Expanded roles of the Fanconi anemia pathway in preserving genomic stability

Repairing double-strand breaks (DSBs)—homologous recombination (HR) vs. nonhomologous end-joining (NHEJ)

The FA pathway plays a fundamental role in coordinating the repair of DNA DSBs. DSBs can be generated in numerous ways. Various genotoxic agents generate DSBs, including ionizing radiation (IR), chemicals that block DNA replication fork progression (such as HU), or byproducts of cellular metabolism (such as reactive oxygen species (ROS)]. DSBs can also be generated as an intermediate of normal cellular processes during immunoglobulin class switching in immune cells. Failure to properly repair DSBs leads to chromosome translocations, genomic instability, and cell death or transformation. Cells have evolved two major specialized mechanisms to cope with these highly deleterious lesions; namely, HR and NHEJ.

Through HR, DNA sequences are exchanged between two similar or identical DNA strands. NHEJ is an alternative pathway in which two broken DSBs are sealed directly, without the need of homologous templates. Cells have evolved specialized proteins that facilitate either HR or NHEJ for DSB repair. Figure 2 describes the simple mechanistic models for HR and NHEJ, both pathways requiring multiple-step reactions. Because HR uses genetically identical complementary DNA strands as templates for repair, it is error-free, while NHEJ can be error-prone, since DNA end ligation can occur without preserving the correct frame of codons, resulting in loss of genetic information. Despite the higher risk of genomic instability, mammalian cells depend primarily on NHEJ, unlike in yeast, where HR is the primary choice for repairing DSBs.

DSB as an intermediate step in ICL repair

In addition to the above-mentioned “direct” DSB-generating means, a DSB can also be generated as an intermediate of the ICL-repairing process (Fig. 3; Kennedy and D'Andrea 2005; Moldovan and D'Andrea 2009). In the current model, ICL hinders the progression of replication fork, which will activate checkpoints and monoubiquitination of FANCD2/FANCI, while the fork is stabilized by repair proteins. The cross-links are thought to be unhooked initially by serial activities of at least two endonucleases (XPF–ERCC1 and MUS81–EME1) (Niedernhofer et al. 2004; Hanada et al. 2006), and the lesion is subsequently bypassed by translesion synthesis (TLS) polymerases, which might be recruited by FANCD2 (Knipscheer et al. 2009; Moldovan and D'Andrea 2009). These processes would generate a DSB, which must be resolved properly for resumption of replication fork progression. The legitimate choice for DSB repair at this point would be HR, mainly because of availability of homologous templates. This hypothesis is strongly backed by reports that ablation of HR genes, but not NHEJ genes, leads to growth sensitivity toward ICL-inducing agents in various organisms (Cole 1973; Grossmann et al. 2001). Consistent with this notion is that Ku deletion chicken DT40 mutant cells are resistant to ICL agents compared with wild-type cells (Takata et al. 1998), and one interpretation could be that elevated HR would be beneficial for cells in repairing ICL. Below, we summarize and discuss the evidence that indicates HR is defective in FA cells, and how the FA pathway regulates resolution of the DSB intermediates during ICL repair.

