ABSTRACT
Targeted mutagenesis is an extremely useful experimental approach in molecular medicine, allowing the generation of specialized animals that are mutant for any gene of interest. Currently the rate determining step in any gene targeting experiment is construction of the targeting vector (TV). In order to streamline gene targeting methods and avoid problems encountered with plasmid TVs, we describe the direct application of [lambda] phage in targeted mutagenesis. The recombination-proficient phage vector [lambda]2TK permits generation of TVs by conventional restriction-ligation or recombination-mediated methods. The resulting [lambda]TV DNA can then be cleaved with restriction endonucleases to release the bacteriophage arms and can subsequently be electroporated directly into ES cells to yield gene targets. We demonstrate that in vivo phage-plasmid recombination can be used to introduce neo and lacZ-neo mutations into precise positions within a [lambda]2TK subclone via double crossover recombination. We describe two methods for eliminating single crossover recombinants, spi selection and size restriction, both of which result in phage TVs bearing double crossover insertions. Thus TVs can be easily and quickly generated in bacteriophage without plasmid subcloning and with little genomic sequence or restriction site information.
Targeted mutagenesis allows specific mutations to be engineered into the mouse germline via homologous recombination of exogenously altered DNA in embryonic stem (ES) cells (1,2). Using this technology, the function of any cloned gene may be examined by its disruption in mice. Thus gene targeting is a critical experiment in molecular medicine, enabling mimicry of human mutations in the mouse for generation of experimental therapeutic models (3).
The original and still the most prevalent gene targeting approach,`the knock-out', uses a replacement vector to direct a positive selectable marker (i.e. neomycin resistance) into a specific chromosomal location via either double reciprocal exchange or gene conversion (4). However, many sophisticated variations on this original technique have become available, including the generation of point mutations, deletionsand translocations and gene substitutions (5-9). Further, the application of cre recombinase from bacteriophage P1 allows additional genomic alterations at loxP target sequences following gene targeting so that mutations can be made tissue- or development-specific (10).
Although targeted mutagenesis provides a powerful tool for analysis of gene function, it is a complex and time consuming procedure. While methods of improving the efficiency of generating targeted ES cell lines (11) and mutant mice (12) have become available, little has been done to streamline construction of the targeting vector (TV). In general, two types of problems are encountered in construction of TVs. First, specific genomic regions undergo rearrangements in plasmid vectors and are difficult to clone either on their own or in combination with the neo and tk selectable marker genes. Second, the use of large genomic fragments in TVs often limits the choice of unique restriction sites available for inserting foreign DNA fragments, such as the neo or lacZ-neo gene cassettes, for modifying function of the test gene. Although both these problems may be alleviated by choosing a smaller genomic region on which to base the TV, the reduction in homology in the vector will likely lower the gene targeting frequency(13).
In order to obviate the difficulties with TV construction in plasmids, we describe the direct application of [lambda] phage in targeted mutagenesis. Using a recombination-proficient bacteriophage, [lambda]2TK, we are able to introduce neo and lacZ-neo mutations into precise positions within genomic subclones via double crossover recombination. In this approach the efficiency of TV construction is improved, as it is not dependent on restriction site availability. Furthermore, we demonstrate that bacteriophage TVs can be directly introduced into ES cells to yield targeted clones, thus avoiding the use of plasmid TVs altogether.
Escherichia coli strains MC1061 (rec+, sup0) and its P2 lysogenic derivative P2MC1061, as well as MC1061[P3], were kindly provided by Dr D.M.Kurnit (University of Michigan, Ann Arbor, MI). LE392 (rec+, supE, supF) and its P2 lysogenic derivative P2392 were obtained from Stratagene (La Jolla).
For routine cloning in phage, [lambda] packaging extracts (Amersham) were split into thirds and either used directly or re-frozen on dry ice for later use. The gene targeting bacteriophage vector [lambda]2TK (Fig. 2A) was preceded by construction of a [lambda]Dash II (Stratagene) derivative containing thymidine kinase genesfrom Herpes simplex viruses 1 and 2 (HSVtk1and tk2; 11,14) in the left and right polylinkers respectively, between NotI and XhoI sites (not shown). A SalI fragment comprising the two tk genes and stuffer region was then transferred into [lambda]Syrinx2A (15) to place the tk1 and tk2 genes adjacent to the short and long arms of the phage respectively. Finally, a XhoI fragment containing the polylinker and stuffer region of [lambda]Gem11 (Promega) was shuffled into this derivative to yield [lambda]2TK (Fig. 2A). The bacteriophage vector used for plasmid-phage recombination, [lambda]2TK:CRABPI (Fig. 2B), was generated by inserting into [lambda]2TK a 9.5 kb XhoI fragment encompassing the first and second exons of the murine cellular retinoic acid binding protein I gene (cRABPI; 16).
