FANCP/SLX4

Introduction

Fanconi anemia is an autosomal recessive and X-linked genetic disorder characterized by congenital abnormalities, progressive bone marrow failure and pronounced cancer susceptibility. This rare disease, with an estimated incidence of 1 in 200,000 to 400,000 live births, was first described in 1927 by the Swiss pediatrician Guido Fanconi. While working at the Kinderspital in Zurich, Fanconi encountered a family of five siblings, three of whom perished from an early-onset condition resembling pernicious anemia. The affected siblings displayed additional clinical features including microcephaly, vitiligo, as well as hypoplasia of the testes.¹ Through the pioneering work of Fanconi and others, this heterogeneous constellation of congenital anomalies and pediatric hematopoietic dysfunction became formally recognized as Fanconi anemia (FA) in 1931. Almost sixty years later the first FA gene was identified: in 1992 Manuel Buchwald and colleagues at the Hospital for Sick Children in Toronto cloned the FANCC gene (biallelic mutations in which underlie FA complementation group C or FA-C for short).² This was followed by discoveries of the FANCA, FANCE, FANCF and FANCG genes.^3–6 The pace of FA gene discovery has accelerated in the 21^st century with an additional eight genes, FANCD2 (2001), FANCD1/BRCA2 (2002), FANCL (2003), FANCB (2004), FANCJ/BRIP1 (2005), FANCM (2005), FANCI (2007) and FANCN/PALB2 (2007), being identified in the first decade alone.⁷ Several exciting recent reports describing the identification of a bona fide new FA gene, FANCP/SLX4, and two very strong candidate FA genes, FAN1 and RAD51C, suggest that the pace of FA gene discovery is set to continue.

The FA-BRCA DNA Damage Response Network

Prior to these recent discoveries, the FA-BRCA DNA damage response network was known to be comprised of thirteen bona fide disease-linked proteins, as well as numerous non-FA proteins, that function cooperatively in ICL repair.^7,8 The hallmark of FA patient cells, as well as BRCA1/2-deficient cells, is hypersensitivity to the clastogenic effects of DNA crosslinking agents, examples of which include mitomycin C (MMC) and cisplatin.⁹ Indeed, the diagnostic test for FA is based directly upon this fact, and involves exposing FA patient cells to DNA crosslinking agents and scoring the percentage of metaphases exhibiting radial chromosome formations.¹⁰ The following model for the FA-BRCA pathway has evolved over the past decade: following exposure to DNA damaging agents and during unperturbed S-phase of the cell cycle, the FANC proteins A, B, C, E, F, G, L and M, the Fanconi anemia associated proteins (FAAP) 100 and 24, and the E2 ubiquitin-conjugating enzyme UBE2T, assemble into an active E2/E3 ubiquitin holoenzyme complex in chromatin.^11–18 This ‘upstream core’ FA complex catalyzes the monoubiquitination of the paralogous FANCD2 and FANCI proteins on K561 and K523, respectively.^9,19,20 Importantly, patient-derived mutations in the FA core complex genes FANCA, B, C, E, F, G and L, lead to a complete failure to monoubiquitinate both FANCD2 and FANCI.^9,19,20 Monoubiquitination of the FANCD2 and FANCI proteins results in their targeting to discrete chromatin-associated foci, where they co-localize with numerous well known DNA repair proteins, including BRCA1, RAD51 and phosphorylated H2AX (γH2AX).^9,21,22 The monoubiquitination of FANCD2/I is reversible: the USP1 ubiquitin hydrolase, in association with the UAF1 protein, deubiquitinates both FANCD2 and FANCI, presumably deactivating these proteins once their function in DNA repair is complete.^23,24

Several FA proteins function ‘downstream’ of FANCD2/I monoubiquitination: namely FANCD1/BRCA2, FANCJ/BRIP1 and FANCN/PALB2. The FANCN/PALB2 and FANCD1/BRCA2 proteins stably associate in chromatin and cooperatively regulate the activation of the RAD51 protein.^25–27 RAD51 is the major eukaryotic homologous recombination (HR) protein. During the process of HR, RAD51 forms a nucleoprotein filament on single-stranded DNA (ssDNA) and catalyzes invasion of a homologous intact DNA template [forming a D-loop (displacement loop) structure in the process] to prime repair DNA synthesis.²⁸ FANCD1/BRCA2 and FANCN/PALB2 promote RAD51 nucleoprotein filament formation and strand invasion.^27,29 Accordingly, cells from FA-D1 and FA-N patients exhibit severely reduced MMC-inducible RAD51 nuclear foci formation.^25,26,30 In certain sub-pathways of HR, D-loop formation is followed by the formation of X-shaped heteroduplex DNA structures known as Holliday junctions (HJs), named after the geneticist Robin Holliday who in 1964 proposed a mechanism for DNA strand exchange during fungal meiotic recombination.³¹ Of particular importance for this article, the resolution of these recombination intermediates necessitates the coordinated activity of HJ resolvases and a host of structure-specific endonucleases.³²

