Fanconi Anemia Proteins Localize to Chromatin and the Nuclear Matrix in a DNA Damage- and Cell Cycle-regulated Manner*

Fanconi anemia (FA) is a genetic disease characterized by congenital defects, bone marrow failure, and cancer susceptibility. Cells from patients with FA exhibit genomic instability and hypersensitivity to DNA cross linking agents such as mitomycin C. Despite the identification of seven complementation groups and the cloning of six genes, the function of the encoded gene products remains elusive. The FancA (Fanconi anemia complementation group A), FancC, and FancG proteins have been detected within a nuclear complex, but no change in level, binding, or localization has been reported as a result of drug treatment or cell cycle. We show that in immunofluorescence studies, FancA appears as a non-nucleolar nuclear protein that is excluded from condensed, mitotic chromosomes. Biochemical fractionation reveals that the FA proteins are found in nuclear matrix and chromatin and that treatment with mitomycin C results in increase of the FA proteins in nuclear matrix and chromatin fractions. This induction occurs in wild-type cells and mutant FA-D (Fanconi complementation group D) cells but not in mutant FA-A cells. Immunoprecipitation of FancA protein in chromatin demonstrates the coprecipitation of FancA, FancC, and FancG, showing that the FA proteins move together as a complex. Also, fractionation of mitotic cells confirms the lack of FA proteins in chromatin or the nuclear matrix. Furthermore, phosphorylation of FancG was found to be temporally correlated with exit of the FA complex from chromosomes at mitosis. Taken together, these findings suggest a role for FA proteins in chromatin and nuclear matrix.

Cells derived from patients with the disease exhibit characteristic hypersensitivity caused by DNA cross-linking agents and generalized decreased survival (16 -20). In addition, a well described G 2 phase cell cycle delay has also been described that is thought to be secondary to a defective S or G 2 checkpoint (21)(22)(23). However, no defined biochemical mechanism for this hypersensitivity has been elucidated. Patient and cellular phenotypes across all the complementation groups are similar, suggesting an interrelatedness or cooperativity between the FA proteins.
This cooperativity has been borne out by work we have done in showing binding of FancA and FancC in a protein complex in both nucleus and cytoplasm (24 -26). Recent work has found the FancG and FancF proteins in the complex as well (27)(28)(29)(30). A large complex is suggested by our recent work 2 , and binding does not occur in any of the complementation groups except the FA-D group.
One clue to FA function lies in the study of the FancA protein, which contains a classic bipartite nuclear localization signal and is phosphorylated. FancA nuclear localization, phosphorylation, and binding to FancC are abolished in all complementation groups except the FA-D group (24 -26, 29). This suggests that a nuclear event is critical to the normal function of the FA proteins, and the aberrant protein in the FA-D group may have a role downstream of the FA complex in the nucleus.
To date, other than the alteration of binding, nuclear localization, and FancA phosphorylation in mutant FA complementation groups, no other consistent biochemical change has been described, although some have described variations in FancC levels in the cell cycle (31,32). In our study, we have found that the FA proteins not only reside in the nucleus but also are closely associated with the nuclear matrix and chromatin.
The nuclear matrix is a loose mechanical framework of proteins that has also been implicated in enzymatic activities in the regulation of transcription, replication, and DNA repair. The nuclear matrix is intimately associated with chromatin. BRG and brm, the human swi-snf homologues, are examples of nuclear matrix proteins involved in transcriptional regulation that change phosphorylation state, shift to chromatin, and become distinct from the nuclear matrix during mitosis (33)(34)(35). Immunofluorescence of FancA reveals a similar shift away from the condensed chromosomes of mitosis, suggesting that they also interact with chromatin and the nuclear matrix. Our studies demonstrate that the FA proteins associate with chromatin and the nuclear matrix in an inducible fashion.

MATERIALS AND METHODS
Cell Culture-Cells were grown at 37°C in a 5% CO 2 incubator. FA fibroblasts cell lines were grown in Dulbecco's modified Eagle's medium * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶  (14,36,37). Synchronization-HeLa cells were treated overnight with 2 mM thymidine, washed, released into regular media, and treated again overnight with thymidine (38). Cells were then released for no time (G 1 -S border), 3 h (S phase), 6 h (G 2 ), or overnight in the presence of 1 M nocodazole (Sigma). Mitotic cells were collected the next morning by shaking the plate and collecting the cells in suspension. Verification of cell cycle state was achieved using FACScan analysis (Becton Dickinson) (39).
