Radiomanganese PET Detects Changes in Functional β-Cell Mass in Mouse Models of Diabetes

Abstract

The noninvasive measurement of functional β-cell mass would be clinically valuable for monitoring the progression of type 1 and type 2 diabetes as well as the viability of transplanted insulin-producing cells. Although previous work using MRI has shown promise for functional β-cell mass determination through voltage-dependent Ca²⁺ channel (VDCC)–mediated internalization of Mn²⁺, the clinical utility of this technique is limited by the cytotoxic levels of the Mn²⁺ contrast agent. Here, we show that positron emission tomography (PET) is advantageous for determining functional β-cell mass using ⁵²Mn²⁺ (t_1/2: 5.6 days). We investigated the whole-body distribution of ⁵²Mn²⁺ in healthy adult mice by dynamic and static PET imaging. Pancreatic VDCC uptake of ⁵²Mn²⁺ was successfully manipulated pharmacologically in vitro and in vivo using glucose, nifedipine (VDCC blocker), the sulfonylureas tolbutamide and glibenclamide (K_ATP channel blockers), and diazoxide (K_ATP channel opener). In a mouse model of streptozotocin-induced type 1 diabetes, ⁵²Mn²⁺ uptake in the pancreas was distinguished from healthy controls in parallel with classic histological quantification of β-cell mass from pancreatic sections. ⁵²Mn²⁺-PET also reported the expected increase in functional β-cell mass in the ob/ob model of pretype 2 diabetes, a result corroborated by histological β-cell mass measurements and live-cell imaging of β-cell Ca²⁺ oscillations. These results indicate that ⁵²Mn²⁺-PET is a sensitive new tool for the noninvasive assessment of functional β-cell mass.

Introduction

Type 1 and type 2 diabetes, although arising from different etiologies, are each associated with the functional loss of insulin-secreting β-cells and can lead to life-threatening complications such as cardiovascular disease and neuropathy (1). As such, a variety of therapeutics are being pursued that aim to preserve, expand, or replace β-cells. An imaging modality capable of monitoring functional β-cell mass in vivo as well as the viability of islet- or stem cell–derived β-cell transplants would therefore be invaluable to future therapeutic investigations.

Human pancreatic islets occupy ∼4.5% of the pancreas volume and are composed of a mixture of β-, α-, γ-, δ-, and ε-cells (2,3). Because individual islets vary in size from 25 to 400 µm in diameter (3–5) and are nonuniformly distributed throughout the pancreas, quantification is challenging through noninvasive anatomical imaging techniques such as MRI or computed tomography (CT) (6). Alternatively, positron emission tomography (PET) is a technique that involves quantifying the in vivo distribution of a biologically relevant moiety by tracking a positron-emitting radioisotope. Compared with MRI and CT, PET has significantly greater imaging sensitivity and inherently probes physiology rather than anatomy (7), which may prove useful in the clinical quantification of functional β-cell mass.

Mn²⁺, in a behavior that mimics Ca²⁺, is freely transported through voltage-dependent Ca²⁺ channels (VDCC) (8,9). Because VDCC activation is required for the release of insulin from β-cells, Mn²⁺ has been proposed as a molecular imaging agent for probing β-cell function and mass using Mn²⁺-enhanced MRI (10–15). However, this technique is limited by significant cellular toxicity and the slow biological clearance of Mn²⁺, which prevents the possibility of repeated administration (9,16). Because the sensitivity of PET is several orders of magnitude higher than MRI, the objective of this work was to assess the feasibility of using PET to probe β-cell mass and function.

Herein, we take advantage of the exceptional sensitivity and quantitation capability of PET and the availability of radiomanganese (⁵²Mn²⁺) to noninvasively image functional β-cell mass in vivo. When intravenously (i.v.) injected in mice, ⁵²Mn²⁺ (t_1/2: 5.6 days) showed a rapid accumulation in the mouse pancreas that was sensitive to glucose and pharmacological manipulation of VDCC. We also explored the potential of ⁵²Mn²⁺-PET to detect changes in functional β-cell mass in mouse models of type 1 and type 2 diabetes. Overall, our results point to the significant potential for ⁵²Mn²⁺ to overcome several limitations of other β-cell imaging agents and potentially become a method of choice for studying β-cell physiology noninvasively.

