Dangerous Liaisons: Fanconi Anemia and Toxic Nonhomologous End Joining in DNA Crosslink Repair

Abstract

The proper choice of repair pathway is critical to tolerating various types of DNA damage. In a recent issue of Molecular Cell, Adamo et al. (2010), along with a second report (Pace et al., 2010), describe how the Fanconi anemia (FA) pathway is involved in preventing aberrant DNA repair. These studies suggest a potentially significant new opportunity for the treatment of FA.

In 1927, the Swiss pediatrician Guido Fanconi described a fatal progressive anemia that had caused the deaths of three brothers. This inherited disease came to be known as Fanconi’s anemia (FA) and is associated with congenital abnormalities, failure of hematopoiesis, and a high predisposition to cancer. Over eighty years later, our knowledge of the genetic basis of FA has progressed, but our ability to treat affected individuals is still limited. FA is classified into 13 subtypes according to the presence of homozygous mutations in any of 13 known FANC genes. Eight of the FANC genes encode factors that make up the FA ““core complex”” (FANCA-C, E-G, L, and M), which catalyzes the monoubiqui-tylation and activation of the FANCD2 and FANCI proteins. Ubiquitylated FANCD2 and FANCI are recruited to chromatin, where they facilitate DNA repair. Three factors that are associated with the homologous recombination (HR) DSB repair pathway—FANCD1, FANCN, and FANCJ—act downstream of FANCD2-FANCI.

Cells from FA patients of all sub-types exhibit chromosome abnormalities when treated with DNA crosslinking agents such as mitomycin C (Moldovan and D’Andrea, 2009; Wang and D’Andrea, 2004), which block replication and transcription. Defective interstrand crosslink (ICL) repair is thought to underlie the clinical and cellular pheno-types associated with FA. DNA crosslink processing utilizes multiple repair pathways that act at different steps, including specialized endonucleases that incise the lesion after replication fork arrest, homologous recombination proteins that repair the resultant DSBs, and translesion polymerases that replicate past the damaged base (Mirchandani and D’Andrea, 2006; Wang and D’Andrea, 2004) (Figure 1). A major function of the FA proteins appears to be to coordinate each of these three independent repair pathways (Knipscheer et al., 2009; Mirchandani and D’Andrea, 2006). In addition to a direct role in promoting efficient ICL repair, it was predicted several years ago that FA proteins might also prevent DNA ends from inappropriately engaging the nonhomologous end joining (NHEJ) machinery (Mirchandani and D’Andrea, 2006). Although in most cases NHEJ suppresses genomic rearrangements by direct ligation of DNA breaks on the same chromosome, if not properly regulated, it can also promote the inappropriate fusion of DNA breaks on different chromosomes, leading to genomic instability and tumorigenesis (Nussenzweig and Nussenzweig, 2010).

Figure 1. Opportunities for NHEJ Factors to Impact DNA Interstrand Crosslink Repair.

(A and B) DNA interstrand crosslinks (ICLs) block DNA replication, causing collapse of DNA replication forks and recruitment of Fanconi anemia (FA) gene products. The FA complex mediates ICL repair and inhibits the activity of nonhomologous end-joining (NHEJ) proteins such as Ku, which can promote genomic instability through ““toxic NHEJ.”” In the absence of FA gene products, Ku can potentially bind double-strand breaks formed (A) after incision of DNA at ICLs by endonucleases, or (B) after replication fork regression. Replication fork regression may be required to allow access of endonucleases to process ICLs. In either case, binding of Ku to double-stranded breaks could prevent completion of repair.

Adamo et al. (2010) and Pace et al. (2010) have now demonstrated that FA proteins govern the decision to channel DSBs into HR in favor of the competing error-prone NHEJ pathway. Adamo et al. (2010) show that FANC deficiency in C. elegans and human cells causes inappropriate DNA repair by the NHEJ machinery, which leads to meiotic defects and sensitivity to DNA crosslinking agents. In certain genetic backgrounds, FANC-deficient C. elegans oocytes have an abnormal chromosome number and contain aberrant structures made up of fused segments of nonhomologous chromosomes. These radial chromosomes resemble genomic rearrangements found in many human cancers and in primary cells from FA patients. By contrast, oocytes that are deficient for both FANC and the NHEJ factor, Lig4, accumulate almost no radial chromosomes. Similarly, they find that inhibition of the NHEJ proteins Ku80 or DNA-PKcs using siRNA or the use of a chemical inhibitor of DNA-PKcs is sufficient to reverse the toxic effects of MMC in FANC-deficient mammalian cells (Adamo et al., 2010; Pace et al., 2010).

