Insights Into Chromatin Modification and FANCD2 Phosphorylation in DNA Damage Repair

Fanconi Anemia (FA) is a rare autosomal X-linked recessive disorder, characterized by congenital abnormalities, pediatric bone marrow failure and cancer susceptibility. FA is caused by biallelic mutation in any one of 22 different genes. The main functions of the FA-BRCA pathway is the resolution of interstrand crosslinks (ICLs) within the DNA. The main activating step of the pathway is the monoubiquitination of the proteins FANCD2 and FANCI. In the first chapter of this dissertation I discuss how methylation and acetylation affect chromatin architecture and activation of the pathway. In the second chapter I will discuss the effects of FANCD2 phosphorylation in a DNA damage-independent, cell cycle-dependent manner. We show that monoubiquitination of FANCD2 is blocked by phosphorylation and that this phosphorylation acts as a molecular switch to alter FANCD2 function.


INTRODUCTION
All organisms are continuously exposed to endogenous and exogenous DNA damaging agents, including reactive oxygen species and aldehydes from normal metabolic processes and UV irradiation from sunlight. The timely and accurate repair of DNA damage is essential for the maintenance of genome stability and organismal survival.
As a consequence, prokaryotic and eukaryotic organisms have evolved complex and highly orchestrated DNA repair pathways to effectively repair damaged DNA.
Chromatin represents the higher order macromolecular complex of DNA and histone proteins, and chromatin plasticity has become increasingly recognized as a major determinant of DNA damage recognition, signaling, and repair [1][2][3]. The nucleosome is the fundamental subunit of chromatin and exhibits plasticity via compositional alteration, translational repositioning, and the posttranslational modification of histone tails. Histone tails are subject to a wide variety of posttranslational modifications including acetylation, methylation, phosphorylation, and ubiquitination [4]. Histone acetylation homeostasis is mediated by histone acetyltransferases (HATs), e.g.
Fanconi anemia (FA) is a rare autosomal and X-linked genetic disease characterized by congenital defects, bone marrow failure, and increased cancer risk in early adulthood [8]. FA is caused by mutation of any one of 21 genes. The FA proteins function primarily in the repair of DNA interstrand crosslinks (ICLs), lesions that block the replication and transcription machineries, which lead to structural and numerical chromosome aberrations if repaired erroneously [8][9][10][11]. A central step in the activation of the FA pathway is the site-specific monoubiquitination of the FANCD2 and FANCI proteins [12][13][14]. Monoubiquitinated FANCD2 and FANCI localize to discrete sites within chromatin where they are hypothesized to promote the recruitment of several structure-specific endonucleases, including FAN1 (FANCD2-associated nuclease 1) and FANCQ/ERCC4 [15][16][17][18]. While FANCD2 and FANCI function primarily within chromatin, the contribution of chromatin plasticity, and specifically, the effects of changes in histone tail post translational modifications, on their activation and function have yet to be determined. Furthermore, while chromatin remodeling at DNA doublestrand breaks (DSBs) has been extensively studied [1][2][3], very little is known about the role of chromatin remodeling in the context of ICL repair.
In this study we have examined the influence of chromatin structure on the activation of the FA pathway. Specifically, we have examined the effects of several histone methyltransferase (HMT), demethylase (HDM), and deacetylase (HDAC) inhibitors on FANCD2 and FANCI monoubiquitination and their assembly into discrete nuclear foci.
We describe potent activation of FANCD2 and FANCI monoubiquitination in chromatin, and enhanced FANCD2 and FANCI nuclear foci formation, following cellular exposure to the HMT inhibitor BRD4770. BRD4770-induced activation of the pathway does not appear to occur via the direct induction of DNA damage perse, or via the inhibition of the G9a histone methyltransferase, a mechanism previously proposed for this molecule [19]. In contrast, our results suggest that BRD4770-induced activation of the pathway may be a consequence of inhibition of PRC2 (Polycomb Repressive Complex 2) and, specifically, its catalytic HMT EZH2. In addition, we demonstrate that inhibition of class I and II HDACs with trichostatin A (TSA) and vorinostat (SAHA) leads to attenuated ICL-inducible FANCD2 and FANCI monoubiquitination and nuclear foci formation. Our results establish that chromatin plasticity, and in particular the posttranslational modification of histone tails, is a critical determinant in the activation of the FA tumor suppressor pathway.

