Introduction of precise alterations into the mouse genome with high efficiency by stable tag-exchange gene targeting: implications for gene targeting in ES cells
Introduction of precise alterations into the mouse genome with high efficiency by stable tag-exchange gene targeting: implications for gene targeting in ES cells
Linda M. Whyatt, Peter D. Rathjen*
Department of Biochemistry, The University of Adelaide, Adelaide, SA 5005, Australia
Received March 7, 1997;Revised and Accepted May 6, 1997
ABSTRACT
The efficiency of tag-and-exchange gene targeting approaches for the introduction of precise genomic modifications is compromised by high levels of non-homologous recombinants which survive selection due to loss of tag gene expression, often by physical loss of the tag gene. We describe a modified approach, termed stable tag-exchange, which incorporates an additional positive selection (stability) cassette to circumvent this limitation. HPRT (tag) and neo (stability) cassettes, separated by 4.9 kb of homologous DNA, were introduced efficiently into the LIF locus of ES cells. The tag cassette was substituted for a [beta]-galactosidase gene in exchange step targeting. Direct comparison of the tag-and-exchange and stable tag-exchange approaches indicated respective targeting efficiencies of 21% and 88%. The increased stable tag-exchange targeting efficiency resulted from elimination of >75% of background lines which survived tag-and-exchange selection due to physical loss of the tag gene. These resulted from reversion of the tagged allele to wild-type which is therefore a major contributor to tag-and-exchange targeting background. Our results extend the application of gene targeting by demonstrating a rationale for single-step integration of multiple regions of extended non-homology, and providing an efficient system for the repeated introduction of precise alterations into the mammalian genome.
INTRODUCTION
The technique of gene targeting (1) in embryonic stem (ES) cells(2,3) allows direct genetic analysis of the mammal by introduction of predetermined alterations into the mouse genome and subsequently germline (reviewed in 4,5). Comprehensive investigation of gene action is anticipated to require the introduction of multiple, distinct, precise mutations into loci of interest. Two step targeting systems known as double-replacement (6,7) or tag-and-exchange (8) permit the introduction of subtle mutations into endogenous loci. These are based on tag selection cassettes, such as hypoxanthine ribosyl transferase (HPRT) (9), for which both positive and negative selection can be applied in ES cells. Positive selection for the tag gene is applied during introduction of this cassette into the locus of interest by homologous recombination (the tag step). In a second round of gene targeting (the exchange step), a precise alteration is introduced into the tagged locus by homologous recombination, resulting in displacement of the tag cassette. Direct (negative) selection for loss of tag cassette expression is used to eliminate non-homologous recombinants. In practise the efficiency of the exchange step has been compromised by a high background of non-targeted lines which survive selection due to loss of tag cassette expression (8) or physical loss of the tag cassette in the absence of homologous recombination (6,7,10). This reduces the efficiency of exchange step targeting, in some cases to below that of tag step targeting (7). Thus, although tag-and-exchange targeting allows the generation of precise genomic modifications in ES cells, the relatively low frequencies of homologous recombinants which survive selection make this strategy impractical for the repeated introduction of independent mutations to a single locus.
The limitations that this imposes on the effective use of gene targeting, and the requirement for development of targeting strategies which eliminate the non-targeted background, have been addressed in detail by Detloff etal. (11). We have designed and characterised a modified tag-and-exchange gene targeting system, termed stable tag-exchange (Fig. 1a), which was designed to improve the efficiency of exchange step gene targeting. A positive selection cassette (stability cassette) is included in the tag step vector and integrated into the endogenous locus during tag step targeting (Fig, 1a; ii, iii). The stability cassette, ideally positioned outside the locus to be altered, remains on the tagged allele during exchange step gene targeting. Selection is applied against the tag cassette and for the stability cassette during exchange step targeting (Fig. 1a; iv, v). In comparison to the tag-and-exchange protocol, selection for the stability cassette eliminates that proportion of selected cell lines which results from physical loss of the tag gene by chromosomal alteration/loss in the absence of homologous recombination.
The stable tag-exchange gene targeting approach was characterised at the leukaemia inhibitory factor (LIF) locus in ES cells. The product of the LIF gene is a pleiotropic cytokine which has been implicated in a range of cellular processes (reviewed in 12,13). A [beta]-galactosidase marker gene was introduced into exon 1D of the endogenous locus (14).
Characterisation of the stable tag-exchange gene targeting system yielded three important findings with respect to the generic design of targeting vectors. Firstly, we demonstrated for the first time that multiple regions of extensive non-homology, in this case >1 kb, can be integrated efficiently into the mammalian genome by homologous recombination using a single replacement gene targeting vector. This has important implications for the design of multi-gene targeting vectors and for our understanding of the mechanism of homologous recombination in mammalian cells. Secondly, we found that a major mechanism of tag gene mutation during exchange step targeting was reversion of the tagged allele to wild-type, possibly as a result of non-reciprocal genetic exchange with the wild-type allele. Finally, the stable tag-exchange system was shown to increase considerably the efficiency with which precise alterations can be introduced into the mouse genome. As such it provides a valuable tool for the sophisticated analysis of mammalian gene function.
