AMP-activated protein kinase is involved in the activation of the Fanconi anemia/BRCA pathway in response to DNA interstrand crosslinks

Fanconi anemia complementation group (FANC) proteins constitute the Fanconi Anemia (FA)/BRCA pathway that is activated in response to DNA interstrand crosslinks (ICLs). We previously performed yeast two-hybrid screening to identify novel FANC-interacting proteins and discovered that the alpha subunit of AMP-activated protein kinase (AMPKα1) was a candidate binding partner of the FANCG protein, which is a component of the FA nuclear core complex. We confirmed the interaction between AMPKα and both FANCG using co-immunoprecipitation experiments. Additionally, we showed that AMPKα interacted with FANCA, another component of the FA nuclear core complex. AMPKα knockdown in U2OS cells decreased FANCD2 monoubiquitination and nuclear foci formation upon mitomycin C-induced ICLs. Furthermore, AMPKα knockdown enhanced cellular sensitivity to MMC. MMC treatment resulted in an increase in AMPKα phosphorylation/activation, indicating AMPK is involved in the cellular response to ICLs. FANCA was phosphorylated by AMPK at S347 and phosphorylation increased with MMC treatment. MMC-induced FANCD2 monoubiquitination and nuclear foci formation were compromised in a U2OS cell line that stably overexpressed the S347A mutant form of FANCA compared to wild-type FANCA-overexpressing cells, indicating a requirement for FANCA phosphorylation at S347 for proper activation of the FA/BRCA pathway. Our data suggest AMPK is involved in the activation of the FA/BRCA pathway.


INTRODUCTION
DNA interstrand crosslinks (ICLs) are a severe form of DNA damage induced by alkylating agents and platinum drugs such as cisplatin and mitomycin C (MMC) [1]. In response to ICLs, cells activate the Fanconi Anemia (FA)/BRCA pathway [2,3]. The FA/ BRCA pathway is composed of FA proteins, BRCA1/2, and other associated proteins [4]. FA is a human syndrome with diverse phenotypes including retarded growth, short stature, neurological degeneration, and a predisposition to cancer [5]. Patients with FA exhibit hypersensitivity to DNA crosslinking agents, and this observation led to the elucidation of the role of the FA/BRCA pathway in the response to ICLs [5]. FA is caused by mutations in one of 19 FA genes, which are all named with the root symbol FANC (e.g. FANCA and FANCB) [6]. The FA nuclear core complex consists of eight FA proteins including FANCA and FANCL E3 ligase, which is activated in response to DNA damage and monoubiquitinates FANCD2 [7][8][9]. Modified FANCD2 forms nuclear foci in regions of DNA damage and plays a role in homologous recombination repair through interaction with BRCA1 [10,11].
FANC proteins have also been implicated in other cellular processes (reviewed in [12]). For example, FANCC and FANCA may participate in the cytokine

Research Paper
response through interactions with signal transducer and activator of transcription 1(STAT1) and IκB kinase, respectively [13,14]. FANCA and other FANC proteins have also been suggested to function in transcriptional regulation via interactions with Hes1 and Brg1 [15,16]. Additionally, FANCC and FANCG are involved in the response to oxidative damage [15]. We previously demonstrated that the interaction between FANCA and the Nek2 kinase is involved in centrosome separation, and suggested a role for the FANCA protein in centrosome maintenance during cell cycle progression [17]. Similarly, other groups have described the involvement of FANC proteins in centrosome maintenance and mitotic progression [18].
To identify novel protein interactions involving FANC proteins, we previously performed yeast twohybrid screening using FANC proteins as bait [17]. Among the candidate interacting proteins, AMP-activated protein kinase (AMPK) was selected for further analysis.
AMPK is an energy sensor that is activated in response to low ATP levels. It phosphorylates various substrates including sterol regulatory element-binding protein 1 (SREBP1), acetyl-CoA carboxylase 2 (ACC2), and tuberous sclerosis 2 (TSC2) [19,20]. AMPK has also been implicated in the DNA damage response and control of mitotic progression [19]. In particular, it is a downstream substrate of ATM and modulates cell cycle regulators including p53 and p21 [21,22]. Activated AMPK localizes to centrosomes during metaphase and the spindle midzone during telophase and cytokinesis [23]. Here, we determined that AMPK is required for the activation of the FA/BRCA pathway in response to DNA damage.

