Real-time imaging of regulated exocytosis in secreting organs can provide unprecedented temporal and spatial detail. Here, we highlight recent advances in 3D time-lapse imaging in Drosophila salivary glands at single-granule resolution. Using fluorescently labeled proteins expressed in the fly, it is now possible to image the dynamics of vesicle biogenesis and the cytoskeletal factors involved in secretion. 3D imaging over time allows one to visualize and define the temporal sequence of events, including clearance of cortical actin, fusion pore formation, mixing of the vesicular and plasma membranes and recruitment of components of the cytoskeleton. We will also discuss the genetic tools available in the fly that allow one to interrogate the essential factors involved in secretory vesicle formation, cargo secretion and the ultimate integration of the vesicular and plasma membranes. We argue that the combination of high-resolution real-time imaging and powerful genetics provides a platform to investigate the role of any factor in regulated secretion.
Exocytosis is the process in which cargo that has been synthesized by specialized cells is delivered to the extracellular environment. This type of secretion typically involves the formation of membranous secretory vesicles within the cell, in which the cargo is packaged and then continually released to the outside (constitutive exocytosis), or is stored until a signal triggers the process of secretion (regulated exocytosis). During secretion, vesicles dock and fuse with the apical plasma membrane (PM) to release their contents (Shitara and Weigert, 2015). This fundamental cellular process occurs across many cell and tissue types and is essential for the release of diverse bioactive molecules that mediate such functions as neurotransmission, reproduction, immunity, development, respiration and digestion (Masedunskas et al., 2012b; Nightingale et al., 2012; Porat-Shliom et al., 2013). Although regulated exocytosis occurs across diverse tissues, the composition of the cargo and morphology of the secretory vesicles can vary dramatically, influencing the biophysics behind secretory vesicle fusion and cargo release. The focus of this Commentary will be on recent developments in three-dimensional (3D) time-lapse imaging of regulated secretion occurring in living secreting organs. This review will highlight recent high-resolution, real-time imaging studies that have revealed the temporal order of events, the localized recruitment of specific factors and unique intermediates in the process of regulated exocytosis, both in vivo and ex vivo. Finally, we will discuss how these techniques can be expanded in the future to address additional mechanistic details of this essential process.
Regulated exocytosis or regulated secretion is an essential process that occurs in specialized secretory cells in response to an extracellular stimulus. Cargo to be delivered to the extracellular environment is packaged into membranous vesicles (secretory granules or vesicles) that fuse with the PM to release their contents. Upon stimulation, fusion can occur between individual secretory granules and the PM or, alternatively, secretory granules can fuse with one another (also known as compound exocytosis) to release their contents (Fig. 1A). Exocytosis of cargo can occur through ‘kiss-and-run’ secretion, in which fused secretory vesicles secrete their contents followed by closure of the fusion pore to reform the vesicle (Wu et al., 2014) (Fig. 1A). Alternatively, after fusion with the PM, secretion of cargo can involve the full collapse of the secretory vesicle, followed by integration of the vesicular membrane with the PM (Fig. 1A). In this instance, dramatic mixing and remodeling of the vesicular membrane and the PM occurs.
The four main types of secretory cells preforming regulated exocytosis are hematopoietic, neuronal, endocrine and exocrine cells (Shitara and Weigert, 2015) (Fig. 1B). Each of these cell types produces and secretes molecules with unique physical and functional properties. In the case of neuronal cells, secretory vesicles containing neurotransmitters are small (∼50 to 100 nm in diameter) and secrete their cargo in a matter of milliseconds to ensure rapid synaptic transmission (Shitara and Weigert, 2015). These secretory vesicles are present along the PM and are protected from premature fusion by an actin coat along the cortex of the cells (Shitara and Weigert, 2015). Upon stimulation, fusion of the secretory vesicles with the membrane and release of the small molecular mass neurotransmitters are thought to occur rapidly and completely. Given the time frame over which these events occur, imaging secretion in real time in this system has been challenging.
