Phenotypic Characterization of MIP-CreERT^1Lphi Mice With Transgene-Driven Islet Expression of Human Growth Hormone

Cre-lox technology for tissue- and time-specific gene ablation is an invaluable tool in molecular biology, yet it is increasingly apparent that introduction or expression of the Cre-recombinase gene may produce confounding artifacts in different cellular and physiological contexts (1,2). In this regard, β-cells appear particularly sensitive. A few widely used transgenic mouse lines created using the promoters of pancreatic and duodenal homeobox 1 (Pdx1) or the rat insulin promoter (RIP) are reported to have significant transgene-mediated effects on insulin secretion and glucose tolerance (2–4). In addition, ectopic Cre-recombinase expression in the central nervous system (CNS) may cause non–β-cell related physiological effects (5,6).

A major challenge in the field has been to identify a specific promoter expressed exclusively in β-cells of the pancreas. Pdx1-driven gene expression turns on early in pancreatic development, leading to expression of transgenes throughout the pancreas (exocrine and endocrine) under the native promoter. This potential limitation for studies targeted toward particular endocrine cell types appeared largely circumvented by the creation of a tamoxifen-inducible Pdx1-CreERT line, which can be used to limit expression to β-cells in adult mice (7,8). The insulin 2 (Ins2) promoter (commonly referred to as RIP) is expressed predominantly in adult β-cells and has also been used to create both constitutive and inducible lines (9–14). Despite efforts to limit CNS expression using inducible systems, it was recently shown using an in vivo reporter system that significant Cre-mediated recombination in the CNS can still occur in many lines (6), confounding interpretation of results if floxed genes are also expressed at appreciable levels in the CNS.

In an effort to eliminate artifacts due to ectopic Cre-recombinase expression, the inducible mouse insulin promoter (MIP)-CreERT^1Lphi mouse line was created using ample upstream regulatory sequence from the mouse insulin 1 promoter (8.5 kb), a gene with limited expression in the CNS compared with the Ins2 gene (15). This promoter drives expression of Cre-recombinase cDNA fused to the hormone-binding domain of a mutant mouse estrogen receptor (Esr1*), preventing nuclear translocation of the recombinase in the absence of tamoxifen (16).

Given the history of confounding issues noted for similar mouse lines and the importance of carefully characterizing each new research model, we closely evaluated the metabolic phenotype and β-cell function of the MIP-CreERT^1Lphi mouse (referred to herein as MIP-CreERT) under normal physiological conditions and common in vivo models of diabetes. We also performed gene expression analysis and high-resolution microscopy on multiple regions of the CNS using a definitive reporter system. Our goal was to clearly define the specificity of Cre-recombinase expression and gene ablation to β-cells and determine whether MIP-CreERT mice exhibit abnormalities in whole-body glucose homeostasis, insulin secretion, and β-cell mass.

Strains used included mT/mG (17), MIP-CreERT^1Lphi (16), RIP-Cre^Herr (10), and Pdx1-Cre^Tuv (18) (see Supplementary Table 1 for details on genetic background). All mice were hemizygous for the Cre-recombinase allele. Data shown are from MIP-CreERT (C57BL/6J) unless otherwise indicated. All mice were male and gavaged at 6 weeks of age for 10 days with 100 mg/kg tamoxifen (Sigma) suspended in 0.05% methylcellulose/distilled water unless otherwise indicated. A minimum 2-week washout period after tamoxifen was provided. Mice were maintained on a 12-h dark/light cycle and given free access to chow and water. Experiments were performed according to approved protocols from the Institut de recherches cliniques de Montréal.

Thirteen-week-old mice were perfused with 10% formalin via cardiac puncture. Brains were fixed overnight in 10% formalin, followed by 10% sucrose/PBS for 24 h prior to freezing in OCT using isopentane and sectioning (7–10 μm). Cells were visualized by fluorescence microscopy at each spectra and images merged.