MATERIALS AND METHODS
Construction of plasmids
A 9.5 kb BglII/HindIII fragment of the mouse LIF gene isolated from a strain 129Sv genomic library (15) was used as the basis for all targeting vectors. The BglII/HindIII fragment was endfilled and cloned into the SmaI site of pT7T3 19U (Stratagene), from which the XbaI site had been removed by endfilling with Klenow fragment. Both the BglII and HindIII sites were lost in this manipulation. The resultant vector pLIFXT7 contained the LIF clone positioned with the 5[prime] end of the LIF gene closest to the HindIII site of the polylinker. The neomycin resistance (neo) gene from pgkNeopA (16) was isolated as an EcoRI/BglII fragment, endfilled and cloned into a Klenow fragment endfilled MluI site contained within the 3[prime] non-translated region of the LIF gene, in the same transcriptional orientation as LIF. The HPRT gene from pnI2(I1s) (9), isolated as a BamHI fragment, was endfilled and ligated into a T4 DNA polymerase blunted SacII site between the alternative LIF D and M first exons (14) in the same transcriptional orientation as the LIF gene to produce the tag step targeting vector pHNLX (Fig. 1b). The HPRT gene in pHNLX was under the transcriptional control of the 5[prime] thymidine kinase (TK) promoter and mutant polyoma enhancer, while the neo gene was expressed from the mouse phosphoglycerate kinase-1 (pgk-1) promoter. All nucleic acid manipulations were carried out as described by Sambrook etal. (17).
Vector pnlacF (kindly provided by Dr Jacques Peschon) was used for the construction of a promoterless, ATG-, polyA+ [beta]-galactosidase reporter cassette which encodes a nuclear localised form of the [beta]-galactosidase gene suitable for use in ES cells (18). The [beta]-galactosidase gene in pnlacF was altered by PCR such that it was contained on a BglII fragment and lacked an initiation codon (18). A unique BglII site was introduced into codon six of LIF exon 1D in pLIFXT7. The [beta]-galactosidase reporter cassette was cloned in frame into the BglII site of pLIFXT7 to produce the targeting vector p1DLacLIF (Fig. 3a). Expression of [beta]-galactosidase is therefore directed by the ATG of the LIF 1D exon.
ES cell culture and electroporation
The feeder independent, HPRT- ES cell line E14TG2a (19) used for gene targeting was kindly provided by Dr Austin Smith. E14TG2a and targeted derivatives were maintained in non-purified LIF in the absence of feeder cells as previously described (20), with the exception that complete ES cell medium containing 15% FCS was used.
Tag step gene targeting with pHNLX
Targeting was carried out using the method of Mountford etal. (15). Electroporation (Bio-Rad Genepulser 0.8 kV, 3 [mu]F in a 0.4 cm cuvette) was carried out with 108 ES cells using 150 [mu]g of SalI linearised pHNLX DNA or no DNA as a control in a total volume of 0.8 ml. One half of the cells transfected with pHNLX were plated into 10 cm diameter plates at a density of 5 × 106 cells/plate. Selection in G418 (175 [mu]g/ml), HAT (100 [mu]M hypoxanthine, 0.4 [mu]M aminopterin, 16 [mu]M thymidine), or HAT + G418 was applied after 24 h and colonies were picked after 7 days.
Exchange step gene targeting with p1DLacLIF
Tag step targeted HG60 ES cells were grown in the absence of selection for 4 days prior to electroporation as previously described (6). Cells were electroporated (as described above) with p1DLacLIF DNA linearised by NheI digestion, or pCH110 (Pharmacia) DNA linearised by BamHI digestion as a control.
Selection with 6-TG must be carried out at low cell density to prevent inappropriate loss of HPRT- ES cells by metabolic co-operation (21). In addition to this, 5-6 days growth in the absence of selection is required after transfection to allow breakdown of residual HPRT activity before 6-TG selection can be applied. This is normally achieved by passaging the cells prior to and during the 6-TG selection process. To allow the analysis of independent targeting events, cell passage after transfection was inappropriate. Selection conditions were therefore established to allow growth for sufficient time without contact between independent colonies.
For the analysis of independent targeting events, 5% of the cells transfected with p1DLacLIF were plated into 150 cm diameter plates at a density of 2.5 × 105 cells/plate and grown in the absence of selection for 5 days without further passaging. Selection with 6-TG (20 [mu]M) or 6-TG + G418 (175 [mu]g/ml) was applied and resistant colonies were picked after 14 days of selection. In a second experiment, in which independent targeting events were not required, one half of the transfected cells were grown in 150 cm2 flasks at a density of 1 × 107cells/flask for 3 days before passaging. Cells were grown for a further 3 days before 1.5 × 107 cells from each of the five pools were seeded into 10 cm diameter plates at a density of 1.5 × 106 cells/plate in 6-TG or 6-TG + G418 selective medium as above. Selective medium was replaced when required and resistant colonies picked after 14 days.
