Structure and Function of Metabolic Networks
Metabolism, the conversion of nutrients into biomass and energy, is executed by a complex network of reactions, catalyzed for the most part by metabolic enzymes. As metabolism is vital for all living organisms at every stage of life, metabolic perturbations, leading to metabolic imbalances can have detrimental consequences. Metabolic perturbations can arise due to environmental changes such as diet or drug exposure or through genetic changes. Many diseases ranging from simple, inborn errors of metabolism, to more complex, type 2 diabetes, obesity and cardiovascular disease, as well as a variety of cancers, have been associated with mutations in metabolic genes leading to changes in metabolic processes. Therefore, organisms employ multiple strategies to deal with these imbalances or perturbations.
While many metabolic enzymes and reactions have been studied in isolation, systematic studies of metabolic gene regulation encompassing the entire network for an organism has been limited. The overarching goal of the Walhout lab is to understand how the metabolic network functions to maintain homeostasis in the organism and how the network fluctuates to compensate under changing conditions. We mainly use the small nematode Caenorhabditis elegans (C. elegans), a soil-dwelling worm, to answer this big question, navigating nimbly between large-scale, systems-level and deep mechanistic studies, combining experimental approaches with bioinformatics and computational modeling.
One major mechanism used to regulate metabolism is the transcriptional regulation of metabolic enzymes. A major focus of the lab is to understand not only how each enzyme is regulated at the transcriptional level but how their regulation is interconnected with that of other metabolic enzymes both at the pathway and the network level and how animals change metabolic gene expression to deal with metabolic perturbations. To address this current projects in the lab focus on:
Computational and Metabolic Network Construction
Understanding which reactions are active at a particular time and place is a key element to determining how the metabolic network is regulated. We reconstructed the first C. elegans metabolic network to generate a genome-scale metabolic network model (MNM) that can be used to computationally model whole animal metabolism using flux balance analysis (FBA)-based methods (YILMAZ 2016) as well as tissue level predictions (YILMAZ, LI 2020). These analyses allow prediction of network activity at the pathway and network levels under various conditions. We continually add and refine this model. The current model is available here: WormFlux.
By manually annotating metabolic enzymes and integrating them into pathways, we have also developed an online catalog of C. elegans metabolic pathway maps. These maps can be used for pathway enrichment analysis in different datasets and are available on our WormFlux website.
To determine the significance of transcriptional regulation driving metabolic network regulation, we recently completed a computational study to determine how much of C. elegans metabolism is under transcriptional control. To this end, we used available gene expression data throughout development and in different tissues, as well as a compendium of more than 150 published gene expression datasets obtained under different conditions. By using variations in mRNA levels as a proxy for gene expression changes, we found that metabolic genes are under as much transcriptional control as non-metabolic genes, and that genes associated with the same WormPaths pathways tend to be co-regulated (NANDA MSB 2023).
From Yilmaz & Walhout, 2016. A C. elegans Genome-Scale Metabolic Network Model
From Nanda et al., 2023, Systems-level transcriptional regulation of Caenorhabditis elegans metabolism
Metabolic Modulation by Diet, Drugs and Genetics
Diet
C. elegans consume a bacterial diet. The nutritional make-up of different bacterial species is highly variable. As metabolism depends on nutrient uptake, different bacterial species could affect worm metabolism in different ways. Using this model of nutritional input on host metabolism we have identified metabolites synthesized by the bacteria that alter the metabolic network leading to changes in animal development and homeostasis. As an example, we have found that when worms are fed the soil bacteria Comamonas aquatica 1877 (hereafter referred to as Comamonas), animals develop faster, have reduced fecundity, and a shorter lifespan than when they are fed the standard laboratory diet of E. coli, OP50 (MACNEIL CELL 2013). Using forward and reverse genetics in the worm we discovered a set of metabolic genes that, are expressed when the animals are fed OP50 but not expressed when they are fed Comamonas (WATSON, MACNEIL CELL2013). Reasoning that there was something specific in the bacteria themselves that could be responsible for the differential gene expression, again using a genetic approach we found that Comamonas supplies the worm with the metabolite, vitamin B12 which E. coli does not synthesize (WATSON CELL 2014).
Vitamin B12 is a cofactor in two metabolic pathways in both humans and worms, the degradation of the short chain fatty acid propionate, and the recycling of methionine in the methionine/S-adenosylmethionine (Met/SAM) pathway, the major methyl generating pathway in the worm. From our differential diet analysis discussed above we discovered an alternate propionate breakdown pathway we named the propionate shunt. This shunt is transcriptionally activated when vitamin B12 is low, as when worms are fed an E. coli diet, or when the animal harbors genetic perturbations in the canonical propionate breakdown pathway (WATSON ELIFE 2016). Through a TF knockdown screen based on the expression of a metabolic gene reporter we found that the propionate shunt is activated by two nuclear receptors (NHRs), nhr-10 and nhr-68, which respond to high persistent propionate levels, a type of regulatory circuit known as a persistence detector (see BULCHA CELL REPORTS 2019 for more information). Finally, we discovered that Met/SAM cycle genes are also activated in response to low vitamin B12 and that this involves yet another NHR, nhr-114 (GIESE ELIFE 2020).
These studies provided one of the first insights into the transcriptional regulation of metabolism by metabolic perturbations. They also showed that our model can uncover metabolic pathways that have not previously been identified.
For additional related studies please see ZHANG CHM 2019 and MIRZA CELL REPORTS 2023.
