SIRT1 Mediates FOXA2 Breakdown by Deacetylation in a Nutrient-Dependent Manner

The Forkhead transcription factor FOXA2 plays a fundamental role in controlling metabolic homeostasis in the liver during fasting. The precise molecular regulation of FOXA2 in response to nutrients is not fully understood. Here, we studied whether FOXA2 could be controlled at a post-translational level by acetylation. By means of LC-MS/MS analyses, we identified five acetylated residues in FOXA2. Sirtuin family member SIRT1 was found to interact with and deacetylate FOXA2, the latter process being dependent on the NAD+-binding catalytic site of SIRT1. Deacetylation by SIRT1 reduced protein stability of FOXA2 by targeting it towards proteasomal degradation, and inhibited transcription from the FOXA2-driven G6pase and CPT1a promoters. While mutation of the five identified acetylated residues weakly affected protein acetylation and stability, mutation of at least seven additional lysine residues was required to abolish acetylation and reduce protein levels of FOXA2. The importance of acetylation of FOXA2 became apparent upon changes in nutrient levels. The interaction of FOXA2 and SIRT1 was strongly reduced upon nutrient withdrawal in cell culture, while enhanced Foxa2 acetylation levels were observed in murine liver in vivo after starvation for 36 hours. Collectively, this study demonstrates that SIRT1 controls the acetylation level of FOXA2 in a nutrient-dependent manner and in times of nutrient shortage the interaction between SIRT1 and FOXA2 is reduced. As a result, FOXA2 is protected from degradation by enhanced acetylation, hence enabling the FOXA2 transcriptional program to be executed to maintain metabolic homeostasis.


Introduction
The mammalian Forkhead transcription factor FOXA2 is a key regulator of hepatic energy metabolism in mammals [1]. Besides its role in regulating liver and other endoderm-derived organ specification during embryonic development [2], in post-natal life Foxa2 controls essential metabolic processes including glucose metabolism [3,4], bile acid homeostasis [5] and lipid oxidation [6,7]. Hepatocyte-specific Foxa2 conditional knockout mice showed impaired glucose homeostasis during the fasting response [3] and Foxa2 heterozygous mice displayed increased adiposity and impaired glucose uptake when challenged with a high fat diet [6,8]. In C. elegans and D. melanogaster, orthologues of FOXA are involved in gut development, illustrating the importance of these proteins for organ specification throughout metazoan evolution [9,10]. Interestingly, the nematode FOXA orthologue PHA-4 was found to control dietary restriction (DR)-mediated lifespan extension in C. elegans [11], suggesting a conserved role for FOXA transcription factors in energy homeostasis and metabolism, although the evidence for the involvement of mammalian FOXA2 in DR-mediated lifespan extension is still absent.
Metabolic processes controlled by Foxa2, such as glucose and lipid homeostasis, are often deregulated in metabolic syndromes such as diabetes and obesity [12,13]. Therefore, it is of great importance to investigate and understand how changes at a metabolic level functionally affect FOXA2-mediated transcriptional regulation of metabolism. Recent reports identified several post-translational modifications which are important in regulating Foxa2 metabolic function and cellular localization. Insulinactivated AKT2/PKB phosphorylates Foxa2 on a specific threonine (T156) adjacent to the DNA binding domain, resulting in nuclear exclusion and subsequent reduced transcriptional output by Foxa2 [13,14], although these findings remain controversial [3,15]. IKKa kinase, a downstream target of the pro-inflammatory cytokine TNFa, was also reported to suppress the FOXA2 transcriptional program by phosphorylating two serines (Ser107/111) adjacent to the DNA binding Forkhead domain [16].
Acetylation was recently proposed as a key post-translational mechanism by which transcription factors involved in metabolism are controlled in a nutrient-dependent manner [17,18]. Acetylation could therefore be a critical post-translational modification to regulate metabolic activity of FOXA transcription factor family members. Indeed, FOXA transcription factors were found to synergize with acetyltransferase p300 towards gene activation [19,20]. Moreover, FOXA1 was shown to be acetylated by p300 [21]. In contrast to acetyltransferases which add acetylation moieties to proteins, protein deacetylation is carried out by several deacetylases. These are grouped to either class I and II histone deacetylases (HDACs) or class III NAD + -dependent deacetylases, or sirtuins, which are activated upon changes in energy levels as manifested by an increase in the cellular NAD + /NADH ratio. Among the sirtuin family members, SIRT1 has the broadest range of substrates and affects the largest number of physiological pathways, including glucose and lipid metabolism [22,23]. Because of its NAD + dependency [24], SIRT1 connects the energy demands of a cell to its metabolic regulation via deacetylation of metabolic transcription factors and co-activators. Therefore, sirtuins may be involved in energy-dependent regulation of FOXA transcription factors by controlling their level of acetylation.
Since FOXA2 functions as a metabolic gauge, we hypothesized that FOXA2 could be regulated in a nutrient-dependent manner by means of acetylation. By using LC-MS/MS, we identified several acetylation sites on FOXA2 which are located adjacent to the DNA binding domain of the protein. Acetylation of FOXA2 was dependent on SIRT1 which was found to interact with and deacetylate FOXA2. Interestingly, the acetylation level of FOXA2 and the interaction of FOXA2 with SIRT1 were nutrientdependent. Moreover, acetylation protected FOXA2 from proteasomal degradation to sustain transcriptional activity of FOXA2 on its target genes. Taken together our data point to a model in which FOXA2 stability is regulated by SIRT1-mediated deacetylation in a nutrient-dependent manner towards the maintenance of whole-body metabolic homeostasis.

