Manganese-Mediated MRI Signals Correlate With Functional β-Cell Mass During Diabetes Progression

Both type 1 and type 2 diabetes are characterized by a loss and/or dysfunction of β-cells (1–5). As long as physiological insulin secretion is maintained by the β-cell, treatment toward the preservation of normoglycemia is easier to achieve. Thus, it is important to detect β-cell failure at a very early stage of the disease or even during the β-cell compensation phase. β-Cell compensation to maintain normoglycemia can be achieved in two ways: by increasing mass or increasing function (or both). Functional β-cell compensation can occur until up to a 65% β-cell loss; only at further β-cell mass (BCM) reduction was insulin secretion found to decline in patients (6). This analysis from autopsy pancreases highlights that the absolute BCM measure in an individual may not provide sufficient information on the status of diabetes progression and that the analysis of the functional BCM is essential for the evaluation of the diabetes risk.

Today, monitoring of functional BCM is achieved by measuring insulin and C-peptide secretion during a glucose as well as arginine tolerance test (7–9). Retrospective studies using human pancreata from autopsy show a strong correlation of fasting blood glucose levels with BCM and a reduction in BCM already before the diagnosis in subjects with impaired fasting glucose levels (1,10). The ideal time of diagnosis is the onset of diabetes at a stage when the functional BCM has just changed.

The noninvasive measurement of functional BCM has enormous potential for diagnostics but is rather challenging to achieve, partly because a ligand for specific β-cell labeling is currently not available (11–14). It is not possible to measure the actual mass of functional β-cells in vivo (15). MRI using manganese (Mn²⁺) shows such monitoring of β-cell functionality in cell culture (16). Similar to Ca²⁺, Mn²⁺ is taken up by glucose-activated β-cells, resulting in a robust signal increase in glucose-stimulated rodent β-cell lines and in islets (16,17). The uptake of Mn²⁺ is controlled by voltage-gated Ca²⁺ channels (18), and Mn²⁺ accumulates in the cytoplasm, primarily in the perinuclear region (19). Mn²⁺ uptake is glucose dependent and can be used in vivo to assess the functionality of both grafted and endogenous pancreatic islets. Mn²⁺ behaves like calcium and will therefore enter metabolically active β-cells (16). MnCl₂-enhanced signals also reflect functional β-cells in vivo; pancreatic MRI signals of mice with streptozotocin (STZ)-induced diabetes were decreased in the case of both high- and low-dose STZ compared with nondiabetic control animals (20) and in BDC2.5 T-cell receptor transgenic nonobese diabetic mice even before the diabetes onset could be measured in the blood (21). Individual islets can be detected by MRI in MnCl₂-injected exteriorized pancreases, exactly correlating with immunohistochemistry performed in parallel (22).

First correlations could be made in humans by a single mangafodipir infusion; MRI data analysis could clearly and significantly distinguish between people without diabetes and patients with type 2 diabetes without differences in MRI signals in other tissues (23).

Mn²⁺ can be used not only in vitro to characterize isolated islet potency but also, more importantly, in vivo to assess the functionality of both grafted and endogenous pancreatic islets (16,24), since Mn²⁺ is also an excellent MRI agent owing to its effect on the longitudinal relaxation (T1) and was used as one of the first MRI contrast agents (25).

As with other contrast agents, Mn²⁺ has limitations mainly linked to its lack of cell specificity and its potential cytotoxicity. Chronic exposure to high concentrations of Mn²⁺ lead to extrapyramidal dysfunction resembling the dystonic movements associated with Parkinson disease, called manganism (26–28). Based on previous studies in mice, 20–35 mg/kg doses MnCl₂ were used for pancreas MRI in rodents (20,22). Although LD₅₀ levels of 38 mg/kg i.v. injections in mice were reported (29), concentrations up to 175 mg/kg were injected to reach a sufficient Mn²⁺ concentration in the brain (30), which also led to systemic toxicity such as loss of temperature regulation, and Mn²⁺-based signals changes in the brain were still measured for >4 days after Mn²⁺ infusion (30).

