How to make ends meet. The repair of double-stranded breaks in DNA and the recombination of antibody gene V(D)J segments share a common pathway ...
D.B. ROTH, T. LINDAHL AND M. GELLERT
REPAIR AND RECOMBINATION
How to make ends meet The repair of double-stranded breaks in DNA and the recombination of antibody gene V(D)J segments share a common pathway involving the Ku protein, which binds DNA ends, and its associated protein kinase. A meeting on 'V(D)J recombination, DNA repair, and hypermutation' in Brighton, England* last fall, brought out striking connections between the general repair of the damage caused to DNA by ionizing radiation and the very specific process of antibody gene rearrangement that takes place only in the immune system, known as V(D)J recombination; V(D)J recombination is a process that normally brings together the 'variable', 'diversity', and joining' regions of antibody genes. As recently as five years ago, these links would have seemed implausible, but since then evidence has begun to accumulate that the two pathways share a number of common factors. The Brighton meeting was the first formal attempt to bring workers in these fields together under one roof, and a number of important new results were unveiled, including the identification of several crucial genes. The story begins a few years ago, with the realization that all cells from mice homozygous for the scid mutation are hypersensitive to X-rays and are defective in the repair of DNA double-strand breaks [1-3]. These 'scid mice' were already known to exhibit an extreme immunodeficiency as a result of a block in V(D)J recombination. As neither V(D)J recombination nor double-strand break repair has any sequence specificity in the coding sequences of the pieces of DNA to be joined, these two seemingly disparate scid phenotypes could be reconciled if there were in fact some overlap in the mechanisms employed for the normal repair of broken DNA and the joining of antibody gene fragments. According to current models for V(D)J recombination (reviewed in [4]), the process is initiated by the introduction of a double-strand break between a recombination signal sequence and the adjacent coding sequence. This leads to the generation of a flush break at the recombination signal - the so called signal end - and a covalently closed DNA hairpin structure at the coding end. In scid mice, the joining of coding ends is severely inhibited and hairpin coding ends accumulate, suggesting that the scid mutation interferes with the resolution of hairpin intermediates [5]. The joining of signal ends, on the other hand, is relatively normal in cells from scid mice. Detailed physical characterizations of the presumed
496
recombination intermediates (the coding and signal ends) was reported at the Brighton meeting by the Roth and Gellert laboratories. These studies indicate that the formation of hairpins involves a precise joining of the two strands of the coding segment, and support the idea of a coupled mechanism, by which hairpins are formed directly as a result of the cleavage reaction that generates the corresponding signal ends [6]. The dual defect seen in scid mice naturally led to a search for other DNA-repair factors that might be needed for V(D)J recombination. Initially, mutations in at least two other genes involved in double-strand break DNA repair were also shown to result in defective V(D)J joining: XRCC4, provided in mutant form by the XR-1 cell line, and XRCCS, from the xrs-5 and xrs-6 mutant cell lines, among others. In addition, XRCC5 mutant cells were found to be missing a protein that binds to DNA ends [7,8]; this protein proved to be the abundant Ku autoantigen - specifically, its 80 kD subunit, Ku80. When the human XRCC5 gene, which encodes Ku80, is expressed in XRCC5 mutant cells, DNA-end binding, V(D)J recombination, and radioresistance [9-12] are all restored. Ku, which was first detected as an autoantigen in several human autoimmune disorders, is a dimer composed of a 70 kD and an 80 kD polypeptide chain. The binding of Ku to DNA is known to require some kind of interruption in the regular DNA structure - a broken end, a single-stranded gap, or a double-strand/single-strand boundary will each serve as substrates for Ku binding. Ku, therefore, has tantalizingly plausible properties for a protein with a role in the repair of double-strand DNA breaks. Furthermore, Ku is the DNA-binding component of a DNA-dependent protein kinase, DNA-PK, which can phosphorylate a wide range of transcription factors as well as itself and Ku. XRCC5 mutant cells have very low DNA-PK activity, presumably because the kinase cannot bind to DNA. The catalytic subunit of the kinase, DNA-PKCs, is a protein with a molecular mass of 450 kD, making it one of the largest polypeptide chains known. Once Ku was implicated in V(D)J recombination and double-strand break repair, it was reasonable to ask whether mutations in the gene encoding DNA-PKcs might also affect repair. The exciting news to emerge at the Brighton meeting was that the factor defective in scid cells (and in the phenotypically very similar CHO V3 cells) has been identified as DNA-PKcs. Two research