Cell culture and transfection
All cell lines were obtained from ATCC. Human embryonic kidney 293T cells (HEK293T cells) were maintained in DMEM containing 1 g/L glucose (Glutamax, Gibco) supplemented with 10% FCS and antibiotics. Human bone osteosarcoma epithelial cells (U2OS) and the human hepatoma HepG2 cell line were maintained in DMEM containing 4.5 g/L glucose (Glutamax, Gibco) supplemented with 10% FCS and antibiotics. In experiments in which cells were starved, cells were washed thrice with PBS prior to culture in glucose-free medium (Gibco) without FCS. All cell lines were cultured in a humidified atmosphere containing 5% CO 2 at 37uC. Transient transfection of HEK293T cells was performed with FuGene 6 (Roche, Basel, Switzerland) according to manufacturer's instructions, while HepG2 cells and U2OS cells were transiently transfected with polyethylenimine (PEI) (Polysciences) at a ratio of 1 mg DNA to 4 ml PEI. Transfection of siRNA was performed with Oligofectamine (Invitrogen) according to manufacturer's instructions.

Luciferase assay and siRNA
To assess FOXA2-mediated transcriptional activity on either the glucose-6-phosphatase (G6pase) or carnitine palmitoyltransferase 1a (CPT1a) promoter, pcDNA3-FLAG-FOXA2, pTK-Renilla and pGL3-G6pase-Luc or pGL3-CPT1a-Luc (and pRSV-MYC-SIRT1 when indicated) were transfected in HEK293T cells and after 48 hours samples were assayed for luciferase activity. To study the effect of knockdown of endogenous SIRT1 expression on FOXA2 transcriptional activity in HEK293T cells, two rounds of SIRT1 knockdown (8-hour interval) were performed with Oligofectamine (Invitrogen) according to manufacturer's instructions, followed by transfection of pcDNA3-FLAG-FOXA2, pGL3-G6pase-Luc and pTK-Renilla after 16 hours. Forty-eight hours after this final transfection, samples were assayed for luciferase activity. Knockdown of SIRT1 was performed using ON-TARGETplus SIRT1 siRNA SMARTpool, while ON-TARGETplus control siRNA SMARTpool (Thermo Scientific, Lafayette, CO, USA) was used as a siRNA transfection control. Luciferase assays were performed with the Dual-Luciferase Reporter Assay System according to manufacturer's instructions (Promega). pTK-Renilla was used for normalization of promoter-driven luciferase expression.

Immunoprecipitation and Western Blot analyses
To assess the acetylation status of FOXA2, wild type pcDNA3-FLAG-FOXA2 was ectopically expressed in HEK293T cells. After 48 hours, cells were lysed in RIPA buffer containing 1% Triton X-100, 0.5% sodium deoxycholate, 20 mM Tris-HCl pH8.0, 167 mM NaCl, 10 mM EDTA, 0.1% SDS and Complete protease inhibitors (Sigma). FLAG-FOXA2 was immunoprecipitated using FLAG-M2 agarose beads (Sigma) at 4uC for 2 hours. Beads were subsequently washed three times in RIPA buffer. To minimize co-elution of the heavy and light chains of the anti-FLAG antibodies, the beads were resuspended in non-reducing sample buffer (containing: 125 mM Tris-HCl pH6.8, 4% SDS, 20% glycerol and 0.004% bromophenol blue) and heated at 95uC for 5 minutes. Subsequently, the supernatant containing dissociated and immunoprecipitated FLAG-FOXA2 was transferred to a new vial and the proteins were reduced by adding DTT to a final concentration of 50 mM. Immunoprecipitated proteins were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) followed by Western Blot analyses. Protein acetylation was detected using an anti-acetyl-lysine-HRP antibody (ab23364) (Abcam). Subsequently, immunoprecipitated levels of FOXA2 were assessed by anti-FLAG antibodies (M2, F1804) (Sigma).