Figure 1 (a) Schematic diagram illustrating the stable tag-exchange gene targeting approach. In the tag step of the stable tag-exchange approach a tag (HPRT) cassette is introduced into the endogenous locus and a stability (neo) cassette is introduced downstream of the locus (i-iii). Selection in HAT + G418 during exchange step targeting (iv) selects simultaneously for replacement of the tag cassette and maintenance of the targeted allele. This overcomes a background of HATr lines which result from physical loss of the HPRT gene in the absence of homologous recombination. The exons of the gene to be altered (closed boxes labelled 1-4), the HPRT tag cassette (grey box), the neo stability cassette (hatched box) and the precise alteration (open box) are indicated. (b-d) Schematic diagrams illustrating tag step gene targeting at the LIF locus. (b) The tag step gene targeting vector pHNLX showing the SalI site used for linearisation and the EcoRI and HindIII sites used for screening targeted lines. (c) Tag step gene targeting at the LIF locus using vector pHNLX. The genomic locus and tag step targeting vector are illustrated. (d) Restriction maps of the wild-type and tag step targeted LIF alleles. The restriction fragments resulting from digestion of the wild-type (i) and tag step targeted (ii) LIF alleles with EcoRI and HindIII are shown. The probes used for Southern analysis (grey boxes 1-4) are indicated. Probe 1, a 5[prime] external LIF probe, was used to screen for targeting at the LIF locus. This probe detects bands of 10.4 and 2.7 kb from the untargeted allele, and 2.7 and 6.5 kb from the tag step targeted allele. Probes 2 and 3 were used to screen for multiple integrations of the targeting vector and to verify the presence of the neo cassette in targeted lines, respectively (results not shown). Probe 4 is a 3[prime] internal LIF probe used to screen correct integration at the 3[prime] end. Loss of the 3[prime] LIF HindIII site during construction of pHNLX means that this probe detects bands of 10.4 kb from the untargeted allele and 1.8 kb from the correctly targeted allele. The exons of LIF (closed boxes) with the first alternate exons of LIF labelled D, M and T (Haines etal., manuscript in preparation), the 3[prime] UTR (thick line), the HPRT and neo cassettes (open boxes), genomic sequence (line) and non-homologous plasmid DNA (dashed line) are indicated.
Southern analysis
Approximately 50% of the DNA from a confluent well of a 24-well plate, 20% of the DNA from a confluent 6-well plate (22) or 10-20 [mu]g DNA from mouse tail biopsies (23) was digested with the appropriate restriction enzyme(s) and run on 0.8% or 1.0% TAE gels. Southern blotting and hybridisation were carried out as described previously (24) except that Hybond N+ filters (Amersham) were used. DNA probes were made with the Gigaprime labelling kit (Bresatec) and [[alpha]-32P]dATP (Bresatec). Filters were washed in 2× SSC/0.1%SDS, 0.5× SSC/0.1%SDS and 0.1× SSC/0.1%SDS for 30 min each.
Karyotypic and germline transmission analysis
Chromosome slides were prepared from ES cell lines by the method of Robertson (25). The chromosomes were Q banded (26) for 30-45 s using a 0.002% solution of quinocrine mustard dihydrochloride in water, rinsed with distilled water and mounted in 0.067 M Sorensen buffer at pH 6.8 (D. Maher, personal communication).
Targeted ES cells were reintroduced into CBA F2 or C57BL F1 (C57BL/6 × C57BL/10) blastocysts by microinjection using established procedures (23,25). E14TG2a ES cells are homozygous cch/cch so that chimaerism was readily detected by the presence of sandy coat colour pigmentation. Gpi-1s analysis of sperm obtained from the epididymis of male chimearae produced with tag step targeted ES cells was carried out to assess germline transmission of the ES cell genotype using the Super Z applicator kit (Helena Laboratories). Chimaeric males produced with exchange step targeted ES cells were crossed to BALB/c females to assess germline transmission of the ES cell genotype.
RESULTS
Tag step gene targeting with pHNLX: efficient introduction of multiple, distinct regions of non-homology into the genome by homologous recombination
The stable tag-and-exchange protocol (Fig. 1a) requires introduction of both a positive/negative selection cassette (tag cassette) within the locus to be modified, and a positive selection cassette (stability cassette) outside this locus by tag step targeting. We used a tag step targeting vector carrying a 6.1 kb HPRT (9) tag cassette and a 1.9 kb neo stability cassette separated by 4.9 kb of homologous DNA. First round targeting, resulting in the generation of cell lines resistant to HAT and G418, was therefore dependent on the introduction of two separate regions of substantial non-homology, 6.1 and 1.9 kb, into the genome by homologous recombination.
The tag step targeting vector pHNLX (Fig. 1b) carried a 9.5 kb LIF genomic clone which contained the entire LIF coding region and transcriptional unit. The HPRT positive/negative selection cassette was positioned between the alternative D and M first exons of the LIF gene (14), and the pgk-neo positive selection cassette was positioned within the 3[prime] untranslated region of the LIF gene. pHNLX therefore contained three distinct regions of homologous LIF genomic DNA: 2.8 kb upstream of the HPRT cassette, 1.8 kb downstream of the neo cassette and 4.9 kb between the cassettes (Fig. 1b and c).
