RAB6 and microtubules restrict secretion to focal adhesions

To ensure their homeostasis and sustain differentiated functions, cells continuously transport diverse cargos to various cell compartments and in particular to the cell surface. Secreted proteins are transported along intracellular routes from the endoplasmic reticulum through the Golgi complex before reaching the plasma membrane along microtubule tracks. Using a synchronized secretion assay, we report here that exocytosis does not occur randomly at the cell surface but on localized hotspots juxtaposed to focal adhesions. Although microtubules are involved, the RAB6-dependent machinery plays an essential role. We observed that, irrespective of the transported cargos, most post-Golgi carriers are positive for RAB6 and that its inactivation leads to a broad reduction of protein secretion. RAB6 may thus be a general regulator of post-Golgi secretion.


INTRODUCTION
To reach the cell surface, secreted proteins are transported along intracellular routes from the endoplasmic reticulum through the Golgi complex.
Cargos exit the Golgi complex in transport carriers that use microtubules to be addressed rapidly to the plasma membrane before exocytosis. Transmembrane proteins are then exposed at the plasma membrane while soluble cargos are released in the extracellular space. Whether delivery of cargos occurs randomly or at specific sites of the plasma membrane is still unclear and the mechanisms that direct exocytosis are still unknown. Microtubules were described to be captured and stabilized by focal adhesions (Kumar et al., 2009). Their targeting to focal adhesions is driven by +TIPs (microtubule plus-end tracking proteins), such as APC, EB and CLASP, which ensure their physical contacts (Lansbergen et al., 2006;Akhmanova and Steinmetz, 2008;Kumar et al., 2012;Stehbens et al., 2014). Additionally, microtubules are linked to the actin network which is a structural component of focal adhesions (Palazzo and Gundersen, 2002).
Notably, microtubules are involved in the regulation of the distribution and dynamics of adhesion sites (Small et al., 2002;Stehbens and Wittmann, 2012;Etienne-Manneville, 2013).
CLASPs (cytoplasmic linker-associated proteins) interact at the plasma membrane with a protein complex made of LL5β, a PI3P-binding protein, and ELKS (also known as RAB6IP2). ELKS is an effector of the Golgi-associated RAB6 GTPase (Monier et al., 2002), which regulates several anterograde and retrograde trafficking pathways to and from the Golgi complex, as well as Golgi homeostasis (Goud, 1999;White et al., 1999;Grigoriev et al., 2007;Mallard, 2002). In particular, RAB6 was shown to be involved in the targeting of post-Golgi vesicles containing the secretory markers VSV-G (vesicular stomatitis virus glycoprotein, a type I transmembrane protein), and NPY (neuropeptide Y, a soluble protein) to ELKS-enriched regions of the plasma membrane (Grigoriev et al., 2011;Miserey-Lenkei et al., 2010). RAB6 has been also shown to regulate the secretion of TNFα in macrophages (Micaroni et al., 2013), and the trafficking of herpes simplex virus 1 (HSV1) (Johns et al., 2014). However, thus far, no systematic study has been performed to characterize the cargos present in RAB6-positive vesicles.
The aim of this study was to investigate the spatial organization of post-Golgi trafficking of a variety of anterograde cargos in non-polarized cells. To this end, we combined the Retention Using Selective Hooks (RUSH) assay (Boncompain et al., 2012) to synchronize anterograde transport of cargos and the Specific Protein Immobilization (SPI) assay to map precisely the sites of arrival of the cargos at the plasma membrane. We show that cargos are transported along microtubules to hotspots of secretion, which are juxtaposed to focal adhesions. Moreover, we found that RAB6-dependent post-Golgi machinery plays a key role in this process and that RAB6 could be a general regulator of post-Golgi secretion.

