Clinical Science

Research article

Conditional expression of the FTO gene product in rat INS-1 cells reveals its rapid turnover and a role in the profile of glucose-induced insulin secretion

Mark A. Russell, Noel G. Morgan


Common polymorphisms within the FTO (fat mass and obesity-associated) gene correlate with increased BMI (body mass index) and a rising risk of Type 2 diabetes. FTO is highly expressed in the brain but has also been detected in peripheral tissues, including the endocrine pancreas, although its function there is unclear. The aim of the present study was to investigate the role of FTO protein in pancreatic β-cells using a conditional expression system developed in INS-1 cells. INS-1 cells were stably transfected with FTO–HA (haemagluttinin) incorporated under the control of a tetracycline-inducible promoter. Induction of FTO protein resulted in localization of the tagged protein to the nucleus. The level of FTO–HA protein achieved in transfected cells was tightly regulated, and experiments with selective inhibitors revealed that FTO–HA is rapidly degraded via the ubiquitin/proteasome pathway. The nuclear localization was not altered by proteasome inhibitors, although following treatment with PYR-41, an inhibitor of ubiquitination, some of the protein adopted a perinuclear localization. Unexpectedly, modestly increased expression of FTO–HA selectively enhanced the first phase of insulin secretion when INS-1 monolayers or pseudoislets were stimulated with 20 mM glucose, whereas the second phase remained unchanged. The mechanism responsible for the potentiation of glucose-induced insulin secretion is unclear; however, further experiments revealed that it did not involve an increase in insulin biosynthesis or any changes in STAT3 (signal transducer and activator of transcription 3) expression. Taken together, these results suggest that the FTO protein may play a hitherto unrecognized role in the control of first-phase insulin secretion in pancreatic β-cells.

  • β-cell
  • fat mass and obesity-associated (FTO)
  • glucose-induced insulin secretion
  • signal transducer and activator of transcription 3 (STAT3)
  • ubiquitin/proteasome


The incidence of obesity in adults and children has increased markedly in recent years and remains on an upward trajectory [1]. Much of this increase has been attributed to changes in eating habits and lifestyles, and it is placing an increasing burden on global health care resources. Although environmental factors clearly play an important primary role in determining the incidence of obesity and Type 2 diabetes, it is clear that the interaction of relevant genes with these environmental triggers is also of importance. The concept that genes can impact on adiposity is not novel, with earlier studies suggesting that between 40 and 70% of the common variation in BMI (body mass index) may have an underlying genetic cause [2,3]. Despite this, common genetic variants associated with modest changes in adiposity have proved elusive, although less common monogenic forms of morbid obesity have been described [4]. In a recent GWAS (genome-wide association study), common polymorphisms within the first intron of the FTO (fat mass and obesity-associated) gene were identified in association with increased BMI and a rising risk of Type 2 diabetes [5]. This association was evident in both adult and child cohorts and has since been extensively replicated [69].

FTO had been originally identified during a study of aberrant limb development in mice, where animals heterozygous for a 1.6-Mb deletion in chromosome 8 were found to develop fused toes as well as thymic hyperplasia [10]. FTO was implicated as one of several genes that might be involved, although its functionality was not fully explored, and it was not linked to altered fat mass at that time [11]. More recently, the FTO gene product has been reported to possess enzymatic activity and to function as a 2-oxoglutarate-dependent oxygenase, which is capable of removing rare cytotoxic methyl lesions from DNA or RNA [12,13]. However, it is unclear whether this activity plays a role in the determination of adiposity.

FTO mRNA is highly expressed in areas of the brain associated with energy balance [12], which is consistent with a possible role in appetite regulation and, as such, would fit with a proposed role in the regulation of fat mass. However, FTO mRNA is also expressed in other tissues including whole pancreatic islets, isolated β-cells [5,6,14] and α-cells [14], where its functional role has not been defined. The aim of the present study was to investigate the regulation and role of FTO in pancreatic β-cells by conditional overexpression of the protein in the clonal β-cell line INS-1.


