![[Feedback]](/icons/banner/feedbackACT.gif)
![[For Subscribers]](/icons/banner/subscriberACT.gif)
![[Archive]](/icons/banner/archiveACT.gif)
![[Search]](/icons/banner/searchACT.gif)
![[Contents]](/icons/banner/tocACT.gif)
Microbiology and Molecular Biology Reviews, December 2002, p. 630-670, Vol. 66, No. 4
1092-2172/02/$04.00+0 DOI: 10.1128/MMBR.66.4.630-670.2002
Copyright © 2002, American Society for Microbiology. All Rights Reserved.
Role of RAD52 Epistasis Group Genes in Homologous Recombination and Double-Strand Break Repair
Lorraine S. Symington*
Department of Microbiology and Institute of Cancer Research, Columbia University College of Physicians and Surgeons, New York, New York 10032
The process of homologous recombination is a major DNA repair pathway that operates on DNA double-strand breaks, and possibly other kinds of DNA lesions, to promote error-free repair. Central to the process of homologous recombination are the RAD52 group genes (RAD50, RAD51, RAD52, RAD54, RDH54/TID1, RAD55, RAD57, RAD59, MRE11, and XRS2), most of which were identified by their requirement for the repair of ionizing-radiation-induced DNA damage in Saccharomyces cerevisiae. The Rad52 group proteins are highly conserved among eukaryotes, and Rad51, Mre11, and Rad50 are also conserved in prokaryotes and archaea. Recent studies showing defects in homologous recombination and double-strand break repair in several human cancer-prone syndromes have emphasized the importance of this repair pathway in maintaining genome integrity. Although sensitivity to ionizing radiation is a universal feature of rad52 group mutants, the mutants show considerable heterogeneity in different assays for recombinational repair of double-strand breaks and spontaneous mitotic recombination. Herein, I provide an overview of recent biochemical and structural analyses of the Rad52 group proteins and discuss how this information can be incorporated into genetic studies of recombination.
Recombination refers to the exchange or transfer of information between DNA molecules and is found in all organisms that have been studied in detail. Homologous recombination involves the exchange of DNA between sequences of perfect or near perfect homology over several hundreds of base pairs. In contrast, nonhomologous recombination occurs between sequences with little or no sequence homology. The process of homologous recombination plays essential roles in the mitotic and meiotic cell cycles of most eukaryotic organisms. In meiosis, the primary function of recombination is to establish a physical connection between homologous chromosomes to ensure their correct disjunction at the first meiotic division. In addition, meiotic recombination contributes to diversity by creating new linkage arrangements between genes, or parts of genes. It is now widely recognized that the primary function of homologous recombination in mitotic cells is to repair double-strand breaks (DSBs) that form as a result of replication fork collapse, from processing of spontaneous damage, and from exposure to DNA-damaging agents. Recombination is also required to repair the DSBs that initiate programmed rearrangements, such as mating-type switching in Saccharomyces cerevisiae.
Most of the genes in the RAD52 epistasis group (RAD50, RAD51, RAD52, RAD54, RAD55, RAD57, RAD59, RDH54 [TID1], MRE11 [RAD58], and XRS2) were identified by their requirement in the repair of ionizing-radiation (IR)-induced DNA damage. Mutations in these genes lead to defects in meiotic and/or mitotic recombination, providing evidence for a link between DSB repair (DSBR) and homologous recombination. Homologues of the RAD52 group of genes have been identified in many other eukaryotes, and in some cases in prokaryotes and archaea, indicating high conservation of the recombinational repair pathway. The recent discovery that several human cancer-prone syndromes, for example, Nijmegen breakage syndrome (NBS) and ataxia-telangiectasia-like disorder (A-TLD), are caused by defects in DSBR has highlighted the importance of this repair pathway in maintaining genome integrity and cancer avoidance (47, 358, 411).
The goal of this review is to provide an update of the role of the RAD52 group genes in spontaneous and DSB-induced homologous recombination. Other aspects of DSBR, cell cycle checkpoints, and meiosis are beyond the scope of this review, and the reader is referred to several excellent reviews on these topics (212, 271, 432). Most of the studies described in this review were performed with S. cerevisiae and with proteins corresponding to the S. cerevisiae gene products. When studies with other organisms are described, the gene or protein is prefixed by the abbreviated species name. The first part of the review describes models for DSBR and physical and genetic assays used to monitor repair and recombination. In the second part, the biochemical functions of the Rad52 group proteins and phenotypes of the rad52 group mutants in various recombination assays are described.
|
CURRENT VIEW OF THE MECHANISMS OF HOMOLOGOUS RECOMBINATION
|
|---|
Much of our understanding of the mechanisms of recombination is based on organisms, such as S. cerevisiae, in which all of the products of an individual meiosis can be recovered for analysis in the form of asci containing four haploid spores. Two types of recombination events have been identified based on the segregation of heterozygous markers during meiosis: crossing over and gene conversion. A crossover between linked heterozygous markers results in new linkage arrangements for two spore products, but the markers are still recovered in Mendelian ratios. Gene conversion represents the nonreciprocal transfer of information between two homologous sequences to duplicate one of the alleles, with the corresponding loss of the other, resulting in a non-Mendelian segregation. Studies with yeast and other fungi have shown that gene conversion events are frequently associated with the exchange of flanking markers (145). This association is thought to be due to alternate resolution of a Holliday junction containing intermediate to generate crossover or noncrossover products (135). Several models have been proposed to explain the molecular mechanisms of recombination, and of these the DSBR and synthesis-dependent strand-annealing (SDSA) models are most consistent with the available genetic data (135, 236, 257, 381).
Double-Strand Break Repair Model
The observation that IR stimulates recombination suggested that recombination is initiated by DSBs. Studies by Orr-Weaver et al. demonstrated that a nonreplicating plasmid containing a DSB made within a region of the plasmid with homology to genomic sequences recombined with high efficiency following transformation of yeast (269). The plasmid DNA integrated at the homologous locus and was flanked by a duplication of the region of yeast DNA carried on the plasmid. When plasmid DNA was digested with restriction endonucleases to delete an internal fragment from the yeast DNA, it was still found to transform yeast at high frequency. The resulting transformants were indistinguishable in structure from those obtained using uncut DNA, indicating that repair of the gapped region had occurred during integration. This repair of a double-strand gap is equivalent to gene conversion without mismatch repair. Using linearized replicating plasmids, Orr-Weaver and Szostak reported the recovery of approximately equal numbers of integrated and non-integrated plasmids and concluded that gene conversion by DSBR can occur with or without crossing over (268). These observations formed the basis for the DSBR model of recombination (381) (Fig. 1A). In this model, the ends of the break are resected to form 3' single-stranded tails that are active in strand invasion with a homologous duplex. Following strand invasion, the 3' end is extended by DNA synthesis. The D-loop formed by strand invasion is able to pair with the other side of the DSB, and the 3' end of the noninvading strand is also extended by DNA synthesis, forming a double-Holliday-junction (dHJ) intermediate. Random resolution of the two Holliday junctions is expected to yield equal numbers of crossover and noncrossover products.