The FANCJ/BACH1/BRIP1 (for BRCA1-Associated C-terminal Helicase or BRCA1-Interacting Protein 1) protein was originally identified as a BRCA1 BRCT (for BReast Cancer C-terminal repeat) domain-interacting protein.³³ FANCJ/BRIP1 is a 5′–3′ DNA helicase that can unwind D-loop structures, yet is incapable of unwinding HJs in vitro.^33–35 Importantly, the FANCD1/BRCA2, FANCN/PALB2 and FANCJ/BRIP1 proteins are all required for the HR-mediated repair of an I-SceI-induced DNA double-strand break (DSB).^27,35,36 Collectively, these findings strongly suggest that one major function of the FA-BRCA DNA damage response network is to promote error-free, conservative HR DNA repair.

The Discovery of FANCP/SLX4

Genome-wide screens in the budding yeast Saccharomyces cerevisiae had previously uncovered an important role for the Slx4p protein in the cellular response to DNA ICLs.^37,38 Similarly, the Drosophila melanogaster Slx4p homolog, MUS312, was also shown to be required for ICL repair.³⁹ Consistent with these findings, in 2009 three independent groups demonstrated that depletion of the human homolog of Slx4p, SLX4 (also known as BTBD12) sensitizes cells to the cytotoxic effects of ICLs.^40–42 In light of these findings, the groups of Johan de Winter and Agata Smogorzewska at Vrije Universiteit Medical Center and the Rockefeller University, respectively, sequenced the SLX4 gene in several FA patients not previously assigned to any of the thirteen known FA complementation groups.^43,44 Biallelic SLX4 mutations were uncovered in a total of six individuals from four unrelated kindreds of distinct geographical origin. The clinical phenotypes of these six individuals, while heterogeneous, were typical of that of classic FA and included congenital abnormalities, pediatric hematopoietic dysfunction, and in the case of one individual, susceptibility to squamous cell carcinoma of the tongue.^43–45 As is characteristic of FA patient cells, either fibroblasts or lymphoblasts, or both, from all six affected individuals displayed hypersensitivity to the cytotoxic and clastogenic effects of ICLs. Interestingly, two of the six patient cell lines also displayed marked sensitivity to the topoisomerase type I poison camptothecin, a unique phenotype of cells from FA complementation groups D1, M and N, as well as provisional FA complementation group O (RAD51C^-/-, see below).^46,47 The cellular ICL hypersensitivity of these patient cells was rescued, albeit partially, by expression of either the wild-type human SLX4 protein,⁴³ or a truncated murine Slx4 protein.⁴⁴ Thus, biallelic mutations in the SLX4 gene underlie FA complementation group P, and FANCP becomes synonymous with SLX4.

In the same issue of Nature Genetics as that reporting the findings of the de Winter and Smogorzewska groups, Crossan and colleagues at the Medical Research Council Laboratory of Molecular Biology in Cambridge describe the phenotype of Btbd12/Slx4 knockout mice.⁴⁸ Slx4-deficient mice are born at sub-Mendelian ratios, display growth retardation and developmental defects, have greatly reduced fertility, and are prone to hematological dysfunction. For example, bone marrow progenitors from Slx4-deficient mice are markedly compromised in their ability to differentiate into both lymphoid and myeloid lineages, compared with those from their wild-type littermates. Thus, Slx4-deficient mice recapitulate many of the hallmark clinical features of FA. At the cellular level, Slx4^-/- mouse embryonic fibroblasts (MEFs) exhibit premature replicative senescence as a consequence of increased spontaneous chromosome instability, which is exacerbated upon treatment with MMC. Furthermore, similar to FA patient cells, Slx4^-/- MEFs are hypersensitive to the cytotoxic effects of MMC.⁴⁸ Importantly, all three groups demonstrated that FANCD2 (and almost certainly FANCI) monoubiquitination and nuclear foci formation are unaffected by biallelic SLX4 mutation, establishing that the FANCP/SLX4 protein functions downstream of this central posttranslational modification in the FA-BRCA DNA damage response network.

