Biology subjects

Ytreberg, F. M.

Publications and source records attributed to Ytreberg, F. M..

15 recordsLinked to original sources

bioRxiv2026

PepXPro: a framework for curating, generating, and optimizing structure-affinity protein-peptide datasets

Protein-peptide interactions are central to cellular signaling and to a growing class of peptide therapeutics, yet the datasets used to develop and benchmark computational methods for protein-peptide modeling remain poorly standardized. Available databases prioritize comprehensive coverage but require task-specific curation, while published benchmarks are typically distributed as static collections built with heterogeneous curation, quality-filtering, redundancy-reduction, and sampling strategies, limiting reproducibility and cross-study comparison. We present PepXPro, a modular framework that transforms publicly available protein-peptide structure-affinity resources into curated datasets and reproducible benchmark collections generated under user-defined criteria. PepXPro is organized into three components: Scrape, for deterministic curation of protein-peptide complex entries from public resources; GenSample, for constructing configurable subsets under explicit quality, redundancy, and sampling constraints; and Benchmark, for evaluating candidate subsets and selecting a nonredundant, representative, general-purpose benchmark for distribution. Starting from PDBbind and complementary resources, the curation pipeline yields a pool of proteinpeptide complex entries that retains chemically complex cases, including disulfide- linked cyclic peptides, which are commonly excluded from existing benchmarks. We release PepXPro Benchmark v1, a benchmark comprising 70 non-redundant protein- peptide complexes with experimentally determined structures and binding affinities. The underlying framework provides an extensible foundation for reproducible protein- peptide benchmark construction.

biophysics↗

bioRxiv2025

{Phi}X174 bacteriophage viability predicted by protein biophysical modeling

The relationship between genotype and phenotype underlies our ability to understand and predict evolution. Efforts to build genotype-phenotype (GP) maps have revealed several unifying rules: epistasis is pervasive, fitness effects are not normally distributed, and the GP map is non-linear and complicated for high-level phenotypes. A critical step in developing GP maps is evaluating how well predictive models do in explaining observed phenotypes. We utilize the simplicity of a bacteriophage ({Phi}X174) study system to test if an intermediate phenotype (predicted stability of the G capsid protein) explains more complex phenotypes. In doing so, we compare the predictive performance of free energies of folding and binding obtained using numerous molecular modeling methods as well as phylogenetic and basic biochemical/biophysical properties of amino acid substitutions. By creating a large mutational library, we find that {Phi}X174 tolerates about 50% of the amino acid substitutions we inserted into the G protein and that molecular modeling compliments other substitution models for predicting viability. Mutations predicted to have large destabilizing effects are especially informative and are almost universally detrimental. These large-effect substitutions often coincide with the most conserved residues in the G protein. Apart from large-effect mutations, our ability to predict {Phi}X174 phenotypes is fairly poor and we explore various potential confounding factors (e.g., codon bias) that could be considered to improve viability predictions.

molecular biology↗

bioRxiv2025

A Comprehensive Structural and Functional Analysis of Saccharomyces Killer Toxins.

Antifungal killer toxins are cytotoxic proteins that have the potential to combat the growing threat of fungi to human health and agriculture. A lack of empirical tertiary structures has placed limitations on understanding their mechanisms of action and targeting of pathogens. AlphaFold and molecular dynamics simulations were used to create tertiary structure models of all canonical Saccharomyces killer toxins. These models have enabled the prediction of the functional domains of killer toxins and postranslational modifications, including sites of proteolytic cleavage and disulfide bonds. They have also revealed unexpected homology between Saccharomyces killer toxins, suggesting that all but K28 are likely ionophores. Structural homology with the well-studied killer toxins K1 and K2 enabled prior empirical data to predict the antifungal and immunity mechanisms of the K1L, K21, K45, K74, and KHS toxins. The understudied killer toxins Klus, KHR, and K62 were found to have homology to bacterial and plant toxins, including members of the aerolysin family and antifungal lectins. These structural similarities provided clues for the mechanisms of killer toxin carbohydrate binding, oligomerization, and membrane attack. Given the hundreds of sequence homologs of the Saccharomyces killer toxin identified across fungi, modeling studies offer an exciting opportunity to characterize novel toxin-like proteins. This approach is strengthened by the continued use of the model yeast S. cerevisiae to study killer toxins and the wealth of functional data gathered in the decades since their first discovery.

