Previous Article | Next Article ![]()
Molecular and Cellular Biology, April 2006, p. 3098-3105, Vol. 26, No. 8
0270-7306/06/$08.00+0 doi:10.1128/MCB.26.8.3098-3105.2006
Copyright © 2006, American Society for Microbiology. All Rights Reserved.
Cheryl A. Miller, and
Jac A. Nickoloff*
Department of Molecular Genetics and Microbiology, University of New Mexico School of Medicine, Albuquerque, New Mexico 87131
Received 29 August 2005/ Returned for modification 11 November 2005/ Accepted 25 January 2006
|
|
|---|
|
|
|---|
Repetitive elements make up one-third of the mammalian genome and consist of coding DNA, and noncoding DNA such as satellites that can exist in thousands of copies. Repetitive sequences are scattered throughout the genome, including SINE and LINE elements, and ribosomal RNA gene repeats. HR between linked or unlinked repetitive elements can result in a variety of rearrangements including translocations, duplications, and deletions (45).
When DNA is broken, there are many potential homologous sequences that could be used as a repair template, including linked or unlinked repeats, sister chromatids in S and G2 phases, and homologous chromosomes. HR can proceed by several pathways and can result in different outcomes depending on the pathway and the locations of the interacting regions. Gene conversion is a conservative HR pathway that involves the nonreciprocal transfer of DNA from a donor to a recipient allele; for DSB-induced events, broken alleles are normally recipients (45). Gene conversion can occur with or without an associated crossover. Gene conversions without crossovers conserve the arrangement of the recombining regions. In linked repeats, crossovers result in either deletion of one repeat plus intervening sequences or inversion of intervening sequences. Single-strand annealing (SSA) is a nonconservative HR pathway and, in the case of direct repeats, SSA also deletes a repeat and intervening sequences (45). For simplicity, we use the terms "gene conversion" to denote events without an associated crossover and "deletion" for crossover and/or SSA events. When there are multiple homologous sequences to use as a repair template, several factors may control template choice, such as proximity and degree of homology. A better understanding of these factors will help clarify HR mechanisms and provide clues to possible causes of genomic instability.
Donor choice has been studied for mating-type switching in the yeast Saccharomyces cerevisiae where DSBs at MAT initiate conversion that depends on interactions with linked homologous sequences located on opposite arms of the chromosome. Donor choice in this highly specialized system involves a recombination enhancer sequence that regulates recombination across an entire chromosome arm (21), although the physical proximity of donor and recipient loci also plays a role (33, 63). A more general case of mitotic DSB-induced gene conversion in yeast was examined by Inbar and Kupiec (25) in which a broken allele could be repaired from either of two ectopic loci on heterologous chromosomes. However, the donor loci in this system were not identical, and the focus of the study was on the efficiency of the homology search and whether the search occurs near or far from broken ends. In mammalian cells, HR efficiency is reduced by heterologies (16, 79), and this is likely to influence donor preference when potential donors differ in the degree of sequence similarity to a recipient locus. In mouse cells, Tremblay et al. (70) showed that DSBs at an I-SceI site within a LINE element were repaired by conversion involving other LINE elements and that the most commonly used donors were those most active in retrotransposition. However, LINE elements are very abundant, and it was not possible to identify the specific donor used in any particular event. A study of LINE donor preference in human cells showed some preference for linked donors, although there were clear preferences for certain donors located both up- and downstream of the DSB, and more proximal donors were often passed over in favor of distal donors (11). These results suggest that donor preference may be influenced by accessibility of different donors in different chromatin environments, as well as "chromosome territoriality" (15).
Although the factors controlling the choice of interaction partner are poorly understood, it is clear that the search for homology can be extensive. In yeast, interactions between closely linked repeats are not favored over interactions between repeats on different chromosomes, suggesting an efficient, genome-wide search by broken ends (22). Richardson et al. (56) analyzed DSB-induced HR in mouse cells and found that repeats on nonhomologous chromosomes are used slightly less efficiently than an allelic locus on a homologous chromosome, indicating that homology searching is efficient even in the largest genomes.
Transcription influences HR in several ways. Transcription stimulates spontaneous HR (20, 46, 47, 58, 69, 74, 75, 78), spontaneous plasmid integration into a chromosome (5), Ty conversion (14, 41, 44), and gene amplification (40). Transcription also affects gene conversion tract spectra in yeast (77). However, a direct test of whether transcription influences donor choice during gene conversion has not been reported.
