A role for FEN-1 in nonhomologous DNA end joining: the order of strand annealing and nucleolytic processing events

Eukaryotic repair of double-strand DNA breaks can occur either by homologous recombination or by nonhomologous DNA end joining (NHEJ). NHEJ relies on Ku70/86, XRCC4, DNA ligase IV, and DNA-dependent protein kinase. NHEJ involves a synapsis step in which the two ends are maintained in proximity, processing steps in which nucleases and polymerases act on the ends, an alignment step in which a few nucleotides of terminal homology guide the ends into preferred alignments, and a ligation step. Some of the steps, such as ligation, rely on a single enzymatic component. However, the processing steps begin and end with a wide array of alternative substrates and products, respectively, and likely involve multiple nucleases and polymerases. Given the alternative pathways that can be catalyzed by the remaining nucleases and polymerases, no one of these processing enzymes is likely to be essential. The only requirement for the processing enzymes, as a collective, is to generate a ligatable configuration, namely a ligatable nick on each strand. Here, we have tested the two major known 5'-specific nucleases of Saccharomyces cerevisiae for involvement in NHEJ. Whereas EXO1 does not appear to be involved to any detectable level, deleting RAD27 (FEN-1 of yeast) leads to a 4.4-fold reduction specifically of those NHEJ events predicted to proceed by means of 5' flap intermediates. Because Rad27/FEN-1 acts specifically at 5' flap structures, these results suggest that the NHEJ alignment step precedes nucleolytic processing steps in a significant fraction of NHEJ events.     

Activation of nuclear factor kappa B inflammatory bowel disease

Nonhomologous DNA end joining (NHEJ) is the major pathway for repairing double-strand DNA breaks. V(D)J recombination is a double-strand DNA breakage and rejoining process that relies on NHEJ for the joining steps. Here we show that the targeted disruption of both DNA ligase IV alleles in a human pre-B cell line renders the cells sensitive to ionizing radiation and ablates V(D)J recombination. This phenotype can only be reversed by complementation with DNA ligase IV but not by expression of either of the remaining two ligases, DNA ligase I or III. Hence, DNA ligase IV is the activity responsible for the ligation step in NHEJ and in V(D)J recombination.     

DNA ligase IV binds to XRCC4 via a motif located between rather than within its BRCT domains

The covalent rejoining of DNA ends at single-stranded or double-stranded DNA breaks is catalyzed by DNA ligases. Four DNA ligase activities (I-IV) have been identified in mammalian cells [1]. It has recently been demonstrated that DNA ligase IV interacts with and is catalytically stimulated by the XRCC4 protein [2,3], which is essential for DNA double-strand break repair and the genomic rearrangement process of V(D)J recombination [4]. Together with the finding that the yeast DNA ligase IV homologue is essential for nonhomologous DNA end joining [5-7], this has led to the hypothesis that mammalian DNA ligase IV catalyzes ligation steps in both of these processes [8]. DNA ligase IV is characterized by a unique carboxy-terminal tail comprising two BRCT (BRCA1 carboxyl terminus) domains. BRCT domains were initially identified in the breast cancer susceptibility protein BRCA1 [9], but are also found in other DNA repair proteins [10]. It has been suggested that DNA ligase IV associates with XRCC4 via its tandem BRCT domains and that this may be a general model for protein-protein interactions between DNA repair proteins [3]. We have performed a detailed deletional analysis of DNA ligase IV to define its XRCC4-binding domain and to characterize regions essential for its catalytic activity. We find that a region in the carboxy-terminal tail of DNA ligase IV located between rather than within BRCT domains is necessary and sufficient to confer binding to XRCC4. The catalytic activity of DNA ligase IV is affected by mutations within the first two-thirds of the protein including a 67 amino-acid amino-terminal region that was previously thought not to be present in human DNA ligase IV [11].     