Unlike neuronal and endocrine cells, many exocrine cells synthesize and secrete bulky highly glycosylated cargo such as mucins. The secretory vesicles carrying this bulky cargo tend to be large (1 to 2 µm diameter in mammals and 3 to 8 µm in diameter in Drosophila). Mucins are known to undergo Ca2+- and pH-dependent unfolding and expansion during the process of secretion (Ambort et al., 2012). Additionally, highly glycosylated mucins become hydrated and expand upon exposure to the lumen, creating additional forces that need to be counter-balanced during secretion (Birchenough et al., 2015). Thus, given the size of the secretory vesicles and the unique biophysical properties of their contents, components of the actomyosin complex are required for productive secretion in many exocrine systems (Shitara and Weigert, 2015).
Previous studies have highlighted diverse and disparate roles for the actin cytoskeleton in the process of regulated secretion. Actin along the cortex of cells is thought to serve as a barrier to prevent premature fusion of secretory granules with the apical PM in certain systems (Brown et al., 2011; Giner et al., 2005; Trifaró et al., 2008). Other studies have shown that actin is essential for proper fusion pore expansion (Larina et al., 2007) and vesicle compression to expel large viscous cargo (Jerdeva et al., 2005; Miklavc et al., 2015, 2012; Nightingale et al., 2011). There is also a suggested role for transient actin cables in directing vesicles to the apical PM (Geron et al., 2013). In vivo imaging studies in rodent salivary glands have shown that F-actin is required for the stabilization and collapse of the secretory granules after their fusion with the PM (Masedunskas et al., 2011). However, the specific factors involved in actin regulation and their spatiotemporal dynamics during regulated exocytosis are not completely understood.
Although the bulk of our knowledge regarding the factors involved in secretion has come from yeast and transformed cells in culture, recent developments in real-time imaging within secretory organs have provided dynamic spatial and temporal information not previously seen in vivo or ex vivo (Masedunskas et al., 2011; Rousso et al., 2016; Tran et al., 2015). Recent real-time imaging studies performed in the major exocrine gland of Drosophila (the salivary gland) have begun to more precisely order the events taking place during regulated exocytosis and to define the factors required for its completion.
The Drosophila larval salivary gland
The Drosophila larval salivary gland system consists of a central common duct and two individual lateral ducts, each of which connects to a gland composed of columnar epithelial secretory cells (Fig. 2A). During the third-instar larval stage, salivary glands produce and secrete highly glycosylated mucins, or ‘glue’ proteins, that will be expectorated from the animals to allow attachment to a substrate just before metamorphosis (Fraenkel, 1952). The glue proteins are made, packaged and secreted in a developmentally regulated fashion in response to the steroid hormone 20-hydroxyecdysone (20E) (Lehmann, 1996). In response to a lower titer pulse of 20E during the mid-third-instar larval period, expression of the glue genes is initiated (Fig. 2B). Glue proteins (also known as salivary gland secretion proteins or Sgs proteins) are synthesized in the secretory apparatus, where they are glycosylated, and then packaged into small immature granules at the trans-Golgi network (TGN) in a clathrin- and AP-1-dependent process (Burgess et al., 2011). As development proceeds, numerous small granules (∼1 µm in diameter) transition to fewer granules of larger diameters (Farkas and Suakova, 1999) until they reach a mature size of 3 to 8 µm in diameter (Rousso et al., 2016; Tran et al., 2015) (Fig. 2B). A high-titer pulse of 20E during the later stages of the third-instar larval period signals the glands to secrete their contents in preparation for metamorphosis (Fig. 2B). It is after this pulse of 20E that secretory granules begin to fuse with the PM and the process of secretion occurs. Thanks to the imaging tools developed by the Drosophila community, many steps in this process can be imaged in real time within the organs of animals. For example, the use of larvae expressing fluorescently-labeled glue proteins (Sgs3–GFP) (Biyasheva et al., 2001) allows one to follow the synthesis, packaging and final secretion of the cargo itself (Burgess et al., 2011; Rousso et al., 2016; Tran et al., 2015). Fluorescent markers that detect actin (Hatan et al., 2011; Riedl et al., 2008) and myosin (Buszczak et al., 2007), as well as the apical PM (Pfeiffer et al., 2010) or phosphoinositides (Verstreken et al., 2009), make it possible to follow the spatial and temporal dynamics of each of these molecules at each step of secretion by imaging. Additionally, Drosophila salivary glands can be cultured ex vivo and stimulated to secrete their contents by the exogenous addition of 20E (Costantino et al., 2008; Tran et al., 2015), thus making it possible to obtain high-resolution images at distinct times during the secretory process.
