A Novel Selection Marker for Efficient DNA Cloning and Recombineering in E. coli

Production of recombinant DNA in bacterial cells is an essential technique in molecular biology. Plasmids are usually maintained in an E. coli host by antibiotic selection. However, there are only a few antibiotic-resistance markers available in common use. Here we report the adoption of a novel selection marker, mfabI (mutant fabI) for plasmid propagation in E. coli. mfabI expands the limited repertoire of selection markers and allows for more efficient molecular manipulation and plasmid propagation in E. coli. We show that mfabI is not only an efficient plasmid selection marker, but it also possesses unique activity that may facilitate molecular manipulation of unstable sequences. Furthermore, we have incorporated mfabI in the recombineering tool kit for generating mouse gene targeting vectors and demonstrate the advantage of using mfabI-containing recombineering vectors.


Introduction
The two most commonly used selection agents in bacteria are ampicillin and kanamycin for the maintenance of plasmids bearing their corresponding resistance genes, beta-lactamase and neomycin phosphotransferase II, respectively [1,2]. More antibioticresistance gene pairs have been discovered but are not commonly used for plasmid selection in bacteria. The chloramphenicol resistance gene is usually used for growth of BACs (bacterial artificial chromosomes) [3]. The tetracyclin resistance gene often serves as a negative selection marker in cloning vectors [4]. Moreover, chloramphenicol, tetracycline, and even kanamycin resistance genes have been frequently used in engineering of the bacterial genome to generate special host strains [5,6,7], and thus are precluded for conducting plasmid selection in engineered strains that already carry resistance to them. Resistance genes to blasticidin [8], hygromycin [9], and Zeocin [10] have also been developed into a variety of commercially available cloning vectors, but the high cost of these antibiotics generally precludes them from popular use. Therefore, the availability of a new selection marker would facilitate more efficient shuttling of recombinant DNA fragments between different vectors and allow for propagation in a broader range of special host strains.
Gene targeting by homologous recombination is an important method for mutating genes in the mouse. In spite of the progress of coordinated efforts to generate targeted mutant alleles for every annotated gene in the mouse genome [11], hypothesis-driven custom-made mutations at specific loci generated in individual laboratories remain an important source of novel discoveries. The recombineering technique is an efficient method commonly adopted by the research community to generate gene-targeting vectors, in which recombination between homologous regions in isolated DNA fragments, plasmids or BACs (bacterial artificial chromosomes) is catalyzed by bacteriophage enzymes expressed by engineered E. coli strains [12,13]. A few antibiotic resistance markers are alternatively or simultaneously selected for at different steps to achieve efficient recovery of desired recombination products. However, the final gene-targeting vectors usually contain large fragments (.5 kb) of mammalian genomic sequence which are sometimes unstable or inefficiently replicated in the bacteria host, and therefore the recombination efficiency varies and desired vectors are sometimes difficult to obtain. We encountered a low efficiency problem associated with large complex sequences, and thus have improved the recombineering system by introducing a novel plasmid selection marker, mfabI.
Triclosan (IUPAC name: 5-chloro-2-[2,4-dichlorophenoxy]phenol) is a synthetic chemical which can inhibit bacterial growth in low concentrations and acts as a biocide in higher concentrations [14]. It exerts antibacterial activity by its inhibitory binding to a bacterial enzyme FabI (enoyl ACP reductase) [15], which is required for lipid synthesis in E. coli [16]. Spontaneous point mutations in FabI affecting the binding of triclosan to FabI allow the mutant bacteria to grow in the presence of triclosan. In the genomic library screens which identified FabI as the triclosan target, plasmids harboring wild-type or a mutant form (G93V) of the fabI gene were found to confer triclosan resistance to the transformed E. coli host [15]-suggesting its potential use as a plasmid selection marker. The wild-type fabI gene was further shown to be an effective plasmid selection marker in E. coli [17]. Here, we demonstrate that the G93V mutant form of fabI (mfabI) is an even more efficient selection marker with a unique growth suppression effect which may lead to stabilization of large or complex cloned sequences. We also developed mfabI-containing recombineering vectors to improve the recombineering tool kit for mouse gene-targeting construct production and demonstrate their advantage in practical use.

