Rab11 regulates exocytosis of recycling vesicles at the plasma membrane

Cells internalize extracellular materials, plasma membrane (PM) proteins and their ligands by endocytosis. Some of endocytosed proteins recycle back to the cell surface for reuse, whereas others are destined for degradation in lysosomes. The most extensively investigated recycling protein is transferrin receptor (TfnR) (Grant and Donaldson, 2009; Maxfield and McGraw, 2004; Mukherjee et al., 1997). It binds diferric transferrin (Tfn) on the cell surface, and is internalized via clathrin-coated vesicles and delivered to early/sorting endosomes. Subsequently, the Tfn–TfnR complex is returned to the PM either directly (rapid recycling) or indirectly via recycling endosomes (REs) (slow recycling). REs, whose subcellular localization varies among cell types, are often located near the nucleus or centrosome and consequently referred to as perinuclear or pericentrosomal REs (Grant and Donaldson, 2009; Mukherjee et al., 1997).

Rab family small GTPases regulate various aspects of membrane traffic through interactions with their effector proteins (Schwartz et al., 2007; Stenmark, 2009). Among them, Rab11 is one of extensively studied Rab GTPases. Rab11 associates primarily with REs and regulates recycling of endocytosed proteins; it has therefore been used as an RE marker in a number of studies (Grant and Donaldson, 2009; Stenmark, 2009; van Ijzendoorn, 2006). However, the recycling step that Rab11 regulates remains poorly understood: For example, previous studies showed that expression of a dominant-negative Rab11a mutant, Rab11a(S25N), inhibited release of endocytosed ¹²⁵I-labelled Tfn to the extracellular medium (Ren et al., 1998; Ullrich et al., 1996) and induced tubulation of compartments containing endocytosed Tfn and TfnR (Hölttä-Vuori et al., 2002; Wilcke et al., 2000). However, the relationship between inhibition of Tfn recycling and tubulation of TfnR-containing compartments has remained unclear; In previous studies with Drosophila cells, mutation or inhibition of Rab11 and the exocyst tethering complex was reported to cause intracellular accumulation of DE-cadherin and Delta, yet it was not determined where these proteins were accumulated (Guichard et al., 2010; Langevin et al., 2005). On the other hand, Rab11 was reported to transiently associate with fusion sites of Fc receptor–containing vesicles with the PM, but it is not yet known whether the Rab11 association with fusion sites is critical for the fusion event (Ward et al., 2005).

Over the course of our experiments, we noticed that, in addition to their distribution in the perinuclear region, endogenous Rab11 and TfnR are found in the peripheral region, particularly around the cellular tips, albeit at a relatively low frequency. In this study, we extend this finding to show that Rab11 not only associates with perinuclear REs, but also participates in exocytosis of recycling vesicles at the PM in concert with the exocyst tethering complex.

In HeLa cells, vesicular RE structures visualized by immunostaining for Rab11 and TfnR are found throughout the cytoplasm, relatively concentrated in the perinuclear region. However, we noticed that Rab11 and TfnR are also found in the peripheral region, particularly around the cellular tips (Fig. 1A). By time-lapse analysis of cells expressing EGFP-tagged Rab11a, we observed vesicular structures throughout the cytoplasm, relatively concentrated in the tip regions, and often moving in an intermittent manner that is suggestive of microtubule-based transport (Fig. 1B; supplementary material Movie 1). When the EGFP–Rab11a signals around the cellular tips were photobleached, the signals were gradually recovered and centrifugal movement of EGFP–Rab11a-positive vesicular structures was often observed in the photobleached area (Fig. 1C; supplementary material Movie 2), suggesting that the peripheral Rab11-positive intermediates are delivered, at least in part, from perinuclear REs.

We next compared localization of Rab11 against organelle marker proteins (supplementary material Fig. S1). Neither EEA1 (early endosomes), Lamp-1 (late endosomes/lysosomes) nor GM130 (Golgi) was significantly colocalized with Rab11. In contrast, EGFP-tagged DMT1-II (type II isoform of divalent metal transporter 1) was extensively colocalized with Rab11 in the tip regions; DMT1-II is localized primarily to REs, and is responsible for transport across endosomal membranes of iron internalized via the Tfn–TfnR complex (Lam-Yuk-Tseung and Gros, 2006; Tabuchi et al., 2002). Taken together, these observations indicate that peripheral Rab11-positive structures represent REs or carrier intermediates originating from REs.

