Alterations in β-Cell Calcium Dynamics and Efficacy Outweigh Islet Mass Adaptation in Compensation of Insulin Resistance and Prediabetes Onset

Obesity and insulin resistance are important risk factors for type 2 diabetes (T2D) (1). However, the majority of obese and insulin-resistant subjects do not develop hyperglycemia (2). This is attributed to the compensatory capacity of the islets of Langerhans, which meet the increased insulin demand by an elevated hormone output (3). Two processes have been implicated in islet compensation in response to obesity and insulin resistance: First, an increased functionality and insulin secretion rate of the individual β-cells, e.g., by enhancing glucose metabolism or insulin gene expression or translation (4–6). In addition, increased systemic insulin output can be achieved by a morphological enlargement of β-cell mass through proliferation (7,8), hypertrophy (5,8), or neogenesis (9,10). Whereas in rodents β-cell mass has been reported to increase substantially in response to insulin resistance (7,11), this seems to be less pronounced in humans (12–14).

Failure of β-cell compensation in insulin resistance and obesity is believed to be a cause of T2D. Both altered β-cell function and mass have been suggested to contribute to hyperglycemia. Decreased β-cell function was observed in islets isolated from organ donors with T2D (15) and animal models (16). In addition, rodent T2D models were shown to exhibit a reduction in β-cell mass (11,17–19). Data on alterations of β-cell mass in humans with T2D are inconsistent and vary from no significant changes up to a 63% decrease in comparison with subjects without diabetes (12,14,20,21). However, the relative contribution of β-cell mass and function to the maintenance of normoglycemia in insulin resistance and their distinct roles in the onset of insulin deficiency and hyperglycemia are unclear. It remains elusive whether the observed β-cell changes in T2D patients and hyperglycemic animal models are cause or consequence of the altered glucose homeostasis. To a large degree this is due to the inability to separately assess β-cell mass and function in vivo.

In this study, we monitored for the first time in vivo dynamics and mechanisms of islet and β-cell mass in relation to systemic glucose homeostasis during long-term diet-induced obesity and insulin resistance. This allowed us to assess the kinetics and relative contribution of β-cell mass during the compensation phase and prediabetes stage of diabetes pathogenesis. Our results show that the functional response of β-cells dominates over islet mass adjustment in compensation of insulin resistance and during the development of prediabetes. To study the underlying mechanism of functional adaptation, we monitored intracellular calcium (Ca²⁺_i) dynamics of β-cells longitudinally in vivo. This revealed increased calcium efficacy as a crucial adaptive mechanism of β-cell function, which was dependent on Epac signaling in isolated islets. Meanwhile, Ca²⁺_i dynamics in vivo showed an early decline preceding the onset of prediabetes. Therefore, our study provides novel insight into the pathogenesis of T2D and potential therapeutic targets.

All experiments were conducted in accordance with the German Animal Welfare Act and approved by the Committee on the Ethics of Animal Experiments of the State Directory of Saxony. C57BL/6N albino mice (B6N-^Tyrc-Brd/BrdCrCrl; Charles River Laboratories) were used as recipients for islet transplantation. Mice expressing green fluorescent protein under control of the mouse insulin I promoter (MIP-GFP) (22) and Pdx1CreER-GCaMP3 mice, both on C57Bl/6J background (The Jackson Laboratory), were used as islet donors. Pdx1CreER-GCaMP mice were generated by crossing Pdx1CreER mice (23) with homozygous GCaMP3 mice (The Jackson Laboratory) (24). For induction of β-cell–specific GCaMP3 protein expression, Pdx1CreER-GCaMP3 mice were injected with tamoxifen (3 × 4 mg s.c. in corn oil over 5 days) 2 weeks before islet isolation.

