Repair of single-stranded DNA nicks, gaps, and loops in mammalian cells.

We studied the ability of mammalian cells to repair single-stranded nicks, gaps, and loops in DNA duplexes. Heteroduplexes prepared from derivatives of the shuttle vector pSV2neo were introduced into monkey COS cells. After replication, the plasmids were recovered and used to transform Escherichia coli. Plasmid DNA from the recovered colonies was tested for repair at each of six different sites. We observed that mammalian cells are capable of repairing single-stranded gaps and free single-stranded ends most efficiently. Regions containing twin loops were recognized, and one of the loops was excised. Portions of the molecules containing small single loops were also repaired. Markers which were 58 nucleotides apart were corepaired with nearly 100% efficiency, while markers which were 1,000 nucleotides or more apart were never corepaired. The mechanisms involved in heteroduplex repair in mammalian cells seem to be similar to those involved in repairing DNA lesions caused by physical and chemical agents.     

Heteroduplex-induced mutagenesis in mammalian cells

We have shown previously that heteroduplexes containing single-stranded loops are repaired efficiently in monkey cells, but not always correctly: 2% of the repair products acquired mutations within a 350 base-pair target (Weiss, U. and Wilson, J.H., Proc. Natl. Acad. Sci. USA 87:1123-1126, 1987). The structures of the mutant genomes, which are described here, are consistent with an error-prone repair system. The spectrum of mutations includes about 25% point mutations and 75% rearrangements, which consist of deletions, duplications, and substitutions. The mutations are clustered in the vicinity of single-stranded loops in the original heteroduplex. The high frequency of mutation, their clustering, and the positions of rearrangement endpoints suggest that the mutations were generated during repair of the heteroduplexes.     

Specific hydrolysis of the cohesive ends of bacteriophage lambda DNA by three single strand-specific nucleases.

Procedures have been worked out for Aspergillus nuclease S1 and mung been nuclease to quantitatively cleave off both of the 12-nucleotide long, single-stranded cohesive ends of lambdaDNA. This cleavage is indicated by the almost complete elimination of the repair incorporation of radioactive nucleotides by DNA polymerase into the digested DNA. With S1 nuclease, cleavage was complete at 10 degrees as well as at 30 degrees. Under the conditions for quantitative cleavage of the single-stranded regions there was no digestion of the double-stranded lambdaDNA. The mung bean nuclease cleaved off the cohesive ends completely at 30 degrees but at 5 degrees, the cleavage was not complete even at high enzyme concentration. The nearest neighbor analysis of the repaired DNA indicates that at 5 degrees about four nucleotides remained undigested. The mung bean nuclease also introduced, under the conditions used, some nicks into double-stranded DNA as determined by the repair incorporation. The Escherichia coli exonuclease VII cleaved off part of the cohesive ends of lambdaDNA, leaving two nucleotides on each end as single-stranded tails.     

The large loop repair and mismatch repair pathways of Saccharomyces cerevisiae act on distinct substrates during meiosis

During meiotic recombination in the yeast Saccharomyces cerevisiae, heteroduplex DNA is formed when single-stranded DNAs from two homologs anneal as a consequence of strand invasion. If the two DNA strands differ in sequence, a mismatch will be generated. Mismatches in heteroduplex DNA are recognized and repaired efficiently by meiotic DNA mismatch repair systems. Components of two meiotic systems, mismatch repair (MMR) and large loop repair (LLR), have been identified previously, but the substrate range of these repair systems has never been defined. To determine the substrates for the MMR and LLR repair pathways, we constructed insertion mutations at HIS4 that form loops of varying sizes when complexed with wild-type HIS4 sequence during meiotic heteroduplex DNA formation. We compared the frequency of repair during meiosis in wild-type diploids and in diploids lacking components of either MMR or LLR. We find that the LLR pathway does not act on single-stranded DNA loops of <16 nucleotides in length. We also find that the MMR pathway can act on loops up to 17, but not >19, nucleotides in length, indicating that the two pathways overlap slightly in their substrate range during meiosis. Our data reveal differences in mitotic and meiotic MMR and LLR; these may be due to alterations in the functioning of each complex or result from subtle sequence context influences on repair of the various mismatches examined.     

