The Fanconi Anemia Protein Interaction Network: Casting A Wide Net

Abstract

It has long been hypothesized that a defect in the repair of damaged DNA is central to the etiology of Fanconi anemia (FA). Indeed, an increased sensitivity of FA patient-derived cells to the lethal effects of various forms of DNA damaging agents was described over three decades ago [1–3]. Furthermore, the cytological hallmark of FA, the DNA crosslink-induced radial chromosome formation, exemplifies an innate impairment in the repair of these particularly cytotoxic DNA lesions [4]. Precisely defining the collective role of the FA proteins in DNA repair, however, continues to be one of the most enigmatic and challenging questions in the FA field. The first six identified FA proteins (A, C, E, F, G, and D2) harbored no recognizable enzymatic features, precluding association with a specific metabolic process. Consequently, our knowledge of the role of the FA proteins in the DNA damage response has been gleaned primarily through biochemical association studies with non-FA proteins. Here, we provide a chronological discourse of the major FA protein interaction network discoveries, with particular emphasis on the DNA damage response, that have defined our current understanding of the molecular basis of FA.

1. RAD51 (December, 1997)

Perhaps the earliest-defined, and arguably most important, interaction between a FA protein and a non-FA protein is that of FANCD1/BRCA2 and RAD51. RAD51 is the mammalian homologue of the Escherichia coli RecA protein, and catalyzes the critical strand invasion step of homologous recombination (HR) DNA repair. HR is an essential cellular DNA double-strand break (DSB) repair mechanism that depends on the existence of an intact ‘homologous’ template sequence on the sister chromatid or homologous chromosome, to copy and synthesize lost or damaged genetic information (for reviews see [5,6]). In 1997, a yeast two-hybrid screen using a T-cell cDNA library revealed that a carboxy-terminus fragment of murine Brca2 could interact with Rad51 [7]. Murine Brca2 and Rad51 were also found to be temporally and spatially co-expressed during early embryogenesis. Furthermore, Rad51−/− and Brca2−/− mice were characterized by early embryonic lethality, while Rad51−/− and Brca2−/− embryos were hypersensitive to the cytotoxic effects of ionizing radiation (IR) [7–9]. In the same year, Mizuta et al described a direct interaction between the human RAD51 protein and murine Brca2 using yeast two-hybrid as well as GST pull-down approaches [10]. The critical importance of these discoveries for both clinical and molecular aspects of FA would not be revealed until five years later when biallelic mutations in the BRCA2 gene were uncovered in two FA-D1 patients [11].

Numerous studies over the past decade have contributed to our current understanding of the mechanism by which the FANCD1/BRCA2 protein regulates the function of RAD51 in HR: FANCD1/BRCA2 encodes a 3418-amino-acid protein that contains eight evolutionarily conserved internal repeats known as BRC motifs. These BRC motifs mediate the direct binding of FANCD1/BRCA2 to RAD51 [12–15]. FANCD1/BRCA2 promotes RAD51 nucleoprotein filament formation and stimulates RAD51-mediated strand exchange (for reviews see [16,17]). The recently identified FANCN/PALB2 protein (for Partner and Localizer of BRCA2) binds to the amino-terminus of FANCD1/BRCA2 and promotes its stabilization in chromatin [18]. Accordingly, both FA-D1 and FA-N patient cells display severely attenuated RAD51 nuclear foci formation [19–21]. The role of the upstream FA proteins (A, B, C, E, F, G, L, and M), as well as FANCD2, in the regulation of RAD51 remains unclear [19,22–24].

Hypomorphic FANCD1/BRCA2 mutations underlie the overwhelming majority of FA-D1 patients characterized to date [11,25,26]. This is consistent with the observation that disruption of murine Fancd1/Brca2 results in early embryonic lethality [7]. Furthermore, the clinical phenotypes of FA-D1 (as well as FA-N) patients are typically markedly more severe than classic FA, and are characterized by pronounced susceptibility to early childhood cancers [11,20,21,27]. In total, 23 biallelic FANCD1/BRCA2−/− patients have been described to date. Among these FA-D1 patients, five cases of Wilms tumor, nine brain tumors, including medulloblastoma, glioblastoma, and astrocytoma, seven cases of acute myeloid leukemia (AML), and three cases of acute lymphoblastic leukemia (ALL), have been recorded [11,25,26,28]. The increased clinical severity of FA-D1 and FA-N patients may be attributable to the direct roles of the FANCD1/BRCA2 and FANCN/PALB2 proteins in the regulation of essential RAD51-dependent HR processes. Conversely, while the upstream FA proteins, as well as FANCD2, may not function directly in HR repair, it seems likely that they at least co-operatively promote this conservative, error-free DNA repair pathway under certain conditions.

2. BRCA1 (August, 1998)

An important association between the protein products of the two major familial breast cancer susceptibility genes, BRCA1 and FANCD1/BRCA2, was suggested by several studies in the latter half of the last decade [29]. In 1996, Rajan et al demonstrated that murine Brca1 and Fancd1/Brca2 mRNA expression were coordinately regulated during mammary epithelial proliferation and differentiation, suggesting that these proteins might function in overlapping pathways [30]. In addition, both proteins were demonstrated to interact with RAD51 [7,10,31]. Furthermore, Brca1−/− mice, like Fancd1/Brca2−/− and Rad51−/− mice, were also characterized by early embryonic lethality [7–9,32–34]. In 1998, Chen et al demonstrated that BRCA1 and FANCD1/BRCA2 co-immunoprecipitate, and co-localize in nuclear foci during S phase of the cell cycle. BRCA1, FANCD1/BRCA2, as well as RAD51 were also shown to co-localize on the axial element of developing synaptonemal complexes in human spermatocytes, strongly suggestive of a cooperative role for these proteins in both mitotic and meiotic HR processes [35].

The seminal FA breakthrough, establishing a biochemical connection between BRCA1 (and hence FANCD1/BRCA2 and RAD51) and the FA pathway, occurred early in 2001 with the positional cloning of the FANCD2 gene and the functional characterization of the FANCD2 protein [36,37]. Garcia-Higuera et al discovered that the FANCD2 protein was post-translationally modified via the covalent linkage of a single ubiquitin molecule to internal K561 [36]. The mono-ubiquitination of FANCD2 was demonstrated to be required for its translocation to discrete nuclear foci following exposure to both IR and UV-irradiation [36]. As the BRCA1 protein had previously been demonstrated to translocate to discrete nuclear foci, where it co-localized with known DNA repair proteins, the association between FANCD2 and BRCA1 was examined [31,38–40]. FANCD2 and BRCA1 were demonstrated to co-localize in nuclear foci, as well as co-immunoprecipitate, following exposure to IR [36]. FANCD2 was subsequently demonstrated to co-localize with BRCA1 and RAD51 during S phase of the cell cycle [41]. Thus, at the molecular level, the FA pathway was simultaneously and unequivocally linked to both ubiquitin-mediated post-translational modification and HR DNA repair.

Using several complementary approaches, the FANCA protein and BRCA1 were also shown to interact [42]. Indeed, a direct interaction between the amino-terminus of FANCA and a central portion of BRCA1 was established using yeast two-hybrid analysis. The FANCA-BRCA1 interaction was confirmed using in vitro transcription/translation and immunoprecipitation analyses. Furthermore, the association between FANCA and BRCA1 was demonstrated to be independent of DNA damage. Interestingly, using this yeast two-hybrid system, Folias et al failed to detect a direct interaction between FANCD2 and BRCA1, suggesting that FANCA may provide a structural link between these proteins [42].

Further highlighting the important connection between BRCA1 and the FA pathway, the FANCJ protein, a DEAH helicase, also known as BRIP1 (for BRCA1-interacting protein) was originally identified through its association with BRCA1 [43]. FANCJ binds directly to the carboxy-terminus BRCT repeats of BRCA1 and facilitates BRCA1s known function in DNA DSB repair [43,44]. It has been hypothesized that the FANCJ/BRIP1 helicase may play a role in the timely displacement of RAD51 from nucleoprotein filaments following RAD51-mediated strand exchange [45]. BRCA1 may directly regulate FANCJ/BRIP1 helicase activity, thereby preventing the premature termination of RAD51-dependent HR processes [45]. FANCJ/BRIP1 is also known to interact with several additional non-FA proteins including the mismatch repair proteins MLH1 and PMS2 [46]. Readers are referred to the Ali et al review article in this issue for a comprehensive description of FANCJ function.