Research Design and Methods

⁵²Mn²⁺-PET Imaging

Aliquots of the buffered ⁵²Mn²⁺ solution (0.01 mol/L NaOAc, pH 6.5) were diluted to the desired injection volume (∼200 µL) with PBS, typically 2–4 MBq (∼50–100 µCi) per subject. The acquisition of PET images was performed with an Inveon µPET/µCT scanner (Siemens Preclinical Solutions). To study the biodistribution and clearance of ⁵²Mn²⁺ in the mouse, 3.7 MBq (100 µCi) of ⁵²Mn²⁺ was i.v. injected into female ICR mice. As a result of the long decay half-life of ⁵²Mn²⁺ (t_1/2: 5.6 days), multiple PET scans were recorded between 1 h and 13 days after injection of the radiotracer. Before each scan, mice were anesthetized with isoflurane (4% induction; 1% maintenance) and placed prone in the scanner, and 30–40 × 10⁶ coincidence events per mouse static PET scans were acquired (time window, 3.432 ns; energy window, 350−650 keV).

The PET images were reconstructed in Inveon Acquisition Workplace (Siemens Preclinical Solutions) workstation using a nonscatter-corrected three-dimensional ordered subset expectation optimization/maximum a posteriori (OSEM3D/MAP) algorithm. Region-of-interest (ROI) analysis was performed after organs were manually delineated on the PET images. Tissue ⁵²Mn²⁺ uptake values are reported as standardized uptake value (SUV), which is normalized to whole-body ⁵²Mn²⁺ concentration to account for weight disparities between animal models. The percentage of injected dose per gram of tissue (%ID/g) is also reported in Supplementary Tables 6–9). To acquire dynamic PET scans, mice were anesthetized with isoflurane and the lateral tail vein was catheterized. Simultaneous with the administration of ∼1.7 MBq (∼50 µCi) of ⁵²Mn²⁺ as a fast i.v. bolus, 1-h scans were recorded, and list-mode files were binned into 46 frames (12 × 5 s, 6 × 10 s, 6 × 30 s, 10 × 60 s, 6 × 150 s, 6 × 300 s) and the images reconstructed using the OSEM3D/MAP algorithm. For comparison, 1.7 MBq (50 µCi) ⁵²Mn²⁺ was continuously infused with a syringe pump (Model 780100l; KD Scientific) during the first 30 min of a 60-min experiment, and list-mode files were binned into 30 frames (2 min each) and reconstructed using the OSEM3D/MAP algorithm.

Effect of Pharmacological VDCC Manipulation on ⁵²Mn²⁺ Uptake In Vivo

Pancreatic ⁵²Mn²⁺ uptake was stimulated using glucose and glibenclamide (Tocris Biosciences), which is known to promote insulin release in β-cells via blockade of ATP-sensitive K⁺ channels (K_ATP). Mice were injected intraperitoneally (i.p.) with 100 µL of 1 g/kg glucose, or glucose plus 5 mg/kg glibenclamide in PBS, 15 min before the i.v. injection of 0.74–1.85 MBq (20 μCi) of ⁵²Mn²⁺. VDCC blockade was achieved via i.p. injection of 20 mg/kg nifedipine (MP Biomedicals) dissolved in DMSO. To activate K_ATP channels, mice received an i.p. injection of 20 mg/kg diazoxide (Tocris Biosciences) in PBS, a clinically used K_ATP agonist, 15 min before injection of 0.74 MBq (20 μCi) of ⁵²Mn²⁺. Whole-body PET scans were acquired 1 h after the injection of the radiotracer, after which ex vivo biodistribution analysis was performed.

⁵²Mn²⁺-PET Studies in a Type 1 Diabetes Model

Type 1 diabetes was induced in female ICR mice via one i.p. injection of 180 mg/kg streptozotocin (STZ; MP Biomedical), a toxin that selectively destroys pancreatic β-cells (17). The injectable STZ solution (12.5 mg/mL) was prepared fresh in PBS. The weight of each mouse was measured daily, and blood glucose levels were recorded every other day with a TRUEresult glucometer (Trividia Health Inc.) using blood samples collected from the tail vein. Mice were considered diabetic after two consecutive blood glucose readings above 250 mg/dL and were used for ⁵²Mn²⁺-PET imaging studies 1 week after the STZ injection. To evaluate ⁵²Mn²⁺ pancreatic uptake in diabetic mice, 0.74 MBq (20 µCi) of radioactivity was administered i.v., and static PET images were recorded 1 h after administration of the tracer. Ex vivo biodistribution was performed after PET acquisition.

⁵²Mn²⁺-PET Studies in a Pretype 2 Diabetes Model

Pancreatic uptake of ⁵²Mn²⁺ was measured in 10-week-old wild-type and ob/ob mice on the C57BL/6J background. For PET imaging, 0.74 MBq (20 µCi) of ⁵²Mn²⁺ was i.v. injected 1 h before PET scan acquisition. Accumulation of ⁵²Mn²⁺ in the pancreas and other organs of interest was also quantified by ex vivo biodistribution analysis.

