Introduction of specific point mutations into RNA polymerase II by gene targeting in mouse embryonic stem cells: evidence for a DNA mismatch repair mechanism

Precision genome editing in plants via gene targeting and subsequent break-induced single-strand annealing

Genome editing via artificial nucleases such as CRISPR/Cas9 has become popular in plants now. However, small insertions or deletions are major mutations and nucleotide substitutions rarely occur when DNA cleavage is induced. To induce nucleotide substitutions, a base editor utilizing dead or nickase-type Cas9 fused with deaminase have been developed. However, the direction and position of practical substitution are still limited. In this context, homologous recombination (HR)-mediated gene targeting (GT) has advantages because any mutations existing on the donor DNA are copied and passed onto the endogenous DNA. As HR-mediated GT is extremely rare in higher plants, positive-negative selection has been used to isolate cells in which GT has occurred. After successful selection, positive selection marker is no longer needed and should ideally be eliminated. In a previous study, we reported a seamless piggyBac-transposon-mediated marker elimination system. Precision marker elimination efficiency in this system is very high. The piggyBac transposon integrates into the host genome at TTAA elements and excises without leaving a footprint at the excised site, so a TTAA sequence is necessary at the location of a positive selection marker. To compensate for this limitation, we have developed a novel marker elimination system using an I-SceI break and subsequent single-strand annealing (SSA)-mediated DNA repair system.

RIM1alpha and RIM1beta are synthesized from distinct promoters of the RIM1 gene to mediate differential but overlapping synaptic functions

At a synapse, presynaptic terminals form a specialized area of the plasma membrane called the active zone that mediates neurotransmitter release. RIM1alpha is a multidomain protein that constitutes a central component of the active zone by binding to other active zone proteins such as Munc13 s, alpha-liprins, and ELKS, and to synaptic vesicle proteins such as Rab3 and synaptotagmin-1. In mice, knockout of RIM1alpha significantly impairs synaptic vesicle priming and presynaptic long-term plasticity, but is not lethal. We now find that the RIM1 gene encodes a second, previously unknown RIM1 isoform called RIM1beta that is upregulated in RIM1alpha knock-out mice. RIM1beta is identical to RIM1alpha except for the N terminus where RIM1beta lacks the N-terminal Rab3-binding sequence of RIM1alpha. Using newly generated knock-out mice lacking both RIM1alpha and RIM1beta, we demonstrate that different from the deletion of only RIM1alpha, deletion of both RIM1alpha and RIM1beta severely impairs mouse survival. Electrophysiological analyses show that the RIM1alphabeta deletion abolishes long-term presynaptic plasticity, as does RIM1alpha deletion alone. In contrast, the impairment in synaptic strength and short-term synaptic plasticity that is caused by the RIM1alpha deletion is aggravated by the deletion of both RIM1alpha and RIM1beta. Thus, our data indicate that the RIM1 gene encodes two different isoforms that perform overlapping but distinct functions in neurotransmitter release.

Mechanisms of double-strand-break repair during gene targeting in mammalian cells

In the present study, the mechanism of double-strand-break (DSB) repair during gene targeting at the chromosomal immunoglobulin mu-locus in a murine hybridoma was examined. The gene-targeting assay utilized specially designed insertion vectors genetically marked in the region of homology to the chromosomal mu-locus by six diagnostic restriction enzyme site markers. The restriction enzyme markers permitted the contribution of vector-borne and chromosomal mu-sequences in the recombinant product to be determined. The use of the insertion vectors in conjunction with a plating procedure in which individual integrative homologous recombination events were retained for analysis revealed several important features about the mammalian DSB repair process:The presence of the markers within the region of shared homology did not affect the efficiency of gene targeting. In the majority of recombinants, the vector-borne marker proximal to the DSB was absent, being replaced with the corresponding chromosomal restriction enzyme site. This result is consistent with either formation and repair of a vector-borne gap or an "end" bias in mismatch repair of heteroduplex DNA (hDNA) that favored the chromosomal sequence. Formation of hDNA was frequently associated with gene targeting and, in most cases, began approximately 645 bp from the DSB and could encompass a distance of at least 1469 bp. The hDNA was efficiently repaired prior to DNA replication. The repair of adjacent mismatches in hDNA occurred predominantly on the same strand, suggesting the involvement of a long-patch repair mechanism.