molecular biology↗

bioRxiv2024

Discovery of Novel Targets for Important Human and Plant Fungal Pathogens via Automated Computational Pipeline HitList

Fungi are a major threat to human health and agricultural productivity, causing 1.7 million human deaths and billions of dollars in crop losses and spoilage annually. While various antifungal compounds have been developed to combat these fungi in medical and agricultural settings, but there are concerns that treatment effectiveness is waning due to the emergence of acquired drug resistance and to novel pathogens. Effectiveness is further hampered due to the limited number of modes of action for available antifungal compounds.To develop new strategies for the control and mitigation of fungal disease and spoilage, new antifungals are needed with novel fungal-specific protein targets that can overcome resistance and prevent host toxicity. New antifungals can add new methods of targeting fungi that have no effective control measures. The increasing availability of complete genomes of pathogenic and spoilage fungi has enabled identification of novel protein and RNA targets essential for viability and not found in host plants or humans. In this study, an automated bioinformatics pipeline utilizing BLAST, Clustal{Omega}, and subtractive genomics was created and used to identify potential new targets for any system of hosts and pathogens with available genomic or proteomic data. This pipeline called HitList allows in silico screening of thousands of possible targets. HitList was then used to generate a list of potential antifungal targets for the World Health Organization fungal priority pathogens list and the top 10 agricultural fungal pathogens. Known antifungal targets were found, validating the approach, and an additional eight novel protein targets were discovered that could be used for the rational design of antifungal compounds.

bioinformatics↗

bioRxiv2024

The Prevalence of Killer Yeasts in the Gardens of Fungus-Growing Ants and the Discovery of Novel Killer Toxin named Ksino.

Killer toxins are proteinaceous antifungal molecules produced by yeasts, with activity against a wide range of human and plant pathogenic fungi. Fungus gardens of attine ants in Brazil were surveyed to determine the presence of killer toxin-producing yeasts and to define their antifungal activities and ecological importance. Our results indicate that 10 out of 59 yeasts species isolated from fungal gardens are killer yeasts. Killer yeasts were less likely to inhibit the growth of yeasts isolated from the same environment but more effective at inhibiting yeast isolated from other environments, supporting a role for killer yeasts in shaping community composition. All killer yeasts harbored genome-encoded killer toxins lacked cytoplasmic toxin-encoding elements (i.e., double-stranded RNA satellites and linear double-stranded DNAs). Of all the killer yeasts associated with attine ants, Candida sinolaborantium (strain LESF 1467) showed a broad spectrum of antifungal activities against 39 out of 69 57% of yeast strains tested for toxin susceptibility. The complete genome sequence of C. sinolaborantium LESF 1467 identified a new killer toxin, Ksino, with similarities in primary sequence and tertiary structure to the Saccharomyces cerevisiae killer toxin named Klus. Surveys of publicly available genome databases identified homologs of Ksino in the genomes of yeast strains of Saccharomycetes and Pichiomycetes, as well as other species of Ascomycota and Basidiomycota filamentous fungi. This demonstrates that killer yeasts can be widespread in attine ant fungus gardens, possibly influencing fungal community composition and the importance of these complex microbial communities for discovering novel antifungal molecules. ImportanceAttine ants perform essential ecosystem services through the harvesting of substrates for fungiculture. The cultured fungi are a food source for attine ants. Characterizing antifungal toxin-producing yeasts (killer yeasts) is vital to understanding how they might protect gardens from invasion by unwanted fungal species. This study describes a new toxin named Ksino from the yeast Candida sinolaborantium, a member of a new group of toxins found across many different species of fungi. This work supports the role of killer yeasts in the ecology of fungicultures and competition between fungi. The observed high prevalence of killer yeasts in fungal gardens also enables the discovery of novel antifungal molecules with the potential to be applied against disease-causing fungi.