Technical considerations for real-time imaging in organs
Culturing glands ex vivo from flies that express various fluorescent markers has provided unprecedented temporal and spatial resolution to perform 3D confocal imaging in real time on individual granules. However, one of the most challenging aspects of imaging dynamic processes at high resolution in living organs is minimizing the amount of organ movement that occurs during the secretory process (also known as sample drift). Sample drift in the x, y and z dimensions can interfere with appropriate image acquisition, particularly when imaging over time. To gain the clearest time-resolved view of these processes, various strategies have been employed to minimize motion artifacts during image acquisition. For intravital imaging of rodent organs, custom holders and microstages are often employed (Masedunskas et al., 2012a). Alternatively, various coatings (Larina et al., 2007) or treatments (Rousso et al., 2016) can be applied to the imaging vessel to facilitate adherence of isolated organ or tissues to the imaging surface. For our studies, we utilize a porous polycarbonate membrane to immobilize the salivary glands on the bottom of the imaging vessel (Fig. 3A). Our approach allows visualization of the secretion of distinct single granules and ensures the maintenance of the health of the organ throughout the duration of the experiment (Tran et al., 2015). Moreover, this method allows the faithful recapitulation of the dynamics of secretion that are seen in vivo (Tran et al., 2015). An additional strength of this system is the ability to image secreting granules either in the same plane as or perpendicular to the apical membrane (Fig. 3B), something that cannot be done with cultured cells. Therefore, by simply imaging different areas of the gland, we can visualize different aspects of secreting granules in different planes without relying on computational reconstruction in the x,z or y,z planes, which would require robust sampling intervals to relay what is being observed and can be subject to visual anomalies due to the poor axial resolution of conventional fluorescent microscopy (North, 2006). Conversely, if x,z or y,z reconstructions are necessary, the ability to image in both planes allows us to confirm the temporal and spatial observations made using this approach. Although our technique negates the majority of the movement associated with a gland that rests freely at the bottom of a medium-filled vessel, one should still determine if there is sample drift before the start of image acquisition. Additionally, the health and integrity of secretory cells within the entire salivary gland must be assessed before the start of each experiment, as it is unknown how unhealthy or damaged cells can affect secretion throughout the rest of the organ.
When performing confocal imaging in real time, attention should also be paid to parameters that can alter the spatial and temporal quality of the image. For example, altering the dynamic range of the image by setting the detector offset value to a more negative number and thus removing detected fluorescent signal information from the image can influence the sensitivity as well as the integrity of the signal seen for certain markers (see example in Fig. 4A). Additionally, image acquisition speed (the time intervals over which images are taken) can influence the ability to separate events temporally; imaging at intervals that are longer than the biological events taking place will make them appear to be simultaneous rather than temporally distinct. As shown in Fig. 4B, imaging at 5-s intervals and visualizing that time series at 10-s intervals is sufficient to visually separate the recruitment of actin and myosin in time, whereas visualizing the same series at 30-s intervals will show their recruitment to be simultaneous. A final consideration is the degree of sampling in the z plane (Fig. 4C). Under-sampling in the z plane can lead to distortion of both the image shape as well as the integrity of the fluorescent signal, as large gaps of information are present in the image acquired (Fig. 4C, left panel versus right panel). Indeed, alteration of detector offset parameters and z sampling can change the apparent distribution of fluorescent markers (making them appear continuous or discontinuous) and can influence the observed shape of cellular structures, such as secretory vesicles. As imaging parameters can influence the images obtained as well as the conclusions drawn, care should be taken to faithfully document all imaging settings for each experiment.
