Defects in β-Cell Ca²⁺ Dynamics in Age-Induced Diabetes

People over the age of 65 years have an increased risk for the development of type 2 diabetes if their pancreatic β-cells fail to compensate for insulin resistance (1,2). The decline in insulin secretion from the β-cell is thus considered a major contributing factor to disease development (3,4). In β-cells, fluctuations in cytoplasmic free calcium concentration ([Ca²⁺]_i) have profound physiological consequences. It has been suggested that aging is associated with compromised mitochondrial function (5,6). If this were to be true also for the β-cell, it would lead to derangements in Ca²⁺ dynamics. The loss of the delicate and precise regulation of the oscillatory Ca²⁺ signal, which includes inositol 1,4,5-trisphosphate (InsP₃)–mediated Ca²⁺ mobilization from intracellular stores (7,8), has been suggested to be an early indicator of diminished islet glucose sensitivity and endoplasmic reticulum (ER) dysfunction (9). Hence, when trying to define a link among aging, β-cell dysfunction, and diabetes, a logical approach is to focus on the Ca²⁺ signal and its regulation at the molecular level. We have now tested the hypothesis that loss of the delicate balance between mitochondrial homeostasis and regulation of the oscillatory Ca²⁺ signaling is of particular importance in underlying disturbed β-cell function in age-induced diabetes. For this purpose, we have used three mouse models that demonstrate how the aging phenotype can be attenuated or accelerated. The first model, representing premature aging, is a homozygous knock-in mouse model expressing a proof reading–deficient form of the catalytic subunit of mitochondrial DNA (mtDNA) polymerase γ (10). This mutant polymerase γ induces error-prone mtDNA synthesis, which in turn leads to accumulation of somatic mtDNA mutations with increasing age. The increased levels of somatic mtDNA mutations lead to reduced life span and premature onset of aging-related phenotypes (i.e., weight loss, reduced subcutaneous fat, alopecia, kyphosis, anemia, osteoporosis, reduced fertility, and heart enlargement) (10). The second mouse model, the naturally mature aging C57BL/6 mouse, represents a standard mouse that has the same genetic background as the mtDNA mutator mouse. The third mouse model, the 129 mouse, has been shown to be less susceptible to the development of diabetes when compared with the C57BL/6 mouse (11–13). We now observe that pancreatic islets from the 129 mouse display a pattern with [Ca²⁺]_i fast oscillations (FastOscs) that remains intact up to old age. This is also accompanied by a maintained mitochondrial function and insulin release, confirming that this model is resistant to aging. In short, the mtDNA mutator mouse displays accelerated aging, whereas the 129 mouse appears more resistant to aging compared with the C57BL/6 mouse. Hence, in the current study we use these three mouse models to demonstrate the associations among aging, mitochondrial dysfunction, [Ca²⁺]_i handling, and a diabetic phenotype.

mtDNA mutator mice and wild-type littermate control mice at 2–10 months of age were generated as described previously (10). The littermate wild-type control mice were from the same mother as mtDNA mutator mice. Normally aging mice (i.e., 24-month-old wild-type littermate control mice of mtDNA mutator mice) as well as C57BL/6 and 129/J mice at 17–24 months of age, were also used. The old and young mice originate from the same colony of mice. We performed experiments on islets from old and young animals within the same period of time in order to have fully comparable experimental conditions. All experiments were approved by the local animal ethics committees at Karolinska Institutet. Pancreatic islets were isolated and cultured, as previously described (14).

We performed glucose tolerance tests in animals essentially as previously described (15), with the exception that 6-h fasting was used before the tests. We performed insulin tolerance tests in animals fasted for 6 h. After the measurements of blood glucose at 0 min (15), 1 unit of insulin (Actrapid; Novo Nordisk A/S) per kilogram of body weight was injected intraperitoneally. Blood glucose levels were measured at 10, 20, 30, and 60 min.

Cytochrome c oxidase (COX) and succinate dehydrogenase (SDH) activities were measured by enzyme histochemical double staining of cryostat sections (4 μm) from pancreas obtained at the ages of 3 and 10 months, as described previously (16).

