Introduction

Succinic acid, a C4-dicarboxylic acid, is a precursor of many important chemicals in the food, chemical, and pharmaceutical industries (Ahn et al. 2016; Olajuyin et al. 2016; Wang et al. 2014). Certainly, the widespread importance of succinate has secured its listing among the top 12 chemical building blocks by the US Department of Energy (Werpy and Petersen 2004). To improve succinate production by using Escherichia coli, researchers have explored various metabolic engineering strategies (Ahn et al. 2016; Forster and Gescher 2014; Kang et al. 2010; Liang et al. 2015) such as activating the glycolytic pathway, overexpressing pyruvate-metabolizing enzymes, eliminating competing pathways, and providing reducing equivalents and energy (Li et al. 2016; Liang and Qi 2014; Meng et al. 2016; Vemuri et al. 2002; Vuoristo et al. 2016).

We previously constructed succinate-producing strains by using a series of genetic modifications, and a high overall volumetric productivity and concentration of succinate were achieved (Li et al. 2013). A whole-phase succinate production strategy was developed in this study. In this method, the engineered strain produced succinate under aerobic, microaerobic, and anaerobic conditions. Compared with this whole-phase fermentation, strain growth uncouples product formation in the typical dual-phase fermentation. However, the overall yield from our constructed engineered strain should be increased further. Recently, we explored the relationships between ATP generation, substrate ratio of xylose to glucose, and succinate production (Zhang et al. 2016). We observed that excess energy exerted a negative effect on succinate production in the engineered E. coli that could not grow anaerobically. Therefore, regulating ATP generation for improving the overall yield of succinate by using glucose as the carbon source in this engineered E. coli is essential.

The NADH/NAD+ ratio is also crucial for succinate production (Singh et al. 2009). Under aerobic growth conditions, NAD+ is regenerated through oxidative phosphorylation. Under anaerobic conditions, NAD+ serves as an electron acceptor during substrate degradation, and NADH provides the reducing power for reductive product formation. A balance of NAD+ reduction and NADH oxidation is a prerequisite for anaerobic fermentation (de Graef et al. 1999). In the engineered E. coli, NADH-dependent lactate dehydrogenase and ethanol dehydrogenase were inactivated. Inactivation limits the means by which NAD+ may be regenerated from NADH formed from aerobic to anaerobic growth conditions, thus affecting the redox balance to increase the NADH/NAD+ ratio (Fig. 1a; Li et al. 2013). Therefore, we assumed that the low yield of succinate production was because of the limited NAD+ regeneration rate, which resulted in the increase in the NADH/NAD+ ratio from the aerobic to anaerobic phase; however, NAD+ could be regenerated in the reductive tricarboxylic acid (TCA) cycle.

Fig. 1
figure 1

Central carbon metabolism and electron transport chain in our engineered E. coli. a Gene deletions in strain YL104 are represented by black stars, and gene overexpression is depicted in red. ATP futile cycling reactions are denoted by purple circles. PEP phosphoenolpyruvate. b Red and blue lines represent the aerobic respiratory chain and anaerobic respiratory chain, respectively. Black stars represent the gene mutation in YL104. FRDS1 is soluble fumarate reductase from S. cerevisiae (color figure online)

Full size image

Another possible cause of the increased NADH/NAD+ ratio at the initial anaerobic condition is the change in NADH dehydrogenases during the aerobic–anaerobic fermentation process. The aerobic respiratory chain of E. coli involves three membrane-bound NADH dehydrogenases (NDH-I, NDH-II, and WrbA) and three ubiquinol oxidases (cytochromes bd-I, bd-II, and bo; Fig. 1b). These enzymes have different efficiencies in coupling electron transfer for the generation of an electrochemical proton gradient (Bekker et al. 2009; Borisov et al. 2011). Under aerobic conditions, E. coli mainly uses NADH dehydrogenases II (ndh), and under anaerobic conditions, it mainly uses NADH dehydrogenases I (nuo). The change in main NADH dehydrogenase in E. coli from the aerobic to anaerobic condition may be time-consuming, and this change also resulted in an elevated NADH/NAD+ ratio. Therefore, we speculated that using one type of NADH dehydrogenase during aerobic and anaerobic conditions could reduce the NADH/NAD+ ratio to increase the overall yield of succinate (Singh et al. 2009; Vemuri et al. 2006).

In this study, by regulating the NADH/NAD+ ratio and ATP generation through various strategies, we explored the relationships between the NADH/NAD+ ratio, ATP level, and overall yield of succinate production with glucose as the carbon source.

