Red^®/ET^® Recombination is a powerful tool for the chromosomal inactivation of genes or complete operons 123. This method is based on the homologous in vivo replacement of a gene/operon of any size and chosen position with a resistance cassette (e.g. a gene conferring antibiotic resistance) in a precise and specific manner by the λ-phage Redγβα recombination system and is applicable to all enteric bacteria. For certain purposes, the introduction of single point mutations is required rather than complete gene replacement or deletion, e.g. to modify a promoter or specifically inactivate the catalytic center of a certain gene product. In most cases, positive phenotypic selection for the introduced mutation is un-achievable, so that mutagenesis must be coupled to a counter-selection approach. A powerful counter-selection system based on the rpsL gene (rpsL-neo) and streptomycin selection for the introduction of point mutations into Bacterial Artificial Chromosomes (BACs) has been described 4, and this method was further useful for the recombination of large DNA-fragments into BACs 5. This counter-selection system is based on the rpsL gene encoding the S12 ribosomal protein, which is the target of streptomycin. Chromosomal mutations within rpsL are responsible for streptomycin resistance 6. Many E. coli strains commonly used for protein overproduction and/or metabolic engineering, including MC4100 7, JM110 8, Rosetta DE3 (Novagen, Darmstadt), HB101 (ATCC 33694), and TOP10 (Invitrogen, Karlsruhe) carry an altered rpsL gene conferring streptomycin resistance. The counter-selection system takes advantage of the fact that mutations within rpsL leading to streptomycin resistance are recessive in a merodiploid strain.

Here we describe a rapid method for site-directed mutagenesis of the E. coli chromosome. We used rpsL counter-selection in combination with Red^®/ET^® Recombination to introduce a single point mutation into the kdpA gene locus of the MG1655 strain genome. We show that rpsL counter-selection is applicable for introducing modifications into the bacterial chromosome. Single-stranded synthetic oligonucleotides can be used for recombination, so any chromosomal modification can be designed.

Introduction of a single point mutation into kdpA was chosen to demonstrate rpsL based counter-selection in combination with Red^®/ET^® Recombination on the E. coli chromosome. KdpA is the K⁺-translocating subunit of the KdpFABC system, a high affinity K⁺ uptake system in E. coli, which is essential for growth under K⁺-limitation 9. Previously, clones obtained by random mutagenesis were screened for their inability to grow under K⁺-limiting conditions. The mutation in one of these clones (E. coli TK2204) was mapped to kdpA, and was designated kdpA4 10. Detailed analysis of kdpA4 revealed a point mutation at position 1033 (G to A). This substitution changes glycine 345 to serine. Glycine 345 is located within the K⁺ selectivity filter of subunit KdpA 11, and the substitution prevents growth under K⁺-limiting conditions. Since E. coli strain TK2240 is not isogenic to E. coli MG1655 (Tab. 1), and the latter strain was already used for systems biological studies of the Kdp system 12, we aimed to introduce the original kdpA4 point mutation into the E. coli MG1655 chromosome. As kdpA4 prevents growth under K⁺-limitation, no positive selection for the mutants was possible. Thus, we applied the rpsL based counter-selection method to screen for mutants.

In contrast to other K-12 E. coli strains (see Introduction), strain MG1655 contains the wild-type rpsL gene and is naturally sensitive to streptomycin. Therefore, an altered rpsL gene conferring streptomycin resistance was introduced first. For this purpose rpsL150 of E. coli MC4100 was amplified and recombined into E. coli MG1655 carrying plasmid pRed/ET(amp). The temperature sensitive (ts) origin of replication of plasmid pRed/ET(amp) restricts replication at 37°C. After each recombination step cells were incubated at 37°C to remove the plasmid, and the colonies obtained were tested for plasmid loss.

In general, it is not absolutely necessary to cure strains from plasmid pRed/ET(amp) after each recombination step. Since araBAD promoter activity is not completely down-regulated in the absence of arabinose, the occurrence of any rearrangement was prevented. Resistant colonies were not obtained on the control plate (cells equally treated without Red^®/ET^® production). The overall recombination frequency was about 4 × 10^-8/μg DNA. All clones obtained were streptomycin resistant and ampicillin sensitive. The efficiency of proper recombination was 100% (i.e. no false-positives). We observed no differences between E. coli MG1655 and E. coli MG1655 rpsL150, neither in growth nor in kdpFABC transcription/KdpFABC translation fidelity (data not shown).