Immunofluorescence-5 l of confluent, trypsinized cells were plated into 250 l of media on 8-chamber slides and allowed to attach overnight at 37°C. The slides were washed in PBS and immediately fixed for 20 min in 4% paraformaldehyde followed by 10 min in 0.3% Triton X-100 in PBS, all at room temperature. The chambers were then exposed to primary antibody in 3% bovine serum albumin in PBS for 1 h at room temperature and washed in 0.1% Nonidet P-40/PBS and PBS. The slides were then exposed to fluorescein isothiocyanate-anti-rabbit secondary antibody and DAPI and then covered with mounting solution with anti-fade (Vectashield) (25).
Subnuclear Fractionation-Procedures for permeabilization and sequential subnuclear extraction were adapted from the methods of Burtelow et al., and Reyes et al. (40,35). In brief, cells pelleted (100 l) from one large plate were resuspended and permeabilized in 5 ml of low salt buffer (10 mM Hepes, pH 7.4, 10 mM KCl, and 50 g/ml digitonin) containing protease and phosphatase inhibitors (2 g/ml leupeptin, 1 g/ml pepstatin, 1 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, and 1 mM Na3VO4) for 15 min, 4°C. Permeabilized nuclei were recovered by centrifugation at 1000 rpm, 5 min at 4°C. The supernatant was termed the soluble fraction. The nuclei were washed two additional times in the same buffer. After wash, the nuclei were resuspended in 200 l of permeabilization buffer containing 30 units of DNase I (RNase free, Roche Molecular Biochemicals) for 15 min, room temperature and an additional 15 min, 37°C. For the purpose of comparing no DNase versus ϩ DNase, the chromatin was incubated for 1 h, 4°C. Chromatin proteins were extracted by adding extraction buffer (1% Triton X-100, 50 mM Hepes, pH 7.4, 150 mM NaCl, and 30 mM Na 4 P 2 O 7 ⅐10H 2 O, 10 mM NaF, and 1 mM EDTA) containing protease and phosphatase inhibitors for 10 min, 4°C. Supernatant collected at 14,000 rpm in a microfuge for 10 min was termed the chromatin fraction. The pellet was extracted in urea buffer (8 M urea, 100 mM NaH 2 PO 4 , and 10 mM Tris, pH 8.0). The supernatant collected after high speed microfuge spin was termed the nuclear matrix. Protein concentration was determined by Bradford assays.
Antibody Production-FancA and FancC antisera were prepared as previously described (24). pGEX-FancG(N) (Alan D'Andrea, Boston, MA) containing the N-terminal half of FancG was transformed into DH5␣, and GST-FancG(N) was prepared and injected into a rabbit (University of Virginia animal protocol 5814). Serum was screened by Western blot and further purified over a column containing GST-FancG(N). Topoisomerase II (Oncogene Science) and lamin B antibodies (Oncogene Science) were obtained commercially. Histone 4 antibody was provided courtesy of Dr. David Allis, University of Virginia.
Immunoprecipitation-A cellular crude lysate or chromatin extract was prepared as above. Two mg of protein in 1 ml of the respective buffers containing protease and phosphatase inhibitors was incubated with 2 g of both anti-N-and anti-C-terminal FancA antibodies for 1 h at 4°C. 50 l of protein A-Sepharose (Amersham Pharmacia Biotech) was added, and the resulting mix was rotated at 4°C for an additional h. Anti-FancA antiserum was added to 2 mg of extract and incubated for 1 h, 4°C. Protein A-Sepharose was added, and the mixture was rotated for 1 h, 4°C. The beads were then washed three times in Tris-buffered saline (TBS; 50 mM Tris-HCl, pH 8.0, and 150 mM NaCl) containing 0.1% Triton X-100 containing protease and phosphatase inhibitors and dried. 50 l of loading buffer was added (24).