Cell-free V(D)J recombination

V(D)J recombination generates diversity in the immune system through the lymphoid-specific assembly of multiple gene segments into functional immunoglobulin and T-cell receptor genes. The first step in V(D)J recombination is cleavage of DNA at recombination signal sequences. Cleavage produces a blunt DNA end on each signal sequence and a hairpin end on adjacent coding gene segments, and can be reproduced in vitro by using purified RAG and RAG2 proteins. The later steps involve processing and joining of the cleaved DNA ends, and until now have been studied only in cells. Here we reconstitute the complete V(D)J recombination reaction in a cell-free system. We find that the RAG proteins are not only involved in cleavage, but are also needed in the later steps for efficient joining of coding ends. Joining is largely directed by short pieces of identical sequence in the coding flanks, but addition of human DNA ligase I results in greater diversity. Coding junctions contain short deletions as well as additions complementary to a coding flank (P nucleotides). Addition of non-templated nucleotides into coding junctions is mediated by terminal deoxyribonucleotidyl transferase. The cell-free reaction can therefore reproduce the complete set of processing events that occur in cells.     

Yeast DNA ligase IV mediates non-homologous DNA end joining

The discovery of homologues from the yeast Saccharomyces cerevisiae of the human Ku DNA-end-binding proteins (HDF1 and KU80) has established that this organism is capable of non-homologous double-strand end joining (NHEJ), a form of DNA double-strand break repair (DSBR) active in mammalian V(D)J recombination. Identification of the DNA ligase that mediates NHEJ in yeast will help elucidate the function of the four mammalian DNA ligases in DSBR, V(D)J recombination and other reactions. Here we show that S. cerevisiae has two typical DNA ligases, the known DNA ligase I homologue CDC9 and the previously unknown DNA ligase IV homologue DNL4. dnl4 mutants are deficient in precise and end-processed NHEJ. DNL4 and HDF1 are epistatic in this regard, with the mutation of each having equivalent effects. dnl4 mutants are complemented by overexpression of Dnl4 but not of Cdc9, and deficiency of Dnl4 alone does not impair either cell growth or the Cdc9-mediated responses to ionizing and ultraviolet radiation. Thus, S. cerevisiae has two distinct and separate ligation pathways.     

Creation of libraries with long ORFs by polymerization of a microgene

We describe a novel method for constructing pools of DNA sequences that encode large proteins with molecular diversity. Sets of primer pairs that form 8 to 10 complementary base pairs in the 3' region and have double mismatch pairs at their 3'-OH ends were designed so that primer dimers recreated short stretches of DNA (microgenes) devoid of termination codons. Cycles of denaturation and elongation reactions with a pair of primers, four dNTPs, and 3'-5' exo+ thermostable DNA polymerase gave head-to-tail polymers of the primer dimer unit (microgene) whose sizes exceeded 12 kb. No template was required in this reaction, but mismatched nucleotides at 3'-OH ends of the primers were critical for efficient polymerization. At end-joining junctions of a microgene, nucleotide insertions and deletions randomly occurred, resulting in combinatorial libraries of three reading frames from a single microgene. Further molecular diversity could be incorporated by using a mixture of primers. The resultant polymers have long ORFs whose products have a repetitious nature that could facilitate the formation of higher structures of translated products. Thus, microgene polymers may be used as a source of libraries for in vitro protein evolution experiments. Ligation of a microgene is apparently related to the nonhomologous recombination of double-strand breaks in DNA that has been shown to be catalyzed by DNA polymerases. We named this polymerization reaction the "microgene polymerization reaction."     

