Enhanced (−)-α-Bisabolol Productivity by Efficient Conversion of Mevalonate in Escherichia coli

(−)-α-Bisabolol, a naturally occurring sesquiterpene alcohol, has been used in pharmaceuticals and cosmetics owing to its beneficial effects on inflammation and skin healing. Previously, we reported the high production of (−)-α-bisabolol by fed-batch fermentation using engineered Escherichia coli (E. coli) expressing the exogenous mevalonate (MVA) pathway genes. The productivity of (−)-α-bisabolol must be improved before industrial application. Here, we report enhancement of initial (−)-α-bisabolol productivity to 3-fold higher than that observed in our previous study. We first harnessed a farnesyl pyrophosphate (FPP)-resistant mevalonate kinase 1 (MvaK1) from an archaeon Methanosarcina mazei (M. mazei) to create a more efficient heterologous MVA pathway that produces (−)-α-bisabolol in the engineered E. coli. The resulting strain produced 1.7-fold higher (−)-α-bisabolol relative to the strain expressing a feedback-inhibitory MvaK1 from Staphylococcus aureus (S. aureus). Next, to efficiently convert accumulated MVA to (−)-α-bisabolol, we additionally overexpressed genes involved in the lower MVA mevalonate pathway in E. coli containing the entire MVA pathway genes. (−)-α-Bisabolol production increased by 1.8-fold with reduction of MVA accumulation, relative to the control strain. Finally, we optimized the fermentation conditions including inducer concentration, aeration and enzymatic cofactor. The strain was able to produce 8.5 g/L of (−)-α-bisabolol with an initial productivity of 0.12 g/L h in the optimal fed-batch fermentation. Thus, the microbial production of (−)-α-bisabolol would be an economically viable bioprocess for its industrial application.

The biological production of (−)-α-bisabolol has been explored using well-studied microbes such as Saccharomyces cerevisiae (S. cerevisiae) [9] and E. coli [12]. Because of the identification of BBS from German chamomile, Matricaria recutita, the microbes expressing the MrBBS enzyme can synthesize an (−)-α-bisabolol as a major terpenoid product [9]. Previously, we engineered an E. coli strain to express the MrBBS enzyme and exogenous MVA pathway. The resulting E. coli produced 9.1 g/L of (−)-α-bisabolol with a productivity of 0.04 g/L h at early stage of fermentation (0-42 h) [12], whereas S. cerevisiae expressing the MrBBS enzyme alone produced 8 mg/L of (−)-α-bisabolol during four days of cultivation [9]. These studies showed the potential of (−)-α-bisabolol production by microbial fermentation. However, productivity remains to be improved for the industrial production of (−)-α-bisabolol using engineered microbes. In our empirical fermentation studies, the initial productivity (0-48 h) of (−)-α-bisabolol was critical to improving its overall productivity, because after 2 days of fermentation, the production rate of (−)-α-bisabolol showed no significant differences among various production strains and fermentation conditions.
In this study, we improve (−)-α-bisabolol productivity in engineered E. coli, which can serve as a promising platform strain for development of an economically feasible bioprocess of (−)-α-bisabolol production. To this end, we first introduced a heterologous MvaK1 from M. mazei that is resistant to FPP feedback inhibition. We then added a copy of the lower MVA pathway genes to the whole MVA pathway for the efficient conversion of MVA to (−)-α-bisabolol. Finally, we optimized the fermentation conditions of the engineered E. coli by tuning the inducer concentrations and aeration for MVA pathway expression and sufficient ATP supply, respectively. Overall, a fed-batch fermentation produced 8.5 g/L of (−)-α-bisabolol with 0.12 g/L h of initial productivity (0-46 h) in the engineered E. coli.