Activation of FANCD2 and FANCI monoubiquitination by the HMTi BRD4770.To
explore the effects of global alterations in histone methylation on the activation of the FA pathway, we exposed the transformed osteosarcoma cell line U2OS and the nontransformed telomerase (hTERT)-immortalized line BJ-TERT to the HMT inhibitors (HMTi) BRD4770 and BIX01294 and the HDM inhibitors (HDMi) GSK-J1 and PBIT, and examined FANCD2 and FANCI monoubiquitination. BRD4770 is a Sadenosylmethionine (SAM) mimetic and competitive inhibitor of PRC2/ EZH2 and G9a [19][20][21]. BIX01294 is a non-SAM mimetic selective inhibitor of G9a [21]. Treatment with BRD4770 resulted in a marked increase in FANCD2 and FANCI monoubiquitination in both U2OS and BJTERT cells, even in the absence of the ICLinducing agent mitomycin C (MMC) ( Figure 1A and B, lane 3). Indeed, the ratio of FANCD2-Ub to FANCD2 (L:S ratio) was higher in cells treated with BRD4770 alone than in cells treated with MMC alone ( Figure 1A and B, compare lanes 2 and 3).
BRD4770-induced activation of FANCD2 monoubiquitination also occurred in a concentration-dependent manner ( Figure 1C). In contrast, no major  effects on levels of spontaneous or ICL-inducible FANCD2/I monoubiquitination were observed for the other inhibitors tested, other than a slight increase in the FANCD2/I L:S ratios following treatment of BJ-TERT cells with GSK-J1 ( Figure 1B).

BRD4770 promotes FANCD2 chromatin localization and nuclear foci formation.
Next, we examined the effects of BRD4770 treatment on the localization of FANCD2 to chromatin and its assembly into nuclear foci. Cells were incubated in the absence or presence of BRD4770, both with and without MMC, and whole-cell (W), soluble cytoplasmic and nuclear (S), and chromatin-enriched (C) protein lysates were prepared and analyzed by immunoblotting. We observed greatly elevated levels of monoubiquitinated FANCD2 and FANCI in the chromatin fraction of cells treated with BRD4770 alone (Figure 2A, compare lanes 3 and 9). In addition, using immunofluorescence microscopy (IF), we analyzed FANCD2 nuclear foci formation, an indicator of the localization of FANCD2 to sites of damaged DNA in chromatin [12,22], following treatment with BRD4770 alone and following combined treatment with BRD4770 and MMC. We observed a striking increase in the percentage of nuclei exhibiting greater than 5 FANCD2 foci in both U2OS and BJTERT cells following BRD4770 treatment ( Figure 2B and Supplementary Figure 1A and B). Levels of FANCD2 nuclear foci formation in cells treated with BRD4770 alone were comparable to that observed following MMC treatment (Supplementary Figure 1A and B). These results identify BRD4770 as a major inducer of FANCD2 monoubiquitination and nuclear foci formation and strongly suggest that changes in histone methylation status are a critical determinant in the activation of the FA signal for cells treated with BRD4770 alone via immunoblotting, this level was markedly lower than that observed following exposure to the topoisomerase type II inhibitor etoposide (VP-16), a well known inducer of DNA DSBs, and no different to that observed following GSK-J1 treatment ( Figure 3B). We also examined levels of RPA S4/8 phosphorylation, a marker of single-stranded DNA [24], following BRD4770 exposure. While MMC treatment led to a strong increase in levels of RPA pS4/8, no increase in levels above that of untreated cells was observed for BRD4770 ( Figure 3C). We also examined the effects of BRD4770 treatment on levels of the FA core complex proteins FANCA and UBE2T/FANCT. UBE2T/FANCT is the FANCD2 E2 ubiquitin-conjugating enzyme [25]. Levels of both proteins decreased following BRD4770 treatment ( Figure 3C). Decreased levels of FANCA were also observed following exposure of HCT116 p53+/+ and p53-/-cells to BRD4770 (see Supplementary Figure 3D). Similarly, we observed a reduction in levels of the USP1 de-ubiquitinating enzyme following BRD4770 treatment ( Figure 3C). Concomitant reductions in the levels of FANCA, UBE2T/FANCT, and USP1 cannot explain the observed BRD4770-induced FANCD2 and FANCI monoubiquitination and nuclear foci formation.
We observed an increase in levels of phosphorylated CHK1 S345 following treatment with BRD4770, albeit to a lesser extent than that observed following MMC treatment ( Figure 3C). Finally, we examined the effects of BRD4770 treatment on cell cycle progression. Following exposure to BRD4770 for 24 h, we observed an increase in the percentage of cells in S-phase at all concentrations of BRD4770 examined ( Figure 3D).
However, following exposure to BRD4770 for 48 and 72 h, where maximal induction of FANCD2 and FANCI monoubiquitination was observed (see Figure 4B), the cell cycle stage profiles did not differ substantially from that of untreated cells ( Figure 3D).
Taken together, these results argue that BRD4770-induced activation of the FA pathway does not appear to be a consequence of direct induction of DNA damage, alterations in expression of the FANCD2 core monoubiquitination proteins, or major changes to cell cycle progression.  [19]. EZH2 is the catalytic HMT of PRC2, and catalyzes the deposition of the transcriptionally repressive mark H3K27me3 [26]. Therefore, to determine if BRD4770-induced activation of FANCD2 monoubiquitination might be a specific consequence of G9a and/or EZH2 inhibition, we treated cells with the DZNep and UNC0646 HMT inhibitors: DZNep treatment has been reported to lead to cellular depletion of EZH2 [27], while UNC0646 is a potent inhibitor of G9a [28]. We observed modest induction of FANCD2 and FANCI monoubiquitination following exposure to 2 and 4 μM DZNep ( Figure 4A). Considerable cell toxicity was observed at 10 μM DZNep (results not shown). Under these conditions, we did not observe a decrease in levels of EZH2 expression. However, reduced levels of EZH2 were observed following incubation with DZNep, and BRD4770 to a lesser extent, for 4 and 8 days (Supplementary Figure 3B). In contrast to DZNep, no induction of FANCD2/I monoubiquitination was observed following treatment with UNC0646, and combined DZNep/UNC0646 treatment resulted in considerable cell death ( Figure 4A and results not shown). Consistent with these findings, we also observed a modest yet statistically significant increase in FANCD2 nuclear foci formation in cells treated with DZNep and not with UNC0646