ES cells (108) were transfected with SalI linearised pHNLX DNA or no DNA as a control. Cells transfected with pHNLX were selected in medium containing G418 (results not shown), HAT (9.2 ×106 cells) or HAT + G418 (1.8 × 107 cells). Sixty HATr and 153 HATr/G418r colonies survived HAT and HAT + G418 selection, respectively. Given that twice as many cells were plated into HAT + G418, these numbers are not significantly different (significance test for comparing two proportions, P > 0.05). Genomic DNA from selected resistant lines was digested with HindIII and EcoRI (Fig. 1d) and screened by Southern analysis. From initial screening results 17 potentially targeted lines were screened more thoroughly by Southern analysis (Fig. 2). 5/16 HATr (31.3%) and 5/45 HATr/G418r (11.1%) lines were correctly targeted. All lines correctly targeted at the 5[prime] end were also found to be correctly targeted at the 3[prime] end, regardless of the selection regime used. The reduced targeting efficiency under HAT + G418 selection may reflect poor expression of the neo cassette within the LIF locus, leading to loss of correctly targeted lines under G418 selection.
Figure 2 Southern blot analysis of tag step targeted cell lines. Genomic DNA from HATr (lanes 1-5) and HATr/G418r tag step targeted lines (lanes 6-17) digested with EcoRI/HindIII was hybridised to probe 1 (upper) and 4 (lower) (Fig. 1d). The size of the detected bands in kilobases is indicated. Filters were exposed to storage Phosphor screens (Molecular Dynamics), processed using a PhosphorImager (Molecular Dynamics) running ImageQuantTM software and images were manipulated using AdobePhotoshopTM software. The relationship between the signal and image intensity is sigmoidal and all the lanes are from a single gel.
Figure 3 Schematic diagram illustrating exchange step gene targeting at the LIF locus. (a) The exchange step targeting vector p1DLacLIF showing the NheI sites used for linearisation and the EcoRV and HindIII sites used for screening targeted lines. Vector p1DLacLIF contains a [beta]-galactosidase reporter gene fused in frame to the first exon of the LIF D transcript. The reporter gene was modified at the N-terminus to remove the ATG initiation codon. The reporter gene contains the N-terminal SV40 nuclear localisation sequence and the intron, 3[prime] untranslated region and polyadenylation sequence from the mouse protamine 1 gene (41) 3[prime] of the coding region. Translation of [beta]-galactosidase is directed by the endogenous ATG of the LIF exon 1D. (b) Exchange step gene targeting at the tagged LIF locus using vector p1DLacLIF. The tagged LIF locus and exchange step targeting vector are illustrated. (c) Restriction maps of the wild-type, tagged and exchange step targeted LIF alleles. The restriction fragments from digestion of the wild-type (i), tag step targeted (ii) and exchange step targeted (iii) LIF alleles with EcoRV and HindIII are shown. Probes used for Southern analysis are indicated. Probe 1, the 5[prime] external LIF probe (Fig. 1d), detects bands of 13.1 kb from the untargeted allele, 11.8 kb from the tag step targeted allele and 7.5 kb from the exchange step targeted allele. Probe 5 is a [beta]-galactosidase probe used to screen for the presence of a single [beta]-galactosidase gene and detects bands of 9.3 kb in exchange step targeted lines. Probe 3 (Fig. 1d) is a neo probe used to verify the presence of the neo cassette in exchange step targeted lines (results not shown). Details as for Figure 1d.
Of particular interest was the observation that 5/5 targeted clones selected in HAT medium alone contained both the HPRT and neo selection cassettes. Homologous recombination 5[prime] of the HPRT cassette is obligatory for integration of the HPRT gene and survival of homologous recombinants in HAT medium. Given that there was no selective pressure for integration of the neo cassette, the 3[prime] recombination event could potentially occur in homologous sequences between the HPRT and neo cassettes (HPRT+neo- lines), or 3[prime] of the neo cassette (HPRT+neo+ lines). If recombination frequency is determined by length of sequence homology alone, then the expected frequencies of HPRT+neo- and HPRT+neo+ homologous recombinants surviving HAT selection would be proportional to the length of sequence homology on the targeting vector between the HPRT and neo cassettes [HPRT+neo-; 4.9/6.7 kb (73%) total homology], and 3[prime] of the neo cassette [HPRT+neo+; 1.8/6.7 kb (27%) total homology], respectively. However, co-integration of the HPRT and neo selection cassettes during homologous recombination was observed in 5/5 HATr ES cell lines. In similar fashion, 2/2 targeted lines selected in G418 alone contained both the neo and HPRT cassettes correctly integrated by homologous recombination (results not shown). These results implicate the ends of the targeting vector in the process of homologous recombination, and have implications for the mechanism of homologous recombination and gene targeting in ES cells as discussed below.
The overall efficiency of 16.4% for tag step gene targeting was similar to frequencies reported previously for the introduction of a single region of non-homology into the LIF locus (15,27). Our results therefore demonstrate that multiple distinct regions of extensive (>1 kb) non-homology can be integrated into a genomic locus by homologous recombination using a single replacement vector.
Germline transmission of tag step targeted ES cells
First round targeted ES cell lines H9, H20, HG26 and HG60 (Fig. 2, lanes 2, 4, 7 and 13) each had a >65% normal karyotype 2N = 40. These lines were microinjected into blastocysts. Chimearae were produced from all targeted lines with the exception of H9. Four out of 12 chimaeric males produced with line HG60 transmitted the ES cell genotype through the germline as assessed by Gpi-1s analysis of sperm (results not shown).
