Insulin secretion from beta cells in intact mouse islets is targeted towards the vasculature
- 2k Downloads
We set out to test the hypothesis that insulin secretion from beta cells is targeted towards the vasculature.
The spatial location of granule fusion was identified by live-cell two-photon imaging of mouse pancreatic beta cells within intact islets, using sulforhodamine B labelling. Three-dimensional (3D) immunofluorescence of pancreatic slices was used to identify the location of proteins associated with neuronal synapses.
We demonstrated an asymmetric, non-random, distribution of sites of insulin granule fusion in response to glucose and focal targeting of insulin granule secretion to the beta cell membrane facing the vasculature. 3D immunofluorescence of islets showed that structural proteins, such as liprin, piccolo and Rab2-interacting molecule, normally associated with neuronal presynaptic targeting, were present in beta cells and enriched at the vascular face. In contrast, we found that syntaxin 1A and synaptosomal-associated protein 25 kDa (SNAP25) were relatively evenly distributed across the beta cells.
Our results show that beta cells in situ, within intact islets, are polarised and target insulin secretion. This evidence for an ‘endocrine synapse’ has wide implications for our understanding of stimulus–secretion coupling in healthy islets and in disease.
KeywordsBeta cell Exocytosis Insulin Islets Secretion Synapse Vasculature
Synaptosomal-associated protein 25 kDa
Soluble N-ethylmaleimide-sensitive factor attachment protein receptor
Whether pancreatic beta cells target insulin secretion towards the vasculature is controversial. An early experiment used chronic stimulation to reduce the numbers of insulin granules; the remaining granules were enriched in one region of the cell, suggesting polarisation of either granule trafficking or secretion . Functional support for targeted secretion was obtained from isolated beta cells, which showed polarised calcium responses , and, using microelectrode-detection, polarised insulin secretion . However, another study, again in single cells, showed granule exocytosis randomly across the cell membrane . These functional experiments used single cultured cells, which does affect secretory control  and removes crucial cell-to-cell contacts found in the islet. One study, in which islets were imaged, suggested that insulin secretion occurs away from the vasculature . But this study was limited to single-plane imaging and did not determine the complex three-dimensional (3D) relationships of the beta cells to the vasculature . Resolution of this issue requires live-cell imaging within intact islets and 3D imaging, something not yet achieved .
If insulin secretion was targeted it would require localisation of proteins to direct the trafficking and fusion of secretory granules. In support of this concept, proteins that organise the targeting of presynaptic neurotransmitter release have been found in beta cells [8, 9, 10, 11, 12]. However, only one study, on the scaffold protein ELKS , immunolocalised the proteins in islets and showed that ELKS appears to be enriched at the vascular face .
Why could targeted insulin secretion be important? By analogy with another endocrine cell, the chromaffin cell, targeted secretion could be profoundly important in understanding stimulus–secretion coupling. Chromaffin cells in situ show exocytic responses to single action potentials  but cultured chromaffin cells need much higher levels of stimulation to elicit secretion . This suggests that in situ the sites of calcium entry are close to sites of exocytosis; an arrangement that is lost in isolated cells. The in vitro data suggest that the cells integrate inputs whereas the in situ data prove the capacity to respond to single inputs. With reference to beta cells the focal targeting of secretion could have similar consequences. While it appears that some targeting of secretion may be maintained in beta cells in culture , studies of models of disease, such as glucose  and lipid toxicity , suggest that there is separation of the calcium entry channels away from sites of exocytosis. Proof of targeted secretion in beta cells could therefore have widespread implications for normal and diseased beta cell behaviour.
Here, we show that in intact islets, insulin granule fusion is non-randomly distributed across the beta cell membrane and is targeted towards the vasculature. We further demonstrate that proteins associated with the cytomatrix of the active zone in neuronal synapses are found in beta cells and are enriched at the vascular face.
We conclude that beta cells possess an ‘endocrine synapse’ that targets secretion of insulin towards the vasculature.
Experiments were performed in an extracellular solution (in mmol/l: 140 NaCl, 5 KCl, 1 MgCl2, 2.5 CaCl2, 5 NaHCO3, 5 HEPES, glucose) adjusted to pH 7.4 with NaOH.
CD1 mice (local colony) were humanely killed according to local, University of Queensland, animal ethics procedures (approved by the University of Queensland, Anatomical Biosciences Ethics Committee).
A range of primary antibodies were used. See ESM Methods for details.
Standard western blot methods were employed. See ESM Methods for details.