Exocytosis takes place in restricted areas, close to the adhesion sites
Secretion of newly-synthesized proteins along the secretory pathway occurs continuously in cells. The RUSH system offers the possibility to synchronize the intracellular transport of cargos fused to the Streptavidin-bindingpeptide (SBP) upon addition of biotin in the culture medium (Boncompain et al., 2012). With this system, it is possible to monitor a wave of secretion of a selected cargo and analyze its transport to the cell surface. Using the RUSH assay, we studied the synchronous secretion of diverse cargos: Collagen type X (ColX), VSV-G, secretory soluble EGFP (ssEGFP), gp135 (podocalyxin), CD59 and placenta alkaline phosphatase (PLAP), two GPI-anchored proteins, and tumor necrosis factor alpha (TNFα). Figure 1A illustrates RUSH-based transport monitoring using ColX as a cargo. As expected, before biotin addition, ColX was retained in the endoplasmic reticulum (ER) ( Figure 1A, 0 min). Upon biotin addition, ColX left the endoplasmic reticulum, reached the Golgi apparatus within 10 min post-release and was then exocytosed at the plasma membrane. About 35 min after biotin addition, most of ColX had been secreted into the medium and almost no signal remained in cells. Time-lapse imaging and temporal projection after Golgi exit suggested that exocytosis did not occur randomly at the cell surface but in preferred domains ( Figure 1A, Supp Movie 1). However, because ColX is a soluble secretory protein, a significant fraction of released proteins diffuses out, which may lead to underestimated levels of exocytosis at these preferred sites. To prevent its diffusion after release, we set-up an assay that we named Selective Protein Immobilization (SPI). In this assay, a GFP moiety is fused to soluble secretory factors or to the luminal part of membrane-bound cargos, and prior to seeding the cells, coverslips are coated with anti-GFP antibodies ( Figure 1B). The interaction between the coated anti-GFP antibodies and the GFP moiety fused to the cargos reduces the diffusion speed of the cargos and eventually immobilizes them. This enables the local accumulation of secreted proteins that were released over an extended period of time.
Combination of the RUSH and SPI assays thus provides a complete overview and localized history of the secretion of a selected cargo. Using SPI, and in contrast to Figure 1A, Figure 1C and Supp Movie 2 show a strong accumulation of secreted ColX visible 35 min after biotin addition. The presence of hotspots of ColX secretion confirmed that some domains of the plasma membrane seemed unable to support exocytosis while others were very active. The localization of the active domains was reminiscent of focal adhesion (FA) sites. We used cells expressing paxillin-mCherry, which localizes to FA (Turner, 1998;Turner et al., 1990), to monitor the synchronized transport of ColX combined with SPI and we found that secreted ColX was clearly enriched on FAs ( Figure 1D). A similar result was obtained for another soluble cargo, ssEGFP, although it appeared more diffuse at the plasma membrane, likely due to rapid diffusion and/or less efficient capture by the antibody ( Figure 1D).
Similar experiments were performed with membrane-bound cargos like VSV-G, gp135, TNFα, and E-cadherin adapted to the RUSH assay. Although no particular enrichment was observed in normal conditions (Supplementary Figure   1), probably due to a rapid diffusion of secreted cargos in the plane of the plasma membrane, topologically-restricted secretion was observed using SPI. As for secreted cargos, exocytosis of VSV-G and gp135 also occurred on hotspots localized to FAs ( Figure 1D). The same results were obtained when monitoring E-cadherin and TNFα secretion (data not shown).
The combination of the RUSH and SPI assays thus demonstrated the existence of secretion hotspots close to focal adhesions for soluble and membrane-bound proteins.

