Reduction of The Pyruvate Decarboxylase Activity Improves Isobutanol Production By Klebsiella Pneumoniae

Klebsiella pneumoniae contains an endogenous isobutanol synthesis pathway. ipdC, annotated as an indole-3-pyruvate decarboxylase (Kp-IpdC), was identied to catalyze the formation of isobutyraldehyde from 2-ketoisovalerate. Compared with 2-ketoisovalerate decarboxylase from Lactococcus lactis (KivD), a decarboxylase commonly used in articial isobutanol synthesis, Kp-IpdC has an 2.8-fold lower K m for 2-ketoisovalerate, leading to higher isobutanol production without induction. However, high level expression of ipdC by induction resulted in a low isobutanol titer. In vitro enzymatic reactions showed that Kp-IpdC exhibits promiscuous pyruvate decarboxylase activity, which adversely consume the available pyruvate precursor for isobutanol synthesis. To address this we have engineered Kp-IpdC to reduce pyruvate decarboxylase activity. From computational modeling we identied 10 residues surrounding the active site for mutagenesis. Ten designs consisting of eight single-point mutants and two double-mutants were selected for exploration. Mutants L546W and T290L showed 5.1% and 22.1% of catalytic eciency on pyruvate, which were then expressed in K. pneumoniae for in vivo test. Isobutanol production by K. pneumoniae T290L was 25% higher than the control strain, and a nal titer of 5.5 g/L isobutanol was obtained with a substrate conversion ratio of 0.16 mol/mol glucose. This research provides a new way to improve the eciency of the biological route of isobutanol production. Enzyme activities of KivD, KP-IpdC, and variants of Kp-IpdC were determined by a coupled enzymatic method. The method was based on the ability of alcohol dehydrogenase, in the presence of NADH, to reduce aldehydes formed from 2-keto acid by decarboxylase. The reaction was measured spectrophotometrically by the decrease in optical density at 340 nm. Pyruvate and 2-ketoisovalerate were used as substrates respectively. The reaction mixture contained 50 mM potassium phosphate, 1 mM MgSO 4 ·7H 2 O, 0.5 mM thiamine pyrophosphate, 0.2 mM NADH, 45 U/ml alcohol dehydrogenase of S. cerevisiae (Sangon Biotech®). The reaction was initiated by adding the substrates. Kinetic data were tted to the Lineweaver-Burk plot, and the parameters such as Km, V max , and Kcat of enzymes were determined from a linear least-squares t.

Indole pyruvate decarboxylases use thiamine diphosphate (ThDP) as a cofactor and require magnesium ion for catalytic activity. ThDP dependent enzymes catalyze a broad range of different reactions involving cleavage and formation of C-C bonds, which are essential in many biosynthetic pathways [19]. The decarboxylase superfamily contains more than ten families of decarboxylases, and their structures are highly conserved [20]. The structures comprise three similarly sized domains: the N-terminal domain which binds the pyrimidine (Pyr) ring of ThDP, a middle domain and the C-terminal domain which binds the diphosphate (PP) moiety. The active site is located at the interface between two monomers, with ThDP interacting with the Pyr domain of one monomer and the PP domain of the second [21].
The objective of the present study was to reveal the mechanism of the decrease of isobutanol production by high level expression of ipdC. Furthermore, the three-dimensional (3D) structure of Kp-IpdC was obtained by homology modeling. Site-directed mutagenesis of Kp-IpdC was carried out to improve the catalytic performance and isobutanol production by K. pneumoniae IpdC.

Strains, plasmids, and primers
Bacterial strains and plasmids used in this study are listed in Table 1. Primers used for PCR are listed in Table S1.
Site-directed mutagenesis of Kp-IpdC and strain construction pMD18-T-ipdC was digested with EcoR I and BamH I to obtain the ipdC fragment, and this fragment was ligated into pET28a to generate pET28a-ipdC. pET28a-ipdC was transformed into E. coli BL21 for protein expression. BL21/ kivD was constructed using the same approach as BL21/ipdC.