Materials and Methods
Plasmid construction pF was constructed by PCR cloning and recombineering to replace the ampicillin resistance marker on pBluescript KS-with fabI. Primer sequences are listed in Table 1. fabI was amplified with Fab-F/Fab-R from genomic DNA of DH10B. The 59 and 39 homologous sequences were amplified from pBluescript KS-(pBS) (Stratagene) with Amp5-F/Amp5-R and Amp3-F/Amp3-R, respectively. The three fragments were PCR-spliced together with Amp5-F/Amp3-R, gel-purified, and then electroporated into heatshocked SW106 [18] electrocompetent cells together with pBluescript KS-. Electroporated cells were selected on 1 mM triclosan plates. Correct clones were identified by restriction enzyme digestion of miniprep DNA and confirmed by sequencing with Amp5-F, Amp3-R, Fab-R1, and Fab-F1. The initial clones all contained mixtures of pBS and pF plasmids. pF DNA was further isolated by retransformation and miniprep screening.
pF2 was constructed by PCR mutagenesis. Using pF as template, two G93V mutation containing fragments were amplified by Fab-BstE2-F/Fab-G93V-R and Fab-G93V-F/Fab-BssH2-R, gel-purified, and then PCR-spliced together with Fab-BstE2-F/ Fab-BssH2-R, resulting in the mfabI fragment. The mfabI fragment was cloned into pF to generate pF2 with BstEII/BssHII digestion. Transformed cells were selected with 50 mM triclosan. Correct clones were identified by restriction enzyme digestion of miniprep DNA and confirmed by sequencing with Fab-BstE2-F and Fab-BssH2-R.

Bacterial transformation and growth
In transformation efficiency assays with DH10B cells, homemade retransformation grade chemical competent cells were used. Twenty-eight microliters of competent cells were heat-shock transformed with 2 ml of 100 ng/ml plasmid DNA, recovered at 32uC for 1.5 hr in 1 ml SOB, from which 50 ml of cells were plated on each plate. For assays with DH5a cells, 28 ml homemade cloning grade chemical competent cells, 2 ml of 1 ng/ml DNA, were used. For recombineering experiments, we follow the published protocols [12]; after 1.5-hr recovery, 200 ml from a 1 ml suspension was plated on each selection plate, incubated at 32uC overnight. Triclosan was purchase from VWR International (Cat. No. 80511-110, Triclosan antibiotic 1 gm, EMD Millipore), dissolved in DMSO, and alliquoted, stored in 220uC as 50 mM and 1 mM (1,0006) concentration stocks.
In the plating assays with clonal broth culture of pF-and pF2transformants, colonies were picked into LB broth with 1 mM triclosan (untransformed DH10B cells in antibiotic-free LB broth), and cultured in 32uC shaker for 21 hrs. At the time of plating, amounts equivalent to 1.6610 23 ml of OD600 = 1 culture were used for each sample on each plate.
In liquid culture assays, overnight clonal cultures were first spun down and resuspended in antibiotic-free LB broth to avoid any remnant antibiotic affecting the subsequent assay. O.D. equalized amounts of culture were inoculated into 15 ml LB broth in 50 ml conical tubes with different concentrations of triclosan, grown with constant 250-rpm shaking. At the end of culture, amounts equivalent to 2.1610 23 ml of OD600 = 1 culture were used for each sample on each plate for the triclosan resistance retention assay.
In the retransformation assay in DH5a cells, 2.5, 5 and 10 ng of pF2, pF2-DTA-Rosa26 and pF2-DTA-Rosa26-Insert plasmids were used to transform 14 ml of homemade cloning grade chemical competent cells. After 1.5-hr recovery at 37uC, 150 ml from a 500 ml resuspension was spread onto 1 mM triclosan LB plates and grown for 24 hrs at 37 or 32uC before observation. Colonies were picked from plates incubated at 32uC and grown in LB broth with 1 mM triclosan at 37uC for 17 hrs before being measured and yielded for DNA extraction.

