|Home | About | Journals | Submit | Contact Us | Français|
Fanconi anemia (FA) is characterized by cellular hypersensivity to DNA crosslinking agents, but how the Fanconi pathway protects cells from DNA crosslinks and whether FA proteins act directly on crosslinks remains unclear. We developed a chromatin-IP-based strategy termed eChIP and detected association of multiple FA proteins with DNA crosslinks in vivo. Inter-dependence analyses revealed that crosslink-specific enrichment of various FA proteins is controlled by distinct mechanisms. BRCA-related FA proteins (BRCA2, FANCJ/BACH1, and FANCN/PALB2), but not FA core and I/D2 complexes, require replication for their crosslink association. FANCD2, but not FANCJ and FANCN, requires the FA core complex for its recruitment. FA core complex requires nucleotide excision repair proteins XPA and XPC for its association. Consistent with the distinct recruitment mechanism, recombination-independent crosslink repair was inversely affected in cells deficient of FANC-core versus BRCA-related FA proteins. Thus, FA proteins participate in distinct DNA damage response mechanisms governed by DNA replication status.
Fanconi anemia (FA) is a recessive genetic disorder, manifesting progressive bone marrow dysfunction, pancytopenia, and highly elevated cancer susceptibility. A total of 13 subtypes of FA have been identified and each has been attributed to mutations in a distinct gene (Wang, 2007). Currently, FA proteins can be organized into three groups according to their roles in FANCD2 monoubiquitination. The FA core complex, consisting of FANCA-C, E-G, L, M, is required for damage-induced monoubiquitination of FANCD2 and FANCI (Ciccia et al., 2007; Smogorzewska et al., 2007; Wang, 2007). FANCD2 and FANCI form the FA I/D2 heterodimeric complex and both are subjected to damage-induced monoubiquitination which is required for their function (Garcia-Higuera et al., 2001; Matsushita et al., 2005; Smogorzewska et al., 2007). FANCD1, FANCJ, and FANCN are involved in homologous recombination (HR), but their functions are not required for monoubiquitination of the FA I/D2 complex. This rare disease has recently attracted broad attention as three FA genes, FANCD1, FANCJ, and FANCN, were found to be identical to breast cancer susceptibility (BRCA) genes BRCA2, BACH1, and PALB2, respectively (Erkko et al., 2007; Howlett et al., 2002; Rahman et al., 2007; Seal et al., 2006).
The cellular hallmark of FA patients is a profound hypersensitivity to agents causing DNA interstrand crosslinks, suggesting an important role of the Fanconi pathway in the DNA damage repair and response. Consistently, FANCD2 and FANCI undergo DNA-damage induced monoubiquitination with the FA core complex acting as the E3 ligase (Alpi et al., 2008; Garcia-Higuera et al., 2001; Meetei et al., 2003; Smogorzewska et al., 2007). Several FA proteins were also found to be downstream targets of the ATM/ATR checkpoint kinases (Andreassen et al., 2004; Smogorzewska et al., 2007). Phosphorylation of FANCI by ATR has been shown to be switch that turns on the FA pathway (Ishiai et al., 2008)
To date, how the FA and BRCA proteins act to protect cells from DNA crosslinking lesions remains unknown. This major gap in understanding results from the absence of knowledge regarding whether and how these proteins directly interact with damaged DNA in the context of chromatin in vivo. Unlike DSBs which can be introduced in a site-specific fashion by homing endonucleases, no analogous method exists that allows the efficient introduction of base-modifying lesions at a specific DNA sequence in vivo. To overcome this deficiency, we developed an episomal replication-based chromatin IP approach we call eChIP. The eChIP method allows the identification of DNA damage response proteins at site-specific DNA lesions in the context of DNA replication. In this report, we applied this approach to analyze the recruitment of various FA proteins to defined DNA interstrand crosslinks. We found that the enrichment of breast cancer-related FA proteins and other FA factors can be distinguished by their differing dependency on DNA replication. This observation suggests that the Fanconi anemia proteins may support diverse mechanisms in dealing with DNA interstrand crosslinks.
