceptional cases. Meanwhile, algorithms predicting 3D RNA models have happened to form knotted structures not so rarely. Interestingly, machine learning-based predictors seem to be more prone to generate knotted RNA folds than traditional methods. A similar situation is observed for the entanglements of structural elements. In this paper, we analyze all models submitted to the CASP15 competition in the 3D RNA structure prediction category. We show what types of topological knots and structure element entanglements appear in the submitted models and highlight what methods are behind the generation of such conformations. We also study the structural aspect of susceptibility to entanglement. We suggest that predictors take care of an evaluation of RNA models to avoid publishing structures with artifacts, such as unusual entanglements, that result from hallucinations of predictive algorithms.

AlphaKnot 2.0: a web server for the visualization of proteins' knotting and a database of knotted AlphaFold-predicted models.

The availability of 3D protein models is rapidly increasing with the development of structure prediction algorithms. With the expanding availability of data, new ways of analysis, especially topological analysis, of those predictions are becoming necessary. Here, we present the updated version of the AlphaKnot service that provides a straightforward way of analyzing structure topology. It was designed specifically to determine knot types of the predicted structure models, however, it can be used for all structures, including the ones solved experimentally. AlphaKnot 2.0 provides the user's ability to obtain the knowledge necessary to assess the topological correctness of the model. Both probabilistic and deterministic knot detection methods are available, together with various visualizations (including a trajectory of simplification steps to highlight the topological complexities). Moreover, the web server provides a list of proteins similar to the queried model within AlphaKnot's database and returns their knot types for direct comparison. We pre-calculated the topology of high-quality models from the AlphaFold Database (4th version) and there are now more than 680.000 knotted models available in the AlphaKnot database. AlphaKnot 2.0 is available at https://alphaknot.cent.uw.edu.pl/.

Knot or not? Identifying unknotted proteins in knotted families with sequence-based Machine Learning model.

Knotted proteins, although scarce, are crucial structural components of certain protein families, and their roles continue to be a topic of intense research. Capitalizing on the vast collection of protein structure predictions offered by AlphaFold (AF), this study computationally examines the entire UniProt database to create a robust dataset of knotted and unknotted proteins. Utilizing this dataset, we develop a machine learning (ML) model capable of accurately predicting the presence of knots in protein structures solely from their amino acid sequences. We tested the model's capabilities on 100 proteins whose structures had not yet been predicted by AF and found agreement with our local prediction in 92% cases. From the point of view of structural biology, we found that all potentially knotted proteins predicted by AF can be classified only into 17 families. This allows us to discover the presence of unknotted proteins in families with a highly conserved knot. We found only three new protein families: UCH, DUF4253, and DUF2254, that contain both knotted and unknotted proteins, and demonstrate that deletions within the knot core could potentially account for the observed unknotted (trivial) topology. Finally, we have shown that in the majority of knotted families (11 out of 15), the knotted topology is strictly conserved in functional proteins with very low sequence similarity. We have conclusively demonstrated that proteins AF predicts as unknotted are structurally accurate in their unknotted configurations. However, these proteins often represent nonfunctional fragments, lacking significant portions of the knot core (amino acid sequence).

Discovery of a trefoil knot in the RydC RNA: Challenging previous notions of RNA topology.

Knots are very common in polymers, including DNA and protein molecules. Yet, no genuine knot has been identified in natural RNA molecules to date. Upon re-examining experimentally determined RNA 3D structures, we discovered a trefoil knot 31, the most basic non-trivial knot, in the RydC RNA. This knotted RNA is a member of a small family of short bacterial RNAs, whose secondary structure is characterized by an H-type pseudoknot. Molecular dynamics simulations suggest a folding pathway of the RydC RNA that starts with a native twisted loop. Based on sequence analyses and computational RNA 3D structure predictions, we postulate that this trefoil knot is a conserved feature of all RydC-related RNAs. The first discovery of a knot in a natural RNA molecule introduces a novel perspective on RNA 3D structure formation and on fundamental research on the relationship between function and spatial structure of biopolymers.