Feedback-Resistant MvaK1
MvaK1 is responsible for the first step of the lower MVA pathway by converting MVA to mevalonate phosphate (MVA 5-P in Figure 1) [10] and is important for the regulation of the entire MVA pathway because it is inhibited by known feedback inhibitors: C5 (IPP and DMAPP), C15 (geranyl pyrophosphate (GPP) and FPP), and longer chain terpenoids [13,14]. FPP is a feedback inhibitor of the widely used Staphylococcus aureus MvaK1 (SaMvaK1) for creating a heterologous MVA pathway [15]. Previously, we have also used SaMvaK1 to produce (−)-α-bisabolol in engineered E. coli [16].
To avoid feedback inhibition of MvaK1 and subsequently improve (−)-α-bisabolol production, we replaced the S. aureus mvak1 gene of the pTSN-Bisa-Sa plasmid with a feedback-resistant mvaK1 gene ( Figure S1) from the versatile methanogen M. mazei, which resulted in a pTSN-Bisa-Mm plasmid. The E. coli DH5α-pTSN-Bisa-Mm strain produced 555 mg/L of (−)-α-bisabolol, which is 1.7-fold higher than that of the E. coli DH5α-pTSN-Bisa-Sa strain ( Figure 2B). This is consistent with MVA accumulation of MmMvaK1 showing 1.7-fold less than that of SaMvaK1 ( Figure 2B, right panel), suggesting that the feedback-resistant MmMvaK1 leads to an increase the MVA utilization efficiency.
Using the E. coli DH5α-pTSN-Bisa-Mm strain, we conducted a fed-batch fermentation by intermittently supplying glycerol ( Figure 2C). Cells were grown exponentially for 24 h and produced 0.9 g/L of (−)-α-bisabolol and 4.9 g/L of MVA along with consuming initially supplied glycerol. A total of 8.2 g/L of (−)-α-bisabolol was yielded with a productivity of 0.06 g/L h, and 10.7 g/L of MVA was accumulated in 140 h. Overall, although feedback-resistant MmMvaK1 was used for (−)-α-bisabolol production, a significant amount of MVA was still accumulated in the fed-batch fermentation. Using the E. coli DH5α-pTSN-Bisa-Mm strain, we conducted a fed-batch fermentation by intermittently supplying glycerol ( Figure 2C). Cells were grown exponentially for 24 h and produced 0.9 g/L of (−)-α-bisabolol and 4.9 g/L of MVA along with consuming initially supplied glycerol. A total of 8.2 g/L of (−)-α-bisabolol was yielded with a productivity of 0.06 g/L h, and 10.7 g/L of MVA was accumulated in 140 h. Overall, although feedback-resistant MmMvaK1 was used for (−)-α-bisabolol production, a significant amount of MVA was still accumulated in the fed-batch fermentation. (B) Improvement of (−)-α-bisabolol production in the engineered E. coli DH5α harboring pTSN-Bisa-Mm compared with the strain expressing pTSN-Bisa-Sa. Cells were grown in TB medium containing 10 g/L glycerol and 20% (v/v) of n-dodecane at 30 °C for 72 h without the addition of isopropyl β-D-1thiogalactopyranoside (IPTG). The error bars represent the standard deviation of the concentrations of (−)-α-bisabolol and MVA from three biological replicates. (C) Fed-batch fermentation of E. coli DH5α harboring pTSN-Bisa-Mm. The fed-batch fermentation was performed in TB medium and 20% (v/v) of n-dodecane using two-phase culture in the absence of IPTG at 30 °C and pH 7.0. After depletion of glycerol initially added, glycerol was fed intermittently into the bioreactor during fermentation. An agitation speed of 280 rpm and an aeration rate of 1 vessel volume per minute (vvm) were maintained throughout the cultivation.

Overexpression of Entire MVA Pathway Genes
Enzymes responsible for (−)-α-bisabolol biosynthesis in E. coli DH5α-pTSN-Bisa-Mm strain are controlled by IPTG-inducible promoters; trc promoter for MrBBS, and ispA genes, and lac promoter for all MVA pathway genes ( Figure 2A). To this end, we explored the effect of IPTG amount on (−)-αbisabolol production and MVA accumulation in batch culture. When 0.025 mM IPTG was used for induction, (−)-α-bisabolol production increased by 1.8-fold along with a 1.4-fold decrease of MVA accumulation compared to those of the control that were not induced by IPTG ( Figure 3A). To scrutinize the effect of IPTG on (−)-α-bisabolol production, a pSEVA231-Bisa-Mm was generated using a medium copy number plasmid, pSEVA231 (pBBR1 ori) ( Figure 3B). Interestingly, the E. coli DH5α-pSEVA231-Bisa-Mm strain produced 926 mg/L of (−)-α-bisabolol without the accumulation of MVA under the induced condition (0.025 mM IPTG), which is 3.7-fold higher than the uninduced condition ( Figure 3B). When the IPTG amount increased up to 0.1 mM, both pTSN-Bisa-Mm (high copy number), and The fed-batch fermentation was performed in TB medium and 20% (v/v) of n-dodecane using two-phase culture in the absence of IPTG at 30 • C and pH 7.0. After depletion of glycerol initially added, glycerol was fed intermittently into the bioreactor during fermentation. An agitation speed of 280 rpm and an aeration rate of 1 vessel volume per minute (vvm) were maintained throughout the cultivation.

