AI revolutions in biology: The joys and perils of AlphaFold

Differential reliance of CTD-nuclear envelope phosphatase 1 on its regulatory subunit in ER lipid synthesis and storage

Lipin 1 is an ER enzyme that produces diacylglycerol, the lipid intermediate that feeds into the synthesis of glycerophospholipids for membrane expansion or triacylglycerol for storage into lipid droplets. CTD-Nuclear Envelope Phosphatase 1 (CTDNEP1) regulates lipin 1 to restrict ER membrane synthesis, but a role for CTDNEP1 in lipid storage in mammalian cells is not known. Furthermore, how NEP1R1, the regulatory subunit of CTDNEP1, contributes to these functions in mammalian cells is not fully understood. Here, we show that CTDNEP1 is reliant on NEP1R1 for its stability and function in limiting ER expansion. CTDNEP1 contains an amphipathic helix at its N-terminus that targets to the ER, nuclear envelope and lipid droplets. We identify key residues at the binding interface of CTDNEP1 and NEP1R1 and show that they facilitate complex formation in vivo and in vitro. We demonstrate that NEP1R1 binding to CTDNEP1 shields CTDNEP1 from proteasomal degradation to regulate lipin 1 and restrict ER size. Unexpectedly, NEP1R1 was not required for CTDNEP1's role in restricting lipid droplet biogenesis. Thus, the reliance of CTDNEP1 function on NEP1R1 depends on cellular demands for membrane production versus lipid storage. Together, our work provides a framework into understanding how the ER regulates lipid synthesis under different metabolic conditions.

The power and pitfalls of AlphaFold2 for structure prediction beyond rigid globular proteins

Artificial intelligence-driven advances in protein structure prediction in recent years have raised the question: has the protein structure-prediction problem been solved? Here, with a focus on nonglobular proteins, we highlight the many strengths and potential weaknesses of DeepMind's AlphaFold2 in the context of its biological and therapeutic applications. We summarize the subtleties associated with evaluation of AlphaFold2 model quality and reliability using the predicted local distance difference test (pLDDT) and predicted aligned error (PAE) values. We highlight various classes of proteins that AlphaFold2 can be applied to and the caveats involved. Concrete examples of how AlphaFold2 models can be integrated with experimental data in the form of small-angle X-ray scattering (SAXS), solution NMR, cryo-electron microscopy (cryo-EM) and X-ray diffraction are discussed. Finally, we highlight the need to move beyond structure prediction of rigid, static structural snapshots toward conformational ensembles and alternate biologically relevant states. The overarching theme is that careful consideration is due when using AlphaFold2-generated models to generate testable hypotheses and structural models, rather than treating predicted models as de facto ground truth structures.

The HADDOCK2.4 web server for integrative modeling of biomolecular complexes

Interactions between macromolecules, such as proteins and nucleic acids, are essential for cellular functions. Experimental methods can fail to provide all the information required to fully model biomolecular complexes at atomic resolution, particularly for large and heterogeneous assemblies. Integrative computational approaches have, therefore, gained popularity, complementing traditional experimental methods in structural biology. Here, we introduce HADDOCK2.4, an integrative modeling platform, and its updated web interface ( https://wenmr.science.uu.nl/haddock2.4 ). The platform seamlessly integrates diverse experimental and theoretical data to generate high-quality models of macromolecular complexes. The user-friendly web server offers automated parameter settings, access to distributed computing resources, and pre- and post-processing steps that enhance the user experience. To present the web server's various interfaces and features, we demonstrate two different applications: (i) we predict the structure of an antibody-antigen complex by using NMR data for the antigen and knowledge of the hypervariable loops for the antibody, and (ii) we perform coarse-grained modeling of PRC1 with a nucleosome particle guided by mutagenesis and functional data. The described protocols require some basic familiarity with molecular modeling and the Linux command shell. This new version of our widely used HADDOCK web server allows structural biologists and non-experts to explore intricate macromolecular assemblies encompassing various molecule types.

Insights into AlphaFold's breakthrough in neurodegenerative diseases

Neurodegenerative diseases (ND) are disorders of the central nervous system (CNS) characterized by impairment in neurons' functions, and complete loss, leading to memory loss, and difficulty in learning, language, and movement processes. The most common among these NDs are Alzheimer's disease (AD) and Parkinson's disease (PD), although several other disorders also exist. These are frontotemporal dementia (FTD), amyotrophic lateral syndrome (ALS), Huntington's disease (HD), and others; the major pathological hallmark of NDs is the proteinopathies, either of amyloid-β (Aβ), tauopathies, or synucleinopathies. Aggregation of proteins that do not undergo normal configuration, either due to mutations or through some disturbance in cellular pathway contributes to the diseases. Artificial Intelligence (AI) and deep learning (DL) have proven to be successful in the diagnosis and treatment of various congenital diseases. DL approaches like AlphaFold (AF) are a major leap towards success in CNS disorders. This 3D protein geometry modeling algorithm developed by DeepMind has the potential to revolutionize biology. AF has the potential to predict 3D-protein confirmation at an accuracy level comparable to experimentally predicted one, with the additional advantage of precisely estimating protein interactions. This breakthrough will be beneficial to identify diseases' advancement and the disturbance of signaling pathways stimulating impaired functions of proteins. Though AlphaFold has solved a major problem in structural biology, it cannot predict membrane proteins-a beneficial approach for drug designing.

Structural studies of intrinsically disordered MLL-fusion protein AF9 in complex with peptidomimetic inhibitors

AF9 (MLLT3) and its paralog ENL(MLLT1) are members of the YEATS family of proteins with important role in transcriptional and epigenetic regulatory complexes. These proteins are two common MLL fusion partners in MLL-rearranged leukemias. The oncofusion proteins MLL-AF9/ENL recruit multiple binding partners, including the histone methyltransferase DOT1L, leading to aberrant transcriptional activation and enhancing the expression of a characteristic set of genes that drive leukemogenesis. The interaction between AF9 and DOT1L is mediated by an intrinsically disordered C-terminal ANC1 homology domain (AHD) in AF9, which undergoes folding upon binding of DOT1L and other partner proteins. We have recently reported peptidomimetics that disrupt the recruitment of DOT1L by AF9 and ENL, providing a proof-of-concept for targeting AHD and assessing its druggability. Intrinsically disordered proteins, such as AF9 AHD, are difficult to study and characterize experimentally on a structural level. In this study, we present a successful protein engineering strategy to facilitate structural investigation of the intrinsically disordered AF9 AHD domain in complex with peptidomimetic inhibitors by using maltose binding protein (MBP) as a crystallization chaperone connected with linkers of varying flexibility and length. The strategic incorporation of disulfide bonds provided diffraction-quality crystals of the two disulfide-bridged MBP-AF9 AHD fusion proteins in complex with the peptidomimetics. These successfully determined first series of 2.1-2.6 Å crystal complex structures provide high-resolution insights into the interactions between AHD and its inhibitors, shedding light on the role of AHD in recruiting various binding partner proteins. We show that the overall complex structures closely resemble the reported NMR structure of AF9 AHD/DOT1L with notable difference in the conformation of the β-hairpin region, stabilized through conserved hydrogen bonds network. These first series of AF9 AHD/peptidomimetics complex structures are providing insights of the protein-inhibitor interactions and will facilitate further development of novel inhibitors targeting the AF9/ENL AHD domain.

