To examine the regulation of FTO expression by LepRb-STAT3 signalling pathway, we used immortalized HuH7 cells, as an in vitro model of hepatocytes. As they poorly respond to leptin treatment (Figure 1A), we transfected them with an expression vector coding for LepRb, and measured leptin-induced STAT3 phosphorylation to validate leptin action in LepRb-transfected HuH7 cells. Acute leptin treatment (5 and 15 minutes) did not significantly modify the Y705 phosphorylation of STAT3 in LepRb-transfected cells (Additional file 1: Figure S1A). However, LepRb overexpression induced STAT3 tyrosine phosphorylation (Y705) upon longer leptin stimulation (Figures 1A and B). Interestingly, LepRb overexpression per se induced FTO protein expression (Figures 1A and C), however 3-hour leptin treatment no further increased FTO expression in LepRb expressing HuH7 cells (Figures 1A and C). Since IL-6, like leptin, can activate STAT3 pathways 22 , we further investigated whether IL-6 could also affect FTO expression in control HuH7 cells. Acute IL-6 treatment (10 ng/ml) had significant but weak effect on Y705 phosphorylation of STAT3 (Additonal file 1: Figure S1B). Indeed, IL-6 treatment induced Y705 phosphorylation of STAT3 in a time-dependent manner (Figure 2A) concomitantly inducing FTO protein expression with a maximal effect at 6 hours of treatment (Figure 2B, +70%, p < 0.05), suggesting that FTO could be a target gene of STAT3 in hepatocytes. Silencing of STAT3 expression using a specific siRNA (−50% of STAT3 mRNA levels, p < 0.001) blocked leptin-induced induction of FTO mRNA levels (leptin effect on FTO/TBP mRNA: -9% vs +51% in STAT3-silenced HuH7 cells or in control cells, respectively, p < 0.05), confirming that FTO is a target gene of STAT3.

To investigate the effects of FTO on the LepRb-STAT3 signalling pathway, we then co-expressed LepRb and FTO in HuH7 cells, and investigated leptin actions on STAT3 phosphorylations. The efficiency of the co-transfections was confirmed by the increased FTO protein levels (Figure 3A), as well as by the ability of leptin to phosphorylate STAT3 on Y705 (Figure 3B). Leptin significantly increased FTO protein expression in control co-transfected cells, but not in cells overexpressing FTO (Figure 3A). Importantly, FTO overexpression reduced Y705 STAT3 phosphorylation by leptin by 1.9 fold, compared to leptin-stimulated control co-transfected cells (Figure 3B). In addition, leptin response on S727 STAT3 phosphorylation was also affected by FTO overexpression, since leptin could not decrease S727 STAT3 phosphorylation in presence of FTO, as in control conditions (Figure 3C). These data indicate thus that FTO interacts with leptin-induced STAT3 phosphorylation both on tyrosine 705 and serine 727 residues in HuH7 cells. Another pathway regulated by LepRb is the phosphorylation of PKB. Then, we investigated whether FTO impact also on leptin-induced PKB phosphorylation (S473) in HuH7 cells. As shown on Figure 3D, leptin induced PKB phosphorylation in LepRb-transfected cells (3 fold, p < 0,005), and FTO overexpression blocked this regulation.

We further investigated the consequences of FTO overexpression on leptin-regulated downstream steps of STAT3. STAT3 tyrosine phosphorylation is required for dimer formation, nuclear translocation, and thus for the DNA binding activity of this transcriptional regulator 23 . Among STAT3-regulated genes, G6P gene is repressed by STAT3 in HuH-7 cells 19 . Thus, we measured FTO overexpression effect on G6P expression. As expected, one hour of leptin treatment repressed the transcription of G6P gene in control condition (Figure 4A). However, G6P mRNA levels were significantly increased in basal conditions of FTO overexpressing HuH7 cells and leptin-induced repression of G6P transcription was totally abolished (Figure 4A). On the other hand, S727 STAT3 phosphorylation has been shown to favor translocation of STAT3 to the mitochondria and to regulate mitochondrial activity 10 . Therefore, we investigated the consequence of FTO-mediated alteration of S727 STAT3 phosphorylation on cellular mitochondrial density in co-transfected HuH7 cells. As shown in Figure 4B, FTO expression significantly increased the mitochondrial DNA/nuclear DNA ratio in co-transfected HuH7 cells (p < 0.05). These finding suggest that FTO expression in liver cells could alter leptin-induced Y705 and S727 STAT3 phosphorylations impacting on metabolic gene expression and mitochondrial density.