BRD4770-induced activation of the FA pathway may occur via inhibition of the
(Supplementary Figure 3C). We note that the degree of DZNep-induced FANCD2/I monoubiquitination was experimentally variable, most likely a consequence of its pleiotropic nature. To further explore the potential role of EZH2 and H3K27me3 in BRD4770-inducible FANCD2/I monoubiquitination, we exposed U2OS cells to BRD4770 for 24, 48, and 72 h and examined levels of EZH2 and H3K27me3 ( Figure   4B). We observed reductions in levels of H3K27me3 following treatment with higher concentrations of BRD4770 for 72 h, when induction of FANCD2 monoubiquitination was maximal ( Figure 4B). Under the same conditions, we did not observe any significant reductions in EZH2 levels ( Figure 4B). Similarly, we also observed a modest reduction in levels of H3K27me3 in HCT116 p53-/-cells treated with BRD4770 for 24 h and a more pronounced reduction in HeLa cells treated with BRD4770 for 72 h (Supplementary Figures 3D and 3E). We again observed induction of CHK1 pS345 upon exposure to higher concentrations of BRD4770 for extended periods ( Figure 4B).

EPZ-6438-mediated inhibition of PRC2/ EZH2 leads to activation of FANCD2
monoubiquitination. To further analyze the role of PRC2 and EZH2 in the activation of the FA pathway, cells were treated with EPZ-6438, an EZH2-specific inhibitor [29,30], and FANCD2 and FANCI monoubiquitination and nuclear foci formation were analyzed. In MCF10A cells, a spontaneously-immortalized, nontransformed, mammary epithelial line, EPZ-6438 treatment led to a pronounced increase in FANCD2 and FANCI protein levels, FANCD2 and FANCI monoubiquitination, and FANCD2 nuclear foci formation ( Figure 5A and Supplementary Figure 4). Interestingly, EPZ-6438 treatment led to an increase in levels of EZH2, possibly a cellular response to chemical inhibition of EZH2, and an overall reduction in levels of H3K27me3 ( Figure   5A). EPZ-6438 treatment also resulted in increased FANCD2 monoubiquitination in isogenic HCT116 p53+/+ and p53-/-cells ( Figure 5B) and HeLa cells ( Figure 5C). In contrast to MCF10A, HeLa, and HCT116, we did not detect increased FANCD2 monoubiquitination in U2OS cells treated with EPZ-6438 (results not shown). Taken together, our results suggest that BRD4770-induced activation of the FA pathway may occur via inhibition of the PRC2 complex, and specifically EZH2 HMT activity, and a consequent decrease in levels of H3K27me3. and vorinostat (SAHA) on FANCD2 and FANCI monoubiquitination and nuclear foci formation. Interestingly, for HeLa cells, we did not observe any appreciable differences in the levels of spontaneous or ICL-inducible FANCD2 or FANCI monoubiquitination when cells were treated with TSA or SAHA ( Figure 6A and 6B). In contrast, when BJ-TERT cells were treated with TSA or SAHA, we observed a marked reduction in the levels of ICL-inducible FANCD2 and FANCI monoubiquitination ( Figure 6C and D).

