Matthew A. Oberhardt

Metabolically re-modeling the drug pipeline

Costs for drug development have soared, exposing a clear need for new R&D strategies. Systems biology has meanwhile emerged as an attractive vehicle for integrating omics data and other post-genomic technologies into the drug pipeline. One of the emerging areas of computational systems biology is constraint-based modeling (CBM), which uses genome-scale metabolic models (GSMMs) as platforms for integrating and interpreting diverse omics datasets. Here we review current uses of GSMMs in drug discovery, focusing on prediction of novel drug targets and promising lead compounds. We then expand our discussion to prediction of toxicity and selectivity of potential drug targets. We discuss successes as well as limitations of GSMMs in these areas. Finally, we suggest new ways in which GSMMs may contribute to drug discovery, offering our vision of how GSMMs may re-model the drug pipeline in years to come.

Harnessing the landscape of microbial culture media to predict new organism-media pairings.

Culturing microorganisms is a critical step in understanding and utilizing microbial life. Here we map the landscape of existing culture media by extracting natural-language media recipes into a Known Media Database (KOMODO), which includes >18,000 strain–media combinations, >3300 media variants and compound concentrations (the entire collection of the Leibniz Institute DSMZ repository). Using KOMODO, we show that although media are usually tuned for individual strains using biologically common salts, trace metals and vitamins/cofactors are the most differentiating components between defined media of strains within a genus. We leverage KOMODO to predict new organism–media pairings using a transitivity property (74% growth in new in vitro experiments) and a phylogeny-based collaborative filtering tool (83% growth in new in vitro experiments and stronger growth on predicted well-scored versus poorly scored media). These resources are integrated into a web-based platform that predicts media given an organisms 16S rDNA sequence, facilitating future cultivation efforts.

Glycan Degradation (GlyDeR) Analysis Predicts Mammalian Gut Microbiota Abundance and Host Diet-Specific Adaptations

ABSTRACT Glycans form the primary nutritional source for microbes in the human gut, and understanding their metabolism is a critical yet understudied aspect of microbiome research. Here, we present a novel computational pipeline for modeling glycan degradation (GlyDeR) which predicts the glycan degradation potency of 10,000 reference glycans based on either genomic or metagenomic data. We first validated GlyDeR by comparing degradation profiles for genomes in the Human Microbiome Project against KEGG reaction annotations. Next, we applied GlyDeR to the analysis of human and mammalian gut microbial communities, which revealed that the glycan degradation potential of a community is strongly linked to host diet and can be used to predict diet with higher accuracy than sequence data alone. Finally, we show that a microbe’s glycan degradation potential is significantly correlated (R = 0.46) with its abundance, with even higher correlations for potential pathogens such as the class Clostridia (R = 0.76). GlyDeR therefore represents an important tool for advancing our understanding of bacterial metabolism in the gut and for the future development of more effective prebiotics for microbial community manipulation. IMPORTANCE The increased availability of high-throughput sequencing data has positioned the gut microbiota as a major new focal point for biomedical research. However, despite the expenditure of huge efforts and resources, sequencing-based analysis of the microbiome has uncovered mostly associative relationships between human health and diet, rather than a causal, mechanistic one. In order to utilize the full potential of systems biology approaches, one must first characterize the metabolic requirements of gut bacteria, specifically, the degradation of glycans, which are their primary nutritional source. We developed a computational framework called GlyDeR for integrating expert knowledge along with high-throughput data to uncover important new relationships within glycan metabolism. GlyDeR analyzes particular bacterial (meta)genomes and predicts the potency by which they degrade a variety of different glycans. Based on GlyDeR, we found a clear connection between microbial glycan degradation and human diet, and we suggest a method for the rational design of novel prebiotics. The increased availability of high-throughput sequencing data has positioned the gut microbiota as a major new focal point for biomedical research. However, despite the expenditure of huge efforts and resources, sequencing-based analysis of the microbiome has uncovered mostly associative relationships between human health and diet, rather than a causal, mechanistic one. In order to utilize the full potential of systems biology approaches, one must first characterize the metabolic requirements of gut bacteria, specifically, the degradation of glycans, which are their primary nutritional source. We developed a computational framework called GlyDeR for integrating expert knowledge along with high-throughput data to uncover important new relationships within glycan metabolism. GlyDeR analyzes particular bacterial (meta)genomes and predicts the potency by which they degrade a variety of different glycans. Based on GlyDeR, we found a clear connection between microbial glycan degradation and human diet, and we suggest a method for the rational design of novel prebiotics.