BRCA1 and FANCD2 Mono-ubiquitination

As the BRCA1 protein harbors a RING finger domain, a domain associated with E3 ubiquitin ligase activity [47], Garcia-Higuera et al examined the effects of BRCA1 expression on the mono-ubiquitination of FANCD2 and its translocation to nuclear foci. Interestingly, in the hypomorphic BRCA1−/− breast cancer cell line HCC1937, both IR-inducible FANCD2 mono-ubiquitination and nuclear foci formation were impaired [36]. Correction of the HCC1937 cells with wild-type BRCA1 restored IR-inducible FANCD2 mono-ubiquitination and nuclear foci formation to wild-type levels [36]. The spontaneous mono-ubiquitination of FANCD2, however, remained intact in HCC1937 cells. Subsequent studies in the chicken DT40 B-lymphoblast line revealed that ablation of the RING domain of BRCA1 had no impact on DNA damage-inducible FANCD2 mono-ubiquitination [48]. These results suggest that BRCA1 may solely promote the accumulation of FANCD2 at sites of DNA damage following exposure to IR, independently of a direct role on FANCD2 mono-ubiquitination. Subsequent studies have since revealed that the FANCL protein, and not BRCA1, is likely to be the catalytic E3 ubiquitin ligase subunit of the core FA complex [48–50].

BRCA1, the FA Pathway and HR DSB Repair

Critical functional roles for the BRCA1 and FANCD1/BRCA2 proteins in HR have been established. For example, Brca1−/− and Fancd1/Brca2−/− mouse embryonic stem cells are markedly compromised in their ability to repair a defined DNA DSB via HR [51–54]. Consequently, several studies have examined the roles of the FA proteins in HR, and varying degrees of defective HR repair capacity have been reported [23,55–59]. There are several plausible explanations for the apparent disparity among studies addressing the role of the FA proteins in HR: 1) the nature of the mutation - hypomorphic FA patient cells versus DT40 null cells, 2) the cell system employed and the extent to which the HR pathway is utilized in these cells - DT40 cells have a high proportion of S phase cells at a given time, thus their dependence on HR may be greater [57], 3) the nature of the HR substrate - the majority of HR assays utilized to date have been based on the repair of a canonical two-ended DNA DSB [56–59], yet hypersensitivity to agents that generate two-ended DNA DSBs, such as IR, is not strictly a phenotype of FA cells. Thus, DNA DSB-based HR assays may not reveal the true extent of the FA HR phenotype. Nonetheless, the overwhelming majority of functional and biochemical studies to date support an important cooperative role for the FA proteins, and BRCA1 and RAD51 in a HR-dependent pathway. Elucidating the precise details of this role will continue to be a focus of major research attention. Furthermore, given the pervasive connections between the BRCA1 protein and the FA pathway, it surely remains a possibility that biallelic mutations in BRCA1 may yet be revealed in a FA patient.

3. ATM (May, 2002)

Bilallelic mutations in the ATM gene underlie the autosomal recessive disorder Ataxia-telangiectasia (AT) [60]. AT is clinically distinct from FA and is characterized by progressive cerebellar ataxia, telangiectases, immune defects, and increased susceptibility to early-onset hematologic cancers of B and T cell origin, as well as central nervous system tumors, astrocytomas, and medulloblastomas (Table 1) [61,62].

Table 1.

Clinical and cellular characteristics of Fanconi anemia and related chromosomal instability syndromes

Phenotype	AT	FA	NBS	BS	SCKL1
Congenital Abnormalities
Short stature	+	+	+	+	+
Radial ray defects	−	+	−	−	−
Microcephaly	−	+	+	+	++
Hypo-/hyper-pigmentation	+	+	+	+	+
Facial dysmorphism	−	+	+	+	+
Mental retardation	−	−	+	±	+
Immunodeficiencies	++	−	++	+	−
Hematological Tumors
Lymphoid tumors	+	−	+	+	−
Myeloid tumors	−	+	−	+	+
Non-hematological Tumors
Solid tumors	+	+	+	+	+
Chromosome Anomalies
Irradiation-breakage	++	±	++	+	−
Mitomycin C-breakage	−	++	+	−	+
Sister chromatid exchanges	−	±	−	+++	−

AT, Ataxia-telangiectasia; FA, Fanconi anemia; NBS, Nijmegen breakage syndrome; BS, Bloom syndrome; SCKL1, Seckel syndrome

At the cellular level AT is defined by hypersensitivity to the lethal effects of DNA damaging agents that generate DNA DSBs. The ATM protein is a member of the phosphatidylinositol-3-OH-kinase-like protein kinase (PIKK) family. ATM recognizes and phosphorylates the consensus sequence SQ/TQ in target proteins, and activates cell cycle checkpoint responses in response to DNA damage (for reviews see [63,64]). For example, the ATM kinase phosphorylates T68 of the CHK2 kinase following exposure to IR [65]. The CHK2 kinase subsequently phosphorylates S123 of the CDC25A phosphatase rapidly targeting it for ubiquitin-mediated proteasomal degradation. The resulting persistent inhibitory phosphorylation of the cyclin-dependent kinase CDK2 on T14 and Y15 blocks entry into S phase, thereby halting cell cycle progression [66,67].

While hypersensitivity to the lethal effects of IR is not a hallmark of FA patient cells, sensitivity to the clastogenic effects of IR has been observed [68,69]. Furthermore, several incidences of clinical radiosensitivity have been reported for FA patients, for example following radiation therapy for solid tumors, as well as during bone marrow transplantation preparative regimens [70,71]. In an effort to determine if the FA and AT pathways converge in the cellular response to IR, Taniguchi et al examined the response of FA-D2 patient cells to IR [72]. FA-D2 patient cells were demonstrated to be moderately sensitive to the lethal effects of IR as well as defective in the IR-inducible S phase checkpoint [72]. This latter phenotype is referred to as radioresistant DNA synthesis (RDS), and reflects a failure to arrest DNA synthesis in the presence of chromosomal DNA DSBs. The RDS phenotype is a hallmark of AT cells, patient cells from the related cancer susceptibility syndromes Nijmegen breakage syndrome (NBS) (see below and Table 1) and Ataxia-telangiectasia-like disorder (ATLD), as well as BRCA1−/− cells [73–75]. Taniguchi et al also demonstrated that the ATM kinase phosphorylates FANCD2 on S222 and S1404 in vivo. The ATM-dependent phosphorylation of FANCD2 S222 occurs 1 to 2 h following exposure to IR, and is required for the establishment of the IR-inducible S phase checkpoint [72]. However, ATM-mediated FANCD2 S222 phosphorylation is dispensable for FANCD2 mono-ubiquitination, nuclear foci formation, as well as resistance to mitomycin C (MMC) [72]. Thus, it appears that the ATM-dependent phosphorylation of FANCD2 and the mono-ubiquitination of FANCD2 are functionally dissociable post-translational modification events. Surprisingly, despite the established interaction between FANCD2 and BRCA1 [36], and BRCA1s important role in the repair of DNA DSBs by HR [51], the ATM-dependent phosphorylation of FANCD2 on S222 does not appear to be required for DNA DSB HR [56]. Nevertheless, these findings strongly suggest an important functional interaction between the FA and AT pathways, and strengthen the need for prudence in the use of radiation therapy for the clinical management of FA [70].

4. NBS1 (December, 2002)

Nijmegen breakage syndrome (NBS), another rare autosomal recessive chromosomal instability syndrome, is caused by biallelic mutations in the Nibrin (NBS1) gene [73,76,77]. Clinically, NBS is characterized by congenital abnormalities including receding mandible, prominent mid-face, short stature, microcephaly, and progressive mental retardation [78]. Unlike FA, immunodeficiency, e.g. decreased serum levels of IgA and IgG, and increased susceptibility to agammaglobulinemia and lymphopenia, is also a feature of NBS [79]. NBS patients are highly susceptible to early-onset hematologic cancers, however the spectrum of tumors differs substantially from that observed for FA patients. For example, NBS patients are highly susceptible to non-Hodgkin’s lymphoma, large B-cell lymphoma, as well as T-cell lymphoma (Table 1) [78,80].

