Sex-Specific Control of Fat Mass and Counterregulation by Hypothalamic Glucokinase Free

Glucose sensing by specialized cells in the central nervous system influences numerous homeostatic processes, such as the counterregulatory response to hypoglycemia, β-cell proliferation and insulin secretion, thermogenesis, and feeding initiation or termination (1,2). Central glucose-sensing cells regulate the function of the liver, of brown and white fat, of muscles, and of pancreatic islet cells by controlling sympathetic and parasympathetic nerve activity. Several cell types in the central nervous system are capable of glucose sensing, including neurons, astrocytes and tanycytes. There are two types of glucose-sensing neurons: glucose-excited (GE) and glucose-inhibited (GI) neurons, whose firing rate is increased by an increase or decrease in extracellular glucose concentrations, respectively (3–5). GE and GI neurons are found in several hypothalamic and brainstem nuclei. Many are found in the ventromedial hypothalamus (VMH), which consists of the ventromedial nucleus (VMN) and the arcuate nucleus (ARC). The role of the VMH in homeostatic regulation was first demonstrated by lesion studies that induced obesity (6), suppressed glucagon secretion in response to hypoglycemia (7), and increased β-cell proliferation and insulin secretion (8). Furthermore, injection of 2-deoxy-d-glucose (2DG) in the VMH induces glucagon secretion, whereas glucose injection during insulin-induced hypoglycemia blocks glucagon secretion (9,10). Because defective counterregulation leading to severe hypoglycemia is a major threat for both patients with type 1 diabetes and patients with type 2 diabetes receiving insulin therapy (11), a better understanding of the glucose-sensing mechanisms controlling glucagon secretion, in particular in the VMH, is an important goal of research (12).

The mechanisms of glucose sensing by GE and GI neurons are multiple (13). GE neurons respond to an increase in glucose by activating a signaling pathway similar to that used by pancreatic β-cells to induce insulin secretion. This is initiated by glucose uptake by glucose transporters, followed by phosphorylation by glucokinase (Gck), which controls the rate-limiting step in glycolysis; the subsequent increase in the intracellular ATP-to-ADP ratio closes K_ATP channels to induce membrane depolarization, Ca⁺⁺ influx, and insulin secretion by β-cells, or firing activity by GE neurons (2). In GI neurons of the VMH, a decrease in glucose concentration induces plasma membrane depolarization through the activation of AMPK, a nitric oxide–dependent amplification of AMPK activity, and the closure of a chloride conductance, possibly the cystic fibrosis transmembrane conductance regulator (14). Although the models for glucose signaling in GE and GI neurons are very different, in the VMH the glucose responsiveness of both types of neurons has been reported to require Gck activity (15–17) and be necessary for proper glucagon secretion and food intake during insulin-induced hypoglycemia (18,19). Thus Gck in the VMH participates in a number of glucoregulatory mechanisms, but its role in specific hypothalamic neurons in integrative physiology has not yet been tested.

A large fraction of VMN neurons are characterized by expression of the transcription factor steroidogenic factor 1 (Sf1), also known as nuclear hormone receptor 5a1 (Nr5a1). These are glutamatergic neurons, some of which project to preautonomic regions of the brainstem (20), consistent with a possible role in controlling autonomic nervous activity. In line with studies of VMH lesions (6), Sf1^−/− mice display an abnormal organization of the VMN and are obese (21–23). The recent development of Sf1-Cre mice (24) helped elucidate the role of specific genes in the control of energy and glucose homeostasis. For instance, inhibition of glutamatergic transmission by inactivation of the synaptic vesicular transporter vGLUT2 gene prevented the normal glucagon response to hypoglycemia (25), thereby supporting a glucose-sensing role of these neurons.

Here we aimed at identifying the role of Gck in Sf1 neurons in body weight control, in the counterregulatory response to hypoglycemia, in the regulation of autonomic nervous activity by glucose, and finally in the glucose-controlled firing activity of these neurons. We found that absence of Gck from Sf1 neurons led to increased white fat mass, decreased lean mass, impaired glucagon secretion in response to insulin-induced hypoglycemia, and suppression of neuroglucopenia-induced sympathetic and parasympathetic activities. However, these phenotypes were only observed in female mice. Importantly, patch clamp analysis of Sf1 neurons revealed that the glucose responsiveness of GE and GI neurons in both male and female mice was unaffected by Gck inactivation.

