We examined the effects of different physiological concentrations of glucose on cytoplasmic Ca2+ handling and mitochondrial membrane potential (Δψm) and insulin secretion in single mouse islet cells. The threshold for both glucose-induced changes in Ca2+ and Δψm ranged from 6 to 8 mM. Glucose step-jumps resulted in sinusoidal oscillations of cytoplasmic Ca2+, whereas Δψm reached sustained plateaus with oscillations interposed on the top of these plateaus. The amplitude of the Ca2+ rise (height of the peak) did not vary with glucose concentration, suggesting a “digital” rather than “analog” character of this aspect of the oscillatory Ca2+ response. The average glucose-dependent elevation of cytoplasmic Ca2+ concentration during glucose stimulation reached saturation at 8 mM stimulatory glucose, whereas Δψm showed a linear glucose dose-response relationship over the range of stimulatory glucose concentrations (4–16 mM). Glucose-dependent increases in insulin secretion correlated well with Δψm, but not with average Ca2+ concentration. These data show that an ATP-dependent K+ channel-independent pathway is operative at the single cell level and suggest mitochondrial metabolism may be a determining factor in explaining graded, glucose concentration-dependent increases in insulin secretion.
both mitochondrial and cytoplasmic metabolism play crucial roles in fuel-stimulated insulin secretion. Fuel metabolism-dependent increases in the ATP-to-ADP ratio close ATP-dependent K+ (KATP) channels, leading to membrane depolarization and Ca2+ influx through voltage-gated Ca2+ channels. A rise in the cytosolic free Ca2+ is a necessary component in the process of insulin secretion in response to elevated glucose concentrations (26, 36). Cytosolic free Ca2+ oscillations, which depend on periodic depolarization of the plasma membrane, occur in parallel with oscillations in insulin secretion during glucose stimulation (2, 38). Imposed Ca2+ pulses caused by repetitive depolarization with extracellular K+ are accompanied by synchronous pulses in insulin secretion (38), but insulin secretion also occurs when elevated cytoplasmic Ca2+ is maintained by sulfonylureas or high K+ in the presence of the KATP channel activator diazoxide (13, 39). It has been shown that this so-called amplifying pathway operates independently of imposed intracellular Ca2+ oscillations in intact islets (31). In the latter study, Ca2+ pulses were imposed by alternating K+ concentration between 4.8 and 30 mM in the presence of diazoxide. Increasing glucose concentrations augmented the amplitude of insulin pulses with no effect on the amplitude of Ca2+ pulses.
Questions arise as to what parameters correlate and therefore are likely determinants of full-scale fuel-dependent insulin secretion, when cytoplasmic Ca2+ does not. We therefore examined two responses [cytoplasmic Ca2+ and mitochondrial membrane potential (Δψm)] in single islet cells and with physiological stimulatory glucose concentrations. There are very few studies examining markers of mitochondrial metabolism at the single cell level. In addition, most studies addressing the relationship between the Ca2+ rise and insulin secretion have been done in intact islets where results may be influenced by intercellular interactions (1, 2, 8). It is well established that various functions of β-cells, including insulin release, vary with glucose concentration in a concentration-dependent manner. Several studies have suggested that the graded response of a whole population might represent concentration-dependent recruitment of individual β-cells, which have different levels of sensitivity to glucose (17, 29). A limited number of studies have addressed this issue on the single cell level and under physiologically relevant glucose concentrations (17). In the present study, we compared single-cell cytoplasmic Ca2+ andΔψm responses to glucose in the range of 4 to 16 mM and compared these responses with insulin secretion under the same conditions.
Swiss-Webster mice (Charles River) were killed with CO2 asphyxiation. All procedures were performed in accordance with the Institutional Guidelines for Animal Care at Boston University (IACUC no. AN-14071) in compliance with United States Public Health Service regulations.
Islet isolation and primary cell culture.
