Introduction

Lactic acid is a widely used organic acid generated by microorganisms (Narayanan et al. 2004; Wasewar et al. 2004). Although lactic acid bacteria naturally accumulate lactate (Konings et al. 2000), metabolically engineered Escherichia coli also accumulates substantial lactate with high productivities (Dien et al. 2001; Zhou et al. 2006; Zhu et al. 2007). For example, a two-phase aerobic–anaerobic process with E. coli aceEF pfl poxB pps frdABCD generated 138 g lactate/l with an overall productivity of 3.5 g/l h (Zhu et al. 2007). Moreover, considering the non-growth phase alone, the process generated lactate at 0.9 g/g h, which corresponded to 6.3 g/l h at a cell density of 12 g/l. This particular two-phase process has two characteristics which offer opportunities for further improvement.

One process characteristic is that both glucose and acetate are required to support microbial growth in an initial aerobic phase. During the subsequent anaerobic non-growth phase about 3 g succinate/l accumulates despite the deletion of frdABCD which encode fumarate reductase. 13C-NMR demonstrated that succinate is biochemically derived from acetate remaining at the end of the aerobic phase (Zhu et al. 2007). This suggests that succinate production might be avoided if acetate were exhausted at the moment of transitioning between aerobic and anaerobic phases. A method is therefore needed to monitor acetate either directly or indirectly and then to use that “signal” as the switch to lactate production.

A second aspect is that the initial growth phase is “unproductive”, and the overall lactate productivity could be increased by prolonging the relative duration of the non-growth production phase. However, lactate formation is limited by ionic strength (Zhu et al. 2007). Therefore, the process might be prolonged by removing lactate and counterions during the production phase.

In the present study, we have monitored acetate indirectly by taking advantage of the growth requirement for this nutrient. Specifically, acetate disappearance should be accompanied by an increase in dissolved oxygen. By ensuring that acetate is nearly exhausted at the time of switching to the production phase, by-products generated from acetate should be minimized. Moreover, we investigate two methods to recycle cells during the anaerobic phase as means to prolong lactate generation.

Materials and methods

Organism and growth

Escherichia coli ALS974 (Hfr zbi::Tn10 poxB1 Δ(aceEF) rpsL pps-4 pfl-1 frdABCD) (see Zhu et al. 2007) was first grown for 6 h in a 250 ml shake-flask containing 30 ml TYA medium before transferring 5 ml to a 250 ml shake-flask containing 50 ml SF medium. After 10 h, these contents were used to inoculate a bioreactor containing GAM medium. All flasks were incubated at 37°C and 250 rpm (19 mm pitch). TYA medium contained (per l): 10 g tryptone, 5 g NaCl, 1 g yeast extract, 1.36 g Na(CH3COO) · 3H2O. SF medium contained (per l): 10 g glucose, 2.3 g sodium acetate · 3H2O, 5.66 g Na2HPO4 · 7H2O, 1.5 g KH2PO4, 0.25 g NaCl, 0.5 g NH4Cl, 0.1 g MgSO4 · 7H2O, 0.013 g CaCl2 · 2H2O, 0.02 g thiamine · HCl, 0.5 g l-isoleucine. GAM medium contained (per l): 20 g glucose, 11.52 g sodium acetate · 3H2O, 1.5 g NaH2PO4 · H2O, 3.25 g KH2PO4, 3.275 g K2HPO4 · 3H2O, 0.2 g NH4Cl, 2.0 g (NH4)2SO4, 1.024 g MgSO4 · 7H2O, 0.01 g CaCl2 · 2H2O, 0.5 mg ZnSO4 · 7H2O, 0.25 mg CuCl2 · 2H2O, 2.5 mg MnSO4 · H2O, 1.75 mg CoCl2 · 6H2O, 0.12 mg H3BO3, 1.772 mg Al2(SO4)3, 0.5 mg Na2MoO4 · 2H2O, 18.29 mg FeSO4 · 7H2O, 0.02 g thiamine · HCl, 0.75 g l-isoleucine. The centrifugation/resuspension buffer contained (per l): 20 g glucose, 1.5 g NaH2PO4 · H2O, 3.25 g KH2PO4, 3.275 g K2HPO4 · 3H2O. The buffered glucose solution used in ultrafiltration contained (per l): 30 g glucose, 1.5 g NaH2PO4 · H2O, 3.25 g KH2PO4, 3.275 g K2HPO4 · 3H2O.