Additional Materials and Methods

Animal studies were conducted under the approval of University of Wisconsin-Madison Institutional Animal Care and Use Committee. Detailed information on animal models, isotope production, ex vivo biodistribution, islet isolation and imaging, and β-cell mass measurement is provided in the Supplementary Data.

Results

In Vivo Whole-Body PET and Biodistribution of ⁵²Mn²⁺ in Normal Mice

The in vivo biodistribution of ⁵²Mn²⁺ was investigated noninvasively with PET and ex vivo gamma counting. Figure 1A shows coronal planes intersecting the pancreas of ICR mice, in PET scans acquired between 1 h and 13 days after i.v. injection of 3.7 MBq (100 µCi) of ⁵²Mn²⁺. A rapid and prominent accumulation of ⁵²Mn²⁺ was observed in the pancreas, kidneys, liver, heart, and salivary glands (5.13 ± 0.38, 5.13 ± 0.02, 3.27 ± 0.36, 2.11 ± 0.20, and 2.30 ± 0.26 SUV at 1 h postinjection, respectively; n = 3) (Fig. 1B). In the subsequent time points during the longitudinal study, ⁵²Mn²⁺ uptake gradually declined in all organs except the salivary gland, where uptake remained stable at an SUV of ∼3 (Fig. 1B and Supplementary Table 1). Uptake of ⁵²Mn²⁺ in the pancreas, which was highest at 1 h postinjection, was notably higher than that in the liver and kidneys at all time points. Such favorable pancreas-to-normal organ contrast ratios, which peaked at ∼3 days after ⁵²Mn²⁺ administration, facilitated an easy delineation of the pancreas (Supplementary Fig. 1). Uptake in the muscle was very low and had little variation during the study. Ex vivo biodistribution was performed after the last PET scan 13 days postinjection, confirming a marked accumulation of ⁵²Mn²⁺ in the salivary gland, pancreas, kidneys, and to a lesser extent, the heart and liver (Fig. 1C and Supplementary Table 2). Other organs, including the brain, lungs, bones, intestines, stomach, and spleen, displayed low ⁵²Mn²⁺ uptake, typically less than 0.5 SUV.

Figure 1.

Tissue distribution and pharmacokinetics of ⁵²Mn²⁺. A: Serial PET images of ICR mice injected i.v. with ⁵²Mn²⁺ (no anesthesia except during the PET scans). Coronal PET image slices were selected to best show pancreatic uptake. Arrows point to P, pancreas; H, heart; L, liver; I, intestines; and SG, salivary gland. B: ROI-based quantification of ⁵²Mn²⁺ uptake in the heart, liver, kidneys, muscle, pancreas, and submandibular salivary gland. C: Ex vivo ⁵²Mn²⁺ biodistribution of euthanized mice after the last PET scans, determined by gamma counting (n = 4).

Because the whole-body distribution of ⁵²Mn²⁺ occurred largely within 1 h after i.v. administration (Fig. 1), a dynamic PET study was designed to investigate the ⁵²Mn²⁺ kinetics during this time. Figure 2 shows the time-activity curves (TACs) resulting from ROI analysis of the dynamic PET data corresponding to the pancreas, liver, kidneys, heart, salivary gland, and muscle under two i.v. administration regimens: rapid bolus injection and 30-min continuous infusion. The analysis of the myocardial TAC revealed extremely fast blood extraction kinetics, with a blood circulation half-life of 10.7 ± 3.5 s in mice administered a rapid i.v. ⁵²Mn²⁺ bolus. Consequently, ⁵²Mn²⁺ uptake was stabilized in the organs of interests within 5 min postinjection. A residual radioactivity of 2.60 ± 0.41 SUV was observed in the heart at 1 h postinjection, which was consistent with the specific uptake of Mn²⁺ ions by myocardial tissue. Compared with the static 1 h postinjection PET scans (Fig. 1), similar ⁵²Mn²⁺ uptake values were observed after rapid bolus injection in the heart (2.11 ± 0.20 vs. 2.60 ± 0.41 SUV) and muscle (0.38 ± 0.03 vs 0.16 ± 0.02 SUV), whereas the liver (3.27 ± 0.34 vs. 5.16 ± 1.46 SUV) and kidneys (5.13 ± 0.02 vs. 7.81 ± 0.51 SUV) were much higher at the end of the dynamic scan after a rapid i.v. ⁵²Mn²⁺ bolus. Interestingly, an ∼50% reduction in pancreatic uptake, from 5.13 ± 0.38 to 2.74 ± 0.59 SUV, was observed in the dynamic studies.

Figure 2.