Materials and methods

Strains and plasmids

All strains, plasmids, and oligonucleotides used in this study are listed in Table 1 and Table S1. All genetic modifications were made in E. coli MG1655.

Table 1 Bacterial strains and plasmids used in this study
Full size table

The one-step inactivation of the chromosomal gene method was used for single gene mutants (Datsenko and Wanner 2000). The primers ndh-pKD4-F and ndh-pKD4-R and the template plasmids pKD4 for ndh knockouts were used to obtain linear DNA fragments with antibiotic-resistant gene cassettes flanked by FLP recognition target sites and 39-bp homologous arms for the construction of the single gene mutants. The purified DNA was electroporated into E. coli K-12 MG1655 cells harboring the helper plasmid pTKRed expressing Red recombinase enzymes (Kuhlman and Cox 2010). Positive clones were selected using appropriate antibiotics. Subsequently, the helper plasmid and resistance gene were eliminated.

For gene overexpression, pCL1920 was used and digested using HindIII–KpnI to obtain the plasmid skeleton and resistant gene. A DNA fragment with frd sequence (GenBank, KX505965) from Saccharomyces cerevisiae S288c was synthesized with codon optimization by Generay Biotechnology (Shanghai, China). This fragment was also digested using HindIII–KpnI, and the sequence was ligated into the vector pCL1920, thus generating the plasmid pCLfrd. Furthermore, nuoC was amplified from wild-type E. coli MG1655 by using the primers nuoC PF and nuoC PR. The vector skeleton and ppsA were amplified from the pCL1920 plasmid and E. coli MG1655 genome by using the primers pCL1920-F and pCL1920-R and trc-ppsA-F and trc-ppsA-R, respectively. The plasmid pCLnuoC was obtained through Gibson assembly with the pCL1920 skeleton and nuoC fragment. The primers trc-RBS-F and trc-RBS-R from the pTrc99a plasmid were annealed to obtain the trc promoter with RBS. Subsequently, the three fragments, plasmid skeleton, trcRBS, and ppsA, were assembled together through Gibson assembly. In addition, pCLppsA/frd was constructed using Gibson assembly with the pCLppsA skeleton and frd fragment.

Culture medium and growth conditions

Luria–Bertani (LB) medium (10 g/L tryptone, 5 g/L yeast extract, and 10 g/L NaCl; pH 7.2) was used for all DNA manipulations. During the construction of the single-gene-deficient mutants, cultures were grown aerobically at 30, 37, or 42 °C in Super Optimal Broth medium (20 g/L tryptone, 5 g/L yeast extract, 0.5 g/L NaCl, 2.5 mM KCl, and 10 mM MgCl2). All cultures were incubated at 37 °C, and the pH was measured using a glass electrode and controlled at 6.4–6.8 by using 5 M K2CO3 and 0.5 M NaOH.

For succinate production, a single clone of each single gene mutant and the parent strain were cultured in a 300-mL Erlenmeyer flask containing 50 mL of LB medium at 37 °C and 250 rpm for 12 h. Subsequently, 4 mL (5% v/v) of the overnight culture was transferred to a 500-mL Erlenmeyer flask containing 80 mL AM1 [2.63 g/L (NH4)2HPO4, 0.87 g/L NH4H2PO4, 0.15 g/L KCl, and 20.9 g/L 3-(N-morpholino)propanesulfonic acid (MOPS)] with an additional 1 g/L yeast extract and 30 g/L glucose and shaken at 250 rpm for 21 h. Antibiotics were added to provide the selective pressure during the cultivation, if necessary (ampicillin 100 μg/mL, kanamycin 25 μg/mL, spectinomycin 25 μg/mL, and chloromycetin 17 μg/mL).

During the succinate fermentation in a multiple mini-fermenter system (INFORS HT, Switzerland), the aforementioned seed culture (10% v/v) was transferred into 800 mL AM1 medium (without MOPS) supplemented with 40 g/L glucose in a 1-L fermenter for batch fermentation. The total fermentation time (70 h) included a 28-h aerobic phase (pass into O2) and a 42-h anaerobic phase (pass into CO2). The air flow rate and agitation were maintained at 1 vvm and 350 rpm, respectively. During the fermentation, 0.05 mM isopropyl-beta-d-thiogalactopyranoside (IPTG) was added at the beginning of anaerobic fermentation with strains harboring plasmids during the batch fermentation. Samples were taken at intervals of 6 or 12 h.