Inhibition of class I and II
TSA and SAHA treatment led to a reduction in ICL-inducible CHK1 S345 phosphorylation in both lines examined RBI, relative band intensity.
( Figure 6). In contrast, incubation with TSA and SAHA did not lead to any observable changes in levels of ICL-inducible CHK2 pT68 ( Figure 6). We observed a concentration-dependent increase in the levels of H4K16ac following treatment with TSA and SAHA confirming their inhibition of histone deacetylation ( Figure 6). We also  (class III) family of HDACs, did not impact cellular sensitivity to MMC ( Figure 8C). Taken together, our results indicate that HDAC1 and HDAC2 positively regulate activation of the FA pathway and that cellular sensitivity to ICL-inducing agents, which are widely used in cancer chemotherapy, may be increased via HDAC1/2 inhibition.

DISCUSSION
In this study, we have established that chromatin state is an important determinant in the activation of the FA pathway. Specifically, we have established that treatment with the HMTi BRD4770 and inhibition of the class I and II HDACs strongly impacts activation of the FA pathway. BRD4770 was recently discovered in a focused screen of a 2-substituted benzimidazole library [19]. While in in vitro biochemical assays, BRD9539, the carboxylic acid derivative of BRD4770, effectively inhibited both G9a and PRC2/EZH2 in a concentration-dependent manner, these studies concluded that BRD4770 functions primarily via the inhibition of G9a [19].  examined [32]. In four out of five lines examined in this study, EPZ-6438 treatment led to an increase in levels of FANCD2 monoubiquitination. The mechanism(s) by which inhibition of PRC2/EZH2 and decreased global levels of H3K27me3 would lead to activation of the FA pathway remain to be clearly elucidated. Recent studies in the silkworm Bombyx mori have shown that PRC2-mediated H3K27me3 increases following exposure to UV irradiation [33]. One possibility is that, upon exposure to DNA damaging agents, transcription may need to be halted at loci that have incurred DNA damage. An inability to catalyze H3K27me3 and arrest transcription could lead to the formation of co-transcriptional RNA-DNA hybrids (R-loops). An important role for the FA proteins in the repair of R-loops has recently been established [34,35].
However, our γH2AX and RPA pS4/8 results strongly suggest a DNA damageindependent mode of action for BRD4770. An alternative hypothesis is that BRD4770 treatment -possibly via both G9a and PRC2/ EZH2 inhibition -leads to the general establishment of a transcriptionally permissive chromatin state, which leads to the recruitment of factors that promote homologous recombination (HR) DNA repair, such as FANCD2 and FANCI [13,36]. Consistent with this hypothesis, Aymard et al have recently shown that HR factors are enriched at transcriptionally active chromatin [37].
It is important to note, however, that BRD4770-induced activation of the FA pathway most likely does not occur solely via the inhibition of PRC2/EZH2: BRD4770 induces FANCD2/I monoubiquitination more robustly than the EZH2-specific inhibitor EPZ-6438, and EPZ-6438 treatment results in more pronounced decreases in levels of H3K27me3. Therefore, other mechanisms are likely to contribute to the observed effects of BRD4770. In our study, we also detected

Cell culture
The Statistical significance was determined using paired two-tailed Student's t-test analysis.

Chromatin fractionation
Cells were plated at density of 3

Cell proliferation assay
Cells were plated at a density of 10,000 cells/well in 96-well dishes, incubated in the absence or presence of drug(s) for 48 h. CellTiter 96® AQueous One Solution Reagent (MTS) (Promega) was added directly to the wells, incubated for a further 2 h, and the absorbance at 490 nm was measured using a 96-well Bio-Rad 680 microplate reader.

