Improved production of 2,3‐butanediol and isobutanol by engineering electron transport chain in Escherichia coli

In this study, we attempted to reduce the electron transport chain (ETC) activity by making multiple mutants to secure availability of NADH, and then heterologous pathways for synthesis of 2,3‐butanediol and isobutanol were introduced. ETC mutants produced higher levels of metabolites with higher NADH/NAD+ ratio. The results proved that engineering of ETC is a useful metabolic engineering approach for production of reduced metabolites.


Introduction
Most of the facultative aerobes catabolize carbon sources through glycolysis and TCA cycle and generate additional ATP through respiration under aerobic conditions, which is called oxidative phosphorylation. It is known that the cellular energy production from oxidative phosphorylation (OXPHO) is a higher-energy-yield metabolic process than substrate level phosphorylation (SLPHO) via fermentative metabolism (Ward, 2015). Therefore, better fitness of host strain is usually observed under aerobic conditions. This is advantageous to operation of industrial bioprocesses because high cell density cultivation of host strain results in increased productivity during the bioprocess operation. However, oxidized intracellular redox state is observed under aerobic conditions. Reduced condition is often required for the production of value-added compounds such as biofuels and commodity chemicals. Cellular reducing cofactors such as NAD(H) are involved in the intracellular redox reaction. Therefore, most bioprocesses are frequently operated in microaerobic conditions, and fine optimization is inevitable to satisfy both cell growth and reducing power (Wong et al., 2014;Chan et al., 2016;Wu et al., 2016). An alternative approach involving the separation of growth phase and production phase, by manipulating oxygen supply, has been attempted for efficient production of target metabolites (Lee et al., 2019). Although several different methods exploring these aspects of the process control have been suggested, problems about inhomogeneous oxygen level in large-scale fermenter and discontinuous oxygen supply sometimes disable the metabolism of host strain (Chubukov et al., 2016).
Recently, a different perspective regarding the efficiency of SLPHO and OXPHO has been suggested in several studies (Chen and Nielsen, 2019;Basan et al., 2015;Jung et al., 2019). Although higher amount of ATP is generated via OXPHO than SLPHO, higher protein costs are required to operate OXPHO. According to a previous study, SLPHO has advantages such as high ATP production rates, which results in increased growth rate and glucose uptake rate (Basan et al., 2015). Some microorganisms exhibit active fermentative pathway despite the presence of oxygen. Saccharomyces cerevisiae, for example, mainly produced a fermentative product, ethanol, using large amounts of glucose under aerobic conditions, which is called Crabtree effect (Pfeiffer and Morley, 2014). Similar metabolic characteristics have been observed in various organisms in substrate rich conditions [lactic acid production in cancer cell; succinate production in trypanosomatids; 2,3-butanediol (2,3-BDO) production in Klebsiella pneumoniae and Enterobacter aerogenes] (Bringaud et al., 2010;Jung et al., 2014;Ji et al., 2011;Goodwin et al., 2015). Microorganisms have evolved to have an active glycolysis and fermentative pathway rather than TCA cycle and respiration to rapidly take up carbon sources from the surroundings. Mimicking these characteristics makes it possible to improve sugar consumption and enhance the production of reduced metabolites under aerobic condition in Escherichia coli.
The electron transport chain (ETC) complex is a significant workhorse for bacterial respiration under aerobic conditions. The oxidation of electron donors and reduction of electron acceptors generates electron flow, and it drives the proton motive force, using which intracellular ATP can be generated. Under aerobic conditions, NADH and oxygen were utilized as electron donor and acceptor respectively. The electron from oxidation of NADH is transferred to oxygen through ETC components forming H 2 O. The ETC of E. coli is composed of two protein complexes, namely NADH dehydrogenase, and terminal oxidase, in which the electron transfer is mediated by membrane quinone (Sousa et al., 2012). The cellular ATP can be synthesized by ATPase using the proton gradient generated by ETC, which is called oxidative phosphorylation. There are two major NADH dehydrogenases in E. coli. The first being NADH dehydrogenase I (NDH-1, encoded by nuo operon), which is composed of 13 protein subunits, generating proton motive force, H + / e À = 2, during electron transfer to quinone pool. The other being NADH dehydrogenase-2 (NDH-2, encoded by ndh), which is a single unit enzyme that is not engaged in proton motive force (H + /e À = 0) but in intracellular redox management (Matsushita et al., 1987). There are several types of membrane quinols, such as demethylmenaquinol (DMK), menaquinol (MK) and ubiquinol (UQ), in E. coli. The expression of ubiquinone synthesis pathway was repressed by global regulators such as Fnr and ArcA under anaerobic conditions (van Beilen and Hellingwerf, 2016). Generally, ubiquinol is dominantly synthesized under aerobic conditions and menaquinol under anaerobic conditions. The three terminal oxidases, namely CyoABCD (H + /e À = 2), CydAB (H + /e À = 1) and CbdAB (H + /e À = 0), are functional in E. coli (Puustinen et al., 1991). The Cyo and Cyd complex can generate proton motive force, but the role of Cbd complex is not fully known. All these components can be candidates for engineering to modulate the respiration activity of E. coli.