Microenzymes: Is There Anybody Out There?

Biological macromolecules are found in different shapes and sizes. Among these, enzymes catalyze biochemical reactions and are essential in all organisms, but is there a limit size for them to function properly? Large enzymes such as catalases have hundreds of kDa and are formed by multiple subunits, whereas most enzymes are smaller, with molecular weights of 20-60 kDa. Enzymes smaller than 10 kDa could be called microenzymes and the present literature review brings together evidence of their occurrence in nature. Additionally, bioactive peptides could be a natural source for novel microenzymes hidden in larger peptides and molecular downsizing could be useful to engineer artificial enzymes with low molecular weight improving their stability and heterologous expression. An integrative approach is crucial to discover and determine the amino acid sequences of novel microenzymes, together with their genomic identification and their biochemical biological and evolutionary functions.

The S-component fold: a link between bacterial transporters and receptors

The processes of nutrient uptake and signal sensing are crucial for microbial survival and adaptation. Membrane-embedded proteins involved in these functions (transporters and receptors) are commonly regarded as unrelated in terms of sequence, structure, mechanism of action and evolutionary history. Here, we analyze the protein structural universe using recently developed artificial intelligence-based structure prediction tools, and find an unexpected link between prominent groups of microbial transporters and receptors. The so-called S-components of Energy-Coupling Factor (ECF) transporters, and the membrane domains of sensor histidine kinases of the 5TMR cluster share a structural fold. The discovery of their relatedness manifests a widespread case of prokaryotic "transceptors" (related proteins with transport or receptor function), showcases how artificial intelligence-based structure predictions reveal unchartered evolutionary connections between proteins, and provides new avenues for engineering transport and signaling functions in bacteria.

Structural basis of the Meinwald rearrangement catalysed by styrene oxide isomerase

Membrane-bound styrene oxide isomerase (SOI) catalyses the Meinwald rearrangement-a Lewis-acid-catalysed isomerization of an epoxide to a carbonyl compound-and has been used in single and cascade reactions. However, the structural information that explains its reaction mechanism has remained elusive. Here we determine cryo-electron microscopy (cryo-EM) structures of SOI bound to a single-domain antibody with and without the competitive inhibitor benzylamine, and elucidate the catalytic mechanism using electron paramagnetic resonance spectroscopy, functional assays, biophysical methods and docking experiments. We find ferric haem b bound at the subunit interface of the trimeric enzyme through H58, where Fe(III) acts as the Lewis acid by binding to the epoxide oxygen. Y103 and N64 and a hydrophobic pocket binding the oxygen of the epoxide and the aryl group, respectively, position substrates in a manner that explains the high regio-selectivity and stereo-specificity of SOI. Our findings can support extending the range of epoxide substrates and be used to potentially repurpose SOI for the catalysis of new-to-nature Fe-based chemical reactions.

AlphaFold2 modeling and molecular dynamics simulations of an intrinsically disordered protein

We use AlphaFold2 (AF2) to model the monomer and dimer structures of an intrinsically disordered protein (IDP), Nvjp-1, assisted by molecular dynamics (MD) simulations. We observe relatively rigid dimeric structures of Nvjp-1 when compared with the monomer structures. We suggest that protein conformations from multiple AF2 models and those from MD trajectories exhibit a coherent trend: the conformations of an IDP are deviated from each other and the conformations of a well-folded protein are consistent with each other. We use a residue-residue interaction network (RIN) derived from the contact map which show that the residue-residue interactions in Nvjp-1 are mainly transient; however, those in a well-folded protein are mainly persistent. Despite the variation in 3D shapes, we show that the AF2 models of both disordered and ordered proteins exhibit highly consistent profiles of the pLDDT (predicted local distance difference test) scores. These results indicate a potential protocol to justify the IDPs based on multiple AF2 models and MD simulations.

Ligand Gaussian accelerated Molecular Dynamics 3 (LiGaMD3): Improved Calculations of Binding Thermodynamics and Kinetics of Both Small Molecules and Flexible Peptides

Binding thermodynamics and kinetics play critical roles in drug design. However, it has proven challenging to efficiently predict ligand binding thermodynamics and kinetics of small molecules and flexible peptides using conventional Molecular Dynamics (cMD), due to limited simulation timescales. Based on our previously developed Ligand Gaussian accelerated Molecular Dynamics (LiGaMD) method, we present a new approach, termed "LiGaMD3", in which we introduce triple boosts into three individual energy terms that play important roles in small-molecule/peptide dissociation, rebinding and system conformational changes to improve the sampling efficiency of small-molecule/peptide interactions with target proteins. To validate the performance of LiGaMD3, MDM2 bound by a small molecule (Nutlin 3) and two highly flexible peptides (PMI and P53) were chosen as model systems. LiGaMD3 could efficiently capture repetitive small-molecule/peptide dissociation and binding events within 2 microsecond simulations. The predicted binding kinetic constant rates and free energies from LiGaMD3 agreed with available experimental values and previous simulation results. Therefore, LiGaMD3 provides a more general and efficient approach to capture dissociation and binding of both small-molecule ligand and flexible peptides, allowing for accurate prediction of their binding thermodynamics and kinetics.

An AlphaFold Structure Analysis of COQ2 as Key a Component of the Coenzyme Q Synthesis Complex

Coenzyme Q (CoQ) is a lipidic compound that is widely distributed in nature, with crucial functions in metabolism, protection against oxidative damage and ferroptosis and other processes. CoQ biosynthesis is a conserved and complex pathway involving several proteins. COQ2 is a member of the UbiA family of transmembrane prenyltransferases that catalyzes the condensation of the head and tail precursors of CoQ, which is a key step in the process, because its product is the first intermediate that will be modified in the head by the next components of the synthesis process. Mutations in this protein have been linked to primary CoQ deficiency in humans, a rare disease predominantly affecting organs with a high energy demand. The reaction catalyzed by COQ2 and its mechanism are still unknown. Here, we aimed at clarifying the COQ2 reaction by exploring possible substrate binding sites using a strategy based on homology, comprising the identification of available ligand-bound homologs with solved structures in the Protein Data Bank (PDB) and their subsequent structural superposition in the AlphaFold predicted model for COQ2. The results highlight some residues located on the central cavity or the matrix loops that may be involved in substrate interaction, some of which are mutated in primary CoQ deficiency patients. Furthermore, we analyze the structural modifications introduced by the pathogenic mutations found in humans. These findings shed new light on the understanding of COQ2's function and, thus, CoQ's biosynthesis and the pathogenicity of primary CoQ deficiency.

