Trends in Biotechnology
Genome-scale microbial in silico models: the constraints-based approach
Section snippets
The constraints-based modeling approach
The challenges of genome-scale model building are being met, in part, by constraints-based models 19, 24. This modeling process involves a multi-step procedure (Fig. 1). The first step is to reconstruct the underlying network 25, 26, 27, 28, 29. For metabolism, this is a well-established procedure [25], whereas methods for the reconstruction of the associated regulatory networks are being developed 30, 31. The second step involves the statement of the constraints under which the reconstructed
Current genome-scale metabolic models
Genome-scale constraints-based models of metabolism have been built for several organisms, and some have appeared in the literature, including for Escherichia coli 45, 46, Haemophilus influenzae [47], Helicobacter pylori [48], and Saccharomyces cerevisiae [49]. Others, including Bacillus subtilis, Pseudomonas aeruginosa, and Pseudomonas putida have been fully built but not yet published (Sung Park and Jeremy Edwards, unpublished results).
‘Genome-scale’ is used to describe these models because
Simulating the results of manipulating gene content and function
There are many experimental methods for changing the gene content of an organism. Genes can be added or deleted, or their functions impaired or enhanced. Reliable computational models that link genotype to phenotype would thus allow for directed manipulation of the gene content of an organism to obtain a desired phenotype. Genome-scale constraints-based models can provide such a link. The ability to predict phenotypic outcomes from genetic inputs would provide a basis for the rational selection
Predictions of optimal growth rates
Experimental validation of in silico predictions has provided increasing evidence that a primary function of the E. coli metabolic network is to maximize growth 42, 44. E. coli was grown on several substrates, including acetate, succinate, malate or glucose minimal media, and the corresponding uptake rates, secretion rates and growth rates were experimentally measured. Good correlation was obtained between the growth rates, uptake rates and secretion rates that were experimentally observed and
Computing minimal reaction sets
A recent study used the E. coli metabolic model to enumerate the number of reactions necessary to maintain the capacity of the metabolic network to produce all of the biomass components necessary for growth [40]. It was discovered that the E. coli metabolic network was able to support growth using only 31% of its known metabolic reactions for growth on glucose and only 17% of its metabolic reactions on a rich medium. Thus, E. coli's metabolic network contains many redundant reactions, making E.
Accounting for regulation
The genome-scale models discussed above are based on network topology and the analysis assumes their unfettered use to achieve assumed optimal performance. Cells use complex regulatory networks to achieve their goals that might or might not be consistent with assumptions of optimal performance. Thus, significant need exists to account for regulation in genome-scale models and initial progress in this regard is being made [59].
Beyond metabolism: constraints-based modeling as means to integrate ‘omics’ data
Genome-scale models provide a framework in which high-throughput biological data can be integrated and thus broadens our capacity to predict phenotypes. In silico methods are being developed that allow for the integration of heterogeneous data sets, including genomics, proteomics, transcriptomics and metabolomics 17, 61 (Fig. 2). In the case of E. coli, the inclusion of metabolism, transcription, translation and regulation will lead to models that account for ∼2000 open reading frames. Such a
In closing
Genome-scale models of metabolism have been developed for several microbial cells. Many useful results have already been derived from these models and as they grow in scope and validation, an even broader set of uses can be anticipated. Perhaps most important is the integration of diverse omics data and the ability to account for the management of the genome and associated regulatory processes. Eventually, such models are expected to become routinely used for a variety of biotechnological
Acknowledgments
We acknowledge the support of the National Science Foundation (BES 01–20363), the National Institutes of Health (GM 57089), and the Whitaker Foundation (Graduate Research Fellowship to JP).
References (61)
- et al.
Metabolic dynamics in the human red cell. Part I–A comprehensive kinetic model
J. Theor. Biol.
(1989) Description and analysis of metabolic connectivity and dynamics in the human red blood cell
Biophys. J.
(2002)- et al.
Extreme pathway analysis of human red blood cell metabolism
Biophys. J.
(2002) - et al.
The virtual cell: a software environment for computational cell biology
Trends Biotechnol.
(2001) - et al.
Mathematical models of metabolic pathways
Curr. Opin. Biotechnol.
(1999) Metabolic modeling of microbial strains in silico
Trends Biochem. Sci.
(2001)Metabolic databases
Trends Biochem. Sci.
(1998)- et al.
Transcriptional regulation in constraints-based metabolic models of Escherichia coli
J. Biol. Chem.
(2002) - et al.
Deciphering gene expression regulatory networks
Curr. Opin. Genet. Dev.
(2002) Energy balance for analysis of complex metabolic networks
Biophys. J.
(2002)