Rapid kinetics of tissue ⁵²Mn²⁺ uptake revealed by single i.v. bolus injection or continuous i.v. infusion. Dynamic PET TACs derived from hand-drawn ROIs for the pancreas, heart/blood, liver, kidneys, salivary gland, and muscle. The blue curves indicate TACs in mice injected with a rapid i.v. bolus of ⁵²Mn²⁺, and the red curves indicate an i.v. infusion of ⁵²Mn²⁺ over the first 30 min of the scans. The inset shows the blood/heart distribution of ⁵²Mn²⁺ within the first 4 min after i.v. bolus injection.

As clearly depicted in Fig. 2 (red curves), continuous infusion of ⁵²Mn²⁺ over a 30-min period resulted in a linear ramping in organ radioactivity, followed by an immediate plateau upon infusion termination that persisted for the remainder of the experiment. This method highlights the rapid distribution kinetics of ⁵²Mn²⁺ and the comparable results obtained by a rapid bolus or continuous infusion of ⁵²Mn²⁺. Only the kidney displayed higher radioactivity at 1 h postinjection in mice administered the rapid bolus versus mice receiving the continuous ⁵²Mn²⁺ infusion, 7.81 ± 0.51 vs. 5.14 ± 1.21 SUV, respectively (n = 4). During the continuous infusion regimen, the pancreas reached an uptake of 3.44 ± 0.69 SUV, which, similar to the rapid bolus injection, was also significantly lower than that in the static PET scans at 1 h postinjection (5.13 ± 0.38 SUV).

Previous reports have shown that many volatile anesthetics, including isoflurane, impair insulin secretion by inhibiting the glucose-dependent inactivation of K_ATP channels (18,19). Plausibly, the observed decrease in pancreatic uptake of ⁵²Mn²⁺ resulted from mice being anesthetized through the full extent of the studies during the dynamic PET scans. To investigate the effect of isoflurane anesthesia on pancreatic uptake of ⁵²Mn²⁺, 1 h postinjection biodistribution experiments were performed under different administration conditions in anesthetized (1% isoflurane) or awake mice. Supplementary Fig. 2 shows that isoflurane significantly inhibited the accumulation of ⁵²Mn²⁺ in the pancreas regardless of the administration regimen (rapid bolus vs. infusion) or glucose stimulation, suggesting that isoflurane indeed acts as an indirect VDCC inhibitor upstream of K_ATP channel closure.

Uptake of ⁵²Mn²⁺ in Isolated Islets

To corroborate the mechanism of ⁵²Mn²⁺ uptake in the pancreas and its dependence on VDCC, an ex vivo ⁵²Mn²⁺ uptake study was performed in islets isolated from ob/ob mice to take advantage of the twofold greater islet yield versus wild-type mice. As a result of the similarities between Mn²⁺ and Ca²⁺ ions, Mn²⁺ uptake by β-cells occurs via influx through VDCC (Fig. 3A) (20,21). Isolated islets were incubated with 0.37 MBq (10 µCi) of ⁵²Mn²⁺ under several conditions that stimulate/inhibit VDCC (Fig. 3B). ⁵²Mn²⁺ was readily taken up by islets, even in the presence of low (1 mmol/L) glucose, as expected for the high basal Ca²⁺ and insulin secretion reported for ob/ob β-cells (22). ⁵²Mn²⁺ uptake was significantly enhanced (P < 0.05) when the islets were stimulated with 10 mmol/L glucose. Importantly, ⁵²Mn²⁺ uptake was completely blocked by the further application of diazoxide (50 μmol/L), which inhibits the opening of VDCC via activation of K_ATP channels. As expected for the clearance of intracellular Ca²⁺ in response to diazoxide (23), intracellular ⁵²Mn²⁺ fell below the basal level. Conversely, glucose administered with the K_ATP channel blocker tolbutamide (250 μmol/L) resulted in significantly higher ⁵²Mn²⁺ retention than glucose alone. Taken together, these experiments demonstrate that pancreatic islet uptake of ⁵²Mn²⁺ depends on the activity of β-cell VDCC.

Figure 3.

Pharmacological manipulation of VDCC in isolated islets. A: Cartoon of the β-cell–triggering pathway. Molecular structures in blue indicate compounds that activate Ca²⁺ influx through VDCC, whereas compounds in red are inhibitory. B: Uptake of ⁵²Mn²⁺ by isolated ob/ob mouse islets. Groups of 50 islets from three preparations were incubated with ⁵²Mn²⁺ (370 kBq) in the presence of glucose and VDCC modulators as indicated. Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001.

