Reconstruction of metabolic pathway for isobutanol production in Escherichia coli

Background The microbial production of useful fuels and chemicals has been widely studied. In several cases, glucose is used as the raw material, and almost all microbes adopt the Embden–Meyerhof (EM) pathway to degrade glucose into compounds of interest. Recently, the Entner–Doudoroff (ED) pathway has been gaining attention as an alternative strategy for microbial production. Results In the present study, we attempted to apply the ED pathway for isobutanol production in Escherichia coli because of the complete redox balance involved. First, we generated ED pathway-dependent isobutanol-producing E. coli. Thereafter, the inactivation of the genes concerning organic acids as the byproducts was performed to improve the carbon flux to isobutanol from glucose. Finally, the expression of the genes concerning the ED pathway was modified. Conclusions The optimized isobutanol-producing E. coli produced 15.0 g/L of isobutanol as the final titer, and the yield from glucose was 0.37 g/g (g-glucose/g-isobutanol). Electronic supplementary material The online version of this article (10.1186/s12934-019-1171-4) contains supplementary material, which is available to authorized users.


Background
The microbial production of valuable compounds, such as chemicals, fuels, and drugs, has been widely researched for the last few decades. Using genetically and metabolically engineered microorganisms as the host strain, these useful compounds were produced mainly from glucose as the raw material [1][2][3][4][5][6]. In nature, almost all microbes follow the Embden-Meyerhof (EM) pathway as the central metabolism system to acquire energy from carbohydrates such as glucose. In the EM pathway, 1 mol of glucose is metabolized to 2 mol of pyruvate via 10-step reactions. In the process, particularly bacteria can acquire 2 mol of adenosine triphosphate (ATP) and 2 mol of reduced nicotinamide adenine dinucleotide (NADH). E. coli and other bacteria follow an alternative metabolic pathway called as the Entner-Doudoroff (ED) pathway. In this pathway, similar to the EM pathway, 1 mol of glucose is metabolized to 2 mol of pyruvate; however, this pathway involved only 4 enzymes. Using the ED pathway, bacteria can acquire 1 mol of ATP, NADH, and reduced NADH phosphate (NADPH) [7].
Some reports concerning microbial production of valuable chemicals using the ED pathway are available. Ng et al. have introduced a synthetic ED pathway from Zymomonas mobilis into E. coli to produce terpenoid [8]. Zhao et al. have reported ethanol production from lignocellulosic biomass using Z. mobilis as the host strain, which originally adopted ED pathway to acquire energy from carbohydrates [9]. In various microorganisms, researches regarding applying the ED pathway for bioproduction have been widely undertaken [10][11][12][13].
Isobutanol, one of the most attractive biofuels, has been widely studied for industrial production. A synthetic pathway of isobutanol has previously been constructed, and the suitable genes were isolated [14,15]. This synthetic pathway initiates from the aldol condensation of 2 mol of pyruvate to 2-acetolactate by acetolactate synthase (AlsS) from Bacillus subtilis. In addition, the overexpression of the enzymes ketol-acid reductoisomerase (IlvC) and dihydroxy-acid dehydratase (IlvD) from E. coli and 2-ketoisovalerate decarboxylase (Kivd) and alcohol

Open Access
Microbial Cell Factories *Correspondence: tomokazu.shirai@riken.jp 1 Center for Sustainable Resource Science, RIKEN, 1-7-22, Suehiro-cho, Tsurumi-ku, Yokohama, Kanagawa 230-0045, Japan Full list of author information is available at the end of the article dehydrogenase (AdhA) from Lactococcus lactis contribute to the synthesis of isobutanol [14]. In this synthetic pathway, 1 mol of NADH and NADPH were consumed (Fig. 1). In the ED pathway of E. coli, 1 mol of NADH and NADPH were produced from 1 mol of glucose, as shown above. Therefore, the combination of ED pathway and the isobutanol synthetic pathway is expected to be one of the powerful strategies to achieve high isobutanol production because of the complete redox balance (Fig. 1). Although isobutanol has reportedly been produced using E. coli carrying a synthetic ED pathway of Z. mobilis [16], the EM pathway and pentose phosphate pathway (PPP) has remained in the metabolic pathway of E. coli used in the report; this indicates that the metabolic pathway for isobutanol production involving the ED pathway should be improved from the perspective of redox balance.
In the present study, we performed isobutanol production using genetically modified E. coli available for the ED pathway as the host strain. For this purpose, we generated E. coli strain strictly available for the ED pathway by inactivating the genes encoding glucose-6-phosphate isomerase (Pgi) and 6-phosphogluconate dehydrogenase (Gnd) passing to the EM pathway and PPP, respectively. Thereafter, the pathways to produce byproducts, such as organic acids, were interrupted from the metabolic pathway of the E. coli strain available for the ED pathway. Finally, the ED pathway was modified by overexpressing the genes concerning the endogenous ED pathway in E. coli. Using the E. coli transformant thus generated, we successfully demonstrated isobutanol production. The redox balance in the metabolic pathway is one of the most important factors, and the use of ED pathway would be a promising tool for microbial production.