Fed-batch process

Replicate two-phase fed-batch processes used a 2.5 l bioreactor (Bioflow 2000, New Brunswick Scientific Co. Edison, NJ, USA) initially containing 1 l medium. O2-enriched air maintained a dissolved O2 (DO) above 50% saturation. Feeding was automatically controlled (YSI 7200 Select Glucose Analyzer, YSI Inc., Yellow Springs, OH, USA) to maintain glucose above 10 g/l. After the cell concentration reached an OD600 of 25 (about 10 g/l DCW), the feed was changed from 400 g glucose/l and 150 g sodium acetate · 3H2O/l to 600 g glucose/l. The anaerobic phase was initiated as described in the results, and subsequently pH was controlled at 7.0 using 30% (w/v) Ca(OH)2.

Cell recycling

Two methods were examined to recycle cells during the anaerobic phase. Centrifugation/resuspension involved removing the bioreactor contents at three intervals. This broth was centrifuged at 10°C at 5,000g for 10 min, cell pellet resuspended in 1 l resuspension buffer and then returned to the bioreactor, a process that required about 45 min. Ultrafiltration involved decreasing the volume from 1.5 l to 500 ml through a membrane filtration cartridge (UFP-500-E-4A, 500,000 NMWC, 420 cm2, Amersham Biosciences Corp., Westborough, MA, USA) at 70 kPa transmembrane pressure. Then, 800 ml buffered glucose was pumped into the bioreactor and another production cycle started.

Analyses

Growth was monitored as the OD600. Concentrations of organic compounds were determined by liquid chromatography (Eiteman and Chastain 1997).

Results and discussion

Controlling acetate concentration at the end of aerobic phase

Acetate is required for growth of E. coli ALS974 due to deletions of both the pyruvate dehydrogenase complex (aceEF) and pyruvate formate lyase (pfl). During aerobic growth, the molar ratio of glucose to acetate consumption rates is about 2.5 (Zhu et al. 2007). To prevent acetate limitation when glucose is monitored, the feed contains more acetate than necessary. However, because the acetate concentration is not controlled, acetate unavoidably remains during the anaerobic lactate production phase. Residual acetate is converted to succinate anaerobically through the glyoxylate shunt (Zhu et al. 2007). To minimize succinate the switch between phases should occur just when acetate is depleted.

We reasoned that dissolved O2 (DO) might be a useful indicator of acetate depletion. Specifically, when the acetate concentration decreases to zero, cell growth should cease, and O2 uptake rate should markedly decrease. This event should be observed by a sudden increase in the DO (Ruottinen et al. 2007). A fed-batch process was conducted to determine if DO could be an indicator for acetate depletion. When the cell density (as OD600) reached about 25, the acetate/glucose feed solution was changed to one containing only glucose. By maintaining a constant agitation, gas flow rate and composition, the DO remained about 70–80 (Fig. 1). When the acetate concentration (as subsequently measured) decreased to about 0.5 g/l, the DO increased to 128% within 10 min (20.1–20.24 h, DO was above 100% because air was supplemented with O2). When the process was switched at the end of this interval, less than 0.1 g acetate/l remained. Over 133 g lactate/l was generated after 26 h, similar to previous results (Zhu et al. 2007). Moreover, with DO monitoring less than 1 g succinate/l and no ethanol were observed, significantly less than when DO was not used as an indicator of acetate exhaustion (Zhu et al. 2007). Six replicate fermentations confirmed that DO begins to increase when acetate decreased below 0.5 g/l, with acetate being depleted within 10 min. Ethanol and succinate formation can thus be minimized by careful control of acetate at the time of the switch to anaerobic conditions.

