Kinetic models, which predict the behaviour of metabolic reaction networks under different conditions, are indispensible to fully and quantitatively understand the relation between a product pathway and connected central metabolism. In this thesis the focus was to develop tools for future in vivo kinetic modeling in Escherichia coli. The relations between fluxome and metabolome at steady-state or transient state but where enzyme levels can be assumed constant are investigated. In addition, we aimed at stoichiometric and thermodynamic analyses that are also fundamental to comprehend the cellular systems. Network stoichiometry is a significant aspect of kinetic models. Stoichiometric metabolic networks usually lack proper biomass composition and the exact stoichiometry of energy generating and consuming processes, e.g. the efficiency of ATP generation in oxidative phosphorylation (P/O ratio) and growth dependent and growth independent maintenance energy requirements (i.e. KX and mATP). Chapter 2 presents the determination of biomass composition of glucose-limited grown E. coli K12 cells, derivation of minimal network subsets from the genome-scale reconstruction for carbon-limited chemostat cultivations on minimal media and the estimation of the in vivo ATP stoichiometry parameters of this metabolic network. This allowed deriving Herbert-Pirt relations, which are of value, e.g. for quantification of transient growth rate (based on ATP balance), as was shown in Chapter 5. Following in Chapter 3, the ATP related energy aspects of a stoichiometric metabolic network was studied within a thermodynamic framework. Case studies were production of dicarboxylic acids; fumaric (in eukaryotes) and succinic (in prokaryotes). Results point out the importance of black box analysis in learning the energy-related capabilities of the network of interest, e.g. Gibbs energy formation of the theoretical anaerobic reaction indicated that the maximal theoretical product yields are anaerobically feasible, also at low pH (< 3). Considering a required high extracellular total acid concentration, thermodynamically feasible active transport mechanisms were found to be H+ antiport at pH 3 and H+ symport at pH 7 for both acids. Subsequent detailed network analysis showed where biologically useful energy could be consumed, leading to novel metabolic engineering targets to enable fully anaerobic production at theoretical yields. Given that a kinetic model needs information on metabolite concentrations, first a proper method is required to get accurate data. Measurement of intracellular metabolite concentrations requires rapid sampling, instantaneous quenching of metabolic enzymes activity with absence of leakage and removal of the extracellular medium as well as extraction of metabolites taking into account the high turnover rate of these compounds. In Chapter 4 different quenching protocols that are variations of the most commonly applied method to remove extracellular metabolites, namely quenching with cold methanol and subsequent washing of the cell pellet, for arresting cellular metabolic activity in E. coli were investigated. From accurate LC-ESI-ID-MS/MS or GC-MS measurements of central metabolites, amino acids and adenine nucleotides, metabolite leakage during cold methanol quenching was quantified using a rigorous balancing approach. Consequently, a differential method, whereby broth samples are rapidly withdrawn from the chemostat with a dedicated rapid sampling device and instantaneously quenched (< 180 ms) and culture filtrate is obtained by direct filtration of broth, was developed. To gain information on in vivo enzyme kinetics, generally stimulus-response experiments are performed. In Chapter 5 a quantitative analysis of flux and metabolite response during a perturbation of a glucose-limited grown E. coli culture with a glucose-pulse directly applied to the bioreactor, is presented. Flux quantification was based on the uptake/secretion rates, which are determined by mass balance equations and ATP balance. Metabolite quantification for 37 metabolites in glycolysis, tricarboxylic acid (TCA) cycle, pentose phosphate pathway (PPP), nucleotides and amino acids was carried out with the proposed differential method (in Chapter 4). A novel approach to resolve fluxes at seconds time scale was based on the calculation of qO2(t) from dynamic dissolved O2 (DO) mass balance and measured DO levels. The obtained dynamic flux and metabolite concentration-time patterns in a time period of 50 s were highly corresponding. However, the observed remarkable increase in the growth rate (3-4 fold within 50 s), and hence protein synthesis rate was not reflected in changed amino acid concentrations. This suggested that the transcripts are increased in this short time frame. In vivo relations between growth rate and fructose-1,6-bisphosphate (FBP) concentration and between oxygen uptake rate and AMP/ATP ratio were found. Since E. coli is shown to respond very quickly (to glucose perturbations), sampling at very short intervals is required. The device BioScope, designed by our group, fulfills this purpose with an additional advantage that pulses are given outside the bioreactor thus without disturbing the running culture and allowing multiple pulse