INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The introduction of predetermined alterations in chromosomal sequences by homologous recombination with transferred DNA (gene targeting) is a powerful technology for modifying gene structure and function. It has applications that include the study of gene expression in its normal chromosomal environment and the creation of animal models of human genetic diseases and, ultimately, it has the potential to be an effective form of gene therapy (3, 33, 35). In addition, the ability to manipulate the transforming DNA makes gene targeting a valuable model system in the study of homologous recombination mechanisms.

Gene targeting can be performed with either insertion ("ends-in" or O-type) vectors or replacement ("ends-out" or -type) vectors. In an insertion vector, a double-strand break is introduced within the homology region creating DNA ends that invade cognate chromosomal sequences. In Saccharomyces cerevisiae, gene targeting using an insertion vector is consistent with the double-strand-break repair (DSBR) model (24, 32). Like yeast, targeted vector insertion in mammalian cells also has features consistent with DSBR (16, 17, 23, 34). In a gene replacement vector, the region of homology is interrupted by a selectable genetic marker. Since the ends of the vector are discontinuous with the chromosome, recombination replaces a region of the chromosome with vector sequences. The mechanism of homologous recombination with a gene replacement vector is unknown. In principle, it might occur by assimilation of a single strand of the vector into the chromosome, as was proposed to explain the replacement of a chromosomal allele with linear duplex DNA in S. cerevisiae (15). However, single-strand assimilation might be impeded by the heterology encoded by the selectable marker. Nevertheless, Negritto et al. (21) found at least a 10-fold increase in marker incorporation in an msh2 mutant even when all flanking markers were identical. Thus, marker assimilation may occur and be corrected by a process that involves mismatch repair (MMR) genes (15, 21). Gene replacement by single strand assimilation predicts that markers in flanking heteroduplex DNA (hDNA) will reside in a cis configuration. Alternatively, gene replacement might involve two crossing-over events in the homologous DNA flanking the selectable marker. This could explain how chromosomal deletions are engineered in the yeast genome using replacement vectors in which the selectable marker is flanked by two very distant homology regions (10, 26, 31). In this instance, markers in flanking hDNA would reside in a trans configuration. Thus, the models make testable predictions about the configuration of hDNA in the recombinants.

An important contribution to the study of hDNA formation during homologous recombination came from studies in S. cerevisiae, where a small palindrome was shown to avoid MMR when encompassed within hDNA, likely as a consequence of it forming a small hairpin structure (20). Semiconservative DNA replication of unrepaired hDNA, followed by cell division, generates a mixed (sectored) recombinant. Thus, the positions of sectored sites mark the location of hDNA formed during recombination. Earlier, we showed that a small palindrome in hDNA is also poorly repairable by the mammalian MMR system (16), a feature that provided some important insights into the formation of hDNA during targeted vector insertion in mammalian cells (16, 17). To obtain information about the presence and position of hDNA formed during mammalian gene replacement, we constructed a gene replacement vector in which the flanking arms of homology were marked by insertions of the small palindrome. Among the several independent recombinants in which hDNA was present in both flanking arms of homology, the palindrome markers were predominantly in a trans configuration. This result is consistent with the mammalian gene replacement reaction involving two crossing-over events.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Recipient hybridoma cells. The haploid, chromosomal immunoglobulin µ- heavy chain locus in the wild-type murine hybridoma cell line, Sp6/HL, serves as the target for gene replacement (see Fig. 2). The origin of Sp6/HL and the methods used for hybridoma cell culture have been described previously (13, 14).

Gene replacement vector. The 13.1-kb omega ()-form, enhancer-trap gene replacement vector, pCµCpal (Fig. 1) was used in these studies. The backbone of pCµCpal consists of a 5.4-kb segment of pSV2neo (28) from which the 372-bp NsiI/NdeI fragment encompassing the simian virus 40 early-region enhancer responsible for neo gene expression was removed. To effect gene replacement, the flanking arms of homology in pCµCpal consisted of a 4.2-kb Bst1107/XbaI Cµ region fragment and a 3.5-kb SpeI/SacI C region fragment positioned to the left and right of pSV2neo, respectively. Both fragments were derived from cloned, wild-type Sp6/HL genomic DNA, and their alignment with the Sp6/HL chromosomal µ- region is indicated by the dashed lines in Fig. 1. The Cµ and C homology regions share a slight (63-bp) overlap between the SpeI and XbaI sites located 3' of Cµ.