Pharmacological Manipulation of ⁵²Mn²⁺ Pancreatic Uptake In Vivo

To verify the specificity of ⁵²Mn²⁺ accretion in the pancreas noninvasively using PET imaging, ⁵²Mn²⁺ pancreatic uptake was pharmacologically manipulated in vivo through the inhibition or stimulation of VDCC. In line with the in vitro results, inhibition of VDCC by direct blockade with nifedipine, or activation of K_ATP with diazoxide, resulted in a significant (P < 0.0001) abrogation of the PET signal within the pancreas. Figure 4A shows coronal PET slices of the pancreas of ICR mice receiving 20 mg/kg i.p. injections of nifedipine or diazoxide 10–15 min before the administration of an i.v. ⁵²Mn²⁺ bolus. Compared with the control group, a clear decrease in pancreatic PET signal was observed 1 h after the injection of ⁵²Mn²⁺. PET quantification revealed a 44% (5.13 ± 0.36 vs. 2.85 ± 0.92 SUV) and 54% (5.13 ± 0.36 vs. 2.36 ± 0.61 SUV) decline in ⁵²Mn²⁺ uptake in groups administered diazoxide and nifedipine, respectively (Fig. 4B and C). The remnant pancreatic signal suggests that i.p. drug injection resulted in the incomplete blockade of VDCC, or more likely, that some ⁵²Mn²⁺ uptake is independent of VDCC (8). However, mice administered glibenclamide (5 mg/kg) exhibited a significant enhancement in pancreatic uptake of ⁵²Mn²⁺. The distribution of ⁵²Mn²⁺ in other organs of interest, including the heart, liver, kidneys, spleen, and salivary gland, remained largely unaltered among the groups (Supplementary Tables 3 and 4). These results demonstrate the potential of ⁵²Mn²⁺ PET imaging to noninvasively detect changes in functional β-cell mass.

Figure 4.

In vivo assessment of functional β-cell mass by ⁵²Mn²⁺-PET. A: Coronal PET images at 1 h postinjection showing the pancreas of ICR mice given i.p. injections of diazoxide (20 mg/kg), nifedipine (20 mg/kg), or glibenclamide (5 mg/kg) before the administration of a ⁵²Mn²⁺ rapid bolus. The pancreas (P) is demarcated by white dashed contours. B: Manual ROI-based quantification of ⁵²Mn²⁺ uptake in various tissues from static PET images acquired at 1 h postinjection. C: Ex vivo biodistribution analysis after PET imaging at 1 h postinjection. Significantly reduced pancreatic uptake of ⁵²Mn²⁺ is observed in mice that received nifedipine and diazoxide before radiotracer administration. Mice that received glibenclamide (5 mg/kg) before radiotracer administration had significantly higher pancreatic uptake of ⁵²Mn²⁺ than the control mice, based on both PET imaging (P = 0.02) and biodistribution (P = 0.047) studies. Data are presented as mean ± SD (n = 3–4 mice per group). ***P < 0.001.

⁵²Mn²⁺ Uptake in Mouse Models of Type 1 and Type 2 Diabetes

We further investigated the correlation between pancreatic ⁵²Mn²⁺ uptake and functional β-cell mass in a murine model of type 1 diabetes. Diabetes was induced in female ICR mice via a single STZ injection (180 mg/kg). Four days after injection (Fig. 5A), mice presented signs of hyperglycemia (blood glucose > 250 mg/dL) and weight loss that indicated a diabetic status. As seen in the PET images (Fig. 5B), ⁵²Mn²⁺ accumulation within the pancreas of diabetic mice was reduced by ∼60%, from 5.13 ± 0.38 SUV (n = 3) in normal mice to 2.04 ± 0.81 SUV (n = 3) in diabetic mice (P < 0.0001) (Fig. 5C, and Supplementary Tables 3 and 4). Ex vivo biodistribution analysis corroborated a very similar (58%) decrease in pancreatic accumulation of ⁵²Mn²⁺ (Fig. 5D). To compare ⁵²Mn²⁺-PET imaging with a direct histological assessment of β-cell mass, three pancreata per treatment group were weighed, sectioned, and immunostained for insulin. On average, β-cell mass in the STZ-treated mice fell by ∼70% compared with the sham-injected ICR mice (Fig. 5E), in good agreement with the ⁵²Mn²⁺-PET data.

Figure 5.

⁵²Mn²⁺-PET imaging in STZ-induced type 1 diabetes. A: After the administration of an acute dose of STZ (180 mg/kg), ICR mice started to show symptoms of diabetes: reduced body weight and high blood glucose level (BGL; >250 mg/dL). B: At 1 h postinjection, coronal PET images of healthy (left panel) or diabetic (center/right panels) ICR mice show clearly reduced PET signal in the pancreas of the STZ-diabetic mice. The pancreas (P) is demarcated by white dashed contours. The significant decline in ⁵²Mn²⁺ uptake in the pancreas of STZ-diabetic mice was confirmed quantitatively by ROI analysis of the PET images (C) and ex vivo biodistribution (D) (n = 3 mice/group). E: Quantification of β-cell mass for control and STZ-treated ICR mice. Data are shown as the mean ± SD (n = 4 mice/group). Scale bars = 400 μm. *P < 0.05, ***P < 0.001. MIP, maximum intensity projection.

