Defective FANCI Binding by a Fanconi Anemia-Related FANCD2 Mutant

FANCD2 is a product of one of the genes associated with Fanconi anemia (FA), a rare recessive disease characterized by bone marrow failure, skeletal malformations, developmental defects, and cancer predisposition. FANCD2 forms a complex with FANCI (ID complex) and is monoubiquitinated, which facilitates the downstream interstrand crosslink (ICL) repair steps, such as ICL unhooking and nucleolytic end resection. In the present study, we focused on the chicken FANCD2 (cFANCD2) mutant harboring the Leu234 to Arg (L234R) substitution. cFANCD2 L234R corresponds to the human FANCD2 L231R mutation identified in an FA patient. We found that cFANCD2 L234R did not complement the defective ICL repair in FANCD2−/− DT40 cells. Purified cFANCD2 L234R did not bind to chicken FANCI, and its monoubiquitination was significantly deficient, probably due to the abnormal ID complex formation. In addition, the histone chaperone activity of cFANCD2 L234R was also defective. These findings may explain some aspects of Fanconi anemia pathogenesis by a FANCD2 missense mutation.

In addition, FANCD2 possesses histone chaperone activity, which modifies the chromatin structure by promoting histone deposition/eviction around the ICL sites [14]. The chicken FANCD2 R305W mutant (cFANCD2 R305W), in which Arg305 is replaced by Trp, is specifically defective in the histone chaperone activity in vitro and in vivo, and complements the ICL repair-defective phenotype of the FANCD2 2/2 DT40 cells with a significantly reduced rate [14]. These results revealed that the histone chaperone activity of FANCD2 may play an important role during ICL repair, probably by modifying the chromatin structure to allow access for the proteins required for the downstream steps of the FA pathway [5,6,14]. Importantly, the human FANCD2 R302W mutation, which corresponds to the chicken FANCD2 R305W mutation, has been identified as an FA causative mutation in a patient [52].
A number of FANCD2 mutations, which are generally considered to reduce protein stability, have been identified in FA patients [52][53][54]. However, the means by which the residual FANCD2 protein functions participate in the ICL repair remain poorly understood. In the present study, we focused on the chicken FANCD2 L234R (cFANCD2 L234R) mutation, which corresponds to the human FANCD2 missense mutation at the Leu231 residue, found in an FA patient [53].

Generation of DT40 cell lines expressing chicken FANCD2 mutants
For the construction of the cFANCD2 fusions, the cDNA encoding FANCD2 was inserted into the GFP or the histone H2B-GFP expression vector, using the Gateway system (Invitrogen). To establish stable cell lines expressing the cFANCD2 fusions, the plasmids were transfected into FANCD2 2/2 DT40 cells [55]. The DT40 clones expressing the cFANCD2-mutant fusions were selected by the previously described method [56,57].

Cisplatin sensitivity assay
The cisplatin sensitivity of the DT40 cell lines was assessed by colony formation, in medium containing the indicated amount of cisplatin (Nihon-Kayaku) and 1.4% methylcellulose, as previously described [56,57].

Detection of FANCD2 and FANCI proteins in cell extracts
The chicken DT40 cells were treated with MMC (500 ng/ml for 6 h) or without MMC, and the chromatin fraction was obtained by the method described previously [51,58]. The GFP-cFANCD2 proteins produced in the DT40 cells were fractionated by 6% SDS-PAGE. The ubiquitinated or non-ubiquitinated cFANCD2 and cFANCI proteins were detected by western blotting with the antichicken FANCD2 and FANCI antibodies, respectively.

Immunoprecipitation
The chicken DT40 cells producing the GFP-cFANCD2 proteins were lysed in lysis buffer, containing 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.5% NP-40, 1 mM phenylmethylsulfonyl fluoride, 5 mM NaF, and protease inhibitor cocktail (Roche). Afterward, the lysates were treated with 50 unit/mL of Benzonase, and were mixed with anti-GFP beads (MBL). After three washes with lysis buffer, the immunoprecipitates were analyzed by western blotting using anti-chicken FANCD2 and cFANCI antibodies.