All of the plasmid constructs used in this study were harbored in the rec+ host MC1061. Plasmids bearing the supF gene were harbored in MC1061[P3]. The P3 episome, which is Kanr, Ampam, Tetam, facilitates selection of supF in media containing kanamycin, ampicillin and tetracycline (17). Since all of the plasmids used in this study were ampicillin resistant, only tetracycline served to functionally select for the presence of supF. The plasmid pGam was generated by excising a 500 bp SalI fragment containing the gam gene coding sequence from bacteriophage [lambda] (18) and inserting it into the SalI site of pBluescriptKS+ (Stratagene). spi+ (sensitive to P2 interference) activity (18,19) conferred by pGam was assessed initially by its inability to be grown in red-, gam- recombinant [lambda] phage plated on P2392.
Construction of the recombination plasmids used in this study (Fig. 3C) was preceded by the assembly of pCRABPIBg/Sp, which is a small 400 bp BglII-SpeI genomic subclone that encompasses the second exon of cRABPI. An SstI site which bisects this genomic fragment was converted to BamHI in order to facilitate cloning of supF-bearing cassettes of MC1neopA (4) and lacZ-MC1neo pA (5) in pCRABPI:neoF and pCRABPI:lacneoF respectively. pGamCRABPI:neoF is the pGam derivative of pCRABPI:neoF.
[lambda]2TK:CRABPI phage (1 × 103 p.f.u.) were passaged via plate lysate over LE392 or MC1061[P3], bearing either pGamCRABPI:neoF or pCRABPI:lacneoF, overnight at 37°C. Supernatant phage were collected the following day in phage dilution buffer and used to infect the indicator strain LE392, MC1061 or P2MC1061. In experiments using pCRABPI:neoF all phage-plasmid recombinants (both double and single crossovers) were scored on MC1061, while double crossovers were scored on P2MC1061. Due to the size limitation of [lambda] phage packaging, only double crossovers could be scored on MC1061 in experiments with pCRABPI:lacneoF. The structure of the resulting recombinants was confirmed by restriction analysis of small scale phage DNA preparations.
Phage targeting vectors were introduced into ES cells by electroporation, as previously described for plasmid targeting vectors (4). Prior to electroporation, the arms of the [lambda] targeting vector were removed by digestion with NotIand the DNA concentration was estimated on the basis of insert only. Targeted cell lines were enriched by positive-negative selection (11) using FIAU (1-[2'-deoxy-2'-fluoro-1-[beta]-D-arabinofuranosyl]-5-iodouridine) instead of gancyclovir for negative selection (20). Clones bearing targeting events were identified by genomic Southern blotting using a probe which flanked the 5'-end of the TV (4). Integrity of the targeted locus was confirmed using a probe internal to the locus which spanned the site of the neo gene insertion.
We have observed that during TV construction certain genomic regions are not amenable to cloning in plasmids, either on their own or in association with neo or tk selectable markers. This cloning difficulty is likely a combination of both the large homology requirement for the TV and addition of difficult regions bearing repetitive sequences. Although we have found that TV stability in plasmids can be augmented by shortening of the homology length or removal of certain difficult regions, we are concerned that either manipulation of the homology region could have serious consequences on targeting efficiency (4,13). We have found, with little exception, that unstable plasmid regions can be maintained in bacteriophage [lambda]. We and others have used [lambda] phage previously to assemble TVs that are subsequently `rolled out' of phage in plasmid form (21,22). In this approach tk and plasmid sequences occur on either side of phage arms such that following construction a plasmid TV is excised from the phage using rare restriction sites or cre recombinase. While this latter approach may be useful in some cases, the use of cre recombinase in TV construction may preclude its subsequent application at the genomic level following gene targeting. In this report we demonstrate that generation of plasmid TVs from phage precursors is redundant, as phage TVs may be used directly in gene targeting. With the elimination of plasmid sequences from the phage TV we demonstrate that double crossover phage-plasmid recombination can be used to direct gene targeting mutations, such as neo or lacZ-neo insertions into specific sites within [lambda] phage vectors, without the use of restriction enzymes.