The metabolic changes owing to mutations of ETC components have been demonstrated in various bacteria such as E. coli, Pseudomonas sp., Corynebacterium glutamicum, Bacillus subtilis, Lactococcus lactis and Zymomonas mobilis (Nies et al., 2019;Koch-Koerfges et al., 2013;Zhu et al., 2011;Kalnenieks et al., 2019;Wu et al., 2015). In most of the cases, the manipulation of respiration level resulted in retardation of growth, decreased oxygen uptake and improved production of organic compounds, such as lactate, acetate, succinate and ethanol, under aerobic condition. Especially, in the case of B. subtilis, the ETC mutant exhibited better fitness than that of the wild-type strain because of the inefficient function of ETC, which contributes to enhanced production of target metabolites such as riboflavin and N-acetylglucosamine (Zamboni et al., 2003;Liu et al., 2014). In a different approach, the activity of ETC was controlled by engineering membrane lipid composition of E. coli, which affected the diffusivity of membrane quinon(ol) (Budin et al., 2018).
In this study, the activity of ETC was controlled to produce reduced compounds under aerobic conditions. The expression of ETC component proteins (NADH dehydrogenase, membrane quinone and terminal oxidase) was manipulated by using CRISPR/Cas9 system to achieve a highly reduced intracellular state in E. coli. The resulting strains exhibited the phenotype of aerobic fermentation with accumulated pyruvate and acetate. When the synthetic pathway for 2,3-BDO and isobutanol was introduced to the mutants, the production of these metabolites was highly improved, regardless of oxygen supply. The intracellular concentration of NADH and ATP was analysed to understand their effects on the production of reduced compounds. The oxygen-insensitive metabolic activities were observed in ETC mutant, which were crucial to large-scale fermentation. It was confirmed that formulation of reduced state is helpful for production of 2,3-BDO and isobutanol in ETC mutants.

Phenotypical changes of ETC mutation
The ETC is one of the major energy generating systems in E. coli under aerobic condition. The NADH supplied from glycolysis and TCA cycle can be oxidized by ETC, during which cellular ATP can be produced. When the ETC of a strain is inactivated, the metabolism of the mutant is similar to that observed under anaerobic conditions for the parent strain. The ETC of E. coli works through three components, namely NADH dehydrogenase, quinone, and terminal oxidase, using NADH as the electron donor and oxygen as the electron acceptor (Fig. 1). The phenotypical changes of mutants with inactivated ETC components were investigated in this study. As shown in Table 1, the two NADH dehydrogenase (Nuo and Ndh) were inactivated in combination with the two terminal oxidases (Cyo and Cyd), thus, forming four mutants (ETC1, ETC2, ETC3 and ETC4). Cbd, one of the three terminal oxidases, was excluded due to its negligible effect on respiration (Bekker et al., 2009). The membrane quinone synthesis pathway was also modulated to reduce the ETC activity. The expression of UbiE (demethylmenaquinone methyltransferase/2-methoxy-6-polyprenyl-1,4-benzoquinol methylase), which is involved in both ubiquinone and menaquinone synthesis pathway of E. coli (Fig. 1A), was controlled by engineering its The overall pathway scheme of electron transport chain in engineered mutants with heterologous gene expression for synthesis of reduced compounds. The redox cofactor NADH is generated by glycolysis and TCA cycle and consumed by electron transport chain under aerobic conditions. The intracellular NADH can be accumulated when the activity of electron transport chain was inactivated despite aerobic conditions. The two NADH dehydrogenases (Nuo and Ndh) and two terminal oxidases (Cyo and Cyd) were removed in combinations, and expression of quinone synthesis pathway (ubiE) was modulated in this study, resulting in the construction of seven ETC mutants. The heterologous pathways were expressed using the electron transport chain inactivated mutants as host strains to produce reduced compounds. The overexpressed heterologous genes are marked in blue (budB, acetolactate synthase; budA, acetolactate decarboxylase; budC, acetoin reductase; ilvC, ketol-acid reductoisomerase; ilvD, dihydroxyacid dehydratase; kivD, a-ketoisovalerate decarboxylase; and adhA, alcohol dehydrogenase). The required cofactors for synthesis of reduced compounds such as 2,3-butanediol and isobutanol are marked in red. Several metabolites were written in abbreviation (2,3-DIV, 2,3-dihydroxyisovalerate; KIV, 2-ketoisovalerate; UQ, Ubiquinone; and MK, Menaquinone). (B) Modification of ubiE 5 0 -UTR using CRISPR/Cas system. The expression levels of UbiE in wild-type strain were predicted and synthetic UTR sequences which are expected to have lower expression levels were designed by UTR designer. The RBS of ubiE was replaced using CRISPR/Cas9 guided cleavage followed by the repair of the broken DNA via homologous recombination. As a result, three engineered strains were constructed (UbiE117, UbiE86 and UbiE26). 