In vitro repair of gaps in bacteriophage T7 DNA

An in vitro system based upon extracts of Escherichia coli infected with bacteriophage T7 was used to study the mechanism of double-strand break repair. Double-strand breaks were placed in T7 genomes by cutting with a restriction endonuclease which recognizes a unique site in the T7 genome. These molecules were allowed to repair under conditions where the double-strand break could be healed by (i) direct joining of the two partial genomes resulting from the break, (ii) annealing of complementary versions of 17-bp sequences repeated on either side of the break, or (iii) recombination with intact T7 DNA molecules. The data show that while direct joining and single-strand annealing contributed to repair of double-strand breaks, these mechanisms made only minor contributions. The efficiency of repair was greatly enhanced when DNA molecules that bridge the region of the double-strand break (referred to as donor DNA) were provided in the reaction mixtures. Moreover, in the presence of the donor DNA most of the repaired molecules acquired genetic markers from the donor DNA, implying that recombination between the DNA molecules was instrumental in repairing the break. Double-strand break repair in this system is highly efficient, with more than 50% of the broken molecules being repaired within 30 min under some experimental conditions. Gaps of 1,600 nucleotides were repaired nearly as well as simple double-strand breaks. Perfect homology between the DNA sequence near the break site and the donor DNA resulted in minor (twofold) improvement in the efficiency of repair. However, double-strand break repair was still highly efficient when there were inhomogeneities between the ends created by the double-strand break and the T7 genome or between the ends of the donor DNA molecules and the genome. The distance between the double-strand break and the ends of the donor DNA molecule was critical to the repair efficiency. The data argue that ends of DNA molecules formed by double-strand breaks are typically digested by between 150 and 500 nucleotides to form a gap that is subsequently repaired by recombination with other DNA molecules present in the same reaction mixture or infected cell.     

Distance and end configuration effects on VDJP-mediated DNA joining

We have previously reported the cloning of a protein, VDJP, that is capable of binding the nonamer element of the V(D)J Recombinational Signal Sequence (RSS) as well as joining linear DNA fragments containing RSS elements in vitro. We show here that the linearized DNA molecules must contain a 5' extension or blunt end in order to be joined by VDJP. DNAs with 3' extensions are not efficiently joined by VDJP. Furthermore, the joining activity of DNAs with 5' extensions is significantly increased as the distance between the end and the RSS decreases. It is not yet clear what role VDJP plays in vivo, because our assay may not mimic exactly the in vivo DNA intermediates.     

Failure of hairpin-ended and nicked DNA To activate DNA-dependent protein kinase: implications for V(D)J recombination

V(D)J recombination is initiated by a coordinated cleavage reaction that nicks DNA at two sites and then forms a hairpin coding end and blunt signal end at each site. Following cleavage, the DNA ends are joined by a process that is incompletely understood but nevertheless depends on DNA-dependent protein kinase (DNA-PK), which consists of Ku and a 460-kDa catalytic subunit (DNA-PKCS or p460). Ku directs DNA-PKCS to DNA ends to efficiently activate the kinase. In vivo, the mouse SCID mutation in DNA-PKCS disrupts joining of the hairpin coding ends but spares joining of the open signal ends. To better understand the mechanism of V(D)J recombination, we measured the activation of DNA-PK by the three DNA structures formed during the cleavage reaction: open ends, DNA nicks, and hairpin ends. Although open DNA ends strongly activated DNA-PK, nicked DNA substrates and hairpin-ended DNA did not. Therefore, even though efficient processing of hairpin coding ends requires DNA-PKCS, this may occur by activation of the kinase bound to the cogenerated open signal end rather than to the hairpin end itself.     

Double-strand break repair in Ku86- and XRCC4-deficient cells

The Ku86 and XRCC4 proteins perform critical but poorly understood functions in the repair of DNA double-strand breaks. Both Ku 86- and XRCC4-deficient cells exhibit profound radiosensitivity and severe defects in V(D)J recombination, including excessive deletions at recombinant junctions. Previous workers have suggested that these phenomena may reflect defects in joining of the broken DNA ends or in protection of the ends from nucleases. However, end joining in XRCC4-deficient cells has not been examined. Here we show that joining of both matched and mismatched DNA ends occurs efficiently in XRCC4-deficient cells. Furthermore, analysis of junctions shows that XRCC4 is not required to protect the ends from degradation. However, nucleotide sequence analysis of junctions derived from joining of mismatched DNA ends in XRCC4-deficient cells revealed a strong preference for a junction containing a 7 nt homology. Similar results were obtained in Ku86-deficient cells. These data suggest that in the absence of XRCC4 or Ku86, joining is assisted by base pairing interactions, supporting the hypothesis that these proteins may participate in aligning or stabilizing intermediates in end joining.     

