The novel cultivation system described here combines an enzymatic glucose release system with an optimized medium (mixture of mineral salts and complex additives) in a way which provides high cell densities together with high recombinant protein yields in shaken cultures. Cell growth can be controlled and a favorable pH level can be maintained during the whole cultivation. Due to the growth control, accumulation of harmful metabolites can be minimized and appearance of anaerobic conditions can be avoided. The cultivation takes one day longer to perform compared to the standard cultivation system, but regarding the "hands-on time" it is even easier and faster to perform. Since the cultivation volumes can be substantially decreased, several cultivations (e.g. many strains or an optimization set) can be performed at the same time. The possibility to use lower cultivation volumes also enables the use of microwell plates or deep well plates, which brings new applications for robot-driven automated cultivations.
With EnBase Flo, 10 to 20-fold higher induction cell density could be used, and the final cell densities were typically 10 times higher compared to the standard Sambrook protocol. This increase was coupled with equal or improved recombinant protein productivity per cell. The higher yield of soluble protein fraction is probably due to the extended protein synthesis period (24 h) under favorable conditions. These described improvements were achieved at a cultivation temperature of 30°C. There was no need to use lower temperatures, since optimal conditions for a long induction and controlled growth were obtained by the new cultivation technology. Cultivation temperatures as high as 37°C have also been applied with EnBase Flo and found to provide high soluble protein yield (data not shown). However, we have chosen to generally use 30°C for EnBase Flo cultures due to higher oxygen solubility and less evaporation. Also the number of functional ribosomes in the cell is known to be higher at lower temperatures . It is also possible that the overall changes in cell physiology occurring at fed-batch cultivation mode and elevated cell density may promote improved protein folding, possibly through increased synthesis of folding-assisting proteins. Protein yield of EnBase Flo cultures was relatively high already after 6 h induction, but often most of it was in insoluble form. This suggests that the bacterial growth rate immediately after boosting (i.e. addition of complex nutrients) was not low enough to provide correct protein folding. However, when a longer induction time (24 h) was applied, the protein produced was usually in soluble form, which was not the case with prolonged induction in MSM, LB and TB media.
An increased proportion of soluble protein by the use of the enzyme-based fed-batch technology has been earlier demonstrated by Panula-Perälä et al. . They reported lower total productivity per cell in their system compared to standard medium (M9 in that case), but achieved 8-10 times increased volumetric protein yield with EnBase due to the drastically higher cell density, and a higher proportion of TIM (triosephosphate isomerase) in a soluble form. Although the slow growth during pre-induction phase may pre-adapt the cell to stress , it also adjusts the protein synthesis system to a very low activity. This may explain why the advantage of high cell yield could not be directly translated to high recombinant protein productivity in the system described by Panula-Perälä et al. Therefore some major improvements were made to reach the good performance of the system described in this article. The new system is not anymore based on plain mineral salt medium composition. The utilization of a low amount of complex medium additives for the first overnight cultivation results in a slightly higher cell yield together with an improved pH balance. Also the addition of a high dose of complex nutrients ("boosting") at the time of induction provides a sufficient supply of the key medium compounds (e.g. amino acids, trace metals, cofactors and vitamins) needed for the setup of efficient protein synthesis. Beneficial effects of complex additives have been earlier reported by some bioreactor studies. For example, Tsai et al.  reported a 10-fold increase in intracellular human IGF-1 accumulation by the addition of yeast extract and tryptone. There are also some indications that product stability may be improved by the use of complex additives. Swartz  hypothesized in a review on recombinant DNA technology that complex additives may also reduce product proteolysis. It is however possible that instead of actually reducing proteolysis, complex nutrients might just compensate for it due to prolonged maintenance of efficient protein synthesis. This view is supported by our observation that an extended protein synthesis period results in higher amount of soluble recombinant protein.
