Fanconi anemia (FA) is a rare recessive disorder characterized by genome instability, congenital malformations, progressive bone marrow failure, and predisposition to hematologic malignancies and solid tumors1. At the cellular level, hypersensitivity to DNA interstrand crosslinks (ICLs) is the defining feature in FA2. Mutations in thirteen distinct FA genes3 have been shown to interfere with the DNA-replication dependent repair of lesions involving crosslinked DNA at stalled replication forks4. Depletion of SLX4, which interacts with multiple nucleases and has been recently identified as a Holliday junction resolvase5–7, results in increased sensitivity of the cells to DNA crosslinking agents. Here we report the identification of biallelic SLX4 mutations in two patients with typical clinical features of FA and show that the cellular defects in the patients’ cells are complemented by wild-type SLX4, demonstrating that biallelic mutations in SLX4/FANCP cause a new subtype of Fanconi anemia, FA-P.
SLX4 is a multidomain scaffold protein interacting with three distinct nucleases SLX1, ERCC4/XPF-ERCC1, and MUS81-EME15–7. While the SLX4-SLX1 interaction is largely responsible for the Holliday junction resolvase activity seen in the complex, SLX4 can also stimulate the activity of ERCC4/XPF and MUS81 nucleases, both of which have been previously implicated in the processing of interstrand crosslinks (ICLs)8. The finding that depletion of SLX4 leads to increased sensitivity to cross-linking agents and camptothecin5–7 prompted us to investigate SLX4 as a candidate gene for Fanconi anemia1.
So far, mutations in thirteen genes are responsible for FA. Eight of the FA proteins (FANCA/B/C/E/F/G/L/M) form a core complex, a nuclear E3 ubiquitin ligase2 which ubiquitinates FANCI and FANCD29,10. These two activated proteins subsequently localize as an FANCI/FANCD2 (ID) complex to chromatin and direct repair4 in part through interaction with the newly identified nuclease FAN111–14. Cells with mutations in the FA core complex (except for FANCM) lack monoubiquitination of FANCD2. The other FA proteins are FANCJ/BRIP1, a helicase, and homologous recombination (HR) effectors, FANCN/PALB2 and FANCD1/BRCA2. Recently, RAD51C, also involved in HR repair, has been found to be mutated in three patients with an FA-like disorder15. Cells mutated in FANCJ/N/D1 and RAD51C have normal FANCD2 monoubiquitination and these genes are thought to work downstream of the ID complex.
As depletion of SLX4 in a U2OS cell line does not affect FANCD2 ubiquitination (Figure 1A and B), we sequenced SLX4 in the families from the International Fanconi Anemia Registry16 with unassigned FA complementation groups and normal FANCD2 modification (Figure 1C) and identified two families carrying germline mutations, IFAR1084 and IFAR414 (Figure 1D). Phenotypes of the two patients are summarized in Table 1. The lymphoblastoid cell line (LCL) (RA3042) and fibroblasts (RA3083) from the patient 1084/1 showed increased genomic instability (Figure 1E and Table 2) and increased sensitivity to Mitomycin C (MMC) (Figure S1A). The 414/1 patient’s LCL (RA 1376) was not sensitive to MMC suggestive of reversion (Figure S1B); however, his skin fibroblasts (RA 3331) displayed a high degree of DEB induced chromosomal instability (Figure 1E and Table 2) and sensitivity to MMC. No UV sensitivity was observed in fibroblasts from either of the patients (Figure S1C and D). Fibroblasts from the patient 414/1 (RA3331), but interestingly not patient 1084/1 (RA3083), were sensitive to camptothecin (CPT), a topoisomerase I inhibitor (Figure S1E and F).
Sequencing of the cDNA from the 1084/1 patient’s cells revealed skipping of Exon 5 (Figure S2A), due to a homozygous point mutation in the canonical splice donor dinucleotide GT in intron 5 (c.1163+2T>A) in the genomic DNA (Figure S2B). Both parents were found to be heterozygous and an unaffected sibling was found to be negative for this mutation (Figure S2B). The predicted effect of this mutation is a 70 amino acid deletion of amino acids (aa) 317 to 387 of SLX4 (p.R317_F387del), leading to an in-frame deletion of the conserved Cys and Leu of the first UBZ domain and the whole second UBZ domain (Figure 2A, Figure S2C). Immunoprecipitation of SLX4 from the cell line RA3083 confirmed the presence of a slightly shorter protein product (Figure 2B, lane 5, Figure S2D)
In patient 414/1, a heterozygous frameshift mutation in exon 2 (c.514delC) was detected by sequencing of the full length RT-PCR product (Figure S3A) and confirmed in the genomic DNA of the patient and his father (Figure S3B). The predicted protein effect of this frame-shift mutation is a truncated protein with N-terminal 171aa of SLX4 followed by 22 non-SLX4 aa (p.L172FfsX22) (Figure 2A). The second allele of SLX4 in the second patient, identified as described in Online Methods, showed a large genomic deletion from intron 9 to exon 12 resulting in c.2013+225_3147delinsCC (Figure S3C, S3D, and S3E). The predicted effect of this mutation is a truncated protein with N-terminal 671aa of SLX4 followed by 119 non-SLX4 aa due to a frameshift (p.L672VfsX119) (Figure 2A). Consequently, immunoprecipitation with the antibody against SLX4 failed to identify the full-length protein in the patient’s fibroblasts RA3331 (Figure 2B, lane 6).