Tuberculosis mortality in England and Wales during 1982-1992: its association with poverty, ethnicity and AIDS

A linear DNA with partial sequence redundancy can be recircularized in cells by either nonhomologous end joining (NEJ) or by homologous recombination (HR). We have studied the relative contributions of these processes in zygotes or early embryos of species that serve as model organisms for developmental genetics. Thus, we have microinjected a linearized plasmid substrate into zygotes of zebrafish (Danio rerio) or into the posterior end of Drosophila melanogaster early embryos before pole cell formation. Similar to the situation observed previously in Xenopus zygotes/early embryos, we detected a large preponderance of DNA-end joining over homologous recombination. A comparison of end-joined junctions revealed that from the three species tested, zebrafish introduced the least number of sequence distortions upon DNA-end joining, while Drosophila produced the largest deletions (average 14 bp) with occasional nucleotide patch insertions, reminiscent of the N nucleotides at V(D)J junctions in mammalian immune receptor genes. Double-strand gap repair by homologous sequences ('homologous recombination') involving a bimolecular reaction was readily detectable in both zebrafish and Drosophila. This involved specifically designed recombination substrates consisting of a mutagenized linear plasmid and DNA fragments carrying the wild-type sequence. Our results show that the basic machinery for homologous recombination is present at early developmental stages of these two genetic model organisms. However, it seems that for any experimental exploitation, such as targeted gene disruption, one would have to inhibit or bypass the overwhelming DNA-end joining activity.     

Maximum pulsed electromagnetic field limits based on peripheral nerve stimulation: application to IEEE/ANSI C95.1 electromagnetic field standards

DNA ligase IV in a complex with XRCC4 is responsible for DNA end-joining in repair of DNA double-strand breaks (DSB) and V(D)J recombination. We found that non-histone chromosomal high mobility group (HMG) proteins 1 and 2 enhanced the ligation of linearized pUC119 DNA with DNA ligase IV from rat liver nuclear extract. Intra-molecular and inter-molecular ligations of cohesive-ended and blunt-ended DNA were markedly stimulated by HMG1 and 2. Recombinant HMG2-domain A, B, and (A + B) polypeptides were similarly, but non-identically, effective for the stimulation of DSB ligation reaction. Ligation of single-strand breaks (nicks) was only slightly activated by the HMG proteins. The DNA end-binding Ku protein singly or in combination with the catalytic component of DNA-dependent protein kinase (DNA-PK) as the DNA-PK holoenzyme was ineffective for the ligation of linearized pUC119 DNA. Although the stimulatory effect of HMG1 and 2 on ligation of DSB in vitro was not specific to DNA ligase IV, these results suggest that HMG1 and 2 are involved in the final ligation step in DNA end-joining processes of DSB repair and V(D)J recombination.     

Presence of double-strand breaks with single-base 3' overhangs in cells undergoing apoptosis but not necrosis

Apoptotic cells in rat thymus were labeled in situ in paraffin-embedded and frozen tissue sections by ligation of double-stranded DNA fragments containing digoxigenin or Texas red. Two forms of double-stranded DNA fragments were prepared using the polymerase chain reaction: one was synthesized using Taq polymerase, which yields products with single-base 3' overhangs, and one using Pfu polymerase, which produces blunt-ended products. Both types of fragment could be ligated to apoptotic nuclei in thymus, indicating the presence in such nuclei of DNA double-strand breaks with single-base 3' overhangs as well as blunt ends. However, in nuclei with DNA damage resulting from a variety of nonapoptotic processes (necrosis, in vitro autolysis, peroxide damage, and heating) single-base 3' overhangs were either nondetectable or present at much lower concentrations than in apoptotic cells. Blunt DNA ends were present in such tissues, but at lower concentrations than in apoptotic cells. In contrast, in all of these forms of DNA damage, nuclei contained abundant 3'-hydroxyls accessible to labeling with terminal deoxynucleotidyl transferase. Thus, although single-base 3' overhangs and blunt ends are present in apoptotic nuclei, the specificity of the in situ ligation of 3'-overhang fragments to apoptotic nuclei indicates that apoptotic cells labeled in this way can readily be distinguished from cells with nonapoptotic DNA damage. These data are consistent with the involvement of an endonuclease similar to DNase I in apoptosis, which is predicted to leave short 3' overhangs as well as blunt ends in digestion of chromatin.     

