![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Cell Biology and Neuroscience and Environmental Toxicology Graduate Program, University of California, Riverside, California 92521,1 Division of Genetics and Mutagenesis, National Institute of Health Sciences, Setagaya, Tokyo, Japan2
Received 30 September 2006/ Returned for modification 2 November 2006/ Accepted 23 April 2007
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Several mechanisms have been proposed to explain the preference for gene conversion without associated crossover in eukaryotic organisms. Crossover is believed to require the formation of Holliday intermediates (17, 36, 44, 76, 77) and their resolution through structure-specific endonuclease scissions (79). Theoretically, half of Holliday junctions that are resolved through scissions would be cut in a configuration that results in crossover (36). However, crossovers can be precluded by well-described recombinational pathways that do not meet these requirements. For example, recombinational exchanges may occur through synthesis-dependent strand annealing (24, 63, 65), in which strand invasion intermediates are removed following repair synthesis but before Holliday junctions can form, thus partially accounting for the absence of associated crossovers. Once formed, Holiday junctions can be resolved without endonuclease scissions, through the unwinding action of a complex involving topoisomerase III and a RecQ family helicase such as BLM (95). A third possibility is that endonuclease resolution inherently favors scissions in a configuration leading to noncrossovers (18). Furthermore, crossovers may require steps that can be suppressed, such as structural isomerization (80). Branch migration, which may facilitate isomerization, also presents a target for regulation (58, 81).
In yeast, the paucity of mitotic interchromosomal crossovers is correlated with the prevalent occurrence of short conversion tracts, suggesting a mechanistic link. One early study reported that specific selection for long, complex conversion tracts resulted in a very high crossover association (73). Furthermore, crossovers are disproportionately associated with relatively rare long tracts in intrachromosomal or plasmid-based systems (1, 2), and a reduction in the overall length of available homology has been found to reduce interchromosomal crossovers to a much greater extent than gene conversion (39, 41). Enzymatic functions which have differential effects on gene conversion and crossover have also been shown to influence conversion tract length (1, 10, 53, 55, 85), thus supporting the hypothesis that the occurrence of crossover is linked to tract length.
In this investigation we evaluated the hypothesis that tract length will also be one of the primary determinants of crossover association in mammalian cells. This issue has been addressed in an intrachromosomal recombination study (30), but no systematic analysis for interchromosomal recombination has been previously reported. In order to study the crossover association of interchromosomal gene conversion in human cells, we employed a haplotype linkage analysis (HLA) methodology (67). The HLA approach enabled linkage determinations within the selectable locus, as well as for flanking heterozygous loci within a study region of 1.9 megabases. For this investigation we developed a pair of isogenic cells lines, each of which contains an I-SceI endonuclease recognition sequence within the endogenous thymidine kinase (tk) locus. In cell line TI-112, where selectable conversion tracts must be a minimum of 7.0 kb in length and complex in tract structure, the crossover association was 48%. In contrast, cell line TSCER2 (37) permits a wider range of selectable conversion tract lengths, including continuous tracts as short as 98 bp; the crossover association in this system was 16%. In some cases, the HLA system also permitted the identification of unselected conversion tracts that were cosegregated into convertant clones with selected conversion tracts. These unselected conversion tracts were often long, even though length was not a prerequisite for their recovery, thus suggesting that tracts of 7 kb and longer can frequently occur in interchromosomal recombination. These results demonstrate that even in the absence of selection, crossovers are particularly associated with long conversion tracts in human cells. The observation of crossovers in the present investigation, which is in contrast to a previous study of this issue (83), is due to the robust ability of the experimental system to recover and fully analyze long and complex conversion tracts.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Cell lines TSCER2 and TI-112 are isogenic tk–/– derivatives of TK6. The structures of the tk locus in the two cell lines are shown in Fig. 1, with the transcriptional direction towards the centromere as shown by human genome sequencing (46). TK6 is mutated in exon 4 of allele B of the tk gene, with a single-base C frameshift (FS) insertion in a run of three Cs at position 4864 of the genomic DNA sequence. There is a phenotypically silent FS insertion on allele A in a run of four Gs at position 12690 in exon 7 (32). TSCER2 was generated as described previously (37). In brief, a two-step gene targeting was performed to introduce an I-SceI site into intron 4 of TK6 to create cell line TSCE5. The 31-bp I-SceI recognition sequence and linker DNA are located in a BglII site at positions 11815 to 11820. Cell lines TSCER2 and TI-112 are TfT-resistant spontaneous tk–/– mutants of TSCE5. TSCER2 is mutated in position 11923, with a G-to-A transition (TS) resulting in a change of codon from a TGC (cysteine) to TAC (tyrosine). TI-112, newly collected for this study, has a G-to-A TS at position 574 resulting in a codon change of GGG (glycine) to AGG (arginine). Cell lines were maintained in RPMI 1640 with 10% iron-supplemented calf serum, 1% L-glutamine, and 1% penicillin-streptomycin mix and cultured at 37°C with 5% CO2, with the cell density kept between 2 x 105 and 10 x 105 cells/ml.
|
Collection of HLA clones. The collection of clones for HLA was performed as previously described (67). In brief, for each convertant clone, a small set of spontaneous TfT-resistant mutants was collected and analyzed for LOH at the flanking microsatellite loci in order to identify at least one clone that was coordinately LOH at tk-flanking microsatellite markers D17S937, D17S802, and D17S836. This convertant was used for all subsequent haplotype analyses.