In the present study we targeted triple neo repeat HR substrates into human cells. DSBs were induced in a recipient neo with I-SceI nuclease and repaired through interactions with either of two donor neo genes. We show that donor preference during gene conversion is influenced by the relative positions of donor and recipient genes and by factors extrinsic to the HR substrate. In one configuration, there was a marked preference for a downstream versus an upstream donor, but this preference was eliminated when the inducible mouse mammary tumor virus (MMTV) promoter was added to the poorly utilized upstream donor. This is the first direct evidence that transcription of a donor sequence enhances its use as an HR repair template.
|
|
|---|
EagI mutations were confirmed by sequencing, and functional inactivation of neo was confirmed as these plasmids no longer conferred resistance to kanamycin in Escherichia coli. HR substrates were constructed from pSV2neo as follows. First, the natural MfeI site was destroyed by cleavage, fill-in, and ligation, then MfeI and XhoI linkers were inserted into filled-in NdeI and BamHI sites, respectively, creating pSV2neoM(N)X(B). Next, a BstBI-EagI fragment of neo with a 29-bp insertion in BanII, including the I-SceI site (66), was transferred to pSV2neoM(N)X(B), and FRT/hyg, and the various neo fragments described above were transferred to the resulting plasmid creating pDP/FRT, p5'3'/FRT, p5'3'Switch/FRT, p5'3'Rev/FRT, and p5'3'MMTV/FRT, each of which has three copies of neo and FRT/hyg (Fig. 1). The I-SceI expression vector pCMV(3xNLS)I-SceI and negative control vector pCMV(I-SceI) were described previously (31, 66).
![]() View larger version (26K): [in a new window] |
FIG. 1. Structures of recombination substrates. Recombination substrates each contain one recipient allele with the I-SceI recognition sequence and two donor alleles carrying either the wild-type BanII recognition sequence or BsaI opposite of I-SceI. Donor alleles also carry the EagI mutation. (A) The distal-proximal substrate has both donors downstream of the recipient. Numbers indicate distances in base pairs between 1.4-kbp neo repeats. (B and C) The 5'3' substrate has donors flanking the recipient allele. 5'3'Switch is identical to 5'3' except that the donors are switched. (D) 5'3'Reverse is identical to 5'3' except that the substrate is oriented in the genome in the opposite direction. (E) MMTV5'3' is identical to the 5'3' substrate except that the MMTV promoter was added to the 5' neo. (F) Configuration of HR substrates within the targeting locus. Targeting of HR substrates disrupts a zeomycin resistance gene and activates a hygromycin resistance gene (for details, see reference 59).
|
DSB-induced HR assays. Parent cells (5 x 105) were seeded to 3.5-cm (diameter) wells, incubated for 24 h, and lipofected with 2 µg of either pCMV(3xNLS)I-SceI to induce DSBs or with the negative control vector pCMV(I-SceI) as described previously (2, 9). After 24 h, the cells were harvested, and 105 cells were seeded to each of four 10-cm dishes. In experiments with the cell line carrying the MMTVneo allele (Fig. 1E), 1 µM dexamethasone (dex) was added to half of the culture dishes 4 h prior to lipofection and was maintained at this concentration until cells were harvested and reseeded. After an additional 24 h, G418 was added to a final concentration of 750 µg/ml, and cultures were incubated for 14 days. G418r colonies were isolated and expanded to confluence in 10-cm dishes, the cells were harvested, and genomic DNA was prepared and analyzed by Southern hybridization with the 1.4-kbp neo fragment and by PCR as described previously (2, 66). Cell viability was measured by seeding dilutions of cells into nonselective media and scoring colonies after 12 to 14 days. HR frequencies were calculated as the ratio of G418r colonies to the number of viable cells plated in medium with G418. Southern hybridization distinguishes gene conversions from deletions. PCR with a simian virus 40 (SV40) promoter-specific primer (5'-GCCCAGTTCCGCCCATTCTC) and a neo-specific primer (5'-CGAAATCTCGTGATGGCAGG) amplifies a 1.4-kbp fragment carrying just the SV40 promoter-driven neo. Digestion of these PCR products with BanII or BsaI identifies interaction partners for gene conversion and deletion events. Approximately 4% of G418r colonies gave mixed product patterns, such as deletion and gene conversion, or independent conversions involving both donor alleles. These likely result from independent DSB repair events in G2 cells and were therefore scored as two events.
