SEARCH

SEARCH BY CITATION

Keywords:

  • cytosolic calcium concentration;
  • insulin granules;
  • insulin secretion;
  • pancreatic islets;
  • plasma membrane depolarization

Abstract

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Like primary mouse islets, MIN6 pseudoislets responded to the depolarization by 40 mm KCl and the resulting increase in the free cytosolic Ca2+ concentration ([Ca2+]i) with a massive increase in insulin secretion, whereas 15 mm KCl had little effect in spite of a clear increase in [Ca2+]i. Analysis of insulin-enhanced green fluorescent protein (EGFP)-labeled granules in MIN6 cells by total internal reflection fluorescence (TIRF) microscopy showed that 40 mm KCl increased the number of short-term resident granules (<1 second presence in the submembrane space), while the total granule number and the number of long-term resident granules decreased. The rates of granule arrival at and departure from the submembrane space changed in parallel and were two orders of magnitude higher than the release rates, suggesting a back-and-forth movement of the granules as the primary determinant of the submembrane granule number. The effect of 15 mm KCl resembled that of 40 mm but did not achieve significance. Both 15 and 40 mm KCl evoked a [Ca2+]i increase, which was antagonized by 10 µm nifedipine. Nifedipine also antagonized the effect on secretion and on granule number and mobility. In conclusion, during KCl depolarization L-type Ca2+ channels seem to regulate two processes, insulin granule turnover in the submembrane space and granule exocytosis.

Physiologically, the stimulated insulin secretion requires the activation of the beta-cell energy metabolism and the depolarization of the plasma membrane potential (1,2). The link between the energy metabolism and the electrical activity of the plasma membrane is the ATP-sensitive K+ channel (KATP channel), the activity of which is inhibited by an increase in the ATP/ADP ratio (3). However, it was found that glucose was still able to increase the secretion rate when KATP channels were blocked by sulfonylureas at maximally effective concentrations (4). The same applied when KATP channels were clamped open by diazoxide in conjunction with a depolarization by a high K+ concentration (5). Apparently, two different pathways affect insulin secretion, one involving KATP channel closure and depolarization of the plasma membrane, the other being independent of KATP channel activity. For the latter pathway, the term ‘amplifying pathway’ was suggested because it appeared unable to stimulate insulin secretion on its own (6). In contrast, the KATP channel-dependent pathway was named ‘triggering pathway’ as it is believed to produce the decisive triggering signal for insulin granule exocytosis, namely the depolarization-induced Ca2+ influx (6), which underlies the action potential spiking in beta cells (7).

As a number of observations suggest that the first phase of glucose-induced insulin secretion is formed by the Ca2+ influx acting on a limited pool of secretion-ready granules, the triggering pathway is held responsible for the generation of the first phase of glucose-induced insulin secretion (8). The second phase is believed to result from the increasing velocity of granule translocation from the reserve pool and the concomitant granule maturation, a function which fits to the characteristics of the amplifying pathway (9). As a consequence of the above, it is generally assumed that the depolarization of the beta-cell plasma membrane generates a first phase-like response without the need of an activated beta-cell energy metabolism (10,11).

Although the existence of an amplifying effect during the first phase has been recently described (12), it was unexpected that 15 mm KCl, which has a depolarizing strength of about 21 mV and elicits a marked continuous elevation of the free cytosolic calcium concentration ([Ca2+]i), produced only a very modest and short-lived insulin secretion in the presence of a basal glucose concentration, much less than the first-phase response during glucose stimulation (13). Specifically, this observation raised questions as to the precise role of the Ca2+ influx and the release-ready granules as the determinants of the initial insulin response. Why is the Ca2+ influx evoked by 40 mm KCl so much more effective to elicit a virtually immediate secretory response (13), when small amounts of Ca2+, sufficient to generate a Ca2+ microdomain around Ca2+ channels, are believed to trigger the fusion of release-ready insulin granules (8,14,15)? We characterized, therefore, insulin secretion, [Ca2+]i and the number and mobility of submembrane insulin granules during sequential stimulation with 15 and 40 mm KCl. To obtain the latter parameters, we used total internal reflection fluorescence (TIRF) microscopy combined with an observer-independent evaluation program.