View larger version (15K):
[in this window]
[in a new window]
|
FIG. 1. Models for the repair of DSBs. (A) In the DSBR model, the ends are processed to yield 3' single-stranded tails. The 3' ends invade the homologous duplex, priming DNA synthesis. After ligation, a dHJ intermediate is formed, which can subsequently be resolved by endonucleolytic cleavage of the two Holliday junctions to generate crossover or noncrossover products. (B) In the SDSA model, the ends are processed to yield 3' single-stranded tails, one of which invades the homologous duplex, priming DNA synthesis. The displacement loop (D-loop) formed by strand invasion could be extended by DNA synthesis or could migrate with the newly synthesized DNA. After displacement from the donor duplex, the nascent strand pairs with the other 3' single-stranded tail and DNA synthesis completes repair. (C) The initial steps in the BIR model are the same as in the SDSA model, but DNA synthesis from the invading strand continues to the end of the DNA molecule. (D) In the SSA model, a DSB made between direct repeats is subject to resection to generate 3' single-stranded tails. When complementary sequences are revealed due to extensive resection, the single-stranded DNA anneals, resulting in deletion of one of the repeats and the intervening DNA. The 3' tails are endonucleolytically removed, and the nicks are ligated. The 3' ends are indicated by arrowheads.
|
|
Further support of the DSBR model as a general model for recombination came from the observation of transient DSBs at meiotic recombination hot spots (46, 369). The ends of meiotic DSBs are processed to yield 3' single-stranded tails of about 600 nucleotides (370). DSBs made by the HO endonuclease in mitotic cells are also resected to generate long 3' single-stranded tails (419). The length of the single-stranded tails at the ARG4 locus correlates well with the length of gene conversion tracts within the ARG4 gene, suggesting that most conversion events arise from repair of heteroduplex DNA and that four-strand branch migration is limited (370). Branched DNA molecules with the topology expected from dHJ intermediates have been detected by two-dimensional agarose gel electrophoresis of meiotic DNA, providing further support for the model (61, 327, 328).
Synthesis-Dependent Strand-Annealing Model
Subsequent studies of plasmid gap repair in yeast and Ustilago maydis and chromosomal DSBR following P element excision in Drosophila melanogaster have shown a lower association of crossing over with gene conversion (20, 96, 257, 291). Similarly, mating-type switching in S. cerevisiae, which is a gene conversion event initiated by a site-specific DSB at the MAT locus made by the HO endonuclease (183, 360), is rarely associated with crossing over. To account for the low levels of associated crossing over observed for some DSB-induced gene conversion events, the SDSA/migrating D-loop model was proposed (96, 257, 360) (Fig. 1B). In this model, strand invasion occurs as envisioned in the DSBR model, but after extensive DNA synthesis primed from the invading strand, the elongated invading strand is displaced and pairs with the other side of the break. DNA synthesis can then be primed from the noninvading 3' end to repair the DSB or gap. An alternative scenario for gap repair involves coupling of lagging-strand DNA synthesis to leading-strand synthesis from the invading strand (137). Further evidence in support of the SDSA model comes from the observation that the donor sequences are generally unchanged during DSB-induced gene conversion (273).
Recent studies of meiotic recombination intermediates have identified a second class of branched molecules, termed single-end invasion intermediates (SEI) (144). These could be precursors to dHJ intermediates or could be processed directly into noncrossover products as predicted by the SDSA model. The relatively long-lived dHJ intermediates appear to be precursors for crossover products. The observation that formation of crossover, but not noncrossover, products is blocked by an ndt80 mutation is consistent with the suggestion that these two classes of recombinants are derived from different intermediates (11). Allers and Lichten propose that noncrossovers are generated by SDSA and crossovers are generated from the resolution of dHJ intermediates that are formed as envisioned by the DSBR model (11).
Other Mechanisms for Homology-Dependent Double-Strand Break Repair
When a DSB is induced at the MAT locus on one chromosome III homologue of diploid cells, the break is generally repaired by gene conversion with about 20% associated crossing over (217). In certain genetic backgrounds, gene conversion is eliminated and repair occurs by invasion of the donor duplex by the broken chromosome followed by replication to the end of the donor chromosome (217) (Fig. 1C). This process is known as break-induced replication (BIR). This nonreciprocal process is likely to be important for telomere maintenance in the absence of telomerase (190, 388). Homology-dependent duplication of an entire chromosome arm to the end of a transformed linearized plasmid has also been detected in yeast and is presumed to occur by BIR (248). Another pathway of homology-dependent repair, single-strand annealing (SSA), is restricted to DSBs that occur between direct repeats (Fig. 1D). These events have been detected in yeast and animal cells by using artificial direct repeats and could conceivably be important for repair in genomes of higher eukaryotes that contain many repeated sequences (101, 202, 224). After formation of the DSB, the ends are resected to produce 3' single-stranded tails, which can anneal when resection is sufficient to reveal complementary single-stranded regions. Single-stranded tails are removed by nucleases, and the resulting gaps and nicks are filled in by DNA repair synthesis and ligation. This process is accompanied by deletion of one of the repeats and the DNA between the direct repeats and is therefore considered to be mutagenic.
|
GENETIC AND PHYSICAL ASSAYS FOR RECOMBINATION
|
|---|
Although sensitivity to IR is a universal feature of rad52 group mutants, the mutants show considerable heterogeneity in different assays for recombinational repair of DSBs and spontaneous mitotic recombination. A brief description of the commonly used recombination assays and the products that are recovered from wild-type cells follows. Later sections discuss in detail the phenotype of the rad52 group mutants in these assays.
Spontaneous and Induced Heteroallelic Recombination
The term "spontaneous recombination" refers to events that occur during normal growth rather than following treatment of cells with DNA-damaging agents such as UV light or IR. These events most probably result from recombination initiated by spontaneous lesions that arise from replication fork arrest or collapse, intermediates in base excision repair, aberrant repair, or transcription. Strand breaks that occur in the context of the replication fork are most likely to be repaired using the sister chromatid, an event that is normally genetically silent. Spontaneous recombination can be detected between chromosome homologues in diploid cells or between artificial duplications in haploids or diploids. In most cases, the two recombining sequences contain mutant alleles (heteroalleles) of a selectable gene to allow selection for recombination events during growth of a culture. Heteroallelic recombination in diploids generally occurs by gene conversion with low (10 to 20%) associated exchange of flanking markers. Understanding the mechanisms of spontaneous recombination is difficult because these events occur at low frequency (around 10-6/cell/generation) and are usually selected. Although both products of a G1 event should be retained, the products of an event initiated during the S or G2 phase of the cell cycle will segregate to different daughter cells 50% of the time and only one of these will be recovered by selection (Fig. 2). The frequency of heteroallelic recombination can be increased by up to 1,000-fold when cells are treated with UV light or IR; these events are referred to as induced recombination. Insertion of the recognition sequence for a rare-cutting endonuclease, such as HO, within one of the repeats results in very high levels of recombination when HO endonuclease is induced and allows the analysis of unselected events. Analysis of HO-induced gene conversion has shown that the level of associated crossing over is similar to that for spontaneous events (217, 261).
