Artificial cell factory design for shikimate production in Escherichia coli

Abstract Shikimate is a key intermediate in high demand for synthesizing valuable antiviral drugs, such as the anti-influenza drug and oseltamivir (Tamiflu®). Microbial-based shikimate production strategies have been developed to overcome the unstable and expensive supply of shikimate derived from traditional plant extraction processes. Although shikimate biosynthesis has been reported in several engineered bacterial species, the shikimate production yield is still unsatisfactory. This study designed an Escherichia coli cell factory and optimized the fed-batch culture process to achieve a high titer of shikimate production. Using the previously constructed dehydroshikimate (DHS)-overproducing E. coli strain, two genes (aroK and aroL) responsible for converting shikimate to the next step were disrupted to facilitate shikimate accumulation. The genes with negative effects on shikimate biosynthesis, including tyrR, ptsG, and pykA, were disrupted. In contrast, several shikimate biosynthetic pathway genes, including aroB, aroD, aroF, aroG, and aroE, were overexpressed to maximize the glucose uptake and intermediate flux. The shiA involved in shikimate transport was disrupted, and the tktA involved in the accumulation of both PEP and E4P was overexpressed. The rationally designed shikimate-overproducing E. coli strain grown in an optimized medium produced approximately 101 g/l of shikimate in 7-l fed-batch fermentation, which is the highest level of shikimate production reported thus far. Overall, rational cell factory design and culture process optimization for microbial-based shikimate production will play a key role in complementing traditional plant-derived shikimate production processes.


Introduction
Shikimate is a key metabolite involved in the aromatic biosynthetic pathway present in most plants and microorganisms, and an important metabolite used as a precursor for the synthesis of many valuable bioactive compounds. The metabolite is a precursor for the chemical synthesis of the neuraminidase inhibitor oseltamivir phosphate (Tamiflu®). This drug is used for the treatment of infection with diverse seasonal influenza viruses, including influenza A and B viruses, the avian influenza virus H5N1, and the human influenza virus H1N1 (Krämer et al., 2003;Dharan et al., 2009;M. Estevez & J. Estevez, 2012;Martinez et al., 2015;Diaz-Quiroz et al., 2018). Shikimate is also a key chiral starting material to many chemical substances used in the chemical and cosmic industries (Ghosh et al., 2012;Rawat et al., 2013).
In the past few decades, the metabolic engineering approach for the overproduction of aromatic amino acids has been reported through the microbial-based shikimate biosynthesis pathway engineering (Kramer et al., 2003;Cortes-Tolalpa et al., 2014;Kruyer & Peralta-Yahya, 2017;Lee & Wendisch, 2017;Averesch & Kromer, 2018;Cao et al., 2020). Most studies optimizing the shikimate pathway focused on Escherichia coli. For example, several attempts have been made to solve DAHP inhibitory feedback to increase the production of aromatic compounds (Pittard & Yang, 2008;Wang et al., 2013). Several engineering strategies were also reported to reduce the consumption of the major precursors, PEP and E4P (Escalante et al., 2010;Rodriguez et al., 2013;Cortes-Tolalpa et al., 2014;Rodriguez et al., 2014;Martínez et al., 2015;Lee et al., 2018). Transcriptome analysis was performed to analyze gene expression in shikimate-producing strains under carbonlimited or phosphate-limited conditions (Johansson et al., 2005;Johansson & Liden, 2006). The highest yield of shikimate in E. coli thus far was approximately 71 g/l shikimate with a yield of 0.27 mol/mol glucose in a 1-l fed-batch culture (Chandran et al., 2003).

Construction of Plasmids and Strains
All constructed plasmids are listed in Table 1, and all primer pairs used in this study are shown in Supplementary Table 1. The suicide vector pKOV (Addgene, USA), which has the sacB gene for providing a markerless system, was used to disrupt the aroL and aroK gene.
For gene overexpression, the plasmids of pPoppA-aroE, pPoppA-aroE-tktA, and pPoppA-aroE-ydiB were constructed. The pPoppA plasmid was amplified from pMESK4, which contained the oppA promoter and used as a backbone. The genes of interest were amplified and cloned into the pPoppA plasmid digested with NdeI for aroE and HindIII for tktA and ydiB.

