New Tool for Metabolic Pathway Engineering in Escherichia coli: One-Step Method To Modulate Expression of Chromosomal Genes

A simple and highly efficient method was developed to producea library of Escherichia coli clones that express a particularchromosomal gene at a wide range of expression levels. The basicstrategy was to replace all or part of the upstream region ofa coding sequence containing the elements involved in its expression(promoter, operator, gene coding for a regulator, ribosome bindingsite, and start codon) with a PCR-generated library of expressioncassettes.

INTRODUCTION

Top

Introduction

Escherichia coli is widely used in the biotechnology industryfor the production of pharmaceuticals, fine chemicals, and feed.In most of these applications, genetic manipulations of theproduction strain have been considered in order to improve theoverall performances of the bioprocess. Particularly, it isoften necessary to tune the level of many enzymes to achievethe expected flux and yield of production of the desired product(10, 13).

A new approach for modulating chromosomal gene expression hasrecently been described by Jensen and collaborators (11, 12,20). The two limitations of this method are that (i) to achievea promoter replacement on the chromosome, a cloning step isneeded, and (ii) it allows only promoter replacement which limitsthe maximal level of expression of a gene present in only onecopy.

Datsenko and Wanner (6) have recently developed a fast and efficientmethod to disrupt chromosomal genes in E. coli, in which PCRprimers provide the homology to the targeted gene(s). In thework presented here, we have extended the use of the tools developedby Datsenko and Wanner to create a simple and highly efficientmethod of producing a library of E. coli clones that expressa particular gene located on the chromosome at a very wide rangeof expression levels. The main advantages of this approach arethat (i) without a cloning step into a plasmid, a library ofclones with different levels of expression of a particular chromosomalgene can be obtained, (ii) it is possible to get both very highand very low levels of expression of a chromosomal gene, (iii)the expression cassette can be moved from strain to strain byP1 transduction (17), and (iv) if needed, markerless strainscan be made (4).

Description of the strategy.

Top

Description of the strategy.

The construction of a library of clones that express a chromosomalgene at different levels was done in three steps. The firststep was the creation of a library of expression cassettes withdefined artificial promoters. This step included the designof two 100-mer oligonucleotides (OL₁ and OL₂) containing variousspecific regions (called pS₁, pS₂, H₁, and P₁) (Fig. 1). Thereverse oligonucleotide was degenerated on specific bases ofthe sequence coding for a very short artificial promoter (P₁)to create a library of promoters. The template plasmid pKD3or pKD4 (6) and the two oligonucleotides (OL₁ and OL₂) werethen used in a PCR (Fig. 1) that produced a first set of expressioncassettes.

View larger version (29K):
[in this window]
[in a new window]
FIG. 1. Strategy for the creation of a library of clones with different levels of expression of a chromosomal gene. pS₁ and pS₂ are the priming sites for pKD3 or pKD4. H₁ and H₂ are the homology extensions. P₁ is the degenerated part of the reverse oligonucleotide OL₂ containing synthetic short GI promoters. OL₃ can contain up to three degenerated sequences: P₂ for mRNA-stabilizing sequences, P₃ for RBSs, and P₄ for start codons (see Results for a description). FRT is the FLP recombinase recognition target (4).

The second step was the extension of the initial promoter libraryby adding modified and defined ribosomal binding site (RBS)and/or mRNA-stabilizing sequences and/or start codons, dependingon the range of expression level needed. This PCR step usedthe first set of expression cassettes as a template. The forwardoligonucleotide was the same as that used for the first step(OL₁). The reverse oligonucleotide (OL₃) contained (i) a primingsequence (pS₃) for the artificial promoter; (ii) defined anddegenerated sequences that encoded mRNA-stabilizing sequences(P₂), RBSs (P₃), and start codons (P₄); and (iii) a flankingsequence (H₂) that was homologous to the chromosomal DNA downstreamof the start codon (Fig. 1). After the second PCR, a full libraryof expression cassettes was obtained.

The third step of the method was the in vivo recombination inE. coli. This last step was achieved by the transformation ofcompetent cells (expressing the Gam, Bet, and Exo proteins fromthe plasmid pKD46) with the library of expression cassettes(6, 7, 16). The result of the in vivo recombination was thereplacement of all or part of the upstream region of a codingsequence containing the elements involved in its expressionby a PCR-generated library of expression cassettes.