We then set out to investigate the effects of Rab11 depletion on perinuclear and peripheral localization of TfnR. In mammals there are two Rab11 isoforms, both of which are expressed in HeLa cells (Fig. 2A). Using a previously characterized double-stranded siRNA (Junutula et al., 2004; Takahashi et al., 2011), we could almost completely deplete Rab11a in HeLa cells (Fig. 2B, top panel). Because of a lack of Rab11b-specific antibody, we were unable to directly measure depletion of Rab11b; however, we believe that Rab11b is also specifically depleted, on the basis of data obtained using siRNAs for Rab11a and Rab11b in combination with Rab11a-specific antibody (top panel) and an antibody that recognizes both Rab11 isoforms (middle panel).

When morphologically examined, Rab11b knockdown did not alter TfnR-positive structures (Fig. 2C, compare column 3 with column 1; Fig. 2D), whereas Rab11a depletion had a subtle, but significant, effect (Fig. 2C, column 2; Fig. 2D); TfnR was often found on tubules emanating from perinuclear structures and tended to accumulate around the cellular tips. Depletion of both Rab11 isoforms had the most dramatic impact (Fig. 2C, column 4; Fig. 2D); the double knockdown reduced the number of perinuclear vesicular structures positive for TfnR while inducing tubules reminiscent of those observed in cells expressing dominant-negative Rab11a (Hölttä-Vuori et al., 2002; Wilcke et al., 2000) (see below). Furthermore, simultaneous depletion of both Rab11 isoforms enhanced TfnR accumulation in the tip regions.

We next evaluated the impact of Rab11 depletion on trafficking through REs. To this end, we treated HeLa cells with Alexa-Fluor-555-conjugated Tfn at 4°C to allow binding to surface TfnR; after washing out excess fluorescent Tfn, we incubated the cells at 37°C to follow Tfn trafficking. In control cells, after 2.5–10 min incubation at 37°C, fluorescent Tfn was found on punctate structures positive for EEA1 (supplementary material Fig. S2A) and TfnR (Fig. 3A) that represent early and recycling endosomes. After 20 min, however, fluorescent Tfn became barely detectable within the cells, indicating release of Tfn molecules to the extracellular medium. In cells depleted of both Rab11 isoforms (Fig. 3B), fluorescent Tfn and TfnR were predominantly found around the tips after 10–30 min at 37°C. After 60 min, however, fluorescent Tfn disappeared almost completely from the cells, indicating that Rab11 knockdown delays, but does not completely block, Tfn release. A significant fraction of fluorescent Tfn reached EEA1-positive punctate structures after 5–10 min incubation at 37°C, indicating that endocytosis of Tfn can occur under the Rab11-depleted conditions (supplementary material Fig. S2B). Moreover, after 10–20 min incubation, Tfn was accumulated at the cell tips where EEA1 was not found. It suggests that the internalized Tfn have left EEA1-positive early endosomes after 10–20 min incubation and thus the accumulation of Tfn at the cell tips are due to the exocytic but not to endocytic defect. Similarly to Rab11 knockdown, expression of dominant-negative Rab11a, Rab11a(S25N), caused accumulation of TfnR in the tip regions and delayed Tfn release as compared with cells expressing Rab11a(WT) (supplementary material Fig. S3A,B).

To further support that the exocytic event of internalized Tfn is suppressed in the Rab11-knockdown cells, we performed another experiment to chase internalized Tfn. Control HeLa cells or those treated with siRNAs for Rab11a and Rab11b were first incubated with Alexa-Fluor-488-conjugated Tfn at 37°C for 5 min, and, after thoroughly washing out surface-bound fluorescent Tfn with acid, further incubated at 37°C to chase internalized Tfn (supplementary material Fig. S4). Just after the 5 min pulse, fluorescent Tfn was found in punctate structures distributed throughout the cytoplasm in both the control and Rab11-knockdown cells (0 min chase), indicating that Tfn is normally internalized in the Rab11-knockdown cells. However, after 10 min chase at 37°C, internalized Tfn was considerably accumulated at the cell periphery in the Rab11-knockdown cells, whereas a significant fraction of internalized Tfn disappeared from the cells in the control cells. A similar trend was observed after 30 min chase. Thus, it is likely that the Rab11 depletion retards the exocytic process of internalized Tfn, but does not significantly affect the internalization.