response experiments. Chapter 6 describes the characterization and application of the new BioScope II device, which was redesigned for E. coli perturbation experiments. Therefore the total observation time window as well as the sample intervals was decreased by shortening the BioScope channel, leading to a time window of 0 - 8 and 0 - 40 s for flow rates of 4 and 2 ml/min respectively. Furthermore the oxygen transfer characteristics were improved by increasing the ratio between the membrane area for gas transfer and the liquid volume. The performance of the modified BioScope is demonstrated by showing that the results of glucose perturbations carried out in the BioScope, coupled to a steady-state chemostat culture of E. coli, and results of similar perturbations carried out directly in the chemostat, are very comparable. Hereby it was ensured that oxygen non-limited conditions were maintained in both systems. For proper metabolite quantification we have applied the differential method, which we described in Chapter 4. The obtained results clearly showed the fast dynamic behaviour of the phosphotransferase system (PTS) in E. coli. The BioScope does not allow flux quantification and therefore a pulse experiment should be performed both in the bioreactor (for flux quantification, Chapter 5) and in the BioScope (for fast metabolite dynamics, Chapter 6). In Chapter 7 in vivo dynamic response of E. coli, grown at glucose-limited chemostat, to glycolytic (glucose) and gluconeogenic (pyruvate, succinate) substrates were investigated in both platforms. Time-resolved flux quantification at seconds time scale was based on the uptake/secretion rates, which are determined from dynamic mass balance equations and the degree of reduction balance. After each added different substrate the system (flux and metabolites) achieved pseudo-steady-state in about 30 - 40 s and a huge oxygen uptake capacity of the cells was observed, and interestingly within 40 s the growth rate reached from its steady-state value of 0.13 h-1 to 0.3 h-1 for each different substrate pulse, indicating a capacity limit in e.g. ribosomes. The observed dynamic responses showed massive reorganization and flexibility (up to 12 - 100 fold change) of intracellular fluxes following the different pulses, which matches with the steep changes in metabolite levels leading to dynamic shifts in mass action ratio’s of pseudo-/near-equilibrium reactions and flux inversions. From the dynamic intracellular metabolite and flux information, the in vivo kinetics was studied based on a simplified thermodynamic approach. It was found that several enzymes showed simple nearequilibrium kinetics as found before in baker’s yeast (Q-linear kinetics). In Chapter 8 a comparative study of steady-state and dynamic metabolite and flux responses to localized pulses for succinate overproducing E. coli mutant is presented. In this mutant two transcription factors of glyoxylate pathway are deleted, which did not seem to affect this pathway but there was a significant effect on improved energy efficiency. The mutant had four functional mutations: increased succinate exporter, a deleted succinate importer, deletion of succinate dehydrogenase and a PEP carboxylase with increased capacity due to a point mutation. These mutations have a clear effect on steady-state and dynamic behaviour. In general the mutant showed much lower maximal uptake rates and succinate export was successfully implemented. Succinate import was 26 fold decreased. Also the mutant showed enormous dynamic flux flexibility. Compared to the wild type a considerable shift occurred from TCA cycle to oxidative PPP, including inversion of pyruvate kinase. Even more flexibility was observed on succinate uptake which seems only possible by an inverted TCA cycle. These flux shifts were in agreement with large and delocalized changes in metabolite levels. Clearly succinate dehydrogenase deletion caused large changes in metabolite levels close to and far from the deleted reaction. The mutant as the wild type showed extreme homeostatic behaviour for energy charge in the pulses. In contrast, large changes in redox level NAD+/NADH occurred where the mutant showed even larger changes. This large redox change can be associated to reversal of flux direction. In Chapter 9 a different application area, which is large-scale mixing problems, of perturbation experiments is presented. Large-scale bioreactors are known to have heterogeneous conditions and several scale-down studies have reported the response of cells to glucose and DO gradients, whereby the glucose gradients have been attributed to cause DO depletion. The situation was mimicked with a two-compartment bioreactor system consisting of an E. coli chemostat culture connected to a BioScope. Fully aerobic transition from a glucose-limiting to a glucose-rich region triggered the overflow metabolism very rapidly (< 2 s). When the glucose gradient was created under anaerobic conditions, many other mixed-acid fermentation metabolites were detected with higher production rates. This study demonstrated that secretion of fermentative by-products occurs even from very short exposures of cells to gradients in large-scale bioreactors and this situation might induce unfavorable process performance.