Fig. 1
figure 1

Dissolved O2 as a percentage of air saturation (DO, ●) and acetate concentration (∆) during a representative aerobic growth phase. The agitation and gas flowrate were fixed prior to the initiation of the anaerobic production phase at 20.37 h. The OD600 of the culture was approximately 25 (~10 g DCW/l)

Full size image

Cell recycling

Although ALS974 does not grow anaerobically, the specific lactate productivity remains high (above 0.8 g/g h) until lactate reaches 120 g/l (Zhu et al. 2007). Decreasing productivity is likely due to excess Ca2+ or lactate. To prolong a high rate of lactate production, cells could be recovered and resuspended in fresh medium containing only glucose. With this strategy cells essentially act as a biocatalyst converting glucose into lactate with minimal by-products. We examined two different means to recycle cells, centrifugation/resuspension and ultrafiltration.

The bioprocesses which included cell recycling were initiated with the same conditions as described above. For centrifugation, the process was paused at intervals during the anaerobic phase, the medium removed, centrifuged, and returned to the bioreactor with fresh medium for another cycle. For an example process during the first 8 h, 76 g lactate/l was generated for a productivity of 9.4 g/l h (Fig. 2). During a second 8 h, 69.6 g lactate/l was generated for a productivity of 8.7 g/l h, and subsequent cycles generated lactate at rates of 5.8 g/l h and 4.2 g/l h, respectively. By prolonging the process to 52 h (and the production phase alone to 34 h), the lactate productivity achieved 4.0 g/l h. Furthermore, because the switch from aerobic to anaerobic conditions coincided with acetate exhaustion (by using the DO) and acetate was absent in resuspension solutions, less than 0.5 g succinate/l and no ethanol were detected during the three cycles. Although centrifugation/resuspension provides fresh medium to almost the entire culture, this manual approach contributed to some cell loss and could potentially cause contamination. Centrifugation/resuspension is also cumbersome (Kim et al. 2004; Kwon et al. 2006).

Fig. 2
figure 2

Comparison of lactate production by ALS974 in representative two-phase fed-batch fermentations with and without cell recycling. The figure shows concentration of lactate in a fed-batch process without cell recycling (▲), the process with intermittent centrifugation/resuspension (◯) and the process with intermittent ultrafiltration (■)

Full size image

Ultrafiltration offers another means to recycle cells and prolong lactate formation (Choi et al. 2000; von Weymarn et al. 2002). The growth phase was conducted as previously, and the cells were ultrafiltered during the anaerobic phase. Ultrafiltration reduced the volume to a third, and then the feed was supplied based on glucose demand. In a representative ultrafiltration process the lactate productivity was 8.6 g/l h for the first 8 h (Fig. 2), and then 7.5 g/l h, 7.2 g/l h and 4.9 g/l h for subsequent, respective cycles. Like the centrifugation process, the anaerobic phase was prolonged to about 34 h. Prolonging this phase increased the overall productivity to 4.2 g/l h similar to the centrifugation and was 20% greater than the fed-batch process without any recycling (3.5 g/l h, Zhu et al. 2007). Again, less than 0.5 g succinate/l and no ethanol were observed. Compared to centrifugation/resuspension, ultrafiltration minimized potential contamination, and this continuous process permitted good control of pH and temperature. These reasons may explain the greater specific lactate productivity (above 0.9 g/g h) for a longer time compared to centrifugation/resuspension.

In summary, by-products succinate and ethanol can be minimized by monitoring the DO during aerobic growth which is an indirect indicator of acetate depletion. By prolonging the anaerobic phase by 50% from 22 to 34 h, we were consistently able to attain a lactate productivity of 4.2 g/l h, a productivity 20% greater than the fed-batch process without recycling. Since no acetate was available in the fresh medium, less than 0.5 g succinate/l and no ethanol were generated.