Endocytic trafficking is regulated by ubiquitylation (also known as ubiquitination) of cargoes and endocytic machineries. The role of ubiquitylation in lysosomal delivery has been well documented, but its role in the recycling pathway is largely unknown. Here, we report that the ubiquitin (Ub) ligase RFFL regulates ubiquitylation of endocytic recycling regulators. An RFFL dominant-negative (DN) mutant induced clustering of endocytic recycling compartments (ERCs) and delayed endocytic cargo recycling without affecting lysosomal traffic. A BioID RFFL interactome analysis revealed that RFFL interacts with the Rab11 effectors EHD1, MICALL1 and class I Rab11-FIPs. The RFFL DN mutant strongly captured these Rab11 effectors and inhibited their ubiquitylation. The prolonged interaction of RFFL with Rab11 effectors was sufficient to induce the clustered ERC phenotype and to delay cargo recycling. RFFL directly ubiquitylates these Rab11 effectors in vitro, but RFFL knockout (KO) only reduced the ubiquitylation of Rab11-FIP1. RFFL KO had a minimal effect on the ubiquitylation of EHD1, MICALL1, and Rab11-FIP2, and failed to delay transferrin recycling. These results suggest that multiple Ub ligases including RFFL regulate the ubiquitylation of Rab11 effectors, determining the integral function of the ERC.
Endocytic trafficking is essential for the maintenance of cell homeostasis by regulating diverse cellular processes including cell adhesion, cell migration, cell polarity and signal transduction (Doherty and McMahon, 2009). Endocytosis is used for the uptake of exogenous and cellular cargo including plasma membrane (PM) proteins such as the transferrin receptor (TfR) with ligands (Grant and Donaldson, 2009). Regardless of the mode of entry, endocytosed cargo is usually delivered to the early endosome (EE), which receives incoming material from primary vesicles generated by clathrin-mediated endocytosis (CME) and clathrin-independent endocytosis (CIE). Cargo can be delivered from the EE to the late endosomes (LEs) and lysosomes (LYs) for degradation, to the trans-Golgi network (TGN) or to recycling endosome (REs), which brings the cargo back to the PM (Maxfield, 2014; Wandinger-Ness and Zerial, 2014). Sorting of membrane proteins from EEs can lead to the entry of cargo into fast recycling pathways. Alternatively, cargo can be trafficked via a slow recycling pathway, which involves traffic through a juxtanuclear endocytic recycling compartment (ERC) and then via the REs before returning to the PM (Li and DiFiglia, 2012). Transfer of cargo from EEs to LYs depends on an endosomal maturation process including a small GTPase Rab5-to-Rab7 switch and a decrease in pH (Huotari and Helenius, 2011). The endosome maturation also involves sorting of membrane cargo destined for degradation into intraluminal vesicles (ILVs), thereby generating LE vacuoles known as multivesicular bodies (MVBs) (Piper and Katzmann, 2007).
In the slow recycling route, the EE maturation also results in the extension of tubules that become the ERC, whereas the remaining EEs eventually become the LEs and MVBs (Grant and Donaldson, 2009). The ERC is defined by the presence of Rab11, and traffic from Rab5-positive EEs to the Rab11-positive ERC is mediated by a family of ATPases called Eps15 homology domain-containing proteins (EHDs) (Naslavsky and Caplan, 2011). Traffic from the ERC back to the PM is complex and may involve multiple routes mediated by Arf6-dependent tubular carriers (Grant and Donaldson, 2009) and EHD1-mediated vesicles (Naslavsky and Caplan, 2011). Rab11 has emerged as an important regulator of endocytic transport (Grosshans et al., 2006; Jordens et al., 2005), especially a slow recycling pathway. Rab11 (of which there are Rab11a and Rab11b forms in mammals) regulates recycling of proteins with several Rab11-family interacting proteins (Rab11-FIPs), which are effectors of Rab11 GTPases (Horgan and McCaffrey, 2009). EHD1, which is recruited to tubular REs by MICAL-like protein 1 (MICALL1) also interacts with Rab11-FIP2, and coordination of these recycling machinery proteins could be crucial for maintaining the integrity of RE function (Naslavsky and Caplan, 2011).
Ubiquitin (Ub) has emerged as an important regulator of cargo endocytosis, endosomal sorting and a number of molecular mechanisms and components regulating these processes (Clague et al., 2012; Piper and Lehner, 2011; Haglund and Dikic, 2012). Ubiquitylation of cargo stimulates the endocytosis mediated by epsin and Eps15 (Clague et al., 2012), and the lysosomal degradation mediated by the ESCRT complex (Piper et al., 2014; Raiborg and Stenmark, 2009; Eden et al., 2012). Ubiquitylation (also known as ubiquitination) of endocytic machineries also regulates the endocytic trafficking. Mono-ubiquitylation of Eps15 regulates epidermal growth factor receptor (EGFR) internalization and lysosomal trafficking (Savio et al., 2016; Gschweitl et al., 2016), inhibits its association with ubiquitylated cargo in endocytic vesicles (Hoeller et al., 2006) and is also required for endosome maturation (Gschweitl et al., 2016). Mono-ubiquitylation of early endosome antigen 1 (EEA1), a membrane-tethering factor required for the fusion and maturation of EEs, regulates endosome fusion and trafficking (Ramanathan et al., 2013). Hrs ubiquitylation induces its degradation and limits cargo trafficking to MVBs for degradation (Pradhan-Sundd and Verheyen, 2015; Zhang et al., 2014). Ubiquitylation of Hrs and STAM1 also prevents their binding to ubiquitylated cargo (Hoeller et al., 2006; Hanafusa et al., 2011). A recent study suggests that ubiquitylation functions as a recycling signal of cargoes such as the v-SNARE Snc1 by mediating COPI binding (Xu et al., 2017). Reversible poly-ubiquitylation of WASH, an actin-nucleating protein essential for recycling, promotes endosomal protein recycling (Hao et al., 2013, 2015). Besides these findings, the role of ubiquitylation in the regulation of recycling endocytic machineries, such as Rab11 effectors is mostly unknown.
Here, we provide evidence that ubiquitylation of recycling endocytic machineries is regulated by the RFFL E3 ligase. An RFFL dominant-negative (DN) mutant inhibited segregation of EEs and REs without affecting LE maturation and induced a clustered ERC phenotype, concomitant with the inhibiting exit of cargoes from the ERC. BioID analysis revealed that RFFL interacts with recycling endocytic machineries including the Rab11 effectors EHD1, MICALL1 and class I Rab11-FIPs. The RFFL DN mutant trapped these Rab11 effectors and inhibited their ubiquitylation. The prolonged interaction of RFFL with Rab11 effectors was sufficient to induce the clustered ERC phenotype and to delay cargo recycling. RFFL directly ubiquitylates these Rab11 effectors in vitro, but RFFL knockout (KO) only reduced ubiquitylation of Rab11-FIP1. RFFL KO had a minimal effect on the ubiquitylation of MICALL1, EHD1 and Rab11-FIP2, and failed to delay transferrin recycling. These results suggest that multiple Ub ligases, including RFFL, regulate the ubiquitylation of Rab11 effectors, determining the integral function of the ERC.
RFFL catalytic inactive mutants induce abnormal recycling endosomes
In the process of a RFFL cellular localization analysis (Okiyoneda et al., 2018), we found that the catalytically inactive H333A and 2CA (C316A, C319A) RFFL mutants, but not a ΔRING mutant were dramatically distributed into a condensed structure (Fig. 1A). This condensed structure seems to be abnormal endosomal compartments because deleting the predicted palmitoylation sites (C5610A, ΔNT, Δ2-10) required for the endosomal localization (Okiyoneda et al., 2018) dispersed the condensed structures of the 2CA mutant into the cytoplasm (Fig. 1A). To determine the cellular localization of the condensed structure, colocalization analysis using endosomal markers was performed using a super-resolution confocal microscopy system (HyVolution). Confocal micrographs showed that wild-type RFFL (RFFL-WT) was partially colocalized with the EE markers Rab5 and EEA1, and the LE markers Rab7 and Lamp1 consistent with the EE and LE localization as previously shown (Okiyoneda et al., 2018) (Fig. 1B; Fig. S1). Intriguingly, RFFL-WT also predominantly colocalized with the RE marker Rab11 (Fig. 1B). RFFL-H333A induced a juxtanuclear localization to Rab5, EEA1, Rab7, Lamp1, Rab11 and trans-Golgi marker TGN46 (also known as TGOLN2) (Fig. 1B; Fig. S1). In contrast, RFFL-H333A had a marginal impact on the cellular localization of the ER marker Sec61, the Golgi marker GM130 (also known as GOLGA2) and the mannose-6-phosphate receptor (CI-M6PR; also known as IGF2R), which is retrieved from the LE to TGN (Fig. S1). Super-resolution micrographs revealed that EEA1, Rab5, Rab7, Lamp1 and TGN46 were partially colocalized with RFFL-H333A, and accumulated in the vicinity of the condensed structure (Fig. 1B; Fig. S1). In contrast, Rab11 was almost completely colocalized with RFFL-H333A in the condensed structure (Fig. 1B). Immunocytochemical analysis showed that RFFL-H333A caused clustering of Lamp1 around the condensed endosome (Fig. S1). These results suggest that the condensed structure is REs surrounded by LYs, and RFFL activity could regulate the integrity of RE function.