Exocytosis is directed between focal adhesions
Next, real-time analysis of exocytic events was carried out using total internal reflection fluorescence microscopy (TIRF). In agreement with results obtained with the SPI set-up, we detected the frequent occurrence of exocytic events in certain regions of the plasma membrane, while other zones were seemingly silent (Figures 2A, B). Moreover, dual-color TIRF microscopy revealed that exocytosis did not exactly occur on focal adhesions, but juxtaposed to them, as confirmed by proximity measurement between the secretion puffs and paxillin signal ( Figure 2C).
To explain how such a restriction of exocytosis may occur, we envisioned two non-exclusive hypotheses. On one hand, directed transport to focal adhesions may bias release toward adhesion domains. On the other hand, the factors essential to sustain exocytosis may only be present at the hotspots. To test the first hypothesis, we performed fast imaging (using a ~200 ms frame rate) of synchronized post-Golgi transport of gp135 to detect potential privileged tracks that may direct transport toward hotspots. Temporal projections revealed that gp135 en route from the Golgi complex to the cell surface used direct tracks toward focal adhesions (Figures 2D,E). Microtubules are involved in the regulation of the distribution and dynamics of adhesion sites, and can be captured and stabilized by focal adhesions (Small et al., 2002;Stehbens and Wittmann, 2012;Etienne-Manneville, 2013). Accordingly, we observed a cooccurrence of transport tracks with the microtubule network ( Figure 2F Altogether, the above results indicate that secretory vesicles use preferential and direct microtubule-based routes for transport to secretion hotspots.

ELKS and RAB6-dependent arrival of secreted cargos at exocytosis hotspots
Microtubules are attached to the cell cortex via the microtubule-stabilizing proteins CLASPs, which interact with a protein complex made of LL5β and ELKS (also known as RAB6IP2) (Lansbergen et al., 2006). We therefore investigated the presence of ELKS in secretion hotspots. As shown in Figure 3A, GFP-ELKS was enriched in zones of the plasma membrane close to FA where immobilized ColX is detected following 30-45 min biotin addition. In cells depleted for ELKS by siRNAs, the pattern of secreted proteins appeared more diffuse although secretion was still biased toward the regions near FA ( Figures 3B, C). This suggests that ELKS, along with microtubules, contributes to the targeting the secretory vesicles to hotspots.
ELKS is a RAB6 effector (Monier et al., 2002), and was shown to be involved in the docking of RAB6-positive secretory vesicles containing VSV-G and neuropeptide Y (NPY) to the plasma membrane (Grigoriev et al., 2007). In addition, the very first study on the dynamics of GFP-RAB6 in living cells reported the presence of peripheral RAB6-positive structures near focal adhesions (White et al., 1999). We thus tested whether RAB6 was also associated with the transport carriers containing the cargos tested in this study, specifically ColX, TNFα and CD59. First, we used TIRF to investigate in cells coexpressing mCherry-RAB6 and either GFP-tagged ColX, TNFα or CD59, whether RAB6-positive vesicles arriving and fusing at the plasma membrane contained these cargos. We observed that about 80% of vesicles arriving at the plasma membrane and containing one of these cargos were positive for RAB6 ( Figures 3D, E). In ELKS-depleted cells, in agreement with data reported previously (Grigoriev et al., 2007), we observed an accumulation of SBP-EGFP-CD59, TNFα-SBP-EGFP, or SBP-EGFP-ColX positive vesicles at the cell periphery as well as an increase in the total number of cytoplasmic vesicles (Figures 3F,G). Altogether, these data show that RAB6 and ELKS play a role in the docking and fusion of cargo-containing secretory vesicles with the plasma membrane.

RAB6 associates with post-Golgi carriers containing GPI-APs, TNFα and ColX
We next investigated whether RAB6 was associated with post-Golgi carriers containing CD59, TNFα or ColX en route to the plasma membrane.
Using live cell imaging, we performed a detailed analysis of the extent of colocalization between mCherry-RAB6 and several cargos during their transport To investigate when RAB6 associates with post-Golgi carriers, we carefully examined them at the exit of the Golgi. As shown in Figure 4D, about 80% of vesicles containing SBP-EGFP-CD59, TNFα-SBP-EGFP or SBP-EGFP-ColX exiting the Golgi complex were positive for RAB6. Importantly, this percentage of co-localization is similar to the 80% colocalization we found above when looking at the whole population of transport carriers.
Altogether, our results show that RAB6 is present on transport vesicles, irrespective of the transported cargo, when they leave the Golgi and remains associated with them until they reach the plasma membrane.

