Computational reconstruction of tissue-specific metabolic models: application to human liver metabolism

Analysing Algorithms and Data Sources for the Tissue-Specific Reconstruction of Liver Healthy and Cancer Cells

Genome-Scale Metabolic Models (GSMMs), mathematical representations of the cell metabolism in different organisms including humans, are resourceful tools to simulate metabolic phenotypes and understand associated diseases, such as obesity, diabetes and cancer. In the last years, different algorithms have been developed to generate tissue-specific metabolic models that simulate different phenotypes for distinct cell types. Hepatocyte cells are one of the main sites of metabolic conversions, mainly due to their diverse physiological functions. Most of the liver's tissue is formed by hepatocytes, being one of the largest and most important organs regarding its biological functions. Hepatocellular carcinoma is, also, one of the most important human cancers with high mortality rates. In this study, we will analyze four different algorithms (MBA, mCADRE, tINIT and FASTCORE) for tissue-specific model reconstruction, based on a template model and two types of data sources: transcriptomics and proteomics. These methods will be applied to the reconstruction of metabolic models for hepatocyte cells and HepG2 cancer cell line. The models will be analyzed and compared under different perspectives, emphasizing their functional analysis considering a set of metabolic liver tasks. The results show that there is no "ideal" algorithm. However, with the current analysis, we were able to retrieve knowledge about the metabolism of the liver.

A Systematic Evaluation of Methods for Tailoring Genome-Scale Metabolic Models

Genome-scale models of metabolism can illuminate the molecular basis of cell phenotypes. Since some enzymes are only active in specific cell types, several algorithms use omics data to construct cell-line- and tissue-specific metabolic models from genome-scale models. However, these methods are often not rigorously benchmarked, and it is unclear how algorithm and parameter selection (e.g., gene expression thresholds, metabolic constraints) affects model content and predictive accuracy. To investigate this, we built hundreds of models of four different cancer cell lines using six algorithms, four gene expression thresholds, and three sets of metabolic constraints. Model content varied substantially across different parameter sets, but the algorithms generally increased accuracy in gene essentiality predictions. However, model extraction method choice had the largest impact on model accuracy. We further highlight how assumptions during model development influence model prediction accuracy. These insights will guide further development of context-specific models, thus more accurately resolving genotype-phenotype relationships.

Degradation Network Reconstruction Guided by Metagenomic Data

Network reconstruction procedures based on meta-"omics" data are an invaluable tool for inferring total and active set of reactions mediated by different members in a microbial community. Within them, network-based methods for automatic analysis of catabolic capacities in metagenomes are currently limited. Here, we describe the complete workflow, scripts, and commands allowing the automatic reconstruction of biodegradation networks using as an input meta-sequences generated by direct DNA sequencing.

Genome scale metabolic modeling of cancer

Cancer cells reprogram metabolism to support rapid proliferation and survival. Energy metabolism is particularly important for growth and genes encoding enzymes involved in energy metabolism are frequently altered in cancer cells. A genome scale metabolic model (GEM) is a mathematical formalization of metabolism which allows simulation and hypotheses testing of metabolic strategies. It has successfully been applied to many microorganisms and is now used to study cancer metabolism. Generic models of human metabolism have been reconstructed based on the existence of metabolic genes in the human genome. Cancer specific models of metabolism have also been generated by reducing the number of reactions in the generic model based on high throughput expression data, e.g. transcriptomics and proteomics. Targets for drugs and bio markers for diagnostics have been identified using these models. They have also been used as scaffolds for analysis of high throughput data to allow mechanistic interpretation of changes in expression. Finally, GEMs allow quantitative flux predictions using flux balance analysis (FBA). Here we critically review the requirements for successful FBA simulations of cancer cells and discuss the symmetry between the methods used for modeling of microbial and cancer metabolism. GEMs have great potential for translational research on cancer and will therefore become of increasing importance in the future.

Towards improved genome-scale metabolic network reconstructions: unification, transcript specificity and beyond

Genome-scale metabolic network reconstructions provide a basis for the investigation of the metabolic properties of an organism. There are reconstructions available for multiple organisms, from prokaryotes to higher organisms and methods for the analysis of a reconstruction. One example is the use of flux balance analysis to improve the yields of a target chemical, which has been applied successfully. However, comparison of results between existing reconstructions and models presents a challenge because of the heterogeneity of the available reconstructions, for example, of standards for presenting gene-protein-reaction associations, nomenclature of metabolites and reactions or selection of protonation states. The lack of comparability for gene identifiers or model-specific reactions without annotated evidence often leads to the creation of a new model from scratch, as data cannot be properly matched otherwise. In this contribution, we propose to improve the predictive power of metabolic models by switching from gene-protein-reaction associations to transcript-isoform-reaction associations, thus taking advantage of the improvement of precision in gene expression measurements. To achieve this precision, we discuss available databases that can be used to retrieve this type of information and point at issues that can arise from their neglect. Further, we stress issues that arise from non-standardized building pipelines, like inconsistencies in protonation states. In addition, problems arising from the use of non-specific cofactors, e.g. artificial futile cycles, are discussed, and finally efforts of the metabolic modelling community to unify model reconstructions are highlighted.