The Epstein-bar virus replication origin utilizes mostly host replication initiation factors. Once cleared of the origin, the elongation complex consists solely of host factors (Chaudhuri et al., 2001; Frappier and O'Donnell, 1991). Episomal DNA supported by the EBV origin is known to possess comparable chromatin structure and has been used extensively in transcription regulation studies (Reeves et al., 1985; Wensing et al., 2001). To construct the eChIP substrate carrying a defined DNA interstrand crosslink and capable of episomal replication, a site-specific DNA interstrand crosslink was placed 488 base pairs downstream of the Epstein-Bar virus (EBV) replication start site in a plasmid substrate (Fig. 1A). When ectopically introduced into cultured cells, the DNA substrate lacking the crosslink modification was able to undergo Epstein-Barr nuclear antigen-1 (EBNA)-dependent episomal replication, while no replication was observed in the absence of EBNA (Fig. 1B) consistent with previous studies of EBV replication (Gahn and Schildkraut, 1989). When the crosslinked substrate was transfected into HEK293 cells, significantly increased FANCD2 ubiquitination could be detected 6 hrs after transfection (Fig. 1C), indicating that the interstrand crosslink on the episomal DNA was able to elicit an endogenous response. Interestingly, EBV origin-directed DNA replication was not essential for FANCD2 ubiquitination, since ubiquitination of FANCD2 was observed in both EBNA− and EBNA+ HEK293 cells.
A DNA interstrand crosslink constitutes an absolute block to DNA replication. Thus, the presence of a psoralen crosslink downstream of the unidirectional EBV origin is expected to create a stalled replication fork positioned at the origin side of the crosslink as depicted in Fig. 1A. To test if the crosslinked pORIP substrate can recapitulate a stalled replication fork in vivo, we performed the eChIP assay to determine whether the MCM2-7 complex, an integral component of an elongating replication fork, is enriched at the site of the crosslink. In 293-EBNA cells, the presence of the psoralen crosslink resulted in a significant enrichment of MCM7 compared with the control substrate undergoing undeterred replication, whereas in HEK293 cells, only a background level of MCM7 was detected at the site of crosslink (Fig. 1D, 1E, and Fig. S1A). Similarly, MCM5, another component of the MCM2-7 complex was also found to accumulate at the site of the crosslink (Wang et al, unpublished data).
Having established that the eChIP assay allows us to identify proteins recruited to the site of a defined crosslink, we systematically investigated the recruitment of FA proteins to the site-specific DNA crosslink. First, we tested the recruitment of FANCA, FANCC, FAAP24, and FANCD2 to the defined psoralen crosslink in HEK293 and HEK293-EBNA cells. Results from the eChIP analyses (Fig. 2A, B and Fig. S1B, C) showed that components of the core complex, such as FANCA, FANCC, and FAAP24, were significantly enriched on the crosslink-bearing substrate compared with the unmodified substrate. Moreover, their enrichment did not require the replication of the crosslinked substrate, as reflected by the comparable levels of FANCA and FAAP24 enrichment in 293 and 293-EBNA cells. Similarly, FANCD2 was recruited to the site of the crosslink independent of the replication of the crosslinked substrate (Fig. 2B and Fig. S1D). These results suggest that both the FA core complex and FANCD2 are recruited to the site of crosslink in the absence of ongoing DNA replication.
In contrast to FA core complex proteins and FANCD2, the recruitment of FANCJ/BACH1 to the crosslinking lesion was found to be highly induced by DNA replication, as indicated by the dramatic increase of its enrichment in 293-EBNA cells compared with that in 293 cells (Fig. 2A, Fig. S1E). Subsequently, we examined FANCI, FANCN, and FANCD1 recruitment to the crosslinked DNA. As shown in Fig. 2C and Fig. S1F, FANCI recruitment was not affected by the replication status of the crosslinked substrate, similar to that of FANCD2 and components of the core complex. Conversely, enrichment of FANCD1/BRCA2 and FANCN/PALB2, a BRCA2-binding protein (Xia et al., 2006), were observed almost exclusively in 293-EBNA cells (Fig. 2C, D, Fig. S1 G, H). This result suggests that recruitment of breast cancer-related FA proteins depend on the presence of stalled replication forks. Moreover, using an eChIP substrate carrying a site-specific mitomycin C interstrand crosslink, we were able to reproduce the above results derived from psoralen-crosslinked DNA substrate (data not shown). Collectively, these observations are consistent with a model in which that the FA core and I/D2 complexes have a different recruiting requirement compared with the BRCA group of FA proteins (D1/BRCA2, N, and J) implicated in homologous recombination.
