Biology subjects

Lercher, M. J.

Publications and source records attributed to Lercher, M. J..

18 recordsLinked to original sources

bioRxiv2025

The fate of horizontally acquired genes: rapid initial turnover followed by long-term persistence

A major driver of bacterial evolution is horizontal gene transfer (HGT), the acquisition of genes from other strains or species. Transfers between closely related taxa are more likely to succeed, while the pervasive deletion bias of bacterial genomes drives frequent turnover of horizontally acquired genes. However, whether the rate of gene loss after acquisition is constant across lineages or time remains unclear. Here, we analyze a comprehensive dataset of bacterial genomes to infer the frequency, distribution, and retention of inter-phylum HGT events. The retention of inter-phylum gene transfers is highly skewed, with only a small subset of bacterial genomes accounting for the majority of such events. Most transferred genes are lost rapidly. Those genes that survive the initial purging are retained over long periods, are biased toward functions such as transport and metabolism, and have larger numbers of protein-protein interactions.

evolutionary biology↗

bioRxiv2025

Growth Mechanics and the emergence of metabolic oscillations in growing cells

Unicellular organisms exhibit oscillations in metabolic activity and gene expression, even when growing in static media with no external dynamic stimuli. The cause and possible function of these endogenous oscillations are still not fully understood, with different hypotheses including its origin as a byproduct of misguided regulatory processes. To investigate whether such oscillations could be functional rather than incidental, we introduce Growth Mechanics (GM) as a general mathematical framework to study the dynamic resource allocation in models of whole self-replicating cells, built exclusively on the first principles of fitness maximization, mass conservation, nonlinear reaction kinetics, and constant cell density. Inspired by the physical theory of classical mechanics, we first find the simplest mathematical description the problem in terms of generalized coordinates, and then solve the optimal dynamical resource allocation for cells growing in any given medium using the Euler-Lagrange equation, resulting in analytical "equations of optimal motion" (EOM) that apply for growing cells in general. We then solve these equations numerically for simple growing cell models and show that, in general, oscillations are necessary to maximize cell fitness by getting more saturated reactions at the time they are most needed. We also show how the EOM predict the emergence of quasi-linear dependencies of the metabolic oscillation frequency and the ribosome protein allocation with the cell growth rate at different growth media, in line with the "growth laws" observed experimentally in microbes. This work contributes to the foundations of theoretical cell biology with general quantitative principles and a bridge to theoretical tools of physics.

biophysics↗

bioRxiv2024

Horizontal Gene Transfer Inference: Gene presence-absence outperforms gene trees

Horizontal gene transfer (HGT) is a fundamental driver of prokaryotic evolution, facilitating the acquisition of novel traits and adaptation to new environments. Despite its importance, methods for inferring HGT are rarely systematically compared, leaving a gap in our understanding of their relative strengths and limitations. Validating HGT inference methods is challenging due to the absence of a genomic fossil record that could confirm historical transfer events. Without an empirical gold standard, new inference methods are typically validated using simulated data; however, these simulations may not accurately capture biological complexity and often embed the same assumptions used in the inference methods themselves. Here, we leverage the tendency of HGT events to involve multiple neighboring genes to assess the accuracy of diverse HGT inference methods. We show that methods analyzing gene family presence/absence patterns across species trees consistently outperform approaches based on gene tree-species tree reconciliation. Our findings challenge the prevailing assumption that explicit phylogenetic reconciliation methods are superior to simpler implicit methods. By providing a comprehensive bench-mark, we offer practical recommendations for selecting appropriate methods and indicate avenues for future methodological advancements.

bioinformatics↗

bioRxiv2024

Horizontal gene transfer rates decrease during transitions to new ecosystems

Bacterial colonization of unfamiliar environments constitutes a drastic evolutionary transition, likely altering the strength and direction of selection. This process is often assumed to increase rates of horizontal gene transfer (HGT), a major driver of bacterial adaptation. Larger genomes are thought to facilitate such adaptations by providing broader functional repertoires and more integration sites for foreign DNA. Here, we systematically test these ideas across a broad bacterial phylogeny linked to environmental transitions inferred from metagenomic data. Contrary to expectation, we find that bacteria entering new environments typically have smaller genomes and experience lower rates of HGT. The reduction in HGT is fully explained by genome size, with no residual effect of environmental transitions once size is controlled for. These findings suggest that successful bacterial colonizers rely less on genomic plasticity through HGT than previously assumed, highlighting gaps in our understanding of microbial evolutionary dynamics.