To determine the physiological relevance of these observations in vivo, we investigated the impact of liver-targeted FTO expression in mice. We infected male C57BL/6 mice by retroorbital injections of recombinant adenovirus coding for FTO or GFP (as control), and followed STAT3 phosphorylation in both nuclear and mitochondrial liver fractions. As shown in Additional file 1: Figure 2, under this experimental setting, FTO was specifically overexpressed in liver, since no change of FTO mRNA levels was observed in epididymal adipose tissue and hypothalamus of Ad-FTO infected mice compared to Ad-GFP mice. In agreement with in vitro data, we found a decreased content of nuclear STAT3 phosphorylated on Y705 (normalized by the nuclear protein SET7) in nuclear fractions of FTO-infected mice compared to GFP-infected mice (Figure 5A, -94%, p < 0.01). Moreover, we found a significant increase of S727 STAT3 phosphorylation (normalized by VDAC1 protein) in mitochondrial fractions of FTO-infected mice compared to GFP-infected mice (Figure 5B, +34%, p < 0,01).

Then we measured the effects of FTO overexpression on downstream events of activated STAT3 in mice liver. In agreement with a decreased content of Y705 STAT3 phosphorylation upon FTO overexpression, we found a decrease of SOCS3 mRNA levels and an increase of G6P expression in liver of Ad-FTO mice (Figure 5C). Furthermore, the mRNA levels of the transcription factor FOXO1α was significantly increased whereas only a tendency was observed for PEPCK and PGC1α, following FTO overexpression (Figure 5C), suggesting that FTO may participate in the control of neoglucogenic genes expression by interacting with STAT3 in liver. We also found that the mtDNA/nuclear DNA ratio was markedly increased (2.5 fold, p < 0.05) in liver of Ad-FTO mice (Figure 6A) in agreement with FTO-induced increase of S727 STAT3 phosphorylation. Furthermore, POLG1 and POLG2 and TFAM, key genes implicated in mitochondrial replication were also induced in liver of Ad-FTO mice (Figure 6B). Moreover, the mRNA levels of the mitochondria-encoded gene COX3 were increased in liver of Ad-FTO mice (Figure 6B). We then measured oxygen consumption by isolated mitochondria from liver of infected mice. As shown in Figure 6C, FTO overexpression significantly increased oxygen consumption stimulated by either octanoyl- or palmitoyl-CoA (+18,2% and +15,5%, respectively, p < 0,05), whereas the respiration under glutamate/malate or succinate substrates were not significantly modified. Altogether, these results confirmed in vivo that FTO may regulate both hepatic neoglucogenenic gene expression and oxidative metabolism by interacting with the STAT3 signaling pathways.

We found that Ad-FTO mice have increased circulating leptin concentration in the fasting state when compared to Ad-GFP mice (Table 1), suggesting compensatory mechanism against a state of leptino-resistance. In agreement, the mRNA levels of LepR and SOCS3, but not STAT3, were decreased in liver overexpressing FTO (Figure 5B). To verify whether leptin action was reduced in vivo, we investigated the effect of FTO on leptin-induced PKB phosphorylation. For that, we infused leptin to fasted infected C57BL/6 mice and measured the repercussion on the phosphorylation of PKB 30 minutes after treatment. As shown on Figure 7A, leptin induced a 3-fold induction of S473 phosphorylation of PKB in Ad-GFP mice and this effect was lost following FTO overexpression, in agreement with in vitro data. Although this was due in part to an increase of basal PKB phosphorylation (Figure 7A), this result indicated that FTO overexpression was able to prevent leptin action in vivo in mice liver.

FTO was overexpressed in liver of mice by adenoviral infection using a recombinant adenovirus encoding human FTO or GFP (as control) proteins for 10 days (2.10⁸ifu/g of body weight). Blood was removed by reorbital punction on isofluran-anesthesized mice, after an overnight fasting. Data are means ± SEM (n = 8/group).

We then verified whether FTO-mediated disruption of leptin-STAT3 signalling could alter glucose homeostasis in mice. We found that both fasting glycemia and insulinemia were significantly higher in Ad-FTO mice compared to Ad-GFP mice (Table 1). Moreover, glucose tolerance test revealed that Ad-FTO mice were glucose intolerant (Figure 7B). However, despite hyperglycemia all along the test, the response of glycemia to insulin injection (insulin tolerance test) was similar between both mice (Figure 7C), suggesting that FTO overexpression in the liver did not impair peripheral insulin sensitivity.