Detailed information about the cellular mechanisms of E. coli during cultivation in EnBase Flo medium is not yet available, but the cultivation supposedly proceeds as follows: bacteria are first cultivated overnight in a balanced medium, where the growth can be tightly controlled by enzymatic glucose feeding (phase 1 in Figure 4). Thereafter, complex nutrients are added ("boosting") at the same time with an inducer to prepare the cells for efficient recombinant protein production. This medium boosting is beneficial since the recombinant protein synthesis has a major effect on growth and cell maintenance . When growth rate is very low most of the metabolic energy is allocated to cell maintenance instead of the recombinant protein synthesis. Additionally, the "boosting" also helps to maintain the required pH. An increase in growth rate is temporarily achieved by the boosting step, and there is approximately 2-fold increase in cell density within 6 hours (phase 2a in Figure 4). Such a moderate increase in biomass implies that despite the presence of rich medium components the growth is not fully uncontrolled. Possibly the enzymatic glucose delivery system keeps the sugar phosphotransferase transport system at least partly saturated , thereby decreasing the utilization of alternative carbon sources and allowing some level of growth control. The increase in growth rate after medium boosting is associated with an increase in protein synthesis, which results in efficient recombinant product formation. However, the product formation rate is probably lower than that of the recombinant protein production phase typical for normal shake flask processes in rich media. With prolonged induction (further 18 h, phase 2b in Figure 4) the cell number approximately only doubles, and protein production switches to a transient phase where the total amount of recombinant protein does not very much change in the cell. However, it is probable that new proteins are slowly synthesized while the old, partly insoluble, proteins are gradually degraded or even re-folded. Carrió and Villaverde [22, 23] have deduced that only a restricted amount of recombinant proteins can be coprocessed (correctly folded) together with the host proteins. Therefore the availability of folding-promoting proteins (chaperones) may limit the correct folding. In this respect, the lower protein synthesis rate in fed-batch cultivation may be very beneficial. Prior to induction, the slow growth rate near to starvation limit may considerably increase the amount of chaperone-type proteins, and the few cell divisions that occur later possibly do not dilute them away. Also the cell physiology during fed-batch high cell density growth can be very different from the batch growth in shake flask cultures, and especially the higher amount of stress proteins related to low growth rate may affect protein folding.
The fastest growth occurs immediately after the addition of booster and inducer, when the cell number is often doubled within 4 to 6 hours. This growth rate (μ = 0.15 to 0.25) is just slightly higher than the growth rate known to induce starvation response , but has been earlier shown to be the optimal growth rate for recombinant protein production. Additionally the complex medium additives in the booster solution provide easily available raw material for protein synthesis. Such a growth rate is much lower than growth rates in the Sambrook process (cell number typically doubled within 20 to 40 minutes). Our results suggest that this reduction in growth rate effectively enhances correct folding of recombinant proteins. The traditional approaches to reduce protein synthesis rate and promote correct folding include the use of low temperatures [24, 25], use of low inducer concentrations , and the use of extreme stress conditions like low or high pH . Our cultivation method can, however, provide the control of protein synthesis rate at the usual cultivation temperatures (30 or 37°C) and with commonly used IPTG concentrations (0.4 or 1 mM). The strategy to reduce the metabolic load of cells during recombinant protein synthesis can be further applied to EnBase Flo as well as to any other cultivation systems.
