Knockdown of the Fat Mass and Obesity Gene Disrupts Cellular Energy Balance in a Cell-Type Specific Manner

Recent studies suggest that FTO variants strongly correlate with obesity and mainly influence energy intake with little effect on the basal metabolic rate. We suggest that FTO influences eating behavior by modulating intracellular energy levels and downstream signaling mechanisms which control energy intake and metabolism. Since FTO plays a particularly important role in adipocytes and in hypothalamic neurons, SH-SY5Y neuronal cells and 3T3-L1 adipocytes were used to understand how siRNA mediated knockdown of FTO expression alters cellular energy homeostasis. Cellular energy status was evaluated by measuring ATP levels using a luminescence assay and uptake of fluorescent glucose. FTO siRNA in SH-SY5Y cells mediated mRNA knockdown (−82%), increased ATP concentrations by up to 46% (P = 0.013) compared to controls, and decreased phosphorylation of AMPk and Akt in SH-SY5Y by −52% and −46% respectively as seen by immunoblotting. In contrast, FTO siRNA in 3T3-L1 cells decreased ATP concentration by −93% (p<0.0005), and increased AMPk and Akt phosphorylation by 204% and 70%, respectively suggesting that FTO mediates control of energy levels in a cell-type specific manner. Furthermore, glucose uptake was decreased in both SH-SY5Y (−51% p = 0.015) and 3T3-L1 cells (−30%, p = 0.0002). We also show that FTO knockdown decreases NPY mRNA expression in SH-SY5Y cells (−21%) through upregulation of pSTAT3 (118%). These results provide important evidence that FTO-variant linked obesity may be associated with altered metabolic functions through activation of downstream metabolic mediators including AMPk.


Introduction
The Fat Mass and Obesity Associated gene (FTO) associates with body mass index (BMI), waist circumference, type II diabetes, and other obesity related traits [1] in individuals that are homozygous for the FTO risk allele [2], 16% in most worldwide populations, establishing itself as the focus of intense research. Substantial evidence suggests that the primary physiological effect of variations in the FTO locus in humans is increased dietary intake without significant effects on physical activity or metabolic rates [3,4]. Murine studies of FTO function demonstrate FTO's strong influence over energy intake but are complicated by metabolic disturbances including alterations in basal metabolic rate and overall energy expenditure not apparent in human studies [4,5,6,7]. Consequently, current studies have not determined a potential mechanism which results in increased energy intake associated with FTO gene variants in humans.
Tissue-specific gene expression profiling suggests that FTO is highly expressed in the hypothalamus [8], an area of the brain strongly linked to eating behavior. FTO is also expressed at a lower level and maintains functionality in adipose and other tissues [9,10]. Further studies suggest that FTO functions as both a sensor of energy status [9,11] and a modulator of metabolic processes [12], and its cellular function may be cell-type specific, since fasting increases FTO expression in white adipose tissue [9] and hypothalamic neurons [11] but decreases FTO expression in brown adipose tissue [9]. Intracellular ATP levels affect cellular function and are tightly controlled through a variety of sensors and signaling pathways [13]. Additionally, intracellular energy status regulated by AMP Kinase plays an important role in controlling organismal energy intake and energy metabolism by hypothalamic neurons [14]. AMPk (59-Adenosine Monophosphate Activated Protein Kinase) appears to be the primary sensor of intracellular ATP levels in the hypothalamus, as it is phosphorylated in the presence of increased AMP/ATP ratio by the tyrosine kinase LKB1 [15], and is activated by starvation and inhibited by leptin [14]. Further, hypothalamic AMPk controls neuropeptide expression, thereby controlling eating behavior [16], and may control adipocyte function as well [13].
Our results suggest that downregulation of FTO by siRNA in both neuronal and adipocyte cells in culture causes a cell-type specific alteration in cellular ATP levels with a corresponding alteration in AMPk phosphorylation. We thus hypothesize that if FTO's metabolic functions control AMPk in the hypothalamus, then AMPk may be the primary driver of the systemic metabolic effects seen in human and murine studies of FTO function.

Validation of siRNA-induced Knockdown of FTO Gene Expression
To confirm appropriate downregulation of FTO in both naïve and differentiated cells, qPCR was performed at several time points after transfection, and one week after induction of differentiation. Naïve (undifferentiated) SH-SY5Y cells treated with siRNA against FTO exhibited a decrease in FTO mRNA expression by 281.8%, and 269.8% at 48 and 72 hours, respectively ( Fig 1A). 3T3-L1 cells exhibited 280.3% and 265.3% decrease in FTO mRNA at 48 and 72 hours after transfection, respectively ( Fig 1B). Additionally, sustained knockdown of FTO mRNA was validated after 1 week of differentiation in both SH-SY5Y (259.3%) and 3T3-L1 cells (259.6%) (Fig 1A  and 1B).