A novel nutritional predictor links microbial fastidiousness with lowered ubiquity, growth rate, and cooperativeness.

Understanding microbial nutritional requirements is a key challenge in microbiology. Here we leverage the recent availability of thousands of automatically generated genome-scale metabolic models to develop a predictor of microbial minimal medium requirements, which we apply to thousands of species to study the relationship between their nutritional requirements and their ecological and genomic traits. We first show that nutritional requirements are more similar among species that co-habit many ecological niches. We then reveal three fundamental characteristics of microbial fastidiousness (i.e., complex and specific nutritional requirements): (1) more fastidious microorganisms tend to be more ecologically limited; (2) fastidiousness is positively associated with smaller genomes and smaller metabolic networks; and (3) more fastidious species grow more slowly and have less ability to cooperate with other species than more metabolically versatile organisms. These associations reflect the adaptation of fastidious microorganisms to unique niches with few cohabitating species. They also explain how non-fastidious species inhabit many ecological niches with high abundance rates. Taken together, these results advance our understanding microbial nutrition on a large scale, by presenting new nutrition-related associations that govern the distribution of microorganisms in nature.

Systems-Wide Prediction of Enzyme Promiscuity Reveals a New Underground Alternative Route for Pyridoxal 5’-Phosphate Production in E. coli

Recent insights suggest that non-specific and/or promiscuous enzymes are common and active across life. Understanding the role of such enzymes is an important open question in biology. Here we develop a genome-wide method, PROPER, that uses a permissive PSI-BLAST approach to predict promiscuous activities of metabolic genes. Enzyme promiscuity is typically studied experimentally using multicopy suppression, in which over-expression of a promiscuous ‘replacer’ gene rescues lethality caused by inactivation of a ‘target’ gene. We use PROPER to predict multicopy suppression in Escherichia coli, achieving highly significant overlap with published cases (hypergeometric p = 4.4e-13). We then validate three novel predicted target-replacer gene pairs in new multicopy suppression experiments. We next go beyond PROPER and develop a network-based approach, GEM-PROPER, that integrates PROPER with genome-scale metabolic modeling to predict promiscuous replacements via alternative metabolic pathways. GEM-PROPER predicts a new indirect replacer (thiG) for an essential enzyme (pdxB) in production of pyridoxal 5’-phosphate (the active form of Vitamin B6), which we validate experimentally via multicopy suppression. We perform a structural analysis of thiG to determine its potential promiscuous active site, which we validate experimentally by inactivating the pertaining residues and showing a loss of replacer activity. Thus, this study is a successful example where a computational investigation leads to a network-based identification of an indirect promiscuous replacement of a key metabolic enzyme, which would have been extremely difficult to identify directly.