Exchange step gene targeting with p1DLacLIF
The exchange step gene targeting vector p1DLacLIF (Fig. 3a) differed from pHLNX in two important respects. Firstly, the HPRT tag gene was replaced by a modified LIF gene. Secondly, homology between p1DLacLIF and the tagged endogenous locus was restricted to sequences upstream of the neo stability cassette which was not carried on p1DLacLIF. As illustrated in Figure 3b, homologous recombination of the exchange step vector leads to exchange of the HPRT cassette for the altered LIF gene, while the neo cassette remains on the endogenous tagged locus. Homologous recombination at the LIF locus could thus be directly selected in medium containing 6-TG, while G418 could be used to select for maintenance of the tagged LIF allele.
The modified LIF gene in p1DLacLIF (Fig. 3a) contained a [beta]-galactosidase reporter gene fused in frame to the mouse 1D exon, surrounded by 2.7 kb 5[prime] homology and 4.0 kb 3[prime] homology with the tagged LIF allele. The germline competent, tag step gene targeted line HG60 (Fig. 2, lane 13), was used in exchange step gene targeting. The relative efficiencies of tag-and-exchange and stable tag-exchange targeting were compared directly by subjecting duplicate pools of transfected cells to two different selection regimes. Selection of exchange step targeted lines in medium containing 6-TG alone is equivalent to using the tag-and-exchange system since this selects only for loss of HPRT expression. Selection in medium containing 6-TG + G418, as described for stable tag-exchange targeting, allows selection for both loss of HPRT expression and maintenance of the tagged LIF allele.
Transfected cells (2.5 × 106) were selected in the presence of 6-TG, or 6-TG + G418. The results of this experiment are shown in Table 1. After 14 days, 103 colonies resistant to 6-TG were observed compared to 23 colonies resistant to 6-TG + G418. This suggested that the stable tag-exchange approach overcame a background of >75% of 6-TGr lines that had survived tag-and- exchange selection due to loss of the tag cassette in the absence of homologous recombination.
This was verified by Southern blot analysis (Fig. 4). Genomic DNA from 6-TGr (Fig. 4, lanes 1-7) and 6-TGr/G418r (Fig. 4, lanes 8-18) lines was digested with HindIII and EcoRV (Fig. 3c). Twenty-eight 6-TGr and 16 6-TGr/G418r lines were screened by Southern blot (Fig. 4). Six 6-TGr and 14 6-TGr/G418r lines were correctly targeted giving respective targeting efficiencies of 21.4% and 87.5% (Table 1).
Figure 4 Southern blot analysis of exchange step targeted cell lines. Genomic DNA from 6-TGr (lanes 1-7) and 6-TGr/G418r (lanes 8-18) exchange step targeted lines digested with EcoRV/HindIII was hybridised to probe 1 (upper) and 5 (lower) (Fig. 3c). The size of the bands in kilobases is indicated. Filters were processed as described in Figure 2. The relationship between the signal and image intensity is sigmoidal and all the lanes are from one gel.
Of the non-targeted lines resulting from tag-and-exchange gene targeting (6-TGr lines), 17/22 (77.3%) had physically lost both the tag and stability cassettes (for example Fig. 4 upper, lanes 1, 2 and 4). Lines which had physically lost one or other of these cassettes were not observed. By contrast, neither of the non-targeted lines resulting from stable tag-exchange gene targeting (6-TGr/ G418r lines) had physically lost the selection cassettes (Fig. 4 upper, lanes 11 and 12). The 5/22 6-TGr and 2/2 6TGr/G418r untargeted lines which had not physically lost the tag cassette presumably arose from loss of tag gene expression by means such as methylation or mutation (11). The improved efficiency of stable tag-exchange targeting therefore results from the use of a selection system which eliminates a significant background of [sim]75% of non-targeted cell lines which result from physical loss of the tag gene by means other than homologous recombination.
Mechanism of physical tag gene loss
Physical loss of thetag gene in the absence of homologous recombination was invariably accompanied by physical loss of the stability cassette. Karyotypic analysis of five ES cell lines which had undergone loss of the tag gene in the absence of homologous recombination was carried out by Q-banding. Each line had a normal diploid chromosome count, 2N = 40, and both chromosomes 11 (on which the LIF gene is located) appeared to be grossly normal. Physical loss of the selection cassettes was therefore not a consequence of gross alteration to chromosome 11.
Loss of the tagged LIF allele by chromosomal microdeletion would result in the presence of a single wild-type LIF allele in non-targeted 6-TGr cells which had physically lost the tag cassette. A second potential mechanism for tag and stability cassette loss, reversion of the tagged allele to wild-type, would result in the presence of two wild-type LIF alleles within the cell. The copy number of LIF alleles within eight non-targeted 6-TGr ES cell lines was estimated by quantitative comparison with the copy number of the autosomal Hesx1 locus which is located on mouse chromosome 14 (24). Comparison with control ES cell lines carrying one (HG60) or two (E14TG2a) wild-type copies of the LIF gene indicated that two wild-type copies of the LIF gene were present (Fig. 5). This indicated that the tag and stability cassettes had been removed precisely from the tagged LIF locus, presumably by non-reciprocal transfer of genetic information from the wild-type locus.