Oligonucleotide-directed site-speci c mutagenesis was carried out on expression plasmids of Kp-IpdC variants. pET28a-L546W was constructed based on pET28a-ipdC. Primer pair L546W-s and L546Wa were used to amplify pET28a-L546W with pET28a-ipdC as the template. The PCR product was transformed to E. coli BL21 to get BL21/L546W. Other mutants of ipdC expression strains were constructed using the same approach.
Enzymatic reaction kinetic parameters determination BL21/ipdC, BL21/kivD, and other E. coli strains expressing mutants of ipdC were cultured for enzyme preparation. Cells lysate was prepared by sonication and puri ed enzyme was obtained through a His tag Ni-NTA-Se nose Column (Sangon Biotech®) by following the protocol given by the manufacturer.
Enzyme activities of KivD, KP-IpdC, and variants of Kp-IpdC were determined by a coupled enzymatic method. The method was based on the ability of alcohol dehydrogenase, in the presence of NADH, to reduce aldehydes formed from 2-keto acid by decarboxylase. The reaction was measured spectrophotometrically by the decrease in optical density at 340 nm. Pyruvate and 2-ketoisovalerate were used as substrates respectively. The reaction mixture contained 50 mM potassium phosphate, 1 mM MgSO 4 ·7H 2 O, 0.5 mM thiamine pyrophosphate, 0.2 mM NADH, 45 U/ml alcohol dehydrogenase of S. cerevisiae (Sangon Biotech®). The reaction was initiated by adding the substrates. Kinetic data were tted to the Lineweaver-Burk plot, and the parameters such as Km, V max , and Kcat of enzymes were determined from a linear least-squares t.

Medium and culture condition
The fermentation medium contained 100 g/L glucose, 5 g/L yeast extract, 4 g/L corn steep liquor, 5 g/L (NH 4 ) 2 SO 4 , 3 g/L sodium acetate, 0.4 g/L KCl and 0.1 g/L MgSO 4 . For the seed culture, 250-mL asks containing 50 mL of LB medium were incubated in a rotary shaker at 37 °C and 200 rpm overnight. The seed culture was inoculated into a 5-L bioreactor (BIOSTAT-A plus Sartorius) with a working volume of 3 L. The culture pH was automatically controlled at 7. The air ow rate and agitation were 2 L/min and 300 rpm respectively. The off-gas was fed through a glass condenser, which was immerged in an ice-bath, and the condensate was collected.

Analytical methods
The biomass concentration was evaluated by determination of optical density (OD 600) with a spectrophotometer.
Chemical compounds in the broth were quanti ed by a Shimadzu 20AVP high performance liquid chromatography system (HPLC) (Shimadzu Corp., Kyoto, Japan) equipped with a RID-10A refractive index detector and a SPD-M20A photodiode array detector. An Aminex HPX-87H column (300×7.8 mm) (Bio-Rad, USA) was used and the column temperature was set up at 65 ºC. The mobile phase was 0.005 mol/L H 2 SO 4 solution at a ow rate of 0.8 ml/min.

Homology modeling of Kp-IpdC
The Rosetta software suite is an academically developed framework for protein structure prediction and design. The three-dimensional structure of Kp-IpdC was modeled with RosettaCM [24,25]. From the NCBI database 10 homologs with ³30% sequence identity to Kp-IpdC were selected as templates to predict the structure. 3D structures of these homologous proteins have been solved empirically. A total of 10,000 structure simulations were run and the structure with the lowest Rosetta energy was chosen.