Cell-cycle analysis
Cells were plated at a density of 1x106 cells in 10 cm2 dishes. The following day, cells

ACKNOWLEDGMENTS
We thank members of the Howlett laboratory for critical reading of this manuscript and for helpful discussions. We also thank Dr. Adrian Bracken of the Smurfit Institute of Genetics at Trinity College Dublin for helpful advice on EZH2 reagents.

CONFLICTS OF INTEREST
The authors declare that they have no competing financial interests.

FUNDING
This work was supported by National Institutes of Health/National Heart, Lung

Introduction
Fanconi anemia (FA) is a rare, X-linked and autosomal recessive genetic disorder. the core complex will be recruited to the ICL and promote the monoubiquitination of FANCD2 and FANCI [3,6,7]. The monoubiquitination of FANCD2 and FANCI is a key activating step in the pathway; a step is absent in over 90% of FA patients. Upon its monoubiquitination FANCD2 and FANCI will work to recruit downstream proteins that will resolve the ICL in a homologous recombination (HR) mediated process [6,[8][9][10]. Without the FA pathway the ICL will be resolved in an error prone process known as non-homologous end joining (NHEJ), resulting in extensive DNA damage [11,12].
In addition to monoubiquitination, FANCD2 and FANCI are also post translationally modified (PTM) through phosphorylation. It has been shown that FANCD2 is phosphorylated by ATM (Ataxia-telangiectasia mutated) in response to ionizing radiation (IR). FANCD2 is also phosphorylated by ATR (ATM and Rad3-related) when cells are treated with ICL inducing agents [13][14][15]. It is thought that DNA damage dependent FANCD2 phosphorylation is key for its monoubiquitination and heterodimerization with FANCI. ATM/ATR are also responsible for FANCI phosphorylation. FANCI is phosphorylated in a DNA damage dependent manner on six SQ/TQ residues that are proximal to its site of monoubiquitination [16]. The phosphorylation of FANCI has been shown to function both upstream and downstream of monoubiquitination [17,18].
Though the main function of FANCD2 is in the DNA damage response (DDR),there is evidence to show that it functions in other cellular processes. For example, FANCD2 is critical in the resolution of RNA:DNA hybrids known as R-loops [19]. There is also mounting evidence that suggests FANCD2 is involved in replication fork firing and stability [20][21][22]. The recent Molecular Cell paper by Madireddy et al. showed that FANCD2 is required for stability at common fragile sites (CFS) and that it may regulate dormant origin firing under replication stress [23]. In addition to this, FANCD2 has been shown to protect stalled replication forks from MRE11 degradation and, via its interaction with CtBP-interacting protein (CtIP), it is involved in replication fork restart [24,25].
In this manuscript, I will present data that shows that FANCD2 is phosphorylated in a DNA damage independent cell cycle dependent manner. This phosphorylation is at its height during unperturbed S-phase of the cell cycle. Both in-silico and ex-vivo studies done by our lab suggest that a cyclin dependent kinase (CDK) is responsible for this phosphorylation. CDKs, along with their cyclin binding partners, are responsible for regulation of the cell cycle, halting it if damage is present [26]. There is precedent for CDKs acting in the DDR, CDKs have been shown to phosphorylate BRCA1 and BRCA2 (both of which are within the FA-BRCA pathway) [27][28][29]. We have identified 3 putative CDK sites that are proximal to the sites of monoubiquitination. Two mutants have been created at these sites. One mutant cannot be phosphorylated (phospho-dead) and the other mutant behaves as if it is constitutively phosphorylated, (a phosphomimetic). Surprisingly, it appears that the phospho-dead mutant is constitutively monoubiquitinated while the phospho-mimetic is deficient in monoubiquitination.This is at odds with the characterization of FANCI phosphorylation, in which phosphorylation promotes monoubiquitination [16,18]. Our data also suggests that phosphorylation of FANCD2 may act as a "molecular switch" to activate other functions, in particular its roles in DNA replication. Characterization of these phosphorylation sites will shed light on the interplay between monoubiquitination and phosphorylation, highlighting new roles for FANCD2 and the importance of temporal regulation of monoubiquitination.