Creation of E. coli strain available for the ED pathway
To create isobutanol-producing E. coli with the complete redox balance, we attempted to modify the metabolic pathway of E. coli K-12 MG1655 (called as CFTi2 in the present study). First, we inactivated pgi to prevent the carbon flux to the EM pathway (CFTi3). To activate the ED pathway, gntR-encoding DNA-binding transcriptional repressor was interrupted. Although the CFTi4 strain thus generated was available for the ED pathway, the carbon flux to PPP pathway had already remained and the redox balance was unsuitable for isobutanol production. To overcome this difficulty, gnd-encoding 6-phosphogluconate dehydrogenase was removed from CFTi4 strain and the completely ED pathway-dependent E. coli was constructed (CFTi5) (Fig. 1). Here, trace experiments with [1-13 C]glucose were performed to verify that the constructed strain metabolized glucose using the ED pathway (Additional file 1: Fig. S1). By analyzing mass isotopomer distributions of TBDMS-derivatized Ala, it was confirmed (Additional file 1: Table S2). The strains constructed in the present study are shown in Table 1.

Inactivation of the pathway to produce byproducts in E. coli
Furthermore, to prevent volatilization of isobutanol, test tubes used for cultivation were screw-capped after 18 h of cultivation. In this case, cultures were assumed to be relatively under anaerobic condition after 18 h of cultivation. In E. coli culture, under anaerobic condition, yielding organic acids, such as lactate, was possible. To further encourage isobutanol production, we attempted to disrupt the carbon flux to some organic aids as byproducts. Some organic acids, such as acetate, lactate, and formate are produced via pyruvate and acetyl coenzyme A (acetyl-CoA) in E. coli. Using CFTi5 as the base strain, pflB, ldhA, and pta (encoding pyruvate formate lyase, d-lactate dehydrogenase, and phosphate acetyltransferase, respectively) were sequentially inactivated (Fig. 1).
Using the modified isobutanol-producing E. coli, we performed isobutanol production. Figure 3 shows the CFTi61, CFTi91, and CFTi101 culture profiles. The maximal levels of cell growth, glucose consumption, and isobutanol production are shown in CFTi91 culture. Although inactivation of ldhA was effective in repressing lactate formation and improving isobutanol production, this production was decreased by removing the gene concerning acetate production ( Fig. 3; Additional file 1: Fig. S3). The titer of isobutanol, yield from glucose, and rate of glucose uptake are summarized in Additional file 1: Table S3. The yield of isobutanol from glucose in CFTi91 between 18 and 96 h of cultivation was 0.28 g/g.

