Metabolism operates under physico-chemical constraints that result in multireaction dependencies. Understanding how multireaction dependencies affect metabolic phenotypes remains challenging, hindering their biotechnological applications. Here, we propose the concept of a forcedly balanced complex that allows to efficiently determine the effects of specific multireaction dependencies on metabolic network functions in constrained-based models. Using this concept, we found that the fraction of multireaction dependencies induced by forcedly balanced complexes in genome-scale metabolic networks followed power law with exponential cut-off. We identified forcedly balanced complexes that are lethal in cancer but have little effect on growth in healthy tissue models. In addition, these forcedly balanced complexes are largely specific to models of particular cancer types. Therefore, multireaction dependencies resulting from forced balancing of complexes represent an innovative means to control cancers that, we argue, can be implemented via transporter engineering. The presented constraint-based approaches pave the way for using multireaction dependencies in metabolic engineering for diverse biotechnological applications.

Unraveling principles of thermodynamics for genome-scale metabolic networks using graph neural networks.

Bacteria live in diverse communities, forming complex networks of interacting species. A central question in bacterial ecology is whether species engage in cooperative or competitive interactions. But this question often neglects the role of the environment. Here, we use genome-scale metabolic networks from two different open-access collections (AGORA and CarveMe) to assess pairwise interactions of different microbes in varying environmental conditions (provision of different environmental compounds). By computationally simulating thousands of environments for 10,000 pairs of bacteria from each collection, we found that most pairs were able to both compete and cooperate depending on the availability of environmental resources. This modeling approach allowed us to determine commonalities between environments that could facilitate the potential for cooperation or competition between a pair of species. Namely, cooperative interactions, especially obligate, were most common in less diverse environments. Further, as compounds were removed from the environment, we found interactions tended to degrade towards obligacy. However, we also found that on average at least one compound could be removed from an environment to switch the interaction from competition to facultative cooperation or vice versa. Together our approach indicates a high degree of plasticity in microbial interactions in response to the availability of environmental resources.

Bacteria live in diverse communities, forming complex networks of interacting species. A central question in bacterial ecology is whether species engage in cooperative or competitive interactions. But this question often neglects the role of the environment. Here, we use genome-scale metabolic networks from two different open-access collections (AGORA and CarveMe) to assess pairwise interactions of different microbes in varying environmental conditions (provision of different environmental compounds). By computationally simulating thousands of environments for 10,000 pairs of bacteria from each collection, we found that most pairs were able to both compete and cooperate depending on the availability of environmental resources. This modeling approach allowed us to determine commonalities between environments that could facilitate the potential for cooperation or competition between a pair of species. Namely, cooperative interactions, especially obligate, were most common in less diverse environments. Further, as compounds were removed from the environment, we found interactions tended to degrade towards obligacy. However, we also found that on average at least one compound could be removed from an environment to switch the interaction from competition to facultative cooperation or vice versa. Together our approach indicates a high degree of plasticity in microbial interactions in response to the availability of environmental resources.

Vibrio parahaemolyticus is a pathogenic bacterium widely distributed in marine environments, posing significant threats to aquatic organisms and human health. The overuse and misuse of antibiotics has led to the development of multidrug- and pan-resistant V. parahaemolyticus strains. There is an urgent need for novel antibacterial therapies with innovative mechanisms of action. In this work, a genome-scale metabolic network model (GMSN) of V. parahaemolyticus, named VPA2061, was reconstructed to predict the metabolites that can be explored as potential drug targets for eliminating V. parahaemolyticus infections. The model comprises 2061 reactions and 1812 metabolites. Through essential metabolite analysis and pathogen-host association screening with VPA2061, 10 essential metabolites critical for the survival of V. parahaemolyticus were identified, which may serve as key candidates for developing new antimicrobial strategies. Additionally, 39 structural analogs were found for these essential metabolites. The molecular docking analysis of the essential metabolites and structural analogs further investigated the potential value of these metabolites for drug design. The GSMN reconstructed in this work provides a new tool for understanding the pathogenic mechanisms of V. parahaemolyticus. Furthermore, the analysis results regarding the essential metabolites hold profound implications for the development of novel antibacterial therapies for V. parahaemolyticus-related disease.

Food fermentation is driven by microbial interactions. This article reviews the types of microbial interactions during food fermentation, the research strategies employed, and their impacts on the quality of fermented foods. Microbial interactions primarily include mutualism, commensalism, amensalism, and competition. Based on these interaction patterns, the safety, nutritional composition, and flavor quality of food can be effectively improved. Achieving precise control of fermented foods' qualities via microbial interaction remains a critical challenge. Emerging technologies such as high-throughput sequencing, cell sorting, and metabolomics enable the systematic analysis of core microbial interaction mechanisms in complex systems. Using synthetic microbial communities and genome-scale metabolic network models, complicated microbial communities can be effectively simplified. In addition, regulatory targets of food quality can be precisely identified. These strategies lay a solid foundation for the precise improvement of fermented food quality and functionality.

