Multivesicular exocytosis in rat pancreatic beta cells

Insulin secretion is associated with ATP release [17]. Figure 1a,b shows examples of ATP-evoked currents (reflecting exocytosis) recorded from rat beta cells infected with P2X₂Rs and infused with 0.2 or 2 μmol/l [Ca²⁺]_i. These currents reflect activation of the P2X₂Rs and are blocked by the purinergic antagonist, suramin [25]. At low [Ca²⁺]_i, the events are infrequent and small (<300 pA). Elevation of [Ca²⁺]_i to 2 μmol/l increases the frequency, amplitude and duration of these events. Figure 1c,d shows the histograms for the integrated currents (Q) of the events recorded at 0.2 and 2 μmol/l [Ca²⁺]_i; non-Gaussian (skewed) distributions were obtained under both experimental conditions (see electronic supplementary material [ESM] text, section 1). Whereas most events recorded at low [Ca]_i were smaller than 5 pC, ∼55% of the events recorded at 2 μmol/l [Ca²⁺]_i were larger than 5 pC. The corresponding cumulative histograms are shown in Fig. 1e. The distribution of the Q of exocytotic events was shifted to the right when [Ca²⁺]_i was elevated from 0.2 to 2 μmol/l (p < 0.001).

Figure 1f shows the relationship between charge and the t _10-90% and HW. The t _10-90% values were <10 ms over a wide range of Q. By contrast, the HWs increased with the event size: from ∼10 ms for small events to ∼40 ms for events >50 pC (Fig. 1g). A longer duration of the ATP-induced P2X₂R current would be consistent with the emptying of an enlarged granule lumen, which may result if several granules combine within the cell and then fuse with the plasma membrane as one unit (multivesicular exocytosis) or rapid sequential exocytosis emptying via a single fusion pore.

If multivesicular exocytosis occurs, then there should be a correlation between the Q and the amount of membrane added (measured as ΔC _m). Figure 2a,b shows examples of events associated with small and large P2X₂R currents and parallel ΔC _m in rat beta cells dialysed with 0.2 μmol/l [Ca²⁺]_i. In both cases, the ATP-evoked whole-cell currents rose and decayed monotonically with no sign of superimpositions of discrete events. Statistical analysis suggests that it is highly unlikely that these events result from superimposition of multiple smaller events (ESM text, section 2). For comparison, Fig. 2c shows an example where a step capacitance increase of ∼10 fF is associated with four discrete spikes that are superimposed. Thus we can estimate that each release event adds ∼2.5 fF of capacitance, close to the 2.9 fF measured by on-cell capacitance measurements [26]. Figure 2d shows the relationship between ΔC _m and Q (only including events without signs of superimpositions). The largest steps averaged ∼35 fF, corresponding to the simultaneous fusion of more than ten granules with the plasma membrane.

It would appear that the increase in capacitance (reflecting the fusion of the granules with the plasma membrane) should precede the emptying of the granules (detected as the activation of the P2X₂Rs). However, intragranular ATP is highly mobile and exits promptly upon membrane fusion, even before complete expansion of the fusion pore [26‐29]. It is accordingly not possible to temporally dissociate the capacitance increase from ATP release. Indeed, the capacitance steps recorded in beta cells not infected with P2X₂Rs exhibited a kinetics (rise time ∼20 ms; Fig. 2e) that was, if anything, slower than that of the events recorded with the P2X₂R-based assay (cf. Fig. 1f). It can also be noted, however, that the large ΔC _m steps develop with a time course only marginally slower than that seen for the small steps, a feature not consistent with the notion that they result from rapid sequential exocytosis of individual secretory granules.

The findings of Fig. 1 indicate that a global elevation of [Ca²⁺]_i (by intracellular dialysis via the recording electrode) promotes multivesicular release. Acetylcholine stimulates insulin secretion by activation of muscarinic M3 receptors [30] and enhances insulin release through a combination of several effects [31] including activation of protein kinase C (PKC) and inositol trisphosphate (IP₃)-dependent mobilisation of intracellular Ca²⁺ [32, 33]. Figure 3a shows changes in [Ca²⁺]_i evoked by 20 mmol/l glucose alone and after addition of 20 μmol/l carbachol in an intact rat pancreatic islet. Glucose produces a non-oscillatory elevation of [Ca²⁺]_i, and carbachol elicits a transient further elevation. We confirmed by confocal imaging of [Ca²⁺]_i (in agreement with previous reports [34]) that carbachol elicited a widespread increase in [Ca²⁺]_i in the cell centre (Fig. 3b,c).