RNA preparation and Northern analysis. neo transcription levels in the strain with the MMTVneo allele were measured by Northern blot as described previously (2, 66). Parallel cultures were treated with 1 µM dex or mock treated for 4 h, the cells were harvested, and total RNA was prepared by using TRIzol reagent as recommended by the manufacturer (Invitrogen). Northern analysis with 32P-labeled neo fragment as probe was performed as described previously (47).
|
|
|---|
EagI mutations were included because the fifth HR substrate has an upstream neo driven by the dex-regulated MMTV promoter and must be inactive to allow selection of neo+ products resulting from I-SceI-induced HR. The
EagI mutation has minimal effect on HR outcomes because conversion tracts are rarely long enough to include this mutation (66). We used FLP targeting to introduce HR substrates into an FRT site in human 293 cells to eliminate potential position effects. Targeting provides highly reproducible gene expression levels (19). Similarly, we found that HR frequencies and outcomes were indistinguishable in cell lines carrying identical HR substrates (data not shown); the results presented below are either from a single cell line or pooled from two such lines. The use of identical, targeted alleles permits comparisons among the five HR substrates.
In the distal-proximal substrate, the two donor loci are downstream of the recipient, (Fig. 1A). In the remaining substrates, donor loci are upstream (5') and downstream (3') of the recipient. The most frequent HR outcomes for these allele configurations are shown in Fig. 2. The distal-proximal substrate gives two types of gene conversions and two types of deletions (Fig. 2A). The 5'3' substrates give two types of gene conversions, one deletion, and a repeat expansion product with four neo genes that may arise by unequal sister chromatid exchange or long-tract sister chromatid conversion (Fig. 2B) (27), hereafter described as "repeat expansions." There are several other possible outcomes when the upstream donor is regulated by the MMTV promoter, however, for events initiated by DSBs at the I-SceI site the outcomes shown in Fig. 2B are expected to predominate because unbroken alleles rarely convert. PCR and Southern blot analyses were used to distinguish among the various product types (see Materials and Methods). DSB-induced HR frequencies of the different substrates are shown in Table 1.
![]() View larger version (29K): [in a new window] |
FIG. 2. Potential HR products. Symbols are as shown in Fig. 1. (A) The distal-proximal substrate yields two types of gene conversions which retain the gross structure of the substrate and two types of deletions. (B) The 5'3' substrates yield two gene conversions, one deletion, and one repeat expansion product. (C) Unequal pairing of sister chromatids in the 5'3' substrate can occur between a recipient allele (center) and either 3' (top) or 5' (bottom) donors. Potential pairing is indicated by shading between alleles.
|
|
View this table: [in a new window] |
TABLE 1. DSB-induced HR frequencies in triple neo repeat substrates
|
97% of products arose by gene conversion (2, 6, 39, 66). Southern analysis of 65 products from the DP19 strain showed that 57% had deleted one or two copies of neo, and 15% displayed complex patterns (Table 3). The complex patterns might result from multiple HR events or coupled HR/NHEJ events (54). This high deletion rate is not intrinsic to the triple repeat structure but is typical of closely spaced direct repeats (59). Among DP19 gene conversion products, there was a modest (2:1) bias in favor of the proximal donor (Table 3), but this is not significantly different from a 1:1 distribution (P = 0.49). |
View this table: [in a new window] |
TABLE 2. Interaction partner among products of distal-proximal strains
|
|
View this table: [in a new window] |
TABLE 3. DSB-induced HR products of distal-proximal strain DP-19
|
|
View this table: [in a new window] |
TABLE 4. DSB-induced HR product spectra of 5'3' strains
|
![]() View larger version (31K): [in a new window] |
FIG. 3. Transcription enhances donor use during gene conversion. (A) Induction of transcription of MMTVneo in the MMTV5'3' substrate. Total RNA was isolated from cultures incubated without () or with (+) 1 µM dex for 4 h. Northern hybridization was performed with a 32P-labeled neo probe; the autoradiograph is shown above; equal loading is shown by ethidium bromide staining of 28S rRNA (B) Percentage of gene conversion events using 5' donor as a repair template for 5'3', the net value for 5'3' and 5'3'Switch, and MMTV5'3' ± dex. P values were calculated by the Fisher exact tests.
