Genome-scale models of metabolism and gene expression extend and refine growth phenotype prediction.

13C metabolic flux analysis at a genome-scale

Genome-scale metabolic models (GMMs) have been recognized as being powerful tools for capturing system-wide metabolic phenomena and connecting those phenomena to underlying genetic and regulatory changes. By formalizing and codifying the relationship between the levels of gene expression, protein concentration, and reaction flux, metabolic models are able to translate changes in gene expression to their effects on the metabolic network. A number of methods are then available to interpret how those changes are manifest in the metabolic flux distribution. In addition to discussing how gene expression datasets can be interpreted in the context of a metabolic model, this chapter discusses two of the most common methods for analyzing the resulting metabolic network. The chapter begins by demonstrating how a typical microarray dataset can be processed for incorporation into a GMM of the yeast Saccharomyces cerevisiae. Once the expression states of the reactions in the model are available, the method of directly trimming the metabolic model by removing or constraining reactions with low expression states is demonstrated. This is the simplest and most direct approach to interpret gene expression states, but it is prone to overvaluing the effects of down regulation and it can propagate false negative errors. We therefore also include a more advanced method that uses a mixed-integer linear programming optimization to find a flux distribution that maximizes agreement with global gene expression states. Sample MATLAB code for use with the COBRA toolbox is provided for all methods used.

Designing metabolic engineering strategies with genome-scale metabolic flux modeling Jiun Y Yen,1,2 Imen Tanniche,1 Amanda K Fisher,1–3 Glenda E Gillaspy,2 David R Bevan,2,3 Ryan S Senger,1 1Department of Biological Systems Engineering, 2Department of Biochemistry, 3Genomics, Bioinformatics, and Computational Biology Interdisciplinary Program, Virginia Tech, Blacksburg, VA, USA Abstract: New in silico tools that make use of genome-scale metabolic flux modeling are improving the design of metabolic engineering strategies. This review highlights the latest developments in this area, explains the interface between these in silico tools and the experimental implementation tools of metabolic engineers, and provides a way forward so that in silico predictions can better mimic reality and more experimental methods can be considered in simulation studies. The several methodologies for solving genome-scale models (eg, flux balance analysis [FBA], parsimonious FBA, flux variability analysis, and minimization of metabolic adjustment) all have unique advantages and applications. There are two basic approaches to designing metabolic engineering strategies in silico, and both have demonstrated success in the literature. The first involves: 1) making a genetic manipulation in a model; 2) testing for improved performance through simulation; and 3) iterating the process. The second approach has been used in more recently designed in silico tools and involves: 1) comparing metabolic flux profiles of a wild-type and ideally engineered state and 2) designing engineering strategies based on the differences in these flux profiles. Improvements in genome-scale modeling are anticipated in areas such as the inclusion of all relevant cellular machinery, the ability to understand and anticipate the results of combinatorial enrichment experiments, and constructing dynamic and flexible biomass equations that can respond to environmental and genetic manipulations. Keywords: genome-scale modeling, genome-scale modeling, flux balance analysis, flux variability analysis, minimization of metabolic adjustment, metabolic bottleneck, pathway optimization

Pore-scale simulation of microbial growth using a genome-scale metabolic model: Implications for Darcy-scale reactive transport

BackgroundConstraint-based analysis of genome-scale metabolic models has become a key methodology to gain insights into functions, capabilities, and properties of cellular metabolism. Since their inception, the size and complexity of genome-scale metabolic reconstructions has significantly increased, with a concomitant increase in computational effort required for their analysis. Many stoichiometric methods cannot be applied to large networks comprising several thousand reactions. Furthermore, basic principles of an organism’s metabolism can sometimes be easier studied in smaller models focusing on central metabolism. Therefore, an automated and unbiased reduction procedure delivering meaningful core networks from well-curated genome-scale reconstructions is highly desirable.ResultsHere we present NetworkReducer, a new algorithm for an automated reduction of metabolic reconstructions to obtain smaller models capturing the central metabolism or other metabolic modules of interest. The algorithm takes as input a network model and a list of protected elements and functions (phenotypes) and applies a pruning step followed by an optional compression step. Network pruning removes elements of the network that are dispensable for the protected functions and delivers a subnetwork of the full system. Loss-free network compression further reduces the network size but not the complexity (dimension) of the solution space. As a proof of concept, we applied NetworkReducer to the iAF1260 genome-scale model of Escherichia coli (2384 reactions, 1669 internal metabolites) to obtain a reduced model that (i) allows the same maximal growth rates under aerobic and anaerobic conditions as in the full model, and (ii) preserves a protected set of reactions representing the central carbon metabolism. The reduced representation comprises 85 metabolites and 105 reactions which we compare to a manually derived E. coli core model. As one particular strength of our approach, NetworkReducer derives a condensed biomass synthesis reaction that is consistent with the full genome-scale model. In a second case study, we reduced a genome-scale model of the cyanobacterium Synechocystis sp. PCC 6803 to obtain a small metabolic module comprising photosynthetic core reactions and the Calvin-Benson cycle allowing synthesis of both biomass and a biofuel (ethanol).ConclusionAlthough only genome-scale models provide a complete description of an organism’s metabolic capabilities, an unbiased stoichiometric reduction of large-scale metabolic models is highly useful. We are confident that the NetworkReducer algorithm provides a valuable tool for the application of computationally expensive analyses, for educational purposes, as well to identify core models for kinetic modeling and isotopic tracer experiments.Electronic supplementary materialThe online version of this article (doi:10.1186/s12918-015-0191-x) contains supplementary material, which is available to authorized users.

