A complete Rab screening reveals novel insights in Weibel–Palade body exocytosis

Cells respond to pathophysiological conditions while maintaining their internal equilibrium. Accomplishment of this task depends on the coordinated communication between cellular compartments, which is mainly achieved by vesicular transport. Each transport process consists of a sequence of events involving the generation of a transport vesicle, by budding from a precursor compartment, transport of the vesicle to its destination and, finally, docking and fusion of the vesicle with the target compartment. Tight regulation of these vesicular transport processes is critical for the correct flow of cargo, which warrants maintenance of the biochemical composition of intracellular compartments. Members of the Rab and SNARE families of proteins are key players in these processes. While SNAREs participate mainly in the final fusion reaction, Rabs possess a more complex role, controlling vesicle biogenesis, their transport and docking to the target membrane, as well as the fusion reaction (Stenmark, 2009; Zerial and McBride, 2001). More than 60 members constitute the Rab family. Each one of them fulfils its specialised role by recruiting in its proximity effector molecules, e.g. motors (Hammer and Wu, 2002) or tethers (Christoforidis et al., 1999; Sztul and Lupashin, 2006), in a coordinated manner.

Specificity of vesicular transport relies largely on the differential localisation of Rab GTPases at distinct intracellular compartments. However, a large number of different Rabs do reside at the same compartment. For example, at least six different Rabs, Rab5, Rab4, Rab11, Rab21, Rab22 and Rab15, have been found at early endosomes (Simpson et al., 2004; Sönnichsen et al., 2000; Zuk and Elferink, 1999). Even more complex, in terms of number of resident Rabs, are synaptic vesicles, on which at least seven Rabs (Rab3, Rab27, Rab4, Rab5, Rab10, Rab11, and Rab14) have been identified (Pavlos et al., 2010). It seems, thus, that a large number of organelle-specific Rab GTPases orchestrates the dynamics of individual compartments.

Exocytosis of secretory vesicles in response to extracellular stimuli is a specialised trafficking pathway that applies to almost every cell type and plays critical role in fundamental cellular processes, such as cell growth and differentiation, as well as in specialised tissue functions, e.g. metabolism, immune response and neurotransmission (Burgoyne and Morgan, 2003). Exocytic vesicles derive from TGN, where bulk sorting of cargo takes place, then, they undergo a maturation process, where membrane and membrane components recycle back to the TGN and finally fuse with the plasma membrane thereby releasing their cargo extracellularly (Burgoyne and Morgan, 2003; Kim et al., 2006). Similarly to other organelles, a large number of Rab GTPases has been identified on secretory vesicles (Fukuda, 2008). This feature reflects the dynamic nature of secretory vesicles as well as the diversity of the organelles they communicate with.

A specialised type of secretory vesicles, Weibel–Palade bodies (WPBs), play critical role in vascular physiology (Valentijn et al., 2011). WPBs are endothelial-specific organelles (Weibel and Palade, 1964) carrying a variety of bioactive molecules, which, upon cellular activation and fusion with the plasma membrane, are released in the blood stream and play important roles in thrombosis, inflammation and angiogenesis (Rondaij et al., 2006). Trafficking of WPBs, after budding from the TGN, comprises of transport along microtubules and trapping into the actin cortex (Manneville et al., 2003). During this process, WPBs recruit regulatory molecules, such as Rab27a, and cargo molecules, e.g. CD63 (Arribas and Cutler, 2000; Hannah et al., 2003; Kobayashi et al., 2000). The latter protein is delivered to WPBs from an endosomal population in an AP3-dependent manner (Harrison-Lavoie et al., 2006).

Given the importance of Rabs in membrane trafficking, in order to understand the mechanisms governing WPB dynamics, we first need to know the complete set of Rabs that localise to these organelles and their implication in cargo secretion. So far, the only members of the large family of Rabs found on WPBs are Rab27a and Rab3d (Hannah et al., 2003; Knop et al., 2004). Since the identification of these Rabs on WPBs has been based on an educated guess approach, similarly to most studies in membrane trafficking, it is expected that this is only a limited view of the diversity of Rab GTPases involved in WPB dynamics and that other unanticipated Rabs may also participate in transport of these organelles. Indeed, studies in different cell types suggest that the group of Rabs present on a given organelle is quite complex as well as organelle and cell-type specific. Thus, to overcome the inherent limitations of the educated guess approach, here we undertook an unbiased strategy by performing a complete Rab screening. This method allowed comprehensive identification of the involvement of Rabs in WPB secretion and led to unexpected findings regarding the identity of Rabs that regulate exocytosis in endothelial cells.

To identify the complete group of Rab GTPases that localise to WPBs, we performed a full Rab screening by transfecting expression plasmids of Rab GTPases (Rab1-43), fused to the green fluorescent protein (GFP), in human umbilical vein endothelial (HUVE) cells. Simultaneous labelling of the cells with antibodies against GFP and vWF, the main cargo of WPBs (Sadler, 2009), followed by analysis of the intracellular localisation of the GFP–Rabs by confocal microscopy, allowed identification of the WPB-resident Rabs. Interestingly, a total of five Rab GTPases, Rab27a, Rab3a, Rab15, Rab33a and Rab37, were identified at the membrane of WPBs (Fig. 1; intracellular localisation of all Rabs in HUVE cells, Rab1-43, including all different isoforms, is depicted in supplementary material Table S1). Quantification of colocalisation showed that the percentage of WPBs that are positive for these five Rabs ranges between 64 and 83% (see values next to the corresponding images in Fig. 1). Additionally, the isoforms of the above five Rabs (Rab27b, Rab3b/c/d) are also localised to WPBs, with the exception of Rab33b (supplementary material Fig. S1). Several lines of evidence highlight the specificity of our screening approach. First, the presence of Rab3 (the Rab3d isoform) and Rab27a at WPBs is consistent with previous reports (Hannah et al., 2003; Knop et al., 2004). Second, the majority of GFP–Rabs localise to intracellular organelles that are vWF negative. Third, a number of Rabs previously identified on other secretory vesicles (e.g. Rab4, Rab8, Rab11, Rab18, Rab32, Rab38), were not found at WPBs (data not shown), which underlines the specialised molecular machinery employed by individual secretory organelles. Fourth, our screening data are consistent with the established localisation of specific Rabs to endosomes (Rab5, Rab4, Rab11), ER (Rab1) and Golgi (Rab6) (supplementary material Table S1).