Evidence for the FA pathway promoting HR

Numerous studies have suggested that the FA pathway promotes HR repair. In DT40 cells, FANCG and FANCD2 knockout was shown to be defective in HR-mediated DSB repair and gene conversion at the IgG locus (Yamamoto et al. 2005). FANCC knockout DT40 cells are defective in HR and it is shown to be epistatic with XRCC2, a RAD51 paralog, in sensitivity to ICL-inducing agents, suggesting that FANCC functions with XRCC in repairing DNA cross-links (Niedzwiedz et al. 2004). Most HR assays in mammalian cells were performed using a reporter-based system. In this system, site-specific DSBs are introduced by the restriction enzyme I-Sce1, resulting in the HR-mediated generation of a functional GFP ORF. Repair by HR is scored as a positive GFP signal (Richardson et al. 1999). Using this system, it was shown that patient-derived FANCA-, FANCG-, and FANCD2-deficient cells are defective, albeit to a mild degree, in HR repair (Nakanishi et al. 2005). Also, siRNA-mediated depletion of FANCI and FANCD2 resulted in reduced HR activities (Smogorzewska et al. 2007). Using an ICL-containing plasmid reporter system, it was further shown that depletion of FA genes, in addition to other DNA repair components, reduced HR activities (Zhang et al. 2007). Consistent with these reports that the FA pathway promotes HR activities, a study using knockout mice did not show an epistatic relationship between FA and NHEJ in repairing DSBs, as FANCD2^−/−/Prkdc^sc/sc double-mutant mice (Prkd has reduced DNS-dependent protein kinase [DNA-PK] activity) are more sensitive to IR than Prkdc^sc/sc single-mutant mice (Houghtaling et al. 2005). USP1, the deubiquitinating enzyme required for deubiquitination of FANCD2/FANCI and completion of the FA pathway (Oestergaard et al. 2007; Smogorzewska et al. 2007; Kim et al. 2009), was also required for efficient HR activity, shown in a mouse knockout system (Kim et al. 2009), consistent with the reports that the FA promotes HR.

How the FA pathway promotes HR is unclear, but most evidence suggests that the monoubiquitination of FANCD2 is the critical regulatory event for promoting HR activity. The modified FANCD2 has been shown to recruit several DNA repair factors involved in HR (such as BRCA1, BRCA2, RAD51, and NBS1) to the damaged chromatin to facilitate repair (Garcia-Higuera et al. 2001; Nakanishi et al. 2002; Hussain et al. 2004; Wang et al. 2004; Zhi et al. 2009). Thus, deregulation of FANCD2 monoubiquitination in FA or USP1-deficient cells could directly result in the failure of recruitment of the DNA repair factors to the DSB lesions, leading to defective initiation of the HR process.

Another line of evidence suggests that FANCD2 may also functionally cooperate with other DNA-metabolizing enzymes, possibly in the later step of HR. It was shown that DNA polymerase POLN and its associated helicase, HEL308, are required for HR activities and resistance to ICL-inducing agents (Moldovan et al. 2010). Notably, an epistatic relationship between POLN and FANCD2/FANCI in HR was observed, and, consistently, a HEL308 and FANCD2 double knockout was similarly sensitive to ICL-inducing agents compared with single knockouts in Caenorhabditis elegans (Muzzini et al. 2008). Interestingly, C. elegans helq-1, an ortholog of HEL308, was shown to promote disassembly of post-synaptic RAD51 filament, a step required for completion of HR (Ward et al. 2010). Collectively, these results suggest a cooperative role of POLN/HEL308 and the FA pathway in promoting HR and ICL repair. In addition to POLN, the DNA synthesis step during HR appears to be mediated by other polymerases, as POLη and POLδ were shown to be capable of synthesizing DNA from D-loop (displacement loop) recombination intermediates in vitro (McIlwraith et al. 2005; Li et al. 2009). These suggest that the multiple polymerases may participate cooperatively or redundantly in the completion of HR.

Last, it is tempting to speculate that there may be an indirect mechanism that contributes to decreased HR activity in FA cells. For instance, FA cells may have increased p53 activity, which in turn suppresses HR (Mekeel et al. 1997). Also, it cannot be ruled out that the “upstream” FA core complex (FANCA, FANCB, FANCC, FANCE, FANCF, FANCG, FANCL, and FANCM) proteins may have additional roles linked to the regulation of HR activity, but separate from their role in FANCD2/FANCI monoubiquitination.