Quantitative RT-PCR
To assess the effect of NAM treatment on the levels of FOXA2 mRNA, HEK293T cells transfected with wild type pcDNA3-FLAG-FOXA2, were cultured for 16 hours with 20 mM NAM of vehicle control and relative FOXA2 mRNA levels were determined by quantitative RT-PCR. RNA was isolated from the cells using a Machery-Nagel NucleoSpin RNA II kit (Bioké, Leiden, The Netherlands) and quantified using a Nanodrop ND-1000 (Wilmington, DE, USA). cDNA was prepared from total RNA using an iScript cDNA Synthesis Kit from Bio-Rad (Bio-Rad Laboratories, Stanford, CA, USA) and gene expression assays were performed according to manufacturer's protocol (SYBR Select Master Mix; Life Technologies Bleiswijk, The Netherlands) to measure mRNA levels of FOXA2 as the gene of interest, and 18 S as the housekeeping gene using the following primers: FOXA2forward 59-TGGAGCAGCTACTATGCAG-39, FOXA2-reverse 59-CGTGTTCATGCCGTTCATC-39; 18 S-forward 59-AGTC-CCTGCCCTTTGTACACA-39 and 18 S-reverse 59-CGATCC-GAGGGCCTCACTA-39.

Mass spectrometry
Wild type FLAG-FOXA2 was ectopically expressed in HEK293T cells. When indicated, cells were cultured for 16 hours prior to cell lysis in 20 mM NAM and 5 mM TSA to inhibit endogenous deacetylase activity. Cells were lysed in RIPA buffer, immunoprecipitated with FLAG-M2 agarose beads and reduced in sample buffer, as described above. Subsequently, immunoprecipitated wild type FLAG-FOXA2 was run on NuPAGE Bis-Tris Mini Gels (Invitrogen, Carlsbad, USA) according to manufacturer's instructions. Bands were excised from the gel, reduced with DTT (0.5 mM; MP Biomedicals), alkylated with iodoacetamide (54 mM; Sigma), and digested with trypsin (Roche), chymotrypsin (Roche) or V8-DE (Roche) as described [27]. A second round of digestion was performed on the gel fragments with elastase (Sigma) to improve protein coverage by MS. Samples were subjected to nanoflow LC (Eksigent) using C18 reverse phase trap columns (Synergi 4 mm Hydro-RP 80 Å , Phenomenex; column dimensions 2 cm6100 mm, packed in-house) and subsequently separated on C18 analytical columns (ReproSil-Pur 120 C18-AQ, 5 mm, Dr. Maisch GmbH; column dimensions, 20 cm650 mm; packed inhouse) using a linear gradient from 0 to 40% buffer B (buffer A = 5% (v/v) acetonitrile, 0.1% formic acid (v/v); buffer B = 95% (v/v) acetonitrile, 0.1% formic acid (v/v)) in 60 min at a constant flow rate of 150 nl/min. Column eluate was directly coupled to a LTQ-Orbitrap-XL mass spectrometer (Thermo Scientific) operating in positive mode, using lock spray internal calibration. Data were processed and subjected to database searches using Proteome Discoverer software (Thermo Scientific) or Mascot software (Matrix Science) against non-redundant SwissProt and NCBI database, with a 10 ppm mass tolerance of precursor and 0.5 Da for the fragment ion.