Creation of transfected INS-1 cells

A cDNA encoding the mouse FTO coding sequence tagged at the 3′ end with HA (haemagglutinin; termed FTO–HA) was obtained from Professor Carol Wicking (Institute of Molecular Bioscience, University of Queensland, St Lucia, Australia). INS-1 cells that conditionally overexpress FTO–HA were constructed using the Flp-In–T-REx gene expression system (Invitrogen) using the methods described by Welters et al. [15]. Briefly, FTO–HA was subcloned into a pcDNA5/FRT/TO expression vector (Invitrogen) using the ApaI and HindIII restriction sites. The plasmid was then transfected into the INS-1 Flp-In–T-REx parental clone #5.3–19 [16] using TransFAST reagent (Invitrogen) to generate cloned INS–FTO cells. The empty pcDNA5/FRT/TO vector was similarly transfected as a control into a parallel clone, and these cells were designated INS–FRT.

Transfected INS-1 cells were cultured in RPMI 1640 medium containing 11 mM glucose supplemented with 10% (v/v) FBS (fetal bovine serum), 2 mM L-glutamine, 100 units/ml penicillin, 100 μg/ml streptomycin, 50 μM 2-mercaptoethanol and 150 μg/ml hygromycin B to apply selection pressure. To induce cDNA expression, cells were incubated in the presence of tetracycline at the concentrations shown in individual experiments. Cells were cultured in 5% CO2 at 37 °C and 100% humidity. Pseudoislets were formed by culturing INS–FTO cells in 1% (w/v) gelatin-coated flasks for 72–96 h [17].

RT–PCR (reverse transcription–PCR)

Total RNA was extracted from cells using TRIzol® reagent (Invitrogen), and the RNA was used to generate cDNA by reverse transcription. cDNA was amplified by PCR in single tube reactions (Abgene) using primers designed for rat or human FTO (rat forward, GCGATGATGAAGTGGACCTT and reverse, GCAGTCTCCCTGGTGAAGAG; human forward, CGGTATCTCGCATCCTCATT and reverse, GGCAGCAAGTTCTTCCAAAG). Cycling conditions (35 cycles) were 95 °C for 30 s, 46 °C for 30 s and 72 °C for 30 s. Products of the PCR reaction were separated by agarose gel electrophoresis and viewed under longwave UV illumination. The identity of PCR products was confirmed by direct sequencing (MWG) following extraction from the gel.

Western blotting

Cells were induced with tetracycline and exposed to the test reagents epoxomicin (Enzo Life Sciences) or PYR-41 (Merck) prior to extraction of whole-cell proteins as described previously [15]. An equal concentration of protein was denatured, loaded and then run alongside markers on precast Tris/HCl-buffered 12.5% (w/v) polyacrylamide gels (Bio-Rad). The proteins were transferred from the gel on to a PVDF membrane by electrotransfer. Following transfer, the membrane was blocked for 1 h in Tris-buffered saline containing 0.05% Tween 20 and 5% (w/v) dried milk powder. Membranes were probed at room temperature (20 °C) for 4 h using either anti-HA (New England Biolabs) or β-actin (Sigma) antibodies (1:1000 and 1:10000 dilution respectively) or overnight at 4 °C with anti-STAT3 (signal transducer and activator of transcription) (New England Biolabs) antibody (1:1000 dilution). After washing, the membrane was probed for 1 h with an AP (alkaline phosphatase)-conjugated anti-(mouse IgG) secondary antibody (Sigma) (1:30000 dilution). Bands were detected by exposure to X-ray film using the chemiluminescent CDP-Star® system.


Cells (4×103) were seeded on to sterile coverslips and exposed to tetracycline plus test reagents as required. Following incubation, cells were fixed for 10 min in 4% (w/v) paraformaldehyde and then washed twice in PBS. The fixed cells were permeabilized using ADS [antibody-diluting solution: PBS, 0.1 M lysine, 10% (v/v) donor calf serum and 0.02% (w/v) NaN3] containing 0.2% (v/v) Triton X-100 for 30 min. Cells were then probed with primary anti-HA antibody (diluted 1:400 in ADS + 0.2% Triton X-100) and incubated at 4 °C overnight. The cells were washed in PBS and incubated for 30 min with anti-mouse secondary antibody [Alexa Fluor® 488 (Invitrogen) at a 1:500 dilution in ADS] and Hoescht 33258 (1:1000 dilution) to stain the nucleus. The cells were washed as before and mounted cell side down on to glass slides, before viewing by fluorescence microscopy (Eclipse 80i; Nikon). All images were captured at ×40 magnification using NIS Elements software linked to a digital imaging head (Nikon).

