- Research article
- Open Access
A systems biology framework for modeling metabolic enzyme inhibition of Mycobacterium tuberculosis
© Fang et al; licensee BioMed Central Ltd. 2009
- Received: 5 April 2009
- Accepted: 15 September 2009
- Published: 15 September 2009
Because metabolism is fundamental in sustaining microbial life, drugs that target pathogen-specific metabolic enzymes and pathways can be very effective. In particular, the metabolic challenges faced by intracellular pathogens, such as Mycobacterium tuberculosis, residing in the infected host provide novel opportunities for therapeutic intervention.
We developed a mathematical framework to simulate the effects on the growth of a pathogen when enzymes in its metabolic pathways are inhibited. Combining detailed models of enzyme kinetics, a complete metabolic network description as modeled by flux balance analysis, and a dynamic cell population growth model, we quantitatively modeled and predicted the dose-response of the 3-nitropropionate inhibitor on the growth of M. tuberculosis in a medium whose carbon source was restricted to fatty acids, and that of the 5'-O-(N-salicylsulfamoyl) adenosine inhibitor in a medium with low-iron concentration.
The predicted results quantitatively reproduced the experimentally measured dose-response curves, ranging over three orders of magnitude in inhibitor concentration. Thus, by allowing for detailed specifications of the underlying enzymatic kinetics, metabolic reactions/constraints, and growth media, our model captured the essential chemical and biological factors that determine the effects of drug inhibition on in vitro growth of M. tuberculosis cells.
- Metabolic Network
- Flux Balance Analysis
- Biomass Growth Rate
- Population Growth Model
- Substrate Uptake Rate
System-level networks of biological processes and functions allow us to draw inferences about the phenotype of an organism that cannot be made by considering each of its individual components [1–4]. In particular, metabolic networks are made up of hundreds to thousands of distinct but interconnected chemical reactions, each processing particular metabolites in different locations of the cell that, taken together, ultimately allow the cell to function and grow . The metabolic network of an organism is assembled, through automated and manual procedures [6, 7], based on known chemical reactions collected from genome annotation databases, such as the Kyoto Encyclopedia of Genes and Genomes . Currently, assemblies of metabolic networks are available for several bacterial species [9–13], yeast , and humans .
Several quantitative approaches have been developed to study different aspects of metabolic networks [16, 17]. Kinetic models comprised of explicit sets of reactants and reactions can be constructed and solved via ordinary differential equations (ODEs), provided the rate constants for each reaction are known. Kinetic models are typically restricted to a selected set of reactions used to study, for example, metabolism in human red blood cells , core components of the anaerobic metabolism in Escherichia coli , and the relation between single nucleotide polymorphisms and anemia . Kinetics of inhibiting enzymes can be incorporated into such models . However, due to the limited number of reactions modeled, only parts of the metabolic network can be taken into account and the whole organism's response (i.e., its phenotype) to the inhibitor cannot be modeled. In more comprehensive studies of the entire metabolic network of organisms for which kinetic information of all reactions is not available, flux balance analysis (FBA) can be used to obtain the optimal steady-state reaction flux distribution and the organism's growth rate under constraints imposed on the directions of the reactions, stoichiometry, and maximal transport fluxes . FBA can predict experimentally determined cellular growth rates in different media [10–12, 23–25] and identify genes, which, when removed, prevent cellular growth [10, 11, 14, 26–29]. Going beyond steady-state conditions of the traditional FBA, cell growth dynamics can be taken into account in a dynamic flux balance model [25, 30]. Such dynamic flux balance models provide a link between temporal changes of nutrient concentrations in a given medium and cell growth, as nutrients are consumed by the cell. Because these conditions are typically encountered in experimental studies of bacterial growth, dynamic flux balance models provide an important mechanism for understanding and representing experimental observations.
Enzyme inhibition kinetics, FBA of metabolic networks, and cell growth dynamics have each been studied separately before. Moreover, enzyme inhibition kinetics has been incorporated into a portion of a metabolic network , and FBA of a metabolic network has been combined with cell growth dynamics . However, integration of all three components has not been attempted before. Here, we present a framework that links together all these three components - enzyme inhibition kinetics, FBA of a metabolic network, and cell growth dynamics - to model the growth inhibition of Mycobacterium tuberculosis, the causative agent of tuberculosis (TB).
TB is a major infectious disease in the world with over 9.2 million new cases and 1.7 million deaths in 2006, and it is estimated that one-third of the human population is infected with the disease . Mycobacteria are aerobic organisms classified as acid-fast Gram-positive bacteria due to their lack of an outer cell membrane. They are a relatively slowly dividing organism compared with other bacteria. Most treatments for tuberculosis directly interfere with mycobacteria-specific physiology . M. tuberculosis is a prototrophic and metabolically flexible organism capable of surviving in a variety of environments. Bacteria that reach the lung alveoli are internalized by resident macrophages, where they are able to replicate in modified vacuoles [33–35]. At the onset of adaptive immunity, activated macrophages keep the infection under control, but the bacteria are not eliminated, and a state of chronic persistence is established . Survival under such conditions requires metabolically active bacteria capable of producing counter-immune effectors [34, 37, 38].
Worldwide efforts to eliminate TB are confronting many obstacles, including drug-resistant pathogens, compliance with complicated drug regimens, and compromised immune systems associated with human immunodeficiency syndrome or acquired immunodeficiency syndrome . Partly to address these issues, renewed efforts have begun in developing drugs that target the intracellular metabolism of M. tuberculosis, for example, by analyzing metabolic pathways to identify potential drug targets that selectively affect M. tuberculosis . Importantly, using the sequenced genome of M. tuberculosis  together with literature data on known metabolic reactions, extensive metabolic network reconstructions have been carried out for this organism [42, 43]. Analyses of these networks based on FBA reveal that they contain sufficient information to predict growth rates and identify genes that are essential for the growth of M. tuberculosis in select media [42, 43].
Novel drug design approaches against M. tuberculosis metabolism exploit the unique and harsh conditions that the pathogen is exposed to in the host environment. After entering a host, pathogens are confronted with a nutrient-poor environment and are often restricted to utilizing fatty acids as their main carbon source [32, 38]. This is accomplished by activation of the glyoxylate shunt pathway and the methylcitrate cycle [44, 45]. Consequently, the ability to inhibit key reactions of these two pathways makes 3-nitropropionate (3-NP) an effective inhibitor for the in vitro growth of M. tuberculosis in fatty acid media as well as for its in vivo growth in mouse macrophage cells . In addition to presenting a limited carbon source, the host environment is also deficient in iron, another nutritional requirement for the invading pathogen. Free iron is strictly controlled in the host environment via host iron-binding proteins, such as human transferrin, as a way to defend against bacterial infections . Thus, many pathogens synthesize siderophores, chemicals with very high affinity for iron, to wrestle iron away from the host . For M. tuberculosis, mycobactin is the necessary siderophore required for growth in macrophages and media containing low concentrations of iron . Therefore, the biosynthesis of mycobactin  and the regulation of this iron uptake mechanism [51, 52] have been extensively studied as a potential drug target. This led to the discovery that 5'-O-(N-salicylsulfamoyl) adenosine (sAMS) can function as an inhibitor to M. tuberculosis via its ability to inhibit mycobactin synthesis .