Genome-scale modeling and human disease: an overview

A commentary on Genome-scale modeling and human disease by Gianchandani, E. P., and Oberhardt, M. A. The last several decades have seen extraordinary progress in the biomedical sciences. The explosion of sequencing and high-throughput data is both welcome and daunting for the study of human disease: while human disease is increasingly understood to be multi-factorial and systemic, the sheer complexity of the data being generated makes unaided interpretation nearly impossible. Meanwhile, genome-scale modeling (GSM) has emerged as a major scaffold and toolkit for contextualizing rich data, and one especially suited to the thousands-of-datapoints-per-measurement reality of contemporary disease research. The archetypal genome-scale model is the genome-scale metabolic reconstruction (GENRE), a predictive network model that contains up to several thousand metabolic reactions, as well as associated genes and enzymes (but not kinetics, due to the scale) (Oberhardt et al., 2009). Recently available GENREs of human metabolism have opened up enormous avenues in disease research (Duarte et al., 2007; Ma et al., 2007; Thiele et al., 2013), especially when integrated with high-throughput data [for an extensive review, see in this topic: (Blazier and Papin, 2012)]. These models rely on extensive manual curation, and annotating understudied or ambiguous parts of metabolism is critical for improving their predictive power. In an effort to address one of the most difficult-to-annotate areas of metabolism, researchers involved in the human metabolic reconstruction efforts have provided for this topic a large analysis of membrane transporters in human metabolism, including a discussion of how transport impacts multiple human diseases (Sahoo et al., 2014). GENREs are contributing to many areas of disease research, as detailed below, and their scope and influence will increase as a result of such contributions. Systemic metabolic disorders such as obesity and diabetes exact a huge toll in the US and worldwide, and GSMs are increasingly being used for their study. Large-scale models of mitochondria, for example, have helped examine obesity-associated aberrations in mitochondrial fatty acid degradation (Van Eunen et al., 2013) and many other aspects of energy metabolism as reviewed in this topic: (Sangar et al., 2012). Similarly, the human GENRE has been used in a number of studies relevant to metabolic diseases [e.g., building a model of human adipocyte—(Mardinoglu et al., 2013); determining biomarkers for inborn errors of metabolism—(Shlomi et al., 2009)], as extensively reviewed here: (Varemo et al., 2013). GENREs are obvious choices for studying metabolically-based diseases, and will likely be relied on more in the future. Another area of increasing interest in human disease is the impact of the microbial organisms that cohabitate our bodies, collectively known as our “microbiome.” The gut microbiome, for example, has been shown to alter the metabolism of many drugs (Kang et al., 2013), and to be a causative factor in maintaining obese or healthy states (Turnbaugh et al., 2006). GENREs have been used to examine prominent members of the gut microbiota (Heinken et al., 2013), to understand interactions between gut microbes (Shoaie et al., 2013), and to explore interactions between gut microbes and epithelial cells (Sahoo and Thiele, 2013). GSMs are still severely limited in this arena due to challenges in community microbial modeling. However, large-scale microbiome modeling efforts will likely have increasing impact as they mature in the coming years, both by driving new knowledge of complex community phenotypes (e.g., Freilich et al., 2011 and reviewed generally in Greenblum et al., 2013) and by including so-far neglected areas such as the oral microbiome, as reviewed in this topic: (Dimitrov and Hoeng, 2013). Cancer is a complex and multifaceted disease, and a hallmark for huge data collection efforts. As such, it is a natural target for systems modeling [for a general review of systems biology approaches, see in this topic: (Hernandez Patino et al., 2012)]. Metabolic deregulation in cancer has generated considerable interest within the genome-scale metabolic modeling community, resulting in a number of cancer-related metabolic reconstructions being recently published [see a review in this topic: (Lewis and Abdel-Haleem, 2013)]. Models of specific cancer subtypes are now being built based on the generic human GENRE (Jerby et al., 2010; Agren et al., 2012), and in a few cases, they have revealed insights with therapeutic potentiality (Frezza et al., 2011; Agren et al., 2014). Due to their lack of kinetic parameters, GENREs alone cannot predict dynamic cell states, nor, surprisingly, can they integrate metabolite concentration data into basic kinetics or allosteric regulation. Since kinetic parameters are difficult to measure and can vary between conditions or cells, ensemble modeling was recently used to estimate kinetic models of human colorectal adenocarcinoma cell lines, and to reveal potential synthetic lethal interactions that could yield new drug targets [see in this topic: (Khazaei et al., 2012)]. Cancer is also a disease marked by the evolutionary process that the cancerous cells undergo. Genomic data and increasingly sophisticated population models are now enabling elucidation of these processes, which are critical for establishing the basis for cures [see a review in this topic: (Stransky and De Souza, 2012)]. These areas have gained a lot of interest, and we expect many more systems-level studies of cancer in the near future. By contrast, neurological disorders constitute a set of diseases that have not received as much attention in the GSM community, despite the significant impacts illustrated in Figure ​Figure1.1. Early attention focused on genome-wide expression analyses and gene-interaction networks, often using yeast pathways conserved in humans and implicated in neurodegenerative diseases such as Parkinsons, Alzheimers, and Huntingtons (Petranovic and Nielsen, 2008; Noorbakhsh et al., 2009; Wall et al., 2009). More recently, efforts have begun to employ GSM with success. For example, (Lewis et al., 2010) integrated gene expression data, proteomics data, and literature-based manual curation to model brain energy metabolism and recapitulate the metabolic interactions between astrocytes and various neuron types relevant to Alzheimers disease. In addition, transcriptomic data from Alzheimers patients were integrated with a genome-scale computational human metabolic model to characterize the altered metabolism in the diseased state, and metabolic modeling methods were employed to predict metabolic biomarkers and drug targets (Stempler et al., 2014). We expect interest in neurological illnesses to continue to rise. Figure 1 Publications in different disease areas vs. morbidity rates. Age standardized disability adjusted life years (DALY), a measure of years of life lost due to disease, is reported as percent worldwide and in the US. Injury (11.3%) and “other non-communicable ... While much of this short review focuses on GENRE-based analyses, GENREs are by no means the only genome-scale models of note. Many alternative topology-based methods for pathway analysis are available and have been reviewed here: (Mitrea et al., 2013). We also include in this topic a promising new Boolean-based model for somatic cell reprogramming: (Flottmann et al., 2012). Somatic cell reprogramming is a new and highly promising field—it first emerged in 2006 with the landmark paper (Takahashi and Yamanaka, 2006)—that could lead to novel therapeutic approaches, such as growing organs from skin cells for self-transplant. GSM-based analysis is now a key asset in studying disease. The works in this topic reflect trends in the biomedical sciences at large, including areas of intense interest (e.g., cancer) as well as those that have been labeled as neglected diseases (e.g., few models have been built for studying parasitic tropical diseases or HIV/AIDS). Although eukaryote reconstructions are more challenging due to genome sizes, knowledge coverage, and the multitude of cellular compartments (Thiele and Palsson, 2010), we expect the successes described in this overview to continue to mount—with a particular focus in coming years on clinical problems with translatable outcomes, in which models will help identify new drug targets or alternate cures. This is already evident from recent DREAM Challenges, which have sought to foster collaboration and build communities around fundamental questions at the intersection of systems biology and translational medicine [see, for example, (Margolin et al., 2013)]. To help guide and contextualize disease study, we have included a chart of the most devastating diseases, along with the amount of focus in GSM studies as well as in science at large toward addressing them (Figure ​(Figure1).1). Shifting focus toward neglected areas is a worthy goal to which we hope this mini-review will contribute.