Overexpression of Entire MVA Pathway Genes
Enzymes responsible for (−)-α-bisabolol biosynthesis in E. coli DH5α-pTSN-Bisa-Mm strain are controlled by IPTG-inducible promoters; trc promoter for MrBBS, and ispA genes, and lac promoter for all MVA pathway genes ( Figure 2A). To this end, we explored the effect of IPTG amount on (−)-α-bisabolol production and MVA accumulation in batch culture. When 0.025 mM IPTG was used for induction, (−)-α-bisabolol production increased by 1.8-fold along with a 1.4-fold decrease of MVA accumulation compared to those of the control that were not induced by IPTG ( Figure 3A). To scrutinize the effect of IPTG on (−)-α-bisabolol production, a pSEVA231-Bisa-Mm was generated using a medium copy number plasmid, pSEVA231 (pBBR1 ori) ( Figure 3B). Interestingly, the E. coli DH5α-pSEVA231-Bisa-Mm strain produced 926 mg/L of (−)-α-bisabolol without the accumulation of MVA under the induced condition (0.025 mM IPTG), which is 3.7-fold higher than the uninduced condition ( Figure 3B). When the IPTG amount increased up to 0.1 mM, both pTSN-Bisa-Mm (high copy number), and pSEVA231-Bisa-Mm (medium copy number) showed a dramatic decrease (90%) in (−)-α-bisabolol production compared to those in the presence of 0.025 mM IPTG ( Figure 3A,B). pSEVA231-Bisa-Mm (medium copy number) showed a dramatic decrease (90%) in (−)-α-bisabolol production compared to those in the presence of 0.025 mM IPTG ( Figure 3A,B).

Reinforcement of the MVA Pathway
Considering the high accumulation of MVA in the production of (−)-α-bisabolol, we reinforced the whole MVA pathway through the expression of an additional copy of lower MVA pathway genes. A newly generated plasmid, pSSN12Didi-MrBBS-IspA, contains the lower MVA pathway genes (mvaK1, mvaK2, mvaD, idi), MrBBS, and ispA ( Figure 4A). The MvaK1 of pSSN12Didi-MrBBS-IspA plasmid was adopted from Streptococcus pneumoniae (S. pneumoniae), which has a 2.6-fold faster turnover number (kcat) than that of MmMvaK1 [17]. In the absence of IPTG, E. coli DH5α harboring both pTSN-Bisa-Mm and pSSN12Didi-MrBBS-IspA plasmids produced 1.2 g/L of (−)-α-bisabolol and 988 mg/L of MVA ( Figure 4B,C), which are 2.2-fold higher and 1.4-fold lower than those of the E. coli DH5α containing the pTSN-Bisa-Mm plasmid alone, respectively. Because the metabolic flux was changed by the introduction of additional lower MVA pathway genes, we probed the effect of IPTG concentrations on the (−)-α-bisabolol production in the E. coli DH5α containing both pTSN-Bisa-Mm and pSSN12Didi-MrBBS-IspA plasmids. Unlike the results from the E. coli DH5α harboring the pTSN-Bisa-Mm alone ( Figure 3A), production of both (−)-α-bisabolol and MVA decreased as the IPTG concentrations increased ( Figure 4B,C), indicating that enzymes for (−)-α-bisabolol biosynthesis were sufficiently expressed in the absence of IPTG to increase the (−)-α-bisabolol production when the lower MVA pathway was reinforced.