Inference of drug off-target effects on cellular signaling using interactome-based deep learning

Many diseases emerge from dysregulated cellular signaling, and drugs are often designed to target specific signaling proteins. Off-target effects are, however, common and may ultimately result in failed clinical trials. Here we develop a computer model of the cell's transcriptional response to drugs for improved understanding of their mechanisms of action. The model is based on ensembles of artificial neural networks and simultaneously infers drug-target interactions and their downstream effects on intracellular signaling. With this, it predicts transcription factors' activities, while recovering known drug-target interactions and inferring many new ones, which we validate with an independent dataset. As a case study, we analyze the effects of the drug Lestaurtinib on downstream signaling. Alongside its intended target, FLT3, the model predicts an inhibition of CDK2 that enhances the downregulation of the cell cycle-critical transcription factor FOXM1. Our approach can therefore enhance our understanding of drug signaling for therapeutic design.

BioNexusSentinel: a visual tool for bioregulatory network and cytohistological RNA-seq genetic expression profiling within the context of multicellular simulation research using ChatGPT-augmented software engineering

Summary

Motivated by the need to parameterize ongoing multicellular simulation research, this paper documents the culmination of a ChatGPT augmented software engineering cycle resulting in an integrated visual platform for efficient cytohistological RNA-seq and bioregulatory network exploration. As contrasted to other systems and synthetic biology tools, BioNexusSentinel was developed de novo to uniquely combine these features. Reactome served as the primary source of remotely accessible biological models, accessible using BioNexusSentinel's novel search engine and REST API requests. The innovative, feature-rich gene expression profiler component was developed to enhance the exploratory experience for the researcher, culminating in the cytohistological RNA-seq explorer based on Human Protein Atlas data. A novel cytohistological classifier would be integrated via pre-processed analysis of the RNA-seq data via R statistical language, providing for useful analytical functionality and good performance for the end-user. Implications of the work span prospects for model orthogonality evaluations, gap identification in network modelling, prototyped automatic kinetics parameterization, and downstream simulation and cellular biological state analysis. This unique computational biology software engineering collaboration with generative natural language processing artificial intelligence was shown to enhance worker productivity, with evident benefits in terms of accelerating coding and machine-human intelligence transfer.

Availability and implementation

BioNexusSentinel project releases, with corresponding data and installation instructions, are available at https://github.com/RichardMatzko/BioNexusSentinel.

Modeling Flexible Protein Structure With AlphaFold2 and Crosslinking Mass Spectrometry

We propose a pipeline that combines AlphaFold2 (AF2) and crosslinking mass spectrometry (XL-MS) to model the structure of proteins with multiple conformations. The pipeline consists of two main steps: ensemble generation using AF2 and conformer selection using XL-MS data. For conformer selection, we developed two scores-the monolink probability score (MP) and the crosslink probability score (XLP)-both of which are based on residue depth from the protein surface. We benchmarked MP and XLP on a large dataset of decoy protein structures and showed that our scores outperform previously developed scores. We then tested our methodology on three proteins having an open and closed conformation in the Protein Data Bank: Complement component 3 (C3), luciferase, and glutamine-binding periplasmic protein, first generating ensembles using AF2, which were then screened for the open and closed conformations using experimental XL-MS data. In five out of six cases, the most accurate model within the AF2 ensembles-or a conformation within 1 Å of this model-was identified using crosslinks, as assessed through the XLP score. In the remaining case, only the monolinks (assessed through the MP score) successfully identified the open conformation of glutamine-binding periplasmic protein, and these results were further improved by including the "occupancy" of the monolinks. This serves as a compelling proof-of-concept for the effectiveness of monolinks. In contrast, the AF2 assessment score was only able to identify the most accurate conformation in two out of six cases. Our results highlight the complementarity of AF2 with experimental methods like XL-MS, with the MP and XLP scores providing reliable metrics to assess the quality of the predicted models. The MP and XLP scoring functions mentioned above are available at https://gitlab.com/topf-lab/xlms-tools.

AlphaFold modeling of nepovirus 3C-like proteinases provides new insights into their diverse substrate specificities

The majority of picornaviral 3C proteinases (3Cpro) cleavage sites possess glutamine at the P1 position. Plant nepovirus 3C-like proteinases (3CLpro) show however much broader specificity, cleaving not only after glutamine, but also after several basic and hydrophobic residues. To investigate this difference, we employed AlphaFold to generate structural models of twelve selected 3CLpro, representing six substrate specificities. Generally, we observed favorable correlations between the architecture and charge of nepovirus proteinase S1 subsites and their ability to accept or restrict larger residues. The models identified a conserved aspartate residue close to the P1 residue in the S1 subsites of all nepovirus proteinases examined, consistent with the observed strong bias against negatively-charged residues at the P1 position of nepovirus cleavage sites. Finally, a cramped S4 subsite along with the presence of two unique histidine and serine residues explains the strict requirement of the grapevine fanleaf virus proteinase for serine at the P4 position.

Cryo-EM structure of a 16.5-kDa small heat-shock protein from Methanocaldococcus jannaschii

The small heat-shock protein (sHSP) from the archaea Methanocaldococcus jannaschii, MjsHSP16.5, functions as a broad substrate ATP-independent holding chaperone protecting misfolded proteins from aggregation under stress conditions. This protein is the first sHSP characterized by X-ray crystallography, thereby contributing significantly to our understanding of sHSPs. However, despite numerous studies assessing its functions and structures, the precise arrangement of the N-terminal domains (NTDs) within this sHSP cage remains elusive. Here we present the cryo-electron microscopy (cryo-EM) structure of MjsHSP16.5 at 2.49-Å resolution. The subunits of MjsHSP16.5 in the cryo-EM structure exhibit lesser compaction compared to their counterparts in the crystal structure. This structural feature holds particular significance in relation to the biophysical properties of MjsHSP16.5, suggesting a close resemblance to this sHSP native state. Additionally, our cryo-EM structure unveils the density of residues 24-33 within the NTD of MjsHSP16.5, a feature that typically remains invisible in the majority of its crystal structures. Notably, these residues show a propensity to adopt a β-strand conformation and engage in antiparallel interactions with strand β1, both intra- and inter-subunit modes. These structural insights are corroborated by structural predictions, disulfide bond cross-linking studies of Cys-substitution mutants, and protein disaggregation assays. A comprehensive understanding of the structural features of MjsHSP16.5 expectedly holds the potential to inspire a wide range of interdisciplinary applications, owing to the renowned versatility of this sHSP as a nanoscale protein platform.