Murine livers and endogenous Foxa2 immunoprecipitation
Wild type Balb/c mice were either fed ad libitum or were restricted from food intake by limiting the chow to 25% of ad libitum levels for 36 hours. Mice were sacrificed, and livers were extracted and stored at 280uC. Whole cell lysates of these livers were generated by homogenizing 50 mg of each liver in RIPA buffer with 0.5 mm Zirconium Oxide beads by Bullet Blender (Next Advance, USA). Immunoprecipitation of Foxa2 from the supernatant of these homogenates was performed by rabbit-anti-FOXA2 antibody (WRAB-1200) (Seven Hill Bioreagents), precoupled to protein G sepharose beads (GE healthcare) for 2 hours at 4uC. Beads were subsequently washed three times in RIPA buffer. Elution of the immunoprecipitated proteins and assessment of Foxa2 and its acetylation levels was similar as described above for immunoprecipitation of FLAG-FOXA2.
This study was carried out following ethical approval for studies on laboratory animals by the Dierexperimentencommissie (DEC) of the UMC Utrecht. All efforts were undertaken to minimize suffering of the animals.

Statistics
Protein expression levels were measured by densitometry using Image J (http://rsbweb.nih.gov/ij/). Protein levels were expressed relative to actin protein levels after subtraction of background intensity. Luciferase data are shown as mean +SEM from 3 independent experiments as analyzed by two-tailed Student's t-test with statistical significance P,0.05.

FOXA2 is acetylated at multiple residues
The FOXA2 transcription factor is an essential player in liver metabolism, although the precise regulation of this transcription factor is not completely understood. We hypothesized that posttranslational modifications (PTMs) such as acetylation could affect and differentially regulate the FOXA2 metabolic program. To investigate whether FOXA2 is an acetylated protein, Liquid Chromatography-tandem Mass Spectrometry (LC-MS/MS) analyses were performed. To this end, FLAG-FOXA2 was ectopically expressed in HEK293T cells, immunoprecipitated, digested by proteases and subjected to LC-MS/MS analysis. Five peptides were identified harboring 3 unambiguously detected acetylation sites: K264, K274 and K275 (Fig. 1A, B). In addition, cells were supplemented with histone deacetylase (HDAC) inhibitors nicotinamide (NAM) and Trichostatin A (TSA), which inhibit endogenous deacetylase activity. Two additional unambiguously identified acetylation sites were detected: K6 and K259 (Fig. 1B). These observations indicate that FOXA2 can be acetylated on at least 5 residues.
The LC-MS/MS analyses identified FOXA2 peptides covering 81% of the whole FOXA2 sequence (data not shown). Although it cannot be excluded that acetylation sites were missed, the MS analyses identified 15 out of all 19 lysine residues, with each lysine covered several times. Next, the position of the acetylated sites was mapped on the FOXA2 structure which revealed that they were localized adjacent to the Forkhead domain (Fig. 1C).

Deacetylation of FOXA2 is mediated by SIRT1
The LC-MS/MS experiments revealed acetylated residues in FOXA2 upon treatment with HDAC inhibitors (Fig. 1B), suggesting that FOXA2 is a substrate for HDAC-mediated deacetylation. To determine which family of HDACs is responsible for deacetylation of FOXA2, FLAG-FOXA2 was ectopically expressed in HEK293T cells which were incubated with TSA (inhibitor of class I and II HDACs) and/or NAM (inhibitor of class III NAD + -dependent deacetylases, or sirtuins). Using a panacetylation-specific antibody (anti-acetyl-K) we observed that FOXA2 was robustly acetylated in cells treated with NAM, but not in cells with TSA alone (Fig. 2A). These data suggest that FOXA2 is a sirtuin-dependent deacetylation substrate.
To investigate whether SIRT1 is capable of deacetylating FOXA2, FLAG-FOXA2 and MYC-SIRT1 were co-expressed in NAM-treated HEK293T cells. Similar as in Fig. 2A, FOXA2 was robustly acetylated in the presence of NAM (Fig. 2B). Ectopic expression of SIRT1 overcame the NAM-mediated inhibition of endogenous sirtuin activity and completely blunted acetylation of FOXA2. These data therefore suggest that FOXA2 is a substrate for SIRT1-mediated deacetylation. SIRT1 interacts with and co-localizes with FOXA2 in the nucleus Next, we assessed whether SIRT1-mediated deacetylation of FOXA2 occurs via a direct interaction of both proteins. Indeed, upon ectopic expression in HEK293T cells, MYC-SIRT1 coimmunoprecipitated with FLAG-FOXA2 suggesting that both proteins interact (Fig. 2C). To further examine the SIRT1-FOXA2 interaction, we performed a Proximity Ligation Assay [28], an immunofluorescence-based method that allows for intracellular detection of two proteins that are in close proximity of each other [29]. A PLA assay performed on the human hepatocellular carcinoma cell line HepG2, which endogenously expresses FOXA2 and SIRT1, showed interaction of these proteins (Fig. 2D). This interaction was strictly observed in the nucleus, which is in agreement with the reported nuclear activity of SIRT1 [30]. These findings were corroborated upon ectopic expression of MYC-SIRT1 and FLAG-FOXA2 in human osteosarcoma U2OS cells (which do not express FOXA2 endogenously), as we found a specific interaction of FLAG-FOXA2 and MYC-SIRT1 by using FLAG-and SIRT1-specific antibodies (data not shown). Collectively, these data show that SIRT1 and FOXA2 interact at an endogenous level exclusively in the nucleus.