FTO Knockdown Disrupts Cellular Energy Balance in a Cell-type Specific Manner
The FTO gene is known to play an important role in controlling energy balance as previous reports show a strong correlation between FTO expression and organismal energy status, including fat accumulation and energy intake [5]. However, little is known about the relationship between FTO expression and cellular energy balance, and substantial research has shown that alterations in neuronal cellular energy status can have large implications on eating behavior [14]. Thus, to determine if FTO plays a role in maintaining or controlling cellular energy status, we examined the effect of FTO knockdown on ATP levels in both SH-SY5Y neuronal cells and 3T3-L1 adipocytes. To fully evaluate ATP levels after siRNA transfection, naïve or differentiated cells were incubated for 30 minutes in glucose-free media, to inhibit glycolysis, with or without oligoymycin A, a specific inhibitor of ATP synthase, to completely inhibit the production of new ATP.

Cellular Proliferation is Unaffected by Exposure to FTO siRNA
To study the effect of FTO siRNA treatment on cellular proliferation in SH-SY5Y and 3T3-L1 cells, we plated and treated cells as described below and subsequently trypsinized and counted viable cells. To confirm that changes in ATP levels in response to FTO siRNA are not due to differences in cell survival after oligomycin treatment, FTO siRNA and Control siRNA treated SH-SY5Y and 3T3-L1 cells were exposed to oligomycin for 30 minutes, collected and counted. Exposure to oligomycin does not alter cell survival in

Effect of siRNA Targeting the FTO Gene on Cellular Signaling and Glucose Uptake
After determining the effect of FTO knockdown on cellular ATP levels, we wanted to examine additional metabolic traits modulated by FTO expression. To determine if FTO downregulation affects glucose uptake in SH-SY5Y neuronal cells and 3T3-L1 adipocytes, cells were treated with FTO siRNA and glucose uptake was determined by measuring uptake of fluorescent 2-NBDG. As shown in Figure 2A, glucose uptake into both naïve and differentiated SH-SY5Y cells was significantly decreased by 227% (FTO siRNA 9804067898 (Fluorescence units (FU) vs Control siRNA 13422064648 FU; p = 0.007; N = 5) and 251% (FTO siRNA 6595267281 FU vs Control siRNA 135202617647 FU; p = 0.015; N = 5), respectively. Additionally, glucose uptake into naïve and differentiated 3T3-L1 adipocytes was decreased by 222% (siRNA 3561662501 FU vs Control siRNA 4574563000 FU; p = 0.017; N = 5) and 230% (siRNA 12276463634 FU vs Control siRNA 17458163643 FU; p = 0.0002; N = 5), respectively ( Fig 3A).
To further explore the metabolic consequences of decreased FTO expression, we immunoblotted for total and phosphorylated AMPK (Thr 172 ) (pAMPk) in FTO deficient SH-SY5Y and 3T3-L1 cells. AMPk is an important sensor of cellular energy status [14] that is phosphorylated in the presence of an elevated AMP/ATP ratio by the serine/threonine kinase, LKB1 [17]. AMPk controls multiple cellular metabolic processes including glucose uptake and fatty acid synthesis and may play a role in hypothalamic energy sensing and modulation of eating behavior [14]. As seen in Figure 2B, FTO knockdown decreases AMPk phosphorylation by 251.8% and 252.3% in both naïve and differentiated SH-SY5Y cells, respectively, without a significant alteration in total AMPk levels. However, FTO-deficient 3T3-L1 adipocytes exhibit a significant increase (204% naïve, 38% differentiated) in pAMPK (Thr 172 ) without a change in total AMPk ( Fig 3B). Since FTO-siRNA does not significantly affect total AMPk protein expression, it is possible that cell-type specific alteration in ATP levels following FTO knockdown are responsible for the cell-type specific alteration in AMPk phosphorylation.