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Gene replacement at the chromosomal immunoglobulin µ- locus. The structure of the haploid, chromosomal immunoglobulin heavy chain µ- region in the recipient wild-type mouse hybridoma cell line, Sp6/HL, is presented, along with this locus in a recombinant hybridoma cell line generated as a result of gene replacement with a single copy of the vector, pCµCpal. In pCµCpal, endogenous restriction enzyme sites in the Cµ and C regions (denoted in normal typeface) have been replaced with the indicated genetic markers (denoted in bold typeface). The vector-borne NotI site is contained within a perfect, 30-bp palindrome insertion. In the Cµ region, marker positions are designated relative to the Bst1107 site that defines the 5' end of this segment, while the single marker position in the C region is numbered relative to the SacI site that marks the 3' end of this segment. For details relating to vector construction, refer to Materials and Methods. Gene replacement has the potential to generate different marker combinations in the various recombinant hybridoma cell lines. Therefore, the five marker positions in the recombinant Cµ and C regions are designated by a question mark. The primer pair AB9703-AB9745 binds outside the Cµ region of homology in the replacement vector and generates a specific 4.8-kb PCR product from the recombinant Cµ region as shown. The sequence of the primers and where they bind have been presented previously (17, 23). The primer pair AB20191-AB20192 binds within the C region of homology and generates a 0.8-kb product. The sequence of the C primers and their binding sites are presented in Materials and Methods. Probe fragments: Cµ-specific probe fragment F is an 870-bp XbaI/BamHI fragment; probe X-R is a 913-bp XhoI/EcoRI fragment from the C region. Abbreviations: Cµ, µ gene constant region; C,  gene constant region; VHTNP, TNP-specific heavy-chain variable region; neo, neomycin phosphotransferase gene. The thickened line represents the vector, pSV2neo (28). The figure is not drawn to scale.

Vector transfer and isolation of independent G418^R transformants. The pCµCpal vector (8.7 pmol) was transferred to 2 × 10⁷ recipient Sp6/HL hybridoma cells by electroporation as described elsewhere (1). Since the enhancer-trap vector significantly reduces the frequency of random transformants but not targeted recombinants (2, 22), independent G418^R transformants could be isolated according to the plating procedure described previously (16, 23).

Recombinant identification and genetic marker analysis. Genomic DNA was prepared from individual G418^R transformants by the method of Gross-Bellard et al. (8). To identify recombinants, individual DNA samples were screened by a PCR assay utilizing the primer pair AB9703-AB9745 that amplifies a specific 4.8-kb Cµ product from recombinant cells as illustrated in Fig. 1. The primer sequences and their binding sites have been reported elsewhere (17, 23). Hybridoma cell lines identified as being recombinants were further characterized by Southern analysis to verify the gene replacement event as detailed in Results. For Southern analysis, restriction enzymes were purchased from New England Biolabs, Inc. (Beverley, Mass.), Amersham Pharmacia Biotech, Inc. (Baie d'Urfé, Québec, Canada), and Canadian Life Technologies, Inc. (Burlington, Ontario, Canada) and used in accordance with the manufacturer's specifications. Gel electrophoresis, transfer of DNA onto nitrocellulose membrane, ³²P-labeled probe preparation, and hybridization were all performed according to standard procedures (27).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Recombinant identification. The purpose of this study was to determine whether genetic markers form hDNA in a cis or a trans configuration during mammalian gene replacement. The gene targeting system is based on the wild-type hybridoma cell line, Sp6/HL, which bears a single copy of the chromosomal immunoglobulin µ- region that serves as the target for homologous recombination with the omega ()-form of the enhancer-trap replacement vector, pCµCpal (Fig. 1). As reported previously (22, 23), enhancer-trap vectors permit efficient isolation of targeted recombinants at the chromosomal µ- locus. The 4.2-kb Cµ and 3.5-kb C flanking arms of homology in pCµCpal were distinguishable from the corresponding regions of the chromosome at several positions as a consequence of inclusion of a 30-bp palindrome containing a unique NotI restriction enzyme site. The palindrome genetic marker is poorly repaired by the mammalian MMR system (16). Thus, following DNA replication and cell division in a recombinant bearing unrepaired hDNA, a sectored (mixed) recombinant is generated. A plating (sectoring) assay was used to recover independent recombinants. As described previously (16, 23), the procedure ensures that each recombinant represents the progeny of a single G418^R cell and that the G418^R product(s) of recombination is retained for molecular analysis.