Reinforcement of the MVA Pathway
Considering the high accumulation of MVA in the production of (−)-α-bisabolol, we reinforced the whole MVA pathway through the expression of an additional copy of lower MVA pathway genes. A newly generated plasmid, pSSN12Didi-MrBBS-IspA, contains the lower MVA pathway genes (mvaK1, mvaK2, mvaD, idi), MrBBS, and ispA ( Figure 4A). The MvaK1 of pSSN12Didi-MrBBS-IspA plasmid was adopted from Streptococcus pneumoniae (S. pneumoniae), which has a 2.6-fold faster turnover number (k cat ) than that of MmMvaK1 [17]. In the absence of IPTG, E. coli DH5α harboring both pTSN-Bisa-Mm and pSSN12Didi-MrBBS-IspA plasmids produced 1.2 g/L of (−)-α-bisabolol and 988 mg/L of MVA ( Figure 4B,C), which are 2.2-fold higher and 1.4-fold lower than those of the E. coli DH5α containing the pTSN-Bisa-Mm plasmid alone, respectively. Because the metabolic flux was changed by the introduction of additional lower MVA pathway genes, we probed the effect of IPTG concentrations on the (−)-α-bisabolol production in the E. coli DH5α containing both pTSN-Bisa-Mm and pSSN12Didi-MrBBS-IspA plasmids. Unlike the results from the E. coli DH5α harboring the pTSN-Bisa-Mm alone ( Figure 3A), production of both (−)-α-bisabolol and MVA decreased as the IPTG concentrations increased ( Figure 4B,C), indicating that enzymes for (−)-α-bisabolol biosynthesis were sufficiently expressed in the absence of IPTG to increase the (−)-α-bisabolol production when the lower MVA pathway was reinforced.

Effect of Aeration on (−)-α-Bisabolol Fermentation
We performed a fed-batch fermentation to produce the (−)-α-bisabolol in E. coli DH5α containing both pTSN-Bisa-Mm and pSSN12Didi-MrBBS-IspA plasmids. The yield of (−)-α-bisabolol was improved by 16% in 46 h compared to when the lower MVA pathway was not additionally overexpressed. However, the MVA still accumulated from the beginning of fermentation and reached 10.1 g/L in 68 h at 280 rpm despite reinforcing the lower MVA pathway ( Figure 6A). It seems that there are other bottlenecks when MVA is converted to (−)-α-bisabolol through the lower MVA pathway. The synthetic MVA pathway requires 3 moles of ATP to convert MVA to (−)-α-bisabolol ( Figure 1) and competes for the ATP with other essential cellular reactions involved in cell growth [18]. Because ATPs are efficiently generated under aerobic conditions using NADHs in oxidative phosphorylation [19], aeration effects were examined by controlling the agitation speed in fed-batch fermentation.
In fed-batch fermentation at an agitation speed of 1,000 rpm, the E. coli strain harboring pTSN-Bisa-Mm and pSSN12Didi-MrBBS-IspA rapidly grew and reached the maximum cell growth within 23 h showing a 1.8-fold improved cell growth relative to those at 280 rpm. As expected, MVA accumulation was considerably reduced, but (−)-α-bisabolol production was not significantly improved ( Figure 6B). It seems that acetyl-CoA was utilized for cell growth and other metabolism, or metabolites downstream of MVA, accumulated. To overcome this problem, we sought to enhance the

