Genome-Scale Metabolic Modeling of the Human Microbiome in the Era of Personalized Medicine.

The human microbiome plays an important role in human health and disease. Meta-omics analyses provide indispensable data for linking changes in microbiome composition and function to disease etiology. Yet, the lack of a mechanistic understanding of, e.g., microbiome-metabolome links hampers the translation of these findings into effective, novel therapeutics. Here, we propose metabolic modeling of microbial communities through constraint-based reconstruction and analysis (COBRA) as a complementary approach to meta-omics analyses. First, we highlight the importance of microbial metabolism in cardiometabolic diseases, inflammatory bowel disease, colorectal cancer, Alzheimer disease, and Parkinson disease. Next, we demonstrate that microbial community modeling can stratify patients and controls, mechanistically link microbes with fecal metabolites altered in disease, and identify host pathways affected by the microbiome. Finally, we outline our vision for COBRA modeling combined with meta-omics analyses and multivariate statistical analyses to inform and guide clinical trials, yield testable hypotheses, and ultimately propose novel dietary and therapeutic interventions.

Systems biology elucidates the distinctive metabolic niche filled by the human gut microbe Eggerthella lenta.

Human gut bacteria perform diverse metabolic functions with consequences for host health. The prevalent and disease-linked Actinobacterium Eggerthella lenta performs several unusual chemical transformations, but it does not metabolize sugars and its core growth strategy remains unclear. To obtain a comprehensive view of the metabolic network of E. lenta, we generated several complementary resources: defined culture media, metabolomics profiles of strain isolates, and a curated genome-scale metabolic reconstruction. Stable isotope-resolved metabolomics revealed that E. lenta uses acetate as a key carbon source while catabolizing arginine to generate ATP, traits which could be recapitulated in silico by our updated metabolic model. We compared these in vitro findings with metabolite shifts observed in E. lenta-colonized gnotobiotic mice, identifying shared signatures across environments and highlighting catabolism of the host signaling metabolite agmatine as an alternative energy pathway. Together, our results elucidate a distinctive metabolic niche filled by E. lenta in the gut ecosystem. Our culture media formulations, atlas of metabolomics data, and genome-scale metabolic reconstructions form a freely available collection of resources to support further study of the biology of this prevalent gut bacterium.

Cardinality optimization in constraint-based modelling: application to human metabolism.

MOTIVATION: Several applications in constraint-based modelling can be mathematically formulated as cardinality optimization problems involving the minimization or maximization of the number of nonzeros in a vector. These problems include testing for stoichiometric consistency, testing for flux consistency, testing for thermodynamic flux consistency, computing sparse solutions to flux balance analysis problems and computing the minimum number of constraints to relax to render an infeasible flux balance analysis problem feasible. Such cardinality optimization problems are computationally complex, with no known polynomial time algorithms capable of returning an exact and globally optimal solution. RESULTS: By approximating the zero-norm with nonconvex continuous functions, we reformulate a set of cardinality optimization problems in constraint-based modelling into a difference of convex functions. We implemented and numerically tested novel algorithms that approximately solve the reformulated problems using a sequence of convex programs. We applied these algorithms to various biochemical networks and demonstrate that our algorithms match or outperform existing related approaches. In particular, we illustrate the efficiency and practical utility of our algorithms for cardinality optimization problems that arise when extracting a model ready for thermodynamic flux balance analysis given a human metabolic reconstruction. AVAILABILITY AND IMPLEMENTATION: Open source scripts to reproduce the results are here https://github.com/opencobra/COBRA.papers/2023_cardOpt with general purpose functions integrated within the COnstraint-Based Reconstruction and Analysis toolbox: https://github.com/opencobra/cobratoolbox.

Circulating metabolites modulated by diet are associated with depression.