© Current Biology 1995, Vol 5 No 5
DISPATCH consortia (represented by C. Kirchgessner, Stanford University School of Medicine, Stanford, USA; P. Jeggo, University of Sussex, Brighton, UK; S. Jackson, Cambridge University, Cambridge, UK; and G. Taccioli, Howard Hughes Medical Institute Research Laboratories, Boston, USA) reported that, when compared to normal controls, scid cells contain very little DNA-PKcs protein or kinase activity. When fragments of human chromosome 8 or yeast artificial chromosomes expressing the kinase are introduced into scid or V3 cells, normal radioresistance and V(D)J recombination are both restored [13,14]. The report that a mutation, called sxi-1, affects Ku70 function and results in defects very much like those seen in XRCC5 mutants (D. Weaver, DanaFarber Cancer Institute, Boston, USA; E.A. Hendrickson, Brown University, Providence, USA) filled in another part of the picture. Thus, all three components of the Ku-DNA-PKcs complex have roles in the repair of double-strand breaks and in V(D)J recombination. How might this kinase complex participate in V(D)J recombination? Because mutation of either of the Ku subunits severely impairs the formation of both the coding and the signal joints [15,16], it seems reasonable to suggest that the binding of the Ku80 and Ku70 proteins to DNA plays an important role in the joining of both signal and coding ends. On the other hand, the scid and V3 mutations chiefly affect coding-joint formation, so DNA-PKcs must play a specific role in this stage of the reaction. Although the mechanism underlying its effects on this part of the process remains unknown, several aspects were discussed at the meeting. Work from the Roth group indicates that coding ends, which are virtually all in the form of hairpins, accumulate to high levels in scid thymocytes [6]. Thus, one obvious possibility is that scid cells lack the ability to unseal the hairpin ends in preparation for joining. However, as reported by S. Lewis (California Institute of Technology, Pasadena, USA), when hairpin DNA molecules are introduced into scid cells by transfection, they are opened normally, indicating that the scid defect may not simply be due to the lack of a hairpinopening activity [17]. Taken together, these results suggest an alternative possibility (discussed below) - that the scid mutation somehow protects hairpin coding ends, but not other hairpins, from nuclease attack [6]. The RAG-1 and RAG-2 gene products also came under scrutiny during the meeting; both are known to be required for the initiation of V(D)J recombination, but it is not clear how they work. Evidence from cell staining and immunoprecipitation experiments argues for an interaction between the Rag-i and Rag-2 proteins (D.G. Schatz, Yale Medical School, New Haven, USA). In addition, the Rch-l protein that is thought to have a role in nuclear structure and protein import has been shown to interact with Rag-1 (M. Oettinger, Harvard Medical School, Boston, USA). Mutational studies have led to the isolation of a RAG-1 mutant that renders the process of V(D)J recombination hypersensitive to the sequence of coding DNA that flanks the recognition signals, implying
that Rag-1 may interact with this site (M.G.). Although these results are only suggestive, they begin to indicate a system in which Rag-i and Rag-2, along with other factors, may locate at recombination sites. U. Storb (University of Chicago, Chicago, USA) reported studies which suggest that the processing of coding ends may depend upon whether they are attached to a recombination signal sequence with a 12 nucleotide or a 23 nucleotide spacer. One possible way by which the type of signal sequence could influence the joining of coding ends would be for all four ends to remain physically associated during joining. An attempt to weave together these findings into a working model is shown in Figure 1. The central postulate is that the hairpin coding ends are sequestered in a DNAprotein complex involving Ku70 and Ku80, which is altered after activation of DNA-PKcs to allow the coding ends to become accessible to the hairpin-opening nuclease. (Complex formation involving multiple factors is a common feature of site-specific recombination and transposition reactions.) It is suggested that cleavage takes place after the DNA substrate has been assembled into an 'initiation complex' that may be composed of Rag-i, Rag-2 and associated factors. DNA cleavage occurs precisely between the recombination signal and the coding segment, producing a pair of blunt signal ends and a pair of