Here we developed a mathematical framework in which enzyme inhibition kinetics, metabolic network simulation, and cell growth dynamics are considered together to produce a system that is able to quantitatively model drug inhibition of cell growth. We separately simulated the effects of two metabolic inhibitors, 3-NP and sAMS, on the growth of M. tuberculosis cells, using an in vitro media model designed to mimic the limited nutritional environment in a host cell. The predicted dose-response curves quantitatively reproduced the observed experimental data, indicating that the developed modular framework was capable of capturing the effects of metabolic inhibitors on bacterial cell growth.
The mathematical framework needed to map the amount of drug to the collective growth response of the M. tuberculosis bacterium required that we connect enzyme inhibition kinetics, metabolic network modeling, and bacterial population growth models. If these components could be modeled and verified by experimental data, we could create a computational system to quantitatively predict how metabolic inhibitors affect bacterial growth. Here, we present the framework that allowed us to generate and reproduce the dose-response curves of two metabolic inhibitors generated from two independent experimental studies.
The inhibition model is defined by the enzyme inhibition kinetics governing the reactants and products of a particular metabolic reaction (i.e., the target reaction) and relates the manner by which the inhibitor concentration [I] modifies or adds constraint to the flux of the target reaction. The mathematical form of the inhibition model depends on the particular enzyme kinetics associated with the specific inhibitor and the metabolic reaction affected by the inhibitor. Mathematically, the inhibition model relates an inhibitor concentration [I] to the resulting metabolite flux ratio in the presence and absence of the inhibitor. Inhibitors affecting more than one reaction can also be considered.
The metabolic network is used to self-consistently calculate the overall biomass growth rate μ, substrate uptake rates v C , and the fluxes of all metabolic reactions. It is coupled to the inhibition model and the population growth model. The inhibition model places constraints on the flux of the target reactions in the metabolic network that affects the total biomass growth rate, and the population growth model adds constraints to the network's substrate uptake rate from the medium. The effect of enzyme deletions (deletion mutants) on growth can be incorporated in the metabolic network model by removal of the particular reactions catalyzed by the specified enzymes [22, 54].
The metabolic network developed by Jamshidi and Palsson  for M. tuberculosis, iNJ 661, is able to quantitatively reproduce observed growth rates under a number of different conditions. We used this network, with some modifications, as the basis for our work (see Additional file 1: Section S1) and verified that the modifications did not affect the growth rate of M. tuberculosis, as originally reported .
The rate of growth of the biomass, the substrate uptake rates, and the reaction fluxes were obtained directly from the metabolic network by applying the FBA . By using a linear programming method, FBA can maximize the biomass growth rate subject to steady-state mass balance of all the intracellular metabolites, and the stoichiometric constraints defined by the reactions. The maximization of biomass growth rate is based on the assumption that bacteria maximize their growth during the exponential stage and early stationary stage conditions. This assumption has been shown by previous studies to generate experiment-compatible results under such conditions [25, 30]. Additional constraints that can be modeled in the FBA include specification of reversible and irreversible reactions, as well as limits placed on substrate uptake and target reaction rates. Here, FBA was performed with the COBRA Toolbox . We also used the COBRA Toolbox to minimize the reaction fluxes while keeping the calculated maximal biomass growth rate. This procedure allowed us to obtain a unique set of minimum fluxes corresponding to the most parsimonious flow of metabolites through the network [56–58]. These fluxes were then used to constrain the target reaction rates.
Population growth model
Given the biomass growth rate μ and the substrate uptake rates v C defined by the metabolic network, the population growth model provides a mechanism for calculating cell [X] and substrate [C] concentrations in the specified medium. This model considers changes in the substrate concentrations over time, which could be used to monitor how different carbon sources are preferentially used in the metabolic process .
where the brackets [.] indicate the concentration of "." and [C] represents the concentration of the limiting substrate C in the medium. The factor 24 simply converts the time t from hours to days, since the units of μ and v C are expressed, respectively, in h-1 and mmol/(h·gDW), that is, mmol per hour per gram dry weight of M. tuberculosis. Equation 1 did not include a cellular death rate because we simulated bacterial growth during exponential stage and early stationary stage conditions, where cellular growth plays a more dominant role than cellular death. Previous studies under similar growth conditions, which also did not explicitly include a death rate term, obtained simulation results that were consistent with experimental data [25, 30].
where K denotes a proportionality constant. Other experimental readouts can be similarly accounted for.
Sensitivity analysis of parameter values
Other observables, different from OD, can be substituted for in Eq. 6. To numerically calculate the sensitivity coefficient for a parameter p, we started from ∂p = +0.5p and repeated the process by reducing ∂p and calculating the sensitivity coefficient until converged, that is, until successive values of ∂p yielded the same . We then repeated the process starting from ∂p = -0.5p until convergence. In the calculation performed here, both processes converged to the same numerical value.
To address the different types of parameters in our framework, we classified the modeled parameters into four groups: group I included parameters obtained from the literature, group II included those determined by matching experimental data, group III included those assumed to be derived from other parameters, and group IV included those that, by definition, were directly determined once the other parameters were defined. During the sensitivity analysis of the parameters in groups I, II, and III, we calculated dose-response curves while increasing and decreasing each parameter by 50% (except for those whose values cannot exceed one) and sensitivity coefficients spanning three orders of magnitude in inhibitor concentration. Although the parameters in group III were assumed to be derived from the parameters in groups I and II, during the sensitivity analysis for the first two groups we held the parameter values in group III fixed. Also, because the parameters in group IV were dependent on other parameters and their values changed as we performed sensitivity analysis on these independent parameters, we did not perform analysis for this group.
Modeling cell growth inhibition by 3-NP
Nutrient-poor environment and in-vivo growth
It has been experimentally shown that 3-NP inhibits the growth of M. tuberculosis in fatty acid media and in mouse macrophage cells . Medium containing propionate (C2H5COO-) an odd-chain fatty acid as the main carbon source was used in an in vitro experimental model system to investigate in vivo inhibition . Here, we similarly modeled the inhibition effect of 3-NP on the M. tuberculosis growth on in vitro media as a prelude to the more complicated study of modeling drug effects in an in vivo host-cell environment.
Experiment-specific mathematical framework
Experiments show that 3-NP inhibits the growth of M. tuberculosis when propionate is the major carbon source in the medium used to grow the bacterium . To quantitatively model this inhibition effect, we needed to assemble the mathematical framework that was specific to this medium and inhibition process. The terminology "inhibitor," "target reaction," and "substrate" in Figure 1 refer to 3-NP, the reactions ICL and MCL, and propionate, respectively. The appropriate specifications needed for the inhibition model, metabolic network, and population growth model are described below.