Figure 5 Copy number analysis of the LIF gene in lines which had physically lost the tag and stability cassettes by means other than homologous recombination. Southern blot analysis (upper) was carried out with BamHI digested DNA from the E14TG2a ES cells (two wild-type LIF alleles; C1), the tag step targeted line HG60 (one wild-type, one tagged LIF allele; C2) and eight lines that had undergone physical loss of the tag and stability cassettes by means other than homologous recombination (1-8). Hybridisation was carried out with a 1.1 kb EcoRI/PstI Hesx1 probe (24) and a 0.9 kb SacII/PstI probe from LIF intron 1 (42). The Hesx1 probe detects a band of 12.0 kb and the LIF probe detects bands of 9.0 and 2.9 kb from the tag step targeted and untargeted LIF genes, respectively. The filter was processed as described in Figure 2 and quantitation was carried out using ImageQuantTM software. The relationship between the signal and image intensity is sigmoidal and all the lanes are from one gel. The ratio of Hesx1/LIF is represented graphically for each cell line (lower).
Germline transmission of exchange step targeted ES cells
The developmental potential of two independent stable tag- exchange step targeted ES cell lines was assessed by blastocyst microinjection. Male chimaerae transmitting the ES genotype through the germline were produced with both lines as determined by coat colour assessment. Southern blots of EcoRV/HindIII-digested tail DNA from the offspring of mice carrying one of the ES cell genotypes probed with a 5[prime] external LIF probe (Fig. 3, probe 1) verified germline transmission of the altered LIF allele (results not shown).
DISCUSSION
Tag gene targeting: implications for the mechanism of homologous recombination and a rationale for the integration of multiple, distinct regions of extensive non-homology on a replacement vector by gene targeting
An HPRT cassette (9) was used as the positive/negative tag cassette in this work. ES cells expressing the HPRT gene can pass through the germline making it possible to assess germline competence or generate null mutant mice after tag step gene targeting. Selection during tag step gene targeting was carried out in medium containing either G418, HAT or HAT + G418. All homologous recombinants carried both the HPRT and neo cassettes integrated correctly at the LIF locus, even when selection for both cassettes was not applied. Functional crossover events between incoming DNA and the endogenous LIF locus were therefore observed in the 5[prime] and 3[prime] but not the central region of homology between pHNLX and the endogenous LIF allele.
This observation indicates a role for vector ends in the initiation of homologous recombination, in contrast to a previous model for homologous recombination of replacement targeting vectors (28), which suggests that crossover events between vector and genomic DNA are initiated by single- or double-strand breaks positioned randomly in the vector or genomic sequence. We suggest a refined version of this model to explain our observations. Recombination machinery attaches to the vector ends and `scans' along the vector DNA leading to strand invasion at the homologous locus. Holliday junction formation between incoming and genomic sequences relies on the existence of homologous sequences and occurs as described by Hasty etal. (28). Thus, although recombination is initiated at the vector ends, non-homologous ends themselves do not participate in Holliday junction formation. Holliday junctions migrate from the ends of the vector until a crossover between vector and genomic sequence is formed. Holliday junction migration is impeded or blocked by extended regions of non-homology, such as those formed by the HPRT and neo cassettes in pHLNX. Thus, the most frequently occurring crossover events between vector and genomic sequences occur at the extremities of the vector leading to integration of the targeting vector at the homologous locus.
This model is consistent with known properties of recombination in bacteria (29,30) where increasing sizes of heterologous DNA have been shown to reduce the efficiency of strand exchange in heteroduplexes. It has also been suggested that large regions of non-homology hinder the normal process of recombination progression in yeast (34). Previous reports of crossovers in mammalian cells between regions of non-homology contained on replacement targeting vectors (31-33) can be explained by the small size of one of the non-homologous regions and suggest that stretches of non-homology of up to 60 bp do not significantly impede Holliday junction migration. The involvement of DNA ends in the process of recombination in mammalian cell lines is well established: DNA ends are involved in extrachromosomal homologous recombination (35), homologous recombination of insertion vectors (28), and non-homologous recombination in mammalian cells (36,37).
We have demonstrated for the first time that multiple, distinct regions of substantial non-homology can be introduced into the mouse genome by homologous recombination with similar efficiency to the introduction of single regions of non-homology. Our interpretation of this observation, that it results from inhibition of Holliday junction migration by large regions of non- homology, points a way to complex modification of endogenous loci by gene targeting. Provided the terminal regions of non- homology exceed a certain undefined length, it should be possible to achieve single step integration of multiple non-homologous regions into a single genomic locus. Applications such as the introduction of gain-of-function mutations (5) or complex biosynthetic pathways into the mammal, or the creation of mammalian models for complex disease states (5), can be envisaged.