microbiology↗

bioRxiv2024

Defining the alpha domains of K1-like killer toxins

Antifungal killer toxins produced by yeasts undergo post-translational processing to enable the formation of mature and active toxins. The preprocessed, immature K1 killer toxin from Saccharomyces cerevisiae has four different domains (delta, alpha, gamma, and beta) that are cleaved to produce a mature toxin consisting of a disulfide-linked alpha and beta heterodimer. K1 homologs from various yeast species of the Saccharomycotina have been previously identified by primary sequence homology. Primary sequence analysis has identified putative cleavage sites across all homologs, with some conservation with cleavage sites identified in K1 and the likely importance of the conserved carboxypeptidases Kex1 and Kex2. The identification of sites of proteolytic processing has enabled an accurate definition of the domain boundaries of delta, alpha, gamma, and beta domains in all K1 homologs. Functional assays revealed that many K1 homologs contain cytotoxic alpha domains. Interestingly, several killer toxins that lacked measurable antifungal activity still exhibited toxic alpha domains when expressed independently. These findings suggest that alpha domain toxicity is conserved across K1 homologs, and that variations in post-translational processing and domain interactions may influence full-length toxin functionality. This work provides insights into the evolutionary conservation and diversity of yeast killer toxins.

biochemistry↗

bioRxiv2024

Leveraging neural networks to correct FoldX free energy estimates

Proteins play a pivotal role in many biological processes, and changes in their amino acid sequences can lead to dysfunction and disease. These changes can affect protein folding or interaction with other biomolecules, such as preventing antibodies from inhibiting a viral infection or causing proteins to misfold. The ability to predict the effects of mutations in proteins is crucial. Although experimental techniques can accurately quantify the effect of mutations on protein folding free energies and protein-protein binding free energies, they are often time-consuming and costly. By contrast, computational techniques offer fast and cost-effective alternatives for estimating free energies, but they typically suffer from lower accuracy. Enhancing the accuracy of computational predictions is therefore of high importance, with the potential to greatly impact fields ranging from drug design to understanding disease mechanisms. One such widely used computational method, FoldX, is capable of rapidly predicting the relative folding stability ({Delta}{Delta}Gfold) for a protein as well as the relative binding affinity ({Delta}{Delta}Gbind) between proteins using a single protein structure as input. However, it can suffer from low accuracy, especially for antibody-antigen systems. In this work, we trained a neural network on FoldX output to enhance its prediction accuracy. We first performed FoldX calculations on the largest datasets available for mutations that affect binding (SKEMPIv2) and folding (ProTherm4) with experimentally measured {Delta}{Delta}G. Features were then extracted from the FoldX output files including its prediction for {Delta}{Delta}G. We then developed and optimized a neural network framework to predict the difference between FoldXs estimated {Delta}{Delta}G and the experimental data, creating a model capable of producing a correction factor. Our approach showed significant improvements in Pearson correlation performance. For single mutations affecting folding, the correlation improved from a baseline of 0.3 to 0.66. In terms of binding, performance increased from 0.37 to 0.61 for single mutations and from 0.52 to 0.81 for double mutations. For epistasis, the correlation for binding affinity (both singles and doubles) improved from 0.19 to 0.59. Our results also indicated that models trained on double mutations enhanced accuracy when predicting higher-order mutations (such as triple or quadruple mutations), whereas models trained on singles did not. This suggests that interaction energy and epistasis effects present in the FoldX output are not fully utilized by FoldX itself. Once trained, these models add minimal computational time but provide a substantial increase in performance, especially for higher-order mutations and epistasis. This makes them a valuable addition to any free energy prediction pipeline using FoldX. Furthermore, we believe this technique can be further optimized and tested for predicting antibody escape, aiding in the efficient development of watch lists.

biophysics↗

bioRxiv2024

Exploring the ability of the MD+FoldX method to predict SARS-CoV-2 antibody escape mutations using large-scale data