Genomic analysis of TK convertant and derivative haplotype clones. DNA purification and PCR of tk intragenic and flanking microsatellite markers was performed as previously described (28, 32, 67). The PCR products containing the intragenic tk markers were sequenced by capillary electrophoresis (ABI) and visualized using standard software (Chromas). Primers were obtained from Invitrogen Inc.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Using these cell lines and HLA (see below), convertants could be categorized as no crossover (No-CO) if LOH or linkage switches were localized to intragenic markers. A linkage switch that occurs in the absence of a crossover of flanking markers is evidence of either double-crossover resolution of flanking Holliday junctions or independent resolution of mismatches within symmetric heteroduplex DNA. In addition, all three possible single-crossover products could be unambiguously distinguished through flanking marker analysis (Fig. 2): crossover with no flanking marker LOH (CO-No-LOH) or crossover with homozygosity for telomeric microsatellite polymorphisms linked to the "A chromosome" (CO-LOH-A) or the "B chromosome" (CO-LOH-B). The A chromosome contains the I-SceI cut site and is therefore the cut chromosome, while the B chromosome is the donor chromosome. Crossovers that lead to flanking marker LOH are expected to represent half of such events, whereas the remaining half of crossovers do not produce LOH (Fig. 2). Typically, this can lead to underrepresentation of crossover estimates, although the HLA system minimizes the likelihood of a significant underestimate. Still, due to the requirements of producing a selectable outcome, some products may be underrepresented, so the crossover measurements reported here should be regarded as minimal.
|
|
|
|
|
|
|
A summary of the conversion tract lengths by crossover classification is presented in Fig. 4B, and allelotypes for individual convertants are provided in Fig. 6. Most of these crossovers were determined to be associated with LOH, and in particular CO-LOH-A (Fig. 4 and Fig. 6f to n), accounting for 0.13 of all convertants (Table 1). Random mitotic segregation should lead to an equal number of crossovers without accompanying LOH (CO-No-LOH), but only 1 CO-No-LOH convertant was observed (Fig. 4 and 6e), compared to 12 CO-LOH-A convertants (Fig. 4A and 6f to n). CO-LOH-B convertants (Fig. 6o), which provide a measurement of long and discontinuous conversion tracts in TSCER2, constituted less than 0.01 of the overall spectrum (Table 1).
CO-LOH-A convertants frequently contain two independent cosegregated I-SceI-induced conversion tracts. The I-SceI site on tk allele A (Fig. 1) is expected to be lost during gene conversion, since extensive heterology is usually removed from the invading ends at the site of a recombinogenic break. However, cosegregation of a second, noncrossover copy of tk allele A in the CO-LOH-A class of convertants (Fig. 2) should result in the recovery of an intact I-SceI site unless this I-SceI site is also cut. Of 19 CO-LOH-A convertants (12 TSCER2 and 7 TI-112), only 1 (Fig. 5p) retained an I-SceI site, whereas 18 other CO-LOH-A convertants from TI-112 (Fig. 5q to t) and TSCER2 (Fig. 6f to n) had lost the I-SceI site on the noncrossover allele. These results indicate that the I-SceI site on the cosegregating chromosome is often cut at some point in the process.
HLA provides a methodology to directly determine the status of the I-SceI site on the noncrossover allele as well as to map a second, independent conversion tract that may result from cleavage of this site (Fig. 7). Selectable CO-LOH-A convertants can be wild type on the crossover allele or the noncrossover allele, which can be discriminated by separate HLA outcomes (Fig. 7, classes 2 and 3); class 4, in which convertants are wild type on both alleles, can be distinguished during the initial genotyping of the parental convertant. HLA would also permit the identification of nonrecombinational DSB repair (DSBR) products. However, in all cases the I-SceI insert was lost without any deletion of flanking sequences, indicating that the cosegregating DSBR product was always a gene conversion rather than a small intragenic deletion. Additionally, in all but three cases (Fig. 5q) additional tk polymorphisms were coconverted, thus further confirming the classification of these cosegregating alleles as recombinational products.