AMPKα interacts with FANCG and FANCA
In a previous study, we performed yeast two-hybrid screening to analyze a novel protein network involving FANC proteins [17]. We determined that the protein kinase Nek2, which plays a role in centrosome separation, interacts with FANCA and that FANCA had a novel function at centrosomes. We also identified AMPKα1 as a FANCG-interacting protein and confirmed a direct interaction between the two proteins using pull-down assays (Supplementary Figure S1). Next, we verified the interaction in human embryonic kidney (HEK) 293T cells by co-immunoprecipitation after expression of V5-AMPKα1 and HA-FANCG ( Figure 1A). In this experiment, V5-AMPKα1 was also detected in the HA-FANCA immunoprecipitates, indicating an interaction with FANCA, possibly in the FA nuclear core complex. The interactions between AMPKα1 and both FANCG and FANCA decreased after treatment with MMC. We also confirmed the interaction between HA-FANCG and endogenous AMPKα in cells transfected with pcDNA3-HA-FANCG ( Figure 1B). Endogenous AMPKα and FANCA had the capacity to interact ( Figure 1C), suggesting AMPK was involved in activation of the FA/ BRCA pathway.

Activation of AMPK upon induction of ICLs
To analyze the involvement of AMPK in the cellular response to ICLs, we investigated whether AMPK was phosphorylated and activated by MMC treatment. Activation of AMPK was indicated by phosphorylation at T174 of AMPKα1 and T172 of AMPKα2. Treatment of U2OS cells with MMC resulted in a dose-dependent increase in phosphorylated AMPKα (Figure 2), suggesting that AMPK was involved in the cellular response to ICLs.

AMPK knockdown inhibits MMC-induced monoubiquitination of FANCD2
To confirm the involvement of AMPK in the activation of the FA/BRCA pathway, we assessed FANCD2 monoubiquitination and nuclear foci formation after MMC treatment in cells transfected with anti-AMPKα1 siRNA. FANCD2 monoubiquitination was readily detected by mobility shift on Tris-acetate gels. The levels of the mobility-shifted form of FANCD2 (FANCD2-L) were lower in AMPK siRNA-transfected U2OS cells (siAMPK#8) than in control siRNA-transfected cells (siControl) ( Figure 3A), indicating AMPKα was required for activation of the FA/BRCA pathway. We also found that FANCD2 expression decreased after siAMPK#8 transfection. This reduction could have been caused by inhibition of FANCD2 transcription, because FANCD2 mRNA levels decreased in siAMPK#8-transfected cells in real-time quantitative reverse transcription (RT)-PCR experiments (Supplementary Figure S2). Treatment with MG132, a proteasome inhibitor, did not restore expression, indicating AMPKα did not control FANCD2 stability (data not shown). To rule out off-target effects of the siRNA, we constructed the expression vectors for siRNA-resistant AMPKα1 (pcDNA3-Res-V5-PRKAA1) by site-directed mutagenesis of pcDNA3-V5-PRKAA1. Overexpression of siRNA-resistant AMPKα1 (Res-PRKAA1) attenuated inhibition of MMC-induced FANCD2 monoubiquitination (Supplementary Figure S3A). In contrast, overexpression of the siRNA-resistant T174A mutant form of AMPKα1 did not rescue the inhibition of FANCD2 monoubiquitination (Supplementary Figure  S3B). These results were confirmed in U2OS cells that stably expressed siRNA-resistant AMPKα1 using a lentiviral system (Supplementary Figure S3C). Finally, co-treatment with MMC and Compound C, a specific inhibitor of AMPK, inhibited MMC-induced FANCD2 monoubiquitination, indicating AMPK was required for FANCD2 monoubiquitination (Supplementary Figure S4). www.impactjournals.com/oncotarget Monoubiquitinated FANCD2 forms nuclear foci around regions of DNA damage. We evaluated FANCD2 nuclear foci formation using confocal microscopy, and found that the number of nuclear foci per cell was lower in siAMPK#8-transfected cells compared to siControltransfected cells ( Figure 3B). Furthermore, MTT assays revealed that MMC sensitivity increased in siAMPK#8transfected cells ( Figure 3C). In contrast, an increase in MMC sensitivity was not observed in cells that stably expressed siRNA-resistant AMPKα1 (Supplementary Figure S5).

