Mechanisms Involved in Targeted Gene Replacement in Mammalian Cells

Corresponding author: Mark D. Baker, Department of Pathobiology, Ontario Veterinary College, University of Guelph, Guelph, Ontario N1G 2W1, Canada., mdbaker{at}uoguelph.ca (E-mail)

Communicating editor: L. S. SYMINGTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The "ends-out" or omega ()-form gene replacement vector is used routinely to perform targeted genome modification in a variety of species and has the potential to be an effective vehicle for gene therapy. However, in mammalian cells, the frequency of this reaction is low and the mechanism unknown. Understanding molecular features associated with gene replacement is important and may lead to an increase in the efficiency of the process. In this study, we investigated gene replacement in mammalian cells using a powerful assay system that permits efficient recovery of the product(s) of individual recombination events at the haploid, chromosomal µ- locus in a murine hybridoma cell line. The results showed that (i) heteroduplex DNA (hDNA) is formed during mammalian gene replacement; (ii) mismatches in hDNA are usually efficiently repaired before DNA replication and cell division; (iii) the gene replacement reaction occurs with fidelity; (iv) the presence of multiple markers in one homologous flanking arm in the replacement vector did not affect the efficiency of gene replacement; and (v) in comparison to a genomic fragment bearing contiguous homology to the chromosomal target, gene targeting was only slightly inhibited by internal heterology (pSV2neo sequences) in the replacement vector.

GENE targeting studies commonly utilize the omega ()-form or "ends-out" gene replacement vector to make predetermined alterations in chromosomal genes (WALDMAN 1992 ; BERTLING 1995 ). Gene replacement vectors usually consist of a dominant selectable drug resistance gene flanked on both sides by homology to the desired genomic locus. Correct homologous recombination replaces endogenous chromosomal sequences with those in the transferred DNA, including the selectable marker. Drug-resistant cells are usually selected in batch culture whereupon targeted cells are identified by appropriate assays such as screening by PCR or Southern analysis. In spite of its routine use as a tool for targeted genome alteration, mechanisms associated with the mammalian gene replacement reaction are not well understood.

Previously, we reported a gene targeting system that detects homologous recombination between transferred vector DNA and the haploid, chromosomal immunoglobulin heavy chain locus in mouse hybridoma cells (BAKER et al. 1988 ). In the present study, this system was modified and exploited to investigate mechanisms of mammalian gene replacement. A gene replacement vector was constructed that contained the selectable neo gene flanked by homology to the constant region of the immunoglobulin heavy chain µ- and -locus in the hybridoma cells (the Cµ and C region, respectively). The SV40 early region enhancer was removed from the vector-borne neo gene creating an "enhancer-trap" gene replacement vector. As shown previously (BAUTISTA and SHULMAN 1993 ; NG and BAKER 1998 ), similar enhancer-trap vectors enrich for hybridoma cells targeted at the chromosomal µ-locus. This occurs because the µ-locus supplies the enhancer (or equivalent) activity required for expression of the enhancerless neo gene in the targeted vector, whereas most random sites of vector integration in the hybridoma genome do not. To address recombination mechanisms, the vector-borne Cµ region was marked with six restriction enzyme sites distinguishable from the corresponding sites in the chromosomal Cµ region. To isolate independent recombinants, immediately following electroporation, the hybridoma cells were segregated to individual wells of tissue culture plates and placed under G418 selection as described previously (NG and BAKER 1999A ; LI and BAKER 2000A ). Use of the enhancer-trap gene targeting vector in association with the segregation of the hybridoma cells immediately following gene transfer has three novel advantages in the study of mammalian recombination mechanisms: (i) it ensures that each G418^R recombinant arises from a single cell deposited in the culture well; (ii) it results in retention of the G418^R product(s) of individual gene replacement events for molecular analysis; and (iii) it permits targeted cells to be identified without selection bias favoring recovery of a functional recombinant µ-gene. Using this system, we have identified and characterized several independent gene replacement events. The results reveal important new insights into the mammalian gene replacement process.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Recipient hybridoma cell line:
The haploid, chromosomal immunoglobulin µ- heavy chain locus in the igm482 hybridoma cell line was used as the target for gene replacement (Fig 1). The hybridoma cell line igm482 was isolated from the wild-type Sp6 hybridoma cell line (subclone Sp6/HL) that makes cytolytic, polymeric IgM(-chain) specific for the hapten trinitrophenyl (TNP; KOHLER and SHULMAN 1980 ; KOHLER et al. 1982 ). The distinguishing feature of the igm482 hybridoma cell line is that it bears a 2-bp deletion in the third constant region exon of the TNP-specific chromosomal µ-gene (Cµ3). The 2-bp Cµ3 deletion destroys an XmnI restriction enzyme site normally present in the wild-type Sp6/HL Cµ3 exon, creating in its place a TfiI site (position 4 in the igm482 Cµ region shown in Fig 1). The igm482 mutation results in the synthesis of a truncated µ-chain lacking the Cµ4 domain that is assembled into a monomeric form of IgM. Unlike the normal TNP-specific IgM synthesized by the wild-type Sp6/HL hybridoma cell line, the mutant IgM made by the igm482 cells is unable to activate complement-dependent lysis of TNP-coupled sheep red cells (TNP-SRC). As described below, the different properties of the IgM made by the mutant igm482 and wild-type Sp6/HL hybridoma cell lines were exploited in the analysis of the recombinant hybridoma cell lines. Aside from the different Cµ3 exons, the chromosomal µ- region in the mutant igm482 and wild-type Sp6/HL hybridoma cell lines is otherwise isogenic. The methods used for hybridoma cell culture have been described (KOHLER and SHULMAN 1980 ; KOHLER et al. 1982 ).

View larger version (29K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Gene replacement at the µ- locus. The structure of the haploid, chromosomal immunoglobulin heavy chain µ- locus in the recipient murine hybridoma cell line, igm482, is presented along with the chromosomal µ- structure in a recombinant hybridoma cell line generated following gene replacement with the vector, pCµM1-6C. The chromosomal and vector-borne Cµ regions are distinguishable by the indicated six pairs of restriction enzyme site polymorphisms at positions numbered relative to the Bst1107 site that defines the beginning of the vector-borne Cµ region of homology (position 0 bp). Markers diagnostic of the vector-borne Cµ region are denoted in boldface type while the corresponding markers from the chromosomal Cµ region are indicated in roman type. As gene replacement has the potential to generate different combinations of chromosomal and/or vector-borne markers, the six corresponding positions in the recombinant Cµ region are designated by a question mark (?). The primer pair AB9703/AB9745 bind outside the Cµ region of homology in the replacement vector at positions described previously (NG and BAKER 1999A ; LI and BAKER 2000A ). They generate a specific 4.8-kb PCR product from the recombinant Cµ region as shown. Probe fragments: Cµ-specific probe fragment F is an 870-bp XbaI/BamHI fragment while probe G is a 762-bp PvuII fragment from the neo gene. Abbreviations: Cµ, µ-gene constant region; C, gene constant region; VHTNP, TNP-specific heavy chain variable region; neo, neomycin phosphotransferase gene. The thick line represents the vector, pSV2neo (SOUTHERN and BERG 1981 ). The figure is not drawn to scale.