Exchange step gene targeting: efficient introduction of precise alterations into the mammalian genome
The design of our experiment allowed direct comparison of the efficiency of tag-and-exchange versus stable tag-exchange gene targeting in the same population of transfected cells by selection in 6-TG, or 6-TG + G418, respectively. Comparison of the numbers of independent clones surviving selection (23 6-TGr/G418r versus 103 6-TGr clones) indicated that the great majority [80/103 (77.6%)] of non-targeted lines surviving tag-and-exchange gene targeting were eliminated by stable tag-exchange selection. Consistent with this, Southern analysis indicated relative targeting frequencies of 21% (tag-and-exchange) and 88% (stable tag- exchange), respectively. Stable tag-exchange gene targeting therefore eliminates a large proportion of the non-targeted colonies that survive selection, the major limitation of tag-and-exchange gene targeting for the repeated introduction of precise modifications into the ES cell genome.
The origin of the 6-TGr/G418s background lines eliminated by stable tag-exchange selection was investigated by Southern blot analysis. 17/22 (77%) of these non-targeted lines survived 6-TG (tag-and-exchange) selection due to physical loss of the tag cassette. In comparison, no lines surviving stable tag-exchange selection had arisen in this manner. Physical loss of the tag (and stability) cassettes was therefore the main contributing mechanism to the non-targeted background surviving tag-and-exchange selection. These lines were shown by cytogenetic and molecular techniques to result from reversion of the tagged allele to wild-type as also shown by Wu etal. (7) for the Colla-1 locus. This could be explained by non-reciprocal genetic exchange with the untagged LIF allele by mechanisms such as gene conversion, chromosome loss and duplication and mitotic recombination (38). The elimination of these lines means that stable tag-exchange gene targeting was [sim]4-fold more efficient than tag-and-exchange gene targeting for the introduction of precise genomic alterations into the LIF locus. However, the increase in targeting efficiency seen with stable tag-exchange targeting is expected to vary at different genomic loci and is possibly underestimated at loci such as LIF where homologous recombination is relatively efficient. This modification will be particularly beneficial at loci where physical loss of the tag gene in the absence of homologous recombination occurs frequently. In the case of Colla-1 (7) for example, where reversion of tagged alleles to wild-type is more common, an increased homologous recombination frequency of closer to 10-fold might be expected.
Germline transmission of the ES cell genotype produced in stable tag-exchange gene targeting indicated that the E14TG2a cell line retained germline competence throughout two rounds of gene targeting, and after HAT, G418 and 6-TG selection. Germline transmission of ES cells which have undergone two rounds of gene targeting has also been reported for J1 (7), HM-1 (6) and CCE (10) lines, suggesting that two-step targeting systems with multiple selection protocols will be applicable to a wide range of ES cell lines.
Comparison with other targeting strategies for the repeated introduction of precise genomic alterations
An alternate two-step targeting system, termed plug and socket (11) can also be used for the repeated introduction of precise alterations to any site of the mouse genome. In this approach the use of positive selection during both rounds of gene targeting overcomes problems associated with selection for loss of gene expression. However, in the published example of this approach, second round targeting resulted in a high background (5/7) of incorrectly targeted lines which arose from integration of recircularised rather than linear targeting vector. Furthermore, the plug and socket selection cassette must be positioned close to the site of the precise genomic alteration, potentially leading to locus disruption. Although the stability cassette of the stable tag-exchange approach remains associated with the exchange step targeted allele, this cassette would ideally be positioned well away from the gene under investigation. The location of the stability cassette within the 3[prime] untranslated region of the LIF gene as reported here is therefore suboptimal for most targeting uses. In cases where the presence of the stability cassette on the exchange step targeted allele is likely to be a particular problem, such as when genes are located within functional clusters (39), an alternative promoter to that of the pgk-1 gene could be used, or the stability cassette could be flanked by loxP sites and removed by cre-mediated recombination (40).
The improved efficiency of the stable tag-exchange system characterised in this work provides an effective means for the repeated introduction of precise alterations to any site of the mouse genome. Using this system it should be possible to dissect the control and function of any mammalian gene in the context of the whole organism leading to a precise, molecular understanding of the gene product.
ACKNOWLEDGEMENTS
We are extremely grateful to Austin Smith for helpful advice on vector design, targeting conditions and critical reading of this manuscript, to Anita Peura and Steven McIlfatrick (Bresagen) who carried out the blastocyst microinjection, to Simon Timke for help with statistical analysis and to Graham Webb for chromosome Q banding and analysis. This research was supported by funding from the Australian Research Council and the Anti-Cancer Foundation of South Australia.
REFERENCES
1. Thomas, K. R. and Capecchi, M. R. (1987) Cell, 51, 503-512.
2. Evans, M. J. and Kaufman, M. H. (1981) Nature, 292, 154-156.
3. Martin, G. R. (1981) Proc. Natl. Acad. Sci. USA, 78, 7634-7638.
4. Bradley, A., Hasty, P., Davis, A. and Ramirez-Solis, R. (1992) Biotechnology, 10, 534-539.
5. Bronson, S. K. and Smithies, O. (1994) J. Biol. Chem., 269, 27155-27158.
6. Stacey, A., Schnieke, A., McWhir, J., Cooper, J., Colman, A. and Melton, D. W. (1994) Mol. Cell. Biol., 14, 1009-1016.