The Structure and Function of FANCP/SLX4

FANCP/SLX4 is an 1,834 amino acid ∼200 kDa multidomain protein (Fig. 1). FANCP/SLX4 is comprised of two N-terminal C2HC zinc finger domains, related to the UBZ4 family of ubiquitin-binding domains (UBDs);⁴⁹ a MEI9^XPF-Interaction-Like Region referred to as the MLR; a Broad-Complex, Tramtrack and Bric a brac/POxvirus and Zinc finger (BTB/POZ) protein-protein interaction domain; a SAF-A/B, Acinus and PIAS (SAP) domain; and a highly conserved C-terminal domain (CCD) containing a helix-turn-helix motif. Proteomic analysis of SLX4 immune complexes from HEK293 cells revealed that, similar to fungal Slx4, human SLX4 interacts with the structure-specific endonucleases XPF-ERCC1, MUS81-EME1 and SLX1, and stimulates their enzymatic activities.^41,42 XPF-ERCC1 and MUS81-EME1 are members of the XPF/MUS81 family of structure specific endonucleases. In vitro, XPF-ERCC1 preferentially cleaves splayed-arm, bubble and stem-loop structures, while MUS81-EME1 preferentially cleaves 3′ flap and replication fork structures as well as nicked HJs.³² SLX1 is a structurally distinct endonuclease that contains an N-terminal UvrC-intron-endonuclease (URI) domain and a C-terminal PHD-type zinc finger, often found in proteins with chromatin-localized functions.^50,51 SLX1, in association with SLX4, is capable of efficiently processing HJ structures.^40–42 XPF-ERCC1, MUS81-EME1 and SLX1 interact directly with SLX4 via the MLR, SAP and CCD domains, respectively (Fig. 1).^40,42 In addition, SLX4 was shown to physically interact with the telomere-binding protein TRF2 and its partner TERF2IP/RAP1, the PLK1 protein kinase, as well as the mismatch repair (MMR) heterodimer MSH2-MSH3.⁴² Thus, FANCP/SLX4 appears to be uniquely poised to coordinate multiple DNA repair activities.

Importantly, the hypomorphic SLX4 mutations uncovered in the FA-P patients did not appear to overtly affect the ability of the mutant SLX4 protein to physically interact with XPF-ERCC1, MUS81-EME1 and SLX1. These findings suggest that the cellular defects observed may be due to low levels of SLX4 in general, and are not likely to be a consequence of a defect in the interaction with one specific nuclease.^43,44 However, subcellular fractionation experiments by Stoepker and colleagues revealed a specific defect in the chromatin localization of XPF-ERCC1 in cells from FA-P patient EUFA1354, while MUS81-EME1 and SLX1 chromatin localization was unaffected. These findings were corroborated by immunofluoresence microscopy experiments demonstrating a striking defect in ERCC1 nuclear foci formation in EUFA1354 cells.⁴⁴ Defective basal and DNA damage-inducible Ercc1 chromatin association was also observed in the Slx4^-/- MEFs.⁴⁸ Furthermore, Slx4 with a C-terminal truncation (defective in the Slx1 interaction) complemented the MMC hypersensitivity of Slx4^-/- MEFs, in contrast to Slx4 with an N-terminal truncation (defective in the Xpf-Ercc1 interaction).⁴⁸ Collectively, these findings strongly suggest that at least one cellular function of SLX4 is to promote the chromatin localization of XPF-ERCC1 and that this function, in particular, could define its role in the FA-BRCA DNA damage response network.