Gene replacement vectors:
The 13.1-kb -form, enhancer-trap vector, pCµ_M1-6C (Fig 1) was used in the gene replacement studies. The backbone of this vector is derived from pSV2neo (SOUTHERN and BERG 1981 ) from which the 372-bp NsiI/NdeI fragment encompassing the SV40 early region enhancer sequence responsible for neo gene expression was removed. On the left and right flanking sides of the enhancer-deleted pSV2neo are genomic DNA segments from the wild-type Sp6/HL hybridoma cell line consisting of a 4.2-kb Bst1107/XbaI Cµ region fragment and a 3.5-kb SpeI/SacI C region fragment, respectively. The position of these DNA segments with respect to the chromosomal µ- region in the mutant igm482 hybridoma cell line is indicated in Fig 1. The Cµ and C homology regions share a 63-bp overlap between the SpeI and XbaI sites located 3' of Cµ. The vector-borne Cµ and C homology regions are isogenic with those of the igm482 hybridoma cell line except for the following modifications. As described previously (NG and BAKER 1999A ), site-directed mutagenesis was used to create the novel vector-borne Cµ region KpnI, EcoRV, DraI, AatII, and ScaI sites located at positions 150 bp, 557 bp, 1117 bp, 1601 bp, and 2041 bp, respectively, relative to the Bst1107 half site that marks the beginning of the vector-borne Cµ region (nucleotide position 0 bp). The vector-borne sites replace the endogenous Cµ region AvaII, SacI, AflII, EarI, and NheI sites located at genomic positions 93 bp, 503 bp, 1053 bp, 1537 bp, and 1968 bp, respectively, according to the numbering in GOLDBERG et al. 1981 . In addition, as indicated above, the 2-bp mutant igm482 deletion creates a TfiI site in the Cµ3 exon as opposed to the wild-type XmnI site in the corresponding position in the vector-borne Cµ region (position 1506 bp). Therefore, the Cµ region in pCµ_M1-6C bears six diagnostic markers that distinguish it from the corresponding endogenous sites in the recipient igm482 hybridoma cell line. Originally, the vector-borne markers were constructed in the Cµ region of the gene targeting sequence insertion vector pCµEn^-_M1-6 by site-directed mutagenesis with all sites being verified by DNA sequencing (NG and BAKER 1999A ). For this study, the markers were moved into the pCµ_M1-6C gene replacement vector by subcloning appropriate Cµ region segments.

To investigate potential effects of the Cµ region markers in pCµ_M1-6C on the efficiency of gene replacement, the vector pCµ_WTC was constructed. This was accomplished by swapping a 1.9-kb Bst1107/DraIII fragment containing the marked Cµ region in pCµ_M1-6C with the corresponding segment from the genomic, wild-type Sp6/HL Cµ region. With the exception of the 2-bp mutant igm482 Cµ3 deletion, the µ-genes of the wild-type Sp6/HL and mutant igm482 hybridoma cell lines are the same. Thus, for illustrative simplicity, the position of the 1.9-kb Bst1107/DraIII wild-type Cµ segment is presented in Fig 1 in the corresponding position of the mutant igm482 chromosomal Cµ region. Restriction enzyme sites that were convenient for this Cµ fragment exchange did not remove the ScaI polymorphism (position 2041 bp in the vector-borne Cµ region), but otherwise the Cµ region in pCµ_WTC was wild type.

To examine possible effects associated with the heterology created by the 5.4-kb enhancer-trap pSV2neo sequences in pCµ_M1-6C and pCµ_WTC, an 8.2-kb Bst1107/NdeI fragment of Cµ-C genomic DNA (Fig 1) was isolated from cloned, genomic DNA of the wild-type Sp6/HL hybridoma and tested in the gene targeting reaction. Again, for illustrative purposes, the position of this segment in the mutant igm482 chromosomal µ-gene is shown in Fig 1.

DNA transfer and isolation of independent G418^R transformants:
The vector pCµ_M1-6C (8.7 pmol) was transferred to 2 x 10⁷ recipient igm482 hybridoma cells by electroporation as described (BAKER et al. 1988 ). Trypan blue staining revealed that typically 50% of the hybridoma cells survived electroporation. Independent G418^R transformants were isolated by a plating procedure described previously (NG and BAKER 1999A ) that ensures each transformant is derived from a single G418^R cell deposited in the culture well and that the product(s) of each gene replacement event are retained in the culture well for analysis.

In other studies, the frequency of gene targeting was measured in a different way. In separate experiments, 8.7 pmol of either pCµ_M1-6C, pCµ_WTC or the genomic Bst1107/NdeI Cµ-C region fragment were transferred to 2 x 10⁷ recipient igm482 hybridoma cells by electroporation (BAKER et al. 1988 ). After 48 hr, the frequency of gene targeting was determined by measuring plaque-forming cells (PFC) in a sensitive, TNP-specific plaque assay as described previously (BAKER et al. 1988 ).

Identification and analysis of targeted, G418^R recombinants:
Genomic DNA was prepared from individual G418^R transformants generated following transfer of pCµ_M1-6C by the method of GROSS-BELLARD et al. 1973 . Individual DNAs were screened by a specific PCR assay utilizing the primer pair AB9703/AB9745 that binds outside the vector-borne Cµ region of homology as described previously (NG and BAKER 1999A ; LI and BAKER 2000A ). As shown in Fig 1, this primer pair generates a specific 4.8-kb PCR product from the Cµ region of targeted recombinants. Hybridoma cells identified in this first screening were further characterized by Southern analysis for the diagnostic restriction enzyme fragment sizes predicted by the targeting event (Fig 1). For Southern analysis, restriction enzymes were purchased from New England Biolabs Inc. (Beverly, MA), Amersham Pharmacia Biotech Inc. (Baie d'Urfé, Québec), and Canadian Life Technologies Inc. (Burlington, Ontario) 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 (SAMBROOK et al. 1989 ).

For analysis of Cµ region genetic markers in the recombinants, the 4.8-kb PCR product was tested for its resistance or sensitivity to digestion with each of the diagnostic restriction enzymes indicated in Fig 1. As described earlier (NG and BAKER 1999A ), each enzyme produces diagnostic fragment sizes that can be resolved by standard gel electrophoresis, thus permitting assignment of the various genetic markers to the correct positions.