Marine sponges are increasingly being recognised for their nutrient cycling ecosystem services, linking pelagic nutrients with the benthic ecosystem. Furthermore, sponges are the most prolific producers of bioactive secondary metabolites in the marine environment. Despite their ecological and commercial value, little is know about the metabolic processes that are responsible. The inability to produce sufficient sponge biomass, and thus specific bioactive compounds, has been repeatedly identified as a major bottleneck towards commercialising sponge holobiont derived drugs. Likewise, nutrient cycling by sponges has been a relatively recent advancement in marine ecology and the scale of this process is unknown. This thesis takes a systems approach to understanding sponge-symbiont metabolic processes by utilising the genomic resources available for the demosponge Amphimedon queenslandica and its bacterial symbiont AqS1. Specifically, I undertake genome-scale modelling and metabolic flux analyses to investigate the metabolic networks present in this sponge as well as its associated vertically-transmitted microbial symbiont. The goal of this thesis is to generate the data required for, and subsequently develop, a dual-species genome-scale metabolic model for A. queenslandica and AqS1. This will provide a framework to further our understanding of how sponges produce their biomass while living in an oligotrophic, tropical reef environment. To develop a genome-scale metabolic model, the biochemical composition of the organism must first be determined. Knowledge of the relative quantities of macromolecules and their respective building blocks (e.g. protein and amino acids) is vital, as each requires different substrates, enzymes and cofactors for their synthesis. I adapted and developed methods to characterise the composition and abundance of DNA, RNA, protein, lipids and carbohydrates in a marine sponge. These methods are described in detail that allows them to be easily transferred to other, non-model, large marine organisms. This is followed by a detailed analysis of A. queenslandica’s macromolecular, amino acid, fatty acid and sterol composition. The biochemical data from this chapter was used to generate a biomass equation that represent the average composition of adult A. queenslandica in the metabolic model. To understand how the metabolic network may work towards producing more biomass, it must be considered in the context of the environmental conditions that naturally constrain growth. The dominant environmental constraint on growth on an oligotrophic reef system is nutrient availability. The abundance of key elements, such as carbon and nitrogen, were quantified throughout the course of a year. To investigate effects on the sponge of any changes in nutrient availability, I concurrently sampled sponges and analysed their biochemical composition to the macromolecular level. Chapter 4 presents these data and discusses a number of trends and correlations in the biochemical composition of the sponges with changing nutrient availability through four seasons of the year. For instance, there was a significant increase in particulate to dissolved organic carbon at the end of summer with a corresponding rise in carbohydrates in A. queenslandica. Of particular importance for the subsequent metabolic modelling, was the separation of carbon into the dissolved and particulate fractions. Chapter 5 presents the metabolic models and their analysis. Initially, I manually constructed a genome-scale metabolic model for the sponge A. queenslandica. The model construction identified 10 amino acids, 4 vitamins and a plant-derived phytosterol for which A. queenslandica is auxotrophic. These represent nutrients that are essential for sponge growth and may in future form the basis of a defined cell culture medium. The microbial community within A. queenslandica is relatively simple and dominated by a species of sulfur oxidising bacteria, called Aqs1. I generated a genome-scale model for Aqs1, which is able to synthesise all 20 proteinogenic amino acids, in addition to having a diverse range of carbohydrate-specific transporters and enzymes. To investigate the interactions between the host sponge and Aqs1, the two models were joined using a shared compartment. This was called the extracellular matrix, and represents the area of interaction within the sponge body where metabolite transfers can occur. I measured the pumping rate of A. queenslandica and calculated the average volume of water pumped per hour, standardised to gram dry weight. This was used to constrain the nutrient uptake rate of the model. To investigate how the metabolic network may respond to different nutrient conditions, low and high nutrient conditions were defined using the ratios of particulate and dissolved carbon from the seasonal environmental profiling. This work represents the first genome-scale model for a sponge-symbiont system, and marine invertebrates in general. The genome-scale metabolic models resulting from this work are an important resource that will guide future work into the metabolic processes of both A. queenslandica and its symbiont, Aqs1.

Genome-scale metabolic network models represent the link between the genotype and phenotype of the organism, which are usually reconstructed based on the genome sequence annotation and relevant biochemical and physiological information. These models provide a holistic view of the organism's metabolism, and constraint-based metabolic flux analysis methods have been used extensively to study genome-scale cellular metabolic networks. It is clear that the quality of the metabolic network model determines the outcome of the application. Therefore, it is critically important to determine the accuracy of a genome-scale model in describing the cellular metabolism of the modeled strain. However, because of the model complexity, which results in a system with very high degree of freedom, a good agreement between measured and computed substrate uptake rates and product secretion rates is not sufficient to guarantee the predictive capability of the model. To address this challenge, in this work we present a novel system identification based framework to extract the qualitative biological knowledge embedded in the quantitative simulation results from the metabolic network models. The extracted knowledge can serve two purposes: model validation during model development phase, which is the focus of this work, and knowledge discovery once the model is validated. This framework bridges the gap between the large amount of numerical results generated from genome-scale models and the knowledge that can be easily understood by biologists. The effectiveness of the proposed framework is demonstrated by its application to the analysis of two recently published genome-scale models of Scheffersomyces stipitis.