Distinct mechanisms of nonhomologous end joining in the repair of site-directed chromosomal breaks with noncomplementary and complementary ends

DNA double-strand breaks (DSBs) are considered the most important type of DNA damage inflicted by ionizing radiation. The molecular mechanisms of DSB repair by nonhomologous end joining (NHEJ) have not been well studied in live mammalian cells, due in part to the lack of suitable chromosomal repair assays. We previously introduced a novel plasmid-based assay to monitor NHEJ of site-directed chromosomal I-SceI breaks. In the current study, we expanded the analysis of chromosomal NHEJ products in murine fibroblasts to focus on the error-prone rejoining of DSBs with noncomplementary ends, which may serve as a model for radiation damage repair. We found that noncomplementary ends were efficiently repaired using microhomologies of 1-2 nucleotides (nt) present in the single-stranded overhangs, thereby keeping repair-associated end degradation to a minimum (2-3 nt). Microhomology-mediated end joining was disrupted by Wortmannin, a known inhibitor of DNA-PKcs. However, Wortmannin did not significantly impair the proficiency of end joining. In contrast to noncomplementary ends, the rejoining of cohesive ends showed only a minor dependence on microhomologies but produced fivefold larger deletions than the repair of noncomplementary ends. Together, these data suggest the presence of several distinct NHEJ mechanisms in live cells, which are characterized by the degree of sequence deletion and microhomology use. Our NHEJ assay should prove a useful system to further elucidate the genetic determinants and molecular mechanisms of site-directed DSBs in living cells.     

Biased short tract repair of palindromic loop mismatches in mammalian cells.

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

In vivo gene repair of point and frameshift mutations directed by chimeric RNA/DNA oligonucleotides and modified single-stranded oligonucleotides

Synthetic oligonucleotides have been used to direct base exchange and gene repair in a variety of organisms. Among the most promising vectors is chimeric oligonucleotide (CO), a double-stranded, RNA–DNA hybrid molecule folded into a double hairpin conformation: by using the cell’s DNA repair machinery, the CO directs nucleotide exchange as episomal and chromosomal DNA. Systematic dissection of the CO revealed that the region of contiguous DNA bases was the active component in the repair process, especially when the single-stranded ends were protected against nuclease attack. Here, the utility of this vector is expanded into Saccharomyces cerevisiae. An episome containing a mutated fusion gene encoding hygromycin resistance and eGFP expression was used as the target for repair. Substitution, deletion and insertion mutations were corrected with different frequencies by the same modified single-stranded vector as judged by growth in the presence of hygromycin and eGFP expression. A substitution mutation was repaired the most efficiently followed by insertion and finally deletion mutants. A strand bias for gene repair was also observed; vectors designed to direct the repair of nucleotide on the non-transcribed (non-template) strand displayed a 5–10-fold higher level of activity. Expanding the length of the oligo-vector from 25 to 100 nucleotides increases targeting frequency up to a maximal level and then it decreases. These results, obtained in a genetically tractable organism, contribute to the elucidation of the mechanism of targeted gene repair.     

Rad52 and Ku bind to different DNA structures produced early in double-strand break repair

DNA double-strand breaks are repaired by one of two main pathways, non-homologous end joining or homologous recombination. A competition for binding to DNA ends by Ku and Rad52, proteins required for non-homologous end joining and homologous recombination, respectively, has been proposed to determine the choice of repair pathway. In order to test this idea directly, we compared Ku and human Rad52 binding to different DNA substrates. How ever, we found no evidence that these proteins would compete for binding to the same broken DNA ends. Ku bound preferentially to DNA with free ends. Under the same conditions, Rad52 did not bind preferentially to DNA ends. Using a series of defined substrates we showed that it is single-stranded DNA and not DNA ends that were preferentially bound by Rad52. In addition, Rad52 aggregated DNA, bringing different single-stranded DNAs in close proximity. This activity was independent of the presence of DNA ends and of the ability of the single-stranded sequences to form extensive base pairs. Based on these DNA binding characteristics it is unlikely that Rad52 and Ku compete as ‘gatekeepers’ of different DNA double-strand break repair pathways. Rather, they interact with different DNA substrates produced early in DNA double-strand break repair.     