High efficiency site-specific modification of the chromosomal immunoglobulin locus by gene targeting

Site-specific modification of the chromosomal immunoglobulin locus by gene targeting is a powerful tool in studying the molecular requirements for immunoglobulin gene structure and function and in the production of engineered antibodies. Here, we describe a two step- integration then excision-gene targeting procedure for introducing planned genetic alterations into the chromosomal immunoglobulin locus. The efficiency of gene targeting with an enhancer-trap vector in which an enhancerless neo and HSV-tk gene were inserted into the vector backbone was compared to that of the corresponding enhancer-positive vector. Both insertion vectors also contained homology to the chromosomal immunoglobulin target locus along with the desired genetic alteration. The first step involved insertion of the transferred vector into the target locus by homologous recombination. An approximately 15-fold enrichment in the frequency of vector insertion was obtained with the enhancer-trap compared to the enhancer-positive vector. The majority of targeted cells (75%) contained a single copy of the vector integrated into the chromosomal immunoglobulin locus. The second step involved excision of the integrated vector by intrachromosomal homologous recombination between the duplicated region of homology that removed the integrated vector, neo and tk genes along with one copy of homologous DNA. Vector excision was very efficient generating G418S, FIAU(R) secondary recombinants at the high rate of approximately 10(-3)/cell generation. In the secondary recombinants, the overall structure of the chromosomal immunoglobulin locus was restored with the desired genetic alteration being present in an expected proportion of the cells.

I-SceI-induced gene replacement at a natural locus in embryonic stem cells

Gene targeting is a very powerful tool for studying mammalian development and physiology and for creating models of human diseases. In many instances, however, it is desirable to study different modifications of a target gene, but this is limited by the generally low frequency of homologous recombination in mammalian cells. We have developed a novel gene-targeting strategy in mouse embryonic stem cells that is based on the induction of endogenous gap repair processes at a defined location within the genome by induction of a double-strand break (DSB) in the gene to be mutated. This strategy was used to knock in an NH2-ezrin mutant in the villin gene, which encodes an actin-binding protein expressed in the brush border of the intestine and the kidney. To induce the DSB, an I-SceI yeast meganuclease restriction site was first introduced by gene targeting to the villin gene, followed by transient expression of I-SceI. The repair of the ensuing DSB was achieved with high efficiency (6 x 10[-6]) by a repair shuttle vector sharing only a 2.8-kb region of homology with the villin gene and no negative selection marker. Compared to conventional gene-targeting experiments at the villin locus, this represents a 100-fold stimulation of gene-targeting frequency, notwithstanding a much lower length of homology. This strategy will be very helpful in facilitating the targeted introduction of several types of mutations within a gene of interest.

Biased short tract repair of palindromic loop mismatches in mammalian cells

Mismatch repair of palindromic loops in the presence or absence of single-base mismatches was investigated in wild-type and mismatch-binding defective mutant Chinese hamster ovary cells. Recombination intermediates with a maximum heteroduplex DNA (hDNA) region of 697 bp contained a centrally located, phenotypically silent 12-base palindromic loop mismatch, and/or five single-base mismatches. In wild-type cells, both loops and single-base mismatches were efficiently repaired (80-100%). When no other mismatches were present in hDNA, loops were retained with a 1.6-1.9:1 bias. However, this bias was eliminated when single-base mismatches were present, perhaps because single-base mismatches signal nick-directed repair. In the multiple marker crosses, most repair tracts were long and continuous, with preferential loss of markers in cis to proximal nicks, consistent with nicks directing most repair in this situation. However, approximately 25% of repair tracts were discontinuous as a result of loop-specific repair, or from segregation or short tract repair of single-base mismatches. In mutant cells, single-base mismatches were repaired less frequently, but the loop was still repaired efficiently and with bias toward loop retention, indicating that the defect in these cells does not affect loop-specific repair. Repair tracts in products from mutant cells showed a wide variety of mosaic patterns reflecting short regions of repair and segregation consistent with reduced nick-directed repair. In mutant cells, single-base mismatches were repaired more efficiently in the presence of the loop than in its absence, a likely consequence of corepair initiated at the loop.