Overview of regulated exocytosis in Drosophila
Using this system and imaging secretory cargo, in this case Sgs3 (Sgs3–GFP), we find that full-sized vesicles are formed through homotypic fusion of smaller vesicles (Fig. 5A; Movie 1); this confirms what has been proposed previously as a result of the still imaging of secretory granules at different stages of granule biogenesis (Burgess et al., 2011; Farkas and Suakova, 1999). Furthermore, by imaging mature Sgs3–GFP-containing secretory vesicles, the process of secretion can be visualized from start to finish. For example, mature secretory granules move to the apical PM to begin secretion only after the high-titer pulse of 20E. Without this hormone pulse, secretory granules will remain in the cytoplasm indefinitely. Secretory granules enlarge upon initial fusion with the PM, probably due to either pressure from the lumen or hydration-related expansion of the highly-glycosylated cargo present within the granules. Additionally, changes in the integrity and structure of the apical membrane [visualized using the apical marker myristoylated tdTomato (Myr–tdTomato)] at the point of secretory vesicle fusion are initially seen, followed by restoration of the membrane after secretion of the cargo is complete. These observations suggest that secretion in this system occurs with the extrusion of the mucinous cargo followed by complete integration of the secretory vesicle membrane with that of the PM (Fig. 5B) (Tran et al., 2015). No evidence of compensatory endocytosis to recover the vesicular membrane was observed. This is likely to be due to the fact that after secretion is complete, the salivary glands undergo histolysis during metamorphosis (Bodenstein, 1950). Real-time imaging of complete secretion at single-granule resolution has also revealed that secretion occurs through the fusion of individual granules with the PM to secrete their contents; we found no evidence of granules fusing with other granules that are docked at the plasma membrane (compound exocytosis). This is similar to what has been seen in rodent salivary glands, where secretion occurs in the absence of compound exocytosis (Masedunskas et al., 2011). Finally, it takes ∼90–120 s for a single granule to dock with the PM, secrete its contents and complete membrane integration in this system, which provides sufficient time to image specific events during the process of secretion.
The role of actin during regulated exocytosis in Drosophila
As mentioned previously, prior studies have demonstrated roles for actin in secretion in diverse systems. However, the spatial and temporal recruitment of actin and the factors that regulate polymerization in the context of exocytosis had not been detailed in real time. By expressing a fluorescent version of the filamentous actin-binding protein Lifeact in Drosophila salivary glands, we and others were able to image actin recruitment and its dynamics at a single granule resolution (Tran et al., 2015; Rousso et al., 2016). We demonstrated that actin is abundantly present along the apical PM and is rapidly cleared at the point of secretory vesicle fusion (Tran et al., 2015) (Fig. 5C), suggesting that, as seen in other secretory systems, actin could serve as a physical barrier to premature granule fusion with the PM. Interestingly, 3D real-time imaging of actin dynamics at the point of vesicle fusion demonstrates that actin is rapidly recruited to the fused secretory vesicle directionally, emanating from the PM (Fig. 5C) (Tran et al., 2015). Actin can be seen to begin to cover the fused secretory vesicle from the point of fusion with the PM and then rapidly move upward until the entire vesicle is coated. Thus, this work suggests a coordination between actin clearance at the PM and directional actin polymerization around the fused secretory vesicle.