Results

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Freshly isolated islets perifused with 5 mm glucose showed only a modest transient increase in secretion when K+ was raised to 15 mm. A subsequent elevation to 40 mm K+ resulted in a prompt overshooting increase in secretion which remained increased as long as 40 mm K+ was present (Figure 1A). When MIN6 pseudoislets were used instead of normal mouse islets, the secretion pattern was practically the same. Again, there was only a minimal increase in secretion in response to 15 mm K+ and a very strong response to 40 mm K+ (Figure 1B). When the same protocol was used to measure [Ca2+]i of perifused NMRI mouse islets, a marked and enduring increase was elicited by 15 mm K+ (Figure 1A). When K+ was raised to 40 mm, a prompt further increase in [Ca2+]i resulted with an initial overshoot (Figure 2). The same response pattern of [Ca2+]i increase was observed when MIN6 pseudoislets were perifused with 15 mm and then 40 mm K+ (Figure 1B). Thus, it was ascertained that the response of the MIN6 cells to moderate and strong K+ depolarization corresponds to that of normal mouse beta cells.

image

Figure 1. Effect of a two-step increase in K+ on insulin secretion and [Ca2+]i of perifused mouse islets (A) or of perifused MIN6 pseudoislets (B). Fifty freshly isolated islets or 100 pseudoislets were used for one secretion measurement, one single Fura-PE3/AM-loaded cultured islet or pseudoislet was used for one [Ca2+]i measurement. The glucose concentration of the perifusion medium was 5 mm (islets) or 3 mm (pseudoislets) throughout. From 60 to 80 min the medium contained 15 mm K+, from 80 to 100 min it contained 40 mm K+, thereafter the K+ concentration was set again at a physiological value (5.6 mm). Values are means ± SEM of four to five experiments.

Download figure to PowerPoint

image

Figure 2. Mobility and exocytosis of secretory granules labeled by insulin-EGFP. A) Identification and tracking (from top to bottom): MIN6 cells transiently transfected with insulin-EGFP were observed using objective-based TIRFM (100× magnification). A 3D surface plot was drawn to exactly localize each fluorescent granule. Granules thus identified are shown superimposed on the raw intensity data. The identified granules could be tracked for the quantitation of mobility. B) Identification of exocytosis: A short (one or two images) transient increase in insulin-EGFP fluorescence, probably due to an alkalinization of intragranular pH, immediately followed by a marked loss of fluorescence intensity is the hallmark of insulin exocytosis. The image sequence below shows the pertaining granule, the dissipating cloud of fluorescent material at the moment of exocytosis and the slow dimming of the residual fluorescent spot.

Download figure to PowerPoint

MIN6 cells transfected with insulin-enhanced green fluorescent protein (EGFP) were used to analyze the behavior of submembrane granules by TIRF microscopy at five time-points during a continuous perifusion: beginning, prior to high K+, one each during 15 and 40 mm K+ and one after washout of high K+. At each time-point, a sequence of 100 images was acquired within 12 seconds. Evaluation of the granule number and mobility was performed by a purpose-made program. Insulin granules were identified by a three-dimensional (3D) polynomial convolution of the intensity profile, the maximal value of which gave the exact and reproducible position of the granule as a requirement for tracking (Figure 2A). The threshold for granule detection was set after moving it through the range of fluorescence intensity which was subdivided in 20 levels. Typically, the cell area in the lower two or three levels was uniformly filled (background noise, probably autofluorescence), then at the third or fourth level a more granular pattern appeared which persisted for the subsequent six or seven levels. Exocytosis was detected by the transient brightening of a granule, immediately followed by a marked loss of fluorescence (Figure 2B). This was done by visual inspection of the movie files, then the granule was identified in the computerized data file and the mobility preceding exocytosis was quantified.

The residence time of the granules within the submembrane space ranged from 0.12 seconds (one image) or less to 12 seconds (entire sequence) and longer (see below). The total itinerary, which corresponds to the mobility in the x/y plane, correlated with the residence time (Figure 3A). However, this correlation was somewhat obscured by the fact that the large majority of the cumulative granule number (sum of all granules identified during one 12-second sequence) is present for less than 1 second. The distribution of the granule residence times was best described by a biexponentially decaying function with half-lives of 0.053 ± 0.006 and 0.211 ± 0.043 seconds (n = 4), respectively (Figure 3B). Assuming that this defines two or potentially more subpopulations of ‘newcomer’ and more stationary granules, we subdivided the granules into three groups. As the faster decaying function approaches zero at image 9, all granules that were visible from 1 to 9 images (ca. 1 second) were defined as short-term residents. Those which were continuously present during the entire sequence (100 images = 12 seconds and potentially longer) were classified as long-term residents and those in between as medium-term residents. Considering that medium-term residents might represent a mixed population, only the data of the short-term and long-term resident granules are given for the below experiments.

image

Figure 3. Properties of secretory granules labeled by insulin-EGFP during a 12-second data acquisition sequence. A) Synopsis of the granule mobility as shown by the total itinerary (z-axis, 1 pixel corresponding to 75 nm) in dependence of the residence time in the submembrane space (x-axis). The number of granules at each point of residence time and itinerary length is given by the y-axis (note the logarithmic scale). The vast majority of the granules in the submembrane space was present for less than 1 second and had a total itinerary of less than 750 nm. The high number of granules at the last time-point (last image) is an artifactual accumulation and not seen with longer acquisition sequences. B) Variability of the granule residence time in the submembrane space. The relation of granule number versus residence time was significantly better described by a biexponential (red curve) than by a monoexponential fit (blue curve). For clarity only the data of the first 50 images are shown.