Effect of Aeration on (−)-α-Bisabolol Fermentation
We performed a fed-batch fermentation to produce the (−)-α-bisabolol in E. coli DH5α containing both pTSN-Bisa-Mm and pSSN12Didi-MrBBS-IspA plasmids. The yield of (−)-α-bisabolol was improved by 16% in 46 h compared to when the lower MVA pathway was not additionally overexpressed. However, the MVA still accumulated from the beginning of fermentation and reached 10.1 g/L in 68 h at 280 rpm despite reinforcing the lower MVA pathway ( Figure 6A). It seems that there are other bottlenecks when MVA is converted to (−)-α-bisabolol through the lower MVA pathway. The synthetic MVA pathway requires 3 moles of ATP to convert MVA to (−)-α-bisabolol ( Figure 1) and competes for the ATP with other essential cellular reactions involved in cell growth [18]. Because ATPs are efficiently generated under aerobic conditions using NADHs in oxidative phosphorylation [19], aeration effects were examined by controlling the agitation speed in fed-batch fermentation.
IPTG at a concentration of 0.025 mM after 6 h of incubation. Similar to the culture in the absence of IPTG, cells rapidly grew for 22 h, but MVA did not accumulate during fermentation, supporting the lower pathway was intensified by IPTG addition. The yield and productivity of (−)-α-bisabolol were improved by 40% and 56%, respectively, as compared to the absence of IPTG, and the final titer of (−)α-bisabolol reached 8.5 g/L ( Figure 6C). In fed-batch fermentation at an agitation speed of 1000 rpm, the E. coli strain harboring pTSN-Bisa-Mm and pSSN12Didi-MrBBS-IspA rapidly grew and reached the maximum cell growth within 23 h showing a 1.8-fold improved cell growth relative to those at 280 rpm. As expected, MVA accumulation was considerably reduced, but (−)-α-bisabolol production was not significantly improved ( Figure 6B). It seems that acetyl-CoA was utilized for cell growth and other metabolism, Catalysts 2019, 9, 432 9 of 15 or metabolites downstream of MVA, accumulated. To overcome this problem, we sought to enhance the overall (−)-α-bisabolol flux through overexpressing all genes in the (−)-α-bisabolol biosynthetic pathway by the addition of IPTG. We carried out the fed-batch fermentation at 1000 rpm and supplied IPTG at a concentration of 0.025 mM after 6 h of incubation. Similar to the culture in the absence of IPTG, cells rapidly grew for 22 h, but MVA did not accumulate during fermentation, supporting the lower pathway was intensified by IPTG addition. The yield and productivity of (−)-α-bisabolol were improved by 40% and 56%, respectively, as compared to the absence of IPTG, and the final titer of (−)-α-bisabolol reached 8.5 g/L ( Figure 6C).

Discussion
In this study, E. coli was engineered for the efficient conversion of MVA to (−)-α-bisabolol using the feedback-resistant MvaK1 and reinforcement of the lower MVA pathway. The feedback-resistant MvaK1 was firstly identified in the archaeon M. mazei. However, only a handful of studies have been carried out on terpenoid production in microbes. Recently, feedback-resistant MvaK1 enzymes were identified and characterized from Methanosaeta concilii (McMvaK) and Methanocella paludicola (MpMvaK) [20]. The McMvaK and MpMvaK enzymes not only showed feedback resistance to DMAPP, IPP, GPP, and FPP, but also exhibited 4.9-and 5.5-fold higher affinity to MVA, respectively, than MmMvaK1 [20]. Therefore, these MvaK1 enzymes may enable the enhancement of (−)-α-bisabolol production in engineered E. coli.
To find the optimal conditions to efficiently convert MVA to (−)-α-bisabolol, we examined the inducer concentrations, cofactor, ATP, and reinforcement of the lower MVA pathway. Adding the inducers for overexpression of MVA or MEP pathway genes has been a controversial issue in the microbial production of terpenoids [21][22][23]. Lycopene production was reduced under all IPTG-induced conditions in E. coli expressing the lower MVA pathway [23]. The leaky expression of all enzymes involved in the (−)-α-bisabolol production without IPTG addition exhibited the highest production among all tested IPTG concentrations [12]. The IPTG-induced overexpression of genes for (−)-α-bisabolol production can inhibit the essential cellular metabolism due to a deficiency of FPP or accumulation of toxic intermediates (IPP and HMG-CoA) of the heterologous MVA pathway [12]. In contrast, isoprene production increased as the IPTG concentration increased from 0.2 to 1.2 mM [24]. In this study, a small amount of IPTG was effective to increase (−)-α-bisabolol production in the engineered E. coli. Concerning the complex regulation of the MVA pathway, balancing the expression of multiple heterologous enzymes is crucial for the optimal production of (−)-α-bisabolol [10].
The availability of reducing cofactors such as NADH and NADPH strongly affects the yield and productivity of terpenoids in bacteria. The strengthening of the reducing power for the increased production of terpenoids has been attempted; modulation of glutamate dehydrogenase increased the production of β-carotene and lycopene through the increased supply of NADPH [25][26][27]. The overexpression of GAPDH of C. acetobutylicum also resulted in the improvement of isoprene production [24]. Moreover, the replacement of NAD + -dependent GADPH of E. coli with the NADP + -dependent GAPDH of C. acetobutylicum showed a 2.5-fold increase of lycopene productivity in the engineered E. coli [28]. We performed the fed-batch fermentation using the strain overexpressing gapC from C. acetobutylicum and repressing the gapA gene through the CRISPRi system under optimized conditions (1000 rpm and the addition of IPTG at a concentration of 0.025 mM) ( Figure S2, Table 1). Contrary to the results in batch fermentations, both strains showed a negative effect on (−)-α-bisabolol production and had slightly reduced cell growth when compared with the strain that did not overexpress gapC and repress gapA. It appears that the overexpression of gapC and the repression of gapA were not effective when sufficient amounts of ATP and NADPH were supplied, owing to the activation of the citric acid cycle and respiration by increasing oxygen in the cells. This observation might be consistent with a previous study that lycopene production in E. coli was improved by decreasing the pentose phosphate pathway flux and increasing the tricarboxylic acid (TCA) cycle flux [26]. Additionally, NADPH has been shown to inhibit 3-hydroxy-3-methylglutaryl-CoA reductase (HMGR), which converts HMG-CoA to MVA [10]. Both the overexpression of gapC and aerobic condition might lead to an accumulation of excess NADPH, thereby repressing HMGR and causing a flux imbalance. As a result, overall (−)-α-bisabolol production decreased.
Results of fed-batch fermentations conducted in this study were summarized in Table 1. Compared with the previous report [12], the final titer of (−)-α-bisabolol (8.5 g/L) is similar in fed-batch fermentation using E. coli expressing pTSN-Bisa-Mm and pSSN12Didi-MrBBS-IspA at an agitation speed of 1000 rpm with IPTG added at a concentration of 0.025 mM. In particular, 5.5 g/L of (−)-α-bisabolol was obtained within two days, indicating that the productivity was improved by 3-fold compared to previous research. In result of fed-batch fermentations, (−)-α-bisabolol was continuously produced after cell growth ceased. Therefore, recycling resting cells is a promising strategy to further improve (−)-α-bisabolol productivity. Although promising results in productivity were obtained in this study, it is necessary to improve the final (−)-α-bisabolol titer for industrial applications. To achieve this, high-cell density culture experiments using living cells to continuously supply cofactors and enzymes should be conducted. Moreover, metabolic modeling of the system used in this study might provide further insight into bottlenecks for (−)-α-bisabolol production. Table 1. Summary of fed-batch fermentations of (−)-α-bisabolol by engineered E. coli.