Animal studies were performed in accordance with the French guidelines for the care and used of animals and were approved by our regional ethic committee. Eight week-old male C57BL/6 mice (Harlan) were housed in controlled environment. Recombinant adenoviruses encoding for FTO or GFP proteins (Ad-FTO and Ad-GFP, respectively) were injected retroorbitally in mice (2.10⁸ infection forming units (ifu)/g of body weight), in order to over-express proteins specifically in liver. Glucose and insulin tolerance test were done respectively on day 7 and 9 after adenoviral infection. During these tests, glycemia were followed every 15 minutes after intraperitoneally injection of glucose (2 mg/g of body weight) or insulin (0.75 mUI/g of body weight). On day 10, blood was collected from animals by retroorbital punction under isofluran anesthesia, and mice were killed by cervical dislocation. Blood glucose levels were measured using a glucometer (Roche Diagnostics). Serum levels of insulin and leptin were determined using murine ELISA kits (ALPCO).

Rat primary hepatocytes were isolated in the presence of collagenase according to the method of Berry and Friend 40 , modified by Groen et al. 41 .

The cDNA sequence encoding full-length human FTO was generated as previously described 3 . Recombinant adenoviral genome encoding human FTO was generated by homologous recombination and amplified as described previously 3 . The plasmid encoding an HA-tagged murine LepRb was generated as previously described 42 .

HuH7 cells were grown in Dulbecco modified Eagle’s medium (PAA) supplemented with 10% fetal bovine serum. Cells were transfected with 1 μg expression plasmids for the FTO gene or for the LepRbgene (pcDNA3-FTO and pCIneo-LepRb, respectively), using EXGEN 500 transfecting reagent (Euromedex). In cotransfection experiments, cells received at the same time 1 μg of both vectors. An empty vector was used as control in each experiment. HuH7 cells were then used for treatments 48 hours post-transfection. Treatments included leptin (Labomics, 100 ng/ml) and IL-6 (Sigma, 10 ng/mL) incubations after a 16 hour serum depletion.

Total RNA from tissues or cell cultures were purified using the TRI Reagent Solution (Sigma). mRNA levels were measured by reverse-transcription followed by real time quantitative PCR using a Rotor-Gene 6000 (Corbett Research), as previously described 3 . Primers are listed in Additional file 2: Table S1. Values were normalized using HPRT or TBP, which were similar among conditions.

Tissues lysis and both separation and revelation of proteins were performed as described previously 3 . The primary antibodies used for protein detection are: STAT3 (Cell Signaling, #4904), Phospho-STAT3 (Tyr705) (Cell signaling, #9145), Phospho-STAT3 (Ser727) (Cell signaling, #9134), FTO (Abcam, ab65366), Actin (Sigma, A5060), SET7/9 (Santa Cruz, sc-56774), VDAC1/Porin (Abcam, ab14734).

Liver was homogenised in isolation buffer (Mannitol 210 mM, Saccharose 70 mM, Tris 50 mM, EDTA 10 mM and BSA 0.5%, pH = 7.4) using a teflon pestle, and centrifuged 10 minutes at 800 g. The pellet was kept for further nuclei isolation whereas the supernatant was centrifuged 10 minutes at 8000 g for mitochondria isolation, as previouly described 43 . The pellet of mitochondria was resuspended in isolation buffer, centrifuged a second time 10 minutes at 8000 g and resuspended in isolation buffer. For nuclei isolation, the pellet from the first centrifugation was resuspended in a hypertonic buffer (Hepes 10 mM, NaCl 0.42 M, MgCl2 1.5 mM, Glycerol 2.5%, EDTA 1 mM, EGTA 1 mM, DTT 1 mM, protease inhibitors cocktail 1X, pH = 7.4) and centrifuged for 30 min at 100 000 g in order to get nuclear extract in supernatant.

Mitochondrial respiration rates were measured at 30°C on freshly isolated liver mitochondria using a closed-thermostated oxygraph (Stratkelvin, UK). Different substrates were used: glutamate 5 mM + malate 2.5 mM as complex 1 substrates; succinate 5 mM + rotenone 5 μM as a complex 2 substrates with inhibition of complex 1 by rotenone; octanoyl-carnitine (110 μM) or palmitoyl-carnitine (55 μM) in presence of 1 mM malate, as β-oxydation substrates. State 3 was measured in the presence of respiratory substrates after the addition of 1 mM ADP and state 4 was measured after the addition of oligomycin (2 μg/mL).

The extraction of total DNA and the measurement of mitochondrial DNA (mtDNA) content by real time PCR was performed as previously described 43 .

All data are represented by means ± SEM. Statistical significance was determined using student unpaired t test. The threshold for significance was set at p < 0.05.