The NBS1 protein is a member of the highly conserved MRN (MRE11-RAD50-NBS1) protein complex. The MRN complex plays a major role in the sensing and signaling of DNA DSBs, and orchestrates several cellular mechanisms to facilitate their repair, including HR, non-homologous end joining (NHEJ), as well as cell cycle checkpoint activation (reviewed in [81]). The MRN complex plays both a structural and enzymatic role in the repair of DNA DSBs: The complex binds to DNA as a heterotetramer and tethers broken DNA ends or sister chromatids [82]. Furthermore, MRE11 possesses both endonuclease and exonuclease activities that are stimulated upon complex assembly [83,84]. The MRN complex acts upstream of the ATM kinase in the response to DNA DSBs, and facilitates its activation by promoting its autophosphorylation on S1981 [85,86].

Intriguingly, a connection between the FA and NBS pathways was uncovered when a NBS patient, EUFA1020, was diagnosed with atypical-FA, on the basis of FA-like congenital abnormalities and heightened MMC-induced chromosome breakage [87]. EUFA1020 cells harbor an atypical homozygous NBS1 1089C>A mutation, and fail to express full-length NBS1 protein [87]. Like AT, the hallmark of NBS patient cells is hypersensitivity to the lethal effects of agents that generate DNA DSBs. However, Nakanishi et al revealed that MMC-hypersensitivity is also a common feature of NBS, as well as ATLD (MRE11−/−), yet not AT, patient lines [87]. Accordingly, MMC was demonstrated to induce the formation of NBS1, MRE11, and FANCD2 nuclear foci. Moreover, these proteins displayed a high degree of co-localization following exposure to both MMC and IR [87]. However, as MRE11 nuclear foci formed in FA-D2 cells, and FANCD2 nuclear foci formed in both NBS and ATLD cells, these proteins appear to have distinct mechanisms of assembly. The carboxy-terminus of NBS1 was demonstrated to be required for cellular resistance to MMC, as a carboxy-terminus truncated NBS1 failed to localize to nuclear foci and failed to support the DNA damage-inducible assembly of the MRN complex [87]. MMC was also shown to induce the phosphorylation of NBS1 on S343, however this post-translational modification did not appear to be required for cellular MMC resistance. Interestingly, the ATM-mediated IR-inducible phosphorylation of FANCD2 S222 was demonstrated to be dependent on the presence of both NBS1 and MRE11, as it could not be detected in cells from NBS or ATLD patients, respectively, consistent with the observation that the MRN complex acts upstream of ATM [86,87]. In a separate study, the phosphorylation of FANCD2 upon treatment with the DNA interstrand crosslinking agent photoactivated 8-methoxypsoralen (8-MOP/UV-A) was also shown to be defective in NBS patient cells [88]. Furthermore, short-interfering RNA (siRNA)-mediated depletion of MRE11 in FA-D2 patient cells did not further exacerbate their S phase checkpoint defect, suggesting that the MRN and FA pathways may be epistatic in the response to DNA crosslinking agents [88].

Thus, while the ATM kinase appears to solely regulate FANCD2s function in the IR-inducible S phase checkpoint, the MRN complex may play dual, yet biochemically-distinct, roles in the regulation of the function of FANCD2 in the IR-inducible S phase checkpoint and in the cellular response to DNA crosslinking agents. It remains to be determined if a structural or enzymatic activity, or both, of the MRN complex underlies its function in DNA crosslink repair. Furthermore, the molecular nature and functional significance of the interaction between FANCD2 and the MRN complex in DNA crosslink repair remains to be clearly determined.

5. BLM (May, 2003)

Bloom Syndrome (BS), caused by biallelic mutations in the BLM gene [89,90], is a rare recessive disorder characterized by pre- and post-natal growth retardation, narrow facies, telangiectasia, hypo- or hyper-pigmentation of the skin, and an increased susceptibility to cancers, including acute leukemias, lymphomas, and medulloblastomas (Table 1) [91]. The protein encoded by the BLM gene is a member of the highly conserved RecQ family of helicases that also includes the WRN and RECQ4 proteins [92]. Mutations in the WRN and RECQ4 genes underlie the progeroid disorders Werner’s syndrome and Rothmund-Thomson syndrome, respectively [93,94]. As is the characteristic of the majority of RecQ helicases, the BLM helicase has ATP-dependent 3′-5′ DNA helicase activity. The in vivo substrate of the BLM helicase is thought to be a branched DNA structure, such as a Holliday junction, that can arise during HR repair [95]. Accordingly, BS patient cells are characterized by markedly elevated levels of sister chromatid exchanges (SCEs) as well as gross chromosomal rearrangements (GCRs) [96].

In 2003 Meetei et al described the purification of an endogenous BLM-associated multiprotein complex from HeLa cells by direct immunoprecipitation with a BLM-specific antibody [50]. Surprisingly, along with the known BLM-interacting proteins Topo IIIα and RPA, five of the core FA complex proteins, FANCA, FANCC, FANCE, FANCF, and FANCG, were demonstrated to associate with the BLM helicase [50]. Using a similar approach, Meetei et al next purified a FANCA-associated multiprotein complex from HeLa cells by direct immunoprecipitation with a FANCA-specific antibody. The FANCA-associated protein complex included BLM, Topo IIIα, RPA70, the five FA core complex proteins (A, C, E, F, and G), as well as five additional FANCA-associated proteins (or FAAPs), FAAP300, FAAP100, FAAP90, FAAP75, and FAAP43. In the presence of 0.7 M NaCl, BLM, Topo IIIα, RPA, and FAAP75 dissociated from the FANCA-associated multiprotein complex, however, FANCA, FANCC, FANCE, FANCF, FANCG, FAAP300 (or 250), FAAP100, FAAP90, and FAAP43 remained tightly associated, revealing the true composition of the FA core complex [49]. In a scientific tour de force by the Wang laboratory, as well as others, FAAP300, FAAP90, and FAAP43 were subsequently determined to be FANCM, FANCB, and FANCL, respectively [49,97,98].

Extending upon these important findings, several groups have attempted to elucidate the functional significance of the BLM-FA pathway interaction [55,99]. In 2004 Pichierri et al demonstrated that BS and FA patient cells are similarly characterized by hypersensitivity to the lethal effects of DNA crosslinking agents as well as agents that inhibit DNA replication [99]. FANCD2 and BLM were demonstrated to co-localize in nuclear foci following exposure to 8-MOP/UV-A as well as the ribonucleotide reductase inhibitor hydroxyurea (HU). Similarly, in immunoprecipitation experiments BLM preferentially associated with mono-ubiquitinated FANCD2. Significantly, the localization of BLM to nuclear foci, and its phosphorylation, were markedly compromised in FA-C and FA-G patient cells, following treatment with 8-MOP/UV-A [99]. Conversely, FANCD2 mono-ubiquitination and nuclear foci formation, as well as the DNA damage-inducible chromatin localization of the FA core complex proteins FANCA and FANCC, remained intact in BS patient cells [99].

Consistent with that reported by Pichierri et al, using DT40 cells Hirano et al demonstrated that FANCD2 and GFP-labeled BLM co-localized in nuclear foci, and co-immunoprecipitated, following exposure to MMC. Furthermore, FANCC and FANCD2 were demonstrated to be required for the efficient MMC-inducible localization of BLM to nuclear foci [55]. Similar to that observed by Pichierri et al, neither FANCD2 nuclear foci formation nor FANCD2 mono-ubiquitination were impaired in DT40 BLM−/− cells [55]. Hirano et al also demonstrated that DT40 FANCC-deficient cells display an approximately 2-fold increased frequency of spontaneous SCEs [55]. However, no differences in SCE frequency or cell survival were observed between the BLM−/− mutant and the FANCC/BLM−/− double mutant, in the absence or presence of MMC [55]. Moreover, similar findings were observed for a FANCD2/BLM−/− double mutant [55].

