Association of SNAREs and Calcium Channels with the Borders of Cytoskeletal Cages Organizes the Secretory Machinery in Chromaffin Cells

t-SNAREs form Plasma Membrane Microdomains in Chromaffin Cells

The initial characterization of SNARE distribution in chromaffin cells using epifluorescence showed annular patters characteristic of a plasma membrane distribution (Roth and Burgoyne 1994; Grant et al. 1999). Further studies using confocal microscopy revealed the uneven distribution of these proteins forming membrane patches (Gil et al. 2001). More recently, it has been demonstrated that these microdomains of SNARES are related with the formation of syntaxin 1/SNAP-25 dimers (Rickman et al. 2004; Lopez et al. 2007). Furthermore, the expression of exogenous SNAREs leads to the formation of these dimers that has been observed with improved resolution under evanescent field fluorescence microscopy (An and Almers 2004; Lopez et al. 2009). The formation of clusters associated with cholesterol-rich regions seems to be the preferential organization of SNARE microdomains as it has been reported in other cellular systems such as PC12 pheochromocytoma cells (Lang et al. 2001; Chamberlain et al. 2001) and β-cell lines (Ohara-Imaizumi et al. 2004). These microdomains may contain hundreds of molecules as their size reported with confocal microscopy results in diameters around 400–600 nm (Lang et al. 2001; Rickman et al. 2004; An and Almers 2004; Lopez et al. 2007), larger than the average chromaffin granule. We have to consider that due to the limited resolution of confocal microscopy these values may be overestimated, and recent calculations using stimulation depletion (STED) microscopy, yield more realistic values of 60–70 nm diameter containing 70 densely packed molecules for SNARE homo-clusters formed by syntaxins (Sieber et al. 2007). The organization of SNARE in clusters may persist during the secretory cycle as syntaxin-1, SNAP-25 and synaptobrevin colocalize with dopamine-β-hydroxylase, a granule membrane marker incorporated to the plasma membrane during exocytosis (Lopez et al. 2007).

SNARE Clusters are Dynamic

The use of evanescent field microscopy recently revealed that the clusters formed by GFP-SNAP-25 (GFP linked to the C-terminus of the synaptosome-associated protein of 25 kDa) present a spatially limited mobility in the proximity of the plasmalemma (Lopez et al. 2009). This motion is characterized by random movement with a 22 nm/s averaged speed, corresponding to relatively slow moving patches displacing one microdomain diameter in 30 s if linear displacement is supposed. Due to the continuous reversal of direction these clusters rarely move more that one cluster diameter in 1 min period. Furthermore, SNARE microdomains present relatively large displacement in the plane perpendicular to the membrane (z-plane), covering 100 nm distances in periods of a few seconds. The expression of GFP-SNAP-25 Δ9 form, the product of botulinum neurotoxin A action (Schiavo et al. 1993; Blasi et al. 1993), forms morphologically similar membrane patches but with increased motion in both the XY- and Z-plane in chromaffin cells (Lopez et al. 2009), therefore the motion of SNARE clusters seems to depend on interactions inside the core complex. The magnitude of this motion is much higher than the expected from the X-ray structure describing a 12-nm bundle (Sutton et al. 1998). It is possible that the interactions among the dozens or hundreds of SNARE complexes forming part of the SNARE microdomains play a fundamental role in this motion in addition to the short range interactions inside individual complexes. Alternatively, SNARE cluster motion could be dependent of interactions with molecular motors like myosins linked to the cortical F-actin cytoskeleton.

Calcium Channel Clusters Localize in the Proximity of SNARE Microdomains

Calcium channels have been show to interact directly with SNAREs in neurons (Catterall 1999), and in chromaffin cells the secretory activity is associated with calcium ‘hot spots’ (Robinson et al. 1995), therefore the proximity of calcium channels to the secretory machinery seems to be implicit. In fact, an analysis of the distribution of SNARE microdomains in relation with the clusters formed L- and P/Q type voltage-dependent calcium channels proved a partial degree of colocalization (Lopez et al. 2007). In this study, it was shown that in cultured chromaffin cells, one of every three SNARE microdomains presents calcium channels cluster colocalization whereas in the majority of cases channel clusters located at a distance shorter than 300 nm. The coupling of calcium channel clusters to SNARE microdomains is tighter in the cells forming part of the adrenal medulla (Lopez et al. 2007). The different proximity of calcium channel cluster to the secretory machinery between the intact tissue an the isolated cells seems to explain the faster coupling of calcium signal to secretory activity in adrenal slices when compared to the slower secretory kinetics found in cultured cells (Moser and Neher 1997).