Results
Isobutanol production by K. pneumoniae using Kp-IpdC or KivD as the decarboxylase K. pneumoniae ΔbudA-ΔldhA is an isobutanol production strain constructed previously. Kp-IpdC has been identi ed to catalyze the reaction of isobutyraldehyde formation from 2-ketoisovalerate. KivD is an L. lactis decarboxylase and has been used in all arti cial isobutanol synthesis pathways. K. pneumoniae ΔbudA-ΔldhA-ipdC and K. pneumoniae ΔbudA-ΔldhA-kivD were constructed to compare the difference of the two decarboxylases on isobutanol production by K. pneumoniae. These two strains were batch cultured in 5 L bioreactors and induced by 1 mM IPTG. The fermentation results are shown in Fig. 2.
Cell growth and glucose consumption of the two strains were comparable. Cells were quickly growing in the rst 10 hours of cultivation and cell densities were kept stable in the remaining cultivation time. 80 g/L of glucose was utilised by both two strains after 30 hours of cultivation. 2.5 and 2.9 g/L of isobutanol were produced by K. pneumoniae ΔbudA-ΔldhA-ipdC and K. pneumoniae ΔbudA-ΔldhA-kivD, respectively. 2-Ketoisovalerate was found to be accumulated in the broth of the two strains with titers of 0.5 and 5.5 g/L, respectively. Ethanol generated by the two strains were 7.4 and 6.8 g/L, respectively. 4.0 and 3.4 g/L of acetic acid were produced by the two strains after 10 hours of cultivation. The acetic acid level decreased to 2.7 g/L for K. pneumoniae ΔbudA-ΔldhA-kivD but its nal level was 4.9 g/L for K. pneumoniae ΔbudA-ΔldhA-ipdC. Formate produced by the two strains were 7.1 and 5.7 g/L, respectively. 2,3-Dihydroxyisovalerate accumulated to levels of 5.0 and 6.4 g/L, respectively.
To further investigate the difference of the two decarboxylases on isobutanol production, K. pneumoniae ΔbudA-ΔldhA-ipdC and K. pneumoniae ΔbudA-ΔldhA-kivD were cultured in 5 L bioreactors without induction, and fermentation results are shown in Fig. 3.
In the conditions without induction, cell growth and catabolites production of the two strains were distinctly different. 80 g/L of glucose was exhausted by K. pneumoniae ΔbudA-ΔldhA-ipdC after 27 hours of cultivation, and the highest cell density of 12.1 OD unit was achieved after 24 hours. While glucose was not exhausted by K. pneumoniae ΔbudA-ΔldhA-kivD until 35 hours of cultivation. Isobutanol produced by the two strains was 4.5 g/L and 0.6 g/L, respectively. In contrast to isobutanol, 2-ketoisovalerate accumulated to levels of 1.0 and 11.0 for K. pneumoniae ΔbudA-ΔldhA-ipdC and K. pneumoniae ΔbudA-ΔldhA-kivD, respectively. Ethanol and acetic acid produced by K. pneumoniae ΔbudA-ΔldhA-ipdC were 6.8 g/L and 3.2 g/L, they were 6.0 g/L and 0.5 g/L for K. pneumoniae ΔbudA-ΔldhA-kivD. Formate produced by the two strains were 8.4 and 7.1 g/L, respectively. 2,3-Dihydroxyisovalerate accumulated to levels of 9.0 and 7.7 g/L, respectively.

Determination of kinetic parameters of Kp-IpdC and KivD
Comparing the results of batch culture of K. pneumoniae ΔbudA-ΔldhA-ipdC and K. pneumoniae ΔbudA-ΔldhA-kivD with and without IPTG induction it can be concluded that Kp-IpdC favors isobutanol production at a low expression level while KivD favors isobutanol production at a high expression level. Low level expression of kivD coincided with a high level of 2-ketoisovalerate accumulation. However, high level expression of ipdC does not result in a high level of 2-ketoisovalerate. Thus, high level expression of ipdC might constrain the metabolic ux of 2-ketoisovalerate synthesis. To clarify this hypothesis, the kinetic parameters of the two enzymes were determined in vitro.