Antibody escape mutations pose a significant challenge to the effectiveness of vaccines and antibody-based therapies. The ability to predict these escape mutations with computer simulations would allow us to detect threats early and develop effective countermeasures, but a lack of large-scale experimental data has hampered the validation of these calculations. In this study, we evaluate the ability of the MD+FoldX molecular modeling method to predict escape mutations by leveraging a large deep mutational scanning dataset, focusing on the SARS-CoV-2 receptor binding domain. Our results show a positive correlation between predicted and experimental data, indicating that mutations with reduced predicted binding affinity correlate moderately with higher experimental escape fractions. We also demonstrate that better performance can be achieved using affinity cutoffs tailored to distinct antibody-antigen interactions rather than a one-size-fits-all approach. We find that 70% of the systems surpass the 50% precision mark, and demonstrate success in identifying mutations present in significant variants of concern and variants of interest. Despite promising results for some systems, our study highlights the challenges in comparing predicted and experimental values. It also emphasizes the need for new binding affinity methods with improved accuracy that are fast enough to estimate hundreds to thousands of antibody-antigen binding affinities.

biophysics↗

bioRxiv2022

Defining the HIV Capsid Binding Site of Nucleoporin 153

The interaction between the HIV-1 capsid (CA) and human nucleoporin 153 (NUP153) is vital for delivering the HIV-1 preintegration complex into the nucleus via the nuclear pore complex. The interaction with CA requires a phenylalanine/glycine-containing motif in the C-terminus of NUP153. This study used molecular modeling and biochemical assays to determine the amino acids of NUP153 that are essential for its interactions with CA. Molecular dynamics, FoldX, and PyRosetta simulations delineated the minimal CA binding motif of NUP153 based on the known structure of NUP153 bound to the HIV-1 CA hexamer. Computational predictions were experimentally validated by testing the interaction of NUP153 with CA using an in vitro binding assay and a cell-based TRIM-NUP153C restriction assay. This multidisciplinary approach identified eight amino acids from P1411 to G1418 that stably engage with CA, with significant correlations between molecular models and empirical experiments. Specifically, P1411, V1414, F1415, T1416, F1417, and G1418 were confirmed as critical amino acids required to interact NUP153 with CA. IMPORTANCEHuman immunodeficiency virus (HIV) can infect non-dividing cells by interacting with host nuclear pores. The host nuclear pore protein NUP153 directly interacts with the HIV capsid to promote viral nuclear entry. This study used a multidisciplinary approach combining computational and experimental techniques to map the essential amino acids of NUP153 required for HIV capsid interaction. This approach revealed that the HIV capsid interacts specifically with only six amino acids of NUP153, suggesting other FG-containing motifs could also interact with the capsid. Based on molecular modeling, naturally occurring polymorphisms in human and non-human primates would be predicted to prevent NUP153 interaction with capsid, potentially protecting from HIV infection.

molecular biology↗

bioRxiv2022

Molecular modeling predicts novel antibody escape mutations in respiratory syncytial virus fusion glycoprotein

Monoclonal antibodies are increasingly used for the prevention and/or treatment of viral infections. One caveat of their use is the ability of viruses to evolve resistance to antibody binding and neutralization. Computational strategies to predict which mutations will result in antibody resistance would be invaluable because current methods for identifying potential escape mutations are labor intensive and system-biased. Respiratory syncytial virus is an important pathogen for which monoclonal antibodies against the fusion (F) protein are used to prevent severe disease in high-risk infants. In this study, we used an approach that combines molecular dynamics simulations with FoldX to estimate changes in free energy in F protein folding and binding to the motavizumab antibody upon each possible amino acid change. We systematically selected 8 predicted escape mutations and tested them in an infectious clone. Consistent with our F protein stability predictions, replication-effective viruses were observed for each selected mutation. Six of the eight variants showed increased resistance to neutralization by motavizumab. Flow cytometry was used to validate the estimated (model-predicted) effects on antibody binding to F. Using surface plasmon resonance, we determined that changes in the on-rate of motavizumab binding were responsible for the reduced affinity for two novel escape mutations. Our study empirically validates the accuracy of our molecular modeling approach and emphasizes the role of biophysical protein modeling in predicting viral resistance to antibody-based therapeutics that can be used to monitor the emergence of resistant viruses and to design improved therapeutic antibodies. ImportanceRespiratory syncytial virus (RSV) causes severe disease in young infants, particularly those with heart or lung diseases or born prematurely. As no vaccine is currently available, monoclonal antibodies are used to prevent severe RSV disease in high-risk infants. While it is known that RSV evolves to avoid recognition by antibodies, screening tools that can predict which changes to the virus will lead to antibody resistance are greatly needed.