7. Wu, H., Liu, X. and Jaenisch, R. (1994) Proc. Natl. Acad. Sci. USA, 91, 2819-2823.
8. Askew, G. R., Doetschman, T. and Lingrel, J. B. (1993) Mol. Cell. Biol., 13, 4115-4124.
9. Reid, L. H., Gregg, R. G., Smithies, O. and Koller, B. H. (1990) Proc. Natl. Acad. Sci. USA, 87, 4299-4303.
10. Gondo, Y., Nakamura, K., Nakao, K., Sasoka, T., Ito, K., Kimura, M. and Katsuki, M. (1994) Biochem. Biophys. Res. Com., 202, 830-837.
11. Detloff, P. J., Lewis, J., John, S. W. M., Shehee, W. R., Langenbach, R., Maeda, N. and Smithies, O. (1994) Mol. Cell. Biol., 14, 6936-6943.
12. Hilton, D. J. (1992) Trends Biochem. Sci., 17, 72-76.
14. Rathjen, P. D., Toth, S., Willis, A., Heath, J. K. and Smith, A. G. (1990) Cell, 62, 1105-1114.
15. Mountford, P., Zevnik, B., Düwel, A., Nichols, J., Li, M., Dani, C., Robertson, M., Chambers, I. and Smith, A. (1994) Proc. Natl. Acad. Sci. USA, 91, 4303-4307.
16. Thonglairoam, V. (1994) PhD thesis, University of Adelaide, Adelaide, South Australia.
17. Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989) Molecular Cloning, 2nd Ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
18. Whyatt, L. M. (1996) PhD thesis, Adeleaide University, Adelaide, South Australia.
19. Hooper, M., Hardy, K., Handyside, A., Hunter, S. and Monk, M. (1987) Nature, 326, 292-295.MEDLINE Abstract
20. Smith, A. G. (1991) J. Tiss. Cult. Meth., 13, 89-94.
21. Hooper, M. L. and Slack, C. (1977) Dev. Biol., 55, 271-284.
22. Laird, P. W., Zijderveld, A., Linders, K., Rudnicki, M. A., Jaenisch, R. and Berns, A. (1991) NucleicAcidsRes., 19, 4293.
23. Hogan, B., Beddington, R., Costantini, F. and Lacy, E. (1994) Manipulating the Mouse Embryo: a Laboratory Manual, 2nd Ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
24. Thomas, P. Q., Johnson, B. V., Rathjen, J. and Rathjen, P. D. (1995) J. Biol. Chem., 270, 3869-3875.
25. Robertson, E. J. (ed.) (1987) Teratocarcinomas and Embryonic Stem Cells: a Practical Approach. IRL press, Oxford, UK.
26. Caspersson, T., Zech, L. and Johansson, C. (1970) Exp. CellRes., 60, 315-319.
27. Stewart, C. L., Kaspar, P., Brunet, L. J., Bhatt, H., Gadi, I., Kontgen, F. and Abbondanzo, S. J. (1992) Nature, 359, 76-79.
28. Hasty, P., Rivera-Perez, J. and Bradley, A. (1992) Mol. Cell. Biol., 12, 2464-2474.
29. DasGupta, C. and Radding, C. M. (1982) Proc. Natl. Acad. Sci., USA, 79, 762-766.
30. Bianchi, M. E. and Radding, C. M. (1983) Cell, 35, 511-520.
31. Le Mouellic, H., Lallemand, Y. and Brûlet, P. (1992) Cell, 69, 251-264.MEDLINE Abstract
32. Gu, H., Zou, Y. and Rajewsky, K. (1993) Cell, 73, 1155-1164.MEDLINE Abstract
33. Michalska, A. E. and Choo, K. H. A. (1993) Proc. Natl. Acad. Sci. USA, 90, 8088-8092.
34. Hayden, M. S. and Byers, B. (1992) Dev. Genet., 13, 498-514.
35. Kucherlapati, R. S. (1984) Proc. Natl. Acad. Sci. USA, 81, 3153-3157.
36. Folger, K. R., Wong, E. A., Wahl, G. and Capecchi, M. R. (1982) Mol. Cell. Biol., 2, 1372-1387.
37. Thomas, K. R., Folger, K. R. and Capecchi, M. R. (1986) Cell, 44, 419-428.
38. Cavanee, W. K., Dryja, T. P., Phillips, R. A., Benedict, W. F., Godbout, R., Gallie, B. L., Murphree, A. L., Strong, L. C. and White, R. L. (1983) Nature, 305, 779-784.
39. Olsen, E. N., Arnold, H.-H., Rigby, P. W. J. and Wold, B. J. (1996) Cell, 85, 1-4.
40. Sauer, B. and Henderson, N. (1988) Proc. Natl. Acad. Sci. USA, 85, 5166-5170.
41. Peschon, J. J., Behringer, R. R., Brinster, R. L. and Palmiter, R. D. (1987) Proc. Natl. Acad. Sci. USA, 84, 5316-5319.
42. Stahl, J., Gearing, D. P., Wilson, T. A., Brown, M. A., King, J. A. and Gough, N. M. (1990) J, Biol. Chem., 265, 8833-8841.
L. van der Weyden, D. J. Adams, and A. Bradley Tools for targeted manipulation of the mouse genome
Physiol Genomics,
December 3, 2002;
11(3):
133 - 164.
[Abstract][Full Text][PDF]
CiteULike 
Connotea 
Del.icio.us What's this?