Analysis of IgM production by the various hybridoma cell lines was accomplished by testing culture supernatants for the presence of either mutant monomeric or wild-type polymeric TNP-specific IgM by hemagglutination and complement-dependent lysis assays of TNP-SRC as described (KOHLER and SHULMAN 1980 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Description of experimental system:
The single copy of the chromosomal immunoglobulin µ- region in the murine hybridoma cell line, igm482, serves as the target for recombination with the 13.1-kb -form or "ends-out" enhancer-trap, gene replacement vector, pCµ_M1-6C (Fig 1). In pCµ_M1-6C, the Cµ and C arms of homology used to effect gene replacement are 4.2 kb and 3.5 kb, respectively, and are separated by pSV2neo sequences. The enhancer-trap vector enriches significantly for gene targeting events at the chromosomal µ- locus (BAUTISTA and SHULMAN 1993 ; NG and BAKER 1998 ), permitting recovery of independent recombinants by a plating procedure described previously (NG and BAKER 1999A ; LI and BAKER 2000A ). The plating procedure ensures that each G418^R recombinant arises from a single cell and that the G418^R product(s) of gene replacement are retained for molecular analysis. To examine recombination mechanisms, the Cµ region of pCµ_M1-6C was marked with six restriction enzyme site polymorphisms, permitting it to be distinguished from the corresponding markers in the mutant igm482 chromosomal Cµ region. The various restriction enzyme site markers, along with their positions relative to the Bst1107 site that marks the beginning of the Cµ region (position 0 bp), are indicated in Fig 1.

Isolation of gene replacement events:
A total of 163 independent G418^R colonies were obtained following electroporation of recipient mutant igm482 hybridoma cells with the enhancer-trap pCµ_M1-6C vector. As hybridoma cell survival averaged 50% following electroporation, the frequency of G418^R transformants/cell was = 1.63 x 10^-5. To identify putative cases in which pCµ_M1-6C had interacted with the haploid chromosomal µ- locus, the G418^R transformants were screened by PCR using the primer pair AB9703/AB9745, which as shown previously (NG and BAKER 1999A ) generate a 4.8-kb product specific for the recombinant Cµ region (Fig 1). No PCR product was detected in 143 of the G418^R transformants but in 20 the correct 4.8-kb PCR product was observed. Southern analysis using Cµ- and neo-specific probe fragments was performed to verify correct gene replacement in these cell lines. Hybridization with Cµ probe F revealed that in 16 of the 20 G418^R transformants, the endogenous 25.0-kb HpaI µ- fragment was replaced with the 14.4-kb HpaI fragment indicative of correct gene replacement. Further verification of correct gene replacement in these hybridoma cell lines was obtained by hybridization of ScaI-digested genomic DNA with neo probe G where the correct 12.6-kb ScaI fragment expected for gene replacement (Fig 1) was observed. Representative examples of correctly targeted recombinants, as determined by the HpaI and ScaI digests, are presented in Fig 2A and Fig B, respectively. In recombinant 25 (Fig 2B), an additional neo-hybridizing band of 6.1 kb is visible, suggesting the possibility that, in addition to gene replacement, a rare random vector integration event might also have occurred. The remaining 4 of the 20 recombinants retained the 25.0-kb HpaI fragment (data not shown), suggesting that the endogenous µ- locus had not been modified by gene replacement. However, in two of these hybridoma cell lines, the 14.4-kb HpaI fragment was also visible. It is possible that these represent cases in which the pCµ_M1-6C vector interacted with the chromosomal target and acquired endogenous sequences including the HpaI site 5' of Cµ (Fig 1), but then ejected from the target locus to integrate elsewhere. These cell lines are still under investigation. Although unlikely, the possibility was considered that the replacement vector may have circularized, suffered a break in either the Cµ or C region, and then undergone single reciprocal crossover within the corresponding region of the chromosome, resulting in targeted vector insertion. In the event crossover occurred between vector-borne and chromosomal Cµ regions, HpaI fragments of 14.4 kb and 23.6 kb would be detected with Cµ-specific probe F whereas crossover between C regions would generate a single 22.1-kb HpaI fragment (predicted fragment sizes not illustrated). However, as is evident from Fig 2A, none of the observed fragment sizes support recombinant generation by this mechanism. In summary, 16/163 (10%) of the G418^R transformants were identified as being correct gene replacement events. Thus, on the basis of the 50% survival of the igm482 hybridoma cells following electroporation (1 x 10⁷ cells), the absolute frequency of gene replacement was = 1.6 x 10^-6 events/cell.

View larger version (139K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Southern analysis of gene targeting events. Genomic DNA from the representative recombinants was digested with (A) HpaI or (B) ScaI electrophoresed through a 0.7% agarose gel and blotted to nitrocellulose. The HpaI blot (A) was hybridized with Cµ-specific probe fragment F. As a control for probe specificity, genomic DNA was included from the recipient igm482 hybridoma cell line in which the chromosomal Cµ-C region is present on a 25.0-kb HpaI fragment (refer also to Fig 1). The ScaI blot (B) was hybridized with neo probe G. As a control for the specificity of probe G, genomic DNA from the hybridoma cell line (49/9) was included. As described previously (NG and BAKER 1999A ), this cell line bears a single targeted copy of a pSV2neo-derived sequence insertion vector and, consequently, probe G is expected to detect a single 20.6-kb ScaI fragment. The sizes of fragments of interest are presented on the left of each blot while relevant DNA marker bands are indicated on the right.

Two features of the isolation procedure are important to re-emphasize. First, plating the hybridoma cells immediately after electroporation makes it highly likely that those destined to become G418^R transformants are segregated to individual culture wells before integration of the transferred DNA into the chromosome. Second, recombinants were recovered from among 3333 wells plated and are expected to follow the Poisson distribution. Accordingly, the probability that the recombinants in a well actually derived from more than one independent recombinant is 0.002. Thus, the G418^R product(s) arising from each independent homologous recombination event are retained for analysis in a single culture well.

Determination of Cµ region marker patterns:
For each recombinant, the specific 4.8-kb Cµ region PCR product (Fig 1) was tested for its sensitivity or resistance to cleavage with restriction enzymes specific for either chromosomal or vector-borne Cµ region markers. As indicated earlier (NG and BAKER 1999A ), the various restriction enzymes generate diagnostic fragment sizes that can be conveniently analyzed by standard gel electrophoresis. This analysis was performed on each of the 16 recombinants (data not shown) and the complete set of results is summarized in Fig 3.

View larger version (32K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Marker patterns in the chromosomal Cµ region of the recombinants. For clarity, the diagram focuses on only the six Cµ region marker positions in the recombinant hybridoma cell lines as determined from restriction enzyme digestion of the Cµ region PCR products. Markers denoting the vector-borne Cµ region are designated in boldface and by a solid circle, while the corresponding markers in the chromosomal Cµ region are indicated in roman type and by an open circle. The Cµ region positions that were heterogeneous (sectored) are denoted by a half-solid circle. Each recombinant was generated from an independent gene replacement event. Recombinant 5 was sectored for five positions in the Cµ region and consisted of four distinct recombinant cell lines in the indicated frequencies. As detailed in the text, the four recombinant cell lines arose from four distinct genotypes determined by the marker heterogeneity at Cµ position 6 (checkered circle).

