Metabolic engineering of a reduced-genome strain of Escherichia coli for L-threonine production

Background Deletion of large blocks of nonessential genes that are not needed for metabolic pathways of interest can reduce the production of unwanted by-products, increase genome stability, and streamline metabolism without physiological compromise. Researchers have recently constructed a reduced-genome Escherichia coli strain MDS42 that lacks 14.3% of its chromosome. Results Here we describe the reengineering of the MDS42 genome to increase the production of the essential amino acid L-threonine. To this end, we over-expressed a feedback-resistant threonine operon (thrA*BC), deleted the genes that encode threonine dehydrogenase (tdh) and threonine transporters (tdcC and sstT), and introduced a mutant threonine exporter (rhtA23) in MDS42. The resulting strain, MDS-205, shows an ~83% increase in L-threonine production when cells are grown by flask fermentation, compared to a wild-type E. coli strain MG1655 engineered with the same threonine-specific modifications described above. And transcriptional analysis revealed the effect of the deletion of non-essential genes on the central metabolism and threonine pathways in MDS-205. Conclusion This result demonstrates that the elimination of genes unnecessary for cell growth can increase the productivity of an industrial strain, most likely by reducing the metabolic burden and improving the metabolic efficiency of cells.


Background
The vast increase in annotated genome information and high-throughput technologies has enabled a systematic improvement of industrial microbes through genome engineering. Restructuring of microbial genomes has been shown to have several advantages over conventional approaches for strain improvement [1][2][3][4][5]. Restructured genomes with the desired functionalities have served as customized industrial strains that display (i) streamlined metabolic pathways for the production of selected biomaterials, (ii) a reduced production of unwanted by-products, and (iii) increased genome stability [6][7][8][9].
Escherichia coli is the most commonly used microbe for both laboratory research and industrial-scale production of metabolites, such as amino acids and proteins, for therapeutic or commercial uses [10][11][12][13][14]. Because E. coli has adapted to a lifestyle that includes residence in animal intestines with frequent exposure to aqueous and soil environments, many of the genes in its genome are unnecessary for growth in the relatively simple environment within industrial fermentors. Further, the complete genome sequence of E. coli has revealed numerous genes whose products have no known functions and genetic materials that may have been acquired from other organisms in the recent past [15]. In an effort to improve E. coli as an industrial host, many researchers have deleted or added limited numbers of selected genes to the genome or modified plasmids to complement the existing genome [16][17][18][19][20][21][22][23][24]. These efforts have helped researchers make significant progress in improving E. coli as a production host, but have not addressed the productivity problems caused by the numerous E. coli genes with potentially detrimental functions.
Recently, an E. coli genome was reduced by the precise deletion of nonessential genes and other DNA sequencesincluding all known recombinogenic and mobile DNA and cryptic virulence genes-to construct a genetically stable strain that displays robust metabolic performance [5]. The resulting strain, E. coli MDS42, has a chromosome that is 14.3% smaller than that of its parental E. coli strain MG1655. MDS42 shows robust growth under normal laboratory conditions and even better growth in high-cell density fermentations, as well as increased transformation efficiency relative to MG1655 [5]. Therefore, the elimination of unnecessary genes and sequences from an E. coli genome appears to have produced a stable reducedgenome strain without physiological compromise. Furthermore, the deletion of all insertion sequence (IS) ele-ments from the genome means that the strain is free of ISmediated mutagenesis and genomic rearrangements.
In this study, we report the results of a reengineering of MDS42 to increase production of the amino acid L-threonine, which is essential for growth and maintenance of commercial livestock [25]. This reengineering included the overexpression of a feedback-resistant threonine operon (thrA*BC) under the control of a recombinant Tac promoter, deletion of the genes that encode threonine dehydrogenase (tdh) and threonine uptake proteins (tdcC and sstT), and introduction of a mutant threonine exporter gene (rhtA23). The reengineered strain, called MDS-205, shows an ~83% increase in threonine production by flask fermentation relative to the wild-type E. coli strain MG1655 that had been engineered to carry the same threonine-specific modifications. Minimization of an E. coli genome by the elimination of genes unnecessary for growth increases the productivity of the strain by reducing the metabolic burden caused by maintenance and expression of unnecessary genes and improving the metabolic efficiency of the cell.