SDS-PAGE and Immunoblotting-SDS-PAGE was conducted followed by gel transfer in 25 mM Tris and 200 mM glycine onto nylonsupported nitrocellulose. Filters were blocked for 1 h in 5% bovine serum albumin in TBS and then were incubated in TBS plus Tween 20 (TBS-T) containing primary antibody overnight at room temperature. Filters were then washed in TBS-T, incubated with horseradish peroxidase-linked secondary antibody (Amersham Pharmacia Biotech), washed again, and visualized by chemiluminescence (24).
Phosphatase Reactions-Phosphatase reactions were adapted from previous protocols (40). FancG immunoprecipitation was conducted on 2 mg of whole cell extracts, as described. Immunoprecipitates were treated with phosphatase (Boehringer) for 30 min at 30°C in 100 l of reaction buffer containing 0, 50, and 100 units of enzyme in the presence or absence of phosphatase inhibitors. All reactions were stopped by addition of 40 l of 4X loading buffer, and samples were subjected to SDS-PAGE and immunoblotting.

FancA Is a Granular Nuclear Protein That Is Excluded from the Nucleolus and Condensed, Mitotic Chromosomes-In
search of clues to FA protein function, we studied the microscopic appearance of the FancA protein. Immunofluorescence was conducted on FA-A mutant cells either containing no detectable FancA or corrected with pMMP-FancA. Visualization of FancA with anti-FancA serum on corrected FA-A cells revealed strong granular nuclear staining and no detectable staining in the nucleoli (Fig. 1, interphase cell). FancA was excluded from condensed chromosomes completely in mitotic cells (Fig. 1, mitotic cell) as evidenced by counterstain with DAPI. Mutant cells were completely unstained by anti-FancA. This appearance is similar to published patterns of chromatinassociated and nuclear matrix proteins, such as brm and BRG, which are involved in DNA repair, maintenance of genomic stability, and transcription regulation (33)(34)(35).
The FA Proteins Are Specifically Released after DNase Treatment-Immunofluorescence microscopy suggests that FA proteins are associated with nuclear matrix and chromatin. Chromatin is arranged in higher order structures with DNA packaged around core histones into nucleosomes (41,42). To determine whether FA proteins are tightly associated with DNA, permeabilized HeLa cells were treated with or without DNase for 1 h at 4°C. FA proteins were predominately released only when treated with DNase (Fig. 2). This result indicates that FA proteins are chromatin-bound and are specifically released after DNase treatment, suggesting an intimate association with DNA.
The FA Proteins Localize to the Nuclear Matrix and Chromatin-Chromatin and nuclear matrix-associated proteins are involved in activities such as transcription, replication, DNA repair, and maintenance of genomic stability. Because the immunofluorescent microscopic appearance of FancA is suggestive of chromatin and nuclear matrix proteins, we analyzed the subnuclear localization of FancA, FancC, and FancG by biochemical fractionation (Fig. 3A). Immunoblotting of prepara- The FA Proteins Coimmunoprecipitate in Chromatin Extract-Our subnuclear fractionation experiments confirmed the immunofluorescence data that FA proteins are found in chromatin and the nuclear matrix of the cell. However, previous work has shown association of the FA proteins only under standard immunoprecipitation conditions that by definition omit chromatin-bound proteins. To confirm chromatin localization and to demonstrate association of the FA proteins, anti-FancA immunoprecipitations were performed using chromatin extract prepared from the indicated cell lines (Fig. 3B). Immunoblotting revealed the coprecipitation of FancC and FancG with FancA. This result was noted in the FA-A corrected line, HeLa, and FA-D corrected line as well as the FA-D mutant line in which FancA, FancC, and FancG have been noted to coprecipitate and localize to the nucleus. The FA protein complex appears intact in a variety of locations including cytoplasm and nuclear extract prepared using standard lysis conditions. However, these data show that the complex, easily coprecipitated in standard lysis conditions, can also be detected intact in a low salt, DNase-liberated chromatin extract.