Conservation of knotted and slipknotted topology in transmembrane transporters.

The existence of nontrivial topology is well accepted in globular proteins but not in membrane proteins. Our comprehensive topological analysis of the Protein Data Bank structures reveals 18 families of transmembrane proteins with nontrivial topology, showing that they constitute a significant number of membrane proteins. Moreover, we found that they comprise one of the largest groups of secondary active transporters. We classified them based on their knotted fingerprint into four groups: three slipknotted and one knotted. Unexpectedly, we found that the same protein can possess two distinct slipknot motifs that correspond to its outward- and inward-open conformational state. Based on the analysis of structures and knotted fingerprints, we show that slipknot topology is directly involved in the conformational transition and substrate transfer. Therefore, entanglement can be used to classify proteins and to find their structure-function relationship. Furthermore, based on the topological analysis of the transmembrane protein structures predicted by AlphaFold, we identified new potentially slipknotted protein families.

Nucleolar Essential Protein 1 (Nep1): Elucidation of enzymatic catalysis mechanism by molecular dynamics simulation and quantum mechanics study.

The Nep1 protein is essential for the formation of eukaryotic and archaeal small ribosomal subunits, and it catalyzes the site-directed SAM-dependent methylation of pseudouridine (Ψ) during pre-rRNA processing. It possesses a non-trivial topology, namely, a 31 knot in the active site. Here, we address the issue of seemingly unfeasible deprotonation of Ψ in Nep1 active site by a distant aspartate residue (D101 in S. cerevisiae), using a combination of bioinformatics, computational, and experimental methods. We identified a conserved hydroxyl-containing amino acid (S233 in S. cerevisiae, T198 in A. fulgidus) that may act as a proton-transfer mediator. Molecular dynamics simulations, based on the crystal structure of S. cerevisiae, and on a complex generated by molecular docking in A. fulgidus, confirmed that this amino acid can shuttle protons, however, a water molecule in the active site may also serve this role. Quantum-chemical calculations based on density functional theory and the cluster approach showed that the water-mediated pathway is the most favorable for catalysis. Experimental kinetic and mutational studies reinforce the requirement for the aspartate D101, but not S233. These findings provide insight into the catalytic mechanisms underlying proton transfer over extended distances and comprehensively elucidate the mode of action of Nep1.

AlphaFold predicts novel human proteins with knots.

The fact that proteins can have their chain formed in a knot is known for almost 30 years. However, as they are not common, only a fraction of such proteins is available in the Protein Data Bank. It was not possible to assess their importance and versatility up until now because we did not have access to the whole proteome of an organism, let alone a human one. The arrival of efficient machine learning methods for protein structure prediction, such as AlphaFold and RoseTTaFold, changed that. We analyzed all proteins from the human proteome (over 20,000) determined with AlphaFold in search for knots and found them in less than 2% of the structures. Using a variety of methods, including homolog search, clustering, quality assessment, and visual inspection, we determined the nature of each of the knotted structures and classified it as either knotted, potentially knotted, or an artifact, and deposited all of them in a database available at: https://knotprot.cent.uw.edu.pl/alphafold. Overall, we found 51 credible knotted proteins (0.2% of human proteome). The set of potentially knotted structures includes a new complex type of a knot not reported in proteins yet. That knot type, denoted 63 in mathematical notation, would necessitate a more complex folding path than any knotted protein characterized to date.

Identification of Novel CB2 Ligands through Virtual Screening and In Vitro Evaluation.