RFFL-H333A induces accumulation of tubular vesicular structures
Super-resolution imaging indicated that the RFFL-H333A-induced condensed RE structure is composed of assembled tubule-like organelles (Fig. 2A). Transmission electron microscopy (TEM) was used to resolve the ultrastructure of the condensed endosomes. In RFFL-H333A-transfected cells, accumulation of tubular vesicular structures was observed at the juxtanuclear region, consistent with the RFFL-H333A–GFP localization (Fig. 2B–E). The tubular vesicular structures were surrounded by electron-dense LYs consistent with immunocytochemical analysis with an anti-Lamp1 antibody (Fig. 2B–E; Fig. S1). The tubular vesicular structures were variable in length with a constant diameter (∼70 nm), which corresponds to the recycling compartment with diameters ranging from 50 to 75 nm (Marsh et al., 1995). Thus, we conclude that the condensed endosome induced by the RFFL mutant is a clustered endocytic recycling compartment (ERC). Consistent with super-resolution microscopy analysis, EEs, visualized as 100–300 nm organelles, were occasionally localized in the clustered ERC (Marsh et al., 1995) (Fig. 2E, arrows).
Scanning electron microscopy (SEM) analysis was also performed to examine the three-dimensional ultrastructure of the ERC. Consistent with the TEM analysis, the clustered tubule-vesicular structure was observed at the juxtanuclear region (Fig. 2F). It was also located in the proximity of LYs (Fig. 2F–H). This structure seemed to be a cluster of tubular vesicular structures rather than one continuous tubular reticulum (Fig. 2G–H).
RFFL-H333A decelerates endocytic cargo recycling, but not lysosomal delivery
To determine whether RFFL-H333A induces not only the morphological change, but also a functional change of the endosomal compartments, its impact on the endosome maturation and cargo sorting were analyzed. Endosomes have distinct functional domains that are organized by Rab proteins (Sönnichsen et al., 2000). During endosome maturation, a Rab switch occurs, where Rab5 is exchanged for Rab7 to form LEs (Rink et al., 2005; Vonderheit and Helenius, 2005; Poteryaev et al., 2010) and Rab11 is also lost (Huotari and Helenius, 2011). Super-resolution confocal analysis revealed that ∼30% of the Rab7 and Rab11 proteins were segregated from Rab5-containing EEs in control and RFFL-WT-transfected cells (Fig. 3A,B). In RFFL-H333A-transfected cells, Rab7 was segregated from Rab5 similar to what was seen in the control cells, but Rab11 failed to be segregated and almost completely colocalized with Rab5 (Fig. 3A,B). Thus, RFFL-H333A may selectively inhibit the segregation of REs from EEs, without affecting LE maturation.
We further examined the impact of RFFL-H333A on the endocytic trafficking of cargoes from the PM. First, we tested the transferrin receptor (TfR), which is internalized by CME and efficiently recycled to the PM. Endocytic trafficking of TfR was monitored by time-lapse imaging after labeling the cell surface TfR with Alexa Flour 647-conjugated transferrin (A647–Tf). After 2.5 h A647–Tf loading with a 1 h chase (T-0 h), TfR was mainly localized at juxtanuclear endocytic compartment in non-transfected (NT), and RFFL–GFP-transfected cells whereas it was accumulated at the condensed ERC in RFFL-H333A–GFP-transfected cells (Fig. 4A). During the additional 1 h chase, cellular Tf intensity gradually decreased to 46% in NT or RFFL–GFP-transfected cells, indicating that approximately half of the internalized TfR was recycled to the PM (Fig. 4B). In RFFL-H333A–GFP-transfected cells, TfR was retained at the ERC and disappearance of the intracellular TfR was significantly lower than in the NT and RFFL-WT–GFP-transfected cells, indicating that the RFFL mutant decelerates the TfR recycling (Fig. 4A,B). Consistent with what was found for TfR, RFFL-H333A expression delayed endocytic recycling of WT-CFTR, a CME cargo (Lukacs et al., 1997), which was found to accumulate at the ERC (Fig. S2A). Flag–CD59, a GPI-anchored protein internalized by CIE and recycled to the PM (Donaldson et al., 2009), was also accumulated in the ERC with RFFL-H333A in the transfected cells after a 4 h chase, indicating delayed endocytic recycling to the PM (Fig. S2B). These results suggest that the RFFL catalytic activity may play an essential role in the endocytic recycling of both CME and CIE cargoes.
We next examined the effect of RFFL-H333A on the endocytic trafficking of cargoes that are delivered to lysosome. Internalized rΔF508-CFTR-3HA, a misfolded PM protein interacting with RFFL Ub ligase (Okiyoneda et al., 2018), disappeared after a 4 h chase in NT and RFFL-WT–GFP-transfected cells because of its lysosomal degradation (Okiyoneda et al., 2010). However, rΔF508-CFTR-3HA was accumulated with RFFL–H333A in the ERC in the transfected cells, implying delayed lysosomal delivery (Fig. 4C). By contrast, lysosomal delivery of the fluid-phase marker dextran was not affected by RFFL-H333A (Fig. 4D; Fig. S2C). Finally, we tested EGFR, which is internalized by CME and recycled to the PM when the EGF concentration is low (1.5–10 ng/ml) but is internalized by CIE and sorted to LY at a high EGF concentration (>100 ng/ml) (Sigismund et al., 2008). Immunocytochemical analysis using anti-EGFR antibody revealed that at a low EGF concentration (10 ng/ml), internalized EGFR was largely disappeared in the NT cells, indicating EGFR recycling (Fig. 4E). RFFL-H333A expression resulted in the accumulation of internalized EGFR with RFFL-H333A in the ERC at a low EGF concentration (10 ng/ml) (Fig. 4E). However, at a high EGF concentration (100 ng/ml) EGFR was not accumulated with RFFL-H333A, and it was delivered to LY as in control cells (Fig. 4E). The EGFR LY delivery was also confirmed by colocalization analysis with the Alexa Fluor 568 (A568)–EGF-labeled EGFR. In the NT cells, the internalized EGF was colocalized with EEA1 after a 15 min chase (Fig. S2D,E), but it was colocalized with Lamp1 after a 60 min chase, indicating the LY delivery of EGFR (Fig. 4F,G; Fig. S4D). Upon RFFL-H333A overexpression, EGF was delivered from EEs to LYs similar to what is seen in the NT and RFFL-WT-expressing cells (Fig. 4F,G; Fig. S4D,E). These results suggest that RFFL-H333A decelerates the endocytic recycling of cargoes from the ERC regardless of the endocytosis pathway without affecting the LY delivery. RFFL-H333A might delay the LY delivery of only a particular type of cargo, including rΔF508-CFTR, which directly interacts with RFFL for the peripheral quality control (QC).
BioID identifies EHD1, MICAL-L1 and class I Rab11-FIPs as RFFL interactors
Since the clustered ERC formed upon transfection with RFFL-H333A or RFFL-2CA, both of which strongly interact with the RFFL substrate rΔF508-CFTR (Okiyoneda et al., 2018), its formation could be due to the substrate trapping by the DN mutants. To examine the molecular mechanism underlying formation of the clustered ERC, RFFL interactome analysis was performed. We used a BioID (proximity-dependent biotin identification) assay, which can identify weak and/or transient interactions (Roux et al., 2012). RFFL fused with BirA* and the HA epitope at the C-terminus (RFFL–BirA*–HA) was stably expressed in CFBE cells. The addition of 50 µM biotin in the medium resulted in the accumulation of biotinylated proteins colocalized with RFFL–BirA*–HA (Fig. 5A). The clustered ERC was also induced by RFFL-H333A–BirA*–HA, and biotinylated proteins were accumulated together, indicating that the RFFL-interacting proteins are biotinylated (Fig. 5A). Immunoblotting also confirmed that RFFL–BirA*–HA expression induced the accumulation of biotinylated proteins (Fig. 5B). We purified the biotinylated proteins from the RFFL–BirA*-expressing CFBE cells under denaturing conditions using NeutrAvidin agarose and visualized these by silver staining (Fig. 5C). The isolated biotinylated proteins were analyzed by mass spectrometry. We identified 100 endosome-associated proteins that were unique to the RFFL–BirA* pulldown. Among them we identified 30 RE -associated proteins including class I Rab11-FIPs (Rab11-FIP1, Rab11-FIP5 and Rab11-FIP2), MICALL1, MICALL2 (also known as JRAB) and EHD1 as novel RFFL interactors (Fig. 5D). These RE-associated proteins are all involved in TfR recycling except for MICALL2 (Sharma et al., 2009; Rapaport et al., 2006; Lin et al., 2001; Peden et al., 2004; Lindsay et al., 2002; Lindsay and McCaffrey, 2002; Terai et al., 2006; Schonteich et al., 2008).