Mitochondrial membrane potential in islets isolated from animals at the age of 10 or 20–24 months was analyzed using rhodamine 123 (Rh123), as described previously (15). Islet autofluorescence from NAD(P)H and flavin adenine dinucleotide (FAD) was monitored, and the NAD(P)H/FAD redox ratio was calculated (17,18). The experimental setup was the same as described below for measurements of [Ca²⁺]_i, but without loading with any indicator. Excitation wavelengths were 360 nm for NAD(P)H and 485 nm for FAD. Fluorescence was detected after band-pass filters 455/110 and 550/80 for NAD(P)H and FAD, respectively.

For [Ca²⁺]_i measurement, islets were incubated with 2 μmol/L Fura-2-acetoxymethyl ester (AM) and changes in [Ca²⁺]_i (i.e., a fluorescence ratio of 340:380) were analyzed, as described previously (14). In addition, some experiments were performed on a microscope system consisting of a Zeiss Axiovert 200M microscope with a fluorescence imaging system using Andor iQ software with an Andor iXon DV887DCS-BV camera and a Cairn monochromator for excitation (Andor Technology Ltd., Belfast, U.K.). Data from the latter system were corrected to become comparable to data obtained from the previous system (14).

For simultaneous measurements of [Ca²⁺]_i and intracellular store lumen calcium concentration ([Ca²⁺]_L), islets were incubated with 10 μmol/L rhod-3 AM (Molecular Probes; K_d = 0.57 μmol/L) and 20 μmol/L mag-fluo-4 AM (Molecular Probes; K_d = 22 μmol/L) at 37°C for 60 min. Excitation wavelengths were 485 nm for mag-fluo-4 and 560 nm for rhod-3. Fluorescence was detected after band-pass filters 455/110 and 550/80 for mag-fluo-4 and rhod-3, respectively (19,20). Because the K_d of each Ca²⁺ indicator matches the Ca²⁺ concentration ranges within the cytoplasm or the organelles, the responses from rhod-3 and mag-fluo-4 are expected to mainly reflect changes in [Ca²⁺]_i and [Ca²⁺]_L, respectively. Rhod-3 offers better cytosolic localization than existing Ca²⁺-sensing dyes since the uptake into organelles has been reduced. Compared with existing red Ca²⁺ dyes, rhod-3 exhibits a large increase (>2.5 fold) in fluorescence upon binding Ca²⁺ and very low fluorescence at rest (without Ca²⁺ binding). Mag-fluo-4 in the cytoplasm, although much less than that in the organelles, may contribute a contaminating [Ca²⁺]_i signal when one attempts to measure [Ca²⁺]_L. After staining with mag-fluo-4, islets were incubated for 30–45 min in a medium without mag-fluo-4 at 37°C to accelerate the extrusion of mag-fluo-4 from the cytoplasm into the extracellular space. Thereafter, the Ca²⁺-responsive mag-fluo-4 signal should predominantly reflect changes in [Ca²⁺]_L. This is also supported by our results (see below) where carbamylcholine (CCh) stimulation leads to an increase in the rhod-3 signal and a simultaneous decrease in the mag-fluo-4 signal, consistent with the release of Ca²⁺ from the ER lumen into the cytoplasm.

All [Ca²⁺]_i data were subjected to visual inspection, and [Ca²⁺]_i oscillations were analyzed using power spectral analysis in Matlab (MathWorks, Lowell, MA) with code adapted for analysis of the oscillation patterns commonly observed in pancreatic islets (M. Köhler, P.-O. Berggren, unpublished data [partly based on spectral analysis]) (21). The amplitude values of FastOsc and slow oscillation (SlowOsc) were calculated as the square root of the total power of periods from 6 to 60 s (FastOsc) and from 60 to 600 s (SlowOsc), respectively (22). These values are directly proportional to the amplitude computed as the root mean square, justifying the use of these values for comparisons of amplitudes between different experiments. Power spectral density for FastOscs was calculated by the method of Welsh (23), and the standard fast Fourier transform power spectrum was used for SlowOscs. The dominant fast and slow periods were obtained from peaks in the respective power spectrum.