Analytical methods

Biomass was detected by measuring the density at 600 nm [1 OD600 ≈ 0.34 g cell dry weight (CDW) L−1] by using a spectrophotometer (Shimadzu, Japan). The organic acid concentrations and the residual glucose concentration were measured using high-performance liquid chromatography (HPLC; Shimadzu, Japan) and an Aminex HPX-87H ion exclusion column (Bio-Rad, USA). H2SO4 solution (5 mM) was applied as a mobile phase at a flow rate of 0.6 mL/min to the column at 65 °C.

The intracellular ATP concentrations were measured using the BacTiter-Glo™ Microbial Cell Viability Assay Kit (Promega, USA), and the intracellular concentrations of NADH and NAD+ were assayed using a cycling method (Leonardo et al. 1996).

Results

Use of sole anaerobic NADH dehydrogenases to increase succinate yield

First, we inactivated the NADH dehydrogenase II encoded by ndh in the previously constructed YL104 (Fig. 1a; Li et al. 2013), thus generating YL104N. In response to this modification, succinate production was similar but the overall yield of succinate in YL104N increased by 7%, reaching 1.05 mol/mol (Table 2). The biomass of YL104N exhibited no substantial change (Table 2), which suggested that the inactivation of NDH-II dehydrogenase did not have a negative effect on the growth of the engineered strains. In addition, we regulated NADH dehydrogenase I by overexpressing an important subunit (encoded by nuoCD) of this enzyme (Singh et al. 2009). However, the titer of succinate decreased, and the overall yield only slightly improved (Table 2). We speculated that excess energy was generated by overexpression of the important subunit of NDH-1 that generates proton gradients across cell membranes (Verkhovskaya et al. 2008). In addition, metabolic burdens resulting from overexpression of enzymes may influence the growth of the engineered strain, leading to decreased succinate production. The aerobic NADH/NAD+ ratio increased in YL104N compared with YL104 (Fig. 2). Moreover, the NADH/NAD+ ratio in YL104N increased less from the aerobic to anaerobic condition compared with that in YL104 (94 and 29% increases in YL104 and YL104N, respectively). These results indicated that the inactivation of main aerobic NADH dehydrogenase (ndh) in YL104 could alleviate the change in the NADH/NAD+ ratio, resulting in a more stable physiological state, which may be beneficial for anaerobic succinate production (Liang et al. 2012).

Table 2 Growth and succinate production of various strains constructed in the study
Full size table
Fig. 2
figure 2

NADH/NAD+ ratio during aerobic and anaerobic fermentation. Standard deviations were calculated from the results of four independent experiments

Full size image

Modulating the NADH/NAD+ ratio and ATP level through overexpression of soluble fumarate reductases FRDS1 from S. cerevisiae

Fumarate reductases (FRDs), which catalyze the reduction of endogenous or exogenous fumarate into succinate, are key factors in the anaerobic metabolism of many organisms (Camarasa et al. 2007). The E. coli FRD is a respiratory enzyme, covalently linked to flavin cofactors (FAD) and membrane-bound enzyme (located in the inner membrane) structurally similar to succinate dehydrogenases. Being associated with the respiratory chain, they are directly involved in the production of ATP through oxidative phosphorylation (Tielens and Van Hellemond 1998). NADH-dependent FRD in E. coli can catalyze the metabolic flux flow to succinate and lower the NADH/NAD+ ratio. However, this type of FRD may generate more energy in the interface (a particular period in which the fermentation is switched from the aerobic to the anaerobic condition) from the aerobic to the anaerobic phase because of the connection of the electron transfer chain in YL104N (Fig. 1b), which has a negative effect on succinate production (Zhang et al. 2016). However, inactivating the main aerobic NADH dehydrogenase in YL104 can increase the ATP level at the initial anaerobic phase (Calhoun et al. 1993). The FRDS1 encoded by frd from S. cerevisiae is a soluble enzyme that noncovalently binds FADH2 (cytoplasmic isozyme) and irreversibly catalyzes the reduction of fumarate independently of the electron transfer chain (Camarasa et al. 2007; Enomoto et al. 1996; Muratsubaki and Enomoto 1998). Thus, we speculated that overexpressing frd from S. cerevisiae in E. coli may change the anaerobic ATP level and NADH/NAD+ ratio of YL104N at the beginning of the anaerobic phase.

