Predicting C4 photosynthesis evolution: modular, individually adaptive steps on a Mount Fuji fitness landscape.

An ultimate goal of evolutionary biology is the prediction and experimental verification of adaptive trajectories on macroevolutionary timescales. This aim has rarely been achieved for complex biological systems, as models usually lack clear correlates of organismal fitness. Here, we simulate the fitness landscape connecting two carbon fixation systems: C3 photosynthesis, used by most plant species, and the C4 system, which is more efficient at ambient CO2 levels and elevated temperatures and which repeatedly evolved from C3. Despite extensive sign epistasis, C4 photosynthesis is evolutionarily accessible through individually adaptive steps from any intermediate state. Simulations show that biochemical subtraits evolve in modules; the order and constitution of modules confirm and extend previous hypotheses based on species comparisons. Plant-species-designated C3-C4 intermediates lie on predicted evolutionary trajectories, indicating that they indeed represent transitory states. Contrary to expectations, we find no slowdown of adaptation and no diminishing fitness gains along evolutionary trajectories.

Deep learning allows genome-scale prediction of Michaelis constants from structural features.

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 on a deep numerical representation of the enzyme's amino acid sequence. We provide genome-scale KM predictions for 47 model organisms, which can be used to approximately relate metabolite concentrations to cellular physiology and to aid in the parameterization of kinetic models of cellular metabolism.

Kinetic profiling of metabolic specialists demonstrates stability and consistency of in vivo enzyme turnover numbers.

Enzyme turnover numbers (kcats) are essential for a quantitative understanding of cells. Because kcats are traditionally measured in low-throughput assays, they can be inconsistent, labor-intensive to obtain, and can miss in vivo effects. We use a data-driven approach to estimate in vivo kcats using metabolic specialist Escherichia coli strains that resulted from gene knockouts in central metabolism followed by metabolic optimization via laboratory evolution. By combining absolute proteomics with fluxomics data, we find that in vivo kcats are robust against genetic perturbations, suggesting that metabolic adaptation to gene loss is mostly achieved through other mechanisms, like gene-regulatory changes. Combining machine learning and genome-scale metabolic models, we show that the obtained in vivo kcats predict unseen proteomics data with much higher precision than in vitro kcats. The results demonstrate that in vivo kcats can solve the problem of inconsistent and low-coverage parameterizations of genome-scale cellular models.

Independent component analysis recovers consistent regulatory signals from disparate datasets.

The availability of bacterial transcriptomes has dramatically increased in recent years. This data deluge could result in detailed inference of underlying regulatory networks, but the diversity of experimental platforms and protocols introduces critical biases that could hinder scalable analysis of existing data. Here, we show that the underlying structure of the E. coli transcriptome, as determined by Independent Component Analysis (ICA), is conserved across multiple independent datasets, including both RNA-seq and microarray datasets. We subsequently combined five transcriptomics datasets into a large compendium containing over 800 expression profiles and discovered that its underlying ICA-based structure was still comparable to that of the individual datasets. With this understanding, we expanded our analysis to over 3,000 E. coli expression profiles and predicted three high-impact regulons that respond to oxidative stress, anaerobiosis, and antibiotic treatment. ICA thus enables deep analysis of disparate data to uncover new insights that were not visible in the individual datasets.

Cellular responses to reactive oxygen species are predicted from molecular mechanisms.

Catalysis using iron-sulfur clusters and transition metals can be traced back to the last universal common ancestor. The damage to metalloproteins caused by reactive oxygen species (ROS) can prevent cell growth and survival when unmanaged, thus eliciting an essential stress response that is universal and fundamental in biology. Here we develop a computable multiscale description of the ROS stress response in Escherichia coli, called OxidizeME. We use OxidizeME to explain four key responses to oxidative stress: 1) ROS-induced auxotrophy for branched-chain, aromatic, and sulfurous amino acids; 2) nutrient-dependent sensitivity of growth rate to ROS; 3) ROS-specific differential gene expression separate from global growth-associated differential expression; and 4) coordinated expression of iron-sulfur cluster (ISC) and sulfur assimilation (SUF) systems for iron-sulfur cluster biosynthesis. These results show that we can now develop fundamental and quantitative genotype-phenotype relationships for stress responses on a genome-wide basis.