The Tn7 transposase is a heteromeric complex in which DNA breakage and joining activities are distributed between different gene products

The bacterial transposon Tn7 translocates by a cut and paste mechanism: excision from the donor site results from double-strand breaks at each end of Tn7 and target insertion results from joining of the exposed 3' Tn7 tips to the target DNA. Through site-directed mutagenesis of the Tn7-encoded transposition proteins TnsA and TnsB, we demonstrate that the Tn7 transposase is a heteromeric complex of these proteins, each protein executing different DNA processing reactions. TnsA mediates DNA cleavage reactions at the 5' ends of Tn7, and TnsB mediates DNA breakage and joining reactions at the 3' ends of Tn7. Thus the double-strand breaks that underlie Tn7 excision result from a collaboration between two active sites, one in TnsA and one in TnsB; the same (or a closely related) active site in TnsB also mediates the subsequent joining of the 3' ends to the target. Both TnsA and TnsB appear to be members of the retroviral integrase superfamily: mutation of their putative DD(35)E motifs blocks catalytic activity. Recombinases of this class require a divalent metal cofactor that is thought to interact with these acidic residues. Through analysis of the metal ion specificity of a TnsA mutant containing a sulfur (cysteine) substitution, we provide evidence that a divalent metal actually interacts with these acidic amino acids.     

Switching from cut-and-paste to replicative Tn7 transposition

The bacterial transposon Tn7 usually moves through a cut-and-paste mechanism whereby the transposon is excised from a donor site and joined to a target site to form a simple insertion. The transposon was converted to a replicative element that generated plasmid fusions in vitro and cointegrate products in vivo. This switch was a consequence of the separation of 5'- and 3'-end processing reactions of Tn7 transposition as demonstrated by the consequences of a single amino acid alteration in an element-encoded protein essential for normal cut-and-paste transposition. The mutation specifically blocked cleavage of the 5' strand at each transposon end without disturbing the breakage and joining on the 3' strand, producing a fusion (the Shapiro Intermediate) that resulted in replicative transposition. The ability of Tn7 recombination products to serve as substrates for both the limited gap repair required to complete cut-and-paste transposition and the extensive DNA replication involved in cointegrate formation suggests a remarkable plasticity in Tn7's recruitment of host repair and replication functions.     

3'-end processing and kinetics of 5'-end joining during retroviral integration in vivo

Retroviral replication depends on integration of viral DNA into a host cell chromosome. Integration proceeds in three steps: 3'-end processing, the endonucleolytic removal of the two terminal nucleotides from each 3' end of the viral DNA; strand transfer, the joining of the 3' ends of viral DNA to host DNA; and 5'-end joining (or gap repair), the joining of the 5' ends of viral DNA to host DNA. The 5'-end joining step has never been investigated, either for retroviral integration or for any other transposition process. We have developed an assay for 5'-end joining in vivo and have examined the kinetics of 5'-end joining for Moloney murine leukemia virus (MLV). The interval between 3'-end and 5'-end joining is estimated to be less than 1 h. This assay will be a useful tool for examining whether viral or host components mediate 5'-end joining. MLV integrates its DNA only after its host cell has completed mitosis. We show that the extent of 3'-end processing is the same in unsynchronized and aphidicolin-arrested cells. 3'-end processing therefore does not depend on mitosis.     

Intravitreal sustained-release ganciclovir implantation to control cytomegalovirus retinitis in AIDS

Immunoglobulin (Ig) heavy chain class switch recombination occurs mainly by joining two switch (S) regions, segments of tandemly repeated DNA sequences that lie upstream of heavy chain constant region genes. The products of this recombination event are a chromosomal DNA joint and a 'looped-out' circular DNA joint. Although a previous study showed that 40% of chromosomal joints in the mu gene switch region (S mu) are found in the flanking regions of S mu, which do not contain typical S mu region repeats [1], other studies revealed that almost all recombination sites on looped-out circular DNA are found within S regions [2-4]. To resolve this discrepancy, we have isolated and sequenced 164 DNA fragments containing recombination joints from both chromosomal and looped-out DNA of a single cell line, the murine B lymphoma line CH12F3, which switches from IgM to IgA production with a high frequency upon cytokine stimulation [5]. The recombination sites were distributed almost evenly in the S mu region and its flanking regions, suggesting that the final joining of DNA ends may not necessarily take place in S regions. In contrast, there were few joining sites in the exon located 5' of the switch region (the I mu exon), suggesting that the 3' end of the I mu exon might be the upstream border of the recombination joint.     