WIV, a protein domain found in a wide number of arthropod viruses, which probably facilitates infection

The most powerful approach to detect distant homologues of a protein is based on structure prediction and comparison. Yet this approach is still inapplicable to many viral proteins. Therefore, we applied a powerful sequence-based procedure to identify distant homologues of viral proteins. It relies on three principles: (1) traces of sequence similarity can persist beyond the significance cutoff of homology detection programmes; (2) candidate homologues can be identified among proteins with weak sequence similarity to the query by using 'contextual' information, e.g. taxonomy or type of host infected; (3) these candidate homologues can be validated using highly sensitive profile-profile comparison. As a test case, this approach was applied to a protein without known homologues, encoded by ORF4 of Lake Sinai viruses (which infect bees). We discovered that the ORF4 protein contains a domain that has homologues in proteins from >20 taxa of viruses infecting arthropods. We called this domain 'widespread, intriguing, versatile' (WIV), because it is found in proteins with a wide variety of functions and within varied domain contexts. For example, WIV is found in the NSs protein of tospoviruses, a global threat to food security, which infect plants as well as their arthropod vectors; in the RNA2 ORF1-encoded protein of chronic bee paralysis virus, a widespread virus of bees; and in various proteins of cypoviruses, which infect the silkworm Bombyx mori. Structural modelling with AlphaFold indicated that the WIV domain has a previously unknown fold, and bibliographical evidence suggests that it facilitates infection of arthropods.

Deep learning applications in protein crystallography

Deep learning techniques can recognize complex patterns in noisy, multidimensional data. In recent years, researchers have started to explore the potential of deep learning in the field of structural biology, including protein crystallography. This field has some significant challenges, in particular producing high-quality and well ordered protein crystals. Additionally, collecting diffraction data with high completeness and quality, and determining and refining protein structures can be problematic. Protein crystallographic data are often high-dimensional, noisy and incomplete. Deep learning algorithms can extract relevant features from these data and learn to recognize patterns, which can improve the success rate of crystallization and the quality of crystal structures. This paper reviews progress in this field.

Advances in biosynthesis of peptide drugs: Technology and industrialization

Peptide drugs are developed from endogenous or synthetic peptides with specific biological activities. They have advantages of strong target specificity, high efficacy and low toxicity, thus showing great promise in the treatment of many diseases such as cancer, infections, and diabetes. Although an increasing number of peptide drugs have entered market in recent years, the preparation of peptide drug substances is yet a bottleneck problem for their industrial production. Comparing to the chemical synthesis method, peptide biosynthesis has advantages of simple synthesis, low cost, and low contamination. Therefore, the biosynthesis technology of peptide drugs has been widely used for manufacturing. Herein, we reviewed the development of peptide drugs and recent advances in peptide biosynthesis technology, in order to shed a light to the prospect of industrial production of peptide drugs based on biosynthesis technology.

ChatGPT in research and health professions education: challenges, opportunities, and future directions

ChatGPT was launched by OpenAI in November 2022 and within 2 months it became popular across a wide range of industrial, social, and intellectual contexts including healthcare education. This article reviews the impact of ChatGPT on research and health professions education by identifying the challenges and opportunities in these fields. Additionally, it aims to provide future directions to mitigate the challenges and maximize the benefits of this technology in health professions education. ChatGPT has the potential to revolutionize the field of research and health professions education. However, there is a need to address ethical concerns and limitations such as lack of real-time data, data inaccuracies, biases, plagiarism, and copyright infringement before its implementation. Future research can highlight the ways to mitigate these challenges; establish guidelines and policies; and explore how effectively ChatGPT and other AI tools can be used in the field of research and healthcare professions education.

A multi-reservoir extruder for time-resolved serial protein crystallography and compound screening at X-ray free-electron lasers

Serial crystallography at X-ray free-electron lasers (XFELs) permits the determination of radiation-damage free static as well as time-resolved protein structures at room temperature. Efficient sample delivery is a key factor for such experiments. Here, we describe a multi-reservoir, high viscosity extruder as a step towards automation of sample delivery at XFELs. Compared to a standard single extruder, sample exchange time was halved and the workload of users was greatly reduced. In-built temperature control of samples facilitated optimal extrusion and supported sample stability. After commissioning the device with lysozyme crystals, we collected time-resolved data using crystals of a membrane-bound, light-driven sodium pump. Static data were also collected from the soluble protein tubulin that was soaked with a series of small molecule drugs. Using these data, we identify low occupancy (as little as 30%) ligands using a minimal amount of data from a serial crystallography experiment, a result that could be exploited for structure-based drug design.

Comprehensive Screening of a Light-Inducible Split Cre Recombinase with Domain Insertion Profiling

Splitting proteins with light- or chemically inducible dimers provides a mechanism for post-translational control of protein function. However, current methods for engineering stimulus-responsive split proteins often require significant protein engineering expertise and the laborious screening of individual constructs. To address this challenge, we use a pooled library approach that enables rapid generation and screening of nearly all possible split protein constructs in parallel, where results can be read out by using sequencing. We perform our method on Cre recombinase with optogenetic dimers as a proof of concept, resulting in comprehensive data on the split sites throughout the protein. To improve the accuracy in predicting split protein behavior, we develop a Bayesian computational approach to contextualize errors inherent to experimental procedures. Overall, our method provides a streamlined approach for achieving inducible post-translational control of a protein of interest.

Differential reliance of CTD-nuclear envelope phosphatase 1 on its regulatory subunit in ER lipid synthesis and storage

The endoplasmic reticulum (ER) is the site for the synthesis of the major membrane and storage lipids. Lipin 1 produces diacylglycerol, the lipid intermediate critical for the synthesis of both membrane and storage lipids in the ER. CTD-Nuclear Envelope Phosphatase 1 (CTDNEP1) regulates lipin 1 to restrict ER membrane synthesis, but its role in lipid storage in mammalian cells is unknown. Here, we show that the ubiquitin-proteasome degradation pathway controls the levels of ER/nuclear envelope-associated CTDNEP1 to regulate ER membrane synthesis through lipin 1. The N-terminus of CTDNEP1 is an amphipathic helix that targets to the ER, nuclear envelope and lipid droplets. We identify key residues at the binding interface of CTDNEP1 with its regulatory subunit NEP1R1 and show that they facilitate complex formation in vivo and in vitro . We demonstrate a role for NEP1R1 in temporarily shielding CTDNEP1 from proteasomal degradation to regulate lipin 1 and restrict ER size. Unexpectedly, we found that NEP1R1 is not required for CTDNEP1's role in restricting lipid droplet biogenesis. Thus, the reliance of CTDNEP1 function on its regulatory subunit differs during ER membrane synthesis and lipid storage. Together, our work provides a framework into understanding how the ER regulates lipid synthesis and storage under fluctuating conditions.