5 0 -untranslated region (UTR) (Ravcheev and Thiele, 2016). The native expression level of ubiE was predicted by bioinformatics web tool UTR designer and three artificial UTRs having lower expression levels were rationally designed (Fig. 1B). The three artificial UTRs were inserted to genomic DNA of parental strain DSM01 using CRISPR/Cas9 system (UbiE117, UbiE86, and UbiE26). For targeting 5 0 -UTR of ubiE, CRISPR RNA-expression plasmid (pZS CRISPR ubiE RBS sacB) and Cas9-expression plasmid (pCas9) were introduced into the parental strain ( Table 1). The three types of doublestranded linear DNA were introduced with homologous recombination for repair of Cas-mediated cleavage (Fig. 1B). The modified UTR of ubiE was confirmed by sequence analysis (Fig. S2).
The seven constructed mutants and parental strain were cultivated in flask. The growth retardation of all mutant strains was identified. The UbiE knockdown mutants exhibited 2-8% lower final OD, while the oxidoreductase mutants (ETC1, ETC2, ETC3 and ETC4) exhibited 15-39% lower final OD ( Fig. 2A). These results corresponded to previous reports that inactivation of ETC inhibited growth of the host strain (Zambrano and Kolter, 1993;Pr€ uss et al., 1994). The glucose uptake of ETC mutants was equal or higher, while their fitness was impeded compared to the parental strain ( Fig. 2B). The improved glucose uptake in non-respirating E. coli has been well discussed in previous reports (Gonzalez et al., 2017;Chen et al., 2011). Interestingly, 6-190% higher accumulation of pyruvate was observed in ETC mutants in the early stationary phase. The pyruvate was completely re-consumed in the parental strain DSM01, but excessive pyruvate was left at the end of fermentation in several ETC mutants (Fig. 2C). Furthermore, higher production of acetate was exhibited in most of the ETC mutants. The increased yields of acetate (32-63%) were seen in UbiE86, ETC1, ETC2, ETC3 and ETC4 after 24 h cultivation (Fig. 2D).
A previous report indicates that enzyme reactions involved in glycolysis, such as the function of the pyruvate dehydrogenase complex (PDHc), can be inhibited by a high redox ratio commonly observed in anaerobic conditions (Kim et al., 2008). Therefore, the high redox state of ETC mutants hinders the conversion of pyruvate to acetyl-CoA, and excessive pyruvate is accumulated in the culture medium. Acetate is another major by-product of ETC mutants. Acetate can be produced through pyruvate oxidase (PoxB)-mediated pathway because the Pta and Ack-mediated pathway had already been eliminated in DSM01. PoxB is a membrane-bound enzyme that oxidizes pyruvate to acetate, which induces proton motive force by transferring electrons to the ETC module (Abdel-Hamid et al., 2001). Therefore, the production of acetate can facilitate energy replenishment by consuming pyruvate. Furthermore, it has been demonstrated that the deletion of ETC components, especially quinone synthesis pathway and terminal oxidase, can increase the formation of superoxide ions (Korshunov and Imlay, 2006). This can be another reason for the aggravation of cellular growth. The pyruvate oxidation through PoxB might be favourable because it can divert the reaction of pyruvate dehydrogenase, which avoids NADH-mediated electron transport system (Moreau, 2004). Accordingly, NADH-consuming synthetic pathway utilizing pyruvate as a precursor could be effectively working in the ETC mutants. Improved production of 2,3-butanediol (2,3-BDO) and isobutanol in ETC mutants We expected that the alteration of cellular metabolism observed in ETC mutants can have a positive effect on the production of reduced compounds. For preliminary experiments, the growth of DSM01 harbouring 2,3-BDO and isobutanol pathway and production of reduced compounds were investigated