Insulin secretion

For static insulin secretion experiments, cells were seeded into 24-well plates at a density of 105 cells/well and induced for 18 h with 1 μg/ml tetracycline. Cells were washed and then incubated in bicarbonate-buffered saline solution [18] containing 0.1% BSA at 37 °C and 0.2 mM glucose. Cells were stimulated in this buffer supplemented with either 0.2 or 20 mM glucose for a further 1 h. After this time, the supernatant was removed and insulin release determined by RIA [19] (coefficient of variation <10%).

For perifusion experiments, INS–FTO pseudoislets were configured and cDNA expression was induced by incubation with 1 μg/ml tetracycline. Pseudoislets were then loaded inside plastic chambers (male–female Luer connectors) plugged with glass wool and perifused with bicarbonate-buffered saline [18] at a temperature of 37 °C and a flow rate of approx. 1 ml/min. On commencement of the experiment, the perifusate was collected every 1 min, and insulin release was measured by RIA.

Statistical analysis

SPSS 16.0 was employed for all statistical analysis, with results expressed as means ± S.E.M. A Student's t test was used to provide statistical comparisons between experimental groups, with P<0.05 considered as significantly different.


Expression of FTO in islet cells

Previous work has suggested that FTO is expressed in pancreatic islet cells [5,6], and we initially confirmed this using cDNA prepared from human islets (Figure 1a) and from two rat β-cell lines [INS-1 (Figure 1b) and BRIN-BD11 (Figure 1c)]. In each case, RT–PCR analysis generated an appropriately sized amplicon, whose identity was verified by direct DNA sequencing.

Figure 1 FTO mRNA expression in clonal β-cell and human islet cDNA

The expression of FTO in cDNA extracted from (a) human islets, (b) INS-1 and (c) BRIN-BD11 cells was determined by RT–PCR. Bands were separated on agarose gels and viewed under UV illumination after staining with ethidium bromide. The lowest arrow indicates the 200-bp DNA marker, with each arrow ascending in 100-bp increments. FTO expression was confirmed after direct sequencing of the extracted amplicons.

Characterization of INS–FTO cells

In order to study the function of FTO in β-cells, the protein bearing a C-terminal HA tag was stably overexpressed in INS-1 cells in a conditional manner using a tetracycline-responsive promoter via the Flp-In–T-REx system [20] to yield INS–FTO cells. Control cells expressing empty vector were generated in parallel and designated INS–FRT. The Flp-In–T-REx system is organized such that all cells incorporate the introduced cDNA within a specific cassette that is present in the parental cell line, by targeted recombination. Hence, unlike the situation in many types of stable transfection of mammalian cells, clonal selection of transfectants is not required, since all transfected cells are structurally and functionally equivalent.

Exposure of INS–FTO cells to tetracycline (50–1000 ng/ml) resulted in the appearance of an intensely positive band of approx. 58 kDa upon Western blotting with an anti-HA antibody, and this corresponded with the expected molecular mass of HA-tagged FTO (Figure 2). This band was absent from uninduced INS–FTO cells (Figure 2) and from INS–FRT cells treated with tetracycline (results not shown). Immunocytochemical analysis using fluorescence microscopy revealed that FTO–HA was localized exclusively within the nucleus of the cells (Figure 3).

Figure 2 Incubation of INS–FTO cells with increasing tetracycline concentrations induces FTO–HA protein production

INS–FTO cells were exposed to increasing tetracycline concentrations (50–1000 ng/ml) for a 24-h incubation period, after which total protein was extracted. FTO–HA expression levels were determined by Western blotting. β-Actin was used as a loading control. Results are representative of three separate experiments.

Figure 3 FTO localizes to the nucleus of INS-1 cells

INS–FTO cells were exposed to 1 μg/ml tetracycline for 24 h. Cells were then fixed and probed with an FITC-labelled secondary antibody directed against the primary anti-HA antibody (green) and with Hoescht stain (blue) to identify the nucleus. Cells were viewed by fluorescence microscopy at ×40 magnification. Scale bar, 10 μm.