5

6

Genetic marker determination. For determination of the Cµ region genetic markers, the 4.8-kb PCR product was tested for its resistance or sensitivity to cleavage with restriction enzymes specific for the vector-borne markers, namely, NotI and ScaI, or the corresponding enzymes specific for chromosomal markers, namely, SacI, AflII, ApaI, and NheI (Fig. 1). As shown in Fig. 2A, the various restriction enzymes generate diagnostic Cµ fragment sizes that can be conveniently analyzed by agarose gel electrophoresis. To determine which genetic marker resided in the C region of each recombinant, Southern analysis was utilized. As illustrated in Fig. 2B, digestion of genomic DNA with the HpaI-NotI combination and, in a separate digest, the combination of HpaI and PaeR71 followed by hybridization with probe X-R distinguishes whether the C marker is the vector-borne NotI palindrome or the corresponding endogenous PaeR71 site.

View larger version (18K): [in this window] [in a new window]   FIG. 2.   Restriction enzyme maps of the Cµ and C region in the recombinants. (A) Diagram of the 4.8-kb PCR product generated from the recombinant Cµ region using primer pair AB9703-AB9745 and the fragment sizes expected if the indicated positions bear either the vector-borne or chromosomal restriction enzyme site markers. (B) Illustration of the Southern blot analysis used to resolve whether the vector-borne palindrome NotI site or chromosomal PaeR71 site resides in the C region. Vector-borne markers are denoted in bold typeface. The diagrams are not drawn to scale.

View larger version (94K): [in this window] [in a new window]   FIG. 3.   Analysis of Cµ and C region marker patterns in recombinants 4/1, 7/1, and 8/1. The results in panels A and B present the analysis of the Cµ and C region genetic markers, respectively, in recombinants 4/1, 7/1, and 8/1. The sizes of fragments of interest are presented on the left of each figure, while the sizes of relevant DNA marker bands (denoted M) are presented on the right.

View larger version (28K): [in this window] [in a new window]   FIG. 4.   Marker segregation patterns in independent recombinants. The results of the Cµ and C region marker analysis for the recombinants are presented. The symbols used to denote the various genetic markers are explained in the text. The results of subcloning analysis of recombinants displaying >1 sectored site, along with the observed frequencies of the individual cell types, are presented in brackets. In the diagrams, the pSV2neo sequences separating the Cµ and C regions are denoted as neo. Positions bearing the vector-borne NotI-palindrome are indicated by a filled circle (); those with a chromosomal marker are indicated by an open circle (), while those sites that are sensitive to clevage with restriction enzymes diagnostic of both the chromosome and vector-borne markers (i.e., heterogeneous sites) are indicated by half-filled circles (). The cross-hatched circle with recombinant 29/2 indicates that neither the vector-borne NotI-palindrome nor the chromosomal PaeR71 marker was present in the C region.

Clonal analysis of sectored recombinants. As described above, the recombinant isolation procedure ensures that the individual product(s) of each gene replacement event is confined to a single culture well. Consequently, the marker configurations in the cell populations comprising each sectored recombinant reflect those present in each strand of the hDNA intermediate. As is evident from Fig. 4, recombinants 4/1, 8/1, 27/1, 29/1, 11/2, 22/2, 14/2, 16/2, 34/2, 36/2, 42/2, and 59/2 were sectored for more than one position in the Cµ and/or C region. Therefore, for these recombinants, it was necessary to establish marker linkage. This was accomplished by cloning each recombinant at 0.1 cell/well in 96-well tissue culture plates and repeating the Cµ and C marker determinations on several independent subclones (data not shown). As for the parental recombinants, PCR and gel analysis methods were used to determine the Cµ region markers in the subclones. For the C region, Southern blotting was utilized as described above. In addition, a more convenient PCR-gel analysis assay was developed. As illustrated in Fig. 1, the primer pair AB20192-AB20191 lies within the vector-borne (and chromosomal) C region of homology. These primers amplify a specific 848-bp PCR product that spans the potential mismatch created by the endogenous PaeR71 and vector-borne NotI-palindrome markers. If either PaeR71 or NotI sites are present in the recombinants, the PCR product will be cleaved to yield diagnostic fragments of 644 and 203 bp. Using these procedures, the C region marker patterns in the subclones of the various parental recombinants was determined. Equivalent results were obtained with either Southern or PCR assays. The parental recombinants, the subclone marker patterns, and the frequency of the various subclone types are presented in brackets in Fig. 4.