Gck^lox mice were generated by homologous recombination in embryonic stem cells (Taconic Artemis, Köln, Germany). Exons 5 and 6 of the Gck gene were flanked by loxP sites, and the positive selection marker puromycin was inserted in intron 4 and flanked by Flp recombinase target sites (Fig. 1). The puromycin gene was removed by crossing mice with germline transmission of the recombined allele with Flp-deleter mice. The Gck^lox mice were then crossed with Sf1-Cre mice (Tg(Nr5a1-cre)7Lowl/J; Jackson Laboratories, Bar Harbor, ME) to generate Gck^lox/lox (control) and Sf1-Cre;Gck^lox/lox (Sf1Gck^−/−) mice. For some experiments, Rosa26tdtomato mice were bred with Gck^lox/lox mice to generate Sf1-Cre;Rosa26-tdtomato;Gck^+/+ and Sf1-Cre;Rosa-26tdtomato;Gck^lox/lox mice. Sf1-Cre mice do not have a phenotype regarding body weight gain, adiposity, glucose homeostasis, or food intake (24,26,27).

Mice were maintained at 23°C with 12-h dark/12-h light cycles and fed normal chow (Diet 3436; Provimi Kliba AG, Kaiseraugst, Switzerland) or a high-fat diet (HFD; 46% energy from fat; Safe Diets 235HF; SAFE, Paris, France).

Mice were perfused first with PBS then with 4% paraformaldehyde. The brains were dissected, incubated for 2 h in 4% paraformaldehyde at 4°C, incubated overnight in 30% sucrose, and frozen in ice-cold isopentane. Cryosections (20 µm) were placed on MMI MembraneSlides (Molecular Machines & Industries, Glattbrugg, Switzerland) and dehydrated with graded ethanol solutions and xylene. Laser capture microdissection was performed with the MMI CellManipulator Laser tweezer and MMI UV Cut Software. RNA was extracted from 10 sections collected in one MMI IsolationCap using the Arcturus Paradise Plus RNA Extraction Kit (Life Technologies Europe, Zug, Switzerland). The integrity of extracted RNA was assessed by equal amplification of the 3′ and 5′ untranslated regions of β-actin using specific primer sets (see Supplementary Table 1).

A Gck riboprobe complementary to exons 7 to 10 of Gck was produced by reverse transcription of an islet’s Gck cDNA subcloned in a pBluescript-KS+ plasmid (28). The riboprobes were labeled with digoxigenin (DIG RNA Labeling Mix; Roche Diagnostics AG, Rotkreuz, Switzerland).

Brains sections were prepared as described above. Hybridization was performed overnight at 60°C in 50% formamide/5xSSC, and the riboprobes were visualized by anti-digoxigenin-alkaline phosphatase and nitro-blue tetrazolium/5-bromo-4-chloro-3′-indolyphosphate (Roche Diagnostics AG) revelation.

Mice were killed by decapitation. Tissues were prepared and RNA extracted with Direct-Zol RNA MiniPrep kit (LucernaChem AG, Luzern, Switzerland). RT-PCR and quantitative RT-PCR were performed as described previously (29). Expression was normalized to β-actin or Gapdh mRNA. Primer sequences are listed in Supplementary Table 1.

Fat pads were fixed with 4% paraformaldehyde and embedded in paraffin; sections were stained with hematoxylin and eosin. Adipocyte size was measured with the ImageJ plugin Adiposoft (http://fiji.sc/Adiposoft).

Blood glucose, plasma insulin (mouse insulin ELISA; Mercodia AB, Uppsala, Switzerland), plasma glucagon (RIA; Millipore AG, Schaffhausen, Switzerland) and body composition measurements, as well as indirect calorimetry and glucose tolerance tests (glucose 2 g/kg i.p.), were performed as described previously (29). Plasma catecholamines were determined by liquid chromatography–tandem mass spectrometry (30).

For insulin-induced hypoglycemia experiments, mice fasted for 6 h received insulin (0.3–0.7 units/kg i.p.) and were bled by submandibular puncture under isoflurane anesthesia; plasma was stored at −80°C. Hyperinsulinemic-hypoglycemic clamps were performed as described elsewhere (31).

Firing activities were recorded as described previously (32). Unipolar nerve activity was recorded continuously over 30 min, and then 30 min after a single injection of 2DG (600 mg/kg i.p.). Data were digitized, amplified, filtered (cutoffs of 200 and 1,000 Hz), and monitored with Chart 8 (ADInstruments, Paris, France).