⁵²Mn²⁺-PET imaging studies were also performed in C57BL/6J mice carrying the lep^ob (ob/ob) spontaneous mutation that results in obesity and pretype 2 diabetes. In this case, ⁵²Mn²⁺ accumulation in the pancreas of ob/ob mice was significantly (P < 0.001) higher than in the wild-type C57BL/6J mice (Fig. 6A). Pancreatic uptake 1 h after administration of ⁵²Mn²⁺ (n = 3) was 4.89 ± 0.68 SUV in wild-type C57BL/6J mice and 7.27 ± 1.03 SUV in ob/ob mice (Fig. 6B). Ex vivo biodistribution analysis corroborated the statistically significant differences in pancreatic SUV between the groups (P < 0.001) (Supplementary Fig. 3 and Supplementary Table 5). Uptake in the liver and salivary gland was very similar in ob/ob and wild-type animals, with SUVs of 3.68 ± 0.25 vs. 3.40 ± 0.73 and 3.11 ± 1.05 vs. 2.40 ± 0.20, respectively. In agreement with ⁵²Mn²⁺-PET imaging, post hoc analysis of insulin-stained pancreatic sections (n = 3) showed a significantly (P < 0.05) increased β-cell mass in ob/ob mice relative to controls (Fig. 6C). We also observed that ob/ob islet Ca²⁺ levels, which oscillate in response to glucose stimulation (Fig. 6D), were strongly left shifted in their glucose dependence relative to controls (Figs. 6E and F), in agreement with the in vivo ⁵²Mn²⁺-PET results and prior functional studies (22). Notably, in contrast to the experiments in Fig. 4, where pharmacological VDCC manipulation was used to strictly define functional β-cell mass, the experiments in Figs. 5 and 6 show that basal ⁵²Mn²⁺ uptake (measured in the absence of exogenous glucose or VDCC inhibitors) remains heavily dependent on changes in β-cell mass. Both approaches, when applied clinically, are likely to yield useful information.

Figure 6.

⁵²Mn²⁺-PET imaging in the ob/ob model of pretype 2 diabetes. A: Coronal PET images acquired at 1 h after ⁵²Mn²⁺ administration in ob/ob mice and C57BL/6J controls. The pancreas (P) is demarcated by white dashed contours. B: Image-derived quantification expressed as SUV indicated a significant difference in ⁵²Mn²⁺ pancreatic uptake between groups (mean ± SD; n = 3). C: Quantification of β-cell mass for wild-type and ob/ob mice (n = 3 mice/group). Scale bars = 400 μm. D: Recordings of islet Ca²⁺ in response to glucose (10 mmol/L and 7 mmol/L) from wild-type and ob/ob C57BL/6J mice. E: Contingency plot shows the range of islet behaviors at each glucose level. F: The oscillatory plateau fraction, reflecting the plasma membrane glucose sensitivity, was calculated as the fraction of time spent in the active state during each oscillation at 10 mmol/L glucose. C57BL/6J, n = 141 islets from 4 mice; ob/ob, n = 126 islets from 4 mice. Results reflect mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Discussion

The loss of functional β-cell mass is a cornerstone of both type 1 and type 2 diabetes (24). Here, we leveraged the ability of β-cells to sequester divalent metal ions (e.g., Ca²⁺, Mn²⁺, Zn²⁺, and Co²⁺), which is essential to the production and release of insulin. Although β-cell function has been widely investigated based on the measurement of Ca²⁺ currents in vitro (25), little progress has been achieved in exploring divalent metal intake for measurement of functional β-cell mass in vivo. Part of this is a result of the lack of effective methods to track the distribution of most of these metals. Fortunately, ⁵²Mn²⁺ has both magnetic and nuclear properties that facilitate its noninvasive detection by MRI, and with greater sensitivity, PET. In the present work, we report the use of ⁵²Mn²⁺ for noninvasive PET imaging of the mouse pancreas. When injected i.v., ⁵²Mn²⁺ accumulated significantly into pancreatic tissue with exceptionally fast blood extraction kinetics. ⁵²Mn²⁺ uptake in the pancreas, which peaked within minutes after administration in healthy mice, was on the order of 5–6 SUV, with a slow clearance half-life of ∼8 days. Our results demonstrate that in vivo measurements of functional β-cell mass by ⁵²Mn²⁺-PET offers complementary information to existing β-cell function assays that lack spatial information (e.g., measurement of c-peptide release), as well as measures of β-cell mass that cannot predict β-cell function (e.g., PET imaging with radiolabeled exendin-4).