This system was used to address the temporal order of fusion pore formation (fusion between the secretory vesicle and the PM) relative to actin recruitment. By infusing the lumens of the salivary glands with a fluorescent dextran (Larina et al., 2007), we can visualize the moment the fusion pore forms, as dextran from the lumen will leak into the fused secretory granule. Performing this experiment in flies expressing Lifeact-GFP demonstrates that the fusion pore forms before actin is recruited to the secretory vesicle membrane (Fig. 5D) (Tran et al., 2015). Thus, secretion begins with the clearance of apical actin at the PM, followed by fusion pore formation and the subsequent directional actin recruitment to the fused secretory vesicle membrane.
What mediates the directional recruitment of actin to the secretory vesicle membrane? Previous studies in other systems have shown that redistribution of membrane lipids can influence actin recruitment and membrane fusion events (Shewan et al., 2011). The membrane phosphoinositide phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P2, or PIP2] has been found to stimulate actin polymerization by activating actin nucleators (Higgs and Pollard, 2000). In another study, PIP2 has been shown to regulate myoblast fusion by influencing the localization of actin nucleators at the fusion site (Bothe et al., 2014). We therefore investigated the temporal order between PIP2 redistribution to the vesicle membrane and actin recruitment using fly lines co-expressing a PIP2 reporter (Verstreken et al., 2009) and Lifeact-RFP. Before vesicle fusion, PIP2 is abundantly present along the apical PM but absent from vesicular membranes, but as secretion begins, PIP2 is rapidly recruited to the membrane of newly fused secretory vesicles (Fig. 5E) (Tran et al., 2015). Interestingly, recruitment of PIP2 precedes that of actin, suggesting that membrane mixing and PIP2 movement to the fused secretory vesicle might act as a signal to initiate actin recruitment (Rousso et al., 2016; Tran et al., 2015). Given the temporal sequence of events, our model suggests that fusion pore formation mediates membrane mixing and PIP2 recruitment, which then might assist in actin recruitment and polymerization (Fig. 6).
The observation that actin coating of the fused secretory vesicles is one of the early events in the secretory process strongly suggests that actin is required for subsequent steps to occur. Previous cell culture studies have provided evidence that actin present on secretory vesicles generates the compressive forces necessary to mediate secretion of cargo (Jerdeva et al., 2005; Miklavc et al., 2015, 2012; Nightingale et al., 2011). Additionally, intravital imaging of secretion in rodent salivary glands using the actin inhibitors latrunculin A (LA) or cytochalasin D (CD) has demonstrated that disruption of actin polymerization interferes with the gradual collapse of the secretory vesicles in vivo (Masedunskas et al., 2011). Our real-time high-resolution imaging of secretion in Drosophila salivary glands that had been treated with either LA or CD shows a complete loss of all actin on fused secretory granules as well as a complete disruption of secretion (Tran et al., 2015). Although vesicles continued to fuse with the apical PM, they failed to collapse and secrete their cargo. Instead, the fused vesicles expanded into the cytoplasm, probably owing to hydrostatic pressure from the lumen or to hydration and expansion of the cargo still present in the vesicle. Interestingly, loss of actin on the fused granules also resulted in compound fusion with other vesicles, suggesting that actin recruitment to fused vesicles is also required for the prevention of compound exocytosis in this system (Tran et al., 2015).