Fed-batch Fermentation
The first growth culture was carried out in a conical tube containing 5 ml of LBG medium (30°C and 200 rpm). After 12 h of culture, a 250-ml baffled flask containing 20 ml of culture medium was inoculated with 0.2 ml of the culture broth and cultured at 30°C and 200 rpm in an incubator. After 6 h, a 7-l fermenter (1% v/v inoculum) was inoculated with the second culture broth. The production culture was carried out using the PB4-md5 medium in a 7-l fermenter (2 l working volume). The PB4-md5 medium included the following: 30 g/l glucose, 10 g/l glycerol, 15.75 g/l yeast extract, 21.375 g/l tryptone, 5.25 g/l KH 2 PO 4 , 1 g/l MgSO 4 •7H 2 O, 0.8 g/l citric acid, 1 ml/l trace metals, and 200 μg/l thiamine hydrochloride. The feeding medium included 600 g/l glucose, 100 g/l yeast extract, 20 g/l MgSO 4 •7H 2 O, and 5 ml/l trace metal. Phosphate was not added to the feeding medium to allow for the regulation of cell growth. The pH was 7.0 in each culture (10N NaOH, 3M HCl), and the DO level was maintained above 30% by controlling the agitation, aeration, and feeding rates. The configuration of the fermenter was as follows: two impellers with a turbine and marine impeller, ring-type sparger with 12 holes, top-driven, and 160 mm tank diameter. The feeding medium was supplied using a peristaltic pump when glucose was depleted (Choi et al., 2019).

Shikimate and DHS Analyses
The E. coli cells were removed from the cultures by centrifugation, and the culture broth was purified using a membrane filter. The metabolites were separated by high-performance liquid chromatography (HPLC) using an Aminex HPX-87H column. The mobile phase was 2.5 mM H 2 SO 4 , and a flow rate of 0.5 ml/min was used for the metabolites. The column was heated to 50°C to detect the metabolites. Shikimate and DHS were analyzed at wavelengths of 215 nm and 236 nm, respectively. Organic acids, including succinic acid, acetic acid, formic acid, and lactic acid, were detected at 210 nm.

Redesign of Shikimate Biosynthetic Pathway in Escherichia coli
To construct a strain capable of accumulating shikimate, an attempt was made to redesign a strain using the previously constructed DHS-overproducing E. coli strain Inha 103 (Choi et al., 2019). As shown in Fig. 1, DHS is another key metabolite present as an immediate precursor of shikimate in the pathway. Inha 103 was constructed rationally to maximize DHS accumulation through the DHS biosynthetic pathway engineering. The tyrR was deleted to inhibit the feedback inhibition of the DAPH synthase isozymes encoded by aroG and aroF by tyrosine. Phosphoenolpyruvate phosphotransferase (encoded by ptsG), which plays a role in converting PEP to pyruvate, was deleted. Pyruvate kinase 2 (encoded by pykA) was also deleted to inhibit the conversion of PEP to pyruvate through glycolysis. Several key DHS biosynthetic pathway genes, including aroG, aroF, aroB, and aroD, were overexpressed through chromosomal integration into the lacI location. The D-galactose transporter (encoded by galP) and PEP synthase (encoded by ppsA) genes were also inserted to maximize DHS overproduction.
In addition to the NADP-specific AroE, there is an additional AroE-like shikimate dehydrogenase called YdiB. YdiB plays a role as a dual-specificity quinate/shikimate dehydrogenase that can use either NAD or NADP as cofactors (Michel et al., 2003;Garcia et al., 2017). The expression of ydiB was increased significantly under carbon-limiting conditions, and it is involved in the catabolism of shikimate (Johansson & Liden, 2006). To check the putative function of YdiB in shikimate production, a strain called Inha 214 was constructed, in which the ydiB gene was deleted. Although the growth pattern was compatible with other strains, the Inha 214 strain failed to produce shikimate under the same culture conditions (Fig. 2, Supplementary Fig. S1). Interestingly, the strain Inha 217, overexpressing the ydiB gene, accumulated approximately 1.62 g/l of shikimate, suggesting that the AroE and YdiB enzymes might act competitively to influence the conversion of DHS to shikimate. The role of shikimate dehydrogenase by AroE is believed to be the most relevant factor in these strains and culture conditions.