Bacterial vaginosis (BV) is the most prevalent vaginal condition among reproductive-age women presenting with vaginal complaints. Despite its significant impact on women’s health, limited knowledge exists regarding the microbial community composition and metabolic interactions associated with BV. In this study, we analyze metagenomic data obtained from human vaginal swabs to generate in silico predictions of BV-associated bacterial metabolic interactions via genome-scale metabolic network reconstructions (GENREs). While most efforts to characterize symptomatic BV (and thus guide therapeutic intervention by identifying responders and non-responders to treatment) are based on genomic profiling, our in silico simulations reveal functional metabolic relatedness between species as quite distinct from genetic relatedness. We grow several of the most common co-occurring bacteria (Prevotella amnii, Prevotella buccalis, Hoylesella timonensis, Lactobacillus iners, Fannyhessea vaginae, and Aerrococcus christenssii) on the spent media of Gardnerella species and perform metabolomics to identify potential mechanisms of metabolic interaction. Through these analyses, we identify BV-associated bacteria that produce caffeate, a compound implicated in estrogen receptor binding, when grown in the spent media of other BV-associated bacteria. These findings underscore the complex and diverse nature of BV-associated bacterial community structures and several of these mechanisms are of potential significance in understanding host-microbiome relationships.

5-Aminolevulinic acid (5-ALA) has been widely used in modern agriculture and therapy as a biostimulant, feed nutrient, and photodynamic drug. Although metabolic engineering strategies have been employed to increase the yield of 5-ALA in Corynebacterium glutamicum, the production of 5-ALA under microaerobic conditions has not been studied. In this paper, we developed, for the first time, overproducing-5-ALA Corynebacterium glutamicum strains under microaerobic conditions, guided by a genome-scale metabolic network model. The engineered strain for the C4 pathway synthesis of 5-ALA was constructed based on the Corynebacterium glutamicum genome-scale metabolic network model iCW773 under different oxygen environmental conditions. The fusion of the key enzymes SucCD and HemA effectively opened the substrate channel and improved the biosynthesis of 5-ALA. Further selection of 5-ALA synthetases alleviated the inhibitory effect of heme, which further improved the titer of 5-ALA. Combinatorial optimization of the lpd, coaA, and ppc genes was employed to enhance the supply of the precursor succinyl-CoA. Finally, a 3.8 g/L 5-ALA titer was achieved in a 5-L bioreactor at 8% dissolved oxygen. This study provides a reference for the synthesis of 5-ALA or other high value-added chemicals with succinyl-CoA as the precursor under microaerobic conditions.

A challenging problem in microbiology is to determine nutritional requirements of microorganisms and culture them, especially for the microbial dark matter detected solely with culture-independent methods. The latter foster an increasing amount of genomic sequences that can be explored with reverse ecology approaches to raise hypotheses on the corresponding populations. Building upon genome-scale metabolic networks (GSMNs) obtained from genome annotations, metabolic models predict contextualized phenotypes using nutrient information. We developed the tool Seed2LP, addressing the inverse problem of predicting source nutrients, or seeds, from a GSMN and a metabolic objective. The originality of Seed2LP is its hybrid model, combining a scalable and discrete Boolean approximation of metabolic activity, with the numerically accurate flux balance analysis (FBA). Seed inference is highly customizable, with multiple search and solving modes, exploring the search space of external and internal metabolites combinations. Application to a benchmark of 107 curated GSMNs highlights the usefulness of a logic modelling method over a graph-based approach to predict seeds, and the relevance of hybrid solving to satisfy FBA constraints. Focusing on the dependency between metabolism and environment, Seed2LP is a computational support contributing to address the multifactorial challenge of culturing possibly uncultured microorganisms. Seed2LP is available on https://github.com/bioasp/seed2lp.

Genetically enhanced microorganisms have wide applications in different fields and the increasing availability of omics data has enabled the development of genome-scale metabolic models (GEMs), which are essential tools in synthetic biology. Bartonella quintana str. Toulouse, a facultative intracellular parasite, presents a small genome and the ability to grow in axenic culture, making it a potential candidate for genome reduction and synthetic biology applications. This study aims to reconstruct and analyze the metabolic network of B. quintana to optimize its growth conditions for laboratory use. A metabolic reconstruction of B. quintana was performed using genome annotation tools (RAST and ModelSEED), followed by refinement using multiple databases (KEGG, BioCyc, BRENDA). Flux Balance Analysis (FBA) was conducted to optimize biomass production, and in-silico knockouts were performed to evaluate growth yield under different media conditions. Additionally, experimental validation was carried out by testing modified culture media and performing proteomic analyses to identify metabolic adaptations. FBA simulations identified key metabolic requirements, including 2-oxoglutarate as a crucial compound for optimal growth. In-silico knockouts of transport genes revealed their essentiality in nutrient uptake. Experimental validation confirmed the role of 2-oxoglutarate and other nutrients in improving bacterial growth, though unexpected decreases in viability were observed under certain supplemented conditions. Proteomic analysis highlighted differential expression of proteins associated with cell wall integrity and metabolic regulation. This study represents a step toward developing B. quintana as a viable chassis for synthetic biology applications. The reconstructed metabolic model provides a comprehensive understanding of B. quintana's metabolic capabilities, identifying essential pathways and growth limitations. While metabolic predictions align with experimental results in key aspects, further refinements are needed to enhance model accuracy and optimize growth conditions.