In 9/16 (56%) of the recombinants (2, 3, 12, 21, 29, 71, 75, 88, and 103), a chromosomal marker was present in every Cµ region position. In recombinants 6, 25, 30, and 32, the Cµ region positions were completely sensitive to cleavage with restriction enzymes specific for either the chromosomal or vector-borne marker. Unlike the other recombinants, one or more Cµ region positions in recombinants 5, 19, and 64 exhibited a mixed cleavage pattern with restriction enzymes diagnostic of either the chromosomal or vector-borne marker, suggesting that these recombinants were sectored (heterogeneous) cultures. As indicated above, the Poisson analysis showed that each recombinant was highly likely to have been derived from a single cell deposited in the culture well. Thus, the presence of a sectored site(s) in these recombinants can be explained in the following way. During gene replacement, heteroduplex DNA (hDNA) may form between vector-borne and chromosomal sequences. In the event mismatches are left unrepaired prior to DNA replication and division of the individual recombinant cell in the culture well, genetically distinct molecules are generated that segregate to different daughter cells. Therefore, each sectored recombinant very likely originated from the failure to completely repair hDNA.

Recombinants 19 and 64 each bore a single mixed Cµ region site. In the case of recombinant 19, the mixed site was at position 6 while for recombinant 64, it was at position 5. The gel analysis revealed that approximately one-half the PCR product was sensitive to digestion with the chromosomal and vector-borne restriction enzymes specific for these sites. This suggested that each recombinant was composed of two distinct cell populations in equal frequency that differed at these sites. Recombinant 5 was mixed for five of the six Cµ region sites and thus it was necessary to determine the linkage relationship between the markers. To accomplish this, recombinant 5 was cloned at 0.1 cell/well and 12 independent G418^R subclones were isolated. The Cµ region marker patterns in the recombinant 5 subclones are expected to reflect the configuration present in the original hDNA intermediate. Thus, restriction enzyme digestion of the Cµ region PCR product was performed for each recombinant 5 subclone. As shown in Fig 3, this analysis revealed that recombinant 5 was composed of not two but four distinct subclone types in approximately equal frequency.

Fidelity of the gene replacement reaction:
The igm482 hybridoma cell line was isolated as a mutant of the wild-type Sp6/HL hybridoma cell line and bears a 2-bp deletion in the Cµ3 exon (KOHLER et al. 1982 ). The 2-bp igm482 deletion destroys an XmnI site normally present in the wild-type Cµ3 exon, creating in its place a novel TfiI site. In addition, it results in production of a truncated µ-protein that when incorporated into the TNP-specific IgM synthesized by the hybridoma cells renders it distinguishable from the normal TNP-specific IgM of the wild-type Sp6/HL hybridoma cell line on the basis of lysis and agglutination of TNP-SRC (KOHLER and SHULMAN 1980 ).

Depending on the outcome of gene replacement, recombinants are expected to bear either the mutant igm482 TfiI or the wild-type XmnI restriction enzyme site in exon Cµ3 and to synthesize the mutant igm482 or wild-type IgM, respectively. As all other markers are located in Cµ introns they are not expected to affect µ-chain synthesis. In this study, recombinants were isolated on the basis of a predicted change in µ-gene structure rather than function. Therefore, an unbiased indication of the fidelity of the gene replacement reaction can be obtained by examining IgM production in the recombinants.

Recombinants 5, 25, 30, and 32 bear the wild-type (vector-borne) XmnI site in the Cµ3 exon (Fig 3). As shown in Table 1, the IgM produced by these recombinants has the same pattern of complement-dependent lysis and agglutination of TNP-SRC as that of the wild-type Sp6/HL hybridoma cell line. In the remaining recombinants, the TfiI site is present in exon Cµ3 and, like the mutant igm482 hybridoma, the IgM made by these cells cannot lyse TNP-SRC and can only agglutinate TNP-SRC in the presence of anti-µ-serum, a distinguishing feature of the IgM made by mutant igm482 hybridoma (KOHLER and SHULMAN 1980 ). Therefore, the results in Table 1 are fully in accord with recombinants possessing a functional Cµ region in which the Cµ3 exon bears either the mutant igm482 or wild-type sequence. These results, together with the Cµ region marker analysis in Fig 3, support a gene replacement mechanism that proceeds with fidelity.

Cµ region markers do not affect the efficiency of gene replacement:
To investigate whether the various Cµ region markers affected the efficiency of gene replacement, the marked Cµ region in pCµ_M1-6C was exchanged for the wild-type Cµ region generating the replacement vector, pCµ_WTC. Available restriction enzyme sites for fragment exchange did not remove the ScaI polymorphism at position 6 in the vector-borne Cµ region. However, with the exception of this site, the Cµ region in pCµ_WTC was otherwise wild type. As an assay, we took advantage of the fact that recombination between transferred DNA bearing a wild-type Cµ region and the mutant igm482 chromosomal µ-gene can regenerate a wild-type µ-gene sequence and restore normal IgM production in the recombinants, permitting them to be detected as PFC in a sensitive TNP-specific plaque assay (BAKER et al. 1988 ). As shown in Table 2, the mean absolute frequency of TNP-specific PFC generated with pCµ_WTC was 0.76 x 10^-6 events/cell whereas with pCµ_M1-6C, it was 0.77 x 10^-6 events/cell. According to a two-sampled t-test, the means are not significantly different (P = 0.95). Thus, the Cµ region markers in pCµ_M1-6C do not adversely affect the frequency of gene replacement. Also, the absolute frequency of TNP-specific PFC generated with pCµ_M1-6C in the plaque assay was similar to an independent determination of this frequency derived from the fraction of recombinants making normal TNP-specific IgM reported in Table 1, multiplied by the absolute frequency of gene replacement as determined from the Southern analysis of G418^R colonies recovered in the 96-well tissue culture plates [i.e., (fraction of recombinants making normal TNP-specific IgM/total recombinants) x the absolute frequency of recombination = () x (1.6 x 10^-6) = 0.4 x 10^-6 events/cell]. This suggests that a common gene replacement mechanism is involved in generating recombinants in the two assay procedures.