To dissect out the specific roles of various actin forms during regulated secretion, we and others took advantage of the genetic tools unique to this system to image the recruitment of factors involved in actin synthesis as well as to disrupt the genes that encode these factors. Actin can exist either in the form of monomers (G-actin) or in a number of polymerized states, including linear, branched and bundled (F-actin). Linear actin formation is initiated by the members of the formin family, including Diaphonous (Dia), which is activated by small Rho-GTPases and phosphoinositides (Faix and Grosse, 2006; Gasman et al., 1998; Goode and Eck, 2007). Branched actin filaments are formed through the coordinated action of nucleation-promoting factors (e.g. WASp, Wash, Whamy and SCAR) and the actin-related protein 2/actin-related protein 3 (Arp2/3) complex (Firat-Karalar and Welch, 2011; Goley and Welch, 2006; Rodriguez-Mesa et al., 2012; Rotty et al., 2013). Arp2/3 binds to linear actin filaments and is then activated by the nucleation-promoting factors to mediate the formation of branched actin structures. Drosophila lines exist that express fluorescently-tagged versions of Dia (Rousso et al., 2016) and Arp3 (Rajan et al., 2009), which, along with Lifeact, can be used to establish the temporal order of acquisition of these factors relative to that of the actin coat on the secretory vesicle. Indeed, Dia is recruited to fused secretory vesicles before the actin coat (as detected by Lifeact) (Rousso et al., 2016), whereas Arp3 is recruited after the actin coat (Tran et al., 2015). These results suggest a model in which the formation of the actin coat on fused secretory vesicles occurs in a two-step process, with Dia forming the initial linear actin coat followed by recruitment of Arp2/3, which then initiates formation of branched actin structures (Rousso et al., 2016; Tran et al., 2015) (Fig. 6).
Such a two-step mechanism of actin coat formation on fused vesicles suggests that different actin structures play unique roles during the process of regulated secretion. The role of Dia has been examined in other secretory contexts, and Dia is suggested to have diverse roles in different systems. In mammals, the murine form of Diaphanous-related formin 1 (mDia1; also known as Diaph1) has been reported to generate actin microfilaments that emanate from the apical PM and serve as tracks to direct secretory vesicles to fusion sites at the apical PM (Geron et al., 2013). Expression of a dominant-negative form of mDia1 does not affect actin localization along the apical PM or actin coating of secreting granules, or indeed secretion, but does result in disorganized vesicle trafficking and vesicles that fuse with one another (Geron et al., 2013). In contrast, in Drosophila, Dia appears to have a primary role in the formation of F-actin along the apical PM and around secreting granules (Massarwa et al., 2009; Rousso et al., 2016). Loss of Dia in the Drosophila salivary gland does not affect the number of fused secretory granules at the apical PM but instead disrupts the accumulation of actin around fused vesicles, as well as the amount of secreted cargo (Rousso et al., 2016). However, it remains unclear why Dia influences actin structures differently in different secretory systems. In this context, high-resolution imaging techniques, such as super-resolution microscopy and/or immunoelectron microscopy, could aid in the future identification of the specific actin structures that are observed to vary between the different experimental systems.
As mentioned earlier, Rho-GTPases regulate Dia-mediated actin polymerization (Spiering and Hodgson, 2011). In the Drosophila salivary gland, Rho1 knockdown disrupts secretion and actin coating of exocytosing secretory granules (Rousso et al., 2016). Interestingly, Rho1 appears on fused secretory granules at the same time as the actin coat, suggesting that Dia is first recruited to granules (possibly through interactions mediated by membrane mixing) and that Rho1 is subsequently recruited to modulate Dia activity (Rousso et al., 2016) (Fig. 6).
In our work, we interrogated the function of the branched actin nucleators, through knockdown of Arp2, Arp3 or their activator WASp. In each case, secretion was disrupted as fused secretory granules continued to enlarge and failed to exocytose their cargo (Tran et al., 2015). Similar results were obtained by Rousso et al. when they examined the effect of knockdown of Arp2 (Rousso et al., 2016). Interestingly, Lifeact could still be seen on fused granules upon knockdown of Arp2, Arp3 or WASp (albeit weaker and less evenly distributed), suggesting that linear actin was still present (Tran et al., 2015). Although fused vesicles failed to secrete their cargo, there was no evidence for compound vesicle fusion such as that observed upon treatment with LA or CD (where all actin on the fused granules was lost). These results suggest that linear actin filaments, first formed on the fused granule by Dia, confer protection from compound fusion. Branched actin structures, formed subsequently from linear actin through the action of Arp2/3 and WASp, then serves as a component of the actomyosin machinery that mediates the compression forces necessary for cargo expulsion and membrane integration. Taken together, these findings suggest specific roles for each type of actin in different steps of the secretion of large highly-glycosylated cargo, such as mucins.