Within the five Rabs found here, three are novel WPB Rabs. Among them, GFP–Rab37 localises almost exclusively to WPBs, while GFP–Rab33a colocalises as well with GalNac-T2 at the Golgi (supplementary material Fig. S2A), in agreement with the known localisation of Rab33 to this compartment (Zheng et al., 1998). Finally, the identification of Rab15 at WPBs is an unexpected finding, since, this Rab is considered to be a Rab of the endocytic pathway (Strick et al., 2002; Zuk and Elferink, 1999; Zuk and Elferink, 2000). Consistently, we found here that, in HUVECs, the perinuclear compartment where Rab15 also localises is positive for internalized transferrin as well as for Rab11 (supplementary material Fig. S2B,C), a marker of the recycling compartment. Additionally, knocking down of Rab15 results in reduced levels of endocytosed peroxidase (supplementary material Fig. S2D), further agreeing with the implication of this Rab in endocytosis (Strick et al., 2002; Zuk and Elferink, 1999; Zuk and Elferink, 2000).

Since HUVECs are specialised cells, we next tested whether the above exogenously expressed GFP–Rabs, those that are localised to WPBs, are actually expressed in these cells. Considering that different Rab isoforms are differentially expressed in different cell types, all possible isoforms of the five WPB Rabs have been included in this test. Q-RT-PCR was employed to check and actually verified the expression of Rab3a/b/d, Rab27a, Rab15, Rab33a/b and Rab37 in HUVE cells (supplementary material Fig. S3A,B), except Rab27b, which was only barely detectable (supplementary material Fig. S3A). Where antibodies were available (Rab27a and Rab3d), expression was confirmed by western blotting (supplementary material Fig. S3C). From all the above, we conclude that Rab3, Rab15, Rab27, Rab33 and Rab37 localise to WPBs and, in addition, all these five Rabs are expressed in HUVE cells.

To explore the role of the above five Rab GTPases on WPB exocytosis, we undertook a knocking down approach using specific siRNAs. Two different siRNAs, used separately, were employed to reduce the expression of each Rab GTPase. Given that Rabs exist in multiple isoforms, which may exert a differential role, we included in our siRNA screening specific siRNAs against all isoforms of the five WPB Rabs (Rab27a/b, Rab3a/b/d, Rab15, Rab33a/b and Rab37). Q-RT-PCR verified the efficiency of the siRNAs, in knocking down their targets (Fig. 2A), while for Rab27a and Rab3d, for which antibodies were available, knocking down efficiency was assessed by western blotting (Fig. 2A,B). Since Rab27b was barely detected by RT-PCR (supplementary material Fig. S3A), knocking down efficiency of the siRNAs against this isoform was evaluated by exogenous expression of myc–Rab27b in human embryonic kidney (HEK)-293 cells, followed by treatment with the siRNAs and analysis by western blotting against the myc epitope (inset in Fig. 2A). To evaluate the effect of each siRNA in WPB exocytosis, we stimulated confluent endothelial cells using a cocktail of ATP, VEGF and basic Fgf and measured vWF secretion by an ELISA assay. Co-stimulation with these three ligands was much more efficient, in comparison to each ligand separately, in inducing WPB exocytosis (supplementary material Fig. S4), thereby allowing a more accurate assessment of the role of the above GTPases. Furthermore, the values of secreted vWF were corrected taking into account the total vWF levels (the sum of secreted and intracellular, see supplementary material Fig. S5A), so as to evaluate the role of the Rabs in secretion independently from possible effects in other cellular functions (e.g. protein synthesis, cell proliferation, apoptosis).

To conclude about the involvement of a Rab in secretion, we had set two main criteria. First, the effect in secretion, either inhibitory or stimulatory, should be statistically significant with regard to control sample and the value of secreted vWF should be at least 20% higher (stimulation of secretion) or lower (inhibition of secretion) than the control. Second, both siRNAs, for any given Rab, should have a consistent effect in secretion. Based on these criteria, the RNAi experiments showed that Rab33a, Rab33b and Rab37 do not interfere with vWF secretion (Fig. 2C). Despite a great reduction of the mRNA of these Rabs, since we lacked antibodies against the endogenous proteins it cannot be excluded that lack of an effect in secretion might be due to inefficient knockdown. On the other hand, knocking down of Rab27a, by two siRNAs, which resulted in ∼80% reduction in protein levels (Fig. 2A,B), led to an almost 50% decrease in regulated vWF secretion (Fig. 2C). Knocking down of the other isoform of Rab27, Rab27b, resulted in a much weaker reduction in WPB secretion (Fig. 2C), consistently with its very low levels of expression (supplementary material Fig. S3A).