The general mechanism of rpsL counter-selection is illustrated in Fig. 1. A prerequisite for the strain to be used is a chromosomal encoded resistance to streptomycin conferred by a mutation in rpsL, which implies that this method can only be applied to bacterial stains with an rpsL homologue. Briefly, cells carrying a mutated chromosomal rpsL gene (e.g. rpsL150), exhibiting streptomycin resistance, are modified by the introduction of linear DNA comprising the rpsL-neo cassette with 50 bp homology arms surrounding the target gene site of interest. The additional wild-type allele of rpsL provided by the rpsL-neo cassette is dominant over rpsL150, and cells become sensitive to streptomycin and resistant to kanamycin, when the cassette has been inserted into the chromosome. In the next step, a DNA-fragment carrying the mutated site in the target gene (e.g. kdpA) is introduced into cells, which recombines and replaces the rpsL-neo cassette. Due to the loss of the rpsL wild-type allele, recombinants regain streptomycin resistance again and can easily be selected. Furthermore, recombinants become kanamycin sensitive due to the loss of the rpsL-neo cassette. Selection for streptomycin resistant clones is carried out on complex medium agar plates, so clones grow within 1–2 days of incubation. Within this short incubation time, clones with spontaneous rpsL mutations rarely appear.

Insertion of the kdpAG1033A point mutation into the chromosome is illustrated in Fig. 2. The homology arms of the rpsL-neo cassette were constructed in a way that a 50 bp homology to kdpA was ensured on each side of the point mutation, while the downstream homology arm contained the point mutation. The PCR product kdpA4-rpsL-neo (1.4 kb) was recombined into E. coli MG1655 rpsL150 carrying plasmid pRed/ET(amp). About half of the colonies obtained represented false positives based on the number of colonies observed on the control plates. Plasmid loss appeared to be a strain MG1655 specific problem, because other E. coli strains (e.g. HB101, W3110, DH5a, DH10B) generally exhibit >90% validated positives. Nevertheless, the overall recombination frequency for the kdpA4-rpsL-neo cassette in E. coli MG1655 was about 3 × 10^-7/μg DNA (25% of clones were correct), which was checked by testing for sensitivity to streptomycin (Strep^S) and ampicillin (Amp^S), and resistance to kanamycin (Kan^R). Introduction of the kdpA4-rpsL-neo cassette was verified for two Strep^S, Kan^R and Amp^S clones by PCR analysis using counterA4_sense and counter_A4 antisense primers. For both clones, a fragment of 1.44 kb was obtained, which corresponded to the inserted kdpA4-rpsL-neo cassette (Fig. 2, Fig. 3a).

Next, the kdpA4-rpsL-neo cassette was replaced by a DNA-fragment containing the kdpA4 mutation, which was obtained by PCR using an E. coli TK2204 chromosomal DNA template. The DNA-fragment was recombined into E. coli MG1655 rpsL150 kdpA4:rpsL-neo carrying plasmid pRed/ET(amp). Depending on the amount of cells plated, limited background growth was visible as a kind of smear at this step. Nevertheless, single colonies were clearly visible. Several hundred clones were obtained, and about half of the number of clones observed on the recombination plate was observed on the control plate. The overall Strep^R, Kan^S and Amp^S clone yield was about 10%, indicating that the recombination frequency was about 2 × 10^-7/μg DNA. The replacement of the kdpA4-rpsL-neo cassette against the mutated kdpA gene fragment (kdpA4) was tested for two clones which were Strep^R, Kan^S and Amp^S by PCR analysis. As expected, the size of the PCR products was about 150 bp confirming the loss of the kdpA4-rpsL-neo cassette (Figs. 2 and 3a).

In principle, this method should work for engineering a mutation into the chromosome using single-stranded synthetic oligonucleotides. To test this, two oligonucleotides were used for recombination, one 79 nucleotides (nt) and one 99 nt in length, consisting of 39 nt and 49 nt length homology arms, respectively, surrounding the point mutation. The recombination frequency for both oligonucleotides was observed to be about 2 × 10^-7/μg DNA, comparable to that observed for double-stranded DNA. Replacement of the kdpA4-rpsL-neo cassette against the mutated kdpA gene fragment (kdpA4) was confirmed for two Strep^R, Kan^S and Amp^S clones from each recombination setup (79 nt and 99 nt oligo) by PCR analysis (Figs. 2 and 3a).