Surprisingly, the induction of FTO–HA protein expression in INS–FTO cells in response to tetracycline was not truly dose-dependent. Incubation of INS–FTO cells with 50 ng/ml tetracycline resulted in the appearance of an appropriately sized band on Western blots, but the intensity of this band did not increase further when higher concentrations of the antibiotic were employed (Figure 2). This contrasts markedly with the response observed previously, when a range of other proteins have been expressed under the control of the tetracycline-responsive promoter in this parental line of INS-1 cells. In each of these cases, protein expression was elevated as the tetracycline concentration was increased progressively between 50 and 1000 ng/ml [15,20]. These data are, therefore, consistent with the possibility that FTO–HA may be subject to rapid turnover in INS-1 cells, thereby limiting its accumulation during induced expression.

Stability of the FTO protein in INS-1 cells

To address this possibility, the stability of FTO–HA was examined in INS–FTO cells by an initial induction with tetracycline followed by withdrawal of the antibiotic and subsequent monitoring of protein expression at timed intervals over 96 h. Western blotting confirmed the expression of FTO–HA during tetracycline treatment, but revealed that its expression was reduced as early as 2 h after tetracycline withdrawal. Within 4 h of removal of the antibiotic, FTO–HA expression was barely detectable (Figure 4a), and this situation was maintained at later time points. These results imply that FTO–HA is turned over very rapidly in INS-1 cells and that the protein does not accumulate, even under conditions of enhanced mRNA expression.

Figure 4 FTO–HA is rapidly degraded in a proteasomal and ubiquitin-dependent manner in INS-1 cells

INS–FTO cells were grown in the presence of 1 μg/ml tetracycline for 12 h, at which point the antibiotic was removed (arrow), and total protein extracted at 2-h increments over a 12-h period (a). Alternatively, following tetracycline removal, cells were cultured in the absence or presence of (b) 2 μM epoxomicin or (c) 50 μM PYR-41. Following incubation, total protein was extracted and FTO–HA production was determined by Western blotting. β-Actin was used as a loading control. Results are representative of three separate experiments.

In order to determine the mechanism by which FTO–HA is degraded under these conditions, specific proteasomal inhibitors were used. FTO–HA expression was induced with tetracycline (1 μg/ml for 12 h), and the antibiotic was then withdrawn and the cells incubated in the absence or presence of either epoxomicin (2 μM) or MG-132 (0.1 μM) for a further 6-h period. Protein was extracted at 2-h intervals and FTO–HA levels analysed by Western blotting. As observed previously, withdrawal of tetracycline was followed by a dramatic reduction in FTO–HA protein expression in INS–FTO cells cultured in the absence of proteasome inhibitors (Figure 4b). However, the presence of either epoxomicin (Figure 4b) or MG-132 (results not shown) attenuated the rate of degradation dramatically. Indeed, in the presence of either inhibitor, FTO–HA protein expression was elevated beyond that observed with tetracycline alone (Figure 3b). Similar results were obtained when protein expression was studied by immunocytochemistry (Figure 5a), and these experiments also revealed that FTO–HA was retained within the nucleus throughout the period of exposure to proteasome inhibitors.

Figure 5 Blockade of FTO–HA degradation by inhibitors of the proteasome and ubiquitin E1-activating enzyme

INS–FTO cells were exposed to 1 μg/ml tetracycline for 12 h (tetracycline), after which time the antibiotic was withdrawn and cells incubated for an additional 6 h with vehicle, epoxomicin (2 μM) or PYR-41 (50 μM). Following incubation, cells were fixed and probed with a primary anti-HA antibody and an FITC-conjugated secondary antibody (green) or with Hoescht stain (blue) to identify the nucleus. Cells were viewed by fluorescence microscopy at ×40 magnification. Scale bar, 30 μm. Micrographs are representative of three separate experiments.

