Introduction

Clinical and epidemiological studies have shown that depression is an underestimated disorder in Type 1 and Type 2 diabetic patients [1, 2]. A higher prevalence has been reported in patients with diabetes mellitus than in non-diabetic individuals [1, 2]. Moreover, depression may impair control of glycaemia and treatment compliance, as well as increasing the risk of vascular complications [1, 3].

Different classes of antidepressants commonly prescribed to depressed patients are also used in patients with depression and diabetes mellitus [2, 4, 5, 6]. Successful drugs include tricyclic antidepressants (imipramine and imipramine-like compounds, which block the neuronal uptake of serotonin and norepinephrine) and the selective serotonin re-uptake inhibitors (typified by fluoxetine and sertraline). Other compounds such as tetracyclic antidepressants, norepinephrine re-uptake inhibitors and monoamine oxidase inhibitors are also effective in treating the various forms of depressive disorders [2, 4, 5, 6].

Treatment with antidepressants has been reported to affect glucose homeostasis in diabetic and non-diabetic individuals [2, 3, 4, 7, 8]. Tricyclic antidepressants but also selective serotonin re-uptake inhibitors have been shown to increase or decrease glycaemia with or without concomitant changes in plasma insulin [2, 3, 4, 7, 9, 10]. Thus, although it is well recognised that antidepressants may affect blood glucose levels, the physiological mechanism(s) underlying these modifications in glucose homeostasis are not completely clear.

The aim of this study, therefore, was to examine, at the insulin-secreting cell level, the ionic and secretory effects of imipramine, a tricyclic antidepressant that has been used clinically for over three decades.

Materials and methods

Measurement of insulin secretion from incubated pancreatic islets

Experiments were performed with pancreatic islets isolated by the collagenase method from fed female Wistar albino rats (180–220 g, Charles River Laboratories, Brussels, Belgium) [11, 12]. Laboratory animal care was approved by the Ethics Committee of the Université Libre de Bruxelles (Free University of Brussels).

Groups of ten islets, each derived from the same batch of islets, were pre-incubated for 30 min at 37 °C in 1 ml of a physiological salt medium (in mmol/l: NaCl 115, KCl 5, CaCl2 2.56, MgCl2 1, NaHCO3 24) supplemented with 2.8 mmol/l glucose, 0.5% (w/v) dialysed albumin (Fraction V; Sigma-Aldrich, Bornem, Belgium) and equilibrated against a mixture of O2 (95%) and CO2 (5%). The islets were then incubated for 90 min at 37 °C in 1 ml of the same medium containing 16.7 mmol/l glucose and, in addition, increasing concentrations of imipramine. Experiments were repeated on different islets populations. Insulin release was expressed as a percentage of the value recorded in control experiments (100%), i.e. in the absence of drug and presence of 16.7 mmol/l glucose. The release of insulin was measured radioimmunologically using rat insulin as a standard [13].

Measurements of 45Ca outflow and insulin release from perifused pancreatic islets

The medium used for incubating, washing and perifusing the islets consisted of a bicarbonate-buffered solution having the following composition (in mmol/l): NaCl 115, KCl 5, CaCl2 2.56, MgCl2 1, NaHCO3 24. The medium was supplemented with 0.5% (w/v) dialysed albumin (Fraction V) and equilibrated against a mixture of O2 (95%) and CO2 (5%).

The methods used to measure 45Ca outflow and insulin release from perifused islets have been described previously [14, 15]. Briefly, groups of 100 islets were incubated for 60 min in a medium containing 16.7 mmol/l glucose and 45Ca ion (0.02–0.04 mmol/l; 3.70×106 Bq/ml). After incubation, the islets were washed four times with a non-radioactive medium and then placed in a perifusion chamber. The perifusate was delivered at a constant rate (1.0 ml/min). From the 31st to the 90th minute of perifusion, the effluent was continuously collected over successive periods of 1 min each. An aliquot of the effluent (0.6 ml) was used for scintillation counting while the remainder was stored at −20 °C for insulin radioimmunoassay [13]. At the end of perifusion, the radioactive content of the islets was also determined. The outflow of 45Ca (cpm) was expressed as a fractional outflow rate (FOR; percent of instantaneous islet content per minute).