Interaction of FOXA2 and SIRT1 is reduced upon nutrient starvation
Because SIRT1 functions as a NAD + -dependent energy sensor we wondered whether the interaction of SIRT1 with FOXA2 is Endogenous FOXA2 and SIRT1 interact in the nucleus. HepG2 cells were subjected to immunofluorescence-based proximity ligation assay [28]. Nuclei were counterstained with DAPI. doi:10.1371/journal.pone.0098438.g002 nutrient sensitive. To assess this, HEK293T cells, which ectopically expressed FLAG-FOXA2 and MYC-SIRT1, were subjected to nutrient stress for 16 hours. Strikingly, when compared to cells grown in standard medium, the interaction between FLAG-FOXA2 and MYC-SIRT1 was decreased upon starvation (deprivation of glucose and serum) (Fig. 3A). These findings were corroborated by observations from a PLA assay, which revealed that the interaction of endogenous FOXA2 and SIRT1 in HepG2 cells was reduced upon nutrient starvation (Fig. 3B). Collectively, these data indicate that the interaction of SIRT1 with FOXA2 is dependent on environmental nutrient levels.

SIRT1 inhibits FOXA2-driven transcription
FOXA2 is involved in regulating transcription of enzymes involved in glucose and lipid homeostasis, such as glucose-6phosphatase (G6pase) and carnitine palmitoyltransferase 1a (CPT1a), respectively [1,3]. To establish whether FOXA2-driven transcription is affected in a SIRT1-dependent manner, we performed a luciferase-driven transcription assay on G6pase and CPT1a promoter constructs. In the presence of SIRT1, FOXA2driven transcription is blunted on both promoters (Fig. 4A, 4B). Conversely, knockdown of endogenous SIRT1 in HEK293T cells by siRNA revealed enhanced FOXA2-mediated transcription on the G6pase promoter (Fig. 4C). Moreover, a progressive decrease in FOXA2-driven transcription on the G6pase promoter was observed which was inversely proportional to the amount of co-transfected MYC-SIRT1 (Fig. 4D), indicating that SIRT1 affects the FOXA2 transcriptional program in a dose-dependent manner. Collectively, these data demonstrate that SIRT1 can inhibit FOXA2-driven transcription. Remarkably, expression analysis revealed that FLAG-FOXA2 protein levels decreased concomitantly with increasing amounts of co-transfected MYC-SIRT1 (Fig. 4D) suggesting SIRT1 negatively affects FOXA2-driven transcription by reducing FOXA2 protein levels, thereby modulating its transcriptional output.

