It’s What’s Inside that Matters

The histological identification of intracellular secretory granules was first made in the exocrine acinus of the pancreas, with the coincident disappearance of granules with the appearance of digestive enzymes in the pancreatic juice (1). However, the adrenal medullary chromaffin has been instrumental in elucidating concomitant cellular secretion of catecholamine and other granular contents without loss of lipids or cholesterol. Secretory granules originate from post-Golgi vesicles that coalesce and fuse with excess membrane removal during a condensation stage where granules mature to objects with increased osmium staining (2) that are actually heavier than postsecretory granular membranes (3). Presumably, the condensation of granule contents must be accompanied by some reduction in solute activity (to preserve osmotic pressure), most easily imagined as the binding of content solutes to form larger complexes. Now, in an innovative pair of articles (4,5), Weiss et al. show that in the case of secretory granules in chromaffin cells (chromaffin granules), the following applies:

• 1.The interior of a dense granule is not crystalline, but fluid;
• 2.Different proteins can have different diffusion constants within an individual granule; and
• 3.The expression of a single protein within the granule can inhibit exocytotic fusion pore expansion, the structure through which protein is released.

Searching for a mechanism to explain why tissue plasminogen activator (tPA) leaves exocytotic chromaffin granules later than other cargo components (6,7), the authors confirmed that fluorescently tagged tPA-cer is discharged very slowly (seconds) in comparison with fluorescently tagged neuropeptide Y-cer (milliseconds) and then devised new FRAP techniques to measure intragranular protein diffusion. They find the mobility of tPA-cer is significantly reduced (1/3000th that of a similar size protein in aqueous solution), compared to NPY-cer. However, the slower diffusion could not alone explain the extremely slow discharge of the protein after fusion, suggesting either inhibition of fusion pore dilatation or a unique viscosity within the fusion pore interior. A kinetic retardation to tPA-cer passage through the fusion pore could result from multiple collisions and interactions with the membrane on the way through a narrow fusion pore, especially if the tPA itself bound to membranes. The authors suggest an ∼14-nm diameter pore (minimum size), stable over many seconds, would be required to retard tPA with a maximal dimension of 13 nm. To directly investigate their predicted slow pore dilatation and curvature changes at the fusion site, the authors used amperometry to follow catecholamine release and polarization total internal-reflection fluorescence microscopy. They determined that the granular remnant membrane remains curved for a long time in tPA-secreting vesicles. Surprisingly, in many cases, the bulk of the tPA can leave the granule through the fusion pore before the vesicle flattens out.

To our knowledge, these new results add to a body of work showing that rapid, all-or-nothing content secretion, postulated by quantal release, is quite complex, depending upon the secretory system. Whereas the quantal hypothesis predicted the release of the entire content, the time course and dynamics of the release process itself were not uniquely specified and were often defined by the temporal resolution of the experiment. In other single-vesicle studies, differential release of membrane proteins from exocytotic vesicles has been documented; e.g., VAMP and IRAP leaving early, compared to Dopamine-β-Hydroxylase and GLUT4 remaining behind, respectively (8,9).

However, it has been impossible to tell if this differential release is due to trapping by the fusion pore or other factors, such as raft association with specific lipids at the fusion site donated by the secretory vesicle, binding to intracellular proteins, or lack of binding of membrane proteins to the granule interior, etc. Indeed, a dense core can be seen protruding from an exocytotic remnant in electron micrographs, clearly indicating that a fraction of a core can remain morphologically intact after exocytosis (Fig. 1 A). Equally problematic has been an ultrastructural definition of the neck, or fusion pore, first discovered by Palade and Bruns (10) in endothelial cell exocytosis: this structure appears to be too wide to restrict content release.

Figure 1.

(A) Release of granule (arrow) from a neurosecretory axon in the corpus cardiacum of 1-day-old meat fly. It is observed that the content of the granule from the upper axon is liberated into the intercellular space. (B) Extrusion of neurosecretory granules in 1-day-old meat fly; Os04, veronal. Two granules have been discharged and are now found in the intercellular space (arrows). The left one has been extruded into the synaptic cleft. (C) Discharge of neurosecretory granule directly into the hemocoele (arrow). Here the granule content has dissolved. From Normann (15).