Adaptations of Escherichia coli strains to oxidative stress are reflected in properties of their structural proteomes.

BACKGROUND: The reconstruction of metabolic networks and the three-dimensional coverage of protein structures have reached the genome-scale in the widely studied Escherichia coli K-12 MG1655 strain. The combination of the two leads to the formation of a structural systems biology framework, which we have used to analyze differences between the reactive oxygen species (ROS) sensitivity of the proteomes of sequenced strains of E. coli. As proteins are one of the main targets of oxidative damage, understanding how the genetic changes of different strains of a species relates to its oxidative environment can reveal hypotheses as to why these variations arise and suggest directions of future experimental work. RESULTS: Creating a reference structural proteome for E. coli allows us to comprehensively map genetic changes in 1764 different strains to their locations on 4118 3D protein structures. We use metabolic modeling to predict basal ROS production levels (ROStype) for 695 of these strains, finding that strains with both higher and lower basal levels tend to enrich their proteomes with antioxidative properties, and speculate as to why that is. We computationally assess a strain's sensitivity to an oxidative environment, based on known chemical mechanisms of oxidative damage to protein groups, defined by their localization and functionality. Two general groups - metalloproteins and periplasmic proteins - show enrichment of their antioxidative properties between the 695 strains with a predicted ROStype as well as 116 strains with an assigned pathotype. Specifically, proteins that a) utilize a molybdenum ion as a cofactor and b) are involved in the biogenesis of fimbriae show intriguing protective properties to resist oxidative damage. Overall, these findings indicate that a strain's sensitivity to oxidative damage can be elucidated from the structural proteome, though future experimental work is needed to validate our model assumptions and findings. CONCLUSION: We thus demonstrate that structural systems biology enables a proteome-wide, computational assessment of changes to atomic-level physicochemical properties and of oxidative damage mechanisms for multiple strains in a species. This integrative approach opens new avenues to study adaptation to a particular environment based on physiological properties predicted from sequence alone.

Causal mutations from adaptive laboratory evolution are outlined by multiple scales of genome annotations and condition-specificity.

BACKGROUND: Adaptive Laboratory Evolution (ALE) has emerged as an experimental approach to discover mutations that confer phenotypic functions of interest. However, the task of finding and understanding all beneficial mutations of an ALE experiment remains an open challenge for the field. To provide for better results than traditional methods of ALE mutation analysis, this work applied enrichment methods to mutations described by a multiscale annotation framework and a consolidated set of ALE experiment conditions. A total of 25,321 unique genome annotations from various sources were leveraged to describe multiple scales of mutated features in a set of 35 Escherichia coli based ALE experiments. These experiments totalled 208 independent evolutions and 2641 mutations. Additionally, mutated features were statistically associated across a total of 43 unique experimental conditions to aid in deconvoluting mutation selection pressures. RESULTS: Identifying potentially beneficial, or key, mutations was enhanced by seeking coding and non-coding genome features significantly enriched by mutations across multiple ALE replicates and scales of genome annotations. The median proportion of ALE experiment key mutations increased from 62%, with only small coding and non-coding features, to 71% with larger aggregate features. Understanding key mutations was enhanced by considering the functions of broader annotation types and the significantly associated conditions for key mutated features. The approaches developed here were used to find and characterize novel key mutations in two ALE experiments: one previously unpublished with Escherichia coli grown on glycerol as a carbon source and one previously published with Escherichia coli tolerized to high concentrations of L-serine. CONCLUSIONS: The emergent adaptive strategies represented by sets of ALE mutations became more clear upon observing the aggregation of mutated features across small to large scale genome annotations. The clarification of mutation selection pressures among the many experimental conditions also helped bring these strategies to light. This work demonstrates how multiscale genome annotation frameworks and data-driven methods can help better characterize ALE mutations, and thus help elucidate the genotype-to-phenotype relationship of the studied organism.

Combining genetic and evolutionary engineering to establish C4 metabolism in C3 plants.