The EBNA-1 protein is present at the origin of the replicating eChIP substrate and could be a potential source for the enrichment of FA proteins if they interact with EBNA-1. To ascertain that recruitment of the FA protein to the crosslinked site is strictly lesion-dependent, IP-western was performed to determine whether EBNA-1 interacts with FANCD2 and FANCJ. As shown in Fig. 2E, under IP conditions identical to that of the eChIP, no interactions were detectable between EBNA-1 and FANCD2 and FANCJ, whereas the interaction between EBNA-1 and RPA (RPA2) was detected in a replication-dependent fashion as previously reported (Zhang et al., 1998). Thus the recruitment of the FA proteins to the crosslinked site is likely mediated by the presence of the crosslink lesion and stalled replication fork.
To determine the mechanism of recruitment among the different FA protein groups, we performed the eChIP assay in a FANCA-deficient lymphoblastoid cell line immortalized by EBV and consequently supporting EBNA-dependent pORIP replication. An EBV-immortalized normal lymphoblastoid cell line, ManEBV, was used as a positive control. We found that the enrichment of FANCD2 and FAAP24 to the crosslink was largely abolished in the mutant cells compared to WT cells or the FANCA mutant complemented with wild type FANCA (Fig. 3A, B and Fig. S2A). Given that a lack of FANCA leads to the dissociation of the FA core complex, this result suggests that the integrity of the core complex is critical for the recruitment of individual FA core component and FANCD2 to the site of DNA crosslink. The recruitment of FANCN and FANCJ, on the other hand, was unaffected by the loss of FANCA, as indicated by the comparable levels of enrichment of both proteins in wt, FANCA+ and FANCA− cells (Fig. 3C, D and Fig. S2B, C). Thus, it appears that the recruitment of FANCJ and FANCN is independent of the core complex and FANCD2/FANCI ubiquitination, suggesting that there may be unrelated functions supported by the BRCA-type FA proteins and the FANC core/ID2 complexes.
Repair of DNA interstrand crosslinks are carried out by both recombination-independent and recombination-dependent mechanisms. For cells in the G1/G0 phase, the absence of DNA replication dictates that the recombination-independent mechanism is the primary mode of crosslink repair (Sarkar et al., 2006; Wang et al., 2001; Zheng et al., 2003). Observing the canonical FA proteins at crosslinks independent of replication raised the possibility that the canonical FA pathway may be involved in recombination-independent repair of crosslinks. Therefore, we asked whether recruitment of FANCA was affected by the loss of key NER components, which prevents the initiation of recombination-independent repair. The XPC and XPA mutants, defective in NER lesion recognition and incision complex assembly, respectively, were analyzed by eChIP for their crosslink-specific enrichment of FANCA. We found (Fig. 4A) that loss of XPC or XPA led to markedly diminished FANCA recruitment compared to complemented wild type cells (WT). This result suggests that the canonical FA pathway may function downstream of NER process during recombination-independent crosslink repair.
To test whether the FA core-I/D2 proteins and the BRCA-related FA proteins affect crosslink repair differently, we examined directly the recombination-independent crosslink repair activities of FANCA and FANCJ mutants. Activity of this NER and lesion bypass-based mechanism was measured by reactivation of a defined crosslink positioned between the transcription initiation and the translation initiation sites (ATG) of a luciferase reporter (Shen et al., 2006; Zheng et al., 2003). Thus, removal of the crosslink can be quantified by the expression of the luciferase gene. Since undamaged sequences homologous to the reporter plasmid are not available from chromosomal DNA, repair of the psoralen crosslink is carried out in a recombination-independent and error-prone fashion (Shen et al., 2006; Zheng et al., 2003).