Recipient mice received either standard normal diet (ND) (ssniff) or rodent high-fat diet (HFD) with 60% of kcal as fat (D12492; Research Diets) for 17 weeks before switching back to ND in the HFD group. Body weight and intraperitoneal glucose tolerance tests (IPGTTs) were assessed at indicated time points. For IPGTTs, mice were fasted for 6 h and injected with 2 g glucose/kg body wt i.p. Blood glucose was measured at 0, 30, 60, and 120 min after glucose injection using a glucometer (Accu-Chek Aviva, Roche). Plasma insulin was measured at 0 and 30 min using a mouse Ultrasensitive Insulin ELISA kit (ALPCO). Prediabetes was defined as significantly increased 2-h blood glucose values at unchanged nonfasting blood glucose levels.

Islet isolation and transplantation to the anterior chamber of the eye were performed as previously described (25–28). Transplanted islets were allowed to engraft fully for at least 4 weeks before the start of any in vivo experiments. In vivo imaging was performed as previously described (26). Briefly, islet recipients were intubated and anesthetized by 2% isoflurane in 100% oxygen. A drop of 0.4% pilocarpine (Pilomann; Bausch & Lomb, Rochester, NY) in saline was placed on the cornea shortly before imaging to limit pupil dilation and iris movement. Animals were fixated and kept on a heating pad during the imaging procedure. Repetitive in vivo imaging was performed at indicated time points on an upright laser scanning microscope (LSM780 NLO; Carl Zeiss, Jena, Germany) with a water dipping objective (W Plan-Apochromat 20×/1.0 DIC M27 75 mm; Carl Zeiss) using vidisic eye gel (Bausch & Lomb) as immersion. The total volume of transplanted islets was assessed by detection of 633 nm laser backscatter, which allows assessment of islet morphology based on its light-scattering properties (27). Total volume of each islet was calculated using surface rendering after three-dimensional reconstruction of collected image stacks (Imaris 7.6; Bitplane AG, Zurich, Switzerland). β-Cell volume in MIP-GFP islets was assessed by two-photon laser excitation of GFP at 910 nm and detection at 500–550 nm using a nondescanned gallium arsenide phosphate detector. β-Cell volume was then quantified by automatic surface rendering from median filtered Z stacks in Imaris 7.6 (Bitplane AG). Qtracker 655 (0.4 μmol/L in 100 µL PBS; Life Technologies) was injected into the tail vein to visualize blood vessels. Qtracker were excited by two-photon laser at 910 nm and detected at 635–675 nm. Vessel volume was calculated by automatic surface rendering. Vessel network length and vessel diameter were measured by filament tracing within the vessel surface using Imaris software.

Pdx1CreER-GCaMP3 islet recipient mice were fasted overnight and anesthetized with an injection of a mixture of fluanisone (12.5 µg/g body wt i.p.; CHEMOS, Regenstauf, Germany), fentanyl (0.788 µg/g body wt i.p.; Hameln Pharma Plus, Hameln, Germany), and midazolam (12.5 µg/g body wt i.p.; Ratiopharm, Ulm, Germany). The anesthetized mice were intubated and respiration was maintained by use of a small animal ventilator with room air (270 µL stroke volume, 250 strokes/min; Hugo Sachs, March-Hugstetten, Germany). Animals were fixed and positioned as described above. In vivo imaging was performed using confocal laser scanning microscopy. The volume of islets was assessed by detection of 633 nm backscatter laser light. The functional index of islets was calculated by dividing stimulated plasma insulin by fold changes in islet mass at corresponding time points in the same mice. GCaMP3 GFP was excited at 488 nm and detected at 468–607 nm. Z stacks of the islet were acquired at 1.5-µm intervals, and time-series recording of GCaMP3 GFP signal from the same islets was carried out with an image sampling rate of 2 s. After acquisition of a baseline recording for 5 min, glucose was injected (1 g/kg body wt i.v.) via the tail vein. Changes in β-cell Ca²⁺_i, as reflected by changes in GCaMP3 GFP fluorescence intensity, were recorded for 30 min after glucose injection. After the recording, mice were maintained on ventilator breathing until autonomous respiration was restored. Z stacks and time series were processed using Imaris 8.1 software (Bitplane AG). The GCaMP3-expressing β-cell area was defined by applying a threshold acquired in the GFP-negative area. Mean GCaMP3 GFP fluorescence per β-cell area was calculated for each frame and normalized to the minimal value in the baseline. Finally, total β-cell Ca²⁺_i dynamics was quantified by calculating the area under the curve of the fluorescence trace for 5 min before glucose injection (basal Ca²⁺_i dynamics) and 30 min after glucose injection (stimulated Ca²⁺_i dynamics). In vivo calcium efficacy was calculated by dividing stimulated plasma insulin by the product of stimulated Ca²⁺_i dynamics per minute and fold changes in islet mass at corresponding time points in the same mice. First-phase peak amplitude was defined by the peak intensity over mean basal intensity within 5 min after glucose injection.

