Comprehensive analysis of metabolic sensitivity of 1,4‐butanediol producing Escherichia coli toward substrate and oxygen availability

Nowadays, chemical production of 1,4‐butanediol is supplemented by biotechnological processes using a genetically modified Escherichia coli strain, which is an industrial showcase of successful application of metabolic engineering. However, large scale bioprocess performance can be affected by presence of physical and chemical gradients in bioreactors which are a consequence of imperfect mixing and limited oxygen transfer. Hence, upscaling comes along with local and time dependent fluctuations of cultivation conditions. This study emphasizes on scale‐up related effects of microbial 1,4‐butanediol production by comprehensive bioprocess characterization in lab scale. Due to metabolic network constraints 1,4‐butanediol formation takes place under oxygen limited microaerobic conditions, which can be hardly realized in large scale bioreactor. The purpose of this study was to assess the extent to which substrate and oxygen availability influence the productivity. It was found, that the substrate specific product yield and the production rate are higher under substrate excess than under substrate limitation. Furthermore, the level of oxygen supply within microaerobic conditions revealed strong effects on product and by‐product formation. Under strong oxygen deprivation nearly 30% of the consumed carbon is converted into 1,4‐butanediol, whereas an increase in oxygen supply results in 1,4‐butanediol reduction of 77%. Strikingly, increasing oxygen availability leads to strong increase of main by‐product acetate as well as doubled carbon dioxide formation. The study provides clear evidence that scale‐up of microaerobic bioprocesses constitute a substantial challenge. Although oxygen is strictly required for product formation, the data give clear evidence that terms of anaerobic and especially aerobic conditions strongly interfere with 1,4‐butanediol production.


| INTRODUCTION
The chemical industry currently undergoes a paradigm change and progress is achieved to replace fossil based resources by renewable ones. [1][2][3] In recent studies a set of chemical compounds have been identified that can serve as intermediates for bridging biological and chemical processes. 4,5 In the list of C4-compounds 1,4-butanediol (BDO) is listed which can serve as chemical synthon besides its direct application as alcohol monomer compound for polyester synthesis. 6,7 Bio-based production of diols has been achieved in several microbial systems, such as 2,3-butanediol in Saccharomyces cerevisiae, 8 Bacillus subtilis, 9 Bacillus licheniformis 10 and many more, 1,3-butanediol in Escherichia coli, 11 1,3-propanediol in Klebsiella pneumoniae, 12 Clostridium butyricum, Lactobacillus brevis, and more, 13 and most successfully 1,4-butanediol in Escherichia coli. [14][15][16] Because of its commercial interest as bio-based monomer for polyester synthesis the economic evaluation clearly shows, that BDO production requires large scale bioreactor operation to meet market demand as well as reasonable process economics. 14 This goes along with increasing bioreactor inhomogeneity, that is, spatio-temporal changes of environmental conditions for certain volume elements. Such effects can be simulated under lab scale conditions following the scale-down approach with compartmented bioreactor systems, providing frequent oscillation in the environmental conditions while parts of the culture pass the additional compartments. 17,18 Hence robustness studies toward oscillating environmental conditions are continuously gaining higher relevance. Such studies provided detailed insight into microbial production of for example valinomycin in E. coli, 19 L-lysine in Corynebacterium glutamicum, 20 preproinsulin in E. coli 21 or even the efficiency of plasmid DNA production in general. 22 Depending on the microbial system and the redox state of substrate and product, microbial production processes can be distinguished. Many of the current production processes are operated under aerobic conditions, 23 but an increasing number of processes were developed which enable product formation under nonaerated or anaerobic conditions, 24 for example, isobutanol, 25 succinate 26 as well as propionate. 27 The decisive factor for successful production processes under oxygen limitation is a suitable substrate/product combination. Such a substrate/product pair is characterized by meeting the energy demand for substrate to product conversion under oxygen limitation including transport mechanisms, possible product excretion and feasibility in commercial scale. 28 In this sense BDO represents an interesting example, since the metabolic pathway from substrate glucose to BDO requires microaerobic conditions. Oxygen deprived conditions are obligatory to enable the network to regenerate sufficient reduced redox equivalents in form of NADH to fuel the reduction of the substrate glucose to the more reduced product BDO.
The heterologous biosynthetic pathway to BDO in E. coli ECKh-422 is illustrated in Figure 1. It is branching from the TCA cycle either from alpha-ketoglutarate or from succinyl-CoA ( Figure 1). The route from succinyl-CoA requires a reduction to succinyl semialdehyde catalyzed by succinate semialdehyde dehydrogenase (from Porphyromonas gingivalis, sucD) in the first step. This reaction consumes NADH and is consequently more expensive for the cell than the decarboxylation from alpha-ketoglutarate via alpha-ketoglutarate decarboxylase (from Mycobacterium bovis, sucA) to succinyl semialdehyde. Hence, the route via decarboxylation of alpha-ketoglutarate is thermodynamically more favorable and seem to be preferred and used up to 95%. 15 Succinyl semialdehyde is further reduced by 4-hydroxybutyrate dehydrogenase (from P. gingivalis, 4hbd) into 4-hydroxybutyrate (GHB).
The following step is catalyzed by the 4-hydroxybutyryl-CoA transferase (from P. gingivalis, cat2) consuming one molecule acetyl-CoA and producing one molecule acetate and 4-hydroxybutyryl-CoA. Subsequently two consecutive reduction steps lead to 4-hydroxybutyraldehyde and F I G U R E 1 Heterologous BDO pathway in Escherichia coli ECKh-422 finally to the target product BDO. The first reduction is done by 4-hydroxybutyryl-CoA reductase (from Clostridium beijerinckii, 025B) and the last step in the pathway is done by native alcohol dehydrogenase of E. coli. 15 Most studies in the field of microaerobic BDO production with