Gene replacement in the absence of the vector-borne neo gene:
Further experiments were conducted to determine whether the heterology created by the pSV2neo vector backbone in pCµ_M1-6C reduced the frequency of gene targeting. To answer this question, an 8.2-kb Bst1107/NdeI fragment spanning the µ- region of the wild-type Sp6/HL hybridoma cell line was isolated from cloned genomic DNA. As shown in Fig 1, this genomic fragment contained the Cµ-C homology region present in pCµ_M1-6C but as an uninterrupted segment. The genomic fragment was transferred to recipient igm482 hybridoma cells by electroporation and the frequency of generating TNP-specific PFC was compared to that obtained with pCµ_M1-6C. As shown in Table 3, the frequency of TNP-specific PFC obtained with pCµ_M1-6C was 0.73 x 10^-6 events/cell, a value similar to that obtained with pCµ_WTC reported in Table 2. A slightly higher frequency of generating TNP-specific PFC (approximately twofold) was obtained with the genomic Cµ-C fragment, which, according to a two-sampled t-test, was significant (P = 0.003). Thus, the results suggested that the 5.4 kb of pSV2neo sequences in the gene replacement vectors had a marginal inhibitory effect on the frequency of gene replacement.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In this study, we investigated targeted gene replacement at the haploid, chromosomal immunoglobulin µ- locus in a murine hybridoma cell line using the -form enhancer-trap vector, pCµ_M1-6C. Six novel restriction enzyme sites replaced the corresponding endogenous sites in the vector-borne Cµ region. These restriction enzyme site polymorphisms served as genetic markers permitting the contribution of vector-borne and chromosomal Cµ region sequences in the recombinant products to be distinguished. Independent recombinants were isolated and identified by a powerful assay system in which the G418^R product(s) of each individual homologous recombination event was retained for molecular analysis. In addition, independent recombinants were isolated without any selection bias favoring recovery of the functional product(s) of recombination.

A common feature of the 16 recombinants analyzed in this study was the loss of one or more vector-borne markers from the beginning of the Cµ region and their replacement with the corresponding chromosomal sequence. In all recombinants, the KpnI site 150 bp from the start of the vector-borne Cµ region was replaced with the chromosomal AvaII site. In 13/16 recombinants, more extensive loss of vector-borne markers occurred. In recombinants 19 and 64, chromosomal markers replaced vector-borne markers up to and including the XmnI site at position 1506 bp, while in recombinant 25, vector-borne markers were replaced up to the DraI site at position 1117 bp. However, in a significant fraction of the recombinants (9/16 or 56%), all vector-borne markers were replaced with chromosomal markers. In principle, vector-borne markers might have been removed by degradation from the end of the replacement vector with the deleted information being replaced by DNA synthesis using the homologous chromosomal sequence as template. While previous studies have shown that degradation can occur at DNA ends during mammalian gene transfer it does not appear to be extensive (SHULMAN et al. 1990 ; HASTY et al. 1992 ; JIANG et al. 1992 ; PFEIFFER et al. 1994 ; RICHARD et al. 1997 ; ELLIOT et al. 1998 ). Further, our previous gene targeting studies using insertion (O)-type vectors also suggested that DNA ends created by the vector-borne double-strand break were usually subject to only slight degradation (NG and BAKER 1999A ; LI and BAKER 2000A , LI and BAKER 2000B ). Thus, although the above studies are consistent with the possibility that degradation might have removed some vector-borne markers near the beginning of the Cµ region, they are not consistent with the extensive amount of degradation (exceeding 2041 bp) required to remove all vector-borne markers in the major class consisting of recombinants 2, 3, 12, 21, 29, 71, 75, 88, and 103. Extensive terminal degradation is also inconsistent with the retention of vector-borne markers near the beginning of the Cµ region in recombinants 5, 6, 30, and 32.

An alternate explanation for the replacement of vector-borne with chromosomal markers is that of mismatch repair (MMR) of hDNA. What is the evidence for hDNA formation in the recombinants? The strongest evidence is the sectoring observed in recombinants 5, 19, and 64. In yeast and fungi, sectoring is indicative of a region of hDNA that did not undergo complete MMR prior to DNA replication and division in the single cell undergoing recombination and that can be detected in both meiosis and mitosis (PETES et al. 1991 ). In this study, the pattern of sectoring observed in recombinant 5 suggested that hDNA formation was extensive beginning prior to the second Cµ marker position (557 bp) and spanning all remaining 3' marker positions over a distance of at least 1484 bp. Unexpectedly, recombinant 5 was composed of not two but four distinct genotypes in approximately the same frequency. A proposed mechanism for the generation of this unusual recombinant is presented below. In recombinant 19, hDNA spanned at least the last two marker positions for a minimum of 440 bp, while in recombinant 64 the evidence suggested that, at a minimum, hDNA encompassed the fifth Cµ marker at position 1601 bp. Although only three sectored recombinants were found, it is unlikely that this was due to any failure to recover hybridoma cells bearing hDNA. That is, the isolation procedure involved hybridoma cells being segregated to the culture wells within 1 hr following electroporation with the replacement vector. This short interval would have provided insufficient time for the hybridoma cells to complete the process of gene replacement and undergo cell division prior to their deposition in the culture well, events that would have obscured evidence of hDNA.

While the position of sectored sites provided the best evidence for hDNA, other evidence was indirect and based on the pattern of MMR. In recombinant 6, it was concluded that hDNA spanned a minimum of 1484 bp encompassing Cµ region markers between the second and sixth positions inclusive (557–2041 bp). Within this region of hDNA, MMR occurred in patches: some mismatches were converted to the chromosomal sequence whereas others underwent restoration to the vector-borne sequence. Patchiness in MMR was also revealed in recombinant 19 where, in the hDNA tract that spanned marker positions 5 and 6, marker 5 was restored to the vector-borne sequence while marker 6 remained unrepaired. It was argued above that, in the major class of recombinants (2, 3, 12, 21, 29, 71, 75, 88, and 103), replacement of the majority of vector-borne with chromosomal markers could not easily be accounted for by a mechanism involving DNA end degradation. If so, then much of the gene conversion observed in this recombinant class can also be explained on the basis of MMR of hDNA. If, as suggested from the data above, hDNA does play an important role in the mammalian gene replacement reaction, then the observation that most Cµ region positions in the recombinants contained either a chromosomal or vector-borne marker suggests that, in individual hybridoma cells undergoing gene replacement, mismatches in hDNA are usually efficiently repaired prior to DNA replication and cell division.