After overnight culture in standard RPMI 1640 medium, islets from female C57Bl/6J mice were handpicked and cultured in 24-well plates in culture medium with a 0.1 mmol/L BSA, 8 mmol/L glucose, and 0.5 mmol/L BSA–conjugated palmitate (G₈P_0.5). Control islets were cultured in RPMI 1640 media in the presence of 5.5 mmol/L glucose without palmitate (G_5.5P₀). The Epac-specific inhibitor ESI-09 (1–5 μmol/L; Sigma-Aldrich) was applied in both basal culture and during glucose stimulated insulin secretion when indicated. Basal insulin secretion was assessed from 24-h culture medium and corrected for release per hour. For assessment of stimulated insulin release, islets were placed for 1 h in their respective culture medium, containing elevated glucose concentrations (25 mmol/L). Islets were lysed in acid/ethanol buffer for insulin content. Total insulin content was defined by the sum of insulin secreted into medium upon stimulation and insulin content in the lysate. Collected medium and islet lysate were stored at −20°C until insulin content was measured using a mouse Ultrasensitive Insulin ELISA kit (ALPCO).

For in vitro imaging of cytosolic free calcium dynamics in β-cells, Pdx1CreER-GCaMP islets were handpicked after overnight culture and embedded in fibrin gels on coverslips according to a previously published protocol (29). Briefly, fibrin gels were prepared by mixing 3 µL Hanks’ balanced salt solution (HBSS) with 1 µL human fibrinogen (10 mg/mL in HBSS; Sigma-Aldrich), after which six islets were placed individually in the gel. Fibrinogen polymerization was induced by adding 1 µL human thrombin (50 units/mL in HBSS; Sigma-Aldrich). Subsequently, the gel-embedded islets were cultured for 24 h in indicated conditions. Calcium imaging was performed in a custom-made perifusion system (30). Briefly, coverslips with gel-embedded islets were moved into a temperature-controlled perifusion chamber where the temperature was set at 37°C. In vitro imaging was performed in the above-described laser scanning microscope system. GCaMP3 GFP was excited at 488 nm and detected at 493–598 nm. Time-series recording of the GCaMP3 GFP signal from the same islets was carried out with an image-sampling rate of 2 s. A basal recording was acquired for 3 min when the islets were perifused in their respective culture medium. For stimulation, islets were perifused with their respective culture medium, containing elevated glucose concentrations (25 mmol/L). Changes in β-cell Ca²⁺_i were recorded for 25 min after stimulation. Mean GCaMP3 GFP fluorescence per β-cell area and total β-cell GCaMP3 GFP fluorescence was processed and quantified using the same protocol as for the in vivo imaging. Basal and stimulated Ca²⁺_i dynamics are displayed as area under the curve per minute in basal and stimulated conditions, respectively. Calcium efficacy was calculated by dividing stimulated insulin by stimulated Ca²⁺_i dynamics per minute in corresponding experiments.