3-NP inhibition model
where v ICL and denote the inhibitor and inhibitor-free reaction fluxes, respectively, w ICL1 and w ICL2 represent the fractions of the overall inhibitor-free ICL reaction flux for the ICL1- and ICL2-catalyzed reaction components, respectively, SUC denotes the succinate substrate, [SUC] indicate its concentration, and K3-NP, ICL1, K3-NP, ICL2, KSUC, ICL1, and KSUC, ICL2denote Michaelis constants .
The parameter values for this model can be partially found in the literature, but some of them need to be calculated or fitted to match experimental conditions. The values of w ICL1 and w ICL2 were estimated from the known rate constants for the ICL reaction catalyzed by the ICL1 and the ICL2 enzymes, respectively. Thus, given the reaction rates of 5.24 s-1 and 1.38 s-1 for ICL1 and ICL2, respectively , the fractions of w ICL1 and w ICL2 were estimated to be 0.79 and 0.21, respectively. The experimental values employed for the succinate concentration were used to set [SUC] to 2.464 mM . Likewise, we directly used the experimentally determined values of 0.003 mM and 0.11 mM for K3-NP, ICL1and K3-NP, ICL2, respectively [21, 61]. The value for KSUC, ICL1was obtained by using the 3-NP specific mathematical framework to match the experimental cell concentration at a specific 3-NP concentration of 0.025 mM (this is described in the next subsection "Obtaining Undetermined Parameter Values"). Based on the experimentally measured range of KSUC, ICL2values , we set this parameter to 10 × KSUC, ICL1.
The variables and parameters in this equation have similar meaning to those given in Eq. 9. The unavailable values for K3-NP, MCL 1, K3-NP, MCL 2, KSUC, MCL 1, and KSUC, MCL 2were set to the same corresponding values used in the ICL model. The fractions wMCL 1and wMCL 2were set to 0.999 and 0.001, respectively, as the associated rate constants of the MCL reaction for the ICL1 and the ICL2 enzymes are 1.25 s-1 and <10-3 s-1, respectively .
Metabolic network considerations
The metabolic network model must take into account the appropriate substrate uptake and target reaction constraints based on the experimental setting . Here, we first focus on substrate uptake in propionate medium used in the experiment. The propionate uptake was constrained based on the propionate concentration in the medium. Because setting the glycerol uptake rate to zero would have caused the biomass growth rate to be zero , we set this uptake rate to a very small value, 0.001 mmol/(h·gDW). In iNJ 661 , glycerol is a biomass component, but there is no pathway to synthesize glycerol, necessitating the addition of a small amount of glycerol uptake to the metabolic network. The uptake rates of other carbon sources, like glucose, were set to zero. Other necessary substrate uptake rates, including phosphate, sulfate, ferric iron, ammonium, and oxygen, were left unconstrained. In the absence of a 3-NP inhibitor in the medium, the fluxes of the ICL and MCL reactions in Eqs. 6 and 7 were unconstrained and the inhibitor-free reaction fluxes and were obtained from a FBA calculation. When 3-NP was present, the ICL (and MCL) reaction fluxes were constrained to be no more than the product of the flux ratios of f ICL ([3-NP]) (or f MCL ([3-NP])), determined from the inhibition model, and the inhibitor-free fluxes f ICL (or f MCL ). We constrained the target reaction fluxes to upper bounds instead of fixing them to specific values in order to avoid an artificial coupling of fluxes. For example, in the case of 3-NP, which inhibits both ICL and MCL reactions, the resultant fluxes may be different and may not be equal to but lower than the constraints. When the M. tuberculosis deletion mutant Δicl1Δicl2 was studied, the fluxes associated with the ICL and MCL reactions were set to zero [11, 12]. The developed metabolic network, including the constrained substrate uptake rates, is available in Systems Biology Makeup Language (SBML) format (see Additional file 2).
Experimental population growth model
The initial values for OD were taken directly from the experimental data . The population growth model defined by Eqs. 11-14 were then iteratively solved by using the results generated from the FBA of the metabolic network.
Obtaining undetermined parameter values
All parameters needed to calculate cellular growth and growth inhibition from the mathematical framework in Figure 1 have not been experimentally determined. However, we could use the combined formalism of the three models to self-consistently determine the unknown parameter values. In particular, we needed to estimate values for the initial propionate concentration [C'] (t = 0), the maximum initial propionate uptake rate V m , the Michaelis-Menten rate constant for the propionate uptake in Eq. 14, and KSUC, ICL1in Eq. 9.
Verification of essentiality of the target reactions
A prerequisite for a good inhibitor is that its target is essential for the survival and homeostasis of the bacterium. In the experimental study, genes icl1 and icl2 whose products catalyze the reactions ICL and MCL, respectively, are deleted from wild-type M. tuberculosis. Figure 3A shows that in the experiment the resultant deletion mutant Δicl1Δicl2 exhibits a lack of growth in propionate medium . We used the mathematical framework to verify that the two 3-NP-inhibited reactions, ICL and MCL, are necessary for the growth of M. tuberculosis in this medium. This was achieved by setting the fluxes associated with these reactions to zero (see Additional file 1: Section S3), leading to a model that predicts complete lack of bacterial growth in the absence of these reactions (dotted line in Figure 3A).
As described above in "Obtaining Undetermined Parameter Values," we used the experimentally determined cell concentration growth at the lowest (0.025 mM) of the four measured inhibitor concentrations 0.025, 0.050, 0.100, and 0.200 mM  to determine the value of KSUC, ICL1. With the values of the rest of parameters established, we then used the mathematical framework to predict the growth of M. tuberculosis at the other three 3-NP inhibitor concentrations (0.050, 0.100, and 0.200 mM) and compared the results to the experimental values (see Additional file 1: Section S4). Figure 3B shows that, for each inhibitor concentration, the predicted and the experimental cell concentrations were, overall, in good agreement with each other. However, Figure 3B also shows that the simulation results may over- or under-predict the experimental data. This was due to the maximization of biomass growth rate and also by not including an explicit cellular death rate, assumptions that are reasonable for modeling growth in exponential and early stationary stages (see section "Development of the Mathematical Framework"). Thus, implementation of our modeling framework under different conditions or for very long time periods may cause mismatches when simulating cellular growth. Figure 3C shows that the simulated dose-response curve, which links the 3-NP concentrations to the cell concentrations after a 16-day growth of M. tuberculosis, provided accurate predictions matching the experimental results.
We used linear regression  to evaluate the fitness between the simulation results and the experimental data. For the 30 experimental data points shown in Figures 3A and 3B, we used our framework to obtain the corresponding simulated values. A linear regression of the data yielded a slope (1.0008) and intercept (0.0001) close to one and zero, respectively. The coefficient of determination (R2) was 0.9867, indicating a strong and significant correlation (P value = 8.2050 × 10-28) between the simulated values and experimental data.