Electrophysiological measurements

Electrophysiological studies were carried out on isolated rat pancreatic islet cells. Pancreatic islets were disrupted in a Ca2+-deprived medium and then centrifuged through an albumin solution to remove debris and dead cells [11, 15]. Insulin-secreting cells were selected on the basis of their larger size [16].

Whole-cell Ca2+ currents were recorded using the perforated patch whole-cell configuration [17] of the patch-clamp technique [18]. The standard extracellular solution was composed of (in mmol/l): NaCl 118, tetraethylammonium chloride 20, KCl 5.6, CaCl2 2.6, MgCl2 1.2, glucose 5 and HEPES 5 (pH adjusted to 7.40 with NaOH). The effect of imipramine was investigated on single beta cells by switching from the standard extracellular solution to the same solution supplemented with 100 µmol/l imipramine. In some experiments, 10 µmol/l tetrodotoxin was added to the extracellular solution to block a potential remaining Na+ current. The pipette was filled with (in mmol/l): Cs2SO4 76, KCl 10, NaCl 10, MgCl2 1 and HEPES 5 (pH adjusted to 7.35 with CsOH). Amphotericin B (0.24 µg/ml) was included in the pipette solution in order to establish electrical contact with the cell interior. Ionic currents were recorded using an EPC-8 amplifier (HEKA, Lambrecht/Pfalz, Germany) and stored on a computer. The pulse software (HEKA) and an ITC-16 AD/DA converter (Instrutech, New York, USA) were used to acquire the data. The current signals were digitised at 2 kHz. Prior to digitising, the signals were filtered at 500 Hz using an 8-pole low-pass Bessel filter (Frequency Devices, Haverhill, Mass., USA). All experiments were carried out at room temperature. The voltage clamp protocol consisted of 450-ms depolarisation pulses starting from a holding potential of −70 mV and ranging from −40 mV to +30 mV.

Measurements of fura-2 fluorescence from single rat pancreatic islet cells

The cells were placed on glass coverslips and maintained in tissue culture for 72 h before use. Islet cells were cultured in RPMI 1640 culture medium (Invitrogen, Merelbeke, Belgium) supplemented with 10% (v/v) newborn calf serum and containing glutamine (2.3 mmol/l), penicillin G (100 U/ml) and streptomycin (100 µg/ml). The cells were then incubated with fura-2 AM (final concentration: 2 µmol/l) for 1 h and, after washing, the coverslips with the cells were mounted as the bottom of an open chamber (1 ml) placed on the stage of the microscope. The medium used to perifuse the cells contained (in mmol/l): NaCl 115, KCl 5, CaCl2 2.56, MgCl2 1, NaHCO3 24, glucose 2.8. It was gassed with O2 (95%) and CO2 (5%). Fura-2 fluorescence of single loaded cells (selected on the basis of their larger size) was measured by dual-excitation microfluorimetry with a Spex photometric system (Optilas, Alphen aan den Rijn, The Netherlands). The excitation wavelengths (340 nm and 380 nm) were alternated at a frequency of 1 Hz, the length of time for data collection at each wavelength being 0.05 seconds. The emission wavelength was set at 510 nm. We calculated [Ca2+]i as described previously [11, 15]. Individual experiments were repeated at least four times, on different cell populations.

Drugs

In some experiments, extracellular Ca2+ was eliminated by omitting CaCl2 from the physiological medium and adding 0.5 mmol/l EGTA (Sigma-Aldrich). Depending on the experiment, the media were enriched with glucose (Merck, Darmstadt, Germany), imipramine (Sigma-Aldrich) or tetrodotoxin (Acros Organics, Kortrijk, Belgium). When high concentrations of extracellular K+ were used, the concentration of NaCl was lowered to keep osmolarity constant.