bioinformatics↗

bioRxiv2023

A general prediction model for substrates of transport proteins

Transport proteins play a crucial role in cellular metabolism and are central to many aspects of molecular biology and medicine. Determining the function of transport proteins experimentally is challenging, as they become unstable when isolated from cell membranes. Machine learning-based predictions could provide an efficient alternative. However, existing methods are limited to predicting a small number of specific substrates or broad transporter classes. These limitations stem partly from using small datasets for model training and a choice of input features that lack sufficient information about the prediction problem. Here, we present SPOT, the first general machine learning model that can successfully predict specific substrates for arbitrary transport proteins, achieving an accuracy above 92% on independent and diverse test data covering widely different transporters and a broad range of metabolites. SPOT uses Transformer Networks to represent transporters and substrates numerically. To overcome the problem of missing negative data for training, it augments a large dataset of known transporter-substrate pairs with carefully sampled random molecules as non-substrates. SPOT not only predicts specific transporter-substrate pairs, but also outperforms previously published models designed to predict broad substrate classes for individual transport proteins. We provide a web server and Python function that allows users to explore the substrate scope of arbitrary transporters.

bioinformatics↗

bioRxiv2023

Nearly half of all bacterial gene families are biased toward specific chromosomal positions

Many bacterial genes encoding components of the translation and transcription machinery occupy genomic positions close to the origin of genome replication. These positional biases are thought to result from selection for high expression levels, as genes close to the origin are the first to be duplicated after the initiation of DNA replication. However, recent work indicates that positional biases of RNA genes involved in translation have evolved to support optimal growth rate dependence of expression levels rather than high expression per se. We hypothesized that, more generally, natural selection may have favored the location of different genes at specific chromosomal positions to optimize the growth rate dependence of their relative gene copy numbers. Here we show that 49% of bacterial gene families are preferentially localized to specific chromosomal regions, with most families biased toward either the origin or the terminus of replication. From our hypothesis, we derive six specific predictions regarding the genes chromosomal positioning and functional categories, as well as the minimum cellular doubling times. All six predictions are robustly supported by comparative genomic analyses of 773 bacterial species and by proteomics data from Escherichia coli and Bacillus subtilis. Our findings reveal a complex relationship between bacterial growth, resource allocation and genome organization, and they provide new insights into the physiological significance and potential functions of a large number of gene families.

genomics↗

bioRxiv2023

A multimodal Transformer Network for protein-small molecule interactions enhances drug-target affinity and enzyme-substrate predictions

Most drugs are small molecules, with their activities typically arising from interactions with protein targets. Accurate predictions of these interactions could greatly accelerate pharmaceutical research. Current machine learning models designed for this task have a limited ability to generalize beyond the proteins used for training. This limitation is likely due to a lack of information exchange between the protein and the small molecule during the generation of the required numerical representations. Here, we introduce ProSmith, a machine learning framework that employs a multimodal Transformer Network to simultaneously process protein amino acid sequences and small molecule strings in the same input. This approach facilitates the exchange of all relevant information between the two types of molecules during the computation of their numerical representations, allowing the model to account for their structural and functional interactions. Our final model combines gradient boosting predictions based on the resulting multimodal Transformer Network with independent predictions based on separate deep learning representations of the proteins and small molecules. The resulting predictions outperform all previous models for predicting drug-target interactions, and the model demonstrates unprecedented generalization capabilities to unseen proteins. We further show that the superior performance of ProSmith is not limited to drug-target interaction predictions, but also leads to improvements in other protein-small molecule interaction prediction tasks, the prediction of Michaelis constants KM of enzyme-substrate pairs and the identification of potential substrates for enzymes. The Python code provided can be used to easily implement and improve machine learning predictions of interactions between proteins and arbitrary drug candidates or other small molecules.