Bacterial strains, plasmids, enzymes, and chemicals
The bacterial strains used in this work are listed in Table  1. Plasmid pKD46 was obtained from B. L. Wanner [26], pST76-ASceP from G. Posfai [27], and pCSI from S. C. Kim [28]. All enzymes were purchased from New England Biolabs (Beverly, MA, USA) except Taq polymerase, which was from Takara Bio Inc. (Shiga, Japan). All antibiotics and chemicals were from Sigma-Aldrich (St. Louis, MO, USA). Ampicillin, chloramphenicol, and kanamycin were used at concentrations of 50, 17, and 25 μg/ml, respectively. All the primers used in this work are listed in Additional file 1.

Construction of threonine-producing E. coli strains
To study the effect of minimization of an E. coli genome on the production of L-threonine, the reduced-genome strain MDS42 [5] and wild-type strain MG1655 were engineered to produce L-threonine. First, to release the feedback inhibition on aspartokinase I and homoserine dehydrogenase I that are encoded by the thrA gene, and the transcriptional attenuation of the thrABC operon by intracellular threonine [29], a feedback-resistant thrA*BC operon under the control of a Tac promoter [30] was introduced to replace the wild-type thrABC operon in the MG1655 and MDS42 genomes (Fig. 1). The feedbackresistant thrA*BC operon (3.5 kb) was amplified by the polymerase chain reaction (PCR) from E. coli ATCC 21277 genomic DNA with the following forward (P-f1) and reverse (P-r1) primers (see Fig. 1). A 1.2-kb fragment containing a chloramphenicol resistant gene (cat) and an I-SceI recognition site was also amplified by PCR from plasmid pCSI [28]. Finally, a 0.5-kb homology fragment that sits to the left of the thrABC operon ( Fig. 1a) was PCRamplified from the chromosomal DNA of MDS42 and MG1655. This amplified 0.5-kb fragment contained a short, 20-base pair (bp) flanking sequence on each side: the 3' end of the 0.5-kb fragment overlapped with the 5' end of the 3.5-kb fragment and the 5' end overlapped with the 3' end of the 1.2-kb fragment described above. This 0.5-kb fragment and 2 PCR fragments above (that is, the 3.5-kb PCR fragment, which contained the feedbackresistant thrA*BC operon under the control of a Tac promoter, and the 1.2-kb fragment, which contained a cat gene and an I-SceI recognition site) were combined by recombinant PCR using the following forward (P-f2) and reverse (P-r2) primers (see Fig. 1A). The constructed 5.3kb DNA cassette was electrotransformed into MG1655 and MDS42, both of which harbored pKD46 expressing λ-Red genes (γ, β, and exo). The recombinants were selected on chloramphenicol-containing LB plates, and then the helper plasmid pKD46 was cured by growing the selected recombinants at 42°C. The cat gene that was introduced as described above was then excised from the constructed recombinant strains by double-strand break (DSB) repair mediated by the I-SceI endonuclease expressed from pST76-ASceP, which generated strain MG-101 from MG1655 and strain MDS-201 from MDS42. To constitutively express the feedback-resistant thrA*BC operon under the control of a Tac promoter, the lacI gene was deleted from the MG-101 and MDS-201 genomes by the markerless deletion method [28], producing MG-102 and MDS-202, respectively. To prevent the degradation of Lthreonine produced inside the cells, the threonine dehydrogenase gene (tdh) was deleted from the MG-102 and MDS-202 genomes by the markerless deletion method, generating strains MG-103 and MDS-203, respectively. To further increase threonine production, two threonine uptake genes, tdcC and sstT, were sequentially replaced by the mutant threonine exporter gene (rhtA23) (see Fig. 1), whose expression level is 10-fold higher than that of the wild-type rhtA gene [31]. These manipulations yielded strains MG-104 and MG-105 from MG-103, and MDS-204 and MDS-205 from MDS-203. At each step of the strain constructions, the modification of each target region was verified by PCR using pairs of primers that flanked the endpoints of each target region.

Bioassay for L-threonine using an E. coli threonine auxotroph
The threonine auxotroph MG1655 thrB::Tn5 [32] was inoculated into 3 ml of LB medium supplemented with kanamycin and grown at 37°C. When the OD 600 reached 0.8, the threonine auxotrophic cells were harvested by centrifugation at 6,000 g for 10 min, resuspended in 3 ml of fresh M9 minimal medium, and cultivated for 5 h at 37°C to deplete any endogenous threonine available to the cells [33]. The threonine auxotrophic culture was then diluted 1:100 into 3 ml of fresh M9 minimal medium to which 300 μl of the filterate of the culture broth of each threonine producing strains was added. After 24 h of cultivation at 37°C, growth of the threonine auxotroph was measured at OD 600 .
A standard growth curve was established in order to correlate the growth of the auxotroph to various concentrations of threonine in M9 minimal media. Linear regression analysis of the plotted data was considered a least-squares fit (R 2 ) of the relationship between threonine supplementation and growth of the auxotroph, and was used as a standard curve when R 2 was greater than 0.95. The R 2 value was derived from the regression line of the resulting plots (OD 600 vs. threonine concentration).