Metabolome reflects the interplay of genome and exposome at molecular level and thus can provide deep insights into the pathogenesis of a complex disease like major depression. To identify metabolites associated with depression we performed a metabolome-wide association analysis in 13,596 participants from five European population-based cohorts characterized for depression, and circulating metabolites using ultra high-performance liquid chromatography/tandem accurate mass spectrometry (UHPLC/MS/MS) based Metabolon platform. We tested 806 metabolites covering a wide range of biochemical processes including those involved in lipid, amino-acid, energy, carbohydrate, xenobiotic and vitamin metabolism for their association with depression. In a conservative model adjusting for life style factors and cardiovascular and antidepressant medication use we identified 8 metabolites, including 6 novel, significantly associated with depression. In individuals with depression, increased levels of retinol (vitamin A), 1-palmitoyl-2-palmitoleoyl-GPC (16:0/16:1) (lecithin) and mannitol/sorbitol and lower levels of hippurate, 4-hydroxycoumarin, 2-aminooctanoate (alpha-aminocaprylic acid), 10-undecenoate (11:1n1) (undecylenic acid), 1-linoleoyl-GPA (18:2) (lysophosphatidic acid; LPA 18:2) are observed. These metabolites are either directly food derived or are products of host and gut microbial metabolism of food-derived products. Our Mendelian randomization analysis suggests that low hippurate levels may be in the causal pathway leading towards depression. Our findings highlight putative actionable targets for depression prevention that are easily modifiable through diet interventions.

Replication Protein A, the Main Eukaryotic Single-Stranded DNA Binding Protein, a Focal Point in Cellular DNA Metabolism.

Replication protein A (RPA) is a heterotrimeric protein complex and the main single-stranded DNA (ssDNA)-binding protein in eukaryotes. RPA has key functions in most of the DNA-associated metabolic pathways and DNA damage signalling. Its high affinity for ssDNA helps to stabilise ssDNA structures and protect the DNA sequence from nuclease attacks. RPA consists of multiple DNA-binding domains which are oligonucleotide/oligosaccharide-binding (OB)-folds that are responsible for DNA binding and interactions with proteins. These RPA-ssDNA and RPA-protein interactions are crucial for DNA replication, DNA repair, DNA damage signalling, and the conservation of the genetic information of cells. Proteins such as ATR use RPA to locate to regions of DNA damage for DNA damage signalling. The recruitment of nucleases and DNA exchange factors to sites of double-strand breaks are also an important RPA function to ensure effective DNA recombination to correct these DNA lesions. Due to its high affinity to ssDNA, RPA's removal from ssDNA is of central importance to allow these metabolic pathways to proceed, and processes to exchange RPA against downstream factors are established in all eukaryotes. These faceted and multi-layered functions of RPA as well as its role in a variety of human diseases will be discussed.

Microbiome Modelling Toolbox 2.0: efficient, tractable modelling of microbiome communities.

MOTIVATION: Constraint-Based Reconstruction and Analysis (COBRA) is a widely used approach for the interrogation and stratification of microbiome samples, yet applications to large-scale cohorts are hampered by limited scalability and efficiency of simulations. RESULTS: We substantially improved the computation speed and scalability of a previous implementation for the construction and interrogation of personalized constraint-based microbiome models as well as implemented additional functionalities for analysis and visualization. AVAILABILITY AND IMPLEMENTATION: Microbiome Modelling Toolbox and tutorials are freely available as part of the COBRA Toolbox at https://git.io/microbiomeModelingToolbox.

MetaboAnnotator: an efficient toolbox to annotate metabolites in genome-scale metabolic reconstructions.

MOTIVATION: Genome-scale metabolic reconstructions have been assembled for thousands of organisms using a wide range of tools. However, metabolite annotations, required to compare and link metabolites between reconstructions, remain incomplete. Here, we aim to further extend metabolite annotation coverage using various databases and chemoinformatic approaches. RESULTS: We developed a COBRA toolbox extension, deemed MetaboAnnotator, which facilitates the comprehensive annotation of metabolites with database independent and dependent identifiers, obtains molecular structure files, and calculates metabolite formula and charge at pH 7.2. The resulting metabolite annotations allow for subsequent cross-mapping between reconstructions and mapping of, e.g., metabolomic data. AVAILABILITY AND IMPLEMENTATION: MetaboAnnotator and tutorials are freely available at https://github.com/opencobra. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

A nutrition algorithm to optimize feed and medium composition using genome-scale metabolic models.