RFFL-H333A traps endocytic recycling machineries in the ERC
To confirm the BioID results, cellular localization of these RE-associated proteins was examined. RFFL–mCherry was partly colocalized with GFP-tagged MICALL1, EHD1, Rab11-FIP1C (also known as RCP), Rab11-FIP2 and Rab11-FIP5 (Fig. 6A). As expected, RFFL-H333A–mCherry expression induced concentration of the GFP-tagged MICALL1, EHD1, Rab11-FIP1C, Rab11-FIP2 and Rab11-FIP5 in the clustered ERC (Fig. 6A). Super-resolution confocal microscopy analysis also confirmed colocalization of these Rab11 effectors with RFFL-H333A in the clustered ERC (Fig. 6A). Pulldown experiments showed that RFFL-H333A was strongly bound to Myc–biotin (MB)-tagged EHD1, MICALL1, Rab11-FIP1C, Rab11-FIP2 and Rab11-FIP5 compared to the WT counterpart (Fig. 6B,C; Fig. S4A–E). Consistent with the colocalization with RFFL-WT (Fig. 6A), EHD1 interacted with RFFL-WT more strongly than the other Rab11 effectors (Fig. S4A–E). Immunocytochemical analysis also revealed that RFFL-H333A expression induced the accumulation of endogenous MICALL1, EHD1, Rab11-FIP1 and Rab11-FIP5 in the clustered ERC (Fig. 6D). Endogenous Rab11-FIP2 was could not be detected with available antibodies. Based on these results, we hypothesize that RFFL-H333A traps these Rab11 effectors on the endosomal membrane, and the defective dissociation may induce the clustered ERC phenotype by inhibiting their physiological function. To test this hypothesis, we used a chemical-induced protein dimerization (CID) assay to trap these Rab11 effectors on RFFL-WT forcibly. The Rab11 effectors and RFFL-WT were fused with FRB–GFP and FKBP–mCherry, respectively. They displayed the typical distribution with dispersed endosomes and tubular endosomes before rapamycin treatment, which induces the dimerization of FRB and FKBP (Fig. 6E; Fig. S4F). Rapamycin treatment induced the clustered ERC, which contained colocalized RFFL-FKBP–mCherry and Rab11 effectors tagged with FRB-GFP (Fig. 6E). Consistent with the RFFL-H333A phenotype, internalized WT-CFTR was accumulated in the CID-induced clustered ERC (Fig. 6F; Fig. S4G). Thus, the sequestration of the Rab11 effectors on RFFL is sufficient to induce the clustered ERC and delay the cargo recycling.
RFFL regulates ubiquitylation of Rab11 effectors
Given that the RFFL DN mutant trapped the Rab11 effectors and resulted in the formation of the clustered ERC, RFFL may regulate the function of the Rab11 effectors by mediating their ubiquitylation. Western blotting following NeutrAvidin pulldown under denaturing condition showed that the Rab11 effectors MICALL1, EHD1, Rab11-FIP1C, and Rab11-FIP2 were ubiquitylated, although Rab11-FIP5 ubiquitylation was undetectable (Fig. 7A). Co-transfection of RFFL-H333A-V5 decreased their ubiquitylation; Rab11-FIP2 ubiquitylation was dramatically inhibited (Fig. 7A,B and C,F, lane 4). These Rab11 effectors were largely mono-ubiquitylated as their ubiquitylation was observed with the HA–Ub-K0 mutant in which all lysine residues in the Ub are replaced with arginine residues (Fig. 7C–F, lane 6). We identified several ubiquitylation sites in MICALL1, EHD1, Rab11-FIP1C and Rab11-FIP2 by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) (Table S1). Whereas RFFL overexpression had a minimal effect on the ubiquitylation of MICALL1 and EHD1, it modestly increased Rab11-FIP1C poly- and/or multiple-ubiquitylation (Fig. 7C–E, lane 3). Intriguingly, RFFL overexpression dramatically increased Rab11-FIP2 poly- and/or multiple-ubiquitylation (Fig. 7F, lane 3).
To examine whether endogenous RFFL regulates the ubiquitylation of Rab11 effectors, their ubiquitylation was also measured in RFFL-knockout (KO) cells generated by the CRISPR/CAS9 system (Fig. S5A–C). As expected, RFFL KO significantly reduced the ubiquitylation of Rab11-FIP1C, and this phenotype was reversed by reintroducing RFFL (Fig. 8A,B). By contrast, RFFL KO did not reduce the ubiquitylation of MICALL1, EHD1 and Rab11-FIP2 (Fig. 8B–E). Intriguingly, RFFL-H333A expression was able to reduce the ubiquitylation of Rab11 effectors in RFFL KO cells as in the case in the parental cells (Fig. 8A and C–E, lane 8). These results suggest that Rab11-FIPC ubiquitylation is predominantly regulated by RFFL, and the ubiquitylation of other Rab11 effectors may be governed primarily by unidentified Ub ligases that could compensate for the RFFL function. RFFL-H333A might inhibit the interaction of the Rab11 effectors with the unknown Ub ligases because of the prolonged interaction. In contrast to the phenotype observed in the RFFL-H333A-transfected cells, RFFL KO had a minimal effect on the TfR recycling (Fig. 8F; Fig. S5D). RFFL-H333A expression delayed the TfR recycling in the RFFL KO cells, similar what was seen in the parental cells (Fig. S5D). These results indicate that the reduced Rab11-FIPC ubiquitylation is insufficient to delay cargo recycling. Diminished ubiquitylation of multiple Rab11 effectors such as EHD1, MICALL1, Rab11-FIP1C, and Rab11-FIP2 might be necessary to delay cargo recycling.
Given that the RFFL DN mutant reduced the ubiquitylation of Rab11 effectors, RFFL may be capable of facilitating their ubiquitylation although its KO had a marginal effect. To clarify this possibility, we performed an in vitro ubiquitylation assay. GST–EHD1, GST–Rab11-FIP1C and ubiquitylation enzymes were purified from E. coli by affinity purification (Okiyoneda et al., 2018) (Fig. S5E). GST–EHD1 or GST–Rab11-FIP1C was incubated with E1 (also known as UBE1 and UBA1), UbcH5c (also known as UBE2D3), RFFL and Ub. After the reaction, the ubiquitylated substrates were isolated by using glutathione beads and analyzed by western blotting with an anti-Ub antibody. As expected, ubiquitylation of GST–EHD1 and GST–Rab11-FIP1C were reconstituted in the presence of E1, UbcH5c, RFFL and Ub, when omitting any components compromised the ubiquitylation (Fig. 8G,H). Because of the technical difficulty of protein purification from E. coli, we attempted an in vitro ubiquitylation assay using MICALL1 and Rab11-FIP2 purified from mammalian cells. Purified MB–MICALL1 or MB–Rab11-FIP2 was incubated with E1, UbcH5c, RFFL and HA-Ub, and the ubiquitylated substrate was pulled down by NeutrAvidin agarose. The pulldown sample and the supernatant were analyzed for the ubiquitylation and confirmation of the ubiquitylation enzymes activity, respectively (Fig. S5F). MICALL1 ubiquitylation was observed in the presence of ubiquitylation enzymes, predominantly as mono-ubiquitylation (Fig. S5G, lane 2). Omitting UbcH5c or HA-Ub completely abolished, and omitting the E1 partially abolished, the ubiquitylation. Intriguingly, omitting RFFL abolished the high-molecular-mass signal derived from poly-ubiquitylation and/or multi-mono-ubiquitylation of MICALL1, but the mono-ubiquitylation was still observed (Fig. S5G lane 5). This RFFL-independent mono-ubiquitylation may be due to the activity of other E3 ligases that continuously interact with the MICALL1 during the purification process because the ubiquitylation activity was observed in the supernatant (Fig. 8G, lane 5). The RFFL-dependent ubiquitylation of Rab11-FIP2 was also seen as poly- and/or multiple-ubiquitylation, which is consistent with the overexpression experiment (Fig. 8I). These results provide evidence that RFFL is capable of ubiquitylating the Rab11 effectors directly.