bioinformatics↗

bioRxiv2023

Metabolic complexity increases adaptability

Bacteria show a striking ability to adapt to new environments through horizontal gene transfer. Anecdotal evidence suggests that some bacteria are more adaptable than others: e.g., intestinal E. coli frequently spin off pathogenic strains adapted to other human tissues, while gastric Helicobacter pylori do not. However, it is unclear what determines the ability of individual strains or species to adapt to new environments. Here, we use pan-genome scale modeling to explore the ability of 102 different unicellular organisms to adapt to each of 5000+ diverse nutritional environments. While the small metabolic systems of specialized endosymbionts typically require 50+ additional metabolic reactions to adapt to new environments, different strains of the generalist E. coli require on average less than 5 new reactions. Thus, there is a positive feedback between metabolic complexity and adaptability, contrary to speculations that complex systems are generally less adaptable.

evolutionary biology↗

bioRxiv2023

Machine learning models for the prediction of enzyme properties should be tested on proteins not used for model training

The recently published DLKcat model, a deep learning approach for predicting enzyme turnover numbers (kcat), claims to enable high-throughput kcat predictions for metabolic enzymes from any organism and to capture kcat changes for mutated enzymes. Here, we critically evaluate these claims. We show that DLKcat predictions become positively misleading for enzymes with less than 60% sequence identity to the training data, performing worse than simply assuming a mean kcat value for all reactions. Furthermore, DLKcats ability to predict mutation effects is much weaker than implied, capturing only 3% of the experimentally observed variation across mutants not included in the training data. These findings highlight significant limitations in DLKcats generalizability and its practical utility for predicting kcat values for novel enzyme families or mutants, which are crucial applications in fields such as metabolic modeling.

bioinformatics↗

bioRxiv2022

Turnover number predictions for kinetically uncharacterized enzymes using machine and deep learning

The turnover number kcat, a measure of enzyme efficiency, is central to understanding cellular physiology and resource allocation. As experimental kcat estimates are unavailable for the vast majority of enzymatic reactions, the development of accurate computational prediction methods is highly desirable. However, existing machine learning models are limited to a single, well-studied organism, or they provide inaccurate predictions except for enzymes that are highly similar to proteins in the training set. Here, we present TurNuP, a general and organism-independent model that successfully predicts turnover numbers for natural reactions of wild-type enzymes. We constructed model inputs by representing complete chemical reactions through difference fingerprints and by representing enzymes through a modified and re-trained Transformer Network model for protein sequences. TurNuP outperforms previous models and generalizes well even to enzymes that are not similar to proteins in the training set. Parameterizing metabolic models with TurNuP-predicted kcat values leads to improved proteome allocation predictions. To provide a powerful and convenient tool for the study of molecular biochemistry and physiology, we implemented a TurNuP web server at https://turnup.cs.hhu.de.

bioinformatics↗

bioRxiv2022

Growth Mechanics: General principles of optimal cellular resource allocation in balanced growth

The physiology of biological cells evolved under physical and chemical constraints such as mass conservation, nonlinear reaction kinetics, and limits on cell density. For unicellular organisms, the fitness that governs this evolution is mainly determined by the balanced cellular growth rate. We previously introduced Growth Balance Analysis (GBA) as a general framework to model such nonlinear systems, and we presented analytical conditions for optimal balanced growth in the special case that the active reactions are known. Here, we develop Growth Mechanics (GM) as a more general, succinct, and powerful analytical description of the growth optimization of GBA models, which we formulate in terms of a minimal number of dimensionless variables. GM uses Karush-Kuhn-Tucker (KKT) conditions in a Lagrangian formalism. It identifies fundamental principles of optimal resource allocation in GBA models of any size and complexity, including the analytical conditions that determine the set of active reactions at optimal growth. We identify from first principles the economic values of biochemical reactions, expressed as marginal changes in cellular growth rate; these economic values can be related to the costs and benefits of proteome allocation into the reactions catalysts. Our formulation also generalizes the concepts of Metabolic Control Analysis to models of growing cells. GM unifies and extends previous approaches of cellular modeling and analysis, putting forward a program to analyze cellular growth through the stationarity conditions of a Lagrangian function. GM thereby provides a general theoretical toolbox for the study of fundamental mathematical properties of balanced cellular growth.