Since proteins are normally labelled for proteasomal degradation by the attachment of multiple ubiquitin molecules, the role of ubiquitination in mediating the turnover of FTO was investigated. Ubiquitination is controlled by the sequential action of three enzymes, the first of which (ubiquitin-activating enzyme or E1) is selectively inhibited by PYR-41 [21]. E1 expression was confirmed in INS-1 cells by RT–PCR (results not shown), and the effects of PYR-41 on FTO–HA turnover were studied. FTO–HA expression was induced with tetracycline; the antibiotic was then removed and the cells grown in the presence or absence of PYR-41 (50 μM) for a further 6 h. As observed with the proteasome inhibitors, the presence of PYR-41 attenuated FTO–HA degradation markedly (Figure 4c). When the subcellular localization of FTO–HA was examined under these conditions by immunocytochemistry, a change was observed when compared with the localization observed under control conditions. Thus, although significant nuclear FTO–HA staining still remained when cells were exposed to PYR-41, clear staining was now also observed in the perinuclear region of all cells (Figure 5b). This might imply that ubiquitination of FTO–HA (or an associated protein) represents an important step for either its entry into the nucleus or for the maintenance of a nuclear localization. To distinguish between these possibilities, INS–FTO cells were pre-incubated with PYR-41 for 1 h prior to induction of FTO expression with tetracycline during a further 2-h period (in the continued presence of PYR-41). In control cells not exposed to PYR-41, FTO–HA expression was evident within 2 h of addition of tetracycline, and the protein was localized to the nucleus. Cells pre-incubated with PYR-41 exhibited an enhanced level of fluorescence by comparison with controls, consistent with the expected attenuation of FTO–HA degradation under these conditions. Some of this protein was again localized in the perinuclear region of the cells, but the majority was nuclear, suggesting that pre-incubation with PYR-41 had not prevented the initial entry of FTO–HA into this organelle (results not shown).

Effect of FTO overexpression on insulin secretion

Having established that FTO is rapidly turned over by the proteasomal pathway in INS-1 cells, the functional effects of overexpression of the protein were then investigated. Initial studies suggested that up-regulation of FTO did not alter either the rate of proliferation of INS-1 cells or their propensity to undergo apoptosis (results not shown). By contrast, changes in expression of the protein did influence the insulin secretory response to glucose.

Short-term (20 min) stimulation of INS-1 cells with 20 mM glucose under static incubation conditions resulted in an equivalent small but significant rise in insulin secretion in both INS–FRT (120 ± 3.5% compared with control; P< 0.001) and INS–FTO cells (126.5 ± 8.1% compared with control; P<0.01) in the absence of tetracycline. The magnitude of this effect was not influenced by the exposure of INS–FRT cells to tetracycline, but unexpectedly, the response to 20 mM glucose was modestly but significantly enhanced in INS–FTO cells (Figure 6). In contrast, the cumulative rise in insulin secretion achieved during a longer period of incubation with 20 mM glucose (60 min) was not significantly increased following FTO expression (Figure 6). These results suggest that up-regulation of FTO may selectively influence the early insulin secretory response to glucose.

Figure 6 Induction of FTO–HA expression enhanced first-phase insulin secretion in response to glucose

INS–FRT (a and b) or INS–FTO cells (c and d) were exposed to low (0.2 mM) or high (20 mM) glucose after 18 h incubation in the presence or absence of 1 μg/ml tetracycline. Cells were incubated in either 0.2 or 20 mM glucose for either 20 min (a and c) or 1 h (b and d) and insulin secretion was determined by RIA. Results are represented as percentage of the control rate (100%) as means ± S.E.M. measured in the presence of 0.2 mM glucose (n=4). **P<0.01 compared with uninduced cells stimulated with 20 mM glucose. NS, not significant.

In order to examine whether the initial response to a non-nutrient stimulus was also amplified following induction of FTO expression, additional experiments were performed in which 30 mM KCl was used as a depolarizing stimulus (Figure 7). Exposure of INS–FTO cells to KCl induced a modest increase in insulin secretion over a 20-min period, which was comparable with that observed in response to 20 mM glucose. Importantly, induction of FTO expression with tetracycline significantly potentiated the KCl-induced rise, thereby confirming the suggestion that FTO may influence the first phase of insulin secretion.

Figure 7 Insulin secretion in response to depolarization with KCl is enhanced by the induction of FTO–HA expression

INS–FTO cells were incubated for 18 h in the absence or presence of 1 μg/ml tetracycline before exposure to either low (0.2 mM) or high (20 mM) glucose, or 30 mM KCl for a 20-min period. The resulting insulin secretion was determined by RIA. Results are represented as percentage of the control rate (100%) as means ± S.E.M. measured in the presence of 0.2 mM glucose (n=2). **P<0.01 compared with uninduced cells stimulated with either 20 mM glucose or 30 mM KCl.