hairpin coding ends. We propose that both coding and signal ends are bound by protein complexes that include Ku70, Ku80, and perhaps other components. Al four ends may remain associated in a large DNA-protein complex for some time prior to joining. Such an arrangement would facilitate inversional recombination, and is consistent with the data from Storb's group (discussed above). The presence of Ku70 and Ku80 bound to DNA ends allows recruitment of DNA-PKcs, which becomes activated upon binding. Activated DNA-PKcs then modifies target molecules (presumably by phosphorylation), resulting in the increased accessibility of the hairpins to the hairpin-opening nuclease. Targets for phosphorylation by the kinase could include components of the complex (such as DNA-PKcs itself, Ku70 or Ku80, or other molecules). Alternatively, DNA-PKcs could act on components outside of the complex, for example by facilitating the recruitment of a specific hairpin-opening nuclease to the complex. Once the hairpins are opened, the coding ends are rapidly joined to each other, perhaps with the participation of Ku70 and Ku80. In cells from scid mice, the absence of DNA-PKcs activity blocks the release of the hairpin coding ends from the DNA-protein complex, so that coding ends accumulate in the form of hairpins. An alternative, or additional, possibility is that DNA-PKcs may be required in some way to organize a complex needed for the joining of coding ends. How might the kinase act in DNA repair? As in V(D)J recombination, DNA-PKcs might play a role in regulating the assembly of multiprotein complexes. There would, however, be competition for binding to DNA ends by
497
498
Current Biology 1995, Vol 5 No 5 other nuclear proteins, especially the very abundant poly(ADP-ribose) polymerase. Alternatively, the activated kinase might act more indirectly, for example by regulating the expression of genes that are involved in the repair process or in a linked process, such as cell-cycle regulation. The possibility that DNA-PKcs is involved in cellcycle control is appealing; activation of DNA-PK by binding to broken DNA ends would provide a logical sensing mechanism that could prevent cells from initiating DNA replication before they had repaired damage in their DNA. Moreover, DNA-PKcs might have other functions in addition to being a kinase; such a large protein would be likely to have several distinct domains with discrete roles. Another possibility is that Ku could interact with repair factors; for example, T. Stamato (Lankenau Medical Research Centre, Wynnewood, USA) reported evidence for its binding to DNA ligase III in cell extracts. Ku also has an intrinsic DNA-helicase activity [18], which may itself play a role in the repair process. There are other aspects of DNA repair in which a role for Ku proteins would be attractive. M. Jasin (Sloan-Kettering Institute, New York, USA) reported progress in the elucidation of how double-strand breaks are repaired in a chromosomal context [19]. M. Fairman (MRC Radiobiology Unit, Didcot, UK) and P. Pfeiffer (University of Cologne, Cologne, Germany) described results from in vitro studies of the repair of double-strand breaks using extracts from mammalian and Xenopus cells [20], respectively. The DNA end-joining reactions in all three systems seem to proceed in a similar fashion, and it has been suggested that joining requires an 'alignment protein' that holds ends together while repair occurs [21]. Could this function be served by the end-binding activity of the Ku complex? The identification of the three subunits of DNA-PK as the products of genes required for normal cellular resistance to ionizing radiation has stimulated interest in other X-ray-sensitive mutant mammalian cells. Mutant human cells with defects in the genes XRCC1-4, for example, are all X-ray sensitive. The 70 kD XRCC1 protein forms a tight complex with DNA ligase III and promotes its ligation activity [22]; DNA ligase III is emerging as a prime candidate for the end-joining reactions in recombination and base excision-repair, whereas ligase I is an essential enzyme active in lagging-strand DNA replication. Ligase II and the recently discovered ligase IV (T.L.; and see [23]) are still searching for roles. Fig. 1. A working model: does DNA-PKcs function as a 'release factor' to allow hairpin coding ends to become available for opening? According to this proposal, Ku70 and Ku80, perhaps along with other factors, bind to the signal and coding ends. The coding ends, which are covalently sealed in a hairpin structure, remain in a DNA-protein complex and are sequestered from hairpin-opening activities. DNA-PKcs binds to this complex, is activated, and phosphorylates target molecules, resulting in the increased accessibility of the coding ends to the hairpin-opening activity. Upon opening, the hairpin coding ends are joined to each other.