The RAB6 machinery is required for the secretion of GPI-APs, TNFα and ColX
To address the functional role of RAB6, we assessed the effect of siRNAmediated knockdown of RAB6 on the secretion of various cargos. A 50 % inhibition of the arrival of TNFα at the plasma membrane was observed 30 min and 60 min after biotin addition under conditions of RAB6 depletion as quantified using SPI ( Figure 5A). Similarly, in the absence of RAB6, a 50 % reduction of ColX secretion was observed 120 min after biotin addition as analyzed by western-blotting of cell lysates and culture medium ( Figure 5B). The arrival at the plasma membrane of placenta alkaline phosphatase (PLAP) was also delayed (Supplementary Figure 4A). Finally, RAB6-positive vesicles are transported from the Golgi to the plasma membrane along microtubules by the KIF5B kinesin motor (Grigoriev et al., 2007;Miserey-Lenkei et al., 2010). Accordingly, cells treated with KIF5B siRNA showed a reduced velocity of CD59-, TNFα-or ColX-positive post-Golgi vesicles ( Figure 5C).
Altogether the above results show that a variety of cargos use the RAB6 machinery for transport from Golgi to the plasma membrane. They also raise the possibility that RAB6 might be involved in the secretory process of all proteins leaving the Golgi complex. To test this hypothesis, we performed experiments on MEF cells derived from RAB6 conditional KO mice embryos (Bardin et al., 2015) and global protein secretion was monitored using the SUnSET assay (Schmidt et al., 2009) Figure 5D). Importantly, after 4 h and 5.5 h of chase, the amount of protein secreted by cells expressing or not RAB6 was similar ( Figure 5D). These results thus showed that RAB6 depletion does not block the secretory process but leads to a delay in total protein secretion, as previously found for exogenous cargos (Grigoriev et al., 2007;Miserey-Lenkei et al., 2010).
Altogether, the above results show that RAB6 function is required for total protein secretion at the cell level.

RAB6 is not involved in sorting of cargos at the exit of Golgi complex
Newly synthesized proteins are thought to be sorted into distinct populations of transport carriers at the TGN (De Matteis and Luini, 2008). The RUSH system allows to synchronize transport of two (or more) cargos at the same time, thus providing a powerful approach to investigate sorting processes in further detail. We imaged cells expressing two cargos, either SBP-EGFP-ColX and TNFα-SBP-mCherry, SBP-EGFP-ColX and SBP-mCherry-CD59, or TNFα-SBP-EGFP and SBP-mCherry-CD59 ( Figure 6A). In all cases, although a majority (60%) of the vesicles contained two cargos, a large fraction of vesicles contained only one cargo. This indicates that, despite the sudden wave of transport imposed by biotin-induced release from the ER, efficient sorting can still occur at the Golgi complex. We then estimated the percentage of co-localization between RAB6 and post-Golgi vesicles containing one or two cargos. As illustrated in Figure 6B in the case of TNFα and ColX, the majority (60%) of vesicles containing the two cargos was positive for endogenous RAB6.
Importantly, similar percentages of vesicles containing only one cargo were positive for RAB6 (50% and 60% of vesicles containing ColX or TNFα, respectively) ( Figure 6B). This suggests that RAB6 is not involved in sorting of cargos at the exit of the Golgi complex but associates to transport carriers, irrespective of the transported cargo, to target them to specific sites at the plasma membrane.