To study the profile of insulin secretion in response to FTO overexpression in more detail, INS–FTO cells were configured as pseudoislets for ease of handling and then exposed to glucose in a perifusion system. When cultured in the absence of tetracycline prior to the perifusion experiment, INS–FTO pseudoislets displayed a modest biphasic insulin secretory response to glucose (Figure 8). The first phase was initiated very quickly following stimulation with 20 mM glucose and was sustained for approx. 8–10 min, before the secretory rate declined towards the basal level. The second phase of insulin secretion was then initiated, and the rate remained elevated for the remainder of the experiment.

Figure 8 Profile of glucose-induced insulin secretion following induction of FTO–HA in INS-1 cells

INS–FTO cells were configured as pseudoislets and either exposed to tetracycline (TET; induced) or left untreated (control) for the final 18 h of culture. They were then transferred to perifusion chambers and perifused with medium containing 0.2 mM glucose prior to the introduction of 20 mM glucose (arrow). Medium was collected at 1-min intervals and assayed for insulin. In all cases, results are means ± S.E.M. relative to the control rate averaged over the 5 min immediately prior to stimulation, and were obtained in five separate experiments. *P<0.05 relative to uninduced cells.

In order to determine the effect of FTO on insulin secretion kinetics, INS–FTO pseudoislets were pre-incubated in either the presence or absence of tetracycline and then perifused in parallel. As observed in short-term static incubation experiments, in the case of pseudoislets overexpressing FTO–HA, the AUC (area under the curve) representing first-phase insulin secretion was increased by comparison with uninduced cells (Figure 8). This was largely due to an apparent increase in the duration of the first-phase secretion, which was more prolonged upon FTO overexpression, such that the nadir preceding the initiation of the second phase of insulin secretion could no longer be readily discerned in these cells. As a result, the differences were most pronounced during the latter 6 min of the first-phase response when the mean AUC was increased from 240 ± 46% above basal in uninduced INS-1 pseudoislets to 431 ± 96% above basal in those overexpressing FTO (P<0.05). In contrast with the first phase, there was no difference in the magnitude of the second phase of insulin secretion between control and FTO-overexpressing pseudoislets. Moreover, when the pseudoislets were subsequently challenged with 30 mM KCl at the end of the perifusion period, no enhancement of response was observed in those expressing FTO–HA (results not shown).

As a means to establish whether changes in insulin content in response to overexpression of FTO were responsible for altering the first phase of insulin secretion, the insulin content of INS–FTO cells was measured in the presence or absence of tetracycline. No change in insulin content was detected following the induction of FTO–HA by the antibiotic when compared with uninduced cells (8.98 ± 0.2 ng/105 cells in uninduced cells compared with 8.51 ± 0.1 ng/105 cells in induced cells).

In view of the finding of an improved first-phase insulin secretion following up-regulation of FTO expression, the subcellular localization of the protein was examined in cells exposed to 20 mM glucose. Immunocytochemical analysis revealed that FTO–HA retained its nuclear localization following exposure of the cells to glucose for 0, 20 or 60 min (results not shown).

STAT3 expression

Recent work has suggested that modulation of FTO mRNA levels may influence STAT3 expression [22], and it is known that alterations in the level of this transcription factor can have an impact directly on the magnitude of the first phase of insulin secretion [23]. Therefore STAT3 expression was measured at the protein level in response to FTO up-regulation in INS–FTO cells. Western blot revealed a band of strong intensity corresponding with the expected molecular mass of the STAT3 protein in uninduced cells. However, in three separate experiments, no significant increase in STAT3 protein expression was discernable in cells that had been cultured in the presence of tetracycline to induce FTO for 24 h (Figure 9).

Figure 9 STAT3 levels following induction of FTO–HA production in INS-1 cells

INS–FTO cells were cultured in the presence or absence of 1 μg/ml tetracycline for 24 h. Following incubation, total protein was extracted and STAT3 was determined by Western blot. β-Actin was used as a loading control. Results are representative of three experiments.