Taken together, three major conclusions can be drawn from these functional studies: 1) the BLM helicase and the FA pathway appear to be epistatic in the cellular response to DNA crosslinking agents, 2) the BLM helicase acts downstream of the mono-ubiquitination of FANCD2, and 3) the translocation of the BLM helicase to nuclear foci following exposure to DNA crosslinking agents is dependent on the presence of FANCD2, FANCC, and FANCG. Interestingly, a BLM-associated immune complex isolated from a FA-A patient cell line retained wild type helicase activity, suggesting that the core FA complex and FANCD2 mono-ubiquitination are not required for BLM helicase activity [50]. Furthermore, the spontaneous frequency of SCEs of FANCC−/−, FANCD2−/−, and FANCG−/− DT40 cells is substantially lower than that of BLM−/− cells, indicating that BLM function is only partially dependent on the FA pathway. Indeed, the FA core complex proteins were detected in only one of three large multi-subunit BLM complexes [50]. Multiple roles for the BLM helicase in the repair of damaged DNA replication forks have been proposed, including replication fork regression as well as the resolution of double Holliday junctions to suppress the formation of crossover recombinants (for reviews see [100,101]). It remains highly important to determine if and how the FA proteins and the BLM helicase act in concert to resolve recombinogenic DNA structures that arise during the repair of arrested DNA replication forks.

6. ATR (August, 2004)

Seckel syndrome (SCKL) is also a rare autosomal recessive disorder characterized by severe growth retardation, proportionate short stature, pronounced microcephaly, mental retardation, and a characteristic ‘bird-headed’ facial appearance (Table 1) [102]. Like FA, SCKL1 syndrome, caused by biallelic mutations in the ATR (for ATM and Rad3-related) gene [103,104], is associated with hematologic abnormalities including acute myeloid leukemia (AML), myelodysplasia (MDS), and aplastic anemia [105,106].

Like ATM, the ATR kinase is also a member of the PIKK protein family. While the ATM kinase responds primarily to DNA DSBs arising throughout the cell cycle, the ATR kinase responds primarily to perturbations of cellular DNA replication, and is the major cellular regulator of the DNA replication checkpoint [107,108]. Examples of genotoxic agents that activate the ATR kinase include HU, the DNA polymerase inhibitor aphidicolin (APH), as well as UV irradiation, which promotes the formation of cyclobutane pyrimidine dimers. Following exposure to UV irradiation the ATR kinase is known to phosphorylate the CHK1 kinase on S317 and S345 [109]. Activated CHK1 phosphorylates the CDC25A phosphatase, targeting CDC25A for poly-ubiquitination by the SCF (for SKP1-Cul1-F-box protein) complex, and subsequent proteasomal degradation [110,111]. Reduction of CDC25A levels slows S phase progression by preventing the dephosphorylation of CDK2 (for reviews see [108,112]).

In support of an important interaction between the ATR kinase and the FA pathway, two independent groups have demonstrated that the ATR and FANCD2 proteins co-localize in nuclear foci following treatment with DNA crosslinking agents [88,113]. Andreassen et al also established that the siRNA-mediated depletion of ATR results in the abrogation of DNA damage-inducible FANCD2 mono-ubiquitination and nuclear foci formation, independent of changes in cell cycle distribution [113]. However, spontaneous FANCD2 mono-ubiquitination remained unaffected. Furthermore, like FA cells, marked increases in levels of MMC-inducible chromosome radial formations were observed in metaphase spreads from ATR-depleted cells. These defects were recapitulated in ATR-deficient SCKL1 patient cells, consistent with a previous report describing the hypersensitivity of SCKL1 cells to the lethal effects of MMC [104,113].

In support of a direct regulatory interaction between the ATR kinase and FANCD2, using an in vitro kinase assay, Andreassen et al demonstrated that the wild-type ATR kinase, and not a kinase-dead mutant, could phosphorylate several GST-tagged FANCD2 peptide fragments [113]. Similarly, Pichierri et al demonstrated that the 8-MOP/UV-A-inducible phosphorylation of FANCD2 was dependent on the presence of a functional ATR kinase domain [88]. ATR was subsequently demonstrated to phosphorylate FANCD2 on T691 and S717 in vitro and in vivo [114]. FANCD2 T691 was also subject to ATM-mediated phosphorylation following exposure to IR, as well as during S-phase of the cell cycle [114]. Interestingly, the FANCD2 T691A/S717A double mutant failed to undergo efficient DNA damage-inducible mono-ubiquitination and failed to correct the MMC hypersensitivity of FA-D2 patient cells [114]. Thus, the DNA damage-inducible ATR/ATM-mediated phosphorylation of FANCD2 on T691 and S717 promotes FANCD2 mono-ubiquitination and DNA repair activity [113,114]. While the underlying mechanism is unknown, the phosphorylation of FANCD2 on T691 and S717 may stimulate its interaction with the FA core complex members FANCE or FANCL, and/or promote its stabilization in chromatin [49,114–116].

Not surprisingly, the role of the ATR kinase in the regulation of the FA pathway is not confined to FANCD2. Indeed, a role for both the ATR and ATM kinases in the phosphorylation of FANCI has also recently been suggested. Using a combined SILAC (stable isotope labeling with amino acids in cell culture)-mass spectrometry approach, three ATM/ATR SQ/TQ phosphorylation sites in the FANCI protein (S730, T952, and S1121) were demonstrated to be phosphorylated following exposure to IR [117]. In addition, a role for the ATR kinase in the phosphorylation of FANCG on S7 has recently been described [118,119]. Furthermore, several FA proteins are known to undergo both cell cycle-regulated and/or DNA damage-inducible phosphorylation, e.g. FANCD1/BRCA2, FANCG, FANCJ/BRIP1, and FANCM [24,98,120,121]. The functional significance of the phosphorylation of these proteins is not fully understood. Moreover, the contribution of the ATR kinase to these phosphorylation events remains unknown. Given the critical role of the ATR kinase in the maintenance of genomic stability during DNA replication [107], and the mounting functional evidence that the intrinsic function of the FA pathway resides in the resolution of damaged DNA replication forks [122–124], it is highly likely that the full contribution of this essential kinase to the regulation of the FA pathway has yet to be revealed.

Finally, several additional proteins in the immediate ATR network also play important, albeit most likely indirect, roles in the activation of the FA pathway. The replication protein A (RPA) complex binds to single-stranded DNA (ssDNA) and is required for the recruitment and activation of ATR [125]. Accordingly, siRNA-mediated depletion of the 70 kDa subunit of the RPA complex, RPA1, strongly impaired DNA damage-inducible FANCD2 mono-ubiquitination and nuclear foci formation [113]. Furthermore, the 32 kDa subunit of the RPA complex, RPA2, co-localizes with FANCD2 in nuclear foci following exposure to MMC, HU, and APH [113,122]. These results strongly suggest that the primary stimulus for activation of the FA pathway may be extended ssDNA regions. In addition, similar to that observed for ATR and RPA, Collis et al recently demonstrated that DNA damage-inducible FANCD2 mono-ubiquitination and nuclear foci formation are markedly impaired in cells depleted of the HCLK2 protein [126]. HCLK2 is the human homologue of the Caenorhabditis elegans biological clock protein CLK-2. HCLK2 associates with ATR and CHK1 and is required for the DNA replication checkpoint [126]. The role of HCLK2 in the activation of the FA pathway was conserved in nematodes, as DNA damage-inducible FANCD2 nuclear foci formation was completely abrogated in two temperature-sensitive C. elegans clk-2 mutants at the restrictive temperature [126]. However, the mechanism by which HCLK2 modulates the ATR-mediated DNA replication stress response remains to be determined.

7. The yeast RAD6 Epistasis Group (circa June, 2004)

In the early 1990s Papadopoulo et al reported that FA-A patient cells have a reduced mutation frequency at the X-linked hypoxanthine phosphoribosyl transferase (HPRT) locus following exposure to DNA crosslinking agents [127]. A subsequent study from the same group revealed that the majority of FA HPRT- mutants had small exonic deletions, compared with a preponderance of point mutations in normal (non-FA) HPRT- mutants. It was therefore proposed that FA might be characterized by a defect in a mutagenic pathway that involves DNA lesion bypass and subsequent gap filling by a HR-dependent process [128]. The hypothesis that the FA pathway might be involved in such a cellular DNA damage tolerance mechanism has been the focus of much recent research attention [57,129,130].