Next, we asked whether the Tfn–TfnR complex that accumulates at the cell periphery is on or beneath the PM. For this purpose, cells treated with siRNAs for Rab11a+Rab11b were incubated with Alexa-Fluor-555–Tfn and immunostained with anti-TfnR antibody, which recognizes the exoplasmic/lumenal region of the TfnR protein, under either non-permeabilized or permeabilized conditions. If TfnR is on the cell surface, it will be detected under both conditions; whereas if it is in cytoplasmic vesicles it will be detected only under permeabilized conditions. In Rab11-knockdown cells, TfnR was detected in the tip regions only under permeabilized conditions (Fig. 3C), indicating accumulation of TfnR beneath the PM. This observation was confirmed by electron microscopy. We incubated Rab11-knockdown cells with horseradish peroxidase (HRP)-conjugated Tfn, and after allowing its endocytosis for 20 min, treated the cells with diaminobenzidine and H₂O₂. Under these conditions, clusters of Tfn-positive vesicles were found beneath the PM (Fig. 3D–F). Furthermore, Tfn-positive vesicles were often found along microtubule-like linear structures (Fig. 3E, yellow arrowheads). In the control cells, peripheral accumulation of Tfn-positive structures was less evident after 5 or 20 min endocytosis of HRP–Tfn (supplementary material Fig. S5, compare A and B), in agreement with the light microscopic data (Fig. 3A,B).

Despite the drastic morphological change in TfnR-positive compartments by Rab11 depletion, the immunofluorescence and electron microscopic data (Fig. 3B–F) suggested that recycling carriers move toward the cell periphery and accumulate there. Both TfnR-positive vesicles in control cells and tubules in Rab11-knockdown cells were frequently found along microtubules (Fig. 3E, yellow arrowheads; Fig. 4A) suggesting that recycling carriers move along microtubules independently of Rab11. In addition, when TfnR–EGFP-expressing cells had been knocked down of Rab11 and recovery of fluorescence signals was followed by time-lapse imaging after photobleaching the peripheral TfnR–EGFP signals, tubular intermediates that appeared to extend from a proximal region gave rise to gradual recovery of peripheral signals (Fig. 4B; supplementary material Movie 3). To examine whether peripheral accumulation of TfnR depends on microtubules, we treated cells with nocodazole and, after removal of the drug, followed the change in localization of TfnR (Fig. 4C–E). When control cells were treated with nocodazole, neither Rab11 (Fig. 4C) nor TfnR (Fig. 4D) was found in the peripheral region (0 min panels). After nocodazole washout, however, both Rab11 and TfnR gradually accumulated around the cellular tips concomitant with microtubule reconstruction depicted by β-tubulin immunostaining (5–20 min panels). A similar time course of TfnR accumulation after nocodazole washout was observed in Rab11-depleted cells (Fig. 4E); a slight difference was that short tubular structures positive for TfnR were prominent during the recovery (10 min panel) in the Rab11-depleted cells (Fig. 4E) as compared with control cells (Fig. 4D). These observations, together with those in Fig. 1B and Fig. 3E and supplementary material Movie 1, support transport of Rab11- and TfnR-positive carriers along microtubules towards the cell periphery; however, the association with and transport along microtubules are independent of Rab11, albeit the drastic morphological change of the carriers in the absence of Rab11.

How does Rab11 perturbation cause considerable accumulation of Tfn–TfnR-positive carrier intermediates beneath the PM? One possible explanation is that the final exocytic events are inhibited by loss of function of Rab11. To address this possibility, we applied total internal reflection fluorescence microscopy (TIRFM) to cells expressing TfnR–EGFP. In control cells, exocytic events of vesicles containing TfnR–EGFP were frequently observed, whereas lateral movement of these vesicles was infrequent (Fig. 5A; supplementary material Movie 4). In stark contrast, exocytic events were barely detectable in cells depleted of both Rab11 isoforms, and lateral movement of vesicular and tubular intermediates, probably along microtubules, was prominent (Fig. 5B; supplementary material Movie 5). Quantification of exocytic events revealed that exocytosis of TfnR–EGFP is ∼20-fold less frequent in Rab11-knockdown cells than in control cells (Fig. 5F; 0.70±0.33 vs 13.8±3.1 exocytic events/cell/min). Thus, in Rab11-depleted cells, TfnR-containing vesicles appear to have no choice but to move laterally, due to their inability to find target membranes.