Mismatch repair by efficient nick-directed, and less efficient mismatch-specific, mechanisms in homologous recombination intermediates in Chinese hamster ovary cells

Repair of single-base mismatches formed in recombination intermediates in vivo was investigated in Chinese hamster ovary cells. Extrachromosomal recombination was stimulated by double-strand breaks (DSBs) introduced into regions of shared homology in pairs of plasmid substrates heteroallelic at 11 phenotypically silent mutations. Recombination was expected to occur primarily by single-strand annealing, yielding predicted heteroduplex DNA (hDNA) regions with three to nine mismatches. Product spectra were consistent with hDNA only occurring between DSBs. Nicks were predicted on opposite strands flanking hDNA at positions corresponding to original DSB sites. Most products had continuous marker patterns, and observed conversion gradients closely matched predicted gradients for repair initiated at nicks, consistent with an efficient nick-directed, excision-based mismatch repair system. Discontinuous patterns, seen in approximately 10% of products, and deviations from predicted gradients provided evidence for less efficient mismatch-specific repair, including G-A-->G-C specific repair that may reflect processing by a homologue of Escherichia coli MutY. Mismatch repair was > 80% efficient, which is higher than seen previously with covalently closed, artificial hDNA substrates. Products were found in which all mismatches were repaired in a single tract initiated from one or the other nick. We also observed products resulting from two tracts of intermediate length initiated from two nicks.

Double-strand breaks at the target locus stimulate gene targeting in embryonic stem cells

Double-strand breaks (DSBs) are recombinogenic lesions in chromosomal DNA in yeast, Drosophila and Caenorhabditis elegans. Recent studies in mammalian cells utilizing the I-Scel endonuclease have demonstrated that in some immortalized cell lines DSBs in chromosomal DNA are also recombinogenic. We have now tested embryonic stem (ES) cells, a non-transformed mouse cell line frequently used in gene targeting studies. We find that a DSB introduced by I-Scel stimulates gene targeting at a selectable neo locus at least 50-fold. The enhanced level of targeting is achieved by transient expression of the I-Scel endonuclease. In 97% of targeted clones a single base pair polymorphism in the transfected homologous fragment was incorporated into the target locus. Analysis of the targeted locus demonstrated that most of the homologous recombination events were 'two-sided', in contrast to previous studies in 3T3 cells in which 'one-sided' homologous events predominated. Thus ES cells may be more faithful in incorporating homologous fragments into their genome than other cells in culture.

More to learn from gene knockouts

Gene targeting by homologous recombination in mouse embryonic stem cells is a powerful technique to determine the physiological function of any gene product in embryonic and postnatal development and in molecular pathogenesis. Although the technique is very demanding and still in its developing stage several knockout mice carrying disrupted genes, which were once thought important for the development or molecular pathogenesis of certain tissues, have given unexpected results. A gene/function redundancy or superfluous and on-functional theory has been advanced by many investigators to explain the unexpected results. These surprising results may teach us a new lesson and lead to a revision of the strongly held view that highly conserved and abundantly expressed genes have a prominent role and function in cell physiology and development. Additional, they may also support the notion that molecular cross-talk among the genes may play an important role in determining the minimal phenotype.

Use of double-replacement gene targeting to replace the murine alpha-lactalbumin gene with its human counterpart in embryonic stem cells and mice

The mouse alpha-lactalbumin gene has been replaced with the human gene by two consecutive rounds of gene targeting in hypoxanthine phosphoribosyltransferase (HPRT)-deficient feeder-independent murine embryonic stem (ES) cells. One mouse alpha-lactalbumin allele was first replaced by an HPRT minigene which was in turn replaced by human alpha-lactalbumin. The end result is a clean exchange of defined DNA fragments with no other DNA remaining at the target locus. Targeted ES cells at each stage remained capable of contributing efficiently to the germ line of chimeric animals. Double replacement using HPRT-deficient ES cells and the HPRT selection system is therefore a powerful and flexible method of targeting specific alterations to animal genes. A typical strategy for future use would be to generate a null mutation which could then be used to produce multiple second-step alterations at the same locus.

Use of double-replacement gene targeting to replace the murine alpha-lactalbumin gene with its human counterpart in embryonic stem cells and mice

The mouse alpha-lactalbumin gene has been replaced with the human gene by two consecutive rounds of gene targeting in hypoxanthine phosphoribosyltransferase (HPRT)-deficient feeder-independent murine embryonic stem (ES) cells. One mouse alpha-lactalbumin allele was first replaced by an HPRT minigene which was in turn replaced by human alpha-lactalbumin. The end result is a clean exchange of defined DNA fragments with no other DNA remaining at the target locus. Targeted ES cells at each stage remained capable of contributing efficiently to the germ line of chimeric animals. Double replacement using HPRT-deficient ES cells and the HPRT selection system is therefore a powerful and flexible method of targeting specific alterations to animal genes. A typical strategy for future use would be to generate a null mutation which could then be used to produce multiple second-step alterations at the same locus.