Download figure to PowerPoint

For comparison between the sequences, the number of submembrane granules per cell was determined in the first image of each sequence and that of the first sequence (89.2 ± 9.7, n = 17, equivalent to 66.3 ± 6.1 granules per 100 and 135 µm2 mean footprint area per transfected MIN6 cell) was normalized to 100%. Exposure to 15 and 40 mm KCl reduced this number to 75 and 56%, respectively, and washout of 40 mm KCl by 5.6 KCl did not lead to a recovery within 10 min (Figure 4). In control perifusions without KCl depolarization, no decrease in granule number occurred (Figure 4). At a given time-point (i.e. during one 50-millisecond exposure) under basal conditions, 65.2 ± 5.1% of the granules were long-term residents, 24.7 ± 3.6% medium-term residents and 10.1 ± 1.8% short-term residents (n = 6). However, this count does not reflect the different kinetics of the granules. When the granule counts for the entire 12-second sequences were summed up, 73.1 ± 4.3% of the granules classified as short-term, 14.7 ± 1.8% medium-term and 12.1 ± 3.4% long-term residents.

image

Figure 4. Analysis of submembrane granule number and z-direction mobility in MIN6 cells during moderate and strong K+ depolarization and washout. At five time-points during an experiment as depicted in Figure 1B, a sequence of 100 images was acquired during 12 seconds. These time-points are indicated by the arrows and the pertaining data are given by the numbers and bar graphs below. The following parameters are given (numbers and bar graph rows from top to bottom): total number of granules (in the first image of each sequence, normalized to 100% for the start value), long-term resident granules (% of total granules in the first image per sequence), short-term resident granules (% of total granule number per sequence), newly arriving granules (% of total granule number per sequence) and number of departing granules (% of total granule number per sequence). The respective control values are given in the same row as clear columns. Values are means ± SEM of three experiments each.

Download figure to PowerPoint

In addition to determining the residence time in the submembrane space, the number of appearing and disappearing granules during one sequence was quantified (Figure 4). Both arrivals at and departures from the submembrane space make up the granule mobility along the z-axis (i.e. orthogonal to the plasma membrane). The number of arrivals increased from 5.0 ± 1.4% initially to 7.7 ± 2.0% during 15 mm K+ and to 12.4 ± 1.6% during 40 mm K+ and 11.7 ± 1.4% washout (n = 3). Very similar percentages applied for departures, which comprise granules returning to the cell interior and granules releasing their insulin content (Figure 4). In consequence, the number of short-term resident granules (≤1 second) increased slightly during 15 mm K+, then significantly during 40 mm K+ (p = 0.002 versus control) and remained so after washout (Figure 4). Assuming that the difference between arrivals and departures might permit an indirect estimate of exocytotic events, a balance sheet was set up. However, the balance of arrivals and departures was not affected by increasing the K+ concentration to 15 or 40 mm (data not shown).

To determine the role of L-type Ca2+ channels in the depolarization-induced insulin granule mobility changes, the effect of 10 µm nifedipine on the [Ca2+]i of perifused islets was measured. Both 15 and 40 mm K+ markedly increased [Ca2+]i, but the increase by 40 mm K+ was significantly larger (Figure 5). After 5 min 10 µm nifedipine was added and diminished the elevated [Ca2+]i to a new steady state. In the presence of 40 mm K+, this level was similar to the steady state of 15 mm K+ alone. In the presence of 15 mm K+, a nearly complete return to baseline was achieved by nifedipine (Figure 5). The same protocol was used to assess the inhibitory effect of nifedipine on the insulin secretion by K+ depolarization. Nifedipine antagonized the modest insulinotropic effect of 15 mm K+, lowering the secretion rate below prestimulatory values (Figure 6). The insulinotropic effect of 40 mm K+ was reduced to 47% during exposure to nifedipine (Figure 6). Taking into account that the full effect of nifedipine was achieved only after 5 min, the nifedipine-resistant secretion area under the curve can be given as 26%.

image

Figure 5. Contribution of L-type Ca2+ channels to the Ca2+ influx elicited by 15 and 40 mm K+. A) Antagonistic effect of 10 µm nifedipine on the [Ca2+]i increase in Fura-PE3/AM-loaded cultured islets caused by 40 mm K+. The gray lines show the [Ca2+]i increase caused by 40 mm K+ without addition of nifedipine. Values are means ± SEM of three experiments. B) Antagonistic effect of 10 µm nifedipine on the [Ca2+]i increase in Fura-PE3/AM-loaded cultured islets caused by 15 mm K+. The gray lines show the [Ca2+]i increase caused by 15 mm K+ without addition of nifedipine. Values are means ± SEM of three experiments.