Understanding Metabolic Regulation at a Systems Level: Metabolite Sensing, Mathematical Predictions, and Model Organisms

Metabolic networks are extensively regulated to facilitate tissue-specific metabolic programs and robustly maintain homeostasis in response to dietary changes. Homeostatic metabolic regulation is achieved through metabolite sensing coupled to feedback regulation of metabolic enzyme activity or expression. With a wealth of transcriptomic, proteomic, and metabolomic data available for different cell types across various conditions, we are challenged with understanding global metabolic network regulation and the resulting metabolic outputs. Stoichiometric metabolic network modeling integrated with "omics" data has addressed this challenge by generating nonintuitive, testable hypotheses about metabolic flux rewiring. Model organism studies have also yielded novel insight into metabolic networks. This review covers three topics: the feedback loops inherent in metabolic regulatory networks, metabolic network modeling, and interspecies studies utilizing Caenorhabditis elegans and various bacterial diets that have revealed novel metabolic paradigms.

Distinct metabolic network states manifest in the gene expression profiles of pediatric inflammatory bowel disease patients and controls

Information on biological networks can greatly facilitate the function-orientated interpretation of high-throughput molecular data. Genome-wide metabolic network models of human cells, in particular, can be employed to contextualize gene expression profiles of patients with the goal of both, a better understanding of individual etiologies and an educated reclassification of (clinically defined) phenotypes. We analyzed publicly available expression profiles of intestinal tissues from treatment-naive pediatric inflammatory bowel disease (IBD) patients and age-matched control individuals, using a reaction-centric metabolic network derived from the Recon2 model. By way of defining a measure of 'coherence', we quantified how well individual patterns of expression changes matched the metabolic network. We observed a bimodal distribution of metabolic network coherence in both patients and controls, albeit at notably different mixture probabilities. Multidimensional scaling analysis revealed a bisectional pattern as well that overlapped widely with the metabolic network-based results. Expression differences driving the observed bimodality were related to cellular transport of thiamine and bile acid metabolism, thereby highlighting the crosstalk between metabolism and other vital pathways. We demonstrated how classical data mining and network analysis can jointly identify biologically meaningful patterns in gene expression data.

Integrative analysis of human omics data using biomolecular networks

High-throughput '-omics' technologies have given rise to an increasing abundance of genome-scale data detailing human biology at the molecular level. Although these datasets have already made substantial contributions to a more comprehensive understanding of human physiology and diseases, their interpretation becomes increasingly cryptic and nontrivial as they continue to expand in size and complexity. Systems biology networks offer a scaffold upon which omics data can be integrated, facilitating the extraction of new and physiologically relevant information from the data. Two of the most prevalent networks that have been used for such integrative analyses of omics data are genome-scale metabolic models (GEMs) and protein-protein interaction (PPI) networks, both of which have demonstrated success among many different omics and sample types. This integrative approach seeks to unite 'top-down' omics data with 'bottom-up' biological networks in a synergistic fashion that draws on the strengths of both strategies. As the volume and resolution of high-throughput omics data continue to grow, integrative network-based analyses are expected to play an increasingly important role in their interpretation.

MetaboTools: A Comprehensive Toolbox for Analysis of Genome-Scale Metabolic Models

Metabolomic data sets provide a direct read-out of cellular phenotypes and are increasingly generated to study biological questions. Previous work, by us and others, revealed the potential of analyzing extracellular metabolomic data in the context of the metabolic model using constraint-based modeling. With the MetaboTools, we make our methods available to the broader scientific community. The MetaboTools consist of a protocol, a toolbox, and tutorials of two use cases. The protocol describes, in a step-wise manner, the workflow of data integration, and computational analysis. The MetaboTools comprise the Matlab code required to complete the workflow described in the protocol. Tutorials explain the computational steps for integration of two different data sets and demonstrate a comprehensive set of methods for the computational analysis of metabolic models and stratification thereof into different phenotypes. The presented workflow supports integrative analysis of multiple omics data sets. Importantly, all analysis tools can be applied to metabolic models without performing the entire workflow. Taken together, the MetaboTools constitute a comprehensive guide to the intra-model analysis of extracellular metabolomic data from microbial, plant, or human cells. This computational modeling resource offers a broad set of computational analysis tools for a wide biomedical and non-biomedical research community.

What Have Metabolomics Approaches Taught Us About Type 2 Diabetes?

Type 2 diabetes (T2D) is increasing worldwide, making identification of biomarkers for detection, staging, and effective prevention strategies an especially critical scientific and medical goal. Fortunately, advances in metabolomics techniques, together with improvements in bioinformatics and mathematical modeling approaches, have provided the scientific community with new tools to describe the T2D metabolome. The metabolomics signatures associated with T2D and obesity include increased levels of lactate, glycolytic intermediates, branched-chain and aromatic amino acids, and long-chain fatty acids. Conversely, tricarboxylic acid cycle intermediates, betaine, and other metabolites decrease. Future studies will be required to fully integrate these and other findings into our understanding of diabetes pathophysiology and to identify biomarkers of disease risk, stage, and responsiveness to specific treatments.

Advances in the integration of transcriptional regulatory information into genome-scale metabolic models

A major goal of systems biology is to build predictive computational models of cellular metabolism. Availability of complete genome sequences and wealth of legacy biochemical information has led to the reconstruction of genome-scale metabolic networks in the last 15 years for several organisms across the three domains of life. Due to paucity of information on kinetic parameters associated with metabolic reactions, the constraint-based modelling approach, flux balance analysis (FBA), has proved to be a vital alternative to investigate the capabilities of reconstructed metabolic networks. In parallel, advent of high-throughput technologies has led to the generation of massive amounts of omics data on transcriptional regulation comprising mRNA transcript levels and genome-wide binding profile of transcriptional regulators. A frontier area in metabolic systems biology has been the development of methods to integrate the available transcriptional regulatory information into constraint-based models of reconstructed metabolic networks in order to increase the predictive capabilities of computational models and understand the regulation of cellular metabolism. Here, we review the existing methods to integrate transcriptional regulatory information into constraint-based models of metabolic networks.