Calculations

Results are expressed as means ± SEM. The increase in 45Ca outflow was estimated in each individual experiment from the integrated outflow of 45Ca observed during stimulation (45th–68th min) after correction for basal value (40th–44th min). Peak 45Ca outflow was estimated from the difference in 45Ca outflow between the highest value recorded during stimulation and the mean basal value (40th–44th min) within the same experiment. The inhibitory effect of imipramine on 45Ca outflow and insulin release from islets perifused in the presence of 16.7 mmol/l glucose was taken as the difference between the mean value for 45Ca outflow or insulin output recorded in each individual experiment between the 40th and 44th and the 60th and 68th min of perifusion. The statistical significance of differences between mean data was assessed by Student’s t test or by analysis of variance followed by a Scheffe test procedure. A p value of less than 0.05 was considered significant.

Results

Effects of imipramine and fluoxetine on insulin release from incubated rat pancreatic islets

In rat pancreatic islets exposed to 5.6 mmol/l glucose, the addition of 100 µmol/l of imipramine did not affect insulin output (data not shown).

By contrast, imipramine inhibited insulin release from pancreatic islets incubated in the presence of an intermediate insulinotropic glucose concentration. Thus, in the simultaneous presence of 8.3 mmol/l glucose and 100 µmol/l imipramine in the incubation medium, the insulin output averaged 55.6±5.4% of the control experiments (p<0.05).

Moreover, the addition of increasing concentrations of imipramine to pancreatic islets incubated in the presence of 16.7 mmol/l glucose provoked a concentration-dependent decrease in insulin release (Fig. 1). After the addition of 1 µmol/l, 10 µmol/l and 100 µmol/l imipramine, the residual insulin release averaged 90.6±5.8% (p>0.05), 38.3±2.6% (p<0.05) and 9.9±0.7% (p<0.05) of the control value. The IC50 value (drug concentration evoking a 50% reduction of the secretory response to 16.7 mmol/l glucose) amounted to 5.2 µmol/l.

Fig. 1
figure 1

Effect of increasing concentrations of imipramine (Im) on insulin release from pancreatic islets incubated in the presence of 16.7 mmol/l glucose (Glu). Insulin release (means ± SEM) is expressed as a percent of the value for control experiments (100%, no added drug, presence of 16.7 mmol/l glucose). The upright lines at the top of the bars correspond to the SEM. Number of samples: 0 µmol/l Im: 31; 1 µmol/l Im: 23; 10 µmol/l Im: 23; 100 µmol/l Im: 25

Full size image

Effects of imipramine on 45Ca outflow and insulin release from perifused rat pancreatic islets

In the presence of 5.6 mmol/l glucose throughout, and regardless of whether the medium contained or was deprived of extracellular Ca2+, the addition of 10 µmol/l imipramine did not affect the rate of 45Ca outflow from prelabelled and perifused rat pancreatic islets (data not shown).

By contrast, when the physiological medium contained 16.7 mmol/l instead of 5.6 mmol/l glucose, exposure to 10 or 100 µmol/l imipramine provoked an immediate and reversible inhibition of 45Ca outflow from islets perifused in the presence of extracellular Ca2+ (Fig. 2a, b). The imipramine-induced reduction in the 45Ca FOR was more rapid and more pronounced at the higher concentration. The paired difference in the 45Ca FOR before (40–44 min) and during (60–68 min) exposure to 10 µmol/l and 100 µmol/l imipramine averaged 0.72±0.01%/min and 1.02±0.07%/min respectively (p<0.05).

Fig. 2
figure 2

Effect of 10 µmol/l (a, c) and 100 µmol/l (b, d) imipramine on 45Ca outflow (a, b) and insulin release (c, d) from pancreatic islets perifused throughout in the presence of 16.7 mmol/l glucose. Basal media contained extracellular Ca2+ (●) or were deprived of Ca2+ and enriched with EGTA (○). Mean values (±SEM) refer to four or five individual experiments. FOR, fractional outflow rate

Full size image

On removal of the drug from the perifusing medium, 45Ca outflow increased. This increase was delayed after removal of the higher drug concentration (compare Fig. 2a and b).

