Escherichia coli is widely cultivated under aerobic conditions in laboratory and industrial processes. The standard procedure for growing E. coli cells to high cell densities is the fed-batch technique, where the carbon substrate, e.g. glucose, controls the growth as a limiting factor. To minimize the volume change in the bioreactor, high concentrated glucose solutions are often used. In large-scale cultivation processes mixing is often insufficient for equal distribution of the substrate resulting in gradients in essential variables such as substrate concentration, dissolved oxygen tension and pH .
Concentrations of glucose far above the saturation constant of the Monod model are connected to high metabolic and respiratory activities. Consequently, due to the low solubility of oxygen in aqueous solutions, the dissolved oxygen tension (DOT) drops to zero already at relatively low cell densities depending on the oxygen transfer rate of the cultivation system which is typical in shake flask cultures . Also in large-scale glucose limited fed-batch processes with limited mixing high glucose concentrations in the feeding zone have been proposed . At typical cell densities, ranging in such reactors from 10 to 100 g L-1 of cell dry weight, the high local volumetric rates for consumption of glucose and oxygen easily cause oxygen depletion .
When exposed to oxygen limitation or anaerobic conditions E. coli shifts to anaerobic respiration, if the corresponding inorganic electron acceptors are available, or to fermentative metabolism. As a result of the fermentative metabolism oxidised fermentation products, such as formate, acetate, lactate, ethanol, and succinate are released to the cultivation medium and consequently the pH of the medium may decrease. The transition to oxygen limitation is rather sharp, as the oxygen uptake follows Michaelis Menten kinetics and the KM value for oxygen is very small, between 10-7 and 10-8 M for E. coli . This correlates to less than 0.2% of oxygen saturation, which is so small that an exact analysis of the critical DOT level is not possible with standard DOT electrodes.
Indication of anaerobic metabolism in the glucose feeding zone of large-scale bioreactors came originally from studies in a scale-down two-compartment bioreactor system, where the synthesis of the above mentioned side metabolites was observed in the plug flow reactor compartment simulating the feeding zone [3, 5]. Interestingly, despite the local oxygen limitation in the feeding zone, these anaerobic products are re-assimilated in the glucose limited, oxygen sufficient parts of the bioreactor, which normally make up more than 90% of the reactor volume [1, 3]. The re-assimilation rates for acetate and lactate appear to be higher than for formate. Consequently, formate can be found as a side product in large-scale processes and is a clear indicator of anaerobic metabolism [1, 6, 7].
Additionally formate has been also found in small-scale processes where the DOT was kept above 30% and that therefore should be clearly aerobic. Castan et al.  showed that formate accumulation in these cultures was due to cell lysis. The authors verified by comparing their results to cultivations where a DNA binding polymer was added, that released DNA bound to cells was the reason of formate production. The DNA formed an extra diffusion barrier around the cells leading to decrease in oxygen transfer.
Physiologically the observation that formate is accumulated under conditions of oxygen limitation is interesting, as formate is toxic and typically further metabolised to dihydrogen. However, only few studies in bioreactors approached the analysis of dihydrogen in the context of reactor mixing of E. coli cultures [9, 10]. Cleland et al.  studied the evolution of dihydrogen gas at various oxygen uptake rates in mineral salt medium cultures. The conclusion was that the relative formation of dihydrogen gas was lower than the formation of other anaerobic metabolites including formate. Dihydrogen evolution increased linearly only during the first hour after the oxygen limitation but was then rapidly diminished. The authors proposed that dihydrogen production was occurring with very low rate or that the produced dihydrogen was consumed in another reaction inside the cell.
The disproportionation of formate to carbon dioxide and dihydrogen without nitrate or oxygen as exogenous terminal electron acceptors is catalyzed by the formate hydrogenlyase (FHL) complex, consisting of formate dehydrogenase (FDH, fdhF gene product) and of six other proteins encoded by the hyc operon. FHL functions as a membrane-integral electron transfer chain which finally releases dihydrogen by the HycE hydrogenase subunit (hycE). In E. coli exist three FDH isoenzymes, all sharing the same mechanism for formate cleavage. These isoforms are differentially expressed in dependence on the availability of exogenous electron acceptors. Dihydrogen is synthesised by FDH-H only, while the other isoforms, FDH-N and FDH-O, perform the reaction with nitrate as terminal electron acceptor. All three isoenzymes of FDH are molybdo-seleno proteins and HycE is a NiFe hydrogenase. Therefore a functional FHL complex is dependent on trace amounts of molybdenum, selenium, and nickel in the growth medium [11, 12].
These essential trace elements for the function of the FHL pathway are generally added into the cultivation medium of anaerobic cultures [13, 14]. However, when checking different cultivation media which are used for high cell density cultivation of E. coli in small or large bioreactor scales, we experienced that none of them contained selenium or nickel. Only few media contained molybdenum (cf. [3, 15–20]).
Oxygen limitation occurs generally in later phases of shake flask cultures and is also experienced by cells in large-scale bioreactors during their passage through the feeding zone. Therefore we considered it highly interesting to study the effects of the addition of the extra trace elements molybdenum, selenium, and nickel on the dynamics of the anaerobic metabolites and the total behaviour of the culture. We expected a lower level of formate accumulation, which would be a positive effect, as formate may be similar toxic to cultures as acetate. However, one also might postulate that the activation of the FHL complex, which leads to release of dihydrogen and carbon dioxide, may negatively affect the carbon yield due to the loss of formate as a re-metabolisable carbon source. Furthermore, the produced carbon dioxide may either positively influence the biomass yield by re-entering the cell's metabolism in carbonylation reactions, or negatively by accumulation in the cultivation medium to toxic concentration.