To feed a world population projected to reach 9 billion people by 2050, the productivity of major crops must be increased by at least 50%. One potential route to boost the productivity of cereals is to equip them genetically with the 'supercharged' C4 type of photosynthesis; however, the necessary genetic modifications are not sufficiently understood for the corresponding genetic engineering programme. In this opinion paper, we discuss a strategy to solve this problem by developing a new paradigm for plant breeding. We propose combining the bioengineering of well-understood traits with subsequent evolutionary engineering, i.e. mutagenesis and artificial selection. An existing mathematical model of C3-C4 evolution is used to choose the most promising path towards this goal. Based on biomathematical simulations, we engineer Arabidopsis thaliana plants that express the central carbon-fixing enzyme Rubisco only in bundle sheath cells (Ru-BSC plants), the localization characteristic for C4 plants. This modification will initially be deleterious, forcing the Ru-BSC plants into a fitness valley from where previously inaccessible adaptive steps towards C4 photosynthesis become accessible through fitness-enhancing mutations. Mutagenized Ru-BSC plants are then screened for improved photosynthesis, and are expected to respond to imposed artificial selection pressures by evolving towards C4 anatomy and biochemistry.

Modelling metabolic evolution on phenotypic fitness landscapes: a case study on C4 photosynthesis.

How did the complex metabolic systems we observe today evolve through adaptive evolution? The fitness landscape is the theoretical framework to answer this question. Since experimental data on natural fitness landscapes is scarce, computational models are a valuable tool to predict landscape topologies and evolutionary trajectories. Careful assumptions about the genetic and phenotypic features of the system under study can simplify the design of such models significantly. The analysis of C4 photosynthesis evolution provides an example for accurate predictions based on the phenotypic fitness landscape of a complex metabolic trait. The C4 pathway evolved multiple times from the ancestral C3 pathway and models predict a smooth 'Mount Fuji' landscape accordingly. The modelled phenotypic landscape implies evolutionary trajectories that agree with data on modern intermediate species, indicating that evolution can be predicted based on the phenotypic fitness landscape. Future directions will have to include structural changes of metabolic fitness landscape structure with changing environments. This will not only answer important evolutionary questions about reversibility of metabolic traits, but also suggest strategies to increase crop yields by engineering the C4 pathway into C3 plants.

Reproductive toxicity in birds predicted by physiologically-based kinetics and bioenergetics modelling.

Effects on the growth and reproduction of birds are important endpoints in the environmental risk assessment (ERA) of pesticides. Toxicokinetic-toxicodynamic models based on dynamic energy budget theory (DEB) are promising tools to predict these effects mechanistically and make extrapolations relevant to ERA. However, before DEB-TKTD models are accepted as part of ERA for birds, ecotoxicological case studies are required so that stakeholders can assess their capabilities. We present such a case-study, modelling the effects of the fluopyram metabolite benzamide on the northern bobwhite quail (Colinus virginianus). We parametrised a DEB-TKTD model for the embryo stage on the basis of an egg injection study, designed to provide data for model development. We found that information on various endpoints, such as survival, growth, and yolk utilisation were needed to clearly distinguish between the performance of model variants with different TKTD assumptions. The calibration data were best explained when it was assumed that chemical uptake occurs via the yolk and that benzamide places stress on energy assimilation and mobilisation. To be able to bridge from the in vitro tests to real-life exposure, we developed a physiologically-based toxicokinetic (PBK) model for the quail and used it to predict benzamide exposure inside the eggs based on dietary exposure in a standard reproductive toxicity study. We then combined the standard DEB model with the TKTD module calibrated to the egg injection studies and used it to predict effects on hatchling and 14-day chick weight based on the exposure predicted by the PBK model. Observed weight reductions, relative to controls, were accurately predicted. Thus, we demonstrate that DEB-TKTD models, in combination with suitable experimental data and, if necessary, with an exposure model, can be used in bird ERA to predict chemical effects on reproduction.

Energetic basis for bird ontogeny and egg-laying applied to the bobwhite quail.