Purification of chicken FANCI, FANCD2, FANCL and human UBE2T
Wild type cFANCD2 and chicken FANCI (cFANCI) were expressed as His 6 -tag fused proteins in the Escherichia coli BL21(DE3) codon(+)RIL strain (Stratagene), and were purified as previously described [14,48]. The Leu234 to Arg mutation in the cFANCD2 cDNA was generated by site directed mutagenesis, using the following primers: 39-gtagc tgctt actga aacag ggcag acggc-59 and 39-gctct ctggc aacct cattt tgctg gg-59. The purification of the cFANCD2 L234R mutant and other cFANCD2 mutants was performed by the same method as for wild type cFANCD2.
The purification of chicken FANCL (cFANCL) and human UBE2T (hUBE2T) was performed by the previously described methods [14,48]. Each protein was expressed as a GST-tagged protein in the Escherichia coli BL21(DE3) codon(+)RIL strain. The GST-tag was removed from hUBE2T by PreScission protease digestion during the purification steps. The concentrations of the purified proteins were assayed by the Bradford method [59], using bovine serum albumin as the standard protein.

Purification of recombinant core histones
Human core histones were purified by the same method described previously [60,61]. Human histones H2A, H2B, and H3 were each overexpressed in Escherichia coli BL21(DE3) cells, and human histone H4 was overexpressed in Escherichia coli JM109(DE3) cells. All core histones were expressed as His 6 -tagged proteins. The histone H3/H4 and H2A/H2B complexes were each formed by the salt dialysis method, and were purified as described previously [62,63]. The His 6tag was removed during the purification, using thrombin protease.

Gel-filtration assay
The purified cFANCD2, cFANCI, and equimolar mixture of cFANCD2 and cFANCI were subjected to Superdex 200 10/300 GL gel filtration column chromatography (GE Healthcare), in buffer containing 20 mM Tris-HCl (pH 8.0), 10% glycerol, 200 mM NaCl, and 5 mM 2-mercaptoethanol. The elution profiles of the purified proteins were monitored by UV absorption at 280 nm. The fractions at 7.5-14.5 mL of elution volume were separated by 7% SDS-PAGE, and the proteins were detected by silver staining.

Supercoiling assay for nucleosome formation
The supercoiling assay for nucleosome assembly was performed as previously described [14]. Before the nucleosome assembly reaction, the H2A/H2B complex (85 ng) and the H3/H4 complex (85 ng) were preincubated with cFANCD2 in the presence or absence of cFANCI at 37˚C for 15 min, in 9 mL of reaction buffer, containing 18 mM Tris-HCl (pH 8.0), 67 mM NaCl, 2.2 mM MgCl 2 , 2.8 mM dithiothreitol, and 3.3% glycerol. To start the nucleosome assembly reaction, 100 ng of relaxed circular wX174 dsDNA, which was preincubated with 2 units of wheat germ topoisomerase I (Promega) at 37˚C for 150 min, in buffer containing 18 mM Tris-HCl (pH 8.0), 8% glycerol, 84 mM NaCl, 2 mM MgCl 2 , 0.2 mM EDTA, and 2.7 mM dithiothreitol, was added to the reaction. The reaction was continued at 37˚C for 60 min. Afterwards, the reaction mixture was deproteinized with 60 mL of a proteinase K solution, containing 20 mM Tris-HCl (pH 8.0), 20 mM EDTA, 0.5% SDS, and 0.5 mg/mL proteinase K. The DNA was then extracted with phenol-chloroform, and subsequently precipitated with ethanol. Topoisomers were separated by 1% agarose electrophoresis in TAE buffer (40 mM Tris-acetate and 1 mM EDTA), and were visualized by SYBR Gold (Invitrogen) staining.