Pyruvate is a central metabolite of the cell and the substrate of the rst reaction of the isobutanol synthesis pathway. Indole-3-pyruvate and pyruvate are both keto acids. Thus, we suspected that pyruvate might be a substrate of Kp-IpdC. High level expression of ipdC can led to more pyruvate to be converted to aldehyde and further to form ethanol or acetic acid, which limited the carbon ux of isobutanol synthesis pathway. In vitro enzymatic reaction of 2-ketoisovalerate and pyruvate decarboxylation catalyzed by Kp-IpdC or KivD were performed. Kinetic parameters were calculated (Fig S1, S2) and results were summarized in Table 2.  Table 2). Thus, F388W was not a suitable enzyme to be used for isobutanol production.
The Km of A387L, V542I, A387I+F388W, Q383M, and A387L for pyruvate were all higher than that of the wild-type Kp-IpdC (4.18 mM shown in Table 2). However, the Km of these enzymes for 2-ketoisovalerate were also higher than that of the wild-type Kp-IpdC (1.48 mM shown in Table 2). These enzymes were all eliminated for further investigations.
Variants L546W and T290L showed lower Km values for 2-ketoisovalerate, 1.01 and 1.17 mM respectively, which is lower than that of the wild Kp-IpdC. The Km of these two variants with pyruvate were 693.27 mM and 11.27 mM, respectively. These values were much higher compared to the 3.13 mM of the wild type Kp-IpdC (shown in Table 2). The Kcat/Km values of T290L and L546W with pyruvate were 40.5 M -1 s -1 and 9.61 M -1 s -1 , respectively. These values were much lower than the value of 5445.39 M -1 s -1 of the wild type Kp-IpdC (shown in Table 2).
The 3D structures of the active center of L546W and T290L docked with 2-ketoisovalerate are shown in Fig 5. The native substrate of Kp-IpdC is indole-3-pyruvate and its catalytic pocket is suitable for the native substrate. The molecule size of indole-3-pyruvate is larger than that of 2-ketoisovalerate. Thus, reducing the size of the catalytic pocket would favor 2-ketoisovalerate as a substrate. The threonine at 290 residue was mutated to leucine in T290L. The side chain of leucine is larger and more hydrophobic than that of threonine. This structure had a smaller catalytic pocket and was more suitable for 2-ketoisovalerate-ThDP-Mg 2+ to be bound. The molecule size of pyruvate is smaller than that of 2ketoisovalerate, the catalytic pocket of T290L might not be suitable for the compound of pyruvate-ThDP-Mg 2+ . The leucine at 546 residue was mutated to tryptophan in L546W. The side chain of tryptophan is closer to 2-ketoisovalerate than that of leucine in the catalytic pocket. This made the compound of 2ketoisovalerate-ThDP-Mg 2+ more stable in the catalytic pocket. L546W and T290L have the characteristics of enhanced a nity interaction with 2-ketoisovalerate and reduced a nity with pyruvate.
These two variants were selected to be used in isobutanol production.
Cell growth and glucose consumption of these strains were similar. After 30 hours of cultivation, the 80 g/L of glucose was all utilised by these strains. Similar to K. pneumoniae ΔbudA-ΔldhA-kivD, 12.3 g/L of 2-ketoisovalerate was accumulated in the broth of K. pneumoniae KivD after 30 hours of cultivation. However, 2-ketoisovalerate levels were less than 1 g/L for the other three strains. Isobutanol produced by K. pneumoniae KivD was 1.5 g/L, which was distinctly lower than that of other strains.
Speci cally, 3.9 g/L, 4.1 g/L, and 3.8 g/L of isobutanol were produced by K. pneumoniae IpdC, K. pneumoniae T290L, and K. pneumoniae L546W, respectively, after 28 hours of cultivation. In addition, 10.2 g/L, 7.7 g/L, and 10.1 g/L of ethanol were produced by these strains, respectively. The decrease in isobutanol and ethanol levels towards the end of the cultivation is probably due to evaporation. All strains produced acetic acid, 2,3-Dihydroxyisovalerate and formate as by-products in similar amounts.