Modification of ED pathway in isobutanol-producing E. coli
To further improve isobutanol production, we attempted to modify the ED pathway. Using a low copy number vector, the gene sets concerning the endogenous ED pathway in E. coli were introduced into isobutanol-producing E. coli transformant CFTi91 generated in the present study. Figure 4 shows CFTi91co, CFTi91zp, and CFTi91zpee culture profiles. The highest amount of isobutanol produced was 15.0 g/L in CFTi91zpee culture, which carries the gene set of zwf, pgl, edd, and eda encoding glucose-6-phosphate dehydrogenase, 6-phosphogluconolactonase, phosphogluconate dehydratase, and KHG/KDPG aldolase, respectively. Although the maximal level of cell growth and the initial rates of glucose uptake and isobutanol production until 18-h cultivation in CFTi91zpee were lower than those in CFTi91co and CFTi91zp, this production was greater in CFTi91zpee than in the others after 18-h cultivation (Fig. 4). Further, we estimated the amount of organic acids produced as byproducts. Among the CFTi91-derivative strains used, acetate and succinate production were primarily confirmed. However, their total amount was < 1.0 g/L in each strain (Additional file 1: Fig. S4). The titer of isobutanol, yield from glucose, and rate of glucose uptake are summarized in Table 2. The yield of isobutanol from glucose in CFTi91zpee between 18 and 96 h cultivation was 0.37 g/g, and the final isobutanol titer was 15.0 g/L.