Previously phage-plasmid recombination has been used to isolate [lambda] phage from genomic libraries by recombination screening (15,17). In this method a [lambda] genomic library (bearing amber mutations) is passaged over a rec+ bacterial strain bearing a small supF (amber suppressing) recombination plasmid. Homology in the recombination plasmid, usually derived from a cDNA sequence, directs the plasmid to integrate into the phage by single crossover, thereby generating supF-bearing phage recombinants capable of growing on a suppressor-free (sup0) host. Depending on homology length, the recombination plasmid can integrate at a frequency of ~10-2. While we considered the possibility that phage TVs could be constructed by single crossover recombination, we were concerned that condensation of this complex could result in reversion of any phenotype generated by the neo or lacZ-neo insertion. Therefore, we sought to introduce more stable changes in phage targeting vectors by selecting for double crossover events that introduce neo or lacZ-neo mutations by replacement.
Two strategies were tested to see whether double crossover events could be enriched by eliminating single crossover phage (Fig. 1). In the first approach a spi selection system was devised to eliminate phage that have integrated the entire recombination plasmid (Fig. 1A). [lambda] phage that carry a functional copy of the gam gene are sensitive to P2 interference (spi+) and cannot be replicated on a P2 lysogenic host (18,19). Thus, if gam function could be maintained in a recombination plasmid, single crossover integrants bearing gam could be eliminated on a P2 lysogen. This approach was used to make a neo `knock-out' TV. The second strategy took advantage of the size limitation in packaging [lambda] phage (23; Fig. 1B). Thus, if a relatively large recombination plasmid was used in the reaction only double crossover phage could be propagated following recombination. This approach was used to make a lacZ-neo `knock-in' TV.
[lambda]2TK (Fig. 2A) is based on the recombination screening vector [lambda]Syrinx2A (15) but differs from that vector in having a gam+ stuffer fragment to enable spi- selection of cloned inserts (18) and HSVtk1 and -tk2 genes adjacent to its small and large arms respectively. [The tk1 and tk2 genes serve as negative selectable makers in gene targeting experiments (11).] Otherwise, as with [lambda]Syrinx2A, [lambda]2TK is Aam, Bam, Sam, requires the amber suppressor tRNA, supF, for lytic growth and is Rap+ for efficient recombination with the plasmid. The vector has unique XbaI and XhoI restriction sites for subcloning of relatively longer genomic pieces and can receive fragments ranging in size from 3.5 to 16.5 kb. Since the XbaI site is compatible with SpeI and AvrII, while XhoI can accommodate SalI, in addition to BamHI, BglII, BclI and Sau3A by partial fill-in of cohesive ends, there is a considerable amount of flexibility in subcloning of genomic regions.
Two basic approaches can be used to construct gene targeting vectors in [lambda]2TK. One approach (not shown) relies on our observation that [lambda]2TK, unlike plasmid TV constructs, are amenable to multiple partner ligations. Thus a TV is generated in one step by ligating a 5' genomic fragment, a neo cassette and a 3' genomic fragment into [lambda]2TK. If supF derivatives of neo or lacZ-neo are used in ligation, selection of correct recombinants is facilitated by plating [lambda]2TK phage on a supF- host, such as MC1061. In a second approach (see below) an 8-10 kb genomic fragment is subcloned into [lambda]2TK and a neo or lacZ-neo cassette is converted into a specific site via in vivo recombination.
Recombination experiments were preceded by construction of a [lambda]2TK phage subclone and a number of test recombination plasmids (Fig. 2B and C). As a model for these experiments we used the gene cRABPI, which has been previously disrupted in mice (24). [lambda]2TK:CRABPI (Fig. 3B) contains a 9.5 kb isogenic genomic fragment bearing the second and third exons of cRABPI (16). The homology region used in the neo and lacZ-neo recombination plasmids (Fig. 2C) was derived from a 400 bp SpeI-BglII genomic fragment bearing exon 2 of cRABPI. In pCRABPISp/Bg an SstI site which bisects this fragment within the coding sequence was converted to BamHI for convenient cloning of the supF-bearing cassettes of MC1neopA and lacZ-MC1neo pA.