In Silico Modeling of Liver Metabolism in a Human Disease Reveals a Key Enzyme for Histidine and Histamine Homeostasis

Primary hyperoxaluria type I (PH1) is an autosomal-recessive inborn error of liver metabolism caused by alanine:glyoxylate aminotransferase (AGT) deficiency. In silico modeling of liver metabolism in PH1 recapitulated accumulation of known biomarkers as well as alteration of histidine and histamine levels, which we confirmed in vitro, in vivo, and in PH1 patients. AGT-deficient mice showed decreased vascular permeability, a readout of in vivo histamine activity. Histamine reduction is most likely caused by increased catabolism of the histamine precursor histidine, triggered by rerouting of alanine flux from AGT to the glutamic-pyruvate transaminase (GPT, also known as the alanine-transaminase ALT). Alanine administration reduces histamine levels in wild-type mice, while overexpression of GPT in PH1 mice increases plasma histidine, normalizes histamine levels, restores vascular permeability, and decreases urinary oxalate levels. Our work demonstrates that genome-scale metabolic models are clinically relevant and can link genotype to phenotype in metabolic disorders.

•

In silico model of liver metabolism reveals global metabolic alterations in PH1

•

Changes in amino acid metabolism in PH1 result in a reduction of histidine and histamine

•

GPT overexpression normalizes histamine levels and reduces oxalate in PH1 mice

A Caenorhabditis elegans Genome-Scale Metabolic Network Model

Caenorhabditis elegans is a powerful model to study metabolism and how it relates to nutrition, gene expression, and life history traits. However, while numerous experimental techniques that enable perturbation of its diet and gene function are available, a high-quality metabolic network model has been lacking. Here, we reconstruct an initial version of the C. elegans metabolic network. This network model contains 1,273 genes, 623 enzymes, and 1,985 metabolic reactions and is referred to as iCEL1273. Using flux balance analysis, we show that iCEL1273 is capable of representing the conversion of bacterial biomass into C. elegans biomass during growth and enables the predictions of gene essentiality and other phenotypes. In addition, we demonstrate that gene expression data can be integrated with the model by comparing metabolic rewiring in dauer animals versus growing larvae. iCEL1273 is available at a dedicated website (wormflux.umassmed.edu) and will enable the unraveling of the mechanisms by which different macro- and micronutrients contribute to the animal's physiology.

Assessment of FBA Based Gene Essentiality Analysis in Cancer with a Fast Context-Specific Network Reconstruction Method

MOTIVATION

Gene Essentiality Analysis based on Flux Balance Analysis (FBA-based GEA) is a promising tool for the identification of novel metabolic therapeutic targets in cancer. The reconstruction of cancer-specific metabolic networks, typically based on gene expression data, constitutes a sensible step in this approach. However, to our knowledge, no extensive assessment on the influence of the reconstruction process on the obtained results has been carried out to date.

RESULTS

In this article, we aim to study context-specific networks and their FBA-based GEA results for the identification of cancer-specific metabolic essential genes. To that end, we used gene expression datasets from the Cancer Cell Line Encyclopedia (CCLE), evaluating the results obtained in 174 cancer cell lines. In order to more clearly observe the effect of cancer-specific expression data, we did the same analysis using randomly generated expression patterns. Our computational analysis showed some essential genes that are fairly common in the reconstructions derived from both gene expression and randomly generated data. However, though of limited size, we also found a subset of essential genes that are very rare in the randomly generated networks, while recurrent in the sample derived networks, and, thus, would presumably constitute relevant drug targets for further analysis. In addition, we compare the in-silico results to high-throughput gene silencing experiments from Project Achilles with conflicting results, which leads us to raise several questions, particularly the strong influence of the selected biomass reaction on the obtained results. Notwithstanding, using previous literature in cancer research, we evaluated the most relevant of our targets in three different cancer cell lines, two derived from Gliobastoma Multiforme and one from Non-Small Cell Lung Cancer, finding that some of the predictions are in the right track.

Reconstruction of genome-scale human metabolic models using omics data

The impact of genome-scale human metabolic models on human systems biology and medical sciences is becoming greater, thanks to increasing volumes of model building platforms and publicly available omics data. The genome-scale human metabolic models started with Recon 1 in 2007, and have since been used to describe metabolic phenotypes of healthy and diseased human tissues and cells, and to predict therapeutic targets. Here we review recent trends in genome-scale human metabolic modeling, including various generic and tissue/cell type-specific human metabolic models developed to date, and methods, databases and platforms used to construct them. For generic human metabolic models, we pay attention to Recon 2 and HMR 2.0 with emphasis on data sources used to construct them. Draft and high-quality tissue/cell type-specific human metabolic models have been generated using these generic human metabolic models. Integration of tissue/cell type-specific omics data with the generic human metabolic models is the key step, and we discuss omics data and their integration methods to achieve this task. The initial version of the tissue/cell type-specific human metabolic models can further be computationally refined through gap filling, reaction directionality assignment and the subcellular localization of metabolic reactions. We review relevant tools for this model refinement procedure as well. Finally, we suggest the direction of further studies on reconstructing an improved human metabolic model.

