Increased Glucose Sensitivity of Stimulus-Secretion Coupling in Islets From Psammomys obesus After Diet Induction of Diabetes

Male DP and DR P. obesus, obtained from Harlan Laboratories (Jerusalem, Israel), were kept on a regular artificial light cycle (0700–1900) and used at 10–16 weeks of age. DP P. obesus were initially fed an LE diet (2.38 kcal/g) (Kafolk, Petach Tikva, Israel) ad libitum to maintain normoglycemia (DP-LE) and were eventually transferred for 5 days to an HE diet (2.93 kcal/g) (Weizmann Institute for Science, Rehovot, Israel) to induce diabetes (DP-HE). DR P. obesus were initially kept on HE diet to ascertain that they remained normoglycemic. Before use, they were transferred to LE diet for 15 days (DR-LE) and were eventually shifted back to HE diet for 5 days (DR-HE). Blood glucose levels (9:30 a.m.) were determined in the fed state using a glucometer (Glucometer Elite, Bayer, NY).

All experiments were performed at 37°C with a bicarbonate-buffered Krebs solution (18), gassed with O₂:CO₂ (94:6) to maintain pH at 7.4, and supplemented with 1 g/l BSA and various glucose concentrations. For islet isolation, P. obesus were anesthetized (100 mg/kg pentobarbital i.p.), laparotomized, and killed by heart incision. The pancreas was inflated with 5 ml Krebs containing 5 mmol/l glucose and 1.5 mg/ml collagenase (Serva, Heidelberg, Germany), removed, minced with scissors, and digested for 4–6 min at 37°C before microdissection of the islets at 4°C under a stereomicroscope.

Except for control experiments in which fresh islets were used, the islets were cultured overnight in RPMI-1640 medium containing 100 IU/ml penicillin, 100 μg/ml streptomycin, 10% FCS, and either 5 or 10 mmol/l glucose, as indicated throughout the report after the P. obesus strain-diet acronym (e.g., DP-LE5, DP-LE10).

After culture, batches of 25–30 islets were perifused (flow rate ∼1 ml/min) with a medium continuously gassed with O₂:CO₂ (94:6). The effluent was collected (2-min fractions) for measurement of insulin by radioimmunoassay using human insulin as standard (19). At the end, the islets were collected for determination of insulin content (18).

[Ca²⁺]_c was measured by microspectrofluorimetry, as previously described (20), using Quanticell 700 m (Visitech, Sunderland, U.K.). The islets were loaded for 3 h with 2 μmol/l fura-PE3 acetoxymethylester (Teflabs, Austin, TX) at 37°C in the presence of the same glucose concentration as that during culture. They were then perifused (∼1 ml/min) at 37°C and stimulated with increasing glucose concentrations. The emission at 510 nm was recorded during alternate excitation at 340/380 nm, and the [Ca²⁺]_c at each pixel was calculated by comparing the 340/380 fluorescence ratio (corrected for background fluorescence) with a calibration curve based on the equation of Grynkiewicz (21). The mean islet [Ca²⁺]_c is the average of all its pixel-[Ca²⁺]_c values.

The autofluorescence of NADH and NADPH [NAD(P)H] was measured by microspectrofluorimetry (excitation/emission 360/470 nm, 400-nm long-pass dichroic mirror) (20). Because NAD(P)H fluorescence increases with cell number, the results obtained in each islet were normalized to the mean fluorescence in the presence of 10 mmol/l glucose before averaging.

Changes in mitochondrial membrane potential were measured by microspectrofluorimetry using rhodamine-123 (excitation/emission 490/530 nm, 505-nm long-pass dichroic mirror) (22). After 20 min loading with 10 μg/ml rhodamine-123 (Molecular Probes, Eugene, OR) in the presence of the same glucose concentration as that during culture, the islets were perifused with a medium containing increasing glucose concentrations. Background-subtracted fluorescence in each islet was normalized to the mean fluorescence in the presence of 10 mmol/l glucose before results from different islets were averaged.