Evaluation of the feasibility of the method with the lacZ gene as a model.

Top

Evaluation of the feasibility...

The initial experiment was designed to evaluate the feasibilityof the method for a small library and the range of expressionlevels that could be obtained from the very short promotersincluded in an oligonucleotide. The LacZF and LacZR oligonucleotides(Table 1) were then designed to delete the lacI repressor andto replace the natural promoter region of the lactose operonby three very short artificial glucose isomerase (GI) promotersof different strengths (Fig. 2) and by an antibiotic resistancegene. The different artificial GI promoters were derived fromthe GI promoter of Streptomyces spp. (9), a promoter that wasshown to be relatively strong in E. coli. The correspondingset of expression cassettes obtained by PCR was used to transformEscherichia coli K-12 MG1655 (F^– LAM rph-1; Genetic StockCenter, New Haven, Conn.) carrying pKD46 competent cells withselection on agar plates containing X-Gal (5-bromo-4-chloro-3-indolyl-ß-D-galactopyranoside)and chloramphenicol. Ten transformants were checked for recombinationat the lacZ locus (by PCR using the LacseqF and LacseqR oligonucleotides[Table 1] located outside the recombination region) and forß-galactosidase activity (15). The sequence analysisof the promoter region of the different types of recombinantclones showed, as expected, that they differed by only one basepair in their –35 region. This experiment also demonstratedthat, with very short artificial promoters, it was possibleto achieve specific ß-galactosidase activities rangingfrom two times higher (2.3 ± 1.05 U/mg of proteins) tofour times lower (0.3 ± 0.11 U/mg of proteins) than thoseproduced with the IPTG (isopropyl-ß-D-thiogalactopyranoside)-induced(1 mM) natural lacZ promoter (1.2 ± 0.54 U/mg of proteins).

View this table:
[in this window]
[in a new window]
TABLE 1. Structures of the different oligonucleotides used in this study

View larger version (13K):
[in this window]
[in a new window]
FIG. 2. Sequences of three very short, artificial GI promoters introduced upstream of the +1 site of lacZ transcription. The –35 and –10 sites are recognition sites for the housekeeping RNA polymerase.

The possibility of using a controlled-degeneracy library ofRBS sequences in combination with an artificial promoter wasthen shown by use of the LacZF, LacZRT, and LacZRBSR oligonucleotides(Table 1). Eight transformants were selected, and their ß-galactosidaseactivities and DNA sequences upstream of lacZ were determined.As expected, the transformants differed only in their RBS sequences,and the ß-galactosidase activity ranged between 0.02± 0.01 and 5.7 ± 1.2 U/mg of proteins (Fig. 3).These values are 60 times lower and 5 times higher, respectively,than the value produced by the IPTG-induced natural lacZ promoter(1.2 ± 0.54 U/mg of proteins).

View larger version (32K):
[in this window]
[in a new window]
FIG. 3. Influence of different RBS sequences on ß-galactosidase activities. (A) CAAGGAGGAAACAGCTATG; (B) CAAGAAGGAAACAGCTATG; (C) CACACAGGAAACAGCTATG; (D) CTCACAGGAGACAGCTATG; (E) CTCACAGGAAACAGCTATG; (F) CACACAGAAAACAGCTATG; (G) CTCACAGAGAACAGCTATG; (H) CTCACAGAAAACAGCTATG. Boldface indicates RBS sequences, and italic indicates start codons.

In order to extend the range of expression of lacZ in the highervalues, the possibility of using a controlled-degeneracy libraryof mRNA-stabilizing sequences with a strong, short artificialpromoter was demonstrated by use of the LacZF and the LacZmRNAoligonucleotides (Table 1). Five transformants were selected,and their ß-galactosidase activities and the DNA sequencesupstream of lacZ were determined (Table 2). For the best clone,the activity was up to 15 times higher than that of the IPTG-inducednatural lacZ promoter. Based on a specific activity of 1,000U/mg of pure ß-galactosidase, it can be calculatedthat in the best case, this enzyme represents approximately2% of the total protein in the cell.