Unconstrained gene conversion tracts are often long. Since only one of the two alleles must be wild type for a convertant to be selected as CHAT resistant (Fig. 7), it was possible to recover and characterize cosegregated conversion tracts which were not constrained by the requirements of selection. Two separate conversion tracts were fully characterized in 16 CO-LOH-A clones derived from either TSCER2 or TI-112 (Table 2). In 14 convertants (Fig. 5q to t and 6f, g, and j to m), one of the two tk alleles is wild type and therefore necessary for growth in selective conditions; the conversion tract on the cosegregating allele, which retains an inactivating mutation, was thus unconstrained. Two cases (Fig. 6 h and i) involve cosegregating wild-type alleles. Both conversion tracts in each of these two cases are considered unconstrained for the purposes of this analysis.
|
Length of intragenic conversion tracts in TI-112. In addition to the unambiguous identification of crossovers associated with convertants, HLA permits the detailed analysis of polymorphic markers within the intragenic conversion tract, enabling the compilation of information regarding the length and complexity of the tracts. This system is not restricted to analysis of tracts on only the cut chromosome; tracts occurring on the donor chromosome are also distinguishable. A full representation of both tracts for all TI-112 clones (Fig. 5) and a summary of the distribution of conversion tract lengths from TI-112 convertants (Fig. 4A) are provided. In the summary, for simplicity, only one tract is presented for each clone. In some convertants (Fig. 5n), the length of the conversion tract is inferred from the location of the crossover relative to the DSB site, since intervening markers remain unaltered. In each case, the longest tract is shown, in order to demonstrate the frequent association between long tracts and crossover. CO-LOH-A clones are an exception, since only one of the tracts is directly associated with the crossover, whereas the other tract occurred independently of the crossover and cosegregated into the convertant. Thus, the crossover-associated tract was selected for display in the summary analysis (Fig. 4).
In TI-112, conversion tracts were all long (Fig. 4A) except in the case of some unconstrained tracts (Table 2; Fig. 5q and r). All remaining conversion tracts were a minimum of 7.0 kb long and as long as 12.1 kb. These are minimal estimates based on the position of the last known marker encompassed within the conversion tract. A large fraction (19/46; 41%) were observed to extend bidirectionally from the I-SceI site, and these occurred in both crossover and noncrossover clones (Fig. 4A).
Complexity of conversion tracts in TI-112. In cell line TI-112, only long and complex conversion tracts can be selected (Fig. 1B), although some unconstrained conversion tracts can also be recovered (Table 2). Selectable complex conversion tracts include a long and discontinuous conversion of the marker linked to the I-SceI site, 11.2 kb away, skipping the intervening exon 4 marker (Fig. 5a to f, k to m, o, s, u, and v). The second type is also long, with conversion of both the recipient I-SceI site and the donor exon 4 FS 7 kb from the cut site (Fig. 5g to i and p to r). A third type is a switch in linkage of either inactivating mutation, which can restore a wild-type tk allele on either chromosome (Fig. 5j, n, and t). The type of tract observed in any particular case is determined by whether the recombined A chromosome, the recombined B chromosome, or both recombined chromosomes segregated into the convertant.
By definition, only one of the two crossed-over chromosomes has segregated into a CO-LOH clone. This necessarily constrains conversion tract types in the CO-LOH-A and CO-LOH-B categories, although the selectable tracts are all complex. CO-LOH-B convertants are produced when the crossed-over A chromosome cosegregates with the non-crossed-over B chromosome (Fig. 2), which always showed the expected nonrecombinant parental haplotype (Fig. 5u to w). Therefore, by necessity all nine CO-LOH-B convertants show discontinuous conversion of the inactivating exon 1 TS on the A chromosome (Fig. 5u to w). Conversely, CO-LOH-A convertants are cosegregation products of the crossed-over B chromosome and the non-crossed-over A chromosome (Fig. 2). Therefore, with one exception (Table 2), selected CO-LOH-A tracts show opposite-direction conversion of the exon 4-inactivating FS on the B chromosome (Fig. 5p to t). Although the aforementioned CO-LOH tracts appear to be donor chromosome gene conversions, a crossover-associated linkage switch could produce the same outcome. The ability to distinguish these possibilities requires that both recombining chromosomes are available to be characterized, which is precluded by the segregation patterns in CO-LOH convertants (Fig. 2).