Activation of DNA-dependent protein kinase by single-stranded DNA ends

DNA-dependent protein kinase (DNA-PK) is involved in joining DNA double-strand breaks induced by ionizing radiation or V(D)J recombination. The kinase is activated by DNA ends and composed of a DNA binding subunit, Ku, and a catalytic subunit, DNA-PK(CS). To define the DNA structure required for kinase activation, we synthesized a series of DNA molecules and tested their interactions with purified DNA-PK(CS). The addition of unpaired single strands to blunt DNA ends increased binding and activation of the kinase. When single-stranded loops were added to the DNA ends, binding was preserved, but kinase activation was severely reduced. Obstruction of DNA ends by streptavidin reduced both binding and activation of the kinase. Significantly, short single-stranded oligonucleotides of 3-10 bases were capable of activating DNA-PK(CS). Taken together, these data indicate that kinase activation involves a specific interaction with free single-stranded DNA ends. The structure of DNA-PK(CS) contains an open channel large enough for double-stranded DNA and an adjacent enclosed cavity with the dimensions of single-stranded DNA. The data presented here support a model in which duplex DNA binds to the open channel, and a single-stranded DNA end is inserted into the enclosed cavity to activate the kinase.     

Visualization of repair of double-strand breaks in the bacteriophage T7 genome without normal DNA replication

An in vitro system based on extracts of Escherichia coli infected with bacteriophage T7 is able to repair double-strand breaks in a T7 genome with efficiencies of 20% or more. To achieve this high repair efficiency it is necessary that the reaction mixtures contain molecules of donor DNA that bracket the double-strand break. Gaps as long as 1,600 nucleotides are repaired almost as efficiently as simple double-strand breaks. DNA synthesis was measured while repair was taking place. It was found that the amount of DNA synthesis associated with repair of a double-strand break was below the level of detection possible with this system. Furthermore, repair efficiencies were the same with or without normal levels of T7 DNA polymerase. However, the repair required the 5'-->3' exonuclease encoded by T7 gene 6. The high efficiency of DNA repair allowed visualization of the repaired product after in vitro repair, thereby assuring that the repair took place in vitro rather than during an in vivo growth step after packaging.     

Rejoining of DNA double-strand breaks as a function of overhang length

The ends of spontaneously occurring double-strand breaks (DSBs) may contain various lengths of single-stranded DNA, blocking lesions, and gaps and flaps generated by end annealing. To investigate the processing of such structures, we developed an assay in which annealed oligonucleotides are ligated onto the ends of a linearized plasmid which is then transformed into Saccharomyces cerevisiae. Reconstitution of a marker occurs only when the oligonucleotides are incorporated and repair is in frame, permitting rapid analysis of complex DSB ends. Here, we created DSBs with compatible overhangs of various lengths and asked which pathways are required for their precise repair. Three mechanisms of rejoining were observed, regardless of overhang polarity: nonhomologous end joining (NHEJ), a Rad52-dependent single-strand annealing-like pathway, and a third mechanism independent of the first two mechanisms. DSBs with overhangs of less than 4 bases were mainly repaired by NHEJ. Repair became less dependent on NHEJ when the overhangs were longer or had a higher GC content. Repair of overhangs greater than 8 nucleotides was as much as 150-fold more efficient, impaired 10-fold by rad52 mutation, and highly accurate. Reducing the microhomology extent between long overhangs reduced their repair dramatically, to less than NHEJ of comparable short overhangs. These data support a model in which annealing energy is a primary determinant of the rejoining efficiency and mechanism.     

Substrate specificity of the DNA unwinding activity of the RecBC enzyme of Escherichia coli

The RecBC enzyme of Escherichia coli promotes genetic recombination of phage or bacterial chromosomes. The purified enzyme travels through duplex DNA, unwinding and rewinding the DNA with the transient production of potentially recombinogenic single-stranded DNA. The studies reported here are aimed at understanding which chromosomal forms allow the entry of RecBC enzyme and hence may undergo RecBC enzyme-mediated recombination. Circular duplex molecules, whether covalently closed, nicked or containing single-stranded gaps of 10 to 774 nucleotides, are not detectably unwound by RecBC enzyme. Linear duplex molecules are readily unwound if they have a nearly flush-ended terminus whose 5' and 3' ends are offset by no more than about 25 nucleotides; molecules with longer single-stranded tails are poorly bound by RecBC enzyme and are infrequently unwound. The single-strand endonuclease activity of RecBC enzyme can slowly cleave gapped circles to produce molecules presumably capable of being unwound. These results provide an enzymatic basis for the recombinogenicity of double-stranded DNA ends established from genetic studies of RecBC enzyme and Chi sites, recognition sites for RecBC enzyme-mediated DNA strand cleavage.     