Cannabinoid receptor type 2 (CB2) is a very promising therapeutic target for a variety of potential indications. However, despite the existence of multiple high affinity CB2 ligands, none have yet been approved as a drug. Therefore, it would be beneficial to explore new chemotypes of CB2 ligands. The recent elucidation of CB2 tertiary structure allows for rational hit identification with structure-based (SB) methods. In this study, we established a virtual screening workflow based on SB techniques augmented with ligand-based ones, including molecular docking, MM-GBSA binding energy calculations, pharmacophore screening, and QSAR. We screened nearly 7 million drug-like, commercially available compounds. We selected 16 molecules for in vitro evaluation and identified two novel, selective CB2 antagonists with Ki values of 65 and 210 nM. Both compounds are structurally diverse from CB2 ligands known to date. The established virtual screening protocol may prove useful for hit identification for CB2 and similar molecular targets. The two novel CB2 ligands provide a desired starting point for future optimization and development of potential drugs.

Are there double knots in proteins? Prediction and in vitro verification based on TrmD-Tm1570 fusion from C. nitroreducens.

We have been aware of the existence of knotted proteins for over 30 years-but it is hard to predict what is the most complicated knot that can be formed in proteins. Here, we show new and the most complex knotted topologies recorded to date-double trefoil knots (31 #31). We found five domain arrangements (architectures) that result in a doubly knotted structure in almost a thousand proteins. The double knot topology is found in knotted membrane proteins from the CaCA family, that function as ion transporters, in the group of carbonic anhydrases that catalyze the hydration of carbon dioxide, and in the proteins from the SPOUT superfamily that gathers 31 knotted methyltransferases with the active site-forming knot. For each family, we predict the presence of a double knot using AlphaFold and RoseTTaFold structure prediction. In the case of the TrmD-Tm1570 protein, which is a member of SPOUT superfamily, we show that it folds in vitro and is biologically active. Our results show that this protein forms a homodimeric structure and retains the ability to modify tRNA, which is the function of the single-domain TrmD protein. However, how the protein folds and is degraded remains unknown.

Amino acid variants of SARS-CoV-2 papain-like protease have impact on drug binding.

The novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused both a health and economic crisis around the world. Its papain-like protease (PLpro) is one of the protein targets utilized in designing new drugs that would aid vaccines in the fight against the virus. Although there are already several potential candidates for a good inhibitor of this protein, the degree of variability of the protein itself is not taken into account. As an RNA virus, SARS-CoV-2 can mutate to a high degree, but PLpro variability has not been studied to date. Based on sequence data available in databases, we analyzed the mutational potential of this protein. We focused on the effect of observed mutations on inhibitors' binding mode and their efficacy as well as protein's activity. Our analysis identifies five mutations that should be monitored and included in the drug design process: P247S, E263D-Y264H and T265A-Y268C.

Identification of CB1 Ligands among Drugs, Phytochemicals and Natural-Like Compounds: Virtual Screening and In Vitro Verification.

Cannabinoid receptor type 1 (CB1) is an important modulator of many key physiological functions and thus a compelling molecular target. However, safe CB1 targeting is a non-trivial task. In recent years, there has been a surge of data indicating that drugs successfully used in the clinic for years (e.g. paracetamol) show CB1 activity. Moreover, there is a lot of promise in finding CB1 ligands in plants other than Cannabis sativa. In this study, we searched for possible CB1 activity among already existing drugs, their metabolites, phytochemicals, and natural-like molecules. We conducted two iterations of virtual screening, verifying the results with in vitro binding and functional assays. The in silico procedure consisted of a wide range of structure- and ligand-based methods, including docking, molecular dynamics, and quantitative structure-activity relationship (QSAR). As a result, we identified travoprost and ginkgetin as CB1 ligands, which provides a starting point for future research on the impact of their metabolites or preparations on the endocannabinoid system. Moreover, we found five natural-like compounds with submicromolar or low micromolar affinity to CB1, including one mixed partial agonist/antagonist viable for hit-to-lead phase. Finally, the computational procedure established in this work will be of use for future screening campaigns for novel CB1 ligands.

AlphaKnot: server to analyze entanglement in structures predicted by AlphaFold methods.