View larger version (31K):
[in this window]
[in a new window]
|
FIG. 2. Mitotic recombination in diploids. A diploid containing leu2 heteroalleles and heterozygous for HIS4 is shown. Gene conversion associated with crossing over gives rise to Leu+ His+ or Leu+ His- recombinants, whereas gene conversion events maintain heterozygosity at HIS4. A BIR event cannot be distinguished from a G2 crossover unless the reciprocal product is recovered.
|
|
Haploid strains containing duplicated genes have been used extensively to study mechanisms and genetic control of mitotic recombination (177). Chromosomal direct repeats can be generated by targeting the integration of a plasmid containing a yeast gene to the homologous chromosomal location to form a nontandem duplication (Fig. 3). Recombination between the repeats can occur by conversion, maintaining the structure of the duplication, or by a deletion event that removes one repeat and the intervening DNA (176, 319, 391). Deletion events between direct repeats can be directly selected if the intervening DNA contains a marker with counterselection, for example URA3, or if the repeats are truncated genes that, by deletion, restore a functional copy of the gene. Conversion events between repeats can occur by an intrachromatid or sister chromatid interaction. Gene conversion between misaligned sister chromatids can generate a triplication on one sister while retaining the direct repeat on the other chromatid. Triplications can also occur by unequal crossing over between sister chromatids, but in this case the reciprocal product is equivalent to a deletion event. Because spontaneous recombination events occur at low frequency, events are usually selected and therefore the reciprocal product is not recovered. Intrachromatid gene conversion associated with crossing over should generate a deletion on the chromosome and an extrachromosomal circular product. It is clear that deletions are produced at high frequency between direct repeats, but the reciprocal product expected from a conservative crossover event is produced in only about 7% of deletions (316). This suggests that intrachromatid gene conversion is rarely associated with crossing over or that most gene conversion events occur between misaligned sisters rather than by intramolecular interactions.
View larger version (19K):
[in this window]
[in a new window]
|
FIG. 3. Use of a direct-repeat substrate to measure mitotic recombination. The substrate contains leu2 heteroalleles separated by a copy of the URA3 gene. Leu+ Ura+ recombinants can be generated by intrachromatid or sister chromatid conversion or by unequal sister chromatid exchange. Ura- recombination events can occur by an intrachromatid or unequal sister chromatid crossover, by SSA, or by replication slip mispairing.
|
|
HO-induced recombination between repeats can occur at high enough frequency to be studied nonselectively in order to recover both products of a recombination event and to monitor the kinetics of DSBR in synchronized populations of cells (260, 312). When the HO-cut site is within one of the repeats, the same spectrum of recombination events found for spontaneous events is recovered, including conversions, deletions, and triplications. The high frequency of triplication or deletion events recovered in one study suggests that sister chromatid interactions are at least as frequent as intrachromatid events (416). Spontaneous deletion events between direct repeats are assumed to occur by SSA, mispairing during replication, or replication slippage rather than by a reciprocal exchange between the repeats. A study of HO-induced recombination involving direct repeats of lacZ contained on a CEN plasmid supports the idea that SSA is a major pathway for DSB-induced deletion formation between direct repeats (101). However, in another study, evidence for DSB-induced reciprocal recombination between direct repeats was presented. An ARS1 element and TRP1 marker were introduced between direct repeats so that an intrachromatid reciprocal exchange could be detected as unstable Trp+ colonies (300). No stable Trp- recombinants were recovered following HO induction, indicating that SSA was not playing a significant role in repair. Of the recombinants, 6% had the unstable Trp+ phenotype indicative of a reciprocal exchange. The reasons for these conflicting results are unclear but could include the length of the repeats, the distance between the repeats, and the location (chromosomal versus plasmid). Deletions are the primary products recovered when the HO-cut site is within unique sequences between the repeats and appear to result from SSA (347, 362) (Fig. 1D).
One concern about the use of direct repeats to study recombination is that deletions can occur by a variety of mechanisms (177). Furthermore, direct repeat recombination still occurs at high frequency in most of the rad52 group mutants (199, 233). In efforts to avoid SSA and replication-associated events, a number of studies have used repeats in inverted orientation (Fig. 4). These generally contain additional sequences between the repeats because perfect palindromes are unstable (118, 313). In most cases the repeats consist of heteroalleles or truncated genes, and in one study inversions were selected by increased expression of a reporter gene present between the repeats (4, 5, 421). To facilitate the detection of recombination events by a colony color assay, Rattray and Symington generated an inverted-repeat substrate from ade2 heteroalleles (299). ade2 mutants accumulate a red pigment, resulting in the formation of red colonies; recombination events that restore a functional ADE2 gene result in white sectors within the red colony. Recombination events within the ade2 reporter can occur by gene conversion or by inversion, and these events occur with equal frequency in wild-type strains (299). Although inversions between inverted repeats could conceivably occur by an intrachromatid crossover, recent studies suggest other mechanisms. Chen and Jinks-Robertson (56) presented evidence in support of long conversion tracts between misaligned sister chromatids to generate inversions (Fig. 4B), as originally suggested by Rothstein et al. (308). Another model suggests that BIR followed by SSA could generate inversions between inverted repeats (166).
View larger version (24K):
[in this window]
[in a new window]
|
FIG. 4. Inverted-repeat substrates. (A) The ade2 heteroalleles can recombine to Ade+ by conversion or inversion (apparent crossover) of the intervening DNA. (B) A long-tract conversion event between sister chromatids can result in an Ade+ product with the intervening DNA inverted.
|
|
Repeats present on nonhomologous chromosomes (ectopic), usually as heteroalleles, have also been used to study recombination (156). If the repeats have the same orientation with respect to the centromeres, reciprocal crossovers can occur to form viable products. The association of crossing over with gene conversion is about 20%, similar to that observed for recombination between homologous chromosomes (156). The advantage of using an ectopic recombination assay is that simple SSA events cannot give rise to recombinants; the disadvantage is the low frequency at which ectopic recombination occurs, compared with direct or inverted repeats.
Mating-Type Switching
The budding yeast mating-type switching system provides a useful tool to study intermediates in DSBR and the genetic control of this process (124) (Fig. 5). The HO gene is normally expressed in late G1, but expression can be regulated using a GAL1-HO cassette so that HO endonuclease can be synchronously induced in a population of cells by addition of galactose to the growth medium (132). After the formation of a DSB at the MAT locus by the HO endonuclease, the ends are processed to form 3' single-stranded tails (419). The 3' single-stranded tail CEN distal to the cut site invades the donor cassette to initiate DNA synthesis by using components of both leading- and lagging-strand synthesis (137). The initial strand invasion and extension step can be monitored by PCR, and the formation of completed products can be detected by using novel restriction fragments (419). Branched DNA molecules have not been reported as intermediates during mating-type switching, suggesting that these intermediates are less stable than intermediates in meiotic recombination. The ability of mutant strains to switch mating type can also be determined by monitoring the survival of cells following HO expression and analyzing the survivors for mating phenotype (221, 414).
As described above, the HO cut site can be inserted at other locations to create an initiation site for recombination. The advantage of this system over inducing events by irradiation is that the precise location of the initiating lesion is known. Furthermore, the high efficiency of cutting by HO endonuclease allows physical monitoring of recombination, as described for mating-type switching, and recovery of all the products of a recombination event. The Haber laboratory has used the HO system extensively to characterize the mechanisms of direct- and inverted-repeat recombination, as well as allelic recombination in diploids (271). The high efficiency of cleavage by HO endonuclease is currently being exploited to identify proteins that associate with DSBs in vivo by the chromatin immunoprecipitation method (91).
To study DSB-induced recombination in mammalian cells, several groups have made use of the rare-cutting HO and I-SceI endonucleases. I-SceI is a site-specific endonuclease responsible for intron mobility in mitochondria of yeast and initiates a gene conversion event that resembles HO-induced gene conversion (292). Adenoviruses expressing the HO gene or containing the HO cut site can be used to infect human cells in culture (262). Coinfection with the two viruses results in efficient cutting of viral genomes containing the cut site, and these are substrates for end-joining repair (262). Direct-repeat recombination substrates containing a cleavage site for I-SceI have been stably transfected into several mammalian cell lines (157, 310). The I-SceI endonuclease was expressed by transient transfection using the constitutive cytomegalovirus promoter. Although the efficiency of cleavage was lower than that found for I-SceI or HO endonucleases in yeast, the frequency of homologous recombination was induced more than 100-fold in the presence of I-SceI.