Both recombining chromosomes are cosegregated in CO-No-LOH convertants, by definition (Fig. 2). HLA is able to clearly identify this category of convertants, since the crossed-over flanking markers identify the presence of reciprocally recombined chromosomes in the parental convertant (Fig. 3). Analysis of the eight CO-No-LOH convertants recovered from TI-112 demonstrated that five were discontinuous conversions and the remaining three had undergone a crossover that included a function-restoring linkage switch of exon 1 (Fig. 5k to o; Table 3). On the other hand, while half of the 22 No-CO convertants (Fig. 5a to j; Table 3) will theoretically contain recombining chromosomes, they cannot be individually distinguished from nonrecombining cosegregants by flanking marker analysis. However, 17 of these No-CO convertants appear to have undergone conversion of exon 1 (Fig. 5a to f), whereas only 1 (Fig. 5j) showed a linkage switch that was localized to the marker in exon 1. This strong pattern suggests that among recoverable No-CO convertants, exon 1 is much more likely to have undergone a conversion than a localized switch, although we cannot rigorously confirm that conclusion for each individual case as we can for each CO-No-LOH convertant. The remaining four No-CO convertants (Fig. 5g to i; Table 3) showed an opposite conversion of the exon 4 FS. However, other No-CO convertants exhibited a localized linkage switch of exon 4 together with discontinuous conversion of exon 1 (Table 3; Fig. 5e and f). These results indicate that localized linkage switching and opposite conversion of the exon 4 marker both occur, although here again individual cases cannot be rigorously confirmed. Overall, with No-CO and CO-No-LOH convertants considered together, there were 22 gene conversions of exon 1 and 4 gene conversions of exon 4 (Table 3). This difference is statistically significant (Fisher's exact test, P = 0.01), suggesting that there is a mechanistic bias in favor of long conversion tracts that discontinuously convert the exon 1 TS (22/30). The preference for discontinuous conversion of exon 1 does not appear to be attributable to selection factors, since opposite-direction conversion of exon 4 was recovered in multiple cases.
|
|
Surprisingly, 7/95 long tracts encompassed the exon 4 FS (Fig. 4B), despite our expectation that this site could not be included within selectable conversion tracts. These seven long tracts included the single discontinuous CO-LOH-B convertant (Fig. 6o) and six crossover-associated tracts from CO-LOH-A convertants (Fig. 6j to m). The six CO-LOH-A tracts were unconstrained (Table 2) and were recoverable only when cosegregating with an independent, selectable conversion tract into a CO-LOH-A recombinant. A further surprise is that long unconstrained tracts occurred in at least half of TSCER2 crossovers; the remaining 5/12 tracts were short CO-No-LOH and CO-LOH-A tracts. Since crossover-associated tracts of any length were recoverable in TSCER2, the high fraction of long tracts in these 12 convertants suggests that long tracts are common among crossovers. This observation is consistent with other lines of evidence, as discussed above (Table 1), that suggest a significant relationship between tract length and the occurrence of crossover.
Rare recombinants retain I-SceI sequence linked to a converted marker. Three convertants were recovered which defied expectations that the I-SceI site should always be converted when linked to the conversion of adjacent markers. In two of these three convertants (Fig. 5o and 6d) the I-SceI insert was partially deleted. Both cases are consistent with end-joining repair of a DSB. In addition, one CO-LOH-B clone retains a complete I-SceI insert on the chromosome which is also crossed over (Fig. 5w). As a group these three convertants could represent spontaneous recombinants, which may have also been cut and repaired by nonhomologous end joining within the same clone. Alternatively, these events could be attributable to triparental recombination involving both a sister chromatid and a homologous chromosome (see Discussion).
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
LOH is the product of crossover. Homozygosity of multiple linked chromosomal loci that is not due to mitotic nondisjunction has been commonly used in mammalian cells as an indication of crossover (60, 91). However, since the associated gene conversions and alternate segregation products cannot be observed, the possibility that other, noncrossover mechanisms could explain observations of LOH has remained. We have reported here interallelic crossovers with and without LOH, as expected for alternative segregation of crossover chromatids in postreplicative cells. I-SceI-induced crossovers were always associated with conversion within the gene (Fig. 5 and 6) and often produced homozygosis of the tk-flanking microsatellite polymorphism linked to the DSB (Fig. 5p to t and 6f to n). This is not what would be expected for the products of an alternate mechanism for extensive homozygosity known as break-induced replication (BIR) (54). Those products that did show homozygosity for the flanking markers on the donor chromosome (CO-LOH-B) were also discontinuous from the site of the DSB (Fig. 5u to w and 6o); again, this is not what has been observed for BIR (45). In this pathway, a single broken end invades and initiates replication of whole chromosomal arms while the terminal fragment produced by the break is lost; BIR thus results in homozygosity for markers on the unbroken chromosome. Therefore, the results presented here confirm that LOH may commonly result from mitotic crossover. This does not exclude the possibility that I-SceI-induced DSBs could also produce the types of terminal multilocus LOH seen in tumor cells (15) by a BIR mechanism, since these would not be selectable outcomes in TSCER2 and TI-112.