Maximal Sum of Metabolic Exchange Fluxes Outperforms Biomass Yield as a Predictor of Growth Rate of Microorganisms

Growth rate has long been considered one of the most valuable phenotypes that can be measured in cells. Aside from being highly accessible and informative in laboratory cultures, maximal growth rate is often a prime determinant of cellular fitness, and predicting phenotypes that underlie fitness is key to both understanding and manipulating life. Despite this, current methods for predicting microbial fitness typically focus on yields [e.g., predictions of biomass yield using GEnome-scale metabolic Models (GEMs)] or notably require many empirical kinetic constants or substrate uptake rates, which render these methods ineffective in cases where fitness derives most directly from growth rate. Here we present a new method for predicting cellular growth rate, termed SUMEX, which does not require any empirical variables apart from a metabolic network (i.e., a GEM) and the growth medium. SUMEX is calculated by maximizing the SUM of molar EXchange fluxes (hence SUMEX) in a genome-scale metabolic model. SUMEX successfully predicts relative microbial growth rates across species, environments, and genetic conditions, outperforming traditional cellular objectives (most notably, the convention assuming biomass maximization). The success of SUMEX suggests that the ability of a cell to catabolize substrates and produce a strong proton gradient enables fast cell growth. Easily applicable heuristics for predicting growth rate, such as what we demonstrate with SUMEX, may contribute to numerous medical and biotechnological goals, ranging from the engineering of faster-growing industrial strains, modeling of mixed ecological communities, and the inhibition of cancer growth.