DNA damage tolerance mechanisms facilitate the timely completion of DNA replication in the presence of DNA damage. DNA lesions can be bypassed in an error-free or error-prone manner in a highly conserved process known as translesion DNA synthesis (TLS), carried out by members of the budding yeast Saccharomyces cerevisiae RAD6 epistasis group. The human counterparts include PCNA, several low fidelity Y- and B-family DNA polymerases, e.g. REV1, DNA Pol η/XPV, and DNA Pol ζ (REV3/REV7), as well as regulatory ubiquitin-modifying enzymes, e.g. RAD6 and RAD18 (for reviews see [131,132]). REV1 possesses deoxycytidyl nucleotidyl transferase activity and inserts a dCMP residue opposite a wide variety of DNA lesions [133]. DNA Pol ζ functions primarily to extend mismatched primer termini, and can also insert nucleotides opposite a variety of DNA lesions [134,135]. Conversely, DNA Pol η/XPV, mutated in Xeroderma pigmentosum variant, can insert the correct nucleotides opposite a potentially mutagenic UV irradiation-induced cyclobutane pyrimidine dimer [136–138]. Genetic analyses have revealed that REV1 and Pol ζ underlie the vast majority of mutagenic DNA polymerase activity in S. cerevisiae (for review see [139]). The RAD6 and RAD18 proteins, an E2 ubiquitin-conjugating enzyme and E3 ubiquitin ligase, respectively, play a major role in error-free TLS by promoting the mono-ubiquitination of PCNA [140]. The mono-ubiquitination of PCNA is hypothesized to signal a switch from DNA Pol δ-mediated processive DNA replication to Pol η/XPV-mediated TLS [141,142].

PCNA

Proliferating Cell Nuclear Antigen (PCNA), the human functional homolog of the β subunit of E. coli DNA Pol III, is an essential processivity factor for the replicative DNA polymerases. The ring-shaped homotrimeric PCNA complex encircles double-stranded DNA and can tether DNA polymerases, thereby increasing their processivity by several orders of magnitude. PCNA is also known to function as a mobile interaction platform for several DNA repair proteins, e.g. MSH3, XPG, REV1 and DNA Pol ζ [134,143–146]. Moreover, studies in S. cerevisiae have revealed that the post-translational modification of PCNA via the covalent conjugation of ubiquitin or the small ubiquitin-like modifier SUMO orchestrates distinct DNA repair mechanisms necessary for the maintenance of DNA replication fork stability [147,148].

Several lines of evidence have suggested an important interaction between the FA pathway and PCNA. FANCD1/BRCA2, like BRCA1 and RAD51, has previously been demonstrated to localize to PCNA-positive nuclear foci following treatment with HU and UV irradiation [35,149,150]. Similarly, a striking co-localization between FANCD2 and PCNA in nuclear foci is observed following treatment of cells with either HU or APH [122,151]. Furthermore, while FANCD2 and PCNA are mono-ubiquitinated by different E3 ubiquitin ligases, both proteins are de-ubiquitinated by the USP1 enzyme [152,153]. Recent results from our laboratory have established an important physical and functional interaction between FANCD2 and PCNA. In agreement with immunofluorescence analyses, FANCD2 and PCNA co-immunoprecipitate in chromatin following exposure to DNA damaging agents. In an effort to identify domains of FANCD2 important for this interaction, we have uncovered a highly conserved putative PCNA-interaction motif (PIP-box) in FANCD2. Interestingly, mutation of the FANCD2 PIP-box motif disrupts the association between FANCD2 and PCNA, and precludes both the spontaneous and DNA damage-inducible mono-ubiquitination of FANCD2. Consequently, the FANCD2 PIP-box mutant fails to correct the MMC-hypersensitivity of FA-D2 patient-derived fibroblasts. Our results suggest that PCNA may transport FANCD2 to sites of stalled DNA replication forks, and may facilitate its mono-ubiquitination by the FA E2/E3 holoenzyme in chromatin.

REV1 & REV3

In 2004 Niedzwiedz et al examined the effects of loss of the DT40 FANCC protein on both HR and TLS. The V gene locus of this B-cell line encodes for the variable antigen-binding site of a surface immunoglobulin (sIgM). The variability of the sIgM antigen-binding site arises as a consequence of three different modes of DNA repair of an endogenously generated abasic site: standard non-mutational repair, templated gene conversion, i.e. HR, and untemplated mutation, i.e. TLS. An approximately 3-fold reduction in templated gene conversion events at the V gene locus was observed for the FANCC−/− cell line, consistent with the proposed role for the FA pathway in HR. However, disruption of the DT40 FANCC gene also resulted in a reduced frequency of untemplated mutations at this locus, suggesting that the FA pathway may additionally promote TLS [57]. It had previously been demonstrated that, like FA cells, DT40 REV1−/− and REV3−/− cells are hypersensitive to the lethal effects of DNA crosslinking agents [154,155]. In support of an important functional interaction between the FA pathway and the TLS polymerases, Niedzwiedz et al demonstrated that FANCC, REV1, and REV3 are epistatic with respect to DNA crosslink-induced lethality [57]. Similarly, unlike wild type DT40 cells, exposure to DNA crosslinking agents failed to increase levels of SCEs in REV1−/−, REV3−/−, or FANCC−/− cells. In addition, the human FANCD2 protein and YFP-tagged REV1 were shown to co-localize in nuclear foci following inhibition of DNA synthesis [57]. However, neither REV1 nor REV3 disruption affected the mono-ubiquitination of FANCD2, indicating that these proteins most likely function downstream, or independently, of this post-translational modification step [57].

To further explore the role of the FA pathway in the regulation of TLS, Mirchandani et al recently examined point mutagenesis in FA-A, FA-G, and FA-D2 patient cells using a bacterial SupF mutagenesis assay [129]. Interestingly, both FA-A and FA-G cells were shown to be hypomutable for spontaneous and UV-C induced mutations, similar to that previously observed by Papadopoulo and colleagues [127,128]. Conversely, FA-D2 cells, as well as cells depleted for FANCD2 using siRNA, were proficient for point mutagenesis. Mirchandani et al also demonstrated that FANCA and FANCG, yet not FANCD2, were required for the efficient assembly of the REV1 protein into nuclear foci, both in the absence and presence of UV-irradiation. The REV1 protein is known to contain an amino-terminus BRCT domain and a carboxy-terminus ubiquitin-binding motif (UBM), both of which have previously been demonstrated to be important for the recruitment of REV1 to nuclear foci [144,145,156,157]. Interestingly, Mirchandani et al demonstrated that the impaired recruitment of REV1 to nuclear foci in FA-G patient cells was not further exacerbated by mutation of its BRCT domain. As BRCT domains typically mediate phospho-peptide interactions, this result suggests that a phosphorylated FA core complex member may interact directly with the REV1 BRCT domain to promote its localization and/or stabilization in nuclear foci (Fig. 2A).

Fig. 2.

Models depicting the proposed functions of the FA protein interaction network in the repair of stalled or collapsed DNA replication forks. Upon encounter of a DNA lesion on the template strand during DNA replication several options are possible. (A) To ensure timely replication fork progression, the template strand DNA lesion can be bypassed in an error-prone REV1- and/or DNA Pol ζ-dependent TLS mechanism. The FA core complex is required for the assembly of REV1 nuclear foci, suggesting that the FA core complex may specifically promote error-prone REV1-dependent TLS. This pathway appears to be independent of both PCNA and FANCD2 mono-ubiquitination. The molecular details of FA core complex promoted REV1-dependent TLS remain to be fully elucidated. (B) The DNA replisome may arrest at the lesion and subsequently resume DNA synthesis downstream of the lesion. The ensuing post-replicative gap (or daughter-strand gap) may be repaired by a combined HR/TLS PCNA mono-ubiqutination-dependent mechanism. We propose that the mono-ubiquitination of FANCD2 and FANCI may also be required for this process by an as-yet-undefined mechanism.