Plasmid Construction
The plasmids and primers used in this study are listed in Table 2 and Table S1, respectively. Standard molecular biological techniques including genomic DNA preparation, restriction digestions of DNA, plasmid transformation were performed as previously described [29]. T4 DNA ligase, and all restriction enzymes were obtained from New England Biolabs (NEB, Ipswich, MA, USA). Polymerase chain reaction (PCR) was carried out following the manufacturer's protocols with a high fidelity KOD-Plus-Neo polymerase (Toyobo, Osaka, Japan). Kits for plasmid preparation and gel extraction were purchased from Promega (Madison, WI, USA) and oligonucleotide synthesis were conducted by Bioneer (Daejeon, Korea). Table 2. Strains and plasmids used in this study.

Name
Description References
For the construction of pSEVA231-Bisa-Mm, the first fragment containing the MVA pathway gene, ispA and MrBBS were amplified in pTSN-Bisa-Mm using primers of pBBR1-IF and pBBR1-IR. The second fragment harboring the kanamycin-resistant gene and pBBR1 origin was amplified from pSEVA231 as a template using pBBR1-VF and pBBR1-VR primers. The fragments were assembled via Gibson Assembly method.
To construct pSSN12Didi-MrBBS-IspA, the MrBBS and ispA including trc promoter gene were amplified using Didi-I-F and Didi-I-R primers from the pTSN-Bisa-Mm plasmid. The vector backbone containing genes encoding enzymes of the lower MVA pathway was amplified using Didi-V-F and Didi-V-R primers from the pSSN12Didi plasmid. Two amplified fragments were then assembled via the Gibson Assembly kit.
The E. coli codon-optimized gapC gene from C. acetobutylicum (GenBank accession number: NP_347346) including the ribosome binding site and SpeI/XbaI restriction enzyme sites was synthesized by Macrogen (Seoul, Korea). The synthesized DNA was then digested with SpeI/XbaI, and the fragment containing the gapC gene was gel-purified. The other fragment was prepared by digesting the pTSN-Bisa-Mm plasmid with XbaI. The two fragments were then ligated byg T4 DNA ligase, which created the plasmid pTSN-Bisa-Mm-GapC.
We used the primers of gapA-gRNA-F and gapA-gRNA-R for amplification of the whole pSECRi plasmid by PCR. The amplified DNA fragment was gel-purified and treated with T4 polynucleotide kinase to phosphorylate it. T4 DNA ligase was used ligate the PCR product. The sequences of all genes associated with the (−)-α-bisabolol biosynthetic pathway were verified by Sanger sequencing (Magcogen).