For the recombination strategy which utilized [lambda]gam and spi selection to eliminate single crossover phage the homology and neoF regions were subcloned in pGam to generate pGamCRABPI:neoF. pGam is a pBluescript derivative that carries a functional [lambda]gam gene and can confer a spi+ phenotype to recombinant phage. For the strategy involving size limitation of [lambda] packaging pCRABPI:lacneoF was constructed, which placed a lacZ-neo-supF cassette in-frame within pCRABPISp/Bg.As this plasmid was 8.7 kb in length, it was theoretically too large to integrate into the vector by single crossover. The 5.7 kb lacZ-neo-supF portion, however, was sufficiently small to be converted into the [lambda]2TK:CRABPI TV by double crossover.
[lambda]2TK:CRABPI phage was passaged via plate lysates through MC1061[P3] bearing pCRABPI: neoF, pGamCRABPI:neoF or pCRABPI:lacZneoF, as well as a supF+ control strain LE392. The resultant phage were plated on LE392, in order to estimate the titer of harvested phage, and on the restrictive host MC1061, to evaluate the titer of supF+ recombinant phage. In addition, to confirm gam gene function in pGamCRABPI:neoF, the supernatants were plated on a P2 lysogen of MC1061. As shown in Table 1, the passage of [lambda]2TK:CRABPI through LE392 resulted in no detectable supF+ phage in 106 that were plated. In contrast, phage passaged through strains bearing the recombination plasmids did result in supF+ recombinants, albeit at differing frequencies. Passage through MC1061[P3, pCRABPI:neoF] resulted in roughly 7% recombinants, as indicated by supF+ phage grown on MC1061. Passage through MC1061[P3, pGamCRABPI:neoF] resulted in ~60% fewer recombinants on MC1061. This result was not unexpected, since gam creates a recBC phenocopy in E.coli (19) and phage-plasmid recombination is reduced in recBC- mutants (25). spi selection appeared to be able to eliminate single crossover phage in this experiment as only10% of these recombinants could grow on the P2 lysogen of MC1061. This result indicated that the occurrence of double crossovers between plasmid and phage is not a product of two independent single crossover events (i.e. 10-2 × 10-2) but instead is a cooperative event.
Double crossover recombinants resulting from passage of [lambda]2TK:CRABPI through MC1061[P3, pCRABPI:lacneoF] occurred at a frequency of ~5 × 10-5. This number was found to be the same when plated on MC1061 and P2MC1061, indicating that, as with pCRABPI:neoF, the P2 lysogen did not affect the plating efficiency of recombinants not carrying the gam gene. Interestingly, when the distance between the homologous sequences is increased from 1.4 to 5.7 kb the frequency of double crossover recombination drops by almost two orders of magnitude. This phenomenon is different from what is seen in gene targeting of mammalian cells, where insert size appears to have little influence on gene targeting frequency (5). Perhaps in phage-plasmid recombination the cooperativity observed in double crossover is abolished by steric hindrance when longer distances occur between recombination sites.
To establish the authenticity of the apparent double crossover recombinants in both experiments, individual plaques were isolated from P2MC1061 plates and phage were grown on a small scale. Restriction analysis of 15 phage clones using BamHI and XhoI indicated that in all cases a double crossover had indeed occurred (Fig. 3). For neo-supF the 2.2 kb XhoI-BamHI fragment from [lambda]2TK:CRABPI is present as two bands as a result of the XhoI sites flanking the supF gene. The 5' fragment which contains the neo gene is 2.3 kb in length, while the 3' fragment has been shifted down to 0.9 kb. Similarly, for lacZ-neo-supF the 2.2 kb parental fragment is shifted to two bands, however, in this case the 5' fragment is shifted up to 6.5 kb as a result of the large lacZ-neo insertion, while the 3' fragment (0.9 kb) is common in both the cRABPI:neoF and cRABPI:lacneoF targeting vectors. The other common fragments (4.7, 1.8 and 1.5 kb) lie outside the recombination site and are common to [lambda]2TK:CRABPI and its neoF and lacneoF derivatives.
These experiments illustrate that flanking homologous sequences as short as 200 bp are sufficient to introduce heterologous sequences ranging from 1.4 to 5.7 kb into phage. We have also discovered that flanking homologous sequences derived from oligonucleotides as short as 25 bp are sufficient to direct double crossover recombination of neo-supF cassettes in phage TVs, albeit at a lower frequency of 10-7 (D.E.Rancourt, unpublished results). This approach, however, has failed to direct double crossover recombination of lacZ-neo-supF cassettes in phage.