RAD6 & RAD18

In 2005, Hirano et al examined the interaction between FANCC and RAD18 by disrupting the FANCC gene in RAD18−/− DT40 cells [55]. Surprisingly, unlike that observed for the FANCC/REV1−/− and FANCC/REV3−/− mutants [57], FANCC/RAD18−/− cells were more sensitive to the lethal effects of DNA crosslinking agents than either single mutant [55]. Furthermore, spontaneous SCEs were elevated in FANCC/RAD18−/− cells compared with the single FANCC−/− and RAD18−/− mutants [55]. In support of independent roles for the FA core complex and RAD6/RAD18 in the cellular response to DNA crosslinking agents, Mirchandani et al also demonstrated that the depletion of RAD6 or RAD18 was markedly more toxic to FANCG-deficient cells, compared with FANCG-proficient cells [129]. In the same study, DNA damage-inducible RAD18-dependent PCNA mono-ubiquitination was shown to be intact in FA-G patient cells, while DNA damage-inducible FANCD2 mono-ubiquitination was shown to be intact in Rad18−/− mouse fibroblasts. A recent report by Zhang and colleagues, however, has suggested a link between the human homologues of S. cerevisiae Rad6, HHR6A and HHR6B, alternatively known as UBE2A and UBE2B, respectively, and FANCD2 [158,159]. Depletion of both HHR6A and HHR6B using siRNA diminished both MMC- and UV-irradiation induced FANCD2 mono-ubiquitination, and sensitized cells to the lethal effects of MMC [159]. However, whether the effect of HHR6 on FANCD2 is direct or indirect remains to be established.

Collectively these results strongly suggest an important role for the FA core complex in the regulation of REV1-, and possibly DNA Pol ζ-, dependent error-prone TLS (Fig. 2A). This function appears to be independent of the mono-ubiquitination of both FANCD2 and PCNA. While FANCD2 mono-ubiquitination-independent functions for the FA core complex have previously been suggested [160], these studies provide the first molecular insight into a FA core complex-specific process [57,129]. Interestingly, it has recently been reported that PCNA mono-ubiquitination and REV1 function in temporally distinct TLS mechanisms [161]: the mono-ubiquitination of PCNA appears to be required for a post-replicative TLS process, for example daughter-strand gap repair, while REV1 is required for the timely maintenance of replication fork progression on damaged DNA [161]. As daughter-strand gap repair is thought to involve both TLS and a HR-mediated template switching process [57,162], the contribution of the FA pathway to this mode of stalled DNA replication fork restart is also worthy of investigation (Fig. 2B).

8. USP1 (February, 2005)

Previous studies had demonstrated that the mono-ubiquitination of FANCD2 is down-regulated during G2 phase of the cell cycle prior to cell division, suggesting that mono-ubiquitinated FANCD2 may be subject to regulation via de-ubiquitination or proteasomal degradation [41]. To investigate the regulation of FANCD2 mono-ubiquitination via de-ubiquitination, Nijman et al screened a de-ubiquitinating enzyme (DUB) RNA interference library targeting 55 known or putative DUBs [153]. Short hairpin RNA (shRNA)-mediated depletion of the ubiquitin-specific protease 1 (USP1) DUB resulted in a marked increase in spontaneous FANCD2 mono-ubiquitination. The effects of USP1 depletion on FANCD2 mono-ubiquitination were demonstrated to be independent of altered cell cycle progression, and dependent on the presence of an intact FA core complex. The over-expression of an USP1 active site mutant also resulted in the accumulation of mono-ubiquitinated FANCD2, further confirming that USP1 is an important regulator of the mono-ubiquitination of FANCD2. Moreover, in immunoprecipitation experiments USP1 was demonstrated to physically associate with FANCD2. Consistent with the fact that the mono-ubiquitination of FANCD2 promotes its chromatin association [115], the USP1 protein is constitutively localized to chromatin. Surprisingly, Nijman et al also demonstrated that the siRNA-mediated depletion of USP1 was protective against the clastogenic effects of MMC. However, as this was the only DNA repair analysis carried out, it was proposed that a persistence of mono-ubiquitinated FANCD2 could also have adverse cellular outcomes. Huang et al subsequently demonstrated that mono-ubiquitinated PCNA is also a substrate of USP1 [152]. Like that observed for mono-ubiquitinated FANCD2, depletion of USP1 resulted in elevated levels of mono-ubiquitinated PCNA. Consequently, as assessed by the SupF mutagenesis assay, an approximately two-fold increased mutation frequency was observed in the absence of USP1 [152]. These findings suggest that a persistence of mono-ubiquitinated PCNA may promote mutagenic TLS via the inadvertent recruitment of the TLS polymersases Pol η/XPV and/or Pol ι [142,152,163].

In an effort to further elucidate the mechanism of regulation of the de-ubiquitination of FANCD2, Oestergaard et al generated a DT40 USP1 null cell line [164]. As expected, ablation of USP1 resulted in constitutively elevated levels of both mono-ubiquitinated FANCD2 and PCNA. However, contrary to what was observed for the SupF mutagenesis assay for human cells [152], the level of point mutations at the endogenous V gene locus was unaffected by the absence of USP1 [164]. The USP1−/− cells also displayed hypersensitivity to both MMC and cisplatin, moderate sensitivity to UV light and were refractory to the lethal effects of ionizing radiation [164], contrasting with the protective effect of USP1 depletion observed in human cells [153]. Furthermore, elegant genetic epistasis experiments revealed that the elevated DNA crosslinking agent sensitivity of USP1−/− cells was attributable to persistent FANCD2 mono-ubiquitination, as opposed to persistent PCNA mono-ubiquitination. Thus, it seems likely that mono-ubiquitinated FANCD2 may represent the pre-dominant USP1 substrate. Intriguingly, in the absence of USP1, mono-ubiquitinated FANCD2 was constitutively present in chromatin, indicating that USP1 also plays a major role in regulating the chromatin-dissociation of FANCD2. Why a persistence of chromatin-associated FANCD2 might restrict DNA repair remains to be elucidated, however, one postulated explanation is that the prolonged association of mono-ubiquitinated FANCD2 with the site of DNA repair might preclude access of subsequent DNA repair factors [164].

Cohn et al subsequently purified a native USP1-associated protein complex using a two-step tandem affinity purification strategy and identified a novel component of this complex, USP1 associated factor 1 (UAF1) [165]. Stoichiometric amounts of USP1 and UAF1 were present in purified protein complexes. UAF1 was demonstrated to stabilize USP1 and to stimulate the enzyme activity of USP1 through formation of a stable heterodimer [165]. By using a fluorescent ubiquitin derivative Ub-AMC, Cohn et al demonstrated that UAF1 does not increase the affinity of USP1 for Ub-AMC, but robustly increases its catalytic rate. Furthermore, the USP1/UAF1 complex can readily, and rapidly, de-ubiquitinate mono-ubiquitinated FANCD2 in an in vitro de-ubiquitination reaction [165].

9. UBE2T (August, 2006)

The promoter region of the E2 ubiquitin-conjugating enzyme encoding gene UBE2T was previously demonstrated to be a target of the E2F transcription factor, suggesting that UBE2T might play an important role in cell cycle progression and proliferation [166]. In an effort to identify the UBE2T E3 ubiquitin ligase(s), Machida et al performed a yeast two-hybrid screen using a UBE2T bait plasmid [167]. Several interacting clones corresponding to the FANCL E3 ubiquitin ligase were isolated. Furthermore, one of the clones harbored only the carboxy-terminus PHD finger domain of FANCL, suggesting that this region was sufficient for the interaction with UBE2T. A direct interaction between UBE2T and the FANCL PHD domain was confirmed in GST pull-down assays. Furthermore, mutations in the FANCL PHD domain abrogated UBE2T-FANCL binding. Similar to the effects observed upon knockdown of FANCA and FANCL, siRNA-mediated depletion of UBE2T resulted in marked decreases in both spontaneous and MMC-inducible FANCD2 mono-ubiquitination, and disrupted FANCD2 nuclear foci formation. Accordingly, both spontaneous and MMC-inducible chromosome aberrations were elevated in UBE2T-depleted cells. The knockdown of several other members of the UBCH5 E2 ubiquitin-conjugating enzyme family had no effect on the mono-ubiquitination of FANCD2. Interestingly, UBE2T was also demonstrated to promote the auto- ubiquitination of FANCL in an in vitro ubiquitination assay. Mutations in the essential cysteine residue of the UBE2T active site C86 (the site of ubiquitin transfer from E1), as well as mutations in the PHD domain of FANCL resulted in the inactivation of UBE2T-stimulated FANCL auto-ubiquitination. Significantly, UBE2T was also demonstrated to be auto-mono-ubiquitinated in vivo on internal K91: Mutation of K91 or C86 abrogated UBE2T mono-ubiquitination. However, while the UBE2T K91R mutant could not be mono-ubiquitinated, it retained the ability to promote FANCL auto-ubiquitination. Moreover, the mono-ubiquitination of UBE2T was demonstrated to inhibit its E2 ubiquitin-conjugating activity. The existence of a regulatory feedback inhibition loop was also suggested by the observation that the mono-ubiquitination of UBE2T in a FA-L patient cell line was stimulated upon restoration of FANCL protein expression [167]. Thus, FANCL may restrict the activity of UBE2T, thereby inhibiting its own activity, providing an additional layer of regulation for the transient activation of the FA pathway.