Batch and Fed-Batch Fermentation
To prepare the pre-culture, recombinant E. coli was cultured in 5 mL of LB medium supplied with appropriate antibiotics at 30 • C and 200 rpm overnight. The batch fermentation was carried out by inoculating 1% (v/v) of the pre-culture into 3 mL of the TB medium with 1% (w/v) of glycerol in a 50 mL mini-bioreactor (SPL Life Sciences, Gyeonggi-do, Korea). 20% (v/v) of n-dodecane was overlaid to extract (−)-α-bisabolol from all fermentation broths. The cultures were incubated at 30 • C and 200 rpm for 72 h. For fed-batch fermentation, the pre-culture was prepared in 5 mL of TB medium with

(−)-α-Bisabolol Quantification
Extraction of (−)-α-bisabolol proceeded in the n-dodecane phase which is initially added to the culture broth throughout the cultivation. The overlaid n-dodecane phase was collected after the pellet, supernatant, and a layer of n-dodecane were fractionated from the culture broth using centrifugation at 13,000 rpm for 3 min. Subsequently, the collected n-dodecane was analyzed for the determination of (−)-α-bisabolol concentration using a gas chromatograph (GC, 7890B, Agilent, SC, USA) which is supplied with a flame ionization detector (FID) with HP-5 column (30 m × 0.320 mm × 0.25 µm, Agilent, SC, USA). As the carrier gas, helium was used at a flow rate of 1 mL/min. Temperatures of an injector and an FID were maintained at 240 • C and 250 • C, respectively. The programmed temperature gradients controlled the column temperature: isotherm at 60 • C for 2 min; increase at a rate of 5 • C/min to 200 • C; isotherm at 200 • C for 2 min; increase at 50 • C/min to 300 • C; and isotherm at 300 • C for 5 min. For the generation of a standard curve, (−)-α-bisabolol was purchased from Sigma-Aldrich. In the GC analysis, there was a peak at 21.7 min in the n-dodecane phase sample of recombinant E. coli as a major peak (>95%) except for a peak of n-dodecane (11.5 min). The peak at 21.7 min corresponded to the standard (−)-α-bisabolol compound dissolved in n-dodecane. The (−)-α-bisabolol concentration produced was determined as follows: (−)-α-Bisabolol (g/L) = ((−)-α-Bisabolol in n-dodecane) × (Volume of n-dodecane) Volume of medium (1)

Determination of Cell Growth and Metabolites
Cell growth was monitored by measuring the absorbance at 600 nm (OD 600 ) using a spectrophotometer (Ultrospec 8000, GE Healthcare, Uppsala, Sweden). After the centrifugation of the culture broth at 13,000 rpm for 3 min, the overlaid n-dodecane phase was removed, and the remaining supernatant was used for analyzing metabolite concentrations. The concentrations of glycerol, acetate, and MVA were measured by high-performance liquid chromatography (HPLC, Agilent Technologies 1200 series) equipped with a refractive index detector (RID) with an Aminex HPX-87H column (1300 mm × 7.8 mm, Bio-Rad, Hercules, CA, USA). The column was eluted with 4 mM of sulfuric acid at a flow rate of 0.5 mL/min at 50 • C. All reagents for the standard solution were purchased from Sigma-Aldrich.

Conclusions
We improved (−)-α-bisabolol productivity from engineered E. coli, which can serve as a promising platform strain for the microbial production of (−)-α-bisabolol at an industrial scale. Metabolic engineering strategies used in this study, including feedback-resistance of MvaK1 enzyme, reinforcement of lower MVA pathway flux, balance of the NADPH and ATP pools, and optimization of fermentation, could be applied to enhance the terpenoid production from engineered microbes. Moreover, metabolic modeling based on genome-wide omics data might provide clues to identify unknown bottlenecks and interpret the results. This experiment will be conducted as a further study.