When cells expressing TfnR–EGFP and mCherry–Rab11a were subjected to dual-color TIRFM, significant populations of puncta observed are positive for both TfnR and Rab11a (Fig. 5C), indicating that TfnR and Rab11 are colocalized at the PM and/or just beneath the PM. This observation, together with the accumulation of TfnR-positive intermediates beneath the PM in the Rab11-knockdown cells, makes it possible that Rab11 participates in tethering and fusion of recycling vesicles with the PM. In the context of the exocytic event involving Rab11, we then focused on the exocyst tethering complex (He and Guo, 2009; Munson and Novick, 2006), because the exocyst has been implicated in Tfn recycling (Oztan et al., 2007; Prigent et al., 2003), and because Sec15, one subunit of the octameric protein complex, directly interacts with Rab11 (Wu et al., 2005; Zhang et al., 2004). In mammals, there are two Sec15 isoforms, Sec15A and Sec15B; we confirmed by RT-PCR that HeLa cells express both Sec15 isoforms (supplementary material Fig. S6A). Via immunoblotting using an available monoclonal anti-Sec15 antibody (Wang and Hsu, 2003), we detected three closely apposed bands in cell lysates (supplementary material Fig. S6B). Treatment of cells with a pool of siRNAs directed against Sec15A or Sec15B mRNA abolished the bottom or middle band, respectively; treating cells with both siRNA pools abolished the lower two bands (supplementary material Fig. S6B), indicating that the bottom and middle bands represent the Sec15A and Sec15B proteins, respectively; the top band may be non-specific. When HeLa cells were depleted of either Sec15A or Sec15B alone, Tfn–TfnR recycling was marginally affected (supplementary material Fig. S6C, columns 2 and 3). In contrast, simultaneous knockdown of both Sec15 isoforms considerably delayed Tfn–TfnR recycling and accumulation of recycled Tfn and TfnR in the tip regions (Fig. 6B; supplementary material Fig. S6C, right column). When examined by TIRFM, TfnR–EGFP-positive carriers barely fused with the PM in cells treated with both Sec15A and Sec15B siRNAs (Fig. 5D; supplementary material Movie 6). Quantification of the exocytic events revealed that the exocytic frequency of TfnR–EGFP-positive carriers in Sec15-knockdown cells is approximately one-seventh of that in control cells (Fig. 5F; 1.87±0.32 vs 13.8±3.1 exocytic events/cell/min).

These observations suggest that the exocyst participates in the final exocytic events, i.e. tethering and fusion of carrier vesicles, downstream of Rab11. Attempting to exclude potential off-target effects of the Sec15 siRNAs, we examined whether exogenously expressed Sec15 could rescue the Sec15-knockdown phenotypes. However, exogenously expressed Sec15 formed aggregates in the cytoplasm and appeared to disturb Tfn recycling (data not shown), probably because the excess Sec15 protein could not form a functional exocyst complex together with other endogenous subunits. To circumvent this problem, and investigate whether the exocyst complex participates en bloc in the exocytic event, we turned to another subunit, Exo70, chosen because an available anti-Exo70 antibody (Vega and Hsu, 2001) works well in both immunoblotting and immunofluorescence. Exo70 was significantly colocalized with Rab11 in the peripheral region (Fig. 7A, left column). Moreover, we found that Exo70 was also colocalized with Rab11 in the perinuclear region (Fig. 7A, right column). This was rather unexpected, because several studies reported that the exocyst components, including Exo70, associate mainly with the PM through interacting with PtdIns(4,5)P₂ in yeast and mammalian cells (He and Guo, 2009; He et al., 2007; Liu et al., 2007), although localization of the exocyst to Tfn-positive endosomes was also reported (Oztan et al., 2007). We therefore examined whether Rab11 depletion influences the Exo70 localization and found that depletion of Rab11a and Rab11b redistributed Exo70 from the perinuclear region to the cytoplasm (Fig. 7C, middle column, cells indicated by asterisks), although apparently did not alter the cellular level of the Exo70 protein (Fig. 7B); this observation indicates that Rab11 is involved in membrane recruitment of the exocyst at perinuclear REs. On the other hand, depletion of Exo70 changed Rab11 localization; specifically, Rab11 was significantly accumulated in the tip regions (Fig. 7C, right column). Furthermore, as with Rab11 depletion, Exo70 depletion resulted in accumulation of recycled Tfn and TfnR at the cell periphery (Fig. 6C) and a decrease in the exocytic frequency of TfnR-positive vesicles (Fig. 5E,F; supplementary material Movie 7). These observations together indicate that Rab11 and probably the exocyst are transported from perinuclear REs through associating with recycling vesicles and participate in tethering and fusion of these vesicles with the PM.