Colony counting and statistics
Plates were scanned on a flatbed scanner with 300 dpi resolution, 8bit, grey scale setting. Scanned images were first processed in Adobe Photoshop to enhance contrast by ''Image.Adjustments.Levels'' Table 1. Primers used in plasmid construction.

Triclosan resistance conferred by mfabI plasmid
We cloned the fabI gene from the DH10B [20] bacterial genome using PCR methods. Vector pF was generated by replacing the ampicillin resistance gene in pBluescript (pBS) (Stratagene) with fabI. By PCR mutagenesis, the G93V point mutation was introduced into the fabI in pF, resulting in mfabI, to generate pF2 (Fig. 1A). The function of fabI and mfabI, and bacteriostatic activity of triclosan was reconfirmed by retransformation and plating assays. Triclosan efficiently inhibited the growth of untransformed bacteria, while mfabI marker conferred a stronger host resistance to triclosan than the wild-type fabI (Fig. 1B).

Suppression of bacterial growth by mfabI plasmid
It was reported fabI in a plasmid can suppress the growth of the host cells in antibiotic-free medium, probably due to an imbalance in lipid metabolism in the cells caused by overproduction of the FabI enzyme [17]. Unexpectedly, mfabI exerts an even stronger growth suppression effect on host cells than that of fabI in the standard growth condition on solid support medium.
When DH10B E. coli cells were transformed by pF or pF2 and grown in 37uC, the pF2 colonies were much smaller than pF colonies ( Fig. 2A, compare i with iv). We also transformed another commonly used E. coli strain, DH5a [20], and observed a similar effect ( Fig. 2A compare vii with x), indicating mfabI-induced growth suppression is not specific to DH10B cells but more general to most strains. However, when we used a lower growth temperature or a higher triclosan concentration during the construction of pF2 or manipulation of other mfabI plasmids, we did not observe a dramatic colony size reduction under these conditions. Therefore we examined the influence of temperature and triclosan on pF2 colony growth. Indeed, lower incubation temperature dramatically rescued the growth phenotype of pF2 colonies ( Fig. 2A v and xi). In addition, the growth suppression by pF2 in 37uC was not permanent but could be subsequently relieved by shifting the cells into a lower temperature ( Fig. 2A vi  and xii). We also found elevated triclosan levels can boost the size of pF2 colonies, albeit reducing the total number of colonies (Fig. 2B).
However, with these results, it remained unclear whether the enhanced growth suppression effect of mfabI is solely due to its reduced sensitivity to triclosan-G93V mutation disrupts the interaction between FabI and triclosan [15], or the mutation has caused additional functional changes to the enzyme-since fabI also has a mild growth suppression effect on the host that can be reversed by triclosan [17]. To distinguish between these two possibilities, we plated clonal pF or pF2 cultures on antibiotic-free medium. While pF colonies showed reduced size compared to the wild-type cells, pF2 had a much stronger effect (Fig. 2C). These results indicate the mFabI enzyme has gained a hypermorphic or neomorphic function, in addition to its reduced sensitivity to triclosan, that contributes to the strong growth suppression effect.
We also analyzed the growth characteristics of pF and pF2 clones in liquid culture. The growth suppression effect of mfabI and its rescue by triclosan was relatively mild and only evident in early growth phase (Fig. 3A). Lower incubation temperature had a marginal effect on the bacterial growth (Fig. 3A). Plasmid yields were not significantly affected by triclosan concentration, but generally higher at 37uC (Fig. 3B). Plasmids were well maintained at the end of culture in all conditions (Fig. 3C).