DNA sequencing confirmed the accurate homologous recombination of the fragments within the kdpA gene, and introduction of the G1033A mutation. Furthermore, strain MG1655 with the point mutation in kdpA was tested phenotypically for growth under K⁺-limitation. As shown in Fig. 3b, cells with a kdpA4 mutation exhibit growth defects under K⁺-limiting conditions (0.02 mM K⁺).

New approaches to introduce point mutations into bacterial chromosomes are important for advancing functional genomics and systems biology investigations as well as for metabolic engineering. In this way, mutations are introduced and copy numbers of the encoded proteins are retained. Moreover, it is important to avoid polar effects from remaining antibiotic resistance cassettes in the gene of interest.

Here we demonstrated that rpsL counter-selection in combination with Red^®/ET^® Recombination is an efficient approach to modify the E. coli chromosome. We introduced a single point mutation into the kdpA locus of E. coli MG1655, one of the bacterial model strains used widely in systems biological approaches. The method described here is convenient because a selectable marker is used in each step. Furthermore, there is no need for cloning the gene fragment into a special vector, which makes the method time efficient. Recombination between the two rpsL alleles is improbable, because Red^®/ET^® Recombination starts recombination at the ends of a linear fragment (Y. Zhang, personal communication), and the wild-type rpsL allele located on the linear PCR-fragment is flanked by about 300 bp 50 of which are responsible for precise recombination. The method that we present works with synthetic oligonucleotides as well. Thus it might also be used for the replacement of multiple nucleotides, for deletions or insertions within the gene of interest.

Other counter-selection methods have been described. The introduction of a single point mutation into the E. coli chromosome (at the galK locus), has been described using λ prophage suicide counter-selection 13. thyA- and galK-based counter-selection systems have been successfully applied to the modification of BACs 1415, and these methods allow modification of DNA without leaving a selectable marker at the modification site likewise rpsL-based counter-selection. While the rpsL marker gene is only slightly modified, these methods require complete deletion of the respective marker gene. Furthermore, in the case of rpsL counter-selection, cells can be grown in complex medium instead of in minimal medium, which speeds up the process and circumvents spontaneous mutations.

Another counter-selection method is recA- dependent, which relies on the integration and resolution of a special shuttle vector, and has been successfully applied to modify BACs 1617. This method requires time-consuming restriction and ligation steps. These same time-consuming steps are necessary for the "gene gorging" method described previously 18, and most of the other counter-selection systems which use recognition sites for rare-cutting restriction endonucleases, like I-SceI. Nevertheless, these methods were successfully utilized for the modification of BACs, and the E. coli genome 1920. Recently, counter-selection using I-SceI rare restriction sites has been applied to modify the Salmonella enteritidis genome using PCR-based recombination cassettes, thereby overcoming the disadvantages described above 21. Another common, simple method is sacB-based counter-selection 22, but this method implies a high frequency of spontaneous point mutations in the selection marker sacB, which significantly increases the background after negative selection 15. Likewise, spontaneous point mutations within the rpsL-neo cassette might also promote the occurrence of false-positive clones. Such false-positive clones can easily be identified by checking for streptomycin sensitivity after recombination of the rpsL-neo cassette.

We present an efficient and non-disruptive approach to introduce point mutations into the E. coli chromosome. Chromosomal modifications performed by rpsL counter-selection may also be used for enteric bacteria that contain an rpsL homologue and in which Red^®/ET^® Recombination is functional. Red^®/ET^® Recombination or analogous methods have been efficiently applied in Salmonella 23, Yersinia 24, Shigella 25, Citrobacter 26, and Serratia 27 species, and all of these species have a rpsL homologue. The versatility of this system illustrates its potential for furthering studies in an important clade of organisms.

BAC: Bacterial Artificial Chromosome; nt: nucleotides; ds: double-stranded; ss: single-stranded.

RH has performed the experiments and drafted the manuscript. TZ contributed ideas for performing the experiments, professional support, and helpful suggestions for improving the manuscript. KJ is the project leader and has improved the manuscript. All authors have read and approved the final manuscript.