Genome-scale models of metabolism are becoming increasingly important to understanding the relationship between genotype and phenotype in an organism on a systems level. There are many tools being developed which rely on a genome-scale metabolic reconstruction as input in order to suggest engineering interventions to improve or modify a strain's metabolism. As more organisms are being sequenced, the information available to reconstruct these models also increases. There are relatively few genome-scale models compared to available genome-scale reconstructions due to the difficulty and time required to create one. In a metabolic engineering context, a genome-scale model can be used to predict engineering interventions that will produce a desired change in an organism's metabolism. Such efforts often consist of iterative small engineering changes to an organism which must be individually analyzed and interpreted, and often updated with the results of analysis on a previous strain. Existing tools for semi-automated genome-scale model generation do not address the issue of updating existing genome-scale models to accommodate new data. A common practice in databases representing metabolic reactions is to represent all possible substrates which are compatible with an enzyme as a single generic metabolite representing the class of substrates which bind to the enzyme. These reactions are referred to as generic reactions, and are not suitable for use in genome-scale modeling, which requires only exact metabolite species to be represented. We have developed a new software for generating genome-scale metabolic models. This software facilitates modification of a base version of a Pathway Genome Database (PGDB) to align it with knowledge of a developed strain. It allows a group to maintain customized data content in an existing BioCyc database, while still being able to integrate newly released updates in a semi-automated fashion. Changes made to the engineered strain also need to be added to any metabolic models of that strain for use in constraint-based analysis. This software can be used to generate metabolic models from a strain specific database in a semi-automated process.

Chinese hamster ovary (CHO) cells are the most commonly used cell lines in biopharmaceutical manufacturing. Genome-scale metabolic models have become a valuable tool to study cellular metabolism. Despite the presence of reference global genome-scale CHO model, context-specific metabolic models may still be required for specific cell lines (for example, CHO-K1, CHO-S, and CHO-DG44), and for specific process conditions. Many integration algorithms have been available to reconstruct specific genome-scale models. These methods are mainly based on integrating omics data (i.e., transcriptomics, proteomics, and metabolomics) into reference genome-scale models. In the present study, we aimed to investigate the impact of time points of transcriptomics integration on the genome-scale CHO model by assessing the prediction of growth rates with each reconstructed model. We also evaluated the feasibility of applying extracted models to different cell lines (generated from the same parental cell line). Our findings illustrate that gene expression at various stages of culture slightly impacts the reconstructed models. However, the prediction capability is robust enough on cell growth prediction not only across different growth phases but also in expansion to other cell lines.

BackgroundThe availability of genome sequences for many organisms enabled the reconstruction of several genome-scale metabolic network models. Currently, significant efforts are put into the automated reconstruction of such models. For this, several computational tools have been developed that particularly assist in identifying and compiling the organism-specific lists of metabolic reactions. In contrast, the last step of the model reconstruction process, which is the definition of the thermodynamic constraints in terms of reaction directionalities, still needs to be done manually. No computational method exists that allows for an automated and systematic assignment of reaction directions in genome-scale models.ResultsWe present an algorithm that – based on thermodynamics, network topology and heuristic rules – automatically assigns reaction directions in metabolic models such that the reaction network is thermodynamically feasible with respect to the production of energy equivalents. It first exploits all available experimentally derived Gibbs energies of formation to identify irreversible reactions. As these thermodynamic data are not available for all metabolites, in a next step, further reaction directions are assigned on the basis of network topology considerations and thermodynamics-based heuristic rules. Briefly, the algorithm identifies reaction subsets from the metabolic network that are able to convert low-energy co-substrates into their high-energy counterparts and thus net produce energy. Our algorithm aims at disabling such thermodynamically infeasible cyclic operation of reaction subnetworks by assigning reaction directions based on a set of thermodynamics-derived heuristic rules. We demonstrate our algorithm on a genome-scale metabolic model of E. coli. The introduced systematic direction assignment yielded 130 irreversible reactions (out of 920 total reactions), which corresponds to about 70% of all irreversible reactions that are required to disable thermodynamically infeasible energy production.ConclusionAlthough not being fully comprehensive, our algorithm for systematic reaction direction assignment could define a significant number of irreversible reactions automatically with low computational effort. We envision that the presented algorithm is a valuable part of a computational framework that assists the automated reconstruction of genome-scale metabolic models.

BackgroundHuman tissues perform diverse metabolic functions. Mapping out these tissue-specific functions in genome-scale models will advance our understanding of the metabolic basis of various physiological and pathological processes. The global knowledgebase of metabolic functions categorized for the human genome (Human Recon 1) coupled with abundant high-throughput data now makes possible the reconstruction of tissue-specific metabolic models. However, the number of available tissue-specific models remains incomplete compared with the large diversity of human tissues.ResultsWe developed a method called metabolic Context-specificity Assessed by Deterministic Reaction Evaluation (mCADRE). mCADRE is able to infer a tissue-specific network based on gene expression data and metabolic network topology, along with evaluation of functional capabilities during model building. mCADRE produces models with similar or better functionality and achieves dramatic computational speed up over existing methods. Using our method, we reconstructed draft genome-scale metabolic models for 126 human tissue and cell types. Among these, there are models for 26 tumor tissues along with their normal counterparts, and 30 different brain tissues. We performed pathway-level analyses of this large collection of tissue-specific models and identified the eicosanoid metabolic pathway, especially reactions catalyzing the production of leukotrienes from arachidnoic acid, as potential drug targets that selectively affect tumor tissues.ConclusionsThis large collection of 126 genome-scale draft metabolic models provides a useful resource for studying the metabolic basis for a variety of human diseases across many tissues. The functionality of the resulting models and the fast computational speed of the mCADRE algorithm make it a useful tool to build and update tissue-specific metabolic models.