DISPATCH The XRCC4 (XR-1) gene [24] is also required for efficient repair of double-strand breaks but does not encode a known component of DNA-PK. Several new X-ray-sensitive Chinese hamster cell mutants have now been isolated, as reported by M. Zdzienicka (University of Leiden, Leiden, The Netherlands) [25], and these promise to become important tools. In addition to many XRCC1 and XRCC5 mutants, two novel types of mutant cells, V-C4 and V-C8, have been recovered. The V-C4 mutant has intriguingly similar properties to human cells derived from patients with ataxia-telangiectasia (AT), or the (similar) Nijmegen syndrome. The cells are very sensitive to X-rays and exhibit radiation-resistant DNA synthesis. However, V-C4 cells were not phenotypically complemented by human chromosome 11, which harbors the AT gene, indicating that V-C4 cells do not carry a mutationally altered AT gene but are defective in another step of this radiation defense pathway. A final link between the V(D)J recombination and DNA repair pathways was suggested by the report from the laboratory of I. Kirsch (National Cancer Institute, Bethseda, USA) that ionizing radiation stimulates aberrant V(D)J recombination events, giving rise to chromosome translocations. These rearrangements are also increased in patients with AT. Thus, it appears that both exposure to ionizing radiation and mutations that result in radiosensitivity can affect the fidelity of V(D)J recombination events; the molecular basis of this phenomenon is far from clear. Nevertheless, the results described at the Brighton meeting demonstrate convincingly that studies of V(D)J recombination and of the repair of doublestrand DNA breaks are rapidly converging, and a clearer picture of the molecular pathways of both processes should soon be available.
8. 9.
10.
11. 12.
13.
14.
15.
16. 17.
18.
19.
20.
21. 22.
23.
References 1. Fulop GM, Phillips RA: The scid mutation in mice causes a general defect in DNA repair. Nature 1990, 347:479-482. Biedermann KA, Sun J, Giaccia AJ, Tosto LM, Brown JM: scid mutation in mice confers hypersensitivity to ionizing radiation and a deficiency in DNA double-strand break repair. Proc Natl Acad Sci USA 1991, 88: 1394-1397. 3. Hendrickson EA, Qin X-Q, Bump EA, Schatz DG, Oettinger M, Weaver DT: A link between double-strand break-related repair and V(D)J recombination: the scid mutation. Proc Natl Acad Sci USA 1991, 88: 4061-4065. 4. Lewis SM: The mechanism of V(D)J joining: lessons from molecular, immunological and comparative analyses. Adv Immunol 1994, 56: 27-150. 5. Roth DB, Menetski JP,Nakajima PB, Bosma MJ, Gellert M: V(D)J recombination: broken DNA molecules with covalently sealed (hairpin) coding ends in scid mouse thymocytes. Cell 1992, 70:983-991. 6. Zhu C, Roth DB: Characterization of coding ends in thymocytes of scid mice: implications for the mechanism of V(D)J recombination. Immunity 1995, 2:101-112. 7. Rathmell WK, Chu G: A DNA end-binding factor involved in doublestrand break repair and V(D)J recombination. Mol Cell Biol 1994, 14:4741-4748. 2.
24.
25.