Differential producibility analysis reveals drug-associated carbon and nitrogen metabolite expressions in Mycobacterium tuberculosis.

Metabolic flux, or the rate of metabolic reactions, is one of the most fundamental metrics describing the status of metabolism in living organisms. However, measuring fluxes across the entire metabolic network remains nearly impossible, especially in multicellular organisms. Computational methods based on flux balance analysis have been used with genome-scale metabolic network models to predict network-level flux wiring1-6. However, such approaches have limited power because of the lack of experimental constraints. Here, we introduce a strategy that infers whole-animal metabolic flux wiring from transcriptional phenotypes in the nematode Caenorhabditis elegans. Using a large-scale Worm Perturb-Seq (WPS) dataset for roughly 900 metabolic genes7, we show that the transcriptional response to metabolic gene perturbations can be integrated with the metabolic network model to infer a highly constrained, semi-quantitative flux distribution. We discover several features of adult C. elegans metabolism, including cyclic flux through the pentose phosphate pathway, lack of de novo purine synthesis flux and the primary use of amino acids and bacterial RNA as a tricarboxylic acid cycle carbon source, all of which we validate by stable isotope tracing. Our strategy for inferring metabolic wiring based on transcriptional phenotypes should be applicable to a variety of systems, including human cells.

Integrating genome-scale metabolic networks with reactive transport models (RTMs) provides a detailed description of the dynamic changes in microbial growth and metabolism. Despite promising demonstrations in the past, computational inefficiency has been pointed out as a critical issue to overcome because it requires repeated application of linear programming (LP) to obtain flux balance analysis (FBA) solutions in every time step and spatial grid. To address this challenge, we propose a new simulation method where we train and validate artificial neural networks (ANNs) using randomly sampled FBA solutions and incorporate the resulting surrogate FBA model (represented as algebraic equations) into RTMs as source/sink terms. We demonstrate the efficiency of our method via a case study of Shewanella oneidensis MR-1. During aerobic growth on lactate, S. oneidensis produces metabolic byproducts (such as pyruvate and acetate), which are subsequently consumed as alternative carbon sources when the preferred nutrients are depleted. To effectively simulate these complex dynamics, we used a cybernetic approach that models metabolic switches as the outcome of dynamic competition among multiple growth options. In both zero-dimensional batch and one-dimensional column configurations, the ANN-based surrogate models achieved substantial reduction of computational time by several orders of magnitude compared to the original LP-based FBA models. Moreover, the ANN models produced robust solutions without any special measures to prevent numerical instability. These developments significantly promote our ability to utilize genome-scale networks in complex, multi-physics, and multi-dimensional ecosystem modeling.

The interpretation of multi-omics datasets obtained from high-throughput approaches is important to understand disease-related physiological changes and to predict biomarkers in body fluids. We present a new metabolite-centred genome-scale metabolic modelling algorithm, the Gene Expression-based Metabolite Centrality Analysis Tool (GEMCAT). GEMCAT enables integration of transcriptomics or proteomics data to predict changes in metabolite concentrations, whichcan be verified by targeted metabolomics. In addition, GEMCAT allows to trace measured and predicted metabolic changes back to the underlying alterations in gene expression or proteomics and thus enables functional interpretation and integration of multi-omics data. We demonstrate the predictive capacity of GEMCAT on three datasets and genome-scale metabolic networks from two different organisms: (i) we integrated transcriptomics and metabolomics data from an engineered human cell line with a functional deletion of the mitochondrial NAD transporter; (ii) we used a large multi-tissue multi-omics dataset from rats for transcriptome- and proteome-based prediction and verification of training-induced metabolic changes and achieved an average prediction accuracy of 70%; and (iii) we used proteomics measurements from patients with inflammatory bowel disease and verified the predicted changes using metabolomics data from the same patients. For this dataset, the prediction accuracy achieved by GEMCAT was 79%.

Genome-wide transcriptomics revealed carbon source-mediated gamma-aminobutyric acid (GABA) production in a probiotic, Lactiplantibacillus pentosus 9D3.