It was curious to note that severe reduction of the levels of Rab27a, a major Rab in exocytosis (Fukuda, 2008), still left 50% of vWF secretion unaffected. Although ablation of Rab27a has been previously suggested to lead to upregulation of Rab27b, which in turn could partially take over the function of Rab27a (Westbroek et al., 2004), there was no further reduction in vWF secretion upon simultaneous knockdown of both isoforms of Rab27, in comparison to knocking down of Rab27a alone (Fig. 2D; specificity of the siRNAs is shown in Fig. 2E). Thus, Rab27a is responsible for almost 50% of WPB secretion, while the remaining secretion is not due to Rab27b function.

The second Rab that we found to be involved in WPB exocytosis is Rab3. Both siRNAs against Rab3d, the only Rab3 isoform that has so far been implicated in vWF secretion (Knop et al., 2004), resulted in significant inhibition of vWF release (Fig. 2C). On the other hand, knocking down of Rab3b had no effect on vWF secretion (Fig. 2C). Similarly to Rab3d, knockdown of the Rab3a isoform, by two different siRNAs, resulted in a concomitant reduction in vWF secretion (Fig. 2C). The effect of Rab3a siRNAs on exocytosis was not due to a non-specific reduction of Rab3d levels, since, the siRNAs against Rab3a had no effect on the amount of Rab3d (see Fig. 2F). Thus, both Rab3 isoforms, Rab3a and Rab3d, are important for vWF secretion. Based on this conclusion and the high amino acid identity between Rab3a and Rab3d (75%), we next tested whether these two isoforms cooperatively regulate vWF secretion. Simultaneous knockdown of Rab3a and Rab3d did not reduce vWF secretion more than any single knockdown did (Fig. 2G). This result indicates that Rab3a and Rab3d regulate vWF secretion in a non-redundant, non-synergistic and non-additive manner. Finally, we were surprised to find that knocking down of Rab15 resulted in a significant reduction, up to 40%, of vWF secretion (Fig. 2C). From all the above, we conclude that three Rab GTPases, Rab27, Rab3 and Rab15, out of the five Rabs located at WPBs, are required for regulated exocytosis of these organelles. Notably, the above data do not seem to depend on the exact stimulant used to trigger exocytosis, since, the same Rabs were found to be involved in PMA-induced secretion (supplementary material Fig. S6A). Finally, although in the present study we focused only in stimulated secretion of WPBs, we also tested, for comparison, the contribution of WPB Rabs in constitutive secretion of vWF, which, within the 30 min that the secretion assay lasts, is four times lower, in terms of cargo release, than stimulated secretion. Among the Rabs tested, we found that Rab27a is also important in constitutive secretion while Rab33b plays an inhibitory role (supplementary material Fig. S5B). On the other hand, the other Rabs did not seem to be significantly involved.

Interestingly, knockdown of the above three Rabs (Rab3, Rab15 and Rab27) did not result in complete inhibition of WPB secretion. Although remaining secretion could be due to residual amount of the proteins that escaped knockdown, other hypotheses are also likely to be responsible. For example, these Rabs might regulate secretion of different pools of WPBs. If this assumption were correct, then one would expect that the three Rabs would localise to different pools of WPBs. However, the majority of WPBs that carry Rab27 are also positive for Rab3 (Fig. 3A) and Rab15 (Fig. 3B), which argues against the above notion. An alternative hypothesis is that the three Rabs might need to cooperate in order to reach the full capacity that drives secretion and that knockdown of individual ones is not enough to block exocytosis. To explore this possibility, we silenced Rab3 (either a or d), Rab27a and Rab15 in combinations of two at a time. Concomitant knockdown of Rab27a and Rab3d (Fig. 3B), or Rab27a and Rab3a (supplementary material Fig. S7A) failed to reduce vWF secretion more than any single knockdown did. On the other hand, simultaneous knockdown of Rab27a and Rab15 resulted in a significantly further reduction in vWF secretion in comparison to the single knockdown of any of these two Rabs (Fig. 3D). Finally, simultaneous knockdown of Rab3d and Rab15 (Fig. 3F), or Rab3a and Rab15 (supplementary material Fig. S7B) did not lead to any further decrease in vWF secretion than any single knockdown did. The above data suggest that, among the three Rabs involved in WPB secretion, only Rab27a and Rab15 are functioning cooperatively.

These findings led us to hypothesise that Rab27 and Rab15 may act through a common effector. To address this issue, we first sought to identify effectors of Rab27a in endothelium and then test whether they interact with Rab15 as well. To search for Rab27a effectors, we used a LexA-based yeast two-hybrid system to screen a cDNA library from human placenta, a tissue that is rich in endothelium and Rab27a (data not shown). Screening of 2.6×10⁶ transformants yielded nine bait-dependent interactors. Among them, the most reliable Rab27a interactor was Munc13-4, a previously reported effector of Rab27a in platelets, cytotoxic T lymphocytes (CTLs) or spleen (Ménager et al., 2007; Neeft et al., 2005; Shirakawa et al., 2004). Endogenous expression of Munc13-4 in endothelial cells was confirmed by western blotting using specific antibodies (Fig. 4A). Furthermore, consistently with its interaction with Rab27a, Munc13-4 colocalises with vWF (Fig. 4B) and with GFP–Rab27a (Fig. 4C). We next examined whether Munc13-4 also interacts with Rab15, or with any of the other WPB Rabs. To this end, we employed co-immunoprecipitation experiments in 293 cells, which were co-transfected with myc–Munc13-4 and each one of the five WPB Rabs separately (GFP–Rab27a, GFP–Rab3a, GFP–Rab15, GFP–Rab33a, GFP–Rab37), or GFP control. Given that Rab15 has so far been reported exclusively as an endocytic Rab, we also included in this experiment other endocytic Rabs (Rab5a, Rab4a, Rab11a), as negative controls. Apart from Rab27a, the only other Rab GTPase interacting with Munc13-4 was Rab15 (Fig. 5A). Consistently with these data, Munc13-4 was found to colocalise with Rab15 at WPBs in HUVE cells (Fig. 5B).