Dissecting the Molecular Basis of Host Leucine-Rich Repeat Containing 15 Mediated Interaction with Receptor Binding Domain of SARS-CoV-2 Spike Protein: A Computational Approach

The detection of leucine-rich repeat containing 15 (LRRC15) as a connecting link with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) underscores the possibility of its involvement in differential restriction activity of SARS-CoV-2 pathways. However, the structure-function mechanism of LRRC15 involving the receptor binding domain (RBD) of the SARS-CoV-2 spike protein and their mode of interaction is largely unknown. Using state-of-the-art AlphaFold2 and all-atom molecular dynamics simulations, our findings provide evidences of alternative binding modes of RBD with LRR units of LRRC15 having varied affinities. Contribution of both the receptor binding regions in RBD, including receptor binding motif in accommodating the LRR domain, towards the C-terminal region, emphasizes its differential role in modulating host cell receptiveness for SARS-CoV-2, the innate immune system, as well as antiviral tone. However, further experimental validations are necessary for unravelling the unknown mechanism and distinctive features of this host receptor in the COVID-19 pandemic, involving both the transmembrane as well as cytoplasmic domain.

Computational analysis of regulatory regions in human protein kinases

Eukaryotic proteins often feature modular domain structures comprising globular domains that are connected by linker regions and intrinsically disordered regions that may contain important functional motifs. The intramolecular interactions of globular domains and nonglobular regions can play critical roles in different aspects of protein function. However, studying these interactions and their regulatory roles can be challenging due to the flexibility of nonglobular regions, the long insertions separating interacting modules, and the transient nature of some interactions. Obtaining the experimental structures of multiple domains and functional regions is more difficult than determining the structures of individual globular domains. High-quality structural models generated by AlphaFold offer a unique opportunity to study intramolecular interactions in eukaryotic proteins. In this study, we systematically explored intramolecular interactions between human protein kinase domains (KDs) and potential regulatory regions, including globular domains, N- and C-terminal tails, long insertions, and distal nonglobular regions. Our analysis identified intramolecular interactions between human KDs and 35 different types of globular domains, exhibiting a variety of interaction modes that could contribute to orthosteric or allosteric regulation of kinase activity. We also identified prevalent interactions between human KDs and their flanking regions (N- and C-terminal tails). These interactions exhibit group-specific characteristics and can vary within each specific kinase group. Although long-range interactions between KDs and nonglobular regions are relatively rare, structural details of these interactions offer new insights into the regulation mechanisms of several kinases, such as HASPIN, MAPK7, MAPK15, and SIK1B.

Seeking the interspecies crosswalk for filamentous microbe effectors

Both pathogenic and symbiotic microorganisms modulate the immune response and physiology of their host to establish a suitable niche. Key players in mediating colonization outcome are microbial effector proteins that act either inside (cytoplasmic) or outside (apoplastic) the plant cells and modify the abundance or activity of host macromolecules. We compile novel insights into the much-disputed processes of effector secretion and translocation of filamentous organisms, namely fungi and oomycetes. We report how recent studies that focus on unconventional secretion and effector structure challenge the long-standing image of effectors as conventionally secreted proteins that are translocated with the aid of primary amino acid sequence motifs. Furthermore, we emphasize the potential of diverse, unbiased, state-of-the-art proteomics approaches in the holistic characterization of fungal and oomycete effectomes.

Updated protein domain annotation of the PARP protein family sheds new light on biological function

AlphaFold2 and related computational tools have greatly aided studies of structural biology through their ability to accurately predict protein structures. In the present work, we explored AF2 structural models of the 17 canonical members of the human PARP protein family and supplemented this analysis with new experiments and an overview of recent published data. PARP proteins are typically involved in the modification of proteins and nucleic acids through mono or poly(ADP-ribosyl)ation, but this function can be modulated by the presence of various auxiliary protein domains. Our analysis provides a comprehensive view of the structured domains and long intrinsically disordered regions within human PARPs, offering a revised basis for understanding the function of these proteins. Among other functional insights, the study provides a model of PARP1 domain dynamics in the DNA-free and DNA-bound states and enhances the connection between ADP-ribosylation and RNA biology and between ADP-ribosylation and ubiquitin-like modifications by predicting putative RNA-binding domains and E2-related RWD domains in certain PARPs. In line with the bioinformatic analysis, we demonstrate for the first time PARP14's RNA-binding capability and RNA ADP-ribosylation activity in vitro. While our insights align with existing experimental data and are probably accurate, they need further validation through experiments.

In the Age of Machine Learning Cryo-EM Research is Still Necessary: A Path toward Precision Medicine

Machine learning has proven useful in analyzing complex biological data and has greatly influenced the course of research in structural biology and precision medicine. Deep neural network models oftentimes fail to predict the structure of complex proteins and are heavily dependent on experimentally determined structures for their training and validation. Single-particle cryogenic electron microscopy (cryoEM) is also advancing the understanding of biology and will be needed to complement these models by continuously supplying high-quality experimentally validated structures for improvements in prediction quality. In this perspective, the significance of structure prediction methods is highlighted, but the authors also ask, what if these programs cannot accurately predict a protein structure important for preventing disease? The role of cryoEM is discussed to help fill the gaps left by artificial intelligence predictive models in resolving targetable proteins and protein complexes that will pave the way for personalized therapeutics.

AlphaFold and the future of structural biology

This editorial acknowledges the transformative impact of new machine-learning methods, such as the use of AlphaFold, but also makes the case for the continuing need for experimental structural biology.

The Realm of Unconventional Noncovalent Interactions in Proteins: Their Significance in Structure and Function

Proteins and their assemblies are fundamental for living cells to function. Their complex three-dimensional architecture and its stability are attributed to the combined effect of various noncovalent interactions. It is critical to scrutinize these noncovalent interactions to understand their role in the energy landscape in folding, catalysis, and molecular recognition. This Review presents a comprehensive summary of unconventional noncovalent interactions, beyond conventional hydrogen bonds and hydrophobic interactions, which have gained prominence over the past decade. The noncovalent interactions discussed include low-barrier hydrogen bonds, C5 hydrogen bonds, C-H···π interactions, sulfur-mediated hydrogen bonds, n → π* interactions, London dispersion interactions, halogen bonds, chalcogen bonds, and tetrel bonds. This Review focuses on their chemical nature, interaction strength, and geometrical parameters obtained from X-ray crystallography, spectroscopy, bioinformatics, and computational chemistry. Also highlighted are their occurrence in proteins or their complexes and recent advances made toward understanding their role in biomolecular structure and function. Probing the chemical diversity of these interactions, we determined that the variable frequency of occurrence in proteins and the ability to synergize with one another are important not only for ab initio structure prediction but also to design proteins with new functionalities. A better understanding of these interactions will promote their utilization in designing and engineering ligands with potential therapeutic value.