The present study aimed at developing a strategy to monitor the functional BCM by measuring Mn²⁺ uptake into β-cells by T1-weighted contrast in vivo Mn²⁺-mediated MRI (MnMRI). By analyzing the Mn²⁺-dependent MRI signal kinetics, we were able to identify early changes of functional BCM during β-cell compensation and failure in two diabetic mouse models in vivo.

C57Bl/6J mice were fed a high-fat/high-sucrose diet (HFD) (“Surwit” [31]) for 12 weeks as previously described (32) or injected with one single high dose of STZ (150 mg/kg i.p.) freshly dissolved in 0.1 mol/L citrate buffer (pH 4.5) (33). All animals were housed in a temperature-controlled room with a 12-h light, 12-h dark cycle and were allowed free access to food and water in compliance with Section 8 of the German animal protection law, the Guide for the Care and Use of Laboratory Animals, and the Bremen Senate in agreement with the National Institutes of Health Animal Care Guidelines (34). Blood glucose (measured with a glucometer) and weight were monitored during the experiments.

Glucose tolerance was monitored by intraperitoneal glucose tolerance tests (ipGTTs) in mice. For ipGTTs, mice were fasted 12 h overnight and injected with glucose (40%; B. Braun, Melsungen, Germany) at a dose of 1 g/kg body wt i.p. Blood samples were obtained at time points 0, 15, 30, 60, 90, and 120 min for glucose measurements using a glucometer. For detection of the effect of Mn²⁺ on glucose tolerance, 25 mg/kg MnCl₂ or vehicle control was injected once weekly in two consecutive experiments and ipGTT measured before and immediately after injection and at days 1 and 3.

At time points 0 and 30 min after 2 g/kg body wt i.p. glucose injection, blood samples were collected for measurement of serum insulin levels. Insulin was determined with a mouse insulin ELISA kit (ALPCO Diagnostics, Salem, NH).

After the serial Mn²⁺ injections and MRI measurements, pancreata of normal diet (ND) and HFD animals were perfused with a Liberase TM (cat. no. 05401119001; Roche, Mannheim, Germany) solution according to the manufacturer’s instructions and digested at 37°C, followed by washing and gradient purification of the islets using a 50:50 mixture of Histopaque1077 and -1119 (Sigma-Aldrich, Munich, Germany) as previously described (35,36). For acute insulin release in response to glucose, islets were washed and preincubated (30 min) in Krebs-Ringer bicarbonate buffer (KRBB) containing 2.8 mmol/L glucose and 0.5% BSA. The KRBB was then replaced by KRBB containing 2.8 mmol/L glucose for 1 h (basal), followed by an additional 1 h incubation in KRBB containing 16.7 mmol/L glucose (stimulated). Stimulatory index was calculated as stimulated divided by basal secretion.

Pancreatic tissues were processed as previously described (32,37). Mouse pancreata were dissected and fixed in 4% formaldehyde at 4°C for 12 h before embedding in paraffin. For Ki67 and insulin staining, 2-µm sections were deparaffinized, rehydrated, and incubated overnight at 4°C with anti-Ki67 and the next day with anti-insulin (both Dako, Hamburg, Germany), followed by fluorescein isothiocyanate– or Cy3-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA). Slides were mounted with glycerol gelatin (Sigma) or with Vectashield with DAPI (Vector Laboratories). For BCM measurement, 10 sections (spanning the width of the pancreas) per mouse were analyzed. Pancreatic tissue area and insulin-positive area as well as nuclear size and nuclear distance were determined by computer-assisted measurements using a Nikon MEA53200 microscope (Nikon GmbH, Düsseldorf, Germany), and images were acquired using NIS-Elements software (Nikon). BCM was obtained by multiplying the ratio of insulin-positive tissue area to whole pancreatic tissue area by the weight of the pancreas as previously described (32).