Modification of shiA, ydiN, tktA, and pta-ackA for Shikimate Overproduction
shiA, one of two putative transporter genes in E. coli, plays a role in importing shikimate and DHS into the cell. The other transporter is encoded by ydiN, even though its precise function is unknown (Knop et al., 2001;Martinez et al., 2015). Shikimate has been proposed to be carried in-and-out of the cell by ShiA and YdiN depending on the C-limitation and P-limitation conditions (Kanehisa & Goto, 2000;Johansson & Liden, 2006). The fed-batch culture showed that the production of shikimate decreased toward the later stages of culture ( Supplementary Fig. S2), suggesting that overproduced shikimate might be reintroduced into the cell for recycling. To determine if the deletion of shiA or ydiN affected the production of shikimate, these two genes were removed from Inha 213 to construct the Inha 215 and Inha 216 strains, respectively. The production yields of shikimate in Inha 215 and Inha 216 strains were approximately 1.1 and 1.3 times higher than that of the parental strain Inha 213, respectively (Fig. 2).
Many studies have reported that the overexpression of the tktA increased the production of metabolites in the shikimate pathway through the accumulation of E4P and PEP, which are important precursors of shikimate biosynthesis; Knop et al., 2001;Escalante et al., 2010). Therefore, the expression vector containing both aroE and tktA was constructed under a strong OppA promoter (Choi et al., 2019), which was then expressed in several shikimateproducing E. coli strains stated above (Table 1). When tktA was overexpressed in the shiA-deleted strain (Inha 219), the shikimate yield was 4.0 g/l, which was 1.4 and 1.1 times higher than that of Inha 212 and Inha 213, respectively. Moreover, the Inha 221 strain, in which the tktA was overexpressed in the shiA and ydiN double knock-out strain, produced 4.10 g/l shikimate, showing the highest production of shikimate among the strains tested. Interestingly, when the Inha 221 strain was cultured in a 7-l fed-batch fermentation, the level of shikimate production was not higher than that from Inha 219 (see below, Supplementary Fig. S3).
Two enzymes, phosphate acetyltransferase (encoded by pta) and acetate kinase (encoded by ackA) are believed to be involved in acetate formation (Fig. 1). Comparative metabolite analysis showed that a small amount of acetate accumulated as a byproduct in the Inha 219 strain at 72 h in the late culture period ( Supplementary Fig. S5). Although these two genes were deleted from the chromosome (called Inha 222), the shikimate production titer was not improved significantly (data not shown).

Fed-batch Fermentation for the aroE-Overexpressing Shikimate Producer
AroE plays a critical role in the metabolic processing of DHS to shikimate. Hence, an inactive chromosomal aroE was replaced with an E. coli K12-derived active aroE in the chromosome of Inha 209 along with a plasmid-driven functional aroE (named Inha 223). Inha 209 and Inha 223 produced 3.24 and 3.73 g/l of shikimate, respectively, suggesting that the simultaneous overexpression of aroE from the chromosome and the plasmid is beneficial for shikimate overproduction. Moreover, the Inha 224 strain, overexpressing both aroE and tktA genes in the plasmid, showed a further increase in shikimate production to 4.39 g/l, which is the highest production level among the strains tested in a miniature culture system. Accordingly, 7-l fed-batch fermentation was performed using the highest shikimate-producing strain of Inha 224. It exhibited the maximum cell dry weight of approximately 28 g/l and a shikimate level of approximately 101 g/l at 102 h of culture (Fig. 3b). The Inha 224 strain showed a slightly higher DCW than the Inha 219 strain, but the maximum yield of shikimate in Inha 224 was increased by approximately 8%. The specific production yield (Y p/s ) was 0.33 g shikimate/g glucose for the strain Inha 213, 0.34 g shikimate/g glucose for the Inha 213 applying an optimized process, 0.48 g shikimate/g glucose for the Inha 219 strain, and 0.47g shikimate/g glucose for the Inha 224 strain (Table 2). Comparing to the Inha 213 strain, the shikimate yield from the Inha 224 strain was improved 1.42 times to 101 g/l, which is the highest level of shikimate production reported for E. coli.