NPY mRNA Expression is Decreased in FTO-deficient Cells
Currently, energy intake is the only obesity-associated trait significantly and repeatedly associated with FTO variants [3,11,19,20,21] in humans. To determine if FTO downregulation could control the expression of neuropeptides associated with energy intake, we analyzed the mRNA expression of an orexigenic neuropeptide, neuropeptide Y (NPY), that encourages energy intake in the fasted state [22], and an anorexigenic neuropeptide, Proopiomelanocortin (POMC) [23], that discourages energy intake after feeding, in both naïve and differentiated FTOdeficient SH-SY5Y cells. We observed a significant decrease in NPY in both naïve (241%) and differentiated cells (221%) (Fig 4A), but no change in POMC mRNA in response to FTO downregulation (data not shown). This corresponds with experiments in mice showing a decrease in NPY expression in FTO knockout mice [7]. Furthermore, we studied the signal transducer and activator of transcription 3 (STAT3), an important regulator of NPY expression [22], that is phosphorylated in the hypothalamus by the leptin receptor to inhibit NPY peptide transcription [22]. In FTO siRNA-treated SH-SY5Y cells, phosphorylated STAT3 (Ser 727 ) is upregulated 118% in differentiated cells, while there is no significant change in total STAT3 (Fig 4B).

Discussion
The study presented in this paper attempts to identify the cellular consequences of FTO downregulation that may account for the influence of FTO variants on obesity. We show that downregulation of FTO gene expression by siRNA significantly modulates cellular concentrations of ATP in both SH-SY5Y neuronal cells and 3T3-L1 adipocytes in a cell-type specific manner. We also demonstrate a cell-type specific response in the phosphorylation of AMPk, an important energy sensor that is involved in controlling both energy expenditure and intake, in response to decreased FTO expression. While these results are preliminary we suggest that our observations may provide the foundation for a mechanism that links FTO expression and the development of human obesity.
Presently, it is unclear how FTO expression may contribute to determining cellular energy status. The FTO gene encodes a 2oxoglutarate-dependent nucleic acid-demethylase [24] that may function as a transcriptional co-activator through CCAAT/ enhancer binding proteins [25]. While previous studies demonstrate that alterations in energy status including fasting and glucose administration [8] may influence FTO gene expression [9,11,26], our study is the first to demonstrate a modulation of cellular energy status by changes in FTO expression. We propose that FTO functions as a modulator of cellular energy status exhibited by alteration in ATP concentrations in response to reduced FTO expression.
Our data shows an increase in total ATP concentration in both naive and differentiated FTO-deficient SH-SY5Y cells. In contrast, FTO deficient 3T3-L1 adipocytes exhibit a significant decrease in ATP concentration in both naïve and differentiated cells (Fig 1). Interestingly, these results are reversed in differentiated SH-SY5Y and 3T3-L1 cells treated with oligomycin A, an inhibitor of ATP synthase. These results provide substantial evidence of the involvement of FTO in determining cellular energy status, and show that FTO may influence energy levels in a cell-type specific nature. Unfortunately, the mechanism of this effect is unclear as the observed changes in ATP levels may be caused by alterations in either production or consumption of ATP. Abolishing FTO function significantly alters the expression of the mitochondrial uncoupling protein 2 (Ucp2) [6], and modulation of FTO expression in human myotubes impacts mitochondrial oxidative phosphorylation and alters the expression of mitochondrial genes including ATP5B, UQCR and SOD2 [26]. These combined results suggest the presence of a link between FTO expression and mitochondrial function. However, a link between FTO expression and alterations in ATP consumption must also be explored.
As seen in Figures 2 and 3, naïve SH-SY5Y and 3T3-L1 cells deficient in FTO mRNA display a significant decrease in glucose uptake (227%, 222%, respectively), with a larger decrease in glucose uptake occurring in differentiated cells (251%, 230%, respectively). These results corroborate previous evidence that FTO expression modulates glucose metabolism [9,12,27]. However, no study has identified a correlation between glucose uptake and FTO gene expression and the mechanism of this effect is unclear. It is possible that the modulation of pAMPk by changes in ATP levels is responsible for the decrease in glucose uptake seen in both SH-SY5Y and 3T3-L1 adipocytes. AMPk phosphorylation decreases glucose uptake in 3T3-L1 adipocytes [28], yet increases glucose uptake in skeletal muscle, neurons, and other tissues [28,29]. Thus, we suggest the opposing effects of FTO knockdown on AMPk phosphorylation may lead to the decrease in glucose uptake demonstrated in 3T3-L1 and SH-SY5Y cells as seen in Figure 2 and 3. However, the role of AMPk in decreased glucose uptake seen in FTO deficient cells is not clear and requires further experimentation. Additionally, GLUT1, the primary glucose transporter of basal glucose uptake in SH-SY5Y cells [15,30], is allosterically inhibited by ATP [31], suggesting that increased ATP levels in FTO-deficient SH-SY5Y cells may also potentially contribute to decreased glucose intake. While the mechanism of decreased glucose uptake in response to FTO knockdown is yet to be determined, these results may provide insight into the correlation between FTO variants and altered glucose metabolism demonstrated in previous studies [9,12,27].
At present energy intake is the only obesity-associated trait significantly and repeatedly associated with FTO variants [3,11,19,20,21] in humans. Energy intake is tightly linked to hypothalamic energy balance [32,33] and is mediated to a significant extent through NPY [14,22], an orexigenic neuropeptide associated with increased energy intake in fasted mice [16]. We show in this study that FTO downregulation in SH-SY5Y neuronal cells leads to a significant decrease in NPY expression (Fig 4). The observed decrease in NPY mRNA may be modulated by both a decrease in pAMPk [16], previously shown to activate NPY neurons through Ca 2+ influx [34], and an observed increase in pSTAT3 [14,22] (Fig 4). While POMC is also modulated by AMPk and STAT3 [23], we observed no significant change in POMC mRNA levels. This data is supported by a decrease in NPY expression observed in FTO knockout mice [7] and an increase in NPY expression observed in FTO knock-in mice [5], and it provides a possible link between modulations in FTO expression and alterations in neuropeptides involved in energy intake.
The conclusions drawn in this study are based upon in vitro experiments performed on immortalized cells and thus, their impact on in vivo feeding behavior are only suggestive/speculative. We encourage further research to confirm a connection between FTO expression and energy intake. However, it is interesting to note that our findings largely support previous findings in which non-functional FTO mutations resulted in decreased food intake and decreased accumulation of fat mass [6], and FTO overexpression resulted in increase in fat accumulation and energy intake in mice [5]. However, Tung et al. demonstrate decreased energy intake in response to stereotactic hypothalamic overexpression of FTO in rats, in stark contrast to the data presented here and demonstrated earlier by others [35]. These results, although very preliminary, coupled with previous findings that attribute variants in the FTO gene to increased food intake [3,19] and increased FTO mRNA translation [36] suggest that alterations in the transcriptional status in FTO may contribute to obesity by disrupting neuronal cellular energy status, and altering the signaling mechanisms that are responsible for controlling energy intake.
While previous studies by Tung et al on FTO have evaluated the effect of modulation of FTO gene expression in the hypothalamus [35] and other tissues [5,6,7], no study has attempted to study the effect of alterations in FTO gene expression in hypothalamic neurons or visceral adipocytes in relation to cellular energy balance and downstream signaling cascades. This study was performed in SH-SY5Y neuroblastoma cells and 3T3-L1 adipocytes, cellular models chosen for their convenience and ability to replicate some characteristics of neurons [37] and adipocytes [38] in vivo. Data for both naïve and differentiated SH-SY5Y neuronal cells and 3T3-L1 adipocytes are presented in this paper since previous results suggest that FTO expression may influence differentiation [39] or FTO inactivity may cause premature senescence or decreased cell survival [25]. Thus, we wanted to determine if differentiation modulated the effect of FTO knockdown on metabolic parameters. In all experiments differentiated cells performed in a similar manner as naïve cells, suggesting a lack of impact of differentiation on FTO knockdown mediated alterations in cellular metabolic activities. However, these results are very preliminary and further experiments must be performed that include knockdown and overexpression of FTO mRNA in hypothalamic neurons and visceral adipocytes with corresponding analysis of cellular energy balance and evaluation of the expression of the proteins modulated in this study.
In summary, this study evaluated the effects of siRNA mediated FTO knockdown on ATP levels and downstream signaling in neuronal and adipocyte cell culture. We are the first to show that alterations in FTO gene expression disrupt cellular energy balance in a cell-type specific manner. Additionally, this study is the first to identify the effects of FTO knockdown on AMPk phosphorylation, and we suggest that AMPk may be a primary mediator of FTO's influence on energy intake and obesity. These findings provide insight into FTO mediated obesity and support the notion that FTO expression is an important driver of the development of obesity.  Cells were maintained in 25 cm2 flasks at 37uC and 7% CO2 to approximately 95% confluency and subsequently passaged with 0.05% Trypsin/EDTA.