AlphaKnot is a server that measures entanglement in AlphaFold-solved protein models while considering pLDDT confidence values. AlphaKnot has two main functions: (i) providing researchers with a webserver for analyzing knotting in their own AlphaFold predictions and (ii) providing a database of knotting in AlphaFold predictions from the 21 proteomes for which models have been published prior to 2022. The knotting is defined in a probabilistic fashion. The knotting complexity of proteins is presented in the form of a matrix diagram which shows users the knot type for the entire polypeptide chain and for each of its subchains. The dominant knot types as well as the computed locations of the knot cores (i.e. minimal portions of protein backbones that form a given knot type) are shown for each protein structure. Based mainly on the pLDDT confidence values, entanglements are classified as Knots, Unsure, and Artifacts. The database portion of the server can be used, for example, to examine protein geometry and entanglement-function correlations, as a reference set for protein modeling, and for facilitating evolutional studies. The AlphaKnot server can be found at https://alphaknot.cent.uw.edu.pl/.

Slipknotted and unknotted monovalent cation-proton antiporters evolved from a common ancestor.

While the slipknot topology in proteins has been known for over a decade, its evolutionary origin is still a mystery. We have identified a previously overlooked slipknot motif in a family of two-domain membrane transporters. Moreover, we found that these proteins are homologous to several families of unknotted membrane proteins. This allows us to directly investigate the evolution of the slipknot motif. Based on our comprehensive analysis of 17 distantly related protein families, we have found that slipknotted and unknotted proteins share a common structural motif. Furthermore, this motif is conserved on the sequential level as well. Our results suggest that, regardless of topology, the proteins we studied evolved from a common unknotted ancestor single domain protein. Our phylogenetic analysis suggests the presence of at least seven parallel evolutionary scenarios that led to the current diversity of proteins in question. The tools we have developed in the process can now be used to investigate the evolution of other repeated-domain proteins.

SARS-CoV-2 Papain-Like Protease Potential Inhibitors-In Silico Quantitative Assessment.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) encodes the papain-like protease (PLpro). The protein not only plays an essential role in viral replication but also cleaves ubiquitin and ubiquitin-like interferon-stimulated gene 15 protein (ISG15) from host proteins, making it an important target for developing new antiviral drugs. In this study, we searched for novel, noncovalent potential PLpro inhibitors by employing a multistep in silico screening of a 15 million compound library. The selectivity of the best-scored compounds was evaluated by checking their binding affinity to the human ubiquitin carboxy-terminal hydrolase L1 (UCH-L1), which, as a deubiquitylating enzyme, exhibits structural and functional similarities to the PLpro. As a result, we identified 387 potential, selective PLpro inhibitors, from which we retrieved the 20 best compounds according to their IC50 values toward PLpro estimated by a multiple linear regression model. The selected candidates display potential activity against the protein with IC50 values in the nanomolar range from approximately 159 to 505 nM and mostly adopt a similar binding mode to the known, noncovalent SARS-CoV-2 PLpro inhibitors. We further propose the six most promising compounds for future in vitro evaluation. The results for the top potential PLpro inhibitors are deposited in the database prepared to facilitate research on anti-SARS-CoV-2 drugs.

GLN: a method to reveal unique properties of lasso type topology in proteins.

Geometry and topology are the main factors that determine the functional properties of proteins. In this work, we show how to use the Gauss linking integral (GLN) in the form of a matrix diagram-for a pair of a loop and a tail-to study both the geometry and topology of proteins with closed loops e.g. lassos. We show that the GLN method is a significantly faster technique to detect entanglement in lasso proteins in comparison with other methods. Based on the GLN technique, we conduct comprehensive analysis of all proteins deposited in the PDB and compare it to the statistical properties of the polymers. We show how high and low GLN values correlate with the internal exibility of proteins, and how the GLN in the form of a matrix diagram can be used to study folding and unfolding routes. Finally, we discuss how the GLN method can be applied to study entanglement between two structures none of which are closed loops. Since this approach is much faster than other linking invariants, the next step will be evaluation of lassos in much longer molecules such as RNA or loops in a single chromosome.