Crossovers are associated with long conversion tracts. Among the few previous studies of interchromosomal gene conversion tracts and crossover in mammalian cells (29, 71, 72, 80, 83), crossovers have been rarely if ever observed. Some authors have therefore hypothesized that crossovers are suppressed in mammalian cells (83). We report here, however, that 16% of convertants in TSCER2 were associated with a crossover (Table 1), indicating that crossovers can be associated with conversion in mammalian cells at a level comparable to that in Saccharomyces cerevisiae (54, 64). For S. cerevisiae it has also been reported that a minimum of 1.7 kb of homology was required to recover DSB-induced crossovers (39). Those authors further proposed that this length represents the minimal region necessary to form a stable double Holliday junction structure. An analysis of our results (Fig. 4) also suggests a strong association of long tracts and crossovers. We observed here that 7/12 TSCER2 crossovers were associated with conversion tracts that were minimally 7.0 kb in length (Fig. 4), which is consistent with expectations that tract length and crossover would extend to mammalian recombination. These results are supported by observations of TI-112 (Fig. 4), in which long tracts are required for selection. Although there is no corresponding requirement for crossover, approximately half of long tracts in the TI-112 system are indeed associated with a crossover (Fig. 4). The results of our study may suggest that crossovers can occasionally occur in association with short conversion tracts (Fig. 4), which are necessarily defined by the inclusion of known polymorphic markers. However, in each case crossover-associated conversion tracts could extend significantly beyond the last polymorphism known to be included within the tract, which would then constitute a tract of at least several kilobases in length. The high fraction of multilocus LOH attributable to crossover in the forward mutant spectrum of TK6 lymphoblasts and their derivatives (47) is similar to that for other somatic cell types in vivo (33, 49, 61, 78, 92) and in vitro (19, 20, 43), suggesting that the crossovers observed in this study are not atypical.
Studies examining the crossover association of DSB-induced interchromosomal gene conversion in mammalian cells are quite limited. A strong relationship between conversion tract length and crossover might explain the absence of crossovers in systems that preclude or select against long conversion tracts. For instance, when homologous substrates were short, crossovers were unobserved (71). Only one other study has been performed to date that could potentially detect both CO-LOH and CO-No-LOH products in association with DSB-induced interallelic gene conversion. In this mouse embryonic stem (ES) cell study (83), crossovers were not observed and conversion tracts were overwhelmingly short; 158/162 were 1 kb long or less, and 145 of these were less than 0.3 kb in length (83). The reason that only a few long tract convertants were observed (83) may in part be due to the potential for coconversion of the closely placed (526 bp) inactivating mutation on the opposite allele. A similar marker configuration has been shown to significantly reduce the recovery of long tracts for interallelic gene conversion in yeast (64). In addition, the restriction fragment length polymorphism markers used for linkage analysis did not permit determination of crossover association for the few long conversion tracts that were recovered (83).
Long (3.2-kb) and one-sided gene conversion tracts were frequently observed in other ES cell studies involving short interchromosomal substrates (69) or a tandem repeat recombinational substrate designed to observe unequal sister chromatid recombination (42). Notably, long conversion tracts in these systems would necessarily extend into regions of nonhomology (42, 69) or through a region of interrupted homology (42). Restrictions on the formation of extensive heteroduplex DNA may thus account for the lack of crossovers in these investigations. Further insight might also be inferred from analysis of common tumor-associated chromosomal translocations, which map to the very long homologous regions (>10 kb to 400 kb) within the genomic features known as low-copy repeats and which are attributed to crossover (6, 8). In contrast translocations in somatic cells involving relatively short sequences such as Alu repeats are generated by single-strand annealing or nonhomologous end joining (22, 70) rather than crossover, and breakpoint analysis of many tumor-associated translocations supports these observations (93).
Recombination following endonuclease-induced sister chromatid breaks. The use of HLA enabled the observation that I-SceI sites from sister chromatids had each been cut and undergone separate conversion events (Fig. 7), at least in a subset of convertants. This would be expected to commonly occur in studies that utilize highly expressed endonucleases for generating DSBs, but it has rarely been possible to rigorously identify and characterize such events (42). Interestingly, 6/12 CO-LOH-A crossovers in TSCER2 (Fig. 6j to m) were selectable only due to the occurrence of a second conversion tract, since the crossover-associated conversion tract did not restore TK function (Fig. 7, class 3). Due to the configuration of the markers in the test locus, this pathway cannot lead to the recovery of CO-LOH-B or CO-No-LOH events. We therefore hypothesize that long continuous tracts are common but are recoverable only in the CO-LOH-A category, thus accounting for the bias in recovery of the CO-LOH-A category among TSCER2 crossovers (Table 1). It should be noted that CO-LOH-B tracts are all necessarily long and complex, suggesting that these occur less frequently than long and continuous tracts. These results also suggest that the TSCER2 system is efficiently scoring crossovers that occur, although there may still be an underestimate for nonfunctional crossover alleles which did not cosegregate with a separately converted allele that had been restored to function.
The higher recovery of crossovers in this study is unlikely to be attributable to the occurrence of two DSB, since other endonuclease-induced mammalian and yeast studies would have a similar circumstance. The difference here is the ability of HLA analysis to dissect out the independent conversion tracts arising from each break (Fig. 7). This analysis reveals no evidence that the occurrence of two breaks affected the nature of the resultant conversion tracts. In contrast, each tract generally appears to be independent of the other based on analysis of internal structure and length (Fig. 5 and 6). The likelihood of crossover association would also appear to be unaffected based on comparison of our results to those on endonuclease-induced recombination in other systems. For example, mammalian cell studies generally report few if any crossovers (42, 83), though the opportunity for two strand breaks is similar to the situation in the present investigation. In Drosophila melanogaster, the crossover fraction has surprisingly been reported to increase when the I-SceI endonuclease was expressed at a lower level (74).