Insulin release from islets was analyzed, as described previously (15). Insulin measurements were performed using a microsphere-based two-photon excitation fluorometer (TPX technology; ArcDia International Oy Ltd., Turku, Finland) using a human insulin standard (Sigma-Aldrich).

All values are given as the means ± SEM. The Student unpaired two-tailed t test was used, and a P value <0.05 was considered to be statistically significant for all comparisons.

We characterized mitochondrial respiratory chain function in pancreatic islets of mtDNA mutator and wild-type littermate control mice (with C57BL/6 background) at the ages of 12 and 40 weeks (Fig. 1A) by using enzyme histochemical double staining for COX and SDH activity. The catalytic subunits of COX are mtDNA encoded, whereas all subunits of SDH are nucleus encoded. The active COX and SDH were visualized as brown and blue staining, respectively. Cells in dark brown are those that have intact COX and SDH enzyme activities since the COX brown staining overwhelms the SDH blue staining (16). Extensive mosaic COX deficiency, as indicated by the blue staining, was found in islets of 40-week-old mtDNA mutator mice in comparison with wild-type control mice (Fig. 1Ac and Ad). These results suggest that the accumulation of mtDNA mutations can lead to an age-associated mitochondrial respiratory chain dysfunction in pancreatic islets. In addition, we investigated the depolarization of mitochondrial membrane potential by carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP) in islets from mtDNA mutator mice at the age of 10 months. Rh123 was used as the fluorescent probe to measure mitochondrial membrane potential (15). The rationale for this experiment is that under conditions of impaired mitochondrial function there is less polarization of mitochondrial membrane potential, and, consequently, the depolarizing effect of FCCP, a mitochondrial uncoupler, should not be as pronounced. In line with this reasoning, the depolarizing effect on mitochondrial membrane potential induced by 1 μmol/L FCCP was decreased in islets of mtDNA mutator mice (Fig. 1B and C), supporting a deficiency in mitochondrial respiratory chain. Moreover, we measured a reduced hyperpolarization of mitochondrial membrane potential in response to glucose stimulation (Fig. 1D and E), indeed supporting a respiratory chain deficiency in islets of mtDNA mutator mice.

To further determine whether the premature aging model of mtDNA mutator mice reflects an accelerated normal aging process in pancreatic β-cells, we first investigated glucose-stimulated hyperpolarization of mitochondrial membrane potential in islets of 24-month-old C57BL/6 mice compared with 6-month-old C57BL/6 mice. The old C57BL/6 mice demonstrated a decreased hyperpolarization of mitochondrial membrane potential in response to both 11 and 16.7 mmol/L glucose compared with the young C57BL/6 mice (Fig. 2A–C). We next measured glucose-stimulated changes in endogenous NAD(P)H/FAD autofluorescence, reflecting the islet cellular redox ratio (17,18,24). After glucose stimulation, islets responded with an initial rapid rise, followed by a sustained elevation in NAD(P)H/FAD (Fig. 2D–G). Compared with 6-month-old C57BL/6 mice, 24-month-old C57BL/6 islets had decreased responses in NAD(P)H/FAD induced by both 11 and 16.7 mmol/L glucose (Fig. 2D, E, and H). An increase in NAD(P)H/FAD stimulated by 16.7 mmol/L glucose, when expressed as a percentage of the maximum response to 3 mmol/L NaN₃, was also diminished in the old C57BL/6 mice (Fig. 2J). We observed that enhanced Ca²⁺ entry in the presence of 10 mmol/L extracellular CaCl₂ led to a substantial elevation in NAD(P)H/FAD (Fig. 2D–I). The old C57BL/6 mice had a decreased response to 10 mmol/L CaCl₂ in the presence of 16.7 mmol/L glucose. Hence, the decreased level of NAD(P)H/FAD, as a result of mitochondrial respiration deficiency in aged pancreatic β-cells, may reflect an impaired Ca²⁺ entry (25,26). In contrast, the 129 mouse islets maintain both glucose-stimulated hyperpolarization of mitochondrial membrane potential (Fig. 2A–C) and NAD(P)H/FAD (Fig. 2F, G, I, and J) during aging. Hence, a compromised mitochondrial function was described in β-cells of both mtDNA mutator mice and old C57BL/6 mice.