One-hour static incubations of isolated islets were performed as previously described (19). Briefly, batches of 10 islets were incubated at 37°C 2 × 20 min in Krebs-Ringer bicarbonate HEPES buffer (KRBH) solution containing 0.1% (w/v) BSA and 2.8 mmol/L glucose prior to incubation for 1 h in 2.8 mmol/L glucose, 16.7 mmol/L glucose, or 16.7 mmol/L glucose with 0.5 mmol/L oleate. Oleate was precomplexed with fatty acid–free BSA to a final molar ratio of 1:5 as previously described (20). Control conditions contained identical concentrations of BSA and oleate vehicle (50% [v/v] ethanol). Intracellular insulin content was measured after acid-alcohol extraction. For pancreatic insulin, half of the pancreas (closest to the spleen) was incubated overnight (O/N) (−20°C) in 5 mL acid-ethanol (1.5% HCl in 70% EtOH) prior to homogenization and refreezing. Thawed homogenate was centrifuged prior to neutralization with 1 mol/L Tris pH 7.5. Insulin was measured by immunoassay kit (PerkinElmer or Alpco). Insulin content was normalized to pancreas weight.

cDNA was synthesized by reverse transcription (Life Technologies) using RNA isolated from hypothalamic nuclei using Trizol (Invitrogen) or from islets (∼120) by RNeasy Mini kit (Qiagen). Gene expression was quantified using SYBR green (Life Technologies) and relative expression calculated by ΔΔct method, normalized to hypoxanthine-guanine phosphoribosyltransferase. Tail genomic DNA or islet cDNA was subject to PCR using primers flanking the start and stop codons of hGH. Cre-recombinase, interleukin-2, and actin primers served as internal controls. Primer sequences are listed in Supplementary Table 2.

Islets (∼150) or INS-1 cells expressing hGH (pUC plasmid) were lysed in radioimmunoprecipitation assay buffer containing protease inhibitors. Proteins were separated by SDS-PAGE and blotted with mouse anti-hGH (1:500; Abcam) or mouse anti–β-actin (1:10,000; Sigma).

After overnight recovery in RPMI (10% FBS, 1% Pen/Strep), 150 islets were equilibrated in 0.275-mL chambers for 1 h (KRBH, 5 mmol/L glucose) at a flow rate of 100 μL/min (Biorep Perifusion system). Islets were then perifused at 24-min intervals in experimental media (KRBH plus indicated conditions) and glucagon concentrations determined by ELISA (Mercodia) and expressed as percentage of baseline.

Mice were maintained on standard chow (Teklad diet 2018) or provided a high-fat/high-sucrose (HFHS) diet (cat. no. D12451i; Research Diets) starting at 8 weeks old. Physiological tests were performed at 13–17 weeks (chow diet) or 28–30 weeks of age (HFHS diet), as indicated, or within 30 days of streptozotocin (STZ) administration. Oral glucose tolerance tests (OGTT) were performed after a 16-h fast following oral administration of glucose diluted in water (1.5 g/kg for chow-fed/HFHS and 1 g/kg for HFHS/STZ mice). Insulin tolerance tests (ITT) were performed after injection of human insulin (1 unit/kg i.p. for chow-fed and 1.5 units/kg for HFHS-fed mice; Lilly) in 4-h-fasted mice. Fasted and refed glucose and insulin were measured after 16 h fasting and 2 h chow refeeding. Blood glucose or serum insulin was measured by tail vein using a standard glucometer (FreeStyle Lite, Abbott Diabetes Care) or the mouse ultrasensitive insulin ELISA (Alpco), respectively. Glucagon levels were measured after a 16-h fast in plasma containing aprotinin by glucagon ELISA (Mercodia).

A combination of HFHS diet and STZ was administered as previously reported (21), with some modifications. Briefly, 8-week-old mice were provided an HFHS diet for 3 weeks prior to administration of two doses of STZ (Bioshop) (day 0: 50 mg/kg; day 5: 75 mg/kg) in 100 mmol/L sodium citrate, pH 4.5, by intraperitoneal injection after a 4-h fast. Mice were maintained on an HFHS diet until sacrifice. When tamoxifen was not preadministered, a third injection of 100 mg/kg STZ was required to potentiate hyperglycemia in control mice. For single, high-dose STZ administration, chow-fed mice were injected with 150 mg/kg STZ at 10–13 weeks of age after a 4-h fast. Random blood glucose was measured as above at indicated times.