Threonine production by flask fermentation
The threonine-producing strains were grown on LB plates overnight and then transferred to a 250 ml flask containing 50 ml of seed medium [ . The fermentation was run for 24 h at 37°C, with vigorous agitation of the culture on a shaker (300 rpm). After cultivation, the amount of threonine accumulated in the broth was analyzed with the bioassay system using the threonine auxotroph described above.

Microarray analysis
Strains MG1655, MDS42, MG-105, and MDS-205 were inoculated from single colonies into 5 ml of seed medium and grown at 37°C overnight. From each overnight culture, 500 μl was used to inoculate 50 ml of fresh fermentation medium-1. These cultures were grown at 37°C, and the cells were harvested at an early log phase corresponding to an OD 600 of 0.4. Total RNA was extracted using the MasterPure™ RNA Purification Kit (Epicentre Technologies, Madison, WI, USA) from 1 ml of the early log phase culture. cDNA synthesis and labeling were performed as described in the Affymetrix GeneChip E. coli Antisense Genome Array Technical Manual [35]. The resulting labeled cDNAs were hybridized to an Affymetrix E. coli antisense genome array. Patterns of hybridization were detected with an Affymetrix Genearray scanner 2500 (Affymetrix, Inc., Santa Clara, CA, USA). The raw data were analyzed using Microarray Analysis Suite version 5.0 (Affymetrix). Every E. coli open reading frame (ORF) was assayed by a set of perfect match (PM) and mismatch (MM) probe pairs. If the PM probe showed an intensity that was at least 200 U higher than that of the MM probe, the probe pair was considered to be present [35]. An ORF was considered to be present with 95% confidence if neighboring probe pairs within an ORF were present.

Construction of L-threonine-overproducing E. coli strains and L-threonine production
The wild-type E. coli MG1655 and reduced-genome E. coli MDS42 were modified in a stepwise manner to overproduce L-threonine. First, we isolated the threonine operon of E. coli strain ATCC 21277, which contains a mutated version of the thrA gene (thrA*) that encodes threonine feedback-resistant aspartate kinase I and homoserine dehydrogenase I; a homoserine kinase-encoding gene (thrB); and a threonine synthase-encoding gene (thrC). The isolated threonine operon, which was expressed under the control of a Tac promoter, was then inserted into the genomes of MG1655 and MDS42 to replace their wild-type threonine operons. Second, the lacI gene, which encodes the E. coli LacI transcriptional repressor, was deleted so that the mutant threonine operon would be expressed constitutively. The resulting strains, MG-102 and MDS-202, produced 34.06 and 36.61 mg/L of L-threonine, respectively (Fig. 2).
Third, the tdh gene, which encodes threonine dehydrogenase, was deleted from MG-102 and MDS-202 to prevent the degradation of L-threonine, generating strains MG-103 and MDS-203, which produced 42.25 and 79.94 mg/ L of L-threonine, respectively (Fig. 2). Finally, to enhance the export of L-threonine into the culture media and block re-uptake, we sequentially replaced the tdcC gene, which encodes the threonine STP importer [36], and the sstT gene, which encodes the DctA dicarboxylate (DAACS) importer [37], in MG-103 and MDS-203 with a mutant threonine exporter gene (rhtA23). The resulting strains, MG-105 and MDS-205, produced 165.54 and 303.23 mg/ L of L-threonine, respectively (Fig. 2). These final strains showed an ~10-and ~15.5-fold increase in L-threonine production, compared to their parental strains, MG1655 and MDS42, respectively (Fig. 2).
When a batch-fermentation was carried out, the final strain MDS-205 produced 40.1 g/L of L-threonine (a yield of 0.401 g threonine/g glucose) after 30 h of fermentation.