Next we investigated how the age-related defect in mitochondrial function affected [Ca²⁺]_i dynamics (Fig. 3). We found a reduced initial decrease in [Ca²⁺]_i (i.e., fura-2 ratio) in islets from 10 month-old mtDNA mutator mice compared with islets from age-matched wild-type littermate control mice when stimulated by 11 mmol/L glucose (Fig. 3A and B, two bar graphs on the left). We thereafter observed a diminished increase in peak [Ca²⁺]_i values in mtDNA mutator mouse islets stimulated by 11 mmol/L glucose compared with islets from wild-type littermate control mice (Fig. 3A and B, two bar graphs on the right). There is no difference in basal [Ca²⁺]_i between 10-month-old wild-type and mtDNA mutator mouse islets (Supplementary Fig. 1). Furthermore, we studied the period and amplitude of [Ca²⁺]_i oscillations, and in this context islets from mtDNA mutator mice had an increased period and a lower amplitude of [Ca²⁺]_i FastOscs compared with islets from wild-type control mice (Fig. 3A, C–E, J, and K). We performed a power spectrum analysis of the [Ca²⁺]_i oscillation data from mice at the age of 10 months (Fig. 3C–E and J–M). The dominant average period for [Ca²⁺]_i FastOscs in wild-type control islets was 18.4 s, whereas the dominant average period in mtDNA mutator islets was 25.7 s (Fig. 3E). Figure 3K and M shows statistics from the total power of the spectrum in the FastOsc range and the SlowOsc range. The oscillation period in the mtDNA mutator islets is compatible with the oscillation period observed in islets from 2-year-old normally aging mice (i.e., wild-type littermate control mice for mtDNA mutator mice and C57BL/6 mice at 2 years of age) (Fig. 3F–I and N). By contrast, no difference in period and amplitude in [Ca²⁺]_i oscillations in response to glucose stimulation was found between 24-week-old mtDNA mutator islets and the age-matched wild-type littermate control islets (Supplementary Fig. 2). Moreover, no difference in KCl-induced [Ca²⁺]_i increase was found between islets from 10-month-old mtDNA mutator mice and islets from age-matched wild-type littermate controls (Fig. 4A). The islets from the 129 mouse did not show any age-induced impairment in [Ca²⁺]_i dynamics when compared with islets from normally aging mice (Fig. 4). These data clearly indicate that the age-dependent increase in [Ca²⁺]_i oscillation period (decrease in frequency) after glucose stimulation is a common feature in aging β-cells, which is well compatible with age-related defects in mitochondrial function (26). SlowOscs in [Ca²⁺]_i have periods of >60 s. It is of interest to note that there was no difference in the frequency of [Ca²⁺]_i SlowOscs in islets from 6- and 10-month-old mtDNA mutator mice compared with islets from age-matched wild-type littermate control mice (Fig. 3L and Supplementary Fig. 2F). Also, there was no age-dependent difference in frequency of [Ca²⁺]_i SlowOscs in either C57BL/6 or 129 mice (Fig. 4E).