In this study, we demonstrate that ubiquitylation of endocytic recycling machinery proteins, including Rab11 effectors, is regulated by the RFFL E3 ligase. The RFFL DN mutant induced a clustered ERC phenotype and delayed exit of cargoes from the ERC. Moreover, it inhibited the segregation of Rab11-positive REs from Rab5-positive EEs without compromising LE maturation. Our biochemical analyses revealed Rab11-FIP1C, -FIP2, -FIP5, EHD1 and MICALL1 as novel RFFL-interacting proteins. The RFFL DN mutant trapped these recycling machineries and inhibited their ubiquitylation. RFFL KO only reduced the Rab11-FIP1 ubiquitylation, but RFFL was capable of ubiquitylating all of the Rab11 effectors directly. These results suggest that the integral function of the endocytic recycling pathway can be regulated by ubiquitylation of the recycling machinery proteins by multiple E3 ligases including RFFL.
RFFL has been proposed to function at EEs and LEs to selectively interact with conformationally defective proteins to direct them to lysosomal degradation as a part of the peripheral QC mechanism (Okiyoneda et al., 2018). Consistent with previous studies (Coumailleau et al., 2004), super-resolution microscopy analysis revealed that RFFL is localized at not only EEs and LEs but also at REs, implying that it has physiological roles in the recycling pathway. In the previous study, RFFL-WT overexpression induced the condensation of endosomal structures displaying TfR, Rab5 and Rab11 into several perinuclear globular structures and delayed TfR recycling (Coumailleau et al., 2004), a similar phenotype observed in the RFFL DN mutant-expressing cells. In fact, we occasionally observed the clustered ERC in RFFL-WT overexpressing cells. This was probably due to the prolonged interaction of RFFL-WT with the Rab11 effectors because of the following reasons: (1) the RFFL DN mutant, but not the ΔRING mutant, which fails to trap substrates (Okiyoneda et al., 2018), induced the clustered ERC phenotype; (2) the RFFL DN mutant trapped the Rab11 effectors; and (3) forced association of RFFL-WT with the Rab11 effectors by CID was sufficient to induce the clustered ERC phenotype. In CFBE cells stably expressing RFFL-H333A–BirA*–HA, the clustered ERC was not frequently observed before biotin addition probably because the expression level of the RFFL DN mutant is insufficient to inhibit the function of the Rab11 effectors (Fig. 5A). The biotin addition dramatically induced clustered ERC formation possibly by inhibiting the function Rab11 effectors because of the biotinylation.
We showed that the RFFL DN mutant failed to affect the internalization, but delayed the recycling of cargoes internalized by CME and CIE, and caused them to accumulate in the clustered ERC. This phenotype is reminiscent of that observed upon EHD1 knockdown (KD) (Jović et al., 2007; Rapaport et al., 2006), MICALL1 KD (Sharma et al., 2009) and overexpression of class I Rab11-FIP mutants (Schafer et al., 2016; Ducharme et al., 2007) or the MYO5B tail (Hales et al., 2002). Thus, the RFFL DN mutant might compromise the function of these recycling machineries. Class I Rab11-FIPs (FIP1, 2 and 5), which competitively interact with Rab11 could influence the dynamic structure of the recycling system membranes and engage in discrete steps of the recycling process, and their cooperative efforts facilitate cargo recycling (Baetz and Goldenring, 2013). Rab11-FIP2 is associated with the plus-end-directed actin-based motor MYO5B, which is crucial for the exit of Rab11 vesicles from the ERC (Hales et al., 2002; Wang et al., 2008; Gidon et al., 2012). EHD1 binds Rab11-FIP2 (Naslavsky et al., 2006) and MICALL1 to regulate ERC trafficking and affect membrane tubulo-vesicle formation and scission, respectively (Roland et al., 2007; Sharma et al., 2009; Cai et al., 2013; Giridharan et al., 2013). Based on these findings, sequestration of these Rab11 effectors by the RFFL DN mutant may compromise the formation of the functional complex (e.g. MYO5B–Rab11–Rab11–FIPs–EHD1) that is necessary for cargo exit from the perinuclear ERC to the PM. Our results suggest that RFFL-mediated ubiquitylation also regulates traffic from the ERC to TGN since the RFFL DN mutant induced TGN46 redistribution into the ERC, possibly by inhibiting Rab11 effectors such as EHD1 and Rab11-FIP1C both of which regulate transport from the EE/RE to the TGN (Lin et al., 2001; Jing et al., 2010). Consistent with the EHD1 and Rab11-FIP1C KD phenotype (Gokool et al., 2007; Jing et al., 2010), the RFFL DN mutant had a marginal effect on the cellular localization of CI-M6PR, which is retrieved from the LE to the TGN.
Intriguingly, the RFFL DN mutant inhibited the segregation of Rab5-positive EEs and the Rab11-positive ERC without interfering with the segregation of Rab7-positive LEs from the EEs. Consistent with this phenotype, the RFFL DN mutant inhibited the recycling of the cargoes, but not the LY delivery of EGFR and dextran. Among the cargoes we tested, only the LY delivery of unfolded rΔF508-CFTR was inhibited by the RFFL mutant. Because RFFL selectively interacts with unfolded rΔF508-CFTR as a part of the peripheral QC mechanism (Okiyoneda et al., 2018), the RFFL DN mutant probably traps the unfolded protein at the EE and/or ERC and consequently delays the LY delivery. The impaired segregation of the Rab11-positive ERC from Rab5-positive EEs upon overexpression of the RFFL mutant is probably a result of the defective function of the Rab11 effectors. Rab11-FIP2 has roles in the modulation of trafficking not only within RE systems but also in elements of early endosomal and pre-recycling trafficking (Ducharme et al., 2011). Moreover, EHD1 and MICALL1 could play a crucial role in the formation of tubular RE, which also participates in the cargo trafficking from peripheral EE to ERC (Sharma et al., 2009; Xie et al., 2016). It has been speculated that accumulation of cargo in perinuclear ERC is due to the coalescence of entire EE populations into the ERC region upon EHD1 or MICALL1 KD (Xie et al., 2016). Since the RFFL DN mutant induced the coalesced EE into the ERC, the timely regulated ubiquitylation of EHD1 and MICALL1 may be essential to segregation of ERC from EEs by producing tubular RE.
Actin- and microtubule-dependent motors play important roles in endocytic recycling. Our BioID assay revealed that RFFL could interact with motor proteins including MYO1B, MYO6, MYO1E, KIF5B and KIF16B (Fig. S3A). MYO1B regulates trafficking from the TGN (Almeida et al., 2011) and EEs (Salas-Cortes et al., 2005), and is involved in membrane recycling (Neuhaus and Soldati, 2000). MYO1E participates in TfR trafficking to the ERC (Cheng et al., 2012). MYO6 has Ub-binding domains (Penengo et al., 2006; He et al., 2016), and its KD induces the perinuclear accumulation of Rab5 (Masters et al., 2017) and inhibits cargo trafficking from the EE to ERC (Chibalina et al., 2007). KIF16B KD induces the clustering of EE at the perinuclear region and delays TfR recycling (Hoepfner et al., 2005). KIF5B could regulate EE-to-RE trafficking (Yi et al., 2015) and its defect induces a perinuclear localization of LYs (Tanaka et al., 1998) as observed upon overexpression of the RFFL DN mutant. Based on these findings, RFFL may also regulate the interaction of these motors with the Rab11 effectors to maintain the integrity of the recycling pathway.
We demonstrate that the Rab11 effectors EHD1, MICALL1, Rab11-FIP1C and Rab11-FIP2 undergo ubiquitylation. This modification does not seem to stimulate their degradation because the RFFL DN mutant inhibited their ubiquitylation, but did not affect their protein level. In vitro assay demonstrates that RFFL can directly ubiquitylate the Rab11 effectors. While the RFFL DN mutant reduced the ubiquitylation of Rab11 effectors, RFFL KO reduced only the ubiquitylation of Rab11-FIP1, but not that of EHD1, MICALL1 or Rab11-FIP2. Hence, Rab11-FIPC ubiquitylation is predominantly regulated by RFFL, and the ubiquitylation of other Rab11 effectors might be regulated mainly by unidentified Ub ligases that could compensate for the RFFL function. Our BioID assay identified several E3 ligases such as RNF34 (CARP1), XIAP, SH3RF1 (POSH) and DTXL3L in the RFFL complex (Fig. S3B). RNF34, a homolog of RFFL, is localized in EEs and LEs (Jin et al., 2014; McDonald and El-Deiry, 2004). SH3RF1, a TGN-localized E3 ligase, regulates Gag viral glycoprotein trafficking from the TGN to the PM (Alroy et al., 2005). DTXL3L is localized in EE and regulates AIP4 E3 ligase activity (Holleman and Marchese, 2014). These E3 ligases may ubiquitylate the Rab11 effectors, and compensate the RFFL function in the KO cells.