Discussion
In this study, shikimate-overproducing E. coli was generated successfully by genetic and metabolic engineering approaches. A strain was constructed to produce shikimate based on a previously designed DHS-overproducing strain that strengthened the biosynthetic pathway of shikimate, resisted feedback inhibition, increased glucose uptake, and induced major precursor accumulation. Furthermore, the shikimate kinase genes, aroL and aroK, were deleted, and the functional aroE gene encoding shikimate dehydrogenase to convert from DHS to shikimate was inserted. YdiB, which encodes shikimate dehydrogenase, such as AroE in E. coli, has dual specificity of quinate/shikimate dehydrogenase using NAD or NADP as a cofactor, unlike the NADP-specific AroE (Michel et al., 2003). One study reported that the molar yield of quinate, a by-product of shikimate biosynthesis, decreased when YdiB was inactivated in the PTS-strain of E. coli, and shikimate production decreased slightly. In contrast, when YdiB was overexpressed, quinate production increased 150% mol/mol compared to shikimate (Garcia et al., 2017). In the present study, shikimate was barely produced when the ydiB gene was deleted. When ydiB was overexpressed, the shikimate productivity was 55% lower than that of the parental strain. Johansson and Lidén (Johansson & Liden, 2006) reported the overexpression of ydiB with respect to aroE in a global microarray-based transcriptomic analysis. They suggested that YdiB, rather than AroE, catalyzes the oxidation of shikimate to DHS in the shikimate-producing E. coli W3110.shik strain grown under carbon-limiting and nonlimiting conditions. Therefore, shikimate production might be affected by the competitive shikimate dehydrogenase role of YdiB and AroE, depending on the culture conditions. This suggests that the NADP-specific AroE should act as the main shikimate dehydrogenase. On the other hand, this effect is different in the state of limited energy and carbon in the cell depending on the expression of YdiB, which requires a broad cofactor. Confirming the production of other intermediate products, such as DHQ and quinate, during the  biosynthesis process is important from various viewpoints, such as the flow of energy in the cell or the limits of carbon or nitrogen in the medium. In this study, an additional increase in the shikimate level was confirmed by deleting ShiA, a shikimate transporter. The Inha 219 strain, which simultaneously overexpressed tktA and aroE, showed higher productivity of shikimate than when shiA was deleted alone. When shikimate enters the cell to be used as a carbon source, it causes catabolite repression and induces the accumulation of intermediates, such as quinate and gallate (Knop et al., 2001;Chandran et al., 2003;Kramer et al., 2003). Therefore, the strategy of inactivation of the shikimate transporter was used to reduce the accumulation of by-products. In addition, the transport of shikimate by ShiA can be bidirectional, and that shikimate can be transported by other transporters, such as YdiN (Johansson et al., 2005). In this study, the shikimate production efficiency could be improved by preventing the re-inflow of shikimate from outside of the cells in small-scale cultivation through the deletion of shiA or ydiN, or both, and allowing the biosynthesis of shikimate from glucose. In addition, a strategy to improve the E4P pool for the synthesis of DAHP through overexpression of the tktA gene encoding transketolase I has proven its effectiveness through many studies (Draths et al., 1999;Chandran et al., 2003;Escalante et al., 2010;Rodriguez et al., 2013). In the present study, the resulting strain Inha 219, which has a shiAdeletion and tktA-overexpression, produced 93 g/l of shikimate at a yield of 0.48 g/g after 78 h in 7-l fed-batch fermentation with the modified medium. In addition, the active AroE expressing Inha 224 strain produced 101 g/l of shikimate at a yield of 0.48 g/g after 102 h in 7-l fed-batch fermentation. To the best of the authors' knowledge, this is the highest shikimate titer and yield obtained using a microbial catalyst reported so far. The transport of shikimate into the cell may be repressed by the glucose catabolism, resulting in fewer by-products formed under