in zinc ion containing culture medium, referring to the report that zinc ion inhibits activity of NADH dehydrogenase in E. coli (Schulte et al., 2014). Approximately 0-2 mM of ZnSO 4 was added in the culture medium. The addition of zinc ion inhibited growth of strains; the strains hardly grew in the presence of 2 mM of zinc ion (Fig. S1A,D). We attempted to produce two reduced chemical products, 2,3-BDO and isobutanol, from engineered strains since the redundant pyruvate was accumulated and higher redox ratio was expected in the ETC mutants. The 2,3-BDO synthesis pathway was constructed by introducing budABC (from E. aerogenes) harbouring plasmid and the isobutanol production pathway by introducing plasmid harbouring ilvDC and budB (from K. pneumoniae) along with kivD and adhA (from L. lactis), to the engineered strains Jung et al., 2017). The 2,3-BDO and isobutanol producing transformants were termed as 'strain-BDO' and 'strain-IB' respectively (Table 1). The production of isobutanol was gradually enhanced as the concentration of ZnSO 4 increased. However, production of 2,3-BDO exhibited negligible difference regardless of the addition of zinc ion (Fig. S1B,E). However, the specific yields of 2,3-BDO and isobutanol improved up to 53% and 174%, respectively, in 12 h cultivation (Fig. S1C,F). These results denoted that inhibition of NADH oxidation was helpful for the production reduced chemical compounds in E. coli. The growth deviation between parental strain and mutants was alleviated when the reduced compound synthesis pathway was expressed (Fig. 3A,C). We postulated that intracellular unused pyruvate and NADH were consumed by the introduced pathways, which relieved metabolic imbalance inside the cell. Therefore, the recovered growth was exhibited, and pyruvate was not produced at all by the strains with the introduced metabolic pathway (data not shown). The production of 2,3-BDO was improved in ETC mutants, but was not in the UbiE knockdown mutants. The titres and yields were increased 13-113% and 6-44%, respectively, in four oxidoreductase mutants (Fig. 3B). Among the oxidoreductase mutants, Cyo mutants (ETC2-BDO and ETC4-BDO) exhibited higher production of 2,3-BDO compared with Cyd mutants (ETC1-BDO and ETC3-BDO). Furthermore, the Nuo mutants (ETC3-BDO and ETC4-BDO) displayed higher production than Ndh mutants (ETC1-BDO and ETC2-BDO). Especially, the highest titre and yield of 2,3-BDO were observed in ETC4-BDO strain, which were 113% and 44%, respectively, more than those in DSM01-BDO (Fig. 3B). On the other hand, the production of isobutanol was significantly improved in all ETC mutants and UbiE knockdown mutants. A gradual increase of isobutanol titre was observed in UbiE117-IB, UbiE86-IB and UbiE26-IB, by 3.1-fold, 5.2-fold and 6.1fold, respectively, compared with the isobutanol titre in DSM01-IB (Fig. 3D). The isobutanol production increased by 5.1-fold, 6.9-fold, 5.74-fold and 7.7-fold in ETC1-IB, ETC2-IB, ETC3-IB and ETC4-IB respectively (Fig. 3D). Similar to 2,3-BDO production case, Cyo mutants (ETC2-IB and ETC4-IB) exhibited more effective production of isobutanol with respect to Cyd mutants (ETC1-IB and ETC3-IB). Also, Nuo mutants (ETC3-IB and ETC4-IB) were more effective for isobutanol production than Ndh mutants (ETC1-IB and ETC2-IB). Higher yields of isobutanol were confirmed in all ETC mutants than those in the parental strain, (DSM01-IB) especially a 5.0-fold increase was observed in ETC4-IB strain (Fig. 3D). It is postulated that Nuo and Cyo are more influential to cell physiology and metabolism than Ndh and Cyd in this culture condition, which led to a synergetic effect for production of reduced compounds.
These results were consistent with previous studies that showed that inactivation of NADH dehydrogenases improved the production of 2,3-BDO and isobutanol in Klebsiella pneumoniae and Corynebacterium glutamicum (Koch-Koerfges et al., 2013;Zhang et al., 2018). It was expected that the improved production of the reduced compounds can be obtained by an increased intracellular NADH/NAD + ratio induced from the inactivation of ETC. Because isobutanol is a chemically more reduced compound than 2,3-BDO, isobutanol production pathway requires more reducing power (1 NADH and 1 NADPH from pyruvate) than the 2,3-BDO production pathway (1 NADH). Thus, the intracellular redox ratio affected the production of isobutanol to a comparatively larger extent than 2,3-BDO in the ETC mutants.