| Cultivation conditions
Cultivations were performed in parallel bioreactor system of DASGIP ® (Eppendorf, Hamburg, Germany) with a working volume of 1 L in M9 minimal medium (6.78 g l −1 Na 2 HPO 4 , 3.0 g l −1 KH 2 PO 4 , 0.5 g l −1 NaCl, 2.0 g l −1 NH 4 Cl, 1.0 g l −1 [NH 4 ] 2 SO 4 , 1 mM MgSO 4 , 0.1 mM CaCl 2 ). The temperature was held at 37 C and the pH was titrated to seven by addition of 2 M NH 3 and 30% H 3 PO 4 . The cells were grown in an initial aerobic growth phase with dissolved oxygen regulated at 30% until an OD 600 of 10, where the culture was induced with 0.25 mM IPTG. One hour later the culture was switched to microaerobic conditions for BDO production. Therefore the base was changed to 2 M Na 2 CO 3 and stirring was set constant at 700 rpm and a gassing rate of 1, 2 or 6 sl hr −1 was adjusted with OTR max of 7.4, 9.8, and 20.9 mmol l −1 hr −1 , respectively.

| Analytical procedures
Cell dry weight was determined from 2 ml cell suspension in dried reaction tubes (> 48 hr, 80 C). After centrifugation at 16,060g for 10 min, the supernatant was discarded and the cell pellet was washed with 0.9% (wt/vol) NaCl. Subsequently, the cell pellet was dried at 80 C for 48 hr following the gravimetric determination of CDW.