Knockdown of FTO Expression Utilizing Small-Interfering RNAs
3T3-L1 and SH-SY5Y cells were transfected with either anti-FTO-siRNA purchased from Dharmacon (Lafayette, CO, Cat No: D-004159-02) or Signal Silence Control siRNA (Cat No: 6568S, Cell Signaling Technology), that does not lead to the specific degradation of any cellular message, using either DharmaFECT 3 (3T3-L1) or DharmaFECT 2 (SH-SY5Y) transfection reagent, respectively, according to the manufacturers specifications. Briefly, cells were plated at appropriate cell concentrations (indicated in specific experiments), and after 24 hours, plated cells were rinsed 26 with Opti-MEM media and pre-incubated for 30 minutes in Opti-MEM media at 37uC and 7% CO 2. A complex of 100 nM siRNA and DharmaFECT transfection reagent in Opti-MEM was then added, and cells were incubated for 24 hours in transfection media. After 24 hours, the transfection media was removed, cells were washed 16 and incubated for 24 hours with the appropriate media as described earlier. After a total of 48 hours from the start of transfection, naïve cells were used for experimentation or differentiation. 48 hours was selected since maximum downregulation of FTO occurred at this point. To achieve stable transfection of siRNA during cellular differentiation, cells were transfected for 48 hours as described above and subsequently exposed to the differentiating agents as described in the previous section. qRT-PCR was used to confirm FTO knockdown using Anti-GAPD siRNA (Dharmacon D-001140-01) as a positive control.