Application of a curated genome-scale metabolic model of CHO DG44 to an industrial fed-batch process

Full text Figures and data Side by side Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Metabolic scaling, the inverse correlation of metabolic rates to body mass, has been appreciated for more than 80 years. Studies of metabolic scaling have largely been restricted to mathematical modeling of caloric intake and oxygen consumption, and mostly rely on computational modeling. The possibility that other metabolic processes scale with body size has not been comprehensively studied. To address this gap in knowledge, we employed a systems approach including transcriptomics, proteomics, and measurement of in vitro and in vivo metabolic fluxes. Gene expression in livers of five species spanning a 30,000-fold range in mass revealed differential expression according to body mass of genes related to cytosolic and mitochondrial metabolic processes, and to detoxication of oxidative damage. To determine whether flux through key metabolic pathways is ordered inversely to body size, we applied stable isotope tracer methodology to study multiple cellular compartments, tissues, and species. Comparing C57BL/6 J mice with Sprague-Dawley rats, we demonstrate that while ordering of metabolic fluxes is not observed in in vitro cell-autonomous settings, it is present in liver slices and in vivo. Together, these data reveal that metabolic scaling extends beyond oxygen consumption to other aspects of metabolism, and is regulated at the level of gene and protein expression, enzyme activity, and substrate supply. Editor's evaluation Key metabolic processes have been shown to scale inversely with the body mass of different animals. This study provides direct evidence for metabolic scaling of key metabolic fluxes in the livers of mice and rats, as well as species-specific differences in the transcription and expression of enzymes involved in energy metabolism that could contribute to metabolic scaling. The finding suggests that metabolic scaling likely reflects multiple levels of regulation and have broad implications for studying animal metabolism and physiology. https://doi.org/10.7554/eLife.78335.sa0 Decision letter Reviews on Sciety eLife's review process Introduction In 1932, Max Kleiber published a seminal study (Kleiber, 1932), integrating prior reports demonstrating a phenomenon that came to be termed 'Kleiber's law,' or the principle of metabolic scaling. Metabolic scaling refers to the phenomenon that the metabolic processes in many animals, if not all, scale inversely to three-quarters of their body mass (West et al., 1997). In simpler terms, there is a reduction in metabolic rate as body size increases. For example, an elephant is 25 million times larger than a fruit fly, yet its energy expenditure is only 20 thousand times higher; thus, from the fruit fly to elephant, the metabolic rate per gram of body weight scales down 1250 times. While there is experimental evidence for metabolic scaling from bacteria to large mammals, data have been generated almost exclusively from observations of caloric intake and oxygen consumption, with gene and protein expression, and substrate fluxes almost entirely unexplored. The concept of hierarchical regulation, whereby gene expression initiates the cascade that allows for the flux of metabolic pathways (Rossell et al., 2005; Suarez and Moyes, 2012), provides a systems framework to begin to understand scaling. Beginning at the transcriptional level, we studied liver gene expression across five species: mice (Mus musculus), rats (Rattus norvegicus), monkeys (Macaca mulatta), humans (Homo sapiens), and cattle (bos taurus), species with a 30,000-fold range of average body weight in adults (from 30 g in mice, to 900 kg in cattle). Numerous metabolic genes related to glycolysis, gluconeogenesis, fatty acid metabolism, oxygen consumption, electron transport, and redox function, and detoxification of oxidative damage, were expressed at levels inverse to body size. Further analysis of liver proteomics revealed that approximately half of the genes in the liver that were expressed inversely proportionally to body size at the transcriptional level, were also expressed at levels inversely proportional to body size at the level of protein expression. To determine if gene and protein expression would correlate with enzyme activity and metabolic flux, we performed a comprehensive assessment of liver metabolism in vivo and in vitro using modified Positional Isotopomer NMR Tracer Analysis (PINTA) (Perry et al., 2017b) and stable isotope-derived turnover (Perry et al., 2015) methods. Our analysis shows that rats exhibit lower metabolic rates when compared to mice in and ex vivo; however, no significant differences were observed when we isolated hepatocytes and cultured them in vitro under identical conditions. Taken together, this study demonstrates the variation of metabolic fluxes according to body size, extending prior studies of metabolic scaling, and provides unique insight into the regulation of metabolic flux across species. Results Genes within the liver that are expressed inversely proportional to body weight are predominantly metabolic genes We examined gene expression in livers from mice (Mus musculus), rats (Rattus norvegicus), monkeys (Macaca mulatta), humans (Homo sapiens), and cattle (Bos taurus). Using recent advances in high throughput mRNA sequencing and bioinformatics tools that allow for intra-species data preprocessing (Bray et al., 2016; Conesa et al., 2016; Ritchie et al., 2015), we searched for a set of genes in the liver, the metabolic hub of mammals, whose expression correlates inversely with body mass. After normalizing for differences in transcript length and abundance across species, we filtered out genes that followed the pattern of mouse >rat > monkey >human > cow. The genes that met these criteria were predominantly related to metabolic pathways, including pyruvate metabolism, amino acid metabolism, and glucose metabolism (Figure 1A). Genes from this list were further restricted to genes involved in amino acid, carbohydrate, energy, lipid, vitamin, and TCA cycle metabolism, and demonstrated a range of degrees of inverse correlation with body mass, with only TCA cycle genes clustering together (Figure 1B). Figure 1 Download asset Open asset Genes that follow the pattern of allometric scaling are most strongly related to metabolism. (A) KEGG Pathway enrichment of all genes that are expressed with an inverse correlation to body mass, and (B) clustering heatmap of scaled genes that belong to one of six Reactome metabolic superpathways. All