Download figure to PowerPoint

image

Figure 6. Contribution of L-type Ca2+ channels to the stimulation of insulin secretion by 15 and 40 mm K+. Fifty freshly isolated islets were perifused with 5 mm glucose throughout. From 60 to 80 min the medium contained 15 mm K+ (open circles) or 40 mm K+ (closed circles), from 65 to 80 min it contained additionally 10 µm nifedipine. The modest increase in secretion by 15 mm K+ was completely suppressed by 10 µm nifedipine, the strong increase by 40 mm was partially antagonized. The gray dotted line denotes the effect of 40 mm K+ alone (adapted from Ref. 13). Values are means ± SEM of four to five experiments.

Download figure to PowerPoint

In light of these results, three time-points were selected to measure the granule mobility parameters as defined above: one prior to the increase in the K+ concentration, one after 2 min of exposure to the high K+ concentration (40 mm) and one after 5-min additional exposure to 10 µm nifedipine. At each time-point, a sequence of 200 images was acquired within 24 seconds (Figure 7). Manual counting of transient fluorescence increases as a measure of exocytosis gave 0–2 events per sequence under basal conditions, 0–5 events in the presence of 40 mm K+ and 0 events in the additional presence of nifedipine (Movies S1 and S2). As a rough estimate, the rate of exocytosis can be given as 20–30 per min and cell in the presence of 40 mm K+. When the granule number of the first image in the first sequence was normalized to 100%, the number decreased to 90% after 2-min exposure to 40 mm K+ and to 75% after 5-min exposure to nifedipine. In the control experiments (continuous exposure to 40 mm KCl for 20 min), the corresponding values were 78 and 58%, respectively.

image

Figure 7. Analysis of submembrane granule number and z-direction mobility during a depolarization by 40 mm K+ and the subsequent addition of 10 μm nifedipine. At three time-points during an experiment as depicted in Figure 5A, a sequence of 200 images was acquired during 24 seconds. These time-points are indicated by the arrows and the pertaining data are given by the numbers and bar graphs below. The following parameters are given (numbers and bar graph rows from top to bottom): total number of granules in the first image of each sequence (% of the normalized initial value), long-term resident granules (% of total granules in the first image per sequence), short-term resident granules (% of total granule number per sequence), newly arriving granules (% of total granule number per sequence) and number of departing granules (% of total granule number per sequence). The respective control values are given in the same row as clear columns. Because of the longer sequences, the percentages of short-term, arriving and departing granules were higher than in the experiments depicted in Figure 4. Values are means ± SEM of four experiments each.

Download figure to PowerPoint

While the normalized granule number did not show a clear-cut effect of nifedipine, the separate determinations of long-term and short-term resident granules did. As in the above experiments, the number of long-term resident granules was decreased by 40 mm K+. Nifedipine led to a recovery to the prestimulatory level which was significantly different from the control value (continued exposure to 40 mm K+, Figure 7). Vice versa, the number of short-term granules per sequence, which was increased by 40 mm K+, was decreased by nifedipine, again resulting in a significant difference from the control value (Figure 7). This pattern was confirmed by separate counts of arrivals and departures (Figure 7).

To describe the effects of KCl depolarization on the movement parallel to the membrane (x/y plane), the itineraries of all granules identified per sequence were measured. The total itinerary was the sum of all movements of one granule during a sequence, the net itinerary was the distance between the points of arrival and departure and the ratio of the net to total itinerary was defined as a measure for the directed granule movement. Plotting this ratio as a function of the total itinerary gave a family of hyperbola curves under all three conditions (Figure 8A). Subtraction of the granule distribution pattern during prestimulatory conditions or that during nifedipine exposure from that during K+ depolarization did not reveal a qualitatively different pattern during K+ depolarization (data not shown). The granules that showed a transient increase in fluorescence (probably indicating exocytosis) had a common property in that their total itineraries were in the maximal range, but their net itineraries in the lower range, resulting in a minimal ratio value (Figure 8A, middle panel).

image

Figure 8. Analysis of granule mobility in the x/y plane during a depolarization by 40 mm K+ and the subsequent addition of 10 μm nifedipine. The original data set is the same as in Figure 7. A) The directedness of granule motion (the net itinerary divided by the total itinerary) is shown as a function of the total itinerary (1 pixel = 75 nm). Under all three conditions, a family of hyperbolas resulted. The short-term resident granules are depicted in green, the medium-term and long-term granules in blue. Considering all granules the relation between itinerary and directedness remained unaltered by exposure to 40 mm K+ or to 40 mm K+ plus nifedipine. The circle in the middle panel indicates the area containing the exocytosed granules. B) Normalized mean values of the total and the net itineraries under the three conditions depicted in (A). In addition, the same parameters were calculated for two control experiments, the prolonged exposure to 40 mm K+ and the continuous perifusion with 5.6 mm K+. Values are means ± SEM of four experiments each.

