Synapsins I and II Are Not Required for β-Cell Insulin Secretion: Granules Must Pool Their Own Weight

Because insulin secretion is defective in type 2 diabetes (1–3), it has been a priority to elucidate the molecular basis of insulin granule exocytosis in normal as well as diabetic β-cells (4–6). From the studies of Grodsky (7), it has been known for some time that insulin secretion from β-cells is biphasic. The fast phase is due to the release of a small number of docked insulin granules [the so-called readily releasable pool (RRP); probably 100 granules or less per cell]. This is followed by a more sustained phase, believed to represent mobilization from a larger, less mobile pool of stored granules [e.g. 10,000 or so per cell (4); the reserve pool (RP)]. It has long been known that a striking feature of type 2 diabetes is the selective decrease in the magnitude of the early fast phase of secretion (3).

To facilitate our understanding of β-cell insulin secretion, it is instructive to examine the more well understood secretion of chemical transmitters from neurons, which exocytose neurotransmitter in a quantal fashion after a rise in intracellular Ca²⁺ concentration or other endocrine cells such as adrenal chromaffin cells. Indeed, comparative studies of these different secretory cells reveal similarities and differences in the mechanisms involved (8) and have identified several key proteins that regulate neurohormone secretion.

One of the earliest synaptic proteins described was synapsin I, an 88-kDa vesicle-specific phosphoprotein first isolated and characterized in 1977 by Tetsufumi Ueda, working in Paul Greengard's laboratory (9). Synapsin I is a highly basic, elongated protein and is phosphorylated at various sites by a number of kinases, classically protein kinase A (9) and calcium/calmodulin-dependent protein kinase (CaMK) (10), but also ERK (11). In neurons, it was hypothesized that synapsin I/II, which colocalizes to the surface of synaptic vesicles (12), interacts with the cytoskeleton (especially elements composed of F-actin), and this interaction tethers vesicles in a storage pool away from presynaptic release sites (13). In this model, Ca ions entering the nerve terminal during action potential-dependent depolarization activate CaMKII, which sequentially activates and phosphorylates synapsin I. The phosphorylation of synapsin I by CaMKII triggers the untethering of vesicles from the cytoskeleton. Vesicles are then free to move to the release sites, facilitating transmitter release. More recently, it has been proposed that synapsins may also play a regulatory role in vesicle docking and recycling through endocytosis (14).

To determine whether a similar model applies to pancreatic β-cells that release insulin contained in large dense-core vesicles (LDCV), Eishichi Miyamoto's lab (15) successfully cloned synapsin I and II from the insulin-secreting MIN-6 cell line and rat islets. Indeed, glucose and other insulin secretagogues triggered CaMKII-dependent synapsin I phosphorylation, in parallel with an increase in insulin secretion (16, 17). Although Miyamoto's data (18) also supported synapsin I/II colocalization with insulin granules, Easom and colleagues (19) could not demonstrate this and suggested, instead, that synapsin might be colocalized with the small, transmitter-containing vesicles also present in β-cells.

Further cracks in the theory that the phosphorylation of synapsin I/II is involved in vesicular secretion began to appear, even in the case of neurons, with the demonstration that neurotransmitter release was normal despite very low levels of synapsin I in Drosophila motor neurons. In addition, F-actin is not observed in the vicinity of the synaptic vesicles (20), even though release is apparently normal. In a mouse model where both synapsin I and synapsin II were deleted globally, no major redistribution of vesicles was noted, and in particular, synaptic vesicles did not appear to be more dispersed in relation to the active zones. A reduction in the total number of vesicles was seen, but there was no change in general synaptic structure. Rosahl et al. (21) did observe changes in neurotransmitter release in double knockout (DKO) neurons, but these were most prominent during repeated stimulation (e.g. using protocols that elicit posttetanic potentiation or facilitation). These authors concluded that the lack of synapsin I/II did not alter the basic mechanisms of synaptic vesicle trafficking or neurotransmitter release evoked by low rates of stimulation.