Ethanol and acetic acid produced by K. pneumoniae T290L were reduced 24% and 7% compared with that of K. pneumoniae IpdC. This indicated the decarboxylation reaction of pyruvate was reduced in K. pneumoniae T290L. However, there was little increase in isobutanol production by K. pneumoniae T290L compared to K. pneumoniae IpdC.
K. pneumoniae T290L was selected for further investigation. This strain, K. pneumoniae KivD and K. pneumoniae IpdC were cultured without IPTG induction, and the results are shown in Fig. 7.
Cell growth and glucose consumption of K. pneumoniae IpdC were slower than that of the other two strains. 5 g/L of glucose was unused after 32 hours of cultivation. While, glucose was exhausted by other two strains, and which were similar to that of the cultivations with IPTG induction.
High level of 2-ketoisovalerate was accumulated in the broth of K. pneumoniae KivD after 32 hours of cultivation with a titer of 12.5 g/L. This was close to that obtained with IPTG induction. However, isobutanol produced by this strain was only 0.32 g/L compared to 1.5 g/L with induction. Acetic acid, and formate produced by K. pneumoniae KivD were higher than that of the other two strains, with the titers of 4.5 g/L and 9.8 g/L, respectively.
Isobutanol produced by K. pneumoniae IpdC and K. pneumoniae T290L were 4.4 and 5.5 g/L, respectively. These titres were both higher than that obtained with IPTG induction. Ethanol produced by K. pneumoniae IpdC was 6.9 g/L, which was lower than that obtained with IPTG induction. Whereas ethanol produced by K. pneumoniae T290L was 7.8 g/L, which was nearly the same as that with IPTG induction. Acetic acid produced by the two strains were all reused by the cells, with nal titers of 1.7 g/L and 1.4 g/L. Acetic acid titers were all lower than that obtained with IPTG induction. In addition, both strains produced 8.7 g/L of 2,3-Dihydroxyisovalerate and 8.5 g/L and 7.3 g/L of formate were produced by K. pneumoniae IpdC and K. pneumoniae T290L, respectively.
Isobutanol production by K. pneumoniae T290L was improved by 24 % compared to that by K. pneumoniae IpdC. However, more ethanol was produced by K. pneumoniae T290L in comparison to K. pneumoniae IpdC. By contrast, acetic acid and formate produced by K. pneumoniae T290L were decreased compared with K. pneumoniae IpdC. The substrate conversion ratio of glucose to isobutanol obtained by K. pneumoniae T290L was 6.7% (w/w) or 0.16 mol/mol. Isobutanol production by K. pneumoniae IpdC with IPTG induction was lower than that obtained without IPTG induction. K. pneumoniae T290L cultures showed a similar tendency indicating that high level expression of T290L still lead to more pyruvate ux into by-products production.