| Influence of substrate availability
The BDO production strain was cultivated under glucose limited and excess conditions. All cultivations were started with an initial aerobic batch phase for biomass growth. Afterward a switch to microaerobic conditions led to steady decrease of growth rate until non growing conditions are reached with a maximal total biomass of 7 and 6.8 g for glucose excess and limiting conditions, respectively ( Figure 2a). To establish microaerobic conditions an aeration rate of 2 sl hr −1 was chosen which was initially derived from Yim, et al. 15 and used for initial bioreactor cultivation experiments (data not shown). For limiting glucose conditions a feed rate of F(t) = 0.01 ml hr −1 * t + 2.4 ml was used, because weak growth was still assumed for the microaerobic phase according to Yim, et al. 15 Feed rate for glucose surplus conditions was set to F(t) = 0.03 ml hr −1 * t + 3 ml. Nevertheless, during cultivation under glucose surplus conditions in the mid of production phase glucose concentration fell below 20 mM and increased again in the later phase up to the final value of 120 mM (Figure 2c). This indicates that glucose consumption rate is higher under oxygen limitation than during aerobic growth phase, 30,31 but seems to stay in a constant range over process time under microaerobic conditions. Although, the glucose concentration showed a strong decrease between 23 and 31 hr glucose excess can be assumed throughout the experiment, since the lowest residual glucose was still measured at 2-3 mM in the cultivation supernatant. In terms of process economics the linear feed rate was not optimal for the microaerobic production phase due to glucose overfeed beginning from 47 hr, but the strain behavior was not negatively affected by higher glucose concentrations. In the glucose limiting experiment the glucose concentration was below level of detection, so that limiting conditions were ensured.
During the microaerobic phase BDO production is observed showing almost linear increase of BDO until the cultivation was stopped. A final BDO concentration of 281 and 198 mM were found for the glucose excess and limiting process after 71 hr production phase, respectively (Figure 2b). The comparative analysis clearly showed that the glucose excess conditions provided better process performance in terms of volumetric product formation with 0.08 g P g CDW −1 hr −1 ± 0.01 for glucose excess and 0.05 g P g CDW −1 hr −1 ± 0.00 for glucose limitation. Furthermore a higher substrate related product yield was measured under glucose excess Y P/S = 0.22 g P g CDW −1 ± 0.01 compared to substrate limited conditions Y P/S = 0.18 g P g CDW −1 ± 0.00 (cf. Hence, it is of high interest to investigate the dependency of the BDO formation with respect to the amount of oxygen supplied.
The bioprocess performance data of different conditions are summarized in Table 1.
The BDO producing strain was cultivated in the presence of glucose excess with the initial aerobic growth phase, followed by the switch to microaerobic conditions with aeration ranging from 1 to 6 sl hr −1 air ( Figure 3). All cultivations showed a dissolved oxygen of 0% in this phase and were therefore stated as microaerobic. Notice that 2 sl hr −1 was used as standard reference value for the experiment with  The initial aerobic growth pattern is very similar for the three conditions, and the microaerobic phase with aeration rate of 1, 2, and 6 sl hr −1 started with similar biomass of~4 g (Figure 3a). In the first 8 hr of the microaerobic phase a weak growth is ascertainable, finally leading to a non-growing state. Despite of this, a remarkable negative production phenotype is observed for 6 sl hr −1 (Figure 3b). Compared to the reference cultivation with 2 sl hr −1 , the higher supply of oxygen reduces final product titer to 63 mM (−77%). Further reduction of the aeration rate to 1 sl hr −1 shows comparable BDO titer at the end, but improved substrate specific product yield up to 0.24 ± 0.03 g P g S    Figure 1).
The carbon balances show a gap in the range of 5-11% of carbon, which is not covered by routine analytics. Therefore an untargeted metabolite profiling of the culture supernatant was done to identify unknown compounds and compare the metabolite spectrum.
In total, 211 metabolites were positively identified in the cul- which decrease from 1 to 6 sl hr −1 . In general, the relative comparison of the metabolites can confirm cultivation data. The experiments with lower aeration rate (1 and 2 sl hr −1 ) already showed a similar pattern in terms of biomass, BDO and CO 2 formation, while higher aeration rate (6 sl hr −1 ) resulted in strong differences in product range especially the massive decrease in BDO production.

| CONCLUSIONS
The results demonstrate the importance of oxygen limiting conditions for the E. coli BDO strain used in this study. E. coli ECKh-422 serves as a prototype for an industrial BDO production strain. Regarding the substrate supply, it was shown that substrate excess is the best choice for the highest productivity, so that potential gradient formation resulting from substrate limited fed-batch operation do not have to be considered for scale-up. On the contrary, changes in oxygen availability revealed strong impact on product formation. The metabolic network of the BDO production strain is tailored for optimal functionality during microaerobic conditions 15 (cf. Figure 1). This work shows that any deviation from the optimal oxygen deprivation toward higher oxygen supply leads to much reduced BDO product formation and high by-product formation, that is, mainly acetate and carbon dioxide.
Microaerobic conditions were tested under different degrees of oxygen deprivation, where in all cases no dissolved oxygen concentration was measurable. Aeration rates of 1, 2, and 6 sl hr −1 were investigated and a reduction of the final BDO concentration up to 77% was observed with the highest aeration rate. Furthermore, the by-product range is strongly influenced by the oxygen supply (cf. Figure 4).