For islet area, 10–12 5-μm sections of pancreas (paraffin embedded), separated by at least 200 μm, were stained with hematoxylin-eosin. Islet density was calculated as the number of islets per section divided by section area. Islet mass was calculated as follows: (total islet area per mouse/total section area per mouse) × pancreas weight. Staining of paraffin sections was performed with guinea pig anti-insulin (1:100; Dako), mouse anti-glucagon (1:50; Sigma), mouse anti-serotonin (1:50; Dako), and rabbit Ki67, clone SP6 (1:100; Thermo Scientific) primary antibodies. Ki67 nuclear staining was quantified in each islet and expressed as percentage of total islet cell number. Islets containing <20 cells were excluded.

Data comparing one variable were analyzed by Student t tests, corrected for multiple comparisons (Sidak-Bonferroni) (*P < 0.05, **P < 0.01, ***P < 0.001). Normality was determined by Shapiro-Wilk test. Two-way ANOVA followed by post hoc analysis (Bonferroni or Fisher least significant differences test) determined significance of individual points. Analysis was performed using GraphPad Prism. Unless indicated, values are mean ± SEM.

A recent study reports that the insulin 1 (Ins1)/MIP promoter can drive transgene expression in cells isolated from the hypothalamus (22). Although transgenic mice using the Ins1 promoter do not appear to exhibit Cre-recombinase–mediated gene excision using staining/imaging of whole brain sections or quantitative PCR (qPCR) (23–25), these studies lack the single-cell resolution required to conclude whether Cre-recombinase activity is present in distinct neuronal populations. To sensitively detect Cre-mediated recombination in single neurons, we crossed MIP-CreERT mice with the global reporter mT/mG mouse line (17), which constitutively expresses red fluorescent protein (tdTomato) in all cells that are replaced by green fluorescent protein (GFP) after Cre-mediated recombination. Coronal and sagittal sections of brain from mT/mG:MIP-CreERT mice given tamoxifen showed no GFP+ neurons (Fig. 1A and B). No GFP+ cells were detected in hypothalamic (paraventricular nucleus, arcuate nucleus, ventromedial hypothalamus, dorsomedial hypothalamus) or hindbrain (dorsal motor nucleus of the vagus nerve, nucleus of the solitary tract, parabrachial nucleus) nuclei (Fig. 1C–G), while pancreata of corresponding mice exhibited a strong GFP signal in islet cells (Fig. 1H). These results provide strong evidence that Cre-recombinase does not drive excision of floxed genes in these CNS regions of MIP-CreERT mice in vivo.

Given concerns that Cre-recombinase expression or activity may lead to artifactual phenotypes (2,4), we next evaluated whole-body glucose homeostasis and in vivo insulin secretion in tamoxifen-treated MIP-CreERT mice on chow and HFHS diets. Compared with wild-type (WT) littermates, chow-fed MIP-CreERT mice showed no significant differences in weight, oral glucose tolerance, insulin tolerance, or fasting or refed glucose (Fig. 2A–D). There were also no differences in circulating insulin after fasting, refeeding, or oral or intraperitoneal administration of glucose (Fig. 2E–G). Likewise, HFHS feeding for 20 weeks did not reveal significant differences in weight, glucose tolerance, insulin sensitivity, or insulin secretion after an OGTT (Fig. 2H–K). However, there was a significant decrease in overnight fasting glucose and 5-h-fasted insulin in HFHS-fed MIP-CreERT mice compared with littermate controls (Fig. 2L), which, taken together with ITT data, may suggest a mild increase in insulin sensitivity.