After overnight culture, batches of five islets were preincubated for 30 min in 500 μl medium containing 0.5 mmol/l glucose and then incubated for 1 h in 1 ml medium containing 0.5 or 10 mmol/l glucose. After removal of 625 μl medium, the islets were disrupted with trichloracetic acid (5% vol/vol final), vortexed, placed on ice, and centrifuged for 5 min. The supernatant (450 μl) was submitted to a triple diethylether extraction, diluted with 450 μl of a buffer containing 20 mmol/l HEPES, 3 mmol/l MgCl₂, and KOH to adjust pH at 7.75, and stored at −80°C. ATP was measured by a luminometric method, as previously described (18). Blanks and ATP standard curves in Krebs buffer were run through the entire procedure. The sum (ATP + ADP) was measured after conversion of ADP into ATP by pyruvate kinase in the presence of phosphoenolpyruvate, and the ATP/(ATP + ADP) ratio was calculated.

Data are shown as means ± SE for the indicated number of islets or batches of islets, except where otherwise specified. Groups were compared using one-way ANOVA followed by Newman-Keuls test. Values are described as higher, lower, or different only when P was at least <0.05. The half-maximal effective concentrations of glucose (EC₅₀) were calculated using Prism 3.02 (Graphpad Software, San Diego, CA).

DP P. obesus, which were normoglycemic on LE diet, became markedly hyperglycemic after 5 days on HE diet (4.4 ± 0.2 and 19.4 ± 1.2 mmol/l glucose, respectively, n = 12–14) (16). In contrast, DR P. obesus remained normoglycemic on either diet (4.1 ± 0.1 and 4.7 ± 0.1 mmol/l glucose on LE and HE diet, respectively, n = 9–10). The islets from normoglycemic P. obesus were cultured overnight in medium containing 5 mmol/l glucose, close to their blood glucose level (DP-LE5, DR-LE5, and DR-HE5 islets). In contrast, to prevent reversal of changes induced by hyperglycemia, the islets from diabetic P. obesus were cultured in the presence of 10 mmol/l glucose (DP-HE10 islets), the maximally effective concentration on insulin secretion from P. obesus islets (16).

During perifusion with 2 mmol/l glucose, insulin secretion from DP-LE5 islets was low and stable (Fig. 1A). Stimulation with 6 mmol/l glucose triggered a biphasic increase in insulin secretion, and subsequent stimulation with 10 mmol/l glucose doubled the secretory rate (Fig. 1A).

In comparison, DP-HE10 islets secreted ∼20 times more insulin during perifusion with 2 mmol/l glucose (Fig. 1B). Stimulation with 6 mmol/l glucose triggered a biphasic increase in secretion, with a plateau only slightly higher than in 2 mmol/l glucose (28.7 ± 0.8 vs. 22.3 ± 1.4 pg · islet⁻¹ · min⁻¹). Subsequent stimulation with 10 mmol/l glucose did not further stimulate insulin secretion (Fig. 1B). Thus, during perifusion with 10 mmol/l glucose, the rate of insulin secretion was similar in DP-LE5 and DP-HE10 islets (27.7 ± 2.1 vs. 24.3 ± 1.3 pg · islet⁻¹ · min⁻¹), despite an ∼80% decrease in the insulin content of DP-HE10 islets. Therefore, relative to the islet insulin content, the rate of insulin secretion during perifusion with 2, 6, and 10 mmol/l glucose was markedly higher in DP-HE10 than DP-LE5 islets (Figs. 1D and E). That difference likely resulted from an increased glucose sensitivity of DP-HE islets. Indeed, a rise in glucose from 0.5 to 2 mmol/l strongly stimulated insulin secretion in DP-HE10 islets but was ineffective in DP-LE5 islets (Figs. 1C and F). A basal glucose concentration of 0.5 mmol/l was therefore used for the rest of the study.

To ascertain that responses of DP-LE5 and DP-HE10 islets were not different because of distinct culture conditions, DP-HE islets were also cultured in the presence of 5 instead of 10 mmol/l glucose (DP-HE5) (Figs. 1B and E). During perifusion with 2 mmol/l glucose, the rate of insulin secretion in DP-HE5 islets (7.8 ± 2.5 pg · islet⁻¹ · min⁻¹) was lower than in DP-HE10 islets but higher than in DP-LE5 islets. However, the rate of insulin secretion during perifusion with 6 and 10 mmol/l glucose was not different between DP-HE5 and DP-HE10 islets, nor was the insulin content.