Fractional pancreatic β-cell area was assessed in cryosections (10 µm) stained for insulin (1:200 dilution; Dako, Hamburg, Germany), glucagon (1:500; Merck Millipore, Darmstadt, Germany), and DAPI (2.5 µg/L; Sigma-Aldrich). Immunostaining was visualized by Alexa Fluor 488 and Alexa Fluor 633 secondary antibodies (1:200; Life Technologies). Images were acquired in a slide scanner (Axio Scan.Z1; Carl Zeiss). Quantification of immunohistochemistry was done manually using Fiji software. Fractional β-cell area was calculated from at least 40 pancreas sections (with a minimal distance of 150 µm between sections) per mouse.

Data are expressed as mean ± SEM. Statistics were analyzed using Prism 6 (GraphPad Software, San Diego, CA) or SPSS 21 (IBM, Armonk, NY). For proper acknowledgment of the two sources of correlation that are inherent to the repeated longitudinal measurements, longitudinal in vivo data were analyzed by linear mixed models (26). Significant differences are indicated at P < 0.05.

We transplanted reporter islets for in vivo imaging into the anterior chamber of the eye of albino C57Bl/6N mice. After engraftment, recipient mice were fed with ND (11% kcal from fat) or HFD (60% kcal from fat) for 17 weeks, followed by 2 weeks of ND. HFD mice showed continuously increasing body weight (37.3 ± 0.6 vs. 26.2 ± 0.5 g for HFD vs. ND, respectively, at 17 weeks) (Fig. 1A). Furthermore, fasting blood glucose levels were slightly, but significantly, increased at several time points during the study (Fig. 1B). Fasting and stimulated plasma insulin levels in response to an IPGTT increased during HFD, reaching a plateau at 9 weeks (fasting 0.68 ± 0.11 vs. 0.16 ± 0.06 ng/mL for HFD vs. ND at 9 weeks; stimulated 1.28 ± 0.16 vs. 0.34 ± 0.06 ng/mL for HFD vs. ND at 9 weeks) (Fig. 1C and D). Also, the fasting and stimulated plasma insulin levels in relation to the given glucose concentration were increased in HFD, indicating an increased sensitivity of β-cells to glucose (Fig. 1E and F). However, despite the elevated insulin release, glucose tolerance continuously deteriorated in HFD mice (Fig. 1G). In addition, the 2-h glucose tolerance test value was significantly higher after 9 weeks of HFD feeding, indicating a stage of prediabetes in these animals (Fig. 1H).

Islets engrafted in the anterior chamber of the eye of the above-described mice allowed longitudinal assessment of total islet volume by backscatter laser light during long-term ND and HFD feeding. The acquired backscatter volume included all endocrine cells and the vascular network of an islet (Fig. 2A and B). While the total volume of islets in ND-fed mice remained unchanged, islets within HFD-fed mice showed a slow but continuous increase in size, doubling in volume after 16 weeks of HFD exposure (1.97 ± 0.14-fold vs. 1.07 ± 0.08-fold increase for HFD vs. ND at week 16) (Fig. 2A and C).

Using MIP-GFP reporter islets in the anterior chamber of the eye during HFD demonstrated that the observed islet mass enlargement was due to β-cell expansion (1.88 ± 0.17-fold increase at week 16) (Fig. 3A and B). This was the result of an increase in β-cell number (1.92 ± 0.15-fold increase at week 16) (Fig. 3C), whereas no significant change in the individual β-cell size was observed (Fig. 3D). Intraislet vessels adapted to islet growth and contributed to total islet volume increase by a lengthening of the vascular network (2.11 ± 0.15-fold increase at week 16) (Fig. 3A and E), thereby maintaining a constant fractional vessel volume within the growing islets (data not shown). In addition, intraislet vessel diameter increased in response to HFD presumably to match the increased secretory activity (8.79 ± 0.09 vs. 8.22 ± 0.11 μm for HFD vs. ND at week 16) (Fig. 3F). Pancreatic β-cell area showed a similar, circa twofold, increase after 16 weeks of HFD feeding in our setting (Fig. 3G and Supplementary Fig. 1), verifying the validity of the anterior chamber of the eye platform to study islet biology during HFD.