Myosin in regulated exocytosis
Myosin is a key factor in the contractile forces that are generated through the actomyosin complex in many systems (Sellers, 2000). Drosophila myosin II consists of the heavy chain Zipper (Zip) and the light chain Squash (Sqh), which is activated by myosin light chain kinase (MLCK). Myosin binds to actin through the N-terminal globular domain of the heavy chain (Sellers, 2000). Using fly lines that express Zip–GFP, the recruitment of myosin to secretory granules in the Drosophila salivary gland has been examined in real time. Although Rousso and colleagues conclude that myosin and actin are recruited simultaneously to fused secretory vesicles (Rousso et al., 2016), we have observed the recruitment of myosin to the fused secretory vesicle to occur after that of actin (Tran et al., 2015). The reason for the temporal difference in myosin recruitment between these two studies is probably due to the rate of image acquisition (Fig. 4B). In addition, Rousso et al. (2016) conclude that myosin is organized in a stripe-like pattern on the fused secretory granules, resulting in non-isotropic (unevenly distributed) compression of the granule as cargo is expelled. However, we have not observed the striped myosin pattern nor have we detected evidence of non-isotropic compression (Tran et al., 2015). Rather, we routinely see a uniform distribution of myosin across the entirety of the granule with granule compression occurring isotropically (evenly around the whole granule) in every instance of granule imaging (Tran et al., 2015). Because both studies utilized the same transgenic lines for detection of myosin (Zip–GFP), the reasons for these disparities are most probably due to differences in image acquisition parameters (Fig. 4). For example, we are able to recreate the myosin stripes (either horizontal or vertical) by altering detector offset parameters and under-sampling in the z plane (Fig. 4D). Given that the dimensions of the reported myosin stripes are close to the limit of resolution of light microscopy, immunoelectron microscopy could be employed in the future to conclusively determine how myosin is arranged on fused secretory granules.
The functional role of myosin in secretion has been examined in a number of systems. For instance, work in rodent salivary glands has previously shown that treatment with the myosin inhibitor blebbistatin causes a decrease in the rate of secretory granule collapse after fusion with the PM, suggesting that myosin II is part of the machinery that drives membrane integration (Masedunskas et al., 2011). Drosophila myosin is insensitive to blebbistatin owing to a mutation in the active site loop switch-2 (Heissler et al., 2015); therefore, the role of myosin was examined by performing RNAi interference (RNAi) targeting zip (Rousso et al., 2016). Rousso et al. note that the fused secretory vesicles did not change in size (did not collapse), and a two-fold reduction in glue secreted into the lumen was observed, suggesting a role for myosin in generating the contractile forces necessary for cargo expulsion and membrane integration (Rousso et al., 2016). Additionally, knockdown of the Rho-associated kinase (Rok) abolished myosin recruitment to vesicles and recapitulated the secretion defects seen upon zip knockdown, highlighting a regulatory role for Rok in myosin recruitment and cargo expulsion (Rousso et al., 2016).