BRCA1, BRCA2, PALB2 (partner and localizer of BRCA2), and FANCJ

More direct evidence for the participation of the FA pathway in HR came from the identification of BRCA2 as a FA subtype, FANCD1 (Howlett et al. 2002). BRCA2 is a critical factor required for initiating HR by directly binding and facilitating RAD51 loading onto ssDNA (Moynahan et al. 2001b; Litman et al. 2005). Subsequently, PALB2, a factor that binds and regulates localization of BRCA2 (Xia et al. 2006), was identified as a FA subtype, FANCN (Xia et al. 2007). BRCA1, a breast and ovarian cancer susceptibility gene, is not a FA gene per se, but it is closely linked to the FA pathway, as BRCA1-deficienct cells also display cross-linker sensitivity, chromosomal breakage, and an HR defect (Moynahan et al. 2001a; Kennedy and D'Andrea 2005). A series of recent studies demonstrated critical interplays among BRCA1, BRCA2, and PALB2 in coordinating the HR process (Sy et al. 2009; Zhang et al. 2009a,b). It was shown that PALB2 also binds directly to BRCA1, thus bridging the interaction between BRCA1 and BRCA2. Importantly, BRCA1 mutations in patients that disrupt its interaction with PALB2 result in impaired HR and DNA damage response. Collectively, a model can be assembled in which BRCA1 acts to recruit PALB2, which in turn recruits BRCA2 onto ssDNA. FANCJ (also known as BACH1 [BRCA1-associated C-terminal helicase] or BRIP1 [BRCA1-interacting protein C-terminal helicase 1]) is also mutated in hereditary breast cancer, and is required for HR (Bridge et al. 2005; Levitus et al. 2005; Levran et al. 2005; Litman et al. 2005). Genetic analyses in C. elegans suggested that dog-1, a FANCJ ortholog, is epistatic with FANCD2 in ICL repair, suggesting that FANCJ functions cooperatively with FANCD2, although FANCJ was suggested to work downstream from RAD51, unlike FANCD2 (Youds et al. 2008). Interestingly, it was also shown that FANCJ is not epistatic with BRC-2, the C. elegans BRCA1, in ICL repair, suggesting that FANCJ participates in two distinct pathways. This observation is consistent with reports that the FANCJ–BRCA1 interaction is not required for HR and ICL repair (Bridge et al. 2005; Peng et al. 2007). How FANCJ functions mechanistically in HR is unclear, although it was suggested to function in limiting promiscuous recombination filament intermediates (Sommers et al. 2009).

A well-known feature that distinguishes BRCA1, BRCA2 (FANCD1), PALB2 (FANCN), and BACH1 (FANCJ) from other FA proteins is that these proteins are not required for monoubiquitination of FANCD2. In addition, these factors are more strictly required for HR activity compared with the rest of the FA proteins. The fact that germline mutations in these genes increase the risk of breast and ovarian cancers (Wooster and Weber 2003; Turnbull and Rahman 2008) suggests that HR-mediated DNA repair is a critical tumor-suppressive mechanism lacking in these and other cancers.

Altogether, there seem to be multiple mechanisms by which the FA pathway promotes the HR pathway. Precisely how these multiple events converge in a coordinated manner for promotion of HR is an important future topic. For instance, how can one link the HR-deficient phenotypes seen in FA cells (e.g., FANCA) to the “downstream” FANCD1/BRCA2 phenotypes? Is the monoubiquitinated FANCD2 a critical link between the “upstream” FA proteins and downstream factors such as BRCA2 or PALB2? At least one report showed that BRCA2 interacts physically with FANCG, and that they are epistatic in cross-linker sensitivity (Wilson et al. 2008). But why are the upstream FA mutant cells only modestly deficient in HR compared with the cells mutated in core HR genes? It is possible that the FA proteins preferentially regulate a “sub-HR” pathway that is dedicated to resolving ICL damage. Along these lines, BRCA2 may be engaged in a wider range of DSB repairs, including the ICL repair and other repairs.