Evidence for extensive hDNA formation and of MMR during gene replacement. The positions of sectored sites revealed that hDNA formation during mammalian gene replacement was extensive (Fig. 4). In recombinants 4/1, 8/1, and 14/2, all positions up to and including the ApaI/NotI-palindrome mismatch located 1,831 bp from the beginning of the Cµ region were sectored, suggesting that hDNA had formed over this distance. In recombinants 5/1, 7/1, 20/1, and several others, the PaeR71/NotI-palindrome mismatch located 2,588 bp from the C terminus was sectored, suggesting that a long hDNA tract had spanned this region. Also, as indicated in the previous section, sectoring was observed for internal markers in both the Cµ and C region of several recombinants. These results indicate that in mammalian cells long regions of hDNA are formed in both flanking arms of homology in the replacement vector.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Eight recombinants (8/1, 27/1, 29/1, 11/2, 22/2, 16/2, 34/2, 36/2, and 42/2) were sectored in both flanking arms of homology. Genetic marker analysis in subclones derived from each recombinant revealed that in seven (27/1, 29/1, 11/2, 22/2, 16/2, 34/2, 36/2, and 42/2), vector-borne and chromosomal markers in the two homology regions were linked in a trans configuration. The trans configuration of markers in hDNA is inconsistent with assimilation of a single strand of the vector into the chromosome but, rather, strongly supports a mechanism of gene replacement in mammalian cells that involves two crossing-over events in homologous flanking DNA. A proposed model is presented in Fig. 5. Essentially, this model is the alternative to the single-strand assimilation model as presented in Fig. 3C of a study by Leung et al. (15). Recombination initiates when the two ends of the vector invade the chromosome and pair with their complementary sequences. For simplicity, invasion by single-stranded (perhaps, 3') ends of opposite DNA strands is shown (30). An alternative, unconventional model might involve invasion of both ends of the same vector strand. Our data do not permit this distinction to be made. In either case, Holliday junction branch migration results in the formation of extensive hDNA in both flanking regions of homology. The entire vector is incorporated into the chromosome by crossover resolution of the two Holliday junctions as a consequence of cutting the DNA strands at positions numbered 1, 4, and 5 and positions number 2, 3, and 6 as indicated. Following strand ligation, genetic markers form hDNA in the trans configuration. The cis configuration of hDNA in the single recombinant (8/1) can also be explained if cuts in the DNA strands are made at the alternate positions of 1 and 6' and of 2, 3, 4', and 5' as shown in the figure. Thus, this simple model readily accounts for the generation of the recombinants. Notable features include pairing between complementary strands of the vector and the chromosomal target, the formation of long regions of hDNA, and the requirement for resolution of two Holliday junctions for vector incorporation into the chromosome. Thus, in these respects the model bears resemblance to the DSBR mechanism of recombination (24, 32), a model that appears consistent with targeted vector insertion in mammalian cells (16, 17, 23, 34).

View larger version (16K): [in this window] [in a new window]   FIG. 5.   Proposed model for mammalian gene replacement. Recombination is depicted as initiating when different DNA strands from the two homologous arms of the vector invade the chromosome. Gene replacement results from crossing-over in each flanking arm and is accompanied by extensive formation of hDNA. For further mechanistic details, refer to the text.