Six- to 10-week-old mice were anesthetized before decapitation, and acute brain sections (250 μm) were prepared. Electrophysiological recordings were performed as described previously (32). Whole-cell recordings were performed in current-clamp mode using a MultiClamp 700B amplifier (Molecular Devices, Berkshire, U.K.). Neurons with an access resistance exceeding 25 MΩ were excluded. A hyperpolarization step (−20 pA, 500 ms) was applied every 30 s to measure input resistance. Signals were filtered at 2 kHz, digitized at 10 kHz, and collected online using a pClamp 10 data acquisition system (Molecular Devices).

Values are reported as mean ± SEM. Data were analyzed with GraphPad PRISM software (GraphPad Software, Inc., San Diego, CA). Statistical significance was determined by 2-way ANOVA or the Student t test; P values <0.05 were considered significant.

All animal care and experimental procedures were approved by the Service Vétérinaire du Canton de Vaud (Switzerland).

Gene targeting in embryonic stem cells was used to create mice with loxP sites flanking exons 5 and 6 of Gck following the strategy delineated in Fig. 1A. Gck^lox/lox mice were then crossed with Sf1-Cre mice (24) to obtain Gck^lox/lox (control [Ctrl]) and Sf1-Cre;Gck^lox/lox (Sf1Gck^−/−) mice.

To verify correct inactivation of Gck in the VMN, we also generated Sf1-Cre;ROSA26tdtomato;Gck^+/+ and Sf1-Cre;ROSA26tdtomato;Gck^lox/lox mice (Fig. 1B and C). We used brain sections from these mice to dissect the VMN or ARC by laser capture microdissection and assess Gck mRNA expression by quantitative RT-PCR analysis. Gck mRNA was detected in the VMN and ARC of male and female control mice and in the ARC, but not the VMN, of male and female Sf1Gck^−/− mice (Fig. 1B and C). Detection of Gck mRNA by in situ hybridization confirmed its absence in the VMN of the Sf1Gck^−/− mice but showed normal expression in the ARC (Fig. 1D). Because there are reports of Sf1 (23) and/or Gck (33) expression in the pituitary and adrenal glands we measured Gck mRNA expression in these tissues and in the whole hypothalamus. We found similar Gck mRNA expression levels in the pituitary and adrenal glands of control and SF1Gck^−/− mice but a 60% reduction in the hypothalamus of Sf1Gck^−/− mice (Supplementary Fig. 1). Because Sf1 and Gck may also be present in the gonads, we assessed the fertility of knockout (KO) male and female mice. The average number of offspring per litter and time between litters was calculated from a total of 22 litters from crosses between Sf1-Cre:Gck^lox/lox males and Gck^lox/lox females (6.4 pups/litter; 22.2 days between litters) and 15 litters from crosses between Sf1-Cre:Gck^lox/lox females and Gck^lox/lox males (6.8 pups/litter; 22.5 days between litters) mice. This is similar to the data for C57Bl/6 mice (6 pups/litter and a gestation period of 21 days). Moreover, 6- and 28-week-old female Ctrl and KO mice had identical plasma estradiol concentrations (Supplementary Fig. 2). Thus Sf1-Cre dependent inactivation of Gck does not affect reproductive function.

Body weight gain was similar in male and female Ctrl and Sf1Gck^−/− mice fed normal chow for up to 20 weeks (Fig. 2A and B). Body composition was assessed by EchoMRI at 12 and 24 weeks of age. The percentage of body fat was increased by 1.4-fold at both time points in female but not male Sf1Gck^−/− mice compared with Ctrl mice (Fig. 2C and E). The percentage of lean body mass was reduced in female Sf1Gck^−/− mice by 3.8% and 10.6% compared with Ctrl mice at both time points, whereas no differences were observed in male mice (Fig. 2D and F). At the end of the experiment gonadal and inguinal fat depot weights had increased by 1.6-fold in Sf1Gck^−/− female mice compared with Ctrl mice, whereas no differences were observed in male mice (Fig. 2G and H). No differences were observed in interscapular brown adipose mass in Ctrl and Sf1Gck^−/− male or female mice (Fig. 2I). Similar data were obtained from a second cohort of male and female Ctrl and Sf1Gck^−/− mice.