We monitored exocytosis before and after the addition of 20 μmol/l carbachol to individual rat beta cells infected with P2X₂Rs using the perforated patch whole-cell technique. The cells were exposed to 10 mmol/l glucose and the membrane potential initially held at −70 mV to keep the voltage-gated Ca²⁺ channels closed. Under these conditions, the rate of exocytosis was low, and P2X₂R currents not observed. Depolarisation to −40 mV resulted in sufficient activation of the Ca²⁺ channels to initiate exocytosis (Fig. 3d). Figure 3e,f shows examples of the P2X₂R currents elicited at −40 mV in the presence of glucose alone and after addition of carbachol. It is evident that addition of carbachol increases the amplitude and duration of the events, but their frequency was not affected (0.22 ± 0.05 events/s and 0.17 ± 0.04 events/s in the absence and presence of carbachol, respectively). The cumulative histograms of Q in the absence and presence of carbachol are shown in Fig. 3g. The distribution was shifted to the right in the presence of 20 μmol/l carbachol (p < 0.001). The effects of carbachol cannot be attributed to changes in the gating of the P2X₂R currents; the integrated current evoked by application of 30 μmol/l ATP in the presence of 20 μmol/l carbachol averaged 101 ± 8% (n = 4) of that evoked in the absence of the agonist. The effect of carbachol was not mimicked by application of the PKC activator phorbol 12-myristate 13-acetate (PMA; 162 nmol/l, not shown), was insensitive to the PKC inhibitor chelerythrine (1.3 μmol/l; not shown), but partially antagonised by pretreatment of the cells with 20 μmol/l the myristoylated CAMKII autoinhibitory peptide (Fig. 3h).

We investigated the occurrence of multivesicular exocytosis in beta cells within intact pancreatic islets using SRB imaging. Figure 4a shows a region of an islet exposed to 20 mmol/l glucose. Exocytotic events were identified as the rapid appearance of fluorescent spots with a diameter of >0.15 μm that are easily distinguished from endocytotic events, most of which have diameters of <0.08 μm [35, 36]. The arrows highlight three exocytotic events. The islet was subsequently exposed to 20 μmol/l carbachol in the continued presence of 20 mmol/l glucose. Figure 4b,c shows magnified views of small and large events observed during carbachol stimulation. The larger event appeared within the cell and seemed connected to the extracellular space via a stalk of dim fluorescence.

The distribution of the corrected diameters of the fluorescent spots seen in the presence of glucose alone and in the simultaneous presence of glucose and carbachol are shown in Fig. 4d,e. The events evoked by 20 mmol/l glucose alone had a diameter of 0.23 ± 0.02 μm, in reasonable agreement with that estimated for insulin granules by electron microscopy [37]. The cumulative distributions of the diameters observed at 20 mmol/l glucose alone and after supplementation of the medium with 20 μmol/l carbachol are shown in Fig. 4f; the distribution is shifted to the right in the presence of carbachol (p < 0.001). In the presence of carbachol, 15% of the events were larger than any of the events observed in the presence of glucose. The diameter of this subgroup of events averaged 1.20 ± 0.23 μm, 5.2-fold larger than the average of the events evoked by glucose alone. The frequency of exocytosis increase averaged 2.4 events/min per islet in both the absence and presence of carbachol in islets exposed to 20 mmol/l glucose (not shown). However, the fraction of events that were of the compound type increased >3.5-fold, from 5% in the presence of 20 mmol/l glucose alone (Fig. 4g) to 18% after addition of 20 μmol/l carbachol (Fig. 4h). In islets exposed to 20 mmol/l glucose, a 1 min stimulation with 20 μmol/l carbachol stimulated insulin secretion by 35% (Fig. 4i). The corresponding stimulation measured over 20 min averaged 49% (not shown). In islets exposed to 1 mmol/l glucose throughout, insulin secretion was <20% of that seen in the presence of glucose.