FANCP/SLX4 and ICL Repair

Here we review our current understanding of cellular ICL repair and consider how FANCP/SLX4 might function in this process: the covalent linkage of DNA nucleobases represents a direct physical block to both DNA replication and RNA transcription processes. Extensive studies in mammalian cells, D. melanogaster, S. cerevisiae and E. coli have established that proteins with defined roles in HR, MMR, NER and translesion DNA synthesis (TLS) also function cooperatively in ICL repair. ICL repair appears to be instigated primarily upon stalling of the DNA replication fork machinery at the covalently linked nucleobases (Fig. 2a and b).⁵² Several studies indicate that the MUS81-EME1 heterodimer is the first structure-specific endonuclease to catalyze nick incision on the template strand, leading to the generation of a one-ended DSB with a 3′ ssDNA overhang (Fig. 2c).⁵³ Indeed, ICL-induced DSB formation is severely compromised in the absence of MUS81.^54,55 While FANCP/SLX4 interacts directly with MUS81-EME1, FANCP/SLX4 does not seem to be required for ICL-induced DSB formation as robust and persistent levels of γH2AX nuclear foci are observed in FANCP/SLX4-deficient cells.^41,44 The FANCM-FAAP24 heterodimer, another member of the XPF/MUS81 family of structure-specific endonucleases, may play several important roles in these early stages of ICL repair. First, FANCM-FAAP24 can promote replication fork reversal and remodeling, and could perhaps facilitate MUS81-EME1 endonuclease activity (Fig. 2c).⁵⁶ Second, FANCM-FAAP24 is required for the recruitment of the ssDNA binding protein RPA to ICL-induced ssDNA, and for the induction of the ATR-mediated ICL-induced checkpoint response.⁵⁷ Many proteins in the FA-BRCA DNA damage response network, including FANCP/SLX4, are phosphorylated by the ATR or ATM DNA damage response kinases.^20,58,59 Third, FANCM-FAAP24 promotes the recruitment of the FA core complex to chromatin and the monoubiquitination of FANCD2/I.^17,60,61 However, Fancd2 monoubiquitination is observed in a Fancm^-/- mouse model generated by targeted deletion of exon 2.⁶² Furthermore, chicken DT40 ΔΔFANCC cells are considerably more sensitive to ICLs than DT40 ΔFANCM cells.⁶¹ These results suggest that FANCM may participate in only a subset of ICL repair events.

The next step of ICL repair, namely unhooking of the ICL, is perhaps the most contentious. ICL unhooking requires structural recognition of the ICL, unwinding of the DNA double helix adjacent to the ICL, incision 3′ and 5′ of the ICL, and displacement of the ICL from the double helix. The NER machinery is thought to play a major role in ICL unhooking, exemplified by the ICL sensitivity of numerous mammalian NER mutants.⁶³ During NER, the catalytic activity for strand unwinding is provided by TFIIH, which is comprised of the ERCC3/XPB and ERCC2/XPD DNA helicases, as well as several other factors.⁶³ As XPB- and XPD-deficient cells exhibit ICL sensitivity, TFIIH may play a similar catalytic function during ICL repair.^63,64 Other candidate helicases for this ICL unhooking step include FANCJ/BRIP1 and BLM (mutated in Bloom syndrome). Indeed, recent studies demonstrate that FANCJ/BRIP1 and BLM can act synergistically to unwind a DNA duplex containing a strand discontinuity.⁶⁵

During NER, XPF-ERCC1 creates a nick incision 5′ of the DNA lesion (most often a UV-C irradiation-inducible cyclobutane pyrimidine dimer).³² XPF- and ERCC1-deficient cells exhibit pronounced ICL hypersensitivity, and by analogy with its function in NER, XPF-ERCC1 is thought to be responsible for nick incision 5′ of the ICL (Fig. 2c or e). However, the precise enzymatic function(s) of the XPF-ERCC1 heterodimer in ICL repair remains a matter of considerable debate.^32,63,66 The observation that the chromatin localization of XPF-ERCC1 is defective in FANCP/SLX4-deficient cells,^44,48 yet ICL-inducible RAD51 nuclear foci formation appears to be unaffected, suggests that XPF-ERCC1 could play a role downstream of DSB formation and RAD51-mediated DNA strand invasion (Fig. 2g). An alternative interpretation of this data is that, in the absence of XPF-ERCC1 and despite the continued presence of the ICL, RAD51-mediated strand invasion proceeds at a non- or micro-homologous DNA sequence, leading to deletion and/or chromosomal rearrangement. Another candidate unhooking nuclease is the recently identified FAN1 protein (for FANCD2-associated nuclease 1).^67–69 FAN1 harbors a VRR_nuc domain and exhibits 5′ flap endonuclease as well as 5′–3′ exonuclease activities.^67–69 FAN1 is recruited to chromatin via a noncovalent interaction between its UBZ domain and monoubiquitinated K561 of FANCD2.^67–69 Like XPF-ERCC1,⁷⁰ and unlike MUS81-EME1,⁵⁵ depletion of FAN1 does not affect ICL-induced DSB formation,⁶⁸ suggesting that FAN1 may also function downstream of the ICL unhooking step (Fig. 2g).