A previous study of 38 I-SceI-induced TSCER2 convertants did not observe any crossovers as measured by multilocus LOH (37). This is in contrast to our observation that 12/95 (0.13) TSCER2 convertants could be classified as CO-LOH-A events (Fig. 4B). To understand the difference in these findings, it is important to appreciate that half of the CO-LOH-A convertants reported here required two conversion tracts to be selected (Fig. 6j to m). We hypothesize that the recovery of this group of crossovers in the earlier study (37) may be specifically diminished due to the utilization of an electroporation technology for transfection of I-SceI expression plasmid that was orders-of-magnitude less efficient (38; E. A. H. Neuwirth and A. J. Grosovsky, unpublished results). This efficiency difference may further be reflected in a 40-fold-higher conversion frequency (Table 1) compared to previous results (37). We thus suggest that the adjusted number of expected crossovers in the previous collection of 38 I-SceI convertants would be (0.13)(38)/2 = 2.47. The failure to observe any crossovers could be attributable to the smaller size of the collection and statistical fluctuations in recovering an anticipated 2.5 crossovers.
Unconstrained gene conversion tracts are often long. Two conversion events within the same recombinant clone made it possible to analyze a conversion tract that was not restricted by the requirements of selection. Significantly, unconstrained tracts were often long, at least 7.0 kb, and included both crossover-associated and non-crossover-associated tracts (Table 2). The data reported here represent the first mammalian cell analysis of interchromosomal homologous recombinational repair tract length in a nonselective context and demonstrate that conversion tracts extend over a much longer region than generally reported with short homologous substrates (21, 23, 87). As discussed above, in a study of mouse ES cell interallelic recombination (83), tracts constrained by selection to be unidirectional were primarily very short. Locus design constraints in TSCER2 also exist, since in the absence of cosegregation with a wild-type allele, coconversion of the exon 4 FS 7 kb from the cut site precludes the selection of long, noncrossover convertants (Fig. 1; Table 4). However, the observation of long and unconstrained tracts among noncrossover conversion tracts (Table 2) suggests that a significant portion of the selectable conversion tract spectrum was actually longer than could be established here due to the distance between available polymorphic markers (Fig. 1; Table 4). Additionally, these findings suggest that long tracts that coconvert exon 4 would represent a significant component of all recombination that occurs, even if they are not represented in the collection of convertants.
Evidence for triparental recombination in mammalian cells. Several convertants (Fig. 5o and w and 6d) which retain all or part of the I-SceI site on the recombined chromosome were recovered. This could suggest the recovery of spontaneous convertants in the collection, although this seems unlikely due to frequency considerations. The background interchromosomal conversion frequency in a closely related cell line is 3 x 10–8 (67; Neuwirth and Grosovsky, unpublished data), whereas the I-SceI-induced conversion frequencies are 4,000 x 10–8 in TSCER2 and 154 x 10–8 in TI-112 (Fig. 6). An alternative explanation is that these convertants are I-SceI induced and are products of a previously described triparental recombination mechanism (27). Under this model, the ends of one break separately invade the homolog and the sister chromatid. Recombination with the homolog eliminates the I-SceI insert but allows conversion of adjacent markers, whereas recombination with the sister restores the I-SceI site on the broken chromatid. The small set of convertants that retain the I-SceI site on the recombined chromosome are consistent with the predictions of this model and thus may be considered evidence for the occurrence of this pathway in mammalian cells.
Potential mechanisms for long tract recombination. The only recombinational model that can explain the full spectrum of observed convertant clones is the DSBR model for recombination (84, 86). Other models that allow for crossovers, such as a migrating D-loop model (24), are one sided and do not allow for convertants with complex alterations such as bidirectional tracts with donor allele alterations (see, for instance, Fig. 5c to f). There are several processes within the DSBR model which might be responsible for the very long tracts observed in TI-112 and TSCER2 convertants. Donor allele conversion in interallelic yeast systems was previously reported as evidence of Holliday junction branch migration and mismatch repair within symmetric heteroduplex DNA (57, 64) and similarly could explain the complex tract structures observed in TI-112 (Fig. 8a1; Tables 3 and 4) and TSCER2 (Fig. 6). In support of this model, symmetric heteroduplex DNA in mammalian cells has been reported in analysis of targeted gene insertion (9), occurring within crossover-associated conversion tracts of as long as 6 kb. If long symmetric heteroduplexes formed in TI-112, we would expect opposite-allele conversion of exon 4, which is relatively close to the I-SceI site, to occur more frequently than discontinuous conversion involving the more distant exon 1 (Fig. 1). However, we found that discontinuous conversion of exon 1 was four- to fivefold more common than donor allele conversion of exon 4 (Table 3).