biophysics↗

bioRxiv2022

Viridiplantae-specific GLXI and GLXII isoforms co-evolved and detoxify glucosone in planta

Reactive carbonyl species (RCS) such as methylglyoxal (MGO) and glyoxal (GO) are highly reactive, unwanted side-products of cellular metabolism maintained at harmless intracellular levels by specific scavenging mechanisms. MGO and GO are metabolized through the glyoxalase (GLX) system, which consists of two enzymes acting in sequence, GLXI and GLXII. While plant genomes encode a large number of different GLX isoforms, it is unclear what their specific functions are and how these arose in evolution. Here, we show that plants possess two GLX systems of different evolutionary origins and with distinct structural and functional properties. The first system is shared by all eukaryotes, scavenges MGO and GO especially during seedling establishment, and features Zn2+-type GLXI, a metal co-factor preference that arose already in the last eukaryotic common ancestor. The GLXI and GLXII of the second system can together metabolize KDG, a glucose-derived RCS, and were acquired by the last common ancestor of viridiplantae through horizontal gene transfer from proteobacteria. In contrast to bacterial GLXI homologs, which are active as dimers, plant Ni2+-type GLXI contain a domain duplication, are active as monomers, and have modified their second active site. The acquisition and neofunctionalization of a structurally, biochemically, and functionally distinct GLX systems indicate that viridiplantae are under strong selection to detoxify a diversity of RCS.

plant biology↗

bioRxiv2022

The substrate scopes of enzymes: a general prediction model based on machine and deep learning

For a comprehensive understanding of metabolism, it is necessary to know all potential substrates for each enzyme encoded in an organisms genome. However, for most proteins annotated as enzymes, it is unknown which primary and/or secondary reactions they catalyze [1], as experimental characterizations are time-consuming and costly. Machine learning predictions could provide an efficient alternative, but are hampered by a lack of information regarding enzyme non-substrates, as available training data comprises mainly positive examples. Here, we present ESP, a general machine learning model for the prediction of enzyme-substrate pairs, with an accuracy of over 90% on independent and diverse test data. This accuracy was achieved by representing enzymes through a modified transformer model [2] with a trained, task-specific token, and by augmenting the positive training data by randomly sampling small molecules and assigning them as non-substrates. ESP can be applied successfully across widely different enzymes and a broad range of metabolites. It outperforms recently published models designed for individual, well-studied enzyme families, which use much more detailed input data [3, 4]. We implemented a user-friendly web server to predict the substrate scope of arbitrary enzymes, which may support not only basic science, but also the development of pharmaceuticals and bioengineering processes.

bioinformatics↗

bioRxiv2021

An optimal growth law for RNA composition and its partial implementation through ribosomal and tRNA gene locations in bacterial genomes

The distribution of cellular resources across bacterial proteins has been quantified through phenomenological growth laws. Here, we describe a complementary bacterial growth law for RNA composition, emerging from optimal cellular resource allocation into ribosomes and ternary complexes. The predicted decline of the tRNA/rRNA ratio with growth rate agrees quantitatively with experimental data. Its regulation appears to be implemented in part through chromosomal localization, as rRNA genes are typically closer to the origin of replication than tRNA genes and thus have increasingly higher gene dosage at faster growth. At the highest growth rates in E. coli, the tRNA/rRNA ratio appears to be regulated entirely through this effect.

systems biology↗

bioRxiv2020

Prediction of Michaelis constants from structuralfeatures using deep learning

The Michaelis constant KM describes the affinity of an enzyme for a specific substrate, and is a central parameter in studies of enzyme kinetics and cellular physiology. As measurements of KM are often difficult and time-consuming, experimental estimates exist for only a minority of enzyme-substrate combinations even in model organisms. Here, we build and train an organism-independent model that successfully predicts KM values for natural enzyme-substrate combinations using machine and deep learning methods. Predictions are based on a task-specific molecular fingerprint of the substrate, generated using a graph neural network, and the domain structure of the enzyme. Model predictions can be used to estimate enzyme efficiencies, to relate metabolite concentrations to cellular physiology, and to fill gaps in the parameterization of kinetic models of cellular metabolism.