Chaperone function in antigen presentation by MHC class I molecules-tapasin in the PLC and TAPBPR beyond

Peptide loading of MHC-I molecules plays a critical role in the T cell response to infections and tumors as well as to interactions with inhibitory receptors on natural killer (NK) cells. To facilitate and optimize peptide acquisition, vertebrates have evolved specialized chaperones to stabilize MHC-I molecules during their biosynthesis and to catalyze peptide exchange favoring high affinity or optimal peptides to permit transport to the cell surface where stable peptide/MHC-I (pMHC-I) complexes are displayed and are available for interaction with T cell receptors and any of a host of inhibitory and activating receptors. Although components of the endoplasmic reticulum (ER) resident peptide loading complex (PLC) were identified some 30 years ago, the detailed biophysical parameters that govern peptide selection, binding, and surface display have recently been understood better with advances in structural methods including X-ray crystallography, cryogenic electron microscopy (cryo-EM), and computational modeling. These approaches have provided refined mechanistic illustration of the molecular events involved in the folding of the MHC-I heavy chain, its coordinate glycosylation, assembly with its light chain, β₂-microglobulin (β₂m), its association with the PLC, and its binding of peptides. Our current view of this important cellular process as it relates to antigen presentation to CD8⁺ T cells is based on many different approaches: biochemical, genetic, structural, computational, cell biological, and immunological. In this review, taking advantage of recent X-ray and cryo-EM structural evidence and molecular dynamics simulations, examined in the context of past experiments, we attempt a dispassionate evaluation of the details of peptide loading in the MHC-I pathway. By critical evaluation of several decades of investigation, we outline aspects of the peptide loading process that are well-understood and indicate those that demand further detailed investigation. Further studies should contribute not only to basic understanding, but also to applications for immunization and therapy of tumors and infections.

AlphaFold, allosteric, and orthosteric drug discovery: Ways forward

Drug discovery is arguably a highly challenging and significant interdisciplinary aim. The stunning success of the artificial intelligence-powered AlphaFold, whose latest version is buttressed by an innovative machine-learning approach that integrates physical and biological knowledge about protein structures, raised drug discovery hopes that unsurprisingly, have not come to bear. Even though accurate, the models are rigid, including the drug pockets. AlphaFold's mixed performance poses the question of how its power can be harnessed in drug discovery. Here we discuss possible ways of going forward wielding its strengths, while bearing in mind what AlphaFold can and cannot do. For kinases and receptors, an input enriched in active (ON) state models can better AlphaFold's chance of rational drug design success.

The confluence of machine learning and multiscale simulations

Multiscale modeling has a long history of use in structural biology, as computational biologists strive to overcome the time- and length-scale limits of atomistic molecular dynamics. Contemporary machine learning techniques, such as deep learning, have promoted advances in virtually every field of science and engineering and are revitalizing the traditional notions of multiscale modeling. Deep learning has found success in various approaches for distilling information from fine-scale models, such as building surrogate models and guiding the development of coarse-grained potentials. However, perhaps its most powerful use in multiscale modeling is in defining latent spaces that enable efficient exploration of conformational space. This confluence of machine learning and multiscale simulation with modern high-performance computing promises a new era of discovery and innovation in structural biology.

Assessing Fairness of AlphaFold2 Prediction of Protein 3D Structures

AlphaFold2 is reshaping biomedical research by enabling the prediction of a protein's 3D structure solely based on its amino acid sequence. This breakthrough reduces reliance on labor-intensive experimental methods traditionally used to obtain protein structures, thereby accelerating the pace of scientific discovery. Despite the bright future, it remains unclear whether AlphaFold2 can uniformly predict the wide spectrum of proteins equally well. Systematic investigation into the fairness and unbiased nature of its predictions is still an area yet to be thoroughly explored. In this paper, we conducted an in-depth analysis of AlphaFold2's fairness using data comprised of five million reported protein structures from its open-access repository. Specifically, we assessed the variability in the distribution of PLDDT scores, considering factors such as amino acid type, secondary structure, and sequence length. Our findings reveal a systematic discrepancy in AlphaFold2's predictive reliability, varying across different types of amino acids and secondary structures. Furthermore, we observed that the size of the protein exerts a notable impact on the credibility of the 3D structural prediction. AlphaFold2 demonstrates enhanced prediction power for proteins of medium size compared to those that are either smaller or larger. These systematic biases could potentially stem from inherent biases present in its training data and model architecture. These factors need to be taken into account when expanding the applicability of AlphaFold2.

Regulatory ligand binding in plant chalcone isomerase-like (CHIL) proteins

Chalcone isomerase-like (CHIL) is a noncatalytic protein that enhances flavonoid content in green plants by serving as a metabolite binder and a rectifier of chalcone synthase (CHS). Rectification of CHS catalysis occurs through direct protein-protein interactions between CHIL and CHS, which alter CHS kinetics and product profiles, favoring naringenin chalcone production. These discoveries raise questions about how CHIL proteins interact structurally with metabolites and how CHIL-ligand interactions affect interactions with CHS. Using differential scanning fluorimetry (DSF) on a CHIL protein from Vitis Vinifera (VvCHIL), we report positive thermostability effects are induced by the binding of naringenin chalcone and negative thermostability effects are induced by the binding of naringenin. Naringenin chalcone further causes positive changes to CHIL-CHS binding, while naringenin causes negative changes to CHIL-CHS binding. These results suggest that CHILs may act as sensors for ligand-mediated pathway feedback by influencing CHS function. The protein X-ray crystal structure of VvCHIL compared with the protein X-ray crystal structure of a CHIL from Physcomitrella patens, reveals key amino acid differences at a ligand binding site of VvCHIL that can be substituted to nullify the destabilizing effect caused by naringenin. Together these results support a role for CHIL proteins as metabolite sensors that modulate the committed step of the flavonoid pathway.

Critical Assessment of Methods for Predicting the 3D Structure of Proteins and Protein Complexes

Advances in a scientific discipline are often measured by small, incremental steps. In this review, we report on two intertwined disciplines in the protein structure prediction field, modeling of single chains and modeling of complexes, that have over decades emulated this pattern, as monitored by the community-wide blind prediction experiments CASP and CAPRI. However, over the past few years, dramatic advances were observed for the accurate prediction of single protein chains, driven by a surge of deep learning methodologies entering the prediction field. We review the mainscientific developments that enabled these recent breakthroughs and feature the important role of blind prediction experiments in building up and nurturing the structure prediction field. We discuss how the new wave of artificial intelligence-based methods is impacting the fields of computational and experimental structural biology and highlight areas in which deep learning methods are likely to lead to future developments, provided that major challenges are overcome.