Analysis of intracellular redox and energy state in ETC mutants
To comprehend the cause of metabolic alteration in ETC mutants, the intracellular redox and energy state were analysed. We tried to compare the NAD(H) and ATP contents of seven ETC mutants and parental strain harbouring reduced compound synthesis pathway. The culture broth was acquired from the middle of exponential phase in which the overall metabolic process could be vigorous involving central carbon metabolism and respiration. The redox ratios of UbiE knockdown mutants were comparable to that of the parental strain, which were in consistence with 2,3-BDO production (Fig. 4A). However, the NADH/NAD + increased by 52%, 55%, 16% and 71% in ETC1-BDO, ETC2-BDO, ETC3-BDO and ETC4-BDO, respectively, compared with the NADH/ NAD + in DSM01-BDO (Fig. 4A). The ratios were especially high in Cyo mutants (ETC2-IB and ETC4-IB), explaining the improved production of 2,3-BDO in those mutants (Fig. 4A). Considering the all UbiE and ETC mutants, we could confirm the positive correlation between redox ratio and 2,3-BDO titres (Fig. 4C). Meanwhile, the ATP contents were similar between UbiE knockdown mutants compared to the parental strain, and were higher in Cyd mutants (ETC1-IB and ETC3-IB) and Nuo mutants (ETC3-IB and ETC4-IB) (Fig. 4B). Considering all mutants, there was no clear correlation between ATP contents and 2,3-BDO production (Fig. 4D).
All ETC mutants harbouring isobutanol pathway exhibited increased NADH/NAD + ratios compared to that of the parental strain. Approximately 1.1-2.5-fold enhanced redox ratios were observed in ETC mutants, among which ETC4-IB showed the highest redox ratio (Fig. 4E). As the expression of ubiquinone synthesis pathway was weakened by introducing synthetic UTRs (UbiE117, UbiE86 and UbiE26), gradually increased redox ratios were observed in mutants (Fig. 4E). Furthermore, highly increased redox ratios were observed in ETC3-IB and ETC4-IB, meaning that Nuo deletion was more crucial for perturbing intracellular redox ratio than that of terminal oxidase inactivation in case of isobutanol production (Fig. 4E). A clear positive correlation was identified between redox ratios and isobutanol titres (Fig. 4G). However, ATP contents were not much changed among UbiE and ETC mutants (Fig. 4F), resulting in no significant correlation between ATP contents and isobutanol titres (Fig. 4H). Even if ATP was not supplied from ETC, it can be supplemented through SLPHO such as glycolysis and acetate formation. This could be the reason for comparable or higher ATP content of several mutants than that of the parental strain (Fig. 4B,F). Moreover, higher amount of glucose consumption and acetate production were observed by ETC mutants, which demonstrated that SLPHO was the major energy forming pathway in ETC mutants. It has been reported that OXPHO is favourable in terms of ATP/substrate yield, but disadvantageous in ATP production rate and enzymatic cost (Chen and Nielsen, 2019;Basan et al., 2015). Therefore, it is speculated that the formation of intracellular ATP was determined by much more complicated factors in ETC mutants.

Scaled-up cultivation to verify oxygen-independent metabolism in ETC mutants
One of the challenging issues in industrial fermentation is the limitation of oxygen transfer in scaled-up fermenters (Garcia-Ochoa and Gomez, 2009). For instance, respiring cells could be present in oxygen-rich spot and fermenting cell in oxygen-limited spot (Chubukov et al., 2016). In other words, various phenotypes could be existing according to the oxygen level in different spots of a batch cultivation. The heterogeneous conditions in the fermenter can cause problems such as retardation of growth, generation of by-products and reduction of product yield. For this reason, the problems of oxygen transfer in large-scale fermentation can be resolved by developing strains carrying out cellular metabolism through oxygen-insensitive way. It can be hypothesized that ETC mutants are less affected by extracellular oxygen concentration than the parental strain. To verify this, scaled-up cultivation was implemented in various agitation conditions. The growth and extracellular metabolites profile of DSM01-BDO and ETC4-BDO were compared in 3 l-scale fermentation because ETC4-BDO exhibited the most significant improved 2,3-BDO production compared with parental strain in flask experiments.