Munc13-4 is a 120 kDa protein, which plays an important role in the regulation of exocytosis of secretory lysosomes (Bossi and Griffiths, 2005; Elstak et al., 2011; Neeft et al., 2005). Genetic defects in Unc13d gene, which encodes for Munc13-4, lead to familial hemophagocytic lymphohistiocytosis (Feldmann et al., 2003). This knowledge, together with our data showing that Munc13-4 is a common effector of Rab27a and Rab15, suggested that this effector might play important role in WPB secretion. Indeed, knockdown of Munc13-4, using three different siRNAs, which all led to efficient reduction of Munc13-4 protein levels (Fig. 6A), resulted in significant inhibition of stimulated WPB exocytosis (Fig. 6B), without apparent effects in basal secretion. Reduction of Munc13-4 protein levels did not significantly influence the total levels of vWF (Fig. 6C). Thus, Munc13-4 is necessary for WPB secretion, similarly to Rab27 and Rab15, the two GTPases that Munc13-4 interacts with. Concomitant knockdown of Munc13-4 and Rab27a, or Munc13-4 and Rab15, did not reduce exocytosis below the level of the single knockdown of Munc13-4 (Fig. 6D), suggesting that Rab27a and Rab15 are on the same pathway with Munc13-4 in controlling WPB secretion. Altogether, the above data indicate that Munc13-4 is implicated in the Rab27- and Rab15-cooperative regulation of WPB secretion.

The identification of Rab GTPases that localise to WPBs has so far been approximated by an educated guess approach. Here, following an unbiased strategy, a localisation-based Rab screening, we identified five Rabs that localise to these organelles. Among them, three belong in the so-called exocytic Rabs (Rab3, Rab27 and Rab37), one is a Golgi-resident Rab (Rab33), and, surprisingly, one is an endocytic Rab (Rab15). The presence of multiple Rabs at WPBs is consistent with the current notion that each intracellular organelle, e.g. endosomes, Golgi, ER, including secretory granules, hosts numerous members of the Rab family (Stenmark, 2009).

Interestingly, if we exclude Rab3 and Rab27, which seem to be well preserved in different secretory compartments (Fukuda, 2008), there are significant differences within the Rab groups in different exocytic vesicles. Thus, Rab4, Rab5, Rab10, Rab11b and Rab14 have been identified on synaptic vesicles, while Rab17, Rab32 and Rab38 on melanosomes, Rab37 on mast cells, Rab33a and Rab37 on dense granules of PC12 cells and Rab11 on insulin granules (Beaumont et al., 2011; Masuda et al., 2000; Pavlos et al., 2010; Sugawara et al., 2009; Tsuboi and Fukuda, 2006; Wasmeier et al., 2006). Furthermore, even in the same cell there are differences in the resident Rabs between distinct secretory vesicles (e.g. α granules compared to dense granules in platelets) (Novak et al., 2002; Shirakawa et al., 2000). Therefore, one can conclude that each type of exocytic granule is characterised by a unique set of numerous Rab GTPases. This distinctive set of multiple Rabs determines the unique identity of secretory vesicles of different cells, via recruitment of effector proteins and lipid modification reactions (Stenmark, 2009; Zerial and McBride, 2001), mediated by each Rab separately. This process allows individual vesicles to accomplish their specialised functions, that is, acquirement of specific cargo, through communication with various intracellular compartments (e.g. early and late endosomes, Golgi, ER), transport of the vesicles on actin or microtubules, and finally, tethering, docking and fusion with the plasma membrane.

Among the five WPB Rabs, knockdown of Rab27, Rab3 and Rab15 inhibited WPB exocytosis, which is largely consistent with the general view that Rabs are positive regulators of membrane fusion (Zerial and McBride, 2001). On the other hand, we found no involvement of Rab33 and Rab37 in exocytosis. However, it cannot be excluded that lack of an effect in secretion might be due to inefficient knockdown (mRNA levels were significantly reduced, but we could not quantify the remaining protein levels since we lacked appropriate antibodies). Rab33 has been previously found to regulate transport between Golgi and endoplasmic reticulum (Kinchen and Ravichandran, 2008), while its localisation on the dense granules of PC12 cells has been proposed to be a remnant of immature vesicles from the TGN without functional significance (Fukuda, 2008). Such a scenario could also account for the WPB Rab33. Rab37, on the other hand, even though this is the first time to be identified on WPBs, has been previously found on secretory vesicles of mast cells, pancreatic cells and natural killer cells (Brunner et al., 2007; Casey et al., 2007; Masuda et al., 2000). Still, the functional role of these localizations remains unknown. Only recently, a role in TNFα release from macrophages has been attributed to this GTPase (Mori et al., 2011). In the case of WPBs, Rab37 could regulate a step distinct from fusion.