Isolated mouse pancreatic tissue was prepared by a collagenase (type IV) (Worthington, Lakewood, NJ, USA) digestion method in Hanks’ buffer (Sigma-Aldrich, Castle Hill, NSW, Australia) adjusted to pH 7.4 with NaOH. Isolated islets were maintained (37°C, 95/5% air/CO2) in RPMI-1640 culture medium (Sigma-Aldrich) containing 10.7 mmol/l glucose, supplemented with 10% FBS (Gibco, Life Technologies, Mulgrave, VIC, Australia), 100 U/ml penicillin and 0.1 mg/ml streptomycin (Invitrogen, Life Technologies, Mulgrave, VIC, Australia).
Isolated islets of Langerhans were cultured for 2–3 days and, before imaging, were bathed in an extracellular solution containing 3 mmol/l glucose for 30 min (37°C, 95/5% air/CO2). Two-photon imaging was performed at 34°C with exocytic events recorded as the entry of extracellular dye into each fused granule.
A custom-made microscope was employed. See ESM Methods for details.
Data are presented as mean ± SEM. Statistical analysis was performed using Microsoft Excel 2010 (Microsoft, Redmond, WA, USA) and GraphPad Prism (ver 6, GraphPad Software Inc, La Jolla, CA, USA). Data were subjected to a Student’s t test. Significance is indicated as *p < 0.05, **p < 0.01 or ***p < 0.001. Islets from at least three mice were used in each experiment.
Live-cell imaging of single insulin granule fusion in islets
This method has been used previously [6, 20] and, in principle, provides a non-specific label for vesicular dye uptake into cells. However, our published evidence  indicates that under our conditions we are likely to be almost exclusively recording insulin granule fusion in beta cells. For example, we record three cell layers into the islet where almost all the cells in a rodent islet are beta cells, the size of the vesicle events we observe is consistent with insulin granule size, the numbers and time course of fusion events are consistent with the amount of insulin secretion, the size of the cells are consistent with beta cells and we record from cells that post-experiment immunostain for insulin .
Glucose-induced exocytosis of insulin granules is non-uniform
Increasing extracellular glucose concentration from 3 to 15 mmol/l induced many insulin granule fusion events. Each fusion event taking place over a 20 min period was identified (using the fluorescence signatures as in Fig. 1a) and its spatial position was located (Fig. 1b). Our images show that the number of exocytic events is unevenly distributed from cell to cell. Simple geometry (assuming spherical cells) shows that the area of cell membrane, sampled across the two-photon imaging slice (∼1–2 μm deep), is the same wherever in the cell the cross section is made . This means the large cell-to-cell differences in the number of fusion events cannot be explained by differences in the sampled cell membrane areas.
A χ 2 analysis showed that the observed distribution of exocytic events in each cell was significantly different (p < 0.01) from a random, Poisson distribution of events (Fig. 1c; 18 islets, six mice, 360 cells responses to 8 and 15 mmol/l glucose).
To further test for an uneven distribution of exocytosis we determined the number of fusion events in one two-photon plane to the number, seen in the same cell, in the other plane (Fig. 2b) and then expressed these numbers as a ratio. An even distribution would predict a 1:1 ratio; in contrast, we measured a 1:6.6 ratio (n = 65 cells, 199 exocytic events, paired Student’s t test p < 0.01). However, exocytic activity changes over time  and differences could be due to capturing a burst of activity in one plane and not in the other. Against this idea, by recording each plane for >6 min, which is longer than a single burst , our data will temporally average any burst responses. Furthermore, the actual data does not show a systematic bias; 33 cells had the same response, 19 cells had more events in the second plane and 13 had more events in the first plane. We conclude that exocytosis is unevenly distributed around a beta cell.
Insulin secretion is targeted to the vasculature
To determine whether these fissures are vascular we counter-stained live cells with isolectin B4, which reacts with α-d-galactopyranosyl groups  in the basement membrane [23, 24]. This stained the fissures (ESM Fig. 1) and, using immunostaining, overlays with laminin (ESM Fig. 1).
The above analysis supports the idea of targeted secretion but further evidence requires 3D real-time acquisition of exocytic events over the entire volume of single cells. This is technically challenging, given the short lifetime and small size of each exocytic event. To facilitate our experiments we employed high-K+ stimulation, in isolated islets, to rapidly induce a large number of exocytic responses and captured images over 9 min of continual cycling of sequential Z planes (2 μm steps, seven steps per stack, 3 s per stack) through a single cell. Isolectin B4 stained the basement membrane and identified the cell’s orientation with respect to the vasculature. The time and location of each individual exocytic event was identified and plotted out in a 3D representation of the cell (Fig. 3b).