When cultivated in 125, 250 and 500 rpm, both of DSM01-BDO and ETC4-BDO produced less 2,3-BDO as the agitation increased. Interestingly, the least oxygen-induced effect was identified in ETC4-BDO (Fig. 5A,B). The two strains produced comparable titres of 2,3-BDO at oxygen-limited condition (125 rpm). DSM01-BDO produced 74% reduced amount of 2,3-BDO at high agitation condition (500 rpm) compared with the amount of 2,3-BDO by ETC4-BDO. The 2,3-BDO yield of DSM01-BDO decreased by 64% but that of ETC4-BDO reduced by 33% as the agitation increased from 125 to 500 rpm (Fig. 5C). Thus, it is confirmed that the ETC mutants can operate fermentative metabolism independent of oxygen supply.
It was also hypothesized that ETC mutants could have lower oxygen utilization than the parental strain. The specific dissolved oxygen (DO) change rate was calculated as previously published protocols (Long and Antoniewicz, 2019). The specific growth rate (l) and specific DO change (YDO/Biomass) were measured during early and late exponential phase. The l of DSM01-BDO and ETC4-BDO were 0.347 and 0.375 h À1 , respectively, while the Y DO/Biomass values were À0.613 mg (g dried cell weight (DCW)) À1 and À0.424 mg gDCW À1 , respectively, in exponential phase (4-8 h) (Fig. 5D). The calculated specific DO change rates (l 9 Y DO/Biomass ) of DSM01-BDO and ETC4-BDO were À0.212 and À0.159 mg gDCW À1 h À1 , respectively (Table S2). This indicated that ETC4-BDO exhibited 26% lower oxygen utilization compared to DSM01-BDO in exponential phase. The difference of specific DO change rates between two strains was even greater in late exponential phase (8-12 h) (Table S2). Considering the specific DO change rate, ETC mutant clearly displayed lower oxygen utilization phenotype than the parental strain.
Although the oxygen-insensitive phenotype appeared to have improved production of reduced compounds, ETC4-BDO exhibited inferior final OD and higher glucose uptake, inducing larger amount of acetate than DSM01-BDO (Fig. S3). The metabolic alteration could be due to energy deficiency and the activation of SLPHO to replenish ATP. The energy deficiency originating from the inactivation of ETC was compensated by increased glycolysis and acetate formation. The accumulation of acetate can cause acidification of culture broth and reduced product yields. In order to relieve acetate overproduction in ETC mutants, the alternative ATP supplementation can be applied. Several alternative ATP generating pathways are demonstrated in a previous report (Unden and Bongaerts, 1997). Furthermore, light or electric energy can be transformed to cellular energy through bacterial rhodopsin and electro-fermentation techniques (Walter et al., 2007;Wu et al., 2019). Further studies utilizing external energy sources can be helpful to resolve the problems related to growth retardation and acetate accumulation in ETC mutants. It has been also documented that the inner membrane space is filled with various membrane proteins such as transporters and components of the respiration machinery (Bernsel and Daley, 2009;Papanastasiou et al., 2013). Thus, the removal of ETC components, which secures considerable volume of the membrane, could increase allocation of sugar transporters in the inner membrane space (Szenk et al., 2017). Therefore, for a comprehensive understanding of the physiology of the ETC mutants, an investigation of the redox and energy metabolism as well as changes of membrane proteome caused by the removal of ETC components are required.

Conclusion
This report described the phenotypical traits (growth, metabolites, and oxygen uptake) and intracellular state (NADH and ATP) in ETC mutants. Because the ETC is engaged in the regeneration of redox cofactors, the attenuation of its activity affected the intracellular redox ratio. The metabolic perturbations in ETC mutants led to improved production of reduced metabolites, 2,3-BDO and isobutanol and positive correlation was observed between redox ratios and production of reduced metabolites in several ETC mutants. The highest production of reduced metabolites was observed with ETC4 (Mnuo Mcyo) mutant. This result suggests that optimization of cellular metabolism can be obtained by modulating the ETC of the host strain. Furthermore, ETC mutants are more reliant on SLPHO than OXPHO compared to parental strain, which resulted in ETC mutants having oxygen-independent metabolism. In conclusion, the discipline has potential for innovative engineering in the future, such as introduction of novel ETC components or efficient ATP generating module beyond innate ETC in E. coli.