Ends-In and Ends-Out Recombination
The early studies by Orr-Weaver et al. demonstrated efficient repair of DSBs and gaps in plasmids by homologous recombination during yeast transformation (268, 269). The repair of breaks and gaps on plasmids is referred to as "ends-in" recombination. Subsequent studies showed that homologous linear DNA fragments could be used to replace chromosomal sequences during transformation (309). This method of gene replacement or gene targeting is routinely used to knock out genes. Since the orientation of the two ends is opposite to that for plasmid gap repair, gene replacement is sometimes referred to as "ends-out" recombination. The major difference between ends-in and ends-out events is that during ends-in recombination one invading end primes DNA synthesis toward the other end of the broken plasmid, much the same as in a chromosomal DSB, whereas replication is primed away from the fragment during ends-out recombination (Fig. 6). There is also evidence that ends-out recombination can occur by assimilation of a single strand from the linear transforming fragment into the genome to form a displacement loop (D-loop) (194). Subsequent repair of the D-loop restores the original marker or incorporates information from the transforming fragment into the chromosomal locus.
View larger version (19K):
[in this window]
[in a new window]
|
FIG. 6. Gene targeting. (A) When yeast cells are transformed with a replicating plasmid (ARS plasmid) containing a double-strand break or gap within a region of the plasmid with homology to the yeast genome, the break or gap is repaired by gene conversion from the chromosomal locus. The plasmid remains episomal if gene conversion without an associated crossover occurs but is integrated into the genome if conversion is associated with crossing over. If the plasmid contains a CEN sequence, only conversion events can give rise to viable transformants. If the plasmid contains no origin of replication (ARS), only integration events are observed. (B) Ends-out gene targeting refers to replacement of chromosomal sequences with sequences present on a linear DNA fragment introduced into cells by transformation. Gene targeting is thought to occur by invasion of the two ends into the chromosomal locus followed by resolution of the resulting Holliday junctions. Extensive DNA synthesis could be primed from the invading 3' ends prior to Holliday junction resolution, or resolution could occur by replication to the end of the chromosome. Gene targeting could also result from integration of a single strand of the targeting fragment followed by trimming the D-loop and mismatch repair.
|
|
Genetic and Physical Assays for Meiotic Recombination
Analysis of the segregation of heterozygous markers during meiosis is used to measure recombination in sporulation-proficient strains. Gene conversion is detected as a departure from the normal 2:2 segregation of heterozygous markers and crossing over by new linkage arrangements between linked heterozygous markers. However, this method is not very useful for strains that fail to sporulate or that give rise to inviable spores. Diploid cells induced to undergo sporulation can return to vegetative growth when plated on rich medium (87, 334). The time at which cells commit to high levels of recombination precedes the time when they commit to completion of the sporulation program; therefore, when diploids are plated on rich medium after several hours of growth in sporulation medium, they experience high levels of recombination but retain the diploid state. The frequency of heteroallelic recombination increases 100- to 1,000-fold when cells initiate meiotic recombination and are returned to vegetative growth. Because the return-to-growth protocol does not rely on the formation of viable spores, it can be used to measure the induction of meiotic recombination in sporulation-defective mutants.
The development of physical assays to monitor the kinetics of formation and processing of DNA intermediates has been invaluable for understanding the mechanisms of recombination (Fig. 7). Also, physical methods can be used to detect recombination intermediates in strains that fail to produce viable spores. In wild-type cells, DSBs are formed about 1 h after DNA synthesis at recombination initiation sites (defined by genetic methods) (46, 369). The ends are processed to form 3' single-stranded tails of about 600 nucleotides that are the substrates for strand invasion to form heteroduplex DNA (370). Heteroduplex DNA can be detected when the mismatches formed are poorly recognized by the mismatch repair system and therefore persist in DNA (198, 255). Restriction site polymorphisms flanking an initiation site can be used to detect branched intermediates by two-dimensional gel electrophoresis and the formation of crossover products (36, 46, 61, 328). Two types of branched-DNA intermediates have been detected during meiosis: SEIs and dHJ intermediates (11, 144). Kinetic studies suggest that SEIs arise before dHJ intermediates and could be precursors to them or could be processed independently to form noncrossover products (11, 144). The formation of crossover products detected by novel restriction fragments occurs almost simultaneously with the meiosis I division, suggesting a tight coupling between resolution of recombination intermediates, exit from pachytene, and chromosome segregation.
View larger version (26K):
[in this window]
[in a new window]
|
FIG. 7. Physical assays for meiotic recombination. (A) The hot spot created by insertion of the LEU2 gene at HIS4 in the SK1 strain background is shown. The end of the LEU2 gene contains a site for meiosis-specific DSBs flanked by restriction sites that can be used to distinguish between DSB fragments, and the two parental and two recombinant fragments. The white and black boxes indicate insertions of heterologous sequences used to distinguish the parental molecules by using strand-specific probes. (B) Three types of joint molecules are expected: Mom-Mom intersister joint molecules, Dad-Dad intersister joint molecules, and Mom-Dad interhomologue joint molecules. Size and hybridization to strand specific probes distinguish the three types of joint molecules. (C) Cartoon of a neutral-neutral two-dimensional gel showing separation of joint molecules from the parental and recombinant fragments. The interhomologue joint molecules are more abundant than the intersister joint molecules in wild-type cells. (Adapted from Fig. 1 of reference 329 with permission)
|
|
|
GENETIC AND BIOCHEMICAL PROPERTIES OF THE RAD52 GROUP GENES AND PROTEINS
|
|---|
The RAD52 group genes can be broadly grouped into the MRE11, RAD50, XRS2 (NBS1) subgroup and the RAD51, RAD52, RAD54, RAD55, RAD57, RAD59, RDH54/TID1 subgroup. MRE11, RAD50, and XRS2 are implicated in the formation and processing of DSBs during meiotic recombination and also function in the end-joining pathway of repair, telomere maintenance, in DNA replication-associated repair, and in the DNA damage checkpoint in mitotic cells. The RAD51 subgroup appears to function only in homologous recombination and is described separately.
Mre11-Rad50-Xrs2 (Nbs1) Complex
Yeast strains with null mutations of MRE11, RAD50, or XRS2 have very similar phenotypes. All of the mutants show poor vegetative growth, high sensitivity to IR, and defects in meiosis (7, 110, 149). The three proteins interact in the two-hybrid system, and coimmunoprecipitation studies have confirmed that they form a stable complex (160, 407). Although all three proteins can be coimmunoprecipitated from wild-type cells with antibodies directed against any one of the components, Rad50 and Xrs2 fail to interact in the absence of Mre11, suggesting a central role for Mre11 in complex formation (407). The Mre11 and Rad50 proteins are conserved in all kingdoms of life, whereas Xrs2 appears to be weakly conserved and is present only in eukaryotes. The human gene, HsMRE11, was fortuitously recovered in a two-hybrid screen for DNA ligase I-interacting proteins (282). Using antibodies against HsMre11, Petrini and coworkers identified a large protein complex containing HsRad50 and a 95-kDa protein (78). The 95-kDa protein was subsequently shown to be mutated in humans with the rare autosomal recessive trait NBS (47, 411). Consequently, p95 is referred to as Nbs1 or Nibrin. Nbs1 is considered functionally analagous to Xrs2. Hypomorphic alleles of MRE11 are found in persons with a different human chromosomal instability syndrome, A-TLD (358). At the cellular level, ataxia-telangiectasia (A-T), A-TLD, and NBS are very similar and are characterized by sensitivity to IR, radioresistant-DNA synthesis, and chromosome instability (primarily translocations) (280).