Organization of the Secretory Machinery in Association with the Cortical Cytoskeleton

As mentioned above the secretory machinery is condensed in clusters that may be distributed through the chromaffin cell plasma membrane with calcium channels ensuring relatively rapid elevation of intracellular calcium in the proximity of releasing sites. But how does this distribution related with the organization of the cortical region in these neuroendocrine cells? Our studies show that the cortical region is structured as an intricate network of F-actin polygonal cages (Giner et al. 2005, 2007; Villanueva et al. 2010). This cortical network re-organizes during cell stimulation changing from an intricate structure imprisoning the vesicles in resting conditions to provide open spaces allowing vesicle access to docking sites during secretion (Giner et al. 2005). During these cortical re-organizations vesicles moved on the surface of the dynamic F-actin-myosin II structures (Giner et al. 2005). Interestingly, we have recently demonstrated that the SNARE clusters formed by exogenous SNAP-25 also associated with the limits of the polygonal F-actin structures in contact with the chromaffin cell plasma membrane (borders of the cytoskeletal cages, Villanueva et al. 2010), therefore these structures seem to organize the secretory machinery in a dynamic environment. The importance of cytoskeletal cage borders is stressed when the site of granule exocytosis is studied in relation with these structures (Villanueva et al. 2010). In effect, simultaneous visualization of vesicle fusion using acridine-orange-loaded granules and cytoskeletal organization using transmitted light reveals that exocytosis occurs in these borders. This is consistent with the precise colocalization of secretory places with the microdomains formed by syntaxin-1/SNAP-25 dimers previously described (Lopez et al. 2007). Taking together these experimental evidences suggest that secretory sites in chromaffin cells are organized in SNARE microdomains co-localizing with calcium channel clusters and placed in the borders of cytoskeletal F-actin polygonal structures that could be limiting calcium diffusion in a relatively closed environment (see model in Fig. 1a).

A Model for the Organization of the Secretory Machinery

What will be the benefit of organizing the molecular components of the exocytotic machinery in the borders of cortical cytoskeletal structures? To approach this, we have developed a model based in Monte Carlo simulations to describe calcium distributions and the secretory activity in the cortical area (Gil et al. 2000). Monte Carlo methods (particle-based methods) are suitable methods when complex models in non-uniform geometries are needed (Gil et al. 2000). The description is microscopic: the movement of individual ions is performed by using random-walk algorithms on a grid and the rest of processes involved in the secretory response are also described probabilistically; for instance, the binding of ions with buffers and vesicle sensors are described through the local probabilities of binding and unbinding per unit of time. In this model, calcium currents through L and P/Q-type calcium channels and computed secretory events in response to short depolarizations are simulated in two different configurations of the secretory machinery (Villanueva et al. 2010); a random spatial distribution of these elements, and their association with the borders of a simulated cylindrical F-actin cage of 0.6 μm in diameter according with our experimental data (Giner et al. 2007) (Fig. 1b). In both configurations we propose that the modelled cage contains one secretory site (SNARE cluster) colocalized with a calcium channel microdomain whereas other two channel clusters are located at a 300 nm distance of the secretory site, thus maintaining the proportion of calcium channel clusters to SNARE microdomains described in a previous work (Lopez et al. 2007). The results of this study allow the quantization of local calcium levels as well as averaged levels in the submembrane domain from several randomly generated sets in both configurations. These results indicate that local levels of calcium in the vicinity of the secretory sensor are higher for the configuration presenting border localization of the secretory machinery whereas the overall averaged calcium levels are similar. In consequence the time course of simulated secretion is faster in a model where calcium channels and SNAREs are localized in the border of a simulated cage when compared with a random distributions of these elements in the interior of the cages (Villanueva et al. 2010; Fig. 1c), indicating a more efficient coupling of the elements constituting the secretory machinery.

The location of the secretory machinery in the borders of F-actin cages predicts the interaction between cytoskeletal elements and granules at the time of the exocytotic fusion, this is consistent with the reported effect of expressing altered forms of myosin II over the single vesicle kinetics and the rate of expansion of the fusion pore (Ñeco et al. 2004, 2008). Alternatively, this type of interactions will make possible to change the fusion mode from full to ‘kiss and run’ under the control of cytoskeletal elements as demonstrated by Doreian et al. (2008).

Therefore, the experimental results and the conclusions obtained in this theoretical study indicate that the organization of the secretory machinery around a particular cytoarchitecture is an important factor to understand how to shape the secretory kinetics in neuroendocrine cellular models.