As a further test of the impact of culture conditions, DP-LE islets were cultured in the presence of 10 instead of 5 mmol/l glucose (DP-LE10) (Figs. 1A and D). During perifusion with 2 and 6 mmol/l glucose, insulin secretion of DP-LE10 islets was similar to that of DP-LE5 islets despite a 75% reduction of islet insulin content. However, the maximal rate of secretion during perifusion with 10 mmol/l glucose was decreased by ∼50% in DP-LE10 islets. Relative to the insulin content, insulin secretion was slightly increased in DP-LE10 islets in the presence of both 6 and 10 mmol/l glucose, but not to the same extent as in DP-HE islets.

The insulin content and insulin secretory response of DR-LE islets were similar to those of DP-LE islets cultured at the same glucose concentration and were not affected by 5 days of HE diet (not shown).

During perifusion with 0.5 mmol/l glucose, DP-LE5 islets exhibited low basal [Ca²⁺]_c (64 ± 2 nmol/l) (Figs. 2A and C). Subsequent stepwise stimulation with glucose-induced [Ca²⁺]_c changes similar to those in normal rodent islets, except for the higher glucose sensitivity of P. obesus islets (23). Thus, in 8 of 29 islets, a slight decrease of [Ca²⁺]_c was observed upon stimulation with as little as 2 mmol/l glucose and, on average, stimulation with 4 mmol/l glucose induced a transient drop followed by a moderate increase in [Ca²⁺]_c. A clear biphasic increase in [Ca²⁺]_c was observed at 6 mmol/l glucose, and a plateau with eventual oscillations was reached at 10 mmol/l glucose. A similar response was observed in DR-LE5 islets, and this response was not altered by an HE diet (DR-HE5 islets, data not shown).

In contrast, DP-HE10 islets had a higher basal [Ca²⁺]_c during perifusion with 0.5 mmol/l glucose (83 ± 4 nmol/l) and were more sensitive to glucose. Strikingly, [Ca²⁺]_c was already increased upon stimulation with 2 mmol/l glucose, showing clear oscillations in 7 of 24 DP-HE10 islets (Figs. 2B and D), and the maximal [Ca²⁺]_c rise occurred at ∼4–6 mmol/l glucose. Upon stimulation with a glucose concentration close to the EC₅₀ (5.6 and 3.3 mmol/l for DP-LE5 and DP-HE10 islets, respectively), islets of either type displayed a biphasic increase in [Ca²⁺]_c, with a plateau followed by regular oscillations with similar characteristics (not shown).

As for insulin secretion studies, DP-HE islets were also cultured in the presence of 5 instead of 10 mmol/l glucose (HE5) (Fig. 2D). The observed changes in [Ca²⁺]_c were similar in DP-HE5 and DP-HE10 islets, except for a lack of [Ca²⁺]_c increase in DP-HE5 islets perifused with 2 mmol/l glucose, indicative of a slightly lower glucose sensitivity. This clearly corresponds with the insulin secretory responses of the two groups of islets described above. Conversely, culturing DP-LE islets in 10 instead of 5 mmol/l glucose (LE10) (Fig. 2C) only slightly increased the sensitivity of their [Ca²⁺]_c response to glucose without reproducing the changes observed in DP-HE islets. Thus, together with the insulin secretion data, these results indicate that overnight culture in 10 mmol/l glucose does not account for the alteration of the insulin and [Ca²⁺]_c responses of DP-HE10 islets. This conclusion is further supported by the observation that freshly isolated DP-HE islets were more sensitive to glucose than DP-LE islets in terms of changes in [Ca²⁺]_c, irrespective of the glucose concentration during fura-PE3 loading (Fig. 3). A similar difference in sensitivity has previously been reported for glucose stimulation of insulin secretion in fresh DP-HE and DP-LE islets (16).

We next measured adenine nucleotides to determine whether glucose metabolism contributes to the increased glucose sensitivity of insulin secretion and [Ca²⁺]_c responses in DP-HE versus DP-LE islets. In DP-LE5 and DP-LE10 islets, the ATP/(ATP + ADP) ratio, an indicator of islet cell energetic level, increased from ∼0.82 to ∼0.94 after incubation in 0.5 and 10 mmol/l glucose (P < 0.001; corresponding ATP/ADP ratios of 4.6 and 15.7, respectively) (Fig. 4A). Compared with these islets, the ATP/(ATP + ADP) ratio in DP-HE10 islets incubated in 0.5 mmol/l glucose was significantly higher (Fig. 4B), and this difference was only partly attenuated by overnight culture in 5 mmol/l glucose (HE5) (Fig. 4B). In contrast, the ATP/(ATP + ADP) ratio in islets incubated for 1 h in 10 mmol/l glucose was virtually similar in all groups. Acute stimulation with glucose thus masked the difference in energetic state that exists between DP-HE and DP-LE islets at low glucose.