Intriguingly, the Rab11 effectors were predominantly mono-ubiquitylated. Moreover, the expression of the RFFL DN mutant reduced their ubiquitylation, and that coincided with the sequestration of Rab11 effectors in the clustered ERC. Although we could not detect the ubiquitylation (Fig. 7A), Rab11-FIP5 might undergo the RFFL-mediated ubiquitylation at an undetectable level because Rab11-FIP5 interacted with the RFFL DN mutant. In fact, similar to other Rab11 effectors, Rab11-FIP5 ubiquitylation is annotated in the PhosphoSitePlus® database (https://www.phosphosite.org/proteinAction?id=5458&showAllSites=true). As mono-ubiquitylation affects intermolecular and/or intramolecular interaction (Hoeller et al., 2006), RFFL might regulate spatiotemporal ubiquitylation of the Rab11 effectors that could then lead to the collapse of the Rab11 complex so it could not undertake subsequent rounds of complex formation. RFFL KO reduced the Rab11-FIP1 ubiquitylation, but it failed to affect TfR recycling. We cannot preclude the possibility that the reduced ubiquitylation affects the Rab11-FIP1 function because TfR recycling is not affected by Rab11-FIP1 KD (Peden et al., 2004; Carson et al., 2013). Although RFFL KO failed to affect the ubiquitylation, over-expression of RFFL dramatically increased the ubiquitylation of Rab11-FIP2, which could link EHD1, Rab11-FIP1, Rab11-FIP5 and myosin motor MYO5b (Hales et al., 2002; Cullis et al., 2002; Naslavsky et al., 2006). We also found the multiple ubiquitylation sites in the MYO5B-binding region of Rab11-FIP2 (Schafer et al., 2014). Since the ERC region could provide a concentrated platform on which vesicles/tubules, motor proteins and the cytoskeleton may be coupled to facilitate efficient transport to the PM (Xie et al., 2016), the RFFL-mediated ubiquitylation of Rab11 effectors may also coordinate these machineries to efficiently transport cargoes and lipids to the PM from ERC.
Intriguingly, we identified deubiquitylating enzymes (DUBs), including VCPIP1 (VCIP135), USP15 and USP43 as RFFL interactors in the BioID analysis (Fig. S3B). VCPIP1 has been reported to regulate SNARE complex formation and Golgi structure assembly by deubiquitylating the mono-ubiquitylation of syntaxin 5 (Huang et al., 2016). USP15 regulates perinuclear localization of endosome by deubiquitylating of SQSTM1 (Jongsma et al., 2016). Therefore, the RFFL-associated DUBs may regulate reversible ubiquitylation of Rab11 effectors, modifying the coordinated function of recycling machineries.
In summary, the RFFL-regulated ubiquitylation of Rab11 effectors could modulate the integral function of the endocytic recycling pathway. This present study could shed new light on the role of ubiquitylation in the regulation of endocytic recycling pathway. Detailed analysis of the regulatory mechanism of the Rab11 effector complex may reveal the precise functions of the ubiquitylation in the endocytic recycling pathway.
MATERIALS AND METHODS
Reagents and plasmids
DAPI (Cat#340-07971), biotin (Cat#023-08716) and ImmunoStar Zeta (Cat#297-72403) were purchased from FUJIFILM Wako Pure Chemical Corporation. Alexa Fluor 647–Transferrin (Cat#T23366), biotin–EGF (Cat#E3477), NeutrAvidin agarose (Cat#29200), Alexa Fluor 568–Streptavidin (Cat#S-11226), HRP–NeutrAvidin (Cat#31001) and SuperSignal West Pico Chemiluminescent Substrate (Cat#34080) were from Thermo Fisher Scientific. TRITC–dextran (Cat#T1162) was from Sigma, and rapamycin (Cat#R0097) was from Tokyo Chemical Industry Co., Ltd.
For the expression constructs, PCR was performed using PrimeSTAR HS or PrimeSTAR Max DNA polymerase (Takara Bio, Japan). pDest-RFFL-GFP (Okiyoneda et al., 2018), pDest-RFFL-Δ2-44-GFP, pDest-RFFL-Δ2-10-GFP, pDest-RFFL-Δ313-363-GFP, pDest-RFFL-H333A-GFP, pDest-RFFL-C316A, C319A-GFP, pDest-RFFL-C5A, -C6A, -C10A, -C316A, -C319A-GFP, pDest-RFFL-Δ2-44, -C316A, -C319A-GFP, pDest-RFFL-Δ2-10, -C316A and -C319A-GFP were constructed by means of the LR reaction using pDest-eGFP-N1 (Addgene #31796; Hong et al., 2010). pDest-RFFL-mCherry (Okiyoneda et al., 2018) and pDest-RFFL-H333A-mCherry were constructed using pDest-mCherry-N1 (Addgene #31907; Hong et al., 2010). pdcDNA-Myc-Bio-EHD1, pdcDNA-Myc-Bio-MICALL1, pdcDNA-Myc-Bio-Rab11FIP1, pdcDNA-Myc-Bio-Rab11FIP2, pdcDNA-Myc-Bio-Rab11FIP5 and pdcDNA-Myc-Bio were constructed from the pdcDNA-Myc-Bio gateway destination vector in which Bio tag sequence from HBH tag (Okiyoneda et al., 2018) was inserted after the Myc tag in pdcDNA-Myc (LMBP 7212, BCCM Plasmid Collection). pLX304-BirA(R118G)-HA, pLX304-RFFL-BirA(R118G)-HA, pLX304-RFFL-H333A-BirA(R118G)-HA, pLX304-RFFL-V5, pLX304-RFFL-H333A-V5, pLX304-RFFL-HB and pLX304-RFFL-H333A-HB were constructed using pLX304-V5 (Addgene #25890; Yang et al., 2011) and pLX304-HB in which the V5 tag in pLX304-V5 was replaced with the His-Bio tag. pEZY-eGFP-EHD1, pEZY-eGFP-MICALL1, pEZY-eGFP-Rab11FIP1, pEZY-eGFP-Rab11FIP2 and pEZY-eGFP-Rab11FIP5 were constructed using pEZY-eGFP (Addgene #18671; Guo et al., 2008). pDest-EHD1-FRB-GFP, pDest-Rab11-FIP1C-FRB-GFP, pDestRab11-FIP5-FRB-GFP and pEZY-GFP-FRB-MICALL were constructed using YFP-tagged FRB (YR) (Addgene #20148; Inoue et al., 2005). pDest-RFFL-FKBP-mCherry was constructed using CFP-FKBP (CF) (Addgene #20160; Inoue et al., 2005). pcDNA3.1(–) HA-Ub and pcDNA3.1(-) HA-Ub K0 were constructed using pcDNA3.1(–) (Thermo Fisher Scientific). pGEX-GST-EHD1 and pGEX-GST-Rab11-FIP1C were constructed using pGEX-4T-2 vector (GE Healthcare, Japan). All constructs were verified by DNA sequencing.
Cell culture and transfection
GripTite 293 MSR cells (293MSR, Thermo Fisher Scientific) and RFFL KO 293MSR cells (see below) were grown in Dulbecco's modified Eagle's medium (DMEM; FUJIFILM Wako Pure Chemical Corporation) supplemented with 10% fetal bovine serum (FBS) and 500 µg/ml G418. HeLa cells stably expressing WT-CFTR-3HA (HeLa-CFTR) or ΔF508-CFTR-3HA (HeLa-ΔF508) were grown in DMEM supplemented with 10% FBS and 2 µg/ml puromycin (Okiyoneda et al., 2010). CFBE41o- cells stably expressing CFTR-3HA variants (Okiyoneda et al., 2018) and RFFL–BirA*–HA or RFFL-H333A–BirA*–HA were grown in MEM (FUJIFILM Wako Pure Chemical Corporation) supplemented with 10% FBS, 200 µg/ml G418, 3 µg/ml puromycin and 10 µg/ml blasticidin. Transient transfection in 293MSR and HeLa cells was accomplished by using polyethylenimine (PEI) max (Polysciences Inc, Warrington, PA), and experiments were performed 24–48 h post transfection.
Cells grown on coverslips were fixed in 4% paraformaldehyde (PFA) for 20 min at room temperature (RT) and permeabilized with 0.1% Triton X-100 in PBS for 5 min. After blocking in 0.5% BSA in PBS for 30 min at RT, cells were incubated with primary antibody in 0.5% BSA in PBS for 1 h at RT, followed by incubating with Alexa Fluor-conjugated secondary antibody in 0.5% BSA for 1 h at RT. Cells were incubated with DAPI for 5 min at RT, and mounted in VECTASHIELD mounting medium (VECTOR Laboratories) or ProLong Diamond Antifade Mountant (Thermo Fisher Scientific). For Lamp1, CI-M6PR, Rab11-FIP1 and Rab11-FIP5 staining, cells were permeabilized with 0.1% saponin for 15 min. For EHD1 staining, cells were permeabilized with 0.2% saponin for 15 min at RT.