We placed frd under the tac promoter and introduced it into YL104N. The resulting strain YL104NF produced more succinate (28.0 g/L), with a 19% higher yield than that of YL104N (24.4 g/L; Fig. 3a). The overall yield of succinate was also improved in YL104NF (an increase of approximately 13%). The biomass and aerobic glucose consumption of these two strains were similar; however, the anaerobic glucose consumption of YL104NF (9.27 g/L) was faster than that of YL104 (8.71 g/L). This may have been the key factor for the increased succinate production. In addition, after 63 h, YL104NF generated less ATP and a lower NADH/NAD+ ratio during anaerobic fermentation (Fig. 3b). The intracellular ATP level of YL104NF decreased by 10%, and the NADH/NAD+ ratio decreased by 24%. The results indicated that overexpression of soluble FRDS1 could certainly reduce the anaerobic ATP level and NADH/NAD+ ratio. Consequently, both anaerobic succinate production and the total succinate production were improved.

Fig. 3
figure 3

Growth, succinate production, NADH/NAD+ ratio, and ATP level. a Growth, glucose consumption, and succinate production of strains YL104N and YL104NF. Glucose consumption (dark blue lines), succinate production (red lines), and OD600 (dark cyan lines). b Measurements of ATP level and NADH/NAD+ ratio of strains YL104N and YL104NF at 63 h. Standard deviations were calculated from the results of four independent experiments (color figure online)

Full size image

ATP futile cycling to reduce the ATP level under anaerobic conditions

In the glycolytic pathway, pyruvate kinase catalyzes the formation of pyruvate from phosphoenolpyruvate (PEP) with the production of one ATP molecule. The reverse reaction catalyzed by PEP synthase (encoded by ppsA) from pyruvate to PEP requires ATP as a cofactor; it is converted into AMP. Thus, a cycle in which one ATP molecule is consumed in net is formed (ATP futile cycling; Fig 1a; Hadicke et al. 2015). We introduced this ATP futile cycling into YL104N to improve the overall yield of succinate by reducing the anaerobic ATP level. The resulting strain YL104NP produced more succinate, with a 7% higher yield than that of YL104N (Fig. 4a), and the overall yield of succinate in YL104NP also increased by 17%. Compared with YL104N, biomass, aerobic succinate production, and glucose consumption decreased, but anaerobic succinate production and glucose consumption were faster than that of YL104N. Moreover, the anaerobic succinate titer of YL104NP was 47% higher than that of YL104N (Fig. 4a). These results suggested that a decreased ATP level resulted in increased succinate production and glucose consumption under anaerobic conditions. The decline in biomass and aerobic succinate of YL104NP may result from the leaky expression of ppsA from the trc promoter under aerobic conditions, which consumed the available ATP for growth. Regarding the by-products of YL104NP, formate, acetate, and lactate, the quantities produced were nearly half of those produced with YL104N (Fig. 4b).

Fig. 4
figure 4

Growth, succinate production, and by-products. a Growth, glucose consumption, and succinate production of strains YL104N and YL104NF. Glucose consumption (dark blue lines), succinate production (red lines), and OD600 (dark cyan lines). b By-products of strains YL104N and YL104NP were measured at the end of fermentation (color figure online)

Full size image

The anaerobic ATP level of YL104NP decreased by 33% compared with that of YL104N at the end of fermentation (Fig. 5a). In addition, the aerobic ATP level in YL104NP decreased compared with YL104N because of the leaky expression of ppsA. Moreover, the anaerobic ATP level in YL104NP decreased more rapidly. The tendency of the regression line (the slope k), which was calculated using the ATP levels throughout the entire fermentation process, is crucial (Stouthamer 1979). In the present study, the slope k 1 of the anaerobic trend line of the ATP curve in YL104N was clearly higher than the slope k 2 of YL104NP (−32.3 and −55.6, respectively), indicating that under the anaerobic condition, YL104NP generated less ATP than YL104N (Fig. 5b; Liu et al. 2013; Zhang et al. 2016).

Fig. 5
figure 5

ATP level and generation in strains YL104N and YL104NP. a Measurements of ATP level of strains YL104N and YL104NP at 70 h. b Trend line of ATP during the anaerobic fermentation of strains YL104N and YL104NP. Standard deviations were calculated from the results of four independent experiments

Full size image

Coexpression of frd and ppsA to increase succinate yield

Overexpression of frd and ppsA could individually reduce the anaerobic intracellular ATP level and increase the overall succinate yield, respectively. Thus, we examined whether coexpression of frd and ppsA had a synergistic effect on improving the succinate yield and titer. The data illustrated that overexpression of frd and ppsA individually produced succinate with titers 28.19 and 29.05 g/L, 12 and 9% lower than that produced by YL104NPF (titer 31.87 g/L), respectively (Table 3). Compared with strains YL104NF and YL104NP, the biomass and glucose consumption of YL104NPF were slightly decreased, whereas the overall yield of succinate in YL104NPF was 1.36 mol/mol, which was 13 and 9% higher than those of strains YL104NF and YL104NP, respectively. The final yield of strain YL104NPF was 39% higher than that of the initial strain YL104. Our results suggested that coexpression of frd and ppsA exhibited a synergistic effect on improving succinate production and yield.