The fusion pore or neck diameter, determined by electron microscopy, varies between 10 and 50 nm in endothelial cells, and can be as small as 10 nm in amoebocyte-dense core vesicle fusion (11). However, such small pores are exceptions rather than the rule. Indeed, the observed scarcity of small fusion pores itself suggests that pore widening is a rapid process compared to the lifetimes of more readily detected, ∼150-nm fusion pore diameters, which are representative of a significant fraction of the fixed fusion events (11). Because electron microscopy often fails to capture events with lifetimes normally as short as fusion pore expansion, the tPA-stabilized fusion pore may be a good one to look for, because a 14-nm pore lasting for 10 s is much larger and longer than the amoebocyte granule fusion pore.

These results suggest that the pore dilatation is specifically controlled by tPA, possibly to regulate cell surface tPA for its physiological role in an autocrine/paracrine negative feedback loop to control nicotinic cholinergic stimulation in the adrenal medulla. There are many different factors that influence the dynamics of the fusion pore and subsequent content release. However, this novel finding that a cargo protein inhibits fusion pore expansion is intriguing and the mechanism unclear. Either the cells expressing tPA are different in their membrane properties, or the tPA itself is causing a change in the membrane properties at the fusion pore itself. It is unlikely that previously released endogenous tPA is acting from the extracellular leaflet because immunostaining reveals very little tPA at the surface other than at puncta (not unlike the released but undissolved granules Fig. 1 A). Several mechanisms of tPA action on fusion pore dilatation are plausible (Fig. 2).

Figure 2.

Schema of the cargo-controlled fusion pore widening. (A) Neuropeptide Y (red dots) does not affect the pore enlargement. (B and C) Hypothetical mechanisms of tPA (green dot) controlled fusion pore widening. (B) tPA regulates membrane curvature of the fusion pore by itself or through an interaction with other protein. (C) tPA sorting influences the lipid composition of the granule membrane that affects the pore expansion. To see this figure in color, go online.

Starting with physical driving forces, fusion pore expansion can be driven either by membrane curvature or surface tension (12). Most membrane-adherent proteins that affect monolayer curvature act to promote positive monolayer curvature. Positive curvature agents will lessen the driving force for fusion pore widening by stabilizing the bent fusion pore. More work is needed to determine if tPA itself can cause monolayer spontaneous curvature to change. For example, we can calculate specific predictions for the ratio of the pore width to the length that is stabilized by inner leaflet spontaneous curvature, depending upon the magnitude of the change in curvature, and these predictions can be tested with cryo-electron tomographic imaging. Additionally, tPA membrane binding can also change surface tension that would affect fusion pore dynamics. Cell-stretching techniques can be used to test whether membrane tension drives pore expansion. However, most cells have little surface tension unless perturbed mechanically.

tPA expressions may also alter protein and lipid composition of granule membranes. More specifically, a plausible tPA interaction with membrane may be facilitated through specific phospholipids that may be enriched in the membrane of the granule. Conical and inverted-conical phospholipids are known to affect fusion pore formation.

Although Weiss et al. (4,5) present the first evidence for cargo protein-controlled fusion pore expansion, the fusion pore seems to be also controlled by the content of envelope viruses. Pores of different sizes induced by influenza virus have been detected by cryo-electron tomography (13). Virion content, which includes nucleic acid, nucleic-acid stabilizing proteins, scaffold or capsid and cellular proteins, is often highly organized and densely packed upon virus assembly. Virus infection relies on successful transport of the genome through a fusion pore between the virus and the host cell. Thus, viral fusion pore expansion often follows dramatic physical changes in the virion content during disassembly of the scaffold and release of the genome (14). Scaffold detachment from the membrane increases membrane flexibility necessary for fusion pore enlargement.

This work demonstrates the benefit of Weiss et al.’s development of new quantitative optical approaches to explore fundamental biophysical phenomenon at the cell membrane. In general, the internal contents of a membrane fusing structure (granule, virus, or vesicle) influences fusion pore dynamics and consequently the release of these contents. The content composition can be considered as an additional regulating factor in membrane fusion that includes fusion proteins, lipid composition, membrane channels, and membrane tension.