DISCUSSION
Cells need to continuously control the transport of various proteins to the cell surface both to ensure homeostasis and to sustain differentiated functions.
The study of various transport steps, like ER to Golgi or Golgi to ER, have shown that mechanisms are at work which enable the transport of a diversity of proteins using a "universal" core machinery (COPI, COPII for example). However, whether Golgi to plasma membrane is similarly controlled by a core machinery was still unclear. One of the main findings of this study is that post-Golgi transport of diverse secretory cargos, of various shape, function and characterized by diverse transport kinetics are handled by a common machinery and reach the membrane in similarly restricted domains.
The concept of "exocytosis hotspots" existing in cells was actually proposed almost 40 years ago by D. Louvard (Louvard, 1980) when analyzing the secretion of fibronectin after cell attachment (Heggeness et al., 1978).
Surprisingly, we found that the secretion of cargos recently released from the Golgi complex occurs in so called "hotspots" close to focal adhesions, and that this is a general process common to several and diverse proteins.
Targeting to focal adhesions does not seem to depend on the particular function or modification of the cargo because it was observed for glycosylated (gp135 and Collagen X) and non-glycosylated proteins (TNFα). This was also observed for soluble secretory GFP, which is non-glycosylated and unlikely to bear any specific transport signal. The secretion of particular cargo types, like metalloproteinases such as MT1-MMP, at adhesion sites was previously proposed (Stehbens et al., 2014;Bravo-Cordero et al., 2007). However, it was not clear if the vesicles containing MT1-MMP were strictly secretory vesicles or vesicles recycled from the endocytic/recycling pathways. Here, by synchronizing the anterograde transport of cargos using the RUSH assay, we unambiguously analyzed the first wave of secretion of transport carriers. The SPI assay that we set-up for capture of secreted proteins prevented transmembrane diffusion in the plasma membrane or release of soluble cargos into the extracellular space. The SPI provides insight into the secretion history of synchronized cargos. By combining these two assays, we were able to show that transport carriers use preferential microtubule tracks to reach secretion hotspots. In addition to this directed movement, the exocytic zones may be restricted close to focal adhesions by the presence of factors necessary for docking/fusion of transport carriers such as ELKS and LL5β proteins (Grigoriev et al., 2011). An important finding of this study is that RAB6 is likely a major regulator of post-Golgi secretion (Figure 7). RAB6 was previously shown to be present on post-Golgi transport carriers containing VSV-G, NPY and TNFα (Grigoriev et al., 2007;Miserey-Lenkei et al., 2010;Micaroni et al., 2013;Johns et al., 2014). Here we show that RAB6 associates with secretory vesicles containing a variety of exogenous cargos and that its depletion delays their arrival at the plasma membrane. The observation that the secretion of most endogenous proteins synthesized by MEF cells is affected by RAB6 depletion further suggests that RAB6 may be present on the majority of post-Golgi vesicles transporting cargos to the plasma membrane. This is in agreement with a model proposing that one of the main functions of RAB6 is to target post-Golgi vesicles to cortical ELKScontaining patches that define secretion domains (Grigoriev et al., 2007;2011).
ELKS may not only dock RAB6-positive vesicles to the plasma membrane but also to other compartments. We have recently shown that ELKS is present on the membrane of mature melanosomes in melanocytes, which allows the diversion of part of the RAB6-dependent secretory pathway to melanosomes, enabling direct transport of a subset of melanosomal enzymes from Golgi (Patwardhan et al., 2017).
It is likely that RAB6 plays another role beyond its function in targeting post-Golgi vesicles to secretion hotspots. We have previously shown that RAB6 not only controls the fission of its own transport carriers (Miserey-Lenkei et al., 2010) but also could select a subset of MTs originating from Golgi membranes for their transport . Although not directly proven, it is likely that this subset of MTs correspond to the ones identified in this study which are used several times for targeting cargos to secretion hotspots.
Evidence exist that several classes of secretory carriers can form at the trans-Golgi, indicating that cargos are sorted prior to be delivered to the plasma membrane (De Matteis and Luini, 2008). Our study highlights that, even when pulsing a massive amount of cargos in the secretory pathway, the Golgi is still able to properly sort a large percent of cargos in dedicated transport intermediates. Such sorting may be even more important in polarized cells, such as epithelial or neuronal cells, or when studying cargos targeted to particular intra-cellular compartments like the endo-lysosomal system. Such an intra-Golgi sorting is also evident when monitoring the time of residence in the Golgi and exit from the Golgi of cargo. Some cargos exit very quickly while others stay for a long time in the Golgi complex, suggesting intra-Golgi segregation. However, our data indicate that irrespective of the cargo, the vast majority of transport intermediates are positive for RAB6. This suggests that RAB6 is not involved in cargo sorting but belongs to a general machinery responsible for fission of