Phosphorylation of FANCD2 is DNA damage independent and cell cycle dependent
To elucidate whether FANCD2 phosphorylation was occurring in a DNA damage independent manner a simple experiment was performed. U2OS cells were treated

FANCD2 phosphorylation is mediated by a cyclin dependent kinase
DNA damage independent phosphorylation of FANCD2 appears to be related to the cell cycle. With this in mind, we began to look for candidate kinases that could be performing this phosphorylation. Since we know the phosphorylation is cell cycle specific, we started investigating cell cycle specific kinases. This led us to CDKs, CDKs are active during S-  An IP was performed using FLAG agarose. In both IPs, the pull down product was probed with an antibody that recognizes serine residues phosphorylated by CDKs. The antibody detected a FANCD2 band in both IPs (Figure 2A), suggesting that FANCD2 is a CDK phosphorylation target. Mass spectrometry analysis of the U2OS 3xFLAG FANCD2 IP showed a phosphorylation of FANCD2 at S692. In addition, our collaborator at the University of Melbourne performed a similar experiment using X. laevis FANCD2 and found a phosphorylation at S726 following mass spectrometry analysis ( Figure 2B). These both represent novel sites of phosphorylation.

FANCD2 contains a putative CDK cluster proximal to the site of monoubiquitination
The CDK consensus sequence is [S/T*]PX[K/R] [30], however, there are examples of proteins be phosphorylated by a CDK that only contain the S/P of the consensus sequence, such as BRCA1 [31]. When the FANCD2 amino acid sequence was analyzed many putative CDK S/P sites were found (data not shown). To hone in on sites that might have an effect on DNA repair, in-silico analysis was performed. When the sites are mapped onto the mouse ID2 heterodimer (PDB ID:3S4W), human ID2 has yet to be crystallized, we see that four of these sites are proximal to K562, the site monoubiquitination in mouse fancd2 ( Figure 3B). In addition, when a multi-species alignmentis performed using Clustal Omega three of these sites, S525, S264 and S726, demonstrate strong evolutionary conservation. The fourth sites was on an uncrystallized loop within mouse fancd2, we did not recognize this site until much later so analysis of this site is ongoing. We hypothesized, based on their proximity to K561, that the three identified sites were the most likely to have an effect on FANCD2 monoubiquitination, based on their proximity to K561. Additionally, two of these sites, S592 and S724, were found in the mass spectrometry results discussed earlier. We next stably expressed two different V5-tagged FANCD2 mutants in a FA-D2 patient cell line. One mutant has S525, S264 and S726 mutated to alanines (TA), this mutant is unable to be phosphorylated at these sites. The second mutant has the same serines mutated to aspartic acids (TD), the negative charge on the aspartic acid acts as phospho-mimetic (FANCD2 appears constitutively phosphorylated at these sites). V5 tagged wildtype (WT) FANCD2 and an empty vector were stably expressed as controls.
An experiment was performed in which each cell line was incubated with and without MMC. When we immunoblotted, probing for FANCD2-V5, we found that the TA mutant appeared to be constitutively monoubiquitinated while the TD mutant was deficient in monoubiquitination ( Figure 3C).

Phosphorylation of FANCD2 plays an important role in foci formation
There is evidence to suggest that monoubiquitination of FANCD2 and aggregation within the nucleus to form foci can be uncoupled, as seen in USP1 autocleavage mutants [32]. To examine if this was occurring in our mutants, cells were treated with both MMC and aphidicolin (APH), a known polymerase inhibitor and replication stressor [33]. The cells were fixed and stained for FANCD2-V5, FANCI and DAPI. The TA mutants had statistically significantly higher FANCD2-V5 and FANCI foci in the no treatment condition than either WT or TD ( Figure 4A, B). Upon treatment with both MMC and APH, the TD mutant had statistically significant lower foci formation than either TA or WT. This indicates that not only is the TA mutant deficient in monoubiquitination but FANCD2 is unable to localize to the sites of damage within the nucleus. In addition, the TA mutant seems to be constitutively monoubiquitinated and form foci even in the absence of damage.
It is noteworthy that the effects on foci formation also occurred in FANCI, indicating that phosphorylation of FANCD2 does not increase the monoubiquitination of FANCI but does increase its foci formation. This brings up questions about the importance of FANCD2 phosphorylation on FANCI-FANCD2 dimerization and aggregation within the nucleus.