Efficient transformation of E. coli by mfabI plasmid
To evaluate mfabI function as a plasmid selection marker, we assessed the transformation efficiency of pF2. When analyzed, pF2 showed transformation efficiency 50% higher than that of pBS (Fig. 4A). We then compared pF2 and pF for their triclosan resistance (Fig. 4B). The transformation efficiency of both pF and pF2 inversely correlates with the triclosan level. This may reflect the increasing bactericidal activity with increasing concentration of triclosan. pF2 consistently showed higher transformation efficiency than pF at any triclosan level examined. Taken together, mfabI appears to be a superior plasmid selection maker with high transformation efficiency and robust triclosan resistance.

Efficient recombineering with mfabI vectors
As our original goal in developing a new plasmid selection marker was to solve recombineering problems, we incorporated mfabI into vectors for recombineering to obtain our desired gene targeting vector. mfabI was used to replace the ampicillin resistance gene in pDTA to create pF2-DTA (Fig. 5A). After correct recombination, the resulting targeting vector should carry both mfabI and Neo, and confer host resistance to both triclosan and kanamycin. Cells were plated with different selective agents to demonstrate improved efficiency with mfabI vectors (Fig. 5B). With ampicillin selection, colonies represent background contamination of the pBS-Insert plasmid. Usually this is not a serious problem, but in our case the restriction enzyme digested insert fragment was large and migrated close to the incompletely digested plasmid in gel electrophoresis, resulting in substantially high number of ampicillin resistant colonies. In the original recombineering scheme, this would interfere with identification of recombined clones, in which pDTA-Rosa26 would also carry ampicillin resistance gene and the double selection with kanamycin and ampicillin would still select for undigested pBS-Insert. When triclosan was applied in the double selection in our improved system, we recovered very few colonies, indicating all the ampicillin resistance background colonies were eliminated. We  screened twelve colonies and identified seven correctly recombined clones (Fig. 5C). The incorrect clones showed restriction enzyme digestion patterns different from that of either pBS-Insert or pF2-DTA-Rosa26, and presumably resulted from aberrantly recombined clones missing unknown regions of the cloned sequences. The insert we constructed contained complex sequence repeats which may have contributed to the high aberrant recombination rates. We also attempted recombineering with pF-DTA-Rosa26, which has the same sequence composition as pF2-DTA-Rosa26 but with the wild-type fabI as the selection marker, but had less successful results (data not shown). Only three out of ten clones screened had the correctly recombined product; the remaining harbored aberrant recombinations. In addition, the yield of DNA from the correct clones was two thirds less than that with mfabI vectors.
Growth advantage conferred by larger mfabI plasmids mfabI appeared to promote preservation of larger constructs, since we had a higher efficiency in obtaining the gene targeting vector with mfabI constructs in our recombineering experiments. To further test the large plasmid stabilization effect of mfabI, we transformed cells with pF2 (3.25 kb), pF2-DTA-Rosa26 (11.2 kb), or pF2-DTA-Rosa26-Insert (21.3 kb) and observed their growth phenotypes. On selection plates, larger colonies were formed with larger constructs, although the differences were reduced at a lower growth temperature (Fig. 6A). In overnight broth culture, higher cell densities were achieved with larger constructs (Fig. 6B); although the DNA yields do not always correlate to cell densities (Fig. 6C). The integrity of larger plasmids was well preserved in the cultures as shown in restriction enzyme digestions (Fig. 6D). In summary, cells transformed with larger mfabI plasmids have a growth advantage on solid medium and in liquid culture. The effect may lead to stabilization of larger plasmids grown in E. coli cells. Figure 4. Superior transformation efficiency and robust triclosan resistance of mfabI. (A) pF2 has a higher transformation efficiency than that of pBS and pF. Equal amounts of plasmid DNA were used to transform DH10B cells, which were then plated on 100 mg/ml ampicillin (for pBS-transformed cells) or 1 mM triclosan plate (for pF and pF2-transformed cells). Transformation efficiency of each plasmid was normalized to that of pBS. Data were collected from three independent replicates (N = 3). pF2 had significantly higher transformation efficiency than that of pBS. (B) More robust triclosan resistance with mfabI plasmid than that with fabI plasmid. DH10B cells were transformed by pF or pF2 and plated with different concentrations of triclosan. Transformation efficiency of each plasmid at each concentration was normalized to that of pF2 at 1 mM triclosan. The graph is generated with data from three independent replicates (N = 3). pF2 shows significantly higher transformation efficiency than that of pF at each concentration tested. *: p,0.05 **: p,0.01 in two-tailed, paired t-test. Error bars show standard deviations. doi:10.1371/journal.pone.0057075.g004