Biochemical reagents
The following reagents were used: Para-nitroblebbistatin (Optopharma Ltd), 4-OH tamoxifen (Sigma). SiR-tubulin (Spirochrome) was used to label microtubule in living cells according to manufacturer's instructions.

DNA and RNA transfection
HeLa or RPE-1 cells were transfected 24 to 48 h before observation with calcium phosphate (Jordan et al., 1996)

Specific protein immobilization assay (SPI)
Coverslips were incubated in sterile conditions in a bicarbonate solution (0,1 M, pH 9,5, 1 h, 37°C) followed by incubation in poly-L lysine 0.01% (diluted in water, 1 h, 37°C, Sigma). Coverslips were then washed in PBS and dried before antibody incubation (diluted in the bicarbonate buffer) 3 h at 37°C (or overnight). Coverslips were washed twice in PBS and then with cell medium before cells were seeded. Antibodies used for coating in this study were: anti-

Live cell imaging
Spinning-disk confocal time-lapse imaging were done at 37°C in a thermostat-controlled chamber using an Eclipse 80i microscope (Nikon) equipped with spinning disk confocal head (Perkin), a 100x objective and either a Ultra897 iXon camera (Andor) or CoolSnapHQ2 camera (Roper Scientific). Fixed sample were on the other hand imaged using a 60x objective using the same setup. TIRF microscopy was done using Leibovitz's medium (Life Technologies) at 37°C in a thermostat-controlled chamber. An Eclipse Ti inverted microscopes (Nikon) equipped with either a TIRF module (Nikon) or an iLAS2 azimuthal TIRF module (Roper Scientific), a 100x TIRF objective, a beam splitter (Roper Scientific) and an Evolve 512 EMCCD camera (Photometrics) was used in this case (Boulanger et al., 2014). All acquisitions were driven by Metamorph (Molecular Devices).

Fluorescent activated cell sorting (FACS)
Cells were incubated at 4°C with an anti-GFP antibody diluted in PBS (Recombinant protein platform, Institut Curie, 1:400). Intra-and extra-cellular EGFP intensity were detected with a BD Accuri TM C6 Cytometer. The ratio between the cell surface signal intensity per the total signal intensity detected in EGFP-PLAP expressing cells measured at 30 and 60 minutes was normalized to the cell surface signal intensity detected at 0 minute (corresponding to background) for each condition.

SUnSET assay (SUrface SEnsing of Translation)
SUnSET assay was carried out as described previously (Schmidt et al., 2009). Briefly, MEF cells were seeded at a confluence of 30 % then treated with ethanol (control) or 1 µM of 4-OH Tamoxifen for 96 h to deplete RAB6. The day of the experiment, cells were cultured without serum and incubated for 30 min at 37 °C with 10 µg/ml of puromycin. Puromycin was then chased at different time points (0, 1, 2, 4, and 5.5 h). The supernatants and the whole cell lysates were collected and processed for western-blotting. The puromycin signal was revealed using an anti-puromycin antibody (clone 12D10 Millipore MABE343).
Quantification of puromycin intensity in the supernatant and the whole cell lysis from four independent experiment using Image Lab software (Bio-Rad) is shown.
The amount of total secreted proteins was determined by the normalization of the intensity of puromycin signal in the supernatant for each time point by the sum of the intensity of puromycin signal in the supernatant and the whole cell lysate for each corresponding time point.

