Two-step Red-mediated recombination for versatile high-efficiency markerless DNA manipulation in Escherichia coli

Introduction

Bacterial artificial chromosomes (BACs), mini-F plasmids into which DNA of up to 300 kb can be cloned (1), have made possible rapid and efficient mutagenesis of large sequences by techniques that were previously restricted to genetic manipulation of small plasmids in Escherichia coli. This advance has accelerated progress in genome sequencing projects, has revolutionized work with DNA viruses (2,3), and has facilitated the generation of transgenic animals through the use of modified clones of genomic BAC libraries (4).

Besides RecA-mediated mutagenesis, Red recombination or RecET cloning is one of the most commonly exploited techniques to engineer sequences in large DNA constructs (5,6). Substrates for Red mutagenesis are DNA double-strand breaks (DSB) wherein homologous recombination is focused close to the ends of the linear double-stranded DNA (dsDNA). Because replication of the substrate is not required, Red recombination allows utilization of PCR-amplified positive selection markers using primers bearing 40–50 bp extensions homologous to the target sequences (5,7–9).

To avoid long residual foreign sequences within modified constructs, FRT or loxP sites flanking selection markers have been used for excision of the positive selection markers with the corresponding recombinases Flp or Cre (6). The persistence of at least one copy of the FRT or loxP sites limits the repeatability of the procedure and may interfere with gene expression in some circumstances. Alternative systems developed for markerless mutagenesis of large DNA sequences in E. coli first insert a cassette consisting of both positive and negative selection markers or bifunctional markers, which allow both positive and negative selection by one gene product. Those cassettes are then replaced with the desired markerless (linear) DNA by a second recombination event and subsequent counterselection (6,10–13). One problem associated with negative selection markers is that they can be lost unspecifically such as by recombinations between repeat sequences, which are common in both mammalian and viral sequences (data not shown). Methods for markerless replacement without selection markers, such as the use of single oligonucleotides harboring modified sequences (14), the “gene gorging” technique (15), or the “hit-and-fix” protocol (16), lack a powerful procedure to screen for desired recombination events. To raise efficiencies, those methods employ PCR screening of pooled samples (17) or a preselection for recombination events with a marker targeting the E. coli chromosome as a second substrate (18).

The 18-bp recognition site of the homing endonuclease I-SceI was used previously as the negative selection marker with an in vivo cleavage as the counterselection step (10). The technique presented here was refined such that I-SceI is utilized not only for counterselection but also for generation of the substrate for a second Red recombination in E. coli. In this modified strategy, the positive selection markers that were used to introduce the desired target modification in a first step are removed in a second step by the combination of I-SceI cleavage and intramolecular Red recombination utilizing a previously introduced sequence duplication. We show that this modification renders the method versatile and very efficient.

Materials and methods

Cells, Viruses, and BAC Clones

Equine herpesvirus type 1 (EHV-1) was propagated on rabbit kidney cells (RK13), maintained in Dulbecco's modified essential medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Two parental BAC clones were used; both are derived from EHV-1 strain RacL11 in which gene 71 located in the unique-short region was replaced by the mini-F sequence pHA1 (resulting in pL1 1 w) or pHA2 harboring an additional gfp gene within vector sequences (resulting in pL11g; also see Supplementary Figure S1) (19,20). Transfection of RK13 cells with pL11w or pL11g DNA isolated from E. coli was used to reconstitute recombinant viruses vL1 1w or vL1 1 g (21).

Bacteria and Plasmids

All plasmids were constructed and maintained in E. coli DH10B cells grown at 37°C in Luria Bertani (LB) medium. BAC maintenance and Red recombinations were performed in E. coli strain EL250 [which contains a λ prophage encoding the recombination enzymes Exo, Beta, and Gam (8)] harboring pL11 w or pL11g.

To construct pEPkan-S (see Supplementary Figure S2), the kanamycin resistance gene aphAI derived from plasmid pACYC 177 (New England Biolabs, Ipswich, MA, USA) was amplified by PCR (Supplementary Table S1), adding I-SceI recognition sites and a sequence encoding the FLAG® epitope (Sigma, St. Louis, MO, USA). The PCR product was cut with BamHI and EcoRV and cloned into the BglII and NruI sites of pcDNA3 (Invitrogen, La Jolla, CA, USA).

The 1.6-kb EcoRV/PstI fragment of plasmid pACBSR (15), harboring the I-sceI gene and a part of araC, was cloned into the EcoRV and PstI sites of pBAD24 (ATCC, Manassas, VA, USA), resulting in pBAD-I-sceI (see Supplementary Figure S3).

To delete a PstI site present in the mRFP1 encoding plasmid (22), pmRFP1 was digested with BglII and BamHI before religation. The resulting plasmid without the multiple cloning site (MCS) was named pmRFP1/-M.