Familial mutations of RAD51C identified in FA-like patients

Identification of the 13 FA genes has provided hints that a major function of the FA pathway is to regulate HR activities during ICL repair. There are still some FA patients with unassigned FA subtypes, and identification of additional FA genes will provide further insights into the molecular functions of the FA pathway. Two recent studies indeed provide such insights. One study reported the identification of RAD51C—a member of the RAD51-like family that is essential for RAD51-mediated HR—as a causative gene biallelically mutated in a family with “FA-like” phenotypes (Vaz et al. 2010). An accompanying study further reported identification of monoallelic (heterozygous) germline mutations in RAD51C genes resulting in an increased incidence of breast and ovarian cancers (Meindl et al. 2010). These two reports reaffirm the close genetic relationship between FA and breast cancer susceptibility (Levy-Lahad 2010). Among the three affected children from the family with the RAD51C homozygous mutation, two siblings died at a very young age, with multiple congenital abnormalities and elevated chromosomal breakages in lymphocytes upon exposure to ICL, strongly suggesting FA. The only surviving patient exhibited many characteristic FA phenotypes, such as ICL-induced chromosomal breakage, congenital abnormalities (short stature), and pronounced cell cycle arrest at G2/M, and the cellular phenotypes were rescued by exogenous RAD51C cDNA expression (Vaz et al. 2010). These results suggest that RAD51C could be the 14th FA complementation group (FANCO). However, the assignment was tentative, since there was only a single family reported so far, and the only surviving patient, at the age of 10, did not show the hematological abnormalities and/or cancers commonly observed in FA patients. However, as Vaz et al. (2010) noted, the age of onset of these phenotypes among FA patients varies. Therefore, a “provisional” assignment of FANCO was given, subject to confirmation with additional families with the mutation (C Mathew, pers comm.).

These reports put RAD51C in the list of cancer susceptibility genes that are classified as classic FA-associated HR genes, including BRCA1, BRCA2 (FANCD1), PALB2 (FANCN), and BACH1 (FANCJ) (Turnbull and Rahman 2008; Levy-Lahad 2010; Meindl et al. 2010). Whether RAD51C has direct functional relationships with FA proteins remains unclear. RAD51C appears to have multiple roles in HR, as it is required for RAD51 foci, suggestive of an upstream function, and for resolving Holliday junction intermediates at a later step of HR (Takata et al. 2001; French et al. 2002; Liu et al. 2004, 2007; Rodrigue et al. 2006). The latter function remains controversial, given the recent identification of GEN1 as the true resolvase responsible for resolution of Holliday junction (Ip et al. 2008). RAD51C contains a functional NLS (nuclear localization signal) at the C terminus, and it was suggested to play a role in nuclear transport of RAD51 following DNA damage (French et al. 2002; Gildemeister et al. 2009). In agreement with this, overexpression of RAD51 partially rescued the Cisplatin sensitivity of RAD51C mutant DT40 cells, suggesting that at least partial function of RAD51C is to promote RAD51 activity during HR (Takata et al. 2001). How RAD51C promotes RAD51 foci formation and HR activity is unknown, although one report indicates that binding of RAD51C to RAD18, a RING domain E3 ubiquitin ligase involved in PCNA monoubiquitination, is required for RAD51 foci and HR activity (Huang et al. 2009).

As different FA proteins are involved in these steps as well, RAD51C may communicate with multiple FA proteins at different steps for promoting HR. Also, as there are other known paralogs of RAD51 (RAD51B, RAD51D, XRCC2, and XRCC3) that are all critical for HR activity, ICL resistance, and genomic stability (Takata et al. 2000; Thompson and Schild 2001; Rodrigue et al. 2006), it is possible that mutation of these genes might be responsible for other unassigned FA subtypes. Altogether, these reports further emphasize the importance of the FA pathway in regulating HR.

