High-flux isobutanol production using engineered Escherichia coli: a bioreactor study with in situ product removal
- 3.6k Downloads
Promising approaches to produce higher alcohols, e.g., isobutanol, using Escherichia coli have been developed with successful results. Here, we translated the isobutanol process from shake flasks to a 1-L bioreactor in order to characterize three E. coli strains. With in situ isobutanol removal from the bioreactor using gas stripping, the engineered E. coli strain (JCL260) produced more than 50 g/L in 72 h. In addition, the isobutanol production by the parental strain (JCL16) and the high isobutanol-tolerant mutant (SA481) were compared with JCL260. Interestingly, we found that the isobutanol-tolerant strain in fact produced worse than either JCL16 or JCL260. This result suggests that in situ product removal can properly overcome isobutanol toxicity in E. coli cultures. The isobutanol productivity was approximately twofold and the titer was 9% higher than n-butanol produced by Clostridium in a similar integrated system.
KeywordsBiofuels Isobutanol E. coli Gas stripping Bioreactor
The growing need for alternative energy calls for new strategies to produce liquid fuels such as alcohols, alkanes, and fatty acid esters (or biodiesel) from renewable sources (Dellomonaco et al. 2010; Peralta-Yahya and Keasling 2010; Solomon 2010). Compared to ethanol, higher alcohols (e.g., 1-propanol, isobutanol, n-butanol, 2-methyl-1-butanol, 3-methyl-1-butanol) have higher energy density and lower hygroscopicity, which make them better candidates as gasoline additives or substitutes. However, no native organisms have been identified to produce these advanced biofuels in appreciable quantities (Clomburg and Gonzalez 2010). For example, isobutanol was identified as a by-product in beer fermentation, but with a titer as low as 16 mg/L (Garcia et al. 1994). Since the feasibility of a bioprocess depends critically on the final product titer (Stephanopoulos 2007), a synthetic approach to produce higher alcohols from non-fermentative pathways in Escherichia coli was previously devised (Atsumi et al. 2008b). In addition, metabolic engineering efforts to produce these alcohols at higher titers and yields have been performed (Atsumi et al. 2008a, b; Cann and Liao 2008; Clomburg and Gonzalez 2010; Connor and Liao 2008; Jarboe et al. 2010; Shen and Liao 2008; Smith et al. 2010). This strategy uses the amino acid biosynthetic pathway and diverts its 2-ketoisovalerate for isobutanol synthesis by overexpression of 2-ketoisovalerate decarboxylase (Kivd) and alcohol dehydrogenase (AdhA). Because of the ubiquity of the amino acid pathways, this strategy is compatible with many organisms and shows significant promise for industrial applications.
Strains and plasmids used in this study
BW25113/F′[traD36, proAB+, lacIq ZΔM15]
Atsumi et al. (2008a)
Same as JCL16 but with ΔadhE, ΔfrdBC, Δfnr, ΔldhA, Δpta, ΔpflB
Atsumi et al. (2008a)
Isobutanol-tolerant strain, JCL260 derivative
Atsumi et al. (2010b)
Same as JCL260 but with ΔpoxB
ColE1 ori; AmpR; P L lacO1: kivd-adhA (Lactococcus lactis)
Atsumi et al. (2010a)
P15 ori; KanR; P L lacO1: alsS (Bacillus subtilis)-ilvCD
Atsumi et al. (2008b)
Although this titer is 85% and 12% higher than those reported by Ezeji et al. (2003) and Qureshi and Blaschek (1999), respectively, for n-butanol produced by Clostridium in batch bioreactors, industrial-scale fermentation processes require higher final titers to be economically feasible. The final product titer can be limited by various factors depending on the particular production strain and bioprocess. Since isobutanol is toxic to the cell, isobutanol production from glucose may be limited by the toxicity of the final product itself. In this sense, improving the tolerance of the biocatalyst becomes a primary necessity to achieve a process with high product titers (Alper et al. 2006; Gonzalez et al. 2003; Jarboe et al. 2010; Miller and Ingram 2007; Yomano et al. 1998). It has been reported that E. coli was unable to grow at isobutanol concentrations >8 g/L (Brynildsen and Liao 2009). Even though the E. coli strain continued to produce isobutanol after it ceased to grow (Atsumi et al. 2008b), product toxicity may eventually damage the cell and limit the final titer (Rutherford et al. 2010). In this regard, Atsumi et al. (2010b) isolated and characterized an isobutanol-tolerant E. coli strain (SA481) able to grow at 8 g/L. SA481 was evolved from the high isobutanol producer strain (JCL260) by a sequential transfer method increasing isobutanol concentration every 15 transfers. The SA481 strain showed superior growth characteristics compared to the high producer (JCL260) strain when they both were cultivated at 6 and 8 g/L of isobutanol. Compared with the control condition, specific growth rate for JCL260 decreased by 35% and 74% at 6 and 8 g/L of isobutanol (calculated from data showed by Atsumi et al. 2010b), respectively, while SA481 growth was not affected at 6 g/L of isobutanol and decreased just 13% at 8 g/L. Interestingly, Atsumi et al. (2010b) also showed that SA481 did not produce more isobutanol than JCL260 under normal conditions. However, the experiments were performed in shake flasks. Accordingly, the performance of SA481 harboring pSA65/pSA69 for isobutanol production in bioreactors will be evaluated in this work.