A key finding of our work was that 50–60% of the observed pancreatic radioactivity resulted from the specific uptake of ⁵²Mn²⁺ through VDCC, implicating β-cells. By pharmacologically manipulating the mechanism of VDCC in isolated islets, we were able to correlate the uptake of radioactive ⁵²Mn²⁺ with Ca²⁺ uptake. This agrees with other in vitro studies using nonradioactive Mn²⁺ that reported a correlation between Mn²⁺ uptake and changes in β-cell functional capacity arising from the triggering pathway (14,15,26). Similarly, in vivo studies demonstrated that chemical inhibition of VDCC, using nifedipine or diazoxide, resulted in a drastic decline in the accumulation of ⁵²Mn²⁺ within the pancreas. In addition, the stimulation of VDCC with glucose and glibenclamide led to an increase in ⁵²Mn²⁺ uptake in the pancreas. Overall, our results indicated that the ⁵²Mn²⁺ uptake observed in the pancreas was largely mediated by and dependent on the functional β-cell mass, keeping in mind that ⁵²Mn²⁺-PET would be blind to defects in the neurohormonal amplifying pathways of insulin secretion, which are largely VDCC independent (27).

Our next question was whether this method was sensitive enough to detect changes in functional β-cell mass that occur during diabetes progression and disease. First, we used an experimental animal model of STZ-induced type 1 diabetes, which showed an ∼60% reduction in the pancreatic accumulation of ⁵²Mn²⁺, while uptake in other organs remained unaltered. Direct measurements of β-cell mass by histology showed a comparable decline (∼70%), indicating the capability of ⁵²Mn²⁺-PET imaging for revealing extreme cases of β-cell loss. Equally relevant results were observed in the obese mouse (ob/ob) model of pretype 2 diabetes, where significant (P < 0.0001) enhancement in pancreatic uptake of ⁵²Mn²⁺ was observed. These results agree with previous reports showing that ob/ob mice do not progress to type 2 diabetes (28,29) but rather present markedly higher functional β-cell mass compared with wild-type mice at 10 weeks of age, as assessed by histology and live-cell Ca²⁺ imaging. Probing the gradual gain/loss of functional β-cell mass using ⁵²Mn²⁺ (t_1/2= 5.6 days) was not feasible as a result of its long radioactive half-life. Rather than trying to account for the remnant ⁵²Mn²⁺ radioactivity in the pancreas, longitudinal studies would benefit from the use of the shorter-lived ⁵¹Mn²⁺ (t_1/2: 45.6 min), which would facilitate regular monitoring of variations in functional β-cell mass without the interference of remnant radioactivity. This is the subject of our future work.

Because significant changes can occur in β-cell function long before changes in anatomical mass are observed (30–32), it is advantageous that the relative uptake of ^52/51Mn²⁺ in vivo depends on the functional β-cell mass. This ability has the potential to shed light on the rate of functional β-cell mass decline in asymptomatic patients with type 1 diabetes as well as the survival of β-cell transplants, including stem cell–derived β-cells. Whether ^51/52Mn²⁺-PET is capable of measuring very subtle changes in this parameter is the pivotal question to its clinical significance. The diffusion of Mn²⁺ through β-cell VDCCs depends on their activation by glucose or drug-based stimulation. This has been shown here through in vitro and in vivo functional suppression and enhancement studies. For a given state of glucose activation, the amount of Mn²⁺ cellular internalization by the pancreas is partly controlled by the number of available VDCCs, which should be proportional to functional β-cell mass. However, accurate quantification of functional β-cell mass will require subtraction of nonspecific exocrine pancreas uptake of the radiotracer. For this reason, the ⁵²Mn²⁺-PET measurements in ob/ob mice underestimate the predicted effect on functional β-cell mass. Calibration may be accomplished through pharmacological manipulation to acutely suppress VDCC-mediated uptake of ⁵²Mn²⁺ using nifedipine, diazoxide, or other suppressing agents. However, same-subject calibration will require the implementation of back-to-back PET scans using the shorter-lived ⁵¹Mn²⁺ or the performance of dynamic PET scanning under a ⁵²Mn²⁺ infusion regimen that allows for the deconvolution of the endocrine signal from the exocrine background.