Discussion
Kp-IpdC is more e cient than KivD in catalysis of 2ketoisovalerate decarboxylation K. pneumoniae has an endogenous isobutanol synthesis pathway, and the structure of this pathway was the same as arti cial isobutanol synthesis pathways constructed in E. coli and other microorganisms [18]. A critical enzyme in the arti cial isobutanol synthesis pathway is 2-keto acid decarboxylase [9], which is common in plants, yeasts, and fungi but less so in bacteria [27]. Kp-IpdC had been identi ed to catalyse the 2-ketoisovalerate decarboxylation reaction in K. pneumoniae. While all arti cial isobutanol synthesis pathways reported using KivD from L. lactis catalyse this decarboxylation reaction [9,[28][29][30][31]. The in vitro experiments results shown in Table 2 indicated the e ciency of Kp-IpdC was higher than that of KivD in catalysis of 2-ketoisovalerate decarboxylation. pDK6 is a high copy vector used for protein expression in K. pneumoniae, this plasmid uses the tac promoter for gene expression. With IPTG induction, the protein expressed would constitute more than 1% of total cellular protein. In the absence of induction, the protein was also expressed to a certain level [22]. Kp-IpdC and KivD were all expressed in K. pneumoniae ΔbudA-ΔldhA without induction of IPTG. Higher level of isobutanol was obtained by K. pneumoniae ΔbudA-ΔldhA-ipdC than that of K. pneumoniae ΔbudA-ΔldhA-kivD (shown in Fig. 3) consistent with the in vitro experimental results. The comparison of isobutanol production by K. pneumoniae KivD and K. pneumoniae IpdC (Fig. 6, 7) showed further clear results. Thus, Kp-IpdC is more e cient than KivD in catalysis of 2-ketoisovalerate decarboxylation. If KivD was replaced by Kp-IpdC in the arti cial isobutanol synthesis pathways, the isobutanol titers might potentially be signi cantly improved.

Catalysis Of Pyruvate Decarboxylation Is A Limitation Of Kpipdc
In the no-induction fermentations, low levels of 2-ketoisovalerate were accumulated in the culture broth of K. pneumoniae ΔbudA-ΔldhA-ipdC (shown in Fig. 3). This indicated the decarboxylation reaction was still a limited step of the isobutanol synthesis pathway. However, a low level of isobutanol was obtained in the induced culture of K. pneumoniae ΔbudA-ΔldhA-ipdC. Furthermore, the 2-ketoisovalerate accumulated in the culture broth of K. pneumoniae ΔbudA-ΔldhA-ipdC was lower than that of K. pneumoniae ΔbudA-ΔldhA-kivD. We can conclude that the total carbon ux of the isobutanol synthesis pathway was more reduced in the IPTG induction conditions compared to the no-induction condition. High levels of ethanol and acetic acid were obtained in the culture broth of K. pneumoniae ΔbudA-ΔldhA-ipdC with IPTG induction, this indicated more pyruvate was converted to ethanol or acetic acid, instead of isobutanol synthesis.
Kp-IpdC and KivD are both ThDP-dependent decarboxylases. However, the substrate range of decarboxylases can be different with some classes, such as pyruvate decarboxylases, benzoylformate decarboxylases and benzaldehyde lyases from bacteria or yeast accepting a broad variety of substrates, including keto acids and aldehydes [19]. The substrates of Ec-IpdC are limited to keto acids. The enzyme has the highest catalytic e ciency to the native substrate indole pyruvate (Km = 20 µM), to 4-Clbenzoylformate and to 4-Br-benzoylformate. Pyruvate is also a substrate of this enzyme, but it has a very low a nity (Km = 3.38 mM) [32]. These data agree with the results obtained in this study, i.e., the Km of Kp-IpdC to pyruvate was found to be 3.31 mM (shown in Table 2).
Pyruvate conversion by Kp-IpdC is a disadvantage for isobutanol production. A high level of Kp-IpdC leads to more pyruvate being decarboxylated to aldehyde and reduces the available pyruvate for isobutanol synthesis. This nding explains the low titer of isobutanol caused by high expression of ipdC (see Fig. 1).
Protein Engineering To Improve The Substrate Speci city Of Kp-ipdc Previously, different variants of decarboxylases have been constructed to alter substrate speci city. For example the I472 residue in the vicinity of the active centre of pyruvate decarboxylase (PDC) from Zymomonas mobilis. I472A enlarges the substrate binding site and allows the decarboxylation of longer aliphatic 2-keto acids (C4-C6) as well as aromatic 2-keto acids besides pyruvate [33]. Benzoylformate decarboxylase (BFD) from Pseudomonas putida favours aromatic 2-keto acids as substrate. The Ala460 residue in BFD is analogous to Ile472 in PDC. This alanine was replaced with isoleucine to obtain BFD A460I. The substrate binding site of BFD A460I was reduced and thus more similar of the wild type of PDC. BFD A460I can use pyruvate as the substrate, while the wild type of BFD is unable to convert it [34].