HepatoDyn: A Dynamic Model of Hepatocyte Metabolism That Integrates 13C Isotopomer Data

The liver performs many essential metabolic functions, which can be studied using computational models of hepatocytes. Here we present HepatoDyn, a highly detailed dynamic model of hepatocyte metabolism. HepatoDyn includes a large metabolic network, highly detailed kinetic laws, and is capable of dynamically simulating the redox and energy metabolism of hepatocytes. Furthermore, the model was coupled to the module for isotopic label propagation of the software package IsoDyn, allowing HepatoDyn to integrate data derived from 13C based experiments. As an example of dynamical simulations applied to hepatocytes, we studied the effects of high fructose concentrations on hepatocyte metabolism by integrating data from experiments in which rat hepatocytes were incubated with 20 mM glucose supplemented with either 3 mM or 20 mM fructose. These experiments showed that glycogen accumulation was significantly lower in hepatocytes incubated with medium supplemented with 20 mM fructose than in hepatocytes incubated with medium supplemented with 3 mM fructose. Through the integration of extracellular fluxes and 13C enrichment measurements, HepatoDyn predicted that this phenomenon can be attributed to a depletion of cytosolic ATP and phosphate induced by high fructose concentrations in the medium.

Transcriptomics resources of human tissues and organs

Quantifying the differential expression of genes in various human organs, tissues, and cell types is vital to understand human physiology and disease. Recently, several large‐scale transcriptomics studies have analyzed the expression of protein‐coding genes across tissues. These datasets provide a framework for defining the molecular constituents of the human body as well as for generating comprehensive lists of proteins expressed across tissues or in a tissue‐restricted manner. Here, we review publicly available human transcriptome resources and discuss body‐wide data from independent genome‐wide transcriptome analyses of different tissues. Gene expression measurements from these independent datasets, generated using samples from fresh frozen surgical specimens and postmortem tissues, are consistent. Overall, the different genome‐wide analyses support a distribution in which many proteins are found in all tissues and relatively few in a tissue‐restricted manner. Moreover, we discuss the applications of publicly available omics data for building genome‐scale metabolic models, used for analyzing cell and tissue functions both in physiological and in disease contexts.

Reconstruction of Tissue-Specific Metabolic Networks Using CORDA

Human metabolism involves thousands of reactions and metabolites. To interpret this complexity, computational modeling becomes an essential experimental tool. One of the most popular techniques to study human metabolism as a whole is genome scale modeling. A key challenge to applying genome scale modeling is identifying critical metabolic reactions across diverse human tissues. Here we introduce a novel algorithm called Cost Optimization Reaction Dependency Assessment (CORDA) to build genome scale models in a tissue-specific manner. CORDA performs more efficiently computationally, shows better agreement to experimental data, and displays better model functionality and capacity when compared to previous algorithms. CORDA also returns reaction associations that can greatly assist in any manual curation to be performed following the automated reconstruction process. Using CORDA, we developed a library of 76 healthy and 20 cancer tissue-specific reconstructions. These reconstructions identified which metabolic pathways are shared across diverse human tissues. Moreover, we identified changes in reactions and pathways that are differentially included and present different capacity profiles in cancer compared to healthy tissues, including up-regulation of folate metabolism, the down-regulation of thiamine metabolism, and tight regulation of oxidative phosphorylation.

Reconstruction and applications of consensus yeast metabolic network based on RNA sequencing

One practical application of genome-scale metabolic reconstructions is to interrogate multispecies relationships. Here, we report a consensus metabolic model in four yeast species (Saccharomyces cerevisiae, S. paradoxus, S. mikatae, and S. bayanus) by integrating metabolic network simulations with RNA sequencing (RNA-seq) datasets. We generated high-resolution transcriptome maps of four yeast species through de novo assembly and genome-guided approaches. The transcriptomes were annotated and applied to build the consensus metabolic network, which was verified using independent RNA-seq experiments. The expression profiles reveal that the genes involved in amino acid and lipid metabolism are highly coexpressed. The diverse phenotypic characteristics, such as cellular growth and gene deletions, can be simulated using the metabolic model. We also explored the applications of the consensus model in metabolic engineering using yeast-specific reactions and biofuel production as examples. Similar strategies will benefit communities studying genome-scale metabolic networks of other organisms.

Constraint Based Modeling Going Multicellular

Constraint based modeling has seen applications in many microorganisms. For example, there are now established methods to determine potential genetic modifications and external interventions to increase the efficiency of microbial strains in chemical production pipelines. In addition, multiple models of multicellular organisms have been created including plants and humans. While initially the focus here was on modeling individual cell types of the multicellular organism, this focus recently started to switch. Models of microbial communities, as well as multi-tissue models of higher organisms have been constructed. These models thereby can include different parts of a plant, like root, stem, or different tissue types in the same organ. Such models can elucidate details of the interplay between symbiotic organisms, as well as the concerted efforts of multiple tissues and can be applied to analyse the effects of drugs or mutations on a more systemic level. In this review we give an overview of the recent development of multi-tissue models using constraint based techniques and the methods employed when investigating these models. We further highlight advances in combining constraint based models with dynamic and regulatory information and give an overview of these types of hybrid or multi-level approaches.

Benchmarking Procedures for High-Throughput Context Specific Reconstruction Algorithms

Recent progress in high-throughput data acquisition has shifted the focus from data generation to processing and understanding of how to integrate collected information. Context specific reconstruction based on generic genome scale models like ReconX or HMR has the potential to become a diagnostic and treatment tool tailored to the analysis of specific individuals. The respective computational algorithms require a high level of predictive power, robustness and sensitivity. Although multiple context specific reconstruction algorithms were published in the last 10 years, only a fraction of them is suitable for model building based on human high-throughput data. Beside other reasons, this might be due to problems arising from the limitation to only one metabolic target function or arbitrary thresholding. This review describes and analyses common validation methods used for testing model building algorithms. Two major methods can be distinguished: consistency testing and comparison based testing. The first is concerned with robustness against noise, e.g., missing data due to the impossibility to distinguish between the signal and the background of non-specific binding of probes in a microarray experiment, and whether distinct sets of input expressed genes corresponding to i.e., different tissues yield distinct models. The latter covers methods comparing sets of functionalities, comparison with existing networks or additional databases. We test those methods on several available algorithms and deduce properties of these algorithms that can be compared with future developments. The set of tests performed, can therefore serve as a benchmarking procedure for future algorithms.