|
Transcription of a donor enhances its use during DSB-induced HR. Having established a strong preference for the 3' donor among conversions with the 5'3'/Switch substrates, we were interested in whether transcription of the poorly utilized 5' donor would enhance its use during gene conversion. To test this, we added the MMTV promoter to the 5' donor, creating strain MMTV5'3' (Fig. 1E). Transcription levels in cells cultured with or without dex were measured by Northern blot. Because the SV40 promoter-driven neo and MMTVneo genes differ in their regulatory, mRNA splicing, and poly(A) signal sequences, mRNAs from these genes give distinct bands on agarose gels. As shown in Fig. 3, dex greatly increases transcription of the MMTVneo gene. The MMTV promoter is leaky in the absence of dex (46), perhaps as a result of serum hormones in the growth medium. Therefore, MMTVneo transcription levels are low in the absence of dex and high in the presence of dex.
Note that the additional MMTV DNA sequences occur outside the homology boundaries defined by the neo repeats. Thus, the relative positions of the three neo genes are identical in the 5'3', 5'3'Switch, and MMTV5'3' substrates (Fig. 1B, C, and E). We characterized 92 and 100 DSB-induced HR products from the MMTV5'3' substrate in the absence or presence of dex, respectively. Because the MMTV promoter regulates the 5' neo, deletions between the 5' neo and either the central or 3' neo alleles could theoretically give rise to a G418r product. However, such events are expected to be rare when HR initiates at DSBs in the central neo and, consistent with this expectation, all 122 deletion products from the MMTV5'3' substrate (±dex) resulted from interactions between the central (broken) and the 3' repeat. The conversion: deletion ratio was similar for the MMTV5'3' and 5'3' substrates, and this ratio was not changed by dex treatment (Table 4). Interestingly, the use of the 5' neo repeat as donor during gene conversion increased with the addition of the MMTV promoter, perhaps owing to leaky transcription, and the effect was further enhanced by treatment with dex (Fig. 3). We conclude that transcription of a donor enhances its use during DSB-induced gene conversion.
|
|
|---|
In the present study we found that repeat configuration influences the choice of interaction partner during DSB repair. When two effectively identical repeats were downstream and closely linked to a third repeat suffering a DSB (distal-proximal substrate), there was no bias in favor of proximal repeat interactions leading to deletions. In addition, although the observed 2:1 bias in favor of the proximal donor during gene conversion is consistent with prior studies in yeast (34, 57), this bias was not statistically significant and probably reflects close spacing of these donors. A bias toward proximal donors is more likely when donors are separated by larger distances; such a bias would serve to stabilize the genome by reducing the probability of interactions over large distances and the associated risk of large-scale deletions and rearrangements. In both yeast and mammalian cells, SSA efficiency depends on the timing with which complementary single strands are exposed in direct repeats (17, 35, 36, 59). Given that crossovers are restricted, most distal-proximal deletions probably arise by SSA, and the lack of bias in favor of proximal deletions suggests that end processing is rapid and extensive, exposing the relatively closely linked distal and proximal repeats at similar times after break induction.