The most striking difference between WPBs and other secretory vesicles, with regard to resident Rabs, is Rab15, a so far considered endocytic Rab (Strick et al., 2002; Zuk and Elferink, 1999; Zuk and Elferink, 2000). Interestingly, from the evolution point of view, this Rab is a very close neighbour of the branch of the secretory Rabs (supplementary material Fig. S8) (Fukuda, 2008). Additionally, Rab15 has been found to interact with Mss4 (Strick et al., 2002), the mammalian homologue of Dss4 (Burton et al., 1993). The later has been shown to act as an exchange factor for Sec4 (Moya et al., 1993), a secretory Rab that controls fusion of secretory vesicles with the plasma membrane (Goud et al., 1988; Salminen and Novick, 1987). Although this is the first time that Rab15, a previously considered endocytic Rab, is implicated in regulated secretion, recycling endosomes and recycling endosome-associated Rabs, such as Rab11, Rab4 and Rab17, have been implicated in secretory pathways in other systems (Manderson et al., 2007; Reefman et al., 2010; Shirakawa et al., 2000; Sugawara et al., 2009). For example, in melanocytes, where recycling endosomes are in close contact with melanosomes and deliver cargo to the maturing organelles (Delevoye et al., 2009), it has been suggested that the endosomal pool of Rab17, and not the melanosome-resident one, contributes to melanosome release (Beaumont et al., 2011). In the present study, it is still uncertain whether the WPB Rab15 or the perinuclear pool of Rab15 contributes to WPB secretion. However, given that, first, knockdown of either Rab15 or Rab27, which is a well-established WPB-resident Rab (Hannah et al., 2003), had a similar influence in secretion; second, Munc13-4, Rab27 and Rab15 colocalise only on WPBs, in accordance with the finding that Munc13-4 is a common effector of Rab27 and Rab15; and third, the effect of Munc13-4 knockdown in WPB secretion parallels that of the combined knockdown of Rab27 and Rab15, we cannot exclude a direct role of Rab15 in the exocytic process. Alternatively, the fact that WPBs recruit cargo from the endocytic compartments, as it has been shown for the endocytic cargo CD63 (Arribas and Cutler, 2000; Kobayashi et al., 2000), suggests that Rab15 may also participate in the pathway that delivers cargo from the endosomes to WPBs. In this scenario, the dual interaction of Munc13-4 with Rab15 and Rab27 could provide the necessary link that mediates directionality of this trafficking route. Together with cargo delivery, this pathway could warrant delivery of molecules that are directly involved in WPB exocytosis, e.g. VAMP3, an endosomal SNARE that is also involved in WPB secretion (Pulido et al., 2011). Regardless of the exact role of Rab15 in WPB dynamics, our data clearly suggest that this member of the Rab family has important functional implications in WPB exocytosis. Interestingly, Munc13-4 has been implicated in tethering of recycling endosomes to Rab27-positive exocytic vesicles in CTLs (Ménager et al., 2007). Our finding that Munc13-4 interacts with both an endocytic and an exocytic Rab suggests that this binary interaction could be the underlying mechanism of the tethering process in CTLs, as well as, in general, of the sorting pathway between endosomes and secretory vesicles in different cell types.

Our results, which show that Rab27a plays a positive role in WPB secretion, confirmed by both a knocking down (Figs 2, 3, 6) and an overexpression approach (see supplementary material Fig. S6B,C), are in agreement with the general role that has been ascribed to this GTPase in exocytosis. Indeed, Rab27a has been reported to be an essential component of exocytosis of a plethora of specialised organelles, such as lytic granules of cytotoxic T lymphocytes (CTLs), insulin and glucagon granules, as well as dense granules of platelets, mast cells and PC12 cells (Fukuda, 2008). Consistently, Rab27a-mutant animals display clear secretion defects (Ménasché et al., 2000; Mizuno et al., 2007; Stinchcombe et al., 2001; Tolmachova et al., 2007; Wilson et al., 2000). Furthermore, Rab27a is required for exocytosis of secretory endo-lysosomes in non-specialised fibroblast-like cells (Laulagnier et al., 2011). Besides, Rab27a is responsible for the peripheral localisation of melanosomes (Hume et al., 2007) as well as of WPBs (Nightingale et al., 2009 and supplementary material Fig. S9). On the other hand, in other cases, Rab27 has been shown to exert a negative role in secretion (Desnos et al., 2003; Johnson et al., 2005; Yi et al., 2002). Consistently, a recent study has shown that Rab27 exerts an overall inhibitory role in exocytosis of WPBs (Nightingale et al., 2009). Although our data are seemingly in disagreement with this study, analysis by Nightingale et al. of the multimeric state of secreted vWF showed that the negative role of Rab27 is restricted to a subpopulation of WPBs that carry immature vWF, while it plays a positive role in secretion of WPBs containing highly multimerised vWF. Similarly, Rab27 has been suggested to differentially control granule movement at the plasma membrane and in the cytosol in NK cells (Liu et al., 2010). Thus, it is possible that the overall role of Rab27 in secretion depends on the ratio of mature versus immature granules. This ratio is known to be severely affected by cell density (Howell et al., 2004), and possibly by other factors, such as cellular activation, proliferation state, migration and others. Differences in these conditions, due to different transfection protocols and culturing methods, may account for the apparent differences between our study and the previous report.

Our finding that 50% of WPB secretion remained unaffected by the single knockdown of Rab27, Rab3 and Rab15 suggested that these Rabs might need to cooperate for achieving full secretion. Double knockdowns between the WPB Rabs were employed to address this issue. We initially thought that Rab3 is the best candidate for cooperating with Rab27, for a number of reasons. First of all, we found that two isoforms of this subfamily, Rab3a and Rab3d are necessary for WPB secretion. The previously attributed negative role of Rab3d in WPB secretion (Knop et al., 2004) could be due to the overexpression approach used to interfere with Rab function, since, in most cell types, both overexpression and knocking down of Rab3 result in inhibition of secretion (Fukuda, 2008). Secondly, Rab27 and Rab3 are known to share common effectors and regulators; and thirdly, these two Rabs cooperatively stimulate exocytosis of granules in PC12 cells (Tsuboi and Fukuda, 2006). However, when we co-silenced Rab27 and Rab3, WPB secretion was not reduced more than the single knockdowns, suggesting that these two Rabs do not cooperate in WPB secretion. These data, together with previous findings in PC12 cells, where simultaneous knockdown of Rab3a and Rab27a reduced stimulated hormone secretion even more than single knockdown, highlight again the differences between different secretory cells.