We have shown here that Rab11 is involved in regulation of the final exocytic event of recycling carriers at the PM, although it is well known to associate with perinuclear REs. Rab11 perturbation (siRNA-mediated depletion of endogenous Rab11 or expression of a dominant-negative mutant) results in accumulation of vesicles containing endocytosed Tfn and TfnR at the cell periphery, particularly in cellular tip regions. The recycling carriers appear to be transported along microtubules from the perinuclear to peripheral regions. The TIRFM data unequivocally demonstrate that Rab11 depletion inhibits exocytic events of recycling vesicles. Rab11 depletion also results in redistribution of Exo70 to the cytoplasm, suggesting that membrane association of the exocyst is, at least in part, under the regulation of Rab11; the exocyst may be recruited onto membranes at perinuclear REs and transported to the cell periphery together with Rab11. As predicted from the interaction of Rab11 with the exocyst, knockdown of Sec15 or Exo70 leads to essentially the same phenotypes (accumulation of recycling carriers beneath the PM) as those exhibited by Rab11-knockdown cells, as well as peripheral accumulation of Rab11. The most plausible explanation for these observations is that inhibition of tethering of Rab11-positive vesicles results in vesicle accumulation beneath the PM; this is consistent with previous studies showing that Rab11 accumulated in the apical portion of epithelial cells in Drosophila exocyst mutants (Langevin et al., 2005), and that Rab11 diffused from vesicle-like dots into the PM, concurrent with exocytosis of Fc receptor–positive vesicles (Ward et al., 2005). Thus, the role of Rab11 at the PM is consistent with the general view of the roles of Rab GTPases in membrane tethering and fusion (Zerial and McBride, 2001).

What is the role of Rab11 at perinuclear REs? The TfnR-positive tubules observed in Rab11-perturbed cells may result from failure of fission of vesicles budding from REs. Thus, in addition to its role in tethering of recycling vesicles with the PM, Rab11 may also participate in vesicle formation from REs. Despite their roles in membrane tethering and fusion, several Rab GTPases have been implicated in controlling vesicle budding (Stenmark, 2009). In particular, the yeast counterparts of Rab11, Ypt31 and Ypt32, are required for vesicle budding from the trans-Golgi network (Jedd et al., 1997). In this context, it is noteworthy that knockdown of Arf family small GTPases, in particular, simultaneous knockdown of class I Arfs (Arf1 and Arf3), causes tubulation of Tfn-positive compartments and delay of Tfn recycling (Volpicelli-Daley et al., 2005); vesicle budding is well known to be regulated by Arfs (D'Souza-Schorey and Chavrier, 2006). It is therefore tempting to speculate that Arfs and Rab11 function along the same sequence of a recycling event from REs. Despite the dramatic tubulation of TfnR-positive REs in Rab11-depleted cells, recycling carriers containing Tfn and TfnR were still delivered to the cell periphery in a microtubule-dependent manner (Fig. 4E). Thus, although Rab11 is required for tethering/fusion of recycling carriers with the PM as shown by the TIRFM analysis (Fig. 5B), it is dispensable for the association of recycling carriers with microtubules.

Taken together, we propose the following model for Rab11 functions in Tfn–TfnR recycling. (1) Rab11 participates in the exit of recycling vesicles from perinuclear REs. (2) Recycling vesicles are transported along microtubules towards the cell periphery, although Rab11 per se is not required for microtubule association of recycling intermediates. (3) Finally, Rab11 regulates tethering of recycling vesicles with the PM in concert with the exocyst. In the future, it will be important to identify the SNARE proteins that are involved in the final fusion of recycling vesicles with the PM.