For MnMRI, mice were fasted overnight and anesthetized with 1.0–2.0% isoflurane in 250 mL/min O₂, received a tail vein catheter, and were positioned in supine position on a warmed animal bed. Breathing was monitored with a pressure sensor at the chest. Just before positioning inside the magnet, mice received 2 g/kg glucose i.p., and two to four baseline images were acquired before MnCl₂ (8 mg/kg i.v.) was injected via a tail vein catheter. (For protocol, see Fig. 1A.) The experiments were performed on a 7 Tesla MR system (BioSpec 70/20 USR with AVIII; Bruker, Ettlingen, Germany) using a quadrature volume coil (72-mm inner diameter) for both radiofrequency transmission and signal reception. The scan repetition time was at least 6 s to ensure full relaxation and thus to avoid any breathing rate–dependent T1 effects (38). Time course experiments were acquired using a 3D Snapshot-FLASH sequence with inversion recovery preparation (39) with the following parameters: time of inversion (TI) = 700 ms, FLASH-type signal readout: echo time = 0.9 ms (thus insensitive to T2 relaxation), repetition time = 1.98 ms, excitation pulse = 10°, bandwidth = 500 kHz, matrix = 192 × 96 × 16, field of view = 70 × 35 × 16 mm, nominal image resolution = 365 × 365 × 1,000 µm, centric phase encoding, time of acquisition = 190 ms (38,40). For reference purposes, one proton-density image was acquired using the same 3D Snapshot-FLASH method but omitting the inversion preparation. During the experiments, all measurement parameters were kept constant (personnel, injection system, Mn²⁺ charge, narcotic setup).

After standard image reconstruction on the Bruker system, the image data were converted from the Bruker to standard NIfTI data format using a home-written software (coded in Matlab 2008b, using “Tools for NIfTI and ANALYZE image” by Jimmy Chen, Matlab File Exchange [http://www.mathworks.de/matlabcentral/fileexchange/8797-tools-for-nifti-and-analyze-image]). These three-dimensional data sets were then subjected to a rigid body motion correction and concatenated to a single four-dimensional data set using SPM8 (Wellcome Department of Imaging Neuroscience, University College London [http://www.fil.ion.ucl.ac.uk/spm/]) for further analysis. The time course data of pancreatic and hepatic tissues were derived from manually drawn regions of interest (ROIs) (38,40). A reduction in breathing rate was often observed directly after bolus MnCl₂ administration, which normalized after ∼3–5 min. ROIs were chosen after Mn²⁺ application. An organ-specific differentiation of MRI signals in the mouse abdomen is feasible, since each organ in the abdomen can be discriminated owing to their time- and signal amplitude–dependent response to the Mn²⁺ application. In the HFD studies, MRI signals were normalized to the 23-min liver signal in order to see the pancreas-specific kinetics. In the STZ studies, pancreatic MRI signals were normalized to the maximal signal of the corresponding ROI (proton-density image) because of the hepatic fibrosis induced by STZ injection.

Stainings/MRI sets were evaluated in a randomized manner independently by the investigators (A.M., K.S., N.L., Z.A., V.K., and E.K.). The results presented are means ± SEM). The significance of difference between individual experiments was tested by Student t test. Significance was set at P < 0.05.

For monitoring of β-cell function and mass during diabetes progression in the HFD mouse model, MnMRI (Figs. 1A and B and 2) together with ipGTT, insulin secretion during the ipGTT, and BCM (Figs. 1C and 3A–C) was analyzed. Figure 1A shows the experimental setting of the MnMRI measurements in the mice. Glucose was delivered intraperitoneally 25 min before intravenous MnCl₂ application to allow glucose-dependent Mn²⁺ uptake into the β-cells.

After Mn²⁺ injection, the signal increased in T1-weighted magnetic resonance images in a diet- and time-dependent manner (Fig. 1B); a time course of MnMRI revealed enlarged MRI signals after 1 week of HFD compared with in ND control mice. The increased signal was only transient, and decreased signals were measured with prolonged diet already after 2 weeks of the HFD. From the 4th week on, MRI signals decreased significantly in the HFD group, compared with ND mice, which had stable MRI signals during the course of the experiment. Such reduced MRI signals in HFD mice persisted over the whole 12-week experimental period, which is also reflected by the area-under-the-curve (AUC) calculation (Fig. 1B) of the MRI kinetics as well as by the continued MRI signal change from one single representative mouse over the 12 weeks of HFD (Supplementary Fig. 1A).