Construction of strains and plasmids
All the strains and plasmids used in this study are listed in Table 1 and the oligomers (Bionics, Seoul, South Korea) are listed in Table S1. The genetic engineering for removal of ETC genes was based on k-red homologous recombination. The mutants were constructed in two types of NADH dehydrogenase and terminal oxidase in a combinatorial way which resulted in four ETC mutants. The 2,3-butanediol synthesis pathway (budABC operon from K. pneumoniae) was inserted to pZS21 MCS vector which is called pZS-BDO. The isobutanol pathway genes, namely, ilvDC and budB from K. pneumonia along with kivD and adhA from L. lactis, were inserted into pZA31 MCS and pBTBX-2 vector, respectively, which were called pZA-DCB and pBT-DA. pZS-BDO, pZA-DCB, and pBT-DA  were introduced into ETC mutants (Fig. 1). In order to modulate the quinone synthesis in E. coli, genome editing was carried out using CRISPR/Cas9 system. The strains for genome editing were cultivated and prepared for electroporation. The 5 0 -UTR sequence of demethylmenaquinone methyltransferase (ubiE) was targeted by crRNA. The crRNA and tracrRNA containing vector (pZS CRISPR ubiE RBS SacB), Cas9 harbouring vector (pCas9), and linear DNA fragment for DNA repair were introduced to parental strain for targeting, cleavage, and modifying ribosome binding sequence of ubiE. Then, artificial 5 0 UTR sequence was inserted by homologous recombination for recovery of cleavaged genomic DNA (Heo et al., 2017). The artificial UTR sequence was designed by 'UTR designer' to reduce the expression of ubiE. To confirm the mutations, the modified sequences were PCR amplified and analysed (Bionics, Seoul, Korea).

Medium and cultivation
Lysogeny broth (LB; 5 g l À1 yeast extract, 10 g l À1 tryptone, 10 g l À1 NaCl) was utilized for all the genetic engineering procedures. Appropriate concentration of antibiotics (100 µg ml À1 of carbenicillin, 50 µg ml À1 of kanamycin and 34 µg ml À1 of chloramphenicol) were applied for selection of transformed strains with antibiotics resistance marker. During flask cultivation of strains containing two or more plasmids, half the concentration of antibiotics was applied to the cultivation medium. For preparation of 2,3-BDO production medium, the modified M9 minimal medium (6 g l À1 Na 2 HPO 4 , 3 g l À1 KH 2 PO 4 , 1 g l À1 NH 4 Cl, 0.5 g l À1 NaCl, 0.01% of Thiamine-HCl) also including 20 g l À1 of glucose, 2 g l À1 of yeast extract and 1 ml of trace elements (2.86 g l À1 H 3 BO 3 , 1.81 g l À1 MnCl 2 Á4H 2 O, 0.22 g l À1 ZnSO 4 Á7H 2 O, 0.39 g l À1 Na 2 MoO 4 Á2H 2 O, 0.079 g l À1 CuSO 4 Á5H 2 O, 49.4 mg l À1 Co(NO 3 ) 2 Á6H 2 O, and 0.9 g l À1 FeCl 3 Á6H 2 O) per litre, was employed for flask cultivation. For preparation of isobutanol production medium, modified 2 9 M9 medium (12 g l À1 Na 2 HPO 4 , 6 g l À1 KH 2 PO 4 , 2 g l À1 NH 4 Cl, 1 g l À1 NaCl, 0.01% of Thiamine-HCl) including 40 g l À1 of glucose was utilized with same concentration of yeast extract and trace elements. One gram of calcium carbonate (CaCO 3 ) was supplied for prolonged cultivation in isobutanol production. The seed inoculum was prepared in 15 ml conical tube by overnight cultivation in production medium at 37°C All host strains were cultivated in 250 ml Erlenmeyer flasks with 40 ml of working volume, at 37°C, and shaking at 250 rpm, with initial OD 600 around 0.05. All the chemical reagents were purchased from Sigma-Aldrich (St. Louis, MO, USA), unless otherwise mentioned.

Analytical methods
The cell growth was measured via turbidity of the fermentation broth at 600 nm (OD 600 ) using a UV-Vis spectrophotometer DU730 (Beckman Coulter, Caguas, Puerto Rico). The DCW was derived from the calculation that DCW = OD 600 9 0.36. For the analysis of extracellular metabolites, culture broth was acquired by centrifugation using CF-10 centrifuge (9000 g) (Daihan Scientific, Seoul, South Korea), followed by filtration (0.2 lm pore) (Whatman, Maidstone, UK), and then, the supernatant was analysed by high-performance liquid chromatography (HPLC) with refractive index detector Waters 2414 (Waters, MA, USA), employed with a SH1011 column (Shodex, Tokyo, Japan). A diluted sulphuric acid solution (10 mM) was applied for the HPLC mobile phase with a flow rate of 0.6 ml min À1 .