microbiology↗

bioRxiv2020

EfgA is a conserved formaldehyde sensor that halts bacterial translation in response to elevated formaldehyde

Normal cellular processes give rise to toxic metabolites that cells must mitigate. Formaldehyde is a universal stressor and potent metabolic toxin that is generated in organisms from bacteria to humans. Methylotrophic bacteria such as Methylorubrum extorquens face an acute challenge due to their production of formaldehyde as an obligate central intermediate of single-carbon metabolism. Mechanisms to sense and respond to formaldehyde were speculated to exist in methylotrophs for decades but had never been discovered. Here we identify a member of the DUF336 domain family, named efgA for enhanced formaldehyde growth, that plays an important role in endogenous formaldehyde stress response in M. extorquens PA1 and is found almost exclusively in methylotrophic taxa. Our experimental analyses reveal that EfgA is a formaldehyde sensor that inhibits translation in response to elevated levels of formaldehyde. Heterologous expression of EfgA in Escherichia coli increases formaldehyde resistance, indicating that its interaction partners are widespread and conserved and may include translational machinery. EfgA represents the first example of a formaldehyde stress response system that does not involve enzymatic detoxification. Thus, EfgA comprises a unique stress response mechanism in bacteria, whereby a single protein directly senses elevated levels of a toxic intracellular metabolite and modulates translational activity.

microbiology↗

bioRxiv2020

Implementing and assessing an alchemical method for calculating protein-protein binding free energy

Protein-protein binding is fundamental to most biological processes. It is important to be able to use computation to accurately estimate the change in protein-protein binding free energy due to mutations in order to answer biological questions that would be experimentally challenging, laborious or time consuming. Although non-rigorous free energy methods are faster, rigorous alchemical molecular dynamics-based methods are considerably more accurate and are becoming more feasible with the advancement of computer hardware and molecular simulation software. Even with sufficient computational resources, there are still major challenges to using alchemical free energy methods for protein-protein complexes, such as generating hybrid structures and topologies, maintaining a neutral net charge of the system when there is a charge-changing mutation, and setting up the simulation. In the current study, we have used the pmx package to generate hybrid structures and topologies, and a double-system/single-box approach to maintain the net charge of the system. To test the approach, we predicted relative binding affinities for two protein-protein complexes using a non-equilibrium alchemical method based on the Crooks fluctuation theorem and compared the results with experimental values. The method correctly identified stabilizing from destabilizing mutations for a small protein-protein complex, but was not as successful to the larger, more challenging antibody complex. In addition, the correlation between predicted and experimental relative binding affinities was high for smaller complex, and low for the other larger complex.

biophysics↗

bioRxiv2020

Assessment of Software Methods for Estimating Protein-Protein Relative Binding Affinities

A growing number of computational tools have been developed to accurately and rapidly predict the impact of amino acid mutations on protein-protein relative binding affinities. Such tools have many applications, for example, designing new drugs and studying evolutionary mechanisms. In the search for accuracy, many of these methods employ expensive yet rigorous molecular dynamics simulations. By contrast, non-rigorous methods use less exhaustive statistical mechanics, allowing for more efficient calculations. However, it is unclear if such methods retain enough accuracy to replace rigorous methods in binding affinity calculations. This trade-off between accuracy and computational expense makes it difficult to determine the best method for a particular system or study. Here, eight non-rigorous computational methods were assessed using eight antibody-antigen and eight non-antibody-antigen complexes for their ability to accurately predict relative binding affinities ({Delta}{Delta}G) for 654 single mutations. In addition to assessing accuracy, we analyzed the CPU cost and performance for each method using a variety of physico-chemical structural features. This allowed us to posit scenarios in which each method may be best utilized. Most methods performed worse when applied to antibody-antigen complexes compared to non-antibody-antigen complexes. Rosetta-based JayZ and EasyE methods classified mutations as destabilizing ({Delta}{Delta}G < -0.5 kcal/mol) with high (83-98%) accuracy and a relatively low computational cost for non-antibody-antigen complexes. Some of the most accurate results for antibody-antigen systems came from combining molecular dynamics with FoldX with a correlation coefficient (r) of 0.46, but this was also the most computationally expensive method. Overall, our results suggest these methods can be used to quickly and accurately predict stabilizing versus destabilizing mutations but are less accurate at predicting actual binding affinities. This study highlights the need for continued development of reliable, accessible, and reproducible methods for predicting binding affinities in antibody-antigen proteins and provides a recipe for using current methods.