The benefit of a quantitative predictive model lies both in the ability to rapidly make predictions once the model is properly parameterized and the additional insights gained in the mechanisms underlying the experimental observables. With a model, we can accurately predict dose-response curves in less than one hour, and the mathematical framework can provide information that cannot be directly obtained from an experiment. For example, we knew that 3-NP inhibits both the ICL and the MCL reactions defined in Eqs. 6 and 7, respectively. However, the degree to which each reaction slows down growth was not known. This question is experimentally difficult to ascertain, since the two reactions are catalyzed by the same enzyme. Theoretically, we can use the developed formalism to answer this question by allowing 3-NP to inhibit only one reaction. The calculated growth of M. tuberculosis in medium with 0.025, 0.050, 0.100, and 0.200 mM 3-NP was very similar to the inhibitor-free growth when only the ICL reaction was affected by the inhibitor. However, when we assumed that 3-NP only inhibited the MCL reaction, the calculated growth was virtually the same as the results in Figure 3A (dashed line) and Figure 3B. Therefore, the simulations suggest that it was primarily the inhibitory effect of 3-NP on the MCL reaction that limited M. tuberculosis growth. This observation is compatible with the metabolism of odd- and even-chain fatty acids. Both the ICL and the MCL reactions are necessary steps in transferring extracellular carbon atoms to intracellular metabolites and obtaining energy from fatty acids. However, these reactions differ in that the ICL reaction is used for even-chain fatty acids, the MCL reaction is used for odd-chain fatty acid with three carbon atoms such as propionate, and longer odd-chain fatty acids require both reactions . Because in the studied medium propionate is the major carbon source, the inhibition of the MCL reaction is key to the inhibitory effect of 3-NP.
Sensitivity analysis of parameter values
Model parameters for cell growth inhibition by 3-NP.
Source of the value
Set to be 0.790 from 
Set to be 2.464 mM from 
K 3-NP, ICL1
K 3-NP, ICL2
w MCL 1
Set to be 0.999 from 
[C'] (t = 0)
Population Growth Model
Obtained by matching experimental cell growth data
Population Growth Model
Population Growth Model
K SUC, ICL1
K SUC, ICL2
Assumed (based on ) to be 10 × KSUC, ICL1
K SUC, MCL 1
Assumed to be equal to KSUC, ICL1
K SUC, MCL 2
Assumed to be equal to KSUC, ICL2
K 3-NP, MCL 1
Assumed to be equal to K3-NP, ICL1
K 3-NP, MCL 2
Assumed to be equal to K3-NP, ICL2
Equal to 1-w ICL1
w MCL 2
Equal to 1-w MCL1
The other parameters of the model had no effect on the calculated dose-response curves. The parameters w ICL1 , K3-NP, ICL1, K3-NP, ICL2, KSUC, ICL1, and KSUC, ICL2, used in the definition of the 3-NP inhibition model in Eq. 8, did not affect the results because 3-NP primarily inhibited growth through the MCL reaction and not the ICL reaction (see previous subsection "Growth Predictions"). Similarly, KSUC, MCL 2and K3-NP, MCL 2, relating to the ICL2 enzyme in the second term of Eq. 10, did not affect the calculated dose-response curves.
Sensitivity coefficients for the parameters in modeling 3-NP inhibition.
K 3-NP, ICL1
K 3-NP, ICL2
w MCL 1
[C' ] (t = 0)
K SUC, ICL1
K SUC, ICL2
K SUC, MCL 1
K SUC, MCL 2
K 3-NP, MCL 1
K 3-NP, MCL 2
Although our framework is capable of modeling growth inhibition as a function of inhibitor concentration, it would still be desirable to reduce the uncertainty in the model parameters by directly obtaining accurate parameter values from experimental studies. The predictive power of our model could be further refined if the relationship between succinate concentrations [SUC] and the fluxes of the ICL and MCL reactions inside M. tuberculosis cells could be ascertained experimentally. This could be done by jointly measuring intracellular metabolite concentrations and metabolic fluxes, as recently done in a study of E. coli metabolism . Similarly, values for K3-NP, MCL1and KSUC, MCL1could be obtained directly from enzyme kinetic experiments .
Modeling cell growth inhibition by sAMS
The importance of iron sequestration
As a response to the invasion of M. tuberculosis, the host immune system reduces the iron levels in pathogen-infected environments by means of iron-binding proteins . M. tuberculosis responds to the changing environment by synthesizing and secreting mycobactin, which has an extremely high iron affinity and helps the pathogen obtain iron from host proteins . The synthesis of mycobactin is thus an essential step for the survival and growth of M. tuberculosis inside the host and provides a potential drug target with broad anti-bacterial applicability.
Experiment-specific mathematical framework
To implement the mathematical framework outlined in Figure 1, we customized the models to account for the action of sAMS ("inhibitor") on the synthesis of mycobactin ("target reaction"). Glycerol, alanine, and salts in the medium used in the experimental studies of this inhibitor  were modeled as "substrates." The appropriate specifications needed for the inhibition model, the metabolic network, and the population growth model are described below.
sAMS inhibition model
where v MS and denote the flux of the mycobactin synthesis reaction in the presence and in the absence of the sAMS inhibitor, respectively, and is an "apparent" reaction rate constant whose value is 0.7 nM . To study the effect of the inhibitor on the isolated MbtA enzyme, we used the same values taken in the in vitro experimental assay and set [E] = 20 nM . For intracellular environment studies, the value of [E] is unknown but can be inferred, as described below (see subsection "Obtaining Undetermined Parameter Values").
Metabolic network considerations
To duplicate the experimental conditions of the sAMS inhibitor study, we constrained the substrate uptake based on the medium used in the experiment . The medium contained glycerol, alanine, salts, and Tween (GAST) and the amount of added iron defined the medium condition as iron-deficient or sufficient . The Tween component of the medium acts as a detergent in the experimental system and was not included as a substrate in the metabolic network. Glycerol and alanine are major carbon sources whose uptake rates were constrained to be no more than 1 mmol/(h·gDW) . The uptake rates of the salts, oxygen, and water were unconstrained in the metabolic network. We also modified the biomass composition for iron-sufficient medium by changing the metabolite "iron(III) chelated carboxymycobactin T" into iron(III), since mycobactin synthesis and chelation were absent in this medium. The constraint placed on the target reaction followed the approaches used in the 3-NP study. Thus, when sAMS was present, the reaction flux was constrained to be no more than the product of f MS and . The developed metabolic network, including the constrained substrate uptake rates, is available in SBML format (see Additional file 3).
Experimental population growth model
where [X0] denotes the inhibitor-free cell concentration, t is set to eight days, and the inhibitor-present biomass growth rate μ and the inhibitor-free biomass growth rate μ0 were inferred from the metabolic network using FBA.
Obtaining undetermined parameter values
Among the parameters needed to study the inhibitory effect of sAMS on M. tuberculosis growth, only the intracellular MbtA-enzyme concentration [E] in Eq. 15 was not readily available from the experimental data. To obtain this parameter value, we selected a relative cell concentration R C of 0.49 at a sAMS concentration [sAMS] of 1.7 μM from the growth-inhibition experiment in iron-deficient medium  and varied the value of [E] in Eq. 15 until our framework reproduced this R C value. This point was selected because 0.49 is close to the mid-range of R C values, 0 ≤ R C ≤ 1. After trying different values for [E], we found that [E] = 40 μM yielded good agreement between the calculated (0.47) and selected (0.49) relative cell concentrations (see Additional file 1: Section S5).