Discussion
The restructuring of microbial genomes by eliminating genes that are unnecessary for a cellular metabolism has received special attention as an important strategy for industrial strain improvement [1][2][3][4][5]. Recently, scientists reported on the advantages of the reduced-genome E. coli strain MDS42 and its applications for bioindustry [5,38]. The MDS42 genome is 14.3% smaller than that of the wild-type strain MG1655 and has ~700 fewer genes.
In this study, we compared the metabolic efficiency of Lthreonine production in the wild-type E. coli strain MG1655 and reduced-genome E. coli strain MDS42 by introducing into each of these strains a series of genetic modifications that altered L-threonine production, degradation, export into the media, and re-uptake from the media. These modifications gave rise finally to MG-105 (from MG1655) and MDS-205 (from MDS42). Although both MG1655 and MDS42 went through the same modifications, the threonine production of MDS-205 strain was an ~2-fold greater than that of the MG-105 strain.
Genome-scale gene expression analysis of MG-105 and MDS-205 revealed that the most of the genes involved in the central metabolism and L-threonine biosynthesis were up-regulated in MDS-205. Among the up-regulated genes, the expression level of the rhtA threonine exporter gene was increased by 4.79-fold in MDS-205. This result indicates that the maximization of L-threonine export combined with deletion of the tdcC and sstT gene involved in the re-uptake of L-threonine across the membrane is one of the important steps for the mass production of L-threonine from E. coli. An additional advantage of the overexpression of the rhtA gene is that it increases the tolerance of E. coli to L-threonine by an ~3-fold [31], which helps the E. coli to withstand high concentrations of L-threonine in the media. This adaptation also might contribute to the higher L-threonine production.
The up-regulation of the ATP-dependent glucose transport and phosphorylation system (mglABC and glk genes) and down-regulation of the PTS system (ptsG gene) in MDS-205 may increase the availability of the phosphoenolpyruvate (PEP) which is a precursor of oxaloacetate (OAA) and aspartate. This result is consistent with the prediction based on flux balance analyses of gene knock-outs in an E. coli metabolic model that the replacement of PTS activity by an ATP-dependent glucose transport system should increase aspartate-family amino acids [39,40]. In addition, down-regulation of poxB, pta, and adhE genes, which are directing PEP to acetate and ethanol, may reduce the PEP flux to acetate and ethanol and further increase the PEP availability.
It is also reported that the up-regulation of pck gene involved in the carboxylation of PEP to OAA leads to increased cellular growth and biomolecular production, since the ppc reaction releases an inorganic phosphate, in contrast to the pck reaction, which produces a high-energy ATP [41][42][43][44]. The effect of the pentose phosphate shunt and glyoxylate bypass on the glycolytic flux to L-threonine production also has been reported [45]. Therefore, the upregulation of the pck gene involved in the PEP carboxylation and tktAB genes involved in the pentose phosphate shunt and the increased expression of the genes involved in TCA cycle and the glyoxylate bypass (mdh, fumABC, sdhABCD, and glcB genes) may increase the OAA synthesis in MDS-205, resulted in higher production of L-threonine in MDS-205 compared to MG-105. The increased PEP and OAA level of MDS-205 by increasing the non-PTS glucose uptake system, PEP carboxylation, the pentose phosphate pathway, and the glyoxylate shunt, and reducing the mixed acid fermentation may resulted in increased production of L-threonine.  The increased productivity of L-threonine in MDS42 might also result from a decrease in the metabolic burden due to the genome reduction. It is well known that the maintenance and expression of unnecessary genes, such as plasmid DNA, impose an uncharacterized metabolic burden on the bacterial host [46]. The metabolic burden could arise due to the extra biosynthetic demands for synthesis and expression of unnecessary genes [47,48], or the perturbation of the E. coli host regulatory system affecting central metabolic pathways [46,49]. The genome-scale gene expression analysis of the wild-type MG1655 and the MDS42 revealed that the genes involved in the non-PTS glucose uptake system and glycolysis were up-regulated in MDS42 compared to MG1655, whereas the genes involved in the glucose PTS-system and TCA cycle were down-regulated in MDS42 compared to MG1655 (Fig 4). It seems that the elimination of unnecessary genes from the E. coli genome might lead to more efficient cellular metabolism and an improved substrate yield coefficient, resulting in nutrient and energy saving of cells without physiological compromise. Researchers reported that the elimination of unnecessary genes actually improves E. coli robustness and enhances carbon metabolism [5,38,50].
In fact, genes involved in mixed acid fermentation were up-regulated in MDS42 compared to MG1655 (Fig 4), leading to the acetate accumulation which indicates a carbon-overflow metabolism [13]. Therefore, the increased production of L-threonine in MDS-205 might be resulted from the redirection of the overflowed carbon metabolism in MDS42 caused by elimination of unnecessary genes from the E. coli genome into the production of Lthreonine.