Antibodies used for the immunocytochemistry were: rabbit monoclonal anti-CI-M6PR (Cell Signaling Technology, D3V8C; 1:400), mouse monoclonal anti-GM130 (MBL, 5G8; 1:200), mouse monoclonal anti-EEA1 (MBL, 3C10; 1:500), rabbit monoclonal anti-Lamp1 (Cell Signaling Technology, D2D11; 1:200), mouse monoclonal anti-Flag (FUJIFILM Wako Pure Chemical Corporation, 1E6; 1:500), mouse monoclonal anti-HA (BioLegend, 16B12; 1:500), mouse monoclonal anti-EGFR (MBL, 6F1; 1:100), rabbit polyclonal anti-MICALL1 (Abnova, H00085377-B01P; 1:500), rabbit polyclonal anti-EHD1 (abcam, ab75886; 1:500), rabbit polyclonal anti-Rab11-FIP1 (GeneTex, GTX117197; 1:500), rabbit polyclonal anti-Rab11-FIP5 (NOVUS Biologicals, NBP1-57009; 1:500), Alexa Fluor 568-Streptavidin (Thermo Fisher Scientific, S-11226; 1:2000). Alexa Fluor 594 AffiniPure Donkey anti-mouse IgG (H+L), Alexa Fluor 647 AffiniPure goat anti-mouse IgG (H+L) and Alexa Fluor 488 AffiniPure Donkey Anti-Mouse IgG (H+L) were purchased from Jackson ImmunoResearch (at 1:500).
Organelle markers mRFP-Rab5 (Addgene #14437; Vonderheit and Helenius, 2005), mRFP-Rab7 (Addgene #14436; Vonderheit and Helenius, 2005), DsRed-Rab11 (Addgene #12679; Choudhury et al., 2002), mCherry-TGN46 (Addgene #55145), Lamp1-RFP (Addgene #1817; Sherer et al., 2003), mTagBFP2-Rab5 and mCherry-Sec61 (Addgene #49155; Zurek et al., 2011) were transiently transfected using PEI Max.
Single optical sections were collected on an inverted laser confocal fluorescence microscope (SP8, Leica) equipped with an HC PL APO 63×/NA 1.40 objective. Images were processed with Photoshop CS6 (Adobe). Colocalization of RFFL–GFP with organelle markers was analyzed by means of the Pearson's correlation coefficient or Mander's correlation coefficient using Volocity 5 (PerkinElmer).
Super-resolution images of fixed cells were obtained using a Leica SP8 confocal system with HyVolution (Leica Microsystems) equipped with HyD hybrid detector. Series of x-y-z images were collected through a 0.5 Airy unit pinhole, and typical voxel sizes were 43 nm/pixel (x, y-axes) and 130 nm/pixel (z-axis). These were deconvolved with Huygens Essential software (Scientific Volume Image, Hilversum, The Netherlands).
Transferrin uptake assay
HeLa-ΔF508 cells were incubated in serum-free medium for 45 min at 37°C, followed by incubating in pre-warmed medium containing 25 µg/m Alexa Fluor 647-Tf (Thermo Fisher Scientific) for 2.5 h at 37°C. After washing twice with PBS, cells were incubated in full medium for the indicated time at 37°C.
Time-lapse imaging of TfR recycling
HeLa-ΔF508 CFTR-3HA cells on round cover glasses were incubated in serum-free medium for 45 min at 37°C, followed by incubated in pre-warmed medium containing 25 µg/ml Alexa Fluor 647-Tf (Thermo Fisher Scientific) for 2.5 h at 37°C and further incubated for 1 h after the Tf washout (T-0 h). Then, cells were immersed in NaKH solution (140 mM NaCl, 5 mM KCl, 10 mM HEPES, 10 mM glucose, 1 mM CaCl2, 0.1 mM MgCl2, pH 7.3) at 37°C. Time-lapse imaging was carried out on an inverted laser confocal fluorescence microscope (SP8, Leica) equipped with a HC PL APO 63×/NA 1.40 objective. A single z-section was imaged for 1 h at 2 min intervals at 37°C. The fluorescence intensities of intracellular Tf were obtained using LAS X software, and data were analyzed using Microsoft Excel.
CFTR uptake assay
HeLa cells stably expressing WT- or ΔF508-CFTR-3HA were incubated in pre-warmed medium containing anti-HA antibody (1:500, 16B12, BioLegend) for 2.5 h at 37°C. After washing twice with PBS, cells were incubated at 37°C for the indicated time. Cells were fixed, permeabilized with 0.1% Triton X-100 and stained with Alexa Fluor 594-conjugated anti-mouse IgG (Thermo Fisher Scientific). For the rescued ΔF508-CFTR experiment, cells were incubated at 26°C for 36–48 h before the anti-HA antibody loading.
CD59 uptake assay
CD59–Flag (Addgene #50378; Rivier et al., 2010) transfected HeLa-ΔF508 cells were incubated with anti-Flag antibody (1E6, FUJIFILM Wako Pure Chemical Corporation) for 1 h at 4°C to selectively label the cell surface with CD59–Flag. After washing twice with PBS, cells were incubated at 37°C for the indicated time. Cells were fixed, permeabilized with 0.1% Triton X-100 and stained with Alexa Fluor. 647-conjugated anti-mouse IgG (Thermo Fisher Scientific).
Dextran uptake assay
HeLa-ΔF508 cells were incubated in pre-warmed medium containing 1 mg/ml TRITC–dextran (Sigma) for 1 h at 37°C. After washing three times with PBS, cells were incubated at 37°C for 3 h. Cells were fixed, permeabilized with 0.1% saponin and stained with ant-Lamp1 antibody to examine the LY delivery of TRITC–dextran.
EGF uptake assay
HeLa-ΔF508 cells were incubated in serum-free medium overnight at 37°C, followed by incubation in medium containing 10 or 100 ng/m EGF (PeproTech), or 200 ng/ml EGF–biotin (Thermo Fisher Scientific) and Streptavidin-Alexa Fluor 594 complex at 37°C for the indicated time. To examine the cellular localization of EGFR, cells were immunostaining with anti-EGFR (6F1, MBL) and anti-EEA1 (3C10, MBL) or anti-Lamp1 (D2D11, Cell Signaling Technology) antibody. For Lamp1 staining, cells were permeabilized with 0.1% saponin.
HeLa-ΔF508 cells transfected with RFFL-H333A-GFP were fixed with 2% PFA and 2% glutaraldehyde (GA) in 0.1 M phosphate buffer (PB, pH 7.4) at 4°C for 30 min. Thereafter, cells were fixed with 2% GA in 0.1 M PB at 4°C overnight. After these fixations, cells were washed three times with 0.1 M PB for 30 min each and were post-fixed with 2% osmium tetroxide (OsO4) in 0.1 M PB at 4°C for 1 h. Cells were dehydrated in graded ethanol solution (50% for 5 min at 4°C, 70% for 5 min at 4°C, 90% for 5 min at RT, and three changes of 100% for 5 min at RT). The sample was transferred to a resin (Quethol-812; Nissin EM Co., Tokyo, Japan), and was polymerized at 60°C for 48 h. The polymerized resin was ultra-thin sectioned at 70 nm with a diamond knife on an ultramicrotome (Ultracut UCT; Leica, Vienna, Austria) and the sections were mounted onto copper grids. They were stained with 2% uranyl acetate at RT for 15 min, and then they were washed with distilled water followed by being secondary-stained with Lead stain solution (Sigma) at RT for 3 min. The grids were observed by a transmission electron microscope (JEM-1400Plus; JEOL Ltd., Tokyo, Japan) at an acceleration voltage of 80 kV. Digital images (2048×2048 pixels) were taken with a CCD camera (VELETA; Olympus Soft Imaging Solutions GmbH, Munster, Germany).