To dissect a possible effect of aging on Ca²⁺ mobilization from intracellular stores, we investigated the effect of CCh, a cholinergic agonist that activates the acetylcholine receptor, on phospholipase C (PLC)/InsP₃–mediated Ca²⁺ release. The response to CCh is quantified as a ratio difference between baseline and peak, meaning that, in that sense, it includes baseline subtraction. The average baselines for the two groups are not different, meaning that the baseline itself cannot explain the difference. One should keep in mind here that [Ca²⁺]_i is monitored by the fura-2 ratio, which does not require further normalization. We found reduced peak [Ca²⁺]_i values in 10-month-old mutator mouse islets after stimulation by 200 μmol/L CCh and perifusion with and without extracellular Ca²⁺ compared with the age-matched control islets (Fig. 5A and B). Moreover, there was also decreased average basal Ca²⁺ influx over the plasma membrane in islets from the prematurely aging mice (Fig. 5A and B). Furthermore, to evaluate the effect of aging on ER Ca²⁺ release, we simultaneously measured [Ca²⁺]_i and [Ca²⁺]_L in islets from 4-month-old and 21-month-old C57BL/6 mice (Fig. 5C–F). Stimulation with CCh led to an increase in [Ca²⁺]_i and a decrease in [Ca²⁺]_L. We found a diminished Ca²⁺ release from intracellular stores and consequently a reduced [Ca²⁺]_i in islets from 21-month-old C57BL/6 mice after stimulation with CCh compared with islets from 4-month-old C57BL/6 mice (Fig. 5C–F). These results demonstrate that aging impairs PLC/InsP₃-mediated Ca²⁺ release from intracellular stores after stimulation at basal levels (3 mmol/L) (Fig. 5A and B) or high glucose levels (16.7 mmol/L) (Fig. 5C–F) in pancreatic β-cells.

To assess the effect of the ER Ca²⁺-ATPase (SERCA) on the replenishment of intracellular Ca²⁺ stores with Ca²⁺ when [Ca²⁺]_i increases, islets from 4-month-old (Fig. 6A) and 21-month-old (Fig. 6B) C57BL/6 mice were initially perifused for 5 min with a Ca²⁺-free medium containing 2 mmol/L EGTA, 16.7 mmol/L glucose, and 250 μmol/L diazoxide. They were then perifused with a medium containing 2.56 mmol/L CaCl₂, 16.7 mmol/L glucose, and 250 μmol/L diazoxide through the experiment. High glucose levels stimulate ATP production in the β-cells, and diazoxide, a potassium channel activator, leads to hyperpolarization of plasma membrane and prevents Ca²⁺ influx. Therefore, this protocol efficiently stimulates Ca²⁺ uptake by the ER in the β-cells (27,28). Readmission of extracellular Ca²⁺ led to an increase in [Ca²⁺]_i and a subsequent increase in [Ca²⁺]_L. We found a reduced [Ca²⁺]_i (Fig. 6C) and decreased Ca²⁺ uptake into intracellular stores (Fig. 6D) after readmission of extracellular Ca²⁺ in islets from 21-month-old C57BL/6 mice compared with islets from 4-month-old C57BL/6 mice. Figure 6E and F show the effect of aging on store-operated calcium entry (SOCE) in islets from C57BL/6 mice. There was a reduced level of SOCE induced by 1 μmol/L thapsigargin in islets from 21-month-old C57BL/6 mice compared with islets from 4-month-old C57BL/6 mice.

Since adequate β-cell Ca²⁺ dynamics is a prerequisite for appropriate insulin release, it is logical that pancreatic islets isolated from mtDNA mutator mice demonstrated an age-dependent progressive reduction in glucose-stimulated insulin secretion (Fig. 7A and B). Also, when normalized to insulin content, insulin release in response to both glucose stimulation and depolarization with KCl was decreased in islets from mtDNA mutator mice compared with islets from control mice (Fig. 7C–F). Since the insulin content is not significantly different between the two groups of islets, this indicates that the reduced insulin secretion cannot be explained by decreased insulin content (Supplementary Fig. 3). Notable, however, is that these values show high variability. A similar decrease in glucose-induced insulin release was observed in normally aging mice (Fig. 7G–I). There was no difference in either basal islet insulin release or in islet architecture in pancreatic sections between 10-month-old mtDNA mutator mice and the age-matched wild-type littermate control mice (Supplementary Figs. 4 and 5). In addition, there was no significant change found in glucose-stimulated insulin secretion in islets from young mutator mice at 2–6 months of age compared with the age-matched wild-type control mice (Supplementary Fig. 6). It is noteworthy that the aging 129 mice demonstrated an intact glucose-induced insulin release (Fig. 7G–I).