View this table:
[in this window]
[in a new window]
TABLE 2. Influence of mRNA-stabilizing sequences on ß-galactosidase activities

Finally, to demonstrate that it would be possible to createa library of E. coli clones with different levels of lacZ expressionby simultaneously introducing different artificial promoters,modified start codons, and modified RBS sequences, the LacZR,LacZF, and LacZRBS2R oligonucleotides (Table 1) were used intwo successive PCRs, as described in Fig. 1 and 2. The resultsshown in Table 3 indicate that the simultaneous modificationof the three elements that affect the expression level of achromosomal gene was achieved in one step. The expression levelsthat were found in five representative members of the libraryshowed that both very low (0.03 ± 0.01 U/mg of proteins)and high (5.7 ± 1.2 U/mg of proteins) levels of lacZexpression could be obtained, demonstrating the validity ofthis approach.

View this table:
[in this window]
[in a new window]
TABLE 3. Comparison of the ß-galactosidase activities of four randomly picked clones from the library and corresponding sequences of the lacZ upstream regions

Creation of a library of E. coli clones with different levels of expression of the glyceraldehyde-3-phosphate dehydrogenase and selection of the best level for growth or for glycerol production.

Top

Creation of a library...

To demonstrate the usefulness of the method for metabolic pathwayengineering, we chose gapA (encoding the glyceraldehyde-3-phosphatedehydrogenase) as a target gene because of its importance inthe central metabolism of E. coli (5). The GapZF and the GapRoligonucleotides were then designed to replace the natural promoterregion of gapA by the three artificial GI promoters that werepreviously used for lacZ. Ten transformants were selected andanalyzed for (i) recombination at the gapA locus (by PCR usingthe GapseqR and GapseqF oligonucleotides that are located outsidethe recombination region) and (ii) GapA activity. The transformantswere then classified in the three expected categories (1.2 GI,1.5 GI, and 1.6 GI), based on the GapA activities and the promotersequence upstream of gapA. One member of each category was thenindividually grown in a shake flask at 34°C in M9 medium(19) supplemented with 10 g of glucose per liter, and the opticaldensity at 550 nm was measured to estimate the specific growthrate (Fig. 4). The best specific growth rate was obtained forthe 1.6 GI gapA clone, which also exhibited the highest GapAactivity and an activity close to that of the wild-type strain.

View larger version (87K):
[in this window]
[in a new window]
FIG. 4. Replacement of the natural promoter region of gapA by three different short GI artificial promoters. (A) Influence of the type of synthetic GI promoter on GapA activity and specific growth rate of the E. coli strain; (B) influence of the type of synthetic GI promoter on average glycerol production rate (grams per liter per hour) and on the glycerol molar yield (mole/mole) on the E. coli(pAH48) strains. Fermentations were done in duplicate at 34°C in shake flasks (200 rpm) on M9 medium supplemented with 10 g of glucose per liter. Fermentation product and glucose were quantified as previously described (8). Variation between duplicate experiments was less than 10%.

To evaluate the effect of the gapA level on the production ofa chemical compound, the three different modified gapA strainswere transformed with the pAH48 plasmid that carries the genesused to convert dihydroacetone-phosphate into glycerol (9).The batch fermentations showed (Fig. 4) that the 1.5 GI gapAstrain gave both the best rate and the best molar yield of glycerolproduction. This result demonstrated that the optimal levelof a metabolic enzyme for production of a chemical may not necessarilybe the highest achievable expression level and emphasizes theneed for the selection of the appropriate expression for thedesired application. Moreover, in a single step, it is possibleto make use of this library to select the best mutant for growthor for glycerol production. For metabolic engineering, the methodhas several advantages over existing inducible expression systems(14). One advantage is that after the optimal expression hasbeen determined, the strain is, in principle, ready for usein the industrial fermentation process. Another important featureis the option to modulate, to different extents, the expressionof many genes located at different positions of the genome inthe same strain. This approach has been extensively and successfullyused to develop an industrial E. coli strain that produces 1,3-propanediolat very high yields (2).

FOOTNOTES

* Corresponding author. Mailing address: Laboratoire de Biotechnologie-Bioprocédés, UMR INSA-CNRS 5504, UMR INSA-INRA 792, INSA, 135 avenue de Rangueil, 31077 Toulouse cedex 4, France. Phone: (33) 5 61 55 94 52. Fax: (33) 5 61 55 94 00. E-mail: soucaille{at}insa-tlse.fr.

REFERENCES

Top