qRT-PCR
To analyze mRNA expression of FTO, NPY and POMC; 3T3-L1 and SH-SY5Y cells were plated at 2610 4 or 3610 5 respectively in 35 mm petri dishes and transfected with siRNA as described previously. After 24 hours, dishes were washed 16with media and replaced with appropriate media. After an additional 24 hours (48 hours after the start of transfection) or after differentiation, cells were rinsed 26 with ice-cold PBS and total RNA was isolated from siRNA treated cells using the PureLink Micro-to-Midi TM Total RNA Purification System from Invitrogen (Cat No: 12183-018). cDNA synthesis and qPCR was performed using the Two- Step qRT-PCR Kit from Invitrogen (

Immunoblotting
To determine the effect of FTO knockdown on expression levels of involved proteins, SH-SY5Y and 3T3-L1 cells were plated at 3610 5 or 2610 4 cells per 35 mm petri dish, respectively, and transfected with siRNA as described previously. After 24 hours, dishes were washed 16 with media and replaced with appropriate media. After an additional 24 hours (48 total hours of transfection) or after differentiation, cells were rinsed 26with ice-cold PBS, and lysed in phospho-protein extraction buffer with protease inhibitor cocktail (Roche) as described previously [40]. Cell lysates were separated by 7.5% or 10% SDS-PAGE under reducing conditions. Proteins were transferred to a nitrocellulose immobilization membrane and immunoblotted using the enhanced ECL.

Determination of Cell Viability
To determine the effect of oligomycin and FTO siRNA on the survival of 3T3-L1 adipocytes and SH-SY5Y neurons, cells were plated at 50,000 cells or 200,000 cells per well, respectively, in 12 well plates in DMEM and appropriate media, as described above. After 24 hours, cells were transfected with either FTO or Control siRNA as described above for 48 hours. Cells were then treated in the presence and absence of 10 mM oligomycin for 30 minutes after which cells were trypsinized and counted with trypan blue exclusion. Each data point was repeated in quadruplicate.

ATP Assay
To determine if changes in FTO expression modulate energy balance in SH-SY5Y and 3T3-L1 cells, we quantified the levels of ATP present within cells with or without FTO downregulation. ATP content of cells was measured using the Perkin Elmer ATPlite luminescence assay system (Waltham, MA, Cat No: 6016941). Briefly, SH-SY5Y and 3T3-L1 cells were plated at 2610 3 and 3610 3 cells/well, respectively, in a 96 well plate, and transfected as described previously. After 24 hours, wells were washed 16 with media and replaced with appropriate media. After an additional 24 hours (48 total hours of transfection) or after differentiation, cells were rinsed 26 with ice-cold PBS and incubated for 30 minutes in glucose-free media with or without 10 mM oligomycin, a specific inhibitor of ATP synthase to reduce levels of endogenous ATP. Cells were then lysed with the provided 16 mammalian cell lysis buffer for 5 minutes on an orbital shaker. After which the provided reconstituted luciferase substrate was then added to the cell lysate and incubated for 5 minutes. The plates were then incubated in dark for 10 minutes, and luminescence was measured using the Bio Tek Synergy 2 multimode microplate reader (Winooski, TN).

Analysis of Results
qPCR results were analyzed by the Applied Biosciences 7300 Real-Time PCR System software with GAPD mRNA serving as a positive internal control. Immunoblotting results were analyzed by densitometric analysis using the Image J software from the National Institutes of Health with values representing percent of control from an average of triplicate experiments. Experiments on cell viability were analyzed using repeated measures ANOVA with a = 0.05. ATP and Glucose experiments were performed in with paired control groups in ''N'' number of replicates as indicated, respectively, and analyzed against the appropriate control siRNA and diluent groups using a one-way ANOVA with post-hoc analysis. Data is presented as FTO siRNA vs Control siRNA, because diluent groups were found be uniformly similar to Control siRNA groups. Significance was established at a = 0.05.