Verification of essentiality of the target reactions
To meet the minimum requirements of an inhibitor, sAMS needs to target a reaction that is essential for cellular survival and function. From the experimental analysis, we noted that, as the sAMS concentration increases to a large value (~103 μM), the measured relative cell concentration becomes close to zero . Similarly, our mathematical framework needs to be capable of reproducing the essentiality of the targeted reaction, mycobactin synthesis in the presence of sAMS, for cellular growth of M. tuberculosis in iron-deficient GAST medium. Thus, we set the flux of the mycobactin synthesis reaction to zero, which, as expected, yielded a relative cell concentration R C of zero (see Additional file 1: Section S6). This suggests that the essentiality of the mycobactin synthesis reaction was duplicated in our framework.
We now turn to predicting the response of M. tuberculosis cells growing in iron-deficient GAST medium exposed to varying sAMS inhibitor concentrations. We used our framework and obtained the dose-response curve in Figure 6B (see Additional file 1: Section S7). The close agreement between predicted and experimental data  indicates that the mathematical framework was successful in coupling the three underlying models (inhibition, metabolic network, and population growth) to quantitatively predict the inhibitory effect of sAMS on M. tuberculosis growth in an iron-deficient medium. Moreover, to evaluate the agreement between the experimental data and their corresponding simulated values, we again performed a linear regression on the data . The obtained slope (0.9223), intercept (0.0661), and coefficient of determination R2 (0.9779) for the 22 data points in Figure 6B (for growth in iron-deficient medium) were commensurate with a P value of 4.9331 × 10-18, suggesting a strong and very similar relation between the simulated values and experimental data.
Similarly, we repeated the calculations for the inhibitory effect of sAMS in an iron-sufficient GAST medium. In this medium, siderophore sequestering is not an issue, because iron is freely available and the direct impact of inhibiting the mycobactin synthesis reaction should be negligible. Accordingly, we predicted that sAMS had no effect on M. tuberculosis growth in an iron-sufficient medium (see Additional file 1: Section S7). Figure 6B shows that our predictions matched the experimental data under relatively low sAMS concentration (<10 μM). At higher inhibitor concentration, however, it is speculated that the growth of M. tuberculosis in iron-sufficient medium is inhibited by sAMS through some other unknown mechanism . Since this inhibitory mechanism is not accounted for in our model, we could not capture this feature. The modeling framework is thus quite powerful when the mechanism of inhibition is known. However, as illustrated, it cannot prospectively predict alternate binding of inhibitors or other cellular inhibition mechanisms not explicitly detailed in the model.
Sensitivity Analysis of Parameter Values
Model parameters for cell growth inhibition by sAMS.
Source of the value
Set to be 0.7 nM from 
Set to be 1 mmol/(h·gDW) from 
Set to be 1 mmol/(h·gDW) from 
t: time length of cellular growth
Population Growth Model
Set to be 8 days from 
[E]: intracellular MbtA concentration
Obtained by matching experimental data
Sensitivity coefficients for the parameters in modeling sAMS inhibition.
We developed a mathematical framework connecting kinetic models of enzyme inhibition with metabolic network analysis and a population growth model. The three components correspond to the three major steps through which a metabolic inhibitor affects bacterial growth. First, the inhibition model describes how the particular inhibitor affects the enzyme kinetics and the flux(es) of one or more metabolic reactions. Second, the metabolic network analysis connects the changes in the affected metabolite flux(es) to the growth rate of the organism. Finally, the population growth model takes the altered growth rate and converts it to an effective bacterial cell concentration. This framework allowed us to quantitatively simulate the effect of two distinct metabolic inhibitors on in vitro bacterial growth under different nutritional conditions.
We applied this framework to model the effect of two separate metabolic inhibitors, 3-NP and sAMS, on the growth of M. tuberculosis cells on propionate medium and on iron-deficient GAST medium, respectively. Both reactions affected by these two inhibitors are required for the survival of the pathogen in the host environment and could potentially become important therapeutic targets. 3-NP inhibits key reactions in the glyoxylate shunt and the methylcitrate cycle, effectively blocking the utilization of fatty acids, the major carbon source of M. tuberculosis in the host environment [45, 46]. sAMS inhibits the synthesis of mycobactin, which is required for iron uptake of M. tuberculosis within an iron-deficient host environment . Our model was capable of quantitatively reproducing the experimentally determined dose-response curves for both inhibitors. Thus, with the proposed mathematical framework, we could analyze the studied system under conditions matching the experimental protocols as they relate to metabolism. We accounted for the underlying kinetics of the inhibition, how this was translated via the metabolic network analysis to metabolite flow and biomass accumulation, and to the growth of the cell population that was used as the experimental readout for drug inhibition. We noted, however, that certain cellular processes or responses that impact drug action in the cell, for example, adaptive responses in the form of altered gene expression of metabolic enzyme and activated drug efflux transport, were not accounted for in the proposed modeling scheme. These processes may play important roles and may need to be accounted for when modeling inhibitor effects of other than those of 3-NP and sAMS.
In this work, the mathematical framework was used to model an inhibitor's effect on cellular growth of a pathogen in different in vitro environments designed to duplicate aspects of the nutritional conditions encountered in the host. However, intracellular pathogens have complex interactions with their hosts  and the conclusions drawn from an in vitro environment may not be operative in the in vivo host environment . The current models in our framework could be coupled to other models that, in turn, determine the medium content by simulating the metabolic nutrients available in a human macrophage cell. Such an embedding of the current modeling framework within other schemes could be used to add further biological complexity to the existing computational platform. Enzyme activity could be further coupled to a gene expression model to modulate protein/enzyme function according to microarray gene expression data . The implementation of additional models is only limited by the availability of experimental data with which to perform rigorous parameter testing and prediction validation.
For the two inhibitors studied, the essentiality of the protein targets was a necessary condition. Essentiality of a gene can be imparted by the network itself or any other condition that alters of restricts the flux of metabolites in the network. Thus, some genes become essential only under specific nutritional conditions, while others may become essential when one or more nonessential genes are knocked out. It is also possible to envision certain scenarios where drugs affecting parts of the metabolic network induce essentiality to uninhibited enzymes in the network. Quantitative models, such as the one developed here, could be used to rapidly investigate such conditions and assist future experimental studies. For example, using our framework, we suggested that 3-NP was effective in fatty acid medium but not in glucose medium (data not shown), which was supported by experimental observations . In addition to the two inhibitors examined in this study, our calculations also suggested that the inhibitor targeting protein TrpD (a drug target discussed in ) will only be effective when tryptophan is absent from the medium (data not shown). This observation calls for further experimental verification.