The ATP/(ATP + ADP) ratio was similar in DR-LE5, DR-LE10, DR-HE5, and DR-HE10 islets, increasing from ∼0.84 to ∼0.97 (P < 0.001) after incubation in 0.5 and 10 mmol/l glucose, respectively. This corresponds to ATP/ADP ratios of ∼5.3 and 32, respectively.

The increase in glucose metabolism was also estimated by measuring the increase in NAD(P)H autofluorescence and the decrease in rhodamine-123 fluorescence, the latter reflecting mitochondrial membrane hyperpolarization (22). Because the ATP/(ATP + ADP) ratio differed in islets incubated in 0.5 mmol/l glucose but was almost identical in islets incubated in 10 mmol/l glucose (Fig. 4), the NAD(P)H and rhodamine-123 fluorescence results were expressed relative to the fluorescence attained during perifusion with 10 mmol/l glucose. In DP-LE5 islets, stepwise stimulation with glucose was followed by stepwise increase in NAD(P)H fluorescence and stepwise mitochondrial membrane hyperpolarization (Fig. 5). In comparison, DP-HE10 islets were more sensitive to glucose, and this difference persisted almost unchanged when DP-HE islets were cultured overnight in 5 instead of 10 mmol/l glucose (not shown).

Glucose-induced changes in NAD(P)H fluorescence and mitochondrial membrane potential of DR-HE5 and DR-LE5 islets were similar to those of DP-LE5 islets (Fig. 6).

In each experiment, NAD(P)H fluorescence, rhodamine-123 fluorescence, and [Ca²⁺]_c were averaged over the last 5 min of each step of glucose concentration, and the mean concentration-response curves were fitted to a sigmoidal relationship (Fig. 6). These curves were clearly shifted to the left in DP-HE10 versus DP-LE5 islets (Fig. 6A, C, and E), with lower (P < 0.001) EC₅₀s for changes in NAD(P)H (3.3 ± 0.2 vs. 5.1 ± 0.1 mmol/l, respectively) and [Ca²⁺]_c (2.7 ± 0.1 vs. 5.6 ± 0.1 mmol/l, respectively). When average [Ca²⁺]_c was calculated over the entire 10-min period at each glucose concentration, a similar shift to the left was observed.

The glucose concentration-response curves of DR-LE5 islets were similar to those of DP-LE5 islets and were not altered by 5 days of HE diet (Fig. 6).

This study shows that hyperglycemia induced by 5 days of HE diet increases the glucose sensitivity of islets from DP P. obesus at the level of glucose metabolism while decreasing their insulin content. The shift in glucose sensitivity was accompanied by a rise in [Ca²⁺]_c and stimulation of insulin secretion at otherwise subthreshold glucose concentrations (2 mmol/l), with maximal β-cell stimulation at a relatively low physiological glucose concentration (4–6 mmol/l). Thus, the glucose concentration-response curves for changes in mitochondrial membrane potential, NAD(P)H fluorescence, and [Ca²⁺]_c were shifted to lower glucose concentrations by ∼2–3 mmol/l in DP-HE10 versus DP-LE5 islets. Already in the presence of 0.5 mmol/l glucose, glucose metabolism was increased in DP-HE10 versus DP-LE5 islets, as shown by a higher ATP/(ATP + ADP) ratio, stronger NAD(P)H fluorescence, and more hyperpolarized mitochondrial membrane potential. In contrast, the ATP/(ATP + ADP) ratio was similar in DP-HE10 and DP-LE5 islets incubated in 10 mmol/l glucose.