Download figure to PowerPoint

For quantitative comparison, the mean values of the itineraries were calculated. After normalization of the values measured during the first sequence, a clearer picture emerged (Figure 8B). In the presence of 40 mm K+, the net but not the total itineraries were longer than those in the continued presence of 5.6 mm K+. Under this condition, both net and total itineraries decreased spontaneously by about 30% during the experiment (Figure 8B). However, the increase in the net versus total itineraries during exposure to 40 mm K+ is most likely secondary to the increased percentage of the short-term resident granules (Figure 7), which have a much higher ratio of net versus total itineraries (Figure 8A).

Finally, the fate of the long-term residents was investigated, which was possible because in this set of experiments the sequences consisted of 200 instead of 100 images. When those granules that were continuously present for the first 100 images were followed up during the next 100 images, it turned out that 26% of them departed under basal conditions (continuous presence of 5.6 mm K+). In the presence of 40 mm KCl, this number was 32% which decreased (marginally significant, p = 0.06, n = 4) to 17% in the additional presence of nifedipine.

Discussion

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

This study is the first to describe the behavior of the entire insulin granule population in the submembrane space. There were two reasons to observe the effects of depolarization by 15 and 40 mm K+. The first was that in contrast to 40 mm K+, 15 mm K+ proved nearly unable to stimulate insulin secretion in the presence of a basal glucose concentration, although it had a marked depolarizing effect (by 20 mV) and clearly increased [Ca2+]i(13). As the [Ca2+]i increase by 15 mm K+ is comparable to that produced by 20 mm glucose (peak values of 630 versus 490 nm; Rustenbeck, unpublished), the discrepancy is not easily explained in the current theoretical framework, which proposes that a depolarization-induced influx of Ca2+ should lead to the release of a number of secretion-ready insulin granules without additional requirements (9). The other reason was that MIN6 cells showed the same response pattern of secretion and of [Ca2+]i to a sequential stimulation by 15 and 40 mm K+ as normal mouse islets and could therefore be used as a substitute for primary beta cells.

Continuous recording times for longer than 1 min proved to be problematic because of focus drift and progressive photobleaching of the fluorescent granule label (insulin-EGFP), both of which are known problems in live-cell TIRF microscopy (16). Instead, image sequences of 12-second duration at five selected time-points were recorded during an experiment which followed exactly the same protocol as used for the [Ca2+]i and the conventional insulin secretion measurements. This way, the focus drift, which was strongly accelerated by the exchange of the perifusion medium, could be corrected for prior to the registration of the next sequence.

To measure the movement, appearance and disappearance of the submembrane granules, the data acquisition rate was as high as compatible with a reliable granule identification one side and the avoidance of photobleaching on the other side. An exposure time of 50 milliseconds, resulting in a work cycle of 120 milliseconds, proved to be a satisfactory compromise but it cannot be excluded that there were faster events that have eluded our registration. The transient increase in fluorescence of the insulin-EGFP-fusion protein is accepted to report exocytosis (17–19), although the reliability has been questioned (20). These events were counted by visual inspection of the image files and were two orders of magnitude less frequent (0–5 events per sequence during stimulation by 40 mm K+) than the automatically counted rates of appearance and disappearance of the granules. Such low numbers do not permit a quantification of the secretory activity, but the overall pattern conformed to the secretion as measured by the batch perifusions of MIN6 pseudoislets.

The mean number of visible submembrane granules at the beginning of the experiments was about 90 per cell. Taking into account that the cellular ‘footprint’, which is visible to the TIRF excitation, makes up about 30–40% of the total cellular surface (21,22), a number of ca. 270 granules can be estimated. Taking further into account that about 80% of the granules are labeled and that primary beta cells are more densely granulated than insulin-secreting cell lines, this is still in reasonable agreement with the data by electron microscopy of primary mouse beta cells. These are 600 for the granules docked at the plasma membrane and 2000 for the granules being no farther away from the plasma membrane than one granule diameter (23). As the insulin granule diameter is about 350 nm (23, for different estimates depending on the fixation method see Ref. 24), which is two to three times the penetration depth of our TIRF field, a number of 300–400 submembrane granules per MIN6 cell seems a reasonable estimate.

From the present data, the time an insulin granule spends in the submembrane space is variable by a factor of at least 100. As the residence time histogram could be fitted by a biexponentially decaying function, one may presume that two different populations exist. Alternatively, it can be assumed that two (and potentially more) factors influence the submembrane residence time. Although the definition of short-term and long-term resident granules may appear somewhat arbitrary, the subdivision and separate evaluation are justified by the subpopulation-specific behavior. The existence of two distinct populations has been suggested recently, proposing that insulin granules exist in low- and high-affinity states for Ca2+. Those in the low-affinity state would be situated close to the Ca2+ channels and correspond to the readily releasable pool (RRP), whereas those in the high-affinity state would be localized farther away from the plasma membrane (18,19,25). Similarly, an earlier investigation has suggested granules of two different Ca2+ affinities to exist in the reserve pool (26). These theories view the secretory granules as rather static entities, which are subject to a fast regulation by different sources of [Ca2+]i. However, compared to the time scale of insulin release, the granules appear to be a highly dynamic population.