Reasoning about hypotheses and updating knowledge through empirical observations are central to scientific discovery. In this work, we applied logic-based machine learning methods to drive biological discovery by guiding experimentation. Genome-scale metabolic network models (GEMs) - comprehensive representations of metabolic genes and reactions - are widely used to evaluate genetic engineering of biological systems. However, GEMs often fail to accurately predict the behaviour of genetically engineered cells, primarily due to incomplete annotations of gene interactions. The task of learning the intricate genetic interactions within GEMs presents computational and empirical challenges. To efficiently predict using GEM, we describe a novel approach called Boolean Matrix Logic Programming (BMLP) by leveraging Boolean matrices to evaluate large logic programs. We developed a new system, BMLP_{active}, which guides cost-effective experimentation and uses interpretable logic programs to encode a state-of-the-art GEM of a model bacterial organism. Notably, BMLP_{active} successfully learned the interaction between a gene pair with fewer training examples than random experimentation, overcoming the increase in experimental design space. BMLP_{active} enables rapid optimisation of metabolic models to reliably engineer biological systems for producing useful compounds. It offers a realistic approach to creating a self-driving lab for biological discovery, which would then facilitate microbial engineering for practical applications.

Sponges harbour complex microbiomes and as ancient metazoans and important ecosystem players are emerging as powerful models to understand the evolution and ecology of symbiotic interactions. Metagenomic studies have previously described the functional features of sponge symbionts, however, little is known about the metabolic interactions and processes that occur under different environmental conditions. To address this issue, we construct here constraint-based, genome-scale metabolic networks for the microbiome of the sponge Stylissa sp. Our models define the importance of sponge-derived nutrients for microbiome stability and discover how different organic inputs can result in net heterotrophy or autotrophy of the symbiont community. The analysis further reveals the key role that a newly discovered bacterial taxon has in cross-feeding activities and how it dynamically adjusts with nutrient inputs. Our study reveals insights into the functioning of a sponge microbiome and provides a framework to further explore and define metabolic interactions in holobionts.

Essential genes are those whose presence is vital for a cell’s survival and growth. Detecting these genes in disease-causing organisms is critical for various biological studies, including understanding microbe metabolism, engineering genetically modified microorganisms, and identifying targets for treatment. When essential genes are expressed, they give rise to essential proteins. Identifying these genes, especially in complex organisms like Plasmodium falciparum, which causes malaria, is challenging due to the cost and time associated with experimental methods. Thus, computational approaches have emerged. Early research in this area prioritised the study of less intricate organisms, inadvertently neglecting the complexities of metabolite transport in metabolic networks. To overcome this, a Network-based Machine Learning framework was proposed. It assessed various network properties in Plasmodium falciparum, using a Genome-Scale Metabolic Model (iAM_Pf480) from the BiGG database and essentiality data from the Ogee database. The proposed approach substantially improved gene essentiality predictions as it considered the weighted and directed nature of metabolic networks and utilised network-based features, achieving a high accuracy rate of 0.85 and an AuROC of 0.7. Furthermore, this study enhanced the understanding of metabolic networks and their role in determining gene essentiality in Plasmodium falciparum. Notably, our model identified 9 genes previously considered non-essential in the Ogee database but now predicted to be essential, with some of them potentially serving as drug targets for malaria treatment, thereby opening exciting research avenues.

Analysis of metabolic networks through extreme rays such as extreme pathways and elementary flux modes has been shown to be effective for many applications. However, due to the combinatorial explosion of candidate vectors, their enumeration is currently limited to small- and medium-scale networks (typically <200 reactions). Partial enumeration of the extreme rays is shown to be possible, but either relies on generating them one-by-one or by implementing a sampling step in the enumeration algorithms. Sampling-based enumeration can be achieved through the canonical basis approach (CBA) or the nullspace approach (NSA). Both algorithms are very efficient in medium-scale networks, but struggle with elementarity testing in sampling-based enumeration of larger networks. In this paper, a novel elementarity test is defined and exploited, resulting in significant speedup of the enumeration. Even though NSA is currently considered more effective, the novel elementarity test allows CBA to significantly outpace NSA. This is shown through two case studies, ranging from a medium-scale network to a genome-scale metabolic network with over 600 reactions. In this study, extreme pathways are chosen as the extreme rays, but the novel elementarity test and CBA are equally applicable to the other types. With the increasing complexity of metabolic networks in recent years, CBA with the novel elementarity test shows even more promise as its advantages grows with increased network complexity. Given this scaling aspect, CBA is now the faster method for enumerating extreme rays in genome-scale metabolic networks. All case studies are implemented in Python. The codebase used to generate extreme pathways using the different approaches is available at https://gitlab.kuleuven.be/biotec-plus/pos-def-ep.

Reconstruction and analyses of genome-scale halomonas metabolic network yield a highly efficient PHA production