These successful examples demonstrate the application potential of decarboxylases and feasibility of changing their substrate speci cities by point mutations.
For biosynthesis of the 6-carbon alcohol, 3-methyl-1-pentanol, engineering of KivD with the aim of achieving a higher selectivity toward 2-keto-4-methylhexanoate was performed. The F381L/V461A variant was the best one and produced 384.3 mg/L of 3-methyl-1-pentanol [35]. KivD was designed computationally to enhance its catalytic e ciency with C8 rather than C5 as a substrate. A triple-residues variant G402V/M538L/ F542V showed a 600-fold improvement in speci city for C8 compared to C5 substrates. But the enzyme activities with the two substrates were both decreased [36]. KivD introduced into Synechocystis results in two products, isobutanol and 3-methyl-1-butanol. To reduce the 3-methyl-1butanol level and improve the isobutanol production many variants of KivD were constructed, and V461I/S286T showed the highest (2.4 times) improvement of the isobutanol to 3-methyl-1-butanol molar ratio [37]. Many protein engineering works were performed using decarboxylases, but the activities of variants with non-native substrates were still much lower than that using native substrates. Km of T290L and L546W of Kp-IpdC to 2-ketoisovalerate detected in this study were 1.17 mM and 1.01 mM. They are both higher than the Km of Ec-IpdC to its native substrate indole pyruvate (Km 20 µM) [32]. Isobutanol production by K. pneumoniae T290L was improved compared with K. pneumoniae IpdC. However, isobutanol production by this strain with IPTG induction was still lower than that without IPTG induction, like that of K. pneumoniae IpdC. The pyruvate decarboxylation activity of T290L still effects isobutanol synthesis, and this disadvantage was not erased totally. Thus, there is still a large potential to improve the performance of Kp-IpdC. Conclusion 5.5 g/L of isobutanol was produced by K. pneumoniae T290L in batch culture, which was 25% higher than that of the control strain. A substrate conversion ratio of 0.16 mol/mol was obtained. However, byproducts of this strain still exhibited high levels and the isobutanol production is constrained by undesirable enzyme promiscuity of IpdC towards pyruvate. The protein engineering work showed promising results but there is scope for further improvement. One target could be to reduce the Km of Kp-IpdC to 2-ketoisovalerate to around 20 µM, near that of Ec-IpdC to indole pyruvate, in order to increase the e ciency of the biological route of isobutanol production further.

Declarations
Ethical Approval and Consent to participate Not applicable.   The cell growth and metabolite production of K. pneumoniae ΔbudA-ΔldhA-ipdC and K. pneumoniae ΔbudA-ΔldhA-kivD in the batch culture with IPTG induction. Cells were cultured in 5 L bioreactors and 1 mM of IPTG was added to the culture broth after 8 hours of cultivation. Data points are the average of n = 3; error bars represent standard error about the mean.

Figure 3
The cell growth and metabolite production of K. pneumoniae ΔbudA-ΔldhA -ipdC and K. pneumoniae ΔbudA-ΔldhA-kivD in the batch culture without IPTG induction. Cells were cultured in 5 L bioreactors and no IPTG was added to the culture broth in the process. Data points are the average of n = 3; error bars represent standard error about the mean.  The cell growth and metabolites production of K. pneumoniae IpdC, K. pneumoniae T290L, K. pneumoniae L546W, and K. pneumoniae KivD in the batch cultures with IPTG induction. Cells were cultured in 5 L bioreactors and 1 mM of IPTG was added to the culture broth after 8 hours of cultivation.
Data points are the average of n = 3; error bars represent standard error about the mean.