The evaluation of the energetic state of islets with markedly different insulin contents may be complicated by the presence of a stable pool of adenine nucleotides within insulin granules. Thus, degranulation of mouse β-cells causes an increase in the apparent ATP/ADP ratio measured on total islet extract (18). This problem was not observed in P. obesus, as shown by the lack of difference in ATP/(ATP + ADP) ratio between DP-LE5 and DP-LE10 islets in spite of a marked reduction in insulin content, thereby confirming that changes in the ATP/(ATP + ADP) ratio accurately reflect changes in islet energetic state. Thus, the energetic state of DP-HE10 islets was increased at low glucose, whereas their capacity to synthesize ATP upon glucose stimulation was preserved. It was thus justified to normalize NAD(P)H and rhodamine-123 fluorescence to the level attained in the presence of 10 rather than 0.5 mmol/l glucose.

An increase in islet glucose metabolism at low glucose has variably been attributed to glycogen or triglyceride accumulation or to increased glucokinase or hexokinase 1 activity in β-cells (24,25). The maximal glucose phosphorylation rate is higher in DP-HE than DP-LE islets, largely due to a rise in the maximal velocities of hexokinase 1 more than glucokinase, with no change in the K_m of these enzymes (16). A similar shift to the left of the glucose concentration-response curve for insulin secretion characterizes islets from hyperglycemic 90% pancreatectomized rats (26,27).

The present results do not conflict with the reduction in fractional glucose oxidation (glucose oxidation/utilization) previously observed in DP-HE islets (17) because this reduction reflected a larger increase in glucose utilization than oxidation, the latter not being smaller than in DP-LE islets (17). We found no evidence that ATP synthesis was impaired in DP-HE islets. The observed imbalance between glycolytic and oxidative metabolism (17) may reflect adaptation to increased demand (overworked β-cells) or consequences of sustained in vivo hyperglycemia. It could be due to excessive expression of lactate dehydrogenase, resulting in a derivation of pyruvate from mitochondrial oxidation to lactate formation. Increased expression of lactate dehydrogenase has been observed in hyperglycemic 90% pancreatectomized rats (28) and in MIN6 cells cultured in the presence of 30 mmol/l glucose (29). Alternatively, more glucose-derived metabolites might be used in anaplerosis rather than oxidation (4), as suggested by an increased anaplerotic flux in INS1 cells cultured for 3 days in the presence of 25 mmol/l glucose (30).

Several observations indicate that the higher glucose sensitivity of DP-HE islets was not due to overnight culture in 5 or 10 mmol/l glucose, but instead related to the in vivo hyperglycemia in DP-HE P. obesus. First, the higher glucose sensitivity of [Ca²⁺]_c responses observed in DP-HE10 and DP-LE5 islets was confirmed in fresh DP-HE and DP-LE islets and corresponded well to the previously reported increase in glucose sensitivity of insulin secretion in fresh DP-HE and DP-LE islets (16). Second, the glucose sensitivity of DP-HE5 islets was only slightly lower than that of DP-HE10 islets, whereas that of DP-LE10 islets was only slightly higher than that of DP-LE5 islets, while being much lower than that of DP-HE5 islets. Thus, it is clear that overnight culture in 10 mmol/l glucose does not fully account for the alterations of [Ca²⁺]_c and insulin secretory responses to glucose of DP-HE10 islets. Third, the glucose sensitivity of islets from DP P. obesus was higher after 3 weeks rather than 5 days on an HE diet (data not shown). This indicates that the impact of different durations of in vivo hyperglycemia can be detected even when islets are tested after overnight culture. Finally, as already stated, the glucose EC₅₀s for metabolic and [Ca²⁺]_c changes in this study are similar to the EC₅₀ for glucose stimulation of insulin secretion in fresh P. obesus islets (16).

The increase in glucose sensitivity and decrease in insulin content of DP-HE10 islets can be attributed to the diabetic state of the animal rather than to the HE diet itself, as indicated by the lack of effect of HE diet on blood glucose and islet function in DR P. obesus. The two β-cell alterations are not tightly linked. Thus, when DP-LE islets were cultured overnight in 10 mmol/l glucose, only a minor increase in glucose sensitivity occurred, whereas the decrease of insulin content approached 80%, close to that of DP-HE islets. The same changes were observed in DR-LE islets cultured in 10 instead of 5 mmol/l glucose. Such a rapid degranulation upon exposure to in vitro hyperglycemia may result from insufficient glucose-dependent insulin biosynthetic activity of P. obesus islets due to the absence of the conserved form of PDX-1 (31), a transcription factor involved in the regulation of insulin gene expression by glucose (32,33).