systems biology↗

bioRxiv2019

The protein translation machinery is expressed for maximal efficiency in Escherichia coli

Protein synthesis is the most expensive process in fast-growing bacteria1,2. The economic aspects of protein synthesis at the cellular level have been investigated by estimating ribosome activity3-5 and the expression of ribosomes3,6, tRNA7-9, mRNA2, and elongation factors10,11. The observed growth-rate dependencies form the basis of powerful phenomenological bacterial growth laws5,12-16; however, a quantitative theory allowing us to understand these phenomena on the basis of fundamental biophysical and biochemical principles is currently lacking. Here, we show that the observed growth-rate dependence of the concentrations of ribosomes, tRNAs, mRNA, and elongation factors in Escherichia coli can be predicted accurately by minimizing cellular costs in a detailed mathematical model of protein translation; the mechanistic model is only constrained by the physicochemical properties of the molecules and requires no parameter fitting. We approximate the costs of molecule species through their masses, justified by the observation that cellular dry mass per volume is roughly constant across growth rates17 and hence represents a limited resource. Our results also account quantitatively for observed RNA/protein ratios and ribosome activities in E. coli across diverse growth conditions, including antibiotic stresses. Our prediction of active and free ribosome abundance facilitates an estimate of the deactivated ribosome reserve14,18,19, which reaches almost 50% at the lowest growth rates. We conclude that the growth rate dependent composition of E colis protein synthesis machinery is a consequence of natural selection for minimal total cost under physicochemical constraints, a paradigm that might generally be applied to the analysis of resource allocation in complex biological systems.

biophysics↗

bioRxiv2019

An Analytical Theory of Cellular Growth

The biological fitness of unicellular organisms is largely determined by their balanced growth rate, i.e., by the rate with which they replicate their biomass composition. Natural selection on this growth rate occurred under a set of physicochemical constraints, including mass conservation, reaction kinetics, and limits on dry mass per volume; mathematical models that maximize the balanced growth rate while accounting explicitly for these constraints are inevitably nonlinear and have been restricted to small, non-realistic systems. Here, we lay down a general theory of balanced growth states, providing explicit expressions for protein concentrations, fluxes, and the growth rate. These variables are functions of the concentrations of cellular components, for which we calculate marginal fitness costs and benefits that can be related to metabolic control coefficients. At maximal growth rate, the net benefits of all concentrations are equal. Based solely on physicochemical constraints, the growth balance analysis (GBA) framework introduced here unveils fundamental quantitative principles of cellular growth and leads to experimentally testable predictions.

biophysics↗

bioRxiv2017

Enzymes And Substrates Are Balanced At Minimal Combined Mass Concentration In Vivo

A fundamental problem in biology is how cells organize their resource investment. Cellular metabolism, for example, typically involves hundreds of enzymes and metabolites, but it is unclear according to which principles their concentrations are set. Reasoning that natural selection will drive cells towards achieving a given physiological state at minimal cost, we derive a general equation that predicts the concentration of a metabolite from the concentration of the most abundant and costly enzyme consuming it. Simulations of cellular growth as well as experimental data demonstrate that costs are approximately proportional to molecular masses. For effectively irreversible reactions, the cell maximizes its metabolic efficiency by investing equally into substrate and unbound enzyme molecules. Without fitting any free parameters, the resulting model predicts in vivo substrate concentrations from enzyme concentrations and substrate affinities with high accuracy across data from E. coli and diverse eukaryotes (R2=0.79, geometric mean fold-error 1.74). The corresponding organizing principle - the minimization of the summed mass concentrations of solutes - may facilitate reducing the complexity of kinetic models and will contribute to the design of more efficient synthetic cellular systems.

systems biology↗