The FA Proteins Shift to Chromatin and Nuclear Matrix upon Induction by MMC-In previous work no inducibility of the FA proteins in location or level has been demonstrated in response to bifunctional alkylating agents even though these agents cause hypersensitivity in mutant FA cells (31,32). However, these studies did not selectively analyze chromatin or the nuclear matrix. Therefore, we analyzed the subnuclear localization of FancA, FancC, and FancG in response to MMC treatment. HeLa cells were treated for 24 h with 0.01, 0.1 (90% cell kill in mutant FA cells, 10% kill in wild-type in 4 days, respectively), and 1 M MMC. The cells were subjected to sequential subnuclear fractionation as above. Increased FA proteins appeared in chromatin (Fig. 4A) and nuclear matrix fractions Analysis again showed no increase in FA protein levels in chromatin (Fig. 4G) and nuclear matrix (Fig. 4H). The induction time course of FA protein localization by MMC seems consistent with the idea that the cell must first encounter the DNA cross-link during S phase as proposed in recent work (43).
Other cell lines were analyzed as well. FA-A mutant and corrected cells were tested similarly by treatment with 0.1 M MMC. Maximal effect was seen after 48 h of treatment. FA-A mutant cells (GM6914) contain no endogenous FancA. Only the corrected cells exhibited increased localization of FA proteins to chromatin and nuclear matrix (Fig. 4I) after treatment. FA-D mutant and corrected cells were also studied by 24 h treatment with 0.1 M MMC. In this case, both the mutant and corrected cells displayed increased FA protein localization to chromatin (Fig. 4J) and nuclear matrix (Fig. 4K). These data are consistent with those previously reported that FA proteins are absent FA Proteins Exit from Chromatin during Mitosis-FA proteins inducibly localize to chromatin and nuclear matrix upon MMC treatment. Because of the immunofluorescence microscopy that demonstrated exit of FA proteins at mitosis, we analyzed the effect of the cell cycle upon subnuclear localization. HeLa cells were synchronized using double thymidine blockade to produce arrest at the G 1 -S border. Upon release, the cells were collected and were subjected to sequential subnuclear fractionation. Synchronization was verified by FACScan analysis (data not shown). FA proteins exhibited little or no change in the chromatin or nuclear matrix fractions at the G 1 -S border, S phase, or G 2 phase (Fig. 5). However, at mitosis, as indicated by the lack of lamin B in the nuclear matrix fraction, no FA proteins were detected in chromatin or the nuclear matrix suggesting exit of the FA proteins from condensed chro- mosomes. Chromatin and the nuclear matrix appeared to remain intact even though the nuclear envelope had dissolved as topoisomerase II was still detectable in both fractions. A higher mobility form of FancG was noticeable at mitosis in the soluble fraction. This nuclear subfractionation experiment is consistent with the immunofluorescent appearance of FancA and its disappearance from condensed chromosomes at mitosis as seen in Fig. 1.
Phosphorylation of FancG Occurs at Mitosis-The FA proteins translocate from condensed chromosomes at mitosis as shown by biochemical fractionation. Interestingly, FancG exists as a higher mobility isoform at mitosis. To determine the nature of the higher mobility isoform of FancG at mitosis, whole cell extract was prepared from mitotic and asynchronous HeLa cells and immunoprecipitated with anti-FancA or anti-FancG antiserum. Subsequent immunoblotting showed that three isoforms of FancG specifically immunoprecipitate with anti-FancG and coprecipitate with FancA in mitotic extract (Fig. 6A). Asynchronous extract contained only one isoform of FancG. To demonstrate that the higher mobility forms of FancG are phosphorylated, anti-FancG-immunoprecipitated beads were incubated with phosphatase. The two higher mobility forms of FancG were removed by phosphatase reaction at 30°C. Incubation at 30°C without phosphatase or incubation with the phosphatase inhibitor sodium pyrophosphate resulted in retention of both forms. Sodium orthovanadate inhibited the removal of only the lower of the two bands. This phenomenon mirrors the regulation and shift in localization of other proteins such as the human swi-snf protein complex. DISCUSSION The FA proteins are of unknown function yet have been shown to localize to the nucleus, a localization which is functionally important to correction of the FA DNA cross-linker hypersensitivity. In mutant cells, FA proteins remain largely in the cytoplasm and do not interact in a complex except in FA-D cells. In this study, we show that the FA proteins FancA, FancC, and FancG associate with the nuclear matrix and chromatin. The proteins specifically shift to chromatin and nuclear matrix in an inducible way by MMC treatment, to which FA cells are hypersensitive. FA proteins are similar to other chromatin and nuclear matrix-bound proteins in their fractionation, appearance by immunofluorescence, and exclusion from condensed chromosomes during mitosis. This exclusion is temporally related to phosphorylation of FancG, which remains bound to the other FA proteins.