The cFANCD2 L234R mutant is defective in ICL repair in DT40 cells
The human FANCD2 Leu231 residue, which is mutated to Arg in an FA patient [53], is evolutionarily conserved, and corresponds to the chicken FANCD2 (cFANCD2) Leu234 residue (Fig. 1A). To study the effect of the Leu to Arg substitution on the ICL repair, we performed the complementation assay with the FANCD2 2/2 chicken DT40 cells producing the cFANCD2 L234R mutant, in which the Leu234 residue was replaced by Arg. We found that the cFANCD2 L234R mutant was not monoubiquitinated in the whole cell extracts, unlike wildtype cFANCD2 ( Fig. 1B and S1A Figure). A trace amount of the cFANCD2 L234R monoubiquitination was detected in the chromatin fraction, but it was quite low, as compared to wild type cFANCD2 ( Fig. 1C and S1C Figure). The cFANCI monoubiquitination, which strictly requires the FANCD2 monoubiquitination [58], was also clearly defective in the cFANCD2 2/2 DT40 cells producing cFANCD2 L234R (Fig. 1B, C, S1B and S1D Figure).
Previous studies established that monoubiquitination is a prerequisite for the chromatin localization of FANCD2, and hence the DNA repair activities. Exogenously produced cFANCD2, when expressed as a fusion protein with histone H2B in DT40 cells, is targeted to chromatin, and the cFANCD2-H2B fusion protein complements the ICL-deficient phenotype of FANCD2 2/2 DT40 cells [51]. To clarify the function of the FANCD2 L234R mutant further, we expressed the cFANCD2 proteins as fusions with histone H2B in the FANCD2 2/2 DT40 cells. Consistent with the previous report [51], cFANCD2-H2B and cFANCD2 K563R-H2B rescued the cisplatin-sensitive phenotype of the FANCD2 2/2 DT40 cells (Fig. 1D). The cFANCD2 K563R mutant, in which the Lys563 residue that is monoubiquitinated in cFANCD2 is replaced by Arg, is reportedly defective in the monoubiquitination, but is targeted to chromatin by an H2B fusion [51]. In contrast, we found that cFANCD2 L234R-H2B did not complement the cisplatin-sensitive phenotype of the cFANCD2 2/2 DT40 cells (Fig. 1D). These results suggested that the cFANCD2 L234R mutant is defective in the ICL repair even if targeted to chromatin.
Consistent with previous studies [48][49][50], the monoubiquitination of wild type cFANCD2 was robustly stimulated in the presence of cFANCI (Fig. 2C, lanes 2  , indicating that the ID complex formation is required for the efficient monoubiquitination of cFANCD2. The monoubiquitination of the cFANCD2 K563R mutant was not detected in the presence of cFANCI (Fig. 2C,  lane 4, and D). This confirmed that the cFANCD2 monoubiquitination specifically occurs on the Lys563 residue. On the other hand, in the cFANCD2 L234R mutant, the Lys563-specific monoubiquitination was also detected, but the level was quite low, even in the presence of cFANCI (Fig. 2C, lanes 5-8, and D). Therefore, consistent with the in vivo results (Fig. 1B), cFANCD2 L234R may be defective in site-specific monoubiquitination.