Phosphorylation of S347 upon DNA damage
To confirm S347 phosphorylation in cells, we generated a phospho-specific antibody against phospho-S347 (P-S347) and used it for immunoprecipitation and western blotting. The levels of P-S347 increased after MMC treatment in cells transfected with HA-FANCA ( Figure 5A). In addition, the AMPK activating chemical A769662 increased P-S347 levels, suggesting S347 was phosphorylated by AMPK. In this experiment, HA-FANCA-S347A transfected samples produced no signal at all, which confirmed the specificity of the antibody. Finally, we demonstrated that phosphorylation of endogenous FANCA increased upon MMC treatment ( Figure 5B).

DISCUSSION
The identification of AMPKα1 as a FANCGinteracting protein prompted us to investigate a possible connection between AMPK and FA/BRCA pathway activation. An association between AMPKα and both FANCA and FANCG was detected by coimmunoprecipitation. We hypothesize that the association between AMPKα and FANCA primarily occurs at the level of the FA nuclear core complex. Co-immunoprecipitation experiments revealed a weaker interaction of HA-FANCA   with V5-AMPKα1 than with HA-FANCG ( Figure 1A). Additionally, we detected an interaction between AMPKα and FANCE, another component of FA core complex in co-immunoprecipitation experiments (data not shown). However, the possibility of a direct association between FANCA and AMPKα could not be completely ruled out in that a FANCA fragment (FANCA-F2) could be phosphorylated by AMPK in vitro. We also detected a direct association between recombinant His-tagged AMPKα1 and GST-tagged FANCA-F2 in pull-down assays (Supplementary Figure S6).
Our results indicate that AMPK, a well-known molecular sensor of energy stress, is involved in the cellular response to ICLs. AMPKα was activated by phosphorylation in response to treatment with MMC. Additionally, AMPKα1 knockdown inhibited full activation of the FA/BRCA pathway, which plays a signaling role in the cellular response to DNA damage. Finally, FANCA may be phosphorylated by AMPK at S347, which is important for activation of the FA/BRCA pathway. Consistent with these findings, a previous study described AMPK activation in response to treatment with cisplatin, which is an ICL-inducing agent [27]. AMPK is also a downstream effector of ATM, a key regulator of the DNA damage response [19,28].
A connection between metabolic pathways and the DNA damage response is emerging. ATM, which is a key regulator in the DNA damage response, contributes to the oxidative stress response and regulates mitochondrial function [28]. In addition, metabolic pathways such as glycolysis and glutaminolysis may promote DNA damage repair in cancer cells [29]. Several transcription factors such as Myc, p53, E2F1, and E4F1 have been shown to control mitochondrial function and cell cycle checkpoints [30]. Thus, the involvement of AMPK in activation of the FA/BRCA pathway upon induction of ICLs suggests that the cellular processes that regulate the metabolic and DNA damage stress responses are closely related.
The interaction between AMPK and the FA/ BRCA pathway suggests that FANC proteins might be involved in the regulation of cellular energy metabolism. Indeed, FA cells had damaged mitochondria and defects in the mitochondrial respiratory complex I [31][32][33]. Thus, FANC proteins may modulate cellular energy metabolism by interacting with AMPK. Notably, FANCA and activated AMPK have been reported to localize to centrosomes [17][18][19]. The interaction between FANCA and AMPK may also function at centrosomes and midzones during mitosis. The S347-phosphorylated form of FANCA could be enriched at these structures where it may function in the coordination of mitosis and cytokinesis.
Inhibition of AMPK with Compound C has anticancer effects in several cancer types including prostate, colorectal, and breast cancer [34][35][36]. Compound C also sensitizes cells to anticancer drugs such as cisplatin [37]. Given that cisplatin has DNA crosslinking activity and activates the FA/BRCA pathway, Compound C may inhibit activation of the FA/BRCA pathway. Therefore, it may be possible to overcome resistance to ICL-inducing therapies (e.g. cisplatin and MMC) by modulating AMPK activity in cancers that highly express FANC proteins in the FA/BRCA pathway [38].
Overall, our results suggest that AMPK may be required for proper activation of the FA/BRCA pathway upon DNA damage via FANCA phosphorylation. This finding may provide a rationale for exploiting AMPK as a target for anticancer chemosensitization strategies.