The finding that mammalian gene replacement occurs by two crossing-over events is of importance in view of current models of gene replacement in yeast in which strand assimilation leading to gene conversion is proposed to occur predominantly (15). In fact, based on the yeast studies, it was suggested that the preferential repair of hDNA in favor of the chromosome might frequently eliminate vector-borne heterologies, including the selectable genetic marker, and thus constitute an efficient barrier to recombination with the -form gene replacement vectors in mammalian cells (15). Thus, the yeast data predict that gene replacement with a linear fragment of DNA contiguous with the chromosome should be much more efficient than that with a replacement vector. Thus, it was surprising that, in our previous study (18), we observed that gene targeting with a linear genomic fragment was only slightly more efficient (~2-fold) than with an -form replacement vector bearing a similar amount of overall homology. The above data suggest that the mammalian gene replacement reaction is not frequently overcome by the cellular MMR machinery. In support of this, a biased elimination of vector-borne markers as a consequence of MMR was not clearly evident in this or our previous gene targeting studies (18, 23). Also, there was no evidence that inclusion of genetic markers in the vector-borne homology regions reduced the efficiency of gene targeting compared to vectors bearing a completely wild-type, isogenic sequence (18, 23). The differences between yeast and mammalian cells can be reconciled if, as suggested by this study, mammalian cells mediate the gene replacement reaction primarily by two crossing-over events. Thus, in contrast to yeast, a factor(s) other than biased action of the cell's MMR machinery may be responsible for the low efficiency of gene targeting relative to random integration in the mammalian genome.

The gene replacement mechanism in Fig. 5 depends on the two ends of the replacement vector locating the correct target sequence and undergoing strand invasion. Since the vector ends point away from each other, there would appear to be a requirement for them to behave independently for this to be accomplished. Therefore, strand invasion may be an efficient process in mammalian gene targeting. In contrast, strand invasion may be inherently inefficient in yeast, as suggested previously (15). Thus, it would be unlikely that both ends of a linear fragment would independently engage the target sequence and undergo strand invasion before strand assimilation from one end spans the entire fragment.

In the gene replacement vector used in this study, the vector ends were completely homologous to the chromosomal target, a feature which presumably made the strand invasion step relatively straightforward. This is not always the case as replacement vectors designed for use in positive-negative selection (PNS) schemes (19) terminate in heterologous sequences that encode a counterselectable marker(s) (for example, the herpes simplex virus thymidine kinase gene). Therefore, in order for these vector designs to undergo correct gene replacement, the terminal nonhomology must be removed. In S. cerevisiae, removal of nonhomologous DNA ends during recombination depends on the activity of the nucleotide excision repair endonuclease Rad1 or Rad10 and also on the mismatch repair proteins Msh2 and Msh3 (25, 29). In mammalian cells, equivalent proteins acting in the same manner may fulfill this role. Terminal nonhomology may be removed prior to any homologous pairing, thus creating homologous ends that can undergo strand invasion. However, it is also possible that pairing between homologous vector-borne and chromosomal sequences occurs first between more internal sequences in each flanking arm with subsequent removal of the terminal "flaps" and resynthesis, as suggested by the ectopic recombination data of Inbar and Kupiec (12).

The positions of sectored sites in several recombinants provided strong evidence for extensive formation of hDNA in both flanking arms of homology during mammalian gene replacement. This observation is novel and pertinent to the model presented in Fig. 5. Long hDNA tracts as revealed by the positions of sectored sites generated by failure to repair mismatches involving the palindrome have also been reported for targeted vector insertion in mammalian cells (16). The formation of long hDNA tracts during mammalian gene targeting provides a conceptual basis for the gene conversion tracts reported previously by us (18, 23) and others (5, 6, 11) which were suggested to have resulted from MMR of hDNA. Similarly, in the present study, the results suggested that some mismatches in hDNA were also subject to repair undergoing either restoration or gene conversion. However, in some cases, there are difficulties in distinguishing MMR of hDNA from other possible explanations. For example, although MMR of hDNA provides an explanation for the presence of vector-borne markers in proximity to the pSV2neo sequences such as in recombinants 3/1, 4/1, 8/1, 16/1, and others, they are also consistent with the possibility of a crossover event in the Cµ region just prior to the markers. Another example is the presence of chromosomal markers in the Cµ region of several recombinants (3/1, 5/1, 7/1, 20/1, and others), a pattern that, while consistent with gene conversion resulting from MMR of hDNA, might also result from deletion of vector sequences from the ends of the transferred DNA followed by their replacement with chromosomal sequences. While several studies in mammalian cells have suggested that, in general, degradation from the ends of transfected DNA is probably not extensive (6, 10, 12, 16, 23, 26, 27, 30), studies in yeast show that there can be extensive 5'-3' exonucleolytic resection of 5'-ending strands to expose long single-stranded regions (9, 30). If such extensive resection is a component of the mammalian gene replacement reaction, it would support the two-stranded invasion model of gene replacement depicted in Fig. 5 and may contribute substantially to the conversion and/or restoration events observed in the recombinants.