Measurement of energy expenditure revealed no differences between Ctrl and Sf1Gck^−/− male or female mice (Fig. 2J and K). Analysis of gonadal (Fig. 3A) and inguinal (Fig. 3B) adipocyte sizes revealed a shift toward larger adipocytes in both white adipose tissue (WAT) depots in female Sf1Gck^−/− mice compared with Ctrl mice (Fig. 3C and D). No differences were observed in male mice (Fig. 3E and F). Quantitative RT-PCR analysis revealed that expression of lipolytic genes was unaltered in gonadal and inguinal WAT in both sexes (Fig. 3G and H; male data not shown). However, expression of the lipogenic genes acetyl-CoA carboxylase (Acc) and fatty acid synthase (Fas) were reduced in both gonadal and inguinal fat of Sf1Gck^−/− female mice (Fig. 3G and H). Fed and fasted glycemia were the same in male and female Ctrl and Sf1Gck^−/− mice (Fig. 4A and B), and plasma insulin concentrations showed no differences in female mice and only slightly higher values in male mice (Fig. 4C and D). Glucose tolerance assessed following intraperitoneal glucose injection in 27-week-old mice (Fig. 4E and F) was the same in male and female mice of both groups; the same results were found when the amount of glucose injected was normalized by lean mass instead of body weight (data not shown).

We next fed Ctrl and Sf1Gck^−/− mice an HFD from 6 until 24 weeks of age. Body weight gain was identical in male and female Ctrl and Sf1Gck^−/− mice (Supplementary Fig. 3A and B), as was body composition assessed by EchoMRI (Supplementary Fig. 3C and D). Moreover, glucose intolerance was similarly increased by an HFD in male and female Ctrl and Sf1Gck^−/− mice, but no differences occurred because of Gck deletion from the VMN (Supplementary Fig. 3E and F; compare with Fig. 4E and F). Thus, Gck expression in Sf1 neurons is involved in the control of WAT mass and adipocyte size when mice are fed a normal chow diet, but only in female mice.

We next assessed whether VMN Gck is involved in the control of glucagon secretion in response to insulin-induced hypoglycemia. Acute intraperitoneal injections of insulin lowered glycemia similarly in both female (Fig. 5A) and male Ctrl and Sf1Gck^−/− mice (Fig. 5C). One hour after insulin injection, plasma glucagon concentrations were increased 3.9-fold in female Ctrl mice and 1.7-fold in female Sf1Gck^−/− mice (Fig. 5B). In male Sf1Gck^−/− mice, insulin-induced hypoglycemia increased plasma glucagon concentrations 2.0-fold in Sf1Gck^−/− mice and 2.3-fold in Ctrl mice (Fig. 5D). In the same samples we tested plasma epinephrine and norepinephrine 1 h after insulin injection. No difference was observed between female Ctrl and Sf1Gck^−/− mice (Fig. 5E and F). We next tested glucagon secretion in female mice at the end of a hyperinsulinemic-hypoglycemic clamp (Fig. 5G and H). The glucose infusion rate during the clamp was similar for Ctrl and Sf1Gck^−/− mice (data not shown), but glucagon plasma concentrations were markedly less increased in Sf1Gck^−/− mice compared with Ctrl mice (Fig. 5H). The pancreatic glucagon content and the overnight-fasted plasma glucagonemia were the same in Ctrl and Sf1Gck^−/− female mice (Fig. 5I and J).

Thus, Gck expression in Sf1 neurons is required for the glucagon response to hypoglycemia, but only in female mice; this defective response is not related to differences in total pancreatic glucagon content nor fasted plasma glucagon concentrations, and it is not associated with changes in catecholamine concentrations.

The increase in lipid storage and the reduced glucagon response during hypoglycemia suggest an alteration in the control of autonomous nervous activity. Therefore we measured the firing activity of the parasympathetic and sympathetic nerves in the fed state and after 2DG-induced neuroglucopenia. Figure 6A and B shows representative tracings and Fig. 6C–F shows the quantification of the firing activities. Male Ctrl and Sf1Gck^−/− mice responded to neuroglucopenia by a 2.0-fold and 1.9-fold increase in parasympathetic activity (Fig. 6E) and a 1.8-fold and 2.0-fold increase in sympathetic activity (Fig. 6F), respectively. In Ctrl female mice, neuroglucopenia increased parasympathetic and sympathetic activities by 2.4-fold and 2.0-fold, respectively, whereas no induction was observed in female Sf1Gck^−/− mice (Fig. 6C and D). Thus Gck in Sf1 neurons is required for the induction of sympathetic and parasympathetic nervous activity by neuroglucopenia, but only in female mice.