Choice between HR and NHEJ

How cells choose one pathway or the other during DSB repair remains poorly understood. The key determinant for the selectivity between the two pathways appears to be the availability of sister chromatids. When DSBs occur during replication fork arrest, the easy accessibility of a homologous template makes HR the best choice. Thus, HR is believed to act primarily on “replication-associated” DSBs. However, random DNA-damaging events, such as chemical modification of DNA (e.g., caused by IR), may occur even when chromatin is in a compact stage, such as during G1 phase of the cell cycle. NHEJ may be an easier and efficient choice in the latter case, where homologous DNA is not necessarily accessible. Accordingly, it is reasonable for cells to depend heavily on HR during S or G2 phases, when the homologous template or sister chromatids are present nearby, while using NHEJ during G1 phases. An experiment using the DT40 system supported this hypothesis by showing that ku70 mutants are extremely sensitive to IR when cells are synchronized in G1, and recover in S/G2 phases, while rad54 mutants are selectively sensitive in S/G2 phases (Takata et al. 1998). There may be multiple mechanisms to regulate differential usage during the cell cycle.

Although one decisive factor for differential usage may be the availability of sister chromatids, increasing evidence suggests that there may be competition between the two pathways (Pierce et al. 2001; Karathanasis and Wilson 2002; Branzei and Foiani 2008; Yun and Hiom 2009). A major mechanism appears to be governed by CDK1-mediated phosphorylation of CtIP (Sae2 in yeast) in S phase. This event activates the nuclease in DNA end resection, thus facilitating ssDNA formation and the initiation of HR (Sartori et al. 2007; Huertas et al. 2008). CtIP phosphorylation in S phase was also shown to be essential for BRCA1 recruitment and HR activities (Yu and Chen 2004; Yun and Hiom 2009). Thus, recognition of DSB ends by CtIP may be a decision point at which the HR pathway is preferred over NHEJ. Although it is not known whether an absence of CtIP phosphorylation leads to increased NHEJ activity in S phase, NHEJ activity does appear to persist throughout the cell cycle (Takata et al. 1998; Kim et al. 2005), further supporting the hypothesis of the existence of interpathway competition. However, certain mechanisms might also exist to actively down-regulate NHEJ activity during S phases (Lee et al. 1997; Chen et al. 2005). A series of recent studies led by Nussenzweig and coworkers (Cao et al. 2009; Bunting et al. 2010) further supports this “competition” hypothesis by demonstrating that the interplay between BRCA1 and 53BP1 may mediate a pathway choice between HR and NHEJ. 53BP1 is an ATM substrate involved in DNA damage checkpoints, and it is required for facilitating certain types of NHEJ processes (DiTullio et al. 2002; Difilippantonio et al. 2008). These studies elegantly showed that an underlying mechanism for elevated genomic instability in BRCA1-deficient cells is the overuse of NHEJ for DSB repair. A critical NHEJ-promoting event is the inhibition of CtIP-mediated DNA end resection by 53BP1. Deletion of 53BP1 restored RAD51 foci and genomic stability in BRCA1-deficient cells. Deletion of LIG4, the end-joining ligase that functions downstream in NHEJ, did not rescue the phenotype, while it still inhibited end-joining, reaffirming the notion that 53BP1 binding is the determinant of the HR inhibitory event. Interestingly, in a more recent independent study, a 53BP1 mutation was found to be a suppressor of proliferation-deficient and ICL-sensitive phenotypes of BRCA1-null cells in a transposon-mediated genetic screen (Bouwman et al. 2010). It was shown that 53BP1 depletion restores HR activity (RAD51 foci), decreases the amount of chromosome breakages in BRCA1-null cells, and abrogates the ATM-mediated DNA damage checkpoint. The decreased DNA checkpoint activity may be due to restored HR and efficient DNA repair activities. Furthermore, Bouwman et al. (2010) found a strong correlation between reduced levels of 53BP1 and more aggressive triple-negative (estrogen, progesterone, and HER2 receptor-negative) breast tumors. This suggests that having a secondary mutation in 53BP1 might be a compensatory mechanism for the BRCA1-deficient tumors to be drug-resistant.