The optimization of animal feeds and cell culture media are problems of interest to a wide range of industries and scientific disciplines. Both problems are dictated by the properties of an organism's metabolism. However, due to the tremendous complexity of metabolic systems, it can be difficult to predict how metabolism will respond to changes in nutrient availability. A common tool used to capture the complexity of metabolism in a computational framework is a genome-scale metabolic model (GEM). GEMs are useful for predicting the fluxes of reactions within an organism's metabolism. To optimize feed or media, in silico experiments can be performed with GEMs by systematically varying nutritional constraints and predicting metabolic activity. In this way, the influence of various nutritional changes on metabolic outcomes can be evaluated. However, this methodology does not guarantee an optimal solution. Here, we develop a nutrition algorithm that utilizes linear programming to search the entire flux solution space of possible dietary intervention strategies to identify the most efficient changes to nutrition for a desirable metabolic outcome. We illustrate the utility of the nutrition algorithm on GEMs of Atlantic salmon (Salmo salar) and Chinese hamster ovary (CHO) cell metabolism and find that the nutrition algorithm makes predictions that not only align with experimental findings but reveal new insights into promising feeding strategies. We show that the nutrition algorithm is highly versatile and customizable to meet the user's needs. For instance, we demonstrate that the nutrition algorithm can be used to predict feed/media compositions that maximize profit margins. While the nutrition algorithm can be used to define an optimal feed/medium ab initio, it can also identify minimal changes to be made to an existing feed/medium to drive the largest metabolic shift. Moreover, the nutrition algorithm can target multiple metabolic pathways simultaneously with only a marginal increase in computational expense. While the nutrition algorithm has its limitations, we believe that this tool can be leveraged in a broad range of biotechnological applications to enhance the feed/medium optimization process.

DEMETER: efficient simultaneous curation of genome-scale reconstructions guided by experimental data and refined gene annotations.

MOTIVATION: Manual curation of genome-scale reconstructions is laborious, yet existing automated curation tools do not typically take species-specific experimental and curated genomic data into account. RESULTS: We developed Data-drivEn METabolic nEtwork Refinement (DEMETER), a Constraint-Based Reconstruction and Analysis (COBRA) Toolbox extension, which enables the efficient, simultaneous refinement of thousands of draft genome-scale reconstructions, while ensuring adherence to the quality standards in the field, agreement with available experimental data and refinement of pathways based on manually refined genome annotations. AVAILABILITY AND IMPLEMENTATION: DEMETER and tutorials are freely available at https://github.com/opencobra. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Integration of a physiologically-based pharmacokinetic model with a whole-body, organ-resolved genome-scale model for characterization of ethanol and acetaldehyde metabolism.

Ethanol is one of the most widely used recreational substances in the world and due to its ubiquitous use, ethanol abuse has been the cause of over 3.3 million deaths each year. In addition to its effects, ethanol's primary metabolite, acetaldehyde, is a carcinogen that can cause symptoms of facial flushing, headaches, and nausea. How strongly ethanol or acetaldehyde affects an individual depends highly on the genetic polymorphisms of certain genes. In particular, the genetic polymorphisms of mitochondrial aldehyde dehydrogenase, ALDH2, play a large role in the metabolism of acetaldehyde. Thus, it is important to characterize how genetic variations can lead to different exposures and responses to ethanol and acetaldehyde. While the pharmacokinetics of ethanol metabolism through alcohol dehydrogenase have been thoroughly explored in previous studies, in this paper, we combined a base physiologically-based pharmacokinetic (PBPK) model with a whole-body genome-scale model (WBM) to gain further insight into the effect of other less explored processes and genetic variations on ethanol metabolism. This combined model was fit to clinical data and used to show the effect of alcohol concentrations, organ damage, ALDH2 enzyme polymorphisms, and ALDH2-inhibiting drug disulfiram on ethanol and acetaldehyde exposure. Through estimating the reaction rates of auxiliary processes with dynamic Flux Balance Analysis, The PBPK-WBM was able to navigate around a lack of kinetic constants traditionally associated with PK modelling and demonstrate the compensatory effects of the body in response to decreased liver enzyme expression. Additionally, the model demonstrated that acetaldehyde exposure increased with higher dosages of disulfiram and decreased ALDH2 efficiency, and that moderate consumption rates of ethanol could lead to unexpected accumulations in acetaldehyde. This modelling framework combines the comprehensive steady-state analyses from genome-scale models with the dynamics of traditional PK models to create a highly personalized form of PBPK modelling that can push the boundaries of precision medicine.