FANCD2 phosphorylation effects chromosomal stability
One of the hallmarks of Fanconi Anemia is hypersensitivity to DNA damaging agents.
Chromosomal breakage analysis of peripheral blood samples treated with damaging agents are often used in clinical setting to diagnose FA [34]. We performed metaphase spread analysis using our mutants, looking for chromosome gaps and breaks, dicentric chromosome and more complex chromosome aberrations such as radials. When our mutants were treated with low doses of APH, the TA mutant had a substantial increase in damage over both the WT and TD. The TD had slightly lower amounts of damage than the WT ( Figure 6). Even though the TA mutant had both an increase in monoubiquitination and foci formation, this did not correlate to decreased chromosomal aberrations and the inverse seems to be true for the TD. We believe that this showcases the importance of controlling the timing of FANCD2 monoubiquitination. It is detrimental to chromosomal stability for FANCD2 to be constantly activated, FANCD2 monoubiquitination must be precisely timed in order for proper DNA repair to take place.

Mutation of the FANCD2 CDK cluster impacts fork restart following damage
Our research has shown that FANCD2 is phosphorylated in S phase of the cell cycle and that this phosphorylation occurs in the absence of any DNA damage. These are interesting observations but they do not answer the basic question of what the purpose of this phosphorylation is. Since this phosphorylation is specific to S-phase, the logical inference is that the phosphorylation at these sites is somehow regulating FANCD2's function in DNA synthesis. To explore this, DNA strand analysis was performed to analyze both fork restart following damage and the affects our mutations have on the total DNA synthesis occurring. Cells were treated with 0.2 μM APH for 12 hours, pulsed for 20 minutes with iodouridine followed by a 20 minute pulse with chlorouridine, these were stained red and green respectively. In addition, a single strand antibody was used that stains the total DNA content of the cell. When these images are overlaid with the red/green images the percent of active replication can be calculated ( Figure S4). When analyzing the red and green staining, we took the ratio of the red to the green. The length of the red portion represents fork progression directly after DNA damage, while the green portion has fully recovered and represents unperturbed synthesis. Without APH treatment there is a red:green ratio of around 1, which should be expected since there is nothing to hinder the replication fork ( Figure 7A). APH treatment results in lower red:green ratios because the cells have to recover from the treatment before progressing, however the TA mutant has a lower red:green than either WT or TD ( Figure 7A). This illustrates replication deficiencies in the TA mutant, in particular where fork restart is concerned.
Interestingly, the TD maintained its red:green ratio of 1 and seemed almost unaffected by the APH treatment. When the proportion of DNA that is actually replicating is analyzed ( Figure 7B), WT and TD seem to be completely unaffected by the APH treatment while TA actually has a higher proportion of DNA replication following APH treatment. This seemed counterintuitive to us until dormant origin firing was taken into account. When a replication fork stalls, it is common for another origin to fire nearby to complete the replication, this is termed dormant origin firing [35]. So based on the way in which we analyzed the data, this would be counted as having a higher proportion of replicating DNA, when in fact the total amount of DNA replicated would remain static and only the number of origins would change.