The NHEJ pathway was also recently shown to exert a pathological effect in FANCC-deficient DT40 (chicken lymphocyte) cells (Pace et al., 2010). As in FA patient lines, these cells are sensitive to DNA crosslinking drugs, but this sensitivity is significantly reduced when the NHEJ factor Ku70 is deleted (Pace et al., 2010). Chromosome aberrations, which are common in FANC-deficient cells treated with MMC, are reduced to WT levels upon deletion of Ku70. Interestingly, in contrast to the study by Adamo et al., loss of DNA-PKcs or Lig4 did not suppress the FA defect, suggesting a potential difference between human and chicken cells.

It may seem counterintuitive that problems caused by the failure of one DNA repair pathway (FA) could be relieved by removing a second DNA repair pathway (NHEJ). However, NHEJ has recently been shown to compete with the HR repair pathway in vivo (Bouwman et al., 2010; Bunting et al., 2010). For example, radial chromosome formation and tumor-igenesis in Brca1 mutant mammalian cells (deficient in HR) can be prevented by deletion of 53BP1, a factor associated with the NHEJ pathway (Bunting et al., 2010). This ““toxic NHEJ”” appears to be caused by 53BP1 clogging the break site, excluding the proteins required for DNA repair by HR. While NHEJ occasionally makes chromosomal rearrangements, HR is an essentially error-free pathway, so genomic stability is favored when HR predominates over NHEJ.

Does a similar mechanism explain why NHEJ has a pathological effect in FA cells? FANC-deficient cells have a mild defect in HR (Mirchandani and D’Andrea, 2006), and in DT40 cells at least, this defect is corrected upon deletion of Ku70 (Pace et al., 2010). The exact step in the FANC pathway where NHEJ interferes is not currently known, however FANCD2-FANCI are thought to cooperate during DNA replication to recruit nucleases to cut DNA around ICLs (Figure 1A) (Knipscheer et al., 2009). Nuclease activity is expected to generate a DNA double-strand break after the ICL has been excised, which can be used to regenerate the replication fork by a process similar to HR. This double-strand break is a potential target for NHEJ factors, which could prevent access of the factors required for replication fork restart (Figure 1A).

An alternative target of NHEJ factors could be a DNA end formed by replication fork regression at an ICL (Figure 1B). When a replication fork encounters an obstacle, it can back up to form a ““chicken-foot”” structure, potentially enabling access of repair factors to remove the crosslink (Atkinson and McGlynn, 2009). Following removal of the ICL, the chicken-foot structure must be processed to restore the replication fork. Nucleolytic processing of the extruded duplex arm could be inhibited by the presence of NHEJ factors at the regressed blunt DNA end. In this scenario, NHEJ proteins would act during replication stalling, perhaps even prior to cross-link removal. In any case, a function for the FANC proteins may be to exclude or remove NHEJ factors from DNA ends produced during ICL repair, thereby favoring error-free repair by HR. Adamo et al. provide support for this notion by suggesting there is increased DNA-PKcs accumulation at sites of replication stress in FANC-deficient cells.

How does the FA pathway divert DSBs away from the error-prone NHEJ pathway and into HR? Pace et al. report that FANCD2 has a previously undetected exonuclease activity. This suggests a mechanism by which the FA pathway could potentiate HR while attenuating NHEJ. In this model, FANCD2-mediated processing of broken DNA ends generates single-stranded DNA, which provides a favorable substrate for HR but cannot be bound by NHEJ proteins. However, the reported exonuclease activity of FANCD2 acts in a 3^′–5^′ direction, unlike most other exonucleases that have been implicated in HR.

While these studies demonstrate a toxic effect of NHEJ in FA cells, some outstanding questions remain. Although ablation of NHEJ appears to increase resistance of several subtypes of FA to DNA crosslinking agents, it may not rescue deficiency in factors such as FANCJ that act downstream of DNA breakage (Adamo et al., 2010). Furthermore, it is not clear whether deletion of any NHEJ factor is sufficient to alleviate the effects of FANC deficiency, or which NHEJ components represent potentially drug-able targets. While loss of NHEJ itself promotes a certain degree of genomic instability, the finding that ICL repair defects of FA cells can be rescued by inhibition of a second DNA-repair pathway offers hope of a completely new approach to therapy for FA patients.