Hypotheses of insulin granule exocytosis have so far only considered a unidirectional transport of the granules to the release sites (9,27). However, the frequency of arrival and departure in the submembrane space that was higher than the frequency of release events by orders of magnitude suggests that an insulin granule present in the submembrane space may not only be released but may also return to the cell interior. Taking into account that the peak rate of secretion by 40 mm K+ is higher than that of the first phase of glucose-stimulated secretion by a factor of 2–3 (100–120 versus 40–60 pg/min × islet; Ref. 13, Rustenbeck, unpublished observations), the estimated number of release events after 2 min of perifusion with 40 mm K+ (ca. 20–30 per cell and minute) is in reasonable agreement with reported rates of exocytosis during the first phase (15 per min and beta cell; Refs 9 and 28). The small number of release events per sequence is probably the reason why, in contrast to our expectations, the balance of arrivals and departures was not systematically affected by increasing the K+ concentration, and thus could not substitute for actual measurements of exocytosis.

The parallel change of arrivals and departures under the various conditions suggests that a constant back-and-forth movement of granules orthogonal to the plasma membrane takes place, which is intensified by the depolarization-induced [Ca2+]i increase. In contrast, the parameters of granule mobility parallel to the membrane (total itinerary, net itinerary, directed movement) remained unchanged when allowance was made for the increased share of short-term resident granules that occurred during K+ depolarization. This conclusion is in apparent contrast to an earlier report that attachment of the insulin granules to the release sites is preceded by random insulin granule diffusion (29). However, this report was based on confocal microscopy which has a lower resolution in the z-axis than TIRF microscopy (30). It is conceivable though that an intensified random diffusion increases the probability of a granule to pass (in either direction) through the F-actin web beneath the plasma membrane, which impedes the transit of secretory granules and appears to control docking (31,32). Such a mechanism would not require a change in the F-actin structure, which has been described only for a glucose stimulus, but not for K+ depolarization (33).

The [Ca2+]i increase by 40 mm K+ was partially reduced by 10 µm nifedipine, whereas the more limited one produced by 15 mm was nearly completely antagonized. This concurs with our observation that depolarization by both 15 and 40 mm K+ activates L-type Ca2+ channel activity, 15 mm being less effective but more specific than 40 mm (Willenborg and Rustenbeck, unpublished observation). The reduction by nifedipine of the secretion stimulated by 40 mm K+ was remarkably effective in view of the limited reduction of [Ca2+]i, while 15 mm K+ produced little that could be antagonized. Recently, it has been reported that the depolarization by 15 mm K+ yields a slower and less extensive increase in Ca2+ concentration in the submembrane space than 40 mm K+, and thus acts on the highly Ca2+-sensitive pool of newly arriving granules, analogous to the second-phase secretion (18,19). While the [Ca2+]i increase by 15 mm K+ is in fact lower than that of 40 mm K+, the resultant modest and transient increase in secretion bears little resemblance to the second phase.

As the increase in the arrivals, departures and in consequence of the short-term resident granules by 40 mm K+ was antagonizable by nifedipine, whereas a continued perifusion with 40 mm K+ led to a further increase together with a continuous elevation of [Ca2+]i, it appears logical that the granule translocation preceding docking and fusion is regulated by Ca2+ entry via L-type Ca2+ channels. However, nifedipine reversed also the reduction of the granule number (per single image) whereas washout of high K+ failed to do so (Figure 4), although both had a similar effect on [Ca2+]i. Either the prolonged depolarization (Figure 4) has led to an intracellular Ca2+ redistribution impeding the reversibility and/or [Ca2+]i may not be the only effector of voltage-dependent (L-type) Ca2+ channel activity. The latter possibility has been recently suggested based on experiments with non-permeable cations and mutated L-type channels (34,35).

The comparatively few granules that underwent fusion had mobility parameters in the x/y plane that were not randomly distributed among those of the total population. Rather, they combined very long total itineraries with short net itineraries. This behavior is compatible with a cage-like confinement as described for docked granules (21,36), but appears paradoxical in view of reports that granules primed for release are nearly immobile (21). A similar observation of increased granular mobility within a restricted space shortly before exocytosis has been made in chromaffin cells (37). This does not necessarily contradict the established sequence of steps in exocytosis but shows that the slow time scale of neuroendocrine and, specifically, beta-cell secretion does not extend to a similar slow kinetic of perfusion events in the submembrane granules With respect to the recently reported fast-fusion modes of insulin granules (38,39), it is worthwhile noting that none of the short-term resident granules were found to fuse during the exposure to 40 mm K+.