Essential Genes Embody Increased Mutational Robustness to Compensate for the Lack of Backup Genetic Redundancy

Genetic robustness is a hallmark of cells, occurring through many mechanisms and at many levels. Essential genes lack the common robustness mechanism of genetic redundancy (i.e., existing alongside other genes with the same function), and thus appear at first glance to leave cells highly vulnerable to genetic or environmental perturbations. Here we explore a hypothesis that cells might protect against essential gene loss through mechanisms that occur at various cellular levels aside from the level of the gene. Using Escherichia coli and Saccharomyces cerevisiae as models, we find that essential genes are enriched over non-essential genes for properties we call “coding efficiency” and “coding robustness”, denoting respectively a gene’s efficiency of translation and robustness to non-synonymous mutations. The coding efficiency levels of essential genes are highly positively correlated with their evolutionary conservation levels, suggesting that this feature plays a key role in protecting conserved, evolutionarily important genes. We then extend our hypothesis into the realm of metabolic networks, showing that essential metabolic reactions are encoded by more “robust” genes than non-essential reactions, and that essential metabolites are produced by more reactions than non-essential metabolites. Taken together, these results testify that robustness at the gene-loss level and at the mutation level (and more generally, at two cellular levels that are usually treated separately) are not decoupled, but rather, that cellular vulnerability exposed due to complete gene loss is compensated by increased mutational robustness. Why some genes are backed up primarily against loss and others against mutations still remains an open question.

Taming the complexity of large models.

At its most basic, science is about models. Natural phenomena that were perplexing to ancient humans have been systematically illuminated as scientific models have revealed the mathematical order underlying the natural world. But what happens when the models themselves become complex enough that they too must be interpreted to be understood? In 2012, Jonathan Karr, Markus Covert and colleagues at the University of California, San Diego (USA) produced a bold new biological model that attempts to simulate an entire cell: iMg [1]. iMg merges 28 sub‐modules of processes within Mycobacterium genitalium , one of the simplest organisms known to man. As a systems biology big‐data model, iMg is unique in its scope and is an undeniable paragon of good craft. Because it is probable that this landmark paper will soon be followed by other whole cell models, we feel it is timely to examine this important endeavour, its challenges and potential pitfalls. Building a model requires making many decisions, such as which …

Data-Driven Metabolic Pathway Compositions Enhance Cancer Survival Prediction.

Altered cellular metabolism is an important characteristic and driver of cancer. Surprisingly, however, we find here that aggregating individual gene expression using canonical metabolic pathways fails to enhance the classification of noncancerous vs. cancerous tissues and the prediction of cancer patient survival. This supports the notion that metabolic alterations in cancer rewire cellular metabolism through unconventional pathways. Here we present MCF (Metabolic classifier and feature generator), which incorporates gene expression measurements into a human metabolic network to infer new cancer-mediated pathway compositions that enhance cancer vs. adjacent noncancerous tissue classification across five different cancer types. MCF outperforms standard classifiers based on individual gene expression and on canonical human curated metabolic pathways. It successfully builds robust classifiers integrating different datasets of the same cancer type. Reassuringly, the MCF pathways identified lead to metabolites known to be associated with the pertaining specific cancer types. Aggregating gene expression through MCF pathways leads to markedly better predictions of breast cancer patients’ survival in an independent cohort than using the canonical human metabolic pathways (C-index = 0.69 vs. 0.52, respectively). Notably, the survival predictive power of individual MCF pathways strongly correlates with their power in predicting cancer vs. noncancerous samples. The more predictive composite pathways identified via MCF are hence more likely to capture key metabolic alterations occurring in cancer than the canonical pathways characterizing healthy human metabolism.