Because the kinetics of islet mass expansion (Fig. 2C) and rising plasma insulin levels (Fig. 1D) within the same animal showed obvious differences, we calculated an approximate index of the relative contribution of β-cell function by dividing stimulated plasma insulin by the increase in islet cell mass. β-Cell functional index peaked within 1 week of HFD feeding with a threefold increase over ND (3.2 ± 1.0 vs. 1.1 ± 0.2 for HFD vs. ND) (Fig. 4). Thereafter, it remained elevated over ND throughout the HFD feeding period at slightly reduced levels (2.6 ± 0.6-fold, 2.8 ± 0.3-fold, and 2.5 ± 0.5-fold at weeks 4, 8, and 16, respectively) (Fig. 4). This revealed that β-cell function showed an immediate strong compensatory response to HFD feeding, which in contrast to islet mass did not further increase during the course of prediabetes development.

We examined the role of Ca²⁺_i dynamics in the functional compensation of β-cells during prolonged HFD feeding by longitudinal in vivo imaging. To this end, we transplanted Pdx1CreER-GCaMP3 islets into albino C57Bl/6N mice. Prior to islet isolation and transplantation, CreER-LoxP–mediated recombination of GCaMP3 expression was induced by tamoxifen application to the islet donor. As previously reported in neurons (24), Ca²⁺_i dynamics in β-cells was assessed by monitoring GFP fluorescence of the GCaMP3 reporter in response to an intravenous glucose challenge at indicated time points before, during, and after long-term ND or HFD feeding (Fig. 5). Under control conditions, β-cells showed little or no basal Ca²⁺_i dynamics (Fig. 5A, panel 1). Upon glucose injection into the tail vein of the recipient mouse, β-cells displayed a strong peak of Ca²⁺_i dynamics (Fig. 5A, panel 2), followed by a lower plateau and/or oscillations in Ca²⁺_i (Fig. 5A, panels 3–5). Islets displayed a continuous plateau, regular oscillations, or a mix of both after the initial peak. The respective pattern of Ca²⁺_i dynamics of an islet was comparable at the various recording time points and was synchronized throughout the image plane. Islets in ND mice showed few alterations in Ca²⁺_i dynamics during repetitive imaging over 19 weeks (Fig. 5B). However, in response to HFD feeding GCaMP3 GFP fluorescence traces displayed considerable changes in basal and stimulated Ca²⁺_i dynamics (Fig. 5C).

Quantification revealed that minimum values of GCaMP3 GFP fluorescence intensity were comparable between islets in ND and HFD mice throughout the study. This suggests that there were no major changes in expression of the fluorescent reporter or resting Ca²⁺_i levels. However, basal islet Ca²⁺_i dynamics, assessed as area under the curve of the GCaMP3 GFP fluorescence trace prior to intravenous glucose injection, continuously increased in islets of the HFD-fed mice (7.93 ± 0.45 vs. 6.25 ± 0.27 arbitrary units for HFD vs. ND at week 16) (Fig. 6A). This parallels the observed rising levels of basal plasma insulin (Fig. 1C), which are most likely the result of elevated basal glucose levels (Supplementary Fig. 2A) and increasing glucose sensitivity (Fig. 1E). In contrast, stimulated Ca²⁺_i dynamics in response to intravenous glucose decreased early and remained significantly lower in the HFD group (42.05 ± 2.32 vs. 52.83 ± 2.60 arbitrary units for HFD vs. ND at week 16) (Fig. 6B). This was associated with a reduced first-phase peak amplitude of calcium activity after glucose injection (Fig. 6C), while second-phase oscillation amplitude and frequency were highly variable among islets of both groups and showed no significant effect by HFD feeding. Increased β-cell function without elevation in stimulated Ca²⁺_i dynamics suggested an augmentation of insulin release by amplification of calcium-induced secretion in vivo to overcome impaired glucose tolerance (Fig. 1G and Supplementary Fig. 2B). This was illustrated by the calculated enhanced efficacy of calcium to induce insulin release during HFD feeding (Fig. 6D).