The current work introduces a systems biology approach using enzyme kinetics, metabolic networks, and population growth models that is capable of capturing the essential chemical and biological variability of the system under study. This enabled us to simulate and understand the underlying chemical and biological factors that give rise to the experimental observables, in this case growth inhibition of M. tuberculosis cells. Our results suggest that this type of inclusive modeling approach would be valuable in proposing new experimental studies by extending, combining, and exploring novel chemical and biological inhibition concepts.
We implemented a systems biology framework, which combines detailed models of enzyme kinetics, a complete metabolic network analysis, and a cell population growth model, to represent and understand cellular growth inhibition in response to drugs. We used this mathematical framework to simulate two separate inhibition mechanisms for the growth of M. tuberculosis cells in an in vitro environment, which was modeled to represent the nutritional challenges encountered in a host cell. We calculated dose-response curves corresponding to the cellular growth versus drug concentration for the growth in a medium whose carbon source was restricted to fatty acids and was infused with varying concentrations of the 3-NP inhibitor. Similarly, we obtained dose-response curves for cells grown in medium with low-iron concentration and exposed to different amounts of the sAMS inhibitor. These results quantitatively reproduced experimentally measured dose-response curves, ranging over three orders of magnitude in inhibitor concentration. The ability of the proposed models to capture in vitro drug inhibition confirms that relevant features of intracellular metabolism of M. tuberculosis can be modeled by a metabolic network-based framework.
The authors were supported, in part, by the Military Operational Medicine Research Area Directorate of the U.S. Army Medical Research and Materiel Command, Ft. Detrick, Maryland. This effort was supported by the U.S. Army's Network Science Initiative. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the U.S. Army or of the U.S. Department of Defense. This paper has been approved for public release with unlimited distribution.
- Ideker T, Galitski T, Hood L: A new approach to decoding life: systems biology. Annu Rev Genomics Hum Genet. 2001, 2: 343-372. 10.1146/annurev.genom.2.1.343View ArticlePubMedGoogle Scholar
- Kell DB: Systems biology, metabolic modelling and metabolomics in drug discovery and development. Drug Discov Today. 2006, 11: 1085-1092. 10.1016/j.drudis.2006.10.004View ArticlePubMedGoogle Scholar
- Hornberg JJ, Bruggeman FJ, Westerhoff HV, Lankelma J: Cancer: a Systems Biology disease. Biosystems. 2006, 83: 81-90. 10.1016/j.biosystems.2005.05.014View ArticlePubMedGoogle Scholar
- Cho CR, Labow M, Reinhardt M, van Oostrum J, Peitsch MC: The application of systems biology to drug discovery. Curr Opin Chem Biol. 2006, 10: 294-302. 10.1016/j.cbpa.2006.06.025View ArticlePubMedGoogle Scholar
- Joyce AR, Palsson BO: Toward whole cell modeling and simulation: comprehensive functional genomics through the constraint-based approach. Prog Drug Res. 2007, 64: 267-309.Google Scholar
- Francke C, Siezen RJ, Teusink B: Reconstructing the metabolic network of a bacterium from its genome. Trends Microbiol. 2005, 13: 550-558. 10.1016/j.tim.2005.09.001View ArticlePubMedGoogle Scholar
- Breitling R, Vitkup D, Barrett MP: New surveyor tools for charting microbial metabolic maps. Nat Rev Microbiol. 2008, 6: 156-161. 10.1038/nrmicro1797View ArticlePubMedGoogle Scholar
- Kanehisa M, Araki M, Goto S, Hattori M, Hirakawa M, Itoh M, Katayama T, Kawashima S, Okuda S, Tokimatsu T, Yamanishi Y: KEGG for linking genomes to life and the environment. Nucleic Acids Res. 2008, 36: D480-484. 10.1093/nar/gkm882PubMed CentralView ArticlePubMedGoogle Scholar
- Suthers PF, Dasika MS, Kumar VS, Denisov G, Glass JI, Maranas CD: A genome-scale metabolic reconstruction of Mycoplasma genitalium, iPS189. PLoS Comput Biol. 2009, 5: e1000285- 10.1371/journal.pcbi.1000285PubMed CentralView ArticlePubMedGoogle Scholar
- Feist AM, Henry CS, Reed JL, Krummenacker M, Joyce AR, Karp PD, Broadbelt LJ, Hatzimanikatis V, Palsson BO: A genome-scale metabolic reconstruction for Escherichia coli K-12 MG1655 that accounts for 1260 ORFs and thermodynamic information. Mol Syst Biol. 2007, 3: 121- 10.1038/msb4100155PubMed CentralView ArticlePubMedGoogle Scholar
- Thiele I, Vo TD, Price ND, Palsson BO: Expanded metabolic reconstruction of Helicobacter pylori (iIT341 GSM/GPR): an in silico genome-scale characterization of single- and double-deletion mutants. J Bacteriol. 2005, 187: 5818-5830. 10.1128/JB.187.16.5818-5830.2005PubMed CentralView ArticlePubMedGoogle Scholar
- Oh YK, Palsson BO, Park SM, Schilling CH, Mahadevan R: Genome-scale reconstruction of metabolic network in Bacillus subtilis based on high-throughput phenotyping and gene essentiality data. J Biol Chem. 2007, 282: 28791-28799. 10.1074/jbc.M703759200View ArticlePubMedGoogle Scholar
- Kjeldsen KR, Nielsen J: In silico genome-scale reconstruction and validation of the Corynebacterium glutamicum metabolic network. Biotechnol Bioeng. 2009, 102: 583-597. 10.1002/bit.22067View ArticlePubMedGoogle Scholar
- Duarte NC, Herrgard MJ, Palsson BO: Reconstruction and validation of Saccharomyces cerevisiae iND750, a fully compartmentalized genome-scale metabolic model. Genome Res. 2004, 14: 1298-1309. 10.1101/gr.2250904PubMed CentralView ArticlePubMedGoogle Scholar
- Duarte NC, Becker SA, Jamshidi N, Thiele I, Mo ML, Vo TD, Srivas R, Palsson BO: Global reconstruction of the human metabolic network based on genomic and bibliomic data. Proc Natl Acad Sci USA. 2007, 104: 1777-1782. 10.1073/pnas.0610772104PubMed CentralView ArticlePubMedGoogle Scholar