The finding that recombinant 5 was sectored in five of the six Cµ positions was very surprising because sectoring was not as frequent in the other recombinants. Another very unusual feature was that this recombinant consisted of four distinct genotypes rather than the two expected if hDNA were generated by a single gene replacement event in which markers had escaped MMR. As indicated in the RESULTS section, according to the Poisson distribution the probability that the recombinants in a well actually derived from more than one independent recombinant is 0.002. This makes it highly unlikely that two gene replacement events occurred in separate hybridoma cells in the culture well. Further, such an occurrence would require unrepaired hDNA to persist in both cells and for markers to be present in the reciprocal linkage pattern observed. A proposed mechanism that more readily accounts for the Cµ region marker pattern in recombinant 5 is depicted in Fig 4. As shown in Fig 4A, a gene replacement event occurs in the µ- locus of a single hybridoma cell with hDNA being generated across the Cµ region. The vector-borne KpnI site may have been replaced with the chromosomal AvaII site by repair synthesis. The Cµ region hDNA escapes MMR and, following DNA replication, two sister chromatids are generated that differ in this region (Fig 4B). A sister chromatid exchange event then occurs, reversing the linkage of the last pair of markers at position 6 [i.e., the chromosomal NheI site (open circle) and the vector-borne ScaI site (solid circle) generating the recombinant in Fig 4C]. Following this, the cell divides to form two daughter cells bearing unrepaired hDNA (Fig 4D). The hDNA remains unrepaired and, following a second division, four genotypically distinct subclones types are formed in equal proportion (Fig 4E). If this interpretation is correct, recombinant 5 is a very interesting hybridoma cell line. The properties of multiple recombination and unrepaired hDNA are similar to some MMR deficiencies in the mouse; for example, inactivation of the mouse Msh2 gene results in a mismatch repair deficiency, hyper-recombination, and predisposition to cancer (DE WIND et al. 1995 ). Recombinant 5 is under further investigation in our laboratory.

View larger version (34K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Proposed mechanism for the generation of recombinant 5. As indicated in RESULTS, the mitotically sectored recombinant 5 bears four distinct genotypes. The following pathway is proposed to explain this unusual mixed genotype. For further details, refer to the text.

In this study, recombinants were isolated and identified by procedures not normally utilized in studies of mammalian homologous recombination, that is, by methods that did not require the recombination products to be functional. This provided the opportunity to investigate whether mammalian gene replacement occurs with fidelity. According to our analysis of diagnostic Cµ region markers and µ-chain production in the recombinants, the overall process appears faithful. These results agree with the majority of recombinants examined in our previous studies of targeted vector integration (BAKER et al. 1988 ; BAKER and SHULMAN 1988 ; NG and BAKER 1999A , NG and BAKER 1999B ) and with the results of other gene targeting studies (ZHENG et al. 1991 ; THOMAS et al. 1992 ). Also, sequencing studies have suggested that intrachromosomal homologous recombination in mammalian cells occurs with fidelity (STACHELEK and LISKAY 1988 ). It is known that illegitimate integration of transfected DNA at random positions in the mammalian genome can be associated with rearrangements near the integration site (ROTH and WILSON 1988 ). However, at least for targeted vector integration, this does not seem to be a problem because the site of vector linearization is usually restored in the vast majority of recombinants in which this has been examined (SMITHIES et al. 1985 ; BAKER et al. 1988 ; BAKER and SHULMAN 1988 ; NG and BAKER 1998 , NG and BAKER 1999A ; LI and BAKER 2000A , LI and BAKER 2000B ). Of further relevance to the issue of whether the genome of targeted cells contains unwanted genetic changes is the observation that gene targeting is usually not associated with illegitimate integration of vector DNA, a potentially mutagenic event (BAKER et al. 1988 ; BAKER and SHULMAN 1988 ; BOLLAG et al. 1989 ; WALDMAN 1992 , WALDMAN 1995 ; BERTLING 1995 ; NG and BAKER 1998 , NG and BAKER 1999A , NG and BAKER 1999B ; LI and BAKER 2000A , LI and BAKER 2000B ). Nevertheless, some studies suggest that gene targeting might be error prone (THOMAS and CAPECCHI 1986 ; DOETSCHMAN et al. 1988 ; BRINSTER et al. 1989 ) and, more specifically, that gene replacement might be less accurate than targeted vector integration (HASTY et al. 1991 ). The precise mechanisms involved in generating these unwanted genetic alterations are not known although, in one case (THOMAS and CAPECCHI 1986 ), it was pointed out that the changes might have reflected peculiarities associated with the recombining sequence. It has since been learned that the efficiency of gene targeting is similar whether insertion or replacement vectors are utilized (THOMAS and CAPECCHI 1987 ; DENG and CAPECCHI 1992 ). It has also been shown that the fidelity of gene targeting is reduced when the vector bears DNA that is nonisogenic to the target (DENG and CAPECCHI 1992 ; TE RIELE et al. 1992 ). In view of the hDNA that is formed during gene replacement, as shown here, and during targeted vector integration, as shown previously (NG and BAKER 1999A , NG and BAKER 1999B ; LI and BAKER 2000A , LI and BAKER 2000B ), it is possible that a low error rate might be associated with MMR processing or perhaps with repair synthesis of double-stranded gaps, as suggested previously (STRATHERN et al. 1995 ). Another critical factor is the length of homology that can affect both the efficiency (SHULMAN et al. 1990 ; DENG and CAPECCHI 1992 ) and accuracy of gene targeting (THOMAS et al. 1992 ). For example, the accuracy of gene replacement can be reduced when the amount of target-homologous DNA on one arm falls below 1 kb (THOMAS et al. 1992 ) and this can generate recombinants bearing one homologous and one nonhomologous junction by a mechanism involving one-sided invasion (BELMAAZA et al. 1990 ; BERENSTEIN et al. 1992 ). Thus, one or more of the above factors may have contributed to the mutations observed in the earlier gene targeting studies (THOMAS and CAPECCHI 1987 ; DOETSCHMAN et al. 1988 ; BRINSTER et al. 1989 ; HASTY et al. 1991 ). A conclusion from most studies and one that appears to be supported by the present work is that gene targeting, like other forms of homologous recombination is largely a faithful process. The issue of fidelity in gene targeting is important from the basic viewpoint of understanding mechanisms of homologous recombination. It is also important from the practical standpoint of its use in directed genome modification because, ideally, gene targeting would involve precise alteration of a chromosomal gene in the absence of extraneous changes.

To determine whether the Cµ region markers in pCµ_M1-6C had any effect on gene targeting we compared the efficiency of generating TNP-specific PFC between pCµ_M1-6C and the vector pCµ_WTC in which the Cµ region was wild type. No significant difference in the gene targeting frequency was observed for the two vectors, suggesting that the Cµ markers did not reduce the recombination frequency. A similar result was also obtained in gene targeting studies using enhancer-trap insertion vectors (NG and BAKER 1999A ). Previous studies have suggested that mismatches between two recombining sequences can reduce the efficiency of homologous recombination although usually greater than a few percentage points mismatch is required (WALDMAN and LISKAY 1987 ; DENG and CAPECCHI 1992 ; TE RIELE et al. 1992 ). The genetic markers in this study contribute <1% heterology to the vector-borne Cµ region. Further, with the exception of the markers at positions 4 and 5, the remaining adjacent markers are separated by a few hundred base pairs of sequence homology, which is greater than the 134–232 bp of uninterrupted homology required for efficient mammalian intrachromosomal recombination (WALDMAN and LISKAY 1988 ). Out of concern that the 5.4 kb of pSV2neo vector sequences in pCµ_M1-6C (and pCµ_WTC) might have exerted a strong inhibitory effect on the frequency of gene replacement, we compared the frequency of generating TNP-specific PFC with pCµ_M1-6C to that obtained with an isolated fragment of genomic DNA in which the Cµ-C homology region was contiguous. The results revealed only a slight (approximately twofold) increase in gene targeting frequency with the genomic Cµ-C fragment.