In the 5'3' and 5'3'Switch substrates we observed a marked preference for the 3' donor during gene conversion. Several factors may influence donor choice during gene conversion, including the extent and degree of homology of available donors. Homology length is thought to influence recognition between homologous sequences in yeast, thus ensuring the best match and minimizing large-scale rearrangements (3). In mammalian cells, interruptions to homology reduce spontaneous HR (2, 76, 79). In the present HR substrates, neither donor length nor the degree of homology differ, and therefore these factors cannot account for the observed preference for the 3' donor. It is thought that direct repeat HR occurs primarily via interactions between sister chromatids (27, 29). Detectable direct repeat HR between sister chromatids requires misalignment of repeats. Misalignment in the 5'3' HR substrates can occur in two directions, with the central (recipient) repeat paired with either the 5' or the 3' donors. However, differential pairing cannot explain the observed preference for the 3' donor because both misalignments have the same pairing potential (Fig. 2C). The limited use of the 5' donor probably does not reflect constraints due to DNA flexibility because the distances between the central neo and 5' and 3' donors are similar. Although transcription initiated in the central neo could "read through" into the 3' donor and thereby enhance its use, this is unlikely because there are transcription termination signals downstream of the central neo. Moreover, a similar bias toward the 3' donor would be expected in the 5'3'Reverse substrate, but this was not observed. The differences in donor preference between the 5'3'/Switch and 5'3'Reverse substrates indicates that donor preference is influenced by factors external to the neo triple-repeat structure, such as the direction of replication or nearby transcription units which might alter the local chromatin environment. Although our HR substrates were targeted to a single chromosomal locus, each cell line is a clonal isolate and differences in donor preference might reflect clonal variation. However, we believe this is unlikely for the following reasons. First, three independent strains with the distal-proximal substrate gave similar product spectra. Second, the 5'3' and 5'3'Switch substrates are effectively identical, and these, too, gave similar product spectra. Third, the enhanced use of the 5' donor with increased transcription in the MMTV5'3' strain cannot result from clonal variation as a single population of cells was used in this experiment.
There are several mechanistic similarities among DNA replication, recombination, and transcription, including DNA unwinding, assembly of DNA/RNA synthesis complexes, and movement of these protein complexes along a template (32). Furthermore, these processes share certain proteins. Some transcription factors operate in both transcription and DNA repair and quickly switch between the two processes (24). Rad52 is essential for HR in yeast (50), and human RAD52 also has roles in HR (31, 51, 53, 61, 72). Human RAD52 associates with the XPB and XPD subunits of transcription factor TFIIH and RNA polymerase II, and it has one domain that activates transcription, and another that represses transcription (37). This association provides a mechanistic link between transcription and HR. In yeast and mammalian cells, transcription enhances spontaneous HR (46, 47, 69, 73), but it does not enhance DSB-induced HR (66, 77) (Table 1). Transcription of a recipient allele has little or no effect on DSB-induced gene conversion tract spectra in yeast or mammalian cells (66, 77), but conversion tract spectra in yeast were altered by transcription of a donor allele (77), perhaps reflecting enhanced strand invasion in transcribed regions. Enhanced invasion of transcribed regions can also account for results obtained in studies of retrotransposons. Yeast Ty and mammalian LINE elements transpose through cDNA copies of mRNA (10, 12). Transcription of Ty elements enhances transposition (10), and this is likely to be the case for LINE elements as well (49). Thus, the more frequent use of actively transposing (i.e., actively transcribed) LINE elements as donors during DSB repair by gene conversion (70) provided indirect evidence that transcription of a donor may enhance its use in gene conversion. The present study provides direct evidence that transcription enhances the use of a donor allele by two- to threefold during DSB-induced gene conversion in mammalian cells (Fig. 3).
Transcriptional enhancement of donor preference and spontaneous HR may be mechanistically linked. Two general classes of models have been advanced to explain how transcription enhances spontaneous HR. One class suggests that transcription alters the repair of preexisting damage, perhaps by facilitating strand invasion or enhancing the recruitment of HR proteins to damaged sites. This latter idea is formally analogous to transcription-coupled repair of other types of DNA damage (24, 65) and is supported by biochemical evidence indicating an association between transcription factors and RAD52 (37). Alternative models suggest that spontaneous DNA damage is enhanced in transcribed regions, perhaps because single strands exposed during transcription are more susceptible to damage or because of torsional stress induced by movement of the transcription machinery along DNA (38). In the latter models, HR increases as a result of repair of the additional lesions or to relieve torsional stress. Our finding that transcription of a donor enhances its use during gene conversion suggests enhanced interactions between transcriptionally active loci during DSB repair, perhaps as a result of transcription- and repair-associated chromatin remodeling (60, 62, 71). Although further studies will be required to determine how transcription enhances donor preference, this preference represents a novel control system for genome stabilization that would minimize the use of nontranscribed pseudogenes as donors. This is particularly important in the highly repetitive genomes of higher eukaryotes, where preventing interactions with most potential donor loci would serve to minimize HR-associated genome rearrangements.
This research was supported by grant CA77693 to JAN from the National Cancer Institute of the National Institutes of Health.
Present address: CerroSci LLC, P.O. Box 177, Cerrillos, NM 87010. ![]()
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»