SIRT1-mediated deacetylation reduces FOXA2 protein stability by targeting it towards proteasomal degradation
To further investigate whether endogenous SIRT1 is able to affect FOXA2 protein levels, FLAG-FOXA2 was ectopically expressed in HEK293T cells. Inhibition of endogenous SIRT1 activity by NAM resulted in enhanced protein levels of FOXA2 whereas FOXA2 mRNA levels remained unaffected (Fig. S1), suggesting that sirtuin-controlled FOXA2 acetylation may be important for FOXA2 protein stability (Fig. 5A).
The deacetylation activity of SIRT1 is mediated by a critical histidine residue (H363) located in its NAD + -binding catalytic site [31]. To investigate whether the acetylation state and concomitant decrease in FOXA2 protein levels is dependent on the catalytic activity of SIRT1, either wild type SIRT1 or the SIRT1-H363Y catalytically-impaired mutant were co-expressed with FLAG-FOXA2 in HEK293T cells. FOXA2 was found to interact with both SIRT1 and SIRT1-H363Y (Fig. 5B), indicating that the interaction is independent of SIRT1 activity. However, in contrast to wild type SIRT1, expression of SIRT1-H363Y reduced FOXA2 protein levels less severely, as FOXA2 expression was 67% (Fig. 5B) and 142% (Fig. 5C) higher upon co-transfection with SIRT1-H363Y than with wild type SIRT1). In addition, acetylation of FOXA2 was predominantly retained upon SIRT1-H363Y co-expression (Fig. 5C). Hence, the NAD + -dependent catalytic activity of SIRT1 appears to be required for the stability of the FOXA2 protein.
To identify which acetylated residue(s) are critical for stabilization of FOXA2, we iteratively mutated all the five identified lysines (K) to arginine (R) residues. These mutant FOXA2 proteins were ectopically expressed in HEK293T cells and endogenous SIRT1 activity was inhibited with NAM. When the 5 acetylation sites identified by our LC-MS/MS analysis (Fig. 1B) were mutated it was noted that protein expression levels were slightly reduced as compared to wild type FOXA2, however a significant level of acetylation was still observed (data not shown). This suggests that other acetylated lysines must be present in FOXA2 that are controlled in a sirtuin-dependent manner, and which were not uncovered by our LC-MS/MS analysis. To investigate this hypothesis, a FLAG-12K-R-FOXA2 mutant was generated, in which 12 out of 19 FOXA2 lysine residues were changed to arginine. K365 and K399 were additionally mutated since these residues were not covered in our LC-MS/MS analyses, K149 was found to be acetylated but with a non-significant peptide score (data not shown), while K226, K229, K253 and K256, although not identified as acetylated by the LC-MS/MS approach, could be potential targets of histone acetyltransferase p300 since they make up the -K-X-X-K-consensus motif of p300 [32]. Ectopic expression in HEK293T cells showed a marked reduction of FLAG-12K-R-FOXA2 protein levels compared to wild type FOXA2 (Fig. 5D). In contrast to 12K-R-FOXA2 protein levels, wild type FOXA2 expression was stabilized by NAM and the acetylation level of FOXA2 was enhanced. Although 12K-R-FOXA2 showed a low basal level of acetylation, little increase was observed when cells were treated with NAM (Fig. 5D). Collectively, these data indicate that these 12 lysines are critical for the SIRT1-mediated regulation of FOXA2 acetylation levels.
Next, we investigated whether the SIRT1-mediated decrease in FOXA2 protein levels is the result of proteasomal degradation of FOXA2. Therefore, FLAG-FOXA2 was transfected in HEK293T cells and treated with the proteasome inhibitor MG132 for 3 and 6 hours. Similar to treatment with NAM, FOXA2 protein levels were increased upon MG132 treatment, suggesting that FOXA2 is subject to degradation in the proteasome (Fig. 5E). Moreover, in HepG2 cells, the observed reduction in endogenous FOXA2 protein levels resulting from MYC-SIRT1 expression was partly abolished upon MG132 treatment (Fig. 5F), suggesting that SIRT1-mediated deacetylation of FOXA2 leads, at least in part, to FOXA2 degradation in the proteasome. Subsequently, we blocked protein translation in HEK293T cells with cycloheximide and found that protein levels of ectopically-expressed wild type FOXA2 decreased substantially within 8 hours after cycloheximide addition. In contrast, prior treatment with NAM abrogated the decrease in FOXA2 protein levels induced by cycloheximide (Fig. 5G). Collectively, these data show that acetylation of FOXA2 protects the protein from degradation.
As FOXA2 is degraded at least in part by the proteasome, we wondered whether the protein is prone to ubiquitination. Therefore, we immunoprecipitated ectopically expressed wildtype or 12K-R FLAG-FOXA2 from HEK293T cells, blocked proteasomal degradation, and assessed the level of FOXA2 ubiquitylation. We found that the 12K-R mutant showed an increased level of poly-ubiquitination than wild-type FOXA2 (Fig. 5H). Of note, we observed that FOXA2 was also monoubiquitinated at three sites ( Fig. 5H; arrows), and that the level of mono-ubiquitination was increased in the 12K-R mutant. These data at least suggest that the higher level of ubiquitination of the 12K-R deacetylation-mimicking mutant results in reduced protein stability by targeting the protein towards degradation in the proteasome.

Nutrient deprivation enhances acetylation of Foxa2 in murine livers
FOXA2 is predominantly expressed in metabolically active tissues such as liver and pancreas. To test if the nutrient-dependent acetylation of FOXA2 also occurs in vivo, we assessed Foxa2 acetylation levels in livers from mice fed with either a isocaloric diet (ad libitum fed, AL) or which had a diet consisting of 25% of ad libitum levels (starved, ST) for 36 hours. Endogenously expressed Foxa2 was immunoprecipitated from liver homogenates and acetylation levels were assessed on Western Blot with a panacetylation-specific antibody. Acetylation levels of Foxa2 in livers from ST mice were increased as compared to those in livers of AL fed mice (Fig. 5I), revealing that Foxa2 is regulated by acetylation in a nutrient-dependent manner in vivo.