These data suggest that the absence of Gck from Sf1 neurons of female, but not male, mice suppresses their glucose responsiveness. We therefore performed whole-cell patch clamp analysis of Sf1 neurons on acute brain slices of Sf1-Cre;ROSA26tdtomato;Gck^+/+ and Sf1-Cre;ROSA26tdtomato;Gck^lox/lox mice. As shown in Fig. 7 and Table 1, typical GE and GI neurons, identified by their firing activity assessed in the presence of 2.5 and 0.1 mmol/L glucose (34,35), were recorded in the VMN of both male and female Ctrl mice. Unexpectedly, glucose responsiveness of Sf1 GE and GI neurons was unaffected by the absence of Gck in VMN slices from male or female mice (Fig. 7 and Table 1). These data therefore indicate that Gck is not required for the glucose responsiveness of GE or GI neurons under the tested conditions in Sf1 neurons from either adult male or female mice.

Here we demonstrate that genetic inactivation of Gck in Sf1 neurons of the VMN led to increased WAT mass and adipocyte size, to impaired hypoglycemia-induced glucagon secretion, and to suppressed activation of parasympathetic and sympathetic nervous activity during neuroglucopenia. These effects were strictly sex-dependent, being present only in female mice. Although these observations suggest that Gck is required for glucose signaling in VMN glucose-responsive neurons of female mice, patch-clamp analysis revealed that the absence of Gck did not suppress the glucose responsiveness of GE and GI neurons in male or in female mice.

The VMN plays an important role in the control of glucose and energy homeostasis, which is, at least in part, a result of the presence of GE and GI neurons and their expression of Gck (15). Here we found that Gck inactivation in Sf1 neurons led to increased white fat mass and increased adipocyte size but no differences in the expression level of lipolytic genes or of Ucp1. At the same time, there was a reduction in lean mass so that body weight was not different between Ctrl and Sf1Gck^−/− female mice. On the other hand, Gck expression in Sf1 neurons was required for normal induction of glucagon secretion in female Sf1Gck^−/− mice in two modalities of insulin-induced hypoglycemia, that is, following intraperitoneal injection of insulin or at the end of a hyperinsulinemic-hypoglycemic clamp. In this last experiment, a surprising observation was that the lower glucagon response in female Sf1Gck^−/− mice compared with Ctrl mice was not associated with an increased glucose infusion rate. This was unexpected and may reflect perhaps a higher sensitivity to stress and catecholamine secretion during the clamp experiments in the mutant mice, or may partly be the result of the difference in body composition, since the Sf1Gck^−/− mice have reduced lean mass and increased fat mass, perhaps affecting glucose partitioning in these experiments. Thus Gck in Sf1 neurons is necessary for the proper response to hypoglycemia in female mice.

A common mechanism that can link Sf1 neurons to fat mass regulation and to glucagon secretion is the glucose control of autonomic nervous activity. White fat depots are innervated by sympathetic nerves (36,37) and pancreatic α-cells by both sympathetic and parasympathetic nerves (1,38). Hypoglycemia activates sympathetic nerve firing to increase adipocyte lipolysis through the activation of β3-adrenergic receptors and glucagon secretion through the activation of the β2-adrenegic receptor. On the other hand, hypoglycemia also increases parasympathetic activity to induce glucagon secretion through cholinergic and possibly also peptidergic receptor signaling (39). We therefore directly tested whether Gck inactivation in Sf1 neurons caused a defect in neuroglucopenia-activated sympathetic and parasympathetic nerve activities. Our data clearly show that the absence of Gck from these neurons prevented neuroglucopenia-induced activation of both branches of the autonomic nervous system in female Sf1Gck^−/− mice.