Discussion
In this study, we have shown that FANCD2 is extensively phosphorylated during S-phase of the cell cycle. We have also presented data showing that this phosphorylation is DNA damage independent. Both our FANCD2-V5 and FANCD2-3XFLAG immunoprecipitations suggest that a CDK is responsible for this phosphorylation, a result also supported by mass spectrometry.. When we examine the mobility shift that is present when FANCD2 is treated with lambda phosphatase, we see that this shift must be from the hyper-phosphorylation of FANCD2 (a single phosphate weighs ~80 daltons and our phosphate shift is 2-3 kilodaltons). This shift is not present in FANCI even upon the addition of MMC, signifying that FANCD2 is more extensively phosphorylated than FANCI. It has been shown that both FANCD2 and FANCI are phosphorylated by the DNA damage kinases ATM/ATR, this is dependent on DNA damage [13,14]. Our data suggest that a CDK may also be acting in the pathway to regulate FANCD2 function independent of DNA damage. There is precedent for CDKs phosphorylating DDR protein however it is unique for them to have a DDR protein substrate in the absence of DNA damage.
When the CDK cluster was mutated to alanine (TA mutant) or aspartic acids (TD mutant) we gained insight into the functional significance of these phosphorylations. It appears that phosphorylation at these sites lowers the overall concentration of monoubiquitinated FANCD2 within the cell ,while inhibiting phosphorylation increases monoubiquitination.
This is the inverse of FANCI, where phosphorylation has been shown to promote monoubiquitination [16,18]. The phospho-dead TA mutant was also able to form foci in the absence of damage while the TD mutant was deficient in foci formation even in the presence of DNA damage. We had expected that constitutive monoubiquitination of FANCD2 may have decreased DNA damage by promoting ICL repair and HR, however, this was not the case. Through chromosomal breakage analysis we see that the TA mutant has higher amounts of damage than either WT or TD and TD had levels of damage slightly lower than WT. When we examined the effects that the mutants had on DNA synthesis, the TA had trouble recovering from APH treatment and appeared to have more active replication occurring, we believe that this is due to dormant origin firing. The TD appeared to be unaffected by damage and continued DNA replication unperturbed.
FANCD2 has been shown to be involved in DNA replication in several recent studies [20,21,22]. In particular, the recent paper by Madireddy et al. highlights the role of FANCD2 in replication across common fragile sites. They state that FANCD2 is important for replication fork stability and efficient replication initiation at these large repeat rich genomic regions [23]. We propose that FANCD2 phosphorylation at these sites acts a "molecular switch" to alter its function from ICL repair to replication fork maintenance. This is supported by both our chromosomal breakage and DNA fiber analysis. Even with high levels of monoubiquitination the TA mutant has extensive chromosomal aberrations and slower fork restart following APH treatment. Since the TA cannot be phosphorylated, it cannot switch roles from ICL to fork maintenance, this results in fork collapse and DNA damage. Upon replication stress it is essential for FANCD2 to be able to switch functions and prevent fork collapse and DNA double strand breaks.
The mechanism by which phosphorylation of this CDK cluster promotes FANCD2's role in replication fork stability is still under investigation, however, we can speculate as to how it is occurring. One possibility, which has been proposed and studied by other groups, is that FANCD2 arrives at points of instability within the genome, such as CFS, and works to resolve R-loops at these sites [19,23]. It is well documented that R-loops occur with a higher frequency during DNA replication and the transcription of genes especially under replication stress; FANCD2 has been shown to resolve R-loops in both murine and human models [19,36]. Phosphorylation of this CDK cluster may block monoubiquitination of FANCD2 and promote its role in R-loop resolution, which would increase the stability of stressed replication forks. It is also possible that by blocking the monoubiquitination of FANCD2, through phosphorylation, we are allowing FANCD2 to be recruited to replication forks and recruit enzymes that prevent fork collapse. Recent evidence points to fanconi associated nuclease 1 (FAN1) as a candidate for this function. FAN1 has been shown to prevent replication fork progression upon fork collapse, thereby preventing DNA damage, this is a completely separate function from its roles in ICL repair [9,37]. It is known that FAN1 associates with FANCD2 at ICLs, a similar association could occur at replication forks to halt the fork upon collapse and prevent any DNA damage.
Our mutants will be key in picking apart the mechanism and function of this novel CDK cluster within FANCD2. We still need to do further analysis to answer some key questions such as 1) Does this phosphorylation inhibit FANCD2 monoubiquitination or promote its deubiquitination? 2) Which CDK is responsible for this phosphorylation? 3) How does this phosphorylation promote fork restart and prevent chromosomal aberrations? Given that CDK inhibitors are in clinical use and that over 90% of FA patients are deficient in FANCD2 monoubiquitination, answering these questions could reveal unexplored treatment opportunities for patients

Cell-cycle synchronization
HeLa cells or the indicated mutants were synchronized by double thymidine block method.
Cells were treated with 2.5 mM thymidine for 18 hours, thymidine-free media for 10 hours,

DNA fiber analysis
For chromosome breakage assays, cells were incubated in the absence or presence of .2µM APH for 12 h. The APH media was removed and media containing 30μM iodouridine(sigma;, this media was removed and the cells were treated with 30µM chloropuridine. Cells were harvested pelleted and the total genomic DNA was purified. The DNA was "stretched" on glass silane coated slides (sigma; S4651-72EA) using a Corning cover glass. The coverslip was then treated with three primary antibodies, against iodouridine, chlorouridine and a ssDNA antibody. A secondary antibody with an attached fluorophore was used against each primary. These slides were mounted using  USP1 is depicted in gray font to signify its uncertain role in this process.