Although the above studies focus only on BRCA1-deficient settings, the role of 53BP1 in promoting NHEJ may be more widespread (Difilippantonio et al. 2008; Dimitrova et al. 2008). Together, these studies suggest that the displacement of 53BP1 by BRCA1 provides a key DNA repair pathway choice, favoring HR. Interestingly, BRCA1 binding to DSB ends also provides a mechanism for inhibiting the exonuclease activity of MRE11, a subunit of MRN complex (Paull et al. 2001). The MRE11 exonulcase activity was shown to be required for both HR and NHEJ (Zha et al. 2009). Thus, BRCA1 may also promote HR by specifically inhibiting the MRE11-mediated DSB end processing event required for NHEJ. These findings further suggest that there may be broader mechanisms that regulate the dynamic competition between HR and NHEJ factors at the site of DSB lesions. Very recent studies indicate that FA proteins may have similar functions at the site of DSBs during ICL repair.

The FA pathway suppresses NHEJ and promotes HR

While multiple mechanisms exist to ensure selective activation of HR in S phases, how this might be achieved during ICL repair is unknown. Since FA-deficient cells are only mildly sensitive to DSB-inducing agents (e.g., IR and HU), the FA pathway is specialized, at least in part, in the regulation of HR activities, specifically during ICL repair. As described above, monoubiquitinated FANCD2-mediated recruitment of HR factors (e.g., BRCA1 and BRCA2) may be one mechanism of promoting HR during ICL repair. Whether the FA pathway also suppresses NHEJ has not been explored until recently. A series of recent studies—using a combination of C. elegans, chicken DT40, and mammalian cell studies—has provided provocative evidence that the FA pathway is indeed involved in the active suppression of the NHEJ pathway (Adamo et al. 2010; Pace et al. 2010). In the former study, the FA phenotypes (e.g., ICL sensitivity, chromosomal breakages, and developmental abnormality) of fcd-2 worms and mammalian FA-deficient cells (FANCD2, FANCA, and FANCC) were all, at least in part, rescued by inhibition of the NHEJ pathway, while, in the latter study, ICL sensitivity and HR defect of the FANCC knockout DT40 cells were partially rescued by additional Ku70 knockout. (The discrepancy here is that the latter study did not observe the suppressive effects by knocking out DNA-PK catalytic subunit [DNA-PKcs] or Lig4, other downstream NHEJ factors.) These results suggest that the FA phenotypes arise not simply due to a weakness in HR, but also to a failure in NHEJ suppression. Although the underlying molecular mechanism remains to be identified, several different scenarios are possible. Loss of the FA pathway may result in impaired recruitment of NHEJ-suppressive factors such as PARP-1 and RAD18 (Saberi et al. 2007), leading to a heightened NHEJ activity. Alternatively, loss of the FA pathway may result in impaired recruitment of early HR factors at DSB sites that might be coupled to the impaired displacement of NHEJ factors from the lesion. Indeed, inappropriate recruitment of DNA-PKcs at the site of DSBs was observed in the FANCD2-deficient cells (Adamo et al. 2010). FANCD2 might also have a cryptic role in blocking the recruitment of NHEJ factors to the site of DSBs, as purified FANCD2 was shown to possess an intrinsic exonuclease activity in vitro (Pace et al. 2010) that might generate specific DNA structures that exclude binding of NHEJ factors. Polyubiquitination and degradation of Ku80 on chromatin is a possible mechanism for its release from chromatin (Postow et al. 2008), and the FA pathway may be required. Either model would be consistent with the observation that more active DNA-PK proteins are associated with damaged chromatin in FA-deficient cells. It appears that the nonfunctional FA pathway may render NHEJ the “default” pathway, leading to more error-prone DNA repair and the accumulation of insertions and deletions. It is interesting to speculate that a hidden function of monoubiquitinated FANCD2/FANCI complex might be facilitating removal of NHEJ factors (e.g., Ku70/80) from damaged chromatin, thus suppressing the downstream end-joining process. This would be similar to the mechanism observed in BRCA1 and 53BP1 studies (Bouwman et al. 2010; Bunting et al. 2010). It is noteworthy that inhibition of LIG4 rescued the FA phenotype, but did not rescue the BRCA1 deficiency (Bunting et al. 2010), potentially suggesting that the FA pathway might be involved in inhibiting both initiation and downstream end-joining steps of the NHEJ pathway. Figure 4 describes the hypothetical mechanisms for suppressing NHEJ during general and ICL-mediated DSB repair processes.