Cell lysis, SDS-PAGE, and Western-blotting
Cells were washed three times in ice-cold PBS, scraped and then lysed in a buffer containing 150 mM NaCl, 50 mM Tris-HCl pH 7.5, 1% Nonidet-P40 (Sigma). Protein concentrations were determined by Quick Start™ Bradford 1x Dye Reagent (Bio-Rad). Equal amounts of proteins were reduced with 1x loading buffer containing 6% β-mercaptoethanol and resolved on 10% SDS-PAGE.
Proteins were transferred onto nitrocellulose Protran BA 83 membrane (Life science), processing for immunoblotting. HRP-conjugated secondary antibodies associated signal was detected with ECL system and an enhanced chemiluminescence system (ChemiDoc Touch System, Bio-Rad). Quantification of the corresponding signal obtained was done using Image lab software.
ColX secretion following RAB6 depletion was measured using westernblotting. After release of the cargo from the ER for different time points, trafficking was stopped by incubating the cells at 4°C and the supernatant was collected.
Samples were then processed for western-blotting.

Image analysis and quantification
For Figures 1A, 2D, 2E Figure 6B) in two different channels was determined manually using Image J software (Synchronize windows). In Figure   S3B, colocalization was measured with Pearson's coefficient using Image J software. Quantification of the velocity and the track length (the covered distance) of vesicles were done with the plug-in "Manual Tracking" from ImageJ software developed by Fabrice Cordelières (Institut Curie). The number of vesicles was determined using Image J software, using the "find maxima" plug-in.
The number of vesicles was normalized either per cell or per area in each cell.

Statistical analysis
All data were generated from cells pooled from at least 3 independent experiments represented as (n), unless mentioned, in corresponding legends.
Statistical data were presented as means ± standard error of the mean (S.E.M.).
Statistical significance was determined by Student's t-test for two or three sets of data using Excel, no sample was excluded. Cells were randomly selected. Only P-value <0.05 was considered as statistically significant.    HeLa-SBP-EGFP-ColX treated or not with RAB6 siRNAs were incubated for 30, 60 or 120 min with biotin to allow cargo release from the endoplasmic reticulum.
The amount of secreted cargo was revealed by western-blotting using an anti-GFP antibody. Beta-tubulin was used as a loading control and as a non-secreted protein. Representative immunoblots are displayed. Quantification of the amount of ColX present at the plasma membrane in cells treated as indicated above (mean ± SEM, n= 3). * p< 0.05 (Student's t-test). (C) HeLa cells expressing SBP-EGFP-CD59, TNFα-SBP-EGFP and SBP-EGFP-ColX were treated for 3 days with control, RAB6 of KIF-5B siRNAs. Cells were incubated for 30 min with biotin to allow cargo release from the ER. Cells were imaged using a time-lapse spinning disk confocal microscope. Vesicles velocity (µm/sec) and the number of vesicles per cell were quantified using ImageJ (mean ± SEM, n= 17-35 cells). *** p< 10 -4 , ** p< 0,005, * p< 0.05 (Student's t-test). (D) The SUnSET assay was used to determine the effect of RAB6 depletion on total protein secretion. Mouse embryonic fibroblasts (MEFs) prepared from RAB6 loxP/Ko Rosa26CreERT2-TG embryos (described in (Bardin et al., 2015)) (named after RAB6-KO) were treated with ethanol or with 4-hydroxytamoxifen (4-OHT) for 96 h to induce RAB6 depletion. Cells were then incubated with puromycin and chased in puromycinfree medium for 0, 1, 2, 4, 5.5 h. Total protein content in cell lysis or supernatant were labelled with an anti-puromycin antibody. Quantification of puromycin intensity in the supernatant and the whole cell lysis using Image Lab software   Vesicles are transported to the plasma membrane in a KIF5B-dependent manner (Grigoriev et al., 2007;Miserey-Lenkei et al., 2010) and ELKS ensures their docking to exocytic hotspots close to focal adhesions (Grigoriev et al., 2007;2011).