Although mre11, rad50, and xrs2 null mutants of yeast are viable, MRE11, RAD50, and NBS1 are all essential for the viability of vertebrate cell lines and for mouse early embryonic development (215, 426, 428, 431). An mre11 conditional chicken cell line was generated by deleting both copies of MRE11 in the DT40 cell line, in the presence of a transgene expressing MRE11 from a tetracycline-repressible promoter (428). Down regulation of the gene resulted in rapid depletion of Mre11 concomitant with increased radiosensitivity, increased levels of spontaneous chromosome breaks, arrest in the G2 phase of the cell cycle, and eventual cell death. Similarly, antibody depletion of Mre11 from Xenopus extracts resulted in aberrant DNA synthesis accompanied by the formation of DSBs (67). These studies suggest an essential role of the Mre11 complex in the repair of lesions generated during S phase in vertebrates.
Localization of the MRN complex to DSBs in vivo.
Immunofluorescence has been used to localize the Mre11-Rad50-Nbs1 (MRN) complex in response to DNA damage in vertebrate cells. In unirradiated cells, Mre11 and Rad50 are uniformly distributed throughout the nucleus, but after exposure to IR, both proteins form discrete nuclear foci (226). Foci that are microscopically visible are thought to represent sites of repair. Discrete foci were not detected after UV irradiation, suggesting that Mre11 and Rad50 associate specifically with DSBs. Formation of Mre11 and Rad50 IR-induced foci was eliminated in cells from NBS and A-TLD patients, consistent with a defect in the DNA damage response (47, 358). In a highly innovative study, soft X rays were used to form discrete areas of DNA damage within the nuclei of fibroblasts and areas of repair localized by incorporation of bromodeoxyuridine (BrdU) (assayed using an anti-bromodeoxyuridine monoclonal antibody) by terminal deoxynucleotidyltransferase (258). This study suggests that DSBs are held in a relatively fixed position during the early stages of DNA repair. At 30 min after irradiation, Mre11 colocalized with BrdU, suggesting that the MRN complex migrates to the sites of DSBs within the nucleus. Discrete areas of DNA damage have also been generated using a laser scissor method (279). The phosphorylated form of H2AX (
-H2AX), a nonessential member of the histone H2A family (48), associated with the sites of damage within 30 min, and HsRad50 was found to colocalize extensively with -H2AX. Treatment of cells with IR also resulted in the formation of -H2AX foci, but in this case Rad50 did not colocalize until 6 to 8 h after irradiation. Preirradiation exposure of cells to the fungal metabolite wortmannin, which inhibits phosphatidylinositol 3-kinases, prevented the formation of -H2AX and Rad50 foci, suggesting that -H2AX is required to recruit Rad50 to sites of damage (279).
Mre11 is associated with chromatin in replicating Xenopus extracts and appears to be important either for prevention of replication-induced breaks or for their repair (67). Immunofluorescence studies of detergent extracted cells also show association of Mre11 with sites of replication based on colocalization of Mre11 with proliferating-cell nuclear antigen during S phase (225). The Mre11 complex could play an important role during replication or could be associated with the fork in preparation for damage signaling or repair.
Structural and biochemical studies.
The 83-kDa Mre11 protein is highly conserved among eukaryotes, and the N-terminal region has several sequence motifs shared by a large family of phosphodiesterases, including the Escherichia coli SbcD and bacteriophage T4 gp46 nucleases and protein phosphatases (333) (Fig. 8). Yeast and human Mre11 proteins have single-stranded DNA (ssDNA) endonuclease and weak 3'-to-5' exonuclease activities (108, 242, 277, 397, 407). Like SbcD (62, 63), the Mre11 nuclease activities are dependent on manganese as a cofactor. An allele of MRE11, mre11-58 (rad58), originally thought to represent a new gene in the RAD52 epistasis group, has a point mutation within the fourth phosphodiesterase motif (H213Y), and the encoded protein is devoid of nuclease activity in vitro (400, 407). Several other mutations have been made at conserved residues within the other phosphodiesterase motifs (mre11-D16A, mre11-D56N, and mre11-H125N), and all abolish the endonuclease and 3'-to-5' exonuclease activities in vitro (108, 241). The aspartic acid residues at positions 16 and 56 are directly involved in coordination of two Mn2+ ions at the active site, and His125 stabilizes the transition state intermediate (139). Two-hybrid and size exclusion chromatography analyses indicate that Mre11 forms a dimer (108, 160). Consistent with these findings, diploid strains expressing an N-terminal mre11 mutation (mre11S) and a C-terminal insertion mutation showed intragenic complementation for meiosis and methyl methanesulfonate (MMS) resistance (256).
View larger version (18K):
[in this window]
[in a new window]
|
FIG. 8. Schematic representation of Mre11, Rad50, and Xrs2 (Nbs1). The phosphodiesterase motifs of Mre11 are labeled MI through MIV, and DNA binding sites are labeled DB site A and B. The Mre11-D16A, Mre11-D56N, Mre11-H125N, Mre11-H213Y, and Mre11-6 mutants are nuclease defective in vitro. Residue Pro84 is mutated in the mre11-S allele, and Pro162 is mutated in the mre11-1 temperature-sensitive allele. The mre11-N113S and mre11-Q623Z alleles correspond to the mutations in the A-TLD patients. Rad50 contains two coiled-coil domains separating the Walker A and B motifs for NTP binding and hydrolysis. The hook domain, containing the conserved CXXC motif, is located between the two coiled-coil domains. The positions of the rad50S alleles, rad50-R20M and rad50-K81I, are shown.
|
|
Rad50, like Mre11, is conserved in all kingdoms of life. In bacteria, the Rad50 homologue, SbcC, forms a complex with SbcD, the homologue of Mre11. The SbcCD complex has ATP-dependent 3'-to-5' exonuclease and ATP-independent single-stranded endonuclease activity (62, 63). The 150-kDa Rad50 protein is related to the SMC proteins, which have the Walker A and B motifs characteristic of nucleotide triphosphate (NTP)-binding proteins separated by a long coiled-coil region (9) (Fig. 8). The SMC proteins, including the Rad50 subgroup, have a conserved hinge region within the coiled region. The hinge region of the Rad50 subfamily is distinct from that of other SMC proteins and contains a conserved Cys-X-X-Cys motif (140). Electron microscopy studies of Rad50 suggest a dimeric structure to bring together the Walker A and B motifs, forming two catalytic sites. The dimer could result from two protomers in an antiparallel configuration or by hinge-mediated interactions between two intramolecularly coiled protomers (12, 139) (Fig. 9). Atomic force microscopy of the human Rad50-Mre11 complex identified two highly flexible intramolecular coiled coils emanating from the globular Rad50-Mre11 DNA binding domain (73). The intramolecular coiled coil places the Walker A and B motifs of one Rad50 monomer together juxtaposed to Mre11. The size of the globular domain is consistent with a dimer of Mre11 and a dimer of the Rad50 ATPase domain. Dimerization of Rad50 is mediated by the CXXC motif within the hinge region (140). Structural analysis of the hinge region revealed a hook structure at the apex of the coiled coil, creating half of a composite metal binding site. Two hook regions coordinate a single Zn2+ ion with the two coiled-coil regions extending in almost opposite directions. This arrangement could potentially bridge sister chromatids because the distance between the head domains of the Mre112Rad502 complex is approximately equal to the distance between sister chromatids (1,200 Å) (140). Another class of SMC proteins, called cohesins, is known to mediate sister chromatid cohesion (128). The Mre11-Rad50 complex could function specifically in sister chromatid interactions during DSBR, as suggested by genetic studies. Electron microscopy studies have provided direct evidence for end binding by the human and yeast Mre11-Rad50 complex and bridging of different DNA molecules (53, 73).