Analysis of intracellular redox and energy state
To measure the intracellular NADH/NAD + ratio, NAD/ NADH-GloTM Assay kit (Promega, WI, US) was utilized. Briefly, the cell broth was prepared by cultivating to exponential phase using production medium. The culture broth was mixed with DTAB solution for 5 min. For measuring NADH, 0.4 N of HCl was added to the reaction mix, whereas for measuring NAD + , nothing was added. The sample was heated to 60°C for 15 min and cooled to 25°C. The HCl/Trizma solution was added to the reaction mix for NADH, whereas Trizma solution was added to that for NAD + . The prepared solutions were mixed with detection reagent at a ratio of 1:1 (v/v) and incubated for approximately 30 min. The luminescence was measured using microplate reader Biotek Synergy H1 (Biotek, VT, USA). The same methods were applied for measuring NAD(H) standard solution. The NADH and NAD + were quantified, and the ratio of them was calculated.
For measuring the intracellular ATP content, BacTiter-GloTM kit was utilized (Promega, WI, USA). Briefly, the cell broth was prepared by flask culture. The cell broth in exponential phase was taken and washed using distilled water. The same volume of cell broth and reaction mix was resuspended together and rested for 5 min in ambient condition. The luminescence was measured using microplate reader Biotek Synergy H1 (Biotek). The same methods were applied for measuring the ATP standard solution.

Large-scale fermentation and dissolved oxygen measurement
The seed inoculum was prepared in 15 ml conical tube using production medium, in a 37°C shaking incubator, by overnight cultivation. The seed was transferred to fresh medium for flask cultivation. When the culture reached mid-exponential phase, 50 ml of culture broth was again inoculated to 3 l-scale fermenter system (CNS, Daejeon, South Korea) with working volume of 1 L. The production medium contains 60 g l À1 of glucose, 1 9 M9 salts, and the same components of production medium in the section 'medium and cultivation'. The temperature and pH (Broadley James, CA, USA) in fermenter were maintained 37°C and 6.5, respectively, and potassium hydroxide (KOH) was used as a base solution. The aeration was fixed to 1.5 vvm (air), and agitation was varied from 125 to 500 rpm. The dissolved oxygen (DO) concentration was measured every 4 min at 600 rpm and 1.5 vvm (air) by using DO probe (Broadley James, CA, USA) equipped in the fermenter. The initial DO value that is saturated by air flow was set as 100% before inoculation in fermenters and the DO value that is calibrated by 2 M of sodium sulphite (Na 2 SO 3 ) was set as 0%. The DO (%) value was measured every 4 min as fermentation proceeded. The DO concentrations (mg l À1 ) of air-saturated culture medium (100%) and 2 M sodium sulphite solution (0%) were measured by DO meter, Orion3-Star Plus, (Thermo Scientific, MA, USA) in order to convert DO (%) to absolute DO concentration (mg l À1 ). The DO concentration (mg l À1 ) of airsaturated culture medium was 6.61 mg l À1 and that of 2 M sodium sulphite solution was 0.21 mg l À1 . The specific DO change rate was calibrated by multiplication of specific growth rate (l) and DO change (Y DO/Biomass ) in DSM01-BDO and ETC4-BDO (Mnuo Mcyo) strain (Long and Antoniewicz, 2019). aerobic formation of reduced chemical products. Metab Eng 13: 704-712.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Table S1. Oligomers used in this study. Table S2. Specific growth rate, specific DO change and specific DO change rate in DSM01-BDO and ETC4-BDO Fig. S1. Effects of ZnSO4 to growth, production and specific yield of 2,3-butanediol and isobutanol. (A) The growth, (B) production titers and (C) specific yield of 2,3-butanediol were exhibited along with the varied concentration of ZnSO 4 . The growth retardation and comparable 2,3-butanediol production were observed as the addition of ZnSO 4 was increased. Thus the specific yield of 2,3-butanediol was improved by addition of ZnSO 4 (N.D. means "Not Detected"). (D) The growth, (E) production titers and (F) specific yield of isobutanol were exhibited along with the varied concentration of ZnSO 4 . The growth retardation and enhanced isobutanol production were observed as the addition of ZnSO 4 was increased. Thus the specific yield of isobutanol was improved by addition of ZnSO 4 . Fig. S2. Sequencing confirmation of UbiE knock down mutants. The modulated 5`-UTR sequence of ubiE were PCR amplified and analyzed to confirm the mutations.