Sample preparation for scanning electron microscopy (SEM) was as described previously (Koga et al., 2012). HeLa-ΔF508 CFTR-3HA cells transfected with RFFL-H333A–GFP were collected with a cell scraper, and the cell suspension was centrifuged at 330 g for 5 min. After removal of the supernatant, the cell pellet was fixed with a mixture of 0.5% PFA and 0.5% GA in 0.1 M PB at pH 7.4 for 30 min at 4°C. The cell suspension was then centrifuged at 330 g for 5 min to collect the cell pellets. The cells were rinsed in 0.1 M PB for 10 min and centrifuged at 330 g for 5 min. After removal of the supernatant, the cell pellets were further fixed with 1% OsO4 in 0.1 M PB for 30 min. They were centrifuged at 330 g for 5 min and rinsed with the same buffer for 10 min after removal of the supernatant. The cell pellets were suspended in 1 ml of 0.1 M PB and mixed with 3 ml of 5% low-melting agarose solution (Sigma) that had been melted and cooled in a 35-mm diameter Petri dish to maintain the temperature at 40°C on a hot plate. The mixture of cells and agarose in the dish was then carefully stirred using the tip of a pipette, and the dish was placed in a refrigerator for 5 min to harden the agarose. The chilled agarose was peeled off from the dish and cut into small blocks with a blade. The agarose blocks were subsequently immersed in 25% and 50% dimethyl sulfoxide (DMSO) for 30 min each, frozen on a flat aluminum block that had been pre-cooled with liquid nitrogen and cracked into two pieces using a screwdriver and a hammer. The fractured pieces were immediately replaced in 50% DMSO for thawing at RT and rinsed in 0.1 M PB for 1 h. For cell maceration, specimens were immersed in 0.1% OsO4 in 0.1 M PB for 96 h at 20°C. The macerated specimens were then further fixed with 1% OsO4 in 0.1 M PB for 1 h, rinsed in 0.1 M PB for 1 h, treated with 1% tannic acid (Nacalai Tesque, Kyoto, Japan) in 0.1 M PB for 1 h, rinsed in the buffer solution for 1 h and immersed in 1% OsO4 in 0.1 M PB for 1 h. The specimens were dehydrated through a graded ethanol series, transferred to isoamyl acetate and dried in a critical point dryer (HCP-2, Hitachi, Tokyo, Japan) with liquid CO2. Dried specimens were subsequently mounted on aluminum stubs with silver paste, and coated with platinum-palladium at a thickness of ∼3 nm in an ion-sputter coater (E1010, Hitachi, Tokyo, Japan). Finally, they were observed in a field emission scanning electron microscope (S4100, Hitachi, Tokyo, Japan) at an accelerating voltage of 5 kV.
Pulldown assay and western blotting
To examine the interaction between RFFL and the Rab11 effectors, transfected 293MSR cells were solubilized in mild lysis buffer (150 mM NaCl, 20 mM Tris-HCl, 0.1% NP-40, pH 8.0, supplemented with 1 mM PMSF, 5 µg/ml leupeptin and pepstatin), and the cell lysates were incubated with NeutrAvidin agarose (Thermo Fisher Scientific) for 2 h at 4°C. After washing four times with mild lysis buffer, the complex was eluted in 2× Laemmli sample buffer supplemented with 10% β-mercaptoethanol and 3 mM biotin at 98°C for 10 min, and analyzed by western blotting as previously (Okiyoneda et al., 2018).
Antibodies used for western blotting were mouse monoclonal anti-V5 (FUJIFILM Wako Pure Chemical Corporation, 6F5; 1:1000), mouse monoclonal anti-Myc (FUJIFILM Wako Pure Chemical Corporation, 9E10; 1:1000), mouse monoclonal anti-GFP (Wako, mFX75; 1:1000), rabbit polyclonal anti-RFFL (SIGMA, HPA019492; 1:200), mouse monoclonal anti-GST (FUJIFILM Wako Pure Chemical Corporation, 5A7; 1:1000), mouse monoclonal anti-ubiquitin (Enzo Life Sciences, FK2; 1:200) and HRP-NeutrAvidin (Thermo Fisher Scientific, 31001; 1:5000).
Measurement of the ubiquitylation level in cells
293MSR cells were transiently transfected with HA–Ub, RFFL–V5 and Myc–Bio (MB)–EHD1, MB–MICALL1, MB–Rab11-FIP1C, MB–Rab11-FIP2 or MB–Rab11-FIP5. Cells were solubilized in lysis buffer (20 mM Tris-HCl pH7.4, 150 mM NaCl, 1% SDS and 1 mM EDTA) and boiled for 10 min at 98°C. The cell lysates were passed through a 27-gauge needle several times to decrease the viscosity. Then, the concentration of Triton X-100 and SDS in the cell lysate was adjusted to 1% and 0.25%, respectively, for NeutrAvidin pulldown overnight at 4°C. After washing five times with RIPA buffer supplemented with 2 M urea, the MB-tagged protein was eluted in 2× Laemmli sample buffer supplemented with 10% β-mercaptoethanol, 3 mM biotin at 98°C for 10 min, and analyzed by western blotting.
To identify the ubiquitylation sites in MB–EHD1, MB–MICALL1, MB–Rab11-FIP1C and MB–Rab11-FIP2, purified MB-tagged proteins were subjected to SDS-PAGE, followed by staining with Bio-Safe Coomassie (Bio-Rad). The gel slices containing the ubiquitylated MB-tagged proteins were reduced with 5 mM TCEP (Thermo Fisher Scientific) and alkylated using 10 mM methyl methanethiosulfonate (MMTS, Tokyo Chemical Industry). In-gel trypsin digestion was performed with 20 ng/µl trypsin (Thermo Fisher Scientific) overnight at 37°C. The resulting peptides, extracted from gel slices with acetonitrile, were desalted using a tC18 96-well plate (Waters). Peptides were then analyzed by LC-MS/MS on EASY-nLC 1000-connected Q-Exactive instruments (Thermo Fisher Scientific). Peptides were trapped on an Acclaim PepMap C18 precolumn (100 Å, Thermo Fisher Scientific) connected to an analytical column (C18, 3 µm particle size×75 µm diameter×125 mm, Nikkyo Technos) using 2 h gradients (0% to 40% over 120 min) with a constant flow of 300 nl/min. Peptides ionization was performed using Nanospray Flex Ion Source (Thermo Fisher Scientific). A ‘Top10’ data-dependent acquisition method was used where the top 10 most abundant ions are selected for MS/MS fragmentation by higher-energy collisional dissociation (HCD). MS raw files were processed by Proteome Discoverer version 2.0 (Thermo Fisher Scientific) for #PSM or MaxQuant version 18.104.22.168, which includes the Andromeda search engine (http://www.coxdocs.org/doku.php?id=maxquant:start) for MS/MS spectra, and searched against the SwissProt reviewed human reference proteome (UniProt). Raw files were subjected to additional database searches for di-gly (K) post-translational modifications.
HeLa-ΔF508 CFTR-3HA cells were transiently transfected with YFP–FRB, EHD1–FRB–GFP, GFP–FRB–MICALL1, Rab11-FIP1C–FRB–GFP or Rab11-FIP5–FRB–GFP, and CFP–FKBP or RFFL–FKBP–mCherry. Cells were incubated in medium containing 500 nM rapamycin for 5 min at 37°C. After changing the medium, the cells were incubated for 16 h at 37°C and analyzed after fixation.
Cellular localization analysis of biotinylated proteins
CFBE-teton-ΔF508 CFTR-3HA cells stably expressing RFFL–BirA*–HA or RFFL-H333A–BirA*–HA were incubated in medium containing 50 µM biotin for overnight at 37°C. Cells were fixed, permeabilized with 0.1% Triton-X100 and stained with Alexa Fluor 568-conjugated Streptavidin and anti-HA antibody with Alexa Fluor 488-conjugated secondary antibody.
CFBE-teton-ΔF508 CFTR-3HA cells stably expressing RFFL–BirA*–HA or RFFL-H333A–BirA*–HA were incubated in medium containing 50 µM biotin overnight (16 h) at 37°C. Cells were solubilized in RIPA buffer, and the cell lysates were incubated with NeutrAvidin agarose (Thermo Fisher Scientific) overnight at 4°C. The beads were washed with wash buffer 1 (1% SDS) twice, wash buffer 2 (0.1% deoxycholic acid, 1% Triton-X100, 1 mM EDTA, 500 mM NaCl, 50 mM HEPES) once, and wash buffer 3 (0.5% deoxycholic acid, 0.5% NP-40, 1 mM EDTA, 250 mM LiCl, 10 mM Tris-HCl pH7.4) once. After washing, biotinylated proteins were eluted in 2× Laemmli sample buffer (containing 10% β-mercaptoethanol) at 98°C for 10 min and analyzed by western blotting and silver staining.
Peptide sample preparation by in-gel digestion and nano flow-liquid chromatography tandem mass spectrometry was performed as described previously (Sadaie et al., 2008). Briefly, the gel slice was subjected to reduction with 10 mM dithiothreitol (DTT) at 56°C for an hour, and alkylation with 55 mM iodoacetamide at RT for 45 min. Then, in-gel digestion with 10 µg/ml modified trypsin (Sequencing grade, Promega) was performed at 37°C for 16 h. The digested peptides were extracted with 1% trifluoroacetic acid and 50% acetonitrile, dried under a vacuum, and dissolved in 2% acetonitrile and 0.1% formic acid. The peptides mixtures were then fractionated by C18 reverse-phase chromatography (ADVANCE UHPLC; AMR Inc.) and measured by a hybrid linear ion trap mass spectrometer (LTQ Orbitrap Velos Pro; Thermo Fisher Scientific) with Advanced Captive Spray SOURCE (AMR Inc.). The mass spectrometer was programmed to carry out 11 successive scans consisting of, first, a full-scan MS over the range 400–2000 m/z by FT-ICR at a resolution of 60,000, and second and eleven automatic data-dependent MS/MS scans of the top ten most-abundant ions obtained in the first scan by ion trap. MS/MS spectra were obtained by setting relative collision energy of 35% collision-induced dissociation and exclusion time of 20 s for molecules of the same m/z value range. The molecular masses of the resulting peptides were searched against the Homo sapiens amino acid sequence dataset (SwissProt downloaded at 2016_10) with the common repository of adventitious proteins (cRAP)* using the MASCOT version 2.6.1 (Matrix Science) with a false discovery rate set at 0.01 via the Proteome discoverer 2.1 (Thermo Fisher Scientific). Carbamidomethylation of cysteine was set as a fixed modification, and N-terminal acetylation and oxidation of methionine were included as variable modifications. The number of missed cleavages site was set as 2. All datasets were integrated using the Scaffold 4.8.4 (Proteome Software Inc.).