Pancreatic islets were isolated from 3- to 4-mo-old animals by collagenase (Roche Diagnostics, Penzberg, Germany) digestion, as previously described (21). Islets were picked by hand four times under a dissecting microscope and after overnight culture dispersed by incubation in Ca2+- and Mg2+-free PBS containing 3 mM EGTA and 0.05 mg/ml trypsin for 10 min at 37°C with occasional agitation. Isolated cells were centrifuged, washed, and suspended in RPMI 1640 culture media (Sigma) supplemented with 5 mM glucose, 10% FCS (HyClone), 100 IU/ml penicillin, and 100 μg/ml streptomycin. A 100-μl aliquot containing an estimated 1 × 105 cell suspension was plated on a poly-d-lysine-coated cover slip in a 35-mm petri dish (MatTek, Ashland, MA). Cells were allowed to adhere for 2 h before replenishment with 2 ml RPMI 1640 media containing 5 mM glucose and 10% FCS. Experiments were performed on individual cells not in contact with other cells after culture for 2–3 days. Under these conditions, β-cells constituted ∼78% of the total cell population, as determined by infecting cells with insulin promoter-directed green fluorescent protein (6). Relatively large cells were selected for study, since mouse β-cells are known to be larger than α- and δ-cells (5).
Fura loading and Ca2+ measurement.
Cells were loaded for 30 min at 37°C with 0.5 μM fura 2-AM (Molecular Probes, Eugene, OR) in Krebs-Ringer-bicarbonate (KRB) buffer containing 5 mM glucose, 140 mM NaCl, 30 mM HEPES, 4.6. mM KCl, 1 mM MgSO4, 0.15 mM Na2HPO4, 0.4 mM KH2PO4, 5 mM NaHCO3, 2 mM CaCl2, and 0.1% BSA, pH 7.4. After being loaded, cells were washed and incubated in this buffer for 15 min to allow cleavage of intracellular fura 2-AM by cytosolic esterases. The intracellular Ca2+ was measured by placing the dish containing cells on the stage of a Zeiss IM 35 microscope using a ×40 glycerin objective in a temperature-controlled cage heated to 37°C. Intracellular fura 2 was excited by a xenon lamp and a dual-wavelength Ionoptix synchronized chopper mirror at 340 and 380 nm. Fluorescence excitation intensity at 380 nm was equalized with that at 340 using a neutral density filter. The fura 2 emission signal at 510 nm was recorded by an intensified charge-coupled detector Ionoptix camera (IonOptix, Milton, MA). Images were collected at 4-s intervals. Fluorescence data were acquired and analyzed using Ion Wizard software (IonOptix). The free Ca2+ was calculated from the fluorescence ratio R as described by Grynkiewicz et al. (11). The ratios of maximum and minimum fluorescence were obtained by exposing cells to ionomycin or EGTA, respectively (30). A dissociation constant of 224 nM for Ca2+ binding to free fura was used in calculations (11, 30).
Mitochondria were labeled using the mitochondria-specific dye tetramethylrhodamine ethyl ester perchlorate (TMRE; Molecular Probes). Cells were incubated for 45 min before visualization in KRB buffer containing 10 nM TMRE. Confocal microscopy was performed on living cells using a Leica Confocal microscope (TCS SP2) with the Kr (568/20 mW) and Ar (488/20 mW) lasers. The Δψm-dependent component of TMRE will accumulate in a Nernstian fashion that can be described by the intensity of its fluorescence (28). The non-Δψm-dependent component of TMRE, also known as the binding component, was ignored, since it is fixed and voltage independent (28). A change in Δψm is in linear relationship to log(FITMRE_A/FITMRE_B), where FITMRE_A and FITMRE_B represent the fluorescence intensities of TMRE in two time points. This formula is based on the Nernst equation and is a modified version for ion concentrations across a potentiated membrane used in experiments that included the use of TMRE (28).