The decreases in the 45Ca FOR induced by the addition of 10 µmol/l and 100 µmol/l imipramine were accompanied by concentration-dependent reductions in insulin output (Fig. 2c, d). In the presence of 10 µmol/l and 100 µmol/l imipramine (60–68 min), the release of insulin represented 31.10±1.08% and 18.62±1.75% of the secretory rate recorded before administration of the drug (40–44 min) (p<0.05). The withdrawal of 10 µmol/l imipramine was followed by a rapid increase whilst the withdrawal of 100 µmol/l imipramine was followed by a weak increase in insulin output (Fig. 2c, d).

To further investigate the effects of imipramine on radioisotopic fluxes, the experiments were repeated in the absence of extracellular Ca2+ (Fig. 2a, b). In islets exposed throughout to 16.7 mmol/l glucose and Ca2+-depleted media, the rate of 45Ca outflow before drug administration (40–44 min) was lower (p<0.05). Under the latter conditions, 10 µmol/l imipramine did not affect 45Ca outflow (Fig. 2a). The addition of 100 µmol/l imipramine, however, provoked a modest but sustained and reversible rise in the 45Ca FOR (Fig. 2b).

Effects of imipramine on KCl-induced changes in 45Ca outflow from perifused rat pancreatic islets

In the presence of 2.8 mmol/l glucose and extracellular Ca2+, a sudden rise in the extracellular concentration of K+ provoked an immediate and pronounced increase in 45Ca outflow from prelabelled and perifused rat pancreatic islets (Fig. 3a, b).

Fig. 3
figure 3

Effect of a rise in the extracellular K+ concentration from 5 to 50 mmol/l on 45Ca outflow from pancreatic islets perifused throughout in the absence (○) (a, b) and the presence (●) of 10 µmol/l (a) or 100 µmol/l (b) imipramine. Basal media contained 2.8 mmol/l glucose and extracellular Ca2+. Mean values (± SEM) refer to four to six individual experiments. FOR, fractional outflow rate

Full size image

When the same experiment was repeated with either 10 µmol/l (Fig. 3a) or 100 µmol/l (Fig. 3b) imipramine in the medium, the basal rate of 45Ca outflow was unaffected (p>0.05). However, the presence of imipramine in the physiological medium markedly reduced the cationic response to K+. The increase in 45Ca outflow evoked by 50 mmol/l K+ averaged 1.18±0.14%/min in the absence and 0.18±0.06%/min in the presence of 10 µmol/l imipramine (p<0.05). The peak 45Ca outflow observed during exposure to K+ averaged 1.66±0.20%/min and 0.92±0.08%/min in the absence and presence of 10 µmol/l imipramine respectively (p<0.05).

In islets exposed throughout to 100 µmol/l imipramine, the stimulatory effect of high extracellular K+ was completely abolished (Fig. 3b).

Effect of imipramine on the inward Ca2+ current in single rat pancreatic islet cells

This effect was investigated by means of the perforated patch configuration of the patch-clamp technique. We investigated the dependence of the Ca2+ current on the voltage using 450 ms depolarisation pulses. Each depolarising pulse was separated from the following by 20 seconds to allow return to the basal calcium level [19]. Depolarisations from a holding potential of −70 mV to voltages between −40 mV and +30 mV evoked an inward current that reached a maximum amplitude around 0 mV, with gating properties consistent with the slowly deactivating Ca2+ current (L-type Ca2+ channels) described in rat beta cells [20]. This inward current was insensitive to the application of 10 µmol/l tetrodotoxin (data not shown), thereby confirming that it was not due to the activation of Na+ channels.

Figure 4a illustrates the effect of 100 µmol/l imipramine on the inward current evoked by depolarisations between −70 and 0 mV in a single islet cell. The decrease of the peak current was marked and reversible.

Fig. 4
figure 4

Effect of 100 µmol/l imipramine on the voltage dependence of the inward Ca2+ current. a. Typical recordings of whole-cell Ca2+ currents observed during a 450-ms depolarising pulse from a holding potential of −70 mV to 0 mV on a single islet cell. b. Corresponding I-V relationships obtained with voltage clamp protocol consisting of 450-ms depolarisation pulses starting from a holding potential of −70 mV and ranging from −40 mV to +30 mV in the absence (control condition, ○) and presence (●) of 100 µmol/l imipramine. Mean values (±SEM) refer to five individual experiments

Full size image

The amplitude of the inward currents was used to construct I-V plots. The application of 100 µmol/l imipramine induced (Fig. 4b) a robust decrease of the Ca2+ current without altering its voltage-dependence. At 0 mV, the peak current intensity dropped from −48.4±4.1 pA in control conditions to −10.9±2.4 pA in the presence of 100 µmol/l imipramine (n=5, p<0.05).