Birds build up their reproductive system and undergo major tissue remodeling for each reproductive season. Energetic specifics of this process are still not completely clear, despite the increasing interest. We focused on the bobwhite quail - one of the most intensely studied species due to commercial and conservation interest - to elucidate the energy fluxes associated with reproduction, including the fate of the extra assimilates ingested prior to and during reproduction. We used the standard Dynamic Energy Budget model, which is a mechanistic process-based model capable of fully specifying and predicting the life cycle of the bobwhite quail: its growth, maturation and reproduction. We expanded the standard model with an explicit egg-laying module and formulated and tested two hypotheses for energy allocation of extra assimilates associated with reproduction: Hypothesis 1, that the energy and nutrients are used directly for egg production; and Hypothesis 2, that the energy is mostly spent fueling the increased metabolic costs incurred by building up and maintaining the reproductive system and, subsequently, by egg-laying itself. Our results suggest that Hypothesis 2 is the more likely energy pathway. Model predictions capture well the whole ontogeny of a generalized northern bobwhite quail and are able to reproduce most of the data variability via variability in (i) egg size, (ii) egg-laying rate and (iii) inter-individual physiological variability modeled via the zoom factor, i.e. assimilation potential. Reliable models with a capacity to predict physiological responses of individuals are relevant not only for experimental setups studying effects of various natural and anthropogenic pressures on the quail as a bird model organism, but also for wild quail management and conservation. The model is, with minor modifications, applicable to other species of interest, making it a most valuable tool in the emerging field of conservation physiology.

Quantitative Morphometric, Physiological, and Metabolic Characteristics of Chickens and Mallards for Physiologically Based Kinetic Model Development.

Physiologically based kinetic (PBK) models are a promising tool for xenobiotic environmental risk assessment that could reduce animal testing by predicting in vivo exposure. PBK models for birds could further our understanding of species-specific sensitivities to xenobiotics, but would require species-specific parameterization. To this end, we summarize multiple major morphometric and physiological characteristics in chickens, particularly laying hens (Gallus gallus) and mallards (Anas platyrhynchos) in a meta-analysis of published data. Where such data did not exist, data are substituted from domesticated ducks (Anas platyrhynchos) and, in their absence, from chickens. The distribution of water between intracellular, extracellular, and plasma is similar in laying hens and mallards. Similarly, the lengths of the components of the small intestine (duodenum, jejunum, and ileum) are similar in chickens and mallards. Moreover, not only are the gastrointestinal absorptive areas similar in mallard and chickens but also they are similar to those in mammals when expressed on a log basis and compared to log body weight. In contrast, the following are much lower in laying hens than mallards: cardiac output (CO), hematocrit (Hct), and blood hemoglobin. There are shifts in ovary weight (increased), oviduct weight (increased), and plasma/serum concentrations of vitellogenin and triglyceride between laying hens and sexually immature females. In contrast, reproductive state does not affect the relative weights of the liver, kidneys, spleen, and gizzard.

Quantitative Comparison of Avian and Mammalian Physiologies for Parameterization of Physiologically Based Kinetic Models.

Physiologically based kinetic (PBK) models facilitate chemical risk assessment by predicting in vivo exposure while reducing the need for animal testing. PBK models for mammals have seen significant progress, which has yet to be achieved for avian systems. Here, we quantitatively compare physiological, metabolic and anatomical characteristics between birds and mammals, with the aim of facilitating bird PBK model development. For some characteristics, there is considerable complementarity between avian and mammalian species with identical values for the following: blood hemoglobin and hemoglobin concentrations per unit erythrocyte volume together with relative weights of the liver, heart, and lungs. There are also systematic differences for some major characteristics between avian and mammalian species including erythrocyte volume, plasma concentrations of albumin, total protein and triglyceride together with liver cell size and relative weights of the kidney, spleen, and ovary. There are also major differences between characteristics between sexually mature and sexually immature female birds. For example, the relative weights of the ovary and oviduct are greater in sexually mature females compared to immature birds as are the plasma concentrations of triglyceride and vitellogenin. Both these sets of differences reflect the genetic "blue print" inherited from ancestral archosaurs such as the production of large eggs with yolk filled oocytes surrounded by egg white proteins, membranes and a calciferous shell together with adaptions for flight in birds or ancestrally in flightless birds.