Discussion
Recently, microbial production using various microbes has been widely studied toward industrial application. The tools of metabolic engineering and synthetic biology could be combined to design and create our desired microorganisms [17][18][19]. The metabolic pathway comprises several enzymatic reactions, particularly redox reactions. Hence, it is important to design the metabolic pathway toward the compounds of interest by considering the intracellular redox balance. We adopted the ED pathway to produce isobutanol using E. coli. In the ED pathway, each 1 mol of NADH and NADPH is produced from 1 mol of glucose, whereas each 1 mol of NADH and NADPH is consumed to produce 1 mol of isobutanol ( Fig. 1) [7]. In the present study, we generated the ED pathway-dependent E. coli transformants and performed isobutanol production using the derivative strains. Several studies regarding isobutanol production using various microbes have been reported. Atsumi et al. have reported isobutanol production using E. coli for the first time [15]. Moreover, Corynebacterium glutamicum, B. subtilis, and Saccharomyces cerevisiae were used for isobutanol production as the host strains; however, the maximal titer and yield levels have been obtained using E. coli [20][21][22]. In these reports, the inactivation of genes concerning byproducts formation, direct evolution of the enzymes in the biosynthetic pathway of isobutanol, and optimization of culture conditions were performed as the general strategies to improve isobutanol production. A study on isobutanol production has demonstrated the use of partially ED pathway-dependent E. coli, which adopted the ED pathway derived from Z. mobilis [16]. However, the strain used in that study was simply introduced to the genes  concerning the ED pathway, and the redox balance in the biosynthesis pathway involved in converting glucose to isobutanol was not completely considered. The yield of isobutanol from glucose (0.31 g/g) was lower than that in the case using the EM pathway and has the potential for further improvements [16,20].
In the present study, we constructed the ED pathway-dependent E. coli transformants by inactivating the negative repressor of the ED pathway and the genes concerning the initial reaction of EM and PPP pathway (Fig. 1). First, we inactivated pgi to repress the carbon flux to the EM pathway. The rates of glucose uptake and isobutanol production in CFTi31 were decreased, and acetate production as the byproduct was lower than that of CFTi21, which is completely dependent on the EM pathway (Additional file 1: Fig. S2). CFTi21 produced isobutanol mainly in growth phase, whereas did not in stationary phase (Fig. 2). This may be because CFTi21 produced a large amount of acetate along with isobutanol and decrease of pH caused lysis of bacterial cell. Although gntR was disrupted after removing pgi to activate the ED pathway, the amount of isobutanol production in CFTi41 was drastically decreased compared with that in CFTi21 or CFTi31 (Fig. 2). This could be attributed to the cell growth of CFTi41, which was greater than that of other strains used in this section; moreover, the PPP pathway and TCA cycle, which are important for cell growth may be activated in CFTi41, whose pgi and gntR were inactivated (Fig. 2). Finally, gnd catalyzing the third reaction of the PPP pathway was inactivated. Although the cell growth of CFTi51 was decreased compared with that of CFTi41, the isobutanol production was almost threefold higher (Fig. 2). This could be attributed to the repression of carbon leak to the PPP pathway and stimulation of the carbon flux to isobutanol from glucose through the ED pathway. The maximal titer and yield of isobutanol were recorded in the ED pathway-dependent CFTi51.
Thereafter, we changed the culture condition to prevent volatilization of isobutanol. Although culturing was initiated in aerobic condition, test tubes used here were screw-capped after 18 h of cultivation. To prevent the formation of organic acids associated with the change in culture condition, we inactivated genes concerning organic acid formation. Although the disruptions of pflB and ldhA concerning formate and lactate formation, respectively, were effective to improve isobutanol production, the inactivation of pta gene resulted in decreasing isobutanol production (Fig. 3). However, the amount of acetate produced in CFTi91 was only 250 mg/L, and the carbon from glucose would be used for isobutanol production (Additional file 1: Fig. S3). Here, we speculated the reason why the deletion of pta gene had a negative effect to isobutanol production, despite of lacking anaerobic source of acetyl-CoA which is a substrate of Pta. As shown in Additional file 1: Fig. S3A, the isobutanol-producing strain whose pta gene was deleted drastically decreased acetate production. We assumed that a small amount of produced acetate until 24 h cultivation (aerobic or micro-aerobic condition) was one of key factor in the isobutanol production using our created strain. The energy balance optimized by deleting both pflB and ldhA genes might be disturbed by omitting the gene concerning acetate production.
To further improve isobutanol production, we attempted to modify the endogenous ED pathway in E. coli. Although the initial isobutanol production and the maximal cell growth were decreased by introducing 4 genes constructing the ED pathway, the final titer of isobutanol and yield from glucose were encouraged ( Fig. 4; Table 2). In the optimized strain CFTi91zpee as well as in CFTi91, acetate production was repressed, whereas succinate production was relatively increased (Additional file 1: Fig. S4). In the present study, the maximal yield of isobutanol from glucose reached 0.37 g/g, which was B. subtilis ΔldhA (EM pathway) 2.62 N.E. [24] higher than that in a previous report using partially ED pathway-dependent strain for isobutanol production and 90% of the maximal theoretical yield (0.41 g/g) [16]. In addition, it would be comparable with other reports on isobutanol production using various host strains (Table 3) [20]. These results indicate that the redox-balanced metabolic design is one of the significant factors in microbial production. Although we reconstructed metabolic pathway suitable for isobutanol production from the aspect of the redox balance, E. coli is known to have the endogenous transhydrogenase encoded in pntAB concerning the conversion of NADH into NADPH which is a candidate reaction to acquire NADPH [21]. In the previous report about isobutanol production, the coexpression of PntAB and NAD kinase contributed to encouraging isobutanol production [22]. This result would mean that those two enzymes needed to be overexpressed to earn the enough activity of converting NADH to NADPH, which is one of essential cofactors for isobutanol biosynthetic pathway.
In the experiment keeping under anaerobic condition after 18 h cultivation, succinate production was confirmed (Additional file 1: Figs. S3C and S4B). the isobutanol-producing strain used in the study lacked the intracellular CO 2 sources the gene pflB, oxidative TCA cycle, and the oxidative pentose phosphate pathway, however; 2 mol of CO 2 was produced in the exogenous biosynthetic pathway of isobutanol (Fig. 1). Succinate production would be attributed to the activation of phosphoenolpyruvate (PEP) carboxylase which produces oxaloacetate from PEP and CO 2 . The produced oxaloacetate might be converted into succinate via the reductive TCA cycle to prevent overflowing the intracellular excess energy.
In conclusion, for effective isobutanol production, we generated ED pathway-dependent E. coli strain, which produced 11.8 g/L of isobutanol, and the yield from glucose was 0.24 g/g under aerobic condition. By inactivating genes concerning organic acids and modifying the endogenous ED pathway, isobutanol production was increased, and the titer and yield from glucose were 15.0 g/L and 0.37 g/g, respectively. The optimization of culture conditions and the collection of volatilized isobutanol are expected to further improve isobutanol production.