In a previous study, fresh DP-HE islets were 90% degranulated compared with DP-LE islets and their maximal rate of insulin secretion was reduced by ∼50% (16). That β-cell degranulation in DP-HE islets reduces their maximal rate of insulin secretion is not evident from the present study. In absolute terms, the maximal rate of secretion of DP-HE islets was similar to that of DP-LE5 islets, despite an ∼80% degranulation of DP-HE islets. It was, however, much higher than that of DP-LE10 islets that were similarly degranulated. Therefore, relative to the islet insulin content, the maximal rate of secretion was markedly higher in DP-HE than in DP-LE islets. However, comparing insulin secretion of DP-LE10 and DP-LE5 islets, it is also clear that an ∼80% degranulation with only minor changes in glucose sensitivity does not affect insulin secretion at basal or intermediate glucose concentrations but markedly reduces it at high glucose (in absolute terms). The observation that the maximal rate of secretion is higher in DP-HE10 than in DP-LE10 islets, despite similar [Ca²⁺]_c, suggests that the amplifying pathway of glucose stimulation (8) is increased, or maximally stimulated, in islets from diabetic P. obesus.

Previous studies on the alterations of stimulus-secretion coupling in β-cells chronically exposed to hyperglycemia have yielded apparently conflicting results. A diminished insulin response to acute glucose stimulation (glucose desensitization) has been reported to characterize islets subjected to chronic hyperglycemia in vitro (34–36) and in vivo (16,37–41). In contrast, an increased glucose sensitivity characterized islets from 90% pancreatectomized rats (42), dexamethasone-treated (43) and 48-h glucose-infused rats (44), and normal rat islets cultured for several days in the presence of high glucose concentrations (45–47). However, the occurrence of maximal stimulation of insulin secretion at low physiological glucose concentration, as observed in DP-HE islets and in other models of chronic hyperglycemia (48), could be easily misinterpreted as a lack of glucose-induced insulin secretion if only tested between 4 and 10–20 mmol/l glucose (42,46). These results call for careful interpretation of a lack of GSIS in in vivo studies, in which plasma glucose cannot be safely decreased below 4 mmol/l.

The glucose stimulus-secretion coupling was, within the limits of our study, similar in DR-LE and DP-LE islets. Why then are DR P. obesus more resistant than DP P. obesus to the diabetogenic effect of HE diet? A higher β-cell mass is unlikely the explanation, because the pancreatic insulin content is similar in DR-HE and DP-LE animals (N.K., unpublished observation). It is more plausible that the lower propensity of DR P. obesus to develop obesity as well as their slightly higher insulin sensitivity account for their lower susceptibility to developing diabetes when fed an HE diet (13). For all parameters measured in the present study, the glucose sensitivity of DR-LE islets was even slightly lower than that of DP-LE islets, which suggests a link between insulin resistance and β-cell glucose sensitivity. A similar conclusion has been reached for rodent models of obesity (24). It is thus possible that the sustained β-cell hyperactivity required to maintain normoglycemia in the face of insulin resistance, rather than hyperglycemia itself, leads to the observed changes in glucose sensitivity. The advantage of such a process is obvious as long as normoglycemia is preserved (42,49), but this physiological adaptation to increased insulin demands seems completely inoperative in the presence of persistent hyperglycemia.

In conclusion, islets from diabetic P. obesus display an increased sensitivity to glucose at the level of glucose metabolism and a profound β-cell degranulation, both of which may affect their in vivo GSIS. Whether such a combination is present in glucose-intolerant or type 2 diabetic patients should be investigated.

This work was supported by Grants 1.5.133.00 and 3.4552.98 from the Fonds National de la Recherche Scientifique (Brussels), by the Interuniversity Poles of Attraction Program (P5/3-20)-Belgian State, by Grant 00/05-260 from the General Direction of Scientific Research of the French Community of Belgium, by Fonds S. and J. Pirart from the “Association Belge du Diabète,” and by grants from the Israel Science Foundation, the Juvenile Diabetes Foundation International, the Hadassah Research Foundation, and the European Community. J.-C.J. is a Research Associate from the National Fund for Scientific Research, Brussels, Belgium.

The authors thank A. Fiévet, F. Knockaert, and M. Nenquin for expert technical help.