A generic avian physiologically-based kinetic (PBK) model and its application in three bird species.

Physiologically-based kinetic (PBK) models are effective tools for designing toxicological studies and conducting extrapolations to inform hazard characterization in risk assessment by filling data gaps and defining safe levels of chemicals. In the present work, a generic avian PBK model for male and female birds was developed using PK-Sim and MoBi from the Open Systems Pharmacology Suite (OSPS). The PBK model includes an ovulation model (egg development) to predict concentrations of chemicals in eggs from dietary exposure. The model was parametrized for chicken (Gallus gallus), bobwhite quail (Colinus virginianus) and mallard duck (Anas platyrhynchos) and was tested with nine chemicals for which in vivo studies were available. Time-concentration profiles of chemicals reaching tissues and egg compartment were simulated and compared to in vivo data. The overall accuracy of the PBK model predictions across the analyzed chemicals was good. Model simulations were found to be in the range of 22-79% within a 3-fold and 41-89% were within 10- fold deviation of the in vivo observed data. However, for some compounds scarcity of in-vivo data and inconsistencies between published studies allowed only a limited goodness of fit evaluation. The generic avian PBK model was developed following a "best practice" workflow describing how to build a PBK model for novel species. The credibility and reproducibility of the avian PBK models were scored by evaluation according to the available guidance documents from WHO (2010), and OECD (2021), to increase applicability, confidence and acceptance of these in silico models in chemical risk assessment.

Modeling photosynthetic resource allocation connects physiology with evolutionary environments.

The regulation of resource allocation in biological systems observed today is the cumulative result of natural selection in ancestral and recent environments. To what extent are observed resource allocation patterns in different photosynthetic types optimally adapted to current conditions, and to what extent do they reflect ancestral environments? Here, we explore these questions for C3, C4, and C3-C4 intermediate plants of the model genus Flaveria. We developed a detailed mathematical model of carbon fixation, which accounts for various environmental parameters and for energy and nitrogen partitioning across photosynthetic components. This allows us to assess environment-dependent plant physiology and performance as a function of resource allocation patterns. Models of C4 plants optimized for conditions experienced by evolutionary ancestors perform better than models accounting for experimental growth conditions, indicating low phenotypic plasticity. Supporting this interpretation, the model predicts that C4 species need to re-allocate more nitrogen between photosynthetic components than C3 species to adapt to new environments. We thus hypothesize that observed resource distribution patterns in C4 plants still reflect optimality in ancestral environments, allowing the quantitative inference of these environments from today's plants. Our work allows us to quantify environmental effects on photosynthetic resource allocation and performance in the light of evolutionary history.

A biochemically-interpretable machine learning classifier for microbial GWAS.

Current machine learning classifiers have successfully been applied to whole-genome sequencing data to identify genetic determinants of antimicrobial resistance (AMR), but they lack causal interpretation. Here we present a metabolic model-based machine learning classifier, named Metabolic Allele Classifier (MAC), that uses flux balance analysis to estimate the biochemical effects of alleles. We apply the MAC to a dataset of 1595 drug-tested Mycobacterium tuberculosis strains and show that MACs predict AMR phenotypes with accuracy on par with mechanism-agnostic machine learning models (isoniazid AUC = 0.93) while enabling a biochemical interpretation of the genotype-phenotype map. Interpretation of MACs for three antibiotics (pyrazinamide, para-aminosalicylic acid, and isoniazid) recapitulates known AMR mechanisms and suggest a biochemical basis for how the identified alleles cause AMR. Extending flux balance analysis to identify accurate sequence classifiers thus contributes mechanistic insights to GWAS, a field thus far dominated by mechanism-agnostic results.

Modeling genome-wide enzyme evolution predicts strong epistasis underlying catalytic turnover rates.