C-rich conditions (Knop et al., 2001). Such a remarkable production yield was achieved by applying several metabolic engineering strategies for shikimate production in E. coli strains. The DCW of Inha 219 was lower than that of Inha 224, but the shikimate production yield was high, suggesting that there is a competitive relationship between cell growth and shikimate production. Although shikimate is a primary metabolite, its production in engineered strains exhibited typical secondary metabolite patterns. In aromatic amino acid production processes, as shown previously in the production of DHS and MA, E. coli and Corynebacterium glutamicum have characteristic metabolites that suppress the cell density at certain concentrations (Cheng et al., 2012;Rodriguez et al., 2014;Lee et al., 2018).
Studies on the production of shikimate from various microorganisms other than E. coli have also been reported. Engineered C. glutamicum strains show the highest shikimate productivity after E. coli. After constructing the genetic modules of the aroG (NCgl2098), aroB (NCgl1559), aroD (NCgl0408), and aroE (NCgl1567) genes based on the selected RBS libraries, C. glutamicum RES167 aroK carrying an efficient genetic module produced 7.4 and 11.3 g/l of shikimate during 5-l batch and fed-batch fermentation, respectively (Zhang et al., 2015). The CRISPRi system has been used to regulate C. glutamicum gene expression at the transcriptional level (Zhang et al., 2016). The titers of shikimate increased from 115% to 7.76 g/l in 250 ml flasks and 23.8 g/l in a 5-l fermenter. The highest productivity from the SKM7 strain (C. glutamicum aroK qsuB qsuD glks: glk1 glk2 ppgk ptsH hpdA overexpressed iolT1 glks gapA tkt tal aroGFBR, aroB, aroD, aroE) was 97.1 g/l shikimate at a 30.1% yield (mol/mol) after 80 h in fed-batch fermentation experiments in rich medium without the switch to a minimal medium (Kogure et al., 2016). The highest yield of 141 g/l shikimate (51% mol/mol) was obtained by the growth-arrested cell reaction method.
Genetically manipulated Bacillus subtilis (Ghosh & Banerjee, 2015;Iomantas et al., 2002) and Citrobacter freundii strains (Shirai et al., 2001) have been exploited successfully to produce shikimate, but the titers did not exceed 20 g/l. A shikimate kinase (aroI)-inactivated B. subtilis was shown to produce 8.5 g/l shikimate together with 9.5 g/l of DHS (Ghosh & Banerjee, 2015). Many studies also have produced various aromatics and derived compounds using Saccharomyces cerevisiae (Averesch & Kromer, 2018). Gao et al. reported that the development of a new platform based on more efficient xylose utilization makes Scheffersomyces stipites particularly suited to produce the shikimate group of compounds. Shikimate was produced at 3.11 g/l, representing the highest level among the products of the shikimate pathway in yeasts (Gao et al., 2017).
In summary, this paper reported the construction of E. coli strains capable of producing shikimate at high concentrations from D-glucose. To accumulate high concentrations of shikimate, tyrR, ptsG, and pykA genes were deleted sequentially from E. coli AB2834, in which the aroE gene was mutated to prevent the conversion of DHS to shikimate. Extra copies of aroB, aroD, galP, ppsA, aroG, and aroF involved in shikimate biosynthesis were also inserted to maximize shikimate accumulation. To convert DHS to shikimate, the aroE gene was introduced so that it could be expressed under a strong oppA promoter. A controlled fed-batch operation was performed with a statistically optimized production medium in a 7-l bioreactor. The redesigned E. coli strain produced 101 g/l shikimate and 9.2 g/l DHS with almost no accumulation of metabolic intermediates, such as acetate and quinate. This study demonstrates the potential of E. coli to produce high levels of intermediate metabolites of aromatic pathways and a rational cell factory design approach to manufacture valuable aromatic compounds.

Supplementary Material
Supplementary material is available online at JIMB (www.academic. oup.com/jimb).

Funding
This work was carried out with the support of "Cooperative Research Program for Agriculture Science and Technology Development (Project No. PJ01318701)," Rural Development Administration, Republic of Korea.