On the other hand, to our surprise, we identified a cooperative function between Rab27a and Rab15 in WPB secretion. Rab27 and Rab15 were identified on the same WPBs, they were both necessary for WPB secretion and, upon their co-silencing, there was further reduction in WPB secretion in comparison to the single knockdowns. The mechanism allowing complementarity between these two Rabs could be mediated via an interaction with a common effector, as has been suggested previously for the cooperation between Rab4 and Rab5 on early endosomes (de Renzis et al., 2002) as well as for Rab5 and Rab7 in early to late endosome progression (Chotard et al., 2010; Rink et al., 2005). Here, by combining yeast two-hybrid screening and affinity chromatography experiments, we identified Munc13-4 as a common effector between Rab27 and Rab15. This is the first time that Munc13-4 is identified as a Rab15 effector. Munc13-4 is a tethering factor that is mainly expressed in cells of the haematopoietic lineage, where it is required, amongst others, for secretion of the lytic granules of CTLs, platelet dense granules, as well as mast cell and neutrophil granules (Bossi and Griffiths, 2005; Fukuda, 2008). The present study adds endothelial cells in the list of blood vessel cells whose regulated secretion is controlled by this effector. The mechanism by which Munc13-4 controls secretion remains, however, unknown. In mast cells, the Munc13-4-mediated priming of secretory lysosomes on the plasma membrane may be accelerated by the Munc13-4–Doc2α interaction (Higashio et al., 2008). Munc13-4 is also thought to regulate conformational opening of the SNARE protein syntaxin11 in CTLs (Hong, 2005). As syntaxin4 and SNAP23 are important for WPB secretion (Fu et al., 2005; Pulido et al., 2011), it would be interesting to determine whether the Doc2α–Munc13-4 complex regulates assembly of these SNAREs.

Overall, the present study, by elucidation of the complete set of Rabs of WPBs, paves the way towards future studies that will aim to reveal the exact role of these small GTPases on WPB dynamics and the detailed mechanism by which they provide fidelity in transport of these organelles.

Human VEGF₁₆₅ and bFgf-2 were from Immunotools, ATP and PMA were from Sigma, while thrombin was from Enzyme Research Laboratories. Rabbit polyclonal antibodies against Rab27a were produced and affinity purified (Sulfolink, Thermo Scientific) in house. Polyclonal antibodies against Rab3d were a generous gift from R. Jahn. Antibodies against Munc13-4 (Murata et al., 2011; Shirakawa et al., 2004) and GFP (Boleti et al., 2010) have been described before. The 9E10 (anti-myc tag) monoclonal antibody was purified from the corresponding hybridoma using standard techniques. The monoclonal anti-tubulin antibody was from DSHB (Iowa), the monoclonal and polyclonal anti-vWF antibodies (M016, A0082 and P0226) were from Dakocytomation, whereas the monoclonal anti-FLAG antibody was from Sigma. Secondary antibodies, Alexa Fluor 488, Fluor 594 and Fluor 680 were from Invitrogen, while HRP-conjugated antibodies were from Jackson Immunoresearch.

Plasmids of GFP–Rabs (1–43) (Tsuboi and Fukuda, 2006), as well as of their isoforms (Matsui et al., 2011), and myc–Munc13-4 (Shirakawa et al., 2004) have been described before, while the expression plasmids of myc–Rab27a wt (rat) and myc–Rab27b wt (human), in the pCMV-Tag3B expression vector, were generated by sub-cloning of Rab27a and Rab27b (Shirakawa et al., 2004). Site-directed mutagenesis of hRab27a, to generate hRab27aQ78L for the yeast two-hybrid screening, was performed using a two-stage PCR strategy. The cDNA of mutant Rab27a was then subcloned into pLexA-N, to generate the bait construct for the yeast two-hybrid screening. FLAG–Rab27a wt was generated by PCR using myc–Rab27a wt (rat) as template and the primers 5′-CCACGCGGCCGCCCATGGGCCACCATGGACTACAAAGACGATGACGACAAGGGAGGTGGAATGTCGGATGGAGATATGAC-3′ and 5′-CCTCTAGAGCTAGCGGATCCTCAACAGCCGCATAACCC-3′. The PCR product was then sub-cloned into pCDNA3. All plasmids were purified with Endotoxin-free kit (Sigma) and sequenced to verify their sequence.