To reveal the mechanism underlying amplified calcium efficacy, we mimicked the acute functional compensation by 24-h culture of isolated mouse islets in 8 mmol/L glucose plus 0.5 mmol/L palmitate (G₈P_0.5). This condition resembled the slightly elevated glucose levels and the presence of increased free fatty acids observed during HFD in vivo. The control islets were cultured in G_5.5P₀ to resemble ND feeding. Compared with control, G₈P_0.5 islets had significantly higher basal and glucose-stimulated insulin secretion (Fig. 7A). At the same time, G₈P_0.5 islets showed significantly higher basal Ca²⁺_i dynamics in vitro, while glucose-stimulated Ca²⁺_i dynamics was slightly but significantly reduced (Fig. 7B). This revealed a significantly higher calcium efficacy of G₈P_0.5 islets upon stimulation (Fig. 7C), identical to what we observed during HFD in vivo (Fig. 6D). Previous data indicate cAMP mediated signaling via the exchange protein activated by cAMP (Epac) to potentiate insulin secretion and mediate the β-cell response to increased metabolic demand (31,32). We therefore assessed the role of Epac in the observed elevated calcium efficacy in our conditions. ESI-09, an Epac-specific inhibitor, dose-dependently (1, 3, and 5 μmol/L) reduced glucose-stimulated insulin secretion when supplemented in G₈P_0.5 islet culture and stimulation, while basal insulin secretion was not significantly affected (Fig. 7D). On the other hand, 5 μmol/L ESI-09 had no effect on basal or stimulated calcium dynamics (Fig. 7E), revealing a significant inhibition of calcium efficacy by 5 μmol/L ESI-09 (Fig. 7F) and indicating a crucial role of Epac in the amplification of insulin secretion by increased calcium efficacy.

Interestingly, after 17 weeks of HFD, a subsequent 2-week period of ND was sufficient to reverse the observed phenotype. Body weight decreased but remained significantly elevated (Fig. 1A). Fasting blood glucose and glucose tolerance returned to control values (Fig. 1B, G, and H). Also, basal and stimulated plasma insulin levels dropped markedly (Fig. 1C and D). Furthermore, islet mass enlargement stopped when HFD mice were switched back to ND (Fig. 2C). Finally, β-cell functional index (Fig. 4) as well as basal and stimulated calcium dynamics (Fig. 6A and B) was restored to prediet values.

In our study, we assessed for the first time islet mass and β-cell function longitudinally throughout the progression from normal glucose tolerance, through compensation, to prediabetes. For this purpose, we correlated systemic glucose homeostasis with longitudinal in vivo imaging of transplanted islets of Langerhans during diet-induced obesity. Although we cannot exclude potential effects of the different local environment on intraocular islets, our own data and previous studies suggest that the behavior of islets in the anterior chamber of the eye closely resembles that of endogenous pancreatic islets (26,28,33–35).

Our results reveal that HFD feeding, a model of early diabetes pathogenesis (36), leads to distinct dynamics of β-cell mass and function preceding the onset of diabetes. Initially, enhanced β-cell function compensates insulin resistance by elevated plasma insulin levels in the absence of any significant islet mass increase. During a second phase, gradual islet mass expansion contributes to rising plasma insulin levels in combination with elevated β-cell function. However, despite a continuous islet mass increase, ongoing HFD feeding subsequently leads to prediabetes as a result of insufficient β-cell function. In our study, islet mass is not decreased at the time of prediabetes onset but, on the contrary, is still increasing. This supports the hypothesis that β-cell function and not islet mass decline is the crucial islet-related mechanism in early T2D pathogenesis. This is in line with the observation that the majority of known T2D susceptibility genes are associated with β-cell function (37). Therefore, observed reduced β-cell mass in individuals with T2D might be a result of stress-induced cell death or dedifferentiation after onset of hyperglycemia (20). However, our results do not rule out a potential role of β-cell mass at later stages of diabetes pathogenesis or the influence of diminished inherent β-cell mass for diabetes development (38).