Aside from VDCC, there are several endogenous mechanisms for Mn²⁺ uptake, including facilitated diffusion, divalent metal transporter-1 (DMT1), Zrt- and Irt-related protein-8 (ZIP-8), transferrin receptor (TfR), and ionotropic glutamate receptor Ca²⁺ channels (8). The relative role that these alternative pathways play in ⁵²Mn²⁺ transport remains unknown, but a lack of effect by nifedipine suggests that VDCC may not be a primary transport route in many normal tissues. In this regard, our studies also provided valuable information on the kinetics of the whole-body distribution of ⁵²Mn²⁺. Besides the pancreas, ⁵²Mn²⁺ also accumulated in the liver, heart, kidneys, and salivary gland. No clear sign of renal excretion was observed despite the significant kidney uptake, and most of the ⁵²Mn²⁺ clearance occurred through the hepatobiliary system. From the analysis of the image-derived dynamic TAC of the heart, we estimated a ⁵²Mn²⁺ circulation half-life of 10.7 ± 3.5 s after a rapid bolus i.v. injection, which aligned with previous studies showing an extremely fast (∼0.8 min) blood clearance of ⁵⁴Mn in dogs (33). Such fast extraction kinetics indicate that compartmental modeling could be used to describe ⁵²Mn²⁺ uptake in a more quantitative manner. A stable residual myocardial uptake was observed after complete blood clearance, ∼1 min after injection. Of note was the marked underestimation of myocardial uptake by PET compared with biodistribution results, with an average recovery coefficient of 0.54. This is an inherent limitation of the PET imaging–based quantification of small volumes such as the myocardial wall or adrenal glands, which in mice are affected by significant partial volume effects (34,35).

Several PET tracers have been studied as potential β-cell imaging agents. [¹¹C]-dihydrotetrabenazine (DTBZ) and the ¹⁸F-labeled DTBZ analog FP-(+)-DTBZ have been shown to have high affinity for the type 2 vesicular acetylcholine transporter (VMAT2), which is expressed on the surface of rodent β-cells but appears to be entirely absent from pancreatic exocrine tissue (36–39). Unfortunately, primate models have shown very low pancreatic uptake of [¹¹C]-DTBZ, which suggests that the degree of VMAT2 expression is species dependent (40–42). More work is needed in quantifying human β-cell expression of VMAT2 and how expression density is related to β-cell function. Another promising PET approach for monitoring in vivo β-cell mass involves compartmental kinetic analysis of [¹¹C]5-hydroxytryptophan (HTP), a compound used clinically for detection of neuroendocrine tumors (43). This compound has shown differential clearance from endocrine versus exocrine pancreatic tissues based on differences in the presence of serotonin biosynthesis machinery, which effect [¹¹C]5-HTP retention in endocrine cells. With this approach, [¹¹C]5-HTP tracer uptake is observed to decrease both longitudinally in patients with type 2 diabetes and cross-sectionally in patients with type 1 diabetes compared with healthy control subjects (44,45). Another category of tracers includes radiolabeled (e.g., ¹⁸F, ⁶⁸Ga, ⁶⁴Cu, ¹¹¹In) derivatives of exendin-4, a glucagon-like protein-1 receptor agonist (46–50). Fluorescence microscopy has shown that glucagon-like protein-1 receptor is only located on β-cells within the human pancreas (51), making it an attractive molecular imaging target as a result of its specificity. In animal models, exendin-4–PET suffers from low pancreatic uptake relative to the surrounding organs (<0.3%ID/g in Sprague-Dawley rats at 1 h postinjection, whereas proximal kidney uptake is >25%ID/g) (52). Although the kidney uptake of ⁵²Mn²⁺ is also significant, the pancreatic uptake of ⁵²Mn²⁺ is comparatively high, such that improved pancreas-to-background ratios are attained by ⁵²Mn²⁺-PET. In general, the implementation of tracers targeting surface receptors of β-cells is extremely challenging, considering the low total mass of β-cells, diffuse pancreatic distribution, and heterogeneous receptor expression.

Aside from following the gain/loss of β-cell mass in patients with prediabetes/diabetes, there is a pressing need for a noninvasive method for longitudinal imaging of β-cell transplants. Islet transplantation has been shown to lead to temporary insulin independence in patients with type 1 diabetes (53,54) but has not been widely adopted because of the need for donor tissues. Recent advances in selective stem cell differentiation techniques will likely lead to wider clinical adoption of β-cell transplantation therapies (55). The ability to noninvasively track the survival and function of transplanted β-cells would enable research into the patient-specific efficacy assessment of immune-modulating therapies and the development of new therapeutic strategies (56). In this case, the use of ⁵¹Mn²⁺-PET for monitoring islet transplant survival is compelling because it can be performed repeatedly and safely over time.

In conclusion, the prominent pancreatic uptake, rapid localization kinetics, and well-characterized metabolic pathways of ⁵²Mn²⁺ make it a promising agent for noninvasive PET imaging of functional β-cell mass and was corroborated by various post hoc experiments such as histological β-cell mass measurements and live-cell imaging of β-cell Ca²⁺ oscillations. With further development, ⁵¹Mn²⁺-PET may also find broad applications in venues ranging from basic research to clinical patient care.