View larger version (25K):
[in this window]
[in a new window]
|
FIG. 9. Models for the Rad50-Mre11 complex. A dimer of Rad50 could form by antiparallel intermolecular interaction to position the Walker A motif from one monomer next to the Walker B motif of the other. A dimer of Mre11 binds adjacent to the head-tail region of Rad50. Recent results are more consistent with an intramolecular interaction between the coiled-coil domains of Rad50, with the dimer held together by a dimer of Mre11. Interactions between the hinge domains could connect two dimers of Rad50 and two dimers of Mre11. The length of the Rad50 tetramer is consistent with the distance between sister chromatids in eukaryotes. The Mre11 and Rad50 head-tail domains are envisioned to interact with DNA.
|
|
Structural studies of the Pyrococcus furiosis Rad50 catalytic domain (Rad50cd) indicate a structure similar to the ATP binding cassette transporter family of ATPases (142). Binding of the Rad50cd to the ATP analog AMP-PNP induces a structural change to align the Walker A and B motifs. The ATP-bound form of the Rad50cd binds DNA more tightly, suggesting that ATP regulates DNA binding by conformational switching. Mutation of the conserved lysine residue within the Walker A box confers a null phenotype in yeast, indicating the importance of ATP binding and/or hydrolysis to function (8). The DNA binding activity of yeast Rad50 is stimulated by ATP, but no ATPase activity has been observed for the purified protein (301).
Rad50 stimulates the nuclease activities of yeast and human Mre11. Unlike the EcSbcC-SbcD complex and the PfMre11-Rad50 complex, the complex of HsMre11 with HsRad50, or HsMre11, HsRad50, and Nbs1 has no demonstrated ATP-dependent exonuclease activity (63, 141). ATP stimulates a weak DNA unwinding activity of the HsMre11-Rad50-Nbs1 complex, and hairpin cleavage has also been observed for the complex (278). However, the yeast Mre11-Rad50 complex cleaves hairpin structures in the absence of Xrs2 (396).
In mammals, Nbs1 (p95) appears to be the functional homologue of Xrs2 in that it is tightly associated with Mre11 and Rad50, but sequence similarity to Xrs2 is limited to the N-terminal 115 amino acids (47, 411). Xrs2 and Nbs1 have no obvious sequence motifs indicative of function. Nbs1 contains two domains in the N-terminal region that are found in DNA damage-responsive cell cycle checkpoint proteins (95). A forkhead-associated (FHA) domain, which is thought to be important for interactions between phosphorylated proteins, is found at the N terminus of Nbs1. The BRCT (breast cancer carboxy-terminal) domain, which is adjacent to the FHA domain in the N-terminal region of Nbs1, is found in a variety of proteins that participate in DNA damage-responsive cell cycle checkpoints. To date, the only biochemical activity assigned to Nbs1 is stimulation of the unwinding and hairpin cleavage activities of HsMre11 and HsRad50 (278).
Meiotic phenotype of mre11, rad50, and xrs2 mutants.
Diploid strains homozygous for mre11, rad50, or xrs2 fail to form meiosis-specific DSBs and thus are unable to initiate meiotic recombination (8, 160). This results in the formation of aneuploid (inviable) spores in some yeast strain backgrounds (e.g., SK1) or a complete block of the sporulation pathway in others (e.g., W303). The spo13 mutation causes cells to bypass the first meiotic division, and thus spo13 strains no longer require recombination for segregation of homologous chromosomes. spo13 suppresses the sporulation and spore viability defects of mutant strains unable to initiate DSB formation, including mre11, rad50, and xrs2 (7, 149, 222). Because of the failure to induce meiosis-specific DSBs, mre11, rad50, and xrs2 null mutants show no induction of meiotic heteroallelic recombination in the return-to-growth protocol or recombinants among viable dyads derived from spo13 meiosis (7, 149, 222). Several separation-of-function alleles of RAD50 have been identified that are proficient for DNA repair in mitotic cells but sporulation defective (8). The sporulation defect conferred by rad50S alleles cannot be suppressed by spo13 but can be suppressed by spo13 plus spo11 (defective for initiation of recombination). Meiotic DSBs are still formed in rad50S mutants, but the ends are not processed to generate 3' single-stranded tails and are stable rather than transient (8, 46, 370).
The role of MRE11, RAD50, and XRS2 in DSB formation is unclear. MRE11 is required for a meiosis-specific alteration of chromatin structure at recombination hot spots, and the chromatin alteration and DSB formation functions of Mre11 are eliminated by a deletion removing the C-terminal 49 residues (108, 266). This region of Mre11 contains one of the two DNA binding sites of Mre11 defined by in vitro studies (407). At least 10 genes are required for DSB formation; one of these, SPO11, is known to play a direct role since it encodes the catalytic subunit of the DSB-forming complex (170). Spo11 is an atypical type II topoisomerase that catalyzes DSB formation by a transesterification mechanism. MRE11 and RAD50 also appear to function after the formation of meiosis-specific DSBs as evidenced by several separation-of-function alleles that are still proficient for DSB formation but are defective in DSB processing. In rad50S mutants, Spo11 remains covalently associated with the 5' ends at break sites (170). Several alleles of MRE11 have been identified that confer a similar phenotype to rad50S in meiosis (mre11S, mre11-58, mre11-6, mre11-D16A, and mre11-H125N) (108, 241, 256, 400, 407). Although covalent attachment of Spo11 at break sites has not been demonstrated for all of these mutants, it is assumed to occur because they all result in the accumulation of unprocessed breaks during meiosis. mre11S was isolated in a screen for mutants that showed RAD50-dependent spore inviability (256). From the same genetic screen, the SAE2/COM1 gene was identified (295), as well as from a screen for SPO11-dependent spore inviability (235). Null mutations in SAE2/COM1 confer a phenotype very similar to that of rad50S and mre11S mutants; accumulation of unprocessed DSBs in meiosis.
Two models have been proposed for the removal of Spo11 from break sites. First, Spo11 could be removed while still covalently attached to the 5' strand by the endonucleolytic activity of the Mre11-Rad50-Xrs2 (MRX) (Sae2?) complex. Second, Mre11, Rad50, and Sae2 could be required for the reversal of the Spo11 transesterification reaction. The first model predicts that resection of the 5' strand is intrinsic to removal of Spo11, whereas the second model predicts that resection of the 5' strand occurs after Spo11 removal. Support for the first model comes from the observation that strains containing nuclease-defective alleles of MRE11 (mre11-D16A, mre11-H125N and mre11-58) have unprocessed DSBs.
Role of the MRX complex in mitotic recombination.
The complex phenotype of mre11, rad50, and xrs2 mutants is even more apparent in vegetative cells. Although it is generally assumed that yeast cells repair IR-induced DNA damage by homologous recombination, mre11, rad50, and xrs2 mutants show little or no defect in spontaneous mitotic recombination. Spontaneous heteroallelic recombination is elevated about 10-fold in mre11, rad50, and xrs2 diploids and still shows some induction by DNA-damaging agents, but not to the extent observed in wild-type cells (7, 149, 314). Because diploids in the G2 stage of the cell cycle preferentially repair lesions from a sister chromatid instead of a homologue, it has been suggested that MRE11, RAD50, and XRS2 are specifically involved in sister chromatid recombination (8, 42, 149). In support of this hypothesis, mre11 diploids are more resistant to irradiation than are haploids; G1 diploids show the same sensitivity to IR as do wild-type strains, and mutant G2 haploids and diploids both show high IR sensitivity (42). In a genetic assay designed specifically to detect sister chromatid recombination (94), mre11 mutants showed a slight decrease in the rate of recombination but not to the extent expected if MRE11 was essential for this process (42). rad52 mutants show a 50-fold reduction in sister chromatid recombination in the same assay system. The hyperrecombination phenotype exhibited by mre11, rad50, and xrs2 mutants for heteroallelic recombination in diploids could be due to channeling lesions from the normal sister chromatid repair pathway into interactions between homologues (8, 149).