Personalized whole-body models integrate metabolism, physiology, and the gut microbiome.

Comprehensive molecular-level models of human metabolism have been generated on a cellular level. However, models of whole-body metabolism have not been established as they require new methodological approaches to integrate molecular and physiological data. We developed a new metabolic network reconstruction approach that used organ-specific information from literature and omics data to generate two sex-specific whole-body metabolic (WBM) reconstructions. These reconstructions capture the metabolism of 26 organs and six blood cell types. Each WBM reconstruction represents whole-body organ-resolved metabolism with over 80,000 biochemical reactions in an anatomically and physiologically consistent manner. We parameterized the WBM reconstructions with physiological, dietary, and metabolomic data. The resulting WBM models could recapitulate known inter-organ metabolic cycles and energy use. We also illustrate that the WBM models can predict known biomarkers of inherited metabolic diseases in different biofluids. Predictions of basal metabolic rates, by WBM models personalized with physiological data, outperformed current phenomenological models. Finally, integrating microbiome data allowed the exploration of host-microbiome co-metabolism. Overall, the WBM reconstructions, and their derived computational models, represent an important step toward virtual physiological humans.

The Virtual Metabolic Human database: integrating human and gut microbiome metabolism with nutrition and disease.

A multitude of factors contribute to complex diseases and can be measured with 'omics' methods. Databases facilitate data interpretation for underlying mechanisms. Here, we describe the Virtual Metabolic Human (VMH, www.vmh.life) database encapsulating current knowledge of human metabolism within five interlinked resources 'Human metabolism', 'Gut microbiome', 'Disease', 'Nutrition', and 'ReconMaps'. The VMH captures 5180 unique metabolites, 17 730 unique reactions, 3695 human genes, 255 Mendelian diseases, 818 microbes, 632 685 microbial genes and 8790 food items. The VMH's unique features are (i) the hosting of the metabolic reconstructions of human and gut microbes amenable for metabolic modeling; (ii) seven human metabolic maps for data visualization; (iii) a nutrition designer; (iv) a user-friendly webpage and application-programming interface to access its content; (v) user feedback option for community engagement and (vi) the connection of its entities to 57 other web resources. The VMH represents a novel, interdisciplinary database for data interpretation and hypothesis generation to the biomedical community.

Parkinson's disease-associated alterations of the gut microbiome predict disease-relevant changes in metabolic functions.