Monoclonal mouse anti-Exo70 and anti-Sec15 antibodies were kind gifts from Shu-Chan Hsu (Rutgers University) and Wei Guo (University of Pennsylvania), respectively (Vega and Hsu, 2001; Wang and Hsu, 2003). Polyclonal rabbit anti-Rab11 antibody, which reacts with Rab11a, and monoclonal mouse anti-TfnR antibody (H68.4), which recognizes a cytoplasmic epitope, were purchased from Zymed (South San Francisco, CA). Monoclonal mouse anti-TfnR/CD71 antibody, which recognizes an exoplasmic/lumenal epitope, was from Sigma-Aldrich (St Louis, MO). Monoclonal mouse anti-Rab11 antibody, which reacts with both Rab11a and Rab11b, and monoclonal mouse antibodies to Rab5, EEA1, Lamp-1 and GM130 were from BD Biosciences (Franklin Lakes, NJ). Monoclonal mouse antibodies to β-tubulin and β-actin were from Millipore (Billerica, MA). Monoclonal rat anti-HA antibody was from Roche Applied Science (Indianapolis, IN). Alexa-Fluor- and HRP-conjugated secondary antibodies were from Molecular Probes (Carlsbad, CA) and Jackson ImmunoResearch Laboratories (West Grove, PA), respectively. Alexa-Fluor- and HRP-conjugated Tfn were from Molecular Probes and Jackson ImmunoResearch Laboratories, respectively. Nocodazole was from Sigma-Aldrich.

An expression vector encoding human DMT1-II tagged with EGFP (Tabuchi et al., 2002) was a kind gift from Mitsuaki Tabuchi (Kagawa University, Japan). N-terminally HA- and EGFP-tagged Rab11a expression vectors were constructed by subcloning a cDNA fragment containing the coding sequence of human Rab11a into pcDNA3-HAN (Shin et al., 1997) and pEGFP-C1 (Clontech, Mountain View, CA), respectively. Construction of an expression vector for N-terminally mCherry-tagged Rab11a was described previously (Takahashi et al., 2011). The S25N mutation was introduced into the Rab11a cDNA using a Quik Change Site-Directed Mutagenesis kit (Agilent Technologies, Santa Clara, CA).

Plasmids for production of replication-defective, self-inactivating lentiviral vectors, pRRLsinPPT, and packaging plasmids (pRSV-REV, pMD2.g and pMDLg/pRRE) (Thomas et al., 2009) were kindly provided by Peter McPherson (McGill University, Canada). A destination cassette from pcDNA3.2/V5-DEST (Invitrogen, Carlsbad, CA) was inserted into pRRLsinPPT in order to convert it to a Gateway system destination vector; the resultant plasmid was named pRRLsinPPT-DEST. A plasmid vector for lentiviral production of human TfnR tagged with MEF (Myc-TEV-Flag; a kind gift from Toshiaki Isobe, Tokyo Metropolitan University) (Ichimura et al., 2005) and EGFP was constructed by subcloning a DNA fragment encoding MEF-TfnR-EGFP into pENTR3C-dual (Invitrogen); the resulting gene was then transferred into pRRLsinPPT-DEST using the Clonase LR recombination reaction (Invitrogen). The resultant plasmid, pRRLsinPPT-MEF-TfnR-EGFP, was used for lentiviral vector production.

Plasmid DNAs were transfected into HeLa cells using the FuGENE6 transfection reagent according to the manufacturer's instructions (Roche Applied Science). The Rab11a and Rab11b isoforms (Takahashi et al., 2011), and Exo70 (Zuo et al., 2006) were knocked down using double-stranded siRNAs purchased from Dharmacon, as described previously. Briefly, HeLa cells were transfected with the siRNAs using Lipofectamine 2000 (Invitrogen) and incubated for 24 h. The cells were then transferred to a new culture dish, incubated for an additional 48 h, and processed for immunofluorescence or immunoblot analysis. For knockdown of Sec15A or Sec15B, pools of siRNAs directed against the 3′-untranslated region of human Sec15A and Sec15B mRNAs (nucleotide residues 2424–3247 and 2454–2990, respectively, when the A residue of the initiation Met codon is assigned as residue 1), were prepared using the BLOCK-iT RNAi TOPO Transcription Kit and BLOCK-iT Dicer RNAi Kit (Invitrogen). HeLa cells were treated with the siRNA pools and processed for immunofluorescence or immunoblot analysis as described above for Rab11 and Exo70 knockdowns.

pRRLsinPPT-MEF-TfnR-EGFP was transfected into HEK293FT cells (Invitrogen) using Polyethylenimine Max (Polysciences, Warrington, PA) along with the packaging plasmids (pRSV-REV, pMD2.g and pMDL/pRRE). Culture medium was replaced 8 h after transfection. Culture media containing the lentiviral vector were collected at 24, 36 and 48 h after transfection, filtrated through a 0.45-µm filter (Millipore) and centrifuged at 32,000 g at 4°C for 4 h in an R15A rotor and Himac CR22G centrifuge (Hitachi Koki, Tokyo, Japan). The precipitated viral vector was resuspended in minimal essential medium.