Alpi and colleagues have recently disrupted the UBE2T gene in DT40 cells [168]. Confirming the findings of Machida et al, UBE2T−/− cells are completely defective for FANCD2 mono-ubiquitination and hypersensitive to cisplatin [168]. Alpi et al also demonstrated that the FA core complex remains intact and stable, and localizes to chromatin, in the absence of UBE2T [168]. Furthermore, UBE2T was demonstrated to be constitutively present on chromatin independent of DNA damage or cell cycle stage [168].

Intriguingly, the number of FA and FA-interacting proteins that undergo mono-ubiquitination in vitro and/or in vivo currently stands at seven - FANCD2, FANCI, FANCL, and UBE2T, as well as BRCA1, PCNA, and REV1. Thus, the coordinated regulation of the FA protein interaction network is likely to comprise a complex series of coupled mono-ubiquitination steps, mediated by the interaction of ubiquitin with conserved ubiquitin-binding domains (UBDs) (Fig. 2B) [169,170]. Defining the molecular details of this coupled mono-ubiquitination network will surely be a major focus of future research attention. Moreover, it remains a strong possibility that mutations in additional genes encoding proteins functioning in ubiquitin-mediated post-translational modification may underlie genetically undefined FA complementation groups.

10. H2AX (March, 2007)

The mammalian core histone variant H2AX is rapidly phosphorylated by members of the PIKK family, including ATM, ATR, and DNA PK_CS on a conserved carboxy-terminus S139 (S136 in mice) in response to DNA damage (for reviews see [171,172]). Phosphorylated H2AX (or γH2AX) plays a major role in the early recruitment of DNA repair proteins, including BRCA1 and NBS1, to sites of DNA damage [173]. Bogliolo et al therefore investigated the role of H2AX in the recruitment and activation of FANCD2 [174]. In contrast to that observed in wild type mouse embryonic fibroblasts (MEFs), FANCD2 failed to localize to sites of UV-irradiation damage in H2AX−/− MEFs. Furthermore, MEFs expressing a phosphorylation-defective H2AX mutant, H2AX^S136A/S136A, the site of ATR-dependent H2AX phosphorylation [175], also failed to support the formation of UV-inducible FANCD2 nuclear foci. Conversely, both spontaneous and DNA damage-inducible FANCD2 mono-ubiquitination were unaffected by the absence of H2AX or by mutation of H2AX S136. The requirement for H2AX in the DNA damage-inducible localization of FANCD2 to nuclear foci was verified in HeLa cells by depleting H2AX using siRNA. FANCD2 and H2AX were also demonstrated to co-immunoprecipitate, and this interaction was dependent on the phosphorylation of H2AX on S136, as well as enhanced upon exposure to DNA damaging agents. Surface plasma resonance experiments confirmed the increased affinity of FANCD2 for phosphorylated H2AX. Consistent with an important role for H2AX in facilitating the recruitment of FANCD2 to DNA damage-inducible nuclear foci, H2AX^−/− and H2AX^S136A/S136A MEFs were demonstrated to be hypersensitive to the clastogenic and cytotoxic effects of MMC. H2AX^−/− MEFs had previously been demonstrated to display elevated sensitivity to γ-irradiation [173]. Finally, as the hypersensitivity of H2AX^−/− and H2AX^S136A/S136A MEFs to MMC was not further increased by the siRNA-mediated depletion of FANCD2, it was proposed that FANCD2 and H2AX are epistatic in the cellular response to MMC [174]. Importantly, in agreement with the findings of Garcia-Higuera et al, Bogliolo et al demonstrated that the FANCD2-γH2AX interaction and the DNA damage-inducible localization of FANCD2 to nuclear foci were dependent on the presence of BRCA1 [36,174]. Collectively, these findings support a cooperative role for γH2AX and BRCA1 in the recruitment, or retention, of FANCD2 to sites of damaged DNA, independent of FANCD2 mono-ubiquitination. The molecular mechanisms by which γH2AX and BRCA1 mediate this process remain to be clearly elucidated.

11. CHK1 (April, 2007)

The serine/threonine protein kinase CHK1 plays a major role in the activation of the DNA replication checkpoint under conditions of unfavorable DNA synthesis [107,176]. As the ATR kinase phosphorylates CHK1 on S317 and S345 [109], and is required for the DNA damage-inducible mono-ubiquitination of FANCD2 [113], Wang and colleagues examined the role of this important effector kinase on the regulation of the FA pathway [177]. Two highly conserved CHK1 phosphorylation consensus sequences were discovered in the carboxy-terminus of the FANCE protein. Mutation of the key residues of these consensus sequences, T346 and S374, abolished the ability of FANCE to rescue the MMC-hypersensitivity of FA-E patient cells. In addition, FA-E cells expressing the FANCE T346A/S374A double mutant displayed elevated spontaneous levels of FANCD2 mono-ubiquitination and nuclear foci formation. Furthermore, DNA damage-inducible FANCD2 mono-ubiquitination and nuclear foci formation were impaired in these cells. Similarly, the siRNA-mediated depletion of CHK1 in HeLa cells resulted in increased basal levels of FANCD2 mono-ubiquitination and nuclear foci formation, as well as a marked attenuation of DNA damage-inducible FANCD2 mono-ubiquitination and nuclear foci formation, suggesting an important role for the CHK1 kinase in the regulation of the FA pathway. Similar findings have recently been described by two independent groups [126,178]. An important functional connection between the FA pathway and CHK1 in the repair of damaged DNA replication forks has also recently been suggested: the FA pathway and CHK1 have been demonstrated to be required for the maintenance of chromosome stability at common chromosomal fragile sites, regions of the genome prone to DNA replication fork collapse and chromosome breakage [122,176,179].

Wang et al also demonstrated that CHK1 directly phosphorylates GST-FANCE peptide fragments containing T346 and S374. In addition, FANCE T346 and S374 phosphospecific antibodies were generated and used to verify CHK1-mediated FANCE phosphorylation in vivo. T346-phosphorylated FANCE was also shown to assemble into discrete nuclear foci following exposure to UV irradiation, where it strongly co-localized with FANCD2. However, while FANCD2 nuclear foci persisted 8 h following exposure to UV irradiation, T346-phosphorylated FANCE nuclear foci could no longer be observed at this point, suggesting that phosphorylated FANCE may be subject to proteasomal degradation. In an elegant series of experiments, Wang et al confirmed that the CHK1-mediated phosphorylation of FANCE promotes its proteasomal degradation. As early as 2 h following exposure to UV-irradiation, wild-type FANCE protein levels decreased concomitantly with an increase in levels of phosphorylated CHK1 S317. Furthermore, the FANCE T346A/S374A double mutant was refractory to UV-irradiation induced proteasomal degradation. Thus, it was proposed that FANCE plays both positive and negative regulatory roles in the FA pathway: FANCE-mediated FA core complex assembly promotes the mono-ubiquitination of FANCD2, while CHK1-mediated FANCE phosphorylation and degradation subsequently prevents additional FANCD2 mono-ubiquitination, thereby possibly precluding inappropriate DNA repair activity.