- Klipp E: Systems biology in practice: concepts, implementation and application. 2005, Weinheim: Wiley-VCHView ArticleGoogle Scholar
- Palsson B: Systems biology: properties of reconstructed networks. 2006, New York: Cambridge University PressView ArticleGoogle Scholar
- Jamshidi N, Edwards JS, Fahland T, Church GM, Palsson BO: Dynamic simulation of the human red blood cell metabolic network. Bioinformatics. 2001, 17: 286-287. 10.1093/bioinformatics/17.3.286View ArticlePubMedGoogle Scholar
- Young JD, Henne KL, Morgan JA, Konopka AE, Ramkrishna D: Integrating cybernetic modeling with pathway analysis provides a dynamic, systems-level description of metabolic control. Biotechnol Bioeng. 2008, 100: 542-559. 10.1002/bit.21780View ArticlePubMedGoogle Scholar
- Jamshidi N, Wiback SJ, Palsson BB: In silico model-driven assessment of the effects of single nucleotide polymorphisms (SNPs) on human red blood cell metabolism. Genome Res. 2002, 12: 1687-1692. 10.1101/gr.329302PubMed CentralView ArticlePubMedGoogle Scholar
- Singh VK, Ghosh I: Kinetic modeling of tricarboxylic acid cycle and glyoxylate bypass in Mycobacterium tuberculosis, and its application to assessment of drug targets. Theor Biol Med Model. 2006, 3: 27- 10.1186/1742-4682-3-27PubMed CentralView ArticlePubMedGoogle Scholar
- Edwards JS, Palsson BO: Metabolic flux balance analysis and the in silico analysis of Escherichia coli K-12 gene deletions. BMC Bioinformatics. 2000, 1: 1- 10.1186/1471-2105-1-1PubMed CentralView ArticlePubMedGoogle Scholar
- Famili I, Forster J, Nielsen J, Palsson BO: Saccharomyces cerevisiae phenotypes can be predicted by using constraint-based analysis of a genome-scale reconstructed metabolic network. Proc Natl Acad Sci USA. 2003, 100: 13134-13139. 10.1073/pnas.2235812100PubMed CentralView ArticlePubMedGoogle Scholar
- Barrett CL, Kim TY, Kim HU, Palsson BO, Lee SY: Systems biology as a foundation for genome-scale synthetic biology. Curr Opin Biotechnol. 2006, 17: 488-492. 10.1016/j.copbio.2006.08.001View ArticlePubMedGoogle Scholar
- Hjersted JL, Henson MA, Mahadevan R: Genome-scale analysis of Saccharomyces cerevisiae metabolism and ethanol production in fed-batch culture. Biotechnol Bioeng. 2007, 97: 1190-1204. 10.1002/bit.21332View ArticlePubMedGoogle Scholar
- Edwards JS, Palsson BO: The Escherichia coli MG1655 in silico metabolic genotype: its definition, characteristics, and capabilities. Proc Natl Acad Sci USA. 2000, 97: 5528-5533. 10.1073/pnas.97.10.5528PubMed CentralView ArticlePubMedGoogle Scholar
- Trawick JD, Schilling CH: Use of constraint-based modeling for the prediction and validation of antimicrobial targets. Biochem Pharmacol. 2006, 71: 1026-1035. 10.1016/j.bcp.2005.10.049View ArticlePubMedGoogle Scholar
- Volpe E, Cappelli G, Grassi M, Martino A, Serafino A, Colizzi V, Sanarico N, Mariani F: Gene expression profiling of human macrophages at late time of infection with Mycobacterium tuberculosis. Immunology. 2006, 118: 449-460.PubMed CentralPubMedGoogle Scholar
- Wang JP, Rought SE, Corbeil J, Guiney DG: Gene expression profiling detects patterns of human macrophage responses following Mycobacterium tuberculosis infection. FEMS Immunol Med Microbiol. 2003, 39: 163-172. 10.1016/S0928-8244(03)00223-2View ArticlePubMedGoogle Scholar
- Mahadevan R, Edwards JS, Doyle FJ: Dynamic flux balance analysis of diauxic growth in Escherichia coli. Biophys J. 2002, 83: 1331-1340. 10.1016/S0006-3495(02)73903-9PubMed CentralView ArticlePubMedGoogle Scholar
- World Health Organization: WHO Report 2008: Global tuberculosis control - surveillance, planning, financing. 2008, World Health Organization. GenevaGoogle Scholar
- Pieters J: Mycobacterium tuberculosis and the macrophage: maintaining a balance. Cell Host Microbe. 2008, 3: 399-407. 10.1016/j.chom.2008.05.006View ArticlePubMedGoogle Scholar
- Ernst JD: Macrophage receptors for Mycobacterium tuberculosis. Infect Immun. 1998, 66: 1277-1281.PubMed CentralPubMedGoogle Scholar
- Vergne I, Chua J, Singh SB, Deretic V: Cell biology of mycobacterium tuberculosis phagosome. Annu Rev Cell Dev Biol. 2004, 20: 367-394. 10.1146/annurev.cellbio.20.010403.114015View ArticlePubMedGoogle Scholar
- Sundaramurthy V, Pieters J: Interactions of pathogenic mycobacteria with host macrophages. Microbes Infect. 2007, 9: 1671-1679. 10.1016/j.micinf.2007.09.007View ArticlePubMedGoogle Scholar
- Munoz-Elias EJ, Timm J, Botha T, Chan WT, Gomez JE, McKinney JD: Replication dynamics of Mycobacterium tuberculosis in chronically infected mice. Infect Immun. 2005, 73: 546-551. 10.1128/IAI.73.1.546-551.2005PubMed CentralView ArticlePubMedGoogle Scholar
- Ng VH, Cox JS, Sousa AO, MacMicking JD, McKinney JD: Role of KatG catalase-peroxidase in mycobacterial pathogenesis: countering the phagocyte oxidative burst. Mol Microbiol. 2004, 52: 1291-1302. 10.1111/j.1365-2958.2004.04078.xView ArticlePubMedGoogle Scholar
- Munoz-Elias EJ, McKinney JD: Carbon metabolism of intracellular bacteria. Cell Microbiol. 2006, 8: 10-22. 10.1111/j.1462-5822.2005.00648.xView ArticlePubMedGoogle Scholar
- Young DB, Perkins MD, Duncan K, Barry CE: Confronting the scientific obstacles to global control of tuberculosis. J Clin Invest. 2008, 118: 1255-1265. 10.1172/JCI34614PubMed CentralView ArticlePubMedGoogle Scholar
- Anishetty S, Pulimi M, Pennathur G: Potential drug targets in Mycobacterium tuberculosis through metabolic pathway analysis. Comput Biol Chem. 2005, 29: 368-378. 10.1016/j.compbiolchem.2005.07.001View ArticlePubMedGoogle Scholar
- Cole ST, Brosch R, Parkhill J, Garnier T, Churcher C, Harris D, Gordon SV, Eiglmeier K, Gas S, Barry CE, et al.: Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature. 1998, 393: 537-544. 10.1038/31159View ArticlePubMedGoogle Scholar
- Jamshidi N, Palsson BO: Investigating the metabolic capabilities of Mycobacterium tuberculosis H37Rv using the in silico strain iNJ661 and proposing alternative drug targets. BMC Syst Biol. 2007, 1: 26- 10.1186/1752-0509-1-26PubMed CentralView ArticlePubMedGoogle Scholar