Systems biology describes cellular phenotypes as properties that emerge from the complex interactions of individual system components. Little is known about how these interactions have affected the evolution of metabolic enzymes. Here, we combine genome-scale metabolic modeling with population genetics models to simulate the evolution of enzyme turnover numbers (kcats) from a theoretical ancestor with inefficient enzymes. This systems view of biochemical evolution reveals strong epistatic interactions between metabolic genes that shape evolutionary trajectories and influence the magnitude of evolved kcats. Diminishing returns epistasis prevents enzymes from developing higher kcats in all reactions and keeps the organism far from the potential fitness optimum. Multifunctional enzymes cause synergistic epistasis that slows down adaptation. The resulting fitness landscape allows kcat evolution to be convergent. Predicted kcat parameters show a significant correlation with experimental data, validating our modeling approach. Our analysis reveals how evolutionary forces shape modern kcats and the whole of metabolism.

Machine learning and structural analysis of Mycobacterium tuberculosis pan-genome identifies genetic signatures of antibiotic resistance.

Mycobacterium tuberculosis is a serious human pathogen threat exhibiting complex evolution of antimicrobial resistance (AMR). Accordingly, the many publicly available datasets describing its AMR characteristics demand disparate data-type analyses. Here, we develop a reference strain-agnostic computational platform that uses machine learning approaches, complemented by both genetic interaction analysis and 3D structural mutation-mapping, to identify signatures of AMR evolution to 13 antibiotics. This platform is applied to 1595 sequenced strains to yield four key results. First, a pan-genome analysis shows that M. tuberculosis is highly conserved with sequenced variation concentrated in PE/PPE/PGRS genes. Second, the platform corroborates 33 genes known to confer resistance and identifies 24 new genetic signatures of AMR. Third, 97 epistatic interactions across 10 resistance classes are revealed. Fourth, detailed structural analysis of these genes yields mechanistic bases for their selection. The platform can be used to study other human pathogens.

Machine learning applied to enzyme turnover numbers reveals protein structural correlates and improves metabolic models.

Knowing the catalytic turnover numbers of enzymes is essential for understanding the growth rate, proteome composition, and physiology of organisms, but experimental data on enzyme turnover numbers is sparse and noisy. Here, we demonstrate that machine learning can successfully predict catalytic turnover numbers in Escherichia coli based on integrated data on enzyme biochemistry, protein structure, and network context. We identify a diverse set of features that are consistently predictive for both in vivo and in vitro enzyme turnover rates, revealing novel protein structural correlates of catalytic turnover. We use our predictions to parameterize two mechanistic genome-scale modelling frameworks for proteome-limited metabolism, leading to significantly higher accuracy in the prediction of quantitative proteome data than previous approaches. The presented machine learning models thus provide a valuable tool for understanding metabolism and the proteome at the genome scale, and elucidate structural, biochemical, and network properties that underlie enzyme kinetics.

BLISTER Regulates Polycomb-Target Genes, Represses Stress-Regulated Genes and Promotes Stress Responses in Arabidopsis thaliana.

HIGHLIGHTS The PRC2 interacting protein BLISTER likely acts downstream of PRC2 to silence Polycomb target genes and is a key regulator of specific stress responses in Arabidopsis. Polycomb group (PcG) proteins are key epigenetic regulators of development. The highly conserved Polycomb repressive complex 2 (PRC2) represses thousands of target genes by trimethylating H3K27 (H3K27me3). Plant specific PcG components and functions are largely unknown, however, we previously identified the plant-specific protein BLISTER (BLI) as a PRC2 interactor. BLI regulates PcG target genes and promotes cold stress resistance. To further understand the function of BLI, we analyzed the transcriptional profile of bli-1 mutants. Approximately 40% of the up-regulated genes in bli are PcG target genes, however, bli-1 mutants did not show changes in H3K27me3 levels at all tested genes, indicating that BLI regulates PcG target genes downstream of or in parallel to PRC2. Interestingly, a significant number of BLI regulated H3K27me3 target genes is regulated by the stress hormone absciscic acid (ABA). We further reveal an overrepresentation of genes responding to abiotic stresses such as drought, high salinity, or heat stress among the up-regulated genes in bli mutants. Consistently, bli mutants showed reduced desiccation stress tolerance. We conclude that the PRC2 associated protein BLI is a key regulator of stress-responsive genes in Arabidopsis: it represses ABA-responsive PcG target genes, likely downstream of PRC2, and promotes resistance to several stresses such as cold and drought.