Spontaneous deletion between direct repeats occurs at wild-type frequency in rad50 mutants (119, 233). However, the recombination products were not studied in sufficient detail to determine whether there were defects in sister chromatid events using this system. Using the ade2 inverted-repeat assay (Fig. 4), a threefold reduction in spontaneous recombination was reported for rad50 and xrs2 mutants, with the same distribution of events as found in the wild-type strain (298). Ectopic recombination between ura3 heteroalleles on chromosomes II and V was unaffected by a rad50 mutation in haploids but showed a threefold increase in rad50 homozygous diploids (357). As suggested above for heteroallelic recombination, the hyperrecombination phenotype could be due to channeling lesions from sisters to interchromosomal interactions.
Studies of HO-induced recombination have also revealed only modest defects in these mutants (151, 400). The most striking finding that emerged from studies of mating-type switching is the decreased extent of processing of the 5' strand at the HO-cut site in the null mutants (151, 362, 400). This result, in combination with the defect in processing of meiosis-specific breaks in certain mre11 and rad50 mutants, led to the suggestion that the MRX complex resects ends to produce 3' single-stranded tails. However, even in mre11, rad50, or xrs2 null mutants, processing does occur, and while there is a delay in mating-type switching, most cells are able to complete the process with fairly high efficiency. HO-induced SSA between chromosomal direct repeats is also delayed by 1 to 2 h in rad50 mutants, but cells are able to repair the break with only a twofold decrease in viability (362).
Role of the Mre11 nuclease in processing DSBs in mitotic cells.
The observation that HO-induced DSBs are processed more slowly in mre11, rad50, and xrs2 null mutants suggests that the MRX complex is directly involved in end processing or regulates the activity of a nuclease. Because the exonuclease activity of Mre11 is of the opposite polarity to that expected for resection of DSBs, the endonuclease activity is thought to be targeted to the 5' strand by an as yet unknown mechanism. The role of the Mre11 nuclease in resection has been investigated by generating nuclease-defective alleles of MRE11 for analysis of end processing in vivo. The phenotype of the mre11-58 (rad58) strain is due to mutation of a conserved residue in phosphodiesterase motif IV (H213Y) (400). The Mre11-58 mutant protein is proficient for DNA binding but lacks exonuclease and endonuclease activities in vitro and fails to interact with Rad50 and Xrs2 when immunoprecipitated from yeast extracts (407). The phenotype of the mre11-58 strain is very similar to that of the mre11 null mutant: high sensitivity to MMS, delayed kinetics of mating-type switching, and elevated rates of mitotic heteroallelic recombination (400). The Mre11-D56N and Mre11-H125N mutant proteins also lack nuclease activity (241), and Mre11-H125N retains interaction with Xrs2 and Rad50 (S. Moreau and L. S. Symington, unpublished data). These two mutations confer much less severe mitotic phenotypes than does the mre11
allele. Strains containing either the mre11-D56N or mre11-H125N allele show intermediate sensitivity to IR and normal levels of spontaneous mitotic recombination and plasmid gap repair, and the kinetics of mating-type switching are indistinguishable from those in the wild type (241, 379). These observations suggest that the Mre11 nuclease may not be involved in the resection of mitotic DSBs or is redundant with another nuclease. The more severe phenotype conferred by the mre11-58 allele could be a consequence of defective complex formation rather than the nuclease defect. Bressan et al. (43) generated mutations within all four phosphodiesterase motifs. The mre11-2 and mre11-4 alleles conferred phenotypes indistinguishable from those conferred by the null mutation for radiation sensitivity and spontaneous mitotic recombination. These two proteins both contain nonconservative substitutions of two amino acids. Both mutants failed to interact with Rad50 in the yeast two-hybrid system, suggesting that the severity of the mitotic DNA repair defect could be due to lack of the MRX complex rather than just loss of the Mre11 nuclease. The mre11-11 and mre11-3 alleles, containing substitutions within motifs I and III, respectively, conferred less severe phenotypes for sensitivity to IR than did the mre11 mutation, and both retained interaction with Rad50. Thus, the severity of the phenotype conferred by these alleles is directly correlated with the ability of the mutant proteins to form complexes. The Mre11-D16A protein lacks exo- and endonuclease activities but retains DNA binding; interaction with Rad50 and Xrs2 has not been determined (108). The mre11-D16A mutant shows intermediate sensitivity to MMS, intermediate-length telomeres, and normal rates of spontaneous mitotic recombination, but has not been tested for processing of HO-induced DSBs in vivo (108).
Because mre11-H125N strains are unable to process meiosis-specific DSBs but are proficient at repair of HO-induced breaks, we have suggested that the complex unwinds ends and that nucleases redundant with Mre11 can process free 5' ends but not ends bound by Spo11 (241). A redundant endonuclease would be expected to substitute for Mre11 in both mitotic and meiotic cells, suggesting that it is a 5'-to-3' exonuclease that processes ends in the absence of Mre11 in mitotic cells. Alternatively, the redundant activity could be an endonuclease that either is not expressed during meiosis or is excluded from the meiotic DSB-processing complex. The EXO1 gene, which encodes a 5'-to-3' exonuclease with a twofold preference for double-stranded over ssDNA (98), was found to suppress the mitotic DNA repair defect of mre11 strains when present at high copy number (195, 242, 399). This suppression was also observed for rad50 and xrs2 mutants, suggesting that EXO1 in more than one copy can bypass the requirement for the MRX complex in DNA repair. Furthermore, exo1 mre11 double mutants have a severe growth defect (only about 30% plating efficiency), higher sensitivity to IR and MMS, and a longer delay in mating-type switching compared with mre11 single mutants (242, 399). The exo1 null mutation alone confers no significant sensitivity to IR or mating-type switching defects, and the exo1 mre11-H125N strain has normal kinetics of mating-type switching (98, 242). This contrasts with the severe defect observed for the exo1 mre11 double mutant. We envision inefficient processing of duplex ends by Exo1 in the absence of the MRX complex, but the efficiency of processing is greater when EXO1 is present on a high-copy-number plasmid. In mre11-H125N cells there must be a single-strand-specific nuclease activity that can substitute for either the Mre11 or Exo1 nuclease activity but is unable to process duplex ends (Fig. 10).