GWASs have implicated polymorphisms within the first intron of FTO as being associated with an increase in BMI and the risk of Type 2 diabetes in various human populations [57]. Additional studies have detected FTO mRNA expression in brain areas that control appetite and feeding behaviour, which would be consistent with a role for this gene product in the regulation of body mass. However, a number of other tissues, including whole pancreatic islets and isolated β-cells, also express FTO at the mRNA level [5,6], although its function in these cells is unclear. It is also uncertain whether variations arising from the polymorphisms in the FTO gene that influence BMI can also alter β-cell function. However, an important finding in the present study is that induced expression of FTO led to enhancement of the first phase of glucose-induced insulin secretion from stably transfected INS-1 cells, whereas the second phase was not altered.

To address the function of FTO in β-cells, we chose to exploit the potential of a conditional expression system based on a parental line of these cells, INS-1 5.3–19, containing an FRT cassette [16]. HA-tagged mouse FTO cDNA was then incorporated into this parental clone using Flp-recombinase-mediated recombination, to allow conditional overexpression of the protein in response to tetracycline. This was confirmed directly in Western blots using an antiserum directed against the HA tag. Importantly, we did not detect expression of the protein in cells that were not exposed to tetracycline, suggesting that the promoter provided tight regulation of expression.

Exposure of INS–FTO cells to elevated concentrations of glucose (20 mM) resulted in an approx. 2-fold increase in insulin secretion which, by perifusion, was found to display a biphasic pattern. Upon expression of FTO–HA, the INS-1 cells responded to the glucose stimulus with a more sustained increase in insulin secretion over the initial 10-min period of stimulation. This was manifest as a loss of the nadir that usually marks the end of the ‘first phase’ of the secretory response such that both phases were effectively merged. No increase in the second phase of secretion was detected.

Therefore, these findings suggest that FTO may play a previously unrecognized role in the regulation of first-phase secretion. In this context, it is important to note that GWASs have previously correlated obesity-associated polymorphisms in FTO with increased insulin secretion, but that this association was invariably lost when corrected for BMI [24,25]. Hence it has been deduced that the effects of FTO on fat mass are determined primarily by its expression at extrapancreatic sites, rather than by alterations in insulin release. However, other work has revealed an association between the risk allele of FTO and increases in the BIGTT-AIR index (β-cell function, insulin sensitivity and glucose tolerance test-acute insulin response), a measure of early-phase insulin secretion [26]. Furthermore, a recent study has indicated that the expression of insulin and FTO are positively correlated in sorted human β-cells [14]. Taken together with the present findings, these results offer circumstantial evidence that FTO may regulate aspects of insulin secretion, which are relevant to its effects on BMI in humans.

Mechanistically, it remains unclear how FTO influences the first phase of insulin secretion from INS-1 cells. FTO induction did not enhance the insulin content of the cells, suggesting that it may act to control relevant steps in exocytosis rather than insulin biosynthesis. Since the first phase of secretion is believed to represent the exocytosis of a readily releasable pool of docked secretory granules, it is possible that FTO promotes the translocation of granules into this pool. We and others [12,27] have shown that FTO occupies a nuclear localization, and it remains uncertain, therefore, how the protein acts to influence the translocation of insulin granules from cytosolic pools to the plasma membrane. The nuclear localization is fully consistent with FTO's suggested role as a nucleic acid demethylase [12,13], and we speculate that it may modulate the expression of one or more genes involved in secretory granule trafficking. This could be mediated by changes in the pattern of DNA methylation, since it is reported, for example, that DNA methylation of CpG islands located within the promoter region of PGC1 (peroxisome-proliferator-activated receptor γ co-activator 1) is decreased in subjects having the FTO risk allele [28]. In that study, the authors did not measure the resultant expression levels of the protein, although changes in DNA methylation status have been reported to regulate the expression of certain genes [29].

In this context, Tung et al. [22] have reported that overexpression of FTO in the hypothalamus gives rise to an approx. ~4-fold increase in STAT3 mRNA levels, and it is known that deletion of STAT3 in the mouse leads to a marked reduction in the first phase of insulin secretion [23]. Hence an indirect regulation of insulin secretion via STAT3 is possible. However, we did not observe any obvious effects of FTO on total STAT3 protein levels in INS-1 cells.

The first phase of insulin secretion is known to be lost or reduced in diabetes [30], and this changes provides early evidence of β-cell dysfunction. A recent study by Kirkpatrick et al. [14] revealed that the mRNA level of FTO was reduced in isolated human β-cells of patients with Type 2 diabetes when compared with controls. When considered together with our present results, showing that an increase in FTO leads to enhancement of first-phase insulin secretion, it seems possible that the reduced levels of FTO found in Type 2 diabetic β-cells may be one factor which contributes to the deterioration in first-phase insulin secretion in these cells.