In mtDNA mutator mice, no impaired glucose tolerance was observed at the age of 30 weeks (Fig. 8A), whereas at an older age mtDNA mutator mice were glucose intolerant (Fig. 8B). We measured a reduced plasma insulin concentration 30 and 60 min after glucose injection during glucose tolerance tests (Fig. 8C), suggesting a deficiency in glucose-induced insulin secretion in vivo. No difference in insulin tolerance was found between 10-month-old mtDNA mutator and the age-matched wild-type littermate control mice (Fig. 8D), suggesting that the deficiency in insulin secretion in mtDNA mutator mice results in impaired in vivo glucose homeostasis. Also, normally aging C57BL/6 mice demonstrated a similar impaired glucose tolerance (Fig. 8E and F). However, the aging 129 mice showed no impairment in overall glucose homeostasis (Fig. 8E and F).

A proper regulation of the [Ca²⁺]_i signal is essential for pancreatic β-cell function and survival. Well-functioning pancreatic islets respond to glucose stimulation with an increase in [Ca²⁺]_i that triggers insulin exocytosis (29). After the initial peak response, [Ca²⁺]_i begins to oscillate. The dynamic behavior of the [Ca²⁺]_i signal involves both Ca²⁺ influx through Ca²⁺ channels in the plasma membrane and Ca²⁺ mobilization from intracellular stores (30–32), and is critically dependent on intact glucose metabolism (i.e., mitochondrial function). We now demonstrate an example of a diabetic phenotype that follows upon aging and that is associated with a defect in [Ca²⁺]_i dynamics.

A focal distribution of respiratory chain deficiency is a well-known phenomenon in normally aging human tissues, such as those in the heart, skeletal muscle, colonic crypts, and neurons (33–37). Interestingly, both an age-related decline in mtDNA copy number in human islets (38) and a correlation of mtDNA mutations with human diabetes (39) have been shown. Our mitochondrial membrane potential and NAD(P)H/FAD measurements obtained in islets from both mtDNA mutator mice and normally aging mice clearly show a compromised mitochondrial function associated with aging. In contrast, islets from the 129 mouse model appear to maintain their mitochondrial function up to old age. An increase in the concentration of extracellular Ca²⁺ to 10 mmol/L was used and is shown in Fig. 2D–G. The reason for increasing extracellular Ca²⁺ concentration to 10 mmol/L was to exaggerate the possible influence of Ca²⁺, and then, particularly, to determine whether this increase in extracellular Ca²⁺ leads to a difference in Ca²⁺-dependent NAD(P)H/FAD fluorescence in young and old islets, respectively. We believe that the results obtained in the presence of 10 mmol/L extracellular Ca²⁺ support our hypothesis, namely, that there is an age-dependent deterioration in β-cell Ca²⁺-regulated mitochondrial function.