A model for regulated secretion
Taken together, the real-time imaging within Drosophila salivary glands highlights a process in which secretory vesicle formation occurs through homotypic fusion of immature granules to generate mature granules (Fig. 6). After a developmentally regulated high-titer hormone pulse, secretion begins with secretory granules fusing with the apical PM at sites of cortical actin clearance, followed in close succession by fusion pore formation and movement of components of the apical PM (PIP2) into the membranes of the fused granules. This membrane mixing is likely to be responsible for the directional recruitment of actin that is seen upon 3D real-time imaging (possibly through directional recruitment of Dia from the PM). After membrane mixing, Dia recruitment to the vesicle membranes is then followed by Rho1 to initiate polymerization of the linear actin coat. Linear actin on the fused vesicle acts to both prevent compound exocytosis and to recruit the branched actin nucleator Arp2/3, which then initiates branched actin formation. Rok-regulated myosin recruitment completes the formation of the actomyosin complex that is required to generate the forces necessary to expel the cargo and complete membrane integration (Fig. 6). This mode of secretion of large mucinous cargo is considerably different from that of hormone secretion in endocrine cells as demonstrated recently (Zhao et al., 2016). Confocal and super-resolution stimulated emission depletion (STED) imaging of secretion in chromaffin cells and pancreatic β-cells has revealed a hemi-fusion intermediate between the secretory vesicle and the PM (Zhao et al., 2016), where fusion between the inner leaflet of the cell and the outer lipid layer of the secretory granule forms in the absence of a complete fusion pore. Interestingly, these intermediates were stable for many seconds before full fusion occurred, suggesting that secretion proceeds through a hemi-fused intermediate. Additionally, fully-fused vesicles have been seen to bud back off from the cell membrane through a hemi-fused intermediate, supporting the ‘kiss-and-run’ model of exocytosis. Whether these reversible intermediates exist in other types of secretory system remains to be determined but is now approachable with these advanced imaging techniques. Interestingly, we and others have not detected evidence of ‘kiss-and-run’ intermediates under normal circumstances in the Drosophila salivary gland system, highlighting that unique structures and mechanisms could be required to accomplish the secretion of diverse cargo from specialized cells and tissues (Rousso et al., 2016; Tran et al., 2015).
Conclusion and perspectives
The real-time imaging techniques described herein combined with the molecular genetic approaches available in Drosophila can be used as a platform to address numerous mechanistic questions that remain in the field of exocrine biology and secretion. For example, the factors that mediate homotypic vesicle fusion, as well as the signals that halt this fusion once granules are of a mature size, remain to be elucidated. Many questions also remain with regard to the dynamics of cytoskeletal proteins, such as how actin clearance at the site of vesicle fusion with the PM is regulated and how the factors that regulate actomyosin assembly on fused granules are recruited and coordinated. This system can also be used to address unanswered questions in mucin biology, such as how mucins are loaded into granules and what role their glycosylation status plays in mucin packaging, unfolding, stability and function.
Although the temporal resolution attainable through resonant scanning confocal and spinning disc imaging affords the ability to image secretory events in real time in vivo and ex vivo, the spatial resolution of fluorescent microscopy will always be limited by diffraction. Although the advent of super-resolution microscopy alleviates some of these constraints, the approaches that allow for the highest gain in resolution also require acquisition times that preclude the live imaging of dynamic events such as those involved in secretion. Additionally, the imaging depth within the organ required to visualize secretion events also complicates image acquisition and negates some of the benefits of super-resolution imaging. Ultimately, to directly visualize certain features, such as linear and branched actin structures, or myosin on secretory granules, resolution at the level of electron microscopy is required. Visualizing and definitively identifying cytoskeletal components or multi-subunit protein complexes by electron microscopy poses its own challenges. However, with immunoelectron microscopy and the development of correlative light and electron microscopy (CLEM) approaches for 2D and 3D visualization of tissues and organs, one can begin to address some of these questions. CLEM approaches allow for the mapping and registration of distinct fluorescently labeled proteins within the landscape of an electron micrograph. Additionally, as live-imaging studies are used to inform future CLEM and immunoelectron microscopy studies, dissecting the mechanisms of regulated secretion at previously unattainable temporal and spatial resolution will be possible.
We would like to thank members of our laboratory and the Shilo laboratory for helpful discussions. We would also like to thank our colleagues from the Drosophila community, the Vienna Drosophila RNAi Center, the Bloomington Stock Center and the Developmental Studies Hybridoma Bank for fly stocks and other reagents that made this work possible.
Work in the Ten Hagen laboratory is supported by the Intramural Research Program of the National Institute of Dental and Craniofacial Research, National Institutes of Health grant Z01-000713 to K.G.T.H. Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.