The nucleotide excision repair proteins through the lens of molecular dynamics simulations

Mutations that affect the proteins responsible for the nucleotide excision repair (NER) pathway can lead to diseases such as xeroderma pigmentosum, trichothiodystrophy, Cockayne syndrome, and Cerebro-oculo-facio-skeletal syndrome. Hence, understanding their molecular behavior is needed to elucidate these diseases' phenotypes and how the NER pathway is organized and coordinated. Molecular dynamics techniques enable the study of different protein conformations, adaptable to any research question, shedding light on the dynamics of biomolecules. However, as important as they are, molecular dynamics studies focused on DNA repair pathways are still becoming more widespread. Currently, there are no review articles compiling the advancements made in molecular dynamics approaches applied to NER and discussing: (i) how this technique is currently employed in the field of DNA repair, focusing on NER proteins; (ii) which technical setups are being employed, their strengths and limitations; (iii) which insights or information are they providing to understand the NER pathway or NER-associated proteins; (iv) which open questions would be suited for this technique to answer; and (v) where can we go from here. These questions become even more crucial considering the numerous 3D structures published regarding the NER pathway's proteins in recent years. In this work, we tackle each one of these questions, revising and critically discussing the results published in the context of the NER pathway.

Using AlphaFold Predictions in Viral Research

Elucidation of the tertiary structure of proteins is an important task for biological and medical studies. AlphaFold, a modern deep-learning algorithm, enables the prediction of protein structure to a high level of accuracy. It has been applied in numerous studies in various areas of biology and medicine. Viruses are biological entities infecting eukaryotic and procaryotic organisms. They can pose a danger for humans and economically significant animals and plants, but they can also be useful for biological control, suppressing populations of pests and pathogens. AlphaFold can be used for studies of molecular mechanisms of viral infection to facilitate several activities, including drug design. Computational prediction and analysis of the structure of bacteriophage receptor-binding proteins can contribute to more efficient phage therapy. In addition, AlphaFold predictions can be used for the discovery of enzymes of bacteriophage origin that are able to degrade the cell wall of bacterial pathogens. The use of AlphaFold can assist fundamental viral research, including evolutionary studies. The ongoing development and improvement of AlphaFold can ensure that its contribution to the study of viral proteins will be significant in the future.

Integrated mass spectrometry strategy for functional protein complex discovery and structural characterization

The discovery of functional protein complex and the interrogation of the complex structure-function relationship (SFR) play crucial roles in the understanding and intervention of biological processes. Affinity purification-mass spectrometry (AP-MS) has been proved as a powerful tool in the discovery of protein complexes. However, validation of these novel protein complexes as well as elucidation of their molecular interaction mechanisms are still challenging. Recently, native top-down MS (nTDMS) is rapidly developed for the structural analysis of protein complexes. In this review, we discuss the integration of AP-MS and nTDMS in the discovery and structural characterization of functional protein complexes. Further, we think the emerging artificial intelligence (AI)-based protein structure prediction is highly complementary to nTDMS and can promote each other. We expect the hybridization of integrated structural MS with AI prediction to be a powerful workflow in the discovery and SFR investigation of functional protein complexes.

NIAS-Server 2.0: A versatile complementary tool for structural biology studies

Increasing the repertoire of available complementary tools to advance the knowledge of protein structures is fundamental for structural biology. The Neighbors Influence of Amino Acids and Secondary Structures (NIAS) is a server that analyzes a protein's conformational preferences of amino acids. NIAS is based on the Angle Probability List, representing the normalized frequency of empirical conformational preferences, such as torsion angles, of different amino acid pairs and their corresponding secondary structure information, as available in the Protein Data Bank. In this work, we announce the updated NIAS server with the data comprising all structures deposited until Sep 2022, 7 years after the initial release. Unlike the original publication, which accounted for only studies conducted with X-ray crystallography, we added data from solid nuclear magnetic resonance (NMR), solution NMR, CullPDB, Electron Microscopy, and Electron Crystallography using multiple filtering parameters. We also provide examples of how NIAS can be applied as a complementary analysis tool for different structural biology works and what are its limitations.

An agnostic analysis of the human AlphaFold2 proteome using local protein conformations

Knowledge of the 3D structure of proteins is a valuable asset for understanding their precise biological mechanisms. However, the cost of production of 3D structures and experimental difficulties limit their obtaining. The proposal of 3D structural models is consequently an appealing alternative. The release of the AlphaFold Deep Learning approach has revolutionized the field. The recent near-complete human proteome proposal makes it possible to analyse large amounts of data and evaluate the results of the approach in greater depth. The 3D human proteome was thus analysed in light of the classic secondary structures, and many less-used protein local conformations (PolyProline II helices, type of γ-turns, of β-turns and of β-bulges, curvature of the helices, and a structural alphabet). Without questioning the global quality of the approach, this analysis highlights certain local conformations, which maybe poorly predicted and they could therefore be better addressed.

Development of a robust HTRF assay with USP7 full length protein expressed in E. coli prokaryotic system for the identification of USP7 inhibitors

Deubiquitinating enzyme ubiquitin-specific protease 7 (USP7) is a promising therapeutic target. Several USP7 inhibitors accommodated in the catalytic triad of USP7 have been reported with the aid of high-throughput screening (HTS) methods using USP7 catalytic domain truncation. However, the drawbacks of previously reported biochemical cleavage assays, including poor stability, fluorescence interference, time-consuming, expensive, more importantly the selectivity issue, have challenged the USP7-targeted drug discovery. In this work, we demonstrated the functional heterogeneity and essential role of different structural elements in the USP7 full activation, highlighting the necessity of USP7 full length in drug discovery. Apart from reported two pockets in the catalytic triad, five additional ligandable pockets were predicted based on the proposed USP7 full length models by AlphaFold and homology modelling. A reliable homogeneous time-resolved fluorescence (HTRF) HTS method was established based on the cleavage mechanism of USP7 towards the ubiquitin precursor UBA10. The USP7 full length protein was successfully expressed in the relatively cost-effective E. coli prokaryotic system and used to simulate the auto-activated USP7 in nature. Via screening our in-house library (∼ 1500 compounds), 19 hit compounds with >20% of inhibition rate were identified for further optimization. This assay will enrich the toolbox for the identification of highly potent and selective USP7 inhibitors for clinical use.

Novel machine learning approaches revolutionize protein knowledge

Breakthrough methods in machine learning (ML), protein structure prediction, and novel ultrafast structural aligners are revolutionizing structural biology. Obtaining accurate models of proteins and annotating their functions on a large scale is no longer limited by time and resources. The most recent method to be top ranked by the Critical Assessment of Structure Prediction (CASP) assessment, AlphaFold 2 (AF2), is capable of building structural models with an accuracy comparable to that of experimental structures. Annotations of 3D models are keeping pace with the deposition of the structures due to advancements in protein language models (pLMs) and structural aligners that help validate these transferred annotations. In this review we describe how recent developments in ML for protein science are making large-scale structural bioinformatics available to the general scientific community.