For small interfering RNA (siRNA) experiments, siRNAs for human Rab27a: Rab27a(1) 5′-GGGAAAAAAGAGUGGUGUAtt-3′, Rab27a(2) 5′-GGACCAGAGAGUAGUGAAAtt-3′; Rab27b: Rab27b(1) 5′-CCGAAUGGAUCUUCAGGGAtt-3′, Rab27b(2) 5′-CCCUUUUGGACUUAAUCAUtt-3′; Rab3a: Rab3a(1) 5′-GAUUCUCAUCAUCGGCAACtt-3′, Rab3a(2) 5′-GGAAUCCUUCAAUGCAGUGtt-3′; Rab3b: Rab3b(1) 5′-CGGGUGCAGAAUCACUUUAtt-3′, Rab3b(2) 5′-GCACAACGUGCUUGUUUCCtt-3′; Rab3d: Rab3d(1) 5′-GCATCGATTTCAAGGTCAAtt-3′, Rab3d(2) 5′-GCCUUUUCCCAUUGUAGAAtt-3′; Rab37: Rab37(1) 5′-CCAGCUUCCAGAUCCGAGAtt-3′, Rab37(2) 5′-GGAUAUGAGCAGCGAAAGAtt-3′; and Munc13-4: Munc13-4(1) 5′-ACUGAAUGGUUCCACCUGAtt-3′, Munc13-4(2) 5′-GGGACAAGAUCUUCCACAAtt-3′, Munc13-4(3) 5′-GAGCUUUGCUACAUGAACAtt-3′ were from Ambion, whereas siRNAs for human Rab15: Rab15(1) 5′-GGCAUGGACUUCUAUGAAACAAG-3′, Rab15(2) 5′-GGUGUUGACUUUAAGAUGAAGAC-3′; Rab33a: Rab33a(1) 5′-CCAGACAAGACUGAAGCCACCAU-3′, Rab33a(2) 5′-ACAGGAAGCUAACAGUAAAtt-3′; Rab33b: Rab33b(1) 5′-CCGCAUCUUCAAGAUAAUCGUGA-3′, Rab33b(2) 5′-GCUAAAAACCCCAAUGAUAtt-3′ and the control siRNA (Random DS) were from Biospring.

Primary HUVECs were isolated (Jaffe et al., 1973) and cultured (Papanikolaou et al., 2011) on collagen type I-coated surfaces. HUVECs between passages two to four were used in all experiments. HEK-293 cells were cultured in RPMI, supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin.

For DNA transfection of HUVEC, the Metafectene-Pro reagent (Biontex, Biontex Laboratories, Martinsried/Planegg, Germany) was used, according to manufacturer's instructions. Briefly, 70% confluent HUVEC were transfected with a DNA-to-lipid ratio of 1∶3. The complex was prepared in plain M199 medium, incubated for 20 min at room temperature and added to the cells containing serum-reduced M199 medium (5% FBS) with no antibiotics, heparin nor ECGS. The transfection medium was replaced, 3 hr later, by full HUVEC medium.

For small interfering RNA (siRNA) experiments, HUVECs were transfected, at a confluence of 70%, with 25 nM siRNAs for 72 hr with Lipofectamine RNAiMax reagent (Invitrogen), according to manufacturer's instructions. In the case of double siRNA transfections, 25 nM per siRNA was used and the cells, at the day of transfection, were at 50% confluence. The control samples of these experiments, where only one Rab was silenced, where supplemented with 25 nM random siRNA, so as to correct for the total amount of siRNA.

HEK-293 cells were transiently transfected with plasmid DNA using PEI (poethyleneimine, Fluka), according to standard protocols. Twelve hours later, cells were transfected with siRNA using Lipofectamine 2000 (Invitrogen) at a final siRNA concentration of 25 nM.

HUVECs transfected with targeted or control siRNAs were re-seeded, 24 hr post transfection, at the appropriate cell density in order to allow the cells to reach a fully confluent and tight monolayer at the end of the 72 hr culture. Cultures that did not meet these criteria were not processed further. Following the three days culture period, the cells were washed three times with HBSS (Hank's Balanced Solution, PAA) containing 0.1% BSA and they were incubated (stimulated secretion) or not (basal secretion) with a mixture of ATP (100 µM), VEGF (50 ng/ml) and bFgf (15 ng/ml), for 30 min. The supernatants were collected, centrifuged for 10 min at 1000 rpm and the relative amounts of vWF were quantified by ELISA (see below). Subsequently, the cells were lysed in 0.5% Triton X-100, 1 mM EDTA in PBS and the amount of vWF in the lysates, corresponding to the non-secreted part of the cargo, was quantified by ELISA. To calculate the percentage of secreted vWF, the values of released vWF were expressed as % of the total (total vWF equals the sum of the released and the non-secreted part of the cargo). Differences in total vWF levels, between the samples, were normalised. To correct for variations between different secretion experiments, we normalised the secretion data by setting as 100% the percentage of secreted vWF from control cells and calculated the secretion values of treated cells as a percentage of this control.

High-binding 96-well plates were incubated overnight with 200 µl anti-vWF antibodies (A0082, Dakocytomation), diluted 1∶400 in PBS, at 4°C. The antibody-coated wells were washed three times with PBS–0.1% Triton X-100 (washing buffer) and blocked with 200 µl PBS, 0.1% Triton X-100, 0.2% gelatine (blocking buffer). Then, 180 µl of culture supernatants, together with 20 µl of blocking buffer, were incubated in the wells for 90 min at room temperature. The wells were then washed three times with washing buffer and they were incubated with HRP-conjugated rabbit anti vWF antibodies (P0226, Dakocytomation) diluted 1∶4000 in blocking buffer. Upon 90 min incubation with the secondary antibody, wells were washed five times with blocking buffer. For detection and quantification of the antigen-antibody conjugates, 200 µl substrate buffer (containing 20 mg o-phenylenediamine in 50 ml phosphate citrate buffer, pH 5.0, supplemented with 20 µl H₂O₂ 30%) was added in each well. The reaction was stopped after 10 to 15 min, by the addition of 50 µl H₂SO₄ 2 mol/l, and the absorbance of the samples was measured at 492 nm.