12. TIP60 (April, 2008)

The role of chromatin remodeling complexes in the orchestration of the DNA damage response is a major focus of current research attention (for reviews see [180,181]). In an unbiased yeast two-hybrid screen using an amino-terminus FANCD2 fragment as bait, Hejna et al identified the chromatin-modifying enzyme TIP60 as a FANCD2-interacting partner [182]. TIP60 is a histone acetyltransferase that has previously been implicated in the DNA damage response: TIP60 acetylates p53 and ATM, and, like FANCD2, has also been shown to co-localize with γH2AX [183,184]. Indeed, the TIP60-mediated acetylation of ATM is required for the ATM-mediated phosphorylation of p53 and CHK2 [183,185], while TIP60-mediated acetylation of p53 is required for p53-dependent apoptosis [184]. The carboxy-terminus of TIP60 containing the acetyl-CoA binding site, and amino acids 248–672 of FANCD2, were required for this interaction. However, the TIP60-FANCD2 interaction occurred independently of the mono-ubiquitination of FANCD2. This interaction was confirmed using a number of different approaches, including GST pull-downs, co-immunoprecipitation, as well as immunofluorescence microscopy, demonstrating co-localization of these proteins in nuclear foci [182]. The depletion of TIP60 using siRNA sensitized cells to the cytotoxic effects of MMC, however FANCD2 mono-ubiquitination and nuclear foci formation were unaffected. Surprisingly, while the depletion of TIP60 did not lead to an elevated level of MMC-inducible radial chromosome formations, depletion of a TIP60-interacting protein TIP49 did. The effect of TIP49 depletion on the mono-ubiquitination of FANCD2, however, was not examined. Finally, Hejna et al depleted TIP60 in FA-A and FA-C patient fibroblasts and demonstrated that depletion of TIP60 did not further sensitize these cells to the cytotoxic effects of MMC, leading the authors to conclude that TIP60 and the FA core complex proteins FANCA and FANCC are epistatic in the cellular response to DNA crosslinking agents [182]. While this study establishes an important connection between a chromatin remodeling complex and the FA pathway, the molecular details of how this important histone acetyltransferase might regulate the pathway remain to be established. For example, it remains to be determined if FANCD2, FANCI, or an FA core complex protein is a target of TIP60-mediated acetylation, and whether the acetylation of these proteins modulates their mono-ubiquitination, phosphorylation, and/or function in the DNA damage response.

Consistent with an important role for chromatin remodeling in the licensing of the FA pathway for DNA repair, Otsuki et al also reported an interaction between the FANCA protein and the SWI/SNF complex component BRG1 [186]. The SWI/SNF multisubunit complex is an ATP-dependent chromatin-remodeling complex implicated in many cellular processes including DNA repair and transcription (for review see [180]). Using yeast two-hybrid analysis a carboxy-terminus fragment of FANCA was demonstrated to interact directly with domain II of BRG1. Immunoprecipitation and immunfluorescence analyses confirmed the FANCA-BRG1 interaction, however, the functional significance of this interaction was not established [180].

13. Conclusions

Over the course of the past decade we have witnessed remarkable progress towards our understanding of the molecular etiology of FA. These advancements have been achieved primarily through the identification and characterization of new FA genes as well as new FA-interacting proteins. It is hard to believe that over the course of approximately eight years the FA pathway has grown from five orphan proteins (A, C, E, F, and G) to a complex network comprising thirteen bona fide disease-associated FA proteins with at least as many non-FA interacting proteins (Fig. 1). An emerging hypothesis, supported by strong functional and biochemical findings, is that the FA protein interaction network plays a critical role in the stabilization of stalled DNA replication forks, and in the orchestration of appropriate repair mechanisms, including HR and TLS (Fig. 2). While DNA replication fork stalling and collapse are known to occur following treatment with DNA crosslinking agents [187,188], a similar, albeit lower, level of occurrence could also be expected under non-perturbed conditions as a consequence of endogenously generated DNA lesions. The erroneous or untimely repair of these structures could lead to the generation of complex, unbalanced chromosomal abnormalities, frequently observed in FA cancer cells. The determination of the endogenous source(s) of chromosomal instability in FA patient cells, as well as precisely defining the collective role of the FA proteins in DNA repair, will continue to provide hope for improved prophylactic as well as targeted therapeutic options for the management of this debilitating disease.

Fig. 1.

A graphical overview of the FA protein interaction network. The network is superimposed on the Massachusetts Bay Transportation Authority (MBTA) T map in recognition of where many of the FA protein interactions were discovered. This complex network is comprised of thirteen FA gene-encoded proteins as well as at least twenty-seven non-FA proteins. While a different biological function is depicted for each line, these functions are often overlapping, with many proteins playing important roles in several cellular functions. Light grey lines depict known or putative protein-protein interactions. TLS, translesion DNA synthesis. IR, ionizing radiation.

Table 2.

Characteristics of the current known interactions between the FA proteins and non-FA proteins

Protein	FANC proteins	Identification mechanism	Organism/Model System	Functional Significance
ATM	FANCD2	IF, IVA	H. sapiens	IR-inducible S phase checkpoint, IR-inducible FANCD2 phosphorylation
	FANCI	IP	H. sapiens	ND
ATR	FANCD2	IF, IVA	H. sapiens	DNA damage-inducible FANCD2 mono-ubiquitination and nuclear foci formation, FANCD2 phosphorylation
	FANCG	ND	H. sapiens	ND
	FANCI	IP	H. sapiens	ND
BLM	FANCA	IP	H. sapiens	DNA crosslink repair, DNA replication fork repair, FANCC, FANCD2, and FANCG required for efficient MMC-inducible localization of BLM to nuclear foci
	FANCD2	IF, IP	H. sapiens, G. gallus
BRCA1	FANCD1/BRCA2	IF, IP	H. sapiens, M. musculus	Mitotic and meiotic recombination, DNA crosslink repair
	FANCD2	IF, IP	H. sapiens	IR- and UV-inducible FANCD2 mono-ubiquitination and nuclear foci formation
	FANCA	IP, Y2H	H. sapiens	ND
	FANCJ/BRIP1	FW, IF, IP, GST-PD	H. sapiens	Homologous recombination, DNA crosslink repair
BRG1	FANCA	Y2H, IP, IF	H. sapiens	ND
CHK1	FANCE	IVA	H. sapiens	DNA damage-inducible FANCE phosphorylation and FANCD2 mono-ubiquitination, DNA crosslink repair
CLK2	FANCD2	ND	H. sapiens, C. elegans	DNA damage-inducible FANCD2 mono-ubiquitination and nuclear foci formation, DNA crosslink repair
H2AX	FANCD2	IF, IP	H. sapiens, M. musculus	DNA damage-inducible FANCD2 nuclear foci formation, Chromatin re-modeling, DNA crosslink repair
HHR6	FANCD2	ND	H. sapiens	DNA damage-inducible FANCD2 mono-ubiquitination, DNA crosslink repair
NBS1	FANCD2	IP, IF	H. sapiens	IR and ICL-inducible S phase checkpoints, IR-inducible FANCD2 phosphorylation, DNA crosslink repair
PCNA	FANCD2	IF, IP, IVA	H. sapiens	FANCD2 mono-ubiquitination and nuclear foci formation, DNA crosslink repair
RAD51	FANCD2	IF	H. sapiens	ND
	FANCD1/BRCA2	IF, Y2H, GST-PD	H. sapiens, M. musculus, C. elegans, U. maydis	FANCD1/BRCA2 Promotes RAD51 nucleoprotein filament formation and promotes RAD51-mediated DNA strand exchange
REV1	FANCA	ND	H. sapiens, G. gallus	FANCA and FANCG promote REV1 nuclear foci formation and REV1-dependent TLS
	FANCG	ND
	FANCD2	IF		ND
RPA	FANCD2	IF	H. sapiens	DNA damage-inducible FANCD2 mono-ubiquitination and nuclear foci formation, DNA crosslink repair
TIP60	FANCD2	Y2H, GST-PD, IP, IF	H. sapiens	Chromatin re-modeling, DNA crosslink repair
UBE2T	FANCL	Y2H, GST-PD, IVA	H. sapiens, G. gallus	E2 ubiquitin-conjugating enzyme, promotes FANCD2 mono-ubiquitination
USP1	FANCD2	IP, IVA	H. sapiens, G. gallus	De-ubiquitination of mono-ubiquitinated FANCD2, DNA crosslink repair

Key: GST-PD, GST pull-down assay. FW, Far western. IF, immunofluorescence microscopy. IP, Immunoprecipitation. IVAIn vitro assay. ND, Not determined. TLS, Translesion DNA synthesis. Y2H, Yeast two-hybrid.