Hic Sunt Dracones: Molecular Docking in Uncharted Territories with Structures from AlphaFold2 and RoseTTAfold

AlphaFold2 and RoseTTAfold impress with their high accuracy in protein structure prediction. However, for structure-based virtual screenings, not only the overall structure but especially the binding sites need to be accurately predicted. In this work, the docking performance for 66 targets with known ligands but without experimental structures available in the protein data bank was elucidated. The results suggest that using an experimental surrogate-ligand complex is often superior over homology models, and only at low sequence identity to the closest homologue AlphaFold2 structures show an equal performance. The generally high fluctuation of receiver operating characteristic area under the curve values obtained for different homology models suggests that multiple combinations of docking programs and homology models should be tested prior to prospective virtual screenings, and in some cases post-processing of crude models might be necessary.

Mechanism of assembly of type 4 filaments: everything you always wanted to know (but were afraid to ask)

Type 4 filaments (T4F) are a superfamily of filamentous nanomachines - virtually ubiquitous in prokaryotes and functionally versatile - of which type 4 pili (T4P) are the defining member. T4F are polymers of type 4 pilins, assembled by conserved multi-protein machineries. They have long been an important topic for research because they are key virulence factors in numerous bacterial pathogens. Our poor understanding of the molecular mechanisms of T4F assembly is a serious hindrance to the design of anti-T4F therapeutics. This review attempts to shed light on the fundamental mechanistic principles at play in T4F assembly by focusing on similarities rather than differences between several (mostly bacterial) T4F. This holistic approach, complemented by the revolutionary ability of artificial intelligence to predict protein structures, led to an intriguing mechanistic model of T4F assembly.

Accelerating crystal structure determination with iterative AlphaFold prediction

Experimental structure determination can be accelerated with artificial intelligence (AI)-based structure-prediction methods such as AlphaFold. Here, an automatic procedure requiring only sequence information and crystallographic data is presented that uses AlphaFold predictions to produce an electron-density map and a structural model. Iterating through cycles of structure prediction is a key element of this procedure: a predicted model rebuilt in one cycle is used as a template for prediction in the next cycle. This procedure was applied to X-ray data for 215 structures released by the Protein Data Bank in a recent six-month period. In 87% of cases our procedure yielded a model with at least 50% of C^α atoms matching those in the deposited models within 2 Å. Predictions from the iterative template-guided prediction procedure were more accurate than those obtained without templates. It is concluded that AlphaFold predictions obtained based on sequence information alone are usually accurate enough to solve the crystallographic phase problem with molecular replacement, and a general strategy for macromolecular structure determination that includes AI-based prediction both as a starting point and as a method of model optimization is suggested.

MRPS36 provides a structural link in the eukaryotic 2-oxoglutarate dehydrogenase complex

The tricarboxylic acid cycle is the central pathway of energy production in eukaryotic cells and plays a key part in aerobic respiration throughout all kingdoms of life. One of the pivotal enzymes in this cycle is 2-oxoglutarate dehydrogenase complex (OGDHC), which generates NADH by oxidative decarboxylation of 2-oxoglutarate to succinyl-CoA. OGDHC is a megadalton protein complex originally thought to be assembled from three catalytically active subunits (E1o, E2o, E3). In fungi and animals, however, the protein MRPS36 has more recently been proposed as a putative additional component. Based on extensive cross-linking mass spectrometry data supported by phylogenetic analyses, we provide evidence that MRPS36 is an important member of the eukaryotic OGDHC, with no prokaryotic orthologues. Comparative sequence analysis and computational structure predictions reveal that, in contrast with bacteria and archaea, eukaryotic E2o does not contain the peripheral subunit-binding domain (PSBD), for which we propose that MRPS36 evolved as an E3 adaptor protein, functionally replacing the PSBD. We further provide a refined structural model of the complete eukaryotic OGDHC of approximately 3.45 MDa with novel mechanistic insights.

AlphaFold accelerates artificial intelligence powered drug discovery: efficient discovery of a novel CDK20 small molecule inhibitor

The application of artificial intelligence (AI) has been considered a revolutionary change in drug discovery and development. In 2020, the AlphaFold computer program predicted protein structures for the whole human genome, which has been considered a remarkable breakthrough in both AI applications and structural biology. Despite the varying confidence levels, these predicted structures could still significantly contribute to structure-based drug design of novel targets, especially the ones with no or limited structural information. In this work, we successfully applied AlphaFold to our end-to-end AI-powered drug discovery engines, including a biocomputational platform PandaOmics and a generative chemistry platform Chemistry42. A novel hit molecule against a novel target without an experimental structure was identified, starting from target selection towards hit identification, in a cost- and time-efficient manner. PandaOmics provided the protein of interest for the treatment of hepatocellular carcinoma (HCC) and Chemistry42 generated the molecules based on the structure predicted by AlphaFold, and the selected molecules were synthesized and tested in biological assays. Through this approach, we identified a small molecule hit compound for cyclin-dependent kinase 20 (CDK20) with a binding constant Kd value of 9.2 ± 0.5 μM (n = 3) within 30 days from target selection and after only synthesizing 7 compounds. Based on the available data, a second round of AI-powered compound generation was conducted and through this, a more potent hit molecule, ISM042-2-048, was discovered with an average Kd value of 566.7 ± 256.2 nM (n = 3). Compound ISM042-2-048 also showed good CDK20 inhibitory activity with an IC50 value of 33.4 ± 22.6 nM (n = 3). In addition, ISM042-2-048 demonstrated selective anti-proliferation activity in an HCC cell line with CDK20 overexpression, Huh7, with an IC50 of 208.7 ± 3.3 nM, compared to a counter screen cell line HEK293 (IC50 = 1706.7 ± 670.0 nM). This work is the first demonstration of applying AlphaFold to the hit identification process in drug discovery.

Accurate Prediction of Enzyme Thermostabilization with Rosetta Using AlphaFold Ensembles

Thermostability enhancement is a fundamental aspect of protein engineering as a biocatalyst's half-life is key for its industrial and biotechnological application, particularly at high temperatures and under harsh conditions. Thermostability changes upon mutation originate from modifications of the free energy of unfolding (ΔG_u), making thermostabilization extremely challenging to predict with computational methods. In this contribution, we combine global conformational sampling with energy prediction using AlphaFold and Rosetta to develop a new computational protocol for the quantitative prediction of thermostability changes upon laboratory evolution of acyltransferase LovD and lipase LipA. We highlight how using an ensemble of protein conformations rather than a single three-dimensional model is mandatory for accurate thermostability predictions. By comparing our approaches with existing ones, we show that ensembles based on AlphaFold models provide more accurate and robust calculated thermostability trends than ensembles based solely on crystallographic structures as the latter introduce a strong distortion (scaffold bias) in computed thermostabilities. Eliminating this bias is critical for computer-guided enzyme design and evaluating the effect of multiple mutations on protein stability.