|
Conclusions. The results presented here suggest that long gene conversion tracts are surprisingly frequent. Furthermore, crossovers were strongly associated with long tracts, even though shorter crossover-associated tracts were just as easily observable. Many of the long tracts observed here were also complex, which could potentially be an additional factor stimulating crossover. It has been suggested (95) that crossovers are suppressed in order to avoid the mutagenic potential associated with LOH or translocations and that crossover may be suppressed in human cells to a greater extent than in yeast (83). However, the findings presented here suggest that yeast and humans may be more comparable than previously known with respect to the regulation of crossover.
|
ACKNOWLEDGMENTS |
|---|
This work was supported by grant NAG2-1638 from the National Air and Space Administration. E.A.H.N. was also supported by a grant from the University of California, Toxic Substances Research and Teaching Program.
|
FOOTNOTES |
|---|
Published ahead of print on 21 May 2007.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2. Ahn, B. Y., and D. M. Livingston. 1986. Mitotic gene conversion lengths, coconversion patterns, and the incidence of reciprocal recombination in a Saccharomyces cerevisiae plasmid system. Mol. Cell. Biol. 6:3685-3693.
3. Aldosari, N., B. K. Rasheed, R. E. McLendon, H. S. Friedman, D. D. Bigner, and S. H. Bigner. 2000. Characterization of chromosome 17 abnormalities in medulloblastomas. Acta Neuropathol. (Berlin) 99:345-351.[CrossRef][Medline]
4. Amundson, S. A., and H. L. Liber. 1991. A comparison of induced mutation at homologous alleles of the tk locus in human cells. Mutat. Res. 247:19-27.[CrossRef][Medline]
5. Amundson, S. A., and H. L. Liber. 1992. A comparison of induced mutation at homologous alleles of the tk locus in human cells. II. Molecular analysis of mutants. Mutat. Res. 267:89-95.[Medline]
6. Barbouti, A., P. Stankiewicz, C. Nusbaum, C. Cuomo, A. Cook, M. Hoglund, B. Johansson, A. Hagemeijer, S. S. Park, F. Mitelman, J. R. Lupski, and T. Fioretos. 2004. The breakpoint region of the most common isochromosome, i(17q), in human neoplasia is characterized by a complex genomic architecture with large, palindromic, low-copy repeats. Am. J. Hum Genet. 74:1-10.[CrossRef][Medline]
7. Benjamin, M. B., and J. B. Little. 1992. X rays induce interallelic homologous recombination at the human thymidine kinase gene. Mol. Cell. Biol. 12:2730-2738.
8. Bi, W., S. S. Park, C. J. Shaw, M. A. Withers, P. I. Patel, and J. R. Lupski. 2003. Reciprocal crossovers and a positional preference for strand exchange in recombination events resulting in deletion or duplication of chromosome 17p11.2. Am. J. Hum. Genet. 73:1302-1315.[CrossRef][Medline]
9. Birmingham, E. C., S. A. Lee, R. D. McCulloch, and M. D. Baker. 2004. Testing predictions of the double-strand break repair model relating to crossing over in mammalian cells. Genetics 168:1539-1555.
10. Blanton, H. L., S. J. Radford, S. McMahan, H. M. Kearney, J. G. Ibrahim, and J. Sekelsky. 2005. REC, Drosophila MCM8, drives formation of meiotic crossovers. PLoS Genet. 1:e40.[CrossRef][Medline]
11. Boley, S. E., E. E. Anderson, J. E. French, L. A. Donehower, D. B. Walker, and L. Recio. 2000. Loss of p53 in benzene-induced thymic lymphomas in p53+/– mice: evidence of chromosomal recombination. Cancer Res. 60:2831-2835.
12. Bollag, R. J., and R. M. Liskay. 1988. Conservative intrachromosomal recombination between inverted repeats in mouse cells: association between reciprocal exchange and gene conversion. Genetics 119:161-169.
13. Bollag, R. J., A. S. Waldman, and R. M. Liskay. 1989. Homologous recombination in mammalian cells. Annu. Rev. Genet. 23:199-225.[CrossRef][Medline]
14. Bradshaw, H. D., Jr., and P. L. Deininger. 1984. Human thymidine kinase gene: molecular cloning and nucleotide sequence of a cDNA expressible in mammalian cells. Mol. Cell. Biol. 4:2316-2320.