Glucose-stimulated insulin secretion was determined by static incubation. Dispersed islet cells were cultured in 48-well plates and stimulated with various glucose concentrations under the same conditions as described for Ca2+ and Δψm studies. The amount of released insulin was determined using an RIA kit (Linco Research, St. Charles, MO) using rat insulin as the standard.
Unless otherwise indicated, all reagents were obtained from Sigma Chemicals (St. Louis, MO).
Cells responded to increases in glucose concentration with transient elevations of Ca2+ that peaked at levels three- to fourfold above the basal value, with an average oscillatory period of 3.4 ± 1.1 min and an average lag of 3.2 ± 0.6 min (n = 59). These responses were heterogeneous, and, depending on applied glucose concentration, included single transient responses and regular and irregular oscillatory responses (Fig. 1). Regular oscillatory responses became more prevalent with increasing glucose concentration (7 mM glucose and above). No sustained elevations of Ca2+ were observed at any applied glucose concentration. Consistent with previous reports, in the majority of responding cells, the Ca2+ elevation was preceded by a brief initial decrease, most likely because of enhanced Ca2+ uptake by endoplasmic reticulum (ER) as a consequence of the increase in the ATP-to-ADP ratio (33, 41).
Recruitment of Ca2+ responses.
The number of responsive cells was dependent on the concentration of glucose (recruitment), with ∼18% of cells responding to 6 mM glucose, and ∼78% responding to 16 mM glucose. We evaluated Ca2+ responses to stimulatory glucose in the following two ways: single step-jumps and gradual glucose challenge. These two protocols did not yield significant differences, consistent with the fact that, in contrast to rat islet cells (9), mouse islet cells are not primed by stimulatory glucose to yield a subsequent increase in the response to the second glucose stimulus (3). The percentage of cells that responded to different glucose concentrations with any type of response (single transient, irregular and regular oscillatory) and only regular oscillatory responses are presented in Fig. 2. Maximal recruitment in both cases was reached by 8–10 mM glucose.
Digitality of the amplitude of the Ca2+ response.
Figure 3A shows the effect of different glucose concentrations on the amplitude (peak value less average basal) of individual Ca2+ transients. Although there was some variation among cells, the mean of the amplitude of the Ca2+ rise was not affected by the glucose concentration. The digital character of this aspect of the oscillatory response was clearly apparent when individual cells were challenged with stepwise increases in glucose concentration (Fig. 3C). As apparent from Fig. 3, neither the amplitude (Fig. 3A) nor the frequency (Fig. 3B) of the oscillations changed between 8 and 16 mM glucose. To ensure that this “digital” effect was not the result of saturation of the fura 2 signal, ionomycin was applied at the end of the trace recording. Ionomycin caused a two- to threefold increase in Ca2+ above the observed rise (data not shown). Illustrations of single-cell Ca2+ and Δψm responses to increasing glucose concentrations are presented in Fig. 3, C and D, respectively.
Average value of the Ca2+ response.
The mean intracellular Ca2+ was calculated from all (responsive and nonresponsive) cells at each glucose concentration and is presented in Fig. 4A. The mean intracellular Ca2+ rise, in contrast to the earlier described amplitude (peak) of the oscillatory wave, is the average intracellular Ca2+ value during the recording period and reflects recruitment (how many cells responded at each glucose concentration), frequency of the oscillatory response, and the amplitude (peak) of this oscillatory response. Average Ca2+ reached saturation at 8 mM glucose, as apparent from single cell analysis, where 88% of cells reached their saturation of Ca2+ response at 8 mM glucose, 7% at 12 mM glucose, and 5% at 16 mM glucose (Fig. 4A).
Changes in Δψm occurred within 30 s after stimulatory glucose addition. In agreement with previously published data (20, 27), the glucose-dependent increase in Δψm reached a steady-state level with slow oscillations on the top of this rise. Amplitude of these oscillations accounted for <8% of the total intensity value of this rise (data not shown). In contrast to the Ca2+ signal, Δψm in individual cells showed a graded response to stimulatory glucose concentrations (Fig. 3D) in 77% of cells. Only 5% of all cells reached their maximal Δψm response at 8 mM glucose and 18% of cells at 12 mM glucose (Fig. 4B). These data are in agreement with data in a different strain of mice (CD-1) in which 69% of 52 cells exhibited linear increases to increasing glucose concentration, 29% cells had nonlinear increments in the response, and 7% were nonresponsive. Only 20% of cells that exhibited nonlinear responses (6% of the total population) exhibited saturation of the Δψm response at 8 mM.