Confocal microscopy
U2OS cells were grown on coverslips (18 mm diameter; Marienfeld Superior, Königshofen, Germany) in 12-well plates, fixed in 3.7% formaldehyde in phosphatebuffered saline (PBS) for 20 min, permeabilized with 0.2% Triton X-100 in PBS for 20 min, and blocked with 1% bovine serum albumin in PBST (PBS with 0.2% Tween-20). The coverslips were sequentially incubated with an anti-FANCD2 antibody (Novus Biologicals) overnight and with an Alexa 488-conjugated donkey anti-rabbit secondary antibody for 2 h. After extensive washing with PBST, the cells were counterstained with 4', 6-diamidino-2-phenylindole, mounted on glass slides, and observed using a Zeiss Axiover LSM780 microscope and ZEN acquisition software (Carl Zeiss, Wetzlar, Germany). The number of foci per cell was counted in > 20 cells per sample. For HA-FANCA-expressing U2OS cells, we used an anti-HA primary antibody (Covance Laboratories, Dedham, MA, USA) and an Alexa 546-conjugated donkey anti-mouse secondary antibody.

Generation of a phospho-specific antibody
Rabbits were immunized with a phospho-peptide containing the CQREWpSFART sequence.

Immunoprecipitation and western blotting to detect phosphorylated FANCA
HEK293T cells transiently transfected with pcDNA3-HA-FANCA-WT or the S347A mutant were treated with 200 ng/mL MMC for 16 h. Cell lysates were immunoprecipitated with 5 μL phospho-specific S347 antisera (P-S347) and the immunoprecipitates were subjected to 3-8% NuPAGE Tris-acetate gel electrophoresis. FANCA was detected by immunoblotting with an HRP-conjugated anti-HA antibody (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA). The presence of equal amounts of HA-FANCA for immunoprecipitation was confirmed by immunoblotting of the lysate inputs. The P-S347 immunoprecipitates were analyzed by immunoblotting with an anti-FANCA antibody to detect endogenous S347-phosphorylated FANCA.

Establishment of a FANCA-expressing stable cell line
U2OS cells expressing the HA-FANCA S347A mutant were established by transfecting U2OS cells with pcDNA3-HA-FANCA-WT or -S347A and selecting stable transfectants with 1 mg/mL G418 sulfate (Invitrogen). We confirmed HA-FANCA expression by immunoblotting. Positive clones were propagated in culture medium supplemented with G418 sulfate.

Statistical analysis
The data are presented as the mean values ± SEM. Comparisons between two groups were performed using the Student's t-test for unpaired data.