As shown in Fig. 4B, lack of FANCD2 reduced the recombination-independent crosslink repair since the ICL repair efficiency was decreased in the FANCD2 mutant cells (FANCD2−) as compared with the FANCD2-complemented cells (FANCD2+), suggesting that FANCD2 plays a role in the recombination-independent repair of ICLs. In contrast, recombination-independent crosslink repair activity was increased by FANCJ deficiency as reflected by the higher luciferase activity in FANCJ mutant cells (FANCJ−) than in the complemented cells (FANCJ+). Collectively, these results showed that recombination-independent crosslink repair is inversely affected by FANCD2 and FANCJ, consistent with the notion that FA proteins may support distinct mode of ICL repair.
The eChIP approach allows DNA damage processing/signaling factors to be directly linked to a defined lesion and detected with significantly improved sensitivity and specificity. Unlike many other viral replication origins, EBV ori possesses a defined replication initiation site with a virtual unidirectionality (Gahn and Schildkraut, 1989) that allows the generation of a crosslink-blocked replication fork at a precise location with enhanced synchrony. This is in contrast to mammalian genomes which lack defined replication origins. The replication elongation complex on the eChIP substrate consists of exclusively endogenous replication factors, enabling the eChIP system to faithfully reflect mammalian DNA replication.
The eChIP system allowed us to uncover distinct recruitment patterns among canonical FA and breast cancer-related FA proteins at the site of a defined crosslink. Our results provide the direct molecular evidence that the FA proteins, including the BRCA-related proteins, are recruited to the sites of DNA interstrand crosslinks. The presence of the FA core complex at a crosslinked site and its indispensable role in the recruitment of FANCD2 suggests that the FA core complex recognize crosslinked DNA or its processed intermediates. Given that the enrichment of both the core and the I/D2 complex occurs in nonreplicative crosslinked substrate, it is likely that FANCD2 ubiquitination/activation can be achieved in the absence of blocked replication. Indeed, we found that the introduction of crosslinked substrate, but not the unmodified control plasmid, resulted in FANCD2 ubiquitination in 293-EBNA as well as in 293 cells.
In G1/G0 cells where DNA replication is dormant, interstrand crosslinks are removed by a recombination-independent mutagenic pathway utilizing the combined actions of nucleotide excision repair and lesion bypass DNA synthesis (Sarkar et al., 2006; Wang et al., 2001; Zheng et al., 2003). Interestingly, recruitment of FANCA exhibited a strong dependence on XPC and XPA (Fig. 4A), both critical for the recognition of DNA crosslinks. Thus, prior recognition of the crosslinks by NER appears to precede the function of the FA core complex during recombination-independent ICL repair. We also observed visible losses of FANCA enrichment in EBV-immortalized XPA and XPC mutants (data not shown), most likely from the nonreplicating substrates subjected to recombination-independent repair. This results further supports that the FA core and I/D2 complexes may play a role in recognizing processed intermediate of recombination-independent mechanism. It is possible that a ubiquitinated FANC I/D2 complex may be required for the loading of a lesion bypass polymerase for gap synthesis. Consistent with this, FANCD2 mutant cells were found in our study to be partially defective in recombination-independent crosslink repair, which can be caused by a defect in the postulated FANC core/ID function of loading lesion bypass polymerase(s). In contrast, we found that loss of FANCJ led to enhanced recombination-independent crosslink repair. This result may reflect, albeit indirectly, a passive competition between the NER/lesion bypass-based homology-independent crosslink repair and the homology-dependent crosslink repair pathways, which presumably is defective in BRCA-related FA mutants (Litman et al., 2005). It is also conceivable that the Brca-related FA proteins are involved in re-establishing the collapsed replication forks after crosslink removal (Fig. 4C), which is likely reflected by the prolonged G2/M arrest in FANCJ mutant cells treated with mitomycin C (Fig. S3).