15. Cavenee, W. K., T. P. Dryja, R. A. Phillips, W. F. Benedict, R. Godbout, B. L. Gallie, A. L. Murphree, L. C. Strong, and R. L. White. 1983. Expression of recessive alleles by chromosomal mechanisms in retinoblastoma. Nature 305:779-784.[CrossRef][Medline]
16. Clive, D. 1973. Recent developments with the L5178Y TK heterozygote mutagen assay system. Environ. Health Perspect. 6:119-125.[CrossRef][Medline]
17. Collins, I., and C. S. Newlon. 1994. Meiosis-specific formation of joint DNA molecules containing sequences from homologous chromosomes. Cell 76:65-75.[CrossRef][Medline]
18. Cromie, G. A., and D. R. Leach. 2000. Control of crossing over. Mol. Cell 6:815-826.[CrossRef][Medline]
19. de Nooij-van Dalen, A. G., V. H. van Buuren-van Seggelen, A. Mulder, K. Gelsthorpe, J. Cole, P. H. Lohman, and M. Giphart-Gassler. 1997. Isolation and molecular characterization of spontaneous mutants of lymphoblastoid cells with extended loss of heterozygosity. Mutat. Res. 374:51-62.[Medline]
20. Dobo, K. L., C. R. Giver, D. A. Eastmond, H. S. Rumbos, and A. J. Grosovsky. 1995. Extensive loss of heterozygosity accounts for differential mutation rate on chromosome 17q in human lymphoblasts. Mutagenesis 10:53-58.
21. Elliott, B., and M. Jasin. 2001. Repair of double-strand breaks by homologous recombination in mismatch repair-defective mammalian cells. Mol. Cell. Biol. 21:2671-2682.
22. Elliott, B., C. Richardson, and M. Jasin. 2005. Chromosomal translocation mechanisms at intronic alu elements in mammalian cells. Mol. Cell 17:885-894.[CrossRef][Medline]
23. Elliott, 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.
24. Ferguson, D. O., and W. K. Holloman. 1996. Recombinational repair of gaps in DNA is asymmetric in Ustilago maydis and can be explained by a migrating D-loop model. Proc. Natl. Acad. Sci. USA 93:5419-5424.
25. Fitzgibbon, J., L. L. Smith, M. Raghavan, M. L. Smith, S. Debernardi, S. Skoulakis, D. Lillington, T. A. Lister, and B. D. Young. 2005. Association between acquired uniparental disomy and homozygous gene mutation in acute myeloid leukemias. Cancer Res. 65:9152-9154.
26. Flemington, E., H. D. Bradshaw, Jr., V. Traina-Dorge, V. Slagel, and P. L. Deininger. 1987. Sequence, structure and promoter characterization of the human thymidine kinase gene. Gene 52:267-277.[CrossRef][Medline]
27. Gilbertson, L. A., and F. W. Stahl. 1996. A test of the double-strand break repair model for meiotic recombination in Saccharomyces cerevisiae. Genetics 144:27-41.[Abstract]
28. Giver, C. R., and A. J. Grosovsky. 2000. Radiation specific patterns of loss of heterozygosity on chromosome 17q. Mutat. Res. 450:201-209.[Medline]
29. Godwin, A. R., R. J. Bollag, D. M. Christie, and R. M. Liskay. 1994. Spontaneous and restriction enzyme-induced chromosomal recombination in mammalian cells. Proc. Natl. Acad. Sci. USA 91:12554-12558.
30. Godwin, A. R., and R. M. Liskay. 1994. The effects of insertions on mammalian intrachromosomal recombination. Genetics 136:607-617.[Abstract]
31. Grist, S. A., M. McCarron, A. Kutlaca, D. R. Turner, and A. A. Morley. 1992. In vivo human somatic mutation: frequency and spectrum with age. Mutat. Res. 266:189-196.[Medline]
32. Grosovsky, A. J., B. N. Walter, and C. R. Giver. 1993. DNA-sequence specificity of mutations at the human thymidine kinase locus. Mutat. Res. 289:231-243.[CrossRef][Medline]
33. Gupta, P. K., A. Sahota, S. A. Boyadjiev, S. Bye, C. Shao, J. P. O'Neill, T. C. Hunter, R. J. Albertini, P. J. Stambrook, and J. A. Tischfield. 1997. High frequency in vivo loss of heterozygosity is primarily a consequence of mitotic recombination. Cancer Res 57:1188-1193.
34. Hagstrom, S. A., and T. P. Dryja. 1999. Mitotic recombination map of 13cen-13q14 derived from an investigation of loss of heterozygosity in retinoblastomas. Proc. Natl. Acad. Sci. USA 96:2952-2957.
35. Haigis, K. M., J. G. Caya, M. Reichelderfer, and W. F. Dove. 2002. Intestinal adenomas can develop with a stable karyotype and stable microsatellites. Proc. Natl. Acad. Sci. USA 99:8927-8931.
36. Holliday, R. 1964. A mechanism for gene conversion in fungi. Genet. Res. 5:282-304.
37. Honma, M., M. Izumi, M. Sakuraba, S. Tadokoro, H. Sakamoto, W. Wang, F. Yatagai, and M. Hayashi. 2003. Deletion, rearrangement, and gene conversion; genetic consequences of chromosomal double-strand breaks in human cells. Environ. Mol. Mutagen. 42:288-298.[CrossRef][Medline]
38. Honma, M., M. Sakuraba, T. Koizumi, Y. Takashima, H. Sakamoto, and M. Hayashi. 2