On the chow diet, there were no significant differences in general islet architecture (insulin+ and glucagon+ cell distribution) or percentage of glucagon+ cells between WT and MIP-CreERT mice (Fig. 3A–C). However, there were trends toward increased basal glucose-stimulated insulin secretion (GSIS) and oleate potentiation of insulin secretion in primary islets and increased islet insulin content that did not reach statistical significance (Fig. 3D and E). Total pancreatic insulin content in MIP-CreERT mice was on average ∼27% higher than in WT controls (Fig. 3F). When plotted by size, islet numbers were significantly increased in a few size ranges (Fig. 3G) and there was a strong trend toward increased overall islet mass and density (Fig. 3H and I). Taken together, these results imply that MIP-CreERT mice had more pancreatic islets containing more insulin, leading to significantly higher pancreatic insulin content. However, this was not accompanied by a corresponding increase in insulin released, in agreement with in vivo data.

MIP-CreERT mice were also challenged with a combination of HFHS diet and STZ, a model of type 2 diabetes (21). Compared with controls, MIP-CreERT mice were significantly protected against resulting hyperglycemia (Fig. 4A) while maintaining similar weights (Fig. 4B). Of note, tamoxifen administration was not required for their resistance to HFHS/STZ (Fig. 4C) and a similar phenotype was observed in MIP-CreERT mice on an FVB background (Fig. 4D). In contrast, the Pdx1-Cre^Tuv mouse line, which also expresses Cre-recombinase in β-cells, remained sensitive to HFHS/STZ-induced hyperglycemia compared with littermates (Fig. 4E). MIP-CreERT mice were also significantly resistant to hyperglycemia induced by a single high dose of STZ (Fig. 4F), which was not the case for RIP-Cre^Herr mice (Fig. 4G). Taken together, these data suggest that expression and/or activity of Cre-recombinase in β-cells alone cannot explain the significant resistance to STZ-induced hyperglycemia in MIP-CreERT mice. HFHS/STZ-treated MIP-CreERT mice also had significantly improved glucose tolerance, yet unexpectedly, no differences in fasted or refed serum insulin levels (Fig. 4H and I), suggesting a non–insulin-dependent mechanism underlying normoglycemia.

It was recently demonstrated that hGH can be translated from the hGH mini-gene commonly used as a polyadenylation sequence in transgene constructs and suggested that this ectopic hGH impairs islet function (3). For determination of whether the MIP-CreERT transgene contained the entire hGH mini-gene, PCR amplicons from genomic DNA were produced using primers flanking the coding sequence of hGH (Fig. 5A) or the 3′ region of the Ins1 promoter to the 3′-UTR of the hGH mini-gene. The sequence contained the entire hGH gene from start to stop codon (Supplementary Fig. 1) (GenBank accession no. KR632635). In silico analysis of the sequenced MIP-CreERT transgene did not identify any previously characterized internal ribosomal entry sites or cryptic promoters (Supplementary Fig. 1). Various spliced variants of the hGH gene were detected in MIP-CreERT islet cDNA, with the major band corresponding to a fully spliced hGH transcript (Fig. 5B). As expected, the hGH cassette was amplified from CD11b-Cre mice (26) but not from Pdx1-Cre^Tuv (18) genomic DNA (Fig. 5A and B) or cDNA of Pdx1-Cre^Tuv or RIP-Cre^Herr (10) mice (not shown). MIP-CreERT primary islets expressed high levels of hGH protein (Fig. 5C) and mRNA (Fig. 5D) that were undetectable in controls. mRNA expression of tryptophan hydroxylase 1 (Tph1) and Tph2, enzymes that produce serotonin and are surrogate markers of hGH signaling via the prolactin receptor (3,27), were also increased in MIP-CreERT (Fig. 5E) but not RIP-Cre^Herr islets (not shown). This was consistent with a high level of serotonin immunoreactivity in MIP-CreERT islets (Fig. 5F), although the intensity of the signal varied greatly throughout the β-cell population.