In our investigation of the mechanisms underlying functional compensation, our findings suggest that increasing fasting plasma insulin levels in HFD are the result of enhanced β-cell function illustrated by the rising basal Ca²⁺_i dynamics. Higher fasting insulin has also been observed in other mouse models of T2D (39) and is most likely due to an increased glucose sensitivity of β-cells in the presence of elevated free fatty acid levels (40). Conversely, whereas plasma insulin levels were increased in response to glucose, we found stimulated Ca²⁺_i dynamics to be reduced under HFD in vivo, which was associated with a decreased first-phase response. In dispersed cells or isolated islets, diminished stimulated Ca²⁺_i dynamics was observed in several models of β-cell stress and linked to altered gene expression of Ca²⁺-channel subunits (41), impaired glucose metabolism (42), or abnormal Ca²⁺_i handling (43). Importantly, blunted stimulated Ca²⁺_i dynamics might be directly related to the increased basal Ca²⁺_i, as both have been observed to occur simultaneously in pathophysiological conditions (39,44). Our observation that stimulated insulin release is elevated despite reduced Ca²⁺_i dynamics suggests that functional compensation in HFD is accomplished by the amplifying pathway of insulin secretion and augmented calcium efficacy (45–47). This is probably mediated by free fatty acids, which rise during HFD feeding (48) and have been shown to potentiate glucose-stimulated insulin secretion (49). Mimicking HFD feeding conditions in vitro by culture of isolated islets in palmitate and slightly elevated glucose, we found Epac to be involved in amplified calcium efficacy. These findings underline the potential of lipid-derived signals to activate the amplifying pathway (50) and the role of Epac in elevated β-cell function during HFD (31).

Interestingly, islet mass expansion stopped but did not reverse after cessation of HFD feeding. However, given the observed slow adaptation of β-cell mass to the changed metabolic environment, this might take place within a longer time frame than assessed in our study. In contrast, HFD-induced alterations in glucose homeostasis and Ca²⁺_i dynamics were quickly reversed by ND feeding. This is in agreement with the rapid recovery of mouse and human islets isolated from subjects with diabetes (51,52) and might explain diverse findings in isolated islets in vitro in comparison with the in vivo results presented here (5).

In conclusion, our data provide the first in vivo evidence that β-cell function outweighs the adaptive response of islet mass during compensation of insulin resistance and progression to prediabetes. Thereby, the amplifying pathway of insulin secretion, via increased calcium efficacy and Epac signaling, plays a crucial role in the β-cells’ capacity to compensate for the higher insulin demand. At the onset of prediabetes, β-cell function is still enhanced but shows signs of weakened compensation. Whether this is primarily the result of reduced stimulated Ca²⁺_i dynamics or of additionally decreasing calcium efficacy requires further investigation. While caution has to be used when translating mouse data to the human situation, it is likely that due to the low proliferative potential of adult human β-cells (13) the role of β-cell function in human diabetes pathogenesis will be even more significant. Thus, we believe that T2D prevention and treatment should aim at protecting and improving β-cell function by addressing calcium dynamics and efficacy.

Acknowledgments. The authors thank Katharina Hüttner, Chrissy Kühn, Angela Hartke, Alin Pfennig, and Claudia Möx, the Paul Langerhans Institute Dresden, for excellent technical assistance.

Funding. This work was partly supported with funds from the Emmy Noether Program of the German Research Foundation (DFG) (www.dfg.de); the DFG–Center for Regenerative Therapies Dresden, Cluster of Excellence (www.crt-dresden.de); and DFG SFB/Transregio 127; and the German Ministry for Education and Research (www.bmbf.de) to the DZD and to the Competence Network Diabetes Mellitus.

Duality of Interest. This work was also supported with funds from Boehringer Ingelheim Pharma GmbH & Co. KG. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. C.C. and S.S. designed the study, analyzed data, and wrote the manuscript. C.C. performed the experiments. H.C., C.M.C., J.A.C., S.R.J., J.S., and I.U. analyzed data and provided intellectual input. S.S. provided funding. S.S. 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.