BACKGROUND: Parkinson's disease (PD) is a systemic disease clinically defined by the degeneration of dopaminergic neurons in the brain. While alterations in the gut microbiome composition have been reported in PD, their functional consequences remain unclear. Herein, we addressed this question by an analysis of stool samples from the Luxembourg Parkinson's Study (n = 147 typical PD cases, n = 162 controls). RESULTS: All individuals underwent detailed clinical assessment, including neurological examinations and neuropsychological tests followed by self-reporting questionnaires. Stool samples from these individuals were first analysed by 16S rRNA gene sequencing. Second, we predicted the potential secretion for 129 microbial metabolites through personalised metabolic modelling using the microbiome data and genome-scale metabolic reconstructions of human gut microbes. Our key results include the following. Eight genera and seven species changed significantly in their relative abundances between PD patients and healthy controls. PD-associated microbial patterns statistically depended on sex, age, BMI, and constipation. Particularly, the relative abundances of Bilophila and Paraprevotella were significantly associated with the Hoehn and Yahr staging after controlling for the disease duration. Furthermore, personalised metabolic modelling of the gut microbiomes revealed PD-associated metabolic patterns in the predicted secretion potential of nine microbial metabolites in PD, including increased methionine and cysteinylglycine. The predicted microbial pantothenic acid production potential was linked to the presence of specific non-motor symptoms. CONCLUSION: Our results suggest that PD-associated alterations of the gut microbiome can translate into substantial functional differences affecting host metabolism and disease phenotype.

Early-Life Adversity Leaves Its Imprint on the Oral Microbiome for More Than 20 Years and Is Associated with Long-Term Immune Changes.

The early-life microbiome (ELM) interacts with the psychosocial environment, in particular during early-life adversity (ELA), defining life-long health trajectories. The ELM also plays a significant role in the maturation of the immune system. We hypothesised that, in this context, the resilience of the oral microbiomes, despite being composed of diverse and distinct communities, allows them to retain an imprint of the early environment. Using 16S amplicon sequencing on the EpiPath cohort, we demonstrate that ELA leaves an imprint on both the salivary and buccal oral microbiome 24 years after exposure to adversity. Furthermore, the changes in both communities were associated with increased activation, maturation, and senescence of both innate and adaptive immune cells, although the interaction was partly dependent on prior herpesviridae exposure and current smoking. Our data suggest the presence of multiple links between ELA, Immunosenescence, and cytotoxicity that occur through long-term changes in the microbiome.

Methanogenic granule growth and development is a continual process characterized by distinct morphological features.

Up-flow anaerobic bioreactors are widely applied for high-rate digestion of industrial wastewaters and rely on formation, and retention, of methanogenic granules, comprising of dense, fast-settling, microbial aggregates (approx. 0.5-4.0 mm in diameter). Granule formation (granulation) mechanisms have been reasonably well hypothesized and documented. However, this study used laboratory-scale bioreactors, inoculated with size-separated granular sludge to follow new granule formation, maturation, disintegration and re-formation. Temporal size profiles, volatile solids content, settling velocity, and ultrastructure of granules were determined from each of four bioreactors inoculated only with small granules, four with only large granules, and four with a full complement of naturally-size-distributed granules. Constrained granule size profiles shifted toward the natural distribution, which was associated with maximal bioreactor performance. Distinct morphological features characterized different granule sizes and biofilm development stages, including 'young', 'juvenile', 'mature' and 'old'. The findings offer opportunities toward optimizing management of high-rate, anaerobic digesters by shedding light on the rates of granule growth, the role of flocculent sludge in granulation and how shifting size distributions should be considered when setting upflow velocities.

The Microbiome Modeling Toolbox: from microbial interactions to personalized microbial communities.

MOTIVATION: The application of constraint-based modeling to functionally analyze metagenomic data has been limited so far, partially due to the absence of suitable toolboxes. RESULTS: To address this gap, we created a comprehensive toolbox to model (i) microbe-microbe and host-microbe metabolic interactions, and (ii) microbial communities using microbial genome-scale metabolic reconstructions and metagenomic data. The Microbiome Modeling Toolbox extends the functionality of the constraint-based reconstruction and analysis toolbox. AVAILABILITY AND IMPLEMENTATION: The Microbiome Modeling Toolbox and the tutorials at https://git.io/microbiomeModelingToolbox.

Predicting gastrointestinal drug effects using contextualized metabolic models.