Finally, for each cell (n = 4 islets, four cells), we trigonometrically measured the distance between each insulin granule fusion site and the location of the nearest point on the vasculature. The histogram shown in Fig. 3d demonstrates a strong bias for fusion events to be targeted to the vascular pole of the beta cells.
Polarisation of beta cells towards the vasculature
We conclude that most pancreatic beta cells respond to glucose. The apparent heterogeneity in cellular responses in (e.g. Fig. 1) is due to an uneven spatial spread of granule fusion events across the cell, with preferential targeting towards the vasculature. Drawing on work in neurons, we next tested for the mechanisms that might support this targeting. Previous work has used single sections to immunolocalise proteins in the stimulus–secretion cascade . However, to understand the 3D relationship within the islet we have employed a method of ‘thick’ sectioning (∼100 μm) of islet tissue and serial confocal imaging . All immunostaining work was conducted on these islet slices.
We show that in intact islets insulin granule fusion is targeted towards the vascular face of beta cells. At the same vascular face we show enrichment of proteins known to be components of the presynaptic region of neurons. Together these findings provide functional and structural evidence that insulin secretion is dependent on an ‘endocrine synapse’ that targets hormonal outflow to the bloodstream.
Functional assessment of targeted insulin secretion
Our work is the first functional assessment of the spatial distribution of insulin granule exocytosis in three dimensions within intact islets. Our conclusion of targeting is consistent with electron microscopy results showing that granules were evenly distributed in resting beta cells (ruling out a simple explanation for polarised secretion) but were asymmetrically clustered in chronically stimulated beta cells . More recent experiments measuring signalling or exocytosis in isolated beta cells either support [2, 3] or go against  the idea of polarised secretion. However, results from single, cultured cells reflect the culture conditions  and are remote from the native environment of the islet.
The work that resembles ours most closely used intact islets and two-photon microscopy, but only took images in one plane . The authors state that ‘most exocytotic events […] occurred in the “abvascular” compartment’ (see their Fig. 2d) (i.e. away from the blood vessels) and this is how their work has been interpreted by others . However, they also state that they ‘detected a polarity of β cells […]: exocytosis occurred preferentially in the vicinity of certain vessels, possibly veins’. These two statements appear contradictory. These experiments were a minor part of this paper and we contest that the use of a single image plane distorts image interpretation. For example, in the most extreme case, an exocytic event could appear to be a beta cell diameter (16 μm ) away from a blood vessel in a single plane but, in our 3D experiments, a blood vessel identified immediately above, or below, in the next image plane, would place the exocytic event only 2 μm away (the depth of our Z plane). 3D analysis is therefore essential to map these spatial relationships and reach conclusions about targeting of exocytosis.
We have yet to consistently record exocytosis in tissue slices and therefore performed the functional experiments on cultured islets. It is known that the number of endothelial cells decreases over time in culture [24, 30] and consistent with this we observe a gradual decrease, but not loss, of isolectin B4 staining over days. We conclude that islets remodel in culture but that laminin is still present along the remaining blood vessels and the beta cells adjoining these regions still maintain targeted secretion.
A further point of discussion is our use of high potassium in the 3D measurements. The relationship between high potassium and glucose-induced responses is not clear [8, 31]. We expect that both would be spatially localised and require the same scaffold proteins, although there may be differences in the contributions of specific SNAREs . However, it is possible that there are spatial differences and, furthermore, it is possible that modulators, such as GLP-1, could control different spatial patterns of exocytosis through influences on proteins like RIM2 . Future experiments will be required to investigate these interesting possibilities.
Molecular mechanisms of the endocrine synapse
There are a few studies identifying proteins in beta cells that are enriched presynaptically in neurons and form a presynaptic stimulus–secretion complex. These include ELKS , RIM2 , Munc 13-1 , SNAP25  and neurexin . However, only ELKS has previously been localised within an islet.
Part of this protein complex is likely to include calcium channels and single-cell work shows calcium entry is polarised . We were unsuccessful in immunolocalising calcium channels but the spatial association of calcium channels is important in secretory control  and is a possible factor in disease .
Since the differential distribution of SNAREs is the dominant hypothesis to explain polarised targeting of secretion in the pancreatic acinar cells [36, 37], it is interesting that syntaxin 1A and SNAP25 are relatively uniformly distributed around beta cells (Fig. 5). However, the specific localisation of just one other crucial component could dictate the targeting of secretion. Our work points to liprin, RIM2 or piccolo as possible candidates for this essential component. An alternative explanation is that syntaxin isoforms other than 1A are regulators of insulin secretion [32, 38, 39].