Discussion
Mammalian Foxa2 regulates glucose and lipid homeostasis in the liver [3][4][5][6][7] in response to fasting, but how this transcription  factor is controlled and how it relays signals in response to extracellular cues, is not completely understood. The present study provides evidence for nutrient-dependent post-translational regulation of FOXA2 via acetylation which is negatively regulated by the sirtuin deacetylase family member SIRT1. By means of LC-MS/MS analyses, we identified 5 acetylated lysine residues on FOXA2 and found that SIRT1 interacts with and deacetylates FOXA2 in a nutrient-dependent manner. Importantly, upon deacetylation by SIRT1, FOXA2 becomes susceptible to proteasomal degradation by enhanced mono-and poly-ubiquitination, hence affecting the FOXA2 transcriptional program due to reduced FOXA2 protein levels. We confirmed these observations in vivo by showing that acetylation levels of Foxa2 in mouse liver were increased upon nutrient starvation, likely resulting from reduced interaction between Foxa2 and Sirt1. Therefore, we propose that acetylation of FOXA2 serves as a rapid mechanism to prevent degradation of FOXA2, when it is needed to regulate metabolic homeostasis in times of nutrient shortage. In addition, these data suggest that FOXA2-mediated transcriptional regulation of metabolism is directly controlled by SIRT1 upon changes in intracellular energy levels.
In our LC-MS/MS analyses, 3 of the 5 identified acetylated lysine residues were found to be specifically present in FOXA2, whereas the conserved K6 and K259 were also found acetylated in  FOXA1. For example, lysine K6 is conserved among all FOXA members and indeed we identified this residue acetylated on FOXA1 as well (data not shown). Interestingly, Belagouli et al. [33] reported that stability of FOXA2 is regulated by sumoylation on residue K6. The metabolic cofactor CRTC2 has recently been shown to be deacetylated by SIRT1 and then subjected to sumoylation-driven breakdown [34]. It is therefore attractive to speculate that competition between these modifications on K6 may regulate protein levels of FOXA2 in response to SIRT1. However, K6 itself seemed not essential to regulate SIRT1mediated FOXA2 stability at least in kidney cells, as mutation of this residue to arginine (K to R) showed relatively similar FOXA2 protein and acetylation levels as the wild type protein (data not shown). It is important to note that mutation of K6 also renders the protein unsusceptible to sumoylation and therefore further research is needed to determine whether any interplay between sumoylation and acetylation on FOXA2 K6 occurs in other cell types.
We found that mutation of at least twelve lysines was required to reduce the overall acetylation signal on FOXA2. Of note, some of the mutated lysines may only represent putative acetylation sites for example due to low stoichiometry and as a result these may not be identified by LC-MS/MS. Some lysines in FOXA2 were not covered by LC-MS/MS despite various digestion strategies, and therefore further confirmation is needed to dissect whether other than the identified residues in FOXA2 are acetylated. Nevertheless, the acetylation sites that were uncovered by our LC-MS/MS analyses play an important role in regulating FOXA2 protein stability.
In our study, FOXA2 was found to be both mono-and polyubiquitinated. Poly-ubiquitination is primarily regarded as a signal that enables transport of the protein to the proteasome for degradation. This notion is in line with our observation that breakdown of FOXA2, mediated by SIRT1 is MG132 sensitive and suggests that at least part of the FOXA2 degradation is mediated by the proteasome. Mono-ubiquitination, however, can impact proteins in various ways and has been described to affect cellular events including DNA repair, gene expression, endocytosis and nuclear export [35]. However, it has also been demonstrated that mono-ubiquitination can also provide a signal for lysosomal degradation of proteins [36]. Although further study is required to determine the exact role for the three mono-ubiquitinated sites in FOXA2, the increase in the level of mono-ubiquitination in the 12K-R mutant may imply their involvement in breakdown of FOXA2.