To demonstrate that phage vectors could be used successfully in gene targeting experiments, double crossover clones bearing neo ([lambda]2TK:CRABPI:neoF) and lacZ-neo ([lambda]2TK:CRABPI:lacneoF) were grown on a large scale and prepared for electroporation into ES cells by excising the phage arms with NotI. Following electroporation cells were grown in medium containing G418 and FIAU (4,20). Individual clones resistant to both drugs were subjected to genomic Southern blotting analysis, using restriction enzymes and probes that distinguish random from homologous integration events. For [lambda]2TK:CRABPI:neoF seven out of 96 cell lines had neo integrated within the cRABPI locus, while for [lambda]2TK:CRABPI:lacneoF three out of 40 were positively targeted. Southern blot analysis of representative cell lines is shown in Figure 4. Parental ES cell DNA digested with EcoRI and hybridized with a 5'-flanking probe yielded a 4.9 kb fragment. In cell lines targeted with neo a portion of this hybridizing band representing the targeted locus was shifted down to 3.4 kb due to the presence of an EcoRI site in the neo gene cassette. With the lacZ-neo targets this hybridizing band shifted up to 7.8 kb because of lacZ sequences upstream of the EcoRI site in neo.
Table 1
We thank Dr Mario R.Capecchi for allowing us to initiate this research independently and acknowledge the technical support of Carol Lenz and Margarie Allen. We are grateful to Drs Gerard Bain, Kostas Iatrou and Karl Riabowol for critical reading of this manuscript. This work was supported, in part, by Grants-in-Aid from the Ministry of Education, Science, Sports and Culture of Japan International Scientific Research Program (08044303), the Alberta Cancer Board and University of Calgary Research Foundation.
Nucleic Acids Research
Pages
Introduction
Materials And Methods
Bacterial strains, bacteriophage and plasmids
Phage-plasmid recombination
Gene targeting
Results And Discussion
Bacteriophage gene targeting vectors
Strategies for generating targeting vectors via phage-plasmid recombination
[lambda]2TK in phage targeting vector construction
neo and lacZ-neo[lambda]2TK TVs via double crossover recombination
Gene targeting with phage replacement vectors
Acknowledgements
References
Original passage
Recombination frequencies(secondary passage)
MC1061
P2MC1061
LE392
0
0
MC1061[P3, pCRABP:neoF]
7.1 × 10-2
7.1 × 10-2
MC1061[P3, pGamCRABP:neoF]
2.5 × 10-2
3.4 × 10-3
MC1061[P3, pCRABP:lacneoF]
5.1 × 10-5
5.1 × 10-5
REFERENCES
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 6 Feb 1998
Copyright© Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. Ohno, N. Ono, F. Seki, M. Takeda, S. Kura, T. Tsuzuki, and Y. Yanagi Measles Virus Infection of SLAM (CD150) Knockin Mice Reproduces Tropism and Immunosuppression in Human Infection J. Virol., February 15, 2007; 81(4): 1650 - 1659. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 X.-F. Li, L. Kiedrowski, F. Tremblay, F. R. Fernandez, M. Perizzolo, R. J. Winkfein, R. W. Turner, J. S. Bains, D. E. Rancourt, and J. Lytton Importance of K+-dependent Na+/Ca2+-exchanger 2, NCKX2, in Motor Learning and Memory J. Biol. Chem., March 10, 2006; 281(10): 6273 - 6282. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. P. Lees-Miller, J. Guo, J. R. Somers, D. E. Roach, R. S. Sheldon, D. E. Rancourt, and H. J. Duff Selective Knockout of Mouse ERG1 B Potassium Channel Eliminates IKr in Adult Ventricular Myocytes and Elicits Episodes of Abrupt Sinus Bradycardia Mol. Cell. Biol., March 15, 2003; 23(6): 1856 - 1862. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Xue, H. A. Tarnasky, D. E. Rancourt, and F. A. van der Hoorn Targeted Disruption of the Testicular SPAG5/Deepest Protein Does Not Affect Spermatogenesis or Fertility Mol. Cell. Biol., April 1, 2002; 22(7): 1993 - 1997. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Woltjen, G. Bain, and D. E. Rancourt Retro-recombination screening of a mouse embryonic stem cell genomic library Nucleic Acids Res., May 1, 2000; 28(9): e41 - e41. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||