The relationship between the Ca2+ responses, Δψm responses, and insulin secretion in a population of single islet cells at each stimulatory glucose concentration was assessed. The mean levels of Ca2+ rise and Δψm at each glucose concentration was calculated from all (responsive and nonresponsive) cells in that set for the entire recording period and correlated with the level of glucose-stimulated insulin secretion, which was measured under the same conditions (Fig. 5). Consistent with previous reports, dispersed islet cells responded to glucose with a lower magnitude of insulin secretion than intact islets (16, 22), but followed the same concentration-dependent pattern. The mean Ca2+ reached saturation at 8 mM glucose, whereas insulin secretion and Δψm did not. Lack of correlation between the mean Ca2+ response and insulin secretion was apparent at 8 mM stimulatory glucose, and this difference increased as the concentration of glucose increased (Fig. 5A). In contrast, the linearity of the glucose-dependent rise in Δψm and the linear correlation between the level of insulin secretion and Δψm is shown in Fig. 5, B and C, respectively.
The principal differences between the responses produced by individual β-cells compared with intact islets lies in their heterogeneity (12, 14, 29, 37). Recently, it has been shown that a similar situation applies to the Ca2+ rise (17). Less attention has been paid to effects of glucose on recruitment and graded responses of mitochondrial metabolism. We evaluated both cytosolic Ca2+, Δψm, and insulin secretion in response to a wide range of physiological glucose concentrations. Our results showed that the Ca2+ response, expressed as the average Ca2+ elevation during the glucose stimulation, was reached by ∼8 mM glucose, similar to the earlier study (17). It has been shown in intact islet studies that increasing glucose concentration can augment insulin secretion pulses without affecting Ca2+ (31). That implies that the magnitude of insulin secretion and the level of Ca2+ cannot correlate under these conditions and support a role for the amplifying pathway to explain this phenomenon.
It has been reported in systems other than islet cells that increases in agonist concentrations augment the frequency rather than the amplitude of the Ca2+ oscillations, suggesting that the cell converts the original continuously graded (analog) signal into a frequency-encoded (digital) signal (4, 40). Although in islet cells we did not observe a frequency change in the single cell, consistent with studies in intact islets, we have shown that the glucose concentration had no effect on the average amplitude of the oscillatory peak of Ca2+ rise in responding cells, suggesting a digital (all or none) type of this aspect of Ca2+ response. This contrasts with a mixed model of cellular activation in insulin-secreting cells (35), according to which both dose-dependent amplification of the individual cell responses and recruitment of individual cells are responsible for the graded response of the cell population. This difference can be explained by the fact that clonal cells (BRIN-BD11) used in the study (35) are derived from rats, which have been shown to exhibit a priming effect during a single-stimulation protocol (9). Our finding is supported by the observation that subsequent application of higher (up to 20 mM) glucose to the same cell already displaying oscillations at a lower stimulatory glucose, did not change the amplitude or frequency of the Ca2+ response. The Ca2+ response was not saturated under our experimental conditions, since the application of ionomycin further increased the intracellular Ca2+ in these cells.