Primers

Primers that we used for the manipulations are listed in Supplementary Table S1. Oligonucleotides were custom-synthesized, and all primers longer than 50 bp were polyacrylamide gel electrophoresis (PAGE) purified (Integrated DNA Technologies, Coralville, IA, USA).

Red Recombination

Red recombination was performed as described earlier (8). PCRs were performed with Taq DNA polymerase (New England Biolabs), DpnI was added to digest the template, and the amplification product was purified from an agarose gel with the Perfectprep® Gel Cleanup Kit (Eppendorf, Westbury, NY, USA). E. coli strain EL250 harboring pL1 1 w or pL1 1g was grown at 32°C in LB with 30 µg/mL chloramphenicol to an A₆₀₀ of 0.5 to 0.7. The EL250 culture was then transferred into a 42°C water bath shaker for 15 min at 220 rpm to induce the expression of the Red recombination system. Finally, bacteria were cooled on ice for 20 min, washed 3 times with ice-cold 10% glycerol in water, and resuspended in 1/100 of the original culture volume of 10% glycerol in water. Recombination-competent EL250 cells were either frozen in aliquots (50 µL) at -70°C or used fresh for electroporation using approximately 100 ng of PCR product (0.1 cm, 1.5 kV, 25 µF, 200 Ω). After electroporation, bacteria were shaken in 1 mL LB medium at 32°C for 1.5 to 2 h and finally spread on appropriate selective agar plates (see Supplementary Protocol).

Results and discussion

Markerless Introduction of Long Sequences

The major problem of markerless DNA modifications utilizing Red mutagenesis has been the absence of efficient protocols for the insertion of longer sequences. We began to optimize the developed mutagenesis procedure to allow for the efficient insertion of sequences of interest such as sequences encoding fluorescent proteins or promoters. Universal transfer constructs were generated that contained the sequences of interest with an inserted positive selection marker (aphAI, which confers kanamycin resistance), an adjoining I-SceI site, and a duplication of a short sequence of the sequences of interest (Figure 1, Supplementary Video S1). The transfer constructs generated by PCR amplification of the I-SceI-aphAI cassette with a forward primer contained an approximately 50 bp DNA duplication of nucleotides present in the sequences of interest. In addition, both the forward and reverse primers contained restriction sites to allow cloning of the PCR product in the correct orientation into a unique restriction site within the sequences of interest. The unique restriction site was chosen such that it is located immediately upstream of the approximately 50 bp sequence duplicated within the forward primer. Supplementary Video S1 summarizes the steps involved in generating and using universal transfer constructs, and Supplementary Figure S4 provides details about the sequences of interest employed in this study.

The transfer plasmids were then used to insert the sequences of interest into desired target sequences present in the BAC. First, the universal transfer constructs were amplified by PCR with primers that each possessed approximately 50 bp of homology to the respective target sequence. The resulting linear DNA fragment was inserted into the target BAC clones by classical Red recombination using E. coli strain EL250 (8). Specifically, 100 ng of plasmid pBAD-I-sceI (for arabinose-inducible expression of I-SceI) were electroporated together with the PCR product into the BAC-containing bacterial cells. Of the kanamycin (selection for aphAI)- and ampicillin (selection for pBAD-I-sceI)-resistant colonies, those in which successful integration of the PCR product into the BAC DNA had occurred were detected by colony PCR or restriction fragment length polymorphism (RFLP) analysis (Table 1, first column).

Table 1.  Efficiency of Two-Step Red-Mediated Recombination for Mutagenesis

The next step of the protocol involved removal of the positive selection markers. Colonies were grown at 32°C for 2–4 h in 2 mL of LB medium with ampicillin and chloramphenicol to maintain the BAC. Two milliliters of LB containing 1% arabinose and antibiotics were added to the culture to induce the expression of I-SceI from plasmid pBAD-I-sceI. Further incubation for 2 h at 32°C allowed cleavage at the I-SceI site, thereby generating a DSB. Cultures were then transferred into a 42°C water bath shaker for 30 min to initiate a second Red recombination event between duplicated sequences (Figure 1, shown in yellow) now present within the modified BAC. Subsequently, bacteria were grown for another 2–4 h at 32°C, and serial dilutions were spread on agar plates containing chloramphenicol, ampicillin, and 1% arabinose. Replica plating was used to detect kanamycin-sensitive colonies (i.e., cells in which aphAI excision had occurred), which were examined by RFLP or colony PCR screening. Frequencies of positive clones are listed in the second column of Table 1; diagrams of the constructs are shown in Figure 2; sample raw RFLP results are presented in Supplementary Figure S5. Finally, the affected loci of candidate positive mutants were PCR-amplified and sequenced to verify successful sequence modification in the BACs.

