Construction and elementary mode analysis of a metabolic model for Shewanella oneidensis MR-1

Abstract

A stoichiometric model describing the central metabolism of Shewanella oneidensis MR-1 wild-type and derivative strains was developed and used in elementary mode analysis (EMA). Shewanella oneidensis MR-1 can anaerobically respire a diverse pool of electron acceptors, and may be applied in several biotechnology settings, including bioremediation of toxic metals, electricity generation in microbial fuel cells, and whole-cell biocatalysis. The metabolic model presented here was adapted and verified by comparing the growth phenotypes of 13 single- and 1 double-knockout strains, while considering respiration via aerobic, anaerobic fumarate, and anaerobic metal reduction (Mtr) pathways, and utilizing acetate, n-acetylglucosamine (NAG), or lactate as carbon sources. The gene ppc, which encodes phosphoenolpyruvate carboxylase (Ppc), was determined to be necessary for aerobic growth on NAG and lactate, while not essential for growth on acetate. This suggests that Ppc is the only active anaplerotic enzyme when cultivated on lactate and NAG. The application of regulatory and substrate limitations to EMA has enabled creation of metabolic models that better reflect biological conditions, and significantly reduce the solution space for each condition, facilitating rapid strain optimization. This wild-type model can be easily adapted to include utilization of different carbon sources or secretion of different metabolic products, and allows the prediction of single- and multiple-knockout strains that are expected to operate under defined conditions with increased efficiency when compared to wild type cells.

Introduction

Species of the genus Shewanella have been known to respire a wide variety of extra-cellular compounds during growth under anaerobic conditions, including many soluble and insoluble substrates, such as Fe(III), Mn(III and IV), U(VI), Co(III), fumarate, and dimethyl sulfoxide, to name a few (Hau and Gralnick, 2007, Hau et al., 2008). The ability to transfer electrons to metal ions and electrodes is facilitated by the metal reduction (Mtr) respiratory pathway (Coursolle et al., 2010, Shi et al., 2007). Briefly, the Mtr respiratory pathway is a complex of membrane proteins that commonly transfers electrons from membrane quinols to external electron acceptors, which can include metal surfaces, electrodes, or external flavins (Hartshorne et al., 2007, Marsili et al., 2008). S. oneidensis MR-1 is a non-fermenting facultative anaerobic γ-proteobacterium isolated from Lake Oneida in New York (Hau and Gralnick, 2007).

The ability of Shewanellae to anaerobically respire a diverse pool of electron acceptors is central to several potential applications in biotechnology, including bioremediation of oxidized metals, electricity generation in microbial fuel cells, and whole-cell biocatalysis applications (Bretschger et al., 2007, Carpentier et al., 2003, Flynn et al., 2010, Marshall et al., 2006). As a result, the development of metabolic models that describe the S. oneidensis anaerobic metabolism is vital to achieving improved product yield, biomass production, and other strain optimization or customization. To this end, the generation and verification of an accurate metabolic model is essential to both improving understanding of Shewanella oneidensis MR-1 metabolism and to enable the optimization of the metabolic network for biotechnology applications. Currently available metabolic network descriptions, such as those in the Kyoto encyclopedia of genes and genomes (KEGG) and Metacyc databases are based on gene annotation. These sources are valuable starting points for model construction, but models made without additional constraints considering conditional enzyme activity or expression may overestimate the number of pathways available under a given set of conditions. This in turn may result in unnecessary gene deletion predictions when attempting to improve product yield. Therefore, it is vital to establish which pathways contain significant carbon or energy flux under the desired conditions while keeping network complexity to a minimum. Publicly available networks may improve conditional pathway accuracy by combining current gene annotation technology with quantitative RNA sequencing to compare key gene expression between standard growth conditions and correlate the results with enzyme activity in a binary way. As RNA-sequencing costs continue to decrease, it may soon be economical to test a subset of organisms in a standardized way to more accurately describe complex reaction networks under common growth conditions.

In this work, we sought to combine available data describing the central metabolism of S. oneidensis under common cultivation conditions to create a metabolic model that describes the metabolism of wild-type cells. To this end, elementary mode analysis (EMA) was utilized to generate testable viability predictions under several cultivation conditions using knockout mutants to evaluate the accuracy of the metabolic model. Previous research describing carbon flux characteristics, enzyme activity assays, and experiments regarding consumption, secretion, and viability of gene knockout mutants were combined with central metabolism information obtained from the KEGG and Metacyc databases to generate a wild-type model describing the S. oneidensis central metabolism under several common culture conditions. This strain is primarily cultivated on lactate, but is also known to oxidize n-acetylglucosamine (NAG) and acetate as carbon and energy sources. Due to the different assimilation mechanisms of these three carbon sources, the growth phenotypes of mutant S. oneidensis on these carbon sources under aerobic and anaerobic conditions has been used to shed light on its central metabolism.

A stoichiometric model of the S. oneidensis is the basis for any quantitative network evaluation such as elementary mode analysis or flux balance analysis. EMA computes all possible, non-divisible pathways through a metabolic network specified by a set of stoichiometric reactions with stated reaction reversibility or irreversibility, assuming pseudo steady state conditions (Schuster et al., 2002, Klamt and Stelling, 2003). This form of metabolic modeling has previously been shown to be effective in modeling E. coli amino acid metabolism, and designing bacterial strains that efficiently produce biomass, biopolymers, ethanol, carotenoids, and amino acid derivatives (Diniz et al., 2006, Kromer et al., 2006, Schuster et al., 1999, Trinh et al., 2006, Trinh et al., 2008, Unrean et al., 2009). Despite a known genome sequence and extensive physiological studies, no publicly available model with predictive capability has been developed to describe the S. oneidensis central metabolism to date under aerobic, anaerobic fumarate, and anaerobic Mtr pathway utilizing conditions, while growing on the common carbon sources lactate, NAG, and acetate. However, a detailed metabolic model subjected to different analysis methods has been recently published (Pinchuk et al., 2010), which focused on aerobic lactate metabolism.