The other approach to solve the cytotoxicity problem and its detrimental effect on the final titer from a bioprocess perspective is the in situ product removal. Several in situ solvent recovery strategies have been developed for n-butanol fermentation (Groot et al. 1992; Nielsen and Prather 2009; Roffler et al. 1987; Qureshi et al. 2005). Since gas stripping is a simple and efficient way to recover the solvent from the fermentation broth (Inokuma et al. 2010; Lee et al. 2008), in the present work, gas stripping integrated with fermentation was used to compare the isobutanol production by the high producer (JCL260), isobutanol-tolerant (SA481), and the parental (JCL16) strains carrying pSA65 and pSA69 plasmids. We hypothesize that if isobutanol could be removed from the fermentation broth, the production can be extended significantly and the final titer will surpass the 22 g/L obtained previously in flask cultures.
Materials and methods
Strains, plasmids, and pre-culture
Strains and plasmids used are shown in Table 1. A single colony of freshly transformed cells was inoculated into a 250-mL screw-cap flask containing 25 mL of Luria-Bertani media with antibiotics (0.05 g/L kanamycin and 0.1 g/L ampicillin). Overnight pre-culture was performed at 37°C on a rotary shaker (250 rpm).
Culture media for bioreactor fermentations
Culture media for bioreactor cultures contained the following composition, in grams per liter: glucose, 55; (NH4)2SO4, 3; K2HPO4, 14.6; KH2PO4, 4; sodium citrate, 2.2; yeast extract, 25; MgSO4⋅7H2O, 1.25; ampicillin, 0.1; kanamycin, 0.05; and 1 mL/L of trace metal solution. Trace metal solution contained, in grams per liter: EDTA, 14.1; CoCl2⋅6H2O, 2.5; MnCl2⋅4H2O, 15; CuCl2.2H2O, 1.5; H3BO3, 3; Na2MoO4⋅2H2O, 2.1; Zn(CH3COO)2⋅2H2O, 33.8; Fe(III)citrate 100.8. The feeding solution contained, in grams per liter: glucose, 500; MgSO4⋅7H2O, 1.25; ampicillin, 0.1; kanamycin, 0.05, 0.1 mM of isopropyl-B-d-thio-galactoside (IPTG); and 1 mL/L of trace elements.
Bioreactor culture conditions
Cell growth was followed by optical density measurements at 600 nm and converted to dry cell weight concentration by calibration plots of samples dried to constant weight in an oven at 80°C. Glucose concentration was determined with YSI biochemical analyzer (YSI Instruments, Yellow Springs, OH, USA). Isobutanol concentration was quantified by a gas chromatograph (GC) model 6850 (Agilent Technologies, Santa Clara, CA, USA) equipped with flame ionization detector 6850 series (Agilent Technologies). A 30-m, 0.32-mm i.d., 0.25-μm DB-FFAP capillary column was used. GC oven temperature was initially held at 85°C for 3 min and raised with a gradient 45°C/min until 225°C and held for 3 min. Helium at 1.7 mL/min was used as the carrier gas with 9.52-psi inlet pressure. A volume of 1 μL supernatant of the culture broth was injected using 1-pentanol as internal standard. Acetate concentration was determined by high-pressure liquid chromatography (Agilent Technologies) with an Aminex HPX-87H column (Bio-Rad Laboratories, Hercules, CA, USA) maintained at 35°C. A mobile phase of 5 mM H2SO4 was used at 0.6 mL/min.
Preparation of cell-free extracts and enzyme activity assays
Cells were harvested by centrifugation at 10,000×g for 10 min at 4°C from bioreactor cultures, washed twice with 50 mM potassium phosphate buffer (pH 7.0), and stored frozen at −80°C. Frozen cells were suspended in the same washing buffer, blended with glass beads, and disrupted in series 4 × 1 min. Crude extract was centrifuged at 10,000×g at 4°C for 30 min and the supernatant used for enzyme assays. The concentration of the protein preparation was determined by the Bradford method. Acetolactate synthase (AlsS) and dihydroxy-acid dehydratase (IlvD) assays were performed as described by Atsumi et al. (2009). KivD activity was assayed by measuring NADPH oxidation at 340 nm in 50 mM phosphate buffer (pH 7.0), 1 mM MgCl2, 1.5 mM TTP, 0.1 μM alcohol dehydrogenase (ADH6) NADPH-dependent, 0.2 mM NADPH, and 30 mM KIV. Since NADPH oxidation was proportional to isobutiraldehyde production, it was possible to determine KivD activity. Alcohol dehydrogenase activity was detected by measuring NADH oxidation at 340 nm in 100 mM MOPS, pH 7.0, 0.2 mM NADH, and 25 mM isobutaraldehyde.