Gastrointestinal side effects are among the most common classes of adverse reactions associated with orally absorbed drugs. These effects decrease patient compliance with the treatment and induce undesirable physiological effects. The prediction of drug action on the gut wall based on in vitro data solely can improve the safety of marketed drugs and first-in-human trials of new chemical entities. We used publicly available data of drug-induced gene expression changes to build drug-specific small intestine epithelial cell metabolic models. The combination of measured in vitro gene expression and in silico predicted metabolic rates in the gut wall was used as features for a multilabel support vector machine to predict the occurrence of side effects. We showed that combining local gut wall-specific metabolism with gene expression performs better than gene expression alone, which indicates the role of small intestine metabolism in the development of adverse reactions. Furthermore, we reclassified FDA-labeled drugs with respect to their genetic and metabolic profiles to show hidden similarities between seemingly different drugs. The linkage of xenobiotics to their transcriptomic and metabolic profiles could take pharmacology far beyond the usual indication-based classifications.

Dynamic genome-scale cell-specific metabolic models reveal novel inter-cellular and intra-cellular metabolic communications during ovarian follicle development.

BACKGROUND: The maturation of the female germ cell, the oocyte, requires the synthesis and storing of all the necessary metabolites to support multiple divisions after fertilization. Oocyte maturation is only possible in the presence of surrounding, diverse, and changing layers of somatic cells. Our understanding of metabolic interactions between the oocyte and somatic cells has been limited due to dynamic nature of ovarian follicle development, thus warranting a systems approach. RESULTS: Here, we developed a genome-scale metabolic model of the mouse ovarian follicle. This model was constructed using an updated mouse general metabolic model (Mouse Recon 2) and contains several key ovarian follicle development metabolic pathways. We used this model to characterize the changes in the metabolism of each follicular cell type (i.e., oocyte, granulosa cells, including cumulus and mural cells), during ovarian follicle development in vivo. Using this model, we predicted major metabolic pathways that are differentially active across multiple follicle stages. We identified a set of possible secreted and consumed metabolites that could potentially serve as biomarkers for monitoring follicle development, as well as metabolites for addition to in vitro culture media that support the growth and maturation of primordial follicles. CONCLUSIONS: Our systems approach to model follicle metabolism can guide future experimental studies to validate the model results and improve oocyte maturation approaches and support growth of primordial follicles in vitro.

DistributedFBA.jl: high-level, high-performance flux balance analysis in Julia.

Motivation: Flux balance analysis and its variants are widely used methods for predicting steady-state reaction rates in biochemical reaction networks. The exploration of high dimensional networks with such methods is currently hampered by software performance limitations. Results: DistributedFBA.jl is a high-level, high-performance, open-source implementation of flux balance analysis in Julia. It is tailored to solve multiple flux balance analyses on a subset or all the reactions of large and huge-scale networks, on any number of threads or nodes. Availability and Implementation: The code is freely available on github.com/opencobra/COBRA.jl. The documentation can be found at opencobra.github.io/COBRA.jl. Contact: ronan.mt.fleming@gmail.com. Supplementary information: Supplementary data are available at Bioinformatics online.

CHRR: coordinate hit-and-run with rounding for uniform sampling of constraint-based models.

SUMMARY: In constraint-based metabolic modelling, physical and biochemical constraints define a polyhedral convex set of feasible flux vectors. Uniform sampling of this set provides an unbiased characterization of the metabolic capabilities of a biochemical network. However, reliable uniform sampling of genome-scale biochemical networks is challenging due to their high dimensionality and inherent anisotropy. Here, we present an implementation of a new sampling algorithm, coordinate hit-and-run with rounding (CHRR). This algorithm is based on the provably efficient hit-and-run random walk and crucially uses a preprocessing step to round the anisotropic flux set. CHRR provably converges to a uniform stationary sampling distribution. We apply it to metabolic networks of increasing dimensionality. We show that it converges several times faster than a popular artificial centering hit-and-run algorithm, enabling reliable and tractable sampling of genome-scale biochemical networks. AVAILABILITY AND IMPLEMENTATION: https://github.com/opencobra/cobratoolbox . CONTACT: ronan.mt.fleming@gmail.com or vempala@cc.gatech.edu. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.