cFANCD2 L234R is defective in the complex formation with FANCI
Previous biochemical studies revealed that the ID complex formation is essential for the site-specific monoubiquitination of FANCD2 [41,[48][49][50]. We then tested whether the cFANCD2 L234R mutation affected the ID complex formation. During ICL repair, FANCD2 directly binds to FANCI, and forms a stoichiometric complex, called the ID complex [10][11][12][13][14]. A gel filtration analysis revealed that the purified cFANCD2 and cFANCI proteins formed the ID complex, which eluted as a single peak corresponding to an apparent molecular weight of 375 kDa ( Fig. 3A  and S2A Figure). This elution profile of the ID complex was different from those of the cFANCI and cFANCD2 proteins alone ( Fig. 3A and S2A Figure). Interestingly, we found that an equimolar mixture of cFANCD2 L234R and cFANCI resulted a peak corresponding to 196 kDa, which covered the regions of the two peaks of cFANCD2 L234R and cFANCI alone ( Fig. 3B and S2B). In fact, cFANCD2 L234R and cFANCI separately eluted from the gel filtration column, and only a few fractions contained both proteins (Fig. 3B, bottom panel, and S2B Figure). Interestingly, a cell-based pull-down assay revealed that cFANCD2 L234R bound to cFANCI with substantially reduced efficiency, as compared to wild type cFANCD2 (Fig. 3C, S2C and S2DFigure ), suggesting its defective ID complex formation in vivo. In this assay, the amount of the input GFP-cFANCD2 L234R was lower than that of wild type GFP-cFANCD2, although the amount of cFANCI was not significantly affected ( Fig. 3C and D). This may be due to the fact that the defective ID complex formation destabilized cFANCD2 L234R in cells. However, similar amounts of GFP-cFANCD2 L234R and GFP-cFANCD2 were obtained in the immunoprecipitated (IP) fractions under our experimental conditions ( Fig. 3C and D). We therefore compared the cFANCI binding by pull-down assays with similar amounts of GFP-cFANCD2 L234R and wild type GFP-cFANCD2 in the IP fraction. In addition, cFANCD2 L234R did not support the cFANCI monoubiquitination, which occurs in the ID complex (Fig. 1C, 3C and S2D Figure). Consistently, the cFANCD2 monoubiquitination, which also occurs in the ID complex, was markedly defective in vivo ( Fig. 1B and C) and in vitro ( Fig. 2C and D). These data strongly suggested that, unlike wild type cFANCD2, cFANCD2 L234R does not properly form the ID complex with cFANCI.