Gene targeting in normal somatic cells: inactivation of the interferon-gamma receptor in myoblasts

Gene targeting in somatic cells represents a potentially powerful method for gene therapy, yet with the exception of pluripotent mouse embryonic stem (ES) cells, homologous recombination has not been reported for a well characterized, non-transformed mammalian cell. Applying a highly efficient strategy for targeting an integral membrane protein--the interferon gamma receptor--in ES cells, we have used homologous recombination to target a non-transformed somatic cell, the mouse myoblast, and to compare targeting efficiencies in these two cell types. Gene-targeted myoblasts display the properties of normal cells including normal morphology, ability to differentiate in vitro, stable diploid karyotype, inability to form colonies in soft agar and lack of tumorigenicity in nude mice.

Genetics of eukaryotic RNA polymerases I, II, and III

The transcription of nucleus-encoded genes in eukaryotes is performed by three distinct RNA polymerases termed I, II, and III, each of which is a complex enzyme composed of more than 10 subunits. The isolation of genes encoding subunits of eukaryotic RNA polymerases from a wide spectrum of organisms has confirmed previous biochemical and immunological data indicating that all three enzymes are closely related in structures that have been conserved in evolution. Each RNA polymerase is an enzyme complex composed of two large subunits that are homologous to the two largest subunits of prokaryotic RNA polymerases and are associated with smaller polypeptides, some of which are common to two or to all three eukaryotic enzymes. This remarkable conservation of structure most probably underlies a conservation of function and emphasizes the likelihood that information gained from the study of RNA polymerases from one organism will be applicable to others. The recent isolation of many mutations affecting the structure and/or function of eukaryotic and prokaryotic RNA polymerases now makes it feasible to begin integrating genetic and biochemical information from various species in order to develop a picture of these enzymes. The picture of eukaryotic RNA polymerases depicted in this article emphasizes the role(s) of different polypeptide regions in interaction with other subunits, cofactors, substrates, inhibitors, or accessory transcription factors, as well as the requirement for these interactions in transcription initiation, elongation, pausing, termination, and/or enzyme assembly. Most mutations described here have been isolated in eukaryotic organisms that have well-developed experimental genetic systems as well as amenable biochemistry, such as Saccharomyces cerevisiae, Drosophila melanogaster, and Caenorhabditis elegans. When relevant, mutations affecting regions of Escherichia coli RNA polymerase that are conserved among eukaryotes and prokaryotes are also presented. In addition to providing information about the structure and function of eukaryotic RNA polymerases, the study of mutations and of the pleiotropic phenotypes they imposed has underscored the central role played by these enzymes in many fundamental processes such as development and cellular differentiation.

Disruption of largest subunit RNA polymerase II genes in Trypanosoma brucei

Two types of largest subunit RNA polymerase II (pol II) genes (pol IIA and pol IIB), differing in 3 amino acid substitutions, are encoded in the Trypanosoma brucei (stock 427-60) genome. As a result, the alpha-amanitin-resistant transcription of the procyclic acidic repetitive protein (PARP) and variant surface glycoprotein (VSG) genes was proposed to involve a modified, alpha-amanitin-resistant form of the largest subunit of pol II. Alternatively, pol I could transcribe the PARP and VSG genes. To discriminate between these two models, we deleted the N-terminal domain (about one-third of the polypeptide), which encodes the amino acid substitutions which discriminated the pol IIA and pol IIB genes, at both pol IIB alleles. The pol IIB- trypanosomes still transcribe the PARP genes and the VSG gene promoter region in insect-form trypanosomes by alpha-amanitin-resistant RNA polymerases, while control housekeeping genes are transcribed in an alpha-amanitin-sensitive manner, presumably by pol IIA. We conclude that the alpha-amanitin-resistant transcription of protein coding genes in T. brucei is not mediated by a diverged form of the largest subunit of pol II and that the presence of both the pol IIA and pol IIB genes is not essential for trypanosome viability. This conclusion was further supported by the finding that individual trypanosome variants exhibited allelic heterogeneity for the previously identified amino acid substitutions and that various permutations of the polymorphic amino acids generate at least four different types of largest subunit pol II genes. The expression of the PARP genes and the VSG gene promoter region by alpha-amanitin-resistant RNA polymerases in the pol IIB- trypanosomes provides evidence for transcription of these genes by pol I.