Strains and plasmid construction
The strains and plasmids used in the present study are listed in Table 1 and Additional file 1: Table S1. E. coli NovaBlue competent cells (Novagen, Cambridge, MA, USA) were used for gene cloning. Polymerase chain reaction (PCR) was performed using the KOD FX Neo DNA polymerase (Toyobo, Osaka, Japan), and the primer pairs are listed in Additional file 1: Table S1. Each gene was assembled with the respective plasmid using Gibson Assembly (New England Biolabs, Ipswich, MA, USA). The construction of plasmids used in this study is described in Table 1. The plasmids were transformed into bacterial strains using the Gene Pulser II (Bio-Rad, Hercules, CA, USA). Wherever applicable, 100 µg/mL ampicillin, 50 µg/mL kanamycin, and 15 µg/mL chloramphenicol were added to the media for selection purpose. All the transformants constructed in this study are listed in Table 1.
pZA23trc was constructed as follows: The nucleotide sequence of trc promoter and pZA23 without the promoter region were amplified by PCR using pTrcHis B and pZA23MCS as the templates with the primer pairs inv_pZA23_no_prom_f and inv_pZA23_no_prom_r. The amplified fragments were fused with each other, and the resulting plasmid was designated as pZA23trc. pZka and pTka were constructed as follows: the synthetic gene fragments of kivd and adhA were directly cloned into the KpnI and HindIII sites of pZE12MCS, respectively, and the resulting plasmid was designated as pZka. Subsequently, the gene set of kivd and adhA and the plasmid fragment of pTrcHis B were amplified by PCR using pZka and pTrcHis B as the template with the primer pair kivd_ptrc_f and adhA_ptrc_r and inv_ptrc_f and inv_ptrc_r, respectively. The amplified fragments were fused with each other, and the resulting plasmid was designated as pTka. p23S and p23SCD were constructed as follows: alsS (NP_391482.2) and pZA23trc fragment were amplified by PCR using B. subtilis genomic DNA and pZA23trc as the template with the primer pairs alsS_ptrc_f and alsS_ptrc_r and inv_ptrc_f and inv_ptrc_r, respectively. The amplified fragments were fused together, and the resulting plasmid was designated as p23S. Thereafter, the gene set of ilvC (NP_418222.1) and ilvD (YP_026248.1) was amplified by PCR using E. coli K-12 MG1655 genomic DNA as the template and the primer pair ilvCD_ptrc_sacI _f and ilvCD_ptrc_sacI_r. The amplified fragment was cloned into the SacI site of p23S, and the resulting plasmid was designated as p23SCD. pSp, pSzp, and pSzpee were constructed as follows: pgl (NP_415288.1) and pSAK fragment were amplified by PCR using E. coli K-12 MG1655 genomic DNA and pSAK as the template and the primer pairs pgl_pZ_ hindIII_f and pgl_pZ_hindIII_r and inv_pZ_hindIII_f and inv_pZ_hindIII_r, respectively. The amplified fragments were fused together, and the resulting plasmid was designated as pSp. zwf (NP_416366.1) was amplified by PCR using E. coli K-12 MG1655 genomic DNA as the template with the primer pair zwf_pZ_kpnI _f and zwf_pZ_kpnI _r. The amplified fragment was cloned into the KpnI site of pSp, and the resulting plasmid was designated as pSzp. edd (NP_416365.1) and eda (NP_416364.1) were amplified by PCR using E. coli K-12 MG1655 genomic DNA as the template with the primer pairs edd_pZ_bamHI _f and edd _r and eda_f and eda_pZ_bamHI_r, respectively. The gene set of edd and eda were amplified by PCR using these amplified fragments of edd and eda as the templates with the primer pair i edd_pZ_bamHI _f and eda_ pZ_bamHI_r. The amplified fragment was cloned into the BamHI site of pSzp, and the resulting plasmid was designated as pSzpee. Plasmid maps used in this study are shown in Additional file 1: Fig. S5.

Deletion of chromosomal genes
The primers used for the deletion of chromosomal genes are listed in Additional file 1: Table S1. E. coli K-12 MG1655 (ATCC700926) was used as the parent strain. Using the Quick and Easy E. coli Gene Deletion Kit (Funakoshi, Tokyo, Japan), each gene was deleted according to the manufacturer's protocol. The gene fragments for deletion and recombination were introduced into each strain using the Gene Pulser II. Other fragments used to inactivate the respective genes were amplified by PCR using the FRT-PGK-gp2-neo-FRT template DNA from the Quick and Easy E. coli Gene Deletion Kit with the appropriate primers, as listed in Additional file 1: Table S1.