In the present investigation, the depolarization-induced reduction of the submembrane granule number correlated with a reduction of the long-term resident granules, which may be regarded as a depletion of the release-ready granules by exocytosis (9). However, a decrease in granule number was also visible after 15 mm K+, which had only a minimal effect on secretion. Also, the observation that 20–30% of the granules defined as ‘long-term resident' during exposure to 40 mm K+ departed during a subsequent 12-second sequence suggests that not only short-term residents but also these granules do return to a pool more distant from the plasma membrane, as this number is about 5–10-fold higher than the detected release events during the same time span. In fact, it has been noted earlier that a strong K+ depolarization (75 mm) led to a stronger reduction of ultrastructurally docked granules than corresponded to the calculated number of granules in the RRP (23). This was explained by postulating the release of an additional pool of ‘nearly’ readily releasable granules, but may have actually been due to a return of these granules to the cell interior.

On the whole, the present investigation suggests that a model can be set up in which a bidirectional transport of granules leads to high turnover of granules in the submembrane space. From there, some of the granules are recruited for exocytosis (Figure 9). The turnover rate may be lower in primary beta cells which showed a considerably lower diffusion velocity of the granules than clonal beta cells (22), and artifacts by use of the insulin-EGFP label cannot be excluded (20). However, such a model is compatible with a rapid appearance of newly formed granules containing recently synthesized insulin (40,41), with a balancing role of crinophagy which eliminates aged granules (42) and with a distal site of action for (amplifying) signals from the energy metabolism (43,44). Hence, the near inability of 15 mm K+ to stimulate insulin secretion in spite of Ca2+ influx via L-type channels may be due to a metabolism-dependent variable threshold for Ca2+-induced exocytosis and/or by effects of L-type Ca2+ channels additional to Ca2+ conductance which are evoked by 40, but not 15 mm K+.

image

Figure 9. The bidirectional granule transport model. The currently predominant model (left) assumes that as in other exocytotic cells the insulin granules (large dense-core vesicles) are translocated to the plasma membrane (double solid line) and, when docked, gain fusion competence (‘priming’), thereby forming the readily releasable pool. Both processes are believed to depend on metabolic activity (squiggle symbol), whereas the readily releasable granules only require an increase in Ca2+ in their immediate vicinity to fuse. Recent variants have added subdivisions and parallel tracks but retained the unidirectional sequence. The bidirectional model (right) assumes that the granules spend a variable time in the submembrane space. A minority fuses with the plasma membrane, whereas the majority returns to a distant pool possibly situated beyond the F-actin web (dotted line). From there, they may recycle to the submembrane space. The chance to gain fusion competence may increase with the residence time in the submembrane space. Repeated recycling may be associated with a decreased ability to gain fusion competence and increased probability of degradation. This model accommodates two salient features described in literature: the preferential release of newly synthesized granules and the release of granules which have spent little time in the submembrane space. Ca2+ influx or, more generally, Ca2+ channel-mediated signaling does not only affect fusion probability but also granule turnover.

Download figure to PowerPoint

Materials and Methods

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Chemicals

Fura-PE3/AM was from TEF Labs. Collagenase NB8 was purchased from Nordmark, DMEM and RPMI-1640 cell culture media and fetal calf serum were from Invitrogen. All other reagents of analytical grade were from E. Merck.

Plasmid construction

Insulin-EGFP (hIns-EGFP): The cDNA of human preproinsulin was amplified by polymerase chain reaction (PCR) generating the restriction sites XhoI and BamHI, which allowed for cloning into the mcs of the expression vector pEGFP-N1 (Clontech). To keep the reading frame intact, two bases had to be inserted. In consequence, the linker between the C-terminus of the A-chain and the N-terminal side of the EGFP molecule consisted of seven amino acids, namely Gly, Asp, Pro, Pro, Val, Ala and Thr.

Tissue and cell culture and transfection

Islets were isolated from the pancreas of NMRI mice by a collagenase digestion technique and hand-picked under a stereo microscope. Single cells were obtained by incubation of the islets for 10 min in a Ca2+-free medium and subsequent vortex mixing for 1 min. Islets and single islet cells were cultured in cell culture medium RPMI-1640 with 10% fetal calf serum (10 mm glucose until attachment, thereafter 5 mm glucose) in a humidified atmosphere of 95% air and 5% CO2 at 37°C. Insulin-secreting MIN6 cells (kindly provided by Jun-Ichi Miyazaki) were cultured in DMEM medium (high glucose, 4.5 g/L) supplemented with 4 mml-glutamine, 15% FBS, 50 µm 2-mercaptoethanol and penicillin/streptomycin at 37°C and 5% CO2. The cells were transfected in suspension using Lipofectamin 2000 (Invitrogen) according to the manufacturer's protocol and cultured on glass coverslips for 54–66 h under the same conditions. MIN6 cells, when growing in suspension, form aggregates of 3000–5000 cells, which can be used like primary islets to study the secretion dynamics (45). To avoid cell attachment to the surface of the plastic material and to promote the formation of cell aggregates, dishes for suspension cell culture were used. Otherwise, pseudoislets were cultured under the same conditions as monolayer MIN6 cells and were harvested after 8 days. By transfecting MIN6 cells with other insulin granule labels (e.g. C-peptide-mCherry) and by comparing this with acridine orange loading (1 µm for 10 min), we estimate that 2 days after transfection fusion proteins label ca. 80% of the insulin granules, the remaining 20% being due to granules formed before the transient transfection.