DNA template requirements for human mismatch repair in vitro

The human mismatch repair pathway is competent to correct DNA mismatches in a strand-specific manner. At present, only nicks are known to support strand discrimination, although the DNA end within the active site of replication is often proposed to serve this role. We therefore tested the competence of DNA ends or gaps to direct mismatch correction. Eight G.T templates were constructed which contained a nick or gap of 4, 28, or approximately 200 nucleotides situated approximately 330 bp away in either orientation. A competition was established in which the mismatch repair machinery had to compete with gap-filling replication and ligation activities for access to the strand discontinuity. Gaps of 4 or 28 nucleotides were the most effective strand discrimination signals for mismatch repair, whereas double strand breaks did not direct repair to either strand. To define the minimal spatial requirements for access to either the strand signal or mismatch site, the nicked templates were linearized close to either site and assayed. As few as 14 bp beyond the nick supported mismatch excision, although repair synthesis failed using 5'-nicked templates. Finally, asymmetric G.T templates with a remote nick and a nearby DNA end were repaired efficiently.     

Heteroduplex DNA correction in Saccharomyces cerevisiae is mismatch specific and requires functional PMS genes.

In vitro-constructed heteroduplex DNAs with defined mismatches were corrected in Saccharomyces cerevisiae cells with efficiencies that were dependent on the mismatch. Single-nucleotide loops were repaired very efficiently; the base/base mismatches G/T, A/C, G/G, A/G, G/A, A/A, T/T, T/C, and C/T were repaired with a high to intermediate efficiency. The mismatch C/C and a 38-nucleotide loop were corrected with low efficiency. This substrate specificity pattern resembles that found in Escherichia coli and Streptococcus pneumoniae, suggesting an evolutionary relationship of DNA mismatch repair in pro- and eucaryotes. Repair of the listed mismatches was severely impaired in the putative S. cerevisiae DNA mismatch repair mutants pms1 and pms2. Low-efficiency repair also characterized pms3 strains, except that correction of single-nucleotide loops occurred with an efficiency close to that of PMS wild-type strains. A close correlation was found between the repair efficiencies determined in this study and the observed postmeiotic segregation frequencies of alleles with known DNA sequence. This suggests an involvement of DNA mismatch repair in recombination and gene conversion in S. cerevisiae.     

Repair bias of large loop mismatches during recombination in mammalian cells depends on loop length and structure

Repair of loop mismatches was investigated in wild-type and mismatch binding-defective Chinese hamster ovary (CHO) cells. Loop mismatches were formed in vivo during extrachromosomal recombination between heteroallelic plasmid substrates. Recombination was expected to occur primarily by single-strand annealing (SSA), yielding 12- or 26-base nonpalindromic loop mismatches, and 12-, 26-, or 40-base palindromic loop mismatches. Nonpalindromic loops were repaired efficiently and with bias toward loop loss. In contrast, the 12-base palindromic loop was repaired with bias toward loop retention, indicating that repair bias depends on loop structure. Among the palindromic loops, repair bias was dependent on loop length, with bias shifting from loop retention to loop loss with increasing loop size. For both palindromic and nonpalindromic loops, repair efficiencies and biases were independent of the general (MSH/MLH) mismatch repair pathway. These results are discussed with respect to the maintenance of large nonpalindromic insertions, and of small and large palindromes, in eukaryotic genomes.     

Action of a mammalian AP-endonuclease on DNAs of defined sequences.