Reduced histone chaperone activity of cFANCD2 L234R
We previously reported that FANCD2 possesses histone chaperone activity, which mediates the incorporation/eviction of histones in chromatin, and this activity is actually important for ICL repair [14]. To evaluate the histone chaperone activity of cFANCD2 L234R, we employed the supercoiling assay. In this assay, nucleosome assembly with histones H2A, H2B, H3, and H4 was mediated by cFANCD2 on relaxed plasmid DNA. Nucleosome formation can be detected as topological changes of the plasmid DNA in the presence of topoisomerase I (Fig. 4A). Consistent with our previous results [14], wild type cFANCD2 exhibited the histone chaperone activity (Fig. 4B, lanes 4-7). In contrast, the histone chaperone activity of cFANCD2 L234R was clearly reduced, as compared with that of wild type cFANCD2 (Fig. 4B, lanes 9-12).
Since the nucleosome assembly activity of FANCD2 is reportedly stimulated by FANCI [14], we next performed the supercoiling assay in the presence of FANCI. As shown in Fig. 4C (lane 5), the histone chaperone activity of cFANCD2 was not detected under the low cFANCD2 conditions (0.15 mM). Consistent with the Experiments with wild type (WT) cFANCD2 and WT cFANCI (A), cFANCD2 L234R and WT cFANCI (B). (C) Immunoprecipitation analysis of the cFANCD2 and cFANCI interaction. The cell extracts from chicken DT40 cells expressing GFP-WT cFANCD2 or cFANCD2 L234R were immunoprecipitated with an anti-GFP antibody. The immunoprecipitates were separated by 6% SDS-PAGE, and cFANCD2 and cFANCI were detected by western blotting using antichicken FANCD2 and FANCI antibodies, respectively. (D) Graphic representation of the relative intensity of the bands corresponding to GFP-cFANCD2 (upper) and endogenous cFANCI (lower) shown in panel (C). The band intensity was normalized relative to that of cFANCD2 or cFANCI. previous results, the histone chaperone activity of cFANCD2 was significantly stimulated in the presence of a stoichiometric amount of cFANCI (0.15 mM) (Fig. 4C, lane 6). However, we found that the nucleosome assembly activity of cFANCD2 L234R was not stimulated by cFANCI (Fig. 4C, lane 11). This may be due to the defective formation of the ID complex between cFANCD2 L234R and cFANCI (Fig. 3).
It remains largely unknown whether these mutations affect the FANCD2 function per se. In the present study, we focused on the Leu231 to Arg mutation. The Leu231 residue in human FANCD2 is conserved as the Leu234 residue in the chicken FANCD2 protein (Fig. 1A). We found that cFANCD2 L234R did not  The residues corresponding to human FANCD2 R302, V427, L456, L457, Q815, and W1268, which are mutated in the FA patients, are colored red. (B) The FANCD2 and FANCI proteins are colored grey and green, respectively. The mouse FANCD2 solenoid 1 and the FANCD2 Leu229 residue, corresponding to the cFANCD2 L234 and human FANCD2 L231 residues, are colored black and red, respectively. The FA-associated residues are colored magenta. Ile210 and Phe243 of mouse FANCD2, which are located adjacent to mouse FANCD2 Leu229, are also colored light blue. The van der Waals surfaces of the FANCD2 Ile210, Leu229 and Phe243 side chain atoms are represented. Defective FANCI Binding by a Fanconi Anemia-Related FANCD2 Mutant complement the ICL repair-deficient phenotype of the FANCD2 2/2 DT40 cells (Fig. 1D), and was defective in the monoubiquitination of cFANCD2 and cFANCI ( Fig. 1B and C). These in vivo results suggest that the FANCD2 2/2 DT40 cells with the cFANCD2 L234R mutant reflect the characteristics of the FA patient cells. We then purified the cFANCD2 L234R protein, and performed biochemical analyses. Interestingly, we found that cFANCD2 L234R is clearly defective in the ID complex formation with cFANCI ( Fig. 3A and B). Intriguingly, a cell-based pull-down assay also revealed that cFANCD2 is defective in the cFANCI binding in vivo (Fig. 3C). Although the original report did not address the stability of the FANCD2 protein bearing this mutation, the level of the protein might be reduced (as found with the other FANCD2 missense mutations) because of the lack of the FANCI interaction [11,53].