Covalently linked nucleobases may also be recognized and initially processed by components of the MMR machinery. Numerous studies have established important physical connections between the FA and MMR proteins. For example, FAN1 and FANCJ/BRIP1 are components of the MLH1 interactome,^69,71 while MSH2 and MSH3 were recently identified in FANCP/SLX4 immune complexes.⁴² From a functional perspective, Peng and colleagues demonstrated that the interaction between FANCJ/BRIP1 and MLH1 is required for correction of the ICL hypersensitivity of FA-J cells.⁷² Furthermore, MutLα (MLH1-PMS2) is capable of incising a nicked mismatch-containing DNA heteroduplex,⁷³ and MSH2-deficient cells are sensitive to the cytotoxic effects of ICLs.⁷⁴ It will be important to clearly define the molecular interplay between FAN1, FANCJ/BRIP1, FANCP/SLX4 and the MMR proteins during ICL repair. Nevertheless, once unhooking is completed, the resulting strand discontinuity/gap is filled by homologs of the yeast RAD6 epistasis group in TLS (Fig. 2e and f). Accordingly, cells with defective TLS display marked sensitivity to ICL-inducing agents.^38,75 The MMR machinery could also function post-TLS to remove any mismatches arising as a consequence of low fidelity TLS DNA polymerases. Recent experiments with ICL-bearing plasmids and Xenopus cell-free extracts suggest an important role for FANCD2 monoubiquitination in both the early endonucleolytic and TLS steps of ICL repair.⁷⁶ Similar to that recently observed for FAN1, perhaps monoubiquitinated FANCD2 and FANCI recruit additional UBD-harboring ligands to perform these functions.^67–69 Following gap filling by TLS, homologs of the yeast RAD52 epistasis group initiate HR through the displacement of RPA from the 3′ ssDNA overhang and the promotion of RAD51 nucleoprotein filament formation and strand invasion of the complementary duplex (Fig. 2g). As mentioned earlier, FANCD1/BRCA2 and FANCN/PALB2 promote RAD51 nucleoprotein filament formation and strand invasion.^27,29 Recently, homozygous hypomorphic mutations in the RAD51 paralogous gene RAD51C were uncovered in three siblings with clinical manifestations resembling FA.⁴⁷ Owing to the absence of hematological deficiencies or cancer in the one surviving affected child, a formal diagnosis of FA has yet to be assigned. Nonetheless, RAD51C is provisionally referred to as FANCO.⁴⁷ Importantly, similar to FANCD1/BRCA2 and FANCN/PALB2, FANCO/RAD51C acts downstream of FANCD2/I monoubiquitination and promotes ICL-inducible RAD51 nuclear foci formation, consistent with previous studies.^47,77,78 Following strand invasion, a D-loop forms and the invading 3′ ssDNA primes de novo DNA synthesis (Fig. 2g). Once strand extension is complete, the newly synthesized DNA will disengage and reanneal with its original contiguous DNA strand, restoring the DNA replication fork (Fig. 2h). Recombination intermediates, including D-loops and HJs, that arise during these latter stages of HR will require the coordinated activity of HJ resolvases and structure-specific endonucleases for their resolution. It is highly likely that FANCP/SLX4 plays a major role in coordinating several of these enzymatic activities. Precisely defining these enzymatic activities and identifying their specific recombination intermediate substrates in vivo poses a considerable challenge.

Concluding Remarks

Over the course of the past two decades, we have witnessed remarkable advancements in our understanding of the genetics and molecular etiology of FA. While the first six identified FA proteins (A, C, D2, E, F and G) harbored no recognizable enzymatic domains, many proteins with putative or proven enzymatic function have now entered the fray. The identification of biallelic mutations in the SLX4 gene in FA-P patients potentially adds three new structure-specific endonucleases to the FA DNA damage response network: MUS81-EME1, SLX1-SLX4 and XPF-ERCC1. Future studies will seek to uncover the in vivo substrates, enzymatic activities and regulation of these structure-specific endonucleases, in particular as they relate to ICL repair. These studies will surely continue to improve our understanding of the molecular basis of this devastating disease and offer hope for much needed therapeutic advancements.