samples were obtained from males. For clarity, the human gene (and style of writing human gene names) are shown. RAPGEF, rap guanine nucleotide exchange factor; ELOVL2, Elongation of Very Long Chain Fatty Acids-Like 2; MDH1, malate dehydrogenase 1; LIPE, hormone-sensitive lipase E; PANK1, pantothenate kinase 1; PGK1, phosphoglycerate kinase 1; SDC4, syndecan 4; ALDH7A1, aldehyde dehydrogenase 7 family member A1; GPX1, glutathione peroxidase 1; GLUL, glutamate-ammonia ligase; SORD, sorbitol dehydrogenase; TDO2, tryptophan 2,3-dioxygenase; DLST, dihydrolipoamide S-succinyltransferase; ACACA, acetyl-CoA carboxylase-alpha; ADIPOR1, adiponectin receptor-1; GPT, glutamic-pyruvate transaminase; HS3ST3B1, heparan sulfate-glucosamine 3-sulfotransferase 3B1; PSMD5, proteasome 26 S subunit, non-ATPase-5; COX8A, cytochrome c oxidase subunit 8 A; NDUFA9, NADH:ubiquinone oxidoreductase subunit A9. Genes encoding enzymes involved in hepatic metabolism are expressed inversely proportionally to body mass and involve metabolite detoxification, intertissue metabolism, substrate metabolism, electron transport, and NAD metabolism In order to further understand the functional aspects of the metabolic genes that are expressed inversely proportionally to body size, the gene list from Figure 1B was categorized into several functional categories, converging on optimizing energy provision, oxidative metabolism, and damage control from oxidative stress and ammonia (Figure 2). Furthermore, to understand whether or not certain genes that are expressed inversely proportionally to body size involved anabolic or catabolic processes, they were further classified by their properties to be energy suppliers or consumers. Eleven of sixteen critical metabolic enzymes that scaled required molecular oxygen, NAD+/NADH, or ATP/ADP for function, possibly indicating exquisite regulation of energy-consuming processes at the individual gene level. Genes involved in the detoxication of lipid peroxidation-derived aldehydes (ALDH7A1), hydrogen peroxide (GPX1), and ammonia (GLUL) suggest scaling of damage control mechanisms that are associated with increased oxidative metabolism across species (Figure 2A). The inverse correlation between body size and expression of genes that are associated with interorgan crosstalk is consistent with scaling in vivo which would not be expected in plated cells. For example, the differentially expressed genes include GPT1, which is involved in recycling skeletal muscle-derived alanine back to liver-derived glucose (Felig and Wahren, 1971; Petersen et al., 2019), and the adiponectin receptor (ADIPOR1), which binds an adipose tissue-derived hormone that regulates gluconeogenesis and fatty acid oxidation (Lin and Accili, 2011; Li et al., 2020 Figure 2B). Genes involved in fatty acid metabolism included the rate-limiting steps of the synthesis of CoA (PANK1), of de novo fatty acid synthesis (ACACA), and of fatty acid elongation (ELOVL2), in addition to the oxidation of diacylglycerols (LIPE) (Figure 2C). NAD and ATP-dependent genes involved in glycolysis (PDK1), fructose/glucose metabolism (SORD1), and DLST of the TCA cycle also correlated inversely with body size (Figure 2D–E). Differentially regulated genes also couple oxygen consumption to NAD provision (MDH1, TDO2), and are involved with the function of the electron transport chain (subunits of complex I, NDUFA9, and complex IV, COX8A, which catalyzes oxygen accepting the final electrons of the electron transport chain) (Figure 2F–G). Figure 2 with 1 supplement see all Download asset Open asset Metabolic genes that are expressed inversely proportionally to body size implicate key pathways in substrate and nucleotide supply, glucose and fatty acid flux, oxygen consumption, and detoxification pathways. mRNA expression of key regulatory genes related to metabolite detoxication (A), intertissue metabolism (B), fatty acid metabolism (C), glucose metabolism (D), tricarboxylic acid (TCA) cycle, NAD metabolism (F), and the electron transport chain (G) in mice, rats, monkeys, humans, and cattle. Bars denote expression levels by an organism, following the same order shown in the cartoon of organisms. Expression was normalized to counts per million and was then further normalized for sequencing depth and transcript length. All genes met an adjusted p-value threshold of 0.01 using a one-way ANOVA with the Bonferroni correction for multiple comparisons. All samples were obtained from males (n=2 replicates per species). ALDH7A1, aldehyde dehydrogenase 7 family member A1; GPX1, glutathione peroxidase 1; GLUL, glutamate-ammonia ligase; GPT, glutamic-pyruvate transaminase; ADIPOR1, adiponectin receptor-1; LIPE, hormone-sensitive lipase E; PANK1, pantothenate kinase 1; ACACA, acetyl-CoA carboxylase-alpha; ELOVL2, Elongation of Very Long Chain Fatty Acids-Like 2; SORD, sorbitol dehydrogenase; PGK1, phosphoglycerate kinase 1; DLST, dihydrolipoamide S-succinyltransferase; TDO2, tryptophan 2,3-dioxygenase; MDH1, malate dehydrogenase 1; NDUFA9, NADH:ubiquinone oxidoreductase subunit A9; COX8A, cytochrome c oxidase subunit 8 A. Figure 2—source data 1 Source data for Figure 2 and Figure 2—figure supplement 1. https://cdn.elifesciences.org/articles/78335/elife-78335-fig2-data1-v1.xlsx Download elife-78335-fig2-data1-v1.xlsx To examine the possibility that the inverse correlation between body mass and gene expression observed in the transcriptomics analysis could be a consequence of global alterations in mRNA (for example, as a consequence of alterations in RNA turnover rates), we performed targeted quantitative polymerase chain reaction (qPCR), measuring in liver tissue abundance of mRNA encoding several enzymes that were found to scale in the five-species transcriptomics analysis, relative to the common housekeeping gene β-actin (Actb). We found that all three enzymes (Glul, Lipe, and Dlst) scaled relative to Actb (Figure 2—figure supplement 1A–C), whereas structural genes (collagenase 3 [Mmp3] and Larp1) did not (Figure 2—figure supplement 1D–E), indicating that the differences in metabolic gene expression observed across species is likely not a result of global changes in RNA levels. In addition to transcriptomics, we assessed proteomics data to evaluate the protein levels corresponding to the genes that were found to be expressed inversely proportionally to body size at the level of mRNA expression. Our proteomics data were limited to mouse, rat, and human, as all the open-source proteomic databases that we identified lacked data from monkey or cow. An important limitation for finding such data is that even with careful post-processing, we cannot combine data from