Collectively, it appears that FA proteins are involved in two distinct or branched processes dealing with crosslinks. Cell biology studies suggested that crosslink-induced BRCA2 and FANCD2 focus formation were independent of each other (Ohashi et al., 2005), although the exact role of replication in this process was not examined. The fact that the recruitment of FANCJ and FANCN does not require the FANC core complex further indicates at the molecular level that a stalled replication fork per se might be sufficient for the recruitment of recombination-related FANC proteins. Despite the similarities between the phenotypes of patients bearing the FA alleles of BACH1, PALB2, and BRCA2 and patients with mutations in the canonical FA genes, there is a shortage of evidence that these two branches directly work together. Our data indicate that they possess independent functions, at least with respect to their abilities to be recruited to sites of damage. This functional deviation between the canonical FA proteins and the BRCA- and homologous recombination-related FA proteins is perhaps reflected also by their distinct tumor spectrum with the latter exhibiting higher prevalence of breast cancer(Wagner et al., 2004). Thus, it appears that the complexities of the FA pathway may resemble those of the xeroderma pigmentosum (XP) proteins whose function encompass three distinct aspects of UV damage response that include lesion bypass synthesis, nucleotide excision repair, and damage signaling via ubiquitination.
The Epstein-Barr virus (EBV)-immortalized wild type ManEBV, FA-A (VU388) and complemented lymphoblast were maintained in RPMI medium (Sigma) supplemented with 10% of fetal calf serum. HEK293, HEK293EBNA (CRL-10852, 293c18) and MCF-7 (SC-9966) were obtained from ATCC. FANCJ (EUFA0030), XPA (XP2OS), XPC (XP4PA) and their complemented derivative (see Supplemental materials for details) were maintained in DMEM with 10% fetal calf serum.
Monoclonal or polyclonal antibodies to MCM7 (Santa Cruz/SC-9966), FANCA (Santa Cruz/SC-28215), FANCC (Santa Cruz/SC-18109), FANCD2 (Abcam/ab5360), FANCJ (Novus/NB100-1580), EBNA-1 (Santa Cruz/SC-100-1581), and RPA70 (Calibio/RPA70-9) were from commercial sources. Affinity-purified FAAP24, FANCD1, and FANCN were raised against recombinant antigens (Ciccia et al., 2007; Xia et al., 2007).
The eChIP substrate contains two essential components, the EBV OriP and site-specific DNA crosslinks downstream of the replication origin (see Supplemental materials for details). Preparation and insertion of psoralen or MMC crosslinked oligonucleotide into plasmid vectors have been described previously (Wang et al., 2001) and is described in Supplemental Materials.
Crosslinked or uncrosslinked control substrates were electroporated into cells (1.5 ×107 HEK293 or HEK293-EBNA cells, 5×106 lymphoblastoid cells) with an Amaxa device (Amaxa Biosystems, Germany) and conditions recommended by the manufacture. Chromatin immunoprecipitation was carried out with established procedures (details in Supplementary materials).
Real time PCR reactions were carried out in triplicates in a final volume of 10 ul with commercial reagent mix. 1% of the input chromatin sample and 10% of the ChIP sample were usually used as template. Details of the reaction and the amplification program are described in the Supplemental materials.
Preparation of the luciferase reporter substrate and performance of the assay were described previously (Shen et al., 2006; Zheng et al., 2003) and in the Supplementary materials. Briefly, reporter substrates with a transcription-blocking crosslink were introduced into cells by electroporation (lymphoblastoid, Amaxa Nucleofector) or FuGENE-6 (attached cells). A β-galactosidase reporter plasmid was included in each sample as an internal control for transfection efficiency. The linear range of luciferase activity was individually determined for each cell line.
The authors wish to thank Drs. John L Yates, Lee Zou, and Alan D’Andrea for helpful discussions and reagents. Dr. W Zhou provided technical help in real-time PCR. J. F. Culajay and J. Liu (Core C PO1-CA97175) assisted in the large scale preparation of crosslinked DNA substrates. This work is supported by grants from the NIH (CA97175, Project 3 to L. L, GM44664 to S. J. E). S. J. E. is an Investigator with the Howard Hughes Medical Institute.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.