In order to compensate for inevitable variations of Mn²⁺ delivery to the tissues, we normalized pancreatic MRI signals to those of the liver. Reported signal intensities reflect the signal changes after subtraction of the mean baseline signal taken from the first three to four images prior to intravenous Mn²⁺ injection (Fig. 2). Before averaging of individual time courses, data were linearly interpolated and regridded to 1-min intervals referenced to t = 0 for Mn²⁺ injection.

The MnMRI signals in the HFD mice correlated well with both glucose tolerance and insulin secretion during the course of HFD feeding (Figs. 1 and 3; Supplementary Fig. 1A). At 1 week after HFD, when MRI signals were increased, glucose tolerance was improved (Fig. 1C), together with increased glucose-stimulated insulin secretion (GSIS) (Fig. 3A and B), which confirms that the increased glucose-stimulated MnMRI signals indeed reflect the improved β-cell function. HFD feeding significantly impaired glucose tolerance and insulin secretion, with significantly higher glucose levels during the glucose tolerance test and higher fasting glucose (Fig. 1C) and reduced GSIS (Fig. 3A and B) after 4 weeks—the same time point when pancreatic MnMRI signals were reduced—compared with ND mice.

In contrast to the functional changes, BCM remained in the compensatory phase during the first 8 weeks of the HFD. Already after 1 week of HFD, BCM was 1.5-fold increased, while BCM was unchanged in ND controls during the 12 weeks (Fig. 3C). Again, the compensatory increase in BCM already after 1 week correlated with the improved function and the increased MRI signal. We did not expect such a fast mass adaptation and therefore analyzed single β-cell and islet size in detail. We observed a significant increase in smaller islets consisting of 1–50 β-cells throughout the pancreas in the HFD mice for 1 week, compared with ND, while larger islets did not significantly increase (Fig. 3D). β-Cell size (Fig. 3E), β-cell nuclear size (Fig. 3F), and nuclear distance (Fig. 3G) as well as Ki67⁺ β-cells (Fig. 3H) were similar in ND and HFD.

In contrast, from 4 weeks on, glucose tolerance and insulin secretion were impaired by the HFD, and MRI signals were reduced and BCM remained increased up to 8 weeks of diet, showing that the functional BCM and not the BCM alone is monitored by MnMRI. A drop in BCM back to the ND level was observed at 12 weeks (Fig. 3C); also here, glucose tolerance test (Fig. 1C) and GSIS (Fig. 3A and B) were lower than in control mice.

All mice fed with the HFD gained weight significantly after 6 weeks of feeding (1.4-fold increase compared with ND) (Fig. 3I). At the end of the experiments, in vitro GSIS was performed on isolated islets from the MnCl₂-injected mice under the ND and HFD (Fig. 3J and K). In line with our previous data (32,34), HFD significantly reduced GSIS compared with ND controls.

A second model was tested for the MnMRI analysis of functional BCM. Three-month-old male C57Bl/6 mice were injected with a single dose of 150 mg/kg STZ or vehicle. On day 14, mice were analyzed by MRI (Fig. 4A–C; Supplementary Fig. 1C) and killed on the next day for BCM measurement (Fig. 4E). MRI signals were clearly reduced in the STZ mice (Fig. 4A), which was confirmed by the AUC analysis (Fig. 4B) (26% reduction). No changes in Mn²⁺ cellular uptake occurred during the first 10 min of MnCl₂ injection, although the majority of the β-cells were destroyed. But during the second phase of Mn²⁺ uptake from 10 to 30 min, we observed a 43% induction of the MRI signal in the control mice, while we could only detect a 9% signal enhancement in the STZ mice (Fig. 4C). Such 79% reduction of Mn²⁺ uptake also correlates with 73% loss in BCM in the STZ mice (Fig. 4D).