Although we initially hypothesized that the absence of regulation of both branches of the autonomic nervous system by neuroglucopenia in female Sf1Gck^−/− mice was caused by the impaired glucose responsiveness of GE and GI neurons in the VMN, our patch clamp analysis clearly showed that this was not the case. Although surprising with regard to previously published data suggesting that Gck is required for the glucose responsiveness of both GE and GI neurons (15), these data may be explained by the particular kinetics of Gck. Indeed, the classical definition of GE and GI neurons is based on their responsiveness over a glucose concentration range of 0.1 to 2.5 mmol/L (34), which should match that experienced by neurons protected by the blood-brain barrier (5). However, the glucose dose-response of Gck activity is sigmoidal, with a threshold at ∼5 mmol/L and a K_m ∼8 mmol/L (40). Thus Gck activity at 0.1 mmol/L glucose is very low and only marginally increased at 2.5 mmol/L glucose; it is not surprising that Gck plays no role in the response of GE or GI neurons as tested in acute brain slices. Nevertheless, here we investigated the role of Gck in Sf1 neurons only, and it cannot be excluded that this enzyme plays a more important role in acute glucose sensing in other neurons of the VMH (the VMN and ARC) (16,18,19).

Since female Sf1Gck^−/− mice have a clear phenotype, what, then, is the role of Gck if not in the acute control of GE and GI neuron firing activity? One possibility is that Gck is required for the glucose responsiveness of GE and GI neurons over a range of higher concentrations, more compatible with the K_m for glucose of Gck. Previous studies have identified classes of GE and GI neurons of the ARC that are activated over a high concentrations range (41) but such neurons have not been described in the VMN. Another possibility is that Gck is required for other aspects of neuron function or development. VMN neurons send long projections to preautonomic regions in the brainstem (20). Establishment of these projections may be directed by Gck expression, for instance, during the weaning transition, a period when glucose appears in the diet and induces adaptive processes to establish the normal control of glucose homeostasis (42); this period is also characterized by active hypothalamic neuron development (43). The weaning transition may therefore be a critical period when Gck-regulated functions are required for normal VMN neuron development.

The sex-specific impact of Gck inactivation in Sf1 neurons suggests that there is interaction between glucose sensing by Gck and sex-hormone signaling. Estradiol (E2) is known to affect recovery to normoglycemia after insulin-induced hypoglycemia (44). In the VMN, E2/estrogen receptor α receptor signaling involves AMPK inhibition (45) and is required to prevent the development of abdominal adiposity and to maintain normal energy expenditure (46). Because hypoglycemia detection in the VMN involves the activation of AMPK (14), this suggests a differential regulation of AMPK activity in male and female mice by E2. E2 concentrations fluctuate during the estrous cycle, with peak levels during proestrous. Here we did not study mice during a selected phase of the estrous cycle. Studying them when estradiol concentrations are highest or lowest, in ovarectomized mice, or in male mice receiving E2 may shed light on the role of sex hormones in Gck-dependent glucose sensing in Sf1 neurons of the VMH.

In summary, our study shows that Gck is required for the control of both branches of the autonomic nervous system during neuroglucopenia, for the control of body composition, and for hypoglycemia-induced glucagon secretion. These phenotypes are only observed in female mice. In addition, we show that Gck is not required for the activity of GE or GI neurons, indicating that the sex-specific physiological deregulations observed in female Sf1Gck^−/− mice cannot be explained by a defect in the firing activity of these classically defined glucose-responsive neurons. This implies that Gck may be involved in a yet undetermined function of glucose sensing in Sf1 neurons. Identifying this function will reveal new sex-specific, central glucoregulatory mechanisms, which may have important implications not only for a better understanding of physiological regulations but also possibly to identify new sites of intervention to control excess adiposity and to improve counterregulation in patients with hypoglycemia-associated autonomic failure.

Acknowledgments. The authors thank Anabela Da Costa, Salima Metref, Xavier Berney, and Wanda Dolci, all from the Center for Integrative Genomics, University of Lausanne, Lausanne, Switzerland, for excellent technical assistance. The authors also thank all current and previous members of the Thorens' group and the Mouse Metabolic Evaluation Facility at the University of Lausanne for their input and discussion.

Funding. This work was supported by grants to B.T. from the Swiss National Science Foundation (3100A0B-128657) and a European Research Council Advanced Grant (INSIGHT).

Duality of Interest. No conflicts of interest relevant to this article were reported.

Author Contributions. L.K.M.S. designed the research studies, conducted experiments, analyzed data, and wrote the manuscript. A.P. and G.L. conducted experiments and analyzed data. M.S.B. and D.B. conducted experiments. B.T. designed the research studies, analyzed data, and wrote the manuscript. B.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the 15th Servier-International Group on Insulin Secretion Symposium, St. Jean Cap Ferrat, France, 27–29 March 2014, and the Annual Dutch Diabetes Research Meeting, Oosterbeek, the Netherlands, 27–28 November 2014.