The possibility was considered that gene replacement as measured by the TNP-specific plaque assay might not occur by the same mechanism that generated the G418^R recombinants identified by Southern analysis of transformants arising in the 96-well tissue culture plate assay. That is, generation of a TNP-specific PFC would require only that the mutant igm482 chromosomal Cµ3 exon be corrected to the wild-type sequence present in the Cµ region of the replacement vector or the genomic Cµ-C fragment and not necessarily that the transfected DNA be incorporated into the chromosome such as occurs in the gene replacement reaction. However, the similarity in the absolute frequency of TNP-specific PFC generated with pCµ_M1-6C and, for the same vector, the absolute frequency of G418^R recombinants making normal TNP-specific IgM recovered from the 96-well tissue culture plates suggests that the two assay systems recover recombinants that are probably generated by a common gene replacement mechanism.

Of the 20 recombinants initially identified in the PCR screening of the G418^R transformants, 16 bore the expected structure for correct replacement targeting at the endogenous µ- locus. However, 4 of the 20 cell lines had an intact endogenous target site but were PCR positive probably as a consequence of the vector acquiring sequences from the target locus prior to random integration. This phenomenon is likely the same or similar to the gene conversion-like process described previously (ADAIR et al. 1989 ; ELLIS and BERNSTEIN 1989 ), which is consistent with one-sided invasion (BELMAAZA et al. 1990 ). In this study, the frequency of this event might be an underestimate as the one-sided invasion process would not necessarily always extend to the primer AB9703 binding site and give a positive PCR signal. The generation of hybridoma cell lines bearing correct replacement events, as well as those in which the transferred vector appears to have interacted transiently with the target locus, suggests that the two ends of the vector are behaving independently during the recombination process. This predicts another class of recombinants which, while not observed in this study, have been described previously: those in which one end of the replacement vector undergoes homologous recombination with the target locus while the other end undergoes illegitimate recombination nearby (BERENSTEIN et al. 1992 ; DELLAIRE et al. 1997 ).

Several features of the gene replacement reaction documented in this study are similar to those obtained in a previous study of targeted vector integration utilizing an enhancer-trap insertion (O-type) vector in which the Cµ region of homology contained the same genetic markers (NG and BAKER 1999A ). That is, the absolute frequencies of gene targeting are similar (10^-6 events/cell), hDNA formation can be extensive in both cases, small heterologies had no measurable effect on the gene targeting frequency, and MMR tended to occur prior to DNA replication in both circumstances. The latter result suggests that the hybridoma cell lines are normally MMR proficient at least for these simple mismatches and that the repair of hDNA is not influenced by the way in which it was generated during replacement or integrative forms of gene targeting. Although targeted vector integration is consistent with the double-strand-break repair (DSBR) model of recombination (VALANCIUS and SMITHIES 1991 ; NG and BAKER 1999A ; LI and BAKER 2000A , LI and BAKER 2000B ), gene replacement is not expected to occur by standard DSBR. Thus, in theory, the two modes of gene targeting could have been quite different. However, as it turns out, the results suggest that the two modes of gene targeting are similar in some respects.

These similarities between gene replacement and targeted vector insertion in mammalian cells appear to contrast with data for the comparable events in the yeast Saccharomyces cerevisiae. In S. cerevisiae, targeted vector insertion is also consistent with the DSBR model of recombination (ORR-WEAVER et al. 1981 ; SZOSTAK et al. 1983 ), whereas assimilation of a single strand of incoming DNA into the chromosome is suggested to be an important mechanism of gene replacement (LEUNG et al. 1997 ). Extensive hDNA is formed during gene replacement in yeast, but during targeted vector insertion, formation of hDNA around the double-strand break might be limited to perhaps a few hundred base pairs (ORR-WEAVER et al. 1988 ; SWEETSER et al. 1994 ). Also, the efficiency of gene replacement and targeted vector insertion may differ in yeast (LEUNG et al. 1997 ).

In principle, the mammalian gene replacement reaction is consistent with three mechanisms. One mechanism involves assimilation of a single strand of the vector into the chromosome, a process that would generate hDNA across the entire region. A second mechanism involves two independent crossover events confined to the homologous ends of the replacement vector. No hDNA is formed and internal double-stranded DNA in the vector is inserted into the chromosome. A third mechanism also postulates two crossing-over events, but in this case associated with extensive hDNA formation. Regarding how vector ends pair with their homologous chromosomal sequence, the marker patterns in recombinants 5, 6, 30, and 32 suggest that, in the case of the Cµ region, pairing initiates near the beginning of the homology region. Pairing between vector-borne and chromosomal strands near the beginning of the Cµ region would allow for potential crossover near the DNA end and thus readily explain the marker pattern in recombinants 30 and 32. In addition, it would provide the opportunity for hDNA formation to begin near the start of the Cµ region and span internal sites as suggested from the marker patterns in at least recombinants 5 and 6.

As indicated above, LEUNG et al. 1997 favored strand assimilation to explain replacement of a chromosomal allele in S. cerevisiae by a single, contiguous 2-kb homologous DNA fragment released by HO endonuclease cleavage from a different position in the yeast genome. In contrast, a crossover-at-ends model was proposed to explain gene replacement in mammalian cells (DENG et al. 1993 ) and in S. cerevisiae (NEGRITTO et al. 1997 ). In the study by DENG et al. 1993 , the conclusion was based on the 50% frequency of cotransfer of a selectable neo marker and a nonselectable ClaI marker located 3 kb away. However, this result would also appear consistent with two crossing-over events if, as shown in the present study, hDNA formation was extensive and encompassed this site whereupon MMR occurred in the direction of the vector-borne ClaI marker. In the investigation by NEGRITTO et al. 1997 , a DNA fragment bearing a region of 17% mismatch to the chromosomal target was released by HO-endonuclease cleavage from a single-copy plasmid in the nucleus of S. cerevisiae. One of the main pieces of evidence supporting crossover at the fragment ends was the inhibition of gene replacement by mismatches, but only when they were positioned near the terminus of the fragment. However, in their article, LEUNG et al. 1997 argued that these results might also be expected if replacement occurred by strand assimilation.