From Giese et al., 2020, C. elegans methionine/S-adenosylmethionine cycle activity is sensed and adjusted by a nuclear hormone receptor
From Garcia-Gonzalez et al., 2017, Worms, bugs and drugs: Caenorhabditis elegans as a model for host-microbe-drug interactions
Drugs
Our worm-diet model has also provided valuable insight into how bacteria, such as those that inhabit our gut (the microbiome), influences drug metabolism which in turn effects the toxicity exerted by the drug. We discovered that C. elegans is resistant to the chemotherapeutic drug FUDR when fed Comamonas, but sensitive when fed E. coli. Interestingly, while FUDR was more toxic to animals fed E. coli, the related drug 5FU was toxic, but not affected by bacterial diet (all compared to Comamonas). By using bacterial mutants, we discovered that 5FU and FUDR are converted by bacteria into 5FUMP, which is toxic to the animal, underscoring how bacterial metabolism can affect the host in which it is associated (GARCIA-GONZALEZ CELL 2017).
We also found that bacterial diet greatly affects toxicity of tamoxifen, a drug used to treat breast cancer in humans. Here we found that tamoxifen is especially toxic on animals fed Bacillus subtilis and that this toxicity can be explained, at least in part, by the different fatty acids provided by the bacteria. This also provided insights into the animals fatty acid content, which mimics that of the bacteria it eats (you are what you eat) (DIOT NAT COMM 2022).
From Diot et al., 2022, Bacterial diet modulates tamoxifen-induced death via host fatty acid metabolism
Genetics
While IEMs are single gene disorders that are deemed genetically ‘simple’, the onset and morbidity of these diseases is highly variable and greatly influenced by genetic background, diet, and environmental factors. Therefore, they are actually more complicated than originally posited. In fact, propionic acidemia, an IEM associated with the inability to breakdown certain proteins and fats, is caused by mutations in the first step of the canonical, vitamin B12-dependent propionate breakdown pathway, which we have characterized as described above.
Additionally, through genetic screens we identified several mutants in C. elegans homologs of human genes that, when mutated, cause IEMs, providing valuable models to study the etiology of these diseases. One such gene is mccc-1, which encodes an enzyme in the breakdown of the branched chain amino acid, leucine. Through a combination of metabolomics and gene expression profiling, we found that mccc-1 mutant worms detoxify leucine breakdown intermediates by conjugating them to several different molecules, some of which are novel! In addition, mevalonate, an essential molecule that is generated from leucine breakdown products, cannot be made in mccc-1 mutant animals. Interestingly, we found that mccc-1 mutant worms upregulate ketone body metabolism genes, which likely serves to generate acetyl-CoA for the synthesis of mevalonate, thereby rewiring metabolism to compensate for defects in the canonical leucine breakdown pathway. (LEE NAT METABOLISM 2024).
We have also studied additional models of IEMs using worms including worms with mutations in dhgd-1, a D-2-hydroxyglutarate (D2HG) dehydrogenase, associated with D-2-hydroxyglutaric aciduria. We found this enzyme to function in the propionate shunt to recycle D2HG back to a-ketoglutarate and the mutant animals have impaired ketone body synthesis leading to reduced animal viability. (PONOMAROVA PLOS BIO 2023). Additionally, studies using a model for human IDH1-neo, a proposed oncometabolite that is associated with increased levels of D2HG, identified metabolic rewiring in the mutant animals which leads to an increased dependence on the one-carbon pool for survival.(PONOMAROVA LIFE SCI ALLIANCE 2024).
From Ponomarova et al., 2023. A D-2-hydroxyglutarate dehydrogenase mutant reveals a critical role for ketone body metabolism in Caenorhabditis elegans development
From Lee et al., 2023. Host-microbe interactions rewire metabolism in a C. elegans model of leucine breakdown deficiency
Inter-individual metabolic variation
In collaboration with the labs of Frank Schroeder and Erik Andersen, we have used four C. elegans strains, the N2 lab strain and three wild isolates with complete genome sequences, to study how metabolism can vary among individuals. This was motivated by the emergence of ‘precision medicine’, which hopes to use individual ‘omic’ data to predict health, disease, and therapy. We found many novel metabolites in these four strains, some of which we validated and characterized further. These include novel conjugates between 3-hydroxypropionate (3HP), a metabolite uniquely generated by the propionate shunt, and different amino acids. We found these specifically in one of the four strains and subsequently found that these animals harbor a coding mutation in the hphd-1 gene, which encodes the enzyme that converts 3HP in the propionate shunt. We then used CRISPR-Cas9 to demonstrate that this mutation is causative for the generation of 3HP-amino acyl conjugates (FOX NATURE 2022) .
From Fox et al. C. elegans as a model for inter-individual variation in metabolism
Older Work
We have extensively annotated and characterized C. elegans and, to a lesser extent, human TFs (REECE-HOYES GENOME BIOL 2005; MACNEIL CELL SYSTEMS 2015; FUXMAN BASS MSB 2016; FUXMAN BASS CELL 2015). In addition, we have developed and used high-throughput yeast one-hybrid assays to identify and study TF-promoter interactions (DEPLANCKE CELL 2006; ARDA MSB 2010; MARTINEZ G&D 2008), and compared molecular interactions for C. elegans TFs (GROVE CELL 2009; REECE-HOYES MOL CELL 2013). Our knowledge and data of worm TFs provide a great platform for mechanistic studies of how metabolism regulates gene expression, and vice versa.
From MacNeil et al., 2015. In vivo gene regulatory network based on transcription factor activity.