Effects of imipramine on the cytosolic free Ca2+ concentration of single rat pancreatic islet cells

A rise in the extracellular glucose concentration from 2.8 to 20.0 mmol/l provoked a biphasic increase in [Ca2+]i, consisting of an initial peak followed by a long-lasting plateau (Fig. 5a). The addition of imipramine (10 µmol/l) during stimulation with 20 mmol/l glucose reduced the cytosolic free Ca2+ concentration (Fig. 5b). The inhibitory effect of the drug was sustained and slowly reversible.

Fig. 5
figure 5

Effect (a) of 20 mmol/l glucose on [Ca2+]i. b. Effect of 10 µmol/l imipramine on glucose-induced (20 mmol/l) increase in [Ca2+]i. Basal media contained 2.8 mmol/l glucose and extracellular Ca2+. Each graph is a representative experiment conducted on a single cell

Full size image

Figure 6 illustrates the effect of depolarising concentrations of K+ on the cytosolic Ca2+ concentration of single islet cells. An increase in the extracellular concentration of K+ from 5 to 50 mmol/l provoked in most cells a monophasic and marked increase in [Ca2+]i (Fig. 6a). In some cells, the [Ca2+]i response to 50 mmol/l K+ was biphasic, an initial peak of short duration being followed by a plateau (Fig. 6b). Whatever the pattern of the K+ response, the subsequent addition of imipramine (10 µmol/l) markedly reduced the [Ca2+]i (Fig. 6b). In the latter condition, the inhibitory effect of imipramine was less readily reversible (Fig. 6b).

Fig. 6
figure 6

Effect (a) of 50 mmol/l KCl on [Ca2+]i. b. Effect of 10 µmol/l imipramine on KCl-induced (50 mmol/l) increase in [Ca2+]i. Basal media contained 5 mmol/l K+, 2.8 mmol/l glucose and extracellular Ca2+. Each graph is a representative experiment conducted on a single cell

Full size image

In another series of experiments, we examined the effects of 100 µmol/l imipramine on the rises in [Ca2+]i induced by glucose (20.0 mmol/l) and K+ (50 mmol/l) (data not shown). In both cases, the addition of 100 µmol/l imipramine again induced marked reductions in [Ca2+]i. At this concentration, however, the drug acted faster. The removal of 100 µmol/l imipramine from the physiological medium was followed by a delayed increase in [Ca2+]i.

Discussion

We showed that the tricyclic antidepressant imipramine provoked a concentration-dependent inhibition of insulin release from rat pancreatic islets incubated in the presence of insulinotropic (8.3 and 16.7 mmol/l) glucose concentrations.

Dynamic experiments using perifused rat pancreatic islets confirmed the capacity of imipramine to reduce the insulin secretory rate. The latter experimental condition further revealed that the secretory response to imipramine was reversible, suggesting that the antidepressant did not damage insulin-secreting cells.

The dose-dependent inhibitory effect of imipramine on the glucose-induced insulin release was reproduced by fluoxetine, a selective serotonin re-uptake inhibitor antidepressant (data not shown). Altogether, these findings confirm those of another study [21] and extend previous observations, indicating that the in vitro reduction of the insulin secretory rate evoked by antidepressants is not restricted to tricyclic antidepressants.

Additional experimental data support the idea that the secretory response to imipramine might be related to modifications in Ca2+ inflow.