For each mouse, the relative pancreas signal at 30 min after bolus MnCl₂ infusion was monitored separately before and after STZ injection and its decrease differed between 30 and 75% (Supplementary Fig. 1A). In line with the MRI data, STZ increased glucose levels to >400 mg/dL after 14 days (Fig. 4E). In contrast to HFD feeding, where liver signals remained unchanged among the treatment groups, liver MRI signals were reduced in the STZ mice (Supplementary Fig. 1B and C). Pancreas MRI results were normalized to the maximal signal of the respective corresponding ROI (100%) (proton-density image).

For exclusion of a direct long-term effect of Mn²⁺ on glucose tolerance during the experiments, 25 mg/kg MnCl₂ or vehicle control was injected once weekly over 2 weeks and ipGTTs were performed immediately after the MnCl₂ injection (Fig. 5A and C) and 2 days later (Fig. 5B and D). After the 1st MnCl₂ injection, glucose concentrations were slightly increased at 15 and 60 min during the ipGTT (Fig. 5A)—an effect that disappeared after 2 days (Fig. 5B). Seven days after the first dose, mice received a second MnCl₂ injection, which showed again an impaired glucose tolerance (Fig. 5C), but this effect also washed off 2 days later (Fig. 5D). The results show that multiple MnCl₂ injections also had no long-term effects on glucose tolerance. We further monitored a possible Mn²⁺ accumulation in the brain but could not detect any MRI signal changes (Supplementary Fig. 1E).

In this study, we established in vivo imaging of functional BCM and show a strong correlation of MnMRI signals with a functional BCM during β-cell compensation as well as β-cell failure. Importantly, a sole association of MnMRI signals with conventionally determined BCM measurements was not observed; rather, MnMRI data displayed the functional BCM. This can be assumed from MnMRI comparisons with ipGTT, GSIS, and BCM during diabetes progression, when changes in glucose metabolism occur, and subsequently, the glucose-dependent uptake of Mn²⁺ is affected in parallel resulting in enhanced MnMRI signals as long as β-cells compensate successfully for the higher demand of insulin during HFD feeding and display adequate GSIS, which normalize glucose levels. In contrast, when GSIS was impaired and glucose levels rose, MnMRI levels were reduced in the HFD mice, even when a higher BCM was still present and declined further with more severe loss of β-cell function. In the STZ model, a more severe model of β-cell loss where loss of function together with a reduction of BCM occurred rapidly after high-dose STZ injection, MnMRI signals were significantly reduced. Initially, we expected that the MRI signals would change linearly and proportional to the BCM, but in none of the tested models did such linearity occur. Definitely, the BCM method is still the most accurate but invasive method to quantify insulin-producing cells. Our aim was to present a noninvasive alternative to the classical BCM measurement, which also allows studying BCM changes longitudinally, but this method has its limitations with respect to absolute accuracy, and we could not reach sensitivity to a single islet level.

While glucose dependent cellular Mn²⁺ uptake correlated with insulin secretion, we still observed glucose stimulated MnMRI signals after 12 weeks of HFD feeding, which were naturally reduced when compared with the ND group, but in contrast, GSIS did not occur at all after 12 weeks. It is obvious that basal insulin secretion raises in response to the insulin resistance, especially when free fatty acid levels are chronically elevated (32) and cause a chronic and glucose independent insulin secretion from the β-cells (41,42). In contrast, Mn²⁺ uptake is not elevated under physiological glucose concentrations at insulin resistance.

We could already see β-cell mass compensation after 1 week, a short time of the diet, together with improved function when basal insulin secretion was not increased yet. While improved function was only seen at week 1, it remained stable until week 4. After longer feeding times, basal insulin secretion was highly increased, as shown before, while GSIS is blunted in HFD mice (43), which resulted in an impaired stimulatory index.

It appears that there are different mechanisms of β-cell adaptation at young and old age as well as in the response to HFD feeding. While β-cell proliferation measured by BrdU incorporation is 10 times higher in 2-week-old than in 4-month-old mice, it is also increased after 2 weeks of HFD feeding but not after 14 weeks (43). This indicates that β-cells can proliferate at a young age (44), but at older age, β-cell compensation is unlikely to be mediated by proliferation. β-Cell hypertrophy is another mechanism reported for β-cell compensation (45). Here, we started the HFD feeding at the age of 12 weeks and observed rapid adaptation in BCM already after 1 week of HFD, which could not be explained by increased proliferation or hypertrophy and was, rather, the result of an increased number of islets in the HFD (Fig. 3D and H). This possibility needs to be further investigated to avoid misinterpretation, as this was not the focus of our current study.