different studies, because differences in methods of tissue preparation may influence results. Therefore, we were limited to a single experiment that had generated proteomics data for mouse, rat, and human using the same experimental procedures. The dataset contained protein expression corresponding to eight of the twenty genes identified to scale in our transcriptomics data analysis. Of these, three (GLUL, GPX1, and MDH1) were found to follow a reverse correlation with body size (Figure 3A–C). Interestingly, one of these proteins (GLUL) was also found to be expressed inversely proportionally to body size in the left ventricle of the heart (Figure 3D). Additionally, we measured liver transaminase concentrations and observed that both alanine aminotransferase (ALT) and aspartate aminotransferase (AST) exhibited lower concentrations in humans as compared to rats and rats as compared to mice (Figure 3E–F), consistent with scaling at the level of protein expression as well as mRNA expression. Finally, we utilized established enzymatic assays to measure the activity of peroxidase and pyruvate carboxylase in the livers of mice and rats. 30–40% lower activity of each enzyme per mg tissue was observed in rats as compared to mice (Figure 3—figure supplement 1A–B), suggesting scaling at the level of metabolic enzyme activity. Figure 3 with 1 supplement see all Download asset Open asset Proteomics reveals a negative correlation between body size and the expression of some liver proteins. Liver (A) glutamate-ammonia ligase (GLUL), (B) glutathione peroxidase 1 (GPX1), and (C) malate dehydrogenase 1 (MDH1) protein expression. (D) GLUL protein expression in the left ventricle of the heart. The proteomics analysis was performed on n=1 per species, so statistical comparisons were not possible. (E) Plasma alanine aminotransferase (ALT) and (F) aspartate aminotransferase (AST) concentrations (for both transaminases, n=5 per species). *p<0.05, ***p<0.001, ****p<0.0001. Figure 3—source data 1 Source data for Figure 3 and Figure 3—figure supplement 1. https://cdn.elifesciences.org/articles/78335/elife-78335-fig3-data1-v1.xlsx Download elife-78335-fig3-data1-v1.xlsx Metabolic rates of mouse vs. rat hepatocytes in vitro are not significantly different Considering prior data reporting higher oxygen consumption per unit body mass in smaller as compared to larger animals (Gilman et al., 2013; Brody, 1945; Urbina and Glover, 2013), we first asked whether these differences were cell-intrinsic, or whether in vivo or hepatocyte-extrinsic signals are required. We incubated plated hepatocytes in [3-13C] lactate and first validated that the data met the assumptions of PINTA, including reaching steady-state in [13C] lactate and glucose enrichment, and producing glucose at a linear rate throughout the 6 hr incubation (Figure 4—figure supplement 1A–C). Consistent with the possibility that hepatocyte-extrinsic signals are primarily responsible for metabolic scaling, when we used PINTA to assess cytosolic and mitochondrial fluxes, we observed no significant differences between species in any of the fluxes measured in plated hepatocytes: glucose production, VPC, VCS, the contribution of glucose or fatty acids to the tricarboxylic acid (TCA) cycle, or lipolysis (Figure 4A–H, Figure 4—figure supplement 1D–F). Similarly, a mitochondrial stress test in plated hepatocytes revealed no difference in any parameter: neither basal mitochondrial and non-mitochondrial respiration, ATP production, maximal (uncoupled) respiration, spare respiratory capacity, nor proton leak differed between plated hepatocytes from mice and rats (Figure 4I). Previous studies have demonstrated scaling in vitro in cell suspensions only when analyzed immediately after hepatocyte isolation (Porter and Brand, 1995), and have suggested that the phenomenon of scaling gradually disappears around 24 hr post removal (Brown et al., 2007), similar to the conditions in which we performed these studies. Most prior in vitro studies have also demonstrated an absence of scaling, in contrast to in vivo (Glazier, 2015), and we extend these results to gluconeogenic and lipolytic fluxes in hepatocytes, glucose production in liver slices, and multimodal flux analysis in vivo. Figure 4 with 1 supplement see all Download asset Open asset Metabolic fluxes are not different between mouse and rat hepatocytes in vitro. (A) Study design. This figure was made using Biorender.com. (B) Tracer labeling strategy. (C) Glucose production. (D) Gluconeogenesis from pyruvate (pyruvate carboxylase flux, VPC). (E) Citrate synthase flux (VCS), i.e., mitochondrial oxidation. (F) Pyruvate dehydrogenase flux (VPDH), i.e., the contribution of glucose via glycolysis to total mitochondrial oxidation. (G) Non-esterified fatty acid (NEFA) production. (H) The contribution of fatty acid oxidation to citrate synthase flux. (I) Oxygen consumption rate (OCR) during a mitochondrial stress test. In all panels, hepatocytes from wild-type males were studied, and groups were compared using the two-tailed unpaired Student's t-test. No significant differences were observed. In all panels, the mean ± SEM. of six biological replicates (averaged from three technical replicates per biological replicate) is shown. Figure 4—source data 1 Source data for Figure 4 and Figure 4—figure supplement 1. https://cdn.elifesciences.org/articles/78335/elife-78335-fig4-data1-v1.xlsx Download elife-78335-fig4-data1-v1.xlsx Glucose production per gram tissue is higher ex vivo in liver slices from mice than in rats Next, considering that hepatocytes comprise approximately 70–80% of liver mass and that their culture in vitro does not replicate in vivo conditions (Krebs, 1950), we asked whether glucose production would be different between mice and rats in slices of liver. Indeed, we found that liver glucose production per gram liver mass was threefold greater in mouse liver slices as compared to rats (Figure 5A–B), suggesting that hepatocyte-extrinsic signals (for example, from other liver cell types) are involved in liver metabolic scaling. Figure 5 Download asset Open asset Glucose production scales ex vivo in liver slices. (A) Study design. This figure was made using Biorender.com. (B) Glucose production. Groups were compared by the two-tailed unpaired Student's t-test. Liver slices from male, wild-type animals (n=4 mice and 2 rats, three technical replicates per biological replicate) were studied. Figure 5—source data 1 Source data for Figure 5. https://cdn.elifesciences.org/articles/78335/elife-78335-fig5-data1-v1.xlsx Download elife-78335-fig5-data1-v1.xlsx Metabolic rates in multiple tissue types are higher in vivo in