The intimate relationship between intact mitochondrial function and the regulation of [Ca²⁺]_i dynamics suggests that the progressive deterioration in [Ca²⁺]_i oscillations that we now observe is indeed a key feature of an age-related decline in pancreatic β-cell function. In this context, it is of interest to note that the 129 mouse model did not show any age-induced impairment in [Ca²⁺]_i dynamics compared with normally aging mice. In addition, we investigated whether the increased period (decreased frequency) and decreased amplitude of [Ca²⁺]_i oscillations in aging islets could be explained by a deteriorated interplay between the two major pathways of increasing [Ca²⁺]_i, namely, glucose-induced Ca²⁺ influx through the voltage-gated L-type Ca²⁺ channels and PLC/InsP₃-mediated Ca²⁺ mobilization from intracellular stores (40–44). Our results regarding intracellular Ca²⁺ mobilization in aging islets from both mtDNA mutator mice (Fig. 5A and B) and C57BL/6 mice (Figs. 5C–F and 6) are compatible, with less releasable Ca²⁺ in the ER pool. This may be explained by the observed decreased average basal Ca²⁺ influx over the plasma membrane in islets from the prematurely aging mice (i.e., less Ca²⁺ is available to fill the ER pool), and as well by a reduction in the ER Ca²⁺ uptake due to compromised mitochondrial function. The reduced activity of SERCA is compatible with the significantly reduced initial decrease in [Ca²⁺]_i observed after glucose stimulation in mtDNA mutator islets (Fig. 3B) and the decreased Ca²⁺ uptake in intracellular stores in aging islets from C57BL/6 mice (Fig. 6C). The β-cell SERCA pump has been demonstrated to serve as a high-affinity sink for Ca²⁺ that is stimulated by ATP (28), and the initial decrease in response to glucose stimulation has been shown to result from an elevation of SERCA activity (45). The observed alterations in [Ca²⁺]_i dynamics in the aging β-cell are also consistent with the observation that SERCA inhibitors transform a mixed oscillatory pattern into a slow pattern (46). Moreover, the reduced PLC/InsP₃-mediated Ca²⁺ release from ER in aging islets from C57BL/6 mice may lead to an attenuated SOCE (Fig. 6E and F) (47). Microarray data from Rankin and Kushner (48) demonstrated reduced mRNA levels in genes vitally important for the maintenance of the electron transport chain of mitochondrial oxidative phosphorylation (i.e., cytochrome c oxidase subunit III and cytochrome b) in aged islets compared with young islets. This is compatible with our results on COX deficiency in the aged β-cell in mtDNA mutator mice (Fig. 1). Moreover, the expression of the β-cell glucose transporter 2 was decreased in an age-dependent manner (48). In summary, our study suggests that the age-dependent reduction of mitochondrial function is sufficient to lower ATP production in β-cells. This in turn leads to reduced SERCA activity and, thereby, altered ER Ca²⁺ handling, which eventually leads to a reduced frequency in [Ca²⁺]_i oscillations and a reduction in insulin exocytosis. However, one should keep in mind that under conditions of decreased mitochondrial respiration other [Ca²⁺]_i regulatory mechanisms might also be compromised and involved in both the decreased basal Ca²⁺ influx and lower [Ca²⁺]_i amplitudes obtained in response to glucose stimulation in the mtDNA mutator mice.

Since β-cell function is strictly dependent on an appropriate regulation of the Ca²⁺ signaling, the age-associated decrease in glucose-induced insulin release is logical and should follow upon a deterioration in [Ca²⁺]_i dynamics. However, the decreased insulin release may also be influenced by other pathways that are altered in the aging β-cell (6). We would expect lower SERCA activity as a result of mitochondrial dysfunction. The deterioration in insulin secretion can result from either altered ER Ca²⁺ uptake or directly from the mitochondrial dysfunction, or from both. Our results show that KCl-induced insulin release is clearly altered in the mtDNA mutator mice, whereas KCl-induced [Ca²⁺]_i increase is not significantly altered (Fig. 4A). This suggests that a similar amplitude of nonoscillating elevated [Ca²⁺]_i under these conditions will still lead to less insulin release from the mutator mouse. Insulin release data normalized to insulin content in islets from mtDNA mutator mice compared with control mice, demonstrating a decrease upon stimulation, suggests that altered total insulin content cannot by itself explain the lower values of insulin secretion. Instead, this further substantiates a decreased insulin exocytotic capacity in association with aging. Interestingly, in accordance with the data on [Ca²⁺]_i dynamics, the aging 129 mice also demonstrated an intact glucose-induced insulin release. Admittedly, there is variability in the insulin content data, and one can therefore not completely rule out a somewhat decreased insulin content in the mtDNA mutator mice. A possible decreased total insulin content or, alternatively, a decreased size or refilling of the readily releasable pool of insulin could be associated with aging and impaired β-cell metabolism per se (29). This is supported by the clear difference seen in the KCl-induced insulin release (49). Altered ER Ca²⁺ handling can also explain a decreased rate of refilling of the readily releasable pool of insulin since 90% of the insulin granules depend on Ca²⁺ efflux from the ER for their mobilization to the readily releasable pool from the reserve pool (50). Decreased insulin content may also be secondary to a compromised positive autocrine feedback loop for insulin gene expression by an impaired glucose-induced insulin release in the mtDNA mutator mice (51), which results from impaired mitochondrial respiration and, thereby, deteriorated [Ca²⁺]_i dynamics.