Polarisation of beta cells
Location of these presynaptic proteins to the vascular side of beta cells suggests that the cells are using orientation cues, one of which may be interaction with endothelial cells. During islet development vascular endothelial growth factor A secreted from beta cells is key to endothelial cell location and vascular development [23, 40]. In turn, the basement membrane protein laminin regulates beta cell functions such as insulin gene expression and proliferation . Interestingly, in islet transplant studies, revascularisation of the donor islets is maximal at 14 days which temporally correlates with a return to normoglycaemia . Factors such as hypoxia may be important, but this recovery of functionality would be consistent with the formation of new beta cell–endothelial cell contacts, something directly shown in electron microscopy studies .
Extending the analogy with neuronal synapses, we speculate that the beta cell–endothelial cell interactions may use similar recognition mechanisms to a synapse. One candidate is the neurexin–neuroligin pathway in which protein–protein interactions span the presynaptic to postsynaptic cells . Neurexins and neuroligins are found in beta cells  and endothelial cells  although their in situ distribution has yet to be determined.
Consequences of secretory targeting for islet function
Our work has implications for the intra-islet effect of insulin. Insulin has feedback actions on beta cells ; secretion into the bloodstream makes it unlikely that cells would sense their own secretory output. Instead, beta cells might sense circulating insulin, an argument recently proposed on the basis of computer modelling . In human islets beta cells are scattered throughout the islet (not just in the core as in rodents)  but, if vascular targeting of insulin secretion occurs in humans, this would still affect its paracrine/autocrine actions.
Given the importance of spatial control of secretory output, there might be further mechanisms of control that selectively target secretion to the lateral, beta cell-to-beta cell, interface. As well as insulin this could also include the secretion of γ-aminobutyric acid, ATP and zinc, which have known actions within the islet [46, 47, 48]. Since disease is associated with changes in beta cell structure  and remodelling of the vasculature , any mistargeting of secretion could be a component in disease.
We conclude that our work provides evidence for targeted insulin secretion, which has wide-ranging implications for the control of insulin secretion and for the role of the beta cell within islets. As such its preservation will be important for islet transplant therapy  and its establishment important in therapies designed around genetically engineered cells.
We thank S. Blomberg (School of Biological Sciences, University of Queensland, Australia) for help with statistical analysis and D. Whitehead (School of Biomedical Sciences, University of Queensland, Australia) for help with tissue sectioning.
This work was supported by an Australian Research Council Grant DP110100642 (to PT) and National Health and Medical Research Council Grants APP1002520 and APP1059426 (to PT and HYG).
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.
All authors made substantial contributions to the conception and design of the experiments. JTL, JMM, MZ and OHD performed research, CJS developed methods and HYG and PT designed the research. All authors were involved in drafting the article and all approved the final version of the manuscript. PT is responsible for the integrity of the work as a whole.
- 12.Ohara-Imaizumi M, Ohtsuka T, Matsushima S et al (2005) ELKS, a protein structurally related to the active zone associated protein cast, is expressed in pancreatic beta cells and functions in insulin exocytosis: interaction of ELKS with exocytotic machinery analyzed by total internal reflection fluorescence microscopy. Mol Biol Cell 16:3289–3300PubMedCentralPubMedCrossRefGoogle Scholar
- 24.Lukinius A, Jansson L, Korsgren O (1995) Ultrastructural evidence for blood microvessels devoid of an endothelial-cell lining in transplanted pancreatic-islets. Am J Physiol 146:429–435Google Scholar
- 27.Takahashi N, Hatakeyama H, Kasai H (2004) Sequential insulin exocytosis and redistribution of snap25 analyzed with two-photon imaging. Diabetologia 47:205Google Scholar
- 34.Vikman J, Svensson H, Huang YC et al (2009) Truncation of SNAP-25 reduces the stimulatory action of cAMP on rapid exocytosis in insulin-secreting cells. Am J Physiol 297:E452–E461Google Scholar
- 37.Behrendorff N, Dolai S, Hong WJ, Gaisano HY, Thorn P (2011) Vesicle-associated membrane protein 8 (VAMP8) is a SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) selectively required for sequential granule-to-granule fusion. J Biol Chem 286:29627–29634PubMedCentralPubMedCrossRefGoogle Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.