View larger version (15K):
[in this window]
[in a new window]
|
FIG. 10. Models for DSB processing by the MRX complex and Exo1. In wild-type cells, ends are processed by unwinding of ends and endonucleolytic cleavage of the 5' strand by the Mre11 nuclease. Unwinding could be mediated by the weak unwinding activity of the Mre11 complex or by association with a DNA helicase. In the absence of the MRX complex, Exo1 inefficiently processes the ends. The M*RX complex in cells expressing the mre11-H125N allele is still able to unwind ends, and other nucleases remove the 5' single-stranded tails.
|
|
An alternative hypothesis to explain the normal processing of HO-induced breaks in the mre11-H125N strain is that the Mre11 nuclease plays no role in resection and the role of the MRX complex is to recruit the resection nuclease to break sites or to clean up ends for the resection nuclease. The mre11-H125N diploid is unable to process DSBs with Spo11 covalently bound to the 5' ends and does show sensitivity to high doses of IR. IR causes base and sugar damage in addition to strand breaks, and termini frequently have phosphate or phosphoglycolate groups. One attractive model is for the endonuclease activity of Mre11 to remove damaged nucleotides or protein-DNA covalent adducts (such as Spo11) from ends to provide the substrate for the resection nuclease. Further support for this model comes from recent studies suggesting that Mre11 removes the terminal protein of adenonvirus during infection (359). Adenoviruses use a protein-priming mechanism for replication, resulting in linear duplexes with a protein covalently bound to the 5' ends. Viruses in which the E4 region is deleted fail to package viruses due to concatemerization of viral genomes (413). Concatemerization of E4-deleted viral genomes was found to require DNA-PKcs, ligase IV, Mre11, and Nbs1, suggesting that concatemers are formed by the end-joining pathway (359). Concatemer formation was restored to the ATLD3 cell line by transfection with wild-type Hs MRE11 but not with the Hs mre11-3 nuclease-defective allele. Infection of cells with wild-type virus results in depletion of Mre11, and this is dependent on the E4 region. Together, these results suggest that one function of E4 is to promote the degradation of Mre11, thus preventing cleavage of the terminal protein from 5' ends and subsequent concatemerization by end joining.
Recent studies suggest the Mre11 nuclease is important for processing unusual DNA structures, such as hairpins. Insertion of a 323-bp quasipalindrome derived from human Alu elements into the yeast LYS2 gene stimulates the rate of ectopic recombination (with a different lys2 allele) 1,000-fold. This stimulation is dependent on MRE11, RAD50, and XRS2 (207). Interestingly, mre11-D56N, mre11-H125N, rad50S, and sae2 mutations also completely eliminate the hyperrecombination induced by the palindrome. The palindrome appears to be extruded to form a cruciform structure, which is cleaved at the base by an unknown activity to form two hairpin ends. These hairpins fail to be resolved in the mutant strains and are replicated, resulting in the formation of an inverted duplication. The failure of the mre11-D56N, mre11-H125N, rad50S, and sae2 mutants to resolve the hairpin intermediate directly implicates the nuclease activity of Mre11, the Rad50 function compromised by the K81I mutation, and Sae2 in cleaving these structures in vivo. rad32 (MRE11)- and rad50-dependent stimulation of mitotic recombination by a 160-bp palindrome is also observed in Schizosaccharomyces pombe (93).
An allele of SAE2 was isolated in a screen for mutants that aberrantly process DSBs within an inverted repeat (297). In the sae2 mutant, DSB-stimulated events occurred normally 46% of the time but the other 54% of the events were characterized by a duplication of part of the inverted repeat. These events were hypothesized to occur by break-induced replication through the inverted repeat followed by an end-joining event to form a palindrome structure. These aberrant events were not detected in the wild-type strain, suggesting that they do not occur, or that the palindrome is not stably maintained, in wild-type cells. The mre11-H125N and rad50S mutations conferred the same phenotype as did the sae2 mutation in this assay, again suggesting that the Mre11 nuclease and Sae2 resolve palindromes. The results of these two studies are consistent with the observation that palindromes are stabilized in sbcC and sbcD mutants of E. coli (49, 115).
The RAD27 gene encodes a flap endonuclease that removes RNA primers from Okazaki fragments during DNA synthesis. rad27 mutants are viable but depend on homologous recombination functions (RAD50, RAD51, RAD52, RAD54, RAD55, RAD57, RAD59, MRE11, and XRS2) for viability (378, 393). The sae2, mre11S, mre11-H125N, and rad50S mutations all cause death in a rad27 strain (72, 241). The Mre11 nuclease could be partially redundant with the Rad27 nuclease or could be required to process aberrant DNA structures that accumulate in rad27 mutants. An alternative explanation is that the large number of lesions generated in a rad27 mutant overloads the homologous recombination system so that mutants with subtle DNA repair defects are unable to repair all of them. The similarity in the phenotypes of mre11S, mre11-H125N, rad50S, and sae2 mutants suggests that the nuclease activity of Mre11 is absent in these mutants. This has clearly been shown for the Mre11-H125N protein, but the Mre11S protein retains endo- and exonuclease activities (E. A. Morgan and L. S. Symington, unpublished data). This suggests the possibility that the Mre11 nuclease is active in vivo only in the presence of Sae2 and when Rad50 is fully functional. An attractive model is for Sae2 to interact with the MRX complex to activate the Mre11 nuclease and for the Sae2-MRX complex to be disrupted by rad50S and mre11S mutations.
Role of the MRX complex in nonhomologous end joining.
The end-joining pathway of repair requires Ku70 and Ku80, encoded by the YKU70 (HDF1) and YKU80 (HDF2) genes, respectively, a specialized DNA ligase encoded by the DNL4 gene, and a ligase stimulatory factor, Lif1 (XRCC4 in mammals) (271). In mammals, the Ku heterodimer associates with a kinase (DNA-PKcs) to form the DNA-dependent protein kinase (DNA-PK), but to date a similar kinase has not been identified in yeast (83, 120). An additional factor, Lif2/Nej1, which interacts with Lif1 and is regulated by mating type, is required for end joining in yeast (104, 171, 267, 408). Defects in any of the components of this pathway, with the exception of MRE11, RAD50, and XRS2, do not cause IR sensitivity but do increase the IR sensitivity of a rad52 strain in stationary phase. This suggests that the homologous pathway is the primary means of repairing IR-induced damage and that end joining can be used as a backup pathway.
Several assays have been used to measure end joining in yeast. The transformation efficiency of autonomously replicating plasmid DNA that has been linearized with a restriction enzyme to produce cohesive ends is used to measure precise rejoining (38). In wild-type cells, ligation occurs with high fidelity, with no loss or gain of nucleotides at the junction most of the time (38). A similar assay to monitor the repair of chromosomal breaks measures cell survival in response to induction of EcoRI endonuclease in a strain containing a GAL1-regulated EcoRI gene (196). Imprecise end joining can be assayed by survival in response to HO endonuclease induction of a strain that cannot repair the break at the MAT locus by homologous recombination (either by deletion of the donor cassettes or by deletion of RAD52) (240). In all three assays, mre11, rad50, xrs2, and yku70 strains have similar phenotypes and appear to be epistatic (37, 240). Although mre11 and yku70 mutations cause a similar reduction in the efficiency of joining cohesive ends of plasmids, the types of products recovered are different. Repaired plasmids recovered from yku70 and yku80 strains have large deletions flanking the break site and rejoin through short sequence homologies, whereas plasmids recovered from mre11, rad50, and xrs2 strains show mainly faithful repair (37, 38). In the HO assay, faithful repair restores the HO-cut site, which can then be recut by HO. Survivors of continuous HO expression have small deletions or insertions at the HO cut site, which prevent further cutting by HO. The frequency of survivors is reduced in mre11, rad50, and xrs2 mutants, but the survivors have large deletions (240). Although it has been suggested that the Mre11 nuclease could function in processing ends for the end-joining pathway, the characterization of junctions produced in mre11 mutants in yeast is inconsistent with this hypothesis. Furthermore, the mre11-H125N strain is proficient for end joining and EXO1 present on a high-copy-number plasmid is unable to suppress the end-joining defect of mre11 strains (
Top
Next
and HMRa and the expressed MAT