Until this study, little evidence has emerged that has shown an inducible response of FA proteins to drug treatment or cell cycle. Previous data have shown variations in FancC (31,32), but change in the total level of complexed FA protein does not seem to change 2 . To perform these experiments, overexpression systems have been used, resulting in increased amounts of cytoplasmic and soluble nuclear proteins. However, we have not observed any increase in the amount of complexed, chromatin-bound, or nuclear matrix FA proteins as a result of overexpression; indeed, HeLa cells contain high, endogenous levels of FA proteins in these subnuclear compartments. Although we are unable to directly immunoprecipitate the complex from the nuclear matrix because of the requirement for urea solubilization, the same pattern of localization, protein level, and shift in response to drug treatment and cell cycle imply association of the FA proteins at least functionally. In addition to the emerging importance of nuclear matrix proteins, recent evidence supports the FA protein data presented here. A report on FancA and FancC binding to a nuclear structural protein, human ␣ spectrin, has been described (44). This complex has been shown to purify with endonuclease activity (45). One attractive hypothesis is that the structural array that spectrin is part of is responsible for the mechanical manipulation in response to DNA damage that allows for the FA complex to translocate to the chromatin where presumably DNA repair can be carried out. In addition, defective non-homologous end joining has been postulated to be abnormal in FA, implying a necessity for DNA binding (46 -49).
Because this shift of proteins seems to occur relatively quickly upon drug treatment and occurs before the well documented G 2 /M accumulation, it is unclear what importance a presumed G 2 /M checkpoint is for FA protein function. Given that the shift is occurring by 12-16 h after drug treatment, this shift could very well be indicative of an S phase checkpoint, the importance of which has been suggested by recent reports (43,50). The stimulus for protein shift could conceivably arise from a cytoplasmic signal as well as from a nuclear signal. Several groups have reported defects in interferon signaling, stat signaling, and association of FA proteins with cytochrome P450 machinery (51-53). The potential for cytoplasmic signaling FIG. 5. FA proteins exit chromatin and the nuclear matrix at mitosis. HeLa cells were synchronized to the G 1 -S border by double thymidine blockade and released. Cells were collected at 0 h (G 1 -S), 3 h (S phase), 6 h (G 2 ), and after overnight exposure to nocodazole (M). Subnuclear fractionation and immunoblotting revealed no gross FA protein level changes except at mitosis. At mitosis, there was complete absence of FA proteins detected in the chromatin or nuclear matrix. Higher mobility forms of FancG were detected at mitosis in the soluble fraction.
FIG. 6. Phosphorylation of FancG occurs at mitosis. A, whole cell extracts (WCE) were prepared from asynchronous (asy) and mitotically (Mit) arrested HeLa cells. Immunoprecipitation (IP) revealed the coprecipitation of three isoforms of FancG with FancA at mitosis. B, immunoprecipitations of FancG prepared from mitotic HeLa cells were subjected to phosphatase under the indicated conditions. The phosphatase removed the two higher mobility forms of FancG when incubated at 30°C. Incubation at 30°C without phosphatase or in the presence of sodium pyrophosphate prevented loss of the two higher mobility forms of FancG. Sodium orthovanadate prevented the removal of only one of the higher mobility isoforms. FancG is phosphorylated at mitosis and correlates with the exit of the FA protein complex from condensed chromosomes. PP, sodium pyrophosphate; Vn, sodium orthovanadate. could resolve the controversial data that support the importance for cytoplasmic versus nuclear events in FA biochemistry.
The shift of FA proteins to chromatin suggests a direct or indirect interaction with DNA. Our next studies will focus on this interaction using naked or DNA templates in nucleosomal form and purified proteins. We will also attempt to define a potential kinase for FancG. A potential kinase that can be tested is the G 2 /M cyclin-dependent kinase cdc2, which has been reported to bind to FancC (32). Attempts at recapitulation of FA protein function in vitro will be crucial in unlocking its biochemical pathway.