Low-level Ins1 promoter activity is detected in the brain (15,22), suggesting that hGH may also be expressed in regions of MIP-CreERT CNS. Although we detected no Cre-mediated gene recombination in the CNS, transgene expression in this inducible system is not controlled by tamoxifen and hGH activity does not depend on drug-induced nuclear localization, as is the case for the Cre-recombinase. mRNA encoding Cre-recombinase and hGH were detected in both primary islets and microdissected arcuate nucleus (ARC) and ventromedial hypothalamus (VMH); however, levels in islets were 100- to 200-fold higher (Fig. 5G and H). Unlike in islets, there were no significant increases in Tph1 or Tph2 mRNA in ARC or VMH of MIP-CreERT mice compared with controls (Fig. 5I), suggesting that levels of hGH in hypothalamic nuclei may not be high enough to elicit physiological effects.

As MIP-CreERT mice did not develop severe hyperglycemia in response to STZ, we first hypothesized that they were resistant to the cytotoxic effects of STZ. In contrast to findings reported in Pdx1-Cre^Late mice (3), we saw no significant difference in Glut2 expression in MIP-CreERT islets, suggesting sufficient entry of the toxin into β-cells (Fig. 6A). In addition, islet mass and density were reduced to comparable levels in both genotypes (mass −78.4% ± 2.5% for WT and −72.7% ± 5.7% in MIP-CreERT; density −58.8% ± 4.8% for WT and −49.7% ± 7.4% in MIP-CreERT) after HFHS/STZ, consistent with the action of STZ to induce β-cell death (Fig. 6B and C). There was an overall increase in the total number of islets in the MIP-CreERT pancreata, with some fractions reaching statistical significance (Fig. 6D). However, when expressed as a percentage of total islet number (not shown), the proportion of islets in each size was identical, suggesting the increase in islet mass resulted from having more, not larger, islets. There were no differences in the incidence of Ki67+ islet cells 30 days after HFHS/STZ (Fig. 6E), and the proportion of these costaining positive for insulin was also similar (not shown), indicating that rates of islet/β-cell proliferation were equivalent.

Similar to mice on chow, there was a significant increase in total pancreatic insulin content in MIP-CreERT mice on HFHS/STZ (Fig. 6F). Furthermore, MIP-CreERT mice subject to HFHS/STZ had fewer α-cells and lower levels of serum glucagon after fasting (Fig. 6G and H), offering a plausible explanation for the significantly reduced glycemia despite similar circulating insulin (Fig. 4I). Consistent with this possibility, glucagon secretion from WT islets in response to low glucose (1 mmol/L) was modestly reduced after addition of serotonin (5-hydroxytryptamine [5-HT]) and rebounded upon serotonin washout, while 5-HT had no inhibitory effect on secretion in response to arginine (Fig. 6I). These data support our hypothesis that high levels of local islet serotonin may have a paracrine effect on α-cell function.

Our study demonstrates that under many physiological circumstances, the MIP-CreERT transgenic mouse line can be a useful tool for in vivo and in vitro studies investigating biological effects of β-cell–specific gene ablation. We report high specificity of tamoxifen-inducible Cre-recombinase activity to islets, with no detectable recombinase activity in multiple regions of the CNS. The MIP-CreERT transgene did not have significant effects on glucose homeostasis, insulin secretion, or insulin sensitivity in chow-fed mice, despite increased pancreatic insulin content and transgene-driven expression of hGH. However, use of these mice in certain contexts should be pursued with caution, as the mice exhibit abnormalities in fasting glycemia and insulin upon high-fat feeding and significant resistance to HFHS/STZ-induced hyperglycemia.