The results of the present study tend to suggest against strand assimilation as the mechanism. Assimilation of a single strand of the pCµ_M1-6C or pCµ_WTC gene replacement vectors into the chromosome would be expected to be strongly impeded by the 5.4 kb of heterologous pSV2neo sequences. In the event such strand assimilation occurred at all, a large looped-out region would form in the hDNA intermediate and to generate G418^R recombinants, MMR would have to favor the pSV2neo sequences. Alternatively, DNA replication and cell division would permit survival of only one of the two daughter cells. This would preclude hDNA, contrary to what was observed in some recombinants in this study. In contrast, strand assimilation is expected to be much more efficient with the genomic Cµ-C fragment because it shares nearly perfect contiguous sequence homology to the chromosomal target. Another reason to anticipate a more efficient gene replacement reaction with the isolated genomic fragment are studies of gene replacement in yeast that suggest that heterologies in the transforming DNA (such as the pSV2neo sequences in the replacement vectors used in this study) are frequently removed by the MMR system in favor of the sequences residing in the chromosome (LEUNG et al. 1997 ). It was surprising then that our results revealed gene targeting with the Cµ-C fragment to be only slightly (approximately twofold) more efficient than with the replacement vectors. In considering the above information, it seems more likely that gene replacement with either vector DNA or contiguous genomic segments in mammalian cells involves two crossovers that are associated with hDNA formation.

It might be possible to distinguish conclusively between the various gene replacement mechanisms. What is required are genetic markers that are poorly repairable by the mammalian MMR machinery such that, when they are included in homologous flanking DNA on both sides of the selectable marker, evidence for hDNA will frequently be preserved. Recently, we have reported that a small palindrome genetic marker when encompassed within hDNA formed in vivo during homologous recombination in the hybridoma cells avoids MMR generating sectored recombinants at high frequency (LI and BAKER 2000A ). Our current work aims to exploit the palindrome genetic marker to elucidate the mechanism of mammalian gene replacement.

	ACKNOWLEDGMENTS

We thank Philip Ng for helpful comments at the inception of this work and Erin Wever and Leah Read for excellent technical assistance. This research was supported by a Post-Doctoral Fellowship from the Medical Research Society (MRC) of Canada to J.L. and an MRC Operating Grant (MT-14416) to M.D.B.

Manuscript received April 14, 2000; Accepted for publication June 26, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ADAIR, G. M., R. S. NAIRN, J. H. WILSON, M. M. SEIDMAN, and K. A. BROTHERMAN et al., 1989  Targeted homologous recombination at the endogenous adenine phosphoribosyltransferase locus in Chinese hamster cells. Proc. Natl. Acad. Sci. USA 86:4574-4578[Abstract/Free Full Text].

BAKER, M. D. and M. J. SHULMAN, 1988  Homologous recombination between transferred and chromosomal immunoglobulin genes. Mol. Cell. Biol. 8:4041-4047[Abstract/Free Full Text].

BAKER, M. D., N. PENNELL, L. BOSNOYAN, and M. J. SHULMAN, 1988  Homologous recombination can restore normal immunoglobulin production in a mutant hybridoma cell line. Proc. Natl. Acad. Sci. USA 85:6432-6436[Abstract/Free Full Text].

BAUTISTA, D. and M. J. SHULMAN, 1993  A hit-and-run system for introducing mutations into the immunoglobulin heavy chain locus of hybridoma cells by homologous recombination. J. Immunol. 151:1950-1958[Abstract].

BELMAAZA, A., J. C. WALLENBURG, S. BROUILLETTE, N. GUSEW, and P. CHARTRAND, 1990  Genetic exchange between endogenous and exogenous LINE-1 repetitive elements in mouse cells. Nucleic Acids Res. 18:6385-6391[Abstract/Free Full Text].

BERENSTEIN, N., N. PENNELL, C. A. OTTAWAY, and M. J. SHULMAN, 1992  Gene replacement with one-sided homologous recombination. Mol. Cell. Biol. 12:360-367[Abstract/Free Full Text].

BERTLING, W. M., 1995 Gene targeting, pp. 1–44 in Gene Targeting, edited by M. A. VEGA. CRC Press Inc., Boca Raton, FL.

BOLLAG, R. J., A. S. WALDMAN, and R. M. LISKAY, 1989  Homologous recombination in mammalian cells. Annu. Rev. Genet. 23:199-225[Medline].

BRINSTER, R. L., R. E. BRAUN, D. LO, M. R. AVARBOCK, and F. ORAM et al., 1989  Targeted correction of a major histocompatibility class II E gene by DNA microinjection into mouse eggs. Proc. Natl. Acad. Sci. USA 86:7087-7091[Abstract/Free Full Text].

DELLAIRE, G., N. LEMIEUX, A. BELMAAZA, and P. CHARTRAND, 1997  Ectopic gene targeting exhibits a bimodal distribution of integration in murine cells indicating that both intra- and interchromosomal sites are accessible to the targeting vector. Mol. Cell. Biol. 17:5571-5580[Abstract].

DENG, C. and M. R. CAPECCHI, 1992  Reexamination of gene targeting frequency as a function of the extent of homology between the targeting vector and the target locus. Mol. Cell. Biol. 12:3365-3371[Abstract/Free Full Text].

DENG, C., K. R. THOMAS, and M. R. CAPECCHI, 1993  Location of crossovers during gene targeting with insertion and replacement vectors. Mol. Cell. Biol. 13:2134-2140[Abstract/Free Full Text].

DE WIND, N., M. DEKKER, A. BERNS, M. RADMAN, and H. TE RIELE, 1995  Inactivation of the mouse Msh2 gene results in mismatch repair deficiency, methylation tolerance, hyperrecombination, and predisposition to cancer. Cell 82:321-330[Medline].

DOETSCHMAN, T., N. MAEDA, and O. SMITHIES, 1988  Targeted mutation of the Hprt gene in mouse embryonic stem cells. Proc. Natl. Acad. Sci. USA 85:8583-8587[Abstract/Free Full Text].

ELLIOT, B., C. RICHARDSON, J. WINDERBAUM, J. A. NICKOLOFF, and M. JASIN, 1998  Gene conversion tracts from double-strand break repair in mammalian cells. Mol. Cell. Biol. 18:93-101[Abstract/Free Full Text].

ELLIS, J. and A. BERNSTEIN, 1989  Gene targeting with retroviral vectors: recombination by gene conversion into regions of nonhomology. Mol. Cell. Biol. 9:1621-1627[Abstract/Free Full Text].

GOLDBERG, G. I., E. F. VANIN, A. M. ZROLKA, and F. R. BLATTNER, 1981  Sequence of the gene for the constant region of the µ chain of the Balb/c mouse. Gene 15:33-42[Medline].

GROSS-BELLARD, M., P. QUDET, and P. CHAMBON, 1973  Isolation of high-molecular weight DNA from mammalian cells. Eur. J. Biochem. 36:32-38[Medline].

HASTY, P., J. RIVERA-PEREZ, and A. BRADLEY, 1991  The length of homology required for gene targeting in embryonic stem cells. Mol. Cell. Biol. 11:5586-5591[Abstract/