Characterizing the optimal flux space of genome-scale metabolic reconstructions through modified latin-hypercube sampling

Sampling of the optimal flux space using modified LHS gives a more uniform coverage than Monte-Carlo Sampling. Analysis of the flux data shows that majority of variation in the flux distribution pattern within the space arises due to the presence of few alternate pathways.

Genome-wide metabolic model to improve understanding of CD4+ T cell metabolism, immunometabolism and application in drug design

Model-based investigation of the metabolism and immunometabolism of CD4⁺ T cells (CD4T1670) and the application of CD4T1670 in drug development.

Computational Modeling of Human Metabolism and Its Application to Systems Biomedicine

Modern high-throughput techniques offer immense opportunities to investigate whole-systems behavior, such as those underlying human diseases. However, the complexity of the data presents challenges in interpretation, and new avenues are needed to address the complexity of both diseases and data. Constraint-based modeling is one formalism applied in systems biology. It relies on a genome-scale reconstruction that captures extensive biochemical knowledge regarding an organism. The human genome-scale metabolic reconstruction is increasingly used to understand normal cellular and disease states because metabolism is an important factor in many human diseases. The application of human genome-scale reconstruction ranges from mere querying of the model as a knowledge base to studies that take advantage of the model's topology and, most notably, to functional predictions based on cell- and condition-specific metabolic models built based on omics data.An increasing number and diversity of biomedical questions are being addressed using constraint-based modeling and metabolic models. One of the most successful biomedical applications to date is cancer metabolism, but constraint-based modeling also holds great potential for inborn errors of metabolism or obesity. In addition, it offers great prospects for individualized approaches to diagnostics and the design of disease prevention and intervention strategies. Metabolic models support this endeavor by providing easy access to complex high-throughput datasets. Personalized metabolic models have been introduced. Finally, constraint-based modeling can be used to model whole-body metabolism, which will enable the elucidation of metabolic interactions between organs and disturbances of these interactions as either causes or consequence of metabolic diseases. This chapter introduces constraint-based modeling and describes some of its contributions to systems biomedicine.

Mathematical models of cancer metabolism

Metabolism is essential for life, and its alteration is implicated in multiple human diseases. The transformation from a normal to a cancerous cell requires metabolic changes to fuel the high metabolic demands of cancer cells, including but not limited to cell proliferation and cell migration. In recent years, there have been a number of new discoveries connecting known aberrations in oncogenic and tumour suppressor pathways with metabolic alterations required to sustain cell proliferation and migration. However, an understanding of the selective advantage of these metabolic alterations is still lacking. Here, we review the literature on mathematical models of metabolism, with an emphasis on their contribution to the identification of the selective advantage of metabolic phenotypes that seem otherwise wasteful or accidental. We will show how the molecular hallmarks of cancer can be related to cell proliferation and tissue remodelling, the two major physiological requirements for the development of a multicellular structure. We will cover different areas such as genome-wide gene expression analysis, flux balance models, kinetic models, reaction diffusion models and models of the tumour microenvironment. We will also highlight current challenges and how their resolution will help to achieve a better understanding of cancer metabolism and the metabolic vulnerabilities of cancers.

Cancer Metabolism: A Modeling Perspective

Tumor cells alter their metabolism to maintain unregulated cellular proliferation and survival, but this transformation leaves them reliant on constant supply of nutrients and energy. In addition to the widely studied dysregulated glucose metabolism to fuel tumor cell growth, accumulating evidences suggest that utilization of amino acids and lipids contributes significantly to cancer cell metabolism. Also recent progresses in our understanding of carcinogenesis have revealed that cancer is a complex disease and cannot be understood through simple investigation of genetic mutations of cancerous cells. Cancer cells present in complex tumor tissues communicate with the surrounding microenvironment and develop traits which promote their growth, survival, and metastasis. Decoding the full scope and targeting dysregulated metabolic pathways that support neoplastic transformations and their preservation requires both the advancement of experimental technologies for more comprehensive measurement of omics as well as the advancement of robust computational methods for accurate analysis of the generated data. Here, we review cancer-associated reprogramming of metabolism and highlight the capability of genome-scale metabolic modeling approaches in perceiving a system-level perspective of cancer metabolism and in detecting novel selective drug targets.

Genome-scale metabolic modelling of hepatocytes reveals serine deficiency in patients with non-alcoholic fatty liver disease

Several liver disorders result from perturbations in the metabolism of hepatocytes, and their underlying mechanisms can be outlined through the use of genome-scale metabolic models (GEMs). Here we reconstruct a consensus GEM for hepatocytes, which we call iHepatocytes2322, that extends previous models by including an extensive description of lipid metabolism. We build iHepatocytes2322 using Human Metabolic Reaction 2.0 database and proteomics data in Human Protein Atlas, which experimentally validates the incorporated reactions. The reconstruction process enables improved annotation of the proteomics data using the network centric view of iHepatocytes2322. We then use iHepatocytes2322 to analyse transcriptomics data obtained from patients with non-alcoholic fatty liver disease. We show that blood concentrations of chondroitin and heparan sulphates are suitable for diagnosing non-alcoholic steatohepatitis and for the staging of non-alcoholic fatty liver disease. Furthermore, we observe serine deficiency in patients with NASH and identify PSPH, SHMT1 and BCAT1 as potential therapeutic targets for the treatment of non-alcoholic steatohepatitis.