Increased insulin content and islet mass and resistance to STZ-induced hyperglycemia in MIP-CreERT mice resemble aspects of a phenotype recently ascribed to Pdx1-Cre^Late mice (3), and these similarities are likely attributable to ectopic expression of hGH from the transgene cassette—not Cre-recombinase. This is consistent with STZ resistance in Pdx1-Cre^Late (3) and MIP-CreERT mice but not Rip-Cre^Herr or Pdx1-Cre^Tuv mice, whose transgenes do not contain the hGH mini-gene. hGH can bind and activate the mouse growth hormone receptor, which can have significant physiological consequences including somatotrophic and lactotrophic effects (28). Consistent with our observations in MIP-CreERT mice, a two- to threefold increase in circulating growth hormone in mice has been shown to increase islet insulin content (29), and transgenic mice expressing hGH under the metallothionein promoter have increased islet size and number (30). Furthermore, culture of β-cells with recombinant hGH increases β-cell replication, insulin secretion, and insulin biosynthesis (31).

In β-cells, hGH can signal via the prolactin receptor (3). Prolactin increases β-cell proliferation and neogenesis during pregnancy and improves β-cell function (GSIS) at physiological doses; however, pharmacological levels can impair insulin secretion (32). Consistent with the former scenario, we observed increased insulin content and a modest trend toward enhanced GSIS in MIP-CreERT islets. In light of described effects of growth and prolactin hormones on β-cells, increased islet neogenesis in MIP-CreERT mice is probable and consistent with the higher islet numbers and total mass. Increased islet numbers were more pronounced 30 days post-HFHS/STZ and not accompanied by higher rates of islet cell proliferation; however, our data cannot rule out potential contributing effects of hGH on β-cell proliferation that may occur at other times during treatment and/or development. Given the role of the prolactin receptor in Pdx1-Cre^Late STZ resistance (3) and our data demonstrating that mice expressing Cre-recombinase without hGH remain sensitive to STZ, we predict that the effects on glucagon secretion, α-cell number, islet number, and insulin content in MIP-CreERT mice are most likely the result of the ectopic hGH expression. However, it remains possible that they occur independently due to other factors specific for this mouse, such as positional effects of transgene insertion.

Interestingly, we did not observe impairment in glucose tolerance or GSIS, as was reported for Pdx1-Cre^Late mice (3), supporting previous studies using MIP-CreERT mice on various genetic backgrounds (16,33–35). This difference may be due to transgene-specific differences in the level of hGH or secretion or other factors independent of hGH and serotonin action. Thus, it remains unclear whether ectopic hGH in β-cells is the underlying cause of cellular and metabolic abnormalities (including impaired β-cell function) reported for other Cre-recombinase transgenic lines (2–4).

Our high-resolution imaging using a definitive reporter (red/green) provides strong evidence that MIP-CreERT mice have no Cre-mediated recombination in the CNS in vivo, extending previous observations (6,23,25,36) and supporting the argument that inducible, Ins1/MIP promoter-driven Cre-recombinase expression is currently the best approach for β-cell–specific gene ablation (5). Nonetheless, we detected, albeit at very low levels compared with islets, transcripts containing both Cre-recombinase and hGH mRNA in the VMH and ARC. Cre-recombinase activity, in this system, depends on sufficient levels of recombinase expression and local tamoxifen to drive nuclear translocation and efficient excision of accessible LoxP loci. Thus, our data suggest that while not sufficient to mediate gene ablation, the Ins1/MIP promoter can still have low or leaky activity in the CNS, which may be of concern if used to drive expression of other biologically active proteins. However, unlike islets, we did not see significant induction of Tph1/2 gene expression, suggesting (at least using this surrogate marker of hGH activity) that transcripts generated in the brain may not be expressed at meaningful levels to have biologically relevant consequences. However, since Pdx1 and RIP promoters can drive significant transgene expression in the CNS (6), it is possible that the disruption of whole-body glucose homeostasis noted for Pdx1-Cre^Late (3) and certain RIP-Cre mouse lines (2) may, at least in part, be due to hGH expression in the brain. Newly generated knock-in mice (25) and transgenic (23) mice with constitutive or conditional Cre-recombinase expression driven by the Ins1 promoter are a promising alternative to the MIP-CreERT mouse to avoid potentially confounding effects of hGH expression on endocrine cell mass. However, efficiency of gene ablation in the β-cell population in these new conditional lines appears lower than achieved with the MIP-CreERT line in our hands. This may be due to differences in drug administration and dosing that will be revealed after further molecular and physiological characterization.