biophysics↗

bioRxiv2020

Predicting the Viability of Beta-Lactamase: How Folding and Binding Free Energies Correlate with Beta-Lactamase Fitness

One of the long-standing holy grails of molecular evolution has been the ability to predict an organisms fitness directly from its genotype. With such predictive abilities in hand, researchers would be able to more accurately forecast how organisms will evolve and how proteins with novel functions could be engineered, leading to revolutionary advances in medicine and biotechnology. In this work, we assemble the largest reported set of experimental TEM-1 {beta}-lactamase folding free energies and use this data in conjunction with previously acquired fitness data and computational free energy predictions to determine how much of the fitness of {beta}-lactamase can be directly predicted by thermodynamic folding and binding free energies. We focus upon {beta}-lactamase because of its long history as a model enzyme and its central role in antibiotic resistance. Based upon a set of 21 {beta}-lactamase single and double mutants expressly designed to influence protein folding, we first demonstrate that modeling software such as FoldX and PyRosetta designed to compute folding free energies can meaningfully, although not perfectly, predict the experimental folding free energies of single mutants. Interestingly, while these techniques also yield sensible double mutant free energies, we show that they do so for the wrong physical reasons. We then go on to assess how well both experimental and computational folding free energies explain single mutant fitness. We find that folding free energies account for, at most, 24% of the variance in {beta}-lactamase fitness values according to linear models and, somewhat surprisingly, complementing folding free energies with computationally-predicted binding free energies of residues near the active site only increases the folding-only figure by a few percent. This strongly suggests that the majority of {beta}-lactamases fitness is controlled by factors other than free energies. Overall, our results shed a bright light on to what extent the community is justified in using thermodynamic measures to infer protein fitness as well as how applicable modern computational techniques for predicting free energies will be to the large data sets of multiply-mutated proteins forthcoming.

biophysics↗

bioRxiv2019

Expanding the watch list for potential Ebola virus antibody escape mutations

The 2014 outbreak of Ebola virus (EBOV) in Western Africa is the largest recorded filovirus disease outbreak and lead to the death of over 11,000 people. This deadly virus still poses a grave epidemic threat as evidenced by the current (since May 2018) EBOV outbreak in the Democratic Republic of the Congo which has already claimed the lives of over 250 people. One important strategy for combating EBOV epidemics is to anticipate how the evolution of EBOV might undermine treatment since the development of vaccines and antibody therapies are typically based on a single strain (often the 1976 Mayinga) of the EBOV envelope glycoprotein (GP). In a previous study we initiated a watch list of potential antibody escape mutations of EBOV by modeling interactions between EBOV GP and the monoclonal antibody KZ52. This watch list was generated using molecular modeling to estimate the effect of every possible mutation of GP. The final watch list containing 34 mutations were predicted to disrupt GP-KZ52 binding but not to disrupt the ability of GP to fold and to form trimers. In this study, we expand our watch list by including three more monoclonal antibodies with distinct epitopes on GP, namely Antibody 100 (Ab100), Antibody 114 (Ab114) and 13F6-1-2. Our updated watch list contains 127 mutations, three of which have been seen in humans or are experimentally associated with reduced efficacy of antibody treatment. We believe mutations on this broad watch list require attention since they may be a signal of an evolutionary response from EBOV to treatment that could diminish the effectiveness of interventions.

biophysics↗