Saccharomyces cerevisiae LIF1: a function involved in DNA double-strand break repair related to mammalian XRCC4

Saccharomyces cerevisiae DNA ligase IV (LIG4) has been shown previously to be involved in non-homologous DNA end joining and meiosis. The homologous mammalian DNA ligase IV interacts with XRCC4, a protein implicated in V(D)J recombination and double-strand break repair. Here, we report the discovery of LIF1, a S.cerevisiae protein that strongly interacts with the C-terminal BRCT domain of yeast LIG4. LIG4 and LIF1 apparently occur as a heterodimer in vivo. LIF1 shares limited sequence homology with mammalian XRCC4. Disruption of the LIF1 gene abolishes the capacity of cells to recircularize transformed linearized plasmids correctly by non-homologous DNA end joining. Loss of LIF1 is also associated with conditional hypersensitivity of cells to ionizing irradiation and with reduced sporulation efficiency. Thus, with respect to their phenotype, lif1 strains are similar to the previously described lig4 mutants. One function of LIF1 is the stabilization of the LIG4 enzyme. The finding of a XRCC4 homologue in S.cerevisiae now allows for mutational analyses of structure-function relationships in XRCC4-like proteins to define their role in DNA double-strand break repair.     

Branch migration during Rad51-promoted strand exchange proceeds in either direction

The Saccharomyces cerevisiae Rad51 protein is important for genetic recombination and repair of DNA double-strand breaks in vivo and can promote strand exchange between linear double-stranded DNA and circular single-stranded DNA in vitro. However, unlike Escherichia coli RecA, Rad51 requires an overhanging complementary 3' or 5' end to initiate strand exchange; given that fact, we previously surmised that the fully exchanged molecules resulted from branch migration in either direction depending on which type of end initiated the joint molecule. Our present experiments confirm that branch migration proceeds in either direction, the polarity depending on whether a 3' or 5' end initiates the joint molecules. Furthermore, heteroduplex DNA is formed rapidly, first at the overhanging end of the linear double-stranded DNA's complementary strand and then more slowly by progressive lengthening of the heteroduplex region until strand exchange is complete. Although joint molecule formation occurs equally efficiently when initiated with a 3' or 5' overhanging end, branch migration proceeds more rapidly when it is initiated by an overhanging 3' end, i.e., in the 5' to 3' direction with respect to the single-stranded DNA.     

A 5'-3' exonuclease activity involved in forming the 3' products of histone pre-mRNA processing in vitro

Histone RNA 3' processing in vitro produces one or more 5' cleavage products corresponding to the mature histone mRNA 3' end, and a group of 3' cleavage products whose 5' ends are mostly located several nucleotides downstream of the mRNA 3' end. The formation of these 3' products is coupled to the formation of 5' products and dependent on the U7 snRNP and a heat-labile processing factor. These short 3' products therefore are a true and general feature of the processing reaction. Identical 3' products are also formed from a model RNA containing all spacer nucleotides downstream of the mature mRNA 3' end, but no sequences from the mature mRNA. Again, this reaction is dependent on both the U7 snRNP and a heat-labile factor. Unlike the processing with a full-length histone pre-mRNA, this reaction produces only 3' but no 5' fragments. In addition, product formation is inhibited by addition of cap structures at the model RNA 5' end, indicating that product formation occurs by 5'-3' exonucleolytic degradation. This degradation of a model 3' product by a 5'-3' exonuclease suggests a mechanism for the release of the U7 snRNP after processing by shortening the cut-off histone spacer sequences base paired to U7 RNA.