MRI kinetics display a rapid signal increase within 5 min after MnCl₂ injection. AUC analyses were used to calculate the signal differences during the 30 min of analysis and showed clear differences in signal intensities of the region of the pancreas, irrespective of normalization to liver signals or proton-density image. For exclusion of a general and unspecific uptake of MnCl₂, signals were measured in muscle during 30 min in response to MnCl₂; only a negligible signal decrease was observed, which was expected owing to lack of calcium uptake in the nonactive skeletal muscle (46).

In the HFD experiments, we normalized pancreatic to hepatic signal, since with its high blood flow, intravenously injected MnCl₂ quickly accumulates in the liver. This makes the normalization to the liver the best option. Although this is still not optimal because changes in the liver signal occurred in the different mice, so liver signals were individual and could not be correlated with any measure: neither to the selected liver ROI nor to the week of feeding in the two different ND and HFD groups.

During obesity and diabetes progression, liver and pancreas tissue accumulated fat droplets after a time period of 12 weeks, which prevented accurate measurement of the ROIs and was therefore excluded from the study. After week 12, it was not possible to mark pancreatic ROIs without the risk of including the fat signals or voxels, which are influenced by the partial volume effect in all mice and would inhibit accurate data analysis under the established experimental conditions during a longitudinal study.

In STZ experiments, the liver is highly affected by alkylation; a significant reduction of the MnMRI signal intensity in the liver was measured and shown in contrast to findings by Antkowiak et al. (20), where no liver signal changes were assumed. At the end, both our study and that by Antkowiak et al. study show an STZ-based MRI signal reduction under glucose stimulation. Hepatic damage evaluated by increased expression of alanine aminotransferase and morphologic analysis shows large necrotic areas in the liver in STZ-injected rats (47). We suspect that the STZ uptake via GLUT2 transporter of the liver is the reason for hepatic damage (48). Therefore, an alternative MnMRI signal normalization strategy in the pancreas was chosen in the STZ experiments. Signals were referenced to the maximal signal of the corresponding ROI in the proton-density image. Proton-density images were acquired without the inversion recovery pulse to see the maximal signal of the ROI, arising from all protons in the ROI (100%), which is constant over the course of measurement.

It has to be pointed out that with the described method, it is not possible to visualize individual islets, which form clusters of roughly 40–300 μm in diameter (49–51). In theory, the partial volume effect is especially strong when imaging the small β-cell organelles by MnMRI because small single spots appear much brighter in smaller measured voxels. Nevertheless, those small structures fall far below the achievable actual spatial resolution of the MnMRI, especially with our kinetic approach, which is limited to measure time and thus resolution. With increased measuring time, individual islets of Langerhans could be detected by ex vivo MnMRI (22), but even ex vivo, an accurate determination of the islet volume is not possible. The goal of the current study was to detect not individual islets but the correlation of MnMRI signals with β-cell function and mass during β-cell compensation and diabetes progression. Therefore, we set up serial measurements enabling the differentiation between kinetics and final signal amplitude.

Mn²⁺ kinetics follows the glucose-induced Ca²⁺ entry in a better way than a single time point. This can especially be observed by the differences in the HFD and STZ models; while a significant Mn²⁺-signal decrease in the pancreas was already observed after 5 min in the HFD mice, it occurred only after 10 min in the STZ model. Nevertheless, in our setting of 30-min Mn²⁺ kinetic measurement at 50 min after glucose, a clear plateau phase was not reached yet, while a plateau phase was already reported 15 min after glucose and Mn²⁺ injection (20) and a maximal signal was observed at 1 h (52), which is in accordance with our studies.