Measurement of insulin secretion

Batches of 50 NMRI mouse islets or 100 MIN6 pseudoislets were introduced into a purpose-made perifusion chamber (37°C) and perifused with a HEPES-buffered Krebs-Ringer medium (115 mm NaCl, 4.7 mm KCl, 2.6 mm CaCl2, 1.2 mm KH2PO4, 1.2 mm MgSO4, 20 mm NaHCO3 and 10 mm HEPES, 2 mg/mL BSA) saturated with 95% O2 and 5% CO2, which contained the respective secretagogue. The insulin content in the fractionated effluate was determined by enzyme-linked immunosorbent assay (ELISA) (Mercodia).

Microfluorimetric measurements of [Ca2+]i

Intact islets from NMRI mice were cultured on collagen-coated glass coverslips in Petri dishes and were used from day 2 to 4 after isolation. Similarly, MIN6 cells were cultured on l-polylysin-coated glass coverslips and were used 2 days after seeding. Fura-PE3/AM was loaded at a concentration of 2 µm for 45 min at 37°C. The coverslip with the attached islets or MIN6 cells was tightly screwed in a purpose-made perifusion chamber on the stage of an iMIC epifluorescence microscope (TILL Photonics). Temperature was maintained at 37.0 ± 0.1°C by overnight prewarming the system with an environmental control chamber. The islets or cells were perifused at a rate of 0.2 mL/min with a HEPES-buffered Krebs-Ringer medium, which was saturated with 95% O2 and 5% CO2. The fluorescence was excited by a xenon arc lamp (Polychrome V, TILL Photonics) at 340 or 380 nm, the emission was collected by a Zeiss Fluar objective [40×, 1.3 numerical aperture (NA)] and recorded by a cooled charge-coupled device (CCD) camera (Sensicam QE, TILL Photonics) under control of the TILLVision program.

TIRF microscopy

Before the start of the experiments, the transfected MIN6 cells were transferred to a HEPES-buffered Krebs-Ringer medium containing 3 mm glucose and perifused for at least 45 min. The microscope setup was the same as for the Fura fluorescence measurements, except for the light source, a 491-nm continuous-wave diode-pumped solid-state laser (75 mW Cobolt Calypso), run at 15%. The objective was a Zeiss α-Plan-Fluar (100×, 1.45 NA), the angle of incidence was 68° and the calculated decay constant (reduction of the initial intensity at the glass membrane interface to 1/e = 37%) of the evanescent field was 84 nm. One image pixel corresponded to 75 nm in the focal plane. Selection of the excitation and emission wavelengths was made with a dual-line filter set for 491/561-nm lasers (AHF Analysentechnik). The exposure time was 50 milliseconds per image, the cycle time for acquisition and storage was 120 milliseconds. The fluorescent spots of the hIns-EGFP-labeled granules were localized, counted and their mobility analyzed by an in-house written program using matlab 7.6.0 (R2008a).

Statistics

Graphpad Prism4 software (GraphPad) was used for statistic calculations. If not stated otherwise, ‘significant’ refers to p < 0.05, based on the two-sided unpaired Student's t-test.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

This work was supported by grants from the Deutsche Forschungsgemeinschaft (Ru 368/5-1), the Deutsche Diabetes-Gesellschaft and the foundation ‘Das Zuckerkranke Kind’. Skillful technical assistance by Verena Lier-Glaubitz and Carolin Rattunde is gratefully acknowledged.

References

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Movie S1: Total internal reflection fluorescence microscopy (TIRFM) movie of insulin-EGFP-labeled granules in MIN6 cells. The excitation wavelength was 491 nm (Cobolt Calypso), the objective was a Zeiss α-Plan-Fluar (100×, 1.45 NA), the angle of incidence was 68° and the calculated decay constant of the evanescent field was 84 nm. The acquisition rate was 8.3 Hz, exposure time 50 milliseconds. The cells were continuously perifused at a rate of 0.2 mL/min, the temperature was 37.0 ± 0.1°C. Three minutes prior to the acquisition of this sequence, the K+ concentration in the Krebs-Ringer perifusion medium was increased by switching to a medium containing 40 mM K+. The lag time before the perifusion chamber was reached by the medium was 1 min. Multiple transient brightenings indicate exocytotic activity.

Movie S2: Same experiment and same acquisition conditions as shown in Movie S1. In addition to 40 mM K+, 10 µM nifedipine was present in the perifusion medium. The exposure time of the MIN6 cells to nifedipine prior to the acquisition of the sequence was 5 min.

FilenameFormatSizeDescription
TRA_1231_sm_SuppMovie1.mov835KSupporting info item
TRA_1231_sm_SuppMovie2.mov612KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.