Disruption of largest subunit RNA polymerase II genes in Trypanosoma brucei

Two types of largest subunit RNA polymerase II (pol II) genes (pol IIA and pol IIB), differing in 3 amino acid substitutions, are encoded in the Trypanosoma brucei (stock 427-60) genome. As a result, the alpha-amanitin-resistant transcription of the procyclic acidic repetitive protein (PARP) and variant surface glycoprotein (VSG) genes was proposed to involve a modified, alpha-amanitin-resistant form of the largest subunit of pol II. Alternatively, pol I could transcribe the PARP and VSG genes. To discriminate between these two models, we deleted the N-terminal domain (about one-third of the polypeptide), which encodes the amino acid substitutions which discriminated the pol IIA and pol IIB genes, at both pol IIB alleles. The pol IIB- trypanosomes still transcribe the PARP genes and the VSG gene promoter region in insect-form trypanosomes by alpha-amanitin-resistant RNA polymerases, while control housekeeping genes are transcribed in an alpha-amanitin-sensitive manner, presumably by pol IIA. We conclude that the alpha-amanitin-resistant transcription of protein coding genes in T. brucei is not mediated by a diverged form of the largest subunit of pol II and that the presence of both the pol IIA and pol IIB genes is not essential for trypanosome viability. This conclusion was further supported by the finding that individual trypanosome variants exhibited allelic heterogeneity for the previously identified amino acid substitutions and that various permutations of the polymorphic amino acids generate at least four different types of largest subunit pol II genes. The expression of the PARP genes and the VSG gene promoter region by alpha-amanitin-resistant RNA polymerases in the pol IIB- trypanosomes provides evidence for transcription of these genes by pol I.

Detection of gene targeting by co-conversion of a single nucleotide change during replacement recombination at the immunoglobulin mu heavy chain locus

A method is described for detecting targeted events at the mu heavy chain gene which relies on co-conversion (or co-exchange) of a point mutation with a selectable marker contained on a replacement vector. The vector, designed for application to IgM producing hybridomas, contains a single nucleotide change within the region of homology with the target gene which encodes a different allotypic determinant of IgM. In a model system where homologous recombination corrected a defective mu gene, the length of homology between this nucleotide change and the position of the double strand break in the vector was found to have a critical influence on the co-conversion frequency. In the vector design ultimately used for targeting in hybridomas, one in 1000-2000 stable transformants produced IgM with the allotype encoded by the exogenous DNA, and Southern blot analysis confirmed that these were derived by targeted integration. The sensitivity of the screening procedure using a monoclonal antibody specific to this allotype enabled a targeted clone to be detected in a pool of stable transformants when present at a frequency at least as low as one per cent. Several different modifications of the target locus were obtained as a consequence of alternative crossover positions and, in some cases, vector DNA concatenation.

Fidelity of targeted recombination in human fibroblasts and murine embryonic stem cells

Targeted recombination in murine embryonic stem cells promises to be a powerful tool for introducing specific mutations into target genes to study development in mice and to create animal models of human disease. Gene targeting also holds potential for correcting genetic defects as an approach to human gene therapy. To precisely modify target genes, homologous recombination must proceed with high fidelity. However, several results have suggested that targeted recombination may be highly mutagenic. To test the accuracy of gene targeting we analyzed 44 independent targeted recombinants at the hypoxanthine phosphoribosyltransferase (HPRT) locus in a human fibroblast cell line and in mouse embryonic stem cells. We surveyed 80 kilobases around the sites of recombination by using chemical cleavage of mismatches. Only two mutations were found: a T----G transversion and a thymidine deletion. Thus, gene targeting in mammalian cells can be extremely accurate. These results demonstrate the feasibility of generating precise modifications of mammalian genomes by gene targeting.