Comparison of high isobutanol producer (JCL260), high isobutanol-tolerant (SA481), and parental (JCL16) strains at 30°C
Isobutanol production continued even after cell growth was stopped (Fig. 2a, c). In fact, around 80% of total isobutanol accumulated by JCL260 cultivated at 30°C was produced after the growth phase. Such behavior was similar to that reported by Atsumi et al. (2008b), but 2.3-fold higher. The high isobutanol producer strain (JCL260) accumulated 50.8 ± 1.1 g/L of isobutanol in 72 h and the parental (JCL16) 41.4 ± 4.5 g/L in the same time (Fig. 2a). Surprisingly, the high isobutanol-tolerant strain (SA481) produced only 23 ± 4.4 g/L of total isobutanol in 68 h. After 22 h, the maximum isobutanol concentration in the fermentation broth for JCL260 and JCL16 was around 11 g/L and for SA481 around 6.5 g/L (Fig. 2b). The maximum cell concentrations for JCL16, SA481, and JCL260 were 8.4 ± 0.3, 7.6 ± 0.7, and 6.7 ± 0.16 g/L, respectively (Fig. 2c).
Time profile of high isobutanol producer strain (JCL260) at 30°C and 37°C
Kinetic and stoichiometric comparison of the strains
Batch fermentations were integrated with gas stripping to evaluate the isobutanol production of the high-producing (JCL260), high isobutanol-tolerant (SA481), and parental (JCL16) strains. For the first time, we were able to produce more than 50 g/L of isobutanol in 72 h. We found that with in situ product removal, E. coli was able to surpass the 22 g/L produced in flask culture (Atsumi et al. 2008b).
Although SA481 has shown superior isobutanol tolerance to JCL260 (Atsumi et al. 2010b), it was not reflected in the final titer of isobutanol obtained in the integrated process. These results suggest that under these culture conditions, isobutanol production was not limited by growth or the product toxicity during growth phase. Atsumi et al. (2010b) found that isobutanol production, cell growth, and glucose consumption rate of SA481 tested in flask cultures were similar to those of JCL260 and that improved tolerance of SA481 did not increase the final titer. In this study, glucose consumption of SA481 was lower than that of JCL260; consequently, isobutanol production rate was also lower (Figs. 2a and 3). Cell growth of SA481 stopped around 22 h, reaching practically the same cellular concentration as JCL260. At this time, isobutanol concentration in the broth was around 7 g/L and cells did not grow even though there was no apparent nutrient limitation.
On the other hand, temperature should be an important factor in isobutanol production since the final titer at 37°C was 53% lower than at 30°C. The observed difference might be caused by a lower Adh activity at 37°C toward the end of the culture (Fig. 6e) In addition, it has been reported that a shift-up in growth temperature decreases the proportion of unsaturated fatty acids in the membrane (Balamurugan 2010; Casadei et al. 2002; Ingram and Buttke 1984). Such changes in membrane composition may determine the tolerance level of cells to a particular stress. For example, Ingram and Buttke (1984) observed that alcohol tolerance of Zymomonas mobilis reduces with increasing growth temperatures. If such behavior will be the same for E. coli, poor isobutanol production at 37°C could be justified by a reduced alcohol tolerance compared with 30°C.
Brynildsen and Liao (2009) reported that 8 g/L (1%) of isobutanol caused a growth arrest in E. coli, but even a lower concentration of solvent (0.8%, v/v, or 6.4 g/L of n-butanol) caused a 50% growth decrease, disruption in respiratory efficiency, increased permeability of the cell wall, and oxidative stress response on E. coli (Rutherford et al. 2010). In this work, cell growth of JCL260 and JCL16 stopped around 10 h (when isobutanol concentration in the broth was roughly 7 g/L) and cells did not grow even after feeding. Such a phenomenon might be attributed to solvent toxicity because isobutanol concentration at 10 h was similar to those tested by Brynildsen and Liao (2009) and Atsumi et al. (2010b).
Even though the present work is not a complete optimization study, the results show that a very simple strategy (gas stripping) significantly improved the effective titer of isobutanol. Gas stripping avoids energy-intensive distillation and appears to be an effective method for product separation.
This work was supported in part by UC MEXUS-CONACYT Postdoctoral fellowship program under the guidelines of the 2009–2010 UC MEXUS-CONACYT Call (I0101/120/07 MOD. ORD.-50-07).
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.
- Chang D, Shin S, Rhee JS, Pan JG (1999) Acetate metabolism in a pta mutant of Escherichia coli W3110: importance of maintaining acetyl coenzyme A flux for growth and survival. J Bacteriol 181:6656–6663Google Scholar
- Farmer WR, Liao JC (1997) Reduction aerobic acetate production Escherichia coli. Appl Environ Microbiol 63:3205–3210Google Scholar