mice relative to rats We utilized multimodal stable isotope metabolic flux analysis to compare rats and mice with respect to a panel of metabolic fluxes (Figure 6A). First, we validated tracer assumptions in vivo, including the metabolic and isotopic steady state in plasma and negligible liver glycogen concentrations, although in the recently hepatic was higher in mice than that in rats (Figure supplement Using PINTA, we found that both glucose production and gluconeogenesis from pyruvate per gram liver were more than higher in mice than rats (Figure although the contribution of pyruvate to gluconeogenesis did not between mice and rats (Figure supplement oxidation scaled threefold in mice as compared to rats studied under the same to in both glucose oxidation (pyruvate dehydrogenase flux, and fatty acid oxidation (Figure associated with an in the of pyruvate carboxylase to citrate synthase flux any difference in the of flux by glucose through (Figure supplement While we did not have the to measure liver fluxes in larger in the glucose production, VPC, and measured using PINTA were lower in humans than in rats et al., 2019), a liver size of g in differences in metabolic fluxes according to body size applied not only to liver metabolism also to adipose tissue fatty acid was higher in mice than in rats (Figure No differences were observed in any of the measured fluxes (Figure supplement Taken together, these data the of common in vitro methods as a of in vivo whereas in vivo mitochondrial oxidation cycle was threefold higher in mice than in rats, in vitro of oxygen consumption throughout a mitochondrial stress TCA cycle flux, and glucose production were not different between the species (Figure Figure 6 with 1 supplement see all Download asset Open asset Analysis of metabolic fluxes suggests in vivo metabolic scaling in mice vs. rats. (A) Study design. (B) glucose production. (C) Gluconeogenesis from pyruvate (D) VCS, i.e., mitochondrial oxidation. (E) i.e., the contribution of glucose via glycolysis to total mitochondrial oxidation. (F) (G) The contribution of fatty acid oxidation to citrate synthase flux. In all panels, groups were compared using the two-tailed unpaired Student's t-test. (n=4 mice and 6 were studied. Figure data 1 Source data for Figure 6 and Figure supplement 1. Download clustering species-specific on in vivo metabolic fluxes not in vitro fluxes clustering was applied to our in vitro flux data and no clustering between species (Figure the in vivo metabolic flux data to clustering of rats and mice (Figure a analysis of in vitro in vivo metabolic flux. Figure 7 Download asset Open asset of in vitro and in vivo results. (A) Study (B) heatmap demonstrating the absence of metabolic differences in vitro. (C) heatmap demonstrating metabolic differences between mice and rats in vivo. In (B) and (C), mouse and rat to the species to the on the of each pyruvate carboxylase flux, citrate synthase flux, pyruvate dehydrogenase flux, fatty acid fatty acid All data in Figures 3 and 5 were utilized in the clustering analysis and are included in this Discussion Oxygen consumption has been shown to scale inversely with body mass in species in mass across 20 of from to et al., et al., et al., et al., et al., et al., This phenomenon has been most studied in mammals, is also been shown to in and et al., et al., et al., et al., et al., et al., et al., et al., and et al., and et al., et al., and (Glazier, et al., 2013; et al., a limitation of prior studies in this has been that observations have been largely limited to oxygen consumption and caloric other metabolic processes unexplored. This study to address this by the of the inverse between body mass and metabolic using both experimental and databases that have not been employed in this is important to that the metabolic processes which we observed to be higher in mice as compared to rats did not to the metabolic scaling with metabolic rates proportional to three-quarters of body mass. This to the that the scaling is it is entirely that oxygen consumption could be proportional to three-quarters of body mass, while other metabolic processes may exhibit a different scaling Further studies across species beyond be required to address this The possibility that gene expression, as by mRNA may also scale with body mass has not been We observed that the expression of key genes in glycolysis, gluconeogenesis, fatty acid metabolism, NAD synthesis and transport, mitochondrial oxygen consumption, and from oxidative damage scale with body mass. however, is the that genes for which an inverse of expression with body mass is are not across the the of genes whose expression is inversely correlated with body mass is for genes related to metabolic processes, and whose corresponding enzymatic are by the of or The that body mass is a related to the level of expression of certain genes has not been as an of metabolic scaling. it be that metabolic scaling cannot be at the transcriptional level, because many rate-limiting enzymes in the metabolic processes measured in vivo did not scale at the transcriptional level, and only approximately half of genes that scaled at the level of mRNA scaled at the level of it is likely that both transcriptional and other mechanisms such as enzyme activity are responsible for in metabolic flux per unit mass, inversely proportionally to body size. Additionally, the data not allow to assess whether the expression of certain of key metabolic enzymes scales differentially across species. is also to contrast the of an inverse between body size and metabolic fluxes per tissue weight consumption, mitochondrial and glucose production in measured in the in vitro to our in vivo all fluxes in mice than in This the to tracer methods in vivo to a comprehensive of differences in metabolic fluxes between species. Our that measurement of oxygen consumption in vitro may to any influence of scaling processes present in vivo. Glucose production was threefold higher in mouse liver slices relative to rat liver slices, did not significantly between plated hepatocytes from mice and rats. studies using metabolic flux analysis may have the further to as to the and of metabolic scaling For example, our data not allow to whether differences in oxygen consumption metabolic as has been suggested in the of et al., or metabolic alterations changes in oxygen Additionally, there are to the that metabolic flux studies were performed only in the most related species included in the transcriptomics analysis monkey and human Our does not have the to flux analysis in larger or smaller species, we that our beyond the range in body size between mice and