The common approaches for pH maintenance during cultivation are based on the use of well-buffered culture media. Although such buffering prolongs the time of efficient bacterial growth, the buffering capacity can be suddenly lost when the amount of H+ or OH- ions exceeds the capacity of the buffer. Furthermore, high concentrations of buffering agents such as 3-(N-morpholino)-propanesulfonic acid (MOPS) and tris-(hydroxymethyl)aminomethane (Tris) may inhibit the growth of bacteria. It should also be noted that no pH buffering system can prevent the accumulation of harmful metabolites such as acetate into the cultivation medium. The pH maintenance system of EnBase Flo is based on another operation principle than buffering system. We have observed that there are remarkable differences between mineral salt media and rich media with respect to the development of pH level (see Table 1 as an example). In mineral salt based media pH tends to drop. This pH drop is not only caused by accumulation of acidic by-products, since it occurs also in fed-batch cultivations, as well as in E. coli strains that do not produce substantial amounts of acetic acid, but is due to the consumption of ammonia during the growth. In contrast, in E. coli cultivations performed in rich cultivation media the pH usually tends to increase. This is very likely a consequence of the utilization of complex compounds (e.g., yeast extract, protein peptones, casamino acids) as a carbon source. During limitation or absence of glucose or other easily-assimilable carbon sources, E. coli uses complex compounds for energy generation, and the useless nitrogen will be secreted into the medium as ammonium ions. Therefore, the use of glucose-limited fed-batch together with an adequate concentration of complex additives provides new possibilities for pH control. The rate of glucose delivery can be optimized together with the amount of complex medium additives to provide a suitable pH level throughout the cultivation in a simple phosphate buffer system.
The shake flask cultivation results described in this article were obtained in normal round-bottom Erlenmeyer flasks. In contrast the Structural Genomics Consortium recommends the use of baffled shake flasks  for efficient aeration and cell growth. We, however, recommend the use of ordinary round-bottom Erlenmeyer flasks for the following practical reasons: 1) the EnBase Flo system brings such an efficient growth control that the improvement in oxygen transfer by baffles is not necessary, and 2) improved mixing by baffles is associated with upwards spilling and increased foaming. For a long-term cultivation it is extremely important that the oxygen-permeable membrane or other closure used is always dry, since the wetting of the closure, as well as extensive foaming, can significantly reduce oxygen transfer , and therefore the use of baffled flasks can actually be counterproductive with respect to oxygen transfer. Closure type  and flask filling volume  are factors that have a profound effect on oxygen transfer and consequently the performance of cultures, but are however often not appropriately considered by researchers.
The EnBase Flo cultivation method is not bound to a certain promoter system; it has been successfully used with arabinose- or tetracycline-inducible expression systems (data not shown). Therefore, additional control of protein synthesis can be obtained by the use of weaker promoters (i.e. other than the T7-promoter system). The possibility to exploit extended induction times could demonstrate the benefits of weak promoters much better than the use of standard shake flask cultures. In this way, it may be possible to get relevant data for the development of bioreactor cultivations already in a small scale. The results with E. coli RB791 [pQE30:Adh] suggest that the standard shake flask-based recombinant protein production process may favor the T7 expression system. This clone is known to produce high amounts of active ADH in glucose-limited fed-batch in mineral salt medium . The gene adh has been cloned under the control of phage c6 promoter which is recognized by the native E. coli RNA-polymerase. High yields of soluble ADH were obtained in EnBase Flo cultures, while the other cultivation media exhibited poor production of soluble ADH. Therefore it seems probable that if the expression system for bioreactor cultivations was selected on the basis of shake flask cultivations with ordinary media, such potent expression strains like the strain RB791 [pQE30:Adh] would possibly be discarded and a T7-polymerase based system selected instead.
The scale-up of EnBase Flo cultivations from 3 ml culture volume (in 24 DWP) to 50 ml culture volume (in 500 ml shake flask) didn't cause deterioration of the protein production. This indicates a good scalability of the system. If required, further fine-tuning can be easily achieved by enzyme dosing. The possibility to obtain high cell densities makes EnBase Flo very appealing for screening and automatic sample treatment purposes. It has also provided very good results with leaky expression systems, provided that appropriate precautions are taken in the preparation of the inoculant cultures (data not shown). A cultivation system that provides a high yield of recombinant proteins in soluble form is also invaluable for researchers performing affinity purification of His-tagged proteins, since this scheme relies entirely on the presence of soluble proteins. Further, EnBase Flo could be a powerful tool for growth rate optimization in small scale. Bioreactor process optimization studies have demonstrated that the highest recombinant protein production level is obtained with a certain growth rate. With enzymatic fed-batch technology, this optimal growth rate can be experimentally determined in shake flasks or deep well plates.