To clarify the extent to which compromised insulin release from aging β-cells also impacted overall glucose homeostasis in the living mouse, we performed intraperitoneal glucose tolerance tests in mtDNA mutator and wild-type littermate control mice at different ages. At the age of 30 weeks, mtDNA mutator mice already displayed the aging-associated phenotypes (10), but no impaired glucose tolerance was found. However, at an older age mtDNA mutator mice were glucose intolerant. It is noteworthy that mtDNA mutator mice did not show any impaired insulin sensitivity compared with wild-type control mice, emphasizing that a defect in glucose homeostasis associated with aging is accounted for by the observed β-cell dysfunction. On the other hand, Fig. 8B shows a normalized level of glucose at 60 min despite very poor insulin secretion (Fig. 8C) and a significant decrease in fasting blood glucose in mtDNA mutator mice compared with wild-type control mice. The mechanisms underlying these observations are unknown. Cooperation between brain and islet plays a critical role in the control of both energy and glucose homeostasis (52). Therefore, mitochondrial mutations/deletions in the central nervous system and other organs in mtDNA mutator mice might be crucial in the brain-centered regulation of glucose homeostasis via insulin-independent mechanisms (53).

The age-induced impairments in β-cells in mutator mice are further corroborated by the findings in the aging 129 mice, which showed maintenance of β-cell [Ca²⁺]_i dynamics and no impairment in either glucose-induced insulin release or overall glucose homeostasis. In addition, the aging 129 mice maintain their mitochondrial metabolism, as demonstrated by glucose-stimulated hyperpolarization of mitochondrial membrane potential and changes in NAD(P)H/FAD. These observations in the aging 129 mice compared with the observations in the mtDNA mutator and normally aging mice also suggest a causal link between adequate [Ca²⁺]_i dynamics and overall β-cell function. Our study thus suggests that a defective metabolism–induced deterioration in [Ca²⁺]_i dynamics reflects an important age-dependent phenotype that may have a critical role in the development of β-cell dysfunction associated with type 2 diabetes, most often having a debut in elderly individuals. In this context, it is conceptually important that minor metabolically driven changes in the specific fine tuning of β-cell [Ca²⁺]_i dynamics over time are associated with major alterations of a process as robust as insulin exocytosis, and thereby the overall regulation of glucose homeostasis in the living animal.

Acknowledgments. The authors thank Jake Kushner for providing microarray data (48) and Yvonne Strömberg for technical assistance.

Funding. This study was supported by the Swedish Research Council; the Novo Nordisk Foundation; Karolinska Institutet; the Swedish Diabetes Association; The Family Knut and Alice Wallenberg Foundation; Eurodia; the European Foundation for the Study of Diabetes; the Diabetes Research and Wellness Foundation; the Bert von Kantzow Foundation; VIBRANT (grant FP7-228933-2); the Skandia Insurance Company Ltd.; Lee Kong Chian School of Medicine, Nanyang Technological University/Imperial College London, Singapore; the Strategic Research Programme in Diabetes at Karolinska Institutet; the ERC-2013-AdG 338936-BetaImage, EU; the Stichting af Jochnick Foundation; and the Erling-Persson Family Foundation.

The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

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

Author Contributions. L.L. helped to design the study and perform and interpret the experiments and contributed to the writing of the manuscript. A.T., Y.W., J.P.B., C.I., L.J.-B. and N.-G.L. performed and interpreted the experiments. M.K. helped to perform and interpret the experiments and contributed to the writing of the manuscript. P.-O.B. helped to design the study and contributed to the writing of the manuscript. P.-O.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented as an oral presentation at the 46th European Association for the Study of Diabetes Annual Meeting, Stockholm, Sweden, 12–16 September 2010.