Network Analysis

Finally, the different pathways are parts of cellular networks. Here the challenge is to define individual pathways in such a network in a clear and mathematical way. This is necessary both for a concise description of the network capabilities and for prediction of the effects of inhibition of a particular enzyme: in many instances alternative routes through the network still allow most of the metabolites to be produced. An enzyme inhibitor which can be compensated by the metabolic network is in most cases compatible with life in prokaryotic cells. Genes which encode such nonessential enzymes are themselves nonessential and lead to a nonlethal phenotype. The nonessential enzymes for a given metabolic network can easily be identified by calculation of the elementary modes. These are nonde-composable ("elementary") sets of enzymes. Each set can sustain a steady state for all internal metabolites used by this set of enzymes as substrates and products. The external metabolites used by each enzyme set need not fulfil this equilibrium condition. Using this mathematical requirement, the software METATOOL [24] calculates all elementary modes for a given metabolic network. Testing conditions are included in this program so that the stable flux modes calculated are elementary, i.e., nondecomposable in subsets which fulfil the steady state condition for internal metabolites. This helps to answer the above questions: any observed network flux state is always a linear combination of the elementary modes. Inhibition of a given enzyme will thus inhibit exactly those elementary modes in which this enzyme occurs [25]. Recent research from our laboratory shows that this perspective of enzymes directing metabolite flows can also be turned around: metabolites also shape the way in which pathways evolve. This is strikingly shown by the observation that metabolite networks tend to be driven in structure and enzyme substrate specificities by the most frequently represented metabolites of this network. This helps and partly explains the observed widespread recruitment of enzymes to new pathways, allowing pathogens to rapidly change and adapt to hostile environments including xenobiotics [26] and antibiotics.

In this way connections and regulatory networks may be sketched, in particular induction of antibiotic resistance and pathways involved in signal transmission and activation of pathogenicity factors. For instance, subtle genome variations within the structure of a 150-kbp pathogenicity island change Enterococcus strains in terms of virulence and most known auxiliary traits that enhance virulence of the organism [27]. Genome-based approaches also identify new targets associated with disease which are not apparent in the commensal behavior of harmless enterococci. Methicillin-resistant Staphylococcus aureus (MRSA) can be identified by typing the Spa ("Staphylococcus aureus encoded protein A") gene. Efficient soft ware for this is available, allowing rapid determination of the dynamics of resistant clones in a hospital setting [28].

After networks have been identified, specific proteins can be singled out as the most promising targets for drug design. A number of bioinformatic strategies such as homology modeling, pharmacophores, and virtual ligand screening are at our disposal today. New work targeted against resistance development also exploits new approaches to differentiating between target structures and related structures in the human host, e.g., rapid identification of structural domains (e.g., using ref. [13]) on a genome scale with subsequent targeting of parasite specific protein domains.

0 0

Post a comment