Inference and Prediction of Metabolic Network Fluxes

In this Update, we cover the basic principles of the estimation and prediction of the rates of the many interconnected biochemical reactions that constitute plant metabolic networks. This includes metabolic flux analysis approaches that utilize the rates or patterns of redistribution of stable isotopes of carbon and other atoms to estimate fluxes, as well as constraints-based optimization approaches such as flux balance analysis. Some of the major insights that have been gained from analysis of fluxes in plants are discussed, including the functioning of metabolic pathways in a network context, the robustness of the metabolic phenotype, the importance of cell maintenance costs, and the mechanisms that enable energy and redox balancing at steady state. We also discuss methodologies to exploit 'omic data sets for the construction of tissue-specific metabolic network models and to constrain the range of permissible fluxes in such models. Finally, we consider the future directions and challenges faced by the field of metabolic network flux phenotyping.

Integrated metabolic modelling reveals cell-type specific epigenetic control points of the macrophage metabolic network

BACKGROUND

The reconstruction of context-specific metabolic models from easily and reliably measurable features such as transcriptomics data will be increasingly important in research and medicine. Current reconstruction methods suffer from high computational effort and arbitrary threshold setting. Moreover, understanding the underlying epigenetic regulation might allow the identification of putative intervention points within metabolic networks. Genes under high regulatory load from multiple enhancers or super-enhancers are known key genes for disease and cell identity. However, their role in regulation of metabolism and their placement within the metabolic networks has not been studied.

METHODS

Here we present FASTCORMICS, a fast and robust workflow for the creation of high-quality metabolic models from transcriptomics data. FASTCORMICS is devoid of arbitrary parameter settings and due to its low computational demand allows cross-validation assays. Applying FASTCORMICS, we have generated models for 63 primary human cell types from microarray data, revealing significant differences in their metabolic networks.

RESULTS

To understand the cell type-specific regulation of the alternative metabolic pathways we built multiple models during differentiation of primary human monocytes to macrophages and performed ChIP-Seq experiments for histone H3 K27 acetylation (H3K27ac) to map the active enhancers in macrophages. Focusing on the metabolic genes under high regulatory load from multiple enhancers or super-enhancers, we found these genes to show the most cell type-restricted and abundant expression profiles within their respective pathways. Importantly, the high regulatory load genes are associated to reactions enriched for transport reactions and other pathway entry points, suggesting that they are critical regulatory control points for cell type-specific metabolism.

CONCLUSIONS

By integrating metabolic modelling and epigenomic analysis we have identified high regulatory load as a common feature of metabolic genes at pathway entry points such as transporters within the macrophage metabolic network. Analysis of these control points through further integration of metabolic and gene regulatory networks in various contexts could be beneficial in multiple fields from identification of disease intervention strategies to cellular reprogramming.

The gut microbiota modulates host amino acid and glutathione metabolism in mice

The gut microbiota has been proposed as an environmental factor that promotes the progression of metabolic diseases. Here, we investigated how the gut microbiota modulates the global metabolic differences in duodenum, jejunum, ileum, colon, liver, and two white adipose tissue depots obtained from conventionally raised (CONV-R) and germ-free (GF) mice using gene expression data and tissue-specific genome-scale metabolic models (GEMs). We created a generic mouse metabolic reaction (MMR) GEM, reconstructed 28 tissue-specific GEMs based on proteomics data, and manually curated GEMs for small intestine, colon, liver, and adipose tissues. We used these functional models to determine the global metabolic differences between CONV-R and GF mice. Based on gene expression data, we found that the gut microbiota affects the host amino acid (AA) metabolism, which leads to modifications in glutathione metabolism. To validate our predictions, we measured the level of AAs and N-acetylated AAs in the hepatic portal vein of CONV-R and GF mice. Finally, we simulated the metabolic differences between the small intestine of the CONV-R and GF mice accounting for the content of the diet and relative gene expression differences. Our analyses revealed that the gut microbiota influences host amino acid and glutathione metabolism in mice.

Metaproteomics and metabolomics analyses of chronically petroleum-polluted sites reveal the importance of general anaerobic processes uncoupled with degradation

Crude oil is one of the most important natural assets for humankind, yet it is a major environmental pollutant, notably in marine environments. One of the largest crude oil polluted areas in the word is the semi-enclosed Mediterranean Sea, in which the metabolic potential of indigenous microbial populations towards the large-scale chronic pollution is yet to be defined, particularly in anaerobic and micro-aerophilic sites. Here, we provide an insight into the microbial metabolism in sediments from three chronically polluted marine sites along the coastline of Italy: the Priolo oil terminal/refinery site (near Siracuse, Sicily), harbour of Messina (Sicily) and shipwreck of MT Haven (near Genoa). Using shotgun metaproteomics and community metabolomics approaches, the presence of 651 microbial proteins and 4776 metabolite mass features have been detected in these three environments, revealing a high metabolic heterogeneity between the investigated sites. The proteomes displayed the prevalence of anaerobic metabolisms that were not directly related with petroleum biodegradation, indicating that in the absence of oxygen, biodegradation is significantly suppressed. This suppression was also suggested by examining the metabolome patterns. The proteome analysis further highlighted the metabolic coupling between methylotrophs and sulphate reducers in oxygen-depleted petroleum-polluted sediments.

Metabolic modeling predicts metabolite changes in Mycobacterium tuberculosis

BACKGROUND

Mycobacterium tuberculosis (MTB) is the causal agent of the disease tuberculosis (TB). Metabolic adaptations are thought to be critical to the survival of MTB during pathogenesis. Computational tools that can be used to study MTB metabolism in silico and prioritize resource-intensive experimental work could significantly accelerate research.