Getts RC, Stamato TD: Absence of a Ku-like DNA end binding activity in the xrs double-strand DNA repair-deficient mutant. J Biol Chem 1994, 269:15981-15984. Taccioli GE, Gottlieb TM, Blunt T, Priestley A, Demengeot J, Mizuta R, Lehmann AR, Alt FW, Jackson SP, Jeggo PA: Ku80: product of the XRCC5 gene and its role in DNA repair and V(D)J recombination. Science 1994, 265:1442-1445. Boubnov NV, Hall KT, Wills Z, Lee SE, He DM, Benjamin DM, Pulaski CR, Band H, Reeves W, Hendrickson EA, et al.: Complementation of the ionizing radiation sensitivity, DNA end binding, and V(D)J recombination defects of double-strand break repair mutants by the p86 Ku autoantigen. Proc Natl Acad Sci USA 1995, 92:890-894. Smider V, Rathmell WK, Lieber MR, Chu G: Restoration of X-ray resistance and V(D)J recombination in mutant cells by Ku cDNA. Science 1994, 266:288-291. Finnie NJ, Gottlieb TM, Blunt T, Jeggo PA, Jackson SP: DNA-dependent protein kinase activity is absent in xrs-6 cells: implications for site-specific recombination and DNA double-strand break repair. Proc Natl Acad Sci USA 1995, 92:320-324. Kirchgessner CU, Patil CK, Evans JW, Cuomo CA, Fried LM, Carter T, Oettinger MA, Brown JIM:DNA-dependent kinase (p350) as a candidate gene for the murine scid defect. Science 1995, 267:1178-1183. Blunt T, Finnie NJ, Taccioli GE, Smith GCM, Demengeot J, Gottlieb TM, Mizuta R,Varghese AJ, Alt FW, Jeggo PA, et al.: Defective DNA-dependent protein kinase activity is linked to V(D)J recombination and DNA repair defects associated with the murine scid mutation. Cell 1995, 80:813-823. Pergola F, Zdzienicka MZ, Lieber MR: V(D)I recombination in mammalian mutants defective in DNA double-strand break repair. Mol Cell Biol 1993, 13:3464-3471. Taccioli GE, Rathbun G, Oltz E, Stamato T, Jeggo PA, Alt FW: Impairment of V(D)J recombination in double-strand break repair mutants. Science 1993, 260:207-210. Lewis SM: P nucleotide insertions and the resolution of hairpin DNA structures in mammalian cells. Proc Natl Acad Sci USA 1994, 91: 1332-1336. Tuteja N, Tuteja R, Ochem A, Taneja P, Huang NW, Simoncsits A, Susic S, Rahman K, Marusic L, Chen J, et al.: Human DNA helicase II - a novel DNA unwinding enzyme identified as the Ku autoantigen. EMBO J 1994, 13:4991-5001. Rouet P, Smih F, Jasin M: Introduction of double-strand breaks into the genome of mouse cells by expression of a rare-cutting endonuclease. Mol Cell Biol 1994, 14:8096-8106. Beyert N, Reichenberger S, Peters M, Hartung M, Gottlich B, Goedecke W, Vielmetter W, Pfeiffer P: Nonhomologous DNA end joining of synthetic hairpin substrates in Xenopus laevis egg extracts. Nucleic Acids Res 1994, 22:1643-1650. Thode S, Schaffer A, Pfeiffer P, Vielmetter W: A novel pathway of DNA end-to-end joining. Cell 1990, 60:921-928. Caldecott KW, McKeown CK, Tucker JD, Ljungquist S, Thompson LH: An interaction between the mammalian DNA repair protein XRCC1 and DNA ligase III. Mol Cell Biol 1994, 14:68-76. Wei Y-F, Robins P, Carter K, Caldecott K, Pappin DJC, Yu G-L, Wang R-P, Shell BK, Nash RA, Schar P, et al.: Molecular cloning and expression of human cDNAs encoding a novel DNA ligase IV and DNA ligase Ill, an enzyme active in DNA repair and recombination. Mol Cell Biol 1995, in press. Giaccia A, Denko N, MaClaren R, Mirman D, Waldren C, Hart I, Stamato TD: Human chromosome 5 complements the DNA doublestrand break-repair deficiency and gamma-ray sensitivity of the XR-1 hamster variant. Am J Human Genet 1990, 47:459-469. Zdzienicka MZ, Verhaegh GWCT, Jongmans W, Morolli B, Jaspers NGJ, Oshimura M: Functional complementation studies with X-ray sensitive mutants of Chinese hamster cells closely resembling ataxia-telangiectasia cells. IntJ Radiat Biol 1994, 66:189-195.
D.B. Roth, Department of Microbiology and Immunology, Baylor College of Medicine, Houston, Texas 77030, USA. T. Lindahl, Imperial Cancer Research Fund, Clare Hall Laboratories, South Mimms, Hertfordshire EN6 3LD, UK. M. Gellert, Laboratory of Molecular Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20892, USA.
499