Radioisotopic experiments conducted on prelabelled rat pancreatic islets indicated that imipramine elicited a concentration-dependent decrease in 45Ca outflow from islets perifused in the presence of 16.7 mmol/l glucose and extracellular Ca2+. In islets exposed throughout to calcium and insulinotropic glucose concentrations, the 45Ca FOR is known to reflect a sustained stimulation of isotopic exchange between influent 40Ca and effluent 45Ca [22]. Thus, under such experimental conditions, the inhibitory effect of imipramine on 45Ca outflow can be viewed as the result of a reduction of 40Ca entry into the islet cells.

The effects of imipramine on the 45Ca FOR responses to high extracellular K+ concentrations also suggest that imipramine dose-dependently reduces Ca2+ inflow. Indeed, in 45Ca-loaded pancreatic islets, the increment in 45Ca FOR evoked by high K+ again results from the stimulation of a 40Ca/45Ca exchange process [12, 22].

A further argument in support of an inhibitory effect of imipramine on Ca2+ entry can be found in the drug’s lack of inhibitory effect on 45Ca outflow from islets exposed to non-insulinotropic glucose concentrations and/or to Ca2+-depleted media.

The electrophysiological experiments conducted on single islet cells confirmed the ability of imipramine to inhibit a tetrodotoxin-insensitive inward current.

The present findings also provide information about the modality of Ca2+ entry affected by imipramine. Indeed, imipramine failed to affect the rate of 45Ca outflow from islets perifused in the presence of 5.6 mmol/l glucose but markedly inhibited the 45Ca FOR response to 16.7 mmol/l glucose. Under the latter condition, the depolarising effect of glucose reaches the threshold potential for activation of voltage-sensitive Ca2+ channels.

The inhibitory effects of imipramine on the KCl-induced changes in 45Ca outflow further suggest that the drug interfered with a voltage-sensitive modality of Ca2+ entry.

Moreover, patch-clamp experiments revealed that imipramine decreased the amplitude of a voltage-sensitive inward current with gating properties similar to the L-type Ca2+ channels equipping rat beta cells [20].

Incidentally, the effects of imipramine on the cationic responses to glucose and KCl depolarisation are reminiscent of those evoked by Cd2+ or verapamil, two calcium channel blockers known to interact with the L-type Ca2+ channels [12, 22, 23, 24].

Thus, all these observations support the view that imipramine reduced Ca2+ entry by inhibiting the L-type voltage-sensitive Ca2+ channels. Previous studies on cardiac and neuronal cells also revealed that imipramine has an inhibitory effect on an inward Ca2+ current [25, 26].

A decrease in Ca2+ entry as mediated by imipramine should be followed by a reduction in the cytosolic Ca2+ concentration. In agreement with this, calcium fluorimetry experiments conducted on single islet cells confirmed the ability of imipramine to counteract the increase in [Ca2+]i elicited either by insulinotropic concentrations of glucose or by high K+ concentrations.

This effect on [Ca2+]i, a key element in the stimulus–secretion coupling process, probably triggers the decrease in insulin output. Such a view is also substantiated in the observation that the secretory responses to imipramine mirrored the 45Ca FOR data.

In addition to inhibitory effects on plasma membrane Ca2+ channels, 100 µmol/l imipramine also elicited a small, sustained and slowly reversible increase in 45Ca FOR from islets exposed to 16.7 mmol/l glucose and Ca2+-free media. This suggests that a high concentration of imipramine might also provoke intracellular calcium redistribution with subsequent changes in 45Ca outflow [11, 27, 28].

In conclusion, our data indicate that the tricyclic antidepressant imipramine reduces the glucose-sensitive insulin output from rat pancreatic beta cells. This effect results from the inhibition of voltage-sensitive Ca2+ channels with subsequent reduction in Ca2+ entry.

Although serum concentrations of imipramine at typical clinical doses are lower (~0.5–1.0 µmol/l), it should be kept in mind that the plasma concentrations of this tricyclic antidepressant can reach higher levels, e.g. through overdosage, deficient elimination and drug interactions [29]. Moreover, numerous data indicate that imipramine, like other antidepressants, is extensively bound to tissue protein and accumulates in the cells [30, 31, 32]. From a clinical viewpoint, therefore, and taking into account the species differences, it is possible that the side-effect profile of imipramine is related, at least in part, to its ability to act as a Ca2+ entry blocker.