Representative recordings of the SRB fluorescence changes in the presence of glucose alone and when the medium was supplemented with carbachol are shown in Fig. 5a,b. Most events recorded in the presence of carbachol were of the same size as those observed with glucose alone. Occasionally, however, much larger events were observed (Fig. 5c). These large events increased monotonically without any irregularities suggestive of smaller events being superimposed. The relationship between the diameter of the event and the rise times is shown in Fig. 5d. Under control conditions (20 mmol/l glucose), the rise times of events averaged 3.9 ± 0.6 s and did not vary much with the diameter. In the presence of carbachol, events with a diameter of <0.4 μm (similar to those evoked by glucose alone) had a mean rise time of 5.10 ± 1.20 s. However, for larger events the rise time increased to 9.1 ± 1.6 s (p < 0.001 vs all events at 20 mmol/l glucose alone). Possible reasons why the rise times reported by the SRB imaging are ∼1,000-fold longer than those indicated by the measurements of ATP release are discussed in the ESM text (section 3). In contrast, the HW of the events was constant over a wide range of spot areas (not shown). Notably, 15% of the events have HWs >20 s and required >30 s to return to the baseline (Fig. 5e).

In addition to multivesicular exocytosis, we also observed examples of sequential exocytosis that took place in rapid succession (10 s) within the same part of the beta cell and almost returned to baseline before the next event (Fig. 5f). These events were easily distinguished from the multivesicular events in displaying temporally well-separated stepwise increases in fluorescence (Fig. 5g); 7% of the events were of this type in islets exposed to 20 mmol/l glucose alone, which increased to 15% in the presence of carbachol (Fig. 4h).

We used the fixable fluorescent styryl dye FM1-43FX [38] to study exocytosis in rat beta cells. This dye cannot cross membranes and only labels membranes in direct contact with the extracellular solution. The cells were exposed to either 20 mmol/l glucose alone or in combination with 20 μmol/l carbachol (for 30 s). The FM1-43FX dye was present only during the last 30 s. Previous observations [18, 39] and the present live cell imaging of exocytosis (Fig. 5e) suggest that many of the release events have sufficiently long lifetimes to be captured by this protocol. Figure 6a shows a 3-D reconstruction of the most common type of structures labelled in the presence of 20 mmol/l glucose alone (ESM Movie 1). Figure 6b summarises the distribution of the length/diameter of these structures labelled by FM1-43FX in cells exposed to glucose alone; the distribution was nearly Gaussian with a mean of 0.60 ± 0.05 μm (n = 140). Fluorescent beads with a diameter of 0.20 μm (i.e. similar to the 0.3 μm diameter of insulin granules in rat beta cells [40]) had a measured diameter of 0.50 ± 0.02 μm (n = 12). The cumulative distribution of these events is shown as the dashed line in Fig. 6c. In cells exposed to carbachol for 30 s (Fig. 6d), >15% of the structures labelled by FM1-43FX were larger than ≥2 μm. In islets exposed to 20 mmol/l glucose alone, the corresponding number was <5%. A 3-D reconstruction of one of the large events is shown in Fig. 6d (ESM Movie 2) that remain connected to the cell surface. In many cases, these exhibited a serpentine-like morphology (hence these data are quoted as lengths rather than diameters). These events are too big to represent endocytotic vesicles (diameter = 70–80 nm [35, 36, 41]). The cumulative distribution of the FM1-43FX-labelled structures in cells exposed to glucose in the presence of carbachol is shown as the continuous line in Fig. 6c. The distribution of the events was shifted to the right in the presence of carbachol (p < 0.001).

We examined whether carbachol increases the prevalence of homotypically fused granules within the cytosol by serial scanning electron microscopy [23]. Figure 7a shows an xy plane of a beta cell together with 3-D rendered xz and yz planes. Regardless of whether the islets had been treated with carbachol or not, most granules were not connected. However, occasionally prefused multigranular structures were observed (Fig. 7b). On average, carbachol increased the number of homotypically fused granules threefold; in the presence of carbachol, 4% of the granules were joined to at least another granule (Fig. 7c). The largest number of fused granules observed was six. The example shown was within ∼400 nm of the plasma membrane. In cells exposed to carbachol, the average distance between the multivesicular structure and the plasma membrane averaged 1 ± 0.2 μm, with about one-third residing within one granule diameter (0.3–0.4 μm) of the plasma membrane. We confirmed the occurrence of multivesicular complexes using transmission electron microscopy, which allows more unequivocal identification of the granule membranes. These experiments provided additional evidence of connected granules (Fig. 7d,e), occasionally adjacent to the plasma membrane.