HEK-293 cells were transfected to express GFP-tagged Rabs and myc–Munc13-4. Then, cells were lysed in a lysis buffer containing 50 mM Tris pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 100 µM GTPγS, 0.5% Triton X-100 and EDTA-free protease inhibitors (Roche Applied Science, USA) and ultracentrifuged at 100,000 g, for 1 hr at 4°C and 1 mg of the supernatant was incubated with anti-GFP antibodies, which had been cross-linked on protein A beads using 20 mM DMP, for 2 hr at 4°C under rotation. The beads were then washed five times with lysis buffer and once with 50 mM Tris pH 7.5, 150 mM NaCl, 0.1% Triton X-100. Bound proteins were eluted with a buffer containing 0.1% Triton X-100 in 0.1 M glycine pH 2.5.

HUVECs were grown on 11 mm collagen-type-I-coated coverslips and were transfected as described above. Indirect immunofluorescence and analysis by confocal microscopy was applied as previously described (Papanikolaou et al., 2011). Samples were viewed with a Leica TCS-SP scanning laser confocal microscope equipped with an argon laser (for excitation at 488), solid state 561 laser line and helium–neon laser (for excitation at 633), or the newer Leica TCS SP5 II scanning confocal microscope, using the Leica 63× HCX PL APO 1.3 NA objective. Images were acquired using Leica Software and files were subsequently processed in LAS AF and Adobe Photoshop. The quantification of colocalisation was done following the object-based method of analysis, using ImageJ software. The percentage of colocalisation refers to the percentage of WPBs that were positive for the corresponding Rab, out of the total number of WPBs counted in the cells.

Rab27aQ78L cloned into pLexA was used as bait to screen a placenta cDNA library. Appropriate controls and specificity tests (including tests for self-activation, growth in selective medium and β-galactosidase activity assays) led to four specific clones. All techniques were according to established protocols (Dual Systems Biotech AG).

RNA was prepared from knockdown and control siRNA-treated HUVEC using the RNeasy micro kit (Qiagen). PCR primers to Rab27a (fwd: 5′-AGCAGGGCAGGAGAGGTTTCGTA-3′, rev: 5′-TGCTATGGCTTCCTCCTCTTTCAC-3′), Rab27b (fwd: 5′-GCTCGGGAACTGGCTGACA-3′, rev: 5′-ACATTTCTTCTCTGGTGGCTTTTC-3′), Rab3a (fwd: 5′-TCGCGCTATGGGCAGAAGGAG-3′, rev: 5′-TCAGCATAGCGGAAGAGGAAGGAC-3′), Rab3b (fwd: 5′-TTCCTCTTCCGCTATGCTGATGAC-3′, rev: 5′-CCCCACGGTAATAGGCTGTTGT-3′), Rab3d (fwd: 5′-TGTGGGCATCGATTTCAAGGTCAA-3′, rev: 5′-CCCATGGCTCCCCGGTAGTAGG-3′), Rab15 (fwd: 5′-GCTGTTCCGGCTGCTGCTGAT-3′, rev: 5′-CCCGCCGATAGTACTGCTTTGTGA-3′), Rab33a (fwd: 5′-ACCTGCCTGACCTTCCGCTTCTG-3′, rev: 5′-GCCCTCGATTTCCACGGTCTTC-3′), Rab33b (fwd: 5′-GCCCGCTCCCGCATCTTCA-3′, rev: 5′-TGCGCTCCCCATCAATCTCCAC-3′) and Rab37 (fwd: 5′-CTGTGGATGGCGTGAGAGTGAAGC-3′, rev: 5′-GGCCCTGATGTTGTCGAAAGAAGA-3′) were purchased from the Microchemistry Laboratory, IMBB Crete, while primers against GAPDH were from MWG (Germany). One step quantitative RT-PCR was performed using Quantitect SYBR Green RT-PCR kit (Qiagen). The reaction was performed using the Light Cycler 2.0 (Roche). The samples were analysed on a 2% agarose gel and single products were verified by melting curve analysis. Quantification was calculated by the ΔΔCt method. Values in siRNA-treated and control-treated samples were normalised according to GAPDH.

FLAG-Rab27a wt (rat) was cloned into the NotI/XbaI site of pShuttle-CMV. Construction of recombinant adenoviruses was performed as previously described (He et al., 1998). The adenovirus expressing β-galactosidase (AdLacZ), as well as all vectors for the adenoviruses production, were kindly provided by Carol Murphy and Theodore Fotsis (BRI/Forth Ioannina, Greece). Plaque assay was employed to quantify the virus particles/µl. HUVEC were infected at 300 MOI for 4 hours. Subsequently, the infection medium was replaced with growth medium and the cells were cultured for 48 hours before the experimental assays.

Statistical significance was determined by one-way ANOVA, followed by Newman–Keuls test using the Graphpad Software.

Total protein levels in lysates were quantified by the Bio-Rad protein assay and used as a control of total cell numbers.

We thank Carol Murphy (BRI-FORTH) and Theodore Fotsis (BRI-FORTH and University of Ioannina, Greece) for the thorough discussions throughout this study, as well as for providing the adenovirus expressing β-galactosidase (AdLacZ) and the vectors for producing adenoviruses. We are also grateful to Marino Zerial (MPI-CBG, Dresden, Germany) for his ideas whenever requested. We also thank Reinhard Jahn (Max-Planck Institute, Gottingen, Germany) for the Rab3d antibodies, Haralabia Boleti (Institute Pasteur, Athens) for the anti-GFP polyclonal antibodies, Kai Simons for the GalNac-T2 expression plasmid and the University Hospital of Ioannina for providing us with the umbilical cords from healthy donors. Finally, we thank the confocal laser microscope facility of the University of Ioannina as well as the microscopy facility of BRI/FORTH for use of the Leica TCS-SP and Leica TCS SP5 scanning confocal microscopes, respectively.