What is the implication for the clinical treatment of FA? Hyperactivation of error-prone NHEJ during ICL repair may elevate the undesirable increase in mutagenesis and random end-joining between neighboring broken chromosomes (Newell et al. 2004). Likewise, improper usage of NHEJ during the hematopoietic program, where HR plays a more critical role, may result in genomic instability, leading to bone marrow failure. These studies also suggest that usage of DNA-PK inhibitors may be beneficial for treating both FA patients and FA-associated tumors, although there may also be a risk of enhanced toxicity for FA patients treated with the drugs. Further investigation of the potential therapeutic approach may have a significant impact on the treatment of FA patients.

Identification of FAN1

Although the key molecular event in the FA pathway is the generation of monoubiquitinated FANCD2/FANCI, the exact molecular functions of the modified FANCD2/FANCI proteins remain enigmatic. Four recent studies (Liu et al. 2010; Kratz et al. 2010; Mackay et al. 2010; Smogorzewska et al. 2010) describe at least one novel function of monoubiquitinated FANCD2. They showed that FAN1 (previously known as KIAA1018), a novel nuclease with a N terminus UBZ (ubiquitin zinc finger) domain and a C terminus nuclease domain, associates with monoubiquitinated FANCD2. The initial identification of FAN1 as a FA-like factor came from different approaches, as it was identified from a genome-wide shRNA screen searching for ICL sensitizer (Smogorzewska et al. 2010), as a binding partner of FANCD2 (Liu et al. 2010), as a binding partner of MLH1 mismatch repair protein (Kratz et al. 2010), and through a bioinformatic search for the UBZ and nuclease domain-containing proteins (Mackay et al. 2010). The groups collectively found that FAN1 (1) possesses a nuclease activity that is required for cellular resistance against ICL agents, (2) is recruited to damaged DNA via its UBZ domain and monoubiquitinated FANCD2, and (3) associates rather specifically with monoubiquitinated FANCD2. As discussed throughout this review, one important function of monoubiquitinated FANCD2 is the promotion of HR. Indeed, depletion of FAN1 resulted in reduced HR efficiency in human cells (Mackay et al. 2010), although the formation of RAD51 or RPA foci was not affected, suggesting that FAN1 might have a role downstream in the HR process.

Collectively, a model can be assembled that FAN1 is recruited to the damaged sites via specific interaction between the UBZ domain of FAN1 and monoubiquitinated FANCD2, followed by the nucleolytic cleavage of DNA (potentially near ICL), which is necessary for subsequent ICL repair. Exactly at which step FAN1 acts to promote ICL repair remains unknown. However, as the Walter laboratory (Knipscheer et al. 2009) demonstrated using Xenopus extracts, monoubiquitinated FANCD2 is required for the nucleolytic incision and unhooking of the ICL. An intriguing question is whether FAN1 represents the cryptic nuclease activity observed in the in vitro system. However, as there are several other nucleases proposed to act during ICL (such as the MUS81–EME1, XPF–ERCC1, SLX4 complex, and the FANCD2 nuclease itself) (Niedernhofer et al. 2005; Knipscheer et al. 2009; Moldovan and D'Andrea 2009; Pace et al. 2010), the exact functions of each nuclease in promoting ICL repair is yet to be elucidated.