Establishment of RFFL KO cells by CRISPR/CAS9 system
Guide RNA sequences targeting the RFFL genetic loci were designed using CRISPRdirect (https://crispr.dbcls.jp/). The 20-bp guide sequence RFFL gRNA #1 (5′-GGCTCCGAACACTTCTTAAT-3′) or RFFL gRNA #2 (5′-CACAATGCTTAGAATGTCGT-3′) was inserted into pSpCas9(BB)-2A-Puro (PX459) V2.0 (Addgene #62988) using a BbsI restriction site to generate the RFFL sgRNA expression vectors. RFFL KO 293MSR cells were generated by co-transfection with the two RFFL sgRNA expression vectors, followed by 3 µg/ml puromycin selection for a day. Individual clones were generated by plating cells at low density and isolating individual colonies. RFFL KO was confirmed by western blotting and DNA sequencing. For sequencing of RFFL, the genomic locus was amplified by PCR using FW primer 5′-GTCCCCAGTACCTGCATTTGATATG-3′ and RV primer 5′-GGGAGGGTGCACACCTAGACACCAT-3′. The PCR product was cloned into pMD20-T using a Mighty TA-cloning Kit (Takara Bio), and the DNA sequence of the six6 colonies was determined by DNA sequencing, which gave in the same result (shown in Fig. S5C).
TfR recycling assay
293MSR (WT) and RFFL KO cells in 24-well plates were incubated in serum-free medium for 45 min at 37°C, followed by incubation in pre-warmed medium containing 25 μg/ml Transferrin Biotin-XX conjugate (Tf-biotin; Thermo Fisher Scientific) for 2.5 h at 37°C to label the internalized TfR. After the labeling, cells were washed with PBS twice and fixed in 4% PFA for 20 min at RT immediately (T-0 h) or after 4 h incubation in cell culture medium at 37°C (T-4 h). The fixed cells were permeabilized with 0.1% Triton X-100 in PBS for 5 min, and incubated with 0.5% BSA in PBS for 30 min at RT, followed by incubation with HRP-NeutrAvidin (Thermo Fisher Scientific) in 0.5% BSA-PBS for 1 h at RT. After washing six times with PBS, the cells were incubated Amplex Red (Thermo Fisher Scientific) for 20 min at RT to measure the HRP activity. The fluorescence was measured on a plate reader (Varioskan, Thermo Fisher Scientific) using 544-nm excitation and 590-nm emission wavelengths. Specific binding of the Tf was calculated by correcting the signal with the nonspecific adsorption of the HRP–NeutrAvidin.
His6–E1 (UBE1, Addgene #34965; Berndsen and Wolberger, 2011), His6–sumo-UbcH5c (Okiyoneda et al., 2018), His6–sumo-RFFL (Okiyoneda et al., 2018), GST–EHD1 and GST–Rab11-FIP1C were expressed in BL21 rosetta2 E. coli strain (Merck Millipore). Cells were lysed by incubation with 1 mg/ml lysozyme for 30 min on ice, followed by sonication. The His-tagged proteins and GST-tagged proteins were purified using Ni-affinity and glutathione-affinity chromatography, respectively, as described previously (Rabeh et al., 2012; Okiyoneda et al., 2018).
In vitro ubiquitylation assay
Purified GST–EHD1 (1.5 µg) was mixed with 0.2 µM His6–E1, 4 µM His6–sumo-UbcH5c, 2 µM His6–sumo-RFFL and 20 µM Ub (Boston Biochem, Cat#U-100H) in 45 µl of reaction buffer (20 mM HEPES pH 7.5, 50 mM NaCl, 5 mM MgCl2, 2.5 mM ATP, 2 mM DTT, and 20 µM MG-132) for 2 h at 37°C. Immediately after the reaction, 5 µl of the sample was analyzed by western blotting to confirm the activity of ubiquitylation enzymes. The rest of the sample (40 µl) was incubated with glutathione Sepharose 4B (GE Healthcare Life Sciences) to purified GST–EHD1, and the purified EHD1 was analyzed by western blotting with an anti-Ub antibody. For Rab11-FIP1C ubiquitylation, GST–Rab11-FIP1C (2 µg) was immobilized on glutathione Sepharose 4B, and incubated with 0.2 µM His6–E1, 4 µM His6–sumo-UbcH5c, 2 µM His6–sumo-RFFL and 20 µM Ub in 20 µl of the reaction buffer for 2 h at 37°C. After the reaction, the supernatant was analyzed by western blotting to confirm the activity of ubiquitylation enzymes. After washing the glutathione beads four times, the GST–Rab11-FIP1C was eluted and analyzed by western blotting.
For the ubiquitylation of MICALL1 and Rab11-FIP2, MB–MICALL1 or MB–Rab11-FIP2 was purified from 293MSR cells at 48 h post transfection. After solubilization in mild lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% NP-40) supplemented with 1 mM PMSF, 5 µg/ml leupeptin, 5 µg/ml pepstatin A, cell lysate was incubated with NeutrAvidin agarose (Thermo Fisher Scientific) for 2 h at 4°C. After being washed four times in mild lysis buffer, purified MB–MICALL1 or MB–Rab11-FIP2 bound to the agarose beads was incubated with components of the putative ubiquitylation machinery (0.2 µM His6–E1, 4 µM His6–sumo-UbcH5c, 4 µM His6–sumo-RFFL, 20 µM HA–Ub (Boston Biochem, Cat#U-110-01M)) in reaction buffer (20 mM HEPES pH 7.5, 50 mM NaCl, 5 mM MgCl2, 2.5 mM ATP, 2 mM DTT, 20 µM MG-132) for 2 h at 37°C. After the reaction, the supernatant containing E1, E2 and RFFL was collected for western blotting to confirm the auto-ubiquitylation. The beads were washed three times in mild lysis buffer, twice in 1% SDS, twoce in high salt buffer (20 mM Tris-HCl, pH 7.4, 0.5 M NaCl) and twice in RIPA buffer supplemented with 2 M urea. All wash steps were performed at 37°C for 5 min on mixing shaker. MB–MICALL1 or MB–Rab11-FIP2 was recovered by elution in 2× Laemmli sample buffer supplemented with 10% β-mercaptoethanol and 3 mM biotin at 98°C for 10 min, and were analyzed by western blotting.
For quantification, data from more than two technical repeats for each of at least two independent experiments were used, and data were expressed as means±s.e.m. Statistical significance was assessed by two-tailed paired Student's t-test using Excel software (Microsoft).
We thank Y. Sato (His6-sumo-UbcH5c), R. Shaw (addgene #31796, #31907), A. Helenius (Addgene #14437, #14436), R. Pagano (Addgene #12679), W. Mothes (Addgene #1817), R. Watanabe (Addgene #50378), M. Davidson (Addgene #54572, #55145), D. Root (Addgene #25890), Y. Zhang (Addgene #18671), T. Meyer (Addgene #20148, #20160), G. Voeltz (Addgene #49155), A. Sorkin (Addgene #32751), C. Wolberger (Addgene #34965) for supplying plasmids via Addgene, as well as DNASU for plasmids, Y. Seki for valuable advice for the establishment of RFFL KO cells and R. Nakagawa (Laboratory for Phyloinformatics in RIKEN Center for Biosystems Dynamics Research) for BioID MS analysis. We also thank T. Harris for carefully reading the manuscript.
Conceptualization: T.O.; Methodology: R.S., A.E., S.K., D.K., T.F., T.O.; Validation: R.S., T.O.; Formal analysis: R.S., R.F., T.O.; Investigation: R.S., R.F., S.U., M.A., Y.O., A.E., D.K., T.O.; Resources: T.O.; Data curation: R.S., T.O.; Writing - original draft: R.S., R.F., T.O.; Writing - review & editing: R.F., T.O.; Visualization: R.S., T.O.; Supervision: M.K., T.O.; Project administration: T.O.; Funding acquisition: T.F., M.K., T.O.
This work was supported by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant numbers JP25893275, 15H05643, 15H01192 to T.O.
The authors declare no competing or financial interests.