An apurinic/apyrimidinic (AP) specific endonuclease from mouse plasmacytoma cells (line MPC-11), was observed to cleave apurinic sites in oligonucleotides 9, 11, 12, 39 and 40 nucleotides in length. However, the enzyme failed to cleave AP-sites in two oligonucleotides 7 nucleotides in length. The maximum rates of digestion observed on these short single-stranded DNA (ssDNA) fragments were approximately 1/30 of the rates observed on double-stranded DNA (dsDNA). In studies using the Maxam-Gilbert DNA sequencing analysis, apurinic sites in purine-rich regions were preferentially cleaved in dsDNA but not in ssDNA, indicating that the enzyme has a sequence preference on dsDNA. These results suggest that some sites on DNA might be more efficiently repaired than others.     

A quantitative study of the flexibility contributed to RNA structures by nicks and single-stranded gaps.

Disulfide crosslinking via thiol-disulfide interchange was applied to quantitate the relative flexibility contributed by nicks and single-stranded gaps in an RNA structure. An RNA duplex comprised of three strands was constructed containing the disulfide crosslink precursors 1 and 2 at opposite ends of the duplex on opposite strands. The third strand was of varying length to yield a nick or single-stranded gaps of 1, 2, or 3 nt. Crosslinking rates Indicated relative flexibilities of the resulting two-helix junctions. Crosslinking in the nicked duplex occurred two orders of magnitude slower than in a duplex containing a 3-nt gap. Rates of crosslinking in duplexes with 3-and 2-nt gaps showed only modest dependence on the gap sequence. Many natural RNAs, including ribozymes, contain two-helix junctions related to the model system described here. The data suggest that two-helix junctions containing a nick in one strand will retain substantial rigidity, whereas one or more single-stranded nucleotides at a two-helix junction allow significant flexibility.     

Multiple pathways for repair of DNA double-strand breaks in mammalian chromosomes

To study repair of DNA double-strand breaks (DSBs) in mammalian chromosomes, we designed DNA substrates containing a thymidine kinase (TK) gene disrupted by the 18-bp recognition site for yeast endonuclease I-SceI. Some substrates also contained a second defective TK gene sequence to serve as a genetic donor in recombinational repair. A genomic DSB was induced by introducing endonuclease I-SceI into cells containing a stably integrated DNA substrate. DSB repair was monitored by selection for TK-positive segregants. We observed that intrachromosomal DSB repair is accomplished with nearly equal efficiencies in either the presence or absence of a homologous donor sequence. DSB repair is achieved by nonhomologous end-joining or homologous recombination, but rarely by nonconservative single-strand annealing. Repair of a chromosomal DSB by homologous recombination occurs mainly by gene conversion and appears to require a donor sequence greater than a few hundred base pairs in length. Nonhomologous end-joining events typically involve loss of very few nucleotides, and some events are associated with gene amplification at the repaired locus. Additional studies revealed that precise religation of DNA ends with no other concomitant sequence alteration is a viable mode for repair of DSBs in a mammalian genome.     

A nonuniform distribution of excision repair synthesis in nucleosome core DNA

We have investigated the distribution in nucleosome core DNA of nucleotides incorporated by excision repair synthesis occurring immediately after UV irradiation in human cells. We show that the differences previously observed for whole nuclei between the DNase I digestion profiles of repaired DNA (following its refolding into a nucleosome structure) and bulk DNA are obtained for isolated nucleosome core particles. Analysis of the differences obtained indicates that they could reflect a significant difference in the level of repair-incorporated nucleotides at different sites within the core DNA region. To test this possibility directly, we have used exonuclease III digestion of very homogeneous sized core particle DNA to "map" the distribution of repair synthesis in these regions. Our results indicate that in a significant fraction of the nucleosomes the 5' and 3' ends of the core DNA are markedly enhanced in repair-incorporated nucleotides relative to the central region of the core particle. A best fit analysis indicates that a good approximation of the data is obtained for a distribution where the core DNA is uniformly labeled from the 5' end to position 62 and from position 114 to the 3' end, with the 52-base central region being devoid of repair-incorporated nucleotides. This distribution accounts for all of the quantitative differences observed previously between repaired DNA and bulk DNA following the rapid phase of nucleosome rearrangement when it is assumed that linker DNA and the core DNA ends are repaired with equal efficiency and the nucleosome structure of newly repaired DNA is identical with that of bulk chromatin. Furthermore, the 52-base central region that is devoid of repair synthesis contains the lowest frequency cutting sites for DNase I in vitro, as well as the only "internal" locations where two (rather than one) histones interact with a 10-base segment of each DNA strand.