Genome-scale models have become indispensable tools for the study of cellular growth. These models have been progressively improving over the past two decades, enabling accurate predictions of metabolic fluxes and key phenotypes under a variety of growth conditions. In this work, an efficient computational method is proposed to incorporate genome-scale models into superstructure optimization settings, introducing them as viable growth models to simulate the cultivation section of biorefinaries. We perform techno-economic and life-cycle analyses of an algal biorefinery with five processing sections to determine optimal processing pathways and technologies. Formulation of this problem results in a mixed-integer nonlinear program, in which the net present value is maximized with respect to mass flowrates and design parameters. We use a genome-scale metabolic model of Chlamydomonas reinhardtii to predict growth rates in the cultivation section. We study algae cultivation in open ponds, in which exchange fluxes of biomass and carbon dioxide are directly determined by the metabolic model. This formulation enables the coupling of flowrates and design parameters, leading to more accurate cultivation productivity estimates with respect to substrate concentration and light intensity.

Genome-scale metabolic models are available for an increasing number of organisms and can be used to define the region of feasible metabolic flux distributions. In this work we use as constraints a small set of experimental metabolic fluxes, which reduces the region of feasible metabolic states. Once the region of feasible flux distributions has been defined, a set of possible flux distributions is obtained by random sampling and the averages and standard deviations for each of the metabolic fluxes in the genome-scale model are calculated. These values allow estimation of the significance of change for each reaction rate between different conditions and comparison of it with the significance of change in gene transcription for the corresponding enzymes. The comparison of flux change and gene expression allows identification of enzymes showing a significant correlation between flux change and expression change (transcriptional regulation) as well as reactions whose flux change is likely to be driven only by changes in the metabolite concentrations (metabolic regulation). The changes due to growth on four different carbon sources and as a consequence of five gene deletions were analyzed for Saccharomyces cerevisiae. The enzymes with transcriptional regulation showed enrichment in certain transcription factors. This has not been previously reported. The information provided by the presented method could guide the discovery of new metabolic engineering strategies or the identification of drug targets for treatment of metabolic diseases.