To explain how the amplitude of Ca2+ oscillations remains constant over a range of stimulatory glucose concentrations, it has recently been suggested that mitochondria regulate the amplitude of Ca2+ oscillations (4, 10, 25, 40). Mitochondria, located in close proximity to the ER release sites (7, 32, 34), are capable of fast Ca2+ uptake, which becomes activated if the average concentration of the cytosolic Ca2+ rises above a level of ∼500 nM (15, 19). Mitochondria therefore act as buffers, with the majority of mitochondrial Ca2+ bound to macromolecules, thus limiting the amplitude of Ca2+ oscillations and therefore keeping them constant. Recently, Ca2+ amplitude regulation was demonstrated for several types of mathematical models of Ca2+ oscillations, including receptor-operated and store-operated models, by incorporating a mitochondrial component in the equations. In these simulations, the amplitude of Ca2+ oscillations remains fairly constant over a broad range of stimulus concentrations (19). Similar findings were documented by Magnus and Kaiser (23, 24) in their model of Ca2+ handling in pancreatic β-cells. The Ca2+ ionophore, ionomycin, which equilibrates Ca2+ across the membranes of intracellular organelles, including ER and mitochondria, disrupts this type of regulation and can therefore increase Ca2+ levels above those we observed after addition of glucose.
Although we did not observe changes in the frequency of Ca2+ oscillations with increased glucose concentration, consistent with whole islet studies, we have noticed that more and more cells seemed to be recruited in the regular oscillatory state (in contrast to the single transient response) as the glucose concentration increased. In our single cell preparations, up to 67% of cells displayed regular oscillations in response to glucose. It has been reported that whole islets or islet clusters readily display oscillations in contrast to single cell preparations (18). It is generally believed that loss of cell-to-cell contact is responsible for the difference, but other explanations might be plausible, such as enzymatic, mechanical, and metabolic stress to which cells are exposed during the isolation and subsequent dissociation protocol, resulting in ∼50% cell death during the procedure (12, 22). Based on our experience, harsher dissociation conditions (higher concentrations of trypsin, longer dissociation periods than used for these studies) resulted in more cell death, and the remaining live cells had a tendency to display more irregular oscillations.
We did not find a good correlation between the intracellular Ca2+ rise and insulin secretion in a population of single islet cells, in contrast to the previously published data (17) in which authors compared clusters and individual cells and found a good correlation between insulin secretion and glucose-stimulated Ca2+ rise. The latter finding is surprising and not consistent with our observations. It is particularly puzzling, since the amplification pathway has been shown to operate at the single islet level, and an explanation is lacking as to why this does not apply to the single islet cell.
Mitochondrially derived ATP represents the major portion of ATP production in the cell. Because the determinant of mitochondrially derived ATP synthesis is largely the Δψm, under normal physiological conditions, where uncoupling is minimal, we wanted to assess how different cells responded to glucose with changes in Δψm. As in the case of average Ca2+, Δψm response was heterogeneous in terms of the percentage of additional increase in individual cells. In contrast to Ca2+, the single cell Δψm response to glucose was graded and continued to rise up to 16 mM glucose. These data also suggest that only a portion of the glucose-derived rise in Δψm, and consequently ATP produced, is used to close KATP channels. The concentration of ATP required to close channels represents a fixed value, whereas any excess ATP may play a role as a component in the amplification pathway.
In conclusion, in this work, we described the digitality of the glucose-induced amplitude of Ca2+ oscillatory responses and showed that the different levels of sensitivity to glucose in individual cells can account for the glucose-induced Ca2+ response in a population. Although the Ca2+ signal reached saturation at 8 mM glucose, Δψm and insulin secretion increased beyond this concentration. These findings correlate with studies in intact islets and are consistent with the role of a KATP channel-independent pathway operating at the single cell level. The correlation between Δψm and insulin secretion suggests that ATP or other mitochondrial products may have other roles in addition to KATP channel closure in insulin secretion. Alternatively, other aspects of mitochondrial metabolism may also play a role.
This work was supported in part by National Institutes of Health Grant RO-1-DK-35914 (B. E. Corkey), DK-63356 (B. E. Corkey and O. Shirihai), and T32 HL-07224 (E. Heart).
We thank M. Meow and Drs. K. Tornheim, J. Deeney, S. Katzman, S. Katz, and H. Mohamed for helpful discussions.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 by American Physiological Society