Testing an "in-out" targeting procedure for making subtle genomic modifications in mouse embryonic stem cells

We have introduced a 4-bp insertion into the hypoxanthine phosphoribosyltransferase (HPRT) gene of a mouse embryonic stem (ES) cell line by using an "in-out" targeting procedure. During the in step, a homologous integration reaction, we targeted a correcting plasmid to a partially deleted hprt- locus by using an integrating vector that carried a 4-bp insertion in the region of DNA homologous to the target locus. HPRT+ recombinants were isolated by direct selection in hypoxanthine-aminopterin-thymidine (HAT) medium. The HATr cell lines were then grown in medium containing 6-thioguanine (6-TG) to select for hprt- revertants resulting from the excision of the integrated vector sequences. The revertants were examined by Southern blot hybridization to determine the accuracy of this out reaction and the frequency of retaining the 4-bp modification in the genome. Of the 6-TGr colonies examined, 88% had accurately excised the integrated vector sequences; 19 of 20 accurate revertants retained the 4-bp insertion in the resulting hprt- gene. We suggest a scheme for making the in-out targeting procedure generally useful to modify the mammalian genome.

Testing an "in-out" targeting procedure for making subtle genomic modifications in mouse embryonic stem cells

We have introduced a 4-bp insertion into the hypoxanthine phosphoribosyltransferase (HPRT) gene of a mouse embryonic stem (ES) cell line by using an "in-out" targeting procedure. During the in step, a homologous integration reaction, we targeted a correcting plasmid to a partially deleted hprt- locus by using an integrating vector that carried a 4-bp insertion in the region of DNA homologous to the target locus. HPRT+ recombinants were isolated by direct selection in hypoxanthine-aminopterin-thymidine (HAT) medium. The HATr cell lines were then grown in medium containing 6-thioguanine (6-TG) to select for hprt- revertants resulting from the excision of the integrated vector sequences. The revertants were examined by Southern blot hybridization to determine the accuracy of this out reaction and the frequency of retaining the 4-bp modification in the genome. Of the 6-TGr colonies examined, 88% had accurately excised the integrated vector sequences; 19 of 20 accurate revertants retained the 4-bp insertion in the resulting hprt- gene. We suggest a scheme for making the in-out targeting procedure generally useful to modify the mammalian genome.

Characterization of nonconservative homologous junctions in mammalian cells

Homologous recombination in mammalian cells between extrachromosomal molecules, as well as between episomes and chromosomes, can be mediated by a nonconservative mechanism. It has been proposed that the key steps in this process are the generation (by double-strand cleavage) of overlapping homologous ends, the creation of complementary single-strand ends (either by strand-specific exonuclease degradation or by unwinding of the DNA helix), and finally the creation of heteroduplex DNA by the annealing of the single-strand ends. We have analyzed in detail the structure of nonconservative homologous junctions and determined the contribution of each end to the formation of the junction. We have also analyzed multiple descendants from single recombination events. Two types of junctions were found. The majority (90%) of the junctions were characterized by a single crossover site. These crossover sites were distributed randomly throughout the junction. The remaining 10% of the junctions had mosaic patterns of parental markers. Furthermore, in 9 of 10 cases, multiple descendants from a single recombination event were identical. Thus, it appears that in most cases few parental markers were involved in junction formation. This finding suggests that nonconservative homologous junctions are mediated mainly by short heteroduplexes of a few hundred base pairs or less. These results are discussed in terms of the current models of nonconservative homologous recombination.

Characterization of nonconservative homologous junctions in mammalian cells

Homologous recombination in mammalian cells between extrachromosomal molecules, as well as between episomes and chromosomes, can be mediated by a nonconservative mechanism. It has been proposed that the key steps in this process are the generation (by double-strand cleavage) of overlapping homologous ends, the creation of complementary single-strand ends (either by strand-specific exonuclease degradation or by unwinding of the DNA helix), and finally the creation of heteroduplex DNA by the annealing of the single-strand ends. We have analyzed in detail the structure of nonconservative homologous junctions and determined the contribution of each end to the formation of the junction. We have also analyzed multiple descendants from single recombination events. Two types of junctions were found. The majority (90%) of the junctions were characterized by a single crossover site. These crossover sites were distributed randomly throughout the junction. The remaining 10% of the junctions had mosaic patterns of parental markers. Furthermore, in 9 of 10 cases, multiple descendants from a single recombination event were identical. Thus, it appears that in most cases few parental markers were involved in junction formation. This finding suggests that nonconservative homologous junctions are mediated mainly by short heteroduplexes of a few hundred base pairs or less. These results are discussed in terms of the current models of nonconservative homologous recombination.