To prove that the mutations identified in SLX4 were causal for the FA phenotype of both patients, we introduced the WT or the mutant SLX4 cDNAs into the patients’ fibroblasts (RA3083 and RA3331) and performed functional complementation assays (Figure 3 and Figure S4). Expression of WT SLX4 in both cell lines almost fully rescued the MMC sensitivity (Figure 3A, Figure S4A, and S4B), the late S/G2 arrest with MMC treatment (Figure 3B, Figure S4C, S4D, and S4E), and the chromosomal instability after treatment with DEB (Figure S4F). Some residual MMC sensitivity, cell cycle arrest, and chromosomal breakage is most likely due to some cells losing expression of the SLX4 as evident by immunofluorescence analysis (data not shown). Introduction of the mutant proteins did not rescue the FA phenotypes of the patients’ cells although a slight improvement was noted in the various assays possibly due to overexpression of the mutant proteins, which might have residual function. These experiments demonstrate that biallelic SLX4 mutations cause a new subtype of Fanconi anemia, FA-P and FANCP becomes an alias for the SLX4 gene.
SLX4 interacts with multiple factors, two of which, ERCC4/XPF and MUS81 have been previously implicated in crosslink repair8. We therefore tested whether the mutant SLX4 proteins from both patients still interacted with the ERCC4/XPF and MUS81 complexes. We found that ERCC4/XPF, MUS81, and ERCC1 co-immunoprecipitate with endogenous mutant SLX4 (p.R317_F387del) from RA3083 fibroblasts (Figure 4A, lane 5 and Figure S5A, lane 4), although the levels of the mutant SLX4 protein were consistently lower in multiple experiments leading to diminished immunoprecipitation of the interacting factors. The second patient’s SLX4 p.L672VfsX119 overexpressed in RA3331 fibroblasts showed diminished but present interaction with ERCC4/XPF and ERCC1, but not with MUS81 (Figure 4B, lane 3). This was consistent with the previous findings that MUS81 interacts with aa 684–1834 fragment of the SLX4 protein7 which are deleted in the p.L672VfsX119 mutant protein. Immunoprecipitation with an antibody recognizing the N terminus of SLX4 from RA3331 cells showed greatly diminished interaction with ERCC4/XPF, ERCC1, and MUS81 (Figure S5B, lane 6).
As UBZ domains are known to interact with ubiquitin17, we hypothesized that the absence of the tandem UBZ domains in the mutant SLX4 from patient 1084/1 might disrupt the binding of the SLX4 complex to ubiquitin chains of repair proteins at the sites of DNA damage, as shown for the tandem UBZ domains of RAP8018. We therefore performed in vitro ubiquitin binding assays (Figure 4C) that showed binding of the isolated UBZ domains of SLX4 to the K63 chains of ubiquitin (Figure 4C, lane 8). When the two conserved cysteines from each UBZ domain were mutated to alanines (Figure S2C) the binding was reduced to background levels seen with GST alone (Figure 4C, compare lane 7 and 9), suggesting the possibility that SLX4 may localize to the sites of damage through binding to K63 ubiquitinated substrates. As SLX4 would localize other proteins including ERCC4/XPF, MUS81 and SLX1 to sites of DNA damage, the SLX4 deficient patient cell lines described here are important tools to understand which interactions of SLX4 are essential for the repair of cross-linked DNA and ultimately to define the importance of the SLX4/FANCP function in the FA pathway. Patient phenotypes also provide an important clue. FA patients with mutations in FANCN/PALB2 or FANCD1/BRCA2 genes, which are essential for homologous recombination, have very early onset of childhood solid tumors and AML19,20. FA patients with SLX4/FANCP mutations show a milder phenotype more akin to that seen in FA patients with mutations in the FA core or the ID complex components. This suggests that the Holliday junction resolution, an integral step of homologous recombination, might not be the essential function of SLX4 in the somatic compartment during crosslink repair and that the repair depends on the other nucleases, ERCC4/XPF and MUS81 that interact with SLX4.
SLX4/FANCP represents a second protein (besides FANCM) that is conserved in lower eukaryotes, which do not have any other FA pathway components. Yeast SLX4, like human SLX4 interacts with orthologs of ERCC4/XPF and SLX1 and the work in this model organism will provide insight into the function of the FA pathway in human cells. Since germ-line mutations in three FA genes (FANCD1/BRCA1, FANCN/PALB2, FANCJ/BRIP1) and RAD51C, mutated in an FA-like disorder, are associated with a high risk of developing familial breast and ovarian cancers21–24, SLX4 should also be sequenced in patients from pedigrees where no other predisposing mutations could be identified.