Into this background comes the paper in this issue by Wendt et al. (22) that examines β-cell insulin secretion again from a mouse model lacking synapsins I and II (DKO). The authors show that synapsins I and II are the major synapsin isoforms of NMRI and C57BL/6 mouse strains using quantitative PCR, with synapsin III being a minor component (Fig. 1, A and B, in Ref. 22). One potentially important quibble is that the authors failed to reexamine the relationship in the DKO animals, leaving open the possibility, however remote, that up-regulation of synapsin III could compensate in the knockouts.

In terms of the localization of synapsin I/II, Wendt et al. (22) also show with immunohistochemistry that synapsin I/II is present in the β-cell where it appears to largely but not completely colocalize to insulin granules.

Using a standard 60-min static incubation protocol to examine islet insulin, glucagon, and somatostatin secretion in wild-type vs. DKO islets (Fig. 2 in Ref. 22), they show that the presence or absence of synapsins I/II is without affect on glucose-dependent secretion of any of the three hormones. Interestingly, they did observe a small but significant increase in the secretion evoked by 1 mm glucose combined with 70 mm KCl, but this was not seen for the other islet hormones assayed or when glucose alone was the stimulus. Although the authors speculated as to the cause of this discrepancy, the modest size of the effect suggests it could possibly be a sampling error. Consistent with finding no marked derangement in insulin secretion, the DKO mice were euglycemic rather than glucose intolerant or diabetic.

The authors extended their analysis of insulin secretion in the DKO by examining exocytosis directly using single β-cells from DKO mouse islets. These experiments were carried out using depolarization-induced capacitance measurements (4–6). In brief, granule release is monitored electrically by detecting incremental increases in membrane capacitance in real time that occur when granules fuse to the β-cell plasma membrane, increasing its surface area (electrically, increasing membrane area after granule fusion increases the total capacitance of the cell) (4). β-Cells were subjected to a train of five 100-msec depolarizations followed by nine 500-msec depolarizations. From these measurements, estimates of the RRP (the sum of first five pulses) and the RP (the sum of next eight pulses) were obtained and were found not to differ between wild-type and DKO β-cells (Fig. 3 in Ref. 22). Voltage-gated Ca²⁺ current also was indistinguishable in both wild-type and DKO β-cells (Fig. 4 in Ref. 22).

Although β-cells are widely believed to release their granules at sites on the plasma membrane where SNARES (Soluble NSF Attachment Protein Receptor) are concentrated (23) along with Ca²⁺ channels (24), β-cells lack the clearly demarcated clusters of vesicles that are clearly visible in electron microscopic images of neuronal presynaptic terminals (25). Instead, it has been argued that morphometric analysis of LDCV profiles within close proximity to the plasma membrane can be used to estimate the number of docked insulin granules. No clear differences were noted as a result of the loss of synapsins I/II.

Thus, this straightforward study offers the most direct functional evidence to date that synapsins I/II are not likely to be crucially important for glucose-dependent insulin secretion from mouse β-cells.

Two major questions are left unanswered by the study. First, if synapsins I and II are not involved in the mobilization of granules from the large reserve pool, how is mobilization able to replenish the docked pool and mediate the second phase of insulin secretion? Of course, it is being assumed that there is indeed a physically remote pool of LDCV in β-cells that is analogous to the vesicle storage pool of neurons. However, the two types of cells might function very differently and use different molecular mechanisms to control their exocytotic processes. For instance, recent evidence from Thurmond and colleagues (26) suggests that a meshwork of cortical actin resides under the plasma membrane of β-cells and acts as a barrier to the movement of granules to their release sites. Their data suggest that the integrity of this barrier to granule movement is dynamic and can be regulated by glucose metabolism and the activity of small G proteins. If this is indeed the case, it may be that the actin barrier controls the replenishment of the RRP from a larger reserve pool of DCV. Thus, the two pools may be functionally separable, but not because the RP is tethered to the cytoskeleton at a remote site by the synapsins.

The second open question, of course, is what are synapsin I and II doing in the β-cell? It would seem premature to completely rule out a secretory role for synapsin proteins at this point, given that the synapsins may mediate more subtle jobs (e.g. regulating the detailed kinetics of granule release or oscillations in insulin secretion) that have not yet been examined. Thus, future studies will be required to more fully understand these interesting, but enigmatic proteins.