RESULTS

We have developed E-Flux-MFC, an enhancement of our original E-Flux method that enables the prediction of changes in the production of external and internal metabolites corresponding to gene expression measurements. We have used this method to simulate the changes in the metabolic state of Mycobacterium tuberculosis (MTB). We have validated the accuracy of E-Flux-MFC for predicting changes in lipids and metabolites during a hypoxia time course using previously published metabolomics and transcriptomics data. We have further validated the accuracy of the method for predicting changes in MTB lipids following the deletion and induction of two well-studied transcription factors (TFs). We have applied the method to predict the metabolic impact of the induction of each of the approximately 180 MTB TFs using a previously generated and publically available expression data set.

CONCLUSIONS

E-flux-MFC can be used to study global changes in MTB metabolites from gene expression data associated with environmental and genetic perturbations. The application of this method to a data set of MTB TF perturbations provides a resource for studying the large number of TFs whose functions remain unknown. Most TFs impact metabolites indirectly through the propagation of gene expression changes through the regulatory network rather than through their direct regulons. E-Flux-MFC is also applicable to any organism for which accurate metabolic models are available.

Cancer cell metabolism as new targets for novel designed therapies

Metabolic processes are altered in cancer cells, which obtain advantages from this metabolic reprogramming in terms of energy production and synthesis of biomolecules that sustain their uncontrolled proliferation. Due to the conceptual progresses in the last decade, metabolic reprogramming was recently included as one of the new hallmarks of cancer. The advent of high-throughput technologies to amass an abundance of omic data, together with the development of new computational methods that allow the integration and analysis of omic data by using genome-scale reconstructions of human metabolism, have increased and accelerated the discovery and development of anticancer drugs and tumor-specific metabolic biomarkers. Here we review and discuss the latest advances in the context of metabolic reprogramming and the future in cancer research.

Longitudinal omics modeling and integration in clinical metabonomics research: challenges in childhood metabolic health research

Systems biology is an important approach for deciphering the complex processes in health maintenance and the etiology of metabolic diseases. Such integrative methodologies will help better understand the molecular mechanisms involved in growth and development throughout childhood, and consequently will result in new insights about metabolic and nutritional requirements of infants, children and adults. To achieve this, a better understanding of the physiological processes at anthropometric, cellular and molecular level for any given individual is needed. In this respect, novel omics technologies in combination with sophisticated data modeling techniques are key. Due to the highly complex network of influential factors determining individual trajectories, it becomes imperative to develop proper tools and solutions that will comprehensively model biological information related to growth and maturation of our body functions. The aim of this review and perspective is to evaluate, succinctly, promising data analysis approaches to enable data integration for clinical research, with an emphasis on the longitudinal component. Approaches based on empirical and mechanistic modeling of omics data are essential to leverage findings from high dimensional omics datasets and enable biological interpretation and clinical translation. On the one hand, empirical methods, which provide quantitative descriptions of patterns in the data, are mostly used for exploring and mining datasets. On the other hand, mechanistic models are based on an understanding of the behavior of a system's components and condense information about the known functions, allowing robust and reliable analyses to be performed by bioinformatics pipelines and similar tools. Herein, we will illustrate current examples, challenges and perspectives in the applications of empirical and mechanistic modeling in the context of childhood metabolic health research.

Drug target identification through systems biology

To rationalise drug target selection, we should understand the role of putative targets in biological pathways quantitatively. We review here how experimental and computational network-based approaches can aid more rational drug target selection and illustrate this with results obtained for microbes and for cancer. Comparison of the drug response of biochemical networks in target cells and (healthy) host cells can reveal network-selective targets.

Using Genome-scale Models to Predict Biological Capabilities

Constraint-based reconstruction and analysis (COBRA) methods at the genome scale have been under development since the first whole-genome sequences appeared in the mid-1990s. A few years ago, this approach began to demonstrate the ability to predict a range of cellular functions, including cellular growth capabilities on various substrates and the effect of gene knockouts at the genome scale. Thus, much interest has developed in understanding and applying these methods to areas such as metabolic engineering, antibiotic design, and organismal and enzyme evolution. This Primer will get you started.

Human metabolic atlas: an online resource for human metabolism

Human tissue-specific genome-scale metabolic models (GEMs) provide comprehensive understanding of human metabolism, which is of great value to the biomedical research community. To make this kind of data easily accessible to the public, we have designed and deployed the human metabolic atlas (HMA) website (http://www.metabolicatlas.org). This online resource provides comprehensive information about human metabolism, including the results of metabolic network analyses. We hope that it can also serve as an information exchange interface for human metabolism knowledge within the research community. The HMA consists of three major components: Repository, Hreed (Human REaction Entities Database) and Atlas. Repository is a collection of GEMs for specific human cell types and human-related microorganisms in SBML (System Biology Markup Language) format. The current release consists of several types of GEMs: a generic human GEM, 82 GEMs for normal cell types, 16 GEMs for different cancer cell types, 2 curated GEMs and 5 GEMs for human gut bacteria. Hreed contains detailed information about biochemical reactions. A web interface for Hreed facilitates an access to the Hreed reaction data, which can be easily retrieved by using specific keywords or names of related genes, proteins, compounds and cross-references. Atlas web interface can be used for visualization of the GEMs collection overlaid on KEGG metabolic pathway maps with a zoom/pan user interface. The HMA is a unique tool for studying human metabolism, ranging in scope from an individual cell, to a specific organ, to the overall human body. This resource is freely available under a Creative Commons Attribution-NonCommercial 4.0 International License.