HeLa cell lines stably expressing MEF-TfnR–EGFP were established by transduction with the lentiviral vector and isolated using cloning cylinders. One of the selected HeLa(MEF-TfnR–EGFP) clones, clone 6, was used in the following experiments.

Immunofluorescence analysis of cells was performed as described previously (Takahashi et al., 2011). Briefly, cells were fixed with 3% paraformaldehyde at room temperature for 10 min, washed three times with phosphate-buffered saline (PBS), quenched with 50 mM NH₄Cl for 20 min, washed three times with PBS, permeabilized with 0.1% Triton X-100 in PBS for 5 min, and washed three times with PBS. For detection of Rab11, cells were fixed with 10% trichloroacetic acid on ice for 15 min. The fixed/permeabilized cells were then subjected to staining with antibodies diluted with Can Get Signal immunostain (TOYOBO, Osaka, Japan), and observed using an Axiovert 200M microscope (Carl Zeiss, Göttingen, Germany) or an A1R-MP confocal laser-scanning microscope (Nikon, Tokyo, Japan).

For time-lapse recording, HeLa cells were plated on a collagen-coated glass-bottom culture dish (Mat Tek Corp.). Twenty-four hours after plating, an expression plasmid encoding EGFP-tagged Rab11a was transfected into the cells using a FuGENE6 reagent. After 48 h incubation, cells were mounted on a Stage Top Incubator (Tokai Hit, Fujinomiya, Japan). Cells were imaged at 37°C on an inverted microscope (Axiovert 200M) using a 63× oil immersion lens and a charge-coupled device camera (C9100-02; Hamamatsu Photonics, Hamamatsu, Japan). Time-lapse images were taken sequentially every 30 sec and analyzed using the MetaMorph imaging software (Molecular Devices, Sunnyvale, CA). Movies correspond to three frames per second.

For TIRFM, HeLa cells stably expressing TfnR–EGFP were culture on a 35-mm culture dish for 24 h, transfected with siRNAs, and further incubated for 24 h. The transfected cells were then transferred to a collagen-coated glass-bottom culture dish. After 48 h incubation, the cells were incubated in HEPES-buffered modified Eagle's medium, placed on a microscope stage pre-warmed to 37°C, and observed using NIS-Elements imaging software on a TIRFM ECLIPSE Ti (Nikon).

For time-lapse analysis after photobleaching (FRAP), cells expressing EGFP–Rab11a or those expressing TfnR–EGFP treated with siRNAs for Rab11a and Rab11b were incubated in HEPES-buffered modified Eagle's medium and placed on a microscope stage that was preincubated at 37°C. The cells were observed and bleached using FV1000D on an inverted microscope IX81 (Olympus, Tokyo, Japan) or the A1R-MP confocal microscope. To visualize FRAP of TfnR–EGFP, images were acquired sequentially every 2 seconds. Movies were recorded at 3 or 10 frames/sec.

Cells treated with control siRNA or siRNAs for Rab11a and Rab11b were incubated with HRP-conjugated Tfn (50 µg/ml) for 50 min at 4°C, washed, and then incubated for 5 or 20 min at 37°C. They were fixed with 3% paraformaldehyde in 0.1 M phosphate buffer (pH 7.2), and incubated first with diaminobenzidine (500 ng/ml) in 0.1 M Tris-HCl (pH 7.6) for 5 h, then with the same solution containing 0.005% H₂O₂ for 15 min at room temperature. The cells were further fixed with 2% glutaraldehyde–2% paraformaldehyde in 0.1 M phosphate buffer (pH 7.2) and embedded in Epon812 for conventional electron microscopy as described previously (Waguri et al., 1999).

We thank Shu-Chan Hsu, Wei Guo, Mitsuaki Tabuchi, Peter McPherson and Toshiaki Isobe for providing materials, and Katsuyuki Kanno and Akane Yamada for electron microscopy.