One surprising feature of the induction of FTO–HA expression in INS-1 cells was that the accumulation of immunoreactive protein was not increased dose-dependently in concert with the concentration of the inducer. Thus the expression levels achieved with 50 ng/ml tetracycline were similar to those obtained with 1 μg/ml. This was unexpected, as a range of other proteins which have been overexpressed using the Flp-In–T-REx system, are accumulated in INS-1 cells in progressively larger amounts in response to a graded stimulus for induction [15,20]. To explain this, we hypothesized that the FTO–HA protein might be turned over rapidly in the cells, such that its rate of synthesis is compensated by a correspondingly high level of degradation. In support of this, FTO–HA expression in INS–FTO cells declined to almost undetectable levels within 4 h of removal of tetracycline from the incubation medium, following an initial 12-h induction period. This decline was completely attenuated by selective inhibitors of either the proteasome or protein ubiquitination. Taken together, these findings suggest that FTO levels are tightly regulated in the β-cell and that accumulation of the protein is minimized by rapid ubiquitination and subsequent proteasomal degradation.

The nuclear localization was unaffected following incubation of INS–FTO cells with proteasome inhibitors, implying that the protein was not trafficked from the nucleus to the cytosol for proteasomal degradation. Nuclear proteolysis has been described previously [31], and it is possible, therefore, that degradation of FTO–HA occurs within this organelle. By contrast, a change in the localization of FTO–HA was observed upon the use of an inhibitor of ubiquitination (PYR-41), when a proportion of the protein became localized perinuclearly. On this basis, we considered whether the ubiquitination of FTO (or an associated protein) might represent an important step in either its nuclear import or for the maintenance of nuclear localization. Pre-incubation of INS–FTO cells with PYR-41 prior to induction of FTO–cDNA expression resulted in an identical pattern of FTO–HA expression compared with that observed when the inhibitor was introduced at a later time point. These results imply that ubiquitination of FTO is not required for nuclear import but, rather, that this process may be important for nuclear retention. In drawing this conclusion, it is important to note that PYR-41 has also been reported to alter protein SUMOylation [21], and it cannot be excluded that this process might be important for directing the localization of FTO [32].

There are some important limitations to the experimental system used in the present study that warrant discussion. First, we have expressed mouse FTO in a rat β-cell line (INS-1), which could, in principle, lead to alterations in response. However, we suspect that this is unlikely, since FTO shares approx. 96% homology between rat and mouse at the amino acid level, and the residues thought to be critical for enzyme function are fully conserved. Secondly, the FTO construct we have used was HA-tagged, and this might influence its distribution or stability. However, although such changes have been described occasionally when larger tags are used [e.g. GFP (green fluorescent protein)], we are not aware of evidence that small tags (such as HA) significantly influence these parameters. Thirdly, it must be emphasized that we have not used antibodies directed against FTO itself to detect the protein in the present studies. Rather, we monitored induction of the HA-tagged FTO. Thus we are unable to draw firm conclusions about the levels of overexpression achieved in the present studies.

Despite these limitations, our findings provide evidence for a functional role of FTO within the pancreatic β-cell and show that FTO protein production appears to be highly regulated by the ubiquitin/proteasome system in these cells. Overexpression of FTO enhanced the first phase of insulin secretion, suggesting a potential and hitherto unrecognized role for this protein in the regulation of insulin secretion.


This work was supported by the Peninsula Medical School Foundation and the Higher Education Infrastructure fund (Ph.D. studentship to M.R.).


Mark Russell conducted all of the experiments. Mark Russell and Noel Morgan designed the experiments and prepared the manuscript for publication.


We thank Professor Carol Wicking for kindly donating the HA-tagged FTO construct. We also thank Professor Gerhart Ryffel for providing the parental Flp-In–T-REx INS-1 clone for use in these studies.

Abbreviations: ADS, antibody-diluting solution; AUC, area under the curve; BMI, body mass index; FTO, fat mass and obesity-associated; GWAS, genome-wide association study; HA, haemagglutinin; RT–PCR, reverse transcription–PCR; STAT3, signal transducer and activator of transcription 3


View Abstract