Table 3 Growth and succinate production of overexpression strains
Full size table

Discussion

Manipulating redox equivalent and energy balance is crucial for the efficient synthesis of target products, particularly for succinate biosynthesis (Singh et al. 2011; Zhu et al. 2014). In the present study, we applied the soluble FRDs from S. cerevisiae for the first time to modulate the ATP level and NADH/NAD+ ratio to enhance biochemical production.

Most FRDs belong to a multimeric complex associated with the respiratory chain and transfer electrons from quinol to fumarate (Van Hellemond and Tielens 1994). However, soluble monomeric FRDs with FADH2/FMNH2 as an electron donor, not linked to the electron transfer chain, have been identified in several Shewanella species, S. cerevisiae, and the protozoan Trypanosoma brucei (Besteiro et al. 2002; Gordon et al. 1998; Muratsubaki and Enomoto 1998). Overexpression of NADH-dependent FRDs from T. brucei were employed to maintain the cellular redox balance and enhance the cytosolic NAD+ and NADPH pools (Besteiro et al. 2002; Coustou et al. 2005; Salusjarvi et al. 2013). Soluble frds1 from S. cerevisiae were inserted into Aspergillus for reoxidizing NADH to NAD+ under anaerobic conditions, thereby increasing the productivity and yield of citrate on glucose (de Jongh and Nielsen 2008; Enomoto et al. 2002). In the present study, the results revealed that expressing soluble FRDS1 from S. cerevisiae in E. coli could modulate not only the NADH/NAD+ ratio but also the ATP level under anaerobic conditions for increasing succinate production (Cabrera et al. 2011; Hadicke et al. 2015; Liang et al. 2012; Singh et al. 2009; Vemuri et al. 2006; Zhang et al. 2016).

In this study, sole anaerobic NADH dehydrogenase and ATP futile cycling were employed to regulate the NADH/NAD+ ratio and ATP level in the interface from the aerobic to anaerobic phase, respectively.

In previous studies, through respiratory chain engineering, combining various NADH dehydrogenases and terminal oxidases with different efficiencies of proton translocation enabled E. coli flexibility to provide diverse growth conditions to produce fermentation products (Becker et al. 1997; Portnoy et al. 2008). Yokota et al. observed the highest rate of glucose metabolism in the double mutant (NAD-1 and cytochrome bo3; Kihira et al. 2012). In a glucose-limited chemostat, a branched electron transport chain was adopted rather than a linear one, and the growth behavior and acetate formation were analyzed under this condition (Steinsiek et al. 2014). The quinone synthesis pathway (Wu et al. 2015) and NDH-II dehydrogenase (Liu et al. 2014) were manipulated for enhancing lactate and polyhydroxybutyrate production, respectively. Furthermore, in succinate production, various strategies for reducing the NADH/NAD+ ratio have been employed. These strategies include regeneration of NAD+ in the reductive TCA cycle (Goldberg et al. 1983; Liang et al. 2011; Millard et al. 1996; Stols and Donnelly 1997) and provision of additional reducing power (Van der Werf et al. 1997). In the present study, only one type of NADH dehydrogenase of the respiratory chain was adopted, thus increasing the overall yield of succinate and alleviating the change in the NADH/NAD+ ratio during the entire fermentation process.

High ATP levels have been proven to inhibit the activities of key enzymes involved in the central metabolism, thus downregulating the glycolytic pathway and TCA cycle (Kihira et al. 2012; Kim et al. 2008; Weitzman 1981). By using enhanced futile cycling to decrease the ATP supply, production rates and yields of fermentation products were increased, and glucose consumption was elevated under aerobic conditions (Chao and Liao 1994; Patnaik et al. 1992). Recently, the same strategy was also applied for improving the production of lactate under anaerobic conditions (Hadicke et al. 2015). In the present study, we proved that ATP futile cycling is an effective strategy for improving succinate yield.

Finally, our results demonstrated that a combination of these methods exerted a synergistic effect on improving the overall yield of succinate, which was 39% higher than that of the strain constructed in our previous study. Moreover, the results indicate that the regulation of the NADH/NAD+ ratio and ATP level, which is not involved in the succinate biosynthetic pathway, is an efficient strategy for succinate production.