As summarized in the final column of Table 1, overall efficiencies for generating the desired long-insertion recombinants ranged from 72%–100%. Furthermore, when tested in RK13 cells, recombinant virus containing RFP inserted at amino acid 51 of VP16 displayed the expected phenotype and localization pattern as assessed by fluorescence microscopy (Supplementary Figure S6A).

Markerless Generation of Point Mutants

To generate point mutations in BAC clones, cloning of transfer constructs was unnecessary to achieve the sequence duplications required for scarless mutagenesis. The I-SceI-aphAI cassette from recombinant plasmid pEPkan-S (Supplementary Figure S2) was PCR-amplified with primers adding extensions of 60–80 bp to the cassette (Figure 3; Supplementary Video S2). The outer approximately 40 bp of primers and PCR products were homologous to the target locus in the BAC. Both primers also contained approximately 40 bp of modified target sequences, which were reverse complementary to each other and—depending on the type of modification—could also be a part of the outer approximately 40 bp (Figure 3). In the first Red recombination, the PCR product containing the positive selection marker flanked by appropriate sequence duplications and modifications was inserted into the locus of interest. In a second step, following the induction of cleavage at the I-SceI site as described above, Red recombination between the duplicated sequences resulted in markerless point mutations (see Figure 3 and Supplementary Video S3 for a general overview and Figure 2 for sequence specifics). A negative control in which arabinose was not added revealed the absence of detectable Red recombination (Table 1), as expected. Western blot analysis of cells infected with recombinant virus was performed for two point mutants: U_L48 M446A and a mutant in which the first start codon was disrupted (U_L48 M1L). The blot showed the presence or absence, respectively, of bands corresponding to a protein of wild-type size (Supplementary Figure S6B).

Markerless Deletion or Insertion of Short Sequences

If the duplications contained within the primers represent a noncontiguous sequence of the target DNA, the point mutations strategy can be adapted to allow markerless deletions. Deletion of shorter sequences appeared to be more efficient than long deletions; however, both procedures required minimal screening to derive positive clones (Table 1; see Figure 2 for explanatory diagrams). As before, the arabinose-negative control confirmed the dependence of the second recombination event on I-SceI expression (Table 1). Western blot analysis of both small deletion mutants of U_L48 revealed proteins of sizes consistent with the expected mutations (Supplementary Figure S6B).

Inclusion of short sequences in the primers (e.g., a sequence encoding a FLAG tag) resulted in the insertion of these short sequences of up to 30 bp into the target sites (Table 1 and Figure 2).

Conclusions

The mutagenesis strategy described here is based on a simple two-step procedure and utilizes various previously established methods such as Red recombination and counterselection with the homing endonuclease I-SceI. In the first step, a linear marker construct is inserted via Red recombination into the target site using positive selection. In the second step, I-SceI cleaves at its recognition site and creates a DSB. The adjoining duplicated sequence is then used as the new substrate for a second intramolecular Red recombination, which results in the loss of the previously introduced marker (Figure 1 and 3). This novel combination and adaptation of various techniques results in an accurate and highly efficient recombination method that enables a wide variety of different manipulations of large DNA molecules in E. coli. Besides the examples shown here, we have prepared a gene expression unit consisting of promoter, multiple cloning site, and poly(A) signal as a universal transfer construct. Sequences encoding various viral antigens were introduced into the generic transfer plasmid, inserted into viral BAC sequences by the described method, and recombinant vaccine viruses were generated (data not shown). The combination of flexible transfer constructs and well-established or novel vector constructs has great potential to allow timely generation of vector-based vaccines or gene transfer constructs for therapeutic purposes.

Probably the biggest advantage of the methodology described here is the efficiency of the protocol, which mainly rests on the fact that Red recombination focuses the recombination event close to a DSB (5) that is generated at an I-SceI site adjoining the duplicated sequence. Because the second Red recombination takes place within this short duplication of sequences, false-positive kanamycin-sensitive colonies resulting from spurious recombinations at other sites were extremely rare (Table 1). A second reason for the efficiency of the method is that both the substrate and the target for the second recombination step are present in all cells of the E. coli culture. Since loss of the positive selection marker and not only negative selection are used to select clones for further screening, error-prone repair at the DSB or lack of the I-SceI cleavage were not of concern. In our experience with EHV-1 and Marek's disease virus (MDV) BAC clones, this strategy has in a number of cases been the only means to generate certain recombinants that could otherwise not be obtained by established methods such as RecA-mediated mutagenesis because these methods resulted in the introduction of illegitimate recombination that mainly occur between tandem repeat sequences (data not shown).

Because the two limiting screening steps have efficiencies of 60% to 100%, an adaptation of the method even for high-throughput generation of libraries of mutant BACs should be possible, and the method may increase the capacities for the production of transgenic animals (7) without the introduction of unnecessary scars, such as loxP sites, within modified sequences (23).