In contrast to the study of Brouwers et al. (3), MIP-CreERT mice did not have decreased Glut2 gene expression, implying that resistance to STZ-induced hyperglycemia was not mediated by inability of STZ to enter β-cells. Instead, we report differences in α-cell mass and glucagon secretion that may explain lower levels of glycemia in MIP-CreERT mice after HFHS/STZ. Ablation of glucagon receptor signaling in mice prevents STZ-induced hyperglycemia, despite equivalent reductions in β-cell mass and circulating insulin (37). We propose that the high levels of serotonin, which is secreted along with insulin (38), inhibit glucagon secretion from neighboring α-cells (39), preventing development of hyperglucagonemia and possibly affecting α-cell number. This hypothesis is supported by our perfusion data in islets, as well as previous demonstration that serotonin antagonists increase circulating glucagon concentrations in men (40).

It still remains a mystery how the hGH protein is translated from these transgenic constructs, highlighting that we have yet to fully grasp or predict all potential molecular and physiological changes caused by manipulation of the genome. Our findings highlight the importance of using appropriate controls, including mice expressing only the Cre-transgene. In conclusion, MIP-CreERT mice and other similar β-cell–specific lines, like most research tools, possess both strengths and caveats, yet remain an invaluable resource to dissect many aspects of β-cell biology when potentially confounding variables are considered within the context of appropriate control experiments.

Acknowledgments. The authors thank Dr. Louis H. Philipson for the MIP-CreERT^1Lphi mice. The pUC:hGH plasmid and the parental INS-1 cell line were generous gifts from Drs. Jacques Drouin (Institut de recherches cliniques de Montreal) and Chris Newgard. The authors give special thanks to Marie-Claude Lavallée, Courtney Reeks, Alexandra Abboud, Noura Kabbara, Naveen Khan, and Edwige Marcinkiewicz for technical assistance and Dawn B. Davis for helpful discussion and comments.

Funding. This study was supported by operating funds to M.E.K. from the National Institutes of Health (NIH) (R01-DK-102598) and American Diabetes Association (114BS115); to D.J.D. from the Canadian Institutes of Health Research (CIHR) (MOP 136942); to R.A.S. from CIHR (MOP111186); to V.P. from CIHR (MOP86454); to M.F. from CIHR (MOP133652), the J.A. DeSève Foundation, and Diabète Québec; to T.A. from CIHR (MOP115042); and to J.L.E. from the Canadian Diabetes Association (CDA) (OG3123843JE), Fonds de recherche du Québec - Santé (FRQS) (26721), Canadian Foundation for Innovation, and the J.A. DeSève Foundation. K.B. was supported by a Diabète Québec studentship. J.L. was supported by a CDA fellowship. J.C.N. was supported by a training grant (University of Wisconsin-Madison Institute on Aging: NIH T32 AG000213). D.J.D. holds a Canada Research Chair in Regulatory Peptides and the Banting and Best Diabetes Centre Novo Nordisk Chair in Incretin Biology. V.P. holds the Canada Research Chair in Diabetes and Pancreatic Beta-Cell Function. M.F. holds the Canada Research Chair in Bone and Energy Metabolism. CIHR and FRQS provided salary awards to T.A. and J.L.E.

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

Author Contributions. D.O., N.J., and J.L.E. conceived and designed the study and wrote the manuscript. D.O., N.J., L.B., J.E.C., K.B., J.L., G.P., V.B., J.C.N., H.K.B., R.J.F, C.M., S.S., M.E.K., D.J.D., R.A.S., V.P., M.F., T.A., and J.L.E. performed experiments and analyzed data. D.O., N.J., M.E.K., D.J.D., R.A.S., V.P., M.F., T.A., and J.L.E. reviewed and revised the manuscript. J.L.E. 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 in abstract form at the 75th Scientific Sessions of the American Diabetes Association, Boston, MA, 5–9 June 2015.