In agreement with our data, a recently published study by Von Meyenn et al. [37] elegantly showed K259 and K275 as acetylated residues in Foxa2. Acetylation of K259 was induced by p300 by glucagon-mediated signaling during fasting, resulting in expression of genes involved in b-oxidation and ketone body formation. Von Meyenn et al. also showed that SIRT1 is able to deacetylate Foxa2. The work reported here corroborates these findings and goes on to add a novel dimension to these observations. Rather than observing effects of acetylation on Foxa2 shuttling, we observe that SIRT1 affects FOXA2 transcriptional output in a nutrientdependent manner by means of regulating FOXA2 protein stability. The nuclear shuttling of Foxa2 in an insulin-dependent manner reported by the Stoffel group [3,15] remains controversial as we (data not shown) and others have not found any evidence for this, which has been discussed elsewhere in detail [6,13]. However, our observation that SIRT1 mediates stability of FOXA2 protein levels does not exclude the nuclear shuttling model, but in contrast represents another layer of regulation of FOXA2 by means of acetylation to execute the appropriate metabolic program during nutrient limitation. Taken together, the maintenance of FOXA2 expression and function by acetylation seems dually regulated during nutrient deprivation. On the one hand, glucagon signaling enhances p300 acetyltransferase activity to induce acetylation of FOXA2. On the other hand, the reduced interaction of SIRT1 and FOXA2 enhances acetylation of FOXA2 by the loss of SIRT1-mediate deacetylation of the protein. It remains unclear how the FOXA2-SIRT1 interaction is reduced upon nutrient deprivation. Possibly alterations in other post-translational modifications, e.g. phosphorylation of either FOXA2 or SIRT1, may reduce the interaction between both proteins. Alternatively, enhanced activation of p300 (increased phosphorylation of Ser89) upon glucagon signaling may compete with SIRT1 for access to the acetylation sites on FOXA2 [34]. Further research on the FOXA2-SIRT1 interaction is required to fully understand this nutrient dependent regulation of FOXA2.
The work reported here also shows additional differences compared to the work presented by Von Meyenn et al. Firstly, Von Meyenn et al. studied murine Foxa2 and showed that mutation of K259 and K275 residues to arginine completely blunted overall Foxa2 acetylation in Hepa1-6 cells. Secondly, pharmacological inhibition by NAM alone was not sufficient to increase Foxa2 K259 acetylation levels, and inhibition of HDACs was additionally required. In contrast, we report here the requirement of at least 12 mutated lysines for reduction of the acetylation signal of FOXA2, and a robust induction of acetylation by NAM treatment alone. Moreover, we identified three additional acetylation sites in FOXA2 when ectopically expressed in HEK293T cells. These differences could be related to species differences of the cell types and constructs that were used. This notion is supported by von Meyenn et al. as they show that only acetylation of one of the two identified acetylation sites (K259) in murine Foxa2 was found in human HepG2 cells whereas both K259 and K275 were found in mouse Hepa1-6 cells [37]. Moreover, the level of coverage likely contributes as 81% of FOXA2 was covered by LC-MS/MS in our study as compared to 37% and 57% of Foxa2 in the study of Von Meyenn et al. Indeed, we identified acetylation on K6, a residue that was not covered by Von Meyenn et al.
In conclusion, we propose a model in which FOXA2 protein levels are regulated by SIRT1 via acetylation-mediated proteasomal degradation (Fig. 6). In the presence of nutrients, FOXA2 is dispensable for metabolic homeostasis and therefore the balance is towards a deacetylated state, mediated by SIRT1, resulting in enhanced degradation of FOXA2 in the proteasome. Conversely, in times of nutrient stress, FOXA2 needs to be switched on to drive gluconeogenesis, b-oxidation and ketone body formation. FOXA2 is then released from the inhibitory interaction with SIRT1 by an, as of yet, unknown mechanism, thereby reducing FOXA2 deacetylation. In addition, glucagon signaling enhances p300 activity in a cAMP-PKA dependent manner to drive Foxa2 acetylation [34,37]. Together with previous published work, our data provides better understanding of the molecular control of metabolism by SIRT1 and FOXA2, and therefore may contribute to the development of intervention strategies for metabolic derangements like diabetes or obesity. Figure S1 NAM does not affect mRNA levels of FOXA2. HEK293T cells were transfected with FLAG-FOXA2 or empty vector. Cells were cultured with or without 20 mM NAM for 16 hours, after which cells were lysed, RNA was isolated and mRNA levels of FOXA2 were determined by quantitative RT-PCR. 18 S mRNA levels were used as a reference. Results depict mean+SEM from three independent experiments. (TIF)