Mg2+-Dependent Methyl Transfer by a Knotted Protein: A Molecular Dynamics Simulation and Quantum Mechanics Study.

Mg2+ is required for the catalytic activity of TrmD, a bacteria-specific methyltransferase that is made up of a protein topological knot-fold, to synthesize methylated m1G37-tRNA to support life. However, neither the location of Mg2+ in the structure of TrmD nor its role in the catalytic mechanism is known. Using molecular dynamics (MD) simulations, we identify a plausible Mg2+ binding pocket within the active site of the enzyme, wherein the ion is coordinated by two aspartates and a glutamate. In this position, Mg2+ additionally interacts with the carboxylate of a methyl donor cofactor S-adenosylmethionine (SAM). The computational results are validated by experimental mutation studies, which demonstrate the importance of the Mg2+-binding residues for the catalytic activity. The presence of Mg2+ in the binding pocket induces SAM to adopt a unique bent shape required for the methyl transfer activity and causes a structural reorganization of the active site. Quantum mechanical calculations show that the methyl transfer is energetically feasible only when Mg2+ is bound in the position revealed by the MD simulations, demonstrating that its function is to align the active site residues within the topological knot-fold in a geometry optimal for catalysis. The obtained insights provide the opportunity for developing a strategy of antibacterial drug discovery based on targeting of Mg2+-binding to TrmD.

Restriction of S-adenosylmethionine conformational freedom by knotted protein binding sites.

S-adenosylmethionine (SAM) is one of the most important enzyme substrates. It is vital for the function of various proteins, including large group of methyltransferases (MTs). Intriguingly, some bacterial and eukaryotic MTs, while catalysing the same reaction, possess significantly different topologies, with the former being a knotted one. Here, we conducted a comprehensive analysis of SAM conformational space and factors that affect its vastness. We investigated SAM in two forms: free in water (via NMR studies and explicit solvent simulations) and bound to proteins (based on all data available in the PDB and on all-atom molecular dynamics simulations in water). We identified structural descriptors-angles which show the major differences in SAM conformation between unknotted and knotted methyltransferases. Moreover, we report that this is caused mainly by a characteristic for knotted MTs compact binding site formed by the knot and the presence of adenine-binding loop. Additionally, we elucidate conformational restrictions imposed on SAM molecules by other protein groups in comparison to conformational space in water.

Genus for biomolecules.

The 'Genus for biomolecules' database (http://genus.fuw.edu.pl) collects information about topological structure and complexity of proteins and RNA chains, which is captured by the genus of a given chain and its subchains. For each biomolecule, this information is shown in the form of a genus trace plot, as well as a genus matrix diagram. We assemble such information for all and RNA structures deposited in the Protein Data Bank (PDB). This database presents also various statistics and extensive information about the biological function of the analyzed biomolecules. The database is regularly self-updating, once new structures are deposited in the PDB. Moreover, users can analyze their own structures.

Knot_pull-python package for biopolymer smoothing and knot detection.

SUMMARY: The biggest hurdle in studying topology in biopolymers is the steep learning curve for actually seeing the knots in structure visualization. Knot_pull is a command line utility designed to simplify this process-it presents the user with a smoothing trajectory for provided structures (any number and length of protein, RNA or chromatin chains in PDB, CIF or XYZ format), and calculates the knot type (including presence of any links, and slipknots when a subchain is specified). AVAILABILITY AND IMPLEMENTATION: Knot_pull works under Python >=2.7 and is system independent. Source code and documentation are available at http://github.com/dzarmola/knot_pull under GNU GPL license and include also a wrapper script for PyMOL for easier visualization. Examples of smoothing trajectories can be found at: https://www.youtube.com/watch?v=IzSGDfc1vAY. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.