In the ID complex, both FANCD2 and FANCI are site-specifically monoubiquitinated by the FA core complex containing other essential FA proteins [10,11,30]. The FANCD2 monoubiquitination is especially important, because a FANCD2 mutant, in which its target Lys residue is mutated, is defective in chromatin localization, and hence also in ICL repair [30,51]. On the other hand, the mutation of FANCI on its target Lys residue may not substantially affect the FA repair pathway [58]. However, proper FANCD2 monoubiquitination strictly requires FANCI [10,11,41,[48][49][50]58]. These facts suggest that the correct FANCD2 monoubiquitination that occurs in the ID complex plays an essential role in ICL repair by the FA pathway. Consistently, we found that cFANCD2 L234R, which exhibits impaired cFANCI binding, is also markedly defective in its monoubiquitination both in vivo and in vitro (Figs. 1B and 2). In addition, the FANCD2 monoubiquitination may function to recruit the other FA proteins, such as the SLX4-XPF (FANCP-FANCQ) nuclease complex, to the damaged chromatin site [44][45][46][47]. The reduced FANCD2 monoubiquitination due to the defective ID complex formation may be responsible for producing a major pathogenic effect in FA patients with the FANCD2 Leu231 mutation.
Our gel filtration and pull-down analyses revealed that cFANCD2 L234R was defective in the ID complex formation (Fig. 3). These results suggest that the Leu234 residue of cFANCD2 may be directly involved in binding to cFANCI. However, in the crystal structure of the ID complex containing mouse FANCD2 and FANCI, the mouse FANCD2 Leu229 residue, corresponding to cFANCD2 Leu234, is not located on the binding surface with FANCI (Fig. 5B) [13]. The Leu229 residue is buried within the solenoid 1 domain of FANCD2, and the Leu (hydrophobic) to Arg (hydrophilic) replacement may drastically change the hydrophobic/hydrophilic environment around the residue within this domain (Fig. 5B, right panel). Therefore, the cFANCD2 L234R mutation may indirectly cause the defective formation of the ID complex, probably through the structural change of solenoid 1.
We previously reported that the human and chicken FANCD2 proteins possess histone chaperone activity, and the chicken FANCD2 R305W mutant (cFANCD2 R305W), which corresponds to the R302W mutant of human FANCD2, exhibited substantially decreased histone chaperone activity in vitro and histone mobility in vivo [14]. The human FANCD2 R302W mutation, in which the Arg302 residue is replaced by Trp, has been identified as an FA-associated mutation [52]. The cFANCD2 R305W mutant is proficient in both the ID complex formation with cFANCI and monoubiquitination in the cFANCI-dependent manner [14], suggesting that the defective histone chaperone activity of the mutant may be responsible for the FA pathogenesis. The cFANCD2 L234R mutant also exhibited defective histone chaperone activity in vitro. However, unlike the cFANCD2 R305W mutant, it is markedly defective in both the ID complex formation and monoubiquitination (Figs. 1, 2, and 3). The in vivo complementation assay using the FANCD2 2/2 DT40 cells also revealed the difference between the cFANCD2 L234R and R305W mutants. The cFANCD2 L234R-expressing cells are quite sensitive to a DNA crosslinking agent, cisplatin (Fig. 1C). On the other hand, the cFANCD2 R305W mutant partially complements the DNA-damage sensitive phenotype of the FANCD2 2/2 DT40 cells [14]. Therefore, the cFANCD2 L234R mutation affects the FANCD2 function more severely than the R305W mutation, both in vitro and in vivo.
Supporting Information S1 Figure. The full-length gel images for Fig. 1B and 1C. (A, B) Cells expressing the GFP-fused wild type (WT) cFANCD2 or cFANCD2 L234R were treated with or without mitomycin C (MMC), and the WCEs were analyzed by western blotting using anti-FANCD2 (left panel) and anti-FANCI antibodies (right panel). S-forms and L-forms indicate non-ubiquitinated forms and monoubiquitinated forms of cFANCD2 and cFANCI, respectively. Non-specific bands are marked by the asterisks. (C, D) Cells expressing GFP-fused WT cFANCD2 or cFANCD2 L234R were treated with or without MMC, and the chromatin fractions were analyzed by western blotting using anti-FANCD2 (left panel) and anti-FANCI antibodies (right panel). ''S'' indicates non-ubiquitinated forms of cFANCD2 and cFANCI, and ''L'' indicates monoubiquitinated forms of cFANCD2 and cFANCI. Asterisks indicate the degradation products of GFP-cFANCD2 generated during the chromatin preparation. doi:10.1371/journal.pone.0114752.s001 (TIF) S2 Figure. The full-length gel images for Fig. 3. (A, B) Gel filtration analysis of the ID complex formation. Fraction numbers are shown in the bottom of gelfiltration profiles, and fractions indicated above the gel images were analyzed by 7% SDS-PAGE, and proteins were visualized by silver staining. Markers are indicated as 'M'. Experiments with wild type (WT) cFANCD2 and WT cFANCI (A), cFANCD2 L234R and WT cFANCI (B). (C, D) Immunoprecipitation analysis of the cFANCD2 and cFANCI interaction. The cell extracts from chicken DT40 cells expressing GFP-WT cFANCD2 or cFANCD2 L234R were immunoprecipitated with an anti-GFP antibody. The immunoprecipitates were analyzed by western blotting using anti-chicken FANCD2 (C) and FANCI antibodies (D). doi:10.1371/journal.pone.0114752.s002 (TIF)