- Beste DJ, Hooper T, Stewart G, Bonde B, Avignone-Rossa C, Bushell ME, Wheeler P, Klamt S, Kierzek AM, McFadden J: GSMN-TB: a web-based genome-scale network model of Mycobacterium tuberculosis metabolism. Genome Biol. 2007, 8: R89- 10.1186/gb-2007-8-5-r89PubMed CentralView ArticlePubMedGoogle Scholar
- Gould TA, Langemheen van de H, Munoz-Elias EJ, McKinney JD, Sacchettini JC: Dual role of isocitrate lyase 1 in the glyoxylate and methylcitrate cycles in Mycobacterium tuberculosis. Mol Microbiol. 2006, 61: 940-947. 10.1111/j.1365-2958.2006.05297.xView ArticlePubMedGoogle Scholar
- Munoz-Elias EJ, Upton AM, Cherian J, McKinney JD: Role of the methylcitrate cycle in Mycobacterium tuberculosis metabolism, intracellular growth, and virulence. Mol Microbiol. 2006, 60: 1109-1122. 10.1111/j.1365-2958.2006.05155.xView ArticlePubMedGoogle Scholar
- Munoz-Elias EJ, McKinney JD: Mycobacterium tuberculosis isocitrate lyases 1 and 2 are jointly required for in vivo growth and virulence. Nat Med. 2005, 11: 638-644. 10.1038/nm1252PubMed CentralView ArticlePubMedGoogle Scholar
- Gobin J, Horwitz MA: Exochelins of Mycobacterium tuberculosis remove iron from human iron-binding proteins and donate iron to mycobactins in the M. J Exp Med. 1996, 183: 1527-1532. 10.1084/jem.183.4.1527View ArticlePubMedGoogle Scholar
- Fischbach MA, Lin H, Liu DR, Walsh CT: How pathogenic bacteria evade mammalian sabotage in the battle for iron. Nat Chem Biol. 2006, 2: 132-138. 10.1038/nchembio771View ArticlePubMedGoogle Scholar
- De Voss JJ, Rutter K, Schroeder BG, Su H, Zhu Y, Barry CE: The salicylate-derived mycobactin siderophores of Mycobacterium tuberculosis are essential for growth in macrophages. Proc Natl Acad Sci USA. 2000, 97: 1252-1257. 10.1073/pnas.97.3.1252PubMed CentralView ArticlePubMedGoogle Scholar
- Quadri LE: Assembly of aryl-capped siderophores by modular peptide synthetases and polyketide synthases. Mol Microbiol. 2000, 37: 1-12. 10.1046/j.1365-2958.2000.01941.xView ArticlePubMedGoogle Scholar
- Chou CJ, Wisedchaisri G, Monfeli RR, Oram DM, Holmes RK, Hol WG, Beeson C: Functional studies of the Mycobacterium tuberculosis iron-dependent regulator. J Biol Chem. 2004, 279: 53554-53561. 10.1074/jbc.M407385200View ArticlePubMedGoogle Scholar
- Crosa JH, Walsh CT: Genetics and assembly line enzymology of siderophore biosynthesis in bacteria. Microbiol Mol Biol Rev. 2002, 66: 223-249. 10.1128/MMBR.66.2.223-249.2002PubMed CentralView ArticlePubMedGoogle Scholar
- Ferreras JA, Ryu JS, Di Lello F, Tan DS, Quadri LE: Small-molecule inhibition of siderophore biosynthesis in Mycobacterium tuberculosis and Yersinia pestis. Nat Chem Biol. 2005, 1: 29-32. 10.1038/nchembio706View ArticlePubMedGoogle Scholar
- Segre D, Vitkup D, Church GM: Analysis of optimality in natural and perturbed metabolic networks. Proc Natl Acad Sci USA. 2002, 99: 15112-15117. 10.1073/pnas.232349399PubMed CentralView ArticlePubMedGoogle Scholar
- Becker SA, Feist AM, Mo ML, Hannum G, Palsson BO, Herrgard MJ: Quantitative prediction of cellular metabolism with constraint-based models: the COBRA Toolbox. Nat Protoc. 2007, 2: 727-738. 10.1038/nprot.2007.99View ArticlePubMedGoogle Scholar
- Edwards JS, Palsson BO: How will bioinformatics influence metabolic engineering?. Biotechnol Bioeng. 1998, 58: 162-169. 10.1002/(SICI)1097-0290(19980420)58:2/3<162::AID-BIT8>3.0.CO;2-JView ArticlePubMedGoogle Scholar
- Raman K, Chandra N: Flux balance analysis of biological systems: applications and challenges. Brief Bioinform. 2009, 10: 435-449. 10.1093/bib/bbp011View ArticlePubMedGoogle Scholar
- Grafahrend-Belau E, Schreiber F, Koschutzki D, Junker BH: Flux balance analysis of barley seeds: a computational approach to study systemic properties of central metabolism. Plant Physiol. 2009, 149: 585-598. 10.1104/pp.108.129635PubMed CentralView ArticlePubMedGoogle Scholar
- Sahle S, Mendes P, Hoops S, Kummer U: A new strategy for assessing sensitivities in biochemical models. Philos Transact A Math Phys Eng Sci. 2008, 366: 3619-3631. 10.1098/rsta.2008.0108View ArticleGoogle Scholar
- van Riel NA: Dynamic modelling and analysis of biochemical networks: mechanism-based models and model-based experiments. Brief Bioinform. 2006, 7: 364-374. 10.1093/bib/bbl040View ArticlePubMedGoogle Scholar
- Schomburg I, Chang A, Schomburg D: BRENDA, enzyme data and metabolic information. Nucleic Acids Res. 2002, 30: 47-49. 10.1093/nar/30.1.47PubMed CentralView ArticlePubMedGoogle Scholar
- Hunter GJ, Hamberg LM, Alpert NM, Choi NC, Fischman AJ: Simplified measurement of deoxyglucose utilization rate. J Nucl Med. 1996, 37: 950-955.PubMedGoogle Scholar
- Al Zaid Siddiquee K, Arauzo-Bravo MJ, Shimizu K: Metabolic flux analysis of pykF gene knockout Escherichia coli based on 13C-labeling experiments together with measurements of enzyme activities and intracellular metabolite concentrations. Appl Microbiol Biotechnol. 2004, 63: 407-417. 10.1007/s00253-003-1357-9View ArticlePubMedGoogle Scholar
- Wensink H, Benito-Lopez F, Hermes DC, Verboom W, Gardeniers HJ, Reinhoudt DN, Berg van den A: Measuring reaction kinetics in a lab-on-a-chip by microcoil NMR. Lab Chip. 2005, 5: 280-284. 10.1039/b414832kView ArticlePubMedGoogle Scholar
- Edwards JS, Ibarra RU, Palsson BO: In silico predictions of Escherichia coli metabolic capabilities are consistent with experimental data. Nat Biotechnol. 2001, 19: 125-130. 10.1038/84379View ArticlePubMedGoogle Scholar
- Nathan C: Role of iNOS in human host defense. Science. 2006, 312: 1874-1875. author reply 1874-1875. 10.1126/science.312.5782.1874bView ArticlePubMedGoogle Scholar
- Becker SA, Palsson BO: Context-specific metabolic networks are consistent with experiments. PLoS Comput Biol. 2008, 4: e1000082- 10.1371/journal.pcbi.1000082PubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.