Context-Specific Metabolic Model Extraction Based on Regularized Least Squares Optimization

Genome-scale metabolic models have proven highly valuable in investigating cell physiology. Recent advances include the development of methods to extract context-specific models capable of describing metabolism under more specific scenarios (e.g., cell types). Yet, none of the existing computational approaches allows for a fully automated model extraction and determination of a flux distribution independent of user-defined parameters. Here we present RegrEx, a fully automated approach that relies solely on context-specific data and ℓ₁-norm regularization to extract a context-specific model and to provide a flux distribution that maximizes its correlation to data. Moreover, the publically available implementation of RegrEx was used to extract 11 context-specific human models using publicly available RNAseq expression profiles, Recon1 and also Recon2, the most recent human metabolic model. The comparison of the performance of RegrEx and its contending alternatives demonstrates that the proposed method extracts models for which both the structure, i.e., reactions included, and the flux distributions are in concordance with the employed data. These findings are supported by validation and comparison of method performance on additional data not used in context-specific model extraction. Therefore, our study sets the ground for applications of other regularization techniques in large-scale metabolic modeling.

Modeling cancer metabolism on a genome scale

Cancer cells have fundamentally altered cellular metabolism that is associated with their tumorigenicity and malignancy. In addition to the widely studied Warburg effect, several new key metabolic alterations in cancer have been established over the last decade, leading to the recognition that altered tumor metabolism is one of the hallmarks of cancer. Deciphering the full scope and functional implications of the dysregulated metabolism in cancer requires both the advancement of a variety of omics measurements and the advancement of computational approaches for the analysis and contextualization of the accumulated data. Encouragingly, while the metabolic network is highly interconnected and complex, it is at the same time probably the best characterized cellular network. Following, this review discusses the challenges that genome-scale modeling of cancer metabolism has been facing. We survey several recent studies demonstrating the first strides that have been done, testifying to the value of this approach in portraying a network-level view of the cancer metabolism and in identifying novel drug targets and biomarkers. Finally, we outline a few new steps that may further advance this field.

Flux balance analysis predicts essential genes in clear cell renal cell carcinoma metabolism

Flux balance analysis is the only modelling approach that is capable of producing genome-wide predictions of gene essentiality that may aid to unveil metabolic liabilities in cancer. Nevertheless, a systemic validation of gene essentiality predictions by flux balance analysis is currently missing. Here, we critically evaluated the accuracy of flux balance analysis in two cancer types, clear cell renal cell carcinoma (ccRCC) and prostate adenocarcinoma, by comparison with large-scale experiments of gene essentiality in vitro. We found that in ccRCC, but not in prostate adenocarcinoma, flux balance analysis could predict essential metabolic genes beyond random expectation. Five of the identified metabolic genes, AGPAT6, GALT, GCLC, GSS, and RRM2B, were predicted to be dispensable in normal cell metabolism. Hence, targeting these genes may selectively prevent ccRCC growth. Based on our analysis, we discuss the benefits and limitations of flux balance analysis for gene essentiality predictions in cancer metabolism, and its use for exposing metabolic liabilities in ccRCC, whose emergent metabolic network enforces outstanding anabolic requirements for cellular proliferation.

Applications of genome-scale metabolic network model in metabolic engineering

Genome-scale metabolic network model (GEM) is a fundamental framework in systems metabolic engineering. GEM is built upon extensive experimental data and literature information on gene annotation and function, metabolites and enzymes so that it contains all known metabolic reactions within an organism. Constraint-based analysis of GEM enables the identification of phenotypic properties of an organism and hypothesis-driven engineering of cellular functions to achieve objectives. Along with the advances in omics, high-throughput technology and computational algorithms, the scope and applications of GEM have substantially expanded. In particular, various computational algorithms have been developed to predict beneficial gene deletion and amplification targets and used to guide the strain development process for the efficient production of industrially important chemicals. Furthermore, an Escherichia coli GEM was integrated with a pathway prediction algorithm and used to evaluate all possible routes for the production of a list of commodity chemicals in E. coli. Combined with the wealth of experimental data produced by high-throughput techniques, much effort has been exerted to add more biological contexts into GEM through the integration of omics data and regulatory network information for the mechanistic understanding and improved prediction capabilities. In this paper, we review the recent developments and applications of GEM focusing on the GEM-based computational algorithms available for microbial metabolic engineering.

Context-specific metabolic network reconstruction of a naphthalene-degrading bacterial community guided by metaproteomic data

MOTIVATION

With the advent of meta-'omics' data, the use of metabolic networks for the functional analysis of microbial communities became possible. However, while network-based methods are widely developed for single organisms, their application to bacterial communities is currently limited.

RESULTS

Herein, we provide a novel, context-specific reconstruction procedure based on metaproteomic and taxonomic data. Without previous knowledge of a high-quality, genome-scale metabolic networks for each different member in a bacterial community, we propose a meta-network approach, where the expression levels and taxonomic assignments of proteins are used as the most relevant clues for inferring an active set of reactions. Our approach was applied to draft the context-specific metabolic networks of two different naphthalene-enriched communities derived from an anthropogenically influenced, polyaromatic hydrocarbon contaminated soil, with (CN2) or without (CN1) bio-stimulation. We were able to capture the overall functional differences between the two conditions at the metabolic level and predict an important activity for the fluorobenzoate degradation pathway in CN1 and for geraniol metabolism in CN2. Experimental validation was conducted, and good agreement with our computational predictions was observed. We also hypothesize different pathway organizations at the organismal level, which is relevant to disentangle the role of each member in the communities. The approach presented here can be easily transferred to the analysis of genomic, transcriptomic and metabolomic data.