We decided on the 30-min experimental setting after Mn²⁺ and 20–50 min after glucose because we did not expect large changes in β-cell Ca²⁺-channel activity at 50 min after glucose injection. In this longitudinal study, we kept the experimental settings constant for the animals’ welfare, which also allowed shorter anesthesia duration for the mice.

One risk under Mn²⁺ exposure is the development of detrimental side effects (manganism), which occur at higher and chronic doses of MnCl₂ (LD₅₀ 38 mg/kg) (29,53), while in our study, 8 mg/kg was used for the MRI measures in one single bolus injection. No Mn²⁺ enhancement was detectable in the brain, which would be necessary to develop the diseases described. Also, serial Mn²⁺ injections did not lead to a magnetic resonance–detectable accumulation of Mn²⁺ in the brain. A high Mn²⁺ accumulation in a small number of cells cannot be ruled out if their volume is much smaller than the voxel volume of the image taken in these experiments.

Although Mn²⁺ caused a static impairment of glucose tolerance in mice, glucose homeostasis was not durably affected by Mn²⁺ injections.

Other endocrine as well as acinar cells in the pancreas express calcium channels; we consider the glucose-stimulated Mn²⁺ uptake to be rather specific for β-cells, but further in vitro studies on a single cell level are necessary to prove this.

α-Cells carry ATP-sensitive K⁺ channels (K_ATP) similar to β-cells (54) as well as Ca²⁺-channels (55,56). K_ATP channels play similar roles in both cell types; the difference is that the K_ATP channel activity in α-cells is very low. Glucose would lead to closure of the channels and membrane depolarization; however, membrane depolarization in α-cells results in lower Ca²⁺ entry versus in β-cells (57,58). The paracrine effect of insulin on the α-cells does not involve electrical activity, and α-cells are rather electrically active at hypoglycemia (below 5 mmol/L glucose).

Somatostatin-releasing δ-cells are the 3rd largest cell population in islets and also express voltage-gated L-type Ca²⁺-channels. While increase of glucose results in the stimulation of somatostatin secretion, it had no effects on intracellular Ca²⁺ levels (59), membrane potential, or electrical activity (60) in in vitro studies. Therefore, we also do not expect large amounts of Mn²⁺ entry into δ-cells in response to glucose.

Amylase secretion by acinar cells is also regulated by Ca²⁺-channels, and Mn²⁺ was shown to increase amylase secretion in vitro in Ca²⁺- and Mg²⁺-free solution; in contrast, under physiological conditions, Mn²⁺ does not induce amylase increase and thus is not expected to give a signal in acinar cells (61); acinar cells show a robust mangafodipir-MRI signal at low glucose (62). By monitoring the difference in the MRI signal from low to high glucose in the mice, we expect a rather β-cell–specific change, as shown previously (52).

Measurements of functional BCM remain a challenge owing to the small size of pancreatic islets, their poor contrast compared with the surrounding tissues, and their position in the body. However, the present study has proven that noninvasive MRI is a reliable tool to detect small changes in the functionality of β-cells in vivo and exhibits the potential for early noninvasive detection of changes in functional BCM. Our approach carries further potential for multifaceted longitudinal studies to monitor functional BCM adaptation during diabetes progression as well as for the evaluation of therapies for diabetes as an alternative to classical BCM analysis. The optimized protocol of T1-weighted MRI of the mouse abdomen fulfills the requirements of least invasive MR imaging.

Acknowledgments. The authors thank Benjamin Pawlik and Amod Godbole (University of Bremen) for help with the MRI analysis and the Maedler laboratory, especially Amin Ardestani and Federico Paroni (University of Bremen), for helpful suggestions.

Funding. This work was supported by the FP7 program In Vivo Imaging of Beta cell Receptors by Applied Nano Technology (VIBRANT; FP7-228933-2), the European Research Council, and the Competence Network Diabetes Mellitus supported by the German Federal Ministry of Education and Research and University of Bremen research funds.

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

Author Contributions. A.M., K.M., and E.K. designed, performed, and analyzed research. K.S., W.D., J.B., V.H.T., N.L., Z.A., and V.K. performed experiments. A.M., K.M., and E.K. wrote the manuscript. K.M. 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.