Discussion
Through detailed characterization, we have demonstrated that mfabI is an efficient plasmid selection marker for molecular cloning applications in E. coli. The introduction of a new selection marker will lead to more efficient molecular manipulations and may facilitate more complicated experimental designs. As demonstrated in our recombineering experiment, the employment of mfabIincorporated recombineering vectors solved the low efficiency problem we encountered associated with high background contamination of undigested ampicillin plasmid, which is significantly better than the original recombineering strategy using only two plasmid selection markers. A practical and valuable advantage with mfabI marker is the low cost of its selective agent triclosan, making it easily applicable in any laboratory.
It is not surprisingly that the mfabI marker shows stronger resistance to triclosan than fabI, since the G93V point mutation was found in the triclosan-resistant E. coli mutant genome. However the stronger growth suppression effect of mfabI plasmid on the host was unexpected. Even when grown in the absence of triclosan, the difference between mfabI and fabI-transformed colonies was dramatic, suggesting the G93V mutation changes the property of the FabI enzyme beyond merely interrupting its interaction with triclosan. The boost of colony size by higher concentrations of triclosan indicated the G93V mutation may enhance enzymatic activity. The mutation may stabilize mFabI protein or render it resistant to cellular regulation, and the resulting elevated or uncontrolled enzymatic activity affects bacterial growth more than wild-type FabI. Lower temperature may reduce the mFabI enzymatic activity or ease cellular stress, and therefore alleviate the growth phenotype. In practical applications, 32uC incubation should be used for positive selection. The strong growth suppression effect in 37uC suggests mfabI's potential use as a negative selection marker.
In the generation of our gene-targeting vector via recombineering, mfabI appears to help retain large and complex inserts. Furthermore, cells transformed with larger mfabI plasmids with more complex sequences grow better than that with a smaller mfabI vector in both solid and liquid media. Based on these results, we propose a model of strong growth suppression activity of mfabI leading to suppression of unwanted aberrant sequence rearrangement events, which happens frequently in cloning of large or complex sequences. In the original recombineering scheme (Fig. 7A), both ampicillin and kanamycin resistance markers promote host cell growth under selection. When the desired plasmid contains complex sequences that do not replicate efficiently, the selection pressure favors loss of sequences, thus producing more resistance gene products. In the mFabI-incorporated system (Fig. 7B), although mFabI product is required for host cell survival in triclosan selection, it also suppresses host cell growth. Aberrant rearrangement events that remove unstable sequence and thus enhance plasmid replication/mFabI production would have a negative impact on cell growth, and thus are disfavored in the selection, thereby becoming less frequent. This mutation suppression effect associated with mfabI should also be applicable in cloning any large complex sequences by other methods. Figure 7. Suppression of undesired sequence rearrangement events with mfabI marker in the recombineering system. (A) In the original recombineering system, undesired sequence rearrangement events are favored in the selection because they promote cell growth. (B) In the recombineering system with mfabI marker, undesired sequence rearrangements are suppressed, since increased mFabI activity can suppress cell growth. doi:10.1371/journal.pone.0057075.g007